In vivo incorporation of an azide-labeled sugar analog to detect mammalian
glycosylphosphatidylinositol molecules isolated from the cell surface
Saulius Vainauskas, Leslie K. Cortes, Christopher H. Taron⇑
New England Biolabs, Inc., 240 County Road, Ipswich, MA 01938, USA
a r t i c l ei n f o
Received 19 July 2012
Received in revised form 11 September
Accepted 13 September 2012
Available online 23 September 2012
GPI side branch modification
Polarized epithelial cells
a b s t r a c t
N-Acetylgalactosamine (GalNAc) linked to the first mannose of glycosylphosphatidylinositol (GPI) core
has been previously reported to be heterogeneously present on some mammalian GPI-anchored proteins.
Here we present a method for profiling GalNAc-containing GPI-anchored proteins in mammalian cells by
metabolic labeling with tetraacetylated N-azidoacetylgalactosamine (GalNAz) followed by biotinylation
of the incorporated sugar analog. We have labeled both endogenous and recombinant GPI-anchored pro-
teins with GalNAz, and demonstrated that the azide-activated sugar gets incorporated into the GPI gly-
can, likely as an unsubstituted side branch of the core structure. GalNAz was detected only on GPI
molecules attached to proteins, and not on GPI precursors, indicating that GalNAc modification takes
place after the GPI anchor is transferred to protein. We have highlighted the utility of this cell labeling
approach by demonstrating the ability to examine specific GalNAc-containing GPI-anchored proteins iso-
lated non-destructively from separate membrane domains (apical and basolateral) in polarized epithelial
cells. This study represents the first demonstration of site-specific in vivo labeling of a GPI moiety with a
synthetic sugar analog.
? 2012 Elsevier Ltd. All rights reserved.
The glycosylphosphatidylinositol (GPI) anchor is an evolution-
arily conserved glycolipid structure that is post-translationally at-
tached to certain eukaryotic secretory proteins to anchor them to
the cell surface. The precise structure of the GPI has been deter-
mined for numerous mammalian, protozoan, yeast, and plant
GPI-anchored proteins (GPI-APs).1,2Comparison of these structures
has revealed a conserved core composed of ethanolamine-P-
Mana1–2Mana1–6Mana1–4GlcN-myo-inositol-lipid that is pre-
sumed to be common to all GPIs. This core structure is synthesized
and transferred to protein in the endoplasmic reticulum, and many
of the enzymes involved in this process have been identified.1,3–6
Overall GPI structure differs considerably between species and
even within an individual organism due to both a diverse and het-
erogeneous array of glycan moieties that can be appended as side-
branches to the core glycan, and variations in the lipid composition
of the phosphatidylinositol.1–3Furthermore, within any given
organism, GPI structure is highly dynamic with each GPI molecule
undergoing remodeling of both the glycan and lipid domains be-
fore and after its attachment to protein.7–9
Understanding GPI structure as it relates to potential functions
of the GPI molecule represents a major challenge. To date, the only
confirmed function of the GPI moiety is as a membrane anchor for
certain proteins, however the GPI has been implicated in several
important cellular processes such as signal transduction, apical tar-
geting in polarized epithelial cells, and cell–cell communication
(see Ref. 2 for a review). The dynamic structural complexity of GPIs
suggests that elements of GPI structure could play a key role in
these important processes, as well as in the biosynthesis and traf-
ficking of GPI-APs. Indeed, roles of certain moieties found as side-
branches to the conserved core of the mammalian GPI glycan are
starting to be elucidated. The presence of phosphoethanolamine
(EtNP) side-branched to the first GPI mannose (Man-1) is a prere-
quisite for the association of a GPI precursor with the human GPI
transamidase complex, the enzyme that catalyzes the attachment
of GPI to protein.10Phosphoethanolamine linked to the second
GPI mannose (Man-2) residue is involved in regulation of mamma-
lian GPI-AP transport from the ER to the Golgi.9More recently, it
has been proposed that the synapse damage induced upon prion
0008-6215/$ - see front matter ? 2012 Elsevier Ltd. All rights reserved.
Abbreviations: GPI, glycosylphosphatidylinositol; GPI-APs, GPI-anchored pro-
teins; EtNP, phosphoethanolamine; Man-1, first GPI mannose; Man-2, second GPI
mannose; Man-3, third GPI mannose; Man-4, fourth GPI mannose; Man3-GPIs,
trimannosyl GPIs; Man4-GPIs, tetramannosyl GPIs; GalNAc, N-acetylgalactosamine;
GlcNAc, N-acetylglucosamine; GalNAz, N-azidoacetylgalactosamine; TEER, trans
epithelial electrical resistance; Gluc, recombinant Gaussia princeps luciferase; Cluc,
recombinant Cypridina noctiluca luciferase; PI-PLC, phosphatidylinositol-specific
phospholipase C; UDP, uridine diphosphatidyl; PDM, protein deglycosylation mix;
b-HexNAc-ase, jack bean b-N-acetylhexosaminidase; HF, aqueous hydrofluoric acid;
PI, Phosphatidylinositol; DMEM, Dulbecco’s modified Eagle’s medium; FBS, fetal
bovine serum; IP, immunoprecipitated.
⇑Corresponding author. Tel.: +1 978 380 7207; fax: +1 978 921 1350.
E-mail address: email@example.com (C.H. Taron).
