Mather and colleagues propose that phasic activation of LC neurons biases perception and memory. They suggest that elevated levels of glutamate at the site of prioritized representations increase local norepinephrine (NE) release, creating “NE hotspots.” At these spots, enhancement of glutamate and release of NE mutually enhance and amplify the activation of prioritized representations. This excitatory effect contrasts with the widespread suppression by NE of weaker representations via lateral and auto-inhibitory processes.
Mather et al. provide a schematic representation at the cellular level (Fig. 6), but is it supported by physiological data obtained in sensory cortices? Although the locus coeruleus (LC) neurons project widely to many cortical areas, recent data indicate that some neurons project more to one area (the prefrontal or motor cortex) than others (Chandler et al. Reference Chandler, Lamperski and Waterhouse2013; Reference Chandler, Gao and Waterhouse2014). Within an area, NE is released in the extracellular space from NE varicosities and reaches the entire cortical network. When sensory stimuli are processed by cortical neurons, glutamate is released by the thalamocortical terminals. How does NE affect cortical processing? Iontophoretic application of NE performed in the somatosensory, visual, and auditory cortices revealed that, in most of the cases, NE depressed evoked responses (e.g., Kolta et al. Reference Kolta, Diop and Reader1987; Manunta & Edeline Reference Manunta and Edeline1997; Reference Manunta and Edeline1998; Videen et al. Reference Videen, Daw and Rader1984), an effect replicated in awake animals (Bassant et al. Reference Bassant, Ennouri and Lamour1990; Foote et al. Reference Foote, Freedman and Oliver1975; Manunta & Edeline Reference Manunta and Edeline1999). Moreover, in awake rats, tonic activation of LC neurons by continuous low-frequency stimulation (1 Hz) triggered similar effects: decreased evoked responses in 63% of cells in the rat somatosensory cortex (Devilbiss & Waterhouse Reference Devilbiss and Waterhouse2004). It has been argued that these inhibitory effects were a consequence of the very high concentrations of NE in the vicinity of the cell (Waterhouse et al. Reference Waterhouse, Devilbiss, Fleischer, Sessler and Simpson1998a), but this seems unlikely given that pronounced depression of evoked responses was also observed with very low ejection currents (Ego-Stengel et al. Reference Ego-Stengel, Bringuier and Shulz2002; Manunta & Edeline Reference Manunta and Edeline1997; reviewed in Edeline Reference Edeline2012). If the hotspot theory were the main mechanism at play, then exogenous application of NA would more likely have increased evoked activity.
But what are the consequences for the neurons' functional properties? In the auditory cortex, the suppressive effect of NE promotes an increase in frequency selectivity in both anesthetized and unanesthetized animals (Edeline Reference Edeline1995; Manunta & Edeline Reference Manunta and Edeline1997; Reference Manunta and Edeline1999). In the visual cortex, application of NE improved the velocity and direction selectivity of cells, without modifying orientation selectivity (Ego-Stengel et al. Reference Ego-Stengel, Bringuier and Shulz2002; McLean & Waterhouse Reference McLean and Waterhouse1994). These results clearly point out that the effects of NE (and other neuromodulators) can differ depending on the stimulus dimension. For example, a dimension that depends on thalamocortical afferences (such as frequency tuning in the auditory cortex or size of the receptive field in the visual cortex) could be more affected than a dimension that relies more on corticocortical afferences (such as frequency modulation tuning in the auditory cortex or velocity tuning in the visual cortex). Yet, glutamate is released in all cases, indicating that the glutamate–norepinephrine interaction is not as straightforward as described by the authors.
One may ask if it is possible for “NE hotspots” to emerge when NE is repeatedly associated with glutamate release at particular synapses? When a stimulus that activates a specific set of synapses is paired with phasic LC stimulation, a predominant decrease in neuronal activity is initially reported in several cortical areas (Olpe et al. Reference Olpe, Glatt, Laszlo and Schellenberg1980; Sato et al. Reference Sato, Fox and Daw1989). However, in the somatosensory cortex, both the excitatory and inhibitory components of evoked responses are facilitated when phasic stimulation of the LC is delivered before tactile stimuli (Snow et al. Reference Snow, Andre and Pompeiano1999; Waterhouse et al. Reference Waterhouse, Moises and Woodward1998b). Rather, in several sensory modalities, LC stimulation affects the temporal organization of evoked responses (Bouret & Sara Reference Bouret and Sara2002), by shortening evoked responses and reducing the first spike latency and its variance (Lecas Reference Lecas2001; Reference Lecas2004). In the auditory cortex, facilitation of evoked responses is the dominant effect when stimulation of the LC is delivered before a particular sound frequency (Edeline et al. Reference Edeline, Manunta and Hennevin2011; Martins & Froemke Reference Martins and Froemke2015), which contrasts with the dominant depressive effect obtained when phasic pulses (1 s) of NE are iontophoretically delivered in the vicinity of the recorded cells (Manunta & Edeline Reference Manunta and Edeline2004). In fact, via its projections onto other brain nuclei, LC stimulation activates other neuromodulatory systems such as the cholinergic (Berridge and Foote Reference Berridge and Foote1991; Berridge et al. Reference Berridge, Page, Valentino and Foote1993) and serotoninergic (Kim et al. Reference Kim, Lee, Lee and Waterhouse2004) systems. The difference between the effects obtained with iontophoretic application of NE and LC stimulation indicates that other neuromodulators likely contribute to the effects observed at the cortical level.
