Expression Analysis of Prestin and Selected Transcription Factors in Newborn Rats

Abstract

Transcription factors (TFs) have a central role to play in regulating gene expression. To analyze the co-expression patterns of selected TFs with the motor protein prestin of the outer hair cells, we applied an real-time PCR approach combining several kinds of information: (i) expression changes during postnatal development, (ii) expression changes by exposure of organotypic cultures of the organ of Corti to factors which significantly affect prestin expression [thyroid hormone (T4), retinoic acid (RA), butyric acid (BA), increased KCl concentration] and (iii) changes along the apical-basal gradient. We found that the mRNA levels of the TF Brn-3c (Pou4f3), a member of the POU family, are significantly associated with the regulation of prestin during postnatal development and in cultures supplemented with T4 (0.5 μM), BA (0.5–2.0 mM), and high KCl (50 mM) concentration. The mRNA level of the constitutively active TF C/ebpb (CCAAT/enhancer binding protein beta) correlates positively with the prestin expression during postnatal development and in cultures exposed to T4 and RA (50–100 μM). The mRNA levels of the calcium-dependent TF CaRF correlates significantly with the prestin expression in cultures exposed to T4 and high KCl concentration. The observed coexpression patterns may suggest that the TFs Brn-3c, C/ebpb, and Carf contribute to regulating the expression of prestin under the investigated conditions.

Introduction

The high sensitivity of hearing depends strongly on the proper functioning of special mechanosensory cells, the outer hair cells (OHCs), which amplify sound-induced vibrations in the organ of Corti. This amplification is mainly the function of the protein prestin, which is highly expressed in the OHCs, but not expressed in the inner hair cells (IHCs; Zheng et al. 2000). The hearing threshold rises by 40–50 dB, when OHCs are damaged or lost (Ryan and Dallos 1975). The molecular mechanisms that regulate the expression of the prestin gene are poorly understood. Prestin expression and incorporation in the plasma membrane begin from postnatal day zero and increase progressively in a time course coinciding with that of electromotility (Belyantseva et al. 2000). In a previous study, we had shown that there is a close correlation between the mRNA levels of Gata-3 and prestin both in vitro and in vivo, which points to a role Gata-3 has to play in regulating prestin expression (Gross et al. 2010). GATA-3 is a member of the zinc finger family and is essential to inner ear development (Karis et al. 2001).

Generally, the regulation of gene expression includes multiple transcription factors (TFs; Farnham 2009). Based on cell biological and regulatory function, Brivanlou and Darnell (2002) divided the TFs into (i) regulatory and (ii) constitutively active TFs. Two broad classes of regulatory TFs were distinguished: developmental, cell-type-specific TFs and signal-dependent TFs. In this study, we compared the mRNA expression patterns of Brn-3c (also known as Pou4f3, developmental, cell-specific), C/ebpb (CCAAT/enhancer binding protein beta), Sp1 (Specificity protein 1; both constitutive), Carf (calcium response factor), and Creb (cAMP response element binding protein; both signal dependent) with that of prestin.

BRN-3c is a member of the POU (Pit-Oct-Unc) family of TFs and is essential to inner ear development and hair cell maintenance (Ryan and Rosenfeld 1997). BRN-3c critically modulates the terminal differentiation of the cochlear and vestibular hair cells. In the cochlea, the expression appears by embryonic day 18 and continues into adulthood (Erkman et al. 1996). Targeted deletion of the Brn-3c gene induces deafness in mice. A mutation of this gene is associated with an autosomal-dominant non-syndromic hearing loss (Dallos 2008; Keithley et al. 1999).

Target-specific gene expression is achieved by interactions between cell-type-specific and constitutively active TFs. C/EBPb is a constitutively active TF involved in numerous cellular responses, including cellular growth and differentiation. It belongs to the basic-leucine zipper (b-ZIP) class of DNA binding proteins and may facilitate IL-6 induced transcriptional activation (Ramji and Foka 2002). C/EBPb is involved in the regulation of many different target genes, for example in PC12 cells of 115 genes (Kfoury and Kapatos 2009). In the brain, C/EBPb promotes inflammation and neuronal death (Yi et al. 2007). The expression level is regulated by inflammatory agents and several cytokines. C/EBPb plays an important part in the gene network of inner ear development and it binds GATA proteins directly (Holley et al. 2007).

SP1 is another constitutively active TF belonging to the family of zinc finger DNA binding TFs. It is one of the best characterized transcriptional activators (Lomberk and Urrutia 2005). This protein interacts with many co-factors and can carry out activating or repressive activities. No data on SP1 expression are known for the inner ear. However, some observations make SP1 a candidate likely to be involved in regulating prestin expression. SP1 interacts with the TFs GATA-3 (Hong et al. 2008). Nolan et al. (2007) found that there are common sequence variants in the Brn-3c 5′-flanking region that modifies binding affinity for the SP1 TF and thus alter transcriptional activity.

