INTRODUCTION

Klotho mutant mice, which harbor a defective klotho gene, have an extremely short lifespan (around 8 weeks). At 4–5 weeks of age, the mice develop multiple age-related syndromes, such as arteriosclerosis, osteoporosis, skin atrophy, infertility, thymic atrophy, and pulmonary emphysema (Kuro-o et al, 1997). Further evidence suggests that the klotho gene is responsible for aging. Introduction of a normal klotho gene into these mutant mice improves their phenotypes (Kuro-o et al, 1997), and overexpression of this gene in normal wild-type mice significantly extends their life spans (Kurosu et al, 2005). Although klotho mutant mice are considered to be a novel animal model for accelerated human aging, these mice do not express certain phenotypes usually seen in aged humans, such as brain atrophy with amyloid or senile plaque deposition (Kuro-o et al, 1997; Nagai et al, 2003).

Our group was the first to observe cognitive impairment in klotho mutant mice (Nagai et al, 2003). We showed that anti-death genes/proteins, Bcl-2 and Bcl-xL, are downregulated while the pro-death molecule, Bax, is upregulated in the hippocampus of klotho mutant mice (Nagai et al, 2003). In addition, Li et al (2004) demonstrated that synaptic structures and synaptophysin are reduced in number and expression, respectively, in the CA3 region of klotho mutant mice. However, very limited information is available regarding underlying mechanisms involved in significant cognitive function impairment in klotho mutant mice, although it is recognized that the klotho gene may be essential for maintaining cognitive function in aging organisms.

Earlier studies suggested that soluble signaling molecules (ie, cytokines and growth factors) and their receptors are expressed in both the developing and mature mammalian CNS (Mehler and Kessler, 1997). One intracellular signaling pathway activated by soluble signaling molecules involves two families of molecules: janus kinase (JAK) proteins and the signal transducer and activator of transcription (STAT) proteins (Ihle et al, 1995). Importantly, p-STAT3 immunoreactivity in hippocampal neurons of young subjects is substantially higher than that of older, cognitively normal humans and rodents (Chiba et al, 2009).

Because McN-A-343 (McN), a M1 agonist appeared to be more effective than either AF102B (AF; Fisher et al, 2003) or talsaclidine (TC; Walland and Pieper, 1998) in improving cognitive impairment in klotho mutant mice (Supplementary Figure S3), we applied McN in the present study. JAK2 is an upstream regulator of STAT3 activation. As inhibition of JAK2 activity by AG490 (AG) results in decreased STAT3 phosphorylation, we applied AG in the present study to inhibit the JAK2/STAT3 signaling axis (Chiba et al, 2009).

In the present study, we investigated whether klotho gene deficiency affects the p-JAK2/STAT3 signaling axis, and also examined whether klotho gene deficiency alters the cholinergic neuronal system. To this aim, we examined the neurochemical, electrophysiological, and behavioral parameters accompanied by aging in klotho mutant mice. We propose here that inactivation of the JAK2/STAT3 signaling axis and M1 mAChR downregulation lead to cognitive impairments in klotho mutant mice.

MATERIALS AND METHODS

Animals

All animals were treated in accordance with the National Institutes of Health (NIH) Guide for the Humane Care and Use of Laboratory Animals (NIH Publication No. 85-23, 1985; www.dels.nas.edu/ila). The present study was performed in accordance with the Institute for Laboratory Research (ILAR) guidelines for the care and use of laboratory animals. Mice were maintained under a 12-h light:12-h dark cycle and fed ad libitum. Since klotho mutant mice are infertile, wild-type and klotho mutant mice were generated by crossing heterozygous klotho mutant mice (Kuro-o et al, 1997; Nagai et al, 2003). The mice were screened by polymerase chain reaction (PCR) analysis using DNA extracted from tail specimens. More details on the gene characterization were described in the supporting information.

Guide Cannula Implantation and Intracerebroventricular Infusion

Forty-day-old mice were anesthetized with pentobarbital (40 mg/kg, i.p.) and a stainless steel guide cannula (AG-4; Eicom, Kyoto, Japan) was implanted into the right lateral ventricle (stereotaxic coordinates: 0.5 mm posterior to bregma, 1 mm right to the midline, and 2 mm ventral to the dura, according to the atlas of Franklin and Paxinos, 2008 and Shin et al, 2009). Microinfusion into the lateral ventricle was performed at a rate of 1 μl/min. More details on the guide cannula implantation and intracerebroventricular (i.c.v.) infusion were described in the supporting information.

