Impact of Diet-Induced Obesity on Intestinal Stem Cells: Hyperproliferation but Impaired Intrinsic Function That Requires Insulin/IGF1

Abstract

Nutrient intake regulates intestinal epithelial mass and crypt proliferation. Recent findings in model organisms and rodents indicate nutrient restriction impacts intestinal stem cells (ISC). Little is known about the impact of diet-induced obesity (DIO), a model of excess nutrient intake on ISC. We used a Sox9-EGFP reporter mouse to test the hypothesis that an adaptive response to DIO or associated hyperinsulinemia involves expansion and hyperproliferation of ISC. The Sox9-EGFP reporter mouse allows study and isolation of ISC, progenitors, and differentiated lineages based on different Sox9-EGFP expression levels. Sox9-EGFP mice were fed a high-fat diet for 20 weeks to induce DIO and compared with littermates fed low-fat rodent chow. Histology, fluorescence activated cell sorting, and mRNA analyses measured impact of DIO on jejunal crypt-villus morphometry, numbers, and proliferation of different Sox9-EGFP cell populations and gene expression. An in vitro culture assay directly assessed functional capacity of isolated ISC. DIO mice exhibited significant increases in body weight, plasma glucose, insulin, and insulin-like growth factor 1 (IGF1) levels and intestinal Igf1 mRNA. DIO mice had increased villus height and crypt density but decreased intestinal length and decreased numbers of Paneth and goblet cells. In vivo, DIO resulted in a selective expansion of Sox9-EGFP^Low ISC and percentage of ISC in S-phase. ISC expansion significantly correlated with plasma insulin levels. In vitro, isolated ISC from DIO mice formed fewer enteroids in standard 3D Matrigel culture compared to controls, indicating impaired ISC function. This decreased enteroid formation in isolated ISC from DIO mice was rescued by exogenous insulin, IGF1, or both. We conclude that DIO induces specific increases in ISC and ISC hyperproliferation in vivo. However, isolated ISC from DIO mice have impaired intrinsic survival and growth in vitro that can be rescued by exogenous insulin or IGF1.

The functional consequences of obesity have been extensively studied in liver, skeletal muscle, and adipose tissue, but much less is known about the effect of obesity on the intestinal epithelium, the initial site of nutrient absorption. The highly proliferative small intestinal epithelium is composed of crypts, containing proliferating cells, terminally differentiated Paneth cells, and some goblet and enteroendocrine cells (EEC), and villi composed of primarily postmitotic differentiated enterocytes but also goblet and EEC. The small intestinal epithelium is renewed every 3 to 7 days depending on the species and region. Constant renewal involves proliferation of intestinal stem cells (ISC) that reside at the crypt base. ISC give rise to more actively dividing progenitors, also termed transit-amplifying cells, that differentiate into postmitotic lineages as they exit the crypts, or migrate to the crypt base (1–3). Intestinal epithelial homeostasis is dependent on a tightly regulated balance between ISC and progenitor proliferation, differentiation, and the constant loss of differentiated cells at the villus tip.

The small intestinal epithelium is highly responsive to changes in nutrient intake or exposure to luminal nutrient. In rodents, fasting or total parenteral nutrition leads to rapid reductions in small intestinal epithelial mass, associated with reduced proliferation in the crypts and increased apoptosis in crypts and villi (4–9). This is a logical physiological adaptation to a reduced need for nutrient absorption. In duodenum and jejunum and to a lesser extent ileum, refeeding can rapidly reverse the fasting-induced atrophy of the epithelium. Until recently, it was not possible to directly assess impact of nutrient status on ISC. Since landmark studies in 2007, Lgr5 and multiple other proteins have been identified as biomarkers of actively cycling ISC (also termed crypt based columnar cells) (10, 11). Development of transgenic reporter mice expressing fluorescent proteins downstream of the promoters driving ISC biomarker expression has permitted direct evaluation of ISC in vivo (10, 12), and isolation and assessment of ISC intrinsic function in vitro. In three-dimensional (3D) culture systems, ISC develop into spherical structures termed enterospheres that are composed of multiple cells, reflecting ISC survival and proliferation. With increased time in culture, enterospheres grow and form more complex structures termed enteroids that show a lumen, crypt buds, and contain ISC and all differentiated lineages (13). Enterosphere and enteroid yield from isolated ISC is a useful measure of ISC survival and growth capacity. A recent study using Lgr5 reporter mice demonstrated that long-term calorie restriction (CR) reduced villus height and proliferation of progenitors but increased both numbers and proliferation of ISC (14). CR also enhanced the ability of isolated ISC to survive, grow, and yield enteroids (14). The ability of CR to enhance ISC number and function was linked to diminished mTORC signaling in Paneth cells, neighboring niche cells that provide trophic support to ISC (13). Other studies performed in Drosophila demonstrated that fasting decreased ISC number that was restored upon refeeding (15, 16), strengthening the concept that ISC respond and adapt to altered nutrient availability. Compared with fasting, the impact of overnutrition as seen in diet-induced obesity (DIO) has not been as extensively studied. Depending on the model and duration of obesogenic diet, DIO has been linked to altered crypt-villus homeostasis, particularly increased villus height but variable effects on crypt cell proliferation (17–19). Importantly, the impact of DIO specifically on ISC is not defined.

