Characterization of an Abundant COL9A1 Transcript in the Cochlea with a Novel 3′ UTR: Expression Studies and Detection of miRNA Target Sequence
- First Online:
- 552 Downloads
EST N66408 represents one of several large unique clusters expressed in the Morton human fetal cochlear cDNA library. N66408 is 575 bp in size and initial BLAST analysis of this sequence showed no homology to any known genes or expressed sequence tags (ESTs) from other organs or tissues. Sequence of the original cochlear clone from which N66408 was derived revealed that the corresponding cDNA was about 700 bp in size, including 125 bp at its 5′ end with homology to the 3′ end of COL9A1 in addition to 575 bp of novel sequence. RT-PCR analysis using primers specific to COL9A1 isoforms 1 and 2 detected expression of both isoforms in human fetal cochlea. Tissue in situ hybridization using the novel 3′ UTR sequence as probe showed abundant expression in spiral limbus and spiral ligament, and a moderate level of expression in the organ of Corti. dbEST analysis of ESTs specific to the 3′ UTR of COL9A1 showed 19 ESTs derived from various tissues; three polyadenylation sites were identified and the majority of these ESTs were derived from overlapping polyadenylation signals at the second site (position 749–758). Comparison of the 3′ UTR of human COL9A1 with its orthologs as well as with dbEST uncovered a highly conserved region around the overlapping polyadenylation signals at position 749–758 in mammals. A search of the microRNA database revealed a highly conserved target sequence for miR-9 immediately preceding the overlapping polyadenylation signals in the novel 3′ UTR of COL9A1, suggesting its role in posttranscriptional regulation of COL9A1.
KeywordsCOL9A1 long form or isoform 1 short form or isoform 2 novel 3′ UTR ESTs cochlea
Hearing loss is the most common sensory disorder in humans. Approximately one in 1,000 children in the United States has congenital deafness (Morton 1991), and at least half of these cases has a genetic etiology. About 70% of cases attributed to genetic causes represent nonsyndromic deafness, and the remaining 30% are syndromic and associated with other clinical findings. Among nonsyndromic deafness, autosomal recessive inheritance accounts for about 80% of cases, autosomal dominant inheritance is observed in about 20%, and X-linked (about 1%) and mitochondrial (less than 1%) forms are comparatively infrequently observed. The unparalled heterogeneity seen in heritable deafness suggests the involvement of many genes in the hearing process. More than 95 loci have been mapped for nonsyndromic hearing loss (i.e., 40 autosomal dominant, 51 autosomal recessive, 4 X-linked, and several mitochondrial loci), and more than 60 genes have been identified (as of February 2006) as responsible for syndromic and nonsyndromic sensorineural hearing loss (http://webhost.ua.ac.be/hhh).
The impressive genetic heterogeneity in nonsyndromic deafness confounds traditional genetic linkage analysis. An alternative method for identification of disease-causing genes is one that utilizes tissue-specific cDNA libraries (Hedrick et al. 1984; Jones and Reed 1989; Gurish et al. 1992). To identify genes responsible for hearing loss and to gain a more fundamental understanding of human hearing at a molecular level, we constructed a human fetal cochlear cDNA library (Robertson et al. 1994; Skvorak et al. 1999). Various methods were undertaken to analyze transcripts in this library, namely, the combined approaches of subtractive hybridization and differential screening and expressed sequence tag (EST) production and analysis (Robertson et al. 1994; Skvorak et al. 1999; Resendes et al. 2002). More than 15,000 ESTs were generated. About 8,500 ESTs corresponded to more than 2,300 known genes, more than 1,770 EST clusters were expressed in other libraries, and approximately 340 EST clusters were unique to the cochlear library (Skvorak et al. 1999; Resendes et al. 2002). Analysis of the cochlear ESTs revealed 788 discrete genetic loci, some of which reside within genomic regions for mapped deafness disorders and represent potential positional candidate genes for hearing impairment (http://hearing.bwh.harvard.edu). With these genomic approaches, several novel auditory genes such as COCH, OTOR, and PFET1 (now designated KCTD12) have been identified (Robertson et al. 1997, 2000; Resendes et al. 2004). Missense mutations in COCH were subsequently shown to be etiologic in the autosomal dominant sensorineural hearing loss and vestibular disorder, DFNA9 (Robertson et al. 1998). OTOR is an inner ear preferentially expressed gene as determined by UniGene dbEST and NCBI BLAST searches (Beisel et al. 2004). The recently described KCTD12 is an intronless gene abundantly expressed in the fetal inner ear and a variety of other fetal organs, but at barely detectable levels in adult organs (Resendes et al. 2004).
