A long non-coding RNA Leat1 mediates the hormone responsiveness of EfnB2 during male urogenital development

The novel long non-coding RNA (lncRNA) Leat1 is extraordinarily conserved in both its location (syntenic with EfnB2, an essential gene in anogenital patterning) and sequence. Here we show that Leat1 is upregulated following the testosterone surge from the developing testis and directly interacts with EfnB2, positively regulating its expression. Leat1 expression is suppressed by estrogen, which in turn suppresses the expression of EfnB2. Moreover, the loss of Leat1 leads to reduced EfnB2, resulting in a severe hypospadias phenotype. The human LEAT1 gene is also co-expressed with EFNB2 in the developing human penis suggesting a conserved function for this gene in urethral closure. Together our data identify Leat1 as a novel molecular regulator of urethral closure and implicate it as a target of endocrine disruption in the etiology of hypospadias.


Introduction
Leat1 associates with the EPHRINB2 protein both in vivo and in vitro 154 Given that Leat1 showed a predominantly cytoplasmic localization, we examined the possibility of it binding to 155 EPHRINB2 protein. Stable TM3 cell lines containing the Leat1a inducible construct were transfected with a 156 V5-tagged-EfnB2 cDNA. After cumate induction of Leat1a mRNA, we used the V5 antibody to precipitate the mesenchyme (Fig 4 H, i and ii), while no foci were detected with Leat1 sense probe ( Supplementary Fig 4). 167 Due to the low abundance of Leat1, this is in line with the amount of staining expected with a Leat1 probe. 168 Together, these data show that Leat1 binds the EPHRINB2 protein both in vitro and in vivo, consistent with its 169 predominantly cytoplasmic and membrane localization. The cytoplasmic localization of Leat1 appears to be 170 facilitated by the presence of a 12bp polyA sequence found at the 3' end of the transcript -as confirmed by 171 rapid amplification of cDNA ends (RACE) -that is present in the genomic DNA ( Supplementary Fig 1C). 172 Coincidentally, 12 adenine residues is the minimum required to bind polyA binding proteins 29 , to stabilize 173 mRNAs in the cytoplasm.  Fig 3B). This experiment was then repeated in 181 the presence of the cumate inducible Leat1a allele described above. There was a significant suppression of 182 mRNA from the endogenous EfnB2 locus in cells expressing both exogenous EfnB2 and Leat1 (Fig 4E).

184
Leat1 is suppressed by estrogen 185 We investigated if estrogen could affect EfnB2 and Leat1 expression in the developing GT. GTs were dissected 186 from wild type and Leat1-deficient embryos at E12.5 and cultured for 48h in hanging drop culture in the 187 presence of dihydrotestosterone to drive virilization (as previously described 31-34 ; controls), and with the 188 addition of estrogen (17β-ethinylestradiol; EE2). 189 At the end of the culture period, the tissues were snap frozen and gene expression levels examined by  PCR. Connective tissue growth factor (Ctgf), a known estrogen responsive gene in the penis 35,36 , was used as a 191 control to indicate a positive estrogenic response. Ctgf was significantly increased compared to controls in both 192 wild type ( Fig 5A) and Leat1 deficient GTs ( Fig 5B) cultured with EE2, indicating that the GTs maintained 193 their hormonal responsiveness. 194 In wild type GTs (Fig 5A), the addition of EE2 significantly decreased both Leat1 and EfnB2 expression. EfnB2 195 levels were reduced to around half that seen in wild type embryos, similar to that in the Leat1 deficient mice (which is sufficient to cause hypospadias 26 ) and similar to levels seen in the developing female GT (see GTs (Fig 5B). 199 200 LEAT1 is expressed in the human penis 201 As described above, LEAT1 is unusually conserved for a lncRNA. Orthologous sequences were readily detected 202 by shared homology across all mammals and always in synteny with EFNB2, including in the human genome 203 (Fig 1C and 1D). We next determined if LEAT1 was a functional gene in humans and capable of producing 204 mRNA transcripts in the human penis alongside EFNB2 37,38 . We examined LEAT1 and EFNB2 mRNA levels 205 in the transcriptomes of urethral plate epithelium (UPE) isolated from patients undergoing repair surgery for 206 mild hypospadias (Fig 6). Although the timing of tissue collection was after the window of urethra 207 internalization in humans, LEAT1 expression was detected in the UPE of the human patient samples ( Fig 6A). 208 Interestingly, LEAT1 expression levels were variable between samples and very low in 3 out of the 9 209 hypospadias patients. EFNB2 was also expressed in the human UPE, as it is in mice, and showed variable levels 210 across patient samples, however, there was no correlation between LEAT1 and EFNB2 expression levels (Fig   211   6B). These data demonstrate that the human LEAT1 locus produces an mRNA transcript that is expressed in the 212 urethral plate of the penis alongside EFNB2, as in mice.

