Abstract
One of the most striking aspects of the sensory epithelium of the mammalian cochlea, the organ of Corti (OC), is the presence of precise boundaries between sensory and nonsensory cells at its medial and lateral edges. A particular example of this precision is the single row of inner hair cells (IHCs) and associated supporting cells along the medial (neural) boundary. Despite the regularity of this boundary, the developmental processes and genetic factors that contribute to its specification are poorly understood. In this study we demonstrate that Leucine Rich Repeat Neuronal 1 (Lrrn1), which codes for a single-pass, transmembrane protein, is expressed before the development of the mouse organ of Corti in the row of cells that will form its medial border. Deletion of Lrrn1 in mice of mixed sex leads to disruptions in boundary formation that manifest as ectopic inner hair cells and supporting cells. Genetic and pharmacological manipulations demonstrate that Lrrn1 interacts with the Notch signaling pathway and strongly suggest that Lrrn1 normally acts to enhance Notch signaling across the medial boundary. This interaction is required to promote formation of the row of inner hair cells and suppress the conversion of adjacent nonsensory cells into hair cells and supporting cells. These results identify Lrrn1 as an important regulator of boundary formation and cellular patterning during development of the organ of Corti.
SIGNIFICANCE STATEMENT Patterning of the developing mammalian cochlea into distinct sensory and nonsensory regions and the specification of multiple different cell fates within those regions are critical for proper auditory function. Here, we report that the transmembrane protein Leucine Rich Repeat Neuronal 1 (LRRN1) is expressed along the sharp medial boundary between the single row of mechanosensory inner hair cells (IHCs) and adjacent nonsensory cells. Formation of this boundary is mediated in part by Notch signaling, and loss of Lrrn1 leads to disruptions in boundary formation similar to those caused by a reduction in Notch activity, suggesting that LRRN1 likely acts to enhance Notch signaling. Greater understanding of sensory/nonsensory cell fate decisions in the cochlea will help inform the development of regenerative strategies aimed at restoring auditory function.
Introduction
Development of the mammalian auditory sensory epithelium, the organ of Corti (OC), requires the interplay of several conserved signaling pathways. For instance, the Hedgehog, Notch, Wnt, Fgf, and Bmp signaling pathways coordinate to partition the early cochlear epithelium into prosensory and nonsensory regions (Groves and Fekete, 2012; Wu and Kelley, 2012). Subsequent signaling events further specify prosensory cells to develop as distinct types of hair cells and supporting cells while nonsensory cells also assume multiple fates (Lanford et al., 1999; Mueller et al., 2002; Ohyama et al., 2010; Jacques et al., 2012). A key step in the initiation of the development of the OC is the onset of differentiation of the single row of inner hair cells (IHCs), located at what will become the medial edge of the OC (Rubel, 1978). The first IHCs appear in the mid-basal region of the cochlear duct, with additional IHCs forming in a contiguous line extending toward the apex. Soon after the formation of the first IHCs, three rows of outer hair cells (OHCs) begin to form laterally (Driver and Kelley, 2020). As is the case for IHCs, OHC development also occurs in a wave that extends from base to apex. While developing OHCs undergo considerable migration and rearrangement (Yamamoto et al., 2009; Driver et al., 2017; Cohen et al., 2020), IHCs initially appear within a few cell diameters of their mature position (McKenzie et al., 2004), suggesting that the medial edge, and therefore the medial-lateral axis, of the OC is defined before the onset of IHC development.
IHCs in the mature OC are arranged in a single row, indicating a precise boundary between sensory and nonsensory regions of the duct, but recent work has demonstrated that the formation and maintenance of this boundary is a dynamic process (Basch et al., 2016). In particular, Notch-mediated interactions play a key role in preventing nonsensory cells located adjacent to the medial boundary from assuming a sensory fate (Kiernan et al., 2001, 2006; Basch et al., 2016). Our understanding of how this process occurs, and the genetic factors that regulate it, remains very limited.
In the Drosophila wing imaginal disk, Notch interacts with two other transmembrane proteins, Tartan and Capricious, to define and maintain a dorsal-ventral boundary (Blair, 2001; Milán et al., 2001, 2005). Similarly, in vertebrates, an ortholog of Tartan and Capricious, Leucine Rich Repeat Neuronal 1 (Lrrn1), has been implicated as a mediator of rhombomere boundary formation (Andreae et al., 2007), and shown to be involved in specifying the chick thalamus-prethalamus and midbrain-hindbrain boundaries (Tossell et al., 2011). In this study, we demonstrate that Lrrn1 is expressed along the medial boundary of the OC before hair cell formation and that deletion of Lrrn1 leads to defects in medial boundary formation. Further, we show that during cochlear development Lrrn1 interacts with the Notch signaling pathway and that Lrrn1 expression is linked to the specification of medial boundary identity. These results demonstrate a key role for Lrrn1 during cochlear medial boundary formation.
Materials and Methods
Animal care
All animals were maintained in the Porter Neuroscience Shared Animal Facility according to the animal care and use guidelines provided by the National Institute of Neurological Disorders and Stroke/National Institute on Deafness and Other Communication Disorders Animal Care and Use Committee of the National Institutes of Health.
Multiplex single molecule fluorescence in situ hybridization (smFISH)
For multiplex smFISH on tissue sections, inner ears were dissected from embryonic day (E)14, E16, and postnatal day (P)0 embryos, fixed in 4% paraformaldehyde (PFA) in PBS overnight (O/N) at 4°C, then cryoprotected in sucrose, embedded in OCT, and frozen for cryosectioning. Sections were cut at a thickness of 10 µm. Following sectioning, samples were pretreated and assayed for the fluorescent hybridization following the ACD RNAscope Fluorescent Multiplex Reagent kit protocol (320850; Kolla et al., 2020). Sections were counter-stained with DAPI to visualize nuclei. For whole mount smFISH, cochlea were dissected from E15 embryos, fixed overnight, and processed following the protocol of Kersigo et al., 2018. The following probes were obtained from Advanced Cell Diagnostics: Pvalb, Sox2, Lrrn1, Hes1, Mfng, and Lfng.
Immunostaining
Cochlear whole-mounts and cryosections were incubated in primary antibodies overnight at 4°C. Sheep anti-human/mouse LRRN1 (R&D Systems AF4990, RRID: AB_2234807) was used at 0.1 µg/ml. Other primary antibodies used were rabbit anti-CDKN1B (Thermo Fisher Scientific PA5-16717, RRID:AB_10977242), rabbit anti-MYOSIN6 (Proteus Biosciences 25-6791, RRID:AB_10013626), rat anti-CDH1 (Santa Cruz Biotechnology sc-59778, RRID:AB_781738), rabbit anti-NPY (Sigma-Aldrich N9528, RRID:AB_260814), rabbit anti-FABP7 (Thermo Fisher Scientific PA5-24949, RRID:AB_2542449), rabbit anti-MYOSIN7A (Proteus Biosciences, 25-6790, RRID:AB_10015251), goat anti-FABP7 (R&D Systems, RRID:AB_2100475), goat anti-NGFR (R&D Systems, RRID:AB_2152638), goat anti-PROX1 (R&D Systems AF2727, RRID:AB_2170716), sheep anti-DLL1 (R&D Systems; RRID:AB_2092830), mouse anti-POU4F3 (Santa Cruz Biotechnology sc-81980, RRID:AB_2167543), goat anti-SOX2 (R&D Systems, RRID:AB_355110), rabbit anti-DsRed (Takara Bio 632496, RRID:AB_10013483), mouse anti-Ki67 (Santa Cruz Biotechnology sc-23 900, RRID:AB_627859), and goat anti-JAGGED1 (R&D Systems, AF599, RRID:AB_2128257). All primary antibodies other than anti-LRRN1 were used at 0.4–1 µg/ml. Alexa Fluor-labeled secondary antibodies (0.5 µg/ml) were used in parallel with Alexa Fluor-conjugated phalloidin to label F-actin (1 U/ml); all were obtained from Thermo Fisher Scientific.
