Background
The family
Filoviridae consists of two genera called ebolavirus (EBOV) and marburgvirus (MARV) that are considered significant public health threats due to their very high morbidity and mortality rates (up to 90% case fatality rate), human-to-human transmissibility, and environmental stability. Due to these characteristics, and the fact that the filoviruses have a low infectious dose [<1 plaque-forming units (pfu)] and can be easily produced to >10
8 pfu/ml
in vitro or
in vivo [
1‐
4], the filoviruses are classified as biosafety level (BSL)-4 agents and Category A biothreat agents by the Centers for Disease Control and Prevention [
5,
6]. Initial symptoms of filovirus infection include nonspecific clinical signs such as high fever, headache, myalgia, vomiting and diarrhea, followed by leukopenia, thrombocytopenia, lymphadenopathy, pharyngitis, edema, hepatitis, maculopapular rash, hemorrhage, and prostration with death generally occurring within 5–10 days of infection [
1,
7].
The first known filovirus outbreaks occurred in simultaneously in both Germany and Yugoslavia in 1967 when laboratory workers became infected from blood and tissues of MARV-infected African green monkeys imported from Uganda [
8,
9]. Subsequent MARV cases or outbreaks have occurred in South Africa, Zimbabwe, Kenya, Democratic Republic of Congo, and Angola with case fatality rates ranging from 20% in Germany in 1967 [
8,
9] to >90% in Angola during 2004–5 [
10]. It is generally considered that transmission of the filoviruses requires direct contact with blood, body fluids, or tissues from an infected individual [
11,
12], although droplet and aerosol transmissions may also occur [
13].
Human-derived Marburg viruses (isolates Musoke, Ravn, and Ci67) are not lethal to immmunocompetent adult mice. Previously, an Ebola Zaire mouse-adapted virus was developed by performing 9 sequential passages of Ebola Zaire '76 virus in suckling mice followed by two sequential plaque picks. The resulting virus was uniformly lethal to mice after intraperitoneal inoculation [
14]. Pathologic evaluation of infected mice identified similarities and differences between this model [
14,
15] and infections in nonhuman primates [
16,
17]. Similarities include the tropism of the virus for monocytes/macrophages and high viral titers in the spleen and liver tissues after infection [reviewed in [
18]]. The mean time to death of infected mice is approximately 5–10 days, which is similar to that observed in infected cynomolgus and rhesus macaques.
A viable lethal mouse model for Marburg virus is critical to the filovirus vaccine research program to understand the immune mechanisms that need to be induced, or avoided, by vaccination. Furthermore, a mouse model would speed the testing and evaluation of new Marburg therapeutic candidates. This effort is currently impeded due to limitations in the numbers of guinea pigs that can be evaluated at one time (based on BSL-4 space limitations, as well as physical demands on investigators and technicians) and the large amounts of compounds that must be synthesized or purified for testing in guinea pigs, which are 20–50× the size of mice. The purpose of this work was to select a marburgvirus that caused death within a similar timeframe as monkeys or humans (7–10 days) in severe combined immunodeficiency (scid) mice. To accomplish this goal, we repeatedly passaged the liver homogenates of MARV-infected scid mice and then recorded their time to death. Once we identified rapidly lethal mouse-adapted viruses, we characterized the models by immunology and pathology studies. These scid mouse-adapted viruses will be used to explore the virulence factors associated with marburgvirus infection. Furthermore, the scid models of MARV infection will be particularly useful for screening candidate therapeutics for their ability to directly diminish viral replication in the absence of adaptive immune responses.
Methods
Virus and cells
Human-derived (wild-type) and mouse-adapted MARV-Musoke, -Ravn, and -Ci67 virus stocks were propagated no more than three passages in Vero or VeroE6 cells. The human-derived (wild-type) and mouse-passaged MARV-Musoke, -Ravn, and -Ci67 plaques were counted by standard plaque assay on Vero cells [
19]. MARV-infected cells and animals were handled under maximum containment in a BSL-4 laboratory at the United States Army Medical Research Institute of Infectious Diseases.
Animals
BALB/c severe combined immunodeficient (scid) mice, aged 4 to 8 weeks, of both sexes, were obtained from National Cancer Institute, Frederick Cancer Research and Development Center (Frederick, MD). Mice were housed in microisolator cages and provided autoclaved water and chow ad libitum. Research was conducted in compliance with the Animal Welfare Act and other federal statutes and regulations relating to animals and experiments involving animals and adhered to principles stated in the Guide for the Care and Use of Laboratory Animals, National Research Council, 1996. The facility where this research was conducted is fully accredited by the Association for Assessment and Accreditation of Laboratory Animal Care International.
