JVI Accepted Manuscript Posted Online 20 January 2016 J. Virol. doi:10.1128/JVI.02827-15 Copyright © 2016 Baer et al This is an open-access article distributed under the terms of the Creative Commons Attribution 4.0 International license.
1
Venezuelan Equine Encephalitis Virus Induces Apoptosis through the Unfolded Protein
2
Response Activation of EGR1
3 4
Alan Baer1, Lindsay Lundberg1, Danielle Swales2, Nicole Waybright2, Chelsea Pinkham1,
5
Jonathan D. Dinman3, Jonathan L. Jacobs2#, Kylene Kehn-Hall1#
6 7 8 9 10
1
National Center for Biodefense and Infectious Diseases, School of Systems Biology, George
Mason University, Manassas, Virginia, USA 2
MRIGlobal, Global Health Security, Rockville, Maryland, USA
3
University of Maryland, Department of Cell Biology and Molecular Genetics, College Park,
Maryland, USA
11 12
#Address correspondence to Kylene Kehn-Hall,
[email protected] and Jonathan Jacobs,
13
[email protected]
14 15
Running Head: Host-Pathogen Interactions in VEEV Infection
16 17
Abstract word count: 230
18
Text word count: 7,677
19 20 21 22 1
23
Abstract
24
Venezuelan equine encephalitis virus (VEEV) is a previously weaponized arthropod-borne virus
25
responsible for causing acute and fatal encephalitis in animal and human hosts. The increased
26
circulation and spread in the Americas of VEEV, and other encephalitic arboviruses such as
27
Eastern equine encephalitis virus and West Nile virus, underscores the need for research aimed at
28
characterizing the pathogenesis of viral encephalomyelitis for the development of novel medical
29
countermeasures. The host-pathogen dynamics of VEEV-TrD infected human astrocytoma
30
U87MG cells were determined by carrying out RNA sequencing (RNA-Seq) of poly(A) +
31
mRNAs. To identify critical alterations in the host transcriptome that take place following
32
VEEV infection, samples were collected at 4, 8, and 16 hours post-infection and RNA-Seq data
33
acquired using the Ion Torrent PGM platform. Differential expression of interferon responsive
34
genes, stress response factors, and components of the unfolded protein response (UPR) were
35
observed. The PERK arm of the UPR was activated as both ATF4 and CHOP (DDIT3), critical
36
regulators of the pathway, were altered after infection. The transcription factor Early Growth
37
Response 1 (EGR1) was induced in a PERK dependent manner. EGR1 -/- MEFs demonstrated
38
lower susceptibility to VEEV induced cell death as compared to isogenic wild type MEFs,
39
indicating that EGR1 modulates pro-apoptotic pathways following VEEV infection. The
40
influence of EGR1 is of great importance as neuronal damage can lead to long-term sequelae in
41
individuals who have survived VEEV infection.
42 43 44
2
45
Importance
46
Alphaviruses represent a group of clinically relevant viruses transmitted by mosquitoes to
47
humans. In severe cases, viral spread targets neuronal tissue, resulting in significant and life
48
threatening inflammation dependent on a combination of viral-host interactions. Currently there
49
are no therapeutics for encephalitic alphaviruses due to an incomplete understanding of their
50
molecular pathogenesis. Venezuelan equine encephalitis virus (VEEV) is an alphaviruses
51
prevalent in the Americas capable of infecting horses and humans. Here we utilized next
52
generation RNA sequencing to identify differential alterations in VEEV infected astrocytes. Our
53
results indicated that transcripts associated with interferon and the unfolded protein response
54
pathways were altered following infection, and demonstrated that Early Growth Response 1
55
(EGR1) contributed to VEEV induced cell death.
56 57
Introduction
58
Venezuelan equine encephalitis virus (VEEV) is a New World alphavirus in the family
59
Togaviridae that is endemic to the Americas. VEEV is a positive strand RNA virus, transmitted
60
by mosquitoes and is naturally present in rodent reservoirs (1). There are six subtypes that are
61
categorized by their geographic range and pathology in equines and humans. The two epizootic
62
strains, IA/B and IC, arose from mutations among the enzootic strains (2). The IA/B and IC
63
strains are of particular concern due to increased rates of morbidity, mortality, and the risks
64
associated with viral amplification and potential species spillover (2). In humans VEEV causes a
65
febrile illness typified by fever, malaise, and vomiting. In some cases, infection progresses to the
66
central nervous system (CNS) and neurological symptoms such as confusion, ataxia, and seizures
67
manifest. Neurological cases have a mortality rate as high as 35% in children and 10% in adults, 3
68
with long-term neurological deficits often seen in survivors (2). In 1995, an outbreak of VEEV in
69
Colombia and Venezuela resulted in over 100,000 human cases (3). In addition to natural
70
outbreaks, VEEV is also a concern from a bioterrorism perspective as it can be grown to high
71
titers, requires a low infectious dose, and contains multiple serotypes. Both the former Soviet
72
Union and the United States previously weaponized the virus, producing large quantities for their
73
now defunct offensive bioweapons programs (4). Currently the vaccine strain TC83 is used in
74
horses and for high risk personnel, however due to its low rate of seroconversion (5) and a
75
reliance on two single attenuating mutations (6), it is considered unfit for mass distribution (7).
76
To date there are no FDA approved therapeutics for VEEV and further studies are required for
77
clarification of the mechanisms associated with its underlying pathogenesis.
