Neurotransmitter phenotype and axonal projection patterns of VIP-expressing neurons in the inferior colliculus

Abstract

Neurons in the inferior colliculus (IC), the midbrain hub of the central auditory pathway, send ascending and descending projections to other auditory brain regions, as well as projections to other sensory and non-sensory brain regions. However, the axonal projection patterns of individual classes of IC neurons remain largely unknown. Vasoactive intestinal polypeptide (VIP) is a neuropeptide expressed by subsets of neurons in many brain regions. We recently identified a class of IC stellate neurons that we called VIP neurons because they are labeled by tdTomato (tdT) expression in VIP-IRES-Cre × Ai14 mice. Here, using fluorescence in situ hybridization, we found that tdT+ neurons in VIP-IRES-Cre × Ai14 mice express Vglut2, a marker of glutamatergic neurons, and VIP, suggesting that VIP neurons use both glutamatergic and VIPergic signaling to influence their postsynaptic targets. Next, using viral transfections with a Cre-dependent eGFP construct, we labeled the axonal projections of VIP neurons. As a group, VIP neurons project intrinsically, within the ipsilateral and contralateral IC, and extrinsically to all the major targets of the IC. Within the auditory system, VIP neurons sent axons and formed axonal boutons in higher centers, including the medial geniculate nucleus and the nucleus of the brachium of the IC. Less dense projections terminated in lower centers, including the nuclei of the lateral lemniscus, superior olivary complex, and dorsal cochlear nucleus. VIP neurons also project to several non-auditory brain regions, including the superior colliculus, periaqueductal gray, and cuneiform nucleus. The diversity of VIP projections compared to the homogeneity of VIP neuron intrinsic properties suggests that VIP neurons play a conserved role at the microcircuit level, likely involving neuromodulation through glutamatergic and VIPergic signaling, but support diverse functions at the systems level through their participation in different projection pathways.

Graphical abstract

A population of stellate neurons in the inferior colliculus express glutamate and VIP. These neurons project to a wide range of targets (shaded green) in the brainstem and thalamus, suggesting a role in a diverse range of behaviors.

1. Introduction

The inferior colliculus (IC) is a major hub of the central auditory system. The IC receives ascending and descending auditory inputs (reviewed by Winer and Schreiner, 2005), neuromodulatory inputs (Schofield and Hurley, 2018), and inputs from non-auditory centers such as the dorsal column nuclei (Aitkin et al., 1981; Robards, 1979), globus pallidus (Yasui et al., 1990), retina (Itaya and Van Hoesen, 1982), and various nuclei of the reticular formation (Andrezik et al., 1981; Kamiya et al., 1988; Herbert et al., 1997). Neurons in the IC integrate these inputs, then send ascending and descending projections to higher and lower auditory centers (Faye-Lund, 1986; Caicedo and Herbert, 1993; Kudo and Niimi, 1980) and to non-auditory centers such as the superior colliculus (Appell and Behan, 1990; Mellott et al., 2018), the periaqueductal gray (Dujardin and Jürgens, 2005; Xiong et al., 2015) and the subparafascicular nucleus (Yasui et al., 1990). The multiplicity of inputs and outputs has complicated efforts to identify which inputs are integrated at the cellular level in the IC and which output pathways receive the results of that integration.

Identification of specific classes of neurons can greatly facilitate the characterization of circuits. This goal has been elusive in the IC. Morphologically, IC cells can be classified as disc- or stellate-shaped (Oliver and Morest, 1984). They can also be classified based on neurotransmitter phenotype as GABAergic or glutamatergic, although there is no clear correlation with the disc/stellate classification (Merchán et al., 2005; Oliver et al., 1994). In vitro, IC neurons can be classified into subtypes based on intrinsic physiology, however the six physiological groups do not correlate with either somatodendritic morphology or neurotransmitter phenotype (Ono et al., 2005; Peruzzi et al., 2000). In vivo responses to sound are similarly diverse, and do not reveal straightforward correlations with other cellular properties (Ono et al., 2017; Palmer et al., 2013; Wallace et al., 2021). Moreover, there can be ambiguity about whether a specific property, such as tuning for sound frequency, indicates separate classes or occurs along a continuum (e.g., Palmer et al., 2013).

In many brain areas, the identification of cell types and the characterization of circuits has advanced considerably with the identification of molecular markers that correlate with other neuronal properties (presumably recognizing functional classes) and that also facilitate experimental manipulation of those neurons. In an earlier study, we found that the neurons labeled in vasoactive intestinal peptide (VIP)-IRES-Cre × Ai14 mice constituted a distinct subtype of IC neuron, which we called VIP neurons. VIP neurons form a class of stellate cells that receive inputs from the contralateral IC and the dorsal cochlear nucleus (Goyer et al., 2019). They represent about 20% of ICc stellate cells, and they participate in most major output pathways of the IC. They were not immunoreactive for markers of GABAergic cells, implying they are glutamatergic. Here, we use RNAscope in situ hybridization to 1) validate the mouse model and confirm that VIP neurons express VIP mRNA, and 2) demonstrate that VIP neurons express Vglut2 and thus are glutamatergic. Additionally, we use Cre-driven fluorescent protein expression to go into further detail describing the unique projection patterns of the VIP subtype of IC neurons. The results reveal a surprisingly wide range of projections for this cell type, suggesting that VIP neurons in the IC play a role in many of the functions attributed to the IC, such as sound localization, speech processing, mediation of defensive behaviors, and modulation of incoming auditory information.

2. Methods

2.1. Animals

All experiments were approved by the University of Michigan Institutional Animal Care and Use Committee and were in accordance with NIH guidelines for the care and use of laboratory animals. Animals were kept on a 12-hour day/night cycle with ad libitum access to food and water. VIP-IRES-Cre mice (Vip^tm1(cre)Zjh/J, Jackson Laboratory, stock #010908; Taniguchi et al., 2011) were crossed with Ai14 reporter mice (B6.Cg-Gt(ROSA)26Sor^{tm14(CAG-tdTomato)Hze}/J, Jackson Laboratory, Stock # 007914; Madisen et al., 2010) to yield F1 offspring that expressed the fluorescent protein tdTomato in VIP neurons (VIP-IRES-Cre × Ai14 mice). Because mice on the C57BL/6J background undergo early onset age-related hearing loss, experiments were restricted to an age range where hearing loss should be minimal (younger than P70; Zheng et al., 1999). Three mice were used for RNAscope in situ hybridization and eight mice underwent intracranial AAV injection to examine the projections of VIP+ cells in the IC.

2.2. RNAscope in situ hybridization

To validate the VIP-IRES-Cre × Ai14 mouse model we performed fluorescent in situ hybridization for tdTomato, VIP and Vglut2 (Slc17a6) using the RNAscope fluorescent multiplex detection kit (Advanced Cell Diagnostics, catalog # 320850) following manufacturer recommendations (Wang et al., 2012). In brief, two male mice (P58) and one female mouse (P45) were anesthetized using isoflurane and decapitated. Brains were quickly dissected, immediately frozen on dry ice (fresh frozen) and kept at −80 °C until the day of slicing. Before slicing, brains were equilibrated at −20 °C, and then 15 μm coronal sections were collected on a cryostat at −20 °C, and each section was mounted on its own Superfrost Plus slide (Fisher Scientific, catalog # 22037246). For each mouse, five non-adjacent sections distributed over the rostral-caudal axis of the midbrain were then randomly selected for in situ hybridization. This resulted in three IC-containing sections each for two mice and five IC-containing sections for the third mouse. The selected sections were fixed using 10% neutral buffered formalin (Sigma-Aldrich, catalog # HT501128) and dehydrated in increasing concentrations of ethanol. After slides were dried, a hydrophobic barrier was drawn around the sections. To block endogenous peroxidase activity, slides were incubated in hydrogen peroxide for 10 min at room temperature followed by application of Protease IV for 30 min. For hybridization, probes targeted to tdTomato, VIP and Vglut2, as well as positive and negative controls were incubated with the slices for 2 hours at 40°C. Next, amplification of probes (AMP 1– 3) was performed by incubating slices for 30 min at 40°C for each amplification step. Signal was developed using the HRP appropriate for each channel. Opal dyes (1:1000) were assigned for each channel: Vglut2 expression was identified by Opal 520 (Akoya Bioscience, catalog # FP1487001KT), VIP expression was identified by Opal 570 (Akoya Bioscience, catalog # FP1488001KT) and tdTomato expression was identified by Opal 690 (Akoya Bioscience, catalog # FP1497001KT). Slices were counterstained with DAPI and coverslipped using ProLong Gold antifade mountant (Fisher Scientific, catalog # P36934). Within two weeks after the assay, representative sections (caudal, mid-caudal-rostral and rostral, 3 – 5 slices per mouse) were imaged at 2 μm depth intervals with a 0.75 NA 20X objective on a Leica TCS SP8 laser scanning confocal microscope (Leica Microsystems). Emission wavelengths were adjusted as previously described (Chan et al., 2018).

