Targeted Disruption of a Murine Glucuronyl C5-epimerase Gene Results in Heparan Sulfate Lacking L -Iduronic Acid and in Neonatal Lethality*

The glycosaminoglycan, heparan sulfate (HS), binds proteins to modulate signaling events in embryogenesis. All identified protein-binding HS epitopes contain L -idu-ronic acid (IdoA). We report that targeted disruption of the murine D -glucuronyl C5-epimerase gene results in a structurally altered HS lacking IdoA. The corresponding phenotype is lethal, with renal agenesis, lung defects, and skeletal malformations. Unexpectedly, major organ systems, including the brain, liver, gastrointestinal tract, skin, and heart, appeared normal. We find that IdoA units are essential for normal kidney, lung, and skeletal development, albeit with different requirement for 2- O sulfation. By contrast, major early developmental events known to critically depend on heparan sulfate apparently proceed normally even in the absence of IdoA. The sulfated polysaccharide, heparan sulfate (HS), 1 is a ubiquitous component of proteoglycans on cell surfaces and in the extracellular matrix. HS

The sulfated polysaccharide, heparan sulfate (HS), 1 is a ubiquitous component of proteoglycans on cell surfaces and in the extracellular matrix. HS chains interact with a variety of proteins, including growth factors/morphogens, cytokines, enzymes, extracellular matrix proteins, and thus modulate important processes in development and homeostasis (1,2). Selective protein binding is mediated by saccharide domains containing sulfate groups in specific patterns, along with Liduronic acid (IdoA) units that promote ligand apposition through their conformational flexibility (3). The biosynthesis of HS is initiated by polymerization of D-glucuronic acid (GlcA) and N-acetyl-D-glucosamine (GlcNAc) residues in alternating sequence and is pursued through a series of modification reactions that include N-deacetylation/N-sulfation of GlcNAc, C5epimerization of GlcA to the C5-epimer, IdoA, and, finally O-sulfation at various positions (4). Mutational studies in Drosophila and targeted disruption of murine genes involved in HS biosynthesis, or encoding HS proteoglycan core proteins, have demonstrated a critical role for HS proteoglycans in developmental processes (5)(6)(7). The observed phenotypes vary dramatically in severity, from failure to gastrulate (8) to selective disturbance of mast cell function in otherwise seemingly healthy mice (9). Some of the biosynthetic enzymes occur in multiple isoforms with potentially different functions and thus different deletion phenotypes (9 -11). IdoA occurs in all proteinbinding HS domains so far identified. Here we show that mouse embryos lacking GlcA C5-epimerase synthesize an abnormal HS, with GlcA but no IdoA residues and a highly distorted sulfation pattern. The resulting phenotype is lethal, with defects that can be differentially ascribed to HS structural alterations. However, it also shows apparently normal organ development, believed to depend on selective HS-protein interactions. These findings raise intriguing questions as to the need for regulation in HS biosynthesis.

EXPERIMENTAL PROCEDURES
Gene Targeting Construct-A 19-kb genomic clone containing 3Јuntranslated region and 3 exons of the GlcA C5-epimerase gene (Hsepi) was isolated from a bacteriophage mouse (strain 129/Sv) genomic library (Stratagene). Three SacI fragments of this genomic clone were separately cloned into Bluescript plasmid (12). A 2-kb fragment immediately upstream of exon 3 was inserted downstream of a PGK-neo cassette as a short arm homologous sequence to the endogenous gene, and a 4-kb fragment downstream of exon 3 was inserted upstream of the PGK-neo cassette as a long arm of homology to the endogenous gene. The targeting vector construct had a total size of 11 kb.
Homologous Recombination in ES Cells and Generation of Hsepideficient Mice-The targeting vector was linearized by restriction enzyme NotI and electroporated into R1 embryonic stem (ES) cells. Clones expressing the neo-resistant gene were selected by including G418 (350 g/ml; Invitrogen) in the cell cultures and analyzed for target gene homologous recombination. The clone showing homologous recombination was detected by Southern blot analysis using an external 700-bp fragment as probe (Fig. 1A). The positive clone was injected into C57BL/6 blastocysts, and chimeric male founder mice were crossed with C57BL/6 females. The offspring was genotyped for the mutation by tail biopsies using a PCR method described below. The heterozygous mice were intercrossed to produce Hsepi mutant mice. Phenotype studies were performed on mice with mixed genetic background (129/SvJ/Sv and C57BL/6).