Drug Treatment and Animal Handling

Drug treatment started after 2 days of recovery from guide cannula implantation. Because we observed, in an earlier study (Nagai et al, 2003), that cognitive impairment did not occur in klotho mutant mice until 6 weeks of age, mice received McN-A-343 (McN, M1, mAChR agonist; Tocris Bioscience, Ellisville, MO, USA; 1.0 μg/μl in saline, i.c.v.) on postnatal day (PND) 42, 46, 49, and 53. AG490 (JAK2/STAT3 inhibitor; Tocris Bioscience; 15 mg/kg/2 ml, i.p.) or k252a (TrkB inhibitor; Tocris Bioscience; 0.3 mg/kg/2 ml, i.p.) was injected 1 h after each McN treatment. Novel object recognition tests or conditioned-fear learning tests were conducted on PND 54 and 55 (Supplementary Figure S1). The dose of each drug used in this study was selected based on previous studies (Chiba et al, 2009; Liu et al, 2005; Wall et al, 2001) and our preliminary study (Bach et al, 2010). Mice were killed 30 min after the test trial for western blot analyses (p-JAK2, p-STAT3, mAChRs, nAChRs, and BDNF), autoradiography, neurochemical measurements, and electrophysiological analyses. To evaluate p-ERK and p-CREB expression, mice were killed 30 min after the training trial. Mouse body weight was recorded on a daily basis from PND 35 to 55 (Supplementary Figure S5). More details on the preparation of each reagent were described in the supporting information.

Novel Object Recognition Test and Conditioned-Fear Learning Test

Novel object recognition test and conditioned-fear learning test were performed with a slight modification as described previously (Jin et al, 2009; Nagai et al, 2003). Detailed procedure is described in the supporting information.

Western Blot Analysis and Immunoprecipitation Assay

Western blot analysis and immunoprecipitation assay were done as described previously (Jin et al, 2009). More details on the procedure and antibody were described in the supporting information.

Autoradiography

Autoradiograms were generated and analyzed with a slight modification as described previously (Jin et al, 2009; Zhang and Buccafusco, 2000). More details on the autoradiography were described in the supporting information.

Level of ACh and Activities of AChE and ChAT

Hippocampi were homogenated in ice-cold 20 mM sodium phosphate buffer (pH 7.4). These were centrifuged at 12 000 g for 30 min, at 4 °C. The supernatant was assayed for ACh levels using Amplex Red Acetylcholine/Acetylcholinesterase (AChE) Assay Kit (A-12217) (Eugene, Oregon, USA), according to the manufacturer’s instruction. Detailed procedure was described in the supporting information. AChE activity was measured using Amplex Red Acetylcholine/AChE Assay Kit (A-12217) (Eugene, Oregon, USA), according to the manufacturer’s instruction. For choline acetyltransferase (ChAT) activity, 5% tissue homogenates were prepared in ice-cold 20 mM sodium phosphate buffer (pH 7.4), and kept frozen overnight at −20 °C. These were thawed the following day and centrifuged at 12 000 g for 1 h, at 4 °C. The supernatant was assayed for ChAT activity according to the method of Chao and Wolfgram (1972). Detailed procedure was described in the supporting information.

Electrophysiological Recordings

Transverse hippocampal slices were prepared from wild-type or klotho mutant mice, according to the previous report with a minor modification (Min et al, 2009). CA1 field potentials (FPs) were evoked by stimulating Schaffer collaterals with 0.2 ms pulses delivered through concentric bipolar stimulating electrodes (FHC, Bowdoinham, ME, USA), and recorded extracellularly in CA1 stratum radiatum. Responses were quantified as the initial FP slope in CA1. Baseline responses were recorded using half maximal stimulation intensity at 0.033 Hz. Long-term potentiation (LTP) was examined using a conventional stimulation paradigm: four episodes of theta burst stimulation (TBS) consisting of eight bursts of four 100 Hz pulses were administered every 200 ms. Because four TBS episodes induced robust LTP that lasted for >60 min, we measured the magnitude of NMDAR-dependent LTP induced by one TBS episode. More details on the hippocampal slice preparation were described in the supporting information.

Statistical Analyses

Data were analyzed using IBM SPSS version 19.0 (IBM, Chicago, IL, USA). One-way ANOVA was applied for the two data groups (Figure 1a and b). Three-way ANOVA was performed for the effect of klotho gene mutation, McN, and AG490 (or k252a) (Figures 1c and d, 2, 3, and 5). The effect of k252a on the McN-mediated pharmacological actions in the klotho mutant mice was analyzed by two-way ANOVA, with McN and k252a as between-subjects factors (Figure 4). The statistical values are presented in Supplementary Tables S1–S4. After analysis by two- or three-way ANOVA, post-hoc multiple pairwise comparisons with Bonferroni’s correction were then performed. P-values of <0.05 were considered statistically significant.

Figure 1
figure 1

Changes in protein expression of p-JAK2, p-STAT3 (a) and cholinergic receptors (b), and AG antagonism against McN-induced effects on reduced M1 mAChR expression (c) and [3H]pirenzepine-binding density (d) in the hippocampus of klotho mutant mice. Veh=Vehicle (50% DMSO in saline, the solvent for AG 490); Sal=Saline; AG=AG490 (15 mg/kg/2 ml, i.p.); McN=McN-A-343 (1.0 μg/μl in saline, i.c.v.). Each value is the mean±SEM of six animals. *P<0.01 vs corresponding wild-type mice; #P<0.01 vs klotho mutant mice treated with Veh and Sal; &P<0.05, &&P<0.01 vs klotho mutant mice treated with Veh and McN (one-way ANOVA (a, b) or three-way ANOVA followed by multiple pairwise comparisons with Bonferroni’s correction (c, d)).

