Apical Transport and Folding of Prostate Specific Membrane Antigen Occurs Independent of ...

23 Pages·2005·2.4 MB·English

JBC Papers in Press. Published on October 12, 2005 as Manuscript M509460200
The latest version is at http://www.jbc.org/cgi/doi/10.1074/jbc.M509460200
Apical Transport and Folding of Prostate Specific Membrane Antigen Occurs
Independent of Glycans Processing

Deborah Castelletti‡, Giulio Fracasso§, Marwan Alfalah‡, Sara Cingarlini§, Marco
Colombatti§ and Hassan Y. Naim‡*

From the ‡Department of Physiological Chemistry, University of Veterinary Medicine Hannover,
Hannover, Germany and §Department. of Pathology, University of Verona, Italy
Running Title: Maturation and Apical Sorting of PSMA
*Address correspondence to: Hassan Y. Naim, Department of Physiological Chemistry, School of
Veterinary Medicine Hannover, Bünteweg 17, D-30559 Hannover, Germany, Tel: [phone]
8780, Fax: + 49 511 953 8585, Email: [email protected]

Prostate-specific membrane antigen Golgi, therefore, is an essential prerequisite
(PSMA) is an integral cell surface for protein trafficking and sorting of
membrane glycoprotein that is PSMA and suggests that altered or
D
overexpressed in prostate carcinomas aberrant glycosylation often occurring o
w
rendering it an appropriate target for during tumorgenesis has no regulatory nlo
a
antibody-based therapeutic strategies. The function on the cell surface expression of de
d
biosynthesis of PSMA in transfected COS-1 PSMA. fro
m
cells reveals a slow conversion of mannose- h
ttp
rich to complex glycosylated PSMA INTRODUCTION ://w
compatible with slow transport kinetics Prostate cancer is one of the most common w
w
from the ER to the Golgi. Importantly, malignancies diagnosed in men and represents .jb
c
mannose-rich PSMA persists as a trypsin- a significant worldwide health problem. No .org
sensitive protein through out its entire life conventional treatment is effective following by/
g
cycle and only Golgi-located PSMA relapse or in case of metastatic disease. In ue
s
glycoforms acquire trypsin-resistance. This fact, in nearly all of the patients relapsing t on
D
resistance, used here as a tool to examine cancer progresses despite initial treatment e
c
e
correct folding, does not depend on the even with the most updated forms of therapy. mb
e
type of glycosylation, since different PSMA Biomarkers of prostate cancer have become r 2
0
glycoforms generated in the presence of available recently for the diagnosis, , 2
0
1
inhibitors of carbohydrate processing in immunotherapy, and predicting disease 8
the Golgi are also trypsin resistant. The progression of prostate cancer (1). In
conformational transition of PSMA to a particular, the Prostate Specific Membrane
correctly folded molecule is likely to occur Antigen (PSMA)1 represents a suitable cell
in the Golgi and does not implicate ER surface marker that is being considered in the
molecular chaperones, such as BiP.

We show here that PSMA is not only
heavily N-, but also O-glycosylated. The
1 PSMA, prostate specific membrane antigen;
question arising is whether glycans, which
SI, sucrase-isomaltase; DMEM, Dulbecco’s
do not play a role in folding of PSMA, are
modified Eagle’s medium; dMM,
implicated in its transport to the cell
deoxymannojirimycin; endo F, endo-β-N-
surface. Neither the cell surface expression
acetylglucosaminidase F; endo H, endo-β-N-
of PSMA nor its efficient apical sorting in
acetylglucosaminidase H; mAb, monoclonal
polarized MDCK cells are influenced by
antibody; ER, endoplasmic reticulum; TX-
modulators of N- and O-glycosylation. The
100, Triton-X 100.
acquisition of folding determinants in the

1
Copyright 2005 by The American Society for Biochemistry and Molecular Biology, Inc.
management of prostate cancer (2). PSMA is cytocidal potential of immunotoxins requires
an integral membrane glycoprotein of 110 therefore elucidation of the properties and
kDa, found initially in LNCaP cells by physiology of PSMA in cells. In particular,
immunoprecipitation (3). The analysis of in phenomena like rate of translation, maturation
vivo distribution and the biological function during biosynthesis, association with
of PSMA led to identify the protein not only membrane microdomains, cell surface
in prostate (4,5) but also in several other expression, internalization and turnover all
tissues, where PSMA acts as a glutamate may have important consequences on the
carboxypeptidase on different alternative sensitivity of prostate tumor and normal cells
substrates (6,7). Although the function of to treatment with immunotoxins.
