Coronaviruses Hijack the LC3-I-positive EDEMosomes, ER-derived vesicles exporting short-lived ERAD regulators, for replication.

Abstract

Coronaviruses (CoV), including SARS and mouse hepatitis virus (MHV), are enveloped RNA viruses that induce formation of double-membrane vesicles (DMVs) and target their replication and transcription complexes (RTCs) on the DMV-limiting membranes. The DMV biogenesis has been connected with the early secretory pathway. CoV-induced DMVs, however, lack conventional endoplasmic reticulum (ER) or Golgi protein markers, leaving their membrane origins in question. We show that MHV co-opts the host cell machinery for COPII-independent vesicular ER export of a short-living regulator of ER-associated degradation (ERAD), EDEM1, to derive cellular membranes for replication. MHV infection causes accumulation of EDEM1 and OS-9, another short-living ER chaperone, in the DMVs. DMVs are coated with the nonlipidated LC3/Atg8 autophagy marker. Downregulation of LC3, but not inactivation of host cell autophagy, protects cells from CoV infection. Our study identifies the host cellular pathway hijacked for supplying CoV replication membranes and describes an autophagy-independent role for nonlipidated LC3-I.

Full text

Page 1

Cell Host & Microbe
Short Article
Coronaviruses Hijack the LC3-I-Positive
EDEMosomes, ER-Derived Vesicles Exporting
Short-Lived ERAD Regulators, for Replication
Fulvio Reggiori,1,6,* Iryna Monastyrska,1,6Monique H. Verheije,2,6Tito Calı `,3,4Mustafa Ulasli,1Siro Bianchi,3
Riccardo Bernasconi,3Cornelis A.M. de Haan,2,* and Maurizio Molinari3,5,*
1Department of Cell Biology, University Medical Centre Utrecht, 3584 CX Utrecht, The Netherlands
2Virology Division, Department of Infectious Diseases and Immunology, Utrecht University, 3508 TC Utrecht, The Netherlands
3Institute for Research in Biomedicine, 6500 Bellinzona, Switzerland
4Department of Biochemistry, University of Padova, 35122 Padova, Italy
5Ecole Polytechnique Fe ´de ´rale de Lausanne, School of Life Sciences, 1015 Lausanne, Switzerland
6These authors contributed equally to this work
*Correspondence: f.reggiori@umcutrecht.nl (F.R.), c.a.m.dehaan@uu.nl (C.A.M.d.H.), maurizio.molinari@irb.unisi.ch (M.M.)
DOI 10.1016/j.chom.2010.05.013
SUMMARY
Coronaviruses (CoV), including SARS and mouse
hepatitis virus (MHV), are enveloped RNA viruses that
induceformationofdouble-membrane
(DMVs) and target their replication and transcription
complexes (RTCs) on the DMV-limiting membranes.
The DMV biogenesis has been connected with the
early secretory pathway. CoV-induced DMVs, how-
ever, lack conventional endoplasmic reticulum (ER)
or Golgi protein markers, leaving their membrane
origins in question. We show that MHV co-opts the
host cell machinery for COPII-independent vesicular
ER export of a short-living regulator of ER-associated
degradation (ERAD), EDEM1, to derive cellular mem-
branes for replication. MHV infection causes accumu-
lation of EDEM1 and OS-9, another short-living ER
chaperone, in the DMVs. DMVs are coated with the
nonlipidatedLC3/Atg8 autophagy marker. Downregu-
lation of LC3, but not inactivation of host cell autoph-
agy, protects cells from CoV infection. Our study
identifies the host cellular pathway hijacked for
supplying CoV replication membranes and describes
anautophagy-independentrolefornonlipidatedLC3-I.
vesicles
INTRODUCTION
Asanearlyandcrucialeventuponinfection,coronaviruses(CoV)
such as the severe acute respiratory syndrome (SARS)-CoV and
mouse hepatitis virus (MHV) induce the formation of double-
membrane vesicles (DMVs) in host cells and target their RTCs
on the limiting membranes of these structures (de Haan and
Reggiori, 2008; Gosert et al., 2002; Miller and Krijnse-Locker,
2008; Salonen et al., 2005). A recent analysis of SARS-CoV-
and MHV-infected cells by electron tomography has revealed
that these DMVs are part of a reticular network of modified
endoplasmic reticulum (ER) membranes and contain double-
stranded RNA (dsRNA) in their interior (Knoops et al., 2008).
The ER origin of the DMVs is supported by the finding that two
nonstructural proteins (nsp3 and nsp4) with transmembrane
segments that are part of the RTCs become N-glycosylated
(Harcourt et al., 2004; Kanjanahaluethai et al., 2007; Oostra
et al., 2008). Moreover, nsp4 localizes to the ER when separately
expressed and moves to the DMVs upon viral infection (Oostra
et al., 2007). Whereas previous work has shown that the early
secretory pathway and CoV-induced DMV biogenesis are
closely connected (Knoops et al., 2010; Oostra et al., 2007; Ver-
heijeetal.,2008),thelackofER,ER-Golgiintermediatecompart-
ment (ERGIC), or Golgi protein markers in CoV-induced DMVs
suggests that their biogenesis does not depend on the conven-
tional routing of proteins through this transport pathway (Oostra
et al., 2007; Snijder et al., 2006; Verheije et al., 2008). Thus, even
though CoV must hijack ER-derived host cell membranes for
replication, the precise origin of the DMV lipid bilayers, the
host protein content, and the identity of the cellular factors
essential for DMV formation remain mysterious (Knoops et al.,
2008). The possible involvement of the autophagy machinery in
the conversion of host membranes into DMVs has been
reported. However, whereas Atg5, an essential component of
the autophagy machinery (Mizushima et al., 2001; Yoshimori
and Noda, 2008), has been shown to be dispensable for MHV
replication (Zhao et al., 2007), contradictory immunofluores-
cence (IF) data report the presence (Prentice et al., 2004; Zhao
et al., 2007) or the absence (de Haan and Reggiori, 2008; Snijder
et al., 2006) of the autophagosome protein marker LC3/Atg8 on
DMVs (de Haan and Reggiori, 2008). In this study, we have used
MHV, the virus prototype for the CoV biology investigation, to
unveil the origin of the virus-induced DMVs and to identify host
cell factors essential for CoV replication.
In the ER, newly synthesized, unstructured polypeptides can
attract the folding as well as the degradation machineries that
are operating in the lumen of this organelle (Calı ` et al., 2008b;
Hebert and Molinari, 2007). Under normal growth conditions,
therefore, the activity of the ERAD machinery must be main-
tained low to avoid premature interruption of folding programs
and favor attainment of the native structure over degradation
of immature polypeptides (Calı ` et al., 2008a). Consistently,
it has been reported that EDEM1, a crucial regulator of ERAD
(Molinari et al., 2003; Oda et al., 2003), is selectively cleared
500 Cell Host & Microbe 7, 500–508, June 17, 2010 ª2010 Elsevier Inc.

