Diverse Expression Patterns of LIM-Homeodomain Transcription Factors (LIM-HDs) in Mammalian Inner Ear DevelopmentMingqian Huang,1 Cyrille Sage,2 Huawei Li,3 Mengquig Xiang,4 Stefan Heller,5 and Zheng-Yi Chen1
Development of the mammalian innerear begins with the otic placode forma-tion, through thickening of the ecto-derm adjacent to the developing hind-brain. Subsequently, the otic placodeinvaginates to form the otic vesicle,which further undergoes a series ofmorphogenetic changes, to ultimatelygive rise to a complex inner ear of sixsensory organs consisting of threecristae, two vestibular maculae, and acochlea (Bryant et al., 2002; Baraldand Kelley, 2004). Delaminating fromthe otocyst are neuroblasts that de-velop into auditory and vestibular
neurons during later stages. Withineach sensory organ, sensory patchesform initially from a population of pro-liferating progenitors that exit the cellcycle as postmitotic committed precur-sors. Production of hair cells, the sen-sory cells of the inner ear, is deter-mined by the function of Atoh1(Math1), a bHLH transcription factor(Bermingham et al., 1999; Zheng andGao, 2000; Kawamoto et al., 2003;Shou et al., 2003). Concomitantly, theNotch signaling pathway plays essen-tial roles in the patterning of the innerear sensory patch, establishing the ar-rayed configuration of hair cells sepa-
rated by supporting cells (Kelley,2006).
LIM-homeodomain transcriptionfactors (LIM-HD) are characterized bytwo zinc finger motifs: the LIM do-main for protein/protein interactions,and a homeodomain for binding to theDNA control elements of target genes.LIM-HDs have been studied exten-sively, in particular in spinal cord sub-type neuron specification and in pitu-itary gland development (Hobert andWestphal, 2000; Shirasaki and Pfaff,2002). For example, in the mouse spi-nal cord, Lhx3 is necessary for thegeneration of motor neurons, whereas
1Eaton-Peabody Laboratory, Massachusetts Eye and Ear Infirmary and Department of Otology and Laryngology, Harvard Medical School,Boston, Massachusetts2Bases Moleculaires et Cellulaires des Maladies Genetiques, Unite Inserm 654, 1er s/sol, Hopital Henri Mondor, Creteil, France3Otology Skull Base Surgery Department, Hearing Research Institute, Fudan University, Shanghai, PR China4Department of Pediatrics, Center for Advanced Biotechnology and Medicine, UMDNJ-Robert Wood Johnson Medical School, Piscataway,New Jersey5Department of Otolaryngology, Head and Neck Surgery, Department of Molecular and Cellular Physiology, Stanford University School ofMedicine, Stanford, CaliforniaGrant sponsor: NIH; Grant numbers: DC-04546, DC-06908, and DC-6167; Grant sponsor: DRF Foundation; Grant sponsor: Pfizer/AFARInnovation in Aging Research.*Correspondence to: Dr. Zheng-Yi Chen, Eaton-Peabody Laboratory, Massachusetts Eye and Ear Infirmary, 243 Charles St.,Boston, MA 02114. E-mail: [email protected]
DOI 10.1002/dvdy.21735Published online 15 October 2008 in Wiley InterScience (www.interscience.wiley.com).
co-expression of Lhx3 and Isl1 resultsin the production of interneurons(Thaler et al., 2002). Such combinato-rial functions of multiple LIM-HDswithin a particular cell population arewidely used in spinal cord neuron sub-type specification (Allan and Thor,2003). Increasingly, LIM-HDs arefound to be essential in defining stem/progenitor cells properties, such asthe role of Isl1 in the specification ofheart progenitor cells and Lhx2 in de-termining the fate of hair follicle stemcells (Laugwitz et al., 2005; Moretti etal., 2006; Rhee et al., 2006). TwelveLIM-HD members have been de-scribed in the mouse genome, andthey can be divided into six sub-groups based on their sequence simi-larities (Hunter and Rhodes, 2005).
We and others have previouslyshown that Isl1, a LIM-HD member,is expressed in the proliferating pro-genitor pools that give rise to the in-ner ear sensory patch (Li et al., 2004;Radde-Gallwitz et al., 2004). Lhx3 ex-pression has been found only in haircells (Sage et al., 2005). A spontaneousmutation in the murine Lmx1a genecauses deafness due to the lack of in-ner ear formation, indicating an es-sential role of Lmx1a during early in-ner ear development (Millonig et al.,2000; Chizhikov et al., 2006). How-ever, most LIM-HDs have not beenexamined for their expression profileand none has been linked to any path-way in the developing inner ear.
In this study, we systematically ex-amined expression of all LIM-HDs inthe developing mouse inner ear. Wefound that many LIM-HD genes areexpressed during inner ear develop-ment, with expression patterns thatencompass early progenitor cells, dif-ferentiating sensory epithelia, andganglion neurons. The expression pat-terns of LIM-HDs suggest that theyplay diverse roles in the developmentand function of multi-inner ear celltypes.
