Pan-otic CRE drivers enable gene regulation throughout the otic placode lineage, comprising the inner ear epithelium and neurons. However, intersection of extra-otic gene-of-interest expression with the CRE lineage can compromise viability and impede auditory analyses. Furthermore, extant pan-otic CREs recombine in auditory and vestibular brain nuclei, making it difficult to ascribe resulting phenotypes solely to the inner ear. We have previously identified Slc26a9 as an otic placode-specific target of the FGFR2b ligands FGF3 and FGF10. We show here that Slc26a9 is otic specific through E10.5, but is not required for hearing. We targeted P2ACre to the Slc26a9 stop codon, generating Slc26a9P2ACre mice, and observed CRE activity throughout the otic epithelium and neurons, with little activity evident in the brain. Notably, recombination was detected in many FGFR2b ligand-dependent epithelia. We generated Fgf10 and Fgf8 conditional mutants, and activated an FGFR2b ligand trap from E17.5 to P3. In contrast to analogous mice generated with other pan-otic CREs, these were viable. Auditory thresholds were elevated in mutants, and correlated with cochlear epithelial cell losses. Thus, Slc26a9P2ACre provides a useful complement to existing pan-otic CRE drivers, particularly for postnatal analyses.
Tg(Pax2-cre)1Akg, an unmapped BAC transgenic mouse strain expressing CRE recombinase in a pattern that largely recapitulates Pax2 expression, except in optic tissues, is a mainstay of inner ear genetic analyses. Starting at E8.5, it reliably recombines LoxP sites throughout the otic placode lineage, which includes the otic epithelium and ganglion neurons, allowing pan-otic conditional deletion or activation of genes from the earliest stages of inner ear development (Ohyama and Groves, 2004). The utility of this strain for postnatal studies of otic genes, however, is limited when modulation of a given gene of interest in the extra-otic part of the CRE lineage affects viability. For example, we studied embryonic inner ear development following inactivation of Fgf10 in the Tg(Pax2-cre) lineage (Urness et al., 2018), but could not analyze such conditional knockout (CKO) mice postnatally because of P0 lethality, which could be caused by loss of Fgf10 in the developing kidney (Michos et al., 2010), a component of the Tg(Pax2-cre) lineage (Ohyama and Groves, 2004). Similarly, we have been unable to study postnatal Fgf8;Tg(Pax2-cre) CKO mice due to P0 lethality suspected to be caused by the requirement of Fgf8 in the developing mid/hindbrain (Chi et al., 2003), which is another component of the Tg(Pax2-cre) lineage (Ohyama and Groves, 2004), or in the kidney (Perantoni et al., 2005). The targeted Foxg1Cre allele provides another option for recombining conditional alleles throughout the otic placode lineage (Hébert and McConnell, 2000), but it also has an expansive extra-otic lineage that can prevent postnatal analyses. Thus, Fgf8;Foxg1Cre CKOs also die at birth (Jacques et al., 2007; Zelarayan et al., 2007). Furthermore, Foxg1Cre is subject to misexpression in some genetic backgrounds (Hébert and McConnell, 2000) and can have genetic interactions as a heterozygote with other otic genes (Pauley et al., 2006). In addition, even if Tg(Pax2-cre) or Foxg1Cre conditional loss- or gain-of-function mice survive postnatally, it can be difficult to unambiguously ascribe auditory or vestibular phenotypes to the inner ear because both Cre lineages also include mid- and hindbrain derivatives, including the dorsal cochlear nucleus, inferior colliculus and the cerebellum (Hébert and McConnell, 2000; Ohyama and Groves, 2004; Zuccotti et al., 2012). Furthermore, Tg(Pax2-cre) drives CRE expression in pharyngeal arch 1 (Lu et al., 2012; Ohyama and Groves, 2004), which gives rise to middle ear ossicles, and Foxg1 is expressed in pharyngeal endoderm, which gives rise to the eustachian tube and tympanic cavity (Anthwal et al., 2013; Chapman, 2011; Sienknecht, 2013). Thus, it is possible that the auditory effects of modulating gene expression using one of these Cre drivers might have a middle ear origin.
