A new recessive autosomal mutation named "compact axial skeleton" (cax), causing kinky and shortened tails and shortened body length, arose spontaneously in a breeding colony of C3H/HeJ mice at The Jackson Laboratory. On the C3H/HeJ background the cax mutation causes severe shortening of the tail and body axis, whereas on the mixed genetic background of our linkage cross the phenotype was variable, and ranged from normal external appearance to almost complete absence of the tail and shortened body axis (Figure 1A–C). Most likely the variability seen in the backcross progeny is attributable to strain background effects. Skeletal preparations showed that even externally normal mice carried mild malformations of the vertebral column (Figure 1D, and data not shown). Skeletal malformations were found along the entire length of the vertebral column (Figure 1E) and included fused ribs (arrowheads in Figure 1F), reduced or missing pedicles (arrows in Figure 1G) and split or hemi-vertebrae (asterisks in Figure 1D, F, G). At earlier stages, mutant embryos had clearly discernable somite borders, however, in some embryos somite size varied considerably (Figure 1H, I, and data not shown). The observed defects are indicative of defective somite formation and compartmentalization suggesting that the cax mutation affects a gene involved in early somite patterning.

To identify the gene affected by the cax mutation we mapped cax genetically by linkage analysis of the mutant phenotype with segregating chromosome markers. First, cax was assigned to chromosome 2 in an interval between the microsatellite markers D2Mit22 and D2Mit409 by analysis of 62 mutant F2 offspring from an intercross of (C3H/HeJ-cax/cax × CAST/Ei+/+) F₁ hybrids. A fine genetic map was established by analysis of 1339 N2 mutant progeny from a backcross of F1 hybrids to cax/cax mutants. The locus order and inter-locus distances (in cM ± SD) for the candidate gene region deduced from our analysis is D2Mit22-(0.4 ± 0.2)-D2Mit140-(0.3 ± 0.2)-D2Mit309-(0.8 ± 0.2)-[D2Mit195, cax]-(0.5 ± 0.2)-D2Mit286-(0.6 ± 0.2)-D2Mit262-(2.1 ± 0.4)-D2Mit409 (Figure 2A). Analysis of the mouse genome sequence showed that D2Mit195 (Chr 2 position 153.082 Mb, NCBI build 36), which did not recombine with cax, resides approximately 120 kb downstream of the Pofut1 gene (Chr 2 position 152.933–152.962 Mb; Figure 2B). Pofut1 appeared as an appealing candidate since O-fucose modification of Notch is essential for receptor function 1819, and Notch signalling is required for normal somite formation and patterning 28. To directly test whether cax affects Pofut1 we crossed homozygous cax mutants with heterozygous mice carrying a targeted null allele of Pofut1, Pofut1^tm1Pst19. Mice carrying one copy of the cax mutation and one Pofut1^tm1Pst allele had a significantly shortened body axis and axial skeleton defects (Figure 2C c, d), indicating that the recessive Pofut1 null allele does not complement the cax mutation and identifying cax as a novel allele of Pofut1, which we refer to as Pofut1^cax from hereon.

To address whether the Pofut1^cax mutation affects the Pofut1 coding sequence, we amplified Pofut1 cDNA from mRNA purified from homozygous Pofut1^cax kidney and C3H wild type mice, and sequenced at least two independently generated cDNA clones. No mutation in the coding sequence was detected (data not shown), suggesting that Pofut1^cax affects Pofut1 transcription. Consistent with this idea, a probe from the coding region revealed by in situ hybridization overall reduced levels of Pofut1 transcripts in a considerable portion of homozygous Pofut1^cax mutant embryos (Figure 2D, b and 2d, and data not shown) that are also apparent in the PSM and developing somites (compare regions indicated by red lines Figure 2D, c and 2d).

To identify potential structural alterations at the Pofut1 locus that might cause the mutant phenotype we scanned 20 kb upstream of the ATG as well as the whole intragenic region of the Pofut1^cax allele by PCR. Primers were designed to amplify overlapping DNA fragments between approximately 500 and 1000 bp in length, and genomic DNAs prepared from two wild type C3H and two mutant Pofut1^cax/cax mice were used as templates. With the exception of one primer pair that consistently failed to amplify a fragment of the 3' end of intron 4 from mutant DNA (fragment 4/11 in Figure 3A) all PCR reactions amplified fragments of indistinguishable size from wild type and mutant DNA. This suggested that the Pofut1^cax mutation disrupts the integrity of the Pofut1 locus in this region. Consistent with this idea, Southern blot analyses using a cDNA probe that contained exons 5 and 6, and the 5' portion of exon 7 revealed restriction fragment length polymorphisms between wild type and mutant genomic DNA (Figure 3B, and data not shown).

