EFTA01076046.pdf

The Journal of lietrosoence. F &nary 17. 2010 • 30(71:2783-2791 • 2783 BehaviorallSystems/Cognitive Neuronal Stability and Drift across Periods of Sleep: Premotor Activity Patterns in a Vocal Control Nucleus of Adult Zebra Finches Peter L Rauske,I Zhiyi Chi? Amish S. Dave,' and Daniel Margoliashi Departments of 'Organismal Biology and Anatomy and 2Statistics, University of Chicago, Chicago, Illinois 60611 How stable are neural activity patterns compared across periods of sleep? We evaluated this question in adult zebra finches, whose premotor neurons in the nucleus robustus arcopallialis (RA) exhibit sequences of bursts during daytime singing that are characterized by precise timing relative to song syllables. Each burst has a highly regulated pattern of spikes. We assessed these spike patterns in singing that occurred before and after periods of sleep. For about half of the neurons, one or more premotor bursts had changed after sleep, an average of 20% of all bursts across all RA neurons. After sleep, modified bursts were characterized by a discrete, albeit modest, loss of spikes with compensatory increases in spike intervals, but not changes in timing relative to the syllable. Changes in burst structure followed both interrupted bouts of sleep (1.5-3 h) and full nights of sleep, implicating sleep and not circadian cycle as mediating these effects. Changes in burst structure were also observed during the day, but far less frequently. In cases where multiple bursts in the sequence changed in a single cell, the sequence position of those bursts tended to cluster together. Bursts that did not show discrete changes in structure also showed changes in spike counts, but not biased toward losses. We hypothesize that changes in burst patterns during sleep represent active sculpting of the RA network, supporting auditory feedback-mediated song maintenance. Introduction court females with "directed" singing: precisely structured, regu- Sleep-dependent behavioral plasticity has been observed in a lar songs comprising introductory notes followed by a sequence broad range of perceptual, motor, and higher-level cognitive of syllables organized into a "motif." Directed songs are even tasks in studies in adult humans (Kami et al., 1994; Stickgold et more highly regulated than the undirected songs males otherwise al., 2000; Fischer et al., 2002; Walker et al., 2002; Fenn et al., 2003; sing (Sossinka and Bohner, 1980; Kao et al., 2005; Glaze and Wagner et al., 2004; Brawn et al., 2008). Electrophysiological Troyer, 2006). studies support a role for active processes during sleep affecting Associated with directed singing are highly structured bursts memory consolidation in humans (Maquet et al., 2000; Peigneux of activity in presumptive projection neurons in the nucleus ro- et al., 2004; Reis et al., 2009), and behavioral and electrophysio- bustus arcopallialis (RA) (Yu and Margoliash, 1996). Each spike logical studies in animals implicate sleep in plastic mechanisms. burst has submillisecond precision in its timing relative to its Sleep modulates plastic changes in ocular dominance histograms corresponding syllable within the motif (Chi and Margoliash, in the developing visual cortex of young cats (Frank et al., 2001; 2001; Leonardo and Fee, 2005). These bursts have a well-defined Aton et al., 2009), the emergence of song system neuronal burst- number of spikes in a well-defined temporal pattern, both of ing in juvenile birds at the onset of song learning (Shank and which vary across bursts emitted at different times in the song. A Margoliash, 2009), and experience-dependent changes in the given burst thus has a specific identity associated with onset time, correlations of activity patterns of rat hippocampal neurons (Poe number of spikes, and pattern of spikes. RA neurons show highly et al., 2000). regulated oscillatory spontaneous activity, become completely These results emphasize changes measured in populations of suppressed about 50 ms before onset of song, and may achieve neurons. Sleep-dependent changes in the individual activity pat- instantaneous firing rates of almost 800 Hz during singing. Thus, terns of single neurons during behavior are not well defined, the nervous system expresses almost the entire dynamic range however, and thus there is little data on the stability of single available to precisely modulate the activity of single RA neurons neuron activity patterns across periods of sleep. In this study, we during singing. address this issue in the birdsong system. Male zebra finches We took advantage of the reliability and precision of this sys- tem to examine neuronal stability over extended periods of time. Ikuwed June 5,2009: tedsed Dec. 13.2609; 5,00165 150.11,2010. RA extracellular recordings can be stable with high signal-to- lkswalmassappaled in part to/ KalIcnai Ir60cats °Math GT/Ell M1159831 ten Mast;awful loPaa noise ratio (SNR), but the technical challenge of maintaining Arnida.Penry D.I.gattaret end Sitpilin D. Shea 55tongalitittv.5 ct thls mint600n. high-quality recordings over the required durations and behav- (cert:pc00fficesti0510 te addressed to DI. Wet L.Pathlo.Ser6o0MccartifoimaxeRcgrankkitateatio0 Inuct6e of 010460 3451. Swerlx Sued, SW< 1476.