Evaluation of thyroid-mediated otolith growth of larval and juvenile tilapia

The accreting and metabolically inert otoliths of teleosts that function as the vestibular and hearing receptor have widespread research application(Begg et al., 2005). Despite the importance of otolith-related techniques and their rapid expansion in applied research, knowledge of the mechanisms of otolith formation and growth is still scarce. Campana (Campana,2005) states, after reviewing 862 recent otolith-related papers,that understanding the physiologically based otolith growth model is a major task for future study.

Otolith growth is influenced by exogenous factors, such as temperature(Volk et al., 1999) and feeding (Baumann et al., 2005)as well as by endogenous factors, e.g. thyroid hormones(Shiao and Hwang, 2004) and neuronal control (Anken et al., 2000). Among endogenous factors, hormones are important in regulating numerous physiological process and developmental events. Mugiya (Mugiya, 1990)demonstrated hormonal influence on otolith growth by hypophysectomizing goldfish (Carassius auratus). Among the hormones, thyroid hormones(THs) mainly function to control the growth, development, metabolism and homeostasis of vertebrates (Brent,1996). Shiao and Hwang (Shiao and Hwang, 2004; Shiao and Hwang, 2006) further confirmed that thyroid hormones are necessary for otolith growth during metamorphosis of leptocephalus tarpon (Megalops cyprinoides). They suggested a positive correlation between thyroid status and otolith growth based on the fact that hyperthyroidism increases otolith growth while hypothyroidism retards or even stops otolith growth during the metamorphosis of tarpons (Shiao and Hwang, 2004; Shiao and Hwang, 2006).

The spectacular metamorphosis of the tarpon, flounder and conger eel is driven by a thyroxine surge (Miwa et al.,1988; Yamano et al.,1991), which is also presumably responsible for the abrupt increase of otolith daily growth rate during leptocephalus metamorphosis(Shiao and Hwang, 2004; Shiao and Hwang, 2006). Most teleosts do not show dramatic changes of morphology from larval to juvenile stage. However, the metamorphosis of many teleostean larvae is also a TH-dependent event, as shown in goldfish (Carassius auratus)(Reddy and Lam, 1992),zebrafish (Danio rerio) (Brown,1997) and grouper (Epinephelus coioides)(de Jesus et al., 1998). Furthermore, gonadal maturation and reproduction may also be mediated by elevated thyroid status (Cyr and Eales,1996). Accordingly, otolith growth may change at certain thyroid-mediated life history events. Otolith growth and its microstructure are usually regarded as the proxy of fish growth and as the recorder of their life histories. So far, the physiological basis, especially for endocrine-mediated growth changes of the otolith is still poorly understood.

Normal embryo development and fish growth depends on the programmed secretion of thyroid hormones and the expression of thyroid hormone receptors(TRs) (Yamano and Miwa, 1998; Liu and Chan, 2002). Disruption of thyroid hormonal secretion causes malformation of developing organisms, including fish (Elsalini and Rohr, 2003). Furthermore, teleosts display unique and characteristic otolith growth patterns, especially during the early stage of development. How the programmed thyroid secretions affect the ontogenic otolith growth is still unclear since no study has simultaneously examined the TH content, TR expression and otolith growth. Therefore, this study aims to evaluate (1) the relationship between otolith growth pattern and the programmed secretion/expression of intrinsic TH and TR, (2) the effects of abnormal thyroid secretion on otolith growth and morphology, (3) the recovery of otolith growth from hypothyroidism. The experiments used tilapia larvae and juveniles (Oreochromis mossambicus) under a manipulated laboratory environment.

Tilapia [Oreochromis mossambicus, Cichlidae(Peters 1852)] larvae,juveniles and adults were reared in fresh water at 28±1°C under a 14 h:10 h light:dark photoperiod. Several pairs of adult tilapia were used to produce the larvae but the fish used in each experiment were from the same progenitors. Fertilized eggs were retrieved from the mouth of the female tilapia and incubated in the aquarium with strong aeration to make the larvae turn around for normal development. After the complete absorption of the yolk,larval tilapia was fed to satiation with commercial fish meal once a day in the morning and the remaining food was cleared after feeding in the afternoon. A stock of local running tap water in a circular tank was prepared for use. All tilapia larvae and juveniles were reared in tap water in 10 l aquaria with a density of approximately 25 fish per liter during the experimental periods. The water was moderately aerated and was replaced daily with fresh tap water containing the same concentration of chemicals. In the following experiments,fish were anesthetized with 0.1 mg ml^–1 MS-222 solution(3-aminobenzoic acid ethyl ester; Sigma, St Louis, MO, USA) before they were killed for measurement and analysis.

