Kcnq 1-5 (Kv7.1-5) potassium channel expression in the adult zebrafish

Abstract

Background

KCNQx genes encode slowly activating-inactivating K⁺ channels, are linked to physiological signal transduction pathways, and mutations in them underlie diseases such as long QT syndrome (KCNQ 1), epilepsy in adults (KCNQ 2/3), benign familial neonatal convulsions in children (KCNQ 3), and hearing loss or tinnitus in humans (KCNQ 4, but not KCNQ 5). Identification of kcnqx potassium channel transcripts in zebrafish (Danio rerio) remains to be fully characterized although some genes have been mapped to the genome. Using zebrafish genome resources as the source of putative kcnq sequences, we investigated the expression of kcnq1-5 in heart, brain and ear tissues.

Results

Overall expression of the kcnq x channel transcripts is similar to that found in mammals. We found that kcnq1 expression was highest in the heart, and also present in the ear and brain. kcnq2 was lowest in the heart, while kcnq3 was highly expressed in the brain, heart and ear. kcnq5 expression was highest in the ear. We analyzed zebrafish genomic clones containing putative kcnq4 sequences to identify transcripts and protein for this highly conserved member of the Kcnq channel family. The zebrafish appears to have two kcnq4 genes that produce distinct mRNA species in brain, ear, and heart tissues.

Conclusions

We conclude that the zebrafish is an attractive model for the study of the KCNQ (Kv7) superfamily of genes, and are important to processes involved in neuronal excitability, cardiac anomalies, epileptic seizures, and hearing loss or tinnitus.

Background

Potassium channels are well-established biological targets for diseases including neuropathic pain, epilepsy, cardiac arrhythmia, hearing loss, deafness, or tinnitus[1]. In particular, mutations in the KCNQ4 potassium gene and perhaps KCNQ3 are associated with progressive high frequency hearing loss[2, 3]. Of the several ion channels used by the sensory hair cell, the K⁺ channel KCNQ4 is thought to modulate the membrane potential of hair cells to adjust the sensitivity of hearing in a variety of mammals[1, 4, 5]. Similarly, KCNQ4 and KCNQ5 are key modulators of L-type Ca²⁺ channel activity in cardiovascular cells[6]. Variants of KCNQ5 are not associated with sensory hearing loss in humans, but there is high abundance in the larval zebrafish ear[7, 8], and thus, may be related to yet to be defined developmental factors related to hearing[9].

A recent study characterized the expression of kcnq2, kcnq3, and kcnq5 in whole larval zebrafish (Danio rerio)[7], but we know little about the expression of the complement of kcnq genes and the K⁺ ion channels that they encode in various organs of the adult zebrafish. Since certain drugs and metal ions affect the function of Kcnq channels[10–12] in a dose-dependent manner, these agents can be used to alter ion permeability across the membrane of zebrafish hair cells and thus create a fish model of sensory cell dysfunction. KCNQ2-5 channels are also regulated by intracellular signal transduction effectors such as phospholipids[13], phosphorylation[14], and calmodulin[15]. However, little is known about how these signaling systems impact the kcnq channels in zebrafish sensory pathways. Thus, the zebrafish offers a unique opportunity to study Kcnq channel modulation, function and dysfunction.

The zebrafish has served as an especially attractive model for the study of the development and function of the vertebrate inner ear[8, 16]. It has three methods of sensing sound within its environment. The first involves the lateral line system, which is comprised of a set of neuromasts containing hair cells arrayed along each side of the body. Neuromasts contain bundles of sensory hair cells beneath a cupula, which are responsible for sensing the displacement of water molecules[17]. The second means of sensing sound are structures of the inner ear composed of the utricle, saccule, lagena and pars neglecta. Each of these anatomical structures house patches of sensory hair cells and supporting cells that are embedded in the epithelial lining of the macula[18]. The hair cells found in these structures are similar to those found in mammals, and contain voltage gated and ligand gated ion channels presumably linked to several signal transduction pathways. Third, there are sets of motion detectors or neuromasts arrayed around the head, particularly the orbital regions. In this report, we have studied Kcnq channel expression and localization in several tissues of the zebrafish. Using the deduced mRNA sequences in the available databases, we probed for the presence of Kcnq channel mRNA transcripts in the ear, brain and heart, and partially characterized the amino acid sequence of one channel protein. The zebrafish genome has two different kcnq4 genes, one of which has been localized to chromosome 19. The mRNA from this gene is also expressed in zebrafish brain and ear. We prepared a specific antibody to zebrafish Kcnq4, quantified its levels using qRT-PCR, and further verified its expression using Western blots of brain and ear tissues.

