"HE JOURNALOF BIOLWICAL CHEMISTRY Vol. 269, No. 33, Issue of August 19, PP 20987-20994, 1994 0 1994 by The American Society for Biochemistry and Molecular Biology, Inc Printed in U.S.A. Molecular Basis for Lewis a(1,3/1,4)- Deficiency (FUT3) Found in Lewis-negative Indonesian Pedigrees*

(Received for publication, March 11, 1994, and in revised form, May 16, 1994)

Rosella MolliconeSQnll,Isabelle ReguigneS, Robert J. Kelly#,Anne Fletcher**, Julie Watt**, Sue Chatfield**, Auda AzizSI, H. Scott Cameron#§, Brent W. Weston§§, John B. Lowe§llll, and Rafael OriolSn From VNSERM U-178, Villejuif 94807, France, the **Red Cross Blood Pansfusion Service, Sydney, New South Wales 2000, Australia, the Svndonesian Red Cross Society, Central Blood Dansfusion Seruices, Jakarta 10002, Indonesia, the $$Department of Pediatrics, Division of Hematology IOncology and the Lineberger Comprehensive Cancer Center, University of North Carolina, North Carolina 27599, and the $Howard Hughes Medical Institute, University of Michigan, Ann Arbor, Michigan 48109

The Le" and Leb human bloodgroup antigens are syn- the majority of Lea-negative individuals was characterized (2). thesized in tissues producing exocrine secretions; they Since then, three red cell phenotypes have been identified in also circulate in plasma, where they are adsorbed by Caucasians: Le(a+b-) 20%, Le(a-b+) 70%, and Le(a-b-) 10% erythrocytes. They are synthesized by twofucosyltrans- (3). These redcell phenotypes are not secondary to two a and b ferases, encoded by Lewis (FUT3) and secretor (FUTZ) alleles at a single Lewis locus as was originally thought, but are loci. This genetic model has been challenged because the result of epistatic interactions of the products of two loci, some erythrocyte Lewis-negative individuals express FUT2 or salivary secretor of ABH (Se,se) and FUTS or Lewis Lewis antigens in saliva. To define the molecular basisof (LeJe),encoding the secretor a(1,2)-fucosyltransferase and the thisapparent discrepancy, we sequenced FUT3 in Lewis a(l,3/1,4)-fucosyltransferase,respectively, in tissues pro- Lewis-negativeindividuals. We identified twosingle ducing exocrine secretions. changes. One,termed L1, yields a Leu-20 +Arg The genetic control of Lewis antigens expression is complex, substitution in the 's transmembrane domain. because FUTS and FUTS each encode a different fucosyltrans- When expressed in COS-7 cells, enzymesubstrate affini- ferase, and thefinal oligosaccharide products are the result of ties are essentially identical to those of wild type. How- the competitive interactions of these on the same oli- ever, the mutant enzyme is found at substantially re- gosaccharide acceptors of type 1 Galpl-.3GlcNAcpl-+R and duced levels in transfected cells. This suggeststhat the type 2 Galpl-.4GlcNAc/3-.R. L1 mutation may alter theGolgi membrane anchoring of The Le' epitope, Gal~l+3(Fucal~+4)G1cNAc~l~R,results the enzyme. It was found alonein double dosein 10 of 30 from the transfer of fucose in a1h4 linkage onto type 1 by the erythrocyte Lewis-negative individuals, nine of whom Lewis enzyme. This terminal product cannot be further glyco- express Lewis antigens in saliva. Therefore, L1 can ac- sylated andis the mainoligosaccharide antigen found in Lewis- count for erythrocytelsaliva-discrepantLewis typing re- positive ABH nonsecretor individuals. It is responsible for the sults. The L2 mutation creates an Ile-356 -+ Lys change in Le" phenotype in saliva and for the Le(a+b-) phenotype on the enzyme's catalytic domain and inactivates the en- erythrocytes. zyme. It was foundin double dosein 18 of 19 individuals The Leb epitope, Fucal~+2Gal~l-.3(Fucal~4)GlcNAcp- bearing the double erythrocyte and salivary Lewis defi- lhR,is the product of the transferof two fucoses onto the same ciency and can account for this phenotype. type 1 precursor, one in a142 linkage onto the terminalgalac- tose transferred by the secretor enzyme and the other ina144 linkage onto the subterminalN-acetylglucosamine transferred In 1946, the first anti-Le" antibody,' agglutinating erythro- by the Lewis enzyme. It isalso aterminal product and the main cytes of about 20% of the population, was described (1).Two epitope found in Lewis-positive, salivary ABH secretor indi- years later, an anti-Lebantibody agglutinating erythrocytes of viduals. It isresponsible for the Leb phenotype in saliva andfor the Le(a-b+) phenotype on erythrocytes. * This work was supported in part by Grant 88.C.0782 from the Min- In Lewis-positive, salivary ABH secretor Caucasians, these- isthre de la Recherche et la Technologie (France) and Grant GM47455 cretor enzyme is highly active and transforms most of type 1 from the National Institutes of Health. The costs of publication of this article were defrayed in part by the payment of page charges. This into H type 1, which is in turn transformed into Leb by the article must thereforebe hereby marked "aduertisement"in accordance Lewis enzyme, giving the Le(a-b+) phenotype on erythrocytes. with 18 U.S.C. Section 1734 solely to indicate this fact, Very few precursor chains are left to make Lea, some of which ll Full-time investigatorsof the CNRS. can be detected in saliva, but usually noton erythrocytes. How- 11 To whom correspondence and reprint requestsshould be addressed: BBtimentINSERM U-178, 16 Av. PaulVaillant-Couturier, Villejuif ever, a fourth erythrocyte phenotype Le(a+b+) hasbeen found 94807, France. Tel.: 33-1-45-59-50-41; Fax: 33-1-46-77-02-33. in Lewis-positive, salivary AJ3H secretors, among Australian I/)/Associate Investigator of the Howard Hughes Medical Institute. aborigines (41,Japanese (5), Chinese (61, and Polynesians (7); it The abbreviations used are: Le", Galpl-3(Fucal--.4)GlcNAc; Le", is due to theexistence of a secretor weak enzyme(Sew) leaving Galpl-,4(Fucal-3)GlcNAc; a(l,3/1,4)-fucosyltransferase,GDP-fu- cose:~-~-N-acetylglucosaminide3/4-a-~-fucosyltransferase; H type 1, a larger proportion of unsubstituted precursor chains which Fucal-2Galpl-3GlcNAc; H type 2, Fucw1+2Gal~l-4GlcNAc; sialyl- can be transformed into Le" by the Lewis enzyme (7). Le", NeuAca2-+3Gal~l~3(Fucal+4)GlcNAc;sialyl-Le", NeuAca2- The cDNA encoding the Lewis a(1,3/1,4)-fucosyltransferase 3Galpl-.4(Fucal-.3)GlcNAc; PCR,polymerase chain reaction; bp, has been cloned (Fuc-TIII; Ref. 8)and registered in the Genome base paifis); RFLP, restriction fragment length polymorphism; ASO, allele-specific oligonucleotide; H gene, FUTI; Se gene, FUT2; Lewis Data Base as FUT3. The sequences of FUT3 alleles of Lewis- gene or Fuc-TIII,FUT3; Fuc-TIV, FUT4; Fuc-W, FUT5;Fuc-TVI, negative individuals in Sweden and Japan have revealed the FUTG. existence of at least threepoint mutations thatlead to a lossof 20987

