In regard to the second alternative, indications that the surrounding stroma is physiologically connected with the mammary epithelium have already been obtained (17). When, after a prolonged dissociation by collagenase, a mammary gland yields cultures of pure epithelium, one is surprised to find no major vacuole or lipidic inclusion develop in the clear cytoplasm of these cells. Even though the mammary glands are, in most cases, in the prelactating stage, lipides are not formed by the epithelium. If the mammary epithelium is explanted while in full secretion, there is an abundance of lipides at the start, but this very abundance wears off as the culture is prolonged. It all appears as if the lipides secreted were of exogenous origin. This becomes more evident when the mammary glands of a mouse taken at mid-pregnancy are explanted, one after complete dissociation by collagenases, which gives pure mammary epithelial cultures, the other after a short enzymatic action, which retains a great part of the adipose tissue of the gland. Numerous lipidic inclusions in the mammary epithelium are observed only in the presence of the adipose tissue; none are seen in cultures of pure epithelium. The elaboration of milk lipoproteins by the mammary epithelium therefore depends on the surrounding tissues of the glandular stroma.

As shown by the results obtained in organ cultures, the multiplication of the milk agent may also depend on the environment. Roller-tube cultures from prelactating mammary glands, partially dissociated with collagenase in order to retain part of the stroma, are then inoculated, as previously, by adding RIII milk in the standard dilution of 10^-3 to the medium. After 3 weeks of maintenance with 2 changes of medium a week, the cultures are ground, centrifuged at low speed, and the supernatant injected into C57BL mice.

Ten of the 15 experimental mice developed a mammary tumor (table 3), the first one appearing less than 9 months after the bioassay. The percentage of tumors induced is even higher than after the inoculation of fresh milk. In comparison, no tumor appeared in the animals injected with the extracts of control cultures which did not receive RIII milk. If one recalls that only two tumors had been induced in 20 mice by cultures of pure mammary epithelium treated in similar conditions, one must conclude that the glandular stroma is somewhat involved in the multiplication of the milk agent.

Table 3. Influence Of The Glandular Stroma On The Maintenance Of The Milk Agent In Vitro


Number of mice inoculated

Number of tumors induced

Tumors (percent)

RIII milk 10^-3








Partially dissociated gland. Control cul-

ture 10^-3




Partially dissociated gland + MTA ** 10^-3




Totally dissociated gland + MTA 10^-3




*IMTA = Incubated mammary-tumor agent (3 days 37° 0.). **MTA - mammary-tumor agent.

Discussion And Conclusions

Because of its far-reaching implications, this last experiment must be further confirmed. There is no doubt, however, that the milk agent can multiply independently in tissue culture, outside its natural host. In this respect, and also because of its structural appearance as revealed by the electron microscope, one has to consider the MTA as an authentic virus.

The mouse mammary virus has been found in the liver, lungs, kidneys, spleen, thymus, heart, blood, and seminal vesicles (18). It is significant that, despite this wide distribution, it induces tumors only in the mammary epithelium. It can therefore survive for long periods in various locations, but finds the conditions necessary for tumor formation only in one elective site, the mammary gland. Tissue-culture techniques indicate another selective characteristic that thus far has not been displayed by other kinds of viruses or expressed in definite terms: The milk agent will develop in the mammary epithelium only if this epithelium is "conditioned" by the proper environment, i.e., by the glandular stroma, possibly by a particular state of this stroma.

We know that the mammary gland is an organ subject to periodicity and that its functional activity, under critical hormonal control, is accompanied by important structural and biochemical modifications (19). Under the influence of the hypophyseal hormones, for instance, the epithelial parenchyma varies tremendously in volume from the resting stage to full lactation, then through involution back to the resting stage. The ovarian hormones, on the other hand, are powerful modifiers of the stroma but, at the same time, spreading factors such as hyaluronidases appear, which are of fundamental importance in the prelactating period. Mam-motropin also initiates an increased fat elaboration in the adipose tissue and so probably influences the reactions of enzyme systems such as diastases, phosphatases, lipases, and dehydrogenases, which have been reported as normally present in this tissue (20). We have shown that secretion by the mammary epithelium is connected with the activity of the adipose tissue (17). If, as suggested by our last experiment, the multiplication of the MTA is influenced by similar factors, its synthesis in the epithelial cells may also depend on the presence of a specific metabolite supplied by the surrounding tissues.

