Title: Editor's evaluation: Ultrafast (400 Hz) network oscillations induced in mouse barrel cortex by optogenetic activation of thalamocortical axons
Abstract: Full text Figures and data Side by side Abstract Editor's evaluation Introduction Results Discussion Materials and methods Data availability References Decision letter Author response Article and author information Metrics Abstract Oscillations of extracellular voltage, reflecting synchronous, rhythmic activity in large populations of neurons, are a ubiquitous feature in the mammalian brain, and are thought to subserve important, if not fully understood roles in normal and abnormal brain function. Oscillations at different frequency bands are hallmarks of specific brain and behavioral states. At the higher end of the spectrum, 150-200 Hz ripples occur in the hippocampus during slow-wave sleep, and ultrafast (400-600 Hz) oscillations arise in the somatosensory cortices of humans and several other mammalian species in response to peripheral nerve stimulation or punctate sensory stimuli. Here we report that brief optogenetic activation of thalamocortical axons, in brain slices from mouse somatosensory (barrel) cortex, elicited in the thalamorecipient layer local field potential (LFP) oscillations which we dubbed “ripplets”. Ripplets originated in the postsynaptic cortical network and consisted of a precisely repeating sequence of 2‑5 negative transients, closely resembling hippocampal ripples but, at ~400 Hz, over twice as fast. Fast-spiking (FS) inhibitory interneurons fired highly synchronous 400 Hz spike bursts entrained to the LFP oscillation, while regular-spiking (RS), excitatory neurons typically fired only 1-2 spikes per ripplet, in antiphase to FS spikes, and received synchronous sequences of alternating excitatory and inhibitory inputs. We suggest that ripplets are an intrinsically generated cortical response to a strong, synchronous thalamocortical volley, and could provide increased bandwidth for encoding and transmitting sensory information. Importantly, optogenetically induced ripplets are a uniquely accessible model system for studying synaptic mechanisms of fast and ultrafast cortical and hippocampal oscillations. Editor's evaluation This study examines the potential neuronal basis for generating ultrafast oscillations (250-600Hz) in the cortex evoked by optogenetic stimulation of thalamocortical afferents in ex vivo brain slices. The authors proposed that these oscillations correlate with sensory stimulation and may be relevant for the perception of relevant sensory inputs and they combined ex-vivo whole-cell patch-clamp recordings, local field potential (LFP) recordings, and optogenetic activation of thalamic afferents to generate ripple-like extracellular waveforms in the cortex, referred to as "ripplets." The authors described the sequences of RS and FS neuron discharge and how they phase-locked to the ripplet, providing a model for the cellular mechanism generating the ripplet. The authors also cited the literature about ultrafast oscillations and carefully compared the novel ripplets to the well-known hippocampal ripples. https://doi.org/10.7554/eLife.82412.sa0 Decision letter Reviews on Sciety eLife's review process Introduction A ubiquitous feature of neuronal networks, in the brains of multiple mammalian species throughout the evolutionary tree, is their propensity to engage in large-scale, synchronous and rhythmic patterns of electrical activity, reflected as oscillations in the local field potential (LFP) (Buzsáki et al., 2013). LFP oscillations occur at frequencies spanning four or more orders of magnitude, with oscillations at different frequency bands often nested together and co-modulating each other (Steriade, 2006). Different oscillatory patterns are hallmarks of specific brain and behavioral states; for example, theta rhythms (4–9 Hz) are typical of exploratory behavior, sleep spindles (~10 Hz) occur during certain sleep stages, and gamma oscillations (30–90 Hz) are the hallmark of cognitive activity and sensory processing (Lopes da Silva, 2013; Singer, 2018). Even faster, transient 150–200 Hz ‘ripples’ in the pyramidal cell layer of the hippocampus occur during quiet immobility and slow-wave sleep (Bragin et al., 