Available via license: CC BY 4.0
Content may be subject to copyright.
Targeting Proteins to the Lumen of Endoplasmic Reticulum Using
N-terminal Domains of 11
b
-Hydroxysteroid Dehydrogenase and the
50-kDa Esterase*
(Received for publication, December 31, 1998, and in revised form, February 17, 1999)
Hassan Mziaut‡, George Korza‡, Arthur R. Hand§, Craig Gerard¶, and Juris Ozols‡i
From the ‡Department of Biochemistry and §Electron Microscopy, University of Connecticut Health Center,
Farmington, Connecticut 06030-3305 and the ¶Department of Pediatrics, Children’s Hospital, Harvard Medical School,
Boston, Massachusetts 02115
Previous studies identified two intrinsic endoplasmic
reticulum (ER) proteins, 11
b
-hydroxysteroid dehydro-
genase, isozyme 1 (11
b
-HSD) and the 50-kDa esterase
(E3), sharing some amino acid sequence motifs in their
N-terminal transmembrane (TM) domains. Both are type
II membrane proteins with the C terminus projecting
into the lumen of the ER. This finding implied that the
N-terminal TM domains of 11
b
-HSD and E3 may consti-
tute a lumenal targeting signal (LTS). To investigate this
hypothesis we created chimeric fusions using the puta-
tive targeting sequences and the reporter gene, Ae-
quorea victoria green fluorescent protein. Transfected
COS cells expressing LTS-green fluorescent protein chi-
meras were examined by fluorescent microscopy and
electron microscopic immunogold labeling. The orienta-
tion of expressed chimeras was established by immuno-
cytofluorescent staining of selectively permeabilized
COS cells. In addition, protease protection assays of
membranes in the presence and absence of detergents
was used to confirm lumenal or the cytosolic orientation
of the constructed chimeras. To investigate the general
applicability of the proposed LTS, we fused the N termi-
nus of E3 to the N terminus of the NADH-cytochrome b5
reductase lacking the myristoyl group and N-terminal
30-residue membrane anchor. The orientation of the cy-
tochrome b5 reductase was reversed, from cytosolic to
lumenal projection of the active domain. These observa-
tions establish that an amino acid sequence consisting
of short basic or neutral residues at the N terminus,
followed by a specific array of hydrophobic residues
terminating with acidic residues, is sufficient for lume-
nal targeting of single-pass proteins that are structur-
ally and functionally unrelated.
A number of proteins are resident and catalytically active in
the lumen of the endoplasmic reticulum (ER).
1
The best char-
acterized motif targeting this compartment is for soluble pro-
teins bearing C-terminal sequence KDEL (1, 2). The molecular
signals responsible for the insertion and retention of mem-
brane proteins in the ER with a lumenal orientation of their
functional domains are poorly understood. Most ER proteins
are targeted for membrane insertion by a hydrophobic signal
peptide at the N terminus (3). During translation, the signal
peptide is recognized by a cytosolic RNA/protein complex
termed the signal recognition particle (SRP) (4). Upon binding
a signal peptide, the SRP arrests the protein synthesis until it
has facilitated transfer of the nascent polypeptide-ribosome
complex to the translocation channel (5). The signals control-
ling folding and eventual topology of ER lumenal proteins after
the nascent chain is transferred to the translocation channel
are unknown. One early critical event in the multistep folding
process is whether or not the signal peptide is removed. If the
signal peptide is removed, then the newly formed N terminus is
generally located in the lumen of the ER. In cases where the
signal peptide is not removed, the resulting N-terminal hydro-
phobic segment may function as a membrane anchor, directing
the transmembrane insertion of the nascent polypeptide.
Therefore, membrane proteins with uncleaved signal peptide
may have either cytosolic or lumenal orientation. Mutagenesis
experiments on a number of membrane proteins have revealed
that the charge distribution, the length of the membrane seg-
ment, as well as the charge of the segments following the
membrane segment are some of the factors that determine the
specific topology of such proteins (6). Detailed analysis of these
factors have been complicated by use of multispanning model
proteins with a lack of agreement on the topology. Moreover,
additional ambiguities are encountered in trying to overex-
press polytopic eukaryotic membrane proteins in prokaryotic
hosts. Previously, we identified the membrane binding anchor
of cytochrome b5 (7), cytochrome b5 reductase (8), cytochromes
P-450 (9–11), epoxide hydrolase (12), D9 stearyl coenzyme A
desaturase (13), and three forms of flavin-containing monooxy-
genases (14). All of these ER proteins appeared to be oriented
toward the cytosolic side of the ER membrane by single or
multispanning membrane segments. In further studies to de-
fine the proteins found or oriented in the lumenal compartment
of the ER, we identified two isoforms of 60-kDa carboxylester-
ases (15) and the lumenal NADP glucose-6-phosphate dehydro-
genase (16). The esterases and the glucose-6-phosphate dehy-
drogenase were devoid of any transmembrane segments. The
lumenal esterases carried a C-terminal segment HIEL- and
HTEL-related KDEL, the ER lumen retention motif (1). The
primary structure of the glucose-6-phosphate dehydrogenase
failed to display sequence segments or a motif that would be
responsible for its lumenal orientation or retrieval. We then
extended our studies to oligosaccharyltransferase complex (OT)
* This work was supported by Grant R01 GM-26351 from the Na-
tional Institutes of Health (to J. O.). 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.
iTo whom correspondence should be addressed: Dept. of Biochemis-
try, School of Medicine of the University of Connecticut Health Center,
Farmington, CT 06030-3305. Tel.: 860-679-2211; E-mail: ozols@sun.
uchc.edu.
1
The abbreviations used are: ER, endoplasmic reticulum; 11
b
-
HSD, 11
b
-hydroxysteroid dehydrogenase; E3, liver microsomal 50-
kDa esterase/N-deacetylase; GFP, green fluorescent protein; LTS,
lumenal targeting signal; PAGE; polyacrylamide gel electrophoresis;
PBS, phosphate-buffered saline; SLO, streptolysin O; OT, oligo-
saccharyltransferase; AEBSF, 4-(2-aminoethyl)-benzenesulfonyl flu-
oride; PCR, polymerase chain reaction; BSA, bovine serum albumin.
