General Concepts

Introduction

Filoviruses were first discovered in 1967 as the causative agents of a hemorrhagic fever outbreak among laboratory workers in Europe. These workers had been exposed to tissues and blood from African green monkeys (Cercopithecus aethiops) imported from Uganda and infected with Marburg virus. Since then, sporadic cases of Marburg hemorrhagic fever in man have occurred in Kenya and Zimbabwe (Table 72-1).

Table 72-1

Outbreaks of filoviral hemorrhagic fever documented by virus isolation.

Ebola hemorrhagic fever was first reported from northern Zaire and southern Sudan in 1976 when two distinct subtypes were isolated during simultaneous epidemics. Ebola Sudan reemerged in 1979 at the same location, causing a smaller epidemic of viral hemorrhagic fever. The third subtype (Reston) was isolated from cynomolgus monkeys (Macaca fascicularis) imported from the Philippines into the United States in 1989 and into Italy in 1992. Another distinct Ebola virus emerged on the Ivory Coast in 1994. The virus was isolated from a nonfatal case, in which a worker was infected during the autopsy of a wild chimpanzee. Recently Ebola reemerged in southwestern Zaire in the city of Kikwit and the surrounding villages in Bandundu Province. The isolated virus was closely related to the 1976 Zairian isolate, and the outbreak in these regions resembled the previous Ebola hemorrhagic fever epidemics (Table 72-1).

Clinical Manifestation

Filoviruses cause a severe hemorrhagic fever in human and non-human primates. Disease onset is sudden, with fever, chills, headache, myalgia, and anorexia. These symptoms may be followed by abdominal pain, sore throat, nausea, vomiting, cough, arthralgia, diarrhea, and pharyngeal and conjunctival vasodilitation. Patients are dehydrated, apathetic, and disoriented. They may develop a characteristic, nonpruritic, maculopapular centripetal rash associated with varying degrees of erythema, which desquamates by day five or seven of the illness. Hemorrhagic manifestations develop at the peak of the illness, and are of prognostic value. Bleeding into the gastrointestinal tract is the most prominent, besides petechia and hemorrhages from puncture wounds and mucous membranes. Laboratory parameters are less characteristic, but the following are associated with the disease: leukopenia (as low as 1000/μl), left shift with atypical lymphocytes; thrombocytopenia (50,000-100,000/μl); markedly elevated serum transaminase levels (typically AST exceeding ALT); hyperproteinemia; and proteinuria. Prothrombin and partial thromboplastin times are prolonged, and fibrin split products are detectable. In a later stage, secondary bacterial infection may lead to elevated white blood counts. There is fever in patients who eventually recover for about five to nine days. In cases ending in death, clinical signs occur at an early stage and the patient dies between day six and 16, from hemorrhage and hypovolemic shock. The mortality rate is between 30 and 90%, depending on the virus, and the highest rate has been reported for Ebola Zaire. Ebola Reston seems to possess a very low pathogenicity for humans or may even be apathogenic. Convalescence is prolonged and sometimes associated with myelitis, recurrent hepatitis, psychosis or uveitis. An increased risk of abortion exists for pregnant women, and clinical observations indicate a high death rate for children of infected mothers.

Classification

Filoviruses are classified as “Biological Level 4” agents (WHO; Risk Group 4) based on their high mortality rate, person-to-person transmission, potential for aerosol infectivity, and absence of vaccines and chemotherapy. Maximum containment is required for all laboratory work with infectious material. Filoviruses can be separated into two clearly distinct types by the features listed in Table 72-2. In general, type Marburg seems unique, having no known subtypes, but at least two different genetic lineages. Ebola, however, can be subdivided into three subtypes—Zaire, Sudan, and Reston—and sequence analysis of the 1994 Ivory Coast isolate strongly indicates the existence of a fourth subtype. There is a lack of antigenic cross-reactivity between the types, but the subtypes of Ebola share common epitopes. Genetic variability among the different virus isolates seems much less in comparison to other RNA viruses, suggesting that variants may not emerge as rapidly in nature. Molecular analyses of the genomes clearly demonstrated that filoviruses are the closest relatives to Rhabdoviridae and Paramyxoviridae, and thus support the concept of the order Mononegavirales comprising the three unique families of nonsegmented, negative-strand (NNS) RNA viruses. Filovirus genomes are more complex than those of lyssaviruses and vesiculoviruses and align organizationally more closely to members of the genera paramyxovirus and morbillivirus.