Since the transcription of genes in neuronal tissue can be regulated by activity, it is of high interest whether signal-dependent TFs are coregulated with prestin expression (Chawla 2002; Flavell and Greenberg 2008; West et al. 2002). Raising outer K⁺ concentration is frequently used to mimic the activity of neurons and to depolarize neuronal cells (Bui et al. 2006). The TF CaRF is a CaRE1 (calcium response element) binding protein, regulated by neuronal activity via changes in calcium signaling (Xia and Storm 2002). The CaRF/CaRE1 transcriptional pathway was found to be involved in the control of BDNF gene expression and to be of critical importance to activity-dependent, long-term changes in the CNS. CaRF cooperates with CREB to mediate the activity-mediated expression of BDNF exon III (Tao et al. 2002). An observation that makes CaRF a candidate for regulator of prestin expression is the fact that Carf and prestin are upregulated by salicylate, a substance known for its effect on hearing and tinnitus formation (Singer et al. 2008; Yu et al. 2008). CREB is a calcium-activated TF which belongs to the basic-leucine zipper family (Kitagawa 2007). An important target gene for calcium-activated TFs is the brain-derived neurotrophic factor BDNF which has an important role to play in neuronal plasticity and survival. CREB and BDNF are important factors in molecular plasticity, a possible mechanism in the development of tinnitus (Kaltenbach et al. 2005).

The developmental, regional, and conditional patterns in which a gene is expressed may contribute to the identification of TFs involved in regulating target genes (Aid-Pavlidis et al. 2009; Schena et al. 1995). On the basis of our own previous investigations and those of others we selected experimental conditions known to increase or decrease prestin expression: (i) prestin expression increases during the early postnatal period; (ii) prestin expression can be modulated in organotypic cultures by changing the culture medium. Addition of thyroid hormone and BA increases prestin expression, whereas the addition of retinoic acid (RA) and an increased KCl concentration cause it to decrease (Gross et al. 2011). These manipulations include pathways known for their effect on the transcriptional activity. For example, thyroid hormone acts via nuclear thyroid hormone receptors (Weber et al. 2002) and RA does so via nuclear retinoic acid receptor (RAR) or retinoid X receptor (RXR) receptors (Bugge et al. 1992; Romand et al. 2006). BA acts via histone acetylation (Garcia-Villalba et al. 1997), and high KCl concentration acts via depolarization and increased cellular or nuclear calcium activity (Boyer et al. 1998). As the organ of Corti is tonotopically organized, it is important to analyze the gene expression in the organ of Corti separately in its apical, middle and basal segments. Real-time PCR was used to quantitate the mRNA levels of prestin and the selected TFs.

Methods

Preparation of the Organ of Corti

All studies were performed in accordance with the German Prevention of Cruelty to Animals Act and permission was obtained from the Berlin State Office of Health and Social Affairs (T0234/00). Apical, middle, and basal segments of the organ of Corti from P2, P5, and P8 old rats were prepared on the basis of the work by Sobkowicz et al. (1993). Stria vascularis and modiolus were separated from the organ of Corti during preparation and were not considered here. The organ of Corti was separated into three parts of about equal length: the basal, middle, and apical segments (Fig. 1). Dissection of the cochlea was performed in buffered saline glucose solution (BSG: 116 mM NaCl, 27.2 mM Na₂HPO₄, 6.1 mM KH₂PO₄, glucose 11.4 mM) at 4°C under a laminar flow hood using a dissecting microscope (Stemi, SV6, Zeiss, Germany). The freshly prepared segments were rapidly transferred into a tube with 100 μl RNAlater (Ambion) to avoid RNA degradation and stored at 4°C for 24 h (control).

Fig. 1.

Images of the cochlear tissue fractions before and after preparation of the organ of Corti segments. a Cochlea with organ of Corti and stria vascularis ready for dissection. b Dissected organ of Corti, stria vascularis, modiolus. c Organ of Corti transected into apical (top), middle (right) and basal segments; tissue was stained by calcein AM. d–f Images of hair cells of the apical (d), middle (e), and basal (f) segments stained by TRITC-phalloidin. Bar (a–c) = 450 μm, bar (d–f) = 20 μm

Explant Cultures

For culturing, segments of P3–P5 old animals were prepared and incubated in four-well tissue culture dishes for 48 h (1.9 cm² culture surface per well, Nunc, Wiesbaden, Germany). The explants were routinely cultured in 500 μl Dulbecco’s Modified Eagle Medium/F12 Nutrient (1:1) Mixtures (DMEM/F12, Gibco, Karlsruhe, Germany) supplemented with 10% fetal bovine serum (FBS, Biochrom AG, Berlin, Germany), 33 mM glucose, insulin-transferrin-Na-selenit-Mix 2 μl/ml (Roche Diagnostics GmbH, Mannheim, Germany), Penicillin 100 U/ml (Biochrom AG, Berlin, Germany) at 37°C, 24 ng/ml IGF-1 (R&D systems, Minneapolis, USA) and 5% CO₂ in a humidified tissue culture incubator (Lowenheim et al. 1999). Four culture groups were analyzed: (i) Cultures grown in medium with the addition of 0.5 μM T4; (ii) Cultures grown in medium with the addition of 0.01–2.0 mM BA which was dissolved in distilled water; (iii) Cultures grown in medium with the addition of 1-100 μM RA (T4, BA and RA from Sigma, Steinheim, Germany). RA was dissolved in DMSO (1 μM RA in 0.01, 5 μM RA in 0.05, 10 μM RA in 0.1, 50 μM RA in 0.5 and 100 μM RA in 1% DMSO); (iv) Cultures grown in medium with increased KCl concentration (50 μM). After 48 h, the cultured segments were washed and transferred into a tube with 100 μl RNAlater (Ambion) to rapidly permeate tissues and to protect cellular RNA. For RNA preparation, 2-3 segments were pooled. In all experiments, one of the ears (from the same animal) was used as a control and was handled identically except for the specific treatment. In the case of RA treatment, the corresponding DMSO concentration was used in the controls. To exclude any hair cell loss from being the cause of prestin changes, we counted the OHC number in freshly prepared samples and in samples cultured for 2-3 days. The OHCs were well preserved in samples cultured for 48 h independent of the treatment (Gross et al. 2011). An example of the appearance of hair cells and their stereocilia is shown in Fig. 1.

