Human and
Drosophila
UDP-galactose transporters transport
UDP-
N
-acetylgalactosamine in addition to UDP-galactose
Hiroaki Segawa*, Masao Kawakita and Nobuhiro Ishida
Department of Physiological Chemistry, The Tokyo Metropolitan Institute of Medical Science (Rinshoken), Honkomagome,
Bunkyo-ku, Tokyo, Japan
A putative Dros ophila nucleotide sugar transpor ter w as
characterized and shown to be the Drosophila homologue o f
the human UDP-Gal transporter (hUGT). When the
Drosophila melanogaster UDP-Gal transporter (DmUGT)
was expressed in mammalian cells, the transporter protein
was localized in the Golgi membranes and complemented
the UDP-Gal transport de®ciency of Lec8 cells but not the
CMP-Sia transport de®ciency of Lec2 cells. DmUGT and
hUGT were expressed in Saccharomyces cerevisiae c ells in
functionally active forms. Using microsomal v esicles isolated
from Saccharomyces cerevisiae expressing these transporters,
we unexpectedly found that both hUGT and DmUGT
could transport UDP-GalNAc as well as UDP-Gal. W hen
amino-acid residues that are conserved among human,
murine, ®ssion yeast a nd Drosophila UGTs, but are distinct
from corresponding ones conserved among CMP-Sia
transporters (CSTs), were substituted by those found in
CST, the mutant t ransporters were still active in transporting
UDP-Gal. One of these mutants in w hich Asn47 was sub-
stituted by Ala showed aberrant intracellular distribution
with concomitant destabilization of the protein product.
However, this mutation was suppressed by an Ile51 to Thr

organisms including yeasts [4±7], protozoa [8], worms [9],
and mammals [10±16]. These genes encode structurally
related hydrophobic membrane proteins. The UDP-Gal
transporter (UGT), UDP-GlcNAc transporter (UGlcN-
AcT) and CMP-Sia transporter (CST) show considerable
similarity with each other, but have distinct substrate
speci®cities. The mechanisms underlying the speci®c sub-
strate reco gnition are intriguing, but remain obscure.
Alignment o f new members o f the NST f amily with other
family members m ay offer clues about the mechanisms of
substrate recognition by NSTs.
In this communication, we d escribe the m olecular
cloning and characterization of a Drosophila homologue
of mammalian NST (DmNST), which we found in the
D. melanogaster expressed sequence tag (EST) database.
The deduced amino-acid sequence of DmNST showed
moderate similarity to hUGT, hUGlcNAcT and hCST, and
heterologous expression in yeast allowed us to identify the
Correspondence to M. Kawakita, Department of Applied Chemistry,
Kogakuin University, 1-24-2 Nishi-Shinjuku, Shinjuku-ku, Tokyo
163-8677, Japan. Fax: + 81 3 3340 0147, Tel.: + 81 3 3340 2731,
E-mail:
Abbreviations: NST, nucleotide sugar transporter; UGT, UDP-
galactose transporter; UGlcNAcT, UDP-N-acetylglucosamine trans-
porter; CST, CMP-sialic acid transporter; UDP-Gal, UDP-galactose;
UDP-GlcNAc, UDP-N-acety lglucosamine; CMP-Sia, CMP-sialic
acid; UDP-GalNAc, UDP-N-acetylgalactosamine; DmNST,
Drosophila melanogaster NST; EST, expressed sequence tag; hUGT,
human UDP-galactose transporter; hCST, human CMP-sialic acid
transporter; hUGlcNAcT, human UDP-N-acetylglucosamine trans-

), UDP-[1-
3
H]Glc (15
Ciámmol
)1
), UDP-[6±
3
H(N)]GlcNAc (60 Ciámmol
)1
),
UDP-[1-
3
H]GlcA (15 C iámmol
)1
), UDP-[
14
C]Xyl (238
mCiámmol
)1
), CMP-[9-
3
H]Sia (15 Ci ámmol
)1
), and GDP-
[2-
3
H]Man (15 Ciámmol
)1
), were purchased from American
Radiolabeled Chemicals Inc. (St Louis, MO, USA).

