Adult Stem Cells in the Small Intestine Are Intrinsically Programmed with Their Location‐Specific Function

Abstract

Differentiation and specialization of epithelial cells in the small intestine are regulated in two ways. First, there is differentiation along the crypt‐villus axis of the intestinal stem cells into absorptive enterocytes, Paneth, goblet, tuft, enteroendocrine, or M cells, which is mainly regulated by WNT. Second, there is specialization along the cephalocaudal axis with different absorptive and digestive functions in duodenum, jejunum, and ileum that is controlled by several transcription factors such as GATA4. However, so far it is unknown whether location‐specific functional properties are intrinsically programmed within stem cells or if continuous signaling from mesenchymal cells is necessary to maintain the location‐specific identity of the small intestine. Using the pure epithelial organoid technique, we show that region‐specific gene expression profiles are conserved throughout long‐term cultures of both mouse and human intestinal stem cells and correlated with differential Gata4 expression. Furthermore, the human organoid culture system demonstrates that Gata4‐regulated gene expression is only allowed in absence of WNT signaling. These data show that location‐specific function is intrinsically programmed in the adult stem cells of the small intestine and that their differentiation fate is independent of location‐specific extracellular signals. In light of the potential future clinical application of small intestine‐derived organoids, our data imply that it is important to generate GATA4‐positive and GATA4‐negative cultures to regenerate all essential functions of the small intestine. Stem Cells  2014;32:1083–1091

Introduction

The mammalian intestine is responsible for the absorption of dietary nutrients and water, and the excretion of waste materials. The intestine also has an important barrier function in preventing direct contact with bacteria that reside in the lumen. To facilitate these diverse functions, the intestinal surface is covered by specialized epithelial cells that can take up nutrients from the lumen, but also form a continuous lining of cells that are linked close together by tight junctions.

The intestinal epithelium is ordered into crypts of Lieberkühn that harbor the stem and Paneth cells and mainly constitute a proliferative compartment, and the villi that represent the differentiated compartment containing the enterocytes, goblet, tuft, and enteroendocrine cells. It has been shown that four signaling pathways control homeostasis of the intestinal epithelium: WNT, Notch, epidermal growth factor (EGF), and bone morphogenic protein (BMP) (reviewed in [1]). WNT and EGF signaling are active in a gradient from the bottom to the crypt‐villus junction, while BMP signaling is active in the villus compartment and Notch signaling determines the differentiation toward the secretory lineage (reviewed in [2]). High expression of WNT induces stem cell survival and proliferation of transit amplifying (TA) cells, while low levels of WNT allow TA cells to fully differentiate into one of the various epithelial cell types [3].

In addition to differentiation along the crypt‐villus axis, there are also structural differences along the length of the intestine, such as cell type distribution and expression of proteins with specialized functions [4-6]. These differences correlate with the functionality of the different regions of the intestine and are referred to as regional or spatial differences along the cephalocaudal axis. This structural organization is established during development and is maintained throughout adulthood [7].

The small intestine is roughly divided into three segments along the cephalocaudal axis: duodenum, jejunum, and ileum [8]. In mice, the duodenum involves the proximal two centimeters directly after the gastric pylorus [8]. Its main function is the digestion and absorption of iron, calcium, and water‐soluble vitamins. The function of the jejunum is the digestion and uptake of nutrients, whereas the ileum absorbs residual nutrients and mediates transport of bile acids and vitamin B12 [8].

To facilitate these highly specialized functions, multiple proteins such as digestive enzymes, nutrient transporters, cytoplasmic carriers, and antimicrobial peptides are differentially expressed along the cephalocaudal axis. For example, duodenal cytochrome b reductase 1 (Cybrd1), involved in iron metabolism, is expressed only in duodenum [9], lactase‐phlorizin hydrolase (Lct) is mainly expressed in jejunum [10], and apical sodium‐dependent bile acid transporter (ASBT, Slc10a2) in ileum [11].

