Cholinergic Cells of the Pontomesencephalic Tegmentum: Connections with Auditory Structures from Cochlear Nucleus to Cortex

2.1. Acetylcholine in the IC

Curtis and Koizumi (1961) looked at ACh effects on spontaneous activity of IC cells. Application of cholinergic agents affected a small minority (28/273) of IC units. Perhaps as a result of this observation, ACh in the IC was largely ignored for many years. More recent studies looked at ACh effects on the responses of IC cells to acoustic stimuli, and found that ACh changes the responses of the majority (60–85%) of IC cells (Farley et al., 1983; Habbicht and Vater, 1996; Watanabe and Simada, 1973). The conclusion, then, is that ACh has little effect on the spontaneous activity of IC cells, but can affect the responses of most IC cells to sound stimuli. Such a pattern of effects can serve as a definition of a neuromodulator, and is consistent with cholinergic effects in other systems (e.g., Timofeeva et al., 2005). The widespread cholinergic effects are consistent with a distribution of nicotinic and muscarinic ACh receptors throughout the IC (Clarke et al., 1985; Glendenning and Baker, 1988; Morley and Happe, 2000; Rotter et al., 1979). It has been suggested that ACh enhances unit responses to specific stimuli (i.e., improves signal/noise ratio) and may set the level of excitability of IC cells in general or in association with particular stimuli (Farley et al., 1983; Habbicht and Vater, 1996; Miko and Sanes, 2009). The cholinergic effects are mediated, at least in part, by direct actions on GABAergic cells or synaptic terminals in the IC (Miko and Sanes, 2009; Yigit et. al., 2003). Whether cholinergic axons contact non-GABAergic IC cells is not known. Interest in cholinergic innervation of the IC has been expanded by the finding that blockade of muscarinic receptors in the IC interferes with some forms of plasticity (e.g., that induced by fear conditioning; Ji et al., 2001; 2005), but not with other forms of plasticity (e.g., that induced by pseudo-conditioning; Ji and Suga, 2009).

There is no evidence for cholinergic cells within the IC (Motts et al., 2008), so the cholinergic innervation must arise from extrinsic sources. We identified cholinergic inputs to the IC in guinea pigs by combining retrograde tracing with immunohistochemistry for choline acetyltransferase (ChAT), a specific marker of cholinergic cells (Motts and Schofield, 2009a). The experiments labeled ChAT-immunopositive cells in the pedunculopontine tegmental nucleus and the laterodorsal tegmental nucleus (PPT and LDT, respectively). The following sections describe these cholinergic nuclei and evidence for extensive connections between the cholinergic cells and many auditory structures.

2.2. Cholinergic cells of the pontomesencephalic tegmentum

The PPT and the LDT are prominent groups of cholinergic cells in the tegmentum of the pons and midbrain (Fig. 1). We will refer to the PPT and LDT collectively as the pontomesencephalic tegmentum, or PMT (Woolf and Butcher, 1989), and refer to distinctions between the PPT and LDT where appropriate. The PMT nuclei have widespread projections to the spinal cord, brainstem sensory and motor nuclei, reticular formation, thalamus, basal forebrain and frontal cortex (Garcia-Rill, 1991; Jenkinson et al., 2009; Jones, 2003; Smith et al., 1988; Steriade et al., 1988). The projections are associated with arousal, sensory gating, control of the sleep-wake cycle, orienting responses, and reward behavior (Chen et al., 2006; Pan and Hyland, 2005; Reese et al., 1995c; Rye, 1997; Swerdlow et al., 2001)². Among the earliest functions to be studied was the role of the PMT as part of the ascending reticular activating system. Stimulation of the PMT leads to an activated EEG pattern (Steriade et al., 1991). The cholinergic cells are active during waking and during paradoxical sleep (a.k.a. REM sleep), and less active during slow wave sleep (Boucetta and Jones, 2009). Their projections to the thalamus can alter the firing mode of thalamic relay cells, and thus modulate information flow in ascending pathways to the cerebral cortex. These projections terminate in most or all thalamic nuclei, including the medial geniculate body (MG), the auditory nucleus of the thalamus. In general, the similarities in the patterns of PMT projections to various thalamic nuclei led to assumptions of a common function across thalamic systems. However, the literature revealed possible differences between PMT projections to the MG versus other thalamic nuclei. The next section describes the apparent differences and new data that re-examine these issues.

