Aging effects on intestinal homeostasis associated with expansion and dysfunction of intestinal epithelial stem cells

Aging-associated decreases in enterosphere formation and enteroid development

Crypt culture in a 3D matrigel system provides a useful and widely used system to assess IESC function [30]. Isolated intestinal crypts form round enterospheres that expand and differentiate to form more complex enteroids with bud structures after several days in culture [31]. Crypt culture revealed that intestinal crypts isolated from both young and old animals formed enterospheres in vitro on day 1 after plating (Figure 1A). Efficiency of crypt culture was calculated by dividing the number of enterospheres and enteroids present at day 4 or 8 by the number of enterospheres present on day 1 in each well. This provides a measurement of how many enterospheres initially plated were able to survive and grow in crypt culture conditions. Efficiency measurements revealed a trend for decreased enteroid survival in old vs. young at day 4 post plating and a significant decrease in enteroid survival in old animals at day 8 post plating (Figure 1B). Crypt-derived enteroids typically begin to show bud structures by 3-7 days post plating [31]. Each bud represents a crypt structure that containing stem and progenitor cells and the number of buds provide a surrogate for IESC function [32]. The numbers of buds per enteroid were counted at days 4 and 8 post plating, categorizing enteroids with 2 buds or fewer as less complex, and enteroids containing 3 or more buds as more complex. Following 4 days in culture there was no difference in the enteroid complexity between young and old (Figure 1C). By day 8 post plating, enteroids from old mice showed a decrease in complexity compared to those from young mice as significantly fewer enteroids from old animals contained 3 or more buds (Figure 1C). At 15 days post plating very complex enteroids had formed from the crypts derived from young mice, while the enteroids formed from old mice were much less complex (Figure 1A).

Increased villus height and Paneth cell number in small intestine of old mice

Jejunal morphology and morphometry and the presence of differentiated cell types were assessed by histology. Results revealed no significant difference in crypt depth, crypt or villus density, total number of cells per crypt, or mucosal circumference between young and old mice, but demonstrated a significant increase in villus height in old vs. young mice (Table 1 and Figure 2A). Immunofluorescence for lysozyme, a Paneth cell biomarker, revealed a small but significant increase in the number of Paneth cells per crypt in old vs. young (Table 1 and Figure 2B). Alcian Blue positive goblet cells were quantified and revealed no change in the number of mucus producing goblet cells between the young and old mice (Table 1 and Figure 2C).

Old mice exhibit increased IESC number and enhanced IESC proliferation

Flow cytometry on IEC isolated from Sox9-EGFP mice revealed significantly more IESC (Sox9-EGFP^Low) and progenitor cells (Sox9-EGFP^Sublow) in old vs.to young (Figure 3A and 3B). However, the Sox9-EGFP^High cell population that contains enteroendocrine cells (EECs) and a reserve stem cell population activatable upon injury and regeneration [25] did not differ between young and old mice (Figure 3A and 3B). Results were confirmed by EGFP immunofluorescence, which revealed no change in the number of Sox9-EGFP^High cells per crypt in old vs. young (Figure 3C and 3E) but demonstrated a significant increase in the number of Sox9-EGFP^Low IESC per crypt in old animals (Figure 3C and 3D). This was further confirmed in the Lgr5-LacZ IESC reporter mouse model [23] where expansion of Lgr5-LacZ IESC was observed in old vs. young (Figure 3F and 3G). Of note, using histology it is not possible to reliably quantify Sox9-EGFP^Sublow progenitor cells by EGFP immunofluorescence because the very low EGFP expression cannot be readily distinguished from background.

