JVI Accepts, published online ahead of print on 20 August 2014 J. Virol. doi:10.1128/JVI.01643-14 Copyright © 2014, American Society for Microbiology. All Rights Reserved.
1
Establishment and characterization of a lethal mouse model for the Angola strain of
2
Marburg virus
3 4
Author names: Xiangguo Qiu1*, Gary Wong1, 2, Jonathan Audet1, 2, Todd Cutts3, Yulian Niu4,
5
Stephanie Booth4, and Gary P. Kobinger1, 2, 5, 6*
6 7
Author affiliations:
8
1
9
Canada, Winnipeg, MB, Canada
Special Pathogens Program, National Microbiology Laboratory, Public Health Agency of
10
2
Department of Medical Microbiology, University of Manitoba, Winnipeg, MB, Canada
11
3
Applied Biosafety and Research Program, JC Wilt Infectious Diseases Research Centre, Public
12
Health Agency of Canada, Winnipeg, MB, Canada
13
4
14
Winnipeg, MB, Canada
15
5
Department of Immunology, University of Manitoba, Winnipeg, MB, Canada
16
6
Department of Pathology and Laboratory Medicine, University of Pennsylvania School of
17
Medicine, Philadelphia, PA, USA
Molecular PathoBiology, National Microbiology Laboratory, Public Health Agency of Canada,
18 19
Keywords: Marburg strain Angola; Mouse; Adaptation; Pathogenesis
20 21 22 23
Running title: Mouse model for Marburg strain Angola
24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43
*Corresponding authors: Xiangguo Qiu, M.D. Special Pathogens Program National Microbiology Laboratory Public Health Agency of Canada 1015 Arlington Street Winnipeg, MB, R3E 3R2 Canada Phone: (204) 789-5097 Fax: (204) 789-2140 E-mail:
[email protected] Gary Kobinger, Ph.D. Special Pathogens Program National Microbiology Laboratory Public Health Agency of Canada 1015 Arlington Street Winnipeg, MB, R3E 3R2 Canada Phone: (204) 784-5923 Fax: (204) 789-2140 E-mail:
[email protected]
44
Abstract
45
Infections with Marburg (MARV) and Ebola virus (EBOV) cause severe hemorrhagic fever in
46
humans and non-human primates (NHPs) with fatality rates up to 90%.
47
experimental vaccine and treatment platforms have previously been shown to be protective
48
against EBOV infection. However, the rate of development for prophylactics and therapeutics
49
against MARV has been slower in comparison, possibly because a small animal model is not
50
widely available.
51
pathogenesis of MARV Angola (MARV/Ang), the most virulent strain of MARV. Infection
52
with the wild-type virus does not cause disease in mice, but the adapted virus (MARV/Ang-MA)
53
recovered from liver homogenates after 24 serial passages in severe combined immunodeficient
54
(SCID) mice caused severe disease when administered intranasally (IN) or intraperitoneally (IP).
55
The LD50 was determined to be 0.015 TCID50 MARV/Ang-MA in SCID mice, and IP infection
56
at a dose of 1000 x LD50 resulted in death between 6-8 days post-infection (dpi) in SCID mice.
57
Similar results were obtained with immunocompetent BALB/c and C57BL/6 mice challenged IP
58
with 2000 x LD50 of MARV/Ang-MA. Virological and pathological analyses in MARV/Ang-
59
MA infected BALB/c mice revealed that the associated pathology was reminiscent of
60
observations made in NHPs with MARV/Ang. MARV/Ang-MA infected mice showed most of
61
the clinical hallmarks observed with Marburg hemorrhagic fever, including lymphopenia,
62
thrombocytopenia, marked liver damage and uncontrolled viremia. Virus titers reached 108
63
TCID50/ml in the blood, and between 106-10 TCID50/g tissue in the intestines, kidney, lungs,
64
brain, spleen, and liver. This model provides an important tool to screen candidate vaccines and
65
therapeutics against MARV infections.