Generation of Lrrn1-CreERT2 reporter line and fate mapping of Lrrn1+ cells
Recombineering was used to replace the coding sequence of Exon 2 of the mouse Lrrn1 gene (NCBI Gene 16 979, MGI 106038, Ensembl: ENSMUSG00000034648), with a CreERT2-Neo cassette in BAC clone RP24-27107. Transgenic mice were generated through pronuclear injection and incorporation of the transgene was confirmed by PCR. The Lrrn1-CreERT2 line was then crossed with homozygous Gt(ROSA)26Sortm14(CAG-tdTomato)Hze mice (Rosa26tdTomato(tdT); The Jackson Laboratory, strain #007914) to generate Lrrn1CreERT2/+; Rosa26tdT/+ transgenic animals. Lrrn1-CreERT2; Rosa2tdT/+ males were crossed with CD1 wild-type (WT) females, with pregnancies timed based on the appearance of a plug on day embryonic day (E)0. At E12 or E16, 400 µg of tamoxifen (Sigma-Aldrich T5648) was dissolved with 20 mg/ml progesterone (Sigma-Aldrich P0130) in a flax seed oil solution and administered to pregnant females by oral gavage. Pregnancies were maintained until litters were delivered. Pups were euthanized at P0, and inner ears were dissected, fixed, and prepared for immunostaining and analysis.
Cochlear explants
For cell fate analysis, Lrrn1-CreERT2; Rosa26tdT/+ cochleae were dissected from E14 embryos and established as explants as previously described (Driver and Kelley, 2010). To induce recombination, 4-OH tamoxifen (Sigma-Aldrich 579002) was added at 100 nm or 1 μm for 24 h beginning at immediately at E14 or after 1 d in vitro (DIV; E15 equivalent). Explants were maintained for 5 DIV total (P0 equivalent), then fixed and processed for immunostaining.
For in vitro inhibition of γ-secretase, cochlear explants were established from Lrrn1+/+, Lrrn1+/−, and Lrrn1−/− embryos at E14 and treated with media containing 500 pm LY411575 (Sigma SML0506) or control media (0.2% DMSO) for the initial 2 DIV. After 2 DIV, the LY411575-containing media was removed and replaced with control media. Explants were fixed after an additional 3 DIV (P0 equivalent) and processed as described above.
For inhibition of GSK3 activity to disrupt medial-lateral boundary formation, explants were established from CD1 embryonic cochleae at E13 and cultured for 2 DIV, either in control media (0.2% DMSO) or in media containing 2 μm CHIR99021 (Tocris 4423). After 2 DIV, explants were fixed and processed for immunostaining.
Generation of Lrrn1-TALEN mutant mice
Five TALENs were designed to target the single coding exon of Lrrn1 using TALEN targeter and assembled using golden gate cloning. Golden Gate TALEN and TAL Effector kit 2.0 was purchased from Addgene (#1000000024). TALEN gene editing efficiency was tested in Neuro2A cells by GeneCopoeia as follows: Neuro2A cells were transfected with plasmids containing the left and right TALEN arms at a concentration of 0.5 µg each. Cells were collected 48 h post-transfection, genomic DNA was extracted, and an 800-bp region around the TALEN target site was PCR amplified. DNA was melted, slowly re-hybridized and digested with T7 Endonuclease I (NEB catalog #M0302S) to detect mismatched base pairs. One chromosome-level validated TALEN pair was prepared for pronuclear injection into mouse embryos as follows: mRNA was generated from TALEN plasmid using SP6 mMessage mMachine kit (Thermo Fisher Scientific catalog #AM1340). mRNA was poly-adenylated using Poly(A) Tailing kit (Thermo Fisher Scientific catalog #AM1350). Poly-adenylated mRNA was cleaned using RNeasy Mini clean up kit (QIAGEN catalog #74104). Pronuclear injection of mouse embryos was performed at the Transgenic Mouse Model Laboratory at Frederick National Laboratory for Cancer Research by Leidos Biomedical Research TALEN founders were identified using genomic DNA from mouse tail snips using the T7 Endonuclease I assay as described above, and three independent mouse lines carrying 2-, 4-, or 31-bp deletions in the 10th leucine-rich repeat in the extracellular domain of Lrrn1. Each mutation was predicted to result in a frameshift leading to an early stop codon. For all three alleles, no significant changes in cochlear morphology or length were observed in heterozygotes, and homozygous mutants all displayed increased numbers of extra IHC and IPhCs (doublets) medial to the sensory/nonsensory boundary. The 2-bp deletion allele was maintained on a CD1 background, and the 4- and 31-bp alleles were maintained on a C57BL/6J background.
Generation of Lrrn1−/−; Notch1+/− mutant mice
Male mice carrying a conditional Notch1 allele (Radtke et al., 1999) were crossed to Zp3Cre (de Vries et al., 2000, RRID:IMSR_JAX:003651) females to generate a germline deletion of Notch1, then crossed to mice carrying the 2-bp deletion allele of Lrrn1. Lrrn1+/−; Notch1+/− mice were crossed to generate Lrrn1+/+; Notch1+/−, Lrrn1+/−; Notch1+/−, and Lrrn1−/−; Notch1+/− animals. These mice were maintained on a mixed background including the outbred CD1 strain.
Cochlear imaging and cell counts
All imaging of cochlear whole-mounts, sections, and explants was performed on a Zeiss LSM710 or LSM900 confocal microscope. Images shown are XY Z-stack projections or orthogonal YZ views, as indicated.
The number of ectopic IHCs and IPhCs per unit length was determined by measuring the total length (µm) of each cochlea, dividing each cochlea into thirds, designated as apical, middle, and basal regions, and then counting the number of ectopic IHCs (MYO7A+) and IPhCs (FABP7+) that appeared next to the single row of IHCs aligned adjacent to the pillar cells and IPhCs along a cochlear length of 1000 µm (∼1/3 of the cochlear length).
To quantify the density of IHCs in γ-secretase inhibitor-treated explants, the total length of each explant was measured. The apical-most 25% of each explant was excluded from analysis because of the high morphologic variability in this region in explants. All IHCs (MYO7A+ cells medial to the pillar cell region) were counted in the basal 75% of each explant, divided by the length of that portion of the explant in µm, and shown as that value per 100 µm.
Comparisons of the intensity of Hes1 expression detected by smFISH were made from images taken 250 µm from the base of E15 Lrrn1+/+ or Lrrn1−/− cochleae. All images were obtained with identical LSM acquisition settings. We generated Z-stack projection images with a depth of 1.5 µm (from five optical slices), starting 1.5 µm from the luminal surface of the HCs. Using ImageJ, we measured the average Hes1 pixel intensity in 40 × 8 µm rectangular regions of interest (ROIs) at equivalent base-to-apex positions, both lateral to the third row of OHCs and medial to the IHCs. The ratio of the average intensity of Hes1 expression in the medial ROI to the lateral ROI was calculated.
Graphs and statistical analyses
Graphs were created in GraphPad Prism. Cell counts and measurements are shown as box and whisker plots, with the box extending from the 25th to 75th percentile, whiskers to the minimum and maximum values, and a line indicating the median. For Lrrn1−/− phenotype analysis, we collected and compared cochleae from eight Lrrn1 mutants with the 4-bp deletion to eight heterozygous littermates, pooled from three different litters for biological replicates. Two 31-bp deletion Lrrn1 mutants and one heterozygous littermate were also included in the analysis. Extra IPhCs were counted in four mutant and four heterozygous Lrrn1 samples from two litters. Analysis of the mutant and heterozygous genotypes was performed by independent, two-tailed, unequal variance Student's t tests, which specifically compared counts of the respective cell type per cochlear region [e.g., ectopic IHCs/(1/3 cochlear length)] to give the estimated number of that cell type per 1 mm in the apex, middle, and base of each sample. Significance annotated by ** (p≤0.01), *** (p≤0.001), and *** (p≤0.0001). A Holm–Šidák test was used to correct for multiple comparisons per analysis.
