Abstract
Aging is often associated with cognitive decline and recurrent cellular and molecular impairments. While life-long caloric restriction (CR) may delay age-related cognitive deterioration as well as the onset of neurologic disease, recent studies suggest that late-onset, short-term intermittent fasting (IF), may show comparable beneficial effects as those of life-long CR to improve brain health. We used a new optogenetic aging model to study the effects of late-onset (>18 months), short-term (four to six weeks) IF on age-related changes in GABAergic synaptic transmission, intracellular calcium (Ca2+) buffering, and cognitive status. We used male mice from a bacterial artificial chromosome (BAC) transgenic mouse line with stable expression of the channelrhodopsin-2 (ChR2) variant H134R [VGAT-ChR2(H134R)-EYFP] in a reduced synaptic preparation that allows for specific optogenetic light stimulation on GABAergic synaptic terminals across aging. We performed quantal analysis using the method of failures in this model and show that short-term IF reverses the age-related decrease in quantal content of GABAergic synapses. Likewise, short-term IF also reversed age-related changes in Ca2+ buffering and spontaneous GABAergic synaptic transmission in basal forebrain (BF) neurons of aged mice. Our findings suggest that late-onset short-term IF can reverse age-related physiological impairments in mouse BF neurons but that four weeks IF is not sufficient to reverse age-related cognitive decline.
SIGNIFICANCE STATEMENT Here, we demonstrate plasticity of the aging brain and reversal of well-defined hallmarks of brain aging using short-term intermittent fasting (IF) initiated later in life. Few therapeutics are currently available to treat age-related neurologic dysfunction although synaptic dysfunction occurs during aging and neurologic disease is a topic of intense research. Using a new reduced synaptic preparation and optogenetic stimulation we are able to study age-related synaptic mechanisms in greater detail. Several neurophysiological parameters including quantal content were altered during aging and were reversed with short-term IF. These methods can be used to identify potential therapies to reverse physiological dysfunction during aging.
Introduction
Age-related cognitive impairment is a substantial burden on society today. The number of individuals over the age of 65 is projected to rise to 88.5 million in 2050, which is more than double that from 2010 (Vincent and Velkoff, 2010). The estimated cost attributable to cognitive decline was $200 billion per year in the United States (Hurd et al., 2013) and $600 billion worldwide (Wimo et al., 2013). It is critical to understand the mechanisms of brain aging and to identify potential therapeutics.
Calorie restriction (CR), whether resulting from reduced calorie intake or intermittent fasting (IF), is a well-established paradigm to promote healthy brain aging and increase longevity (Witte et al., 2009; Van Cauwenberghe et al., 2016; Horie et al., 2016; Leclerc et al., 2020; Hwangbo et al., 2020; Green et al., 2022). Although improved brain health is well established, the effects of CR and IF on cognitive function are less well understood. Studies show short-term IF demonstrates variable or no improvement in cognitive function in humans (Gudden et al., 2021) and in rodents (Ingram and de Cabo, 2017). However, other studies demonstrate that fasting or a diet that mimics fasting enhances cognitive performance and health span in mice (Brandhorst et al., 2015) and improves hippocampal neurogenesis and enhances cognition (Dias et al., 2021).
A wide range of mechanisms are thought to contribute to the efficacies of CR or IF. Cellular/molecular pathways encompass: delayed oxidative stress and improved bioenergetics (Merry, 2004; Hunt et al., 2006; Mattson and Arumugam, 2018; Bayliak et al., 2021; Green et al., 2022), increased protein chaperone activity (Arumugam et al., 2010), reduced inflammation (Spaulding et al., 1997; Bhattacharya et al., 2006; Ugochukwu and Figgers, 2007), or by increased neurogenesis and upregulation of neurotrophic factors (Lee et al., 2002; Cheng et al., 2003; Joseph D'Ercole and Ye, 2008; Mattson, 2010; Vasconcelos et al., 2021; Dias et al., 2021). Interestingly, synaptic function was shown to improve following only days (Campanelli et al., 2021) or months (Dasgupta et al., 2018) of diet restriction or IF. Because many of these cellular mechanisms are ultimately reflected in changes in synaptic signals and/or calcium homeostasis, we focused on synaptic transmission and calcium signaling during IF.
We have been studying calcium (Ca2+) homeostasis, synaptic physiology and cognitive status during aging of rat basal forebrain (BF) neurons for many years (Griffith and Murchison, 1995; Murchison and Griffith, 1996, 1998, 2007; Murchison et al., 2009; Griffith et al., 2014). The BF plays an important role in cognitive functions such as attention, arousal, and learning and memory and projects to widespread throughout the brain (Záborszky et al., 2018). We have previously reported that enhanced Ca2+ buffering during aging is associated with age-related deficits in the frequency of spontaneous IPSCs (sIPSCs) in rat BF neurons, and that these changes were most prominent in cognitively impaired subjects (Murchison et al., 2009; Griffith et al., 2014). An increase in Ca2+ buffering was prevented in aged rats by life-long 40% reduced calorie intake (Murchison and Griffith, 2007).
We have extended these studies to mice using a bacterial artificial chromosome (BAC) transgenic line with stable expression of the channelrhodopsin-2 (ChR2) variant H134R (VGAT-ChR2(H134R)-EYFP) specific for GABAergic neurons (Zhao et al., 2011; Bang et al., 2021). We used acutely dissociated BF neurons and a reduced synaptic preparation (Griffith et al., 2014; Bang et al., 2021), to examine the effect of late-onset short-term IF diet (24-h alternate-day fasting for four to six weeks) on age-related changes in inhibitory synaptic transmission and Ca2+ buffering in behaviorally characterized mice. This optogenetic aging model enabled us to selectively stimulate GABAergic nerve terminals on voltage-clamped BF neurons, and calculate quantal content (m) of inhibitory neurotransmitter release. We report that age-related changes in Ca2+ homeostasis and inhibitory synaptic transmission are correlated with cognitive status. We show that this IF protocol can reverse age-related neurophysiological changes, but it is not sufficient to improve cognitive function.
Materials and Methods
Animals and treatments
Male wild-type (WT) C57 Bl/6 mice and VGAT-ChR2(H134R)-EYFP optogenetic mice of two ages, young (four to eight months), and aged (18–25 months) were used for all late-onset, short-term IF experiments. Breeding pairs were purchased from The Jackson Laboratory (stock #014548) and an aging colony was maintained for both genotypes as previously described (Bang et al., 2021). Mice were identified as WT or vesicular GABA transporter (VGAT) by standard tail-clip genotyping methods after weaning (EZ tissue/tail PCR genotyping kit, EZ BioResearch) as described previously (Bang et al., 2021). In aged animals, IF was initiated later in life. Calcium imaging experiments used WT mice, while VGAT mice were used for whole-cell patch-clamp recordings. Mice were randomly assigned to a control diet (ad libitum; AL) or an IF diet (alternate-day fasting) and diet groups were age-matched. Both AL and IF mice were singly housed from a week before the initiation of the IF diet. IF mice received the AL diet until the initiation of the IF diet and for 24 h on alternate days during the fasting regimen. After four weeks from the start of the diet, mice underwent two weeks of behavioral testing for a total of six weeks of diet treatment. Individual mice were then used daily for either calcium imaging or whole-cell patch-clamp experiments and this required approximately two weeks. During this time, mice were maintained on their respective diets. Mice were fed the NIH-31 chow which contained crude protein (18%), crude fat (5%), crude fiber (5%), minerals and vitamins. Water was available AL. The body weight and food consumption were recorded daily in both IF and control groups. Food intake was calculated on feeding days for IF mice. For calcium imaging in lifelong CR mice, we acquired male WT C57 Bl/6 mice of aged lifelong CR (aged CR, 21 months, N = 6), aged AL (aged control, 21 months, N = 6), and young AL (young control, 3 months, N = 6) from the National Institute of Aging (NIA) colony. Upon arrival at Texas A&M, we continued the same regimen until mice were killed for experiments, including at least three weeks of habituation. According to NIA protocol, both CR and control mice were individually housed. CR was initiated at the NIA colony at 14 weeks of age at 10% restriction, increased to 25% restriction at 15 weeks, and to 40% restriction at 16 weeks where it was maintained throughout the life of the animal. Animals were given NIH-31, the autoclavable form of a commonly used cereal, open formula feed if AL controls, or NIH-31 fortified with extra vitamin supplementation (1.67 times the vitamin mix) if CR. Water was available AL in individual water bottles in each cage of all mice.