Quantification of labeled cells was performed manually using Neurolucida 360 (MBF Bioscience). For each of the IC sections selected for in situ hybridization, the left or right side of the IC was randomly selected for quantitative analysis. Labeled cells were counted by placing a marker on the top of each cell containing puncta indicative of the mRNA of interest (i.e., one marker each for tdTomato, VIP, and Vglut2). To avoid bias in the analysis, fluorescent channels for each mRNA were assessed separately. Additional markers were then added to indicate cells that were positive for specific combinations of markers. All labeled cells on the selected side (left or right) of the IC were counted. To ensure that mRNA puncta were labeling a cell body, analyses were performed only when there was co-labeling with DAPI.

2.3. Intracranial AAV injection

Intracranial virus injections were performed on eight VIP-IRES-Cre × Ai14 mice of either gender aged P21 – P35 using standard aseptic techniques. Throughout the procedure, mice were anesthetized with isoflurane and their body temperature was maintained with a homeothermic heating pad. An injection of the analgesic carprofen (5 mg/kg, CarproJect, Henry Schein Animal Health) was delivered subcutaneously. The scalp was shaved and a rostro-caudal incision was made along the midline to expose the skull. A single craniotomy was created using a micromotor drill (K.1050, Foredom Electric Co.) with a 0.5 mm burr (Fine Science Tools).

Viral construct was injected with a Nanoject III nanoliter injector (Drummond Scientific Company) connected to an MP-285 micromanipulator (Sutter Instruments). Glass injection pipettes were prepared by pulling capillary glass (Drummond Scientific Company) with a P-1000 microelectrode puller (Sutter Instruments). The injector tip was cut to an opening of ~20 μm and beveled at 30° with a BV-10 pipette beveller (Sutter Instruments). Injectors were back-filled with mineral oil then front-filled with AAV1.CAG.FLEX.eGFP.WPRE.bGH (Addgene #51502-AAV1, University of Pennsylvania Vector Core, Lot# CS0922, 2.465e13 genome copies (GC)/ml; Oh et al., 2014). Injections of 20 nl each were made at one to seven sites within the IC, guided by stereotaxic coordinates. Total viral deposits ranged from 20 nl to 140 nl.

After injections were completed, the scalp was sutured with Ethilon 6-0 (0.7 metric) nylon sutures (Ethicon USA LLC), and the wound was treated with 0.5 – 1 ml 2% lidocaine hydrochloride jelly (Akorn Inc). Once mice were ambulatory, they were returned to the vivarium where they were monitored daily until sutures fell out and the wound was completely healed. Mice were perfused after a survival time of 3–4 weeks.

2.4. Immunohistochemistry

AAV-injected mice were deeply anesthetized and perfused transcardially with 0.1M phosphate-buffered saline (PBS; 0.9% NaCl in 0.01 M phosphate buffer, pH 7.4), for 1 minute and then with a 10% buffered formalin solution (Millipore Sigma, cat # HT501128) for 10 minutes. Brains were collected and post-fixed in the same fixative for 2 hours and cryoprotected overnight at 4° C in 0.1M PBS containing 20% sucrose. Brains were cut into 40 μm sections in the transverse, sagittal, or horizontal plane, and sections were collected in three series.

To identify major IC subdivisions, one series from select cases was stained for GAD67 and GlyT2. Tissue was washed in PBS then permeabilized in 0.3% Triton X-100 in PBS for 30 minutes at room temperature. Non-specific staining was blocked in a solution containing 0.1% Triton X-100 and 10% normal donkey serum (NDS) at room temperature for one hour. Then, tissue was incubated in primary antibodies anti-GAD67 (1:250, MAB5406, Millipore; RRID: AB_2278725) and anti-GlyT2 (1:2500, 272-004, Synaptic Systems; RRID: AB_2619998) in PBS with 0.1% Triton X-100 and 1% NDS overnight at 4° C. The following day, tissue was washed in PBS and incubated in secondary antibodies (Alexa Fluor 488-conjugated donkey anti-mouse to label the anti-GAD67 and Alex Fluor 647-conjugated donkey anti-guinea pig to label the anti-GlyT2; both at 1:100, Molecular Probes, A21202 and A21450, respectively; RRID: AB_141607 and AB_141882, respectively). After rinsing tissue several times in PBS, sections were mounted from a 0.2% gelatin solution onto gelatin-subbed slides and coverslipped with DPX mountant (Sigma).

One series of sections was counterstained using a fluorescent Nissl stain to help identify target structures. Tissue was washed several times in PBS, then in 0.1% Triton X-100 in PBS for ten minutes at room temperature. Neurotrace 640/660 (1:100, Molecular Probes, N21483; RRID: AB_2572212) in PBS was applied for 20 minutes at room temperature. Tissue was then washed in 0.1% Triton X-100 for ten minutes, followed by several washes in PBS. Tissue was mounted from a 0.2% gelatin solution onto gelatin-subbed slides and coverslipped with DPX mountant (Sigma).

2.5. Data analysis, imaging, and figure preparation

Quantification of injection sites in three cases and mapping of axons was completed with Neurolucida software (MBF Bioscience) and a Zeiss AxioImager.Z1 or Zeiss AxioImager.Z2 fluorescence microscope. Neurolucida Explorer was used to export data for quantitative analysis using Microsoft Excel and preparation of figures with Adobe Illustrator.

Images were taken using a Neurolucida system and a Zeiss AxioImager.Z2 fluorescence microscope (Carl Zeiss, Inc.) equipped with a Hamamatsu Orca Flash 4.0 camera. Low-magnification images of the entire IC were collected using a 5X objective in a single focal plane. High-magnification images of VIP+ cells at the injection site and of axons in target structures were captured as z-stacks using a 1.40 NA 63X oil-immersion objective and an Apotome 2 (Carl Zeiss, Inc.) to provide optical sectioning at 0.5 μm intervals. Presented images are maximum intensity projections of image stacks. Reported measurements of axon diameter were collected by loading image stacks into Neurolucida 360 software (MBF Bioscience) and reconstructing VIP axon fragments. Adobe Photoshop (CC 2022) was used to colorize images and adjust levels globally when needed. Final figures were assembled with Photoshop or Adobe Illustrator (CC 2022).

3. Results

3.1. IC VIP cells express VIP and VGLUT2

3.1.1. VIP-expressing neurons represent 6.9% of IC glutamatergic neurons

We previously found that VIP-IRES-Cre × Ai14 mice expressed the red-fluorescent protein tdTomato (tdT) in a distinct population of IC neurons that we called VIP neurons (Goyer et al., 2019). However, we did not previously determine whether VIP neurons express VIP. To test this possibility, we performed fluorescence in situ hybridization in IC sections from VIP-IRES-Cre × Ai14 mice with probes targeted to VIP and tdT mRNA. Using brain sections from two male mice (P58) and one female mouse (P47), we counted tdT+ and VIP+ neurons in representative coronal sections spaced along the rostral-caudal axis of the IC (n = 3 – 5 slices per mouse. Counting was performed manually using Neurolucida 360, as detailed in the Methods. We found that 95.2% of tdT+ neurons expressed VIP (n = 453 tdT+ VIP+ neurons compared to n = 23 tdT+ VIP− neurons, Table 1, Fig. 1). This indicates that the VIP-IRES-Cre × Ai14 mouse line is highly selective for labeling IC neurons that express VIP mRNA. Interestingly, we also found that 81% of VIP-expressing neurons in the IC did not express tdT (n = 1981 tdT− VIP+ neurons compared to n = 453 tdT+ VIP+ neurons). Thus, while VIP-IRES-Cre × Ai14 mice selectively label VIP-expressing neurons, this mouse line appears to underrepresent the total VIP-expressing population in the IC. This is consistent with past reports about the efficacy of IRES transgenic animals (Song and Palmiter, 2018).

Table 1.

VIP-IRES-Cre × Ai14 mice selectively label VIP-expressing neurons

Mouse #	Slice Location	tdT+ cells	tdT+ VIP+ cells	% tdT+ VIP+ / tdT+ cells	tdT− VIP+ cells	% tdT+ VIP+ / VIP+ cells
Male 1 P58	Rostral	6	6	100	271	2.2
	Rostral	1	1	100	243	0.4
	Rostral	5	4	80	319	1.2
Female P47	Caudal	54	53	98.1	39	57.6
	Caudal	78	77	98.7	28	73.3
	Middle	66	60	90.9	13	82.2
Male 2 P58	Caudal	69	68	98.6	290	19.0
	Caudal	109	107	98.2	367	22.6
	Middle	24	19	79.2	117	14.0
	Middle	34	31	91.2	164	15.9
	Rostral	30	27	90.0	130	17.2
Total		476	453	95.2	1981	18.6

Table 1 presents the numbers of cells that expressed tdT, VIP, or both across a series of coronal IC sections collected from three mice on which an RNAscope assay was performed. Percentages of tdT-expressing cells that also expressed VIP and percentages of VIP-expressing cells that also expressed tdT are also shown.

