Strong CA2 pyramidal neuron synapses define a powerful disynaptic cortico-hippocampal loop

Abstract

Neurons propagate information through circuits by integrating thousands of synaptic inputs to generate an action potential output. Inputs from different origins are often targeted to distinct regions of a neuron's dendritic tree, with synapses on more distal dendrites normally having a weaker influence on cellular output compared to synapses on more proximal dendrites. Here, we report that hippocampal CA2 pyramidal neurons, whose function has remained obscure for 75 years, have a reversed synaptic strength rule. Thus, CA2 neurons are strongly excited by their distal dendritic inputs from entorhinal cortex but only weakly activated by their more proximal dendritic inputs from hippocampal CA3 neurons. CA2 neurons in turn make strong excitatory synaptic contacts with CA1 neurons. In this manner, CA2 neurons form the nexus of a highly plastic disynaptic circuit linking the cortical input to the hippocampus to its CA1 neuronal output. This circuit is likely to mediate key aspects of hippocampal-dependent spatial memory.

Figure 1. Identification of CA2 pyramidal neurons

A: Typical firing of CA1, CA2 and CA3 pyramidal neurons in response to a 1 second depolarizing step. Inset shows an expanded trace to illustrate that the slow hyperpolarizing potential following the action potential in CA1 neurons (arrow) is lacking in CA2 neurons. B: Voltage response to a 1 second hyperpolarizing step (from -70 to -100 mV). CA1 neurons displayed a much larger depolarizing sag (mediated by activation of Ih) than CA2 or CA3 neurons. C: Top left: Low power fluorescence image of a CA2 pyramidal neuron (red arrow) filled with Alexa 594 during whole cell recording. Bottom left: Higher power fluorescence image showing same neuron filled with Alexa 594 (red) with superimposed immunolabeling for -actinin 2 (green), a protein enriched in CA2 neurons. Right: Higher magnification view of same neuron showing typical morphological features of a CA2 neuron, including the presence of two apical dendrites with a bifurcation point near the soma, very few oblique dendritic branches in SR, and a dense branching in SLM. SO: Stratum Oriens; SP: Stratum Pyramidale; SR: Stratum radiatum; SLM: Stratum Lacunosum Moleculare. See also Figure S1.

Figure 2. Distal Layer III EC inputs strongly depolarize CA2 neurons but are mostly inhibitory in CA1 neurons

A: Top: Arrangement of recording and stimulating electrodes and sample whole cell current-clamp EPSPs recorded from CA1 and CA2 pyramidal neurons in response to stimulation of LIII EC inputs at 4, 8 and 16 V intensities. Bottom: averaged EPSP input-output curves in CA1 and CA2 neurons. Inhibition was blocked with 100 M picrotoxin and 2 M CGP 55845A (PTX/CGP). B: Sample traces and input-output curves obtained with extracellular recordings of fEPSPs in SLM of CA1 and CA2 areas with inhibitory transmission blocked (as above). C: Top: representative synaptic voltage responses from CA1 and CA2 pyramidal neurons in response to LIII stimulation before (Control) and after GABA receptor blockade (PTX/CGP). The inhibitory postsynaptic potential (IPSP) component of the response was isolated by taking the difference between control responses and pure EPSPs obtained in the presence of PTX/CGP. Bottom: Averaged IPSP input-output curves in CA1 and CA2 neurons. D: Summary graph of the EPSP/IPSP amplitude ratio from the populations of cells shown in (C). All synaptic responses were recorded with membrane potential held at -73 mV using constant current injection. Error bars show SEM. See also Figure S2

Figure 3. Proximal SC inputs efficiently depolarize CA1 neurons but provide a large inhibitory drive in CA2

A: Sample EPSPs and input-output curves in CA1 and CA2 neurons in response to stimulation of SC inputs with inhibition blocked by picrotoxin and CGP 55845A. Data are not shown for stimulation intensities above 16 V for CA1 neurons because the large EPSPs triggered spikes. B: Sample traces and input-output curves obtained for extracellular recordings of fEPSPs in SR of CA1 and CA2 with inhibition blocked. C: Top: Sample traces of synaptic responses obtained before (control) and after GABA receptor blockade (PTX/CGP). The IPSP was isolated by subtracting traces in the presence of PTX/CGP from control traces. Bottom: Averaged IPSP input-output curves in CA1 and CA2 neurons. D: Summary graph of the EPSP/IPSP amplitude ratio from cells recorded in (C). All synaptic responses were obtained with resting membrane potential set at -73 mV using constant current injection.