All mice were maintained in the AAALAC-accredited vivarium at the Texas A&M. University Health Science Center under controlled conditions (22–25°C; lights 7 A.M. to 7 P.M.) in accordance with policies of the Texas A&M. University Laboratory Animal Care Committee and NIH guidelines.
Enzyme-treated, acutely dissociated neurons for calcium imaging
Mice were euthanized after isoflurane anesthetization and the brain quickly removed and sectioned at 320-μm thickness in a coronal plane. Slices were cut using a carbogen-equilibrated ice-cold cutting solution containing the following: 120 mm NaCl, 2.7 mm KCl, 0.5 mm CaCl2, 6 mm MgCl2, 5 mm HEPES, 26 mm NaHCO3, 11 mm D-glucose, and 0.3 mm kynurenic acid (pH 7.4 with NaOH, 310–330 mOsm). BF slices were then micro dissected to isolate the medial septum and nucleus of the diagonal band (MS/nDB) and enzymatically treated (trypsin, Sigma, Type XI or Type I), as described previously (Murchison and Griffith, 1996, 1998; Murchison et al., 2009). Enzyme-treated tissues were continuously suspended in a holding solution: 140 mm NaCl, 2.7 mm KCl, 0.5 mm CaCl2, 1.2 mm MgCl2, 10 mm HEPES, and 33 mm D-glucose (pH 7.4 with NaOH, 310–330 mOsm) by agitation of a small magnetic stir bar and were kept oxygenated with 100% O2. MS/nDB pieces were then gently triturated using a series of increasingly smaller diameter fire-polished glass pipettes. The dissociated cells were plated on the glass floor of a recording chamber containing bath solution: 140 mm NaCl, 2.7 mm KCl, 2 mm CaCl2, 1.2 mm MgCl2, 10 mm HEPES, and 33 mm D-glucose, then perfused at a rate of ∼2 ml/min. Experiments were performed at room temperature within 5 h of dispersal.
Calcium imaging
In order to measure intracellular Ca2+ buffering in mouse BF neurons, we used a dual wavelength ratiometric fura-2 microfluorimetry system (Murchison and Griffith, 1998, 2007; Murchison et al., 2009). Enzyme-treated, acutely dissociated mouse BF neurons were loaded with ratiometric fluorescent Ca2+-indicator fura-2 AM (1 μm, Invitrogen) for 12 min in the presence of 10 μm pluronic acid (Invitrogen), after which they were washed for 40–45 min to allow for fura de-esterification. The fura-2 fluorescence was excited using alternating 340- and 380-nm wavelengths and the resulting 340/380 intensity ratio was determined. The intracellular calcium concentrations ([Ca2+]i) were estimated using in vitro calibration as described previously (Murchison and Griffith, 1998). The Ca2+-transients (Δ[Ca2+]i) were triggered by pressure application of elevated [K+] (20 mm) solution (high [K+]), using a focal picospritzer (Picospritzer II, General Valve). Application of different stimulus durations (0.2–1.2 s) resulted in increasing amplitudes of Δ[Ca2+]i. The amplitude of the transients was then plotted against the stimulus intensity to calculate a buffering slope. Unlike the traditional buffering curve generated by voltage-clamp stimulation (Murchison et al., 2009), this less invasive technique produces plots of transient amplitude versus stimulus intensity calculated in arbitrary units (stimulus duration × 10) with a greater slope value indicating less intracellular Ca2+ buffering. Fluorescent signals were recorded using a Sutter λ DG-4 excitation switcher (Sutter Instruments), a Hamamatsu ORCA flash2.8 CCD camera (Hamamatsu Photonics), an Olympus IX70 fluorescent scope and Metafluor imaging software (Molecular Devices). The resting [Ca2+]i was determined before any stimulation of the cell.
Reduced synaptic preparation using enzyme-free, acutely dissociated neurons to study synaptic transmission across aging
We used a reduced synaptic preparation to examine synaptic transmission in acutely dissociated BF neurons across aging (Griffith et al., 2014; Bang et al., 2021). In this preparation, there is a very low frequency of spontaneous glutamatergic EPSCs which show rapid kinetics with decay time constant (τd = ∼5 ms). In contrast, much slower GABAergic sIPSCs are easily identifiable with decay constants of τd = 20–30 ms. With CsCl in the recording pipette, the IPSCs have a reversal potential of 0 mV and are often larger than the faster, occasional sEPSCs. No excitatory amino acid antagonists were used. All recordings were visually inspected and analyzed. Slices from the MS/nDB region (320 μm) were cut and neurons were acutely dissociated as described above, but without enzyme pretreatment. These neurons retain functional presynaptic boutons (Fig. 1).
Electrophysiology
Standard whole-cell patch voltage-clamp recording was used for measuring sIPSCs and optically-evoked IPSCs (oIPSCs). Recordings for sIPSCs and oIPSCs were made from a holding potential of −60 mV, low-pass filtered at 2 kHz and digitized at 10 kHz with Digidata 1440A interface, Multiclamp 700B amplifier (Molecular Devices), and pClamp 10 software (Molecular Devices). Membrane capacitance (pF) was obtained by digital cancellation of whole-cell capacitance transients. Recording electrodes were pulled from glass capillary tubing (KG-33, 1.5 mm OD, King Precision Glass) on a Flaming/Brown P-87 pipette puller (Sutter Instruments) with final electrode resistance 4–8 MΩ when filled with a pipette solution. The internal pipette solution for whole-cell recordings contained the following: 130 mm CsCl, 10 mm EGTA, 2 mm MgCl2, 10 mm HEPES, 4 mm Mg-ATP, and 0.1 mm GTP (pH 7.2 with CsOH; 295–300 mOsm).
Mini Analysis (6.0.7; Synaptosoft) and Clampfit 10 (Molecular Devices) software were employed for the analysis and we visually identified and measured sIPSCs. The optically-evoked response was calculated as the peak current over 20 ms from the time when response begins. Mean data were compared using a two-way ANOVA followed by Fisher's LSD test (GraphPad Prism 9, GraphPad software). Cumulative probability curves were compared using a Kruskal–Wallis test.
