Age‐associated mitochondrial complex I deficiency is linked to increased stem cell proliferation rates in the mouse colon

Abstract One of the hallmarks of aging is an accumulation of cells with defects in oxidative phosphorylation (OXPHOS) due to mutations of mitochondrial DNA (mtDNA). Rapidly dividing tissues maintained by stem cells, such as the colonic epithelium, are particularly susceptible to accumulation of OXPHOS defects over time; however, the effects on the stem cells are unknown. We have crossed a mouse model in which intestinal stem cells are labelled with EGFP (Lgr5‐EGFP‐IRES‐creERT2) with a model of accelerated mtDNA mutagenesis (PolgA mut/mut) to investigate the effect of OXPHOS dysfunction on colonic stem cell proliferation. We show that a reduction in complex I protein levels is associated with an increased rate of stem cell cycle re‐entry. These changes in stem cell homeostasis could have significant implications for age‐associated intestinal pathogenesis.


| INTRODUC TI ON
The accumulation of molecular damage to adult stem cells over time results in reduced capacity for self-renewal and tissue regeneration, and is thought to play a causative role in age-related frailty and disease. Mitochondrial dysfunction is a molecular hallmark of aging (Lopez-Otin et al., 2013) and the generation of a progeria mouse model with an error-prone version of the mitochondrial DNA (mtDNA) polymerase (PolgA mut/mut mice) provided evidence that mtDNA mutations (Kujoth et al., 2005;Trifunovic et al., 2004) and resulting defects in oxidative phosphorylation (OXPHOS) (Baines et al., 2014;Vermulst et al., 2007) could cause premature aging. Subsequent analysis of the underlying mechanisms showed that stem cell dysregulation played a major role (Ahlqvist et al., 2012;Chen et al., 2009;Fox et al., 2012). mtDNA mutations resulting in OXPHOS defects are observed in aging human stem cell populations (Fellous et al., 2009;McDonald et al., 2008;Taylor et al., 2003), including the colonic epithelium (~15% of crypts are OXPHOS-deficient by age 70 (Greaves et al., 2010;Taylor et al., 2003)). We have recently shown that age-associated defects in OXPHOS complexes I (CI) and IV (CIV) can increase tumour cell proliferation rates in a mouse model of intestinal tumorigenesis, accelerating colonic adenoma growth (Smith et al., 2020). However, the effects of CI and CIV deficiency on normal colonic stem cell proliferation rates are unknown.
We investigated this by crossing the PolgA mut/mut mice with Lgr5-EGFP-IRES-creERT2 mice, which have an EGFP under the Lgr5 (a wellaccepted stem cell marker) promoter (Barker et al., 2007). For ease, mice are referred to by their PolgA genotype only. Levels of NDUFB8 (CI) and MTCO1 (CIV) were quantified in PolgA +/mut and Polg mut/mut mice at 12 months of age and compared with age-matched PolgA +/+ controls using a validated immunofluorescent protocol (Rocha et al., 2015;Smith et al., 2020) (Figure 1a). Mean intensities were normalized to the mitochondrial mass marker TOMM20, and z-scores relative to the PolgA +/+ mice were generated (Figure 1b,c). Crypts with z-scores of <−4.5 were defined as deficient. In the PolgA +/mut mice, 31.88% of crypts were CI-deficient, 4.24% CIV-deficient and 11.06% CI + CIV-deficient. In the PolgA mut/mut mice, 53.1% of crypts were CI-deficient and 46.9% CI + CIV-deficient ( Figure 1d). No crypts were classified as deficient in TOMM20 (Figure 1e), indicating no differences in mitochondrial mass. This was confirmed using a second mass marker, VDAC1 (Figure 1f,g).
Total crypt cell and LGR5 High stem cell proliferation indices were analysed using multiple thymidine analogue labelling (Smith et al., 2020;Stoll et al., 2011). Mice were injected daily with CldU for 4 days, and a final injection of IdU was given 15 h before death. These times were based on the reported murine colonic crypt turnover time of 4-5 days and a 24-h stem cell cycle (Barker et al., 2007). CldU+ The proportion of Ki-67+ cells was significantly lower in PolgA mut/mut mice (p < 0.001, Figure 2f) suggesting a smaller proliferative zone with a higher rate of cell turnover. Specifically comparing LGR5 High stem cells per crypt from the PolgA +/+ and PolgA mut/mut mice, there was no significant difference in the proportion of CldU incorporation; however, a significantly higher proportion of PolgA mut/mut LGR5 High stem cells incorporated IdU (p = 0.0189) and CldU+IdU compared with PolgA +/+ mice (p < 0.001, Figure 2g-i). Next, we focussed on the PolgA +/mut mice, exploiting the mosaic pattern of OXPHOS defects to make intra-mouse comparisons. Three serial sections were analysed: CI and CIV levels were quantified in the first; thymidine analogue incorporation and Ki-67 labelling were quantified in the second  Figure 2q). There was a significantly higher frequency of CldU + IdU incorporation in LGR5 High stem cells with either isolated CI deficiency (p = 0.0373) or combined CI + CIV deficiency (p = 0.00069) compared with those with normal OXPHOS protein levels ( Figure 2r). These data suggest that CI deficiency is driving a ~20% increase in LGR5 High stem cell division rates in the colon.
We have directly examined the effect of age-related OXPHOS defects on colonic stem cell proliferation rates in vivo and show for the first time that CI deficiency results in an increased frequency of cell cycle re-entry in LGR5 High stem cells of the colon of PolgA +/mut and PolgA mut/mut mice. These data could either reflect a general increase in stem cell division rate or a change in the ratio of rapidly dividing stem cells to non-dividing or slowly dividing stem cells within the stem cell niche (Barriga et al., 2017). Despite certain limitations, such as the model-specific mechanism of mutation and mutational spectrum of the PolgA mice (Kujoth et al., 2005;Trifunovic et al., 2004), the functional consequences of the increased rate of mutagenesis (CI and CIV deficiency) make them a useful model in which to study the effects of defects in these proteins in stem cells. Previous work looking at cell cycle progression in the small intestine of PolgA mut/mut mice at 3 months of age showed a decrease in proliferation rates, suggesting there are either differential effects of OXPHOS deficiency between the colon and small intestine, or adaptive changes with age. Our previous studies have shown an increasing upregulation of de novo serine synthesis and one-carbon metabolic pathway enzymes in the PolgA mut/mut colon from 6 months onwards (Smith et al., 2020). These pathways are associated with biomass production in rapidly dividing cells and may influence stem cell division rates (Maddocks et al., 2017).
We do not know the molecular effects of CI deficiency on LGR5 High stem cell function beyond the changes in proliferation rates shown by this study; however, given the importance of control of the stem cell compartment to maintain epithelial crypt homeostasis, further work is required to determine the underlying molecular mechanisms.