1999a; Bragin et al., 1999b; Csicsvari et al., 1999a), and are thought to be critical for memory consolidation (Girardeau and Lopes-Dos-Santos, 2021). Extracellular voltage signals, including oscillations, are driven by currents entering (sinks) or leaving (sources) the intracellular brain compartment, and reflect both synaptic activity and action potential firing (Mitzdorf, 1985; Liebe et al., 2011; Buzsáki et al., 2012; Schomburg et al., 2012; Reimann et al., 2013; Pesaran et al., 2018). Indeed, excitatory and inhibitory neurons in the hippocampus fire at precise phases of theta, gamma, and ripple oscillations in a subtype-specific manner (Klausberger et al., 2003; Klausberger and Somogyi, 2008; Cardin, 2018), and synaptic interactions between excitatory and inhibitory cells are implicated in the generation of oscillations across the spectrum, although the details may vary between frequency bands, between different brain areas, and even between cortical layers (Kopell et al., 2010; Wang, 2010; Whittington et al., 2018). At the high end of the frequency range, ultrafast, 250–600 Hz oscillations have been observed in both hippocampus and neocortex. In the hippocampus, such oscillations (often called ‘fast ripples’) are almost exclusively paroxysmal (reviewed in Gulyás and Freund, 2015; Lévesque and Avoli, 2019), although their underlying cellular and network mechanisms remain under debate (Dzhala and Staley, 2004; Foffani et al., 2007; Engel et al., 2009). In the neocortex, however, non-paroxysmal 600 Hz wavelets have been repeatedly observed in scalp recordings from human subjects in response to peripheral nerve stimulation (reviewed in Curio, 2000). Similar 600 Hz LFP oscillations in response to peripheral nerve stimulation were also observed in monkeys and piglets (Ikeda et al., 2002; Baker et al., 2003). Notably, LFP oscillations in the 300–500 Hz range were also consistently observed in rats in response to whisker deflections or to auditory clicks (Jones and Barth, 1999; Jones et al., 2000; Jones and Barth, 2002); however, their underlying mechanisms and functional significance remain largely unexplored. We recorded ex vivo extracellular and intracellular activity evoked by optogenetic activation of channelrhodopsin (ChR2)-expressing thalamocortical axons in thalamorecipient layers of mouse somatosensory (barrel) cortex. We report here that brief light pulses (as short as 1 ms) elicited a transient (<25 ms), highly reproducible LFP oscillation or ‘ripplet,’ which closely resembled hippocampal ripples but, at ~400 Hz, was more than twice as fast. Paired whole-cell recordings from fast-spiking (FS) inhibitory interneurons and regular-spiking (RS) excitatory cells revealed precise phase relationships between FS spikes, RS spikes and ripplets, and between FS spikes and sequences of alternating EPSCs and IPSCs in RS cells, suggesting that phasic RS→RS and RS→FS excitation underlies ripplet generation. Ripplets may be a stereotypical cortical response to strong, punctate stimuli that activate thalamocortical afferents with a high degree of synchrony and could be used by the cortex to encode specific features of a sensory event. Importantly, optogenetically evoked ex vivo ripplets are a uniquely accessible model for studying the synaptic basis of fast and ultrafast network oscillations, including hippocampal ripples. Results We recorded extracellular and intracellular responses in layer 4 (L4) of barrel cortex brain slices, evoked by widefield optogenetic activation of thalamocortical synapses. Slices were prepared from brains of 3–7-postnatal-week-old mice of both sexes expressing ChR2 in the somatosensory thalamus (see ‘Materials and methods’). We applied brief light pulses through the epi-illumination path of the microscope, likely activating most of the thalamocortical axons and terminals within the surrounding L4 barrel. To verify that ChR2 in L4 was expressed by thalamocortical and not intracortical axons, we imaged and quantified the extent of tdTomato reporter expression in the cortex (Figure 1—figure supplement 1). Within the barrel cortex, there was a sparse population of reporter-expressing cell bodies in L2/3 and L5, but virtually none within L4 (Figure 1—source data 1). Since the thalamus and L4 