EBOLA AND MARBURG VIRUS MORPHOLOGY AND TAXONOMY
FREDERICK A. MURPHY (1), GUIDO VAN DER GROEN (2) SYLVIA G. WHITFIELD (1), JAMES V. LANGE (1)
1. Center for Disease Control, Atlanta, Georgia 30333, USA.
2. Prince Leopold Institute of Tropical Medicine, Antwerp, Belgium.
In 1967, in an era when it was felt with a degree of savoir-faire that every possible morphologic form of pathogenic viruses had already been visualized, the first electron microscopic observations of Marburg virus were absolutely hair-raising. Despite the temperate choice of terms used in the literature, the same sense of wonder was felt in each of the laboratories involved as they found the bizarre filamentous agent. Following the 1967 disease episode, Marburg virus morphology and morphogenesis were studied in detail in several countries by negative contrast and thin-section electron microscopy after propagation in cell cultures, and in the organs and body fluids of humans, monkeys, and guinea pigs (1-7).
Electron microscopic studies following the 1975 Marburg virus infections in South Africa were limited to cell culture preparations and liver specimens from the single fatal case (CDC and South African Institute for Medical Research, Johannesburg). Ebola virus morphology studies have been concentrated on human liver tissue (8) (CDC and Prince Leopold Institute of Tropical Medicine, Antwerp), guinea pig tissues (9) (Microbiological Research Establishment, Porton Down) and cell culture preparations (8,9,10) (all laboratories). After careful comparison of morphologic and morphogenetic details of the viruses isolated in the three disease episodes, it seems clear that they are indistinguishable and only separable by the antigenic properties described elsewhere in this Colloquium. Minor differences in structure which have been noted can be attributed to variations in 1) the condition of virus preparations, 2) the nature Of background material, 3) the effects of fixatives used for biohazard containment, and 4) the methods used for electron microscopic preparation in various laboratories. Therefore, the following morphologic and morphogenetic descriptions are meant to be generally representative and illustrations of details are taken from studies with viruses from both Marburg episodes and with Ebola virus from Zaire and the Sudan.
In their native state, Marburg and Ebola virus particles are pleomorphic, appearing in negative contrast preparations as either long filamentous forms or "U"-shaped, "6"-shaped, or circular forms (Figures 1,2,3). The virus particles are composed of an internal helical structure which is presumed to be the nucleocapsid, a unit-membrane envelope, and a surface projection layer (Fig.4), Particles are approximately 80nm in diameter and extremely variable in length. Lengths of Marburg virus particles measured at CDC in 1967 ranged randomly between 130 and 2600nm. Peters and his colleagues 4, however, found that a median length of Marburg virus particles was 665nm and that longer particles appeared to occur in multiples of this length. In 1967 Marburg virus particles as long as 8,000nm were found, and this year Ebola virus particles up to 14,000nm were measured. The shorter and circular-shaped or '6"-shaped particles predominate in tissues and body fluids of man and experimentally infected animals, and in cell cultures there are more anomalous particles containing bizarre windings of nucleocapsid strands, more branched particles, and more very long particles (Figures 5,6). In a serial harvest series of Ebola virusinfected Vero cell culture supernatant fluids, the proportion of the various particle shapes and lengths did not vary between 1 and 4 days postinfection.
The virus particle surface projection layer is composed of spikes about 10nm long; the character of the spikes is significantly affected by fixation and staining conditions. The envelope layer is clearly formed of host cell membrane and along the length of the filamentous particles this envelope is rather closely apposed to the nucleocapsid. The envelope is distended over the wound nucleocapsid of "6"-shaped and circular-shaped particles, and over terminal windings of nucleocapsid which are common at one end of long particles. The envelope layer is often incomplete over the terminal bleb of long particles; this probably results from avulsion of virus particles rather than neat pinching off at the end of the budding process. Envelope blebbing can occur elsewhere along the length of the filamentous particles, probably as a result of osmotic shock during preparation for electron microscopy. The nucleocapsid structure is complex; it consists of a dark (stain penetrated) central axis, 20-30nm in diameter, surrounded by a light helically wound capsid with a diameter of 40-50nm and a cross-striation interval of 5nm, The outer edge of the nucleocapsid often appears fuzzy and thick so that in intact particles the separation between the main nucleocapsid layer and the envelope is often indistinct. Peters and his colleagues 4 defined another "intermediate layer" beneath the envelope. Particles occur which have a uniform diameter of 80nm and surface projections but no internal structure; in other cases long particles may contain segments of the internal structure and often the envelope diameter is reduced beyond the ends of the internal structure or between segments.
