Afferent connections to the dorsal nucleus of the lateral lemniscus of the mustache bat evidence for two functional subdivisionsкод для вставкиСкачать
THE JOURNAL OF COMPARATIVE NEUROLOGY 373~575-592 ( 1996) Afferent Connections to the Dorsal Nucleus of the Lateral Lemniscus of the Mustache Bat: Evidence for Two Functional Subdivisions LICHUAN YANG, &IN LIU, AND GEORGE D. POLLAK Department of Zoology, University of Texas at Austin, Austin, Texas 78712 ABSTRACT The dorsal nucleus of the lateral lemniscus (DNLL) of the mustache bat, Pteronotus parnellii, was found to consist of two divisions. The neurons in each division were distinguished by their temporal discharge patterns evoked both by tone bursts and sinusoidal amplitudemodulated (SAM) signals. Neurons in the anterior one-third of the DNLL responded to tone bursts with an onset discharge pattern and only phase-locked to SAM signals with low modulation frequencies ( < 300 Hz). Neurons in the posterior two-thirds of the DNLL responded to tone bursts with a sustained discharge pattern and phase-locked to SAM signals with much higher modulation frequencies (400-800 Hz). In addition, there was a different frequency representation in the two divisions. The frequency representation in the posterior division was from about 15 to 120 kHz, whereas in the anterior division it was only up to 62 kHz. The physiological differences were further supported by data from experiments that revealed the sources of afferent projections to the two DNLL divisions. Retrograde labeling showed that the afferent projections to the two divisions were from different neuronal populations. Input differences were of two types. Some nuclei projected to one or the other DNLL division, but not to both. For instance, the ventral nucleus of the lateral lemniscus projected predominately to the anterior DNLL and provided little or no inputs to the posterior DNLL, whereas the medial superior olive innervated the posterior but not the anterior DNLL. Other lower nuclei projected to both DNLL divisions. These include the contralateral cochlear nucleus, the ipsi- and contralateral lateral superior olives, the intermediate nucleus of the lateral lemniscus, and the contralateral DNLL. However, the projections to each division of the DNLL originate from different neuronal subpopulations in each lower nucleus. The functional implications of these findings are discussed in the context of the possible impacts that the two DNLL divisions exert on their postsynaptic targets in the inferior colliculus. 1~ Wiley-LL<s, Inc. Indexing terms: amplitude modulation, temporal discharge patterns, differential projections, auditory pathway The dorsal nucleus of the lateral lemniscus (DNLL) occupies a strategic position in the ascending auditory pathway. Located immediately below the inferior colliculus, it receives innervation from a complement of lower nuclei similar to those that innervate the inferior colliculus, and it sends strong bilateral projections to the central nucleus of the inferior colliculus (ICc; Glendenning et al., 1981; Kudo, 1981; Zook and Casseday, 1982, 1985, 1987; Ross et a]., 1988; Shneiderman et al., 1988; Ross and Pollak, 1989; Hutson et al., 1991; Bajo et al., 1993; Merchan et al., 1994; Huffman and Covey, 1995; Vater et al., 1995). DNLL neurons are also distinguished by two additional features: 1) They are largely binaural, and those neurons tuned to ( 1996 WILEY-LISS, INC. high frequencies are driven by stimulation of the contralatera1 ear and inhibited by stimulation of the ipsilateral ear (Brugge et al., 1970; Buckthought eta]., 1993; Covey, 1993; Markovitz and Pollak, 1994; Yang and Pollak, 1994a); and 2) they are strongly immunoreactive for antibodies against conjugated y-aminobutyric acid (GABA) or glutamic acid decarboxylase (GAD), the rate-limiting enzyme for the synthesis of GABA (Adams and Mugnaini, 1984; Thompson et al., 1985; Moore and Moore, 1987; Roberts and Ribak, Accepted May 5, 1996. Address reprint requests to Dr. George D. Polkdk, Department of Zoology, University of Texas at Austin, Austin, T X 78712. 576 1987; Glendenning and Baker, 1988; Vater et al., 1992a; Covey, 1993; Vater, 1995; Winer et al., 1995). The prominent immunoreactivity of DNLL neurons for GABA suggests that the DNLL exerts a powerful inhibitory influence on its targets in the ICc. Indeed, Oliver and Shneiderman (1989) report that in the cat, DNLL projections account for roughly one-third of the inhibitory synaptic contacts in the ICc. The putative inhibitory impact of the DNLL on its targets in the ICc, suggested by its connections and neurochemistry, has been confirmed by neurophysiological studies. In several studies, binaural properties of high-frequency ICc neurons were changed dramatically either when GABAergic inhibition was blocked by the iontophoretic application of bicuculline or when the DNLL was reversibly inactivated with pharmacological agents (Faingold et al., 1989, 1993; Li and Kelly, 1992; Vater et al., 1992b; Park and Pollak, 1993; Mug et al., 1995). The evidence provided by these studies leaves little doubt that the inhibitory influences of the DNLL play a substantial role in forming the binaural properties of many ICc neurons. The DNLL may play an even more pervasive and complex role in the processing of acoustic information than indicated by the studies cited above. Recent studies show that DNLL neurons respond in unusual ways to multiple binaural stimuli (Yang and Pollak, 1994a,c). An initial stimulus that is more intense at the inhibitory ear than at the excitatory ear evokes a long-lasting inhibition in DNLL cells. The long duration of the inhibition prevents DNLL cells from discharging to subsequent stimuli that would normally be excitatory to them. Thus, whether a binaural signal can drive DNLL cells depends upon the cell’s immediate history (whether an initial signal received prior to the second signal evoked an inhibitory response). These response properties, and the way in which they would impact their targets in the ICc, may well have a profound influence on the way the auditory system processes the reception of multiple signals that emanate from various regions of space. Although the binaural properties of high-frequency DNLL neurons are homogenous in that almost all are excitatoryinhibitory (EI), their discharge patterns evoked by tone bursts presented to the contralateral (excitatory) ear are more variable. Some EI neurons discharge only at the onset of a tone burst, whereas others respond with a sustained discharge whose duration corresponds to the duration of the tone burst (Aitkin et al., 1970; Covey, 1993; Markovitz and Pollak, 1993; Yang and Pollak, 199413). Heterogeneity among the population of DNLL neurons is also indicated by anatomical studies. Several investigators report that the DNLL in the cat and rat is composed of several types of neurons and that each type is localized to a particular DNLL region (Kane and Barone, 1980; Iwahori, 1986; Shneiderman et al., 1988; Bajo et al., 1993). The significance of the various cell types, however, is unclear. In a recent intracellular study of rat DNLL neurons in brain slices, Wu and Kelly (1995) report that the various cell types all have similar intrinsic membrane characteristics and respond in a similar way to sustained depolarizing currents, suggesting that all of the cell types have a similar complement of voltage-sensitive channels. Here we report on new features of the DNLL of the mustache bat. This animal captures prey and orients in the dark by emitting loud echolocation calls and listening to the echoes reflected from objects in its environment (Griffin, L. YANG ET AL. 1958; Novick and Vaisnys, 1964; Pollak and Casseday, 1989). Its echolocation calls are characterized by a long constant-frequency component and are terminated by a brief downward-sweeping frequency-modulated component. The constant-frequency component of each call is emitted with four harmonics, at 30, 60, 90, and 120 kHz. The 60-kHz component is always dominant. When a flying insect crosses its path, the insect’s moving wings impose both frequency and amplitude modulations on the long constant-frequency component of the echo reflected from the insect. It is from these modulation patterns on the echoes of the 60-kHz constant frequency component that the bat not only distinguishes its prey from background objects but also identifies its prey (Goldman and Henson, 1977). Because of the emphasis placed on this component, 60 kHz is overrepresented in every auditory nucleus. This is one reason that the DNLL of the mustache bat is enlarged relative to the DNLL of other bats as well as other mammals (Ross et al., 1988; Markovitz and Pollak, 1993; Winer et al., 1995). For the past several years, we have been studying the physiology of the mustache bat’s DNLL as well as its impact on its targets in the ICc (Yang and Pollak, 1994a-c). During the course of our recent studies, we noted that the DNLL is divided into two functional divisions based on temporal discharge patterns evoked by tone bursts and more complex signals. Here we illustrate the two functional divisions and show that each division receives a unique afferent innervation from cell groups in lower nuclei. MATERIALS AND METHODS Surgical procedure Thirteen mustache bats, Pteronotus parnellii parnellii, were used in this study. Prior to surgery, each animal was anesthetized with methoxyflurane inhalation (Metofane, Pitman-Moore, Inc.), and 