It has been shown that several transcription factors, for example, GATA4 and CDX2, control the expression of genes along the cephalocaudal axis [12, 13]. As such, GATA4 is expressed in proximal small intestine where it inhibits expression of ileum‐specific genes [12], whereas CDX2 is expressed throughout small and large intestine and has an important role in formation of normal intestinal identity [13, 14]. Furthermore, syngeneic and xenogeneic transplantation of intestinal tissue into murine hosts have demonstrated that donor fetal intestine retains its positional information [15-18]. However, it is not known whether the identity of the intestinal epithelium is a hard‐wired program within the stem cells or if the microenvironment, such as surrounding mesenchymal cells or luminal content, provides location‐specific signals to the stem cells.

Recently, we developed a method to grow intestinal epithelial tissue in vitro from isolated crypts of both murine [19] and human origin [20]. We characterized the functional properties of organoids that were derived from different segments of the small intestine of mice and human. We found that the functional fate of both murine and human differentiated cell types is intrinsically programmed within location‐specific stem cells as specific gene expression programs are maintained in the absence of location‐specific external signals from mesenchyme or microbiota.

Materials  and Methods

Mice

C57BL/6 wild‐type mice (6–12 weeks old) were used for experiments. The mice were specific pathogen free and maintained in an environmentally controlled facility. Experiments were approved by the local Ethical Committee. Murine small intestines were dissected and cut into three equal parts. We isolated small parts of duodenum (proximal 2 cm of the proximal part), jejunum (proximal 2–4 cm of the middle part), and ileum (distal 2–4 cm of the distal part). Each segment was opened longitudinally, washed with cold phosphate buffered saline (PBS), and further processed for crypt isolation. From other mice, similar pieces of intestinal segments were used for ex vivo RNA isolation and histology.

Human Material

Approval for this study was obtained by the Ethics Committee of the University Medical Centre Utrecht. Duodenal biopsies were obtained by flexible gastroduodenoscopy and ileal biopsies were obtained from the terminal ileum by colonoscopy. The biopsies were macroscopically and pathologically normal.

Crypt Isolation and Organoid Culture

Murine organoids were generated from isolated small intestinal crypts and maintained in culture as described previously [19]. Crypt isolation and culture of human intestinal cells from biopsies have been described previously [20, 21]. In short, human organoids were maintained long‐term in expansion medium (EM) containing RSPO1, noggin, EGF, A83‐01, nicotinamide, SB202190, and WNT3A. For induction of differentiation, cultures were maintained for 5 days in differentiation medium (DM), which is EM without nicotinamide, SB202190, and WNT3A. We used conditioned media for RSPO1 (stably transfected RSPO1 HEK293T cells were kindly provided by Dr. C. J. Kuo, Department of Medicine, Stanford, CA), noggin, and WNT3A. The medium was changed every 2–3 days and organoids were passaged 1:6 every week.

RNA Isolation and qPCR

From dissected tissue or cultured organoids, RNA was isolated with the RNeasy minikit or microkit (Qiagen, Venlo, Netherlands), respectively, and quantified by optical density. cDNA was synthesized from 1 µg of RNA by performing reverse‐transcription (Invitrogen, Carlsbad, CA or iScript, Biorad, Hercules, CA). Messenger RNA (mRNA) abundances were determined by real‐time PCR using validated primer pairs (Supporting Information Table S1) with the SYBR Green method (Bio‐Rad, Hercules, CA). Glyceraldehyde‐3‐phosphate dehydrogenase (Gapdh) or ACTIN mRNA abundance was used to normalize mouse or human data, respectively.

RNA Sequencing

Freshly isolated villi, crypts, or organoids (cultured for 12 weeks) were used for mRNA sequencing. Total RNA was isolated using Trizol LS reagent (Invitrogen). Total RNA was purified using the Poly(A)Purist MAG Kit (Life Technologies, Ghent, Belgium) and subsequently repurified using the mRNA‐ONLY Eukaryotic mRNA Isolation Kit (Epicentre, Madison, Wisconsin) to obtain full length mRNA. Sequencing libraries were constructed using the SOLiD Total RNA‐Seq Kit (Life technologies, Ghent, Belgium) and sequenced on a SOLiD 4 sequencer to produce 50‐bp long reads. Reads per 1,000 base pairs of transcript per million reads sequenced (RPKM) values were calculated for all RefSeq annotated genes and quantile normalized. A location‐specific gene was defined as a gene with expression >10 RPKM (crypt) or >5 RPKM (villi) in at least one part of the intestine and with >1.5‐fold higher expression in a specific part compared to any other part separately in both replicates (crypt) or in single replicate (villi). The gene ontology (GO) analysis was performed using the ToppGene Suite website [22]. The mRNA‐seq datasets can be accessed via GEO (www.ncbi.nlm.nih.gov/geo) accession number GSE53297.

