Our current approach has been to delineate regulated zebrafish genes in order to provide direction for future investigations into auditory hair cell regeneration in zebrafish and mammals. Distinct patterns of gene expression were evident two and four days after acoustic trauma, suggesting that sound-induced damage in the zebrafish inner ear is a good model system for understanding pathways involved in hair cell regeneration. Transcripts showing the most dramatic regulation over the time course of our study include transcripts encoding growth hormone, major histocompatibility complex, class I, ZE, a light chain myosin, a heavy chain myosin, and a protein similar to atrial myosin light chain (zgc:66286).
The short time period within which these transcripts were examined following acoustic trauma coincided with a sharp increase in cell proliferation and partial recovery of hair cell bundle density, which was observed in our previous experiment with zebrafish , suggesting that these genes, as well as others listed in the datasets, may play a role in the regulation of cell proliferation and/or cellular repair. Genes associated with transport, kinase activity, transcription factor activity, signal transduction, hormone activity, nucleobase, nucleoside, nucleotide and nucleic acid metabolic process, extracellular region, cellular component, and calcium ion binding were also significantly regulated during this time period. However, a number of genes could not currently be assigned to any process or functional category. The roles of these transcripts during hair cell repair and regeneration remain undetermined. Further work is needed to elucidate the specific roles of many of the genes uncovered in this study.
A. Role of growth hormone in hair cell regeneration
Mammalian growth hormone (GH) and insulin-like growth factor 1 (Igf1) affect growth in postnatal animals through independent and common pathways , influencing final stature [39, 40] and facilitating neuron development and survival . No previous study has been published concerning the effect of growth hormone in the inner ear, but other growth-related factors are known to affect hair cell production and survival in mammals. Igf1-null mice exhibit altered inner ear maturation, abnormal innervation of the sensory cells in the organ of Corti, and increased apoptosis of cochlear neurons . Vestibular hair cell proliferation can be stimulated in mammals through exposure to transforming growth factor-alpha and epidermal growth factor . The zebrafish homologs of these genes were not listed among the differentially regulated transcripts in our study, but gh1 was dramatically upregulated 64-fold on Day 2 and remained upregulated over five-fold on Day 4, indicating that growth hormone played a prominent role in post-sound exposure recovery of the inner ear of zebrafish. We speculated that the activity of gh1 in the zebrafish might induce proliferation in the ear, since administration of growth hormone can increase cell proliferation in cultured trout leukocytes  and increase body mass in zebrafish . To characterize the effect of gh1 on cells of the inner ear, we injected zebrafish with salmon growth hormone. The significant increase we observed in cell proliferation in non-sound-exposed zebrafish inner ear following injection with growth hormone suggests at least three things: 1) growth hormone has the ability to stimulate proliferation in the inner ear of zebrafish, 2) under normal conditions, cells of the utricle, a vestibular organ, are more sensitive to growth hormone-mediated signaling than are cells of the saccule, and 3) that the rostral portion of the saccule may be more sensitive than the caudal portion. The difference in growth hormone sensitivity between the zebrafish utricle and saccule, and the potential difference between the rostral and caudal portions of the saccule, was unexpected but intriguing.
A difference in growth hormone sensitivity may reflect differences in proliferative capacity among the inner ear end organs. The regenerative capacity of hair cells in the fish utricle and lagena has not yet been determined. However, other non-mammalian vertebrates are capable of regenerating hair cells in both the vestibular and auditory portions of the inner ear [3, 4, 10, 14, 16], and rates of proliferation differ in vestibular and auditory systems in the absence of damage. For instance, supporting cells in the auditory portion of the chick inner ear, the basilar papilla, are normally quiescent in the absence of a damaging stimulus . Conversely, hair cells in the vestibular organs of chicks have a relatively short life span (approximately 2-6 weeks), undergoing spontaneous apoptosis and replacement though proliferation and differentiation of epithelial supporting cells [47–50]. Hair cells in the mammalian auditory system do not regenerate, but vestibular hair cells exhibit a limited regenerative capacity [51–54]. Low levels of apoptosis occur throughout the development of the zebrafish saccule , but no data comparing the rate of apoptosis in the uninjured zebrafish saccule and utricle is currently available. This would be useful in elucidating whether the dissimilar sensitivity of different portions of the zebrafish inner ear to proliferation corresponds with dissimilar rates of apoptosis. In noise-exposed goldfish, apoptosis peaked in the saccule one day before a peak in apoptosis in the lagena, suggesting that patterns of cell damage can vary between different endorgans of the teleost ear .
