The stop transfer sequence of the human UDP-
glucuronosyltransferase 1A determines localization to the
endoplasmic reticulum by both static retention and
retrieval mechanisms
Lydia Barre
´
, Jacques Magdalou, Patrick Netter, Sylvie Fournel-Gigleux and Mohamed Ouzzine
UMR 7561 CNRS-Universite
´
Henri Poincare
´
Nancy I, France
Biosynthesis of integral membrane proteins involves sev-
eral events such as targeting to the endoplasmic reti-
culum (ER), translocation of certain domains into the
ER lumen and integration of transmembrane domains
(TMD) into the lipid bilayer. These proteins are then
maintained in the ER by two modes, static retention or
dynamic retention by continuous retrieval of the escaped
proteins from the post-ER compartments. Retrieval sig-
nal sequences have been identified in both soluble [1]
and transmembrane [2] ER resident proteins. For
soluble ER resident proteins, a C-terminal tetrapeptide
KDEL in mammals and a closely related sequence
HDEL in yeast were shown to serve as specific ER
retention signals. The mechanism is based on the KDEL
receptor, which binds the escaped proteins in the Golgi
complex and returns them back to the ER [3]. For trans-
membrane type I ER resident proteins, a retrieval signal
KXKXX has been identified in the cytosolic tail (CT)
allowing for retrieval from the Golgi to the ER in a

teins contain a C-terminal stop transfer sequence with a transmembrane
domain (TMD), which anchors the protein into the membrane, followed
by a short cytosolic tail (CT). Here, we investigated the mechanism of ER
residency of UGT1A mediated by the stop transfer sequence by analysing
the subcellular localization and sensitivity to endoglycosidases of chimeric
proteins formed by fusion of UGT1A stop transfer sequence (TMD ⁄ CT)
with the ectodomain of the plasma membrane CD4 reporter protein. We
showed that the stop transfer sequence, when attached to C-terminus of
the CD4 ectodomain was able to prevent it from being transported to the
cell surface. The protein was retained in the ER indicating that this
sequence functions as an ER localization signal. Furthermore, we demon-
strated that ER localization conferred by the stop transfer sequence was
mediated in part by the KSKTH retrieval signal located on the CT. Inter-
estingly, our data indicated that UGT1A TMD alone was sufficient to
retain the protein in ER without recycling from Golgi compartment, and
brought evidence that organelle localization conferred by UGT1A TMD
was determined by the length of its hydrophobic core. We conclude that
both retrieval mechanism and static retention mediated by the stop transfer
sequence contribute to ER residency of UGT1A proteins.
Abbreviations
CT, cytosolic tail; Endo H, endoglycosidase H; ER, endoplasmic reticulum; FITC, fluoresceine isothiocyanate; PGNase F, peptide-N-
glycosidase F; TMD, transmembrane domain; UGT, UDP-glucuronosyltransferase.
FEBS Journal 272 (2005) 1063–1071 ª 2005 FEBS 1063
retention ⁄ retrieval motif CVLF has been described for a
splice variant SV1 of the voltage- and Ca
2+
-activated
K
+
channel alpha-subunit preventing plasma mem-

gene locus consisting of 16 exons. The isoforms are
generated by alternative splicing of exon 1 to the
four common exons 2–5 resulting in isoforms with
an identical C-terminal half of the protein [15] and a
unique N-terminal end. UGT1A proteins are predic-
ted to be type I membrane proteins of the ER with
a glycosylated lumenal domain. It has also been
reported that UGT2B7 and UGT1A6 were expressed
in nuclear membrane [16]. The proteins contain a
stop transfer sequence at the C-terminus consisting
of a TMD of 17 residues followed by a short CT of
25 residues containing a KXKXX ER retrieval sig-
nal. In this study, we investigated the role of the
UGT1A stop transfer sequence (TMD ⁄ CT) in ER
residency. We showed, using a CD4 plasma mem-
brane protein as a reporter, that the UGT1A stop
transfer sequence acts as an ER retention signal. We
demonstrated that ER residency is determined by
both the retrieval mechanism mediated by the
KSKTH motif at the C-terminus of the CT and by
a static retention mediated by the hydrophobic
domain of the TMD. Furthermore, we showed that
the major determinant accounting for ER residency
conferred by the TMD is related to the length of its
hydrophobic core.
Results
The stop transfer sequence of UGT1A functions
as an ER targeting and retention signal in
mammalian cells
In order to analyse the ER retention capacity of the

