Nutrient sensing by absorptive and secretory progenies of small intestinal stem cells

Small intestinal stem cells differentiate into several cell types transiently populating the villi. We used specialized organoid cultures each comprised of a single cell type to demonstrate that 1) differentiation seems required for nutrient sensing, 2) secretory goblet and Paneth cells along with enterocytes sense fructose, suggesting that sensing is acquired after differentiation is triggered but before divergence between absorptive and secretory lineages, and 3) forcibly dedifferentiated enterocytes exhibit fructose sensing and lifespan extension.

Abstract

Nutrient sensing triggers responses by the gut-brain axis modulating hormone release, feeding behavior and metabolism that become dysregulated in metabolic syndrome and some cancers. Except for absorptive enterocytes and secretory enteroendocrine cells, the ability of many intestinal cell types to sense nutrients is still unknown; hence we hypothesized that progenitor stem cells (intestinal stem cells, ISC) possess nutrient sensing ability inherited by progenies during differentiation. We directed via modulators of Wnt and Notch signaling differentiation of precursor mouse intestinal crypts into specialized organoids each containing ISC, enterocyte, goblet, or Paneth cells at relative proportions much higher than in situ as determined by mRNA expression and immunocytochemistry of cell type biomarkers. We identified nutrient sensing cell type(s) by increased expression of fructolytic genes in response to a fructose challenge. Organoids comprised primarily of enterocytes, Paneth, or goblet, but not ISC, cells responded specifically to fructose without affecting nonfructolytic genes. Sensing was independent of Wnt and Notch modulators and of glucose concentrations in the medium but required fructose absorption and metabolism. More mature enterocyte- and goblet-enriched organoids exhibited stronger fructose responses. Remarkably, enterocyte organoids, upon forced dedifferentiation to reacquire ISC characteristics, exhibited a markedly extended lifespan and retained fructose sensing ability, mimicking responses of some dedifferentiated cancer cells. Using an innovative approach, we discovered that nutrient sensing is likely repressed in progenitor ISCs then irreversibly derepressed during specification into sensing-competent absorptive or secretory lineages, the surprising capacity of Paneth and goblet cells to detect fructose, and the important role of differentiation in modulating nutrient sensing.

NEW & NOTEWORTHY Small intestinal stem cells differentiate into several cell types transiently populating the villi. We used specialized organoid cultures each comprised of a single cell type to demonstrate that 1) differentiation seems required for nutrient sensing, 2) secretory goblet and Paneth cells along with enterocytes sense fructose, suggesting that sensing is acquired after differentiation is triggered but before divergence between absorptive and secretory lineages, and 3) forcibly dedifferentiated enterocytes exhibit fructose sensing and lifespan extension.

the mammalian small intestinal mucosa is highly folded, with finger-like projections into the lumen called villi, which are surrounded by invaginations called crypts. The crypt contains proliferative intestinal stem cells (ISC), which divide into absorptive and secretory progenitor cells that migrate into the villus to become terminally differentiated, generally postmitotic cell types acting in a coordinated manner to digest and absorb nutrients, secrete mucus and hormones, as well as aid in immune and barrier defense. Nutrient digesting and absorbing enterocytes are the sole descendants of absorptive progenitors, and the roles make up ~85% of cells lining the intestinal villi, along with mucus-secreting goblet cells (~5% of total cell population), enteroendocrine (~1%), and Tuft (<1%) cells that are derived from secretory progenitors (5, 17, 35). Secretory Paneth cells (~5% of cells), which produce antimicrobial peptides that regulate the gut microbiota as well as growth factors that maintain the neighboring ISC (<1%), paradoxically migrate down to the crypts. The high proportion of enterocytes reflects the importance of maintaining a large surface area for optimal nutrition, and common diseases like gluten enteropathy that compromise enterocyte numbers often result in malnutrition.

The apical membranes of all cell types are exposed to luminal food contents, and the ability of the cells to perceive and adapt to these environmental cues is critical so that appropriate endocrine signals are released and that specific types as well as numbers of enzymes and transporters are synthesized to match the variety and fluctuating concentrations of dietary constituents. Nutrient sensing regulates, not only release of hormones that modulate metabolism as well as feeding behavior, but also rates of intestinal nutrient transport that ultimately determines blood levels and availability of nutrients to other organ systems (19). Although optimal nutrition requires epithelial cells to sense and respond appropriately to luminal signals, the roles of cell heterogeneity and differentiation in nutrient sensing have not yet been investigated.

Two general types of nutrient sensing in the small intestine have recently been identified, one typically involving a chemosensing G protein-coupled receptor in the luminal side of enteroendocrine cells that then releases endocrine and paracrine factors to initiate a physiological response, such as the glucose-induced release of incretins that regulates insulin secretion and modulates expression of the glucose transporter SGLT1 (Slc5a1) in neighboring enterocytes (23, 29). Second, a metabolic receptor involves sensing by enterocytes of nutrients that are metabolized in the cytosol after transport by facilitative carriers, such as the fructose transporter GLUT5 (Slc2a5) (19, 21), the dimeric, neutral, and cationic amino acid transporter rBAT/b^0,+AT (Slc3a1/slc7a9) (33), and the fatty acid translocase FAT (CD36) (22), to initiate intracellular signaling and physiological response. The transporters involved in sensing are often called transceptors.

During differentiation and migration from crypt to villus, the signaling system Delta-Notch nudges adjacent cells to increasingly either have high Notch expression and an absorptive destiny or low Notch expression and a secretory fate (14, 24). It is not known whether ISC differentiation affects nutrient sensing. Because enteroendocrine cells and enterocytes represent secretory and absorptive lineages, respectively, we tested the hypothesis that the progenitor ISCs already possess nutrient sensing ability that is inherited by their progenies.

Until recently, the heterogeneity of epithelial cells in vivo and the absence of a cell type-specific primary culture system precluded clear answers to these questions. Primary intestinal cells can now be cultured as organoids that faithfully replicate the full three-dimensional differentiation process and the cell type distribution of intestinal epithelia in vivo (8, 25). In this study, we, not only made typical organoids comprised of all cell types, but also directed the strict differentiation of these cultured organoids so that they contain specific cell types comprised mostly of ISC, enterocyte, goblet, or Paneth cells (38) at relative proportions much higher than in vivo, allowing us to examine whether ISCs and their progenies have similar capacities to adapt to dietary cues.