Results

Detection of Kcnq Expression in Zebrafish

Amplicons representing kcnq1-5 were detected by RT-PCR analyses and observed under UV illumination. Table 1 shows the primers, amplicon size, and primer sequences used for all PCR reactions. Primers and nested primers were designed to cross several exons of the specific PCR template sequence.

Table 1 Summary of primers designed to amplify kcnq 1-5 and β-actin mRNAs

Several of the kcnq RNA transcripts were expressed in the zebrafish brain (Figure 1A) consistent with observations in other species[19]. Kcnq1 and kcnq 5 were absent in the gel (Figure 1A), but were detected in the quantitative data (Figure 2, top), at the same approximate levels. The PCR data for kcnq5 were done with kcnq5b primers while the quantitative PCR was done with kcnq5a primers. Kcnq 4 was probed by three different sets of primers (KCNQ4-a, KCNQ4-b, KCNQ4-c) for downstream cDNA sequencing. Bands for kcnq4 for the three sets of primers were readily detected. Kcnq2 and kcnq3 showed the strongest signal, while kcnq 3 displayed two almost overlapping bands. A negative control (PCR reaction without RNA, or “No RNA”) is shown in the last lane. The expression strength of the mRNA transcript was compared to the intensity of the β-actin control.

Figure 1

Expression of kcnq in zebrafish brain (A), ear (B) and heart (C). Lanes labeled M are the 100 bp ladder molecular standards. Lane 9 and 10, respectively, is a positive control with β-actin, and a negative control without RNA.

Figure 2

qRTPCR of Kcnq channel transcripts in brain, heart, and ear. Real time PCR thresholds (Delta CT) were first normalized to GAPDH or β-actin for each tissue. These data were then plotted relative to the lowest expressed mRNA in each tissue using a logarithmic scale. Depicted also is +/-1.0 standard error.

The inner ear tissue of zebrafish included the sensory epithelium (culled from 6 fish, both ears), consisting also portions of the utricle, saccule and lagena tissues, but not semicircular canals. Figure 1B shows mRNA expression of kcnq1-5 in the zebrafish ear - all kcnq transcripts were detected. However, kcnq1 was somewhat weak, while kcnq5b provided a much stronger signal. Figure 1C shows the expression pattern in zebrafish heart. Except for kcnq5b, transcripts for kcnq1-4 were detected.

As mentioned in the Introduction, the partial sequence of a kcnq4 gene has also been mapped to chromosome 19. We detected expression of transcripts based upon this gene in brain and ear (Figure 3A) and the heart (Figure 3B). The sequence is located at the 5′ end (519 bp) of the transcript encoding a 173 amino acid sequence, homologous to KCNQ4 from human as well as other species (See Figure 4). Using various combinations of primers in the RT-PCR experiments, all attempts to link sequences of the chromosome 19 transcripts to the more 3′ kcnq4 sequences in our mRNA, which is not yet mapped to a chromosome, failed to show up in our data (not shown). Therefore, we conclude from these observations that there are two separate kcnq4 genes expressed in the zebrafish.

Figure 3

Expression of kcnq4 transcripts from chromosome 19. A. PCR products from reverse transcriptase PCR reactions using primers (Table 1) for kcnq 4. Lanes 2 and 5 are brain mRNA products, while lanes 3 and 4 are zebrafish ear mRNA-derived products. Lane 1 is a 100 bp ladder standard. B. PCR products from reverse-transcriptase PCR reactions using primers (Table 1) for Chr 19 (Lanes 2 and 3) and the Zv9_NA546 scaffold kcnq 4 (lanes 4 and 5) of mRNA from zebrafish heart. Lane 1 is a 500 bp ladder standard.