PSMA in prostate has not been clarified yet, As many other type-II transmembrane
its expression is enhanced in higher-grade proteins, PSMA is highly glycosylated. In
cancers, metastatic disease, and hormone- particular, PSMA has ten potential N-
refractory prostate carcinomas (8). Moreover, glycosylation sites. At least nine of them,
PSMA is present at the surface of endothelial located in the PSMA ectodomain, are indeed
cells found in tumors of different histotype modified by carbohydrates that contribute to
but absent from normal endothelial cells (9). the 20-25 % of the molecular weight of the D
o
w
Serum studies have suggested that PSMA native PSMA (16). As expected, a severe n
lo
levels may be linked to a more aggressive impairment of glycosylation, induced for ad
e
d
clinical phenotype (10). instance by treatments with tunicamycin, fro
m
Radioimmunoconjugates of anti-PSMA causes the generation of an enzymatically
h
monoclonal antibodies (mAb) are currently inactive protein that is retained intracellularly ttp
://w
used as imaging agents for prostate cancer (16,17). This is likely to be due to misfolding w
w
(11). Moreover, it has been observed that or inhibition of dimerization, which are .jb
c
following contact with antibodies, PSMA is crucial for the generation of an active protein .o
rg
rapidly and efficiently internalized by the (18). b/
y
target cell and localizes in a juxtanuclear In the present study we report first on the gu
e
s
region of the cell (12). PSMA represents biosynthesis, processing and transport of t o
n
therefore a very attractive marker for passive PSMA in transfected COS-1 cells, where D
e
c
immunotherapy with mAb or their armed PSMA is expressed as a complex em
b
derivatives (13). Indeed, results obtained in glycosylated protein on the cell surface, er 2
0
our laboratory demonstrated that PSMA- similar to PSMA molecules found in tumor , 2
0
positive cells are intoxicated by anti-PSMA tissues or serum (15). The possible role of 18
antibody linked to the ricin toxin A-chain glycans on the trafficking of PSMA toward
with high efficiency (14). Most types of the cell surface was investigated by using
immunotoxins need to be internalized and specific inhibitors of O- and N-glycosylation.
transported through intracellular A similar experimental approach was also
compartments before exerting their cytotoxic used in polarized MDCK cells to determine
effect. This requires that a complex series of whether or not apical sorting signals of PSMA
phenomena take place in an orderly fashion in are located in its O- and N-glycans.
appropriate locations within the cell. These Furthermore, the folding status of PSMA in
events in turn are likely to be influenced by the early biosynthetic stages and in particular
the properties and physiology of the relevant its association with typical ER-resident
target molecules. The enzymatic activity of chaperones were examined.
PSMA and presumably also its folding and
antibody recognition may be influenced by a EXPERIMENTAL PROCEDURES
differential glycosylation in tumor or normal Materials and Reagents – Tissue culture
cells (15). The optimization of the overall dishes were obtained from Greiner (Hamburg,
2
Germany). Streptomycin, penicillin, Canada). The anti-rabbit peroxidase-
glutamine, Dulbecco's modified Eagle's conjugated secondary antibody and the
medium (DMEM), methionine-free DMEM Hybond-P PVDF membranes were obtained
were purchased from Invitrogen (Eggenstein, from Amersham Bioscience. The
Germany). Fetal calf serum, pepstatin, SuperSignal®ELISA Fempto Maximum
leupeptin, aprotinin, trypsin inhibitor, Sensitivity Substrate kit was from Pierce
molecular mass standards for SDS-PAGE, (Rockford, IL) and Kodak BioMax XAR
Biotin-N-hydroxy-succinimide ester (Biotin- Films were from Sigma-Aldrich (Steinheim,
NHS), streptavidin horseradish peroxidase, Germany).
inhibitors of specific N-glycosylation Cloning of human PSMA – A cDNA encoding
reactions (deoxymannojirimycin, swainsonine the human PSMA (hPSMA) was cloned from
and monensin) and an inhibitor of O- LNCaP cells using as a template the mRNA
glycosylation, benzyl-N-acetyl-galactosamine purified from LNCaP cells. The following
(benzyl-GalNAc) were purchased from Sigma primers were designed based on the
Chemical Co. (Deisenhofen, Germany). nucleotide sequence reported by R.S. Israeli
Phenylmethylsulfonyl fluoride, antipain, and et al. (5) and used to obtain two PSMA
soybean trypsin inhibitor were purchased fragments of 1471 bp and 1426 bp in length, D
o
w
from Roche Diagnostics (Mannheim). L- respectively. The primers n
lo
[35S]-methionine (>1000 Ci/mmol) and TGCAGGGCTGATAAGCGAG and ad
e
protein A-Sepharose were obtained from AGCCACGCCACGCTCTTG were used to d fro
m
Amersham Biosciences Inc., (Freiburg, generate a fragment spanning the 5’ region of h
Germany). Acrylamide, N,N'- PSMA that was subcloned into the pBS vector ttp://w
methylenebisacrylamide, and TEMED were (Invitrogen, La Jolla, CA) in correspondence w
w
purchased from Carl Roth GmbH (Karlsruhe, of a SmaI and a EcoRI restriction enzyme .jb
c
Germany). SDS, ammonium persulfate, sites. Two other oligonucleotides (the forward .o
rg
dithiothreitol and TX-100 were obtained from primer TCATCCAATTGGATACTATG and b/
y
Merck, (Darmstadt, Germany). Endo-β- the reverse primer gu
e
s
acetylglucosaminidase H (endo H) and endo- TCTTTCTGAGTGACATAC) enabled to t o
n
β-N-acetylglucosaminidase F glycosidase amplify a second PCR product that was De
c
(endo F) were purchased from New England cloned into the same vector (between a em
b
e
BioLabs (Frankfurt, Germany). Trypsin, EcoRV and a EcoRI restriction enzyme sites) r 2
0
chymotrypsin was purchased by Sigma- and overlapped with the first fragment of , 2
0
Aldrich (Taufkirchen, Germany). The anti- PSMA. The complete cDNA sequence of 18
PSMA mAb J591, described by Liu H. et al. PSMA was finally digested with BamHI and
(12), was kindly supplied by Dr. N.H. Bander XhoI enzymes and subcloned into the
(Medical College of Cornell University, New pcDNA3.1 vector (Invitrogen, Carlsbad, CA)
York). The murine 7E11c mAb (3), to obtain the pcDNA 3.1-hPSMA construct.