2.3. Acetylcholine in the MG

Retrograde tracing in several species (rats, cats, macaques) shows that the cholinergic projection to most thalamic nuclei arises bilaterally in the PPT and LDT (Hallanger et al., 1987; Shute and Lewis, 1967; Steriade et al., 1988). The ipsilateral projection is predominant and, on each side, the PPT contributes more cells than the LDT. However, both retrograde and anterograde tracing experiments in rats suggest that there is no projection to the MG from the LDT (Cornwall et al., 1990; Hallanger et al., 1987; Satoh and Fibiger, 1986). The lack of a projection from the LDT to the MG suggests a unique cholinergic innervation for the auditory thalamus.

We combined retrograde tracing and immunohistochemistry to identify cholinergic projections to the MG in guinea pigs (Figure 2; Motts and Schofield, 2010). Retrogradely-labeled, immunopositive (i.e., cholinergic) cells were located in the PMT bilaterally. Contrary to expectations based on the rat experiments, our data in guinea pigs include a projection from cells in the LDT, revealing a cholinergic innervation of MG similar to that of other thalamic nuclei.

2.4. Collateral projections from cholinergic cells

The PMT contains a relatively small number of cells – about 40,000 cells in humans (Manaye et al., 1999); 7000 cells in guinea pigs (Leonard et al., 1995) and 6000 cells in rats (Jones, 1990). Given the widespread projections, perhaps it is not surprising that PMT cells often have collateral projections to multiple targets. Numerous studies in rats show that individual PMT cells project to the superior colliculus and the lateral geniculate nucleus (Billet et al., 1999), to the fastigial nucleus and rostral ventrolateral reticular nucleus (Ruggiero et al., 1997), or to the thalamic reticular nucleus and the cerebral cortex (Jourdain et al., 1989). One study suggests that individual PMT cholinergic cells can innervate bilaterally symmetrical targets (e.g., left and right lateral geniculate nuclei; Turlejski et al., 1994). Collateral projections could allow a small number of cells to 1) innervate a large number of targets, and, 2) to coordinate activity in those targets, and, possibly, 3) to enhance transmission of information through a pathway by modulating the responsiveness of cells at multiple points along the pathway. We found evidence for collateral projections to both ICs (Fig. 3A; Motts and Schofield, 2009a). We were interested in determining whether the PMT cells that project to the IC are the same ones that project to the MG. We tested this by combining multiple retrograde tracers with ChAT immunohistochemistry in guinea pigs (Motts and Schofield, 2009b).

We tested several possible patterns of projections to two targets: 1) to an MG and an IC on the same side of the brain; 2) to an MG and an IC on opposite sides; 3) to the left and right MGs. We found examples of each pattern. The most commonly observed pattern was from a PMT cell to the ipsilateral MG and ipsilateral IC (Fig. 3B). The second most common pattern was to the left and right MGs. Other patterns included cells that project to 3 targets (one IC and both MGs). We conclude that individual cholinergic cells can innervate bilateral targets as well as targets at different levels of the auditory pathways.

2.5. Acetylcholine in the cochlear nucleus (CN)

Cholinergic innervation of the CN is well established (Caspary et al., 1983; Comis and Whitfield, 1968; Fujino and Oertel, 2001; Godfrey et al., 1987, 2000; Happe and Morley, 1998; Henderson and Sherriff, 1991; Sherriff and Henderson, 1994). Sherriff and Henderson (1994) provide the only study in which retrograde tracers were injected into the CN and combined with immunohistochemistry to identify the source(s) of ACh. They identified cholinergic cells in the superior olivary complex, but not in the PMT, as the source of ACh in the CN of rats. We sought to expand the observations to guinea pigs, in which we had identified PMT connections to the IC and MG, and thus to provide a basis for contrasting within a single species the two auditory-related cholinergic systems in the brainstem (Motts and Schofield, 2005). The results were unexpected.

