The middle ear muscle (MEM) reflexes are major efferent systems to the auditory periphery (Liberman and Guinan,1998). The stapedius and tensor tympani muscles comprise the MEMs and are innervated by the facial and trigeminal nerves, respectively. Although in some mammals, both MEMs contract in response to sound, the stapedius reflex is felt to be the dominant sound-evoked MEM pathway in humans and will be the focus of this study (Murata et al.,1986; Liberman and Guinan,1998). The basic circuit diagram of the stapedius reflex is shown in Figure 1. The stapedius muscle is the target organ of the stapedius reflex and is innervated by efferent fibers originating in motoneurons located near the facial nerve nuclei of the brainstem. The tensor tympani muscle is innervated by efferent fibers originating in motoneurons located near the trigeminal nerve nuclei. Although the afferent and efferent pathways of the MEM reflexes are well described, the central neural connections (called “interneurons”) that project from the cochlear nucleus to the stapedius (or tensor tympani) motoneurons have not been clearly defined. The aim of this study was to provide a review of what is known about the central pathways mediating the MEM reflexes, develop a physiologically based assay of the MEM reflexes in a mammalian model, and use the assay to determine whether selective manipulations of the auditory brainstem could provide clues about the neural pathways of the MEM reflexes.
The middle ear muscle (MEM) reflexes function to protect the inner ear from intense acoustic stimuli and to reduce acoustic masking. Sound presented to the same side or to the opposite side activates the MEM reflex on both sides. The ascending limbs of these pathways must be the auditory nerve fibers originating in the cochlea and terminating in the cochlear nucleus, the first relay station for all ascending auditory information. The descending limbs project from the motoneurons in the brainstem to the MEMs on both sides, causing their contraction. Although the ascending and descending pathways are well described, the cochlear nucleus interneurons that mediate these reflex pathways have not been identified. In order to localize the MEM reflex interneurons, we developed a physiologically based reflex assay in the rat that can be used to determine the integrity of the reflex pathways after experimental manipulations. This assay monitored the change in tone levels and distortion product otoacoustic emissions within the ear canal in one ear during the presentation of a reflex-eliciting sound stimulus in the contralateral ear. Preliminary findings using surgical transection and focal lesioning of the auditory brainstem to interrupt the MEM reflexes suggest that MEM reflex interneurons are located in the ventral cochlear nucleus. Anat Rec Part A, 2006. © 2006 Wiley-Liss, Inc.
What Is the Function of the MEM Reflexes?
Activation of the MEM reflex pathways results in contraction of the MEMs bilaterally in response to high-level sound stimuli in either ear. Maximal excitation occurs following binaural acoustic input, with reflex thresholds in humans ranging from 75 dB SPL for noise and 90 dB SPL for tones (Liberman and Guinan,1998). Contractions of the stapedius and tensor tympani muscles exert forces perpendicular to the stapes and malleus, respectively, to stiffen the ossicular chain (Borg,1972; Moller,1983; Pang and Peake,1986). Thus, the MEM reflex increases middle ear impedance and reduces sound transmission (Borg and Moller,1967; Borg,1971; Moller,1974). This attenuation of acoustic energy reaching the cochlea likely controls the dynamic range of hearing, as most auditory nerve fibers have limited dynamic ranges (Guinan,1996). The MEM reflex also preferentially filters low-frequency sounds (Moller,1984; Pang and Peake,1986) while higher-frequency sounds are allowed to pass to the inner ear (high band-pass filter). This frequency-specific attenuation supports the hypothesis that the MEM reflex preserves speech frequencies, which are higher in frequency, from masking by intense background noise, which is often lower in frequency (Stevens and Davis,1938; Borg and Zakrisson,1974, 1975; Mahoney et al.,1979; Borg et al.,1984; Pang and Guinan,1997). The MEM reflex also protects against inner ear damage from intense acoustic stimuli (Zakrisson and Borg,1974; Brask,1979; Nilsson et al.,1980; Ferraro et al.,1981; Borg et al.,1983). For example, ears in human subjects with an absent stapedius reflex secondary to facial nerve palsy sustain more temporary hearing loss when exposed to industrial noise compared to the ear with an intact reflex (Zakrisson et al.,1980). Finally, the stapedius muscle also contracts in response to internally or self-generated vocalization (Borg and Zakrisson,1975) and may thus serve to prevent self-stimulation.
