1. Introduction
Leucine-rich repeat-containing G-protein-coupled receptor 5 positive (Lgr5
+) intestinal stem cells (ISCs), which are located at the bottom of intestinal crypts and herein are also called crypt base columnar (CBC) cells, function not only to generate other intestinal cell types via differentiation but also to maintain themselves via self-renewal. The careful balance between self-renewal and differentiation of ISCs allows for the constant turnover of the intestinal epithelium, which occurs every 3-5 days under normal homeostatic conditions [
1]. But in response to various forms of injury, ISCs can undergo limited differentiation with enhanced self-renewal to promote epithelial regeneration [
2]. This plasticity of ISCs is weakened during ageing for impaired intestinal function, whereas improved in tumor for the resistance to medical therapies. Although several signaling pathways have been reported to modulate such plasticity [
3], the mechanisms are more complicated and require further investigation.
Metabolic pathways have been widely acknowledged to adjust cells in response to circumstances changes, and thus metabolic reprogramming plays an important role in post-injury regeneration. Cells acquire energy mainly from two metabolic pathways: mitochondrial-dependent oxidative phosphorylation (OXPHOS) and mitochondrial-independent glycolysis. These two metabolic programs typically define different cell functions, with OXPHOS usually involved in homeostatic growth, while glycolysis is often associated with more-rapid cell growth, such as tumor formation. Along these lines, ISCs contain abundant mitochondria and display high mitochondrial activity, relying on OXPHOS for their function in homeostasis [
4], [
5], [
6], while the metabolic mode required for tissue regeneration driven by ISCs has not been determined. Increasing evidence has suggested that mitochondrial-based metabolic pathways involved in OXPHOS, such as fatty acid oxidation (FAO), also modulate ISC properties [
7]. Recently, FAO has been reported to be indispensable for ISC maintenance. FAO activation induced by fasting was found to improve ISC function but with no change in proliferation [
8]. In high fat diet (HFD)-associated tumorigenesis, FAO was considered responsible for ISC proliferation [
9]. In contrast, decreased FAO in ISCs caused by genetic alterations has been shown to promote proliferation and impair differentiation [
10], [
11]. Accordingly, although it is generally accepted that FAO plays an important role in the regulation of ISC fate, conflicts exist with regard to the complexity of FAO function under different circumstances, and it remains unclear how metabolic pathways are involved in ISC plasticity during epithelial regeneration.
For better understanding the ISC metabolic plasticity in response to injury as well as determining intervention strategy, it is an important issue to find the endogenous regulators that contribute to this biological process. To date, in addition to key enzymes directly participating in metabolic pathways, several metabolic regulators have been identified that drive ISC behavior [
12]. Among them, microRNAs (miRNAs) attract wide interest considering their unique roles in epigenetics, both flexible to react with environmental changes and efficient to be targets as one miRNA regulates a series of proteins. Recently, microRNA-277 has been reported to control FAO-related genes in ISCs [
13]. Meanwhile, it has been found that microRNA-31 acts in response to intestinal injury and thus drives ISC fate [
14], [
15]. These clues highlight the potential of miRNAs as modulators between metabolism and damage of ISCs. One family of miRNAs, microRNA-29 (miR-29), displays such potential. The three members of the miR-29 family, miR-29a, miR-29b, and miR-29c, share the same seed sequence. They are located on two different chromosomes, one of which contains
miR-29a and
miR-29b1, while the other contains
miR-29c and
miR-29b2 (
miR-29b1 and
miR-29b2 are from the same gene). The effects of miR-29a and miR-29b (miR-29a/b) on metabolic disorders have been shown to be related to lipid homeostasis via targeting the messenger RNAs (mRNAs) encoding enzymes or receptors controlling lipid absorption or breakdown in the heart, liver, lung, and kidney [
16], [
17], [
18], [
19]. These results imply that miR-29a/b are potential metabolic regulators, but their role in the metabolic regulation of intestinal homeostasis and regeneration has yet to be explored. Meanwhile, there are some
in vivo clues showing that miR-29a/b have an important function in alleviating intestinal inflammation via immunoregulation [
20], [
21], [
22], but the effects on intestinal epithelial cells have only been reported by
in vitro studies [
23], [
24], [
25], [
26]. In addition, our previous study found that supplementation with miR-29b could accelerate the recovery of intestinal epithelium after inflammatory damage in mice [
27], likely relevant to ISC biology but lacking in verification. Together, miR-29a/b present potential involving both metabolic regulation and intestinal epithelial regeneration, standing out as a possible regulator for ISC metabolic plasticity upon injury.
To investigate the role of miR-29a/b in ISC-mediated epithelial regeneration and to determine if miR-29a/b can regulate ISC fate via an effect on metabolic pathways, we created Mir29ab1-knockout (KO) mice. Analysis of these mice allowed us to identify a novel metabolic regulator of ISC fate under homeostatic and injury conditions.
2. Materials and methods
2.1. Mouse models
Male or female mice were housed six per cage (with the same gender) and provided with a standard diet in a specific pathogen free (SPF)-class housing laboratory belonging to China Agriculture University. For irradiation experiments, mice were exposed to γ-ray radiation from a 60Co source (12 Gy; Peking University, China). Animal work was approved (approval No. AW31012202-4-3) by the Laboratory Animal Welfare and Animal Experimental Ethical Inspection Committee of China Agricultural University. Villin-Cre and Lgr5-eGFP-IRES-CreERT2 transgenic mice were gifts from Zhengquan Yu’s laboratory (China Agricultural University). Mir29ab1flox/+ and Mir29ab1+/− transgenic mice were constructed by Model Organisms (Shanghai, China). Mir29ab1+/− strain were crossed to each other to obtain Mir29ab1+/+ mice and Mir29ab1−/− mice. The Mir29ab1flox/+ strain was crossed to each other to obtain Mir29ab1+/+ mice and Mir29ab1flox/flox mice. Mir29ab1+/+ and Mir29ab1−/− strains were crossed to Lgr5-eGFP-IRES-CreERT2 strain. Mir29ab1flox/flox strain was crossed to Villin-Cre strain to obtain Mir29ab1villin-KO mice. All the experimental mice had a C57BL/6N background. Mice were 3-5 months old when performing experiments. Littermate controls were used in this study.
2.2. Human samples
Twenty-three human fresh colorectal polypus and adjacent normal tissue samples were collected from patients undergoing colonoscopy in The Chinese People’s Liberation Army General Hospital, and the experiment was approved (CAUHR2021020) by the Human Research Ethics Committee of China Agricultural University. Tissues were immediately saved in liquid nitrogen.
2.3. Crypt isolation and organoid culturing
Crypts were isolated from small intestine of mice utilizing a protocol as previously described [
28]. Briefly, the small intestine was harvested from mice. After shave and washing, the intestine was digested in 5 mmol·L
−1 ethylenediaminetetraacetic acid (EDTA). Crypts were collected and passed through 70 μm cell strainers. The pelleted crypts were then resuspended with DMEM/F12 (Gibco, USA) and calculated to be seeded into 24-well plates at a density of 2000 crypts per well. Each well received crypts in a 50 μL droplet consisting of a 3:7 ratio of DMEM/F12 and Matrigel (Corning (USA), 356237) and bathing in 450 μL of Intesticult organoid growth medium (GM; StemCell (USA), 06005) with 1× penicillin-streptomycin-gentamicin. Primary organoids were formed after seven-day culturing of crypts.
For organoid passaging, droplets were destroyed with ice-cold phosphate-buffered saline (PBS), and the organoids were centrifuged (200g for 5 min), resuspended, pipetted up and down to disperse them, and seeded into 24-well plates in droplets in the same manner as the primary organoids.