In this study, we defined nutrient sensing as the ability to increase mRNA expression of fructose-responsive genes when an organoid type is exposed to fructose in the incubation medium. We used GLUT5 and fructolytic enzymes as representative transceptor system because the small intestine is particularly sensitive to levels of dietary fructose in vivo, and exposure of intestinal cells to fructose during in vivo feeding or by ex vivo perfusion specifically upregulates fructose-responsive genes, including Slc2a5 or Glut5, glucose-6-phosphatase (G6Pase), as well as fructokinase [ketohexokinase (Khk)] (21, 31). Organoids are excellent tools for studying nutrient sensing because they survive for days and hence can be exposed to fructose for several hours or days; in contrast, isolated intestinal tissues and cells survive only ~1 h at 37°C in buffer solutions (11). These are also better models in studying regulation than the often nonphysiological transformed cell lines that have been used for decades in previous work (10).

Differentiated, postmitotic secretory and absorptive cells spend most of their lives exposed to the harsh luminal environment in vivo and may lose sensing ability over time. We next studied whether postdifferentiation age affects expression levels of differentiation biomarkers as well as the fructose response of differentiated cells by examining fructose sensing as a function of time after differentiation from ISC to enterocyte or goblet organoids. Finally, we forcibly dedifferentiated enterocyte organoids and examined the effect of dedifferentiation on the ability of the organoids to sense fructose, as dedifferentiated precancerous and cancerous cells often overexpress Glut5 (2–4). Our findings will increase our understanding of daughter cell fate determination and development of nutrient sensing, of the role of luminal nutrient signals in influencing daughter cell phenotype, and of conditions that can cause committed progenitor as well as differentiated cells to regain stem cell properties.

MATERIALS AND METHODS

Animals

All procedures conducted in this study were approved by the Institutional Animal Care and Use Committee, New Jersey Medical School (NJMS), Rutgers University. Small intestinal crypts were isolated from 6- to 8-wk-old wild-type (WT; C57BL/6; Taconic Laboratories, Hudson, NY) mice and genetically modified Glut5^−/− as well as Khk^−/− mice. Glut5^−/− were obtained from Jian Zuo (St. Jude’s Children’s Research Hospital), and Khk^−/− were obtained from Richard Johnson (University of Colorado).

Intestinal Organoid Isolation

Intestinal organoids were isolated as previously described (38). Briefly, crypts were harvested, processed, and suspended in 200 μl of crypt culture medium (CCM), which consisted of advanced DMEM/F12 (Life Technologies, Carlsbad, CA) supplemented with serum- free B27 (1:50; Life Technologies), N2 (1:100; Life Technologies), N-acetylcysteine (1 mM; Sigma-Aldrich, St. Louis, MO), recombinant murine epithelial growth factor (EGF, 50 ng/ml; Peprotech, Rocky Hill, NJ), Noggin (100 ng/ml; Peprotech), R-Spondin (500 ng/ml; Peprotech), Glutamax-I supplement (1:100; Life Technologies), penicillin/streptomycin (500 μg/ml; Life Technologies), and HEPES (10 μM; Life Technologies). Crypts were placed in Matrigel (density ~300 crypts per well) and allowed to solidify at 37°C. CCM (300 μl) was then added to each well. Cultures were incubated in a humidified incubator at 37°C, and CCM was changed daily.

Time Course Experiment

Typical organoids were cultured and then fed fructose for 24, 48, or 72 h to determine the optimal duration of 5 mM fructose incubation that resulted in the greatest response.

Directed Differentiation of Intestinal Organoids

We directed the differentiation of crypts into ISC, enterocyte, goblet, and Paneth cell organoids. Briefly, Wnt and Notch activators/inhibitors CHIR99021 (CHIR) (3 µM; Stemgent, Cambridge, MA), valproic acid (VPA, 2 mM; Tocris Bioscience, Bristol, UK), C59 (2 µM; Stemgent), and DAPT (10 µM; Tocris) were added in required combinations to the CCM (Fig. 1A) following Yin et al. (38). To stimulate fructose responsive genes, organoids were incubated for 24 h with 5 mM glucose (control), 5 mM or 10 mM fructose in their respective media.

Fig. 1.

Bright-field images during directed organoid differentiation of typical (TYP), enterocyte (ENT), stem cell (intestinal stem cell, ISC), and secretory goblet (GOB) organoids over time from crypt isolation (day 0) until day 7 (A). Only 1 image is shown at day 0, as all organoids came from the same crypt precursors of the same mouse. Thus directed differentiation of organoids arises only because of different treatments not because of differences in source of crypt precursors. Bar = 50 μm for bright-field images. L, lumen of the organoid. Day of experiment refers to time in days since precursor crypts were harvested, whereas day of differentiation refers to time in days since shift from ISC media to either enterocyte, goblet, or Paneth media. Where applicable, forced dedifferentiation would usually begin on day 6 of an experiment when enterocyte organoids are 3 days old. Expression profile of biomarkers in typical, enterocyte, ISC, and goblet organoids compared with that in freshly isolated intestinal crypts (CRY) and in intestinal tissue homogenates (HOM) (B). Expression of each marker was normalized to that in TYP organoids (1.0). The biomarker for stem cells, leucine-rich-repeat-containing G-protein coupled receptor 5 (Lgr5), was highly expressed in both TYP and ISC (i). Paneth cells were marked by lysozyme (Lyz) that was abundant in TYP, ISC, and CRY (ii). Enterocytes were marked by intestinal alkaline phosphatase (Alpi) expressed markedly in ENT and modestly in CRY (iii). The enteroendocrine biomarker, Chromogranin A (Chga), was only expressed in GOB (iv). Mucin 2 (Muc2) marked goblet cells and was also mainly expressed in GOB (v). Olfm4 was another biomarker of stem cells (vi). ^a,b,cP ≤ 0.05, n = 5. Because synthesis of Paneth organoids was not predictable, we had to examine nutrient sensing by this cell type in a separate series of experiments. VPA, valproic acid; CHIR, CHIR99021. D0a, day of crypt isolation when these are cultured with EGF, Noggin, and R-spondin (ENR) only to make typical organoids, or to make ISC organoids with ENR + CHIR (Wnt activator) + VPA (Notch activator). On day 3 of experiment, 1 set of ISC organoids are converted to enterocytes by adding C59 (Wnt inhibitor) + VPA, and another to goblet cells by adding C59 + DAPT (Notch inhibitor). For details, see methods.