In this work, we present characterization of an EST from one of the unique EST clusters identified from the sequence analysis of the human fetal cochlear cDNA library. This EST was subsequently determined to represent a novel 3′ UTR of the short form of a collagen gene, COL9A1, and was selected for further investigation based on the fundamental role of collagens in the development and maintenance of hearing. Collagens expressed in the inner ear and essential for hearing are illustrated by the association of defects in several syndromic disorders including hearing loss, such as mutations in COL1A1 and COL1A2 in osteogenesis imperfecta (Kuivaniemi et al. 1991, 1997); COL2A1 and COL11A1 in classic Stickler syndrome (Ahmad et al. 1991; Winterpacht et al. 1993; Richards et al. 1996); COL11A2 in nonocular Stickler syndrome (Sirko-Osadsa et al. 1998) and otospondylomegaepiphyseal dysplasia (OSMED) syndrome (van Steensel et al. 1997); COL11A1 in Marshall syndrome (Griffith et al. 2000); and COL4A3, COL4A4, and COL4A5 in Alport syndrome (Barker et al. 1990; Lemmink et al. 1994; Mochizuki et al. 1994). Mutations in COL11A2 have also been found in DFNA13, an autosomal dominant nonsyndromic sensorineural hearing loss (McGuirt et al. 1999).
Materials and methods
EST sequence analysis
A unique EST cluster representing the 3′ UTR of COL9A1 was derived from the unsubtracted, nonnormalized Morton fetal cochlear cDNA library. Briefly, mass in vivo excision of the human fetal (16–22 weeks gestational age) cochlear cDNA library, constructed in the UniZap vector, was performed according to the manufacturer's protocol (Robertson et al. 1994) (Stratagene, La Jolla, CA, USA). The cDNA library was then contributed to the IMAGE Consortium to generate human cochlear ESTs, and sequences subsequently deposited in GenBank (Skvorak et al. 1999).
Nucleotide sequence of cDNA clones was determined by using an ABI PRISM dye-terminator cycle-sequencing system (PE Applied Biosystems, Foster City, CA, USA). Sequence analysis was performed using the University of Wisconsin Genetics Computer Group software (Devereux et al. 1984). The nucleotide sequence of N66408 was compared to sequences contained in various databases including the GenBank primate database and sequences in the EST, STS, and nonhuman mammalian databases (Skvorak et al. 1999) using the BLAST network service of the National Center for Biotechnology Information (Altschul et al. 1997).
Northern blot analysis
Total cellular RNAs were extracted (Chirgwin et al. 1979) from second trimester human fetal cochlea, brain, skeletal muscle, testis, eye, placenta, thymus, spleen, tongue, liver, kidney and bone marrow. All human tissues were obtained according to guidelines established by the Partners Human Research Committee. Ten micrograms of each of the RNAs was electrophoresed in 1% agarose–formaldehyde gels and transferred to GeneScreen (DuPont, Wilmington, DE, USA) filters (Thomas 1980). Filters were prehybridized for 2–4 h and hybridized overnight at 42°C with a 32P-labeled probe derived from the original human EST N66408 clone. After incubation, filters were washed in 0.1× SSC in 0.1% SDS at 42–55°C prior to autoradiography overnight using XAR-5 film (Kodak, Rochester, NY, USA) with intensifying screens at −80°C.
