Enhancer trap lines with GFP driven by smad6b and frizzled1 regulatory sequences for the study of epithelial morphogenesis in the developing zebrafish inner ear

Abstract Live imaging in the zebrafish embryo using tissue‐specific expression of fluorescent proteins can yield important insights into the mechanisms that drive sensory organ morphogenesis and cell differentiation. Morphogenesis of the semicircular canal ducts of the vertebrate inner ear requires a complex rearrangement of epithelial cells, including outgrowth, adhesion, fusion and perforation of epithelial projections to generate pillars of tissue that form the hubs of each canal. We report the insertion sites and expression patterns of two enhancer trap lines in the developing zebrafish embryo, each of which highlight different aspects of epithelial cell morphogenesis in the inner ear. A membrane‐linked EGFP driven by smad6b regulatory sequences is expressed throughout the otic epithelium, most strongly on the lateral side of the ear and in the sensory cristae. A second enhancer trap line, with cytoplasmic EGFP driven by frizzled1 (fzd1) regulatory sequences, specifically marks cells of the ventral projection and pillar in the developing ear, and marginal cells in the sensory cristae, together with variable expression in the retina and epiphysis, and neurons elsewhere in the developing central nervous system. We have used a combination of methods to identify the insertion sites of these two transgenes, which were generated through random insertion, and show that Targeted Locus Amplification is a rapid and reliable method for the identification of insertion sites of randomly inserted transgenes.


| INTRODUC TI ON
Morphogenesis of the complex labyrinth of the vertebrate inner ear from a simple ball of epithelial cells exemplifies many core developmental processes (reviewed in [Alsina & Whitfield, 2017]). In zebrafish, the formation of the three semicircular canal ducts involves the formation of finger-like projections of otic epithelium that grow into the lumen of the otic vesicle, where they adhere, fuse, and perforate to form pillars of tissue (Waterman & Bell, 1984). Each pillar subsequently widens to contribute to the inward-facing wall of the curved semicircular canal duct. To understand these dynamic processes at the tissue and cellular level, it is most informative to image them in the live embryo, in real-time. Techniques such as light-sheet microscopy are able to image gently over long time courses with low photobleaching (Huisken et al., 2004;Power & Huisken, 2017), and the use of spinning disk confocal or Airyscan technologies allow visualisation of subcellular details in live specimens (Lam et al., 2014;Wu & Hammer, 2021).
A variety of transgenic lines have been successfully exploited for live imaging of the developing zebrafish inner ear. The ubiquitous actb2 promoter drives transgene expression in all cells of the embryo, including the otic epithelium, and the use of tags on fluorescent proteins, or fusion to other proteins, can be used to highlight different subcellular compartments (Mosaliganti et al., 2019;Munjal et al., 2021;Swinburne et al., 2018). Promoters for cldnb (Haas & Gilmour, 2006), sp7 (DeLaurier et al., 2010), and sox10 (Carney et al., 2006) can also be used to drive transgene expression throughout the developing otic vesicle. A few cell-type-specific transgenic lines are available for studying the developing ear in zebrafish, including those utilising promoters that drive expression in neuronal progenitors (isl2; neurod), sensory hair cells (pou4f3; myo6b) and supporting cells (agr2) (reviewed in [Baxendale & Whitfield, 2016]). The identification of additional transgenic lines that define specific regions of the inner ear, including non-sensory epithelial structures, would be advantageous, facilitating the analysis of regional cell behaviours.
One way of identifying new transgenic marker lines is through random insertion, where lines of interest can be selected on the basis of expression pattern. Identification of the insertion site of the transgene can reveal both the genomic enhancer sequences controlling transgene expression and the endogenous gene or genes that they regulate. Many PCR-based techniques are available for mapping insertion sites; examples include inverse PCR (Brown et al., 2022), linkermediated PCR (Ellingsen et al., 2005;Wu et al., 2003), or amplification between transgene-specific primers and degenerate primers (Liu & Whittier, 1995;Ma & Zhang, 2015;Parinov et al., 2004). Sequencebased approaches can also be used, but require a high coverage to be certain to find the region of interest. This problem can be overcome by combining cross-linking, proximity ligation, and selective amplification to target the insertion region prior to DNA sequencing (De Vree et al., 2014;Hottentot et al., 2017).
We describe two enhancer trap lines generated by random insertion that form useful markers for characterising epithelial morphogenesis in the developing zebrafish inner ear. We have used both PCR-based and Targeted Locus Amplification approaches to identify the insertion sites for these transgenes. We show them to be driven by regulatory sequences for smad6b, which codes for a BMP regulator, and fzd1, which codes for a Wnt receptor. These lines provide useful tools for imaging the developing zebrafish ear and understanding the morphogenetic events that happen during the formation of the semicircular canal system.