Flow cytometry of dissociated jejunal IEC from young and old Sox9-EGFP reporter mice detected incorporation of the S-phase marker EdU in 3 distinct EGFP-expressing cell populations (Figure 4B). As expected, rapidly dividing Sox9-EGFP^Sublow progenitor cell population showed the most EdU positive cells compared to other Sox9-EGFP populations (Figure 4). Based on percentage of cells incorporating EdU, there was no detectable difference in proliferation of Sox9-EGFP^Sublow progenitors between young (33.7±2.7%) and old (37.6±5.0%) (Figure 4). However, there was a significant increase in the percentage of EdU-positive Sox9-EGFP^Low IESC in old (39.5±4.3%) vs. young (26.3±3.8%), indicating hyperproliferation of IESC in old mice. The percentage of EdU-labeled Sox9-EGFP^High cells was significantly higher in old (7.8±1.6%) vs. young (2.4±0.5%) (Figure 4). This provides evidence that old mice have a greater proportion of “activatable reserve” IESC in S phase.

Immunofluorescence for EdU revealed no difference in the total number of EdU positive cells per crypt in old vs. young (Figure 4C, 4D). However, co-immunofluorescence for EdU and GFP revealed a significant increase in the number of EdU positive Sox9-EGFP^Low IESC located at the crypt base in old mice (Figure 4C, 4D). Results were confirmed in an independent IESC reporter mouse model, Lgr5-EGFP [23] (Figure 5A-C), which revealed significantly more Lgr5-EGFP IESC co-labeled with EdU in old (2.3±0.2) vs. young (1.5±0.2) (Figure 5C). Together these data provide strong evidence that aging is associated with a selective increase in proliferation and/or altered cell cycle dynamics in IESC of old mice.

Old mice exhibit increased apoptosis of the small intestinal epithelial cells

Given the age-associated IESC dysfunction indicated by our results in the ex vivo enteroid culture system, we tested whether apoptosis was increased in the small intestinal crypt of old vs. young mice. Cleaved caspase-3 immunohistochemistry revealed a significant increase in the number of apoptotic cells per crypt section in old vs. young (Figure 6A, 6B-right and 6C). The location of cleaved caspase-3 positive cells within the crypt was assessed by dividing the crypt in half lengthwise and counting the position of apoptotic cells from the base of the crypt (Figure 6D). Most apoptotic cells in old animals were at the base of the crypt in positions 1-4 (Figure 6C and 6D), indicating preferential increases in apoptosis within the IESC zone. Old mice showed a significant increase in cleaved caspase-3 positive cells in the villi (Figure 6A and 6B-left), that were strikingly located along the villus epithelium rather than the villus tip.

Aging-associated changes in gene expression

To gain insight into molecular mechanisms or biomarkers of age associated functional changes in IESC and other cell populations between old and young mice, we used FACS (Fluorescence Activated Cell Sorting) coupled with Fluidigm mRNA profiling. First we validated that profiles of Sox9 mRNA and biomarkers of specific IEC subtypes exhibited predicted profiles in isolated cell populations from young and old Sox9-EGFP mice (Figure 7A-D). Our data verified that the expression profiles of Sox9 mRNA in FACS-isolated cells matched the profiles of Sox9-EGFP used to FACS isolate cell populations (Figure 7A). Furthermore, Sox9 mRNA profiles across Sox9 negative, sublow, low and high populations were consistent between young and old. The IESC specific marker Lgr5 was enriched in the Sox9-EGFP^Low population from both young and old, consistent with phenotypic similarities between Sox9-EGFP^Low and Lgr5-EGFP-expressing IESC (Figure 7B). Sox9-EGFP^High populations from young and old were enriched for EEC specific mRNA ChgA (Figure 7C) and Hopx, a putative biomarker of activatable reserve IESC (Figure 7D). To gain insight into the mechanisms driving aging-associated IESC expansion and hyperproliferation, we examined expression of mRNAs involved in cell cycle, apoptosis and maintenance of homeostasis in cell populations isolated from Sox9-EGFP mice. Consistent with histology and flow cytometry data suggesting increased proportion of active Lgr5-expressing and activatable reserve IESC in S-phase in old mice, we observed a significant increase in Ccnd1 and Cdk6 mRNAs in Sox9-EGFP^Low cells and a dramatic increase in Ccnd1 mRNA in Sox9-EGFP^High cells from old vs. young (Table 2A, 2B). Ccnd1 and Cdk6 are required for cell cycle progression and activate G1 to S phase transition [33,34]. Additional mRNAs involved in regulating the cell cycle and proliferation were significantly different in cell populations from old vs. young including Max in Sox9-EGFP^High and p21 in Sox9-EGFP^Low (p21) (Table 2A, 2B). Consistent with increased levels of apoptosis observed based on localization of cleaved caspase-3 in the crypt base of old animals, Fluidigm data revealed increased levels of mRNAs encoding p53 and Perp specifically in Sox9-EGFP^Low IESC from old mice (Table 2B). We found a significant increase in the mRNA encoding Nrf2, a marker of oxidative stress, in Sox9-EGFP^Low IESC from old mice (Table 2B). Significantly decreased levels of mRNAs associated with proliferation, apoptosis and homeostasis were observed in the old Sox9-EGFP^Sublow progenitor cell population, including Pten, Yap1, Ephb2, Notch2 and Bbc3 (Table 2C). H2afx mRNA levels were increased specifically in Sox9-EGFP^Negative cells from old mice, consistent with and providing a potential biomarker of age-associated increases in apoptosis of differentiated cells on the villus. Decreased levels of mRNAs known to be associated with tissue homeostasis with aging in Sox9-EGFP^Negative cells from old mice are consistent with increases in villus height, a sign of altered homeostasis.