(hog0116051S brt01. iors in freely moving animals required by this design limited the 1)01.1015230/151.0050 3112402010 size of the dataset. Nevertheless, we were able to directly compare Com6ght 02010 the authors D2706174/10/10/761-12515.000) premotor activity of the same single neurons before and after EFTA01076046 2781 • .I. ileurosci., Febuary 17,2010 • 30I71:1783-2791 Rauske et al. • Xeuronal Stability and Mt across Sleep periods ofsleep. To the best ofour knowledge, such comparisons recordings until single-unit isolation was lost. Auditory stimulation en- have not been reported in any premotor system. abled us to verify the responsiveness to the bird's own song that RA neurons exhibit exclusively during sleep (Dave et al., 1998). Further- Materials and Methods more, this was a preliminary experiment to test the hypothesis that sleep- To examine the effects of sleep on the stability of premotor burst patterns related changes in singing behavior result from drift arising from neural in RA neurons, we recorded neuronal activity in three types of experi- replay during sleep activity without concomitant auditory feedback mental sessions: short-sleep (or interrupted-sleep) sessions, long-sleep (Deregnaucourt et al., 2005). We hypothesized that playback would pro- (or normal circadian-sleep) sessions, and awake-only sessions. For both vide structured activity during sleep, possibly preventing sleep-related types of sessions including sleep, we recorded the activity of the same changes, but failed to see systematic differences between short-sleep (au- single RA neurons whilebirdssang or produced learned calls (sec below) ditory stimulation) and long-sleep (no stimulation) sessions (see Re- both before and after the period of sleep. We developed algorithms to sults), a null result with respect to the sleep-drift hypothesis. We do not identify changes to burst patterns across periods of sleep, as well as sta- consider this hypothesis further in this study. tistical techniques to compare the frequency of such changes with that In some additional, exceptional cases = 5 neurons, 3 birds), we observed in the absence of sleep. successfully gambled on our ability to maintain stable unit isolation Eleetrophysiology and design of the experiments. All animal procedures across a full night of sleep (8 or 10 h), maintaining the normal light/dark were approved by an Institutional Animal Care and Usc Committee. cycle. All but one of these cases involved single-unit isolation, with the Adult male zebra finches (n = 13) were habituated to either a 16/8 h or exception being a site in which a pair of units could be reliably distin- 14/10 h light/dark cycle. We found no systematic differences between the guished from background activity but not from each other; this "double- two conditions, and combine the data for aggregate statistical analyses. unit" site was treated similarly to single units in our analysis. No auditory The birds were implanted with microdrives with electrodes targeting PA; stimuli were presented during sleep for these sites, but ongoing activity the implant design and surgical procedures have been described in detail was sampled throughout the night to verify the presence of bunting previously (Dave et al., 1999). Briefly, a recording device carrying four activity in RA that indicated the bird remained asleep. When the next glass-coated Pt-Ir electrodes (impedance, 1.2-2.0 MS/ at 1 kHz) was day's light cycle began, recordings continued until unit isolation was lost. implanted under modified Equithesin anesthesia over RA. During re- Finally, we augmented this data set with additional recordings during cordingsessions starting 2-4 d later, a flexible cable connected the head- vocalizations in recording sessions that did not indude sleep (see below). gear to an overhead commutator to allow the bird free movement within Analysis of deep. Sleep was objectively defined behaviorally (eye clo- the cage. Differential recordings were used to minimize movement arti- sure, body posture), and we also developed a quantitative measure of facts. Recording sites were obtained by audiovisual monitoring of the spontaneous bursting in RA neurons to use as an assay for sleep. We first recordings while using a drive screw to manually advance the electrodes. established a baseline for a neuron's spontaneous spiking activity during Birds were manually restrained during this procedure, then carefully periods before and after darkness when the bird was awake and active, released into the cage while trying to maintain unit isolation. but not vocalizing. We used 1-4 min segments of neuronal activity both Recording sessions began at various times during the day, and we before and after darkness, dividing the spiking activity into 3 s segments. recorded only sites with at least one unit that could be well isolated. In all For each segment, the distribution of interspike intervals (1S1s) was ap- cases, a conspecific female was introduced into an adjacent half-cage to proximately Gaussian because of the highly regular spiking activity of RA elicit directed singing and calling. [In male zebra finches, contact or neurons in awake, nonvocalizing birds. We cakulated for each segment's so-called "long" calls are learned vocalizations whose production in- ISI distribution the mean (IS1-MEAN) and standard deviation (1SI-SD). volves PA premotor activity (Zann, 1985; Simpson and Vicario, 1990), The resulting range of values across all awake segments for each single and they are treated equivalently with song syllables in this study.' unit provided an estimate of the baseline variability in spiking activity in After collecting high SNR spike data during vocalizations comprising the awake bird. at least 10 song motifs and/or contact calls, or in the normal circadian To quantify the amount of sleep during darkness, we similarly divided rhythm depending on experimental design (see below), the cage lights spiking activity into 3 s segments, calculating the ISI-MEAN and IS1-SD were doused. After the bird was quiescent for several minutes, activity in for each segment. Any segment whose ISI-MEAN and 151-517 both fell RA entered a characteristic bursting mode. This distinct state was never within a 95% confidence interval as determined by the baseline awake observed in an awake bird, and bursting disappeared whenever the bird distributions was labeled "awake"; all other segments were labeled as was disturbed or became active. Spontaneous bursting in RA and its `sleep" (for exampk,see supplemental Fig. I, available at unvw.jneurosci.org efferent sensorimotor control nucleus (HVC) has come to be used as an as supplemental material). Such labeling agreed well with visual inspec- assay for sleep. It is reliably associated with the onset of sleep postures and tion of spiking activity, with segments including sleep-typical depressed strong, selective auditory responses (Dave et al., 1998; Dave and Margo- firing rates and/or bunting reliably labeled as sleep. Video surveillance Hash, 2000; Nick and Konishi, 2001; Hahnloser ct al., 2002, 2006; Cardin under infrared illumination verified that the bird was quiescent with and Schmidt, 2003; Rauske a al., 2003; Shank and Margoliash, 2009) closed eyelids in >95% of sleep-labeled segments. We had not developed and has been correlated with EEC measures of sleep (Nick and Kon- reliable EEG recording techniques and an understanding of sleep staging ishi, 2001; Hahnloser et al., 2006; Shank and Margoliash, 2009). Dur- in zebra finches except toward the end of these studies (Low et al., 2008); ing recording sessions including 1.5-3 h darkness (labeled "short sleep"; nevertheless, our analysis reliably distinguished sleep from wakang. n = 10 neurons, 4 birds), we recorded continuously from the isolated RA Song syllables, spike bursts, and a definition of bunt opts. Vocalizations single units, enabling us to estimate the amount of time birds actually and onset and offset times for each syllable were identified by manual slept by examining the bursting activity (or lack thereof) during the dark inspection of spectrographs. A syllable was defined as a stereotyped vocal period. We used a quantitative measure of spontaneous PA bursting as a gesture containing no silent interval 710 ms: in addition to the tradition- sleep assay, described below. During some recording sessions, we also ally defined song syllables that comprise song "motifs" (stereotyped se- verified by direct observation (infrared monitoring) that the