Dedifferentiation

Crypts were isolated and differentiated into enterocyte organoids as described above. In the first experiment to evaluate the effect of dedifferentiation on fructose sensing, fully differentiated 3-day-old enterocyte organoids were dedifferentiated by placing them into ISC media containing 3 or 6 μM CHIR + 2 mM VPA for 24 h before an overnight 5 mM fructose challenge and harvest. In the second experiment determining organoid lifespan, enterocyte organoids were incubated in 6 μM CHIR + 2 mM VPA until death. Unfortunately, we were unable to dedifferentiate enterocyte organoids by modulating the NF-κB pathway as suggested (26).

Real-Time PCR

Total RNA was extracted from intestinal organoids (RNeasy Micro, Qiagen, Hilden, Germany). Although we were able to successfully differentiate organoids to specific cell types, the number of organoids produced was small, and we were able to harvest only ~100 organoids per well (~50–100 ng protein), an amount sufficient for RNA expression assays. Attempts to increase the number of precursor crypt populations and the volume of Matrigel resulted either in death of the organoids or in disconcordant development. Real-time PCR (Mx3000P; Stratagene, La Jolla, CA) was used to analyze cDNA using Maxima SYBR green (ThermoFisher Scientific, Grand Island, NY) and primers (Integrated DNA Technologies; Coralville, IA) (Table 1). All samples were standardized to β–actin expression (there were no changes in results if other standard housekeeping genes, Gapdh, Ef-1α, or 18S, were used).

Table 1.

Primer sequences of target genes

Gene Name	Sense (5′-3′) - Forward	Antisense (5′-3′) - Reverse
β-Actin (Actb)	TTGTTACCAACTGGGACGACATGG	CTGGGGTGTTGAAGGTCTCAAACA
Alkaline phosphatase, intestinal (Alpi)	CTCATCTCCAACATGGAC	TGCTTAGCACTTTCACGG
Chromogranin A (Chga)	CGATCCAGAAAGATGATGGTC	CGGAAGCCTCTGTCTTTCC
Glucose-6-phosphatase (G6pase)	CTTTCCCATCTGGTTCCATC	GCGTTGTCCAAACAGAATCC
Glucose transporter 2 (Glut2)	CTGCTCTTCTGTCCAGAAAGC	TGGTGACATCCTCAGTTCCTC
Glucose transporter 5 (Glut5)	AACTCTCCCTCAGAGTTCATGCAG	CAGGGCCCTTTTTTTGCCCATTT
Ketohexokinase (Khk)	TGTCCTTTCCTTGCTTGGAG	AAGTCAGCCACCAGGAAGTC
Leucine-rich repeat-containing G protein- coupled receptor 5 (Lgr5)	TGAGCGGGACCTTGAAGATT	AGGTGCTCACAGGGCTTGAA
Lysozyme (Lyz)	ATGGCTACCGTGGTGTCAAG	CGGTCTCCACGGTTGTAGTT
Mucin-2 (Muc2)	CTTCTGTGCCACCCTCGT	TTCGGGATCTGGCTTCTT
Olfactomedin 4 (Olfm4)	GCCACTTTCCAATTTCAC	GAGCCTCTTCTCATACAC
Sodium-glucose transporter 1 (Sglt1)	GAAGCTACTGCCCATGTTCCTCAT	ACTGGTGTGCCGCAGTATTTCTGA
Sucrase isomaltase (Si)	ATCCAGGTTCGAAGGAGAAGCACT	TTCGCTTGAATGCTGTGTGTTCCG

Immunocytochemistry

Organoids were grown on Nunc Laboratory-Tek II chamber slides (ThermoFisher) and stained for biomarkers as previously described. Organoids were fixed in 4% paraformaldehyde, then rinsed twice with PBS/glycine for 5 min. Cells were then permeabilized with PBST for 10 min, 24°C and then blocked with immunofluorescence (IF) buffer containing 0.1% BSA, 0.2% Triton X-100, 0.05% Tween-20, 0.05% sodium azide, and 10% serum for 1 h (24°C). Following blocking, primary antibody (Table 2) was added in IF buffer overnight at 24°C; then organoids were washed with IF buffer (3 times, 10 min each), followed by a 2-h incubation with fluorescence-conjugated secondary antibody plus a nucleic acid stain (TO-PRO-3). After the washing, chambers were detached from the microscope slide and then mounted with SlowFade Gold Antifade Mountant (ThermoFisher).

Table 2.

Primary antibodies used for immunofluorescence

Protein	Company	Cat No.	Dilution
Glucose-6-phosphatase (G6PASE)	Santa Cruz Biotechnology	sc-25840	1:200
Sucrase isomaltase (SI)	Santa Cruz Biotechnology	sc-27603	1:500
Lysozyme (LYZ)	Santa Cruz Biotechnology	sc-27958	1:100
Mucin-2 (MUC2)	Santa Cruz Biotechnology	sc-15334	1:200
Olfactomedin 4 (OLFM4)	Cell Signaling	D6Y5A	1:1,000
Chromogranin A (CHGA)	Santa Cruz Biotechnology	sc-13090	1:200

Statistical Analyses

Data are presented as means ± SE. To analyze the significance of cell type, genotype, organoid age, incubation time, fructose, and/or inhibitor concentrations on fructose response, a multiway ANOVA was used with Tukey’s multiple range tests. Differences were considered significant at P ≤ 0.05 (STATVIEW).

RESULTS

Directed Development of Intestinal Organoids into Distinct Cell Types

We cultured primary intestinal tissue organoids and directed their differentiation, using Notch and Wnt signaling modulators, into distinct cell lineages (38) (Fig. 1A). Bright-field images show the increasing size and morphological complexity of organoids treated with Wnt and Notch modulators to direct their development. Typical organoids are similar in morphology to previously described mouse organoids (30). Because the directed differentiation of Paneth cells was highly unpredictable, we had to examine nutrient sensing by this cell type in a separate series of experiments.

Expression of mRNA biomarkers relative to typical organoids.

Expression levels of specific mRNA biomarkers (38) of ISC, Paneth, enterocyte, enteroendocrine, and goblet cells were examined in organoids that were directed to be typical (control, undirected, containing all cell types) or to be comprised primarily of enterocyte, ISC, and goblet cells, along with the starting crypt material as well as homogenates of intestinal mucosa (Fig. 1B). Because primer properties of these biomarkers were different, they could not be compared within an organoid type. Instead, biomarker mRNA expression relative to that in typical cells was compared among organoids.