0.02 mgig neuroleptic (InnovarVet, Pitman-Moore, Inc.) was injected intraperitoneally. The hair on the head was removed with a depilatory, and the head was secured in a head holder with a bite bar. The muscles and skin overlying the skull were reflected, and Lidocaine (Elkins-Sinn, Inc.) was applied topically to all open wounds. The surface of the skull was cleared of tissue, and a foundation layer of cynoacrylate and small glass beads was placed on the surface. A small hole was then drilled around the central portion of the inferior colliculus using the landmarks visible through the skull for orientation. The bat was then transferred to a heated recording chamber where it was placed in a restraining apparatus attached to a custom-made stereotaxic instrument (Schuller et al., 1986). A small metal rod was cemented to the foundation layer on the skull and then attached to a bar mounted on the stereotaxic instrument to ensure uniform positioning of the head. A ground electrode was placed between the reflected muscle and the skin. Recordings were begun after the bat recovered from the anesthetic. Supplementary doses of the neuroleptic were given if the bat appeared to be in discomfort. These methods have been approved by the Institutional Animal Care and Use Committee of the University of Texas. Localization of the DNLL A change in tonotopic sequence was used to determine when the electrode had left the ICc and entered the DNLL, SUBDIVISIONS IN DNLL as was done in previous studies (Markovitz and Pollak, 1993; Yang and Pollak, 1994a-c). As the electrode was advanced through the ICc, a repetitive tone burst was presented to the contralateral ear that evoked background, multiunit activity. This background activity was more prominent with the larger-tipped pipettes used for the injection of tracers, but it was also discernible with the finer-tipped pipettes used to record single-unit activity. The best frequency, the frequency to which the background activity was most sensitive, was then determined every 50-100 IJ-malong the penetration. A reversal in the best frequency of the background activity at a depth of about 2,200-2,400 pm signaled that the electrode had left the ICc and entered the DNLL. For the next 400-500 pm, there was a systematic change in best frequency from low to high. A second abrupt change in best frequency indicated that the electrode had left the DNLL and entered the intermediate nucleus of the lateral lemniscus (INLL). These changes in best frequency, as well as the tonotopic organization of the DNLL, served as reliable indicators of electrode location. Systematic mappings of the best frequencies and temporal discharge patterns of single- or multiunit activity were conducted in the DNLL of four bats. In these experiments a single micropipette filled with horseradish peroxidase (HRP; 5% in 0.9% NaCl) was used for making multiple penetrations along the anterior-posterior axis of the DNLL. The electrode position at each recording site was recorded by digital positioners attached to the stereotaxic instrument. In the last penetration of each experiment, a small iontophoretic injection of HRP (1.5 pA for 3 minutes, electrode positive) was made at a selected site in the DNLL. The location of the HRP injection site was later used as an adjustment point to reconstruct the distribution of best frequencies and temporal discharge patterns in the DNLL. 577 tone bursts and SAM signals were fed to a window discriminator and then to a Macintosh 7100 computer for generating raster displays. Administration of retrograde tracers Injections of tracers were made with pipettes having tip diameters of 5-10 IJ-m that were filled with 10% lysine fixable biotinylated dextran (BD; 3 kD, pH 7.4) or 10% lysine fixable dextran conjugated to fluorescein ( 3 kD, pH 7.4) or tetramethylrhodamine ( 3 kD, pH 7.4). The dextrans were obtained from Molecular Probes Inc. These fluorophores were chosen because they can be applied iontophoretically, they transport well retrogradely, and they fluoresce brightly. The electrode was positioned on the surface of the inferior colliculus by using visual landmarks viewed with an operating microscope. Subsequently, the electrode was advanced from outside of the experimental chamber with a piezoelectric microdrive (Burleigh 7121W). Injections of dextrans were made by passing positive current (2.5-5 pA) that was pulsed (5 seconds on, 5 seconds off) for 8-20 minutes. After a 48-hour survival time, the animal was deeply anesthetized with methoxyflurane inhalation (Metofane, Pitman-Moore, Inc.). It was then perfused transcardially with phosphate-buffered saline (PBS; 0.1 M, pH 7.4) followed by a fixative solution of 4% paraformaldehyde and 1.25% glutaraldehyde in phosphate buffer (0.1 M, pH 7.4). For animals that were injected with dextran amines conjugated to fluorescein or rhodamine, 1.25% glutaraldehyde was spared from the fixative solution because it caused autofluorescence in the tissue. After fixation, the brain was dissected out and postfixed for 2 hours. It was then transferred to a 30%sucrose solution overnight for cryoprotection. The brain was sectioned on a freezing microtome at a thickness of 40 pm. With BD injections, Acoustic stimuli and data acquisition sections were first incubated in 0.3%Triton X-100 diluted Acoustic stimuli were presented through %-inchBruel & with PBS for 1 hour. The sections were rinsed three times Kjaer condenser microphones driven as loudspeakers. Micro- in PBS and then reacted with ABC procedure (Vectastain pipettes whose tips were broken to a diameter of about 2 ABC Elite kit, Vector Laboratories) followed by diaminobenIJ-mwere used for recording multiunit activity, and micropi- zidine reaction (DAB peroxides substrate kit, Vector pettes with tip diameters of less than 1.0 pm were used to Laboratories). Sections were then mounted, air-dried, record the action potentials of single units. When a unit was and counterstained with cresyl violet. With injections of isolated, its best frequency was determined with tone dextran amines conjugated to fluorescein or rhodamine, bursts that were 20 msec in duration and had 0.5-msec sections were mounted on the slides without further reacrisefall times. The tone bursts were generated by shaping tion and coverslipped with FITC-Guard mounting medium a sinewave from a Wavetek function generator (model 136) (TESTOG). with a custom-made analog switch. The binaural property Sections from animals that received injections of BD were was then assessed by driving the cell with a tone burst examined under brightfield illumination. Sections from presented to the contralateral ear at 20 dB above the animals injected with fluorophores were evaluated with threshold of the best frequency. With the intensity fixed at the contralateral ear, the same frequency was then pre- epi-illumination using appropriate filter cubes. The fluoressented simultaneously to the ipsilateral ear, and the inten- cence of cells filled after injecting dextran conjugated to sity of the ipsilateral signal was increased from 10-20 dB tetramethylrhodamine was typically a little more intense below the intensity at the contralateral ear to 40 dB above than cells filled after injections of dextran conjugated to that intensity in steps of 10 dB. After determining the fluorescein. Moreover, wavelengths passed by the fluoresneuron’s binaural property, the ipsilateral signal was turned cein cube could also evoke a slight fluorescence in cells off, and sinusoidal amplitudemodulated (SAM) signals of strongly stained by rhodamine. To ensure that double70-msec duration were presented to the contralateral ear. labeled cells could be distinguished from such bleed-over, in The SAM signals were created digitally on a Macintosh one experiment we injected both rhodamine- and fluorescein7100 computer. The carrier frequencies of the SAM signals conjugated dextran in the same site. The labeled cells in this were set at the neuron’s best frequency. A set of SAM case fluoresced intensely when viewed with both cubes. signals, each with a different modulation frequency, was This served as a standard, and we considered cells as presented to each unit. The modulation frequencies ranged doubled-labeled only when clear fluorescence was observed from 100 to 800 Hz in 100-Hz steps. Spikes evoked both by with both cubes. I’ a 3 579 SUBDIVISIONS IN DNLL RESULTS General features of the mustache bat DNLL As seen in sagittal and transverse sections (Fig. 1 B,C), the DNLL has well-defined boundaries that distinguish it from surrounding structures. Dorsally it is separated from the inferior colliculus by a nearly acellular space filled with the fibers entering the ventrolateral border of the central nucleus of the inferior colliculus (ICc). Ventrally it is demarcated from the intermediate nucleus of the lateral lemniscus (INLL) by crossing fibers of the lateral lemniscus. Medially and anteriorly it is distinct from the paralemniscal nucleus by the ascending fiber bundles of the lateral lemniscus. Laterally and posteriorly the DNLL extends almost to the edges of the brain surface where it is encapsulated by a thin layer of lateral lemniscal fibers. A three-dimensional view suggests that the DNLL of the mustache bat is cone-shaped with a large, rounded posterior region that tapers into a smaller anterior region (Fig. 1D). The anterior-posterior axis is about 850 pm, and the lateral-medial axis is about 500 pm. The dorsal-ventral