fragment of p MKIT-neo-hUGT1-cHA. An in¯uenza virus
hemagglutinin (HA) epitope tag encoding the sequence
Table 1. Oligonucleotides used in mutagenesis i n this study. Bold letters indicate mismatched bases.
Mutagenic PCR primer set 1
Primers for the ®rst PCR:
Upstream primer:
NI254 : 5¢-GTCTTTGTTTCGTTTTCTGTTCTG-3¢
Downstream primer: one of the following mutagenic primers
V45L: 5¢-GGCATTCTGGAGCACCAGCA-3¢
N47A: 5¢-GGCAGCCTGGACCACCAGCA-3¢
I51T: 5¢-TGCTGAGGGTGAGGGAGGC-3¢
Q89E: 5¢-ACCCCTCTTCTCTGCGAAGAGC-3¢
Q129A: 5¢-GGCAACATACGCGAGGTTATT-3¢
L174M: 5¢-GCTGCAGTGGGCCTCCCTGCTGATGCTCTTCACTGG-3¢
Primers for the second PCR:
Upstream primer:
mega primers obtained from the ®rst PCR
Downstream primer:
NI255: 5¢-TGCCAGGCCTGCCCCAGGGTTCTG-3¢
Mutagenic PCR primer set 2
Primers for the ®rst PCR:
Upstream primer: one of the following mutagenic primers,
I181L: 5¢-GGCGTCGCCCTTGTCCAGGCAC-3¢
Q185K: 5¢-AGGCAAAGCAAGCCGGTGGG-3¢
F265Y: 5¢-GGTTTCTTTTATGGGTACACACCTGC-3¢
V286T: 5¢-CGGCGGGCTACTGACGGCTGTGGTTGTCA-3¢
Downstream primer:
11±5: 5¢-ACCCTTTAAGCCCCGCCCCATTTA-3¢
Primers for the second PCR:
Upstream primer:

(BamHI, EcoRI, SmaIandNot I) was generated in this way.
The modi®ed plasmid, pYEX-BESN, was utilized to
construct pYEX-hUGT-cHA and pYEX-DmNST-cHA,
in which HA-tagged hUGT1 and HA-tagged DmUGT
cDNAs, respectively, were inserted into the EcoRI±NotI
site. S. cerevisiae YPH500 cells (MATa ura3-52 lys2-801
ade2-101 trp1-D63 his3-D200 leu2-D1) were transformed
with these expression plasmids by the lithium acetate
method [18].
Subcellular fractionation and nucleotide sugar
transport assay
The subcellular fractionation and transport assay were
performed as described previously [19,20]. The membrane
fractions obtained by centrifugation at 10 000 g and
100 000 g were combined and used in the transport
assay. Microsomes (50 lgofprotein)wereincubatedin
0.1 mL of TSM buffer [10 m
M
Tris/HCl (pH 7.0), 0.8
M
glucitol, 1 m
M
MgCl
2
,50m
M
dimercaptopropanol] con-
taining 1 l
M
radioactive substrate (6400 Ciámol

(EY L aboratories, San Mateo, CA, USA), and further
incubated with monoclonal anti-HA Ig to detect the
transporter protein expressed in the cells. T he cells were
then incubated with the secondary antibody, Alexa594-
conjugated anti-(rat IgG) Ig. Fluorescence labeling w as
visualized under a Carl Zeiss laser scanning confocal
microscope LSM510.
Western blot analysis
Western blot analysis was carried out as described previ-
ously [14]. Brie¯y, transfected cells were lysed in an
extraction buffer [10 m
M
Tris/Hepes (pH 7.4), 10 m
M
KCl, 1 m
M
EDTA, 0 .2% Nonidet P-40, 2 mgámL
)1
of
aprotinin, 2 mgámL
)1
of pepstatin A, 2 mgámL
)1
of
leupeptin, 0.5 m
M
phenylmethanesulfonyl¯uoride], and the
samples were fractionated b y electrophoresis on a 12%
SDS/polyacrylamide gel. The separated polypeptides were
electotransferred to a poly(vinylydene di¯uoride) mem-