Reverse transcription-polymerase chain reaction
Oligonucleotide primers used in COL9A1 RT-PCR
Isoform 1 specific
COL9A1F1-F: CCA GAA CAC ATA GTC CTA GGG TAA
COL9A1F1-R: ATG ATC TTA TGC CAC TGG GAA T
COL9A1F2-F: AAC AGC AGC CTT TTC GAA TTT
COL9A1F2-R: AGG GTC ACC AAC AGG TCC TAC
Isoform 2 specific
Iso2-F: ACG CTT TGG CAA CCG CTA CTC
Iso2-R: ACA AGC AGA GCC CCA ACA
COL9A1F3-F: GTA TTC CTG GAC CCC CTG GT
COL9A1F3-R: CTA CCC TTT TCA CCA GCA ACA C
COL9A1F4-F: ATT GCA GGG GTT ACC AGG TCT A
COL9A1F4-R: GAG TCT GGA CGC TTA AGA CTG G
COL9A1F5-F: ATT AAG CAG GTT TGC ATG AGA GT
COL9A1F5-R: GCA CCG TTC TTC TAT ATG TGA TTG
COL9A1F6-F: GAA CCA CGC CTG GTG AAG
COL9A1F6-R: CCA AAC AGT TGT TTT TGT TTA TTG G
Novel 3′ UTR
COL9A1novel -F: AGT TGC AGT AGT TAT TTT CAT TTA G
COL9A1novel-R: AAT AAG CTC AGT AAC TTC TTT TCA TTG
Tissue in situ hybridization
Radioactive in situ hybridization was carried out with a 450-bp PCR-generated probe synthesized from the novel 3′ UTR of COL9A1, amplified by using gene-specific primers with T7 and SP6 RNA polymerase-specific promoter sequences at the 5′ end of forward and reverse primers, respectively. Primers and conditions used for amplification were as follows: forward primer (5′-GAA TTG TAA TAC GAC TCA CTA TAG GGA GTT GCA GTA GTT ATT TTC ATT TAG-3′) and reverse primer (5′-CAA GCT ATT TAG GTG ACA CTA TAG AAC ATG CAA GAG ATT AAG AAA CCA-3′); initial denaturation at 96°C for 5 min; 30 cycles of 97°C for 30 s, 50°C for 30 s, and 72°C for 30 s, and a final extension at 72°C for 7 min. The gel-purified PCR product with a final concentration of 1 μg/μL was used to prepare 35S-labeled riboprobe using an in vitro transcription kit (Roche Molecular Biochemicals, Mannheim, Germany). The efficiency of transcription and incorporation of 35S-UTP into riboprobe were evaluated by 5% acrylamide-urea gel electrophoresis and quantification of counts by using a scintillation counter, respectively.
Cochlear tissue from a 20-gestational-week-old human fetus was fixed in 4% paraformaldehyde/1× PBS for 3 days, decalcified in 0.1 M EDTA/1× PBS, pH 7.3 for 7 days, and dehydrated before being embedded in paraffin by standard histologic procedures. Sections (17 μm) were cut with a cryomicrotome (Leica, Malven, PA, USA), collected on silyated slides (PGC Scientific, Gaithersburg, MD, USA), dried at 37°C for 45 min, and stored frozen at −20°C until use.
In situ hybridization
Tissue sections were deparaffinized, fixed in 4% paraformaldehyde in PBS, and treated with proteinase K. After washing in 0.5× SSC, sections were prehybridized in 50% deionized formamide, 0.3 M NaCl, 20 mM Tris (pH 8.0), 5 mM EDTA, 1× Denhardt's solution, 10% dextran sulfate, and 10 mM dithiothreitol for 2 h at 60°C. 35S-labeled antisense and sense RNA probes (3×106 cpm/slide) were then added separately to the hybridization solution and incubated for 12–18 h at 60°C. After hybridization, sections were washed for 20 min in 2× SSC, 10 mM β-mercaptoethanol, and 1 mM EDTA, treated with RNase A (20 μg/mL) for 30 min at room temperature, and washed at high stringency (0.1× SSC, 10 mM β-mercaptoethanol, and 1 mM EDTA) for 2 h at 60°C. Sections were dehydrated, dipped in photographic emulsion NTB-3 (Kodak), and maintained at 4°C for 2–4 weeks of exposure. Sections were then developed and counterstained with hematoxylin and eosin (H&E), and coverslips were applied with 50% Canada balsam in methyl salicylate (Sigma, St. Louis, MO, USA).