Figure 1. tdT+ neurons in the IC of VIP-IRES-Cre × Ai14 mice express VIP and Vglut2 mRNA.

A. Confocal image showing the widespread expression of tdT (yellow), VIP (magenta) and Vglut2 (cyan) mRNA in an IC section treated with the RNAscope in situ hybridization procedure. B-E. Higher magnification confocal images of the section in A showing that tdT+ neurons co-labeled with VIP and Vglut2 (white arrows). These images also illustrate the presence of a sizeable subset of VIP-expressing neurons that expressed Vglut2 but not tdT (yellow arrows). Scale bar in E applies to B-E.

Our previous data also showed that VIP neurons in the IC do not express the GABA-synthetic enzyme GAD67 (Goyer et al., 2019). Since IC neurons are either glutamatergic or GABAergic (Merchan et al., 2005; Oliver et al., 1994), this suggested by exclusion that VIP neurons are glutamatergic. Here, to more directly test whether VIP neurons are glutamatergic, we performed fluorescence in situ hybridization in IC sections from VIP-IRES-Cre × Ai14 mice using a probe against Vglut2, a marker for IC glutamatergic neurons (Ito et al., 2011), along with probes against VIP and tdT. We found that VIP-expressing neurons were glutamatergic, regardless of whether they expressed tdT (98.5% of tdT+ neurons co-labeled with Vglut2, n = 469/476, and 99.4% of tdT− VIP+ neurons co-labeled with Vglut2, n = 1969/1981, Table 2). We then quantified all glutamatergic cells in our slices, and we found that VIP-expressing neurons comprised 6.9% of glutamatergic neurons in the IC (n = 2434 VIP-expressing neurons out of 35,022 cells that expressed Vglut2, Table 2).

Table 2.

VIP neurons comprise 6.9% of glutamatergic cells in the IC

Mouse	Slice Location	tdT+ cells	tdT+ Vglut2+ cells	% co-labeled cells	VIP+ tdT− cells	VIP+ tdT− Vglut2+ cells	% co-labeled cells	Total VIP+ cells	Total Vglut2+ cells	% co-labeled cells
Male 1 P58	Rostral	6	6	100	39	37	94.9	45	2221	2.0
	Rostral	1	1	100	13	12	92.3	14	2511	0.6
	Rostral	5	5	100	28	28	100	32	2452	1.3
Female P47	Caudal	54	52	96.3	271	268	98.9	324	2834	11.4
	Caudal	78	76	97.4	319	318	99.7	396	3395	11.7
	Middle	66	66	100	243	241	99.2	303	3868	7.8
Male 2 P58	Caudal	69	66	95.7	290	290	100	358	1923	18.6
	Caudal	109	109	100	367	367	100	474	3771	12.6
	Middle	24	24	100	117	117	100	136	3840	3.5
	Middle	34	34	100	164	163	99.4	195	4109	4.7
	Rostral	30	30	100	130	126	96.9	157	4098	3.8
Total		476	469	98.5	1981	1967	99.3	2434	35,022	6.9

Table 2 presents the numbers of cells that expressed tdT, Vglut2, or both across a series of coronal IC sections collected from three mice on which an RNAscope assay was performed. Percentages of cells that co-labeled for various combinations of markers are also shown (from left to right: tdT+ Vglut2+/tdT+, VIP+ tdT− Vglut2+/ VIP+ tdT-, and total VIP+/ total Vglut2+).

3.1.2. tdT+ and VIP+ tdT− neurons are similarly distributed in the IC

Since our above results showed that VIP-expressing neurons were glutamatergic regardless of whether they expressed tdT, we hypothesized that tdT expression occurs stochastically within the population of VIP-expressing neurons. As a first step toward testing this hypothesis, we compared the distributions of tdT+ and VIP+ tdT− neurons within the IC of VIP-IRES-Cre × Ai14 mice (Fig. 2). First, we evaluated the distribution across the central nucleus (ICc), dorsal cortex (ICd) and lateral cortex (IClc) of the IC. We found that 59.0% of tdT+ and 52.3% of VIP+ tdT− neurons were localized in the ICc (n = 281/476 and 1036/1981, respectively), 27.5% of tdT+ and 33.0% of VIP+ tdT− neurons were in the ICd (n = 131/476 and 653/1981, respectively), and 13.4% of tdT+ and 14.7% of VIP+ tdT− neurons were in the IClc (n = 64/476 and 292/1981, respectively). Therefore, our data suggest tdT+ and VIP+ tdT− neurons were similarly distributed across the major IC subdivisions.

Figure 2. tdT+ and tdT− VIP-expressing neurons are similarly distributed in the IC.

Plot showing the distribution of tdT+ VIP+ neurons (magenta circles) and tdT− VIP+ neurons (cyan triangles) throughout a series of sections spanning the IC from caudal (top row, left) to rostral (bottom row, right). The major IC subdivisions (ICc, ICd, and IClc) are indicated with black lines.

Next, we analyzed the distribution of tdT+ and VIP+ tdT− neurons along the rostro-caudal axis of the IC. We found that both groups of neurons were most prevalent in caudal IC sections (65.1% of tdT+ neurons, n = 310/476, and 62.8% of VIP+ tdT− neurons, n = 1245/1981), moderately prevalent in mid-rostral-caudal IC sections (26.1% of tdT+ neurons, n = 124/476, and 26.5% of VIP+ tdT− neurons, n = 524/1981), and less common in rostral sections (8.8% of tdT+ neurons, n = 42/476, and 10.6% of VIP+ tdT− neurons, n= 210/1981). Since tdT+ and VIP+ tdT− are glutamatergic and are similarly distributed throughout the IC, these results are consistent with the hypothesis that tdT+ neurons represent a random subset of VIP+ neurons in the IC.

3.2. IC VIP cells project to the thalamus and many brainstem centers

3.2.1. AAV injection induced eGFP expression in VIP cells across the IC.

In order to identify targets of IC VIP cells, we injected a virus carrying a Cre-dependent eGFP gene into the IC of VIP-IRES-Cre mice that also expressed tdT in VIP cells throughout the brain. An example of a viral injection site is shown in Fig. 3A. VIP-tdT cells (reporter-mediated expression, magenta) were present throughout the IC, as well as other brain areas, including the cerebral cortex. Cells expressing eGFP (virally-mediated expression, green) were present only within the IC. The vast majority of eGFP-expressing cells at injection sites also expressed tdT, indicating that eGFP expression was appropriately restricted to VIP neurons (Fig. 3B, arrows). In each case, many tdT-expressing cells that did not express eGFP were intermingled with the virally-labeled cells, indicating that many IC VIP axons remained unlabeled (Fig. 3B, arrowhead). The locations of eGFP-expressing cells were plotted in three cases that were used to collect images of VIP axons (Fig. 3C). In each case, eGFP-expressing cells were present through the rostro-caudal extent of the IC in a manner consistent with the distribution of VIP-expressing cells in the IC (Goyer et al., 2019). Table 3 shows the distribution of labeled cells in three cases that showed robust labeling of axons and, together, allowed us to examine projections from VIP neurons in all three major IC subdivisions. Additional cases typically had fewer eGFP-expressing cells and axons but the results were qualitatively similar.

Figure 3. AAV injection induced eGFP expression in VIP cells across the IC.

A. A photomicrograph showing tdT-expressing cells (magenta, reporter-mediated) and eGFP-expressing cells (green, virally-mediated) in the IC of a VIP-IRES-Cre mouse. Many tdT-expressing cells that do not express eGFP are present, however eGFP-expressing cells typically exhibit greater somatic labeling than tdT-labeled cells. D – dorsal; M - medial. Scale = 1 mm. B. High magnification photomicrographs showing eGFP-expression (green, virally-mediated) and tdT-expression (magenta, reporter-mediated) in VIP cells in the IC. Both eGFP-expressing cells also express tdT (arrows), indicating that eGFP expression is restricted to appropriate cells. A tdT-expressing cell that does not express eGFP is also present (arrowhead) indicating that some VIP cells do not express eGFP. Scale = 20 μm. C. Plots showing eGFP-expressing cells in every third section through the IC in three example cases. Green circles indicate individual eGFP-expressing cells. The distribution of eGFP-labeled cells is consistent with the distribution of VIP cells in the IC, and cells within each IC subdivision were labeled in each case. Across cases, our sample of the VIP population in the IC is robust. Sections are arranged caudal to rostral. D – dorsal; M - medial.

Table 3.