To expand the options for gene manipulation in the otic placode lineage, we sought to target Cre to a gene that is active in the otic placode, but not in the brain, and has extra-otic expression that differs from Pax2 and Foxg1. We noticed that Slc26a9, a gene we identified as an otic placode-expressed target of the FGFR2b ligands FGF3 and FGF10, appeared specific to the otic placode, as assessed by in situ hybridization of whole E8.5 embryos (Urness et al., 2010). Slc26a9 is located on chromosome 1 and encodes a membrane protein expressed in adult stomach and lung epithelia that functions as a chloride ion transporter (Walter et al., 2019). Null mutants are viable and fertile, with defects in gastric acid secretion (achlorhydria) and airway clearance in response to inflammation (Anagnostopoulou et al., 2012; Chang et al., 2009; Xu et al., 2005, 2008). We show here that Slc26a9 is specific for the otic placode, cup and vesicle from E8.5 to E10.5. We used Easi-CRISPR (Quadros et al., 2017) to insert P2ACre into the Slc26a9 stop codon, producing a correctly targeted allele designed to express CRE from the same transcript as SLC26A9. Reporter analyses of embryonic and early postnatal Slc26a9P2ACre lineages showed that recombination in otic tissues starts at E9.5, 1 day later than Tg(Pax2-cre), but nevertheless encompassed the entire otic epithelium and ganglion neurons. There was very little recombination in the brain or middle ear, but perinatal CRE activity was detected as expected in stomach and lung epithelia, as well as in pancreas, kidney and intestinal epithelia. In contrast to Fgf10 or Fgf8;Tg(Pax2-cre) CKOs, Fgf10 or Fgf8;Slc26a9P2ACre CKOs survived postnatally. Auditory brainstem response thresholds were elevated in both CKOs. We also used Slc26a9P2ACre to drive activation of a dominant-negative FGFR2b ligand trap from E17.5 to P3, thereby briefly inhibiting both FGF3 and FGF10 function in the otic lineage, and observed partially penetrant hearing loss correlating with a reduction in the cochlear outer sulcus. These data indicate that Slc26a9P2ACre mice provide a valuable alternative to Tg(Pax2-cre) and Foxg1Cre, particularly for postnatal analyses of the inner ear, and might also be useful for gene regulation in other FGFR2b-dependent epithelia.
Slc26a9 is otic specific from E8.5 to E10.5, but is not required for auditory function
At E8.5, Slc26a9 is an otic placode-specific gene that is dependent upon FGF3 and FGF10 signaling (Urness et al., 2010). To further investigate Slc26a9 expression, we conducted in situ hybridization in whole-mount embryos at additional stages (Fig. 1). As the otic placode matured from the 7- to 8-somite stage (Fig. 1A) through the otic cup (Fig. 1B) and vesicle stages (Fig. 1C,D), Slc26a9 expression became dorsally regionalized within the vesicle by E10.0-E10.5 (Fig. 1E,E′,F,F′), but remained otic specific, unlike either Pax2 or Foxg1, which are also expressed in neural and other tissues from early stages (Hébert and McConnell, 2000; Ohyama and Groves, 2004). Notably, we did not detect expression in the otic ganglion. We conducted in situ hybridization using inner ear sections at later stages (E12.5, E14.5, E17.5 and P3), but did not detect a specific Slc26a9 signal, suggesting that Slc26a9 otic expression is downregulated after E10.5-E11.5. To assess possible requirements for Slc26a9 in the inner ear, we intercrossed mice heterozygous for an Slc26a9 null allele (Xu et al., 2008), observed posture/movements and measured auditory brainstem response (ABR) thresholds, an indicator of hearing, to several stimuli at P21-P49. None of the intercross offspring showed abnormal posture or movements, and there were no significant differences between genotypes with respect to auditory thresholds (Fig. 1G), showing that Slc26a9 is not uniquely required during inner ear development or function.
Targeting P2ACre to Slc26a9
The substantial differences in extra-otic embryonic Slc26a9 expression relative to Pax2 and Foxg1, and the lack of a requirement for Slc26a9 in inner ear development prompted us to target this gene for CRE expression. To avoid disrupting Slc26a9 expression, we targeted an in-frame P2A ‘self-cleaving peptide’-coding sequence (Kim et al., 2011) followed by a codon-optimized Cre sequence (Ohtsuka et al., 2013; Shimshek et al., 2002) into the Slc26a9 stop codon (Fig. 2A-D). In this way, a transcript encoding both SLC26A9 (including 21 extra C-terminal amino acids from P2A) followed by CRE, with an N-terminal proline was expected. We used Easi-CRISPR for gene targeting (Miura et al., 2018; Quadros et al., 2017). In this case, a single-stranded DNA donor, together with a ribonucleoprotein complex composed of CAS9, a tracer RNA and a CRISPR RNA designed to cleave inside the Slc26a9 stop codon, was injected into mouse zygotes. Out of the seven founder pups, four had the expected 5′ and 3′ junctions with Slc26a9 (Fig. 2E) and these had exactly the expected sequence (data not shown). Founders 6 and 7 were used for germline transmission of the Slc26a9P2ACre allele and initial crosses of male heterozygotes to female CRE-dependent LacZ reporter mice (Rosa26lslLacZ) revealed similar patterns of CRE activity in the offspring, so all further characterization was arbitrarily confined to line 7. When a female heterozygote was crossed to a male LacZ reporter mouse, four out of five offspring exhibited ubiquitous βGAL activity, even though none carried the Slc26a9P2ACre allele, suggesting that the germline of the female contained CRE, which caused recombination in most zygotes and was propagated to all tissues, similarly to Tg(Sox2-cre) (Hayashi et al., 2003). Thus, for all lineage and CKO studies reported here, Slc26a9P2ACre was contributed by the male parent. Intercrosses of Slc26a9P2ACre/+ mice generated offspring of all three expected genotypes (Fig. 2F). The allele can be maintained as homozygous, with no obvious fertility or longevity decrement for either sex.