In this study, we sought to define the effects of DIO, specifically on ISC using the Sox9-EGFP reporter mouse model. In the intestine of this model, different expression levels of the Sox9-EGFP transgene mark different intestinal epithelial cell types (12, 20). The highest expression levels of Sox9-EGFP (Sox9-EGFP^High) are found in cells coexpressing or enriched for EEC markers and Bmi1, Hopx, and Dcamkl1, markers linked to a reserve population of cells that can function as ISC in some situations (21–24). Lower Sox9-EGFP expression marks cells that have been termed Sox9-EGFP^Low ISC because they are highly enriched for Lgr5 and other biomarkers of actively cycling ISC and are capable of self-renewal and multipotency in vivo and in vitro (12, 20, 24, 25). In healthy, ad libitum fed Sox9-EGFP mice, only Sox9-EGFP^Low cells can survive and form enteroids in 3D culture. Sublow EGFP expression (Sox9-EGFP^Sublow) mark progenitors, and cells negative for EGFP (Sox9-EGFP^Negative) correspond to enterocytes, Paneth, and goblet cells based on enrichment of markers of these differentiated cells (12, 20, 24). This model therefore permits us to test the impact of dietary interventions on Sox9-EGFP^Low ISC and also on the other cell populations expressing different levels of Sox9-EGFP. Our study tested the hypothesis that DIO promotes ISC proliferation and expansion to increase small intestinal epithelial mass.

It is well established that changes in crypt cell proliferation and intestinal mass induced by fasting and refeeding correlate with changes in circulating insulin-like growth factor 1 (IGF1) and local intestinal expression of Igf1 that derives from the pericryptal mesenchyme (7). Enterotrophic effects of IGF1 have been demonstrated and IGF1 is able to prevent atrophy of the small intestinal epithelium induced by total parenteral nutrition (26). Since DIO due to long-term high-fat diet (HFD) promotes insulin resistance associated with elevated circulating insulin (27) and increased free IGF1 in the circulation (28, 29), we examined whether DIO-induced changes in ISC correlated with plasma insulin or IGF1 and whether ISC from DIO mice showed altered responsiveness to insulin or IGF1 in vitro.

Materials and Methods

Animals/diet

Sox9-EGFP mice contain a BAC transgene with approximately 226.5kb of Sox9 genomic regulatory region that drive EGFP expression (12, 20) and are maintained as heterozygotes on an outbred CD-1 background. Genotyping was performed as described in (12). Adult male and female Sox9-EGFP mice (7–10 weeks old) were randomly divided to receive either low-fat chow (14% kcal from fat; Prolab RMH3000) or HFD (45% kcal from fat; Research Diets D12451) ad libitum. Published literature on inbred C57Bl/6J mice demonstrated frank obesity and insulin resistance after 16 weeks on the 45% kcal from fat HFD (27). To ensure obesity and hyperinsulinemia, Sox9-EGFP mice were maintained on the diet for 20 weeks. Body weight was measured weekly. Body composition was assessed by magnetic resonance imaging (EchoMRI, Houston, TX) after 20 weeks on diet. All mice were killed between 9:30–11:00 AM with a lethal dose of Nembutal (150 μg/g body weight). The Institutional Animal Care and Use Committee of the University of North Carolina at Chapel Hill approved all animal studies.

Plasma hormone measurements

Animals were not fasted prior to euthanasia; therefore any measured hormone or glucose levels were in mice allowed to feed ad libitum. Blood was collected by cardiac puncture. Plasma was obtained by centrifugation at 2500 rpm for 6 minutes. Plasma for IGF1 assays was processed by acid-ethanol extraction to remove IGF binding proteins as described in (30). ELISA kits were used to measure plasma glucose (Cayman Chemical), insulin (Mercodia), and IGF1 (R&D Systems) levels, according to manufacturer's instructions.

Tissue harvest

Entire small intestine was collected, flushed to remove contents, and weight and length measured. To mark cells in S-phase, 5-ethynyl-2′deoxyuridine (EdU, Sigma) was administered by ip injection (100 μg/25g body weight) to animals 90 minutes prior to killing. Small intestine was divided into three segments: the most proximal and distal 10cm were considered duodenum and ileum, respectively. The remaining middle segment was considered jejunum and used in all subsequent studies.