This is an Open Access article under the CC BY license. 20988 Lewis Fucosyltransferase Deficiency enzyme activity in transiently transfected cells. The first, Thr- Barr virus was made on frozen buffycoats (13), and thewhole genomic 105 + Met, was found to be homozygous in 5 of 18 Swedish DNA was extracted (16) and dissolved at 700 pg/ml in Tris-EDTA, pH blood donors of erythrocyte Le(a-b-) phenotype (9). A second 7.5 (20 m~ "is, 5 mM EDTA). Erythrocyte Phenotypes-Routine red cell agglutination tests were mutation, Gly-170 + Ser, was reported to be homozygous in 4 performed on fresh blood samples with polyclonal anti-A, anti-B (CSL of 8 Lewis-negativeindividuals by twoJapanese teams (10,ll). Ltd.,Melbourne, Australia), lectins anti-H UEez europaeus (Vector, Finally, a third missense mutation, Asp-336 + Ala, was found Burlingame, CAI and anti-A1 Dolichos biflorus (Gamma Biologicals, on a heterozygous girl, who had inherited the Gly-170 + Ser Houston, TX) and six anti-Lea and anti-Leb including the monoclonal mutation from her father andthe Asp-336 + Ala mutation from antibodies from the Second International Workshop on MonoclonalAn- her mother (11). These last two teams also reported another tibodies Against Human Red Blood Cells (Lund, 1990) (17) and goat point missense mutation in the transmembrane region of the polyclonal antibodies (Ortho Diagnostics, Raritan, NJ). ABH and Lewis Salivary Phenotypes-Salivas from all family mem- enzyme, changing Leu-20 + kg, but they detected Lewis en- zyme activity in COS cells transfected with this FUT3 allele bers were collected, boiled for10 min within 2 h of collection, and cen- trifuged, and the supernatants were kept frozen until tested. Tests were and concluded that this mutation was irrelevant for enzyme performed by inhibition of the corresponding erythrocyte agglutination activity (10, 11). reactions with U. europaeus, anti-A, anti-B, anti-Le", and anti-Leb. The Lewis a(1,3/1,4)-fucosyltransferaseenzyme is expressed Molecular Cloningof FUT3 Alleles and Corresponding FUT3 Expres- in epithelial cells producing exocrine secretions and its oligo- sion Vectors from a Lewis-deficient Individual-PCR was used to am- saccharide final products (Le", Leb, Le', or Ley) are shed into plify the coding and 3' flanking regions of the FUT3 gene of the B4 secretions, mainly as glycoproteins and intoplasma, mainly as individual, who is plasma a(l,3)-fucosyltransferase-deficient and glycolipids, but there is no detectable Lewis enzymeactivity in Lewis-negative onerythrocytes and saliva. Genomic DNA1200 ng) was amplified in a DNAThermal Cycler (Perkin-Elmer), using a Gene-Amp plasma, irrespective the Lewis phenotype of the individual. of PCR kit (Perkin-Elmer) and 0.5 p of each primer. The PCR program In contrast, large amounts of plasma a(1,3)-fucosyltransferase included a 5-min, 94 "C initial denaturation step, followed by 30 cycles encoded by the FUTG gene (Fuc-TVI; Ref. 12) are found in consisting of 1.5 min at 94 "C and 3.5 min at 72 "C.The sense primer of plasma. FUT3 anneals to nucleotides 1-31, beginning immediately with the In alarge series of plasma a(l,3)-fucosyltransferase-deficient ATG codon. The sequence of this primer contains additional nucleotides individuals in Indonesia, we found that this enzyme is inacti- (lowercase)at its5' end including an EcoRI restriction site (underlined): vated by two single base pair mutations inthe coding regionof 5'-gcgcgaattcATGGATCCCCTGGGTGCAGCCAAGCCA3'.__ The antisense primer is complementary of nucleotides 1125-1096, and the FUTG. A missense mutation yields Glu-247 3 Lys and a non- extraneous nucleotides (lowercase) at the 5' end contain an XbaI re- sense mutation yields Tyr-315 + stop, which truncates the striction site (underlined): 5'-gcgc~GGCAGATGAGG'M'CCCG- COOH terminus of the enzyme by 45 amino acids.Each muta- GCAGCCCAGGCAC-3' (sequence numbering according to Ref. 8). tion, when present in double dose, produce a plasma a(1,3)-fu- The PCR product was digested with EcoRI and XbaI, and the 1125-bp cosyltransferase-deficient phenotype. Furthermore, the muta- fragment was cloned betweenthe EcoRI and XbaI restriction sites of the tion Glu-247 + Lys in double dose cosegregatedwith the plasma mammalian expression vector pCDNAI (Clontech). Twelverepresenta- enzyme defect in the five Indonesian pedigrees tested (13). tive plasmids were selected and both strands of their inserts were sequenced in their entirety, by dideoxy chain termination, using T7 The of three human a(l,3)- (FUT3, DNA polymerase (Sequenase, U. S. Biochemical Corp.). We first used FUT5 also known as Fuc-TV (141, and FUTG) share 85-90% oligonucleotides correspondingto flanking plasmid sequences, and sub- identity (12), they are expressed in different tissues and have sequently we used primers corresponding to internal sequences of the different acceptor specificity patterns. FUT3 and FUTG are wild type FUT3 gene (8). closely linked, within 13 kilobases of each other on the short The wild type pFUT3 vector, containing a sequence of 1125 bp, was arm of 19 (15). Typing of 198 Indonesian random prepared by amplifylng the wild type FUT3 cDNA (81. To diminish the risk of PCR errors, 50 ng of cloned wild type cDNA were used as tem- blood donors has shown that 17 of 18 plasma a(l,3)-fucosyl- plate for 20 cycles of PCR with the same program. After amplification, -deficient individuals are alsoLewis-negative on the fragment was digested with EcoRI andXbaI and cloned in pCDNAI, red cells. These data are compatible with FUT3 and FUTG in the same way as the mutated clones. Inserts of two representative being in linkage disequilibrium (13). clones of pFUT3 were sequencedto confirm the absence of PCR-induced We have now studied the five original plasma a(1,3)-fucosyl- mutations. One plasmid containing mutations L1 and L2 (see "Re- transferase-deficient families (13), plus three new Indonesian sults"), but otherwise identical to a wild type pFUT3 allele, was desig- nated pL1,LZ. Constructions of other mutant FUT3 expression vectors families ascertained by a propositus with the FUTG defect, for were accomplished via restriction fragment interchange. Vector pL1, both FUTG and FUT3 mutations and corresponding pheno- containing only the L1 mutation, was constructed by replacing the wild types. The molecular basis of this FUT3 defect is shown here to type AflIII restriction fragment encompassing this sequence positionin be a consequence of two point mutations inthe coding regionof pFUT3, with the corresponding AflIII fragment isolated from the plas- FUT3. One, termed L1, is identical to the Leu-20 + Arg mu- mid pLl,L2. The AflIII restriction enzyme cuts at position 1532 in the tation reported above (10, ll).The other, termed L2, is a new plasmid vector, upstream of the cytomegalovirus promotorand at posi- tion 656 in thecoding regionof the gene, after the Ll mutation. Vector missense mutation, in the catalytic domain of the protein. We pL2 containing only the L2 mutation was prepared by insertion of the show here that theL1 mutation in double dose produces weak wild type Afl111 restriction fragment into the AflIII-digested pL1,LB Lewis enzyme activity characterized by erythrocyte Le(a-b-) vector. The structure of each of these constructs was confirmed by phenotype with secretion of Lewis antigens in saliva; the L2 restriction endonuclease digestion, by DNA sequencing across the re- mutation, in double dose, inactivates the enzyme giving the striction sites used for cloning, and by DNA sequencing across each Lewis-negative phenotype on both erythrocytes and saliva. polymorphic nucleotide position (Fig.I). Detection of the Single Base Differences GiuingLI and L2 Alleles of the FUT3 Gene, Using Allele-specificOligonucleotides CM0J"he EXPERIMENTALPROCEDURES FUT3 alleles of individuals from the eight pedigrees, the wild type Plasma or(l,3)-Fucosyltransferase-deficientPedigrees-Eight defi- pFUT3 construct, and the constructs containing the L1 and/or L2 mu- cient plasma a(1,3)-fucosyltransferase blood donors from Jakarta were tated alleles were amplified by PCR with the same primers used for selected forthis study. Family trees were drawn giving a serial numeric cloning FUT3. Hybridization with wild type and mutated AS0 probes order to each individual, starting from the oldest members. Missing (Table I) was performed as described (13). numbers in the pedigrees indicate that the individuals were not avail- fiansfection of cos-7Cells-Cells were transfected with DE--dex- able or that they refused to participate in the study. Pedigrees were tran (18).An expression vector, containing the coding region of the bac- validated by segregation of HLA haplotypes using serological and RFLP terial chloramphenicol acetyltransferase (pCDM7-CAT)(19), was simul- analysis by the Tissue Typing Department of the Red Cross Blood taneously transfected to allow normalization for transfection efficiency. Transfusion Service (Sydney, Australia). Transformation by Epstein- Transfected cells were harvested after a 72-h expression period (12). Lewis DeficiencyFucosyltransferase 20989

