Using biochemical techniques and confocal laser scanning microscopy analysis, we demonstrated that GD3 and GM1, after being synthesized at the Golgi apparatus, were transported and accu-
Trang 1MDCK cells and internalized by clathrin-independent
endocytosis
Pilar M Crespo, Natalia von Muhlinen, Ramiro Iglesias-Bartolome´ and Jose L Daniotti
Centro de Investigaciones en Quı´mica Biolo´gica de Co´rdoba (CIQUIBIC, UNC-CONICET), Departamento de Quı´mica Biolo´gica, Facultad de Ciencias Quı´micas, Universidad Nacional de Co´rdoba, Argentina
Gangliosides are complex glycosphingolipids
contain-ing one or more sialic acid residues, which are mainly
located at the outer leaflet of the plasma membrane of
eukaryotic cells They participate in cell surface events,
such as the modulation of growth factor receptors and
cell-to-cell and cell-to-matrix interactions [1–6] The
synthesis of gangliosides is carried out in the lumen of the Golgi complex by a complex system of membrane-bound glycolipid acceptors, glycosyltransferases and sugar nucleotide transporters [7,8] After synthesis, gangliosides leave the Golgi complex via the lumenal surface of transport vesicles In this context, we have
Keywords
gangliosides; glycolipids; intracellular
trafficking; MDCK cells; polarized cells
Correspondence
J L Daniotti, Centro de Investigaciones en
Quı´mica Biolo´gica de Co´rdoba (CIQUIBIC,
UNC-CONICET), Departamento de Quı´mica
Biolo´gica, Facultad de Ciencias Quı´micas,
Universidad Nacional de Co´rdoba, Haya de
la Torre y Medina Allende, Ciudad
Universitaria, X5000HUA Co´rdoba, Argentina
Fax: +54 351 433 4074
Tel: +54 351 433 4168⁄ 4171
E-mail: daniotti@dqb.fcq.unc.edu.ar
(Received 4 August 2008, revised
21 September 2008, accepted 8 October
2008)
doi:10.1111/j.1742-4658.2008.06732.x
Gangliosides are glycosphingolipids mainly present at the outer leaflet of the plasma membrane of eukaryotic cells, where they participate in recogni-tion and signalling activities The synthesis of gangliosides is carried out in the lumen of the Golgi apparatus by a complex system of glycosyltrans-ferases After synthesis, gangliosides leave the Golgi apparatus via the lumenal surface of transport vesicles destined to the plasma membrane In this study, we analysed the synthesis and membrane distribution of GD3 and GM1 gangliosides endogenously synthesized by Madin–Darby canine kidney (MDCK) cell lines genetically modified to express appropriate gan-glioside glycosyltransferases Using biochemical techniques and confocal laser scanning microscopy analysis, we demonstrated that GD3 and GM1, after being synthesized at the Golgi apparatus, were transported and accu-mulated mainly at the plasma membrane of nonpolarized MDCK cell lines More interestingly, both complex gangliosides were found to be enriched mainly at the apical domain when these cell lines were induced to polarize
In addition, we demonstrated that, after arrival at the plasma membrane, GD3 and GM1 gangliosides were endocytosed using a clathrin-independent pathway Then, internalized GD3, in association with a specific monoclonal antibody, was accumulated in endosomal compartments and transported back to the plasma membrane In contrast, endocytosed GM1, in associa-tion with cholera toxin, was transported to endosomal compartments
en route to the Golgi apparatus In conclusion, our results demonstrate that complex gangliosides are apically sorted in polarized MDCK cells, and that GD3 and GM1 gangliosides are internalized by clathrin-indepen-dent endocytosis to follow different intracellular destinations
Abbreviations
CHO, Chinese hamster ovary; CTx, cholera toxin; DiI, 1,1¢-dioctadecyl-3,3,3¢,3¢-tetramethylindocarbocyanine perchlorate; Eps15, epidermal growth factor receptor pathway substrate clone 15; GalNAc-T, UDP-GalNAc:LacCer ⁄ GM3 ⁄ GD3 N-acetylgalactosaminyltransferase; Gal-T2, UDP-Gal:GA2 ⁄ GM2 ⁄ GD2 ⁄ GT2 galactosyltransferase; GFP, green fluorescent protein; GPI, glycosylphosphatidylinositol; HA, hemagglutinin; LacCer, lactosylceramide; MDCK, Madin–Darby canine kidney; PI, propidium iodide; Sial-T2, CMP-NeuAc:GM3 sialyltransferase; Tf,
transferrin; TGN, trans-Golgi network; YFP, yellow fluorescent protein.