Slc26a9P2ACre directs CRE activity in the otic lineage starting at E9.5
We surveyed Slc26a9-directed CRE activity in sagittal sections of Slc26a9P2ACre/+;Rosa26lsltdTom/+ heads at P1 (Fig. 3). tdTom fluorescence was found throughout both the cochlear and vestibular epithelia of the inner ear (Fig. 3A,B). Magnified views showed that virtually all otic epithelial cells were labeled, with only a few tdTOM-negative gaps (Fig. 3C-F). Cochlear (spiral) and vestibular ganglion neurons were also tdTOM-positive (Fig. 3A,C,C′,J,K) and their central processes, which form the 8th nerve, were evident (Fig. 3A). Labeling of spiral ganglion neurons was assessed by immunostaining for PROX1. The vast majority of PROX1-positive spiral ganglion neurons were also tdTOM-positive (Fig. 3J). Vestibular ganglion neurons were also labeled with tdTOM; however, as PROX1 was expressed at much lower levels than in spiral ganglion neurons, we examined co-labeling with antibodies directed against neurofilament (NFH). Virtually all NFH-positive cell bodies in the vestibular ganglion were also tdTOM-positive (Fig. 3K). Unlike Tg(Pax2-cre), which is active in the derivatives of the mid-hindbrain junction, including the dorsal cochlear nucleus and cerebellum (Ohyama and Groves, 2004; Zuccotti et al., 2012; data not shown), except for the incoming 8th nerve axons, we did not detect the Slc26a9P2ACre lineage in those structures (Fig. 3A,B) or indeed in any brain nuclei (e.g. Fig. 3G). There were, however, scattered tdTOM-positive cells in the brain that could be microglia (Fig. 3A, carets). In addition, the anterior pituitary and eustachian tube were tdTom positive (Fig. 3G). Finally, although a few tdTOM-positive cells were seen in the manubrium of the malleus and in the tympanic membrane (middle ear components, Fig. 3H), these were much less numerous than in sections of P3 Tg(Pax2-cre) heads (Fig. 3I).
To determine when Slc26a9-directed CRE activity was first detectable, we examined Slc26a9P2ACre/+;Rosa26lslLacZ/+ embryos stained with X-gal. Somewhat surprisingly, there was no βGAL activity detected at otic placode or cup stages (Fig. 4A,B), but robust otic-specific activity was present in the otic vesicle and ganglion at E9.5 and E10.5 (Fig. 4C,D). By E11.5, βGAL activity was detected not only in the otic lineage, but also in the lung and kidney, and diffusely in the lower head and neck (Fig. 4E). Notably, βGAL activity was not detected in the brain.
As otic lineage labeling was first detected at the otic vesicle stage, which is after the time when otic neurons begin to delaminate from the anteroventral otic cup epithelium (Carney and Silver, 1983; Li et al., 1978), we asked whether all or only a fraction of otic neuroblasts exhibit CRE activity at E9.5 and E10.5. Slc26a9P2ACre/+;Rosa26lsltdTom/+ embryos were sectioned through the otic region and stained simultaneously with antibodies directed against Neurod1 to detect neuroblasts and E-cadherin to outline the otic epithelium. At E9.5 a few Neurod1-positive cells were detected ventral to the otic epithelium and most appeared to be tdTOM-positive (Fig. 4F). By E10.5, much larger numbers of Neurod1-positive cells had formed a large anteroventral ganglion. The vast majority of these cells were tdTOM-positive (Fig. 4G). Thus, despite the later onset of otic CRE activity in Slc26a9P2ACre mice relative to Tg(Pax2-cre) mice, cochlear and vestibular ganglion neurons are effectively labeled.
Slc26a9P2ACre directs CRE activity in a variety of extra-otic epithelia
To identify additional sites of Slc26a9-directed CRE activity, we detected tdTOM in early postnatal tissues or βGAL activity in embryos from the reporter crosses. As expected from SLC26A9 expression studies (Amlal et al., 2013; Chang et al., 2009; Lohi et al., 2002; Xu et al., 2005), we detected tdTOM-positive cells in sections of early postnatal lung and stomach epithelia, as well as in kidney and pancreas (Fig. 5A-D). Interestingly, the gastric lineage includes only the distal (pyloric) region of the epithelium (Fig. 5B,F). In addition, the kidney lineage appears very restricted (Fig. 5C), potentially comprising only the parietal cells of the medullary collecting ducts where SLC26A9 expression has been documented (Amlal et al., 2013). Each of the above-mentioned CRE-positive lineages was also detectable in whole-mount tissues at E16.5 (Fig. 5E-G; pancreas not shown), and the lung epithelium was labeled as early as E11.5 (Fig. 5E,I). Rathke's pouch, which is the precursor of the anterior pituitary, was strongly labeled at E13.5 and E11.5 (Fig. 5H,L). CRE activity was not assessed in postnatal intestine, but was detected in E13.5 intestinal epithelium (Fig. 5J). Finally, the Slc26a9P2ACre lineage was quite strong in the E13.5 cornea (Fig. 5K). CRE-negative tissues included heart (Fig. 5E), adrenal gland (Fig. 5G) brain (Fig. 5H), spleen and liver (data not shown).
