Testing the Calcium Hypothesis of Aging in the Rat Hippocampus In Vivo Using Manganese-enhanced MRI

Abstract

In this longitudinal study, we non-invasively tested the hypothesis that Mn²⁺-enhanced MRI (MEMRI) is sensitive to age-related changes in Ca²⁺ influx occurring in the hippocampal region CA1. Uptake of Mn²⁺, an MRI contrast agent and Ca²⁺ surrogate with low cellular efflux rates (days to weeks), was measured in longitudinal MEMRI studies involving two separate groups of male Long-Evans rats: One group was studied at ages 2.5 and 7 mo, while the other was studied at 7 and 19 mo. Separate or combined analysis revealed that the extent of Mn²⁺ accumulation in CA1 significantly increased with age (P<0.05). These results provide first-time in vivo confirmation of the calcium hypothesis of aging and justify future longitudinal studies combining MEMRI with behavioral testing to investigate mechanisms of age-related cognitive decline.

1. Introduction

Healthy aging is associated with declines in cognition that have been linked to increases in hippocampal CA1 Ca²⁺ influx through L-type voltage-gated Ca²⁺ channels (L-VGCCs) (Campbell et al., 1996; Thibault and Landfield, 1996). To-date, there have been no attempts to non-invasively measure CA1 Ca²⁺ influx during senescence, and this has greatly limited translational studies aimed at understanding its natural history, consequences, and potential modifiers. Here, we tested whether Mn²⁺-enhanced MRI (MEMRI) is sensitive to age-related increases in CA1 Ca²⁺ influx. In MEMRI, the contrast agent Mn²⁺ is systemically administered, accumulates in active neurons while the animal is awake and freely moving, and is later measured with MRI (Bissig and Berkowitz, 2011). Importantly, Mn²⁺ uptake occurs through L-VGCCs (Bissig et al., 2013; Cross et al., 2007; Drapeau and Nachshen, 1984).

2. Methods

Here, we analyze brain images collected from Long-Evans rats (Hilltop Labs; Scottdale, PA) that were part of a study focused on retinal physiology (Bissig et al., 2013). Two groups of rats were studied longitudinally with MEMRI. One group (n=8) was imaged as young adults (aged 2.5 mo), then imaged again in mid-adulthood (7 mo). A second group (n=7) was imaged as mid-adults (7 mo) then imaged again as old adults (19 mo). For both groups, we anticipated full clearance of accumulated Mn²⁺ between imaging sessions, given the ~12 d half-life for brain Mn²⁺ (Chuang et al., 2009). A third group of animals (n=7) were used to measure baseline MRI signal in the absence of Mn²⁺ injection. Those rats were studied cross-sectionally at either young (n=3), mid- (n=3), or old adulthood (n=1). All rats with a complete set of high-quality brain scans are included in the present report, encompassing a representative and relatively large subset of those studied in Bissig et al. (2013). All experiments were run in accordance with the NIH Guide for the Care and Use of Laboratory Animals.

Brain scans were collected after the retinal imaging protocol in Bissig et al. (2013) was complete. As part of that study, rats were awake, free-moving, housed alone, and periodically examined visually for signs of discomfort or distress during the 4 hr after intraperitoneal injection of 44 mg MnCl₂∙4H₂O / kg body weight. They were then anesthetized for the duration of MRI scans, and kept warm with a recirculating water blanket. The first 2.5 hr of scanning (mean(SD) 2.5(0.1) hr for young, 2.6(0.3) hr for mid-adult, and 2.5(0.4) hr for old adults) were spent acquiring data from each eye. Brain scans were started immediately after eye scans were complete, 7 hr after Mn²⁺ injection. Because the pattern of Mn²⁺ accumulation can be altered if immediately preceded by anesthesia (Itoh et al., 2008) the timing between Mn²⁺ injection and anesthesia in the present experiments was designed to allow Mn²⁺ accumulation to reach plateau levels in the retina (<4 hr; Tofts et al., 2010) and hippocampus (4–6 hr; Lee et al., 2005) while the animal remained awake and free-moving. Although modest additional Mn²⁺ enhancement has been reported beyond 12 hr post-injection for some brain regions, it is attributed to factors other than uptake via L-VGCCs, such as axonal transport of Mn²⁺ (Chuang and Koretsky, 2009). Because the rate of axonal transport decreases with age (Cross et al., 2008), delaying brain scans much beyond the present 7 hr post-injection time would have risked obscuring the hypothesized age-related changes in Mn²⁺ uptake via L-VGCCs.