2.4. Cell line culturing and transfection
LS174T cells (ATCC, USA) were cultured in minimum essential medium (MEM; Gibco, 11095080) supplemented with 1 mmol·L−1 sodium pyruvate (Gibco) and 10% fetal bovine serum (FBS; Biological Industries, Israel). HEK293T cells were cultured in Dulbecco’s modified eagle medium (DMEM; Gibco, 11995065) with 10% FBS. In both cases, 1× penicillin-streptomycin was added. For miR-29a/b overexpression or inhibition, cells were treated with Opti-MEM (Gibco) and transfected with 40 nmol·L−1 hsa-miR-29a-3p or b-3p mimics or inhibitor (Biolino, China) together with Lipofectamine 2000 reagent (Invitrogen, USA) for 6 h. Inhibitors are chemically modified RNAs that can bind to specific miRNAs and thus suppress their function on target genes. RNA was extracted after another 24 h incubation in culture medium while protein was extracted after 48 h. For Hnf4g knockdown, cells were transfected with 50 nmol·L−1 small interfering RNA (siRNA) (si-Hnf4g, GenePharma, China). For Hnf4g overexpression, plasmids containing human HNF4G DNA (pcDNA-Hnf4g, GenePharma) were established, and cells were transfected with 2 μg·mL−1 plasmids.
2.5. Quantitative polymerase chain reaction (Q-PCR)
TRIzol method was used for RNA extraction from tissues or cell lines and RNAprep Pure Micro Kit (TIANGEN, China) following the manufacturer’s instructions for isolated cells or organoids. RNA was converted to complementary DNA (cDNA) with 5× All-In-One RT MasterMix (Applied Biological Materials, Canada). The expression of target genes was analyzed using Q-PCR according to the testing protocol of TB Green Premix Ex Taq (Takara, Japan). Results were normalized by reference genes with comparable CT method. Primers employed are listed in Table S1 in Appendix A.
For miR-29, extracted RNA was converted in a 10 μL system with 1 μL of miR-29a/b-RT, 1 μL of U6-RT for 10 min at 70 °C and then converted for 1 h at 42 °C with 0.6 μL of deoxynucleotide triphosphate (dNTP), 0.6 μL of Moloney Murine Leukemia Virus Reverse Transcriptase (M-MLV), 1 μL of Recombinant RNasin Ribonuclease Inhibitor (RRI) and 1 μL of 5× M-MLV buffer (Promega, USA) added into the system. Primers employed are listed in Table S2 in Appendix A.
2.6. RNA-sequencing
Crypts were isolated and RNA was extracted. Illumina libraries were prepared using NEBNext® Ultra™ Directional RNA Library Prep Kit for Illumina® (USA). Concentration was first quantified with Qubit2.0 Fluorometer (Thermo Fisher Scientific, USA) and diluted to 1.5 ng·μL−1. After checking insert size by Agilent 2100 bioanalyzer (USA), accurate qualification was performed with Q-PCR to ensure the valid concentration over 2 nmol·L−1. Qualified libraries were then fed into HiSeq machines (USA) and sequenced by Illumina NovaSeq 6000. The reads were mapped to the reference genome with HISAT2 (v2.0.5). Differential expression analysis was conducted using DESeq2 R (v1.20.0). The whole process was performed at Novogene. The data have been deposited at the Gene Expression Omnibus (accession No. GSE216670). One of the samples from wild-type (WT) mice and another sample from Mir29ab1−/− mice were excluded from analysis.
2.7. Cell sorting
Isolated crypts from small intestine of Lgr5-GFP mice were incubated in Dispase II (Roche, Germany) with 100 U of DNaseI (TIANGEN) at 37 °C for 4 min and passed through 40 μm cell strainers to obtain single cells. After centrifugation at 500g for 5 min, cells were resuspended in 1 mL of PBS with 7-aminoactinomycin D (AAD) on ice for 15 min. Lgr5high, Lgr5low, CD24−Lgr5− and CD24+ side scatter (SSC)high Lgr5− cells were collected by flow cytometry (BD).
2.8. Cell cycle analysis
Cells were fixed with 70% ethanol for 2 h at −20 °C and stained with propidium iodide for 30 min at 37 °C using Cell Cycle and Apoptosis Analysis Kit (Beyotime, China). Then, cells went through flow cytometry (Beckman, USA) to determine the proportion of cells in different cell cycle phase.
2.9. Histological analysis
Small intestinal tissues were formalin-fixed, gradient-ethanol-dehydrated and paraffin-embedded. Sections of 5 μm were dewaxed, stained with hematoxylin for 30 s and eosin for 10 s. For goblet cell analysis, the dewaxed sections were subjected to periodic acid-Schiff (PAS) staining (LEAGENE, China) according to the manufacturer’s instructions.
2.10. Immunohistochemistry and immunofluorescence
After dewaxing and antigen retrieval, the sections were blocked and stained with primary antibodies at 4 °C overnight, followed by horseradish peroxidase (HRP)-conjugated (ZSGB-BIO, China) or florescence-labeled (Abcam, UK) antibodies. After incubation with diaminobenzidine and hematoxylin for immunohistochemistry, or 4′,6-diamidino-2-phenylindole (DAPI) (Beyotime) for immunofluorescence, the slides were observed using a microscope (Leica, Germany). Antibodies used for immunohistochemistry or immunofluorescence: Ki67 (Abcam, ab15580), OLFM4 (Cell Signaling Technology (USA), 39141), green fluorescent protein (GFP) (Cell Signaling Technology, 2956), lysozyme (Abcam, ab108508), mucin 2 (MUC2) (Servicebio (China), GB11344), chromogranin A (CHGA) (Abcam, ab15160), cleaved caspase3 (Cell Signaling Technology, 9664), sex-determining region Y-box 9a (SOX9) (Abcam, ab185230), and proliferating cell nuclear antigen (PCNA) (Santa Cruz Biotechnology (USA), sc-56).
2.11. EdU staining
Mice were treated with 50 μg·g−1 body weight (BW) of 5-ethynyl-2′-deoxyuridine (EdU) by intraperitoneal injection. Small intestines were processed to sections as described above and subjected to dewaxing. Washed sections were stained using BeyoClick™ EdU Cell Proliferation Kit (Beyotime) by incubation with click reaction buffer for 30 min, and Hoechst 33342 was employed for nuclear staining.
2.12. Alcian blue staining
Cells were fixed in 4% paraformaldehyde for 15 min and stained with Alcian blue (Solarbio, China) for 15 min. Fast red (Solarbio) was used for nuclear staining.
2.13. In situ hybridization
The dewaxed and hydrated sections were deproteinized with 10 μg·mL−1 proteinase K at 37 °C for 10 min and 0.1 mol·L−1 triethanolamine at room temperature for 10 min. For hybridization, the sections were equilibrated in prehybridization buffer at 60 °C for 2 h and hybridized with 30 nmol·L−1 probe (Exiqon, Danmark) at 65 °C for 16 h. The sections were then washed with 2× saline-sodium citrate (SSC; 20× SSC was 0.15 mol·L−1 sodium chloride and 0.017 mol·L−1 sodium citrate in RNase-free water, pH 9.5) at 65 °C (2 × 10 min), 50% formamide along with 2× SSC (3 × 30 min), and 1% Tween-20 in PBS (5 × 5 min), followed by incubation with anti-digoxigenin conjugated to alkaline phosphatase at room temperature for 2 h and staining with color developing agent (0.35% nitro blue tetrazolium (NBT), 0.35% 5-bromo-4-chloro-3-indolyl phosphate (BCIP) and 1.17 mmol·L−1 levamisole) in the dark for over 48 h. The samples were counterstained with nuclear fast red.