RNA Preparation

The segments were washed two times using 100 μl PBS (phosphate buffered saline) followed by centrifugation at 14,000 rpm for 2 min. Total RNA was isolated using the RNeasy Mini Kit (Qiagen, Hilden, Germany) according to the manufacturer’s protocol. The RNA samples were digested by DNase according to the protocol. The RNA content was determined with the Quant-iT™ RiboGreen RNA Assay Kit (Invitrogen/Molecular Probes, Göttingen, Germany). The RNA was stored in 30 μl RNAse-free water at −80°C.

Reverse Transcription

Reverse transcription was performed in three steps in a Perkin Elmer Thermocycler. (i) RNA was linearized using 11.2 μl RNase-free water, 0.8 μl RNasin and 10 μl RNA sample at 70°C for 5 min and cooling to 4°C. (ii) 8 μl RT-buffer (5×), 0.5 μl RNasin, 0.25 μl M-MLV-Reverse Transcriptase, 8 μl dNTP mix (10 μM) and 2 μl oligo-dT-mix (75.8 μM) were added and the reaction mixture was incubated at 42°C for 60 min. (iii) The mix was heated to 95°C and cooled to 4°C. The c-DNA was stored at −20°C up to when the mRNA level was determined in the PCR.

PCR

Primers were designed using the primer designer Primer3 (http://frodo.wi.mit.edu/primer3/input.htm). PCR primers were custom-synthesized by BIOTEZ (Berlin, Germany). The specificity of the primers was checked by a BLAST search and revealed an exact match with the indicated genes. Primers used for RT-PCR (Acc. No., forward, reverse, product length): Tbp, NM_001004198, 5′-acc ccc ttg tat cct tca cc-3′; 5′-cat gat gac tgc agc aaa cc-3′ (201); Prestin, NM_030840, 5′-cac aga gtc cga gct aca cag tc-3′; 5′-tca gtg cgc tgc tgt aca ag-3′ (162); Brn-3c, NM_001108889.1, 5′-gtc tca gcg atg tgg agt ca-3′; 5′-gcg aca ggg taa gag act cg-3′ (186); C/ebp-beta, NM_024125, 5′-gac aag ctg agc gac gag ta-3′; 5′-agc tgc tcc acc ttc ttc tg-3′ (158); Sp1, NM_012655, 5′-tga atg ctg ctc aac tgt cc-3′; 5′-ctc cac ctg ctg tct cat ca-3′(195 bp); Carf, NM_001106915, 5′-ctt gaa gaa cct ttg ctg gc-3′; 5′-tga gac ttc cac atc aag cg-3′, (233); Creb1, NM_031017, 5′-cat gga ctc tgg agc aga ca-3′, 5′-tac agt ggg agc aga tga cg-3′ (154).

Real-time PCR was performed using the LightCycler FastStart DNA Master^Plus SYBR Green I (Roche Diagnostics GmbH, Penzberg, Germany) in the Light Cycler System (Roche Diagnostics, Basel, Switzerland). The glass capillaries were filled with 15 μl of master mix (9.0 μl water, 1 μl 3′-primer, 10 μM, 1 μl 5′-primer, 10 μM, 4 μl LightCycler FastStart DNA Master^Plus SYBR Green I, Roche). Then, 5 μl of cDNA was added as PCR template. After centrifugation, capillaries were placed inside the cycler. A typical setting for targets of 100–280 bp in length included the following steps: (1) preincubation at 95°C for 10 min; (2) amplification (35–40 cycles) including denaturation at 95°C for 10 s, annealing at 66°C for 10 s, extension at 72°C for 15 s; (3) melting curve program at 60°C with a heating rate of 0.1°C per second up to 95°C; (4) cooling down to 40°C.

Three criteria were used for an optimal RT-PCR: (i) Efficiency within the range of E = 1.82–1.95. The amplification efficiency was estimated using serial dilutions of RNA samples. The real-time RT-PCR efficiency was calculated from the slope according to the equation: E = 10^(−1/slope). (ii) A single band resolved during agarose electrophoresis (single product of the desired length) was observed for all genes. (iii) A well-defined melting curve (single product-specific melting temperature) was obtained for all genes.