Page 2

from the ER to tune down the ERAD activity (Calı ` et al., 2008a).
This regulation mechanism has been named ERAD tuning. It
relies on the selective sorting of EDEM1 and probably other
short-living ER chaperones in 200–800 nm vesicles called EDE-
Mosomes. These vesicles, coated with nonlipidated LC3/Atg8
(LC3-I) (Calı ` et al., 2008a, 2008b), emerge from the ER through
a COPII coat-independent mechanism (Zuber et al., 2007) and
deliver their content to endosomal compartments for disposal
in a series of poorly characterized events (Calı ` et al., 2008a,
2008b; Le Fourn et al., 2009; Zuber et al., 2007).
We have discovered that MHV exploits the pathway of EDE-
Mosome formation to generate the DMVs required for viral repli-
cation. In doing so, MHV interferes with the degradation of
EDEM1 and OS-9, another short-living chaperone that we iden-
tify here as a second EDEMosome cargo, by trapping them into
theDMVs.OurdataareconsistentwiththemodelinwhichDMVs
are generated from the host ER (Knoops et al., 2008) and explain
the absence of conventional ER resident chaperones in their
interior. In addition, we show that, whereas the autophagy
pathway is not essential for MHV infection, nonlipidated LC3-I
coats CoV-induced DMVs and is essential for MHV replication.
RESULTS AND DISCUSSION
MHV Infection Does Not Require an Intact Autophagy
Machinery
To conclusively establish whether autophagy is required for MHV
infection, we assessed the consequences of deleting ATG7,
a gene essential for autophagy (Komatsu et al., 2005), on DMV
biogenesis. IF analysis of two RTC components, nsp2 and
nsp3, in MHV-infected wild-type (Atg7+/+) and ATG7 knockout
mouse embryonic fibroblasts (Atg7?/?MEF) revealed that these
two proteins were similarly distributed to numerous punctuate
structures in both cell lines (Figure 1A). These puncta represent
the virus-induced DMVs that contain viral dsRNA (colocalization
in Figure 1A; Harcourt et al., 2004; Knoops et al., 2008). The
presence in both Atg7+/+and Atg7?/?MEF of these double-
membrane structures with a diameter of approximately 200–350
nm (Knoops et al., 2008; Stertz et al., 2007) was confirmed by
conventional electron microscopy (Figure 1B). Moreover, cells
withandwithoutATG7wereequallysusceptibletoMHVinfection.
Thiswasestablishedbyassessingthevirusreplicationusingare-
combinant luciferase-expressing MHV (Figure 1C) and by deter-
mining the titer of a virus stock on these cells by using the mean
tissue culture infection dose (TCID50) test (Figure 1D). MHV repli-
cation in Atg7+/+and Atg7?/?MEF during the course of an
infection was also very similar (Figure 1E). Thus, the conventional
host cellautophagy is not required for formation of MHV-induced
DMVs, nor for viral replication and production of viral progeny.
Nonlipidated LC3/Atg8 Associates with CoV-Induced
DMVs
Because contrasting data have been published on the presence
(Prentice et al., 2004; Zhao et al., 2007) or the absence (de Haan
and Reggiori, 2008; Snijder et al., 2006) of LC3/Atg8 on CoV-
induced DMVs, we examined this issue. Our analysis by IF
showed that endogenous LC3 extensively colocalized with the
DMV protein markers nsp2 and nsp3 (Figures 2A and 2C). This
colocalization was observed during the entire course of the
MHV infection (Figures S1A and S1B available online). In
contrast, ectopically expressed GFP-LC3, a conventional
protein marker for autophagosome membranes (Klionsky et al.,
2008; Mizushima et al., 2004), did not colocalize with nsp2 and
nsp3 (Figures 2B and 2C). These apparently conflicting data
were confirmed in other cell lines (e.g., in HeLa cells; Figures
S1CandS1D)andexplainthecontradictorydataintheliterature.
LC3 is present in the cell predominantly in a cytoplasmic form
(LC3-I)that,uponautophagyinduction,isconvertedintoanactive
lipidated form (LC3-II) by specific covalent linkage to the phos-
phatidylethanolamine present on autophagosomal membranes
(Klionsky et al., 2008; Mizushima et al., 2004). The lipidation of
LC3-I and the formation of LC3-II-coated autophagosomes
require several proteins, including Atg7 (Komatsu et al., 2005).
However, Atg7 was not necessary for the association of endoge-
nous LC3 to DMVs (Figure 2D and 2E). Consequently, Atg7 and
LC3 lipidation are not essential for the formation of LC3-positive
DMVs. Analysis of the protein content in DMVs induced upon
MHV-nsp2GFP infection of HeLa cells and separated on contin-
uous density gradients as described (Calı ` et al., 2008a) showed
the presence of LC3-I in the denser fractions containing the
DMV protein marker nsp2-GFP (Figure 2F). These fractions
wereclearlyseparatedfromthelighterautophagosomescontain-
ingLC3-IIthatfloatedatthetopofthegradient.Moreover,IFanal-
yses revealed that ectopically expressed C-terminally HA-
tagged, nonlipidable LC3 localizes on DMVs (Figure 2G). Taken
all together, these data show that an intact host autophagy
machineryisdispensablefortheviruslifecycleandthatlipidation
is not required for LC3 association with the DMV membranes.
Analogies between MHV-Induced DMVs
and EDEMosomes
The ER origin of the MHV-induced DMVs (Knoops et al., 2008;
Oostra et al., 2007), the absence of conventional ER markers in
their membranes and lumen (Oostra et al., 2007; Snijder et al.,
2006; Verheije et al., 2008), their association with LC3-I, and
the fact that DMVs are stained with antibodies against endoge-
nous LC3, but not with ectopically expressed GFP-LC3 (Figures
2, S1C, and S1D) are features reminiscent of those describing
the EDEMosomes (Calı ` et al., 2008a). Significantly, the LC3-I
coat distinguishes DMVs (this study) and EDEMosomes (Calı `
et al., 2008a; Figure S3D) from autophagosomes, which are
associated with LC3-II and can be decorated with GFP-LC3
(Klionsky et al., 2008; Mizushima et al., 2004).
As in the case of formation of the ER-derived, CoV-induced
DMVs, it is unclear whether an active autophagy machinery is
required for formation of ER-derived EDEMosomes and/or for
disposal of EDEM1 (Calı ` et al., 2008a versus Le Fourn et al.,
2009). To better understand this, we compared variations in the
intracellular levels of EDEM1 and of p62, a canonical substrate of
autophagy(Bjørkøyetal.,2005),underconditionsthateitherinac-
tivate, e.g., ATG7 deletion (Figures 3A and 3B) or cell incubation
with chloroquine (CQ) (Figures 3C and 3D) (Klionsky et al., 2008;
Komatsuetal.,2005)oractivateautophagy,e.g.,rapamycintreat-
ment (Figures 3C and 3D) (Klionsky et al., 2008). Deletion of ATG7
inhibits the p62 turnover (Waguri and Komatsu, 2009), thus
substantially increasing the intracellular level of this autophagy
substrate (Figure 3A). On the contrary, deletion of ATG7 did not
result in substantial variations of the level of EDEM1 (Figure 3A),
Cell Host & Microbe
Host Cell Factors for Coronavirus Replication
Cell Host & Microbe 7, 500–508, June 17, 2010 ª2010 Elsevier Inc. 501

