Toxoplasma gondii Uses Unusual Sorting Mechanisms to Deliver Transmembrane Proteins into the...

16
# 2008 The Authors Journal compilation # 2008 Blackwell Munksgaard doi: 10.1111/j.1600-0854.2008.00793.x Traffic 2008; 9: 1665–1680 Blackwell Munksgaard Toxoplasma gondii Uses Unusual Sorting Mechanisms to Deliver Transmembrane Proteins into the Host-Cell Vacuole Claire Gendrin 1 , Corinne Mercier 1 , Laurence Braun 1 , Karine Musset 1 , Jean-Franc xois Dubremetz 2 and Marie-France Cesbron-Delauw 1, * 1 Laboratoire Adaptation et Pathoge ´ nie des Micro-organismes, CNRS UMR 5163, Universite ´ Joseph Fourier GRENOBLE 1, Institut Jean Roget, BP 170, 38042 Grenoble cedex 9, France 2 CNRS UMR 5235, Bt 24, CC 107, Universite ´ de Montpellier 2, Place Euge ` ne Bataillon, 34095 Montpellier cedex 05, France *Corresponding author: Marie-France Cesbron-Delauw, [email protected] A critical step in infection by the apicomplexan parasite Toxoplasma gondii is the formation of a membrane- bound compartment within which the parasite prolifer- ates. This process relies on a set of secretory organelles that discharge their contents into the host cell upon invasion. Among these organelles, the dense granules are specialized in the export of transmembrane (TM) GRA proteins, which are major components of the mature parasitophorous vacuole (PV) membrane. How eukary- otic pathogens export and sort membrane-bound pro- teins destined for the host cell is still poorly understood at the mechanistic level. In this study, we show that soluble trafficking of the PV-targeted GRA5 TM protein is parasite specific: when expressed in mammalian cells, GRA5 is targeted to the plasma membrane and behaves as an integral membrane protein with a type I toplogy. We also demonstrate the dual role of the GRA5 N-terminal ectodomain, which is sufficient to prevent membrane integration within the parasite and is essen- tial for both sorting and post-secretory membrane inser- tion into the vacuolar membrane. These results contrast with the general rule that states that information con- tained within the cytoplasmic tail and/or the TM domain of integral membrane proteins dictates their cellular localization. They also highlight the diversity of sorting mechanisms that leads to the specialization of secretory processes uniquely adapted to intracellular parasitism. Key words: Apicomplexa, dense granules, parasitopho- rous vacuole, sorting, Toxoplasma gondii, transmem- brane proteins Received 9 March 2008, revised and accepted for publica- tion 3 July 2008, uncorrected manuscript published on- line 9 July 2008, published online 29 July 2008 Modifications of the host-cell vacuole in which the patho- gen establishes a safe residence is a strategy commonly used by the apicomplexan parasites Toxoplasma and Plasmodium or the bacterial pathogens Legionella and Chlamydia. Given that some of these modifications involve parasite-encoded membrane-bound proteins, the patho- gen has to solve the problem of both their delivery and membrane targeting beyond its own plasma membrane. In contrast to the well-studied prokaryotic secretion systems (1), how eukaryotic pathogens export and sort membrane- bound proteins is still poorly understood (2). Toxoplasma gondii is a widespread obligate intracellular parasite able to infect almost any nucleated mammalian or avian cell type. It replicates within a parasitophorous vacuole (PV) that is isolated from host vesicular traffic. This vacuole is surrounded by host-cell microtubules, endoplasmic reticulum (ER) and mitochondria. The PV is formed upon invasion and originates from both the host- cell plasma membrane and parasite secretory products. As any eukaryotic cell, Toxoplasma possesses a well-developed secretory apparatus that includes a perinuclear ER (3) and a single-stacked Golgi apparatus (4). Moreover, like other members of the Apicomplexa phylum, Toxoplasma also possesses specialized secretory organelles, namely the micronemes, the rhoptries and the dense granules (DG). Both entry into the host cell and formation of the PV are orchestrated by sequential secretion from these three types of secretory organelles (5). While micronemes and rhoptries drive early formation of the PV (6), maturation of the Toxoplasma-replicative vacuole is correlated with the intravacuolar release of DG proteins, which are targeted either to the PV-surrounding membrane or to the mem- branous nanotubular network formed within the vacuolar space (7,8). Among the DG proteins, nine proteins, named GRA1 to GRA9, do not present any homology with proteins of known function. Recent data using knock-out parasites support a role in metabolic exchanges between the parasites and the host cell (9,10) and in the spatial organization of the parasites within the PV (11,12). Each GRA protein contains an N-terminal (Nt) signal peptide (SP), which mediates entry into the secretory pathway (13). Except for GRA1, which is a soluble calcium-binding protein (14), all the GRA proteins identified so far are predicted to contain specific membrane association domains, which consist of a single typical a-helical transmembrane (TM) domain in the case of GRA3, 4, 5, 6, 7 and 8 or of amphipathic alpha-helices (GRA2 and 9) (13). Despite the presence of these membrane-association domains, GRA proteins remain excluded from the parasite www.traffic.dk 1665

Transcript of Toxoplasma gondii Uses Unusual Sorting Mechanisms to Deliver Transmembrane Proteins into the...

# 2008 The Authors

Journal compilation # 2008 Blackwell Munksgaard

doi: 10.1111/j.1600-0854.2008.00793.xTraffic 2008; 9: 1665–1680Blackwell Munksgaard

Toxoplasma gondii Uses Unusual Sorting Mechanismsto Deliver Transmembrane Proteins into theHost-Cell Vacuole

Claire Gendrin1, Corinne Mercier1,

Laurence Braun1, Karine Musset1,

Jean-Francxois Dubremetz2 and

Marie-France Cesbron-Delauw1,*

1Laboratoire Adaptation et Pathogenie desMicro-organismes, CNRS UMR 5163, Universite JosephFourier GRENOBLE 1, Institut Jean Roget, BP 170,38042 Grenoble cedex 9, France2CNRSUMR5235,Bt24,CC107,UniversitedeMontpellier2,PlaceEugeneBataillon, 34095Montpellier cedex05,France*Corresponding author: Marie-France Cesbron-Delauw,[email protected]

A critical step in infection by the apicomplexan parasite

Toxoplasma gondii is the formation of a membrane-

bound compartment within which the parasite prolifer-

ates. This process relies on a set of secretory organelles

that discharge their contents into the host cell upon

invasion. Among these organelles, the dense granules

are specialized in the export of transmembrane (TM) GRA

proteins, which are major components of the mature

parasitophorous vacuole (PV) membrane. How eukary-

otic pathogens export and sort membrane-bound pro-

teins destined for the host cell is still poorly understood

at the mechanistic level. In this study, we show that

soluble trafficking of the PV-targeted GRA5 TM protein is

parasite specific: when expressed in mammalian cells,

GRA5 is targeted to the plasma membrane and behaves

as an integral membrane protein with a type I toplogy.

We also demonstrate the dual role of the GRA5

N-terminal ectodomain, which is sufficient to prevent

membrane integration within the parasite and is essen-

tial for both sorting and post-secretory membrane inser-

tion into the vacuolar membrane. These results contrast

with the general rule that states that information con-

tained within the cytoplasmic tail and/or the TM domain

of integral membrane proteins dictates their cellular

localization. They also highlight the diversity of sorting

mechanisms that leads to the specialization of secretory

processes uniquely adapted to intracellular parasitism.

Key words: Apicomplexa, dense granules, parasitopho-

rous vacuole, sorting, Toxoplasma gondii, transmem-

brane proteins

Received 9 March 2008, revised and accepted for publica-

tion 3 July 2008, uncorrected manuscript published on-

line 9 July 2008, published online 29 July 2008

Modifications of the host-cell vacuole in which the patho-

gen establishes a safe residence is a strategy commonly

used by the apicomplexan parasites Toxoplasma and

Plasmodium or the bacterial pathogens Legionella and

Chlamydia. Given that some of these modifications involve

parasite-encoded membrane-bound proteins, the patho-

gen has to solve the problem of both their delivery and

membrane targeting beyond its own plasmamembrane. In

contrast to the well-studied prokaryotic secretion systems

(1), how eukaryotic pathogens export and sort membrane-

bound proteins is still poorly understood (2).

Toxoplasma gondii is a widespread obligate intracellular

parasite able to infect almost any nucleated mammalian or

avian cell type. It replicates within a parasitophorous

vacuole (PV) that is isolated from host vesicular traffic.

This vacuole is surrounded by host-cell microtubules,

endoplasmic reticulum (ER) and mitochondria. The PV is

formed upon invasion and originates from both the host-

cell plasma membrane and parasite secretory products. As

any eukaryotic cell, Toxoplasma possesses awell-developed

secretory apparatus that includes a perinuclear ER (3) and

a single-stacked Golgi apparatus (4). Moreover, like other

members of the Apicomplexa phylum, Toxoplasma also

possesses specialized secretory organelles, namely the

micronemes, the rhoptries and the dense granules (DG).

Both entry into the host cell and formation of the PV are

orchestrated by sequential secretion from these three

types of secretory organelles (5). While micronemes and

rhoptries drive early formation of the PV (6), maturation of

the Toxoplasma-replicative vacuole is correlated with the

intravacuolar release of DG proteins, which are targeted

either to the PV-surrounding membrane or to the mem-

branous nanotubular network formed within the vacuolar

space (7,8).

