2. LA EVALUACIÓN EDUCATIVA COMO COMPONENTE CURRICULAR EN LA TRANSFORMACIÓN DE LAS
2.4. Hacia la construcción de una evaluación crítica y formativa de los aprendizajes de los
Here we discuss the evidence for the existence of both phasic inhibition and excitation and our results on their phase relationships.
Phasic inhibition Our results challenge the prevailing view that sharp wave-asso-
ciated ripples are shaped by phasic synaptic inhibition alone. This view is based on several observations about ripples in vivo:
1. At the intracellular level, using sharp microelectrode recordings on CA1 pyra- midal cells in anesthetized rats, Ylinen et al. (1995) varied the pipette Cl−con-
centration and showed that ripple-associated postsynaptic potentials displayed phase shifts as expected from inhibitory PSPs. 2. In the same study, a current source density (CSD) profile was produced from an extracellular shank electrode that showed ripple modulation in the pyramidal layer. Since peri- somatic afference to CA1 pyramidal cells is mostly inhibitory and the LFP patterns are widely considered to be governed by local synaptic activity, the observed extracellular ripple was suggested to be caused by the inhibitory inputs to the soma. Hindering dissection of the actual responsible, the observed current source could be indeed active, due to inhibitory-driven efflux, or pas- sive, due to the compensation of a concurrent active sink at the dendrites caused by excitatory-driven ion influx there.
3. Third, extracellular recordings in vivo revealed that soma-targeting interneu- rons increase their discharge rate during ripples and fire rhythmically with the network oscillation (Csicsvari et al., 1999b). The increase in rate is still only about half of that for pyramidal neurons (factor 3.8 vs. 8.6), but a larger pro- portion of the interneuronal population is recruited during ripples (about 60% vs. 30%, Csicsvari et al., 2000). Interneurons start firing earlier and cease later in the ripple (Csicsvari et al., 1999b). More importantly, single putative interneurons fire ripple-locked at high frequencies, whereas the paucity of pyramidal firing makes it necessary to conduct any locking analysis at the pop- ulation level, or on extremely long recordings. These studies also showed, how- ever, that pyramidal rates couple better to ripple amplitude (in SD of baseline activity) than interneuronal rates do (see in particular Csicsvari et al., 1999a, Fig. 3).
4. Ripple-locked firing of soma-targeting basket cells has been confirmed more recently by juxtacellular recordings with post hocmorphological reconstruc- tions in anaesthetized rats (Klausberger et al., 2003).
Phasic excitation Phasic excitation is to be expected from the abundant reports using
in vivo multielectrode recordings, which demonstrate that pyramidal cells, as a pop- ulation, fire phase-locked to ripples. Our study adds that phasic excitatory inputs at
∼200 Hz are reliably expressed and prominent during ripples as seen from other pyramidal cells. Further, they are effective in regulating spike timing. This is demon- strated by the ripple-locked spiking of cells whose inhibitory inputs have been blocked pharmacologically with DNDS. Phasic involvement of the majority cell population in the hippocampus (>90%; Buhl and Whittington, 2006) does not come as a surprise. Indeed, while oscillations are theoretically possible in an inhibitory-only population
when driven by a stationary excitatory forcing, it is likely that in a mixed population with reciprocal connections both types of cells entrain each other into the rhythm (Brunel, 2000; Isaacson and Scanziani, 2011).
Preferred phase of excitatory and inhibitory inputs It would be tempting
to compare our results on preferred current phases with the in vivo reports of phase locking of individual pyramids (Buzsáki et al., 2003; Csicsvari et al., 1999b; O’Neill et al., 2006) and interneurons (Klausberger and Somogyi, 2008). Unfortu- nately, the times are incomparable without additional knowledge: different delay lines are involved in carrying an emitted spike to an extracellular electrode and to a the soma of a voltage-clamped postsynaptic cell. Those delays include the axonal prop- agation delay, the synaptic transmission delay and the electrotonic ramp-up of the current inside the postsynaptic neuron all the way to the soma. Additionally, the phases of steep slopes used in our locking analyses are delayed by ca. 1 ms with respect to the fitted onsets. These considerations prevent direct comparisons. At most, phase differences of excitation and inhibition could be compared between our analysis and the extracellular electrode-based literature, but this would still require assuming that the delays are similar for both types of input.
