3. Gene Profiling in the Vertebrate Inner Ear

3.1. Developmental Shifts in Gene Expression
Since it has been hypothesized that regeneration recapitulates development, understanding temporal shifts in gene expression patterns during the normal development of the auditory system may provide clues to important cellular pathways used for hair cell regeneration. One of the first two studies to use microarray technology to examine gene expression in the inner ear was by Chen and Corey [38,39]. They examined mouse cochleae at two developmental stages (postnatal days 2 and 32) to find differential gene regulation between developing and mature (quiescent) auditory tissues. Since then, gene expression at a number of other developmental stages has been studied in the mouse ear (e.g., E9–E15 [41], P3 and adult [42]). Similarly, Rivolta et al. [40] quantified the time course of gene expression following induced differentiation in conditionally-immortal cells derived from mouse cochleae. These microarray studies confirmed previous evidence that key signaling pathways, such as Notch and Wnt, are important to inner ear development. 
For example, Notch1 and Notch3, as well as downstream effectors of the notch cascade, such as Hes1 and Hes3, were significantly regulated during differentiation of cochlear cells [40]. Nrarp (Notch-regulated ankyrin repeat protein), which is thought to be part of a negative feedback pathway to attenuate Notch signaling, was upregulated in P3 mouse cochlea relative to adult tissue [42]. A number of Wnt genes were expressed in P2 and/or P32 mouse cochleae, including Wnt-4, Wnt-5a, Wnt-5b, Wnt-7b and Wnt10a [38], but the roles of different Wnt genes vary during development. Four Wnt genes were upregulated only in the early developmental stages of the mouse ear, while eleven were upregulated only in the later stages of inner ear development [39].
Microarray studies have also confirmed the importance of cell cycle regulation genes, such as those for cyclin-dependent kinase inhibitors for inner ear development. For instance, p27Kip1, p27Kip2, p19Ink4d and p15Ink4b [38,40,41] were regulated during development of the mouse inner ear. In general, these genes were downregulated in early developmental stages during significant cell proliferation and upregulated in later stages during cell differentiation [40,41]. 
The power of microarray analysis goes beyond verifying genes that are already known to be expressed in the inner ear, to establishing networks of genes and finding novel genes and pathways. Some examples of such novel genes and pathways discovered via gene expression analysis to be regulated in the inner ear during development include semaphorins [40], Hmga2 (high mobility group AT-hook 2), Nrarp, Prl (prolactin) and Ar (androgen receptor) [42] and circadian rhythm and estrogen receptor signaling pathways [41].

3.2. Cell- and Tissue-Specific Transcript Profiling
One of the first steps to understanding pathways involved in hair cell regeneration is to separate genes that are expressed in auditory sensory tissues compared to other reference tissues. One of the first array studies to be applied to inner ear tissue compared regions of the rat cochlea to the cochlear nucleus, inferior colliculus and hippocampus [58]. Greater differences in gene expression were found between the cochlea and the central nervous system regions than between the central auditory regions and the hippocampus, showing that gene expression patterns in peripheral and central nervous tissues differ. Genes that were expressed at higher levels in the cochlea included insulin-like growth factor binding proteins, matrix metalloproteinases and tissue inhibitors of metalloproteinases [58].
Within the inner ear itself, gene expression can vary among different end organs (i.e., cochlea, utricle, saccule and cristae in mammals). Sajan et al. [41] found unique gene expression signatures for the mouse cochlea, utricle and saccule. This is relevant to hair cell regeneration research, because there is evidence for limited hair cell regeneration in the mammalian utricle [47,48,49], but not in the cochlea. Thus, differences in gene expression between the separate auditory end organs are not surprising. In contrast to the mammalian utricle, the avian utricle is in a constant process of apoptosis and regeneration [59,60,61,62]. The avian cochlea is similar to the mammalian cochlea, though, in that it is normally in a quiescent state [63]. Thus, comparing gene expression between different end organs may highlight potential therapeutic targets that may help guide mammalian cochlear sensory epithelia into a proliferative state, allowing for potential hair cell regeneration. Hawkins et al. [27] found 20 different inner ear genes and 80 transcription factors (TF) that were significantly different between the avian cochlea and utricle. Bmp4, Gata3, Gsn, Foxf1 and Prdm7 were some of the genes that were upregulated in the cochlea, while Smad2, Kit, β-amyloid, Loc51637, Hmg20b and Crip2 are examples of genes that were upregulated in the utricle. While some of these genes are well known to be involved in the development of the inner ear (e.g., Gata3 [64]), some of them were novel TF, like Loc51637 and Hmg20b, about which little was previously known.
At an even finer scale, the transcriptome of hair cells can be examined. Cristobal et al. [28] used laser capture microdissection to collect hair cells and supporting cells separately from the rat cristae to compare expression profiles between the two cell types. There were 97 and 78 annotated genes with greater than a five-fold expression difference in hair cells relative to supporting cells and supporting cells relative to hair cells, respectively [28]. Another means of separating hair cells from supporting cells for a pure hair cell transcriptome is to dissociate them from the sensory epithelia using proteases. McDermott et al. [43] isolated a population of pure hair cells from the zebrafish lagena and compared the hair cell transcriptome to that of control liver tissue. They found 1,037 hair cell-specific genes supporting a range of functions, including synaptic transmission, transcriptional control, membrane transport, cellular adhesion, cytoskeletal organization and signal transduction, as well as candidate deafness genes, such as KIDINS220.

