• Open Access

Calcium buffering systems and calcium signaling in aged rat basal forebrain neurons

Authors

  • David Murchison,

    1. Department of Neuroscience and Experimental Therapeutics, College of Medicine, Texas A&M University System Health Science Center, College Station, TX, USA
    Search for more papers by this author
  • William H. Griffith

    1. Department of Neuroscience and Experimental Therapeutics, College of Medicine, Texas A&M University System Health Science Center, College Station, TX, USA
    Search for more papers by this author


William H. Griffith, Department of Neuroscience and Experimental Therapeutics, College of Medicine, Texas A&M University System Health Science Center, College Station, TX 77843-1114, USA. Tel.: (979) 845 4991; fax: (979) 845 0790; e-mail: griffith@medicine.tamhsc.edu

Summary

Disturbances of neuronal Ca2+ homeostasis are considered to be important determinants of age-related cognitive impairment. Cholinergic neurons of the basal forebrain (BF) are principal targets of decline associated with aging and dementia. During the last several years, we have attempted to link these concepts in a rat model of ‘normal’ aging. In this review, we will describe some changes that we have observed in Ca2+ signaling of aged BF neurons and the reversal of one of these changes by dietary caloric restriction. Our evidence supports a scenario in which subtle changes in the properties of voltage-gated Ca2+ channels result in increased Ca2+ influx during aging. This increased Ca2+, in turn, triggers an increase in rapid Ca2+ buffering in the somatic compartment of aged BF neurons. However, this nominal ‘compensation’, along with other changes in Ca2+ handling machinery (notably mitochondria) alters the Ca2+ signal with age in a way that is dependent on the magnitude of the Ca2+ load. By combining whole-cell patch clamp electrophysiology, ratiometric Ca2+-sensitive microfluorimetry and single-cell reverse transcription-polymerase chain reaction, we have determined that age-related rapid buffering changes are present in identified cholinergic BF neurons and that these changes can be prevented by a caloric restriction dietary regimen. Because caloric restriction extends lifespan and retards the progression of age-related dysfunction, these findings suggest that increased Ca2+ buffering in cholinergic neurons may be relevant to cognitive decline during normal aging. Importantly, calcium homeostatic mechanisms of BF cholinergic neurons are amenable to dietary interventions that could promote cognitive health during aging.

Introduction

It has been 20 years since the first formulations of the Ca2+ hypothesis of brain aging (Khachaturian, 1984; Gibson & Peterson, 1987; Landfield, 1987) focused research attention on the details of Ca2+ buffering and signaling in aged mammalian neurons. During that time, a clear distinction has emerged between mechanisms responsible for age-related neuronal pathologies (often characterized by cell death) and mechanisms underlying the more subtle changes associated with cognitive impairment in ‘normal’ aging (Foster & Norris, 1997; Morrison & Hof, 1997; Toescu, 2005; Burke & Barnes, 2006). While there have been several excellent reviews of neuronal Ca2+ homeostasis and aging (e.g. Thibault et al., 1998; Foster & Kumar, 2002; Toescu & Verkhratsky, 2003; Toescu et al., 2004), this review will briefly describe basic Ca2+ buffering mechanisms before focusing on Ca2+ buffering in basal forebrain (BF) neurons as a model of how age-related changes impact Ca2+ signaling functions. Understanding changes in Ca2+ buffering in cholinergic neurons of the BF is important because these cells are principal targets of decline associated with aging and dementia. Finally, we will suggest that age-related enhancement of rapid Ca2+ buffering in cholinergic (ChAT+) BF neurons is relevant to aging because it can be blocked by caloric restriction, the only intervention known to prevent or delay the physiological and cognitive dysfunctions of age.