Immunohistochemistry

Intestinal segments and organoids were fixed in neutral buffered formalin and sectioned. After deparaffination and dehydration, epitope retrieval was performed by boiling for 15 minutes at 120 °C in target retrieval solution pH = 6.0 (DAKO, Glostrup, Denmark). Slides were blocked for endogenous peroxidase with 3% H₂O₂/PBS for 20 minutes and blocked with 1% bovine serum albumine (BSA)/PBS for 30 minutes at RT. Primary antibodies were diluted in 0.05% BSA/PBS and incubated for 1 hour at RT or o/n at 4°C. Next, slides were incubated with secondary antibody (Envision anti‐Rabbit‐HRP or Envision anti‐mouse‐HRP, DAKO) 30 minutes at RT, with Vector‐Nova Red (Vector Labs, Burlingame, CA) to develop staining and counterstained with hematoxylin. Primary antibodies were monoclonal mouse anti‐human proliferating cell nuclear antigen (NCL‐PCNA, Leica Biosystems, Wertzlar, Germany), polyclonal rabbit anti‐human sucrose‐isomaltase (SI, HPA011897, Sigma, St. Louis, MO), and polyclonal rabbit anti‐human SLC10A2 (ASBT, HPA004795, Sigma). Alcian Blue–periodic acid Schiff (AB‐PAS) staining was performed according to standard methods.

Statistics

Data were expressed as means ± SEM and analyzed using one‐way ANOVA and Tukey’s Multiple Comparison Post‐test.

Results

Location‐Specific Signatures Within the Murine Small Intestinal Crypts Are Maintained in Long‐Term Epithelial Cell Cultures

To obtain a systematic overview of location‐specific gene expression, we performed mRNA sequencing on crypts and villi isolated from three equal parts of the small intestine. For analysis, we listed genes that were most strongly expressed in duodenum (A), jejunum (B), and ileum (C) or in two adjacent compartments (AB, BC). Next, we generated organoids from similar crypt preparations of the three small intestinal segments and cultured them for 12 weeks. mRNA sequencing profiles of these cultures were compared to their location‐specific crypt‐derived (Fig. 1A) and villus‐derived (Fig. 1B) ex vivo counterparts. For example, we found that from those genes that were specifically expressed in crypts or villi of duodenum, 78% or 43% of these genes was also specifically expressed in organoids derived from duodenum, respectively (Fig. 1 and Table 1). Similarly, 87% versus 83% of duodenal‐jejunal, 70% versus 65% of jejunal, 77% versus 73% of jejunal‐ileal, and 65% versus 49% of ileal crypt‐ versus villi‐specific genes were maintained in organoids derived from their ex vivo counterparts, respectively (Fig. 1 and Table 1). GO analysis of the selected genes was performed (Supporting Information Tables S2, S3) and showed that the location‐specific crypt‐ or villus‐derived genes are mainly involved in digestion, transport, and metabolic processes. The complete list of location‐specific crypt‐ and villi‐genes is deposited in the GEO database.

In summary, we confirmed differential expression of several genes that are known to be involved in location‐specific function of the small intestine, such as genes coding for the proteins ferroportin (FPN) (Slc40a1) in duodenum, lactase (Lct) in jejunum, GATA4 (Gata4) in duodenum‐jejunum, and ASBT (Slc10a2) in ileum (Fig. 1). These data show that crypt‐derived genes were highly maintained in organoid cultures, indicating that mouse organoids are mainly crypt‐based structures. However, organoids do contain differentiated cells that express villus‐derived genes, albeit in low levels. As the organoids were cultured in the same conditions, that is, in Matrigel and in presence of EGF, noggin, and R‐spondin1, these data indicate that epithelial stem cells maintain their identity, without location‐specific external signals from mesenchyme or luminal content.

