Hearing requires the transduction of vibrational forces by specialized epithelial cells in the cochlea known as hair cells. The human ear contains a finite number of terminally differentiated hair cells that, once lost by noise-induced damage or toxic insult, can never be regenerated. We report here that sphingosine 1-phosphate (S1P) signaling, mainly via activation of its cognate receptor S1P2, is required for the maintenance of vestibular and cochlear hair cells in vivo. Two S1P receptors, S1P2 and S1P3, were found to be expressed in the cochlea by reverse transcription-PCR and in situ hybridization. Mice that are null for both these receptors uniformly display progressive cochlear and vestibular defects with hair cell loss, resulting in complete deafness by 4 weeks of age and, with complete penetrance, balance defects of increasing severity. This study reveals the previously unknown role of S1P signaling in the maintenance of cochlear and vestibular integrity and suggests a means for therapeutic intervention in degenerative hearing loss.
Lysophospholipids are an important class of extracellular signaling molecules that are involved in many aspects of normal mammalian physiology and pathophysiology. Sphingosine 1-phosphate (S1P) is a major lysophospholipid species that elicits effects in organisms as diverse as yeast, Dictyostelium, Drosophila, zebrafish, rodents, and humans (Saba et al., 1997; Kupperman et al., 2000; Li et al., 2001; Yang et al., 2002; Herr et al., 2003; Spiegel and Milstien, 2003). In vertebrates these responses are thought to be mediated predominantly by a family of five G-protein-coupled receptors (GPCRs) known as S1P1–5 (Anliker and Chun, 2004; Ishii et al., 2004; Saba and Hla, 2004). These receptors influence a range of biological functions including vascular development, angiogenesis, cancer growth, immunity, and cardiovascular and nervous system function (Gardell et al., 2006), but previously they have not been reported to be important for cochlear or vestibular function. S1P1–3 are expressed widely in the developing and adult organism. Functional studies have been facilitated greatly by the generation of receptor-null mice. Although s1p1−/− mice die embryonically from blood vessel malformation (Liu et al., 2000), s1p2−/− and s1p3−/− single-null mice show relatively mild phenotypes with no gross anatomical or behavioral abnormalities (Ishii et al., 2000, 2001; MacLennan et al., 2001).
To assess the functional redundancy of S1P2 and S1P3 signaling, we generated s1p2/s1p3 double-null mice (Ishii et al., 2002). Whereas s1p3 single-null mice displayed no readily detectable physiological defects despite a measurable reduction in S1P responsiveness of fibroblasts (Ishii et al., 2001), double-null mice were characterized by subtle but consistent phenotypes. There was a marked reduction in litter size and an increase in perinatal lethality among s1p2−/−/s1p3−/− pups. Additional analyses identified postural defects that led us to examine vestibular function. Here we report significant and progressive degeneration of auditory and vestibular organs that result in complete deafness and balance defects in s1p2−/− and s1p2−/−/s1p3−/− mice. Our data identify a novel function for receptor-mediated S1P signaling in inner ear hair cell maintenance and suggest that a loss of S1P signaling may be a factor in degenerative hearing loss in humans.
Materials and Methods
Mice were housed in ventilated cages in the vivarium at The Scripps Research Institute. s1p2/s1p3 double-null mice were generated and maintained as described (Ishii et al., 2002) in a 129/SvJ, C57BL/6N mixed background.
Mice were placed in a 1 or 2 L beaker three-fourths full of warm (30–32°C) tap water and observed for a 1 min test period. Mice that sank to the bottom of the beaker or were unable to keep their nose above the surface of the water for a period of 5 s were rescued and scored as a swim failure. In each group 5–17 mice were tested at each time point.
Mice were allowed to ambulate freely in an open field chamber for a 30 min test period. The chamber has infrared sensor beams placed at a height of 75 mm that are broken when the mice rear on their hind legs. Rears per minute reflect the number of beam breaks occurring in the center of the chamber during the time spent in this area; n = 9 mice aged 12–14 weeks in each group.
