The fine structure of the different elements of the normal cell need not be detailed here, since numerous and excellent reviews exist on the subject (1, 2). Some characteristics, which are particularly striking in virus-induced tumor cells, will be discussed.

Nucleus

We would like to draw attention to some features of the nucleus of virus-induced tumor cells that may be related to a viral etiology though unspecific in themselves.

Among these, coarseness of chromatin networks associated with patchy destruction of nucleoplasm is extremely frequent. Such features are well illustrated by comparison of the relatively well-preserved nucleus of a human breast cell, figure 1, with the nucleus of a Rous sarcoma cell shown in figure 2. The possibly important meaning of such aspects of the nucleus is difficult to appreciate, as far as virus involvement is concerned, because this picture is also that of a badly preserved nucleus.

The recent observations in the study of the Shope papilloma of the rabbit (IS) and of the mouse polyoma (14-17) must, nevertheless, lead us to revise our former attitude toward such images.

Nuclei in thin sections of rabbit papilloma cells appeared at first sight to be poorly preserved. The nuclear membrane was thickened with remnants of chromatin appended to it, while most of the nucleoplasm was often a clear, empty space. High resolution pictures, however, then showed that some of these remnant threads of chromatin constituted the same particles that had been identified in pure preparations of the virus (18). When these small amounts of virus were not there, as in most nuclei, the specificity of the destructive lesion could not be established. Yet a relation to the development of viral particles is almost certain.

The same is valid for polyoma. Nuclei of renal tumors of mice appear damaged to the electron microscopist who is looking for undistorted tissues, yet most of the time no virus is found in these tumors (19). It is possible, though, that this appearance of the nuclei is linked with the disease as suggested by the nuclei in cells infected in vitro.

Another structural characteristic that raises the same problems is the peculiar distribution of chromatin along the nuclear membrane known as margination of chromatin, an arrangement often found in tumors.

It is difficult to believe that this phenomenon, as viewed in the electron microscope, is a specific indication of viral involvement, because it is always accompanied by lytic changes comparable with those due to poor preservation; yet it is a classic and conspicuous lesion in infectious viral diseases, even in well-preserved, osmium-fixed nuclei (20, 21). Its value as an indication of direct viral involvement cannot be doubted in this example, since it is accompanied by development of clear areas in the center of the nucleus in which the virus particles will later appear.

An unusual feature sometimes present in the nuclei of tumor cells, virus-induced and otherwise, is also noteworthy, though its significance is unknown. It consists of areas of peculiar granules larger in size than the normal nuclear granular constituents (figs. 2 and 3). They are present sometimes in Rous sarcomas and have been well illustrated in a fowl tumor induced by 1,2,5,6-dibenzanthracene (dibenz[a,A]anthracene) (22). A similar observation has also been made in a mouse lymphoma (23) and in human breast cancer (personal observation). The appearance of such particulate elements in the electron microscope should be known to the morphologist. They are in the range of size of the smaller viruses, and since no membrane has been detected up to now in some types of polyoma virus particles, it is not completely irrelevant to consider the possibility of confusion, especially where virus particles are few in number. Their resemblance, also, to granules found in the nuclei of cells infected with herpes B virus, which have been described as possible "templates" for the formation of that agent, should be considered (21).

As to the nucleolus, its fine structure which the electron microscope has revealed (24, 25) is not modified in cancer cells of human origin or in those of virus-induced animal tumors. But hyperplasia brings out some of its features in a striking manner (figs. 4 to 7). This is especially true when the cancer cells are grown in tissue culture, and the thinness and good preservation of the specimens allow a true unrolling of the nucleolar ball. The "nucleolonema" then becomes conspicuous and condensation of some of its segments is revealed, while the "pars amorpha" may accumulate in small pools as if microsecretion had occured (fig. 4).

With the exception of the granules illustrated in figures 2 and 3, all these characteristics of the nucleus-coarseness of chromatin network, marginatum of chromatin, and hypertrophy of nucleoli-have their counterpart in the optical microscope. Much more impressive, therefore, is the following trait.

Recent studies have brought out one striking character of nucleoli and nucleoplasm of virus-induced tumor cells. It consists of the occurrence of very dense, deeply stained bodies distributed at random in the nucleoplasm or in the nucleolus (figs. 3 and 5). These, when seen at high magnification, sometimes consist of granules similar in size to the nucleolar granules. Sometimes they seem opaque. Although such bodies are found outside the nucleolus, in the chromatin network, their relationship to the nucleolus is unquestionable. Dense bodies were observed for the first time in the adenocarcinoma of the leopard frog (26) and later in the benign human-skin tumors of molluscum contagiosum (27, 28). They may be numerous in Rous sarcoma cells cultivated in vitro (29). In addition to molluscum contagiosum and Rous sarcoma, "dense bodies" are present also in great quantities in polyoma virus-infected cells in vitro (14, 15). What makes them important with respect to our problem here is that they are found also in some "classical" virus infections such as herpes B {20), adenovirus (SO), and varicella (31). Although they do occur in normal cells-especially when rapidly growing-they are only scattered and scarce and never present in such an amount as that depicted here. It looks as if the "dense bodies" were capable of becoming hyperplastic under certain conditions, and because of their occurrence in a few virus-infected cells, they may sometimes be truly indicative of viral involvement. In this respect it is important to note that they have been found in human breast cancer (32).

To conclude this survey concerning the nucleus, one should also briefly mention the nuclear membrane. This is not altered in tumors, virus-induced or otherwise. General hyperplasia of the nucleus often leads to its profound indentation. In tangential thin sections, the characteristic nuclear pores then become remarkably conspicuous, a feature of which the morphologist in search of virus particles in tumor cells should be well aware (see fig. 20).

A recent relevant observation, probably related to the presence of virus, if not to virus tumors, is the remarkable reduplication of the nuclear membrane described in adenovirus-infected HeLa cells by Gregg and Morgan (33). But none of these changes may be considered as specific.