axis is about 450 pm posteriorly and tapers to about 100 pm anteriorly. In Nissl-stained material the DNLL is organized as vertically oriented parallel columns of cell somata separated by 30-50 km of “acellular” spaces (Figs. 1B, C, 2A). Each column is about 20-30 pm in width and contains one to two rows of somata. The “acellular” space contains the dendrites and axons of DNLL cells as well as the ascending fibers of the lateral lemniscus. Figure 2B shows retrogradely labeled cells in the DNLL found after injecting biotinylated dextran amines (BD) into the 60-kHz region of both the contra- and ipsilateral ICc. The labeled cells present a Golgi-like appearance, and in this case the cells presumably correspond to the 60-kHz representation in the DNLL. The dendrites of the labeled DNLL cells were horizontally oriented and traversed the spaces between columns of somata. Because the ascending fibers of the lateral lemniscus also travel vertically in these spaces, it seems that the crossovers at these spaces, between the dendrites and the ascending lemniscal fibers, provide a substrate for the DNLL cells to capture the ascending information from lemniscal fibers. From evaluations of labeled cells we could discern no obvious cellular segregations, but rather the neuronal architecture of the mustache bats’ DNLL presented a fairly homogenous appearance. Comparison of DNLL filled cells were made following injections of BD in either the contralateral DNLL or in both the 60-kHz regions of the ipsi- and contralateral ICc. Labeled neurons were evaluated in terms of their size, shape of their somata, and their dendritic arborization. DNLL neurons were classified into one of three groups: fusiform shaped, round or oval-shaped, and multipolar (Fig. 2C). The somata of the fusiform cells were either medium (9-10 Fm) or large (11-13 pm), with their dendrites projecting laterally (Fig. 2Ca,b). Although we did not do cell counts, fusiform cells appeared to be slightly more prevalent in the posterior two-thirds of the DNLL than in the anterior one-third of the DNLL. The round or oval-shaped cell type typically had a single primary dendrite which gave rise to two or more secondary dendrites oriented in the same direction (Fig. 2Ce,f). The somata of round cells varied in size from small (5-8 pm) to medium (9-10 pm). Fewer had large somata (11-13 pm). Round cells of each size appeared to be distributed throughout the nucleus. The third type was multipolar cells (Fig. 2Cc,d). The somata of these cells ranged from medium (9-10 pm) to large (11-13 pm), and they appeared to be slightly more frequent in the anterior one-third of the DNLL. Afferent projections to the DNLL The nuclei that provide afferent inputs to the DNLL were determined from the locations of retrogradely labeled cells following injections of either BD or rhodamine-conjugated dextran that filled almost all of the DNLL. As illustrated in Figure 3, the injected tracer was confined to the DNLL with little or no discernable spread to adjacent structures. The clear retrograde labeling that resulted from these injections is illustrated in Figure 4 by the cells in the lateral superior olive (LSO) labeled after an injection of BD or fluoresceinconjugated dextran. Three bats received large injections of this sort, and the results were similar among the cases. As shown for one case in Figure 5, labeled cells were found in the contralateral cochlear nucleus, the ipsilateral medial superior olive (MSO), the ipsilateral ventral nucleus of the lateral lemniscus (VNLL), the INLL, and in the contralateral DNLL. A small number of labeled cells was also found in the ipsilateral paralemniscal nucleus (not shown in Fig. 5) and in some ventral periolivary nuclei. The largest number of labeled cells, however, were found in the ipsiand contralateral LSOs. This complement of nuclei that provide afferent input to the DNLL in the mustache bat is very similar to the nuclei that project to the DNLL in other bats and other mammals (Glendenning et al., 1981; Kudo, 1981; Shneiderman et al., 1988; Hutson et al., 1991; Bajo et al., 1993; Merchan et al., 1994; Huffman and Covey, 1995; Vater et al., 1995). Finally, labeled cells in the ICc were rare, and those that were found were always in its ventral portion. This suggests that if the mustache bat’s DNLL receives any descending projections from the ICc, that projection is much weaker than the descending projection from the 1Cc that has been reported in the DNLL of rat and cat (Carey and Webster 1971; Hutson et al., 1991; Caicedo and Herbert, 1993). There are two subdivisions in the DNLL of the mustache bat Fig. 1. A: Brain of the mustache bat with the cerebellum partially removed to reveal the inferior colliculus (IC) and nuclei of the lateral lemniscus (NLL).Also shown are the auditory nerve (AN) and cochlear nucleus (CN).B: Sagittal section stained with cresyl violet showing the dorsal nucleus of the lateral lemniscus (DNLL) in relation to the intermediate and ventral nucleus of the lateral lemniscus (INLL and VNLL). D and P refer to dorsal and posterior, respectively. C: Transverse section showing the nuclei of the lateral lemniscus. D and M refer to dorsal and medial, respectively. D: Three-dimensional drawing of the DNLL. D, R, and L refer to dorsal, rostral, and lateral, respectively. The DNLL is enclosed by dashed lines. Scale bars = 400 Fm in B (corresponds to 1.5mm in A), 300 Km in C, 200 Fm in D. The DNLL has at least three features that are common at any locus and which serve to define it as a nucleus. First, as mentioned previously, we did not observe any obvious architectonic subdivisions in the DNLL. Second, all of the neurons that we sampled in the DNLL had the same binaural property: They were excited by stimulation of the contralateral ear and inhibited by stimulation of the ipsilateral ear and thus were excitatoryinhibitow or El. The third common feature was that there was only one tonotopic arrangement in the mustache bat DNLL: Neurons Figure 2 C SUBDIVISIONS IN DNLL Fig. 3. Transverse section through the brain of one bat showing that the injection of biotinylated dextran was confined largely to the DNLL. D and L refer to dorsal and lateral, respectively. Scale bar = 300 pm. tuned to low frequencies were represented dorsally, whereas neurons tuned to progressively higher frequencies were represented ventrally. This tonotopic pattern was also seen in previous studies (Markovitz and Pollak, 1993; Yang and Pollak, 1994a,b) and is illustrated in Figure 6. There are, however, two features that distinguish neurons in the anterior one-third of the DNLL from those in the posterior two-thirds of the DNLL. The first and principal feature is a pronounced difference in temporal discharge pattern evoked by tone bursts and sinusoidal amplitudemodulated (SAM)signals (Figs. 6, 7). When tone bursts at each neuron’s best frequency were presented to the contralateral ear, most neurons in the posterior DNLL responded with a sustained temporal discharge pattern, in which the discharge train lasted for the duration of the tone burst (Fig. 6C). In contrast, most neurons in the anterior DNLL responded with an onset pattern, in which discharges were evoked only at the beginning of the tone burst (Fig. 6C). As mentioned above, both onset and sustained neurons were inhibited by stimulation at the ipsilateral ear, and thus were EI. Examples of sustained and onset neurons are shown in Figure 6C. Onset and sustained neurons were Fig. 2. A: Transverse Nissl-stained section showing cells in the DNLL (enclosed by dashed lines) organized into vertically oriented columns. One column is indicated by arrows. B: A row of cells in DNLL labeled by injections of biotinylated dextran amines into the 6O-kHz region of both the contra- and ipsilateral central nucleus of the inferior colliculus (ICc).The labeled cells presumably correspond to the 60-kHz representation in the DNLL. Notice that neurites of cells are horizontally oriented and cross columns of somata. C: Three main types of DNLL cells. Fusiform (a,b), multipolar (c,d), and round cells (e,D. Arrows show axons. Scale bars = 100 pm in A and B, 50 bm in C. 581 often intermingled at the juncture of the two regions, but one or the other type was dominant at the more posterior and anterior portions of the DNLL (Fig. 6A, B). Sustained and onset neurons also responded differently to SAM signals (Fig. 7). Of the 77 DNLL neurons from which we recorded, the majority responded to SAM with discharges that were phase-locked to each cycle of the amplitude-modulated signal. The difference was that the discharges of the 45 sustained neurons phase-locked to modulation frequencies as high as 400-600 Hz, and a few phase-locked to rates as high as 800 Hz; on the other hand 75% (24132) of onset neurons only phase-locked to low modulation frequencies, typically below 300 Hz. Twentyfive percent of the onset neurons (8132) did not to phaselock to any of the SAM signals that we presented. The three neurons in Figure 7 are representative and illustrate the pronounced disparity in the phase-locking of sustained and onset neurons to the time variations of SAM signals. The second feature was a difference in the frequency representation between the anterior DNLL, the region dominated by onset neurons, and the posterior DNLL, the region dominated by sustained neurons. Although both regions had a similar tonotopic sequence, in which neurons tuned to low frequencies were located