Trang 2demonstrated, in Chinese hamster ovary (CHO)-K1
cells, that gangliosides traffic from the trans-Golgi
network (TGN) to the plasma membrane by a
Rab11-independent and Brefeldin A-insensitive exocytic
pathway [9] Gangliosides have been found to reside in
glycosphingolipid-enriched microdomains (also called
detergent-resistant membranes or rafts), dynamic
assemblies of cholesterol, saturated phospholipids and
sphingolipids [10–12]
The renal epithelial Madin–Darby canine kidney
(MDCK) cell line is a recognized cellular model system
for the study of protein targeting because of its ability,
when grown to confluence, to form a polarized
mono-layer [13] Polarized epithelial cell surface membranes
are divided into apical and basolateral domains
possess-ing distinct protein and lipid compositions that are
sepa-rated by tight junctions It is known that proteins can be
targeted directly to the apical or basolateral membrane
from the TGN via the exocytic pathway [14–16] They
can also be targeted indirectly by being delivered to one
domain, typically the basolateral domain, endocytosed
and then redirected to the opposite domain in a process
termed ‘transcytosis’ [17] Alternatively, proteins can be
randomly targeted to both domains and achieve their
asymmetric distribution by selective stabilization at one
plasma membrane [18] In MDCK cells, newly
synthe-sized apical and basolateral membrane proteins are
seg-regated into separate transport vesicles within the TGN
by virtue of sorting signals within the protein [13]
Spe-cifically, the apical targeting of proteins within MDCK
cells can be mediated by sequestration into apical
trans-port vesicles via association with sorting platforms rich
in cholesterol and sphingolipid-rich lipid rafts at the
level of the TGN [19,20] The prevention of the
associa-tion of apical proteins with lipid rafts perturbs the apical
sorting of these proteins [21]
Polarized distribution of sphingolipids has been
reported in different cell types In migrating
lympho-cytes, GM1 localizes to the uropods, whereas another
form, GM3, segregates to the leading edge [22] In
fully polarized human hepatoma HepG2 cells,
C6-NBD-GlcCer, a fluorescent sphingolipid analogue
of glucosylceramide, recycles between the subapical
compartment and the apical, bile canalicular
membrane By contrast, C6-NBD-SM, a fluorescent
sphingolipid analogue of sphingomyelin, initially
accumulates in the subapical compartment, but is
ulti-mately transported to the basolateral membrane [23]
Apical membranes from MDCK cells have generally
been found to be enriched mainly in neutral
glycos-phingolipids and sphingomyelin, whereas
phosphati-dylcholine is concentrated in the basolateral domain
(for a discussion, see [24–26])
In this study, we analysed the subcellular distribu-tion of GD3 and GM1 gangliosides endogenously expressed in both polarized and nonpolarized MDCK cell lines stably transfected to express CMP-NeuAc: GM3 sialyltransferase (Sial-T2, GD3 synthase) and UDP-GalNAc:LacCer⁄ GM3 ⁄ GD3 N-acetylgalactos-aminyltransferase (GalNAc-T, GM2 synthase) glyco-syltransferases GD3 and GM1 gangliosides were found to be enriched mainly at the apical domain when these cell lines were induced to polarize In addi-tion, we found that GD3 ganglioside, in association with a specific monoclonal antibody, was endocytosed using a clathrin-independent pathway and recycled back to the plasma membrane GM1 ganglioside, in association with cholera toxin (CTx), was also actively endocytosed in MDCK cells and transported to the en-dosomal compartment en route to the Golgi complex
Results
MDCK cells synthesize and express GD3 and GM1 gangliosides at the cell surface after transfection with Sial-T2 and GalNAc-T MDCK cell lines expressing different gangliosides were generated by stable transfection with the cDNA encod-ing either chicken full-length Sial-T2 [tagged with the hemagglutinin (HA) epitope] or human full-length Gal-NAc-T (tagged with the c-myc epitope) under the con-trol of constitutive promoters The expression of the recombinant glycosyltransferases was characterized by double immunostaining GalNAc-T and Sial-T2 were found to be located predominantly in a region near the cell nucleus, colocalizing with the area of immunostaining of GM130, a cis-medial Golgi marker [27] (Fig 1A) Moreover, full-length Sial-T2 and Gal-NAc-T colocalized with their respective fluorescent truncated versions (Sial-T2-YFP and GalNAc-T-YFP; YFP, yellow fluorescent protein), which were also found to concentrate at the Golgi apparatus in another epithelial cell [28] (results not shown)
Wild-type MDCK cells predominantly express the ganglioside GM3, as shown in the pattern of radioactive lipids metabolically labelled from d-[U-14C]galactose (Fig 1B, MDCK wt) Cells stably transfected with the cDNA encoding chicken Sial-T2 (Clone 13) mostly synthesize GM3 and GD3 (MDCK Sial-T2); a minor amount of GT3 was also observed, probably also synthe-sized by Sial-T2, as reported previously [29] The notice-able decrease in lactosylceramide (LacCer) labelling was most probably a result of conversion to GM3, which is transformed to GD3 and GT3 by the activity of trans-fected sialyltransferase In contrast, MDCK cells stably
Trang 3expressing the human full-length GalNAc-T cDNA
(Clone 22) synthesize GM3 and GA2 (MDCK
GalNAc-T) The expression of GM1 was below the limit of
detec-tion, even at longer exposure times, probably indicating
inefficient coupling between GalNAc-T and GM3, its
endogenous substrate, and⁄ or a very low activity of
endogenous UDP-Gal:GA2⁄ GM2 ⁄ GD2 galactosyltrans-ferase (Gal-T2) in MDCK cells to catalyse the conversion
of GM2 to GM1 In addition, no expression of GM1 was observed in an extract from GalNAc-T-expressing MDCK cells when analysed by TLC immunostaining using an antibody to GM1 (results not shown)
A
B
C
Fig 1 Characterization of stably transfected
MDCK cell clones (A) MDCK clones stably
expressing Sial-T2-HA (MDCK Clone 13) or
GalNAc-T-myc (MDCK Clone 22) were
dou-ble immunostained with antibodies to HA
(green) and GM130, a Golgi apparatus
mar-ker (red, top panels), or antibodies to myc
(green) and GM130 (red, bottom panels).