Quantification of mRNA

To minimize the technical variability and to make the expression patterns reliable two reference criteria were used for normalization of the target mRNA levels: total RNA and the expression of the housekeeping gene Tbp (TATA box binding protein; Hendriks-Balk et al. 2007). Tbp is a TF that binds specifically to a DNA sequence called the TATA box. Thus, the expression level of mRNA is indicated as AU/RNA (arbitrary units/ng total RNA) and AU/Tbp (arbitrary units/Tbp mRNA). The calculation was carried out as follows: (i) The AU/RNA levels were calculated on the basis of the formula: AU/RNA = 1/E ^CP(tn)_t where E _t is the amplification efficiency and CP_(tn) the crossing point of the target molecules at the normalized RNA mass input used in the PCR (Gross et al. 2007). (ii) The AU/Tbp were calculated on the basis of the Pfaffl method (Pfaffl 2001; Pfaffl et al. 2002). Tbp has been used as a housekeeping gene in several studies (Nygaard et al. 2007; Roon-Mom et al. 2005). The calculation incorporates the amplification efficiencies (E) of the target (Ta) and the reference genes (Ref) according to the following formula:

Symbols: Ratio—expression of the target gene normalized to that of the reference gene; ΔCPTa [Cali-Sample]—difference of the CP values (normalized to one ng RNA input) of the target genes, the calibrator (Cali) and experimental samples (Sample); ΔCPRef [Cali-Sample]—difference of the CP values (normalized to one ng RNA input) of the reference gene Tbp, the calibrator and experimental samples.

Analysis of DNA Binding Sites

Various programs were used to analyze the promoter region of prestin for the occurrence of putative binding sites. Gene2Promoter (www.genomatix.de, accessed 2009) was used to locate the promoters of prestin (Cartharius et al. 2005). The MochiView (http://johnsonlab.ucsf.edu/sj/mochiview-start, accessed 2010) program was used to screen the promoter region of prestin for the occurrence of putative binding sites (Homann and Johnson 2010). The search for binding sites included the following motifs: BRN-3c, 5′-AAATGA-3′ (Clough et al. 2004); C/EBP, 5′-TKNNGNAAK-3′ (Osada et al. 1996); Sp1, Sp-A (5′-GGGGCGT-3′), Sp-B (5′-GGGCGG-3′), and Sp-C (5′-CCTCCT-3′) (Ahn et al. 2005); CaRF, 5′-YSANAACGAGGC-3′ (Pfenning et al. 2010; Tao et al. 2002); CREB, 5′-TGAYGTCA-3′ (Smith et al. 2007) [Y = C,T; S = C,G; N = any; K = G,T]. To reduce the number of falsely predicted sites we analyzed the occurrence of these binding sites in randomized data sets of the corresponding promoters. The rearrangement was carried out in 30 different data sets of the promoters generated with the “shuffleseq” program from EMBOSS (http://emboss.bioinformatics.nl/cgi-bin/emboss/shuffleseq, accessed 2010). The program allows the sequences to be reshuffled, while maintaining the single-nucleotide frequency of the nucleotides. Furthermore, we tested the conservation of sites across different species. The alignments were carried out using the ClustalW2 program (http://www.ebi.ac.uk/Tools/msa/clustalw2/, accessed 2010).

Statistics

Mean ± standard errors of the mean (SEM) and the Pearson Product–Moment Correlation Coefficient were calculated from the RT-PCR data. The significance of the differences between the different regions and treatment groups was assessed using the analysis of variance (ANOVA) procedure for the developmental data and the nonparametric Wilcoxon paired test (StatSoft, Statistika 6.0, Tulsa, USA) for the ratio data. A significant ANOVA was usually followed by a post hoc Scheffé’s or LSD test (see legend). Differences of P < 0.05 were considered to be significant. Changes were regarded as significant only, when P < 0.05 was observed in both reference systems. Quality control was carried out by aliquots from an organ of Corti pool from several newborn rats. The day-to-day variation coefficient over all genes and over a period of about 4 months was 12.9 ± 3.9% (n = 8–13).

Results

Reference Criteria

To reduce any potential error in the mRNA expression, we used two different reference criteria for the normalization of the RT-PCR data, (i) total RNA and (ii) the mRNA level of the housekeeping gene Tbp. The suitability of these criteria was validated under each experimental condition by analyzing their responses in control and experimental samples (Hendriks-Balk et al. 2007). Figure 2 shows the changes of the RNA mass per segment and of the Tbp expression level (normalized to the total RNA) in freshly prepared tissue of 2-, 5-, and 8-day-old rats. The total RNA per segment amounts to 200–400 ng and shows a slight increase between postnatal day (PD) 2 and PD8. Because measuring total RNA gives a good indication of the amount of available cellular material we assume that this increase reflects the growth of the organ of Corti and the adjacent tissue. The Tbp expression level is similar in the apical, middle and basal segments on all days tested. Unexpectedly, a 40% decrease of the expression level of Tbp was observed in all segments between PD2 and PD8.

Fig. 2.

Changes of total RNA mass and expression of Tbp mRNA levels during postnatal development in the apical, middle and basal segments of the organ of Corti. Apical (a), middle (m), and basal (b) segments were combined for the purpose of significance calculation. PD postnatal day. *¹Tbp, P < 0.0001 vs. PD2, *²Tbp, P < 0.0001 vs. PD5, *³RNA, P < 0.006 vs. PD2 (Scheffe); n = 6 each

Expression Pattern During Postnatal Development

During postnatal development, prestin expression increases continuously in all segments and forms an apical–basal gradient beyond PD5 (Fig. 3a), as previously reported (Gross et al. 2005). In contrast, Brn-3c mRNA levels do not change statistically between PD2 and PD5, but increase between PD5 and PD8 in parallel to prestin expression (Fig. 3b). For Brn3c at PD8 a lower level of expression was observed in the basal segment. The expression pattern of C/ebpb is very similar to that of Brn-3c (Fig. 4a), whereas the Sp1 levels decrease between PD2 and PD5 in all regions (Fig. 4b). Carf and Creb do not show any developmental or regional changes (data not shown). In general, the expression patterns of all genes analyzed are similar in both reference systems.