program in
1998 (SPTREMBL accession number O76865), but coin-
cided with the prediction made in 2000 (SPTREMBL
accession number: O9W4W6). The cDNA clone contained
an ORF encoding 357 amino acids with a calculated
molecular mass of 38 635.3 Da. The putative product was
very hydrophobic and the hydropathy pro®le resembled
130 H. Segawa et al. (Eur. J. Biochem. 269) Ó FEBS 2002
Fig. 1. Sequence analysis of the DmNST/DmUGT. (A) Nucleotide and deduced amino-acid sequences of DmNST/DmUGT. The GenBank/
EMBL/DDBJ accession number of the nucleotide sequence is AB055493. T he putative exon junctions were deduced from comparison with
genomic DNA data (accession numbers O76865 and O9W4W6) and i ndicated by the arrowheads. The s ymbol ÔVÕ indicates a po ten tial N -
glycosylation site. A putative polyadenylation signal is enclosed by a box. (B) Hydrophobicity plot of DmNST/DmUGT. The plot was calculated
with a window size of 10 amino acids using the hydrophobicity values of Kyte & Doolittle [29].
Ó FEBS 2002 UDP-Gal transporter transports UDP-GlcNAc (Eur. J. Biochem. 269) 131
those of other NSTs (Fig. 1B). As shown in Fig. 2,
comparison of the amino-acid sequence of DmNST with
those of human NSTs indicated that DmNST is equally
similar to those three transporters. DmNST had 74 residues
in common with UGT, 69 residues with UglcNAcT, and 40
residues with CST, in addition to 90 residues conserved
Fig. 2. Alignment of DmNST/DmUGT and human NST sequences. hUGT1, human UDP-Gal transporter 1 (GenBank accession number D84454)
[10]; h CST, human CMP-Sia transporter (D87969) [11]; h UGlcNAcT, human UDP-GlcNAc transporter (AB021981) [15]. Thick bars, putative
transmembrane helices as proposed by Eckhardt et al. [27]. Asterisks indicate the Ôsubstrate speci®cÕ residues described previously [15]. The solid
asterisks indicate ÔUGT-speci®cÕ residues conserved in DmUGT. Underlining indicates a potential glycosylation site of DmUGT.
Fig. 3. Expression of DmNST/DmUGT in Lec2 and Lec8 cells. (A) Lec2 a nd Lec8 cells were transfected with appropriate plasmids as speci®ed
below, and CST and UGT activities of cDNA product s were assessed using FITC-labeled lectins as described in Materials and methods. a, pMKIT-
neo; b, pMKIT-neo-hCS T-cHA; c, pMKIT -neo-DmNST-c HA; d, pMKIT-neo ; e, pMKIT- neo-hUGT-cH A; f, pMK IT-neo-DmNS T-cHA. Bar,
10 lm (B) Western blot analysis of DmNST/DmUGT protein expressed in Lec2 (lanes 1 and 2) and Lec8 (lanes 3 and 4) cells. Cell extracts were
prepared from cells transfected with appropriate plasmids as speci®ed below, and were s ubjected to Western blot analysis. Lanes 1 and 3, pMKIT-
neo; lanes 2 and 4, pMKIT-neo-DmNST-cHA.