hydrophobic Ala ⁄ Leu residues, respectively.
Retention of human UGT1A in endoplasmic reticulum L. Barre
´
et al.
1064 FEBS Journal 272 (2005) 1063–1071 ª 2005 FEBS
which was converted, after peptide-N-glycosidase F
(PGNase F) treatment, to a band  4 kDa smaller,
corresponding to the expected size of the unglycosyla-
ted fusion protein (Fig. 2A). Interestingly, a similar
band was generated upon treatment with Endo H indi-
cating that the protein was also sensitive to Endo H
digestion (Fig. 2A), thereby demonstrating that CD4–
TMD ⁄ CT was retained in the ER of HeLa cells. To
ascertain the intracellular localization of the CD4–
TMD ⁄ CT chimeric protein, immunofluorescence analy-
ses were carried out using monoclonal anti-CD4 sera.
Cells expressing recombinant full-length CD4 were
used as a control for cell surface expression (Fig. 2B).
Cells expressing CD4–TMD ⁄ CT were not labelled in
the absence of Triton X-100 permeabilization (Fig. 2B)
suggesting that CD4–TMD ⁄ CT protein was not
exposed to the cell surface but was retained in an
intracellular compartment. Interestingly, analysis of
Triton X-100-permeabilized cells showed a reticular
staining pattern characteristic of the ER localization of
CD4–TMD ⁄ CT protein. This location was confirmed
by the colocalization of the chimeric protein with the
ER marker protein, calnexin (Fig. 2B). Together, these
data showed that the UGT1A TMD ⁄ CT domain was
able to retain the CD4 plasma membrane protein in

CD4–TMD ⁄ CT
ser
protein leaked from the ER into the
latter compartment in the secretory pathway. A similar
behaviour was observed in the case of CD4–
TMD ⁄ CT
myc
(data not shown). Immunofluorescence
A
B
Fig. 2. The TMD and CT of the UGT1A stop
transfer sequence determine subcellular
localization. (A) Sensitivity of CD4–TMD ⁄ CT
chimeric proteins to Endo H and PGNase F
treatment. CD4–TMD ⁄ CT construct has
been stably expressed in HeLa cells and
microsomal membranes of recombinant
cells were prepared as described in Experi-
mental procedures. Microsomal proteins
were subjected, or not, to Endo H and
PGNase F digestion and chimeric proteins
were then analysed by SDS ⁄ PAGE and
detected by Western blot analysis using
anti-CD4 sera. Nontransfected HeLa cells
were used as controls. (B) Cells expressing
native (CD4) and CD4–TMD ⁄ CT chimeric
protein were analysed by immunofluore-
scence microscopy. Cells were fixed with
paraformaldehyde, permeabilized or not (to
monitor cell surface expression) with Tri-

