Mitochondrial function controls intestinal epithelial stemness and proliferation

Abstract

Control of intestinal epithelial stemness is crucial for tissue homeostasis. Disturbances in epithelial function are implicated in inflammatory and neoplastic diseases of the gastrointestinal tract. Here we report that mitochondrial function plays a critical role in maintaining intestinal stemness and homeostasis. Using intestinal epithelial cell (IEC)-specific mouse models, we show that loss of HSP60, a mitochondrial chaperone, activates the mitochondrial unfolded protein response (MT-UPR) and results in mitochondrial dysfunction. HSP60-deficient crypts display loss of stemness and cell proliferation, accompanied by epithelial release of WNT10A and RSPO1. Sporadic failure of Cre-mediated Hsp60 deletion gives rise to hyperproliferative crypt foci originating from OLFM4⁺ stem cells. These effects are independent of the MT-UPR-associated transcription factor CHOP. In conclusion, compensatory hyperproliferation of HSP60⁺ escaper stem cells suggests paracrine release of WNT-related factors from HSP60-deficient, functionally impaired IEC to be pivotal in the control of the proliferative capacity of the stem cell niche.

Figure 1. Formation of hyperproliferative crypt nodules in the small intestine of Hsp60^Δ/ΔIEC mice.

(a) Schematic illustration of the targeted genomic modifications to generate mice carrying a conditional knockout allele for Hsp60 (Flox allele). Expression of Cre recombinase and induction of Cre activity by tamoxifen generate the Hsp60 knockout specifically in intestinal epithelial cells (IEC) or intestinal stem cells (ISC). (b) Schedule for oral tamoxifen administration to induce HSP60 deficiency in IEC of adult mice. Agarose gels showing the presence of the knockout allele specifically in IEC isolates. (c) Representative H&E and corresponding HSP60 IHC stainings of Hsp60^flox/flox and Hsp60^Δ/ΔIEC mice along the intestinal tract. Images of HSP60 IHC in higher magnification show HSP60-deficient villus versus HSP60-positive crypt regions of the jejunum. HSP60-positive crypt nodules in Hsp60^Δ/ΔIEC mice were counted along the intestinal tract using HSP60 IHC stainings. The graph represents quantifications of Hsp60^Δ/ΔIEC mice (N=6) with >100 crypts counted per animal. Lines indicate mean numbers. One-way analysis of variance (ANOVA) followed by Dunn's test was used to test for significance. (d) Genomic DNA was isolated from villi and hyperproliferative crypt nodules of Hsp60^Δ/ΔIEC mice using laser-dissection microscopy (N=3). Presence and absence of the Hsp60 knockout allele and Cre transgene was determined via PCR. DNA control PCR were run to check for equal loading (e) Correlation (Pearson) of MT-UPR marker gene expression with Hsp60 mRNA levels in IEC isolated from jejunal fractions of villus tip and crypt bottom of Hsp60^Δ/ΔIEC mice (N=6). P values indicate one-sided significance. Asterisks indicate significant differences *P<0.05, **P<0.01, ***P<0.001; NS, not significant.

Figure 2. Formation of hyperproliferative crypt nodules and MT-UPR activation is independent of CHOP-mediated signalling.

(a) Schedule for oral tamoxifen administration to induce HSP60 deficiency in IEC of adult Chop^−/− Hsp60^Δ/ΔIEC mice. Agarose gels showing the presence of the knockout allele specifically in IEC isolates. (b) Representative H&E and corresponding HSP60 IHC stainings of Chop^−/− Hsp60^Δ/ΔIEC mice along the intestinal tract. Images of HSP60 IHC in higher magnification show HSP60-deficient villus versus HSP60-positive crypt regions of the jejunum. HSP60-positive crypt nodules in Chop^−/− Hsp60^Δ/ΔIEC mice were counted along the intestinal tract using HSP60 IHC stainings. The graph represents quantifications of Hsp60^Δ/ΔIEC mice (N=6) with >100 crypts counted per animal. Lines indicate mean numbers. One-way analysis of variance (ANOVA) followed by Dunn's test was used to test for significance. (c) Quantification of HSP60-positive and HSP60-negative crypt numbers and total area of HSP60-positive crypt nodules of Hsp60^Δ/ΔIEC and Chop^−/− Hsp60^Δ/ΔIEC mice revealed no differences between genotypes (N=6); unpaired t-tests. Bars represent means+s.e.m. Lines in the dot plot indicate mean numbers. (d) qRT–PCR analysis of MT-UPR-associated genes in villus (upper panel) and crypt (low panel) IEC isolated from Hsp60^Δ/ΔIEC (N=6) and Chop^−/− Hsp60^Δ/ΔIEC mice (N=6) and corresponding controls (N=5 and N=6). Statistical analysis was performed using unpaired t-tests. Asterisks indicate significant differences *P<0.05, **P<0.01, ***P<0.001; NS, not significant.

Figure 3. HSP60 levels determine the proliferative capacity of the crypt.