bird's eyes quence of syllables), wealso included introductory notes at the beginning were closed and respiration slowed when PA activity indicated sleep of singing bouts and isolated "long" calls, both of which recruit RA (Dave et al., 1998). bunting activity, in our definition of "syllable" for this study. Syllable During the short-sleep recording sessions, we also presented playback onset times and spike times were merged for each site to create a raster of the bird's own song. Recordings of the bird's own song were scaled to plot of spiking activity associated with each syllable type. For each sylla- 70 dB root-mean-squared amplitude and presented randomly at 10-30 s ble, we included the spiking activity beginning 50 ms before syllable onset intervals beginning immediately after turning out the lights. After 50- and ending with the syllable offset. 250 repetitions of song playback, we recorded 20-60 min of ongoing We used simple thresholding techniques to identify spike times for spiking activity while the bird remained asleep. Thereafter, the lights most, extremely well-isolated single units. For a few sites with more were then turned back on, rousing the bird, after 1.5-3 h of sleep. Birds ambiguous isolation, we used theSpiicesort program, which uses a Bayes- then directed singing toward the adjacent female, and we continued ian approach to identify putative spikes with distinct spike-shape models EFTA01076047 Rauske et al. • Neuronal Stability and Drift across Sleep 1. Neurovi, February 17,2010 .30(71:2783-2794. 278S (Lewicki, 1994). To confirm single-unit isolation in all cases, we visually pain, while preserving each individual burst's interspilce intervals. inspected overbid waveforms from all identified spike times to confirm Pre-sleep and postsleep burst stacks were then aligned with each other that spike shapes were consistent throughout our recordings, and we according to similar procedures, with all of the bunts in each stack used ISI distributions to confirm the hallmarks of single-unit isolation in shifted as a whole so that the relative timing within each stack was RA (i.e., an approximately Gaussian distribution of ISIs during behav- preserved. ioral quiescence and a lack ofISIs <1ms). For the majority ofsites (28 of The L, metric used in the L,-MIN method measures the difference 42 single units), we were able to confidently identify 100% of all spikes between two spike sequences obtained by averaging over all spikes the after manual inspection. The remaining single-unit sites, as well as the temporal difference between each spike and its closest corresponding "double-unit" site, included a small number ofambiguous spikes, so we spike in the other sequence, so that optimal alignment would be achieved estimate that we achieved 98-99% correct classification. In these cases, by minimizing this measure. To generate a cross-correlation measure for the ambiguities were attributable to either the extreme attenuation of the CC-MAX method, we used the biweight kernel F(x) = (I — (W0)2] 2 spike amplitude during bursting (Yu and Margoliash, 19%) or sporadic for all < D, where D is a time window corresponding to the temporal background spikingactivity that could not be reliably distinguished from precision of the cross-correlation measure (set to 1.5 ms, a value chosen attenuated spikes in the recordings with the lowest SNR. These sites, to approximate the apparent temporal precision of RA premotor spike however, did not show any greater or lesser stability of temporal patterns patterns).The total CC ofspikc trains S,,..., Sk was defined as!,,,,K(S„ ofspike bursts—the principal dependent variable of this study—than did S,), where K(S„ = !Rs — t) over s in S, and tin The alignment those sites with completely reliable spike identification. maximized the total CC by shifting each spike train S, while preserving RA activity during singing is characterized as having high-frequency each individual bunt's interspilce intervals. bunts of spikes organized into trains of bursts. Each burst in the train of Once bursts within a stack were aligned, fine temporal structure was bunts is distinguished from the others both by the pattern of spikes and expressed as the tightly aligned spikes across renditions. A "feature" the timingof the bunt relative to the syllable (Yu and Margoliash, 1996; within a bunt was defined as a canonical spike, i.e.,a spike produced with Dave and Margoliash, 2000; Leonardo and Fee, 2005). In this study, we reliable timing relative to the other spikes in the bunt across many or all defined a bunt as a sequence