Zika Virus and Dengue Virus Genomics PDF

Resource
Identification of Zika Virus and Dengue Virus

Dependency Factors using Functional Genomics
Graphical Abstract
Authors
George Savidis, William M. McDougall,
Paul Meraner, ..., Sharone Green,
Timothy F. Kowalik, Abraham L. Brass
Correspondence
abraham.brass@umassmed.edu
In Brief
Savidis et al. identify DENV and ZIKV
dependencies using orthologous RNAi
and CRISPR/Cas9 approaches. Multiple
host factors involved in endocytosis and
transmembrane protein processing,
including the endoplasmic reticulum
membrane complex (EMC), are important
for flaviviral replication. Together, their
studies generate a systems-wide view of
human-flavivirus interactions.
Highlights
d
RNAi and CRISPR/Cas9 screens were used to find flavivirus

dependencies
The screens recovered host factors involved in endocytosis

and heparin sulfation
The EMC is required by DENV and ZIKV in the early stages of

replication
These studies give a systems-wide view of human-flavivirus

interactions
Savidis et al., 2016, Cell Reports 16, 115

July 19, 2016 2016 The Author(s).
http://dx.doi.org/10.1016/j.celrep.2016.06.028
Please cite this article in press as: Savidis et al., Identification of Zika Virus and Dengue Virus Dependency Factors using Functional Genomics, Cell
Reports (2016), http://dx.doi.org/10.1016/j.celrep.2016.06.028
Cell Reports
Resource
Identification of Zika Virus and Dengue Virus
George Savidis,1,5 William M. McDougall,1,5 Paul Meraner,1,5 Jill M. Perreira,1 Jocelyn M. Portmann,1 Gaia Trincucci,1
Sinu P. John,2 Aaron M. Aker,1 Nicholas Renzette,1 Douglas R. Robbins,1 Zhiru Guo,3 Sharone Green,3
Timothy F. Kowalik,1 and Abraham L. Brass1,4,*
1Department
of Microbiology and Physiological Systems (MaPS), University of Massachusetts Medical, School, Worcester, MA 01655, USA
Systems Unit, Laboratory of Systems Biology, National Institute of Allergy and Infectious Diseases, National Institutes of Health,
Bethesda, MD 20892, USA
3Division of Infectious Diseases and Immunology, Department of Medicine, University of Massachusetts Medical School, Worcester,
MA 01655, USA
4Division of Gastroenterology, Department of Medicine, University of Massachusetts Medical School, Worcester, MA 01655, USA
5Co-first author
*Correspondence: abraham.brass@umassmed.edu
http://dx.doi.org/10.1016/j.celrep.2016.06.028
2Signaling
SUMMARY
The flaviviruses dengue virus (DENV) and Zika virus

(ZIKV) are severe health threats with rapidly expanding ranges. To identify the host cell dependencies of
DENV and ZIKV, we completed orthologous functional genomic screens using RNAi and CRISPR/
Cas9 approaches. The screens recovered the ZIKV
entry factor AXL as well as multiple host factors
involved in endocytosis (RAB5C and RABGEF), heparin sulfation (NDST1 and EXT1), and transmembrane protein processing and maturation, including
the endoplasmic reticulum membrane complex
(EMC). We find that both flaviviruses require the
EMC for their early stages of infection. Together,
these studies generate a high-confidence, systemswide view of human-flavivirus interactions and provide insights into the role of the EMC in flavivirus
replication.
INTRODUCTION
The New Millennium has brought a rapid expansion of human flavivirus infections, including dengue virus (DENV), yellow fever
virus (YFV), West Nile virus (WNV), and Zika virus (ZIKV) (Bhatt
et al., 2013). Given that global warming is predicted to enlarge
the range of the insect vectors that carry these viruses, it is critical that we understand the biology of these viruses in order to
design effective therapies against them. DENV and ZIKV are single-stranded, positive-sense RNA viruses that are transmitted
to humans by Aedes mosquitos. Both are fast-growing health
threats that are producing an escalating number of infections
in the Americas and worldwide.
Each year, 390 million people are infected with DENV, with
500,000 individuals hospitalized with severe dengue, the majority of those being young children (Bhatt et al., 2013). ZIKV, first
isolated from an infected macaque in Uganda in 1947, suddenly
emerged in Micronesia in 2007 and expanded its range to Southeast Asia. In May 2015, ZIKV was identified in Brazil coincident
with an upsurge in neurologic and fetal abnormalities. With its
rapid spread to Central and South America, ZIKV has emerged
as a severe health threat by virtue of its fast-paced global spread
and associated morbidities, including microcephaly and Guillain-Barre syndrome. (DOrtenzio et al., 2016; Driggers et al.,
2016; Haug et al., 2016; Lazear and Diamond, 2016; Musso
and Gubler, 2016; Rasmussen et al., 2016). These events have
led to ZIKV being declared a public health emergency by the
World Health Organization. Recent animal models have demonstrated that ZIKV infects the placentas of pregnant mice, with
transmission to fetal mice resulting in death or severe growth
impairment (Cugola et al., 2016; Miner and Diamond, 2016;
Miner et al., 2016; Li et al., 2016). There are no specific therapies
for flavivirus infection, although a DENV vaccine has recently
been approved in some countries. There is no approved vaccine
or therapy for ZIKV infection.
Flavivirus replication begins with the virus binding to host cell
receptors and undergoing endocytosis (Fernandez-Garcia
et al., 2009). A number of proteins have been implicated to
facilitate DENV attachment and entry, including TIM1 and
AXL (Jemielity et al., 2013; Meertens et al., 2012; Morizono
and Chen, 2014; Perera-Lecoin et al., 2014; Richard et al.,
2015), the latter having also been identified as an important
ZIKV entry factor (Hamel et al., 2015). Subsequent to initial viral
entry, late endosomal acidification triggers the fusion of host
and viral membranes and permits the virus positive sense
RNA genome (viral RNA [vRNA]) to enter the host cell cytosol.
Upon cytosolic entry, the vRNA is translated into a large polyprotein on the rough endoplasmic reticulum (RER). This polyprotein is processed by both host and viral proteases into three
structural proteins (premembrane [prM], capsid [C], and the
glycoprotein envelope [E protein]), and seven non-structural
(NS) proteins (NS1, NS2A, NS2B, NS3, NS4A, NS4B, and
NS5). DENV has been demonstrated to extensively remodel
the ER into replication centers (RCs), where progeny viruses
are created. The newly synthesized flaviviruses then traffic
from the RER to the cell surface via the Golgi, where they
Cell Reports 16, 115, July 19, 2016 2016 The Author(s). 1
This is an open access article under the CC BY-NC-ND license (http://creativecommons.org/licenses/by-nc-nd/4.0/).
Figure 1. MORR Screens for Identifying Human Proteins that Modulate DENV Replication
(A) Schematic workflow diagram of the DENV-HF screen.
(B) MAGI HeLa cells were transfected with the indicated negative (non-targeting [NT]) and positive (ATP6V0B, IFITM3) controls for 72 hr and then infected with
DENV2-NGC (MOI 0.5) for 30 hr. Cells were then fixed, permeabilized, and immunostained with anti-E protein antibody (4G2, green) or stained for DNA (blue).
Percent infected cells SD are shown at 43 magnification.
(legend continued on next page)
2 Cell Reports 16, 115, July 19, 2016
undergo exocytosis, thus spreading the infection to neighboring cells.

The flaviviruses have a complex lifecycle that relies on the host
cells resources. Earlier efforts have addressed the role of
arthropod DENV-host factors (DENV-HFs) (Sessions et al.,
2009) and human factors required by YFV (Le Sommer et al.,
2012) and WNV (Krishnan et al., 2008; Ma et al., 2015). Nonetheless, fundamental questions regarding how human proteins
modulate flavivirus replication, including ZIKV infection, remain.
Therefore, we investigated human-flavivirus interactions with a
combined functional genomic approach employing multiple
orthologous RNAi reagent (MORR) screens for DENV replication
together with a pooled CRISPR/Cas9 cell survival enrichment
screen to find host factors that are required for ZIKV replication.
To limit false positives, we used gene-expression filtering to remove host cell candidates that were not expressed in the cells
used for the screens. To address some of the shortcomings of
functional genomics (false positives and false negatives), we integrated the MORR screen datasets using an existing RNAi
analysis program, RIGER (Luo et al., 2008), which quantitatively
determines what the likelihood is of a gene contributing to the
phenotype of interest. This orthologous functional genomic effort
recovered the recently discovered ZIKV entry factor AXL (Hamel
et al., 2015) as well as multiple proteins involved in endocytosis
(RAB5C and RABGEF), heparin sulfation (NDST1 and EXT1) and
transmembrane protein processing and maturation, including
the endoplasmic reticulum membrane complex (EMC). Unlike
what has been shown with WNV, we find that DENV, ZIKV, and
YFV strongly depend on the EMC for their replication and that
this requirement occurs at an early stage of infection. Using proteomics, we show that the EMC associates with the translocon
and oligosaccharyl-transferase (OST) complex. Together, these
studies generate a high-confidence, systems-wide view of human-flavivirus interactions and deepen our understanding of flavivirus pathogenesis.
RESULTS
MORR Screens for Identifying DENV Host Factors
We developed an RNAi screening platform for DENV-HFs (Figures 1A and 1B). After 72 hr of siRNA transfection, HeLa cells
were infected with DENV serotype 2-New Guinea C strain
(DENV2-NGC, MOI 0.5; Figure 1A). At 30 hr post infection, the
cells were stained using a monoclonal antibody, 4G2, directed
against a conserved fusion loop in the flaviviral E protein
(E) and stained for DNA. Cell number and percent infection
were determined for each well. Positive control siRNAs were
against ATP6V0B and IFITM3. This assay was used to screen
in triplicate three siRNA libraries (Dharmacon siGENOME 4;
SMARTpool, 21,121 genes targeted), Ambion Silencer Select
(21,584 genes targeted), and the Dharmacon revision 5
(SMART-Rev, 4,506 genes targeted; Figure 1C). These libraries
were selected because of their complementary design and cu-
mulative comprehensive coverage. A comparison of >1,000

siRNA pools from the Silencer Select and SMARTpool libraries
demonstrated <5% similar siRNAs showing that the reagents
are largely orthologous. The SMART-Rev library is based on a
more recent RefSeq database annotation and thus in part supplements the earlier designed SMARTpool library with which it
shares <10% siRNA sequence overlap. siRNA pools were chosen as hits if the percentage of infected cells was % 50% or R
150% of the plate mean and cell number was R 50% of the
plate mean. Selected DENV-HF candidate pools were validated
by independently testing the individual oligos from each pool
(Table S1). In the validation round, pools with two or more
siRNAs that satisfied these criteria were denoted as higher confidence because such results are most consistent with the
depletion of the intended target (Echeverri et al., 2006). Collectively, these screens identified over 150 high-confidence candidate DENV-HFs that were validated with two or more individual
siRNAs (Tables S1 and S2). The validity of this approach was
supported by the enrichment of multiple pathways and macromolecular complexes in the high-confidence set (e.g., the vacuolar ATPase [vATPase], the OST complex, and endocytic
trafficking; Figures S1AS1D; Tables S1 and S2), arguing that
the majority of these candidates are true positives. Overlap of
specific genes between our high-confidence list with either a
Drosophila-based DENV-HF screen (Sessions et al., 2009) or a
human-cell-based whole-genome screen using WNV (Krishnan
et al., 2008) was low (discussed below). Consistent with known
DENV dependencies, all of the screens independently identified
various vATPase components, and two or more screens also
detected a role in replication for SEC61B, an endoplasmic reticulum protein translocase component, and SCFD1, which is
necessary for transmembrane protein processing. The low specific gene overlap between the screens may reflect different
screening strategies and reagents and is consistent with previous screen comparisons revealing that, although pathways and
complexes are more likely to be detected across screens, the
overlap of specific genes is less pronounced. Although we did
find significant overlap in pathways and gene clusters when
comparing our own Silencer Select and SMARTpool library
screens, the exact gene overlap was low (13%14%; Table
S2). This is consistent with the differences between the various
siRNA libraries influencing the genes recovered. Therefore, our
use of multiple orthologous RNAi reagents is useful in attempting to saturate a given screen.
RIGER Analysis of the MORR DENV-HF Screens
In an attempt to address the caveats of RNAi, false negatives, and
false positives, we employed the MORR screening approach and
used an existing informatics program, RIGER (Luo et al., 2008), to
integrate the primary screen data into rankings of dependency
factors (DFs; RIGER3-DF) or competitive factors (CFs; RIGER3CF; Table S3). We compared the RIGER and primary screen rankings for their ability to detect genes in the EMC, OST complex or
(C) The results of the DENV-HF screens with the siRNA pools ranked in order of their normalized percent infection (log2 scale). The highlighted genes are selected
hits from each respective library.
(D) Based on the RIGER3 screen dataset, a hypothetical model cell was created highlighting the DENV lifecycle as well as where 259 of the DENV-HFs might
function on the basis of the available literature (Table S3).
Cell Reports 16, 115, July 19, 2016 3
the EC (Figure S2A). For this, we calculated the area under the
curve (AUC) generated by plotting the percentage of complexrelated genes from the total set that is detected moving from
the top down along the respective gene rankings (Figure S2A;
Table S3). The AUC analysis shows graphically how well an
siRNA library or RIGER analysis performed in identifying the expected components. The AUC analyses show that the Silencer
Select screen was superior, outperforming both the RIGER analysis and the SMARTpool screen for both the EMC and OST complex. For the EC, the RIGER analyses were more efficient than the
independent screens, revealing its usefulness. Using the flavivirus replication cycle as a starting point and with the referenced
literature as a guide, 259 (137 DFs and 122 CFs) of the top 150
of both the DF and CF RIGER3 gene lists expressed in the
HeLa cell were assembled into a speculative model of how they
might modulate viral infection (Figure 1D; Table S3).
DENV-HF Pathways and Complexes
The screens enriched for several pathways and complexes
needed by the virus (e.g., OST, conserved oligomeric complex
of Golgi, and nonsense-mediated decay pathway), whose components scored across the primary screen datasets (Figures S1
and S2; Tables S2 and S3). For the OST complex, DENV required
seven of eight components (Figure S1D; Table S2). A similarly
strong enrichment was seen with the vATPase, with 13 subunits
needed by the virus. In addition to DENV-HFs, the screens also
identified the EC as well as others, whose loss enhanced viral
infection (CFs) in keeping with inactivation of the hosts transcriptional machinery being a common viral attack mechanism
(Tables S1, S2, and S3).
Among the most enriched gene sets was the EMC, a highly
conserved, although poorly understood, oligomeric complex
comprised of nine to ten distinct proteins that predominantly
reside in the ER (Figures 2A and 2B). The EMC is needed for transmembrane protein folding (Jonikas et al., 2009) and has been
implicated in regulating the expression of the rhodopsin receptor
in flies (Satoh et al., 2015) as well as orchestrating the transfer of
lipids from the ER to the mitochondria in yeast (Lahiri et al., 2014).
Notably, a pooled CRISPR/Cas9 screen for WNV host factors
found three components of the EMC to be needed for WNVinduced cell death but not for viral replication (Ma et al., 2015).
In contrast to what has been reported for WNV, we found that
DENV strongly required EMC1, EMC2, EMC4, and EMC5 for its
replication (Figures 2A2C), suggesting that these related flaviviruses utilize the EMC differently, a topic we discuss further below.
DENV-HF MORR Screens Have False Positives and False
Negatives
Comparison of candidate lists from similar siRNA screens reveals
shared pathways and complexes, but in the majority of instances,
there is low exact gene overlap (Perreira et al., 2015, 2016; Zhu
et al., 2014). The MORR DENV-HF screens behaved similarly,
with a low percentage of same-gene overlap detected in the primary screens (for DFs, with % 50% plate mean infected cells
and R 50% plate mean number of cells: Silencer Select 13.5%,
SMARTpool 14.4%, and SMART-Rev 10.6%; Table S2). A comparison across the screens for siRNAs targeting components of
the OST complex, the EC, or the EMC showed estimated false
negative rates ranging from 25% to 50% (Table S2). We generated

