Department of Medicine Division of Gastroenterology and Hepatology, The University of North Carolina at Chapel Hill, North Carolina, USA
Department of Cell Biology and Physiology, The University of North Carolina at Chapel Hill, North Carolina, USA
Biomedical Engineering, The University of North Carolina at Chapel Hill, Chapel Hill, USA
Ph.D., University of North Carolina at Chapel Hill, 111 Mason Farm Rd., CB# 7032, MBRB Rm. 4337, Chapel Hill, North Carolina 27599, USA. E-mail: firstname.lastname@example.org Telephone: 919-966-6816; Fax: 919-843-6899
Author contributions: A.D.G. and S.T.M.: conceived and designed the study, wrote the manuscript, and carried out culture and gene expression experiments; M.K.F.: developed and optimized protocols for epithelial isolation and FACS; F.W. and L.L.: provided GSK-inhibitor conditions; M.S., J.C.Y.D., and M.G.M.: provided critical review of the manuscript. A.D.G. and M.K.F. contributed equally to this article.
Recent seminal studies have rapidly advanced the understanding of intestinal epithelial stem cell (IESC) biology in murine models. However, the lack of techniques suitable for isolation and subsequent downstream analysis of IESCs from human tissue has hindered the application of these findings toward the development of novel diagnostics and therapies with direct clinical relevance. This study demonstrates that the cluster of differentiation genes CD24 and CD44 are differentially expressed across LGR5 positive “active” stem cells as well as HOPX positive “facultative” stem cells. Fluorescence-activated cell sorting enables differential enrichment of LGR5 (CD24−/CD44+) and HOPX (CD24+/CD44+) cells for gene expression analysis and culture. These findings provide the fundamental methodology and basic cell surface signature necessary for isolating and studying intestinal stem cell populations in human physiology and disease. STEM Cells2013;31:2024-2030
Lgr5 was the first validated intestinal epithelial stem cell (IESC) biomarker shown to be expressed in actively cycling mouse crypt base columnar cells (CBCs) . Subsequent studies demonstrated a secondary, “reserve” population of mouse IESCs marked by Bmi1, Hopx, mTert, and Lrig [[2-5]]. Emerging evidence indicates overlapping expression of Lgr5 with these reserve IESC biomarkers; however, Lgr5-negative cell populations have also been shown to dedifferentiate in response to damage, suggesting the existence of one or more functionally competent “facultative” IESC populations [[6-8]]. Despite these advances in IESC biomarker discovery, fluorescence-activated cell sorting (FACS) isolation and functional characterization of putative “active” and “reserve/facultative” IESC populations from human intestinal tissue have been limited by the lack of validated human IESC biomarkers and in vitro assays to functionally test stemness at the single-cell level.
Investigators in other stem cell fields have used FACS-based approaches, which rely on multiple cell surface antigens, to isolate target stem cell populations of varying purity. Notably, biomarkers comprised of cluster-of-differentiation (CD) genes have long been used to identify hematopoietic stem cells and their progenitors . We recently adopted a similar strategy to demonstrate that low levels of CD24 facilitate FACS of Sox9Low/Lgr5+ murine IESCs capable of forming enteroids in vitro . Similarly, CD44 is expressed in the stem cell zone of the murine small intestine and can be used to enrich for Lgr5+ CBCs (Magness et al., unpublished data). In this study, we explored whether CD24 and CD44 could be used to FACS-isolate human IESCs.