Figure 4. LIII EC inputs drive spiking in CA2 but not CA1 pyramidal neurons

A–C: Somatic voltage responses of CA1 and CA2 neurons to a brief burst of stimuli applied to LIII EC inputs. A: Top: Sample whole-cell voltage recordings in CA1 and CA2 pyramidal neurons in response to a brief train of stimuli to LIII EC inputs (5 stimuli at 100 Hz), recorded in control conditions or in presence of PTX/CGP. B: Probability of action potential firing for CA1 and CA2 pyramidal neurons as a function of EPSP number during the 5-stimuli bursts using 10 or 20 V stimulus strengths (inhibition was intact). C: Probability of eliciting an action potential during a 5 pulse burst in presence of PTX/CGP. D–F: Somatic voltage responses of CA1 and CA2 neurons to a burst of 5 stimuli applied to SC inputs. Panels and other conditions as in A-C. Experiments performed at initial resting membrane potential (−70.5 mV for CA1 and−72.4 mV for CA2 in control; −67.8 mV for CA1 and −67.7 mV for CA2 in PTX/CGP).

Figure 5. LII and LIII EC inputs converge onto CA2 pyramidal neurons

A: Left: Position of stimulating electrode in the middle molecular layer of DG and whole recording pipette in CA2. 1. Sample whole-cell voltage responses from a CA2 neuron following stimulation of LII axons. Synaptic responses under control conditions (black trace, no blockers) or after blocking inhibition (PTX/CGP, grey trace). Note polysynaptic EPSPs with inhibition blocked (arrows). The polysynaptic response in PTX/CGP (grey trace) was abolished after addition of the Type II mGluR agonist DCG-IV (1 μM, black trace). A2. Extracellular recordings of the fEPSP from SLM of CA2 (grey trace) show that LII stimulation in DGelicited an early negative voltage response (current sink) followed by a delayed, positive response (current source). Only the late positive response was abolished by DCG-IV application (black trace). B Left: Whole cell EPSPs recorded from CA2 or CA3 pyramidal neurons in response to 10, 15 and 20 V intensity stimulation of LII inputs in DG. Right: Mean EPSP input-output curves for CA2 and CA3 pyramidal neurons in response to stimulation of LII inputs. Inhibition was blocked with PTX/CGP. C. Schematic drawing illustrating the location of the different hippocampal areas. LII and LIII EC inputs were stimulated independently and extracellular field responses were recorded in the somatic layer at different regions of the hippocampus. D. Sample traces recorded in the different hippocampal areas following LII and LIII stimulation. Left, LII stimulation evoked the largest response in CA2 with a smaller response in CA3 and little response in CA1. Positive extracellular voltage responses in the somatic layer is caused as negative synaptic current entering the dendrite leaves the soma as positive membrane current. Right, LIII responses are largest in CA2, with a smaller response in CA1 and little or no response in CA3. E. Plot of mean somatic fEPSP amplitude in different hippocampal subfields in response to stimulation of LII and LIII inputs. Inhibition was intact. Error bars show SEM.

Figure 6. Integration of LII and LIII cortical inputs onto CA2 pyramidal neurons

A: Sample voltage responses (left) and mean whole cell voltage responses (right) evoked in CA2 neurons following stimulation of LII or LIII independently (white and grey circles, respectively) or simultaneously (black circles). The predicted linear sum is shown by triangles. Inhibition was intact. B: Sample voltage traces (left) and summary graph (right) of whole cell EPSP amplitudes following paired stimulation of LII and LIII (~5 V stimulus) at different intervals (time of stimulation of LIII minus time of stimulation of LII) before (Control, white circles) and during block of inhibition (PTX/CGP, black circles). The second EPSP was normalized to the amplitude of the first EPSP.

