Conserved miRNA targets in the chicken Scrt2 3′UTR (ENST00000246104.6) were searched using TargetScan v7.2 (Agarwal et al., 2015¹. An overlapping site for miR-125b and miR-200bc/-429 and another one for miR-204/-211 were identified and the expression pattern of these miRNAs was verified at GEISHA (Gallus Expression in situ Hybridization Analysis, Darnell et al., 2006)². Only miR-125b and miR-200b were expressed in the neural tube.

Corresponding regions of the chicken genome (galGal6) miR-125/-200b overlapping site in other species were retrieved from the UCSC Genome Browser multiple alignment of 77 vertebrates. The resulting alignment showed conservation among amniotes only. To confirm the candidate homologous elements in X. tropicalis and zebrafish Scrt2, their predicted cDNA sequences were scanned for miR-125b/-200b sites and sequences surrounding the candidates identified were aligned to human, mouse, chicken and painted turtle REs obtained previously. The zebrafish candidates were also identified using TargetScan. The phylogenetic tree was constructed based on multiple alignment using PhyloP (UCSC Genomics; Siepel et al., 2005) and edited on iTOL v4 (Letunic and Bork, 2019)³ to propose the appearance of miR-125b and 200b responsive elements during the evolution of Scrt2 (Supplementary Figure S1).

To clone the 3′UTR region of Scrt2 for the in vivo luciferase assay, we designed primers based on a 521-bp region of the putative chicken Scrt2 3′UTR generated by 3P-seq tags (Jan et al., 2011). The region was amplified by PCR of cDNA from HH23 embryos. The primers (F- 5′CTCGAGACCGGAGGCGGATCGCCGTGC and R- 5′ TCTAGATAGTGGCAGAAGTCCCTTTTATA) contained XhoI and XbaI restriction sites at their 5′ ends. The product was cloned into pGEM-T Vector (Promega) and subcloned downstream of the luciferase coding region in the pmiR-Glo vector (Promega). The resulting plasmid was named pmiR-GLO-cScrt2UTR.

For depleting miRNAs in vivo, we designed and ordered sponges sequences (GenScript, United States)⁴ containing seven MREs in tandem for each miRNA, flanked by restriction sites for XbaI and PmeI for subsequent subcloning into the pRNA-U6.1 vector (Addgene # 35664). The sponge sequence for miR-125b was 5′- TCACAAGTTACCACTCAGGGACGATTCACAAGTTACCAC TCAGGGAACCGGTTCACAAGTTACCACTCAGGGATCACT CACAAGTTACCACTCAGGGATCACAAGTTACCACCGATT CAGGGACGATTAG and for miR-200b was 5′- TCACAAGTTACCACCAGTATTACGATTCACAAGTTACCAC CAGTATTAACCGGTTCACAAGTTACCACCAGTATTATCAC TCACAAGTTACCACCAGTATTATCACAAGTTACCACCGAT CAGTATTACGATTAG.

For CRISPR/Cas9 genome editing, the sgRNA was designed to target the miR-125b site within the cScrt2 3′-UTR and annealed from single stranded oligos (sgRNA-F 5′- AGTCGTGCAATTCAGGGATATAAAA; sgRNA-R 5′- AAACTTTTATATCCCTGAATTGCAC) designed to form overhangs compatible with pcU6.3-sgRNA vector digested with BsmBI (NEB, cat. R0580S). The oligos were cloned into the pcU6.3-sgRNA vector as described (Williams et al., 2018). A scrambled control guide was generated using the “RNA sequence scrambler” tool (GenScript, United States)⁴, with the chicken genome as a base for the search of off-targets, and cloned from the oligos sgRNA Scrambled-F 5′-AGTCGGCAGGAATCAATTAGATAAT and sgRNA Scrambled-R 5′-AAACATTATCTAATTGATTCCTGCC). All final clones were sequenced to ensure that the cloned guide had no mismatches.

Fertilized eggs from Gallus gallus Leghorn (Yamaguishi Farm, São Paulo, Brazil) or White Leghorn hens (University of Connecticut, Department of Animal Science – United States) were incubated at 37.8°C and 50% humidity until the desired developmental stages according to Hamburger and Hamilton (Hamburger and Hamilton, 1992). All procedures were approved by our institutional ethic committees (CEUA ICB/USP n° 025/2013).

