Otitis media (OM), inflammation of the middle ear, is the most common cause of hearing impairment in children, potentially causing language delays and learning and behavioural disruption 12. A significant number of children with acute OM will go on to develop OM with effusion or chronic OM. The high prevalence of the disease, coupled with its recurrent and chronic nature, accounts for the large number of tympanostomies, the insertion of ventilation tubes or 'grommets' in the tympanic membrane, undertaken in affected children. OM is still the most common cause of surgery in children in the developed world. However, this and other treatments are largely ineffective.

There is evidence from studies of the human population that there is a significant genetic component predisposing to recurrent or chronic OM 345, yet little is known about the underlying genetic pathways involved. From a deafness screen as part of the MRC Harwell mouse mutagenesis programme 6 we have identified two novel dominant mutants, Jeff and Junbo, which develop a conductive deafness due to a chronic suppurative OM 789. Both these mutants represent the first models for chronic forms of middle ear inflammatory disease in humans, and both of these mutants have now been cloned 89.

The gene underlying the Jeff mutant was identified as Fbxo11, a member of the F-box family 8. The Jeff mutant carries a non-conservative glutamine to leucine change at amino acid 491. F-box proteins function as part of an SCF (SKP1-cullin-F-box) E3 protein ligase complex, recognizing and binding phosphorylated proteins and promoting their ubiquitination and degradation 1011. However, the substrate of Fbxo11 is unknown. It has been demonstrated that Caenorhabditis elegans DRE-1, an orthologue of human FBXO11, and the SKP1-like homologue SKR-1 function as part of an E3 ligase complex, as does its human counterpart 12. There is also evidence that FBXO11 has arginine methyltransferase activity, catalyzing arginine methylation, but with a structure different from all other known protein arginine methyltransferases (PRMTs) 13. PRMT activity was not however detected for DRE-1 12. Recently it has been demonstrated that FBXO11 can function as a Nedd8-ligase for the tumour suppressor protein p53, promoting the neddylation of p53 and inhibiting its transcriptional activity 14. p53 is a partner of Smad2 in the activation of multiple transforming growth factor β (TGF-β) target genes 15.

We previously reported that mice homozygous for the Jeff mutation die within a few hours of birth 8. Newborn Jeff homozygotes have cleft palate, facial clefting, impairment of respiratory function and an eyes-open at birth (EOB) phenotype 8. TGF-β signalling has been shown to be involved in all of these processes 161718. For these reasons it will be important to understand the role of Fbxo11 in mouse developmental processes and in particular the impact of mutations on the TGF-β signalling pathway.

The TGF-β superfamily is composed of a large number of cytokines involved in a variety of cellular processes such as proliferation, differentiation, epithelial mesenchymal transformation and apoptosis 1920. They mediate their effects from membrane to nucleus through combinations of type I and type II serine/threonine kinase receptors (TGFβR-I and TGFβR-II) and their downstream effectors, Smad proteins. Certain receptor-regulated Smads (R-Smads) become phosphorylated by activated type I receptors and form a heteromeric complex with a common-partner Smad4. Once formed, this R-Smad/Smad4 complex translocates to the nucleus and, in conjunction with other nuclear cofactors, regulates the transcription of target genes 19. Two different Smad signalling branches have been described. The TGF-β sub-family ligands TGF-β, Activins and Nodals, are transduced by Smad2 and Smad3. In contrast, BMP sub-family ligands are transduced by Smad1, Smad5 and Smad8 19.

TGF-β signalling is controlled by many mechanisms, including ubiquitin-mediated proteosomal degradation 192122. A key component of ubiquitination, the ubiquitin ligase (E3), controls the specificity and timing of Smad ubiquitination. The E3 ubiquitin ligases, Smad ubiquitination-related factor 1 and 2 (Smurf1 and 2), have been identified as regulators of TGF-β signalling targeting Smad1 for ubiquitination and degradation 2324. At higher expression levels, Smurf2 also lowers protein levels of Smad2, but not Smad3 24. In addition to regulation of steady-state levels of R-Smads, the ubiquitin-proteosome pathway is also involved in the degradation of activated R-Smads 2122. Smurf2 exhibits higher binding affinity to activated Smad2 upon TGF-β stimulation, and Smurf2 is a candidate E3 ligase for activated Smad2 degradation 25. Roc1, a component of an SCF complex, interacts and promotes degradation of activated Smad3 26. MFB-1, a novel F-box-type ubiquitin ligase, negatively regulates Dauer formation in C. elegans by modulating DAF-7/TGF-β-like signalling pathway 27. More recently, Arkadia, a RING-domain E3 ubiquitin ligase, has been shown to interact with and ubiquitinate phospho-Smad2/3 28. At the same time, Arkadia enhances transcription, thus coupling turnover of phospho-Smad2/3 to activity.