Page 3

thus showing that Atg7 and conventional autophagy are
dispensable for EDEM1 turnover. Consistently, when wild-type
and Atg7?/?MEF were metabolically radiolabeled and chased
for 10–90 min, the amount of residual endogenous EDEM1
decreased with similar kinetics in the two cell lines, confirming
B
Atg7+/+
Atg7-/-
CD
0
40
80
120
160
Atg7+/+
Atg7-/-
Relative luciferase
expression [%]
0
2
4
6
Atg7+/+
Atg7-/-
TCID50/ml [log10]
A
dsRNAnsp2/nsp3 mergedinset
Atg7+/+
Atg7-/-
E
5
0
2
6
10
14
18
67
Time [h]
Relative luciferase units [log10]
89 10
Atg7+/+
Atg7-/-
Figure 1. MHV Replication, Assembly, and
Release Is Autophagy Independent
(A) Atg7+/+and Atg7?/?MEF were infected with
MHV before being processed for IF at 7 hr postin-
fection (p.i.). The RTC components nsp2 and nsp3
colocalize with dsRNA to cytoplasmic puncta in
both wild-type and knockout cells.
(B) DMVs have normal morphology in autophagy-
deficient cells. Cells analyzed in (A) were also pro-
cessed for EM. DMVs were absent in noninfected
cells (data not shown). Arrowheads highlight
DMVs. Scale bar, 500 nm.
(C) The Atg7+/+and Atg7?/?MEF were inoculated
with recombinant MHV-2aFLS, and luciferase
expression was measured at 7 hr p.i. The graph
showstherelativeluciferaseexpressioncompared
to Atg7+/+cells after correction for cell viability
from an experiment performed in triplicate.
(D) End point 10-fold dilutions of an MHV stock
were titrated on Atg7+/+and Atg7?/?MEF. Similar
titers were observed indicating that ATG7 deletion
doesnot affectMHV
assembly. Values presented in the graph are
calculated and expressed as the log10of TCID50
units per ml of supernatant, and the plotted values
represent the average of two experiments.
(E) The experiment described in (C) was repeated
inatimecourse manner,and luciferase expression
was determined at 5, 7, and 10 hr p.i.
Error bars in (C–E) indicate the standard deviations
between experiments.
entry,replication,or
similar rate of disposal in the presence or
absenceofAtg7(Figure3B).Cellexposure
to CQ inhibited the p62 degradation,
resulting in higher levels of this protein, as
expected for an autophagy substrate
(Figure 3C; Bjørkøy et al., 2009). CQ de-
layed EDEM1 turnover and resulted in
intracellular accumulation
(Figures 3C and 3D; Calı ` et al., 2008a).
Finally, induction of autophagy with rapa-
mycin reduced the intracellular levels of
p62, as expected for an autophagy
substrate (Figure 3C; Bjørkøy et al., 2009),
but increased those of EDEM1 (Figure 3C)
by delaying its turnover (Figure 3D). These
results confirmed that, as reported above
forCoVreplication,thepathwayregulating
EDEM1 turnover is clearly distinct from
autophagy. They also highlight another
analogy between
pathway and the CoV infection. Cell expo-
sure to the autophagy-inducer rapamycin
negatively affected both EDEM1 turnover
(Figures 3C and 3D) and MHV replication,
of EDEM1
the ERADtuning
as shown by measuring the levels of both the N nucleocapsid
levels and luciferase in cells infected with the recombinant lucif-
erase-expressing MHV (Figure S2). All together, these observa-
tions led us to hypothesize that MHV hijacks the ERAD tuning
machinery to co-opt cellular membranes for DMV generation.
Cell Host & Microbe
Host Cell Factors for Coronavirus Replication
502 Cell Host & Microbe 7, 500–508, June 17, 2010 ª2010 Elsevier Inc.