2.14. Western blot
Radio-immunoprecipitation assay (RIPA) buffer or the Nuclear and Cytoplasmic Protein Extraction Kit (Beyotime) was used for extraction of total protein or nuclear protein. After concentration regulation and boiling with 5 × loading buffer, protein was separated by sodium dodecyl sulfate (SDS)-polyacrylamide gel electrophoresis (PAGE), transferred onto polyvinylidene fluoride (PVDF) membranes, and incubated with primary antibodies. The blot was developed with the membranes treated by secondary antibody (Beyotime) and chemiluminescence reagent and exposed using Amersham Imager 600 instrument (GE, USA). The primary antibodies were as follows: PCNA (1:500, Santa Cruz Biotechnology, sc-56), single chain antibody 1 (SCA1) (1:2000, Abcam, ab109211), β-tubulin (1:2000, YEASEN (China), 30301ES60), carnitine palmitoyltransferase 1A (CPTA1) (1:1000, Abcam, ab234111), 3-hydroxy-3-methylglutaryl-coenzyme A (CoA) synthase 2 (HMGCS2) (1:2000, Abcam, ab137043), hepatocyte nuclear factor 4γ (HNF4γ) (1:1000, Novus (USA), NBP2-98898, reacted with mouse), HNF4γ (1:1000, Lsbio (USA), LS-C164948, reacted with human), pyruvate dehydrogenase kinase 4 (PDK4) (1:1000, Abcam, ab214938), cluster of differentiation 36 (CD36) (1:1000, Abcam, ab133625), acyl-CoA dehydrogenase medium chain (ACADM) (1:10000, Abcam, ab92461), histone H3 (1:2000, Cell Signaling Technology, 4499), and glyceraldehyde-3-phosphate dehydrogenase (GAPDH) (1:1000, Novus, NB300-211).
2.15. Dual-luciferase reporter assay
Human embryonic kidney 293T (HEK293T) cells were seeded in 24-well plates and cultivated with DMEM. The miRGLO control luciferase (miRGLO basic, Promega), miRGLO-wild-type Hnf4g 3′- untranslated regions (UTR) (Hnf4g-WT) and miRGLO-mut Hnf4g 3′- UTR (Hnf4g-Mut, with deletion of the putative miR-29a/b binding sites in the 3′-UTR of Hnf4g) luciferase plasmids were generated by GenePharma. Cells were transfected with 40 nmol·L−1 negative control (NC) mimics/inhibitor or miR-29a/b-3p mimics/inhibitor (GenePharma) and 50 ng plasmids using Lipofectamine 2000 reagent for 6 h. After 24 h, Dual-Luciferase Reporter Assay System (Promega) was used for the activities of luciferase, which were normalized by relative light unit of Renilla. The sequences of mimics, inhibitors, and plasmids are listed in Tables S3-S5 in Appendix A.
2.16. Seahorse analysis
Seahorse analysis of organoids was conducted according to a prior study [
29]. Briefly, organoids were passaged to the fourth culture, and four wells of them were seeded into 24 wells of the Seahorse XFe24 cell culture microplates (Agilent, 102340-100) after 36 h. Each well received a 3 μL droplet containing organoids. For the mitochondrial stress test, following another 36-hour incubation, culture medium was replaced by assay medium which contained 10 mmol·L
−1 glucose, 5 mmol·L
−1 pyruvate, and 2 mmol·L
−1 glutamine in Seahorse XF DMEM (pH 7.4, Agilent, 103575-100) for 1 h at 37 °C before measurement as follows: 5 μmol·L
−1 oligomycin (MCE (USA), HY-N6782), 2 μmol·L
−1 Carbonyl cyanide-
p-trifluoromethoxyphenylhydrazone (FCCP) (MEC, HY-100410), and 4 μmol·L
−1 rotenone (MCE, HY-B1756) were added in turn, and each phase contained three cycles (4 min mixing, 2 min measuring) except that oligomycin was injected followed by 5 min mixing, 10 min waiting, and 2 min measuring for the first cycle. For the glycolysis stress test, culture medium was replaced by assay medium consisting of 2 mmol·L
−1 glutamine in Seahorse XF DMEM (pH 7.4) and the drugs (10 mmol·L
−1 glucose, 5 μmol·L
−1 oligomycin, 100 mmol·L
−1 2-deoxy-
D-glucose (2-DG) (MCE, HY-13966)) were added for three measurements (4 min mixing, 1 min waiting, 2 min measuring; 5 min mixing, 10 min waiting, 2 min measuring for the first cycle and 4 min mixing, 2 min measuring for the other 2 cycles; 4 min mixing, 2 min measuring, respectively). For FAO stress test, medium was replaced by substrate-limited medium (0.667× B27 (Gibco), 1× N2 (Gibco), 50 ng·mL
−1 epidermal growth factor (EGF) (Peprotech (USA), 315-09-100), 100 ng·mL
−1 noggin (Peprotech, 250-38), 500 ng·mL
−1 R-spondin1 (Sinobiological (China), 50316-M08S), 0.5 mmol·L
−1 glucose, 1 mmol·L
−1 glutamine (Gibco), and 0.5 mmol·L
−1 L-carnitine (MCE, HY-B0399) in Seahorse XF DMEM) for 6 h (5% CO
2) and changed to assay medium consisting of 2.5 mmol·L
−1 glucose and 0.5 m mol·L
−1 L-carnitine for 45 min. Then, 100 mmol·L
−1 palmitate (Sigma, Germany) or bovine serum albumin (BSA) was added, and the assay was run according to the protocol of the mitochondrial stress test, except that 40 μmol·L
−1 etomoxir (MCE, HY-50202) was injected prior to oligomycin for three measurement cycles (4 min mixing, 2 min measuring). Oxygen consumption rates (OCR) as well as extra-cellular acidification rates (ECAR) was measured.
For the mitochondrial stress test on LS174T cells, assay medium consists of 10 mmol·L−1 glucose, 1 mmol·L−1 pyruvate, and 2 mmol·L−1 glutamine in Seahorse XF DMEM (pH 7.4). Measurement was as follows: 1.5 μmol·L−1 oligomycin, 1 μmol·L−1 FCCP, and 1 μmol·L−1 rotenone were added in turn, and each phase contained three cycles (3 min mixing, 2 min waiting, 3 min measuring).
2.17. 13C-palmitate labeling and LC-MS methods
Metabolic products were labeled and analyzed as previously described [
9]. The isolated crypts or transfected LS174T cells were incubated in 500 μL of Roswell Park Memorial Institute (RPMI) 1640 medium (no glucose nor glutamine, Gibco) with 100 μmol·L
−1 13C-palmitate (Sigma, 605573) per well of 24-well plates for 1 h at 37 °C, 5% CO
2. After washing by PBS and centrifugation (200
g for 5 min), liquid chromatography-mass spectrometry (LC-MS) grade methanol, acetonitrile and water (1:1:2) was added together with chemical standards of substances to be detected, followed by being vortexed for 10 min and centrifugated for 10 min at 4 °C. The spun down was dried and resuspended with 100 μL of LC-MS grade water. The LC-MS analysis was performed using a zwitterionic hydrophilic interaction liquid chromatography (ZIC-pHILIC) column (2.1 mm × 150 mm, 5 μm; EMD Millipore, Germany). Buffer A was 20 mmol·L
−1 NH
4HCO
3, 0.1% NH
3·H
2O; buffer B was acetonitrile. The proportion of buffer B followed a linear gradient from 80% to 20% for 20 min, then from 20% to 80% for 30 s, and finally hold at 80% for 9.5 min. The flow rate was 0.15 mL·min
−1. The mass spectrometer was operated with a Q Exactive orbitrap instrument (Thermo Fisher Scientific) in muti reaction monitoring (MRM), and the conditions were set as follows: spray voltage as 4500 V (negative) and 5500V (positive), temperature as 500 °C, ion source gas1/2 as 55 psi (1 psi = 6894.76 Pa). Standards were purchased from Sigma.
2.18. Fatty acid uptake and consumption
LS174T cells were cultured upon glass slides and treated with 100 μmol·L−1 BOTIBY-labeled palmitate (Invitrogen, D3821) for 12 h after transfection of NC inhibitor or miR-29a/b-3p inhibitor. Then, medium was changed to be fresh, and images were taken after another 24-hour culturing.