Mouse adaptation
The general approach to adapt MARV to mice was based on virus passage in scid (BALB/c background) mice to avoid usage of suckling mice to develop a lethal mouse-adapted Marburg virus. The goal was to isolate the viral population that was capable of migrating to target tissues/organs (i.e., liver) at the earliest time point. Each group consisted of 10 mice that were inoculated intraperitoneally (IP) with 1000 pfu of Marburg virus (isolate Musoke, Ci67, or Ravn). Two mice were euthanized on day 7, the livers removed, pooled, and homogenized in 10 ml of PBS. The liver homogenates were blindly passed (200 μl IP) and used to infect new mice to evaluate lethality of the next virus passage. Lethality was monitored in the remaining eight mice of each passage. The supernatants of the liver homogenates from each passage were introduced onto Vero cells to determine the viral titers by plaque assay [
19].
Viral challenge with 'scid-adapted' MARV
For the characterization studies, scid mice were injected IP with ~1000 pfu of 'scid mouse-adapted' MARV-Musoke, passage (P)10; Ravn P(10); or Ci67, P(15) diluted in PBS. After challenge, mice were observed at least twice daily for illness and death and in some experiments, daily weights were determined for each infected group.
Hematologic studies
For mice, blood samples were obtained under anesthesia by cardiac puncture. Viremia was assayed by traditional plaque assay [
19]. Hematological, cytokine, and D-dimer levels, as well as liver-associated enzymes, were measured as previously described [
20,
21].
Pathologic sampling
Four animals from each group were randomly chosen for euthanasia on 2, 4, 6, and 8 days postchallenge for gross necropsy. A full complement of tissues from each mouse was fixed in 10% neutral buffered formalin and held in the BSL-4 laboratory for >21 days. The tissues were embedded in paraffin, sectioned for histology, and stained with hematoxylin and eosin for routine light microscopy or were stained by an immunoperoxidase method (Envision System – DAKO Corporation, Carpinteria, CA), using a mixture of two mouse monoclonal antibodies against MARV nucleoprotein (NP) and glycoprotein, or by the TUNEL method to detect apoptotic cells within the tissue samples.
Adminstration of antisense PMO and filovirus-specific antisera
Two groups of 10 scid mice were each administered 1 ml of convalescent sera from guinea pigs that had survived either EBOV or MARV infection. The antibodies were administered IP 1 h after challenge. Both pools of antisera had 80% plaque reduction-neutralization titers of >1:160 against the homologous virus, but <1:20 against the heterologous virus. Alternately, another group of 10 scid mice were administered IP with 1 mg of a mixture of four MARV-specific phosphodiamidate morpholino oligomers (PMOs) targeting the AUG start site of VP24, VP35, NP, and L (kind gift of Dr. P.L. Iversen of AVI BioPharma, Inc., Corvallis, OR) 1 h after challenge. A control group received saline (i.e., vehicle) alone. The mice were challenged with 1000 pfu of 'scid-adapted' MARV-Ci67 and monitored for survival.
Discussion
In previous studies, scid mice became ill and died within 3–4 weeks after inoculation with ZEBOV ('76), Sudan EBOV, or GP-adapted MARV-Ravn, but not with the other viruses [
25]. However, the scid mice in these studies were only observed for 40 days after the infection – a much shorter time than we found required to produce lethal disease with the human-derived, wild-type viruses. The MTD of scid mice infected with the wild-type MARV isolates was not previously reported elsewhere. We found the time-to-death using wild-type MARV infections in scid mice much too long (50–70 days) to feasibly screen the efficacy of a large number of potential therapeutics
in vivo. Therefore, we passaged the viruses until the time to death was consistently in the range of 7–10 days. These more virulent 'scid-adapted' viruses will allow for more rapid and efficient testing of candidate prophylactic and therapeutic treatments against multiple MARV isolates.