78
Viral and host transcriptomic studies can provide a wealth of information on the underlying
79
pathogenic mechanisms and interactions following the course of an infection. The use of high-
80
throughput next generation sequencing has led to the discovery of previously uncharacterized
81
viruses and the establishment of numerous novel experimental systems redefining viral-host
82
interactions. To date there has been a number of studies examining alterations in the host
83
transcriptome following VEEV infection. A comparative microarray analysis between cells
84
persistently infected with VEEV and cells able to clear VEEV resulted in the identification of
85
PARP12L as an antiviral factor (8). A molecular comparison utilizing microarrays among host
86
based responses to the TC83 strain was able to identify biomarkers differentiating between
87
vaccine responder and nonresponder groups, as well as the involvement of interferon, interferon-
88
induced pathways, toll like receptor (TLR), and interleukin 12-related pathways (9). A study
89
examining the role of adhesion and inflammatory factors in VEEV infected CD-1 mice found
90
viral modulation of extracellular matrix and adhesion genes such as integrins (ItgαX, Itg2, 3, and 4
91
7), cadherin 1 and 2, vascular cell adhesion molecule-1, and intracellular adhesion molecule-1
92
(ICAM-1) in the brains of VEEV infected mice (10). Follow-up experiments utilizing ICAM-1
93
knockout mice demonstrated reduced inflammation in the brain and a subsequent delay in the
94
onset of neurological sequale (10). A study by Sharma et al. utilizied microarrays to analyze
95
gene expresion changes in the brain tissue of VEEV infected mice over the course of an
96
infection, discovering numerous immune pathways involved in antigen presentation,
97
inflammation, apoptosis and the traditional antiviral response (Cxcl10, CxCl11, Ccl5, Ifr7, Ifi27,
98
Oas1b, Fcerg1,Mif, Clusterin and MHC class II) (11). A second study by the same group
99
identified regulation of miRNAs in the brains of VEEV infected mice which enabled the
100
correlation of the miRNA changes with earlier mRNA expression data (11, 12). These analyses
101
suggest that VEEV may be utilzling cellular miRNAs in order to regulate downstream mRNA,
102
which may correspond with VEEV induced histological changes to the nervous system (11, 12).
103
In the current study, next generation RNA sequencing (RNA-Seq) was used to identify clinically
104
relevant alterations in the mRNA transcriptome of human astrocytes infected with wild type
105
VEEV- Trinidad Donkey (TrD). The analysis of host mRNA’s by RNA-Seq provides novel
106
insight into how a host responds to a viral infection through the identification of a wide and
107
dynamic range of transcripts in an unbiased manner. Selectively sequencing mRNA’s,
108
specifically polyadenylated (poly-A) transcripts, which are responsible for encoding the genome
109
and account for ~1% of the entire transcriptome, enhances the detection of the most relevant and
110
low abundant transcripts (13). As VEEV has been shown to productively infect astrocytes both
111
in vitro and in vivo (14, 15) we chose astrocytes as our model of interest. Astrocytes are the most
112
abundant cell in the brain, outnumbering neurons by at least fivefold (16), providing an abundant
113
resource for viral replication within the brain. In addition to their well described structural role in 5
114
neuronal tissue, astrocytes play critical roles in other processes including regulation of blood
115
flow and of the blood brain barrier, synapse transmission, and response to infection (16). VEEV
116
infected astrocytes have been shown to produce multiple cytokines including IL-8, IL-17, IFN-
117
gamma and IP-10, all of which were associated with viral attenuation (14).
118
In order to obtain a dynamic view of the viral-host interactome, RNA-Seq was used to monitor
119
changes in gene expression in VEEV-TrD infected astrocytes at 4, 8 and 16 hours post infection.
120
By viewing alterations among multiple early time points using triplicate biological replicates, a
121
robust and dynamic range of information is generated, providing an increase in both the power
122
and accuracy of detection for differentially expressed transcripts in a highly relevant clinical
123
model (17). Among VEEV infected cells, an increase in interferon regulated genes including
124
IFIT1, IFIT2, IFIT3, and OASL was observed. Increased expression of genes involved in the
125
stress induced unfolded protein response (UPR) pathway was also noted. Interestingly, VEEV
126
infection resulted in an increase in early growth response protein 1 (EGR1), which may serve as
127
a link between the two pathways. The identification of altered host mRNAs, specifically EGR-1
128
and its up and downstream interactors following VEEV replication, may provide novel host
129
based therapeutic targets critical for VEEV replication, while providing a greater understanding
130
of the underlying mechanisms underpinning alphavirus replication.
131
132
Materials and Methods
133
Viral infections and plaque assays. VEEV TrD was obtained from BEI Resources. All
134
experiments with VEEV TrD were performed under BSL-3 conditions. All work involving select
135
agents is registered with the Centers for Disease Control and Prevention and conducted at 6
136
George Mason University’s Biomedical Research Laboratory, which is registered in accordance
137
with Federal select agent regulations. For infections, VEEV was added to supplemented DMEM
138
to achieve an MOI of 0.05, 0.5 or 5. Cells were infected for one hour at 37°C and rotated every
139
fifteen minutes to ensure adequate coverage. Cells were then washed with PBS and complete
140
growth media added back to the cells. Viral supernatants and cells were collected at various
141
times post infection for further analysis. Plaque assays were performed as previously described
142
(18).
143 144
mRNA isolation and Poly A library preparation. RNA from U87MG cells were purified from
145
both VEEV-TrD (biosafety level 3) infected and mock infected U87MG cells at 4, 8, and 16
146
h.p.i. utilizing the mirVana isolation kit (Life Technologies). Quality control of purified RNA
147
was then performed using an Agilent 2100 Bioanalyzer with a RNA integrity number (RIN)
148
cutoff of 8 utilized for all samples. An ERCC RNA Spike-In Control mix was then added to the
149
total RNA inputs (10ug RNA) before poly(A) selection using Life Technologies Dynabeads
150
mRNA DIRECT Kit. Whole transcriptome RNA library preparation from purified mRNA was
151
then performed using the Ion Total RNA-Seq Kit v2 (Life Technologies). Quality control of
152
cDNA libraries was then performed using the Agilent 2100 Bioanalyzer along with sterility
153
testing for removal of libraries for sequencing from a BSL-3 to BSL-2 laboratory.
154 155
RNA Sequencing. Library template preparation was performed on the One Touch 2 platform
156
(Life Technologies). Next generation RNA sequencing was performed on the Ion Torrent PGM
157
platform and carried out for each sample to assess differential gene expression of infected vs.
158
uninfected cells over time.
7
159
Data Filtering & RNA-Seq Analysis Pipeline. A total of ~119M sequencing reads were used as
160
input into our analysis pipeline, an average of 6.6M reads per sample. Unless otherwise noted,
161
downstream RNA-Seq analysis was carried out using CLC bio Genomics Workbench v7. Raw
162
RNA-Seq reads were trimmed to remove any residual sequencing adapter fragments that
163
remained on the 5’ or 3’ ends after sequencing. In addition, end trimming of reads was done
164
using the modified Mott algorithm with a Q20 quality score, and any reads below 15 bp were
165
discarded. Following read trimming, reads were mapped to Human Genome hg19 with the
166
following RNA-Seq parameters: a 10 hit limit for multiple mapped positions; similarity fraction
167
0.8; length fraction 0.8; mismatch cost 2; and indel cost 3. The expression level of individual
168
genes and transcripts was calculated using the RPKM (Reads Per Kilobase of exon model per
169
Million mapped reads) method of Mortazavi et al. (19). In addition, unmapped reads were also
170
mapped to the ERCC92 synthetic RNA sequence set (20), as well as to the VEEV reference
171
genome (GeneBank Accession #L01442). In all samples, the R2 correlation coefficient between
172
expected and mapped number of reads for ERCC92 spike in controls was above 0.90. A
173
summary of the overall sequencing results is shown in Table 1. The raw sequencing data for all
174
RNA-seq runs included in this work are publically available at the NCBI BioProject
175
PRJNA300864 (http://www.ncbi.nlm.nih.gov/bioproject/PRJNA300864).