Histological analyses

All quantitative histological analyses were performed by an investigator blinded to diet groups. Morphometric analyses were performed on zinc formalin fixed, paraffin embedded, hematoxylin, and eosin stained cross sections as described in (31, 32). Crypt density was calculated by dividing the number of well-oriented crypts per millimeter of submucosal circumference. Immunofluorescence analyses were performed as described in (24). Briefly, jejunum was opened longitudinally and fixed in fresh 4% paraformaldehyde followed by 24-hour incubations in 10% and 30% sucrose. Frozen 5–7 μm sections were cut and mounted for staining. To visualize cells in S-phase, sections were stained with EdU using the Click-iT EdU AlexaFluor 594 Kit following manufacturer's instructions (Invitrogen). For other immunofluorescence stains, sections were incubated overnight in the following primary antibodies: chicken α-GFP (1:500; Aves Labs), rabbit α-chromogranin-A (1:400; Abcam), rabbit α-lysozyme (1:500; Leica Biosystems), and rabbit α-mucin2 (1:200; Santa Cruz Biotechnology). The following secondary antibodies were used at 1:500: goat α-chicken-AlexaFluor 488 (Invitrogen) and goat α-rabbit Cy3 (Jackson ImmunoResearch Laboratories). DAPI containing mounting medium was used to visualize nuclei (Electron Microscopy Sciences). Images were captured using an inverted fluorescence microscope (Olympus IX83) fitted with a digital camera (ORCA-Flash4.0 C11440). The number of positively stained cells per crypt or villus section was counted in at least 20 crypts or villi/animal. Individual positive cells were confirmed by DAPI nuclear staining. Confocal images were photographed using the Leica SP2 Laser Scanning Confocal Microscope (Leica Microsystems).

Intestinal epithelial cell dissociation for flow cytometry and fluorescence activated cell sorting (FACS)

Preparation of jejunal epithelial cells for flow cytometry and FACS was carried out as described in (24). For flow cytometry studies, dead cells were excluded based on uptake of propidium iodide (Sigma) and quantification of Sox9-EGFP^Low ISC was performed using a Cyan flow cytometer and Summit version 4.3 software (Beckman Coulter). Sox9-EGFP cell populations were sorted using the MoFlo XDP cell sorter (Beckman Coulter) using gating parameters as described in (20, 24). CD45⁺ (BioLegend), CD31⁺ (BioLegend), and Annexin-V⁺ (Life Technologies) cells were excluded prior to sorting. Sort efficiency was assessed by postsort cell analysis to establish the percentage of populations that fall within the previously established gates.

RNA isolation and high throughput quantitative real-time PCR (qRT-PCR) by Fluidigm

Total RNA was isolated from whole thickness jejunum or FACS isolated epithelial cells using the RNeasy Mini Kit (Qiagen) per manufacturer's instructions. For high throughput qRT-PCR of FACS isolated cells, RNA quality was assessed by the 2100 Bioanalyzer (Agilent Technologies) and high quality RNA was included for gene expression studies using the Fluidigm BioMark HD system per manufacturer's instructions (Fluidigm). Igf1 is expressed in intestinal mesenchyme (33, 34) and so Igf1 mRNA was quantified on total RNA extracted from whole thickness segments of jejunum. Total RNA was reverse transcribed using the High Capacity cDNA Reverse Transcription Kit with RNase inhibitor (Applied Biosystems). qRT-PCR reactions were performed in duplicate using the Rotor-Gene 3000 and analyzed using Rotor-Gene software version 6.0.23 (Qiagen). All qRT-PCR data were normalized to the invariant control gene ActB. For FACS isolated samples, data were also normalized to nonsorted intestinal epithelial samples or Sox9-EGFP^Low ISC from control animals sorted on the same day in the same run to control for any technical variability across cell preparations and/or sorting procedures. The following TaqMan primer/probesets were used ActB: Mm00607939_s1, Igf1: Mm00439561_m1, Sox9: Mm00448840_m1, Chga: Mm00514341_m1, Hopx: Mm00558630_m1, Lct: Mm01285112_m1, Lgr5: Mm00438890_m1, Ccnd1: Mm00432359_m1, Igf1r: Mm00802831_m1, Insr: Mm01211881_m1, Myc Mm00487804_m1, and Axin2 Mm00443610_m1 (Applied Biosystems). EGFP mRNA was assessed as described in (24).

RT-PCR for insulin receptor (IR) isoform expression

A quantity of 2 μg cDNA was used for RT-PCR using primers that amplify the A and B isoforms of the IR as described in (35). Densitometry was performed using ImageJ (http://imagej.nih.gov/ij).

In vitro culture of FACS-isolated Sox9-EGFP cell populations

Culture experiments were carried out using methods originally described by Sato et al (36) for Lgr5⁺ cells and adapted for Sox9-EGFP^Low ISC by Gracz et al (20). In all studies, ISC were plated at low density and cultured in growth factor reduced Matrigel (BD Biosciences) with a standard growth factor cocktail (EGF: R&D Systems, Noggin: PeproTech, and R-Spondin 1: R&D Systems). To assess responsiveness to insulin or IGF1, ISC from control or DIO mice were cultured plus or minus insulin (Sigma), IGF1 (Genentech), or insulin and IGF1, added at a concentration of 50 ng/mL every other day. Enterospheres are predominantly seen at day 2–4 post plating and starting at day 6–8, they typically grow and develop into enteroids. Number of enterospheres/enteroids formed was counted every other day by an investigator blinded to treatment groups until day 12 post plating (end of study). Quantification and photographs were taken at 10x objective.