Carbohydrate Research 362 (2012) 62–69
Contents lists available at SciVerse ScienceDirect
journal homepage: www.elsevier.com/locate/carres
protein clustering is mediated in part by the sialic acid-containing
side branch of its GPI anchor.11
The significance of other moieties side-branched to the GPI gly-
can is unknown. For example, a fourth mannose (Man-4) may be
added as a a1,2-linked side-branch of the third mannose (Man-3)
on the GPI anchors of Trypanosoma cruzi, yeast, slime mold, and
mammalian proteins. Addition of Man-4 is essential for GPI biosyn-
thesis in yeast, whereas mammalian cells do not have this strict
requirement and both Man3- and Man4-GPIs are transferred to pro-
teins.12Interestingly, in mammalian cells the fourth GPI mannosyl-
transferase (PIG-Z) is expressed in a tissue-specific manner,
suggesting that formation of tetramannosyl GPIs (Man4-GPIs) is
highly regulated on a tissue- or cell-specific level.12In mammalian
cells, N-acetylgalactosamine (GalNAc) linked to Man-1 by a b1–4
linkage has been reported to be present on certain GPI-anchored
proteins.13–17These studies involved the analysis of large amounts
of purified protein through enzymatic and chemical digestion,
thin-layer chromatography, gas- and mass spectrometry, and
NMR.13,15–18Moreover, it has been demonstrated that the GalNAc
addition to the GPI anchor is heterogeneous on an individual pro-
tein,13,15–18which makes their detection by conventional methods
even more challenging. In some GPIs, GalNAc may be terminally
capped with galactose and sialic acid.17,19The relative abundance
of side-chain modifications on GPI anchors of mammalian proteins,
the biological significance of the GalNAc modification, and the sub-
cellular localization and mechanism of GalNAc addition are not
known. The study of this modification is impeded in part by the
lack of a simple and effective method to visualize GalNAc-contain-
ing GPIs without the requirement for purification of large amounts
In this study, we investigated the in vivo incorporation of a syn-
thetic analog of N-acetylgalactosamine into the GPI glycan to visu-
alize the GalNAc side-branch modification known to be present on
some GPI-APs. Tetraacetylated N-azidoacetylgalactosamine (Gal-
NAz) has been utilized in previous studies to label different classes
of glycans in living cells, including N- and O-glycans.20–24This su-
gar analog is able to cross the cell membrane in living cells and is
converted by the endogenous cellular machinery into its cognate
nucleotide sugar (UDP-GalNAz) donor prior to incorporation into
glycans.23It has also been reported that UDP-GalNAz can be epi-
merized into UDP-GlcNAz prior to integration into glycans in
mammalian cells.28The work presented here demonstrates that
GalNAz can be incorporated specifically into the GPI glycan. Using
metabolic labeling of GPIs with GalNAz, we demonstrate the pres-
ence of the GalNAc side-chain modification on GPI-APs, but not GPI
precursors. The utility of this method is highlighted by the demon-
stration of the ability to isolate GalNAc modified GPI anchors from
different plasma membrane domains in intact polarized epithelial
cells. Our findings provide the framework for the future study of
the synthesis and function of GalNAc-containing side-branches of
2. Results and discussion
2.1. A model system for sugar analog incorporation into GPIs
Metabolic incorporation of tetraacetylated N-azidoacetylgalac-
tosamine (GalNAz), a synthetic azide-labeled GalNAc analog, has
been used to label both N- and O-glycans in living cells.20–24GPI-
anchored proteins (GPI-APs), but not GPI precursors, have been re-
ported to heterogeneously contain a GalNAc side chain modifica-
tion, suggestingthatthis modification
attachment to protein. Therefore, we examined whether metabolic
GalNAz incorporation can also be used to label the GPI-glycan of
GPI-APs in living cells. Two important considerations for this
experimentation were the selection of a model GPI anchored pro-
tein and of a model cell line.
We sought a GPI-anchored protein that had been demonstrated
to be robustly expressed in cells and to contain GalNAc on the GPI
glycan. Furthermore, to minimize the challenges presented by the
incorporation of the GalNAc analog into N- and O-linked glycan
chains, we also sought a GPI-anchored model protein with a
well-characterized glycosylation pattern and preferably lacking
O-glycans. Based on these criteria, we chose CD73 (50-nucleotid-
ase) from a list of mammalian GPI-anchored proteins with charac-
terized GPI structures. CD73 is N-glycosylated, but does not have
O-linked glycans.25,26Additionally, the presence of HexNAc on
the GPI anchor of the bovine CD73 has been reported previously.18
Expression of CD73 was examined in HeLa cells and the human
retinal pigment epithelial cell line, ARPE-19, which can be grown
either as an unpolarized monolayer, or cultured in a polarized fash-
ion on membrane supports. These cell lines were selected based on
observations that the majority of the GPI anchors from these cells
detected by mass spectrometric analysis contain HexNAc attached
to the first mannose residue of the core glycan (Nakayasu, E.; Vain-
auskas, S.; Taron, C.; Almeida, I. unpublished results). Consistent
with previous reports,27we found that CD73 was highly expressed
in ARPE-19 cells, but was virtually undetectable in HeLa crude cell
lysates (Fig. 1A). Furthermore, CD73 could be efficiently released
from the ARPE-19 cell surface by treatment with phosphatidylino-
sitol-specific phospholipase C (PI-PLC) (Fig. 1B). Therefore, endog-
enously expressed CD73 from ARPE-19 cells was selected as a
model system in which to test GalNAz incorporation into GPI
2.2. GalNAz can be incorporated into the GPI glycan of CD73
To determine if GalNAz can be metabolically incorporated into
glycoproteins in APRE-19 cells, and specifically into the glycans
of CD73, non-polarized ARPE-19 cells were grown in culture med-
ium supplemented with GalNAz, after which GPI-anchored pro-
teins were released from the cell surface with PI-PLC. Putative
GalNAz-containing glycans were subsequently labeled with bio-
Figure 1. Expression of CD73 in different cell lines and metabolic labeling of ARPE-
19 with GalNAz. (A) Expression of CD73 in HeLa (lane 1) and ARPE-19 (lane 2) cells.
Whole cell lysates were analyzed by western blotting with anti-CD73 antibody.
Note the strong expression of CD73 in ARPE-19 cells. (B) CD73 can be released from
live ARPE-19 cells by treatment with PI-PLC. Cells were incubated in the absence
(lane 1) or presence (lane 2) of PI-PLC, and then collected medium was analyzed by
blotting with anti-CD73 antibody. (C) GalNAz is able to label total glycoproteins in
ARPE-19 cells. Cells were incubated in the growth medium with (lane 2) or without
(lane 1) 50 lM GalNAz. Cell lysates were prepared after 48 h, labeled with biotin–
PEG3–phosphine via the Staudinger ligation and analyzed by blotting with anti-
biotin-HRP antibody. (D) CD73 can be specifically labeled with GalNAz. ARPE-19
cells were incubated in the presence (lane 2) or absence (lane 1) of 50 lM GalNAz.