THE JOURNAL OF BIOLOGICAL CHEMISTRY Vol. 274, No. 20, Issue of May 14, pp. 14122–14129, 1999
© 1999 by The American Society for Biochemistry and Molecular Biology, Inc. Printed in U.S.A.
This paper is available on line at http://www.jbc.org14122
by guest on December 29, 2015http://www.jbc.org/Downloaded from
(18). The OT from mammalian tissues was found to be a tri-
molecular complex consisting of ribophorins I and II and a
50-kDa protein. The predicted membrane orientation of OT
indicated that both ribophorins and the 50-kDa proteins have
large N-terminal lumenal domains and a short C-terminal
cytoplasmic domain (18). Studies on the orientation of the
11
b
-HSD (17) and the 50-kDa esterase/N-deacetylase (E3) (19)
revealed that both proteins shared a short N-terminal cytoplas-
mic domain and a large C-terminal lumenal domain.
In the present study, we investigated whether the N-termi-
nal membrane binding segment of 11
b
-HSD and E3 may act as
a targeting sequence for the lumenal orientation of proteins in
the ER. Here we show that a transmembrane segment consist-
ing of basic residues at the N terminus followed by an array of
specific 17 hydrophobic residues terminating with acidic resi-
dues constitute a lumenal targeting signal for a set of single-
membrane-spanning proteins that are otherwise structurally
and functionally unrelated.
EXPERIMENTAL PROCEDURES
Materials—Chemical products were purchased from Sigma unless
otherwise noted. COS-7 cells were from American Type Culture Collec-
tion (ATCC). Cell reagents, restriction, and modification enzymes were
from Life Technologies, Inc. The GFPS65T mutant protein in
pEGFP-N1 vector and an anti-GFP mouse monoclonal antibody were
from CLONTECH. Anti-stearyl CoA-desaturase polyclonal antibody
was raised in a rabbit by injecting keyhole limpet hemocyanin-conju-
gated 20-residue synthetic peptide corresponding to residues 338–358
of stearyl CoA-desaturase antigen (20). Avian anti-b5 and anti-reduc-
tase were raised against the polar moieties of rat cytochrome b5 and
cytochrome b5 reductase (19). Anti-calreticulin rabbit polyclonal anti-
body was from Affinity Bioreagents. Alkaline phosphatase-conjugated
anti-mouse and anti-rabbit IgG were from Sigma. Biotin-coupled goat
anti-mouse IgG was from Molecular Probes. Iodocarbocyanine (Cy3)-
streptavidin and Cy3-conjugated goat anti-rabbit IgG were from Jack-
son Immunoresearch Laboratories. Proteinase K and 4-(2-aminoethyl)-
benzenesulfonyl fluoride (AEBSF) were from Roche Molecular
Biochemicals. Gold-labeled secondary antibody was from Amersham
Pharmacia Biotech.
Construction of Plasmids—All expression vectors were driven by the
cytomegalovirus promoter-enhancer contained in the pEGFP-N1 vec-
tor. The N tag-GFP chimeras were constructed by inserting the coding
region of GFP at the C terminus of the first 23 and 34 amino acids
corresponding to lumen targeting signal (LTS) of 11-
b
HSD and esterase
3, respectively and variants thereof, as follows (see Fig. 1). First, cDNA
encoding residues 1–34 of human esterase 3 preceded by Kozak se-
quence and EcoRI site at 59end and extended by BclI site at 39was
synthesized using two long overlapping oligonucleotides (59-CTTCGA-
ATTCCCACCATGGGAAGAAAATCGCTGTACCTTCTGATTGTGGG-
GATCCTCAATTT-39as sense long oligo and 59-TGGATTGATCATTC-
TCCATGGCTCCTCAACGTTATCTGGGAGAGGCGTATAAATATAAT-
G-39as antisense long oligo) and two short oligonucleotides (59-CTTC-
GAATTCCCACCATG-39as sense short oligo and 59-TGGATTGATCA-
TTCTCCA-39as antisense short oligo) and the chain polymerase reac-
tion (PCR). Second, the coding sequence of GFP was amplified from
pEGFP-N1 by PCR using a 59-oligonucleotide, which contains a BclI
restriction site (59-ATGGTGATCAAGGGCGAGGAGCTG-39) and 39-
oligonucleotide containing NotI site (59-CCTCTACAAATGTGGTATGG-
C-39). This was done in order to remove the Kozak sequence preceding
the first ATG codon of GFP in pEGFP-N1 vector so that the initiation
codon introduced in the N-tag sequences will be used by N-tag-GFP
chimeras.
The resulting PCR products were purified and digested with EcoRI-
BclI and BclI-NotI for N-tags and GFP, respectively and inserted into
EcoRI-NotI-digested pEGFP-N1 vector in one-step ligation. All the N-
tag-GFP chimeras and their mutated or deleted variants were con-
structed using the experimental procedure described above.
The expression vector pEN1.rab E3 N-b5 red.(31–300) encodes a
fusion protein consisting of rabbit esterase 3 LTS, followed by three
novel amino acids (PPV) encoded by the restriction site for AgeI, by
amino acids 31–300 of bovine liver NADH-cytochrome b5 reductase,
and FLAG epitope TAG (LARIKRTGDGSHKSS). This fusion protein
was expressed in pEN1 vector obtained by removing the coding se-
quence of GFP from pEGFP-N1. The rab E3 N-b5 red.(31–300) was
constructed as follows. First, the sequence corresponding to amino acids
31–300 of reductase was amplified by PCR using as a template the
940-bp DNA fragment coding for entire bovine liver microsomal redu-
ctase inserted in pGEM3zf(1) (21) with a pair of primers, 59-GATCC-
ACCGGTCCCGGCCATCACGCTCGAGAAT-39matching with the
amino acids 31–37 of beef reductase preceded by AgeI site, and 59-AG-
AGTCGCGGCCGCTTTAGCTACTCTTGTGGCTCCCATCTCCAGTTC-
TCTTAATCCTGGCTAAGAAGGCGAAGCAGCGTTCTTT-39matching
with amino acids 294–300. This oligonucleotide included the desatu-
rase epitope, stop codon, and NotI site. The PCR fragment was purified,
digested with AgeI-NotI and inserted along with the EcoRI-AgeI frag-
ment corresponding to LTS of rabbit esterase3 into EcoRI-NotI-digested
pEN1 vector. The integrity of the constructs were confirmed by DNA
sequencing.
