Long term potentiation, but not depression, in interlamellar hippocampus CA1

Synaptic plasticity in the lamellar CA3 to CA1 circuitry has been extensively studied while interlamellar CA1 to CA1 connections have not yet received much attention. One of our earlier studies demonstrated that axons of CA1 pyramidal neurons project to neighboring CA1 neurons, implicating information transfer along a longitudinal interlamellar network. Still, it remains unclear whether long-term synaptic plasticity is present within this longitudinal CA1 network. Here, we investigate long-term synaptic plasticity between CA1 pyramidal cells, using in vitro and in vivo extracellular recordings and 3D holography glutamate uncaging. We found that the CA1-CA1 network exhibits NMDA receptor-dependent long-term potentiation (LTP) without direction or layer selectivity. By contrast, we find no significant long-term depression (LTD) under various LTD induction protocols. These results implicate unique synaptic properties in the longitudinal projection suggesting that the interlamellar CA1 network could be a promising structure for hippocampus-related information processing and brain diseases.

In vivo Field Recording. C57BL/6 J mice (6-12 weeks old) were anesthetized with Urethane (0.06 g per 25 g weight) and supplemented with Atropine. After a craniotomy, recording electrodes were placed above the dorsal hippocampus. Stimulation electrodes were placed lateral to the recording electrodes in a longitudinal direction. Electrodes were lowered to hippocampal CA1 layer according to the response of characteristic single unit and evoked postsynaptic potentials. Local field potentials were sampled at 24 kHz and filtered 500 Hz with a Tucker-Davis Technologies system. I-O curve was performed to determine half the maximal stimulus intensity of evoked field excitatory postsynaptic potentials (fEPSPs). LTP was induced with currents eliciting 75% of the maximal EPSP response by high frequency stimulation (HFS, four 100 Hz pulse, 10 s interval); LTD was induced with low frequency stimulation (1 Hz pp-LFS; with or without 50 ms paired-pulse intervals, 900 pairs of stimuli during 15 min). Response change was measured in % of the slope baseline. After recording, 10 μA current was passed through both recording and stimulating electrodes for 30 s to lesion the recorded areas. Then, the mice were trans-cardially perfused with 4% PFA. The brains were harvested, sliced and stained with Cresyl Violet. Slice preparation. The mouse brains were quickly removed and placed into chilled (4 °C), oxygenated (5% CO 2 and 95% O 2 ) slicing medium containing the following ingredients: 212 mM sucrose, 5 mM KCl, 1.23 mM NaH 2 PO 4 , 26 mM NaHCO 3 , 11 mM glucose, 1.5 mM MgCl 2 , and 2.5 mM CaCl 2 . Transverse slices (300-400 μm) were cut orthogonal to the septo-temporal axis of the hippocampus, whereas longitudinal slices (300-400 μm) were cut parallel to the septo-temporal plane (Fig. 1a). The slices were then stored, submerged in aCSF at room temperature for 1-2 h before transferring them to the recording chamber.
In vitro Field Recording. Transverse and longitudinal hippocampal slices were transferred from the recovery beaker to the recording chamber, where they were submerged in aCSF (30 °C) with a flow rate of 2 ml/min. For the transverse slices, extracellular field potentials were recorded in the CA1 region using glass electrodes containing NaCl (3 M). A stimulating electrode in CA2 (Schaffer collateral pathway) was used to evoke field EPSPs (constant voltage, 100 μs duration, repeated at 30 sec intervals). For the longitudinal slices, recording and stimulating electrodes were placed at stratum oriens (S.O.) or stratum radiatum (S.R.) along the longitudinal axis to record extracellular field potentials and evoke field EPSPs. LTP was induced by two trains of tetanus stimuli (HFS; 100 Hz, 1 sec, repeated after a 30 sec interval). LTD experiments were performed with 4 protocols: 5 Hz LFS (900 stimuli during 3 min), 1 Hz LFS (900 stimuli during 15 min), 1 Hz paired-pulse LFS (50 ms paired-pulse interval, 3D digital holography uncaging. The procedures for digital holographic photolysis have been described in detail elsewhere [18][19][20] . Unlike conventional uncaging techniques, the holographic method permits glutamate photolysis to be directed precisely and simultaneously at multiple sites and depths. Briefly, the holographic beam was brought into the optical axis of an upright fluorescence microscope (Olympus BX51) below the epi-fluorescence unit, with a long-pass dichroic mirror. The output beam of a 150 mW, 405 nm diode laser (CNI Laser) was expanded with a beam expander (3x) to fill the short axis of a reflective spatial light modulator (SLM) (LCOS Hamamatsu, model X10468-05). The SLM plane was projected onto the back aperture of the microscope objective through a telescope (L1, f1 = 750 mm; L2, f2 = 500 mm). The magnification of the telescope was chosen in order to match the SLM short axis with the diameter of the objective's back aperture (Olympus, 60x, W 0.9NA). The undiffracted component (zero order spot) was removed by placing a small (<0.5 mm) anodized metal plate on antireflective coated glass plate at the focal plane of L1. The algorithm for the phase hologram calculation and calibration of the temporal spatial resolution have been described previously 18 . Fresh MNI-caged-L-glutamate (Tocris, Ellisville, MO) at final concentration in physiological solution was prepared each day. All agonists and antagonists were purchased from Sigma (St. Louis, MO) or Tocris (Ellisville, MO).

Statistics.
All data are shown as mean ± standard error (SE). An ANOVA of Fisher's PLSD post hoc test was performed for between-group comparison, while a t-test was done for within-groups (significance, *P < 0.05; **P < 0.01).