Among the DG proteins, nine proteins, named GRA1 to

GRA9, do not present any homologywith proteins of known

function. Recent data using knock-out parasites support

a role in metabolic exchanges between the parasites and

the host cell (9,10) and in the spatial organization of the

parasites within the PV (11,12). Each GRA protein contains

an N-terminal (Nt) signal peptide (SP), which mediates entry

into the secretory pathway (13). Except for GRA1, which is

a soluble calcium-binding protein (14), all the GRA proteins

identified so far are predicted to contain specific membrane

association domains, which consist of a single typical

a-helical transmembrane (TM) domain in the case of GRA3,

4, 5, 6, 7 and 8 or of amphipathic alpha-helices (GRA2 and 9)

(13). Despite the presence of thesemembrane-association

domains, GRA proteins remain excluded from the parasite

www.traffic.dk 1665

endomembranous system (15). Instead, they form high-

molecular-weight complexes and traffic to the DG in both

a soluble and an aggregated state (8,15–17). They are

secreted into the vacuolar space as soluble complexes (17)

and further associate with distinct PV membranous sub-

compartments: GRA2, 4, 6 and 9 associate with the

intravacuolar membranous nanotubes, while GRA3, 5, 7

and 8 associate with the PV membrane (13).

In eukaryotic cells, TM-domain-bearing proteins with an

ER-type signal sequence at the Nt enter the secretory

pathway and are delivered to the plasma membrane in

absence of any additional signal. In contrast, sorting of

TM-domain-bearing proteins to specific compartments is

mediated by signals present within their cytoplasmic

domain (18), specific residues or motifs present in the

TM domain (19,20) or its length (21). In Toxoplasma, while

it is well established that DG constitute the default

pathway for soluble proteins (22–24), the default route

for TM proteins has not yet been identified. In particular,

sorting of TM proteins to the vacuolar membranes versus

the parasite plasmamembrane remains to be investigated.

GRA5 is the prototypic protein chosen to analyze the

unusual export of parasite-encoded TM proteins to the

host-cell vacuole (15). It is a 21-kDa protein (120 amino

acids, aa) composed of two hydrophobic domains, a SP

and a single a-helical TM domain, the latter being flanked

by the Nt and C-terminal (Ct) hydrophilic domains (25).

Following soluble release into the PV, GRA5 spans the

vacuolar membrane and exposes its Nt to the host-cell

cytosol, while its Ct faces the vacuolar space (15). In this

study, we report that in mammalian cells, GRA5 is targeted

to the plasma membrane with a type I topology, providing

evidence that soluble trafficking of GRA5 in Toxoplasma is

peculiar. To investigate which are the determinants of

GRA5 DG targeting and subsequent PV membrane target-

ing, we substituted each domain of GRA5 with those of

a surface-targeted TM protein (26). We showed that in the

parasite, the Nt ectodomain of GRA5 mediates DG target-

ing and is essential for its selective sorting to the PV

membrane. These results contrast with the broad accep-

tance that targeting signals are present within the Ct tail of

TM proteins.

Results

When expressed in mammalian cells, GRA5 is targeted

to the plasma membrane with a type I topology

To examine if the intracellular soluble trafficking that is

observed for GRA5 in Toxoplasma is evolutionary con-

served, we compared in themammalian human embryonic

kidney (HEK) 293 T-cell line, the behavior of C-terminally

hemagglutinin (HA)-tagged GRA5 (GRA5-H; Figure 1A)

with that of C-terminally HA-tagged CD46 (CD46-H; Fig-

ure 1A) as a TM protein control. CD46 is an ubiquitous

mammalian surface protein, which also contains a single

TM-domain and displays a type I topology at the plasma

membrane (27).

Immunofluorescence microscopy on permeabilized cells

showed that GRA5 is targeted to the plasma membrane,

as indicated by the rim of fluorescence observed at the

cell periphery when using both a mouse serum directed

against the Nt part of GRA5 (anti-GRA5 Nt) (15) and the

anti-HA serum (Figure 1B, left). This staining was very

similar to that observed for CD46-H (Figure 1B, left).

Computational analysis using topology prediction methods

(TMPRED and TMHMM) indicates that GRA5 is expected to

behave as a type I TM protein in mammalian cells, with its

Nt domain extending outside the cell and its Ct domain

within the cytoplasm. To confirm biochemically that GRA5

behaves as an integral membrane protein in mammalian

cells, soluble proteins were separated from membranous

material following mechanical disruption of cells tran-

siently expressing C-terminally HA-FLAG-tagged GRA5

(GRA5-HF), and the partitioning of the protein was ana-

lyzed after treatments of the membranous pellet with

several denaturing agents (Figure 1C). GRA5 was exclu-

sively detected in the high-speed pellet (HSP), which

reflects its membrane association. This association was

not disrupted by treatments capable of releasing peripheral

membrane proteins such as high salt concentration or high

pH. Additionally, GRA5 was not solubilized by 6 M urea but

was totally solubilized by 1% Nonidet P-40 (NP-40), similar

to the CD46 TM protein. As a control, the transiently

expressed mammalian soluble protein was detected ex-

clusively in the high-speed soluble fraction (HSS) of HEK

293 T-cells (MLO2 in Figure 1C). These results demon-

strated that GRA5 behaves as a TM protein when ex-

pressed in a mammalian cell system. Cell surface labeling

was clearly observed when non-permeabilized HEK 293-T

cells expressing GRA5-H were incubated with the anti-

GRA5 Nt serum (Figure 1B, right). In contrast, no signal

was obtained when using the anti-HA antibody, which re-

cognizes the Ct tag (Figure 1B, right). In parallel, immuno-

detection of CD46-H expressed in HEK 293-T cells was

achieved with the TRA2-10 antibody, which recognizes the

Nt domain of CD46 (28), while no staining was observed

with the anti-HA serum (Figure 1B, right). Altogether,

these results showed that the Toxoplasma GRA5 protein

behaves like the TM control CD46 when expressed in HEK

293 T-cells: it is targeted to the plasma membrane with

a type I topology.

In Toxoplasma cells, GRA5 is targeted to the DG,

whereas both CD46 and a TM variant of SAG1 follow

a different route to the parasite surface

To further compare the trafficking of GRA5 with that of

CD46, we stably expressed the latter in Toxoplasma. By

contrast to what was previously observed in mammalian

cells, CD46-HF (Figure 2A) did not colocalize with endog-

enous GRA5 or GRA5-HF in Toxoplasma: in extracellular

parasites of the RH strain stably expressing the protein,

CD46-HF localized predominantly around the nucleus and

1666 Traffic 2008; 9: 1665–1680

Gendrin et al.

was associated to vesicle-like structures throughout the

cytoplasm (Figure 2B, right, lower panel, white arrow-

heads). These vesicles were distinguishable from the DG,

as indicated by the labeling of endogenous GRA5 (Figure

2B, right, lower panel, yellow arrowheads). Overnight in-

fected cells were permeabilized with saponin, which

allows selective permeabilization of both the host cell and

the vacuolar membranes. Under these conditions, no

staining of the parasite surface was observed neither with

the anti-Nt TRA2-10 antibody nor with the anti-Ct anti-HA

serum (Figure 2B, lower panel, left), while in these con-

ditions, the glycosyl-phosphatidylinositol (GPI)-anchored

SAG1 surface antigen was detected (data not shown).

These results suggested that the vesicles containing

CD46-HF were not fusing with the parasite plasma mem-

brane. Interestingly, a TM variant of the surface protein

SAG1, in which the GPI anchor was replaced by the TM and

Ct domains of human CD46 (construct SAG1/CD46TM/

CD46), was described to be targeted to the parasite plasma

membrane (26). A C-terminally HA-FLAG-tagged version of

this construct (SAG1/CD46TM/CD46-HF; Figure 2A) was

stably expressed in Dsag1 parasites. When infected cells

were fully permeabilized by Triton, the staining patterns

obtained with the anti-SAG1 monoclonal antibody (mAb)

and the anti-HA serum were similar, pointing out a surface

labeling as well as a labeling of the nucleus periphery

(Figure 2C, upper right panel). Selective saponin permeabi-

lization of overnight infected cells led to the detection of

the fusion protein at the parasite surface with the anti-

SAG1 mAb only, while no labeling was observed with the

anti-HA serum (Figure 2C, upper left panel). These results

indicated that the protein finally ends up at the parasite

plasma membrane with a type I topology, even if it is

delayed at the level of the parasite ER.

Following permeabilization with Triton-X-100, labeling of

extracellular parasites stably expressing SAG1/CD46TM/

CD46-HF indicated that the fusion is found in intracellular

Figure 1: GRA5 behaves as a type I

TM protein in mammalian cells. A)

Schematic representation of the pro-

teins GRA5-H/HF and CD46-H. The

genes from which the SP, the TM, the

Nt and the Ct domains were amplified

are indicated within the boxes. The

number of aa of each segment is indi-

cated between parentheses, and the

length of the TM domain is provided

below. Stars indicate the HA (H) or HA-

FLAG (HF) tag. B) GRA5-H or CD46-H

transiently expressed in HEK 293-T cells

was detected in cells permeabilizedwith

0.1% Triton-X-100 (permeabilized cells)

or without any prior treatment (living

cells). The Nt domain of each protein

was detected using the TRA-2-10 anti-

CD46mAb and the anti-GRA5 Nt serum,

respectively. The anti-HA rabbit serum

was used to detect the Ct tag. Bars,

10 mm. C) Solubility profile of GRA5-HF

in HEK cells. Transiently transfected

cells were fractionated into a high-speed

membrane pellet (HSP) and a HSS. The

pellet (P) and the soluble fraction (S)

obtained after treatment of the HSP

with several denaturing agents (PBS,

NaCl, CO3, urea and NP-40) were ana-

lyzed by immunoblot. Proteins were

detected using the anti-FLAG mAb

(GRA5-HF and MLO2-F) or the anti-

MCP rabbit serum (endogenous CD46).