Page 4

MHV Infection Interferes with ERAD Tuning and Results
inAccumulationofERADTuningSubstratesintheVirus-
Induced DMVs
Asmentionedabove,despitetheirERorigin(Knoopsetal.,2008;
Zuber et al., 2007), CoV-induced DMVs and EDEMosomes do
not contain conventional ER chaperones or protein markers of
the early compartments of the secretory pathway (Calı ` et al.,
2008a; Oostra et al., 2007; Snijder et al., 2006; Stertz et al.,
2007; Verheije et al., 2008). By IF analysis of MHV-infected
wild-type and Atg7?/?
knockout cells, we systematically
assessed the presence in DMVs of several conventional protein
markers of the early compartment of the secretory pathway,
essentially confirming that none of them were in the virus-
induced DMVs (Figure S3A and data not shown). Only two anti-
bodies stained the virus-induced dsRNA- or nsp2/nsp3-positive
compartments, namely those recognizing EDEM1 and OS-9
A
B
D
C
E
FG
endogenous LC3
Endogenous
LC3
Atg7+/+
Atg7-/-
0
0
20
40
60
80
100
10
20
30
40
50
60
70
GFP-LC3
nsp2/nsp3 mergedinset
GFP-LC3nsp2/nsp3mergedinset
*
endogenous LC3nsp2/nsp3 mergedinset
LC3 puncta co-localizing
with DMVs [%]
Atg7-/-
Atg7+/+
2 3 4 5 6 7 8 9 101
LC3-I
LC3-II
DMVs autophagosomes
LC3-HA nsp2/nsp3 mergedinset
nsp2-GFP
LC3-I
EDEM1
LC3 puncta co-localizing
with DMVs [%]
Figure 2. Autophagy-Independent Recruitment of LC3-I onto MHV-Induced DMVs
(A and B) HEK293 cells stably transfected (B) or not (A) with a plasmid expressing GFP-LC3 were infected with MHV-Srec and processed for IF at 7 hr p.i.
(C) Summary statistics of the samples shown in (A) and (B) expressed as the percentage of LC3 or GFP-LC3 puncta colocalizing with the nsp2/nsp3 signals.Error
bars represent the standard error of the mean percentage from counting 40 cells in three independent experiments. The asterisk indicates that the two samples
are significantly different (tdf = 78= 10.4; p < 0.00001).
(D) Atg7+/+and Atg7?/?MEF were infected with MHV before being processed for IF at 7 hr p.i.
(E) Statistical analysis of the samples shown in (D) performed as described in (C).
(F) HeLa-CEACAM1a cells were infected with MHV-nsp2GFP for 7 hr before fractionating a cell extract on a continuous Optiprep gradient. Ten fractions were
collected from the top to the bottom of the gradient and probed with antibodies against EDEM1, GFP, and LC3. The fractionation profile was confirmed by per-
forming this experiment three times.
(G) HeLa-CEACAM1a cells were transiently transfected with a plasmid expressing C-terminally HA-tagged, nonlipidable LC3 before being infected with MHV.
Cells were fixed at 7 hr p.i. and processed for IF. DMVs and LC3-HA were detected with antibodies against nps2/nsp3 and HA, respectively.
See also Figure S1.
Cell Host & Microbe
Host Cell Factors for Coronavirus Replication
Cell Host & Microbe 7, 500–508, June 17, 2010 ª2010 Elsevier Inc. 503

Page 5

A
E
CD
dsRNA
dsRNAOS-9 mergedinset
dsRNA
Atg7+/+
Atg7+/+
Atg7-/-
Atg7-/-
OS-9mergedinset
EDEM1merged inset
Atg7
+/+
mock CQrap
-/-
EDEM1
p62
LC3-I
LC3-II
tubulin
H
EDEM1
OS-9.1
tubulin
12
EDEM1
p62
LC3-I
LC3-II
tubulin
10’60’90’10’60’90’
EDEM1
100 85 48
100 8142
10’60’ 120‘
120’
CQ
EDEM1
1006329 87
30’60’90‘
90’
Rap
100 573351
B
F
G
100 143
mock MHVmock MHV
12
123
100 151
OS-9.2
Atg7
+/+
100
-/-
78
mock CQ rap
100126 120
Cell Host & Microbe
Host Cell Factors for Coronavirus Replication
504 Cell Host & Microbe 7, 500–508, June 17, 2010 ª2010 Elsevier Inc.