dorsally and neurons tuned to progressively higher frequencies were located ventrally, the total range of frequencies represented in the anterior DNLL was smaller than the range represented in the posterior DNLL (Figs. 6 and 8, top panel). Specifically, the anterior DNLL had neurons tuned from about 15 kHz to about 62 kHz. Neurons tuned to frequencies higher than about 62 kHz were rarely encountered even in the most ventral regions of the anterior DNLL (Fig. 8, top panel). In contrast, the frequency representation in the posterior DNLL ranged from about 15 kHz to 120 kHz and encompassed the mustache bat’s entire hearing range (Fig. 8, top panel). A summary figure illustrating the general tonotopic arrangement of the DNLL and the difference in frequency representation between the anterior and posterior divisions is shown in Figure 6D. In summary, the DNLL is functionally divided into a smaller anterior division and a larger posterior division, in which neurons in each division have a different temporal discharge pattern to tone bursts and SAM signals. However, the neurons in both divisions have the same binaural property, they have a similar complement of cell types, and they display a comparable tonotopic organization, although the posterior division encompasses the bat’s entire hearing range whereas the anterior division has a more restricted frequency representation. Afferent inputs to each of the DNLL divisions We next determined whether the same set of lower nuclei provided inputs to both divisions of the DNLL. We did this in six bats by first mapping the DNLL to identify the two DNLL divisions based on the discharge patterns of their neurons evoked by tone bursts. After verifying t h a t 4 e electrode was in one or the other division, the electrode was advanced or retracted so that it was 100-200 pm from the dorsal surface of the DNLL, which was about in the middle of the dorsal-ventral axis of either the anterior or posterior subdivision. In four bats BD was injected in only one DNLL division (in the anterior division in two bats and in the posterior division in two bats). In two other bats both divisions of the DNLL were injected; rhodamine-conjugated dextran was injected into one DNLL division, and fluores- L. YANG ET AL 582 Fig. 4. Labeled cells in the contralateral lateral superior olive (enclosed by dashed lines) in two bats following injections of fluorescein conjugated to dextran (left) or biotinylated dextran (right) in DNLL. Scale bar = 100 pm. cein-conjugated dextran was injected into the other division. The time between the injections of the two fluorophores was 30-45 minutes. We refer to the experiments that used two fluorophores as double-labeling experiments. Examination of the histological sections showed that the injections were confined to the desired division of the DNLL. The main finding is that the two DNLL divisions receive different afferent projections. The different projections derive in part from different lower nuclei and in part from different cell groups of some nuclei that project to both DNLL divisions. The percentage of cells in each lower nucleus that were labeled is shown in Figure 9. Figure 10 shows the location of the labeled cells in one case in which fluorescein-conjugated dextran was injected into one DNLL division and rhodamine-conjugated dextran was injected into the other division. Below, we turn first to the nuclei that project to both DNLL divisions and show that the cell groups in each nucleus that project to the anterior DNLL are different from those that project to the posterior DNLL. We then consider the projections from the VNLL and MSO, the two nuclei that project to either the anterior or posterior DNLL divisions but not to both divisions. Nuclei that project to both DNLL divisions Both DNLL divisions received projections from the contralateral cochlear nucleus, the LSOs on both sides, the ipsilateral INLL, and from the opposite DNLL (Figs. 9, 10). The projections from the cochlear nucleus were sparse in that only a small number of cells were labeled in most of the cases. The projections to the anterior DNLL were from the anteroventral cochlear nucleus and occasionally from the posteroventral cochlear nucleus. The projections to the posterior division were divided equally between the anteroventral cochlear nucleus and the dorsal cochlear nucleus (Fig. 10). Although the anteroventral cochlear nucleus sent projections to both the anterior and posterior DNLL, it was unclear from the BD cases whether the same cells in the anteroventral cochlear nucleus projected to both DNLL divisions. In the double-labeling experiments, in which rhodamine-conjugated dextran was injected into one DNLL division and fluorescein-conjugated dextran was injectec into the other DNLL division, cells in the anteroventra cochlear nucleus were labeled either with fluorescein 01 with rhodamine, but none were double labeled (Fig. 10) This suggests that those cell groups in the anteroventra’ cochlear nucleus that project to the anterior DNLL arc distinct from those that project to the posterior DNLL. The projections from LSO were of particular interest because they provided the heaviest input to both DNLL divisions. The main result is that different LSO cell groups projected to the two DNLL divisions. We point out first that one component of the difference in projections from LSO t c the two divisions of the DNLL can be explained by a tonotopic projection system, in which LSO neurons tuned to a particular frequency send axonal projections to regions of higher nuclei that represent the same frequency. Thri tonotopic arrangement of the mustache bat LSO is parti tioned into three general frequency regions, each of approximately equal size: The dorsolateral limb represents frequencies from about 10 to 59 kHz, the ventromedial limh represents frequencies from about 63 to 120 kHz, and sandwiched between these is a central limb that is isofrequency for frequencies within 60-63 kHz-the exact frequency varies with the individual animal (Ross et al., 1988; Covey et al., 1991).Because the frequency representation in the anterior DNLL extends only to about 62 kHz, the ventromedial limb of the LSO, which represents high frequencies, would not be expected to project to the anterior DNLL. On the other hand, the ventromedial limb of the LSO should project to the posterior DNLL, which has a prominent representation of high frequencies, from 63 to 120 kHz. The two other LSO limbs, the dorsolateraland central limbs, should project tc both the anterior and posterior DNLL, because the twc DNLL divisions have “low”-frequency representations at well as 60-kHz representations. These predicted tonotopic projection patterns were con firmed. The lower panel of Figure 8 shows the counts oa‘ labeled cells in the contralateral LSOs in four bats. Thew cells were from selected sections in which all three LSO SUBDIVISIONS IN DNLL 583 6 Fig. 5. Distribution of cells labeled by a large injection of biotinylated dextran in the DNLL of one bat. The injection was made in the left DNLL. indicated by shading. The most caudal section is in the lower right and is labeled 1, and the most rostra1 section is in the top left and is labeled 6. AVCN, Anteroventral cochlear nucleus; DCN, dorsal cochlear nucleus; INLL, intermediate nucleus of the lateral lemniscus; LSO, lateral superior olive; MNTB, medial nucleus of the trapezoid body; MSO, medial superior olive; N. VII, nucleus of the seventh cranial nerve; PVCN, posteroventral cochlear nucleus; VNLL, ventral nucleus of the lateral lemniscus; VIII, auditoly nerve. limbs were apparent in each section. In two bats, BD was injected in the anterior DNLL, and in two other bats BD was injected in the posterior DNLL. Anterior DNLL injections resulted in a large number of labeled cells in the low-frequency, dorsolateral limbs of the LSOs and in the central limbs that represent 60 kHz. Virtually no labeled cells were found in the high-frequency, ventromedial limbs. In contrast, there were labeled cells in all three limbs of the LSO following injections in the posterior DNLL. There were, however, fewer labeled low-frequency (in the dorsolatera1 limb) and more labeled 60-kHz cells (in the central limb) following posterior DNLL injections than there were with anterior DNLL injections. This pattern was most likely due to the difference in frequency representations of the anterior and posterior injection sites. All injections were made in the center of the DNLL, midway along its dorsoventral extent. In the posterior DNLL, the central region along the dorsoventral axis corresponds to the 60-kHz representation (Fig. 6D), and injections there resulted in a large number of labeled cells in the central limb of the LSO. In contrast, “low frequencies” are represented in the center of the anterior DNLL (Fig. 6D), which could explain the large number of labeled cells in the dorsolateral (“low”-frequency) limb of the LSO following anterior DNLL injections. Thus, the pattern of labeled LSO cells is consistent with the differential frequency representations in the anterior and posterior DNLLs. The most surprising result was that different populations of LSO cells projected to the anterior and posterior DNLL, and this result was obtained after taking the difference in L. YANG ET AL. 584 4 R c B P Onset Sustained contrd = 30 dB SPL contra = 30 dB SPL = 10 dFl SPL lpl= 10 dB SPL , lpl=20 dB SPL 1p.l p i = 30 dB SPI. = 20 dE SPL Ip;i = 3odB SPL '2 