Injection Summary

	M 4–4 3RL		M 4–16 3RL		M 718F
	eGFP cells	% of Total	eGFP cells	% of Total	eGFP cells	% of Total
ICc	238	84%	121	51%	169	58%
ICd	38	13%	116	49%	74	25%
IClc	6	2%	2	1%	48	16%
Total	282		239		291

Table 3 shows the total number of eGFP-expressing cells in every third section of the injection site for three cases that showed robust labeling of axons. The percentage of the total number of cells that were present in each subdivision for each case is also shown. The viral deposits labeled a similar number of cells in these cases, however the relative distribution across IC subdivisions varied from case to case. The combination of cases allowed us to examine projections from VIP neurons in all three major IC subdivisions.

3.2.2. IC VIP axons travel to extrinsic targets via multiple routes

Fig. 4 shows the axons labeled in a case that had virally labeled IC cells spread through much of the rostro-caudal extent of the IC and included substantial portions of the cells in the main IC subdivisions (M718F, Table 3). The densest collection of axons occurred in the injected IC (Fig. 4; asterisks indicate sections that contained labeled cell bodies). The labeled axons extended through the ipsilateral IC, but were densest in the caudal areas, reflecting the distribution of the labeled VIP cells (compare with Fig. 3C, Case M 718F). Beyond the injected IC, labeled axons could be followed in several directions. Some axons collected along the lateral edge of the IC to enter the brachium of the IC. Label in the brachium could be followed rostrally into the thalamus where many axons ended in or near the medial geniculate nucleus (MG). Along the way, labeled axons were visible leaving the medial border of the brachium to enter the intercollicular tegmentum (ICt) and, more rostrally, the deep superior colliculus (SC, Fig. 4, sections 39, 45).

Figure 4. IC VIP axons leave the IC via several routes.

Plots showing VIP axons (green lines) travelling from the IC to targets in the thalamus and the brainstem. Axons can be seen leaving the IC via the brachium (ascending pathways), the lateral lemniscus (descending pathways), and the commissure of the IC (to the contralateral IC). Asterisks indicate sections that contained virally-labeled VIP neurons (not shown). Sections are numbered caudal to rostral. Note that more rostral sections are more widely spaced; distance between sections can be calculated by subtracting the section numbers and multiplying the difference by 120 μm.

Labeled axons also left the IC medially. Here, many axons collected within the commissure of the IC, forming a dense bundle (Fig. 4, section 33). Axons also left the medial IC more ventrally to enter the periaqueductal gray. On the contralateral side, axons from the commissure of the IC spread into the IC and reached all the major subdivisions. Some axons could also be seen in the brachium of the IC on the contralateral side, where the population of fibers could be followed rostrally as far as the MG. A small population of labeled axons was visible in the SC on this contralateral side, though the exact route taken by these axons was not discernible.

Labeled axons also exited the ventral side of the injected IC. These descending projections were substantially less numerous than ascending or commissural axons. Many of the axons collected along the lateral margin of the IC and continued ventrally into the lateral side of the lateral lemniscus (Fig. 4, sections 33, 30). Immediately ventral to the IC, some of these axons appear to travel medially to enter the cuneiform nucleus, which contains a substantial collection of axons just ventral to the IC (CnF, Fig. 4, sections 33, 30). Their transverse orientation suggested that some of these axons continued medially to enter the PAG (e.g., Fig. 4, section 30).

Other labeled axons that entered the lateral lemniscus from the IC maintained a descending trajectory, travelling through the dorsal, intermediate and ventral nuclei of the lateral lemniscus (DNLL, INLL and VNLL, respectively) and also along the medial and lateral margins of those nuclei. Moving ventrally and caudally, the number of axons dwindled substantially; bundles of axons were no longer present but individual axons could be observed in the superior olivary complex (SOC) and, occasionally, in the caudal pontine reticular nucleus (PnC, Fig. 4, sections 33, 30). Within the SOC, labeled axons were most numerous in the ipsilateral ventral nucleus of the trapezoid body (VNTB), where the thicker axons traveled primarily in a rostro-caudal direction. A few axons were present in the trapezoid body and scattered in other nuclei of the ipsilateral SOC and, less often, in the contralateral SOC. Finally, a few axons could be seen ipsilaterally in the cochlear nuclei and in the lateral paragigantocellular nucleus.

3.2.3. IC VIP axons terminate in many targets across the brainstem and thalamus

Figure 5 shows the distribution of labeled boutons in the same case for which axons are illustrated in Figure 4. Because the boutons represent likely sites of transmitter release, they provide a better indication of innervation density and demonstrate likely functional significance for the target area. This is particularly important for targets that are closely associated with dense fiber tracts, such as the nucleus of the brachium of the IC or the nuclei of the lateral lemniscus; each of these areas contained readily identifiable labeled boutons, often associated with thin side branches off parent axons or sometimes with more complex branching arbors. Labeled boutons were spread widely. The most prominent target is within the IC, both ipsilateral and contralateral to the viral deposit. In addition, many boutons were present in the NBIC, ICt, sagulum and SC. Additional boutons were present in the PAG and CnF, in the nuclei of the lateral lemniscus as well as surrounding regions (especially the sagulum, but also the paralemniscal area). Projections to lower brainstem areas were less dense but included the superior olivary complex, the dorsal cochlear nucleus (DCN) and the lateral paragigantocellular nucleus (LPGi). The density of the boutons, the morphology of the axons and boutons, and their association with neuronal cell bodies varied from target to target. The following section provides further details of axon morphology and distribution in the various target areas.

Figure 5. Boutons of IC VIP axons are distributed among many nuclei in the thalamus and brainstem.

Plots showing labeled VIP boutons (magenta) in the same sections illustrated in Figure 4. Asterisks indicate sections that contained virally-labeled VIP neurons (not shown). Sections are numbered caudal to rostral. Note that more rostral sections are more widely spaced; distance between sections can be calculated by subtracting the section numbers and multiplying the difference by 120 μm. Scale = 1 mm.

3.2.3.1. Inferior colliculus

In addition to labeled cells, the injected IC exhibited a dense collection of axons that distributed boutons to each of the IC subdivisions (Fig. 5, sections 21–36). Most of the axons were < 0.8 μm in diameter (i.e., “thin to medium”), but a small minority were notably thicker (“large diameter”, mostly 0.8 – 1.2 μm; rarely larger; Fig. 6). Axonal branching was very common, with both en passant boutons and terminal boutons located in each area (ICc, ICd and IClc; Fig. 6A–F). Comparison with a Nissl counterstain (Fig. 6, yellow) revealed that boutons could be located in the neuropil between cell bodies and in some cases appeared in close apposition to cell bodies (Fig. 6 A–C, insets).

Figure 6. IC VIP axons terminate throughout ipsilateral and contralateral IC.

High magnification photomicrographs showing VIP axons and boutons (green) in the IC ipsilateral (A-F) and contralateral (G-I) to a viral deposit. A-C. Labeled VIP axons are shown in each IC subdivision: central IC (ICc), lateral cortex (IClc) and dorsal cortex (ICd). Insets show selected regions (dashed boxes) at higher magnification to illustrate VIP boutons (green) in close apposition to Nissl-stained somas (yellow arrowheads) or located in the neuropil between somas (white arrows). D-F. Some VIP boutons appeared to contact other VIP cells in the IC. The images show several examples of virally-labeled VIP axons (green) with boutons in apparent contact with tdTomato reporter-labeled neurons (magenta) in the ICc. Insets show apparent contacts (white arrowheads) at higher magnification. G-I. Virally-labeled VIP axons and boutons (green) in the ICc (G), the IClc (H) and the ICd (I) contralateral to a viral deposit. Insets show apparent contacts (yellow arrowhead) between VIP boutons and Nissl-stained neurons (yellow) and apparent contacts (white arrowheads) between VIP boutons and tdT reporter-labeled VIP somas and dendrites (magenta). In each panel, the scale bar = 20 μm for the main image and 10 μm for insets.

Because these experiments were completed in VIP-IRES-Cre × Ai14 animals, presumptive VIP neurons were labeled in the IC, allowing us to assess possible relationships of the virally-labeled VIP axons with other, non-virally-labeled, VIP cells. Fig. 6 D–F shows close appositions (white arrowheads) between virally labeled boutons (green) and other VIP neurons (magenta) within the ICc. Such appositions were associated with both cell bodies and labeled dendrites. These results suggest direct inputs from VIP axons to the somas and dendrites of VIP cells in the ipsilateral ICc.