Fgf10;Slc26a9P2ACre conditional mutants are viable, and show vestibular and auditory dysfunction
Fgf10 is required for inner ear morphogenesis, but germline null or conditional mutants generated with Tg(Pax2-cre) do not survive past birth (Min et al., 1998; Sekine et al., 1999; Urness et al., 2018). To assess postnatal Fgf10 inner ear phenotypes, we generated Fgf10 conditional mutants using Slc26a9P2ACre and observed motor behavior and measured ABR thresholds. Fgf10;Slc26a9P2ACre CKOs survived to weaning in somewhat more than the expected proportion (8/21=38%, expected 25%), indicating that the Slc26a9 lineage does not overlap any sites of Fgf10 expression required for viability, e.g. the lung mesenchyme (Bellusci et al., 1997; Min et al., 1998) or metanephric mesenchyme (Michos et al., 2010; Ohuchi et al., 2000). Seven out of eight Fgf10 conditional mutants exhibited mild head tossing or circling behavior and all showed increased ABR thresholds averaging 24-36 decibels above control means for all stimuli. The differences were significant for all comparisons between the CKO and control animals, except at 32 kHz, where variability in two of the control genotypes was high (Fig. 6A).
To evaluate cochlear development in Fgf10;Slc26a9P2ACre CKOs relative to previous studies of Fgf10−/− and Fgf10;Tg(Pax2-cre) CKOs (Urness et al., 2018, 2015), we examined histological cross-sections of CKO and littermate control cochleae at E18.5 (Fig. 6B-G). The cochlear phenotype of Fgf10;Slc26a9P2ACre CKOs was variable. One cochlea completely lacked Reissner's membrane and the outer sulcus (Fig. 6C,C′), and was similar to all of the three Fgf10−/− or four Fgf10;Tg(Pax2-cre) CKO cochleae examined previously (Fig. 6G; Urness et al., 2015, 2018). Another cochlea had a shortened Reissner's membrane and lacked the outer sulcus (Fig. 6D), and a third cochlea had a shortened Reissner's membrane as well as a reduced outer sulcus (Fig. 6E).
To assess vestibular development in Fgf10;Slc26a9P2ACre CKOs, we paintfilled a few inner ears from the conditional cross and categorized phenotypes according to the scheme used previously for Fgf10−/− and Fgf10;Tg(Pax2-cre) CKOs. Relative to controls (Fig. 6H), all three Fgf10;Slc26a9P2ACre CKO inner ears entirely lacked the posterior semicircular canal (PSCC; Fig. 6I). This was the most prevalent phenotype observed in Fgf10;Tg(Pax2-cre) CKOs (8/16; Urness et al., 2018), but in that case we also observed some more severely affected inner ears that lacked the PSCC and had an obviously shortened cochlear duct (6/16; Fig. 6J). Thus, histological and paintfill analyses show that the Fgf10;Slc26a9P2ACre CKO inner ear structural phenotype is very similar to, but appears milder than, that of Fgf10;Tg(Pax2-cre) CKO inner ears.
Fgf8;Slc26a9P2ACre conditional mutants are viable and have mild low-frequency hearing loss associated with defective mid-apical inner pillar cell differentiation
Fgf8, which is expressed by inner hair cells (IHCs), is required for complete differentiation of the adjacent inner pillar cells (IPCs) (Jacques et al., 2007; Zelarayan et al., 2007). However, the Fgf8;Foxg1Cre conditional mutants that were studied die at birth, as do Fgf8;Tg(Pax2-cre) conditional mutants, so auditory phenotypes of such mutants cannot be assessed. We generated Fgf8;Slc26a9P2ACre conditional mutants and found that they survived in the numbers expected from the cross (9/35 were CKOs; expected=8.75). These conditional mutants had modestly, but significantly, elevated ABR thresholds for click and 8 kHz stimuli relative to the Cre-negative control genotype tested (Fig. 7A). To assess IPC differentiation, we stained E18.5 whole-mount cochleae using antibodies directed against p75NTR, and counterstained with phalloidin to reveal IPCs and HCs, respectively (Fig. 7B). As expected from studies of the Fgf8;Foxg1Cre conditional cross (Jacques et al., 2007; Zelarayan et al., 2007), the apical surface of Fgf8−/c control cochleae showed continuous basal-to-apical expression of p75NTR between the inner and outer HCs (n=3). Similar results were obtained with the other control genotypes, Fgf8c/+ and Fgf8c/+;Slc26a9P2ACre/+ (data not shown). In contrast, Fgf8;Slc26a9P2ACre CKO cochleae (n=4) had gaps in p75NTR staining indicative of incomplete IPC differentiation. The gaps were more numerous in middle and apical segments, and had a similar distribution to those seen in Fgf8;Tg(Pax2-cre)/0 cochleae (n=3). Immunostaining of cochlear cross-sections with antibodies directed against p75NTR and MYO7A (n=2 each) was consistent with these results, showing that in the base of all genotypes, the IPC head extended above the level of the HCs, but in mid and apical regions of both types of Fgf8 CKOs, the IPC head was level with the HC surface. In addition, the distance between an IHC and the first OHC was narrow (Fig. S1). Thus, the more severe mid-apical disruption in pillar cell differentiation correlates with the modest low-frequency hearing loss.