Initial serial sampling studies to characterize the pathology of these more virulent, scid-adapted MARV strains indicate similarities to the filovirus disease observed in other models. After parenteral challenge, the incubation period for MARV is 2 to 6 days, with death typically occurring between 7 and 11 days after infection in both guinea pigs and nonhuman primates [
26‐
30]. Initial indicators of MARV disease in all the animal models include fever, anorexia, rash, huddling, weight loss, dehydration, and diarrhea. More severe complications such as prostration, failure to respond to stimulation, hind limb paralysis, and bleeding from injection sites and/or body orifices develop at later times after infection (i.e., 6–10 days) [
26‐
30]. As noted here and in other models, the liver and spleen are tissues most consistently affected by MARV, as assessed by gross appearance, microscopy and histology. Based on pathology studies of the scid mice, guinea pigs, and nonhuman primates, the primary targets of MARV infection appear to be phagocytic cells, followed by hepatocytes, endothelial cells and fibroblastic cells [
26‐
30]. Clinically, the scid mouse model appears to also be similar to the guinea pig and nonhuman primate models. MARV virus was present at increasingly high titers in the blood (Figure
2A), liver, spleen, kidneys, and other major organs (data not shown). Furthermore, early hematological and immunological changes included lymphopenia, variable neutrophilia, and profound thrombocytopenia [Figure
4 and [
26‐
30]]. Notable alterations in serum chemistry levels, especially liver enzymes, occurred with increasing severity after infection (Figure
3). However, unlike nonhuman primates, rodents such as mice, guinea pigs, and hamsters are not susceptible to primary human isolates of MARV virus directly from blood or organ homogenates derived from infected patients [
27,
29‐
31].
Rodents infected with filoviruses appear to have slightly different coagulopathic responses than filovirus-infected nonhuman primates [
14,
26‐
30,
32]. Similarities of the models include profound and rapid loss of circulating platelets, increased D-dimer levels, and uncontrolled bleeding (Figure
3D, data not shown, and [
32]). For EBOV, rodents do not display all the characteristics of disseminated intravascular coagulation (DIC) that filovirus-infected nonhuman primates show including prolongation of PT and aPTT, circulating fibrin degredation products (FDPs), decreased plasma fibrinogen, and increased tissue fibrin deposition [
32]. Not all these parameters have yet been tested for the MARV scid mouse model and will surely be the subject of future work.
Sequence comparisons of the original wild-type and more virulent scid-adapted MARV are required. Based on previous reports with mouse and guinea pig-adapted EBOV [
18,
33,
34], we predict changes in VP24, VP35, NP, and L are likely to be important for enhanced virulence of the 'scid-adapted' MARV. VP24 was recently implicated in host pathogenicity as VP24 is an interferon antagonist that functions by binding karyopherin-α1 and blocking nuclear accumulation of the interferon signaling molecule stat1 [
35,
36]. The NP, VP35, and L proteins are all critical for viral replication and alterations in these proteins may lead to advantages in viral replication/growth within a given host species. NP is the viral nucleoprotein that tightly couples with the viral RNA [
37]. Together, the L protein, VP30, and VP35 form the filovirus RNA-dependent RNA polymerase [
37]. The VP35 is also implicated in blocking interferon (IFN) type-I responses in filovirus-infected cells by inhibiting double-stranded RNA-mediated activation of interferon regulatory factor 3, a transcription factor which triggers expression of interferon and interferon-stimulated genes [
38‐
41]. Future experiments using reverse genetics could help demonstrate which of the acquired mutations were important for adaptation to mice.
This scid mouse model of MARV infection has obvious uses as a model for analysis of therapeutics and candidate antibodies. It will be much more efficient for the purposes of quickly screening lead compounds and neutralizing antibodies than guinea pigs or nonhuman primates. Before the development of this novel scid mouse model of MARV, a large quantity of antiviral compound and/or antibodies was required to achieve relevant physiological levels in guinea pigs, which are 20–50 times larger than mice, for the purposes of initial in vivo efficacy studies. Furthermore, guinea pigs require much larger cages, limiting the number of animals within a study and are also much more difficult and dangerous to handle under BSL-4 conditions than mice, requiring at least two laboratory personnel for treatments and challenges. A delay in time to death in this newly developed scid mouse model of MARV infection will indicate a positive result that should be followed up in the more intensive and expensive guinea pig studies. Thus, this novel MARV mouse model will allow for faster and more efficient in vivo screening of potential MARV prophylactics and therapeutics.
Acknowledgements
The authors thank C.A. Mech, J. Wells, M.T. Cooper, N.A. Posten and C. Rice for excellent technical assistance, Dr. Patrick L. Iversen of AVI BioPharma for providing MARV-specific PMOs, and Drs. A.L. Schmaljohn, D.L. Swenson, M.J. Aman and K.E. Steele for suggestions and helpful discussions. A portion of the research described herein was sponsored by the Defense Threat Reduction Agency JSTO-CBD and the Medical Research and Material Command. Opinions, interpretations, conclusions, and recommendations are those of the authors and are not necessarily endorsed by the U.S. Army.