Electroporation procedures followed standard protocols (Harada et al., 2017). The plasmids pmiR-GLO-cScrt2UTR, pmiR-GLO, pRNA-6.1-125b-Sponge or pRNA-6.1-200b-Sponge (3 μg/μL) were mixed with pCDNA3.1-mGFP (2 μg/μL), together with 0.2% Fast Green dye (Sigma Aldrich, United States). For CRISPR/Cas9 experiments, pCAG-Cas9-2A-Citrine was mixed with pcU6.3-3′UTR-sgRNA or pcU6.3-3′UTR-sgRNA-Scrambled. Electroporation parameter were: 5 pulses of 20 V, 50 ms of duration and 100 ms of interval. Embryos were reincubated, screened for successful transfection, collected at stage HH23 and processed further accordingly.

Forty-eight hours post-electroporation, the neural tube of HH23 embryos were collected in Ringer′s solution. Three control and three experimental halves were collected and matched by electroporation extension and GFP intensity. Each sample was lysed in 1× lysis buffer from Dual-Luciferase Reporter Assay System kit (Promega, cat. #E1910) and luciferase activity detection was performed according to manufacturer’s instructions in a Synergy HT luminometer (Biotek, United States). Three technical triplicates were read for each biological sample.

Neural tube halves of HH23 embryos unilaterally electroporated with CRISPR/Cas9 system were individually microdissected with a tungsten needle (0,125 mm) and kept in Ringer’s solution until dissociation in Accumax (Accutase cat. #SCR006) cell dissociation solution for 40 min at room temperature under mild agitation. After dissociation, cells were passed through a 40 μm cell strainer (Pluriselect USA, Mini Cell Strainer II, 45-09840-50) and centrifuged at 400 g for 10 min. The supernatant was carefully discarded and cells were resuspended in 200 ml of Hank’s Balanced Salt Solution (HBSS) with 50 mM EDTA, 100 mM HEPES, pH 8.0 and 0.5% BSA. At least 4000 Citrine-positive or negative cells were sorted through fluorescence-activated cell sorting (FACS) process from experimental or control side, respectively, directly into 50 μL of lysis buffer from Power SYBR Green Cells-to-CT kit (Thermo Fisher Scientific, 4402953) using BD AriaFusion cell sorter. Control-side cells were randomly selected to match the number of contralateral citrine-positive cells.

For miRNAs absolute quantification, the truncal portions of three neural tubes from HH22 embryos were microdissected and pooled for lysis and RNA isolation with TRIzol (Invitrogen, United States). For miRNA cDNA synthesis, we used the Taqman miRNA Assay kit (Applied Biosystems, United States) with sequence-specific primers for miR-125b and miR-200b and 24 ng total RNA as input. For quantification, we used Taqman probes (Taqman miRNA Assay; Applied Biosystems) to detect hsa-miR-125b (ID 000449) and gga-miR-200b (ID 006005). We performed the quantification in three technical replicates for each sample with 6 ng cDNA per reaction. The mimics of both miRNAs (miRVANA hsa-miR-125b-5p, cat. MC10148 and gga-miR-200b-3p, cat MC1050 – Thermo Fisher Scientific, EUA) were used to generate a standard curve with serial dilutions from 7.2 to 000.72 μM/μL.

For data acquisition and analysis, we used the QuantStudio 12K Flex Real-Time PCR System (Applied Biosystems, United States) equipment. The absolute quantification was analyzed by the QuantStudio 12K software (v1.4).

For Scrt2 quantification in CRISPR-edited neural tubes, cDNA was synthesized from FACS-sorted samples with a cell-to-cT kit (Invitrogen, United States). The quantification was performed with SyBr Green (Applied Biosciences, cat. 4368577) in technical triplicates on the ViiA 7 Real-Time PCR System (Applied Biosystems, United States). The primers used were cScrt2-F 5′ CTGCTGCAGGGCCACATGCGTTCGCACA and cScrt2-R 5′ GCACTGCTTGCACTTGTAGTGCTT. HPRT was used as an endogenous control and detected with the primers cHPRT-F 5′ TGGTGAAAGTGGCCAGTTTG and cHPRT-R 5′ TCATTGTAGTCGAGGGCGTATC.