Materials and Methods
Patients/Tissue Collection and Preparation
Deidentified tissue from female patients ranging between 33 and 53 years of age with body mass indices of 39–60 kg/m2 was used in this study. Tissue was obtained from laparoscopic roux-en-y gastric-bypass surgery and represents jejunal segments of approximately 4 cm in length. Following resection, tissue was placed in a specimen cup on ice until a mucosectomy was performed, aided by injecting ice-cold saline between the mucosa and submucosa prior to careful dissection. Single-cell dissociation was carried out on a small portion of the total mucosa (1 cm × 1 cm) for gene expression studies and a larger tissue area (4 cm × 4 cm) was dissociated for culture experiments. For an informative comparison, the mass of mucosa used for this preparation represents approximately 300- and 1,200 times the mucosal mass of an average biopsy from endoscopy or colonoscopy at University of North Carolina Hospitals (13 mg/biopsy; Drs. Tope Keku/Robert Sandler, unpublished data), respectively. Following dissection, mucosa was placed in 3 mM EDTA in 1× Phosphate buffered saline (PBS) for 45 minutes at 4°C on a rocker to remove villi. The villus fraction was discarded (Supporting Information Fig. S1A) and the remaining mucosa was then transferred into 5 ml of PBS and lightly shaken by hand (approximately one shake per second for 2 minutes) to remove the remaining epithelium (Supporting Information Fig. S1B). An equal volume of 2% Sorbitol made in 1× PBS (Sigma, St. Louis, MO) was added. To further deplete the solution of contaminating villi, the epithelial solution was passed through a 70 μm filter. This procedure results in a “crypt-enriched” epithelial preparation (Supporting Information Fig. S1C). The crypts were pelleted at 150g for 10 minutes at 4°C. Crypts were then digested to single cells by resuspending the pellet in 5 ml of Hanks' balanced saline solution containing 0.3 U/ml of dispase (Worthington Biochemical, Lakewood, NJ) followed by incubation at 37°C for 10 minutes. The crypt solution was then manually shaken for 30 seconds (three to four shakes per second). The solution was then checked for extent of dissociation to single cells. If cell clumps remained, shaking cycle was repeated every 5 minutes, then checked for extent of dissociation. Shaking cycles were stopped at the earliest time point at which 80%–90% of crypts were dissociated to completion or up to 30 minutes maximum. An average of 1 × 107 cells was obtained from a 1 cm × 1 cm mucosal segment. Single cells were filtered using a 40 μm filter to remove undissociated clumps. For FACS, 1 × 107 cells were placed in 500 μl of IESC staining media (as described in Flow Cytometry, below) for antibody staining. Human tissue used in this study was deemed exempt from full Institutional Review Board review (approval #09-2159).
A 3 cm square piece of jejunum from each case was fixed with freshly made 4% paraformaldehyde (PFA) for a 24–48 hours at 4°C. The tissues were then prepared for cryosectioning by immersion in 30% sucrose for at least 24 hours at 4°C. Tissues were embedded in Tissue-Tek optimal cutting temperature (OCT) medium (Sakura, Torrance, CA) and frozen on dry ice. Eight to ten micrometer sections were cut on a cryostat and placed on positively charged microscope slides. Prior to immunostaining, tissue sections were rinsed twice in PBS to remove OCT. Nonspecific binding was blocked by applying Dako Protein Block (Dako, Carpinteria, CA, X0909) to tissue sections for 30 minutes at room temperature. Primary antibodies were applied in Dako Antibody Diluent (Dako, S0809) and incubated for 2 hours at room temperature. Dilutions were as follows: CD326/EpCAM (1:250, clone 9C4, BioLegend, San Diego, CA), CD44 (1:250, clone IM7, BioLegend), CD24 (1:100, clone ML5, BioLegend), Lysozyme (1:500, Diagnostic Biosystems, Pleasanton, CA), Mucin2 (1:100, Santa Cruz Biotechnology, Santa Cruz, CA), Sucrase Isomaltase A-17 (1:100, Santa Cruz), and Chromogranin A (1:500, Immunostar, Hudson, WI). Anti-Rabbit-Cy3 (1:500 Sigma, C2306) and anti-Rat-Cy3 (1:500 Jackson Immunoresearch, Carlsbad, CA, 112-165-003) secondary detection antibodies were diluted in Dako Antibody Diluent and applied to tissue for 30 minutes at room temperature. Nuclei were stained for 10 minutes with bisbenzamide (1:1,000, Sigma) diluted in PBS. Background staining was negligible as determined by nonspecific IgG staining. Images were collected by capturing ∼1 μm optical sections using a Zeiss LSM 710 confocal microscope.