For lineage tracing analysis in Lrrn1-CreERT2; Rosa26tdTomato mice, in vivo cell labeling was sparse, so we performed our quantification with respect to all of the cells labeled per time point (E12 or E16). Following characterization and quantification of the distinct cell types labeled in the cochlear sensory epithelium, we then calculated the proportion of each type relative to the total labeled. For multiple comparisons, a two-way ANOVA was performed, comparing the effect of both the stage of Cre induction (E12 and E16) and the specific cell types labeled (IHC, IPhC, Border, Deiters, Pillar). This was followed by Tukey's multiple comparison test. Statistically significant differences between genotypes or treatments were determined using Welch's t-test.
A mixed-model (Type III) two-way ANOVA and Tukey's test were similarly used to assess the effect of Lrrn1 and Notch1 genotypes and their potential interaction on the number of extra IHCs, endogenous IHCs (Fig. 6B,C) and OHCs (data not shown); p-values were adjusted for multiple comparisons according to the Holm–Šidák method. To quantify the changes in cell shape and density observed in Lrrn1−/−; Notch1+/− cochleae (N = 54 cells, 7 cochleae) relative to Lrrn1−/−; Notch1+/+ cochleae (N = 76 cells, 7 cochleae), we analyzed high magnification images of the middle regions of each sample. Data on parameters including the area and perimeter of individual, intact cells in each image were collected using the Analyze Particles and Measure tools in ImageJ.
Differences in the density of total IHCs in LY411575-treated explants compared with control explants were determined by ordinary one-way ANOVA, followed by Šidák's multiple comparisons test. Differences in the ratios of medial to lateral Hes1 expression were determined by unpaired, two-tailed t test.
Results
Lrrn1 is expressed in cells at the medial boundary of the embryonic and neonatal OC
Analysis of single-cell RNAseq data from the developing cochlear duct indicated that Lrrn1 expression is predominantly, but not exclusively, expressed in cells located at the medial boundary of the OC at embryonic day (E)14, E16, and postnatal day (P)1 (Fig. 1A,B; data not shown). To determine the timing of onset and localization of Lrrn1 during cochlear development, we performed single molecule fluorescence in situ hybridization (smFISH) on inner ear sections from E14, E16, and P0. E14 sections were also hybridized with a probe against Sox2, which marks the prosensory domain (Kiernan et al., 2005). In the cochlear base at E14, Lrrn1 mRNA is expressed in a narrow band, possibly only the width of a single cell, that overlaps with expression of Sox2 near the medial edge of the prosensory domain (Fig. 1C).
Expression of Lrrn1 is restricted to the medial edge of the developing OC. A, t-distributed Stochastic Neighbor Embedding (tSNE) plot of P1 cochlear single cells based on single-cell RNAseq results generated using the gEAR website (Orvis et al., 2021). Left side image depicts different cell types as labeled. Circle indicates inner phalangeal cells (IPhC). Right side image shows expression of Lrrn1 in the same plot. Note increased expression in the IPhC group (circle). B, Violin plot for expression of Lrrn1 in the same data as in A generated using gEAR. Note that only IPhCs show consistent increased expression of Lrrn1. C–E, Confocal images of smFISH sections through the basal turn of the cochlea from WT mice at E14, E16, and P0. C, C', At E14, expression of Lrrn1 overlaps with Sox2, a marker of the prosensory domain (asterisk, bracket). In C'', the bracket indicates the prosensory domain (PS) and cell nuclei are labeled with DAPI. D, At E16, expression of Lrrn1 overlaps with Pvalb, a marker of developing IHCs (asterisk). The bracket in D'' indicates the developing organ of Corti (OC). E, At P0, Lrrn1 is still co-localized with Pvalb but the intracellular labeling pattern appears different, likely indicating Lrrn1 expression in IPhCs at this stage (arrows). Scale bar in C (same in D, E): 25 µm. IHC: inner hair cells, OHC: outer hair cells, IPhC: inner phalangeal cells, HeC: Hensen's cells, IPC: inner pillar cells, OS: outer sulcus cells, DC1/2: Deiters' cells row 1 or 2, DC3: Deiters' cells row 3, OPC: outer pillar cells, LKO1–3: lateral Kölliker's organ cells groups 1–3, MKO: medial Kölliker's organ cells, IS: inner sulcus cells, IdC: interdental cells, OC90: otoconin 90+ cells.
LRRN1 protein is expressed in IHCs and IPhCs/BCs. A, E14 cochlear whole-mount stained with anti-LRRN1 antibody (grayscale). LRRN1 is found in a single row of cells (arrow) extending ∼3/4 of the way from the base to apex. A', Same image as A, showing LRRN1 expression at the medial edge of the prosensory domain, marked by CDKN1B staining. B–D, Confocal Z-stack projections of the surface of the basal region of E14 WT cochleae. B, LRRN1+ cells are mostly negative for surface CDKN1B staining (arrows), but some cells are both LRRN1+ and CDKN1B+ (B', B'', arrowheads). C, Most LRRN1+ cells are also weakly positive for MYO6 (arrows). LRRN1+, MYO6− cells (C', open arrowheads) are often medial to the row of presumptive IHCs and may be developing IPhCs. D, Cadherin-1 (CDH1) is excluded from the medial edge of the prosensory region (bracket), and LRRN1 and CDH1 staining are nonoverlapping. E, E16 whole-mount showing LRRN1 expression along the length of the cochlea (arrows). Anti-LRRN1 signal is also detected in the spiral ganglion (bracket). F–H, Confocal Z-stack projections of the surface of the indicated region from E16 WT cochleae. F, At the apex, LRRN1 is still found in most IHCs (arrows), which have bright phalloidin staining of the cell cortex (F'). Some IHCs are not LRRN1+ (open arrowheads). G, LRRN1 in the mid-base is still found medial to the pillar cells, marked by NPY (G', large arrowhead), but anti-LRRN1 staining appears to mainly occur in IPhCs/BCs (arrowheads) though some IHCs are also positive. H, At the P1 apex, LRRN1 staining is weak and mostly overlapping with FABP7 (H') in the IPhCs/BCs (H'', brackets). IHCs are LRRN1negative (arrows). Scale bar in A, E, 200 µm; B, H, 20 µm.
Fate mapping indicates that Lrrn1+ cells mainly become medial OC cell types. A, Confocal image of Lrrn1-CreERT2; Rosa26tdTomato whole-mount cochlea induced in vivo at E16. tdTomato+ cells (magenta) are present in both periotic mesenchyme (MES; arrowheads) and in the sensory epithelium (marked with anti-FABP7 in green and anti-POU4F3 in blue). Arrows indicate Lrrn1+ cells in the sensory epithelium. B, Orthogonal views of Lrrn1+ cells in the OC labeled as in A. Examples of an IHC, IPhC, and border cell (BC) are illustrated. Right-hand panels of tdTomato alone illustrate cellular morphology. C, Quantification of Lrrn1+ cell types in cochleae induced in vivo at E12 (n = 8 cells from 4 cochleae) or E16 (n = 55 cells from 9 cochleae). Most Lrrn1+ cells developed as IHCs, IPhCs, or BCs. A small number of all Lrrn1+ cells acquire the lateral fates of pillar cells (2) or Deiters' cells (4), suggesting limited mixing of Lrrn1+ cells. D, Low-magnification confocal image of a Lrrn1-CreERT2; Rosa26tdTomato cochlear explant established at E14 and induced with 100 nm 4-OH tamoxifen on the day of culture. Explants were fixed and stained after 5 DIV. Lrrn1+ cells are in magenta while supporting cells (SOX2) and hair cells (POU4F3) are marked in green and blue, respectively. E, Orthogonal views of the three main Lrrn1+ cell types, IHC, IPhC, and BC, from in vitro fate mapping. F, Quantification of Lrrn1+ cell fates separated by region along the cochlea (n = 91 cells from 4 explants). In the apical, middle, and basal regions, most Lrrn1+ cells developed as IHCs, IPhCs, or BCs. A small number of Lrrn1+ Deiters' cells or OHCs (4/91) were observed suggesting minimal cell mixing. G, Confocal image of a Lrrn1-CreERT2; Rosa26tdTomato cochlear explant established at E14 and induced at E15 with 1 μm 4-OH tamoxifen and analyzed after 5 DIV, labeled as in D. Note the stripe of magenta cells along the medial edge of the OC. H, Higher magnification of the boxed region in G. Lrrn1+ cells form a continuous stripe along the medial edge of the OC, with some positive cells in Kölliker's organ. Scale bar in G (same in A, D), 200 µm; B (same in E), 10 µm; H, 50 µm.