In Western blot analysis, the DmNST was detected as a
broad band with an apparent molecular mass ranging from
30 to 36 kDa (Fig. 3B, lanes 2 and 4). The broadening of the
bands might be due to N-linked glycosylation at Asn311
(Fig. 1A), as this broadening was not observed with human
nucleotide sugar transporters (data not shown) that lack the
glycosylation motif at the corresponding sites (Fig. 2). The
DmNST expressed in Lec8 migrated slightly slower than
that expressed in Lec2 (Fig. 3B, lanes 2 and 4). This may be
explained by the fact that expression of DmNST comple-
mented the defect in UDP -Gal transp ort of Lec8 cells, and
that this would lead to the formation of f ully processed
oligosaccharide chains attached to t he protein.
hUGT and DmUGT both transport UDP-Gal
and UDP-GalNAc
To examine the substrate speci®city of DmUGT more
extensively, we utilized a yeast expression system.
HA-tagged DmUGT cDNA was inserted into the copper-
inducible yeast expression vector pYEX-BESN and trans-
fected into S. cerevisiae YPH500, and a t ransformant was
obtained. We prepared the microsomes from the transfor-
mant, and analyzed them for the presence of the DmNST
protein by Western blot analysis using anti-HA Ig (Fig. 4).
DmUGT-cHA migrated as a broad band with an apparent
molecular mass ranging from 28 to 36 kDa (lane 4).
Microsomes were prepared from transformants carrying
vectors with and without the DmUGT insert, and investi-
gated for their activity to transport nucleotide sugars
(Fig. 5). We also examined the substrate speci®city of
human UDP-Gal transporter extensively using microsomal

Mutagenesis of hUGT1 cDNA and assessment
of expression and NST activities of mutant proteins
DmUGT indicated signi®cant similarity to both hCST and
hUGlcNAcT c omparable with that to hUGT. Its substrate
speci®city w as, how ever, e xactly the same with that of
Fig. 4. Expression of DmNST/DmUGT in mammalian and yeast
microsomal membranes. Microsomes were prepared from Lec8 or yeast
cells expressing DmNST/DmUGT, and samples containing 30 lgof
protein were subjected to Western blot analysis. Lane 1, pMKIT-neo-
transfected Lec8; lane2, pMKIT-neo-DmNST-cHA-transfected Lec8;
lane 3, pYEX-BESN-transformed YPH500; lane 4, pYEX-BESN-
DmNST-cHA-transformed YPH500.
Ó FEBS 2002 UDP-Gal transporter transports UDP-GlcNAc (Eur. J. Biochem. 269) 133
hUGT as far as examined. In one of our previous
communications we noted that 10 amino-acid residues
seemed to be Ôsubstrate speci®cÕ in that they we re conserved
among transporters with identical substrates, but were
different b etween those speci®c for different substrates [15].
As shown in Fig. 2, a mong these 10 residues, only three, one
and three residues were shared by DmNST and hUGT,
hUGlcNAcT, or hCST, respectively. The three remaining
residues were not conserved between these transporters. To
see if these few conserved residues may be critical in
discriminating between speci®c substrates, we have chosen
the hUGT molecule as the representative of UDP-Gal
transporters, and altered the ÔUGT-speci®cÕ residues of
hUGT to their corresponding ÔCST-speci®cÕ residues by site-
directed mutagenesis. We paid particular attention to N47,
L174 and V285, which were shared by DmUGT and
hUGT.

radioactivities trapped by vector control
microsomes were subtracted as background
values from corresponding experimental
values. The double reciprocal plot of the data
obtained in (A) and (C) and the results of
linear regression analyses are shown in (B) and
(D), respectively.
134 H. Segawa et al. (Eur. J. Biochem. 269) Ó FEBS 2002
were introduced into Lec2 and Lec8 cells. The hUGT1
(N5aaCST)-cHA mutant carried V45L, N47A, I51T, Q89E,
and Q129A substitutions, and hUGT1(C5aaCST)-cHA
carried L174M, I181L, Q 185K, F265Y, and V285T substi-
tutions. In the hUGT1(10aaCST)-cHA mutant all of the
ÔUGT-speci®c Õ residues were replaced by the corresponding
ÔCST-sp eci®cÕ ones [15].
The expression and the transport activities of each mutant
were assessed by immuno¯uorescence and FITC-labeled
lectin binding as in Fig. 3. Figure 7 shows t hat N47A,
L174M and V285T mutants retained UDP-Gal transport
activity but were unable to transport CMP-Sia. Other single
substitution mutants as well as three multiple substitution
mutants gave essentially the same results as V45L (Figs 7d,l)
and hUGT1(10aaCST) (Figs 7h,p), and were active in UDP-
Gal transport, but not in CMP-Sia transport (data not
shown). Most of the mutant proteins, except h UGT1
(N47A)-cHA, were localized in the Golgi apparatus as was
the wild-type p rotein. The hUGT1(N47A)-cHA mutant
protein was not con®ned to the Golgi region, but was dis-
tributed more diffusely in the perinuclear region. This mutant
showed UDP-Gal transport activity, but the frequency of