The primers used to amplify fragment 4/11 are located in fragments 4/10 and fragment E5, and should therefore be present in mutant DNA (Figure 3A, lanes c, d, and k, l). Thus, we reasoned that an insertion into fragment 4/11 might prevent amplification of this fragment from mutant DNA by conventional PCR. Therefore, we employed long range PCR with fragment 4/11 primers, and amplified an approximately 6 kb fragment from mutant DNA (arrowhead in Figure 3C). Sequencing of this fragment identified an intracisternal A particle (IAP) insertion into Pofut1^cax intron 4 (Figure 3D). The presence of the insertion in the Pofut1^cax genomic DNA was independently verified by PCR reactions that specifically amplify the junction fragments (lanes c, d and g, h in Figure 3D). In E10.5 cax mutant embryos (n = 12) Pofut1 mRNA levels were reduced to approximately 25% of wild type (n = 6) levels (Figure 3E) as determined by TaqMan real time PCR. Western blot analyses with anti-POFUT1 antibodies 29 showed also reduced POFUT1 protein levels in cell lysates from Pofut1^cax embryos (Figure 3F). The protein reduction could not precisely be determined due to the normally low levels of endogenous POFUT1 in embryos and additional background bands. Thus, the Pofut1^cax allele carries an insertion of a retrotransposon that most likely underlies variably reduced Pofut1 mRNA and POFUT1 protein levels. This variable reduction might also explain the variable phenotype of cax mutant embryos and mice.

To analyze in more detail how the cax allele affects Notch activity and Notch-dependent processes we first analyzed expression of the Notch target genes Hes1, Hey1, and HeyL, which are severely down-regulated in Pofut1^{tm1Pst/tm1Pst}embryos (n = 3, respectively; Figure 4A d, h, l, p, t, x, zb, zf, zj). In cax mutant embryos expression of Hes1, Hey1, and HeyL was also obviously affected in the paraxial mesoderm and somites (Figure 4A f, r, zd). In contrast, expression in other domains, for example of Hes1 in the optic vesicle (compare Figure 4A i and 4j), and of Hey1 and HeyL in the branchial bars (compare Figure 4A u and 4v, and 4A zg and 4zh, respectively), was similar to wild type both with respect to expression levels and patterns. In the case of Hes1, the clearly visible stripes of expression in posterior somite compartments in wild type embryos (Figure 4A e) were not detected in cax mutants (n = 10; Figure 4A f)). Similarly, the strong distinct expression domains of Hey1 in posterior somite halves of wild type embryos (Figure 4A q) appeared fuzzy in cax mutants (n = 28; regions indicated by arrows in Figure 4A r), and expression in the PSM was reduced (Figure 4A r). In mutant embryos (n = 10) HeyL expression was reduced in the anterior PSM (red line in Figure 4A zd) and in newly formed somites, and expressed in a smaller expression domain in anterior somites (regions indicated by arrows in Figure 4A zd). The expression patterns of Hes1, Hey1, and HeyL in the paraxial mesoderm of embryos heteroallelic for the targeted null and the cax allele of Pofut1 were more severely disrupted (n = 3, respectively; Figure 4A g, s, ze), consistent with expected further reduced Pofut1 levels. In these embryos Hes1 expression in the optic vesicle (Figure 4A k) was not obviously affected, whereas Hey1 (Figure 4A w) and HeyL (Figure 4A zi) expression in the branchial bars appeared reduced, indicating that further reduction of Pofut1 levels in Pofut1^cax/tm1Pst embryos affects Notch activity also outside the paraxial mesoderm.