PowerPoint slide

Full size image
Figure 2
figure 2

AG antagonism against McN-induced changes in acetylcholine level (a), and choline acetyltransferase (ChAT) protein expression (b), ChAT activity (b), AChE protein expression (c), AChE activity (c), p-JAK2 (d), p-STAT3 expression (e), and NMDAR-dependent LTP (f, g) in the hippocampus of klotho mutant mice. (f) Field EPSP slope (% of baseline) was plotted against time. LTP was induced by application of one TBS episode. Insets: representative traces recorded before (thin line), 60 min (thick line) after one TBS episode. A: Wild-type/Sal+Veh (n=11), B: klotho mutant/Sal+Veh (n=11), C: klotho mutant/McN+Veh (n=10), D: klotho mutant/McN+AG490 (n=11), E: wild-type/McN+Veh (n=8). (g) Comparison of average early-LTP magnitude 58–60 min after TBS application. AG490 blocked the reversal effects of McN on NMDAR-dependent LTP. Veh=Vehicle (50% DMSO in saline, the solvent for AG490); Sal=Saline; AG=AG490 (15 mg/kg/2 ml, i.p.); McN=McN-A-343 (1.0 μg/μl in saline, i.c.v.). Each value is the mean±SEM of six animals (a–e) or 8–11 animals (f, g). *P<0.05, **P<0.01 vs corresponding wild-type mice; #P<0.05, ##P<0.01 vs klotho mutant mice treated with Veh and Sal; &P<0.05, &&P<0.01 vs klotho mutant mice treated with Veh and McN (three-way ANOVA followed by multiple pairwise comparisons with Bonferroni’s correction).

PowerPoint slide

Full size image
Figure 3
figure 3

AG490 antagonism against McN-induced effects on PKCβII (a), p-ERK (b), p-CREB (c), and BDNF (d) expression in the hippocampus of klotho mutant mice. Veh=Vehicle (50% DMSO in saline, the solvent for AG490); Sal=Saline; AG=AG490 (15 mg/kg/2 ml, i.p.); McN=McN-A-343 (1.0 μg/μl in saline, i.c.v.). Each value is the mean±SEM of six animals. *P<0.05, **P<0.01 vs corresponding wild-type mice; #P<0.01 vs klotho mutant mice treated with Veh and Sal; &P<0.05, &&P<0.01 vs klotho mutant mice treated with Veh and McN (three-way ANOVA followed by multiple pairwise comparisons with Bonferroni’s correction).

PowerPoint slide

Full size image
Figure 4
figure 4

K252a antagonism against McN-induced effects on p-JAK2 (a), p-STAT3 (b), choline acetyltransferase (c), AChE (d), and M1 mAChR (f), and the level of acetylcholine (e) in the hippocampus of klotho mutant mice. Veh=Vehicle (5% DMSO in saline, the solvent for k252a); Sal=Saline; k252a=k252a (0.3 mg/kg/2 ml, i.p.); McN=McN-A-343 (1.0 μg/μl in saline, i.c.v.). Each value is the mean±SEM of six animals. #P<0.05, ##P<0.01 vs klotho mutant mice treated with Veh and Sal; &P<0.05, &&P<0.01 vs klotho mutant mice treated with Veh and McN (two-way ANOVA followed by multiple pairwise comparisons with Bonferroni’s correction).

PowerPoint slide

Full size image

RESULTS

p-JAK2 and p-STAT3 Expression Was Significantly Decreased in the Hippocampus of klotho Mutant Mice

Our experimental protocol is shown in Supplementary Figure S1. First, we investigated whether klotho gene deficiency affected p-JAK2/p-STAT3 expression, since p-JAK2/p-STAT3 decreases in an age-dependent manner (De-Fraja et al, 1998). p-JAK2 and p-STAT expression were significantly decreased (p-JAK2, P=0.00257; p-STAT3, P=0.000779) in klotho mutant mice (Figure 1a; Supplementary Table S1).

M1 mAChR Expression Was Significantly Decreased in the Hippocampus of klotho Mutant Mice

Next, we investigated whether mAChR expression was decreased in klotho-deficient mice, as M1 mAChR is downstream of the p-JAK2/p-STAT3 signaling axis (Chiba et al, 2009).

Of the five muscarinic acetylcholine receptors (M1–M5 mAChRs) examined, only M1 mAChR expression was significantly decreased (P=0.00464) in klotho mutant mice. Conversely, nicotinic acetylcholine receptor subtypes (α4 nAChR, α7 nAChR, β2 nAChR, or α4 β2 nAChR) were not significantly different in the mutant mice (Figure 1b; Supplementary Table S1).