itself are the two main sources of excitatory input to L4 neurons (Feldmeyer et al., 1999; Petersen and Sakmann, 2000; Lefort et al., 2009), we conclude that the network activity we observed was predominantly evoked by thalamocortical activation, although we cannot rule out that a small number of corticothalamic neurons in L6, which send axonal collaterals to L4 (White and Keller, 1987; Kumar and Ohana, 2008), also expressed ChR2 and provided a minor excitatory contribution. Optogenetic activation of thalamocortical axons evoked ripple-like extracellular wavelets A brief light pulse (1–10 ms, typically 2–5 ms) evoked a stereotypical field potential wavelet in L4 (Figure 1A). In this and all other electrophysiological traces, the light-evoked response is represented by a 25-ms-long record beginning 1 ms before light onset, and the timing of the light pulse is indicated by a cyan bar above or below the trace. At all stimulus durations, the LFP waveform consisted of an initial low-amplitude negative component (solid arrowhead), followed by 2–5 (typically 3–4) larger negative transients (hollow arrowheads) riding on a slow, positive-going envelope. In the example slice in Figure 1A, the fourth transient was barely discernible in response to a 1 ms light pulse (left), but became noticeably larger with a 5 ms pulse (right). A low-amplitude negative shoulder or ‘hump’ (arrow) was often observed on the rising phase of the first large transient. These optogenetically evoked waveforms were reminiscent of hippocampus ripples, LFP wavelets recorded from the cell body layer of areas CA1 or CA3; we therefore named them RIPPle-Like Extracellular Transients or ‘ripplets’. Figure 1 with 2 supplements see all Download asset Open asset Local field potential oscillations (ripplets) induced by optogenetic thalamocortical activation. (A) Averaged ripplets evoked in an example slice; light pulses (1 and 5 ms duration for left and right traces, respectively) represented by cyan bars above traces, and apply to (A–C). Solid arrowhead indicates the early presynaptic TC volley; hollow arrowheads indicate the peaks of the postsynaptic components; the arrow points to the hump on the rising phase of the first postsynaptic transient. (B) The iGluR-independent (i.e. presynaptic) component: same stimuli as in (A) but in the presence of CNQX + APV. Arrowheads indicate a single presynaptic volley in response to a 1 ms stimulus and three volleys (of decreasing amplitude and coherence) with a 5 ms stimulus. (C) The difference between (A) and (B), revealing the purely postsynaptic component of the ripplet. Symbols as in (A). (D) Superposition of the presynaptic (blue, left) and postsynaptic (red, right) components of the ripplets at three stimulus durations. (E) Superposition of presynaptic components (blue, left) and ripplets (red, right) from slices from four different animals, normalized to the same maximal amplitudes to facilitate comparison of their time course. (F) Histogram of peak times (from light onset) of presynaptic volleys (left) and of ripplet transients (right) in slices from 8 and 11 animals, respectively (5 animals are included in both plots). Presynaptic volleys peaked at 2.3 ± 0.07, 6.0 ± 0.22, and 10.8 ± 0.24 ms from light onset. Ripplet transients peaked at 4.4 ± 0.11, 6.5 ± 0.26, 8.9 ± 0.30, and 11.9 ± 0.38 ms. (G) The histograms from (F) modeled as normal distributions, with the same mean, SD and integral as the corresponding component. Left, presynaptic volleys; right, ripplet transients (red) with presynaptic volleys overlaid as gray outlines for comparison. Figure 1—source data 1 Counts of Cre-expressing neurons in the KN282 mouse. Four male mice (4–6 weeks old) were deeply anesthetized with 2.5% Avertin and transcardially perfused with saline followed by 50 ml 4% paraformaldehyde (PFA). Brains were removed, post-fixed in 4% PFA at room temperature for 4 hr, stored for at least 72 hr in 30% sucrose solution in phosphate buffered saline (PBS) at 4°C, and sectioned coronally on a freezing stage sliding microtome. For imaging, five 60 μm sections were sampled per brain at 240 μm intervals, in total extending 800–2000 μm posterior to bregma, as determined by comparison with the Paxinos and Franklin atlas (Paxinos and