Interestingly, although gh1 was up regulated approximately 64-fold at 2 dpse in our experiment with zebrafish, Gh1 in the rat cochlea is down-regulated two-fold following temporary threshold shift induced by noise-exposure, and to a smaller extent following permanent threshold shift . Previous studies with mammals have shown that acute or chronic stresses can reduce GH levels in blood serum as well as in the brain [56–58]. While it is possible that exposure to sound produces a different growth hormone regulatory response in mammalian and non-mammalian vertebrates like zebrafish, it is more likely that this is due to differential timing of sound exposure. In the rat cochlea experiment, noise exposures were only for 90 minutes whereas in the current study, zebrafish were exposed for 36 hours followed by a two day recovery period. Thus, in zebrafish there could have been a decrease in GH during the initial stress of the acoustic exposure, followed by a subsequent increase during the recovery and regeneration phase. Future experiments with more time points following acoustic trauma are needed to determine this.
B. Other transcripts associated with cell proliferation
Other genes that may have been up- or down- regulated in order to enable cell proliferation include signal transducer and activator of transcription 1 (stat1), stearoyl-Coa desaturase (scd), diacylglycerol O-acyltransferase (dgat), and major histocompatablility complex class II (MHC II) genes. The function of stat1 may be connected with gh1, as it is in mammals, since growth hormone is known to activate signaling pathways that include STAT proteins [Figure 4; . The STAT activation process is transient and influences a broad range of physiological processes depending on the activating ligands and tissue type . The STATs that are activated by growth hormone exposure can vary by cell type, possibly contributing to the specificity of the growth hormone response .
Proteins Scd and Dgat appear to regulate lipid biosynthesis, and possibly phospholipid membrane synthesis. Proliferation depends in part on the ability to incorporate oleate with free long-chain fatty acids in order to form membrane phospholipids . Since the Scd protein synthesizes the oleate necessary for the biosynthesis of membrane phospholipids , the Danio rerio scd gene may be up-regulated on day two in order to increase production of membrane phospholipids as required by cell proliferation. The protein encoded by dgat, another gene up-regulated at 2 dpse and down-regulated between days 2 and 4, also participates in the regulation of membrane lipid synthesis. DGAT proteins interact with diacylglycerols, which are common intermediates for both triacylglycerol and phospholipid synthesis. DGAT tips diacylglycerol toward triacylglycerol synthesis. For instance, in vitro overexpression of DGAT1 gene in human lung SV40-transformed fibroblasts reduces synthesis of the membrane phospholipids phosphatidylcholine, phosphatidylethanolamine, and sphingomyelin by 30-40%, and reduces cell growth rate . It is not clear why dgat was upregulated on Day 2, given that cell proliferation peaks at this time, but one possibility is that up-regulation of dgat occurred as part of the system to regulate proliferation.
Several genes associated with immune function were identified in the microarray. These genes may play roles in cell proliferation following apoptosis. Major Histocompatibility (MHC) class II molecules are found on professional antigen-presenting cells such as macrophages, dendritic cells and B cells. MHC class II molecules are observed in the cochlear cells of adult mice following a damaging event and may promote cell proliferation in the inner ear of organs that possess proliferative capability .
Deoxyspergualin, a drug that inhibits de novo cell surface expression of MHC class II antigens, blocks cell proliferation in the kidney . Zebrafish MHC complex class II integral membrane alpha chain gene (mhc2a) was significantly regulated on 2 dpse and between days 2 and 4 dpse. Even more notable is MHC complex, class I, ZE (mhc1ze), which was down-regulated more than 67-fold on 2 dpse, but was not significantly regulated by 4 dpse. At this time, the function of mhc1ze has not been determined, but since MHC class I proteins are involved in antigen presentation on nearly all cell types in mammals, it seems probable that mhc1ze functions similarly in zebrafish. Antibodies that bind human MHC Class I molecules (HLA) and prevent them from presenting antigens induce increased proliferation of airway epithelial cells . Down-regulating mhc1ze in zebrafish may have a similar effect, encouraging proliferation by the reduction of antigen presentation.