AG490 Antagonism Against McN-Induced Effects on Reduced M1 mAChR Expression and [3H]-Pirenzepine-Binding Density in the Hippocampus of klotho Mutant Mice

Since M1 mAChR expression and p-JAK2/p-STAT3 signaling is decreased in animal models of Alzheimer’s disease (Chiba et al, 2009), we examined whether McN, a specific M1 mAChR agonist, affected M1 mAChR expression or [3H]-pirenzepine-binding density in the hippocampus of klotho mutant mice. In addition, we investigated whether AG, an inhibitor of JAK2/STAT3 signaling, affected McN-induced changes.

Three-way ANOVA indicated significant effects of klotho gene mutation, McN and AG490, and a significant interaction between klotho gene mutation and McN (Supplementary Table S1). A post-hoc test revealed that McN treatment significantly attenuates (P=0.00308) the reduction in M1 mAChR expression in klotho mutant mice. This attenuation was significantly inhibited (P=0.0268) by AG490. However, AG490 per se did not significantly alter M1 mAChR expression in either wild-type or klotho mutant mice treated with vehicle and saline (Figure 1c). Representative autoradiograms showing the distribution of [3H]-pirenzepine binding are presented in Figure 1d. Three-way ANOVA showed significant effects (CA1, CA3, and dentate gyrus (DG)) of klotho gene mutation, McN and AG490, and a significant interaction between klotho gene mutation and McN (DG; Supplementary Table S1). Strong binding was noted in the CA1 region and DG, whereas mild binding was observed in the CA2/3 region in the hippocampus of wild-type mice with or without McN. However, a post-hoc test revealed that [3H]-pirenzepine-binding densities were significantly lower (CA1, P=6.82 × 10−7; CA2/3, P=3.14 × 10−5; DG, P=7.59 × 10−8) in the hippocampus of klotho mutant mice. AG490 did not affect these reduced binding densities per se in klotho mutant mice. However, McN significantly reversed (CA1, P=0.00156; CA2/3, P=0.00898; DG, P=7.82 × 10−4) the decreases in [3H]-pirenzepine binding in the hippocampus of klotho mutant mice. AG490 significantly counteracted (CA1, P=0.0209; CA2/3, P=0.0416; DG, P=0.00960) McN effects (Figure 1d).

AG490 Antagonism Against McN-Induced Effects on Altered Cholinergic System and p-JAK2/p-STAT3 in klotho Mutant Mice

Since earlier reports demonstrated that M1 mAChR agonists reverse cholinergic impairment (Calabresi et al, 1998; Fisher et al, 2003; Walland et al, 1998), we next investigated whether McN attenuated changes in klotho mutant mice cholinergic parameters. We subsequently examined whether AG affected McN pharmacological effects. Three-way ANOVA showed significant effects of klotho gene mutation (ACh, ChAT expression ChAT activity, AChE expression, and AChE activity), McN (ACh, ChAT expression ChAT activity, AChE expression, and AChE activity), and AG490 (ACh, ChAT expression, and ChAT activity), and a significant interaction between klotho gene mutation and McN (ACh, AChE expression, and AChE activity; Supplementary Table S2). A post-hoc test indicated that acetylcholine ACh level (P=4.24 × 10−5), ChAT protein expression (P=5.15 × 10−5), and ChAT activity (P=8.59 × 10−4) were significantly decreased in the hippocampus of klotho mutant mice (Figure 2a and b). In contrast, AChE protein expression (P=2.15 × 10−4) and AChE activity (P=6.38 × 10−4) were significantly increased in klotho mutant mice (Figure 2c). McN treatment significantly attenuated (ACh level, P=7.38 × 10−4; ChAT expression, P=9.87 × 10−4; ChAT activity, P=0.00673; AChE expression, P=0.0160; AChE activity, P=0.00998) these alterations in klotho mutant mice. AG treatment significantly inhibited McN-induced effects on decreased ACh level (P=0.0182), ChAT expression (P=0.0187), and its activity (P=0.0200). However, AG failed to significantly inhibit McN-induced effects on increased AChE expression and its activity in klotho mutant mice (Figure 2a–c).

Although there is no existing evidence that an M1 mAChR agonist per se increases p-JAK2/STAT3 signaling in an animal model of aging or neurodegeneration, IL-6, a STAT3 activator, increases carbachol-induced p-ERK (Chiba et al, 2009). Additionally, carbachol plays a role as a mAChR agonist via the M1 type mAChR (Anagnostaras et al, 2003). Therefore, we examined whether McN altered p-JAK2/STAT3 signaling in the hippocampus of klotho mutant mice. We then investigated whether AG affected pharmacological effects of McN. Three-way ANOVA revealed significant effects of klotho gene mutation, McN and AG490 on p-JAK2 and p-STAT3 expression. In addition, a significant interaction between klotho gene mutation and McN was shown in the p-JAK2 expression (Supplementary Table S2). A post-hoc test indicated that McN significantly attenuated reduced p-JAK2 (P=0.00191) and p-STAT3 (P=0.00234) expression in klotho mutant mice, although McN treatment did not significantly alter p-JAK2 or p-STAT3 expression in wild-type mice. Additionally, AG significantly inhibited McN-induced effects on reduced p-JAK2 and p-STAT3 in klotho mutant mice (p-JAK2, P=0.00837; p-STAT3, P=0.0236). However, AG490 per se did not significantly affect p-JAK2 and p-STAT3 expression in either wild-type or klotho mutant mice (Figure 2d and e).