Figure 4

A. Alignment of the partial zebrafish kcqn4 sequence (A), with KCNQ4 sequences from chimpanzee (B), frog (C) and human (D). The sequences on top of the KCNQ4 groups are the translation of the sequences that we obtained from the sequencing of PCR products (Table 2). The carboxyl region that contains the sequences implicated in the assembly of homo- vs. hetero-tetramers is indicated with bars, with the Head-Linker-Tail designations. B. Alignment of the translation of the Chr 19 sequence of zebrafish Kcnq4 (A) with the amino termini of Frog (B), Chimpanzee (C) and Human (D).

qRTPCR of kcnq1-5 expression in brain, ear, or heart

End point PCR is not applicable to quantitative measures of expression, and detection of bands can be variable using electrophoretic separation. Therefore, we performed qRTPCR using reverse-transcribed mRNA templates from each tissue. Different primers were designed to produce amplicons (100-200 bp) suitable for SYBR green-based real time quantitative analysis (Table 2). As shown in Figure 2, the brain has very high expression of kcnq2 and kcnq3 compared to other tissues. Kcnq2-5 transcripts are lower than kcnq1 in the heart, while kcnq5 a is particularly elevated in the ear, when compared to the brain.

Table 2 Partial sequences of KCNQ4 obtained from PCR products

Kcnq4 protein expression using Western Blots

To further verify the presence of Kcnq4 protein, we probed Western blots of ear (Figure 5, lane A) and brain (Figure 5, lane B) tissue with rabbit polyclonal antipeptide antisera. A major band at ~80 kDa corresponding to the size of Kcnq4 in mammalian species was readily detected in ear, and brain tissue extracts. The band was absent (Figure 5, lane C) when the immunizing peptide was present along with the primary antibody. Pre-immune sera from the rabbit showed no reactivity to proteins in zebrafish brain or ear tissue extracts (not shown). Using appropriately designed primers, we also performed DNA sequencing on the kcnq4 PCR products that are summarized in Table 2. The sequenced data maps to various zebrafish genomic clones, and the translation of the mRNA provide amino acid sequences consistent with KCNQ4 (Figure 4), indicating that the clones represent active transcripts of a Kcnq4 protein.

Figure 5

Kcnq4 protein expression in zebrafish brain and ear tissues. Western blots of Brain (Lane A) and Ear (Lane B) extracts probed with rabbit antibody to zebrafish Kcnq4. Lane C is zebrafish ear tissue probed with antisera containing 100 ng/mL of the synthetic peptide used for immunogen production added.

Genetic and comparative analysis of Kcnq4 Proteins (channels)

We checked the putative kcnq4 cDNA sequence against the zebrafish genome assembly and clones by BLAST searching. We retrieved the sequences of two genomic clones that contained kcnq4 sequences (Figure 6A). Using two exon prediction programs, Net2Gene[20] and HMM[21], we generated a mRNA sequence comparable to the GenBank sequence, except that the 300 bp of 5′ end and the 200 bp of 3′ end sequence were not found. In the genomic clones, NW_001881069 contains three exons, while the scaffold Zv9_NA546:9, 520-23,017 contains 4 exons. We originally analyzed an earlier sequence deposited in GenBank (NW_00188744) that contains the same exons as the scaffold Zv9_NA546: 9,520-23,017. Two cryptic exons in NW_00188074 detected by the exon predictions programs are not used in the zebrafish Kcnq4. Moreover, ORFs within these two cryptic exons do not contain amino acid sequences related to the Kcnq family of channels. For comparison, the exon structure of the partial kcnq 4 gene from chromosome 19 is shown in Figure 6B. Our PCR data confirmed that the transcript contains a sequence from the three exons in this partial gene.

Figure 6

Analysis of kcnq4 genomic structures. A. Schematic representation of zebrafish genomic clone containing kcnq4-related sequences - these have not at the time of the publication of this work been mapped to a chromosome. B. Partial structure of the Chr 19 Kcnq4 genomic clone. Only the first three exons have been identified. Blue boxes indicate active exons, while white boxes indicate silent exons. A yellow box shows the 5′ untranslated region of Kcnq4 from chromosome 19. Arrows indicate the location of primers used for mRNA detection across the largest number of exons.