Cell Culture and Transient Transfection—COS-7 cells were cultured
in Dulbecco’s modified Eagle’s medium (DMEM) supplemented with
10% heat-inactivated fetal bovine serum, 100 units/ml penicillin, and
100 mg/ml streptomycin. The cells were seeded at approximately 30%
confluence onto 35-mm dishes at 37 °C under 5% CO
2
. After 24 h, when
the cells became 60–70% confluent, they were transfected using the
LipofectAMINE reagent according to the manufacturer’s instructions.
The cell staining was observed 48–72 h after transfection under Zeiss
LSM 410 confocal microscope.
Immunoblot Analysis—Particulate components and the soluble frac-
tion of the cells were separated as described previously (22). Insoluble
material was removed by centrifugation, samples resolved by SDS-
PAGE, and Western blot analysis was performed using anti-mouse
monoclonal antibody against GFP. Alkaline phosphatase-conjugated
antibodies were used as secondary antibodies for protein visualization.
Isolation and Protease Digestion of Microsomes from COS-7 Cells—
Microsomal membranes were prepared as described by Ref. 23. The
pellet fraction was enriched in cytochrome b5 by adding 10
m
l of 0.5 mM
pure detergent free cytochrome b5 (24). The mixture was resuspended
in a solution containing 250 mMsucrose, 10 mMpotassium-Hepes, pH
7.2, 50 mMpotassium chloride, 2 mMmagnesium chloride, and 2 mM
calcium chloride to a final volume of 0.2 ml. Protease digestions were
performed at 4 °C for 30 min with increasing concentrations of protein-
ase K in presence or absence of 0.5% Triton X-100. AEBSF (4 mM) was
used to terminate the protease digestions. After 15 min of incubation on
ice, the membranes were solubilized with SDS gel loading buffer, and
the proteins subjected to SDS-PAGE and immunoblotted as described
above.
Selective Permeabilization and Immunofluorescence Analysis—Per-
meabilization of cells using digitonin, streptolysin O, and saponin was
performed 48 h after transfection, as described previously (25). Immu-
nocytofluorescent staining was conducted by incubation of the cells with
a 1:100 dilution of anti-GFP mouse monoclonal antibody in the perme-
abilization buffers containing 0.1% BSA. The cells were then rinsed
three times and then incubated with a 1:100 dilution of biotin-coupled
goat anti-mouse immunoglobulin. After the wash, cells were incubated
with streptavidin coupled-iodocarbocyanine (Cy3) diluted 1:100 in the
permabilization buffers containing 3% BSA. The cells were rinsed sev-
eral times before mounting in Mowiol 4.88 and then viewed under Zeiss
LSM 410 confocal microscope.
Electron Microscopic Immunogold Labeling—Transfected COS-7
cells with rab E3 N-GFP were fixed with 3% paraformaldehyde plus
0.15% glutaraldehyde in 0.1 Msodium cacodylate buffer, pH 7.4. The
cells were allowed to fix for about1hatroom temperature, the fixative
was removed, and the cells stored at 4 °C in 1% paraformaldehyde in 0.1
Mcacodylate, pH 7.4. The cells were removed from the flask by scraping
and then pelleted in 0.2% BSA in PBS. The cells were resuspended in
0.1% BSA/PBS, transferred to a tube containing 3% low-gelling agarose
and pelleted again. The agarose was chilled on ice, and the tip of the
tube was cut off. The embedded cell pellet was rinsed in 0.1 Mcacodylate
buffer, pH 7.4, dehydrated in methanol, infiltrated, and embedded at
220 °C in Lowicryl K4M resin. After polymerization with ultraviolet
light (365 nm), thin sections of the embedded cells were cut with a
diamond knife and collected on Formvar-coated, 200-mesh nickel grids.
Immunogold labeling was done essentially as described elsewhere
(26). Nonspecific binding was blocked with 1% BSA plus 1% instant
milk in PBS. The sections were labeled overnight at 4 °C with anti-GFP
mouse monoclonal antibody diluted 1:50 in 1% BSA and 5% normal goat
serum in PBS. After rinsing with PBS, the bound immunoglobulins
were visualized by incubating with 10-nm diameter gold-labeled sec-
ondary antibody diluted 1:15 in PBS for1hatroom temperature. After
thorough rinsing with PBS and distilled water, the sections were
stained with uranyl acetate and lead citrate and observed in Philips
CM10 TEM.
Targeting Proteins to the ER Lumen 14123
by guest on December 29, 2015http://www.jbc.org/Downloaded from
RESULTS
The N Termini of 11
b
-HSD and Esterase 3 (E3) Determines
Their Subcellular Localization—The proposed LTS of rat, rab-
bit, and human 11
b
-HSD aligned with sequences of E3 of
rabbit and human are shown in Fig. 1A. The two protein
families share a short positively charged N termini (region I),
followed by an array of some 17 hydrophobic residues contain-
ing an aromatic cluster Ala-Tyr-Tyr-X-Tyr or Gly-Tyr-Tyr-Tyr
cluster (region II), terminating with di-glutamyl residues (re-
gion III). Sequence similarity beyond the N termini could not be
found in these two protein families (19). To ascertain whether
N terminus of 11
b
-HSD is a determinant of its subcellular
localization in mammalian cells, constructs were generated
encoding N terminus of 11
b
-HSD fused to GFP (11
b
-HSD N-
GFP) (Fig. 1B). The A. victoria GFP is widely recognized as a
powerful tool in cell biology serving as a reporter for monitoring
localization and dynamics of intracellular proteins and or-
ganelles (27). When rab 11
b
-HSD N-GFP (Construct a) was
transiently expressed in COS-7 cells and examined by fluores-
cent microscopy, a staining pattern characteristic of ER local-
ization was seen (Fig. 2A). Expression of native GFP showed
green fluorescent signals in both the cytoplasm and nucleus
(Fig. 2B) consistent with the previous studies (28). To explore
whether the N-terminal domain of E3 also imparts GFP a
membrane localization, GFP constructs hum and rab E3N-GFP
(Constructs c and e, respectively) were generated and ex-
pressed in COS-7 cells. Cells expressing Constructs c and e
showed fluorescence pattern of ER localization, which was sim-
ilar to that seen with Construct a (Fig. 2, A,C, and D). Analysis
of the membrane fractions of cells expressing the Construct e
by SDS-PAGE, followed by immunoblotting with GFP antibody
showed the localization of GFP construct in the insoluble frac-
tion. Upon SDS-PAGE, a decreased mobility of rab E3 N-GFP
as compared with untagged GFP was observed, implying that
the signal sequence like N terminus of the construct was not
removed by the signal peptidase (Fig. 2E). As anticipated,
untagged GFP expressed in cells was found only in the soluble
fraction.