Genotype Analysis-Genotypes were determined by genomic Southern blot analysis (Fig. 1A) or by PCR. Genomic DNA from ES cells and tail biopsies were isolated as described previously (14) and digested with BglII and SalI. The resulting fragments were separated in agarose gel and subsequently blotted onto a nylon membrane, followed by hybridization with a 700-bp external probe as shown in Fig. 1A. The PCR amplification was performed using the following primers: 5Ј-AGTGT- HS Structure Analysis-Pregnant mice (17.5-18.5 d.p.c., intercross of heterozygous mice) were injected with 0.5 mCi of [6-3 H]GlcN and were sacrificed after 6 h. Whole embryos were homogenized in 4 M urea, 1% (v/w) Triton X-100, 50 mM Tris-HCl, pH 7.4. Proteoglycans in the extracts were purified by chromatography on DEAE-Sephacel, digested with chondroitinase ABC, and subjected to alkaline ␤-elimination. Released HS chains were recovered by anion-exchange chromatography. The purified HS samples were N-deacetylated by hydrazinolysis and treated with HNO 2 at pH 1.5 and 3.9 to convert the chains to disaccharides, which were subsequently reduced with NaB 3 H 4 (15). The metabolically and chemically 3 H-labeled disaccharides were analyzed by anion-exchange HPLC on a Partisil-10 SAX column. Products obtained on deamination of HS at pH 1.5 only (cleavage of chains at N-sulfated GlcN units (15)) were analyzed by gel chromatography on a Bio-Gel P-10 column (1.3 ϫ 185 cm). The column was eluted with 0.5 M NH 4 HCO 3 at a flow rate of 2 ml/h.
Morphological Analysis-Embryos from heterozygote matings were dissected under microscopy. Skeletal preparations of newborn mice were generated using an alcian blue-alizarin red staining method as described previously (16). For histological analysis, newborn mice were immediately fixed in 10% formalin. The fixed animals were sectioned transversally at seven levels, each level yielding nine consecutive 4-m sections. The sections were stained with hematoxylin and esoin.

Targeted Disruption of Hsepi Results in Neonatal Lethality
with Respiratory Failure-The GlcA C5-epimerase is encoded by a single Hsepi gene with three exons and accounts for the formation of IdoA units in HS biosynthesis (12,13). To assess the functional significance of IdoA residues in HS, we have generated Hsepi-deficient mice by gene targeting in ES cells. A targeting vector was constructed to create a functional mutation by deletion of exon 3 (Fig. 1A) corresponding to the 341 C-terminal amino acid residues of the 618-residue protein.
Following electroporation, homologous recombination in the Hsepi gene was found in one ES cell clone, identified by Southern blot analysis, with no additional sites of integration (not shown). Microinjection of this clone into C57BL/6 blastocysts yielded chimeric animals and germ line transmission. No overt defects were seen in heterozygous animals.
HS from Hsepi Ϫ/Ϫ Mice Lacks IdoA-HS from Hsepi ϩ/ϩ , Hsepi ϩ/Ϫ , and Hsepi Ϫ/Ϫ embryos (d.p.c. 17.5-18.5) was analyzed for composition, with particular regard to the occurrence of IdoA. Samples were cleaved to disaccharides by exhaustive deamination with nitrous acid, and the products were reduced with NaB 3 H 4 and analyzed by anion-exchange HPLC ( Fig. 2A). Notably, all three samples showed the same overall proportions of non-O-sulfated versus O-sulfated 3 H-labeled disaccharide units (Table I). The Hsepi ϩ/ϩ and Hsepi ϩ/Ϫ patterns of Osulfated disaccharides were indistinguishable and typical of normal HS, with a predominance of mono-O-and di-O-sulfated, IdoA-containing species. By contrast, these disaccharides were completely absent from the Hsepi Ϫ/Ϫ samples, which instead showed increased proportions of GlcA-containing species, including a disaccharide with 2-O-sulfated GlcA ( Fig. 2A; Table  I). Gel chromatography of metabolically [ 3 H]GlcN-labeled HS indicated similar chain length for the various samples (not shown), whereas the Hsepi Ϫ/Ϫ polysaccharide emerged more retarded than the corresponding Hsepi ϩ/ϩ and Hsepi ϩ/Ϫ products on anion-exchange chromatography (Fig. 2B). This difference in charge density was reflected by changes in N-substitution pattern, as demonstrated by size analysis of fragments obtained by selective deaminative cleavage of the HS chains at N-sulfated glucosamine residues (Fig. 2C). The proportion of disaccharides, derived from contiguous N-sulfated domains in mutant HS, was increased relative to wild type HS, whereas that of intermediate sized (4-to 14-mer) fragments was decreased. Total N-sulfation was calculated to 51 and 43% of disaccharide units in Hsepi Ϫ/Ϫ and Hsepi ϩ/ϩ HS, respectively. These findings indicate a change in HS structure, from a largely intermixed distribution of N-substituents to a pattern of extended N-sulfated domains, along with a switch from -IdoA(2-OSO 3 )-GlcNSO 3 -(upper structure in Fig. 2D) to -GlcA-GlcNSO 3 (6-OSO 3 )-(lower structure in Fig. 2D) as the predominant sulfated disaccharide unit.