Mouse models

Sox9-EGFP mice on an outbred CD-1 background were maintained as heterozygotes and genotyped as previously described [24,28]. Lgr5-EGFP-IRES-CreERT2 (Lgr5-EGFP) mice on a C57BL/6 background were ordered from The Jackson Laboratory (Bar Harbor, ME) and maintained as heterozygotes. Original Lgr5-LacZ mice on a mixed background were generously provided by Lexicon Pharmaceuticals (The Woodlands, TX). Lgr5-LacZ mice on C57BL/6 background that were used in this study were generated after 10 generations of backcrossing the original Lgr5-LacZ mice with C57BL/6 wild type mice. Young mice studied were 2 to 4 months old. Old mice were 18 to 24 months old. Young and old female and male mice were studied. For the old study groups, females were 20-24 months old and males 18-20 months old. These ages were selected based on pilot studies demonstrating no overt signs of disease in female or male mice at these ages.

Crypt isolation and culture

Jejunal crypts were isolated and in vitro crypt culture was performed as previously described [32]. Approximately 30 crypts were plated into 24 well plates in 10μl Low Growth Factor Matrigel (BD Biosciences, Franklin Lakes, NJ) containing 50ng/ml EGF (R&D Systems, Minneapolis, MN), 100ng/ml Noggin (Peprotech, Rocky Hill, NJ) 1μg/ml R-spondin (R&D Systems, Minneapolis, MN) and 10uM Y-27632 (Sigma-Aldrich, St. Louis, MO). After polymerization 250μl Advanced DMEM-F12 containing N2 supplement (Invitrogen, Carlsbad, CA), B27 supplement without vitamin A (Invitrogen, Carlsbad, CA), 10mM HEPES (Invitrogen, Carlsbad, CA), Glutamax (Gibco-Invitrogen, Carlsbad, CA) and Pen/Strep (Gibco-Invitrogen, Carlsbad, CA) was added to each well. Growth factors were added every 2 days at the same concentrations as the initial plating except R-spondin, which was reduced to 500ng/ml. Medium was changed every 4 days. The number of enterospheres or enteroids and their complexity was recorded at days 1, 4 and 8-post plating. Representative images of cultures were taken using an inverted fluorescent microscope (Olympus IX83).

Tissue harvest

Sox9-EGFP, Lgr5-EGFP and Lgr5-LacZ mice were euthanized with a lethal dose of Nembutal (150µg/g body weight). To identify cells in S-phase, 5-ethynyl-2’ deoxyuridine (EdU; Sigma-Aldrich, St. Louis, MO) was administered by intraperitoneal injection at a dose of 100µg/25 g of body weight 90 minutes prior to Nembutal injection. The entire intestine was then removed and flushed with cold PBS. The jejunum was dissected on ice for further analyses.