216
Leat1 is a novel, hormonally regulated lncRNA which binds to and facilitates EfnB2 function. Deletion of Leat1 217 leads to a suppression in EfnB2 expression and a complete lack of urethral closure resulting in hypospadias. In 218 addition, exogenous estrogen supresses EfnB2 expression in a Leat1 dependent manner. Finally, we show that 219 LEAT1 was conserved in humans and produces a mRNA that was co-expressed alongside EFNB2 in the human 220 penis. Together, our in vitro and in vivo data identify a new, hormonally regulated driver of urethra 221 internalization that sits at the interface between the genome and the environment in the development of 222 hypospadias.

224
Leat1 binds to and regulates EfnB2 and is required for urethral closure 225 We have shown that Leat1 mRNA physically binds the EPHRINB2 protein both in vitro and in vivo in the 226 developing penis. This is consistent with the predominantly cytoplasmic localization of Leat1 and its functional 227 poly-A tail. EPHRINB2 sits in the plasma membrane 17 and has a large intracellular tail which is the only region 228 to which cytoplasmic Leat1 could bind. To explore how Leat1-EPHRINB2 binding might affect EfnB2 gene 229 regulation we examined the impact of this interaction on EfnB2 autoregulation. In the presence of Leat1, 230 exogenous EfnB2 overexpression was able to suppress transcription of the endogenous EfnB2 gene. However, in 231 the absence of Leat1, EfnB2 was unable to autoregulate. Therefore, Leat1 likely mediates Efnb2 autoregulation 232 in the developing penis where both genes are co-expressed.

233
Loss of Leat1 in mice caused a complete lack of urethral closure, observable from birth and persisting through 234 to adulthood. The location of the urethral meatus at the base of the penis in homozygous mutant mice, in 235 conjunction with the unfused scrotal bulges, reduced AGD and dorsal foreskin hood, corresponds to the clinical 236 classification of a severe hypospadias phenotype in humans (Fig 2B)  Leat1 is a potential target of endocrine disruptors in the etiology of hypospadias 246 Leat1 expression was sexually dimorphic and peaks in the male GT at E13.5, directly after the onset of 247 androgen production from the fetal Leydig cells 43 . We next investigated the impact of estrogen on Leat1 and 248 EfnB2 expression in the developing GT during the window of urethral closure. Male GTs cultured in the 249 presence of DHT showed normal urethral colure and normal upregulation of Leat1. In contrast, GTs exposed to 250 estrogen and DHT showed a significant reduction in the expression levels of both Leat1 and EfnB2 to around 251 half of that seen in the normal GT. This was equivalent to Leat1 and EfnB2 expression levels in the female GT 252 (Fig 1B, Supplementary Fig 3A) and an equivalent level of suppression of EfnB2 to that seen in the Leat1 null 253 mutant in vivo ( Fig 3C). Furthermore, our GT culture system demonstrates that Leat1 (and EfnB2) are directly 254 supressed by estrogen in the developing penis itself. This suggests EDCs can directly target genes regulating 255 urethral closure in the penis to cause hypospadias, rather than indirectly impacting androgen output from the 256 developing testis 44 . In addition, we showed that the suppression of EfnB2 by estrogen was Leat1 dependent.

257
EfnB2 was not supressed by estrogen in the developing GTs from Leat1 null mutants.

258
Estrogens and estrogen-mimicking endocrine disruptors such as BPA and genistein are well known to increase 259 the risk of hypospadias in mice and humans 11,45,46 . However, the molecular targets of these chemicals are   Fig 5). In each case, Leat1 was located in an identical 272 genomic location syntenic with EfnB2 ( Fig 1D). Marsupials last shared a common ancestor with mice and 273 humans 160 million years ago 50 indicating an extremely conserved function for Leat1 in mammals that is both 274 location and sequence dependent. Given that Leat1 can bind the EPHRINB2 protein, we suggest that the high 275 degree of nucleotide conservation may be important for mediating this interaction. Despite being syntenic with EfnB2, Leat1 overexpression in trans can still drive EfnB2 upregulation, indicating that Leat1 can regulate 277 EfnB2, even outside of its genomic context 51 . 278 We demonstrated that LEAT1 is a functional gene in humans, producing an mRNA that was expressed in the 279 developing penis alongside EFNB2. Interestingly, there was variable LEAT1 expression in the UPE of humans 280 with mild hypospadias and an almost complete absence in one third of the cases examined (Fig 6). This, 281 together with the sequence conservation of LEAT1, suggests a conserved interaction with EPHRINB2 to 282 mediate urethra internalization and implicates it as a potential candidate gene in human hypospadias.