AG490 Antagonism Against McN-Induced Effects on Impaired NMDAR-Dependent LTP in klotho Mutant Mice

After behavioral testing, we measured LTP magnitude in the CA1 area of hippocampal slices prepared from klotho mutant or wild-type mice using a maximal LTP protocol (Supplementary Figure S2; Figure 2f and g). We tested whether the reduced cholinergic activity in klotho mutant mice was a possible reason for impaired LTP induction with the use of McN. McN treatment significantly reversed the LTP impairment in klotho mutant mice (Figure 2f and g). Three-way ANOVA showed significant effects of klotho gene mutation and AG490 (Supplementary Table S2). The magnitude of NMDAR-dependent LTP in slices from klotho mutant mice treated with McN was 142.38±10.04% of baseline (n=10) 1 h after TBS application. These values were higher than those in slices from klotho mutant mice treated with vehicle (116.80±6.59% of baseline, 1 h after TBS application, P=0.0389; Figure 2g). Although the magnitude of NMDAR-dependent LTP in slices from wild-type mice treated with McN alone was slightly higher than in slices from control mice treated with vehicle, the two groups were not significantly different (154.13±15.84% vs 143.82±7.71% of baseline, respectively, P=0.434).

We then investigated whether the M1 agonist required JAK2/STAT3 signaling to attenuate the NMDAR-dependent LTP impairment. In the presence of AG490, the McN effect on LTP induction was not reversed; the average LTP magnitude in slices from klotho mutant mice treated with McN and AG490 was 117.08±4.60, significantly lower than that in slices from klotho mutant mice treated with MCN or vehicle (142.38±10.04% of baseline; P=0.0406). These results suggest that M1 mAChR and JAK2/STAT3 interaction is required to induce normal NMDAR-dependent LTP.

AG490 Antagonism Against McN-Induced Effects on PKCβII, p-ERK, p-CREB and BDNF Expression in the Hippocampus of klotho Mutant Mice

M1 mAChR stimulation activates a G-protein belonging to the pertussis toxin-insensitive Gq/11 family. Gq/11 subunits stimulate phospholipase C, inducing IP3-dependent calcium release and protein kinase C (PKC) activation (Fisher et al, 2003). PKC pathways are strong candidates for mediating the molecular changes that underlie spatial learning, as they play critical roles in neurotransmitter release and synaptic plasticity, and deletion of specific PKC genes results in learning deficits. To understand the role of PKC in our experimental system, we examined PKC isozyme protein expression in the hippocampus of klotho mutant mice (Figure 3a; Supplementary Figure S4). Three-way ANOVA indicated significant effects of klotho gene mutation, McN and AG490 on PKCβII expression (Supplementary Table S3), but not on PKCα, PKCβI, PKCδ, or PKCζ expression (Supplementary Figure S4; Supplementary Table S6). Post-hoc analysis showed that PKCβII expression was significantly and selectively decreased (P=1.45 × 10−6) in klotho mutant mice (Figure 3a), whereas PKCα, PKCβI, PKCδ, and PKCζ (Supplementary Figure S4) were not significantly different from wild-type. McN significantly attenuated (P=0.00215) the decreased PKCβII expression in klotho mutant mice. Consistently, AG490 significantly counteracted McN-mediated attenuation (P=0.0122).

ERK1/2 signal transduction pathways are downstream of the PKC pathway (Fisher et al, 2003) and are also involved in synaptic plasticity, learning, and memory. ERK is necessary for the development of several forms of memory, such as fear conditioning, conditioned taste aversion, spatial memory, step-down inhibitory avoidance, and object recognition memory (Alkam et al, 2010). ERK regulates CREB function and BDNF expression (Williams et al, 2008). CREB is a critical mediator of activity-induced transcriptional signaling of dendrite growth in response to neuronal activity (Wayman et al, 2006). Three-way ANOVA indicated significant effects of klotho gene mutation, McN and AG490, and a significant interaction between klotho gene mutation and McN on the expression of p-ERK, p-CREB, and BDNF (Supplementary Table S3). No significant change in ERK1/2 expression was observed among the groups in this study. However, a post-hoc test revealed that p-ERK1/2 expression was significantly decreased (P=5.22 × 10−8; Figure 3b) in klotho mutant mice. Similar to ERK, p-CREB and BDNF expression were significantly decreased (p-CREB, P=6.02 × 10−8; BDNF, P=4.00 × 10−6) in the hippocampus of klotho mutant mice (Figure 3c and d). McN significantly attenuated these decreases in p-ERK (P=4.26 × 10−5), p-CREB (P=9.18 × 10−5), and BDNF (P=0.00269) expression. Consistently, AG490 significantly reversed McN-mediated attenuations in klotho mutant mice (p-ERK, P=0.00163; p-CREB, P=0.0192; BDNF, P=0.0306; Figure 3b–d).