Calcium buffering and signaling

Intracellular Ca2+ signaling is highly regulated by buffers that modulate the amplitude, duration, and extent of an influx of free Ca2+ in the cytoplasm (reviewed by Pozzan et al., 1994; Berridge et al., 2000, 2003). Calcium influx can arise from external sources via ligand- or voltage-gated channels (or store-operated/capacitative channels) and from internal release of Ca2+ sequestered in organelles. Figure 1A illustrates Ca2+ entry through voltage-gated calcium channels (VGCC) and subsequent sequestration and/or release from intracellular stores. Binding and release of Ca2+ by the rapid and slow buffers is a dynamic process that shapes net Ca2+ levels and permits various discrete signaling with the same signaling ion. Rapid buffering is usually thought to be mediated by calcium binding proteins (CaBP), while slow buffering is accomplished by pumps and exchangers. The slow buffers may expel Ca2+ through the plasma membrane (plasma membrane calcium ATPase, PMCA; Na+–Ca2+ exchanger) or they may sequester Ca2+ within an organelle [endoplasmic reticulum (ER), mitochondria]. These organelles themselves contain buffering and release mechanisms that are critical contributors to Ca2+ signaling (e.g., the smooth ER calcium ATPase, calreticulin, and inositol triphosphate (IP3) or ryanodine receptors of the ER). In some cases, traditional slow buffers can contribute to rapid buffering as well. For example, mitochondria participate in rapid buffering of large Ca2+ influxes (see below). Stationary or low-mobility rapid buffers restrict the spatial spread of a Ca2+ signal, while mobile buffers can enhance the spread. Slow buffers (Ca2+ clearance mechanisms) control the duration of a Ca2+ signal by removing Ca2+ from the cytoplasm and restoring the basal [Ca2+]i (~100 nm in most neurons). Restoration of the basal [Ca2+]i requires that the same amount of Ca2+ that entered the cell be expelled or sequestered.

Figure 1.

Summary of intracellular Ca2+ buffering and signaling. (A) Simplified diagram showing several major components of Ca2+ buffering in central neurons. Ca2+ enters the cytoplasm through voltage-gated Ca2+ channels (VGCC) when the cell is depolarized. Rapid buffers, possibly calcium binding proteins (CaBP), initially reduce the peak [Ca2+] with some contribution by endoplasmic reticulum (ER) and mitochondria for larger Ca2+ influx. Exchange may take place between the different intracellular organelles depending upon the Ca2+ load, and Ca2+ may be contributed to the cytoplasm also by release from ER or mitochondrial stores. Ca2+ is then cleared from the cytoplasm by uptake and retention in the organelles and by extrusion through the cell membrane by the plasma membrane Ca2+ ATPase (PMCA) and Na+–Ca2+ exchanger. (B) Idealized Ca2+ signals expected in the somatic cytoplasm in cases of different rapid buffer strengths. With no rapid buffer present, a 100-µm Ca2+ influx would peak near 100 µm and show fast decay kinetics. As rapid buffering capacity is added, peak responses are decreased and kinetics are slowed. Note that there is a delay in the peak intracellular Ca2+ level with increased buffering. (C) Examples of both small and large Ca2+ signals and the effects of the organelle blocking drugs thapsigargin (ER Ca2+ ATPase pump blocker) and carbonly cyanide m-chlorophenylhydrazone (CCCP, protonophore and mitochondrial blocker). Following a small Ca2+ influx (left), thapsigargin (green), and CCCP (orange) have little effect on the peak Ca2+ signal but prolong the decay. In contrast, following a large Ca2+ influx, both thapsigargin and CCCP dramatically alter both the amplitude and duration of the Ca2+ signal, suggesting the involvement of the ER and mitochondria in buffering large Ca2+ influxes. Therefore, different Ca2+ loads (small vs. large) will dictate how the intracellular buffering systems contribute.

Following a Ca2+ influx, rapid buffers bind Ca2+ almost instantaneously and limit the amplitude (peak free Ca2+ concentration) of the signal to a tiny fraction of the actual Ca2+ influx. Figure 1B shows an example of how global calcium signaling can be interpreted with the addition of different rapid buffers. Idealized Ca2+ signals are depicted for the theoretical case of no rapid buffering compared to the case of ‘weak’ rapid buffering and ‘strong’ rapid buffering. Note that there is a decrease in peak with increased buffering strength. The binding affinity of the rapid buffers can influence the rate of clearance by the slow buffers, with higher affinity buffers slowing the removal of Ca2+.