RESULTS
As the first step in elucidating poten-tial roles for LIM-HDs during innerear development, we systematicallyexamined their expression patterns indeveloping mouse inner ears usingRNA in situ hybridization. We chosestages from E10.5 to P6, encompass-
ing otocyst development, inner earcompartmentalization, hair cell fatespecification, and cell differentiation.For early developmental stages, weused ISL1 and PAX2 immunostainingto mark the presumptive sensory epi-thelia (Li et al., 2004; Radde-Gallwitzet al., 2004). In addition, antibodies toISL2 and LHX3 were used. For in situhybridization, we validated the ribo-probes for each LIM-HD gene by dis-tinct expression patterns in the brain.Lhx1, 2, 4, 5, 6, 9, and Lmx1b labeleddifferent areas of the medulla oblon-gata, and Lhx2, 4, and Lmx1b alsolabeled areas of the retina. Lhx2staining was strong in olfactory epi-thelium and Lhx8 was primarily ex-pressed in the first branchial arch andnotochord at E10.5 (data not shown).The expression patterns of theLIM-HD probes in brain matched pub-lished results (Hunter and Rhodes,2005). In the inner ear, fragments ofall the LIM-HD gene-coding regionswere amplified by RT-PCR (data notshown), yet Lhx1, 2, 4, and 8 were notdetected by in situ hybridization. It islikely that the expression levels of thefour LIM-HDs are below the thresholdof detection sensitivity by in situ hy-bridization.
At E10.5, the otocyst is composed ofprogenitor cells destined to becomesensory and non-sensory epithelialparts of the inner ear. Of the LIM-HDs examined, two of them, Isl1 andLmx1a, were distinctly expressed atthis stage (Fig. 1A,B). As shown pre-viously, labeling of ISL1 delineatedthe future sensory epithelium and oticganglions (OG), whereas PAX2-posi-tive domain defined the non-sensoryand a portion of the sensory epithe-lium (overlapping partially with Isl1)(Fig. 1A, bracket) (Burton et al., 2004;Radde-Gallwitz et al., 2004). At thisstage, expression of Lmx1a was prom-inently detected in the medial portionof the otocyst that overlapped almostcompletely with the PAX2-positive do-main, indicating that Lmx1a likelymarked the future non-sensory epi-thelium (Fig.1B). Unlike Isl1, Lmx1aexpression was not detected in the oticganglions. At E10.5, most otocyst cellsare proliferating, suggesting that Isl1and Lmx1a have roles in proliferatingsensory and non-sensory progenitorcells, in addition to their functions inpostmitotic cells (see below).
At E12.5, the otocyst undergoesmorphological changes with the for-mation of the immature vestibular or-gans. The future cochlea is not yetdeveloped and the primordial cochlearregion is mainly comprised of progen-itor cells exiting cell cycle (Chen et al.,2002). In the vestibular system, thesensory epithelium begins to differen-tiate into hair cells and supportingcells, by the appearance of nascenthair cells expressing Atoh1 andPou4f3, two hair cell–specific tran-scription factors (Xiang et al., 1998;Bermingham et al., 1999). ProminentISL1 labeling was restricted to a re-gion of differentiating sensory epithe-lium, which partially overlapped withPAX2 labeling (Fig. 1C, arrow). Ex-pression of additional LIM-HDs be-came detectable at this stage includ-ing Lhx3, 5, 9, and Lmx1b (Fig. 1D–F,H). Following hair cell specification byAtoh1, Pou4f3 is one of the earliesttranscription factors expressed in alldifferentiating hair cells (Xiang et al.,1998). To delineate the timing be-tween Lhx3 and Pou4f3 expression,double labeling was performed withanti-POU4F3 and anti-LHX3 antibod-ies. POU4F3 was detected in all haircells yet only a subset of them wasfound to produce LHX3 (Fig. 1D).Therefore, developmentally, the ex-pression of Lhx3 was likely down-stream of Pou4f3. Weak expression ofLhx5 and Lhx9 was detected in thedeveloping saccule outside the sensoryepithelial region and cochleovestibu-lar ganglions (CVG), with Lhx5 butnot Lhx9 detected in primordial co-chlea (Fig. 1E,F). Consistent withE10.5, Lmx1a was present only in thenon-sensory epithelial region of sac-cule, and absent in CVG or cochlea(Fig. 1G). Comparison between Figure1C and G showed that expression ofIsl1 and Lmx1a is mutually exclusive.Lmx1b was weakly expressed in CVGand saccule, but not in the cochlea(Fig. 1H). Among all the LIM-HDs ex-pressed at this stage, only Isl1 andLhx5 showed expression in the pri-mordial cochlear region (Fig. 1C,E).