Figure 7. Robust LTP is present at cortico-CA2 but not CA3-CA2 synapses

A: Stimulation of SC inputs induces LTP of the extracellular fEPSP in CA1 but only transient potentiation in CA2. Left: Experimental set-up showing stimulating electrode in middle of SR in CA1 midway between extracellular field recording electrodes placed in SR of CA2 and CA1. Sample traces of fEPSPs recorded simultaneously in CA1 and CA2 before and after delivery of tetanic stimulation to SC inputs are shown below. Right: summary graph of normalized fEPSP slope simultaneously recorded from CA1 (open circles) and CA2 (filled circles) before and after delivery of tetanic stimulation (at arrow; n = 9). B: Stimulation of LIII inputs produces larger LTP in CA2 than CA1. Left: Experimental set-up showing stimulating electrode placed in middle of SLM midway between two field recording electrodes placed in SLM of CA2 and CA1. Sample fEPSPs in CA1 and CA2 in response to stimulation of LIII EC inputs before (black) and after (grey) induction of LTP are shown below. Right: Summary graph showing normalized fEPSP responses before and after delivery of tetanic stimulation to LIII inputs (n = 7 and 9 for CA1 and CA2, respectively, with 6 simultaneous recordings). C: Stimulation of LII inputs produces larger LTP in CA2 than CA3. Left: Experimental set-up showing stimulating electrode placed in middle molecular layer of DG and two field recording electrodes placed in SLM of CA3 and CA2. Sample CA2 and CA3 fEPSPs before (black) and after (grey) induction of LTP are shown below. Right: Summary graph of normalized CA2 and CA3 fEPSPs in response to stimulation of LII EC inputs, before and after tetanic stimulation (n = 6 simultaneous recordings). All experiments performed in PTX/CGP. Error bars show SEM. See also Figure S3 and S4

Figure 8. Unitary recordings from pairs of synaptically connected CA2 and CA1 neurons reveal multiquantal responses

A: Antidromic action potentials are recorded extracellularly in CA2 and CA3 cell body layers in response to extracellular stimulation in SO or SR of CA1. Left, Experimental set-up showing placement of field recording electrodes in CA2 and CA3 cell body layer and stimulating electrodes in SO (electrode S1) and SR (electrode S2) of CA1. Both excitatory and inhibitory transmission were blocked with CNQX/APV and PTX/CGP, respectively. Right, Stimulation through electrode S1 in SO led to an extracellular population spike (PS, following stimulus artifact at arrow) in CA2 but not CA3. Stimulation through electrode S2 in SR led to a PS in both CA2 and CA3. B: Left, Experimental set up showing placement of patch pipettes during dual whole cell recordings from CA2 and CA1 pyramidal neurons. Right, Inset, Whole-cell voltage-clamp currents in a CA1 neuron showing two EPSCs (successes) and one failure in response to the firing of single spikes in the CA2 neuron (not shown). CA1 EPSC amplitude histogram shows a prominent peak of failures at 0 pA and peaks of successes at approximately −4 and −8 pA. CA1 neuron was held at -73 mV. C: EPSC amplitude histogram from a second paired whole-cell recording between a presynaptic CA2 and postsynaptic CA1 neuron, showing 4–5 peaks of EPSC successes at integral multiples of −5 pA. Cell held at −73 mV. D: Plot of EPSC failure rate and potency (average EPSC amplitude of a success) in CA1 neurons in response to firing a single action potential in a synaptically connected CA2 neuron. Black circles show values for individual experiments and white circles give means. Values for CA3-CA1 paired recordings obtained in similar conditions (McMahon et al., 1996) are shown for comparison (grey triangle). Error bars show SEM.

Figure 9. Direct evidence for cortico-CA2-CA1 loop

A: Stimulation of LIII inputs in the SLM of CA1 elicits spikes in CA2 that trigger a suprathreshold EPSP in CA1. Bottom: Experimental set-up showing stimulating electrode in SLM of CA1, extracellular field recording electrode in the CA2 cell body layer, and the whole cell recording electrode in the CA1 cell body layer. Top: A brief train of stimuli to LIII (3 at 100 Hz) elicited a PS burst in CA2 and a polysynaptic EPSP in CA1. Response in CA2 is shown for a 30 V stimulus. Upward deflections are stimulus artifacts and downward deflections are extracellular PS (arrows). Intracellular voltage recordings from a CA1 neuron show that: a 15 V stimulus elicits a monosynaptic response (light grey trace); a 25 V stimulus elicits a subthreshold polysynaptic response (dark grey trace); and a 30 V stimulus elicits a suprathreshold polysynaptic response (black trace). Experiment performed in PTX/CGP. To avoid any contribution from SC CA3 inputs, the slice was cut between CA3 and CA2. B: Experiment similar to that in (A) except that the slice was cut between CA2 and CA1 from SO to SR (SLM was kept intact). Note the lack of suprathreshold polysynaptic response in the CA1 neuron at high stimulation intensities (40 and 60 V) that elicited pronounced spike firing in CA2.