To quantify CA1 Mn²⁺ signal intensity enhancement, we applied the ratio approach of (Van de Moortele et al., 2009) using our previous brain imaging parameters (Bissig and Berkowitz, 2011). A transmit-only whole body coil and a 2×2 phased-array receive-only surface coil placed directly on the dorsal surface of the head were used to collect brain scans on a 7 T ClinScan system controlled by Siemens’ Syngo software (Bruker BioSpin, Billerica, MA). Images were collected using a 3D turbo-FLASH sequence (flip angle 3°; repetition time (i.e. echo spacing) 7.77 ms; echo time 3.03 ms; 192 × 192 × 128 matrix; 2.50 cm × 2.50 cm × 3.32 cm field of view; providing a mediolateral × dorsoventral × rostrocaudal resolution of 130 μm × 130 μm × 260 μm) both with (TI 1500 ms) and without an inversion pulse (1 acquisition of each, requiring 11 min 12 s total). The magnetization prepared rapid acquisition gradient echo (MPRAGE) image, collected using the inversion pulse, is strongly T₁ weighted. In contrast, the proton density weighted gradient echo (PDGE) image, collected without the inversion pulse, has no T₁ weighting (Haase, 1990). At each time point, we divided the MPRAGE by the PDGE image, thereby removing non-biological sources of signal variability shared by both images (Van de Moortele et al., 2009), creating a map of heavily T₁-weighted ratio values (RVs) for each subject (Fig.1).

Figure 1: Image Analysis.

A representative coronal slice — zoomed-in to detail the hippocampus — is shown for a control (left column) and a Mn²⁺-injected (middle column) mid-adult rat. The same area is shown in PDGE (top row) and MPRAGE (middle row) images, and the resultant MPRAGE/PDGE ratio map (bottom row). The bottom-most ratio maps are zoomed-out from the hippocampus (white box) to show muscle (large circle) and pituitary (small circle) regions-of-interest. In the zoomed-in images, five of the thirty 520 μm-wide perpendiculars used to sample the hippocampus (white rectangles) of each brain are shown. As denoted by the small dashes within, perpendiculars are centered on the interior white/grey matter border identified in PDGE images. This border is obscured in MPRAGE and ratio images by Mn²⁺-enhancement of gray matter signal (lighter grayscale; blue to green to yellow to red). The average PDGE, MPRAGE, and RV profiles from the pictured perpendiculars (figure right) show the Mn²⁺-independent PDGE profiles converging midway between local minimum/maximum values for proton density (i.e., the interior white/grey matter border; 0 on x-axis). Mn²⁺-enhancement is evident throughout MPRAGE and ratio profiles, including the CA1 layer analyzed in Figure 2 (py; green shaded area). Positions of cortex (cortical grey matter), dcw (deep cortical white matter), alv (alveus), ori (oriens layer), py (pyramidal layer), rad (radiatum layer) are derived from (Paxinos and Watson, 2007) and noted at the bottom of the profiles.

CA1 RVs were quantified in three adjacent coronal slices (within bregma −6.5 to −5.7 mm; (Paxinos and Watson, 2007)) selected by local anatomical landmarks. They were 260 μm (the thickness of a coronal slice), 520 μm, and 780 μm rostral to the caudal-most coronal slice that contained the granular layer of dentate gyrus and the pyramidal layer of CA1. In each of the three coronal slices per animal per time point, we selected ten (five per hemisphere; Fig.1) 520 μm-wide bands oriented perpendicular to local white matter. The dorsal-most of these evenly-spaced bands was positioned just lateral to the dorsal hippocampal commissure, which appears in coronal sections as an abrupt dorsomedial thickening of local white matter. The ventral-most band was placed at the level of the inferior colliculus (Fig.1). RVs were averaged within these bands to produce a profile of RV as a function of distance from local white matter, which includes both the alveus (alv) of the hippocampus and the deep cerebral white matter (dcw). These RV profiles were aligned to the interior, sub-cortical, white/grey matter border, identified based on PDGE signal intensity profiles collected with identically-placed bands. Although PDGE images lack the T₁ tissue contrast that is altered by Mn²⁺ (Haase, 1990), the location of white matter is easily identified by its relatively lower proton density (i.e., lower signal) than surrounding grey matter. This strategy avoids confounding Mn²⁺ accumulation with the identification of structural borders.