At E16, Parvalbumin (Pvalb) is expressed in developing IHCs located in the basal and middle turns of the cochlea (Xiang et al., 1998), and can be used to locate the medial boundary. In sections that are double labeled for Pvalb and Lrrn1, the two markers strongly overlap (Fig. 1D), confirming localization of Lrrn1 to the medial edge of the prosensory domain. Finally, in sections of P0 cochleae, Lrrn1 expression continues to overlap partially with Pvalb in cells along the medial edge of the OC (Fig. 1E). However, the intracellular pattern of Lrrn1 expression does not fully overlap with Pvalb (Fig. 1E''), suggesting that Lrrn1 may be expressed in cells other than hair cells, most likely inner phalangeal cells (IPhCs), based on location. These results indicate that Lrrn1 expression is primarily found at the medial boundary of the cochlear sensory epithelium between E14 and at least P0.
LRRN1 expression switches from inner hair cells to inner phalangeal cells
To confirm smFISH results and to localize LRRN1 expression to specific cell types, immunofluorescence was performed on cochlear whole-mounts, using a commercial polyclonal antibody against the N-terminal portion of human LRRN1. At E14, LRRN1 expression appears as a narrow band extending from the base of the cochlea ∼75% to the apex (Fig. 2A). Double labeling with the prosensory marker CDKN1B (formerly p27kip1) indicates that LRRN1 is localized to the medial edge of the prosensory domain (Fig. 2A'). A high-magnification view of the epithelial surface shows LRRN1+ cells forming a single or, in some cases, double row of cells (Fig. 2B). Most LRRN1+ cells are negative for surface labeling of CDKN1B, which is an early indicator of HC differentiation as cells exit mitosis (Chen and Segil, 1999; Atkinson et al., 2015). Co-labeling with MYO6, another early hair cell marker, indicates that the majority of LRRN1+ cells at E14 are developing IHCs (Fig. 2C). However, some LRRN1+ cells are MYO6 negative. Whether these cells represent immature IPhCs remains undetermined, as definitive markers of IPhCs are not expressed at this stage. Expression of LRRN1 and CADHERIN1 (also known as E-Cadherin), a cell-adhesion protein that is excluded from the medial edge of the prosensory domain (Chacon-Heszele et al., 2012) were also nonoverlapping (Fig. 2D). At E16, LRRN1 expression is present along the full length of the cochlear spiral (Fig. 2E), and the anti-LRRN1 antibody also labels cells within the spiral ganglion. Examination of cellular expression indicates a transition from IHCs in the apex to IPhCs and the more medial border cells (BCs) in the base (both marked by FABP7; Fig. 2F,G). In addition, LRRN1-dependent fluorescence intensity indicates a decrease in overall LRRN1 expression along the apical-to-basal gradient. By P1, expression of LRRN1 protein in the cochlea appears to be very low but is detected in apical IPhCs (Fig. 2H).
Fate-mapping confirms that Lrrn1+ prosensory cells predominantly develop as IHCs, IPhCs, and border cells
To definitively identify the fate of Lrrn1+ cells, we generated a Lrrn1-CreERT2 transgenic mouse line (refer to Materials and Methods). Lrrn1-CreERT2 was crossed to the Rosa26tdTomato transgenic reporter (Madisen et al., 2010), and CreERT2 activity was induced by administering tamoxifen at E12 or E16. Cochleae were collected at P0 and labeled for tdTomato, POU4F3 (a HC marker), and FABP7. We observed sparse labeling of Lrrn1+ cells in several regions of the cochlea (Fig. 3A). A substantial number of mesenchymal cells located beneath the basement membrane were positive for tdTomato, as well as a limited number of cells within the cochlear epithelium (Fig. 3A,B). Morphologic analysis using high magnification images indicated that tdTomato+ epithelial cells were primarily IHCs, IPhCs, or BCs (Fig. 3B). Quantification of Lrrn1+ cell fates following Cre-mediated recombination indicates that the vast majority (87.5%, E12 and 90.1%, E16) of all labeled cells develop as cells located along the medial border of the OC (Fig. 3C). Of the 61 Lrrn1+ cells characterized across both induction stages, the mean percentage of IHCs (18.5), IPhCs (29.5), and border cells (41.0) was far greater than that of Deiters' (2.83) and pillar cells (1.89).
Since induction in vivo generated a small number of labeled cells within the cochlear epithelium, we sought to induce broader recombination using cochlear explants. Cochleae were dissected at E14 and CreERT2 activity was induced by adding 4-hydroxytamoxifen (4-OH) to the culture media the same day. Explants induced by the addition of 100 nm 4-OH on E14 for 24 h showed levels of recombination similar to those obtained in vivo at E16. A significant number of mesenchymal cells were labeled as well as a limited number of cells within the cochlear epithelium (Fig. 3D). Consistent with in vivo results, analysis of Lrrn1+ cell fates in vitro indicated that just under 90% of all cells quantified (93) developed as medial cell types (Fig. 3E,F). To determine whether Lrrn1+ cell fates varied according to cochlear region, cell fates were re-analyzed based on their location along the cochlear spiral. In the basal region of the cochlea, over 93% of all Lrrn1+ cells developed with medial cell fates. Overall, the majority of Lrrn1+ cells developed as supporting cells, a result that contrasted with the observation that many of the LRRN1+ cells in the E14 cochlea appear to be hair cells. This difference is most likely a result of the fact that hair cells differentiate before support cells. It is also possible that the anti-LRRN1 antibody is less sensitive by comparison with either the smFISH or Lrrn1-CreERT2 fate mapping assays or that there is a delay between expression of mRNA and protein.
By comparison with the base, in the less mature mid and apical regions, the percentage of Lrrn1+ cells that developed as medial cell types decreased to ∼90% and 88%, respectively (Fig. 3F). Since the smFISH analysis indicated that all Lrrn1+ cells at E14 are located at the medial edge of the prosensory domain, these results suggest that some cell movement across the medial-lateral boundary of the OC may occur. However, this effect may also be a result of patterning defects that are known to occur in vitro.
Finally, to generate a more comprehensive understanding of the fates of Lrrn1+ cells, explants were established at E14 and induced for 24 h beginning on the equivalent of E15 with 1 μm 4-OH. This higher level of induction resulted in a labeled stripe (not quantified) of cells at the medial edge of the epithelium that includes nearly all IHCs, IPhCs and BCs (Fig. 3G). Yet while only sporadic Lrrn1+ cells developed with lateral cell fates (pillar cells, OHCs, Deiters' cells), multiple cells located in Kölliker's organ were also labeled (Fig. 3G,H). To determine whether this result could represent an artifact of the culture system or ectopic recombination because of infidelity in expression of the Lrrn1-CreERT2 transgenic, we examined Lrrn1 expression in a recently published single-cell RNAseq dataset for cochlear epithelial cells at E16 and found that Lrrn1 is expressed in a small number of cells located in Kölliker's organ (Kolla et al., 2020), and some signal is also detected in these cells by smFISH (Fig. 1D). Overall, these results are consistent with expression of Lrrn1 being largely, but not entirely, restricted to cells at the medial boundary of the developing organ of Corti, a bias that is stronger in the more differentiated basal region at any specific time point.