The right-hand panels show merged images
from the glycosyltransferases (green) and
GM130 (red) Single confocal sections were
taken every 0.7 lm parallel to the coverslip.
Scale bars ¼ 10 lm (B) A schematic
repre-sentation of the pathway of glycolipid
bio-synthesis is shown on the left Wild-type
MDCK cells (MDCK wt) and cells from
clones 13 (MDCK Clone 13) and 22 (MDCK
Clone 22) were metabolically labelled with
D -[U- 14 C]galactose for 24 h Lipid extracts
were purified, chromatographed on an
HPTLC plate and visualized as indicated in
Experimental procedures The positions of
co-chromatographed radioactive glycolipid
standards are indicated (St) GM1 was also
co-chromatographed and visualized by
exposing the plate to iodine vapour The
position of GM1 is indicated on the right of
the plate Lipids migrated as multiple bands
on the HPTLC plate because of the
hetero-geneity of the fatty acyl chains of the
mole-cules (C) Wild-type MDCK cells (MDCK wt)
and cells from clones 13 (MDCK Clone 13)
and 22 (MDCK Clone 22) were labelled for
GD3 with R24 antibody (green) or for GM1
with Alexa 555 -CTx b subunit (red) The
image contrast in wild-type MDCK cells
was reduced to show the presence of cells.
Single confocal sections were taken
every 0.7 lm parallel to the coverslip.
Scale bars ¼ 10 lm.
Trang 4We next attempted to characterize the expression and
subcellular localization of GD3 and GM1,
representa-tive gangliosides from the ‘a’ and ‘b’ series, for whose
identification we have valuable and highly sensitive
research methods As shown in Fig 1C, GD3
immunostaining with the mouse monoclonal antibody
R24 (IgG3) [30] was typical of a plasma membrane
con-stituent with a patchy distribution, as also observed in
other cell lines [10] A minor fraction was also observed
in internal membranes The expression of GD3 was
below the limit of detection in wild-type MDCK cells,
which only express GM3, thus confirming the specificity
of the antibody The expression of GM1 was evaluated
by staining with the Alexa555-CTx b subunit, a protein that binds specifically and with a high affinity (kd= 5· 10)12m) to the monosialoganglioside [31,32] Interestingly, using this approach, we were able to detect the expression of GM1 in GalNAc-T-expressing MDCK cells (Fig 1C) Thus, like GD3 expression, GM1 was predominantly expressed at the plasma membrane and
no expression of GM1 was observed in wild-type MDCK cells It should be mentioned that no apprecia-ble morphological alterations were observed in MDCK cells ectopically expressing either Sial-T2 or GalNAc-T
A
Fig 2 GM1 and GD3 gangliosides are apically localized in polarized MDCK cells (A) The steady-state distribution of GM1 or GD3 in polar-ized MDCK cells was determined using immunofluorescence confocal microscopy Polarpolar-ized MDCK cells stably expressing GalNAc-T-myc (left panels) were fixed, permeabilized and incubated with CTx to label GM1 (CTx, green) and PI to label cell nuclei (PI, red), or with CTx to label GM1 (CTx, green) and fluorescent DiI to label plasma membrane (DiI, red) Polarized MDCK cells stably expressing Sial-T2-HA (right panels) were fixed, permeabilized and incubated with R24 antibody to label GD3 (R24, green) and PI to label cell nuclei (PI, red) or with R24 (R24, green) and DiI (DiI, red) to label plasma membrane Cells were treated using laser scanning confocal microscopy with serial confocal sections (xy, 0.2 lm) collected from the top to the bottom of the cell monolayer Then, xz sections were displayed using the ortho mode in
LSM5 PASCAL software The bottom rows in each set of images show merged images from CTx and R24 (green) with PI or DiI (red) The top (T) and bottom (B) of the cell monolayer are indicated Scale bars ¼ 10 lm (B) Domain-selective surface labelling of ganglioside in MDCK cells GM1-expressing (left) and GD3-expressing (right) MDCK cells were grown on transmembrane filters to acquire polarity Cells were then chilled and CTx or R24 was added to either the apical or basolateral chamber at 4 C for 60 min Next, cells from each filter were pro-cessed as indicated in Experimental procedures, and the amounts of CTx and R24 antibody bound to the apical or basolateral surface were analysed by western blotting The expression of tubulin in the same membrane was analysed as a control of protein loading (Tub) (C) The relative contribution of bands in each condition was calculated using the computer software SCION IMAGE on the scanned film shown in (B).