The derived coding sequence from the 7 exons of NW_001881069 + Zv9_NA546: 9, 520-23,017 matches very well to other Kcnq4 amino acid sequences from chimpanzee, frog, and humans (Figure 4A). Similarly, the translation of the 5′ mRNA from chromosome 19 KCNQ4 is also very well conserved in the same species (Figure 4B). Table 3 summarizes all of the known Kcnq genes in zebrafish as of January 2013. Ensembl IDs are used for the gene, mRNA, and translated protein sequences, as these are referenced directly in the ZFIN database. Cross-references to entries in GenBank and Uniprot are also tabulated. Kcnq1 on chromosome 7 has three potential splice products of differing lengths that have not been fully characterized. Kcnq1 also has another gene identified on chromosome 25, but no transcripts have yet to be reported. Kcnq2 has two genes, one on chromosome 6 that we also detected in this work, and another on chromosome 8 that was detected in zebrafish larvae[7]. Kcnq5 has also two genes, and in this paper, we focused mostly on transcripts derived from chromosome 13 (kcnq5a). The chromosome 1 derived kcnq5b has also been detected in zebrafish larvae[7].

Table 3 Summary of kcnq genes in zebrafish (2013)

Discussion

In this study, we characterized Kcnq-type proteins/channel expression in brain, heart, and ear tissues of the zebrafish. We show that members of the Kcnq (Kv7.x) family of mRNAs are present in these tissues. Further, we demonstrated mRNA as well as the protein for Kcnq4 in ear and brain extracts from adult zebrafish. Although signals for kcnq1 and kcnq5 were weak using end-point PCR, the transcripts were readily detected in all tissues using qRTPCR. These data are consistent with previous reports of the Kcnq1 channel expressed during development[22].

As previously found in mammals[1, 3–5, 23–25], kcnq2 was expressed in zebrafish brain, heart and ear. Similarly, in mammals, kcnq3 is usually found co-expressed in the same tissues. Kcnq4 was detected in ear and brain tissue using a Kcnq4 selective antibody. KCNQ4 is found in auditory hair cells in mammals and we suggest that it may be present in homologous cells in the zebrafish.

The amino acid sequences of zebrafish Kcnq4, as well as other members of the KCNQ channel family, are conserved across phylogeny[19]. One distinguishing characteristic of KCNQ2 and KCNQ3 is the presence of a clustering domain that allows interaction of KCNQ channels with Na⁺ channels in the nodes of Ranvier[26]. Another characteristic of KCNQ channels is that the structural assembly (homotetramer vs. heterotetramer) is dependent upon amino acid sequences in the carboxyl-terminal region[23]. In the case of zebrafish Kcnq4, the translated amino acid sequence that we derived (Figure 4A) is consistent with the head-linker-tail structure of KCNQ4 that supports a homotetrameric structure[23].

The more highly abundant KCNQ transcripts expressed in the brain (KCNQ2, KCNQ3, and KCNQ5) are possible contributors to a number of important electrophysiological functions that are necessary for normal cognitive function. That is, dysfunction of these channels has been associated with dementia, stroke, and epilepsy[24]. Very similar to the mammalian cochlea[4, 25], but perhaps more similar to the vestibular system[27], our results show that the zebrafish inner ear sensory tissues do express the kcnq 2-5 genes. The inhibition of KCNQ4 activity in the mammalian cochlea[28] or knockout mouse[29] causes sensory cell degeneration followed by deafness. However, unlike mammals, the zebrafish hair cells are capable of regeneration after acoustic or chemical insult[30, 31], and selected transcription factors among other putative molecules are key mediators of the regeneration[32, 33]. No variants of KCNQ5 are associated with sensory hearing loss in humans so perhaps its high abundance in the zebrafish ear is associated with regenerative capabilities.

Studies of the effects of exogenous regulators of zebrafish hair cell regeneration are at various stages of investigation[34]. Our identification of Kcnq channels in zebrafish may offer a new in vivo model system for screening KCNQ channel modulators/drugs and their effects on regeneration. Certain classes of drugs are being designed to modulate the activity of specific KCNQ-type channels[35–37], and our work suggests that screening this class of chemotherapeutic agents for functional[38–40], as well as for adverse effects (such as behavioral abnormalities) in the zebrafish is promising. Further, expression of the channels cloned from the zebrafish in heterologous systems[15, 41, 42] provides an attractive platform for electrophysiological studies since dissociated hair cells from the inner ear of the zebrafish are extremely difficult to patch (Moore, unpublished observations, 2010; however, see[43]).

