|Home | About | Journals | Submit | Contact Us | Français|
Memory and its underlying neural plasticity play important roles in sensory discrimination and cortical pattern recognition in olfaction. Given the reported function of slow-wave sleep states in neocortical and hippocampal memory consolidation, we hypothesized that activity during slow-wave states within the piriform cortex may be shaped by recent olfactory experience. Rats were anesthetized with urethane and allowed to spontaneously shift between slow-wave and fast-wave states as recorded in LFPs within the anterior piriform cortex. Single-unit activity of piriform cortical layer II/III neurons was recorded simultaneously. The results suggest that piriform cortical activity during slow-wave states is shaped by recent (several minutes) odor experience. The temporal structure of single-unit activity during slow-waves was modified if the animal had been stimulated with an odor within that cell’s receptive field. If no odor had been delivered, the activity of the cell during slow-wave activity was stable across the two periods. The results demonstrate that piriform cortical activity during slow-wave state is shaped by recent odor experience, which could contribute to odor memory consolidation.
Spatial memories are replayed within the hippocampal formation during sleep (Pavlides and Winson, 1989; Skaggs and McNaughton, 1996; Louie and Wilson, 2001). Neurons encoding sequences of recently experienced spatial patterns are re-activated during slow-wave sleep (Lee and Wilson, 2002). This replay is hypothesized to facilitate consolidation of spatial memory, as well as help link hippocampal activity to neocortical activity (Buzsaki, 1996; Maquet, 2001; Lee and Wilson, 2002; Steriade and Timofeev, 2003; Eschenko et al., 2008; Cai et al., 2009; Wierzynski et al., 2009).
Like the hippocampal formation, the piriform cortex is a three-layered cortical structure with highly distributed, sparse encoding of input patterns (Haberly, 2001; Rennaker et al., 2007; Poo and Isaacson, 2009). The inputs to the piriform cortex are spatiotemporal patterns generated by the olfactory bulb in response to odor (Wachowiak and Cohen, 2001; Lin et al., 2006; Johnson and Leon, 2007). Through a combination of afferent convergence, a highly auto-associative intrinsic fiber system and Hebbian synaptic plasticity, the anterior piriform cortex (aPCX) is hypothesized to develop experience-dependent memories of previously experienced input patterns to facilitate odor discrimination and recognition (Barkai et al., 1994; Wilson, 2001; Linster et al., 2009).
Olfactory experience induces a variety of changes throughout the olfactory pathway, including changes in local (e.g., within cortical areas)(Wilson et al., 1987; Grajski and Freeman, 1989; Ravel et al., 2003; Quinlan et al., 2004) and global (e.g., between cortical areas) (Martin et al., 2006; Martin et al., 2007; Cohen et al., 2008) circuit connectivity. One of the consequences of these changes is olfactory perceptual learning, which is an enhancement in perceptual acuity for the learned odor and enhanced discrimination of that odor from similar odors (Stevenson, 2001; Fletcher and Wilson, 2002; Wilson, 2003; Wilson and Stevenson, 2003; Li et al., 2008). Olfactory perceptual learning and modulation of odor acuity are associated with changes in both the olfactory bulb (Fletcher and Wilson, 2003; Mandairon et al., 2006; Doucette et al., 2007; Doucette and Restrepo, 2008; Chaudhury et al., 2009) and aPCX (Wilson, 2003; Li et al., 2008). In aPCX, both single-units (Wilson, 2000, 2003) and single-unit ensembles (Kadohisa and Wilson, 2006) show enhanced de-correlation of familiar odors even following passive exposure to those odors (Wilson, 2003). In fact, under urethane anesthesia, brief passive exposure to novel odors is sufficient to improve piriform cortical single-unit discrimination (Wilson, 2003). Disruption of normal aPCX association fiber synaptic transmission and plasticity, for example by disrupting cholinergic input (Barkai and Hasselmo, 1994; Patil et al., 1998), impairs olfactory perceptual learning at both the cortical (Wilson, 2001; Linster et al., 2009) and behavioral (Fletcher and Wilson, 2002) level.
We hypothesized that activity during slow-wave states within the aPCX may be shaped by recent olfactory experience. Specifically, we took advantage of the fact that urethane anesthetized rats undergo spontaneous shifts between fast-wave and slow-wave states (Murakami et al., 2005). Both single-unit and local field potential (LFP) activity were monitored within the aPCX during slow- wave states before and after a fast-wave state during which repeated odor stimulation was delivered. The results suggest that slow-wave state activity within the aPCX reflects recent odor experience.