Cells were stained for 90 minutes on ice in IESC staining media (Advanced Dulbecco's modified Eagle's medium [DMEM]/F12 [Gibco, Grand Island, NY], N2 [Gibco], B27 [Gibco], Glutamax [Gibco], Penicillin/Streptomycin [Gibco], 10 μM Y27632 [Selleck Chemicals, Houston, TX], 500 mM N-acetyl-cysteine [Sigma], and 10% Fetal Bovine Serum (FBS) [Gemini Biosciences, West Sacramento, CA]) with the following antibodies or isotype controls: AlexaFluor 647 (Alexa647) conjugated anti-CD24 (1:100) (clone ML5, BioLegend #311109, San Diego, CA), Pycoerythrin-Cy7 (PE-Cy7) conjugated anti-CD45 (1.6:100) (BioLegend #304015), Fluorescein isothiocyanate conjugated anti-EpCAM (4:100) (clone 9C4, BioLegend #324203), and Brilliant Violet 421 (BV421) conjugated anti-CD44 (0.6:100) (clone IM7, BioLegend #103039). For initial analysis by flow cytometry, 1 × 106 cells were stained, fixed in 2% PFA for 10 minutes at room temperature, then rinsed (with PBS) and resuspended in 500 μl of PBS for analysis using a Beckman Coulter (Dako) CyAn ADP (Supporting Information Fig. S2). For sorting experiments, cells were rinsed and resuspended in IESC sort/culture media (Advanced DMEM/F12 [Gibco], N2 [Gibco], B27 [Gibco], Glutamax [Gibco], Penicillin/Streptomycin [Gibco], 10 mM HEPES [Gibco], 10 μM Y27632 [Selleck Chemicals], and 500 mM N-acetyl-cysteine [Sigma]). FACS was conducted using an iCyt Reflection (Visionary BioScience, Champaign, IL) for RNA collection or FACSAria (BD Biosciences, San Jose, CA) for cell culture experiments. Dead cells and debris were first excluded based on size via bivariate plot of forward scatter (FSC) versus side scatter (SSC) (Supporting Information Fig. S3A). Doublets/multimers were excluded using a bivariate plot of FSC peak versus FSC length (Supporting Information Fig. S3B). Epithelial cells were FACS-enriched by sorting EpCAM (CD326) positive, CD45-negative cells (Supporting Information Fig. S3C). The remaining cell events were analyzed for CD24 and CD44 expression on a bivariate plot (Supporting Information Fig. S3D). Five cell populations: CD45−EpCAM+ (whole), CD45−EpCAM+CD24−CD44−(negative), CD45− EpCAM+CD24+CD44− (CD24+CD44−), CD45−EpCAM+ CD24−CD44+(CD24−CD44+), CD45−EpC AM+CD24+CD 44+(CD24+CD44+) were collected directly into 500 μl of RNA lysis buffer (Ambion RNAqueous Micro, Grand Island, NY) for gene expression analysis. For cell culture experiments, cells were collected into 500 μl of IESC staining media.
Primary Isolation of Human Small Intestinal Myofibroblasts
Following mucosectomy for epithelial prep, remnant submucosa (∼4 cm × 4 cm) was diced with a sterile razor blade. To eliminate blood cells and remnant epithelium, tissue chunks were washed in 20 ml DMEM (Gibco), shaken for 2 minutes, and then allowed to settle before supernatant was removed and discarded. This process was repeated eight times. The submucosal tissue was then resuspended in 5 ml DMEM containing Penicillin/Streptomycin, 0.3 U/ml dispase, and 300 U/ml collagenase I (Sigma) and rotated for 25 minutes at room temperature. Ten milliliters of DMEM +10% FBS was added to quench the reaction and the tissue suspension was pipetted vigorously ∼50 times to further mechanically dissociate myofibroblasts. The tissue suspension was centrifuged at 300g for 5 minutes and the resulting supernatant and tissue remnants were plated separately in DMEM +10% FBS. Media was changed every 24 hours. Cultures initiated from the supernatant of the prep produced myofibroblasts, which were passaged three times before use in IESC culture experiments.