After 48 h, cells were lysed in RIPA buffer, CD73 was immunoprecipitated (IP) using
the CD73 antibody, and then ligated to biotin–PEG3–phosphine. Finally, the CD73 IP
samples were analyzed by SDS–PAGE and immunoblotting with anti-biotin and
S. Vainauskas et al./Carbohydrate Research 362 (2012) 62–69
tin–PEG3–phosphine via Staudinger ligation, and the biotinylated
protein was subjected to western blot analysis using an anti-biotin
antibody. Multiple proteins were labeled with GalNAz, suggesting
that azide-containing sugar was efficiently incorporated en masse
into multiple glycoproteins in ARPE-19 cells (Fig. 1C). For the anal-
ysis of CD73, PI-PLC released CD73 was affinity purified using an
anti-CD73 antibody prior to biotin labeling and western blot anal-
ysis. Biotin was only found associated with CD73 isolated from
cells that were exposed to GalNAz (Fig. 1D) confirming that GalNAz
was efficiently incorporated into its glycans.
It has been reported that mammalian cells can epimerize the
nucleotide sugar derived from GalNAz, uridine diphosphatidyl
(UDP)-GalNAz, to UDP-GlcNAz.28In our study, both GalNAz and
its epimerized form, UDP-GlcNAz, may be incorporated into the
N-linked glycans of CD73. Because CD73 contains two different
classes of appended carbohydrates (N-glycans and a GPI anchor),
we sought specific evidence that GalNAz was being incorporated
into the GPI moiety. We undertook two independent lines of exper-
imentation on non-polarized cells to make this determination: (i)
chemical and enzymatic treatment of glycans on CD73 and (ii)
analysis of an engineered GPI-less version of CD73.
2.2.1. Chemical and enzymatic removal of glycans on CD73
demonstrates GalNAz labeling on the GPI glycan
To demonstrate that GalNAz is incorporated into the CD73 GPI
glycan, we eliminated the biotin signal from azidosugars incorpo-
rated into non-GPI glycans by treating the PI-PLC-released and
immunoprecipitated protein samples with protein deglycosylation
mix (PDM; a commercial cocktail of enzymes including PNGase F,
O-glycosidase, neuraminidase, b-galactosidase and b-N-acetylglu-
cosaminidase) which removes both N- and O-linked glycans
(Fig. 2A). PDM-treated CD73 decreased in molecular weight, con-
sistent with the removal of N-glycans from the protein (Fig. 2B,
lane 2). However, CD73 was still detected by western blotting with
the anti-biotin antibody (Fig. 2B, lane 2), suggesting that biotinyl-
ated GalNAz remained associated with the protein, likely via the
GPI anchor. Further treatment of PDM-deglycosylated CD73 with
jack bean b-N-acetylhexosaminidase (b-HexNAc-ase) that liberates
terminal b-linked HexNAc (Fig. 2A) completely eliminated the anti-
biotin signal (Fig. 2B, lane 3). This indicates that the CD73-associ-
ated GalNAz is terminally exposed and is consistent with the no-
tion that it is side-branched to the GPI glycan.
Lastly, we used aqueous hydrofluoric acid (HF) to cleave the
phosphodiester bond and release the carbohydrate portion of the
GPI from CD73 (Fig. 2A).13PDM-treated CD73 was immobilized
by transfer onto PVDF membrane and treated with aqueous HF.
Western analysis indicated that the anti-biotin signal was almost
completely abolished following exposure of PDM-treated CD73 to
HF (Fig. 2C, lane 2, top panel), whereas the control anti-CD73 signal
showed little change following HF treatment (Fig. 2C, lane 2, lower
panel). This data supports the conclusion that the GalNAz signal
remaining after PDM treatment was associated with CD73 via a
phosphodiester linkage, likely via the GPI anchor.
2.2.2. An engineered GPI-less CD73 molecule illustrates GalNAz
labeling of the GPI anchor
Additional evidence that GalNAz was being incorporated into
the GPI glycan was provided through expression of an engineered
recombinant form of CD73 that lacked a GPI anchor. Two CD73
expression constructs were created that encoded: (i) a full length
CD73 having its native GPI anchor (termed CD73–GPI) and (ii) a
chimeric CD73 protein having its C-terminal GPI attachment signal
peptide replaced with the transmembrane domain of membrane
cofactor protein (MCP/CD46) (termed CD73–TMD).29The ex-
pressed CD73 proteins both correctly localized to the cell surface
as visualized by immunostaining on transiently transfected, non-
permeabilized HeLa cells (Fig. 3A), indicating that the GPI anchor
of CD73 can be replaced with a transmembrane domain without
detriment to the protein’s folding or trafficking. Next, the CD73–
GPI and CD73–TMD constructs were each individually expressed
in ARPE-19 cells and subjected to GalNAz labeling. The cells were
lysed with RIPA buffer, and the resulting immunoprecipitated
and deglycosylated CD73 proteins were labeled with biotin–
PEG3-phosphine and analyzed by Western blot with anti-CD73
and anti-biotin antibodies. The GPI-anchored forms of endogenous
and overexpressed CD73 (Fig. 3B) were detectable with both anti-
CD73 and anti-biotin antibodies. Cells transfected with the CD73–
TMD expression vector produced a slightly larger form of CD73 due
to the presence of the appended transmembrane domain (Fig. 3,
lane 3, lower panel). Importantly, this GPI-less form of CD73 did
not contain biotinylated GalNAz (Fig. 3, lane 3, top panel) clearly
demonstrating that GalNAz is incorporated into the GPI moiety.
While it is formally possible that GalNAz could also become
incorporated into the GPI anchor as GlcNAz due to the reported
epimerization of UDP-GalNAz to UDP-GlcNAz inside the cell,28
the evidence strongly suggests that it is GalNAz. Only two amino-
sugars have been described to be associated with the mammalian
GPI. The first is a GlcN residue that is an integral component of the
core GPI glycan. GlcN is first added to the phosphatidylinositol (PI)
as GlcNAc, after which its amino group is deacetylated prior to
addition of the first GPI mannose. Thus, it is likely that GlcNAz
incorporated at this position would either have its azide group
similarly removed (also resulting in GlcN) thus becoming unde-
tectable by Staudinger ligation with biotin–PEG3–phosphine, or
GlcNAz-PI would become a dead-end biosynthetic intermediate
Figure 2. Demonstration of GalNAz labeling on the GPI glycan through chemical
and enzymatic removal of glycans on CD73. (A) GPI anchor schematic, including
sites of cleavage and the site of synthetic sugar analog incorporation and
conjugation to biotin–PEG3–phosphine. (B) Hexosaminidase-mediated removal of
GalNAz to show incorporation into the GPI glycan. To verify GalNAz labeling within
the GPI glycan, immunoprecipitated CD73 was either non-treated (lane 1), treated
with PDM (lane 2), or treated with PDM followed by b-HexNAc-ase (lane 3) before
labeling with biotin–PEG3–phosphine. (C) Aqueous HF release of the GPI anchor
results in the loss of the GalNAz signal. GalNAz-biotin-labeled CD73 was analyzed
by SDS–PAGE and transferred onto PVDF membrane. Then the PVDF membrane
with CD73 was incubated in the absence (lane 1) or presence (lane 2) of aqueous HF,
followed by immunoblotting with anti-biotin and anti-CD73 antibodies. Glucosa-
mine (blue/white half shaded square), mannose (green circle), and N-azidoacetylga-
lactosamine (yellow square-N3). PDM = protein deglycosylation mix), HF = aqueous
hydrofluoric acid, PI-PLC = phosphatidylinositol-specific phospholipase C, b-Hex-
NAc-ase = b-N-acetylhexosaminidase.