Franklin, The Mouse Brain in Stereotaxic Coordinates, 4th edition, Academic Press 2012). Confocal image stacks were taken with a ×10 objective at 2.5 μm Z-steps on a Zeiss LSM 710 or a Nikon A1R confocal microscope. Labeled cells were counted by visual inspection of the full stack and summed per area over the five sections from each animal; counts thus represent the number of cells in a composite 300 μm section. Areal boundaries follow the designations in the Paxinos and Franklin atlas. https://cdn.elifesciences.org/articles/82412/elife-82412-fig1-data1-v2.xlsx Download elife-82412-fig1-data1-v2.xlsx Synchronous spike volleys in thalamocortical terminals generate relatively large current sinks in layer 4, and these are discernible as negative transients in the LFP (Morin and Steriade, 1981; Agmon and Connors, 1991; Swadlow et al., 2002; Bruno et al., 2003). To reveal any presynaptic thalamocortical components in the ripplets, we blocked fast ionotropic glutamate receptors (iGluRs) pharmacologically with CNQX + APV. This eliminated the large negative transients but left unchanged the early negative component (Figure 1B, left panel, solid arrowhead), which therefore reflected the initial thalamocortical volley. Interestingly, the presynaptic response to 5 ms or longer light pulses revealed one or two additional presynaptic transients of decreasing amplitudes (Figure 1B, right panel, solid arrowheads), suggesting that thalamocortical axons fired additional volleys of decreasing magnitude and coherence. Subtracting the waveforms in CNQX + APV from those in control ACSF isolated the postsynaptic component of the response (Figure 1C). Superposition of the isolated presynaptic (Figure 1D, left) and postsynaptic (Figure 1D, right) contributions at 1, 5, and 10 ms stimulus durations demonstrated that prolonging the light pulse added late components to the response waveform but did not affect the timing of the earlier peaks. To confirm that the iGluR-independent signals reflected action potentials, we added TTX to the bath, which blocked all remaining responses except for a small square waveform (Figure 1—figure supplement 2). The latter was coincident with the light pulse and was most likely an electrical or optoelectrical artifact, and is subtracted from all traces in Figure 1. Ripplet waveforms, as well as their isolated presynaptic components, were remarkably consistent between animals in their temporal structure, as illustrated in Figure 1E by the superposition of responses in slices from four different animals, drawn at different vertical scales to facilitate time course comparison. To quantify the variability in the temporal structure of ripplets between animals, we measured the peak times (relative to light onset) of the isolated presynaptic volleys (Figure 1F, left; eight slices from eight animals tested in CNQX + APV) and of the postsynaptic transients (Figure 1F, right; 11 slices from 11 animals with ripplet amplitudes ≥0.5 mV). Standard deviations (SDs) of the presynaptic volleys ranged from 0.2 to 0.6 ms, and SDs of ripplet peaks ranged from 0.3 to 1.0 ms. Calculated over the three volleys, the presynaptic axons appeared to fire at an average frequency ( = 1/average ISI) of 239 ± 6 Hz. This is consistent with the observed burst frequency in thalamocortical neurons of the same mouse genotype when activated optogenetically ex vivo, which rarely exceeds 300 Hz (Hu and Agmon, 2016, their Figure 4E). Calculated over the four ripplet peaks, ripplet frequency averaged 408 ± 15 Hz, that is, nearly twice as fast as the apparent presynaptic volleys. To facilitate comparison of presynaptic components and ripplets, we modeled the histogram peaks in Figure 1F as Gaussians with the same mean, SD, and integral as that of the respective event (Figure 1G). Overlaying the Gaussians (Figure 1G, right) highlighted the lack of temporal coherence between the presynaptic volleys and ripplets, indicating that ripplets were not a direct, 1:1 response to bursts in thalamocortical axons but were likely generate de novo within the L4 network. Optogenetic stimulation of thalamocortical axons evoked precisely timed FS spike bursts Since ripplets required intact glutamatergic neurotransmission, they most likely reflected population activity