a common candidate list for genes that scored in two or more
of the screens (Table S2). To minimize the effects of false positives
and off-target effects, we performed microarray analysis (Affymetrix GeneChip human 2.0 ST array) to determine the genes expressed in the HeLa cells (Tables S1, S2, and S3). The microarray
probe-set values were matched to the genes present in the siRNA
libraries: 17,070 (80.8%) and 17,168 (79.5%) genes with expression data in the SMARTpool and Silencer Select screens. The median of the negative control intron probe set served as a cutoff for
gene expression, producing a list of 12,461 common genes in the
siRNA libraries that are expressed in the HeLa cells.
A CRISPR/Cas9 Screen to Identify ZIKV Host Factors
The advent of bacterially derived CRISPR/Cas9 gene modulation
has revolutionized the field of in vitro mammalian pooled genetic
screening (Hsu et al., 2014; Sanjana et al., 2014; Shalem et al.,
2014, 2015). In optimization experiments, we found that the
ZIKV MR766 strain kills >95% of H1-HeLa cells in 8 days at an
MOI of 5, with other ZIKV strains not being as cytopathic, making
this combination of virus and the host cell ideal for a pooled
CRISPR/Cas9 screen (data not shown). Therefore, to find additional flavivirus host factors, we performed a CRISPR/Cas9
screen using the H1-HeLa cell line and ZIKV MR766 (Figure 3A).
We first stably expressed a human-codon-optimized cDNA of
S. pyogenes Cas9 in a population of HeLa cells (Shalem et al.,
2014). Next, we stably transduced the H1-HeLa-Cas9 cells at an
MOI of 0.2 with a complex lentiviral pool expressing the human
GeCKO v.2 single-guide RNA (sgRNA) library (Shalem et al.,
2014), which targets 19,052 genes in the human genome with
six sgRNAs per gene across two half-libraries (libraries A and B).
Libraries A and B each possess three unique sgRNA per gene
and these two half-libraries were used in parallel to screen for
ZIKV host factors in independent triplicate screens. The Cas9-expressing H1-HeLa cells were infected with ZIKV MR766 at an MOI
of 5 and cultured for 8 days until >95% of the cells were dead, as
determined by their detachment from the plate surface as seen
with an inverted microscope. The surviving cells, appearing as
tightly clustered clonal colonies, were expanded, and genomic
DNA was prepared. Proviruses containing the sgRNA stably integrated into each of the surviving cells were amplified and identified
from genomic DNA using PCR and next-gen sequencing. Similar
to RNAi screens, we used the reagent redundancy principle
(Echeverri et al., 2006) to select for candidate genes which
had >6 sequencing reads for three or more independent sgRNAs
(Table S4). We then repeated the screen and compared the data to
identify candidates that are high confidence in both screens (Table
S4). There was a high degree of sgRNA and exact-gene overlap
among the top candidates found by the independent screens (Figure S2B; Table S4), suggesting that the screen was approaching
saturation under these conditions.
EMC Is Required for the Replication of ZIKV, DENV,
and YFV
Among the top ZIKV-DF candidates (>4 independent sgRNAs,
with >6 sequencing reads per sgRNA; Table S4) from the
CRISPR/Cas9 screens was the recently published ZIKV entry
factor, AXL (six of six sgRNAs) (Hamel et al., 2015; Jemielity
Figure 2. The EMC Is Needed for DENV Replication

(A) MAGI HeLa cells were transfected with the indicated Silencer Select siRNAs (Table S1) for 72 hr and then infected with DENV2-NGC (MOI 0.5). At 48 hr post
infection, the cells were fixed, permeabilized, and immunostained for the viral E protein. Imaging analysis software was used to determine the percentage
infection. Values indicate the mean infected cells normalized to the NT negative control siRNA-transfected samples of n = 3 independent experiments SD.
(B) Schematic of the EMC with components 110 depicted as circles whose relative sizes are proportional to their respective molecular weights (Table S1).
Components 8 and 9 are shown as one subunit because of their similarity and variable expression. Subunits are indicated in color if they scored in one or more of
the DENV-HF MORR screens.
(C) MAGI HeLa cells were transfected with the indicated pooled Silencer Select siRNAs (three siRNAs per gene; Table S1) for 72 hr and then infected with DENV2NGC (MOI 0.5). At 48 hr post infection, the cells were fixed, permeabilized, and immunostained for the viral E protein. The nuclei of the cells were also stained for
DNA using Hoechst 33342 dye (blue). Imaging analysis software was used to determine the percentage infection (numbers shown SD). Values indicate the
mean percent infected cells of n = 3 independent experiments SD at 43 magnification.
Results throughout are the mean of three independent experiments SD.
Figure 3. CRISPR/Cas9 Screens to Identify Human Proteins that Modulate ZIKV Replication
(A) Schematic workflow diagram of the ZIKV-HF CRISPR/Cas9 screen.
(B) Ranking of genes from the ZIKV-HF CRISPR/Cas9 screen. Relative sgRNA frequencies were detected using next-gen sequencing (pooled data from two
independent screens) were plotted against relative frequencies of sgRNAs in the unselected starting population of H1-HeLa cells containing the human GeCKO
v2.0 library. For top-ranking genes, sgRNA frequencies are highlighted.
(C) Ranking of genes from the ZIKV-HF CRISPR/Cas9 screen. Next-gen sequencing reads from two independent screens were mapped to the human GeCKO
v2.0 library using bowtie2. Only sgRNAs with six or more sequencing reads were used for analysis. Identified genes were ranked according to the number of total
reads from the two screens combined and according to the number of retrieved independent sgRNAs per gene (the human GeCKO v2.0 library contains six
independent sgRNAs per gene). Gene symbols are shown for a selection of the highest-ranking genes.
et al., 2013; Meertens et al., 2012; Perera-Lecoin et al., 2014),

providing both an unbiased confirmation of AXLs role in ZIKV
replication as well as an attestation to the validity of the
screening approach (Figures 3B and 3C; Table S4). The screen
also recovered candidate ZIKV-HFs involved in endocytosis
(RAB5C, RABGEF1, WDR7, and ZFYVE20), transmembrane protein production (the translocon-associated components SSR2
and SSR3), and post-translational modification (STT3A, EXT1,
and EXTL3) as well as two microRNAs, HSA-MIR-451A and
HSA-MIR-451B (Figures 3B and 3C; Table S4).
Consistent with the MORR DENV-HF screens, the CRISPR/
Cas9 ZIKV-HF screen effectively saturated the EMC, finding all
nine components with three or more sgRNAs, including R5
sgRNAs out of six total for EMC1EMC6 (Figure 3D; Table S4).
Importantly, the EMC was needed for the replication of a more
recently isolated ZIKV strain provided by a pediatric patient in
Cambodia in 2010 based on the complexs depletion leading to
a marked decrease in the levels of either viral double stranded
RNA (dsRNA) or E protein in H1-HeLa cells (Figures 3E and
S3A), with similar results seen using A549 adenocarcinoma
epithelial cells (data not shown).
The role of 12 of the top candidates found in the ZIKV-HF
screen was then assessed in a replication assay (E protein
expression) using three independent siRNAs targeting each
gene (Figure 4A). In matched studies, we determined the level
of mRNA target depletion produced by each of the siRNAs (Figure 4B). These paired assays showed a strong correlation
between the level of mRNA knockdown and the observed
phenotype. All of the ZKV-HF candidates, including EMC1
EMC6, decreased E protein levels to %50% of control with
two or more siRNAs, and this effect was seen with multiple
strains of ZIKV (MR766 Africa 1947, ZIKV FSS3025 Cambodia
2010, and ZIKV Puerto Rico 2015) as well as DENV2-NGC.
Low-passage clinical isolates of DENV1 and DENV4 from
Thailand, as well as the YFV vaccine strain YF17D, also required
the EMC for replication (Figure 4C).
A Role for the EMC in Early DENV and ZIKV Replication
These data revealed that the EMC was needed for DENV and
ZIKV replication. In contrast, WNV has been shown to not require
the EMC for its replication, with loss of the complex resulting in
an increase in viral replication intermediates in some instances
(Ma et al., 2015). Although the role of the EMC in WNV replication
remains to be tested in our system, these data suggest that the
DENV- and ZIKV-HF screens reported here have identified a previously unappreciated role for the EMC in flavivirus replication.
To determine where in the DENV and ZIKV lifecycles the EMC
is needed, we tested how depletion of either EMC1, EMC2,
EMC4, or EMC5 impacted the early stages of DENV infection us-
ing synchronized time course infectivity experiments. To monitor

viral infection in these assays, we employed immunostaining for
E protein combined with a fluorescence-in-situ-hybridizationbased imaging assay (viewRNA) for vRNA (Feeley et al., 2011).
DENV2-NGC (MOI 50) was first incubated on ice with siRNAtransfected H1-HeLa cells to allow viral binding but not endocytosis. Warm media was then added, and, at the given times, the
cells were washed with cold PBS, incubated with or without
trypsin (T), fixed, processed, and confocally imaged using either
an antibody that binds E protein (4G2) or a probe set that recognizes DENV or ZIKV RNA (ViewRNA). The trypsin step is designed to remove the virus remaining at the cell surface after
warming has triggered the virus endocytosis. Within the time
course of this imaging assay (0 to 90 min post infection), the viral
signals arise predominantly from the incoming viruses that are
either surface bound, trafficking within the endosomal pathway,
or have recently fused and transferred their contents, including
the vRNA, into the cytosol. However, we cannot rule out the possibility that in addition to the vRNA that has recently entered the
cytosol from incoming viruses, this assay is also detecting the
products of de novo vRNA transcription that has occurred at
60 and 90 min post infection. Given our experiences using the
viewRNA assay to detect the RNA of other viruses, e.g., influenza
(Chin et al., 2015; Feeley et al., 2011), it is unlikely that this assay
would detect vRNA within intact virions due to their limited
accessibility to the probes. In support of this, our attempts to
detect vRNA fixed to the cell surface immediately prior to warming and infection (time zero) in samples that were not treated with
trypsin (-T) showed little to no vRNA signal (data not shown).
At 0 min with no trypsin treatment, the levels of surface-bound
virus were not appreciably altered by loss of the EMCs (Figures
5A5C and S3B). In contrast, at 40 min post infection, loss of the
EMC decreased the levels of both intracellular E protein and
vRNA as compared to the control cells. A small amount of virus
was seen arrested at or very near the surface of the EMCdepleted cells at 40 min post infection subsequent to trypsin
treatment; therefore, we attempted to rescue DENV2-NGC entry
using acidified buffer. However, this did not produce an observable effect (data not shown). Similar results were obtained using
ZIKV MR766, with EMC depletion decreasing intracellular vRNA
levels but not the levels of bound virus at 0 min (Figures S4A
S4D). The depletion of AXL also resulted in less intracellular E
and vRNA apparent at both 60 and 90 min post infection but little
to no decrease in the levels of surface bound virus at time zero
(Figures S5AS5E). The lack of an effect on viral binding with
AXL depletion seen with this confocal imaging-based assay
was surprising given its known role as a viral entry factor; however, it is partly consistent with a published study that detected
only a modest loss, 20%, of DENV binding to host cells in the
(D) Schematic of the EMC with components 110 depicted as circles whose size is representative of their relative molecular weights. Components 8 and 9 are
shown as one subunit because of their similarity and variable expression. Subunits are indicated in color if they scored in two or more of ZIKV-HF CRISPR/Cas9
screens repeats with the indicated number of independent sgRNAs (right).
(E) H1-HeLa cells were transfected with the indicated pooled Silencer Select siRNAs (three siRNAs per gene; Table S1) for 72 hr. then infected with ZIKV
Cambodia (moi 0.5). At 48 hr post infection the cells were fixed, permeabilized, and immunostained for viral double-stranded RNA (dsRNA; recombinant J [rJ]
antibody, green). The nuclei of the cells were also stained for DNA using Hoechst 33342 dye (blue). Imaging analysis software was used to determine the
percentage infection (numbers shown SD). Values indicate the mean percent infected cells of n = 3 independent experiments SD shown at 43 magnification.
Figure 4. ZIKV-HF Are Required for the Replication of Multiple Viral Strains
(A) H1-HeLa cells were transfected with the indicated Silencer Select siRNAs (three independent siRNAs [#1, #2, and #3] per gene; Table S1) for 72 hr and
then infected with the indicated flaviviruses (MOI 0.31). At 48 hr post infection, the cells were fixed, permeabilized, and immunostained for the viral E
protein. Imaging analysis software was used to determine the percentage infection. Values indicate the mean percent infected cells of n = 3 independent
experiments SD.
Figure 5. Loss of the EMC Halts DENV Early