i p i = 4 0 W SPL : \ antcnor DNLL Ipl=40 dB SPL , Tonotopic Organization of DNLL % I 1:: 'PI D 60-62kHz i postenor DNIL . I lp.1 =so dB SPL ' ipi = SO dB SPL I Ipl=60 dB SPL 70 rns dB SPL 1p.l = 60 70 rns Fig. 6 . A. Best frequencies of multiunits recorded in several penetrations along the rostrocaudal axis of the DNLL in one bat. In all cases low frequencies were represented dorsally and high frequencies ventrally. Note that the best frequencies range from 40.7 kHz up to 112 kHz in the posterior DNLL, whereas the highest best frequency in the anterior DNLL was only 61.9 kHz. The discharge patterns evoked by tone bursts at each recording site are listed below the best frequencies. Open circles indicate sites at which onset responses were recorded, and solid circles indicate sites at which sustained responses were recorded. Dorsal and rostra1 are indicated in the upper left. B: Best frequencies of single units recorded in the more dorsal portions of the DNLL in one bat. Although units in the more ventral DNLL were not recorded, there is a clear tonotopic progression from low to higher frequencies. Low frequencies between 16 and 30 kHz are represented in the dorsal DNLL along the entire rostrocaudal axis. C: Raster displays of the discharges from two DNLL units showing that both onset neurons from the anterior DNLL and sustained units from the posterior DNLL were excited by stimulation of the contralateral ear and inhibited by stimulation of the ipsilateral ear. The intensities at the contralateral ear were held constant at 30 dB SPL in both neurons, and intensities at the ipsilateral ear were varied. The ipsilateral intensities are indicated below each histogram. Each raster display was generated by 20 repetitions of a binaural stimulus. Tone burst durations were 70 msec. D: Schematic illustration of tonotopic arrangement of the DNLL showing the loci of the three major frequency bands constituting the mustache bat's hearing range: 1) < 60 kHz; 2) 60-62 kHz; and 3 ) > 62 kHz. Also shown is the difference in frequency representation between the anterior and posterior DNLL. SUBDIVISIONS IN DNLL Sustained Neuron Posterior DNLL 585 Onset Neuron Anterior DNLL Onset Neuron Anterior DNLL Raster displays of discharges evoked by sinusoidal amplitude modulated (SAM) signals from a sustained neuron in the posterior DNLL (left 1 and from two onset neurons in the anterior DNLL (middle and right). The waveform of either the tone burst or SAM signal is shown below each raster. The top panels show discharge patterns evoked by 70-msec tone bursts at the cell's best frequency. The panels below show the discharges evoked by SAM signals of increasing modulation frequency, shown at the far left. The carrier frequencies were a t the unit's best frequency. Note the difference in the phaselocking as the SAM modulation frequency increased. All signals were 70 msec in duration and were 40 dB SPL (sound pressure level is sound pressure re 0.0002 dynes/cm2).The number at the right corner of each raster display is the total spike count. The numbers a t the top of each panel show how many neurons of each type were in our sample. The best frequencies were 85.2 kHz (left),61.25 kHz (middle),and 60.6 kHz (right1. frequency representation of the two DNLL divisions into account. This segregation of projections was confirmed in two double-labeling experiments. Although a n occasional LSO cell was labeled with both rhodamine and fluorescein, colabeled cells were rare in all cases. The vast majority of cells were labeled with either rhodamine or fluorescein but not with both fluorophores. In double-labeled experiments, unlike the BD experiments described above, the injections in both DNLL divisions were centered in locations that represented similar frequencies, between 52 and 56 kHz. Furthermore, both rhodamine- and fluorescein-labeled cells in the LSO occurred throughout the dorsolateral limbs, which encompassed low frequencizs, as well as the enlarged 60-kHz representation in the central limb (Fig. 10).These results provide strong support for concluding that the LSO cell groups which innervate the anterior DNLL are separate from those that innervate the posterior DNLL. However, we did not evaluate the morphology of the LSO cells labeled with each fluorophore, and thus are unable to conclude whether a particular LSO cell type projects to the anterior DNLL and a different LSO type projects to the posterior DNLL. Comparable results were also found in the INLL. This nucleus, like the LSO, sends projections to both DNLL divisions. BD injections in either division of the DNLL resulted in substantial labeling in the INLL (Fig. 9). From these cases, however, we could not determine whether the same or different INLL cells projected to the two DNLL divisions. This question was answered by two cases in which both rhodamine and fluorescein injections were made in the two divisions in the same animal. In both cases the INLL had both rhodamine- and fluoresceinlabeled cells, and like the LSO, there were few doublelabeled cells (Fig. 10). The DNLL on one side also projected to both DNLL divisions on the opposite side. Injections of BD in the anterior division resulted in labeling throughout the opposite DNLL, in both the anterior and posterior regions. The labeling, however, was largely confined to the dorsolateral aspects of both DNLL divisions. With injections in the posterior DNLL, labeling was also found in both DNLL divisions on the opposite side, although fewer cells were labeled than with anterior injections. The labeled cells, moreover, were also largely in the dorsolateral aspects of L. YANG ET A 586 Anterior Division Posterior Division 50- 40 A U n 30-59 60-64 65-79 c-CN c-LSO I-LSO i-MSO I-INLL I-VNLL c-DNLL I-PL c-CN c-LSO i-LSO I-MSO i-INLL I-VNLL c-DNLL I-PL 80-110 Best Frequency (kHz) Anterior injection Posterior Injection U Q Q r Anterior c] Posterior Fig. 8. Top: Distribution of best frequencies recorded from several penetrations through the anterior and posterior DNLL in one bat. Recordings were of multiunit activity. Part of these are shown in Figure 6A. Bottom: Number of labeled cells in each of the three limbs from selected sections of the LSO after injections of dextran in the anterior (two cases) or posterior divisions (two cases) of the DNLL. the opposite DNLL. The same pattern of labeling was also seen with rhodamine- and fluorescein-conjugated dextran: when each of these tracers was injected into one of the DNLL divisions in the same animal, there were few if any doubled-labeled cells, which suggests that separate DNLL populations project to the anterior and posterior divisions in the opposite DNLL. Lower nuclei that project to one or the other division of the DNLL but not to both Two nuclei, the VNLL and MSO, provided inputs to only one of the divisions of the DNLL (Figs. 9, 10). The VNLL projected almost exclusively to the anterior DNLL, and the MSO projected almost exclusively to the posterior DNLL. These differences were seen both with BD injections and when fluorescent tracers were injected into both DNLL divisions in the same bat. With injections of BD in the c-CN c-LSO I-LSO I-MSO I-INLL inject ioi injectic I-VNLL c-DNLL i-PL Fig. 9. Histograms illustrating the percentage of labeled cells i each Lower nucleus after injecting retrograde tracers into the anteric (A)and posterior (B) division of the DNLL. C: The percentage c labeled cells in each lower nucleus averaged for four cases. Cells i nuclei contralateral to the injected DNLL are indicated by c, whered cells in nuclei ipsilateral to injected DNLL are indicated by i. PL refer to paralemniscal nucleus. anterior division, substantial numbers of labeled cells wert found in the VNLL but few if any in the MSO. Conversely BD injected into the posterior DNLL resulted in labeling 11 the MSO, but virtually no cells were labeled in VNLL. 11 MSO, labeled cells were found throughout the nucleus anc thus encompassed much of the tonotopic representation. 11 the more caudal regions, labeled cells were predominately in thl ventromedial portion of the MSO, where high frequencies ar* represented (Ross et al., 1988 Covey et al., 1991). In morc rostral regions, labeled cells tended to be located in the mi&€ gion of the MSO where 60 kHz is represented, and in the moe' rostral region labeled cells were located dorsally, in the 10% frequencyrepresentation.The findingthat MSO projects large]! SUBDIVISIONS IN DNLL 587 Fig. 10. Locations of labeled cell bodies in selected transverse sections in one double-labeling case in which the dextran conjugated to fluorescein (green) was deposited in the posterior division, whereas the dextran conjugated to rhodamine (red) was deposited in the anterior division of the DNLL. The most caudal section is in the lower right and is labeled 1,and the most rostral section is in the top left and is labeled 6. Injections were in the left DNLL. Abbreviations are the same as in Figure 5. to the posterior DNLL is consistent with the results of a study by Vater et al. (1995)that traced the efferent projections from the MSO in the mustache bat. When the posterior DNLL was injected with fluorescein and the anterior DNLL with rhodamine, the results were consistent with those obtained when BD was injected into one or the other DNLL division (Fig. 10). In these double- labeling experiments, the injection of fluorescein in the posterior DNLL resulted in labeled cells in the MSO but not in the VNLL. Conversely, the