The N Termini Targets ER Localization—To confirm that the
fluorescence distribution along the reticular network observed
in Fig. 2 is characteristic of ER localization, we compared the
fluorescence pattern of cells expressing rab E3N-GFP (Con-
struct e) with a known ER marker, calreticulin. The fluores-
cence pattern of Construct e (Fig. 2F) is almost identical to the
calreticulin staining (Fig. 2G). Indeed, the combined fluores-
cence revealed a yellow staining indicative of the colocalization
of the rab E3N-GFP-associated green fluorescence and the
calreticulin-associated red fluorescence as evidenced by
Fig. 2H.
Electron Microscopic Immunogold Labeling—To confirm
these results at the ultrastructural level, ultrathin cryosections
were analyzed in double-labeling experiments with the immu-
nogold technique, using anti-GFP antibodies to identify the ER.
Electron microscopic examination of thin sections of rab E3
N-GFP (Construct e) transfected cells labeled with anti-GFP
antibodies revealed the presence of gold particles over the
nuclear envelope and the ER (Fig. 3, Aand B). Clusters of
heavily labeled ER membranes were found in the perinuclear
region. The gold particles were clearly associated with periph-
ery of the ER cisternae. Although the labeling was most fre-
quently seen in association with ribosome-studded membranes,
cisternae of apparently smooth ER were also labeled (Fig. 3C).
The labeling obtained with the polyclonal and monoclonal anti-
GFP preparations was approximately equivalent. Nonspecific
background labeling was relatively low; very few particles were
found over the nuclear chromatin, mitochondria, or other
structures.
Transmembrane Topology of Rab E3N-GFP, Construct e—To
determine the orientation of Construct e in the membrane, we
performed immunofluorescent staining of selectively permeabi-
lized COS-7 cells by saponin, digitonin, and streptolysin O
(SLO). As previously reported, digitonin (20
m
g/ml) and SLO
(200 units/ml) selectively permeabilized plasma membrane
leaving internal membranes including ER intact, whereas sap-
onin permeabilized all the membranes (29). For control exper-
iments, in cells expressing untagged GFP treated with digito-
nin (Fig. 4A), or saponin (Fig. 4C), the antibodies stained the
nucleus as well as the cytoplasm as expected. Cells treated with
SLO (Fig. 4B) showed a similar staining except for the nuclear
membrane, which was not permeabilized by SLO. Cells trans-
fected with cDNA for rab E3N-GFP (Construct e) and perme-
abilized with either digitonin (Fig. 4D) or SLO (Fig. 4E) failed
to stain with the antibodies to GFP even though these cells
expressed the GFP construct as indicated by the green fluores-
FIG.1.N-terminal sequences of native proteins and GFP constructs. A, N-terminal sequences of rat, rabbit, and human 11
b
-HSD are
aligned with those of rabbit and human E3. The positions of N-terminal amino acid residues are numbered below the sequence. Conserved amino
acids are enclosed in boxes.Dashes were inserted to maintain alignment of conserved residues. The LTS is denoted by regions I, II and III as
indicated above. The Protein Identification Resource accession numbers for the rat, rabbit, and human 11
b
-HSDs are A34430, A55573, and
A41173, respectively. Rabbit and human E3 accession numbers are A58922 and A53856. B, amino acid sequences of LTS of native proteins and
the derivatives thereof with altered LTS fused to GFPs, which were prepared for the experiments described here. Amino acid residues or segments
that were altered are indicated by letters in boldface or by dashed line.
Targeting Proteins to the ER Lumen14124
by guest on December 29, 2015http://www.jbc.org/Downloaded from
cent emission. However, after permeabilization with saponin
(Fig. 4F), all the cells expressing GFP chimeras became reac-
tive to the GFP antibody. The fact that the GFP fluorescence
pattern uniformly overlapped with the GFP antibody staining
pattern only when all the cell membranes were completely
permeabilized with saponin indicates that the GFP tagged with
E3 LTS is oriented toward the lumenal side of ER (Fig. 4F).
Evidence for Lumenal Localization by Proteinase K As-
say—To confirm the lumenal topography of rab E3N-GFP (Con-
struct e), membranes from cells expressing Construct e were
subjected to proteinase K digestion in the presence and absence
of detergent. As seen in Fig. 4G, in the intact membranes
Construct e was not susceptible to proteinase K digestion.
Proteolysis of Construct e could be demonstrated when deter-
gent was added to intact membranes prior to incubation with
the protease. Taken together, these results are consistent with
the lumenal orientation of the expressed chimera. The prote-
olysis of cytochrome b5 was used as a positive control in this
proteolysis protection assay. Cytochrome b5 is an integral
membrane protein oriented toward the cytosolic side of ER, and
is readily released from the membrane upon proteolysis (19,
24). Cytochrome b5 added to a membrane preparation is spon-
taneously incorporated into the membrane. As shown in Fig.
4G, proteolysis of membrane fractions in the presence or ab-
sence of detergent results in the conversion of cytochrome b5to
a species with faster mobility upon SDS-PAGE. The latter
species lacks the C-terminal 40 residue membrane anchor, but
retains its catalytic activity (7). Cleavage of cytochrome b5 was
observed in intact microsomes of cells containing Construct e at
protease concentration where no cleavage of the expressed
protein occurred (Fig. 4G).
