BRAIN RESEARCH ELSEVIER
Brain Research 660 (1994) 181-191
Histochemical and scanning electron microscopic studies of supernumerary hair cells in embryonic rat cochlea in vitro A. Abdouh, G. Despres, R. R o m a n d * Laboratotre de Neuroblologw et Phy~lologte du ddt eloppernent, Umverstte Blaise Pa~ al-Clerrnont I1, 03177 Aubti're ( eden. [ t a m e Accepted 28 June 19q4
In the embryonic organ of Corh supernumerary hair cells were observed when developed m organotyptc cultures. Hair cells ranging in up to two rows of inner hair cells (IHCs) and up to nine rows of outer hair cells (OHCs), were obscrved by phalloldln hlstochemistry The total number of hmr cells may double in some explanted cochleae compared to control ones Cutlcular plates of hair cells &splayed an actin-free zone corresponding to the kmoclhum location, differently located and indmatlng different degrees of differentiation and maturation Moreover, some hmr cells had a small apical surface area and a ccntrally located klnocihum, revealing ~mmatunty. Under scanning electron microscopy, stereocdla appeared to dfffcrentlatc normally, as compared to the m VlVOdevelopment The staircase pattern of the stereocfliary bundles was reached on most of thc hair cells with a 'V" shape on the OHCs and hemispherical one on the IHCs. Hair cell polarity was not homogeneous along the length of the t~ssue Organs of Corti explanted at birth developed a weaker number of supernumerary hair cells showing a decrease ot supernumerary hmr cells w~th the developmental stage of the explant These results provide evidence for supernumerary hair cells in the mammahan cochlea in culture, w~thout loss or injury to preexisting hair cells
Keywords Au&tory receptor, Development; Fetus; Overproduction; Organotyp~c culture
It has been recently shown that an organotypic culture system of embryonic rat cochlea maintained several days in vitro (DIV) can produce supernumerary hair cells . These cochleae contained differentiating hair cells in which the organization and the structure associated showed varying degrees of development. However, no information was provided at the electron microscopic level concerning the degree of differentiation of the specialized apical surface of hairs ceils with its stereocilia and klnocilium. In fact, in the adult the stereocilia bundle present a very sophisticated organization [14,23,28] which is the basis for transducing acoustic stimuli into electrical activity. Along with this organization, the apical pole of hair cells is polarized with a specific orientation of the stereocilia bundle and a precise location of the kinocilium in developing hair
* Corresponding author, Fax (33) 73 40 78 112 0006-8993/94/$07 00 © 1994 Elsevier Science B V All rights reserved SSDI 0 0 0 6 - 8 9 9 3 ( 9 4 ) 0 0 8 0 2 - 7
cells or basal body in adult. The establishment of polarized cytoarchltecture is fundamental to the specialized functions of many cells . It is known in the bird that the hair cell acquires its polarity by reonentatton of the stereocilia bundle and changing location of kinocilium during specific phases of hair cell development [11,12,42,43]. It was of interest to study tf hair cells still achieve their normal polarity during in vitro development when explanted during the period of stereocilia ontogenesis . Some information were given from organotyplc cultures of postnatal cochlea where stereocilia bundles developed normally, except maybe for some links between ciha which retained some immature features . In order to determine the degree of differentiation of the apical pole of hair cells of embryonic cochleae maintained in vitro, surface preparations with phalloidm staining and scanning electron microscopy (SEM) were made of these cultures and the morphology of the apical surface of hair cells was studied. The results of this study indicate that the polarity and the stmrcase pattern of the stereocilia bundle were reached by most
t 4hdo.h ct al ~Brain Re~ear(h tnSO (1994) 181-191
hair cells, although some of them presented different polarities. Moreover, fetal explants presented the ability to generate extra hair cells that may be of interest to hair cell regeneration.