High concentrations of 3,5,3′-triiodothyronine (T₃; 100 p.p.m.; Sigma) were dissolved in absolute ethanol then diluted to 10 p.p.b. or 25 p.p.b. in fresh water for use. Ethanol alone had no effect on fish growth(data not shown). A stock solution of 10 000 p.p.m. thiourea (TU) was prepared by dissolving 10 g TU powder in 1 l fresh water. TU is an anti-thyroid hormone drug that inhibits the production of 3,5,3′,5′-thyroxine(T₄) and T₃ in the thyroid tissue. The stock solution was diluted to 300, 600 and 900 p.p.m. for use.

A batch of larvae from the same progenitors was reared in normal tap water until 30 days post-hatching (d.p.h.). Every 2 days after hatching, five fish were randomly picked for analysis and frozen in –80°C until T₃ quantification by radioimmunoassay (RIA; see below). Every 3 days after hatching, 8–10 fish were randomly selected for total RNA extraction and their TRα and TRβ mRNA were quantified by real-time PCR (see below). The remaining fish after 30 d.p.h. were used to examine otolith growth.

In order to evaluate the effects of T₃ and TU administration on thyroid hormone content of tilapia juveniles, the hatched larvae were reared for 13 days to the juvenile stage, then fish were randomly transferred to one of three aquaria that contained either normal tap water or tap water with 10 p.p.b. T₃ or 300 p.p.m. TU for another 13 days. Then the fish were stored at –80°C until T₃ quantification.

Hatched larvae were reared for 13 days to the juvenile stage. At 14 d.p.h.,juveniles were immersed in 300 p.p.m. tetracycline solution (Sigma; pH adjusted to 7.4) in the dark for 12 h to create a fluorescent mark in the otolith. Fish were subjected to treatment the next day (15 d.p.h.). The juveniles were reared in 0, 300, 600 and 900 p.p.m. thiourea (TU),respectively for 13 days until 27 d.p.h. Before the otolith was removed for observation, total length (from the tip of the mouth to the end of caudal fin)was measured using a digital caliper (Mitutoyo, Kawasaki, Japan). Fish was blotted on the Kimwipes tissue to remove the water on the body surface and wet mass was measured using a digital balance.

Another batch of larvae was also reared for 13 days and their otoliths were marked with 300 p.p.m. tetracycline at 14 d.p.h. as described above. At 15 d.p.h. the fish were subjected to 0 p.p.b. T₃, 0.1 p.p.b. T₃, 10 p.p.b. T₃, 0.1 p.p.b. T₃ + 300 p.p.m. TU or 10 p.p.b. T₃ + 300 p.p.m. TU for 13 days, until 27 d.p.h. All the fish from experiments 4 and 5 were sacrificed at 28 d.p.h. for morphological measurements i.e. total length (to 0.01 mm) and wet mass (to 0.01 mg), and for otolith examination.

Fertilized eggs were incubated until hatching. Larvae at 3 d.p.h. were immersed in 300 p.p.m. tetracycline solution for 12 h before the treatment. Then fish were reared in normal tap water, 25 p.p.b. T₃, 300 p.p.m. TU or 25 p.p.b. T₃ + 300 p.p.m. TU from 4–16 d.p.h. Fish were sacrificed for measurement of total length, wet mass, otolith and T₃ contents on 17 d.p.h. However, otolith and T₃contents were not measured for the group treated with 25 p.p.b. T₃+ 300 p.p.m. TU because of the loss of the samples during preparation.

Tilapia juveniles at 14 d.p.h. were immersed in 300 p.p.m. tetracycline solution for 12 h before being reared in 300 p.p.m. TU for 13 days. The fish were transferred to water without TU at 28 d.p.h. and the fish were immersed in 300 p.p.m. tetracycline solution for 12 h on 29, 31 and 33 d.p.h. (the second, fourth and sixth days after recovery from TU treatment). The fish were reared until 41 d.p.h. Total length and wet mass were measured and otoliths were extracted for examination.