Mather et al. also suggested that local “NE hotspots” bias perception. But is there any evidence that NE facilitates the discrimination performance of cortical neurons? When the responses of cortical neurons are tested with a set of conspecific and heterospecific vocalizations having the same spectral content (thereby activating the same sets of synapses), NE application induces either an increase or a decrease in the response, and, on average, discrimination of overall performance of cortical neurons is unaffected. However, a small population of neurons displaying the largest increase in responses exhibit enhanced discrimination (quantified by the mutual information) between communication sounds (Gaucher & Edeline Reference Gaucher and Edeline2015). This population of neurons corresponds to recordings showing smaller receptive fields; they are not located in a particular layer and are distributed across the whole cortical map. As for any natural stimulus, representations of these vocalizations are distributed among largely overlapping cortical and subcortical networks. Therefore, having local “NE hotspots” in such networks will not help perception of such stimuli. Subtle, and transient, synchronizations between distributed populations of neurons are potentially more relevant for discriminating natural stimuli.
To conclude, based on physiological results from sensory cortices, it is clear that the dominant effect of NE is widespread suppression of excitatory inputs when responses are tested both with artificial and with natural stimuli. NE promotes facilitated responses for only a small proportion of cells; this effect is more common with LC activation, probably because of activation of other neuromodulatory systems. Only a small fraction of these facilitated responses displayed enhanced discriminative performance. Therefore, the local “NE hotspot” theory receives little support from current physiological results. Although we cannot dismiss that the “hotspot theory” might account for certain specific forms of perceptual alterations, we believe this theory is inappropriate to account for facilitations of perceptive abilities. It remains to be demonstrated that local interactions between NE and glutamate promote enhanced perceptive abilities, particularly because of the inherent distributed representation of any natural stimuli.
You have
Access
Mather and colleagues propose that phasic activation of LC neurons biases perception and memory. They suggest that elevated levels of glutamate at the site of prioritized representations increase local norepinephrine (NE) release, creating “NE hotspots.” At these spots, enhancement of glutamate and release of NE mutually enhance and amplify the activation of prioritized representations. This excitatory effect contrasts with the widespread suppression by NE of weaker representations via lateral and auto-inhibitory processes.
Mather et al. provide a schematic representation at the cellular level (Fig. 6), but is it supported by physiological data obtained in sensory cortices? Although the locus coeruleus (LC) neurons project widely to many cortical areas, recent data indicate that some neurons project more to one area (the prefrontal or motor cortex) than others (Chandler et al. Reference Chandler, Lamperski and Waterhouse2013; Reference Chandler, Gao and Waterhouse2014). Within an area, NE is released in the extracellular space from NE varicosities and reaches the entire cortical network. When sensory stimuli are processed by cortical neurons, glutamate is released by the thalamocortical terminals. How does NE affect cortical processing? Iontophoretic application of NE performed in the somatosensory, visual, and auditory cortices revealed that, in most of the cases, NE depressed evoked responses (e.g., Kolta et al. Reference Kolta, Diop and Reader1987; Manunta & Edeline Reference Manunta and Edeline1997; Reference Manunta and Edeline1998; Videen et al. Reference Videen, Daw and Rader1984), an effect replicated in awake animals (Bassant et al. Reference Bassant, Ennouri and Lamour1990; Foote et al. Reference Foote, Freedman and Oliver1975; Manunta & Edeline Reference Manunta and Edeline1999). Moreover, in awake rats, tonic activation of LC neurons by continuous low-frequency stimulation (1 Hz) triggered similar effects: decreased evoked responses in 63% of cells in the rat somatosensory cortex (Devilbiss & Waterhouse Reference Devilbiss and Waterhouse2004). It has been argued that these inhibitory effects were a consequence of the very high concentrations of NE in the vicinity of the cell (Waterhouse et al. Reference Waterhouse, Devilbiss, Fleischer, Sessler and Simpson1998a), but this seems unlikely given that pronounced depression of evoked responses was also observed with very low ejection currents (Ego-Stengel et al. Reference Ego-Stengel, Bringuier and Shulz2002; Manunta & Edeline Reference Manunta and Edeline1997; reviewed in Edeline Reference Edeline2012). If the hotspot theory were the main mechanism at play, then exogenous application of NA would more likely have increased evoked activity.