We have set out to explore the role of Fbxo11 in development and to relate the Jeff homozygous mutant phenotype to the underlying mechanisms of Fbxo11 function. Given the findings of a palatal and EOB phenotype in Jeff mice and the role of TGF-β signalling in these processes, we have focused our studies on salient members of the TGF-β family signalling pathway and demonstrated that pSmad2 is upregulated in the epithelia of Jeff homozygotes. Moreover, we have utilized compound mutants to assess genetic interactions and throw light on the genetic pathways affected. Mice heterozygous for both Jeff and Smad2 mutations recapitulate the Jeff homozygote phenotype in the palate and lungs. However, we failed to detect any interaction between Fbxo11 and Smad2, suggesting they are indirect partners in the development of these tissues. Our observations support a model whereby in palate, eyelid and lung Fbxo11-dependent modification is required to limit the accumulation of pSmad2 in the nucleus of epithelial cells of palatal shelves, eyelids and airways of the lungs. Overall, we conclude that Fbxo11 is involved with the regulation of TGF-β signalling, a finding that has implications for the chronic middle ear inflammatory phenotype that is observed in Jeff heterozygotes.

Jeff homozygote mutant mice develop a variety of epithelial developmental abnormalities, including palatal and EOB phenotypes. We used immunohistochemistry (IHC) to study protein expression and localization of members of the TGF-β signalling pathway (TGFβ-3, TGFβR-I, Smad2, Smad3 and Smad4) in Jeff mutants in both palatal and eye tissues at relevant developmental stages. Given the apparent involvement of Fbxo11 in epithelial development, we also investigated the expression of these TGF-β signalling pathway members in lung development.

We previously identified a mutation in Fbxo11 in the Jeff mutant mice 8. Jeff heterozygote mice develop chronic OM 7 and Jeff homozygotes demonstrate perinatal lethality, cleft palate and an EOB phenotype 8. We investigated the impact of the Jeff mutation on the TGF-β signalling pathway, paying particular attention to those members of the pathway that have a role in epithelial development.

In mammalian development the formation of the palate is a multi-step process, involving palatal shelf growth and elevation above the tongue, followed by fusion of the shelves and the disappearance of the MEE 31. Defects in any of these steps can lead to cleft palate, one of the most common birth defects in humans. Murine palatogenesis takes place between E11.5 and E15.5 31. All TGF-β ligands are known to be expressed at that stage in the developing palate 32. Amongst them, TGF-β3 is likely to be the most important in palatogenesis since mutation of the gene causes cleft palate in mice 2933 and humans 34. In palate cultures from these mutant embryos, the fusion defect is rescued by endogenous TGF-β3 35. It has been demonstrated that TGF-β3 selectively regulates the disappearance of MEE during palatal fusion: TGF-β3 disintegrates the MEE basement membrane 36. Studies on the mechanism for MEE disappearance during palatal fusion suggests a role of TGF-β3 as an inducer of apoptosis – cell death in TGF-β3 null palates is reduced at the time of fusion 37. Smad2 and Smad3, mediators of TGF-β signalling, are expressed in the MEE cells, but only Smad2 is phosphorylated during palatal fusion. Smad2 phosphorylation is temporo-spatially restricted to the MEE and correlates with the disappearance of the MEE 30. TGF-β3 is required for phosphorylation of Smad2 in the MEE and the inhibition of MEE proliferation during palatal fusion 30.

Our studies using IHC to co-localize the TGF-β3 ligand, TGF-β receptors and Smads in developing palates revealed one major difference between wild-type and Jf/Jf mice. At E15.5 more epithelia cells are positive for pSmad2 in the homozygote palatal shelves than in the wild type. Moreover, we found that there was a substantial increase in nuclear localization of pSmad2 in Jf/Jf mice. This might suggest that the turnover of pSmad2 is stalled and pSmad2 accumulates in the nucleus, with consequent effects for TGF-β signalling in the epithelia.

TGF-β3 plays a role as an inhibitor of MEE proliferation during palatal fusion 30 and an inducer of apoptosis 37. Using Ki67 as a marker for proliferation we did not observe increased proliferation in E14.5 and E15.5 palates of the Jeff homozygote MEE. However, we also stained for apoptotic cells using a caspase-3 antibody and found many fewer positive cells in Jeff homozygote E15.5 palates. It appears that in the presence of a mutation in Fbxo11, palatal fusion is inhibited and MEE cells are not able to progress to epithelial programmed cell death or epithelial-mesenchymal transformation.