Table 72-2

Characteristics of type ‘Marburg’ and ‘Ebola’ of filoviruses.

Virus Structure

(A) Genome

Genomes of filoviruses consist of a single negative-stranded, linear RNA molecule. The RNA is noninfectious, not polyadenylated, and complementary to polyadenylated viral subgenomic RNA species. Filovirus genomes are approximately 19 kb long (Marburg, 19.1 kb; Ebola, 18.9 kb) and are organized as illustrated in figure 72-1A. Genes are defined by highly conserved transcriptional start signals at their 3′ (3′-CUNCNUNUAAUU-5′; consensus motif) and termination signals at their 5′ ends (3′-UAAUUCUUUUU-5′) (Fig. 72-3). They are separated by intergenic regions varying in length and nucleotide composition. Some genes overlap, but the positions and numbers of overlaps differ between the two types of viruses (Ebola type, VP35-VP40, GP-VP30 and VP24-L; Marburg type, VP30-VP24) (Fig. 72-1A). The length of overlaps is limited to 5 highly conserved nucleotides (3′-UAAUU-5′) within the transcriptional signals (Fig. 72-3). Most genes tend to possess long noncoding sequences at their 3′ and/or 5′ ends, which contribute to the increased length of the genome. Upstream of the N gene start site and downstream of the L gene stop site are extragenic sequences present which are thought to be templates for very small viral, nonpolyadenylated subgenomic RNAs. The genomes are complementary at the very extreme ends.

Figure 72-1

(A) Genome organization of Marburg and Ebola viruses. (B) Expression strategies of gene 4 of Ebola type viruses. (A) Filoviral genomes consist of a single, negative-stranded, linear RNA molecule. Differences in the organization between Marburg and Ebola (more...)

Figure 72-3

Replication of filoviruses. The nonsegmented negative-stranded RNA genome is transcribed into subgenomic RNAs which are polyadenylated at their 3′ and presumably capped at their 5′ ends. Replication works via a full length (+)-strand antigenome (more...)

(B) Viral proteins

NP - nucleoprotein

NP proteins possess an unusually high Mr with 95 kDa for Marburg and 105 kDa for Ebola isolates. The protein is the major structural phosphoprotein, and only the phosphorylated form is incorporated into virions, as demonstrated for Marburg. It is the major component of the ribonucleoprotein complex (RNP). Although RNA binding has not yet been demonstrated, there is little doubt that this protein is the functional analogue of the nucleocapsid proteins of other NNS RNA viruses (Fig. 72-2).

Figure 72-2

Structure of filoviral particles. (A) Electron micrograph of Marburg virus. Ultrathin sections obtained from primary cultures of human endothelial cells three days post-infection. Particles consist of a nucleocapsid surrounded by a membrane in which spikes (more...)

GP - glycoprotein

GPs are directed into the endoplasmic reticulum by an N-terminal hydrophobic domain which is cleaved by signal peptidases, as shown directly for Marburg. They are anchored via a C-terminal hydrophobic domain in the membrane (type I transmembrane protein) and contain N- and O-glycans that account for up to 50% of the Mr of the mature proteins. Oligosaccharide side chains differ in their terminal sialylation pattern which seems to be isolate- and cell line-dependent. Detailed structural analyses are available for Marburg which include oligomannosidic and hybrid type N-glycans, as well as bi-, tri-, and tetraantennary complex species, and high amounts of neutral mucin-type O-glycans. N- and C-terminal ends of GP are conserved and carry most of the cysteine residues. Two cysteine residues (positions 671 and 673) are acylated as shown for Marburg. The middle region is variable, extremely hydrophilic, and carries the bulk of the N and O-glycosylation sites. For Marburg it has been shown that the mature GP is inserted in the membrane as a homotrimer and oligomerization seems to be mediated by intramolecular disulfide bridges. GP may mediate binding to cellular receptors and subsequent fusion with cellular membranes. It is further discussed as the major viral antigen (Fig. 72-2).

L - large protein

L proteins show significant homologies to L proteins of other NNS RNA viruses. Homologies are mainly located in the N-terminal half of the protein and concentrated within three common boxes (A, B, C). A highly conserved peptide motif -GDNQ-, located at the C-terminal end of domain B (positions 744-747) flanked by hydrophobic amino acid residues, possibly correlates with enzymatic functions of the protein. Even though transcriptase and replicase activities have not yet been demonstrated, the L protein is regarded as an RNA-dependent RNA polymerase (Fig. 72-2).

