We studied human brainstems from the Witelson Normal Brain Collection; the details of subject selection and tissue acquisition were described in Witelson and McCulloch (1991). Briefly, the subjects were patients diagnosed with metastatic cancer. At the time of enrollment in the study, they had no diagnosed neurological disease. They then underwent neuropsychological testing focused on cognitive tasks and experimental dichotic listening tasks. Standard audiological testing was done for each subject, and each had hearing levels within normal limits for each ear and did not wear hearing aids. Patients were then followed medically and periodically screened for the development of neurologic dysfunction. At death, brains were removed by pathologists and examined for neuropathology. We selected cases with short post-mortem intervals (PMI) and no documented neurological dysfunction in life or neuropathology after death.

Table 1 shows the case number, age, sex, and PMI (in hours) of the human brain specimens we used. We have previously reported data on protein expression in various brainstem nuclei of these cases, including the CN (Baizer et al., 2007, 2011a,b, 2013b, 2014; Baizer and Broussard, 2010).

Our methods for processing this tissue have been described previously (Baizer et al., 2007, 2013b; Baizer and Broussard, 2010). Briefly, tissue blocks containing the brainstem and cerebellum were dissected away from the cerebrum, and all tissue was stored in 10% formalin. We further separated the brainstem and cerebellum and then cryoprotected the brainstems with 15% sucrose and 30% sucrose in 10% formalin. Prior to sectioning, we made a small slit along one side of the ventral brainstem to allow identification of the left and right sides of the brain. About 40-μm-thick frozen sections were cut on an American Optical (AO) sliding microtome in a plane transverse to the brainstem. All sections were collected and stored in plastic compartment boxes, five sections/compartments, at 4°C in 5% formalin. Initially, sets of sections 2 mm apart from each case were stained with a Nissl stain, cresyl violet, following a standard protocol (LaBossiere and Glickstein, 1976). Additional sections were stained as needed in different studies to define the limits of structures of interest.

We also processed archival sections from three chimpanzees (Pan troglodytes; cases AN, f, age 15; MT, m, age 25; and WM, f, age 13). Chimpanzees had been housed at the Emory National Primate Center according to Association for the Assessment and Accreditation of Laboratory Animal Care International (AAALAC) guidelines. All died from natural causes. None of the animals showed evidence of neurological disorder and appeared normal on routine assessments for pathology. Details of the histology for these cases have been published in earlier reports (Baizer et al., 2011b, 2013a, 2018b). Blocks of tissue, including the cerebellum and brainstem, were first cryoprotected in 15% and then 30% sucrose in phosphate-buffered saline (PBS), and then frozen, transverse sections of the brainstems were cut with an AO sliding microtome at 40 μm. Every section was collected and stored in a cryoprotection solution of 30% ethylene glycol, 25% glycerol, and 45% 0.1 M phosphate buffer at −20°C.

We immunostained archival sections of a cat brainstem that had been prepared for earlier studies (Baizer and Baker, 2005, 2006; Baizer et al., 2010). For those studies, 10 adult cats of both sexes were obtained from commercial breeders, maintained in the Northwestern University animal care facility, and handled according to the guidelines of the Northwestern University Animal Care and Use Committee. Frontal sections 50 μm thick were cut on the AO sliding microtome and stored in tissue culture wells in the cryoprotection solution in a −20°C freezer. We selected sections containing the CN and/or SOC and used the same antibodies and immunohistochemistry (IHC) protocols as in human and chimpanzee tissue, except for omitting the antigen retrieval protocol.

We also examined sections from cats, chimpanzees, and humans that had been stained for cresyl violet or immunostained for non-phosphorylated neurofilament protein (NPNFP) or calretinin (CR) in the course of earlier studies (Baizer and Baker, 2005; Baizer and Broussard, 2010; Baizer et al., 2013a). NPNFP is a structural protein expressed in subsets of pyramidal cells and their axons in the cerebral cortex and in subsets of projection neurons in the brainstem (Hof et al., 1995; Hof and Morrison, 1995; Baizer et al., 2011a, 2013a). CR is a calcium-binding protein expressed by subsets of interneurons in the cerebral cortex (Glezer et al., 1992; Condé et al., 1994). In the auditory brainstem, neurons in the AVCN and octopus cells in the PVCN express CR (Lohmann and Friauf, 1996; Felix et al., 2017; Baizer et al., 2018a). Some of these sections had been immunostained with a DAB visualization protocol, giving brown staining in contrast to the gray/black staining seen with the glucose oxidase visualization protocol.