Injection of a retrograde tracer into the CN labeled many cells bilaterally in the superior olivary complex, in a pattern similar to that described in rats (Sherriff and Henderson, 1994). As expected, a subset of these cells were immunopositive for ChAT. In addition, there were a substantial number of retrogradely-labeled, ChAT-positive cells in the PMT. The PMT cells constituted about one quarter of the retrogradely-labeled, ChAT-positive cells. This result indicates that, at least in guinea pigs, the CN receives cholinergic innervation from two different sources: the superior olivary complex and the PMT (Motts and Schofield, 2005). It also demonstrates that PMT cholinergic projections innervate the auditory pathways from the CN to the MG.

3.1. Inputs from auditory cortex

Several studies in rats focused on inputs to the PMT (Cornwall et al., 1990; Satoh and Fibiger, 1986; Semba and Fibiger, 1992; Steininger et al., 1992). Inputs from cerebral cortex originate in frontal and limbic areas. There had been no indications of inputs from the AC until, in the course of studying AC projections to other subcortical targets, we discovered direct projections from the primary auditory cortex (A1) to the PMT in guinea pigs (Fig. 4A, B; Schofield and Motts, 2009). One limitation of both the anterograde and the retrograde tracing studies is that they do not directly associate the inputs with the cholinergic cells in the PMT. We combined anterograde tracing from A1 with immunohistochemistry for ChAT in guinea pigs. We observed presumptive synaptic contacts between AC boutons and cholinergic cells in all the PMT nuclei (i.e., PPT and LDT, bilaterally). The results are the first report of AC projections to the PMT and, to the best of our knowledge, the first report to associate projections from any primary sensory cortical area with cholinergic cells in the PMT.

3.2. Lack of inputs from visual cortex

We were curious about the possibility that other primary sensory areas project to the PMT. In experiments not previously published, we made large injections of fluorescein dextran into the right visual cortex and FluoroRuby into the left visual cortex in each of two guinea pigs (see section 8 for Methods). Figure 4C-E shows a fluorescein dextran injection site and resulting label in the thalamus and the pontine nuclei. We could find no evidence for labeled axons in the PMT (Fig. 4F). This finding is consistent with previous studies of inputs to the PMT (Cornwall et al., 1990; Semba and Fibiger, 1992; Steininger et al., 1992). We conclude that the recently discovered projection to PMT from the AC is not representative of a general projection from all primary sensory cortices. Nevertheless, it may be of interest to use the same anterograde tracing methods that revealed AC projections to examine possible projections from higher order visual areas or other sensory regions (e.g., somatosensory cortex).

3.3. Cells of origin of cortical projections to the PMT

Retrograde tracers injected into the PPT labeled pyramidal cells in layer V of the AC, confirming the anterograde results and suggesting that the projection to the PMT is glutamatergic (Schofield and Motts, 2009). No labeled cells were present in the visual cortex, consistent with the results from the anterograde tracer injections described in the previous section.

6.1. Habituation and novelty detection

Approximately half of the neurons in the PMT respond to auditory stimuli (Koyama et al., 1994; Reese et al., 1995a, b). Many PMT cells are also multimodal, responding to two or more sensory modalities (audition, touch and vision). The presence or absence of habituation (along with other distinctions) served to differentiate two populations of PMT cells. One population responded consistently to repeated stimuli and a second population responded only to the first one or two in a series of identical stimuli. This habituation may play an important role in an organism’s ability to detect and attend to novel stimuli (Koyama et al., 1994). Interestingly, “novelty” neurons (a.k.a. neurons that show “stimulus-specific adaptation, or SSA”) have been described recently in the IC (Malmierca et al., 2009; Pérez-González et al., 2005). A majority of IC cells show some degree of SSA. Those that show the strongest SSA have some intriguing similarities with PMT cells that habituate. First, these IC cells respond only at the onset of a stimulus, like the PMT cells. Second, both groups of cells show broad frequency tuning. If this form of adaptation in the IC depends on the PMT cholinergic input, one would predict that application of cholinergic blockers to the IC cell would eliminate the selectivity for novel stimuli (i.e., the IC neuron would either cease responding at all or would lose its adaptation and continue to respond to a repeated stimulus). Given the multimodal nature of many PMT cells, it would be interesting to determine whether a non-auditory stimulus – say, a flash of light embedded in a series of auditory stimuli - would overcome the habituation of an IC cell to the auditory stimulus (i.e., the cell would respond to the “same” auditory stimulus when it follows the flash of light), even if the IC cell did not respond directly to the visual stimulus.