S. Vainauskas et al./Carbohydrate Research 362 (2012) 62–69
that could not be further elongated or attached to proteins.30The
second aminosugar known to be present on mammalian GPIs is a
heterogeneously present GalNAc residue that is side-branched to
Man-1 in a b1-4 linkage.13,15,16In our experiment, if this side-
branching sugar was epimerized GlcNAz instead of GalNAz we
would have observed a complete loss of biotin signal upon treat-
ment with PDM mix alone, which contains a GlcNAc-specific b-
N-acetylglucosaminidase that would remove a terminally exposed,
side-branching GlcNAz. Instead, it was only upon treatment with
both PMD and jack bean b-N-acetylhexosaminidase that we were
able to achieve a nearly complete removal of the GalNAz–biotin
signal, suggesting that it is terminal GalNAz, not GlcNAz that is
incorporated into the GPI anchor.
Considered together, these data clearly indicate that GalNAz can
be incorporated into the GPI glycan in vivo in mammalian cells,
and that the sugar analog can be exploited to interrogate the pres-
ence or absence of GalNAc on individual GPI anchored proteins,
such as CD73.
2.3. GalNAz is incorporated into the GPI after attachment to
Thus far, we have examined the ability of cells to incorporate
GalNAz into the specific GPI-AP, CD73. However, while the GalNAc
modification has not been previously observed on mammalian GPI
precursors prior to their attachment to protein, the exact timing of
GalNAc addition is unknown. Interestingly, biosynthesis of GPI
intermediates containing GalNAc or Glc-GalNAc residues was pre-
viously reported in Toxoplasma gondii.31Therefore, we analyzed
whether complete mammalian GPI precursors can be labeled with
GalNAz prior to their attachment to protein.
HeLa cells were pulse-labeled with D-[2-3H] mannose in the
presence of GalNAz and then chased for 18 h with or without Gal-
NAz. Cell lysates were subjected to Staudinger ligation with biotin–
PEG3–phosphine. Glycolipids were then extracted from the lysates
and analyzed by thin layer chromatography (TLC). The mature GPI
precursor H8 was the major labeled GPI accumulating in mamma-
lian cells during metabolic radiolabeling, while H6 and H7 precur-
sors were also detected as minor GPI species (Fig. 4, lanes 1, 3). All
three detected precursors consist of a conserved core glycan
(Mana1–2Mana1–6Mana1–4GlcN) linked to the 6-position of
the D-myo-inositol ring of PI and differ only by a number of EtNP
residues (1 (H6), 2 (H7), and 3 (H8)) appended to the core glycan.
With increased incubation times, all of the H6 and H7 precursors
were converted in the cell to H8 (Fig. 4, lanes 2, 4). If GalNAz could
become incorporated into the GPI precursor, it would be expected
that the mobility of the GalNAz-modified, radiolabeled GPIs would
be altered relative to the non-GalNAz labeled GPIs due to the ap-
pended sugar and biotin tag. However, the mobility and relative
abundance of the complete GPI precursor H8 extracted from cells
grown in the presence or absence of GalNAz remained the same,
indicating that GalNAz was not incorporated into GPI precursors
(Fig. 4, lanes 2 and 4). Our results suggest that GalNAc modification
occurs on protein-bound GPIs, but not on biosynthetic GPI inter-
mediates or complete GPI precursors. This is in agreement with re-
sults from other studies where the GalNAc side-chain modification
in higher eukaryotes was detected only on GPI-APs.13,15–18Thus, it
is probable that side-branching GalNAc is added to the GPI post-
attachment to protein, most likely in the Golgi. Our method and
model system for observing GalNAz incorporation into CD73 thus
provide a needed framework for ongoing investigation into the
precise location and mechanism of GalNAc addition to the mam-
malian GPI glycan.
2.4. Detection of GalNAc-modified CD73 in polarized ARPE-19
An association of GPI-APs with detergent-resistant microdo-
mains in combination with their oligomerization32–34or N-glyco-
sylation35–37has been implicated in the preferential localization
of GPI-APs to specific plasma membrane domains in polarized
cells. It is not known whether precise GPI structures, or, more spe-
cifically, the presence or absence of side-branching substituents on
the GPI glycan, contribute to the polarized trafficking of GPI-APs. In
Figure 3. Biotinylated GalNAz is found on GPI-anchored CD73, but not transmembrane anchored CD73. (A) CD73–GPI and CD73–TMD correctly localized to the cell surface.
HeLa cells were transfected with pcDNA3.1, CD73–GPI, and CD73–TMD plasmid constructs. Cells were immunostained under non-permeabilizing conditions with anti-CD73
and AlexaFluor488 anti-mouse IgG antibodies. (B) CD73–GPI contained biotinylated GalNAz, while CD73–TMD did not, demonstrating the specific incorporation of GalNAz
into the GPI anchor. ARPE-19 cells were transfected with pcDNA3.1 (lane 1), CD73–GPI (lane 2), or CD73–TMD (lane 3) constructs and grown in the presence of GalNAz. Cells
were lysed 48 h later and the immunoprecipitated CD73 proteins were deglycosylated and labeled with biotin–PEG3–phosphine. The resulting CD73 IP samples were
analyzed by SDS–PAGE and immunoblotting with anti-biotin and anti-CD73 antibodies. Scale bar = 10 lm.
