Molecular Cloning, Expression, Chromosomal Assignment, and Tissue-specific Expression of a Murine (cid:97) -(1,3)-Fucosyltransferase Locus Corresponding to the Human ELAM-1 Ligand Fucosyl Transferase*

Terminal Fuc (cid:97) 1–3GlcNAc moieties are displayed by mammalian cell surface glycoconjugates in a tissue-spe- cific manner. These oligosaccharides participate in se-lectin-dependent leukocyte adhesion and have been implicated in adhesive events during murine embryogenesis. Other functions for these molecules remain to be defined, as do the tissue-specific expression patterns of the corresponding (cid:97) -(1–3)-fucosyltransferase ( (cid:97) 1– 3FT) genes. This report characterizes a murine (cid:97) 1–3FT that shares 77% amino acid sequence identity with hu- man ELAM ligand fucosyltransferase (ELFT, also termed Fuc-TIV). The corresponding gene maps to mouse chromosome 9 in a region of homology with the Fuc-TIV locus on human chromosome 11q. In vitro , the murine (cid:97) 1–3FT can efficiently fucosylate the trisaccharide Gal (cid:97) 1–3Gal (cid:98) 1–4GlcNAc In Analysis— A segment of Fuc-TIV gene encompassing nucleotide positions

Oligosaccharides represent major components of animal cell surfaces and are believed to function in cellular interactions during development and differentiation (1,2), oncogenic transformation (3), and inflammation (4). Identification of specific oligosaccharide ligands for the selectin family of cell adhesion molecules directly links cell surface carbohydrates to cell-cell communication in the context of inflammatory response (5)(6)(7). The proposed ligands for E-selectin and P-selectin are fucosylated oligosaccharides (for review, see Refs. 4 and 8), whose biosynthesis is catalyzed by ␣-(1-3)-fucosyltransferases (␣1-3FTs). 1 These enzymes are encoded by one or more distinct and tightly regulated ␣1-3FT genes.
Much of the interest in discovering functional roles for oligosaccharides during development is derived from studies documenting precise temporal-spatial expression patterns for some oligosaccharides during human and murine embryogenesis (9 -12). The murine stage-specific embryonic antigen-1 (SSEA-1; (Gal␤1-4[Fuc␣1-3]GlcNAc; Lewis x)), for example, is expressed coincident with morula compaction at the 8 -16 cell stage of the preimplantation mouse embryo (13,14). Since SSEA-1 structural analogs appear to inhibit compaction, it has been suggested that this antigen may participate in this process (1,2). While it has been proposed that the SSEA-1 determinant functions to promote homotypic adhesion (15), neither the physiological relevance of this interaction during compaction nor the existence of other preimplantation-specific receptors for SSEA-1 have been demonstrated. Furthermore, it has not been possible to identify and directly demonstrate functional correlates for SSEA-1 expression patterns during early embryogenesis.
Virtually nothing is known about the molecular mechanisms that determine the tissue-specific and developmentally regulated expression patterns of the oligosaccharides implicated in morphogenic events during early murine embryogenesis. Expression of surface-localized SSEA-1 molecules may be regulated by differential expression of ␣1-3FT(s) required for their synthesis (16) and of sialyltransferases (17) and an ␣-(1-3)-galactosyltransferase (18) that may mask SSEA-1 expression (9,10,18). The relative contributions of these and other regulatory mechanisms, in the context of the developing embryo, remain undefined.
To begin to define, in detail, the enzymes and mechanisms that determine expression of Fuc␣1-3GlcNAc linkages in murine cell surface oligosaccharides, we have isolated and characterized a murine gene that corresponds to a human ␣1-3FT gene termed Fuc-TIV (Refs. 19 and 20; also known as ELFT for ELAM-1 ligand fucosyl transferase, Ref. 21). The mouse enzyme differs from human Fuc-TIV/ELFT in its relatively higher apparent affinity for ␣-(2-3)-sialylated type acceptor substrates in vitro, and is able to efficiently synthesize an unusual tetrasaccharide (Gal␣1-3Gal␤1-4[Fuc␣1-3]GlcNAc) whose existence in periimplantation mouse tissues has been previously inferred (10). Northern blot analyses confirm that transcripts corresponding to this gene accumulate in leukocytic cell lines and in leukocyte-rich tissues in the mouse. However, these analyses, and companion in situ hybridization studies, demonstrate that Fuc-TIV/ELFT transcripts are unexpectedly abundant in epithelial cells lining the gastrointestinal and reproductive tracts and suggest that this sequence, and the cognate Fuc␣1-3GlcNAc linkages whose expression it determines may have unexpected functions in these tissues.