Immunostaining

Intestinal segments were splayed open and fixed overnight in freshly made 4% paraformaldehyde at 4°C, washed, incubated in 10% sucrose in PBS at 4°C overnight and incubated in 30% sucrose in PBS at 4°C until embedding. Tissues were embedded in Optimal Cutting Temperature compound (OCT) and sectioned at 5-7-μm thickness. Antigens were retrieved in 10mM sodium citrate buffer (pH=6) followed by blocking with 10% normal goat serum. After blocking, samples were incubated overnight at 4°C with each of the following primary antibodies: GFP (1:500, chicken; Aves Labs, Tigard, OR), lysozyme (1:500, rabbit; Leica Biosystems, Buffalo Grove, IL), cleaved caspase-3 (1:500, rabbit; Abcam, Cambridge, UK). For immunofluorescence, slides were washed and incubated with the following secondary antibodies: goat-anti-chicken-AlexaFluor 488 (1:400; Invitrogen, Carlsbad, CA) or goat-anti-rabbit-Cy3 (1:400; Jackson ImmunoResearch Laboratories, West Grove, PA). Slides were then mounted using DAPI containing mounting medium (Electron Microscopy Sciences, Hatfield, PA) to visualize nuclei. Peroxidase with Vectastain avidin-biotin complex (ABC kit; Vector Laboratories, Burlingame, CA) was used for immune peroxidase based staining. Morphometric analyses were performed on formalin fixed, paraffin embedded tissue cross-sections. Tissues were washed well and slides (5-7 μm) were stained with hematoxylin and eosin or Phloxine Tartrazine and Alcian Blue. Fluorescent images were captured using an inverted fluorescence microscope (IX83; Olympus, Center Valley, PA). White light images were captured using an upright light microscope (AX10; Zeiss, Oberkochen, Germany). At least 20 crypts were counted per animal and number of animals per group is indicated in figure legends.

Phloxine Tartrazine and Alcian Blue staining

Alcian Blue Phloxine Tartrazine staining was performed based on previous reports [55]. Briefly, slides were stained with Alcian blue (Sigma-Aldrich, St. Louis, MO) for 30 minutes, followed by alum hematoxylin (BioCare, Concord, CA) for 20 seconds, staining with 0.5% phloxine in calcium chloride (Fisher Scientific, Waltham, MA) for 20 minutes, and finally incubate in a saturated solution of tartrazine in Cellosolve (Fisher Scientific, Waltham, MA) for 3-5 minutes. Stained sections were observed using light microscopy.

β-Galactosidase staining

Jejunal cross sections from Lgr5-LacZ mice were dissected, washed with PBS, then fixed at 4º C using 2% paraformaldehyde (PFA)/ 0.2% glutaraldehyde in PBS for 2 hours. Tissues were then rinsed for 20 minutes at room temperature in 0.1M Phosphate Buffer, pH7.3 and permeabilized by incubation in 0.1M Phosphate Buffer, pH7.3 containing 0.1% Triton-X 100. Tissues were then transferred to X-gal solution (1g/liter X-Gal) in 0.1M Phosphate Buffer, pH7.3 with 5 mmol/liter of potassium ferrocyanide, 5 mmol/liter of potassium ferricyanide and 0.1% Triton X-100) overnight in the dark at room temperature. Tissues were then washed twice with PBS and post-fixed in 4% paraformaldehyde at 4ºC overnight. Subsequently, tissue was washed, dehydrated in 70% ethanol and embedded in paraffin. 5-7 μm sections were counterstained with nuclear fast red and observed using light microscopy. At least 40 crypts were counted per animal and number of animals per group is indicated in figure legends.