176 177
Post-mapping filtering was carried out next for all RNA-Seq data to include only genes with at
178
least one uniquely mapped read (26,230 genes remained across all datasets) and only those with
179
a non-zero interquartile range across the entire experiment. Principal component analysis of the
180
resulting “filtered dataset” (13,906 genes in total) was carried out using raw counts of uniquely
181
mapped reads (Fig. 2A). The remaining RPKM expression values for each gene included in the
8
182
filtered dataset were subjected to quantile normalization with a 5% cutoff. A box-plot of log2
183
transformed RPKM values for each sample before normalization is shown in Fig. 2B. Pairwise
184
sample to sample variation R2 value within each biological replicate set was observed to range
185
from 0.89 to 0.99, indicating that our biological replicates were consistent and showed no strong
186
bias (data not shown).
187 188
Differential Gene Expression Analysis. Differentially expressed genes were identified using
189
two approaches. First, the empirical analysis of differential gene expression algorithm, part of the
190
edgeR Bioconductor package (21), was applied to the integrated dataset of all 18 experiments
191
using the default parameters and a false-discovery rate corrected p-value. At each time point,
192
infected vs. mock sample comparisons were made and genes whose expression differed by more
193
than 2-fold with a significance value of p ≤ 0.05 were provisionally considered differentially
194
expressed.
195 196
In addition to the method above, an orthogonal statistical test of differential expression was
197
applied to the data using a statistical test developed by Baggerly et al. (22) for counting the
198
number of expressed sequence tags associated with individual genes, a common feature of both
199
serial analysis of gene expression (SAGE) data and RNA-Seq data. Comparing infected vs. mock
200
samples, individual genes were provisionally considered differentially expressed with a greater
201
than 2-fold change and a significance of p ≤ 0.05. Differentially expressed genes found to be in
202
the intersection of the sets of genes identified these both of the methods outlined above were
203
considered high-quality candidates and used as the starting point for further investigation.
204
9
205
Clustering & Gene Set Enrichment Analysis. Filtered, normalized expression data was
206
subjected to k-means clustering using a Euclidian distance metric where genes were grouped by
207
means of normalized gene expression (RPKM) values for each experimental condition.
208
Clustering was fitted to 20 distinct clustering groups, and individual gene expression profile
209
clustered were further tested for enrichment of gene ontology (GO) terms associated with
210
individual genes. Gene annotations were obtained from Reactome a database of biological
211
pathway and gene functional annotations (23). Enrichment analysis was performed using two
212
approaches. First, a hypergeometric test on GO annotations was carried out using an
213
implementation of GOStats on each of the individual clusters obtained from k-means clustering
214
(24). In addition, gene set enrichment analysis (GSEA) was carried out on the entire filtered
215
dataset using 100,000 permutations, while removing duplicates and applying an ANOVA test. A
216
total of 1,419 categories passed a minimum feature size of 10, and were used for further
217
investigation.
218 219
Pathway Analysis. Cohorts of genes with shared patterns of expression over time were
220
identified by k-means clustering. Those found to be enriched for differentially expressed genes
221
were subsequently subjected to pathway analysis using the GeneMania system (25). Using an ad
222
hoc manual approach, relevant pathways and the connections between them were identified
223
based on existing data in the literature coupled with the temporal gene expression data obtained
224
from this study.
225 226
qRT-PCR analysis. Purified mRNA was converted to cDNA using the High Capacity RNA-to-
227
cDNA kit (Life Technologies) according to the manufacturer’s instructions. Analysis of the viral
10
228
copy numbers was performed by qRT-PCR as previously described (26). Host gene expression
229
was assayed with the following Taqman assays: ATF3 (Hs00231069_m1), ATF4
230
(Hs00909569_g1), CEBPB (Hs00270923_s1), CEBPD (Hs00270931_s1), DDIT3
231
(Hs00358796_g1), FOS (Hs04194186_s1), JUN (Hs01103582_s1), EGR1 (Hs00152928_m1),
232
IFI6 (Hs00242571_m1), IFIT1 (Hs01911452_s1), IFIT2 (Hs01922738_s1), IFIT3
233
(Hs01922738_s1), ISG15 (Hs01921425_s1), ISG20 (Hs00158122_m1), OASL
234
(Hs00984387_m1), BIRC5 (Mm00599749_m1) and XIAP (Mm01311594_mH). 18S assays
235
(Hs99999901_s1 or Mm04277571_s1) were used for normalization. Assays were performed
236
according to the manufacturer’s instructions using an ABI StepOne Plus instrument.
237
Treatment with PERK inhibitor and collection for western blots. U87MG cells were pre-
238
treated for two hours with 10 µM of the PERK inhibitor (PERKi) GSK2606414 (EMD Millipore,
239
516535) or DMSO in DMEM prior to infection with VEEV TrD (MOI 5). After one hour, the
240
viral inoculum was removed and cells were washed with sterile PBS (1X). Media containing the
241
inhibitor or DMSO was replaced. Sixteen hours post-infection media was removed, cells washed
242
with PBS, and then collected for western blot analysis.
243
Knockdown of EGR1 with siRNA. U87MG seeded at 6.7 x 104 cells per well in a 12-well plate
244
were transfected with 50 nM of siGENOME SMARTpool EGR1 (Dharmacon, M-006526-01) or
245
AllStar Negative Control siRNA (Qiagen, 1027280), using 1.33 µl of Dharmacon DharmaFECT
246
1 Transfection Reagent (T-2001-02). Twenty-four hours post-transfection, cells were infected
247
with VEEV TrD (MOI 5) for one hour. Fresh media was replaced after infection. Twenty-four
248
hours after infection supernatants or cells were collected for analysis.