The contralateral IC also contained labeled axons and boutons across IC subdivisions (Fig. 5, sections 21–36). The axons and boutons were less numerous than those on the ipsilateral side, but otherwise they were similar on the two sides, with frequent branching and numerous en passant and terminal boutons. The case illustrated in Fig. 5 showed a slight bias toward the IC shell areas (with more boutons in the ICd and IClc, compared to the ICc), but comparison across cases showed substantial projections to all three areas. As on the ipsilateral side, the axons ranged in diameter, with thin and medium axons most common. Thicker axons were observed more often in the IClc than in the other subdivisions. Fig. 6H shows virally labeled boutons in the neuropil (Inset 2, white arrows) and also in close association with cell bodies (Inset 2, yellow arrowhead). Virally labeled boutons were also found in apposition to VIP neuronal cell bodies and dendrites (Fig. 6G–I, white arrowheads).

3.2.3.2. Intercollicular tegmentum and IC rostral pole

Labeled axons provided boutons to several areas immediately adjacent to the injected IC, including the periaqueductal gray (PAG), the cuneiform nucleus (CnF) and two regions just rostral to the ICc, the rostral pole (ICrp, often considered a rostral extension of the ICc) and the intercollicular tegmentum (ICt), a multisensory region between the IC and the superior colliculus (SC). The ICt contained a large number of labeled thin to medium diameter axons, often with extensive branching and many boutons (Fig. 7A, B). Boutons were located bilaterally in the ICt but were much more common on the ipsilateral side (Fig. 5, sections 39, 45). The axons were distributed typically in a non-uniform pattern within the ICt, leading to areas with dense innervation and other areas with little innervation. Comparison with the Nissl stain indicated that boutons could be located in the neuropil or be in close apposition with a cell body (Fig.7A, B insets). Labeled axons were also present in the ICrp, but these were much less common than in the surrounding ICt; most were located ipsilateral to the injected IC. The axons could branch, and boutons could be located in the neuropil or in apposition to cell bodies (Fig. 7C, Inset).

Figure 7. IC VIP axons terminate in rostral subdivisions of the IC as well as the periaqueductal gray (PAG) and cuneiform nuclei (CnF).

High magnification photomicrographs showing VIP axons and boutons (green) in regions adjacent to the main IC subdivisions. A-C. VIP axons in the intercollicular tegmentum (A,B) and the rostral pole of the IC (C). Insets show selected regions (dashed boxes) at higher magnification to illustrate VIP boutons (green) in close apposition to Nissl-stained somas (yellow arrowheads) or located in the neuropil between somas (white arrows). D-F. VIP boutons in three zones of the periaqueductal gray, including the dorsal PAG (D), the dorsolateral PAG (E) and the lateral PAG (F). Insets illustrate contacts as in A. G-I. VIP axons and boutons in the cuneiform nucleus. Insets illustrate contacts as in A. In each panel, the scale bar = 20 μm for the main image and 10 μm for insets.

3.2.3.3. Periaqueductal gray

The PAG contained labeled medium and thin axons and, occasionally, a thick axon (Fig. 7D–F). The axons were distributed bilaterally with an ipsilateral dominance. While the PAG extends for greater distances rostrally and caudally, the VIP axons and boutons were largely limited to levels adjacent to the IC and the caudal one-half to two-thirds of the SC (Fig. 5, sections 30–63). Boutons were located in the dorsal, dorsolateral and lateral subdivisions of the PAG bilaterally. Some of the axons traversing the PAG were among the thickest observed in any area (Fig. 7E); these axons had few or no boutons on them. In each of the areas, thin and medium axons provided boutons that were primarily en passant; branching occurred but appeared rarer than in other areas. Fig. 7 D–F shows boutons in the neuropil (white arrows) or in close apposition to Nissl-stained cell bodies (yellow arrowheads).

3.2.3.4. Cuneiform nucleus

Labeled VIP boutons were present in the cuneiform nucleus (CnF), which lies ventral to the IC, medial to the lateral lemniscus and lateral to the PAG (Fig. 5, sections 30–36). Axonal branching occurred but was not prominent. Boutons were present bilaterally with a strong ipsilateral predominance. Boutons were located in the neuropil or in close apposition to Nissl-stained cell bodies (Fig. 7G–I).

3.2.3.5. Superior colliculus

Labeled axons were present in the SC bilaterally with a very strong ipsilateral dominance (Fig. 5, sections 39–69). Axons and boutons were most numerous in the deep layers, less so in the intermediate layers and rare in the superficial layer. Most axons were thin and often exhibited many en passant boutons as well as terminal boutons (Fig. 8 A–C). The boutons were located in the neuropil or in apposition to Nissl-stained cell bodies (Fig. 8A–C insets). Some axons appeared rather isolated, but it was common for an area to have a high density of boutons. Such clusters appeared to include boutons from multiple axons, but we could not rule out particularly extensive arbors of single axons.

Figure 8. IC VIP axons terminate in the superior colliculus and nucleus of the brachium of the IC in the rostral midbrain.

High magnification photomicrographs showing VIP axons and boutons (green) in the superior colliculus (A-C) and in the nucleus of the brachium of the inferior colliculus (D-F). Insets show selected regions (dashed boxes) at higher magnification to illustrate VIP boutons (green) in close apposition to Nissl-stained somas (yellow arrowheads) or located in the neuropil between somas (white arrows). In each panel, the scale bar = 20 μm for the main image and 10 μm for insets.

3.2.3.6. Nucleus of the brachium of the IC

The NBIC is notable for its high density of labeled VIP axons, rivaling any other area except possibly the injected IC (Fig. 5, sections 39–45). To some extent this reflects the presence of labeled axons travelling within the brachium of the IC to reach the thalamus. However, many of the axons possess en passant boutons, and could also exhibit branches that give rise to additional boutons within the NBIC (Fig. 8D–F). Boutons were present bilaterally with an ipsilateral dominance. Boutons could be located in the neuropil or apposed to Nissl-stained cell bodies.

3.2.3.7. Thalamus

The thalamus represents another major target of the VIP projections, with boutons located in a number of nuclei (Fig. 5, sections 63–78). Within the ipsilateral MG, boutons were located in MGv and MGm, less often in the MGd and only rarely in the MGsg. Projections to the contralateral thalamus were much less dense, with most boutons in the MGv. On both sides the boutons could be en passant or terminal and were associated with thin and medium diameter axons (Fig. 9A–D). Some of the larger boutons were particularly prominent because they originated from the thinner axons (e.g., Fig. 9B, D). Another notable feature was clustering of axons, sometimes but not always involving the large boutons. Whether or not associated with the clusters, the boutons were associated with Nissl-stained cell bodies or located within the neuropil.

Figure 9. IC VIP axons terminate broadly in the auditory thalamus.

High magnification photomicrographs showing VIP axons and boutons (green) in various regions of the thalamus. A-D. Examples from the ventral, dorsal and medial subdivisions of the medial geniculate nucleus (MGv, MGd, MGm, respectively). E-I. Examples in other auditory thalamic regions: E. posterior intralaminar nucleus (PIN). F. peripeduncular nucleus (PP). G-I. subparafascicular nucleus. In all panels, virally labeled VIP axons are shown in green. Insets show selected regions (dashed boxes) at higher magnification to illustrate VIP boutons (green) in close apposition to Nissl-stained somas (yellow arrowheads) or located in the neuropil between somas (white arrows). In each panel, the scale bar = 20 μm for the main image and 10 μm for insets.

Additional boutons were located outside the MG proper in other thalamic regions associated with auditory circuits. Ventral to the MG, substantial projections were observed in the posterior intralaminar nucleus (PIN, Fig. 9E) and in the peripeduncular nucleus (PP, Fig. 9F). The subparafascicular nucleus (SPF) also contained labeled boutons in association with cell bodies and the intervening neuropil (Fig. 9G–I). Finally, a few axons and boutons appeared to travel rostrally beyond the MG proper (Fig. 5, section 78).

3.2.3.8. Lateral lemniscus and surrounding areas

Ventral to the IC, thin and medium diameter axons entered the lateral lemniscus (Fig. 4, sections 30–33). The sagulum contained many such axons. The medium axons could give rise to terminal or en passant boutons as well as to thinner side branches (Fig. 10A–C). Both the side branches and thin axons travelling in parallel to the lateral lemniscus gave rise to additional boutons such that the sagulum appears to be a major target of IC VIP projections. The boutons in the sagulum most often appeared in the neuropil, although association with Nissl-stained cell bodies occurred occasionally (Fig. 10A, C, yellow arrowheads).

Figure 10. IC VIP axons terminate in nuclei in and around the lateral lemniscus.

High magnification photomicrographs showing VIP axons and boutons (green) in the sagulum (A-C), in the dorsal nucleus of the lateral lemniscus (D-F), in the ventral nucleus of the lateral lemniscus (G, H) and in the ventrolateral tegmental nucleus (I). In all panels, virally labeled VIP axons are shown in green. Insets show selected regions (dashed boxes) at higher magnification to illustrate VIP boutons (green) in close apposition to Nissl-stained somas (yellow arrowheads) or located in the neuropil between somas (white arrows). In each panel, the scale bar = 20 μm for the main image and 10 μm for insets.