Total RNA was extracted from the whole juvenile tilapia following the standard protocol using TRIzol reagent (Invitrogen, Carlsbad, CA, USA). Total RNA (5 μg) was reverse-transcribed for cDNA synthesis using a kit (Qiagen,Hilden, Germany). The resulting cDNA was amplified by PCR in a total volume of 50 μl with 2 μl cDNA template, 5 μl of 10× PCR buffer (100 mmol l^–1 Tris–HCl, 500 mmol l^–1 KCl, 15 mmol l^–1 MgCl₂), 5 μl of 2.5 mmol l^–1 dNTP, 2 μl of forward and reverse primers, 0.5 μl Taq DNA polymerase (5 i.u. μl^–1) and 35.5 μl distilled water. We used the same specific primer sequences as Marchand et al.(Marchand et al., 2001) for the species Oreochromis niloticus. For the TRαamplification, the forward primer sequence was 5′-GCTGCATCATCGACAAGATC-3′ and the reverse primer sequence was 5′GATCTGAGCTCATGAGAAGC3′. For TRα amplification,the forward primer sequence was 5′AATGTGTTATTGACAAAGTG3′ and the reverse primer sequence was 5′-GATCGGATGAAAGCAGGATA-3′. The amplified cDNA fragments were inserted into the pGEM-T easy vector (Promega,Madison, WI, USA) and transformed into competent cells (ECOS9-5) for amplification. The purified plasmids were subjected to DNA sequencing using an automatic DNA sequencer (ABI 3700, Applied Biosystems, Wellesley, MA, USA). Quantitative real-time PCR (qPCR) was carried out using a SYBR Green dye(Qiagen, Hilden, Germany)-based assay with an ABI Prism 7000 Sequence Detection System (Perkin-Elmer, Applied Biosystems, Wellesley, MA, USA)according to the manufacturer's instructions. Primers targeting the TRα and TRβ and the endogenous control gene,β-actin, were designed using Primer Express 2.0 software (Applied Biosystems). In each assay, 25 ng cDNA was amplified in a 20 μl reaction containing 2×SYBR Green Master mix, 300 nmol l^–1 of forward and reverse primers, and nuclease-free water. The primers designed for the consensus of TRα and TRβ isoforms were as follows: TRβ-forward: 5′-GCTCAGGGCTCACAGTGGAA-3′, TRα-reverse: 5′-AACGACACGGGTGATGGC-3′; TRβ-forward: 5′-GGCAACCACTGGAAGCAGAA-3′, TRβ-reverse:5′-TGATAATTTTTGTAAACTGACTGAAGGCT-3′.

Sagittal otoliths were removed under a stereomicroscope (Olympus SZX 12,Tokyo, Japan), dried in air and embedded with Epofix resin (Struers,Copenhagen, Denmark). The embedded otoliths were then sectioned using a low-speed circular saw (Buehler Isomet, Evanston, IL, USA) to remove excess resin. The otoliths were then ground and polished on a grinder-polisher machine (Buehler Metaserv 2000, Evanston, IL, USA) at a speed of 300 r.p.m. with wet-polisher paper of 2 000-grit for the initial grinding and 2 400-grit for the final grinding until the core was exposed. The otolith was finally polished with a polishing cloth and 0.05 μm alumina (Buehler) to smooth the surface. During the grinding and polishing process, the otolith was periodically checked under a compound light microscope. A fluorescence microscope (Axioplan 2 Imaging MOT, Zeiss, Germany) with incident light from a 50 W mercury lamp and FITC filter sets was utilized to detect the fluorescent ring on the ground surface of otoliths. Then, dilute HCl (0.05 mol l^–1) was used to etch the otolith for 20 s. Etching increased the visibility of otolith increments by enhancing the contrast under a compound light microscope. Images of fluorescent rings and the whole etched otolith were recorded at 200× magnification under the light microscope equipped with a digital camera (AxioCam HRm Zeiss). From the images,measurements were made of otolith length and the width of individual increments, as well as counts of daily growth increments and these were processed on a personal computer using Image-Pro plus software (Media Cybernetics Inc. 1994, Silver Spring, MD, USA). The measurements were made along the maximum radius from the core to the posterior end of each otolith. To observe the otolith topology, the sagittal otoliths were removed from the fish, dried in the oven, and gold coated for observation by scanning electron microscopy (FEI Quanta 200 SEM, FEI, Hillsboro, OR, USA).