1 Kb pFUT3 361 aa

f f i. 154 185 1 AflIII AfllII L1 L,2 pLI,L2 L1 L2 I1 (Leu 20 - Arg) (Ile 356 + Lys) FIG.2. Location of L1 and L2 mutations and predictedglyco- sylation sites within theLewis a(l,3/1,4)-fucosyltransferase.The Lewis enzyme is depicted as a rectangle; its NH, transmembrane seg- ment, containing the L1 mutation, is represented by the small gray rectangle. Positions and predicted amino acid sequence changesfor each mutation are indicated below the scheme. Predicted N-linked glycosy- FIG.1. Mutated FUT3 allele expression vectors.The FUTB gene lation sites are indicated by the symbol I/ and the residue number for is represented by open rectangles representing its coding region (large the corresponding asparagine. rectangle)and 3'-untranslatedregion (small rectangle to the right).The line represents adjacent pCDNAI vector sequences. The relative posi- Lewis-negative individual (B4)was used for cloning the FUT3- tions of the mutations within thewild type sequenceare indicated above deficient allele. An XbaI-EcoRI PCR fragment of 1125 bp was the coding region of each vector, whose designation is given at right. Vectors with one or two mutations were constructed with restriction amplified from genomic DNA with FUT3-specific primers. The fragment interchange procedures using AflIII restriction sites as de- inserts of 12 recombinant plasmids were sequenced and found tailed in "Experimental Procedures." The relative positions of the AflIII to differ from the wild type FUT3 allele (8) by single base sites within thevector sequence or within the FUT3 coding region are changes at two positions in the coding region (Fig. 2). indicated by the arrows. The L1 mutationis a T--j G substitution at position 59. This