Optically-induced current recordings and quantal content analysis
Optically-induced IPSCs (oIPSCs) were recorded using whole-cell clamp from a holding potential of −60 mV. Enzyme-free, acutely dissociated neurons from VGAT mice were stimulated with 5-ms duration light stimulation (470 nm) at 15-s intervals using an optogenetic illumination system (DC2100, Thor Labs). Power density was calculated by dividing the measured power using a PM200 optical power and energy meter (ThorLabs) connected to an S120C photodiode sensor (ThorLabs) by the illuminated area. Initial measurements using a 400-μm cannula revealed a maximum power density of ∼10 mW/mm2, and a power density of 1.76 mW/mm2 corresponding to 100-mA drive current. LED intensity was controlled by adjusting the drive current (1 mA to 1 A). For the quantal content analysis, 50–100 consecutive traces of oIPSCs were recorded, and minimal light stimulation was used to ensure a sufficient number of failures. Using the method of failures [m = ln(N/N0)] in the Poisson model, quantal content (m) was calculated where, N is the total number of stimuli, and N0 is the number of failures where stimulation did not evoke a synaptic response (Bang et al., 2021).
Barnes maze (BM)
Cognitive function was assessed using the BM spatial learning task (Barnes, 1979; Bizon et al., 2012; Pitts, 2018). All food was provided and removed at 10 A.M. each day. BM testing started at 12 P.M., therefore on fasting days, IF mice had food removed at 10 A.M. but food was available until 2 h before testing. This experimental design was used to avoid hunger and foraging behavior during the behavior test. The apparatus of BM is a circular platform (92 cm in diameter) with 20 circular holes (5.1 cm in diameter) evenly spaced around the periphery and an escape box is placed under one of the holes. The platform is located 92 cm above the floor to discourage mice from jumping down. Four bright floodlights were affixed above the maze as a negative reinforcement to motivate the mice to search for the escape chamber using spatial cues which surround the Barnes table. Video was acquired using an EQ610 Polestar II Everfocus camera and data were quantified using Ethovision XT version 15 video tracking software (Noldus). Mice were habituated by allowing each mouse to freely explore the maze without an escape box under modest light for 5 min to become accustomed to the room, table, and extra- and intra-maze layout. On the pretraining day which is 48 h after habituation, mice learned the presence of the escape hole by exploring the maze under bright light for 3 min and then being guided gently toward the escape hole positioned randomly on the maze and allowing the mouse to enter the hole and remain inside for 1 min. Acquisition training began 24 h after the conclusion of pretraining and started with mice in a 2-l transparent glass beaker in the center of the table. A trial started when the beaker was lifted, and mice were allowed 3 min to locate and enter the escape box. The trial session ended when the mouse entered the escape hole or after 3 min had elapsed. If the mouse was unable to find the escape box, it was gently guided and allowed to remain there for 1 min. All mice received four trials per day for 4 d. 72 h after the last acquisition training, reversal learning trials were performed in which the escape chamber position was changed 180° from its previous location in the acquisition training. Similar to the acquisition training, all mice were given four trials per day for 4 d. The platform and escape box were cleaned with 70% ethanol and water to prevent odor cues between subsequent trials. Distance traveled to the escape hole and search strategies used per trial during acquisition and reversal training were measured from the recorded video. Search strategies were determined by examining each mouse's daily session and separating them into six main categories (Illouz et al., 2016): (1) direct (spatial), navigating directly to target hole; (2) corrected (spatial), making a minor correction (±1 hole away from the target hole); (3) long-correction (spatial), re-angulating from a distal region of environment and making a direct movement toward the target; (4) focused search (spatial), searching the region surrounding the target (±3 holes away from the target hole); (5) serial search (nonspatial), systematic sequential hole searches; and (6) random search (nonspatial), localized searches of holes separated by paths crossing the maze center. Strategies during each trial session were scored with a numerical value to quantify a cognitive score for each animal: direct = 1, corrected = 0.75, focused search = 0.5, long-correction = 0.5, serial = 0.25, random = 0. Failure of a mouse to find the target hole was also assigned a score of “0”.
Statistical analysis
Comparisons were performed using a two-way or repeated measures three-way ANOVA with Fisher's LSD post hoc test using GraphPad Prism 9 (GraphPad software). For cumulative probability graph in Figure 5B, Kruskal–Wallis test was used with Dunn's multiple comparison post hoc test. Values are expressed as mean ± SEM. Differences were considered significant at p < 0.05.
Results
Optogenetic mouse model and reduced synaptic preparation
Experiments were conducted using WT C57/Bl6 mice and VGAT-ChR2(H134R)-EYFP optogenetic mice on a C57/Bl6 background. The expression of the channelrhodopsin-EYFP construct is under control of the VGAT promotor, and is only expressed in GABAergic neurons (Fig. 1A). This permits specific excitation of GABAergic neurons by 470-nm light stimulation and oIPSCs can be recorded in postsynaptic targets, as shown in Figure 1B. For our experiments we employed a reduced synaptic preparation that consists of acutely dissociated BF neurons that receive no enzyme treatment and presynaptic terminals remain attached. In this preparation, GABAergic neurons can be identified by their bright EYFP fluorescence, while non-GABAergic neurons (presumably cholinergic) feature fluorescent puncta that are putatively GABAergic terminals (Fig. 1C).
Reduced synaptic preparation in VGAT-ChR2-EYFP BAC optogenetic mice allows stimulation of light-evoked IPSCs (oIPSCs) A, Schematic diagram of the transgenic construct present in the GABAergic neurons of these mice. B, Examples of oIPSCs (Vh = –60 mV). Top records show superimposed traces with individual responses and failures below. Minimal intensity light stimulation of 5-ms duration (470 nm) is indicated at blue arrow. A stylized diagram of a dissociated neuron with blue light excitation is shown also. Presynaptic GABAergic terminals are depicted in red to illustrate how they can be activated with light. C, Confocal DIC and fluorescent images of GABAergic and non-GABAergic acutely dissociated neurons in the reduced synaptic preparation. Top panels show a DIC image of a live (non-GABAergic) neuron and colored arrows point out the position of presumed presynaptic terminals. The EYFP fluorescence image to the right shows more clearly the location of the fluorescent puncta that are thought to be GABAergic synaptic terminals (arrows). In this preparation non-GABAergic neurons are likely cholinergic and display only very dim autofluorescence. In contrast below, live GABAergic neurons are seen as DIC images (left) and displaying bright EYFP fluorescence (right). Note that in this single confocal plane scan, ChR2-EYFP fluorescence is predominately located around the cell surface where functioning ChR2 channels are located.
Changes in body weight and food consumption by IF diet
Young and aged mice were maintained on AL diets before the start of the experimental protocols. Control subjects continued on the AL diet for the duration of the protocols, while mice on IF diets commenced four weeks before BM training and continued during the remaining protocol (six weeks total), as shown in Figure 2A. Therefore, aged cohorts received late-onset IF that was started four weeks before the start of behavioral testing and six weeks before they were used for physiological studies. Body weight and food consumption were monitored throughout the experiment. Aged mice on IF diet gradually decreased their body weight during the six-week IF and showed 7% reduction in body weight after the six-week IF period compared with their initial body weight, while aged control maintained their body weight throughout the six-week diet period. Young control mice showed a gradual increase of their body weight during the six-week period, resulting in a 3% increase of body weight. The young IF group also increased body weight during the first two weeks of the IF diets, however, their body weights eventually began to decrease and were reduced by 1.4% at the end of six weeks. This change in body weight was significantly different for the aged IF diet group (Fig. 2B). Both young and aged mice placed on IF diet were observed to binge on feeding day, and their daily food intake on feeding day was 67% and 63% higher than young and aged control, respectively (Fig. 2C, left). Despite binge eating behavior, total food consumption during the six-week diet period was reduced by 14% and 17% in young and aged IF mice compared with respective controls (Fig. 2C, right).