Lrrn1 expression is regulated through medial-lateral patterning
Expression of LRRN1 is mostly restricted to cells located at the medial boundary of the OC, suggesting that Lrrn1 expression could be regulated through specification of the medial-lateral cochlear axis. An early developmental gradient of BMP4, originating at the extreme lateral edge of the cochlear duct, is thought to act as a morphogen that specifies this axis (Ohyama et al., 2010). To determine whether LRRN1 expression is regulated through this patterning mechanism, cochlear explants were established at E13, and specification of the medial-lateral axis was disrupted by treating explants with CHIR99021 (CHIR), a GSK3 antagonist, that downregulates Bmp4 signaling in the cochlea (Ellis et al., 2019). Explants treated with CHIR develop multiple rows of IHCs, BCs, and IPhCs, in part through expansion of the medial OC (Ellis et al., 2019). Labeling of CHIR-treated explants with the HC marker anti-POU4F3 and anti-LRRN1 indicated an expansion in LRRN1 expression that paralleled its expansion in the IHC region (Fig. 4A–D). These results demonstrate that expression of LRRN1 is regulated by the same factors that specify the medial-lateral cochlear axis. Whether Lrrn1 is a direct target of Bmp4 signaling or is indirectly influenced by an intermediate factor remains to be determined.
Disruption of the medial-lateral boundary expands LRRN1 expression. A–D, Confocal images of cochlear explants established at E13 and maintained in 0.2% DMSO (control) or 2 μm CHIR99021 for 2 DIV, and then stained for LRRN1 and POU4F3. A, Control explant has mostly normal morphology of one to two rows of POU4F3+ IHCs. B, CHIR-treated explant has additional IHCs and a wide gap between IHCs and OHCs. C, View of the mid-apical region of the explant in A. POU4F3+ IHCs (bracket) are also LRRN1+ (arrows). D, Same as C for the explant in B. Many more POU4F3+ cells are found in the expanded IHC region (bracket), and there is a corresponding increase in LRRN1+ cells. Scale bar in A (same for B), 100 µm; C (same for D), 20 µm.
The medial boundary of the OC is disrupted in Lrrn1 mutant mice
To determine whether Lrrn1 plays a role in cochlear development, we used TALEN-mediated mutagenesis to target the 10th LRR repeat in Lrrn1. Three different lines were generated using different targeting sequences (see Materials and Methods), resulting in 2-, 4-, or 31-bp deletions within the Lrrn1 gene. The resulting frameshifts and truncations are predicted to generate large N-terminal extracellular fragments which likely act as loss-of-function mutations (Fig. 5A,B). All three lines showed phenotypically and statistically similar cochlear phenotypes so data and images from all three lines were merged for experimental analysis, and mutants are referred to as Lrrn1−/−. Immunolabeling of cochleae from embryos with either the 2- or 4-bp Lrrn1 deletion with anti-LRRN1 antibody indicated no labeling, consistent with an absence of full-length LRRN1 (Fig. 5). At E14, LRRN1 staining in the base of a Lrrn1+/− cochleae co-localizes with POU4F3 in a single row of developing IHCs, but no LRRN1 signal is apparent in the Lrrn1 mutant (Fig. 5C–F). Similarly, at E15, LRRN1 protein is detected in a Lrrn1+/− section at the medial edge of the sensory epithelium, but not in a Lrrn1−/− section (Fig. 5G–J). The lack of detectable anti-LRRN1 staining in Lrrn1−/− tissue suggests these truncating mutations generate null alleles of Lrrn1.
Lrrn1−/− male and female mice are viable and fertile with no obvious gross developmental defects. Lrrn1+/− heterozygous animals showed no differences from wild-type and, therefore, were used as littermate controls. Examination of the cochleae from Lrrn1−/− and Lrrn1+/− mice at P0 indicated a significant increase in the number of IHC doublets along the length of the epithelium in Lrrn1−/− animals (Fig. 6A). To determine whether these additional IHCs represented a phenotypic conversion of adjacent supporting cells into hair cells or resulted from a change in the boundary between sensory and nonsensory regions of the cochlear duct, IPhCs and BCs were labeled with anti-FABP7. We observed an increased number of IPhC/BCs, suggesting a change in the sensory/nonsensory boundary (Fig. 6B). Quantification of the number of additional IHCs and IPhC/BCs in basal, mid, and apical regions along the cochlear spiral indicates significant increases in both cell types at each position (Fig. 6C). The Lrrn1 mutant phenotype is more pronounced in the apical region, an effect that may be attributed to the late maturation of the apical OC and its corresponding susceptibility to developmental mutations (Fig. 6C).
Patterning defects, including IHC doublets, can also occur because of inhibition of cochlear outgrowth. To determine whether the phenotype observed in Lrrn1 mutants is a result of outgrowth defects, the density of normally aligned IHCs and OHCs and the total length of the cochlear duct were determined for Lrrn1+/− and Lrrn1−/− cochleae at P0 (Fig. 6D,E). No significant changes in either cochlear length or hair cell density were found, suggesting that cochlear outgrowth is unaffected in the absence of Lrrn1. Finally, we examined patterning of pillar cells and Deiters' cells in Lrrn1−/− cochlea, both of which were indistinguishable from control (Fig. 6F,G).
To determine when the initial appearance of the patterning defect occurs in Lrrn1 mutants, E14 cochleae were labeled with anti-DLL1, an early marker of IHCs, and CDKN1B, a marker of the prosensory domain. In controls, a single row of DLL1+ cells was present at the medial edge of the prosensory domain (Fig. 6H). In contrast, in Lrrn1−/− cochleae, doublets of IHCs were already present (Fig. 6H, arrows). These results suggest that specification of the medial boundary of the prosensory domain is disrupted early in Lrrn1 mutant OC development.
In some cases, additional medial OC cells can arise because of a delay in cell cycle exit of prosensory cells (Prajapati-DiNubila et al., 2019). To determine whether an increase in proliferation contributed to the formation of ectopic IHCs and SCs in Lrrn1 mutants, we examined the expression of the proliferation marker Ki67 at E14, the earliest stage that Lrrn1 expression is detected. Ki67 should normally be excluded from the prosensory domain as cells exit the cell cycle between E12 and E14 (Chen and Segil, 1999). Within the prosensory domain, marked by overlapping expression of CDKN1B and SOX2 (Driver et al., 2017), we found a total of 4 Ki67+ cells in 8 Lrrn1+/− cochleae, and 4 Ki67+ cells in 10 Lrrn1−/− cochleae. While we cannot rule out that a modest increase in proliferation may contribute to the extra medial cells observed in Lrrn1 mutants, it does not appear to be a significant factor.