2.19. DNA copy number measurement
Total DNA of organoids was extracted with TIANamp Micro DNA Kit (TIANGEN). The copy number of mitochondrial DNA (mtDNA) was quantified using Q-PCR through amplification of mtDNA fragments (Atp6, Nd2, Cytb, Dl) against nuclear gene β-actin. Primers of mtDNA are listed in Table S1.
2.20. JC-1 staining
Organoids were cultured upon glass slides and stained on the third day after passaging. 10 μg·mL−1 5,5′,6,6′-tetrachloro-1,1′,3,3′-tetraethylbenzimidazolylcarbocyanine iodide (JC-1) (Invitrogen, T3186) was injected into the culture medium and incubated at 37 °C for 20 min. Then, the organoids were uncovered with fresh medium for live imaging using a Zeiss confocal microscope and Structure Illumination Microscopy (Dong Li’s Lab, Institute of Biophysics, Chinese Academy of Sciences).
2.21. Electron microscopy
The intestinal tissues were fixed in 2.5% glutaraldehyde (Solarbio) for 1 h at room temperature and then placed at 4 °C. They were further fixed in 1% OsO4 for 4 h and went through dehydration with concentration-increasing acetone (50%, 70%, 80%, 90%, 95% for 15 min, respectively; 100% for 20 min twice). After embedding and polymerization (37 °C for 12 h, 45 °C for 12 h, 60 °C for 48 h), samples were cut into 70-100 nm-thick sections and stained with 2% uranyl acetate followed by lead citrate. Images were obtained with a JEM-1200EX electron microscope (JEOL, Japan).
2.22. Adenosine triphosphate (ATP) measurement
The cell ATP level was measured 48 h after transfection using Enhanced ATP Assay Kit (Beyotime). Relative luminescence unit of supernatant reacted with assay buffer was detected using an Infinite M200 Pro multi-functional microplate reader (Tecan, Switzerland). The amount of ATP was normalized with protein concentration using bicinchoninic acid (BCA) assay (Beyotime).
2.23. Metabolomics
Crypts were isolated for small intestine and frozen in liquid nitrogen. After vortexing for 30 s in 200 μL of water, samples were frozen and thawed three times and sonicated in ice for 10 min. According to the measured protein concentration, samples were moved to tubes with water and 450 μL of methanol (containing isotope labeled internal standard mixture) added and incubated at 40 °C for 1 h. Then, the samples were centrifuged at 13 800g at 4 °C for 15 min. The supernatant was injected onto an ultra-performance liquid chromatography (UPLC) using high strength silica (HSS) T3 2.1 mm × 100 mm column (1.8 μm particle size) in UHPLC system (Vanquish, Thermo Fisher Scientific). Buffer A was 5 mmol·L−1 ammonium acetate, 5 mmol·L−1 acetic acid; buffer B was acetonitrile. The mass spectrometer was operated with Orbitrap Exploris 120 mass spectrometer (Orbitrap MS) with resolution set to 60 000 (full MS) and 15 000 (MS/MS), capillary temperature held at 320 °C, collision energy to be 10/30/60 in normalized collision energy (NCE) mode, spray voltage set to 3400 V (negative) and 3800 V (positive), sheath gas flow rate to be 50 arbitrary units (Arb) and auxiliary (Aux) gas flow rate to be 15 Arb. The whole process was performed at Biotree (China).
2.24. Statistical analysis
All the experiments were conducted three times independently. For organoid assays and in vivo studies, samples from at least three different mice were used in each experimental group. Results were all presented as mean ± standard deviation (SD), except the change of miR-29a/b and Hnf4g expression in different days post irradiation (IR), which was presented as mean ± standard error of the mean (SEM), and the survival data that were presented by Kaplan-Meier plot. Data analysis was carried out with SPSS (IBM, USA), Excel (Microsoft, USA) and GraphPad Prism (GraphPad Software, USA). Image J (National Institutes of Health, Germany) was used for blot and image analysis. FlowJo 10.4 (FlowJo LLC., USA) was used for flow cytometry analysis. GSEA 4.1.0 (Broad Institute, USA) was used for gene set enrichment analysis (GSEA). The significance of the differences between two groups was analyzed by Student’s t-test. Significant differences were established at the level of P < 0.05. *P < 0.05, **P < 0.01, ***P < 0.001. Non-significant was represented by ns. For multiple groups, the significance was analyzed with Duncan’s post hoc test following one-way analysis of variance (ANOVA). Significant differences were established at the level of P < 0.05 and represented by a, b. Animals were randomly assigned to groups based on genotype. The animal experiments were not carried out in a blinded fashion.
2.25. Data availability
Uncropped blots were deposited to Mendeley (DOI: 10.17632/8kkgwgy4xj.1).
3. Results
3.1. MiR-29a/b are required for epithelial regeneration of the small intestine
Colorectal cancer (CRC) is initiated from hyperplastic adenoma associated with expanded intestinal epithelial stemness. According to the miRNA sequencing data from The Cancer Genome Atlas-Colon Adenocarcinoma/Rectal Adenocarcinoma (TCGA-COAD/READ) cohort, we found that both hsa-miR-29a-3p and hsa-miR-29b-1-3p highly express in tumors compared to normal tissues (
Fig. 1(a)). In addition, the survival analysis of the prognostic value after radiation treatment displayed worse outcomes in CRC patients with higher expression of these two miRNAs (
Fig. 1(b)). Among different stages of COAD and READ, the expression of miR-29a/b remained at a stable high level from the beginning of tumorigenesis and became even higher at Stage IV, the most severe stage (
Fig. 1(c)). Moreover, miR-29a/b accumulate in tumors with high purity (
Fig. 1(d)). We identified numerous genes correlated to miR-29a/b expression in COAD and READ, and a large amount of the correlated genes belonging to mitochondrial biological processes based on the gene ontology (GO) analysis (
Fig. 1(e)). Furthermore, GSEA points out energy and lipid metabolism may change with miR-29a/b expression. In view of these statistical analysis, miR-29a/b expression is positively related to stemness upregulation and negatively related to prognostic value of radiotherapy, and they are likely to regulate intestinal metabolic processes.
The results above indicated that miR-29a/b are important in intestinal stemness and post-destruction reoccurrence, thus we speculated that miR-29a/b also contribute to epithelial regeneration mediated by ISCs. High conservation of miR-29a/b among different species (
Fig. 2(a)), particularly among mammals, makes mice an ideal model organism to study these two miRNAs. To clarify the role of miR-29a/b in murine intestine, we first performed
in situ hybridization and found that intestinal miR-29a/b are predominantly expressed in the epithelium, and compared to villi, the expression levels are generally higher in the crypts of the small intestine (SI), which mainly contain ISCs and transit amplifying cells (
Fig. 2(b)). Next, we applied 12 Gy of γ-ray irradiation (γ-IR) to destroy ISCs and examined miR-29a/b expression in response to such injury. MiR-29a/b expression remained at a low level within 24 hours (DNA damage/apoptosis period), but the expression was sharply upregulated in regenerated crypts 2-3 days post-IR (initiation of regenerative proliferation from the refilled ISCs), and finally returned to baseline levels five days after initial injury (switch towards differentiation and epithelium reconstruction) (
Figs. 2(c)-(e)). These data suggested that miR-29a/b might function in intestinal crypt cells and play a role in epithelial regeneration post-injury.
To determine such a role, we generated mice lacking miR-29a/b by genetic deletion of
Mir29ab1 (Figs. S1(a)-(c) in Appendix A), which resulted in no significant difference in intestinal structure compared to WT mice (Fig. S1(d) in Appendix A), but it did lead to a lower rate of epithelial turnover mediated by ISCs during homeostasis in the mutant mice (
Figs. 2(f) and
(g)).
