In this study, we have for the first time successfully over-expressed Grhl3 in adult mice, allowing us to uncover previously unsuspected developmental events that are dependent on stringent regulation of GRHL3. We achieved this using a Grhl3-null allele (Grhl3Cre) to induce expression of a Grhl3 transgene inserted into the murine Rosa26 locus. Mice homozygous for the null allele (Grhl3Cre/Cre) and lacking transgene expression exhibited the classic defects described in the original Grhl3 constitutive knockout8, of neural tube defects, short longitudinal body axis, failure of the skin barrier to form and peri-natal lethality. These defects were completely rescued in mice carrying two copies of the Rosa26-based transgene, with the animals surviving to adulthood. However, multiple novel phenotypic abnormalities were observed as a consequence of dysregulated Grhl3 expression, with digit defects, impaired epidermal homeostasis with alopecia, the Shaker-Waltzer phenotype, hearing impairment, and inner ear malformations all evident.
De Castro, et al. 20 reported the first successful attempt of over-expressing Grhl3 in late-stage mouse embryos, utilising the hypomorphic curly tail strain to generate the line9. Grhl3 expression was driven by a bacterial artificial chromosome (BAC), encompassing the endogenous Grhl3 locus. A ~ 1.5 to 2-fold increase in total Grhl3 mRNA expression above WT levels was documented in the caudal region of E9.5 and E10.5 embryos and a less then 1.2-fold increase in the epidermis of E18.5 embryos that were double homozygotes for Grhl3ct allele and the BAC transgene (Grhl3ct/ct;TgGrhl3/TgGrhl3). Interestingly, 67% of these embryos displayed NTDs, a phenotype we never observed in our transgenic embryos, despite comparable levels of Grhl3 over-expression. None of the De Castro, et al. 20 embryos displayed digit defects or abnormal epidermal morphology, and phenotypes in adult mice were not reported. The phenotypic discrepancies between the two lines, particularly the NTDs, cannot be attributed to differences in the initial timing, location or magnitude of expression of Grhl3. Both lines restricted expression to tissues that usually express Grhl3 at the appropriate developmental time-point; De Castro, et al. 20 through use of the endogenous locus in a BAC, and our line via Cre expression from the endogenous locus activating the transgene. However, several other differences may underlie the phenotypic discrepancies. Firstly, Grhl3 once over-expressed in our line would remain so, as the Rosa26 locus lacks any endogenous regulatory elements that are retained in the BAC, and may influence Grhl3 expression. Secondly, and in our view more likely, Grhl3 mRNA and protein is derived from a cDNA construct in our transgene, whereas it is derived from the endogenous locus in the BAC. Several studies have highlighted the importance of intronic RNA for GRHL3 regulation31–33, particularly in the context of epidermal differentiation31. Lack of intronic elements in our transgenic embryos may lead to further dysregulation of Grhl3 expression. This implies that correct regulation of Grhl3 expression is crucial for embryonic development of multiple systems. Similarly, over-expression of Grhl3 has previously been shown to induce different developmental consequences in different models20,21.
A surprising finding in our model, was the under-representation of embryos homozygous for the transgene (irrespective of the presence or absence of a Cre allele), suggesting that “leaky” expression of the transgene in early embryogenesis had dire developmental consequences. This finding is in keeping with the previous reports on the ActbCre;CAGLSL−Grhl3 mouse line, which displayed early lethality at E5.522. We postulate that this may result from random transmission of Cre recombinase from the sperm to the oocyte where the unexpected deletion of the upstream loxP-STOP-loxP sequence of the transgene occurred even if the consequential embryos are Cre ‘negative’ genetically34, or the inheritance of delta allele (activated transgene allele) by the offspring due to germline recombination occurred during gametogenesis in male Grhl3Cre/+;Rosa26Tg/+ mice35, leading to certain level of Grhl3 expression that is incompatible with early embryonic survival. Therefore, surviving homozygous transgenic mice presumably had faithful transgene expression, that commenced only after Grhl3-Cre was activated at E8.5. This is consistent with our PCR and Q-RT-PCR analyses showing no deletion of the loxP-STOP-loxP cassette, or Grhl3 transgene expression, in E18.5 Grhl3+/+;Rosa26Tg/Tg embryos.
