A Missense Mutation in the FUT6 Gene Results in Total Absence of α3-Fucosylation of Human α1-Acid Glycoprotein

The major α3-fucosyltransferase activity in human plasma is encoded by the gene for fucosyltransferase VI (FUT6). A missense mutation (Gly-739 → Ala) in this gene is responsible for deficiency of enzyme activity in plasma. To examine whether this fucosyltransferase is the sole enzyme responsible for the α3-fucosylation of serum glycoproteins in the liver, we studied the fucosylation of three glycoproteins in sera of individuals with or without inactivated FUT3 and/or FUT6 gene(s) but with a functional FUT5 gene. α1-Acid glycoprotein was used as the principal reporter protein for liver α3-fucosyltransferase activity, because of its high fucose content. In all individuals with the FUT6 missense mutation Gly-739 → Ala in double dose, no fucosylation of α1-acid glycoprotein was found. This α1-acid glycoprotein was not intrinsically resistant to fucosylation, since it was susceptible to in vitro fucosylation using an α3/4-fucosyltransferase isolated from human milk. The same result was found for α1-antichymotrypsin and α1-protease inhibitor. On the other hand in all individuals with α3-fucosyltransferase activity in plasma, α3-fucosylated glycoforms of the glycoproteins studied were found. The degree of fucosylation of α1-acid glycoprotein was correlated with α3-fucosyltransferase activity (Rs = 0.82). These data indicate that the product of FUT6, but not of FUT3 or of FUT5, is responsible for the α3-fucosylation of glycoproteins produced in liver and suggest that this organ is a major source of α3-fucosyltransferase activity in plasma.

The tissue origin of the ␣3-FT activity in plasma is not known. In an ␣3-FT-deficient individual with a congenital kidney anomaly only about 10% of ␣3-FT activity remained in plasma, and this activity appeared to have a myeloid enzyme character (6,14). As the FucT-IV activity in the leukocytes of this individual was normal, the remaining activity in plasma may originate from leukocytes. The majority of the ␣3-FT activity in plasma must therefore have another tissue origin. In the above-mentioned patient, however, there was no expression of Le x antigen in the kidney (14), suggesting that this organ may contribute to the ␣3-FT activity in plasma (6,14). In addition to the kidney the liver is a likely candidate because it expresses the plasma-type enzyme (FUT6 gene product) (6,7,13). This organ is also known to be the source of the majority of plasma proteins as well as of another plasma glycosyltransferase, the ␣6-sialyltransferase (15,16). Plasma ␣3-FT activity is mainly controlled by the FUT6 gene, and the Gly-739 3 Ala inactivating missense mutation results in the deficiency of this ␣3-FT activity (11). We have exploited this experiment of nature to examine in sera of individuals with or without inactivation of FUT3 and/or FUT6 genes, but with functional FUT5 gene (11,17), whether the ␣3-fucosylation of ␣ 1 -acid glycoprotein (AGP), ␣ 1 -antichymotrypsin (ACT), and ␣ 1 -protease inhibitor (PI), three serum glycoproteins secreted by the liver, was affected. This would identify the ␣3-FT responsible for the ␣3-fucosylation of glycoproteins in the liver. AGP was selected as the principal reporter protein in this study because of its high fucose content and because it carries fucose residues in ␣3-linkage only (18 -22). The extent of ␣3-fucosylation of AGP, ACT, and PI was measured in the sera of individuals with and without inactivating point mutation(s) in the FUT3 and FUT6 * This work was supported in part by Grant 8266 from the Swedish Medical Research Council, Grant 57/94 from GREG (France), and Grant 9514111 ACCSV14 from MENESR (France). The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.