All IHC was performed on free-floating sections. Sections were rinsed in PBS (all rinses were 3 × 10 min). Sections from humans and chimpanzees were then treated with an antigen retrieval (AR) protocol. Each section was placed in a separate small glass jar with 20 mL of citrate buffer (pH 6). The jars were heated in a water bath at 85°C for 30 min. The jars were removed from the bath and cooled to room temperature. Sections were rinsed in PBS, and non-specific labels were blocked by incubating sections in a solution of PBS, 1% Triton X-100, 1% bovine serum albumin, and 1.5% normal serum (from the appropriate Vector Elite kit). The primary antibody was added, and sections were incubated overnight at 4°C on a tissue rocker. Further processing was with the Vector “ABC” method using the appropriate Vector Elite kit (mouse or rabbit; Vector Laboratories, Burlingame, CA), followed by visualization with a 3,3′-diaminobenzidine (Sigma-Aldrich now Thermo Fisher Scientific) protocol, giving brown staining, or a glucose oxidase modification of the protocol, giving gray-black staining (Shu et al., 1988; Van der Gucht et al., 2006). Sections were mounted on gelled slides, dehydrated in 70, 95, and 100% alcohol, cleared in xylene, and coverslipped with Permount (Fisher Scientific). Table 2 shows the primary antibodies and dilutions used. The figure legends indicate which antibody was used for the sections shown.

We examined sections with a Leitz Dialux 20 light microscope, using objectives 1.6× (numerical aperture 0.05), 2.5× (0.08), 2.5× (0.08), 6.3× (0.22), 10× (0.30), and 25× (0.65). We captured digital images (1,600 × 1,200 pixels) with a SPOT Insight Color Mosaic camera. Brightness, contrast, and color of the images were adjusted, and figures were assembled with Adobe Photoshop software (San Jose, CA).

We found expression of glycine receptors on somata and processes of neurons in all CN subdivisions, the DCN, PVCN, and AVCN, in each case examined. Figure 1A shows the immunolabel in the DCN. There is a band of label (between the white arrowheads) about 200 μm below the surface of the DCN and about 400 μm wide, running parallel to the surface. This band is composed of immunolabeled somata (example in Figure 1B, arrow) embedded in a meshwork of immunostained processes running at all orientations. Figure 1C shows immunolabeled neurons in the DCN of a second case. Figures 1D,E show the PVCN on this section. There are many immunolabeled somata of a variety of shapes; the proximal dendrites of many neurons are labeled (example in Figure 1E, arrow). There is a range of soma sizes, shapes, and orientations of both somata and dendrites relative to the dorsoventral axis of the PVCN. Figure 1F shows immunolabeled somata and puncta in the PVCN of another case. Figures 1G,H show immunostaining in the AVCN. There are immunostained neurons with mainly round somata (example in Figure 1H, arrow) throughout, as well as scattered puncta. A similar staining pattern is shown for another case in Figure 1I. The diversity of GLYR-ir neuron types in the PVCN is also illustrated in Figure 2, which shows higher magnification images of nine GLYR-ir neurons from the section shown in Figures 1D,E.

Anatomical and physiological studies in mice and rats have suggested that octopus cells in those species receive little, if any, glycinergic input (Golding et al., 1995; Zafra et al., 1995a; Friauf et al., 1999). We asked whether that might also be true in humans and chimpanzees. Octopus cells in several species have been shown to express the calcium-binding protein calretinin (CR; Lohmann and Friauf, 1996; Adams, 1997; Kulesza Jr, 2014; Felix et al., 2017). We compared the distribution of glycine receptor (GLYR)-immunoreactive (ir) cells in the PVCN with the distribution of CR-ir neurons. Figure 3 shows immunostaining with the two antibodies on two sections about 200 μm apart. Immunoreactivity to GLYR (Figures 3A,B) shows a ventral and lateral cluster of immunolabeled cells in the PVCN. The white arrows (Figure 3A) show the medial border of the PVCN. The higher magnification image (Figure 3B) shows GLYR-ir neurons more laterally in the PVCN. The black line encloses a medial region in which no immunolabeled neurons are found. The distribution of these neurons is compared with the distribution of CR-ir neurons on a section about 200 mm caudal (Figures 3C,D). The white arrows in Figure 3C show the lateral border of PVCN with CR-ir neurons found relatively medially. The higher magnification image shows that there is a cluster of CR-ir neurons within the region in which there were no GLYR-ir neurons (black outline). This result is consistent with a lack of glycinergic input to at least part of the OCA in humans.