Next, to further characterize the effect of miR-29a/b in intestinal epithelial regeneration, we utilized the 12 Gy of γ-IR injury model in WT and
Mir29ab1−/− mice (
Fig. 3(a)).
Mir29ab1−/− mice displayed more weight loss and lower survival rate than WT mice (
Figs. 3(b) and
(c)). Histological analysis showed an impaired regenerative response from three days post IR in
Mir29ab1−/− mice compared to WT mice (
Fig. 3(d)). Further, a reduction in the number and the proliferation of regenerated crypts was found in the mutant mice (
Figs. 3(e)-(g); Fig. S2(a) in Appendix A). As the origin of these proliferating crypts, the number of regenerated ISCs was also significantly lower in the mutant mice at the beginning of repair period (
Figs. 3(h) and
(i)). These results were further confirmed by lower expression of proliferation-related and CBC markers in the mutant mice (
Figs. 3(j) and
(k)), and there was also a decreased expression of repair-related markers (
Sca1,
Clu,
Anxa1) and epithelial tight junction markers (
Figs. 3(j) and
(k)). Consequently, the number of differentiated epithelium cells in regenerated crypts, and thus the ability to reconstruct epithelial structure with adequate crypt cells, was lower in
Mir29ab1−/− mice than in WT mice five days after IR (
Figs. 3(l) and
(m); Figs. S2(b)-(d) in Appendix A).
As the stroma and other organs apart from the intestine could also exert an impact on epithelial regeneration, we genetically deleted
Mir29ab1 specifically in the intestinal epithelium via the use of
Villin-Cre, thus creating
Mir29ab1villin-KO mice, to exclude the effect of miR-29a/b via other tissues (
Fig. 3(n)). We found the effects on IR in the epithelium-specific-knockout mice were consistent with those in the global-knockout mice (
Figs. 3(o)-(z)). Briefly, both the number of regenerative ISCs, as well as proliferative cells, and the differentiated cells originating from regenerated ISCs, were lower in
Mir29ab1villin-KO mice compared to WT mice, with the villi being shorter at the end of repair period resulting from a lack of source for epithelial reconstruction (
Figs. 3(q) and
(r)). Taken together, our results demonstrate that miR-29a/b play a direct role in promoting ISC-mediated epithelial regeneration of the SI and mainly function in regenerated crypts in response to IR.
3.2. Deletion of miR-29a/b contributes to impaired stemness of ISCs
Lgr5
+ ISCs support both the proper turnover of intestinal epithelium in homeostasis and epithelial reconstruction after self-regeneration when facing damage. As we found that miR-29a/b are highly expressed in ISC-enriched regenerated crypts and are required for both epithelial turnover in states of homeostasis and epithelial repair post injury, we questioned if miR-29a/b regulate the fate of ISCs. Firstly, we found that
Mir29ab1 deletion resulted in a significant downregulation in the expression of the CBC stem cell markers (
Lgr5 and
Ascl2) and reserved stem cell markers (
Dclk1,
Lrig1,
Bmi1, and
Msi1) (
Fig. 4(a)). Also, we employed the
Lgr5-GFP mice to label ISCs (
Fig. 4(b)), and we indeed found that in the SI of
Mir29ab1−/− mice, the Lgr5
+ cell population was smaller compared to that of WT mice (
Figs. 4(c)-(f)), which supports the idea that
Mir29ab1 deletion results in the loss of Lgr5
+ ISCs. We further found that the percentage of crypt cells in the G0/G1 phase was greater in the mutant mice (
Fig. 4(g); Fig. S3(a) in Appendix A), indicating less cell division and thus impaired self-renewal of ISCs.
To assess ISC function in epithelium regeneration in the absence of environmental impacts, we assayed the SI organoid-forming capacity of crypts isolated from WT and
Mir29ab1−/− mice. Notably,
Mir29ab1 deletion in crypts restrained the forming potential of mini-SI along with less budding organoids (
Figs. 4(h)-(j)), and limited daughter organoids cultured from primary ones (Fig. S3(b) in Appendix A). To further evaluate the regenerative capacity of ISCs, we employed Wnt3a (WENR medium) to obtain spherical organoids (crypts disappeared) whose stemness (relevant to ISC number and self-renewal potential) was destroyed. We then changed the WENR medium to standard (ENR) culture medium to assay the potential of crypt reestablishment (
Fig. 4(k)). After modeling with WENR, a switch from spherical phenotypes to budding ones was observed during ENR culturing in WT organoids, while the crypt formation was diminished in
Mir29ab1 deletion organoids (
Figs. 4(l) and
(m)). In addition, the expression of CBC cell markers was lower in
Mir29ab1-null organoids compared to WT organoids (Fig. S3(c) in Appendix A), consistent with the results
in vivo. While the number of Lgr5
+ cells in the mutant organoids was lower (
Figs. 4(n)-(p)), more differentiated cells, including Paneth cells and goblet cells, were found compared to the WT organoids (
Figs. 4(q) and
(r); Figs. S3(d) and (e) in Appendix A), indicating a loss of stemness due to skewing from self-renewal towards differentiation.
To determine the effect of the miR-29a/b on ISCs directly, we transfected human LGR5
+ LS174T cells [
1] with inhibitors of both miR-29a-3p and miR-29b-3p (
Fig. 4(s); Fig. S3(f) in Appendix A) and found more cells in the G0/G1 phase compared to use of a control (
Fig. 4(t)). This effect of miR-29a/b inhibition could account for the lower number of ISCs and slower regeneration rate seen in the mutant mice. Also, we found that there was enhanced differentiation of the LGR5
+ cell line after treatment with miR-29a/b inhibitors (
Figs. 4(u) and
(v)), showing a switch from self-renewal towards differentiation. In summary, lack of miR-29a/b caused impaired stemness with reduced ISC self-renewal via cell cycle arrest that was associated with a greater degree of differentiation.
3.3. Lack of miR-29a/b leads to altered mitochondrial metabolism in ISCs
To gain mechanistic insight into how miR-29a/b participate in the regulation of ISC stemness, we carried out RNA sequencing analysis on SI crypts from WT and
Mir29ab1−/− mice (
Figs. 5(a)-(d)). By GO analysis we found that the genes with altered expression were involved in metabolic processes, especially metabolite transportation (
Fig. 5(e)), and that differentially expressed genes between crypts from WT mice and those from
Mir29ab1−/− mice contained numerous targets related to metabolism of nutrients (
Fig. 5(f)). GSEA showed an association with mitochondrial metabolism, especially with regard to OXPHOS (
Fig. 5(g); Fig. S4(a) in Appendix A).
Mitochondria are abundant in ISCs, and mitochondrial metabolism defines the behavior of ISCs during self-renewal and differentiation. To elucidate the underlying relationship between miR-29a/b and mitochondrial energy metabolism in ISCs, we first focused on energy production. Treatment with either an miR-29a or an miR-29b inhibitor resulted in greater ATP generation in LGR5
+ cells (
Fig. 5(h)), which is likely mainly derived from OXPHOS. We then further evaluated the organoids for OXPHOS (
Fig. 5(i)) and found mitochondrial respiratory capacity, as well as oxygen consumption associated with OXPHOS, was greater in
Mir29ab1-null organoids (
Figs. 5(j)-(m)) and miR-29a/b-inhibited LGR5
+ cells (Figs. S4(b)-(d) in Appendix A) compared to controls. Consistent with this result, the live imaging of mitochondria for measuring the mitochondrial membrane potential (MMP) showed enhanced power for OXPHOS in crypt cells that lacked
Mir29ab1 (
Figs. 5(n)-(p); Movie S1).
