Characterization of transgenic mouse lines for labeling type I and type II afferent neurons in the cochlea

The cochlea is innervated by type I and type II afferent neurons. Type I afferents are myelinated, larger diameter neurons that send a single dendrite to contact a single inner hair cell, whereas unmyelinated type II afferents are fewer in number and receive input from many outer hair cells. This strikingly differentiated innervation pattern strongly suggests specialized functions. Those functions could be investigated with specific genetic markers that enable labeling and manipulating each afferent class without significantly affecting the other. Here three mouse models were characterized and tested for specific labeling of either type I or type II cochlear afferents. Nos1CreER mice showed selective labeling of type I afferent fibers, Slc6a4-GFP mice labeled type II fibers with a slight preference for the apical cochlea, and Drd2-Cre mice selectively labeled type II afferent neurons nearer the cochlear base. In conjunction with the Th2A-CreER and CGRPα-EGFP lines described previously for labeling type II fibers, the mouse lines reported here comprise a promising toolkit for genetic manipulations of type I and type II cochlear afferent fibers.

Spiral ganglion neurons (SGNs) receive inputs from hair cells, mechanoreceptors of the cochlea, to encode acoustic information into action potentials that travel into the central nervous system (CNS). SGNs are divided into two major groups based on their morphology and cochlear innervation pattern. Type I SGNs are larger diameter, myelinated neurons that constitute ~95% of the total auditory nerve fibers. They send a single dendrite to contact one inner hair cell (IHC). The remaining 5% are smaller diameter, unmyelinated type II afferent fibers that contact numerous outer hair cells (OHCs) as they spiral hundreds of microns towards the cochlear base 1,2 . Type I SGNs are responsible for encoding the salient parameters of sound 3 . Type II SGN function remains an area of active inquiry, with recent studies supporting a role in signaling tissue damage 4,5 .
Genetically engineered mouse lines that allow selective targeting and manipulation of specific neuronal groups are valuable tools for in vivo functional studies, fate-mapping during development, regeneration experiments and more. Since type II afferent fibers are few in number, small in caliber and unmyelinated, mouse genetic tools will be especially useful for defining their function in vivo. A variety of mouse lines have been described that label SGNs, for example: Shh (Sonic hedgehog)-Cre 6,7 , Neurog1 (Neurogenin1)-Cre 8 , Neurog1-CreER T2 9 , Bhlhb5-Cre 8 , PV (Parvalbumin)-Cre 10 . However, Cre drivers such as these don't distinguish type I from type II SGNs. The present work shows that all type I, but not type II SGNs express the enzyme neuronal nitric oxide synthase, making this a specific marker for future studies of type I SGNs.
Previous work has shown that tyrosine hydroxylase (TH) is preferentially expressed by apically-located type II afferents, while calcitonin gene related peptide alpha (CGRPα) is preferentially expressed by type II afferents in the cochlear base 11,12 . In the present work, two additional mouse lines are shown to specifically label type II SGNs, the serotonin transporter (SERT/Slc6a4) and a subunit of the dopamine receptor, Drd2. Furthermore, previously in different cell types in the cochlea, including but not limited to the inner and outer hair cells, SGNs and olivocochlear efferents [16][17][18][19] . However, the labeling pattern observed here was specific to the SGNs, and not found in hair cells or olivocochlear efferents. This discrepancy in the labeling patterns between nNos antibody and Nos1 CreER ; Ai9 mice could be due to various factors, such as the lack of antibody specificity, low expression of CreER in other cell types at the time of tamoxifen induction, or the timing difference between the Nos1 gene expression and nNos protein accumulation. The present results show that when induced in the second postnatal week, the Nos1 CreER mouse line can be used to label type I cochlear afferents specifically.