The palatal shelves fusion is a permanent fusion, such as the fusion of the neural tube. During mammalian development some temporary fusions also occur for example, eyelid fusion, fusion of the digits and fusion of the pinnae of the ears to the scalp. The disjunction of the temporary fusions takes place after birth 38. The eyelids develop at approximately the same time as the palatal shelves. They start to form at about E11.5, grow across the eye from E14 to E16 and, as the fusion progresses, the diminishing gap fills with a profusion of rounded cells that are extruded, flattened, and sloughed off from the area of completed fusion 3839. Failure of the eyelids to grow and fuse in mice leads to the EOB defect. Mouse knockout studies have identified several signalling molecules involved in the control of embryonic eyelid closure and some of them are from the TGF-β family. Mutation in activin βB results in an EOB phenotype 40 similar to that observed in MEKK1-deficient mice 41. TGF-β/activin signalling via the MEKK1-mediated JNK pathway is one pathway controlling eyelid closure 1742. MEKK1 is not required however for the classical TGF-β/activin pathway, which involves nuclear translocation of Smad proteins 17.

Normally in mouse development, the eyelid fusion is complete by E16, but the eyelids of Jeff homozygote mice remain wide apart. The eyelids start to form the leading edge of the developing margins, but the leading edge is not extended and the eyelids do not reach each other. We applied the same panel of antibodies, employed on the palates, on sections of E16 eyelids. The expression pattern of TGF-β3 and the signalling receptor TGFβR-I are consistent with localization observed in the cytoplasm of the cells of the epidermis of both the developing wild-type and mutant eyelids. The distribution of non-activated Smads is also similar, but with pSmad2 there is a clear difference between the wild-type and homozygote eyelids. Activated Smad2 was present as a nuclear stain in the majority of the cells of the epidermis, the basal layer and also some cells from the dermis of the Jeff homozygote eyelids. In contrast, in wild-type eyelids the pSmad2 is localized in the cytoplasm of the epidermis. The difference is also time specific. At E15.5 before eyelid fusion the pattern and the localization of activated Smad2 is similar in wild type and Jf/Jf. These findings mirror and are consistent with the expression changes observed in the palate.

In the developing eyelid, as with the palate, we also studied proliferation and apoptosis in the Jeff homozygote. Using Ki67 as a marker for proliferation, we found that there was little difference between wild-type and mutant homozygotes in the number or distribution of cells stained. However, we found significantly lower staining with caspase-3 in the Jeff homozygote, indicating reduced rates of apoptosis. It would appear that, as with the palate, the absence of epithelial fusion prevents the transition to epithelial programmed cell death.

We show here that, in addition to palatal and eyelid development defects, the development of the lung in Jeff homozygotes is severely compromised. In the mouse, early lung development begins at about E9.5 with the formation of the paired lung buds and continues through the pseudoglandular phase at E9.5-16.5 when the bronchial and respiratory tree develops, and the canalicular phase E16.5-17.5 when vascularization occurs. During late lung development at E17.5-P4 the distal airways form saccular units (saccular stage) and the secondary septae divides these units during the alveolar stage at P5-28 4344.

Several studies have implicated TGF-β signalling in early lung development. Immunohistochemical studies on localization of the three TGF-β isoforms in the developing mouse embryo suggest that they play an important role in lung formation 18. TGF-β signalling is implicated in the negative regulation of lung growth and development during early lung organogenesis 45. Recent expression data also indicates that active TGF-β signalling is required for normal late lung development 46. Gene-targeting studies also demonstrate the involvement of the TGF-β family in early and late lung morphogenesis. TGF-β3 null newborn mice have under-developed and poorly inflated lungs 33. The developmental delay is detectable as early as day E12.5, which implicates a role of TGF-β3 in early lung development. Retarded lung alveolarization and subsequent emphysema in Smad3 knockout mice, on the other hand, suggests a role for the TGF-β/Smad3 pathway in postnatal lung growth and emphysema prevention 47. As the TGF-β family is involved in lung morphogenesis and reflecting the changes in pSmad2 expression and localization in the palate and eyelid, we have explored pSmad2 in the developing lung. Again, we find a significant increase in the expression of pSmad2. Most notably, there is a very marked increase in localization of pSmad2 to the nuclei of epithelial cells lining the airways.