Data were expressed as means ± s.d. (N=number of fish) but means ± s.e.m. were used in the Figs 1, 3 and 4 for clarity. Correlations between T₃ and daily growth increment (DGI) profiles as well as T₃ profiles and TRα mRNA expression patterns were analyzed by Pearson product moment correlation. Statistical differences among treatments were analyzed using one-way analysis of variance (ANOVA). Tukey's pairwise comparison was used to identify groups that differed from others if the data satisfactorily met the assumptions of normal distribution and equal variance. Otherwise, the Kruskal–Wallis test on ranks and Dunn's pairwise comparison were used to isolate the groups that differed from the others. Statistical significance was set at α<0.05.

T₃ contents of tilapia changed considerably from early larval to juvenile stage. The temporal profile during the experimental period between 3 and 30 d.p.h. was classified into five phases. Phase 1: a gradual decline during the larval stage from 3 to 9 d.p.h. Phase 2: rapid increase from 11–15 d.p.h., which corresponded to the transition of the tilapia from larval to juvenile stage. During the transition from the larval to juvenile stage, the yolk is completely absorbed and the fish starts active feeding and free swimming (Holden and Bruton,1992; Anken et al.,1993). Phase 3: a high plateau around 15–21 d.p.h. Phase 4:a second decline from 21 to 25 d.p. Phase 5: a minor increase from 25 to 30 d.p.h. (Fig. 1).

The trend of otolith growth was generally similar to the T₃profile. The width of the daily growth increment (DGI) slightly declined from 3 to 7 d.p.h., followed by a gradual increase to approximately 3 μm around 14 d.p.h. A high growth plateau was maintained until 22 d.p.h. then the daily otolith growth gradually decreased to approximately 1.3 μm by 29 to 31 d.p.h. (Fig. 1). A significant median correlation (correlation coefficient=0.681, P=0.021) occurred between the T₃ and DGI profiles during the period from 3 to 25 d.p.h.

The partial DNA sequence of tilapia TRα and TRβ genes were cloned and sequenced. The cloned sequences are 703 nucleotides and 730 nucleotides for TRα and TRβ, respectively. The sequences were submitted to GenBank with the accession no. EU048544 for TRα and EU048545 for TRβ. The sequences were confirmed from the NCBI database and phylogenetic tree analysis of the TR family from published species. Quantitative analysis of the TRα mRNA expression suggested a trend similar to the T₃ contents and otolith growth; decreasing from 3 to 9 d.p.h., followed by an increase until 15 d.p.h., then gradually decreasing until 27 d.p.h. However, TRβ mRNA expression was low,with no evident variation throughout the experimental period(Fig. 2). A significant median correlation was also observed for TRα and DGI profiles during the period from 6 to 27 d.p.h. (correlation coefficient=0.63, P=0.021).

Compared with the control group (1.4±0.01 ng g^–1),300 p.p.m. TU treatment caused a significantly low T₃ content(0.3±0.07 ng g^–1) in tilapia juveniles but immersion in 10 p.p.b. T₃ significantly increased the T₃ contents(9.7±2.13 ng g^–1) of tilapia juveniles by approximately sevenfold (P<0.05).

Tilapia juveniles, immersed in 300, 600 and 900 p.p.m. TU for 13 days,showed a retarded otolith growth in a dose-dependent manner(Table 1). The new growth in otolith radius was approximately 56%, 43% and 41% of the control group for the 300, 600 and 900 p.p.m. TU group, respectively. The newly formed otolith radii of TU groups were significantly smaller than for the control group (all P<0.05). Fish growth, i.e. TL and mass, was also significantly retarded by TU treatment. By contrast, 10 p.p.b. T₃ moderately increased the TL and mass and significantly increased the DGI(P<0.05), but 0.1 p.p.b. T₃ did not promote fish growth(Table 2). However, a high dose of T₃ (10 p.p.b.) compared with a low dose of T₃ (0.1 p.p.b.), slightly, but not significantly, increased otolith growth in the presence of 300 p.p.m. TU. This result was in general agreement with previous observations of TU-induced effects on tilapia larvae yolk absorption, growth and development (Reddy and Lam,1992). The smaller otolith growth was attributed to fewer newly formed DGI (an average of two to four fewer rings; Table 3) and a slower otolith growth rate compared with the control group (see below).