Serotonin reuptake transporter (SERT) is a membrane protein encoded by the Slc6a4 gene that recycles the neurotransmitter serotonin from the synaptic cleft into presynaptic neurons in a sodium-dependent manner 20 . In the auditory periphery, immunolabel of SERT has been reported in the olivocochlear efferent system 21 , auditory afferent fibers of developing marmoset 22 and embryonic (E15.5) rat cochlear nucleus 23 . Serotonergic synaptic activity also has been demonstrated in the cochlea by the use of biochemical inhibitors 24 . Here Slc6a4-GFP (also known as SERT-GFP), a BAC transgenic mouse line expressing GFP under the Slc6a4 promoter was used to study the expression of SERT in the cochlea. Whole mount fluorescence microscopy of Slc6a4-GFP mouse cochlea (P30) showed the expression of GFP in fibers in the organ of Corti along the three rows of outer hair cells from which short branches with bouton endings connect with the OHCs; a pattern typical for type II afferents 1,11,25 (Fig. 2a). When co-immunostained with β-tubulin 3 (TuJ1), GFP-expressing neurons did not overlap with TuJ1-positive type I SGNs (Fig. 2b-e), again supporting their identity as type II, but not type I, cochlear afferent neurons.
The expression pattern of Slc6a4-GFP cochleas was examined in pre-hearing (P7-9) and hearing mice (P30-45). GFP-expressing SGNs were counted in bins by dividing the cochlear whole mounts into 10 segments along the tonotopic axis using ImageJ, as described previously 12 . Representative images of apical, mid and basal turns of cochlea with labeled SGNs in pre-hearing mice are shown in Fig. 3a-d. GFP-expressing neuronal somata (arrowheads point to example type II SGNs) were present in all cochlear turns at both ages. Numerous small-diameter cells (arrows) were observed in Slc6a4-GFP mouse cochleas at both ages that were easily distinguished from SGNs by their size. Since SERT is expressed in platelets 26,27 and involved in regulating blood pressure 28 , these are likely to be platelets. Consistent with that conclusion, the putative platelets were found aligned within blood vessels (See Supplemental Fig. S3). The distribution of GFP-expressing SGNs peaks around 1/3 of the cochlear length from the apex and drops at both ends (Fig. 3e). Additional non-neuronal GFP-expressing cells were seen in the stria vascularis (Table 1).
Consistent with the observations using Slc6a4-GFP mice, another knock-in mouse line, Slc6a4 Cre (see Materials and Methods), also specifically labeled type II fibers when crossed with the Ai9 reporter line and analyzed at ages between P7-45 (see Supplementary Fig. S4). Similar to Slc6a4-GFP, cochleas in Slc6a4 Cre ; Ai9 mice also labeled non-neuronal cells in the osseous spiral lamina and stria vascularis, however, no expression was observed in the platelets (Table 1) (see Supplementary Fig. S4). Finally, we examined the cochlear labeling pattern of Slc6a4-Cre BAC transgenic mice (see Materials and Methods). This line showed a less specific labeling pattern, that included expression in both type I and type II SGNs and cochlear efferents, and therefore the expression was not investigated further.
Drd2-Cre mouse line labels type II afferents in the mid-basal region of the cochlea. Dopamine is a neurotransmitter of lateral olivocochlear efferents that regulate type I afferent signaling. Previous studies have reported the expression of dopamine receptor subtypes (D1-5) in spiral ganglion neurons by RT PCR and immunohistochemical analysis 29,30 . Type II-like morphology of fibers and their terminal boutons on OHCs could clearly be visualized in the mid and basal turns of Drd2-Cre; Ai9 mouse cochleas at P30 (Fig. 4a). When Drd2-Cre; Ai3 (R26 LSL-EYFP ) cochleas were co-immunostained with α3 Na + /K + ATPase, which is expressed specifically in myelinated type I afferents and medial efferents but not in unmyelinated type II afferents and lateral efferents 31 , SGNs positive for EYFP (i.e., Drd2-Cre driven) or α3 Na + /K + ATPase were mutually exclusive, as shown in the spiral ganglion region (Fig. 4b-e).
The expression pattern of Drd2-Cre was examined in pre-hearing (P7-9) and hearing mice (P30-45). Interestingly, the expression of Drd2-Cre was found only in the mid and basal type II afferents, although occasionally also in a few medial efferents in the apical cochlea (Fig. 5b, arrow) and presumably glia cells in the osseous spiral lamina (see Supplemental Fig. S5; Table 1). This expression gradient in type II SGNs across cochlea spiral is illustrated for pre-hearing cochlear whole mounts ( Fig. 5a-d) and is similar for cochleas from hearing animals (Fig. 5e). www.nature.com/scientificreports www.nature.com/scientificreports/  www.nature.com/scientificreports www.nature.com/scientificreports/

Comparison of expression between different molecular markers for type II afferents. Thus far,
four molecular/genetic markers have been shown to label type II afferent neurons, described here or in Wu et al. 12 .