The distribution of glycinergic input to the CN, as shown by immunoreactivity to the neuronal glycine transporter (GLYT2), aligns well with the distribution of neurons expressing the glycine receptor. We saw immunolabeled fibers and puncta in the DCN (Figures 4A–C), PVCN (Figures 4D–F), and AVCN (Figures 4G–I). Figure 4A shows that immunolabeling for GLYT2 identifies a band of label in the DCN that is similar in width and position to the band shown in Figure 3A. This band consists of labeled processes, many running parallel (Figure 4B, arrow; Figure 4C), and many scattered labeled puncta. Figure 4D shows immunolabel medially in VCP on the same section. The arrow indicates a neuron surrounded by labeled fibers, as shown at higher magnification in Figure 4E. The arrowhead in Figure 4E shows a small band of immunolabeled fibers Immunolabeled neurons are also shown in Case 168 in Figure 4F (example at arrow). More rostrally, in the AVCN (Figures 4G–I), there are somata encircled by immunolabel (Figures 4H,I examples at arrows) as well as labeled fibers throughout (example at arrowhead in Figure 4H).

We used Figures 4–6 in Strominger et al. (1977) and Figure 1 in Heiman-Patterson and Strominger (1985) as well as our own Nissl-stained sections, to identify the CN subdivisions in chimpanzees. Immunoreactivity to the neuronal glycine transporter suggests glycinergic input to DCN, PVCN, and AVCN. Figure 5A shows immunolabel in the DCN dorsally and the PVCN ventrally; the white outline shows a region of sparse immunolabel in the PVCN. The rectangles show the locations of the higher magnification images of the DCN (B), PVCN (E), and the region of light immunostaining (H). Figure 5B shows immunolabeled fibers (example at arrow) and puncta consistent with glycinergic axons in the DCN. Immunolabel also surrounds some somata (the arrowhead shows an example in deep DCN). Figure 5E shows the immunolabel in the PVCN. There are immunolabeled fibers and puncta, and many neurons are also surrounded by immunolabels (example at arrowhead). Figure 5H shows a higher magnification image of the region with a light GLYT2 immunolabel. The arrowhead shows a neuron at the dorsolateral edge of this region that is surrounded by immunolabel. Only a few immunolabeled fibers cross the rest of that region. Figure 5C shows a section about 1 mm caudal immunostained for NPNFP. The white outline shows the location of the region defined in A with a very light immunolabel. There are NPNFP-ir neurons, some resembling octopus cells (Figures 5C,D, neuron at asterisk) in the region outlined in white. These images are consistent with the suggestion that the OCA in the chimpanzee may lack glycinergic input. In the AVCN (Figures 5F,G), there is also immunolabeling surrounding many neurons. The lower magnification image shows that the immunoreactivity in the AVCN is distributed throughout. The higher magnification image (Figure 5G) shows immunolabel surrounding round or oval somata; these are likely bushy cells (example at arrow in Figure 5G).

As expected, immunolabels for the glycine receptors and the neuronal glycine transporter suggest that there is glycinergic input to DCN, AVCN, and PVCN in cats. Figure 6A shows immunolabels for the glycine receptor in the DCN and PVCN. In the DCN, there is a very dark band of label about 250 μm wide extending to the surface of the brainstem (Figures 6A,B, white arrowheads). Below that band are scattered immunolabeled somata (Figure 6B, examples at arrows). In the PVCN, immunolabeled somata are sparser dorsomedially (large white arrow in Figure 6A, possible location of OCA) and denser ventrally. The immunolabeled somata are of a variety of shapes (example at arrow in Figure 6C). Proximal processes are also labeled, suggesting glycinergic input to both dendrites and somata. More rostrally, in the AVCN (Figures 6D,E), there are many labeled neurons throughout. The immunolabel outlines somata (Figure 6E, example at arrow); there are also immunostained fibers (arrowhead).