in Replication
(A) H1-HeLa cells were transfected with the indicated pooled Silencer Select siRNAs (three siRNAs
per gene, vATP = vATPase subunit ATP6V0B) for
72 hr and then incubated with DENV2-NGC on ice
(MOI 100). The viral supernatant was removed, and
warm media was added at time zero. At the indicated time points after warming, cells were washed
and treated with (+) or without ( ) a 1 min incubation
with trypsin (T) to remove or leave surface-bound
virus. The cells were then fixed, permeabilized, and
immunostained with anti-E antibody (4G2, green)
and confocally imaged (633). At 40 min post
warming, vRNA was also detected in matched
samples using a viewRNA probe set (red). Cellular
DNA was stained with DAPI (blue). The scale bar
represents 20 mM. Corresponding differential
interference contrast (DIC) images are provided in
Figure S3B.
(B and C) Quantitation of DENV E and DENV RNA
signal intensity normalized to the NT control is
provided (normalized E or vRNA mean intensity
SD) for R 12 cells from each of n = 3 experiments.
presence of an anti-AXL antibody using a flow-cytometry-based

assay (Meertens et al., 2012).
These studies showed that DENV and ZIKV infection was
halted early in the setting of lowered EMC components or with
the loss of AXL but did not impact viral binding under these settings. Therefore, we modified the above assay by omitting the
trypsin step at 80 min post infection in order to assess what
happens to the bound virus we observed at time zero in the
EMC- and AXL-depleted cell samples. We also omitted a permeabilization step in the no-trypsin (-T) 80 min post infection samples so as to preferentially detect cell-surface-bound viruses.
These studies showed that, at 80 min

post infection, the majority of the ZIKV
MR766 remained bound to the cell surface
of the EMC and AXL siRNA-transfected
cells, whereas, in the negative control, the
vast majority of the surface-bound virus
had undergone endocytosis and entered
the cells interior (Figures 6A6C).
Given these data, we determined
whether the EMC was involved with the
cell-surface levels of either AXL or heparin
sulfate, both of which are important for
the initial stages of DENV and ZIKV replication (Jemielity et al., 2013; Meertens et al.,
2012; Richard et al., 2015). Depletion of
EMC1, EMC2, EMC4, or EMC5 did not
decrease surface expression of either
AXL or heparin sulfate (Figures S6A and
S6B). We also evaluated the levels of a broad range of surface
protein glycosylation moieties using flow cytometry and a panel
of conjugated lectins and found no differences between the
negative control and the EMC-depleted cells (data not shown).
Together, these data suggest that the loss of the EMC blocks
DENV and ZIKV replication after the initial virus binding to the
host cell but prior to the virus endocytosis; however, given the
limitations of this imaging assays noted above, we cannot rule
out the possibility that a portion of the block to infection involves
the inhibition of the early amplification of vRNA in the cytosol.
This last possibility will require further evaluation in future studies
(B) mRNA was made from cells in (A), and qPCR was performed in order to determine the relative abundance of the indicated mRNAs. Values indicate the mean
expression normalized to the NT control.
(C) Experiments performed as in (A) using the YF17D and low-passage DENV1 and DENV4 viruses.
Figure 6. ZIKV Remains at the Cell Surface

with Loss of the EMC or AXL
(A) H1-HeLa cells were transfected with pooled
Silencer Select siRNAs (three siRNAs per gene)
targeting the indicated EMC components or AXL for
72 hr and then incubated with ZIKV MR766 on ice
(MOI 50). The viral supernatant was removed and
warm media was added at time 0. To preferentially
detect internalized virus at the indicated time points
post warming, cells were subjected to a one minute
incubation with trypsin (+ T, top) to remove surfacebound virus followed by fixation and permeabilization. In order to selectively detect surface-bound
virus, matched sets of cells were processed without
the trypsin step (-T) and then fixed with no permeabilization (bottom). Both sets of cells were then
immunostained for E protein (green) and confocally
imaged (633). Cell plasma membranes are outlined
in white on the basis of DIC images. Cellular DNA
was stained with DAPI (blue). The scale bar represents 20 mM.
(B and C) Quantitation of ZIKV E signal intensity
normalized to the NT control is provided (normalized mean E intensity SD) for R 12 cells for each
panel in (A) from each of n = 3 experiments. The
legend indicates the minutes post infection (post
warming).
using additional assays such as flavivirus replicon-expressing

cells.
The EMC Associates with the OST Complex
The EMC is conserved from yeast to humans, participates in ER
protein folding response (with its loss eliciting the unfolded
protein response (UPR) [Jonikas et al., 2009]), and is needed
for ER-to-mitochondrial lipid trafficking (Lahiri et al., 2014) and
the expression of several multi-pass transmembrane proteins,
including rhodopsin, in the Drosophila eye (Satoh et al., 2015)
and the acetylcholine receptor in C. elegans (Richard et al.,
2013). EMC1EMC6 were shown to form a complex after their
identification in a screen in yeast for ER protein folding factors
(Jonikas et al., 2009). EMC7EMC10 were added to these initial
functionally defined EMC components based on proteomic
studies (Christianson et al., 2012). Our data now demonstrate
that, in addition to WNV-induced cell death (Ma et al., 2015),
the loss of the EMC arrests DENV and ZIKV early in their replication cycles. To further elucidate the role of the EMC in viral infection and host-cell biology, we identified proteins that associate
with EMC1, EMC2, EMC4, or EMC5, using affinity purification
coupled to mass spectrometry (AP-MS).

EMC1, EMC2, EMC5HA, and EMC4FLAG fusion protein expression in H1HeLa cells were induced for 72 hr with
doxycycline, and affinity purification (AP)
was performed (Figures 7A7D). Co-purified proteins were then identified with
shotgun-proteomics. Proteins associating
with EMC components (hits) have increased total spectral counts from affinity
purifying EMC components compared to
vector control cells (Figure 7E; Table S5). These experiments
identified multiple components of the translocon and OST complexes, including both STT3A and STT3B subunits, RPN1/RPN2,
and DDOST, as associating with the EMC (Figures 7E7G; Table
S5). From this AP-MS dataset, common pathways emerge as
being potentially important for the EMCs function (Figure 7F;
Table S5). Glycolysis and glucose-metabolism pathway components are the most enriched (23 of 72 proteins). The next most
represented pathway is the citric acid cycle and electron transport, where 26 of 155 proteins are present in the EMC affinity purifications. Protein metabolism (57 of 687), protein processing in
the ER (23 of 169), and protein folding (CCT/TriC) pathways (8 of
21) are also highly represented.
The higher-order assembly of the EMC proteins with one
another, and how such differing assemblages might function, is
poorly understood. Consistent with the notion that different
EMC components have distinct functions, proteins that co-purify
with EMC4 showed the least amount of overlap with those associating with EMC1, EMC2, and EMC5. EMC4s affinity purification
alone co-purified EMC2, whereas EMC1, EMC2, and EMC5 all
associated with multiple other EMC components, including
EMC7, EMC8, and EMC10. These data suggest that EMC1,

EMC2, and EMC5 are core components of the EMC complex,
whereas EMC4s association may be more peripheral or restricted
to a particular variant. Consistent with this idea, EMC4 alone copurified with multiple COP1 (COPB1, COPB2, and COPG1) and
COPII (SEC13, SEC23A, and SEC22) components, which are
coat-protein complexes responsible for anterograde and retrograde transport to and from the ER to the Golgi (Brandizzi and Barlowe, 2013), suggesting that this subunit is involved in trafficking
between these closely interacting organelles.
DISCUSSION
The rapid spread of flavivirus infections represents a severe
and escalating threat to global health (Lazear and Diamond,
2016). To obtain a systems-level understanding of DENV and
ZIKV host factor dependencies and identify potential therapeutic targets, we completed orthologous functional genomic
screens using both MORR/RIGER and CRISPR/Cas9 strategies coupled with validation experiments and proteomics.
We chose this combinatorial approach in order to take advantage of the best that each of the these technologies has to
offer, thereby offsetting the caveats of genetic screening, false
positives, and false negatives. This combined functional
genomic approach generated a comprehensive overview of
flavivirus dependencies in human cells. The strong overlap of
enriched complexes and pathways between the MORR RNAi
screens and the CRISPR/Cas9 screens, as well as the
CRISPR/Cas9 screens themselves, suggest that the flavivirus
host factor screens are approaching saturation (Tables S6
and S7). These datasets will now hopefully provide a useful
resource in helping the research community find and exploit
flavivirus weaknesses.
Both screening approaches (MORR and CRISPR/Cas9) enriched for common complexes and pathways, including the
EMC; however, it was only in the CRISPR/Cas9 screen that the
reported ZIKV and DENV entry factor AXL was detected as
important for viral replication (Hamel et al., 2015). This is consistent with pooled screens using cell survival as a readout displaying limited sensitivity (few candidates recovered) but excellent
specificity in finding host genes that act very early in viral replication, for instance host factors needed for viral entry (Perreira
et al., 2016). In contrast, RNAi screens generate a larger list of
host factor candidates contributing to a broader understanding
of viral requirements; however, this comes at the high cost of
increased false leads. Therefore given these complementary
strengths and weaknesses, a combined RNAi and CRISPR/
Cas9 screening effort is likely to yield the most comprehensive
insight into human-viral interactions.
Overlap with Flaviviral Screens
siRNA screens for both WNV and YFV host factors in human
cells, as well as a Drosophila cell-based siRNA screen for
DENV host factors, have been previously published (Krishnan
et al., 2008; Kwon et al., 2014; Le Sommer et al., 2012; Ma
et al., 2015; Sessions et al., 2009). We compared our MORR
screen datasets with the published candidate lists from these efforts (Table S6). We found that the overlap between these efforts
was low, most likely due to false positives and false negatives
that vary significantly between siRNA libraries and platforms
(Table S6). The greatest overlap was with the fly-based screen
(Sessions et al., 2009), which, although surprising, may speak
to lower levels of artifacts being produced using that approach.
Notably, neither AXL nor the other published DENV receptor
TIM1 (Jemielity et al., 2013; Meertens et al., 2012) were recovered as a viral dependency factor in any of these RNAi-based
efforts.
A comparison between the published flavivirus screens and
the ZIKV CRISPR/Cas9 screen for overlap showed that the
WNV CRISPR/Cas9 screen detected EMC2, 3 and 7 and the
translocon-associated protein, SSR2, as being needed for
WNV cytopathicity (Ma et al., 2015). The YFV siRNA screen found
EMC3, WDR7 (endocytosis), EXT1 (heparin sulfation) to be
needed for viral replication (Le Sommer et al., 2012); with the
requirement for EMC3 in YFV replication being consistent with
the results presented here. WDR7 was also needed for WNV
replication in the siRNA-based screen, along with SH3GLB2
and the vATPase subunit, ATP6V0A1, which are both involved
in endocytosis (Krishnan et al., 2008).
As discussed, the WNV CRISPR/Cas9 screen found that three
EMC subunits (EMC2, EMC3, and EMC7) were necessary for
virus-induced cell death but not for viral replication (Ma et al.,
2015), which is in contrast to what we had found with DENV,
ZIKV, and YFV. The lack of detection of any EMC subunit in
the WNV siRNA replication screen (Krishnan et al., 2008),
together with the EMC-modulating viral cytopathicity, but not
viral replication, as shown with the WNV CRISPR/Cas9 screen,
are consistent with the notion that the EMC plays an important
but differing role in the WNV lifecycle as compared to those of
DENV, ZIKV, and YFV.
Overlap with Related Screens
A role for the EMC in yeast was discovered using a genetic
screen to find genes needed for phospholipid (PL) transfer
from the ER to the mitochondria (Lahiri et al., 2014). A comparison of the yeast PL and ZIKV CRISPR/Cas9 screens respective candidate lists showed that only the four EMC components were shared in common, arguing against a shared
pathway in PL transfer underlying the observed phenotype
(Table S6). The flaviviruses enter host cells via clathrin-mediated endocytosis (CME), and the loss of the EMC blocks flavivirus infection early, after viral binding but likely before viral
entry. Therefore, we checked for any overlap between the
ZIKV CRISPR/Cas9 high-confidence dataset and one from a
whole-genome siRNA screen for CME using the SMARTpool
library and human cells (Kozik et al., 2013). This comparison
revealed one common gene, LRRC29. The lack of overlap between our screens and the CME RNAi screen, together with
the EMC only registering as needed for select flaviviruses
and not any other virus we or others have investigated
(HIV-1, HRV, IAV, and HCVl; Table S6) strongly suggests that
the block in flaviviral replication found in the EMC-depleted
cells is rather unique to this subset of viruses and does not
involve a general disruption of CME. In support of this,
YF17D has recently been shown to enter via a clathrin-independent pathway (Fernandez-Garcia et al., 2016).
Figure 7. Identification of EMC-Interacting Proteins and Pathways using AP-MS

(A) EMC1, EMC2, and EMC5-HA expression in H1-HeLa cells after induction with 5 ug/ml doxycycline and immunostaining with anti-HA antibody.
(B) EMC4-FLAG expression in H1-HeLa cells after induction with 5 ug/ml doxycycline and immunostaining with anti-FLAG antibody.
(C) Whole-cell lysates from cells in (A) showing EMC1, EMC2, and EMC5-HA expression. Actin (ACTB) serves as a loading control.
(D) Whole-cell lysates from cells in (B) showing EMC4-FLAG expression. ACTB serves as a loading control.
The Role of AXL in Flavivirus Replication