Within an individual neuron, the characteristics of a Ca2+ signal can differ substantially, depending on the magnitude of the influx and the location and/or concentration of intracellular buffers. Changes in any of these parameters could alter the quality of the Ca2+ signal experienced from an otherwise identical stimulus. Figure 1C shows an example of changes in the Ca2+ signal following application of blockers of intracellular organelle buffering systems. Following a small Ca2+ influx (left), thapsigargin, which blocks ER Ca2+ pumps [sarcoplasmic-ER Ca2+-ATPase (SERCA)] and prevents Ca2+ uptake by the ER (Murchison & Griffith, 1998) and carbonly cyanide m-chlorophenylhydrazone (CCCP), which dissipates the mitochondrial membrane potential (Murchison et al., 2004), have little effect on the peak Ca2+ signal, although the decay is slightly prolonged. Small Ca2+ loads are not sufficient to induce buffering by these organelles. In contrast, following a large Ca2+ influx, both thapsigargin and CCCP dramatically alter the amplitude and duration of the Ca2+ signal, suggesting the involvement of both the ER and mitochondria in buffering large Ca2+ influxes (Fig. 1C, right). Notice that the stereotypical ‘peak and shoulder’ features associated with the mitochondrial uptake and release of a large Ca2+ load are abolished by CCCP (Murchison & Griffith, 2000). Different Ca2+ signals will dictate how much the intracellular buffering systems contribute (for details, see Murchison & Griffith, 1998, 2000; Murchison et al., 2004).

Quantitative measurement of Ca2+ buffering

Investigation of Ca2+ buffering and signaling has been facilitated greatly by the application of Ca2+-sensitive fluorescent indicators, such as fura-2 (Grynkiewicz et al., 1985; Neher, 1995). Fura-2 changes the intensity of its fluorescent emissions differentially at excitation wavelengths of 340 and 380 nm when it binds Ca2+. As the [Ca2+] increases, the emissions from 340 nm excitation increase, and the emissions from 380 nm decrease. Thus, with fura-2 inside a cell, changes in the intracellular [Ca2+] (Δ[Ca2+]i) can be quantitated by recording the ratio of emissions from the two excitation wavelengths (340/380 nm). This ratiometric method enhances the resolution of Δ[Ca2+]i relative to that available when only a single excitation/emission is monitored. It also controls for certain inherent variables, such as light path length (i.e., cell thickness). Fura-2 is considered as a high-affinity indicator (for Ca2+, in vivo Kd≈ 250 nm in BF neurons), and as such, it has the drawback of becoming saturated at [Ca2+] above about 3 µm. However, a high-affinity indicator has the advantage of excellent resolution of small Δ[Ca2+]i (< 100 nm). In cases where very large (Δ[Ca2+]i) occur, such as in cell death or in restricted dendritic spaces, a lower affinity indicator is desirable in order to avoid saturation artifacts. Because we record the spatially averaged Ca2+ signal in the somatic compartments of acutely dissociated rat BF neurons (Fig. 2C) subjected to ‘physiological’ (rather than pathological) levels of Ca2+ influx, we have used fura-2 for most of our experiments.

Figure 2.

Ca2+ buffering values in F344 rat basal forebrain (BF) neurons. (A) Voltage-clamped Ca2+ currents and fura-2 fluorescence ratio records from an acutely dissociated neuron. Voltage steps (–60 to 0 mV) of different durations created increasing amplitudes of Ca2+ influx. Six current records and six ratio records are each superimposed and color coded. Neurons were loaded with 50 µm fura-2 K+5 via the whole-cell pipette. (B) A buffering curve is constructed for each neuron. The Ca2+ entry is calculated from the integrated currents. The Δ[Ca2+]i was measured from the averaged peak of each calibrated fura-2 record with the averaged pre-stimulus baseline subtracted. The buffering value β represents the ratio of bound to free ions, a larger β indicating stronger buffering. (C) An acutely dissociated cholinergic BF neuron is shown with patch pipette and colored circle representing the area from which the fluorescent Ca2+ signals are recorded.

