In the cytoplasm the situation is similar, though a certain number of features indicative of modified metabolism are present with unusual frequency in cancerous cells of virus-induced tumors and of human cancers. They are similar to those found in "classical" virus infections.

Although mitochondria may retain a perfectly normal aspect, they may also often exhibit pathologic features (34). "Cloudy swelling" is the most frequent lesion (figs. 8 and 12). Transformation into "microbodies" is also extremely common (35). This phenomenon is not necessarily indicative of chondrioma hyperactivity. The view that "microbodies" are precursors of adult mitochondria is no longer held by modern biochemists since the discovery of "lysosomes" (36*).

Other types of transformation also occur, some of which are undoubtedly of a degenerative nature (figs. 9, 10, and 11). To this category belong large bodies presenting the structure of lipides which have undergone the modification known as "myelin figure."

In "classical" virus infection, such bodies should be well known to electron microscopists. Their dimensions and density meet some of the criteria of an "inclusion body" as observed in the optical microscope, and errors have been made in interpreting them as viral inclusions. Yet they are not related to a viral process but most certainly represent a general cellular reaction to abnormal conditions.

Still another type of structure should be described here, because it is at the heart of our matter. In some virus-induced tumors and in avian myeloblastosis in particular, virus develops within formations called " gray bodies" by Bonar et al. (37). There is impressive evidence that gray bodies correspond to mitochondrial precursors (38).

In avian erythroblastosis (39), also, remnants of characteristic double membranes and cristae may be seen at the periphery of similar bodies, and they have been observed recently in Rous sarcoma cells grown in vitro (40) (see fig. 25). Undoubtedly, gray bodies play an important though not yet explained role in virus development. When they are found without viral particles, relationship to viral etiology is difficult to perceive but should be kept in mind.

The other classical constituents of the cytoplasm, ergastoplasm, Golgi apparatus, and endoplasmic reticulum, are not especially modified in virus-induced tumor cells or in human cancer.

As is now well established (41), the ergastoplasm comprises in its "organized form" three components: the membrane system, the granules attached to it, and the substance contained between the membranes. Cytoplasmic basophilia is linked to the granular component which represents the ribonucleoprotein as demonstrated by the brilliant work of Palade (42) and Siekevitz (6).

In many cancers cytoplasmic basophilia and ribonucleic acid (RNA) content increase. It is then not surprising to witness in some tumors an increase of ergastoplasm, especially in the form of scattered ribonucleoprotein granules. Yet no general rides are valid, since this increase may not occur. When it does, it is probably linked to rapid multiplication of a dedifferentiated cell in a state comparable to that of the embryonic cell (43). One morphologic fact is certain: A striking increase in ribonucleoprotein granules may sometimes be observed in the vicinity of virus particles or in cells where virus formation is occurring, at least in Rous sarcoma (figs. 13, 14, and 15). A similar observation has been made in the study of the Western strain of equine encephalomyelitis. (Morgan- personal communication.) A relationship between this increase and the development of viral particles is thus strongly suggested, which is in good accord with our knowledge of viral chemical composition.

The Golgi zone (44) deserves special attention because its hyperplasia in some cancers was noted a long time ago. As all cytologists know, it corresponds to the cellular center which becomes conspicuous in cultures of some virus-infected cells or in cancer cells grown in vitro (45, 46). Hyperplasia of the paranuclear zone is beautifully illustrated in Rous sarcoma cells grown in tissue culture (fig. 16) and was even considered by Tenenbaum and Doljanski in their fundamental work on the cytopath-ology of the Rous sarcoma cells (47) as the main change occurring in the cytoplasm.

There is no doubt that, like the giant mitochondrial bodies noted, such areas, because of their often well-demarcated limits, have been believed to be "inclusion bodies." Yet there is no difference between the ultrastructure of this area and that of a normal Golgi zone in a cell in which it is well developed, such as the plasma cell (archoplasm). The "inclusion body "-like effect and demarcation from the rest of the cytoplasm are due to dense packing of the Golgi elements (fig. 17) and often to distention of the Golgi vacuoles by a substance of unknown significance. It is of interest to note, particularly in the Rous sarcoma in which this hypertophy is a marked feature, that no relationship between it and virus formation has been observed.

This consideration is important because a relationship between Golgi apparatus and virus particles has been stressed in some strains of mouse mammary cancers (48). There is no doubt in this case that the intracellular particles termed type A by Bernhard (7) are concentrated in the Golgi region where they may accumulate in such numbers that an "inclusion body" is visible in the light microscope (49). But aside from the presence of particles at this site, the participation of the Golgi apparatus in the formation of such particles is not clear. Furthermore, the meaning of the intracellular type of particle in mouse mammary cancer is subject to much discussion, since it has been shown that this was not the infective particle (50-62) and that the latter could develop in some mouse strains without the necessary intermediate occurrence of the A type of particle (50).

One should now discuss still another type of particulate formation, which is concerned with the complex of "particle-containing" vacuoles described as "compound vacuoles" by Low and Freeman (53). This structure is found in many types of normal cells, often in the Golgi region and, typically, vesicles are always present in the cytoplasm immediately adjacent to the vacuole membrane (fig. 18).

In mouse mammary cancer comparable particle-filled vacuoles are frequent among the type A particles. They are striking in K-virus-infected cells (54). Aggregations of comparable vacuoles have been described in Lucke's renal adenocarcinoma (26). In human breast cancer, attention has been drawn (82) to formations that may resemble this, though the absence of vesicles around the vacuolar membrane or even the disappearance of this membrane casts some doubt as to their similar nature (fig. 19). The regularity sometimes of the particles themselves has even led-on purely morphologic grounds-to discussion of their possible viral nature. Arguments in favor of their being a perfectly regular structure of a normal cell are equally strong. It is important, nevertheless, to be aware of them.