Results
In vivo LTPs in the interlamellar CA1 network. In order to test whether the interlamellar CA1 connection shows long-term synaptic potentiation, recording and stimulus electrodes were placed along the longitudinal septal-temporal axis of hippocampus CA1 in anesthetized animals. Loci of recording (in a septal side) and stimulating (in a temporal side) electrodes were identified by lesion marks after completion of the electrophysiological recordings (Fig. 1a). 100 Hz HFS potentiated the synaptic responses in 10 unit recordings ( Fig. 1b; 124 ± 4%, t-test, t = −4.79, p = 0.001). This result demonstrates that LTP exists in the interlamellar network of live animals. In vitro LTPs in interlamellar CA1 network. To further investigate direction-and layer-specific LTP within the interlamellar plane, longitudinal brain slices were prepared at a nearly perpendicular angle of the transverse slices according to our established protocol 12 . Synaptic responses were monitored with field recordings in both transverse and longitudinal slices (Fig. 2a). Longitudinal LTPs, induced by employing 100 Hz HFS in the area of the stratum radiatum (S.R.) of CA1 pyramidal neurons, were comparable to transverse LTPs (Fig. 2b, longitudinal LTP, 144 ± 5%, n = 24; transverse LTP, 159 ± 11%, n = 9). Next we tested, whether longitudinal LTPs are selective for direction or layer. Synaptic responses were induced by placing a stimulating electrode on either the temporal or the septal side of the stratum radiatum (SR). Electrical activation of both sides induced LTP without a significant difference between the two directions ( Fig. 2c; temporal: 146 ± 5%, n = 12; septal: 143 ± 7%, n = 12; P > 0.1). Similar to LTPs stimulated in S.R., LTPs observed in stratum oriens (SO) had no directional specificity, suggesting an absence of layer selectivity ( Fig. 2d; temporal: 160 ± 6%, n = 10; septal: 147 ± 6%, n = 9; P > 0.1).

LTPs in CA1-CA1 pyramidal cells.
To confirm that LTPs in CA1-CA1 pyramidal cells is mediated by direct interaction between longitudinally connected excitatory neurons, we combined glutamate uncaging with a digital hologram system in longitudinal slices (Fig. 3a). After establishing whole-cell recording on a CA1 pyramidal neuron, we induced glutamate-mediated synaptic responses of the neuron by photostimulation of caged glutamate on neighboring pyramidal neurons. We monitored the uncaging-induced synaptic responses before and after the same 100 Hz tetanic stimulation we used for in vivo unit and in vitro field recordings. Electrical tetanic stimulation potentiated the uncaging-induced synaptic responses to 135.60 ± 11.03% (Fig. 3b; n = 10 neurons, P < 0.05). These data indicate that excitatory CA1-CA1 synaptic connection mediates postsynaptic long-term synaptic plasticity.

Discussion
The CA1 region on the transverse axis of hippocampus is a common model for studying the cellular basis of cognitive functions. Research into CA1 function has produced an extraordinarily rich body of knowledge on hippocampus physiology 3 . Yet, it has remained unclear, whether CA1 efferent fibers project along the long axis of hippocampus 36,37 and that being the case, whether they play a role in synaptic plasticity. Here we show that the longitudinal network exhibits robust LTP but no significant LTD under many known LTD induction protocols.
The synaptic potentiation of interlamellar CA1 network could contribute to hippocampus-related memory. Hampson et al. 8 carried out multi-electrode recordings in the hippocampus of rats that were performing  the septotemporal axis of CA1. Also, hippocampal sharp wave and ripple (SPW/R) signals, which play an essential role in the formation and consolidation of episodic memory, are known to be generated and propagated along the CA1 longitudinal connectivity 38,39 . Consistent with these findings, the longitudinal projection may serve as an anatomical substrate of this behaviorally relevant signal processing 12 .
The mechanism regarding the absence of LTD in longitudinal network is not clear. It has been known that LTD requires the involvement of both presynaptic (such as CB1 receptors and calcineurin) and postsynaptic mechanisms (such as NMDA receptors, mGluRs, phospholipase C, and endocannabinoid synthesis) 21,40,41 . For example, when mGluRs and CB1 receptors are blocked, this leads to a suppression of lamellar LTD in vitro 42,43 and in vivo 44,45 . We speculate that the absence of LTD in the interlamellar connection is due to the absence or lack of these LTD inducing machineries. Besides the molecular mechanisms, the absence of LTDs can be explained by structural factors such as an axon properties (thick axonal diameter and low varicosity density in the transverse projection vs. thin axonal diameter and high varicosity density in the longitudinal projection 12 . For now, our finding just showed favorable LTPs but not LTDs in the longitudinal projection, presumably being involved in strong connectivity and hyperexcitability. Longitudinally oriented synaptic networks were identified as potential foci in patients with hippocampal epilepsy. Seizure activity was found to synchronize along the entire length of hippocampus via the longitudinal projection 46 . Transverse cuts that severed the longitudinal pathway in human hippocampus are an effective treatment to prevent hippocampal seizures from spreading 46,47 . Both the length and number of axon collaterals of CA1 neurons are increased in rats that experience chronic limbic seizures following kainic acid or pilocarpine application 48,49 . These results implicate that epileptiform activity may travel along the longitudinal axis of hippocampus. Based on ours and other studies, we speculate that the thin axonal diameter (probably having high impedance) and high varicosity density can be favorable for fast signal spread in multiple spots simultaneously. These axonal properties can promote a prompt spread of epileptiform activity and enhance memory function. However, we await further researches concerning favorable LTPs but not LTDs in the longitudinal networks. In conclusion, we pose that the CA1 longitudinal network is a promising candidate to investigate memory-related information processing and brain diseases.