The stapedius muscle plays the dominant role in the sound-evoked MEM reflex in humans. By contrast, EMG recordings of tensor tympani muscles have shown minimal electrical activity in response to contralateral sound (Salomon,1963; Djupesland,1965,1967), although they may respond to binaural sound (Simmons,1965). Interestingly, a patient who lacked stapedius muscle activity secondary to facial nerve palsy showed an ipsilateral reflex attributable to the tensor tympani (Stach et al.,1984). However, the relative contributions of the MEMs to the acoustic reflex vary among mammals. For example, the rat demonstrates sound-evoked activity of both the tensor tympani and stapedius muscles with mean thresholds of 57 db SPL based on EMG measurements (Murata et al.,1986). Measurable activity of the tensor tympani muscle also appears to be associated with acoustic stimuli that cause a startle reaction (Borg et al.,1984; Gelfand,1984,1998; Moller,1984). Finally, cortical input in addition to auditory feedback pathways may help to modulate responses of both MEMs (Nomura et al.,1979; Burns et al.,1993).
What Is Known About the Wiring Diagram of the MEM Reflexes?
Figure 1 shows the general wiring diagram of the stapedius reflex. The afferent (cochlea → auditory nerve → cochlear nucleus) and efferent (motoneurons → middle ear muscle) limbs are well described (Joseph et al.,1985; McCue and Guinan,1988; Strutz et al.,1988; Guinan et al.,1989; Vacher et al.,1989; Wiener-Vacher et al.,1999). The MEM reflexes are triggered by monaural or binaural sound stimuli, and therefore the first-order neurons are the spiral ganglion cells of the cochlea. The central axons of these cells synapse in the cochlear nucleus (Fig. 1) (Fekete et al.,1984). Thus, the second-order neurons of the reflexes reside in the cochlear nucleus. We will call these second-order neurons “interneurons” of the MEM reflexes.
The cochlear nucleus neurons consist of a number of cell types with unique somatic and dendritic characteristics (Osen,1969; Brawer et al.,1974). These neurons also have distinct response properties to sound stimulation (Pfeiffer,1966a,1966b; Evans and Nelson,1973; Rhode et al.,1983a,1983b) and project to different targets in the auditory brainstem and beyond (Van Noort,1969; Warr,1982; Schofield and Cant,1996). The major subdivisions of the cochlear nucleus, the dorsal cochlear nucleus (DCN), the anterior ventral cochlear nucleus (AVCN), and the posterior ventral cochlear nucleus (PVCN), each has a restricted population of cell types. The subdivision that is physiologically important for the MEM reflex and the cell type of the reflex interneurons in the cochlear nucleus are not known.
The morphologic features of the efferent limb of the reflex have been well described at least for the stapedius muscle in cats (Joseph et al.,1985; McCue and Guinan,1988; Guinan et al.,1989; Vacher et al.,1989; Wiener-Vacher et al.,1999) as well as the guinea pig (Strutz et al.,1988). HRP-labeling studies of stapedius motoneurons reveal dense innervation of the stapedius muscle. In cats, over 1,100 motoneurons innervate approximately 1,730 stapedius muscle fibers (Blevins,1964; Joseph et al.,1985). Stapedius motoneurons are spread over several perifacial and periolivary regions in the cat (Lyon,1978; Joseph et al.,1985) and both ventromedially and dorsomedially to the facial nucleus in guinea pigs (Strutz et al.,1988). They are spatially organized around the motor nucleus of the facial nerve in a pattern consistent with their physiologic responses to sound (Lyon,1978; Shaw and Baker,1983; Joseph et al.,1985). As described in Figure 1, stapedius motoneurons are segregated based on whether they respond to contralateral (red), ipsilateral (blue), either ear (purple), or only by binaural sound (not shown) (McCue and Guinan,1988; Guinan et al.,1989; Vacher et al.,1989). Most motoneuron axons travel with the course of the facial nerve and ultimately travel within primary stapedial nerve fascicles to the stapedius muscle (Wiener-Vacher et al.,1999). However, axons of the accessory stapedius motoneurons bypass the brainstem genu of the facial nerve (Fig. 1, blue motoneurons) (Guinan et al.,1989). Unlike the functionally related distribution of stapedius motoneurons in the brainstem, however, there does not appear to be spatial segregation of innervation to separate zones of the stapedius muscle (Wiener-Vacher et al.,1999). The brainstem locations of stapedius motoneurons in humans is unknown, but in Rhesus monkeys, the pattern appears to be very similar to the cat (Strominger et al.,1981).