MCP, membrane cofactor protein.

MLO2, the human homolog of the

ml02, SP:Q8N806.

Traffic 2008; 9: 1665–1680 1667

Trafficking to the Toxoplasma Parasitophorous Vacuole

vesicles (Figure 2C, lower panel, white arrowheads) dis-

tinct from the DG (Figure 2C, lower panel, yellow arrow-

heads). Thus, it seems that both CD46-HF and SAG1/

CD46TM/CD46-HF follow a vesicular route that could be

the default pathway for surface-targeted TM proteins. By

contrast, the DG targeting of TM proteins like GRA5

requires expression in Toxoplasma cells and may rely on

intrinsic properties of the GRA proteins.

Figure 2: Legend on next page.

1668 Traffic 2008; 9: 1665–1680

Gendrin et al.

Targeting of GRA5 to the DG does not depend on its

SP or on its Ct tail

To study whether (a) specific domain(s) of GRA5 would

be responsible for its targeting to the DG, we com-

pared the traffic of endogenous GRA5 with that of the

parasite surface protein SAG1/CD46TM/CD46-HF by sub-

stituting specific domains of GRA5 by those of SAG1 or

CD46.

All the GRA proteins identified up to now present a cleav-

able SP (13), which allows them to enter the secretory

pathway at the ER level. Replacement of the GRA5 SP by

that of CD46 (construct CD46SP/GRA5/GRA5TM/GRA5-HF;

Figure 3A) did not affect the DG targeting of GRA5 in

transiently transfected Dgra5 parasites (Figure 3B). Con-

versely, substitution of the SAG1 SP by that of GRA5 in the

context of the fusion protein SAG1/CD46TM/CD46-HF

(construct GRA5SP/SAG1/CD46TM/CD46-HF; Figure 3A)

and transient expression of this new construct in Dsag1parasites resulted in parasite plasma membrane localiza-

tion with a type I topology (Figure 3C). These results

indicated that the GRA5 SP is neither necessary nor

sufficient to target the protein to the DG.

In higher eukaryotic cells, sorting of TM proteins is often

mediated by signals present within their cytoplasmic tail,

that is the Ct domain in the case of type I TM proteins

(18,29). The potential role of the GRA5 Ct domain was

investigated by stably expressing the fusion protein

GRA5/GRA5TM/CD46-HF (Figure 4A) in Dgra5 parasites.

Immunofluorescence analysis showed that the chimeric

protein colocalized with endogenous GRA3 within the

DG (Figure 4B). Conversely, stable expression of SAG1/

CD46TM/GRA5-HF (Figure 4A) in Dsag1 parasites re-

sulted in the insertion of the fusion in the parasite plasma

membrane with a type I membrane topology (Figure 4C).

Together, these results indicated that the Ct domain of

GRA5 is not involved in the targeting of TM proteins to

the DG.

The Nt domain of GRA5 is sufficient to prevent

membrane insertion within Toxoplasma cells

In a previous study (30), fusion proteins based on the

bacterial alkaline phosphatase (BAP) reporter protein com-

bined with various TM and Ct domains were used to show

that in Toxoplasma, the length of the TM domain is a key

factor that determines targeting to the DG versus the

parasite plasma membrane. We thus examined if specifi-

cities of the GRA5 TM domain are involved in the soluble

targeting of the protein to the DG. Both the fusion proteins

SAG1/CD46TM/CD46-HF (Figure 2A) and SAG1/CD46TM/

GRA5-HF (Figure 4A), which display the 23-aa TM domain

of CD46, were targeted to the parasite surface when

stably expressed in Dsag1 parasites (Figures 2C and 4C).

In contrast, the fusion protein SAG1/GRA5TM/GRA5-HF

(Figure 5A), which contains the 18-aa TM domain of GRA5,

was targeted to the DG when stably expressed in the

same strain (Figure 5C, upper panel). Extracellular para-

sites stably expressing SAG1/CD46TM/GRA5-HF or SAG1/

GRA5TM/GRA5-HF submitted to Triton-X-114 partition pre-

sented the same solubility profile, with both fusion pro-

teins partitioning into both the detergent and the aqueous

phase (Figure 5D). Yet, when parasites were submitted

to cellular fractionation, meaningful differences were

observed (Figure 5D). SAG1/CD46TM/GRA5-HF was exclu-

sively associated with themembranous fractions [HSP and

low-speed pellet (LSP)], and the LSP association was

solubilized by NP-40 only, which reflects a membrane

integration. In contrast, SAG1/GRA5TM/GRA5-HF behaved

as the GRA1 soluble control, being mainly soluble (HSS)

and also present in the LSP, which reflects an aggregated

state characteristic of DG proteins (Figure 5D) (15,16). This

LSP association presented the same characteristics as that

of GRA1 because SAG1/GRA5TM/GRA5-HF was fully sol-

ubilized by both urea and NP-40 (Figure 5D). These results

indicated that the nature of the TM domain is of import-

ance for the soluble targeting to the DG versus the para-

site surface. They were thus in agreement with the data

previously published using the BAP protein (30).

Figure 2: Both CD46-HF and SAG1/CD46TM/CD46-HF traffic through vesicles distinct from the DG, but only SAG1/CD46TM/CD46-

HF reaches the parasite surface. A) Schematic representation of (HA-FLAG tagged) GRA5 and of the fusion proteins CD46-HF and SAG1/

CD46TM/CD46-HF. Stars indicate the HA-FLAG (HF) tag. Annotations and symbols are described in Figure 1. B) Immunolabeling of CD46-

HF compared with that of endogenous GRA5 and GRA5-HF. Endogenous GRA5 was labeled with the anti-GRA5 mAb in intracellular RH

parasites (upper panel). GRA5-HF expressed in the Dgra5 knocked-out strain was labeled with the anti-HA rabbit serum (middle panel).

Upon 0.002% saponin treatment of infected cells, the host-cell membrane and the PVM were selectively permeabilized (host-cell

mbn þ PVmbn perm., left), whereas total permeabilization was obtained by treatment with 0.1% Triton-X-100 (full perm., right). Parasites

stably expressing CD46-HF were labeled as intracellular parasites following saponin permeabilization of infected cells (lower panel, left) or

as extracellular parasites permeabilized with Triton-X-100 (lower panel, right). The TRA-2-10 mAb and the anti-HA rabbit serum were used

to detect the Nt and Ct parts of the fusion protein, respectively. In extracellular parasites, localization of the fusion protein (white

arrowheads) was compared with that of endogenous GRA5 detected with the anti-GRA5 mAb (yellow arrowheads). Bars represent 2 and

5 mm for extracellular and intracellular parasites, respectively. C) Immunostaining of SAG1/CD46TM/CD46-HF in intracellular (upper panel,

bars 5 mm) and extracellular (lower panel, bars 2 mm) Dsag1 parasites stably expressing the fusion protein. In infected cells, the fusion

protein was detected with the anti-SAG1 mAb directed against its Nt domain, while the anti-HA serum was used to detect the Ct tag.

In extracellular parasites permeabilized with Triton-X-100, the anti-HA serum was used to detect the fusion protein (white arrowheads),

while endogenous GRA5 was detected with the anti-GRA5 mAb (yellow arrowheads).

Traffic 2008; 9: 1665–1680 1669

Trafficking to the Toxoplasma Parasitophorous Vacuole

To determine whether these results were relevant in the

case of GRA5, fusion constructs derived from the GRA5

gene and differing in their TM domain were stably trans-

fected in Dgra5 parasites. Surprisingly, DG targeting was

still observed upon substitution of the GRA5 18-aa TM

domain by the CD46 23-aa TM domain (GRA5/CD46TM/

GRA5) or the CD8 27-aa TM domain (GRA5/CD8TM/GRA5)

(Figure 5A,B,E). Subcellular fractionation analysis of extra-

cellular parasites showed that both the fusion proteins

GRA5/CD46TM/GRA5 and GRA5/CD8TM/GRA5 behaved as

endogenous GRA5 (15): (i) they were mainly associated

with the detergent fraction following Triton extraction and

(ii) they were not only present in the soluble fraction but

also in an aggregated state that was fully solubilized by NP-

40 only (Figure 5F). Taken together, these results showed

that in the context of the SAG1 Nt domain, the properties

of the TM domain influence targeting to the DG versus the

parasite surface. They also showed that the GRA5 Nt

domain contains a very strong signal that prevents TM-

domain-bearing proteins from being inserted into the

parasite plasma membrane and targets them to the DG.

Finally, very importantly, they demonstrated that the signal

contained within the GRA5 Nt domain predominates over

the influence of the TM domain.

Figure 3: The SP domain of GRA5 is neither necessary nor sufficient for DG targeting. A) Schematic representation of endogenous

GRA5 and of the fusion proteins CD46SP/GRA5/GRA5TM/GRA5-HF and GRA5SP/SAG1/CD46TM/CD46-HF. Annotations and symbols are

described in Figure 1. B) Immunostaining of CD46SP/GRA5/GRA5TM/GRA5-HF transiently expressed in Dgra5 parasites (upper panel).