The pattern of immunolabeling with GLYT2 (Figures 6F–N) is also consistent with glycinergic input to the DCN (Figures 6F,G), PVCN (Figures 6F,H), and AVCN (Figures 6J,K). In the DCN, there is a band of darker label along the outer margin (Figure 6F, arrowheads), similar to the pattern seen with the GLYR antibody. Deeper in the DCN are immunolabeled puncta and somata outlined by immunoreactivity (Figure 6G, example at arrow). There is a region of PVCN with a sparse label (Figure 6F, white arrowheads); this region may correspond to the OCA. Outside of that region, there are outlined somata and puncta (Figure 6H, examples at arrows). Figure 6I shows a section immunolabeled for NPNFP; there is a good label of somata and dendrites of octopus cells. The OCA is indicated by white arrowheads. In the AVCN, immunolabeled somata are distributed throughout (Figures 6J,K). As in PVCN, there are neurons outlined by immunolabeling (Figure 6K, examples at arrows). To compare our data on the cat with earlier reports, we examined immunoreactivity to GLYT2 in the cat SOC. We found immunolabels in the LSO, MSO, and MNTB (Figure 6L). Figure 6M shows an example of a neuron in the MSO outlined by immunolabel. Figure 6N shows a very darkly immunolabeled neuron in the MNTB. This label may reflect cytoplasmic staining, similar to what was seen in glycinergic neurons in rats (Friauf et al., 1999).

Our data suggest that, despite differences in DCN organization, glycine is an inhibitory transmitter in the CN of humans and chimpanzees. We saw widespread expression of glycine receptors on somata and processes of neurons in the DCN, PVCN, and AVCN, with a parallel distribution of glycinergic fibers and terminals. Our results in the cat are consistent with earlier anatomical and neurochemical studies of the CN in that species and also consistent with the differences in laminar organization of the DCN between cats and primates.

In the human DCN, immunoreactivity to GLYRs showed a dense band of label that was composed of immunolabeled somata and processes. The location of the immunolabel was similar to antibodies to the neuronal glycine transporter. We found a similar band of immunolabel in the DCN with an antibody to NPNFP (Baizer et al., 2014; Figures 4A–C). In that study, we showed that the major neuronal components of this band are the fusiform cells and their processes. In humans, these somata and dendrites are found at all different orientations (Moore and Osen, 1979; Moore, 1980; Baizer et al., 2014), in contrast to the orderly arrangement of somata seen in other species. The similarity of the band of labels with GLYR, GLYT2, and NPNFP antibodies suggests that there are glycinergic inputs to the fusiform cells of the DCN. Such a result is consistent with what has been shown in other mammals (Caspary et al., 1987; Golding and Oertel, 1997; Ostapoff et al., 1997; Doucet et al., 1999; Caspary et al., 2005; Wang et al., 2011).

The pattern of immunolabel for GLYT2 in the DCN was similar in chimpanzees, with a band of label below the surface and parallel to it.

We saw GLYR-immunolabeled neurons and processes in both the PVCN and AVCN. In the PVCN, there were labeled somata of several different shapes and sizes, likely including multipolar, bushy, and spindle-shaped neurons. The diversity of these neurons is similar to the diversity of Golgi-stained cells in the PVCN of the cat (see Figure 7 in Brawer et al., 1974) and mouse (see Figures 21, 22 in Webster and Trune, 1982).

In rats, two studies showed very little immunoreactivity to the neuronal glycine transporter in the OCA of rats (Zafra et al., 1995b; Friauf et al., 1999), suggesting that the OCA in that species lacks glycinergic input, a result consistent with electrophysiological data (Golding et al., 1995). However, Piechotta et al. (2001) showed glycine receptors on neurons around the periphery of the OCA; this may reflect an intermingling of octopus cells and other neurons outside of the OCA center. A lack of glycinergic input to octopus cells may not be true in all species; Wenthold et al. (1988) found expression of glycine receptors in octopus cells in the guinea pig. Octopus cells have been described in caudal PVCN in humans and chimpanzees (Heiman-Patterson and Strominger, 1985; Adams, 1986; Kulesza Jr, 2014). Several studies have found that octopus cells express the calcium-binding protein calretinin (CR; Adams, 1997; Bazwinsky et al., 2008; Felix et al., 2017), and our data from cats suggested that they also express NPNFP. We used those markers to try to locate octopus cells and the OCA in humans and chimpanzees. The borders of the OCA cannot be determined on the basis of immunoreactivity to CR or NPNFP since both are expressed in multiple neuron types (see Figure 6I, also Kulesza Jr, 2014; Baizer et al., 2018a). Comparison of the distributions of GLYR-ir and CR-ir neurons in humans and of NPNFP-ir neurons with GLYT2 immunoreactivity in chimpanzees does suggest that there is little or no glycinergic input to at least the central part of the OCA in both species.