11
249
Protein lysate preparation and western blot analysis. Protein lysates and western blot analysis
250
were performed as previously described (27). The primary antibodies were used: EGR1 (44D5)
251
(Cat#4154, Cell Signaling), Polyclonal Anti-Venezuelan Equine Encephalitis Virus, TC83
252
(Subtype IA/B) Capsid Protein (BEI Resources), CHOP (L63F7) (Cat#2895, Cell Signaling), p-
253
eIF2a (Ser51) (D9G8) (Cat#3398, Cell Signaling), ATF4 (D4B8) (Cat#11815, Cell Signaling),
254
Activated Caspase 3 (Asp175) (Cat#9661, Cell Signaling), and HRP conjugated β-actin (Cat#
255
ab49900-100, Abcam).
256
Immunofluorescence analysis. U87MG cells were grown on coverslips in a 6-well plate,
257
infected with VEEV TrD as described above, washed with PBS (without Ca and Mg) and then
258
fixed with 4% formaldehyde. Cells were permeabilized with 0.5% Triton X-100 in PBS for 20
259
minutes then washed twice with PBS. The cells were blocked for 10 minutes at room
260
temperature in 3% BSA in PBS. Primary antibodies, VEEV-capsid (BEI Resources, NR-9403)
261
diluted 1:600, and EGR1 (44D5) (Cat#4154, Cell Signaling) diluted 1:400 were incubated in
262
fresh blocking buffer at 37°C for 1 hour and washed 3 times for 3 minutes in 300 mM NaCl with
263
0.1% Triton X-100. Alexa Fluor 568 donkey anti-goat secondary (Invitrogen, A11057) and
264
Alexa Fluor 488 donkey anti-mouse secondary (Invitrogen, A21202) dilution 1:400 were used as
265
secondary antibodies and treated in the same manner as the primary antibody. DAPI, dilution
266
1:1000, was used to visualize nuclei. Cover slips were mounted to glass slides using 10 µL of
267
Fluoromount G (Southern Biotech, 0100-01). A Nikon Eclipse TE 2000-U was used for
268
fluorescence microscopy. Images were viewed using an oil-immersion 60X objective lens. Each
269
sample was subjected to at least five images, with a representative image shown. All images
270
were subjected to four line averaging. The images were processed through Nikon NIS-Elements
271
AR Analysis 3.2 software. 12
272 273
Cell TiterGlo and Caspase 3/7 Glo assays. Wild-type and EGR1 -/- MEFs were infected with
274
TrD at various MOIs for an hour then washed with PBS and media replaced. Cell viability was
275
measured at 24 hours post-infection using Promega’s CellTiter Luminescent Cell Viability Assay
276
(G7571) according to the manufacturer’s protocol. Luminescence was read using Beckman
277
Coulter’s DTX 880 Multimode Detector with an integration time of 100 ms per well. Similarly,
278
caspase activation at 24 hours post-infection in infected wild-type and EGR1 -/- MEFs was
279
measured using Promega’s Caspase 3/7 Glo assay (G8090) according to the manufacturer’s
280
protocol. Luminescence was read using the DTX 880 with an integration time of 100 ms per
281
well.
282 283
Results
284
VEEV replication kinetics in U87MG astrocytes. VEEV replicates in vivo in monocytes,
285
macrophages, neurons, and astrocytes (14). Common cell lines used to study VEEV infection
286
include Vero and BHK cells; in this study U87MG astrocytes were chosen as an in vitro model
287
due to their physiological relevance and greater clinical significance. Initial experiments were
288
performed to characterize viral replication in U87MG cells. VEEV replication kinetics in
289
U87MG cells were measured using plaque assays, and by monitoring viral protein and RNA
290
expression levels, and cytopathic effect (CPE) on the infected cells (Fig. 1). Viral release was
291
observed as early as 4 hours post-infection (h.p.i.), with ~4 logs of virus observed, followed by a
292
consistent increase in replication at 8 and 16 h.p.i. (Fig. 1A). Viral replication peaked at sixteen
293
h.p.i. with no additional increase in viral titers observed at 24 h.p.i. Viral capsid expression
294
followed a similar pattern, with protein being detected at 8 h.p.i. and plateauing at 16 h.p.i (Fig.
13
295
1B). Among infected U87MG cells, significant CPE was observed by microscopy at 24 h.p.i.,
296
with little to no CPE detected at 16 h.p.i. (data not shown). Consistent with these observations,
297
increased caspase 3/7 activity was only observed at 24 h.p.i. (Fig. 1C). Based on these data, 4, 8,
298
and 16 h.p.i. were selected for RNA-Seq analysis, reflecting early, mid, and late stages of the
299
viral life cycle, in order to provide a dynamic view of the host-pathogen transcriptome profile.
300
RNA sequencing analysis of VEEV infected astrocytes. mRNA from triplicate sets of mock
301
and VEEV infected U87MG cultures was isolated and purified at 4, 8, and 16 h.p.i. and used to
302
prepare cDNA libraries for downstream RNA-Seq (see Materials & Methods). A high level
303
summary of the RNA-Seq results is shown in Table 1. VEEV RNA samples were assayed by
304
quantitative RT-PCR at each time point as a control to demonstrate the increasing viral RNA
305
load over time (Fig. 1D), consistent with increasing number of RNA-Seq reads mapped to the
306
VEEV genome at later time points (Table 1).
307
For RNA-Seq analysis, individual genes were expressed as reads per kilobase per million
308
mapped reads (RPKM) (19). Log2 normalized RPKM expression values for each experimental
309
sample are shown in Fig. 2A and can be found in Dataset S1. Minimal sample-to-sample
310
variation expression values within biological replicates was consistently low (R2 > 0.89 for all
311
replicates; data not shown). In addition, intersample variation was also found to be minimal
312
when tested pairwise across the entire experiment by using RPKM values for ERCC97 synthetic
313
spike-in control RNAs (R2 > 0.90 for all comparisons; data not shown).
314
As anticipated, two-component principal component analysis of RNA-Seq data for mock vs.
315
VEEV infected cells showed a clear separation of 16 h.p.i. samples compared to earlier time
316
points (Fig. 2B). However, the clustering of VEEV infected samples with mock samples at
14
317
earlier time points suggested that the response to viral infection was limited to a narrow subset of
318
early response genes, thus placing a higher burden of proof on identifying differentially
319
expressed genes (DEGs) during the first few hours of infection. Along these lines two orthogonal
320
methods were used to identify DEGs suitable for further characterization: the edgeR method
321
(21), and that developed by Baggerly et al. (22). Genes identified by one test were provisionally
322
considered as DEGs, and those identified by both were candidate DEGs to be confirmed by qRT-
323
PCR. In addition to comparing individual gene expression values for mock vs. VEEV infected
324
cells at each time point, gene expression values were also compared serially within each time
325
series of VEEV infected cells for genes that did not show any statistically significant changes in
326
mock infected cells. A schematic of the comparative analysis is shown in Fig. 2C. The number of
327
statistically significant DEGs identified by each of these comparisons is shown in Fig. 2D.