Cells were pelleted and resuspended in ES-qualified Matrigel (BD Biosciences) containing IESC culture growth factors (Supporting Information Table S2). For feeder-free cultures, 10 μl Matrigel droplets were plated in 96-well plates and overlaid with 100 μl IESC sort/culture media, with or without 2.5 μM CHIR99021 (Selleck Chemical) following 15-minute of polymerization at 37°C. For feeder coculture, 25 μl of Matrigel droplets were allowed to polymerize in 12-well transwell inserts (BD Biosciences) before being placed in wells containing fibroblast feeder cells and 500 μl IESC sort/culture Media. An additional 500 μl of the same media was placed in the transwell to prevent drying of the Matrigel droplet. CHIR99021 was not used in coculture experiments. To facilitate differentiation of enteroids, Wnt3a, SB202190, and nicotinamide were withdrawn at 14 days of culture, as previously described for whole crypt cultures .
cDNA Preparation/Real-Time PCR Analysis
mRNA from sorted cell populations was purified using RNAqueous Micro Kit (Ambion) according to the manufacturer's protocols. cDNA was generated using iTaq Reverse Transcription Supermix (Bio-Rad, Hercules, CA). Real-time PCR was conducted for each sample in triplicate on 1/20,000 of the total amount of cDNA generated. Taqman probes (18S, HS99999901; CD24 Hs00273561_s1; CD44, Hs01075861_ m1; LGR5 Hs00173664_m1; OLFM4 HS00197437_m1; HOPX Hs04188695_m1; DEFA6 Hs00427001_m1; LYZ Hs00 426232; MUC2 Hs00159374_m1; CHGA Hs00900373_m1) for each gene were obtained from Applied Biosystems (Pleasanton, CA) and used in reactions according to the manufacturer's protocol.
Quantitative reverse transcriptase polymerase chain reaction (qRT-PCR) data were normalized for the expression of 18S. ΔΔCt values were then calculated using the CD24−CD44− (negative) cell population as the comparator. Statistical analysis compared gene expression across all cell populations by gene for each patient via one-way ANOVA followed by Bonferroni post-test for multiple comparisons between the population of interest and all other populations. Statistical analysis was performed in Graph Pad Prism (version 4.0, La Jolla, CA).
Results and Discussion
CD24 and CD44 expression was assessed on human jejunum derived from patients who had undergone roux-en-y gastric-bypass surgery. Immunostaining demonstrates that CD44 is expressed on cells from the base of the crypt to the crypt-villus junction (Fig. 1A, 1C, 1E). By contrast, the villus epithelium does not show appreciable CD44 staining (Fig. 1B, 1D). CD24 demonstrates similar expression to CD44 in the epithelium with the notable exception that staining is primarily distributed along the apical membrane (Fig. 1F, 1H, arrows). A minority population of crypt-based cells expresses high levels of cytoplasmic CD24 (Fig. 1G, 1I, arrows). While CD24/44 expression is highly restricted to the stem cell zone in the epithelium, there is broad expression in nonepithelial cells in the lamina propria, submucosa, and muscle (Fig. 1A–1I). EpCAM (CD326) expression is unique to all crypt and villus epithelial cells and was deemed useful for positive FACS selection to separate epithelial from nonepithelial CD24/44-expressing cells (Fig. 1J–1L).
Next, we analyzed CD24, CD44, CD45, and CD326 on dissociated single epithelial cells (Supporting Information Fig. S2). Positive selection for epithelial cells (EpCAM+) and negative selection for lymphocytes (CD45−) was included in the FACS parameters for robust epithelial enrichment (Supporting Information Fig. S3C). Flow cytometry data demonstrate four distinct epithelial cell populations based on CD24/44 expression status (Supporting Information Fig. S3D). Each of the four populations (CD24−/44−; CD24+/44−; CD24−/44+; CD24+/44+) was collected for gene expression analysis. Semi-quantitative RT-PCR validated the FACS by showing enriched gene expression for CD24/44 in the appropriate populations (Supporting Information Fig. S4A, S4B). The data demonstrate that CD24−/CD44+ populations are most enriched for the active cycling IESC markers, LGR5 and OLFM4 (Fig. 2A, 2B); and the CD24+/CD44+ population is most enriched for the reserve/facultative IESC marker HOPX (Fig. 2C). Interestingly, all CD24+ and CD44+ populations demonstrate significant de-enrichment for Paneth cell markers DefensinA6 (DEFA6) (Supporting Information Fig. S5A) and Lysozyme (LYZ) (Supporting Information Fig. S5B), and goblet cell marker, Mucin 2 (MUC2) (Supporting Information Fig. S5C), all of which associate with CD24 expression in mice [[7, 12, 13]]. However, the CD24+/CD44+ population is highly enriched for the enteroendocrine cell marker Chromogranin A (CHGA) which is consistent with observations made in mice (Supporting Information Fig. S5D) .