A number of
Here we report the development of a mouse model for studying the
66
Importance
67
The Angola strain of Marburg virus (MARV/Ang) was responsible for the largest outbreak ever
68
documented for Marburg viruses. At a 90% fatality rate, it is similar to Ebola virus, which
69
makes it one of the most lethal viruses known to humans. There are currently no approved
70
interventions for Marburg virus, in part because a small animal model that is vulnerable to
71
MARV/Ang infection is not available to screen and test potential vaccines and therapeutics in a
72
quick and economical manner. To address this need, we have adapted MARV/Ang so that it
73
causes illness in mice resulting in death. The signs of disease in these mice are reminiscent of
74
wild-type MARV/Ang infections in humans and nonhuman primates. We believe this will be of
75
help in accelerating the development of life-saving measures against Marburg virus infections.
76
Introduction
77
Infection with filoviruses cause severe hemorrhagic fevers with high mortality rates. The family
78
Filoviridae is divided into three genera: Ebolavirus, Marburgvirus, and Cuevavirus. Within the
79
Ebolavirus genus, Zaire ebolavirus (Ebola virus; EBOV) is the most pathogenic in humans, with
80
a case fatality rate of up to 90%, followed by the Sudan (54%) and Bundibugyo (34%) viruses.
81
Together, ebolaviruses have been responsible for more than 1500 deaths since their discovery in
82
1976 (4). In contrast, live Cuevavirus have not yet been isolated from insectivorous bats (17)
83
and is not known to cause illness in humans, whereas outbreaks of Marburg virus (MARV) have
84
been responsible for less than 400 deaths since its discovery in 1967 (5). Although MARV is a
85
less visible threat to humans in terms of total number of outbreaks and deaths, it should be noted
86
that one of the largest and deadliest filovirus outbreaks in history was caused by MARV. An
87
outbreak of the Angolan variant of MARV (MARV/Ang) in Uige, Angola during 2004-2005
88
resulted in 227 deaths out of 252 total cases, a 90% mortality rate (31). This rate is similar to
89
that of the most devastating EBOV outbreak on record, and MARV/Ang infections currently
90
account for over half of all documented MARV cases and deaths (5).
91 92
The development of a mouse-adapted virus for modeling EBOV infections in mice (3) has
93
provided an economical and efficient method for the screening of candidate vaccines and post-
94
exposure therapies. Several promising strategies from these experiments have also demonstrated
95
efficacy in higher animal models, highlighted by the vesicular stomatitis virus-based vaccine
96
(VSVǻG/EBOVGP) (14) in addition to monoclonal antibody cocktails (ZMAb and MB-003)
97
(23) (24) (18) (21). However, the number of promising prophylactics and therapeutics against
98
MARV infections has lagged behind its EBOV counterpart; likely in part due to the lack of a
99
widely-available rodent-adapted virus that recapitulates MARV disease.
100 101
A rodent-adapted Ravn virus (RAVV) was developed previously to address this need. RAVV, a
102
virus within the Marburgvirus genus, was first adapted to severe combined immunodeficient
103
(SCID) mice (34), and then in immunocompetent BALB/c mice (35). Sequential passaging of
104
the Musoke and Ci67 variants of MARV (MARV/Mus and MARV/Ci67, respectively) in SCID
105
mice have also yielded lethal host-adapted variants to SCID mice (34); however, it is unknown
106
whether the mouse-adapted MARV/Mus and MARV/Ci67 are also lethal to immunocompetent
107
mice. Complicating matters further, these adapted viruses are not yet widely available for use
108
amongst Biosafety Level 4 (BSL-4) labs. Additionally, RAVV has 21% nucleotide divergence
109
from the MARV/Mus, whereas MARV/Ang is 7% divergent from MARV/Mus based on whole-
110
genome analysis (34). The glycoprotein (GP) gene shows 22% divergence between RAVV and
111
MARV/Ang in both nucleotide and amino acid sequences (31), suggesting that vaccines and
112
therapeutics raised against RAVV GP may not be protective against MARV challenge, and vice
113
versa. Indeed, studies have shown that vaccines based on Venezuelan equine encephalitis virus
114
(VEEV) expressing the GP and nucleoprotein from MARV/Mus are protective against
115
homologous virus challenge in cynomolgus macaques (13) but not protective against a RAVV
116
challenge (6), indicating that the differences between the two virus lineages are sufficient to
117
negatively impact the efficacy of potential candidate vaccines.