2. Materials and methods Experiments were carried out on Sprague-Dawley rats bred m our ammal faohty Pregnant females were hmed" the presence of spermatozoa m vaginal smears, after 12 h of contact with male, was recorded as day of gestatmn 0 (0DG) Birth, between 21DG to
22DG, was considered as postnatal day U (I)PI)) -Fh~s study was carrmd out on cochleae explanted at 16DG (n = 25L 18DG (n = 5ill 20DG (n = 12), 0PD (n = 25) and 5PD (n = lfll The detaded procedure concerning cultures and h~stochem=cal staining has been pubhshed previously  The mare steps are as follow rat embryos were removed, under sterde cond~hon~, from deeply anesthehzed pregnant females (sodmm pentobarb~tal, 5(1 m g / k g body wmght) B~rth and 5PD ammals were decapitated alter cryo-anesthesla Dtssecuon procedure A gross dissection was rapidly pertormed the head was cut sag~ttally and temporal bones, gently carved w~th a pmr of scissors, were transported m a Petn d~sh containing Hanks' balanced salt solution (HBSS, S~gma) Microd~ssection of the t~ssues was performed with a stereom]croscope installed under an horizontal
15%-18DG Og6-18DG Ref-18DG Ref-SPO
15%-180G Oe~-ISDG Ref-18DG Ref-SPD
O0 1 2 0
m Ba~ a Middle
F~g. 1 Histograms of the average number of hair cells at different stages of explantation. A and B: stage 181M3 with its three segments of the cochlea. Supernumerary IHCs (A) and OHCs (B) can be compared in two different culture conditions; with 15% horse serum (15%-18DG) and without serum (0%-18DG) into the culture medium. Only IHCs from the in vivo basal cochleae at 18DG (Ref-18DG) can be observed at this stage of development and used as reference Cochleae at 5PD (SPD) are also used as control C and D: s t a g e 20DG in serum-free culture medmm compared with 20DG m vwo cochleae (20DG) and 5PD (5PD) taken as reference. E and F- newborn stage with two different culture conditions, 15% horse serum (15%-B) and without serum (0%-B) compared to newborn m vlvo cochlea (Ref-B) and 5PD (5PD) taken as reference Vertmal bars = 1 S.D
A Abdouh et al/Brain Research 600 (1994) 181-191
subsequently dehydrated in graded series of ethanol (30%, to 100%) and absolute acetone The specimens were then critical point dried, coated with gold and observed with a Cambridge 350 scanning electron microscope Quantttatwe analysis The cochleae were divided into three pieces, basal, middle and apical and maintained 7 DlV as described above Each segment consists of approximately one half turn Four stages were studied, 16DG which correspond to the stage before hair cell differentiation m the rat (Romand et al ), tSDG and 20DG at the onset of IHC and OHC dlfferentmtlon respectively For comparison two postnatal in VlVO stages were used as references, I e birth and 5PD Experiments were carried out with two culture conditions, with 15% horse serum or with a serum free medium made of MEM, where serum was replaced by MEM Hair cell counts were made trom a standardized length of 200 /xm along the three segments of the cochlea At least 8 cochleae were used for each stage from which more than 30 counts were obtained For each explanted stage, the number of hair cells was obtained from corresponding in vwo control cochleae
laminar flow hood Spiral laminae were delicately extirpated from cochleae and transferred in a Petrl dish containing a minimum essential medium (MEM, Sigma) Stria vascularis, Relssner's and tectorlal membranes were removed Fetal spiral laminae were explanted in their entirety, the neonatal ones were dwlded into three turns apical, middle and basal In some cases, the spiral ganghon was removed The explants were then transferred onto collagen floating on feeding medmm into previously prepared Falcon culture dishes Management of ~ulture~ Falcon culture dishes were prepared 24 h before dlssechon. A drop of rat tad collagen, prepared accordmg to the method of Mazurovsky and Peterson , was slowly layered on 700/zl of MEM in the center well of a Falcon dmsh, photo-reconstituted by exposure to UV for 30 mm and placed 24 h in an Heraeus incubator set at 37°C and 5% CO 2 atmosphere Ten minutes before the dissection, the prepared culture dishes were transferred under the laminar flow hood and the medmm was replaced by a feeding solution prewarmed at 37°C (50c£ MEM with 20 mM Hepes, 29% HBSS. 6 8 g glucose/l, 2 mM L-glutamme and 15% horse serum, all purchased from Sigma) In some experiments, horse serum was replaced by MEM. As soon as the explant was put on the collagen floating over the feeding medmm, the Falcon dish was covered and transferred m the CO 2 incubator at 37°C Cultures were checked every day and fed with fresh feeding medmm every 2 days Htstochemtstry After different t~mes m culture, explants were rinsed with HBSS to remove the excess of feeding medmm and fixed for 1 h m 4% paraformaldehyde prepared in 0 1 M phosphate buffered sahne (PBS, pH 7 3), and rinsed several times in PBS, In order to label hair cells, wholemount explants were incubated w~th tetramethyl rhodamme labeled phalloldm (Sigma, 3 /xg/ml m PBS) for 30 mm. This toxin is a specific probe for F-actm  In order to observe the mnervat~on of the organ of Cort~ an anti-neurofdament monoclonal antibody (a gift of Dr Dahl, Harvard Medical School, Boston) was used The antibody was diluted 1/100 m PBS containing 20% fetal calf serum, 0.3% Triton × 100 and 0 05% thlmerosal After several rinses, samples were incubated with a secondary anti-mouse IgG antibody conjugated to fluorescem isothlocyanate (Sigma), diluted 1/200 m PBS After many rinses m PBS, explants were mounted m 90% glycerol in carbonate buffer (pH 8 0) with 0 1% p-phenylenedlamme as an ant~-bleach agent Preparahons were observed w~th a N~kon Optlphot eplfluorescence microscope Scanmng electron microscopy (SEM) Exp[ants of different stages and m VlVOdissected spiral laminae of 6PD were fixed for 1 h m 2% paraformaldehyde, 2% glutaraldehyde m 0 12 M Sorensen phosphate buffer, postfixed m 1% osmium tetrox~de m the same buffer and
3.1. Quantltattt'e results The number of rows of explanted cochleae from 16DG and 18DG ranges from 1 to 3 for IHCs and 3 to 9 for OHCs, whereas in newborn and 5PD control cochleae, the number of IHC rows is stable at 1 and is from 3 to 4 for OHCs. For all three segments of explanted cochleae irrespecttve of the stage compared to controls, the number of IHCs and OHCs is statistically higher ( P < 0.001, Student's-t-test), (Fig. 1). For example, there are almost double the number of OHCs from 18DG explants compared with 5PD cochleae (Fig. 1B). Moreover, cochleae explanted at birth for 7 DIV present a statistically higher number of hair cells when compared with reference cochleae ( P < 0.001), (Fig. 1E-F). The supernumerary hair cells are present
0 6O lO-
Z 0 16DG
2 0 DG
Fig. 2 Modification of supernumerary hair cells with cochlear development Mean number of hair cells in relation to stages of explantatton A and B correspond to the mean number of hair cells from the basal part of the cochlea maintained in serum free medium for 7 DIV Note, the decrease of supernumerary hair cells related to stages of explantatlon for IHCs (A) and OHCs (B) The number of IHCs and OHCs is statistically different ( P < 0 001, Student's-t-test), from the reference m vwo newborn cochlea at every stage of explantatlon Vertical bars represent one standard dewation
I 4b~h)uh et u/ ' Brain Research O00 (1994) 181-191
whatever the culture conditions and the stage of explantation observed (Figs. 1 and 2). A comparison of the mean number of hair cells at different stages of explantation shows a decrease of supernumerary hair cells with the developmental stage of the explant (Fig. 2).