k252a Antagonism Against McN-Induced Effects on p-JAK2/p-STAT3, ChAT Protein Expression, ChAT Activity, AChE Protein Expression and Activity, ACh Level, and M1 mAChR Protein Expression in klotho Mutant Mice

BDNF, a neurotrophin, activates a high-affinity cell surface receptor (TrkB) that is coupled to activation of phosphatidylinositol-3-kinase and protein kinase B (Akt). BDNF is increased in the hippocampus of rats during and after performance of a spatial learning task (a radial-arm maze), and both acquisition and maintenance of spatial memory are impaired when BDNF levels are decreased using antisense methods (Mizuno et al, 2003). BDNF specifically binds to TrkB receptors and induces receptor dimerization, phosphorylation, and activation of the intracellular tyrosine kinase domain (Mizuno et al, 2003). Therefore, we examined whether McN-induced effects on p-JAK2 and p-STAT3 expression and cholinergic parameters (ie ChAT, AChE, Ach, and M1 mAChR) in klotho mutant mouse hippocampus were via BDNF-TrkB signaling. Two-way ANOVA indicated significant effects of McN (all parameters in Figure 4) and k252a (p-JAK2, ChAT expression and activity, ACh level, and M1 mAChR expression), and a significant interaction between McN and k252a (p-JAK2) (Supplementary Table S3). A post-hoc test revealed that McN treatment significantly increased p-JAK2 (P=1.18 × 10−4), p-STAT3 (P=0.00222), ChAT expression (P=9.71 × 10−4), ChAT activity (P=1.79 × 10−4), ACh level (P=0.00709), and M1 mAChR expression (P=1.97 × 10−4) in klotho mutant mice, whereas McN treatment significantly decreased AChE expression (P=0.0109) and its enzymatic activity (P=0.0307). In contrast, k252a treatment, an inhibitor of TrkB receptor tyrosine phosphorylation, resulted in significant inhibition of McN-induced p-JAK2 (P=0.00233), p-STAT3 (P=0.0368), ChAT expression (P=0.0137), ChAT activity (P=0.0107), and M1 mAChR expression (P=0.0195), and ACh level (P=0.0342) in klotho mutant mice. However, k252a per se did not significantly decrease AChE expression or its enzymatic activity (Figure 4).

Antagonism of AG490 or k252a Against McN-Induced Effects on Cognitive Impairment in klotho Mutant Mice

As the hippocampus is important in the formation of recognition memory in both humans and animals (Zola-Morgan et al, 1986), we evaluated visual recognition memory in a novel-object recognition memory task. The recognition test is based on the natural tendency of rodents to investigate a novel object instead of a familiar one. The choice to explore the novel object reflects the use of learning and (recognition) memory processes. Three-way ANOVA revealed significant effects of klotho gene mutation and AG490 or k252a (Supplementary Table S4). A post-hoc test indicated a significant decrease (P=1.92 × 10−4) in the exploratory preference in klotho mutant mice, suggesting cognitive impairment. McN attenuated this decrease (P<0.00558; Figure 5a).

Figure 5
figure 5

Antagonism by AG490 or k252a against McN-induced effects on novel object recognition (a) and conditioned-fear learning performance (b) of klotho mutant mice. Veh=Vehicle (50% DMSO in saline, the solvent for AG490); Sal=Saline; AG=AG490 (15 mg/kg/2 ml, i.p.); k252a=k252a (0.3 mg/kg/2 ml, i.p.); McN=McN-A-343 (1.0 μg/μl in saline, i.c.v.). Neither 50% DMSO nor 5% DMSO solution affect significantly memory performance; thus, shown as Veh for AG490. Each value is the mean±SEM of 15 animals. *P<0.01 vs corresponding wild-type mice; #P<0.01 vs klotho mutant mice treated with Veh and Sal; &P<0.05, &&P<0.01 vs klotho mutant mice treated with Veh and McN (three-way ANOVA followed by multiple pairwise comparisons with Bonferroni’s correction).

PowerPoint slide

Full size image

We then examined associative fear memory in a conditioned-fear task. In contextual fear conditioning, which assesses hippocampus-dependent memory, animals learn to fear the context in which they are trained to associate conditioned and unconditioned stimuli. Three-way ANOVA indicated a significant effect of klotho gene mutation, McN and AG490 or k252a, and a significant interaction between klotho gene mutation and McN (Supplementary Table S4). A post-hoc test revealed that contextual freezing was significantly lower in klotho mutant mice (P=1.97 × 10−8 vs wild-type). McN treatment significantly attenuated (P=9.47 × 10−5) the lowered contextual freezing in klotho mutant mice (Figure 5b). Both AG490 and k252a treatment significantly counteracted McN-mediated improvements in novel objective recognition (AG490, P=0.0320; k252a, P=6.08 × 10−4) and conditioned-fear learning performance (AG490, P=0.0443; k252a, P=0.00218) (Figure 5).