Prior ex vivo demonstrations of age-related L-VGCC changes used pyramidal neurons within CA1 (Campbell et al., 1996; Thibault and Landfield, 1996). We therefore focused our analysis on CA1 RVs that encompass the ~110 μm-wide pyramidal layer (py), which is densely packed with the somas of pyramidal neurons. Limited by the present 130 μm image resolution, this was accomplished by averaging profile RVs in the section +130 to +260 μm from the interior white/grey matter border (Fig.1). Although some partial-volume averaging with adjacent layers of the hippocampus is anticipated, those within one voxel-width of the selected span — the oriens (ori), alveus (alv), and radiatum (rad) layers — are largely composed of pyramidal cell axons and dendrites. It was therefore unsurprising that alternative sampling strategies, such as a wider or narrower profile section centered on py, or choosing a section slightly off-center to py, had no effect on the conclusions of the present work (not shown).

We considered other analysis strategies to assure consistent localization and sampling of py across ages and groups. Performing a non-rotational rigid-body registration prior to region-of-interest analysis would have been based on the same local anatomical landmarks used presently, and so would not alter findings. Adding a rotation step to the rigid-body registration would correct for small (a few degrees) variability in alignment observed between subjects, but that approach would have modified signal intensities (and RVs) of py due to signal averaging (interpolation) with adjacent structures. In either case, between-subject differences in the exact positioning and curvature of py discouraged attempts to sample images based on fixed set of spatial coordinates across all subjects and times, and suggested the present approach of sampling py based on the nearby and easily-identified layer of white matter. Although a fixed set of spatial coordinates might be used if data were first spatially warped to a common atlas image, that procedure likely compromises layer-specific analysis of MEMRI data due to substantial signal averaging (interpolation) with adjacent tissue layers (Bissig and Berkowitz, 2009). Spatial warping might nevertheless have some appeal if there were dramatic age-related changes in brain size — large enough to compromise tissue identification based on fixed (e.g., rather than proportional) distances from anatomical landmarks. To assess this possibility in longitudinally-studied rats, we measured the dimensions of a box spanning from (i) the caudal border of dentate gyrus to the rostral border of the fimbria of the hippocampus, (ii) the widest mediolateral expanse of the brain within the coronal slices used for analysis, and (iii) the dorsoventral span in that same slices (dorsal-most cortex to ventral-most midbrain). As detailed in the Results section, averaged across subjects, the largest longitudinal size increase in any dimension was roughly 3%. If that percent increase were applied to the py region, we estimate that its rostrocaudal extent would increase by less than the thickness of a coronal slice (i.e., <260 μm), and its dorsoventral extent would increase by less than the available resolution in the coronal plane (i.e., <130 μm). Mediolaterally, we estimate that the thickness of py and the distance from py to its local (white matter) landmark would each increase by <4 μm (e.g., ~3% of 110 μm). In short, since estimated changes in py size were below the available image resolution, it is unlikely that brain growth compromised localization of py based on fixed distances from anatomical landmarks.

To determine if systemic Mn²⁺ handling was significantly altered with age, in each ratio image we measured average RVs of skeletal muscle and the anterior pituitary, tissues which do not have blood barriers. Muscle RV was measured with two spherical volumes of interest (780 μm radius; one for each side of the head) in muscle directly lateral to our CA1 measurements. Average muscle RV from these regions was used to normalize brain RVs (i.e., normalized brain RV = brain RV / muscle RV). Anterior pituitary RV was measured with two spherical volumes of interest (390 μm radius; one on each side of the midline; Holt et al., 2010).