Subcloning and DNA Sequence Analysis-A 1.4-kb NcoI-SspI fragment homologous to the human FucT-IV probe (468-bp AvaI-PvuII fragment, Ref. 20) was isolated from the phage DNA and subcloned into pTZ19R plasmid vector. A 2.6-kb SacI fragment homologous to the human Fuc-TIII probe (representing the murine ␣1-3FT pseudogene described under "Results") was isolated from phage DNA and subcloned into the SacI site of pTZ19R. The DNA sequences of these fragments were determined by dideoxy method using Sequenase sequencing kit (U. S. Biochemical Corp.). Sequence analysis was performed using the sequence analysis software package (GCG) of the University of Wisconsin Genetics Computer Group (39) and the MacVector version of the IBI Pustell Sequence Analysis Software package (International Biotechnologies, Inc.). Sequence alignments were assembled with the GAP and BESTFIT functions of the GCG package.
Northern Blot Analysis-Total RNA was prepared from mouse (FVB strain) tissues and cultured cells, using procedures described previously (38). Poly(A) ϩ RNA was isolated with oligo(dT)-cellulose (Collaborative Research) chromatography (38). RNA samples were electrophoresed through 1.0% agarose gels containing formaldehyde (38) and were transferred to a nylon membrane (Hybond-N, Amersham Corp.). Northern blots were prehybridized for 2 h at 42°C in 1 ϫ PE (35), 5 ϫ SSC, 0.5% sodium dodecyl sulfate, and 150 g/ml sheared salmon sperm DNA. Blots were hybridized for 18 h at 42°C in prehybridization solution containing a 32 P-labeled (36) 1.4-kb NcoI-SspI fragment. This sequence begins at the most proximal putative initiation codon shown in Fig. 1 and ends at an SspI site approximately 90 bp distal to the termination codon. Blots were stripped in boiling 0.1% SDS and rehybridized with a ␤-actin probe to confirm that RNA samples were intact and loaded in equivalent amounts.
Transfection and Expression of Murine Fuc-TIV Gene-The 1.4-kb NcoI-SspI fragment was subcloned into the plasmid pcDNAI. A plasmid containing a single insert in the sense orientation relative to the plasmid's cytomegalovirus promoter-enhancer sequences was designated pcDNAI-mFuc-TIV. COS-7 cells were transfected with this plasmid, with the control plasmid pcDNAI, or with pcDNAI-hFuc-TIV (20), using a DEAE-dextran procedure (38) as described previously (37,40). A 1.8-kb HpaI-XbaI fragment derived from the murine putative pseudogene was cloned between the EcoRV and XbaI sites in the mammalian expression plasmid pcDNAI. A plasmid containing a single insert in the sense orientation was designated pcDNAI-MPFT.
Flow Cytometry Analysis-COS-7 cells transfected with plasmid pcDNAI-mFuc-TIV were harvested (40) 72 h after transfection and stained with monoclonal antibodies diluted in staining media, as described previously (37,40). Anti-Lewis a, anti-H, and anti-sialyl Lewis x antibodies (mouse IgM) were used at 10 g/ml. Anti-Lewis x antibody (mouse IgM anti-SSEA-1; ascites) was used at a dilution of 1:1000. Anti-sialyl Lewis a (mouse IgG; ascites) was used at a dilution of 1:500. Cells were then stained with fluorescein isothiocyanate-conjugated goat anti-mouse IgM or anti-mouse IgG and subjected to analysis on a FACScan (Becton Dickinson) as described previously (37). Thresholds for antigen positivity were set at a fluorescence intensity level that excludes 99% of transfected COS-7 cells that had been stained with the test antibody (anti-H).