Inducing dnFGFR2b from E17.5 to P3 in the Slc26a9 lineage causes hearing loss and disrupts the outer sulcus
To assess the utility of Slc26a9P2ACre for activating gene expression, we crossed Rosa26lslrtTA/lslrtTA to Slc26a9P2ACre/+;Tg(dnFgfr2b)/0 mice and treated pregnant/lactating females from E17.5 to P3 with doxycycline (DOX). The cross generated all four expected genotypes, of which only one [Slc26a9P2ACre/+;Rosa26lslrtTA/+;Tg(dnFgfr2b)/0] was expected to express dnFGFR2b, a ligand trap that inhibits signaling by FGF3 and FGF10 in the inner ear (Urness et al., 2018). Offspring were tested at 3-5 weeks for ABR thresholds. Control pups with any of the three genotypes lacking one or both of the components needed to produce dnFGFR2b had normal ABR thresholds for all stimuli, with considerable variability at the highest test frequency (32 kHz; Fig. 8A). In contrast, Rosa26lslrtTA/+;Slc26a9P2ACre/+;Tg(dnFgfr2b)/0 mice had significantly elevated ABR thresholds (37-44 dB) for all stimuli except 32 kHz, at which the only significant difference was with the corresponding Cre-negative control (Fig. 8A).
To determine the potential cause of this moderate dnFGFR2b-induced hearing loss, we immunostained P3 cochlear cross-sections with markers of sensory and supporting cells. Normal patterns of MYO7A (sensory hair cells), PROX1 (pillar and Deiters' supporting cells) and CD44 (outer pillar and non-sensory outer sulcus cells) staining were seen in controls [Fig. 8B; only Rosa26lslrtTA/+;Tg(dnFgfr2b)/0 is shown, n=2]. However, although Slc26a9P2ACre/+;Rosa26lslrtTA/+;Tg(dnFgfr2b)/0 cochleae had normal sensory and supporting cells, the outer sulcus, a non-sensory tissue required to maintain endolymph composition and dependent on Fgf10 for its specification around E11.5 (Urness et al., 2015), was missing, as indicated by the loss of CD44 staining lateral to the supporting cells (n=3; Fig. 8C). Similar results were seen when Tg(Pax2-cre) was used to induce rtTA expression (n=1, Fig. 8D). Thus, Slc26a9P2ACre readily activates rtTA expression, and FGFR2b ligands are required between E17.5 and P3 to maintain outer sulcus development and normal auditory thresholds.
Our data show that Slc26a9P2ACre is a promising new tool for otic gene manipulation. Like the classic pan-otic Cre alleles Tg(Pax2-cre) and Foxg1Cre, Slc26a9P2ACre recombines LoxP sites efficiently throughout the otic inner ear epithelia and ganglia, but unlike those Cres, it does not recombine appreciably in the CNS, particularly in the auditory and vestibular nuclei or cerebellum, or in the middle ear ossicles. Thus, it may prove useful for separating peripheral sensory and central origins for auditory and vestibular phenotypes. Furthermore, the extra-otic recombination sites are very different from those of Tg(Pax2-cre) and Foxg1Cre, potentially allowing postnatal survival of conditional mutants, as we found with Fgf10 and Fgf8. This property allowed us to show for the first time that both of these genes are indeed required for normal auditory function.
In comparison with Tg(Pax2-cre) CKOs, however, we found that, whereas the Fgf8; Slc26a9P2ACre IPC phenotypes were similar, the Fgf10;Slc26a9P2ACre morphological phenotypes were variably milder than their Tg(Pax2-cre) counterparts. Two factors may contribute to this difference. Although Slc26a9P2ACre-directed recombination is pan-otic, it starts at E9.5, 1 day later than with Tg(Pax2-cre). As Fgf8 is not expressed in the cochlea until E15.5 (Hayashi et al., 2007), the deletion with either CRE driver occurs many days before Fgf8 would normally start to be expressed. Thus, in both conditional mutants, it is unlikely that any Fgf8 is ever produced in the cochlea and the pillar cell phenotypes are the same. In contrast, Fgf10 is expressed in a very dynamic pattern, starting at or before formation of the first somite in mesenchyme underlying the preplacode and subsequently in the otic cup epithelium starting at E9.0-E9.25 (Pirvola et al., 2000; Urness et al., 2018; Wright and Mansour, 2003). Thus, in both cases, CRE is activated after Fgf10 has already been produced. The differences between the Fgf10;Slc26a9P2ACre and Fgf10;Tg(Pax2-cre) CKO phenotypes could reflect a specific requirement for Fgf10 between E8.5 and E9.5 in subsequent otic morphogenesis and/or perdurance of FGF10 protein after its gene has been inactivated, which is later in the case of Slc26a9P2ACre than for Tg(Pax2-cre), giving more opportunity for FGF10 accumulation. Unfortunately, antibodies able to localize and quantify FGF10 in otic tissues to address this issue more directly have not been described.