3.2 Hlstochemtcal studle~s After seven days m culture 16DG explants developed a morphological structure different from that seen in vivo at birth. Observed from surface preparations labeled with phalloidin, the IHCs were more
A Abdouh et a l / B r a m Research 660 (1994) 181-101
Fig 4 Six-day-old cochlear explant taken from a 18DG fetus double labeled A with phalloldln for actm and B with antl-neurofllament antibody for fibers Supernumerary hair cells are well present for IHCs (1) and OHCs (2) The locahzatlon of fibers with neurofllament antibody (B) shows that basal pole of hair cells are shifted toward the external border of the spiral lamina relative to the apical pole of hair cells Some radial fibers reach the IHCs' rows and ramify into branches which terminate m endings (black arrows) Other fibers elongate toward OHC rows (thin arrows) and start to bifurcate (open arrow) for forming maybe the future outer spiral bundle It is interesting to notice that extra rows of OHCs can be innervated by fibers as shown by the ending depicted by the large arrow Scale bar 20/zm n u m e r o u s a n d w e r e o r g a n i z e d in o n e t o t h r e e r o w s ( F i g . 3 A ) . M o s t I H C s w e r e n o r m a l l y o r i e n t e d as s e e n by the location of the kinocilium indicated by the free actin area on the cuticular plate. Few cells presented an aberrant polarity as observed by various positions of the kinocilium (Fig. 3A). Some IHCs were very close together without any visible separation by supporting cells, w e c a l l e d t h e m ' S i a m e s e t w i n s h a i r c e l l s ' w h e n plasmic membranes of the medial side were connected together presenting sometimes a single membrane, and 'twin hair cells' when two small cells were close together (Fig. 3A). 'OHCs were tightly packed together showing up to 9 rows of cells with different degrees of polarity. Some OHCs presented a smaller apical surface and a differe n t p o l a r i t y as s e e n b y t h e p o s i t i o n o f t h e k i n o c i l i u m
(Fig. 3A). Those smaller cuticular plates looked s q u e e z e d b e t w e e n l a r g e r cells. Seven-day-old cultures of 18DG rat cochlea showed a l m o s t t h e s a m e f e a t u r e as d e s c r i b e d a b o v e . S e v e r a l rows of IHCs and OHCs were observed with different polarities of apical surfaces of both types of hair cells (Fig. 3B,C). The OHCs were not so tightly packed t o g e t h e r as f o r t h e 1 6 D G e x p l a n t . A t h i g h e r m a g m f i c a tion, some OHCs presented a smaller apical surface w i t h t h e a r e a d e v o i d o f a c t i n l o c a l i z e d in t h e c e n t e r o f t h e c u t i c u l a r p l a t e (Fig. 3C). The supernumerary hair cells modified the regular a r r a n g e m e n t m r o w s . T h e y w e r e m o r e o r less d i s r u p t e d a t t h e l e v e l o f b o t h t y p e s o f h a i r c e l l s in c o m p a r i s o n with an explant taken at birth and maintained 7 DIV, s h o w i n g a n o r m a l l o o k i n g o r g a n i z a t i o n (Fig. 3 D ) .
3. Surface preparation of cochlear explants labeled by phalloldm A 7-day-old explant of a 16DG rat's cochlea without spiral ganglion The hair cells are arranged in one to three rows of IHCs (bracket) and up to seven rows of OHCs. The plasmlc membrane of hair cells is well visible, especially for the IHCs. The focus made at the reticular lamina shows the polarity of the hair cells as indicated by the area devoid of staining m their apical surfaces. This region is assumed to correspond to the position of the klnocfllum (black arrows). The polarity of hair cells can be very different from one to another, especially for OHCs (large white arrows). Two IHCs very close together, were called 'Siamese twins hair cells' because their plasmlc membrane of the medial sides seem to be connected together Moreover, they present an opposite polarity (open arrow) Some smaller OHCs in pack of two squeezed between larger ones were called 'twin hair cells' (stars) B cochlear explant of a 18DG fetus maintained 7 DIV. Supernumerary hair cells is also present as shown by the number of IHCs and OHCs' rows OHCs are less densely packed together than m A, the plasmic membrane is visible as well as the CUtlcular plate Although many cells display a normal polarity, some of them presented a variable one (arrows) C' cochlear explant of a 18DG fetus maintained 7 DIV Supernumerary of hair cells is seen at the IHCs level where two or three rows are present, A cell displays an opposite polarity (large black arrow) as observed by the position of the kmocfilum Small OHCs are present between larger ones with a central position of the area devoid of actln (black arrows) Some Smmese twins hair cells are visible from the two types of hair cells (open arrows) D 6-day-old cochlear explant from the basal turn of a newborn rat No supernumerary of hair cells Is visible The regular arrangement of IHCs and OHCs is well preserved as well as the polarity of the cutlcular plate where the kmoclhum posmon is oriented toward the external side of the spiral lamina (arrows) E' 6-day-old explant from the apical turn of a newborn rat's cochlea Supernumerary hair cells are visible and restricted to the OHCs. Note the varmt~on of cell polarity and compare with the previous micrograph from the basal turn of the cochlea Scale bar 10/xm for all micrographs Fig.