By E14.5, the vestibular organs arefurther developed, with an increasednumber of differentiating hair cells.ISL1 was predominantly stained insupporting cells, with decreased pro-duction in hair cells (Fig. 1I), suggest-ing that Isl1 expression may no longer
3306 HUANG ET AL.
be required for future hair cell differ-entiation. ISL2 labeling started to ap-pear in hair cells, but was also foundweakly in supporting cells and stroma(Fig. 1J). Labeling of LHX3 was up-regulated and confined to hair cells(Fig. 1K), whereas Lhx5, 9, andLmx1b were weakly expressed in the
sensory epithelium, mainly in haircells (Fig. 1L,M,O). Lmx1a was signif-icantly upregulated in the non-sen-sory epithelium region including theendolymphatic duct (Fig. 1N).
At E16.5, LIM-HD expression pat-terns became more defined. In the ves-tibular system, ISL1 labeling was
most prominent in the supportingcells, and remained weak in some haircells (Fig. 2A). ISL2 was detected at alevel similar to E14.5, higher in haircells than in supporting cells orstroma (Fig. 2B). Prominent LHX3 la-beling was confined to hair cells (Fig.2C). Lhx5 expression showed an inter-
Fig. 1. Expression of LIM-HDs in the developing inner ear: E10.5 (A,B), E12.5 (C–H), and E14.5 (I–O). Each stage was labeled with sidebars. A: ISL1labeling in the region of future sensory epithelia (bracket) was distinct from primarily non-sensory PAX2 expression (red) in the medial portion of otocyst(Ot). OG, otic ganglions. B: In an adjacent section, prominent Lmx1a expression was in a region overlapping with PAX2 expression. C: At E12.5, ISL1expression was prominent in primordial cochlea (coch), cochleovestibular ganglions (CVG), and the developing saccular sensory epithelia, whichoverlapped slightly with PAX2 expression (arrow). D: Labeling with antibodies against LHX3 and POU4F3 showed immunoreactivities in nascentutricular hair cells. Most hair cells were LHX3 and POU4F3 double-positive (arrows to show example), whereas occasional early hair cells were onlypositive for POU4F3 (arrowhead). E–H: Expression of Lhx5, Lhx9, Lmx1a, and Lmx1b in the inner ear at E12.5. Among them, only Lhx5 was expressedin the primordial cochlea (Fig.1E). Lmx1a was detected in the non-sensory epithelial region, but undetectable in the sensory epithelium positive for ISL1(arrows, comparing to 1C) or in the cochleovestibular ganglions (Fig.1G). The sensory epithelial region of saccule is indicated by brackets, whichcorrespond with ISL1 staining in an adjacent section (data not shown). I: ISL1 expression in utricle at E14.5 was mainly in supporting cells (SC), withlittle expression in hair cells (HC). J: ISL2 expression was higher in saccular hair cells, weak in supporting cells and stroma. K: LHX3 expression wasonly in the utricular hair cells, coinciding with Myosin VIIa (MYO7A). L, M, O: Lhx5, Lhx9, and Lmx1b were detected in the sensory epithelial cell region.N: Robust Lmx1a expression was restricted to non-sensory epithelia including the endolymphatic duct (ED), and was excluded in the vestibularganglions (VG). Ut, utricle; Sac, saccule; Cri, crista; SE, sensory epithelia. Scale bar � 20 �m.
LIM-HD EXPRESSION DISTRIBUTION IN THE INNER EAR 3307
esting pattern: its expression alter-nated between hair cell and support-ing cell regions, in which theexpression in two discrete hair cell re-gions (Fig. 2D, below the lumen,brackets) was separated by expressionin the supporting cell region (Fig. 2D,above the basal lamina, bracket). It isunclear about the structural signifi-cance of the expression pattern, as itdoes not entirely correlate with striolaversus extra-striola regions (data notshown). Lhx6, 9, and Lmx1b wereweakly expressed in the sensory epi-thelium (Fig. 2E,F,H). Expression ofLmx1a, however, was markedly re-duced and restricted to the transi-tional epithelium and dark cells (Fig.2G, brackets).
In the E16.5 cochlea, hair cells havecompleted fate determination fromthe base to the apex, and are under-going differentiation. Distinct sup-porting cell subtypes in the organ ofCorti are clearly visible. ISL1 wasprominently detectable in all cochlearepithelial cells (Fig. 2I). This observa-tion contrasted with a previous studyin which Isl1 was no longer detectableat this stage (Radde-Gallwitz et al.,2004). In fact, Isl1 expression was alsodetected at much later stages, includ-ing the P6 postnatal inner ear (Fig.3I,J). It is likely that the protocols weused for immunohistochemistry andin situ hybridization were more sensi-tive, allowing detection of Isl1 at laterdevelopmental stages. ISL2 waswidely detected, with a slightly higherlabeling level in hair cells than inother cells including supporting cellsand the greater epithelial ridge (GER,Fig. 2J). Similar to the vestibular sys-tem, LHX3 only appeared in the co-chlear hair cells (Fig. 2K), while Lhx5,6, 9, and Lmx1b were detected in haircells and GER (Fig. 2L,M,N,P). Lmx1awas exclusively expressed in the non-sensory epithelial cell (NSE) regionsthat will give rise to spiral promi-nence, primordial stria vascularis,and Reissner’s membrane (Fig. 2O,bracket). All the LIM-HDs main-tained their expression in the spiralganglions, with the exception ofLmx1a and Lhx3 (data not shown).