328
Furthermore, k-means clustering was employed (against normalized RPKM values) to identify
329
gross changes in gene expression over time for cohorts of genes potentially sharing the same
330
pathway or regulatory triggers (Fig. 3 and Dataset S2). Gene set enrichment analysis (GSEA, see
331
Material & Methods and Dataset S3) was carried out on each k-means cluster. In particular,
332
Cluster_20 (Table 2) was significantly enriched for genes involved in translational control, type I
333
interferon-mediated signaling pathway, and the unfolded protein response (UPR) pathway
334
(GSEA p-value < 0.01). Although there is a well-established connection between translational
335
control and UPR, a novel connection between UPR and the type-I mediated interferon response
336
in response to viral replication was suggested by pathway analysis (see Materials & Methods)
337
implicating EGR1 as a potential bridge between these two pathways (Fig. 4). EGR1 belongs to
338
Cluster 20, and is strongly induced during VEEV infection; and several other genes associated
339
with the interferon response belong to the same cluster: IRF1, IFIT1, IFIT2, ISG15, and ILF3.
15
340
EGR1 has been associated with increases in ATF3 (28), a key component of the UPR, and also
341
belongs to Cluster 20. This connection represented a potential between the UPR pathway and the
342
interferon response pathway, with EGR1 as one of the potential key transcription factors driving
343
this connection. Consequently, 15 genes were selected from this analysis for further
344
characterization by qRT-PCR (see below): ATF3, ATF4, CEBPB, CEBPD, DDIT3/CHOP,
345
EGR1, FOS, IFI6, IFIT1, IFIT2, IFIT3, ISG15, ISG20, JUN, and OASL. The expression values
346
of these genes, as measured by RNA-Seq, are shown in Figs. 5A and 5B. Confirmatory qRT-
347
PCR analysis indicated concordant gene expression (Figs. 5C and 5D). The induction of
348
interferon responsive genes is in agreement with previously published studies (11, 29, 30),
349
serving as an internal positive control. Moreover, the link between EGR1 and the interferon
350
pathway has been demonstrated; EGR1 is induced by IFN-γ in mouse fibroblasts, and by IFN-α,
351
β and γ in human fibroblasts (31, 32). EGR1 and the UPR pathway were selected for further
352
analysis as their role in VEEV infection has not been elucidated.
353
VEEV infection induces the UPR late in infection. The RNA-Seq and pathway analysis data
354
indicated that UPR and stress response genes were induced after VEEV infection. During an
355
infection, host cells respond to cellular stresses resulting from increased viral protein translation
356
and secretion by triggering the onset of the UPR pathway. The UPR pathway is an adaptive
357
cellular response activated by endoplasmic reticulum (ER) stress due to protein misfolding. In
358
order to regulate cellular homeostasis during protein folding and secretion, the UPR pathway has
359
developed three classes of sensors to ensure proper cellular regulation: inositol-requiring enzyme
360
1 (IRE1), protein kinase RNA-like ER kinase (PERK), and activating transcription factor 6
361
(ATF6) (33, 34). During VEEV infection, the PERK arm of the UPR appeared to be altered, as
362
two critical regulators of this pathway were differentially expressed: activating transcription 16
363
factor 4 (ATF4) and CHOP (DDIT3) (35). To determine if DEG’s altered subsequent protein
364
expression, western blot analysis was performed for CHOP, ATF4 and phosphorylated-eIF2α (p-
365
eIF2α). Tunicamycin, a glycosylation inhibitor and inducer of UPR (36), was included as a
366
positive control. A time course analysis of U87MG cells treated with 1 μM of Tunicamycin
367
indicated that 8 hours of treatment provided the most robust induction of UPR proteins (data not
368
shown). VEEV infected, but not mock or UV-inactivated VEEV (UV-VEEV) infected cells
369
displayed a dramatic increase in p-eIF2α and a modest but consistent increase in CHOP and
370
ATF4 at 16 h.p.i. (Fig. 6A). No change in protein expression was observed at 4 h.p.i. (data not
371
shown). Confocal microscopy confirmed CHOP and ATF4 upregulation, demonstrating a more
372
robust and nuclear staining pattern in VEEV infected cells as compared to mock infected cells
373
(Fig. 6C-E). While ATF4 protein expression levels increased, ATF4 mRNA abundances
374
decreased following VEEV infection (Fig. 5B and 5D). These results are consistent with the
375
observation that ATF4 expression is regulated at the translational level upon UPR induction (37).
376
As eIF2α can be phosphorylated by multiple kinases [PERK, PKR (protein kinase double-
377
stranded RNA-dependent), GCN2 (general control non-derepressible-2), and HRI (heme-
378
regulated inhibitor)] (38), the PERK inhibitor GSK2606414 (PERKi) was used to determine if
379
the observed phosphorylation was PERK dependent. Treatment of VEEV infected cells with
380
PERKi resulted in a marked decrease in eIF2α phosphorylation (Fig. 6B). These results indicate
381
that PERK contributes to eIF2α phosphorylation, but that there is likely an additional kinase
382
contributing to the phosphorylation event. Collectively, these findings indicate that the PERK
383
arm of the UPR pathway is induced at later time points following VEEV infection.