recognizing the cytosolic tail of PSMA, was Transient transfection of COS-1 cells,
purified from the supernatant of the biosynthetic labeling, and
hybridoma HB-10494 (American Type immunoprecipitation of the total cellular
Culture Collection, Rockville Pike, MD) extract – COS-1 cells were cultured in 10-cm
according to standard chromatographic Petri dishes and in DMEM medium
methods. The anti-sucrase-isomaltase (SI) supplemented with 10% FBS and 2 mM L-
antibodies, mGlu1, mGlu2 and mGlu3, were a glutamine. When the cells reached a
generous gift of Dr. D.Swallow (The Galton confluence of about 30-40%, they were
Laboratory, University College London, UK). transiently transfected with 5 µg of pcDNA
The anti-BiP and anti-calnexin antibodies 3.1-hPSMA. The transfection was carried out
were purchased from Stressgen (Victoria, according to the DEAE-dextran-based method
3
described previously (19). Briefly, cells were When required, immunoprecipitated proteins
incubated in the presence of DNA and were subjected to enzymatic digestion with
DEAD-dextran for 1 h at the standard either endo-β-N-acetylglucosaminidase H
culturing conditions and then left for further 3 (endo H) or with endo-β-N-
h in the complete DMEM medium acetylglucosaminidase F (endo F) according
supplemented with 1 µg/ml chloroquine. to a method described previously (21).
After washings, the cells were Furthermore, the effect of proteases on PSMA
biosynthetically labeled with 100 µCi [35S]- immunoprecipitated in its native form was
methionine for different time intervals, investigated. After washings, the protein A-
according to the experimental protocol. In Sepharose beads binding the
pulse-chase experiments, the cells were immunocomplexes were resuspended in 25 µl
pulsed for 30 min and then incubated for of 0,1% TX-100/PBS containing different
further 30 min, 1, 1.5, 2, 4 or 6 h at 37°C in amounts of either trypsin or chymotrypsin
the presence of fresh medium containing cold (see the Results section) and 10 µg bovine
methionine. serum albumin, used as a carrier. The reaction
The cells were solubilized on ice for 1 h by was prolonged for 30 min at 37°C and
using a combination of 0.5% TX-100 and blocked by boiling the samples at 95°C in the Do
w
0.5% sodium deoxycholate, dissolved in PBS, presence of additional 25 µl of Laemmli- nlo
a
supplemented with a cocktail of protease containing reducing buffer. In parallel, de
d
inhibitors (1 mM phenylmethylsulfonyl sensitivity of sucrase-isomaltase (SI) to 100 fro
m
fluoride, 1 µg/ml pepstatin, 5 µg/ml leupeptin, µg trypsin was also assayed (22). After h
ttp
5 µg/ml aprotinin, 1 µg/ml antipain and 50 transfection of COS-1 cells with a cDNA ://w
µg/ml trypsin inhibitor). After removal of cell encoding human SI cloned into pSG8-vector ww
debris by centrifugation and pre-clearing of (23), immunoprecipitation was performed by .jb
c
.o
the total extract with protein A-Sepharose (as using a combination of mGlu1, mGlu2 and rg
b/
described in more detail in Ref. 20), the mGlu3 anti-SI mAb. The intensity of the y
g
u
supernatant was immunoprecipitated with a protein bands obtained by autoradiography e
s
combination of the J591 and 7E11c anti- was then quantified by using the Quantity t on
D
PSMA mAb. In some experiments One software (BIO-RAD). ec
e
m
immunoprecipitation with each individual Co-immunoprecipitation with BiP and b
e
antibody, i.e. 7E11c or J591, was performed. calnexin – To assess the possible interaction r 2
0
Solubilization of cells was also performed of PSMA with BiP or calnexin, total cellular , 20
1
under denaturing conditions. Here, the cells extracts from four plates of transiently 8
were treated for 10 min at room temperature transfected COS-1 cells were
with 0.1% SDS in PBS and the mixture of immunoprecipitated with either anti-BiP or
protease inhibitors. TX-100 was then added to anti-calnexin mAb (24). To study the
the extracts at a final concentration of 1% to interaction with BiP, pulse-chase experiments
