Loss of Kit^low progenitors, reduced stem cell factor and high oxidative stress underlie gastric dysfunction in progeric mice

Abstract

Gastrointestinal functions decline with ageing leading to impaired quality of life, and increased morbidity and mortality. Neurodegeneration is believed to underlie ageing-associated dysmotilities but the mechanisms have not been fully elucidated. We used progeric mice deficient in the anti-ageing peptide Klotho to investigate the contribution of key cell types of the gastric musculature to ageing-associated changes in stomach function and the underlying mechanisms. Klotho expression, enteric neurons, interstitial cells of Cajal (ICC), smooth muscle cells and electrical activity were assessed by immunofluorescence, confocal microscopy, 3-dimensional reconstruction, flow cytometry, quantitative RT-PCR, Western immunoblotting and intracellular recordings. Gastric emptying of solids was analysed by the [¹³C]octanoic acid breath test. Circulating and tissue trophic factors were measured by enzyme immunoassays and quantitative RT-PCR. The role of oxidative stress was investigated in organotypic cultures. Klotho expression was detected in gastric glands, myenteric neurons and smooth muscle cells. Progeric Klotho-deficient mice had profound loss of ICC and ICC stem cells without a significant decrease in neuron counts, expression of neuronal nitric oxide synthase or smooth muscle myosin. Slow wave amplitude and nitrergic inhibitory junction potentials were reduced while solid emptying was unchanged. Klotho-deficient mice were marantic and had low insulin, insulin-like growth factor-I and membrane-bound stem cell factor. Klotho deficiency accentuated oxidative stress and ICC loss. We conclude that Klotho-deficient, progeric mice display a gastric phenotype resembling human ageing and involving profound ICC loss. Klotho protects ICC by preserving their precursors, limiting oxidative stress, and maintaining nutritional status and normal levels of trophic factors important for ICC differentiation.

Introduction

The number of people in the United States aged 85 or older is expected to increase ∼4-fold by 2050 (Camilleri et al. 2008). Organ functions inevitably decline in ageing. Changes in the gastrointestinal tract include increased prevalence of gastroesophageal reflux, silent aspiration, achlorhydria, postprandial hypotension, irritable bowel syndrome, constipation and faecal incontinence (Bhutto & Morley, 2008; Camilleri et al. 2008), as well as more subtle dysfunctions such as early satiation and consequent decline in caloric intake, the so-called ‘physiological anorexia of ageing’ (Parker & Chapman, 2004; Hays & Roberts, 2006; Bhutto & Morley, 2008). Although these disorders and dysfunctions are usually not themselves fatal, they negatively affect general health, quality of life, and the ability to maintain independence (Camilleri et al. 2008) and represent a significant health care burden (Bhutto & Morley, 2008). The subtle physiological changes such as increased satiation also lead to reduced ability to maintain nutritional status in response to metabolic challenge (Hays & Roberts, 2006) and, therefore, predispose the elderly to severe complications from other diseases (Bhutto & Morley, 2008).

The mechanisms of ageing-associated gastrointestinal dysfunctions are incompletely understood. Enteric neurons decline with age, particularly in the distal gastrointestinal tract (Phillips & Powley, 2007; Camilleri et al. 2008; Bernard et al. 2009), but neurodegeneration is less evident in the stomach (Phillips & Powley, 2007). Smooth muscle function also diminishes with age but there is insufficient information about its significance (Bitar & Patil, 2004). Another key cell type of the gastrointestinal neuromuscular apparatus is interstitial cells of Cajal (ICC), which generate electrical pacemaker activity (slow waves) underlying phasic contractions, partially mediate neuromuscular signalling and mechanotransduction, and set smooth muscle membrane potential and tone (Sanders et al. 2006; Ward & Sanders, 2006; Kraichely & Farrugia, 2007; Huizinga et al. 2009). Through these functions and in concert with the autonomic nervous system and the smooth musculature, ICC regulate key aspects of motility including accommodation, mechanical processing of food, peristalsis and waste excretion. ICC loss has been implicated in several neuromuscular disorders (Huizinga et al. 2009; Ordog et al. 2009); however, their role in ageing has not been studied extensively. We recently described a profound, age-related decline in ICC throughout the gastrointestinal tract of human subjects including the stomach (Gomez-Pinilla et al. 2010). Therefore, ICC loss may underlie the impaired electrical slow wave activity detected in elderly subjects (MacIntosh et al. 2001; Shimamoto et al. 2002) and reduce the capacity of the gastrointestinal tract to properly adapt to various homeostatic challenges.

Our goal was to investigate the role of ICC, enteric neurons and smooth muscle cells in ageing-associated gastric neuromuscular dysfunctions as well as the underlying mechanisms in a mammalian model of ageing. Since a decline in tissue-specific stem/progenitor cell pools is believed to be central to the loss of organ function in the elderly (Rossi et al. 2008), we also enumerated Kit^lowCd44⁺Cd34⁺ ICC stem cells (Lorincz et al. 2008; Bardsley et al. 2010). As a model of ageing we studied the klotho mouse, which is homozygous for a hypomorphic mutation that results in severely reduced expression of Klotho (α-Klotho; Kl) (Kuro-o et al. 1997), a pleiotropic polypeptide with anti-ageing properties (Kurosu et al. 2005). Klotho is a single-pass transmembrane protein expressed in a limited number of tissues that serves as co-receptor for fibroblast growth factor 23 (FGF23) and a source of free Klotho peptide, which can influence the function of a broad range of organs (Kuro-o, 2009). Klotho expression declines with age in mice, rats and monkeys (Duce et al. 2008) and KL gene variations affect the human life span (Arking et al. 2002). Klotho-deficient mice display a wide array of premature ageing phenotypes and have an average life span of ∼60 days (Kuro-o et al. 1997; Kuro-o, 2009), whereas mice overexpressing Klotho live 20–30% longer than their wild-type littermates (Kurosu et al. 2005). By oligonucleotide microarray analysis we previously detected Kl expression in ICC of the deep muscular plexus of the murine small intestine (Chen et al. 2007) but no expression has previously been observed in the stomach (Kuro-o et al. 1997). Here we demonstrate that Klotho is expressed in the mouse stomach, and decreased expression of Klotho results in a dramatic depletion of ICC and their stem/progenitor cells without an overt loss of neurons or smooth muscle cells. We also found increased oxidative stress which, together with a decrease in circulating and tissue factors that regulate ICC differentiation and survival, contributes to profound depletion of mature ICC and impaired gastric function. This study reveals previously unidentified mechanisms whereby Klotho preserves ICC and their precursors and highlights ICC loss as an important mechanism of ageing in the gastrointestinal tract.

Methods

Ethical approval

Experiments were performed in accordance with the National Institutes of Health Guide for the Care and Use of Laboratory Animals, the American Physiological Society's Guiding Principles in the Care and Use of Animals and The Journal of Physiology's ethical policies and regulations as outlined in Drummond (2009). All protocols were approved by the Institutional Animal Care and Use Committee of the Mayo Clinic. A total of 73 mice were used in the completion of this study.