Conclusions

Recent advances in sequencing the zebrafish genome have provided further insight into modeling human diseases[44, 45]. Nevertheless, the chromosomal localizations and/or complete sequencing of the kcnq4 gene remain to be completed. Western blots demonstrated that the Kcnq4 protein is expressed in the brain as well as the ear. Thus, using the zebrafish with its rapid developmental period as a laboratory specimen may accelerate genetic screening for more specific KCNQ channel mutants, and perhaps foster drug discovery strategies for chemotherapeutic intervention in diseases associated with mutations in the Kv (x) family of genes, e.g., conditions manifested in humans such as hearing loss, and tinnitus.

Methods

Animals

Animal procedures were approved by the NU-ACUC (Approval number 2006 - 1034) and were performed in accordance with regulations for the care and use of laboratory animals. Adult zebrafish (initial stock was a kind gift from Dr. Jacek Topczewski, Ann Lurie Children’s Research Medical Center, Northwestern University, Chicago, IL) were kept in an aquarium that was maintained at 25°C, filtrated, pH balanced, with frequent removal of excess nitrate, nitrite, ammonia, chloramines, and chloride. Exchange of conditioned tap water occurred at regular intervals. Two bottom feeder fish (Bristle nose catfish, Ancistrus temmincki) were kept in the aquarium to reduce the accumulation of waste. Animals were fed twice daily using a combination of flake or morsels that had been sterilized (UV illumination overnight) before usage. Wild type embryos were collected from natural matings in our lab, or ordered from ZIRC (University of Oregon, Eugene, OR), and were kept in 12-well clusters (~n = 6 each well) at 28.5°C in an air-only incubator. Stages were referred to in hours post-fertilization (hpf) or days post-fertilization (dpf)[9]. After 24-hpf, some larvae were maintained in 0.03% phenylthiourea to prevent melanin pigment formation[46] to ease the identification of a normally developed lateral line.

Tissue extraction

Zebrafish were sacrificed using a combination of Tricaine Methylsulphonate (MS-222) and ice, and the various tissues were rapidly removed. Zebrafish brain, heart, and ear were dissected, used immediately for experimentation, or pooled separately in 1.5 ml Cryovials and placed in liquid nitrogen until use. Total RNA was extracted from the tissues using TRIzol reagent (Invitrogen, Carlsbad, CA). Ear tissues were placed in 6.0 ml of low calcium saline (LCS, 10 mM HEPES, 100 μM CaCl2, 110 mM NaCl, 2.0 mM KCl, 2.0 mM MgCl2, 3.0 mM D-glucose, pH 7.3). EDTA and MgCl₂ (12 μl) were added to the ear tissues and incubated for 15 min to prevent calcium carbonate leakage from the inner ear otolithic structures.

RT-PCR and PCR

Primers for kcnq1-5 and controls were designed based on zebrafish DNA sequences found in publically available databases such as the NCBI (GenBank), and Ensembl. The nucleotide sequence was searched using BLASTN to determine the number and location of different exons. Primers were designed using PrimerQuest (IDT, Coralville, IA). Nested primers were designed to cross exon boundaries and selected to amplify a 200 – 600 bp fragment of the desired kcnq mRNA. Primers were selected for optimum base content and annealing properties to the desired mRNA. Searching the zebrafish genome and expressed mRNAs was conducted using the BLAST suite of programs[47].

Total RNA pellets were washed with 75% ethanol, dried, suspended in RNAse free water and stored at a temperature of 4°C. The RNA concentration and purity was quantified by spectrophotometry (Beckman DU-7500, Fullerton, CA) using the absorbance ratio of A260/A280. Individual PCR reactions contained forward and reverse primers for the desired target, and control reactions contained primers for β-actin. The one-step RT-PCR system (Invitrogen, Carlsbad, CA) was used for amplification of target mRNAs. For each reaction, 200 - 400 ng of total RNA was used for the RT-PCR.

Reactions were performed in a thermocycler for 30 cycles (Techne, TC-312, Minneapolis, MN) with recommended denaturation (94°C, 2 min), annealing (55°C, 30 s), extension (72°C, 2 min), and hold (72°C, forever). The PCR products were separated using 1.0 – 2.0% agarose gel electrophoresis with gels containing ethidium bromide. Gel photographs were taken (Kodak, DC 290, Rochester, NY), transferred and stored to a microcomputer (Dell Dimension 8200). The molecular ladder (M) of the gels was separated by 100 bp bands with the first band at the bottom being 100 bp; the brightest band near the top of the ladder is at 600 bp.