Before testing the central hypothesis of the present work, we ran several statistical comparisons to verify the sensitivity of our imaging parameters to Mn²⁺ uptake, and to gauge the appropriateness of both muscle normalization and calculation of RVs from MPRAGE and PDGE signal intensities. Furthermore, since the time between Mn²⁺ injection and scan differed between young and old adults (mean(SD) of 6.7(0.1) hr for young, 6.9(0.6) hr for mid-, and 7.4(0.5) hr for old adults), and animal body weights increased with age (304(81) g for young, 468(53) g for mid-, and 623(76) g for old adults) we tested whether our Mn²⁺ uptake measurement for py was affected by either animal body weight or time between Mn²⁺ injection and scanning. Data from age 7 mo (the age with the most subjects and widest range of time spans between Mn²⁺ injection and scanning (6.4–8.5 hr)) were used for these preliminary tests, which were restricted to Mn²⁺-injected rats unless otherwise noted.

There are several viable statistical approaches that may be used on the present data, which were collected as an add-on to a previous study and include both cross-sectional and longitudinal components. In our preferred approach, we first test whether RVs from Mn²⁺-injected rats increased with age (log-transformed in all analyses to reduce skewness) using a generalized estimating equation (GEE) analysis to accommodate longitudinal measurements (Burton et al., 1998; Hanley et al., 2003; geepack library in R 3.0.1 (http://www.r-project.org)). Finding an age-related increase in py RVs among Mn²⁺-injected rats would support the present hypothesis — that pyramidal cell Mn²⁺ uptake increases with age — but should be distinguished from an age-related change in baseline MRI signal (though none is expected; El Tannir El Tayara et al., 2006). To address that possibility, we used linear regression to test for an effect of age on the RVs from uninjected controls. That step was followed by combining data from control and Mn²⁺-injected rats into broader GEE analyses to test for a Mn²⁺ injection × age interaction. A positive result for the interaction would demonstrate an age-related change in Mn²⁺-injected animals in significant excess of any trend present in the control group. In an alternative to the GEE analyses, we tested whether RVs from Mn²⁺-injected rats increased with age using a mixed-effects model (age as fixed effect, individual as random effect; lme4 and lmerTest libraries in R). Slope estimates from Mn²⁺-injected rats (mixed-effects model) versus control rats (linear regression) were then compared using the t test for parallelism (Kleinbaum and Kupper, 1978), such that a positive result would demonstrate an age-related change in Mn²⁺-injected animals in significant excess of any trend present in the control group. The effects of Mn²⁺ injection, age, and the interaction were also tested on muscle RVs and muscle-normalized pituitary RVs. Where these analyses showed an effect of age, we clarified changes from young to mid- and mid- to old adulthood with t-tests. Results were considered significant at P<0.05.

3. Results

We evaluated age-related changes in brain size relevant to analysis of CA1 by measuring the dimensions of a box spanning (i) from caudal dentate gyrus to rostral fimbria, and (ii) the mediolateral expanse and (iii) the dorsoventral expanse of the brain in coronal slices. At age 2.5 mo, these dimensions (mean±s.e.m.) were (i) 6.18±0.10 mm, (ii) 15.22±0.04 mm, and (iii) 8.15±0.08 mm. Among those rats, between 2.5 and 7 mo dimensions increased (i) by 0.20±0.13 mm (3.2%; paired two-tailed t-test P>0.1) to reach 6.37±0.07 mm, (ii) by 0.42±0.12 mm (2.7%; P=9e-3) to reach 15.64±0.09 mm, and (iii) by 0.18±0.07 mm (2.2%; P=0.04) to reach 8.33±0.06mm. Those rats first studied at age 7 mo initially showed similar dimensions (each P>0.05); (i) 6.39±0.05, (ii) 15.31±0.13, and (iii) 8.23±0.10. Between 7 and 19 mo, dimensions non-significantly trended higher (i) by 0.19±0.11 mm (2.9%; P>0.1) to reach 6.57±0.09 mm, (ii) by 0.32±0.19 mm (2.1%; P>0.1) to reach 15.63±0.15 mm, and (iii) by 0.08±0.10 mm (0.9%; P>0.4) to reach 8.30±0.07 mm. In short, we detected only minimal longitudinal growth of the gross brain area containing hippocampus.