284
A model for Leat1 function in urethra internalization 285 We have shown that Leat1 binds to the EPHRINB2 protein in vitro. We further demonstrate that Leat1 internalization, establishes EPHRINB2 autoregulation leading to expression of EfnB2 mRNAs to the level 290 necessary for urethra internalization ( Fig 1B, Fig 3C, Fig 7A). A loss of Leat1 expression reduces EPHRINB2 291 feedback, downregulating EfnB2 mRNA expression. This, in turn, prevents urethral closure, resulting in 292 hypospadias. Exposure to exogenous estrogen during this developmental window can reduce Leat1 expression, 293 leading to inhibition of EfnB2 mRNA below required levels resulting in hypospadias ( Fig 7B). Together these 294 data demonstrate an essential regulatory mechanism required for normal urethral closure, and one which can be 295 impacted by exogenous estrogen to cause hypospadias in both mice and humans. In conclusion, we have identified a novel lncRNA Leat1 that regulates urethral closure in mice. Leat1 regulates 299 the function of its neighboring gene EfnB2 by affecting its autoregulatory feedback through protein-RNA 300 interactions. This relationship between Leat1 and EfnB2 was demonstrated both in vivo and in vitro. Leat1 301 shows extraordinary sequence conservation for a lncRNA, indicating conservation of domains we propose 302 mediate its binding to EPHRINB2. We further show that Leat1 expression was supressed by estrogen and that 303 this mediates the suppression of EfnB2. Importantly, we showed that LEAT1 and EFNB2 are also co-localised in  To produce TM3-PiggyBac Leat1a/pcEfnB2 stable cell lines, a TM3-PiggyBac Leat1a cell line was transfected 336 with 1 µg of pcDNA-V5-EfnB2 using Fugene 6 transfection reagent (Promega, Sydney, Australia). 24h after 337 transfection culture media was supplemented with 0.4 µg/ml Geneticin (Life technologies, Sydney, Australia) to select for positive clones. When indicated cumate treatment was performed for 48h at a concentration of 30 339 µg/ml (1X) in culture medium.

340
Genome mutation analysis 341 We mapped the site of the transgene insertion in the mutant mice to chromosome 8 using standard methods 25 .

342
The transgene insertion event caused a 50kb deletion in a coding-gene deficient region and was located 343 approximately 300kb downstream from the EfnB2 gene ( Supplementary Fig 1H. We used genomic PCR to 344 verify the transgene insertion boundaries and show that the associated deleted region was specific to our mutant 345 line compared to the background FVB/NJ mice ( Fig 1I). The transgene boundaries were identified using inverse 346 PCR according to published methods 25 . Genomic DNA was extracted and purified using phenol-chlorophorm-  EfnB2 ORF was amplified by PCR from E14.5 mouse genital tubercle cDNA using primers ClmEfnB2F and 372 ClmEfnB2R and subcloned into pcDNA3.1D/V5-His-TOPO® vector. pGem-EfnB2 was generated using the 373 pGEM®-T Easy Vector Systems kit (Promega, Sydney, Australia) following manufacturer instructions. A 374 736bp fragment of EfnB2 was amplified from E14.5 mouse genital tubercle cDNA using primers ClimEfnB2F 375 and ClimEfnB2R and subcloned into pGEM®-T Easy vector. All plasmids were sequenced by the CTP Sanger

384
In vitro transcription 385 pGem-mEfnB2 plasmid was used as template for PCR amplification using T7 and SP6 primers (Supplementary   386   table 1). The amplicon was purified using the QIAquick PCR Purification kit (Qiagen) and used as a template 387 for transcription and labeling using the DIG RNA labeling Kit (SP6/T7) (Roche, Sydney, Australia). SP6 388 transcription produced antisense probe whereas T7 transcription produced sense probe (control).

389
Whole mount in situ hybridization 390 Mouse embryos collected at E14.5 were fixed in 4% paraformaldehyde (PFA) in PBS for 24 hours at 4°C.

391
Whole mount in situ hybridization (WISH) was carried out as described previously 52 . Probe signal detection 392 was performed using an anti-digoxigenin antibody from sheep, conjugated with alkaline phosphatase (Roche).

393
Imaging was performed on an Olympus SZX16 microscope equipped with a Nikon DS-Fi2 Digital Sight 394 Camera and using NIS Element software (Nikon).  RNAseq data was assessed for quality using FastQC 461 (http://www.bioinformatics.babraham.ac.uk/projects/fastqc). Reads were aligned to the hg38 Human genome 462 using the Subread aligner. Ten bases were trimmed from both read ends prior to alignment and only uniquely mapping reads were returned from the alignments. Read counts were performed using featureCount available 464 from the Subread package for R. Only primary alignments with a mapping quality greater than 10 were 465 considered. LEAT1 counts in human were performed by extracting reads aligning to the LEAT1 locus on 466 Chromosome 13 using Samtools. The resulting counts were concatenated to the featureCount data prior to 467 normalization. For within-sample expression quantification, counts per million (CPM) were calculated using the 468 edgeR package. The full LEAT1 transcript length is unknown in humans so other length correction 469 normalization methods were considered inappropriate.

471
RNA protein immunoprecipitation was performed using a Magna RIP kit (Millipore, Sydney, Australia).     (pm). b) Foci within the urethra are non-specific staining (d, distal; pr, proximal; u, urethra). Mouse, Human and Wallaby Leat1 nucleotide sequences were performed using T-COFFEE 56 .