Intestinal epithelial cell isolation, flow cytometry and FACS

Intestinal epithelial cells were isolated from the jejunum of young and old Sox9-EGFP mice for flow cytometry as previously described [25]. Dissociated intestinal epithelium was stained with propidium iodide solution (Sigma-Aldrich, St. Louis, MO) to exclude dead cells. To assess the proportion of Sox9-EGFP cells in S-phase by flow cytometry, an equal number of dissociated intestinal epithelial cells was fixed and stained for EdU using the Click-it EdU AlexaFluor 594 Kit and following the manufacturer’s instructions (Invitrogen, Carlsbad, CA). Gates were placed based on levels of Sox9-EGFP expression and back gating was performed to assess the percentage of cells incorporating EdU in each Sox9-EGFP expressing cell population. Quantification of different Sox9-EGFP expressing populations was performed using a Cyan flow cytometer and Summit software v4.3 (Beckman Coulter, Brea CA). FACS on Sox9-EGFP cells for population analysis was performed using the MoFlow XDP cell sorter (Beckman Coulter, Brea, CA). FACS on Sox9-EGFP cells for single cell analysis was performed using the SH800Z cell sorter (Sony Biotechnology Inc., San Jose, CA). Immune cells were excluded from the sort by gating out CD31+ (BioLegend, San Diego, CA) and CD45+ (BioLegend, San Diego, CA) cells. Dead/dying cells were excluded from the sort by gating out SytoxBlue+ (ThermoFisher Scientific, Waltham, MA) and Annexin V+ (Invitrogen, Carlsbad, CA) cells. Cells were sorted into Sorting Medium composed of Advanced DMEM/F12, 100x N2 Supplement, 50x B27 Supplement, 1mM HEPES), 100x Penicillin/Streptomycin, 100x Glutamax all from (Gibco-Invitrogen, Carlsbad, CA) and supplemented with 10uM Y27632 (Invitrogen, Carlsbad, CA), and 1mM N-Acetylcysteine (NAC; Sigma-Aldrich, St. Louis, MO).

High throughput cell population qRT-PCR using Fluidigm

Total RNA was isolated from FACS sorted Sox9-EGFP populations using the RNAqueous-Micro Total RNA Isolation Kit (Ambion-ThermoFisher Scientific, Waltham, MA) according to the manufacturer’s instructions. RNA was quantified by measuring absorption at 260 nm using a Nanodrop photometer (Tecan, Morrisville, NC) and underwent DNase treatment to remove genomic DNA. Reverse transcription was performed on 25 μg RNA using Cells Direct One-Step qRT-PCR Kit (Invitrogen, Carlsbad, CA). Specific target amplification (STA) of 96 genes was performed on cDNA from the different FACS sorted populations isolated from Sox9-EGFP mice using Taqman primers purchased from Applied Biosystems. These target genes included population specific markers, genes that stimulate cell proliferation, apoptosis, DNA damage and maintain homeostasis. Preparation and loading of Fluidigm 2 x 48.48 Dynamic Array integrated fluidic circuit (IFC) was performed according to the manufacturer’s instructions (Fluidigm, South San Francisco, CA). Briefly, the chip was primed, loaded with samples and primer reaction mixes and run on the BioMark™ HD System. Data analysis was accomplished with the Fluidigm Real-Time PCR Analysis software. Resulting cycle threshold (Ct) values were analyzed by calculating the relative gene expression (fold change) using the 2^-ΔΔCt method [56]. β-actin was used as the reference gene to normalize mRNA input and allow accurate comparisons of results between samples. An unsorted, pooled RNA sample was set as the control sample to which all other samples were compared. GraphPad Prism 5 was used to perform a 1-way ANOVA with Bonferroni post-hoc test to test for differences in levels of mRNA across different cell populations and between cells from young and old mice.

Statistical analysis

Data are expressed as mean ± SEM. For histology morphometry, histology, flow cytometry and in vitro culture experiments, unpaired Student’s t-tests were performed to compare data from young and old mice. For high throughput qRT-PCR experiments one-way ANOVA with Bonferroni post hoc test was performed for comparisons of multiple groups. A P value of less than 0.05 was considered statistically significant.