Animals and tissue preparation

Homozygous klotho mice (Kuro-o et al. 1997) and age-matched wild-type (WT) and heterozygous (Het) littermates were obtained from heterozygous breeders and their genotype verified by PCR (Kuro-o et al. 1997). Experiments were performed between 50 and 70 days of age. Data from WT and Het mice were pooled when the results indicated no significant differences between these groups. Experimental groups were balanced for sex in all physiological and endocrine studies. Mice were killed by decapitation performed under deep isoflurane (Baxter Healthcare) inhalation anaesthesia. Intact gastric tunica muscularis tissues were prepared as described (Horvath et al. 2005, 2006; Lorincz et al. 2008) and used immediately or after culturing.

In vivo and ex vivo physiological studies

Body weights were obtained and physiological studies started between 09.00 and 10.00 h, and before feeding. Whole-gut transit was estimated by gavaging 0.3 ml carmine red (6% w/v in 0.5% w/v methylcellulose in water; Sigma-Aldrich) and measuring time until the excretion of the first coloured faecal pellet. Faecal output was quantified by collecting pellets for 3 h and weighing them after drying to weight constancy. Gastric emptying of solids (200 mg baked egg yolk) was measured by [¹³C]octanoic acid breath test using a non-invasive, non-restraining technique (Choi et al. 2008). Electrical slow waves and postjunctional responses to electrical field stimulation (EFS) of enteric nerves in the circular muscle of the distal stomach (corpus + antrum) and the fundus, respectively, were recorded by an intracellular technique (Ordog et al. 1999, 2000). N^ω-Nitro-l-arginine (l-NA) and atropine were from Sigma-Aldrich.

Blood glucose and circulating hormones

Blood samples were taken from the submandibular vascular bundle or from trunk blood collected after decapitation performed under deep isoflurane inhalation anaesthesia between 09.00 and 10.00 h. Blood glucose was measured from fresh whole blood using an Accu-Chek Complete monitor (Roche) (Horvath et al. 2006). Serum total insulin-like growth factor-I (IGF-I), insulin and corticosterone was measured by immunoassays in the Mayo Clinic Immunochemical Laboratory using a Quantikine Mouse IGF-I Immunoassay (R&D Systems), Linco Rat/Mouse Insulin ELISA Kit (Millipore), and AssayMax Corticosterone ELISA Kit (AssayPro), respectively.

Immunohistochemistry, confocal microscopy and image analysis

Whole-mounts of freshly dissected or cultured, intact gastric corpus + antrum tunica muscularis tissues were processed using established techniques (Horvath et al. 2005, 2006). Briefly, the tissues were fixed with cold acetone (10 min) and blocked with 1% bovine serum albumin (Sigma-Aldrich). ICC were detected with rat monoclonal anti-murine Kit antibodies (ACK2; 48 h at 4°C; see Supplemental Table S1 for detailed antibody information, available online only in Supplemental material) and Alexa Fluor 488- or 594-goat anti-rat IgG. Enteric neurons and nerve fibres were detected with human serum containing anti-HuC/D (ELAV (embryonic lethal, abnormal vision, Drosophila)-like 3/4) antibodies (ANNA1; a generous gift from Dr Vanda Lennon, Mayo Clinic; 24 h at room temperature) and rabbit polyclonal anti-protein gene product 9.5 (anti-PGP 9.5) antibodies, respectively, applied in PBS containing 50 mg ml⁻¹ beef liver powder (a generous gift from Dr Vanda Lennon, Mayo Clinic). Bound antibodies were detected with Alexa Fluor 633-goat anti-human IgG and Alexa Fluor 594-chicken anti-rabbit IgG.

Klotho protein was detected in whole-mounts or in 5 μm cryosections (Gomez-Pinilla et al. 2009) prepared from undissected stomachs. The cryosections were fixed with 4% paraformaldehyde (30 or 60 min), subjected to antigen retrieval using Dako Target Retrieval Solution (Gomez-Pinilla et al. 2009) and labelled with rat monoclonal or goat polyclonal anti-α-Klotho antibodies (4°C overnight; see Supplemental Table S1 for details). Smooth muscle cells, neurons and ICC were labelled with rabbit anti-smooth muscle myosin, rabbit anti-PGP 9.5 and either rabbit polyclonal anti-Kit or rat monoclonal anti-Kit (2B8) antibodies. Nuclei were counterstained with 4′,6-diamidino-2-phenylindole (DAPI). Whole-mounts were fixed with acetone or paraformaldehyde and labelled with goat polyclonal anti-α-Klotho and either rat monoclonal anti-Kit (ACK) or rabbit polyclonal anti-PGP 9.5 antibodies as described above.

Wide-field fluorescence images of cryosections were captured with a Nikon Eclipse TS-100F microscope and a Jenoptik MFCool CCD camera. Whole-mounts were imaged with an Olympus FV300 or FV1000 confocal system equipped with an Olympus ×40, 1.00 NA oil immersion objective. The confocal images (512 pixels × 512 pixels) were collected using the optimal pinhole size and Z-axis step (0.7 or 0.84 μm). In the corpus and antrum, images were obtained within three fields-of-view (530 μm) from the greater curvature. The volume of Kit⁺ ICC networks in the circular and longitudinal muscle layer and in the myenteric region of the corpus and antrum was calculated by 3-dimensional (3-D) reconstruction using ANALYZE software (Mayo Foundation) (Bernard et al. 2009) and expressed as percentage of layer volume. Mast cells were excluded from this quantification. HuC/D⁺ enteric neurons were visualized with an Olympus BX61 microscope equipped with a Zeiss ×25, 0.80 NA water immersion objective and an Olympus Magnafire camera and counted frame-by-frame with the aid of a Prior automated stage controller in an area defined by the distance between the fundus–corpus border and the pylorus and three fields-of-view (1428 μm) centred on the greater curvature. Neuron numbers were expressed as average counts/field-of-view and total count (sum of all fields-of-view).

Specificity of immunolabelling was verified by omitting the primary antibodies and by examining the samples with filter sets not designed for the fluorochrome used. Post-acquisition modification of images was limited to maximum-transparency projection, noise reduction, assignment of pseudocolour, and adjustment of brightness and contrast, which were always applied to the entire image. Personnel performing quantitative analyses were blinded to the identity of the samples.