Expression relative to loading controls was calculated using the 2-ΔΔCt method (Livak and Schmittgen, 2001).

After unilateral electroporation of Cas9 and UTR-sgRNA, the electroporated side of three neural tubes were dissected independently and dissociated as described above. We then performed FACS to select at least 3000 Citrine-positive cells from each neural tube. Genomic DNA (gDNA) was extracted with TRIzol (Invitrogen, United States) and total RNA with Arcturus PicoPure RNA Isolation Kit (Thermo Fisher Scientific, United States). The gDNA edited region was PCR-amplified (F – 5′ ACCGGAGGCGGATCGCCGTGC and R – 5′ CCACCGCCGCGTGCACAAACA, Figure 2A), gel-purified and cloned into pGEM-T vector (Promega, United States). Thirty-five positive clones were Sanger sequenced to verify successful edition of the target region. The sequences were aligned in UniPro uGene software (v1.29.0) (Supplementary Figure S2). For evaluation of CRISPR-edited transcripts, the total RNA was first converted into cDNA with SuperScript IV RT (Invitrogen, United States, cat. 18090200) and a 250 bp fragment including the sgRNA target region was PCR-amplified with primers containing overhang adapter sequences (F – 5′ TCGTCGGCAGCGTCAGATGTGTATAAGAGACAGAGGGTC ACTTTGAGCCCCGTG and R – 5′ GTCTCGTGGGCTCGGA GATGTGTATAAGAGACAGCTGGTAGTGGCAGAAGTCCC). The products were gel-separated and purified using Wizard SV Gel and PCR Clean-Up System (Promega, United States). The libraries were prepared with Nextera XT Index kit v2 (Illumina, United States) and deep-sequenced using paired-end reads (2 × 250 bp) in Illumina MiSeq System equipment at the Laboratory of Animal Biotechnology (ESALQ/University of São Paulo, Brazil). FastQ reads were submitted to quality analysis and mapped to the cScrt2 3′UTR in galGal5 reference genome using BWA v. 0.7.17 (Burrows-Wheeler Aligner; Li and Durbin, 2009). Transcript variants were determined using the R package CrispRVariants v.1.16.0 (Lindsay et al., 2016).

In situ hybridization was performed on stage HH22-23 embryos post-electroporation of miRNA sponges or CRISPR/Cas9 plasmids, as previously described (Acloque et al., 2008). The whole mount embryos were imaged with Nikon SMZ1500 stereomicroscope and then processed for gelatin-sucrose embedding (Bronner-Fraser et al., 1996). Cross sections with 25 μm of trunk neural tube were collected by cryosectioning (Fisher Scientifics, Waltham, MA, United States).

For immunohistochemistry, we used a rabbit IgG anti-GFP antibody (1:200, Molecular Probes, cat. A-6455) and Alexa 488 goat anti-rabbit IgG antibody (1:400, Molecular Probes, cat. A-11008). The anti-GFP antibody was used to detect the expression of citrine (Nagai et al., 2002). Stained sections were mounted in FluoroShield Mounting (Abcam, United States) and imaged with Zeiss Axio Imager.D2 coupled with an Axiocam 503 Color camera.

We used GraphPad Prism v.7 for statistical calculations. For 3′-UTR luciferase reporter assay, three control and three experimental embryos were assayed as technical triplicates and analyzed with unpaired Student’s t-test where p < 0.05 was considered as significant. For cScrt2 detection post-CRISPR/Cas9 editing, seven embryos were analyzed and the results were statistically evaluated with Student’s t-test.

To identify functional MREs in Scrt2 genes, we searched conserved responsive elements (REs) in the 3′UTRs of vertebrates. Despite the 3′UTRs sequence variations, we found an element containing overlapping REs for miR-125b and miR-200b that was conserved among amniotes (Figure 1A). The conservation of these overlapping sites in amniotes and the presence of miR-200b sites in orthologs of other jawed vertebrates suggests biological significance.