Labeled boutons were present in dorsal and ventral nuclei of the lateral lemniscus but rare in the intermediate nucleus (INLL). In the DNLL, axonal branching was quite prominent, with thin axons providing a robust network of boutons (Fig. 10D–F). These boutons were located in the neuropil or closely associated with Nissl-stained cell bodies (Fig. 10D–F insets). In the VNLL, labeled boutons were less numerous but there were clear examples of medium axons aligned with the lateral lemniscus giving rise to daughter branches, at ninety degree branch angles, that supplied boutons to the VNLL (Fig. 10G). Thin axons in the lemniscus could also give rise to branches and boutons (Fig. 10H).

In addition to the boutons within the nuclei of the lateral lemniscus there were labeled boutons in the medially adjacent reticular formation. This region has been referred to as the paralemniscal area (PL) in other species and appears to be a target of IC VIP axons (not shown). Another nucleus of the reticular formation is the ventrolateral tegmental nucleus (VLTg), which has also been associated with auditory circuits. Labeled boutons were present in the VLTg, often arising from short branches off a parent axon (Fig. 10I).

3.2.3.9. Superior olivary complex

Labeled boutons in the SOC were most numerous in the ipsilateral VNTB. Rostro-caudally oriented parent branches gave rise to thinner axon branches with en passant and terminal boutons (Fig. 11A–C). The boutons could be clustered among the cell groups of the VNTB (i.e., between the fascicles of the trapezoid body) and could appear in the neuropil or closely adjacent to Nissl-stained cell bodies (Fig. 11A–C insets).

Figure 11. IC VIP axons terminate in lower auditory centers, including the superior olivary complex, cochlear nucleus and lateral paragigantocellular nucleus.

A-F. High magnification photomicrographs showing VIP axons and boutons (green) in the various nuclei of the superior olivary complex. A-C. the ventral nucleus of the trapezoid body. D,E. the superior paraolivary nucleus. F. the medial nucleus of the trapezoid body. G,H. VIP axons and boutons in the dorsal cochlear nucleus. I. VIP axons and boutons in the lateral paragigantocellular nucleus. In all panels, virally labeled VIP axons are shown in green. Insets show selected regions (dashed boxes) at higher magnification to illustrate VIP boutons (green) in close apposition to Nissl-stained somas (yellow arrowheads) or located in the neuropil between somas (white arrows). In each panel, the scale bar = 20 μm for the main image and 10 μm for insets.

There were occasional axons with boutons in other nuclei of the ipsilateral superior olivary complex, including the superior paraolivary nucleus (SPN; Fig. 11D, E) and less often the medial nucleus of the trapezoid body (MNTB; Fig. 11F). Labeled boutons in the contralateral SOC were rare. We also noted a small ipsilateral projection to the caudal pontine reticular nucleus (PnC), just dorsal to the rostral half of SOC (e.g., Fig. 5, sections 30–36). The PnC contains neurons involved in the startle reflex, but it was not possible to determine if the labeled VIP boutons were associated with cells related to startle or to other functions of the PnC.

3.2.3.10. Cochlear nucleus

The cochlear nucleus typically contained almost exclusively thin axons in relatively small fragments, suggesting limited arborizations. Labeled boutons were usually small and almost all were located in the DCN, where they were largely limited to the deep layer and fusiform cell layer (Fig. 11G–H).

3.2.3.11. Lateral paragigantocellular nucleus

Labeled boutons were also present in the LPGi, a small nucleus of the reticular formation located caudal to the superior olivary complex and lateral to the medullary pyramids. While little is known of the function of the LPGi, it has connections with numerous auditory nuclei (e.g., SOC, IC, auditory cortex and cochlear nucleus; Yasui et al., 1990; Beebe, Zhang et al., 2021) suggesting a role in ascending and descending auditory circuitry. The present study revealed labeled boutons in the LPGi ipsilateral to the injected IC (Fig. 11I, J).

4. Discussion

In this study, we have shown that Cre-expressing neurons in the IC of VIP-IRES-Cre mice express VIP mRNA and are glutamatergic. We have also shown that IC VIP neurons have broad projections, with targets including higher and lower auditory nuclei, brainstem areas involved in a variety of behaviors and regions containing dopaminergic and cholinergic neurons. Figure 12 summarizes these projections. Innervation by IC VIP neurons is lightest in lower brainstem nuclei, such as the CN and the SOC, and heaviest within the ipsilateral IC and sagulum and in ascending pathways to the ipsilateral NBIC and the ipsilateral MG.

Figure 12. Schematic summary showing the widespread projections of IC VIP neurons.

A series of coronal sections are shown to depict the major components of the auditory pathway and other areas targeted by projections from IC VIP cells. The image depicts the relative density of projections from VIP neurons in the IC on the left side of the figure, allowing comparison of ipsilateral versus contralateral projections (darker shading indicates more axons and boutons in the target area). Overall, the projections from VIP neurons, known to be glutamatergic stellate cells, appear to contribute to most or all of the known projections from the IC, suggesting that these cells participate in a wide range of auditory analyses and behavior.

4.1. Technical issues.

Our viral injections led to eGFP expression limited to cells that also expressed tdTomato in the reporter animals, indicating that virally labeled axons originated from VIP cells. It is regularly observed that Cre-dependent expression is less than 100% efficient, meaning that some Cre-containing cells do not express fluorescent protein so the results necessarily reveal only a portion of the projections from the population of neurons being studied. In the present experiments, our FISH data indicate that Cre is expressed in a minority of VIP neurons, suggesting that our viral labeling under-represents the projections of the full VIP+ population. Nonetheless, the virally labeled axons spread extensively to innervate the auditory thalamus and many brainstem centers. These results indicate a wide range of influence for IC VIP neurons, presumably supporting a wide range of functions.

4.2. Functional implications: cellular.

4.2.1. VIP IC cells are glutamatergic and likely VIPergic

We found that 95.2% of tdT-expressing cells in the IC of VIP-IRES-Cre × Ai14 mice express VIP mRNA, indicating that the neurons we previously called “VIP neurons” based on their labeling in these mice do in fact express VIP (Goyer et al., 2019). The expression of VIP mRNA in these cells makes it likely that IC VIP neurons use VIP signaling to modulate activity in the brain regions they target. Consistent with this, Prönneke and colleagues showed that expression of VIP mRNA was tightly correlated with expression of VIP peptide in cortical VIP neurons (2015). Furthermore, VIP receptors are highly expressed in several of the brain regions VIP neurons target, including MG, PAG, and SOC, and superior colliculus (Joo et al., 2004). Since VIP signaling enhances neuronal excitability in other regions of thalamus that have similarly dense expression of VIP receptors (Lee and Cox, 2003, Sun et al., 2003), it is likely that VIP release from IC VIP neurons would have excitatory effects in the MG. Effects of VIP signaling in the MG and the other brain regions targeted by IC VIP neurons are an important focus for future studies.

In addition, our in situ hybridization experiments showed that 99.3% of VIP+ cells and 98.5% of tdT+ cells co-labeled with Vglut2. Combined with our previous results showing that tdT-expressing cells did not immunolabel for the GABAergic marker GAD67 (Goyer et al., 2019), this provides high confidence that IC VIP neurons are glutamatergic. Our confirmation that IC VIP neurons are glutamatergic is an interesting contrast to forebrain regions like neocortex, hippocampus, and suprachiasmatic nucleus, where VIP-expressing cells are GABAergic (e.g., Klausberger and Somogyi, 2008, Fan et al., 2015, Tremblay et al., 2016). Notably, however, VIP-expressing cells in the dorsal raphe nucleus and periaqueductal gray, brain regions that neighbor the IC, also express Vglut2 (Zhao et al., 2022). Thus, VIP-expressing cells in the midbrain appear to differ from those in the forebrain by being glutamatergic.

Overall, the present results indicate that VIP neurons likely use a combination of glutamatergic and VIPergic signaling to shape computations in their postsynaptic targets. Since VIP receptors are G-protein coupled receptors (Dickson and Finlayson, 2009), these results suggest that IC VIP neurons may influence their targets over multiple timescales, ranging from fast excitatory neurotransmission at glutamatergic synapses to slower neuromodulation through VIP signaling.