Flow cytometry and fluorescence-activated cell sorting (FACS)

ICC and ICC stem/progenitor cells were quantified by flow cytometry in the haematopoietic marker-negative fraction of dissociated gastric corpus + antrum tunica muscularis as Kit⁺Cd44⁺Cd34⁻ and Kit^lowCd44⁺Cd34⁺ cells, respectively, and sorted by FACS. Previously published techniques were used with modifications (Horvath et al. 2006; Lorincz et al. 2008; Bardsley et al. 2010) (see online Supplemental material: Supplemental Fig. S1 for gating scheme and Supplemental Table S2 for detailed antibody information). Briefly, intact gastric corpus + antrum muscles were incubated with allophycocyanin (APC)-ACK2 at 4°C for 3 h then dissociated with collagenase digestion and trituration and filtered. Single-cell suspensions were incubated in 100 μl with anti-mouse Cd16/32 antibody (Fc block). Haematopoietic cells were identified with phycoerythrin (PE)-cyanine (Cy) 7-coupled anti-mouse Cd11b, anti-Cd45 and anti-F4/80 antibodies. Cells were also labelled with APC-anti-Kit (2B8), PE-anti-Cd34 and APC-Cy7-anti-mouse/human Cd44. Two additional aliquots of the same samples were labelled in an identical manner but without the anti-Cd44 or the anti-Cd44 and the anti-Cd34 antibodies. Our experimental design was extensively verified by using single-stained and fluorescence-minus-one controls as well as by sorting and RT-PCR (Horvath et al. 2006). Samples were analysed using a Becton Dickinson LSR II flow cytometer (see Supplemental Table S3 for configuration). Data files were analysed by FlowJo software (Treestar).

ICC, identified as Kit⁺Cd44⁺Cd34⁻, haematopoietic marker⁻ cells as described above (Horvath et al. 2006; Lorincz et al. 2008), were sorted using a Becton Dickinson FACS Aria Cell Sorter configured to match the LSR II cytometer.

Quantitative and qualitative RT-PCR

Total RNA was isolated using the Norgen Biotek Total RNA Purification Kit including on-column RNase-free DNase treatment and reverse transcribed into first-strand cDNA using the SuperScript VILO cDNA Synthesis Kit (Invitrogen). The cDNA was amplified on a Bio-Rad MyiQ real-time PCR detector using the SYBR GreenER qPCR SuperMix (Invitrogen) and specific, intron-spanning primers published previously (Kit, total Kitl (the gene encoding stem cell factor, SCF), soluble Kitl, Actb (β-actin) (Horvath et al. 2006) or designed for this study (Igf1: IGF-I transcript variants 1 and 2; RefSeq accession numbers: NM_010512 and NM_184052; amplicon: 171 bp; reference position: 58; Nos1: nitric oxide synthase 1, neuronal; RefSeq accession number: NM_008712.2; amplicon: 206 bp; reference position: 76). Primers for mouse Klotho (Kl) were purchased from SABiosciences (RefSeq accession number: NM_013823.2; amplicon: 189 bp; reference position: 4219). PCR was performed using the following amplification profile: 50°C for 2 min (uracil DNA glycosylase (UDG) incubation); 95°C for 10 min (UDG inactivation and DNA polymerase activation); then 40 cycles of 95°C for 15 s and 60°C for 1 min. Transcriptional quantification was obtained by the ΔΔC_T method. In experiments not requiring quantification, cDNA was amplified with AmpliTaq Gold polymerase (Applied Biosystems) and the products were separated by 2% agarose electrophoresis and visualized by ethidium bromide staining. Optimal concentrations of input cDNA and the efficacy of the PCR have been determined by standard curves for each primer set. We tested for non-specific amplification and primer–dimer fragments by omitting the template from the PCR amplification; by running melt curve after each PCR, and by separating the amplified products by electrophoresis. Potential amplification of genomic DNA sequences was controlled for by designing primers that span an intron large enough to yield an amplicon that would not amplify under the PCR conditions used. We also tested for genomic DNA contamination by omitting the reverse transcriptase.

Western immunoblotting

Homogenates of gastric tissues were lysed in a solution consisting of 50 mmol l⁻¹ Tris-HCl, 1% NP-40, 0.25% sodium deoxycholate, 150 mmol l⁻¹ NaCl, 1 mmol l⁻¹ EDTA, 1 mmol l⁻¹ activated Na₃VO₄, 1 mmol l⁻¹ NaF and protease inhibitors including 1 μg ml⁻¹ of aprotinin, leupeptin, pepstatin and 1 mmol l⁻¹ phenylmethylsulphonyl fluoride (Choi et al. 2007). From the tissue lysates, 30 μg of protein was resolved by SDS-PAGE on a 12% gel and transferred to immunoblot polyvinylidene fluoride (PVDF) membranes (Bio-Rad). Rabbit polyclonal anti-Kit (1:200; Santa Cruz Biotechnology), anti-Nos1 (1:2000; Chemicon) and anti-glyceraldehyde-3-phosphate dehydrogenase (Gapdh; 1:2000; Santa Cruz Biotechnology) were used as primary antibodies. Bound primary antibodies were visualized with horseradish peroxidase-conjugated secondary antibodies and ECL Western Blotting Detection Reagents (GE Healthcare Bio-Science). Blots were imaged with a Bio-Rad ChemiDoc XRS imager.

Organotypic culture studies

Intact gastric corpus + antrum tunica muscularis tissues from klotho mice and age-matched WT littermates were harvested between 52 and 68 days of age and cultured with serum- and growth factor-free Medium 199 (Sigma-Aldrich) containing 100 or 500 mg dl⁻¹ glucose as described (Horvath et al. 2005, 2006; Lorincz et al. 2008). After 46–49 days, oxidative stress was determined in 16 h culture media by measuring thiobarbituric acid-reactive substances as malondialdehyde (MDA) equivalents using an Oxi-Tek kit (Zeptometrix) (Choi et al. 2008). The stomachs were cut along the greater curvature and one-half was used for the assessment of ICC by Kit immunohistochemistry (see above); in the other half, Kit mRNA and Kit protein was determined by quantitative RT-PCR and Western immunoblotting (see above and Choi et al. 2008).

Statistical analyses

Data are expressed as means ±s.e.m. or median [interquartile range (IQR)] and analysed by Student's t test, Mann–Whitney rank sum test, or one-way ANOVA and all-pairwise multiple comparisons. P < 0.05 was considered significant.

Results

First we examined whether Klotho was expressed in the stomach. By RT-PCR we detected Kl mRNA in the WT mouse gastric corpus + antrum tunica muscularis, whereas in hypomorphic klotho mice, no or only very little Kl mRNA was found (Fig. 1A). We were unable to amplify Kl in highly purified gastric ICC from WT mice (Fig. 1B). Immunohistochemistry in WT gastric whole-mounts or cryosections verified the lack of Klotho protein in ICC and revealed specific expression in myenteric neurons and smooth muscle cells of both the external muscle layers and the tunica muscularis mucosae (Fig. 1C–G). Additionally, we found strong Klotho expression at the base of gastric glands. Klotho was not detected in intraepithelial myofibroblasts. Expression in the stomach occurred at a lower level than in the kidney (not shown). Thus, direct and paracrine effects may complement the effects of circulating Klotho protein on cells of the stomach.

Figure 1. Klotho is expressed by epithelial cells, enteric neurons and smooth muscle cells, but not by ICC, in the murine stomach.