The effect of IF on body weight and food consumption. A, Overview of the timeline and experimental design of the IF experiments. All mice were initially fed AL then maintained for six weeks on either an IF diet or continued on the AL (control) diet. To assess cognitive function, all mouse groups (young control and IF, aged control, and IF) were trained in the BM task starting four weeks after the implementation of the IF diets. BM training continued for two weeks before mice were used to measure either calcium buffering or synaptic transmission. Diets were maintained until killing. B, Plot of percent (%) change in body weights of control and IF mice in both young and aged during the six-week diet period [weeks: F(3.084,163.4) = 10.29, p < 0.0001, age: F(1,53) = 13.53, p = 0.0005, diet: F(1,53) = 5.070, p = 0.0285, interaction (weeks × age): F(6,318) = 4.701, p = 0.0001, interaction (weeks × diet): F(6,318) = 7.810, p < 0.0001, interaction (age × diet): F(1,53) = 3.109, p = 0.0836, interaction (weeks × age × diet): F(6,318) = 1.070, p = 0.3806, N = 12–16 per group]. For IF mice, the body weights were measured after feeding days. C, Graphs of average of daily food intake per mouse (left, interaction: F(1,53) = 0.09,763, p = 0.7559; age: F(1,53) = 11.47, p = 0.0013, diet: F(1,53) = 395.4, p < 0.0001, N = 12–16 per group) and average of total food consumption per mouse during the six-week diet period (right, interaction: F(1,53) = 0.45, p = 0.5052; age: F(1,53) = 7.986, p = 0.0066; diet: F(1,53) = 29.72, p < 0.0001, N = 12–16 per group). All data are expressed as mean ± SEM and are pooled from the WT and VGAT cohorts. The data in panel B were analyzed by a repeated-measures three-way ANOVA with Fisher's LSD post hoc test. The data included in panel C were analyzed by a two-way ANOVA with Fisher's LSD post hoc test; *p < 0.05. Values of N indicate number of mice.
If reverses age-related alterations in calcium buffering in WT mouse BF neurons
In order to determine the functional consequences of a six-week IF diet, we started by testing the previously described age-related enhancement of rapid intracellular Ca2+ buffering that occurs in rat BF neurons and is associated with age-related cognitive impairment (Murchison et al., 2009) and reversed by life-long caloric restriction (CR; Murchison and Griffith, 2007). We then hypothesized that age-related increases in rapid Ca2+ buffering can be reversed by late-onset short-term IF diet. We used a simplified method (Bang et al., 2021) to assess the relative Ca2+ buffering of the young and aged mice in each diet group. Acutely dissociated BF neurons from behaviorally characterized (BM) WT mice were loaded with the ratiometric calcium-sensitive fluorescent probe fura-2 AM and Ca2+ transients were evoked by focal picospritzer application of elevated [K+] (20 mm) solution (high [K+]), as shown in Figure 3A. Representative Ca2+ transients from visually identified neurons of each diet group are shown in Figure 3B. Transients of increasing amplitude were generated by stimuli of increasing duration. The amplitude of the transients (peak Δ[Ca2+]i – baseline [Ca2+]i) was plotted against the stimulus intensity (stimulus duration × 10) to calculate buffering slopes (Fig. 3C). The stimulus intensities in these plots are in arbitrary units and a greater slope value is indicative of lower rapid Ca2+ buffering. The calculated slopes for all neurons from each group were plotted in Figure 3D, left, and the slope values averaged from individual animals is shown at right. Consistent with our earlier findings, Ca2+ buffering in aged control is significantly greater (lower slope) compared with young control. However, Ca2+ buffering plots in aged late onset IF showed significantly higher slope when compared with aged control but was not different from young control and IF. This demonstrates that increased calcium buffering during aging was reversed by late-onset short-term IF diet. There was no significant difference in calcium buffering values between young control and young IF. In addition, we did not observe any differences in somatic baseline [Ca2+]i between groups both in neurons and animals: young control: 88.18 ± 3.85 nm, young IF: 83.74 ± 5.15 nm, aged control: 89.84 ± 3.95 nm, aged IF: 84.83 ± 4.83 nm, interaction, F(1,116) = 0.004037, p = 0.9494, age, F(1,116) = 0.0928, p = 0.7612, diet, F(1,116) = 1.092, p = 0.2983, ANOVA, n = 19–38 per group. Finally, we replicated the reversal of the age-related increase in Ca2+ buffering in life-long CR mice (Fig. 4). These mice were obtained from the NIA aging colony and were maintained on a 60% of control diet starting in adolescence as described in Materials and Methods.
Increased Ca2+ buffering in WT mouse BF neurons during aging is reversed by late-onset short-term IF. A, DIC image of an acutely dissociated mouse BF neuron with an illustration of the focal picospritzer method for application of elevated [K+] (20 mm) solution (high [K+]). The red circle depicts the region of interest from which the fura-2 fluorescent intensity data are collected. B, Representative traces of ratiometric fura-2 Ca2+-sensitive fluorescence in BF neurons from young and aged control and IF mice. Ca2+-transients were generated by depolarizing the neurons with picospritzer applications of high [K+] with increasing durations (red arrows). Note that smaller transients in aged neurons indicate increased [Ca2+]i buffering. Estimated [Ca2+]i was determined using standard in vitro calibration (see Materials and Methods). C, Summary data of buffering slope plots for BF neurons from each mouse group (n = 19–38 per group). The stimulus duration was multiplied by 10 to avoid fractional slopes. Greater slopes indicate lower calcium buffering. The dashed line depicts the aged IF data. D, Bar graphs and scatter plots show the mean buffering slopes in both young and aged control and IF diet groups (left panel from individual neurons, interaction, F(1,116) = 4.216, p = 0.0423, n = 19–38 per group; right panel by animals, interaction, F(1,21) = 6, p = 0.0232, N = 4–8 per group). All data are mean ± SEM. The data in panel D were analyzed by a two-way ANOVA with Fisher's LSD post hoc test; *p < 0.05. Values of n indicate number of neurons and values of N indicate number of mice.
Increased Ca2+ buffering during aging is prevented by life-long CR. Bar graphs and scatter plots show mean (±SEM) buffering slopes in young control, aged control, and aged CR groups (young control: 49.49 ± 3.92, aged control 34.43 ± 5.58, aged CR 78.45 ± 10.42, n = 17–20 per group, young control vs aged control: p = 0.0304, young control vs aged CR: p = 0.0132; aged control vs aged CR: p = 0.0012, unpaired t test). *p<0.05.