What Is Known About the MEM Reflex Interneurons?
Although previous investigations (Borg,1973; Lyon,1978; Rouiller et al.,1986,1989) have provided a basic framework for understanding the reflex interneurons, many questions have been left unanswered. What region of the cochlear nucleus contains the reflex interneurons? Borg (1973) concluded that they reside within the ventral cochlear nucleus (VCN) on the basis of tracings of degenerating nerve fibers following a mechanical lesion of the cochlear nucleus. However, such a lesioning approach jeopardizes fibers of passage, such as the auditory nerve fibers, confounding the results. Furthermore, lesion techniques cannot determine which neuronal cell types are involved in the MEM reflex pathway. In addition, whether the reflex interneurons project directly to the motoneurons or through an additional relay station is not completely understood. Borg's lesion studies concluded that the reflex pathway passed through a region of the medial superior olive (MSO) but whether axons of passage or synaptic relay stations were affected has not been settled (Borg,1973). For instance, electrical stimulation experiments by Guinan (data not shown) have shown that the MSO has the highest threshold for activating the stapedius muscle and thus this nucleus is unlikely to mediate the cat stapedius reflex. Viral transneuronal techniques seem to indicate that the stapedial and tensor tympani pathways have an additional synapse after the cochlear nucleus (Rouiller et al.,1986,1989). However, studies in cats suggest that there may be direct connections from the cochlear nucleus to the tensor tympani motoneurons (Itoh et al.,1986; Ito and Honjo,1988). More experimental studies will be needed to address this issue.
MEM Reflex Assay
Relkin and colleagues have recently developed assays for the MEM reflex based on suppression of the distortion product otoacoustic emission (DPOAE) (Azeredo et al.,2000; Relkin et al.,2001,2005; Smith et al.,2005). Specifically, the DPOAE is an emission of sound at frequency 2f1–f2 when the cochlea is stimulated with two primary tones of frequency f1 and f2 (Kemp and Chum,1980). The use of DPOAEs in the testing of cochlear reflexes stems from its ease of measurement even in awake animals (Boyev et al.,2000). Presentation of broadband noise at moderate levels to one ear was associated with a significant change in DPOAE magnitude in the measured ear. These effects were almost completely eliminated following surgical sectioning of the middle ear muscles (Relkin et al.,2005), implying that the assay may be specific for the MEM reflex.
Our reflex metric was patterned on ones developed by Relkin (Azeredo et al.,2000; Relkin et al.,2001,2005; Smith et al.,2005). Our assay tested only the reflex in response to contralateral stimulation, a crossed pathway that traverses the midline of the brainstem (Fig. 1, red pathway) and infers middle ear muscle contraction and changes in middle ear impedance via measured changes in primary tone and DPOAE levels in the ear canal. In the present study, we examine the central pathways of the MEM reflex by measuring the effects on the strength of the reflex assay produced by selective surgical transection or focal chemical lesioning of the auditory brainstem. These findings give us insight to the location and identity of the MEM reflex interneurons in the cochlear nucleus.