Statistical analysis

Data are expressed as mean ± SEM. For control and DIO groups, n ≥ 20 for the entire study. Subsets of animals were used for different experiments. All experimental results include ≥ 3 independent pairs of animals. Body weights were compared between diet groups by repeated measures ANOVA. For high throughput qRT-PCR experiments, differences between Sox9-EGFP populations were compared using 1-way ANOVA followed by pairwise comparisons using Holm-Sidak post hoc test. Impact of diet on individual Sox9-EGFP populations was assessed using paired t test on cells isolated from DIO and chow-fed control littermate pairs sorted in the same run. ISC culture studies were performed on ISC isolated from Sox9-EGFP littermate pairs fed either HFD or chow. ISC were isolated from these littermate pairs by FACS and plated on the same day in any given experiment. A paired t test was therefore used to assess impact of diet on cultured ISC or their response to insulin, IGF1, or both. All remaining data were compared using Student's t test as appropriate. In all analyses, P < .05 was considered statistically significant. All statistical analyses were performed using SigmaPlot 12.0.

Results

Diet-induced obesity in Sox9-EGFP mice

Sox9-EGFP mice were fed a low-fat standard chow or HFD for 20 weeks to induce DIO. Male and female mice responded similarly to HFD feeding and were combined in all analyses. Sox9-EGFP mice fed HFD showed significantly higher body weight by 7 weeks on diet (+23.6 ± 4.7%). This effect was maintained and more dramatic (+37.2 ± 4.8%) by 20 weeks on diet (Figure 1A) and accompanied by an increase in percent fat mass and a decrease in percent lean mass (Figure 1B). Following 20 weeks on diet, DIO mice exhibited elevated plasma glucose, insulin, and IGF1 (Figure 1, C–E). DIO mice also exhibited significantly higher jejunal Igf1 mRNA levels (Figure 1F).

Figure 1.

High-fat diet feeding for 20 weeks increases body weight and fat mass leading to diet-induced obesity (DIO), elevated plasma glucose, insulin, and IGF1 levels. A, Body weight over 20 weeks in mice fed low-fat chow (control) or high-fat diet to induce DIO. B, Body composition of control vs DIO mice measured by magnetic resonance imaging after 20 weeks on diet. C–E, Circulating plasma concentrations of (C) glucose, (D) insulin, and (E) insulin-like growth factor 1 (IGF1). F, qRT-PCR measured Igf1 mRNA from jejunum. Data expressed as mean ± SEM. *, P < .05 DIO vs control, repeated measures ANOVA (A) or unpaired t test (B–F), n ≥ 5.

Impact of DIO on small intestine weight, length, and jejunal crypt and villus morphometry

As shown in Table 1, DIO resulted in a small but significant decrease in intestinal length and a nonsignificant trend for decreased overall intestinal weight so that weight per unit length did not change significantly. Crypt-villus morphometry was measured in the jejunum, a major region of nutrient absorption. Villus height was increased in DIO mice (+18.3 ± 6.7%; Figure 2, A and B). Crypt depth did not differ between DIO and controls, but crypt density was significantly increased in DIO mice (+12.0 ± 3.8%; Figure 2, A and C), indicating an increase in the number of crypts feeding onto the heightened villi. Percentage of fissioning crypts was similar in DIO and controls.

Table 1.

Measures of Intestinal Morphology and Morphometry in Control Versus DIO Mice

Measure	Control	DIO	P Value
	Mean (sem)	Mean (sem)
Small intestine length (cm)	46.7 (1.0)	43.6 (1.3)	.02a
Small intestine weight (g)	1.9 (0.09)	1.8 (0.09)	.35
Small intestine weight/length (g/cm)	0.04 (0.002)	0.04 (0.002)	.73
Jejunal crypt depth (μm)	70.1 (2.5)	66.1 (0.8)	.14
Jejunal villus height (μm)	282.1 (8.0)	333.8 (18.8)	.02a
Jejunal crypt density (#/mm submucosal circumference)	18.4 (0.6)	20.6 (0.7)	.04a
Jejunal crypt fission (% of total cross section)	0.92 (0.1)	1.15 (0.2)	.31

^a, Significant P < .05

Figure 2.

Increases in villus height, crypt density, ISC number, and proportion of ISC in S-phase in DIO vs control mice. A, Representative hematoxylin and eosin stained photographs depicting crypt-villus architecture of intestinal epithelium from control and DIO mice. Images were taken at 10x magnification. Scale bar, 100 μm. B, Villus height in control and DIO mice. C, Crypt density quantified by number of crypts per mm submucosal circumference in control and DIO mice. D, Number of cells per crypt in control and DIO mice. E, Number of cells in S-phase measured by EdU positive staining per crypt. F, Representative images of crypt sections stained with DAPI (blue), GFP (Sox9-EGFP; green), and Chromogranin-A (ChgA; red). Sox9-EGFP^Low ISC (white arrows) and Sox9-EGFP^High cells (open triangles) were defined by intensity of EGFP staining. G, Quantification of Sox9-EGFP^Low and Sox9-EGFP^High cells. H, Representative images of crypt sections stained with DAPI, GFP, and EdU (red). Dual positive Sox9-EGFP^Low and EdU cells are denoted by the filled triangles. EdU: 5-ethynyl-2′-deoxyuridine. Images were taken at 63x magnification. Scale bar, 20 μm. I, Percentage of Sox9-EGFP^Low or Sox9-EGFP^High cells positive for EdU. J, Relative abundance of Sox9-EGFP^Low cells assayed by flow cytometry and expressed as the percentage of total cells that fall in Sox9-EGFP^Low gates. Data expressed as mean ± SEM. *, P < .05 DIO vs control, unpaired t test, n ≥ 5.