S. Vainauskas et al./Carbohydrate Research 362 (2012) 62–69
the present study, we utilized our GalNAz labeling method to con-
duct the first experiment aimed at comparing a structural feature
of a GPI derived from the same protein isolated from two different
membrane surfaces of polarized epithelial cells.
In several polarized epithelial cell lines, CD73 has been ob-
served primarily on the apical membrane domain, but with lesser
amounts also being present on the basolateral domain.27,38Using
confocal microscopy, we confirmed that in polarized ARPE-19 cells,
CD73 predominantly localized to the apical domain (Fig. 5A, aster-
isks). Some CD73 expression was also observed on the basolateral
domain as indicated by partial co-localization with b-catenin, a
marker of the basolateral compartment (Fig. 5A, arrows).39
To query the presence or absence of GalNAc on CD73 derived
from both the apical and basolateral faces of polarized ARPE-19
cells, the cells were labeled with GalNAz for 48 h and CD73 was
selectively released from each membrane by incubation with PI-
PLC in the top (apical) and bottom (basolateral) chambers of the
transwell filter. To ensure tight junction integrity was maintained
during PI-PLC treatment, we developed a novel method using a
highly sensitive luciferase-based assay as a means to monitor pro-
tein diffusion across the cell monolayer. Two luciferases of differ-
ent sizes (20 kDa Gaussia princeps luciferase and 75 kDa Cypridina
noctiluca luciferase) were added to the apical chamber concurrent
with PI-PLC, and the diffusion of each enzyme to the basolateral
chamber was measured over time (Fig. 4B). The experiments dem-
onstrated no increase in enzyme diffusion during the PI-PLC incu-
bation period, whereas a disruption of tight junctions (which are
critical to maintaining barrier integrity) with EDTA and Triton X-
10040led to a significant increase in epithelial permeability
(Fig. 5B). This indicates that the tight junctions in the ARPE-19
polarized monolayer were maintained throughout PI-PLC treat-
ment and ensures the integrity of our apical and basolateral PI-
Following PI-PLC release, CD73 was collected by immunopre-
cipitation, PDM-treated to remove N-glycans, and incorporated
GalNAz was labeled with biotin–PEG3–phosphine. GalNAz was de-
tected on both apically- and basolaterally-derived CD73 (Fig. 5C),
though the signal for both CD73 and biotin was significantly lower
on samples from the basolateral compartment, consistent with the
predominant apical localization of CD73. This observation suggests
that (i) GalNAc is added to the GPI glycan upstream of sorting of
apically- or basolaterally-bound GPI-APs and (ii) that the presence
or absence of side-branching GalNAc does not alone discriminate
apical versus basolateral targeting of GPI-APs in human ARPE-19
cells. Furthermore, this experiment provides an example of the
utility of sugar nucleotide analog incorporation into the GPI moiety
as a means to probe for specific glycan modifications on two differ-
ent surfaces of living cells.
In this study, we established a method to easily identify the
presence of a GalNAc residue side-branched to the GPI glycan moi-
ety of mammalian GPI anchored proteins. We utilized a metabolic
labeling approach with tetraacetylated N-azidoacetylgalactos-
amine (GalNAz), an unnatural analog of GalNAc that crosses the
cell membrane and is converted into UDP-GalNAz for incorpora-
tion into various cellular glycans. The incorporated glycan analog
can be further modified by chemical coupling to a variety of phos-
phine-containing compounds (biotin, peptide epitopes, fluoro-
phors) that permit its detection. We showed that mammalian
cells are capable of metabolically incorporating a side-branching
GalNAz in place of GalNAc in the GPI glycan moiety. In vivo label-
ing with GalNAz not only directly detects GPI-anchored proteins
containing a GalNAc-modified glycolipid, but can also be used for
specific enrichment of these GPI-APs through the immobilization
of GalNAz containing GPI proteins via the azide group.
This approach will be useful for probing the structures of low
abundance GPIs that are difficult to study using conventional
mass spectrometry and NMR based techniques. Furthermore, this
method may also aid in the identification of the glycosyltransfer-
ases responsible for the addition of the GalNAc side modification
in a heterogeneous manner. We have also exemplified the utility
of the method by showing that GalNAz is incorporated into the
GPI moiety of CD73 separately isolated from different membrane
domains of polarized epithelial cells. This highlights perhaps the
most interesting application of this technique; the ability to
examine the presence of specific GPI molecules and/or side
branch modifications in living cells. This could be applied to
such analyses as the investigation of GPI-APs on different mem-
brane compartments of polarized cells (as demonstrated here) or
the examination of GPI-APs on the cell surface during develop-
mental or drug time course studies. The ability to specifically
and non-destructively label and isolate the GalNAc-containing
GPI anchors should therefore prove to be a useful tool for the
analysis of GPI anchor structure, and even more importantly,
may allow us to begin to elucidate the functional roles of GPI
side branch modifications.
4.1. ARPE-19 cell culture
ARPE-19 cells were obtained from American Type Culture Col-
lection (ATCC). ARPE-19 non-polarized cultures were maintained
Figure 4. GalNAz does not get incorporated into GPIs prior to attachment to
protein. HeLa cells were labeled with D-[2-3H] mannose, then chased for 18 h in the
presence (lane 4) or absence (lane 2) of 50 lM GalNAz. Cell lysates were prepared
and followed by Staudinger ligation with biotin–PEG3–phosphine. Non-protein
associated GPI precursors were isolated by organic extraction and analyzed by TLC.
O = origin, f = solvent front.
S. Vainauskas et al./Carbohydrate Research 362 (2012) 62–69
in Dulbecco’s modified Eagle’s medium (DMEM): Ham’s F12 med-
ium with HEPES buffer containing 10% (v/v) fetal bovine serum
(FBS; Thermo ScientificHyclone),
100 lg/mL streptomycin. The cells were sub-cultured once a week.
All cells were maintained in a humidified 5% CO2atmosphere at
For polarized cell cultures, the cells were cultured on filters by
seeding at a density of 1.6 ? 105cells/cm2on 75 mm diameter,
0.4 lm pore size, polycarbonate or 24 mm diameter, 0.4 lm pore
size, polyester Transwell permeable supports (Costar). The cells
were grown in DMEM/F12 medium (see above) with 10% FBS for
1–2 days, after which the medium was replaced with DMEM/F12
medium containing low serum (1%) and cells cultured for 3–
Trans epithelial electrical resistance (TEER) was used to monitor
epithelial barrier formation and polarization using an EVOM with a
STX2 electrode (World Precision Instruments). A plateau in TEER
100 U/mL penicillinand
(35–45 X cm2) was reached in 8–10 days and, thereafter, it re-
mained essentially unchanged. Polarized cells were used for exper-
iments after culturing for 3–4 weeks.