DISCUSSION

This study revealed that klotho gene deficiency impairs hippocampal cholinergic neural systems as evidenced by significant decreases in M1 mAChR gene expression, M1 mAChR-binding density, ACh levels, ChAT activity, and ChAT gene expression, and by significant increases in AChE activity and AChE gene expression. These changes in the cholinergic neuronal system may be due to klotho gene deficiency in presynaptic cholinergic nerve terminals and postsynaptic neurons expressing M1 mAChR in the hippocampus. Klotho gene deficiency consistently decreased PKCβII, p-ERK, p-CREB and BDNF expression, and NMDAR-dependent LTP (Supplementary Figure S6). To further understand the role of M1 mAChR in our experimental system, we examined the effects of McN, an M1 agonist.

McN-induced M1 mAChR stimulation activates PKCβII-, p-ERK-, p-CREB-, and BDNF-dependent pathways in the hippocampus of klotho mutant mice. BDNF activation leads to JAK2/STAT3 and TrkB signaling, followed by enhanced ChAT activity and ACh level, and further stimulation of M1 mAChR. Moreover, McN enhanced NMDAR-dependent LTP via JAK2/STAT3 signaling. Interestingly, the ChAT promoter contains a STAT consensus binding site (Cattaneo et al, 1999). We observed here that ChAT loss was pronounced in klotho mutant mice, similar to that observed in the aged brain (O’Neil et al, 1987). Thus, it is possible that presynaptically altered levels of ChAT and JAK2/STAT3 in klotho mutant mice are concomitant with basal hippocampal neuronal apoptosis (Nagai et al, 2003), and may contribute to both cognitive and behavioral symptoms. In contrast, although McN inhibited AChE expression and its activity, neither AG nor k252a counteracted these McN effects. Therefore, this phenomenon remains to be characterized.

In addition, we showed in the present study that klotho gene deficiency significantly deteriorated p-JAK2/p-STAT3 expression in hippocampal neurons, suggesting a critical interaction between the JAK2/STAT3 signaling axis and the cholinergic neurotransmitter system. This is the first evidence that JAK2/STAT3 inactivation in klotho mutant mice contributes to the pathogenesis of cholinergic dysfunction via dual mechanisms (ie, presynaptic downregulation of cholinergic genes such as ChAT and postsynaptic desensitization of M1 mAChR; refer to Supplementary Figure S6).

The decreased NMDAR-dependent LTP observed in klotho mutant mice may result from either decreased M1 mAChR expression or from downregulation of the machinery, such as PKC coupling to M1 mAChR, since LTP induction is impaired in M1 receptor knockout mice (Anagnostaras et al, 2003). M1 receptors are coupled to the Gq/11 family of G-proteins and activate PKC-mediated signaling cascades (Lu et al, 1999). In contrast, NMDA receptors are regulated by PKC (MacDonald et al, 2001). Interestingly, M1 receptors co-localize with NR1 on the dendrites and soma of hippocampal pyramidal neurons, and activation of M1 receptors potentiates NMDA receptor currents (Marino et al, 1998). Similar to previous reports (Sun and Alkon, 2005), our data might link postsynaptic cholinergic dysfunction with other substrates (PKCβII and NMDA receptor) known to play essential roles in cognitive processes and provide further insight into the mechanisms by which impairment of M1-mediated signaling may underlie the cognitive decline of klotho mutant mice. Interestingly, mice lacking PKCβ suffer from deficits in both cued and contextual fear conditioning (Weeber et al, 2000), which may be in line with our current results.

We observed here that McN facilitated LTP induction by activation of the JAK2/STAT3 signaling axis in klotho mutant mice. This suggests that NMDAR activation could be a mechanism for LTP facilitation through activation of the JAK2/STAT3 signaling axis by M1 mAChR. Reduced JAK2 activation in klotho mutant mice may result in impaired NMDAR function. As a consequence of NMDAR dysfunction, Ca2+ influx and extrusion across the plasma membrane, cytosolic buffering, and uptake into organelles, may be altered in neurons. Evidence showing that calcium influx is mediated by JAKs/STATs (Orellana et al, 2005) supports this notion. However, Nicolas et al (2012) showed that a JAK2 inhibitor, AG490, does not block the induction of NMDAR-dependent LTP. These results seem to contradict our observation that AG490 treatment blocked McN-induced recovery of NMDAR-dependent LTP. It suggests that JAK2/STAT3 activation differentially regulates synaptic plasticity and/or would be affected under klotho gene dysfunction.

