We investigated the effects of cortical desynchronization on spontaneous firing rates in identified neural populations of auditory cortex. Both PCs in superficial layers and putative FSs reduced their activity during desynchronized state, but diverse effects were seen in PCs of other layers. The ability of desynchronization to enhance neuronal spiking was positively correlated with a baseline firing rate of PCs, but not FSs. Results were generally consistent for different types of desynchronization (PPT-evoked and spontaneous desynchronization under urethane anesthesia, spontaneous desynchronization in unanesthetized animals).
Superficial layer PCs are sparsely active (Sakata and Harris, 2009
). Their decreased firing rate during the desynchronized state is in a sense surprising: whereas during the synchronized state neurons can only fire during up states, this constraint is removed by desynchronization. This observation, together with the diverse rate changes observed in other cell-types, suggests that the effects of cortical state on mean spontaneous firing rate is not simply due to a change in the total amount of “up state time,” but reflects a true state-dependent modulation of excitability that differs between cell classes.
A number of studies have investigated the effects of cortical state on auditory cortical function (Edeline et al., 2001
; Issa and Wang, 2008
; Lakatos et al., 2008
; Curto et al., 2009
; Otazu et al., 2009
; Harris and Thiele, 2011
; Marguet and Harris, 2011
), though to our knowledge none have yet identified its effects on different morphological cell classes. This issue has however been addressed in other modalities. In the somatosensory barrel cortex, active whisking, a behavior that is accompanied by cortical desynchronization (Crochet and Petersen, 2006
; Poulet and Petersen, 2008
), was found to suppress the spontaneous activity of L2/3 PCs but cause enhanced firing in a subset of L5 PCs (De Kock and Sakmann, 2009
). Whisking-related desynchronization suppressed FS activity while enhancing spiking of non-FSs (Gentet et al., 2010
). Application of cholinergic agonists to thalamus under urethane anesthesia also causes a desynchronized state in the barrel cortex in which L2/3 PCs are suppressed; however, in contrast to our results and observations in superficial layers (Gentet et al., 2010
), FSs in thalamic recipient layer were strongly enhanced by this manipulation (Hirata and Castro-Alamancos, 2010
). In the visual cortex, basal forebrain stimulation causes a decrease in multiunit firing rates in superficial layers, but a mean increase in all others (Goard and Dan, 2009
), largely consistent with our results; however, a second study (Niell and Stryker, 2010
) found that running in a head-restrained condition (also expected to cause desynchronization) had little effect on spontaneous firing rates, except for a subset of putative FSs which showed a dramatic increase. The differences between these results—such as diverse effects on FSs—may result from either differences in the form of desynchronizing conditions (e.g., drug administration, induced cortical activations, active behaviors), or from differences between cortical areas and/or layers. Nevertheless, a common picture does appear to emerge from these multiple studies, in particular with desynchronization evoking suppression of PC activity in the superficial layers and diverse effects elsewhere, consistent with our results. In addition, although effects of electrical PPT stimulation on downstream structures are likely complex (Steriade et al., 1991
; Jones, 2003
; Winn, 2006
; Schofield et al., 2011
), we observed consistent results across different conditions.
It has been suggested that spiking phases (up states) of the synchronized state consist of fragmentary moments of cortical activity similar to wakefulness (Destexhe et al., 2007
). Our data indicate at least one way in which this notion is incomplete (see also Ushimaru et al., 2012
). Superficial-layer activity during up states is sparse (Sakata and Harris, 2009
); however, desynchronized state firing rates are rather lower than synchronized state rates (which also include silent phases of the synchronized state). The mechanisms of this decrease in rate are currently uncertain. It seems improbable that it results from inhibition, at least from fast-spiking cells, as putative FSs also reduce their firing rates during desynchronized epochs. Neuromodulatory systems are active during desynchronization (Pace-Schott and Hobson, 2002
; Jones, 2003
; Harris and Thiele, 2011
), which may hyperpolarize certain classes of neuron (Gulledge and Stuart, 2005
; Gulledge et al., 2007
). Another possibility is firing rate adaptation. If superficial layer neurons showed particular tendency of spike adaptation, one might expect them to fire strongly in a spiking phase of the synchronized period, but not be able to sustain this activity in prolonged desynchronized state activity. Indeed, superficial PCs show stronger adaptation than at least fast-spiking cells in the auditory cortex (Schiff and Reyes, 2012
), and lower PC activity might in turn lead to lower activity of fast-spiking cells.
What function might laminar-dependent changes in firing rates have for information coding? The reduction of spontaneous firing in superficial cells is likely to increase signal-to-noise ratios (SNRs) to encode sensory signals (Livingstone and Hubel, 1981
). Since L2/3 population activity is sparse and spatially localized (Sakata and Harris, 2009
; Bathellier et al., 2012
; Harris, 2012
) and sparse coding is in general beneficial to readout signals for downstream networks (Olshausen and Field, 2004
), the improvement of SNRs could be further beneficial for their downstream targets. In deep layers, on the other hand, increased baseline firing in a subset of PCs might appear to increase baseline “noise”. However, because this baseline activity has a tonic structure rather than exhibiting alternating up and down phases, desynchronization could instead result in the reduction of output fluctuations from those cells compared to the synchronized state. We therefore suggest that it could allow downstream networks to detect subtle deviations, with the increased baseline activity helping to reach threshold for weak signals as well.
In conclusion, we assessed the effect of cortical states on spontaneous spiking activity across cell-types of the auditory cortex. Spiking activity in superficial PCs and putative FSs is suppressed during desynchronization. We hypothesize that this reduction of internal noise is beneficial for the brain to detect behaviorally relevant stimuli.
Conflict of interest statement
The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.