of consecutive spikes with all interspike renditions. To identify and quantify features, we defined for each group intervals <10 ms. This simple definition reliably identified all bunts of ofspike trains an adjusted rate function,R(t) = — t11D), where two or more spikes ofan RA neuron during singing. In all cases, we also s is the time of an individual spike within the spike train S, and G(x) = could readily identify a canonical sequence ofbursts for each syllable (Yu (1 — x2) for all Ix] < D and 0 for all Ix] > D with the predefined time and Margoliash, 19%). Aligning multiple renditions of the sequences of window D = 1.2 ms. (Note that D = 1.2 ms results in a more precise bunts relative to the onset of a given syllable (as in a raster plot) created firing rate estimate than the 1.5 ms time window used for the original stacks ofbursts, with each "burst stack" associated with a particular time bunt alignment, achieving a coarse-to-fine alignment procedure.) We relative to syllable onset and a particular temporal pattern of spikes. We then identified peaks in the rate function. This method captures the identified 2.1 ± 1.3 bursts for each syllable across all the neurons, with changes in features we visually observed but is sensitive to the definition some syllables not eliciting any bunts and one neuron reliably emitting ofpeaks in the rate function, for example, slight changes in the temporal eight bunts for a particularly long and complex syllable. jitter of a given spike. The principal data set consisted of 115 distinct burst stacks emitted For a sample of N presleep spike trains, time T was identified as a during singing both before and after sleep by 15 RA neurons (seven feature location ifit satisfied four criteria: (1) the averaged adjusted rate birds). To compare the stability of temporal structure in premotor activ- function had a local peak at time Tli.e.,r(7) Z r(s) for s between T ± D, ity in the absence of sleep, we also examined the activity of RA neurons wherer( 7) = mean(R( Mover the sample]; (2) the peak at time Twas of recorded in periods of singing and/or calling that did not include sleep. significantly high amplitude compared with the variability of the rate We included in this data set the same 15 neurons used in the sleep function, Ir(T) 2 0.3 + , (0.975) X o (7), where a ( = SD(R( 7)) analysis, separating out the pre-sleep activity and postsleep activity into over the sample, and t,,,_, the inverse s-distribution function with N — I distinct sessions, each ofwhich did not include sleep (i.e., 115 bunt stacks from presleep recordings, and 115 burst stacks from postsleep record- degrees of freedom]; (3) the variability of spike times within the pre- ings, for a total of 230 burst stacks). To expand our data set to include defined time window around T was sufficiently low, IsN_ ,(0.975) X a sessions oflonger duration without sleep, we included the additional 28 ( 7)5 D, where a (7) = SD (spike times between T ± 0)1; and (4) the PA neurons recorded from 10 birds (six new, four that were also repre- average value of the adjusted rate function on either side of the peak fell sented in our sleep-inclusive recordings) in experiments where the lights off sufficiently quickly such that If] 5 2 ms, where 1equals the maximal were not turned out and the birds remained awake throughout, yielding interval containing T over which r(s) 2 r(T)13. Under these criteria, an additional 321 burst stacks. Thus, this "augmented" data set com- —65% of all spikes in premotor bunts were identified with located fea- prised a total of 551 distinct burst stacks recorded from 13 birds. tures (4.8 ± 3.1 total spikes/burst; 3.1 ± 1.8 features/burst). During one awake-only session, we also briefly recorded one putative We judged each burst stack as having a `structural change" across the RA interneuron characterized by a low baseline firing rate and an espe- sleep interval if three criteria were met. Pint, features in the presleep and cially narrow spike width (0.13 ms peak to trough,compared to a range of postsleep adjusted rate functions did not align well. Each feature was 0.19-0.41 ms for all other RA neurons we recorded), but we did not evaluated to determine whether we could rule out the existence of a include this unit in our analyses because of insufficient spike isolation corresponding spike in the corresponding stack (i.e., presleep vs during singing. postsleep). If for any feature there was no corresponding feature in the Analysis of burst structure and definition of features and structural corresponding stack within 0.25 