Overnight infected cells permeabilized with Triton-X-100 were incubated with the anti-HA serum and the anti-GRA3 mAb to detect the

chimeric protein and endogenous GRA3, respectively. For a comparison, endogenous GRA5 (detected with the anti-GRA5 mAb) was co-

detected with endogenous GRA3 (detected with the anti-GRA3 serum) in RH parasites (lower panel). Bars, 5 mm. C) Immunolocalization of

the chimeric protein GRA5SP/SAG1/CD46TM/CD46-HF in infected cells permeabilized either with saponin (left) or with Triton-X-100 (right).

The construct was transiently expressed in Dsag1 parasites, which allowed the use of the anti-SAG1 mAb as an anti-Nt antibody. The Ct

part of the fusion protein was detected with the anti-HA serum. Bars, 5 mm.

1670 Traffic 2008; 9: 1665–1680

Gendrin et al.

The Nt domain is essential for the selective post-

secretory targeting of GRA5 to the PV membrane

Following soluble secretion into the vacuolar space, the

GRA5 TM protein is selectively targeted to the PV mem-

brane (15,25). We (this study) and others (30) have shown

that the TM domain of GRA proteins is of importance in the

soluble targeting to the DG. Whether it is also involved in

post-secretory targeting events was investigated by com-

paring the behavior of the fusion proteins GRA5/CD8TM/

GRA5 and GRA5/CD46TM/GRA5 stably expressed in Dgra5parasites with that of endogenous GRA5. When infected

cells were selectively permeabilized with saponin, the

three proteins detected with the anti-GRA5 mAb colocal-

ized with endogenous GRA3 at the PV membrane (Fig-

ure 6A). This indicated that targeting to the PV membrane

does not rely on peculiar properties of the TM domain.

After removing parasites from overnight infected cells, the

cellular debris and vacuolar materials were fractionated

into a soluble fraction (HSS) and a membranous fraction

(HSP). The fusion proteins partitioned into the HSP frac-

tion, like wild-type GRA5. This association was resistant to

urea, but membrane-associated proteins were solubilized

by NP-40 (Figure 6B). Together, these results showed that

the TM domain of either CD8 or CD46 allows efficient

targeting and membrane insertion of GRA5 into the PV

membrane, indicating that specific properties of the GRA5

TM domain are not involved.

To examine the potential role of the GRA5 Ct domain in

targeting to the PV membrane, the fusion protein GRA5/

GRA5TM/CD46-HFwas stably expressed in Dgra5 parasites,

and its vacuolar localization was compared with that of

endogenous GRA5. Incubation of infected cells permeabi-

lized with saponin with the anti-GRA5mAb showed that the

fusion protein colocalized with GRA3 at the PV membrane,

like endogenous GRA5 (Figure 6A). After selective perme-

abilization of the host-cell plasma membrane with digitonin

(15), the PV was not stained with the anti-HA serum,

whereas under the same condition, a staining of the PV

membrane was observed with the anti-GRA5 mAb directed

against the Nt domain of GRA5 (Figure 6C). These data

confirmed that, similar to the GRA5/GRA5TM/GRA5-HF

fusion protein, GRA5/GRA5TM/CD46-HF is inserted into

the PV membrane, with its Nt protruding into the host-cell

cytoplasm. These data thus demonstrated that the Ct

domain of CD46 does not modify the post-secretory traf-

ficking of GRA5 and thus argue against a role of the GRA5Ct

domain as a targeting determinant to the PV membrane.

Having excluded a role of both the TMand the Ct domains in

the post-secretory membrane targeting of GRA5, we finally

examined the role of the GRA5 Nt domain in these events.

To determine if this domain is sufficient to mediate target-

ing to the PV membrane, the fusion protein GRA5/CD46TM/

CD46-HF (Figure 7A) was stably expressed in Dgra5 para-

sites. Staining of extracellular parasites showed that the

fusion protein was targeted to the DG (Figure S1A). When

infected cells were permeabilized with saponin, a clear

staining of the PV membrane, overlaying that obtained with

the anti-GRA3 mAb (Figure 7B), was observed using the

anti-HA serum. Fractionation analysis performed on over-

night infected cells showed that GRA5/CD46TM/CD46-HF

was associatedwith themembrane fraction (HSP) (Figure 7C).

However, in contrast to GRA5, this association was not

totally released by treatment with NP-40. This may reflect a

stronger interaction with lipids as it can be observed within

raft domains. Together, these results demonstrated that

the Nt domain of GRA5 is sufficient to mediate both

targeting andmembrane insertion of the protein into the PV.

To study the necessity of the GRA5 Nt domain for insertion

into the PV membrane, the chimeric protein SAG1/

Figure 4: The Ct domain of GRA5 is

not involved in the targeting to the

DG. A) Schematic representation of

the fusion proteins GRA5/GRA5TM/

CD46-HF and SAG1/CD46TM/GRA5-

HF. Annotations and symbols are

described in Figure 1. B) Extracellular

parasites of the Dgra5 strain stably

expressing GRA5/GRA5TM/CD46-HF

were permeabilized with Triton-X-

100. The fusion protein and endoge-

nous GRA3 were detected with the

anti-HA rabbit serum and the anti-

GRA3 mAb, respectively. Bars, 2 mm.

C) Immunolocalization of the chimeric

protein SAG1/CD46TM/GRA5-HF in

infected cells permeabilized either

with saponin (left) or with Triton-X-100

(right). The anti-SAG1 mAb was used

as an anti-Nt antibody, while the Ct

part of the fusion protein was detected

with the anti-HA serum. Bars, 5 mm.

Traffic 2008; 9: 1665–1680 1671

Trafficking to the Toxoplasma Parasitophorous Vacuole

GRA5TM/GRA5-HF (Figure 7A) was stably expressed in

Dsag1 parasites. Unexpectedly, immunofluorescence ana-

lysis of overnight infected cells showed that the protein,

detected with the anti-HA serum, colocalized with GRA1 in

the vacuolar space (Figure 7B). Biochemical analysis of the

vacuolar fraction of infected cells showed that SAG1/

GRA5TM/GRA5-HF, similar to the soluble control protein

GRA1, was present in the PV HSS fraction only (Figure 7D).

The GRA5 Nt domain is thus necessary to mediate the

targeting of a fusion protein to the PV membrane.

The flanking regions of TM domains have been shown to

be of importance in both the targeting and the solubility of

TM proteins (31). The influence of the GRA5 TM domain

Figure 5: Legend on next page.

1672 Traffic 2008; 9: 1665–1680

Gendrin et al.

could thus have been diminished because of the SAG1 Nt

domain environment. To address this possibility, we stably

expressed the construct SAG1/8aa/GRA5TM/GRA5-HF,

which preserved the 8 aa Nt to the GRA5 TM domain

(Figure 7A), in Dsag1 parasites. Following targeting to the

DG (Figure S1B), the fusion protein was detected within

the vacuolar space (Figure 7B) as a fully solubilized protein

(Figure 7D). This indicated that the peculiar properties of

the GRA5 Nt domain do not reside in the Nt boundary of the

TM domain. Taken together, these results demonstrated

that the Nt domain of GRA5 is both necessary and sufficient

tomediate selective targeting to the PVmembrane and that

the membrane association of the protein does not rely on

the Nt juxtamembranous region of the TM domain.

Discussion

During invasion, Toxoplasma secretes numerous proteins

destined to modify host-cell compartments. This process

includes the export of the GRA proteins that are integrated

into the membranes of the vacuole in which the parasite

replicates (PV membrane and intravacuolar nanotubular

network). Such transport of TM proteins outside the

parasite does not involve prior delivery to its own plasma

membrane. Instead, we showed that TM proteins en route

to the parasite surface versus the PV membrane are

segregated throughout the secretory pathway and that

the parasite has evolved specialized sorting systems to

ensure targeting of single-pass TM proteins into PV

membranous subcompartments. Using the TM GRA5

protein as a model, we demonstrated the dual role of its

Nt domain, which prevents membrane insertion in the

parasite early secretory pathway, while it mediates PV

membrane insertion when the protein reaches the con-

fines of the parasite-surrounding vacuole.

Sorting to the DG

T. gondii DG possess two unusual properties: they consti-

tute the default pathway for soluble proteins (22–24) and

part of their luminal content is constituted by single-pass

TM proteins packaged in both a soluble form and aggre-

gates (8,15–17).

We first showed in this study that, when expressed in

mammalian cells, one of these TM proteins, GRA5, dis-

plays biochemical properties similar to those of the integral

membrane protein CD46 and adopts the same type I

topology in the plasma membrane. GRA5 thus possesses

all the elements required for efficient cotranslational inser-

tion into the ERmembrane of mammalian cells. In addition,

because the plasma membrane constitutes the default

compartment for membrane proteins in animal cells, we

concluded that GRA5 does not contain any of the signals

recognized by the mammalian sorting machinery.

By contrast, when expressed in Toxoplasma, both CD46

and a TM variant of the parasite surface antigen SAG1

(SAG1/CD46TM/CD46-HF) behave differently from GRA5:

both fusion proteins localize at the ER and in vesicles

distinct from the DG. Moreover, despite its ER retention,

SAG1/CD46TM/CD46-HF ends up at the parasite surface.