DIO selectively increases number of ISC and ISC in S-phase

Total number of cells per crypt section and total number of cells labeled with the S-phase marker EdU per crypt section did not differ in DIO mice vs controls (Figure 2, D and E). We used the Sox9-EGFP reporter mouse to assess if DIO affected numbers of Sox9-EGFP^Low ISC or crypt-based Sox9-EGFP^High cells that express the EEC marker Chromogranin A (ChgA), but also contain a population of reserve ISC-like cells activated to proliferate after injury (24). Sox9-EGFP^High cells were distinguished from Sox9-EGFP^Low cells by both high intensity of EGFP and ChgA expression (Figure 2F). As shown in Figure 2, F and G, there was a small but significant increase in the number of Sox9-EGFP^Low ISC but not Sox9-EGFP^High cells per crypt section. Colabeling with EdU was used to quantify the numbers of Sox9-EGFP^Low and Sox9-EGFP^High cells in S-phase. DIO resulted in a significant increase in the percentage of Sox9-EGFP^Low ISC colabeled with EdU, but no significant change in the proportion of Sox9-EGFP^High EdU positive cells per crypt section (Figure 2, H and I). Whereas the total numbers of Sox9-EGFP^Low cells per crypt section were increased by 24.6 ± 6.2%, there was a greater increase in the number of EdU positive Sox9-EGFP^Low cells per crypt section (+48.1 ± 19.3%) indicative of ISC hyperproliferation and/or altered cell cycle time. Flow cytometry, performed as an independent measure of the proportion of Sox9-EGFP^Low ISC in entire jejunum of control and DIO mice, revealed a significant increase in the percentage of Sox9-EGFP^Low ISC in DIO mice compared to controls (Figure 2J). Collectively, the morphometry, histology, and flow cytometry data demonstrate that DIO results in increased total numbers of Sox9-EGFP^Low ISC and percentage of ISC in S-phase, which may be required to support the increased crypt density and villus height even though total numbers of cells per crypt are unchanged.

DIO mice have decreased numbers of Paneth and goblet cells

To evaluate if increased villus height and increased ISC proliferation in DIO mice were associated with changes in differentiated lineages, we compared numbers of Paneth, goblet and EEC in DIO mice vs controls. Quantification of lysozyme positive cells revealed a 25.7 ± 3.1% decrease in the number of Paneth cells per crypt section in DIO mice compared to controls (Figure 3, A and B). Mucin2 staining revealed that numbers of goblet cells were significantly decreased in DIO mice in both crypts (−33.6 ± 4.6%) and villi (−24.4 ± 4.6%) compared to controls (Figure 3, C and D). EEC number evaluated by the number of ChgA positive cells revealed no differences between control and DIO mice in either crypt or villus (Figure 3, E and F). Taken together, these data suggest DIO mice exhibit significantly reduced numbers of Paneth and goblet cells and no change in EEC, providing indirect evidence that the increased villus height likely involves increased number or size of enterocytes, the other major differentiated villus lineage.

Figure 3.

Decreased Paneth and goblet cells but no change in EEC in DIO mice vs controls. A, C, and E, Representative immunofluorescent images of jejunal sections stained with (A) lysozyme (Lyz), (C) mucin2 (Muc2), or (E) Chromogranin-A (ChgA). B, D, and F, Quantification of (B) Lyz positive Paneth cells, (D) Muc2 positive goblet cells, and (F) ChgA positive EEC in DIO vs control mice. All images were taken at 20x magnification. Scale bar, 50 μm. Data expressed as mean ± SEM. *, P < .05 DIO vs control, unpaired t test, n ≥ 4.

Sox9-EGFP^Low ISC isolated from DIO mice have reduced ability to form enterospheres and enteroids in culture

ISC were isolated by FACS using dispersed epithelial preparations from control and DIO Sox9-EGFP reporter mice. High throughput qRT-PCR on isolated Sox9-EGFP cell populations demonstrated an appropriate and similar gradient of EGFP and Sox9 mRNA across the sorted cell populations from control and DIO mice (Figure 4, A and B). Sox9-EGFP^High cells were enriched for mRNAs encoding EEC marker Chga and reserve ISC marker Hopx (Figure 4, C and D) and Sox9-EGFP^Negative cells were enriched in Lct mRNA, a brush border enzyme expressed by absorptive enterocytes (Figure 4E). Importantly, Sox9-EGFP^Low cells isolated from both control and DIO mice were both similarly enriched for the mRNA encoding the ISC marker Lgr5 (Figure 4F). Consistent with selective hyperproliferation of ISC in DIO mice observed by histology and EdU, cyclin D1 (Ccnd1) mRNA was increased only in Sox9-EGFP^Low cells from DIO mice vs controls and not in any other cell population (Figure 4G).

Figure 4.