4 ,qhdouh ct a//' Bratn lCesear¢h 660 (1994) 181-191
When the cochlea was explanted at birth, often the results were different depending of the region from which the epithelium was taken. When the explant was from the basal turn of the cochlea sometime supernu-
merary hair cell was observed (Fig. 3D). However, most of the time the normal configuration of the sensory epithelium was preserved, with one row of IHCs and three to four rows of O H C s with a normal cell polarity
A Abdouh et a l / B r a m Research 660 (1994) 181-191
as can be observed in vivo. Explantation made from the apical turn presented more often supernumerary hair cells as observed by the five rows of OHCs (Fig. 3E). It was interesting to compare the polarity of hair cells between these two explants. Explants from the basal cochlea presented a very regular position of kinocllium location toward the external side of the spiral lamina (Fig. 3D), while explants from the apical cochlea showed a less regular orientation of kinocilia (Fig. 3E). In order to understand factors that might influence the observed supernumerary hair cells we made in vitro experiments without serum in the culture medium. The absence or presence of serum did not influence the hair cell overproduction. In fact, supernumerary hair cells were observed in serumless medium (Fig. 3A,B) or in serum supplemented medium (Fig. 3C-E). On the other hand, the absence (Fig. 3A, B) or presence of spiral ganglion neurons (Figs. 3C, 4A, B) did not modify the hair cell overproduction. In the former experiment, double labeling using phailoidin and anti-neurofilament antibody showed nerve fiber endings at the level of the supernumerary hair cells (Fig. 4B). At the level of the two rows of IHCs, numerous radial fibers with their numerous branches were distinctly visible. Usually radial innervation is largely preserved. Some fibers reached the OHCs region, where some of them, crossing the first row of OHCs, extended to the upper ones. These fibers may represent afferent spiral fibers selectively connecting the OHCs in vivo. 3.3. S E M observattons Observation of the supernumerary hair cells by SEM showed disruption of the regular arrangement of rows as compared with a control cochlea taken 6 days after
birth (Fig. 5A). This stage was used as reference in this case because it may roughly correspond to a cochlea taken at 18DG and maintained 10 days in culture. A 18DG explant after 10 DIV presented several important modifications that can be seen on Fig. 5B. Despite these modifications, stereocilia are present on both types of hair cells. Hair cells showed a smaller size, at least at their apical surface, especially for the OHCs compared with the control cochleae (Fig. 5A,B). IHCs presented characteristic stereocilia of different sizes, vestigial stereocilia were still present as in the control. However, the arrangement and the polarity of hair cells as seen by the position of kinocllia showed some varxations between cells (Fig. 5C). The Siamese twms and twin hair cells observed by phalloidin staining can also be seen by SEM. In Fig. 5C and D the Siamese twins IHCs (Fig. 5C) presented two kmocilia and a large common cuticular plate, that may be due to the fusion of two cells. In fact, no supporting cells were visible at this level between these Siamese twins cells as can be observed by the absence of microvilli that normally top supporting cells at this stage of development. However, the absence of mlcrovilli between these Siamese twins hair cells does not totally role out the possibility of supporting cell processes between the lateral wall of hair cells. More intriguing, hair cells may present an opposite orientation between the kinocilium and stereocilia, in other words, stereoclha were located in the opposite direction on the cuticular plate in respect to the klnocilium (Fig. 5C). Interesting enough, each supporting cell still presented a well visible kinocilium. OHCs showed the same changes as described for the IHCs, especially with sets of twin hair cells (Fig. 5B and F) and variations in the polarity between hair cells. Moreover, small cells with a cuticular plate devoid of
Fig. 5 SEM mlcrographs of cochlear surface preparations A 6-day-old rat taken as control Hair cells are orderly arranged in one row for IHCs and three rows for O H C s The polarity of hair cells is given by the position of the k m o c d m m that is visible on most cells (arrows) The cutlcular plate of IHCs is topped by s t e r e o o h a (large arrow) of different length, vestigial calla and a k m o o h u m O H C s m comparison seem more mature, the cutlcular plate is almost devoid of remains of cilia and the s t e r e o o h a bundle shows the typical V shape Each O H C displays a k m o o h u m (arrows) on the surface of the cutlcular plate oriented toward the external side of the spiral lamina IHCs as well as O H C s are separated by supporting cells topped by n u m e r o u s mlcrovflh (thin arrows), and a klnocdlum Scale bar 10/xm