A possible explanation is that the CD46 cytoplasmic

tail contains an ER localization signal that is recognized in

the parasite but not in mammalian cells, as previously

described (31). Alternatively or additionally, the CD46 Ct

domain may carry export signals that are not recognized by

the Toxoplasma machinery. The paucity of N-glycosylation

processes in T. gondii (32) may indeed obviate the lectin-

dependent ER export system that occurs in most eukary-

otic cells. In addition, as shown by urea treatment, GRA5 is

not membrane associated in the DG [(15), this study],

whereas the SAG1/CD46TM/CD46-HF variant behaves as

a TM protein within the parasite. Previous work had shown

that in Toxoplasma, the length of the TM domain is a critical

feature in the segregation of membrane proteins to the DG

versus the Golgi (30). Our results did not contradict these

conclusions as SAG1/GRA5TM/GRA5-HF (TM domain of 18

aa) and SAG1/CD46TM/GRA5-HF (TM domain of 23 aa) are

Figure 5: The Nt domain of GRA5 predominates over the TM domain in the targeting to the DG. A) Schematic representation of the

fusion proteins SAG1/GRA5TM/GRA5-HF, SAG1/CD46TM/GRA5-HF, GRA5/CD46TM/GRA5 and GRA5/CD8TM/GRA5. Annotations and

symbols are described in Figure 1. B) Sequences of the TM domains of GRA5, CD46 and CD8. C) Comparative immunolocalization of

SAG1/GRA5TM/GRA5-HF and SAG1/CD46TM/GRA5-HF stably expressed in Dsag1 parasites. SAG1/GRA5TM/GRA5-HF was detected with the

anti-HA rabbit serum, while endogenous GRA5 was revealed with the anti-GRA5 mAb in extracellular parasites permeabilized with Triton-X-

100 (upper panel, bars 2 mm). After saponin permeabilization of infected cells, the Nt domain of SAG1/CD46TM/GRA5-HF was detected with

the anti-SAG1 mAb, whereas the anti-HA rabbit serum was used to detect the Ct tag (lower panel, bars 5 mm). D) Solubility profile of stably

expressed SAG1/GRA5TM/GRA5-HF and SAG1/CD46TM/GRA5-HF in extracellular parasites. After treatment with Triton-X-114, proteins were

fractionated into an insoluble pellet (I), a detergent phase (D) and an aqueous phase (A). The parasite lysate was also separated by low-speed

centrifugation into cell ghosts (LSP) and a soluble fraction (low-speed supernatant, LSS). This fraction was further separated by a high-speed

spin into a membrane pellet (HSP) and a soluble fraction (HSS). The LSP was submitted to treatments by denaturing agents, and solubilized

proteins (S) were separated from residuals (P) by high-speed centrifugation. The fusion proteins were detected with anti-HA serum. Both

endogenous GRA5 and GRA1, used as controls, were detected using the anti-GRA5mAb and the anti-GRA1mAb, respectively. E) Localization

of GRA5/CD46TM/GRA5 and GRA5/CD8TM/GRA5 stably expressed inDgra5 parasites. Following complete permeabilization with Triton-X-100,

fusion proteinswere detected using the anti-GRA5mAb, and their localizationwas comparedwith that of endogenousGRA3 revealedwith the

anti-GRA3 serum. Bars, 5 mm. F) Cell fractionation analysis of both GRA5/CD46TM/GRA5 and GRA5/CD8TM/GRA5 in extracellular parasites.

Stable transformants were submitted to the cell fractionation procedure described in (D). Fractions were analyzed using the anti-GRA5 mAb,

and the behavior of fusion proteins was compared with that of endogenous GRA5 analyzed in the wild-type RH strain.

Traffic 2008; 9: 1665–1680 1673

Trafficking to the Toxoplasma Parasitophorous Vacuole

targeted to the DG and the parasite surface, respectively.

However, we showed in this study that targeting to the

DG does not solely depend on peculiar properties of the

TM domain because both GRA5/CD8TM/GRA5 (TM domain

of 27 aa) and GRA5/CD46TM/GRA5 (TM domain of 23 aa)

were targeted to the DG pathway. Our results thus

demonstrated that the Nt domain of GRA5 predominates

over the TM domain in DG targeting. Synergy of sorting

signals was previously demonstrated in mammalian cells

to confer robustness to the sorting system [for a review on

sorting of proteins to dense core secretory granules, see

Dikeakos and Reudelhuber (18)]. It is likely that Toxo-

plasma has also developed security systems to ensure

targeting of single-pass TM proteins to the DG using

characteristics of both the TM domain and the Nt domain

as sorting elements.

Biochemical analysis of GRA5/CD8TM/GRA5 (TM domain

of 27 aa) and GRA5/CD46TM/GRA5 (TM domain of 23 aa)

showed that these chimeric proteins behave similarly to

endogenous GRA5 and other TM GRA proteins in being

found as both soluble and aggregated forms, the latter

Figure 6: Neither the TM domain nor

the Ct domain of GRA5 is necessary for

insertion into the PV membrane. A)

Immunolocalization of GRA5/CD46TM/

GRA5, GRA5/CD8TM/GRA5 and GRA5/

GRA5TM/CD46-HF within the PV. HFF cells

were infected either by parasites of the RH

strain or by Dgra5 parasites stably express-

ing the chimeric proteins and selectively

permeabilizedwith saponin. The anti-GRA5

mAb allowed detection of both endoge-

nous GRA5 and fusion proteins. GRA3 co-

staining was obtained with the anti-GRA3

serum. Bars, 5 mm. B) Solubility profile of

vacuolar GRA5/CD46TM/GRA5 and GRA5/

CD8TM/GRA5 compared with that of en-

dogenous GRA5. Cells were infected over-

night by Dgra5 parasites stably expressing

the chimeric proteins. After removal of

intracellular parasites, the membranes

(HSP) were separated from soluble pro-

teins (HSS) by a high-speed spin. The HSP

was treated with denaturing agents, and

membrane pellets (P) were separated from

soluble fractions (S) by a second high-

speed spin. Fractions were analyzed using

the anti-GRA5 mAb. C) Topological study

of GRA5/GRA5TM/CD46-HF in the PV

membrane (upper panel) compared with

that of GRA5-HF (lower panel). Cells were

infected with Dgra5 parasites stably ex-

pressing the chimeric proteins and per-

meabilized with 0.002% digitonin. The

anti-GRA5 mAb and the anti-HA serum

were used as anti-Nt and anti-Ct antibod-

ies, respectively. Bars, 5 mm.

1674 Traffic 2008; 9: 1665–1680

Gendrin et al.

being partially solubilized by urea treatment. We have

recently shown that one prominent feature of TM GRA

proteins is to form large multimeric complexes (>1 MDa)

within the parasite, whereas the other GRA proteins, such

as soluble GRA1, are included into much smaller com-

plexes. It was proposed that formation of such large

complexes would prevent TM insertion of GRA proteins

within the secretory pathway by sequestering their

Figure 7: The Nt domain of GRA5 is both necessary and sufficient to mediate post-secretory membrane insertion into the PV

membrane. A) Schematic representation of the fusion proteins GRA5/CD46TM/CD46-HF, SAG1/GRA5TM/GRA5-HF and SAG1/8aa/

GRA5TM/GRA5-HF. The stripped box represents the 8 aa of GRA5 bordering the GRA5 TM domain. Annotations and symbols are described

in Figure 1. B) Immunolocalization of GRA5/CD46TM/CD46-HF stably expressed in Dgra5 parasites and of the SAG1-derived fusion proteins

SAG1/GRA5TM/GRA5-HF and SAG1/8aa/GRA5TM/GRA5-HF stably expressed in Dsag1 parasites. Infected cells were selectively

permeabilized with saponin. The fusion proteins were detected with the anti-HA serum, and endogenous GRA3 or GRA1 were detected

with the anti-GRA3 or the anti-GRA1 mAb, respectively. Bars, 5 mm. C) Fractionation analysis of vacuolar GRA5/CD46TM/CD46-HF. HFF

cells were infected overnight with a Dgra5 mutant stably expressing the fusion protein and treated as described in Figure 5B. The anti-

GRA5 mAb was used to detect the fusion protein. As a control, endogenous GRA5 was detected with the anti-GRA5 mAb in RH parasites

submitted to the same treatments. D) Analysis of the solubility of vacuolar SAG1/GRA5TM/GRA5-HF and SAG1/8aa/GRA5TM/GRA5-HF.

Both fusion proteins were detectedwith the anti-HA serum, whereas endogenous GRA5 and GRA1were detected with the anti-GRA5 and

anti-GRA1 mAbs, respectively.

Traffic 2008; 9: 1665–1680 1675

Trafficking to the Toxoplasma Parasitophorous Vacuole

hydrophobic TM domain within the interior of the complex

(17). Our results showed that inclusion into large com-

plexes is likely to depend on the presence of the GRA5 Nt

domain within chimeric TM-bearing proteins. We thus

propose a model in which the Nt domain of TM-domain-

bearing GRA proteins is involved in their incorporation into

multimeric complexes. Consistent with previous work

showing that the DG constitute the default pathway for

soluble proteins (22–24), the formation of such complexes

through mediating aggregation/solubilization would be

sufficient to target the TM-domain-bearing GRA proteins

to the DG (Figure 8). Given that GRA proteins are highly

expressed, they could act as regulators in the conden-

sation process, leading to the formation of dense core

granules. According to this model, the cell specificity of DG

biogenesis would rely on the unique and intrinsic proper-

ties of GRA proteins.