obtain a 10-fold concentration of TX-100 were performed by chasing the cells for 30
versus SDS. Immunoprecipitation was then min, 1, 2, 4 or 6 h after 30 min labeling with
carried out using the 7E11c antibody. [35S]-methionine. After the indicated time
Antigen-antibody complexes were recovered intervals, the cells were washed with cold
by using 30 µl of protein A-Sepharose beads PBS and lysed in a buffer containing 1% TX-
and then eluted by boiling in 1% SDS and 100 and supplemented with a cocktail of
processed by SDS-PAGE on 7% protease inhibitors (see above) and 10 µg of
polyacrylamide gels. Proteins were finally apyrase (Sigma). When a possible interaction
detected by exposure of the gel to a Kodak X- with calnexin was assayed, cells were pulsed
Omat AR film and autoradiography (using a for 1 hr at 37°C and chased for 4 h either at
PhosphorImager purchased by Bio-Rad). 37°C or at 15°C. Immunoprecipitated proteins
4
were eluted from protein A-Sepharose by supernatant by J591 and 7E11c mAb used in
boiling for 5 min at 95°C in the presence of combination.
1% SDS. Following dilution with 1% TX- To test the role of inhibitors on the maturation
100, samples were further processed to detect process of membrane proteins, inhibitors of
PSMA with J591 and 7E11c. Protein fractions the glycosylation were separately added to the
depleted from BiP or calnexin were separately medium during the preincubation of
immunoprecipitated to detect non interacting transfected COS-1 cells in methionine-free
PSMA molecules. The possible interaction of medium as well as during the labeling with
PSMA and SI was investigated also by [35S]-methionine. Benzyl-GalNAc was used
western blot. PSMA and SI were first as an inhibitor of the O-glycosylation at 4
immunoprecipitated from transiently mM final concentration, whereas the
transfected COS-1 cells, according to the following inhibitors of the N-glycosylation
method described above. Proteins were process were considered:
subjected to SDS-PAGE on 7% deoxymannojirimycin (dMM), which inhibits
polyacrylamide gels and blotted onto a PVDF the cis-Golgi mannosidase I, swainsonine and
membrane according to manufactures’ monensin, active in the medial-Golgi and the
instructions. Anti-BiP and anti-calnexin trans-Golgi, respectively. dMM was used at 5 D
o
w
antibodies were added to the membranes mM, whereas swainsonine and monensin n
lo
followed by anti-rabbit peroxidase-conjugated were given to cells at 4 µg/µl and 1 µM final ad
e
d
secondary antibodies. The antigenic detection concentration, respectively. fro
m
in a chemiluminescent reaction employed Detection of cell surface antigens by h
SuperSignal®ELISA Fempto Maximum biotinylation – COS-1 cells were transfected ttp://w
Sensitivity Substrate kit. Western blot with pcDNA 3.1-hPSMA as described above w
w
analysis was also performed using total and incubated 2 h posttransfection with the .jb
c
lysates from COS-1 cells that were glycosylation inhibitors dMM (5 mM), .o
rg
transfected with pcDNA 3.1-hPSMA. Here, swainsonine (4 µg/µl), benzyl-GalNAc (4 b/
y
g
50 µg cellular proteins were used. The mM) or without the inhibitors. The media u
e
s
antibody 7E11c was used for antigenic were changed 3 times during a period of 48 h t o
n
detection. This antibody recognizes denatured and fresh glycosylation inhibitors were added D
e
c
e
forms of both the mannose-rich and complex at the same concentrations. Thereafter, the m
b
e
glycosylated form of PSMA (see “Results”). cells were washed twice with ice-cold PBS to r 2
0
Cell surface immunoprecipitation from COS- remove amine-containing culture media and , 2
0
1 cells – COS-1 cells were transfected and cellular proteins. Biotin-NHS in PBS at 1 18
labeled following the same procedure µg/ml was added to the cells and left on ice
described above. Two days post-transfection, for 30 min. The cells were then washed twice
cells were labeled for 6 h and surface- with ice-cold PBS, quenched with 100 mM
expressed PSMA molecules were glycine and solubilized as described above.