A, detection of Kl mRNA by RT-PCR in whole gastric corpus + antrum tunica muscularis. Results from quantitative RT-PCR experiments (box plots: median and IQR; filled circles are individual data points; WT: n = 6; klotho: n = 5) and a representative qualitative RT-PCR experiment performed in a WT stomach are shown. No amplification was detected in any sample generated without the RT (RT–). B, lack of detectable Kl expression in mature, Kit⁺Cd44⁺Cd34⁻ ICC (n = 119,493 cells) purified from the haematopoietic marker-negative fraction of 3 WT stomachs by FACS (see gating scheme in Supplemental Fig. S1, available online). Actb, β-actin; Kit, Kit oncogene, an ICC marker. C–G, immunohistochemical localization of Klotho protein. Seven WT tissues were processed either as whole-mounts or cryosections and labelled with rat monoclonal or goat polyclonal anti-α-Klotho antibodies. C and D, representative confocal images taken in the perimyenteric region (Z thickness: 7.56 μm) of gastric whole-mounts. Tissues were labelled with goat polyclonal anti-Klotho (green pseudocolour) and rat monoclonal anti-Kit (ACK2; red pseudocolour) antibodies. C, distal corpus in the vicinity of the greater curvature. D, orad antrum, anterior surface. Scale bars, 50 μm. E–G, representative cryosections labelled with rat monoclonal anti-Klotho antibodies (green pseudocolour). The primary antibody was omitted from G. Enteric neurons and smooth muscle cells were stained with anti-PGP 9.5 (red pseudocolour in E) and anti-smooth muscle myosin (red pseudocolour in F and G), respectively. Nuclei were counterstained with DAPI (blue pseudocolour). Scale bars, 25 μm. Klotho was detected in the mucosal glands (Muc), myenteric neurons (asterisks), and smooth muscle cells of the muscularis propria (SM) and the muscularis mucosae (MM) but not in ICC. Intraepithelial myofibroblasts (arrows in F) also did not express Klotho.

Next, we investigated whether Klotho deficiency affected key cell types of the gastric tunica muscularis. Upon dissection, klotho stomachs did not appear to be smaller but had prominent mucosal and submucosal calcification (Kuro-o et al. 1997; Kurosu et al. 2005). Surprisingly, we found no decrease in HuC/D⁺ myenteric neurons in the gastric corpus and antrum (Fig. 2A and B) and neither mRNA nor protein levels for neuronal nitric oxide (NO) synthase (Nos1) changed significantly (Fig. 2C and D). Gene expression for the smooth muscle-specific contractile protein myosin heavy chain 11 (Myh11) also remained unaffected (Fig. 2E). In contrast, confocal microscopy and quantitative analysis of Kit⁺ structures reconstructed in 3-D showed that ICC network volumes were diffusely and grossly reduced throughout the klotho corpus (Fig. 3A and B) and antrum (Fig. 3C and D), both in the myenteric region and in the circular and longitudinal muscle layers. Intramuscular ICC were also reduced in the fundus (Fig. 3E). Low ICC network densities reflected, to a significant degree, loss of ICC (i.e. not only a reduction in ICC arborization) as flow cytometry indicated an ∼35% decrease in the frequency of mature (Kit⁺Cd44⁺Cd34⁻; Lorincz et al. 2008) ICC (Fig. 3F and G). ICC loss occurred without a significant change in median cellular Kit fluorescence (1.19[0.89;1.50]-fold vs. WT; n = 6; P = 0.37). There was an even more pronounced (∼67%) decrease in the frequency of ICC stem cells (Kit^lowCd44⁺Cd34⁺ cells (Lorincz et al. 2008; Bardsley et al. 2010; Fig. 3F and G). These findings indicate that in the gastric tunica muscularis, a major depletion of ICC precursors and consequent loss of ICC are the main results of Klotho deficiency.

Figure 2. Myenteric neurons, Nos1 mRNA and protein and smooth muscle-specific gene expression are not reduced in the gastric corpus and antrum of klotho mice relative to their age-matched, wild-type littermates.

A, detection of perikarya of myenteric neurons by HuC/D immunostaining. Representative wide-field fluorescence image from a total of 6 klotho and 9 WT mice are shown. Ganglion boundaries were determined by PGP 9.5 immunostaining of the same tissue samples (not shown). Scale bar, 100 μm. B, neuron numbers quantified in an area defined by the distance between the fundus–corpus border and the pylorus and 3 fields-of-view (1428 μm) centred on the greater curvature. Neuron counts were expressed as total count (sum of all fields-of-view; left panel) and average counts/field-of-view (right panel). There was no evidence of net neuron loss in the stomach of klotho mice (n = 3/group). C, expression of Nos1 determined by quantitative real-time RT-PCR (n = 3 stomachs/group). Actb, β-actin used as reference. D, Nos1 protein expression detected by Western immunoblotting (n = 3 stomachs/group; panel on the right shows representative immunoblots). Gapdh, glyceraldehyde-3-phosphate dehydrogenase used as loading control. The three bands probably represent splice variants described in several tissues (Smith et al. 2009). All bands were included in the densitometric analysis. Neither Nos1 mRNA nor protein was significantly reduced in klotho mice. E, expression of smooth muscle myosin heavy chain 11 (Myh11) determined by quantitative real-time RT-PCR (n = 6 stomachs/group). There was no evidence of smooth muscle dystrophy in the stomach of Klotho-deficient mice.

Figure 3. ICC and ICC stem cells are depleted from the stomach of Klotho-deficient mice.

A, representative confocal stacks showing Kit⁺ ICC in the longitudinal muscle (LM), myenteric region (MY) and circular muscle layer (CM) of the corpus of WT and klotho mice. Scale bar, 50 μm. B, relative volume of Kit⁺ ICC networks in the corpus obtained by 3-D volume rendering in WT (n = 3) and klotho mice (n = 4). C and D, representative confocal stacks and ICC network volumes in the antrum. Labels are as above. ICC were profoundly depleted in all layers of both the corpus and the antrum. E, representative confocal stacks showing Kit⁺ ICC in the fundus of WT and klotho mice. Scale bar, 50 μm. Intramuscular ICC were also reduced in the fundus. F and G, quantification of ICC and ICC stem cells, identified as Kit⁺Cd44⁺Cd34⁻ and Kit^lowCd44⁺Cd34⁺ cells, respectively, by flow cytometry (see gating sequence in Supplemental Fig. S1). F, Kit and Cd34 immunofluorescence of ICC (green) and ICC stem cells (red) in a representative WT and klotho mouse back-projected to the Cd44⁺ subset of non-haematopoietic population (grey). G, cell frequencies (% of cells with light scatter properties characteristic of live cells (LS⁺)). Both ICC and ICC stem cells were profoundly reduced in the klotho mice.