Subjects were placed in a standard 150 mm Petri dish so that each mouse was in contact with a flat surface both dorsally and ventrally. When the dish was inverted, wild-type mice sensed that they were upside-down and immediately righted themselves. Mice with vestibular lesions, however, became disoriented and often experienced a delay in righting behavior.
Acoustic startle response.
The acoustic startle test was performed by using a startle response system (SR-LAB apparatus, San Diego Instruments, San Diego, CA) as described previously (Risbrough et al., 2003). Subjects were restrained inside a Plexiglas tube with a 40 mm inner diameter in a ventilated, darkened chamber and then were exposed to clicks of defined, variable decibels at random intervals. Response was measured in arbitrary units with a force transducer and accompanying SR-LAB software. Responses were discarded when preceded by detectable motion in the restraint chamber before click stimulus. Then 7–15 individual mice, 4–12 weeks of age, were analyzed in each group. Wild-type or heterozygous littermate controls were included with all knock-out mice at the time of testing.
RNA was purified from the indicated tissues with a Qiagen RNeasy kit (catalog number 74104, Hilden, Germany) and was reverse transcribed with Superscript II (catalog number 18064-022, Invitrogen, Carlsbad, CA). Quantitative real-time reverse transcription-PCR (qRT-PCR) was performed in the presence of SYBR Green (catalog number S4438, Sigma–Aldrich, St. Louis, MO) with the use of a Rotor-Gene 3000 (Corbett Research, Sydney, Australia). Copy number was determined for each receptor by comparing the cycle threshold to a standard curve of known target concentrations and then standardized to a reference gene (β-actin). Nonquantitative RT-PCR was performed as described (Lee et al., 2006).
In situ hybridization
Pups aged 0–4 d were killed by isoflurane inhalation and decapitated. Heads were bisected, fixed overnight in 4% paraformaldehyde (PFA)/PBS, and cryoprotected in 30% sucrose. Cochleas from mice >7 d old were treated similarly, with the addition of a decalcification step for 2 d in 5% EDTA/4% PFA/PBS before cryoprotection. All tissues then were embedded in Tissue-Tek OCT compound (catalog number 25608-930, VWR, West Chester, PA), sectioned to 12 μm, and hybridized with gene-specific probes as described previously (Grillet et al., 2003). The open reading frames of mouse S1P2 and S1P3 cDNAs were subcloned into pBluescript SK (Stratagene, La Jolla, CA) and used as templates for RNA probe synthesis.
Mice were killed by deep anesthesia with isoflurane inhalation and then perfused transcardially with 4% PFA. Whole inner ear structures and surrounding tissue were removed intact, fixed overnight in Bouin's fixative, dehydrated in ethanol, decalcified overnight in 4% hydrochloric acid/4% formic acid, and neutralized in dilute ammonia. The tissues were embedded in paraffin, sectioned, and stained via routine techniques. Whole-mount surface preparation and staining of neonatal cochleas were performed as described previously (Senften et al., 2006).
S1P receptors are expressed in the murine cochlea
Nonquantitative RT-PCR analysis identified the expression of three S1P receptor subtypes in the murine cochlea (Fig. 1A). To identify any developmentally relevant changes in the expression of these genes, we performed qRT-PCR (Fig. 1B). Notably, S1P2 has a much higher level of expression when compared with S1P1 and S1P3 and is upregulated significantly at the later ages, with expression approximately sixfold greater at postnatal day 28 (P28) as compared with P0 (p = 0.0039). In situ hybridization suggests that cochlear expression of S1P2 is localized exclusively inside the organ of Corti, primarily in the supporting cells, with a weak signal sporadically detected in the inner and outer sensory hair cells (Fig. 1C) (supplemental Fig. S1, available at www.jneurosci.org as supplemental material). In contrast, S1P3 is expressed primarily outside the organ of Corti in a pattern that is consistent with the vasculature and possibly glial populations of the cochlea (Fig. 1E,F).
s1p2−/−/s1p3−/− mice have progressive loss of vestibular function
As we have reported previously, s1p2 null and s1p2/s1p3 double-null mice have no gross anatomical or behavioral differences from their heterozygous or wild-type littermates (Ishii et al., 2002). However, since that report we noted some subtle but consistent postural defects initially in older (> 6 months) s1p2−/−/s1p3−/− animals. These defects included a persistent head tilt (Fig. 2A) and defects in tail hanging posture (Fig. 2B). Because previous studies have suggested that such posturing is indicative of vestibular defects (Seoane et al., 2001; Paffenholz et al., 2004), we ran these mice through a series of behavioral tests to evaluate dexterity including forced swim, rearing behavior, and contact righting.
