Identification of transcripts derived from the Skt/Etl4 locus
To identify the gene detected by the Etl4lacZ insertion we isolated and sequenced overlapping BAC and PAC clones of the genomic Etl4 region on mouse chromosome 2 (prior to publication of the mouse genome sequence and identification of the Skt gene). Within this genomic region we identified two cDNA clones, mpm09263 and mbg07236, from the Kazusa Mouse cDNA Project [www.kazusa.or.jp/rouge [22], which contained exons flanking both, the Etl4lacZ and the SktGt insertions identified by Semba et al [20]. Both cDNA clones extensively overlap with other Skt/Etl4 mRNA sequences (Fig. 1a), and contain two additional so far unknown exons (red lines in Fig. 1a) located around 360 kb and 168 kb upstream of the thus far known most 5’exon in the genomic region. These cDNA clones contained 21 exons that are distributed over approximately 871 kb of genomic sequence (Fig. 1a), and extend the genomic region of the Skt/Etl4 gene listed in GenBank entries at NCBI (Fig. 1a). The longest open reading frame starts in exon 4 and ends in exon 21 (Fig. 1b). According to this gene structure the Etl4lacZ and the SktGt insertions occurred in intron 4 and 15 of the gene, respectively.
To verify that the newly identified exons 1 and 2 are indeed part of the Skt/Etl4 gene we analysed poly (A+) RNA isolated from adult brain and stage E9.5 embryos by RT-PCR using primers pair located in exon 1 and 3 (Fig. 1b). We obtained the expected PCR-Fragment (validated by sequencing, data not shown) from adult brain mRNA (Fig. 1c lane 1), but not from E9.5 mRNA (Fig. 1c lane 2). Furthermore a probe encompassing exon 1–3 hybridized to the same transcripts of around 7 and 6 kb as probes specific for exons 5–8, exons 9–13 and exons 15–18 (Fig. 1d) in Northern blot hybridizations of poly (A+) RNA from adult testis, supporting that the newly identified exons contained in mbg07236 are part of the Skt/Etl4 gene. In addition to the 7 and 6 kb transcripts detected in testis we observed an approximately 1 kb transcript in spleen with the exon 5–8 probe (Fig. 1d, Sp), an approx. 3 kb transcript in kidney with the exon 9–13 probe (Fig. 1d, Ki) and an approx. 4 kb transcript in testis with the exon 15–18 probe (Fig. 1d, Te), suggesting the generation of complex tissue-specific transcripts potentially derived from different promoters. In mRNAs from various embryonic stages we detected two transcripts of around 8 kb and 6 kb (Fig. 1e) with the exon 5–8 (Fig. 1e) but not with the exon 1-3 probe (not shown), suggesting that exon 1–3 are specific for testis (and brain as detected by RT-PCR) expressed transcripts of Skt/Etl4. Consistent with the Northern Blot results whole-mount in situ hybridizations (WISH) on E9.5 embryos with the exon 5-8 (n = 30) but not with the exon 1–3 (n = 10) probe detected expression in E9.5 embryos (Fig. 1f a, b).
Expression of Skt/Etl4 mRNA
To get a comprehensive picture of the expression pattern and to detect expression domains that might have been missed in previous experiments using the lacZ reporter gene in Etl4lacZ or SktGt mice [19, 20] we performed in situ hybridizations with a Skt/Etl4 specific probe. Consistent with the β-galactosidase staining pattern of the lacZ alleles [19, 20] we found expression in the notochord at similar levels along its entire length by WISH of E8.5 to E8.75 embryos (Fig. 2a, b). However, beginning at E8.75 up to approximately E11.5 Skt/Etl4 expression in the notochord appeared graded with high expression in the caudal region (Fig. 2c-g). Furthermore we detected Skt/Etl4 expression within the otic placode (Fig. 2c), the branchial arches (Fig. 2c) and the AER of the fore- and hindlimbs (Fig. 2e, f).