In our assessment of image sensitivity to Mn²⁺ using mid-adult rats, we found that Mn²⁺-injected rats had higher RVs than uninjected controls at muscle, anterior pituitary, and the py layer of CA1 (all P<0.05). Consistent with expectations (Haase, 1990), this effect was driven by group differences in brain MPRAGE signal intensities (all locations P<0.05), since the T₁-independent PDGE signal intensities were no different in Mn²⁺-injected and control rats (all locations P>0.2). PDGE and MPRAGE signal intensities were strongly correlated at each location (all r>0.92, P<0.05), consistent with their shared sources of hardware-related signal variability, which are removed by calculating RVs. Modest correlations between each regions’ pre-normalization RVs were expected (muscle and py, r=0.82, P=2e-4; muscle and pituitary, r=0.46, P=0.08; py and pituitary r=0.49, P=0.06), representing a source of variability removed by muscle normalization. After muscle normalization, there was no correlation between pituitary and py RVs (r=0.15, P>0.5). We found no relationship between muscle-normalized py RVs and either body weight (r=0.07, P>0.7) or the time between Mn²⁺ injection and scanning (r=−0.39, P>0.1).

In support of the present hypothesis, GEE analysis of data from Mn²⁺-injected rats demonstrated that RVs from the py layer of CA1 significantly increased with age (P=2.2e-5; slope estimate ± standard error of 0.046±0.011 per unit age, in ln(months); Fig.2). The same analysis applied to muscle and pituitary RVs demonstrated that both were stable with age (both P>0.2; Fig.2), arguing that systemic handling of Mn²⁺ is independent of age. Mixed-effect models similarly showed a significant effect of age on py RVs (P=4e-4; 0.046±0.011 per ln(months)) but not muscle and pituitary RVs (both P>0.3). Increases in py RVs from Mn²⁺-injected rats were noted both from young to mid-adulthood and mid- to old adulthood (paired two-tailed t-tests; respectively P=0.020 and P=0.047). Comparing Mn²⁺-injected rats first studied at age 2.5 mo with those first studied at age 7 mo (unpaired two-tailed t-test), we similarly found py RVs increased from young to mid-adulthood (P=0.034). That finding is consistent with the similar 7 mo values from each sub-group (i.e., in Fig.1, ● vs ○ at age 7 mo; P>0.7), suggesting that the history of Mn²⁺ exposure had no effect on the present results. When pituitary and muscle RVs of Mn²⁺-injected rats were analyzed with the same battery of t-tests, no significant differences were found (all P>0.1), consistent with GEE analyses showing no effect of age on RVs from these negative-control regions.

Figure 2: Analysis of Age-related Changes in CA1 Mn²⁺ Uptake.

Muscle RVs and muscle-normalized pituitary RVs were higher in Mn²⁺-injected (● and ○, depending on age at first scan) than control rats (grey ■), but were not affected by age. Muscle-normalized RVs from CA1 pyramidal cells (py layer, see Figure 1) increased significantly with age in Mn²⁺-injected rats, but not in controls. This demonstrates that CA1 pyramidal cell Mn²⁺ uptake increases with age. In each plot, a black line connects longitudinal measurements from the same subject. Background gray lines and shaded areas denote best-fit lines and 95% confidence intervals.

Visual inspection (Fig.2) suggested that the robust age effect at py was unrelated to changes in tissue baseline. Among uninjected controls, no effect of age on muscle, pituitary, or py RVs was found in linear regression analyses (for each, −0.50<r<0.11 and P>0.2). For py, the aging pattern of uninjected controls (slope estimate ± standard error of −0.002±0.012 per ln(month)) appeared distinct from the pattern in Mn²⁺-injected rats. This impression was supported by GEE models including both Mn²⁺-injected and uninjected control data, which showed that Mn²⁺-injected rats’ age-related increase in py RVs was in significant excess of the control trend (Mn²⁺ injection × age interaction; P=5e-5). Neither muscle nor pituitary showed this pattern (interaction P>0.05 for both) despite clear evidence of Mn²⁺ enhancement (i.e., higher RVs in Mn²⁺-injected than controls; main effect; P<4e-7 for muscle, pituitary, as well as py). Tests for parallelism between mixed-model slope estimates (e.g., 0.046±0.011 at py) and slope estimates for uninjected controls (e.g., −0.002±0.012 at py) were in good agreement with GEE results: Mn²⁺-injected and control rat RVs had significantly different relationships to age at py (P=0.042) but not at pituitary or muscle (both P>0.2).