From E18.5 to P6, the vestibularsystem undergoes maturation withhighly differentiated hair cells by P6.In the cochlea, the patterning of theorgan of Corti and hair cell specifica-
tion is complete by E18.5 (Chen et al.,2002), with hair cells expressing func-tional channel genes by P6 (Kros,2007). Expression patterns of LIM-HDs at E18.5 were generally main-tained as in E16.5, but with changedexpression levels. In the vestibularsystem, ISL1 labeling was mainly insupporting cells and ISL2 was slightlyup-regulated in hair cells (data notshown). The expression patterns ofLhx3, 5, 6, 9, and Lmx1b were consis-tent with that found in E16.5, withLmx1a expression further reduced(data not shown). In E18.5 cochlea,ISL1 labeling was greatly reduced inhair cells but maintained in other co-chlear cell types and GER (Fig. 3A).ISL1 expression was subsequentlydown-regulated at P6 in all cochlearcells, with the exception of Hensencells and GER (Fig. 3I). In situ hybrid-ization with an Isl1 ribo-probe showedan expression pattern at P6 thatmatched the antibody study (Fig. 3J).In contrast, ISL2 level was main-tained in hair cells, supporting cells,and GER from E18.5 to P6 (Fig. 3B,K).As ISL2 labeling was generallyweaker in comparison to ISL1, we fur-ther performed in situ hybridizationto verify the immunostaining results.
For both E18.5 and P6, in situ hybrid-ization of Isl2 showed patterns thatmatched the immunostaining ob-tained by using the antibody (Fig.3C,L). Prominent hair cell labeling ofLHX3 was maintained in both E18.5cochlea (Fig. 3D) and P6 utricle (Fig.3M). Lhx5, 6, 9, and Lmx1b were con-tinuously expressed in cochlear epi-thelium from E18.5 to P6 (Fig. 3E,F,H,N,O). There was continuous down-regulation of Lmx1a in E18.5 cochlea,with expression maintained in the spi-ral prominence (Fig. 3G, arrow), butdown-regulated in marginal cells (Fig.3G, arrowhead). Little Lmx1a expres-sion remained in the P6 utricular darkcells and transitional epithelium (Fig.3P, bracket). Table 1 summarizesLIM-HD expression patterns duringinner ear development.
DISCUSSION
We comprehensively examined the ex-pression patterns of all twelve knownLIM-HDs in the developing mouse in-ner ear. The overlapping yet distinctexpression profile of individual LIM-HDs, from the early proliferating pro-genitors (Isl1 and Lmx1a) to differen-tiating sensory epithelia (Lhx3, 5, 6, 9,
Fig. 2. Expression of LIM-HDs in the E16.5 vestibular system (A–H) and cochlea (I–P). A: ISL1expression was mainly in utricular supporting cells. B: ISL2 expression was similar to E14.5, slightlystronger in hair cells than in supporting cells or stroma. C: LHX3 expression was upregulated in haircells. D: In utricle, Lhx5 expression was in two regions associated with hair cells (brackets near thelumen), which was separated by a region with supporting cell expression (bracket close to the basallamina). E, F, H: Lhx6, 9, and Lmx1b showed expression primarily in the hair cell region. G: Lmx1aexpression was greatly reduced and restricted to the transitional epithelium and dark cells of utricleand crista (brackets). I: ISL1 was expressed in all cochlear epithelial cells including hair cells andspiral ganglions (SG). J: ISL2 was widely expressed in the cochlea, with the labeling moreprominent in hair cells. K: LHX3 was specifically expressed in hair cells. L,M,N,P: Lhx5, 6, 9, andLmx1b were weakly expressed in cochlear hair cells, GER, and SG. O: Lmx1a was expressed in thenon-sensory epithelia (NSE), including primordial spiral prominence, stria vascularis, and Reiss-ner’s membrane, as demarcated by the bracket. Scale bar � 20 �m.