Identification of polyadenylation sites in the 3′ UTR of COL9A1
Fifteen hundred bases of sequence of the 3′ UTR starting from the first base after the stop codon of COL9A1 were retrieved from the Ensembl genome browser (Gene ID no. ENSG00000112280) and analyzed to identify the number and location of polyadenylation sites using the ESTparser program http://tagc.univ-mrs.fr/bioinfo/ESTparser/ (Beaudoing and Gautheret 2001). ESTparser predicts potential polyadenylation signals in the sequence, performs BLAST searches against the EST databases, and filters the output to produce a general picture of alternative polyadenylated forms and the tissues in which they occur. The sequence was also compared to the human EST database to retrieve COL9A1 3′ UTR-specific ESTs expressed in various tissues. Because the actual 3′ ESTs are not consistently annotated in the database, we used criteria proposed by Beaudoing et al. (2000) to select ESTs for further analysis: (1) ESTs of at least 40 nucleotides (nt) matching the COL9A1 3′ UTR with at least 95% similarity over the entire length of the EST sequence, except for the allowed 25 and 5 nt mismatches at the EST 5′ and 3′ ends; (2) internal priming assessed by adenine stretches in the COL9A1 3′ UTR flanking the 3′ extremity of the EST [i.e., internal priming as a result of cDNA oligo d(T) primers hybridizing to internal poly (A) stretches instead of the actual poly (A) tail]; and (3) ESTs with poly (A) or poly (T) termini (in antisense strand) of length 10 nt or more at the end of the transcript. ESTs that fulfilled the first two criteria were aligned to their corresponding positions in the 3′ UTR of COL9A1 to identify tissue-specific polyadenylation sites.
A large fraction of human polyadenylation sites are flanked by upstream and downstream sequence elements, located around positions 0 to −50 and +20 to +60, respectively, relative to the polyadenylation signal and distinguishing true polyadenylation sites from randomly occurring AAUAAA hexamers. The vicinity of the polyadenylation sites in the 1.5-kb 3′ UTR of COL9A1 was also analyzed with the ERPIN program (http://tagc.univ-mrs.fr/erpin/) to identify true polyadenylation sites within this sequence based on sequence present upstream of the polyadenylation signal and downstream of the cleavage site (Legendre and Gautheret 2003).
Identification of regulatory motif in the novel 3′ UTR
About 1.5 kb of sequence from the 3′ UTR of Col9a1 was retrieved from the Ensembl genomic sequences of other species, including mouse (ENSMUSG00000026147), rat (ENSRNOG00000012920), dog (ENSCAFG00000002589), chicken (ENSGAGG00000015970), fugufish (SINFRUG00000146962), pufferfish (GSTENG0030712001), and western clawed frog (ENSEXTG00000017926). To identify conserved regulatory elements, the 3′ UTR of COL9A1 was compared with that of its orthologs retrieved from Ensembl as well as with EST databases of mouse and other species using BLAST (Altschul et al. 1997). The microRNA database (http://www.microrna.org) was searched for the presence of a microRNA target in the 3′ UTR of COL9A1 and in its rat and mouse orthologs. The 3′ UTR of the canine ortholog was also evaluated for any conserved microRNA target sequences by using the miRranda computer program (John et al. 2004).
Northern blot analysis
Five overlapping sets of short form COL9A1 transcript-specific primers (Table 1, Fig. 1a) used for RT-PCR of cDNA synthesized from total RNA extracted from 21-gestational-week-old human fetal cochlea revealed expression of the short form (data not shown). Furthermore, sequence analysis showed that this transcript contained the novel 3′ UTR with no recognized sequence similarity to COL9A1 transcripts. RT-PCR analysis performed to assess expression of the novel 3′ UTR sequence in human fetal tissues showed an intense amplification product of this sequence in cochlea and in patella, a notably less intense product in testis, brain, eye, trachea, and skeletal muscle, and no detectable expression in placenta, epiglottis, and skull (Fig. 2B). RT-PCR was also carried out to evaluate expression of both long and short forms of COL9A1 in various tissues using 5′ end isoform-specific primers. Expression of the long form of COL9A1 was seen in trachea, patella, cochlea, and skull (Fig. 2C). The short form of COL9A1 was found to be expressed in human fetal cochlea, eye, patella, testis, brain, skeletal muscle, epiglottis, trachea, and skull, with no detectable expression in placenta (Fig. 2D). For a positive control, RT-PCR analysis using GAPDH-specific primers showed products of equal intensities in all tissues, confirming the quality of cDNAs (Fig. 2E).
In situ hybridization
Multiple polyadenylation signals in the 3′ UTR of COL9A1
Analysis of a 1.5-kb region from the 3′ UTR of COL9A1 with the ESTparser program detected 17 potential polyadenylation signals: an AATAAA hexamer at two sites and seven of its single-base variants in 15 additional sites. A single nucleotide polymorphism at position 266 was found to result in an additional primary polyadenylation signal at position 264–269 (AAC/TAAA). The ESTparser program detected only 19 ESTs from dbEST with sequence identity/similarity to the 1.5-kb region of the 3′ UTR of COL9A1; the majority of ESTs were found to have putative polyadenylation sites either at positions 774 or 853.