Consistent with obvious alterations of Hes1, Hey1, and HeyL expression in the paraxial mesoderm also the expression patterns of genes that are important or indicative for somite patterning and polarity, and whose normal expression patterns depend on Notch activity, were altered in Pofut1 mutants. Dll1 expression was virtually abolished in the somites but upregulated in the neural tube of Pofut1 null mutant embryos (n = 3; Figure 4B d). Similarly, Uncx4.1, which is normally expressed in a regular pattern delineating posterior somite halves (Figure 4B i), was severely downregulated in somites of Pofut1 null mutants (n = 5; Figure 4B l), and expression of Tbx18, which normally delineates anterior somite halves (Figure 4B e), was expanded throughout somites (n = 3; Figure 4B h) indicating loss of somite polarity.

In cax mutants anterior-posterior somite patterning was also abnormal, as indicated by the reduced and broadened Dll1 expression domains in somites (n = 11; arrows in Figure 4B b), fuzzy expression domains and disorganized stripes of expression of Tbx18 (n = 27) and Uncx4.1 (n = 10; arrows in Figure 4B f, j). In heteroallelic Pofut1^cax/tm1Pst embryos anterior-posterior somite patterning was more severely affected than in Pofut1^cax/caxembryos: the somitic Dll1 stripes were essentially lost (Figure 4B c), Tbx18 expression domains were fuzzy and expanded (Figure 4B g) and, Uncx4.1 stripes were irregular and scrambled (Figure 4B k). Similarly, expression of Cer1, Mesp2, and Papc was disrupted by the cax mutation further supporting the notion that somite compartmentalization is affected. Cer1, whose expression is normally restricted to the anterior somite compartments of the prospective and most recently formed two somites (Figure 4B m), was expressed in one broad domain (n = 25; red line in Figure 4B n) similar to Pofut1 null mutants (n = 2; Figure 4B p). Mesp2 which is normally expressed in one or two distinct stripes of variable width (Figure 4B q, and data not shown) showed always only one not clearly delineated expression domain in cax mutant embryos (n = 31; red line in Figure 4B r), and the distinct domains of Papc expression (Figure 4B u) appeared as one blurred domain in cax (n = 36; red line in Figure 4B v). In heteroallelic Pofut1^cax/tm1Pst embryos (n = 3, respectively), the expression patterns of these genes were similarly disrupted (Figure 4B o, s, w). In Pofut1 null mutants (n = 6) expression of these genes was severely downregulated in addition to their abnormal patterns (Figure 4B t, x).

Since the establishment of somite polarity depends on the function of cyclic Notch pathway genes in the PSM 30313233 we analyzed the expression of Hes7 and Lfng in cax mutant embryos. Hes7 could be assigned to the three described phases both in wild type and cax mutant (n = 26) embryos albeit expression in cax mutants was only detected after prolonged staining suggesting that levels are reduced (Figure 5). Likewise, Lfng expression patterns reflecting the different phases were found in cax mutants (n = 9; Figure 5). In contrast, in additional 24 embryos we observed rather broad domains of expression in the PSM (Figure 5m, n) that could not be assigned to certain phases, suggesting that reduced POFUT1 levels in cax mutant embryos impinge on the dynamic expression of cyclic Notch target genes in particular on Lfng. Collectively, these data suggest that the reduced Pofut1 mRNA levels in cax mutant embryos attenuate Notch activity below a threshold required for patterning the presomitic mesoderm, leading to defects in regular somite spacing and compartmentalization that underlie the observed skeletal malformations.