Diffusion of 0.2 ng of recombinant Gaussia princeps luciferase
(Gluc, 20 kDa) and 0.1 ng of recombinant Cypridina noctiluca lucif-
erase (Cluc, ?75 kDa) from the apical to basolateral chamber was
used to measure the intactness of the cell monolayer during PI-
PLC treatment. Every 15 min for 1.5 h, 30 lL aliquots were col-
lected from the basolateral and apical chambers. Gluc and Cluc
activities were measured in 5 lL of sample using BioLux Gaussia
Luciferase and BioLux Cypridina Luciferase Assay Kits (New Eng-
land Biolabs). Luminescence was read with a Berthold Centro LB
960 luminometer (Berthold Technologies). Diffusive permeability
rates (Po) in centimeters per second were calculated as previously
Figure 5. Examination of the GalNAc modification on the CD73 GPI anchor in live polarized cells. (A) CD73 preferentially localized to the apical membrane in polarized ARPE-
19 cells. Cells were grown on filter supports and stained by indirect immunofluorescence followed by confocal microscopy. b-catenin staining was used as a marker for
basolateral membrane expression. Note the expression of CD73 on the apical surface (asterisks) and the partial co-localization with b-catenin (arrows) on the basolateral
surface. (B). The polarized cell monolayers were fully formed, as revealed by determining the permeability of the monolayer over time by measuring the diffusion of Gaussia
princeps and Cypridina noctiluca luciferases (Gluc and Cluc, respectively) from the apical to the basolateral compartments (see Section 4). Only upon dissociation of the tight
junctions by addition of EDTA and Triton X-100 was significant luciferase permeability observed. (C) CD73 containing a GalNAz-labeled GPI anchor was detected on both
apical (AP) and basolateral (BL) membranes in polarized cells. ARPE-19 cells were grown on filter supports in the presence of GalNAz and then the live cells were treated on
each membrane surface with PI-PLC. GalNAz-labeled CD73 released from apical and basal membranes was collected and immunoprecipitated separately. Deglycosylated
CD73 was labeled with biotin–PEG3–phosphine probe and analyzed by immunoblotting with anti-biotin and anti-CD73 antibodies. Scale bar = 5 lm.
S. Vainauskas et al./Carbohydrate Research 362 (2012) 62–69
where Pois diffusive permeability (cm/s), DLBis the change in baso-
lateral luminescence, LAis apical luminescence, Dt is the change in
time, VBis basolateral chamber volume (cm3), and A is filter surface
cDNA was generated via the SuperScriptIII reverse transcription
kit (Invitrogen) from 2.5 lg total RNA isolated from ARPE-19 cells
(RNeasy RNA extraction kit, Qiagen). To generate the CD73-GPI
construct, PCR amplification of the full length gene was performed
with Phusion DNA polymerase (New England Biolabs) using the
following primers: CD73 Forward 50-CATGAATTCCACAGCCATGTGT
CCCCGAGCC-30and CD73 Reverse 50-CATGTCTAGACTATTGGTATA
AAACAAAGATCACTGCC-30. The resulting PCR fragment was then
cloned into the EcoRI-XbaI sites of pcDNA3.1(+) (Invitrogen). To
generate the CD73–TMD construct, a knitting PCR approach was
used. The CD73 gene upstream from the GPI attachment site was
amplified using the following primers: CD73 Forward and CD73
30. The C-terminal domain of MCP, including the transmembrane
domain, was amplified from ARPE-19 cDNA with the following
GAGGCCTACTTACAAGCCTCCAGTCTC-3’ and MCP Reverse 5’-CAT-
fusion was then generated using the amplified CD73 and MCP frag-
ments as templates and the CD73 Forward and MCP Reverse prim-
ers. The resulting PCR fragment was subsequently cloned into the
EcoRI-XbaI sites of pcDNA3.1(+).
4.3. Incorporation of GalNAz into CD73
ARPE-19 cells (4–5 ? 106cells per 100 mm plate) were grown
for 48 h in DMEM/F12 medium containing 10% (v/v) FBS with or
without 50 lM tetraacetylated N-azidoacetylgalactosamine (Gal-
NAz, Invitrogen). After two days, the medium was replaced with
5 mL fresh DMEM/F12 without serum and cells were treated with
recombinant Bacillus cereus PI-PLC (3 U/mL final concentration) for
45 min at 37 ?C. The recombinant Bacillus cereus PI-PLC was ex-
pressed in Escherichia coli and purified as described elsewhere.43
Spent culture medium was harvested after PI-PLC treatment and
centrifuged for 10 min at 3000g to remove any cell debris. The
supernatant was incubated with 2 lg CD73 antibody (mouse
monoclonal, Santa Cruz Biotechnology) and 50 lL Protein G mag-
netic beads (New England Biolabs) for 16 h at 4 ?C. CD73 was
eluted from the beads with 50 lL of 1% SDS/100 mM Tris?HCl, pH
7.5. A 20 lL aliquot of eluate (CD73 IP) was treated with protein
deglycosylation mix (New England Biolabs) for 4 h as recom-
mended by the manufacturer. Staudinger ligation was performed
by reacting 25 lL of deglycosylated CD73 IP samples with 2 lL of
10 mM biotin–PEG3–phosphine (Pierce Thermo Scientific) for 4 h
at 37 ?C. Then CD73 IP samples were run on a SDS–PAGE, trans-
ferred onto nitrocellulose membrane and blotted with anti-bio-
tin-HRP (Cell Signaling Technologies), or anti-CD73 and anti-
mouse IgG HRP-conjugated light chain specific (Jackson Immuno-
research) antibodies. The membrane was developed using Super-
Signal West Pico or Dura Chemiluminescent Substrate (Pierce
To study GalNAz incorporation into recombinant CD73 con-
structs, 1 lg of pcDNA3.1(+), CD73–GPI or CD73–TMD DNA was
transfected into one well of a 6-well dish of ARPE-19 cells using
Lipofectamine2000 (Invitrogen). Growth media was changed 6 h
post-transfection, and GalNAz (Invitrogen) was added to the med-
ium to a final concentration of 50 lM 24 h after transfection. Cells
were harvested 72 h post-transfection by scraping in cold PBS and
pelleting at 4 ?C for 10 min at 1500 rpm. The cell pellet was
resuspended in 100 lL PBS + 2 U PI-PLC and incubated for 45 min
at 37 ?C. Subsequently, 900 lL RIPA buffer containing protease
inhibitors was added. Cells were vortexed and incubated on ice
for 30 min, then spun down for 20 min at 4 ?C at 14000 rpm. The
supernatant was precleared with 25 lL protein G magnetic beads
for 1 h at 4 ?C, then was subjected to immunoprecipitation with
CD73 antibody, followed by deglycosylation, labeling with bio-
tin–PEG3–phosphine via the Staudinger ligation, and detection of
CD73 by western blotting as described above.