Page 6

(Figures 3E and 3F). The presence of EDEM1 during the entire
course of the infection (Figures S3B and S3C) confirmed our
working model postulating that CoV hijack the ERAD tuning
pathway to promote formation of their DMVs. Our model was
further corroborated by the observation that cell infection with
MHV substantially interfered with ERAD tuning, thus causing
intracellular accumulation of EDEM1 and OS-9 (Figure 3H) due
to the relocalization and confinement of these chaperones into
the lasting MHV-induced DMVs (Figures 2F, 3E, 3F, S3B, and
S3C). In noninfected cells, EDEM1 has a half-life of about
60 min (Figures 3B and 3D; Calı ` et al., 2008a). Because MHV
infection induces a host translational shutoff (Raaben et al.,
2007), the persistence of EDEM1 (and OS-9) for several hours
confirms the defective clearance of this protein in infected cells
(Figures 3H, S3B, and S3C). Entrapping of a fraction of cellular
OS-9 in the DMVs was unexpected but significant. Like
EDEM1, OS-9 is a regulator of protein disposal from the ER
and it is expressed in two splice variants, OS-9.1 and OS-9.2
(Bernasconi et al., 2008; Christianson et al., 2008). Immunode-
tection of OS-9 in the MHV-induced DMVs (Figures 3F and 3G)
led us to verify by subcellular fractionation whether OS-9 local-
izes to the same LC3-I-positive vesicular structures containing
EDEM1 in noninfected cells (Calı ` et al., 2008a). Our analysis
showed that, atsteady state, about 20%of OS-9.1 and a smaller
amount of OS-9.2 were indeed found in LC3-I-positive EDEMo-
somes, which sedimented in fractions 7–9 of a continuous Opti-
prep gradient, whereas other conventional ER chaperones were
fully excluded from these fractions (Figure S3D; Calı ` et al.,
2008a).
Despite their presence in DMVs, EDEM1 and OS-9 are not
required for MHV replication. In fact, MHV infectivity was not
affected by their knockdown, as measured by examining the
synthesis of the viral structural N protein, the DMV formation
by IF, and by assessing virus replication by either determining
the TCID50value of a virus stock on these cells or assaying the
luciferase expression levels at different p.i. time points
(FigureS4).Takentogether, these resultssupport the hypothesis
that CoV hijack the ERAD tuning machinery to generate the repli-
cative DMVs.
LC3-I Is Required for CoV Replication
The association of LC3-I with MHV-induced DMVs is supported
by data showing that these vesicles are decorated with anti-LC3
antibodies in cells lacking Atg7 and LC3-II (Figure 2D), they co-
sediment with LC3-I, but not LC3-II, in density gradients
(Figure 2F), and they colocalize with LC3-HA (Figure 2G). We
therefore verified whether LC3 is required for viral replication
even though the autophagy protein Atg7, which is required for
its covalent membrane association, is dispensable (Figure 1).
Efficient depletion of LC3A and LC3B was obtained by specific
RNA interference (Figure 4A). LC3 downregulation protected
cells from MHV infection as assessed by inhibition of the
synthesisofthestructuralNprotein(Figure4A)andbyasubstan-
tial decrease in luciferase levels measured in both single time
point and time-course experiments (Figures 4B and 4C). The
inhibition of MHV replication observed in LC3 knockdown cells
was caused by a defect in DMV biogenesis, as no nsp2 and
nsp3 signal was detected by IF (Figure 4D). The initial DMVs
and RTCs generated upon MHV infection are necessary for the
massive synthesis of extra nsp proteins, which, in turn, leads
to the formation of a multitude of additional DMVs and RTCs.
Consequently, a defect in DMV biogenesis results in a severe
block of nsp production. Crucially, back transfection of LC3
knockdown cells with the plasmid-expressing C-terminally HA-
tagged, nonlipidable LC3 restored MHV replication measured
by monitoring N protein synthesis (Figure 4E).
During evolution, pathogens have developed molecular
devices to exploit conserved cellular pathways to optimally
infect, replicate, and/or leave host cells. Until now, no host
protein has been directly implicated in CoV replication. The
fact that both CoV-induced DMV formation and EDEM1/OS-9
turnover rely on analog mechanisms leads us to postulate that
CoV hijack the ERAD tuning machinery for the generation of
DMVs, which provide the membranous support for viral RTCs.
Based on our data, our current working model is that one or
more CoV nsps do associate with a still elusive EDEMosome
cargoreceptor that normally mediates segregation and vesicular
export from the ER of EDEM1 and OS-9. Accordingly to what is
known about other vesicular transport pathways (Sato and
Nakano, 2007), we postulate that the EDEMosome cargo
receptor interacts with subunits of a cytosolic vesicle protein
coat (nonlipidated LC3-I seems a good candidate). It is unlikely
that the viral nsps recruit LC3-I at the cytosolic surface of
DMVs because LC3-I is associated to EDEMosomes in unin-
fected cells (Calı ` et al., 2008a) and because we have not found
direct interaction between nsp and LC3-I (unpublished data).
Astonishingly, whereas EDEMosomes are transient structures
that end their journey in late endosomes and/or lysosomes,
MHV-induced DMVs are persistent cytoplasmic organelles
(Ulasli et al., in press). Therefore, the presence of either nsps or
Figure 3. Components of the ERAD Tuning Pathway Are Associated with DMVs
(A) Cell extracts from Atg7+/+(lane 1) and Atg7?/?(lane 2) were separated by SDS-PAGE and western blot membranes probed with antibodies against EDEM1,
p62, LC3, and tubulin. Repetition of the analysis showed no significant differences in the EDEM1 level in Atg7+/+versus Atg7?/?MEF. The percentages (right)
indicate the relative EDEM1 levels in the knockout cells compared to wild-type cells and represent the average of two experiments.
(B) Atg7+/+and Atg7?/?MEF were metabolically labeled and chased for the times indicated before lysis and EDEM1 immuno-isolation. The residual radiolabeled
EDEM1present ineachlanewasquantified and indicated beloweachband.Repetition oftheanalysisshowednosignificantdifferences intheEDEM1turnoverin
Atg7+/+versus Atg7?/?MEF.
(C) Atg7+/+MEFwere untreated or treated with100mM CQor 1mMrapamycin (rap) for 4hrbefore preparation of cell extracts and analysisas in(A). Thepercent-
ages (right) indicate the relative EDEM1 levels in drug-treated cells compared to mock-treated cells and represent the average of two experiments.
(D) Same as (B) to confirm, in a pulse-chase radiolabeling experiment, that CQ and rap delay EDEM1 turnover.
(E and F) HeLa cells were infected with MHV-Srec and processed for IF at 7 hr p.i. using antibodies against dsRNA and (E) EDEM1 or (F) OS-9.
(G) Atg7+/+and Atg7?/?MEF infected with MHV-Srec were fixed at 7 hr p.i. and processed for IF using antibodies against OS-9 and dsRNA.
(H) Cell extracts were analyzed by western blot using antibodies against EDEM1 or OS-9. The percentages (right) indicate the relative EDEM1 and OS-9 levels in
infected cells compared to control cells and represent the average of two experiments.
See also Figures S2, S3, and S4.
Cell Host & Microbe
Host Cell Factors for Coronavirus Replication
Cell Host & Microbe 7, 500–508, June 17, 2010 ª2010 Elsevier Inc. 505