Higher concentrations of TU, i.e. 600 p.p.m. and 900 p.p.m. evidently caused higher mortality of tilapia juveniles. Lower TU i.e. 300 p.p.m., 0.1 p.p.b. and 10 p.p.b.T₃ have no evident effects on fish survival compared with the control treatment (Tables 1 and 2).

A prominent growth difference of normal tilapia juveniles between experiments 3 (Table 1) and 4(Table 2) was noticed. The larvae in each of these experiments were produced from different adults. The condition of the progenitors – sizes, nutrient levels and health –may affect development of their embryos.

For tilapia larvae, the simultaneous administration of 300 p.p.m. TU and 25 p.p.b. T₃ had no significant effect on fish or otolith growth. TU(300 p.p.m.) also had no negative effect on fish or otolith growth. T₃ of 25 p.p.b. significantly increased the otolith growth but not the DGI number. T₃ content of tilapia larvae were also significantly increased (approximately threefold) by 25 p.p.b. T₃administration, but not significantly reduced by 300 p.p.m. TU(Table 4). In preliminary trials, 10 p.p.b. T₃ had no effect on fish and otolith growth (data not shown).

The mortality of larval tilapia was higher than that of juveniles. However,compared with the control group, 300 p.p.m. TU and 25 p.p.b. T₃have no evident effect on larval survival(Table 4).

TU treatment retarded otolith growth and slightly changed otolith morphology. After 13 days immersion in 300 p.p.m. TU, otolith growth was severely retarded on the ventral area (Fig. 3C,D,F). In most samples, the ventral part of the otolith stopped growing shortly after the beginning of the 300 p.p.m. TU treatment although otolith growth in other directions continued for another 6–9 days(Table 3). Otolith growth in the ventral direction did not resume even after fish were returned to normal water for 2 weeks. This indicated that TU treatment not only caused a temporary inhibition but also permanent damage to this organ, at least in the ventral area. The severe reduction of otolith growth in the ventral direction resulted in a flat margin (Fig. 3F). The normal otolith showed the convex margin in both ventral and dorsal directions. Furthermore, every tilapia larva and juvenile was successfully marked by tetracycline. All fluorescent rings were very distinct in the normal otolith (Fig. 3B)but in the experimental group, however, the second, third and fourth fluorescent rings, laid down during the recovery period, were less discernible, faint or only partially visible compared with the first prominent fluorescent ring laid down before exposure to 300 p.p.m. TU(Fig. 3D). This result indicated that CaCO₃ mineralization in the inner ear had not recovered to a normal level during the first week of recovery from hypothyroidism.

TU-treated fish showed a slower otolith growth rate than their counterparts in normal water (Fig. 4). There were one, six, eight and nine fish whose otolith only had 9, 10, 11 and 12 newly deposited rings, respectively, during the 13 day TU treatment(23–27 d.p.h.). A probable explanation for this was that respective otoliths stopped growth after 9 (N=1), 10 (N=6), 11(N=8) and 12 (N=9) days of TU treatment. After the TU-treated fish were returned to normal water, otolith growth was observed on the second day in all individuals, as determined by the tetracycline marking. It was unclear if the growth occurred on the first day during the recovery period since the fish was not marked by tetracycline on that day. Only three DGI were discernible among the three fluorescent rings deposited in the 5 day period. This indicated that otolith increase did not have a daily cycle during the approximately 1 week recovery from TU treatment. Furthermore, the otolith growth rate of the experimental group was only half that of the control group during the first 5 days of recovery. The otolith growth rate of TU-treated fish did not reach the same level as the control group after recovery for 1 week and remained approximately 2 μm day^–1 slower than that of the control group. An otolith growth increment was not formed daily until approximately 1 week after recovery from the 300 p.p.m. TU treatment.