Although the basal-to-apical increase in IPC head gaps observed in cochlear surface preparations of Fgf8;Tg(Pax2-cre) and Fgf8;Slc26a9P2ACre CKOs appeared similar, as did the shortening and narrowing of the IPC space observed in cross-sections, these phenotypes seemed milder than the more frequent and non-graded appearance of IPC head gaps and the cross-sectional IPC area seen in Fgf8;Foxg1Cre CKOs (Jacques et al., 2007; Zelarayan et al., 2007). It is possible that this is a consequence of differing genetic backgrounds. Another possibility, however, is that as Foxg1 is expressed in the IHC along with Fgf8, the previously reported Fgf8 CKO phenotypes reflect a genetic interaction between Foxg1, which is heterozygous in the CKOs, and Fgf8 to regulate IPC differentiation, similar to the genetic interaction seen between Foxg1 and Fgf10 regulating morphogenesis of the inner ear (Pauley et al., 2006). If so, it may be necessary to re-evaluate cochlear CKO phenotypes generated using Foxg1Cre with this possibility in mind.
One of the characteristics of Fgf10 germline null cochleae is the absence of both Reissner's membrane and most or all of the outer sulcus. Based on developmental studies of marker genes in those mutants, it appeared that the phenotypes initiated sequentially between E12.5 and E15.5. Thus, it was surprising to find such a severe effect on the outer sulcus after inducing dnFGFR2b starting at E17.5, which is well after the start of outer sulcus specification. This suggests an ongoing requirement for FGFR2b/1b ligands in the maintenance of this tissue: a requirement that is not shared with Reissner's membrane. The more highly penetrant loss of outer sulcus tissue and somewhat more severe hearing loss following dnFGFR2b induction compared with the Fgf10 conditional knockout suggests that FGF3, which is expressed just medial to the outer sulcus, presumably in the third row of Deiters' cells (Kolla et al., 2020; Urness et al., 2018), and which is also a ligand for FGFR2b (Zhang et al., 2006), may be involved in this maintenance function.
The losses/reductions of Reissner's membrane and the outer sulcus, non-sensory tissues involved in maintaining the unique ionic composition of cochlear endolymph, which is critical for sensory cell function (Jagger and Forge, 2013; Kim and Marcus, 2011; Wangemann, 2006), may explain the reduced ABR thresholds in Fgf10 CKO and dnFGFR2b-expressing mice. However, the relatively moderate hearing loss found in both cases was somewhat surprising and suggests that the systems for endolymph maintenance are highly redundant. It would be interesting to conduct more detailed studies of the mutant endolymph and determine whether the observed hearing loss progresses in severity over time.
Finally, it is interesting that the Slc26a9P2ACre extra-otic lineages are primarily epithelia that, like the inner ear, are dependent on Fgf10 for development. For example, Slc26a9 lineage was found in lung, stomach, intestine, kidney and pancreas epithelia, as well as in Rathke's pouch, all of which require adjacent Fgf10 for development (Lv et al., 2019; Min et al., 1998; Ohuchi et al., 2000; Sekine et al., 1999). In addition, corneal epithelial proliferation depends on FGFR2, which is likely stimulated by adjacent mesenchymal FGF10 (Zhang et al., 2015). Thus, not only will Slc26a9P2ACre be useful for gene regulation in those tissues, but Slc26a9 expression might serve as a sensor for FGFR2b/1b signaling in many contexts, and Slc26a9 promoter elements could be useful in constructing FGFR2b/1b-dependent reporters.
MATERIALS AND METHODS
Mouse alleles, genotyping and crosses
Generation and germline transmission of Slc26a9P2ACre mice complied with protocols approved by the UNMC Institutional Animal Care and Use Committee (IACUC). Subsequent studies of these and all other mice complied with protocols approved by the University of Utah IACUC. The Slc26a9P2ACre strain will be submitted to a public repository for distribution. Until then, requests should be directed to S.L.M.
The Slc26a9-null allele (Slc26a9−; Slc26a9tm1Sole; MGI 3822918) was obtained on a C57Bl/6 background from Dr Manoocher Soleimani (University of Cincinnati, OH, USA) and genotyped using a 3-primer PCR assay modified from Xu et al. (2008). A forward primer in Neo (5′TGCGAGGCCAGAGGCCACTTGTGTAGC3′) and a forward primer in exon 5 (5′TGACTGCTGTCATCCAGGTGAGC3′) were combined with a reverse primer in the exon 5-6 intron (5′AAGGGGCATCTGAAAGGTTCTAGGC3′), generating a 325 bp product from the null allele and a 250 bp product from the wild-type allele. Heterozygotes from the first or second generation of backcrossing to CBA/CaJ were intercrossed to generate pups of all three genotypes for ABR testing.
The Fgf10 null (Fgf10−; Fgf10tm1.1Sms; MGI:3526181) and Fgf10 conditional (Fgf10c; Fgf10tm1.2Sms; MGI:4456398) alleles and genotyping methods were described previously (Urness et al., 2010, 2015). The Fgf8 null allele (Fgf8−;Fgf8tm1Mrc; MGI:2150350) was obtained from Moon and Capecchi (2000) and genotyped as described (Mansour et al., 2013). The Fgf8 conditional allele (Fgf8c; Fgf8tm2Moon; MGI:3639320) was obtained from Park et al. (2006) and genotyped as described (Fresco et al., 2016). All Fgf mutant alleles were kept on a mixed genetic background comprised of C57Bl/6 and various 129 substrains. To generate conditional mutants and controls, Fgf(8 or 10)c/c females were crossed to Fgf(8 or 10)−/+;Slc26a9P2ACre/+ males.
