Olivocochlear neurons sending axon collaterals into the ventral cochlear nucleus of the rat

Olivocochlear neurons

Following FB or DY application into the cochlea, retrogradely labeled cells were encountered bilaterally in the SOC as described earlier (White and Warr, 1983; Aschoff and Ostwald, 1988). Labeled cells were found in the facial, vestibular, and reticular nuclei, but these will not be dealt with any further in this study. Small (diameter approximately 10 μm) and fusiform neurons inside the LSO (intrinsic LOCs) were labeled virtually exclusively on the ipsilateral side. The majority of larger (approximately 15 μm) multipolar cells of the VNTB (MOCs) resided on the contralateral side. Additionally, larger (about 15 μm) neurons in the periolivary region around the LSO were labeled with an ipsilateral dominance. These neurons were located mainly dorsal, caudal, and rostral to the LSO, and their dendrites were seen to extend almost always into the LSO. This group of olivocochlear cells may be called shell neurons (Vetter and Mugnaini, 1992), or shell LOCs. Figure 1 shows the distribution of neurons in the SOC that were retrogradely labeled following FB application into the left cochlea of one representative animal. There was no difference in the distribution of labeled cells after the intracochlear application of DY or FB. The total number of labeled olivocochlear cells was 901.0 ± 17.7 SEM after tracing with FB and 1,129.0 ± 13.0 SEM after DY application. Averaging over six cases, 509.2 ± 19.8 SEM cells were labeled in the ipsilateral LSO, 54.2 ± 5.0 SEM cells in the ipsilateral shell, and 170.0 ± 13.7 SEM cells in the ipsilateral VNTB. Contralaterally, no labeled neurons were found in the LSO in three cases, but up to two labeled cells were seen in others. There were 14.3 ± 1.3 SEM labeled contralateral shell neurons and 228.3 ± 17.7 SEM labeled cells in the contralateral VNTB. DY-labeled cells could be classified only according to the location of cells and the size of the nucleus as the dendritic morphology of the labeled cells was not revealed by the tracer. Table 1 summarizes the counts of retrogradely labeled olivocochlear neurons over all six experiments.

Double-labeling experiments

In the second series of experiments, we combined the retrograde labeling of olivocochlear cells with the visualization of cells that project to the VCN. DY was applied to the left cochlea to trace olivocochlear cells and FB was injected into the VCN of the same side. The average number of labeled olivocochlear cells (929.0 ± 42.6 SEM) was indistinguishable to the counts obtained after a single application of FB or DY into the cochlea as detailed above (977.0 ± 49.5 SEM). In five experiments, the injection site of FB, including their diffusion range, was restricted to the VCN and included both the posteroventral (PVCN) and anteroventral cochlear nucleus (AVCN). Figures 2 and 4 document the injection sites of FB in the VCN of two representative cases. The most medial and rostral parts of the VCN were not filled with the tracer.

Following the FB injection into the VCN, a large number of periolivary cells were labeled, showing an ipsilateral dominance (ipsilateral vs. contralateral 1.5:1). Principal neurons of the ipsilateral medial nucleus of the trapezoid body (MNTB) were also labeled. The ipsilateral LSO had a few (8.8 ± 2.3 SEM), the contralateral had no labeled neurons. Additionally, multipolar neurons of the contralateral VCN, some cells in the deeper layers of the contralateral DCN, and neurons of all three subnuclei of the inferior colliculus were also labeled. Figure 2 provides data of one representative experiment, showing the distribution of neurons in the SOC that were retrogradely labeled after injection of FB into the VCN. In five other experimental cases, the injection site extended beyond the VCN and covered part of the white matter immediately medial of it. As axons of the olivocochlear neurons run in this bundle, tracer uptake by its fibers resulted in retrograde labeling of cells in several regions of the brainstem, including the ipsilateral LSO. These experiments were not considered for our quantitative evaluations.

Olivocochlear neurons with collaterals to the VCN

The combination of DY and FB allowed us to identify olivocochlear cells giving off collaterals into the VCN. The combined application of these tracers resulted in partly intermingled populations of blue and/or yellow-labeled neurons (Figs. 3A,B). Single-labeled neurons showed only yellow fluorescence of the nucleus (Fig. 3C) or blue staining of the cytoplasm (Fig. 3D). Double-labeled neurons were clearly distinguishable by the intense yellow fluorescence of their nucleus and their blue cytoplasm (Figs. 3E–G). We found double-labeled cells in the periolivary region around the LSO (shell LOCs) and in the VNTB (MOCs) of both sides of the brain. Across five experiments, only six double-labeled neurons were found in the ipsilateral LSO. Figure 4 shows the distribution of single and double-labeled olivocochlear cells in the SOC. Fifty-six percent of the MOCs in the VNTB and 39% of shell neurons were seen by virtue of their double labeling to send collaterals into the VCN. These percentages resulted from retrograde transport originating in a restricted area of the VCN. They are expected to grow, possibly up to 100%, with a larger or even complete coverage of the VCN with tracer, which is difficult to achieve for practical reasons. Table 2 summarizes counts and SEMs of single and double-labeled olivocochlear neurons in the LSO, shell region, and the VNTB.