Levels of the ISC biomarkers Leucine-rich repeat-containing G protein-coupled receptor 5 (Lgr5) and Olfactomedin 4 (Olfm4) (36) were highest in organoids directed to be ISC, ~1.3-fold higher than those of typical cells and 20- to 500-fold higher than those of enterocyte, goblet, and homogenates (Fig. 1B, i and vi). Expression of the Paneth biomarker lysozyme (Lyz) was high in ISC organoids as well as in isolated crypts expected to contain large numbers of Paneth cells (Fig. 1Bii). Relative to typical organoids, expression of the enterocyte biomarker alkaline phosphatase (Alpi) was greatest in organoids directed to be enterocytes, with modest expression in isolated crypts (Fig. 1Biii). Levels of the enteroendocrine biomarker chromogranin A (Chga) and the goblet cell marker mucin 2 (Muc2) were greatest in goblet organoids (Fig. 1B, iv and v).

Expression of immunocytochemical biomarkers.

We also confirmed cell type distribution with immunofluorescence (Fig. 2), an approach that allowed us to compare the proportion of specific cell types in the same organoid. To our knowledge, this is the first set of comparisons of cell numbers in specialized organoids. An antibody against OLFM4 was used to mark ISC by immunocytochemistry because antibodies against biomarkers LGR5⁺ and ASCL2 did not work in ISC organoids. Likewise, sucrase isomaltase (SI) was used as immunocytochemical biomarker of enterocytes.

Fig. 2.

Immunofluorescence staining of proliferation and differentiation biomarkers in typical (A), ISC (B), enterocyte (C), and goblet (D) organoids. Nuclei are stained blue. Organoids were stained with stem cell marker OLFM4 (red), Paneth cell marker lysozyme (LYZ; green), enterocyte marker sucrase isomaltase (SI; red), goblet cell marker mucin 2 (MUC2; green), or enteroendocrine marker chromogranin-A (CHGA; green). Representative typical, ISC, enterocyte, and goblet organoids are shown at ×60 magnification. Bars = 25 μm. Occasional basolateral staining has been shown (personal observations) to be caused largely by autofluorescence of the Matrigel. Frequency of marked cells from at least 5 organoids from 3 mice (vi of each organoid type, ^a,b,c,dP ≤ 0.05). Immunocytochemistry of the fructose-responsive enzyme glucose-6-phosphatase (G6Pase) in enterocyte, ISC, and secretory goblet organoids is shown (E). All organoid types were analyzed on the 7th day of experiment, which for goblet, Paneth, and enterocyte organoids, is the 4th day after differentiation (see Fig. 1). The control panel is an image of an enterocyte organoid incubated without the primary antibody.

In typical organoids, each cell type was represented, as would be expected (Fig. 2A). There were large numbers of OLFM4- as well as SI-positive cells, and modest amounts of LYZ- and MUC2-positive, cells. OLFM4 is continuously synthesized and secreted by ISCs (12) and therefore accumulates within the closed lumen of the organoids where there are no pancreatic enzymes to digest exfoliated cells and secretions. Percentage of cell marker might exceed 100% (Fig. 2, Avi, Bvi, Cvi, and Dvi), as a single cell could coexpress different biomarkers.

In organoids directed to be mainly ISC, there was virtually no SI, MUC2, or CHGA immunofluorescence, suggesting that differentiated absorptive, goblet, and enteroendocrine cells, respectively, were generally absent (Fig. 2B). Instead, a large proportion of cells was positive for OLFM4 and accounted for the strong luminal staining, indicating accumulated secretions. Edu staining (not shown) confirms active proliferation of cells in ISC organoids. Lyz-positive Paneth cells required for in vivo ISC maintenance were also present in modest numbers (Fig. 2B, ii and vi). Thus ISC organoids contained dramatically more stem cells (~90%) compared with published in vivo levels (<1%) (5, 17, 35).

Organoids directed to be enterocytes were lined by large numbers of cells containing significant amounts of SI (Fig. 2Ciii). MUC2-expressing and, rarely, CHGA-expressing cells were represented, but there were virtually no OLFM4- and LYZ-positive cells (Fig. 2C, i, ii, and vi). The small amount of luminal staining is likely an undigested remnant of ISC secretion before ISCs were forcibly differentiated to ENT. Thus enterocyte organoids had similar or an even a greater number of enterocytes (~90%) than most published estimates of enterocyte number in vivo (~85%) (5, 17, 35).

Goblet organoids had high numbers of MUC2-expressing cells, low levels of CHGA- and SI-expressing cells, virtually no OLFM4-expressing ISCs, and almost no LYZ-expressing Paneth cells (Fig. 2D). The absence of Paneth cells by immunocytochemistry (Fig. 2D, ii and vi) confirmed the absence of Lyz mRNA in goblet organoids (Fig. 1Bii). Few enterocytes were present. Thus goblet organoids had a markedly greater number of goblet cells (~85%) than most published estimates of in vivo levels (4–8%) (5, 17, 35).

Results depicted in Figs. 1B and 2 are similar to the ex vivo distribution of biomarkers along the crypt-villus axis of cryosectioned intestines (37). Thus our procedure for directed differentiation worked in that ISC organoids were comprised mainly of OLFM4⁺ cells, enterocyte organoids mostly of SI⁺ cells, and goblet organoids primarily of MUC2⁺ cells at proportions dramatically higher than observed in vivo. We then interrogated these specialized organoids for their ability to sense nutrients.

We also examined by immunocytochemistry the expression of a representative, highly fructose-responsive gene, glucose-6-phosphatase (G6Pase) (21, 31), which has been immunolocalized in the cytosol (15). G6Pase is modestly expressed in ISC organoids, is highly expressed in ENT organoids, and is significantly expressed in GOB organoids (Fig. 2E). Luminal contents also seem to indicate immunoreactive G6Pase in exfoliated cells.

Role of Differentiation in Fructose Sensing

Although same-age differentiated organoids were observed to be of similar size and number, comparisons of enteroids enriched in different cell types were conducted at the same postdifferentiation duration to minimize confounding effects arising from potential differences in organoid size. Baseline expression of genes involved in fructose transport and metabolism was greatest in enterocyte and goblet organoids incubated in glucose (Fig. 3). Enterocyte and typical, but not ISC, organoids responded to an overnight fructose exposure. Much to our surprise, goblet organoids also displayed a robust response to fructose incubation. Highly significant cell type (P < 0.0001) and fructose (P < 0.0001 to 0.01) effects were observed for fructose-responsive genes Glut5, G6pase, and Khk (Fig. 3), as well as for Si. There was a significant interaction (P < 0.01), suggesting that cell type confounded the fructose effect on Glut5, G6Pase, and Khk, as indicated by the higher expression in and greater fructose-sensing capacity of goblet and enterocyte organoids. In contrast, there were no fructose effects on the expression of the Na⁺-dependent glucose transporter Sglt1 (Fig. 3iv), a gene that does not respond to fructose signals (19). Expression of the biomarker genes Alpi, Lgr5, Lyz, Muc2, and Chga was not affected (P > 0.50) by 5 or 10 mM fructose in typical, enterocyte, ISC, and goblet organoids (not shown).