Fig. 3.

Expression levels of prestin and Brn-3c on PD2, PD5, and PD8 in the apical, middle and basal segments of the organ of Corti during postnatal development. For abbreviations, see Fig. 2. Significances are valid for both reference systems. A-prestin, B-Brn-3c. *¹ P < 0.007 vs. PD2, *² P < 0.0001 vs. PD5, ^#2 P < 0.002 vs. apical, *³ P < 0.008 vs. PD5, ^#3 P < 0.02 vs. apical

Fig. 4.

Expression of C/ebpb and Sp1 mRNA on PD2, PD5, and PD8 in the apical, middle and basal segments of the organ of Corti during postnatal development. For abbreviations, see Fig. 2. Significances are valid for both reference systems. a C/ebp, b Sp1. *¹ P < 0.02 vs. PD2, (apical and middle segments, Scheffe), ^# P < 0.008 vs. apical (LSD),*² P < 0.0001 vs. PD2 (all segments), n = 6

To compare the expression levels of the different genes in the two reference systems, we plotted the expression levels normalized to total RNA on the abscissa and the levels normalized to the house keeping gene on the ordinate (Fig. 5). Thus, the x-axis allows expression levels of the different genes to be compared, whereas the y-axis shows the relative range of expression of the different genes due to development or regional (apical, middle, basal) changes. Sp1 has the highest expression level (range 1000–10,000 AU); prestin, Brn-3c, and Creb have an intermediate expression level (100–1000 AU) and Carf and C/ebpb display the lowest expression levels. The highest changes have been observed for prestin. As expected, significant correlations between the expression levels related to the RNA mass and the housekeeping gene were found for prestin and for each of the studied TF (see legend Fig. 5).

Fig. 5.

Gene expression diagram. The diagram shows the expression levels, using total RNA (x-axis) and Tbp (y-axis) as reference criteria. Each point represents the mean of the apical, middle and basal segments of the organ of Corti from newborn rats during postnatal period. Semi-logarithmic trend lines fit best to the data. Significant correlations between AU/RNA and AU/Tbp were observed for all genes (r = 0.80–0.98; P < 0.01, n = 9)

Expression Pattern in Organotypic Cultures

In culture, the total RNA mass/segment decreases by about 50% (Fig. 6a). A similar response was observed in cultures supplemented with BA, RA and high KCl concentration (data not shown). The RNA decrease has most probably been the result of tissue degradation during culture preparation. In T4-treated cultures, the RNA mass per segment remained statistically unchanged, possibly due to the growth supporting effects of T4 (data not shown). The Tbp expression level was similar in the starting material, the cultures and the apical, middle, and basal segments under all conditions studied. During 48 h culturing, the prestin expression increases in the apical and middle segments (Fig. 6b; factor 1.3–2.0). Roughly, the increase rate in culture is similar or somewhat lower than in vivo (Gross et al. 2005). In the present study, in the basal segment no increase of prestin levels was observed, possibly due to non-optimal growth conditions or increased mRNA degradation.

Fig. 6.

Changes of total RNA mass, Tbp, and prestin mRNA levels in the apical, middle, and basal segments of the organ of Corti in 48 h culture. a RNA mass and Tbp mRNA levels; apical (a), middle (m), and basal (b) segments were combined for the purpose of significance calculation; RNA, *P < 0.006 vs. PD2 (Scheffe); Tbp—ns. b Prestin mRNA, *P < 0.000 versus combined apical and middle segments, n = 6 each

In culture, prestin expression can be modulated by the addition of different substances to the medium, with T4 and BA addition increasing and RA and a high KCl concentration both decreasing the prestin mRNA levels (Gross et al. 2011; Mazurek et al. 2011). In these experiments, using one ear as control and the second one as an experimental ear enabled us to calculate the fold changes of expression on the basis of a pairwise comparison. Addition of T4 to the culture medium significantly increases the expression of prestin in the apical and middle segments and the Brn-3c expression in the apical segment (Fig. 7). Pronounced changes are observed in the C/ebpb expression in all segments. Carf expression decreases slightly. No changes were observed in the Sp1 and Creb mRNA levels. The addition of BA affected prestin and Brn-3c expression only, with no effect on the other TFs. BA induced an increase in prestin and Brn-3c expression in all regions at BA concentrations above 500 μM (Fig. 8).

Fig. 7.

Effects of T4 on the expression levels of the studied genes in the apical, middle, and basal segments of the organ of Corti after 2 days in culture. The data are the means of the fold changes using both reference systems, n = 6 each; *P < 0.05 vs. controls (Wilcoxon-paired test)

Fig. 8.