Intracellular electrophysiology indicated significantly reduced slow wave amplitudes throughout the corpus and antrum and the peak voltage reached during the plateau was also significantly lower in the antrum (Fig. 4A, Table 1). Slow wave frequency was significantly increased in the corpus but there was no evidence of irregular arrhythmias (Table 1). Diminished slow wave amplitudes signify reduced electrical drive to smooth muscle contractions and peristalsis (Sanders et al. 2006). EFS of enteric nerves in the WT fundus evoked stimulus frequency-dependent fast and slow inhibitory junction potentials (IJPs) and a highly variable, late-developing, atropine-sensitive excitatory junction potential (EJP) (Supplemental Fig. S2, Fig. 4B). The slow IJPs and, to a much lesser extent, the fast IJPs could be significantly inhibited by 100 μm l-NA (Supplemental Fig. S3) indicating the involvement of nitrergic mechanisms (Burns et al. 1996; Beckett et al. 2002). Partial inhibition of the fast IJPs by l-NA probably reflected incomplete temporal separation of the nitrergic and non-nitrergic components (e.g. as in Supplemental Fig. S2B). In klotho tissues, EJPs did not change significantly (Table 2), although accentuated excitatory responses were occasionally observed (Fig. 4B). The lack of a significant change in EJPs should be interpreted with caution since these responses were not uniformly detected in either WT or klotho mice. In contrast, IJPs were significantly reduced (Fig. 4B, Table 2) and l-NA did not have a significant effect (Supplemental Fig. S3). Thus, reduced inhibitory neuromuscular neurotransmission in the klotho mice was mainly due to loss of nitrergic signalling, which occurred in the absence of significantly reduced Nos1 protein expression. Reduced inhibitory neuromuscular neurotransmission results in an impaired relaxation in response to filling (Dixit et al. 2006). Gastric emptying of solids remained unchanged (Fig. 4C), probably reflecting opposing influences of reduced peristalsis and smooth muscle relaxation (Ordog et al. 2009).

Figure 4. Reduced slow wave amplitudes and inhibitory neuromuscular neurotransmission and normal gastric emptying of solids in klotho mice.

A, representative recordings demonstrating reduced slow wave amplitudes in the corpus and antrum of WT and klotho mice. Quantitative analysis of slow wave parameters obtained from 6 WT and 5 klotho mice are shown in Table 1. B, representative recordings of electrical responses to EFS (square pulses delivered by parallel platinum electrodes; 0.1 ms; 5–20 Hz for 1 s; supramaximal voltage). See Supplemental Fig. S2 for interpretation of electrical responses. Quantitative analysis of IJPs and EJPs from 6 WT and 7 klotho mice is shown in Table 2; the effects of l-NA on IJPs are shown in Supplemental Fig. S3. C, gastric emptying of solids (T_1/2) measured by [¹³C]octanoic acid breath test in klotho mice (n = 15) and WT or Het controls (n = 24).

Table 1.

Parameters of gastric slow waves in klotho mice and their age-matched WT or Het littermates

Region	Parameter	WT or Het	klotho	P
Corpus	RMP (mV)	−57.5[−61.4;−52.9]	−56.8[−64.3;−47.7]	n.s.
	Frequency (cycles min−1)	9.4[8.9;10.2]	10.7[8.9;11.1]	<0.001
	s.d. of ΔPP (% of ΔPP)	9.3[6.8;13.2]	9.9[6.7;13.0]	n.s.
	Amplitude (mV)	9.0[7.6;10.3]	6.7[4.9;9.1]	<0.001
	s.d. of amplitude (%)	20.6 ± 1.0	14.5 ± 1.1	<0.001
	Peak voltage (mV)	−48.3[−52.0;−45.1]	−49.2[−55.8;−42.7]	n.s.
	Rate of rise (mV s−1)	2.8[2.3;3.3]	2.6[1.8;3.4]	n.s.
Antrum	RMP (mV)	−57.9 ± 1.0	−62.6 ± 1.7	0.014
	Frequency (cycles min−1)	8.8[8.3;9.7]	9.8[7.8;10.3]	n.s.
	s.d. of ΔPP (% of ΔPP)	8.4[5.6;13.8]	7.8[4.8;15.2]	n.s.
	Amplitude (mV)	12.4[10.6;14.4]	8.5[6.2;12.0]	<0.001
	s.d. of amplitude (%)	19.1 ± 0.8	18.7 ± 1.2	n.s.
	Peak voltage (mV)	−45.1[−50.2;−38.5]	−51.8[−60.9;−46.8]	<0.001
	Rate of rise (mV s−1)	3.3[2.9;3.9]	2.8[2.1;3.9]	0.02

Impalements were obtained according to a grid with inter-square distances of 1.5 mm laid over the intact corpus + antrum tissues. Each square was probed only once. Data are mean ± s.e.m. or median[IQR] from 6 WT or Het (corpus: n = 75 recordings; antrum: n = 67) and 5 klotho mice (corpus: n = 48 recordings; antrum: n = 31). RMP, resting (diastolic) membrane potential; ΔPP, peak-to-peak interval.

Table 2.

Postjunctional electrical responses to EFS in the fundus of klotho mice and their age-matched WT littermates

EFS frequency	Parameter	WT	klotho	P
5 Hz	Fast IJP	16.5 ± 1.9 (9)	5.9 ± 1.0 (21)	<0.001
	Slow IJP	2.3[0.0;11.3] (9)	0.0[0.0;0.5] (21)	0.032
	Fast + slow IJP	18.1[14.0;31.1] (9)	5.8[2.8;9.5] (21)	0.001
	EJP	2.6[0.0;26.8] (7)	0.0[0.0;3.9] (21)	n.s.
10 Hz	Fast IJP	24.5[17.2;34.6] (23)	7.6[5.5;13.9] (24)	<0.001
	Slow IJP	13.0[1.2;30.1] (22)	0.0[0.0;2.8] (23)	<0.001
	Fast + slow IJP	40.6[33.8;65.0] (22)	10.3[5.8;18.3] (23)	<0.001
	EJP	7.9[0.0;17.9] (17)	0.9[0.0;29.9] (22)	n.s.
20 Hz	Fast IJP	35.1[19.8;52.8] (20)	9.2[6.5;23.1] (20)	<0.001
	Slow IJP	18.5[3.7;38.5] (18)	1.0[0.0;10.1] (19)	<0.001
	Fast + slow IJP	58.4[29.5;83.7] (18)	16.4[8.0;26.4] (19)	<0.001
	EJP	14.4[0.0;29.3] (13)	11.5[0.0;39.8] (18)	n.s.

Responses to square-wave electrical pulses (0.1 ms; 5–20 Hz for 1 s; supramaximal voltage) are expressed as areas over or under the curve (mV s), i.e. the areas enclosed by the pre-stimulus RMP and the junction potentials (Supplemental Fig. S2). Data are mean ±s.e.m. or median[IQR] from 6 WT and 7 klotho mice; numbers of responses analysed are shown in parentheses.