B cochlear explant of a 18DG fetus maintained 10 D I V Supernumerary hair cells can be observed on both types of hair cells IHCs and O H C s The regular orgamzatlon seen m the control explant (see A) is disrupted and several modifications of the apical surface of the organ of Cortl are well visible In some cases, stereoclha bundle of several hmr cells are connected together (arrows) Scale bar 1 0 / x m C higher magmflcatlon at the level of IHCs S t e r e o o h a are graded in height with the tallest ones facing the kmocfilum Vestlgml small s t e r e o o h a (black arrows) are still present m the center of the stereocdm bundle T h e polarity of hmr cells shows variations between one to another O n e IHC presented an intriguing feature (large arrow) with stereocdla located m the opposite side m reference to the kmocfilum Smmese twins hmr cells are also well ws~ble (open arrows) They present two cuhcular plates connected together, as shown by the presence of two k m o o h a and the absence of supporting cells between them Notice the presence of kmoclhum on the supporting cells (thin arrows) Scale bar 5/,~m D higher magmficatlon at the level of a twin IHCs showing two k l n o o h a (large arrows) on the cutlcular plate separated by vestigml mlcrovfih (small arrow) that might correspond to the r e m n a n t surface of supporting cell between the two cutlcular plates This separation can be compared with a normal one (open arrow) Scale bar 5 / x m E. Higher magmficahon at the level of O H C s These cells present a CUtlcular plate topped by the typical V shaped stereocfim bundle and a kmoclhum A cell d~splays a stereocdla bundle of uniform height located m the center of the cutlcular plate (large arrow) The k m o o h u m , although not very ws~ble on th~s mlcrograph (black arrow) is m the center of the s t e r e o o h a bundle Supporting cells present n u m e r o u s mlcrowlll, their k l n o o h u m is stdl present (thin arrows) Scale bar 5 / x m F two O H C s with their two kmoclha (arrows) The cutlcular plates do not seem to be separated by supporting cells as shown by the absence of mlcrovflh (small arrow), compared with the surrounding hmr cells (open arrows) Scale bar 5 p,m
mlcrowlli and topped with the center of the cuticular (Fig. 5E). Supporting ceils m~crowlli and a kinocilium
~hdouh ('t a/ ,'Bratn Research 060 (1094) 181-101
a bundle of cilia located in plate also can be observed can be recognized by their in their center.
The results of the present study show a supernumerary production of hair cells m embryonic cochlear explants without loss or injury to preexisting cells. Previous studies dealing with organotypic cultures for studying cochlear development in mammals [16,38,40] have not shown supernumerary hair ceils. Sobkowicz et al. [40,41] reported a slight and occasional increase in the number of O H C rows in the apical turns of cochleae explanted from newborn mice. Kelley et al.  presented evidence of supernumerary hair ceils in embryonic explanted mouse cochleae when treated with retmoic acid. However, we recently demonstrated without adding drugs, that supernumerary hair ceils from the cochlear epithelium was closely related to the age and the origin of the tissue at the time of explantation . Based on this observation and the present one, the supernumerary hair cells can be classified into three responses: (1) an important overproduction occurring in the explants established from embryonic cochleae (Le 16-20DG) in the rat, (2) a slight hair cell proliferation when explanted around birth, (3) no supernumerary hair cell was seen when explantation was made some days after birth. From our observations, supernumerary hair cells do not depend on serum in culture medium that might contain growth factors or related molecules, since we obtained the same results with a culture medium without serum. Also, the supernumerary hair cells are independent of innervatlon because we obtained the same results with or without spiral ganglion neurons. However, interesting enough, supernumerary hair cells when present received fiber projections from the spiral ganglion neurons (see Fig. 4B). Thus, it appears that epithelial cells from K611iker's organ can develop a normal hair cell phenotype without direct contact with spiral ganghon neurites. In fact, cochleae explanted at 16DG, before hair cell differentiation , and despite lack of innervation, present a hair cell phenotype (Fig. 3A). This observation confirms previous findings in cultured mouse and chick otocysts [3,8,32,46] as m non-innervated sense organs of the lateral line in salamanders , where sensory hair cells can differentiate and develop without direct contact with neurites. However, our observations do not rule out that spiral ganglion neurites has absolutely no influence, because it is possible that before the 16DG, the few neurites below the sensory epitheha (personal observations) may have triggered the subsequent programmed developmental sequences of hair cells.