Age-related decrease in frequency of sIPSCs is reversed by late-onset IF in VGAT optogenetic mice
Our laboratory has reported that the frequency of sIPSCs in BF neurons is reduced with age both in rats (Griffith et al., 2014) and mice (Bang et al., 2021) and that this reduction in sIPSC frequency is associated with cognitive impairment in rats (Griffith et al., 2014). We further proposed that increased intracellular calcium buffering contributes to cognitive impairment by affecting synaptic function during aging (Murchison et al., 2009; Griffith et al., 2014). In order to examine whether late-onset short-term IF can reverse the age-related decrease in sIPSC frequency, we used standard whole-cell voltage-clamp (Vh = −60 mV) in the reduced synaptic preparation and measured sIPSCs in BF neurons from VGAT mice that had been behaviorally characterized by BM training. Figure 5 shows that late-onset IF reverses the decline in frequency of sIPSCs observed during aging with no significant effect on amplitude. Example currents from each test group are shown in Figure 5A, while Figure 5B shows cumulative probability plots of sIPSC interevent intervals (n = 17–23 neurons per group; *p < 0.0001, Kruskal–Wallis test) along with reversal of the age-related increases in interevent duration by late-onset IF. Figure 5C shows scatter plots and mean ± SEM of sIPSC frequency (left panel from individual neurons, interaction, F(1,76) = 4.484, p = 0.0375, n = 17–23 per group; right panel by animals, interaction, F(1,28) = 4.645, p = 0.0399, N = 8 per group). The sIPSC frequencies of aged late-onset IF diet mice were not significantly different from that of young mice of either diet regimen, while showing a significantly higher sIPSC frequency than aged controls. Figure 5D shows scatter plots with mean ± SEM of sIPSC amplitude (left panel from individual neurons, n = 17–23 per group; right panel by animals, N = 8 per group). There were no significant differences in sIPSC amplitudes in either neurons or animals regardless of age and diet. The data included in Figure 5C,D were analyzed by a two-way ANOVA with Fisher's LSD post hoc test. Our results demonstrate that late-onset short-term IF can reverse age-related reductions in sIPSC frequency in the mouse BF.
The effect of late-onset short-term IF on frequency and amplitude of sIPSCs in BF neurons of VGAT optogenetic mice in the reduced synaptic preparation. A, Representative whole-cell voltage-clamp recordings (Vh = –60 mV) from an acutely dissociated BF neuron in each group of mice. B, Cumulative probability plots of sIPSC interevent intervals (n = 17–23 neurons per group; *p < 0.0001, Kruskal–Wallis test). C, Bar graphs and scatter plots show mean ± SEM of sIPSC frequency (left panel from individual neurons, interaction, F(1,76) = 4.484, p = 0.0375, n = 17–23 per group; right panel by animals, interaction, F(1,28) = 4.645, p = 0.0399, N = 8 per group). D, Bar graphs and scatter plots with mean ± SEM of sIPSC amplitude (left panel from individual neurons, interaction, F(1,76) = 0.8325, p = 0.3644; age, F(1,76) = 0.4848, p = 0.4884; diet, F(1,76) = 1.061, p = 0.3063, n = 17–23 per group; right panel by animals, interaction, F(1,28) = 0.2625, p = 0.6124; age, F(1,28) = 0.983, p = 0.3299; diet, F(1,28) = 1.634, p = 0.2116, N = 8 per group). There were no significant differences in sIPSC amplitudes in either neurons (left) or animals (right) regardless of age and diet. The data included in panels C, D are expressed as mean ± SEM and were analyzed by a two-way ANOVA with Fisher's LSD post hoc test; *p < 0.05. Values of n indicate number of neurons, and values of N indicate number of mice.
Late-onset short-term IF restores age-related changes in quantal content of VGAT optogenetic mice BF neurons
Changes in the frequency of synaptic transmission are often mediated by changes in the probability of neurotransmitter release at the presynaptic sites. Therefore, we next examined whether there are observable presynaptic changes in inhibitory synaptic mechanisms during aging and whether these alterations can be restored by late-onset short-term IF diet. We used whole-cell voltage-clamp (Vh = −60 mV) to record oIPSCs in the reduced synaptic preparation of BF neurons from behaviorally characterized VGAT mice. By applying 5-ms minimal intensity light stimulation, quantal content (m) was calculated using the method of failures (see Materials and Methods). Figure 6A shows representative traces of oIPSCs from each test group with traces superimposed (top) and distributed (below). Note the failures (lack of response) visible in the individual traces. Figure 6B shows scatter plots and mean quantal content (m) data calculated by the method of failures for both young and aged diet groups. The data on the left is from individual neurons (interaction, F(1,51) = 4.085, p = 0.0485, n = 9–19 per group) and data on the right is averaged by individual animals (interaction, F(1,24) = 1.947, p = 0.1757, age, F(1,24) = 2.42, p = 0.1329, diet, F(1,24) = 2.871, p = 0.1031, N = 6–8 per group). The m values are significantly reduced in the BF neurons of aged control as compared with young control and late-onset IF, suggesting that there is an age-related decrease in quantal content at GABAergic synapses in mouse BF neurons and that this decrease is reversed by late-onset IF. The data averaged by animals (Fig. 6B, right) did not show statistical significance of either interaction between aging and diet or main effects for aging and diet on quantal content, possibly because of the small number of animals and the large scatter. None-the-less, the animal data very much resemble the individual neuron data. Overall, these results suggest that late-onset short-term IF diet can reverse the age-related reduction in presynaptic quantal content (m) of BF neurons. We observed no significant changes in the averaged amplitude of oIPSCs (Fig. 6C). The data in Figure 6B,C were analyzed by a two-way ANOVA with Fisher's LSD post hoc test; *p < 0.05.
Late-onset short-term IF reverses an age-related reduction in quantal content of inhibitory synaptic transmission in BF neurons of VGAT optogenetic mice in the reduced synaptic preparation. A, Representative whole-cell voltage-clamp recordings (Vh = −60 mV) of oIPSCs superimposed (top) and distributed (below). Blue arrow indicates minimal intensity light stimulation (470 nm, 5 ms). B, Graphs showing mean quantal content (m) calculated by the method of failures for both young and aged diet groups. The data on the left are from individual neurons (interaction, F(1,51) = 4.085, p = 0.0485, n = 9–19 per group), and data on the right are averaged by individual animals (interaction, F(1,24) = 1.947, p = 0.1757, age, F(1,24) = 2.42, p = 0.1329, diet, F(1,24) = 2.871, p = 0.1031, N = 6–8 per group). C, Amplitude data of similar bar graphs for averaged oIPSCs show no significant differences for age or diet (left panel from individual neurons, interaction, F(1,51) = 1.611, p = 0.2101; age, F(1,51) = 0.2256, p = 0.6369; diet, F(1,51) = 1.008, p = 0.3202, n = 9–19 per group; right panel by animals, interaction, F(1,24) = 1.165, p = 0.2912; age, F(1,24) = 1.018, p = 0.3230; diet, F(1,24) = 0.519, p = 0.4782, N = 6–8 per group). All data are mean ± SEM. The data in B, C were analyzed by a two-way ANOVA with Fisher's LSD post hoc test; *p < 0.05. Values of n indicate number of neurons, and values of N indicate number of mice.