To validate the mRNA sequencing data, we determined expression of several location‐specific genes by quantitative reverse transcription PCR (qRT‐PCR). We used small pieces (∼2 cm) of each compartment of the small intestine for direct ex vivo RNA isolation or generation of organoid cultures that were maintained short term (3 weeks) or long‐term (10–12 weeks).

The transcription factor GATA4 has been shown to be a key regulator of location‐specific gene expression in the mouse small intestine by repressing ileum‐specific genes [12, 23]. In concordance, mRNA sequencing as well as qRT‐PCR analysis confirmed that Gata4 expression is high in duodenum and jejunum but almost absent in ileum. We found that this expression pattern was maintained in organoids that were cultured for 3 and 10–12 weeks (Figs. 1, 2A).

Duodenum‐Specific Gene Expression

The uptake of nutritional iron involves the reduction of Fe³⁺ into Fe²⁺ by ferric reductases such as CYBRD1 (encoded by Cybrd1) in the intestinal brush border and the subsequent transport of Fe²⁺ across the apical membrane of enterocytes by the divalent metal transporter 1 [9]. Iron is eventually exported across the basolateral membrane into the bloodstream via the solute carrier and Fe²⁺ transporter FPN (encoded by Slc40a1) [24]. Transcripts of these genes were only found in our villus‐based mRNA sequencing data (Fig. 1B) because they are expressed at the tips of the villi in vivo [9]. We determined mRNA expression levels of Cybrd1 and Slc40a1 by qRT‐PCR and found that their expression ex vivo is indeed restricted to the duodenum and this cephalocaudal expression pattern remained stable in cultured organoids (Fig. 2B).

Jejunum‐Specific Gene Expression

It has been shown that disaccharidases, like lactase (encoded by Lct) and sucrase‐isomaltase (encoded by Sis), show the highest expression and activity in the jejunum [25]. We confirmed these expression patterns in ex vivo tissue and for Lct we also found highest expression in jejunum‐derived organoid cultures, whereas the specific expression pattern of Sis was less obvious (Fig. 2C).

Ileum‐Specific Gene Expression

The bile acid transporter ASBT (encoded by Slc10a2) is expressed apically on enterocytes in the ileum and can take up bile from the intestinal lumen, while organic solute transporter (OST)α/β form a pore at the basolateral side of the cell to export bile acids from the enterocytes to the blood [26]. We found by mRNA sequencing and qRT‐PCR that the expression of both Slc10a2 and Ostb was highly specific for the ileum, and that expression was maintained throughout organoid culture (Figs. 1, 2D). We observed low Ostb expression in ex vivo jejunum, which was also reflected in the cultures. Collectively, these data show that the functional fate of mature intestinal epithelial cells is intrinsically programmed within the stem cell niches of each location and that these signatures are maintained even after long‐term organoid culture where stem cells are not exposed to exogenous signals from mesenchyme or luminal content.

Mouse Organoids Do Not Require Exogenous WNT for Long‐Term Maintenance

We have shown that mouse stem cells maintain their location‐specific origin during organoid culture by analyzing expression patterns of genes that are involved in location‐specific functions along the cephalocaudal axis of the small intestine.

However, mouse organoids consist mainly of crypt compartments and contain only limited amounts of fully differentiated enterocytes that fulfill the location‐specific function [19]. As such, the percentages of location‐specific genes that were maintained in organoids were higher for crypt‐derived genes compared to villus‐derived genes (Fig. 1 and Table 1). Furthermore, it has been shown that mouse small intestinal organoids have an intrinsic WNT gradient, which allows proliferation but limits differentiation [19], resulting in low levels of gene transcripts for the genes of interest.

By adding WNT3A‐conditioned media to the location‐specific mouse cultures, we show that WNT inhibits the expression of differentiation markers, such as Sis and Slc10a2, whereas the expression of Gata4 remains unchanged (Supporting Information Fig. S1A). Morphologically, the addition of WNT results in more spherical organoids compared to the standard culture condition (Supporting Information Fig. S1B), an indication for induced proliferation.