The origin and genotyping of Tg(tetO-dnFgfr2b) (MGI 5582625) was as described previously (Urness et al., 2018). The Rosa26lslrtTA allele [Gt(ROSA)26Sortm1(rtTA,EGFP)Nagy; MGI:3583817] was obtained courtesy of Dr Charles Murtaugh (University of Utah, USA) and genotyped as described previously (Belteki et al., 2005). To enable DOX-mediated induction of dnFGFR2b, we crossed Rosa26lslrtTA/lslrtTA;Tg(tetO-dnR2b)/0 females with Slc26a9P2ACre/+ males. dnFGFR2b expression was induced with an intraperitoneal injection of E17.5 pregnant dams with 0.1 ml/10 g body weight of 0.45 mg/ml (4.5 mg/kg body weight) doxycycline hyclate (Sigma-Aldrich) prepared in PBS, followed by provision of DOX chow (6 g/kg, Custom Animal Diets) ad libitum until P3, when standard chow replaced DOX chow.
Tg(Pax2-cre)1Akg (MGI:3046196) substituted for Slc26a9P2ACre where indicated and was genotyped as described previously (Urness et al., 2018). The sex of embryonic and early postnatal samples was not determined. The sex of animals used for ABR measurements was noted and was roughly equal within genotypic classes.
RNA in situ hybridization
Embryos were harvested and fixed in 4% paraformaldehyde solution (PFA) and stored in methanol at −20°C. Whole-mount RNA in situ hybridization of embryos to detect Slc26a9 mRNA, and fixation, embedding and cryosectioning of stained embryos was conducted as described previously (Urness et al., 2010). Whole embryos were photographed using a stereomicroscope (Zeiss Discovery V12) fitted with a digital camera (QImaging Micropublisher 5.0). Tissue sections were photographed under DIC illumination (Zeiss Axioskop) using a digital camera (Lumenera Infinity3).
Auditory brainstem response testing
Slc26a9−/+ intercross offspring of both sexes were tested at P21-P49 for auditory brainstem response thresholds to broadband click and 8, 16 and 32 kHz pure tone pip stimuli as described previously (Mansour et al., 2009). An additional stimulus (24 kHz pure tone pip) was added to the tests of the Fgf10 (P28-P31) and Fgf8 (P31-P51) conditional crosses and to DOX-induced dnFGFR2b (P20-P39) offspring. The Fgf10 CKOs were compared with each of the three control genotypes (Fgf10c/+, Fgf10c/+;Slc26a9P2ACre/+ and Fgf10c/−). The Fgf8 CKOs were compared with a single control genotype (Fgf8c/−). The Tg(tetO-dnFgfr2b)/0;Rosa26lslrtTA/+;Slc26a9P2ACre/+ genotype was compared with the Tg(tetO-dnFgfr2b)/0;Rosa26lslrtTA/+ control genotype. Testing of all pups was performed blind with respect to genotype, but most of the Fgf10 conditional cross CKO offspring could be identified as such based on head tossing/circling behavior. Statistical comparisons between genotypes were made using Prism 8.3.1 (GraphPad) to compute two-way repeated measures ANOVA with Tukey's or Sidak's correction for multiple comparisons as indicated to adjust P-values.
Generation of Slc26a9P2ACre mice
We used Easi-CRISPR to generate Slc26a9P2ACre founders in the C57Bl/6 background as described in detail previously (Miura et al., 2018; Quadros et al., 2017). Briefly, CAS9 protein, tracer RNA and the Slc26a9-specific CRISPR RNA (5′-GGAACCAAAUGGGUCCUCACA-3′) were purchased from Integrated DNA Technologies (IDT) and used to form a ribonucleoprotein complex designed to cleave Slc26a9 inside the stop codon (TG/A). The long single-stranded donor DNA was custom synthesized by IDT (Megamer) and comprised the 66-base P2A sequence (Kim et al., 2011) in-frame with a codon-optimized Cre sequence (Ohtsuka et al., 2013; Shimshek et al., 2002) designed from pBBI (Addgene, 65795). These were flanked on the 5′ and 3′ sides with 93 bases and 95 bases, respectively, of homology to Slc26a9.