We also saw immunoreactivity with glycine receptor antibodies on round somata, presumably spherical bushy cells, in the rostral AVCN. Immunoreactivity to glycine receptors has been seen on the somata of AVCN bushy cells in multiple species, including mice, guinea pigs, and macaque monkeys (Wenthold et al., 1988; Gomez-Nieto and Rubio, 2011; Lin and Xie, 2019).

Unlike humans, cats have a laminar DCN with an orderly arrangement of fusiform neurons (Brawer et al., 1974). There was a major difference in the pattern of immunolabeling in the DCN between cats and humans. In cats with both antibodies, there was an outer band of darker immunoreactivity extending to the surface. This is the layer in which the dendrites of the fusiform cells are found, and the label may reflect glycinergic input to the dendrites of fusiform cells. The pattern was different from the pattern in humans, in which there was a band of immunoreactivity below the surface, consistent with the differences in laminar organization of the DCN and the orientation of the fusiform cells between the two species.

Immunolabeling with both antibodies showed glycinergic input to PVCN and AVCN in cats. The presence of glycine receptors in the CN of cats was in agreement with Glendenning and Baker (1988). Glycinergic input to VCN bushy cells has also been reported in rhesus monkeys (Gomez-Nieto and Rubio, 2011). There also seemed to be little or no glycinergic input to octopus cells of the OCA, a result consistent with findings in the rat (Zafra et al., 1995b; Friauf et al., 1999). The immunolabeling pattern in the SOC we observed with the GLYT2 antibody was quite similar to the pattern of immunostaining seen in cats using an antibody to glycine (Spirou and Berrebi, 1997, cat, Figure 4). There were labeled neurons in the MNTB, as well as evidence for glycinergic input to neurons in the MSO.

There is clear evidence for glycinergic input to the human CN; however, determining the source of that input is difficult in human tissue. Many studies have used antibodies to glycine itself to visualize glycinergic neurons (Aoki et al., 1988, rat; Gates et al., 1996, rat; Spirou and Berrebi, 1997, cat; Doucet et al., 1999, rat). However, such antibodies recognize glycine conjugated to glutaraldehyde and depend on having perfused animals with glutaraldehyde in the perfusate, which is not possible for the formalin-fixed tissue we used in our studies.

We must therefore rely on data from other species to consider possible sources of glycinergic input. There are glycinergic neurons in the DCN of other species; however, the existence of two classes of glycinergic neurons, tuberculoventral cells, and cartwheel cells, in human DCN has been challenged (Adams, 1986). In the VCN, commissural neurons and two populations of stellate cells are glycinergic (the D-stellate cells and the L-stellate cells; Wenthold, 1987; Ngodup et al., 2020). While multipolar/stellate cells have been identified in humans (Adams, 1986), available techniques have not allowed identification of the three classes of stellate cells (D-stellate, L-stellate, and M-stellate) or of their transmitters. In baboons, Moore et al. (1996) found many glycinergic neurons, presumably stellate cells, in the DCN and fewer in the VCN; such may exist in chimpanzees and humans. Another possible source of input to the CN is from the glycinergic neurons of the SOC. Glycinergic neurons have been found in several SOC nuclei: the MNTB, the LNTB, and the LSO (Aoki et al., 1988; Spirou and Berrebi, 1997). Neurons in various SOC nuclei project to the CN (Winter et al., 1989; Schofield, 1991, 1994); some of these neurons may be glycinergic. For both humans and chimpanzees, then, there is strong evidence of glycinergic input to the CN, but whether the glycinergic neurons are in the CN or SOC is not known.