injection of rhodamine in the anterior DNLL labeled cells in VNLL but not in MSO. Complementary results were obtained when we switched the tracers, injecting fluorescein in the anterior DNLL and rhodamine in the posterior DNLL. L. YANG ET AL. 588 DISCUSSION There are three main findings in this study. The first is that the mustache bat DNLL has two divisions: a n anterior division in which almost all neurons respond with a n onset discharge pattern to tone bursts and Only phase-lock to SAM signals with low modulation frequencies, and a posterior division whose neurons respond to tone bursts with a sustained dischargepattern and Phase-lock to much higher modulation rates. The second finding is that although the two DNLL divisions have a common tonotopic Organization, there is a different frequency representation in the two divisions. The frequency representation in the posterior division is from about 15 to 120 kHz, which encompasses the bat's entire hearing range, whereas the representation in the anterior division is more restricted, and only represents frequencies up to 62 kHz. The third finding is that the afferent projections to the two divisions are from different neuronal populations. Input differences are of two types, and are schematically illustrated in Figure 11. The VNLL projects predominately to the anterior DNLL and provides little or no inputs to the posterior DNLL, whereas the MSO innervates the posterior DNLL but not the anterior division. Other lower nuclei, especially the LSO, INLL, and contralateral DNLL, project to both DNLL divisions, However, the projections to the two divisions The neuoriginate from different neuronal subpopu~at~ons~ rons in the LSO, INLL, and the contralateral DNLL that project to the anterior DNLL do not project to the posterior DNLL, and the neurons in those nuclei that project to the posterior DNLL do not project to the anterior DNLL. A question that emerges with regard to the differential innervation from the VNLL and MSO is whether this, in fact, reflects a difference in the topography of the innervais a conse. tion by the two nuclei, or whether the quence of damage to the fibers from MSO that selectively passed through the posterior DNLL and fibers from VNLL that selectively passed through the anterior DNLL. If these results were due solely to the uptake of tracers by damaged fibers rather than to a differential innervation, then the fibers from both the MSO and VNLL should provide no collateral innervation to the DNLL. This is unlikely for several reasons. The first reason follows from the observations by Iwahori (1986).From Golgi-impregnated material in the rat, he observed that most ascending fibers provide collateral innervation to the DNLL. The second reason is that all previous connectional studies found that DNLL cells receive innervation from both MSO and VNLL. These studies are discussed below. The third reason follows from the recent anterograde study by Vater et al. (19951,mentioned previously. They showed that projections from the mustache bat MSO terminate in the posterior aspect of the DNLL. Their findings with anterograde tracers are concordant with our results obtained with retrograde tracers: that only the posterior DNLL is innervated by the MSO. Taken together, these observations support the conclusion that the innervation provided by the MSO and VNLL are topographically segregated in the mustache bat DNLL. Comparison with previous studies The principal physiological, neurochemical, and anatomical features of the mustache bat's DNLL are similar to those that have been described in other mammals. The binaural properties of DNLL cells have been studied in a variety of mammals including cat, rat, and in two species of bats (Brugge et al., 1970; Metzner and Radtke-Schuller, 1987; Buckthought et al., 1993; Covey, 1993; Markovitz and pollak, 1994;Yang and pollak 1gg4a),All of these studies have shown that the DNLL dominated by binauand that DNLL ral tuned to high frequencies are almost entirely EI. Immunocytochem~ca~ studies in a variety of species have also shown that the DNLL is dominated by that stain intensely with antibodies directed against either GABA or glutamic acid decarboxy]1984;Thompson et al., 1985; ase (Adams and Moore and Moore, 1987;Roberts and Ribak, 1987;Glendenning and Baker, 1988; Vater et al., 1992a, 1995; Covey, 1993;Winer et a]., 1995).Additionally, the DNLL in the rat (Merchan et al., 1994), cat (Atkin et al., 1970), and mustache bat (Markovitz and Pollak, 1993) possesses a pronounced tonotopic organization. The big brown bat is the only animal for which a tonotopic organization has not been observed in the DNLL (Covey, 1993). appear to be The afferent connections Of the DNLL there are Some among minor differences. Studies in the cat and rat have shown that the DNLL receives projections from the contralateral from the ipsilateral MSO, and strong bilateral projections from the LSO. In addition it also receives inputs from the ipsilateral VNLL and INLL and from the contralateral DNLL via the commissure of Probst (Glendenninget a]., 1981;Kudo, 1981;Shneiderman et al., 1988;Hutson et al., 1991;BaJo et al., 1993;Merchan et al., 1994;Huffman and Covey, 1995;Vater et al., 1995).These are the same connections that were consistently observed with large injections of dextran amines in the present is Another area Of general agreement the types in the DNLL. Previous anatomical studies of the DNLL in rat and cat report that there are various cell types in the DNLL and each type is regionally distributed (Kane and Barone, 1980;Iwahori, 1986;Shneiderman et al., 1988; BaJo et al., 1993).Most investigators group DNLL cells into three major types, fusiform, multipolar, and round O r oval Cells, where each type more or less dominates different regions o f t h e DNLL. These are the Same general cell types that we observed in the mustache bat DNLL. Moreover, we also observed that fusiform cells tend to be slightly more common in the Posterior DNLL and that multipolar cells are slightly more common in the anterior DNLL. As mentioned in the opening section, the functional significance of the three cell types is at the moment unclear. In a recent intracellular study of rat DNLL neurons in brain slices, w u and Kelly (1995)report that all three types of' DNLL cells express similar types of intrinsic membrane properties. We point out that our observations of DNLL cell types are somewhat limited, as a complete description of the neuronal architecture of the DNLL was not the primary goal of this study. Our evaluations of DNLL cell types, for example. were made from material sectioned only in the transverse plane and only from filled cells in the putative 60-kHz contour of the DNLL. But even with these limitations, our observations concerning cell types in the mustache bat DNLL are consistent with those seen in other animals. There are, however, two differences between the projec.tions that we found and those reported in other animals. One is the presence of a small projection from the dorsal cochlear nucleus that we found after injections of retrograde tracer in the posterior DNLL. Whereas inputs from SUBDIVISIONS IN DNLL 589 Projections to Anterior DNLL DNLL d DNLL 3 VNLL -% 4 Cochlear Nucleus Q 7 Cochlea MNTB Projections to Posterior DNLL DNLL / MNTB Fig. 11. Schematic diagram showing projections to the anterior and the posterior divisions of the DNLL. The thickness of the arrows indicates the strength of the projections. The neuronal population in each nucleus is indicated by solid and open circles. Notice that the population of cells in each nucleus that projected to the anterior DNLL (indicated by open circles) was different from the population that projected to the posterior DNLL (indicated by solid circles). VNLL in the top panel is highlighted, and its projection is shown as a dashed line to indicate axons from VNLL projected only to anterior DNLL and not to posterior DNLL. In the lower panel, MSO is highlighted, and its projection is shown as a dashed line to indicate its axons projected only to the posterior DNLL and not to the anterior DNLL. Abbreviations are the same as in Figure 5. 