(a) IF co-staining of HSP60 and Ki67 on jejunal sections including detailed images in higher magnification of Hsp60^flox/flox controls and Hsp60^Δ/ΔIEC mice (DAPI stains nuclei in cyan; dotted lines indicate localization of cryptal epithelium). Quantification of Ki67-positive cells in HSP60-positive and HSP60-negative crypts (N=5; 2 regions per mouse). (b) Parallel evaluation as in (a) for Chop^−/− Hsp60^Δ/ΔIEC mice and Chop^−/− Hsp60^flox/flox controls. Lines in the dot plots indicate mean numbers. (a,b,c), significantly different from each other. One-way analysis of variance (ANOVA) and appropriate post-hoc test were used for statistical analysis.

Figure 4. HSP60 deficiency impairs mitochondrial function.

(a) Correlation (Pearson) of expression levels of mtCoxI and Otc, involved in mitochondrial function, with Hsp60 mRNA levels in isolated IEC from villus (left) and crypt (right) compartment of Hsp60^Δ/ΔIEC mice (N=6 per genotype). P values indicate one-sided significance. (b) Plasma citrulline (left), glutamine (middle) and arginine (right) level in Hsp60^Δ/ΔIEC mice at d₂ (N=6) and Hsp60^flox/flox controls (N=5). Statistical analysis was performed using unpaired t-tests. (c) Agarose gel showing the presence of the Hsp60-knockout allele in genomic DNA isolated from organoids at 1 day after tamoxifen (4-OHT) addition. (d) Schematic illustration of the experimental set-up using small intestinal organoids ex vivo. Organoids were isolated from Hsp60^flox/flox, VillinCreER^T2-Tg and according VillinCreER^T2 negative control mice and distributed to three protocols. One-way analysis of variance and appropriate post hoc test were used for statistical analysis. (e) mRNA (left) and protein (right) expression levels of Hsp60 in organoids after 4-OHT treatment. Protein levels of mitochondria-located citrate synthase and β-Actin serving as loading control are also shown. (f) qRT–PCR analysis of target genes involved in MT-UPR. (g) qRT–PCR analysis of target genes involved in mitochondrial function. (h) ATP content of organoids relative to life cell protease activity measured by a fluorescence assay. (i) Respiratory capacity of mitochondria was measured by high-resolution respirometry. To exclude toxicity-related artefacts of 4-OHT, effects on CreER^T2-Tg organoids were directly compared to CreER^T2-negative organoids both receiving 4-OHT. Statistics were performed by unpaired t-test. Data from organoid experiments derive from at least 3 independent experiments. Bars represent mean+s.e.m. Asterisks indicate significant differences *P<0.05, **P<0.01, ***P<0.001; NS, not significant.

Figure 5. HSP60 loss antagonizes stemness while HSP60-positive escaper stem cells hyperproliferate.

(a) Images showing representative in situ hybridizations for Olfm4 mRNA on jejunal sections at d₂ including detailed images in higher magnification and quantification of Olfm4-positive crypts (N=6); dotted lines indicate localization of cryptal epithelium; red arrows indicate zones of Olfm4 expression). The dot plot indicates number of Olfm4⁺ crypts in Hsp60^flox/flox and Hsp60^Δ/ΔIEC mice at d₂. (b) qRT–PCR analysis of stem cell markers Olfm4 and Lgr5 was performed on isolated crypt bottom IEC of Hsp60^flox/flox (N=5) versus Hsp60^Δ/ΔIEC mice (N=6). Lines in the dot plots indicate mean numbers. All statistical analyses were performed via unpaired t-tests comparing genotypes. (c) Schedule for oral tamoxifen administration to induce HSP60 deficiency in IEC. IF images show HSP60 and OLFM4 expression at two different time points (d₀ and d₂) in jejunal sections of Hsp60^Δ/ΔIEC and Hsp60^flox/flox mice (DAPI stains nuclei in cyan). Representative pictures of N=5 per genotype. (d) Parallel analysis as in c for Chop^−/− Hsp60^Δ/ΔIEC mice and corresponding Chop^−/− Hsp60^flox/flox controls. Asterisks indicate significant differences *P<0.05, **P<0.01, ***P<0.001; NS, not significant.

Figure 6. HSP60 deficiency in Lgr5-positive cells induces intestinal stem cell loss.

(a) Schedule for oral tamoxifen administration to induce HSP60 deficiency in intestinal stem cells (ISC, left) in Hsp60^flox/flox, Lgr5CreER^T2-Egfp^Tg mice. Agarose gel validating the presence of the knockout allele in IEC isolates (middle). (b) Quantification of EGFP-positive, Lgr5 expressing ISC containing crypts at different time points after ISC-specific Hsp60 deletion (N=4 per genotype and time point) (right). (c) Representative IHC stainings for EGFP at d₄ are shown and the corresponding percentages of EGFP-positive crypts are indicated (lower panel).

Figure 7. Stem cell hyperproliferation is associated with epithelial induction of WNT-related signals.