For experiments in patch-clamped neurons, we commonly add 50 µm fura-2 K+5 to the intracellular solution. For experiments in ‘intact’ unpatched neurons, we load the cells via the membrane permeable acetoxymethyl ester (fura-2 AM), which is cleaved to the active salt by endogenous esterases. The loading protocol provides a fluorescent emission intensity equivalent to that from 50 µm fura-2 K+5. Background and cellular autofluorescence are subtracted at each wavelength (340 and 380 nm) prior to computation of the fluorescence ratio. In vitro and in vivo calibrations are used to estimate [Ca2+]i according to the standard equations (Grynkiewicz et al., 1985).

We quantified Ca2+ buffering by relating the [Ca2+]i expected to result from activation of VGCCs in the theoretical absence of buffering (Ca2+ entry) with the Δ[Ca2+]i measured from the fura-2 fluorescence signal (Tse et al., 1994; Murchison & Griffith, 1998). A buffering value for a neuron is obtained by constructing a buffering curve as follows: increasing levels of Ca2+ entry are generated in voltage-clamped neurons by activating VGCCs with voltage steps (–60 to 0 mV) of different durations and then plotting the resulting Δ[Ca2+]i against the calculated Ca2+ entry. Ca2+ entry is determined by integrating ICa measured during the step depolarization to yield total charge movement, then dividing by 2 to obtain the ionic Ca2+ flux, dividing by Avogadro's number to find the molar flux, and dividing by the estimated cell volume. Ca2+ entry is plotted against the mean peak amplitude of the Δ[Ca2+]i measured with fura-2. The slope (determined by linear regression) of this graph reflects the ratio of free to bound (buffered) Ca2+ ions, and the reciprocal of the slope (–1) is the rapid buffering value (β). Cell volume is estimated from the membrane capacitance, assuming the capacitance of biological membranes is 1 µF cm−2, and the cell is modeled as a hollow sphere. Because the actual accessible volume of the cell is smaller than the above model, the true Ca2+ influx is likely to be of greater concentration than we estimate, and therefore the calculated buffering values are likely to represent minimums. The concentration of the exogenous buffer (50 µm fura-2) is low enough that the contribution to the calculated buffering value should be minimal and constant without risking indicator saturation. Issues relating to this technique have been discussed by Neher (1995).

An example of data and plotted buffering curve in a BF neuron is shown in Fig. 2, where a sequence of voltage-clamped Ca2+ currents (steps from –60 to 0 mV, 20–200 ms) provides incrementally increasing levels of Ca2+ entry and the corresponding fluorescence ratio records representing Δ[Ca2+]i (Fig. 2A). Short duration voltage steps were employed in order to limit the levels of Ca2+ influx below those that result in substantial involvement of the ER or mitochondria. Rapid buffering in this ‘linear’ region of the buffering curve (Fig. 2B) then reflects the operation of the ‘other’ rapid buffers (presumably CaBPs). Note that the calculated Ca2+ entry reaches micromolar concentrations, while the measured Δ[Ca2+]i is restricted to nanomolar concentrations. Thus, a buffering value (β) of 204 in this neuron indicates that there was one free Ca2+ ion for every 204 that were buffered. Figure 2C shows the somatic area of an acutely dissociated rat BF neuron where the averaged Ca2+ signal is measured (red circle).

Calcium buffering in BF neurons

We utilize BF neurons as a model system because dysfunction of the BF cholinergic system occurs during aging, senile dementia, and Alzheimer's disease (Bartus, 2000). The BF interacts with cortical and subcortical sites to control cognitive processes such as attention, arousal, and some forms of memory (Sarter et al., 2003). BF contains a heterogeneous population of neurons, including cholinergic and GABAergic cell types (Han et al., 2005). It is hypothesized that properties specific to BF cholinergic cell types cause them to be susceptible to age-related changes that affect cognitive function (Wenk & Willard, 1999; McKinney, 2005).