Effects of butyric acid on the expression levels of prestin and Brn-3c in the apical, middle and basal segments of the organ of Corti. Data of 500–2000 μM concentrations were combined for the purpose of significance calculation. n = 6–8 each. Prestin, apical a *¹ P < 0.004; middle b *¹ P < 0.006; basal c *¹ P < 0.0008. Brn-3c; *² P < 0.003 for all segments (Wilcoxon-paired test)

In contrast to T4 and BA, addition of RA to the culture medium induced a decrease of prestin expression. Figure 9 illustrates the effect of increasing concentrations of RA on the changes of prestin, Brn-3c, and C/ebpb in the apical, middle, and basal segments of the organ of Corti. At concentrations of 50 and 100 μM, RA significantly decreases prestin and Brn-3c expression in all segments. In contrast, C/ebpb showed a dose-dependent increase in all regions. Sp1, Carf, and Creb expression remained unchanged (data not shown). Under culture conditions of 50 mM KCl concentrations the prestin mRNA levels decreased by about 50% compared to the controls in the apical and basal segments (Fig. 10). Brn-3c and Sp1 expression remained unchanged, whereas a remarkable decrease was found for Carf in parallel to prestin levels. C/ebpb and Creb mRNA levels increased.

Fig. 9.

Effects of retinoic acid on the prestin, Brn-3c, and C/ebpb expression levels in the apical, middle, and basal segments of the organ of Corti. Data from 50 and 100 μM concentrations were combined for the purpose of significance calculation. n = 6–8 each. Apical a prestin, *¹ P < 0.000; Brn-3c, *² P < 0.02; C/ebp, *³ P < 0.002. Middle b prestin, *¹ P < 0.007; Brn-3c, *² P < 0.004; C/ebpb, *³ P < 0.002. Basal c prestin, *¹ P < 0.003; Brn-3c, *² P < 0.005; C/ebpb, *³ P < 0.002 (Wilcoxon-paired test)

Fig. 10.

Effects of high KCl concentration on the expression levels of the studied genes in the apical, middle, and basal segments of the organ of Corti after 48 h in culture. The data are the means of the fold changes using both reference systems, n = 6 each; *P < 0.05 vs. controls (Wilcoxon-paired test)

To further analyze the relations between the TFs and prestin we used correlation analyses. A close correlation was observed between Brn-3c and prestin mRNA levels obtained in PD5-PD8 samples and in samples from T4, BA, and KCl culture groups. Significant correlations were also found between C/ebpb and prestin mRNA levels in samples obtained from the PD5 to PD8 animals and samples from the T4 and RA culture groups. Interestingly, the correlation between C/ebpb and Prestin mRNA levels was negative in the RA culture group. A significant positive correlation between Carf and prestin mRNA levels was observed in the KCl and T4 treated culture group (Table 1).

Table 1.

Correlation between the expression level of TFs and prestin

TF	Condition	r; P (n)
Brn-3c-prestin	PD5-PD8	0.84; 0.0000 (36)
	Culture, T4	0.76; 0.0016 (24)
	Culture, BA	0.73; 0.01 (15)a
	Culture, KCl	0.63; 0.001 (24)
C/ebpb-prestin	PD5-PD8	0.74; 0.0000 (36)
	Culture, T4	0.61; 0.003 (24)
	Culture, RA	−0.76; 0.01 (15)a
Carf-prestin	Culture, T4	0.55; 0.008 (24)
	Culture, KCl	0.65; 0.0006 (24)

TF transcription factor, PD postnatal day, T4 cultures exposed to thyroid hormone, BA butyric acid, RA retinoic acid, r correlation coefficient, P statistical significance, n number. Significant correlations were observed to occur also within each segment group (all of the PD5–PD8 groups, n = 12; nearly all culture groups, n = 8)

^aCorrelation of the means of different concentration groups

Discussion

Reference Criteria

Any interpretation of the RT-PCR data of prestin has to consider the effects exerted by both the number of damaged OHCs and the reference criteria used to normalize the gene expression data. Hair cell loss due to the experimental conditions could have a critical impact, because prestin is expressed in the OHCs only. No significant changes of OHC numbers were observed under the conditions studied in this work (Gross et al. 2011). Accordingly, changes in the prestin levels observed in this study have been caused by changes in gene expression or mRNA degradation. To reduce any potential error in the expression of prestin and the studied TFs by the housekeeping gene, we used two reference criteria, total RNA mass and a housekeeping gene. Our previous study had shown that the frequently used Gapdh gene is not suited as a reference gene for the developing organ of Corti, because its activity increases by a factor of two (Gross et al. 2010). Instead, we used Tbp. The Tbp expression level is similar in the apical, middle and basal segments in all conditions tested. However, in the present study Tbp unexpectedly decreased by 40% in samples prepared on PD8 compared to those of PD2, but did not change in culture. As the developmental decrease of Tbp expression is relatively low compared to the conditional or regional changes of prestin, no significant effect on the expression patterns was observed. It should be noted that steady-state levels of mRNA are regulated by degradation as well as by gene expression. Data of mRNA degradation are still largely unknown, for example for prestin. The C/ebpb mRNA stability is known to be regulated by ARE binding proteins, which interact with motifs in the 3′ UTR (Yim et al. 2006). Calissano et al. (2007) showed that the 3′ UTR of the Brn-3b mRNA does contain regulatory sequences that mediate mRNA degradation in neuroblastoma cells.