Fig. 3.

Effect of cell type on the mRNA expression of classical fructose-responsive genes Glut5, G6Pase, and Khk. Typical and ISC organoids were incubated in their respective culture media for 6 days after generation from crypt precursors. Enterocyte and goblet organoids were incubated in their respective culture media for 3 days after directed differentiation from ISC precursors. Stem cells with very few Paneth cells constituted ISC organoids; enterocytes made up almost entirely enterocyte organoids, with few goblet and occasional enteroendocrine cells, whereas goblet organoids consisted primarily of goblet with some enteroendocrine cells. Media in each organoid type was supplemented with an additional 5 mM glucose, 5 mM fructose, or 10 mM fructose for 24 h. Levels of mRNA in all organoids were normalized to that in typical organoids exposed to 5 mM glucose (1.0). Only typical, enterocyte, and goblet organoids responded to supplemental fructose. Expression of the control gene Sglt1 did not vary with fructose. These results were confirmed 3 other times, using ISC, enterocyte, and goblet organoids from different mice. Analysis was by 2-way ANOVA, ^a,b,cP ≤ 0.05 within cell type, n = 3.

To determine whether secretory cells other than goblet can also sense fructose, we created CHIR+DAPT-treated organoids with mostly Paneth cells as indicated by >100- increases in Lyz expression over other types of organoids (Fig. 4i). Muc2 expression was >10-fold lower in Paneth compared with other organoids (Fig. 4ii). There was significant expression of Glut2 in all cell types (Fig. 4iii). By immunocytochemistry, Paneth organoids contained ~65% Paneth cells, a proportion much higher than in vivo (Fig. 4iv). Enteroendocrine and ISC cells constituted 15 and 5%, respectively, whereas remaining cell types each constituted <10%. Paneth organoids sensed fructose, which increased Glut5, G6Pase, Khk, and Si mRNA levels by 3- to 30-fold, without corresponding significant changes in Glut2 and Sglt1 mRNA levels (Fig. 4v).

Fig. 4.

Effect of fructose incubation of organoids enriched in Paneth cells on fructose-responsive genes Glut5, G6Pase, and Khk. Because Paneth cells also arise from secretory progenitors, we made (i to iv) then challenged Paneth cell-enriched organoids with 10 mM fructose (10 mM glucose as control, v). *P < 0.05, n = 3; for i-iii, ^a,b,c,dP < 0.05). Fructose induced marked increases in expression of Glut5, G6Pase, Khk, and Si genes. vi: Challenging ISC-enriched organoids with 20 mM fructose to increase the concentration gradient still had no effect on glucose-responsive genes.

The absence of fructose sensing by ISC organoids was likely not due to reduced relative mRNA expression of the basolateral fructose and glucose transporter Glut2 because mRNA levels of this gene were also similarly low in fructose-responsive typical and Paneth organoids (see Fig. 4, iii and v). Because passive flux into cells is a function of transporter number and chemical gradient, we compensated for a potentially low number of GLUT2 by incubating ISC organoids in fourfold greater concentrations of fructose (20 mM) compared with those used in Fig. 3i. There was, however, still no fructose response from ISC organoids (Fig. 4vi), suggesting that the absence of a response was not due to decreased fructose transport.

To determine the possibility that stem cells could sense but not respond to fructose or whether it could retain dietary information acquired before differentiation, we incubated ISC organoids in fructose for 3 days and then forcibly differentiated them to either enterocyte or goblet organoids overnight without fructose. There was still no significant (P > 0.50) difference in relative percentage of expression of fructose-responsive genes in glucose- and fructose-incubated organoids [in enterocytes (means ± SE), 100 ± 30 and 155 ± 25% for Glut5, 100 ± 5 and 95 ± 20% for Khk, and 100 ± 40 and 70 ± 4% for Glut2; in goblet, 100 ± 15 and 75 ± 30% for Glut5, 100 ± 20 and 90 ± 15% for Khk, and 100 ± 15 and 110 ± 25% for Glut2]. Because fructose-enhanced levels of enterocyte GLUT5 mRNA can be retained for hours after fructose consumption (6), the absence of a fructose response here was likely not due to withdrawal of the fructose signal. Thus stem cells probably did not sense fructose and did not transmit dietary information to their progeny.

Effect of fructose concentration and incubation duration.

When 10 mM fructose was added to the media, there tended to be higher responses from fructose-sensing typical, enterocyte, and goblet organoids, compared with those from 5 mM fructose (Fig. 3). This response was not due to increased osmolality because 5 mM fructose but not 5 mM glucose added to the media induced a fructose response. Increasing the duration of incubation from 1 to 3 days had little or modest effects on the magnitude of response of typical organoids to 5 mM fructose: 100 ± 5% for glucose-incubated organoids, then 225 ± 40, 280 ± 20, and 240 ± 10% for relative Glut5 mRNA expression in 1-, 2-, and 3-day fructose-incubated organoids. In contrast, relative expression for the control gene Lgr5 was 100 ± 10% for glucose-incubated organoids, then 105 ± 25, 95 ± 8, and 130 ± 10% for 1-, 2-, and 3-day fructose-incubated organoids. The test concentrations reflect a compromise between postprandial blood (~1 mM) and luminal (~30 mM) fructose concentrations (13, 20).

No effect of decreased glucose concentration in the incubation media.

Organoids are bathed in media already containing 17.5 mM glucose before the fructose challenge. To determine the confounding effect of high glucose levels in the medium on the fructose response, we cultured typical and enterocyte organoids in low (11.5 mM) glucose media for either the entire experimental duration [(see Fig. 1A) culture time plus incubation time in additional 5 mM glucose (control) or fructose, or ~6 days total] or only during the fructose “feeding” period of 24 h before harvest. Reducing glucose concentration in the medium over 6 days (not shown) or for 24 h (Fig. 5, i and ii) had no effect on fructose-induced Glut5 and G6Pase expression by typical and enterocyte organoids. For both genes, there were strong cell type (P < 0.0001 by 3-way ANOVA) and fructose (P < 0.005) effects but no effect of incubation media (P > 0.5). Reductions in medium glucose concentration had no effect on expression of non-fructose-responsive genes, like the enterocyte biomarker Alpi (P > 0.5) (Fig. 5iii). Thus organoids in general can be cultured at glucose concentrations lower than presently prescribed. Moreover, changes in glucose concentrations in the organoid medium do not confound findings related to the fructose response.