It is not yet clear where proteins destined to the DG are

segregated from proteins destined to the parasite surface

and from the other secreted proteins. In mammalian

secretory cells, sorting of secreted proteins occurs classi-

cally at the trans Golgi network. Considering that the

solubilization process of TM-domain-bearing proteins (i.e.

prevention from stable membrane insertion within the

parasite) is likely to rely in part on the formation of multi-

meric complexes, the simplest model is that oligomeriza-

tion would occur as soon as proteins are synthesized, that

Figure 8: Proposedmodel of both GRA5 soluble export and its post-secretorymembrane insertion into the PVmembrane. GRA5

would enter the ER through a translocon-type apparatus (1). Whether GRA5 inserts temporarily into the ER membrane (2) or whether it is

solubilized at the very moment of its synthesis (20) is not known. GRA5 solubilization would occur through protein–protein interactions,

most probably at the ER level (3). Its Nt domain would allow self-assembly of the protein (GRA5 multimers) or hetero-oligomers formation

through interaction with chaperone proteins such as TgHsp90 or with other TM-domain-bearing GRA proteins (TM GRA oligomers) (3).

Consecutively to their transit through the Golgi complex (4), these high-molecular-weight complexes would aggregate and lead to the

formation of the DG (5). In contrast, surface-targeted TM proteins, such as SAG1-CD46TM-CD46-HF, would follow an alternative route to

the parasite plasmamembrane. Following secretion into the vacuolar space, GRA hetero-oligomerswould dissociate (6) and individual GRA

proteins would interact with specific accessory factors (7 and 70), leading to their proper targeting either to the PV membrane (8) or to the

intravacuolar membranous nanotubular network (80). PPM, parasite plasma membrane; PVM, parasitophorous vacuole membrane; X and

Y, unidentified vacuolar factors. The GRA5 Nt domain is in red. The black rectangles indicate the GRA TM domain, and the pink rectangles

indicate the CD46 TM domain.

1676 Traffic 2008; 9: 1665–1680

Gendrin et al.

is at the ER level (Figure 8). This would fit with the

observation of GRA5 interacting with TgHSP90b (17),

which is likely an ER-resident protein as indicated by the

presence of both a Nt SP and an ER retention sequence

KDEL at its Ct (33). The solubilized/aggregated TM-domain-

bearing GRA proteins would be targeted to the DG by bulk

flow, together with soluble proteins that lack sorting

determinants to other organelles, as was previously sug-

gested (24,34). Whether high-level condensation of GRA

proteins occurs already in the ER or at the trans Golgi

network remains to be investigated.

Targeting to the PV-surrounding membrane

Following soluble secretion from the DG into the PV, GRA5

is targeted to its final destination, the PV delimitating

membrane, where it adopts a TM conformation. Our

results demonstrated that the Nt domain of GRA5 is

necessary and sufficient to mediate membrane insertion

into the PV membrane. Whereas both GRA5/CD46TM/

CD46-HF and GRA5/GRA5TM/GRA5-HF behaved as TM

proteins in the PV membrane, substitution of the Nt

domain of GRA5 by that of SAG1 led to a chimeric protein

(SAG1/GRA5TM/GRA5-HF) that was exclusively soluble

within the vacuolar space. Moreover, given that the Nt

domain of GRA6 was also found to be essential to target

the protein to the membranous nanotubular network

(unpublished data), it is thus likely that the Nt domain of

TM-domain-bearing GRA proteins contains a sorting deter-

minant specific to the different membranous subcompart-

ments of the PV. These results contrast with the broad

acceptance that targeting signals are present within the Ct

tail of TM proteins.

The mechanism by which the Nt domain of GRA5 pro-

motes a conformational change from a soluble stage to

a TM conformation remains to be determined. Interest-

ingly, we have recently shown that the soluble stage of

GRA proteins observed within the parasite and the PV is

correlated with the formation of high-molecular-weight

complexes in which several TM-domain-bearing GRA

proteins associate intimately and that these interactions

are not preserved following insertion into the PV mem-

branes [(17); Ruffiot, AB, JG, K. M., AD,M-F. C-D. and C. M.,

UMR5163, Grenoble, unpublished data]. A modification in

the GRA proteins conformation leading to exposure of their

TM domain could explain their spontaneous insertion into

the PV membranes following secretion. In this regard, it is

important to consider that the vacuolar space constitutes

an environment enriched by extensive exocytosis of

parasite proteins, including a cyclophilin homologue that

may participate in protein refolding (35). Yet, the sole

exposure of a GRA TM domain is not sufficient for

membrane insertion because SAG1/GRA5TM/GRA5-HF

was unable to insert into the PV membrane. Interactions

between the GRA5 Nt domain and the specific vacuolar

cofactors (Figure 8) could thus contribute to post-secre-

tory membrane insertion. Because the majority of vacuolar

components are delivered from the DG at the same time

as GRA5, it is likely that establishment of new interactions

between the GRA5 Nt domain and the vacuolar proteins

relies on conformational modifications of GRA5 and/or of

protein partners following their secretion. Such changes

could be induced by new environmental conditions within

the PV. Alternatively, given that the PV membrane is

a mixed compartment derived from both the host cell

and the parasite, it is important to consider the possibility

of an interaction with host-cell components. Some of

these components could constitute accessory partners

of GRA proteins during their translocation into the PV

membrane. Of particular interest are the GPI-anchored

proteins of host origin. These proteins are selectively

included in the PV membrane during the invasion process,

while the vast majority of host integral membrane proteins

are excluded from it (36,37). Moreover, these host-derived

GPI-anchored proteins are likely to contribute to the

ordering of PV-membrane-associated proteins into mem-

brane microdomains (36,37). Therefore, another possibil-

ity, not necessarily exclusive from the previous one, is that

the GRA5 Nt domain would function as a specific raft-

targeting signal in a manner similar to the proteinaceous

raft-binding determinants found either within the Nt ecto-

domain of the prion protein (38) or within the juxtamem-

brane region of the epidermal growth factor receptor

ectodomain (39). Physiologically, such interaction with

host-cell components would be of particular evolutionary

interest in terms of host–parasite cross talk.

In conclusion, this work provided evidence for a novel

specialized sorting mechanism evolved by T. gondii to

ensure trafficking of single-pass TM proteins within its

host-cell residence. Such diversity in sorting mechanisms

is needed to comprehend the specialization of secretory

pathway in a cell-specific fashion.

Materials and Methods

Parasites, cell culture and transfectionTachyzoites of the RH hxgprt� strain (40), the mutants Dgra5 (41) and

Dsag1 (42) and transgenic lines derived from these strains were maintained

by serial passages in human foreskin fibroblasts [HFF, American Type

Culture Collection (ATCC)-CRL 1634] in DMEM (GibcoBRL) supplemented

with 1 mM glutamine, 10% FBS, 50 U/mL penicillin and 50 mg/mL strepto-

mycin (referred to as D10). Parasite transfection was performed by

electroporation, and phleomycin was used to select mutants derived from

the Dsag1 strain as previously described (43), whereas those derived from

the RH hxgprt� and Dgra5 strains were selected using both xanthine and

mycophenolic acid (40). HEK 293-T cells (ATCC-CRL 11268) were also

maintained in D10 and transfected using the calcium phosphate precipita-

tion method (44).

AntibodiesThemammalian protein CD46was detected using either the TRA-2-10 mAb

that recognizes the first ligand-binding motif in the CD46 Nt region (28) or

an anti-CD46 rabbit polyclonal antibody, both donated by Dr K. Liszewski

(Washington University School of Medicine, St Louis, MO, USA). The

various fusion proteins were detected using an anti-HA rabbit serum (gift

fromM.A. Hakimi, CNRS, UMR 5163-Universite Joseph Fourrier, Grenoble,

France), the mAbs BioM2 anti-FLAG� (Sigma-Aldrich), TG 17–113

Traffic 2008; 9: 1665–1680 1677

Trafficking to the Toxoplasma Parasitophorous Vacuole

anti-GRA5 that recognizes the Nt region of GRA5 (15,45), TG 05–54 anti-

SAG1 (46) and an anti-GRA5 Nt mouse serum (15). The mAbs TG 17–43

anti-GRA1 (45) and TG2H11 anti-GRA3 (47) as well as an anti-GRA3 rabbit

serum were also used.

Plasmid constructsSpecific primers (Table S1) were designed from the sequences of GRA5

(25), CD8 (48), CD46 (27) and SAG1 (49).

DNA constructs for transfection into HEK cellsThe coding regions of GRA5 and CD46 were cloned into a pBluescript KSþ

vector modified by the insertion of a HA-FLAG (HF) tag (50). These vectors

were used as templates to amplify GRA5-H (GRA5 in fusion with the HA

tag), GRA5-HF (GRA5 in fusion with the HA-FLAG tag) and CD46-H. After

cleavage with both BamHI and XhoI, the polymerase chain reaction (PCR)

fragments were cloned into the vector pcDNA3.1 (Invitrogen Life Sciences)

under the control of the cytomegalovirus promoter. The MLO2-F vector

was provided by Dr M.A. Hakimi.

DNA constructs for transfection into T. gondiiThe CD46 open reading frame (ORF) was cloned into the modified KSþ

under the control of the strong GRA1 promoter and its 50-untranslated

region (50-UTR) (50). All the other fusion constructs were cloned under the

control of the regulatory regions [promoter, both 50-UTR and 30-untranslated

region (30-UTR)] of endogenous GRA5. To obtain the constructs GRA5/

CD8TM/GRA5 and GRA5/CD46TM/GRA5, specific fragments of GRA5, CD8

or CD46 were amplified using primers containing 15–20 bases comple-

mentary to the DNA fragment to be amplified and tailed with 15–20 bases

complementary to the adjacent DNA fragments in the final construct,

allowing hybridization of adjacent fragments in further PCRs. The final

products, including the GRA5 regulatory regions, were then cloned into the

BamHI–NotI sites of the pBluescript SKþ vector expressing the selection

marker HXGPRT (provided by Dr D.S. Roos, University of Pennsylvania,

Philadelphia, PA, USA). For all the other constructs, the GRA5 regulatory

regions were first amplified and cloned into the modified pBluescript KSþ

vector bearing the HF tag. Following amplification of specific fragments

from the GRA5, SAG1 or CD46 ORF, the final fusion PCR products were

obtained using the same hybridization PCR technique. The final products

were cloned into the SmaI and EcoRI sites of the modified KSþ plasmid, in

frame with the GRA5 50-UTR and the 30-HF tag.