Sox9-EGFP^Low ISC from DIO mice are enriched for appropriate biomarkers and show elevated cyclin D1 mRNA. High throughput qRT-PCR on FACS isolated Sox9-EGFP cells from control and DIO mice assessed levels of mRNAs encoding (A) EGFP, (B) Sox9, (C) Chga, (D) Hopx, (E) Lct, (F) Lgr5, and (G) Ccnd1. Data expressed as mean ± SEM. * or **, P < .05 vs all other Sox9-EGFP populations; ^#, P < .05 in DIO vs control Sox9-EGFP^Low ISC. Differences between Sox9-EGFP populations compared by 1-way ANOVA, Holm-Sidak and between diets by paired t test, n ≥ 3.

To assess whether the in vivo increase in Sox9-EGFP^Low ISC proliferation in DIO mice translated to increased intrinsic ISC function, we compared the ability of Sox9-EGFP^Low cells isolated from DIO and control mice to survive, expand, and form enterospheres/enteroids in 3D culture. In this in vitro assay, low density ISC are plated at day 0 and the number of structures formed are counted every other day for 12 days. Because of low-density plating, enterosphere/enteroids derive primarily from single ISC. This assay is a useful in vitro readout for intrinsic function of stem cells in terms of survival and growth. In addition to Sox9-EGFP^Low cells, we also tested Sox9-EGFP^High, Sox9-EGFP^Sublow, and Sox9-EGFP^Negative cells to establish if DIO altered stemness in populations other than actively cycling Sox9-EGFP^Low ISC. Consistent with previous findings in chow fed mice, Sox9-EGFP^Negative, Sox9-EGFP^Sublow, and Sox9-EGFP^High cells from control mice did not survive or expand to form enterospheres/enteroids, and this was also the case in DIO mice (data not shown). Sox9-EGFP^Low ISC isolated from both control and DIO mice were able to survive and form enterospheres/enteroids in 3D culture (Figure 5A). Enterospheres were quantified starting at day 4 when these structures are large enough to count reliably. No significant difference was seen in the percentage of enterospheres formed from Sox9-EGFP^Low ISC isolated from DIO or control mice at day 4-post plating, although there was a trend for lower numbers in DIO mice (Figure 5B). By day 6 post plating, Sox9-EGFP^Low ISC isolated from DIO mice yielded fewer enteroids than controls and numbers of enteroids formed from DIO ISC remained significantly lower until the end of the culture at day 12 post plating (Figure 5B). Whereas sizes of 3D enteroids are difficult to quantify, qualitative evaluation indicated a reduced size of enterospheres/enteroids formed from ISC isolated from DIO mice vs controls as shown in the examples in Figure 5A. These data indicate that despite increased numbers and proliferation of ISC in DIO mice in vivo, the intrinsic in vitro survival and growth capabilities of ISC isolated from DIO mice are impaired relative to ISC isolated from controls. This provides novel evidence that DIO alters ISC intrinsic function.

Figure 5.

Sox9-EGFP^Low ISC from DIO mice exhibit reduced enteroid-forming ability. A, Time course of enterosphere/enteroid formation by ISC isolated from control and DIO mice. Images were taken at 10x magnification. Scale bar, 100 μm. B, Quantification of enterosphere/enteroids formed starting at day 4 until day 12. Data expressed as mean percentage vs enteroids formed in control at day 4 ± SEM. *, P < .05 DIO vs control, paired t test, n ≥ 6 independent pairs performed in duplicate or triplicate.

Insulin and/or IGF1 treatment rescues functional defect of ISC isolated from DIO mice