The TMD of UGT1A is sufficient for ER retention
In order to investigate the role of the TMD of
UGT1A in ER retention, a CD4–TMD chimeric pro-
tein (CD4 ectodomain fused to the TMD of UGT1A)
was stably expressed in HeLa cells (Fig. 1). As des-
cribed above, glycosylation was used as a marker to
determine whether this protein without CT was
retained in the ER or moved forward to the distal
organelles in the secretory pathway. Endoglycosidase
analysis showed that CD4–TMD protein was Endo H
sensitive, as digestion with the endoglycosidase pro-
duced a single polypeptide, whose mass was reduced
by 4 kDa (Fig. 4A). Similar results were obtained after
PGNase F treatment (Fig. 4A). These data suggested
that CD4–TMD was retained in the ER. In agreement,
immunofluorescence studies showed that CD4–TMD
presented a typical reticular staining pattern and
colocalized with the ER marker protein, calnexin
(Fig. 4B). Taken together, these data indicated that the
TMD domain of UGT1A was sufficient to retain the
ectodomain of CD4 protein in the ER. Because
the carbohydrate moieties of proteins that are trans-
ported to the Golgi become resistant to Endo H, these
results also indicated that CD4–TMD was retained
in the ER without recycling from post-Golgi com-
partment.
TMD length determines the subcellular
localization
It has been proposed that the length of the TMD of
Golgi and plasma membrane proteins was in part

tion, and analysed by Western blot using
anti-CD4 sera. Nontransfected HeLa cells
were used as controls. (B) Cells expressing
CD4–TMD ⁄ CT
ser
protein were Triton X-100
permeabilized, or not, and analysed by
immunofluorescence microscopy using anti-
CD4–FITC conjugated sera. Cells expressing
CD4–TMD ⁄ CT
ser
were also stained for cal-
nexin and for GM130 as ER and Golgi
marker, respectively, using rhodamine-conju-
gated secondary antibodies. Merge corres-
ponds to colocalization (yellow) of the
chimeric protein with each subcompartment
marker.
Retention of human UGT1A in endoplasmic reticulum L. Barre
´
et al.
1066 FEBS Journal 272 (2005) 1063–1071 ª 2005 FEBS
CD4–TMD
26
was partially sensitive as about half of
the polypeptides acquired resistance to Endo H
(Fig. 5A). However, these polypeptides were sensitive
to PGNase F (Fig. 5A). Taken together, these data
suggested that the CD4–TMD
26

Human UGTs are transmembrane type I glycoproteins
with an N-terminal cleavable signal peptide and a
C-terminal stop transfer sequence. Our laboratory has
been deeply involved in the identification of protein
domains that are determinant for membrane assembly
in the ER. We previously showed that deletion of the
signal peptide alone or in combination with that of the
TMD did not prevent membrane targeting and
insertion of the enzyme. These findings resulted in the
identification of an internal signal sequence localized
between residues 140 and 240 and led us to suggest
that the membrane assembly of UGT1A6 may involve
several topogenic elements [17]. This prompted us to
investigate the topogenic role of the stop transfer
sequence, which comprises a TMD followed by a short
cytosolic tail with the common KXKXX ER
retrieval ⁄ retention signal.
It is widely accepted that there are two mecha-
nisms for the localization of ER resident proteins;
one is the dynamic retrieval mechanism from post-
ER compartments, and the other is the static retent-
ion mechanism that prevents exit from the ER. In
type I membrane proteins such as UGTs, the retrie-
val signal has been defined as two lysine residues at
positions )3 and )5 from the C-terminus exposed
on the cytosolic side of the ER membrane [18]. We
report in this study that disruption of the dilysine
motif KSKTH of UGT1A by mutation of lysine to
serine residues or by extending the length of the
cytoplasmic tail to relocate the dilysine from the crit-

indicates that the retrieval mechanism is not suffi-
cient to ensure ER residency, suggesting that an
additional mechanism may be involved.
TMDs of membrane proteins have often been shown
to contain important information for localization in
the ER [19]. We found that the TMD of UGT1A pro-
teins appended to the C-terminus of the ectodomain of
CD4 plasma membrane glycoprotein was able to retain
the chimeric protein in the ER as indicated by endo-
glycosidase treatments which showed that CD4 ⁄ TMD
was sensitive to Endo H, which removes the high-
mannose oligosaccharides that are found in the ER,
and by immunofluorescence analysis. These data sug-
gest that the TMD contains sufficient information for
ER retention probably acting via a static mechanism.
In the same manner, it has been demonstrated that the
TMD of transmembrane proteins cause ER localiza-
tion of a yeast type VI transmembrane protein UBC6
[9] and a rat type II membrane protein cytochrome b5
[20] by static retention. Further experiments showed
that extension of the TMD of UGT1A appended to
the CD4 ectodomain resulted in a chimeric protein
that was partially resistant to Endo H treatment, but
sensitive to PGNase F, indicating that complex glyco-
sylation occurred. This observation suggested that the
protein effectively moved through the medial-Golgi
compartment. In agreement, immunofluorescence
localization studies confirmed that lengthening the
TMD resulted in Golgi and cell-surface expression of
CD4–TMD chimera. The concept that TMD length