To test whether the CD24−/CD44+ and CD24+/CD44+ populations had functional properties of stemness, we subjected isolated single cells to culture conditions similar to those that have been successfully used to grow small intestine crypts and single colonic stem cells from human tissue [[11, 14]]. Both IESC populations formed appreciable cystic enterosphere structures by 48 hours (Fig. 3A), while the other populations (CD24−/CD44− and CD24+/CD44−) failed to do so. Enterospheres derived from CD24−/CD44+ cells continued to develop over the first week of culture, while CD24+/CD44+ derived enterospheres exhibited a limited increase in size (Fig. 3A). Nevertheless, both populations persisted in culture at 7 days (Fig. 3B, 3C). In an attempt to increase culture efficiency, we added glycogen synthase kinase (GSK)-inhibitor CHIR99021, which promotes self-renewal of embryonic stem cells and was recently used to enhance in vitro growth and survival of human colon cancer stem cells and intestinal crypts (Wang et al., unpublished data) [[15, 16]]. Interestingly, initial GSK-inhibition significantly improved 7-day enterosphere survival in the CD24−/CD44+ population, but had no appreciable effect on the CD24+/CD44+ population (Fig. 3B, 3C). Initial GSK-inhibition greatly increased 14-day survival of the CD24−/CD44+ population, but was insufficient for the maintenance of CD24+/CD44+ derived enteroids, which did not survive to 14 days, regardless of GSK-inhibition (Fig. 3C). Previous studies have shown that Lgr5-negative populations can dedifferentiate and function as facultative stem cells when presented with the proper extrinsic cues [[7, 8]]. Additionally, coculture of primary human intestinal crypts with myofibroblasts enhances culture efficiency . In an attempt to “activate” the CD24+/CD44+ population, we cocultured the cells with myofibroblasts isolated from human jejunal submucosa. Surprisingly, when grown in these conditions, both CD24−/CD44+ and CD24+/CD44+ cells produced long-lived enteroids, at rates of 0.07% and 0.76%, respectively (Supporting Information Fig. S6).
To assess multipotency, we retrieved and processed enteroids for immunohistochemistry at 21 days. CD24−/CD44+ derived enteroids were epithelial in nature (Supporting Information Fig. S7) and produced enteroendocrine (CHGA), goblet (MUC2), and Paneth cells (LYZ) as well as absorptive enterocytes (SIM) (Fig. 4B). CD24+/CD44+ cells grown in coculture with primary human myofibroblasts exhibited the same characteristics of multipotency as CD24−/CD44+ cells (Fig. 4C and Supporting Information Fig. S7). Enteroids derived from both populations exhibited increased numbers of lysozyme positive cells, consistent with results in mice demonstrating that persistent WNT signaling, specifically through CHIR99021 treatment, increases Paneth cell production in enteroid culture . This biased secretory lineage allocation likely explains the rare occurrence of MUC2-positive cells in the enteroids. Importantly, coculture with myofibroblasts did not induce growth in CD24−/CD44− and CD24+/CD44− populations, demonstrating that functional stemness under all tested growth conditions remains restricted to CD24−/CD44+ and CD24+/CD44+ populations.