DIO mice are hyperinsulinemic, display elevated plasma IGF1 levels (Figure 1, D and E), and have significantly higher local intestinal Igf1 mRNA expression (Figure 1F). Additionally, plasma insulin levels were found to be positively and significantly correlated with abundance of Sox9-EGFP^Low ISC evaluated by flow cytometry in the same animals (Figure 6A). We therefore hypothesized that the reduced in vitro survival and growth of ISC from DIO mice might reflect an acquired dependence on elevated insulin or IGF1. To test whether the decrease in enterosphere/enteroid formation seen in ISC from DIO mice could be rescued by supplementation of cultures with insulin and/or IGF1, isolated ISC from control and DIO mice were treated with standard growth factor conditions (EGF, Noggin, and R-Spondin 1) alone or plus insulin, IGF1, or insulin and IGF1 combined. Consistent with prior results (Figure 5), in this independent series of experiments, Sox9-EGFP^Low ISC from DIO mice formed fewer enteroids than controls when plated in standard growth factor conditions alone (Figure 6B). Neither insulin nor IGF1, nor both affected enteroid yield from ISC isolated from control mice. In contrast, insulin, IGF1, or insulin and IGF1 combined resulted in a significant > 3-fold increase in the number of enteroids formed from Sox9-EGFP^Low ISC isolated from DIO mice (Figure 6B) compared to ISC from DIO mice cultured in standard growth factor conditions. In fact, the mean number of enteroids formed from IGF1 treated ISC from DIO mice significantly exceeded the number formed from ISC of control mice cultured under standard growth factor conditions (P < .05) and a similar trend was observed in DIO ISC treated with insulin (P = .07) or insulin and IGF1 (P = .17) (Figure 6B). Insulin, IGF1, or both therefore rescued the defect in survival and growth of DIO ISC. qRT-PCR on FACS isolated Sox9-EGFP^Low ISC and other populations revealed that IGF1 receptor (Igf1r) mRNA was significantly enriched in Sox9-EGFP^Low ISC, suggesting that ISC may be particularly responsive to IGF1 (Figure 6C). In contrast, Ir mRNA was significantly enriched in Sox9-EGFP^High cells (Figure 6D). However neither levels of Igf1r mRNA nor Ir mRNA differed in Sox9-EGFP^Low ISC isolated from DIO vs control mice suggesting that receptor downregulation, at least at the mRNA level, did not accompany differences in enteroid forming ability or dependence on exogenous insulin/IGF1 in isolated ISC from DIO mice. Our prior studies demonstrated that IR isoform A (IR-A), a receptor that can mediate proliferative actions of insulin or IGFs, is the predominant IR isoform in Sox9-EGFP^Low ISC compared to the metabolic IR-B isoform (35). We performed RT-PCR to evaluate if DIO resulted in changes in relative proportions of the two IR isoforms. We found no significant changes in relative expression of the two isoforms, with IR-A being expressed at > 2-fold the levels of IR-B in ISC from control and DIO mice. (Figure 6, E and F). Thus, ISC from DIO mice require the presence of elevated insulin and/or IGF1 to maintain their intrinsic function in vitro but this does not reflect reduced levels of either Igf1r or Ir at least at the level of mRNA expression. Wnt signaling is a key pathway involved in ISC function. We therefore assessed levels of two known Wnt targets, Myc and Axin2 in Sox9-EGFP^Low ISC isolated from DIO and control mice. Both Myc and Axin2 mRNAs were significantly lower in ISC isolated from DIO mice compared to controls (−20.5 ± 8.5% and −24.4 ± 9.3%, respectively). Decreased Wnt activation in ISC isolated from DIO mice could therefore contribute to their reduced enteroid forming ability.

Figure 6.

Plasma insulin positively correlates with percentage of Sox9-EGFP^Low ISC and treatment of ISC from DIO mice with insulin, IGF1, or both rescues decreased intrinsic in vitro function. A, Linear regression analysis of plasma insulin levels and percentage of Sox9-EGFP^Low ISC in individual mice. B, Enteroid formation at day 12 from Sox9-EGFP^Low ISC isolated from control or DIO mice cultured with standard growth factors (Std. GF) as described in Materials and Methods or Std. GF plus insulin (Ins.), IGF1, or Ins. and IGF1. Data expressed as mean ± SEM. *, P < .05 vs control cultured with Std. GF alone; **, P < .05 vs DIO Std. GF alone, paired t test, n = 3 for ISC isolated from 3 independent littermate pairs each cultured in duplicate. C and D, High throughput qRT-PCR measured (C) Igf1r and (D) total Ir mRNA. E, RT-PCR on Sox9-EGFP^Low ISC measured IR isoform expression. F, Ratio of IR-A to IR-B expression quantified in control vs DIO Sox9-EGFP^Low ISC. Data expressed as mean ± SEM. *, P < .05 vs all Sox9-EGFP cell populations, 1-way ANOVA, Holm-Sidak, n ≥ 3.

Discussion

In this study, we used the Sox9-EGFP reporter model to directly examine the impact of DIO due to chronic HFD exposure on Sox9-EGFP^Low ISC, which share a similar gene signature to actively cycling Lgr5⁺ ISC (24) and are able to survive and form enteroids in culture (20). We provide new evidence that DIO leads to increased numbers of jejunal Sox9-EGFP^Low ISC, increased ISC proliferation as measured by EdU, and increased expression of Ccnd1 mRNA specifically in Sox9-EGFP^Low ISC, but does not affect total crypt cell number, crypt depth, or total EdU positive cells per crypt. DIO was associated with increased crypt density and villus height, supporting a model whereby the increased numbers of ISC and ISC in S-phase support the increased number of crypts that in turn feed cells onto longer villi as an adaptation to hypercaloric load (Figure 7).

Figure 7.

Proposed model of impact of DIO on ISC number, function and crypt-villus homeostasis. We provide a model where intestinal adaptation to DIO increases the number of total and proliferating ISC associated with elevated plasma insulin and IGF1 and local Igf1 mRNA. The expansion in the ISC pool is associated with increases in the number of crypts feeding onto each villus resulting in heightened villi. DIO also decreases crypt-based Paneth cells and crypt and villus goblet cells that may favor a greater mass of villus enterocytes that may enhance absorptive capacity during hypercaloric conditions.