or UGT-de®cient Lec8 (panels i±p) cells. Lec2
cells were stained with FITC-labeled PNA,
and Lec8 cells with FITC-labeled GS-II to
assess CST and UGT activities of the trans-
porters and mutants. The expression of pro-
teins was detected by immunostaining with
anti-HA Ig, which was visualized by using
Alexa596-conjugated anti-(rat IgG) Ig.
(a and i), pMKIT-neo; b and j, pMKIT-neo-
hCST-cHA; (c and k), pMKIT-neo-hUGT1-
cHA; (d and l), pMKIT-neo-hUGT1(V45L)-
cHA; e a nd m, pMKIT-neo-hUGT1(N47A)-
cHA; (f and n), pMKIT-neo-
hUGT1(L174M)-cHA; (g and o), pMKIT-
neo-hUGT1(V285T)-cHA; (h and p),
pMKIT-neo-hUGT1(10aaCST)-cHA. Bar,
10 lm.
Ó FEBS 2002 UDP-Gal transporter transports UDP-GlcNAc (Eur. J. Biochem. 269) 135
intracellular distribution (Fig. 9A, panel e) and expression
level (Fig. 9B, lane e) of the double-mutant protein.
DISCUSSION
In this study we determined the primary structure of a
putative nucleotide sugar transporter of D. melanogaster,
and identi®ed it as the D. melanogaster homologue
(DmUGT) of human UDP-Gal transporter (hUGT). The
cDNA complemented the genetic defect of UGT-de®cient
Lec8 ce lls, and its product was detected in the Golgi region
of the transfected cells. Heterologous expression of the
cDNA in S. cerevisiae cells allowed us to demonstrate
directly that the cDNA product was able to transport UDP-

N-acetylgalactosamine residues are involved in the wingless
signaling [25]. The DmUGT protein was localized in the
GolgiregionwhenthecDNAwasexpressedinLec2andLec8
cells, and transported both UDP-Gal and UDP-GalNAc.
The subcellular localization and its substrate speci®city are
consistent with its possible involvement in this process. RNA
interference experiments [26] may help to answer this
intriguing question about the physiological role of DmUGT.
We found a single possible N-glycosylation site in the
DmUGT during analysis on the primary structure of the
fruit ¯y NST (Fig. 1A). Based on the 10-segment trans-
membrane model proposed by Eckhardt et al.[27],the
N-glycosylation site resides at the boundary between th e
ninth and tenth putative transmembrane regions (Fig. 2).
Eckhardt et al. were not able to decide whether these
hydrophobic regions (Fig. 2; Hxs9 and 10) traverse the
membrane, are enbedded in the membrane without being
exposed to the lumen side, or are just tightly membrane
associated, as anti-HA epitope antibodies failed to detect an
HA epitope introduced to this boundary region [27]. The
expressed DmUGT proteins were glycosylated in both
CHO (Fig. 3A ) and S. cerevisiae (Fig. 4) cells indicating
that the N-glycosylation site found is faced to Golgi lumen
and accessible to glycosyl transferases. These results suggest
that both the ninth and tenth hydrophobic regions form
discrete membrane-spanning domains.
Three amino-acid residues of hUGT, namely N47, L174,
and V285, are conserved among human, murine, ®ssion
yeast, and Drosophila UGTs, but are distinct from the
corresponding residues conserved among CMP-Sia trans-