In order to identify additional tissues expressing Skt/Etl4, we performed section in situ hybridizations (SISH) with a Skt/Etl4 specific probe on transverse sections of stage E15.5 wt embryos (n = 3) when most of the vital organs are present or developing. In E15.5 embryos when notochord cells are incorporated into the intervertebral discs Skt/Etl4 expression persisted in the developing NP and AF throughout the whole vertebral column (Fig. 2i and j, [20]). In addition to the expression in the emerging IVDs we verified the expression in the kidney (Ki, Fig. 3a, b) with strong signals within the cortical region presumably in developing nephrons (Ne, Fig. 3a, b). Also the epithelium of the ureter (Ur, Fig. 3a), the mesonephric duct (MeDu, Fig. 3b) and the bladder (Bl, Fig. 3c) were positive for Skt/Etl4. In testis expression was strongest within the cortical region (Te, Fig. 3b) and less within the outer region of the seminiferous tubules (SeTu, Fig. 3b). We additionally identified tissues were Skt/Etl4 expression was not known hitherto: the epithelium of the lung (Lu, Fig. 3d, e) the bronchus (Br, Fig. 3e), as well as the epithelium of the cochlea (Co, Fig. 3f), the tympanic cavity (TyCa, Fig. 3f) and the nasopharynx (NaPh, Fig. 3g). In the developing eye we found that in addition to the lens (Le, Fig. 3h, [20] and inner layer of the retina [20], Skt/Etl4 also is expressed within the optic nerve (OpNe, Fig. 3h). The primordia of follicles of vibrissae associated with the upper lip (FoVi, Fig. 3j) as well as the ducts of the submandibular gland (SuGl, Fig. 3i), the developing thymus gland (ThymGl, Fig. 3k) and thyroid gland (ThyrGl, Fig. 3k) display Skt/Etl4 expression. Expression was also detected within the pancreatic primordium most likely within the exocrine acini (Pa, Fig. 3l) and the wall and epithelium of the gut (Gu, Fig. 3l).
Targeted insertion of lacZ reporter genes into Skt/Etl4 exons 1 and 5
In both mouse lines, Etl4lacZ and SktGt, the insertion of a lacZ transgene resulted in similar very mild phenotypic changes [19, 20]. Both integrations occurred within introns (Fig. 1a) and in the SktGt allele a truncated protein of 998 aa fused to β-galactosidase could potentially be generated from the mutated locus [20], raising the possibility that these mutations represent hypomorphic alleles. To evaluate the phenotype of the complete loss of the gene we set out to generate a bona fide null allele of Skt/Etl4.
The presence of various transcripts presumably expressed from different promoters [Fig. 1 d and e; and 20], and the large genomic distance between many coding exons (Fig. 1a) precluded a simple targeting strategy to ensure the complete elimination of Skt/Etl4 function. Therefore, we decided to introduce lacZ reporter genes with strong transcriptional termination signals (triple poly (A) [23] and loxP sites into two different regions of the Skt/Etl4 locus that are parts of differentially expressed transcripts. The termination signal should prevent the expression of further downstream sequences from a given promoter, the loxP sites should allow us to delete portions of the gene by site directed recombination. Moreover the integration of the lacZ reporter into different exons should allow us to readily examine the transcriptional activity of the gene from different transcriptional start sites.