Such a mechanism would explain some of the oddities of this virus. Because of the periodic function of the mammary gland, any specific metabolite may be available only at one definite stage of the glandular cycle. The synthesis of virus proteins cannot therefore be continuous, but strictly limited to this period. Before, or after, the virus just survives as it does in the other parts of the body, but in the mammary gland it will resume active multiplication whenever the conditions permit. This would be consistent with the fact that tumors appear generally late in the life of the mouse, usually after several pregnancies. It is also consistent with the heritable hormonal pattern that has been advanced to explain the differences in tumor incidence between susceptible and resistant strains (21). Variations in the normal secretory output of hormones involved in the mammary-gland control will evidently "condition" the tissues of this organ differently. In this respect, it is significant that the RIII strain of mice, which has a high incidence of spontaneous tumors, is a great milk producer, while the C57BL strain, resistant to the milk agent, gives in similar conditions a very small yield. It is no less significant that hyperplastic nodules with a typical lobulo-alveolar structure develop in virgin RIII mice and not in the C57BL. Pitelka et al. have shown these hyperplastic nodules to be a site of virus multiplication (22), but Nandi (23) demonstrated that lobulo-alveolar development is possible only under the combined action of ovarian hormones, deoxycorticosterone acetate, and either prolactin or growth hormone. These hormones, probably more readily available in the young RIII than in the C57BL strain, appear as factors preparing the ground for virus multiplication rather than as agents directly involved in mammary carcinogenesis.

Another fact in support of an exogenous complement necessary for the epithelial cells to synthesize new virus proteins is given by cultures of mouse mammary tumors. In the 48 hours after the primary explanation of neoplastic cells, a large amount of viral particles is produced, which can be demonstrated by electron microscopy in the culture medium itself. Since we considered this method as a possible way of producing large amounts of virus, we checked systematically the medium of tumor tissues maintained in continuous culture. It was observed that the viral particles, so abundant in the first days of cultivation, were thinning away from the 10th day down to the 15th or 20th day, after which particles were seldom seen. Since the cells themselves remain in excellent morphologic condition and the composition of the medium is constant, it seems that the loss of the viral particles is caused by default of a necessary component explanted with the primary tissue but exhausted after a few days of life in vitro.

Several points remain to be clarified to have a precise picture of the mechanisms involved in mouse mammary carcinogenesis. The results we have presented, however, offer an opportunity for a more logical experimental approach. If we succeed in obtaining a continuous production of the mammary-tumor virus in vitro, several of these remaining problems will eventually be solved.

To summarize: Evidence of multiplication of the milk agent has been obtained in organotypic cultures of embryonic mouse skin in which mammary rudiments differentiated from the epidermis. However, conventional roller-tube cultures of embryonic skin epithelium failed to maintain the virus for any length of time. Maintenance without multiplication was achieved in cultures of pure mammary epithelium from the adult mouse. On the other hand, there are definite indications that the virus will grow on adult mammary epithelium, if a part of the adipose tissue from the glandular stroma is retained in the culture. In the light of this last observation, it is suggested that synthesis of virus proteins by the mammary epithelium is mediated by specific metabolites supplied by the surrounding tissues.

Plate 24

Figure 1. Hardy's technique of organ culture. Fragments of embryonic skin from 12-day mouse embryos are placed on the nutritive gel with the germinal layer down. Mammary tubes differentiate in underlying connective and adipose tissues.

Figure 2. 'Expanding mouse mammary epithelium; 7-day culture from adult C57BL mammary epithelium explanted at mid-pregnancy. X 240.

Hardy's technique of organ cultureHardy's technique of organ cultureExpanding mouse mammary epithelium