Late-onset short-term IF did not reverse age-associated decline in cognitive function
After four weeks on the IF diets, BM training was performed to assess cognitive function in all mice (young and aged, VGAT and WT). In order to familiarize mice with the room and maze table, as well as to decrease their anxiety which may affect their behavior, all the mouse subjects went through habituation and pretraining before the acquisition and reversal training sessions (Fig. 7A). The BM diagrams for acquisition and reversal training are shown in Figure 7B. Note that the location of the escape hole in reversal training is 180° opposite of the acquisition location. During the acquisition and reversal training days, distance traveled (cm) was measured and analyzed with repeated measurements ANOVA. In the acquisition training (Fig. 7C, left), all four tested groups showed a decrease in the distance traveled to reach the escape hole from the start position (maze center) between days 1 and 4. However, the aged controls took significantly longer path-lengths than young controls. Although both IF groups displayed path-lengths generally intermediate between the young and aged controls, the aged IF was not different from aged control. Both aged control and aged IF showed delayed learning but eventually acquired the task such that there were no significant differences in distance traveled between the groups by day 4. Interestingly, young IF showed significantly longer distance on day 1 and day 2 when compared with young control, suggesting that the IF regimen in this study may be stressful and have some detrimental effect on the cognitive function in young animals. On the other hand, in the reversal training where animals are required to learn the new location of the escape box (Fig. 7C, right), there was no difference in the performance of young control and young IF. Aged controls took longer paths than young controls or young IF during reversal training, and this deficit was only marginally improved by IF diet. These data suggest that aged mice showed delayed learning and less cognitive flexibility as compared with young mice and that the IF regimen in the present study was not sufficient to fully reverse age-related cognitive impairment. Graphs in Figure 7C show distance traveled per day to the escape hole in acquisition training (left, day, F(2.211,117.2) = 69.14, p < 0.0001; age, F(1,53) = 8.281, p = 0.0058; diet, F(1,53) = 0.9082, p = 0.3449; no interaction between day and age, day and diet, age and diet, or day, age and diet) and reversal training (right, day, F(2.376,125.9) = 7.445, p = 0.0004; age, F(1,53) = 9.125, p = 0.0039; diet, F(1,53) = 0.3886, p = 0.5357; no interaction between day and age, day and diet, age and diet, or day, age and diet). All data are mean ± SEM and are pooled from WT and VGAT cohorts. The data included in Figure 7C were analyzed by a repeated measure three-way ANOVA with Fisher's LSD post hoc test. Significance (p < 0.05) is shown in the figure with multiple comparisons indicated by letters a–f. Values of N indicate number of mice.
Age-related learning delay is not reversed by late-onset short-term IF. A, Diagram of the experimental timeline for BM training is shown. Numerals represent days of training. B, Schematic representation of BM is depicted. Acquisition training (T) is shown on the left and reversal training is shown on the right. Note that escape hole position in the reversal training is changed 180° from its previous location in the acquisition training. The nonescape holes lack the escape compartment. C, Graphs are shown of distance traveled per day to the escape hole in acquisition training (left, day, F(2.211,117.2) = 69.14, p < 0.0001; age, F(1,53) = 8.281, p = 0.0058; diet, F(1,53) = 0.9082, p = 0.3449; no interaction between day and age, day and diet, age and diet, or day, age and diet) and reversal training (right, day, F(2.376,125.9) = 7.445, p = 0.0004; age, F(1,53) = 9.125, p = 0.0039; diet, F(1,53) = 0.3886, p = 0.5357; no interaction between day and age, day and diet, age and diet, or day, age and diet). All data are mean ± SEM and are pooled from WT and VGAT cohorts. The data included in panel C were analyzed by a repeated measure three-way ANOVA with Fisher's LSD post hoc test. Significance (p < 0.05) is shown in the figure with multiple comparisons indicated by letters a–f. Values of N indicate number of mice.
We also investigated possible age-related differences in search strategy to navigate to the escape hole. Search strategies are defined in the Materials and Methods, but basically, different search strategies were divided into two categories: spatial and nonspatial, as shown in Figure 8. During the acquisition training, all four groups gradually used a more spatial strategy across the training days, although both aged diet groups relied less frequently on spatial strategies than young groups throughout the training (Fig. 9A, left). The pattern of search strategies was very similar between the aged IF and aged control. Young IF showed a delay in acquiring spatial search strategies relative to young control and continued to show a slight relative deficit throughout acquisition training. During the reversal training (Fig. 9A, right), similar to the acquisition phase, all four groups displayed an increased use of a spatial strategy between days 1 and 4, although the aged control and aged IF did not achieve the levels of spatial strategies observed in the young diet groups. Both aged control and aged IF showed more random searches and failures throughout reversal training. Young IF displayed a slight delay in switching to a predominately spatial strategy relative to young control. These results show that the four-week IF diet does not appear to improve cognitive performance in aged mice.
Spatial search strategies in mouse BM test. Examples shown range from highly spatial (top) to nonspatial and failures. Spatial strategies are shown in the box and include direct, corrected, focused search, and long correction. Nonspatial strategies include serial search, random search, and failure. Each strategy was scaled to quantitate cognitive score and values are shown in parentheses. Higher total cognitive scores indicate greater use of spatial search strategies. The focused search strategy is distinguished from the serial search when the mouse does not visit holes on the opposite side from the escape hole.
Late-onset short-term IF did not improve an age-related reduction in the use of spatial search strategies. A, The percentage of search strategy for each mouse group in the BM task was recorded during the acquisition training (left) and the reversal training (right). The percentages for the spatial search strategies (blue hues) are pooled. B, Line graphs show cognitive scores for each mouse group per day during acquisition training (left, day, F(2.953,156.5) = 62.91, p < 0.0001; age, F(1,53) = 13.20, p = 0.0006; diet, F(1,53) = 0.1104, p = 0.7410; no interaction between day and age, day and diet, age and diet, or day, age and diet) and during reversal training (right, day, F(2.542,134.7) = 5.256, p = 0.0032; age, F(1,53) = 3.655, p = 0.0613; diet, F(1,53) = 0.02,204, p = 0.8826; no interaction between day and age, day and diet, age and diet, or day, age and diet). Values are mean ± SEM and are pooled from WT and VGAT cohorts. The data in panel B were analyzed by a repeated measure three-way ANOVA with Fisher's LSD post hoc test. Significance (p < 0.05) is indicated by letters a–f in the figure. Values of N indicate number of mice.
To quantify the cognitive capacity of individual mice, a cognitive score was calculated based on the data from Figure 9A and the cognitive score numerical values shown in Figure 8 (further details in Materials and Methods). All four tested groups showed an increase in their cognitive score between days 1 and 4 of acquisition and reversal training (Fig. 9B). Young controls showed significantly higher cognitive scores than aged control and/or aged IF on the day 1, 2, and 4 in acquisition and on day 1 of reversal training, with young IF showing intermediate scores (Fig. 9B), indicating age-related deficits in spatial learning. In addition, aged IF showed higher cognitive score on most of days in acquisition and reversal training when compared with aged control, but it was not statistically significant, suggesting that short-term IF diet regimen on aged animals is not sufficient to reverse age-related cognitive impairment. In contrast, by day 4 in the reversal training, there were no statistically significant differences in cognitive score between all tested groups (Fig. 9B, right). The legend for Figure 9 contains the summarized statistical analysis.