Sequencing of kcnq4 PCR products

Gel bands were excised and purified using purification columns (Catalog #K2100, Invitrogen, Carlsbad, CA) designed for agarose gel extracts. The purified DNA products were sequenced at Sequetech (Mountain View, CA). The 5′ PCR primers for each product were used as sequencing primers (see Table 1).

Quantitative RTPCR (qRTPCR)

qRTPCR was conducted using a Biorad MyIQ detection system. Second strand synthesis was conducted using the IScript cDNA synthesis kit (Biorad, Grand Island, NY) starting with 0.45 to 0.74 μg of mRNA from the target tissues (brain, ear, or heart). An aliquot of the cDNA (5-10 ng) was then subjected to quantitative PCR in 96 well plates in the BioRAD Icycler using IQ supermix (Biorad, Grand Island, NY) and primers (100 nM each) selected to generate 100-200 bp amplicons for kcnq1-5. GAPDH was used as an internal control (Table 4). After initial denaturation, samples were subjected to 40 cycles of PCR (95°: 15 s, 60°: 60 s) with SYBR green dye fluorescence read at the end of each cycle. Melting curves were obtained at the end of the runs to verify that single melting point species were generated in each reaction. Well factors were collected to compensate for differences in responses across the plate. Cycle times were calculated, and data were transferred to a spreadsheet for calculation of deltaCTs. Dilutions of the mRNA (1:2, 1:10, 1:50, and 1:250) of the mixed cDNA were used to determine amplification efficiency, as reported previously[48].

Table 4 Sequences of primers used for qRTPCR

Antibody preparation

A synthetic peptide CSGKMGFRDRIRMNNSRSS based upon the reported partial cDNA, and putative amino acid sequences for zebrafish Kcnq4 was prepared and conjugated to Kehoe Limpet hemocyanin (KLH) for immunization. Two rabbits were used for antibody production by Bio-Synthesis (Lewisberg, TX). Antisera were obtained 6 -10 weeks after immunization and characterized for immunoreactivity against zebrafish tissue extracts.

Western blotting

Zebrafish brain and ear were homogenized in a lysis buffer (20 mM Tris, 150 mM NaCl, 1.0 mM EDTA, 1.0 mM EGTA, 2.5 mM NaPyrophosphate, 1.0 mM Na Vanadate, 1.0 mM betaglycerol phosphate, 1.0 mg/mL leupeptin, and 1.0% Triton X-100, pH 7.5) using a motor driven pestle (5 bursts of 10 s) on ice. The crude homogenate was centrifuged at 1.4 × 10³ rpm (Eppendorf microfuge, Hauppauge, NY). The supernatant fraction was removed, saved and the pellet extracted again using lysis buffer plus 1.0% SDS. After centrifugation to remove insoluble material, the soluble pellet fraction was saved. Protein concentrations in each fraction were determined using a reagent (Pierce BCA, Pittsburgh, PA) as suggested by the manufacturer. Electrophoresis was conducted on 4 - 12% SDS Page minigels (Invitrogen, Grand Island, NY). A total of 25 - 30 μg of brain fractions, ear and heart fractions (10 - 12 μg) were loaded into separate lanes on the gel. After electrophoresis, the gel was blotted to PVDF membranes (Millipore, Billerica, MA) using a Biorad transfer cell and Tris-Glycine-20% methanol transfer buffer. Transfer was accomplished at a constant voltage (40 V) for 1.5 hrs at room temperature. Membranes were blocked in 5.0% nonfat dry milk in Tris-buffered saline (20 mM Tris -150 mM NaCl, pH 7.5) containing 0.1% Tween 20 (TBS-T). Blots were then treated with anti-Kcnq4 peptide antiserum (1:2000) in TBS-T with 5.0% nonfat dry milk at room temperature for 2.0 hrs. After washing 5x with TBS-T, the blots were then incubated with HRP-conjugated Goat anti-rabbit antibody (Bio-Rad, 1:5000) for 1.0 hr at room temperature. After washing 5x with TBS-T, the blot was developed with electrochemiluminescent substrate (Pierce West Pico, Pittsburgh, PA) and bands were detected on film (Kodak Biomax, Rochester, NY).

Data analysis

Expression intensity of gene transcripts was analyzed using Image J (NIH Image, Bethesda, MD). Data display was accomplished using Origin (v8.0 Origin Software, Northampton, MA).