MATERIALS AND METHODS
All procedures were conducted in accordance with guidelines of the National Institutes of Health and were approved by the Animal Care and Use Committee of the Massachusetts Eye and Ear Infirmary. Experiments were conducted within a soundproof chamber. A total of 14 adult albino rats (374–596 g) were used: 2 rats for tubocurarine administration, 2 rats for transection of the dorsal and intermediate acoustic striae, and 10 rats for KA lesion experiments. Anesthesia was performed using ketamine (100 mg/kg i.p.) and xylazine (10 mg/kg i.p.). Atropine (0.4 mg/kg) was given to minimize respiratory secretions. Booster injections of ketamine (30 mg/kg i.p.) and xylazine (3 mg/kg i.p.) were given every 1–2 hr as needed. Rectal temperature was maintained at 37°C. After tracheotomy, each pinna was removed. The meatus and lateral external auditory canal of the rat are tortuous and pinna removal ensures precise placement of the microphone or sound source coupler in the medial external auditory canal. The pinna is not known to be directly involved in the MEM reflex and the muscles that move the pinna for sound orientation are felt to be innervated by the upper cervical nerves (Abrahams et al.,1984a,1984b). It is not known whether distinct branches of the facial nerve are also involved in orienting the pinna in response to sound in the rat.
For experiments involving manipulation of the auditory brainstem, the left CN was exposed by a posterior craniotomy and partial cerebellar aspiration.
The reflex elicitor, a broadband noise stimulus (28- to 48-sec duration), was presented to the left ear using a reverse-driven 1 inch condenser microphone at 74 dB SPL. The output of the broadband noise stimulus was attenuated by 30 dB and is at a level below the threshold of the acoustic startle reflex (Lee et al.,1996). DPOAEs were assayed in the opposite (right) ear at 4-sec intervals before, during, and after presentation of the reflex elicitor. DPOAEs at 2f1–f2 were measured with an Etymotic ER-10c transducer using a Labview controller. The frequency of the primary tone, f2, was 10 kHz, and that of the second primary tone, f1, was 0.83 × f2. The f1 primary tone was presented 10 dB higher than f2, and DPOAE 2f1–f2 was recorded; the acoustic pressures of the primary tones were also recorded in the external ear. Primary levels were adjusted to f2 = 25 to 35 dB to keep 2f1–f2 at a constant baseline.
Two rats were ventilated via a tracheotomy tube and tubocurarine was intramuscularly injected (1 mg/kg). The reflex assay was used to compare responses before and after paralysis with tubocurarine.
In two rats, a posterior craniotomy was performed and cerebellum aspirated to expose the DCN. Using the DCN as a landmark, a surgical scalpel was used to cut the dorsal acoustic stria (DAS) and intermediate stria (IAS; output pathways of the CN) of the left ear. Postexperiment histology confirmed transection of these pathways.
In 10 rats, a posterior craniotomy and cerebellar aspiration was performed on the left side to expose the DCN and caudal PVCN. Kainic acid (KA; 5 mM in phosphate-buffered saline, pH 7.0) was delivered with a micropipette (20 μm tip) attached to a 5 μl Hamilton syringe mounted on a micromanipulator. The tip was aimed into the CN of the left ear by visual guidance and the micromanipulator scale was used to determine the depth of penetration. Pressure injections of KA were made using a hydraulic master/slave remote injection system. Injections were made in small increments (usually 0.1 μl over a period of 15–20 min for a total injection of 0.2–0.3 μl). After KA injection, DPOAE responses and primary tone levels were measured during the survival period (150–433 min) in the right ear. The reflex elicitor, a broadband noise stimulus, was presented to left ear, the side of the KA lesion. Control experiments were performed presenting the reflex elicitor to the right ear while measuring primary tone level and DPOAE changes in the left lesioned ear to rule out any generalized effects of KA or surgical trauma on auditory brainstem function. Postinjection survival times allow for development and stabilization of neuronal damage at the injection site (Bird et al.,1978; Melcher et al.,1996a,1996b; de Venecia et al.,2005).
At the end of the survival period, animals were perfused with 4% paraformaldehyde. Serial frozen sections were cut (40 μm) in the transverse plane and counterstained with methylene blue. Lesion sites were defined as the regions with almost complete absence of neurons and the presence of cell fragments. Lesions were projected onto template sections of a rat CN for comparison.