4.4. Treatment with b-N-acetylhexosaminidase and aqueous HF
CD73 IP samples were treated with 0.01 U of jack bean b-N-
acetylhexosaminidase (Sigma–Aldrich) for 20 h at 25 ?C in 50 mM
sodium citrate, pH 4.5. CD73 samples transferred onto PVDF mem-
brane were treated with 48% aqueous hydrofluoric acid for 60 h on
For immunohistochemistry, HeLa cells cultured in Nunc Lab Tek
8-well chamber slides (Thermo Scientific) were transfected with
pcDNA3.1, CD73–GPI and CD73–TMD plasmid constructs using
Lipofectamine2000. Cells were rinsed once with PBS and fixed for
15 min in 4% PFA at room temperature 48 h after transfection.
For staining, all steps were performed at room temperature. Fol-
lowing fixation, cells were washed with PBS, then blocked in PBS
containing 5% BSA (w/v) for 15 min, and then incubated with
anti-CD73 antibody for 1 h diluted 1:100 in PBS containing 1%
BSA (w/v). Cells were washed with PBS and incubated for 1 h
AlexaFluor 488 anti-mouse IgG (Invitrogen) diluted 1:500 in PBS
containing 1% BSA (w/v). Cells were washed with PBS, DAPI (Sig-
ma–Aldrich) stained, and mounted with ProLong Gold Antifade
mounting media (Invitrogen).
ARPE-19 cells were maintained in culture for 3–4 weeks on fil-
ters as mentioned above. Cells were washed once with DPBS con-
taining calcium chloride and magnesium chloride (Invitrogen),
and then fixed for 15 min in PBS containing 4% PFA (w/v). Cells
were washed in PBS, quenched for 15 min in PBS with 75 mM
ammonium chloride, 25 mM glycine, and the membrane filters
were cut from the plastic inserts and placed into 24-well dishes
for staining. Membranes were blocked in PBS containing 5% BSA
(w/v) for 1 h at room temperature, and then incubated for 16 h
at 4 ?C in mouse anti-CD73 (Santa Cruz Biotechnology) primary
antibody diluted in PBS containing 1% BSA (w/v). Filters were
washed with PBS and then incubated with goat anti-mouse IgG
AlexaFluor 488 in PBS containing 1% BSA (w/v) for 1 h. Filters were
washed again with PBS and post-fixed with PBS containing 4% PFA
(w/v) for 15 min at rt. Filters were washed with PBS, then blocked
and permeabilized with PBS containing 5% BSA (w/v), 0.3% Triton
X-100 (v/v) for 30 min. Cells were then incubated with rabbit
anti-b-catenin (Cell Signaling Technologies) primary antibodies
for 1 h at rt. Filters were washed in PBS, then incubated in anti-rab-
bit IgG 568 secondary antibodies (Invitrogen) in PBS containing 1%
BSA (w/v) for 1 h. After washing in PBS, cells were DAPI stained and
filters mounted onto glass slides with ProLong Gold Antifade
mounting media. Images were acquired on a Zeiss LSM 510 Meta
confocal microscope (Carl Zeiss, Inc.) with LSM software and ana-
lyzed using ImageJ and Adobe Photoshop CS4.
4.6. Labeling of HeLa cells with [2-3H]mannose; extraction and
analysis of radiolabeled GPIs
HeLa (?1 ? 107) cells were labeled with D-[2-3H]mannose and
radiolabeled GPIs were extracted with chloroform–methanol–
water (10:10:3, by volume), desalted by partitioning between
S. Vainauskas et al./Carbohydrate Research 362 (2012) 62–69
n-butanol and water, and analyzed by TLC as described previ-
ously.44,45The chromatograms were visualized and analyzed using
a Typhoon PhosphorImager and ImageQuant image analysis soft-
ware (GE Healthcare Biosciences).
The authors wish to acknowledge Dr. Donald Comb of New Eng-
land Biolabs for financial support.
1. Ferguson, M. A. J.; Kinoshita, T.; Hart, G. W. In Essentials of Glycobiology; Varki,
A., Cummings, R. D., Esko, J. D., Freeze, H. H., Stanley, P., Bertozzi, C. R., Hart, G.
W., Etzler, M. E., Eds., 2nd ed.; Cold Spring Harbor Laboratory Press: Cold Spring
Harbor, 2009; pp 143–161.
2. Paulick, M. G.; Bertozzi, C. R. Biochemistry 2008, 47, 6991–7000.
3. Fujita, M.; Kinoshita, T. FEBS Lett. 2010, 584, 1670–1677.
4. Orlean, P.; Menon, A. K. J. Lipid Res. 2007, 48, 993–1011.
5. Pittet, M.; Conzelmann, A. Biochim. Biophys. Acta 2007, 1771, 405–420.
6. Ferguson, M. A. J. Cell Sci. 1999, 112, 2799–2809.
7. Houjou, T.; Hayakawa, J.; Watanabe, R.; Tashima, Y.; Maeda, Y.; Kinoshita, T.;
Taguchi, R. J. Lipid Res. 2007, 48, 1599–1606.
8. Maeda, Y.; Tashima, Y.; Houjou, T.; Fujita, M.; Yoko-o, T.; Jigami, Y.; Taguchi, R.;
Kinoshita, T. Mol. Biol. Cell 2007, 18, 1497–1506.