EXPERIMENTAL PROCEDURES
Source of Sera and Milk-Sera were obtained from normal healthy blood donors (Indonesian and European ancestry), positive and negative for the Lewis phenotype on erythrocytes and in saliva, as described previously (11). Human milk was obtained from healthy lactating mothers and stored frozen until use.
Crossed Affinoimmunoelectrophoresis (CAIE)-CAIE was performed according to the Bg-Hansen method (23,24). The differently fucosylated glycoforms of AGP, ACT, and PI were separated by electrophoresis of sera (0.5-9.0 l) through the AAL-containing polyacrylamide slab gel (2.5 mg/ml of an AAL preparation with a heme agglutination titer of 1024) (22). Immunoelectrophoresis in a second perpendicular dimension in the presence of a precipitating monospecific antiserum (Dakopatts, Glostrup, Denmark) in a 1% agarose (Bio-Rad) gel resulted in precipitation of the separated glycoforms, which were stained with Coomassie Brilliant Blue R250 (Sigma), and the areas under the curves, representing the relative amounts of glycoprotein, were determined with a Summagraph (ACEDAD D-9000) (25).
Statistics-The correlation between ␣3-FT activity in plasma and reactivity of AGP from these sera with the fucose-specific lectin AAL was determined using the Spearman rank correlation coefficient.

