Infectious RNA is perhaps most readily tested in an in vitro system. Extracellular ribonuclease can be removed by washing; the conditions under which cells and RNA interact can be carefully controlled, and the results can readily be quantitated. In vivo demonstration of activity is not likely to be successful unless the target organ is readily accessible so that the nucleic acid has a chance of immediate contact and entry into the appropriate cells types. Infectivity of viral RNA has been demonstrated in the mouse only when the nucleic acid is injected directly into the brain or muscle. RNA from the virus of foot and mouth disease, for example, is inactive when given by intraperitoneal injection, but will produce an infection when given by the intramuscular or intracerebral route (11). We have found that the Ehrlich ascites tumor cannot be infected by the intraperitoneal injection of Mengo RNA into tumor-bearing mice. However, these cells can readily be infected by removing them from the animal, washing them several times in a balanced salt solution, and exposing them to the RNA in the absence of serum. It may be of significance with regard to these observations that, among mouse tissues, muscle, brain, and cells of the Ehrlich ascites tumor are exceptional in their low level of intracellular ribonuclease (12). It would be of interest to know the levels of intracellular ribonuclease in the various primary explanted tissue cultures used in this work. Preliminary estimations in our laboratory on some stable cell lines suggest that they, like all ascites tumors that we have examined, have very low levels of ribonuclease activity at physiological hydrogen ion concentrations.

A potentially active viral RNA must then overcome many obstacles before it can initiate an infection in an experimental animal. Degradation by extracellular-and perhaps intracellular-ribonuclease is an ever-present hazard, as is nonspecific immobilization by proteins. To this, one must add the inherent thermal instability of the RNA molecule. At 37° C, and in the presence of 0.001 m Versene, Mengo RNA has a half life of more than 90 minutes. In the absence of Versene, it has a half life of not more than 30 minutes.

These considerations concerning the in vivo labdity of viral RNA may also apply to DNA. Although unpurified transforming principle has been found to be stable for at least 24 hours in the mouse (13), deproteinized preparations, while still effective in vitro, will not transform in vivo.

Of the several reports concerning the infectivity of DNA extracted from virus-infected tissues, the most convincing is the recent one by DiMayorca et al. (14) These investigators used the duponal method (15) and the phenol method of Kirby (16) to deproteinize concentrates of tissue cultures infected with polyoma virus. The preparations produced cytopathogenic changes in primary cultures of mouse-embryo fibroblasts accompanied by the appearance of deoxyribonuclease-insensitive polyoma virus, identified by its ability to produce tumors in the hamster. The activity of the nucleic acid preparations was abolished by deoxyribonuclease, but was unaffected by ribonuclease.

While no other tumor virus has yielded an infectious nucleic acid, there seems no reason to believe that other viruses which have been studied should prove any exception to the basic notion that the viral genome is contained within the structure of the nucleic acid. Tumor viruses cannot be grouped together as DNA-containing particles. While the polyoma, Shope papdloma, and rabbit fibroma and myxomatosis viruses apparently belong to this group, the best preparations of the Rous sarcoma (17) and myeloblastosis (18) viruses have been found by chemical analysis to contain only RNA.

There appear to be many aspects in common between the biology of tumor and non-tumor viruses. The differences between them are mainly quantitative rather than qualitative (19). The major distinguishing feature between them-the ability to produce tumors in animals-reflects only one aspect of the virus-cell relationship. Thus the polyoma virus, which produces multiple tumors in rodents, kills mouse-embryo fibroblast cells in vitro with the release of large quantities of virus from the dying cells. This agent also induces in vitro a proliferation of fusiform cells in hamster-embryo cultures (20) during which production of the virus is poor. A similar relationship between cell proliferation and virus production is seen in the Shope papilloma virus. Proliferation of the basal layer of the rabbit epidermis is accompanied by low virus production, whereas increased virus production occurs in the nonproliferating, keratinizing layer (21).

In an analogy with the behavior of virulent as opposed to temperate phages, Dulbecco (19) has called the "cytocidal-high virus yield" relationship between virus and cell nonintegrative, and the "cell proliferative-low virus yield" relationship integrative. He points out that the important deviations from the temperate phage-bacterium relationship are: (1) Virus-induced tumors usually contain either virus-specific protein or intact virus particles, and (2) instead of the discontinuity which exists between the prophage and the vegetative state in lysogenic cells, there is an apparent continuity of the changes in the properties of virus-carrying cells.

The observation that different strains of Rous sarcoma virus can produce consistent, hereditable changes in the morphological appearance of chicken fibroblasts is important (22). In this case an RNA virus has converted its host cell with respect to at least three hereditable characteristics: (1) the ability to produce Rous sarcoma virus of the type that caused the infection, (2) the property of stability to subculture, and (3) a change in the morphological appearance of the cell-a change not determined exclusively by the 2 preceding characteristics, since cells possessing these by virtue of infection with different strains of the Rous virus may have different and equally unique appearances. This phenomenon of conversion is not the prerogative of tumor viruses.

Infection with cytocidal viruses may also result in conversion. As examples, one may cite the acquisition of hemagglutinin on the surface of influenza-infected cells (23), the production of interferon accompanying influenza-virus multiplication (24), the protein factor produced during adenovirus replication which induces cytopathic changes in HeLa cells (25, 26), and the virus-induced formation of a number of enzymes related to the special requirements of T2 phage DNA synthesis in Escherichia coli (27, 28).

Although conversion by non-tumor viruses is not heritable when the agents are cytocidal, in the sigma virus of Drosophila, which is responsible for CO2 sensitivity in this fly, the evidence for a hereditable phenotypic change in the organism seems indisputable (29). The evidence so far obtained with the Rous sarcoma virus suggests that the ability to produce virus is transmitted to the progeny of an infected cell as an intracellular event, the virus continuing to be produced at a slow rate in succeeding generations (SO). The moderate behavior of these viruses suggests a stable addition to the genetic information of the infected cell, in line with Luria's concept of viruses as infective genetic materials (31). In a sense then there has been an addition to the genotype of the cell.