Given the effects of the Fbxo11 mutation on pSmad2 in a variety of tissues, we decided to examine genetically the interaction between these two genes and generated Jf/+ Smad2/+ compound heterozygotes. Intriguingly, we found that Jf/+ Smad2/+ mice recapitulated many of the features of Jeff homozygotes, including the cleft palate, lung phenotype and post-natal mortality. The phenotypes of the Jf/+ Smad2/+ compound heterozygote are not fully penetrant. Around one-third of Jf/+ Smad2/+ mice survive but are much smaller than wild-type or single-mutant littermates. The developmental abnormalities observed in the Jf/+ Smad2/+ mice contrasts with the normal phenotype seen in Jf/+ mice and underlines the effects of the Fbxo11 mutation on pSmad2. It would appear that a complete absence of Fbxo11 function, with consequent effects on pSmad2 expression and localization, can lead to the epithelial developmental defects that we have observed. Alternatively, compromising levels of gene expression at both the Fbxo11 and Smad2 loci leads to the same phenotype.

Despite the effects of the Fbxo11 mutation, and the genetic interaction observed between Fbxo11 and Smad2, we failed to observe any biochemical interaction between Fbxo11 and Smad2. As expected we were able to demonstrate the known interaction between Smad2 and p53. It has been reported that Fbxo11 neddylates p53, inhibiting its transcriptional activity 14. However, immunoprecipitations did not reveal an interaction in the tissues we tested. Indeed, Jf/+ p53/+ compound heterozygotes did not show any developmental phenotypes. Nevertheless, we found that Jeff homozygotes have markedly reduced levels of p53 and heterozygotes appeared to have intermediate levels. It appears that Fbxo11 is required for stabilisation of p53.

p53 is required for TGF-β responses through its interactions with Smads 15. p53-deficient mammalian cells demonstrate impaired responses to TGF-β signals. Most importantly, p53 and Smad2 cooperated synergistically at target promoters for TGF-β signalling, and indeed they directly interact in vivo in a TGF-β-dependent fashion 15. We propose that in the Jeff homozygous mutant, loss of Fbxo11 function leads to destabilisation of p53 by an unknown mechanism. It is possible, given the synergistic interaction between p53 and Smad2, that the loss of p53 at transcriptional targets leads to stalling and accumulation of pSmad2 in the nuclei of epithelial tissues of Jeff homozygotes. Mutant p53 leads to attenuation of TGF-β1 signalling including a reduction in Smad2/3 phosphorylation, inhibition of Smad2/Smad4 complex formation and Smad4 translocation to the nucleus 48. Overall, given the interaction between p53 and TGF-β signalling pathways, it seems likely that the effects of the Fbxo11 mutation on TGF-β signalling may be mediated through effects on p53.

Chronic and recurrent OM in the human population is known to have a very significant genetic component, yet little is known about the underlying genes or pathways involved. Moreover, although a number of mouse strains are available that show OM, they are complicated by poor penetrance or the complex syndromic nature of the disease. Jeff heterozygotes develop a highly penetrant chronic suppurative OM in the absence of any other significant pathology 78 and thus the Jeff mutant represents a powerful model for studying the genetic and pathophysiological bases of chronic/recurrent OM. The discovery of the Fbxo11 gene underlying the Jeff mutant identified an important candidate for the study of OM in the human population. Indeed, initial studies with FBXO11 SNPs in human OM families have demonstrated nominal evidence of association, indicating the genetic involvement of human FBXO11 with chronic and recurrent OM 49.

The studies reported here support a role for Fbxo11 in TGF-β signalling and suggest that perturbations in this pathway may underlie chronic inflammation in the middle ear. This is particularly pertinent given the studies of a similar mouse mutant with chronic OM, Junbo 9, caused by a mutation in the transcription factor Evi1 which is a co-repressor of Smad3. Overall, there is increasing evidence that defects in TGF-β signalling or associated pathways may underlie the development of chronic OM, and genes in these pathways represent good candidates for future human association studies.

The authors declare that they have no competing interests.

HT carried out the immunostaining analysis, some of the proteomics analysis, contributed to the design of the study and the interpretation of the results and participated in drafting the manuscript. RH contributed to the analysis and discussion of the results. SM participated in the phenotyping and the genotyping of some of the mice, and MRR and HH performed most of the immunoprecipitations and western blot analysis and contributed to the interpretation of the results from these studies. CD helped with the study of the lung phenotype of the mice and contributed to the interpretations of the results. SDMB contributed to the design of the study, interpretation of results and participated in drafting the manuscript. All authors read and approved the final manuscript.