MEM Reflex Assay
Our MEM reflex assay represents a test of the contralateral or crossed reflex pathways and would include effects of both stapedius and tensor tympani contractions (Fig. 2A). Presentation of the reflex elicitor (broadband noise) to the left ear excites MEM reflex interneurons in the left cochlear nucleus (large star). Ultimately, the pathway provides input to the motoneurons innervating the stapedius muscle and tensor tympani muscles of the right ear. The reflex metric shown in Figure 2B monitors the right ear for effects during the period of the reflex elicitor. If the MEM reflex is activated, contractions of MEMs would stiffen the ossicular chain, producing an increase in impedance at the tympanic membrane, resulting in an increase in the primary tone levels, and a decrease in forward and reverse middle ear transmission, resulting in DPOAE reduction. As shown in Figure 2B, presentation of the noise at 74 dB SPL to the left ear changes the level of the primary tones and the DPOAE as measured in the right ear (red and green plot, Fig. 2B). A reflex elicitor intensity of 74 dB SPL was chosen in order to produce a consistent primary tone level change and DPOAE suppression without causing acoustic trauma or fatigue of the MEM reflex. Although we observed DPOAE reduction with contralateral noise levels as low as 54 dB SPL, primary tone level changes were inconsistent when the intensity of the reflex elicitor was decreased from 74 dB SPL (data not shown). No reflex effects in response to broadband noise were observed at 44 dB or lower. For pure tones, we observed the largest MEM reflex effects in the right ear when frequencies of 4 or 6 kHz were presented as the reflex elicitor to the left ear. These pure tone effects are generally consistent with previous studies (Murata et al.,1986; Pilz et al.,1997). The thresholds needed to produce primary tone level changes and DPOAE reduction were higher when using pure tones compared with broadband noise.
Is the Assay Specific for the Middle Ear Muscle Reflex?
DPOAE suppression due to MEM reflex effects has been observed in awake rabbits and anesthetized rats (Whitehead et al.,1991; Azeredo et al.,2000; Relkin et al.,2001,2005; Luebke et al.,2002; Smith et al.,2005). In contrast, DPOAE effects in anesthetized cats, guinea pigs, or mice are primarily due to another auditory feedback pathway, the medial olivocochlear reflex (MOC) system (Puel and Rebillard,1990; Liberman et al.,1996; Kujawa and Liberman,2001).
To determine if our assay was specific for the MEM reflex, we utilized a noninvasive technique to compare the relative effects of the MEM or MOC systems on primary tone level changes and DPOAE suppression. Tubocurare is a nondepolarizing paralytic that competitively binds to acetylcholine receptors at the neuromuscular junction. Tubocurare would be expected to paralyze the activity of the MEMs and have no effect on the MOC efferent system. We administered systemic tubocurare intramuscularly in two subjects and ventilated them following tracheotomy (Fig. 3). Ten minutes following systemic tubocurare administration, DPOAE input/output functions were unchanged and no DPOAE suppression or primary tone level changes were observed (Fig. 3). This result demonstrates that our assay is primarily a test of the MEM reflexes and that the MOC system does not contribute significantly to our observed reflex effects. These findings are consistent with previous observations (Relkin and Doucet,1989; Relkin et al.,2005). It is possible that for different anesthesia levels or test conditions, MOC-induced changes may affect the DPOAE. However, this influence will be negligible on changes in primary tone levels that are specific for MEM activity.
Transection of Acoustic Striae
Cochlear nucleus neurons project to other regions of the brainstem by way of pathways known as striae. The three striae are ventral, intermediate, and dorsal (VAS, IAS, DAS), corresponding to their location as they exit the cochlear nucleus (Held,1893; Masterton and Granger,1988).
To investigate the central pathways taken by the axons of the reflex interneurons, transection experiments were performed (Fig. 4). We surgically sectioned both the dorsal and intermediate acoustic striae (DAS/IAS) on the side of the reflex elicitor. The ventral acoustic stria was preserved. The DAS/IAS transection did not result in the attenuation or elimination of noise-evoked DPOAE or primary tone level changes (Fig. 4). We thus conclude that MEM reflex interneurons project through the remaining ventral acoustic striae. These findings suggest that the location of the MEM reflex interneurons is in the VCN, since most VCN neurons project out the ventral acoustic stria.