CD4–TMD ⁄ CT
26
were treated, or not, with
Endo H and PGNase F, and analysed by
Western blot using anti-CD4 sera. (B) Cells
expressing CD4–TMD ⁄ CT
ser
were analysed
by immunofluorescence microscopy using
anti-CD4–FITC conjugated sera. Cells were
also stained for calnexin and for GM130 as
ER and Golgi markers, respectively, using
rhodamine-conjugated secondary antibodies.
Merge corresponds to colocalization (yellow)
of the chimeric protein with each sub-
compartment marker.
Retention of human UGT1A in endoplasmic reticulum L. Barre
´
et al.
1068 FEBS Journal 272 (2005) 1063–1071 ª 2005 FEBS
Experimental procedures
Chemicals were from Merck (Darmstadt, Germany) or
Sigma (St. Louis, MO, USA). Vent DNA polymerase,
restriction enzymes, Endo H, PGNase F and Phototope
Ò
-
HRP Western detection system were from New England
Biolabs (Beverly, MA, USA). Escherichia coli JM109 was
from Promega (Madison, WI, USA). ExGen 500 transfec-
tion reagent was from Euromedex (Souffelweyersheim,

TMD
21
and pCD4–TMD
26
expression vectors with the
hydrophobic segment of the TMD extended from 17 to 21
and 26 residues, respectively, amino acids LALA and
LALALALAL were inserted into TMD sequence
1
VIG
FLLAVVLTVAFITF
17
between Val9 and Leu10 residues
by two rounds SOE-PCR [16] using pCD4–TMD as tem-
plate. Mutants were systematically checked by sequencing.
Schematic representation of the constructs is shown in
Fig. 1.
Site-directed mutagenesis
Mutation of lysine residues of the cytoplasmic tail motif
KSKTH to serine residues was performed by PCR using a
sense primer 5¢-GGATCCGTGATTGGTTTCCTCTTG-3¢
containing a BamHI site and UGT1A6 nucleotides 1467–
1482 and an antisense primer 5¢-CTCGAGTCAATGG
GTACTGGAACTGTGGGCTTTCTT-3¢ introducing the
mutations and encoding for the last six residues of human
UGT1A6 (1593–1576) followed by a stop codon and XhoI
site. The PCR product was cloned into BamHI and XhoI
sites of pTM1 expression vector in frame with CD4 ecto-
domain to generate pCD4–TMD ⁄ CT
ser

France) and centrifuged at 12 000 g for 20 min. Membranes
were pelleted from the supernatant for 1 h at 100 000 g at
4 °C. The pellet was resuspended by Dounce homogeniza-
tion in sucrose–Hepes buffer. The protein concentration of
the homogenate was evaluated by the method of Bradford
[23]. Membrane proteins (50 lg) were boiled for 10 min in
denaturing buffer (0.5% (w ⁄ v) sodium dodecyl sulfate, 1%
(v ⁄ v) 2-mercaptoethanol), then digested with Endo H or
PNGase F for 2 h at 37 °Cin50mm sodium citrate buffer
(pH 5.5) or 50 mm sodium phosphate buffer (pH 7.5) con-
taining 1% NP-40, respectively. Endo H cleaves aspara-
gine-linked high mannose structures generating a peptide
with one attached N-acetylglucosamine residue. PNGase F
is an amidase that cleaves between the innermost N-acetyl-
glucosamine and the asparagine residue removing all types
of N-glycan chains from glycopeptides and glycoproteins.
Endoglycosidase-digested and nontreated samples were elec-
trophoresed on 10% (w ⁄ v) SDS ⁄ polyacrylamide gels and
transferred to ImmobilonPÒ membrane. Proteins were then
immunostained using a polyclonal anti-CD4 sera and Pho-
totopeÒ-HRP labelled anti-rabbit secondary sera for chemi-
L. Barre
´
et al. Retention of human UGT1A in endoplasmic reticulum
FEBS Journal 272 (2005) 1063–1071 ª 2005 FEBS 1069
luminescence detection (Cell Signaling Technology, Beverly,
MA, USA).
Immunofluorescence microscopy
Immunofluorescence was performed as described by Louvard
et al. [24]. Briefly, cells were grown on glass coverslips and