To address whether also other early processes regulated by Notch are affected by the cax mutation we first analyzed establishment of left-right asymmetry, the earliest reported developmental process requiring Notch activity in mice 3435. Loss of POFUT1 activity appears to completely block all early Notch signalling 1936. However, thus far a requirement for POFUT1 during establishment of left-right asymmetry has not been evaluated. Consistent with an absolute requirement of POFUT1 for Notch activity, Pofut1^{tm1Pst/tm1Pst} embryos (n = 6) showed defects in establishment of left-right asymmetry as indicated by the loss of Nodal expression (Figure 6C, F), which is directly regulated by Notch signalling 35. In contrast, none out of 25 Pofut1^cax/cax embryos showed abnormal Nodal expression (Figure 6B, E), and 50 analyzed heteroallelic Pofut1^Tm1Pst/cax embryos at E9.5 had normal turning and heart looping (data not shown), suggesting that POFUT1 activity derived from one hypomorphic allele modifies Notch signalling to a level sufficient for this process. Vascular remodelling is another early process that critically depends on Notch signalling 3738394041 and Pofut1^{tm1Pst/tm1Pst} embryos (n = 6) show severe vascular malformations 19. PECAM1 staining showed a highly irregular vascular network in Pofut1 null mutant embryos in all regions of the body (Figure 6O, R, and data not shown). In contrast, Pofut1^cax/cax embryos had a vascular network that was virtually identical to wild type embryos (Figure 6N, Q), except for occasional minor irregularities of the intersomitic vessels, which most likely arise secondarily to somite patterning defects. In addition, overall neuronal differentiation assessed by expression of neurofilament (NF165) as a pan-neuronal marker was not obviously affected by reduced Pofut1 expression (Figure 6T, V) except for irregular spacing and width of spinal nerves (Figure 6V), which also are most likely secondary to and indicative of disrupted somite patterning. Likewise, expression of NeuroD and Neurogenin1 was apparently unaltered in cax mutants (Figure 6H, K) whereas both genes were clearly upregulated in Pofut1 null mutants (Figure 6I, L) indicative of enhanced neuronal differentiation due to loss of Notch activity. Thus, establishment of left-right asymmetry, angiogenesis and neuronal differentiation proceed apparently normal in cax mutant embryos whereas Lfng expression appeared abnormal in most mutants and anterior-posterior somite patterning was consistently affected.

The recessive cax mutation arose spontaneously in a C3H/HeJ colony of mice at The Jackson Laboratory and has been maintained on this strain background. The inbred mouse strain carrying the mutation is designated C3H/HeJ-Pofut1^cax/J, Jackson Laboratory Stock Number 7782. Phenotypic analysis was performed with cax mutants on a mixed genetic background due to the low breeding performance of the inbred line.

Intersubspecific F1 hybrids were generated by mating C3H/HeJ-cax/cax mutants with CAST/Ei-+/+ mice. An intercross of these F1 hybrids produced 62 F2 mice with mutant phenotypes (cax/cax genotype) that were used to determine the initial map position of the cax mutation. A backcross of (C3H/BL6xCAST/Ei) F1 hybrids to C3H/B6-cax/cax mutant mice produced 1339 N2 mice with an unambiguous mutant tail phenotype that were analyzed for high resolution mapping. Initially, we mapped cax with respect to the flanking simple sequence length polymorphism (SSLP) markers D2Mit22 and D2Mit409 using all 1339 backcross animals. We identified 62 recombinants, which were genotyped for SSLP markers located between D2Mit22 and D2Mit409.

Whole mount in situ hybridization was performed according to 59. The probes used were originally obtained from Dr. Manfred Gessler (Hey1, HeyL), Dr. Martyn Goulding (NeuroD, neurogenin), Dr. Tom Gridley (Lfng), Dr. Ryoichiro Kageyama (Hes1, Hes7), Dr. Andreas Kispert (Uncx4.1, Tbx18), Dr. Janet Rossant (Cer1, nodal), Dr. Yumiko Saga (Mesp2), or isolated in our laboratory (Dll1, Pofut1, Papc).

Embryos were fixed in 4% PFA, dehydrated, embedded in paraffin, sectioned at 10 μm, stained with Nuclear Fast Red, and embedded in Mowiol (Calbiochem).

For immunohistochemical analysis embryos were fixed in 4% paraformaldehyde, and in case of NF165 staining subsequently treated with proteinase K. Primary antibodies used were anti-PECAM (Pharmingen) diluted 1:100, and anti-NF165 (DSHB clone 2H3) diluted 1:50. Secondary antibodies were biotinylated anti-rat and anti-mouse (Vectastain) diluted 1:200. The signal was intensified with the ABC system (Vectastain) and bound antibodies were detected with DAB (Sigma).

To search for mutations in the open reading frame of Pofut1 mRNA from C3H/HeJ-cax/cax or C3H/HeJ embryos was prepared either with magnetic Dynabeads (Dynal Novagen) or with a Direct mRNA kit (Qiagen). cDNA was synthesised with Superscript II Reverse transcriptase (Invitrogen or Promega). cDNA was amplified in two overlapping fragments using the following gene specific primers: for Exon 2-Exon4: CTG CTT CTG CTG CTG TTG CTG C and CAG TGC GAG CAC AGG ATG CTC, and for Exon 5-Exon7: CAA TGG ACC CAG AGA TTT CCT GCA and GGT TGA GGG TGG GAG GTG GG. Exon1 was amplified from genomic DNA with the primers GCC ATT GTG CGG TGC ATT G and AAG CAG AGG GTT CCG GAG GC. PCR fragments from at least two independent PCR reactions were subcloned either into TpGEM easy or pCRII Topo vectors and sequenced.