Interplay of excitation and inhibition during the ripple We compared our results
at the reversal of inhibition(−66 mV)to those at higher holding potentials(−45 mV), reaching up to the reversal of inhibition (−6 mV; Cs-based solution). We found
that, as potentials were increased, the polarity of the currents flipped from inward to outward in a cell-dependent manner. Those cells that showed a partial polarity change at an intermediate holding potential systematically exhibited an inward-signed overall current first, followed by an outward current5.1 with the somewhat longer tail-off also seen in fully inhibitory traces. This is reminiscent of the gradual tilt towards inhibition observed in sensory cortices in response to impulse-like stimuli (Isaacson and Scanziani, 2011).
To refine this initial observation we improved on our relatively assumption-free slope-based technique to detect PSCs embedded in cPSCs by developing a time- domain iterative reconstruction method with two stages. Detection, first, is performed by deconvolution. Peeling, second, exploits the information garnered from extrap- olating the uncorrupted section of the first incoming PSC to subtract it altogether from the remaining trace. Using this technique, we were able to decompose the vast majority of the monophasic cPSCs (excitatory or inhibitory) into individual PSCs, and collect statistics on their amplitude, kinetics, and timing by ripple cycle. The large ampli- tudes of PSCs observed around the SWR peak could be explained either by superlinear time integration (whereby single presynaptic partners see their input amplified when it occurs in fast sequence) or by the well-orchestrated coactivation of a presynaptic
5.1. This is somewhat similar to the “biphasic” events reported by Ellender et al. (2010) in their Figure 3 albeit with the caveat that we study mouse, CA1 and currents that follow a consistent pattern for each cell whereas these authors recorded in CA3 of the rat the membrane potential of one pyramidal cell expressing EPSPs, IPSPs or mixed PSPs in a single recording. They also observed a distance-dependent ratio of excitation to inhibition (their Fig. 3 H), with excitation predominant for sharp-waves originated close to the patched cell, which would be in line with our observations. In general results like those of Ellender et al. seem to beg for a study of the synaptic composition of sharp-wave ripples that takes into account the coupling between location in the slice and time along the ripple that is induced by the propagation of the ripple (Csicsvari et al., 2000; our own observations indicating lags of at most one ripple cycle between distant locations in mouse CA1).
assembly. According to the assembly hypothesis, virtually every pyramidal cell would be capable of sensing the activation of any assembly, but only selected ones belonging in the sequential schedule of activation would be driven strongly enough to surmount the shunt inherent to the SWR-induced high-conductance, and fire.
The kinetics of the in-ripple and spontaneous PSCs were largely compatible and homogeneous across cells, which supports the view that cPSCs are built from PSCs drawn from the network-wide pool of afference.
Finally, ripple-associated excitatory and inhibitory currents express an exquisite temporal precision and converge in phase, as inhibitory phasic input reduces its ini- tial lag with respect to excitation over the course of ripples. This may be interpreted as a quenching mechanism. Indeed, with recurrent excitation, synaptic or otherwise, there exists the risk that activity avalanches out of control into an epileptic-like state. Interneurons could switch from emphasizing the oscillation at the beginning by firing in phase opposition, to bridling it by discharging in phase alignment towards the end. This phenomenon could help explain the partial polarity reversal noted above for some cells at an intermediate holding potential. The diversity of interneuronal ripple phase-tuning according to cell phenotype (Klausberger and Somogyi, 2008) could be instrumental in achieving this staggeringly precise phase drift of about a quarter of a cycle over a few cycles. The dampening of the ripple has been so far thought to occur passively as the depolarization from CA3 wears off. Since our data shows that ripples are also tran- sient in the minislice, i.e., in absence of CA3 input altogether, the active interplay and phase alignment of excitation and inhibition appears as an alternative or complement to explain their limited duration.
Note that the short delays between excitation and inhibition are in principle com- patible with both feedforward (excitatory afference incoming to both pyramids and interneurons) and feedback inhibitory circuits (local interneurons fed only by local pyramidal cells). However, in a very simplistic model where axonal delays in the local circuit would be of the same order for excitatory and inhibitory cells one would expect feedforward inhibition to be faster by one delay with respect to feedback inhibition. This runs counter to the observed initial delay of inhibition. Notwithstanding and since the real situation is much more complex, it cannot be excluded that the network archi- tecture provides yet another route to explain the observed synchronization.