Distribution of polyadenylation sites in COL9A1 ESTs
Putative polyadenylation site
Site 1 at position 371
Eye: BU726888, Chondrosarcoma: CA447254
Site 2 at position 774
Cochlea: BI492913, BI494749, BI495178, BI493009, BI497335, BI496483, N69335, BQ007876
Chondrosarcoma: BQ004883, CA419175, BQ003610
Malignant melanoma: AI986458
Pooled tumors: AI745096
Pooled fetal tissues: BF196880
Site 3 at position 853
Cochlea: N66408, BI492246, AW021120
Analysis of 3′ UTR sequences revealed two AAUAAA sites at positions 343–348 and 753–758 as positive sites with scores of 12.14 (E value of 4.42 × 10−4) and 13.86 (E = 4.01 × 10−5), respectively. The score is computed directly from the lodscore profile and E is the number of solutions expected by chance at this score level. These E values indicate that these two sites are very likely to be true sites (Gautheret, personal communication).
miR-9 target sequence in the 3′ UTR
The 3′ UTR of COL9A1 was compared with that of its orthologs from mouse, rat, dog, chicken, fugufish, zebrafish, and pufferfish. This analysis revealed ∼80% to 90% homology at two regions within the novel 3′ UTR in dog, mouse, and rat. One of these sites contained the highly conserved overlapping polyadenylation signals, where most of the transcripts were found to be cleaved during polyadenylation. This region was also conserved in pig and cow (Fig. 4b). We did not observe any such similarity between the 3′ UTR of human COL9A1 and its orthologs in chicken, fugufish, zebrafish, and pufferfish. A search for regulatory motifs in the 3′ UTR of COL9A1, especially within the highly conserved region, revealed a highly conserved target sequence for one of the microRNAs, miR-9, immediately preceding the overlapping polyadenylation signals at position 749–758 in the novel 3′ UTR of COL9A1 and conserved orthologs in mouse and rat. Analysis of the 3′ UTR of the dog ortholog with the miRranda program demonstrated a highly conserved potential miR-9 target sequence, which is also highly conserved in cow and pig ESTs homologous to the 3′ UTR of COL9A1.
Analysis of EST N66408 revealed homology to COL9A1 at the 5′ end and showed no homology to any known gene transcript in the remainder of the sequence. An independent cDNA library prepared with human fetal cochlear RNA obtained from our cochlear RNA collections, using suppression subtractive hybridization against a mixture of cDNAs from various fetal tissues to increase the proportion of tissue-specific genes in the cDNA library, also revealed abundant expression of COL9A1-specific clones (Luijendijk et al. 2003).
Type IX collagen is one of the important components in the tectorial membrane of the organ of Corti, along with three other collagens (collagen types II, V, and XI) and noncollagenous proteins (alpha-tectorin, beta-tectorin, otogelin, and glycoproteins) (Goodyear and Richardson 2002). Type IX collagen is a heterotrimer of polypeptide chains encoded by three genes, COL9A1, COL9A2, and COL9A3, and these chains fold into the triple-helix structure, an important characteristic feature of members of the collagen family of extracellular matrix proteins (Olsen 1997). The triple helix of COL9A1 consists of COL1, COL2, and COL3 domains, which are separated and flanked by noncollagenous segments, domains NC1–NC4. COL9A1 is synthesized in two forms (Nishimura et al. 1989): a short form of about 2.6 kb and a long form of approximately 3.3 kb. The COL9A1 long form (also known as v1 or COL9A1_v1), found in hyaline cartilage, is transcribed from an upstream promoter and generates a product that contains an additional N-terminal globular domain of 245 amino acids, called NC4 (Wu et al. 1992). In contrast, the COL9A1 short form (v2 or COL9A1_v2), expressed in vitreous and several other tissues (Svoboda et al. 1988; Fitch et al. 1988; Wright and Mayne 1988; Perris et al. 1991; Swiderski and Solursh 1992), uses an alternative downstream promoter and an alternative exon; the globular NC4 domain is replaced with a short alternative amino acid sequence at the N-terminal end (Nishimura et al. 1989). The two other polypeptide chains, transcribed from COL9A2 and COL9A3, also lack this NC4 domain (Mayne and Brewton 1993).