VP35 and VP30

VP35 and VP30 are components of the RNP, but the association of VP35 is weak compared with NP and VP30. Ebola VP30 has been identified as the minor (second) phosphoprotein of virions, whereas VP35 is not phosphorylated in virion particles. Expression studies of Marburg VP35 and VP30 in insect cells (SF9 cells), however, revealed weak phosphorylation for both proteins, whereas VP35 expressed in HeLa cells using the vaccinia virus-driven T7 polymerase system was not phosphorylated. Despite inconsistent data on phosphorylation among filoviruses and the lack of sequence homology, the genome position of the corresponding gene combined with the association in the RNP suggest that VP35 is functionally analogous to the P proteins of other NNS RNA viruses. VP30 may work as a functional unit in encapsidation (minor nucleoprotein) and/or play a role as a cofactor for the transcriptase/replicase (Fig. 72-2).

VP40 and VP24

VP40 and VP24 are membrane-associated proteins and not associated with the RNP. The predominantly hydrophobic profile, the abundance in virion particles, and the genome localization of the corresponding gene suggest VP40 is the matrix protein. VP24 presumably serves as a second matrix protein and may link the other membrane proteins to the RNP (Fig. 72-2).

Nonstructural protein of unknown function

A nonstructural glycoprotein has only been discovered with viruses of the Ebola type. This protein, designated sGP, shares ~ 300 N-terminal amino acids with GP, but has a different C terminus (~70 amino acids) containing many charged residues as well as conserved cysteines. The protein is directed into the endoplasmic reticulum, becomes N- and O-glycosylated, and is secreted into culture medium (Fig. 72-1B).

Virus Replication

Cell entry is possibly mediated by GP as the only surface protein of virion particles. Studies on Marburg (strain Musoke) infection of hepatocytes have identified the asialoglycoprotein receptor as a receptor candidate. However, one has to postulate additional receptors, since this receptor is not expressed on many virus-susceptible cells, and Marburg GP does not generally lack sialic acids. It is not known if the next step in virus entry involves a fusion process at the plasma membrane or fusion following endocytosis of virus particles. Neither has the uncoating mechanism been studied. Filovirus transcription and replication take place in the cytoplasm of infected cells (Fig. 72-3). Transcription probably starts with a short (+)-leader sequence, and subsequently genomes are transcribed into monocistronic subgenomic RNA (mRNA) species complementary to viral genomic RNA. The 3′ ends of the transcripts carry a poly(A) tail generated by a stuttering mechanism of the viral polymerase at a run of uridine residues located in the transcription termination signals. The 5′ ends of the transcripts seem to be capped, and the sequences show a potential for formation of stable hairpin structures, which may play a role in transcript stability and ribosome binding. The characteristic pentamer (3′-UAA UU-5′), present in all transcriptional signals, could serve as the recognition site for the polymerase complex. The surrounding semiconserved sequences may then mediate the exact initiation of transcription and termination/polyadenylation events. The role of gene overlaps in regulating transcription is unknown, but transcription may be reinitiated by reposition of the polymerase at the downstream start site (back up-mechanism). Alternatively, the polymerase may occasionally terminate transcription at the overlap and initiate transcription of the downstream gene without polyadenylation of the upstream gene. The switch mechanism between transcription and replication has not been studied. The fact that the extremities of the genomes are complementary suggests a single identical encapsidation site on the genome and antigenome and an identical entry signal for the polymerase complex for both transcription and replication. Replication works via a full-length (+)-strand antigenome which will be encapsidated and serve as the template for synthesis of (-)-strand genome molecules. Virions usually bud at the plasma membrane. Mature particles exit preferentially in a vertical mode, but budding via the longitudinal axis has also been observed.

With the Ebola GP gene, transcription occurs from two open reading frames. The primary gene product is a small nonstructural glycoprotein that is secreted from infected cells. In order to express the full-length GP, two independent mechanisms are discussed: transcriptional editing of a single nucleotide at a run of uridine residues or translational frame shifting (-1) at or just past the editing site of unedited transcripts (Fig. 72-1B). Marburg GP, however, is expressed in a single frame, and the gene does not contain sequences favoring mechanisms such as editing or frame shifting.

Epidemiology

The reservoir of filoviruses remains a mystery. Species such as guinea pigs, primates, bats, and hard ticks have been discussed as possible natural hosts; however, no non-human vertebrate hosts or arthropod vectors have been identified.