This study demonstrates that somatic and otolith growth show differential susceptibility to thyroid hormone levels at different developmental stages of tilapia larvae and juveniles. In this study, we only measured the functional thyroid hormone, T₃ since T₄ secreted in fish, as in other vertebrates, has to be converted into T₃ in order to bind to a nuclear receptor and exert its full biologic activity. We also only administrated T₃ to tilapia since T₄ supplementation via food had no effect on plasma T₃ and T₄concentrations in tilapia (Geyten et al.,2005).

T₃ and TU treatments have no evident effects on somatic and otolith growth of tilapia larvae. As tilapia grows, T₃ treatment induces somatic growth and otolith growth during the juvenile stage, which can be retarded by TU treatment. TU (300 p.p.m.) and T₃ treatment does not cause an acute stress to fish, which is supported by the absence of any evident marks or checks in their otolith. We did not observe behavioral differences in T₃-treated fish compared with the control fish. Hypothyroidism induced by TU inhibition of the synthesis of thyroid hormones is progressive until the existing thyroid hormones are degraded. After TU treatment for several days, tilapia juveniles, but not the larvae, clearly show symptoms of hypothyroidism: less mobility, with stiff swimming behavior near the bottom and slow response to disturbances. The appetite of tilapia juveniles was also evidently affected after immersion in TU for several days. Reduced feeding slows somatic and otolith growth, but this factor alone cannot cause the cessation of otolith growth, particularly on the ventral side of the otolith after a few days of TU treatment. Furthermore, the retarded somatic and otolith growth is only partially counteracted by T₃administration, suggesting TU is toxic to tilapia juveniles. Extrathyroidal effects of TU, although not fully understood, have been found in killifish(Chambers, 1953), rainbow trout (Eales, 1981) and flounder (Schreiber and Specker,1999). Furthermore, TU-like goitrogenic phenythiourea evidently delays the hatching of zebrafish (Elsalini and Rohr, 2003). TU was found to reduce protein synthesis in the liver and muscle of freshwater catfish (Heteropneustes fossilis)(Singh, 1979). If TU also inhibits protein secretions of otolith chambers, this would retard and change the otolith shape during the process of biomineralization as observed here in tilapia juveniles. Nevertheless, tilapia larvae (this study), metamorphosing flounder (Miwa and Inui, 1987)and metamorphosing tarpon leptocephali(Shiao and Hwang, 2006) are insensitive to TU, suggesting that the extrathyroidal effects of TU in teleosts may vary in different species as well as developmental stages.

Experiments on salmon have suggested that otolith growth is more closely correlated with metabolic rate than with somatic growth(Wright, 1991; Yamamoto et al., 1998). The coupling of metabolic rate and otolith growth can be detected as early as the embryonic stage of zebrafish (Bang and Grønkjær, 2005). At similar temperatures, developing zebrafish embryos can show significant intraspecific variation in metabolic rate (Bang et al., 2004). The relationship between metabolic rate and thyroid level are not fully known in fish. Some studies suggested that thyroid hormones do not increase metabolic rate in fish (Weirich et al.,1987; van Ginneken et al.,2007). Accordingly, thyroid hormones may increase otolith growth by stimulating somatic growth and differentiation rather than by directly enhancing the metabolic rate of the fish. Kobuke et al.(Kobuke et al., 1987) first reported the presence of thyroid hormones in fish eggs, and their entry into tilapia oocytes is probably via diffusion from the maternal plasma(Tagawa and Brown, 2001). Maternal thyroid hormones are involved in the regulation of development and growth of teleosts (Tagawa and Hirano,1987; Brown, 1997). However, T₃ has no biological function without binding to thyroid hormone receptors. Somatic T₃ content and TRα mRNA expression decrease during the larval stage in tilapia. The decrease in TRα mRNA expression suggests a low demand for T₃ by tilapia larvae and this phenomenon may explain why the overdose of T₃ induced by 25 p.p.b. T₃ administration has no prominent effects on somatic growth of tilapia larvae(Table 4). Differentiation of the thyroid gland can be detected in tilapia larvae as early as 3 d.p.h. by histological staining (S.-M.W., unpublished data), which is similar to the observation in zebrafish (Elsalini and Rohr, 2003). Therefore, the thyroid gland may start to synthesize T3 at an early stage in tilapia. In experiment 5, larval tilapia were reared in the presence of 300 p.p.m. TU from 4–16 d.p.h. TU, like phenylthiourea (Elsalini and Rohr,2003) can almost completely inhibit the new synthesis of thyroid hormones. The measured thyroid content in TU-treated fish was 0.8±0.3 ng g^–1 (Table 4), which is not significantly different from the normal fish(1.4±0.4 ng g^–1, Table 4). This result suggests that the maternal thyroid hormones in TU-treated fish may be still active, at least until 16 d.p.h., and be sufficient to supply larval development. In addition, the newly synthesized T₃, if there is any, is limited in larval tilapia since TU does not significantly reduce the somatic T₃ content of tilapia larvae(Table 4). Although the transition of tilapia larvae to juvenile does not involve a striking morphological change, as observed in flounder and eel, a thyroid hormone surge and increasing expression of TRα mRNA are found during larval metamorphosis at 9–15 d.p.h. (Figs 1, 2). This indicates that thyroid hormones have an evolutionarily conserved function in manipulating the metamorphic process from larval to juvenile stage in teleosts. The coupling of thyroid hormone surges and high TRα mRNA expression may stimulate tilapia growth and development, which is reflected in faster otolith growth. The disruption of T₃ levels, either by hypothyroidism or hyperthyroidism, at this stage can easily change somatic growth, and the changes are recorded in otolith growth.