Flow CytometryAnalysis-Transfected COS-7 cells were labeled with sequence alteration yields a leucine --j arginine substitution different monoclonal antibodies: afinity-purified anti-H, anti-Le", and (Leu-20 + Arg) in the transmembranedomain of the protein. anti-Leb (Chembiomed, Alberta Research Council, Edmonton, Canada; The 12 clones sequenced contained this mutation, suggesting Ref. 17); anti-Le" (SSEA1, D. Solter, WistarInstitute, Philadelphia,PA); that theB4 individual is homozygous LlL1. anti-sialyl-Le" (19.9 Centocor, Malvern, PA); and anti-sialyl-Lex (CSLEX1, UCLA Tissue Typing Laboratory, Los Angeles, CA; Ref. 20). The L2 mutation is a T + A substitution at position 1067. After incubation with the first antibody, the cells were washed and This sequence alteration changes an isoleucine codon to a ly- stained with ffuorescin isothiocyanate-conjugated goat anti-mouse im- sine codon (Ile-356 + Lys) 6 amino acids before the COOH munoglobulins and analyzed with a Becton Dickinson FACScan (12). terminus of the protein. Eleven of 12 clones contained the L2 a(1,3/- and a(l,4)-FucosyltransferaseEnzyme Activities of Dans- mutation, suggesting thatL2 may also be homozygous in this fected COS-7 Cells with DifferentAcceptors-Fucosyltransferase assays individual. Homozygosity was further ascertained by the AS0 were performed with GDP-['4Clfucose (Amersham Corp., 300 mCil mmol) and Galpl--.4GlcNAc, Galpl-.4Glc, Galpl-3GlcNAc, Fucal- technique. Only the two L1 andL2 mutant AS0 probes (Table 2Galpl-+4Glc,and NeuAca2-3Galfi1-.4GlcNAc (Oxford Glycasys- I) hybridized with genomic DNA, from the B4 individual, am- tems, Oxford, United Kingdom) as described previously (12). plified with the same primersused for cloning FUTB, whereas Plasma a(l,3)-Fucosyltransferase-Enzyme activity was measured neither of the two wild type AS0 probes hybridized, confirming with H type 2 acceptor, synthesized as a 8-methoxycarbonyloctyl glyco- that this person is homozygous for both mutations. side, Fuca1-+2Galp1-.4GlcNAcp1-.(CH,),COOCH, (Chembiomed, Al- Correlation of Lewis Genotypes with the Phenotypic Expres- berta Research Council, Edmonton, Canada) by the Sep-Pak C,, product isolation procedure (21). sion of Lewis Antigens on Erythrocytes and Saliva-The same Kinetics of a(1,3)- and a(l,4)-Fucosyltransferase Enzyme Actiuities- AS0 probes were usedto ascertain the L1 andL2 genotypes of Kinetics were measured in 1%Triton X-100 extracts from transfected all family members typed for Lewis antigens on erythrocytes COS-7 cells using GDP-fucose and the 8-methoxycarbonyloctyl glyco- and saliva (Table 11). sides H type 2 and H type l, Fuca1-,2Galpl-tSGlcNAcfil-.(CH,),- It was predicted that inactivating mutations in the coding COOCH, (Chembiomed). Enzyme assays were performed with amounts region of FUT3, responsible for lack of expression of the fuco- of protein corresponding to initial velocity and giving similar fucose transfer for each cell extract (21). syltransferase products, should be homozygous. Therefore, the Detection of Specific mRNA Expression--Total RNA was isolated individual results ofASO typing were sorted for the presence of from transfected COS-7 cells by the guanidium isothiocyanate-CsC1 homozygous L1L1 or L2L2 genotypes (Table 11). Four groups procedure (22). For RNA PCR analysis, 5 pg of total RNA were treated were defined: I, two individuals were homozygous L2L2 only with 15 units of DNase I (Promega) for 15 rnin. The synthesisof the first and did not express Lewis antigens, either on erythrocytes or cDNA strand was then performed on a final volume of 20 pl with the saliva; 11, 16 individuals were homozygous L2L2 and LlL1, oligo(dT)12-18primer and a cDNA synthesis kit (RPN 2275, Amersham Corp.). The reaction mixtures were incubated for 2 h at 37 "C. The and all of them also had Lewis-negative phenotypes on eryth- double-stranded cDNA was synthesized by PCR using 38 cycles of the rocytes and saliva; 111, 10 individuals were only homozygous above described program. The Gene-Amp PCR kit with 5-pl aliquots of L1L1 andwere also Lewis-negativeon erythrocytes, but nine of each reverse transcriptase mixture and 0.5 PM of each of the sense them expressed Lewis antigens in saliva andone (A61 was neg- primer used for cloning FUTB alleles and the antisense primer 5'- ative on both erythrocytes and saliva; IV, 25 individuals had CAGCCAGCCGTAGGGCGTGAAGATGTCGGA-I, corresponding to neither L1 nor L2 mutations in double dose, and they all ex- the complement of nucleotides 600-571. The amplified 600-bp frag- ments were analyzed in 1.6% NuSieve agarose gel. For RNA dot-blot pressed Lewis antigens insaliva. Only two individuals from this hybridization, 20 pg of total RNA were denatured in 1 mM glyoxal (23) group typed as erythrocyte Le(a-b-). One expressed A and Le" and then immobilized onto nylon membrane (Hybond-N, Amersham). in saliva(G9), suggesting that hemay be a weak secretor(Sew) The dot-blot was prehybridized at 42 "C in the presence of 50% form- (7); the