To identify gross abnormalities of the Pofut1 gene about 20 kb of the promoter region, the introns and the untranslated 3'region were amplified by PCR as overlapping fragments of about 0.5 to 1 kb in length. Primer sequences were selected based upon http://www.ensembl.org. Primer sequences used to amplify the fragments shown in the figures are as follows: fragment 4/10: TCCATTTTGCCCTTTCAAAGGT and ACACAGAATCCTTTCTGCAATCTTTC; fragment 4/11: GCACTGCCACTGGGGCTAGT and CCCAGGCAGTGCGAGCA; fragment E5: AGATTTCCTGCAAAAGAGCATCCT and GAGCTAAAATCCAGACTTGGTGGA; fragment 5/1: CATTCATCTGCGCATTGGCT and AGTGGGACTGCAGATCACTCCC. Sequences of all other primers are available upon request.

A DNA fragment containing the insertion in the cax mutation was amplified with the long range PCR Kit from Qiagen using primers GCACTGCCACTGGGGCTAGT and CCCAGGCAGTGCGAGCA, followed by a nested PCR reaction with primers GAAAGATTGCAGAAAGGATTCTGT and AGGATGCTCTTTTGCAGGAAATCT. The fragment was subcloned in the TopoXL vector (Invitrogen) and sequenced. Primer pairs used to specifically detect the 5' and 3' ends of the IAP insertion in the cax allele were: AGGGCTCTTTTTGCGTCCTGT and TGGCGCTGACATCCTGTGTT, and CCCAGGCAGTGCGAGCA and TCAAGATCAGACTTACCTCGTTCC, respectively.

Southern Blot hybridizations were performed according to standard procedures with a Pofut1 cDNA probe containing exons 5 and 6 and the 5' region of exon7.

RNA was prepared from individual embryos using the RNeasy Minikit (Quiagen), and reverse transcribed using the High Capacity cDNA Reverse Transcription Kit (Applied Biosystems) according to the manufacturers' instructions. Pofut1 cDNA was quantified on an ABI 7900HT using two gene specific assays detecting portions of exons 1 and 2, and exons 6 and 7, respectively, (TaqMan gene expression assays Mm 01240157 m1 (for Pofut1 exon 6–7), Mm 00475567 m1 (for Pofut1 exon 1–2)) each measured in quadruplicate using wild type embryo cDNA as biological calibrator, GAPDH and HPRT as endogenous controls (TaqMan gene expression assays Mm 00446968 m1 (for Hprt exon 6–7), Mm 999999159 g1 (Gapdh exon 1–2), and additional no-template, no reverse transcription and blank controls. Data were evaluated using the SDS RQ Manager (V1.2 ABI) and the delta-delta Ct method 6061. Analysis of variance (ANOVA) was carried out using the statistical software package BIAS for Windows Version 8.6.3.

12 wild type and cax mutant embryos, respectively, and four Pofut1^{tm1Pst/tm1PSt} E9.5 embryos were pooled, and lysed in 2× Lämmli buffer. The equivalent of one embryo from pooled embryo lysates was loaded per lane. Proteins were separated by SDS-polyacrylamid gel electrophoresis and blotted onto transfer membranes (Immobilon, Millipore). After blotting membranes were incubated with Qentix™ Western Blot Signal Enhancer (Pierce), blocked with 10% milk powder in TBST at 4°C over night, incubated with rabbit anti-POFUT1 antibodies 29 diluted 1:1000 in TBST with 1% milk powder for 3 hr at 4°C, washed five times in TBST, incubated with HRPOD conjugated donkey anti-rabbit (GE Healthcare, 1:10000 in TBST with 1% milk powder), and washed five times in TBST. Bound antibodies were detected using ECL Western Blotting Detection Reagents (GE Healthcare) using a Fujifilm LAS 3000 gathering signals every 3 min over an 21 minute interval.