Lesioning Studies of Cochlear Nucleus
KA was used to create discrete chemical lesions of the cochlear nucleus to identify the specific region involved in the MEM reflex. KA has the advantage of killing cells locally while sparing nerve fibers of passage (Coyle et al.,1978; McGeer et al.,1978; Wuerthele et al.,1978). We focally injected KA into the cochlear nuclei of 10 rats to correlate lesion data with alterations in MEM reflex strength. Six cases had insufficient survival times or no visible lesion of the cochlear nucleus and so were not included in our analysis. As shown in the circuit diagrams of Figures 5 and 6, KA was injected into the left cochlear nucleus (on the side of the reflex elicitor) and both acute and long-term effects on primary tone levels and DPOAE suppression were monitored from the opposite ear. In our paradigm, an ineffective or reflex-sparing lesion of the cochlear nucleus would be associated with no long-term changes in the MEM reflex assay, although acute effects could be due to spread of KA that causes transient interruption of function at sublethal concentrations. An effective or reflex-interrupting lesion involving the reflex interneurons would be expected to produce a permanent attenuation in reflex strength.
In all cases, KA injections into the cochlear nucleus produced transient attenuation of DPOAE suppression and primary tone level changes (Figs. 5B and 6B). In Figure 5, KA was injected into the DCN and resulted in full recovery of reflex strength several hours following administration (Fig. 5C). Such long-term recovery of reflex strength was observed in three cases.
Long-term attenuation of the MEM reflex was seen in one case (Fig. 6). A KA injection was made in the ventral cochlear nucleus, eliminating most of the reflex effects up to 7 hr following lesioning. This reflex strength reduction indicates that the reflex interneurons were lesioned. To test whether permanent attenuation was secondary to experimental trauma, the opposite reflex pathway was tested by presenting the reflex elicitor to the unlesioned right ear while monitoring for reflex effects in the lesioned left ear. We observed robust DPOAE suppression and increase in primary tone levels (data not shown), demonstrating that there was no sign of generalized trauma that blocked reflex effects.
The transient and long-term changes measured by our reflex metric were correlated with postexperiment histology. Figure 7 is a representative section from R-135, where long-term reduction in reflex strength was seen following lesioning. KA injections create well-defined regions of cell loss surrounded by poorly defined areas of partial cell loss (Fig. 7A). The blue dashed line defines the anatomical boundaries of a KA lesion, a region of complete cell loss with cell fragments and reactive gliosis. Typically, such lesions were surrounded by a halo of partial cell loss. The lesion in this case is centered in the caudal PVCN and had some involvement of DCN. In contrast, the control side that was not injected with KA (Fig. 7B) from the same brainstem section as Figure 7A shows a normal population of neurons in the DCN and PVCN.
Lesion data for both reflex-sparing and reflex-interrupting cases were summarized on an atlas of a rat cochlear nucleus (Fig. 8). The three cases (R-122, 126, 127) where only transient effects were seen are plotted in blue (Fig. 8A). The lesion from R-122 is a small area centered between the caudal PVCN/DCN boundary. The AVCN was uninvolved. R-126 has a large lesion centered on the DCN with involvement of the granular region of AVCN and the dorsal edge of PVCN. The AVCN was otherwise not lesioned by KA. The lesion from R-127 spans most of DCN while sparing PVCN and AVCN. Based on these lesions and the DAS-IAS transection experiments, we conclude that the MEM reflex interneurons are not found in the DCN and probably not in the superficial granular layer of AVCN.
Figure 8B (red plot) shows the data from one subject (R-135) that demonstrated a long-term reflex-interrupting lesion. Unlike the reflex-sparing lesions seen in Figure 8A, this lesion is centered in the caudal PVCN and also involves the rostral PVCN, the dorsal nerve root region, and a part of ventral DCN (Fig. 8B). The AVCN is spared in this case. It is unlikely that the DCN lesion caused interruption in this case given the reflex-sparing cases presented in Figure 8A. The lesioning data suggest that the MEM reflex interneurons are found in the VCN.
We applied a noninvasive assay that measures rat MEM reflex effects to determine the integrity of the central pathways following experimental manipulation. Transection of the dorsal and intermediate acoustic striae (DAS and IAS) did not change the reflex effects. This result eliminates as prospective interneurons all those cochlear nucleus neurons that project out through these striae (e.g., all DCN cells, PVCN octopus cells). It also demonstrates that the output axons of the reflex interneurons project through the remaining stria, the ventral acoustic stria. Interestingly, behavioral studies in cats following selective surgical sectioning of the DAS and IAS suggest that these central pathways may not be critical in reducing the masking effects of background noise compared with the fibers in the VAS (Masterton et al.,1994). Perhaps the sparing of the VAS in the cat preserves the MEM reflex pathway that contributes to this antimasking effect. The recent identification of the MOC reflex interneurons in the PVCN (which proceed through the VAS) is also supported by these behavioral studies (de Venecia et al.,2005).