But what are the consequences for the neurons' functional properties? In the auditory cortex, the suppressive effect of NE promotes an increase in frequency selectivity in both anesthetized and unanesthetized animals (Edeline Reference Edeline1995; Manunta & Edeline Reference Manunta and Edeline1997; Reference Manunta and Edeline1999). In the visual cortex, application of NE improved the velocity and direction selectivity of cells, without modifying orientation selectivity (Ego-Stengel et al. Reference Ego-Stengel, Bringuier and Shulz2002; McLean & Waterhouse Reference McLean and Waterhouse1994). These results clearly point out that the effects of NE (and other neuromodulators) can differ depending on the stimulus dimension. For example, a dimension that depends on thalamocortical afferences (such as frequency tuning in the auditory cortex or size of the receptive field in the visual cortex) could be more affected than a dimension that relies more on corticocortical afferences (such as frequency modulation tuning in the auditory cortex or velocity tuning in the visual cortex). Yet, glutamate is released in all cases, indicating that the glutamate–norepinephrine interaction is not as straightforward as described by the authors.
One may ask if it is possible for “NE hotspots” to emerge when NE is repeatedly associated with glutamate release at particular synapses? When a stimulus that activates a specific set of synapses is paired with phasic LC stimulation, a predominant decrease in neuronal activity is initially reported in several cortical areas (Olpe et al. Reference Olpe, Glatt, Laszlo and Schellenberg1980; Sato et al. Reference Sato, Fox and Daw1989). However, in the somatosensory cortex, both the excitatory and inhibitory components of evoked responses are facilitated when phasic stimulation of the LC is delivered before tactile stimuli (Snow et al. Reference Snow, Andre and Pompeiano1999; Waterhouse et al. Reference Waterhouse, Moises and Woodward1998b). Rather, in several sensory modalities, LC stimulation affects the temporal organization of evoked responses (Bouret & Sara Reference Bouret and Sara2002), by shortening evoked responses and reducing the first spike latency and its variance (Lecas Reference Lecas2001; Reference Lecas2004). In the auditory cortex, facilitation of evoked responses is the dominant effect when stimulation of the LC is delivered before a particular sound frequency (Edeline et al. Reference Edeline, Manunta and Hennevin2011; Martins & Froemke Reference Martins and Froemke2015), which contrasts with the dominant depressive effect obtained when phasic pulses (1 s) of NE are iontophoretically delivered in the vicinity of the recorded cells (Manunta & Edeline Reference Manunta and Edeline2004). In fact, via its projections onto other brain nuclei, LC stimulation activates other neuromodulatory systems such as the cholinergic (Berridge and Foote Reference Berridge and Foote1991; Berridge et al. Reference Berridge, Page, Valentino and Foote1993) and serotoninergic (Kim et al. Reference Kim, Lee, Lee and Waterhouse2004) systems. The difference between the effects obtained with iontophoretic application of NE and LC stimulation indicates that other neuromodulators likely contribute to the effects observed at the cortical level.
Mather et al. also suggested that local “NE hotspots” bias perception. But is there any evidence that NE facilitates the discrimination performance of cortical neurons? When the responses of cortical neurons are tested with a set of conspecific and heterospecific vocalizations having the same spectral content (thereby activating the same sets of synapses), NE application induces either an increase or a decrease in the response, and, on average, discrimination of overall performance of cortical neurons is unaffected. However, a small population of neurons displaying the largest increase in responses exhibit enhanced discrimination (quantified by the mutual information) between communication sounds (Gaucher & Edeline Reference Gaucher and Edeline2015). This population of neurons corresponds to recordings showing smaller receptive fields; they are not located in a particular layer and are distributed across the whole cortical map. As for any natural stimulus, representations of these vocalizations are distributed among largely overlapping cortical and subcortical networks. Therefore, having local “NE hotspots” in such networks will not help perception of such stimuli. Subtle, and transient, synchronizations between distributed populations of neurons are potentially more relevant for discriminating natural stimuli.
To conclude, based on physiological results from sensory cortices, it is clear that the dominant effect of NE is widespread suppression of excitatory inputs when responses are tested both with artificial and with natural stimuli. NE promotes facilitated responses for only a small proportion of cells; this effect is more common with LC activation, probably because of activation of other neuromodulatory systems. Only a small fraction of these facilitated responses displayed enhanced discriminative performance. Therefore, the local “NE hotspot” theory receives little support from current physiological results. Although we cannot dismiss that the “hotspot theory” might account for certain specific forms of perceptual alterations, we believe this theory is inappropriate to account for facilitations of perceptive abilities. It remains to be demonstrated that local interactions between NE and glutamate promote enhanced perceptive abilities, particularly because of the inherent distributed representation of any natural stimuli.