Page 7

AB
N protein [%]
0
20
40
60
80
100
120
siLC3
siSCR
N protein
LC3-I
LC3-II
actin
Relative luciferase
expression [%]
0
20
40
60
80
100
120
140
siSCRsiLC3
siSCR
siLC3
D
nsp2/nsp3
LC3
merged
siSCR
siLC3
C
Relative luciferase expression [%]
0
57
Time [h]
10
20
40
60
80
100
120
140
160
E
N protein
actin
LC3-I
LC3-II
LC3-HA
siLC3
siSCR
siLC3
+ LC3-HA
Figure 4. LC3-I Is Required for the Formation of DMVs
(A and B) HeLa-CEACAM1a cells were transfected with either siRNA directed against LC3A and LC3B (siLC3) or nontargeting (siSCR) siRNA. After 48 hr, cells
were infected with MHV-2aFLS and analyzed at 7 hr p.i. This analysis has been repeated five times.
(A) Celllysateswereanalyzedby westernblotusingantibodies againsttheNprotein,LC3,andactin.Thepercentage indicates therelativeNproteinlevelsinLC3-
depleted cells compared to control cells and represent the average of three experiments.
(B) Luciferase activity was measured as described in the Experimental Procedures, and error bars indicate the standard deviations for an experiment made in
triplicate.
(C) The experiment described in (B) was repeated three times in a time course manner, and luciferase expression was determined at 5, 7, and 10 hr p.i. Error bars
represent the standard deviations between experiments.
(D) Localization of nsp2/nsp3 and LC3 was examined by IF in cells treated as in (A) and (B).
(E) HeLa-CEACAM1a cells were transfected with either siRNA directed against LC3A and LC3B (siLC3) or nontargeting (siSCR) siRNA. After 24 hr, one of the
samples was transfected with a plasmid expressing C-terminally HA-tagged, nonlipidable LC3, and MHV was inoculated at 48 hr. Cell lysates were prepared
at 7 hr p.i. and analyzed by western blot using antibodies against the N protein, LC3, HA, and actin. Note that the higher levels of LC3-I in the siLC3+LC3-HA
lane are due to LC3-HA, which run as LC3-I in the used SDS-PAGE gel.
Cell Host & Microbe
Host Cell Factors for Coronavirus Replication
506 Cell Host & Microbe 7, 500–508, June 17, 2010 ª2010 Elsevier Inc.