6.2. Sensory gating

The PMT has also been associated with sensory gating. A popular model for studying gating is prepulse inhibition of the acoustic startle reflex. This reflex includes a blink, contraction of facial, neck and skeletal muscles, and an acceleration of heart rate in response to a sudden sensory stimulus (Koch, 1999). Startle can be induced by various modalities, including sight, sound and touch. For acoustic startle, the circuitry includes neurons in the CN that synapse onto cells in the caudal pontine reticular nucleus (PnC), which in turn contact cranial and spinal motor neurons that drive the motor response (Koch, 1999). The startle response can be inhibited by a less intense stimulus presented before the startling stimulus, a phenomenon known as prepulse inhibition of startle (Hoffman and Ison, 1980). This inhibition is mediated by projections of PMT cholinergic cells to the PnC (Bosch and Schmid, 2008; Koch, 1993; Swerdlow et al., 2001; Yeomans et al., 2006). It has been proposed that the prepulse inhibition allows the organism to continue processing sounds without the interruption that would be induced by a startle response (Graham, 1975). One can speculate that the PMT projections to structures in the ascending auditory pathways could enhance responsiveness or transmission of the sensory information during the time when the motor components of the startle are being inhibited. Wang et al., (2007) have shown that a “pre-pulse” of weak noise can alter response thresholds and frequency tuning of IC cells. These changes are dependent on GABAergic circuits in the IC. The possibility that cholinergic inputs to the IC can activate the same GABAergic circuits is supported by evidence that ACh activates IC GABAergic cells directly (Yigit et al., 2003).

Interestingly, “higher” functions can modulate prepulse inhibition of startle (Li et al., 2009). Both attention and emotional salience of the prepulse can affect the subsequent inhibition. Various forebrain areas, including the prefrontal cortex, AC and amygdala, have been implicated. Prefrontal cortex projects directly to the PMT (Sesack et al., 1989). The amygdala could potentially affect the PMT via direct projections or via projections to the superior colliculus or locus ceruleus. Finally, the AC projects to the IC (part of the circuit for prepulse inhibition) and, as shown previously, also projects directly to the PMT. Deciphering the contributions of these inputs to sensory gating is an ongoing challenge. One reason for continued interest in this issue is the well-documented abnormalities in sensory gating in such disorders as schizophrenia (Braff et al., 2001; Geyer et al., 2001). Continued advancement in identifying cholinergic circuits and the descending pathways that modulate them may be particularly important for understanding sensory processing in various disorders.

6.3. Motivation and reward

The PMT may serve as an interface between sensory and motivational systems (Pan and Hyland, 2005). The PMT has been associated with reward behaviors, including conditioned place preference, intravenous drug self-administration, and intracranial electrical self-stimulation (Chen et al., 2006). An important part of the PMT circuitry related to reward is the pathway from cholinergic PMT cells to dopaminergic cells in the ventral tegmental area (Chen et al., 2006; Pan and Hyland, 2005). It would be interesting to determine whether PMT cells that participate in reward behaviors project to auditory nuclei. For example, do individual cells project to both the ventral tegmental area and the auditory thalamus or midbrain? Such projections may help to identify locations in the auditory pathways where reward signals are integrated with sensory processing.