It is not surprising that genes related to immune function were regulated following acoustic trauma since macrophages, a type of leukocyte, are recruited to sites of damage and may be involved in initiating wound healing and repair . Within hours of trauma to hair cell sensory epithelium, macrophages and other leukocytes are recruited to the area of damage. This has been reported in the lateral line of amphibians  and zebrafish , avian inner ear sensory epithelia [71–73], and the mammalian organ of Corti . Macrophages recognize and destroy cells undergoing apoptosis via phagocytosis  and may secrete substances such as growth factors that could affect cell proliferation and other functions [68, 76]. It has long been recognized that there is an interaction between the endocrine and immune systems in mammals. This appears to be true in fishes as well, and GH may be an important mediator between the two systems. For example, plasma GH levels and phagocytic activity are positively correlated in brown trout (Salmo trutta) during sea-water transfer [77, 78], and GH causes proliferation in leukocyte cultures of chum salmon, Onchorynchus keta . Reciprocal effects are also evident. Stress induces a rapid decrease of plasma GH levels in several fish species [80–82].
Another group of proteins that were highly regulated in our dataset were myosins. The most highly regulated were atrial myosin light chain (zgc:66286, -30 fold on Day 2, and -62 fold on Day 4) and slow muscle myosin heavy chain, like (smyhc1l, -36 fold on Day 2, and - 33 fold on Day 4). Mutations in non-muscle myosins MYH9, MYH14 and myosin VIIa have been implicated in deafness in mammals [83–85]. Myosins are a large superfamily with many shared domains among the members and are important regulators of the actin cytoskeleton, a prominent component of hair cell bundles. A large number of different myosins are expressed in developing neurons and sensory cells, helping to carry out a range of functions including morphogenesis, axonal transport, and synaptic and sensory functions [reviewed in , although the functions of many myosins are not known .
It is not clear why smyhc1l was down-regulated following acoustic trauma; however, smyhc1l may play a role in the regulation of immune response in the inner ear. Smyhc1l is a TMPIT-like protein, which is induced by TNF-alpha . Since TNF-alpha is a cytokine involved in inducing immune response, apoptosis and inflammation , it is reasonable to assume that the down-regulation that we see in smyhc1l may be associated with the down-regulation in TNF-alpha and other cytokines that one would expect during the recovery from inflammation. In support of this, a number of genes that are negative regulators of immune response were up-regulated two days post-trauma, including TCF family B cell activation factor (TC277656), C1q tnf1 protein (TC276192), and complement C1q tumor necrosis factor-related protein 4 precursor (TC298139; Additional file 1).
Atrial myosin light chain (zgc:66286) possesses an EF-hand domain . EF hands are a superfamily of calcium sensors and calcium signal modulators. Calcium-binding proteins such as calretinin, calmodulin, and parvalbumin have been used as markers for inner ear ganglion neurons and hair cells [90–94]. Calmodulin is known to mediate inflammation, apoptosis, immune response, and cell cycling [95, 96], but it is unclear at this point if the calcium-binding properties of atrial myosin light chain are serving similar roles in the zebrafish inner ear.
C. Genes associated with induced hair cell regeneration in mammals
Zebrafish homologs of genes that have been used to induce hair cell regeneration in mammals, specifically, cyclin-dependent kinase inhibitor p27(kip1)/cdkn1b, retinoblastoma1 (rb1), and atonal homolog 1 (atoh1) were found to be regulated at the P-value 0.05 level, but not at fold changes ≥1.4. Two days following sound exposure, cdkn1b was down-regulated slightly (-1.12-fold, see Additional file 1), while cdkn1b and rb1, both suppressors of cellular proliferation, showed up-regulation (1.60- and 1.36-fold, respectively) at 4 dpse following the peak in proliferation (Additional file 3). A similar pattern was evident for atoh1, which was down-regulated at 2 dpse (-1.20 fold) and up-regulated at 4 dpse (1.24 fold). Thus, more work will need to be done to rule them out as players in the process of proliferation and differentiation of zebrafish hair cells.
In this study, we used RNA isolated from whole ear tissue because of the very small size of the sensory epithelium of the zebrafish inner ear. RNA collected only from sensory maculae or specific cell types may reveal significant regulation of low-abundance transcripts that was not detectable in whole ear samples. Additionally, regulation of proteins, which would not be detected via microarray, likely affects cellular processes during regeneration in the inner ear. Levels of existing p27Kip1 protein may have been altered by ubiquitinylation in order to allow proliferation to occur. Analysis of p27Kip1 protein alteration in the sound-exposed inner ear will be necessary to ascertain whether p27Kip1 protein regulation plays a significant role in naturally occurring hair cell regeneration in the zebrafish. Interestingly, p27 Kip1 was not found to be a part of the zebrafish hair cell transcriptome , although it is a supporting cell marker in the mammalian organ of Corti that inhibits cell cycle progression . Knock-out mice without this gene exhibit cell proliferation in the organ of Corti .