Our laboratory has described a number of age-related changes in F344 rat BF neurons, including several involved in Ca2+ buffering and signaling (last reviewed in Griffith et al., 2000). Because VGCCs are major contributors to Ca2+ signaling and were possibly involved in age-related dysfunction (Landfield, 1987), we began by examining the properties of VGCCs in acutely dissociated BF neurons from young (1–4 months) and aged (20–28 months) rats with whole-cell and perforated-patch voltage clamp techniques (Murchison & Griffith, 1995, 1996). The most relevant findings were that low-voltage activated currents increased and that high-voltage activated currents had reduced inactivation during aging. These results supported the increased Ca2+ hypothesis, although not by the same increased L-type current mechanism found in hippocampal neurons (reviewed in Thibault et al., 1998). Interestingly, in mouse BF neurons, there appears to be a substantial increase in L-type current with age (Griffith & Etheredge, unpublished observation). By using single-cell reverse transcription-polymerase chain reaction (RT-PCR) techniques, we have determined recently that the low-voltage activated channels involved in the age-related increase of current density in rat BF cholinergic neurons probably contain the CaV 3.2 (α1H) subunit (Han et al., 2005).

When we compared buffering in young and aged BF neurons (Murchison & Griffith, 1998; Murchison et al., 2004), several features were observed. First, the resting [Ca2+]i is not different. We have examined literally hundreds of acutely dissociated BF neurons in both age groups, and we find the average resting [Ca2+]i to be ~90 nm. Therefore, aged BF neurons are not impaired in their intrinsic ability to regulate their basal Ca2+ levels. More importantly, rapid buffering values are markedly increased with age. The buffering values of aged BF neurons are 50–100% greater than those of young neurons (Murchison & Griffith, 1998; Murchison et al., 2004). This difference is reflected in the idealized buffering curves shown in Fig. 3A, where the slope of the linear portion of the aged curve (gray solid line) is reduced relative to that of the young curve (black solid line). Note that the differences in the curves are most pronounced in the linear portions (lower levels of Ca2+ entry, small arrow) and the differences are reduced as Ca2+ entry increases (large arrow) over the sublinear portions of the curves.

Figure 3.

Age-related differences in Ca2+ buffering depend upon the Ca2+ load. (A) To the left, idealized buffering curves (as described in Fig. 2) are shown for a young (black) s in the presence of carbonly cyanide m-chlorophenylhydrazone (CCCP). Note that the initial slopes of the curves are different, with the aged cell showing a reduced slope indicative of increased buffering. In the presence of CCCP, the curves become linear for the entire range of Ca2+ influxes. Arrows denote small and large Ca2+ influxes, respectively. At either age group, the differences between the control and CCCP curves are most pronounced at large Ca2+ influxes. In young, the difference between the control and CCCP curves is greater compared to aged, indicating a greater mitochondrial contribution to buffering large Ca2+ loads. To the right at top, a small Ca2+ signal (arrow) results in a larger peak response in young compared to aged, however, CCCP (dashed line) has very little effect on the peak response. The decreased response during aging is due to the increased rapid buffering. In contrast, following a large Ca2+ influx (large arrows below), the young and aged control responses are similar (solid lines) while, in the presence of CCCP (dashed lines), the contribution of mitochondrial buffering is much greater in young. (B) The greatest age-related change in Ca2+ buffering occurs with small Ca2+ influxes. An idealized graph of net intracellular Ca2+ buffering vs. the level of Ca2+ influx is shown. Buffering improves with increased entry as mitochondria and endoplasmic reticulum (ER) come into play, while the age-related difference declines with increased entry. Dashed lines emphasize the difference in buffering with age for small and large Ca2+ loads (arrows).

We then attempted to determine if increased Ca2+ uptake by ER or mitochondria was responsible for the difference in the aged buffering curve. We used thapsigargin (400 nm) to block ER Ca2+ pumps (SERCAs) and prevent Ca2+ uptake by the ER (Murchison & Griffith, 1998), and we prevented mitochondrial Ca2+ uptake by using the protonophore, CCCP (5 µm), which dissipates mitochondrial membrane potential (Murchison et al., 2004). Application of each of these drugs linearized the buffering curves without altering the slope relative to control. We interpret these results to mean that the ER and mitochondria do not make a net contribution to rapid buffering in BF neurons until relatively large Ca2+ loads are achieved, and so cannot be responsible for the difference in the aged buffering values, which are determined by smaller Ca2+ loads.

