Does the preceding information as to where and when the polyoma virus elicits a proliferative response give us material from which to construct a working hypothesis as to how it acts? Our knowledge is still so fragmentary as hardly to permit the question, much less the answer. The following comments are therefore entirely speculative:

We have seen that the range of species susceptibility to the agent, so far as proliferative response is concerned, is limited to 4 or possibly 5 rodents, each of which responds with a different pattern of tissue response. The kidney and subcutaneous connective tissues are the only sites that may be involved in all. This rather narrow range of species specificity is similar to that seen in general for other tumor viruses, in contrast to the much wider range that is shown by many "nontumor" viruses, e.g., influenza, rabies, and lymphocytic choriomeningitis (LCM) viruses.

The tissue-specificity range, though broad for the mouse, is nevertheless limited. With few exceptions, the tissues that most frequently develop tumors are examples of higher-degree differentiation far out on the ontogenetic tree, where versatility of response is limited, e.g., salivary and lacrimal glands, thymus, thyroid gland, hair follicles, mammary glands, and dental epithelium. Possibly a connection exists between the facts that the species range is wider for kidney tumors, and the morphogenetic response of metanephrogenic mesenchyme also has a low heterospecificity to induction by ureteral bud or spinal cord (73, 74)- Is there also a clue, hidden in the hydrocephalic response (66) to polyoma virus, relating the inductive capacity of spinal cord to susceptibility of some as yet undetermined part of the central nervous system to polyoma virus? Not much is known of the interspecific responsiveness of the epithelial rudiment of the submandibular gland, but it has been shown that the morphogenetic response of this epithelium is highly specific for capsular mesenchyme of the same organ (82).

If one can judge from tissue-culture observations, the proliferative type of response in the submandibular gland, at least, is correlated with the degree to which the epithelium of this organ has acquired its tissue-characteristic properties. When infected in the earliest stages of morphogenesis, this tissue responds predominantly by necrosis. A parallel may exist between these relationships in the mouse-polyoma virus system and the hemorrhagic versus proliferative response of the chicken-Rous virus system (83).

Since morphogenesis of the salivary-gland epithelium of the mouse is strictly dependent on homologous mesenchyme (82), and proliferative response in turn is correlated with the developmental stage of this epithelium, it would appear that the tumor-inducing capacity of polyoma virus is indirectly dependent on the establishment and perhaps the maintenance for some time of the specific interrelationship between the mesenchymal and epithelial components. In connection with this point it is pertinent that in tissue-culture systems using disaggregated embryonic mouse cells, no proliferative phenomenon has been reported, whereas in organ cultures it does occur, along with the cytolytic phenomenon. A disturbance of a specific mesenchymal-epithelial relationship could also account for the irregularity of results in transplantation of many of the polyoma-induced tumors (13, 45, 84).

The superiority of the newborn period for exhibition of proliferative response to polyoma virus would seem to be attributable to the ideal overlap of two critical biological phases in chronologic time. Thus, defectiveness of immune response allows infection and "boosting" of virus concentration at a period when the susceptible tissues have reached a potential for proliferative response. A continued proliferative response might not be possible, however, were it not for timely intervention of the immune reaction, which prevents complete dissolution of the intercellular dependency postulated for the proliferative response. Too late an immune response would lead to death of the animal, or the run ting phenomenon, the latter as a result of extensive loss of mesenchymal tissues, e.g., bone.

To look at the composite picture from this point of view, the salivary-gland tumors early in their development would be doubly dependent: (1) on establishment within the epithelial cells by virus, and (2) on the existence of the proper relationship between mesenchymal and epithelial components of the organ. In this tripartite relationship, the mesenchyme might act temporarily as the "balancing influence" which has been emphasized (85) as necessary to prevent virus from "gaining the upper hand" and to produce cytolysis of the host cell. Thus there would exist within the epithelial cell a competition between virus and cell for a complex material supplied by the mesenchyme and presumably identical with factors normally controlling morphogenesis and/or differentiation. The morphogenetic factor supplied by the mesenchymal cell could be analogous to the specific repressor substance responsible for maintenance of phage in a prophage state in bacterial systems (86). In specialized multicellular organisms, it would be logical to expect that the ability to produce or induce the production of repressor substance should reside in cells other than those carrying the virus, or normally responding to the repressor substance in morphogenetic or differentiative phenomena. Upon separation of epithelium from specific mesenchyme, the virus would multiply and/or become complete, and destroy the host cell. Some other virus-tumor systems might also fit this concept. For example, it is in accord with the observation that Shope papilloma virus is abundantly present in the dying cells of the outer layers of the epithelium, whereas it is either not present in complete form or is present in undetectable quantities in the basal, proliferating cell layers (87).

The recent report of Hearn (88) is relevant to this discussion. He found that, in tissue cultures chronically carrying Venezuelan equine encephalomyelitis, removal of glutamine resulted in the prompt appearance of a cytopathic effect, and simultaneously with this a great increase of virus production. In this system the virus-suppressing role of a second cell type would have to be filled by a relatively simple nutrient, glutamine, or more likely by a complex material produced by the infected cell itself, utilizing glutamine. An analogy was drawn by Hearn between this system and lysogenic phage systems.

To summarize: Differences in sensitivity and pattern of response to polyoma virus by different species, strains, tissues, and different cells within these tissues at various stages of biologic development have been reviewed in an attempt to see some order in a picture that at first appears chaotic.

Development of progressively growing tumors in the intact animal is limited, in present knowledge, to 4 rodents: mouse, rat, hamster, and Mastomys. Each of these species has a distinct pattern of tumor response, with Mastomys developing some, but not all, tumors characteristic of both hamster and rat. The apparent gap between the mouse and the other 3 rodents with respect to failure of the latter to respond with epithelial tumors has been partly bridged by Rabson's observation of epithelial lung tumors in hamsters receiving large doses of virus intra-tracheally. A common theme in the response of these 4 animals is the occurrence of connective-tissue tumors of the kidney and subcutaneous tissues in all. Whether the tumors at these sites arise from the same cell type in each animal is unknown. As with other viruses, it appears that species barriers, at least concerning tissue-response differences, may be partially broken down by dosage and route-of-administration factors.