␣1-3FT Assays-Cell extracts containing 1% Triton X-100, 25% glycerol were prepared from transfected COS-7 cells using procedures described previously (37). ␣1-3FT assays were performed in a total volume of 20 l. In preliminary experiments designed to optimize the conditions for activity determination, the reaction mixture contained 3 M GDP-[ 14 C]fucose, 20 mM acceptor (N-acetyllactosamine), and a quantity of cell extract protein sufficient to yield linear reaction conditions. Neutral acceptors were purchased from Sigma (N-acetyllactosamine, lactose, lacto-N-biose I, 2Ј-fucosyllactose) or from V-labs (Gal␣1-3Gal␤1-4GlcNAc). The reaction mixture utilized for kinetic calculations included 50 mM Tris-HCl buffer, pH 6.8, 5 mM ATP, 10 mM L-fucose, 5 mM MnCl 2 , 3 M GDP-[ 14 C]fucose, and 0.5-1.5 g of protein extract. Concentrations of acceptor substrates, or GDP-fucose, were varied as indicated in the legends to Figs. 3, 4, and 5. Synthesis and characterization of unlabeled GDP-fucose utilized for GDP-fucose concentration activity determinations has been described previously (41). The concentrations of GDP-fucose in stock solutions were calculated from the UV absorbance at 254 nm of an aliquot diluted in water. The molar extinction coefficient of GDP (⑀ ϭ 13,800 at 254 nm, pH 7.0, Ref. 42) was used for this calculation since the extinction coefficient of GDP-fucose is not available. Reactions were incubated at 37°C for 1 h. Blanks were prepared by omitting the acceptor in the reaction mixture, and their values were subtracted from the corresponding reaction that contained acceptor. This background radioactivity reproducibly represented less than 1% of the total radioactivity in the assays and corresponds to the [ 14 C]fucose present in the GDP-[ 14 C]fucose as obtained from the manufacturer.
Reactions containing neutral acceptors (N-acetyllactosamine, lactose, lacto-N-biose I, 2Ј-fucosyllactose, Gal␣1-3Gal␤1-4GlcNAc) were terminated by the addition of 20 l of ethanol and 560 l of water. Samples were centrifuged at 15,000 ϫ g for 5 min, and a 50-l aliquot was subjected to scintillation counting to determine the total amount of radioactivity in the reaction. An aliquot of 200 l was applied to a column containing 400 l of Dowex 1-X2-400, formate form (35,37). The column was washed with 2 ml of water, and the radioactive reaction product, not retained by the column, was quantitated by scintillation counting. Reactions with the acceptor ␣-(2-3)-sialyl-N-acetyllactosamine were terminated by adding 980 l of 5.0 mM sodium phosphate buffer, pH 6.8. Samples were then centrifuged at 15,000 ϫ g for 5 min, and a 500-l aliquot was applied onto a Dowex 1-X8 -200 column (1 ml) prepared in the phosphate form. The reaction product was collected in the eluate and quantitated as described previously (43).
In Situ Hybridization Analysis-A segment of the murine Fuc-TIV gene encompassing nucleotide positions 270 -756 was amplified by the polymerase chain reaction and subcloned into plasmid vector pcDNAI (Clontech). Antisense or sense RNA in situ hybridization probes were generated using this plasmid as a template in Sp6 or T7 RNA polymerase-directed in vitro transcription mixtures containing ␣-[ 35 S]UTP (Amersham Corp.). Probe lengths were adjusted to an average of 250 nucleotides by limited alkaline hydrolysis. Alternatively, antisense or sense RNA in situ hybridization probes were generated in a similar manner, were derived from a segment of the murine Fuc-TIV gene encompassing nucleotide positions 491-746, and were used directly without prior alkaline hydrolysis.
Cryosections (10 m thick) were prepared with Jung Frigocut 2800N cryostat (Leica) from organs of 129/J strain mice. Tissue sections were fixed in 4% paraformaldehyde/phosphate-buffered saline for 30 min at room temperature. After proteinase K treatment and then acetylation by acetic anhydride, sections were incubated with antisense or sense probes (2 ϫ 10 4 cpm/l) for 18 h at 55°C in 20 mM Tris-HCL, pH 8.0, 300 mM NaCl, 5 mM EDTA, 10 mM sodium pyrophosphate, 50% formamide, 10% dextran sulfate, 5 ϫ Denhardt's solution, 50 mg/ml heparin, 0.1 M dithiothreitol, and 0.5 mg/ml Escherichia coli tRNA. The slides were washed for 30 min at 65°C in 50% formamide, 2 ϫ SSC, 10 mM dithiothreitol. The sections were then subjected to digestion with 10 g/ml of ribonuclease A at 37°C to eliminate residual nonbase-paired probe and were then washed again in the wash solution described above. Slides were exposed for 2 weeks using Kodak NBT-2 emulsion and processed using D-19 developer and fixer (Eastman Kodak Co.). The sections were subsequently stained with hematoxylin and eosin.