Klotho-deficient mice have small body sizes and lower postprandial blood glucose and serum insulin and IGF-I levels (Fig. 5A and B) partly reflecting reduced growth hormone (GH) production (Kuro-o et al. 1997). However, relative to constitutively dwarf mice deficient in GH or GH receptors, klotho mice have reduced body fat, hepatic glycogen stores and weight-normalized food intake (Mori et al. 2000; Coschigano et al. 2003) suggesting other abnormalities in appetite/weight control. We confirmed that klotho mice had reduced absolute food intake but their very low daily consumption (<50% of WT/Het intake) made exact quantification difficult. Therefore, we inferred nutritional status from measurements of dry faecal output (Bhutto & Morley, 2008). Dry faecal output was significantly reduced (Fig. 5A), despite accelerated whole-gut transit times (including defaecation, 177 ± 9 min; n = 9; WT: 302 ± 16 min; n = 10; P < 0.001) signifying reduced food consumption (Bhutto & Morley, 2008). As reported for food intake (Mori et al. 2000), weight-adjusted dry faecal output was not different in klotho and WT mice (1.27 ± 0.18 vs. 1.45 ± 0.14 mg h⁻¹ (g body wt) ⁻¹, respectively; P = 0.453), indicating that klotho mutants, unlike constitutively dwarf mice (Coschigano et al. 2003), do not consume the extra calories required for maintaining energy balance in the presence of increased heat loss from their greater surface area-to-volume ratios, a geometric consequence of smaller body sizes (Longo et al. 2010). These data, together with the observed reduced adiposity and marantic appearance, signify caloric deficiency in the klotho mice. Thus, impaired electrical pacemaking, reduced nitrergic inhibitory control of smooth muscles and inadequate food intake are the main dysfunctions that accompany ICC loss in Klotho-deficient mice.

Figure 5. Circulating and tissue factors important for ICC differentiation and maintenance are reduced in klotho mice.

A, body weight, dry faecal output and non-fasting blood glucose are reduced in klotho mice. B, circulating hormone levels. Serum insulin and IGF-I levels were reduced without an increase in serum corticosterone (n = 6–22/group). C, tissue growth factors. Expression of IGF-I mRNA (Igf1) and soluble SCF mRNA (soluble Kitl) was not different in the gastric musculature of klotho and WT mice (n = 3–6/group). Total (membrane bound + soluble) Kitl expression was reduced indicating reduced expression of membrane-bound SCF in these tissues (n = 9/group).

Self-renewal of ICC stem/progenitor cells and differentiation and survival of mature ICC depend on the production by smooth muscle cells of soluble and membrane-bound stem cell factor (SCF; Kit ligand; Kitl), respectively (Lorincz et al. 2008; Ordog et al. 2009). Insulin and IGF-I (both circulating and locally expressed), besides directly supporting ICC stem/progenitor cell proliferation (Lorincz et al. 2008), control the expression of SCF isoforms (Horvath et al. 2006; Ordog et al. 2009). Therefore, we examined whether reduced serum insulin and IGF-I in the klotho mice were also associated with reduced gastric Igf1 and SCF (Kitl) expression. Unlike systemic IGF-I, Igf1 mRNA in gastric muscles was unchanged. However, there was a significant decrease in mRNA for membrane-bound SCF (inferred from a significant decrease in total (membrane-bound + soluble) Kitl in the absence of a significant change in soluble Kitl (Horvath et al. 2006; Fig. 5C), a key regulator of ICC differentiation and maintenance (Lorincz et al. 2008; Ordog et al. 2009). These findings identify loss of membrane-bound SCF as a likely factor of ICC depletion in klotho mice.

ICC in normal tissues are resistant to oxidative stress caused, e.g., by hyperglycaemia (Horvath et al. 2005) but can become vulnerable to glycaemic stress when antioxidant defences are compromised (Choi et al. 2008). Klotho plays a major role in protecting tissues from oxidative damage in part by limiting signalling mediated by insulin and IGF-I receptors (Kurosu et al. 2005; Kuro-o, 2009) but it is unclear if loss of Klotho could lead to increased oxidative stress and ICC loss even when insulin/IGF-I levels are low. Therefore, we cultured whole gastric corpus + antrum musculature for 46–49 days with low and high glucose (100 and 500 mg dl⁻¹, respectively) and in the absence of insulin, IGF-I or serum. In Klotho-deficient tissues, reactive oxygen species (ROS), measured as MDA equivalents, were significantly increased and Kit mRNA, Kit protein and ICC networks were significantly reduced, particularly in the hyperglycaemic cultures (Fig. 6, Supplemental Fig. S4). These results indicate that reduced protection from oxidative stress contributes to ICC loss in klotho mice despite low insulin and IGF-I levels.

Figure 6. Klotho deficiency accentuates ICC loss from serum and growth factor deprivation by increasing oxidative stress and its effects.

Organotypic cultures of whole gastric corpus + antrum tunica muscularis tissues were started from klotho mice and age-matched wild-type littermates between 52 and 68 days of age and maintained with serum- and growth factor-free Medium 199 in the presence of normal (100 mg dl⁻¹) or high glucose (500 mg dl⁻¹) for 46–49 days (n = 3/group). A, oxidative stress assessed by measuring thiobarbituric acid-reactive substances as MDA equivalents. Groups not sharing the same superscript are significantly different by multiple comparisons. B, expression of Kit mRNA. C, Kit protein expression detected by Western immunoblotting. The panel on the right shows representative immunoblots. Representative confocal images are shown in Supplemental Fig. S4. Klotho deficiency increased oxidative stress and reduced Kit protein and mRNA expression in both normo- and hyperglycaemic cultures.

Discussion

In this study we show that mice deficient in the anti-ageing peptide Klotho have dramatic depletion of ICC stem cells leading to loss of mature ICC and reduced slow wave amplitudes signifying impaired electrical control of phasic smooth muscles of the stomach. Diminished phasic activity, together with profoundly reduced nitrergic control of smooth muscle relaxation, limits the functional reserves of the stomach and probably contributes to the marasmus of klotho mice (MacIntosh et al. 2001; Shimamoto et al. 2002; Parker & Chapman, 2004; Hays & Roberts, 2006; Bhutto & Morley, 2008). The observed changes occurred without neuron loss, significantly reduced Nos1 expression or overt smooth muscle dystrophy. The major age-related decline in ICC mirrors our recent finding of a similar change in humans without gastrointestinal motor disorders (Gomez-Pinilla et al. 2010) and the functional changes seen are also consistent with those in human subjects (MacIntosh et al. 2001; Shimamoto et al. 2002; Madsen & Graff, 2004; Parker & Chapman, 2004; Bhutto & Morley, 2008). Thus, reduced ICC differentiation from senescent precursors is a major cause of ageing-associated gastric dysfunctions including reduced food intake (Parker & Chapman, 2004; Hays & Roberts, 2006; Bhutto & Morley, 2008) and the klotho mouse is a suitable model for studying the mechanisms of these changes. We also found Klotho-deficient mice to have reduced levels of trophic factors required for the proper differentiation and maintenance of ICC (Horvath et al. 2005, 2006; Lorincz et al. 2008) and increased oxidative stress that further reduced ICC survival (Choi et al. 2008).