Person-to-person transmission by intimate contact is the main route of infection in human outbreaks. Transmission seems to be inefficient, as documented by secondary attack rates rarely exceeding 10%. Nosocomial transmission via contaminated syringes and needles has been a major problem, especially in the 1976 and 1995 Ebola outbreaks in Zaire. Transmission by droplets and small-particle aerosols was observed in outbreaks among experimentally infected (Marburg) and quarantined imported monkeys (Ebola Reston, 1989-90). This is supported by identification of filovirus particles in alveoli of naturally and experimentally infected monkeys. Courses of human outbreaks, however, indicate that aerosols and droplets do not seem important modes of transmission.

Marburg and subtypes Sudan and Zaire of Ebola appear to be indigenous to the African continent, and both Ebola subtypes have been isolated from human patients only in Africa. Marburg has been isolated from human patients in Africa and Europe; however, the European cases were caused by a virus which originated in Africa. The Ebola Reston outbreak documented for the first time the presence of a filovirus in Asia. Serological studies (IFA) among captive macaques in the Philippines indicated that the source of Ebola Reston might be wild non-human primates. However, IFA-detected antibodies seem to be spurious and latent infection in non-human primates has never been observed (Table 72-1).

Serological studies suggest filoviruses are endemic in many countries of the central African region. Recent serosurveys imply that filoviruses might also be endemic in other countries (Germany, United States, Philippines). Although serological data based on IFA are only of limited reliability, they at least suggest the possible occurrence of subclinical infections caused by known or unknown filoviruses.

Pathogenesis

The incubation period for rhesus and African green monkeys inoculated with Marburg and Ebola, subtype Zaire, is 4 to 16 days. High virus titers can be detected in liver, spleen, lymph nodes, and lungs by the time of onset of clinical symptoms. All of these organs, especially the liver, are necrotized by virus replication in parenchymal cells. There is little inflammatory response at those sites, which suggests that classical immunopathology may not be an important pathogenic consideration. Interstitial hemorrhage occurs and is most prominent in the gastrointestinal tract. In infected non-human primates, thrombocytopenia, accompanied by aggregation disorders of remaining platelets in response to agonists such as ADP and collagen, has been found. Histopathological damage of the target organs is at odds with serum transferase levels showing increase of ALT and AST in a ratio of AST:ALT of 7:1. Recent morphologic studies on Ebola Reston-infected monkeys from the 1989 outbreak demonstrated extensive virus replication in tissue macrophages, interstitial fibroblasts of many organs, and in circulating monocytes/macrophages. These were seen less frequently in endothelial cells, hepatocytes, adrenal corticoid cells, and renal tubular epithelium. Similar results have been reported from experimental infected monkeys.

In human cases resulting in death, generalized hemorrhage was found macroscopically in most organ systems. Microscopic changes included focal necrosis in liver, lymphatic organs, kidneys, testes, and ovaries. The liver, while universally involved with large eosinophilic intracytoplasmic inclusion bodies in hepatocytes and Councilman-like bodies within necrotic foci, was not the site of massive, potentially fatal necrosis. Generalized lymphoid necrosis is characteristic of the disease, and renal tubular necrosis is commonly found in agonal stages. A diffuse encephalitis, as described for many viral infections, has been observed in these patients. Activation of the clotting system occurred and intravascular fibrin thrombi were observed. Viral antigen was detected in many organs, but predominantly in the liver, kidneys, spleen, and adrenal glands. Viral persistence has been demonstrated for Marburg cases by isolation of virus from liver biopsy material and the anterior chamber of the eye (at 4 to 5 weeks), and from semen (at 12 weeks), despite an apparently normal immune response.

Pathophysiological changes that make filovirus infections so devastating is just beginning to be unraveled. Clinical and biochemical findings support the anatomical observations of extensive liver involvement, renal damage, changes in vascular permeability, and activation of the clotting cascade. Visceral organ necrosis is a consequence of virus replication in parenchymal cells. However, no organ, not even the liver, shows sufficient damage to account for death. The role of disseminated intravascular coagulation (DIC) in pathogenesis is still controversial, since there has been no laboratory confirmation of DIC in human infections. Laboratory para-meters in the crucial early stage of filoviral hemorrhagic fever suggest extrahepatic targets in the infection. Fluid distribution problems and platelet abnormalities are dominant clinical manifestations indicating dysfunction or damage to endothelial cells and platelets. At post mortem, there is little monocyte/macrophage infiltration at sites of parenchymal necrosis, suggesting that a dysfunction of white blood cells, such as macrophages, occurs. Monocytes/macrophages and fibroblasts may be the preferred sites of virus replication in early stages. Other cell types such as endothelial cells may become involved as the disease progresses.