Prediction of membrane-spanning regions

and their orientationThe TMHMM and TMPRED softwares (http://www.cbs.dtu.dk/services/

TMHMM-2.0/ and http://www.ch.embnet.org/software/TMPRED_form.html)

available at the Expert Protein Analysis System proteomics server (ExPasy

server) (http://www.expasy.org/) were used to predict the topology of GRA

proteins in mammalian cells.

Immunofluorescence stainingHFF cells were grown on 12-mm coverslips in 24-well plates until

confluency and then infected. Overnight infected cells were washed in

PBS, fixed for 20 min in 5% formaldehyde in PBS at room temperature and

permeabilized for 10 min either selectively with 0.002% saponin (8) or

completely in 0.1% Triton-X-100. Extracellular parasites were left to adhere

to a cell monolayer for 30 min at 378C and permeabilized with 0.1% Triton-

X-100 following fixation. Transfected HEK cells were either permeabilized

with 0.1% Triton-X-100 after fixation or processed as living cells. After

rinsing, cells were incubated with primary antibodies and with anti-

immunoglobulin G (HþL) goat secondary antibodies coupled to Alexa

(Invitrogen, Molecular Probes). Coverslips were mounted using Prolong

antifade reagent (Molecular Probes) and observed using a Zeiss Axioplan II

equipped for phase-contrast and epifluorescence microscopy. Photographs

were taken at the magnification�100 using a Zeiss AxiocamMRm coupled

to the AXIOVISION 4.5 software and processed with ADOBE PHOTOSHOP 6.0.

Cell fractionationFor HEK cell fractionation, cells were lysed by syringing through a 22-gauge

needle. After removal of nuclei and debris by a 10-min centrifugation

at 1000 � g, the supernatant was centrifuged for 30 min at 130 000 � g.

The resulting membrane pellet was dispersed in PBS and submitted to

treatment with either 1 M NaCl, 0.1 M sodium carbonate (pH 11.3), 6 M urea

or 1% NP-40. After incubation on ice for 30 min, the samples were

centrifuged at 130 000 � g for 30 min to separate solubilized proteins

from residuals.

Cell fractionations of both extracellular and intracellular parasites were

performed as described previously (15): extracellular parasites were lysed

by three freeze/thaw cycles, and the lysate was submitted to Triton-X-114

extraction (51). Briefly, the lysate was treated with 10% Triton-X-114, and

after removal of the insoluble material, the aqueous phase was separated

from the detergent phase by a 10-min incubation at 378C. The parasite

lysate was also centrifuged for 10 min at 2500 � g. The low-speed

supernatant was submitted to ultracentrifugation at 100 000 � g for

1.5 h. The LSP resuspended in 50 mM Tris (pH 8.0) was treated with

denaturing agents (6 M urea or 1% NP-40) for 30 min at 48C, followed by

a 1.5-h ultracentrifugation at 100 000 � g. For intracellular parasites, over-

night infected HFF cells were mechanically disrupted by syringing through

a 27-gauge needle. After removal of intact parasites and cell debris by

centrifugation at 2500 � g for 10 min, the soluble fraction was separated

from themembrane-associated fraction by ultracentrifugation at 100 000 � g.

Aliquots of the membrane pellets were treated with 6 M urea or 0.1%

NP-40 for 30 min at 48C, followed by centrifugation at 100 000 � g, as

described previously (8). In all fractionation experiments, supernatants

were concentrated by trichloroacetic acid or acetone precipitation. Each

experiment was repeated three times.

Western blottingProteins were boiled for 2 min in SDS–PAGE sample buffer, separated by

electrophoresis on 13% polyacrylamide gels and transferred to nitrocellu-

lose membranes. Immunologic detection was achieved using appropriate

horseradish-peroxidase-conjugated secondary antibodies. Peroxidase activ-

ity was visualized by chemiluminescence using the Supersignal ECL

system (Pierce Chemical).

Acknowledgments

The authors thank Drs J. Dubuisson, M.A. Hakimi, K. Liszewski and D.S.

Roos for sharing reagents and acknowledge the contribution of E. Glyde for

the first constructs. They are also very grateful to Drs M. Meissner and

J. Gagnon for their critical reading of the manuscript. This study was

supported by the CNRS and the AC program ‘Dynamique et Reactivite des

Assemblages Biologiques’ (DRAB04/027). C. G. was the recipient of a PhD

fellowship from the French Ministry of Research.

Supporting Information

Additional Supporting Information may be found in the online version of

this article:

Figure S1: DG targeting of GRA5/CD46TM/CD46-HF and SAG1/8aa/

GRA5TM/GRA5-HF in extracellular parasites by immunofluorescence

double labeling. Dgra5 parasites expressing GRA5/CD46TM/CD46-HF (A)

or Dsag1 parasites expressing the SAG1-derived construct (B) were

permeabilized with Triton-X-100 and incubated with the anti-HA serum.

Co-detection of GRA3 (using the anti-GRA3 mAb) or GRA5 (using the anti-

GRA5 mAb) is presented. Bars, 2 mm.

Table S1: Sequence of the primers used in this study

1678 Traffic 2008; 9: 1665–1680

Gendrin et al.

Please note: Wiley-Blackwell are not responsible for the content or

functionality of any supporting materials supplied by the authors. Any

queries (other than missing material) should be directed to the correspond-

ing author for the article.

References

1. Galan JE, Wolf-Watz H. Protein delivery into eukaryotic cells by type III

secretion machines. Nature 2006;444:567–573.

2. Cesbron-Delauw MF, Gendrin C, Travier L, Ruffiot P, Mercier C.

Apicomplexa in mammalian cells: trafficking to the parasitophorous

vacuole. Traffic 2008;9:657–664.

3. Hager KM, Striepen B, Tilney LG, Roos DS. The nuclear envelope

serves as an intermediary between the ER and Golgi complex in the

intracellular parasite Toxoplasma gondii. J Cell Sci 1999;112:2631–

2638.

4. Pelletier L, Stern CA, Pypaert M, Sheff D, Ngo HM, Roper N, He CY,

Hu K, Toomre D, Coppens I, Roos DS, Joiner KS, Warren G. Golgi

biogenesis in Toxoplasma gondii. Nature 2002;418:548–552.

5. Carruthers VB, Sibley LD. Sequential protein secretion from three

distinct organelles of Toxoplasma gondii accompanies invasion of

human fibroblasts. Eur J Cell Biol 1997;73:114–123.

6. Lebrun M, Carruthers VB, Cesbron-Delauw MF. Toxoplasma secretory

proteins and their roles in cell invasion and intracellular survival. In:

Weiss LM, Kim K, editors. Toxoplasma gondii: TheModel Apicomplexan:

Perspectives and Methods. Elsevier Science Ltd, Academic Press,

London, UK; 2007, pp. 265–307.

7. Sibley LD, Krahenbuhl JL, Adams GM, Weidner E. Toxoplasma

modifies macrophage phagosomes by secretion of a vesicular network

rich in surface proteins. J Cell Biol 1986;103:867–874.

8. Sibley LD, Niesman IR, Parmley SF, Cesbron-Delauw MF. Regulated

secretion of multi-lamellar vesicles leads to formation of a tubulo-

vesicular network in host-cell vacuoles occupied by Toxoplasma gondii.

J Cell Sci 1995;108:1669–1677.

9. Coppens IJ, Dunn D, Romano JD, Pypaert M, Zhang H, Boothroyd JC,

Joiner KA. Toxoplasma gondii sequesters lysosomes from mammalian

hosts in the vacuolar space. Cell 2006;125:261–274.

10. Mercier C, Cesbron-Delauw MF, Ferguson DJP. Dense granules of the

infectious stages of Toxoplasma gondii: their central role in the host/

parasite relationship. In: Soldati D, Ajioka J, editors. Toxoplasma:Molecular

and Cellular Biology. Horizon Bioscience Norfolk, UK; 2007, pp. 475–492.

11. Mercier C, Dubremetz JF, Rauscher B, Lecordier L, Sibley LD,

Cesbron-Delauw MF. Biogenesis of nanotubular network in Toxo-

plasma parasitophorous vacuole induced by parasite proteins. Mol Biol

Cell 2002;13:2397–2409.

12. Travier L, Mondragon R, Dubremetz JF, Musset K, Mondragon M,

Gonzalez S, Cesbron-DelauwMF,Mercier C. Functional domains of the

Toxoplasma GRA2 protein in the formation of the membranous nano-

tubular network of the parasitophorous vacuole. Int J Parasitol 2008;38:

757–773.

13. Mercier C, Adjogble KD, Daubener WH, Cesbron-Delauw MF. Dense

granules: are they key organelles to help understand the parasitophorous

vacuole of all Apicomplexa parasites? Int J Parasitol 2005;35:829–849.

14. Cesbron-Delauw MF, Guy B, Torpier G, Pierce RJ, Lenzen G, Cesbron

JY, Charif H, Lepage P, Darcy F, Lecocq JP, Capron A. Molecular

characterization of a 23-kilodalton major antigen secreted by Toxo-

plasma gondii. Proc Natl Acad Sci U S A 1989;86:7537–7541.

