Metastatic dissemination of cancer cells, the major cause of cancer mortality, is traditionally viewed as a late-stage event (1
), although mammary epithelial cells have been shown to disseminate systemically from early neoplastic lesions in transgenic mice and from ductal carcinoma in situ in women (2
). There is ample evidence that the ability of fully transformed tumor cells to metastasize depends on the regulation of developmental programs and external environmental cues (3
), but to what extent the seeding or growth of tumor cells at the ectopic site is dependent on the initiating transforming event(s) is a subject of debate (12
). We have developed a system that separates the process of seeding cells in the lung from the process of malignant growth at an ectopic site by using animals engineered to express potent oncogenes in a doxycycline-dependent mammary-specific manner. After intravenous (IV) injection of marked mammary cells that have different genetic potentials [no oncogenes, or oncogenes that will be expressed only after cells have taken up residence in an ectopic site (the lungs)], normal mammary cells can lodge in the lungs, grow slowly, and become frank metastatic malignancies once potent oncogenes are turned on.
We recently described tri-transgenic TetO-MYC;TetO-KrasD12;MMTV-rtTA
) mice that coordinately express MYC
and mutant Kras
oncogenes in mammary epithelial cells when fed doxycycline (13
). Doxycycline-naïve animals do not express the transgenic oncogenes and have morphologically and functionally normal mammary glands, but they develop diffuse autochthonous tumors within 3 to 4 weeks after doxycycline exposure. Tumors that form because of the expression of these oncogenes display malignant characteristics, such as transplantability and metastasis (fig. S1
). Therefore, this model provides primary mammary cells that can be switched from a normal to a neoplastic state with simple experimental manipulation.
To investigate whether mammary cells from these mice can be induced to form metastasis in the absence of transformation at the primary site, we modified the traditional experimental metastasis assay (14
). In the modified approach, instead of IV delivery of tumor cells from doxycycline-treated animals into new recipients, we injected dissociated morphologically normal mammary cells from mature TOM;TOR;MTB
animals never exposed to doxycycline into the tail veins of Rag1−/−
) females on a doxycycline diet. In this way, the injected cells can become transformed only in the bloodstream or tissues of the recipient mouse, an experimental situation that has not been previously examined. Magnetic resonance imaging (MRI) was used to survey Rag1−/−
recipients for evidence of tumor foci in the lungs. Solitary nodules were observed in four out of four recipients 6 weeks after injection (), and histological sections showed foci of mammary adenocarcinoma in the lung (); in contrast, no tumors were found in control mice that did not receive doxycycline ( and table S1
). The experimental metastases were histologically identical to the spontaneous metastases observed in tumor-bearing TOM;TOR;MTB
mice (fig. S1
) as well as to the primary mammary tumors arising in donor animals (13
), and pulmonary nodules contained cells positive for MYC (), keratin 8 (K8), smooth-muscle actin (SMA), and keratin 6 (K6) (fig. S2
). These findings demonstrated that the tumorigenic capacity conferred on mouse mammary cells by coexpression of MYC
can be realized in the ectopic environment of the lung.
Fig. 1 Untransformed mouse mammary cells form lung metastases after IV injection and induction of oncogenes. (A to D) Lung metastases develop from intravenously injected phenotypically normal mammary cells upon activation of MYC and KrasD12 transgenes. (A) Metastases (more ...)
We confirmed this observation in a different mouse model, TetO-PyMT:IRES:Luc;MMTV-rtTA
), recently generated in our lab (14
). Expression of the polyoma middle T (PyMT
) oncogene in this line is mammary gland–specific and doxycycline-dependent and can be monitored through the coordinate expression of the reporter gene (Luc
) encoding firefly luciferase (fig. S3
). When dissociated mammary cells from mature TOMT:IRES:Luc;MTB
animals never exposed to doxycycline were injected intravenously in Rag1−/−
females on a doxycycline diet, a bioluminescence signal was apparent over the thorax of recipient mice within 2 weeks (). The development of mammary tumors in the lungs of injected mice was documented by MRI imaging () and by histology (). Ectopic mammary tumor foci derived from either TOM;TOR;MTB
transgenic lines displayed characteristics associated with oncogene activation at the primary site, including robust proliferation () and dense vasculature (). Therefore, upon activation of potent oncogenes, previously untransformed mouse mammary cells delivered to the systemic circulation produce metastatic-like disease in the pulmonary parenchyma without having undergone transformation at the primary site.
Fig. 2 Delay in oncogene activation does not preclude the development of ectopic mammary tumors. (A) Bioluminescence in Rag1−/− mice after IV delivery of 1 × 105 mammary cells from doxycycline-naïve TOMT:IRES:Luc;MTB mice is undetectable (more ...)