Most miRNAs repress gene function by binding to a specific sequence at the 3′UTR of a target mRNA to cause transcript degradation. Thus, we verified the ability of cScrt2 3′UTR to mediate modulation of mRNA levels by electroporating either a reporter chimera, where this portion of the UTR is downstream of luciferase (pmiR-GLO-cScrt2UTR), or a control plasmid (pmiR-GLO), into the neural tube of HH12 embryos. Both plasmids were co-electroporated with membrane GFP (pCDNA3.1-mGFP) to assess electroporation efficacy prior to tissue lysis. As shown in Figure 1B, our results indicated that cScrt2 3′UTR reduced luciferase production to 32% (a 3-fold reduction; p < 0.05) of levels produced in control embryos.

To investigate how miR-125b and -200b might affect endogenous cScrt2 transcripts, we looked at the effect of deleting their MREs in the cScrt2 3′UTR genomic region with CRISPR/Cas9 (Figure 2A). A scrambled sgRNA was used as control and edition by our MRE-targeted guide was verified by sequencing (Supplementary Figure S2). The transfected cells from the experimental and control neural tube sides of single HH23 embryos were isolated by FACS and analyzed with RT-qPCR. Six of seven embryos analyzed revealed a significant increase in Scrt2 levels (Figure 2B), suggesting that the REs for miR-125b and miR-200b can regulate cScrt2 levels post-transcriptionally. Deletion of these MREs with CRISPR/Cas9 also expanded the cScrt2 expression field in the neural tube (Figure 2C), phenotype observed both macroscopically (in the whole embryo – Figure 2C) and in histological sections, more pronouncedly in the dorsal domain. Electroporation of scrambled sgRNA elicited no phenotype (Figure 2D). Moreover, ablation of the MREs also disrupted the centrifugal movement of the electroporated cells, which became distributed throughout the center-peripheric layers of the neural tube. In control neural tubes, the electroporated cells were located toward the outer layers of the neural tube.

We also reduced miRNA availability with miRNA sponge constructs. These constructs produce RNAs that contain tandemly repeated MREs to miR-125b or miR-200b (Ebert and Sharp, 2010), to sequester each endogenous miRNA specifically and decrease their availability to act on endogenous targets, including cScrt2.

The miR-125b sponge caused a pronounced increase in the cScrt2 expression field (Figure 3A), such that transcripts were no longer restricted to the intermediate zone (IZ), and there was an increase in expression levels in dorsal root ganglia (DRG). The miR-200b sponge increased cScrt2 expression in the DRG in a similar manner, but only slightly expanded the neural tube expression field (Figure 3B). The efficiency of miRNA titration depends on both the abundance of the target and the endogenous miRNA levels (Mukherji et al., 2011). Thus, we hypothesized that the difference in sponge-generated phenotype could be the result of differences in miRNA levels. Our results show that the absolute copy number for miR-125b was 10-fold higher (5.67 × 10⁸) than miR-200b (5.44 × 10⁷) in the neural tube of HH22 embryos (Figure 3C).

Due to the close proximity of miR-125b and -200b REs (14 bp apart) and the unpredictable outcome associated with CRISPR/Cas9-mediated NHEJ (Williams et al., 2018), we could not guarantee that one sgRNA would only target one of the REs singly. Although our sgRNA target was centered at the miR-125b binding site (Figure 2A), our gDNA sequencing results showed that, in several events, both MREs were edited (Supplementary Figure S2).

To investigate if the same bias was observed in the edited transcripts, we used deep-sequencing to analyze the variations in the 3′UTR sequence of cScrt2 transcripts from CRISPR-edited embryos. From the 32.405 reads that were analyzed, 29.962 (92.4%) counts were non-variants and 2.443 (7.5%) were edited transcripts (Figure 4). Considering only the modified reads, the miR-125b RE was edited in 62% of them and the miR-200b RE was edited in 3.6%. Thus, these data indicate that most of the edited transcripts lacked the miR-125b target site. Further, it suggests that our data on CRISPR-mediated variation of Scrt2 expression is due to reduction of miR-125b action on Scrt2 transcripts.