New insights into the regulation of synaptic plasticity from an unexpected place: hippocampal area CA2

Abstract

The search for molecules that restrict synaptic plasticity in the brain has focused primarily on sensory systems during early postnatal development, as critical periods for inducing plasticity in sensory regions are easily defined. The recent discovery that Schaffer collateral inputs to hippocampal area CA2 do not readily support canonical activity-dependent long-term potentiation (LTP) serves as a reminder that the capacity for synaptic modification is also regulated anatomically across different brain regions. Hippocampal CA2 shares features with other similarly "LTP-resistant" brain areas in that many of the genes linked to synaptic function and the associated proteins known to restrict synaptic plasticity are expressed there. Add to this a rich complement of receptors and signaling molecules permissive for induction of atypical forms of synaptic potentiation, and area CA2 becomes an ideal model system for studying specific modulators of brain plasticity. Additionally, recent evidence suggests that hippocampal CA2 is instrumental for certain forms of learning, memory, and social behavior, but the links between CA2-enriched molecules and putative CA2-dependent behaviors are only just beginning to be made. In this review, we offer a detailed look at what is currently known about the synaptic plasticity in this important, yet largely overlooked component of the hippocampus and consider how the study of CA2 may provide clues to understanding the molecular signals critical to the modulation of synaptic function in different brain regions and across different stages of development.

Figure 1.

Hippocampal area CA2 is often excluded from circuit diagrams illustrating the flow of information through the hippocampus. (A) The simplified schematic diagram is of a typical coronal section through the dorsal hippocampus and highlights how area CA2 (shown in red) fits within the traditional “trisynaptic” circuit. Arrows indicate the primary direction of information flow through the circuit (D, dorsal; L, lateral). Granule cells in the dentate gyrus (DG, purple) form synapses on pyramidal cells in area CA3 via the mossy fibers. In turn, CA3 neurons (green) synapse with neurons in both CA2 and CA1 (blue) via the Schaffer collaterals. CA2 pyramidal neurons project mainly to the stratum oriens of area CA1 and synapse onto the basal dendritic arbors of CA1 pyramidal cells. Note, CA2 axons also branch and project to the stratum radiatum in CA1, as well as the stratum oriens of proximal CA3, though these projections are less dense (shown in B). (B) Induction of activity-dependent LTP differs along the proximal-distal axis of CA2 pyramidal neurons. Schaffer collateral projections to the proximal dendritic compartment of CA2 neurons do not typically support activity-dependent LTP (green arrow, no LTP). This is in contrast to temporoammonic inputs from the entorhinal cortex to the distal dendrites of CA2 neurons which readily express LTP (black arrow, LTP). (so) Stratum oriens; (sp) stratum pyramidale; (sr) stratum radiatum; (slm) stratum lacunosum moleculare.

Figure 2.

The lack of activity-dependent LTP at Schaffer collateral synapses in area CA2 is due mainly to robust calcium handling in CA2 pyramidal neurons. (A,B) Pairing 3-Hz afferent synaptic stimulation of the Schaffer collaterals with the depolarization of CA2 pyramidal cells to 0 mV fails to induce activity-dependent LTP in area CA2. Group data in B show the averaged amplitudes of EPSCs (normalized to baseline) before and after the pairing protocol (arrow) to induce LTP in CA2 neurons. Bars indicate the mean ± SEM in this and subsequent figures. Sample EPSCs shown in A are from the time points marked by the corresponding numbers in B. (C,D) In contrast, the same pairing protocol induces robust LTP in Schaffer collateral inputs to area CA1. The intrinsic biophysical properties of neurons in CA2 differ significantly from those in CA1. Relative to pyramidal cells in area CA1, CA2 pyramidal neurons show greater leak currents at holding potentials less than −60 mV (E, red circles, arrow), as well as have a lower resting membrane potential (F, red bar), higher rheobase current (G, red bar), and lower action potential threshold (H, red bar). Note, stars indicate P < 0.05 in this and subsequent figures. Data modified from Zhao et al. (2007). Differences in intrinsic excitability alone, however, do not account for the lack of LTP in CA2 neurons (data not shown). CA2 pyramidal neurons display higher calcium buffering and extrusion relative to cells in CA1. (I) A temporary increase in the amount of extracellular calcium from 2 mM (gray circles) to 10 mM (red circles) for 3 min (indicated by the red bar) permits induction of LTP in CA2 pyramidal cells following tetanic stimulation (200 Hz HFS; arrow) of the Schaffer collaterals. (J) Loading CA1 neurons with a functional analog of the calmodulin-regulating protein Pep-19 blocks induction of LTP (blue circles). Data modified from Simons et al. (2009).