4.2.2. Prevalence of VIP neurons in the IC

Our in situ hybridization data also showed that VIP-expressing cells accounted for one out of every 14–15 glutamatergic cells in the IC (6.9%). This indicates that VIP cells are more prevalent in the IC than we previously determined based on tdT fluorescence (Goyer et al., 2019). Since ~20–27% of IC neurons are GABAergic (Oliver et al., 1994, Merchan et al., 2005, Beebe et al., 2016), and therefore by extension 73–80% of IC neurons are glutamatergic, our present results suggest that VIP+ cells account for ~5.0–5.5% of IC neurons. Does this percentage represent a functionally relevant proportion of the IC cells? In cerebral cortex, VIP neurons comprise ~2% of the neurons (Rudy et al., 2011). These neurons play demonstrable roles in cortical processing and animal behavior (e.g., Lee et al., 2013; Pfeffer et al. 2013; Cichon et al., 2017; Kuchibhotla et al., 2017). Our results suggest that VIP neurons in the IC are sufficiently numerous to 1) affect IC physiology and auditory behavior, and 2) to be experimentally tractable (e.g., via optogenetic manipulation).

Interestingly, while tdT expression in VIP-IRES-Cre × Ai14 mice was highly selective for VIP-expressing cells (95.2% of tdT+ cells were VIP+), only about one in five VIP-expressing cells expressed tdT. This inefficiency in driving reporter expression is not unexpected for an IRES-Cre transgenic animal (Song and Palmiter, 2018). In the VIP-IRES-Cre mouse, the IRES-Cre transgene was inserted in the 3’ untranslated region of the native VIP gene (Taniguchi et al., 2011). Since we observed that VIP mRNA was generally abundant in VIP+ cells (Fig. 1), we suspect the IRES-Cre construct is relatively inefficient at yielding bicistronic translation from VIP mRNA in the IC. Alternatively, because VIP-IRES-Cre × Ai14 mice are hemizygous for the IRES-Cre transgene, the transgenic VIP allele might be transcribed at a much lower rate than the wild type allele. These scenarios are not mutually exclusive, and either or both would result in inefficient expression of Cre and suboptimal expression of tdT in crosses with Ai14 reporter mice. Based on these putative mechanisms, we suspect that tdT expression stochastically labels VIP+ neurons in VIP-IRES-Cre × Ai14 mice, although we cannot rule out the possibility that VIP+ tdT− neurons somehow differ from VIP+ tdT+ neurons.

Less than 5% of all tdT+ cells did not co-label for VIP. This is not unexpected, as there are reports of low levels of tdT expression in Ai14 mice in the absence of Cre expression (Getz et al., 2022, Ivanova et al., 2021, Jackson Laboratory Website, 2022). In addition, since transient Cre expression can be sufficient to excise the loxP-flanked STOP cassette located upstream of the tdT allele in Ai14 mice, it is possible that the tdT+ VIP− cells we observed expressed VIP at an earlier developmental stage but no longer expressed VIP in adult mice (Heffner et al., 2012). It is also possible that the tdT+ VIP− cells expressed VIP mRNA at levels that were too low to reliably detect with the RNAscope assay. Whatever the cause, the finding that >95% of tdT+ cells were VIP+ indicates that VIP-IRES-Cre × Ai14 mice are a reliable resource for identifying VIP expressing cells in the IC.

4.2.3. VIP neurons as modulators of circuit output

While VIP neurons represent a relatively small population of IC neurons, several factors suggest that they may exert outsize influence on the circuits they participate in. First, several intrinsic physiologic properties suggest that among IC neurons, the VIP neurons would be among the more responsive ones, with a correspondingly stronger output for many auditory stimuli (Goyer et al., 2019). These properties include stellate morphology, which suggests they respond to a broader range of frequencies than, for example, disk cells of the IC, and intrinsic properties that support sustained firing with little or no adaptation.

Additional factors arise from physiological studies suggesting that VIP can have substantial effects on target neurons. Thalamocortical cells exhibit two main firing patterns – bursting and tonic – that are closely related to behavioral state and reflect different modes of information transmission to the cerebral cortex (e.g., Steriade et al., 1993; McCormick and Bal, 1997). VIP activates HCN channels and depolarizes thalamocortical cells, shifting them from bursting to tonic firing mode (Lee and Cox, 2003; Sun et al., 2003). The ability of VIP to switch the firing mode of target cells indicates a substantial effect on thalamocortical physiology. Another factor regarding VIP function relates to its identity as a neuropeptide. Like many neuropeptides, VIP can have long-lasting postsynaptic effects. Sun et al. (2003) described long-lasting depolarization (many minutes in vitro) that the authors suggested might be mediated by a VIP-activated second messenger. Such effects may allow VIPergic/glutamatergic neurons to elicit longer and stronger depolarizations than could be accomplished by glutamatergic neurons that do not release VIP.

A final factor for interpreting VIP function is homogeneity within the IC VIP neuron population. We previously found that VIP neurons had a narrow range of physiological properties, regardless of where they were found in the IC (Goyer et al., 2019). The present study revealed variation in axon diameter of VIP neurons, raising a question of VIP neuron subtypes. While we cannot rule out that possibility, we were unable to find other characteristics that correlate with axon diameter. Instead, it appears likely that variation in axon diameter reflects, at least in part, that thin daughter axons arise from thick parent axons that continue on to more distant targets. For example, thick parent axons in the VNLL give rise to thin branches within the VNLL while the parent axon presumably continues to lower centers such as the superior olivary complex or cochlear nucleus (Fig. 10G). We conclude that the evidence available so far, both anatomical and physiological, is consistent with a homogenous class of VIP IC neurons.

The relative homogeneity of VIP IC neurons suggests that they play similar computational roles in each of the IC circuits in which they participate. The relatively slow membrane time constant of VIP neurons (~15 ms) suggests they integrate synaptic input across an extended time window compared to the average IC neuron (~10 ms) (Goyer et al., 2019). The present results extend our concept of the functional role of VIP neurons by indicating that VIP neurons uniformly express markers for glutamatergic and VIPergic signaling. We therefore propose that VIP neurons might act like “widgets” that can be “plugged into” diverse circuits to support a similar neuromodulatory outcome at the circuit level, regardless of the exact nature of the sensory or other information that is processed by that circuit. In this scheme, different functional roles would be accomplished via projections to different targets rather than by altering the physiological features of VIP neurons. Put another way, the homogeneity of VIP neuron properties at the cellular level suggests a conserved role at the microcircuit level, while the diversity of VIP axon targets suggests an equally diverse set of functions at the systems level.

4.3. Functional implications: circuit level

The IC is often referred to as a hub of the subcortical auditory system, reflecting not only a multitude of inputs but also a wide range of targets for IC outputs. A striking feature of the VIP projection is its participation in most or all of these outputs. This diversity suggests a role in a wide range of functions.

4.3.1. Intrinsic and commissural projections

The IC, ipsilateral and contralateral, was a prime target of the VIP projections, suggesting a substantial role for VIP projections within the IC. Within the ipsilateral IC, most IC neurons have local axon collaterals (Oliver et al., 1991). Functional studies indicate that both excitatory and inhibitory neurons contribute to local IC circuits (Sivaramakrishnan et al., 2013, Sturm et al., 2014) and that intrinsic circuits can reorganize following noise exposure (Sturm et al., 2017). These results suggest that local IC circuits play important roles in sound processing, but an inability to manipulate specific circuit components has made it difficult to determine how specific neuron types shape local circuit operations. With the finding that VIP neurons provide excitatory input to other IC neurons, it will now be possible to use targeted manipulations of VIP neurons to determine how local VIP projections influence coding in other IC neurons. Importantly, our observation that VIP neurons contact other VIP neurons in the ipsilateral IC supports the presence of recurrent excitatory networks in the IC (Sivaramakrishnan and Oliver, 2006; Sivaramakrishnan et al., 2013), which could contribute to the central role of the IC in initiating audiogenic seizures (Faingold, 1999).

IC commissural projections have been implicated in enhancing the discrimination of tones and sound localization cues, which they accomplish by altering the gain of input-output functions in the contralateral IC (Malmierca et al., 2003, 2005; Orton and Rees, 2014; Orton et al., 2016). Additional roles for the IC commissure are likely, as the function of this circuit has rarely been explored. Commissural connections between the two ICc’s are generally thought to be homotopic, connecting mirror-image regions of the left and right ICc (González-Hernández et al., 1986; Saldaña and Merchán, 1992; Malmierca et al., 1995, 2005). However, we recently found that IC NPY neurons, which are GABAergic stellate neurons, make more heterotopic commissural projections than non-NPY neurons (Anair et al., 2022). Since VIP neurons are also stellate neurons, it will be interesting to determine whether VIP commissural projections are more like those of NPY neurons or are homotopic, as expected for the majority of commissural neurons. Furthermore, as described above for local IC circuits, our results provide impetus to use targeted manipulations of VIP neurons to start disentangling how specific sources of commissural input shape computations in the contralateral IC, including how VIP neurons influence computations in contralateral VIP neurons.

4.3.2. Ascending projections

The present results show that projections to higher centers represent the bulk of extrinsic projections from IC VIP neurons. The targets include multiple nuclei in the thalamus as well as the superior colliculus and the NBIC. These ascending projections likely contribute to a variety of auditory functions.