Correlation between cognitive behavioral scores and age-related changes in Ca2+ buffering and inhibitory synaptic transmission
Previously, we reported that age-related changes in calcium buffering and in sIPSC frequency are associated with cognitive impairment during aging in rats (Murchison et al., 2009; Griffith et al., 2014). We therefore examined the relationship between three cellular parameters that we measured in the present study (intracellular Ca2+ buffering slope, sIPSC frequency and quantal content) and cognitive function (cognitive score) across aging. Only control animals of both ages were used in this analysis to eliminate the effects of the IF diet. In addition, we used the cognitive score from the day 1 of acquisition and reversal training which showed the most significant age-related differences in cognitive function. Figure 10 shows the correlation plots for both acquisition (top graphs) and reversal training (bottom graphs). Each point on the graphs represents an individual animal. The intracellular Ca2+ buffering slope values were positively correlated with cognitive scores, suggesting higher slopes (less buffering) is beneficial for cognitive function (Fig. 10A, top, acquisition, r = 0.5355, p = 0.0346; bottom, reversal, r = 0.7861, p = 0.0005). Likewise, Figure 10B shows a positive correlation between sIPSC frequency and cognitive score (Fig. 10B, top, acquisition, r = 0.6468, p = 0.0083; Fig. 10B, bottom, reversal, r = 0.6900, p = 0.004). No correlation between quantal content and cognitive score was found (Fig. 10C, top, acquisition, r = 0.1643, p = 0.5397; Fig. 10C, bottom, reversal, r = 0.4821, p = 0.0608). When all animals (control and IF diet) were included in the analysis, there were no correlations (data not shown). This is predictable in that the aged IF diet reversed the changes in the neurophysiological parameters, but not the cognitive score deficits. Consistent with the previous rat data, we observed significant positive correlation between calcium buffering slope/sIPSC frequency and cognitive score. It is unclear why quantal content (m) from GABAergic synapses failed to correlate significantly with cognitive score (Fig. 10C), other than that the aged mice had similar m values regardless of cognitive score. Future studies are required to test whether IF may affect different releasable pools of neurotransmitter as an explanation for the differences between sIPSC frequency and evoked quantal release. Overall, our results suggest that changes in intracellular Ca2+ buffering and sIPSC frequency in the mouse BF are relevant to age-related changes in cognitive function.
Cognitive scores in the BM task correlate with physiological measures of calcium buffering and synaptic transmission. A, Intracellular calcium buffering slopes show a significant correlation with cognitive scores, such that greater slopes (lower buffering) are associated with higher cognitive scores (top, acquisition, r = 0.5355, p = 0.0346; bottom, reversal, r = 0.7861, p = 0.0005). B, Scatter plots show that sIPSC frequency is significantly positively correlated with cognitive score (top, acquisition, r = 0.6468, p = 0.0083; bottom, reversal, r = 0.6900, p = 0.0040). C, Scatter plots show that quantal content of inhibitory synaptic transmission is not significantly correlated with cognitive score (top, acquisition, r = 0.1643, p = 0.5397; bottom, reversal, r = 0.4821, p = 0.0608). All data were analyzed by Spearman's correlation with significance p < 0.05. Values of N indicate number of mice.
Discussion
The substantial burden of physiological and cognitive impairments associated with aging has led to an increased focus on potential therapeutics and lifestyle changes to alleviate age-related deterioration. CR has shown significant promise to increase longevity and improve health span. Clinical studies show benefits of CR and IF for numerous neurologic diseases including epilepsy, Alzheimer's disease, and multiple sclerosis on disease symptoms and progress (Gudden et al., 2021) and age-associated physiological decline (Green et al., 2022). Rodent studies show IF improves longevity and health status particularly in rats and mice (Longo and Panda, 2016; Hwangbo et al., 2020; Green et al., 2022), although potential limitations should also be considered (Ingram and de Cabo, 2017; Sohal and Forster, 2014). On balance, the general thrust of literature suggests that IF has beneficial effects during aging, however, the molecular mechanisms involved are multifactorial and still under investigation.
In the present study, we show that aged mice displayed cognitive decline, increased intracellular Ca2+ buffering, decreased sIPSC frequency, and decreased quantal content of GABA release from presynaptic terminals. We further show that late-onset short-term (four to six weeks) IF reverses these physiological changes in aged mice, but fails to reverse cognitive impairment in aged mice. Different cellular mechanisms may mediate the positive effects of IF and have been studied extensively, including increased heart and kidney function, improved glucose tolerance metabolism, activation of transcription factors that enhance neuronal resistance to metabolic, oxidative and proteotoxic stresses (Camandola and Mattson, 2017; Ingram and de Cabo, 2017; Bayliak et al., 2021). Ultimately, IF may act through metabolic, cellular and circadian mechanisms leading to positive anatomic, functional and synaptic changes in the brain (Gudden et al., 2021; Camandola and Mattson, 2017; Mattson and Arumugam, 2018). Many of these mechanisms appear to act through Ca2+ signaling and synaptic mechanisms, and for this reason we focused our IF studies on intracellular Ca2+ buffering, synaptic transmission and quantal analysis. We have already shown that life-long CR can prevent the increase in Ca2+ buffering in aged rat BF neurons (Murchison et al., 2009) and aged BF mouse neurons (Fig. 4) suggesting these Ca2+ signaling mechanisms are mutable and subject to modification in the BF. The contribution of Ca2+ buffering to the synaptic functions in BF neurons was demonstrated with the addition of the exogenous Ca2+ buffer, BAPTA-AM (Griffith et al., 2014). In these experiments, increased Ca2+ buffering with BAPTA-AM in synaptic terminals reduced the frequency of sIPSCs in the young BF neurons, suggesting that increased Ca2+ buffering at synaptic terminals could mediate a deficit in BF synaptic transmission with age and contributes to cognitive impairment. In the present study, IF reversed the age-related change in all three parameters of increased Ca2+ buffering, inhibitory synaptic transmission and decreased quantal release highlighting the importance of Ca2+ signaling during aging. Overall enhancement of Ca2+ influx was observed in the BF during aging and this increase occurred through larger amplitude low-voltage activated (LVA) currents, and reduced current inactivation of high-voltage activated (HVA) currents (Murchison and Griffith, 1995, 1996). Interestingly, basal [Ca2+]i of aged BF neurons was unaltered, and the peak amplitude of Ca2+ transients (Δ[Ca2+]i) induced by either High K+ or Ca2+ current stimulation was decreased in BF cholinergic neurons of aged rats (Murchison and Griffith, 1998, 1999; Griffith et al., 2000). These findings suggest that the age-related increase in rapid Ca2+ buffering in aged BF neurons may be a possible compensatory mechanism to limit the rise of intracellular Ca2+ concentrations. It is unclear whether this compensatory increase in Ca2+ buffering is beneficial or detrimental although these mechanisms may interfere with synaptic functions. Studies have shown alterations in Ca2+-dependent physiology during aging contributes to changes in synaptic function and account for age-related cognitive impairment (Disterhoft et al., 1996; Thibault and Landfield, 1996; Foster and Norris, 1997; Foster, 1999, 2007, 2012; Tombaugh et al., 2005; Murphy et al., 2006; Gant et al., 2015, 2018; Kumar and Foster, 2019; Abu-Omar et al., 2018). In hippocampal CA1 pyramidal neurons, increased spike-mediated Ca2+ accumulation and slow after-hyperpolarization (sAHP) with age were prevented by CR, which was independent of Ca2+ clearance mechanisms (Hemond and Jaffe, 2005), while another study reported that CR offers neuroprotective effects under excitotoxic condition by increasing both calcium uptake rates and maximal uptake capacity in mitochondria (Amigo et al., 2017).