KA lesions also indicate that the reflex interneurons are in the VCN. KA is an excitatory neurotoxin that prolongs ion channel opening at the cell membrane by binding to glutamate receptors (cell membrane neurotransmitter receptors). Prolonged excitation by kainic acid results in disruption of intracellular homeostasis and cell death (Olney et al.,1971,1974). The advantage of KA over other techniques is its selective activity; KA is only toxic to neurons whose cell bodies, or somata, reside near the injection site (Coyle et al.,1978; McGeer et al.,1978; Wuerthele et al.,1978) while sparing axons of passage. The important axons of passage for our studies are the auditory nerve fibers that pass through various regions of the cochlear nucleus. Cell loss in transiently interrupted cases was mainly in the DCN or granular regions of AVCN, indicating that the interneurons are not localized to these sites. The single KA injection that showed long-term reflex reduction produced neuron loss in the PVCN. The cell types localized to the PVCN include multipolar cells, bushy cells, and octopus cells. Octopus cells are unlikely to be the reflex interneurons because we cut the IAS that carries the outflow axons from these neurons and did not observe any MEM reflex effects. In addition, the responses of octopus cells are transient (Rhode et al.,1983b; Rouiller and Ryugo,1984), unlike the sustained responses of stapedius motoneurons (Kobler et al.,1992). In contrast, the projections of globular bushy cells out the VAS as well as the areas in which some of their collaterals end make them possible interneurons (Friauf and Ostwald,1988; Smith et al.,1991).
Our studies only tested the reflex in response to contralateral sound. Hypothetically, the interneurons for the motoneuron groups responsive to ipsilateral or either-ear sound could be different (Fig. 1). Continuing studies will further narrow the candidates for the interneurons of several reflex loops. Based on these preliminary data, the pathway of the MEM reflex is a 3- or 4-neuron circuit composed of the spiral ganglion cell, a reflex interneuron in the ventral cochlear nucleus, a possible second interneuron, and MEM motoneurons.
Analogous experiments have been performed to describe other sound-evoked reflex circuits. In the rat acoustic startle reflex, the second-order neurons are cochlear root neurons embedded in the dorsal auditory nerve root (Lee et al.,1996). Based on bilateral KA lesions of the dorsal nerve root, the proposed wiring diagram is a 4-neuron circuit consisting of the auditory nerve, the cochlear root neurons, neurons in the nucleus reticularis pontis caudalis, and spinal motoneurons. Our MEM reflex elicitor is below the thresholds required to elicit the acoustic startle reflex in these subjects. However, intense sound stimuli would be expected to activate both the MEM reflex and acoustic startle pathway simultaneously and additional experiments may reveal a common pool of neurons between these two circuits. The MOC reflex is another sound-evoked reflex that, like the MEM reflex, provides efferent feedback to the inner ear (Liberman and Guinan,1998). The MOC circuit diagram is a three-neuron reflex loop whose interneurons have been localized to the PVCN (Thompson and Thompson,1991; de Venecia et al.,2005). More complex pathways to the MOC neurons are also known (Thompson and Thompson,1993; Vetter et al.,1993).
Our preliminary data suggest that the VCN may contain the MEM reflex interneurons, a region common to MOC reflex interneurons (as well as the acoustic startle reflex). This degree of overlap would not be surprising, as the MEM and MOC pathways are triggered by sound and provide complementary antimasking effects to the auditory periphery (Liberman and Guinan,1998). In contrast to the stapedius reflex pathway, however, the MOC reflex pathway has been shown to project directly from the cochlear nucleus to the output neurons (MOC neurons) (Thompson and Thompson,1991). Future studies will be necessary to clarify the central pathways mediating the MEM reflex pathway.
The authors thank Dr. Wen Xu for his expert technical assistance in this project.