384
EGR1 is upregulated in infected cells and localizes to the nucleus. Early Growth Response 1
385
(EGR1) is a transcription factor that can be induced by numerous signals including oxidative 17
386
stress, hypoxemia, and growth factors (39, 40). It can also be activated upon infection by both
387
DNA and RNA viruses, including Epstein - Barr virus, Mouse Hepatitis Virus, Murine
388
Coronavirus and Japanese Encephalitis Virus (41-43). Treatment of MEFs with the UPR
389
activator, thapsigargin, has been shown to induce EGR1 expression in a PERK dependent
390
manner (44). Given the link between EGR1 and UPR, and the robust induction of EGR1 mRNA
391
expression following VEEV infection (Figs. 4 and 5), EGR1 was chosen for further study. EGR1
392
protein expression was analyzed after VEEV infection by western blot analysis. As previous
393
studies have indicated that EGR1 can be activated by Mouse Hepatitis Virus independent of
394
virus replication (likely due to cellular membrane disruption following entry) (41), a UV-
395
inactivated virus control was included (UV-VEEV). EGR1 protein levels were increased
396
following VEEV infection as compared to mock infected cells and UV-VEEV infected cells
397
(Fig. 7A, compare lanes 3, 6, and 9). The most dramatic upregulation of EGR1 occurred at 16
398
h.p.i.; this correlates with the highest levels of VEEV capsid production (see Fig. 1B). Following
399
induction, EGR1 has been shown to translocate to the nucleus to induce gene expression through
400
binding to the Egr binding sequence (EBS) [(GCG(G/T)GGCG] (40, 45). Confocal microcopy
401
revealed high levels of EGR1 in the nuclei of infected cells, whereas only low levels of both
402
nuclear and cytoplasmic EGR1 were detected in mock cells (Fig. 7B). PERKi treatment of
403
VEEV infected cells resulted in a complete loss of EGR1 induction (Fig. 7C), indicating that
404
EGR1 was induced in a PERK dependent fashion. These results demonstrate that EGR1 protein
405
levels and nuclear localization are increased following VEEV infection and that the induction of
406
EGR1 is dependent on PERK.
407
Loss of EGR1 inhibits VEEV induced apoptosis but does not alter VEEV replication
408
kinetics. As EGR1 influences cell survival and apoptosis (46), the impact of EGR1 on VEEV 18
409
induced cell death was assessed. Caspase 3 cleavage was observed in WT MEFs at 24 h.p.i.
410
when infected at an MOI of 0.5, and started as early as 16 h.p.i. when infected at an MOI of 5
411
(Fig. 8A). In contrast, EGR1 -/- cells showed little to no detectable caspase cleavage following
412
infection with VEEV. Two sets of experiments were performed to quantitatively confirm these
413
results: Cell Titer Glo assays to measure total cell viability (ATP production) and Caspase 3/7
414
Glo assays to measure caspase 3/7 activity. Both WT and EGR1 -/- MEFs displayed dose
415
dependent decreases in cell viability following VEEV infection, with EGR1-/- cells having
416
significantly more viable cells at each MOI examined (Fig. 8B). Concordantly, a dose dependent
417
increase in caspase 3/7 activity was observed following VEEV infection, with EGR1 -/- cells
418
demonstrating reduced caspase 3 activity at MOIs of 0.5 and 5 (Fig. 8C). These results were
419
replicated in EGR1 siRNA transfected U87MG cells (Fig. 8D).
420
EGR1 has been shown to negatively regulate the transcription of BIRC5 (Survivin), an inhibitor
421
of apoptosis (IAP) family member (47). RNA-Seq data indicated that BIRC5 gene expression
422
was decreased following VEEV infection, -1.16, -1.18, and -1.50 log2 transformed fold change
423
values of normalized gene expression at 4, 8, and 16 h.p.i. respectively (see Table S1 and NCBI
424
BioProject PRJNA300864). WT and EGR1 -/- MEFs were used to determine if EGR1 was
425
influencing BIRC5 gene expression following VEEV infection. BIRC5 expression was
426
significantly decreased at 16 h.p.i. in VEEV infected WT MEF, but this reduction was not
427
observed in VEEV infected EGR1 -/- MEFs (Fig. 8E). Gene expression of X-linked inhibitor of
428
apoptosis (XIAP), another IAP family member, was not significantly differentially altered after
429
infection (data not shown). Collectively these results demonstrate that EGR1 contributes to
430
VEEV induced apoptosis.
19
431
VEEV replication kinetics were determined for both EGR1 -/- and WT MEFs to determine the
432
relevance of EGR1 in viral replication. Cells were infected at two different MOIs (0.5 and 5),
433
with viral supernatants collected at 4, 8, 16, and 24 h.p.i., and analyzed by plaque assay. The
434
replication kinetics were similar between EGR1 -/- and WT MEFs at both MOI’s with titers
435
peaking at 16 h.p.i. (Fig. 9A). Lack of EGR1 expression was confirmed by western blotting (Fig.
436
9B). These results were replicated in EGR1 siRNA transfected U87MG cells. EGR1 siRNA
437
transfection resulted in a >90% decrease in EGR1 protein expression (Fig. 9D) without any
438
significant effect on viral replication (Fig. 9C). These results suggest that the observed decreased
439
in apoptosis in EGR1 -/- MEFs was not due to altered VEEV replication kinetics.
440 441
Discussion
442
Despite being recognized as an emerging threat, relatively little is known about the virulence
443
mechanisms of alphaviruses, largely due to a knowledge gap in the host-pathogen interactome.
444
VEEV infection often results in fatal encephalitis and is known to inhibit both cellular
445
transcription and translation in order to downregulate the innate immune response (1, 48). In
446
contrast, in the CNS VEEV has been shown to upregulate numerous genes in both the
447
inflammatory response and apoptotic pathway (1, 48). Specifically, numerous pro-inflammatory
448
cytokines including interleukin-1β (IL-1β), IL-6, IL-12, GSK-3β, iNOS and tumor necrosis
449
factor -α (TNF-α) have all been shown to play a role in VEEV pathogenesis (49-53). The use of
450
high throughput next generation sequencing technologies such as RNA-Seq allows for an in
451
depth and unbiased look into the viral-host transcriptome, thus enabling changes in the
452
expression of specific mRNAs to be connected with phenotypic outcomes. To this end,
453
identifying critical differentially expressed transcripts among clinically relevant infected cells 20
454
will help lead to a greater understanding of viral pathogenesis and may prove beneficial for the
455
identification of therapeutic targets.
456 457
In this study, network analysis/ RNA-Seq data and protein expression studies revealed that
458
VEEV infection resulted in activation of the PERK arm of the UPR pathway, including ATF4,
459
CHOP, and eIF2α phosphorylation. Several alphaviruses have previously been reported to
460
hijack key components of the UPR pathway in order to promote viral replication, as the reliance
461
of enveloped viruses on the ER for the synthesis of viral envelope associated glycoproteins and
462
their transport to the plasma membrane often stress the ER due to rapid viral protein production
463
(54, 55). Modulation of the UPR is not unique to alphaviruses; rather it is a shared trait of many
464
positive sense RNA viruses. Dengue virus has been shown to suppress PERK by inhibiting
465
continued eIF2α phosphorylation in order to inhibit immediate apoptosis, increasing viral protein
466
translation, and extending the length of productive viral replication (34). Studies with Hepatitis E
467
virus (HEV), have demonstrated that expression of HEV capsid protein open reading frame 2
468
(ORF2) activates the expression of CHOP and ATF4 (56). In HEV, ORF2 was shown to
469
stimulate CHOP through both ER stressors and amino acid response elements (AARE) through
470
interaction with ATF4 (56).