Based on these findings, we surmised that
Mir29ab1-deletion induces altered mitochondrial function. Thus, we analyzed the ISC mitochondrial morphology by electron microscopy but found no difference between mitochondria in WT and those in
Mir29ab1−/− organoids (
Fig. 5(q); Fig. S4(e) in Appendix A). In addition, mtDNA copy number were similar between mutant and WT organoids and between LS174T cells treated with inhibitors of miR-29a/b and those treated with a control (Fig. S4(f) and (g) in Appendix A). These results illustrate that loss of miR-29a/b promotes OXPHOS but posed no significant impact on mitochondrial structure or mtDNA content.
3.4. Depletion of miR-29a/b boosted FAO in crypts and ISCs
As OXPHOS was stimulated by miR-29a/b deletion without a change in mitochondrial structure or mtDNA copy number, we hypothesized that alterations in metabolic pathways that provide electrons for mitochondrial respiration and drive OXPHOS might explain the effects of the depletion of the miRNAs. Metabolic pathways contributing to OXPHOS mainly consist of pyruvate oxidation and FAO. Based on principal component analysis, we found that the metabolome of SI crypts isolated from WT and
Mir29ab1−/− mice depicted different clustering of metabolic profiles (
Figs. 6(a) and
(b)). Most of the metabolites we obtained showing a significant difference in flux in
Mir29ab1-null crypts compared to WT crypts belonged to lipids or lipid-like molecules (
Figs. 6(c)-(f); Fig. S5(a) in Appendix A). Even other molecules in the differential metabolites were highly correlated with the lipid-like molecules (
Fig. 6(g)). Moreover, there were less long-chain fatty acids, which can be consumed by mitochondrial metabolism via FAO, in
Mir29ab1-null crypts, while there were more phosphatidylcholines that can accelerate FAO in the mutant crypts (
Fig. 6(h)). Likewise, based on the RNA-sequencing analysis, we discovered a trend in elevated mRNA levels for genes involved in FAO, while those involved in pyruvate oxidation were lower in the mutant crypts (Figs. S5(a)-(c) in Appendix A). Consistent with these findings, GO analysis of the RNA-sequencing data highlighted a change in fatty acid metabolism upon
Mir29ab1 deletion (
Fig. 6(i)). The results were confirmed by expression of PDK4, witch functions inhibit the conversion of pyruvate to acetyl-CoA, thereby inhibiting the glycolysis-derived acetyl-CoA into tricarboxylic acid (TCA) cycle and pulling FAO-generated acetyl-CoA into TCA cycle (Fig. S5(a)). The expression of PDK4 was elevated in mutant crypts and organoids, as well as in LS174T cells treated with either an miR-29a or miR-29b inhibitor (Figs. S5(d)-(g) in Appendix A). Taken together, FAO activation might be responsible for the higher OXPHOS levels observed in the absence of miR-29a/b function.
Then, we performed the following experiments to verify such hypothesis (
Fig. 6(j)). We performed Seahorse analysis of organoids treated with etomoxir, an irreversible inhibitor of FAO. We found an increase in the sensitivity to etomoxir of
Mir29ab1-null organoids compared to that of WT organoids (
Fig. 6(k)), indicating a higher independence for FAO to drive OXPHOS. To further verify the FAO elevation, we undertook a labeled substrate metabolomics study to evaluate the utilization of substrates via FAO. After incubation with [U-
13C] palmitate,
Mir29ab1−/− crypts had higher fractional labeling of (M+2) acetylcarnitine and (M+2) citrate (
Fig. 6(l)), both of which derive from palmitate via FAO. To identify whether the FAO activation took place in ISCs, we detected FAO-related targets in isolated Lgr5
high cells from crypts, whose expression became higher in
Mir29ab1−/− ISCs (
Fig. 6(m); Fig. S5(h) in Appendix A), while the effect was not noticed in Lgr5
− cells (Fig. S5(i) in Appendix A). The targets encode important FAO-related enzymes. Carnitine palmitoyltransferase 1A (Cpt1A) is a regulator that functions to transfer active long-chain acyl-CoA to mitochondria for oxidation. Acyl-CoA dehydrogenase, medium chain specific (Acadm) is responsible for acyl-CoA oxidation. 3-Hydroxy-3-methylglutaryl-CoA synthase 2 (Hmgcs2) functions in ketone body synthesis following FAO. The same results were also found in the LGR5
+ LS174T cells where miR-29a/b were inhibited (
Figs. 6(n)-(p)). Lastly, we employed fluorescent palmitate to treat LS174T cells for 12 h and refreshed the medium for another 24 h of culturing. There was less fluorescent palmitate left in cells subjected to miR-29a/b inhibitors (
Figs. 6(q) and
(r)), indicating improved consumption of palmitate via FAO. Thus, miR-29a/b depletion or their inhibition contributed to FAO activation in ISCs. To ascertain whether elevated FAO was sufficient for OXPHOS promotion, we employed an FAO agonist to treat LS174T cells (Fig. S5(j) in Appendix A) and found FAO activation could enhance the degree of OXPHOS (
Figs. 6(s) and
(t)).
Apart from FAO, we also paid attention to glycolysis, the other pathway important for cellular energy metabolism. The MiR29ab−/− organoids displayed an enhanced preference to converting glucose into lactate instead of pyruvate oxidation (Figs. S5(k)-(m) in Appendix A), excluding the contribution of glycolysis to enhanced OXPHOS. Moreover, in absence of glucose, Mir29ab1-null organoids still showed higher mitochondrial respiration related to OXPHOS (Figs. S5(n) and (o) in Appendix A). These results highlight that FAO is the key pathway accounting for higher mitochondrial OXPHOS in ISCs.
To confirm that enhanced FAO is responsible for miR-29a/b-mediated regulation of ISCs, we analyzed the effect of FAO activation on ISC stemness. We found that activation of FAO was associated with greater LS174T cell differentiation (
Figs. 6(u) and
(v)). However, despite the promotion of FAO (Fig. S5(j) in Appendix A), when OXPHOS was inhibited (Fig. S5(p) in Appendix A), the increase in differentiation declined (
Figs. 6(w) and
(x)), reinforcing the notion that miR-29a/b-deficiency-mediated regulation of LGR5
+ cell stemness occurs via elevated FAO and its promotion of OXPHOS.
Next, we asked whether elevated FAO and its promotion of OXPHOS was indispensable for miR-29a/b-mediated post-injury regeneration. To investigate how miR-29a/b regulates ISCs during the post-IR repair period, we conducted RNA-sequencing analysis on SI regenerated crypts from WT and
Mir29ab1−/− mice three days after IR (
Fig. 7(a)). We found a notable change in the expression of genes associated with metabolic pathways as determined by Kyoto Encyclopedia of Genes and Genomes (KEGG) analysis (Fig. S6(a) in Appendix A). Notably, the expression of genes that were significantly upregulated in SI regenerated crypts from
Mir29ab1−/− mice compared to those from WT mice post IR showed an enrichment in pathways related to ATP production or mitochondria (
Fig. 7(b)), as well as fatty acid metabolism (
Fig. 7(c)). Similarly, targets in peroxisome proliferator-activated receptors (PPAR) pathways, which regulate FAO activity, showed a marked accumulation in
Mir29ab1-depleted cells compared to WT cells (
Figs. 7(d)-(g)). Next, we validated the induction of selected enzymes involved in FAO at the protein level in the IR-damaged crypts (
Fig. 7(h); Figs. S6(b) and (c) in Appendix A). Further, we detected a change in expression of intestinal epithelial FAO-related enzymes after injury, with a reduction in FAO three days post IR (
Fig. 7(i)), which correlated with the increase of miR-29a/b expression at that time point, while also accounting for the impaired repair of intestinal epithelium in
Mir29ab1−/− mice post IR.
Together, we found that a lack of miR-29a/b results in enhanced FAO-mediated OXPHOS in ISCs, thus causing their impaired stemness. We also infer that after injury induction miR-29a/b results in lower FAO and OXPHOS that leads to accelerated ISC regeneration.