The phosphatidyl serine (PS) binding transmembrane proteins
AXL and TIM1 have been shown to be required for the entry
of a broad range of viruses, including DENV and ZIKV, via a
general interaction with the PS present on the viral membrane
(Hamel et al., 2015; Jemielity et al., 2013; Meertens et al.,
2012). Although multiple screens for flavivirus host factors
have not detected these proteins, including our own MORR
screens, the ZIKV CRISPR/Cas9 found AXL to be one of the
strongest candidates in terms of both orthologous sgRNAs as
well as total number of reads. To date, no flavivirus host-factor
screen has detected TIM1 as important for viral replication,
indeed we did not find it to be expressed in the HeLa cells
used in these studies (Table S1; data not shown). We found
the ZIKV requirement for AXL to occur early in infection, after
viral binding but prior to viral entry into the endosomal pathway
and/or the early amplification of vRNA in the cytosol. Interestingly, the requirement for AXL was considerably less with the
replication of DENV and YF17D, demonstrating that, for this
subset of flaviviruses, an AXL- and TIM-1-independent means
of viral entry exists.
EMCs Cellular Function
The EMC was first identified through a genetic interaction
screen in yeast as playing a critical role in protein folding, the
loss of which induced the UPR (Jonikas et al., 2009). More
recently, the EMC has been postulated to serve as a tethering
mechanism that facilitates PL transfer from the ER to the mitochondria in yeast (Lahiri et al., 2014) and to mediate the folding
of multi-pass, but not single-pass, proteins in the ER (Richard
et al., 2013; Satoh et al., 2015). We became interested in the
EMC because it was the top complex enriched for in all of our
flavivirus MORR and CRISPR/Cas9 screens. This viral dependency on the EMC was unique among the multiple viruses we
and others have investigated in human-cell-based screens,
suggesting a specific role for the EMC in the flavivirus lifecycle
(Tables S6 and S7). To improve our understanding of the EMC,
we used EMC proteins as baits in AP-MS studies to find proteins with which they associate. These studies showed that
the EMC interacts with both the translocon and the OST complex. Given the known close interaction between the translocon
and OST (Rapoport, 2007) and the role of the EMC in ER protein
folding (Jonikas et al., 2009), our data suggest that, similar to
the OST complex, the EMC is poised near the intraluminal opening of the translocon, where it would have immediate access to
the nascent transmembrane protein being extruded into the ER
lumen. Additional EMC interactions included multiple nuclearmembrane-associated proteins as well as mitochondrial proteins (Table S5). Although the EMC has been thought to reside
predominantly in the ER, these studies suggest that it may also
contribute to the functions of additional organelles, warranting
further investigation.
The Role of the EMC in Flavivirus Infection

The UPR has been implicated in modulating flavivirus replication
(Blazquez et al., 2014; Pena and Harris, 2011, 2012), in particular
during the generation of viral RCs beginning several hours post
infection. We therefore compared our gene lists with those of
the yeast screen that first identified the EMC as being important
for protein folding because its loss induces the UPR (Jonikas
et al., 2009). Interestingly, we did not identify a strong overlap
in genes that were found to play a role in the induction of the
UPR in yeast and those that are required for flavivirus replication
(Table S6). These findings point away from a triggering of the
UPR as the mechanism underlying the EMC-dependency of flaviviruses. Indeed, this comparison leaves open the possibility
that the EMCs role in transmembrane protein folding is needed
for the expression of one or more additional glycosylated multipass transmembrane proteins required for DENV, ZIKV, and YFV
replication.
A CRISPR/Cas9 screen found that the EMC was needed for
WNV-induced cytopathicity but not for viral replication (Ma
et al., 2015). Although we have not tested these observations
directly in our system, they suggest that there are at least two
distinct roles for the EMC in flaviviral infection: an early one,
potentially acting at viral entry and impacting DENV, ZIKV, and
YFV, and a second later one which may result in the death of
cytopathic flavivirus-infected cells. EMCs role in cell death appears to predominate in WNV replication, but we note that
both roles (replication and cell death) could contribute to what
we observe with ZIKV, which is also cytopathic. An early requirement for the EMC could potentially occur via the loss of an additional human protein or lipid that is needed for the early stages of
viral replication and/or the triggering of an anti-viral state; if the
latter is the case, then the induced phenotype would act in a
restricted manner because we and others have not found that
the loss of the EMC inhibits the replication of the diverse viruses
interrogated via genetic screening. In addition, multiple EMCs
are themselves predicted to be transmembrane proteins,
including EMC1, EMC3, EMC4, and EMC5, and so there also exists the possibility that one or more of the EMCs are interacting
directly with the flaviviruses, potentially as entry factors. WNV
is more closely related to DENV and ZIKV than YFV (Blitvich
and Firth, 2015), arguing against the differing roles of the EMC
being explained by phylogeny. DENV, ZIKV, and YFV are all
transmitted by Aedes mosquitos, whereas WNV is transmitted
by mosquitos of the genus Culex, suggesting that a vector-specific viral adaptation may have resulted in these two distinct roles
for the EMC in human cells. A prediction arising from this notion
is that Japanese encephalitis virus, which is also transmitted by
Culex mosquitos, should not require the EMC for its early replication. Additional studies are now needed in order to more fully
elucidate the differing actions of the EMC in flavivirus infection.
Collectively, these results provide an early evaluation of the
role of the EMC in the viral lifecycle (Blitvich and Firth, 2015;
(E) EMC interaction networks revealed through AP-MS of EMC1, EMC2, EMC4, and EMC5 to identify interacting proteins. The top 20 enriched proteins for each
EMC subunit (blue) were used to generate networks in Cytoscape 3.1 (Table S5). Proteins that interacted with two or more EMCs are shown in yellow.
(F) Pathway enrichment of proteins identified in (E) using ConsensusPath DB.
(G) Schematic of the OST complex, with components depicted in color if they were found to be enriched (p % 0.5) in AP-MS experiments involving EMC1, EMC2,
EMC4, or EMC5 versus the negative control samples.
Huang et al., 2014) together with a comprehensive and systematic view of human-flavivirus interactions intended to serve as a
useful resource in combating these emerging viral infections.
EXPERIMENTAL PROCEDURES
MORR RNAi Screens
The RNAi screens were done in triplicate using four siRNA libraries:
SMARTpool, Dharmacon (21,121 pools, three oligos per pool), Silencer Select,
Ambion (21,584 siRNA pools, three oligos per pool), and Dharmacon
RefSeq27 Reversion Pools (4,506 siRNA pools, four oligos per pool). We
created a high-throughput image-based screening platform to find host factors needed for DENV replication as follows: human cervical cancer cells
(MAGI cells, NIH AIDS Reagent Repository) were transiently transfected with
siRNA (50 nM concentration) using a reverse transfection protocol employing
0.44% Oligofectamine (siRNA delivery reagent, Invitrogen) in a 384-well plate
format. After 72 hr of siRNA-mediated gene knockdown, the cells were infected with DENV2-NGC (VR-1584) at an MOI of 0.5. At 30 hr post infection,
the cells were fixed with formalin, permeabilized with 0.1% Triton X-100,
and immunostained using the 4G2 monoclonal antibody against the E protein.
The cells were then incubated with an Alexa Fluor 488 goat anti-mouse secondary antibody and stained for DNA with Hoechst 33342. The cells were
imaged on an automated Image Express Micro microscope at 43 magnification. Images were analyzed with MetaXpress in order to determine the total
cells per well and the percentage of infected cells in each well. Positive control
siRNA SMARTpools targeting the V-ATPase subunits ATP6V0B or IFITM3
were used for the high-throughput screen. siRNA pools were classified as
hits (decreased infection) if the average of the triplicate plates showed that
the percentage of infected cells was > 50% of the plate mean and the cell
number was not > 50% of the plate mean. Pools which increase infection
by >150% of the plate mean were also selected as hits (increased infection).
Next, in the validation round each siRNA from selected hit pools (Ambion, three
siRNAs, and Dharmacon, four siRNAs) was screened individually as above. In
the validation datasets, pools with two or more individual siRNAs that met the
above criteria were deemed higher confidence, as per the reagent redundancy
principle.
RIGER Analysis
Z score robust normalization was applied prior to using the RIGER algorithm as
previously described (RNAi Gene Enrichment Ranking, GENE-E [Luo et al.,
2008]). Three RIGER methods were employed: second best (SB) method,
weighted sum (WS) method, and the Kolmogorov-Smirnov (KS) method, all
of which provide a p value for the phenotypic significance of each gene. To
compare the individual RIGER analyses to one another and the individual
DENV-HF screen datasets (Silencer Select, endoribonuclease prepared siRNA
[esiRNA], SMARTpool), we used several test sets of complexes or pathways.
For these comparisons, AUCs were derived by integration with Microsoft
Excel employing the following equation: = (SUMPRODUCT(Y1:YL-1,X2:XL)
SUMPRODUCT(X1:XL-1,Y2-YL)+XLYL-X1Y1)/2, where X and Y are a plotted
X and Y value, respectively, with the subscript representing each coordinate
and L representing the last value. Thus, L-1 represents the second to last
value. If a component was not in a dataset, then it was taken out of that dataset; in this way, alterations were made in the total number of dataset parts so
that, at the end of the dataset, the percentage of the complex was 100%.
SUPPLEMENTAL INFORMATION
Supplemental Information contains Supplemental Experimental Procedures,
six figures, and seven tables and can be found with this article online at
http://dx.doi.org/10.1016/j.celrep.2016.06.028.
AUTHOR CONTRIBUTIONS
G.S., W.M.M., P.M., J.M.P., J.M. Portmann, G.T., S.P.J., A.M.A., N.R., D.R.R.,
Z.G., S.G., T.F.K, and A.L.B. conceived and conducted the experiments,
analyzed the results, and wrote the paper.
ACKNOWLEDGMENTS
We thank R.H. Scheuermann and B. Pickett (J. Craig Venter Institute), the University of Maryland Medical System (UMMS; R. Fish, L. Benson, and J. Barrett), ICCB-L (C. Shamu, J. Smith, S. Rudnicki, S. Johnston, K. Rudnicki,
and D. Wrobel), and T. Soellner of Heidelberg University for help and inspiration. This work was supported by an Investigators in the Pathogenesis of Infectious Disease award from the Burroughs Wellcome Foundation to A.L.B. and a
subcontract to A.L.B. from contract HHSN272200900041C to Northrup Grumman Corporation from NIAID, NIH. A.L.B. is grateful to the UMMS Center for
Clinical and Translational Science and to the Bill and Melinda Gates Foundation for their support.
Received: May 12, 2016
Revised: June 2, 2016
Accepted: June 10, 2016
Published: June 21, 2016
REFERENCES
Bhatt, S., Gething, P.W., Brady, O.J., Messina, J.P., Farlow, A.W., Moyes,
C.L., Drake, J.M., Brownstein, J.S., Hoen, A.G., Sankoh, O., et al. (2013).
The global distribution and burden of dengue. Nature 496, 504507.
Blazquez, A.B., Escribano-Romero, E., Merino-Ramos, T., Saiz, J.C., and Martn-Acebes, M.A. (2014). Stress responses in flavivirus-infected cells: activation of unfolded protein response and autophagy. Front Microbiol 5, 266.
Blitvich, B.J., and Firth, A.E. (2015). Insect-specific flaviviruses: a systematic
review of their discovery, host range, mode of transmission, superinfection
exclusion potential and genomic organization. Viruses 7, 19271959.
Brandizzi, F., and Barlowe, C. (2013). Organization of the ER-Golgi interface for
membrane traffic control. Nat. Rev. Mol. Cell Biol. 14, 382392.
Chin, C.R., Perreira, J.M., Savidis, G., Portmann, J.M., Aker, A.M., Feeley,
E.M., Smith, M.C., and Brass, A.L. (2015). Direct Visualization of HIV-1 Replication Intermediates Shows that Capsid and CPSF6 Modulate HIV-1 Intranuclear Invasion and Integration. Cell Rep. 13, 17171731.
Christianson, J.C., Olzmann, J.A., Shaler, T.A., Sowa, M.E., Bennett, E.J.,
Richter, C.M., Tyler, R.E., Greenblatt, E.J., Harper, J.W., and Kopito, R.R.
(2012). Defining human ERAD networks through an integrative mapping strategy. Nat. Cell Biol. 14, 93105.
Cugola, F.R., Fernandes, I.R., Russo, F.B., Freitas, B.C., Dias, J.L.M., Guimaraes, K.P., Benazzato, C., Almeida, N., Pignatari, G.C., Romero, S., et al. (2016).
The Brazilian Zika virus strain causes birth defects in experimental models.
Nature 534, 267271.
DOrtenzio, E., Matheron, S., Yazdanpanah, Y., de Lamballerie, X., Hubert, B.,
Piorkowski, G., Maquart, M., Descamps, D., Damond, F., and Leparc-Goffart,
I. (2016). Evidence of Sexual Transmission of Zika Virus. N. Engl. J. Med. 374,
21952198.
Driggers, R.W., Ho, C.Y., Korhonen, E.M., Kuivanen, S., Jaaskelainen, A.J.,
Smura, T., Rosenberg, A., Hill, D.A., DeBiasi, R.L., Vezina, G., et al. (2016).
Zika Virus Infection with Prolonged Maternal Viremia and Fetal Brain Abnormalities. N. Engl. J. Med. 374, 21422151.
Echeverri, C.J., Beachy, P.A., Baum, B., Boutros, M., Buchholz, F., Chanda,
S.K., Downward, J., Ellenberg, J., Fraser, A.G., Hacohen, N., et al. (2006). Minimizing the risk of reporting false positives in large-scale RNAi screens. Nat.
Methods 3, 777779.
Feeley, E.M., Sims, J.S., John, S.P., Chin, C.R., Pertel, T., Chen, L.M., Gaiha,
G.D., Ryan, B.J., Donis, R.O., Elledge, S.J., and Brass, A.L. (2011). IFITM3 inhibits influenza A virus infection by preventing cytosolic entry. PLoS Pathog. 7,
e1002337.
Fernandez-Garcia, M.D., Mazzon, M., Jacobs, M., and Amara, A. (2009). Pathogenesis of flavivirus infections: using and abusing the host cell. Cell Host
Microbe 5, 318328.
Fernandez-Garcia, M.D., Meertens, L., Chazal, M., Hafirassou, M.L., Dejarnac,
O., Zamborlini, A., Despres, P., Sauvonnet, N., Arenzana-Seisdedos, F.,
Jouvenet, N., and Amara, A. (2016). Vaccine and Wild-Type Strains of Yellow
Fever Virus Engage Distinct Entry Mechanisms and Differentially Stimulate
Antiviral Immune Responses. MBio 7, e01956e15.
Hamel, R., Dejarnac, O., Wichit, S., Ekchariyawat, P., Neyret, A., Luplertlop, N.,
Perera-Lecoin, M., Surasombatpattana, P., Talignani, L., Thomas, F., et al.
(2015). Biology of Zika Virus Infection in Human Skin Cells. J. Virol. 89,
88808896.
Haug, C.J., Kieny, M.P., and Murgue, B. (2016). The Zika Challenge. N. Engl. J.
Med. 374, 18011803.
Hsu, P.D., Lander, E.S., and Zhang, F. (2014). Development and applications
of CRISPR-Cas9 for genome engineering. Cell 157, 12621278.
Huang, Y.J., Higgs, S., Horne, K.M., and Vanlandingham, D.L. (2014). Flavivirus-mosquito interactions. Viruses 6, 47034730.
Jemielity, S., Wang, J.J., Chan, Y.K., Ahmed, A.A., Li, W., Monahan, S., Bu, X.,
Farzan, M., Freeman, G.J., Umetsu, D.T., et al. (2013). TIM-family proteins promote infection of multiple enveloped viruses through virion-associated phosphatidylserine. PLoS Pathog. 9, e1003232.
Jonikas, M.C., Collins, S.R., Denic, V., Oh, E., Quan, E.M., Schmid, V., Weibezahn, J., Schwappach, B., Walter, P., Weissman, J.S., and Schuldiner, M.
(2009). Comprehensive characterization of genes required for protein folding
in the endoplasmic reticulum. Science 323, 16931697.
Kozik, P., Hodson, N.A., Sahlender, D.A., Simecek, N., Soromani, C., Wu, J.,
Collinson, L.M., and Robinson, M.S. (2013). A human genome-wide screen
for regulators of clathrin-coated vesicle formation reveals an unexpected
role for the V-ATPase. Nat. Cell Biol. 15, 5060.
Krishnan, M.N., Ng, A., Sukumaran, B., Gilfoy, F.D., Uchil, P.D., Sultana, H.,
Brass, A.L., Adametz, R., Tsui, M., Qian, F., et al. (2008). RNA interference
screen for human genes associated with West Nile virus infection. Nature
455, 242245.
Kwon, Y.J., Heo, J., Wong, H.E., Cruz, D.J., Velumani, S., da Silva, C.T., Mosimann, A.L., Duarte Dos Santos, C.N., Freitas-Junior, L.H., and Fink, K.
(2014). Kinome siRNA screen identifies novel cell-type specific dengue host
target genes. Antiviral Res. 110, 2030.
Lahiri, S., Chao, J.T., Tavassoli, S., Wong, A.K., Choudhary, V., Young, B.P.,
Loewen, C.J., and Prinz, W.A. (2014). A conserved endoplasmic reticulum
membrane protein complex (EMC) facilitates phospholipid transfer from the
ER to mitochondria. PLoS Biol. 12, e1001969.
Lazear, H.M., and Diamond, M.S. (2016). Zika Virus: New Clinical Syndromes
and Its Emergence in the Western Hemisphere. J. Virol. 90, 48644875.
Le Sommer, C., Barrows, N.J., Bradrick, S.S., Pearson, J.L., and GarciaBlanco, M.A. (2012). G protein-coupled receptor kinase 2 promotes flaviviridae
entry and replication. PLoS Negl. Trop. Dis. 6, e1820.
Li, C., Xu, D., Ye, Q., Hong, S., Jiang, Y., Liu, X., Zhang, N., Shi, L., Qin, C.F.,
and Xu, Z. (2016). Zika virus disrupts neural progenitor development and leads
to microcephaly in mice. Cell Stem Cell 19, S1934-5909(16)30084-4. Published online May 11, 2016. http://dx.doi.org/10.1016/j.stem.2016.04.017.
Luo, B., Cheung, H.W., Subramanian, A., Sharifnia, T., Okamoto, M., Yang, X.,
Hinkle, G., Boehm, J.S., Beroukhim, R., Weir, B.A., et al. (2008). Highly parallel
identification of essential genes in cancer cells. Proc. Natl. Acad. Sci. USA 105,
2038020385.
Ma, H., Dang, Y., Wu, Y., Jia, G., Anaya, E., Zhang, J., Abraham, S., Choi, J.G.,
Shi, G., Qi, L., et al. (2015). A CRISPR-Based Screen Identifies Genes Essential
for West-Nile-Virus-Induced Cell Death. Cell Rep. 12, 673683.
Meertens, L., Carnec, X., Lecoin, M.P., Ramdasi, R., Guivel-Benhassine, F.,
Lew, E., Lemke, G., Schwartz, O., and Amara, A. (2012). The TIM and TAM
families of phosphatidylserine receptors mediate dengue virus entry. Cell
Host Microbe 12, 544557.
Miner, J.J., and Diamond, M.S. (2016). Understanding How Zika Virus Enters
and Infects Neural Target Cells. Cell Stem Cell 18, 559560.
Miner, J.J., Cao, B., Govero, J., Smith, A.M., Fernandez, E., Cabrera, O.H.,
Garber, C., Noll, M., Klein, R.S., Noguchi, K.K., et al. (2016). Zika virus infection
during pregnancy in mice causes placental damage and fetal demise. Cell 165,
10811091.
Morizono, K., and Chen, I.S. (2014). Role of phosphatidylserine receptors in
enveloped virus infection. J. Virol. 88, 42754290.
Musso, D., and Gubler, D.J. (2016). Zika Virus. Clin. Microbiol. Rev. 29,
487524.
Pena, J., and Harris, E. (2011). Dengue virus modulates the unfolded protein
response in a time-dependent manner. J. Biol. Chem. 286, 1422614236.
Pena, J., and Harris, E. (2012). Early dengue virus protein synthesis induces
extensive rearrangement of the endoplasmic reticulum independent of the
UPR and SREBP-2 pathway. PLoS ONE 7, e38202.
Perera-Lecoin, M., Meertens, L., Carnec, X., and Amara, A. (2014). Flavivirus
entry receptors: an update. Viruses 6, 6988.
Perreira, J.M., Aker, A.M., Savidis, G., Chin, C.R., McDougall, W.M., Portmann, J.M., Meraner, P., Smith, M.C., Rahman, M., Baker, R.E., et al.
(2015). RNASEK Is a V-ATPase-Associated Factor Required for Endocytosis
and the Replication of Rhinovirus, Influenza A Virus, and Dengue Virus. Cell
Rep. 12, 850863.
Perreira, J.M., Meraner, P., and Brass, A.L. (2016). Functional Genomic Strategies for Elucidating Human-Virus Interactions: Will CRISPR Knockout RNAi
and Haploid Cells? Adv. Virus Res. 94, 151.
Rapoport, T.A. (2007). Protein translocation across the eukaryotic endoplasmic reticulum and bacterial plasma membranes. Nature 450, 663669.
Rasmussen, S.A., Jamieson, D.J., Honein, M.A., and Petersen, L.R. (2016).
Zika Virus and Birth DefectsReviewing the Evidence for Causality. N. Engl.
J. Med. 374, 19811987.
Richard, M., Boulin, T., Robert, V.J., Richmond, J.E., and Bessereau, J.L.
(2013). Biosynthesis of ionotropic acetylcholine receptors requires the evolutionarily conserved ER membrane complex. Proc. Natl. Acad. Sci. USA 110,
E1055E1063.
Richard, A.S., Zhang, A., Park, S.J., Farzan, M., Zong, M., and Choe, H. (2015).
Virion-associated phosphatidylethanolamine promotes TIM1-mediated infection by Ebola, dengue, and West Nile viruses. Proc. Natl. Acad. Sci. USA 112,
1468214687.
Sanjana, N.E., Shalem, O., and Zhang, F. (2014). Improved vectors and
genome-wide libraries for CRISPR screening. Nat. Methods 11, 783784.
Satoh, T., Ohba, A., Liu, Z., Inagaki, T., and Satoh, A.K. (2015). dPob/EMC is
essential for biosynthesis of rhodopsin and other multi-pass membrane proteins in Drosophila photoreceptors. eLife 4, 4.
Sessions, O.M., Barrows, N.J., Souza-Neto, J.A., Robinson, T.J., Hershey,
C.L., Rodgers, M.A., Ramirez, J.L., Dimopoulos, G., Yang, P.L., Pearson,
J.L., and Garcia-Blanco, M.A. (2009). Discovery of insect and human dengue
virus host factors. Nature 458, 10471050.
Shalem, O., Sanjana, N.E., Hartenian, E., Shi, X., Scott, D.A., Mikkelsen, T.S.,
Heckl, D., Ebert, B.L., Root, D.E., Doench, J.G., and Zhang, F. (2014).
Genome-scale CRISPR-Cas9 knockout screening in human cells. Science
343, 8487.
Shalem, O., Sanjana, N.E., and Zhang, F. (2015). High-throughput functional
genomics using CRISPR-Cas9. Nat. Rev. Genet. 16, 299311.
Zhu, J., Davoli, T., Perriera, J.M., Chin, C.R., Gaiha, G.D., John, S.P., Sigiollot,
F.D., Gao, G., Xu, Q., Qu, H., et al. (2014). Comprehensive identification of host
modulators of HIV-1 replication using multiple orthologous RNAi reagents. Cell
Rep. 9, 752766.
Cell Reports, Volume 16
Supplemental Information
Identication of Zika Virus and Dengue Virus