Expression of the novel short form of COL9A1 in cochlea
Northern blot analysis of RNAs from various human tissues showed a strong hybridizing band corresponding to an ∼3.5-kb transcript in fetal cochlea. Further RT-PCR analysis confirmed expression of the full-length transcript of the short form of COL9A1 with the novel 3′ UTR, representing a new alternative transcript of COL9A1, designated as v3 or COL9A1_v3. Sequence analysis of 22/26 COL9A1-specific clones obtained from a suppression subtracted human fetal cochlear cDNA library (Luijendijk et al. 2003) also showed the novel 3′ UTR in 15 clones. In addition to human fetal cochlea, several other tissues showed expression of this novel 3′ UTR by RT-PCR analysis. Cochlea, trachea, skull, and patella were found to express both short and long forms; it remains to be determined whether the novel 3′ UTR is specific to the short form or present in both isoforms. This new variant with the novel 3′ UTR might have been derived from alternative polyadenylation sites present in the 3′ UTR. Northern blot analyses of RNA from chicken sternal cartilage was previously reported to show two different transcripts of Col9a1 produced by usage of alternative polyadenylation sites (Svoboda et al. 1988).
Expression of the COL9A1 short form with the novel 3′ UTR (v3) has not been described previously, although expression of the short isoform (v2) has been detected in other tissues including chicken cornea (Svoboda et al. 1988; Fitch et al. 1988; Nishimura et al. 1989), vitreous humor (Wright and Mayne 1988; Yada et al. 1990), notochord, axial skeleton, chondrocranium, Meckel's cartilage (Swiderski and Solursh 1992), and perinotochordal matrix (Perris et al. 1991). Swiderski and Solursh (1992), while comparing the spatial and temporal expression patterns of transcripts of avian type II collagen and the long and short forms of the alpha 1 type IX collagen by in situ hybridization, observed expression of the short form of Col9a1 in the developing chicken (stages 25–28) nonchondrogenic notochord along with type II collagen mRNA. They also found expression of the short form of Col9a1 to be more restricted in the developing chondrogenic vertebrae of the axial skeleton, chondrocranium, and Meckel's cartilage, although expression of both long and short forms of Col9a1 was seen along with type II collagen transcripts (Swiderski and Solursh1 1992). Using Northern blot and RNase protection assays, Savontaus et al. (1998) detected expression of both isoforms and only the short form of Col9a1 in the developing mouse embryonic limbs and eyes, respectively. Expression of the short form of Col9a1 was also observed in bovine nucleus pulposus, the gel-like central zone of the young intervertebral disk that has a similar collagen phenotype to that of hyaline cartilage (Wu and Eyre 2003).
Expression of novel 3′ UTR of COL9A1 in human fetal cochlear sections
Our tissue in situ hybridization studies detected abundant expression of the novel 3′ UTR of COL9A1 in the spiral ligament and limbus. The high level and widespread expression of the novel 3′ UTR of COL9A1 in the cochlea is in agreement with a strong hybridizing band in human fetal cochlea by Northern blot analysis. Immunolocalization studies of type IX collagen have demonstrated its expression in the tectorial membrane, an acellular structure that overlies the cochlear sensory hair cells and plays an important role in the transduction process, along with type II collagen in the mouse inner ear (Slepecky et al. 1992; Goodyear and Richardson 2002; Suzuki et al. 2005). The tectorial membrane is composed of two different components: the radial collagen fibril and striated sheet matrix (Goodyear and Richardson 2002). Immunoelectron microscopic findings showed that types II and IX collagens colocalize in collagen fibril bundles, indicating the contribution of type IX collagen to the three-dimensional integrated structure of type II collagen molecules (Asamura et al. 2005).