Single labeling of olivocochlear cells

In the present study, we could identify and quantify two groups of LOCs — one in the ipsilateral LSO (intrinsic LOCs) and another in the periolivary region around the LSO of both sides (shell LOCs). MOCs were found in the VNTB of both sides. The distribution and number of olivocochlear cells are in general agreement with previous studies. White and Warr (1983) described the dual (i.e., lateral and medial) origin of the OCB in the rat by using axonal transport of HRP. Injection of free HRP into the cochlea resulted in the retrograde labeling of 350 cells in the LSO and 300 cells in the VNTB. The numbers of labeled olivocochlear cells in the present study are significantly larger than these counts (509.2 ± 19.8 cells in the LSO and 398.3 ± 30.8 cells in the VNTB), but somewhat smaller than found previously with the same fluorescent tracers (540 cells in the LSO and 556 in the VNTB, Aschoff and Ostwald, 1988; cp. Robertson et al., 1989). In addition, we also recognized labeled shell neurons first described by Vetter et al. (1991). Shell neurons were also traced by intracochlear injection of cholera toxin tracers. It was shown that their dendrites penetrate into the LSO and surrounding areas (Vetter and Mugnaini, 1992). In six cases, we counted 68.5 ± 5.3 SEM shell neurons, 54.2 ± 5.0 SEM cells around the ipsilateral, and 14.3 ± 1.3 SEM cells around the contralateral LSO. Injection of cholera toxins resulted in 111 labeled shell neurons, 105 ipsilateral, and 6 contralateral (Vetter and Mugnaini, 1992). These differences may be explained by a difference in uptake efficiency of the different tracers.

Following tracer application into the cochlea, marginal to low levels of FB remained in the middle ear cavity. This may have resulted in labeling of a comparatively sparse population of motor neurons of middle ear muscles (Spangler et al., 1982; Keller et al., 1983; Rouiller et al., 1989) and salivatory glands (Hiura, 1977). However, areas of overlap with olivocochlear cells only exist for stapedial motor neurons and are marginal even there (Rouiller et al., 1989). Should this labeling have occurred, our counting error would not exceed 5%.

Double-labeling experiments

In our second experimental group, we have combined the retrograde tracing of the olivocochlear cells with DY and the injection of FB into the VCN. This tracer combination has been proven useful in numerous neuroanatomical studies dealing with virtually all projectional systems in the mammalian brain. Double-labeled cells can be easily identified by a yellow nucleus and blue cytoplasm (Figs. 3E–G). This combination of tracers has also been used to study the collateral projections of olivocochlear cells in the guinea pig (Robertson and Winter 1988; Winter et al., 1989).

Quantitative evaluation was based only on those experiments in which the injection and local diffusion of FB were confined to the VCN and included both the PVCN and the AVCN (Figs. 2, 4) but not adjacent white matter. Regions that were found to project to the VCN included the ipsilateral DCN, the ipsilateral and contralateral SOC, the contralateral DCN and VCN, and the inferior colliculus of both sides. These projections are known and were analyzed in previous studies in the cat (Adams and Warr, 1976; Cant and Gaston, 1982; Adams, 1983; Spangler et al., 1987), guinea pig (Wenthold, 1987; Schofield, 1991, 1994; Shore et al., 1991, 1992; Ostapoff et al., 1997), and rat (Weedman and Ryugo, 1996a,b; Warr and Beck, 1996; Alibardi, 1998). Our study specifically focuses on olivocochlear cells projecting to the VCN.

MOCs with collaterals to the VCN

Our results show that the majority of MOCs, located in the VNTB of both sides, send collaterals into the VCN (Fig. 4, Table 2). Double-labeled cells were found along the whole rostrocaudal extent of the VNTB and the percentage of double-labeled cells was almost the same on the two sides (ipsilateral 58%, contralateral 54%). Because the tracer injection into the VCN could not cover the whole nucleus, the counts we obtained must necessarily be minimal estimates. We suggest that in fact 80–100% of MOCs send collaterals into VCN. This estimate would be consistent with previous observations in some other species. Using HRP injection into the cochlea, 67–100% (depending on the species) of thick MOC axons were found to give off collaterals into the VCN in the cat (Brown at al., 1988), mouse (Brown et al., 1988; Brown and Benson, 1992), and gerbil (Brown et al., 1988; Ryan et al. 1990). In the rat, HRP injection into the cochlea showed that collaterals leave the OCB and enter the VCN at two points: at the junction between the DCN and PVCN and from the main OCB as it descends on the medial side of the vestibular nerve, entering the AVCN (White and Warr, 1983). These collaterals are also apparent with stains for acetylcholinesterase (AChE; White and Warr, 1983). Injection of ³H-leucine and biotinylated dextran amine into the VNTB of the rat resulted in strong anterograde labeling in the cochlear nuclei and identification of the OCB (Warr and Beck, 1996). However, indirect evidence from other studies suggested that collateral projections from the OCB to the VCN are very sparse in the rat. Cutting the entire OCB causes only slight reductions in levels of AChE in the cochlear nucleus (Osen et al., 1984). In microassays for choline acetyltransferase (ChAT), a reduction of only about 15% was observed in the cochlear nucleus granule cell region of the rat (Godfrey et al., 1987); a reduction of 70% was found in the cat (Godfrey et al., 1990).