The translation of scientific findings made in genetically homogeneous mouse models presents investigators with a daunting challenge when attempting to understand the vast heterogeneity underlying physiology in human populations. This report provides new methodology for epithelial dissociation, isolation, and FACS enrichment of IESCs from human small intestine through the use of a combinatorial CD-marker “signature.” The ability of CD24 and CD44 to enrich for IESCs is highly consistent with observations in mice; however, in stark contrast to findings in mouse models, human Paneth cells do not appear to express high levels of CD24, highlighting the need to validate mouse genetic biomarkers on human samples [[7, 10, 12, 13]]. Furthermore, our results demonstrate variable levels of gene enrichment between some patients. This observation could be attributed to a wide range of factors, including phenotypic heterogeneity between samples. In this study, tissue procurement protocols required full deidentification of samples; thus, a patient's dietary habits, state of illness, or medication history could not be correlated to gene expression variations. This highlights the need for comprehensive studies with controlled enrollment criteria, using the foundation for isolation of LGR5 and HOPX-enriched populations, presented here.
In addition to FACS-enrichment of active LGR5+ IESCs, the CD24/CD44 approach facilitates differential isolation of HOPX+ cells, which emerging evidence suggests may represent facultative IESCs . While several of the markers examined here exhibit a degree of variability between samples, it is clear that the population of cells most enriched for LGR5 is distinct from the population most enriched for HOPX, unlike recent observations of Hopx expression in murine Lgr5+ cells . Additionally, CD24−/CD44+ and CD24+/CD44+ populations exhibit differential behavior in vitro, both in response to GSK-inhibition and in basal culture conditions required for growth, further supporting the existence of phenotypically distinct IESC populations. Although overlap of LGR5 expression with putative reserve IESC markers in mice remains controversial, our data suggest that more distinct populations may exist in the human small intestinal epithelium [[4-6, 19]]. Further work is needed to characterize the CD24+/CD44+ population and determine whether it is analogous to proposed facultative IESCs observed in mouse models of intestinal damage and regeneration. Multiple reports have demonstrated the ability of Lgr5-negative facultative IESCs to drive regeneration in murine intestine following irradiation, suggesting that an analogous population may be of particular interest in human pathophysiology, especially in patients undergoing radiation treatment for tumors [[7, 8]]. Understanding the genetics and patient-to-patient heterogeneity of phenotypically distinct human IESC populations may provide valuable insight toward the development of novel clinical diagnostics and therapeutic interventions.
We thank the UNC flow cytometry facility (P30CA06086) and the neuroscience confocal facility, which is cofunded by the National Institute of Neurological Disorders and Stroke and the National Institute of Child Health and Human Development (P30NS045892). Drs. Timothy Farrell and Wayne Overby, and Myra Jones, Karen Colton, and Lisa Prestia, for coordinating the collection of deidentified remnant jejunal tissue from gastric-bypass operations. The Duke Human Vaccine Institute flow cytometry core, and Dr. John Whitesides and Ian Cumming, for technical assistance. Drs. Tope Keku and Robert Sandler for unpublished data regarding mucosal biopsy weight. Drs. P. Kay Lund, Susan Henning (SJH), and Christopher Dekaney for useful discussions and critical review of the manuscript. This work was funded by the NC TraCS Institute Pilot Grant Program, 2KR271103 (Gracz/Fuller) and 2KR381202 (Gracz), the Center for Gastrointestinal Biology and Disease, 5P30DK034987-27 (Magness), and a University Cancer Research Fund Innovation Award (Magness). The NC TraCS Institute is supported by Grants UL1RR025747, KL2RR025746, and TLRR025745 from the NIH National Center for Research Resources and the National Center for Advancing Translational Sciences, National Institutes of Health. The National Institute of Diabetes and Digestive and Kidney Diseases/National Institute of Allergy and Infectious Disease Intestinal Stem Cell Consortium (U01DK085541-SJH; U01DK85507-LL/FW; U01DK85535-MGM/JCYD/MS). The content is solely the responsibility of the authors and does not necessarily represent the official views of the National Center for Research Resources or the National Institutes of Health.
Disclosure of Potential Conflicts of Interest
The authors declare that they have no conflicts of interest.