George Savidis, William M. McDougall, Paul Meraner, Jill M. Perreira, Jocelyn M.
Portmann, Gaia Trincucci, Sinu P. John, Aaron M. Aker, Nicholas Renzette, Douglas R.
Robbins, Zhiru Guo, Sharone Green, Timothy F. Kowalik, and Abraham L. Brass
A
80S Ribosome
29/48
43S Ribosome
B
Ribosome
29/46
17/24
Pre-rRNA Complex
31/70
Exon Junction Complex
20/68
Spliceosome
28/88
Spliceosome C Complex
19/65
SF3B Complex
SF3B Complex
4/4
snRNP
U12 snRNP
5/8
EIF3 Complex
3/3
8/20
OST Complex
4/6
ATAC B Complex
9/17
CCT/TriC
3/3
WDR5-ASH2L-RBBP5MLL2 Complex
3/4
OST Complex
4/6
COG Complex
4/8
EIF3 Complex
4/6
Large Drosha Complex
10
15
20
25
47/96
40s Ribosome
33/54
NMD
30/52
15
20
25
DENV MORR Silencer Select

and SMARTpool Combined
127/720
Gene Expression
10
CORUM complexes
38/115
mRNA Metabolism
-Log (p-value)
Reactome complexes
Translation
6/17
0
-Log (p-value)
7/7
75/385
Protein Metabolism
20/69
mRNA Splicing
Phagosomal Maturation
9/21
Endocytosis
8/22
3/5
COPI Transport
0
10
DENV MORR/RIGER Top 1000

15
20
25
-Log (p-value)
Reactome pathways
Fig. S1
Fig. S2
A NT
EMC1
EMC2
EMC4
EMC5
ZIKV Cambodia
48.0+/-7
4.5+/-1.9
4.2+/-1.1
5.8+/-1.3
5.5+/-1.4
NT
EMC1
EMC2
EMC4
EMC5
40m+T
DENV2-NGC
0m-T
0m+T
vATP
DNA
E
Fig. S3
A NT
EMC1
EMC4
EMC5
0m +T
EMC2
90m+T
ZIKV MR766
60m +T
ZIKV RNA
Normalized vRNA
signals Per Cell
5
4
3
Normalized E Intensity
NT
EMC1
EMC2
EMC4
EMC5
2
1
0
0
NT
ZIKV MR766
0m -T
30
MPI
EMC1
60
90
EMC2
1.6
1.4
1.2
1
0.8
0.6
0.4
0.2
0
NT
EMC1 EMC2 EMC4 EMC5
EMC4
EMC5
Fig. S4
Vector NT
NS1
AXL-1 NS3
AXL-2
AXL-3
ZIKV MR766
60m
1.4
1.2
60m
1
0.8
90m
0.6
0.4
0.2
0
NT
1.6
1.4
1.2
1
0.8
0.6
0.4
0.2
0
NT EMC4 EMC5 AXL-1 AXL-2 AXL-3