other expressed H and Leb in saliva (C4), and it isnot amide and hybridized with [a-32P]dCTP-labeled pFUT3 probe (Mega- clear why he had erythrocyte Le(a-b-) phenotype (Table 11). prime kit from Amersham) for 24 h at 42 "C, with 40% formamideand All the homozygous L2L2 (groups I and were plasma 10% dextran sulfate. The blot was rinsed once in 2 x SSC, 0.5%SDS, 11) washed twice in the samebuffer for 15 min at room temperature, and c~(1,3)-fucosyltransferase-deficient(<300 dpm), while only 3 of washed twice in 0.1 x SSC, 0.1% SDS at 65 "C for 20 min. 35 individuals (9%) without the L2 mutation in double dose were plasma a(l,3)-fucosyltransferase-deficient (H11 from RESULTS group I11 and J1 and 53from group IV, Table 11). Missense Mutations within the FUT3 Alleles of a Lewis-neg- A significant correlation was alsoobserved between the pres- dive Individual-DNA from an erythrocyte Le(a-b-), salivary ence of L2 in double dose and incidence of natural anti-Lewis 20990 Fucosyltransferase Lewis Deficiency TABLEI and L2 mutations yield the Lewis-deficient phenotypes, three Allele-specific oligonucleotides (ASO) used to determine Ll and L2 series of experiments were performed (i) flow cytometry anal- genotypes of all members ofthe eight Indonesian families ysis of COS-7 cells transfected with FUTB expression vectors Positions 59 (L1) and 1067 (L2) are underlined. and stained withmonoclonal antibodies specific for cell surface Mutation Wild Mutation type Mutated allele oligosaccharide antigens synthesized by the Lewis enzyme, (ii) L1 GCC GCA CTG CTA TTT GCC GCA C@ CTA TTT measurement of the specific activity of the recombinant fuco- L2 CGC AGC ATA GCG GCT CGCAGC GCG GCT syltransferases in these cells toward different low molecular weight oligosaccharide acceptor substrates, and (iiij determi- antibodies (anti-Le" or anti-Leb)in plasma. Sevenof 18 homozy- nation of the apparentMichaelis constants of these recombinant gous L2L2 (groups I and 11) had anti-Lewisantibodies in enzymes for GDP-fucose, H type 1 and H type 2 substrates. plasma (39%), while only one of 35 individuals (3%) without COS-7 cells were transfected with expression vectors encod- this mutation in double dose (groups I11 and N)had weak ing wild type FUT3 (pFUT3), or mutant forms of FUTS con- anti-Leb activity (G9, Table 11). It is interesting to note that taining the L1 substitution (pLl), theL2 substitution (pL2), or none of the individuals ingroup I11 had anti-Lewis antibodies both changes (pLl,L2). After a %day expression period, these in plasma, in spiteof the fact that they alltype as erythrocyte cells, or cells transfected with a negative control vector (pCD- Le(a-b-), confirming that they are genetically Lewis-positive, NAI), were stainedwith specific antibodies and subjected to flow as suggested by the presence of Lewis antigens in saliva 9in of cytometry analysis (Fig. 4). Cells transfected with thewild type 10 individuals from this group111. FUT3 vector (pFUT3) express allfour Lewis antigens, whereas Segregation of L1 and L2 Mutations-The a(1,3)-fUcosyl- cells transfected with thenegative control vector (pCDNAI) do transferase was measured in plasma, but Lewisthe a(1,3/1,4)- not express these antigens. Cells transfected with vectors con- fucosyltransferase enzyme activity was only deduced from the taining the L2 mutation alone (pL2) or L2 in conjunction with presence of Le" and/or Leb antigens insaliva, becausethe Lewis the L1 mutation (pLl,L2)do not expressdetectable levels of the enzyme in the available saliva samples was heat inactivated four Lewis antigens. Similarly, we found no detectable a(1,3)- during preparation. or activity in extracts preparedfrom Cosegregation of lack of expression of Lewis antigens in sa- ol(l,4)-fucosyltransferase liva with the L2 mutation in double dose was confirmed in all cells transfected with pL2 or with pL1,LB (Table 111). families, with the exception of the A6 individual, who was The specific FUT3 mRNA alleles were detected in all the Lewis-negative on both saliva and erythrocytes, but was het- transfected cells by reverse transcriptase-PCR andby hybrid- erozygous at the L2 position (Table I1 and Fig. 3). This indi- ization with the FUTS probe, confirming that pL2- and pL1,LS- vidual was homozygous L1 and expressed very low levels of transfected cells contained the corresponding FUT3 alleles, but plasma a(f,3)-fucosyltransferase (440 dpm). The Lewis-nega- the mutationL2 inactivates the cognate enzyme. tive phenotypes of this A6 individual could be secondary to By contrast, the pL1vector determines expression of the Lea incomplete maturation of expression of Lewis antigens, be- and sialyl-Le" antigens on transfected COS-7 cells, although cause hewas only 5 years old when the sampleswere collected the fraction of cells that express these determinants is lower (24). However, in this particularfamily, it is possible that other than the fraction of antigen-positive pFUT3-transfected cells. mutations occur in the FUT3 