Interspecific Mouse Backcross Mapping-Interspecific backcross progeny were generated by mating (C57BL/6J ϫ Mus spretus) F 1 females and C57BL/6J males as described previously (46). A total of 205 N 2 mice were used to map the Fut4 locus (see "Results" for details). DNA isolation, restriction enzyme digestion, agarose gel electrophoresis, Southern blot transfer, and hybridization were performed essentially as described previously (47). All blots were prepared with Hybond-N ϩ nylon membrane (Amersham Corp.). A probe corresponding to base pairs 172-1203 (see Fig. 1) of the mouse Fuc-TIV coding region was labeled with [␣-32 P]dCTP using a nick translation labeling kit (Boehringer Mannheim); washing was done to a final stringency of 1.0 ϫ SSC, 0.1% SDS, 65°C. A fragment of 5.8 kb was detected in HindIII-digested C57BL/6J DNA, and a fragment of 4.7 kb was detected in HindIIIdigested M. spetus DNA. The presence or absence of the 4.7-kb M. spretus-specific HindIII fragment was followed in backcross mice.
A description of the probes and RFLPs for the loci linked to Fut4 including murine macrophage metalloelastase (Mmel), low density lipoprotein receptor (Ldlr), and erythropoietin receptor (Epor) has been reported previously (48). Recombination distances were calculated as described previously (49) using the computer program SPRETUS MAD-NESS. Gene order was determined by minimizing the number of recombination events required to explain the allele distribution patterns.

Molecular
Cloning of a Murine ␣1-3FT Gene-A hybridization screen for a murine ␣1-3FT gene (see "Experimental Procedures") yielded numerous phages that cross-hybridize with probes derived from human ␣1-3FT genes. Sequence analysis of the insert in a representative of a group of phages with similar restriction maps identified a region with a substantial amount of primary sequence similarity to the coding portion of the human Fuc-TIII gene (62% sequence identity; data not shown; sequence deposited in GenBank, accession number U33458). However, translation of this murine sequence to maintain primary amino acid sequence similarity to Fuc-TIII required the conceptual suppression of multiple frameshift and nonsense mutations in the murine sequence. Furthermore, this sequence did not yield detectable ␣1-3FT activity when expressed in COS-7 cells (data not shown). These observations indicate that this murine sequence represents a pseudogene; this sequence was not further analyzed.
This hybridization screen also identified phages containing a 1.4-kb NcoI-SspI fragment that cross-hybridizes with the human ␣1-3FT gene encoding Fuc-TIV (19 -21). Sequence analysis of the gene fragment identifies a single long open reading frame (Fig. 1), which begins with a methionine codon located within a sequence context largely consistent with Kozak's consensus rules for mammalian translation initiation (52). Hydropathy analysis (53) of the protein sequence predicted by this open reading frame identifies a single 22-amino acid hydrophobic segment at the NH 2 terminus, implying that the polypeptide has a type II transmembrane topology typical of mammalian glycosyltransferases (16,54). Sequence comparisons made between the predicted murine protein and human ␣1-3FTs identify significant primary sequence similarities; the murine protein maintains approximately 33, 33, 39, and 37% amino acid sequence identity with the human Fuc-TIII (37), Fuc-TVI (45), Fuc-TVII (55,56), and Fuc-TV (57) enzymes, respectively (data not shown). However, it is most similar to human Fuc-TIV (19 -21), which shares 77% amino acid sequence identity with the murine protein (304 identities at 396 aligned residues; Fig. 1). This sequence similarity includes the conservation of two consensus sites for asparagine-linked glycosylation (Fig. 1).
Maximal sequence similarity is achieved by aligning the murine DNA sequence in a colinear manner with the sequence of the human Fuc-TIV gene (19,20) and its cDNA (21). Like the human Fuc-TIV/ELFT locus, the murine gene apparently maintains a single coding exon. Primer extension experiments and RNase protection analyses designed to define the transcriptional initiation site for this murine gene have failed because of the tendency for its transcript to form secondary structure that is resistant to denaturation (data not shown). This is most probably a function, in part, of the extraordinarily high GϩC content within the 5Ј end of this gene. We also believe these observations explain our inability to isolate full-length murine Fuc-TIV cDNAs from multiply screened high quality cDNA libraries (data not shown). Nonetheless, comparison of the DNA sequences of the murine and human genes through positions exceeding 400 base pairs proximal to their respective initiation codons discloses that their respective 5Ј-flanking regions are identical at 66% of the aligned positions (Fig. 1). This high level of DNA sequence identity is maintained throughout the region of the human gene where it is co-linear with Fuc-TIV/ELFT mRNA transcripts (Fig. 1), as defined by cDNA cloning experiments (21), which in turn yield Fuc-TIV/ELFT or a longer form of this enzyme (ELFT-L; Ref. 21). The murine gene also has the potential to yield a longer form of its polypeptide product. This longer sequence is represented by a polypeptide initiating at a methionine codon located 33 codons 5Ј to the human Fuc-TIV/ELFT initiator methionine codon, extending the shorter murine polypeptide by 33 amino acid residues at its NH 2 terminus (Fig. 1). This longer murine sequence maintains 52% amino acid sequence identity with the ELFT-L polypeptide in a region immediately proximal to the Fuc-TIV/ELFT initiator methionine residue (Fig. 1). These observations further support the conclusion that the murine gene maintains a structural organization essentially identical to that of the human Fuc-TIV/ELFT gene.