To explore whether untransformed mammary cells can survive in the bloodstream and in the ectopic environment of the lung without oncogene expression yet still be induced to develop tumors at a later time, we injected dissociated mammary cells from TOMT:IRES:Luc;MTB
animals never exposed to doxycycline into the lateral tail veins of Rag1−/−
females on a doxycycline-free diet. No bioluminescence signal was observed in the lungs of recipient mice in up to 4 months of monitoring on this diet (). When recipients were instead placed on doxycycline 1.5, 8, or 17 weeks after IV injection of transgenic mammary cells, we detected bioluminescence in the chest within 2 weeks of the start of doxycycline exposure (). The histological appearance and the total number of foci in the lungs were similar in animals placed on a doxycycline diet 1 day before IV injection and 10 days after the injection [32 ± 6 (SD), n
= 5 mice and 26 ± 6 (SD), n
= 4 mice, respectively] (). Conversion to malignancy at later times of induction was not measured because expression of the MMTV-rtTA
transgene in the MTB
line becomes non-uniform with age (16
). These observations show that cells responsible for development of the ectopic mammary tumors can persist in the lung for up to 17 weeks in the absence of oncogene expression.
To rule out the possibility that low-level or transient expression of the transgenic oncogene(s) occurs in the absence of doxycycline, we carried out the modified experimental metastasis assay with mammary gland preparations from animals lacking any transgenic oncogenes (C57BL6/J mice) or from those expressing a gene encoding the “enhanced” green fluorescent protein (GFP) from the chicken β-actin
). Three weeks after the injection of mammary cells from β-actin-GFP
mice, green foci were observed with fluorescent microscopy of the whole lungs in all recipients (). These foci lacked the nodular appearance of metastatic tumors and were inconspicuous on routine histological inspection (, top left). Staining with antibody to K8 facilitated the detection of the ectopic foci and confirmed their epithelial origin, whereas in the lungs of uninjected mice only the bronchial epithelium was stained (, right).
Fig. 3 Mammary cells without an oncogenic transgene can persist in the lung. (A) Focus of green cells observed under excitation light (14) in a whole fresh lung of a Rag1−/− recipient 3 weeks after IV injection of 5 × 105 dissociated (more ...)
To determine whether the ectopic foci of normal epithelial cells persist and grow in the foreign environment of the lung, we counted the total number of discrete foci in lung sections at different times after injection and looked at proliferation markers in these foci. The total number of foci found in lung sections from C57BL6/J recipients injected with 4 × 105
syngeneic mammary cells was similar in the animals surveyed at 3 weeks (n
= 3 mice) and those surveyed at 10 weeks (n
= 3 mice) after injection (42 ± 7 and 56 ± 22 in 10 paraffin lung sections, respectively). Moreover, the efficiency with which the wild-type cells were able to form these small epithelial clusters was similar to the efficiency with which we were able to induce ectopic tumors after injecting cells from doxycycline-naïve TOM;TOR;MTB
donors [1.2 ± 0.4 (SD) versus 1.7 ± 1.4 (SD) per 10,000 cells injected, n
= 6 and 8 mice, respectively; measured as described in (14
)]. This result strongly argues that most or all of the mammary cells that are capable of surviving in the lung are able to respond to the initiating oncogene expression by forming an ectopic mammary tumor.
In both nontransgenic C57BL6/J– and β-actin-GFP
–derived foci, occasional cells displayed mitotic activity (). Consistent with this result, the green foci found under excitation light in the lungs of animals injected with mammary cells from β-actin-GFP
mice 16 weeks after injection were larger in size than those found in recipients of the same preparation 1 week after injection (). Ectopic epithelial outgrowths contained K8- and SMA-positive cells, such as observed in intact mammary glands (fig. S4A
), and the outgrowths occasionally displayed a glandular appearance. Despite prolonged residence in the lung (up to 4 months), the green cells recovered from the recipients' lungs were competent to form hollow acinar structures in three-dimensional morphogenesis assays (fig. S4B
) and secondary mammary outgrowths in cleared fat pads of Rag1−/−
females (). These findings establish that the ectopic cells residing in the lungs are indeed of mammary origin, that they are viable and mitotically active, and that at least some of them are multipotent and able to support full mammary development.
The experiments described here show that, in the absence of an active oncogene, dissociated cells from an untransformed mouse mammary gland can establish residence in the ectopic environment of the lung, grow slowly, and remain clinically undetectable after IV injection. The same cells can give rise to metastatic malignancies upon activation of oncogenes that can produce mammary tumors in an intact gland. It is widely acknowledged that multiple steps are required to establish metastases, including intravasion of cells from primary tumors into blood vessels or lymphatics; survival in the circulation, extravasation, and establishment of cells at ectopic sites; and malignant growth. Because we have injected mammary cells from transgenic mouse donors into tail veins of recipient mice, we have not examined the requirements for intravasation. We have, however, demonstrated that activated oncogenes and cellular transformation are not required for any of the subsequent steps, save for malignant growth at ectopic sites. These findings indicate that properties inherent in normal cells are sufficient for negotiating a substantial portion of the metastatic cascade. Considerable experimental and clinical evidence favors the idea that cells from small cancers may spread to distant sites early in tumorigenesis and account for dormancy and late relapse in human breast cancer (2
). Although we do not know whether premalignant cells can enter the systemic circulation during these early stages and become sources of later metastatic tumors, our observations argue that this hypothesis should be tested. The finding that metastatic disease can arise from untransformed mammary cells in the circulation refines our conception of cancer progression, and suggests that each step in the metastatic cascade should be examined to establish its functional requirements, including those performed by normal cells. Such functions might be susceptible to inhibitory strategies that can ablate disseminated pre-malignant or malignant cells and thereby diminish the mortality caused by cancer.