The endoplasmic reticulum, as might be expected from its importance in nutrition, conduction, and membrane synthesis within the cell, does not seem to play a direct role in tumor virus replication transmission and extrusion. The variations in endoplasmic reticulum observed in infected, transforming, and tumor cells seem to be more related to metabolic variations and membrane synthesis than to virus production. In transforming human amnion cells, the endoplasmic reticulum changes from smooth-profiled to rough-profiled vesicles and tubules during growth of the cell (12). In FL cells in horse serum, as also in certain tumor cells, fenestrated lamellae of the reticulum have been observed (18). High magnification shows the dense regions of this structure to be intermembrane "desmosomes," hence normal components. In nonviral tumors, e.g., bronchogenic carcinoma (22,28) and squamous-cell carcinoma (24), the endoplasmic reticulum may enter into the formation of cytoplasmic filaments or myelinoid forms. However, such forms occur in some normal cells and hence cannot be directly related to cell abnormality. Perhaps the closest relation of the endoplasmic reticulum to a viral form is that seen in pobovirus-infected cells, in which the polio particles in the cytoplasm appear to be laid down in patterns determined by the ergasto-plasmic membranes (21).

The involvement of the Golgi complex in infected and in tumor cells is consistent with its normal function in any cell. The Golgi complex comprises dictyosomal lamellae, macro- and microvesicles, and micro-vesicular bodies. The dictyosome and macrovesicles appear to play a role in normal feed-back mechanisms within the cell. The microvesicles and microvesicular bodies apparently are more concerned with secretion and elimination. Microvesicles are most commonly seen in those cells which normally are secretory, e.g., pituitary, pancreas, hypodermal gland cells, and the like. Microvesicular bodies most commonly appear during hyperactive stages of cell growth or metabolism, e.g., during the early stages of tissue-culture growth, during chloride secretion of gastric mucosa cells, and during virus release from infected tissue-culture cells. In its function as an excretory organelle, then, it is not surprising that the Golgi complex should contribute to the release of viruses from the cell, and perhaps actively contribute membrane material, as in the formation of herpes (14, 15) and mammary-tumor viruses (25).

The principal cytoplasmic organelle involved in parotid tumor cells and leukemic thymus, as in myeloblastosis and swine pox-infected cells, is the mitochondrion. In normal and transforming cells, as in most virus-infected tissue-culture cells, the number, size, and shape of mitochondria appear to vary directly with metabolic stage. In the polyoma- and leukemia-infected cells the mitochondria also play a role in the formation of inclusion bodies (S, 26). I would like at this point to redefine "inclusion body" on the basis of the evidence to follow. Inclusion bodies as defined by light microscopy are identified on the basis of their size and stainability. In electron microscopy they have been variously defined but probably most often given as virus-containing vacuoles of the endoplasmic reticulum or Golgi complex, or as virus-containing, dense structures originating from mitochondria. The present evidence supports the latter view, and suggests that not only virus-containing structures, but also the previously described "gray bodies," "dense bodies," and other similar structures found in virus-tumor cells should be included within the category of inclusion bodies. In osmium, and possibly chromium, stained leukemic thymus, and polyomatous parotid cells, it is possible to identify not only the definitive virus-containing inclusion body, but also the transforming mitochondrion and early stage inclusion body, even though no particles are visible within the matrix.

The apparent sequence of events in the formation of the inclusion body in the parotid tumor of the polyoma-infected mouse is as follows: The normal mitochondria are small, roughly cylindrical, possess few cristae, and little matrix (fig. 8). Some show mitochondrial granules; others are smaller, possess fewer cristae, and appear to be more spherical in outline. Modified, but still clearly recognizable as mitochondria, are structures with dense double outer membranes, vestiges of cristae, but with densely filamentous matrix. Large structures in such an area appear to be formed by the aggregation of these latter forms. In some of these large forms there may still be seen vestiges of outer mitochondrial membranes and remnants of cristae. An apparent later stage is represented by similar structures showing no membranes within, but only a finely granular to filamentous dense matrix. Within such forms dense spots are sometimes observable, usually at the periphery (fig. 8). In others, identifiable single virus particles, or even crystalline arrays of virus particles, may appear (fig. 10).

In the leukemic mouse thymus the inclusion bodies are apparently also derived from modified mitochondria, albeit either in two different manners or representative of two types of body. One type of formation is similar to that seen in polyoma-infected cells. The second type shows characteristic fractionation of the mitochondrial cristae with subsequent symmetrical arrangement of the subunits, and a secondary transformation of these into filaments possessing an extremely fine axial period (fig. 9). In all definitive inclusion bodies there have been observed orderly arrays of dense, fine, periodic filaments, and aggregates of tiny dense spots apparently representative of viral particles or precusory viral material. The structure and density of the filaments are comparable to those of the filaments seen within the nucleoid of the mature leukemia virus particle.

The significance of the viral material, or crystalline arrays of particles within the cytoplasmic inclusion body, is still uncertain. It is even possible that the crystals are of nonviral protein. However, on the evidence given, it is most logical that the formation of viral particles and their crystallization in the inclusion body are merely due to the assumption of minimum form under conditions of high concentration or supersatura-tion. The transmission of the virus from the inclusion body to the cell border and its subsequent release become the next questions. It is doubtful that, in the Gross leukemia agent and polyoma, the particles observed in the inclusion bodies actually migrate as such throughout the cytoplasm. It is more suggestive that the elementary units of the component filaments are the migrating material, that these units recombine in whatever locale be best suited, and that the form so assumed would be the minimum possible for aggregated twisted filaments.