Brn-3c and Prestin

Parallel changes of Brn-3c and prestin are observed during the postnatal period and in most of the in vitro manipulations. Our findings add support to the assumption of Xiang et al. (1997) who suggested that BRN-3c is required for correct postnatal development of hair cells. We assume that BRN-3c contributes together with other TFs, e.g., Gata-3 to the postnatal regulation of prestin expression. Interestingly, there are remarkable differences between the expression of Gata-3 and that of Brn-3c: (i) Brn-3c is localized in hair cells only, but it is not expressed in supporting cells (Xiang et al. 1997), whereas Gata-3 was found in both hair cells and adjacent tissue (Van der Wees et al. 2004). (ii) During postnatal development, Brn-3c expression increases beyond PD5, whereas the level of Gata-3 decreases or remains stable (Gross et al. 2010). These different patterns of expression may indicate cooperative interactions between both TFs. For example, the function of Brn-3c may be limited to hair cells, whereas Gata-3 performs regulatory functions in supporting cells as well.

C/ebpb, Sp1, and Prestin

The response of the constitutively active TFs C/ebpb and Sp1 is completely different. Whereas C/ebp expression increases during development, that of Sp1 decreases. In all cultures, except those of BA-treated ones, a dramatic increase of C/ebp expression levels was observed, whereas Sp1 expression remained largely unchanged. Parallel changes and significant correlations of C/ebpb and prestin mRNA levels were observed in samples obtained from PD5 and PD8 and in thyroid hormone treated cultures. A significant negative correlation was observed in RA treated cultures. Interestingly, C/ebpb shows a clear apical–basal gradient on PD8, similar to the prestin gradient. Interactions between C/EBPb, thyroid hormones, retinoids, and increased KCl concentrations were observed in various experimental models. For example, C/EBPb, together with thyroid hormones and retinoids, regulates the expression of stromelysin-3, a protein involved in tissue remodeling (Ludwig et al. 2000). Further, RA was found to inhibit C/EBPb protein expression and thus contribute to reducing ischemia-induced cerebral injury (Choi et al. 2009). C/EBP pathways are also involved in the depolarization or excitotoxicity-mediated damages of brain tissue (Cortes-Canteli et al. 2008). Application of 25 mM KCl to 14-day rat cortical culture neurons resulted in inducing 19 genes (among them C/EBPb) and reducing 42 genes (Sun et al. 2007). Gata-3 forms protein complexes with C/EBPb and regulates adipocyte differentiation (Tong et al. 2005). These observations together with our data led us to the conclusion that C/EBP is a constitutively active TF involved in regulation of prestin expression. In contrast to C/ebpb, Sp1 did not show any parallel changes or correlation to prestin expression. This Sp1 expression pattern is in line with its function of a constitutive activator of housekeeping genes which are usually not regulated (Wierstra 2008).

Carf, Creb, and Prestin

Carf expression correlates with prestin expression and decreases significantly in T4 and KCl-treated cultures. In all other experimental conditions, Carf expression remained unchanged, in terms of both development and the apical–basal gradient. The principles of the regulation of gene expression by calcium were described in different experimental models. Briefly, the increase in extracellular K⁺ concentration to 30–50 mM depolarises the cell via voltage-sensitive calcium channels being associated with the influx of Ca⁺⁺ across the plasma membrane (Fields et al. 2005; Kingsbury and Krueger 2007; Shieh et al. 1998). Segal (1990) identified Ca⁺⁺ as the first messenger for the action of the thyroid hormone and showed Ca⁺⁺ uptake to occur as an early event following the binding of the hormone to its membrane receptor. The increased intracellular Ca⁺⁺ concentration may trigger a signal to the nucleus to activate TFs (Xia and Storm 2002). Our findings are of high interest considering the fact that Carf and prestin were found to undergo parallel expressional changes in the salicylate model. Long-term administration of salicylate may increase prestin expression and OHC electromotility (Yu et al. 2008). CaRF1 is upregulated in these conditions (Singer et al. 2008). The authors assume that the dose-dependent effects of salicylate on CaRF1 may result from indirect dose-dependent effects of salicylate on L-type calcium channels in cochlear neurons. These data together with the present findings suggest that CaRF may be involved in regulating prestin under the influence of thyroid hormone and depolarizing conditions.

In contrast to Carf, Creb mRNA levels increase in KCl-treated cultures without statistically significant correlations with prestin expression. It was postulated that the CaRF activity requires cell-type specific regulatory events. Xia and Storm (2002) found that CaRF contributes to calcium stimulation of the BDNF gene and that the CaRF/CaRE1 transcriptional pathway may control the expression of genes critical to activity-dependent, long-term changes in the central nervous system (CNS). Specific pathways seem to be necessary to activate the Carf expression. For example, Tao et al. (2002) did not observe induction of CaRF activation following ionomycin-mediated elevation of cellular calcium. The calcium-dependent activation of CaRF is also independent of PKA, whereas the calcium-dependent activation of CREB is dependent on PKA (Tao et al. 2002; Xia and Storm 2002). It remains to be determined whether the activation of calcium metabolism, for example, by intensive noise will affect the Carf expression and thus influence the hearing process. The fact that Creb and Carf responded in diametrically opposed ways underlines the specificity in activating calcium-dependent TFs.