In Vitro Enzyme Assay of Mouse and Human Fuc-TIVs-COS-7 cells transfected with pcDNAI-mFuc-TIV also contain a substantial amount of ␣1-3FT activity that utilizes the acceptor N-acetyllactosamine. This murine enzyme has a pH optimum of 7.5 (Fig. 3) and is maximally stimulated by Mn 2ϩ (5-fold) at a concentration of 15 mM (Fig. 3). In comparison, the human Fuc-TIV enzyme (20) is optimally active at pH 7.3 and is maximally stimulated by Mn 2ϩ (4-fold) at a concentration of 10 mM, when assayed with N-acetyllactosamine (Fig. 3).
Kinetic analyses of murine and human Fuc-TIV enzymes indicate that they exhibit typical substrate concentration-dependent Michaelis-Menten kinetics (Figs. 4 and 5). The calculated apparent Michaelis constants for the donor substrate GDP-fucose are 16.6 and 27.0 M for the murine and human enzymes, respectively (Fig. 4). The apparent Michaelis constants for the acceptor substrates N-acetyllactosamine and ␣-(2-3)-sialyl-N-acetyllactosamine are 2.05 and 1.78 mM, respectively, for the murine enzyme, and 3.3 mM and 6.74 mM, respectively, for human Fuc-TIV (Fig. 5). Both enzymes utilize the neutral type II acceptor molecules 2Ј-fucosyllactose and lactose with efficiencies that are substantially less than those obtained with either ␣-(2-3)-sialyl-N-acetyllactosamine or Nacetyllactosamine. For example, a single preparation of the murine enzyme utilized 2Ј-fucosyllactose and lactose at rates that were 7% (1.1 nmol/mg⅐h) and 1% (0.2 nmol/mg⅐h), respectively, of the rate obtained with N-acetyllactosamine (16.6 nmol/mg⅐h). Only trace amounts of fucosylated product were produced when the neutral type I acceptor lacto-N-biose I was assayed with the same preparation of murine enzyme (0.05 nmol/mg⅐h). These results are virtually identical to those obtained when extracts containing human Fuc-TIV activity are assayed with the latter three substrates (20,55). Thus, when considered together with the DNA and protein sequence comparisons and the flow cytometry analyses, these biochemical data indicate that the open reading frame shown in Fig. 1 encodes an ␣1-3FT activity that corresponds to a human "myeloid"-type ␣1-3FT (Fuc-TIV/ELFT) (19 -21, 58, 59).
The Gal␣1-3Gal␤1-4GlcNAc trisaccharide is a cell surface oligosaccharide epitope expressed by many murine tissues (60). There is indirect evidence for a fucosylated form of this epitope FIG. 1. Nucleotide and deduced amino acid sequences of the murine ␣1-3FT gene and comparison to the human Fuc-TIV/ELFT gene and its cDNAs. The DNA sequence of the murine gene, and its predicted protein sequences, correspond to MFT-IV DNA, and MFT-IV AA, respectively. The DNA sequence of the human Fuc-TIV/ELFT gene, and predicted protein sequences, correspond to lines denoted by HFT-IV DNA, and HFT-IV AA, respectively. The A residues of the putative initiation codon of the murine ␣1-3FT gene and of the human ELFT/Fuc-TIV sequence are assigned as residue 1 of their nucleotide sequence. The methionine residues corresponding to these codons are assigned as residue 1 of the corresponding protein sequences. These positions are further denoted by an arrow pointing to these aligned initiator methionine codons (start of ELFT AA). The initiation codon for the "long" form of ELFT (21) is also indicated by arrows (start of ELFT AA; Initiator codon for ELFT-L protein). The initiator methionine for the long form of the murine polypeptide is also indicated by an arrow (start of "long" mouse polypeptide). This methionine codon is encompassed within the NcoI site (dotted underlined) used to construct the vector pcDNAI-mFuc-TIV. Arrows also indicate the 5Ј-most residues found in the ELFT and ELFT-L cDNAs (21) (start of ELFT cDNA; start of ELFT-L cDNA). The sequence alignment was generated using the BESTFIT and GAP programs of the University of Wisconsin Genetics Computer Group (39). The GAP program generates symbols between aligned amino acids, according to the evolutionary distance between them, as measured by Dayhoff (50) and normalized by Gribskov (51). Amino acid sequence identities are assigned a score of 1.5, denoted by a vertical bar; related amino acid residues with scores from 0.5 to 1.4 are denoted by a two dots, less strongly related amino acid residues with scores between 0.1 to 0.4 are denoted by one dot; no symbol