4. Discussion

Our major finding is that in vivo Mn²⁺ uptake significantly increased with age in the py portion of CA1, where the pyramidal cell somas are located. Our scope was tightly focused by previous work demonstrating age-related changes CA1 pyramidal cell L-VGCCs (Campbell et al., 1996; Thibault and Landfield, 1996). Since L-VGCCs are a dominate influx path for Mn²⁺ uptake (Bissig et al., 2013; Cross et al., 2007; Drapeau and Nachshen, 1984), our results are consistent with this literature. Based on the present results supporting this ‘Ca²⁺ hypothesis of aging’, it is now possible to test, for example, whether the level of Mn²⁺ uptake in mid-adulthood predicts the magnitude of behavioral declines into old adulthood. These results highlight, for the first time, the potential usefulness of MEMRI in the context of the aging hippocampus.

Although the present study is the first to assess age-related changes in hippocampal ion influx with MEMRI, a prior study by Kuo and colleagues (2010) described an age-related increase in hypothalamic Mn²⁺ uptake in mice. These findings, combined with the work Das and Ghosh (1996) — who reported that the rat hippocampus, cortex, and putamen all show similar age-related increases in systemically injected 45Ca²⁺ uptake — imply that age-related increases in ion influx are widespread throughout the brain. Future MEMRI studies may evaluate this possibility, but would need to address several factors not applicable to the hippocampus. First, even from young to mid-adulthood, age-related increases in blood brain barrier (‘BBB’) permeability have been documented in the cerebellum, hypothalamus, and parts of the neocortex (Saija et al., 1992). As discussed by Kuo and colleagues (2010), such an increase in BBB permeability may complicate interpretation of age effects by increasing neuronal exposure to circulating Mn²⁺ and thereby potentially inflating uptake. Conversely, age-related increases in Mn²⁺ uptake via L-VGCCs may be harder to detect in regions like the parietal cortex, basal ganglia, and superior colliculus, which reportedly undergo age-related decreases in BBB permeability (Goldman et al., 1992). One advantage of the present study’s narrow focus on the hippocampus is the widely-reported stability of hippocampal BBB permeability throughout healthy aging (Goldman et al., 1992; Rapoport et al., 1979; Saija et al., 1992; Sankar et al., 1983; Öztas et al., 1990). Non-neuronal factors besides BBB permeability, including imaging methods and age-related changes in hemodynamics or systemic availability of Mn²⁺, are somewhat easier to address because of their widespread effects. The present finding that there are no age-related changes in Mn²⁺ uptake at the anterior pituitary or muscle (both lack a BBB) argue that the findings in py are not confounded by a change in Mn²⁺ availability or hemodynamics, and are somewhat specific to the selected hippocampal region.