Fig. 3. Expression of LIM-HDs in E18.5 cochlea (A–H) and P6 inner ear (I–P). Each stage waslabeled with sidebars. A: ISL1 was down-regulated in hair cells at E18.5, while it was maintained inall other cochlear cells. B: ISL2 showed the same expression pattern as in E16.5, with slightly higherexpression in hair cells than in other cochlear cells. C: Isl2 in situ hybridization showed the samepattern as shown with immunostaining, including spiral ganglions (SG). D: LHX3 showed the samehair cell expression pattern as in E16.5. E, F, H: Expression of Lhx6, Lhx9, and Lmx1b was similarto their respective expression patterns in E16.5. G: Expression of Lmx1a was maintained in thespiral prominence (SP, arrow) and down-regulated in the marginal cells (MC, arrowhead). I: At P6,ISL1 expression was completely absent in hair cells and most supporting cells, while it wasmaintained in Hensen cells (HeC) and GER. J: Isl1 in situ hybridization showed the result thatmatched the immunostaining in 3I with the exception in Hensen cells, which may indicate that theimmunostaining had higher sensitivity. K: ISL2 expression was up-regulated in hair cells while itwas maintained at a lower level in supporting cells and GER. L: Isl2 in situ hybridization showedexpression patterns nearly identical to that obtained from immunostaining study (3K). M,O: Ex-pression of Lhx3 and 9 in P6 utricle and cochlea. N: Lhx5 expression was in the P6 cochlear haircells, spiral ganglion neurons and very weakly in the GER. P: Lmx1a expression was significantlydown-regulated in the transitional epithelium of utricle (bracket). Scale bar � 20 �m.
3308 HUANG ET AL.
Fig. 2.
Fig. 3.
Isl2, and Lmx1b), suggests that theseproteins are involved in a number ofdevelopmental events in different celltypes at different developmentalstages. We used immunostaining andin situ hybridization to study Isl1 andIsl2 expression patterns, and observedthe same pattern with the two meth-ods, which further supported our re-sults in general.
Over two thousand transcriptionfactors have been identified in themammalian genome, with only ahandful studied in the inner ear fortheir expression patterns or func-tional roles (Gray et al., 2004). LIM-HDs are a well-studied protein family,particularly for their functions in theneuron sub-type specification, pitu-itary gland development, and stemcell properties (Hobert and Westphal,2000; Allan and Thor, 2003; Hunterand Rhodes, 2005; Andersson et al.,2006; Rhee et al., 2006). All LIM-HDsexamined by in situ hybridization orimmunostaining, with the exceptionof Lmx1a and Lhx3, are expressed ininner ear ganglion neurons at various
stages, suggesting their roles in thedevelopment of cochleovestibular gan-glions. Mouse knockout models havebeen produced for all the LIM-HDs(Hunter and Rhodes, 2005; Liodis etal., 2007). In all cases, LIM-HD knock-outs showed discernable phenotypechanges, with many showing abnor-malities in neuronal development,demonstrating important functionalroles of LIM-HDs in general. How-ever, the effects on inner ear develop-ment in the LIM-HD knockout micehave not been investigated with theexception of Lmx1a. Our systematicanalysis of LIM-HDs expression in theinner ear sets the stage for furtherinvestigation of their functional roles.
Lmx1a expression, which is presentfrom the E10.5 otocyst to later in in-ner ear development, has been consis-tent with its non-sensory epithelial or-igin. Its early expression largelyoverlaps with that of Pax2, which isinvolved in specification within the co-chlear duct, in particular in the regionbetween the stria vascularis and co-chlear sensory epithelia (Burton et al.,
2004). However, unlike Pax2, which isalso expressed in prosensory patchesand hair cells at later stages, Lmx1aexpression is exclusively in the non-sensory epithelia, including the ves-tibular transitional epithelia, the sac-cular roof (data not shown), dark cells,strial vascularis, and strial promi-nence. The expression pattern ofLmx1a makes it a good marker for alineage tracing study of non-sensoryepithelial cell types. In a spontane-ous homozygous Lmx1a mouse mu-tant, dreher, both cochlea and vestib-ular maculae fail to develop and themice exhibit abnormalities includingcircling, head-tossing, and deafness(Millonig et al., 2000) (http://www.informatics.jax.org/javawi2/servlet/WIFetch?page�alleleDetail&key�51666), suggesting that normal devel-opment of non-sensory epithelia is dis-rupted in the absence of LMX1A, andLmx1a is essential for the genesis ofthe inner ear. Future studies of thedreher mutant inner ear, based on theLmx1a expression pattern, shouldhelp to elucidate the mechanism un-derlying the specification of non-sen-sory epithelium and its relationship tothe development of the inner ear ingeneral.
Early Isl1 expression in the otocystsuggests that it may have a role in celllineage specification of the prosensoryprogenitors (Li et al., 2004; Radde-Gallwitz et al., 2004), and its subse-quent downregulation in differentiat-ing hair cells may be important fortheir proper differentiation. This issupported by a separate study inwhich Lhx3 overexpression results insuppression of Isl1 expression in thecochlea (Z.Y. Chen, unpublisheddata). The non-overlapping, sensoryversus non-sensory expression pat-terns of Isl1 and Lmx1a make themgood candidates to label progenitors inthe pro-sensory and pro-non-sensoryregions respectively. The potentialrole of Isl1 in inner ear progenitorcells is consistent with recent studiesdemonstrating that Isl1 is associatedwith cardiac multipotent progenitorcells, which are capable of giving riseto all the major heart cell types accom-panied by the reduction of Isl1 expres-sion (Laugwitz et al., 2005; Moretti etal., 2006). Interestingly, ISL1 labelingis generally absent in late postnatalcochlea sensory epithelia yet is main-
TABLE 1. Summary of LIM-HD Expression in Developing Inner Ear
Level of expression is determined by observation of signal of a particular LIM-HDthroughout the developmental stages. The expression levels are classified �, detected;�, no expression; Ot, otocyst; OG, otic ganglions; Ves, vestibule, Coch, cochlea; CVG,cochleovestibular ganglions; GN, ganglion neurons; SE, sensory epithelia (includeboth hair cells and supporting cells); H, expression in hair cells; S, expression insupporting cells; GER, the greater epithelial ridge; NSE, non-sensory epithelia; HeC,Hensen cells.