To better illustrate and compare the tonotopic distribution of the labeled type II SGNs using these different strategies, graphical representations of these distributions at hearing age are shown in Fig. 6. Similar to Th and Cgrpα (also known as Calca) 12 , Slc6a4-GFP and Drd2-Cre also showed specific expression gradients along the cochlear coil (as summarized in Fig. 6c). Both TH antibody labeling as well as the distribution of Slc6a4-GFP labeled type II SGNs showed an apical preference, although the peak of TH labeling was found further apically (Fig. 6a). Slc6a4-GFP labeling is largely absent from the apical and basal extremes, but otherwise is distributed along the cochlear spiral, with a maximum at approximately 1/3 of the cochlear length from the apical end (Fig. 6a). Both CGRPα-EGFP and Drd2-Cre showed distributions biased towards the cochlear base. However, the Drd2-Cre labeled roughly half as many type II neurons compared to CGRPα-EGFP (P30, Fig. 6b). Given that the expression of Cgrpα is downregulated in type I SGNs during the first postnatal month 12 , there could be an overestimation for the number of type II SGNs based on CGRPα-EGFP labeling in one-month-old mice.
To assess further type II afferent heterogeneity and to understand better how to utilize the different molecular and genetic markers for manipulating type II afferents, labeling patterns were compared at the level of individual SGNs. For example, two genetic markers with similar tonotopic distribution might be expressed by separate populations of type II SGNs. However, type II SGNs that co-express candidate genes were found in the overlapping cochlear regions. While type II SGNs expressing Drd2-Cre; Ai9 and Slc6a4-GFP are largely segregated along the cochlear coil, individual SGNs co-expressing Slc6a4-GFP and Drd2-Cre; Ai9 can still be found in the cochlear middle turn (Fig. 7a,a1,a2,a3 asterisk). Drd2-Cre and the previously reported CGRPα-EGFP mouse lines both label basal type II afferents. Cross-bred Drd2-Cre; Ai9; CGRPα-EGFP mice showed co-expression of the reporter proteins in some SGNs (asterisks in Fig. 7b,b1,b2,b3). ' Apical' reporters were examined in Slc6a4-GFP mouse cochleas labeled with antibodies against TH (Fig. 7c). Most of the type II SGNs in the cochlear apical region were co-labeled by both markers (Fig. 7c1,c2,c3). SGNs expressing TH and Drd2-Cre were largely restricted to apex or base, respectively. However, a few co-labeled SGNs could be found in the middle turn (Fig. 7d,d1,d2,d3, asterisk). Finally, as previous reported, TH and CGRPα-GFP expressing neurons could show co-expression in the middle turn of the cochlea 12 . We were not able to look for co-expression of CGRPα-EGFP and Slc6a4-GFP, because both mouse lines express the same reporter protein.