RESULTS AND DISCUSSION
In 26 individuals of different FUT3 and FUT6 genotypes the ␣3-FT activity in plasma was compared with the extent of fucosylation of AGP. Table I shows Lewis a/b antigen expression, mutations in FUT3 (17,(27)(28)(29), and FUT6 (11) genes, ␣3-FT activity, and relative distribution of the differently fucosylated glycoforms of AGP. No inactivating mutation in FUT5 has been found as yet (11).
The extent of ␣3-fucosylation of AGP in sera of these individuals was established by CAIE with the fucose-specific lectin AAL. By this method AGP is fractionated into a nonreactive glycoform A0, containing no fucose, and weakly reactive (Aw) and strongly reactive glycoforms (As), containing fucose in increasing amounts (Fig. 1c). In all individuals without ␣3-FT activity in plasma, 1-8, only AGP-A0 was present (Fig. 1a), i.e. the AGP molecules did not contain fucosyl residues. This AGP-A0 was not intrinsically resistant to fucosylation, as fucose could be incorporated into this nonfucosylated AGP by in vitro fucosylation using an ␣3/4-FT isolated from human milk, as shown in Fig. 1b by the appearance of Aw and As glycoforms.  Refs 17,[27][28][29]. ␣3-FT activity in plasma (in dpm), FUT6 mutations in nucleotide 739 described in Ref. 11, and the relative distribution of the differently fucosylated glycoforms of AGP. A0, AGP fractions that are not reactive with AAL, Aw, AGP fractions that are weakly reactive with AAL; As, AGP fractions that are strongly reactive with AAL (cf. Fig. 1). E, European ancestry; I, Indonesian ancestry. The same results were obtained with two other serum glycoproteins, ACT (Fig. 1, d and e) and PI (Fig. 1, g and h). In vitro fucosylation of PI in individuals with the Gly-739 3 Ala mutation gives an almost similar fractionation pattern as found for PI of normal sera (cf. Fig. 1, h and i). The retarded glycoforms of PI present in sera of individuals with the Gly-739 3 Ala mutation is due to glycan core ␣6-fucosylation of PI, which is also retarded by AAL (22,30).
Only the FUT6 Gly-739 3 Ala mutation in double dose results in lack of ␣3-FT activity in plasma (11). FUT3 alleles containing Thr-1067 3 Ala or Gly-508 3 Ala mutations or the allele containing both Thr-202 3 Cys and Cys-314 3 Thr mutations encode for proteins with no Lewis enzyme activity (17,(27)(28)(29). Inactivating mutations are present in individuals 1-6 in both FUT3 and FUT6 genes. Therefore, from the results obtained with these individuals alone it cannot be concluded whether the product of FUT3 or FUT6 is responsible for the fucosylation of these glycoproteins in liver. However, the fact that AGP, ACT, and PI are not ␣3-fucosylated in individuals 7 and 8, who show the FUT6 mutation in double dose and the FUT3 mutations in single dose, proves that these proteins are only ␣3-fucosylated by the FUT6-encoded enzyme (FucT-VI) and not by the FUT3-encoded enzyme (FucT-III) in liver. This conclusion is further supported by the fact that individuals 9 -14, who have inactivating mutations in their FUT3 gene but not in their FUT6 gene, all have fucosylated glycoforms of the proteins studied in plasma. These results also show that, when FucT-VI is inactive, the ␣3-fucosylation of AGP, ACT, and PI is not taken over by FucT-III or FucT-V. The exclusive action of FucT-VI in the process of ␣3-fucosylation of glycoproteins secreted into the circulation by the liver is in accordance with the known acceptor specificity and tissue expression of this enzyme (3, 6 -13). The activities found in the liver are much higher for FucT-VI than for FucT-III. Furthermore, FucT-III prefers type 1 (Gal␤133GlcNAc) glycan chains, whereas FucT-VI transfers fucose with high preference to type 2 (Gal␤134GlcNAc) glycan chains (3, 6 -13), and the glycan chains that occur on AGP, ACT, and PI are of type 2 (18,19,30,31). The tissue distribution of FucT-V is still unknown (11), but this study indicates that this enzyme is not involved in the ␣3-fucosylation of glycoproteins in the liver, at least as far as it concerns the glycoproteins secreted into the circulation.
In addition our data suggest that the liver is a major source for the ␣3-FT activity in plasma, which has been suggested before (13). In contrast with the reporter serum glycoproteins of individuals 1-8, all deficient in ␣3-FT activity in plasma, these proteins were ␣3-fucosylated (Fig. 1, c, f, and i) in individuals 9 -26, all showing ␣3-FT activity. Furthermore, this ␣3-FT activity in plasma in the latter group of individuals strongly correlates to the extent of fucosylation of AGP (Rs ϭ 0.82, Fig.  2). This correlation in apparently healthy individuals is in accordance with earlier findings in individuals with certain pathologies. Increased activity of human plasma ␣3-FT activity has been described in liver cirrhosis (32) and for various cancers (33)(34)(35). An increase in fucosylation of human AGP (36) and other serum glycoproteins has also been found in these conditions (37)(38)(39). During acute inflammation we found an increase in fucosylation of AGP (22,24) and concomitantly an increase in ␣3-FT activity in plasma (data not shown). So far, no particular pathology was found in individuals showing no ␣3-FT activity in plasma and having nonfucosylated AGP. Several immunomodulatory activities have been described for this glycoprotein that are dependent on its glycosylation (40). Therefore, the implications of the presence of nonfucosylated AGP should be further investigated.
In conclusion, this study shows that AGP, ACT, and PI are ␣3-fucosylated by the product of the FUT6 gene in liver and that no other fucosyltransferase genes, like FUT3 or FUT5, FIG. 1. Reactivity of AGP (a-c), ACT (d-f), and PI (g-i) with AAL. Sera were subjected to CAIE as described under "Experimental Procedures." Only the second-dimension gels are shown. The lower right corner of each gel corresponds to the site of application in the first-dimension gel. Electrophoresis was performed from right to left for the first-dimension and from bottom to top for the second dimension. A0, protein fractions that are nonreactive with AAL; Aw, protein fractions that are weakly reactive with AAL; As, protein fractions that are strongly reactive with AAL. In the absence of AAL all protein molecules were recovered at the site of A0. a, d, and g, an individual deficient in ␣3-FT activity in plasma. b, e, and h, AGP, ACT, and PI from the same individual as in a, d, and g, respectively, after in vitro fucosylation as described under "Experimental Procedures." c, f, and i, an individual with ␣3-FT activity in plasma. take over when inactivating mutations in the FUT6 gene are present.

FIG. 2. Fucosylation of AGP versus ␣3-FT activity in plasma.
The fucosylation of AGP was correlated with the ␣3-FT activity in plasma in the individuals with this plasma activity. The Spearman rank correlation coefficient was determined between these two parameters in individuals with fucosylation of AGP (open circles) and was 0.82 (p Ͻ 0.001) between these two parameters.