Targeted Interruption of the Glucuronyl C5-epimerase Gene 28364 cage and sternum and often cleft palate (not shown). However, the brain, heart, liver, gastrointestinal tract, pancreas, and skin all appeared normal. Given the recognized importance of HS in early embryonic patterning and morphogenesis (5,6), the selective nature of the IdoA-deficient phenotype, albeit severe, was unexpected. The requirements for defined HS structure in several functionally important interactions thus would not seem to include the presence of IdoA units.
Comparison of the numerous defects in the Hsepi Ϫ/Ϫ mice with phenotypes due to elimination of other enzymes involved in HS biosynthesis shows similarities as well as differences. HS synthesized by mice deficient in N-deacetylase/N-sulfotransferase isoform 1 (NDST-1) is low in N-sulfate, hence in IdoA (sulfated and non-sulfated) and total O-sulfate compared with    (20), whereas the overall content of IdoA is similar to that of wild type HS (21). 2 The Hs2st Ϫ/Ϫ mice further show several skeletal abnormalities indistinguishable from those seen in Hsepi Ϫ/Ϫ animals (17,18). On the other hand, contrary to the NDST-1 Ϫ/Ϫ and Hsepi Ϫ/Ϫ mice, the Hs2st Ϫ/Ϫ mutants had apparently normal lungs and remained alive for some hours (17). Differential Dependence of Development on HS Fine Structure-The accumulated information provides some clues to structure/function relations in HS biology. The multiple functions ascribed to HS proteoglycans throughout embryonic development involve several signaling networks, including FGF, Wnt, TGF-␤, and Hedgehog pathways (6,22). The corresponding signaling mechanisms appear either unaffected or perturbed by the lack of IdoA in HS. Indian Hedgehog is expressed during gastrulation in mammals and is an important regulator of developmental processes. Elimination in mice of the EXT1 protein, which is essential for HS biosynthesis, resulted in loss of HS, disruption of gastrulation, and embryonic lethality before E8.5 (23). In EXT1 Ϫ/Ϫ embryos Indian Hedgehog failed to bind to the appropriate target cells, whereas wild type embryos showed strong binding. This critical interaction can apparently be mediated also by grossly perturbed HS lacking either IdoA or 2-O-sulfate residues. The up-regulation in N-sulfation and 6-O-sulfation observed in both Hsepi Ϫ/Ϫ (Table I) and Hs2st Ϫ/Ϫ HS (20) could conceivably "compensate" for the loss of IdoA and 2-O-sulfate residues, respectively, in some interactions crucial to development. On the other hand, we have now identified certain phenotype traits that critically depend on the presence of IdoA. Some of these features can be differentially related to the sulfation state of this monosaccharide component, as revealed by the Hsepi Ϫ/Ϫ and Hs2st Ϫ/Ϫ phenotypes. Lung maturation thus requires a minimal proportion of IdoA units (or a structure associated with such units) that do not need to be sulfated. Other events, including kidney, iris, and skeletal development, depend on HS with 2-O-sulfated IdoA residues. The distinctive NDST-1 Ϫ/Ϫ phenotype points to signaling mechanisms with yet other requirements for HS structure (5). The regulation and temporal aspects of the expression and functional recruitment of HS epitopes continue to provide challenges for future work.