Trang 5GD3 and GM1 gangliosides are predominantly
expressed at the apical surface of polarized
MDCK cells
Gangliosides are synthesized in the lumen of the Golgi
apparatus by a complex system of membrane-bound
glycolipid glycosyltransferases After synthesis,
ganglio-sides leave the Golgi apparatus via the lumenal surface
of transport vesicles, which are mainly targeted to the
plasma membrane [9] More specifically, in polarized
epithelial cells, gangliosides can be transported
selec-tively to apical or basolateral surfaces, or can be
homogeneously distributed in both specialized plasma
membrane domains To investigate the steady-state
dis-tribution of GD3 and GM1 gangliosides in a polarized
monolayer of MDCK clones, cells were grown to
con-fluence and the expression of glycolipids was analysed
by immunocytochemistry and fluorescent confocal
microscopy MDCK cells ectopically expressing either
Sial-T2 or GalNAc-T manifested no noticeable
morphological disruption of their epithelial
organiza-tion and no altered transepithelial electrical resistance
when grown on filter cultures (data not shown)
Inter-estingly, both the disialoganglioside GD3 and the
monosialoganglioside GM1 were mainly localized to
the apical domain of MDCK cells (Fig 2A, see top of
the cell monolayer) To further confirm and quantify
the steady-state distribution of GD3 and GM1
gangliosides in apical and basolateral domains of
MDCK cells, we used a domain-selective labelling
assay in which polarized MDCK cells grown on filter
cultures were selectively labelled from either surface
with the antibody R24 and CTx, specific for GD3 and
GM1, respectively The binding of R24 antibody and
CTx in both apical and basolateral domains was
evalu-ated by western blotting In agreement with
immuno-fluorescence labelling, we found that more than 90%
of GM1 was selectively expressed at the apical surface
(Fig 2B,C) Approximately 70% of GD3 ganglioside
was found to be expressed apically (Fig 2B,C),
compa-rable with the percentage observed for a
glycosylphos-phatidylinositol (GPI)-anchored protein used as apical
marker in MDCK cells [16] Together, these results
indicate that, after synthesis at the Golgi complex,
both GD3 and GM1 gangliosides are preferentially
transported to the apical surface of polarized MDCK
cells
GD3 and GM1 gangliosides are rapidly
endocytosed in MDCK cells
It has been observed in different cell types that
gangliosides can undergo endocytosis after arrival at
the plasma membrane [8,33] It has also been observed
in MDCK cells that apically delivered cargos, such as the raft-associated HA of influenza virus, are very poorly internalized [34] To specifically explore whether plasma membrane-expressed GD3 and GM1 ganglio-sides undergo endocytosis in MDCK cells, we used an antibody- and CTx b subunit-binding technique to track the fate of GD3 and GM1, respectively, after their internalization Briefly, subconfluent GD3- and GM1-expressing MDCK cells were incubated on ice for 10 min to inhibit intracellular transport, and then with R24 or the Alexa555-CTx b subunit on ice for
45 min Afterwards, cells were washed extensively with cold buffer in order to remove unbound antibody and toxin, and the temperature was changed to 37C in order to restore transport and thereby allow the endo-cytosis of GD3 and GM1 for different times Confocal microscopy analysis revealed that shortly (15 min) after allowing endocytosis by shifting the temperature
to 37C, GD3 and GM1 were found in vesicles all around the cytoplasm (Fig 3A) Almost the same sub-cellular distribution for GD3 was observed at 30 and
45 min after shifting the temperature to 37C, associ-ated with a noticeable decrease in the fluorescence intensity at later times In contrast, the GM1-CTx
b subunit began to acquire a perinuclear distribution
at 15 min After 30 min at 37C, the intracellular pool
of GM1 became more concentrated in the perinuclear region and the plasma membrane mark had almost disappeared (Fig 3A) By double immunofluorescence,
we demonstrated that the perinuclear region stained with the GM1-CTx b subunit colocalized with the area
of staining of GM130 and GalNAc-T-YFP, markers
of the Golgi apparatus [27,35] (Fig 3B) However, no colocalization was observed between endocytosed GD3 and GalNAc-T-YFP at any given time (Fig 3B) Endocytosis of R24 and the Alexa555-CTx b subunit seems to be specifically mediated by GD3 and GM1 gangliosides, as wild-type MDCK cells, which only express GM3 ganglioside, did not bind and internalize the two ligands (results not shown) In addition, endo-cytosis of apically expressed GD3 was observed in fully polarized MDCK cells (Fig S1)