15. Lecordier L, Mercier C, Sibley LD, Cesbron-Delauw MF. Transmem-

brane insertion of the Toxoplasma gondii GRA5 protein occurs after

soluble secretion into the host cell. Mol Biol Cell 1999;10:1277–1287.

16. Labruyere E, Lingnau M, Mercier C, Sibley LD. Differential membrane

targeting of the secretory proteins GRA4 and GRA6 within the para-

sitophorous vacuole formed by Toxoplasma gondii. Mol Biochem

Parasitol 1999;102:311–324.

17. Braun L, Travier L, Kieffer S, Musset K, Garin J, Mercier C, Cesbron-

Delauw MF. Purification of Toxoplasma dense granule proteins reveals

that they are in complexes throughout the secretory pathway. Mol

Biochem Parasitol 2008;157:13–21.

18. Dikeakos JD, Reudelhuber TL. Sending proteins to dense core secre-

tory granules: still a lot to sort out. J Cell Biol 2007;177:191–196.

19. Bonifacino JS, Cosson P, Shah N, Klausner RD. Role of potentially

charged transmembrane residues in targeting proteins for retention

and degradation within the endoplasmic reticulum. EMBO J 1999;10:

2783–2793.

20. Zaliauskiene L, Kang S, Brouillette CG, Lebowitz J, Arani RB, Collawn

JF. Down-regulation of cell surface receptors is modulated by polar

residues within the transmembrane domain. Mol Biol Cell 2000;11:

2643–2655.

21. Bretscher MS, Munro S. Cholesterol and the Golgi apparatus. Science

1993;261:1280–1281.

22. Karsten V, Qi H, Beckers CJ, Reddy A, Dubremetz JF, Webster P,

Joiner KA. The protozoan parasite Toxoplasma gondii targets proteins

to dense granules and the vacuolar space using both conserved and

unusual mechanisms. J Cell Biol 1998;141:1323–1333.

23. Striepen B, He CY, Matrajt M, Soldati D, Roos DS. Expression,

selection, and organellar targeting of the green fluorescent protein in

Toxoplasma gondii. Mol Biochem Parasitol 1998;92:325–338.

24. Striepen B, Soldati D, Garcia-Reguet N, Dubremetz JF, Roos DS.

Targeting of soluble proteins to the rhoptries and micronemes in

Toxoplasma gondii. Mol Biochem Parasitol 2001;113:45–53.

25. Lecordier L, Mercier C, Torpier G, Tourvieille B, Darcy F, Liu JL, Maes P,

Tartar A, Capron A, Cesbron-Delauw MF. Molecular structure of

a Toxoplasma gondii dense granule antigen (GRA 5) associated with

the parasitophorous vacuole membrane. Mol Biochem Parasitol 1993;

59:143–153.

26. Seeber F, Dubremetz JF, Boothroyd JC. Analysis of Toxoplasma gondii

stably transfected with a transmembrane variant of its major surface

protein, SAG1. J Cell Sci 1998;111:23–29.

27. Lublin DM, Liszewski MK, Post TW, Arce MA, Le Beau MM,

Rebentisch MB, Lemons LS, Seya T, Atkinson JP. Molecular cloning

and chromosomal localization of human membrane cofactor protein

(MCP). Evidence for inclusion in the multigene family of complement-

regulatory proteins. J Exp Med 1988;168:181–194.

28. Manchester M, Valsamakis A, Kaufman R, Liszewski MK, Alvarez J,

Atkinson JP, Lublin DM, Oldstone MB. Measles virus and C3 binding

sites are distinct on membrane cofactor protein (CD46). Proc Natl Acad

Sci USA 1995;92:2303–2307.

29. Bonifacino JS, Traub LM. Signals for sorting of transmembrane proteins

to endosomes and lysosomes. Annu Rev Biochem 2003;72:395–447.

30. Karsten V, Hegde RS, Sinai AP, Yang M, Joiner KA. Transmembrane

domain modulates sorting of membrane proteins in Toxoplasma gondii.

J Biol Chem 2004;279:26052–26057.

31. Hoppe HC, Ngo HM, Yang M, Joiner KA. Targeting to rhoptry

organelles of Toxoplasma gondii involves evolutionarily conserved

mechanisms. Nat Cell Biol 2000;2:449–456.

32. Fauquenoy S, Morelle W, Hovasse A, Bednarczyk A, Slomianny C,

Schaeffer C, Van Dorsselaer A, Tomavo S. Proteomics and glycomics

analyses of N-glycosylated structures involved in Toxoplasma gondii –

host cell interactions. Mol Cell Proteomics 2008;7:891–910.

33. Gupta RS. Phylogenetic analysis of the 90 kD heat shock family of

protein sequences and an examination of the relationship among

animals, plants, and fungi species. Mol Biol Evol 1995;12:1063–1073.

Traffic 2008; 9: 1665–1680 1679

Trafficking to the Toxoplasma Parasitophorous Vacuole

34. Reiss M, Viebig N, Brecht S, Fourmaux MN, Soete M, Di Cristina M,

Dubremetz JF, Soldati D. Identification and characterization of an

escorter for two secretory adhesins in Toxoplasma gondii. J Cell Biol

2001;152:563–578.

35. Carey KL, Donahue CG, Ward GE. Identification and molecular charac-

terization of GRA8, a novel, proline-rich, dense granule protein of

Toxoplasma gondii. Mol Biochem Parasitol 2000;105:25–37.

36. Mordue DG, Desai N, Dustin M, Sibley LD. Invasion by Toxoplasma

gondii establishes a moving junction that selectively excludes host cell

plasma membrane proteins on the basis of their membrane anchoring.

J Exp Med 1999;190:1783–1792.

37. Charron AJ, Sibley LD. Molecular partitioning during host cell penetra-

tion by Toxoplasma gondii. Traffic 2004;5:855–867.

38. Walmsley AR, Zeng F, Hooper NM. The N-terminal region of the prion

protein ectodomain contains a lipid raft targeting determinant. J Biol

Chem 2003;278:37241–37248.

39. Yamabhai M, Anderson RG. Second cysteine-rich region of epidermal

growth factor receptor contains targeting information for caveolae/

rafts. J Biol Chem 2002;277:24843–24846.

40. Donald RG, Carter D, Ullman B, Roos DS. Insertional tagging, cloning,

and expression of the Toxoplasma gondii hypoxanthine-xanthine-gua-

nine phosphoribosyltransferase gene. Use as a selectable marker for

stable transformation. J Biol Chem 1996;271:14010–14019.

41. Mercier C, Rauscher B, Lecordier L, Deslee D, Dubremetz JF, Cesbon-

Delauw MF. Lack of expression of the dense granule protein GRA5

does not affect the development of Toxoplasma tachyzoites. Mol

Biochem Parasitol 2001;116:247–251.

42. Rachinel N, Buzoni-Gatel D, Dutta C, Mennechet FJ, Luangsay S,

Minns LA, Grigg ME, Tomavo S, Boothroyd JC, Kasper LH. The

induction of acute ileitis by a single microbial antigen of Toxoplasma

gondii. J Immunol 2004;173:2725–2735.

43. Messina M, Niesman I, Mercier C, Sibley LD. Stable DNA trans-

formation of Toxoplasma gondii using phleomycin selection. Gene

1995;165:213–217.

44. Sambrook J, Russel DW. Calcium-phosphate-mediated transfection of

eukaryotic cells with plasmid DNAs. In: Irwin N, Curtis S, Zierler M,

McInerny N, Brown D, Schaefer S, editors. Molecular Cloning: A

Laboratory Manual. Cold Spring Harbor, New York: Cold Spring Harbor

Laboratory Press; 2001, 16.14–16.20.

45. Charif H, Darcy F, Torpier G, Cesbron-Delauw MF, Capron A. Toxo-

plasma gondii: characterization and localization of antigens secreted

from tachyzoites. Exp Parasitol 1990;71:114–124.

46. Rodriguez C, Afchain D, Capron A, Dissous C, Santoro F. Major surface

protein of Toxoplasma gondii (p30) contains an immunodominant

region with repetitive epitopes. Eur J Immunol 1985;15:747–749.

47. Achbarou A, Mercereau-Puijalon O, Sadak A, Fortier B, Leriche MA,

Camus D, Dubremetz JF. Differential targeting of dense granule

proteins in the parasitophorous vacuole of Toxoplasma gondii. Parasi-

tology 1991;103:321–329.

48. Littman DR, Thomas Y, Maddon PJ, Chess L, Axel R. The isolation and

sequence of the gene encoding T8: a molecule defining functional

classes of T lymphocytes. Cell 1985;40:237–246.

49. Burg JL, Perelman D, Kasper LH, Ware PL, Boothroyd JC. Molecular

analysis of the gene encoding the major surface antigen of Toxoplasma

gondii. J Immunol 1988;141:3584–3591.

50. Saksouk N, Bhatti MM, Kieffer S, Smith AT, Musset K, Garin J, Sullivan

WJ, Cesbron-Delauw MF, Hakimi MA. Histone-modifying complexes

regulate gene expression pertinent to the differentiation of the pro-

tozoan parasite Toxoplasma gondii. Mol Cell Biol 2005;25:10301–

10314.

51. Bordier C. Phase separation of integral membrane proteins in Triton

X-114 solution. J Biol Chem 1981;256:1604–1607.

1680 Traffic 2008; 9: 1665–1680

Gendrin et al.