The forced swim test is a simple, discrete assay for vestibular function. Mice with significant vestibular lesions are unable to orient themselves spatially in the absence of tactile cues and therefore are unable to swim when placed into a beaker of water. All s1p2−/−/s1p3+/− and s1p2+/−/s1p3−/− animals were able to swim normally at all ages. Similarly, very young s1p2−/−/s1p3−/− mice (4 weeks) had no difficulty in this assay (Fig. 2C) (supplemental video S2, available at www.jneurosci.org as supplemental material). However, the double-null mice displayed a progressive loss of swimming ability and were almost uniformly deficient by 18 weeks of age. Additionally, s1p2−/−/s1p3−/− mice had a significant reduction in rearing behavior, consistent with an inability to balance on the hindlimbs (Fig. 2D). Double-null mice were also defective in contact righting (supplemental video S3, available at www.jneurosci.org as supplemental material). Differences in body size made quantitation of this assay unreliable, because smaller mice are exposed to additional tactile cues in the righting chamber. However, s1p2+/−/s1p3−/− mice invariably displayed normal righting behavior, but mature s1p2−/−/s1p3−/− mice always showed significant righting delays.
s1p2−/− mice do not respond to sound
Given the close relationship between cochlear and vestibular function, we evaluated the double-null mice for their sound-evoked sensorimotor response by measuring their acoustic startle response (ASR). Mice heterozygous for S1P2 exhibited a normal startle reflex, responding to acoustic stimuli >90–100 dB. This response was the same both in the presence and absence of S1P3 (Fig. 2E). In contrast, all s1p2−/−/s1p3+/− and s1p2−/−/s1p3−/− mice completely lacked a startle reflex even at the youngest age that was tested (4 weeks).
Hair cells develop normally in s1p2−/−/s1p3−/− mice
To determine whether the observed behavioral defects were the result of improper hair cell development, we assessed neonatal mice for morphological abnormalities in the inner ear. Examination of hematoxylin and eosin-stained sections showed that both the cochlea and vestibular end organs were normal in appearance until at least 14 d of age (Fig. 3A–D). To assess more accurately the morphology of the cochlear hair cells, we labeled whole-mount cochlear tissue with FITC-phalloidin to reveal actin-rich stereocilia (Fig. 3E,F). The overall patterning of the sensory epithelium and the polarity and organization of the stereociliary bundles were indistinguishable between the wild-type and knock-out animals at P4. Furthermore, expression of cadherin 23 (CDH23), a marker of differentiating hair cells, was undisturbed in the knock-out mice (Fig. 3G,H), indicating that the hair cells had undergone normal differentiation (Siemens et al., 2004).
s1p2−/−/s1p3−/− mice have progressive degeneration of inner ear structures
After a histological examination of older knock-out mice (>4 weeks) we observed significant defects in both vestibular end organs and cochleas. Although young mice appeared to be grossly normal, there was profound disorganization of the sensory epithelia of the vestibular maculas in older double-null mice (Fig. 4A,B). We noted a marked decrease of afferent calyces, complete loss of stereocilia, and general disorganization of the epithelial sheet. These defects became pronounced by 4 months of age.
Similarly, structures in the cochlea displayed a progressive, age-dependent increase in severity of the double-null phenotype. Until 2 weeks of age the cochleas appeared to be completely intact. Shortly thereafter, we observed degeneration of the organ of Corti, characterized by loss of inner and outer hair cells and all of their supporting cells (Fig. 4C,D). This degeneration proceeded in a progressive basal-to-apical pattern and was nearly complete by 4 months of age. In addition, there was a striking loss of afferent neurons in the spiral ganglia, with a similar basal-to-apical pattern of degeneration (Fig. 4E,F).