Variations in negative contrast methods can affect the appearance of virus particles and this can add some confusion to virus identification. Virus particles from fresh unfixed cell culture supernatant preparations exposed to negative contrast media for short times at near neutral PH (as with droplet methods with phosphotungstate or silicotungstate stains) are usually unpenetrated and appear as smooth membrane-bound forms with surface projections. In some cases fixation of fresh virus (gluteraldehyde or formaldehyde) seems to stabilize the viral envelope so that the same forms are obtained. Caution must be exercised in distinguishing these unpenetrated particles from the normal microvillous projections of plasma membranes common in many cell cultures. When the negative contrast method favors stain penetration, the resolution of the unique nucleocapsid structure makes identification of Marburg/Ebola viruses unquestionable. In a recent study at CDC, Ebola virus in Vero cell culture supernatant fluid was fixed with 0.5% gluteraldehyde for 1 hour and subjected to several negative contrast techniques in order to determine which methods yielded the most clearly identifiable particles. We found that all methods in which stain exposure was extended, PH was lowered (as with uranyl acetate staining - PH 4.5), or there was exposure to a lipid solvent (as with pseudoreplication from an agar surface with a formvar film cast in ethylene dichloride), nucleocapsids were well resolved and identification was certain. The glutaraldehyde fixation slightly obscured cross-striation detail of nucleocapsids; in comparison, the formaldehyde fixation schemes used in 1967 seem to have left fine structural details intact. In any case, there is no reason for carrying out negative staining on viable organisms. Finally, in this same Ebola virus study some further insight was gained into the practical sensitivity of negative contrast electron microscopy in a diagnostic setting. When we diluted daily harvests of Vero cell culture supernatant fluids fourfold with gluteraldehyde (final concentration 0.5") in water, and then ultracentrifuged these preparations at 25,000 PRM for one hour, virus particles were easily found even in 24 hour postinfection specimens, and extraordinary numbers of particles were present in all subsequent specimens. In this case the virus inoculum had been passaged in Vero cells previously, so high MOIs and rapid growth rates were obtained, but in a diagnostic setting a search for Particles in inoculated Vero cell cultures should start at the same time immunofluorescence testing begins -- that is at 24 to 48 hours.
Virus Morphogenesis and Cytopathology
The ultrastructural events involved in Marburg and Ebola virus morphogenesis and the associated changes in host cells have been examined in human liver tissue, in guinea pig organs, in monkey organs (so far only Marburg virus), and in cell cultures 3,4,6,7,8,9. in each case the viruses have been shown to be constructed from preformed nucleocapsids, which develop within cytoplasm, and envelopes which are added via budding through plasma membranes (Figure 7). Surface projections are inserted in the viral envelope at the bud site. The budding process is not seen often in human or experimental animal tissues, partly because of the convolutions of host cell membranes in relation to plane of section and the asynchrony of infection. Budding is seen commonly in infected cell cultures; the apparently regulated formation of simple filamentous particles contrasts with the violent plasma membrane deformations involved in envelopment of pleomorphic particles. The avulsion of plasma membrane at the termination of the budding process is not like the usual pinching off of other enveloped viruses.