AXL-1
AXL-2
AXL-3
ZIKV MR766
0m -T
0m +T
NT
AXL-1 AXL-2 AXL-3

EMC4
EMC5
Normalized E Intensity
Normalized vRNA Intensity
90m
ZIKV RNA
E
E
kDa NT 1
150
AXL
2
3
AXL
ACTB
Fig. S5
NT
NDST1
EMC1
EMC2
EMC4
EMC5
Heparin Sulfate
IgG
B
IgG
AXL
NT
EMC1
AXL 1
EMC2
AXL 2
EMC4
AXL 3
EMC5
Fig. S6
Supplemental Information
Supplemental Figure Legends
Fig. S1 The DENV-HF MORR screens identify complexes and pathways that
modulate flaviviral replication, Related to Fig. 1.
(A-C) Gene enrichment analyses of the DENV-HF MORR screen RIGER3 dataset.
Highly enriched (p > 0.008) Reactome and Corum complexes and pathways among the
top scoring 550 RIGER3 DENV-HFs are shown (Table S3). Tables present gene
enrichment data for dependency factors (DF), whose loss decreases viral replication.
The Reactome and Corum databases report expertly curated functional and physical
gene associations. Overlying fractions to the right of each pathways or complexs name
indicate the number of DENV-HFs identified in the analysis over the total number of
genes in the set. The red line indicates a Log (p-value) of 1.3 (p-value of 0.05).
(D) Schematic of the OST complex, with components that scored in one or more of the
DENV-HF MORR screens (top), or in the top 1000 genes of the RIGER3 analysis
(bottom), depicted in color (right).
Fig. S2 Comparisons of MORR DENV-HF screens and the RIGER3 analysis
Related to Fig. 1, 3
(A) The RIGER3 analyses (weighted sum [WS], second best [SB], and KolmogorovSmirnov [KS], Table S3) and the individual MORR screen datasets were compared by
assessing their respective levels of enrichment for components of the EMC (top), the
OST complex (middle), or the elongation complex (EC). An enrichment score for each
screen was calculated by determining the AUC generated by plotting the percent
fraction (Fraction) of component proteins encountered moving from the lowest to
highest p value on the ranked gene lists (Rank). Numbers represent the percent
enrichment of the total gene set at <60% of the ranked gene list.
(B) Reproducibility of sgRNA detection in the 2 independent ZIKV-HF CRISPR/Cas9
screens. All mapped reads from screen 1 and screen 2 were used for this analysis,
without thresholding of read numbers. Plotting the read frequencies for sgRNAs
detected in both screens (red) shows a strong correlation (Adjusted R-Squared =
0.211), especially among guides with abundant reads (red). Guides detected only in
screen 1 (green) or only in screen 2 (blue) show a narrower distribution and a lower
number of reads.
Fig. S3 Loss of the EMC halts ZIKV and DENV early in infection, Related to Fig. 3,
5
(A) H1-HeLa cells were transfected with the indicated pooled Silencer Select siRNAs (3
siRNAs per gene, Table S1) for 72 hr. then infected with ZIKV Cambodia (moi 0.5). At
48 hr post-infection the cells were fixed, permeabilized and immunostained for the viral
E protein. The nuclei of the cells were also stained for DNA using Hoechst 33342 dye
(blue). Imaging analysis software was used to determine the percentage infection
(numbers shown +/- SD). Values indicate the mean percent infected cells of n=3
independent experiments +/- SD. 4X magnification.
(B) DIC images corresponding to the data shown in Fig. 5 are presented. We have also
again provided the images from Fig. 5 to facilitate evaluation.
Fig. S4 Depletion of the EMC stops ZIKV early in replication, related to Fig. 6
(A) H1-HeLa cells were transfected with pooled Silencer Select siRNAs (three siRNAs
per gene) targeting the indicated EMC components for 72 hr. and then incubated with
ZIKV MR766 on ice (moi 50). The viral supernatant was removed and warm media was
added at time 0. Cells were treated at the indicated time points post warming, with (+) a
one minute incubation with trypsin (T) to remove surface bound virus. The cells were
then fixed, permeabilized and processed to detect vRNA using a ViewRNA probe set
(red). Cell plasma membranes are outlined in white based on corresponding DIC
images. Cellular DNA was stained with DAPI (blue). (x63). Scale bar represents 20 M.
(B) Quantitation of ZIKV RNA signals (foci) normalized to the NT control is provided
(normalized vRNA mean foci SD) for 12 cells from each of n = 3 experiments. MPI =
min post-infection.
(D) Time zero images without typsin treatment (- T) matched to the data provided in (A).
H1-HeLa cells were transfected with pooled Silencer Select siRNAs (three siRNAs per
gene) targeting the indicated EMC components for 72 hr. and then incubated with ZIKV
MR766 on ice (moi 50). Cells were then washed, fixed, permeabilized and processed to
detect E protein (E) using the 4G2 antibody (green). Cellular DNA was stained with
DAPI (blue). (x63). Scale bar represents 20 M.
(D) Quantitation of ZIKV E signal intensity normalized to the NT control from (C) is
provided (normalized E mean intensity SD) for 12 cells from each of n = 3
experiments.
Fig. S5 Loss of AXL stops ZIKV early in replication but does not alter viral binding
to the cell surface, related to Fig. 6
(A) H1-HeLa cells were transfected in parallel with three independent Silencer Select
siRNAs (not pooled) targeting AXL for 72 hr. and then incubated with ZIKV MR766 on
ice (moi 50). The viral supernatant was removed and warm media was added at time 0.
At the indicated time points, the cells were fixed, permeabilized and processed to detect
ZIKV RNA using a ViewRNA probe set (red). Cell plasma membranes are outlined in
white based on corresponding DIC images. Cellular DNA was stained with DAPI (blue).
(x63). Scale bar represents 20 M.
(B) Quantitation of ZIKV RNA signal intensity normalized to the NT control is provided
(normalized vRNA mean intensity SD) for 12 cells from each of n = 3 experiments.
Legend indicates the minutes post-infection (post warming).
(C) Cells matched to those in (A), plus those transfected with pooled Silencer Select
siRNAs against EMC4 or EMC5 for 72 hr, were treated as in (A) but were washed and
treated with (+ T) or without trypsin (- T), then fixed, permeabilized and immunostained
for viral E protein (green). Cellular DNA was stained with DAPI (blue). (x63). Scale bar
represents 20 M.
(D) Quantitation of ZIKV E signal intensity normalized to the NT control for the cells in
(C) is provided (normalized E mean intensity SD) for 12 cells from each of n = 3
experiments.
(E) Immunoblot of the cells in (A) using the indicated antibodies. ACTB = Actin.
Fig S6 EMC depletion does not decrease AXL or heparin sulfate levels at the cell
surface, related to Fig. 5,6
(A, B) H1-HeLa cells were transfected with the indicated pooled Silencer Select siRNAs
targeting EMC 1, EMC2, EMC4, EMC5, NDST1 (positive control), and non-targeting
siRNA (NT, negative control) for 72 hr. Cells were then treated with cell dissociation
buffer (Gibco) and incubated at 37C for 10 minutes. Samples were then either stained
with an isotype matched IgG FITC conjugated antibody or with either anti-Heparin
Sulfate FITC (US Biological, H1890-10, A) of anti-AXL (B) FITC antibody for 30 min. on
ice. Samples were then washed three times with cold buffer and fixed with cold 1%
formalin in PBS. Samples were finally washed and re-suspended in MACS buffer. Flow
cytometry data was acquired on MACSQuant Analyzer using blue laser 488 nm and
filter 585/40 nm. Data was analyzed using FlowJo V10.
Supplemental Table Legends
Table S1. DENV-HF MORR primary and validation round screen data and siRNAs
used in this work, related to Fig. 1
The normalized data for the DENV-HF primary screens (Silencer Select, SMART-Rev,
and SMARTpool) is provided. The expression data from the a HeLa cell Affymetrix
microarray are also provided, as well as the Affy probe ID and the net expression value
(gene value minus the intron probe mean value). The red highlight indicates that the net
expression value for that probe set was negative, suggesting that the gene was not well
expressed under these conditions. The information and sequences for the siRNAs used
for follow-up experiments are provided
Table S2. Gene enrichments, false negatives, and exact gene overlap information,
related to Fig. 1
The collective primary screen data for the EMC, the OST complex, and the elongation
complex components from the DENV-HF MORR screens is provided. Genes were
designated as scoring in the primary screen if their normalized percent infectivity was <
0.5 (50%) of the plate mean and/or if one or more siRNAs from the validation round
performed similarly and their cell number was >0.5 (50%) of the plate mean (highlighted
in blue if they fall below this toxicity cutoff). The total number of times a gene scored
across all 3 screens is noted (green heat map highlights). RIGER3 WS scores and
combined p values are provided for each gene. Affy probe, expression, and net
expression values are also provided for each gene. DENV-HF false negative analyses:
The primary screen data for the gene sets, as well as the estimated percentage of false
negatives, are provided for each the DENV-HF MORR screens. For these analyses, any
component that scored in one or more primary screen datasets was considered an ontarget candidate, and, as such, its not being detected in a screen represents a false
negative for that effort. True negatives are genes that did not meet selection criteria in
any screen but are expressed in HeLa cells based on microarray data. Comparison of
the exact gene overlap (DFs or CFs) across the primary screen datasets from the
DENV-HF MORR screens is provided. Genes were designated as scoring in the primary
screen (black) if their normalized percent infectivity was < 0.5 (50%). Exact genes found
in two or more of the candidate lists are shown in red. The common DENV-DFs (all 3
screens) or common DENV-DFs (all 3 screens) that scored in two or more screens are
provided on the right.
Table S3. RIGER analyses, RIGER gene enrichments, AUC comparisons, RIGER
compared with the individual MORR screens, related to Fig. 4
RIGER3 DENV-DF or -CF analysis of the SMARTpool, Silencer Select, and SMARTRev screens. The mean cell number across the three datasets is provided. Both the
separate (WS, KS, and SB) and combined p values are provided. Microarray geneexpression data (Affy expression value) and probe ID are given. RIGER3 DENV-CF and
-DF gene enrichments: Enriched pathways and complexes for the analyses are
provided. AUC comparison data: The DENV-HF screen gene test sets are noted. AUC
analyses for the SMARTpool, and Silencer Select libraries and the RIGER3 dataset are
given for each gene test set. The last column contains two different AUC values; the
first is the AUC of the entire library, and the second is the AUC value for the top 60%
highest ranked candidates from each respective library. For the top 150 RIGER3 DFs
and CFs, the data are provided from the primary literature regarding cellular function
(PMID citations), as well as cell number and expression level in HeLa cells based on
microarray analysis. For the included genes, PubMed identification numbers are
provided (Citations [PMID]) to support their designated location and putative function in
the cell and potentially in the viral lifecycle. Genes were also excluded if insufficient data
were available regarding their cellular function.
Table S4 ZIKV-HF CRISPR/Cas9 screen results, related to Fig. 3
H1-HeLa cells expressing Cas9 and containing the human GeCKO v2.0 sgRNA library
were challenged with ZIKV strain MR766 in two independent screens. For each screen,
genomic DNA was prepared from the surviving population, and chromosomally
integrated sgRNAs were PCR-amplified and sequenced using the Ion Torrent
sequencing platform. Reads were mapped to the GeCKO v2.0 library using bowtie2
software on UMass Medical School's Green Hill High Performance Computing Cluster
(GHPCC) and analyzed with proprietary Matlab scripts. For analysis, sgRNAs detected
by 6 or more reads were selected, and genes were considered for further analysis only
when detected by at least 3 independent sgRNAs. Reads remaining after applying the
above selection criteria are listed individually for each screen in two tabs on this table
(screen 1, screen2). The column abbreviations are as follows:
gene_symbol official gene symbol
sgRNA_ID_gene_symbol string combining sgRNA ID and official gene symbol
sgRNA_reads_libA sgRNA read numbers obtained by screening library A
sgRNA_reads_libB sgRNA read numbers obtained by screening library B
reads_per_gene
sum of all reads mapped to sgRNAs targeting one gene
indep_sgRNAs
number of recovered independent sgRNAs targeting one gene
gene_symbol_unique
official gene symbol (iterated for algorithm control purposes)
gene_ID_NCBI
NCBI gene ID
gene_ID_UniProt
UniProt gene ID
gene_name_NCBI NCBI gene name

gene_name_UniProt
UniProt gene name
To integrate results, reads from both screens were pooled and reads mapping to the
same sgRNA summed up, as listed in the tab 'ZIKV screen sum'. Abbreviations are as
given above, plus one column reporting the sum of all reads corresponding to one
sgRNA.
Table S5 EMC AP-MS, Related to Fig. 7
Proteins with multiple spectral counts specific for the EMC1 (tab A), EMC2 (tab B),
EMC4 (tab C), EMC5 (tab D) cell lines versus the vector control cell line are provided.
Proteins that are represented in multiple EMC-AP-MS experiments (tab E). Enriched
pathways and complexes found using ConsensusPath DB and identified proteins in
(A),(B),(C), and (D, tab F).
Table S6 DENV-HF MORR screen comparisons, Related to Fig. 1, 2
Summary table of host factors (HF) found to overlap in the comparison tables described
below. The Excel spreadsheet tab for each screen comparison is denoted as (T #).
Drosophila DENV-HF RNAi screen (Sessions et al., 2009) vs. MORR DENV-HF primary
dataset. All identifiable human homologs are cross referenced to the provided primary
normalized percent infection, cell number, ranking, and microarray gene-expression
data for the MORR DENV-HF screens. RIGER3 data for the DENV-HF MORR screens,
given at the far right, includes p-value (WS dependency), average normalized percent
infection, average normalized cell number, and ranking based on p-value. Red
highlighting denotes 0.5 normalized percent infection values, blue denotes a
normalized cell number 0.4 (toxic), and yellow denotes negative net expression values
(values less than the median of the intron probe set). Genes that do not appear in the
screens do not have values. Genes considered hits in any of the primary screens are
marked as green. A summary of the overlapping hits is given in the first column.
Tab 3: YFV-HF RNAi screen (Le Sommer et al., 2012) vs. MORR DENV-HF primary
dataset. The YFV RNAi screen data is provided, consisting of 395 host factors ranked
by p-value (column P). We provide similar primary screen and RIGER3 data analysis as
in (T2).
Tab 4: Yeast phospholipid (PL) transfer screen (Lahiri et al., 2014) vs. MORR DENV-HF
primary data is shown from a S. cerevisiae synthetic genetic array (SGA) screen for the
CHO2 gene in the absence and presence of choline to identify genes that functioned in
PL transfer between the ER and the mitochondria. Human homologues were identified
and compared to MORR DENV-HF screen as in (T2).
Tab 5: Kinome DENV-HF screen (Kwon et al., 2014) vs MORR DENV-HF and Tab 6:
Kinome DENV-CF vs MORR DENV-HF primary dataset. These screens detected 22
DFs and 8 CFs which modulated DENV infection. Kinome screen HFs are ranked based
on normalized percent infection. DFs and CFs are in tabs 5 and 6 respectively. Primary
screen data is ranked according to normalized percent infectivity; lower infectivity
resulted in higher ranking. DFs were analyzed similar to (T2). Top CFs in primary
screens have higher normalized percent infectivity and, thus, lower ranking. Rankings
for RIGER3 CFs reflect ranking based on RIGER3 p-values (WS competitive).
Tab 7: WNV-HF CRISPR/Cas9 (Ma et al., 2015) vs MORR DENV-HF primary screen
dataset is provided for genes with 2000 reads and 3 gsRNA in the second round of
screening from (Ma et al., 2015). Ranking was based on total reads. We provided
similar primary screen and RIGER3 data analysis as in (T2).
Tab 8: WNV-HF siRNA vs MORR DENV-DFs and Tab 9: WNV siRNA vs MORR DENVCFs Primary data from a WNV RNAi screen (Krishnan et al., 2008) detects host factors
required for only WNV or both WNV and DENV. DFs and CFs were analyzed as in (T5)
and (T6) respectively.
Tab 10: Yeast screen for genes needed for ER protein folding (Jonikas et al., 2009) vs.
DENV-HF Silencer Select Screen. We identified the highest confidence human
homologs of the top 200 S. cerevisiae genes from (Jonikas et al., 2009) responsible for
inducing the UPR (based on provided log 2 reporter levels in the deletion strain).
Human homologs were identified using the DIOPT-DRSC Integrative Ortholog
Prediction Tool from the Drosophila RNAi Screening Center (DRSC). S. cerevisiae
genes were ranked according to log2 GFP/RFP reporter levels and compared to the top
200 hits of the DENV-HF Silencer Select screen (this work). Overlapping genes are
denoted in green.
Tab 11: DENV-HF Silencer Select vs Yeast ER protein folding screen. Comparison of
all identifiable yeast homologs from the top 200 hits of the DENV-HF Silencer Select
screen and the entire S. cerevisiae UPR screen dataset. Yeast log2 GFP/RFP reporter
levels are noted, as well as whether or not deletion of the gene induces the UPR (yes or
no). Genes that were hits in the primary Silencer Select screen and that induced the
UPR in (Jonikas et al., 2009) are marked as hits in green.
Tab 12: ZIKV-HF CRISPR/Cas9 screen vs. Yeast ER protein folding screen (Jonikas et
al., 2009). Comparison of all identifiable yeast homologs from the top 40 hits of the
ZIKV-HF CRISPR/Cas9 screen and the entire S. cerevisiae UPR screen dataset. Yeast
log2 GFP/RFP reporter levels are noted, as well as whether or not deletion of the gene
induces the UPR (yes or no). Genes that were hits in the primary Silencer Select screen
and that induced the UPR in (Jonikas et al., 2009) are marked as hits in green.
Table S7 ZIKV-HF CRISPR/Cas9 Screen Comparisons, Related to Fig. 3, 4
Data for the top 35 ZIKV CRISPR/Cas9 screen hits are compared to data from other
screens we have performed:
the MORR DENV-HF Screens (Silencer Select,
SmartPool, SMARTRev this work), MORR HRV-HF Screens (Silencer Select,