To further characterize the intracellular structures decorated by GD3 and GM1 gangliosides, we performed colocalization experiments with markers of apical and basolateral early endosome (GTPase Rab5-GFP; GFP, green fluorescent protein), apical recycling endosome (GTPase Rab11a-GFP) and common recy-cling endosome [labelled with endocytosed Alexa647 -transferrin (Tf)] As shown in Fig 3C, internalized GD3 is detected in apical early endosome, partially detected in apical recycling endosome, but not detected
Trang 6A
B
C
GM1
Fig 3 GD3 and GM1-CTx are rapidly and specifically endocytosed in MDCK cells (A) MDCK cells stably expressing GD3 (GD3, top panels)
or GM1 (GM1, bottom panels) were incubated with R24 or Alexa 555 -CTx b subunit, respectively, at 4 C The temperature was then shifted
to 37 C to allow endocytosis of GD3-R24 and GM1-CTx, and cells were fixed at 0, 15, 30 or 45 min and processed for immunostaining Sin-gle confocal sections (xy) of 0.7 lm were taken parallel to the coverslip Scale bars ¼ 10 lm (B) Left panel: MDCK cells stably expressing GD3 and transiently expressing GalNAc-T-YFP, a Golgi apparatus marker (pseudo-coloured green), were allowed to internalize R24 antibody for 30 min Then, R24 antibody was visualized using Alexa546-conjugated goat anti-mouse IgG (red) Middle panel: MDCK cells stably expressing GM1 and transiently expressing GalNAc-T-YFP (pseudo-coloured green) were allowed to internalize Alexa 555 -CTx b subunit for
30 min (red) Right panel: cells were processed as shown in the middle panel, except that they were immunostained with antibody to GM130, another Golgi apparatus marker (green) (C) MDCK cells stably expressing GD3 (top panels) or GM1 (bottom panels) were allowed
to internalize R24 antibody (red) or Alexa 555 -CTx b subunit (red), respectively, for 15 min (left panels) or 45 min (middle and right panels) As indicated, the cells were transiently transfected to express wild-type Rab5-GFP (left panels, green) or wild-type Rab11a-GFP (middle panels, green) or allowed to internalize Alexa647-Tf (Tf-Alexa647, pseudo-coloured green) Insets in the merged panels show the details at higher mag-nification Single confocal sections (xy) of 0.7 lm were taken parallel to the coverslip The expression of fusion proteins and endocytosed Tf-Alexa 647 was analysed by the intrinsic fluorescence Scale bars ¼ 10 lm.
Trang 7in common recycling endosome at any given time Like
GD3, GM1 was also observed to colocalize with
wild-type Rab5 and Rab11a (Fig 3C), as well as with
common recycling endosome (Fig 3C) and the Golgi
apparatus (Fig 3B) It is probable that, in analogy
with the results obtained in other cells [31,32,36],
inter-nalized CTx can traffic to apical early endosome from
where it is differentially sorted A fraction is recycled
back to the plasma membrane via the apical recycling
endosome (Rab11a positive); the remainder is sorted
to the Golgi apparatus probably via the common
recycling endosome
The association of internalized GD3-R24 with the
apical early endosome and apical recycling endosome
in MDCK cells is compatible with its probable
recy-cling back to the plasma membrane, as recently
observed in nonpolarized epithelial CHO-K1 cells [33]
To explore this hypothesis, GD3-expressing MDCK
cells were incubated on ice for 10 min to inhibit
intra-cellular transport, and then with R24 antibody at 4C
for 60 min Afterwards, cells were allowed to
interna-lize the antibody for 30 min at 37C, and then the
temperature was shifted again to 4C The cell surface
was then stripped of any remaining antibody with an
acid wash (0 min) Finally, the cells only contained
R24 antibody in intracellular compartments
Subse-quently, prewarmed culture medium was added to the
cells and they were maintained at 37C to restore intracellular transport Cells and culture medium were recovered at different times, and the presence of the R24 antibody in both samples was analysed by western blotting As shown in Fig 4, at the beginning of the time-course experiment (stripped cells, 0 min), the anti-body was present only in the cell fraction At 15 min, the antibody was detected in both fractions (cells and culture medium), and, at 60 min, most of the R24 anti-body was recovered from the culture medium The antibody recovered from the culture medium was found to have the expected molecular mass (whole molecule) in gels run under nonreducing conditions Together, these results indicate that GD3-R24 anti-body, once internalized in MDCK cells, is recycled back to the plasma membrane and released to the culture medium