471 472
The results shown here indicate that during VEEV infection, initiation of the UPR pathway and
473
subsequent activation of EGR1 play a role in the outcome of viral induced apoptosis. During the
474
initial detection of ER stress, PERK is able to identify misfolded proteins in the lumen of the ER
475
and phosphorylates eIF2α in order to initiate pro-survival pathways in the UPR through the
476
general inhibition of protein synthesis (33, 34). VEEV appears to be inducing the UPR and
21
477
promoting increased eIF2α phosphorylation, which results in translational inhibition of most
478
mRNAs, while selectively increasing translation of ATF4. ATF4 is responsible for the
479
expression of genes that encode proteins involved in apoptosis, redox processes, amino acid
480
metabolism, and ER chaperone recruitment and is a well-known mediator of the PERK pathway
481
and CHOP (33, 34). CHOP activation facilitates the increased expression of cellular chaperones
482
in order to counteract the buildup of misfolded proteins (57). Failure to suppress protein
483
misfolding in persistently stressed cells, such as during a viral infection, can then result in
484
activation of the pro-apoptotic transcription factor CHOP, leading to suppression of the anti-
485
apoptotic protein B cell lymphoma-2 (Bcl-2). CHOP can also function as a pro-survival
486
transcription factor by dephosphorylating eIF2α through activation of the DNA damage-
487
inducible protein (GADD34) in a self-regulating feedback look (33, 34). However, the data
488
presented here supports a model whereby VEEV infection leads CHOP to function in its pro-
489
apoptotic role, as no change in GADD34 gene expression was detected by RNA-Seq analysis.
490 491
While the UPR was induced following VEEV infection, robust activation was not observed until
492
later time points after infection. This is somewhat surprising as VEEV infection is expected to
493
induce significant ER stress due to the massive production of viral proteins during the course of
494
an acute robust infection. The structural proteins of VEEV are translated from the viral
495
subgenomic RNA into polyproteins on the rough ER. The E1 and pE2 precursor glycoproteins
496
are then assembled as heterodimers in the ER, undergoing conformational changes requiring
497
numerous chaperones (1, 58). It is possible that VEEV has developed mechanisms to subvert the
498
induction of the UPR. In order to counteract the UPR, the non-structural proteins (nsPs) of
499
Chikungunya virus (CHIKV) have been shown to inhibit expression of ATF4 and other known
22
500
UPR target genes including GRP78/BiP, GRP94 and CHOP (59). Through nsPs activity, CHIKV
501
has developed a means of suppressing UPR activity resulting from viral glycoprotein induced ER
502
stress, thus preventing immediate autophagy and apoptotic activation. VEEV capsid is
503
responsible for interfering with nucleocytoplasmic trafficking, inhibiting rRNA and mRNA
504
transcription, and has been implicated in the regulation of type-I IFN signaling and the antiviral
505
response through regulation of both viral RNA and protein production (1, 48, 60). Therefore we
506
hypothesize that the ability of VEEV capsid to inhibit cellular transcription and block
507
nucleocytoplasmic trafficking results in delayed induction of the UPR.
508 509
A detailed network analysis, based on existing data in the literature, coupled with the temporal
510
gene expression profiles obtained from this study, pointed towards EGR1 as an important node
511
in the novel link between VEEV activation of the type-1 interferon response and UPR. EGR1 is
512
known to form a DNA binding complex with C/EBPB, a critical dimerization partner of CHOP
513
(61). Previous studies have demonstrated that nuclear localization of CHOP may act as an
514
inducer of EGR1 and as a transcriptional cofactor for regulation of C/EBPB-EGR1 target genes
515
(61). The western blot and microscopy analysis presented in this study support this model as
516
VEEV infection was found to increase both overall levels and the nuclear distribution of CHOP
517
along with ERGR1. Previous studies demonstrated EGR1 mRNA induction by IFN-γ in mouse
518
fibroblasts, and by TNF-α, TNF-β, IL-1, IFN-α,β and γ in human fibroblasts (32, 62). EGR1,
519
also known as Zif268 and NGF1-A, is a zinc finger protein and mammalian transcription factor.
520
It has been implicated in cellular proliferation and differentiation, but may also have pro-
521
apoptotic functions, depending on the cell type and stimulus (63). Of particular interest, EGR1
522
directly controls proliferation when activated by the MAP kinase/ERK pathway in mitogen
23
523
stimulated astrocytes (64). Virus-induced changes in EGR1 expression have been observed in
524
several in vitro systems. In HIV-1 infected astrocytes, EGR1 upregulation was found to be
525
induced by Tat through transactivation of the EGR1 promoter, leading to cellular dysfunction
526
and Tat-induced neurotoxicity (65). Increased amounts of EGR1 mRNA have also been
527
demonstrated to act in a region specific manner, corresponding temporally with viral RNA
528
production in the brain tissues of rats infected with both rabies and borna disease virus (66).
529 530
In summary, the current study demonstrates a potential link between UPR activation and EGR1.
531
EGR1 -/- MEFs demonstrated lower susceptibility to VEEV induced cell death as compared to
532
wildtype MEFs, indicating that EGR1 modulates pro-apoptotic pathways following infection.
533
Studies are underway to determine if alteration of UPR through small molecule inhibitors or
534
siRNA interference influences VEEV replication and/or cell death. To date the mechanisms
535
underlying VEEV pathogenesis and subsequent neuronal degeneration have only been partially
536
elucidated. Therefore, determining the role of EGR1 and UPR may play a significant role in the
537
development of a novel therapeutic target resulting in decreased neuronal death and the
538
subsequent neuronal sequela resultant from infection.
539 540
Funding Information
541
This work was funded through a Defense Threat Reduction Agency grant, HDTRA1-13-1-0005,
542
to KK and JJ. The content of the information does not necessarily reflect the position or the
543
policy of the federal government, and no official endorsement should be inferred.
544
24
545
Acknowledgments
546
The following reagents were obtained through the NIH Biodefense and Emerging Infections
547
Research Resources Repository, NIAID, NIH: 1) Polyclonal Anti-Venezuelan Equine
548
Encephalitis Virus, TC83 (Subtype IA/B) Capsid Protein (antiserum, Goat), NR-9403 and 2)
549
Venezuelan equine encephalitis virus Trinidad Donkey (subtype IA/B), NR-332. We thank Dr.