3.3. Gene Expression Following Inner Ear Trauma
A number of microarray studies have examined gene expression following trauma to the non-regenerative mammalian cochlea. These and other inner ear microarray studies are more thoroughly reviewed by Hertzano and Elkon [29]. Gene expression in mammalian (non-regenerative) models can be compared to shifts in gene expression patterns following trauma to the non-mammalian (regenerative) inner ear to highlight functional pathways involved in hair cell death and regeneration. Although the process of regeneration of adult inner ear tissue may recapitulate some of the same processes of initial sensory epithelial development, it is highly likely that there are important differences, as well. Thus, measuring gene expression in tissues that are going through the regeneration process is the most direct way to discover what pathways are activated during hair cell regeneration. 
Trauma to the ear can be produced by ototoxic chemicals [24], acoustic overstimulation [25,44,65] or laser ablation [24]. Hawkins et al. [24] performed the first large-scale microarray experiment on regenerating auditory tissues. They examined gene expression of TF in cultured avian utricles and cochleae following trauma induced by either a pulsed laser microbeam or the ototoxic antibiotic, neomycin. Although there were differences in expression patterns between tissue types and treatments, there were a number of identical gene expression patterns found across treatments during the process of regeneration. Some of the identified signaling pathways were TGFβ, PAX, NOTCH, WNT, NFKappaB, INSULIN/IGF1 and AP1. In addition, p27KIP and genes that regulate its expression and other apoptotic and cell cycle control pathways, were significantly regulated during regenerative proliferation.
Following noise exposure, Schuck et al. [25] examined microarray gene expression patterns in zebrafish ears at two and four days post-exposure. Transcripts that showed the greatest regulation on day 2 compared to control tissues included growth hormone 1 (gh1, upregulated) and major histocompatibility complex, class I, ZE (mhc1ze, downregulated). Many genes that were upregulated on day 2 were downregulated at day 4 and vice versa. Follow-up experiments showed that growth hormone (GH) injection following acoustic exposure led to an increase in cell proliferation and a decrease in apoptosis in the zebrafish inner ear [65]. Insulin-like growth factor 1 (IGF-1) is secreted mainly by the liver and is stimulated by GH. It is required for normal post-natal survival, maturation and differentiation of cochlear ganglion and hair cells [66]. Thus, it is likely that the effects of GH on the zebrafish inner ear are also mediated by IGF-1. A number of genes involved with immune function, including MHC class I and II molecules, were also significantly regulated in the zebrafish ear post-acoustic exposure. These genes may play a role in cell proliferation following hair cell death. When MHC Class I molecules are bound by antibodies, it prevents them from presenting antigens and promotes cell proliferation [67]. Thus, the downregulation of mhc1ze may have promoted increased cell division, as it coincides with a peak in cell proliferation in the zebrafish ear following acoustic trauma [9].
Liang et al. [44] also examined gene expression in the zebrafish ear after noise exposure, but at more time points (immediately after two days of exposure and one, two and four days post-exposure). They used digital gene expression (DGE), which utilizes tag sequence profiling. Immediately after noise exposure, stat3 and socs3a were significantly upregulated and the stat3/socs3a pathway was the dominant signaling pathway that was regulated. Genes related to this pathway were also significantly regulated (e.g., socs3b, jak1, mmp9). Interestingly, the same or similar genes were found regulated in the zebrafish ear by Schuck et al. [25]. These included socs3b, socs1, mmp13 and stat1b. The stat/socs pathways are activated by GH [68], so it makes sense that these genes are regulated as GH is upregulated following acoustic trauma to the ear.