The present study was suggested by a body of literature describing an age-related increase in L-VGCCs on hippocampal neurons (Thibault and Landfield, 1996; Veng and Browning, 2002), which we hypothesized would increase Mn²⁺ uptake. Of note, MEMRI has previously been used to measure neuronal activity (e.g., Bissig and Berkowitz, 2011) based on the preferential opening of L-VGCCs in active versus inactive neurons. For this reason, it was important to consider the potential influence of neuronal activity on the present results. Anesthesia could change neuronal activity, and because each rat was anesthetized for a retinal imaging protocol (Bissig and Berkowitz, 2013) completed immediately prior to brain imaging, we considered the possible influence of protocol anesthesia and timing on the present findings. The period of anesthesia prior to brain scans was very consistent between age groups, with mean durations falling between 2.5 and 2.6 hr, making it an unlikely confound on the present findings. Moreover, it is likely that hippocampal Mn²⁺ uptake reached plateau in the hours between Mn²⁺ injection and the start of anesthesia (Lee et al., 2005), and the present measurements would therefore reflect uptake in the period when rats were awake and free-moving. Because py RVs were not correlated with the total time between Mn²⁺ injection and scanning, we find it unlikely that the somewhat longer duration in old adults influenced results. We also note that total time between Mn²⁺ injection and scanning was very similar between young and mid-adults, and both cross-sectional and longitudinal comparisons of Mn²⁺ uptake showed a robust increase in py uptake in that period of adulthood, consistent with previous work (Karst et al., 1997). Although we noted no gross age differences in behavior in the time between Mn²⁺ injection and anesthesia, it is difficult to rule out subtle behavioral differences without objective measures (e.g. video recording). We nevertheless find it unlikely that the age-related ~10% increase in py RVs can be explained by changes in neuronal activity: Activity-related Mn²⁺ uptake appears to alter brain RVs by a more modest ~3% (visual system; measured using identical image parameters; Bissig and Berkowitz, 2011), and, critically, baseline rates of adult rat CA1 neuronal activity appear unaffected by aging (Barnes et al., 1983).

Our study design utilized both cross-sectional and longitudinal comparisons. One of the concerns in longitudinal MEMRI studies is the potential for Mn²⁺ toxicity, which has been reported at much higher doses than used in the present work (Silva et al., 2004). To test whether a history of Mn²⁺ exposure affected results, we compared 7 mo rats that had previously been injected with Mn²⁺ (at 2.5 mo) to those that were first studied at 7 mo. We found no differences in RVs at muscle, pituitary, or py, nor in measurements of brain size, suggesting that a single dose of 44 mg MnCl₂∙4H₂O / kg (i.e., 0.22 mmol Mn²⁺/kg) is benign. This is consistent with previous reports showing no effect on a sensitive behavioral measurement of hippocampal function following a similar dose (0.20 mmol Mn²⁺/kg; Jackson et al., 2011), nor histological signs of hippocampal toxicity from a higher dose than used presently (0.50 mmol Mn²⁺/kg, Eschenko et al., 2010). The apparently lack of toxicity with the present dose of Mn²⁺ suggests the possibility of future human MEMRI studies aimed at testing the calcium hypothesis of aging. Despite the apparent safety of low-dose MnCl₂ as a human MRI contrast agent for some organs (e.g., for liver ~0.05 mmol Mn²⁺/kg orally (Albiin et al., 2012), or for heart 0.005 mmol Mn²⁺/kg intravenously (Fernandes et al., 2011)), we are not aware of any study exploring use of this compound in human brain imaging. Human studies are nevertheless plausible using the FDA-approved Mn²⁺-based contrast agent Teslascan (Mangafodipir trisodium, Mn-DPDP): In rats, we have previously shown human-safe doses of Teslascan are sufficient to detect activity-dependent differences in Mn²⁺ uptake at the neural retina (Tofts et al., 2010). If human MRI studies can demonstrate a similar level of central nervous system Mn²⁺ uptake using Teslascan, it should be possible to non-invasively test the calcium hypothesis of aging in humans.

Although we did not investigate the underlying molecular causes of the age-related increase in CA1 ion influx, it is plausibly related to an age-related increase in expression of the α_1D L-VGCC isoform at CA1 (Veng and Browning, 2002): We recently reported a similar age-related increase in α_1D expression at the retina that coincided with increased Mn²⁺ uptake (Bissig et al. 2013). Further study is warranted to investigate whether the hippocampus and retina share common mechanisms of aging. The present findings lay the foundation for future MRI studies using Mn²⁺ as a biomarker to non-invasively explore the natural history of, and interventions aimed at altering, age-related changes in neuronal Ca²⁺ regulation.

Abbreviations:

MEMRI

manganese-enhanced magnetic resonance imaging

L-VGCC

L-type voltage-gated Ca²⁺ channel

MPRAGE

magnetization prepared rapid acquisition gradient echo

PDGE

density weighted gradient echo

RV

ratio value

SI

signal intensity

dcw

deep cortical white matter

alv

alveus

ori

oriens layer

py

pyramidal layer

rad

radiatum layer

lmol

lacunosum moleculare

GEE

generalized estimating equation