gene, which may result in low However, the Le" and sialyl-Le" antigens areexpressed on pL1- enzyme activity, as has been proposed for the closely linked transfected cells at levels barely above background. This obser- FUTG gene product (13). vation is at odds with enzyme activity assays done on extracts Segregation of the L1 mutation in double dose with the prepared from pLl-transfected cells (Table 111). These results erythrocyte Le(a-b-) phenotype and salivary Lewis-positive indicate that the L1 mutantenzyme is, in fact, capable of uti- phenotype is illustrated in family B, where B2 and B3 indi- lizing type 2 acceptor substrates in vitro, although apparently viduals are homozygous L1 and both have the erythrocyte not efficiently within the cell. Le(a-b-) phenotype with salivary Lewis-positive phenotype. We explored this issue furtherby comparing the kineticprop- The remaining membersof this family are eitherheterozygous erties of the wild type FUTS enzyme and the mutant L1 en- L1 andLewis-positive on erythrocytes (Bl,B5, B7, B8, and B9) zyme (Table N).We found that both enzymes maintain similar or double homozygous L1 and L2 and Lewis-negative on both apparent Michaelis constants for H type 1, H type 2, and GDP- erythrocytes and saliva(B4, B10, and B11) (Table I1 and Fig. 3). fucose. However, the specific activity (and Vm,j of the fucosyl- All individuals with thedouble homozygous L1 andL2 gen- transferase found in pLl-transfected cells was consistently otype had the same salivary and erythrocyte Lewis-negative 2-3-fold lower than theactivity in pFUT3 transfectants (Tables phenotypes in the eightfamilies. The inactivatingeffect of the 111 and IV). L2 mutation indouble dose was dominantover the effect of L1, Taken together, these results suggest that the L1 mutant as illustrated familyin H, where all members werehomozygous fucosyltransferase exhibits a catalytic activitythat is similar to LlL1, and thesix members heterozygousL2/* expressed Lewis that for the wild type enzyme, at least with respect to its af- antigens in saliva (H2, H3, H6, H9, H11, and H15), while the finity (K,) toward three different substrates. However, the L1 six members homozygous L2L2 were all negative for Lewis mutation seems to decrease the overall specific activity of the antigens in saliva (Hl,H5, H7, H12, H14, and H16) (Table I1 enzyme. The wild type Lewis a( 1,3/1,4)-fucosyltransferaseis and Fig. 3). less active on type 2 acceptor substrates; it is therefore not The plasmaa( 1,3)-fucosyltransferase-negativephenotype CO- surprising that diminution in expression efficiency in trans- segregated with the Lewis-negative phenotype and homozy- fected cells is associated with a greater loss in expression of cell gous L2L2 genotype in most cases. However, there were three surface type 2 antigens than type 1 antigens. exceptions, J1, 53, and H11, who were plasma cy(1,3)-fucosyl- Since the L1 mutation is localized to the transmembrane transferase-deficient but expressed Lewis antigens on erythro- segment of this enzyme, its decreased activity might be a con- cytes and saliva and were heterozygous L2, suggesting a cross- sequence of inappropriate membrane insertion. To test this ing over between FUTB and FUTG loci in the H family and hypothesis, total enzyme activityof the culture mediumof pL1- another in the ancestryof J1 (Fig. 3). transfected cells was tested and wasfound to contain 2.3 times Lewis Enzyme Activity of COS-7 Cells Dansiently Dans- more activity than thecorresponding cells, whereas the differ- fected with FUT3"To determine themechanisms by which L1 ence was only of 1.5 times for wild type pFUT3-transfected Lew is Fucosyltransferase Lewis Deficiency 20991 TABLEI1 AB0 and Lewis phenotypes and Lewis genotypes of all family members (Fig. 31 Erythrocyte phenotypes were obtainedby direct hemagglutination, and saliva phenotypes were obtainedby inhibition of hemagglutination of the anti-A, -B, -H, -Lea, and -Lebreagents. Lewis genotypes at the 59 (L1) and 1067 (L2) positions were determinedbyAS0 tests(* indicates wild type). The presenceof natural anti-Lea and anti-Leb antibodies u(l,3)-fucosyltransferaseand activity expressedas dpm oftransferred[14Clfucose (a-3-FT) were determined in plasma. After sorting for L2 and L1 homozygous genotypes, four groups could be distinguished: I, homozygous L2L2; 11, homozygous L2L2 and Ll/Ll;111, homozygous Ll/Ll; Iv, all others. Individuals within each group were sortedby increasing amount of plasma a(l,3)-fucosyltransferase activity (FUT6). _____~~~~~~~ ~ ~ ~ ~ Identity Erythrocytes - Saliva Plasma Genotype Lewis Group Family No. AB0 Le(a,b) A B H Le" Leb L1 L2 Antibody a-3-FT I E 3 A1B A B 212 Leb 108 E 5 B B 212 122 I1 H 12 B B 212 Leb, Le" 26 H 7 B B 22 43 H 14 B B 212 Leb, Le" 55 H 1 0 212 84 H 5 B B 212 87 D 1 A1B A B 212 Leb 109 B 10 0 212 Le" 120 D 2 A1 A 2/2 121 J 4 AI A 2/2 124 H 16 0 212 128 C 1 0 212 130 A 3 0 2/2 Le" 132 B 4 A1 A 212 172 B 11 AI A 212 Leb, Le" 178 G 4 AIB 212 198 G 5 A1B 212 232