Our results are consistent with recent findings that report increases in villus height and numbers of Ki67 or BrdU positive cells per crypt with long-term HFD feeding (18, 19). However, our study provides, to our knowledge, the first evidence that DIO preferentially expands and promotes proliferation of Sox9-EGFP^Low ISC. Our prior studies on Sox9-EGFP^High cells, which are enriched for EEC markers (24), and evidence from the literature indicate that a subpopulation of secretory cells, EEC or Paneth cells or their immediate progenitors can be activated to proliferate and adopt a stem cell phenotype upon injury (24, 37). Our findings indicate that DIO does not increase EdU positive Sox9-EGFP^High cells and that Sox9-EGFP^High cells from DIO mice do not acquire the ability to generate enterospheres/enteroids in culture suggesting that in contrast to injury models, DIO does not activate Sox9-EGFP^High cells to expand, proliferate, or adopt functional characteristics of stem cells. Thus DIO selectively promotes hyperproliferation of the more actively cycling Sox9-EGFP^Low ISC.

Our findings show that concomitant with increased villus height and ISC hyperproliferation, DIO mice have reduced numbers of Paneth cells in crypts, reduced goblet cells on crypts and villi, and no change in EEC suggesting that increased villus height likely reflects increased numbers or size of absorptive enterocytes. Other studies have reported reduced goblet cells in obese mice (38, 39) whereas some reports have suggested changes in numbers of specific EEC (38, 40, 41). To our knowledge, this is the first report of reduced Paneth cell number in DIO mice. This is potentially interesting since Paneth cells are reported to secrete factors such as Wnts that provide trophic support to ISC. However in vivo, despite reduced Paneth cell numbers, DIO ISC still hyperproliferate and expand and the ISC expansion correlates with plasma insulin levels.

We used a 3D Matrigel based culture system to directly test if the in vivo hyperproliferation of Sox9-EGFP^Low ISC was associated with enhanced ability of these cells to generate enterospheres/enteroids in vitro. Surprisingly, the number of enterospheres/enteroids derived from isolated Sox9-EGFP^Low ISC was reduced in DIO mice vs controls indicating impaired rather than enhanced intrinsic function of DIO ISC when isolated from their in vivo environment. This suggested that some extrinsic signal in the in vivo setting that promotes survival or proliferation of ISC from DIO mice may be deficient in the in vitro system. In support of this possibility, supplementation of culture medium with insulin and/or IGF1 that are increased in plasma of DIO mice, was able to rescue the impaired enteroid-forming ability of Sox9-EGFP^Low ISC from DIO mice to levels equal to or greater than that observed in Sox9-EGFP^Low ISC from controls. Interestingly, the addition of insulin, IGF1, or both factors combined did not affect the yield of enteroids from control Sox9-EGFP^Low ISC. Because DIO mice exhibited elevated circulating levels of insulin and IGF1 and elevated locally expressed Igf1 mRNA in vivo, we interpret these findings as novel evidence that ISC from DIO mice develop a dependence or responsiveness to exogenous insulin or IGF1 for their survival or increased proliferation normally not seen in control ISC. At the mRNA level, we found no evidence for reduced levels of Igf1r, Ir, or IR-A in Sox9-EGFP^Low ISC from DIO mice vs controls suggesting that the dependence on exogenous insulin/IGF1 may result from a mechanism other than downregulation of receptor expression. We cannot exclude altered expression of Igf1r or Ir at the protein level, but this is difficult to quantify considering the small numbers of ISC that can be collected. The significant decrease in Wnt targets Myc and Axin2 mRNAs in Sox9-EGFP^Low ISC isolated from DIO mice suggest that reduced Wnt activation may contribute to impaired enteroid-forming ability of Sox9-EGFP^Low ISC. This could reflect reduced Wnt ligand exposure in vivo due to reduced Paneth cell numbers. Impaired Wnt signaling is also seen in other tissues during obesity such as the hypothalamus and bone (42, 43). In other systems, insulin and IGFs are known to activate β-catenin/TCF, known downstream mediators of Wnt activation (44–46). Thus the ability of insulin and IGF1 to promote the enteroid-forming ability of DIO ISC may reflect their ability to compensate for Wnt downregulation. Additional studies of the Wnt pathway or insulin/IGF/Wnt interactions in ISC of DIO mice represent an interesting future direction.

It is noteworthy that CR reduced progenitor proliferation and led to a reduction in villus height of similar magnitude to the increase in villus height observed here with DIO, but similar to DIO, CR resulted in selective ISC expansion and hyperproliferation based on numbers of cells labeled with the ISC biomarker Olfm4 and BrdU (14). However the mechanisms leading to selective increases in ISC numbers and proliferation appear to differ in CR and DIO. In CR, increased ISC number and proliferation in vivo were associated with enhanced enteroid formation in vitro that depended on paracrine interactions with Paneth cells and increased responsiveness of Paneth cells to insulin appeared to enhance intrinsic ISC enteroid forming abilities (14). In contrast, our study suggests that despite increased ISC numbers and hyperproliferation in DIO in vivo, isolated ISC exhibit impaired intrinsic function that can be rescued with insulin or IGF1.

In summary, our study provides novel evidence in rodents that in vivo DIO increases ISC number and this correlates with elevated plasma insulin and is associated with ISC hyperproliferation and decreased Paneth and goblet cell number. However in vitro analyses of isolated ISC from DIO mice demonstrate impaired intrinsic function that can be reversed by insulin and/or IGF1, indicating that DIO leads to an acquired dependence of ISC on insulin or IGF1.