| Generation of the Et(smad6b:EGFP-CAAX) line
The Et(smad6b: EGFP-CAAX) line was generated in an enhancer trap screen by ligating a 4.1 kb fragment upstream of the zebrafish en2a coding region to a minimal c-fos (fosab; zfin.org) promoter driving membrane-(CAAX)-tagged EGFP in the Gateway vector pDestTol2, generating the construct en2acfospromEGFPMycpDest. This was injected into the KWT strain using the Tol2 system (Kawakami, 2007).
Injected embryos were grown to adulthood and their offspring were selected on the basis of GFP expression in the ear.

| Generation of the Et(fzd1:EGFP) line
The Et(fzd1:EGFP) line was originally generated during the establishment of a transgenic zebrafish assay for enhancer activity, using the T2K-gata2-EGFP-C1 destination vector (Ishibashi et al., 2013), generating the construct T2K-gata2-EGFP-miR-137. Fish were selected that showed expression of GFP (cytoplasmic) in the ventral pillar of the developing otic vesicle.

| Thermal asymmetric interlaced (TAIL)-PCR
Thermal Asymmetric Interlaced (TAIL)-PCR (Liu & Whittier, 1995;Parinov et al., 2004) alternates a higher annealing temperature for Tol2-specific primers with a lower annealing temperature for degenerate primers designed to bind to the genomic sequence.
This approach was used to generate short fragments flanking the transgene insertion site. TAIL-PCR products from secondary and tertiary nested PCR reactions were separated by gel electrophoresis, and fragments with the correct size shift-corresponding to the distance between the nested Tol2 primer sites-were isolated and sequenced. For the Et(smad6b: EGFP-CAAX) transgene, Tol2specific and degenerate (AD) primer sequences used were: Tol2 AWA GNC SWCAA, as described (Liu & Whittier, 1995). Sequences from either side of the insertion site were mapped to the zebrafish GRCz11 genome assembly. For the Et(smad6b: EGFP-CAAX) line, two fragments (Tol2 5′-3/AD11 and Tol2 3′-3/AD5) were identified from the 5′ and 3′ ends of the insertion site at Ch18:19,702,439 (GRCz11), at the 3′ end of the smad6b locus (+3378 bp). A third fragment was also identified that mapped to a second site on chromosome 16.

| Targeted locus amplification (TLA)
Targeted Locus Amplification (TLA) (Cergentis B.V., Utrecht, Netherlands (De Vree et al., 2014;Hottentot et al., 2017)) is a crosslinking-based technique that enables the selective amplification of local genomic sequence surrounding a transgene insertion. For cell preparation from zebrafish embryos, the protocol was adapted as follows. For each transgene, two hundred zebrafish embryos at the 15-20 somite stage were dissociated in a clean Petri dish under a coverslip in a minimal volume of Ca 2+ -free Ringer's solution, as previously described (Baxendale et al., 2009). Cells were collected in a 1.5 mL reaction tube and separated from yolk platelets by centrifugation at 300 g for 1 min. The cell pellet was gently resuspended in Accumax™ cell dissociation medium (Invitrogen/Thermofisher) and incubated until cells were fully dissociated. DNA from ~10 7 cells was crosslinked using the Cergentis kit, before shipping to Cergentis B.V.
Two insertion sites were identified for Et(smad6b: EGFP-CAAX), which matched the sites identified by TAIL-PCR, and one site was identified for Et(fzd1:EGFP). Insertion sites were amplified by PCR between a genome-specific primer and a Tol2-specific primer on both sides of each insertion, using the following genome-specific