I11 H 11 B B *I2 94 A 6 A1 A */2 440 H 9 B B */2 1430 E 4 0 */2 1810 H 15 B B */2 2030 H 6 0 */2 2790 B 3 0 */2 3490 B 2 AI A */2 4300 H 3 B B */2 4720 H 2 B B */2 15160 Iv J 3 A1 A */2 52 J 1 A1 A */2 69 A 4 0 */2 576 A 1 0 */2 910 A 2 A1 A */* 1600 J 2 A1 A */2 1665 B 9 AI A */2 2480 E 1 AI A */2 2830 G 9 A1 A *I2 Le * 2860 B 8 A1 A */2 2960 B 5 0 */2 3150 A 5 0 */* 3230 B 1 A2 A */2 3500 C 6 0 */2 3530 C 5 0 */2 4050 C 3 A1 A */2 4060 C 4 0 */2 4080 G 8 A1 A */2 4080 B 7 AI A */* 4280 C 2 A1 A */* 5290 G 3 A1 */* 6400 G 7 B *I2 10970 G 2 Alb A B */2 11260 G 1 0 */* 13260 G 6 0 */* 15660

cells. This result favors the proposed hypothesis but does not an all or none tissue expression of two fucosyltransferases, account for all the difference between pL1 and pFUT3. New encoded by FUT2 and FUT3 loci (25) has recently been chal- experiments have to be performed in carefully controlled cell lenged, because Lewis enzyme activity and Le" or Leb antigens viability conditions, with polarized epithelial cells, since some were detected in saliva of 3 of 6 individuals with the erythro- transfected COS-7 cells dye and can liberate theirenzyme con- cyte Le(a-b-) phenotype, who were called nongenuine tent in the culturemedium. Le(a-b-) individuals for this reason (26). The occurrence of Lewis antigens in saliva of individuals DISCUSSION with Le(a-b-) erythrocyte phenotype can now be seen to be due The original genetic model of Lewis antigens resulting from to the mutation L1. This mutation in double dose was found epistatic interactions of the Le-le and Se-se loci considered as (without thehomozygous L2 mutation) in 10 of 30 erythrocyte 20992 Lewis Fucosyltransferase Deficiency

A B

13TIt li 2 , 3*jib

576 3230

6 -i 1440

4 J 'e 109 69 tl I I 4 5 3 4 6 5

4080 4050 3530 40504060 4080 122 121 1810 1 08 124 tt 44 di 4t it

0 H

1 1 2 3 T 6 7 13260 11260 *6400 1 198 g* 232 15660 10970

8 14

4920 2860 55 2030 128 it t4 tt id it FIG.3. Segregation of FUT3 alleles in eight Indonesian families. Solid symbols for males (W) and females (01) correspond to salivary Lewis-negative phenotype. The halfblack symbols(0, (3) indicate Lewis-positive individuals, assumed to be Lewis heterozygous from the structure of the family tree. Open symbols(0, correspond 0) to Lewis-positivephenotypes, which can be either homozygous or heterozygous. Across (t)inside symbols indicates that the person died before the study. The plasma a(l,3)-fucosyltransferaseactivity of each individual is expressed below the symbol, as disintegrationdminute of fucose incorporated onto H type 2. The L1 (A, A) and L2 (V, V) polymorphic alleles are displayed under enzyme activity numbers. Mutated alleles are represented by solid triangles and wild type alleles by open triangles.