clues to investigate th e mechanisms of integration, sorting
and substrate-recognition of this polytopic membrane
protein.
ACKNOWLEDGEMENTS
This work was supported in part by Grants-in-Aid for Scienti®c
Research no. 11480172, Grants-in-Aid for Scien ti®c Research on
Priority Area no . 12033222 f rom the Ministry of Education, Science,
Sports and Culture of Japan and a Grant from Mizutani Foundation
for Glycoscienc e.
REFERENCES
1. Kawakita, M., Ishida, N ., Miura, N., Sun-Wada, G H. &
Yoshioka, S. (1998) Nucleotide sugar transporters: elucidation of
their molecular identity and its implication f or future studies.
J. Biochem. (Tokyo). 12 3, 777±785.
2. Seppo, A. & Tiemeyer, M. (2000) Function and structure of
Drosophila glycans. Glycobiol. 10, 751±760.
3. Herman, T. & Horvitz, H.R. (1999) Three proteins involved in
Caenorhabditis elegans vulval invagination are similar to compo-
nents of a glycosylation pathway. Pr oc. Natl A cad. Sci. USA 96,
974±979.
4. Abeijon,C.,Robbins,P.W.&Hirschberg,C.B.(1996)Molecular
cloning of the Golgi apparat us uridine diphosphate-N-acetyl-
glucosamine transporter from Kluyveromyces lactis. Proc. Natl
Acad.Sci.USA93, 5963±5968.
5. Dean, N., Zhang, Y.B. & Poster, J.B. (1997) The VRG4 gene is
required for GDP-mannose transport into the lumen of the Golgi
in the yeast, Saccharomyces cerevisiae. J. Biol. Chem. 272, 31908±
31914.
6. Segawa, H., Ishida, N., Takegawa, K. & Kawakita, M. (1999)
Schizosaccharomyces pombe UDP-galactose transporter: identi®-

the hamster CMP-sialic acid tran sporter. Eur. J. Biochem. 24 8,
187±192.
14. Ishida, N., I to, M., Yoshiok a, S., Sun-Wada,G H. & Kawakita, M.
(1998) Functional expression of human Golgi CMP-sialic acid
transporter in the Golgi complex of a transporter-de®cient Chinese
hamster ovary cell mutant. J. Biochem. (Tokyo). 124, 171±178.
15. Ishida, N., Yoshioka, S., Chiba, Y., Takeuchi, M. & Kawakita,
M. (1999) Molecular cloning and functional expression of the
human Golgi UDP-N-acetylglucosamine transporter. J. Biochem.
(Tokyo) . 126, 68±77.
16. Muraoka, M., Kawakita, M. & Ishida, N. (2001) Molecular
characterization of hum an UDP-glucuro nic acid/UDP- N-acety-
lgalactosamine transporter, a novel nucleotide sugar transporter
with dual substrate speci®city. FEBS Lett. 495, 89±93.
17. Sarker, G. & Sommer, S.S. (1990) The Ômegaprimer Õ method of
site-directed mutagenesis. Biotechniques 4, 404±407.
18. Ito, H., Fukuda, Y., Murata, K. & Kimura, A. (1983) Transfor-
mation o f intact yeast cells treated w ith alkali cations. J. Bacteriol.
153, 163±168.
19. Sun-Wada, G H., Yoshioka, S., Ishida, N. & Kawakita, M.
(1998) Functional expression of the human UDP-galactose
transporters in the yeast Saccharomyces cerevisiae. J. Biochem.
(Tokyo) . 123, 912±917.
20. Yoshioka, S., Sun-Wada, G H., Ishida, N. & Kawakita, M.
(1997) Expression of the human UDP-galactose transporter in the
Golgi membranes of murine Had-1 cells that lack the endogenous
transporter. J. Biochem. (Tokyo). 122 , 691±695.
21. Aoki, K., Ishida, N. & Kawakita, M. (2001) Substrate recognition
by UDP-galactose and CMP-sialic acid transporters: dierent sets
of transmembrane helices are utilized for the speci®c recognition of