Development of the hair bundle Ls well documented m the bird and follows preczse steps [4.11,12,43,45]. In the mouse and the rat, hair bundles have mainly been studied after birth when the stereociha are already present [2,25]; few observations were made before thls stage about the earliest ontogenesxs of cilia . Our observations from phalloidin staining  and SEM studies  show that hair cells are not differentiated at 16DG in the rat and therefore cells of K611iker's organ do not present any signs of hair bundle. Then, when explantation is made at 16DG, despite the In v~tro condition, stereocllia are present on both types of hair ceils 7 days later: this suggests that the explant contains the appropriate information to independently direct the differentiation of hair cell stereociha bundles as observed in the embryomc avian cochlear organ in culture . Moreover, organotypic cultures can lead to the sophisticated development of stereocilia bundle as observed in vitro in the newborn mouse  and embryonic chick . However, in our case, with longer m vitro cultures and supernumerary hair cells, some peculiar stereocilia bundles can be observed despite the normal development of stereocilia bundles for many hair cells. The most frequent observation ~s the variation of polarity between hair cells (see Figs. 3 and 5). This contrasts with early development of the apical region of hair cell where the tallest stereocdia and the kinocdium present a correct polarity oriented toward the external border of the cochlea [34,50]. One proposed explanation is that certain hair cells are still immature after 6 to 10 DIV, and hair bundles may reorient later, as observed during the maturation of stereocilia in the chick cochlea . One more likely explanation in our case, is that the aberrant polarity of hair cells will persist, because one can already observe the same phenomenon two days after explantation without apparent amelioration a few days later (personal observation). In this case, cell polarity may be determined very early during differentiation by modifications of epigenetic factors  or during mitosis by modification of the axis of mitosis . The functional consequence of this aberrant geometry of stereocilia bundles is that the mechanosensitivity of hair ceils, which is polarized, would be different between cells in the same location. In fact, ~t is well known that deflection of the stereociha toward the kinocdium, and so in the direction of the tallest stereocilia, opens the mechanosensltive channels, and deflection in the opposite direction closes them [19,28]. So, this supernumerary hair cells may produce some kinds of aberrant functioning that are difficult to predict precisely at the present time. The mechanisms that govern the formation of the highly organized mammalian organ of Corti are stdl unknown. In the mouse, the cells that differentiate in hair cells were believed to complete their terminal
A Abdouh et a l / B r a m Research 660 (1994) 181-101
mitosis during a precise period of embryonic development , so their number is limited before birth. Although the same study has not been made in the rat, it is assumed that terminal mitosis is well finished before birth. Studies have demonstrated that many species of fish and amphibians produce hair cells throughout life [5,6]. Recent observations in avian auditory organs show that epithelium retains some capability to restore itself after acoustic damage [7,17,39] or drug ototoxicity [13,18] and that repair is undertaken by production of new hair cells [10,36]. It has also been shown that injured mammalian adult vestibular receptors can regenerate [15,47] after exposure to noxious agents. Reports on the regeneration of auditory receptors in mammals are scarce by comparisoja. A recent work showed that embryonic mouse explants presented proliferation of hair cells in vitro when retinoic acid was added to the culture medium , while another recent in vitro work claimed that rat neonatal auditory receptors might need retinoic acid for regeneration after antibiotic treatment . In any case, the regeneration of newly hair cells was dependent on a preceding loss, injury of preexisting hair cells or added drugs [10,22,24,36]. In our cases, no drug, no serum was added and no Injury was made, unless unintentionally during explantation by unnoticed traumatic mechanical manipulations, so we cannot exclude the possibility of some damage to the epithelium. However, when sensory epithelia is observed by phase contrast microscope after explantation, the sensory epithelium is still well organized. Moreover, observation of the organ of Cortl along with its lnnervation after a week in vitro, showed the sensory epithelium and its afferent innervation were well preserved  (Fig. 4). From our study, it seems that the capability of sensory epithelia to produce supernumerary hair cells depends on the developmental stage of the cochlea at which the explant is taken, as shown by the larger number of hair cells produced when explanted from embryonic cochleae  (Fig. 2). Although, we did not specifically study the origin of cell proliferation in this report, one can speculate however on the possible origin of this phenomenon. The supernumerary hair cells could be formed according to several possibilities. (i) There is maybe an overproduction of hair cells during the embryonic period. This overproductlon could be followed by a period of cell death. It could be possible that explantation and in vitro growing conditions would bypass this period, thus, producing the supernumerary hair cells. However, this hypothesis has not be tested for hair cells, and moreover, no supernumerary rows of IHCs and OHCS were observed during the early ontogenesls of the organ of Corti 
(ii) Supernumerary hair cells may develop lrom polypotent cells in the future sensory epithelium that changed their normal developmental fates in response to explantatlon and in vitro conditions. This would suggest that cues in the cell's environment can influence its final commitment to specialized hair cells [9,26]. The determination of cells as hair cell phenotype can be controlled by several factors such as lateral inhibitory cues between cells, mediated for example by cell adhesion molecules [30,33] Growth factors may act as developmental regulators after mitosis as postulated by Corwin et al. . (l) Supernumerary hair cells could be formed by cells already engaged in the way of differentiation at the time of explantation, but dedifferentiate and reenter the mitotic cycle for producing a symmetric differentiation resulting in what we called 'Siamese twins hair cells' and 'twin hair cells' (Figs. 3A and 5C,D). This observation may be related to a recent finding in the lateral line of two species of fish where hair cells appear to develop almost exclusively in pairs . (i) Division of a progenitor cell which differentiates asymmetrically as a hair cell and a supporting cell has been suggested for the avian vestibular organ  and the lateral line organ in salamanders . This hypothesis may be supported by our observations where a single hair cells with obvious signs of immaturity such as stereoclha of the same height and the central position of kinocihum between cilia  is present among supporting cells and more differentiated hair cells (Figs 3C, 5E,F). Although, our observation is not d~rectly related to auditory receptor regeneration, ~t ~s relevant because ~t shows that embryonic auditory sensory epithelium may produce extra hair cells without previous loss or injury of preexisting cells. It would be interesting to know what the factors are that trigger the supernumerary hair cells in embryonic explants m order to apply this reformation to the regeneration of hair cells in adult injured cochleae. Moreover, studying the progenitor cells and mechanisms by which they give rise to supernumerary hair cells along with the elucidation of the intrinsic a n d / o r extrinsic signals setting off th~s phenomenon will provide information about the genesis of the sophisticated organization of the organ of Corh in mammals.