is placed between dissimilar amino acid pairs with scores less than 0.1. Gaps in the amino acid sequence alignment are denoted by a dash. Nucleotide sequence identity is indicated by vertical lines between aligned corresponding residues. Gaps in the nucleotide sequence alignment are indicated by a dotted line. The predicted transmembrane domain of the murine enzyme (amino acids 53-74) is double underlined. Consensus sites for asparagine-linked glycosylation are underlined. In vitro studies using a human ␣1-3FT suggest that fucosylation can be a terminal step in the synthesis of this tetrasaccharide (61). We find that the murine ␣1-3FT can efficiently utilize the trisaccharide Gal␣1-3Gal␤1-4GlcNAc (with an apparent K m ϭ 0.71 mM; Fig. 5C) to form a fucosylated product corresponding to the tetrasaccharide Gal␣1-3Gal␤1-4[Fuc␣1-3]GlcNAc (see "Experimental Procedures"). This observation suggests that murine glycoconjugates containing the Gal␣1-3Gal␤1-4GlcNAc trisaccharide represent one authentic acceptor substrate for this murine ␣1-3FT. Although the human Fuc-TIV enzyme can also form this tetrasaccharide product in vitro (data not shown), this reaction has no apparent physiological relevance since the human genome does not encode a functional ␣(1-3)-galactosyltransferase capable of synthesizing Gal␣1-3Gal␤1-4GlcNAc precursors (62).
Chromosomal Location of the Murine ␣1-3FT Locus-The mouse chromosomal location of this Fuc-TIV-like gene (locus designation Fut4) was determined by interspecific backcross analysis using progeny derived from matings of ((C57BL/6J ϫ M. spretus)F 1 ϫ C57BL/6J) mice. This interspecific backcross mapping panel has been typed for over 1600 loci that are well distributed among all of the autosomes as well as the X chromosome (46). C57BL/6J and M. spretus DNAs were digested with several enzymes and analyzed by Southern blot hybridization for informative restriction fragment length polymorphisms using a mouse Fuc-TIV probe (see "Experimental Procedures"). The 4.7-kb M. spretus HindIII restriction fragment-length polymorphism ("Experimental Procedures") was used to follow the segregation of the Fut4 locus in backcross mice. The mapping results indicated that Fut4 is located in the proximal region of mouse chromosome 9 linked to Mmel, Ldlr, and Epor. Although 138 mice were analyzed for every marker and are shown in the segregation analysis (Fig. 6), up to 173 mice were typed for some pairs of markers. Each locus was analyzed in pairwise combinations for recombination frequencies using the additional data. The ratios of the total number of mice exhibiting recombinant chromosomes to the total number of mice analyzed for each pair of loci and the most likely gene order are as follows: centromere -Mmel -3/173 -Fut4 -4/150 -Ldlr -1/161 -Epor. The recombination frequencies (expressed as genetic distances in centimorgans Ϯ the standard error) are as follows: The proximal region of mouse chromosome 9 shares a region of homology with human chromosomes 11q and 19p (summarized in Fig. 6). The human Fuc-TIV gene (locus designation FUT4) was initially assigned to human 11q12-qter (63). More recently, the human map position has been refined to 11q21 (64). These studies provide additional support for concluding that this murine gene is the human homologue of Fuc-TIV gene, and confirm and extend the region of homology between mouse chromosome 9 and the long arm of human chromosome 11.
Tissue-specific Expression of the Murine Fuc-TIV Gene-Northern blot analyses were completed to define the tissuespecific expression patterns of this murine ␣1-3FT gene. A single transcript, approximately 4.4 kb in length, was detected on blots prepared with polyadenylated RNA isolated from various murine tissues and blood-derived cell lines (Fig. 7). This transcript is most abundant in the stomach and colon; substan- tial amounts are also detected in the lung, testes, uterus, and small intestine. Lesser amounts of the Fuc-TIV mRNA are detected in the thymus, spleen, and ovary, and only trace amounts are observed in the brain, heart, smooth muscle, kidney, thymus, and bone marrow. The murine Fuc-TIV transcript is not found in liver, salivary gland, or pancreas at levels detectable by Northern blot analysis using up to 3 g of polyadenylated mRNA (Fig. 7).