Page 8

other viral products inhibits the fusion of EDEMosomes/DMVs
with the degradative compartments of the endosomal system.
A main goal of future research is the identification of the elusive
EDEMosome cargo receptor and the characterization of its
involvement in CoV infection. It also remains to be established
whether the ERAD tuning and CoV infection share mechanistic
analogies with the recently characterized Atg5/Atg7-indepen-
dent type of autophagy (Nishida et al., 2009). Importantly, in
our study, we also show that blocking DMV formation by
depleting LC3 severely impairs MHV infection. Consequently,
future investigations on the involvement of viral proteins and
ERAD tuning components, as well as the unconventional inter-
vention of certain factors of the host cell autophagy machinery
in the formation of CoV-induced DMVs, might provide valuable
therapeutic targets.
EXPERIMENTAL PROCEDURES
Cell Culture, Viruses, and Expression Plasmid
Cell culture, viruses, construction of the expression plasmid, and statistical
analyses are described in the Supplemental Experimental Procedures.
Gene Silencing with siRNA
Sets of three different siRNA duplexes targeting different sites within the
coding sequences of LC3A and LC3B were designed by and obtained from
Applied Biosystems/Ambion (Nieuwerkerk a/d lJssel). One day after seeding,
the HeLa-CEACAM1a cells were transfected with 40 nM of siRNA using Lipo-
fectamine 2000. Experiments were conducted at 48 hr posttransfection. Anti-
bodies against LC3 (NanoTools) were used to assess by western blot the
depletion of the targeted gene.
Immunofluorescence Microscopy
Cells were grown and processed for IF on coverslides as described (Verheije
et al., 2008). Fluorescence signals were visualized with a DeltaVision RT fluo-
rescence microscope equipped with a CoolSnap camera (Applied Precision).
Images were generated by collecting a stack of 20 pictures with focal planes
0.20 mm apart in order to cover the entire volume of the cell and by subse-
quently deconvolving them using the SoftWoRx software (Applied Precision).
A single focal plane is shown at each time. Primary immunological reactions
were carried out with polyclonal anti-EDEM1 (Sigma), anti-OS-9 (Novus Bio-
logicals),anti-nsp2andnsp3(Schilleretal.,1998),andanti-Sec23(AffinityBio-
reagents)antiseraandmonoclonalanti-LC3(NanoTools), anti-HA(akindgiftof
G. Bu, Washington University), anti-dsRNA K1 (English and Scientific Consul-
ting Bt.), anti-KDEL (Calbiochem) and anti-ERGIC53 (Alexis Biochemicals)
antibodies.Allexperimentswererepeatedtwoorthreetimes.Selectedimages
show a representative fluorescence profile.
Western Blots
The cell extracts wereprepared with lysis buffer (200 mMNaCl, 50mM HEPES
[pH 6.8], 2% CHAPS, and protease inhibitors) and boiled for 5 min in sample
buffer. Proteins were separated on a SDS-PAGE gel and transferred onto
a PVDF membrane before analysis with antibodies against EDEM1, OS-9,
N protein (a gift of S. Siddell, University of Bristol), Erp57 (Solda ` et al., 2006),
calnexin (a gift of A. Helenius, ETH Zurich), p62 (Progen Biotechnik), mouse
BiP (Stressgen), tubulin (Applied Biological Materials), and actin (MP Biomed-
icals). Alexa Fluor680-conjugated goat anti-rabbit or rabbit anti-mouse
secondary antibodies (Molecular Probes) were used for the visualization of
the immunoblots using an Odyssey system (Li-Cor Biosciences). Tubulin
and actin represent the loading controls.
Miscellaneous Procedures
Electron microscopy, quantification of the virus replication using firefly lucif-
erase-carrying virions MHV-2aFLS, measurement of the TCID50, subcellular
fractionation on Optiprep gradients, and pulse-chase radiolabeling experi-
ments followed by immunoprecipitations are described in Calı ` et al., 2008a;
Slot and Geuze, 2007; Verheije et al., 2008.
SUPPLEMENTAL INFORMATION
Supplemental Information includes Supplemental Experimental Procedures
and four figures and can be found with this article online at doi:10.1016/j.
chom.2010.05.013.
ACKNOWLEDGMENTS
TheauthorsthankS.Baker,A.Helenius,K.Kirkegaard,M.Komatsu,S.Siddell,
and S. Tooze for reagents; P. van Kerkhof and E. te Lintelo for technical
advices;and I.Pen forguidanceon thestatistics. C.A.M.d.H.and F.R. aresup-
ported by the Utrecht University (High Potential grant). M.M. is supported by
grants from the Foundation for Research on Neurodegenerative Diseases,
the Fondazione San Salvatore, the Swiss National Center of Competence in
Research on Neural Plasticityand Repair, the Swiss National Science Founda-
tion, and ONELIFE Advisors SA.
Received: October 27, 2009
Revised: April 9, 2010
Accepted: May 11, 2010
Published: June 16, 2010
REFERENCES
Bernasconi,R., Pertel, T.,Luban, J., and Molinari, M. (2008). A dual task for the
Xbp1-responsive OS-9 variants in the mammalian endoplasmic reticulum: in-
hibiting secretion of misfolded protein conformers and enhancing their
disposal. J. Biol. Chem. 283, 16446–16454.
Bjørkøy, G., Lamark, T., Brech, A., Outzen, H., Perander, M., Overvatn, A.,
Stenmark, H., and Johansen, T. (2005). p62/SQSTM1 forms protein aggre-
gates degraded by autophagy and has a protective effect on huntingtin-
induced cell death. J. Cell Biol. 171, 603–614.
Bjørkøy, G., Lamark, T., Pankiv, S., Øvervatn, A., Brech, A., and Johansen, T.
(2009). Monitoring autophagic degradation of p62/SQSTM1. Methods Enzy-
mol. 452, 181–197.
Calı `, T., Galli, C., Olivari, S., and Molinari, M. (2008a). Segregation and rapid
turnover of EDEM1 by an autophagy-like mechanism modulates standard
ERAD and folding activities. Biochem. Biophys. Res. Commun. 371, 405–410.
Calı `, T., Vanoni, O., and Molinari, M. (2008b). The endoplasmic reticulum
crossroads for newly synthesized polypeptide chains. Prog. Mol. Biol. Transl.
Sci. 83, 135–179.
Christianson, J.C., Shaler, T.A., Tyler, R.E., and Kopito, R.R. (2008). OS-9 and
GRP94 deliver mutant alpha1-antitrypsin to the Hrd1-SEL1L ubiquitin ligase
complex for ERAD. Nat. Cell Biol. 10, 272–282.
deHaan,C.A.,andReggiori,F.(2008).Arenidoviruseshijackingtheautophagy
machinery? Autophagy 4, 276–279.
Gosert, R., Kanjanahaluethai, A., Egger, D., Bienz, K., and Baker, S.C. (2002).
RNA replication of mouse hepatitis virus takes place at double-membrane
vesicles. J. Virol. 76, 3697–3708.
Harcourt, B.H., Jukneliene, D., Kanjanahaluethai, A., Bechill, J., Severson,
K.M., Smith, C.M., Rota, P.A., and Baker, S.C. (2004). Identification of severe
acute respiratory syndrome coronavirus replicase products and characteriza-
tion of papain-like protease activity. J. Virol. 78, 13600–13612.
Hebert, D.N., and Molinari, M. (2007). In and out of the ER: protein folding,
quality control, degradation, and related human diseases. Physiol. Rev. 87,
1377–1408.
Kanjanahaluethai, A., Chen, Z., Jukneliene, D., and Baker, S.C. (2007).
Membrane topology of murine coronavirus replicase nonstructural protein 3.
Virology 361, 391–401.
Klionsky, D.J., Abeliovich, H., Agostinis, P., Agrawal, D.K., Aliev, G., Askew,
D.S., Baba, M., Baehrecke, E.H., Bahr, B.A., Ballabio, A., et al. (2008).
Cell Host & Microbe
Host Cell Factors for Coronavirus Replication
Cell Host & Microbe 7, 500–508, June 17, 2010 ª2010 Elsevier Inc. 507