Single-unit recordings were made from Layer II/III neurons in aPCX of urethane (1.5 g/kg) male Long-Evans hooded rats as previously described (Wilson, 1998). Rats were obtained from Charles River Labs. Recordings were made during the light phase of the 12:12 light:dark cycle. Animals had ad lib access to food and water and were cared for according to NIH guidelines. All procedures were approved by the IACUC’s at both the Nathan Kline Institute and the New York University Medical School. Single-units (filtered 300Hz – 3kHz) and LFP’s (filtered 0.1–300 Hz) were recorded simultaneously with a single tungsten microelectrode (1–5 mOhm). Signals were digitized at 10kHz with a CED micro1401, and analyzed with Spike2 software (CED, Inc.). Electrode placement in the aPCX was guided by recording potentials evoked by electrical stimulation of the lateral olfactory tract, and confirmed histologically at the end of the experiment. Respiration was monitored with a piezoelectric sensor strapped to the chest. This allowed both recording of the respiration and as a trigger for controlling odorant stimulation onset. Anesthetic plane was such that no odorant-evoked respiratory responses were detected. Odorants were delivered in 2 sec pulses initiated at the transition from inhalation to exhalation. Monomolecular odorants were delivered at 100 PPM based on vapor pressure and diluted in mineral oil. At least 30–60 sec inter-stimulus intervals were used to reduce cortical adaptation. Odorants included: isoamyl acetate, ethyl valerate, 5-methyl-2-hexanone, 1-pentanol, heptanal, 2-heptanone, and 1,7-octadiene.
The work was divided into two experiments. First, in 4 animals odor stimuli were delivered during both slow-wave (SWA) and fast-wave activity (FWA). State changes were readily apparent by eye in the LFPs, and were confirmed quantitatively by FFT analyses. SWA was dominated by strong delta frequency (1–4 Hz) power and large slow-waves that occurred at 1–2 Hz. FWA had reduced delta, higher theta (4–12 Hz) and low beta (15–30 Hz) frequency activity and no slow-waves. This experiment was a replication of Murakami et al. (2005) to confirm differential cortical responsiveness to afferent input during SWA and FWA.
The main experiment involved odor stimulation only during FWA in order to allow comparisons between spontaneous SWA before and after FWA odor stimulation. This design was meant to mimic the analyses performed in the hippocampal formation wherein spontaneous activity is monitored during SWA before and after maze learning which occurs when the animal is awake (e.g., (Pavlides and Winson, 1989; Skaggs and McNaughton, 1996). Once a single-unit, or multiple single-units were isolated, the animal was allowed to transition into SWA. In order to be included, a SWA period had to last long enough to have at least 90 slow-waves. Once the animal had transitioned to FWA, cells were randomly assigned to either receive odor stimulation or no odor stimulation. In order to be included in the odor-stimulation dataset, units had to show a significant and reliable response to at least one odor. If a cell was tested but shown to be non-responsive to any odor (no more than 5 total test stimuli, or tested at the end of a control session), it, and the animal, were excluded from the analysis. In some cases, multiple single-units were recorded simultaneously, and data was acquired from all. However, in no case were multiple cells within the same animal tested sequentially, since previous odor experiences may have affected later tests.
Single-unit activity during SWA was assessed with Peri-Slow-Wave Time Histograms (PSWTH). Slow-waves were extracted from continuous LFPs which were low-pass filtered at 50Hz. The time stamps of slow-wave peaks (identified by thresholding in each individual animal) were used as triggers to build PSWTH from ongoing single-unit activity (Fig. 1 A and B). PSWTH’s (5 ms bin width) were built for each cell during two SWA periods. Two measures were calculated for comparison of activity during these two periods. First, total activity within ± 100 ms of the slow-wave peak was determined. Second, the peak of single-unit activity during the PSWTH (± 100 ms of slow-wave peak) was compared between the two SWA periods to determine whether single-unit activity during the slow-wave shifted between time periods. The time bin containing the highest spike count was considered the peak. If 2–3 nearby time bins were of equal counts, the average time between them was used to calculate the peak shift. If 4 or more peaks were equal or if the peaks were widely scattered, that cell was eliminated from this calculation. Five single-units in the control group and three single-units in the odor-stimulated group were dropped for this reason from this specific calculation.
Slow-wave morphology was determined by calculating an average LFP triggered on the slow-wave peak (± 500 ms). Individual animal mean waveforms were then averaged within sample periods and within experimental groups for statistical comparisons (ANOVA).