Figure 3A shows an example of idealized buffering curves in both young and aged cells during control and CCCP application. Dashed lines show the curves in the presence of CCCP. The age-related differences with regard to Ca2+ load can be seen clearly in this example. The initial slopes of the curves are different, with the aged cell showing a reduced slope indicating increased buffering. In the presence of CCCP, the curves become linear for the entire range of Ca2+ influxes. The small arrow denotes a small Ca2+ influx, while the large arrow shows a large influx. For both age groups, the differences between the control and CCCP curves are most pronounced at large Ca2+ loads. In ‘young’, the difference between the control and CCCP curves is greater compared to aged, indicating a greater mitochondrial contribution to buffering for large Ca2+ loads. The impact of mitochondrial buffering can be seen more clearly in the idealized signals to the right in Fig. 3A. As the level of Ca2+ entry increases, the age-related difference in the amplitude of the Ca2+ signal is reduced. By triggering large Ca2+ influx in the presence of CCCP, we observed a significant deficit in the ability of aged mitochondria to buffer Ca2+, at least partly attributable to reduced mitochondrial membrane potential (Murchison et al., 2004). Thus, the aged buffering curve is shaped by the combination of increased rapid buffering at low levels of Ca2+ influx and reduced mitochondrial buffering at higher levels. Notice in Fig. 3A that the age-related difference in the low-level rapid buffer contributes to a difference in the level of Ca2+ entry at which mitochondrial uptake becomes involved (where the curves become sublinear). A larger Ca2+ influx is needed in aged BF neurons to reach the level of [Ca2+]i at which net mitochondrial buffering is evident (500–600 nm in both young and aged).

As summarized in Fig. 3B, the age-related difference in net rapid Ca2+ buffering from all sources is greatest for low levels of Ca2+ influx where buffering is dominated by a nonmitochondrial, non-ER mechanism that is enhanced with age. However, as influx levels increase, the age-related differences in net Ca2+ buffering are reduced because of a decrease in mitochondrial uptake with age. Despite the similarity of net buffering between the age groups when larger Ca2+ loads are incurred, the quality of the Ca2+ signal propagated to targets within the soma of BF neurons is likely to differ with age for all levels of Ca2+ entry; for example, in aged BF neurons, a greater Ca2+ load would be required in order to provide the same level of mitochondrial Ca2+ signal experienced by a young neuron, as in Fig. 3A, lower right. This situation occurs because of the interactions between buffering mechanisms, the efficiency of which are changing in opposite directions with age.

Age-related Ca2+ buffering changes in ChAT+ BF neurons are prevented by caloric restriction

Recently, we have attempted to provide some insight into the functional impact at the organismal level of altered Ca2+ buffering with age in BF neurons by comparing the buffering of cells from young animals to those of ad libitum (AL) diet aged and caloric restricted diet aged animals. Our hypothesis is that if age-related buffering changes are detrimental, then caloric restriction might prevent them; while if the changes are functionally compensatory, then they might be enhanced by caloric restriction. We used single-cell RT-PCR to identify F344 rat BF cholinergic neurons following buffering data acquisition. We found that BF neurons expressing mRNA for the cholinergic marker choline acetyltransferase (ChAT) had significantly elevated rapid Ca2+ buffering values in AL aged animals. The buffering values in cholinergic neurons from young animals and in those from caloric restricted aged were not different (Fig. 4). The mean buffering values of all ChAT+ neurons sampled and the mean values averaged for each individual subject were significantly increased in the aged AL rats (Fig. 4B). Several other cellular properties were unchanged with age or dietary regimen, including basal [Ca2+]i and properties of the Ca2+ current (not shown). However, the reduction in membrane capacitance (Cm) that is known to occur in BF neurons from AL aged rats was not observed in aged caloric restricted rats (young: 18.3 ± 0.9 pF, n = 31; caloric restricted: 18.7 ± 0.9 pF, n = 21; AL: 15.2 ± 0.6 pF, n = 37; P < 0.05).

Figure 4.

Age-related changes in rapid buffering are prevented by caloric restriction. (A) Mean buffering curves for neurons from young, aged, and aged caloric restricted subjects. Mean values for each group are shown in parentheses. Summary graphs of buffering values in cholinergic (ChAT+) neurons. Increased buffering values observed in ChAT+ neurons from aged ad libitum (AL) diet rats are not seen in neurons from aged caloric restricted rats. (B) Summary graphs show that aged AL diet rats have a significant (*P < 0.05) increase in Ca2+ buffering values of ChAT+ basal forebrain (BF) neurons that is not present in aged caloric restricted animals. This is true whether the results are analyzed by neuronal numbers (left, number of cells in parentheses) or by averaging the cells within each subject (right, number of subjects in parentheses). See text for additional experimental details.