ms, and there was no other peak within changes. To evaluate changes to the temporal structure of premotor 0.5 ms in the opposite stack's rate function with a magnitude statistically bunts across many renditions, we (I) developed a procedure to align all indistinguishable from that of the feature being evaluated, then the burst presleep or postsleep bunts for a given bunt stack, (2) generated func- stack was judged to meet this criterion. Second, there was a statistically tions that captured the temporal features of the aligned bursts, and (3) significant change in mean spike count of at least 0.5 spikes/burst. This evaluated the significance of any temporal or spike count differences criterion arises from the observed loss (or, rarely, gain) in spikes across between presleep and postsleep groups of spikes. sleep intervals (see Results). Third, to reduce the effect of artifacts in To optimally align bunt renditions within a presleep or postsleep alignment, structural changes were flagged only when the first two crite- bunt stack, we used two procedures: / 1-distance minimization (l.,- ria were satisfied under both L,-MIN and CC-MAX alignment proce- MIN), as described by Chi and Margoliash (2001), and cross-correlation dures. Overall, under both alignment procedures, 37 burst stacks maximization (CC-MAX). In both cases, the alignment of spike se- satisfied the first criterion, and 60 satisfied the second, with 33 satisfying quences was accomplished by iteratively shifting each burst rendition to both. Thus, changes in spike timing typically were associated with either globally minimize the summed L, distances (L,-MIN) or maxi- changes in spike rate, but the reverse was not generally the case. Those mize summed cross-correlation measures (CC-MAX) across all bunt bunt stacks found to undergo structural changes under these criteria EFTA01076048 2786 • 1. Neurosci., February 17,2010 • 3017):2783-2794 Rauske et al. • Neuronal Stability and Drift across Sleep corresponded well to those burst stacks that appeared to have altered pre-sleep . post-sleep spiking patterns under visual inspection. A Analysis for separator intern& other than sleep. Sleep is a natural sepa- !HUHU rator between groups of vocalizations, but we also explored whether . • II•01. • . . changes to premotor activity occurred at times other than sleep. To this ep ON' iN • ON end, for each burst stack we sought to identify the interval between con- OD MO • • II • •• PIMIC • • • -.1k I • secutive renditions of bunts that was most likely to correspond to a Man • • • • .. •• • .. . so • loss .. ay. , . . . . change in burst structure, referring to the interval thus identified as the • •• •• • • ,•. • • . similar `separator interval." We began by measuring the similarity ofall possible • • x • ••. • .. ... • . post-sleep • r 1 pairings ofindividual bursts within each burst stack, using theL, distance • • • • II.1 more metric described above; greater L, distance implies less similarity. Then, • I.' similar • • we considered each interval between bursts as a candidate separator in- terval, except that we excluded the first four and last four such intervals to avoid boundary effects. For each candidate interval, we divided the bursts 0.9 across-group into preinterval and postinterval groups, and from the collection of can- didate intervals we identified the one interval that maximized the differ- mean LI ence between the mean L, distances of across-group comparisons distance (preintenel vs postinterval) and within-group comparisons (preinterval 0.5 ten roux vs preinterval, or postinterval vs postinterval). This procedure tended to 10 30 identify two groups ofmost-similar bursts,one exclusivelybefore and the interval other exclusively after the interval, dividing the bunt stack at a moment time in time that often corresponded to a visibly noticeable change in burst structure (Fig. IA). B 15 We also used a modified procedure better suited to quantify a subset of the transitions in bunt structure. For these cases there was a distinct transition between distinct states, but with one of the states exhibiting less variability than the other. In these cases, the candidate interval that % of maximized the difference between mean L, distances did not always bursts correspond to the visually observed transition. We found that for these cases, the transition typically coincided with a candidate interval that maximized the differences comparingL, variances for across-group and postinterval (or preinterval) candidate intervals as well as maximizing the differences between L, means for