Clathrin is not required for the internalization of GD3 and GM1 gangliosides in MDCK cells Previously, we have described, in nonpolarized epithe-lial CHO-K1 cells, that endocytic recycling of GD3 is sensitive to Brefeldin A [33], consistent with the requirement of clathrin-coated vesicles for efficient GD3 recycling To further explore the early endocytic process involved in the internalization of GD3 in
A
B
Fig 4 R24 antibody is recycled back to the plasma membrane and released to the culture medium (A) GD3-expressing MDCK cells were incubated with R24 antibody for 60 min on ice Afterwards, cells were allowed to internalize the antibody for 30 min at 37 C, and the tem-perature was then shifted again to 4 C The cell surface was then stripped of any remaining antibody with an acid wash (0 min) The cells were then incubated at 37 C to restore intracellular transport, and cells and the culture medium were recovered at 15, 30 and 60 min The presence of R24 antibody in both samples was analysed by western blotting under nonreducing conditions, as indicated in Experimental pro-cedures (B) The relative contribution of the bands in each condition was calculated using the computer software SCION IMAGE on the scanned film shown in (A) Tubulin expression was used to normalize the level of proteins seeded in each lane The band intensity for R24 antibody
at 0 min (cellular fraction) was arbitrarily taken as unity The results are representative of three independent experiments.
Trang 8MDCK cells, we used a dominant-negative mutant of
epidermal growth factor receptor pathway substrate
clone 15 (Eps15), which selectively affects
clathrin-mediated endocytosis Eps15 is an established
compo-nent of clathrin-coated pits that is ubiquitously and
constitutively associated with adaptor proteins; the
expression of dominant-negative Eps15 selectively
inhibits clathrin-mediated endocytosis [37] First, we
demonstrated that the expression of dominant-negative
Eps15 in subconfluent GD3-expressing MDCK cells
inhibits the internalization of Alexa647-Tf, which is
known to be a clathrin-dependent process [13,38]
(Fig 5A) Next, subconfluent GD3-expressing MDCK
cells transiently expressing dominant-negative Eps15 were incubated on ice with the R24 antibody for
45 min The cells were then washed to remove unbound antibody, prewarmed culture medium was added and the cells were transferred to 37C to allow endocytosis for 30 min The results shown in Fig 5B indicate that the expression of dominant-negative Eps15 did not affect GD3 internalization, as the frac-tions of internalized and accumulated antibody at
30 min were similar in both control and dominant-negative Eps15-expressing cells This results indicate that clathrin-coated vesicles do not participate in the endocytic process of GD3 in MDCK cells
A
B
C
Fig 5 GD3 and GM1-CTx are internalized in MDCK cells by clathrin-independent endocytosis (A) GD3-expressing MDCK cells were tran-siently transfected to express the dominant-negative form of Eps15-GFP (DN-Eps15-GFP, green) After 24 h, uptake of Alexa 647 -Tf (Tf-Alexa 647 , pseudo-coloured red) was monitored for 20 min (B) GD3-expressing MDCK cells were transiently transfected to express the dominant-negative form of Eps15-GFP (DN-Eps15-GFP, green) After 24 h, the cells were incubated with R24 antibody to GD3 at 4 C, and then allowed to internalize the antibody for 30 min by shifting the temperature to 37 C Cells were fixed and the presence of R24 was anal-ysed using Alexa 546 -conjugated goat anti-mouse IgG (GD3, red) (C) GM1-expressing MDCK cells were transiently transfected to express the dominant-negative form of Eps15-GFP (DN-Eps15-GFP, green) After 24 h, cells were incubated with the Alexa555-CTx b subunit at 4 C, and then allowed to internalize the complex GM1-TCx for 15 min by shifting the temperature to 37 C Cells were fixed and the presence of CTx was analysed by confocal microscopy (GM1, red) Single confocal sections (xy) of 0.7 lm were taken parallel to the coverslip Arrows indicate DN-Eps15-GFP-transfected cells The expression of DN-Eps15-GFP and endocytosed Tf-Alexa647 was analysed by the intrinsic fluorescence Right panels show merged images from the two acquired channels Scale bars ¼ 10 lm.