TALEN-mediated mutagenesis of Lrrn1 results in loss of LRRN1 protein. A, Schematic of the protein structure of LRRN1. LRRN1 contains leucine-rich N-terminal and C-terminal domains (LRRNT and LRRCT, blue boxes), 12 leucine-rich repeats (LRR, black boxes), an immunoglobulin domain (Ig, orange crescent), a fibronectin Type III domain (Fn3, green box), a transmembrane domain, and a short C-terminal cytoplasmic tail. The 10th LRR was targeted for TALEN-mediated mutagenesis (arrow). B, Deletions in the genomic sequence of Lrrn1 of 2, 4, or 31 bp. All three deletions result in frameshifts predicted to generate truncated LRRN1 proteins. C, D, Confocal Z-stack projections of the basal region of E14 WT cochleae. C, Surface view of LRRN1+ cells in a Lrrn1+/− cochlea forming a single row, overlapping with developing HCs marked by POU4F3 (bracket). D, Orthogonal view from C. LRRN1 is localized to the luminal surface of the cell (arrow), above the POU4F3+ nucleus (arrowhead). E, F, Similar views to C, D from a Lrrn1−/− cochlea. POU4F3+ cells are still present (bracket, arrowhead), but LRRN1 signal is absent (open arrow). G–J, Confocal images of cryosections through Lrrn1+/− or Lrrn1−/− cochlear ducts at E15. G, LRRN1 staining is found in the OC and the spiral ganglion (arrows). H, LRRN1 at the base appears to be expressed in a single cell (arrow) at the medial edge of the OC (bracket), marked by CDH1 and CDKN1B (H'). I, No anti-LRRN1 staining is found in the OC of a Lrrn1 mutant cochlea, but faint signal is present in the spiral ganglion, probably representing background staining, which may contribute to the LRRN1 signal found in Figure 2E. J, No LRRN1 signal is present in a basal section of a Lrrn1−/− cochlea. Scale bar in E (same for C), 20 µm; F (same for D), 10 µm; I (same for G), 200 µm, J (same for H), 50 µm.
The medial boundary of the organ of Corti is disrupted in Lrrn1 mutants. A, Surface images of the apical (top) and basal (bottom) regions of the OC from Lrrn1+/− and Lrrn1−/− mice at P0. Lrrn1+/− cochleae have a single row of IHCs (MYO7A, magenta) with surrounding IPhCs (FABP7, green). Two ectopic IHCs (arrows) are present in the apical region. Lrrn1−/− cochleae show a significantly increased number of ectopic IHCs (arrows) in both apical and basal regions, and the vast majority of IHCs are surrounded by ectopic IPhCs. B, Cross-sections through the OC from the basal region of Lrrn1+/+ and Lrrn1−/− cochleae at P0. Asterisk in each upper panel marks the endogenous IHC. Note the presence of an additional IPhC (arrow in lower panel) adjacent to the ectopic IHC in the Lrrn1 mutant. C, The number of ectopic IHCs and IPhCs is significantly increased in all three regions of the cochlea in Lrrn1 mutants D, In contrast, the number of aligned IHCs and OHCs did not differ significantly between Lrrn+/− and Lrrn1−/− cochleae. N for each genotype: Lrrn1+/−, 9, Lrrn1−/−, 10 (IHCs, OHCs); Lrrn1+/−, 4, Lrrn1−/−, 4 (IPhCs). E, No significant difference in cochlear length between Lrrn1+/− (N = 9) and Lrrn1−/− (N = 10) cochleae. F, Surface views from the middle region of cochleae from mice with the indicated genotypes. Anti-NGFR labels inner pillar cells which appear similar in both samples. G, Similar view as in F, but from cochleae labeled with anti-MYO7A and anti-PROX1 which labels nuclei of pillar and Deiters' cells. Plane of focus is at the level of the SC nuclei. PROX1 expression appears comparable between Lrrn1+/− and Lrrn1−/−. H, Labeling with anti-DLL1 (magenta), an early marker of IHCs, in the base of the cochlea at E14. In control a single line of Dll1+-IHCs is present while in the Lrrn1 mutant several ectopic IHCs are already present (arrows). ****p < 0.0001, **p < 0.01, Scale bars for all, 25 µm.
Lrrn1 genetically interacts with Notch1 to mediate medial boundary formation
The disrupted medial boundary phenotype in Lrrn1 mutant cochleae strongly resembles the cochlear phenotype described in response to mild disruption of Notch signaling during the same developmental period (Basch et al., 2016). Moreover, Lrrn1 has been reported to interact with the Notch signaling pathway in other developing systems (Andreae et al., 2007; Tossell et al., 2011). These results suggest that the cochlear phenotype in Lrrn1 mutants could be a result of modulation of the Notch pathway during medial boundary formation. To test this hypothesis, we generated an allelic series of different Notch1 and Lrrn1 deletions. In agreement with previous results (Zhang et al., 2000), Notch1+/− cochleae showed no significant increase in IHC doublets at P0 (Fig. 7A,B). IHC doublets were also not significantly increased in Lrrn1+/−; Notch1+/− cochleae, despite the increased variability of the phenotype (Fig. 7A,B). Consistent with the findings described for the Lrrn1 mutant, a significant increase in IHC doublets was observed in Lrrn1−/−; Notch1+/+ cochleae compared with Lrrn1+/− littermates. However, the number of doublets was further, and statistically significantly, increased in Lrrn1−/−; Notch1+/− animals relative to Lrrn1−/−; Notch1+/+ mice (Fig. 7A,B). These results are consistent with a genetic interaction between Notch1 and Lrrn1 and suggest that Lrrn1 normally acts to enhance Notch1 signaling.
Lrrn1 interacts genetically with Notch1 to mediate medial boundary formation. A, Representative surface images of the middle turns from P0 cochleae with the indicated genotypes. HCs are labeled with anti-MYO7A and SCs are labeled with anti-FABP7. Increased IHC doublets and corresponding supporting cells are observed in Lrrn1−/−; Notch1+/+ and Lrrn1−/−; Notch1+/− genotypes. B, Number of extra IHCs per genotype, calculated per mm based on the total number of extra IHCs in the middle region of each sample. The Lrrn1 genotype underlies greater than half (51.6%) of the total variation between samples while only a small fraction of this variation may be attributed to the Notch1 genotype (3.90%). The mean number of extra IHCs does not differ significantly between Lrrn1+/+ (0.64, 1.45) and Lrrn1+/− (4.22, 4.37) cochleae in the Notch1+/+ versus Notch1+/− background. Heterozygosity at the Notch1 locus induces a significant increase in the mean number of IHC doublets in Lrrn1 mutants (8.37–14.7). C, In addition to increased IHC doublets, Lrrn1−/−; Notch1+/− cochleae also have a significant increase in endogenous IHCs (157–179). D, Surface view of cochleae from the indicated genotypes. Circles illustrate decreased endogenous IHC surface area in Lrrn1−/−; Notch1+/− cochleae, possibly underlying the increased number of cells. E, F, Quantification of endogenous IHC luminal surface area and hair cell packing indicate a significant reduction in cell surface area (4.74 ± 0.729 µm2) and a significantly greater density (29.9 ± 6.41 µm/40 cells) of IHCs in Lrrn1−/−; Notch1+/− cochleae. ****p < 0.0001, ***p < 0.001, **p < 0.01, N for each genotype: (Notch1+/+ background) Lrrn1+/+, 11, Lrrn1+/−, 9, Lrrn1−/−, 10; (Notch1+/− background) Lrrn1+/+, 5, Lrrn1+/−, 14, Lrrn1−/−, 10. Scale bar in A, 50 µm; D, 25 µm.
If Lrrn1 does augment Notch signaling, then another potential phenotype in Lrrn1−/−; Notch1+/− cochleae would be an increase in the number of IHCs aligned directly adjacent to pillar cells, as Notch1 has been shown to inhibit IHC formation (Lanford et al., 1999; Kiernan et al., 2005). To test this hypothesis, the number of aligned IHCs was determined for each genotype in the allelic series. In accordance with the expected reduction of Notch1 activation along the medial border, results indicated a significant increase in the number of aligned IHCs only in Lrrn1−/−; Notch1+/− cochleae (Fig. 7C), while the length of the cochleae was statistically unchanged (data not shown). Higher magnification of labeled hair cells revealed noticeably smaller IHC surface areas in Lrrn1−/−; Notch1+/− animals (Fig. 7D), a result that was confirmed quantitatively (Fig. 7E,F). Overall, these phenotypes support the hypothesis that decreased Notch1 signaling in the absence of Lrrn1 leads to both increased boundary disruptions and IHC formation.