Mutagenesis of the Lumenal Targeting Sequence—Sequences
of 11
b
-HSD from different mammalian species contain two Lys
residues in region I. The region I of human E3 contains an
Arg-Lys sequence. In the rabbit E3 sequence, the Arg residue is
replaced by Val, implying that in region I a single lysyl residue
rather than two basic residues do not alter the orientation of E3
in the membrane. Replacement of the single Lys residue by Ile
as in rab E3N(K4/I)-GFP (Construct f) led to ER localization of
the resulting chimera as evidenced by immunofluorescence
(Fig. 5A). The fluorescence image of cells expressing Construct
f is similar to that seen for rab E3N-GFP (Construct e). The
membrane localization of Construct f was confirmed by subcel-
lular fractionation and immunoblot analysis. As seen in Fig.
FIG.2. Expression of rab11
b
-HSD N-GFP, hum E3 N-GFP, rab E3 N-GFP, and native GFP in COS-7 cells. Evidence for the
colocalization with the ER marker calreticulin. Representative fluorescence images of COS 7 cells transfected with GFP and the indicated fusion
proteins are shown. A, Rab 11
b
-HSD N-GFP (Construct a). B, GFP control. C, Hum E3 N-GFP (Construct c). D, Rab E3 N-GFP (Construct e). The
cells were cultured for 48 h after transfection and observed under Zeiss LSM 410 confocal microscope (magnification, 340). E, immunoblot analysis
of rab E3 N-GFP (Construct e) and native GFP in cellular subfractions. Cells transfected with cDNA for GFP and rab E3 N-GFP were harvested
48 h after transfection, disrupted, and subjected to centrifugation. The whole cell extract (W), soluble fraction (S), and pellet fraction (P) (50
m
gof
protein) were subjected to SDS-PAGE and immunoblotted with 5
m
g/ml of an anti-GFP mouse monoclonal antibody. F,G, and H, colocalization of
rab E3 N-GFP (green) with the ER marker calreticulin (red) in COS-7 transfected cells permeabilized with 0.2% saponin. F, fluorescence image of
COS-7 cells expressing rab E3 N-GFP (green). G, the same cell as in Fwas costained by incubation with a rabbit polyclonal antibody against the
ER-resident protein calreticulin followed by incubation with Cy3-conjugated goat anti-rabbit IgG. The cells were then observed under confocal
microscope to visualize the staining pattern of the fluorescent dye Cy3 (red). H, merged images of Fand Greveal considerable colocalization of
brightest signal for rab E3 N-GFP and calreticulin in yellow (magnification 340).
Targeting Proteins to the ER Lumen 14125
by guest on December 29, 2015http://www.jbc.org/Downloaded from
5B, the expressed protein was present in pellet and absent in
the soluble fraction. To establish the orientation of the ex-
pressed protein, cells were selectively permeabilized with sap-
onin, digitonin, SLO, and analyzed by immunofluorescence
microscopy (Figs 5, C–E). Again, the staining pattern of such
permeabilized cells was identical to that obtained with Con-
struct e (Fig. 4, D–F). This result indicates that positively
charged residues are not an essential topogenic signal for the
lumenal orientation.
When the single Gly residue in region II was replaced by Glu,
and the resulting chimera (Construct b) expressed in cells, such
mutation led to the expression of the protein with an apparent
cytosolic and nuclear localization (Fig. 6A). Deletion of six
residues in region II as in the Construct d also led to a cytosolic
and nuclear localization of the protein (Fig. 6B).
Reorientation of NADH Cytochrome b5 Reductase Conferred
by the LTS—To test the utility of LTS as an in vivo trafficking
signal for proteins other than GFP, we made a fusion protein
between the polar segment of microsomal NADH cytochrome
b5 reductase and the N terminus of rab E3. The catalytically
active polar segment of the reductase consists of some 270
residues, and lacks the hydrophobic 28-residue N-myristoy-
lated membrane-anchoring domain at the N terminus present
in the intact protein (8). The native reductase has a cytosolic
orientation. To assess whether the LTS can redirect the reduc-
tase derivative to the lumen of ER, we fused the LTS of rabbit
E3 to the N terminus of the polar segment of the reductase. In
addition, we added to the C terminus of the reductase the
highly antigenic C-terminal sequence of stearyl CoA desatu-
rase as an epitope tag. We termed the chimera rab E3-N-b5 red
(31–300) or Construct g. The orientation of Construct g in the
ER membrane was examined by proteinase K digestion. As
shown in Fig. 7, digestion of intact microsomes (lane 1–3 and 5)
revealed a protease-resistant fragment approximately of 35–40
kDa in mass, indicating that the construct was intralumenal.
When microsomal membranes were disrupted with 0.5% Triton
X-100, the protein became sensitive to proteinase K digestion
and was completely degraded (lane 4). The proteinase K pro-
tection assay results confirm our previous results of the GFP
chimera orientation and show that soluble as well as membra-
nous proteins with a cytoplasmic orientation wearing 11
b
or E3
LTS are oriented toward the lumenal side of ER.
DISCUSSION
11
b
-HSD and E3 are two unrelated ER lumenal proteins
with type II orientation (17, 19). Analysis of their covalent
structures implied that the LTS for 11
b
-HSD and E3 is encoded
in the N-terminal segment. The N-terminal amino acid se-
quence of 34 residues of this group of proteins from several
mammalian species is shown in Fig. 1A. For discussion pur-
poses, the LTS sequence is divided into regions I, II, and III. In
addition to the positive charge at the N-terminal residue, fea-
tures of region I that are shared by several species include a
single lysyl residue. A stretch of aliphatic and aromatic hydro-
phobic residues (region II) follows region I, and continues into
a segment of negatively charged residues (region III). A strik-
ing feature of region I is the variability of the first three
residues and the conservation of a single lysyl residue. Region
II is a hydrophobic segment containing a cluster of tyrosyl
residues. In region III, there is a conserved Asn-X-Glu-Glu
segment. The remaining sequence of the C-terminal ectodo-
main shared no obvious homology between the dehydrogenases
and the esterases. Despite that the bulk of the polypeptide
chain is translocated across the membrane, the N-terminal
leader peptide like sequence in this group of proteins escapes
signal peptidase cleavage and serves as the membrane anchor.
Fig. 1Bdisplays the mutations and deletions of the chimeras
constructed by fusion of N-terminal domain of 11
b
-HSD or E3
to the N terminus of GFP. These constructs were expressed in
COS 7 cells, and analysis of their cellular targeting was accom-
plished by combination of fluorescence microscopy and electron
microscopic immunogold labeling. The topology of expressed
proteins was established by immunocytofluorescent staining of
selectivity permeabilized COS cells. Protease protection assay
of membranes in the presence and absence of detergents was
also used to distinguish the membrane sidedness of the GFP
chimeras.