We did not investigate the exact mechanism for improvement in age-related synaptic dysfunction but the IF regimen we used here restored the age-related reduction in synaptic transmission (sIPSC frequency of BF neurons), and because changes in the synaptic frequency represents changes in probability of neurotransmitter release from presynaptic sites, we further investigate possible presynaptic mechanisms with age and IF by measuring quantal content of inhibitory synaptic transmission across aging. Quantal content was significantly reduced with age and reversed by IF. A reduction in quantal content could be one explanation of reduced sIPSC frequency with age and restored sIPSC frequency by IF. Furthermore, the probability of neurotransmitter release is directly affected by the level of Ca2+ within the synaptic terminal (Zucker, 1993; Wu and Saggau, 1997), hence the improved Ca2+ buffering by IF may influence positively on the presynaptic function in the BF neurons. It was previously reported that, in the animals with obesity-induced diabetes, CR improved diabetes-induced cognitive deficits by reversing high-fat diet-induced expression of neurogranin which regulates Ca2+/calmodulin-dependent synaptic function (Kim et al., 2016). Additional mechanisms are undoubtedly involved such as modulating synaptic functions by changes in neurotrophic factors or reduced oxidative stress both of which have been studied extensively with CR (Granger et al., 1996; Lee et al., 2002; Merry, 2002, 2004; Mattson et al., 2004; Gredilla and Barja, 2005; Hunt et al., 2006; Hyun et al., 2006; Rex et al., 2006; Tapia-Arancibia et al., 2008; Kishi et al., 2015; Longo and Mattson, 2014; Arumugam et al., 2010; Camandola and Mattson, 2017). Previous electrophysiological study also provides another possible mechanism of influence GABAergic inhibitory synaptic transmission by IF. In this study, the CA1 hippocampal neurons from mice on IF diets for one month showed upregulation of GABAergic synaptic activity via a mechanism requiring the mitochondrial protein deacetylase sirtuin 3 (SIRT3) and involving a reduction in mitochondrial oxidative stress (Liu et al., 2019). SIRT3 protects mitochondria and neurons against metabolic and excitotoxic stress (Cheng et al., 2016).
Reduction in neurotrophic factors and their signaling is one of the most consistent synaptic plasticity-associated impairments during aging, and CR protects against age-related deficits in synaptic plasticity and neuronal networks via upregulation of brain-derived neurotrophic factor (BDNF)/TrkB (Granger et al., 1996; Mattson et al., 2004; Rex et al., 2006; Tapia-Arancibia et al., 2008; Kishi et al., 2015). Our findings on the presynaptic mechanisms may be explained by either direct or indirect BDNF modulation on synaptic transmission (Tanaka et al., 1997; Frerking et al., 1998; Crozier et al., 1999; Tyler et al., 2006; Porcher et al., 2018). BDNF is negatively correlated to oxidative stress (Jain et al., 2013). In hypertensive rats, CR alone could not improve cognitive decline (Kishi and Sunagawa, 2012) but administration of antioxidant reagent could protect against cognitive decline via BDNF/TrkB in the hippocampus of hypertensive rats (Kishi et al., 2012), indicating CR might increase BDNF with reduction of oxidative stress. Another finding showed that the beneficial effect of CR on cognitive function was attenuated by blockade of BDNF receptors in the hippocampus of metabolic syndrome model rats but CR-mediated improvements in BDNF expression and oxidative stress were not altered (Kishi et al., 2015).
At a molecular level, age-related Ca2+ dysregulation and cognitive impairment in the hippocampus was reversed by viral-mediated transgene delivery of FK506-binding protein 12.6/1b (FKBP1b) to pyramidal neurons (Gant et al., 2015). The authors show that hippocampal overexpression of FKBP1b alleviates the age-related decline in this protein and reverses both age-related Ca2+ dysregulation and spatial memory impairment. Because KFB1b regulates ryanodine receptor (RyR)-mediated Ca2+ release and homeostasis, its decline with age and reversal with overexpress strongly suggests the importance of ER Ca2+ signaling in many of the mechanisms of age-related decline in hippocampal neurons. In BF neurons, both ER and mitochondria function change with age, however, these organelles do not impact Ca2+ buffering until large Ca2+ loads are supplied to the neurons (Murchison and Griffith, 1999, 2007; Murchison et al., 2004). This is in contrast to our data with Purkinje cells which demonstrate significant ER mediate calcium-induced calcium release and experience profound ER effects on rapid intracellular Ca2+ buffering (Dove et al., 2000). Undoubtedly, the complexity of different Ca2+ signaling pathways in neurons in different regions of the brain only add to the multiple cellular mechanisms responsible for age-related Ca2+ dysfunction and cognitive impairment.
Despite our finding that CR and IF reverse age-related synaptic and Ca2+ buffering impairment in mice, we do not see an improvement in cognitive function using the BM task. These data are consistent with a growing body of knowledge that short-term CR or IF may have little or no effect on cognitive function (Cao et al., 2001; Ingram and de Cabo, 2017; Kaur et al., 2008; Gudden et al., 2021; McQuail et al., 2021). Reasons for failure of IF to improve cognition are complex and may include the age of IF initiation, duration of fasting, genetic and dietary factors as well as the specific behavioral tasks used (Sohal and Forster, 2014; Ingram and de Cabo, 2017; McQuail et al., 2021). Early studies showed that long-term (12 month or lifelong) CR improved some forms of learning, memory and motor coordination in mice (Idrobo et al., 1987; Ingram et al., 1987) as well as spatial memory in F344 rats (Stewart et al., 1989; Pitsikas and Algeri, 1992). More recent studies demonstrate the complexity of cognitive reserve and resilience in aging mice and reinforce the importance of longitudinal studies, better genetic models and inclusion of diet and bioenergetics to help explain potential discrepancies across outcomes (Ingram and de Cabo, 2017; Dunn et al., 2020; Hwangbo et al., 2020; Green et al., 2022; McQuail et al., 2021). One additional factor that could influence BM activity is the mild stress that results from BM testing (Harrison et al., 2009). Although using the BM is less stressful than the water maze (Harrison et al., 2009), any mild stress may add to the mild stress resulting from food restriction resulting in apparent impairments especially early in learning training. In rodents, it is known that food restriction increases exploratory and foraging activities (Hebebrand et al., 2003; Carter et al., 2009). In our studies, young IF mice showed increased path length during the first day of training compared with young AL-fed mice (Fig. 7C), suggesting mild stress and increase exploratory behavior in this group may explain the first day of training and may underlie these initial differences observed.
In conclusion, our findings provide evidence that late-onset short-term IF can reverse age-related changes in calcium homeostasis and synaptic physiology in the mouse BF neurons. We did not observe an improvement in age-related cognitive decline with the four weeks of IF treatment. A single BM task may not be sensitive enough to determine reversal of age-related impairment in cognitive function. Longitudinal behavioral paradigms may be more sensitive to measure age-related changes. Together, these results suggest that short-term IF, even initiated in the late life, is sufficient to improve age-related alterations in cellular signaling in the BF neurons, but not enough to change behavioral cognitive function. We suggest that late-onset IF may have significant beneficial effects on age-related function without a lifetime commitment to diet.
Footnotes
This work was supported by the National Institutes of Health Grant AG047652 (to W.H.G.) and the Marek Family Fund for Alzheimer's Research (W.H.G.).
The authors declare no competing financial interests.
- Correspondence should be addressed to William H. Griffith at whgriff{at}tamu.edu