550
Stephen Hann, Vanderbilt University School of Medicine, for the EGR1 -/- cells.
551 552
Figure Legends
553
Fig. 1. VEEV replication kinetics in U87MG cells. U87MG cells were infected with VEEV
554
TrD (MOI 5). A) Viral supernatants were collected at 4, 8, 16, and 24 h.p.i. and viral titers
555
determined by plaque assays. B) Protein lysates were collected at 4, 8, 16, and 24 h.p.i. and
556
western blot analysis performed with anti-capsid and anti-actin antibodies. C) U87MG cells were
557
mock treated or infected as in panel A. Caspase 3/7 activity was analyzed using the Caspase 3/7
558
Glo assay (Promega). * p-value < 0.01 D) RNA was extracted using the miRVana kit and VEEV
559
RNA quantitated by qRT-PCR. Data shown are fold change values (compared to the VEEV 4
560
h.p.i. sample) of normalized gene expression (ΔΔCt method).
561
Fig. 2. RNA-Seq Data Analysis. A) Box-plot of log2 transformed RPKM expression values
562
from each individual biological replicate. B) Principal component analysis of mock vs. VEEV
563
infected U87MG cells at 4, 8, 16 h.p.i. C) Process diagram of comparative analysis approach to
564
identify differentially expressed genes (DEG) at each sampling time point. D) Total number of
565
DEGs identified for each comparison shown in C using two different statistical methods (see
25
566
Materials & Methods). The union of DEGs identified by both methods is shown in orange. V and
567
M are used for brevity for VEEV infected and mock infected cells respectively.
568
Fig. 3. K-means Clustering of Expressed Genes. K-means clustering (20 groups) of genes
569
based Euclidean distance metrics of expression over time for both mock and VEEV infected
570
U87MG cells. Data was normalized based on group means and plotted as relative transformed
571
expression level for each gene. Each cluster has six data points per gene (VEEV 4, 8, 16 h.p.i.
572
followed by mock 4, 8, 16 h.p.i). See Dataset S2 for a complete list of expressed genes within
573
each cluster.
574
Fig. 4. Network of Type I Interferon Response & UPR Related Genes. Differentially
575
expressed genes (large circles), no significant change (small circles), type-I interferon response
576
factors (red circles), regulation of DNA transcription (yellow circles), unfolded protein response
577
genes (blue circles), physical protein-protein interactions (red lines), common pathway (blue
578
lines). This network was seeded with kmeans Clusters 18 and 20 – many ribosomal protein genes
579
were removed.
580
Fig. 5. Next-generation RNA sequencing (RNA-Seq) results and confirmatory qRT-PCR
581
analysis. A) and B) RNA-Seq data for Interferon Responsive Genes (panel A) and UPR &
582
Stress Response Genes (panel B). Data shown in panels A and B are log2 transformed fold
583
change values of normalized gene expression (RPKM). Error bars are ratiometric standard
584
deviation of a two-component log function (VEEV/mock)*. C) and D) RNA was extracted
585
using the miRVana kit and gene expression determined by qRT-PCR using Taqman assays for
586
Interferon Responsive Genes (panel C) or for UPR and Stress Response Genes (panel D). Data
26
587
shown in panels C and D are log2 transformed fold change values of normalized gene expression
588
(ΔΔCt method).
589
Fig. 6. UPR is activated in VEEV infection at later time points. A) U87MG cells were mock,
590
VEEV, or UV-VEEV infected (MOI 5) and protein lysates collected at 4, 8, and 16 h.p.i.
591
Western blot analysis was performed using antibodies against CHOP, ATF4, p-eIF2α and actin.
592
T8=U87MG cells treated with Tunicamycin for 8 hours as a control for UPR induction. Results
593
are representative of three independent experiments. B) U87MG cells were pre-treated with
594
DMSO or PERK inhibitor (PERKi) for two hours prior to infection with VEEV TrD (MOI 5)
595
and replacement of drug media. Western blot analysis was performed using antibodies against p-
596
eIF2α and actin. C) U87MG cells were treated with Tunicamycin for 8 hours (T=8) as a control
597
for UPR induction. D) and E) U87MG cells were mock or VEEV infected (MOI 5). Sixteen
598
hours after infection, cells were fixed and probed with DAPI, anti-VEEV capsid, anti-ATF or
599
anti-CHOP primary antibodies, and Alexa Fluor 488 and 568 labeled secondary antibodies.
600
Slides were imaged on a Nikon Eclipse TE2000-U after immunofluorescent staining. Results are
601
representative of two independent experiments.
602
Fig. 7. EGR1 is upregulated in infected cells and localizes to the nucleus. A) U87MG cells
603
were mock, VEEV, or UV-VEEV infected (MOI 5) and protein lysates collected at 4, 8, and 16
604
h.p.i. Western blot analysis was performed using antibodies against EGR1, capsid, and actin.
605
Results are representative of two independent experiments. B) U87MG cells were mock or
606
VEEV infected (MOI 5). Sixteen hours after infection, cells were fixed and probed with DAPI,
607
anti-VEEV capsid and anti-EGR1 primary antibodies, and Alexa Fluor 488 and 568 labeled
608
secondary antibodies. Slides were imaged on a Nikon Eclipse TE2000-U after
609
immunofluorescent staining. Results are representative of two independent experiments. C) 27
610
U87MG cells were pre-treated with DMSO or PERK inhibitor (PERKi), infected with VEEV
611
TrD (MOI 5), and post-treated with drug media. Protein lysates were collected at 16 h.p.i.
612
Western blot analysis was performed using antibodies against EGR1, capsid, and actin.
613
Fig. 8. Loss of EGR1 reduces VEEV induced apoptosis. A) EGR1 -/- and WT MEFs were
614
infected with VEEV at an MOI of 0.5 or 5. Protein lysates were prepared at 4, 8, 16, and 24 h.p.i.
615
and separated by SDS-PAGE and western blot analysis was performed using antibodies against
616
cleaved caspase 3 and actin. Mock infected cells are included as a control. Results are
617
representative of two independent experiments. B) EGR1 -/- and WT MEFs were infected with
618
VEEV at an MOI of 0.05, 0.5 or 5. Twenty-four h.p.i. cells were analyzed for viability using the
619
Cell Titer Glo assay (Promega). Mock infected cells were included as a control and set to 100%
620
viability. *p-value