Absence of Lrrn1 enhances the effects of γ-secretase inhibition in cochlear explants
To examine the regulation of Notch signaling by Lrrn1 more directly, we asked whether the absence of Lrrn1 leads to increased sensitivity to the potent Notch inhibitor LY411575 (Mizutari et al., 2013). Cochlear explants from WT mice were established at E14 and treated with different dosages of LY411575 to determine the maximum concentration of LY411575 that did not cause an increase in IHC numbers relative to untreated controls (500 pm; Fig. 8A,B). Next, cochlear explants from Lrrn1+/+, Lrrn1+/−, and Lrrn1−/− mice were established at E14 and treated with 500 pm LY411575 for 2 d in vitro (DIV), equivalent to E14–E16, then maintained for an additional three DIV. An increased frequency of IHC doublets was observed in Lrrn1−/− explants treated with LY411575 relative to both control explants treated with LY411575 and to untreated Lrrn1−/− cochleae (Fig. 8A,B). Quantification of IHC density indicated a significant increase in Lrrn1−/− explants treated with LY411575, relative to both untreated Lrrn1−/− explants and to Lrrn1+/+ explants treated with LY411575 (Fig. 8C).
Lrrn1 mutants show disruptions to Notch signaling. A, Surface views of the middle regions from cochlear explants established on E14 and maintained in 0.2% DMSO (control) or 500 pm LY411575 (LY) for 2 DIV, then in control media for an additional 3 DIV. MYO7A+ HCs, magenta, FABP7+ IPhCs, green. Occasional IHC doublets (arrows) are observed in Lrrn1+/+ explants treated with LY (A') and in Lrrn1−/− control explants (B). However, significantly more doublets are observed in LY-treated Lrrn1−/− explants (B'). C, Density of both aligned and ectopic IHCs for cochlear explants maintained in the indicated conditions. Density in Lrrn1+/+ or Lrrn1+/− explants treated with LY was unchanged from control explants of the same genotype (p = 0.99, 0.19), but was significantly increased in LY-treated Lrrn1−/− explants versus untreated Lrrn1−/− explants. LY treatment of Lrrn1−/− explants induced a significant increase in IHC density by comparison with control or LY-treated wild-type explants (****p < 0.0001, n.s., not significant). N for each genotype, both conditions: Lrrn1+/+, 10, Lrrn1+/−, 11, Lrrn1−/−, 9. D, E, Confocal Z-stack projection of the surface view of whole-mount smFISH for Mfng (magenta) and Lfng (green) in the basal region of E15 cochleae. No difference was observed between Lrrn1+/− and Lrrn1−/−. F, G, Sections of the E15 cochlear base showing JAG1 staining (green) in the differentiating OC, also marked by CDKN1B (magenta). JAG1 appears similar in Lrrn1+/− (F) and Lrrn1−/− (G) cochleae. F', G' JAG1 in SCs in the developing OC (grayscale). H, I, smFISH for Hes1 (green) shown in a confocal Z-stack projection near the surface of the cochlear base at E15. Differentiating HCs are stained with anti-MYO6 (magenta). The intensity of the lateral Hes1 signal (H', I', arrows) is equivalent in Lrrn1−/− to wild-type, but medial Hes1 is reduced (brackets). H'', I'', Boxes showing the ROIs used to compare the intensity of the Hes1 smFISH signal. L, lateral; M, medial. J, Average pixel intensity of medial Hes1 signal, normalized to the paired lateral ROI. Medial Hes1 is significantly reduced in Lrrn1−/− cochleae (**p = 0.002). N = 8 paired (medial + lateral) ROIs from 4 cochlea of each genotype. Scale bar in B (same for A), 25 µm; E (same for D), 20 µm; G, G' (same for F, F'), 25 µm, I (same for H), 20 µm.
As a final examination of the effects of Lrrn1deletion on Notch signaling, smFISH was used to examine the expression of Notch pathway genes. In mice with deletion of both Manic fringe (Mfng) and Lunatic fringe (Lfng), a similar duplication of medial cochlear sensory cells has been observed (Basch et al., 2016). We did not detect any difference in expression of either Mfng or Lfng in E15 Lrrn1−/− cochleae compared with Lrrn1+/− (Fig. 8D,E), or in the Notch ligand JAG1 (Fig. 8F–G'). Deletion of the transcription factor Hes1, a Notch target and effector, also leads to an expansion of IHCs and medial SCs (Zine et al., 2001; Tateya et al., 2011). At the base of the E15 cochlea, Hes1 is normally expressed in a strong, narrow stripe lateral to the OHCs, and a more diffuse domain medial to the IHCs (HCs labeled with MYOSIN6; Fig. 8H). While lateral Hes1 expression was unchanged in Lrrn1−/− cochleae, the medial Hes1 signal appeared to be reduced (Fig. 8I). To quantify the change in Hes1, we compared the ratios of medial to lateral Hes1 intensity in Lrrn1+/+ and Lrrn1−/− samples and found that medial Hes1 intensity was significantly reduced in Lrrn1 mutant cochleae (Fig. 8J). Overall, these results are consistent with decreased Notch activation in Lrrn1 mutants, suggesting that from its position at the medial boundary, Lrrn1 typically enhances Notch signaling in adjacent nonsensory cells, an effect that may be mediated through Hes1 expression.
Discussion
Straight cellular boundaries within populations of epithelial cells often specify signaling centers that influence unique cell fates on one or both sides of the boundary (Wurst and Bally-Cuif, 2001; Wang and Dahmann, 2020). The OC contains at least three boundaries: a medial boundary between the single rows of IHCs/IPhCs/BCs and adjacent nonsensory cells (Basch et al., 2016), a similar lateral sensory/nonsensory boundary between the third row of Deiters' cells and nonsensory cells (Ohyama et al., 2010), and a boundary between IHCs and pillar cells. Two of these, the medial sensory/nonsensory and the IHC/pillar cell boundaries, are characterized by a straight, highly organized alignment of unique cell types on each side (Driver and Kelley, 2020). The lateral interface between sensory and nonsensory cells, while also organized, is less regular and can feature deviations in alignment. Our understanding of how these boundaries are formed remains limited.
A likely first step in the creation of the OC boundaries is establishment of the medial-lateral axis of the cochlear duct. BMP4 expression in the lateral region of the cochlea has been proposed to create a morphogen gradient that specifies cellular identities along this axis (Ohyama et al., 2010; Munnamalai and Fekete, 2016). Deletion of the BMP receptors Alk3 and Alk6 within the duct leads to an expansion of Kölliker's organ at the expense of the outer sulcus (Ohyama et al., 2010). While morphogen gradients play key roles in specifying differing cellular identities (Bier and De Robertis, 2015; Irizarry and Stathopoulos, 2021), the dynamic nature of diffusive processes, especially at increased distance from the source, makes precisely positioning a straight boundary from a morphogen gradient alone difficult (Lander, 2007). Counter gradients of FGF10 or an unidentified WNT may act with BMP4 to improve spatial resolution, but data to support this hypothesis have been inconsistent (Urness et al., 2015; Munnamalai and Fekete, 2016; Ellis et al., 2019).
Work from Basch et al. (2016) examined cell-cell interactions at the medial prosensory/nonsensory boundary. Based on transient expression of Atoh1, Jag2, Dll1, Lfng, and Mfng in cells located along this boundary, they suggested that a peak of hair cell-inducing factors at the medial edge of the prosensory domain directs upregulation of the above genes. Differences in Notch signaling then arise as a result of various cis(-) and trans-interactions of Notch ligands and receptors. The negative feedback loop between expression of Notch and Notch ligands leads to a difference in activity (higher in medial nonsensory cells, lower in lateral prosensory cells) which specifies the medial boundary through inhibition of prosensory induction (Hartman et al., 2010; Pan et al., 2010, 2013).