Injecting FB into the cochlear nucleus and DY into the cochlea of the guinea pig, Winter et al. (1989) saw only 3.5–9.9% of MOCs projecting to the cochlear nucleus. This difference to the data obtained in the present study could either be an interspecies difference between guinea pig and rat or the result of methodological differences. In three experimental cases, Winter et al. (1989) injected a 2% aqueous solution of FB into the cochlear nucleus (including the DCN). However, the 9.9% double labeling of MOCs was reported in only one experiment, whereas the others showed even fewer double-labeled cells.

The local targets of the collateral projections from MOCs could not be determined in the present study as our injection site occupied both the peripheral and central VCN (Figs. 2, 4). Previous ultrastructural investigations showed that cochlear nucleus collaterals of MOCs terminate on varicose dendrites of small cells and on large dendrites of multipolar cells in mice (Benson and Brown 1990; Benson et al., 1996). The massive collateral projection of MOCs found in the present study indicates that MOCs must play a crucial role in feedback control of auditory processing. These cells can directly influence the motility of OHCs (Dallos et al., 1990; Liberman et al., 1990) as well as the secondary sensory neurons in the cochlear nucleus. They are themselves under the influence of an excitatory input from the contralateral cochlea by way of the cochlear nucleus, thus serving to link both cochlea and both cochlear nuclei (Robertson and Winter, 1988; Thompson and Thompson, 1991). This regional feedback loop, together with descending pathways from the inferior colliculus (Faye-Lund, 1986; Caicedo and Herbert, 1993; Vetter et al., 1993) and auditory cortex (Feliciano et al., 1995; Weedman and Ryugo, 1996a,b), constitutes the anatomical basis for the efferent control of auditory processing in the lower brainstem.

LOCs with collaterals to the VCN

The second main finding of the present study is that the two types of lateral olivocochlear neurons differ in their collateral projection pattern. Although intrinsic LOCs in the LSO, apart from extremely rare single cells, do not send collaterals into the VCN, a majority of shell neurons around the LSO may send axon branches into the VCN. Previous studies have shown that the projections from the LSO to the cochlear nucleus are very sparse in the guinea pig (Winter et al., 1989; Schofield, 1994; Ostapoff et al., 1997) and cat (Spangler et al., 1987). Weedman and Ryugo (1996a) did not observe any labeled cells in the LSO following FB injection into the rat cochlear nucleus. On average, we found only 8.8 ± 2.3 SEM LSO cells that project to the VCN. Across all five experiments, only six of these neurons (1.2 ± 0.7 SEM) proved, by virtue of their double labeling, to be olivocochlear cells. When our injection extended medially to the VCN and touched the OCB, nearly all LOCs in the LSO were double labeled. This is evidence that the thin unmyelinated axons are capable of transporting either tracer. The failure of labeling these cells by tracer injections strictly confined to the VCN does imply the absence of axon terminals there.

There is virtually no projection from intrinsic LOCs to the VCN in the normal adult rat. Therefore, we must conclude that the cochleotomy-induced reemergence of GAP-43 in the neuropil of the VCN and in cell bodies of intrinsic LOCs reported earlier (Illing et al., 1997, 1999) are independent consequences of the cochlear lesion — unless there occurs a cochleotomy-dependent sprouting of the axons of these olivocochlear neurons into the cochlear nucleus, a possibility currently investigated in this laboratory.

Our results support the evidence that LOCs are divided into two subclasses in the rat: the small (diameter approximately 10 μm) and fusiform intrinsic neurons in the LSO and the larger (diameter approximately 15 μm) and multipolar shell neurons around the LSO (Vetter and Mugnaini, 1992). It has already been reported that these groups of cells have different terminal arborizations in the cochlea (Warr et al., 1997). Here we show that the axonal projection pattern of these neurons is also different: many shell LOCs send collaterals into the VCN, whereas the intrinsic LOCs in the LSO project only to the cochlea. Apparently, shell neurons share size, dendritic morphology, and the collateral projection pattern including the possibility to project to both cochleae, with MOCs (Aschoff and Ostwald, 1988; Robertson et al., 1989), but their localization close to the LSO, and their innervation of IHCs in the cochlea with the LOCs.