The lack of evidence for massive direct vascular involvement in infected hosts supports the role of active mediator molecules in the pathogenesis of the disorders. Although the source of these mediators during filovirus infections is still unknown, candidate cells exist. Besides the endothelium, the common denominator remains the macrophage, which is known as a pivotal source of different protease, H₂O₂, and mediators such as tumor necrosis factor-alpha (TNF-a). Recently supernatants of filovirus-infected monocyte/macrophage cultures have been shown to increase paraendothelial permeability in an in vitro model. Examination for mediators in those supernatants revealed increased levels of secreted TNF-a, the prototype cytokine of macrophages. These data support the concept of a mediator-induced vascular instability, and thus increased permeability may be a key mechanism for the development of the shock syndrome in severe and fatal cases. The bleeding tendency could be due to endothelial damage caused directly by virus replication, as well as indirectly by cytokine-mediated processes. Furthermore, the combination of viral replication in endothelial cells and virus-induced cytokine release from macrophages may also promote a distinct proinflammatory endothelial phenotype that then triggers the coagulation cascade (Fig. 72-4).

Figure 72-4

Possible role of monocytes/macrophages and endothelial cells in the development of filoviral hemorrhagic fever. (i) Infection of monocytes/macrophages leads to activation and release of various cytokines and mediators. (ii) Cytokines may cause upregulation of (more...)

Host Defenses

The mechanisms of recovery from filovirus infections in humans and in wild as well as laboratory animals are unknown. In vitro neutralization by serum antibody has never been demonstrated by plaque reduction in cell culture systems, and protection by convalescence sera has never been evaluated by controlled clinical trials. Fatal filovirus infections usually end with high viremia and no evidence of an immune response. In humans and monkeys, infection leads to an extensive disruption of the parafollicular regions in the spleen and lymph nodes that contain the antigen-presenting dendritic cells. Ebola Reston infection in monkeys is an exception in that it shows a rise in nonneutralizing antibodies shortly before death. Thus, cell-mediated immunity may mediate recovery from filovirus infections.

GP is assumed to be the major antigenic molecule of virion particles, but its interaction with the host immune system may be modulated by a high carbohydrate content. For Ebola, sGP production and secretion might interfere with the host immune response by neutralizing effective antibodies. In addition, GP molecules carry a sequence near their C-termini resembling immunosuppressive domains found in retroviruses. Peptides synthesized according to that 26 amino acid-long region inhibited the blastogenesis of lymphocytes in response to mitogens, induced cytokine production, and increased mononuclear cell proliferation in vitro. Infected animals showed increased levels of mediators, in particular of interferon and TNF. Activation of natural killer (NK) cells has been observed at earlier stages of the infection, but ultimately there are no NK cells. These findings are in line with the observation of immunosuppression in monkeys experimentally infected with Marburg and Ebola and of proliferation of filoviruses in macrophages and monocytes in vivo and in vitro.

Diagnosis

In tropical settings, filoviral hemorrhagic fever may be difficult to identify, since the most common causes of severe, acute, febrile disease are malaria and typhoid fever. A wide range of infectious diseases have to be considered before making a diagnosis of filovirus. Travel, treatment in local hospitals, and contact with sick persons or wild and domestic monkeys are useful historical features in returning travelers, especially in those from Africa. Diagnosis of single cases is extremely difficult, but occurrence of clusters of cases with prodromal fever followed by cases of hemorrhagic diatheses and person-to-person transmission are suggestive of viral hemorrhagic fever, and containment procedures must be initiated. In filoviral hemorrhagic fever prostration, lethargy, wasting, and diarrhea are usually more severe than that observed with other viral hemorrhagic fever patients. The rash is characteristic and extremely useful in differential diagnosis.

Laboratory diagnosis can be achieved in two different ways: by measurement of the host-specific immunological response to the infection and by detection of viral antigen and genomic RNA in the infected host. The most commonly used assays to detect antibodies to filoviruses are the indirect immunofluorescence assay (IFA), immunoblot, and enzyme-linked immunosorbent assays (ELISA) (direct IgG and IgM ELISA, and IgM capture assay) (Table 72-3). Direct detection of viral particles, viral antigen and genomic RNA can be achieved by electron microscopy (negative contrast, thin-section), immunohistochemistry, immunofluorescence on impression smears of tissues, antigen detection ELISA, and reverse transcriptase-polymerase chain reaction (RT-PCR) (Table 72-3).