The subcellular localization and orientation of the con-
structed GFP chimeras is summarized in Fig. 8 (Aand B).The
most important finding of this study is that fusion of the N-
terminal sequences from both 11
b
-HSD and E3 proteins to
GFP resulted in the lumenal localization of the chimera in the
ER membrane. This finding implied that the information (LTS)
for the lumenal orientation of 11
b
-HSD and E3 is encoded in
the N-terminal segment of some 20 to 25 residues.
As displayed in the amino acid sequences of the native pro-
teins from different species, the presence of two neutral or two
basic residues in region I of LTS are not essential for the
lumenal targeting. This is based on the assumption that the
proteins from different animal species share an identical ori-
entation. The proposed essential di-arginine sequences (30) at
the cytoplasmic N terminus of type II membrane proteins are
not indispensable for the ER localization for the group of pro-
teins described here. Deletion of the single lysyl residue in
region I (Construct f) led to ER targeting of the GFP chimera
with a lumenal ER orientation (Fig. 5). Therefore, the absence
of basic residues in region I does not alter the lumenal target-
FIG.3. Electron microscopic immunogold labeling of rab E3
N-GFP transfected cells with anti-GFP antibodies. Aand B, illus-
trate clusters of ER in the perinuclear region. In A, the asterisks (*)
indicate dilated regions of ER with membrane-associated gold particles.
In B, the arrows indicate regions where labeling of the ER is particu-
larly prominent. In both panels, the arrowheads indicate gold particles
associated with the nuclear envelope. Nucleus is denoted by n.InC,
several smooth-surfaced membranes are labeled (arrows). Scale bar 5
0.25
m
m.
Targeting Proteins to the ER Lumen14126
by guest on December 29, 2015http://www.jbc.org/Downloaded from
ing of the protein. Clearly, the positively charged residues are
not required for the ER localization, but the extreme negative
charge of region III would conform to the N .C charge role for
the lumenal orientation.
In order to further characterize the essential residues of the
LTS, several chimeras of LTS fused to GFP were constructed
and their subcellular localization analyzed. The Construct b
bearing a single negative charge in the center of region II led to
the expression of protein with fluorescence pattern similar to
the control GFP (Fig. 6A). Immunoblot of the subcellular frac-
tions of cells expressing Construct b indicated its electro-
phoretic mobility upon SDS-PAGE identical to native GFP,
implying that a truncation of the N terminus of the chimera
had occurred (Fig. 6C), but the protein failed to enter the
FIG.4. Immunocytofluorescent
staining of selectively permeabilized
COS-7 cells and the protease sensitiv-
ity of the expressed chimeras. COS-7
cells transfected with cDNA coding for
GFP (A–C), and rab E3 N-GFP (D–F) were
permeabilized either with digitonin (20
m
g/ml), streptolysin O (200 units/ml), or
saponin (0.2%) before incubation with an-
ti-GFP mouse monoclonal antibody.
Bound antibody was visualized by incuba-
tion with biotin-coupled goat anti-mouse
and then with Cy3-streptavidin as de-
scribed under “Experimental Procedures”
(magnification, 340). G, protease sensi-
tivity of rab E3 N-GFP and cytochrome b5
in membranes. The microsomal mem-
branes of COS-7 cells transfected with
cDNA coding for rab E3 N-GFP fusion
protein and enriched in cytochrome b5
were incubated with the indicated concen-
tration of proteinase K for 30 min on ice in
the absence (lanes 1–3) or presence (lane 4
and 5) of 0.5% Triton X-100. The mobility
of rab E3 N-GFP and cytochrome b5 was
analyzed by SDS-PAGE and immunoblot
with 5
m
g/ml of an anti-GFP mouse mono-
clonal antibody or with 1:1,000 dilution of
an anti-cytochrome b5 chicken polyclonal
antibody.
FIG.5.Expression of rab E3 N(K4/I)-
GFP in COS-7 cells to analyze the im-
portance of positively charged resi-
due of region I in the targeting to the
ER. A, subcellular localization of rab E3
N(K4/I)-GFP (Construct f) determined by
fluorescence microscopy (magnification,
340). The (K4/I) symbol indicates the po-
sition of positively charged residue within
E3 region I in which the single lysine was
mutated to an isoleucine residue. B, im-
munoblot analysis of COS-7 cells express-
ing Construct f and fractionated as de-
scribed in Fig. 2. The fluorescence image
and the immunoblot show the same intra-
cellular localization as the constructs E3
N-GFP in Fig. 2 (Cand D). Selective per-
meabilization with digitonin (C), SLO (D),
and saponin (E) of cells expressing Con-
struct f was done as described in Fig. 4.
Targeting Proteins to the ER Lumen 14127
by guest on December 29, 2015http://www.jbc.org/Downloaded from
secretory pathway as anticipated. This implied that the signal
peptidase was not responsible for the cleavage of the N-termi-
nal signal. What may have compensated for the translocation
deficiency of Construct b? It is most likely that the introduction
of a glutamyl residue in region II of the LTS resulted in a
conformational change interpreted by the cytosolic quality con-
trol system as a misfolded protein. Proteolytic elimination of
malfolded proteins, including uncomplexed subunits of protein
assemblies, is a common defense mechanism of the cell. Mutant
forms of cystic fibrosis transmembrane conductance regulator
expressed in mammalian cells, or carboxypeptidase Y mutants
that fail to translocate across the membrane and accumulate
on the surface of ER are readily degraded by the cytoplasmic
proteosome pathway (31, 32). One reason we observed the GFP
tag in cells expressed with Construct b could be the resistance
of the GFP molecule toward proteolysis. Expression of the N
terminus of Construct b fused to a polypeptide that is more
sensitive to proteolysis than the GFP molecule most likely
would result in the degradation of the entire construct.
A striking feature of region II of the parent proteins is the
presence of a cluster of tyrosyl residues (AYYXY or GYYY).