Fig. 5.

Effect of glucose concentration in the incubation medium on fructose-responsive genes. The media in typical and enterocyte organoids was changed from 17.5 to 11.5 mM glucose (control: 17.5 to 17.5 mM) at the same time that fructose was added 24 h before the termination of the experiment. We added 5 mM glucose or 5 mM or 10 mM fructose to organoids in normal 17.5 mM or low 11.5 mM glucose media (results separated by a dashed line). Although there was an expected effect of fructose on expression of Glut5 (i) and G6Pase (ii), there was no effect of glucose concentration in the media. Alpi (iii) expression was independent of media and sugar in both typical and enterocyte organoids. There was also an effect of cell type on the expression of these genes that are highly expressed in enterocytes in vivo. Expression was normalized to that in 5 mM glucose of the control 17.5 to 17.5 mM experiment (1.0). Analysis was by 3-way ANOVA (see text for details), n = 3.

No effect of Wnt and Notch modulators on fructose sensing.

To steer the development of precursor crypts to ISC organoids, these had to be incubated with Wnt and Notch activators CHIR and VPA (Fig. 1A), respectively, which might have inhibited the fructose response. To confirm that the absence of fructose sensing in ISC organoids was independent of CHIR, CHIR was removed from ISC media when fructose was added. Without CHIR, ISC organoids still did not respond to fructose (relative G6Pase expression: 100 ± 10% for glucose and 95 ± 2% for fructose). Because enterocyte organoids already respond to fructose, CHIR was added to the enterocyte media when fructose was added. Enterocyte organoids still continued to respond to fructose (relative G6Pase expression: 100 ± 5% for glucose and 400 ± 50% for fructose). Similar ISC and enterocyte results were obtained for Glut5. Thus CHIR had no effect on fructose sensing. Moreover, SGLT1 expression still did not respond to fructose with or without CHIR (not shown).

We could not remove VPA without affecting the generation of ISC organoids. However, the absence of fructose sensing by ISC cannot be due to VPA because both enterocyte organoids incubated with VPA and goblet organoids grown without VPA responded to fructose (Fig. 3). Taken together, these results clearly indicate that enterocytes, Paneth, and goblet, but not stem, cells can respond to fructose, and that response is not confounded by Wnt and Notch modulators.

Role of Fructose Transport and Metabolism in Fructose Sensing

We proceeded to determine whether the fructose response by organoids requires GLUT5-mediated fructose transport and KHK-mediated fructose metabolism observed in vivo (19). The morphology exhibited by typical organoids from Glut5^−/− and Khk^−/− mice was similar to that in WT mice, indicating that deletion of genes involved in fructose transport and metabolism did not affect organoid formation.

Although fructose clearly induced Glut5, G6Pase, Khk, and Si in typical organoids obtained from WT mice, it failed to induce a fructose response in Glut5^−/− mice (Fig. 6), suggesting that GLUT5 expression and function were required for the upregulation of fructose-responsive genes. An interesting exception was G6pase. Because intestinal G6pase was typically the most sensitive gene to changes in fructose levels (31), perhaps a small amount of intracellular fructose entering via basolateral GLUT2 was sufficient to upregulate this gene.

Fig. 6.

Effect of Glut5 and Khk deletion on fructose sensing by typical organoids. Typical organoids from wild-type (WT), Glut5−/−, and Khk−/− mice were cultured for 6 days then incubated in media supplemented with 5 mM glucose, 5 mM fructose, or 10 mM fructose for 24 h on the final day of culture. There was no effect of genotype on expression of undeleted genes. Levels of mRNA in 5 and 10 mM fructose were normalized to that in 5 mM glucose (1.0) for each genotype. Except for G6Pase expression, which did not require glucose transporter 5 (GLUT5), fructose transport and metabolism were clearly necessary for fructose sensing. ^a,b,cP ≤ 0.05 within genotype, n = 3.

Fructose also did not upregulate Glut5, G6Pase, Khk, and Si in organoids from Khk^−/− mice (Fig. 6), suggesting that fructose metabolism was required for fructose sensing. Expression of Glut2, Sglt1 (Fig. 6), and all biomarker genes Alpi, Lgr5, Muc2, and Lyz (not shown) remained independent of fructose in all mice, suggesting that their expression was not affected by fructose and was also not affected by the deletion of Glut5 and Khk. Thus the effect of Glut5 and Khk deletion is specific for fructose-responsive genes.

Fructose Response over the Lifespan of Enterocyte and Goblet Organoids

When differentiated secretory and absorptive cells migrate to the villus in vivo, they die after arriving at the tip where they are exfoliated. The maximum lifetime of enterocyte organoids is 5.0 ± 0.2 days, whereas that of goblet organoids is 4.5 ± 0.2 days. In contrast, ISC and typical organoids, after reaching large sizes with numerous buds, can be dispersed and reseeded into new Matrigel and thus are able to theoretically live (be passaged) indefinitely.

When 1- to 3-day-old goblet organoids were challenged with 5 mM fructose overnight, baseline expression of genes associated with sugar transport and digestion (Glut5 and Sglt1, P < 0.001–0.01 by 2-way ANOVA, Fig. 7A) as well as Si increased with age of goblet cells. Moreover, fructose increased the expression of the typical fructose-responsive genes [Glut5 and G6pase (P < 0.001–0.09)] as well as Khk regardless of age. Expression of ISC and Paneth biomarkers Lgr5 (Fig. 7A) and Lyz decreased dramatically with age of goblet organoids. In contrast, expression of goblet and enteroendocrine biomarkers Muc2 and Chga, respectively, increased with age. These results suggest that goblet organoids continued to differentiate and sense fructose over time.

Fig. 7.

Effect of organoid age on expression of genes in goblet (A) and enterocyte (B) organoids. Organoids were cultured in their respective media supplemented with 5 mM fructose 1–3 days after differentiation from ISC to goblet cells or 2 and 3 days after differentiation from ISC to enterocytes. Gene expression was normalized to glucose-incubated, 2-day-old organoids for enterocytes, or to 1-day-old organoids for goblet. Expression of most genes and the magnitude of the fructose response tended to increase with organoid age. Analysis was by 2-way ANOVA, n = 3. The enterocyte series of experiments was repeated in a different batch of enterocyte organoids from different mice, with similar results.