Discussion
This work is part 3 of a series of studies to identify and validate mouse genetic tools for labeling and separately manipulating type I and type II afferents, the spiral ganglion neurons (SGNs) of the mammalian cochlea. As for previous studies in this series 11,12 genetically modified mice obtained from commercial vendors and local laboratories were examined for expression of reporter proteins in cochlear afferents. This work describes a mouse with CreER coupled to the Nos1 promoter that drove reporter expression specifically in type I but not type II SGNs. In addition, three mouse lines, Slc6a4-GFP, Slc6a4 Cre and Drd2-Cre could be used to target type II, but not type I, SGNs. Between tested genes, different apical-to-basal expression patterns and different amounts of overlap were found between markers, suggesting that subpopulations of type II neurons exist. www.nature.com/scientificreports www.nature.com/scientificreports/ Does the labeling of specific groups of SGNs by these different mouse lines represent the endogenous expression pattern of these genes? Nos1 CreER was constructed by inserting CreER into the endogenous Nos1 locus. Therefore, its expression most likely reflects the actual gene expression. The presence of nitric oxide synthase (NOS) has been described in cochlear tissue, including SGNs 17,18,32,33 . Slc6a4-GFP and Drd2-Cre were both made by random insertion of bacterial artificial chromosomes (BAC) containing the promoter and regulatory sequence of these two genes. Depending on where the BAC integrates in the genome, the expression may or may not reflect the endogenous pattern. However, the labeling patterns of these two lines have been validated in the central nervous system (CNS) by the GENSAT project (www.gensat.org). The specific expression of SERT in type II afferent neurons has been replicated with another Slc6a4 Cre knock-in line. For the Drd2-Cre line, the expression patterns in the CNS have been further validated by in-situ hybridization 34 . Additional supporting evidence comes from three recently published single-cell RNA sequencing (scRNAseq) studies on SGNs [35][36][37] . NOS1 was identified in all three studies as a gene that is expressed in type I but not type II SGNs. In addition, Nos1 was expressed in all three subtypes of type I SGNs (based on principal component clustering), corresponding with the present observation of universal expression of reporter protein in type I SGNs. Similarly, TH and SERT were identified as marker genes for the type II SGNs in all three studies. Evidence for differential CGRPα expression in type II versus type I SGNs was reported in two of the studies. CGRPα was expressed at higher levels in type II SGNs than in type I SGNs 37 and was among the genes expressed differentially by type I and type II SGNs 36 . Drd2 expression was not reported in these publications, possibly due to low RNA levels, but the online data repository 37 showed that Drd2 is expressed at a low level in one of the subtypes of type I SGNs. The discrepancy between the scRNAseq result and the Drd2-Cre mouse line labeling could be due to various factors. It is possible that basal type II afferent neurons express Drd2 during development and downregulate its expression in adults. Alternatively, because the Drd2-Cre mouse line is constructed by random insertion of BAC in the genome, which are prone to internal rearrangements, the reporter expression induced by this line may not represent endogenous expression of Drd2 gene in the cochlea. To sum up, the expression patterns of NOS1, TH, CGRPα and SERT genes in SGNs in these transgenic mouse lines is largely consistent with scRNAseq data. We do see different types of unidentified cells www.nature.com/scientificreports www.nature.com/scientificreports/ in Nos1 CreER , Drd2-Cre and SERT-Cre expressing mouse cochleas that were not traceable in the literature. These unidentified cell types were found in different regions of the cochlea, have different shapes and molecular foot prints and need further analysis for identification. Whether basal type II afferents express Drd2 mRNA or protein awaits further investigation. It bears repeating that even without such confirmation, these mouse lines still can serve as experimental models for the study of SGNs.
Does the expression of these marker genes in the SGNs tell us anything about function? Nitric oxide regulates voltage-gated ion channels of hair cells 38,39 and possibly can act as a retrograde transmitter to increase the probability of transmitter release from efferent terminals on inner hair cells prior to the onset of hearing 40 . Soluble guanylyl cyclase, the principal target of NO, is expressed in olivocochlear efferents 41 . Do the expression of TH, SERT and CGRPα suggest that type II afferent neurons use dopamine, serotonin and CGRP as neurotransmitters? Because they are all suggested olivocochlear efferent neurotransmitters 42 , previous studies of these neurotransmitters have focused logically on efferent neurons (in addition to glutamate transmission from hair cells) but made no mention of type II cochlear afferents. Dopamine release from lateral olivocochlear efferents can suppress the activity of type I afferents [43][44][45] but the cellular effects of CGRP and 5-HT remain to be determined. Expression of TH is not in itself a guarantor of dopaminergic transmission. If the Drd2-Cre labeling represents the endogenous expression, the opposing patterns of Th and Drd2 along the tonotopic axis further complicates any functional interpretation. Using RT-PCR, Drd2 receptor transcripts were identified in the OHCs 46 . Whether type II afferents could release dopamine in a retrograde fashion to act on OHCs requires further investigation. Thus, while these expression patterns may prove useful for future experimental strategies, they don't change our present understanding of OHC to type II afferent synaptic function. Synaptic currents evoked in type II afferents by high potassium depolarization of cochlear tissue are due to glutamate release from outer hair cells that acts on AMPA [47][48][49] and possibly kainate receptors 50 . Besides those synaptic signals, type II afferents respond to extracellular ATP with P2X and P2Y type receptors 4 . Intracellular recording from OHCs has yet to reveal any synaptic currents other than those due to acetylcholine release from cholinergic medial olivocochlear terminals [51][52][53][54] . If dopamine, or CGRP or 5-HT are released from type II afferent terminals in the cochlear nucleus, their actions there remain to be determined.