immunoprecipitated from intact cells by using Immunoprecipitation with anti-PSMA
the J591 antibody. Unbound antibodies were antibodies (J591/7E11c) was performed and
removed by washing the cells three times with the immunoprecipitates were subjected to
cold PBS and were further subtracted by western blot analysis. The chemiluminescent
using a cold-lysate of COS-1 cells transfected detection of PSMA was carried out using
with PSMA. Cells were lysed for 1 h at 4°C streptavidin horseradish peroxidase and
and the antigen-antibody complexes from the SuperSignal®ELISA Fempto Maximum
cell surface were recovered with the addition Sensitivity Substrate kit.
of protein A-Sepharose. Intracellular PSMA Indirect immunofluorescence – Subcellular
was then immunoprecipitated from the localization of PSMA in transfected COS-1
cells in the presence or absence of
5
glycosylation inhibitors was performed using RESULTS
indirect immunofluorescence. Here, the Biosynthesis and glycosylation of PSMA –
transfected cells were incubated on cover slips COS-1 cells were transiently transfected with
with or without the glycosylation inhibitors a cDNA encoding human PSMA (pcDNA
dMM, swainsonine and benzyl-GlaNAc using 3.1-hPSMA) and biosynthetically-labeled
similar procedure as described above. The with [35S]-methionine for 4 h. After
cells were fixed with 4% paraformaldehyde solubilization with a combination of TX-100
and permeabilized with 0.5% saponin to and sodium deoxycholate, PSMA was
visualize intracellular PSMA. The cell surface immunoprecipitated from the cell lysates by
localization was analysed in non- using a combination of two anti-PSMA mAb,
permeabilized cells. Immunolabeling was J591 and 7E11c. Preliminary experiments
carried out using anti-PSMA antibodies elucidated that mAb J591, directed against a
(combination of J591 and 7E11c) at 1:100 luminal epitope of PSMA,
dilution. The secondary antibody was rabbit immunoprecipitated both the immature
anti-mouse conjugated to Alex Fluor 488. The precursor and the complex glycosylated form
fluorescence images were visualized using a (of approximately 100 and 110 kDa,
Leica TCS SP2 confocal laser microscope respectively) with comparable efficiency, D
o
w
with an X63 water planapochromat lens whereas 7E11c, recognizing the cytosolic tail, n
lo
(Leica Microsystems). preferentially binds to the immature protein. ad
e
d
Generation of MDCK stable cell lines The immature and the mature forms were fro
m
expressing hPSMA and cell surface discriminated based on their sensitivity to
h
immunoprecipitation – Stable MDCK cells treatment with endo H (Fig. 1). Endo H ttp
://w
expressing hPSMA were obtained by using an digests mannose-rich N-linked glycans, w
w
Eppendorf microinjector (FemtoJet whereas complex carbohydrates are resistant .jb
c
Eppendorf, Hamburg, Germany). Selection of to this enzyme. The substantial shift in the .o
rg
cells containing pcDNA 3.1-hPSMA were size of the lower 100-kDa band upon endo H b/
y
positively selected by adding neomycin (500 treatment confirmed that this band gu
e
s
µg/ml) to the DMEM medium. PSMA corresponds to the ER-localized precursor of t o
n
expression was finally confirmed by PSMA. To gather information on the D
e
c
immunoprecipitation (see above). trafficking rate of the PSMA between the ER em
b
To immunoprecipitate selectively PSMA and the Golgi and the kinetics of the er 2
0
molecules from the apical or from the glycosylation event, transfected COS-1 cells , 2
0
basolateral surface of polarized MDCK, cells were labeled with [35S]-methionine for 30 min 18
were cultured to complete confluence on followed by several chase times (Fig. 1). 1 h
transmembrane filters (25). During the pre- into the chase, the biosynthetic forms of
incubation with methionine-free medium and PSMA comprised a major mannose-rich form
the 4 hr pulse, cells were treated either with 5 and a faint protein band that has acquired a
mM dMM or with 4 mM benzyl-GalNAc, as substantial endo H-resistance indicative of
inhibitors of the N- and the O-glycosylation, complex glycosylation and maturation. The
respectively. The J591 anti-PSMA mAb was half life for the conversion of mannose-rich
used to immunoprecipitate PSMA from both PSMA to a complex glycosylated species was
the apical and the basolateral surface of about 120 min. After 4 h and 6 h the
MDCK cells, whereas a combination of both proportion of the mature protein progressively
J591 and 7E11c mAb precipitated increased. The persistence of major
intracellular PSMA molecules. proportions of PSMA as a mannose-rich
glycosylated polypeptide after 6 h of chase
supports the notion that the transport rates of
PSMA are very slow.