We introduced an IRES driven lacZ reporter gene fused to triple poly (A) into the most 5’exon (SktEx1IRESLacZ allele, Fig. 4a b) to disrupt the transcript specifically detected in brain and testis (Fig. 1c, d). Likewise we introduced a lacZ triple poly (A) cassette into exon 5 (SktEx5LacZ allele, Fig. 4a c), to disrupt transcripts expressed during embryogenesis. Correctly targeted ES cells were used to establish mouse lines carrying the SktEx1IRESLacZ and the SktEx5LacZ alleles (Fig. 4f, and data not shown). The presence of exon 1 containing transcripts was analysed by lacZ staining of embryos and adult tissues of mice carrying the SktEx1IRESLacZ allele. Consistent with RT-PCR and WISH results (Fig. 1c and f) we found no lacZ activity during embryogenesis (data not shown). LacZ staining of adult SktEx1IRESLacZ organs (brain, testis, epididymis, skeletal muscle, spleen, heart, salivary gland, small intestine, stomach, kidney, liver, lung) revealed only specific lacZ staining in the testis (n = 2) and the epididymis (n = 2; Additional file 1: Figure S1 A, and data not shown). The staining results confirm the Northern blot results that we obtained with the exon 1-3 probe (Fig. 1d) but are at odds with the RT-PCR results obtained with brain mRNA, where exon 1 potentially is expressed at low levels. Homozygous mice carrying the insertion of the lacZ-stop cassette into exon1 were viable and fertile and had no obvious skeletal phenotypes (data not shown) indicating that disruption of Skt/Etl4 in its 5’most exon did not affect the function during vertebrae development.
With the SktEx5LacZ allele we observed specific lacZ reporter gene expression in stage E10.5 (n = 6) and E11.5 (n = 10) embryos in the notochord (white triangles in Additional file 1: Figure S1 B b, c, e and f), the optic vesicle, the otic placode, the eye, the AER of the fore- and hindlimbs and on the surface of the branchial arches (Additional file 1: Figure S1 B b, c, e, f). This expression pattern was nearly identical to endogenous Skt/Etl4 expression detected with the exon 5 to 8 probe in whole mount in situ hybridizations (Fig. 2e-g) and similar to Etl4lacZ embryos of the same age [19]. Similar to SktEx1IRESLacZ mice homozygous SktEx5lacZ mice were viable and fertile without obvious external phenotype (data not shown). To test if the termination signal 3’to the lacZ insertion in exon 5 prevents transcription of downstream exons we analysed the presence of exon sequences upstream and downstream of exon 5 in Skt/Etl4 mRNA by RT-PCR using RNA from wt, heterozygous and homozygous SktEx5LacZ mutant E10.5 embryos (for primer positions see Additional file 2: Figure S2A). As expected exons 4 and 5 upstream of lacZ were detected in wt and mutant embryos (PCR1, Additional file 2: Figure S2 B), as well as a Skt/Etl4-lacZ fusion transcript in embryos containing the SktEx5LacZ allele (PCR2, Additional file 2: Figure S2 B). RT-PCR with primers binding to exon 4 and 7 detected a fragment in which exon 4 was fused to exon 6 revealing an alternative splicing event, which removes the targeted exon 5 (PCR3, Additional file 2: Figure S2 B). In addition, further downstream sequences were still expressed in mutant embryos (PCR4, Additional file 2: Figure S2 B). Removal of exon 5 causes an in-frame deletion of 96 of the total 1373 amino acids. Thus, most likely, this allele does not represent a null allele and therefore was not analysed further.
Deletion of the whole locus to eliminate Skt/Etl4 gene function
Since the SktEx5LacZ allele did not prevent the generation of Skt/Etl4 transcripts that can give rise to likely functional protein (s) we set out to remove the vast majority of the coding region using a two step deletion strategy based on the loxP sites present in the targeted alleles and an additional targeting event into exon 21 (outlined in Fig. 4a, b).