In Silico Analysis

To test whether the prestin promoter contains binding sites for the TFs analyzed here, we searched the promoter region for the occurrence of putative binding sites using the MochiView program. We identified core sequences for the BRN-3c, C/EBP, and CaRF motifs but not for SP1 and CREB. Table 2 illustrates the sequences and the number of occurrences of these motifs in the rat, mouse, cattle, and human PS3 promoter region. The randomly rearranged sequence sets were used as controls. In general, parts of the recognized binding sites occurred several times. In the prestin-PS3, a significant enrichment of these motifs was observed compared to the rearranged PS3 of the analyzed species. This information increases the significance of the experimental data concerning a regulatory role played by BRN-3c, C/EBP, and CARF. It is interesting to note that the binding site for BRN-3c corresponds to that of BDNF and NT-3, genes expressed both in hair cells and supporting cells (Clough et al. 2004). The binding sites for C/EBP show a high match to the binding sequences of acute phase proteins (Osada et al. 1996). The characteristic 5′-GAGGC-3′ sequence of the CaRF consensus sequence also occurred several times in the analyzed promoter regions. However, the GAGGC sequence is merely part of the recognized motif, and the 5′ elements most commonly observed in the full motif are not present in the prestin promoter. As DNA recognition by TFs is marked by high diversity and complexity, it remains open whether this 5-mer may be too small and not specific enough to enable the binding site for CaRF to be reliably predicted (Badis et al. 2009; Pfenning et al. 2010). Sp1 and Creb mRNA levels did not correlate with the prestin expression. It is well known that TFs act on the protein level and may be regulated at the posttranscriptional and posttranslational levels (Farnham 2009). The presence or absence of binding sites in target genes is indicative of a possible role in regulation, but no proof of it. For example, it may well be that regulation occurs primarily on the level of phosphorylation. Whether the binding sites identified by in silico analysis are functionally relevant need direct experimental evidence.

Table 2.

Putative binding sites for BRN-3c, C/EBP, and CaRF in the promoter region 3 (PS3) of prestin

Motif	Sequence	Occurrence	Species
BRN-3c	5′-AAATGA-3′	BDNF and NT-3a
	5′-GAAAATGAAA-3′ (+)	3 (0, 0–2)	Rat, 89–98d
	5′-GCAAATGAAA-3′ (+)	5 (2, 1–6)	Mouse
	5′-GCAAATGAGA-3′ (+)	2 (1, 0–5)	Cattle
	5′-AAAAATGACC-3′ (−)	1 (0, 0–2)	Human
C/EBP	5′-TKNNGNAAK-3′	Consensus sequenceb
	5′-TGGAGGAAG-3′ (−)	2 (1, 0–4)	Rat, 56–48d
	5′-TGGAGGAAG-3′ (−)	8 (3, 0–10)	Mouse
	5′-TGGAGGGGC-3′ (+)	7 (2, 0–6)	Cattle
	5′-TGGAGAGGA-3′ (+)	4 (1, 0–3)	Human
CaRF	5′-ANAACGAGGC-3′ (+)	CaRE1 of BDNFc
	5′-CAGGAGAGGC-3′ (+)	3 (1, 0–3)	Rat, 450–441d
	5′-CTTGGGAGGC-3′ (−)	7 (2, 0–6)	Mouse
	5′-CCTGTGAGGC-3′ (+)	3 (2, 0–6)	Cattle
	5′-GGGAAGAGGC-3′ (+)	5 (2, 0–5)	Human

^aClough et al. (2004); ^b Osada et al. (1996); ^c Pfenning et al. (2010); ^d Location of binding site in the promoter region according to the database of Genomatix (GXP_19572). The alignment shows sequence conservation between rat, mouse, cattle and human in PS3 prestin promoter. Invariant core sequences that are highly conserved in all three species are underlined. In brackets, occurrence of motifs in rearranged sequences of the promoters (median, min–max), n = 30 each). The Wald–Wolfowitz test resulted in significant difference between observed and random probabilities (P < 0.001 each). The homologue sequence of PS3 of rats can be identified under NCBI/HomoloGene/SLC26a5, accession number NW_047687 (position 2914958–2915609; 652 bp). BLAST search identifies the homologues sequences of PS3 under the accession number NT_165760.2 (Mus musculus), NW_001494886.1 (Bos taurus) and NT_007933.15 (Homo sapiens)

Conclusions

The comparison of the expression patterns of prestin gene and the TFs studied under different experimental conditions contribute to our understanding of the regulation of this gene. The observed coexpression patterns and the multiple correlative changes in gene expression may suggest that the TFs BRN-3c, C/EBPb, and CaRF contribute to regulating the expression of prestin under the investigated conditions. The differential response of these TFs under differing experimental conditions may be an indication of their hierarchical role in the regulation of prestin expression. Thus, Brn-3c and C/ebpb may have a dominating role to play in increasing prestin expression in postnatal development and the formation of the apical–basal gradient, whereas Carf may play a key part under depolarizing conditions. We are aware of several other TFs regulating prestin expression. For example, Yu et al. (2008) observed that NF-kB signaling pathway up-regulates c-fos and egr-1, which could also account for regulation of prestin expression. The present data may have implications for further studies concerning the understanding of the role of prestin for hearing and tinnitus.