We challenged 2- and 3-day-old enterocyte organoids with 5 mM fructose overnight. Age and fructose increased the expression of Glut5, G6pase, and Si (Fig. 7B) (P < 0.0001–0.05, by 2-way ANOVA). These results suggest that enterocyte organoids also continued to differentiate and to sense fructose with increasing age. Expression of ISC and Paneth cell biomarkers, Lgr5 and Lyz, respectively, decreased with enterocyte age (Fig. 7B). Relative Muc2 and Chga expression did not change with age and fructose, suggesting that numbers of goblet and enteroendocrine cells, respectively, did not change with age of enterocyte organoids.

Effect of Dedifferentiation on Lifespan of and Fructose Sensing by Enterocytes

Here we triggered dedifferentiation of enterocyte organoids by activating both Wnt and Notch pathways after expression of enterocyte biomarkers had reached a peak (~3–4 days). Replacement of C59 with 3 or 6 μM CHIR triggered dedifferentiation of 3-day-old enterocyte organoids within 36 h; the greater the concentration of, and the longer the incubation in, CHIR, the greater the degree of dedifferentiation. CHIR (6 μM) extended organoid lifespan more than twofold, to 11 days after addition of C59 + VPA.

Forced enterocyte dedifferentiation affected the expression of all genes. mRNA levels of fructose-metabolizing genes (Fig. 8) decreased dramatically with dedifferentiation (P < 0.0001–0.01). Muc2 expression was already relatively low in enterocyte organoids but decreased further with dedifferentiation (P < 0.0001–0.01). In fact, Muc2 could not be detected anymore in enterocyte organoids 6 days after dedifferentiation. In contrast, expression of the classical ISC biomarker Lgr5 and the Paneth biomarker Lyz increased markedly. Although Lgr5 expression increased ~100-fold with dedifferentiation, it was still ~20-fold less than that in ISC organoids (Fig. 8).

Fig. 8.

Effect of CHIR-induced dedifferentiation of enterocyte organoids on expression of genes involved in the fructose response and in marking cell types. From ISC precursors, enterocyte organoids were cultured for 3 days with C59 +VPA to full differentiation as indicated by peak expression of fructose-responsive genes. To force dedifferentiation, the Wnt inhibitor C59 was removed from the medium, and enterocyte organoids were then exposed for 36 h to 3 (dENT1) or 6 μM CHIR (dENT2) or for 6 days to 6 μM CHIR (dENT3). ENT, dENT1, dENT2, dENT3, and ISC organoids from the same mouse were then challenged with 5 mM fructose in the last 12 h before harvest. Gene expression was normalized to glucose-incubated, enterocyte organoids. Although expression of fructose-responsive genes decreased markedly with CHIR treatment, Glut5 and G6Pase expression still responded to fructose incubation in all dedifferentiated enterocyte organoids. Lgr5⁺ levels increased almost 100-fold with CHIR-induced dedifferentiation but never reached levels observed in ISC organoids. Lyz and Muc2 expression were also affected by dedifferentiation (n = 3 per treatment).

Surprisingly, fructose incubation still triggered remarkable increases in Glut5 and G6Pase expression in all dedifferentiated cells, even as total Glut5 and G6Pase expression decreased (Fig. 8). For Glut5 and G6Pase, there was a significant interaction between dedifferentiation treatment and fructose sensing, suggesting that the fructose response depended on the magnitude of dedifferentiation. Expression of Lgr5, Lyz, and Muc2 (Fig. 8) as well as Glut2 did not change with fructose incubation. Although Wnt activation by CHIR generally induced ISC characteristics, dedifferentiated enterocyte organoids retained some enterocyte features, such as nutrient sensing, and acquired unusual characteristics, such as increased Alpi expression. Alpi is a biomarker of differentiated enterocytes (Fig. 1B), and its expression should decrease with dedifferentiation. Instead, Alpi expression increased proportionally with levels of dedifferentiation, culminating in relative percentage expression from just 100 ± 10% for glucose and 90 ± 5% for fructose in differentiated enterocyte organoids, to 5,500 ± 450% for glucose and 8,500 ± 400% for fructose, in dedifferentiated organoids.

DISCUSSION

The main findings in our study are that 1) differentiation is required for intestinal cells to sense nutrient signals; 2) secretory goblet and Paneth cells along with enterocytes can sense fructose, and thus fructose sensing is likely acquired after ISC differentiation is triggered but before divergence between absorptive and secretory lineages; and 3) forced dedifferentiation of mature enterocyte organoids results in retention of fructose sensing, lifespan extension, and reacquisition of some stem cell characteristics. These findings increase our understanding of nutrient sensing during normal and dysregulated gut cell development. Our secondary findings that intestinal and secretory organoids have finite lifetimes and that organoids can flourish in media containing 33% less glucose are also important because these expand our ability to utilize organoids in various experimental approaches to understand intestinal homeostasis and function.

Nutrient Sensing and Differentiation State

We have previously suggested that GLUT5 is probably acting as a metabolic transceptor taking up fructose, leading to KHK-dependent, phosphoinositide 3-kinase-involving changes in gene expression of fructose-responsive genes in enterocytes (7, 19). Because the basolateral membrane of organoids faces the incubation medium, the requirement of apical GLUT5 for fructose sensing was unexpected, as basolateral GLUT2 would initially take up fructose. However, glucose and dipeptide uptake were each markedly reduced in typical organoids from Sglt1−/− and Pept1−/− mice lacking apical glucose and dipeptide transporters, respectively (39), suggesting that apical transporters are actually required for uptake by organoids even though only the basolateral membrane faces the incubation medium. In our study, perhaps the fructose in the medium entered the organoid lumen via a highly permeable paracellular pathway (39) so that apical GLUT5 would be required to absorb fructose into the cells of typical organoids. Taken together, these findings suggest that sensing by typical organoids requires the apical transport via GLUT5 of fructose from the lumen followed by its KHK-dependent metabolism in the cytosol. Thus one reason for the failure of ISC to respond to fructose may be due to the low expression of Glut5 and Khk, probably arising from the few non-ISC cells in the ISC organoids. It likely cannot be due to low Glut2 expression in ISC organoids because Paneth and typical organoids that have similarly low Glut2 levels respond to fructose and because increasing the fructose concentration gradient to compensate for low transporter abundance still failed to induce GLUT5. In contrast, enterocyte, goblet, and Paneth organoids have robust responses to fructose, as these express high levels of Glut5 and Khk. Typical organoids with mixed cell populations are able to respond because of the presence of enterocyte, goblet, and Paneth cells. Dedifferentiated enterocytes sense fructose, as they also have higher levels of Glut5 and Khk mRNA.