RT-PCR, cholinergic neurons, and calcium-binding proteins

We have employed molecular techniques, including single-cell RT-PCR (Griffith et al., 2006) to identify cholinergic neurons and study the expression of CaBPs. We probed for mRNA expression of the cholinergic marker ChAT and the GABAergic marker glutamic acid decarboxylase (GAD), as well as for the positive cell marker glyceraldehyde-3-phosphate dehydrogenase (GAPDH) and for the CaBPs calbindin (CB), calretinin (CR), and parvalbumin (PV). As in previous investigations of BF neurons in which single-cell RT-PCR was used to identify cell types, ChAT, GAD, both ChAT and GAD (ChAT+/GAD+) or neither mRNA was detected in individual neurons (Tkatch et al., 1998; Han et al., 2002; Williams et al., 2002). We have accumulated evidence that most acutely dissociated medial septum/nucleus of the diagonal band (MS/nDB) neurons expressing ChAT mRNA have properties consistent with a cholinergic phenotype, regardless of the detection of GAD in the same cell (Han et al., 2005; Murchison et al., 2006). We have been unable to identify correlations between the expression of CB, CR, or PV with any age-related changes in Ca2+ signaling. In the data set for the caloric restricted study, the detection of CaBP mRNA expression in ChAT+ neurons was not correlated with physiology (not shown). No CaBP mRNA was detected in 13/31 young, 8/21 caloric restricted, and 20/37 AL aged neurons. For cells in which CaBP expression was detected, all but one were CR+. At the level of BF tissue, there was no age-related difference in the expression of CaBP mRNA or protein (Han et al., 2002).

Possible buffering mechanims

The mechanism of the age-related rapid buffering difference remains elusive. We have shown that it is not due to Ca2+ uptake by ER (Murchison & Griffith, 1998) or to mitochondrial uptake (Murchison et al., 2004). Also, we found that it was not associated with the level of mRNA or protein expression of CB, CR, and PV in BF tissue (Han et al., 2002). Furthermore, the detection of CaBPs in ChAT+ neurons was not correlated with buffering values. It could be that some other CaBP is involved (there are dozens), or even that membrane clearance mechanisms could contribute. The reduced cytoplasmic volume implied by reduced Cm in AL aged neurons could functionally concentrate the rapid buffers so that they are in closer proximity to VGCC influx sites, and therefore buffer more efficiently. Decreased BF cholinergic neuron size with age is well documented (Fischer et al., 1989; Niewiadomska et al., 2002; Veng et al., 2003). Cytoplasmic Ca2+ buffering has not been measured directly in any other aged neurons, however, sufficient cumulative evidence exists to support a conclusion of generally reduced buffering in aged CA1 hippocampal neurons. In these neurons, increased activity of L-type VGCCs is well documented (Moyer & Disterhoft, 1994; Campbell et al., 1996; Thibault & Landfield, 1996), and increased Ca2+ transients have been observed (Thibault et al., 2001; Hemond & Jaffe, 2005; Gant et al., 2006). Furthermore, CaBP expression is reduced in aged hippocampus (Villa et al., 1994; De Jong et al., 1996; Krzywkowski et al., 1996). Importantly, age-related deficits in hippocampal synaptic transmission can be alleviated by the addition of exogenous Ca2+ buffers (Ouanounou et al., 1999). Possibly, the increase in native buffering in aged BF cholinergic neurons is an attempt to maintain the integrity of synaptic function by a means not available to hippocampal neurons.

