A role of the double-stranded RNA-binding protein PACT in mouse ear development and hearing

Results

For probing functions of a mouse protein, disruption of the corresponding gene is often used as a tool. This approach is quite feasible for single-copy genes, as is the case for the Pact gene. However, it is often difficult to decide which exon of the gene to disrupt. Disruption of the exon containing the translation start site is the popular choice, but in many such cases, truncated proteins are expressed from the disrupted allele through the use of cryptic internal translation initiation sites. To avoid this problem for ablating Pact expression in mice, we targeted exon 8, which encodes its domain 3, because this domain is necessary and sufficient for activating PKR (4). Thus, even if the disrupted PACT allele were expressed in the knockout mice, the truncated protein containing only domains 1 and 2 would not be able to activate PKR. To the contrary, experiments in cultured cells suggested that such a protein should act as an inhibitor of PKR activation (14). With the above reasons in mind, we constructed a targeting vector containing a Neo cassette flanked by a part of exon 7, intron 7, and a part of exon 8 on the 5′ side. Part of exon 8 and the downstream genomic sequence were at the 3′ side, followed by a TK cassette. Moreover, we introduced a translation stop codon and a polyadenylylation signal in exon 8 just 5′ to the Neo cassette (Fig. 1B). Homologous recombination of this targeting vector with the resident gene should delete ≈500 bp from exon 8 and introduce the Neo gene in its place (Fig. 1C). Digestion of the disrupted Pact gene with SacI should produce a 6.4-kb DNA fragment, instead of a 5-kb fragment, that can hybridize with an exon 7-specific probe. An ES cell clone with a disrupted allele was isolated, and chimeric mice were generated from this clone. Successful propagation of the disrupted allele produced the founder Pact^+/− mice. After several generations, these mice of mixed genetic background were bred with pure C57BL/6 mice, and the Pact^+/− progenies were successively back-crossed with C57BL/6 partners for 10–14 successive generations. This crossing led to the establishment of a Pact^+/− line in the C57BL/6 genetic background. The genotyping of wild-type, Pact^+/−, and Pact^−/− mice is shown in Fig. 1D.

Unexpectedly, no truncated protein was expressed from the disrupted Pact allele (Fig. 2A). For detecting Pact expression, Western blot analysis was performed with extracts of several tissues in which PACT is expressed at a high level; the antibody used in these experiments recognizes specifically domain 2 of PACT. As positive controls, we used extracts of tissue culture cells expressing transfected wild-type PACT or Pact Δ3, which was the expected product of the targeted allele in mice. As shown in Fig. 2A, the expected full-length PACT was expressed in the brain of wild-type mice, but not in Pact^−/− mice. The authenticity of the protein band was established by competing it out with the peptide against which the antibody was raised. No PACTΔ3 protein was detectable in the brain extracts of either mouse, although it was readily detected in the transfected cells. The observed absence of PACT-related protein expression in Pact^−/− mice prompted us to investigate whether the corresponding mRNA was expressed. For mRNA analyses, we used a combination of RT-PCR, 3′ RACE, and cDNA sequencing. When the structures of the designed truncated mRNA and the actual mRNA from the disrupted allele were compared (Fig. 2B), it was apparent that the additional polyadenylylation signal of the targeting vector (at base 934) was not used for generating the mRNA; instead, the polyadenylylation signal of the Neo cassette was used. Moreover, intron 7 was retained in this aberrant PACT mRNA that was ≈5.5 kb long, instead of the expected 1-kb length. Thus, this PACT/neo hybrid mRNA contained 1–922 bases of authentic PACT mRNA, intron 7, 923–934 of the disrupted PACT exon 8, and all of the Neo cassette. This mRNA contained multiple translation termination signals in all three reading frames in the 5′ end of intron 7. The PACT ORF is terminated at a TGA codon right in the beginning of intron 7, and hence the encoded protein, if expressed, would have been almost identical to the one designed. Although we did not investigate the reason for the lack of translation of the aberrant mRNA, enhanced degradation, inhibition of cytoplasmic export, or inhibition of translation initiation due to the presence of the unnatural 3′ sequences could all have contributed to this effect. Regardless of the mechanism, because no PACT-related protein was expressed in the PACT^−/− mouse, a true null strain of mouse line had been established.