The gene rb1, was also not significantly regulated in this study at the 1.4 fold cut-off level, but since Rb1 function is regulated by phosphorylation, significant changes in overall transcription levels may not be necessary to promote proliferation. Hypophosphorylated Rb1 is an active proliferation repressor, but Rb1 loses all repression function if sufficiently phosphorylated . The phosphorylation state of pRb following noise exposure will need to be delineated to determine whether pRb is an active regulator of cell proliferation in the zebrafish inner ear.
Similarly, regulation of zebrafish atoh1a, homolog of the hair cell differentiation gene Atoh1/Math1, was weak at 2 and 4 dpse in our study. Atoh1a is a key regulator of differentiation of precursor cells that become hair cells in mice [24, 25]. Atoh1a and b are also necessary for hair cell differentiation in zebrafish . The time points investigated in this study may have been too early in the recovery process for Atoh1 detection, as Atoh1 only promotes the final stages of hair cell development [24, 100] and may have peaked in the majority of regenerating hair cells later than 4 dpse.
D. Hair cell genes
Comparison of our microarray dataset with the zebrafish hair cell transcriptome  revealed common hair cell genes. We identified significant regulation in zebrafish hair cell genes encoding proteins such as creatine kinase, alpha-tubulin, keratin 8, and v-fos FBJ murine osteosarcoma viral oncogene homolog. Two zebrafish genes encoding creatine kinase (creatine kinase, muscle (ckm) and creatine kinase, mitochondrial 2 (ckmt2)) were significantly regulated in our microarray dataset. Muscular creatine kinase performs a variety of functions, even in non-muscle tissues and cells . In the inner ear, creatine kinase (or its mitochondrial creatine kinase isoform) is required to maintain energy homeostasis through ATP delivery to plasma-membrane Ca2+-ATPase isoform 2 (Pcma2), an ion pump required for normal sensory transduction in stereocilia of mammals and birds . In the avian utricle, creatine kinase B is primarily localized in hair cells, and creatine kinase/mitochondrial creatine kinase isoform double knockout mice exhibit elevated hearing thresholds of 20-30 dB at 8 and 16 kHz .
Significant regulation of transcripts encoding zebrafish inner ear structural proteins was noted in our study. Alpha-tubulin and beta-tubulin dimers are components of all polymerized microtubules. Strong labeling for alpha tubulin is seen in sensory and supporting cells of the guinea pig inner ear . Keratin 8 is one of the major intermediate filaments, which provide structural support throughout many tissue systems. Keratin 8 is thought to confer resistance to apoptosis induced by Fas ligand or TNF family receptors , both of which are implicated in cisplatin- and ethacrynic acid-induced apoptosis of hair cells in chinchillas .
V-fos genes (the viral homologue of c-fos genes) are highly inducible in response to a variety of growth factors and differentiation-specific inducers, and can induce bone tumors in mice . Members of the fos and jun protein families can combine to form a complex called activating protein-1 (AP-1). AP-1 induction by the hair cell-toxic antibiotic gentamicin is transient and occurs exclusively in hair cells in rat organ of Corti explants . Inhibitors of the upstream pathway for AP-1 rescue hair cells . It should be noted that the up-regulation of some genes found in our microarray data, such as c-fos, are indicative of a general neuronal stress response in fishes , and acoustic stimuli can induce a short-term stress response in goldfish . Thus, it is unclear if such regulation is the response from hair cell damage or auditory nerve overstimulation, but it should not be indicative of neuronal changes in brain activity since our samples only contained ear tissue.
Some of the regulated genes in the current study are similar to genes highly regulated in the hair cells of other model organisms as well. Avian utricular hair cell genes include parvalbumin, which serves as a mobile Ca2+ buffer in the avian inner ear, alpha-tubulin, creatine kinase, heat shock protein 90 (HSP90), and an isoform of Ca2+ transporting ATPase . Additionally, POU domain transcription factors, thyroid hormone receptor , heat shock proteins , and collagen IV alpha chain 4  have been noted in mammalian hair cells. Bcl-2, another regulated gene in our dataset, is believed to play an essential role in prevention of sensory cell death in guinea pigs . Thus, a number of the gene products that were regulated in the zebrafish ear following acoustic trauma have been found in hair cells or have been found to regulate hair cells.