590 the anteroventral cochlear nucleus have been reported in the cat (Glendenning et al., 1981; Shneiderman et al., 19881, and were seen in this study as well, inputs from the dorsal cochlear nucleus have been questioned. Of interest in this regard are observations by Shneiderman et al. (1988). They never found terminals in DNLL following injections of anterograde tracers in dorsal cochlear nucleus. They suggest that the labeled cells in dorsal cochlear nucleus reported by others (Glendenning et al., 1981) may have been a consequence of HRP uptake in cut fibers traveling through the DNLL. We cannot rule out the possibility that cut fibers contributed to the labeled dorsal cochlear nucleus cells seen in the present study, but we also cannot rule out the possibility that the projection is real and simply represents a small species-specificdifference. The other difference concerns the presence of descending projections from the ICc that have been reported in studies of rat and cat (Carey and Webster, 1971; Hutson et al., 1991; Caicedo and Herbert, 1993). We found very few labeled cells in the ICc even after large injections of BD in the DNLL. These data suggest that the mustache bat DNLL does not receive significant descending inputs from the ICc and further suggest that such a descending projection may be a species-specificfeature. The overall view, then, is that although there may be some minor species-specific features, the general features of the mustache bat DNLL are consistent with those seen in the DNLL of other mammals. What is not clear, and is very difficult to evaluate, is whether the DNLLs in other mammals are also divisible into the two major subdivisions that we observed in the mustache bat. One reason is that there have been only a limited number of neurophysiological studies of the DNLL in other mammals. Most of those studies, however, did not map the response properties in DNLL, nor did they evaluate response properties evoked by S A M signals. The only exception was the study by Covey (1993) of the DNLL in the big brown bat. She identified a region just anterior to the main body of the DNLL that she called the dorsal paralemniscal area. Although most of the cells in the dorsal paralemniscal area were GABA positive, as were the cells in the DNLL, the distinguishing feature of the dorsal paralemniscal area is that it is dominated by monaural neurons. Because the neurons in the anterior DNLL of the mustache bat are all binaural and EI, and because this region receives similar afferent inputs as the posterior DNLL, we conclude that anterior DNLL in the mustache is indeed a division of the DNLL and does not correspond to the dorsal paralemniscal area. Functional implications of the differential afferent connections of the two divisions of the DNLL The results presented here provide a detailed picture of the functional divisions of the mustache bat DNLL and the afferent connections of each division. Below we discuss some functional consequences of the afferent connectional specificity and point out why, for the most part, the results reported here raise more questions about the functional impact of the connectional patterns than they answer. Projections that target only one DNLL division. Given the pronounced differences in response properties of neurons in the anterior and posterior DNLL, it was not surprising to find that there are some differences in the subset of lower nuclei that innervate the two DNLL divisions. One prominent difference in afferent inputs is L. YANG ET AL. that the anterior region receives a substantial projection from the VNLL, whereas the posterior region receives little or no innervation from the VNLL. It would seem reasonable to suppose that the prominent projections from the VNLL to the anterior DNLL play a significant role in shaping its neuronal response properties. VNLL neurons stain intensely for antibodies against conjugated glycine and presumably exert inhibitory influences on their targets (Park et al., 1991; Winer et al., 1995). Additionally, VNLL neurons are strikingly similar to octopus cells in the cochlear nucleus in that they discharge only once at the onset of a tone burst, and each discharge has a remarkably constant latency (Covey and Casseday, 1991). As VNLL cells respond with only one initial discharge, the inhibitory effect of that discharge probably has only a minor influence on the initial phasic discharges of anterior DNLL neurons evoked by tone bursts. Furthermore, the single discharges from VNLL cells probably play no role in suppressing the discharges to the later portions of the tone burst of anterior DNLL neurons. With SAM signals, however, the VNLL could provide repetitive bursts of inhibition to the anterior DNLL that are tightly locked to the phase of each cycle of the SAM signal. If this is the case, then it might well be that the glycinergic inhibition provided by the VNLL could contribute to the shaping of the responses of its DNLL targets to SAM signals having different modulation rates. In other words, it seems reasonable to assume that inhibition from VNLL plays at least some role in limiting the phase-locking to SAM signals of higher modulation rates in the anterior region of the DNLL. The MSO is the other projection that targeted only one DNLL division, in this case the posterior division. We point out that the MSO in the mustache bat is largely monaural, receiving inputs predominately from the contralateral cochlear nucleus and almost no inputs from the ipsilateral cochlear nucleus (Covey et al., 1991; Grothe et al., 1992). Additionally, Grothe (1994) showed that MSO neurons in the mustache bat respond phasically to tone bursts and phase-lock only to SAM signals with low modulation rates. Paradoxically, although we found that the MSO projects selectively to the posterior DNLL, the response properties of MSO neurons are similar to those we observed in the anterior DNLL, rather than to those of the posterior DNLL. Further complicating interpretations of MSO projections are the results from recent studies which indicate that the mustache bat MSO has GABAergic, glycinergic, and excitatory principal cells, all of which may project to the ICc and presumably to the DNLL as well (Vater et al., 1995; Winer et al., 1995). In previous reports we pointed out that the roles of either excitatory or inhibitory MSO neurons in shaping the responses of their DNLL targets is unclear (Yang and Pollak, 1994a-c). Unfortunately, a more detailed view of the projections from MSO to DNLL provides no additional insights into this question. Projections that target both DNLL divisions. The finding that both DNLL divisions receive strong bilateral inputs from the LSO was expected but also had some unexpected features. On the one hand, we anticipated finding prominent inputs from the LSO as they are one of the primary projections found in every previous connectional study of the DNLL, and their binaural, EI properties are consistent with the EI response properties of DNLL neurons. EI properties are first created in the LSO (Boudreau and Tsuchitani, 1968; Caird and Klinke, 1983; Moore and Caspary, 1983; Covey et al., 19911, and our previous studies SUBDIVISIONS IN DNLL suggest strongly that the EI properties in the majority of DNLL neurons are imposed upon DNLL neurons by a well-established excitatory projection from the contralatera1 LSO (Yang and Pollak, 1994a,c). Furthermore, glycinergic inhibitory inputs from the ipsilateral LSO and GABAergc inputs from the other DNLL are consistent with other more complex properties evoked in DNLL neurons by several stimuli that follow each other in close temporal succession (Yang and Pollak, 1994~).These features were seen in DNLL neurons that responded with onset discharges as well as those that responded with sustained discharges to tone bursts. In view of the above results, our expectation was that the axonal projections from an LSO cell would innervate DNLL neurons in both the anterior and posterior divisions because we could see no reason for segregating projections from cell groups in each LSO. We were, therefore, surprised to find that different LSO cells project to the anterior and posterior regions of the DNLL. Indeed, the strict segregation of INLL neurons as well as the segregation of commissural neurons from the opposite DNLL that project to the two DNLL regions were equally surprising and represent one of the principal findings of this study. The prominence of the segregated projections suggests that they must be functionally relevant, although exactly what the functional consequences of such a segregated projection system might be is unclear to us. Functional impact of the two DNLL divisions on their targets in the inferior colliculus Although the pattern of efferent projections will be described in a future report, both divisions of the DNLL target the inferior colliculus bilataterally, and the projections from each division target similar or even overlapping regions in the two colliculi. As mentioned previously, the GABAergic influences of the DNLL play a substantial role in shaping the properties of many EI cells in the ICc (Li and Kelly, 1992; Vater et al., 199213; Faingold et al., 1993; Park and Pollak, 1993; Klug et al., 1995). These DNLL influences are thought to be mainly, although not exclusively, on the contralateral ICc. In view of the differences in the ways neurons in the two DNLL divisions respond to both tone bursts and SAM signals, the inhibitory impact of projections from the anterior division must be considerably different from those of the projections from the posterior division, and this holds for the projections from both the ipsi- and contralateral DNLLs. It would follow, therefore, that the influences of the DNLL on the ICc are more varied and complex than was previously thought. Specifically, the results suggest that each DNLL division may differentially shape not only the binaural properties of their targets in the inferior colliculus, but also their coding of the time variations in complex signals. ACKNOWLEDGMENTS We thank Albert Berrebi for his advice on the use of dextran tracers used in this study and Mr. Carl Resler for designing and implementing the electronics for data acquisition. We also thank Eric Bauer, Michael Burger, Achim Hug, Carl Resler, Albert Berrebi, Brett Schofield, and John Zook for their critical comments on the manuscript. This work is supported by National Institutes of Health grant DC 20068. 591 LITERATURE CITED Adams, J.C., and E. Mugniani (1984) Dorsal nucleus of the lateral lemniscus: a nucleus of GABAergic projection neurons. Brain Res. Bull. 13t585-590. Aitkin, L.M., D.J. Anderson, and J.F. Brugge (1970) Tonotopic organization and discharge characteristics of neurons in the nuclei of the lateral lemniscus. J. Neurophysiol. 