A hallmark of the cochlea is the 'tonotopic' organization of macro-and microscopic features that underlie acoustic frequency selectivity. For example, the basilar membrane increases in stiffness from apex to base, giving rise to the mechanically-tuned traveling wave described by von Békésy 55 . The neuronal innervation of the cochlea also varies along the tonotopic axis 56,57 . Afferent and efferent innervation is highest in mid regions of the cochlea (serving the most sensitive range of hearing) and there is a general tendency for greater numbers of both afferent and efferent contacts in the higher frequency cochlear base.
Since individual type I afferents contact a single inner hair cell in the mature cochlea, one might predict these neurons to be specialized according to their acoustic frequency selectivity. Indeed, there is already evidence supporting tonotopic variation of type I SGNs. For example, intracellular recording from dissociated type I afferents  www.nature.com/scientificreports www.nature.com/scientificreports/ showed that basic membrane properties differed between those originating apically versus those from the cochlear base 58 . Moreover, within each of the three major subgroups of type I SGNs clustered by scRNAseq, there was tonotopic variation of expression for a subset of genes 35 . Taken together these results support the hypothesis that type I SGNs may fall into functional subgroups; with minor functional differences along the tonotopic axis.
Type II afferents are known to vary morphologically along the cochlear length 59 . Possibly due to the relatively small number of type II afferent neurons, very few studies have addressed heterogeneity among type II SGNs. Similarly, the recent scRNAseq studies also did not provide additional insights regarding subgroups within type II SGNs. However, expression patterns of different transgenic mouse lines have revealed 'tonotopic' heterogeneity among type II SGNs. Previous studies showed that promoters for TH and CGRPα were active in apical and basal type II afferents, respectively 11,12 . In the present work the promoters for SERT and DRD2 also drove tonotopic gradients in reporter protein expression by type II afferents. Each of these four genetic drivers had distinctive patterns of expression along the cochlea's tonotopic axis. TH predominates in the apical half, SERT is expressed in a bell-shaped expression gradient with a bias toward the cochlear apex, while CGRPα and DRD2 appear preferentially in the cochlear base. In mid-cochlear regions of overlap, CGRPα and TH can be co-expressed by single type II neurons 12 . The present work also shows that single type II SGNs can co-express more than one label in regions of overlap, suggesting that these are not mutually exclusive populations. Rather, it suggests that the expression patterns found here reveal some underlying genetic differentiation related to cochlear position.
Why do type II afferents differ along the cochlea's tonotopic axis? Comparisons to somatosensory afferents might be illuminating. Pain-sensing C-fibers of skin express CGRP and this signaling molecule plays a role in damage-triggered inflammation 60 . The preferential expression of CGRP by type II afferents in the higher frequency base of the cochlea would be consistent with the greater sensitivity of this region to acoustic trauma and so might serve an analogous function. In contrast, TH is expressed by unmyelinated low threshold mechanoreceptors (C-LTMRs) of skin 61 that play a role in 'emotional touch' 62 and TH predominates in the low frequency cochlear apex where tissue trauma is less likely. It will be of interest to explore further the tonotopic diversity of type II afferents.

Methods
All animal experiments were carried out in according to the guidelines approved by the Johns Hopkins Animal Care and Use committee (ACUC). Mice from both sexes were used in experiments. No obvious differences were observed between males and females in this study. For every finding, at least three experiments with animals from three litters were performed. Tamoxifen Injection. Tamoxifen stock (Sigma #T5648) was prepared by dissolving tamoxifen in corn oil (Sigma #C8267) at a concentration of 10 mg/ml for sonication at room temperature (2 h). Stock solutions were stored at 4 °C in the dark and were used within 4 days of preparation. For studying the phenotype of postnatal day (P) 7 animals, tamoxifen was administrated through intragastric injection 64 at P3 and P5 (0.2 mg each time) using an insulin syringe with an ultrafine needle (BD, 22 G). For analysis at 3-7 weeks, tamoxifen (1 mg) was administered by intra-peritoneal injections in the second postnatal week.