118 119
There is a need to develop MARV models for immunocompetent small animals using variants
120
that are antigenically distinct from RAVV. This is important for facilitating the development of
121
MARV vaccines or therapeutics so that they can first be tested in small animals with a high
122
predictability of protective efficacy, before follow-up studies in a higher animal species such as
123
non-human primates (NHPs). A guinea pig-adapted MARV/Ang has been previously described
124
(32), however sequence data for this MARV variant is not available. The aim of this study was
125
to adapt MARV/Ang, the MARV variant most pathogenic in humans, to immunocompetent mice
126
such that a challenge with a sufficient dose will kill the host within days. Here, we describe the
127
development and characterization of a mouse-adapted virus (MARV/Ang-MA) for use in the
128
immunocompetent BALB/c mouse. The adapted virus is characterized with regards to different
129
inoculation routes, various clinical parameters, cytokine, chemokine and growth factor
130
responses, as well as the pathology within the various organs during the course of disease.
131
MARV/Ang-MA was also sequenced and compared to the full length MARV/Ang sequence, in
132
order to distinguish the molecular differences between the two viruses.
133
Materials and Methods
134
Ethics statement
135
All animal work was performed according to Animal use document (AUD) #H-11-007, which
136
has been approved by the Animal Care Committee (ACC) based at the Canadian Science Center
137
for Human and Animal Health (CSCHAH), in accordance with the guidelines outlined by the
138
Canadian Council on Animal Care (CCAC). All infectious work was performed in the Biosafety
139
Level 4 (BSL-4) facility at the National Microbiology Laboratory in Winnipeg, Canada.
140 141
Cells and viruses
142
The wild-type virus was Marburg virus, strain Angola (Marburg virus H.sapiens-
143
tc/AGO/2005/Angola), termed MARV/Ang, which was isolated from a patient during the 2005
144
Marburg outbreak in Uige, Angola.
145
NML/M.musculus-lab/AGO/2005/Ang-MA-P2 (MARV/Ang-MA). Stocks of MARV/Ang and
146
MARV/Ang-MA were grown on T-150 flasks (Corning) of VeroE6 cells (ATCC) for use in the
147
studies.
The adapted virus is designated Marburg virus
148 149
Animals and the virus adaptation process
150
Severe combined immunodeficient (SCID) and wild-type BALB/c mice, female, 4-6 weeks old
151
(Charles River) were used for these studies. For passaging experiments, two SCID mice were
152
initially injected intraperitoneally (IP) with ~106 plaque forming units (PFU) of MARV/Ang in
153
200μL of Dulbecco’s modified Eagle’s medium (DMEM), supplemented with 2% heat-
154
inactivated fetal bovine serum (FBS) (Sigma). Serial passaging of MARV/Ang in SCID mice
155
was then performed as follows. Livers were removed from euthanized mice at 7 days post-
156
infection (dpi), pooled, and then homogenized by hand by grinding the livers against a steel
157
mesh with a plastic plunger. The cells were suspended in 10mL of phosphate buffered saline
158
(PBS) and centrifuged at 400 x g for 5 minutes. The supernatant was passed through a cell
159
strainer, and the cell pellet was further mechanically homogenized with a tissue homogenizer.
160
After centrifugation at 400 x g for 5 minutes, the supernatant was also passed through the cell
161
strainer. Two naïve SCID mice were injected IP with 200μL of the liver homogenate, and the
162
process was repeated again at 7 dpi until a MARV/Ang variant (termed MARV/Ang-MA) lethal
163
to both SCID and BALB/c mice was produced after 24 passages. The virus was not plaque
164
purified.