Northern blot analysis of cultured blood cell-type cell lines indicates that the Fuc-TIV transcript is relatively abundant in the murine erythroleukemia cell line MEL (26,27), substantially less abundant in the RAW macrophage-derived line (28,29), and not detectable in the P388 macrophage-derived cell line (30)  types within some of the organs where relatively abundant levels of this transcript are present (Fig. 8). These experiments demonstrate that the Fuc-TIV transcript accumulates to substantial levels within the epithelial cells lining the stomach and colon (Fig. 8). The Fuc-TIV transcript accumulates to a lesser degree within the epithelial cells (both absorptive cells and goblet cells) lining the small intestinal villi, and the mucus glands within the small intestine. Fuc-TIV transcripts are also evident in the epithelial cells lining the epididymus (Fig. 8) but are not visible within the testis proper (data not shown). These observations correspond well with Fuc-TIV transcript abundance seen on the Northern blot analyses (Fig. 7). No Fuc-TIV transcripts are detected in the kidney or the lung by in situ hybridization (data not shown), even though low to moderate levels of the Fuc-TIV transcript are evident on Northern blot analysis. It remains to be determined if this apparent discrepancy may be accounted for by the relatively insensitive nature of the in situ hybridization method and/or by Fuc-TIV transcripts in blood cells within the larger vessels of these organs. These cells may contribute to the Northern blot signals but may be eliminated from the tissues prior to or during preparation for in situ hybridization. Fuc-TIV transcripts are not detected in murine neutrophils (data not shown); we have yet to define which blood cell lineages, and which maturation stage(s) of those lineages, are responsible for the Fuc-TIV transcripts observed in bone marrow mRNA. DISCUSSION Biochemical studies indicate that selectin ligand expression in myeloid and lymphoid lineages is controlled in part by cell type-specific expression of one or more ␣1-3FTs and that these enzymes consequently play a pivotal role in the human inflammatory response (5)(6)(7)(8). Evidence also suggests that the aberrant expression of these enzymes in malignancy may facilitate the spread of transformed cells via selectin-dependent metastatic processes (5)(6)(7)(63)(64)(65)(66)(67). The developmentally regulated expression patterns of ␣-(1-3)-fucosylated oligosaccharides in mammalian embryos (2, 9 -12) further imply that these molecules have additional, undefined functions. Given logistical and ethical considerations, the laboratory mouse is a useful system to analyze embryonic and adult ␣1-3FT gene expression patterns and to perturb these patterns through transgenic and embryonic stem cell approaches.
The polypeptide product of the murine gene described here shares 77% amino acid sequence identity with human Fuc-TIV (19 -21). Inter-species sequence comparisons have not previously been made for any ␣1-3FT sequences; the degree of sequence identity observed is comparable with levels reported for other cross-species comparisons of glycosyltransferase sequences (68 -70) and suggest that this murine gene is the orthologous homologue (71) of human Fuc-TIV. This assignment is further supported by their corresponding chromosomal localizations and by the genomic organization of the murine gene and the human Fuc-TIV gene, both of which apparently maintain intronless coding sequences.
Comparison of the catalytic properties of the murine enzyme with those of the human Fuc-TIV enzyme provides additional evidence for their homologous nature. Transfection studies indicate that both enzymes can determine expression of surfacelocalized Lewis x molecules, but not sialyl Lewis x moieties or products based on type I precursors. The mouse and human enzymes maintain similar affinities for the type II acceptor N-acetyllactosamine, in vitro (2.05 and 3.3 mM, respectively), but differ somewhat in their apparent affinities for ␣-(2-3)sialyl-N-acetyllactosamine. Specifically, the human enzyme exhibits a rather higher apparent K m (6.7 mM) for the sialylated substrate than does the mouse enzyme (1.8 mM). Nevertheless, the murine enzyme does not utilize oligosaccharides terminating with ␣-(2-3)-sialylated type II chains when it is expressed in the cultured cell lines used here. The apparent discrepancy presented by the ability of Fuc-TIV to utilize ␣-(2-3)-sialylatedtype II oligosaccharides in vitro, but not in vivo, may be resolved by considering recent results indicating that Fuc-TIV resides in a Golgi compartment proximal to ␣-(2-3)-sialyltransferase and thus may not have an opportunity to operate upon such substrates within a cell (72). These observations further emphasize that it is difficult to reliably predict the spectrum of oligosaccharide products that will be constructed in a specific cell lineage by a given glycosyltransferase solely by considering the results obtained from in vitro assays using low molecular weight oligosaccharide acceptors. This notion is strongly reinforced by results indicating that the human Fuc-TIV gene can, under some circumstances, determine cell surface sialyl Lewis x expression in transfected cells, and that this outcome is critically dependent upon the glycosylation phenotype of the host cell in which Fuc-TIV is expressed (73). Thus, while it seems likely that the murine Fuc-TIV enzyme creates Fuc␣1-3GlcNAc linkages in murine tissues, it is not yet possible to predict which murine glycoconjugate acceptors will be utilized by this ␣1-3FT, and, consequently, we cannot yet predict the cell surface products created by this enzyme.