Trang 9It has been demonstrated that GM1-CTx can exploit
three different internalization pathways en route to the
Golgi apparatus and endoplasmic reticulum [39]
GM1-CTx can be internalized by a nonclathrin-mediated,
noncaveolar, cholesterol-sensitive pathway or by
clath-rin-mediated endocytosis or by caveolae Following
essentially the experimental protocol described above
for the internalization of GD3, we observed that the
GM1-Alexa555-CTx b subunit is internalized in
domi-nant-negative Eps15-expressing MDCK cells (Fig 5C)
Thus, these results suggest that GM1-CTx
internaliza-tion in MDCK cells occurs through a
clathrin-indepen-dent process, or that clathrin-mediated internalization
is a relatively minor pathway in these epithelial cells
Discussion
We have developed a panel of cell lines derived from
MDCK cells whose glycolipid composition has been
genetically modified by the transfection of key
ganglio-side glycosyltransferases MDCK cells express GM3 as
the sole ganglioside species at the cell surface
Trans-fection with GalNAc-T or Sial-T2, two enzymes that
act on GM3 at the branching point of the synthesis
pathway, directs the flow of GM3 to either GM2 and
series ‘a’ gangliosides [40] or to GD3 (series ‘b’) and
GT3 (series ‘c’) [9], respectively In addition, the
expression of GalNAc-T resulted in the synthesis of
GA2, a neutral glycolipid belonging to series ‘o’, by
catalysing the transfer of N-acetyl-d-galactosamine to
the galactose residue of LacCer As observed
previ-ously in other cell lines [9,41], it is expected that the
synthesis in the Golgi apparatus, intracellular transport
and topological disposition of the new ganglioside
species in the plasma membrane in transfected MDCK
cells will follow the physiological mechanisms
estab-lished in the parental cells
In this report, we have focused on the examination
of the subcellular distribution and intracellular
trans-port of GD3 and GM1, representative complex
gangliosides from series ‘a’ and ‘b’, respectively Using
biochemical and fluorescent confocal microscopy, we
have demonstrated that GD3 and GM1 gangliosides,
after synthesis at the Golgi apparatus, are mainly
delivered to the apical plasma membrane of polarized
MDCK cells It is known that, for proteins, apical
sorting signals are localized in exoplasmic, membrane
or cytoplasmic domains, and comprise moieties as
divergent as lipids (GPI), sugars (N- and O-glycans)
and peptide motifs located at both the transmembrane
and cytoplasmic domains [13] Much evidence has
demonstrated that some of these apical sorting motifs
mediate the incorporation of proteins into lipid rafts,
membrane microdomains that are enriched in choles-terol and sphingolipids, a hypothesis put forward by van Meer and Simons [19] According to the lipid raft hypothesis, lipids rafts and their associated proteins form sorting platforms at the TGN that are incorpo-rated into apical transport intermediates and delivered
to the apical surface More recently, it has been reported that protein oligomerization modulates raft partitioning and apical sorting of GPI-anchored proteins [42]
Do GD3 and GM1 gangliosides associate with lipid rafts for trafficking to the apical surface? It is accepted that some biochemical properties associated with polar-ized pathways are preserved in fibroblast cell lines [43]
In this sense, we found that newly synthesized ganglio-sides, including GD3 and GM1, did not partition into lipid raft domains in the Golgi apparatus of nonpolar-ized CHO-K1 cells [9,10] Nevertheless, the presence of GD3 in lipid rafts was also evaluated in MDCK cells according to the usual criterion of insolubility in non-ionic detergent at 4C We observed that membrane-expressed GD3 was about 70% soluble to detergent extraction (Fig S2) Comparatively, GM3, an endoge-nously synthesized ganglioside enriched in the basolat-eral domain of MDCK cells [44], partitioned in equal percentages between the Triton-X100-soluble and Tri-ton-X100-insoluble fractions in both wild-type and GD3-expressing MDCK cells Thus, the partitioning of GD3, and probably GM1, into lipid rafts does not account for the polarized distribution of this glycolipid
in MDCK cells Further work using biophysical tech-niques, which measure lipid raft association more accu-rately than does detergent insolubility, is required However, progress in this area is slow because of tech-nical difficulties with available fluorescent lipids Short-chain and fluorescent lipid analogues have been employed as useful tools to delineate the potential pathways and mechanisms of intracellular transport and sorting [45] Nevertheless, quantitative comparison with natural lipids remains to be determined, as it has been reported previously that the quantitative and qualitative behaviour of analogous lipids is quite differ-ent from that of long-chain cellular lipids [46]
In the current study, we have shown that GD3 ganglioside, in association with a specific antibody, is actively endocytosed in subconfluent and polarized MDCK cells Endocytosed GD3 was detected in both Rab5-positive apical early endosome and Rab11a-posi-tive apical recycling endosome, which is compatible with its recycling back to the plasma membrane, as recently observed in nonpolarized CHO-K1 cells [33] Indeed, this hypothesis was confirmed in MDCK cells GD3-R24 antibody was rapidly endocytosed, recycled