Multiple polyadenylation sites in the 3′ UTR of COL9A1
About 29% of human mRNAs have multiple polyadenylation sites (Beaudoing et al. 2000) and the choice of poly (A) site usage may influence stability, translational efficiency, or localization of an mRNA in a tissue- or disease-specific manner (Edwalds-Gilbert et al. 1997). Polyadenylation sites are primarily defined by a polyadenylation signal of an AATAAA hexamer or its single-base variants, located between 10 and 35 nt upstream of the actual cleavage and polyadenylation site. Analysis of ESTs specific to the COL9A1 3′ UTR revealed three putative polyadenylation sites present within the novel 3′ sequence expressed in cochlea at positions 371, 774, and 853 following the stop codon. Our analysis demonstrated that position 774 was the site used most commonly and that this site comprises two overlapping polyadenylation signals, AAGAAA/AATAAA. It is unknown which of these two overlapping signals was involved in cleavage and polyadenylation. However, given that the AATAAA hexamer is the most commonly observed polyadenylation signal (58.2%) (Colgan and Manley 1997; Beaudoing et al. 2000) and AAGAAA is the least commonly observed variant, it seems likely that AATAAA is involved in the cleavage and polyadenylation of the transcripts at position 774. Three cochlear ESTs were derived from the putative polyadenylation site at 853. It is unknown whether expression of two alternative transcripts in cochlea is cellular-specific or whether expression of both isoforms occurs in all cell types within the cochlea. Because the difference between the two transcripts of COL9A1 is about 80 bases in size, it is understandable as to why two distinct bands were not detectable by Northern blot analysis of human fetal cochlear RNA.
Highly conserved miR-9 target in the novel 3′ UTR
A high degree of sequence conservation seen in mammals in the 3′ UTR of COL9A1, particularly around one of the polyadenylation signals, suggests the presence of important key regulatory elements necessary for proper polyadenylation. A search of the microRNA (miRNA) database (http://www.microrna.org), as well as an analysis using miRanda software (John et al. 2004) to identify potential regulatory motifs in the 3′ UTR of COL9A1, revealed a putative target sequence for one of the microRNAs, miR-9, in the immediate upstream vicinity of the AATAAA hexameric signal at position 753–758. MicroRNAs are about 22 nt single-stranded RNAs that regulate expression of target genes by interacting with complementary sites in the 3′ UTR of the target mRNAs (Bartel 2004). When target sites are not perfectly matched to the miRNA, which is seen in the majority of 3′ UTRs, the miRNA suppresses translation of the mRNA by an unknown mechanism (Zeng et al. 2002; Zeng and Cullen 2003). In cases with perfect or near-perfect sequence matches, miRNAs mediate cleavage of the target mRNA (Yekta et al. 2004). These miRNAs have critical roles in a variety of biological processes. In Caenorhabditis elegans, the let-7 RNA acts to promote the transition from the late-larval to adult cell fates in the same manner that lin-4 RNA acts earlier in development to promote progression from the first larval stage to the second stage (Reinhart et al. 2000; Slack et al. 2000). In mammals, miRNAs were shown to be involved in developmental regulation of subsets of lymphoid lineages (Chen et al. 2004). As of February 2006, 326 human miRNA sequences giving rise to 319 unique mature miRNAs are found in the microRNA database (miRBase) http://www.sanger.ac.uk/Software/Rfam/mirna/index.shtml. miR-9 is encoded by three genes, miR-9-1, miR-9-2, and miR-9-3, located on human chromosomes 1, 5, and 15, respectively (Lagos-Quintana et al. 2002). Recently, miR-9 was shown to play an important functional role during neural cell differentiation (Krichevsky et al. 2003, 2005; Smirnova et al. 2005).
As many miRNA target sequences are conserved across large evolutionary distances, they must be subjected to strong functional constraints (John et al. 2004). The target sequence, which we found in the 3′ UTR of the short form of COL9A1, was highly conserved in dog, cow, pig, mouse, and rat, and this region was shown to have a potential candidate target for miR-9 (John et al. 2004). The expression pattern of miR-9 in inner ear and its functional importance remain to be elucidated.
Importance of COL9A1 in hearing
To date, mutations and polymorphisms in COL9A1 have not been reported with any form of hearing loss. However, a small insertional mutation in intron 8 has been shown in affected members of a family with autosomal dominant multiple epiphysial dysplasia (Czarny-Ratajczak et al. 2001). Recently, a knockout mouse model lacking both isoforms of Col9a1 showed progressive hearing loss as well as an abnormality in the shape of the tectorial membrane (Suzuki et al. 2005). Electron microscopic analysis of the tectorial membrane revealed disorganized, aggregated, and fused fiber-like structures in the 1-month-old homozygous knockout mice (Suzuki et al. 2005). Deficiency of alpha 1 type IX collagen was also found to lead to a functional knockout of the other two Col9 polypeptide chains, alpha 2 and alpha 3, which were transcribed normally (Hagg et al. 1997). Type II collagen, which is colocalized with type IX collagen, was likewise not immunohistochemically detected in the tectorial membrane of the homozygous Col9 knockout mice, suggesting that lack of type IX collagen may also affect the three-dimensional structure of type II collagen (Asamura et al. 2005). Humans and mice with mutations in COL2A1 have hearing loss associated with several other phenotypic abnormalities (Richards et al. 2000; Donahue et al. 2003; Lopponen et al. 2004). Given the expression pattern of COL9A1 and the electrophysiological and morphological abnormalities of the inner ear shown in the homozygous knockout mice for type IX collagen, it is likely that type IX collagen plays an important functional role along with type II collagen in the development and function of the inner ear, and that mutations in COL9A1 may be expected to result in hearing loss in humans.