In four rats separate from those used in the primary experiment here, we confirmed that aPCX sensory-evoked activity is markedly different during FWA and SWA. As previously described (Murakami et al., 2005) and as shown in Fig. 1A, LFP’s recorded in the aPCX spontaneously shifted between periods of FWA and SWA, with transitions in either direction often occurring rapidly over 1–2 sec. As shown in Fig. 1B, during SWA single-units recorded in aPCX Layers II/III fired in phase with the slow-waves, most commonly with firing peaking near its nadir. Furthermore, during SWA, single-units showed reduced responsiveness to odors (Fig. 1C) and reduced entrainment to the respiratory cycle (Fig. 1D) compared to during FWA. Single-unit response magnitude (numbers of spikes/stimulus above baseline rate) to odors presented during SWA was reduced to 31.3 ± 13.8% of response magnitude to the same odors presented during FWA (paired t-test, t(7) = 3.83, p < 0.01). The reduction in responsiveness to afferent input has been shown to be due to changes intrinsic to the aPCX, since responsivity of cortical afferent mitral cells to odors is not reduced during SWA (Murakami et al., 2005). For data presented below, odors were not presented during SWA.
The main question to be examined here was whether spontaneous activity during SWA was modified by recent FWA odor experience, similar to that described for the hippocampus. Data from 44 single-units (22 odor-exposed and 22 control) from 29 rats are described. Slow-wave associated spontaneous single-unit activity was monitored during two SWA periods. In odor-exposed cells, during the FWA between these two SWA periods, odor stimuli to which the unit responded were repeatedly delivered. The mean delay between the last odor stimulus and the onset of SWA measurements was 360 sec, with a range of 1–30 minutes. In control cells, no stimulation was delivered during the intervening FWA. Periods of SWA were chosen to equate the number of slow-waves that occurred during the two periods within a cell, and between the two groups (Table 1). As shown in Table 1, there was no significant difference between slow-wave rate or duration of the intervening FWA period between groups (t-tests).
Single-unit activity associated with slow-waves was significantly modified by prior odor experience compared to control cells that were not odor stimulated. Mean slow-wave associated activity (PSWTH) across all single units within each group is plotted in Fig. 2. In control cells, mean slow-wave associated activity did not significantly change between the two periods of SWA (Fig. 2B). However, in odor-stimulated animals, mean slow-wave associated activity was significantly reduced (paired t-tests, p < 0.05) during SWA after odor stimulation compared to SWA before odor stimulation (Fig. 2C).
The observed decrease in mean slow-wave associated activity in the odor-exposed cells could be due to either a decrease in individual unit activity after odor exposure or a shift in the temporal structure of individual unit activity during the slow-wave, with some cells firing earlier and some later than under baseline conditions. The change in slow-wave associated activity was not due to a change in activity rate. A repeated measures ANOVA of total single-unit spiking during the 200 ms period surrounding (±100 ms) a slow-wave peak showed no significant effect of either group (odor-exposed vs. control, F(1,38) = 0.30, N.S.) or time (pre vs. post, F(1,38) = 0.01, N.S.), nor a significant interaction between group or time (F1,38) = 0.43, N.S.).
Rather, the change in slow-wave associated activity in the odor-exposed group was linked to a shift in temporal structure of single-unit activity after odor exposure. To analyze this, we focused on the histogram bin showing the maximal (peak) activity for each unit during the two slow-wave sampling periods. An example of a single-unit shifting its slow-wave associated activity is shown in Fig. 2D. While single-units in both conditions showed some variability in the timing of peak activity during slow-waves, single-units in odor-exposed animals showed a significantly greater shift in slow-wave associated peak firing during the two sampling periods (Odor exposed mean shift = 50.4 ± 9 ms; control shift = 20.9 ± 5 ms, t-test, t(33) = 2.43, p = 0.02). The direction of the shift in peak slow-wave associated activity was equally balanced between earlier or later times. That is, some cells fired earlier during the slow-wave after odor exposure and some fired later (50% earlier and 50% later in the odor-exposed cells). In a small set of 5 cells, odors were delivered during FWA that were outside of the cells’ receptive field and thus produced no odor-evoked change in firing. Though a small sample, the activity of these cells during the subsequent SWA period was similar to that of the unstimulated controls, with a small mean shift of 27 ms (compared to 21 ms for the unstimulated controls and 50 ms for the stimulated cells).