| In situ hybridisation
Embryos were fixed by incubating in 4% paraformaldehyde overnight at 4°C. Antisense probe synthesis and in situ hybridisation were carried out as described (Thisse & Thisse, 2008). In some samples, embryo pigment was removed after the fixation step by incubation in a bleaching solution (PBS, 3% H 2 O 2 , 0.5% KOH) for 30 min.

| Microscopy and image analysis
Images of in situ hybridisation stains were taken on an Olympus BX51 compound microscope equipped with differential interference contrast (DIC) optics, a C3030ZOOM camera, and CellB software, or a Micropublisher 6 camera and Ocular software. (water immersion) objectives or a Nikon CSU-W1 spinning disk confocal microscope using 10×/0.45 N.A. and 20×/0.75 N.A. objectives.
Airyscan images were processed using Zen Black 2.3 SP1 software using Auto Airyscan processing settings.
Time-lapse imaging was performed using a Zeiss Z.1 light-sheet microscope and ZEN Black 2014SP1 software using a 20×/1.0 N.A.
water immersion objective with 1.0× or 1.5× optical zoom. Embryos were mounted vertically (either head up or head down) in 0.8% low melting point agarose in E3 medium in glass capillaries and imaged at 28.5°C in E3 medium with 4% tricaine as an anaesthetic in the imaging chamber. Images were acquired every 5 min using 5% 488 nm and 5% 561 nm laser power, 99.9 ms exposure time, light sheet thickness of 4.49 μm, and a z-interval of 1 μm.
For multiview imaging in the Zeiss Z.1 light-sheet microscope, embryos were mounted in 0.8% agarose containing a 1/8000 dilution of red fluorescent beads (0.5 μm, Sigma). Z-stacks encompassing the whole depth of the sample were acquired from four different angles (at 90° intervals) using a 20×/1.0 N.A. water immersion objective and 0.36× optical zoom, 2.5% 488 nm and 2.0% 561 nm laser power, 99.9 ms exposure time, light-sheet thickness of 7.32 μm and a z-interval of 1.652 μm. Multiview images were registered using landmark alignment in Zen Black 2014SP1 software. The registered datasets were then fused and deconvolved using mean fusion followed by fast-iterative deconvolution using PSFs extracted from the bead channel. Fused, deconvolved datasets were 3D rendered using arivis Vision4D (v3.4.0). Images were processed in Fiji (Schindelin et al., 2012) and fluorescence images were assembled using the QuickFigures Plugin (Mazo, 2021).

| A smad6b:EGFP Enhancer trap line marks cells of the zebrafish otic epithelium
The Et(smad6b: EGFP-CAAX) enhancer trap line expresses GFP in most cells of the otic epithelium (Figures 1 and 2; Movie S1).
Expression in the otic vesicle is present from 24 h post fertilisation (hpf) and remains throughout development, more strongly on the lateral side of the ear. As the otic vesicle grows, the GFP expression level varies; fluorescence in the three cristae is particularly bright and much stronger than in the two maculae (Figure 1b