3.5. MiR-29a/b regulate ISCs through negatively targeting Hnf4g
MiRNAs exert bioactivities through downregulating gene translation by targeting the 3′-UTR of their target genes to induce their degradation. To specifically seek out the target(s) of miR-29a/b involved in their reprogramming of ISC metabolism, and their regenerative competence, we performed
in silico analysis to identify the potential target genes within differentially expressed genes between SI crypts from WT and
Mir29ab1−/− mice three days after IR injury. According to three databases (Targetscan, miRD8, PicTar),
Hnf4g is the only gene that contains a predictable miR-29a/b consensus sites in the 3′-UTR among all the differentially expressed genes (
Fig. 8(a)), and the binding sites are conserved among various species (Fig. S7(a) in Appendix A). To validate that miR-29a/b targets
Hnf4g we performed a luciferase assay in HEK293T cells and found that mimics/inhibitors of miR-29a/b reduced/enhanced the level of a WT form of
Hnf4g but did not affect the levels of a mutant form of the gene lacking the seed sequence for these miRNAs (
Fig. 8(b)). Interestingly,
Hnf4g was much lower in Lgr5
high ISCs compared to Lgr5
− cells, but
Mir29ab1-null Lgr5
high ISCs had a similar level of
Hnf4g with Lgr5
− cells from WT mice (
Fig. 8(c)), indicating that miR-29a/b modulates
Hnf4g levels especially in Lgr5
+ ISCs rather than other epithelial cells. Next, we found in human intestinal biopsy specimens that
HNF4G expression was negatively associated with
MIR29A/B levels (
Figs. 8(d) and
(e)). In mouse organoids, knockout of
Mir29ab1 led to higher expression of Hnf4γ (Fig. S7(b) in Appendix A). Simultaneously, we found a higher expression of
HNF4G/HNF4γ in LS174T cells when miR-29a/b were inhibited (
Fig. 8(f); Fig. S7(c) in Appendix A). Importantly, as a transcription factor, HNF4γ accumulated in the nuclear fraction upon miR-29a/b inhibition (
Figs. 8(g) and
(h)).
Notably,
Mir29ab1 deletion enhanced Hnf4g/Hnf4γ expression in crypts three days post IR (
Figs. 8(i) and
(j); Figs. S7(d) and (e) in Appendix A), while the expression of
Hnf4g in regenerated crypts post IR was sharply lower compared to that in crypts without injury in WT mice (
Fig. 8(k)), contrary to that of miR-29a/b and consistent with that of FAO-related enzymes. These results imply a possible role of Hnf4γ in intestinal regeneration mediated by miR-29a/b and its inhibition of FAO during this process.
To determine whether HNF4γ is responsible for the metabolic modulation of ISCs, the induction of FAO-related proteins following
HNF4G overexpression (Fig. S7(f) in Appendix A) was recapitulated in LGR5
+ LS174T cells transfected with plasmids containing
HNF4G DNA (
Fig. 8(l); Fig. S7(g) in Appendix A). We found that
HNF4G overexpression gave rise to higher basal mitochondrial respiration and OXPHOS level in LS174T cells (
Figs. 8(m) and
(n)), which simulated the effect of miR-29a/b inhibition. Finally,
HNF4G upregulation showed a similar effect as miR-29a/b inhibition in driving LS174T cells towards differentiation (
Figs. 8(o) and
(p)).
To clarify whether targeting
HNF4G is sufficient for the consequences brought about by depletion of miR-29a/b in ISCs, we investigated whether knockdown of
HNF4G could abolish the effect of miR-29a/b suppression (Fig. S7(j) in Appendix A). The induction of FAO-related proteins upon miR-29a/b inhibition was normalized upon
HNF4G siRNA-mediated depletion (
Fig. 8(q)). Similarly, the changes in mitochondrial respiration associated with OXPHOS of LS174T cells upon miR-29a/b inhibitor treatment were normalized upon
HNF4G knockdown (
Figs. 8(r) and
(s)). Moreover, upon siRNA-mediated interference of
HNF4G expression, miR-29a/b inhibition failed to cause over-differentiation (
Figs. 8(t) and
(u)). Overall, these data show that the function of miR-29a/b in evoking a metabolic program to intervene with ISC stemness is dependent on Hnf4g.
4. Discussion
The regulation of the balance between self-renewal and differentiation among ISCs under different biology processes is essential for intestinal health. A number of studies have shown the occurrence of metabolic regulation of ISCs in homeostasis, but lack of them focus on how metabolism is involved in rapid ISC self-refilling, and thus the tissue regeneration. Moreover, regarding to colorectal cancer treatment, therapies typically include radiation, chemotherapy and surgery, while metabolic plasticity of ISCs provides tumor with radio/chemoresistance and cause reoccurrence. The effective targets for controlling such plasticity remain unclear. In addition, during ageing, the turnover rate of intestinal epithelium is reduced and bring increased risk for intestinal barrier injury as well as function loss due to the impaired regenerative response. In the present study, we found that ISC-driven regeneration requires the induction of miR-29a/b to downregulate FAO-mediated OXPHOS by targeting
Hnf4g, and that this metabolic alteration fuels the renewal of the ISC population at the expense of their differentiation (
Fig. 9). Accumulation of these two miRNAs in adenocarcinoma is also associated with worse prognostic value of CRC patients, which may attribute to reoccurrence of tumors after radiotherapy. Our findings indicate the metabolic plasticity of depleted ISCs and reveal potential targets for improving mucosal healing in disease and ageing or preventing tumor revival.
4.1. MiR-29a/b act as the switch between self-renewal and differentiation
Lgr5
+ ISCs drive intestinal epithelial reconstruction following tissue damage. Nevertheless, considering the high sensitivity to injury and massive destruction, the pool of Lgr5
+ ISCs needs to be refilled [
30], which comes from quiescent ISCs and dedifferentiated progenitor cells or mature epithelium cells [
31], [
32], [
33], [
34], [
35], [
36]. It implies that the switch from differentiation to self-renewal of ISCs is indispensable for the regeneration. In the present work, we found that miR-29a/b deficiency leads to impaired stemness of ISCs as determined by their excessive differentiation at the expense of self-renewal. The skewing towards ISC differentiation might be related to reduced dedifferentiation upon injury, accounting for the weakened repair of damaged epithelium and reduced cell number in the Lgr5
+ pool. In addition, in view of our result that miR-29a/b were upregulated in regenerated crypts during intestinal epithelial post-injury repair, we infer that upregulation of miR-29a/b enhances ISC self-renewal by driving dedifferentiation.
Next, we wonder how miR-29a/b contributes to this skewing. Given that Lgr5
+ ISCs actively cell cycle, they reside predominantly in the S or G2/M phase, and their constant asymmetric division maintain a balance between self-renewal and differentiation [
37]. In contrast, cells in the G0/G1 phase take longer to divide and can finally enter differentiation. Here, we found that a deficiency of miR-29a/b resulted in more crypt cells or LGR5
+ cells in the G0/G1 phase, implying a lower rate of cell division and more opportunities for differentiation, and thus accounting for reduced Lgr5
+ ISC stemness as well as their slower migration. Nevertheless, how miR-29a/b regulates the cell cycle was not immediately obvious. Reports have demonstrated that cell cycle progression is related to metabolic control [
38], [
39], thus cell cycle arrest might attribute to miR-29a/b-mediated metabolic reprogramming.