Germany).
Acknowledgements
This work was supported by grants from Fonds National
pour la Science, Ligue Re
´
gionale Contre le Cancer,
Re
´
gion Lorraine, Communaute
´
Urbaine du Grand
Nancy and Institut Fe
´
de
´
ratif de Recherche 111 (Bio-
inge
´
nierie). Dr N. Venkatesan is gratefully acknowledged
for critical reading of the manuscript and Dr D. Dumas
for performing confocal laser scanning microscopy.
References
1 Munro S & Pelham HR (1987) A C-terminal signal pre-
vents secretion of luminal ER proteins. Cell 48, 899–907.
2 Nilsson T, Jackson M & Peterson PA (1989) Short cyto-
plasmic sequences serve as retention signals for trans-
membrane proteins in the endoplasmic reticulum. Cell
58, 707–718.
3 Lewis MJ & Pelham HR (1990) A human homologue of
the yeast HDEL receptor. Nature 348, 162–163.

culum localization of Sec12p is achieved by two
mechanisms: Rer1p-dependent retrieval that requires the
transmembrane domain and Rer1p-independent
retention that involves the cytoplasmic domain. J Cell
Biol 134, 279–293.
11 Rayner JC & Pelham HR (1997) Transmembrane
domain-dependent sorting of proteins to the ER and
plasma membrane in yeast. EMBO J 16, 1832–1841.
12 Munro S (1995) A comparison of the transmembrane
domains of Golgi and plasma membrane proteins. Bio-
chem Soc Trans 23, 527–530.
13 Harding D, Fournel-Gigleux S, Jackson MR &
Burchell B (1988) Cloning and substrate specificity of
a human phenol UDP-glucuronosyltransferase
expressed in COS-7 cells. Proc Natl Acad Sci USA 85,
8381–8385.
14 Ebner T & Burchell B (1993) Substrate specificities of
two stably expressed human liver UDP-glucuronosyl-
transferases of the UGT1 gene family. Drug Metab
Dispos 21, 50–55.
15 Ritter JK, Chen F, Sheen YY, Tran HM, Kimura S,
Yeatman MT & Owens IS (1992) A novel complex
locus UGT1 encodes human bilirubin, phenol, and other
UDP-glucuronosyltransferase isozymes with identical
carboxyl termini. J Biol Chem 267, 3257–3261.
16 Radominska-Pandya A, Pokrovskaya ID, Xu J, Little
JM, Jude AR, Kurten RC & Czernik PJ (2002) Nuclear
UDP-glucuronosyltransferases: identification of
UGT2B7 and UGT1A6 in human liver nuclear mem-
branes. Arch Biochem Biophys 399, 37–48.

izing the principle of protein–dye binding. Anal Biochem
72, 248–254.
24 Louvard D, Reggio H & Warren G (1982) Antibodies
to the Golgi complex and the rough endoplasmic
reticulum. J Cell Biol 92, 92–107.
L. Barre
´
et al. Retention of human UGT1A in endoplasmic reticulum
FEBS Journal 272 (2005) 1063–1071 ª 2005 FEBS 1071