6
Implication of N- and O-glycosylation in the affected by incubating transfected COS-1
transport and cell-surface expression of cells during biosynthesis with dMM,
PSMA – Altered and aberrant glycosylation swainsonine and monensin, all of which are
has been described to occur in essentially all reagents acting at different sites in the Golgi
types of experimental and human cancers and apparatus. Qualitative differences in the
many glycosyl-epitopes constitute tumor- electrophoretic mobility of PSMA were
associated antigens (26). It is still unclear obtained in the presence of these inhibitors
whether altered glycosylation in tumors compatible with altered carbohydrate patterns.
affects the trafficking and therefore the Nevertheless, the modified glycoforms of
exposure of tumor-associated antigens at the PSMA reached the cell surface as the fully
cell surface. We wanted therefore to glycosylated PSMA indicating that
investigate the glycosylation pattern of PSMA modification of N-glycosylation is most likely
and its possible implication in the targeting not implicated in the transport of PSMA to
and sorting of the protein. the surface of COS-1 cells (Fig. 3A). A
Treatment of PSMA with endo F, which possible role of O-glycans was analyzed by
cleaves both mannose-rich as well as complex using benzyl-GalNAc. Here again, the
glycans, has resulted in a drastic shift in the modified protein reached the cell surface as D
o
w
apparent molecular weight of PSMA its control counterpart suggesting that also O- n
lo
concomitant with a high level of N-linked glycans do not influence the trafficking of ad
e
d
glycosylation of the protein (Fig. 2). The PSMA. fro
m
diffuse heterogeneous pattern of the protein We corroborated these data by employing cell
h
band after endo F treatment is most likely due surface biotinylation followed by western blot ttp
://w
to the sialylation of endo F-resistant O- detection of PSMA glycoforms. As shown in w
w
glycans. This type of glycosylation was Fig. 3B the modified forms of PSMA were .jb
c
examined by utilizing the sugar analogue detected at the cell surface and in almost .o
rg
benzyl-GalNAc in biosynthetic labeling similar amounts supporting the previous b/
y
experiments of transfected COS-1 cells. biosynthetic data. gu
e
s
Benzyl-GalNAc strongly interferes with the Finally we investigated the cell surface and t o
n
O-glycosylation events by competing with intracellular distribution of PSMA in the D
e
c
GalNAc (27) and has been frequently used to presence or absence of the glycosidase em
b
examine the presence of O-glycan units in modulators using indirect er 2
0
glycoproteins (28,29). Fig. 2 demonstrates immunofluorescence. In non-permeabilized , 2
0
that benzyl-GalNAc treatment resulted in a transfected cells PSMA was found at the cell 18
significant reduction in the size of PSMA surface (Fig. 3C, upper panel). There was no
indicative of the existence of O-glycans in significant difference in the images obtained
PSMA. in the presence or absence of the inhibitors
To investigate whether the extensive N- and dMM, swainsonine or benzyl-GalNAc
O-glycosylation of PSMA may modulate providing another support to the notion that
intracellular transport and delivery of PSMA modified glycoforms of PSMA are also
to the cell surface, we utilized a panel of transported to the cell surface. The
inhibitors of N- and O-glycosylation in intracellular localization of PSMA was also
combination with cell surface analyzed in saponin-permeabilized cells (Fig.
immunoprecipitation experiments. The J591 3C, lower panel). Here, the cell surface as
mAb was used to immunoprecipitate PSMA well as the Golgi membranes and ER reticular
from intact cells, whereas a combination of networks were labeled supporting the
J591 and 7E11c precipitated PSMA retained biosynthetic pulse-chase data and the
in the intracellular compartments from total constitutive trafficking pathway of PSMA.
cell lysates (Fig. 3A). N-glycosylation was
7
Sorting of PSMA in stably transfected MDCK found (Fig. 5B). The band corresponding to
cells – For some glycoproteins N- and O- immature PSMA was discerned when a
glycans have been proposed to act as signals temperature block at 15°C for 4 h was
for apical targeting in polarized cells performed, which facilitated the accumulation
(28,30,31). Predominant apical transport of of newly synthesized proteins in the ER. The
PSMA has been recently demonstrated (32) in binding to BiP and calnexin was also probed
MDCK cells. Here, we address the question by western blotting (Fig. 5C). Here, PSMA
whether impairment of N-glycosylation or was immunoprecipitated from total cell
prevention of O-glycosylation can cause a lysates containing all forms of PSMA at
missorting of PSMA. steady state and the immunoprecipitates were
Fig. 4 illustrates the results of cell surface blotted with anti-BiP (Fig. 5C, left) or anti-
immunoprecipitation of PSMA from the calnexin antibodies (Fig. 5C, right). Here
apical (A) and basolateral (B) membranes of a again, PSMA did not associate with BiP and
MDCK cell line stably transfected with calnexin bound weakly to the protein. SI (and
PSMA. PSMA was predominantly found at also lactase phlorizin hydrolase, data not
the apical membrane (left panel). The shown), whose interaction with both BiP and
basolateral membrane was almost devoid of calnexin was previously investigated (24,34), D
o
w
PSMA, whereas the corresponding cellular were used here as positive controls. The n
lo
lysates contained the protein (right panel). In binding of both proteins to calnexin was many ad
e
d
the presence of the O-glycan inhibitor, folds stronger than that of PSMA. BiP is fro
m
benzyl-GalNAc, a reduction in the size of therefore not directly involved in the folding
h
PSMA occurs, but no alteration in the cell events of PSMA. The weak binding of PSMA ttp
://w