Differentiation biomarkers like Si and Alpi are typically regulated by both DNA methylation and histone modifications, whereas most other genes are regulated by histone modifications (37). Thus CpG dinucleotides in the proximal promoter of an intestinal amino acid transporter were highly methylated in the crypt, suggesting that its transcription was repressed, but were fully demethylated in the villus, suggesting that its transcription was permitted (34). Expression of fructose-sensing and unidentified fructose-regulating genes in ISC organoids may likely be repressed but is derepressed in differentiated progenies.

Crypt Villus Site of Nutrient Sensing

The surprising, fructose-induced response of goblet and Paneth cell organoids, in combination with findings that enterocyte organoids can also respond to fructose, strongly suggests that fructose sensing may be conferred when ISC progenies leave the crypt and enter the transit-amplifying zone, whose highly adaptable cells have no well-accepted, specific biomarkers. Here, secretory and absorptive progenitor cells divide every 12–16 h to finally give rise to terminally differentiated cell types while simultaneously migrating upward, toward the villus column (1).

At least two and most likely all three major secretory cell types can sense fructose. The main secretory product of goblet cells, mucin, consists of several oligosaccharide side chains with fructose moieties. Goblet cells may be sensitive to altered fructose metabolism, as fructose malabsorption has been linked with respiratory and intestinal mucus hypersecretion (9). Likewise, enteroendocrine cells seem able to sense luminal fructose via taste receptors and respond by releasing glucagon-like peptides in vivo (28). It is unclear why Paneth cells retain fructose-sensing ability when these migrate down into the crypt, but perhaps this ability, once obtained during divergence to absorptive and secretory lineages, is likely conserved and irreversibly retained.

Fructose-induced GLUT5 regulation in vivo involves de novo mRNA and protein synthesis in mature cells lining the villus (13), but these studies could not specifically examine effects of cell age on fructose sensing. Our results suggest that postdifferentiation age increases the expression of genes coding for proteins that carry out the absorptive and digestive functions of the intestine, including genes that seem to enhance the cellular response to fructose.

Hallmarks of Forced Dedifferentiation of Enterocyte Organoids

Dedifferentiated enterocyte organoids do not seem to attain the hallmarks that define ISC organoids containing mostly undifferentiated cells. For example, the lifespan of dedifferentiated enterocyte organoids, though dramatically longer than that of normal enterocytes, is still finite, unlike that of ISC. This suggests that Wnt activation via CHIR is not the sole factor required for transforming enterocyte into ISC organoids and that Wnt initiates but does not complete dedifferentiation. Although Wnt activation is known to lead to colonic dysplasia (26), small intestinal dedifferentiation may involve the canonical Wnt pathway inhibited by β-catenin and initiate intestinal tumorigenesis. Alpi, which is low in ISC but paradoxically increases with enterocyte dedifferentiation, can be induced by histone deacetylase inhibitors (27). Thus Alpi may mark, not only differentiated, but also dedifferentiated enterocytes.

It is unclear why dedifferentiated cells can sense fructose, unlike ISC. However, it is interesting to note that fructose and its transceptor GLUT5 have been associated with colorectal (2, 4), small intestinal (32), breast, and pancreatic (16, 18) cancers, whereas inhibition of GLUT5 expression in breast cancer repressed proliferation of cancer cells (3).

Limitations and Future Work

Although there are mixed cell populations in the organoids that could potentially confound conclusions, each organoid type representing a specific cell contains a much greater proportion of that cell type compared with in situ populations in the small intestine. The >10-fold increase in proportion of stem cells in ISC organoids and goblet or Paneth cells in secretory organoids afforded us a method of examining nutrient sensing in various cell types that would otherwise not be possible in vivo and even in vitro, as isolated intestinal cells have extremely short lifespans (~1 h) (11). The possibility that enteroendocrine cells representing ~10–15% of cells in goblet and Paneth organoids were the sole responders to fructose cannot be eliminated but is highly unlikely because it would require these cells to increase expression of fructose-responsive genes by >100-fold, for the average fructose response of the entire organoid to increase by 10- to 50-fold. Likewise, although fructose-sensing Paneth cells are present in ISC organoids, their numbers are too small to significantly affect average expression of fructose-responsive genes. Thus fructose sensing by enterocyte organoids is most likely due to enterocytes, by goblet organoids to goblet cells, and by Paneth organoids to Paneth cells.

Because organoids develop with the lumen inside and GLUT5 is apical, it is not possible to examine GLUT5-mediated fructose uptake in organoids. We were able to interrogate fructose sensing by these organoids because GLUT2, though unresponsive to fructose levels, is basolateral and thus can transport fructose into the cell, where it can be sensed.

Taken together, the data strongly indicate that fructose sensing is likely conferred during early differentiation, certainly before divergence of secretory and absorptive lineages (Fig. 9). Thus fructose sensing is likely repressed in progenitor stem cells and is derepressed in their progenies. Ongoing epigenetic (31) and RNA sequencing studies of specialized organoids will attempt to identify the developmental mechanisms regulating fructose sensing.

Fig. 9.

Model depicting the role of differentiation in nutrient sensing. Lgr5⁺-ISC cells cannot sense fructose perhaps because of low GLUT5 and ketohexokinase (KHK) levels. During differentiation and migration from crypt to villus, Notch-dependent lateral inhibition and Delta-Notch feedback loops nudge adjacent cells to increasingly either have high Notch expression and an absorptive destiny or low Notch expression and a secretory fate. All differentiated cells in normal epithelia express high levels of GLUT5 and KHK and are able to respond to fructose, suggesting that fructose sensing is acquired after ISC cells leave the crypt bottom, perhaps during Delta-Notch-mediated differentiation occurring in transit-amplifying (TA) cells.

GRANTS

This work was supported by NSF Grants IOS-1121049 (R. P. Ferraris) and 1456673 (R. P. Ferraris). R. P. Ferraris also received support from NIH Grant R01-DK-102934 (N. Gao).

DISCLOSURES

No conflicts of interest, financial or otherwise, are declared by the authors.

AUTHOR CONTRIBUTIONS

K.K. and S.C.P. performed experiments; K.K., S.C.P., S.Y., N.G., and R.P.F. analyzed data; K.K., S.C.P., S.Y., N.G., and R.P.F. interpreted results of experiments; K.K. and S.C.P. prepared figures; K.K. and S.C.P. drafted manuscript; K.K., S.C.P., S.Y., N.G., and R.P.F. edited and revised manuscript; K.K., S.C.P., S.Y., N.G., and R.P.F. approved final version of manuscript; R.P.F. conceived and designed research.