6.4. Arousal

The cholinergic projections to the thalamus have been associated with arousal and control of the sleep-wake cycle. It seems likely that PMT projections to other auditory structures could serve a similar function. Velluti and colleagues have shown that the responses of neurons at many levels of the auditory system change as an animal passes from waking to slow-wave sleep to paradoxical sleep (Goldstein-Daruech, 2002; Pedemonte et al., 1994; Peña et al., 1992, 1999). These results demonstrate the continued processing of acoustic stimuli even while an individual sleeps.

A loud sound could presumably activate the PMT and lead to waking of a sleeping individual or increased vigilance in an awake individual. As discussed above, PMT projections may contribute to novelty detection, or stimulus specific adaptation, in auditory nuclei. The novelty of a stimulus may itself be a factor in inducing arousal, including waking from sleep (Ruby et al., 2008). Several additional aspects of arousal are of interest in view of the descending projections from forebrain to the PMT. The salience of a stimulus clearly affects the ability of a stimulus to arouse or to capture attention. This may occur in the cocktail party effect, whereby a signal, such as one’s name, “pops out” of the background noise and can redirect a listener’s attention. A similar phenomenon occurs during sleep, when a person’s own name is more effective at waking him or her than someone else’s name (Oswald et al., 1960; Bastuji et al., 2002). It seems likely that cholinergic projections from the PMT to the auditory nuclei could play a role in these aspects of arousal and attention.

6.5. Top-down effects and attention

An area of particular interest concerns the broad descending projections from the AC that terminate directly at all levels of the central auditory system (reviewed by Schofield, 2010b). These pathways provide a basis for top-down effects on auditory processing, allowing the AC to modify how sounds are analyzed and to control how, and what, information ascends to higher levels. Functions attributed to this pathway include binaural processing (Nakamoto et al. 2008), encoding of stimulus frequency, timing and amplitude (Suga 2008; Xiong et al. 2009) and behaviorally-driven learning (Bajo et al. 2010). The underlying mechanisms are only beginning to be identified, and may include plasticity at both cortical and subcortical levels (Suga 2008; Winkowski and Knudsen 2008; Xiong et al. 2009; He and Yu 2010).

Knudsen and colleagues emphasized the association between cholinergic systems and mechanisms of attention that could be particularly relevant for experience-dependent plasticity (e.g., Keuroghlian and Knudsen, 2007; Winkowski and Knudsen, 2008). The projections from the AC to brainstem cholinergic cells that project to subcortical auditory nuclei could assure that ACh is released concurrent with the direct effects of corticofugal activity in those auditory nuclei. It is interesting to note in this context that cholinergic projections appear to terminate in the major sites targeted by direct AC projections (i.e., the MG, IC and CN). More experiments are needed to explore the relationships of the cholinergic inputs and activity in the corticofugal pathway.

6.6. Tinnitus

A maladaptive form of plasticity may underlie tinnitus, a common condition that has been associated with changes - usually an increase but in some cases a decrease - in the firing rates of neurons in several auditory areas (Brozoski et al., 2002; 2005; Eggermont and Kenmochi, 1998; Kaltenbach and Afman, 2000; Norena and Eggermont, 2003; Salvi et al., 2000; Tzounopoulos, 2008). Animal models of tinnitus demonstrate changes in cholinergic mechanisms in auditory nuclei (e.g., Jin et al., 2006). Interestingly, ACh has been suggested to adjust excitability and set the level of activity in the IC (Farley et al., 1983; Habbicht and Vater, 1996), one of the regions reported to show hyperactivity in tinnitus. Miko and Sanes (2009) recently concluded that ACh could affect IC excitability via actions on GABAergic IC cells. Could hyperactivity associated with tinnitus be a result of the cholinergic system trying to “reset” the responsiveness of the auditory system broadly (i.e., across frequencies), with a result that a subset of cells (those associated with the frequencies of the tinnitus) become hyper-excitable? A better understanding of the mechanisms of actions of ACh on excitatory vs. inhibitory cells will be of particular interest in this regard.