Location‐Specific Genes Are Induced Upon Differentiation of Human Organoid Cultures

In contrast to mouse small intestinal organoids, human small intestinal organoid cultures produce little endogenous WNT. Therefore, their proliferation is dependent on exogenous WNT and differentiation is only initiated upon withdrawal of WNT [20]. Here, we use the human organoid system to delineate the effects of WNT on human location‐specific small intestinal organoid cultures and show the full differentiation capacity of location‐specific stem cells.

Human organoid cultures were established from crypts that were isolated from duodenal and ileal small intestinal biopsies. The organoids were maintained in so‐called EM, containing WNT3A conditioned medium, which induces expansion of stem cells and repression of differentiation. We cultured the organoids for 3 (data not shown) or 7 weeks in EM, after which the medium was replaced by so‐called DM for 5 days. DM lacks WNT3A, Nicotinamide, and SB202190 in order to inhibit expansion and induce differentiation, demonstrated by the loss of LGR5 mRNA (Fig. 3A left), more dense morphology (Supporting Information Fig. S2), and loss of PCNA staining on immunohistochemistry (Fig. 4). Incubation in DM also induced the differentiation of Goblet cells and production of glycoproteins as shown by AB‐PAS staining of both duodenal and ileal organoids (Fig. 4).

We confirmed GATA4 as a specific marker for duodenum and found that the expression levels did not change in the absence of WNT3A (Fig. 3A right). This was in concordance with the previously reported finding that GATA4 protein in proximal small intestine is expressed in all cells from crypt to tips of the villi [27].

We found that induction of CYBRD1 and SLC40A1 was equal or even higher in ileal organoids compared to duodenal organoids (Fig. 3B) and that SI (coding for SI) expression was only induced in organoids derived from human ileum and not in duodenal organoids (Figs. 3C, 4). In contrast, LCT was induced in both duodenum and ileum, but expression was rather low. The ileal markers SLC10A2 and OSTB were highly restricted to differentiated organoids derived from ileum (Figs. 3D, 4). Notably, not all the location‐specific gene expression profiles in human intestinal organoids corresponded with the profiles in mice.

Collectively, these data show that WNT signaling inhibits expression of genes involved in the function of differentiated cells (LCT and SLC10A2) but not those that are involved in spatial patterning (GATA4). Furthermore, using the organoid technique we show that location‐specific stem cell fate is programmed within the stem cell niche, as the removal of WNT only induces expression of those genes that are not repressed by GATA4.

Discussion

It has been shown that WNT is expressed by Paneth cells in the crypt compartments of the small intestine to enable proliferation (and inhibit differentiation) in the crypt [28, 29]. WNT signaling activity tapers off toward the crypt‐villus junction, thereby reducing proliferation and inducing the differentiation into enterocytes, goblet, tuft, or enteroendocrine cells. Conversely, spatial compartmentalization of the small intestine along the cephalocaudal axis is regulated by expression gradients of transcription factors, such as GATA4 and CDX2 [12, 13]. These transcription factors are expressed in epithelial cells and have been shown to regulate expression of genes that are involved in functional properties of differentiated enterocytes.

As such, WNT expression is regulating differentiation along the crypt‐villus axis, whereas GATA4 is regulating regional specification and the linked differentiation along the cephalocaudal axis. To date, it was not known whether the expression of location‐specific genes is programmed in a hard‐wired fashion within the stem cells or if continuous signaling from mesenchymal cells is necessary to maintain the location‐specific identity.

Using the organoid culturing technique, we now show that pure epithelial cell cultures that originated from location‐specific stem cells are intrinsically programmed to differentiate into enterocytes with defined regional gene expression profiles. The organoid culture technique is based on the fact that Lgr5+ murine stem cells can be grown in laminin‐rich Matrigel in the presence of the WNT‐agonist and Lgr5‐ligand R‐spondin‐1, the BMP‐antagonist noggin, and EGF [2, 19]. In the presence of these extracellular signals, which are in vivo derived from Paneth cells and underlying mesenchyme, isolated crypts from the intestine can generate crypt‐villus structures, which can be maintained in long‐term culture [19]. Others have shown that mesenchymal cells are required for long‐term culture of intestinal stem cells using myofibroblasts as well as stem cell niche signaling molecules to establish long‐term cultures of intestine and colon [30].