In Grhl3Cre/Cre;Rosa26Tg/Tg embryos, transgene-mediated expression of Grhl3 fully rescued the epidermal barrier defect, and largely normalised keratinocyte terminal differentiation in both embryos and adult mice. Surprisingly, it did not correct basal cell hyperproliferation and expansion. Previous studies have shown that diverse mechanisms centred on reducing Grhl3 RNA levels are critical for maintaining low levels of GRHL3 protein expression in epidermal stem cells, thereby preventing differentiation31,33. Consistent with this, constitutive Grhl3-knockout mice display failed epidermal differentiation associated with expansion of a proliferative stem cell pool10,12,14. Therefore, the persistence of a hyperproliferative and expanded basal layer, despite normal differentiation shown here, would suggest that it is not simply the level of Grhl3 mRNA that is important for epidermal stem cell behaviour. In keeping with this, hyperproliferation in the Grhl3Cre/Cre;Rosa26Tg/+ epidermis was far greater than in Grhl3Cre/+;Rosa26Tg/+ epidermis, despite both lines having comparable total Grhl3 mRNA levels. Similarly, Grhl3Cre/+;Rosa26Tg/Tg mice exhibited less basal cell proliferation than Grhl3Cre/Cre;Rosa26Tg/Tg mice despite slightly higher levels of total Grhl3 expression. These findings suggest that either the endogenous transcript is important for stem cell homeostasis, or that higher levels of transgene expression are deleterious to stem cell behaviour.
Another unexpected skin phenotype, severe alopecia, was also observed as a consequence of Grhl3 over-expression. Our previous studies in adult conditional Grhl3-knockout mice revealed no alopecia27, whereas both Grhl3Cre/Cre;Rosa26Tg/Tg and Grhl3Cre/+;Rosa26Tg/Tg mice displayed dorsal hair loss that mimicked the grooming alopecia phenotype we had previously observed in constitutive Grhl1-knockout mice28. In that line, alopecia was due to poor hair anchorage as a result of detachment of the IRS from the ORS, due to loss of expression of the GRHL1 target gene, Dsg1a. Neither Grhl1 nor Dsg1a levels were altered in the Grhl3 over-expressing lines compared to WT, suggesting that the hair anchorage phenotype may be due to perturbed expression of other Desmoglein genes36,37. As this phenotype was more severe in mice with higher levels of transgene expression, and given the complete absence of a comparable phenotype in adult Grhl3 conditional knockout mice27, it is unlikely that the presence of the endogenous transcript exerts any influence on hair anchorage.
Although GRHL1 deficiency causes inner ear malformation in zebrafish, and GRHL2 has been identified as an autosomal-dominant deafness gene in humans, Grhl3 has not previously been linked to inner ear structural defects or hearing impairment38–40. Our previous research had identified a key role for Grhl3 with Vangl2 in planar cell polarity mediated orientation of cochlear hair cell stereocilia5. Here, we found that misexpression of Grhl3 had no impact on cochlear hair cell orientation, but did result in severe bony labyrinth dysmorphogenesis. Normally, the semi-circular canals provide sensory input for rotary movement, with the end of each canal extending into an ampulla in which hair cells excite the afferent fibre of the vestibular nerve upon head and body movement41,42. The lateral and superior canals detect the movement in vertical axis and lateral axis respectively, and structural defects result in head tilting in mice. The posterior canal detects movement in left-right axis and defects result in circling behaviour in mice41,42. The transgenic mice displayed vestibular disturbance, and a Shaker-Waltzer phenotype that mirrored the degree of severity of the ampulla and semi-circular canal malformation. The ABR test also revealed a positive correlation between hearing impairment and the severity of inner ear structural defects. Interestingly, severity was again linked to the levels of expression of the transgene rather than total Grhl3 mRNA levels, with Grhl3Cre/Cre;Rosa26Tg/+ mice more severely afflicted than Grhl3Cre/+;Rosa26Tg/+ animals. Given that the Grhl3Cre/Cre;Rosa26+/+ inner ear showed normal morphological appearance, it appears that the levels of expression of the transgene rather than loss of the endogenous transcript is detrimental to inner ear development.
A recent study identified that intron 1 of the nascent human GRHL3 RNA provides binding sites for the Cleavage and Polyadenylation Specificity Factor (CPSF) complex and HNRNPA3. This suppresses exonic splicing and promotes intronic polyadenylation, lowering GRHL3 expression and preventing the premature differentiation of primary human keratinocytes31. In humans, four alternatively spliced isoforms of GRHL3 have been characterised2,17,43. Although no murine Grhl3 splice isoforms have been discovered to date, it is likely that Grhl3 alternative splicing is conserved in mouse to generate transcript variants44. Early in embryonic development, the precursor of the inner ear, the otic placode, is derived from the pre-placodal region (PPR), located in the territory of the neural plate border7,45. Notably, expression of the PPR marker, Six1, overlaps with Grhl3 spatiotemporally in mouse embryos6,46. We propose that the cDNA-based Grhl3 transgene may induce abnormal differentiation of the PPR and consequently, inner ear defects and hearing impairment. This may indicate that a specific Grhl3 mRNA isoform allows this process to occur, and expression of the incorrect splice isoform from the transgene perturbs inner ear development.
In conclusion, our study has highlighted that stringent regulation of Grhl3 expression is an absolute requirement for numerous developmental processes. In particular, epidermal differentiation and hair anchorage, digit formation, and formation of the cochlea and vestibular apparatus are affected. In addition, it has raised the concept that not only are the levels of the transcription factor important, but also that isoform-specific roles may govern different morphogenetic events.