3310 HUANG ET AL.
tained in some utricular supportingcells even in the adult (data notshown). It is tempting to speculatethat sustained Isl1 expression in adultutricles may contribute to the limitedproliferation potential in supportingcells after hair cell death (Forge et al.,1993; Warchol et al., 1993; Lambert etal., 1997). Absence of Isl1 during earlydevelopment may therefore result infailure of the inner ear to develop,whereas sustained Isl1 expression inhair cells may lead to impaired differ-entiation. These hypotheses can beaddressed by conditional deletion ofIsl1 in the otocyst, and by hair cell–specific activation of Isl1 in a trans-genic mouse model.
Initiation of expression of Isl2,Lhx3, 5, 6, 9, and Lmx1b in differen-tiating sensory epithelia starts be-tween E12.5–E13.5. Their expressionis generally maintained throughoutthe developmental stages examined.ISL2 labeling is slightly up-regulatedin hair cells after E12.5 but remainsweak in supporting cells at laterstages. Interestingly, the ISL2 stain-ing pattern, to a certain degree, is in-versely correlated with the expressionpattern of ISL1 from E12.5 onwards.In the spinal cord motor neurons,ISL2 is expressed after expression ofISL1, and Isl2 mutant mice displaydefects in the migration and axonalprojections of motor neurons (Thaleret al., 2004). In the inner ear, Isl2 mayplay a role in cells that initially ex-press Isl1, to further define subse-quent differentiation.
The sequence of expression of someLIM-HDs in the inner ear is differentfrom other tissues such as neurons.For instance, during spinal cord devel-opment of chicks and mice, motor neu-ron progenitors express Lhx3, andpostmitotic motor neurons expressIsl1 (Thaler et al., 2002). In contrast,in the developing inner ear, the oppo-site is true, with dividing sensory ep-ithelia progenitors expressing Isl1and postmitotic hair cells expressingLhx3. The switch in timing and cellsubtypes of Lhx3 and Isl1 expressionin the inner ear may reflect the context-dependent functions of both LIM-HDs.
While Isl1 and Lmx1a are likely im-portant in the specification of progen-itor cell populations of sensory andnon-sensory epithelia, what is thefunctional significance of expression of
other LIM-HDs in the differentiatinghair cells and supporting cells? In theCNS, combinatorial functions of LIM-HDs play critical roles in sub-typeneuron specifications (Thor et al.,1999; Shirasaki and Pfaff, 2002; Allanand Thor, 2003). The primary roles ofLIM-HDs in early differentiation pro-cesses are consistent with recent stud-ies showing that most LIM-HDs arebound by the Polycomb repressivecomplex subunit SUZ12, which sup-presses genes involved in differentia-tion of embryonic stem cells (Lee etal., 2006). Within the inner ear, com-binatorial functions of LIM-HDs maybe important in defining more cellsubtypes beyond what is alreadyknown. Indeed, expression of a specificLIM-HD is far from uniform in a giveninner ear cell type. Isl1, for instance,is predominantly expressed in the ves-tibular supporting cells, but can stillbe seen in occasional hair cells. Lhx3is expressed in all hair cells, but withvariable expression levels. The combi-nation of subtle differential expres-sion levels and patterns of LIM-HDsmay stoichiometrically specify a par-ticular cell subtype. Future experi-ments including expression profilingat single cell resolution and studies ofmultiple LIM-HD gene conditionalknockout mice should reveal the func-tional roles of LIM-HDs and may iden-tify novel inner ear cell types.
EXPERIMENTALPROCEDURES
RT-PCR and In SituHybridizationSpecific primers corresponding to eachLIM-HD gene were designed (Table 2).RT-PCR was performed using a cDNAmixture prepared from mouse inner earsfrom E10.5 to P6, which included bothutricle and cochleas. The conditions forRT-PCR were as follows: 94°C, 2 min;94°C, 30 sec; 67°C, 1 min; 72°C, 1 min 30sec for 35 cycles; 72°C, 10 min. All LIM-HDs were amplified by RT-PCR, indi-cating their expression in at least one ofthe stages examined. All the RT-PCRamplified the fragments of expected size(data not shown). RT-PCR fragmentswere further cloned into TOPO TA-Cloning vector (Invitrogen) and two in-dependent clones for each LIM-HD RT-PCR were isolated and sequenced. Thesequencing data showed all the LIM-HDs PCR products were proper clones(data not shown). For in situ hybridiza-tion, timed pregnant CD1 female micewere purchased from the Charles RiverLaboratories for embryo collection.Transverse sections of mouse innerears, from E10.5 to P6, were preparedas described (Sage et al., 2006). Ribo-probes preparation and subsequent insitu hybridization follow the exact pro-cedure as described (Sage et al., 2006).