| A frizzled1:EGFP Enhancer trap line marks the ventral epithelial projection and pillar of the developing zebrafish ear
We have also characterised a second enhancer trap line, fortuitously generated during establishment of a zebrafish reporter assay for human enhancer activity (Ishibashi et al., 2013). Occasionally, ran- We initially attempted to use TAIL-PCR to identify the insertion site for this transgene, but this was unsuccessful, most likely because the insertion site turned out to be in a repetitive site close to a telomere (see Figure S3). We next turned to the TLA approach.
Sequencing reads revealed that the transgene is located on chro- To validate the insertion site further, we generated antisense RNA probes to fzd1, gfp, and cdk14, and used these to compare expression of these genes by in situ hybridisation, and to the GFP fluorescence. Expression of fzd1 mRNA in wild-type embryos is a close temporal and spatial match to both the gfp mRNA and GFP expression pattern in transgenic embryos in the brain and ear In zebrafish, fzd1 expression has only previously been characterised at earlier embryonic stages (Nikaido et al., 2013). cells in the cochlea (Yu et al., 2010). Expression of Frizzled1 mRNA has been reported in the otocyst of chick (Sienknecht & Fekete, 2009) and mouse (Borello et al., 1999;Durruthy-Durruthy et al., 2014), and at later stages in the mammalian cochlea (Daudet et al., 2002;Yu et al., 2010). It is not yet known which Wnt acts as a ligand for Fzd1 in the zebrafish ear, although Wnt3, facilitated by the co-receptor Lrp5, interacts with Fzd1 in the zebrafish brain (Veerapathiran et al., 2020). Transcripts for wnt3 are expressed in the zebrafish ear, at least from 4 days post fertilisation (4 dpf) (Clements et al., 2009); wnt4 is also expressed in ventral otic epithelium (Thisse & Thisse, 2005).

| CON CLUS ION
Enhancer trapping provides an unbiased approach to generating reporter lines for the live imaging of specific cell types. This technique can have different advantages compared with the more recent targeted approaches that use gene editing techniques to knock-in reporter genes at a specific locus (see, for example, Wierson et al., 2020).  (Waterman & Bell, 1984) or conventional histology (Haddon & Lewis, 1996); the use of live imaging will now make it possible to determine the origin and fate of individual cells within the pillar.
In summary, the Et(smad6b:EGFP-CAAX) and Et(fzd1:EGFP) lines are useful tools for live imaging of the otic vesicle in the developing zebrafish embryo. Our results also demonstrate that Targeted Locus Amplification is a robust method for the rapid identification of the genomic insertion sites of randomly-inserted transgenes, even where the insertion site is close to a telomere or in repetitive sequence, and should be widely applicable.

AUTH O R CO NTR I B UTI O N S
DB and SB performed the experimental work and data analysis, with imaging support from NvH and technical support from MM and EG.
The Et(smad6b: EGFP-CAAX) line was originally generated and provided by RK; the Et(fzd1:EGFP) line was originally generated by FS-G, in the lab of T.S. Becker (University of Sydney). DB, SB, NvH and TTW prepared the figures. DB, SB and TTW designed the study and wrote the article.

ACK N OWLED G M ENTS
We thank the Nieto lab for use of the Tg(xEF1A:H2B-RFP) line. We thank the aquarium team at Sheffield for expert animal care and the team at Cergentis for helpful discussion.

This work was funded by the Biotechnology and Biological Sciences
Research Council (BBSRC: BB/J003050 to TTW, BB/M01021X/1 and BB/S007008/1 to TTW and SB, and BB/D020433/1 to RDK).
DB was funded by an Anatomical Society Ph.D. studentship.
Light-sheet imaging was carried out in the Sheffield Wolfson Light Microscopy Facility, supported by a BBSRC ALERT14 award (BB/ M012522/1) to TTW and SB. The funders had no role in study design, data collection and analysis, decision to publish, or preparation of the article.

CO N FLI C T O F I NTER E S T S TATEM ENT
The authors have no conflicts of interest to declare.

DATA AVA I L A B I L I T Y S TAT E M E N T
The data that support the findings of this study will be made available on reasonable request. The Et(smad6b:EGFP-CAAX) and Et(fzd1:EGFP) lines will be maintained as live fish and/or preserved as cryogenic sperm samples at the University of Sheffield and will be made available on reasonable request.

LI CEN CI N G
For the purpose of open access, the authors will apply a Creative Commons Attribution (CC BY) licence to any Author Accepted Manuscript version arising from this study.