within L4; but in which cells? To answer this question, we examined light-induced whole-cell responses in inhibitory FS interneurons and excitatory, RS cells, two subclasses known to receive direct thalamocortical inputs (Agmon and Connors, 1992; Gibson et al., 1999; Porter et al., 2001; Bruno and Simons, 2002; Gabernet et al., 2005; Sun et al., 2006; Cruikshank et al., 2007; Shigematsu et al., 2019). These two subtypes are readily identifiable in slice recordings by their firing patterns and also fall into distinct, nonoverlapping clusters in post hoc analysis of their electrophysiological parameters (Figure 2—figure supplement 1, Figure 2—source data 1). A brief (typically 2–5 ms) light pulse at 60–90% maximal intensity elicited a strong (10–30 mV) depolarization in L4 FS cells, which, in 80% of cells (N = 44 cells from 35 animals), gave rise to a stereotypical burst of 2–7 (typically 3–5) spikes. Representative bursts in an example FS cell, evoked by 2 and 5 ms light pulses, are illustrated in Figure 2A and B. Spiking almost never persisted beyond the illustrated 25 ms time window, although the underlying subthreshold depolarization often lasted for 50 ms or more. We performed detailed analysis on all FS cells that consistently fired bursts of four or more spikes (25 cells from 20 animals). Bursts in this dataset had three distinguishing properties: Very high frequency: Averaged over these 25 cells, the first three interspike intervals (ISIs) were 2.4 ± 0.05, 2.2 ± 0.08, and 2.7 ± 0.13 ms (mean ± SEM), respectively, for an average burst frequency ( = 1/average ISI) of 418 ± 9.5 Hz, not significantly different from the extracellular ripplet frequency (p=0.56). Very high temporal precision: Bursts in a given cell were precisely reproducible, as illustrated in Figure 2A and B by the near-perfect registration of consecutive sweeps, the alignment of spikes in the raster plots and the very low jitter in spike times indicated above the raster plot. In the full dataset, spike jitter averaged 23 ± 4, 63 ± 10, 110 ± 12, and 183 ± 21 μs for the four spikes, which translates to coefficients of variation (CV = SD/mean) of 0.7 ± 0.1%, 1.1 ± 0.2%, 1.4 ± 0.1% and 1.8 ± 0.2%, respectively. This low jitter is also illustrated in Figure 2H and I, in which the vertical extent of the boxed regions indicates 10–90th percentiles of spike time jitter and CV, respectively. High animal-to-animal reproducibility: Across slices from different animals, the distributions of peak spike times for each of the four spikes in the burst was very narrow: 3.1 ± 0.06, 5.5 ± 0.08, 7.7 ± 0.12, and 10.3 ± 0.14 ms after light onset. This is also illustrated in Figure 2H and I, in which the horizontal extent of the boxed regions indicates 10–90th percentile of spike times. Figure 2 with 2 supplements see all Download asset Open asset Precise light-induced spike bursts in fast-spiking (FS) interneurons. (A) Upper panel: spike bursts evoked in an example L4 FS interneuron by 5 ms (blue traces) and 2 ms (red traces) light pulses (cyan bars) at 90% intensity; 10 consecutive sweeps repeated at 8 s intervals at each duration. Arrowhead indicates a late, low-reliability spike evoked by the longer light pulse. Lower panel: raster plots of 20 consecutive bursts from the same neuron, evoked by 2 and 5 ms pulses (red and blue symbols, respectively). Spike time jitters (SD of spike peak times), in μs, are indicated above the raster, in the respective colors. (B) As in (A) but for bursts evoked by 5ms light pulses at 20 and 90% intensities (red and blue traces and symbols, respectively). (C) Average number of spikes/stimulus in seven FS interneurons tested with both 2 and 5 ms light pulses. Medians indicated by horizontal black lines. *p=0.016, sign test. (D) The number of spikes/stimulus fired by 14 FS interneurons tested at 4–5 different light intensities. Error bars indicate SEM. ***p<0.0001; *p=0.02. (E) Jitter of the second spike in the burst, and the first interspike interval (ISI), compared between 5 and 2 ms light pulses, from the dataset of panel (C); n.s., not significant. (F) Same as (E), but comparing the lowest and highest light intensities from the dataset of panel (D). Note significant difference in spike jitter (***p<0.0001) but not