Lewis-negative individuals, and 9 of them expressed Lewis an- the Golgi, which synthesizes the Lewis glycolipid epitopes shed tigens in saliva. Therefore, this kind of mutation in the trans- into the plasma compartment. Enzyme kinetics favor this hy- membrane domain is probably responsible for the Le" pheno- pothesis since similar apparent K, values were found in cell type, previously described in individuals typing as Le(a-b-) on extracts of COS-7 cells transfected with wild type pFUT3 and erythrocytes and Lewis-positive in saliva (27). pL1 constructs, but the pLl-transfected COS-7 cells had con- The reduction in expression of Le" and Leb antigens on red sistently lower V,,, values, suggesting a 2-3-fold reduction of cells, induced by the L1 mutation,might be related toits loca- intracellular enzyme in the case of L1 mutant. tion in the putative transmembranedomain. The change of a In addition, the so-called genuine Lewis-negative individuals neutral leucine for a charged arginine in thishydrophobic do- are not completely negative, since they express small amounts main can impair the proper anchoring of the enzyme in the of Lewis epitopes inother tissues (26, 28). Indeed, trace Golgi membrane. The expected consequences of such an im- amounts of glycolipids bearing thepentasaccharide Le" and the pediment would be a decreased amount of enzyme resident in hexasaccharide Leb were identified in genuine Lewis-negative Lewis Fucosyltransferase Deficiency 20993 TABLEV 0 Anti-H Summary ofall reported mutations affecting thecoding region of H Anti-Lea FUT3, which can modify the cognate Lewis enzyme activity and give 0 Anti-sLea erythrocyte Le(a-b-l phenotype 0 Anti-Lex Nucleotides Amino acids Geographic W Anti-sLex Ret-. I Position"Change Position" Change area

59T + 20G Leu -, ArgIndonesia This study (L1) Japan (IO, 11) 314 C -> T Thr105 + SwedenMet (9) 508 G-170 A Gly 3 JapanSer (10, 11) 1007A -,336 C Asp -, Ala Japan (11) 1067T + 356A Ile + Lys IndonesiaThis study (L2) Position numbering according toRef. 8.

Alternatively, other enzymes encoded by FUT4, FUT5, or an- other stillunknown gene, maybe responsible for the synthesis

0 of the small amountsof Le" or Leb epitopes found in genuine Lewis-negative individuals. The presence of small amounts of Lewis antigens in Lewis- FIG.4. Flow cytometry analysis of COS-7 cells transfected with negative individuals shed light on one of the most mysterious FUT3 expression vectors. COS-7 cells were transfected and sub- differences between AB0 (31) and Lewis systems. The AT30 jected to flow cytometry analysis as described under "Experimental alloantigens are a clear-cut, all-or-none alloantigensystem, Procedures," using vectors indicated below the histograms. Cells were where all the individuals lacking A or B antigens have the stained with monoclonal antibodies specific for Le" (Lea), sialyl-Le" (sku), Le" (Lex),and sialyl-Le" (sLex)determinants, or with negative corresponding anti-Aor anti-B alloantibodies in plasma. Thisis control antibodiesas anti-H andanti-Le" (not shown). Data are normal-not the case amongLewis-negative individuals, wherethe pres- ized for transfection efficiencies, as determined by an internal control ence of anti-Lewis antibodies is not the rule. Only a small chloramphenicol acetyltransferase assay. proportion of Lewis-negative individuals have natural anti- TABLE111 Lewis antibodies in plasma, and they areusually of lower titer Enzyme activity inpmollmglh on different oligosaccharide acceptors than theanti-A or anti-B naturalantibodies (32). Homozygous of extracts of transfected COS-7 cells with FUT3 constructs, performed LlL1 have Lewis antigens in secretions; therefore, anti-Le"or as described under "Experimental Procedures"(12) anti-Leb would be autoantibodies, and, as expected, these an- Lacto-N-bioseI, Galpl+3GlcNAc; 2'-fucosyllactose, Fucal+PGal- tibodies were not found (Table 11). However, even the so-called pI+4Glc; lactose, Galpl-4Glc; LacNAc, Galpl-4GcNAc; sialyl-Lac- genuine Lewis-negativeindividuals can stillmake small NAc, NeuAca2+3Galpl+4GlcNAc. amounts of Lewis antigens in tissues, and thisfact accountsfor Lacto-N- 2'-Fucosyl- Lactose LacNAc Sialyl- Constructs bioseI lactose LacNAc the low incidence of natural anti-Lewis antibodies amonggen- uine Lewis-negative individuals. pFUT3 1990 221410,880 2380 5020 The Lewis enzyme is the only a(1,3)-fucosyltransferase that PL1 820 1655 3995 722 620 also has 1,4)-fucosyltransferase activity and therefore can PL2