across-group and preinterval (or postinterval) candidate intervals. Therefine, in these cases, we designated the interval thus defined as the separator interval; in all other cases, we simply used the interval maximizing the L,-mean distances between separator interval/ across-group and within-group comparisons as the separator interval. total # burst renditions Estimating occurrences of sleep-separator comparisons attributable to chance. Finally, we also developed a statistical procedure to compare the Figure 1. Identifying intervals with possible changes to premotor burst patterns. A, Corn- location of separator intervals in recordings that did and did not include parisons of pairs of burst renditions for a burst stack recorded before and after a period of sleep, sleep. To this end, we first identified separator intervals for the awake- using the L, distance metric. Each row and column represents a single burst rendition Dem. only recording sessions. Then, for each bunt stack, we calculated the seated by rasters to the left of rows a above columns). Each colored box represents Mel, proportion of bursts occurring before the separator to the total number &stance between the bursts denoted by the given row and column according to a color map ofbursts. A histogram of the resulting distribution suggested a quadratic with red indicating the highest distances (less similarity) and blue representing the lowest L, distribution, so we used a quadratic fit to generate a baseline probability &stances (more similarity). The sleep Interval is denoted by black dashed Ines. Note that burst density function (PDF) (Fig. I B). comparisons above and to the left of the sleep.interval Imes (I.e., comparisons between The PDF estimate allowed us to test the hypothesis that the distribu- preskep and postsleep bursts) show less similarity than 63 comparisom between bursts taking tion ofL,-optimized separator intervals was the same in awake-only and place exclusively before or after sleep. The graph at the bottom right shows the mean L, dis- sleep-inclusive recordings. In a bootstrap procedure, we sampled 115 tances between all pairs of burst renditions taking place across the separator interval(red line) fractions from the PDF and respectively multiplied these by the total &exclusively before& after the separator interval (blue line) for all possibk intervals. The sleep number ofrenditions for each of the 115 burst stacks in our data set to get intervalis &noted by the dashedline, where thedifferenceinmeanLi &stance between these a random separator interval. We repeated this procedure 10,000 times to twogroupsreaches a <leas peak; such a peak defines the optimized separator interval. 8, Esti- obtain a distribution of simulated separator intervals. This distribution mated probability distreutem function for the location of the optimized separator interval in was used to evaluate the likelihood that the number ofseparator intervals recording sessions that did not include sleep (551 burst sucks). the histogram shows the dia. we observed to correspond with the period of sleep (either exactly or tribution ofopeimized separator intervals relative to the total number &burst renditions within within one interval) would occur simply by chance. each recording session. A quadratic fit (line) was used to determine thePDF. The quadratic shape of the PDF can be explained as follows. The greater likelihood of locating separator intervals near the endpoint of an ings was maintained through a period of sleep and subsequent experiment rather than in the middle is most likely attributable to the vocalizations (Table I) (see Materials and Methods). It is likely exaggerated effect ofoutliers on small groups ofbunts. Theasymmetrical that all of these cells were projection neurons targeting the brain- shape of the PDF (see Results) may reflect a slightly increased variability stem, given their fast (>30 Hz), regular baseline spiking activity across bunt renditions later in experiments, when more time sometimes passed between singing bouts as birds became desensitized to the pres- and bursting activity during singing (Spiro et al., 1999; Leonardo ence of the adjacent female and tended to sing less frequently. and Fee, 2005). Each cell reliably burst with consistent timing relative to specific vocalizations such as a particular syllable or Results call; thus, raster plots of the neuronal activity aligned to vocaliza- We recorded from 43 RA single units while birds vocalized (sang tion onsets produced "stacks" of bursts, which were the basis for or called; including one "double