165 166
Median lethal dose challenge and pathogenesis experiments
167
For the median lethal dose (LD50) studies, 10-fold dilutions of MARV/Ang-MA ranging between
168
27.2 to 2.72 x 10-3 TCID50 was administered per SCID or BALB/c mouse (n=3 or 4 per dilution).
169
Mice were weighed daily and monitored for survival, weight loss and clinical signs of disease.
170
For the pathogenesis experiments, groups of 10 C57BL/6 mice were given 2000 x LD50 (100
171
TCID50) of MARV/Ang-MA IP, whereas groups of 10 BALB/c mice were administered 2000 x
172
LD50 of MA-MARV IP, intramuscularly (IM), intranasally (IN) or subcutaneously (SC). Mice
173
were weighed daily and monitored for survival, weight loss and clinical signs of disease.
174 175
Serial sampling experiments
176
BALB/c mice were infected either with 2000 x LD50 MARV/Ang-MA, or an equivalent titer of
177
MARV/Ang. Four to 6 mice from each group were randomly chosen to be euthanized at planned
178
time points between 0 and 6 dpi for necropsy. Blood was sampled from anaesthetized mice by
179
retro-orbital bleeding.
Whole blood was collected in K2 ethylenediaminetetraacetic acid
180
(EDTA) plus blood collection tubes (BD Biosciences) for blood counts and determination of
181
viremia, and in SST plus blood collection tubes (BD Biosciences) for blood biochemistry
182
studies. Whole blood from mice was also collected in EDTA plus blood collection tubes and
183
spun at 1000 x g for 10 minutes at 4oC in order to isolate plasma for cytokine responses. Parts of
184
organs including liver, spleen, kidney, lung, brain, and intestine were also harvested, weighed,
185
put in 1mL of DMEM supplemented with 2% heat-inactivated FBS, and then homogenized in a
186
tissue homogenizer. Samples were then spun at 400 x g for 5 minutes and frozen at -80oC until
187
processing. For histopathological and immunohistochemical analysis, whole livers and spleens
188
harvested from mice, and fixed in 10% phosphate buffered formalin in the BSL-4 laboratory for
189
at least 28 days before processing.
190 191
Blood counts and blood biochemistry
192
Complete blood counts were performed with the VetScan HM5 (Abaxis Veterinary Diagnostics).
193
The following parameters were included in the results: levels of white blood cells (WBC),
194
lymphocytes (LYM), percentage of lymphocytes (LY%), monocytes (MO%), neutrophils
195
(NE%), and levels of platelets (PLT). Blood biochemistry was performed with the VetScan VS2
196
(Abaxis Veterinary Diagnostics). The following parameters were included in the results: levels
197
of alkaline phosphatase (ALP), alanine aminotransferase (ALT), amylase (AMY), total bilirubin
198
(TBIL), blood urea nitrogen (BUN), and glucose (GLU).
199
200
Cytokine, chemokine and growth factor responses
201
Mouse cytokine, chemokine and growth factor levels were quantified with the Cytokine Mouse
202
20-Plex Panel for the Luminex platform (Life Technologies) and performed according to
203
manufacturer instructions. The following parameters were included in the results: fibroblast
204
growth factor (FGF-basic), interferon gamma (IFN-Ȗ), interleukin (IL)-1ȕ, IL-2, IL-4, IL-5, IL-6,
205
IL-10, IL-12, IL-13, IL-17, IFN-Ȗ induced protein (IP)-10, keratinocyte chemoattractant (KC),
206
monocyte chemoattractant protein (MCP)-1, monokine induced by IFN-Ȗ (MIG), macrophage
207
inflammatory protein (MIP)-1Į, tumour necrosis factor (TNF)-Į and vascular endothelial growth
208
factor (VEGF).