The Slc26a9-directed RNP (10 ng/µl) was mixed with single-stranded donor DNA (10 ng/µl) and microinjected into 38 C57Bl/6 zygotes, of which 34 were implanted into the oviducts of two foster mothers following standard procedures described previously (Harms et al., 2014). The seven resulting pups were tested for correct targeting at both 5′ and 3′ sides of the CAS9 cleavage site by PCR amplification of tail DNA using primer pairs Slc5′F plus CreSeq5′R and CreSeq3′F plus Slc3′R. The primer locations and expected PCR product sizes are shown in Fig. 2D. The sequences of the primers are: Slc5′F, GGAGGAACACAGTTCACAGT; CreSeq5′R, CTTCTGCCACCTCCCAACTG; CreSeq3′F, GTGTCTGGTGTGGCTGATGACC; Slc3′R, ATGGGTTCACCAGAGTCTCATC. Both PCR products from each of the four pups with correctly sized products were sequenced and had no deviations from the expected sequence (data not shown). Two correctly targeted female founders (6 and 7) were crossed to C57Bl/6 males to transmit the new allele. Male heterozygotes were crossed to Rosa26lslLacZ/lslLacZ females to assess CRE activity at E10.5; as both lines produced the same results, only one (7) was selected for further study. Heterozygous intercrosses were genotyped using a 3-primer mix of Slc5′F+CreSeq3′F+Slc3′R. The wild-type and P2ACre alleles produce 514 and 598 bp products, respectively. Homozygotes were selected to start a maintenance colony and heterozygotes were used for lineage and conditional crosses.
Immunostaining of embryonic and postnatal cryosectioned tissues
P1 heads were bisected in the sagittal plane and fixed in 4% PFA, embedded in sucrose/gelatin and cryosectioned at 8 µm as described previously (Mansour et al., 2013). E18.5 or P3 cochleae were isolated and prepared similarly. Whole E9.5 and E10.5 embryos were fixed in 4% PFA, embedded in sucrose/gelatin and cryosectioned in the transverse plane as described previously (Urness et al., 2018). Primary antibodies were diluted into PBS/5% normal serum of the secondary antibody species/0.2% TritonX-100 and applied at the following dilutions: PROX1 (R&D, AF2727, 1:100), NFH (Millipore, AB1989, 1:1500), NEUROD (Santa Cruz, sc-1084, 1:50), p75NTR (Millipore, 07-476, 1:600), MYO7A (Proteus, 25-6790, dilution or DSHB 138-1, 1:100) and CD44 (BD-Pharmingen, 550538, 1:1000). Secondary antibodies [Alexa Fluor 488 goat anti-rabbit IgG (Invitrogen, A11034), Alexa Fluor 594 or 488 goat anti-mouse IgG (Invitrogen, A11032 or A11001, respectively), and Alexa Fluor 594 goat anti-rat IgG (Invitrogen, A11007)] were diluted 1:400 in PBST/5% normal serum. DAPI was included in the mounting medium (Vectashield, Vector Labs). Fluorescent signals were observed under epi-illumination on a Zeiss Axioskop and captured using an Infinity3 camera (Lumenera) driven by InfinityAnalyze software. Channels were overlaid using Photoshop CC 2019/2020.
Histology at E18.5 and paintfilling of E15.5 inner ears
E18.5 heads were bisected in the sagittal plane, fixed in 4% PFA overnight and prepared for paraffin embedding, sectioning and staining with Hematoxylin and Eosin as described previously (Mansour et al., 2009). Filling of E15.5 inner ears with latex paint and photography was as described previously (Urness et al., 2015).
Staining of whole-mount cochlear surface preparations
Temporal bones were isolated and fixed in 4% PFA overnight and stored in PBS at 4°C. Cochleae were freed from the otic capsule and modiolus using fine dissection tools, and Reissner's membrane, the lateral wall of the cochlear duct, and remnant ganglion were trimmed with fine scissors to reveal the sensory epithelium. Cochleae were rinsed in PBS/0.1% TritonX-100 (PBST), blocked for 1 h in 5% normal goat serum in PBST and incubated overnight with rabbit anti-p75NTR (Millipore, 07-476, 1:1000) and phalloidin-Alexa Fluor 488 (Invitrogen, A12379, 1:2000). The following day, cochleae were washed three times for 10 min in PBST and incubated with Alexa Fluor 594 goat anti-rabbit IgG (Invitrogen, A11012) at 1:400 in PBST/5% normal goat serum for 1-2 h at room temperature. Cochleae were washed as previously described and mounted in 20 µl Fluoromount-G (Invitrogen, 00-4958-02). Fluorescent signals were observed and captured as described above.
We thank Dr Manoocher Soleimani for providing the Slc26a9 KO mice, Drs Scott Rogers, Susan Chapman, Kristen Kwan and Shannon Odelberg, respectively, for consultation on the brain, middle ear, eye and statistics.
Conceptualization: L.D.U., C.B.G., S.L.M.; Methodology: L.D.U., R.M.Q., D.W.H., C.B.G., S.L.M.; Formal analysis: S.L.M.; Investigation: L.D.U., X.W., C.L., R.M.Q., D.W.H., S.L.M.; Writing - original draft: L.D.U., S.L.M.; Writing - review & editing: L.D.U., R.M.Q., C.B.G., S.L.M.; Visualization: L.D.U., X.W., C.L., S.L.M.; Supervision: L.D.U., C.B.G., S.L.M.; Project administration: C.B.G., S.L.M.; Funding acquisition: S.L.M.
This work was supported by grants from the National Institutes of Health [R01DC011819, R21DC014779 and R01DC014470 to S.L.M.]. Deposited in PMC for release after 12 months.
The authors declare no competing or financial interests.