Page 9

Guidelines for the use and interpretation of assays for monitoring autophagy in
higher eukaryotes. Autophagy 4, 151–175.
Knoops, K., Kikkert, M., Worm, S.H., Zevenhoven-Dobbe, J.C., van der Meer,
Y., Koster, A.J., Mommaas, A.M., and Snijder, E.J. (2008). SARS-coronavirus
replication is supported by a reticulovesicular network of modified endo-
plasmic reticulum. PLoS Biol. 6, e226.
Knoops, K., Swett-Tapia, C., van den Worm, S.H., Te Velthuis, A.J., Koster,
A.J.,Mommaas,A.M.,Snijder,E.J.,andKikkert,M.(2010).Integrityoftheearly
secretory pathway promotes, but is not required for, severe acute respiratory
syndrome coronavirus RNA synthesis and virus-induced remodeling of endo-
plasmic reticulum membranes. J. Virol. 84, 833–846.
Komatsu, M., Waguri, S., Ueno, T., Iwata, J., Murata, S., Tanida, I., Ezaki, J.,
Mizushima, N., Ohsumi, Y., Uchiyama, Y., et al. (2005). Impairment of starva-
tion-induced and constitutive autophagy in Atg7-deficient mice. J. Cell Biol.
169, 425–434.
Le Fourn, V., Gaplovska-Kysela, K., Guhl, B., Santimaria, R., Zuber, C., and
Roth, J. (2009). Basal autophagy is involved in the degradation of the ERAD
component EDEM1. Cell. Mol. Life Sci. 66, 1434–1445.
Miller,S.,and Krijnse-Locker, J.(2008). Modificationof intracellular membrane
structures for virus replication. Nat. Rev. Microbiol. 6, 363–374.
Mizushima, N., Yamamoto, A., Hatano, M., Kobayashi, Y., Kabeya, Y., Suzuki,
K., Tokuhisa, T., Ohsumi, Y., and Yoshimori, T. (2001). Dissection of autopha-
gosome formation using Apg5-deficient mouse embryonic stem cells. J. Cell
Biol. 152, 657–668.
Mizushima, N., Yamamoto, A., Matsui, M., Yoshimori, T., and Ohsumi, Y.
(2004). In vivo analysis of autophagy in response to nutrient starvation using
transgenic mice expressing a fluorescent autophagosome marker. Mol. Biol.
Cell 15, 1101–1111.
Molinari, M., Calanca, V., Galli, C., Lucca, P., and Paganetti, P. (2003). Role of
EDEM in the release of misfolded glycoproteins from the calnexin cycle.
Science 299, 1397–1400.
Nishida, Y., Arakawa, S., Fujitani, K., Yamaguchi, H., Mizuta, T., Kanaseki, T.,
Komatsu, M., Otsu, K., Tsujimoto, Y., and Shimizu, S. (2009). Discovery of
Atg5/Atg7-independent alternative macroautophagy. Nature 461, 654–658.
Oda, Y., Hosokawa, N., Wada, I., and Nagata, K. (2003). EDEM as an acceptor
of terminally misfolded glycoproteins released from calnexin. Science 299,
1394–1397.
Oostra, M., te Lintelo, E.G., Deijs, M., Verheije, M.H., Rottier, P.J., and de
Haan, C.A. (2007). Localization and membrane topology of coronavirus
nonstructural protein 4: involvement of the early secretory pathway in replica-
tion. J. Virol. 81, 12323–12336.
Oostra, M., Hagemeijer, M.C., van Gent, M., Bekker, C.P., te Lintelo, E.G.,
Rottier, P.J., and de Haan, C.A. (2008). Topology and membrane anchoring
of the coronavirus replication complex: not all hydrophobic domains of nsp3
and nsp6 are membrane spanning. J. Virol. 82, 12392–12405.
Prentice, E., Jerome, W.G., Yoshimori, T., Mizushima, N., and Denison, M.R.
(2004). Coronavirus replication complex formation utilizes components of
cellular autophagy. J. Biol. Chem. 279, 10136–10141.
Raaben, M., Groot Koerkamp, M.J., Rottier, P.J., and de Haan, C.A. (2007).
Mouse hepatitis coronavirus replication induces host translational shutoff
and mRNA decay,withconcomitant formationof stress granulesand process-
ing bodies. Cell. Microbiol. 9, 2218–2229.
Salonen, A., Ahola, T., and Ka ¨a ¨ria ¨inen, L. (2005). Viral RNA replication in asso-
ciation with cellular membranes. Curr. Top. Microbiol.Immunol. 285, 139–173.
Sato, K., and Nakano, A. (2007). Mechanisms of COPII vesicle formation and
protein sorting. FEBS Lett. 581, 2076–2082.
Schiller, J.J., Kanjanahaluethai, A., and Baker, S.C. (1998). Processing of the
coronavirus MHV-JHM polymerase polyprotein: identification of precursors
and proteolytic products spanning 400 kilodaltons of ORF1a. Virology 242,
288–302.
Slot, J.W., and Geuze, H.J. (2007). Cryosectioning and immunolabeling. Nat.
Protoc. 2, 2480–2491.
Snijder, E.J., van der Meer, Y., Zevenhoven-Dobbe, J., Onderwater, J.J., van
der Meulen, J., Koerten, H.K., and Mommaas, A.M. (2006). Ultrastructure and
origin of membrane vesicles associated with the severe acute respiratory
syndrome coronavirus replication complex. J. Virol. 80, 5927–5940.
Solda `, T., Garbi, N., Ha ¨mmerling, G.J., and Molinari, M. (2006). Consequences
of ERp57 deletion on oxidative folding of obligate and facultative clients of the
calnexin cycle. J. Biol. Chem. 281, 6219–6226.
Stertz, S., Reichelt, M., Spiegel, M., Kuri, T., Martı ´nez-Sobrido, L., Garcı ´a-
Sastre,A.,Weber,F.,andKochs,G.(2007). Theintracellular sitesofearly repli-
cation and budding of SARS-coronavirus. Virology 361, 304–315.
Ulasli, M., Verheije, M.H., de Haan, C.A., and Reggiori, F. (2010). Qualitative
and quantitative ultrastructural analysis of the membrane rearrangements
induced by coronavirus. Cell. Microbiol. 12, 844–886.
Verheije, M.H., Raaben, M., Mari, M., Te Lintelo, E.G., Reggiori, F., van Kuppe-
veld, F.J., Rottier, P.J., and de Haan, C.A. (2008). Mouse hepatitis coronavirus
RNA replication depends on GBF1-mediated ARF1 activation. PLoS Pathog.
4, e1000088.
Waguri, S.,and Komatsu, M.(2009).Biochemicaland morphologicaldetection
of inclusion bodies in autophagy-deficient mice. Methods Enzymol. 453,
181–196.
Yoshimori, T., and Noda, T. (2008). Toward unraveling membrane biogenesis
in mammalian autophagy. Curr. Opin. Cell Biol. 20, 401–407.
Zhao, Z., Thackray, L.B., Miller, B.C., Lynn, T.M., Becker, M.M., Ward, E.,
Mizushima, N.N., Denison, M.R., and Virgin, H.W., 4th. (2007). Coronavirus
replicationdoesnotrequiretheautophagygeneATG5.Autophagy3,581–585.
Zuber, C., Cormier, J.H., Guhl, B., Santimaria, R., Hebert, D.N., and Roth, J.
(2007). EDEM1 reveals a quality control vesicular transport pathway out of
the endoplasmic reticulum not involving the COPII exit sites. Proc. Natl.
Acad. Sci. USA 104, 4407–4412.
Cell Host & Microbe
Host Cell Factors for Coronavirus Replication
508 Cell Host & Microbe 7, 500–508, June 17, 2010 ª2010 Elsevier Inc.