After removal of TU, synthesis of thyroid hormones is gradually resumed, so that fish as well as otolith growth slowly recovers. There were only three tetracycline-marked increments deposited in the otolith(Fig. 3C,D) during an approximate 6 day period. This result suggests that an otolith growth increment is not deposited daily but in a 2 day cycle during the first week of recovery. This suggests that fish age may be underestimated due to the uncoupling of otolith growth increments and the real daily age of the fish under conditions of hypothyroidism. Thyroid hormone levels take about a week to return to normal based on observations of the otolith growth period. It is worth noting that otolith growth on the ventral side is not resumed after TU removal. The otolith grows by CaCO₃ mineralization on the protein matrix, which is directly regulated by the epithelial cells of the otolith sac(Payan et al., 1997; Payan et al., 1999; Takagi, 2000; Tohse and Mugiya, 2001). Failure to grow indicates that the function of the otolith sac might be permanently damaged on the ventral side when fish are exposed to 300 p.p.m. TU for 13 days. Ionocytes on the otolith sac are the cells responsible for transepithelial transport of HCO^–₃ and Ca²⁺ into the endolymph (Mugiya and Yoshida, 1995; Shiao et al., 2005), although Payan et al.(Payan et al., 2002) suggested a passive diffusion through a paracellular pathway for Ca²⁺transportation. It is likely that TU treatment causes death or dysfunction to some ionocytes, leading to severely retarded otolith growth in the ventral direction. Disruption of ion and protein supplies from the plasma or cells into the endolymph may also cause the allometric growth of the otolith during hypothyroidism. These hypothetical explanations require further studies to determine the effects of hypothyroidism on the otolith sac at the cellular and molecular levels.

To our knowledge, this is the first study revealing consistent patterns among T₃ content, TRα expression, and otolith growth during the early stages of fish development. Yamano and Miwa(Yamano and Miwa, 1998) found ubiquitous expression of TR genes in the fish body, suggesting that development of each tissue of the flounder is controlled by thyroid hormone at the receptor level. Under TU-induced hypothyroidism, the slower otolith growth rate could be attributed to the retarded somatic growth, whereas the permanent changes on the ventral side of the otolith might be due to the cellular toxicity of TU. Dramatic changes in otolith increment width often occur at larval metamorphosis and settlement(Victor, 1986; Sponaugle and Cowen, 1994; Shiao et al., 2002). These changes cannot simply be attributed to environmental factors. This study demonstrates that otolith growth is influenced by T₃, especially during the larval to juvenile stages. To a certain extent, the characteristic ontogeny of otolith growth for each species may be also determined by programmed T₃ secretion and TRα expression.

The present work was financially supported by the Thematic Project of the Academia Sinica, contract no. AS-95-TP-B05 and by National Taiwan University to J.-C.S. We thank B. M. Jessop for suggestions on the draft.