The cellular defect underlying gastric dysfunction in the elderly is unclear. In the distal gastrointestinal tract, ageing is associated with progressive neurodegeneration affecting mainly excitatory neurons (Phillips & Powley, 2007; Camilleri et al. 2008; Bernard et al. 2009). However, neurodegeneration is much less common or severe in the stomach (Phillips & Powley, 2007). In this study we found no evidence of neurodegeneration in the stomach of progeric klotho mice whereas our preliminary results indicate that it does occur in the colon. Thus, the degree of neuron involvement may be organ-dependent. Expression of Nos1 mRNA or protein also did not change significantly, despite a clear loss of nitrergic inhibitory neuromuscular neurotransmission. However, we cannot exclude the possibility that Nos1 dysfunction, for example, from reduced Nos1 dimerization (Gangula et al. 2007), could have contributed to some of the observed changes. We also found no change in the expression of smooth muscle-specific contractile protein Myh11 suggesting that, unlike in type 1 diabetic mice (Horvath et al. 2006), frank smooth muscle dystrophy is not likely to be a key factor in the gastric phenotypes of klotho mice. Again, more subtle abnormalities in contractile mechanisms (Bitar & Patil, 2004) cannot be excluded.

In contrast to the preservation of neurons and smooth muscle cells, we detected profound, diffuse loss of all major ICC classes throughout the gastric fundus, corpus and antrum. The functional abnormalities detected in the klotho mice are consistent with this ICC depletion. First, in the distal stomach, electrical slow waves are driven by pacemaker ICC in the myenteric region and further amplified by intramuscular ICC (Ordog et al. 1999; Dickens et al. 2001). Primary loss of intramuscular ICC and reduction in myenteric ICC in Kit mutant mice (W/W^v) has been shown to cause reduced slow wave amplitudes closely resembling our findings in klotho mice (Dickens et al. 2001). Reduced peak slow wave amplitudes are expected to lead to diminished voltage-dependent activation of Ca²⁺ influx into smooth muscle cells and consequent reduced phasic contractile activity (Sanders et al. 2006). Impaired peristalsis, in turn, may lead to symptoms and contribute to the inadequate food intake (Parker & Chapman, 2004) detected in the klotho mice. Second, early satiation and weight loss in the elderly may also reflect abnormal reflex relaxation of the proximal stomach in response to food intake (Parker & Chapman, 2004; Tack et al. 2004) and reduced intramuscular ICC play roles in both the afferent and efferent limbs of this reflex (Ward & Sanders, 2006). For example, intramuscular ICC have been shown to contribute to mechanotransduction via their association with vagal afferent nerve endings (Fox et al. 2002). Notably, in SCF mutant Sl/Sl^d mice, primary loss of intramuscular ICC and consequent reduction in vagal intramuscular arrays causes early satiation with decreased meal size and reduced body weight (Fox et al. 2002), which is consistent with findings in patients with functional dyspepsia (Tack et al. 2004). Primary loss of fundic intramuscular ICC in Sl/Sl^d and W/W^v mice may also interfere with the efferent limb of the accommodation reflex by reducing nitrergic inhibitory neuromuscular neurotransmission (Burns et al. 1996; Beckett et al. 2002; Ward & Sanders, 2006); loss of ICC was also accompanied by a dramatic decrease in nitrergic IJPs in the klotho fundus. However, no ICC involvement in these responses has been found by other investigators (Dixit et al. 2006; Huizinga et al. 2008). Nevertheless, the end result in the klotho animals is probably diminished stretch-induced relaxation of the whole stomach (Dixit et al. 2006), which may contribute to the observed reduced food intake. These changes may be further aggravated by gastric calcification.

In the klotho mouse, as well as in ageing humans (Gomez-Pinilla et al. 2010), the degree of ICC loss is disproportionate to symptoms or the changes in gastric emptying, which are modest or inconsistent. For example, gastric emptying in elderly people has been reported to be moderately delayed, particularly after large solid meals, unchanged, or even accelerated with consequent episodes of postprandial hypotension (Madsen & Graff, 2004; Bhutto & Morley, 2008). The reasons are unclear but it is striking that in ageing, the ICC depletion is diffuse and uniform across the regions and the layers of the stomach, whereas, e.g., in diabetic mice, the ICC loss occurs in foci, leading to heterogeneities in slow wave activity and resting membrane potentials (Ordog et al. 2009) which, in turn, can trigger irregular slow wave arrhythmias (Lammers et al. 2008) and cause grossly delayed gastric emptying (Ordog et al. 2009). Diffuse ICC loss is also associated with reduced slow wave amplitudes but no irregular arrhythmias in W/W^v mice (Ordog et al. 2002). Another possible explanation for the lack of delayed emptying in our model and in aged humans is that opposing actions of reduced phasic electrical activity predisposing to delayed gastric emptying and loss of intramuscular ICC, reduced nitrergic control of smooth muscle relaxation and calcification predisposing to impaired accommodation could cancel each other's effects (Ordog et al. 2009). This hypothesis explains how slight imbalances in the contributing factors could lead to the inconsistent changes in gastric emptying detected in various cohorts (Bhutto & Morley, 2008) while uniformly reducing overall functional capacity of the stomach (Hays & Roberts, 2006). Yet another possibility is that loss of excitatory and inhibitory electromechanical control via intramuscular ICC could potentially be compensated by increased sensitivity of the fundic smooth muscles to various neurotransmitters (Sergeant et al. 2002) and pharmacomechanical coupling (Sarna, 2008). Whether such compensation occurs in klotho mice remains to be investigated.

ICC loss in klotho mice appears to involve multiple mechanisms that reduce the self-renewal of ICC stem cells, impair their differentiation and limit the survival of mature ICC (Fig. 7). Klotho functions both as a cell-surface co-receptor for FGF23 and, after cleavage by the membrane-anchored proteases ADAM10 and ADAM17, as a circulating molecule. Here we report expression of Klotho in gastric glands, enteric neurons and smooth muscle cells. We did not detect Klotho expression in gastric ICC, therefore all effects on these cells must be mediated, directly or indirectly, by circulating or locally released Klotho. This is in contrast to deep muscular plexus-associated ICC of the small intestines, which were found to express Klotho mRNA by oligonucleotide microarrays (Chen et al. 2007). However, this result remains to be confirmed by other techniques.

Figure 7. Proposed model for the regulation of ICC differentiation and maintenance by Klotho.

The phenotype of ICC and ICC stem cells and the roles of insulin, IGF-I and SCF are from Lorincz et al. (2008) and Bardsley et al. (2010); the effects of ROS are from Choi et al. (2008). Ano1 represents anoctamin 1, an ICC marker (Gomez-Pinilla et al. 2009); Insr, insulin receptor; Igf1r, IGF-I receptor. Klotho may prevent ICC loss by limiting oxidative stress and by supporting ICC stem cell self-renewal and differentiation into ICC via an increase in circulating insulin and IGF-I and consequent increase in membrane-bound SCF in the gastric musculature. Klotho may maintain normal insulin and IGF-I levels by slightly reducing insulin/IGF-I signalling (Kurosu et al. 2005) as well as by preventing progeric marasmus and associated GH insensitivity (Thissen et al. 1994; Kuro-o et al. 1997) and reduction of GH secretion (Kuro-o et al. 1997). Klotho does not seem to have an effect on soluble SCF (open characters) but may prevent ICC stem/progenitor cell depletion by maintaining normal insulin/IGF-I levels and other mechanisms.