Recently, it has been shown that in vivo, murine mesenchymal cells produce various WNT proteins [28], whereas Paneth cells produce EGF and WNT3A [31]. However, Paneth cells within organoids only produce WNT3A [28]. Endogenous Paneth cell‐derived WNT3A is sufficient to maintain mouse small intestinal organoids in the presence of EGF, noggin and Rspondin‐1 and allow generation of all differentiated epithelial cell types concomitant with stem cell self‐renewal [28]. The addition of WNT3A to these mouse organoid cultures interferes with intestinal differentiation and yields organoids that largely consist of undifferentiated progenitors ([19] and Supporting Information Fig. S1).

Previously, it has been shown that intestine‐specific deletion of Gata4 results in expression of Slc10a2 in the proximal intestine. As such, GATA4 is normally restricting the expression of Slc10a2 to the distal small intestine by repressing its expression in the proximal part [12]. Furthermore, reduction of GATA4 activity in the small intestine results in an induction of bile acid absorption in the proximal small intestine that is sufficient to correct bile acid malabsorption associated with ileocecal resected mice [32]. In contrast, CDX2 has been shown to be a master regulator in the posterior endoderm by repressing the foregut differentiation program in the posterior gut [13, 33].

In the human organoid system, we demonstrate that GATA4 is expressed both under expansion and differentiation conditions, while the expression of differentiation genes, such as SLC10A2, is only induced in DM. We can explain this by the fact that EM contains WNT3‐conditioned medium, which is inhibiting differentiation.

Recently, these data were supported by the finding that genetic regulatory networks and compartment‐specific stem cell differences direct regional diversity in the Drosophila midgut [34, 35]. Furthermore, the GATAe transcription factor, a master regulator of midgut development, was shown to control compartment‐specific gene expression along the length of the adult intestine, whereas gradients of WNT activity were shown to refine the location of boundaries [34].

As our key observation, we found differences in location‐specific gene expression patterns, even though all cells were exposed to the same extracellular factors. This suggests that the cephalocaudal programming in the small intestine is not dependent on mesenchymal or niche factors, but rather on the location‐specific expression of patterning genes such as Gata4.

Conclusions

Here, we show that adult stem cells of the small intestine are regionally specified in a cell‐autonomous fashion and retain their location‐specific gene expression profiles even after long‐term culture. Furthermore, our data are also of importance with respect to the differentiation of intestinal tissue from induced pluripotent stem cells [36, 37]. To date, there is a generally accepted tendency to define small intestinal tissue as CDX2+, but we here show that especially the expression of GATA4 will determine the functional fate of the established cultures. Moreover, in light of the potential future clinical application of organoids, either derived from iPS cells or intestinal adult stem cells [38, 39], our data imply that it is important to generate GATA4‐positive and GATA4‐negative cultures to regenerate all essential functions of the small intestine.

Acknowledgments

We thank Stieneke van den Brink, Hans Teunissen, and Joyce Blokker for practical guidance and providing noggin‐ and WNT3A‐conditioned media; Akifumi Ootani for providing the R‐spondin1‐producing cell line; the gastroenterologists of the Wilhelmina Children’s Hospital and the department of gastroenterology of the University Medical Center Utrecht for obtaining biopsies; Sylvia Brugman and Sabrina Roth for critically reading the manuscript. This work was supported by ZonMW (VIDI 91786322) to E.E.S.N., ZonMW (TAS 40–41400‐98–1108) to H.C. and E.E.S.N., and WKZ foundation (2011) to S.M. K.S. and C.L.W. contributed equally to this article.

Author Contributions

S.M.: conception and design, financial support, collection and/or assembly of data, data analysis and interpretation, and manuscript writing; K.S., C.L.W., M.M., and R.D.L.A.: collection and/or assembly of data and data analysis and interpretation; S.v.W.: collection and/or assembly of data; H.C.: financial support and material support; E.E.S.N.: conception and design, financial support, and final approval of manuscript.

Disclosure  of Potential Conflicts  of Interest

The authors indicate no potential conflicts of interest.

References

Author notes