Figure 3.

Blockade of A₁ receptors potentiates synaptic responses in CA2 but has no lasting effect on synaptic transmission in areas CA1 or CA3. (A) Bath-application of caffeine (100 µM) or the selective A₁R antagonist DPCPX (10 nM) for 5 min potentiates synaptic responses in CA2. The red bar in A marks the onset and duration of caffeine or DPCPX application, and the inset traces show example currents recorded at the latencies marked by the corresponding numbers. Brief exposure to DPCPX increases the volume of spines located on apical dendritic branches of CA2 pyramidal neurons. (B) Two-photon confocal images of a spine-containing segment of secondary apical dendrite for a CA2 neuron loaded with Alexa Fluor 594 are shown before (0 min; baseline) and after (+20 min; washout) 5-min application of DPCPX (10 nM). Arrows in B mark spines that showed a significant change in volume after treatment with DPCPX (calibration bar = 1 µm). (C) Group data for all spines are shown comparing the average change in spine volume at 20-min post-vehicle or post-DPCPX treatment. There was no change in the amplitude of synaptic responses induced by caffeine or DPCPX (D) in areas CA1 or CA3 (E). Conventions in D and E are the same as in A (calibration bars for inset traces in A, D, and E: 50 pA, 25 msec). Caffeine consumption in vivo induces synaptic potentiation in hippocampal CA2 but not in CA1. (F,G) Oral administration of caffeine induces a dose-dependent increase in the amplitude of synaptic responses in CA2 in vitro across a range of stimulation intensities. (H,I) Treatment with caffeine, however, had no effect on responses in CA1. Group data in G and I show average synaptic responses at each stimulation intensity and dose of caffeine tested. Colored bars in G and I indicate the intensity used to evoke the corresponding colored currents in F and H from single neurons in slices from rats dosed with 20 mg/kg caffeine or vehicle (CA2, F; CA1, H). Data modified from Simons et al. (2011).

Figure 4.

Expression of Avpr1b or RGS14 in hippocampal CA2 is linked to social recognition memory and spatial memory, respectively. (A) The photomicrograph illustrates the high expression of the Avpr1b gene in area CA2 determined using in situ hybridization in a sagittal section of mouse brain. (Image source: Allen Brain Atlas, http://mouse.brain-map.org.) Arrows point to the borders between CA2 and adjacent areas CA3 and CA1. (B) Knockout of Avpr1b causes a deficit in social recognition memory. Wild-type mice spend significantly less time investigating a familiar littermate vs. a novel conspecific (gray bars). In contrast, Avpr1b KO mice spend the same amount of time investigating a familiar mouse as they do a novel one (red bars; ns = not significant). Data adapted from DeVito et al. (2009). Knockout of RGS14 in CA2 enhances spatial memory acquisition. (C) Photomicrograph shows the high expression of the RGS14 gene in CA2 using in situ hybridization (http://mouse.brain-map.org). (D) Mice lacking the full-length RGS14 protein learn the Morris water maze task faster than wild-type controls. Although similar on the first day of training, RGS14 KO mice (red circles) find the hidden platform significantly faster on subsequent training days relative to wild-type mice (gray circles). Data adapted from Lee et al. (2010).