(a) Representative pictures of in situ hybridization for WNT-target gene Axin2 mRNA in jejunal sections at d₂ including detailed images in higher magnification (red arrows indicate zones of Axin2 expression) (b) Correlation analysis (Pearson) of Axin2 and Hsp60 mRNA expression levels in IEC isolated from the crypt of Hsp60^Δ/ΔIEC (upper graph) and Chop^−/− Hsp60^Δ/ΔIEC (lower graph) mice (N=6). P values indicate one-sided significance. (c) qRT–PCR analysis of WNT ligands including the WNT enhancer Rspo1 in IEC isolated from the crypt bottom of Hsp60^Δ/ΔIEC mice (N=6) versus Hsp60^flox/flox control mice (N=5). Statistical analyses were performed using unpaired t-tests. (d) Correlation analysis (Pearson) of significantly regulated WNT factors with Hsp60 levels in Hsp60^Δ/ΔIEC mice (N=6). P values indicate one-sided significance. Asterisks indicate significant differences *P<0.05, **P<0.01, ***P<0.001; NS, not significant; ND, not detectable.

Figure 8. HSP60 loss is associated with epithelial induction of WNT-related signals.

(a) IF images show epithelial RSPO1 (red) and HSP60 (yellow) expression at d₂ in jejunal sections from Hsp60^Δ/ΔIEC and Hsp60^flox/flox mice. E-Cadherin (grey) and αSMA (green) were used as epithelial and mesenchymal cell marker, respectively (DAPI stains nuclei in cyan, asterisks indicate Paneth cells). (b) IF images show epithelial WNT10A (red) and HSP60 (yellow) expression at d₂ in jejunal sections from Hsp60^Δ/ΔIEC and Hsp60^flox/flox mice. Lysozyme (green) and E-Cadherin (grey) were used as Paneth cell and epithelial cell marker, respectively (DAPI stains nuclei in cyan). WNT10A and Lysozyme co-stained (yellow, lower panel). (c) Quantification of WNT10A-positive cells in the jejunum of Hsp60^Δ/ΔIEC and Hsp60^flox/flox mice (N=6). Lines in the dot plot indicate mean numbers. Statistical analysis was performed using unpaired t-test. (d) Experimental scheme to induce Hsp60 loss in small intestinal organoids. qRT–PCR analysis of Lgr5 and Wnt10a mRNA expression in organoids following gene knockout. Data from organoid experiments derive from at least 3 independent experiments. Bars represent mean +s.e.m. One-way analysis of variance and appropriate post-hoc tests were used for statistical analysis. Asterisks in c, d indicate significant differences *P<0.05, **P<0.01, ***P<0.001; NS, not significant.

Figure 9. WNT10A and RSPO1 but not ROS scavenging rescues intestinal organoid growth after Hsp60 knockout.

Organoids were isolated from Hsp60^flox/flox, VillinCreER^T2-Tg mice and distributed to four protocols. (a) Experimental scheme to show the effects of RSPO1 and WNT10A supplementation (100 ng ml⁻¹) on Hsp60-deficient small intestinal organoids. Lower panel: representative pictures of the indicated treatments and time points. (b) Measurement of organoid area (left; N>60 per treatment) and life cell protease activity measured by fluorescence (right) following indicated treatments. (c) Quantification of de novo crypt formation (left); (a–d) significantly different from each other, Kruskal–Wallis test on ranks followed by Dunn's test. Right: qRT–PCR analysis of Lgr5 mRNA expression in organoids in response to WNT10A treatment. (d) Experimental scheme to show the effects of the ROS scavenger Euk-134 (100 μM) on Hsp60-deficient small intestinal organoids. Lower panel: representative pictures of the indicated treatments and time points. (e) Organoid area (left; N>60 per treatment) and life cell protease activity measured by fluorescence (right) following indicated treatments. (f) Quantification of de novo crypt formation (left); (a,b), significantly different from each other, Kruskal–Wallis test on ranks followed by Dunn's test. Right: qRT–PCR analysis of Lgr5 mRNA expression in organoids in response to Euk-134 treatment. Bars represent mean+s.e.m. Asterisks indicate significant differences *P<0.05; **P<0.01; ***P<0.001; NS, not significant. Unless otherwise indicated, one-way analysis of variance and appropriate post hoc tests were used for all statistical analyses. Data from organoid experiments derive from at least three independent experiments. Scale bars, 200μm.

Figure 10. Schematic model of hyperproliferative crypt formation in Hsp60^Δ/ΔIEC mice.

Intestinal crypts loose stem cells and the proliferative capacity in response to HSP60 loss. WNT-related factors are produced by HSP60-deficient IEC including Paneth cells to compensate diminished proliferation. This epithelial microenvironment causes HSP60-positive escaper stem cells to hyperproliferate and form HSP60-positive crypt nodules, finally leading to tissue reconstitution. In contrast, HSP60 deficiency in Lgr5-positive stem cells causes transient loss of stem cells without alterations in tissue morphology.