Nucleocapsid formation and accumulation in cytoplasm leads to massive inclusion bodies, both in vivo and in cell cultures. Early condensation of nucleocapsids occurs in amorphous or granular matrices, and most young inclusions consist of variable proportions of matrix (? constituent ribonucleoprotein and nucleic acid) and filamentous nucleocapsids (Figure 8). Judging from the daily harvest series of Ebola virus-infected Vero cells, these inclusions seem more likely to be the "factories" for nucleocapsids going on to form virus particles rather than accumulations of leftover constituents; that is, these inclusions appear early -- at the time when virus particle formation is starting. In Ebola virus-infected Vero cells the median length of nucleocapsids (with entire lengths in plane of section) associated with these early inclusions was 750nm. Variations in inclusion body structure have been seen commonly, but there is still no understanding of their nature. For example, some early inclusions consist of masses of 50 to 60nm spheres (Figure 9). Some late inclusions consist of crystalline arrays of nucleocapsid cylinders, and others consist of amorphous dense material (Figures 10,11). In one case (Marburg virus of 1975) infected cells accumulated extremely dense flat sheets of unrecognizable material. Perhaps all the late inclusions represent anomalous variations in the condensation of viral constituents, but adequate studies have not been done.
The cytopathic change in tissues and in cultured cells infected with Marburg or Ebola viruses is striking, especially because cells are not arrested in late stages of the common terminal pathway of cytonecrosis. The progression of cel destruction has been shown to be similar in cell culture (for example, in an Ebola virus-infected Vero cell harvest series) and in vivo (for example, in monkey liver; 7). Infection processes, of course, form a continuum, but for convenience 4 stages can be distinguished. In the first stage of infection, virus particle budding and inclusion body formation occurs without apparent effect upon the morphologic appearance of cell organelles. Large amounts of virus are formed in this stage (Figures 7,10). In the second stage of infectio virus particle budding and inclusion body formation continue in cells with dilated endoplasmic reticulum, beginning intracytoplasmic vesiculation, and mitochondrial damage (loss of cristae and swelling) (Figures 12,13,14). In the third stage of infection, virus production ceases as the breakdown of cytoplas mic organelles and associated endophagocytosis (lysosomal response) continues. In this stage there is a change in cytoplasmic and nuclear density -- the alternate pathways to cell death appear as condensation or rarifaction. In the fourth stage of infection, destruction of membrane systems, including the nuclear and plasma membranes, reduces cells to debris (Figures 15,16). The progression of infection to this last stage is extreme with the two viruses. The degree of dissolution of infected cells in the liver of monkeys inoculated wit Marburg virus is so pervasive that focal sites contain only the vestiges of cellular structure. In Ebola virus-infected Vero cells the progression of infection in individual cells is rapid; between 48 and 96 hours postinfection, when more and more cells are still becoming infected; there is no shift in the proportion of cells observed in late versus ealy stages of infection, and ther is no build-up of intact dead cells as would be the case with most other viral infections.
Overall, the morphologically visible events in Marburg and Ebola virus infections at the cellular level seem as devastating as the effects of infection at the clinical level.
At the time of the initial characterizations of Marburg virus, morphologic similarities with rabies, vesicular stomatitis and other rhabdoviruses were noted(4,5,6). In the years since then, the isolation of many more viruses with physicochemical and morphologic characteristics very similar to the prototype rhabdoviruses has led to a more precise definition of the taxon, the Rhabdoviridae family. At the same time the differences in construction of the rhabdoviruses and Marburg and Ebola viruses have become more widely appreciated
Physicochemical characterization data must be comprehensive if they are to be of value for taxonomic consideration, but most properties of Marburg and Ebola virus remain untested, suggestive, or unconfirmed. Our lack of progress has been due entirely to the biohazards involved in viral biochemistry laboratories.