SmartPool, esiRNA, SmartRev) and MORR HDF Screens (Silencer Select, SmartPool,
esiRNA) is provided. Red highlighting indicates a hit that overlaps between the ZIKVHF CRISPR/Cas9 screen and a given screen. A summary table of overlap genes is
also provided.
Data for the top 35 ZIKV-HF CRISPR/Cas9 screen hits are compared to data from other
flavivirus screens specifically, WNV siRNA Screen (Krishnan et al., 2008), WNV
CRISPR/Cas9 screen (Ma et al., 2015), YFV siRNA screen (Le Sommer et al., 2012),
and DENV Drosophila-based screen (Sessions et al., 2009).
Yellow highlighting
indicates a hit that overlaps between the ZIKV-HF CRISPR/Cas9 screen and a given
screen. A summary table of overlap genes is also provided.
Data for the top 35 ZIKV-HF CRISPR/Cas9 screen hits are compared to data from a
CME screen (Kozik et al., 2013) is provided. Yellow highlighting indicates a hit that
overlaps between the ZIKV-HF CRISPR/Cas9 screen and the CME screen. A summary
table of overlap genes is also provided.
Supplemental Methods
Gene enrichment analysis
Functional enrichment analysis was performed using Consensus PathDB-human
(http://cpdb.molgen.mpg.de), Reactome and KEGG databases were utilized to
determine enriched pathways while Reactome and CORUM databases were employed
for enriched protein complex identification.
RNAi reagents
All siRNAs were purchased from either Ambion (Thermo) or Dharmacon (GE) and were
used at 50 nM final concentration using Oligofectamine transfection lipid (Life
Technologies). The control siRNA is a pool of Silencer Negative non-targeting (NT)
siRNAs (No.1-4, AM4611, AM4613, AM4615 and AM4641, Ambion), or siGENOME
Non-Targeting siRNA Control Pools (Dharmacon). Information for gene-specific
followup siRNAs is provided in Supplementary Table 1.
Viral translocation assay
Briefly, HeLa cells were infected with the indicated viruses at the indicated moi at 72 hr
post transfection of siRNAs and processed as previously described (Feeley et al.,
2011). Trypsin treatment and permeabilization were varied as per the text and legends.
Generation of Cas9-expressing H1-HeLa cells and sgRNA library construction.

The human codon-optimized sequence Cas9 from of S. pyogenes was subcloned from
plasmid lentiCas9-Blast (Addgene #52962) into plasmid pQCXIH (Clontech) and
transduced into H1 HeLa cells, which were subsequently expanded in the presence of
200 ug/ml hygromycin B (Life Technologies). Cas9 expression was verified using
Western blot, flow cytometry, and fluorescence microscopy, and Cas9 functionality was
demonstrated by efficient knockdown of select target genes upon stable sgRNA
transduction.
The Cas9-expressing H1-HeLa cell line (subsequently referred to as H1-Cas9) served
as the parental line harboring the human GeCKO v2 sgRNA library in plasmid
lentiGuide-Puro (Shalem, Zhang Science 2014; Addgene #1000000049). The sgRNA
library targets the human genome with 6 sgRNAs per gene. H1-Cas9 cells were
transduced with lentivirus-packaged GeCKO v2 sgRNA library at a MOI of about 0.2.
On average, each sgRNA was overrepresented about 650 times, as calculated from
titration plates that were set up in parallel with the library.
CRISPR/Cas9 ZIKV-HF Screen
For each GeCKO v2.0 library A and B, we plated cells in 15-cm dishes to achieve a 500
to 1000-fold representation on average of each sgRNA in the screened cell population.
The cells were then infected with ZIKV MR766 (MOI 5) and cultured at 37 C for 8 days.
To follow the progress of the infection, cytopathic effect (CPE) was monitored by eye
using light microscopy. Control plates were run in parallel using the H1-HeLa-Cas9 cell
parent population which does not contain the GeCKO v.2 library. 8 days after infection
the majority of cells, >95%, had died. The remaining surviving cells were washed
extensively and transferred to 37 C with fresh medium. The surviving cells were
expanded and genomic DNA prepared. Proviruses containing the sgRNA stably
integrated into each of the surviving cells were amplified and identified from genomic
DNA using PCR and next-gen sequencing using an Ion Torrent sequencer.
sgRNA sequencing and analysis
Ion Torrent sequencing technology was used to characterize the complexity of the
human GeCKO v2 sgRNA library at different stages, i.e. as plasmid library (sgRNA in
plasmid lentiGuide-Puro) after amplification in E. coli, as well as after transduction into
H1-HeLa cells. Likewise, the sgRNA complement in virus-resistant H1 populations was
determined. For sgRNA sequencing from cells, genomic DNA (gDNA) was prepared
using the DNeasy Blood & Tissue Kit. 50 ul PCR reactions were set up with Herculase II
Phusion DNA polymerase (Agilent), and primers placed around the sgRNA integration
site
to
give
final
product
of
~200
bp
(upstream
primer,
5'-
AATGGACTATCATATGCTTACCGTAACTTGAAAGTATTTCG-3'; downstream primer,

5'-TAAAAAAGCACCGACTCGGTGCCACTTTTTCAAG-3').
The
gel-purified
PCR
product was 5'-phosphorylated and ligated to Ion Proton adapters and barcodes, again
gel-purified, quantified by qPCR, and loaded onto the Ion Proton chip. Ion Protongenerated sequencing read files were analyzed on the UMass Green High Performance
Computing Cluster (GHPCC) using the software modules cutadapt, samtools, and
bowtie2, as well as custom-made Linux scripts. Collapsed read lists were further
analyzed with customized scripts in Matlab.
Cells
HeLa H1 (#CRL-1958) and U2OS (#HTB-96) cells were from ATCC. HeLa Magi cells
were from NIH AIDS Repository (#3522). Cells were cultured in DMEM (Sigma) with
10% fetal bovine serum (GIBCO). Cells were grown in complete Dulbeccos Modified
Eagle Media (Sigma) with 10% FBS (Invitrogen) and 2mM L-glutamine (Invitrogen).
Viruses
Zika virus strains were kindly provided by Dr. Robert Tesh at the World Reference
Center for Emerging Viruses and Arboviruses at the University of Texas Medical Branch
in Galveston Texas: FSS13025, isolated from a patient in Cambodia in 2010 (Haddow
et al., 2012), MR766 obtained originally from a Rhesus macaque in Uganda in 1947.
ZIKV Puerto Rico (VR-1843) and DENV2-NGC (VR-1584) were obtained from ATCC.
YF17D was the kind gift of Dr. Bali Pulendran, Emory University. DENV1 and DENV4
are low passage isolates obtained from patients living in Thailand and were kindly
provided by Dr. Oscar Perng, National Cheng Kung University Medical College.
Viruses were propagated in Vero cells (ATCC) (Dick et al., 1952) and the titer
determined by standard plaque assays and immunofluorescence imaging assays for E
protein expression (Brass et al., 2009; John et al., 2013).
Plasmids
Human EMC 1, 2, and 5 cDNAs were cloned in frame with a C terminal HA epitope tag
into the lentiviral vector, pLVX-puromycin. The EMC4 cDNA expression construct was
cloned together with a C terminal FLAG epitope tag. Lentiviruses were packaged using
a published protocol and used to stably transduce the H1-HeLa cells expressing a tetactivator (tet-3G, Clontech).
Immunoblotting
Whole-cell lysates were prepared using Laemmli buffer and resolved by SDS-PAGE,
then transferred to PVDF membrane (Thermo, #88518) and probed with the indicated
antibodies.
Primary:
Anti-HA-7
Sigma (H3663)
Anti-FLAG M2
Anti-TMEM85
Anti-AXL
Anti--Actin
Sigma (F1804)
WB (1:1000)
WB (1:1000)
Abcam (ab184544) WB (1:10,000)
Cell Signaling (8661S)

Sigma (A2228)
WB (1:1000)
WB (1:5000)
Secondary:
Anti-rabbit horseradish peroxidase (HRP)-conjugated antibody (catalog no. 111035-003; Jackson ImmunoResearch)
WB (1:10,000)
Anti-mouse horseradish peroxidase (HRP)-conjugated antibody (catalog no. 115035-003; Jackson ImmunoResearch)
Flow Cytometry
WB (1:10,000)
In order to determine the functional role of EMCs in the context of heparin sulfate, HeLa
H1 cells were transfected with siRNA targeting EMC 1, EMC2, EMC4, EMC5, NDST1
(positive control), and non-targeting siRNA (negative control) for 72 hr. Post-transfected
cells were treated with Cell Dissociation Buffer (Gibco, 13151-014) and incubated at
37C for 10 minutes. Samples were then either stained with an isotype matched IgG
antibody FITC or probed with anti-Heparin Sulfate FITC (US Biological, H1890-10) for
30 minutes on ice. Samples were then washed three times with cold MACS buffer (1X
PBS, 2% FBS, 1 mM EDTA) and fixed with cold 1% formalin in PBS. Samples were
finally washed and re-suspended in MACS buffer. Flow cytometry data was acquired on
MACSQuant Analyzer using blue laser 488 nm and filter 585/40 nm. Data was analyzed
using FlowJo V10.
qPCR assays
RNA was isolated using the RNeasy Plus Mini Kit (Qiagen, 74134). Reverse
transcription and quantitative PCR were performed with the one-step QuantiTect Probe
RT-PCR kit (Qiagen, 204443) and gene specific TaqMan probes (Life Technologies).
The assay was performed on a Light Cycler 480 Thermocycler (Roche) with the
following settings: 50C for 30 min, 94C for 10 min and 45 cycles of 94C for 30
seconds followed by 60C for 1 minute. Expression levels were normalized to the
housekeeping gene (Glyceraldehyde 3-phosphate dehydrogenase, GAPDH) and
calibrated to the samples transfected with non-targeting siRNA using the Cp method.
The assay was performed in triplicate for each sample and the normalized expression
levels were averaged for each sample. Results were expressed as average standard
deviation (SD).
Taqman reagents for the indicated human genes were used as following: KIAA0090
(Assay ID Hs00299327_m1, Cat. # 433118), EMC2 (Assay ID Hs00206458_m1, Cat. #
4331182) TMEM111 (Assay ID Hs00218383_m1, Cat. # 4331182), EMC4 (Assay ID
Hs00212821_m1, Cat. # 4331182), MMGT1 (Assay ID Hs00953953_m, Cat. #
4331182), EMC6 (Assay ID Hs00364020_g1, Cat. # 4331182), AXL (Assay ID
Hs01064444_m1, Cat. # 4331182), NIPAL (Assay ID Hs00398027_m1, Cat. #
4331182), STARD10 (Assay ID Hs00246405_m1, Cat. # 4331182), STT3A (Assay ID
Hs00537619_m1, Cat. # 4331182), VPS45 (Assay ID Hs00273144_m1, Cat. #

4331182), WDR7 (Assay ID Hs00362983_m1, Cat. # 4331182), ZFYVE20 (Assay ID
Hs00223482_m1, Cat. # 4331182) and GAPDH (Assay ID Hs02758991_g1, Cat. #
4331182).
Affinity purification coupled to mass spectroscopy