9. Fujita, M.; Maeda, Y.; Ra, M.; Yamaguchi, Y.; Taguchi, R.; Kinoshita, T. Cell 2009,
10. Vainauskas, S.; Menon, A. K. J. Biol. Chem. 2006, 281, 38358–38364.
11. Bate, C.; Williams, A. J. Biol. Chem. 2012, 287, 7935–7944.
12. Taron, B. W.; Colussi, P. A.; Wiedman, J. M.; Orlean, P.; Taron, C. H. J. Biol. Chem.
2004, 279, 36083–36092.
13. Homans, S. W.; Ferguson, M. A.; Dwek, R. A.; Rademacher, T. W.; Anand, R.;
Williams, A. F. Nature 1988, 333, 269–272.
14. Mehlert, A.; Varon, L.; Silman, I.; Homans, S. W.; Ferguson, M. A. Biochem. J.
1993, 296, 473–479.
15. Nakano, Y.; Noda, K.; Endo, T.; Kobata, A.; Tomita, M. Arch. Biochem. Biophys.
1994, 311, 117–126.
16. Mukasa, R.; Umeda, M.; Endo, T.; Kobata, A.; Inoue, K. Arch. Biochem. Biophys.
1995, 318, 182–190.
17. Brewis, I. A.; Ferguson, M. A.; Mehlert, A.; Turner, A. J.; Hooper, N. M. J. Biol.
Chem. 1995, 270, 22946–22956.
18. Taguchi, R.; Hamakawa, N.; Harada-Nishida, M.; Fukui, T.; Nojima, K.; Ikezawa,
H. Biochemistry 1994, 33, 1017–1022.
19. Stahl, N.; Baldwin, M. A.; Hecker, R.; Pan, K. M.; Burlingame, A. L.; Prusiner, S. B.
Biochemistry 1992, 31, 5043–5053.
20. Zaro, B. W.; Yang, Y. Y.; Hang, H. C.; Pratt, M. R. Proc. Natl. Acad. Sci. U.S.A. 2011,
21. Hang, H. C.; Yu, C.; Pratt, M. R.; Bertozzi, C. R. J. Am. Chem. Soc. 2004, 126, 6–7.
22. Dube, D. H.; Prescher, J. A.; Quang, C. N.; Bertozzi, C. R. Proc. Natl. Acad. Sci. U.S.A.
2006, 103, 4819–4824.
23. Laughlin, S. T.; Bertozzi, C. R. Nat. Protoc. 2007, 2, 2930–2944.
24. Laughlin, S. T.; Baskin, J. M.; Amacher, S. L.; Bertozzi, C. R. Science 2008, 320,
25. Zimmermann, H. Biochem. J. 1992, 285, 345–365.
26. Fini, C.; Amoresano, A.; Andolfo, A.; D’Auria, S.; Floridi, A.; Paolini, S.; Pucci, P.
Eur. J. Biochem. 2000, 267, 4978–4987.
27. Reigada, D.; Zhang, X.; Crespo, A.; Nguyen, J.; Liu, J.; Pendrak, K.; Stone, R. A.;
Laties, A. M.; Mitchell, C. Purinergic Signal. 2006, 2, 499–507.
28. Boyce, M.; Carrico, I. S.; Ganguli, A. S.; Yu, S. H.; Hangauer, M. J.; Hubbard, S. C.;
Kohler, J. J.; Bertozzi, C. R. Proc. Natl. Acad. Sci. U.S.A. 2011, 108, 3141–3146.
29. Lublin, D. M.; Coyne, K. E. J. Exp. Med. 1991, 174, 35–44.
30. Watanabe, R.; Ohishi, K.; Maeda, Y.; Nakamura, N.; Kinoshita, T. Biochem. J.
1999, 339, 185–192.
31. Azzouz, N.; Shams-Eldin, H.; Niehus, S.; Debierre-Grockiego, F.; Bieker, U.;
Schmidt, J.; Mercier, C.; Delauw, M. F.; Dubremetz, J. F.; Smith, T. K.; Schwarz, R.
T. Int. J. Biochem. Cell Biol. 2006, 38, 1914–1925.
32. Paladino, S.; Lebreton, S.; Tivodar, S.; Campana, V.; Tempre, R.; Zurzolo, C. J. Cell
Sci. 2008, 121, 4001–4007.
33. Paladino, S.; Sarnataro, D.; Pillich, R.; Tivodar, S.; Nitsch, L.; Zurzolo, C. J. Cell
Biol. 2004, 167, 699–709.
34. Paladino, S.; Sarnataro, D.; Tivodar, S.; Zurzolo, C. Traffic 2007, 8, 251–258.
35. Benting, J. H.; Rietveld, A. G.; Simons, K. J. Cell Biol. 1999, 146, 313–320.
36. Pang, S.; Urquhart, P.; Hooper, N. M. J. Cell Sci. 2004, 117, 5079–5086.
37. Imjeti, N. S.; Lebreton, S.; Paladino, S.; de la Fuente, E.; Gonzalez, A.; Zurzolo, C.
Mol. Biol. Cell 2011, 22, 4621–4634.
38. Strohmeier, G. R.; Lencer, W. I.; Patapoff, T. W.; Thompson, L. F.; Carlson, S. L.;
Moe, S. J.; Carnes, D. K.; Mrsny, R. J.; Madara, J. L. J. Clin. Invest. 1997, 99, 2588–
39. Nathke, I. S.; Hinck, L.; Swedlow, J. R.; Papkoff, J.; Nelson, W. J. J. Cell Biol. 1994,
40. Cohen, C. J.; Shieh, J. T.; Pickles, R. J.; Okegawa, T.; Hsieh, J. T.; Bergelson, J. M.
Proc. Natl. Acad. Sci. U.S.A. 2001, 98, 15191–15196.
41. Harhaj, N. S.; Barber, A. J.; Antonetti, D. A. J. Cell. Physiol. 2002, 193, 349–364.
42. Phillips, B. E.; Cancel, L.; Tarbell, J. M.; Antonetti, D. A. Invest. Ophthalmol. Vis.
Sci. 2008, 49, 2568–2576.
43. Koke, J. A.; Yang, M.; Henner, D. J.; Volwerk, J. J.; Griffith, O. H. Protein Expr.
Purif. 1991, 2, 51–58.
44. Vainauskas, S.; Menon, A. K. J. Biol. Chem. 2004, 279, 6540–6545.
45. Vidugiriene, J.; Menon, A. K. Methods Enzymol. 1995, 250, 513–535.
S. Vainauskas et al./Carbohydrate Research 362 (2012) 62–69