If the available physicochemical data were examined at this time in an objective context by the International Committee on Taxonomy of Viruses (ICTV), it is likely that all taxonomic considerations would be deferred. However, we have an immediate need for a nomenclature which will avoid the perpetuation of competing terms -- this need has been made quite clear in the past year. Although the names of the two viruses, Marburg and Ebola, are settled into general use, nomenclature for the "group" is confusing. For example, use of the term "African hemorrhagic fever" to describe the disease syndrome caused by the two viruses may have clinical value, but the term does not sit well as a virus genus or family designation. Alternately, the term "tuburnavirus", as advanced by Simpson and Zuckerman(11) as a taxonomic designation, has not been submitted to the ICTV for consideration by the virology community. In response to this situation the ICTV has asked its Vertebrate Virus Subcommittee (F.A. Murphy, chairman) to undertake a study of the matter. Toward this end, the Rhabdovirus Study Group (F. Brown, Animal Virus Research Institute, Pirbright, chairman) is solic iting, from working virologists, virus characterization data and opinions regarding nomenclature. Dr. Brown would welcome all experimental data and personal opinions. If the data warrant construction of a new taxon this will be done, but in any event the matter of nomenclature will be settled democratically and as soon as possible.
This is not a comprehensive listing of the Marburg virus pathology literature. Extended bibliographies are found in references 4,5,6 and 7 and in the book Marburg Virus Disease (edited by Martini, G.A., and Siegert, R.) Springer Verlag, New York 1971.
1. Siegert, R., Shu, H,-L., Slenczkj, W., Peters, D., Müller, G.(1968) Deutsche Medisinische Wochenschrift, 93: 2163-2165.
2. Peters, D., Müller, G. (1966) Deutsche Arzteblatt, 65: 1827-1834.
3. Kissling, R.F., Robinson, P.Q., Murphy, F.A., Whitfield, S.G. (1968) Science, 160: 888-890.
4. Peters, D., Müller, G., Slenczka, W. (1971) in Marburg Virus Disease (edited by Martini, G.A., and Siegert, R.) Springer-Verlag, New York, pp. 68-83.
5. Almeida, JED., Waterman, ALP., Simpson, D.I.H, (1971) in Marburg Virus Disease (edited by Martini, G.A., and, Siegert, R,) Springer Verlag, New York, pp. 84-97.
6. Kissling, R.E., Murphy, F.A., Henderson, B.E. (1970) Annals of New York Academy of Sciences, 174: 932-945.
7. Murphy, F.A., Simpson, D.I.H., Whitfield, S.G., Zlotnik, I.,Carter, G.B.(1971) Laboratory Investigation, 24: 279-291.
8. Johnson, K.M., Lange, J.V., Webb, P.A., Murphy, F.A. (1977) The Lancet, 1: 569-571.
9. Bowen, E.T.W., Lloyd, G., Harris, W.J., Platt, G.S., Baskerville, A.,Vella, E.E. (1977) The Lancet, 1: 571-573.
10. Pattyn, S., Van der Groen, G., Jacob, W., Piot, P., Courteille, G. (1977) The Lancet, 1: 573-574.
11. Simpson, D.I.H., Zuckerman, A.J. (1977)Nature, 266: 217-218.
Fig. 1. Ebola virus. Unfixed diagnostic specimen from first Vero cell passage, showing elongated particle shape, but no internal tail. Sodium phosphotungstate; X 90,000.
Fig. 2. Ebola virus. Glutaraldehyde fixed particle from Vero cell culture supernatant; particles up to 14,000nm long were found in such preparations. Uranyl acetate; X 28,000.
Fig. 3. Ebola virus. Glutaraldehyde fixed particles from Vero cell culture supernatant; typical "6-shaped" configuration such as was common in blood of Marburg virus infected animals. Uranyl acetate; X 66,000.
Fig. 4. Ebola virus. Unfixed diagnostic specimen from first Vero cell passage, showing cross-striations of internal helical structure and surrounding envelope layer. Sodium phosphotungstate; X 156,000.
Fig. 5. Ebola virus. Glutaraldehyde fixed particle from Vero cell culture supernatant; there was more evidence of branching in such preparations than previously found. Uranyl acetate; X 43,000.
Fig. 6. Ebola virus. Glutaraldehyde fixed particle from Vero cell culture supernatant; elaborate windings of the internal structure occurred within the envelope blebs of many particles. Sodium phosphotungstate; X 39,000.