The overall (i.e., VIP and non-VIP) set of projections from the IC to the thalamus have been conceptualized as lemniscal and extra-lemniscal systems that terminate in different thalamic nuclei and thus support different aspects of hearing (see Hu 2003; Lee and Sherman, 2010; Anderson and Linden, 2011; Mellott et al., 2014). We have shown that the VIP projections to the thalamus target both lemniscal (MGv) and extra-lemniscal nuclei (MGd, MGm, PP, PIN). Through these projections, the output of IC VIP neurons is likely to affect information flow to a variety of thalamic targets that provide auditory input to a multitude of auditory cortical areas as well as the amygdala and the striatum. These projections are critical for auditory perception as well as many aspects of the emotional content of sound and the integration of hearing with attention and movement.

IC projections to the SC are critical for auditory-cued attention (Hu and Dan, 2022). IC projections to the PAG and the CnF, along with projections to the SC, have been implicated in orienting and defensive behaviors (Mitchell et al., 1988; Schenberg et al., 2005; Xiong et al., 2015; Assareh et al. 2016; Tovote et al., 2016; Wang et al., 2019; Leffler et al., 2020). The PAG has also been implicated in control of vocalizations (e.g., Kyuhou and Gemba, 1998; Dujardin and Jurgens, 2005). The present results demonstrate projections to each of these targets, suggesting that IC VIP neurons could contribute to these behaviors.

The NBIC is one of the primary targets of IC VIP projections. The NBIC is a large nucleus best known as a source of ascending projections to the thalamus and to the superior colliculus. It receives input from auditory, somatosensory and visual systems and its neurons exhibit multisensory responses (Berkley et al., 1980; Calford and Aitkin, 1983; Kudo et al., 1984; Le Doux et al., 1985; Thiele et al., 1996; Jiang et al., 1997; King et al., 1998). Responses to acoustic stimuli can be sensitive to sound location and are thought to contribute to orienting (and perhaps other) behaviors through its projections to the SC (Aitkin and Jones, 1992; Schnupp and King, 1997; Slee and Young, 2013). The projections to the MG target the extra-lemniscal pathways, consistent with multisensory responses (Calford and Aitkin, 1983; Kudo et al., 1984). In the present results, IC VIP projections terminate densely in the NBIC and could contribute to functions associated with the NBIC projections to the thalamus and the SC.

4.3.3. Descending projections

IC VIP cells also project to lower auditory nuclei, including the NLL, SOC and CN. Projections to the NLL and surrounding regions were surprisingly dense (e.g. Fig. 10). Previous studies have suggested such a projection but technical limitations have prevented firm conclusions that axons labeled in the NLL originate from IC cells and not from cells in other areas that were labeled via retrograde transport from the IC (e.g., Caicedo and Herbert, 1993). Our methods provide clear resolution of this issue. While the viral labeling method can sometimes lead to retrograde labeling, we observed no such labeling in cases with substantial axonal labeling in the NLL. This shows clearly that the axons labeled in the NLL originated from IC neurons. The VIP projections terminated in both the DNLL and VNLL, two large nuclei associated with circuits for binaural and monaural processing, respectively (reviewed in Schofield, 2005). The major projections from DNLL and VNLL are to the IC, so the VIP projections are likely modulating the ascending input to the IC.

Overall projections (not limited to VIP neurons) from the IC to the SOC terminate densely in the VNTB and less so in other nuclei (reviewed by Thompson and Schofield, 2000). The present experiments labeled axons in the SOC and the CN, but these VIP projections appeared much less dense than the overall projections from the IC (cf. Saldaña, 1993; Caicedo and Herbert, 1993; Suthakar and Ryugo, 2017). The SOC projects to a wide range of targets so IC inputs could modulate many circuits; published data indicate that IC axons contact olivary cells that project to the cochlea or the cochlear nucleus (Vetter et al., 1993; Schofield and Cant, 1999; Mulders and Robertson, 2002). The VIP axons labeled in the present study terminated in the VNTB and other periolivary nuclei, so they are in a position to target the same circuits (i.e., projections to the cochlea or cochlear nucleus). However, the VIP axons appeared much less dense than overall IC projections. We know that our methods labeled only a minority of IC VIP neurons, so it is possible that the projection to the SOC (or other targets) are more substantial than we observed. It is also possible that the IC VIP neurons play a minimal role in the IC-SOC projections. A similar argument applies to the IC projections to the cochlear nucleus. The VIP axons labeled in the present study reflect the general pattern of IC-CN projections (reviewed by Saldaña, 1993), but may represent only a small part of this projection. VIP projections to the SOC and the CN provide an opportunity for modulating the earliest stages of monaural and binaural processing (and perhaps cochlear function if VIP axons also contact olivocochlear cells). Further work is needed to identify the hindbrain circuits and cell types contacted by the IC VIP axons.

4.3.4. VIP association with modulatory systems

The axons of IC VIP neurons also terminate in several areas associated with some of the classical modulatory neurotransmitters, including acetylcholine and dopamine. The SPF is a source of dopaminergic projections to the IC and to the SOC (Nevue et al., 2016a, b). We observed VIP axons terminating in the SPF where they are in a position to affect the activity of SPF neurons and their modulatory projections. The IC VIP projections also terminate in lower auditory centers that contain modulatory cells. Both the SOC and the LPGi contain cholinergic cells that modulate other auditory regions in the brainstem and cochlea (Schofield and Hurley, 2018; Schofield and Beebe, 2019). Further experiments will be needed to determine whether IC VIP axons directly contact dopaminergic or cholinergic cells in these different areas.

4.4. Conclusions

The projections from IC VIP neurons arise from glutamatergic stellate cells that presumably excite their targets. While VIP cells represent only a subset of stellate cells, their projections reach a wide range of targets; in fact, the present results show at least a small contribution of VIP cells to all the known targets of IC projections. The extra-lemniscal pathways are a prominent target, as indicated by substantial VIP projections to the sagulum and extra-lemniscal regions of the auditory thalamus (MGd, MGm, PP, PIN). The NBIC and SC and related areas are also targeted, suggesting VIP contributions to orienting and defensive behaviors. Descending VIP projections are less substantial, but are in a position to modulate processing at early levels of the auditory system. Finally, VIP projections target a number of modulatory centers, including medullary locations of cholinergic cells (SOC and LPGi) and thalamic loci of dopaminergic cells (i.e., SPF). As suggested above, the diversity of targets may reflect a diversity of functions while the physiologic homogeneity of the VIP cells suggests a common cellular role in each of these pathways.

Highlights.

• A subset of inferior collicular (IC) stellate neurons expresses glutamate and VIP

• IC VIP neurons project locally to the ipsilateral and contralateral IC

• IC VIP neurons contribute to all IC output pathways

• Diverse projections suggest IC VIP neurons contribute to many behaviors

Abbreviations:

AAV

adeno-associated virus

Aq

cerebral aqueduct

Cb

cerebellum

cic

commissure of the IC

CN

cochlear nucleus

CnF

cuneiform nucleus

cp

cerebral peduncle

Cre

Cre-recombinase

DCN

dorsal cochlear nucleus

DNLL

dorsal nucleus of lateral lemniscus

eGFP

enhanced green fluorescent protein

fr

fasciculus retroflexus

IC

inferior colliculus

ICc, ICd, IClc

central nucleus, dorsal cortex, lateral cortex of the IC

ICrp

rostral pole of inferior colliculus

ICt

intercollicular tegmentum

INLL

intermediate nucleus of lateral lemniscus

LG

lateral geniculate nucleus

LP

lateral posterior nucleus of thalamus

LPGi

lateral paragigantocellular nucleus

LSO

lateral superior olivary nucleus

MG

medial geniculate nucleus

MGd, MGm, MGsg, MGv

dorsal, medial, suprageniculate and ventral subdivisions of the MG

ml

medial lemniscus

MNTB

medial nucleus of the trapezoid body

NBIC

nucleus of the brachium of the IC

NDS

normal donkey serum

PAG

periaqueductal gray

PBS

phosphate buffered saline

PIN

posterior intralaminar nucleus (of the thalamus)

PL

paralemniscal area

PP

peripeduncular nucleus

PnC

caudal pontine reticular nucleus

py

pyramid

Sag

sagulum

SC

superior colliculus

scp

superior cerebellar peduncle

SCs, SCi, SCd

superficial, intermediate and deep layers of SC

SN

substantia nigra

SOC

superior olivary complex

sp5

spinal trigeminal tract

SPF

subparafascicular nucleus

SPN

superior paraolivary nucleus

tdT

tdTomato

VCN

ventral cochlear nucleus

VGLUT2

vesicular glutamate transporter 2

VIP

vasoactive intestinal polypeptide

VLTg

ventrolateral tegmental nucleus

VNLL

ventral nucleus of lateral lemniscus

VNTB

ventral nucleus of the trapezoid body