Pact^−/− MEF lines were generated and used for examining PACT's role in PKR activation in cells infected with vesicular stomatitis virus (Fig. 3). In wild-type MEF, virus infection caused increasing eIF-2α phosphorylation. Because this is mediated by activated PKR, it did not occur in Pkr^−/− cells; however, Pact^−/− cells behaved like wild-type cells. These results suggest that dsRNA, not PACT, is the activator of PKR in vesicular stomatitis virus-infected MEF.

When Pact^+/− male and female mice were mated, Pact^−/− mice were born at the expected frequency, indicating normal Mendelian inheritance of the disrupted allele. The Pact^−/− mice, both of the original mixed genetic background and pure C57BL/6 background, were smaller in size than wild-type mice. The difference in size was less noticeable in the first two weeks after birth, but after 4–6 weeks of age, the Pact^−/− mice weighed ≈40% less than the wild-type mice, whereas Pact^+/− mice weighed the same as wild-type mice. This difference in weight did not change over the whole life span of the Pact^−/− mice. Despite the smaller size, anatomy of the Pact^−/− mice is relatively normal except for the craniofacial areas: the rostrum is rounded, the nose is shortened, the turbinates are hypoplastic, and most remarkably, there is severe reduction of the size of the outer ear (Fig. 4A). Microscopic examination of the ear tissues revealed smaller pinna and external auditory canals and malformed ossicles, as well as a very small middle ear space and bulla (Fig. 4B). In contrast to the defects of the outer and middle ear, the cochlea of the Pact^−/− mice is normal. The noted microtia of the Pact^−/− mice was accompanied by hearing defects. When auditory brainstem responses were performed, hearing thresholds were elevated by 30–45 dB across all frequencies (4–60 kHz) in experimental mice as compared to the Pact^+/+ and Pact^+/− littermates (Fig. 5). This threshold shift did not change with age up to mice of 6 months. Pact^−/− mice are clearly hearing impaired as a result of the congenital anomaly of the ear.

The above observations suggested that PACT might be expressed in the ear tissues of wild-type mice and required for their normal development. To examine PACT expression in different regions of the ears of 3-month-old adult C57BL/6 mice, the middle ear, the cochlea, and the pinna were dissected, and their extracts were analyzed for the presence of PACT protein by Western blotting. PACT was expressed in all three parts of the ear (Fig. 6A). To further localize the sites of PACT expression in the ear, we used in situ hybridization for detecting PACT mRNA expression in middle and inner ear sections. PACT mRNA was clearly detectable in the middle ear and the cochlea. There were strong signals with the antisense probe and no signals with the sense probe (Fig. 6B). In the middle ear, PACT mRNA was detected in the middle ear mucosa throughout the bulla in the respiratory epithelium. In the cochlea, PACT mRNA was observed in the marginal cell layer of the stria vascularis (Fig. 6B) and in the spiral prominence (data not shown). PACT mRNA was also expressed in the supporting cell layer of the sensory epithelium (Claudius and Hensen cells), but it was not expressed in the spiral limbus or the spiral ganglion (not shown).

In situ hybridization for PACT mRNA was performed in mouse embryos to determine the timing and location of PACT expression during development. PACT mRNA was present in the developing ear at embryonic day 12 (E12) as seen in the whole-mount embryo (Fig. 6C). Although the signal was weak, it was specific; no signal was obtained with the sense probe (data not shown). Strong signals were obtained in the mandible and anterior skull base at E16 in parasagittal sections (Fig. 6D). There were no signals when the sense probe was used (data not shown). Both the mandible and external ear are derived from the first branchial arch and PACT mRNA was expressed in both. These results demonstrated that PACT is expressed early in mouse development in structures derived from the first branchial arch (external ear and mandible) and the anterior skull base, and it is clearly important for normal development of the ear, mandible, and craniofacial skeleton. Additionally, PACT expression in the anterior skull base is likely responsible for normal development of the nose, as the PACT knockout mouse demonstrates a foreshortened nose and hypoplastic nasal turbinates.