33.42 1 4 4 0 . Bajo, V.M., M.A. Merchan, D.E. Lopez, and E M Rouiller (1993) Neuronal morphology and efferent projections of the dorsal nucleus of the lateral lemniscus in the rat. J. Comp. Neurol. 334341-262. Boudreau, J.C., and C. Tsuchitani (1968)Binaural interaction in the cat superior olivary S-segment. J. Neurophysiol. 31:442454. Brugge, J.F., D.J. Anderson, and L.M. Aitkin (1970) Responses of neurons in the dorsal nucleus of the lateral lemniscus to binaural tonal stimuli. J. Neurophysiol. 33r44 1455. Buckthought, A.D., L. Li, and L.B. Kelly (1993) Physiological properties of single neurons in the rat’s dorsal nucleus of the lateral lemniscus. Abstr. Assoc. Res. Otolaryngol. 16-126. Caicedo, A., and H. Herbert (1993) Topography of descending projections from the inferior colliculus to auditory brainstem nuclei in the rat. J. Comp. Neurol. 328.377-392. Caird, D., and R. Klinke ( 1983) Processing of binaural stimuli by cat superior olivary complex neurons. Exp. Brain Res. 52.385-399. Carey, C.L., and D.B. Webster (1971) Ascending and descending projections of the inferior colliculus in the kangaroo rat (Dipodomys rnerriami). Brain Behav. Evol. 4:401-412. Covey, E. (1993) Response properties of single units in the dorsal nucleus of the lateral lemniscus and paralemniscal zone of an echolocating bat. J. Neurophysiol. 69.842-859. Covey, E., and J.H. Casseday (1991) The monaural nuclei of the lateral lemniscus in an echolocating bat: parallel pathways for analyzing temporal features of sound. J. Neurosci. 11:345&3470. Covey, E., M. Vater, and J.H. Casseday (1991) Binaural properties of single units in the superior olivary complex of the mustached bat. J. Neurophysiol. 66: 1080-1093. Faingold, C.L., G. Gehlbach, and D.M. Caspary (1959)On the role of GABA as an inhibitory neurotransmitter in inferior colliculus neurons: iontophoretic studies. Brain Res. 500:302-312. Faingold, C.L., C.A. Boersma-Anderson, and M.E. Randall (1993)Stimulation or blockade of dorsal nucleus of the lateral lemniscus alters binaural and tonic inhibition in contralateral inferior colliculus neurons. Hearing Res. 69t98-106. Glendenning, K.K., and B.N. Baker (1988) Neuroanatomical distribution of receptors for three potential inhibitory neurotransmitters in the brainstem auditory nuclei of the cat. J. Comp. Neurol. 275261-285. Glendenning, K.K., J.K. Brunso-Bechtold, G.C. Thompson, and R.B. Masterton (1981) Ascending auditory afferents to the nuclei of the lateral lemniscus. J. Comp. Neurol. 1973573-703. Goldman, L.J., and O.W. Henson, Jr. (1977) Prey recognition and selection by the constant frequency bat, Pteronotus p . parnellii. Behav. Ecol. Sociobiol.2:411419. Griffin, D.R. (1958) Listening in the Dark. New Haven, CT: Yale University Press. Grothe, B. (1994) Interactions of excitation and inhibition in processing of pure tone and amplitude-modulated stimuli in the medial superior olive of the mustached bat. J. Neurophysiol. 71:70&721. Grothe, B., M. Vater, J.H. Casseday, and E. Covey (1992) Monaural interaction of excitation and inhibition in the medial superior olive of the mustache bat: an adaptation for biosonar. Proc. Natl. Acad. Sci. USA 89t5105-5 112. Huffman, R.F., and E. Covey (1995) Origin of ascending projections to the nuclei of the lateral lemniscus in the big brown bat, Eptesicus fuscus. J. Comp. Neurol. 357:532-545. Hutson, K.A., K.K. Glendenning, and R.B. Masterton (1991) Acoustic chiasm: IV. Eight midbrain decussions of the auditory system in the cat. J. Comp. Neurol. 312105-131. Iwahori, N. (1986) A Golgi study on the nucleus of the lateral lemniscus in the mouse. Neurosci. Res. 3:196-212. Kane, E.S., and L.M. Barone (1980) The dorsal nucleus of the lateral lemniscus in the cat: neuronal type and their distributions. J. Comp. Neurol. 192797-826. Klug, A., T.J. Park, and G.D. Pollak (1995) Glycine and GABA influence binaural processing in the inferior colliculus of the mustache bat. J. Neurophysiol. 74: 1701-1713. 592 Kudo, M. i 19811 Projections of the nuclei of the lateral lemniscus in the cat: an autoradiographic study. Brain Res. 221:57-69. LI, L., and J.B. Kelly (1992) Inhibitory influences of the dorsal nucleus of the lateral lemniscus on binaural responses in the rat’s inferior colliculus. J . Neurosci. 124530-4539. Markovitz, N.S., and G.D. Pollak (1993) The dorsal nucleus of the lateral lemniscus in the mustache hat: monaural properties. Hearing Res. 71:s 1-63. Markovitz, N.S., and G.D. Pollak (1994) Binaural processing in the dorsal nucleus of the lateral lemniscus. Hearing Res. 73:121-140. Merchan, M.A., E. Saldan’a, and I. Plaza (19941 Dorsal nucleus of the lateral lemniscus in the rat: concentric organization and tonotopic projections to the inferior colliculus. J . Comp. Neurol. 341:l-20. Metzner, W., and S. Radtke-Schuller (1987) The nuclei of the lateral lemniscus in the rufous horseshoe hat, Rhinolophus rouri. J. Comp. Physiol. ffi0:395-411. Moore, M.J., and D.M. Caspary (19831Strychnine blocks binaural inhibition in lateral superior olivary nucleus. J. Neurosci. 3237-242. Moore, J.K., and R.Y. Moore (1987) Glutamic acid decarboxylase-like immunoreactivity in brainstem auditory nuclei of the rat. J . Comp. Neu rol. 260: 157- 174. Novick. A,. and J.R. Vaisnys (19641Echolocation of flying insects by the bat, Ctzdonycteris parnellzz. Biol. Bull. f27:478-488. Oliver, D.L., and A. Shneiderman (1989) EM autoradiographic study of projections from the dorsal nucleus of the lateral lemniscus: possible source of inhibitory inputs to the inferior colliculus. J . Comp. Neurol. 28ma-4 7 . Park, T.J.. G.D. Pollak, J.A. Winer, and D.T. Larue (1991) Colocalization of inhibitory transmitters and HRP in projections of the mustache hat’s auditory brainstem: projections from the I S 0 to the inferior colliculus. Abstr. Assoc. Res. Otolaryngol. I4:89. Park. T.J., and G.D. Pollak (1993) GABA shapes sensitivity to interaural intensity disparities in the mustache hat inferior colliculus: implications for encoding sound location. J . Neurosci. I3:2050-2067. Pollak. G.D., and J.H. Casseday (1989) The Neural Basis of Echolocation in Bats. Berlin: Springer-Verlag. Roberts, R.C., and C.E. Ribak (1987) GABAergic neurons and axon terminals in the brainstem auditory nuclei of the gerbil. J . Comp. Neurol. 25S267-280. Ross, I,.S..and G.D. Pollak (1989) Differential projections to aural regions in the 60 kHz isofrequency contour of the mustache hat’s inferior colliculus. J. Neurosci. 9:2819-2834. Ross. L.S..G.D. Pollak, and J.M. Zook (1988)Origin of ascending projections to an isofrequency region of the mustache hat’s inferior colliculus. J. Comp. Neurol. 270:488-505. L. YANG ET AL. Schuller, G., S. Radtke-Schuller, and M. Betz (1986) A stereotaxic method for small animals using experimentally determined reference profiles. J . Neurosci. Methods 18:339-350. Shneiderman, A,, D.L. Oliver, and C.K. Henkel (1988)The connections of the dorsal nucleus of the lateral lemniscus. An inhibitory parallel pathway in the ascending auditory system? J . Comp. Neurol. 276:188208. Thompson, G.C., A.M. Cortez, and D.M.K. Lam (19851Localization ofGABA immunoreactivity in the auditory brainstem of guinea pigs. Brain Res. 339: 119-122. Vater, M. (1995) Ultrastructural and immunocytochemical observations on the superior olivary complex of the mustache hat. J . Comp. Neurol. &58:155-180. Vater, M., M. Kossl, and A.K.E. Horn (1992al GAD- and GABA-immunoreactivity in the ascending auditory pathway of horseshoe and mustache bats. J. Comp. Neurol. 32183-206. Vater, M., H. Hahbicht, M. Kossl, and B. Grothe (1992b)The functional role of GABA and glycine in monaural and binaural processing in the inferior colliculus of hnrseshoe bats. J . Comp. Physiol. [A]171:541-553. Vater, M., J . H . Casseday, and E. Covey (1995) Convergence and divergence of ascending binaural and monaural pathways from the superior olives of the mustache hat. J . Comp. Neurol. 351:632-646. Winer, J.A., D.T. Larue, and G.D. Pollak (1995) GABA and glycine in the central auditory system of the mustache bat: structural substrates for inhibitory neuronal organization. J. Comp. Neurol. 355:317-353. Wu, S.H., and J.B. Kelly i19951 In vitro brain slice studies o f t h e rat’s dorsal nucleus of the lateral lemniscus. 11. Physiologxal properties of biocytinlabeled neurons. J . Neurophysiol. 73:794-809. Yang, L., and G.D. Pollak (1994a) The roles of GABAergic and glycinergic inhibition on binaural processing in the dorsal nucleus of the lateral leminscus of the mustache bat. J . Neurophysiol. 71:1999-2013. Yang, L., and G.D. Pollak (1994133GABA and glycine have different effects on monaural response properties in the dorsal nucleus of the lateral lemniscus of the mustache hat. J . Neurophysiol. 71:2014-2024. Yang, L., and G.D. Pollak ( 1 9 9 4 ~Binaural ) inhibition in the dorsal nucleus of the lateral lemniscus of the mustache hat affects responses for multiple sounds. Auditory Neurosci. f:1-17. Zook, J.M., and J.H. Casseday (1982) Origin of ascending projections to inferior colliculus in the mustache bat, Pteronotus parrzeiliz. J . Comp. Neurol. 207:14-28. Zook, J.M., and J . H . Casseday (1985)Projections from the cochlear nuclei in the mustache hat, Pteronotusparnellii. J . Comp. Neurol. 237:307-324. Zook, J.M., and J . H . Casseday (1987) Convergence of ascending pathways a t the inferior colliculus of the mustache bat, Pteronotusparneilzi. J. Comp. Neurol. 2 6 ~ 3 4 7 - 3 6 1 .