Repeats of tyrosyl residues (AYPYYA) are found in the trans-
membrane domain of the 48-kDa subunit of microsomal OT. OT
has a short C terminus oriented toward the cytosol, and a
single transmembrane domain with the bulk of the protein
molecule positioned in the lumen of ER (18). The T-cell receptor
delta chain displays a GYYYYV sequence in its membranous
segment (33). That a tyrosine-containing motif mediates ER
retention of CD3-
e
chain has also been reported (34). To deter-
mine the topogenic importance of the tyrosyl residue cluster
GFP chimera, Construct d lacking the tyrosyl residues in re-
gion II was generated. Construct d upon its expression in COS
cells displayed fluorescence in the nucleus and cytosol (Fig.
6B). This fluorescence pattern of intracellular structures was
similar to that observed by the untagged GFP. These findings
suggest that a cluster of tyrosyl residues is essential for the
correct folding and the lumenal targeting. Further analysis will
be needed to establish whether replacement of these residues
with phenylalanines or glycines provides an appropriate con-
text for the function of the conserved tyrosyl residues. Deletion
of five to six residues in region III of the 11
b
-HSD-GFP con-
structs does not affect the lumenal orientation of the constructs
(Fig. 2A).
Presently, the 11
b
-HSD and the E3 protein are the only type
II, static lumenal ER proteins with a single N-terminal trans-
membrane segment that have been characterized. The asialo-
FIG.6.Effect of mutations in the hydrophobic region of 11
b
-HSD and E3 LTS on the localization of GFP chimera proteins. A,
fluorescence microscopy of COS-7 cells transfected with rab 11
b
-HSD N(G10/E)-GFP (Construct b) in which the glycine residue in position 10 was
mutated to glutamic acid residue. B, fluorescence image of COS-7 cells transfected with hum E3 N(D16–21)-GFP (Construct d) in which the
hydrophobic segment was shortened by deletion of the six residues AYYIYT. A strong labeling pattern of nucleus, ER, and cytoplasm is seen for
the mutant chimera proteins as compared with the ER labeling pattern seen for the wild chimera proteins in Fig. 2 (Cand D) (magnification, 340).
C, immunoblot analysis of cells expressing GFP, Construct b, and Construct a.
FIG.7. Evidence for the lumenal orientation of rab E3 N-b5
red.(31–300) in microsomal membranes of transfected cells. Mi-
crosomes containing the chimeric protein (Construct g) were incubated
with the indicated concentrations of proteinase K in the presence (lane
4) or absence (lanes 1–3 and 5) of 0.5% Triton X-100. Samples of the
proteolysis reaction mixture were subjected to SDS-PAGE analysis,
followed by immunoblot with 1:2,000 dilution of anti-FLAG polyclonal
antibody.
FIG.8. Targeting proteins to the lumen of ER with LTS. A,
summary of subcellular localization of all the constructs expressed. The
constructs are abbreviated a-g as shown in Fig. 1B. The abbreviations
for the subcellular localizations are N, ER, C, S, L, and Cyt, nucleus,
endoplasmic reticulum, cytosolic, secreted, lumen, and cytoplasmic
side, respectively, proposed membrane topology of the LTS-GFP con-
structs. Nand Crepresent the N and C termini of the native and
chimera proteins. The chimeras are displayed in the upper part, and the
native proteins in the lower part of the model. My at the N terminus of
native cytochrome b5 reductase denotes a myristoyl residue.
Targeting Proteins to the ER Lumen14128
by guest on December 29, 2015http://www.jbc.org/Downloaded from
glycoprotein receptor (35) and the influenza virus neuramidi-
nase (36) are type II plasma membrane proteins. The
glycoprotein, paramyxovirus hemagglutinin-neuroaminidase
(HN) is a type II membrane protein localized to the Golgi
cisternae (37). The predicted amino acid sequence of the simian
virus HN protein includes a 17-residue cytoplasmic tail, a
19-residue membranous segment, and a large 523-residue C-
terminal ectodomain (38). Newly synthesized HN oligomerizes
into tetramers before transport from ER to Golgi, and alter-
ations of the C-terminal ectodomain can prevent ER to Golgi
transport (39). The large family of glycosidases and glycosyl-
transferases, although type II transmembrane proteins are all
located in the Golgi cisternae. Their common features include
an N-terminal cytoplasmic tail, a transmembrane domain, and
a large catalytic domain oriented toward the lumenal side of
the membrane (40). Two mutually complementary models have
been proposed to explain the mechanism of Golgi retention of
the glycosyltransferases mediated by their transmembrane do-
mains. One model postulates the retention through oligomer-
ization, which prevents proteins from entering transport vesi-
cles (41, 42). The other model suggests that their retention
depends on the length of a membrane-spanning domain, which
can be accommodated by the specific lipid composition of Golgi
complex membrane (43, 44). It has been pointed out that nei-
ther the oligomerization nor the membrane lipid composition
alone can explain the sorting of Golgi proteins.
Having established that the N-terminal domain of 11
b
-HSD
or E3 can mediate lumenal localization of the following down-
stream GFP molecule, we sought to test the general applica-
bility of this finding to targeting of ER proteins that display a
cytosolic orientation in their native state. The flavoprotein
NADH cytochrome b5 reductase has an amphiphilic structure
in which the hydrophilic, catalytic domain of some 270 residues
is linked to a membrane-anchoring, hydrophobic domain that
serves to orient the catalytic site of the reductase at the mem-
brane-aqueous interface to permit a rapid electron transfer to
cytochrome b5 (8). The hydrophobic domain at the N terminus
of the reductase consists of some 28 residues as well as N-
myristoylation of the N-terminal glycine residue (45) (Fig. 8B).
Fusion of the LTS of rabbit E3 to the N terminus of the polar
segment of the reductase (Construct g) resulted in the lumenal
targeting of the fusion protein (Figs. 7 and 8B).
Whether the topogenic signal described here is position-in-
dependent as regards to lumenal targeting of proteins, in ad-
dition to identifying the structural features that prevent these
constructs transiting to the Golgi, are the topics for further
studies. In conclusion, a static lumenal targeting signal of
proteins reported here should add to the repertoire of tech-
niques for studies on the structure, organization, and processes
of the lumen of ER.
Acknowledgment—We thank Dr. C. S. Ramaro (University of Alabama
at Birmingham) for constructing and expressing Constructs a and b.