Given that the slow-waves recorded in the LFP reflect activity of a large population of neurons, and a subset of those neurons change their activity relative to the population based on recent odor experience (Fig. 2), there may be a concomitant change in the slow-waves themselves. As shown in Fig. 3, there was no change in mean peak amplitude or overall morphology of the slow-waves between the two SWA sampling periods. However, an analysis of the pre- versus post difference in waveforms (mean post waveform subtracted from mean pre-waveform within each animal) showed a significant increase in variability of this measure in the odor-exposed group (ANOVA group × time interaction, F(9,117) = 65.66, p < 0.001). Specifically, despite the lack of change in mean morphology of slow-waves pre- versus post-odor, variability in the early phase of the waveform was significantly enhanced by odor experience. Variability in the control waveforms was constant across the waveform and significantly less than that observed in the early phase of the slow-waves in odor-exposed animals. As with the PSWTH analyses, this enhancement in variation may reflect the fact that some waveforms increased and some decreased in early phases after odor exposure.
Activity within the aPCX during slow-wave states is shaped by recent odor experience. The timing of single-unit activity relative to the population of cells active during a slow-wave was modified if the animal had recently (past several minutes) been stimulated with an odor within that cell’s receptive field. If no odor had been delivered, or the cell did not respond to presented odors, the activity of the cell during SWA was stable. Furthermore, the slow-wave itself became more variable, especially along its leading edge, after odor stimulation. No such change was observed in control animals. These results are consistent with memory for recently experienced odors being expressed during spontaneous SWA within the aPCX. Whether this experience-dependent SWA represents a replay or re-activation (Skaggs and McNaughton, 1996; Lee and Wilson, 2002) of a learned odor image, or whether SWA is involved in consolidation of olfactory perceptual learning which can occur using similar stimulation protocols in anesthetized rats (Wilson, 2003), is currently being examined. These results suggest, however, a potential similarity between the role of slow-wave states in the aPCX and hippocampal formation.
During SWA, aPCX single-unit activity occurred largely in phase with LFP slow-waves. Individual cells varied in the specific temporal structure of their firing activity relative to the slow-waves under baseline conditions, and in control cells, this temporal structure was relatively stable over many minutes and between consecutive slow-wave periods, as measured by overall PSWTH and PSWTH peaks. However, if cells had been activated by odor stimulation in FWA, those cells shifted their activity relative to the population (LFP slow-wave) during the subsequent slow-wave period, just as might be expected if the activity of those cells was modified by the previous odor exposure.
Previous work has demonstrated that aPCX acts much like thalamus during SWA to serve as a state-dependent sensory gate (Murakami et al., 2005). During SWA, despite relatively maintained input from mitral/tufted cells of the olfactory bulb, piriform cortical neurons show reduced responsivity to afferent input ((Murakami et al., 2005); Fig. 1). This reduced cortical activity may underlie the observed decrease in perceptual responses to odors during slow-wave sleep (Carskadon and Herz, 2004). The present results suggest that, similar to thalamocortical and hippocampal systems, during this period of reduced sensitivity to afferent input, the aPCX turns inward, with activity driven by intracortical associative synapses. A change in the effectiveness of intrinsic synapses during SWA could occur via changes in cholinergic modulation (Hasselmo and Bower, 1992; Liljenstrom and Hasselmo, 1995; Hasselmo, 1999). Acetylcholine suppresses intrinsic association fiber synapses in aPCX (Hasselmo and Bower, 1992) effectively enhancing the relative importance of afferent input on cortical activity during periods of high cholinergic activity (FWA). Critically, cholinergic input to many forebrain regions drops during SWA, allowing a shift toward intrinsic activity dominance (Liljenstrom and Hasselmo, 1995; Hasselmo, 1999). The most effective synapses during this time may be those that have been strengthened by recent experience. Reactivation of these recently strengthened synapses during slow-wave states may help consolidate the newly initiated changes, and in so doing, help consolidate memory critical for enhanced acuity or odor-associations.
The present results have at least two important implications. First, they suggest that the role of slow-wave states in memory observed in hippocampal circuits may be a common feature of other similar circuits such as aPCX. Second, the olfactory system offers some unique advantages for the study of SWA and memory. Odors, which are complex stimulus patterns (Lin et al., 2006) can be quantitatively manipulated, delivered to animals in different behavioral states, and can be rapidly learned and remembered for long times. In fact, odor memory can even be induced rapidly under anesthesia (Wilson, 2003; Wilson et al., 2004). Thus, the olfactory system may be an especially powerful new model system for the study of slow-wave sleep and memory consolidation.
This work was funded by grant R01 DC003906 to D.A.W.