During the generation of the SktEx1IRESLacZ and SktEx5LacZ alleles a floxed PGK-Neo cassette was introduced in both cases 3’to the lacZ reporter. After removal of the neo gene by Cre-mediated recombination a single loxP site (black triangle in Fig. 4a b, c) in the same orientation remained in the SktEx1IRESLacZ and SktEx5LacZ alleles. We used these loxP sites in combination with Cre expression during oogenesis to delete 595 kb of genomic DNA between exon 1 and exon 5 by targeted meiotic inter-chromosomal recombination [TAMERE 21]. Heteroallelic SktEx1IRESLacZ/Ex5LacZ females carrying a ZP3::Cre transgene were mated with wild type males and offspring analysed for the presence of the inter-chromosomal recombination event (Fig. 4a d SktΔEx1-5 allele, and 4b). Among the first 20 offspring we identified a female with the desired deletion by PCR analyses using four different primer pairs that amplified DNA fragments specific for the various alleles (Fig. 4c, Primer pair position in Fig. 4a). Successful inter-chromosomal recombination was indicated by PCR products obtained with PCR 1 and 4 and lack of products with PCR 2 and 3, which differentiates between the deletion and the presence of the initial SktEx1IRESLacZ and SktEx5LacZ alleles (Fig. 4c). The correct recombination event that removes the first 4 exons of the gene was validated by Southern blot analysis, which showed the expected polymorphisms between the three alleles (Fig. 4e). Mice homozygous for the 595 kb SktΔEx1-5 deletion were viable and fertile and did not show any obvious phenotype (data not shown), indicating that the portions of Skt/Etl4 essential for vertebral development were still functional.
To delete the major part of the coding sequence contained in further 273 kb of genomic DNA we generated ES cells from homozygous SktΔEx1-5 blastocysts, targeted exon 21 by introducing a floxed PGK-Neo cassette (SktΔEx1-5; Ex21GFP allele, Fig. 4a e), and generated mice carrying this allele. Intrachromosomal recombination (deletion of exons 5–20) was achieved in female mice harbouring the SktΔEx1-5; Ex21GFP allele together with the ZP3::Cre transgene. Offspring carrying the deletion were identified by PCR analyses with 4 different primer pair combinations (Fig. 4d, Primer pair positions in Fig. 4a). Only mice with the second genomic deletion of 273 kb between exon 5 and 21 (Fig. 4b) generate PCR products with PCR1, 5 and 6 in combination with lack of a product with PCR4, which distinguishes the second deletion and the initial SktΔEx1-5 and SktΔEx1-5; Ex21GFP alleles (SktΔEx1-5 allele, Fig. 4a f and d). Southern blot analysis using a lacZ specific and a genomic probe located downstream of exon 21 (3’Exon21) and three different restriction digests of genomic mouse DNA (Fig. 4e and Additional file 3: Figure S3) confirmed successful deletion of exons 1 to 20 of Skt/Etl4. Despite the deletion of about 88 % of the N-terminal portion of the protein homozygous SktΔEx1-20 mice were viable and fertile.
To test whether the truncated C-terminal portion of the protein comprising 167 AA encoded by exon 21 is generated we performed Western blot analysis with a polyclonal SKT/ETL4-specific antibody directed against a C-terminal peptide (anti-NGS, see Materials and Methods), which detects SKT/ETL4 protein expressed in CHO cells with the expected size of approximately 150 kDa (data not shown). In protein lysates of adult wild type brain we detected in addition to several other cross-reacting proteins a 200 kDa and a 150 kDa protein species, which were not present in lysates of homozygous SktΔEx1-20 mice (Fig. 4f, red triangles in the left panel) and likely represent two variants of the SKT/ETL4 protein. We did not see any truncated version of the SKT/ETL4 protein at the predicted approximate size of 16,4 kDa in brain lysates of homozygous SktΔEx1-20 mice in a Western blot after separation of proteins by SDS-PAGE in higher percentage gels (Fig. 4f, right panel), indicating that no truncated SKT/ETL4 protein is present in these mice. Thus, the established the SktΔEx1-20 mouse line carrying a 868 kb deletion that removes about 90 % of the N-terminal coding sequence should represent a bona fide null allele of the Skt/Etl4 gene.