Acknowledgements We thank Dr. M.-Y. Mu for sharing some data from the reference cochleae. The work was supported by 'La Fondatlon pour la Recherche MEdicale' as well as by l'Association Recherche et Partage'.
4 4bdouh et a l / Brain Research 660 ( 19941 181-191
References  Abdouh. A , Despr6s, G and Romand, R., Hair cell overproduction m the developing mammahan cochlea m culture, Neuroreport, 5 (1903) 33-36  Anmko, M., Postnatal maturation of cochlear sensory hairs m the mouse, Anat Embryol, 166 (1983) 355-368  Ard, M D . Morest, D K and Hauger, S.H., Trophlc interaction between the cochleovestlbular ganglion of the chick embryo and its synaptlc targets in culture, Neuroscwnce, 16 (1985) 151-170  Cohen, G M. and Cotanche, D A , Development of the sensory receptors and their mnervatlon in the cjatck cochlea In R. Romand (Ed), Del'elopment of Auditory and Vestibular System~. 2, Elsevier, Amsterdam, 1992, pp, 11/1-138  Corwm, J T, Production and aging of tuner ear cells m sharks, J Comp Neurol. 201 (19811 541-553  Corwln, J T., Perpetual production of hair cells and maturatlonal changes m hair cell ultrastructure accompany postembryonlc growth in an amphibian ear, Proc Natl Acad Sct USA, 82 ( 19851 3911-3915  Corwin, J T and Cotanche, D A , Regeneration of sensory hair cells after acoustic trauma, Science, 240 (1988) 1772-1774  Corwln, J T and Cotanche, D.A, Development of Iocahonspecific hair cell stereocdm m denervated embryonic ears, J Comp Neurol, 288 (1989) 520-537  Col-win, J T , Jones, J E , Katayama, A , Kelley, M.W. and Warchol, M.E, Hair cell regeneration the identities of progenitor cells, potential triggers and Instructive cues. In R Bock and J Whelam (Eds), Regeneration of Vertebrate Sensor)' Receptor Cells, Wiley and Son, New York, 1991, pp 103-119  Corwm, J.T, Warchol, M.E and Kelley, M.W, Hair cell development, Curr Opm Neuroblol, 3 (1993) 32-37  Cotanche, D A , Development of hair cell stereocflla in the avian cochlea, Hear Res, 28 (1987) 35-44  Cotanche, D A and Corwm, J T , Stereociha bundles reorient during hair cell development and regeneration m the chick cochlea, Hear Res, 52 (19911 379-402  Duckert, L G and Rubel, E W., Ultrastructural observations on regenerating hair cells m the chick basllar papilla, Hear Res, 48 (19901 161-182  Engstrom, H and Engstrom, B, Structure of the hairs on cochlear sensory cells, Hear Res, 1 (19781 49-66.  Forge, A , Ll, L, Corwln, J.T and Nevlll, G , Ultrastructural evidence for hair cell regeneration m the mammalian tuner ear, Scwn~e, 259 (1993) 1616-1619  Furness, D N, R~chardson, G P. and Russell, I J , Stereoclha bundle morphology m organotyp~c cultures of the mouse cochlea, Heat Re~, 38 (1989) 95-110  G~rod, D A , Duckert, L G and Rubel, E W , Possible precursors ot regenerated hair cells m the awan cochlea following acoustic trauma, Hear Res, 42 (1989) 175-194  Hashmo, E , Tanaka, Y, Salv~, R J, and Sokabe, M., Hair cell regeneration in the budgerigar after kanamyein ototoxicity, Hear Re~ , 59 (1092) 46-58  Hudspeth, A J and Corey, D P., SensltlWty, polarity, and conductance change m the response of vertebrate hair cells to controlled mechanical shmuh, Proc Natl Acad Set USA, 74 (19771 2407-2411  Jones, J E and Corwln, J T , Replacement of lateral hne sensory organs during taft regeneration m salamanders" identification of progenitors cells and analysis of leucocyte activity, J Neurosct, 13 (1993) 1022-1034  Jorgensen,J M and Flock, A , Non-innervated sense organs of the lateral hne development in the regenerating tail of the Salamander Ambystoma mexwanum, J Neurocytol, 5 (1976) 33-41  Kelley, M W , Xu, X - M , Wagner, M A , Warchol, M E and
  