The authors wish to recognize the support of this research by the National Institutes on Deafness and Other Communication Disorders, DC03402 (to C.C.M) and F32DC00405 (to B.L.R). Also, the authors thank Dr. Hannie Kremer for communication of the sequence analysis of the COL9A1-specific clones from the suppression subtracted human fetal cochlear cDNA library, Drs. Massimo Loda and Yu Yang for technical assistance with the in situ hybridization provided through the Dana Farber–Harvard Cancer Center in situ hybridization core, Dr. Charles Lee for his help in photography of the cochlear in situ hybridization, and Drs. Bjorn Olsen, Kirk Beisel and Garret Soukup for critical review of the manuscript.
- Czarny-Ratajczak M, Lohiniva J, Rogala P, Kozlowski K, Perala M, Carter L, Spector TD, Kolodziej L, Seppanen U, Glazar R, Krolewski J, Latos-Bielenska A, Ala-Kokko L. A mutation in COL9A1 causes multiple epiphyseal dysplasia: further evidence for locus heterogeneity. Am. J. Hum. Genet. 69:969–980, 2001.CrossRefPubMedGoogle Scholar
- Donahue LR, Chang B, Mohan S, Miyakoshi N, Wergedal JE, Baylink DJ, Hawes NL, Rosen CJ, Ward-Bailey P, Zheng QY, Bronson RT, Johnson KR, Davisson MT. A missense mutation in the mouse Col2a1 gene causes spondyloepiphyseal dysplasia congenita, hearing loss, and retinoschisis. J. Bone Miner. Res. 18:1612–1621, 2003.CrossRefPubMedGoogle Scholar
- Krichevsky AM, Sonntag KC, Isacson O, Kosik KS. Specific microRNAs modulate ES cell-derived neurogenesis. Stem Cells, 2005.Google Scholar
- McGuirt WT, Prasad SD, Griffith AJ, Kunst HP, Green GE, Shpargel KB, Runge C, Huybrechts C, Mueller RF, Lynch E, King MC, Brunner HG, Cremers CW, Takanosu M, Li SW, Arita M, Mayne R, Prockop DJ, Van Camp G, Smith RJ. Mutations in COL11A2 cause non-syndromic hearing loss (DFNA13). Nat. Genet. 23:413–419, 1999.CrossRefPubMedGoogle Scholar
- Mochizuki T, Lemmink HH, Mariyama M, Antignac C, Gubler MC, Pirson Y, Verellen-Dumoulin C, Chan B, Schroder CH, Smeets HJ et al. Identification of mutations in the alpha 3(IV) and alpha 4(IV) collagen genes in autosomal recessive Alport syndrome. Nat. Genet. 8:77–81, 1994.CrossRefPubMedGoogle Scholar
- Richards AJ, Baguley DM, Yates JR, Lane C, Nicol M, Harper PS, Scott JD, Snead MP. Variation in the vitreous phenotype of Stickler syndrome can be caused by different amino acid substitutions in the X position of the type II collagen Gly–X–Y triple helix. Am. J. Hum. Genet. 67:1083–1094, 2000.PubMedGoogle Scholar
- Robertson NG, Lu L, Heller S, Merchant SN, Eavey RD, McKenna M, Nadol JB Jr., Miyamoto RT, Linthicum FH Jr., Lubianca Neto JF, Hudspeth AJ, Seidman CE, Morton CC, Seidman JG. Mutations in a novel cochlear gene cause DFNA9, a human nonsyndromic deafness with vestibular dysfunction. Nat. Genet. 20:299–303, 1998.CrossRefPubMedGoogle Scholar