TABLE 2. Primers Used to Amplify Individual LIM-HDs by RT-PCR
LIM-HD EXPRESSION DISTRIBUTION IN THE INNER EAR 3311
Immunohistochemistry
Antibodies against POU4F3 (1:100)(Xiang et al., 1997), PAX2 (1:100, Co-vance, Princeton, NJ), LHX3, ISL1 (1:100, Developmental Studies Hybrid-oma Bank, University of Iowa, IowaCity, IA), ISL2 (1:1,000, Dr. SamuelPfaff, Salk Institute, La Jolla, CA),and Myosin VIIa (1:2,000, ProteusBioSciences, Ramona, CA) were usedfor immunohistochemistry. Frozen sec-tions of the inner ear tissues were driedfor 15 min at 37°C, and re-hydrated in1�PBS for 5 min. Then the sectionswere subjected to an antigen unmask-ing treatment using the Antigen Un-masking Solution (Vector Laboratories)according to the manufacturer’s proto-col. The blocking, primary and second-ary antibody incubation followed thestandard protocol. The secondary anti-bodies were anti-rabbit Alexa Fluor 594and/or anti-mouse Alexa Fluor 488 (Mo-lecular Probes) for fluorescent labeling.
ACKNOWLEDGMENTSWe thank Dr. Joe Adams for his con-structive comments on the manu-script. The project was supported byNIH grants DC-04546, DC-06908(Z.Y.C.) and DC-6167 (S.H.). M.H. wassupported in part by a grant from theDRF foundation. The research wasconducted while Z.-Y.C. was a Pfizer/AFAR Innovation in Aging ResearchGrant recipient.
REFERENCES
Allan DW, Thor S. 2003. Together at last:bHLH and LIM-HD regulators cooperate tospecify motor neurons. Neuron 38:675–677.
Andersson E, Tryggvason U, Deng Q,Friling S, Alekseenko Z, Robert B, Per-lmann T, Ericson J. 2006. Identificationof intrinsic determinants of midbrain do-pamine neurons. Cell 124:393–405.
Barald KF, Kelley MW. 2004. From placodeto polarization: new tunes in inner ear de-velopment. Development 131:4119–4130.
Bermingham NA, Hassan BA, Price SD,Vollrath MA, Ben-Arie N, Eatock RA,Bellen HJ, Lysakowski A, Zoghbi HY.1999. Math1: an essential gene for thegeneration of inner ear hair cells. Sci-ence 284:1837–1841.
Bryant J, Goodyear RJ, Richardson GP.2002. Sensory organ development in theinner ear: molecular and cellular mech-anisms. Br Med Bull 63:39–57.
Burton Q, Cole LK, Mulheisen M, ChangW, Wu DK. 2004. The role of Pax2 inmouse inner ear development. Dev Biol272:161–175.
Chen P, Johnson JE, Zoghbi HY, Segil N.2002. The role of Math1 in inner eardevelopment: Uncoupling the establish-ment of the sensory primordium fromhair cell fate determination. Develop-ment 129:2495–2505.
Chizhikov V, Steshina E, Roberts R, Ilkin Y,Washburn L, Millen KJ. 2006. Moleculardefinition of an allelic series of mutationsdisrupting the mouse Lmx1a (dreher)gene. Mamm Genome 17:1025–1032.
Forge A, Li L, Corwin JT, Nevill G. 1993.Ultrastructural evidence for hair cell re-generation in the mammalian inner ear.Science 259:1616–1619.
Gray PA, Fu H, Luo P, Zhao Q, Yu J, Fer-rari A, Tenzen T, Yuk DI, Tsung EF, CaiZ, Alberta JA, Cheng LP, Liu Y, Sten-man JM, Valerius MT, Billings N, KimHA, Greenberg ME, McMahon AP, Row-itch DH, Stiles CD, Ma Q. 2004. Mousebrain organization revealed through di-rect genome-scale TF expression analy-sis. Science 306:2255–2257.
Hobert O, Westphal H. 2000. Functions ofLIM-homeobox genes. Trends Genet 16:75–83.
Hunter CS, Rhodes SJ. 2005. LIM-home-odomain genes in mammalian develop-ment and human disease. Mol Biol Rep32:67–77.
Kawamoto K, Ishimoto S, Minoda R,Brough DE, Raphael Y. 2003. Math1gene transfer generates new cochlearhair cells in mature guinea pigs in vivo.J Neurosci 23:4395–4400.
Kelley MW. 2006. Regulation of cell fate inthe sensory epithelia of the inner ear.Nat Rev Neurosci 7:837–849.
Kros CJ. 2007. How to build an inner haircell: challenges for regeneration. HearRes 227:3–10.