209 210
Infectious virus titrations
211
Titration of live MARV/Ang or MARV/Ang-MA was performed as follows. Whole blood or
212
supernatants from harvested organs were first serially diluted 10-fold in DMEM supplemented
213
with 2% heat-inactivated FBS. 100ȝL of the dilutions were then added in three replicates to a
214
96-well plate (Corning) of pre-seeded VeroE6 cells at 95% confluence and incubated at 37oC for
215
1 hour. The samples were removed and 100ȝL of fresh DMEM with 2% FBS was added to the
216
wells and incubated for 14 days. The plates were scored for the presence of cytopathic effects
217
(CPE) at 14 dpi, and titers were calculated with the Reed and Muench method (28) and
218
expressed as TCID50/mL, or TCID50/g of tissue. The lower detection limit for this assay is 10
219
TCID50/mL or 10 TCID50/g of tissue.
220
221
Histology and Immunohistochemistry
222
For histology studies, fixed liver and spleen tissues were embedded in paraffin wax to make
223
paraffin blocks.
224
microscope slides (Fisher) and incubated overnight at 37oC, deparaffinized with two changes of
225
xylene, immersed in 100%, 90% and then 70% ethanol for 2 minutes each, washed with distilled
226
water, and then stained with Hematoxylin (Thermo) as well as Eosin-Y (Therno) for 2 minutes
227
and 30 seconds, respectively, with a 2 minute wash of distilled water in between. The sections
228
are then dehydrated with 70%, 90% and 100% ethanol, cleared in two changes of xylene and
229
mounted with Permount (Fisher) for viewing by routine light microscopy.
The blocks were then cut 5 microns thick and mounted on Superfrost
230 231
For immunohistochemistry studies, paraffin blocks were deparaffinized, and washed with
232
distilled water as described above. Sections were first incubated with 10mM sodium citrate
233
buffer, pH 6.8 at 100oC for 10 minutes with the NxGen Decloaking Chamber (Inter Medico)
234
before immunohistochemistry was performed with the UltraVision Quanto mouse on mouse
235
HRP kit (Thermo) in the Lab Vision Autostainer (Thermo) following manufacturer instructions.
236
The in-house monoclonal antibody 3H1, which recognizes MARV GP was used at a 1:1000
237
dilution as a primary antibody.
238 239
Statistical analysis
240
Statistical comparison of the blood biochemistry, blood counts, and cytokine data was carried out
241
using R 3.1.1 (26) along with the reshape2 (37), ggplot2 (36), and car (9) packages. The data
242
used for the analysis consisted of 6 parameters for the blood biochemistry, 8 parameters for the
243
blood counts, and 20 parameters for the cytokines. Only data from days where both groups were
244
sampled was used. In order to account for the number of comparisons, three multiple analysis of
245
covariance (MANCOVA) were used, one for the biochemistry data, one for the blood count data,
246
and one for the cytokines.
247
considered as categorical with 4 values: 1, 3, 5, and 6 (three for the cytokines: 3, 5, and 6). In all
248
cases, the MANCOVAs were significant and were followed by individual ANCOVAs for each
249
outcome. In the case where the ANCOVA was significant, a post-test consisting of Tukey’s
250
Honest Significant Differences was used to compare the viruses at different times.
251
assumption of homoscedasticity was tested for each outcome using Levene’s test. For the blood
252
biochemistry data, all outcomes except GLU showed heteroscedasticity so a log-transform was
253
used on all outcomes including GLU in order to keep the scales consistent.
254
transformation, AMY still showed significant heteroscedasticity and a number of transformations
255
were attempted (see output in Supplementary Material for details), but none corrected the
256
problem. As a consequence, weights (1/y2) were applied to the regression to compensate for the
257
increased impact of the highly variable values. For the blood count data, all outcomes passed
258
Levene’s test, so no transformations were applied. For the cytokine data, all cytokines were
259
heteroscedastic and three (IFNȖ, GM-CSF, and IL-1ȕ ) were still heteroscedastic after a log
260
transformation. The source of heteroscedasticity for those samples appeared to be groups where
261
all or most measurements were at background level; because there are no well-documented
262
alternatives to the ANCOVA procedure, we elected to analyse those cytokines in the same
263
manner but the p-values obtained were interpreted more conservatively. Scripts and output are
264
provided in the supplementary materials. For all analyses, statistical significance was set at p