Consistent with this hypothesis, deletion of Lfng and Mfng, resulting in decreased Notch signaling to medial nonsensory cells across the boundary, causes duplication of both IHCs and IPhCs (Basch et al., 2016). The increase in both cell types is key, as it distinguishes a change in the boundary from a defect in lateral inhibition, in which the increase in hair cells comes at the expense of supporting cells (Kiernan et al., 2005). Similar defects in medial boundary formation are observed in Jag1 heterozygotes and in other mutants that decrease, but do not abolish, Notch signaling (Kiernan et al., 2001; Zine et al., 2001; Brooker et al., 2006). These results demonstrate roles for Notch1, Jag1, Lfng, and Mfng in boundary formation. However, a central premise of the hypothesis proposed by Basch et al. (2016) is that Lfng and Mfng enhance signaling by the Dll1 ligand. Two observations argue against an exclusive role for Dll1 in boundary formation. First, Dll1 is only expressed in hair cells, while the boundary includes hair cells and supporting cells. Second, while the phenotype has not been thoroughly examined, images of the medial boundary from Dll1 mutant mice seem more consistent with impaired lateral inhibition rather than disrupted boundary formation. Dll1 mutants include examples of what appear to be IHCs touching each other (Kiernan et al., 2005; Brooker et al., 2006), and Jag1 labeling in Dll1 mutant cochleae did not indicate an increased number of IPhCs or BCs (Brooker et al., 2006). These results suggest that other factors interact with or directly modulate Notch activity in cells on one side or the other of the medial boundary.
In this study, we demonstrate that the timing and location of Lrrn1 expression coincides with the formation of the medial boundary (Basch et al., 2016). Lrrn1 has been shown to regulate Notch activity across boundaries in developmental and cancerous contexts (Tossell et al., 2011; Y. Zhang et al., 2021) and our data imply a similar role in the cochlea. Disruptions in medial boundary formation occur in response to weakened, not absent, Notch signaling (Basch et al., 2016). Using multiple approaches, we demonstrated that the loss of Lrrn1 decreases Notch signaling, leading to duplications at the medial boundary. As illustrated in Figure 9, these results suggest that Lrrn1 expression in developing prosensory cells likely acts to augment Jag1-mediated induction of the Notch target gene Hes1 in adjacent nonsensory cells, an interaction that is supported by recent findings from a study of pancreatic carcinoma (Y. Zhang et al., 2021). However, while unlikely, we cannot completely rule out the possibility that Lrrn1 acts via a separate, parallel pathway that simultaneously regulates medial boundary formation. The demonstrated decrease in Hes1 is consistent with modulation of Notch signaling, but whether this is the primary effector of Lrrn1 signaling could only be demonstrated definitively by restoring Hes1 expression to WT levels in a Lrrn1 mutant background.
Model of cellular interactions at the developing cochlear medial boundary. Left panel, Before Lrrn1 expression at E14, cells on both side of the boundary express roughly equal levels of Notch1 (N1) and the Notch ligand Jagged1 (J1) leading to roughly equivalent levels of Notch activation and expression of the Notch target Hes1. Right panel, Because of differences in positional identity, cells on one side of the boundary begin to express Lunatic Fringe (Lfng), Manic Fringe (Mfng), and Lrrn1 (L1). Lfng and Mfng glycosylate Jagged1, leading to decreased autocrine Notch signaling, while expression of Lrrn1 enhances Notch activation in neighboring cells. This results in heterogeneous levels of Notch pathway activation. More medial cells are prevented from assuming a sensory cell fate by the increased level of Notch activity.
The mechanisms by which Lrrn1 enhances Notch activity in neighboring cells is unclear. Previous studies examining the role of Lrrn1 during formation of the chick midbrain/hindbrain boundary demonstrated that Lrrn1 induces autocrine expression of Lfng (Tossell et al., 2011), prompting Lfng-mediated glycosylation and enhancement of transcellular Notch signaling (Panin et al., 1997; Brückner et al., 2000; Moloney et al., 2000). However, no change in expression of Lfng or Mfng was observed in Lrrn1−/− cochleae. An alternative mechanism is through a direct interaction of Lrrn1 with Notch or Notch ligands in the same or adjacent cells. Lrrn1 is a transmembrane protein with an extracellular leucine-rich repeat domain (Taguchi et al., 1996; Andreae et al., 2007) and the Drosophila homologs Tartan and Capricious have been shown to confer differential adhesion through transcellular binding (Milán et al., 2001; Mao et al., 2008). Therefore, interactions with Notch, Notch ligands, or other extracellular molecules could lead to modulation of Notch signaling. While Notch1 has not been reported to interact physically with Lrrn1 or other leucine-rich repeat molecules, a small leucine-rich proteoglycan, Biglycan, has been shown to bind to Notch3 (X. Zhang et al., 2015), suggesting possible physical interactions that could modulate Notch signaling.
The restricted expression of Lrrn1 to a narrow row of cells within the cochlea is striking and suggests precise regulation. In Drosophila, Tartan and Capricious are regulated by Apterous (Lhx2; Milán and Cohen, 2003; Milán et al., 2005). However, neither Lhx2, nor the related Lhx9, have been reported to be expressed in the developing cochlea. In another context, Lrrn1 expression was shown to be controlled through binding of Mycn, a bHLH transcription factor, to two E-box motifs in its promoter and first intron (Hossain et al., 2008), so Lrrn1 may be directly regulated by Atoh1 or other bHLHs. Atoh1 expression is not restricted to the medial boundary, though, so additional regulatory mechanisms would be required. Another possible regulatory mechanism for Lrrn1 expression is through the BMP4 gradient thought to specify the cochlear medial-lateral axis. A direct response requiring precise interpretation of the BMP4 gradient seems unlikely, suggesting instead that Lrrn1 expression is regulated through establishment of the medial-lateral cochlear axis. Finally, as noted by Basch et al. (2016), differentiation of the OC requires inductive signals from mesenchymal cells located on the opposite side of the cochlear basilar membrane (Montcouquiol and Kelley, 2003; Doetzlhofer et al., 2004). These yet unidentified inductive signals could also play a role in regulation of Lrrn1 expression.
In summary, the data presented here demonstrate that the leucine-rich repeat protein Lrrn1 is expressed at the medial boundary of the OC beginning around E14. Deletion of Lrrn1 leads to defects in maintenance of this prosensory/nonsensory boundary resulting in ectopic IHCs and surrounding supporting cells. The effects of Lrrn1 are mediated through enhancing Notch signaling across this boundary, as the effects of mild inhibition of Notch signaling are enhanced in a Lrrn1 mutant background and medial expression of the Notch target gene Hes1 is downregulated in Lrrn1 mutants. While the factors that limit expression of Lrrn1 to the medial edge of the OC are unknown, the precise nature of this expression pattern suggests that cellular patterning and axial specification within the cochlear duct have already occurred by E14. Finally, the precise expression of Lrrn1 and Lfng along this boundary strongly suggests the existence of a process of cellular resolution that extends beyond a single morphogen gradient. Subsequent studies examining the factors the regulate Lrrn1 expression in the cochlea could provide valuable insights regarding axial specification and cell fate determination.
Footnotes
This work was supported by the National Institute on Deafness and Other Communication Disorders Division of Intramural Research Grant DC000059 (to M.W.K.). We thank the staff in the Porter Neuroscience Shared Animal Facility for outstanding animal care and Dr. Weise Chang for technical assistance. We also thank Dr. Basile Tarchini and Dr. Doris Wu for reading an earlier version of this manuscript and the staff at the Frederick Transgenic Mouse Model Laboratory for assistance in the generation of Lrrn1 mutant mice.
The authors declare no competing financial interests.
- Correspondence should be addressed to Elizabeth Carroll Driver at drivere{at}nidcd.nih.gov or Matthew W. Kelley at kelleymt{at}nidcd.nih.gov