Sickle tail null mice display malformations of caudal IVDs
External observation (at least n = 12) and skeletal preparations (n = 3) revealed that most of the adult homozygous mutant animals displayed kinks within the tails (Fig. 5a) similar to the phenotype described for Etl4lacZ [19] and SktGt [20] mice. For histological analysis paraffin sections of tails from 3-week old wt (n = 4) and homozygous Skt/Etl4 mutant mice (n = 6) were stained with Haematoxylin-Eosin (HE, Fig. 5b). We found aberrations in the morphology of the intervertebral discs (IVDs) that arose mostly in caudal vertebrae as shown in two examples of SktΔEx1-20 tails (Fig. 5b c-f). Normally in wt mice the NP of the IVDs is centrally located (Fig. 5b a and a’, b and b’), which we also observed in IVDs of SktΔEx1-20 mutants (Fig. 5b e and e’). However, in several SktΔEx1-20 mutant IVDs the NP was shifted to the periphery (arrowheads Fig. 5b c, d and higher magnification in c’, d’) or in rare cases not present at all (arrowhead in Fig. 5 b f and higher magnification in f’) in homozygous SktΔEx1-20 mutants. In addition the fibrous layer of the AF surrounding the NP was reduced in size or not present at all (Fig. 5b c’) resulting in some cases in the direct contact of adjacent vertebral bodies. In more anterior IVDs (Fig. 5c, upper caudal vertebrae) we found only a slight lateral shift of the nucleus pulposus in one IVD of one of four analysed mutant animals (Fig. 5c d and higher magnification in d’) but no other abnormalities compared to wt (n = 2). In IVDs of the sacral region no structural alterations of the NP or AF in homozygous mutant (n = 4) mice were detected (Fig. 5c, sacral vertebrae). Together our results highly correspond with the phenotype described for the Etl4lacZ and SktGt mutants, where only defects in size and position of the NP and AF, and corresponding vertebral malformations were observed caudally, but not in other regions of the vertebral column [19, 20].
Skt/Etl4 null mice do not display major defects in the notochord
During mouse embryogenesis between E12 and E13 notochordal cells in the region of the future disc transform and form the later NP and AF. The appearance of the previously described abnormalities in caudal IVDs of Skt/Etl4 deletion mutants may arise from defects in the formation or differentiation of the notochord. To analyse if the Skt/Etl4 deletion leads to obvious notochord defects early during development we hybridized stage E9.5 embryos with probes for the notochordal markers Sonic hedgehog [Shh 3] and Brachyury [T 24]. We observed no differences in the expression pattern of Shh in wt (n = 7) and in homozygous Skt/Etl4 (n = 5) deletion mutant embryos (Fig. 6a, b). Likewise expression of T at embryonic stage E9.5 was indistinguishable between wt (n = 7) and Skt/Etl4 (n = 6) mutants with the exception of an ectopic expression domain at the forelimb level in one out of six analysed mutants (red arrowhead Fig. 6d) indicating that the loss of Skt/Etl4 expression does not have a major impact on early notochord development.
Histological analysis of adult Skt/Etl4 mutants
As described earlier Skt/Etl4 is a gene expressed in multiple tissues during embryogenesis. Therefore, we analysed various tissues of juvenile Skt/Etl4 deletion mutants for histologically detectable abnormalities. In HE stained longitudinal kidney sections we found no obvious differences in the overall appearance (Fig. 7b) or structure of the cortex (Fig. 7 c, d), the medulla (Fig. 7 e, f) and papilla (Fig. 7 g, h), or the number of glomeruli (data not shown) between wild type (n = 2) and mutants (n = 2). Other organs that exhibited a thus far not described expression of Skt/Etl4 were epithelia of the lung and the cochlea. We analysed lung tissue of 2 week old wt (n = 3) and homozygous SktΔEx1-20 (n = 2) mice with HE staining and could not detect any obvious difference in the tissue structure between both phenotypes (compare Fig. 7i, k, m with j, l, n). Likewise cochleae from the same stage did not exhibit any obvious variations between wt (n = 2) and homozygous SktΔEx1-20 mice (n = 2; compare Fig. 7o, q, s with p, r, t).