Mechanisms of action of circulating Klotho are not well understood but may involve modification of cell surface glycoproteins via its putative sialidase activity (Kuro-o, 2009). Main known targets of Klotho via this mechanism include insulin and IGF-I receptors and Trpv5/6 ion channels (Kuro-o, 2009), but these are not expressed by ICC (Horvath et al. 2006; Chen et al. 2007). SCF is a glycosylated peptide but it is unclear if its glycans are modified by Klotho or whether such a modification would change its signalling properties. Alternatively, ICC could be affected indirectly by one or more of the many systemic effects of circulating Klotho (Kuro-o, 2009). Here we show that ICC loss from Klotho deficiency involves reduced expression of SCF, the natural ligand for Kit receptor tyrosine kinase and a critical differentiation and survival factor for ICC (Ward & Sanders, 2006; Huizinga et al. 2009; Ordog et al. 2009). Reduced expression of SCF in klotho mice probably reflects low circulating insulin and IGF-I, which are established stimulators of SCF/Kitl expression (Horvath et al. 2006). SCF exists as a membrane-bound or locally secreted peptide (Ashman, 1999). While secreted SCF can stimulate the proliferation of Kit^low ICC stem/progenitor cells (Lorincz et al. 2008), local presentation of membrane-bound SCF is needed for their differentiation and for maintenance of mature ICC (Lorincz et al. 2008). Since in klotho mice only the expression of membrane-bound SCF was decreased, reduced SCF/Kit signalling contributes to the loss of mature ICC and to impaired differentiation from ICC stem/progenitor cells (Fig. 7).

An additional mechanism of mature ICC depletion in the absence of Klotho comes from reduced protection from oxidative stress (Choi et al. 2008; Kuro-o, 2009; Fig. 7). Klotho suppresses oxidative stress by limiting insulin/IGF-I-induced inactivation of forkhead box transcription factors (Kuro-o, 2009). However, since insulin and IGF-I levels are low in klotho mice, it is likely that Klotho also has antioxidant effects independent of these hormones. Indeed, here we show that Klotho's protective effects are demonstrable after ∼50 days of culturing gastric muscles without insulin, IGF-I or serum. While IGF-I is expressed in gastric muscles, its effects decline within ∼30 days in adult tissues (Lorincz et al. 2008) leading to ICC loss fully preventable by insulin or IGF-I treatment (Horvath et al. 2005). Thus, the loss of ICC in the WT cultures (Supplemental Fig. S4) reflects insufficient IGF-I production by the cultured muscles. Therefore, the limitation of oxidative damages seen in the WT cultures relative to the klotho tissues was unlikely to be mediated by the restriction of insulin/IGF-I signalling by gastric Klotho.

The most striking effect of Klotho deficiency in this study was the depletion of Kit^low ICC stem cells (Lorincz et al. 2008). Considering a constant attrition of mature cells, loss of precursors could explain the progressive decline of ICC (∼13% per decade of life) we recently detected in normal human subjects (Gomez-Pinilla et al. 2010). Stem cell senescence is an important mechanism of ageing (Rossi et al. 2008) and has been reported in the skin, intestinal crypts, testis and bone marrow of the klotho mouse (Liu et al. 2007). Klotho has been shown to prevent stem cell senescence by limiting Wnt signalling (Castilho et al. 2009). Another potential mechanism of Kit^low ICC stem cell loss is reduced insulin and IGF-I, which cause depletion of these cells from gastric muscles (Lorincz et al. 2008) (Fig. 7). Future studies will determine the significance of these pathways in ICC stem cell senescence in ageing.

The detection of Klotho expression in the stomach also raises the question of whether it plays a specific role besides being a local source of secreted peptide. Cell-surface Klotho is required for FGF23 to induce phosphate excretion and downregulation of active vitamin D (Kuro-o, 2009). Hypervitaminosis D, hyperphosphataemia and hypercalcaemia that develop in the absence of Klotho or FGF23 underlie the ectopic calcification prominent in the stomach and the hyperphosphataemia that occurs in both Klotho-deficient and Fgf23^−/− mice (Shimada et al. 2004). Interestingly, surgical removal of the gastric corpus in rats also resulted in chronic hypervitaminosis D, hyperphosphataemia and osteopenia (Gepp et al. 2000). While these results could be interpreted as an indication that gastric Klotho expression could influence global phosphate metabolism, the partially gastrectomized rats also had increased phosphaturia and bone turnover, suggesting that other mechanisms were also involved. Thus, a role for gastric Klotho in the regulation of calcium and phosphate metabolism remains to be established.

In summary, Klotho is expressed in the mouse stomach, and loss of Klotho results in a dramatic depletion of ICC stem cells and mature ICC, leading to gastric motor dysfunctions including abnormal electrical pacing. ICC loss in Klotho-deficient mice appears to reflect changes in multiple factors known to regulate ICC stem cell maintenance, ICC differentiation and survival including a reduction in the expression of the key ‘ICC growth factor’ SCF, low circulating insulin and IGF-I levels and increased oxidative stress. This study reveals Klotho as a key factor protecting ICC and their precursor cells and also indicates a role for ICC loss in ageing-associated gastric dysfunctions.

Glossary

Abbreviations

3-D

3-dimensional

APC

allophycocyanin

DAPI

4′,6-diamidino-2-phenylindole

EFS

electrical field stimulation

EJP

excitatory junction potential

FACS

fluorescence-activated cell sorting

FGF23

fibroblast growth factor 23

GH

growth hormone

Het

heterozygous

HuC/D

ELAV (embryonic lethal, abnormal vision, Drosophila)-like 3/4 (Hu antigen C/D)

ICC

interstitial cells of Cajal

IGF-I

insulin-like growth factor 1

Igf1r

IGF-I receptor

IJP

inhibitory junction potential

Insr

insulin receptor

IQR

interquartile range

Kitl

gene/mRNA encoding stem cell factor

Kl

gene/mRNA encoding Klotho

l-NA

N^ω-nitro-l-arginine

MDA

malondialdehyde

Nos1

NO synthase 1, neuronal

RMP

resting (diastolic) membrane potential

ROS

reactive oxygen species

SCF

stem cell factor

WT

wild-type

Author contributions

All authors contributed to the conception and design of the study and/or the analysis and interpretation of the results, drafting or revising the manuscript, and have approved the final version. F.I., D.T.A. and A.L. contributed equally to this work. All experiments were done in the Enteric Neuroscience Laboratories at the Mayo Clinic.

Supplemental material

Supplemental Table S1

Supplemental Table S2

Supplemental Table S3

Supplemental Figure S1

Supplemental Figure S2

Supplemental Figure S3

Supplemental Figure S4

As a service to our authors and readers, this journal provides supporting information supplied by the authors. Such materials are peer-reviewed and may be re-organized for online delivery, but are not copy-edited or typeset. Technical support issues arising from supporting information (other than missing files) should be addressed to the authors