Trang 10back to the plasma membrane and released to the
cul-ture medium At this point, we cannot discard the
recycling of the GD3-R24 complex to the plasma
membrane from Rab5-positive apical early endosome
through the direct return pathway Interestingly, the
endocytosis of GD3 in MDCK cells was found to be a
clathrin-independent process, in line with previous
results showing the caveolae-mediated endocytosis of
glycosphingolipid analogues [47] However, the
mecha-nism(s) of apical endocytosis of GD3 ganglioside in
MDCK cells remains to be investigated, as caveolae
are formed only at the basolateral surface [13] It
should be mentioned that sorting and recycling
endo-some compartments have an acidic pH (6–6.5), which
could dissociate the ganglioside–antibody complex
However, we demonstrated that the association of
antibody R24 with GD3 ganglioside was only slightly
affected, even after 1 h at pH 6 or 7 (Fig S3),
suggest-ing that the itinerary of the R24 antibody reflects the
intracellular transit of the complex
It has been demonstrated in different cell types that
CTx can be internalized by clathrin-, caveolar- and
CDC42-dependent pathways en route to the
endoplas-mic reticulum [47,48] In the present study, we
observed that GM1-CTx can be efficiently endocytosed
in dominant-negative Eps15-expressing MDCK cells,
suggesting that GM1-CTx internalization occurs
through a clathrin-independent process, or that
clath-rin-mediated internalization is a relatively minor
path-way in these epithelial cells After GM1-CTx
internalization, we essentially observed that it follows
the intracellular itinerary described for other cell types
(plasma membrane fi endosomes fi Golgi
appara-tus) However, in MDCK cells, we observed that the
binding of CTx mainly occurs in the apical domains,
whereas, in polarized CaCo-2 cells, the endocytosis of
CTx is essentially the same at both apical and
basolat-eral surfaces [49] As also observed for the
GD3–anti-body complex, the association of CTx with the
monosialoganglioside GM1 was not affected at pH 6
(Fig S3), suggesting that the intracellular trafficking of
CTx through acidic compartments occurs in
associa-tion with GM1 Moreover, it has been suggested that
the monosialoganglioside is the vehicle that directs the
toxin to its final destination [31,48]
In conclusion, our results demonstrate, for the first
time, that complex gangliosides are apically sorted in
polarized MDCK cells, and that GD3 and GM1
gan-gliosides are internalized by clathrin-independent
endo-cytosis to follow different intracellular destinations A
potential function of ganglioside recycling involves the
selective endocytosis of particular cell surface
compo-nents during the generation and maintenance of
mem-brane polarity A further potential role involves the regulation of signal transduction processes Previous work has reported that cell surface receptors and sig-nalling molecules are recruited into specialized mem-brane domains enriched in glycosphingolipids, and that gangliosides modify the biological effects of sev-eral trophic factors [1,4,50] Consequently, the internal-ization of these domains may well play an important role in signalling events
GD3 and GM1 internalization and trafficking in MDCK cell lines were monitored using a specific anti-body to GD3 and CTx, respectively Therefore, the results described in this work provide the basis to gain further insight into the molecular mechanisms that operate in the intracellular trafficking and pathological effects of bacterial toxins and antibodies to ganglio-sides, which are associated with autoimmune disorders [51,52] In addition, the genetically modified MDCK cell lines expressing both complex gangliosides and neutral sphingolipids should provide excellent model systems in which to study the synthesis, trafficking and polarity of glycolipids in epithelial cells
Experimental procedures
Expression plasmids The expression plasmid for SialT-2-HA (carboxy-terminal epitope-tagged chicken SialT-2 with the nanopeptide epi-tope of viral HA; pCEFL-SialT-2-HA) has been described previously by Daniotti et al [53] GalNAc-T-myc (3¢ end of the human GalNAc-T cDNA epitope tagged with human c-myc; pCIneo-GalNAc-T-myc) has been described previ-ously by Giraudo et al [54] The GTPase Rab11a-GFP wild-type construct was kindly provided by M Colombo (Universidad Nacional de Cuyo, Mendoza, Argentina); the plasmid coding for Rab5-GFP was supplied by J Bonifa-cino (NICHD, National Institutes of Health, Bethesda,
MD, USA) The construct containing the cDNA coding for the N-terminal domain (cytosolic tail, transmembrane domain and few amino acids of the stem region) of Gal-NAc-T fused to the N-terminus of YFP (GalGal-NAc-T-YFP) was obtained by subcloning the corresponding cDNA frag-ments into the plasmid pEYFP-N1 (Clontech, Mountain View, CA, USA) [55]
Cell culture and generation of stable MDCK cell lines
MDCK II cells (generously provided by A Gonza´lez, Pontificia Universidad Cato´lica de Chile, Chile) were main-tained in DMEM (Gibco BRL, Carlsbad, CA, USA) sup-plemented with 7.5% fetal bovine serum (HyClone, Logan,