| Animals
Female mice were group-housed in individually ventilated cages at 25°C with a 12-h light/dark cycle. All mice were on a C57BL6/J background. Animal experiments were conducted in compliance with the UK Home Office (PPL P3052AD70) and the Newcastle University Animal Welfare Ethical Review Board (AWERB 425).

| Thymidine analogue labelling
Mice (n = 4 per group, age: 49-51 weeks) were injected with 50 mg/ kg of CldU at 10am and 6 pm on days 1-3. On day 4, they were injected with 50 mg/kg of CldU at 10am followed by 50 mg/kg IdU at 6 pm (Smith et al., 2020). Mice were humanely killed by cervical dislocation 15 h later. The intestines were removed and fixed in 10% neutral buffered formalin for 24 h before standard dehydration and paraffin embedding.

| Immunofluorescence
OXPHOS immunofluorescence was performed on 3-µm serial sections of colon using the same antibodies as previously described (Rocha et al., 2015), and immunofluorescence was performed to detect the thymidine analogues as described (Smith et al., 2020).
Antibody specificity was determined using mice singly injected with either CldU or IdU, and no cross-reactivity was observed. Sections were imaged using a Zeiss Axio Imager M1 fluorescent microscope and labelled proteins quantified as previously described (Rocha et al., 2015). For the thymidine analogue labelling analysis, crypts were selected where a full longitudinal section was visible from crypt base to apex. The total number of cells per crypt, and the number of LGR5 High , Ki-67+, CldU+ and IdU+ cells per crypt were counted. GLMMs were performed using R programming, code is available on request.

ACK N OWLED G M ENTS
We thank Professor Tomas Prolla for donating the PolgA +/mut mice and Professor Karen Vousden for donating the Lgr5-EGFP-IRES-creERT2 mice, staff in the Newcastle University Comparative Biology Centre for animal husbandry, and the Newcastle University BioImaging Unit for assistance with imaging.

CO N FLI C T O F I NTE R E S T
The authors declare no competing financial interests.

AUTH O R CO NTR I B UTI O N S
CS and CB bred and maintained the mice. CS, JCW, DH ALMS and EAS performed experimental work and collected data. JCW and APB performed statistical analysis. DMT and LCG designed the study, supervised the project and wrote the manuscript.

DATA AVA I L A B I L I T Y S TAT E M E N T
Source data for this study are available from the corresponding author upon request. Negative binomial or Poisson GLMM, *p < 0.05, **p < 0.001. and ***p < 0.0001