4.2. Temporary inhibition of FAO/OXPHOS axis contribute to refilling of ISCs
As the canonical signals that determine ISC fate, including wingless-related integration site (WNT), bone morphogenetic protein (BMP), and Notch [
40], [
41], [
42], [
43], were excluded (data not shown), we paid attention to the indispensable position of metabolic regulation in miR-29a/b function. ISC differentiation has been reported to be tightly associated with mitochondrial metabolism [
5], [
6]. During differentiation, mitochondrial activity, mtDNA copy number and reactive oxygen species level dramatically increase in ISCs as OXPHOS drives ISC differentiation. On the other hand, inhibition of OXPHOS by knockout of ISC mitochondrial pyruvate carrier 1 (Mpc1) was found to improve ISC stemness while overexpression of Mpc1/Mpc2 had the converse effect [
44], [
45], in agreement with our result that miR-29a/b depletion augmented ISC differentiation but restrained stemness with OXPHOS elevated. In addition, a previous study examining Sox9-EGFP
high/Lgr5
low progenitors in regeneration showed that they were stimulated to hyperproliferate in post-IR repair, which was accompanied by a lower expression of OXPHOS-related genes, although actual changes in respiration were not measured [
46]. These results, as well as ours, indicate that in proliferation-featured intestinal epithelial repair, OXPHOS is suppressed to promote regeneration. Taken together, upregulation of miR-29a/b after damage to the intestinal epithelium may downregulate OXPHOS in ISCs to allow for their proliferation and dedifferentiation to give rise to more successful epithelial regeneration.
With the function of FAO in ISC biology attracting more attention, its positive role in ISC fate regulation has been gradually uncovered, but our work challenges the previously suggested model. Mihaylova et al. [
8] showed that long-term deletion of PPAR/Cpt1a, and thus chronic inhibition of FAO, abrogated ISC stemness, indicating that FAO is required for ISC survival and function. And with regard to HFD-induced intestinal tumorigenicity and fasting-driven ISC function, PPAR/Cpt1a-mediated FAO was found to be associated with improved ISC stemness [
9]. However, both HFD feeding and fasting regulate ISCs through other pathways besides FAO. For example, HFD interferes with the interaction between ISCs and immune cytokines via gut flora [
47], and fasting affects Paneth cells as well as quiescent ISCs [
8], [
48]. Moreover, both diets promote ketogenesis, which acts downstream of FAO, and ketone bodies have been shown to stimulate ISC renewal and suppress secretory cell differentiation through Notch signaling, but the change in OXPHOS level was not measured [
49], [
50]. In the absence of ketogenesis, post-IR regeneration was hindered, but in this case, FAO activity was not changed and OXPHOS was evaluated. These clues suggest that FAO improves stemness with enhanced ketogenesis, rather than OXPHOS, as a prerequisite, although the effect of ketogenesis on ISCs is also controversial in different studies [
51], [
52]. Conversely,
Mex3a knockout was found to impair Lgr5
+ ISC renewal by inducing the PPAR pathway [
53], indicating that PPAR-mediated FAO activation could pose negative impacts on ISC regeneration. In addition, a reduction in FAO caused by
Hnf4a/g deletion was found to blunt stemness but trigger a burst in proliferation among crypt cells with reduced mitochondrial OCR [
10]. In addition, Lgr5
+ ISC-specific deficiency of Prdm16, a transcription factor that drives mitochondrial respiration, diminishes FAO and ISC differentiation [
11]. Taken together, FAO inhibition along with failure in OXPHOS in ISCs is beneficial for epithelial proliferation and detrimental for the differentiation. In the present study, we did not find miR-29a/b regulate the Notch pathway but, rather, OXPHOS was affected via HNF4γ-mediated FAO. We also found that FAO activation fueled differentiation of LGR5
+ cells, which is consistent with Prdm16-mediated promotion of FAO in upper intestinal progenitors compared to ISCs [
11]. Thus, we infer that FAO-mediated OXPHOS inhibition is required during regeneration to expand the population of ISCs at the expense of differentiation.
4.3. MiR-29a/b specially regulate the FAO/OXPHOS axis in ISCs and function when regeneration is required
Importantly, metabolic programs vary between ISCs and differentiated cells [
5], and thus regulation of different cells by the same factor may be inverse [
54], [
55]. As to our findings here,
Mir29ab1 deletion may only stimulate FAO in Lgr5
+ ISCs, while the deletion in Lgr5
− differentiated cells had no significant effect or even an opposite one with respect to expression of FAO-related genes, indicating a diverse impact of miR-29a/b on different cell types. As a result, the RNA-sequencing analysis of crypts, which are a mixture of Lgr5
+ and Lgr5
− cells, did not indicate a clear difference in the FAO pathway. However, after IR-injury, regenerated crypts showed FAO inhibition via miR-29a/b upregulation as those crypts are occupied by Lgr5
+ cells, and in the absence of miR-29a/b, ISCs failed to suppress FAO, resulting in impaired regeneration. These findings emphasize the unique role of FAO-driven OXPHOS in ISCs and thus epithelial regeneration.
4.4. Loss of Hnf4γ is required for post-injury repair of intestinal epithelium
Hnf4γ has been shown to be a major driver of enterocyte differentiation and abundant in enterocyte lineages [
56], in line with the higher expression we detected in Lgr5
− cells compared to that in ISCs and the elevated differentiation we found in LGR5
+ cells overexpressing
HNF4G. ISCs lacking miR-29a/b had a similar
Hnf4g expression with that in differentiated cells, highlighting a shift of ISCs toward progenitors. Hnf4α and Hnf4γ belong to the same family and have overlapping functions. Knockout of merely
Hnf4a or
Hnf4g has no significant influence on ISC homeostasis as the other gene can offset the absence of one. Loss of both Hnf4α and Hnf4γ, which display higher expression in differentiated-cell enriched organoids, compromised FAO and suppressed epithelial differentiation with proliferation enhanced in crypts [
57], showing potential roles in intestinal epithelial recovery. With regard to Hnf4α, knockdown of
Hnf4a1 in colonic organoids has been found to increase Lgr5 expression as well as the number of Lgr5
high cells [
58], indicating that Hnf4α inhibition encourages ISC stemness. Consistently, we found that a single intervention of Hnf4γ is also enough to promote stemness, and explained the mechanism. Furthermore, we found that downregulation of
Hnf4g occurs naturally in the proliferating pool of regenerated crypts upon injury, and the downregulation was attributed to miR-29a/b. However, Hnf4α was previously shown to be upregulated in SI-derived organoids during post-IR regeneration [
59], implying inhibition of Hnf4γ might be responsible for the post-injury repair while HNF4α expression increases as feedback.
5. Conclusions and implications
Overall, our work provides a comprehensive insight into the metabolic plasticity of ISCs during regeneration. We demonstrate that temporal reductions in FAO and OXPHOS are required for the acute self-renewal of ISCs, challenging previous studies mainly focusing on homeostasis and the long-term effects. We identify miR-29a/b as the essential regulators to serve for such metabolic plasticity of ISCs in response to injury. Targeting of miR-29a/b or Hnf4γ has great potential in addressing diseases such as irritable bowel disease or Crohn’s disease and improving intestinal renewal during ageing. It also has potential in preventing tumor reoccurrence after medical treatment, though further studies need to be done for verification.
Acknowledgments
We thank The Chinese People’s Liberation Army General Hospital for the human samples. We thank Maggie Kane (Department of Genetics and Development, Columbia University Medical Center) for constructive suggestions for our work. This study was supported by the National Natural Science Foundation of China (32372247) and the National Key Research and Development Program of China (2023YFF1104501) to Huiyuan Guo.
Compliance with ethics guidelines
Yingying Lin, Yao Lu, Yuqi Wang, Cong Lv, Juan Chen, Yongting Luo, Heng Quan, Weiru Yu, Lining Chen, Ziyu Huang, Yanling Hao, Qingyu Wang, Qingfeng Luo, Jingyu Yan, Yixuan Li, Wei Zhang, Min Du, Jian He, Fazheng Ren, and Huiyuan Guo declare that they have no conflicts of interest or financial conflicts to disclose.
Appendix A. Supplementary material
Supplementary data to this article can be found online at
https://doi.org/10.1016/j.eng.2024.08.008.
PDF
(7619KB)