Fig. 7. Ebola virus. Vero cell culture, day 2; virus particles budding from plasma membrane (arrows) with "nucleocapsids" in cytoplasm. Thin section; X 37,000.
Fig. 8. Ebola virus. Vero cell culture, day 2; inclusion body (arrows) consisting primarily of amorphous matrix within the cytoplasm of a cell in an early cytopathic state. Thin section; X 9,000.
Fig. 9. Marburg virus. Vero cell culture, day 3; uncommon configuration of intracytoplasmic inclusion body in which 50-60nm spheres occurat the edges of an amorphous matrix. Thin section; X 39,000.
Fig. 10. Ebola virus. Vero cell culture, day 2; this most typical inclusion body configuration, consisting of precise cylinders in an amorphous matrix, is present in an otherwise normal cell. Thin section; X 19,000.
Fig. 11. Ebola virus. Vero cell culture, day 4; high magnification of the cylindrical structures which form most inclusions and constitute the internal structure of virus particles. Thin section; X 789000.
Fig. 12. Ebola virus. Vero cell culture, day 3; starting cytopathology marked by mitochondrial swelling and destruction while virus particle production continues.Thin section; X 13,000.
Fig. 13. Ebola virus. Vero cell culture, day 4; early cytopathology with mitochondrial destruction associated with massive cytoplasmic replacement with viral material. Entire plasma membrane involved in virus budding. Thin section; X 16,000.
Fig. 14. Ebola virus. Vero cell culture, day 4; high magnification of the surface of cell illustrated in Figure 13, showing nascent budding along whole plasma membrane as supplied by massive cytoplasmic infection. Thin section; X 46,000.
Fig. 15. Ebola virus. Vero cell culture, day 2; late cytopathology with organelle destruction and destruction of plasma membrane (top). Thin section; X 37,000.
Fig. 16. Ebola virus. Vero cell culture, day 4; terminal cytopathology marked by nuclear and cytoplasmic rarifaction, organelle destruction and frank dissolution of the plasma membrane.Thin section; X 16,000.
S.R. Pattyn : what about the name Toroviruses that has been proposed some time ago ?
F.A. Murphy Dr.Almeida proposed that name in 1970 or so, and nothing never happened. The I.C.T.V. will never involve itself in the names of the viruses per se and it seems to me that the terms Marburg, Ebola are entrenched. We need a family or genus term, but it should come from people who are working in the field and the Taxonomy Committee is there only to see that democracy is respected. Each of the common virus names has come out a different way and any one who has got a good name should advance it.
G.A. Eddy : It is possibly a little early to name this group of viruses, if indeed it is a group, someone should first Zook at the virion polypeptides.
F.A. Murphy : In considering the rhabdovirus study group, I would guess that they wouldn't want to do very much in a formal taxonomic matter, until more is known about the proteins and the nucleic acid. They might be willing to echo a nomenclature which seems to be needed now.
P. Brès : I would discourage the use of the term African Haemorrhagic Fever which may be rather confusing because this would include yellow fever. For reasons of symmetry you would have to say American Haemorrhagic Fever which would include Argentinian and Bolivian H.F. as well. This has been discussed with some people and we thought that Ebolavirus Haemorrhagic Fever would better describe the syndrome.
J. Casals : Is there enough virus in the blood of a patient so that you could do the same procedure with blood serum, and have a diagnosis within six hours maybe ?
F.A. Murphy : It was done in Germany on Marburg virus. Another thing that is much quicker, the Liver itself of the case in South Africa, 1975, fluoresced so brightly and specifically that the results could have been available in hours.
T. Muyembe : Why is the name Ebola jirus and not Yambuku ? Ebola is a river and was not involved in the epidemics.
F.A. Murphy : That's a long story. The people who were there at the time, not any individual, had the privilege of naming the virus. That name came out of a lot of discussion and I think if any one doesn't like the name now it's almost too late to discuss it.