surface expression of PSMA was observed. to calnexin strongly suggests that it occurs w
w
Similarly, dMM, used here as a modulator of transiently for a short time before the third .jb
c
N-linked glycosylation, did not affect either glucose is trimmed in the ER (35) and .o
rg
the sorting profile of the modified PSMA proposes that PSMA does not enter into b/
y
glycoform. Together, the results indicate that consecutive glucosylation and gu
e
s
processing of N- and O-glycosylation in the deglucosylation cycles until it has acquired t o
n
Golgi apparatus has no marked effect on the correct folding. D
e
c
sorting of PSMA. The next step was to examine the folding of em
b
Folding of PSMA – Overexpression of PSMA PSMA and the time point of attainment of er 2
0
in prostate cancer could generate aberrantly PSMA to a correct folding. We analysed this , 2
0
folded PSMA molecules that escape the issue by evaluating the sensitivity or 18
quality control machinery in the ER (33). We resistance of PSMA towards proteases during
wanted therefore to analyze the folding events intracellular trafficking. Here, two different
of PSMA along its route to the cell surface in proteases with distinct specificities were
an overexpression cell system such as COS-1 tested. Fig. 6A shows a representative
cells and in MDCK cells, in which low copy experiment performed by treating the
of the PSMA gene is expressed. immunoprecipitated proteins with trypsin and
We first studied a possible association of chymotrypsin. Strikingly, PSMA was found
overexpressed PSMA in COS-1 cells with the either resistant or completely degraded by
ER-resident proteins BiP and calnexin. In a trypsin without producing a defined pattern of
pulse-chase analysis (Fig. 5A), PSMA did not smaller fragments upon enzymatic digestion.
co-immunoprecipitate with BiP, whereas it Comparison between treated- and non treated-
was detected in the fractions that were samples showed that the mannose-rich form
immunodepleted from BiP. A very weak of PSMA is fully sensitive to trypsin, while
association with calnexin, a lectin-like the complex glycosylated PSMA acquires a
membrane-bound chaperone of the ER, was substantial resistance towards the protease.
8
An almost similar pattern was revealed by dMM inhibits the function of α-mannosidase
chymotrypsin. The complete sensitivity of I in the cis-Golgi (37) and a mannose-rich like
mannose-rich PSMA to trypsin did not species is the predominant form along the
change in pulse-chase experiments (Fig. 6B) secretory pathway as has been observed for
strongly suggesting that no conformational many proteins (38,39). It should be noted that
alteration in mannose-rich glycosylated this mannose-rich form of PSMA is slightly
PSMA has occurred. In MDCK cells similar different from that obtained in the absence of
results were obtained indicating that dMM, since it shows a faint smear upon endo
overexpression of PSMA in COS-1 cells has H treatment, due to partial processing of the
no relevance to the acquisition of malfolded mannose-rich PSMA. In the first 30 min of
conformation by PSMA molecules in the ER chase the dMM-treated PSMA was entirely
(data are not shown). The trypsin sensitivity sensitive to trypsin as shown for the non-
pattern of PSMA is unique to this protein treated protein. By contrast, 2 h into the chase
since its mannose-rich form never acquires at which time PSMA has arrived the Golgi the
trypsin-resistance during its life cycle as other endo H-sensitive mannose-rich form was
many other membrane proteins (for example retained, but has nevertheless acquired a
see 22, 24, 30,36). substantial trypsin resistance. Strikingly, the D
o
w
The complete sensitivity of mannose-rich trypsin-resistance increases with chase times n
lo
PSMA and the resistance of its mature form from 2 h to 5 h (from 15% to 60% ad
e
d
towards trypsin led us to ask whether these approximately). Obviously, the acquired fro
m
differences are glycosylation-dependent or are resistance of complex glycosylated PSMA to h
rather the consequence of altered folding trypsin is not due to the type of glycosylation ttp://w
determinants that occur during intracellular per se, otherwise the dMM-treated mannose- w
w
transport after ER exit. To address this rich PSMA would have been also degraded. It .jb
c
question, the pattern of N-glycosylation was is rather the folding pattern of PSMA that .o
rg
modulated by labeling transfected COS-1 dictates its sensitivity or susceptibility to the b/
y
cells in the presence of dMM or swainsonine protease. As such the acquired resistance of gu
e
s
that inhibit processing of the PSMA chains in the dMM mannose-rich PSMA to trypsin at 2 t o
n
the cis- and medial Golgi, respectively. Cells h or 5 h of chase is solely due to correct D
e
c
were pulsed for 30 min (Fig. 7A, left panel) in folding of the protein that should be localized em
b
order to visualize only the mannose-rich form in the Golgi. This is supported by the er 2
0
of PSMA in the ER, or for 5 h (Fig. 7A, right observation that the half life of the mannose- , 2
0
panel) to detect also further processed PSMA rich PSMA form is approximately 2 h, at 18
forms. All the mannose-rich forms obtained in which time the protein should have traversed
the presence or absence of the inhibitors after the ER to the Golgi (see pulse-chase data
30 min of pulse labeling were completely above in Fig. 1). To further substantiate these
degraded with trypsin confirming the data we performed a 20°C temperature block
sensitivity of this form to the protease. After 5 to retain and accumulate PSMA molecules in
h of labeling in the presence of swainsonine the Golgi followed by examining their trypsin
the modified glycoforms acquire trypsin- sensitivity. PSMA was visualized as a diffuse
resistance comparable to the non-treated band that possibly comprises multiple
counterpart. Here again, the mannose-rich processed glycoforms of the protein.