H1-HeLa cells expressing either EMC1,2,5-HA, EMC4-FLAG or the empty vector alone
were grown to 75% confluence, and expression was induced for 72 hr using 5 ug/uL
doxycycline.
Cells were collected and then washed in cold PBS three times with
centrifuging at 1000 rpm for 1 min each, then lysed in 0.5 ml MCLB (50 mM Tris, pH
7.5, 150 mM NaCl, and 0.5% NP40) containing protease inhibitors (PIs, Sigma) with
shaking at 4C followed by clarification by centrifugation.
Anti-HA agarose beads
(Sigma) or M2-FLAG agarose beads (Sigma) were prepared by washing the beads in
0.1M glycine pH 3.0 followed by washing three times with cold TBS, and a final wash
with MCLB+PI.
The supernatant was mixed with prepared M2-FLAG or Anti-HA
agarose beads (Sigma), and incubated overnight at 4C. The beads were subsequently
washed in cold MCLB+PIs five times followed by two cold PBS washes. The proteins
were then eluted using Laemmli Sample Buffer (62.5 mM Tris pH 6.8, 10% Glycerol,
2%SDS, 5% - mercaptoethanol, 0.004% Bromophenol Blue) and incubated for 5 min
at 95 C. Samples were then loaded onto a 4-20% Tris-Glycine SDS-PAGE gel, and
electrophoresed for approximately 20 min. Coomassie stained bands were excised,
and submitted to the UMMS Mass Spectrometry and Proteomics Facility for
identification. Spectral data from experiments was analyzed using Scaffold 4 analysis
software. A fishers exact test P-value >0.05 was used to determine enriched proteins
in the EMC expressing samples. Interaction networks were generated using Cytoscape
3.1 with the top 20 enriched proteins from each data set.
Enriched pathways and
complexes were generated using the ConsensusPathDB Over-Representation

Analysis tool.
Confocal Imaging Assays
Cells were fixed in 4% formalin (Sigma) in D-PBS (Invitrogen) for 10 min followed by a
wash with D-PBS. Cells were then permeabilized in 0.1% Triton X-100 (Fisher, BP151)
and 0.1% Tween 20 (Sigma, P7949) in D-PBS for 20 min followed by 3 washes in DPBS. Blocking is performed for 30 min in 1%BSA (BioPharm, 71-010) in D-PBS
containing 0.3M glycine (Sigma, G7126). Cells were next incubated in primary
antibodies (see Antibodies) diluted in 1%BSA in D-PBS for at least 1 hr, followed by 3
washes with D-PBS. Cells were then incubated in secondary antibodies (see
Antibodies) diluted 1:1000 in D-PBS containing 1% BSA. Cells were washed 3 times
with D-PBS before mounting to slides using Vectashield either with DAPI (4, 6diamidino-2-phenylindole)
(Vector
Laboratories,
H-1200)
or
without
(Vector
Laboratories, H-1000). Samples were then imaged using confocal microscopy using a
Nikon A1 inverted microscope using a pinhole of 0.9AU. All microscope settings were
kept constant throughout the entirety of a given experiment. Image analysis and
quantification were done using the FIJI software.
Antibodies used for Immunofluorescence Imaging Assays

Primary antibodies used in this study are as follows: 4G2 (ATCC HB-112 hybridoma
supernatant 1:2), Anti-dsRNA (Kerafast ES2001, 1:50), Anti-LBPA (BMP, Echelon, ZPLBPA, 1:500), Anti-HA (Sigma-Aldrich, H6908, 1:500), Anti-Flag (Novus, NB600-344,
1:1,000).
Secondary antibodies used in this study are as follows: Alexa Fluor 488 goat antimouse IgG (H+L) (Life Technologies, A11001), Alexa Fluor 488 goat anti-rabbit IgG
(H+L) (Life Technologies, A11034), Alexa Fluor 488 donkey anti-goat IgG (H+L) (Life
Technologies, A11055).
Fluorescence Intensity Quantification

Fluorescence intensity is quantified using the FIJI software. The intensity is measured
as the average fluorescence intensity per cell. First, the image file was opened in the
FIJI software and the cell to be measured was outlined with the freehand selections
tool to define the ROI (region of interest) using the DIC image as a guide. After setting
the ROI, the intensity was quantified using the measure tool (selected from
Analyze>Measure). Next a region outside of the cells was selected to determine the
background fluorescence of the image. The measured values were then recorded into a
Microsoft Excel file for further analysis. The intensity value measured for the cell had the
background intensity subtracted from it to get the net intensity for the cell. Multiple cells
for each condition were measure and the average standard deviation of the net
intensity was calculated for each condition across the experiment. To make differences
across the conditions easier to appreciate the average net intensity values were then
normalized to the control condition of each experiment by simply dividing the average
and standard deviation for each condition by the average of the control condition. Data
was then represented graphically using bar graphs in Microsoft Excel.
Zika Virus E Viral RNA Signal (Foci) Quantification:

12-bit multichannel images were opened using the free online software program FIJI.
The cell of interest was outlined using the freehand selections tool using the DIC
channel as a guide. The threshold was set to a minimum of 900 and maximum of 4096
for the viral staining (E protein or viral RNA). The number of individual foci was
quantified using the analyze particles tool (Analyze>Analyze Particles). The size range
of counted foci was 1-infinity pixels, to remove background in the quantification. Foci
counts per cell were recorded for all cells in each condition (>12 cells per condition per
experiment) in a Microsoft Excel document and represented as a bar graph using the
mean and standard deviation.
ViewRNA Assays
Prior to staining, the hybridization oven (Illumina) was set to 40C. The probe set was
placed on ice and allowed to thaw. Coverslips were transferred to a 100mm round
petridish (Fisher Cat No. 08-757-13) labeled and lined with parafilm. 50 L of D-PBS
was added to each coverslip immediately to prevent them from drying out. The
Affymetrix 1X detergent solution was mixed and prepared with D-PBS. D-PBS on each
coverslip was aspirated off and samples were incubated with the detergent solution for
5 min. at room temperature before being washed twice with D-PBS. Affymetrix Protease
was diluted at 1:1000 in D-PBS and the coverslips were incubated for 15 min. at room
temperature. After two washes, the probe set was diluted (see probe details below) in
pre-warmed (40C) probe set diluent and added to the coverslips. Samples were then
placed in the 40C humidified hybridization chamber for 3h. During this time, the mixed
probe set PreAmp, Amp and Label Probe (wrapped in aluminum foil) were thawed on
ice. PreAmp and Amp are diluted in Amplification Diluent at 1:25 just prior to application
(during the 3 washes in between each step). After 3 washes with ViewRNA Wash Buffer
(Affymetrix), coverslips were treated with the pre-warmed (40C) PreAmp solution for 30
min. in the 40C hybridization chamber. This is followed up by 3 washes with ViewRNA
wash buffer and incubation in pre-warmed (40C) Amp solution for 30 min. Label Probe
was diluted in Label Probe Diluent at 1:25, then added to coverslips for 30 min. in the
hybridization oven. The coverslips were washed 3 times in ViewRNA wash buffer, and
left in the last wash for 10 min. The coverslips were protected from the light as much as
possible during this time. The coverslips are then mounted using VectaShield mounting
media with DAPI (Vector Laboratories, H-1200).
ViewRNA Probe Sets:
Dengue Virus VF1-10726 use at 1:25
Zika Virus Polyprotein VF1-19981 use at 1:50
Supplemental References
Brass, A.L., Huang, I.C., Benita, Y., John, S.P., Krishnan, M.N., Feeley, E.M., Ryan,
B.J., Weyer, J.L., van der Weyden, L., Fikrig, E., et al. (2009). The IFITM proteins
mediate cellular resistance to influenza A H1N1 virus, West Nile virus, and dengue
virus. Cell 139, 1243-1254.
Dick, G.W., Kitchen, S.F., and Haddow, A.J. (1952). Zika virus. I. Isolations and
serological specificity. Transactions of the Royal Society of Tropical Medicine and
Hygiene 46, 509-520.
Haddow, A.D., Schuh, A.J., Yasuda, C.Y., Kasper, M.R., Heang, V., Huy, R., Guzman,
H., Tesh, R.B., and Weaver, S.C. (2012). Genetic characterization of Zika virus strains:
geographic expansion of the Asian lineage. PLoS neglected tropical diseases 6, e1477.
John, S.P., Chin, C.R., Perreira, J.M., Feeley, E.M., Aker, A.M., Savidis, G., Smith, S.E.,
Elia, A.E., Everitt, A.R., Vora, M., et al. (2013). The CD225 domain of IFITM3 is required
for both IFITM protein association and inhibition of influenza A virus and dengue virus
replication. Journal of virology 87, 7837-7852.
Jonikas, M.C., Collins, S.R., Denic, V., Oh, E., Quan, E.M., Schmid, V., Weibezahn, J.,
Schwappach, B., Walter, P., Weissman, J.S., et al. (2009). Comprehensive
characterization of genes required for protein folding in the endoplasmic reticulum.
Science 323, 1693-1697.
Kozik, P., Hodson, N.A., Sahlender, D.A., Simecek, N., Soromani, C., Wu, J., Collinson,
L.M., and Robinson, M.S. (2013). A human genome-wide screen for regulators of
clathrin-coated vesicle formation reveals an unexpected role for the V-ATPase. Nature
cell biology 15, 50-60.
Krishnan, M.N., Ng, A., Sukumaran, B., Gilfoy, F.D., Uchil, P.D., Sultana, H., Brass,
A.L., Adametz, R., Tsui, M., Qian, F., et al. (2008). RNA interference screen for human
genes associated with West Nile virus infection. Nature 455, 242-245.
Kwon, Y.J., Heo, J., Wong, H.E., Cruz, D.J., Velumani, S., da Silva, C.T., Mosimann,
A.L., Duarte Dos Santos, C.N., Freitas-Junior, L.H., and Fink, K. (2014). Kinome siRNA
screen identifies novel cell-type specific dengue host target genes. Antiviral research
110, 20-30.
Lahiri, S., Chao, J.T., Tavassoli, S., Wong, A.K., Choudhary, V., Young, B.P., Loewen,
C.J., and Prinz, W.A. (2014). A conserved endoplasmic reticulum membrane protein
complex (EMC) facilitates phospholipid transfer from the ER to mitochondria. PLoS
biology 12, e1001969.
Le Sommer, C., Barrows, N.J., Bradrick, S.S., Pearson, J.L., and Garcia-Blanco, M.A.
(2012). G protein-coupled receptor kinase 2 promotes flaviviridae entry and replication.
PLoS neglected tropical diseases 6, e1820.
Ma, H., Dang, Y., Wu, Y., Jia, G., Anaya, E., Zhang, J., Abraham, S., Choi, J.G., Shi,
G., Qi, L., et al. (2015). A CRISPR-Based Screen Identifies Genes Essential for WestNile-Virus-Induced Cell Death. Cell reports 12, 673-683.
Sessions, O.M., Barrows, N.J., Souza-Neto, J.A., Robinson, T.J., Hershey, C.L.,
Rodgers, M.A., Ramirez, J.L., Dimopoulos, G., Yang, P.L., Pearson, J.L., et al. (2009).
Discovery of insect and human dengue virus host factors. Nature 458, 1047-1050.

Zika Virus and Dengue Virus Genomics PDF

Uploaded by

Document Information

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Zika Virus and Dengue Virus Genomics PDF

Uploaded by

Copyright:

Available Formats

Resource

Identification of Zika Virus and Dengue Virus

RNAi and CRISPR/Cas9 screens were used to find flavivirus

The screens recovered host factors involved in endocytosis

The EMC is required by DENV and ZIKV in the early stages of

These studies give a systems-wide view of human-flavivirus

Savidis et al., 2016, Cell Reports 16, 115

The flaviviruses dengue virus (DENV) and Zika virus

(legend continued on next page)

2 Cell Reports 16, 115, July 19, 2016

undergo exocytosis, thus spreading the infection to neighboring cells.

mulative comprehensive coverage. A comparison of >1,000

Cell Reports 16, 115, July 19, 2016 3

4 Cell Reports 16, 115, July 19, 2016

negative rates ranging from 25% to 50% (Table S2). We generated

Figure 2. The EMC Is Needed for DENV Replication

Cell Reports 16, 115, July 19, 2016 5

(legend continued on next page)

6 Cell Reports 16, 115, July 19, 2016

et al., 2013; Meertens et al., 2012; Perera-Lecoin et al., 2014),

ing synchronized time course infectivity experiments. To monitor

Cell Reports 16, 115, July 19, 2016 7

(legend continued on next page)

8 Cell Reports 16, 115, July 19, 2016

Figure 5. Loss of the EMC Halts DENV Early

presence of an anti-AXL antibody using a flow-cytometry-based

These studies showed that, at 80 min

Cell Reports 16, 115, July 19, 2016 9

Figure 6. ZIKV Remains at the Cell Surface

using additional assays such as flavivirus replicon-expressing

10 Cell Reports 16, 115, July 19, 2016

coupled to mass spectrometry (AP-MS).

EMC7, EMC8, and EMC10. These data suggest that EMC1,

Cell Reports 16, 115, July 19, 2016 11

Figure 7. Identification of EMC-Interacting Proteins and Pathways using AP-MS

(legend continued on next page)

12 Cell Reports 16, 115, July 19, 2016

The Role of AXL in Flavivirus Replication

The Role of the EMC in Flavivirus Infection

Cell Reports 16, 115, July 19, 2016 13

14 Cell Reports 16, 115, July 19, 2016

Cell Reports 16, 115, July 19, 2016 15

Cell Reports, Volume 16

Identication of Zika Virus and Dengue Virus

Exon Junction Complex

Large Drosha Complex

DENV MORR Silencer Select

DENV MORR/RIGER Top 1000

EMC1 EMC2 EMC4 EMC5

NT EMC4 EMC5 AXL-1 AXL-2 AXL-3

AXL-1 AXL-2 AXL-3

Normalized vRNA Intensity

sum of all reads mapped to sgRNAs targeting one gene

number of recovered independent sgRNAs targeting one gene

official gene symbol (iterated for algorithm control purposes)

gene_name_NCBI NCBI gene name

UniProt gene name

the MORR DENV-HF Screens (Silencer Select,

SmartPool, SMARTRev this work), MORR HRV-HF Screens (Silencer Select,

Generation of Cas9-expressing H1-HeLa cells and sgRNA library construction.

AATGGACTATCATATGCTTACCGTAACTTGAAAGTATTTCG-3'; downstream primer,

Abcam (ab184544) WB (1:10,000)

Cell Signaling (8661S)