Table 72-3

Laboratory diagnosis.

Attempts to isolate the virus from serum and/or other clinical material should be performed using Vero or MA-104 cells (monkey kidney cells) (Table 72-3). However, most filoviruses do not cause extensive cytopathogenic effects on primary isolation. The most useful animal system, besides non-human primates, are guinea pigs which develop fever within 10 days upon primary infection. Several passages, however, are necessary to produce a uniformly fatal disease.

Patient Management, Therapy, and Control

A virus-specific treatment does not exist. Supportive therapy should be directed towards maintaining effective blood volume and electrolyte balance. Shock, cerebral edema, renal failure, coagulation disorders, and secondary bacterial infection must be managed and may be life-saving for patients. Heparin treatment should only be considered when there is clear evidence of disseminated intravascular coagulopathy (DIC). Human interferon and human reconvalescence plasma have been used to treat patients in the past. Use of both therapies would be reasonable; however, there is a lack of experimental data showing efficacy. On the contrary, filoviruses are resistant to the antiviral effects of interferon, and interferon administration to monkeys has failed to increase survival rate or to reduce virus titer. Ribavirin does not affect filoviruses in vitro and thus is probably not of clinical value, in contrast to its efficacy against other viral hemorrhagic fevers. Isolation of patients is recommended, and protection of medical and nursing staff is required. This can be achieved by strict barrier nursing techniques and use when fea sible of HEPA-filtered respirators for protection against aerosols.

Even though filoviral hemorrhagic fever outbreaks have been rare and were mainly restricted to a small number of cases, vaccines would be of value for both medical personnel in Africa and for laboratory personnel. Cross protection among different Ebola subtypes in experimental animal systems has been reported suggesting a general value of vaccines. Inactivated vaccines have been developed by treatment with formalin or heat of cell culture-propagated Marburg and Ebola, subtypes Sudan and Zaire. Protection, however, has only been achieved by carefully balancing of the challenge dose and virulence. Because of the biohazardous nature of the agents, recombinant vaccines would be an attractive approach in the future. Immunization of monkeys with purified NP and GP has demonstrated the induction of the humoral and cellular immune responses and protection of animals against challenge with lethal doses.

Monkeys caught in the wild are an important source for the introduction of filoviruses. Quarantine of imported non-human primates and professional handling of animals will help prevent introduction into humans.

References

1. Feldmann H, Klenk H-D, Sanchez A. Molecular biology and evolution of filoviruses. In Kaaden OR, Eichhorn W, Czerny CP, (eds.). Unconventional Agents and Unclassified Viruses, Arch. Virol. Supplementum. 1993;7:81. [PubMed: 8219816]
2. Feldmann H, Mühlberger E, Randolf A. et al. Marburg Virus, a filovirus: messenger RNAs, gene order, and regulatory elements of the replication cycle. Virus Res. 1992;24:1. [PubMed: 1626422]
3. Feldmann H, Sanchez A, Klenk H-D: Filoviruses. In Topley & Wilson's: Microbiology and Microbial Infections, Edwald Arnold, London, in press .
4. Martini GA, Siegert R (eds.): Marburg virus disease. New York, Springer Verlag, 1971 .
5. Pattyn SR (ed.): Ebola virus haemorrhagic fever. Amsterdam, Elsevier/North- Holland. 1978 .
6. Peters CJ, Johnson ED, McKee KT: Filoviruses and management of viral hemorrhagic fevers. In Belshe RB (ed.): Textbook of Human Virology, Mosby Year Book, St. Louis; 159-175, 1991 .
7. Peters CJ, Sanchez A, Feldmann H, Rollin PE. et al. Filoviruses as emerging pathogens. Seminars in Virology. 1994;5:147.
8. Peters CJ, Sanchez A, Rollin, PE, et al: Filoviridae: Marburg and Ebola Viruses. In Fields BN, Knipe DM, (eds.): Virology, Raven Press, New York, in press.
9. Sanchez A, Kiley MP, Holloway BP, Auperin DD. Sequence analysis of the Ebola virus genome: organization, genetic elements, and comparison with the genome of Marburg virus. Virus Res. 1993;29:215. [PubMed: 8237108]