Besides being of fundamental importance, the question of tumor virus constitution raises a number of related issues each of which is highly-interesting in its own right. These fall into two main groups, one, perhaps the more pedestrian, being concerned with problems of methodology and technique, and the other with basic biological concepts and the exciting speculations to which, in our present state of knowledge, their consideration leads.

The practical problems involved are not peculiar to the study of tumor viruses but are those which apply throughout the virus field and, in that context, they are both familiar and obvious. Nevertheless it is worth recalling them at the present time for two reasons: (1) Recent discoveries (1-6) have led to an enormous expansion in the study of tumor viruses and to the recruitment of many new workers, and (2) we are clearly at the start of an era in which the search for tumor agents will be extended to human neoplasia (6, 7). As this extension goes forward, experimental work cannot be too carefully controlled or too cautious, for one will be dealing with an emotionally explosive and sensational subject.

The actual nature of the practical considerations at issue is, however, straightforward; there is the problem of virus identification, the problem of virus isolation, and, finally, the problem of determining virus composition.

The present communication is concerned mainly with virus composition, but, before one can investigate this in a given virus, the agent itself must be identified. The recognition of so-called "virus-like" particles with the electron microscope is no more than a preliminary step, for this instrument alone can never serve to identify a virus, distinguish it from another morphologically similar virus, or establish its activities. When particles are found in association with a particular neoplastic condition, they must be linked to some sort of biological activity before any conclusions can be drawn as to their nature. Evidence is constantly accumulating which indicates that a wide variety of viruses can be and are present in normal cells (8-10), thus suggesting that certain tissues support a normal viral flora analagous to the bacterial flora of healthy individuals. This being so it is more than ever necessary that if virus-like particles are detected in tumors, they should be shown to be there in an active capacity and responsible for the disease process, rather than as casual loafers merely looking on. To this end, morphological investigations must be integrated with biological tests; if this is done the contribution of morphology in this field can be vital, whereas, alone, its usefulness is restricted and may even be seriously misleading.

The second practical question requiring consideration is that of virus isolation. Exact information on the composition of a virus can be reliably obtained by chemical analysis, but for this complete purification is a prerequisite which in itself can be troublesome to assess and extremely difficult or even impossible to attain. On the other hand, information on virus composition can also be obtained, and this where only partially purified material is available, if cytochemical digestion methods are used and their effects on individual virus particles observed with the electron microscope. But whichever of these approaches is used, the necessity for isolating a given virus remains, so that the requirement regarding purification for either technique is merely one of degree. In addition, experiments of quite a different kind can also throw light on viral composition. Where protein-free extracts of viral nucleic acid can be shown to possess infectivity, the type of nucleic acid involved may be demonstrated by abolishing the infectivity with a known specific nuclease.

Turning to the actual determination of virus constitution, relatively little has been achieved so far with true tumor agents. Recently, however, the virus of avian myeloblastosis has been found to occur in enormous quantities in the blood of infected chickens and has been successfully isolated from this source by differential centrifugation Bonar and Beard (12) have applied microchemical methods of analysis to material obtained in this way and have shown that the virus contains nucleic acid of the ribose type and that in other respects its chemical constitution is comparable to that of the generality of infectious viruses.

For our own work with the Rous virus we have preferred to rely on thin-sectioning techniques for electron microscopy, coupled with specific enzymic digestions. As has already been pointed out, this approach can be used without complete purification of the virus and in addition it has a further advantage. Chemical analysis can show of what a virus particle consists as a whole, but it cannot give information about the composition of morphological structures in the particle. This type of knowledge can only be obtained by examining these structures in situ, and combined electron microscopy and digestion methods provide a way of doing so.

The starting material for the experiments consisted of suspensions of Rous virus prepared from egg-grown nodules of Rous tumors by treatment with a fluorocarbon; the treatment freed the suspensions of almost all formed host substances, but certain other host-cell contaminants remained (13). However, when the suspensions were subjected to high speed centrifugation, pellets resulted in which characteristic areas were composed almost entirely of uniform particles with which tumor-producing activity could be shown to be associated (13).

Virus-containing fragments of such Rous pellets were fixed with potassium permanganate (14) and prepared directly for thin-sectioning for electron microscopy. When examined in this way the virus was found to be a spherical particle about 75 ray. in diameter, with a central dense nucleoid surrounded by a less dense outer zone or viroplasm, the whole being limited externally by a double membrane (fig. 1).

Besides being an excellent fixative for the electron microscopy of viruses, potassium permanganate has the further important attribute, unlike the more commonly used osmium fixative (15), of allowing subsequent digestion of nucleic acids by appropriate specific nucleases (16). In further experiments virus-containing samples were treated with ribonuclease; they were first fixed, as before, with potassium permanganate and then incubated for 2 hours at 37° C. in 30 percent alcohol containing 0.1 percent by weight of thrice-crystallized, protease-free ribonuclease. At the same time control samples were incubated in alcohol without enzyme. All samples were then dehydrated, embedded, and sectioned for electron microscopy. The enzyme digestions were applied in an alcoholic medium, in order to adhere closely to the series of treatments recommended by Luft (14) for the best possible morphological preservation of specimens. It was known, of course, from preliminary experiments (16) that both types of nuclease were fully active under such conditions.

Examination showed that the incubation procedure made the particles appear ragged and extracted as compared to samples fixed and prepared directly (fig. 1). Nevertheless, in the control preparations, which were not exposed to ribonuclease (fig. 2), the nucleoids and outer-Umiting membranes of the particles remained intact. In marked contrast, the nucleoids were digested from the particles which had been subjected to the ribonuclease (fig. 3), and these particles were also slightly swollen as if some central bracing structure had been removed. Thus, it can be concluded that the Rous virus contains a substantial amount of ribonucleic acid and the methods used have demonstrated that the site of localization of this material is the nucleoid.