form remained trypsin-sensitive and was Treatment of this protein form with trypsin
degraded as observed before. In the presence revealed a similar type of band indicating an
of dMM an endo H-sensitive protein of a acquisition of correct trypsin-resistant
similar apparent molecular weight as its non- conformation (Fig. 7C). On the contrary,
treated mannose-rich counterpart was the molecules that were retained in the ER were
predominant glycoform of PSMA (Fig. 7B). again completely degraded. Altogether, the
9
data unequivocally demonstrate that the polypeptide through out its life cycle and also
mannose-rich ER form of PSMA does not the slow transport rate between the ER and
acquire trypsin resistance at any time along its the Golgi. At first glance these characteristics
biosynthetic pathway. suggest that PSMA is retained for a prolonged
We further assessed the folding of PSMA by time in the ER where it associates with
utilizing the 7E11c antibody. Fig. 8 shows members of the quality control machinery,
that the mAb 7E11c immunoprecipitates such as BiP and calnexin, until it has acquired
almost exclusively the mannose-rich form of a correct conformation and transport-
PSMA under native conditions from Triton competence before ER egress.
X-100 extracts of cells that were Strikingly, however, a conformational
biosynthetically labeled for 5 h. For maturation of mannose-rich PSMA does not
comparison, the J591 antibody has a wider occur as evidenced by protease sensitivity
specificity and reacts strongly with both tests and antibody binding. Remarkably, the
biosynthetic forms of PSMA, the mannose- initial mannose-rich biosynthetic form of
rich as well as the complex glycosylated PSMA is readily degraded with trypsin (or
species. On the other hand, western blot chymotrypsin) and does not acquire a
analysis of PSMA using mAb 7E11c clearly different proteolytic pattern through out its D
o
w
demonstrates that this antibody has reacted entire biosynthetic life cycle. This strongly n
lo
not only with the mannose-rich form, but also suggests that no gross conformational ad
e
d
with the complex glycosylated PSMA protein alterations have occurred in mannose-rich fro
m
(right lane). Likewise, when denatured lysates PSMA. By contrast, complex glycosylated
h
from 5 h biosynthetically labeled cells were PSMA is predominantly resistant to the ttp
://w
immunoprecipitated with the 7E11c antibody proteases indicative of acquisition of a correct w
w
an appreciable amount of the complex folding or of a protective function of the .jb
c
glycosylated PSMA became now detectable complex oligosaccharides. Nevertheless, this .o
rg
(middle panel). The 7E11c antibody is latter possibility could be excluded, since b/
y
capable therefore of discriminating between modulation and inhibition of complex gu
e
s
differently folded forms of complex glycosylation of PSMA in the Golgi, for t o
n
glycosylated PSMA, while it reacted with the example, by dMM or swainsonine results in D
e
c
mannose-rich form under native as well as equally trypsin-resistant glycoforms. This em
b
denaturing conditions. This result is in line finding is surprising for an endogenous er 2
0
with the observations above that the folding protein in view of current knowledge on , 2
0
of the complex glycosylated PSMA differs protein anterograde transport and the 18
from that of the mannose-rich protein and that consensus that proteins fold and acquire
gross alterations in the folding of this latter transport-competent conformation in the ER
form do not occur. before leaving this organelle (40). Also the
binding patterns of early biosynthetic PSMA
DISCUSSION forms to typical ER chaperones, such as BiP
Understanding the molecular mechanisms and calnexin, do not conform to well known
underlying the biosynthesis, folding, concepts of quality control mechanism in the
intracellular transport and polarized sorting of ER (41). While BiP does not associate with
the type II membrane glycoprotein PSMA are PSMA at all, calnexin does albeit to a very
essential and crucial prerequisites for its low extent, suggesting that it does not enter
utilization as a therapeutically suitable target into a glucosylation and deglucosylation
in prostate cancer. One interesting feature in calnexin cycles (42). An implication of other
the biosynthetic features of PSMA is the chaperones, such as ERp57 (43), in an early
presence of high proportions of PSMA in an binding to PSMA can not be completely
early immature mannose-rich glycosylated excluded. A role of the ER and its chaperones
10