ISI. (G) Another example cell stimulated at 90% intensity (upper trace, five superimposed sweeps) and at 20% intensity (lower traces, three single traces and eight superimposed sweeps). Note that at 20% intensity, spikes drop out sporadically and reveal subthreshold excitatory postsynaptic currents (EPSPs) (arrows), with little change in the temporal structure of the burst (vertical guidelines are aligned with the spikes at 90% intensity). (H) Spike times in all 25 FS cells with each spike order in a different color, plotted by its mean time measured from light onset (X axis) and by its jitter (Y axis); crosses indicate medians, boxes indicate 10–90th percentile range. (I) As in (H) but plotted along the Y axis by CV (jitter/mean spike time). (J) Histogram of all average spike times using the same color scheme as in (H); each peak in the histogram is overlaid by a Gaussian with the same mean, SD, and integral (blue curves). (K) The Gaussians from panel (J), representing FS spike bursts, superimposed with the Gaussian from Figure 1G representing thalamocortical spike volleys (blue and red curves, respectively). (L) The Gaussians representing FS spike bursts (blue) and those from Figure 1G representing ripplets (red) are superimposed; note the antiphase relationship between FS spikes and ripplets. Figure 2—source data 1 Electrophysiological parameters of fast-spiking (FS) and regular-spiking (RS) neurons. Analysis was done on a subset of 32 FS interneurons and 25 RS cells in slices from 25 and 22 animals, respectively, 3–6.5 weeks old (14 animals were common to both subsets). A total of eight electrophysiological parameters were analyzed per cell. Single-spike parameters were measured at rheobase (minimal current evoking an action potential). All current steps were 600 ms long. Electrophysiological parameters definitions: Vrest: resting potential upon break-in, with no holding current applied. Vthreshold: the voltage where dv/dt = 5 V/s. Spike height: Spike peak-Vthreshold. Spike width at half-height (SWHH): spike width measured half-way between Vthreshold and spike peak. After Hyperpolarization (AHP): Vthreshold -spike trough. Input Resistance (Rin): the slope of the I-V plot, calculated from 4 to 6 positive and negative subthreshold current steps, at membrane potentials up to ±15 mV from rest. Imax: the maximal current step applied before a noticeable reduction in spike height. Fmax: the maximal steady-state firing frequency, computed as the reciprocal of the average of the last five interspike intervals (ISIs) in a spike train elicited by Imax. https://cdn.elifesciences.org/articles/82412/elife-82412-fig2-data1-v2.xlsx Download elife-82412-fig2-data1-v2.xlsx Since thalamocortical axons also form a lower tier of terminations in L5B (Frost and Caviness, 1980; Herkenham, 1980; Agmon et al., 1993; Meyer et al., 2010), we also tested for occurrence of light-evoked spike bursts in infragranular FS cells. Brief light pulses evoked in L5B FS interneurons spike bursts that were indistinguishable, in spike number and frequency, from those in L4; of 13 cells stimulated with 2 ms light pulses, 11 cells from nine animals fired consistently four spikes/stimulus, at an average frequency of 405 ± 13 Hz (Figure 2—figure supplement 2). In the remainder of this study, we focus on L4. To examine if burst patterns were dependent on stimulus parameters (light duration and intensity), we first compared bursts elicited by 2 and 5 ms light pulses. In the example cell in Figure 2A, a 2 ms stimulus elicited a precisely reproducible four-spike burst (10 superimposed red traces); increasing stimulus duration to 5 ms left the timing and jitter on these four spikes nearly unchanged but occasionally elicited a fifth, long-latency spike (arrowhead) with considerably higher jitter compared to the first four spikes (blue traces; note increased jitter in the raster plot in the lower panel). We refer to such spikes, with higher jitter and occasional failures, as ‘low-reliability spikes.’ All seven FS cells tested at both stimulus durations fired more spikes in response to a 5 ms compared with 2 ms stimulus, with a median increase of one spike/stimulus (Figure 2C) but with no appreciable change in timing and precision of the earlier spikes, as seen by the identical median values of the second spike jitter and the first ISI at the two stimulus durations (Figure 2E). We then compared the number and precision of spikes in response to different light intensities. Bursts elicited by 20 and 90% intensity, in the same example cell from Figure 2A, are illustrated in Figure 2B. At the lower intensity, there was again one less spike, but in addition the jitter of all spikes except the first was >3× higher. In a subsample of 14 cells tested with 5 ms light pulses at five different intensities (10, 20, 40, 60, and 90% of maximal intensity), the average number of spikes/trial increased from 2.9 at 10% to 4.8 at 90% intensity (Figure 2D). As seen by comparing the second spike jitter and first ISI between at 20 and 90% light pulses (Figure 2F), overall spike timing did not differ between the two intensities (p=0.38) but at lower intensities spikes were much less precise (second spike jitter: 106 ± 13 vs. 43 ± 6 μs, p<0.0001). That ISIs were largely independent of stimulation intensity suggested that the burst pattern was not intrinsically generated in the FS cell but was imposed on it by extrinsic inputs. Consistent with this conclusion, reducing the light level resulted in some cells in spikes dropping out sporadically, revealing apparent subthreshold EPSPs that remained aligned with the temporal structure of the burst (Figure 2G, guidelines). That EPSPs elicit FS spikes with such high precision is consistent with the properties of excitatory synaptic inputs onto FS interneurons in neocortex and hippocampus (Geiger et al., 1997; Galarreta and Hestrin, 2001). What were the temporal relationships between FS spike bursts, thalamocortical volleys, and ripplets? We plotted all FS spike times in our dataset as a histogram and then modeled each spike order as a Gaussian (Figure 2J). Overlaying Gaussians representing FS spikes with those representing thalamocortical volleys from Figure 1G (Figure 2K) revealed a close temporal relationship between the first FS spike and the initial thalamocortical volley, with the postsynaptic spike following the presynaptic volley at a latency of 0.8 ms, consistent with a monosynaptic response. Later FS spikes, however, actually preceded the closest presynaptic volley, consistent with the conclusion above that the additional thalamocortical volleys did not drive the postsynaptic bursts in a 1:1 manner. In contrast, overlaying the distributions of FS spikes and LFP transients (Figure 2L) revealed a 1:1 relationship between these events, however, with the two oscillations almost exactly out of phase: each ripplet transient lagged behind the corresponding FS spike peak by 0.44–0.54 of a cycle, translating to 1.0–1.3 ms. FS cells synchronized their firing with near-zero lag and submillisecond precision Given the precise temporal relationship between FS spikes and the population activity reflected as ripplets, one would expect different FS cells in the same barrel to fire in tight synchrony with each other. We therefore examined intra-barrel FS-FS temporal relationships by simultaneous paired recordings. Indeed, when stimulated by a light pulse, simultaneously recorded FS-FS pairs exhibited highly precise spike synchrony. In the two example pairs shown in Figure 3A and B, the peaks of the first three spikes in each burst aligned to within 0.1–0.3 ms between the two cells. To quantify the magnitude and precision of pairwise synchrony, we used the Jitter-Based Synchrony Index (JBSI), which is a normalized measure of spike-spike synchrony in excess of that expected by chance (Agmon, 2012; see ‘Materials and methods’). Synchrony is defined as the fraction of spikes co-occurring within a predetermined ‘synchrony window’ (SW), and chance synchrony is the average synchrony remaining after shifting each spike in one train by a random jitter ≤J, J = 2·SW. When calculated for decreasing SW values, the JBSI typically increases to a maximum and then precipitously drops off to zero, as J falls below the intrinsic precision of the system and the applied jitter no longer disrupts the pairwise synchrony. We defined ‘pairwise precision’ as the smallest SW value with JBSI > 0.5. Since neurons can maintain a precise temporal relationship even when they fire a