Relevance to cognitive aging

The evidence that increased Ca2+ buffering during aging in BF cholinergic neurons is preventable by caloric restriction supports the conclusion that this buffering change is physiologically relevant to the process of cognitive aging. In several investigations (but not all, e.g., Markowska, 1999), caloric restriction has been shown to prevent age-related cognitive and behavioral deficits (Ingram et al., 1987; Stewart et al., 1989; Gyger et al., 1992; Pitsikas & Algeri, 1992), as well as some physiological, metabolic, and gene expression changes with age (Mattson et al., 2003; Weindruch, 2003). We have proposed that this buffering increase is a compensatory response to increased Ca2+ influx with age (Griffith et al., 2000). Our latest findings suggest that this compensation may be functionally detrimental based on the results from the calorically restricted animals. Experiments are in progress to test directly these conclusions using cognitively characterized rats. Additionally, an inverse correlation has been reported between BF cholinergic cell size and cognitive impairment (Fischer et al., 1987; Martinez-Serrano et al., 1995; Backman et al., 1996), such that the prevention of age-related BF cholinergic neuron atrophy by caloric restriction could be an important mechanism by which cognition is preserved.

The quality of a Ca2+ signal from a given stimulus is likely to be altered in aged BF neurons. The relevance to cognitive aging of seemingly subtle changes in the quality of neuronal Ca2+ signaling has been demonstrated convincingly in the hippocampus. There, increases in L-type Ca2+ channel activity with age (Moyer & Disterhoft, 1994; Campbell et al., 1996; Thibault & Landfield, 1996) can be associated with enhanced after hyperpolarization (AHP) (Power et al., 2002) and synaptic plasticity deficits (Norris et al., 1998; Thibault et al., 2001), as well as cognitive deficits (Disterhoft et al., 1996; Tombaugh et al., 2005).

Caloric restriction is perhaps the most intriguing way in which age-related changes may be prevented. This paradigm, in which adult animals have their diet limited to 60–80% of the average caloric intake, extends lifespan in a wide variety of species (Weindruch, 2003) and is the only ‘treatment’ that extends mammalian lifespan (Barger et al., 2003). Additionally, caloric restriction delays the onset of age-related diseases in animal models, and protects against neurodegeneration and cognitive impairment (Mattson et al., 2003). Caloric restriction has been envisioned as a ‘mild’ stress on cells, stimulating a complex physiological response that enables the cell to resist age-dependent insults (Masoro, 2000). Several mechanistic pathways have been proposed that that could be involved simultaneously in different tissues (see reviews, Patel & Finch, 2002; Hursting et al., 2003; Bordone & Guarente, 2005). Because caloric restriction can protect neurons from degeneration, cell death mechanisms, including disrupted Ca2+ homeostasis, are probable target processes in brain (Mattson et al., 2001).

Relevant to the above, Hemond & Jaffe (2005) showed that caloric restriction can reduce age-related Ca2+ accumulation and the associated AHP in CA1 neurons. This could explain how an age-related reduction in hippocampal long-term potentiation (LTP) is rescued in caloric restricted animals (Hori et al., 1992; Eckles-Smith et al., 2000). In an investigation of the mechanism of LTP maintenance by caloric restriction, Okada et al. (2003) observed that there was no difference between the amplitudes of depolarization-induced Ca2+ transients in CA1 neurons of young rats and those of aged caloric restricted rats. Such findings support the interpretation that neuronal Ca2+ homeostatic mechanisms are an important target of caloric restriction.

Concluding remarks

Calcium signaling is so important to cellular function that there is a considerable amount of compensatory plasticity and redundancy contained in the regulatory systems that maintain it. This sort of complexity, combined with the relatively small number of easily interpretable experimental manipulations that can be performed on Ca2+ buffering systems has certainly impeded our understanding of the mechanisms relating altered neuronal Ca2+ signaling to age-related cognitive impairment. However, as more sophisticated techniques are applied (knockout/in technology, gene expression arrays, and proteomics approaches), specific targets may be identified for therapeutic interventions designed to maintain quality of life into old age by reducing cognitive impairment. That Ca2+ homeostatic mechanisms of BF neurons can be manipulated by caloric restriction during aging is an intriguing result suggesting a number of potential target mechanisms. As a larger and larger portion of the population survives to ever greater ages, the importance of finding cognitive therapies increases from both social and moral perspectives.

Acknowledgements

This research was supported by National Institute on Aging grant no. AG07805. The assistance of Drs Sun-ho Han and Alan Parrish is appreciated greatly.

Ancillary