S1P2 has been shown to mediate a number of diverse responses in a variety of cell types (for review, see Ishii et al., 2004; Gardell et al., 2006), but to date no reports implicate this receptor in cochlear or vestibular function. This study identifies a previously unknown role of S1P2 and adds to a growing list of biological functions mediated by S1P signaling.
Our previous work with s1p2 null and s1p2/s1p3 double-null mice revealed significant perinatal lethality of s1p2−/− pups, but survivors developed into grossly normal adults in a 129/SvJ, C57BL/6N background (Ishii et al., 2002). After backcrossing mice into a pure C57BL/6J (>N6), we occasionally observed mortality in young adults along with rare seizure activity, consistent with reports of an independently generated s1p2 knock-out line (MacLennan et al., 2001).
Here we report that s1p2−/−/s1p3−/− mice also are characterized by progressive basal-to-apical degeneration of cochlear hair cells, resulting in complete loss of ASR before 4 weeks of age. It is possible that an absence of ASR could be attributable to defects in the CNS or motor function; however, this is unlikely because s1p2−/−/s1p3−/− mice exhibit normal ambulation and behavior except as noted above. The histological abnormalities described here are consistent with profound hearing and vestibular dysfunction. This phenotype is degenerative, rather than developmental, because the fine bundle organization of stereocilia on the top of hair cells is morphologically normal at P4 (Fig. 3I,J) and the cochlear histology appears to be undisturbed until at least 2 weeks of age. In addition, functional and anatomical defects also are found in the closely related hair cells of the vestibular maculas.
The mRNA expression patterns of S1P2 and S1P3 suggest indirect mechanisms leading to the loss of sensory hair cells in the knock-out animals. An absence of S1P2 may result primarily in loss of function of the cochlear-supporting cells that appear to be important for survival of sensory hair cells (Ramirez-Camacho et al., 2004) and afferent neurons (Sugawara et al., 2005). This defect may be exacerbated by compromised normal cochlear blood flow via loss of S1P3 in the vascular bed and/or abnormal glial function. The vascular role of S1P3 is supported by the observation that S1P may regulate blood flow to the cochlea (Scherer et al., 2006). It is interesting to note that, for both the cochlear and the vestibular systems, additional data (data not shown) suggest that behavioral deficits precede overt hair cell degeneration. This may indicate indirect effects of S1P signaling on hair cell function before the cells degenerate, which may provide a therapeutic window for intervention.
While this manuscript was in preparation, another group described nearly identical vestibular and cochlear phenotypes in an independently generated s1p2 null mouse in the absence of the S1P3 mutation (MacLennan et al., 2006). Notably, in both cases s1p2−/− mice demonstrated profound deafness at the youngest ages that were tested and progressive degeneration of the organ of Corti and spiral ganglia. A major difference between the two lines is penetrance of the vestibular phenotype (43 vs 100% observed here), likely explained by the concomitant loss of S1P3, which exacerbates the phenotype.
Overall, these data support the involvement of S1P receptor-mediated signaling in hair cell function and survival. This finding raises the possibility that selective modulation of S1P receptors may prevent degeneration associated with exposure to noise or ototoxic drugs such as cisplatin, furosemide, or aminoglycoside antibiotics. A number of compounds with varying specificities for the five known S1P receptors are currently under investigation for other clinical indications and may be applicable to human diseases of cochlear and vestibular origins (Hla, 2004; Gardell et al., 2006).
This work was supported by National Institutes of Health Grants MH051699, MH01723, NS048478, and HD050685 (J.C.). We thank Christine Paczkowski, Danielle Letourneau, and Suzanne Peterson for a critical review of this manuscript and Anna Sczaniecka for expert technical assistance.
- Correspondence should be addressed to Jerold Chun, Department of Molecular Biology, The Scripps Research Institute, 10550 North Torrey Pines Road, La Jolla, CA 92037.