Lambert PR, Gu R, Corwin JT. 1997. Anal-ysis of small hair bundles in the utriclesof mature guinea pigs. Am J Otol 18:637–643.
Laugwitz KL, Moretti A, Lam J, Gruber P,Chen Y, Woodard S, Lin LZ, Cai CL, LuMM, Reth M, Platoshyn O, Yuan JX,Evans S, Chien KR. 2005. Postnatal isl1�cardioblasts enter fully differentiated car-diomyocyte lineages. Nature 433:647–653.
Lee TI, Jenner RG, Boyer LA, GuentherMG, Levine SS, Kumar RM, Chevalier B,Johnstone SE, Cole MF, Isono K, KosekiH, Fuchikami T, Abe K, Murray HL,Zucker JP, Yuan B, Bell GW, Herbol-sheimer E, Hannett NM, Sun K, OdomDT, Otte AP, Volkert TL, Bartel DP,Melton DA, Gifford DK, Jaenisch R,Young RA. 2006. Control of developmen-tal regulators by Polycomb in humanembryonic stem cells. Cell 125:301–313.
Li H, Liu H, Sage C, Huang M, Chen ZY,Heller S. 2004. Islet-1 expression in thedeveloping chicken inner ear. J CompNeurol 477:1–10.
Liodis P, Denaxa M, Grigoriou M, Akufo-AddoC, Yanagawa Y, Pachnis V. 2007. Lhx6 ac-tivity is required for the normal migrationand specification of cortical interneuronsubtypes. J Neurosci 27:3078–3089.
formation of the roof plate in the verte-brate CNS. Nature 403:764–769.
Moretti A, Caron L, Nakano A, Lam JT,Bernshausen A, Chen Y, Qyang Y, Bu L,Sasaki M, Martin-Puig S, Sun Y, EvansSM, Laugwitz KL, Chien KR. 2006. Mul-tipotent embryonic Isl1(�) progenitorcells lead to cardiac, smooth muscle, andendothelial cell diversification. Cell 127:1151–1165.
Radde-Gallwitz K, Pan L, Gan L, Lin X,Segil N, Chen P. 2004. Expression of Is-let1 marks the sensory and neuronal lin-eages in the mammalian inner ear.J Comp Neurol 477:412–421.
Rhee H, Polak L, Fuchs E. 2006. Lhx2maintains stem cell character in hair fol-licles. Science 312:1946–1949.
Sage C, Huang M, Karimi K, Gutierrez G,Vollrath MA, Zhang DS, Garcia-AnoverosJ, Hinds PW, Corwin JT, Corey DP, ChenZY. 2005. Proliferation of functional haircells in vivo in the absence of the retino-blastoma protein. Science 307:1114–1118.
Sage C, Huang M, Vollrath MA, BrownMC, Hinds PW, Corey DP, Vetter DE,Chen ZY. 2006. Essential role of retino-blastoma protein in mammalian hair celldevelopment and hearing. Proc NatlAcad Sci USA 103:7345–7350.
Shirasaki R, Pfaff SL. 2002. Transcrip-tional codes and the control of neuronalidentity. Annu Rev Neurosci 25:251–281.
Shou J, Zheng JL, Gao WQ. 2003. Robustgeneration of new hair cells in the ma-ture mammalian inner ear by adenoviralexpression of Hath1. Mol Cell Neurosci23:169–179.
Thaler JP, Lee SK, Jurata LW, Gill GN,Pfaff SL. 2002. LIM factor Lhx3 contrib-utes to the specification of motor neuronand interneuron identity through cell-type-specific protein-protein interac-tions. Cell 110:237–249.
Thaler JP, Koo SJ, Kania A, Lettieri K,Andrews S, Cox C, Jessell TM, Pfaff SL.2004. A postmitotic role for Isl-class LIMhomeodomain proteins in the assign-ment of visceral spinal motor neuronidentity. Neuron 41:337–350.
Thor S, Andersson SG, Tomlinson A,Thomas JB. 1999. A LIM-homeodomaincombinatorial code for motor-neuronpathway selection. Nature 397:76–80.
Warchol ME, Lambert PR, Goldstein BJ,Forge A, Corwin JT. 1993. Regenerativeproliferation in inner ear sensory epithe-lia from adult guinea pigs and humans.Science 259:1619–1622.
Xiang M, Gan L, Li D, Chen ZY, Zhou L,O’Malley BW Jr, Klein W, Nathans J.1997. Essential role of POU-domain fac-tor Brn-3c in auditory and vestibularhair cell development. Proc Natl AcadSci USA 94:9445–9450.
Xiang M, Gao WQ, Hasson T, Shin JJ.1998. Requirement for Brn-3c in matu-ration and survival, but not in fate de-termination of inner ear hair cells. De-velopment 125:3935–3946.
Zheng JL, Gao WQ. 2000. Overexpressionof Math1 induces robust production ofextra hair cells in postnatal rat innerears. Nat Neurosci 3:580–586.