Discussion

Genes encoding several dsRNA-binding proteins have been ablated in mice, and every ablation has caused a major change in their phenotypes. Mice lacking spermatid perinuclear RNA-binding protein show a high rate of mortality, reduced weight, and defects in spermatogenesis (15). Those lacking Prbp, encoded by Tarbp2, are sterile because of a failure to synthesize protamine (16). Disruption of the gene encoding RNA helicase A led to early embryonic lethality because of apoptotic death of embryonic ectodemal cells (17). The RNA-editing enzyme ADAR1 was shown to be required for embryonic erythropoiesis, and disruption of the corresponding gene caused embryonic lethality (13). On the other hand, the ADAR2 mice were viable, but prone to seizure and early death (18). These reports clearly indicated the critical importance of members of this family of proteins in mouse development.

PACT was not required for PKR activation and the resultant eIF-2α phosphorylation caused by vesicular stomatitis virus infection of MEF, indicating that, in the context of the specific cell type and the virus, dsRNA, not PACT, is the activator of PKR. It remains to be seen whether the same is true for other viruses and cells of other lineages. The observations presented here on Pact^−/− mice demonstrate a critical function of PACT in ear development. Because Pkr^−/− mice (12) do not share any of the defects of Pact^−/− mice reported here, it is fair to assume that the corresponding functions of PACT are not mediated by PKR activation. Because PACT is a strong RNA-binding protein as well, we postulate that the missing functions are mediated by that property of PACT. This idea is strongly supported by what is known about several other dsRNA-binding proteins, most notably Staufen. In Drosophila, Staufen functions as an mRNA chaperon and transports and regulates the translation respectively of bicoid mRNA and Oskar mRNA at the anterior and the posterior poles of oocytes (9, 19, 20). Consequently, disruption of the Staufen gene is lethal to Drosophila development (11). In mammals, the protein is ubiquitously expressed and has been implicated in mRNA transport and translational regulation in hippocampal neurons (21, 22). It also interacts with telomerase RNA affecting telomerase functions (23, 24). Staufen promotes decay of specific mRNAs as well (25), thus down-regulating their cellular expression. It remains to be seen whether PACT functions in the above fashion, but the observed expression of PACT in different parts of the ear, which are developmentally affected in Pact^−/− mouse, is consistent with this notion. This effect is obviously tissue specific and developmental lineage specific, because, although PACT was expressed in all three parts of the ear, the structure of the cochlea was not affected by the absence of PACT, whereas those of the pinna and the middle ear were profoundly affected. It should be noted that the outer and middle ears have a different embryologic origin than the inner ear (26). The Pact^−/− mice must have additional developmental anomalies that cause the observed dwarfism. However, despite both of these defects, the mice have normal life span.

The Pact^−/− mice may represent the first animal model for human microtia, a congenital otologic malformation of unknown etiology. People with microtia have small malformed pinnae, narrow or atretic ear canals, malformed ossicles, and associated conductive hearing loss (27), all of which are present in the Pact^−/− mouse. Because ear development is intimately connected to craniofacial development, microtia is often accompanied by craniofacial abnormalities, such as in Treacher Collins syndrome or Goldenhar's syndrome (28). In the same vein, Pact^−/− mice displays flattening of the midface, shortening of the nose, and hypoplastic nasal turbinates. In humans, microtia is relatively common (1 in 6,000 births). Although many cases are sporadic, others are clearly inherited, but the pattern of inheritance remains unclear (29). Our study suggests that some patients with microtia may have defects in PACT expression, a hypothesis that can be tested experimentally in the future.

Materials and Methods