unit" that we treat equivalently our analysis (see Materials and Methods). In the 37 cells for which to the single units in our analyses). Only a subset of these record- we recorded singing, there were 10.1 ± 4.2 unique burst stacks EFTA01076049 Rauste et al. • Neuronal Stability and Drift across keep J. Reurovi, February 17, 2010.30(71:2783-2794.2787 Table 1. Distribution of recordings across different experimental conditions cases. The changes persisted for as long as we could hold the Recording session type Birds Neurons Burst stacks recording—in one case, for several hours after waking (Fig. 3). Whatever the cellular or network effects that led to these changes, Sleep inclusive Tall) 7 15 115 Sleep inclusive (shod) 4 lob 83 they achieved their suprathreshold effects during sleep or imme- Sleep inclusive (long) 3 5 32 diately after awakening, and persisted thereafter. No sleep° 13 43° 551 To quantitatively assess these changes in burst structure, we 'Includesboth the pleserpenbt and pssuletpenls pleas el tit serpent:Mitt sessions as &Una no-slett aligned all presleep and postsleep bursts for each of the 115 burst moons. stacks using two algorithms—Le-distance minimization and 'Includes cee"dothkunt °turned as a snub nese:con outatubsts. cross-correlation maximization—converted these into probabi- listic rate functions, located features within those functions, and identified reliable changes in features, which we call "structural per song, and 1.9 -± 1.1 unique burst stacks per call. For the changes" (see Materials and Methods). Using these criteria, we remaining six cells for which we only recorded calls, there were found that for recordings spanning a sleep interval, 33 of 115 one to two unique burst stacks per call. burst stacks showed structural changes (Fig. 4). These 33 burst We examined the effects of sleep on vocalization-related neu- stacks were distributed across 10 of 15 neurons recorded across ral activity in adult male zebra finches under two protocols: short, sleep, with each neuron exhibiting one (four neurons), two (three interrupted periods of sleep and full, uninterrupted nights of neurons), or six (two neurons) bursts with structural changes, sleep. The distribution of recording sessions according to exper- but with one neuron exhibiting 11 bursts with structural changes. imental protocol is reported in Table I. Birds in the short-sleep The statistical significance of all the results that follow was main- design (at = 10 neurons in 4 birds) experienced a period of dark- tained even with the neuron with 11 bursts with structural ness lasting 90-179 min (average, 136 -± 31 min), whereas the changes removed. birds in the second design (n = 5 neurons in 3 birds) experienced Considering the sequence of syllables within motifs or the a full 8 -10 h of darkness. For all birds, during the first 2-10 min sequence of bursts within a syllable, there was no apparent ten- of darkness, birds typically rapidly transitioned between short dency for structural changes to be associated with bursts that periods of wake and sleep. Initially, in some cases, sudden awak- occurred in any particular syllable within the motif, or within any enings apparently resulted from playback of the bird's own song particular burst within a syllable. There was also no clear differ- that was presented during short-sleep sessions (see Materials and ence in the number of changes in burst stacks associated with Methods), but birds quickly habituated to the song playback and contact calls (5 of 21; 24%) compared to those associated with began to reliably sleep through the stimulus. Birds were judged to song syllables (28 of 94; 30%; p = 0.79, Fisher's exact test). begin an extended period of sleep when a full minute passed with We tested the hypothesis that the two different experimental no two consecutive 3 s intervals classified as "awake" (see Mate- designs affected the rate of occurrence of structural changes. rials and Methods) (see supplemental Fig. 1, available at www. There were no significant differences in the frequency of burst jneurosci.org as supplemental material). Based on the measure of changes between short-sleep and long-sleep birds. The frequen- spontaneous activity, the onset of extended periods of sleep began cies of structural changes [27% (22 of 83) short sleep, 34% (1 I of 10.5 -± 8.8 min (range, 1.3-29.5 min) after the start of subjective night, and represented 78.8 -± 10.5% (range 67.8 —94.9%) of the 32) long sleep; p = 0.40; x2 = 0.701 were similar under both total dark phase after sleep onset. Thus, whereas the recording