In general, compensative upregulation of M1 receptors expressed on postsynaptic sites could occur if presynaptic ACh levels and ChAT levels decreased concomitantly with increased AChE expression. Unexpectedly, in our study, hippocampal M1 mAChR was downregulated in klotho mutant mice. Interestingly, Nomura et al (1997) demonstrated that SAMP8 also show decreased hippocampal M1 mAChR-binding activity. These findings raise the possibility that it may be difficult to induce compensatory adaptation in response to cholinergic neuronal dysfunction in models of aging, although this idea remains to be fully elucidated.

mAChRs undergo desensitization or downregulation in response to prolonged muscarinic agonist stimulation, resulting in a decline in mAChR density (Svedberg et al, 2004). However, McN treatment in the present experimental condition led to upregulated M1 mAChR in the hippocampus of klotho mutant mice. Functionally selective muscarinic agonists (ie, McN, oxotremorine, pilocarpine, and AF102B) antagonize the stimulatory effect of full agonist (ie, carbachol) on adenylate cyclase in cells expressing muscarinic M1 (Gurwitz et al, 1994). Thus, many of the functionally selective muscarinic receptor agonists seem to fall within the category of partial agonists (Hoyer and Boddeke, 1993). Bursa et al (1995) demonstrated that treatment of corticostriatal neurons with McN upregulates M1 mAChR mRNA in primary rat cortical neuronal cultures, suggesting that McN as a partial agonist could exert an antagonistic effect on this receptor transcript. Moreover, rapid and transient changes of striatal M1 mAChR mRNA levels are observed after the single administration of muscarinergic agents. For example, the full agonist carbachol dose-dependently upregulated M1 mAChR mRNA expression, while the antagonist trihexylphenidyl dose-dependently downregulated M1 mAChR mRNA expression (Chou et al, 1993), suggesting that short-term treatment with muscarinic agents positively regulates M1 mAChR mRNA expression in the early stage after treatment, contrary to negative regulation in the long-term treatment. Our results may be, at least in part, in line with earlier findings (Bursa et al, 1995; Chou et al, 1993). In addition, we cannot rule out the possibility that McN possesses other unknown pharmacological mechanisms, which are independent of M1 mAChR-like effects (Wang et al, 2008).

ERK1/2 family members were originally identified by their responsiveness to growth factor receptor tyrosine kinases. However, they are also activated by many G-protein-coupled receptors (GPCRs). One family of GPCRs that activates ERK in several systems is the mAChR family (M1–M5), although limited information is available regarding how mAChR activates ERK in the brain or the receptor subtype responsible. One important function of ERK activation relates to the localization of nuclear downstream targets, such as CREB (Sgambato et al, 1998). Although each mAChR subtype can activate ERK in cell culture, M1 significantly activates ERK in CA1 pyramidal neurons of the hippocampus (Berkeley et al, 2001).

It was reported (Pepitoni et al, 1997) that regulatory elements of the M1 mAChR gene located at the promoter region lack a CRE sequence (TGACGTCA), but contain an activator protein-1 (AP-1) binding element sequence (TGACTCA). Since CREB and AP-1 can form a heterodimer to activate the transcription of target genes (Shaulian and Karin, 2001), CREB phosphorylation may be related to M1 receptor stimulation, as shown in our McN effects.

BDNF expression may be coupled to CREB-mediated transcription. Interestingly, we observed that CREB protein levels were decreased in neurons in the cerebral cortex and hippocampus of aged rats (Chung et al, 2002). Asanuma et al (1996) indicated that CREB DNA-binding activity is decreased in several different brain regions of aged rats compared with young adult rats. We cannot rule out the possibility that reduced BDNF levels in klotho mutant mice are due to impaired CREB-mediated transcription of the BDNF gene.

Klotho mutant mice have impaired visual recognition memory and associative fear memory at 7 weeks of age, while the animals have normal memory function at 6 weeks of age (Nagai et al, 2003). Performance in the light–dark shuttle box and elevated-plus maze box tasks is normal in klotho mutant mice (Nagai et al, 2003). There is also no apparent difference in the nociceptive threshold to electric foot shock in the conditioned-fear task (Nagai et al, 2003), suggesting that there are no major changes in exploratory activity, emotionality, or shock sensitivity in klotho mutant mice. Therefore, it is possible that the impaired performance of mutant mice is mainly due to memory. In addition, we speculate that neuropathologic changes secondary by klotho gene deficiency can negatively affect homeostatic regulation in the M1 mAChR system and related cognitive network (see Supplementary Discussion for more details).

In conclusion, we propose that underlying mechanisms (Supplementary Figures S6 and S7) accountings for memory impairment in klotho mutant mice are as follows: (1) The JAK2/STAT3 axis not only upregulated ChAT expression but also sensitized M1 mAChR and increased NMDAR-mediated LTP in the hippocampus of wild-type mice. (2) Klotho gene deficiency resulted in inactivation of the JAK2/STAT3 axis, dysfunction of the cholinergic neuronal system and decreased NMDAR-mediated LTP. (3) Desensitized M1 mAChR impaired PKCβII-, ERK/CREB-, and BDNF-dependent signaling. (4) McN, an M1 agonist, attenuated these impairments induced by klotho gene deficiency. Thus, it is possible that the klotho gene may be essential for maintaining cholinergic neuronal function in aging organisms, and that an M1 agonist may be useful in treating cognitive impairment associated with aging-related disorders.