The Gal␣1-3Gal␤1-4[Fuc␣1-3]GlcNAc tetrasaccharide represents a noncharged analogue of the sialyl-Lex determinant, prompting a suggestion that it may represent a possible ligand for the selectins (61). It is not yet known if this tetrasaccharide can participate in selectin-dependent cell adhesion processes or if it is expressed by murine neutrophils, although the precursor trisaccharide determinant is displayed by these cells. 2 It is interesting to note in this context that the oligosaccharide portion of the murine E-and P-selectin counter-receptors are not yet defined, although the sialyl Lewis x determinant is not expressed on murine leukocytes (75), at least as defined by the use of the monoclonal antibody CSLEX. The tissue-specific expression patterns, and functions, if any, of this tetrasaccharide molecule thus remain an open and interesting question.
Northern blot analyses identify a single mouse Fuc-TIV transcript in all tissues where the gene is expressed, whereas the human Fuc-TIV gene generates multiple transcripts (19 -21). By analogy to other mammalian glycosyltransferase mRNAs (16,54), the relatively large size of the mouse Fuc-TIV transcript suggests that it contains substantial amounts of 3Јand/or 5Ј-untranslated segments. The structure of this transcript remains to be determined by cloning and sequencing of the cDNA(s) derived from this gene. While the murine transcript is detectable in bone marrow, the specific marrow cell types that express this gene remain to be precisely defined. It is likely, however, that the marrow transcripts are derived in part from myeloid-lineage cells since the human Fuc-TIV gene is expressed in myeloid cells (19 -21). The murine erythroid lineage may also contribute to expression in the bone marrow since transcripts are detected in the mouse erythroleukemia line MEL. The murine transcript is also relatively abundant in the epithelial cells lining the colon and stomach, with somewhat lesser amounts in small intestinal epithelial cells. The function of this enzyme in these locations remains unknown, although it is interesting to speculate that it serves to participate in the synthesis of fucosylated mucins that may operate to protect the gastrointestinal lining from destructive effects of digestive enzymes or from ingested pathogens. The low levels of the Fuc-TIV transcript in the testes is accounted for by transcripts that accumulate in the epididymus. The low levels of Fuc-TIV message detected in the kidney, and the trace amounts identified in brain and heart, may represent transcripts derived from specific cell types within the parenchyma of these organs. Further in situ hybridization experiments will be required to confirm this possibility.
These results indicate that while the murine and human Fuc-TIV genes are similarly organized, there may be substantial interspecies differences in the structural, functional, and regulatory properties of the orthologous ␣1-3FTs. Evidence suggests that the functional roles of their oligosaccharide products may also differ in significant ways. The expression patterns exhibited by the mouse ␣1-3FT gene are different from those shown by a rat ␣-(2-6)-sialyltransferase gene (76,77), a constitutively expressed mouse ␤-(1-4)-galactosyltransferase gene (78,79), a mouse ␣-(1-3)-galactosyltransferase gene (60), and a series of human sialyltransferase genes (80). When considered together with these data, our results provide additional support for the notion that transcriptional control of glycosyltransferase gene expression regulates cell surface glycosylation (54,80).
Functional correlates for the regulation of this ␣1-3FT remain to be explored. In this context, we have compared our interspecific map of chromosome 9 with a composite mouse linkage map that reports the map location of many uncloned mouse mutations. 3 Fut4 maps in a region of the composite map that lacks mouse mutations with a phenotype that might be expected for an alteration in this locus (data not shown). The task of assigning function to this locus 2 A. Thall and J. B. Lowe, unpublished data. may be facilitated by a more complete understanding of the types, structures, and expression patterns of this and other murine glycosyltransferase genes and their cognate oligosaccharide products, in concert with approaches that utilize transgenic and gene ablation technologies in the mouse.