Corwln, J T, The developing organ ot COlh contain,, retlnol~_ acid and forms supernumerary hair cells in ~esponse to exogenous retlnOlC acid in culture, Decelopment, 119 11993) 10411053 Klmura, R S, Hairs of the cochlear sensory cells and their attachment to the tectorlal membrane, Aeta Otolar),ngol 61 (19661 55-72 Lefebvre, P P , Malgrange, B, Staecker, H, Moonen, G and Van De Water, T R., Retlnolc acid stimulates regeneration of mammalian auditory hair cells, Science, 260 (1993) 692-695 Lenoir, M,, Puel, J - L and Pujol, R , Stereoclha and tectorlal membrane development in the rat cochlea A SEM study, Anat Emb~ol, 175 (1987) 477-487 Lewis, J , Rules for the production of sensory cells In R Bock and J Whelam (Eds), Regeneration of Vertebrate Semory Receptors Cells, Wiley and Sons, New York, 199l. pp 25-39 Lim, D J and Rueda, J , Structural development of the cochlea In R Romand (Ed.), Decelopment of Auditory and Vestibular Systems, 2, Elsevier, Amsterdam, 1992, pp. 33-58 Lowensteln, O. and Wersall, J , A functional interpretation ot the electron microscopic structure of sensory hairs in the cnstae of the elasmobranch Raja claL'ata m terms of directional sensitivity, Nature, 184 (1959) 1807-1808 Mazuro,~sky, E.B and Peterson, E R., Photoreconstltued collagen gel for hssue culture substrats, Exp Cell Res, 76 (19731 447-448 Raphael, Y , Volk, T, Crossm, K L, Edelman, G,E and Gelger, B, The modulation of cell adhesion molecule expression and Intercellular junction formation In the developing avian Inner ear, Dec Btol, 128 (19881 222-235 Rastek D., Abdouh, A , Dahl, D and Romand, R , An original organotyp~c culture method to study the organ of Cort. of the newborn rat in vitro, J. Neurosct Meth, 47 (19931 123-131 Raymond, J., In vitro differentiation o1 mouse embryo statoacoustic ganghon and sensory epithelium, Hear Re~, 28 (19871 45-56 Rodrlguez-Boulan, E and Powell, S K, Polarity of epithehal and neuronal cells, Annu Ret Cell Btol, 8 (1992) 395-427 Romand, R , Zme, A E A and Hafidl, A., Ontogenesls of F-actm m hair cells, Cell Motd Cvto~k, 25 (1993) 213-222 Rouse, G.W and Pickles, J O , Paired development of hair cells in neuroblasts of the teleost lateral line, Proe R Soc Lond B, 246 (1991) 123-128 Rubel, E W, Regeneration oI hair cells in the avian inner ear In Dancer et al (Eds), No~se-lndu¢ed Hear Loss, Mosby, 1992, pp 204-227 Ruben, R J , Development of the tuner ear of the mouse A rad~oautographic study of terminal mitosis, Acta Otola~ngol, Suppl 220 (1967) 1-44 RusseLI J and Richardson, O P , The morphology and physiology of hair cells in organotyplc cultures of the mouse cochlea, Hear Re~, 31 (19871 9-24 Ryals, B M and Rubel, E W, Hair cell regenerahon alter acoustic trauma m adult Coturnlx quad, Scwnce, 24(1 (19881 1774-1776 Sobkowlcz, H M., Bereman, B and Rose, J E , Organotyplc development of the organ of Cortl in culture, J Neurocytol, 4 (19751 543-572 Sobkowlcz, H M., Loftus, J M and Slapnic, S M., Tissue culture of the organ of Cortl, Acta Otolaryngol, Suppl. 502 (19931 3-36. Stone, J S and Cotanche, D A., Hair cell dffferentmtion m the developing chick cochlea and m embryonic cochlear organ culture, J Comp Neurol, 314 (1991) 614-625. TIlney, L G , Cotanche, D A and Tflney, M S, Actm filaments, stereocflla and hmr cells in the bird cochlea VI How the number and arrangement of stereoclha are determined, Det,elopment, 116 (19921 213-226
A Abdouh et al ~Brain Research 660 (1994) 181-191  Tllney, G L and Tllney, M S , Actm filaments, stereoclha, and hair cells how cells count and measure, Annu Rel' Cell Btol, 8 (1992) 257-274  Tllney, L . G , Tllney, M S, Saunders, J C and de Rosmr, D . J , Actm filaments, stereomha, and hair cells ot the bird cochlea III The development and differentiation of hair cells and stereoclha in embryos, Det' Btol, 116 (1986) 100-118  Van De Water, T R , Effects ot removal of the statoacoustm ganglion complex upon the growing otocysts, Ann Otol Rhmol Laryngol, 85 Suppl 33 (1976) 1-32  Warchol, M E , Lambert, P R , Goldstem, B.J, Forge, A and
Corwln, J T , Regenerative prohferatlon in inner ear sensory epltheha lrom adult guinea pigs and humans, Science, 259 (1993) 1619-1622  Wmsleder, P and Rubel, E W , Hair cell regeneration m the avian vestibular eplthehum, E.tp Neurol, 115 (1992) 2 - 6  Wulf, E , Deboden, A , Bautz F A , Faulstlch, H and Wmland, T , Fluorescent phallmdm, a tool for Vlsuahzatlon ol cellular actln, Pro~ Natl Acad S~* US4, 76 (1979) 4498 4502  Zme, A E A , Desprds, G and Romand, R , Ontogenesls of hair cell stereocJha m the neonatal rat, Abstract 4RO ~leetmg, Feb 1994