REFERENCES
1. Munro, S., and Pelham, H. R. B. (1987) Cell 48, 899–907
2. Scheel, A. A., and Pelham, H. R. (1996) Biochemistry 35, 10203–10209
3. Blobel, G. (1980) Proc. Natl. Acad. Sci. U. S. A. 77, 1496–1500
4. Schatz, G., and Dobberstein, B. (1996) Science 271, 1519–1526
5. Hegde, R. S., Voigt, S., Rapoport, T. A., and Lingappa, V. R. (1998) Cell 92,
621–631
6. Gafvelin, G., Sakaguchi, M., Andersson, H., and von Heijne, G. (1997) J. Biol.
Chem. 272, 6119–6127
7. Ozols, J., and Gerard, C. (1977) Proc. Natl. Acad. Sci. U. S. A. 74, 3725–3729
8. Ozols, J., Korza, G., Heinemann, F. S., Hediger, M. A., and Strittmatter, P.
(1985) J. Biol. Chem. 260, 11953–11961
9. Ozols, J., Heinemann, F. S., and Johnson, E. F. (1985) J. Biol. Chem. 260,
5427–5434
10. Ozols, J. (1986) J. Biol. Chem. 261, 3965–3979
11. Heinemann, F. S., and Ozols, J. (1982) J. Biol. Chem. 257, 14988–14999
12. Heinemann, F. S., and Ozols, J. (1984) J. Biol. Chem. 259, 797–804
13. Thiede, M. A., Ozols, J., and Strittmatter, P. (1986) J. Biol. Chem. 261,
13230–13235
14. Ozols, J. (1994) Biochemistry 33, 3751–3757
15. Ozols, J. (1989) J. Biol. Chem. 264, 12533–12545
16. Ozols, J. (1993) Proc. Natl. Acad. Sci. U. S. A. 90, 5302–5306
17. Ozols, J. (1995) J. Biol. Chem. 270, 2305–2312
18. Kumar, V., Heinemann, F. S., and Ozols, J. (1994) J. Biol. Chem. 269,
13451–13457
19. Ozols, J. (1998) Biochemistry 37, 10336–10344
20. Heinemann, F. S., and Ozols, J. (1998) Mol. Biol. Cell 9, 3445–3453
21. Strittmatter, P., Kittler, J. M., Coghill, J. E., and Ozols, J. (1992) J. Biol.
Chem. 267, 2519–2523
22. Mori, M., Morita, T., Ikeda, F., Amaya, Y., Tatibana, M., and Cohen, P. P.
(1981) Proc. Natl. Acad. Sci. U. S. A. 78, 6056–6060
23. Haynes, R. L., Zheng, T., and Nicchitta, C. V. (1997) J. Biol. Chem. 272,
17126–17133
24. Ozols, J. (1989) Biochim. Biophys. Acta 997, 121–130
25. Otto, J. C., and Smith, W. L. (1994) J. Biol. Chem. 269, 19868–19875
26. Hand, A. R. (1995) in Introduction to Biophysical Methods for Protein and
Nucleic Acid Research (Glasel, J. A., and Deutscher, M., eds) pp. 205–260,
Academic Press, New York
27. Chalfie, M., Tu, Y., Euskirchen, G., Ward, W. W., and Prasher, D. C. (1994)
Science 263, 802–805
28. Yano, M., Kanazawa, M., Terada, K., Namchai, C., Yamaizumi, M., Hanson,
B., Hoogenraad, N., and Mori, M. (1997) J. Biol. Chem. 272, 8459–8465
29. Schulz, I. (1990) Methods Enzymol. 192, 280–300
30. Schutze, M. P., Peterson, P. A., and Jackson, M. R. (1994) EMBO J. 13,
1696–1705
31. Kopito, R. R. (1997) Cell 88, 427–430
32. Hiller, M. M., Finger, A., Schweiger, M., and Wolf, D. H. (1996) Science 273,
1725–1728
33. Hein, W. R., and Dudler, L. (1993) EMBO J. 12, 715–724
34. Mallabiabarrena, A., Jimenez, M. A., Rico, M., and Alarcon, B. (1995) EMBO
J. 14, 2257–2268
35. Spiess, M., and Lodish, H. F. (1986) Cell 44, 177–185
36. Bos, T. J., Davis, A. R., and Nayak, D. P. (1984) Proc. Natl. Acad. Sci. U. S. A.
81, 2327–2331
37. Parks, G. D. (1996) J. Biol. Chem. 271, 7187–7195
38. Hiebert, S. W., Paterson, R. G., and Lamb, R. A. (1985) J. Virol. 54, 1–6
39. Ng, D. T., Hiebert, S. W., and Lamb, R. A. (1990) Mol. Cell. Biol. 10, 1989–2001
40. Paulson, J. C., and Colley, K. J. (1989) J. Biol. Chem. 264, 17615–17628
41. Colley, K. J. (1997) Glycobiology 7, 1–13
42. Dahdal, R. Y., and Colley, K. J. (1993) J. Biol. Chem. 268, 26310–26319
43. Pelham, H. R. B., and Munro, S. (1993) Cell 75, 603–605
44. Munro, S. (1995) Biochem. Soc. Trans. 23, 527–530
45. Ozols, J., Carr, S. A., and Strittmatter, P. (1984) J. Biol. Chem. 259,
13349–13354
Targeting Proteins to the ER Lumen 14129
by guest on December 29, 2015http://www.jbc.org/Downloaded from
Hand, Craig Gerard and Juris Ozols
Hassan Mziaut, George Korza, Arthur R.
Dehydrogenase and the 50-kDa Esterase
-HydroxysteroidβDomains of 11
Endoplasmic Reticulum Using N-terminal
Targeting Proteins to the Lumen of
CELL BIOLOGY AND METABOLISM:
doi: 10.1074/jbc.274.20.14122
1999, 274:14122-14129.J. Biol. Chem.
http://www.jbc.org/content/274/20/14122Access the most updated version of this article at
.JBC Affinity SitesFind articles, minireviews, Reflections and Classics on similar topics on the
Alerts:
When a correction for this article is posted• When this article is cited•
to choose from all of JBC's e-mail alertsClick here
http://www.jbc.org/content/274/20/14122.full.html#ref-list-1
This article cites 44 references, 31 of which can be accessed free at
by guest on December 29, 2015http://www.jbc.org/Downloaded from
