To develop an immune-competent mouse model of early-stage disease and breast cancer progression, we used a previously established MIND model developed by our group (16, 28), which allows for cancer cell lines to be injected directly up the teat of mice in the absence of surgical manipulation and surgery’s associated pathologic inflammation. Using this method, we characterized tumor formation and progression of six murine mammary tumor cell lines (D2A1, D2.OR, 4T1, EMT6, Py230, EO771), in two strains of immune-competent mice (BALB/c, C57BL/6).
Murine mammary cell lines delivered intraductally model human invasive breast cancer
We next sought to compare the histology of mammary tumors derived from mammary fat pad injections—the most common transplant model approach—to the histology of tumors derived from intraductal injection. To achieve this, murine mammary cancer cell lines, Py230 or EO771, were intraductally injected into the 4th right mammary teat of C57BL/6 mice. In the contralateral 4th left mammary gland, Py230 or EO771 tumor cells were injected directly into the mammary fat pad. Tumors were collected 14 days (EO771) or 37 days (Py230) later, and assessed through H&E staining.
Injection of EO771 (Fig. 2A) and Py230 (Fig. 2B) cells directly into the mammary fat pad resulted in the formation of solid, invasive tumors with necrotic cores. In contrast, intraductal injection of these same cells resulted in the development of heterogeneous tumors, with phenotypes more consistent with invasive human breast carcinoma. Specifically, MIND model EO771 tumors were characterized by an infiltrating fat phenotype (Fig. 2C), similar to some human breast cancers (Fig. 2E). MIND Py230 tumors were characterized as having phenotype more comparable to ‘pushing’ border human breast cancers (Fig. 2D,F). These data are consistent with previously published work demonstrating that tumors which arise from intraductal tumor cell delivery more closely reflect human breast cancer histology than tumors derived from mammary fat pad injections (19, 20, 22).
Intraductal injection of murine mammary cell lines into BALB/c immune-competent hosts does not recapitulate human DCIS
We next determined whether intraductal delivery of murine mammary tumor cells into immune-competent hosts form DCIS-like lesions. BALB/c mammary tumor cell lines, D2A1, D2.OR, 4T1, or EMT6, were intraductally delivered into the mammary gland of immune-competent BALB/c mice. Based on preliminary experiments designed to find optimal cell numbers and time to overt tumor development, cells were intraductally injected at 1x104 (10K) or 5x104 (50K) cells per gland, and glands collected between 11–33 days post tumor collection (Supplementary Table 2). These time-points were prior to the development of overt lesions as assessed by palpation, but concurrent with micro-lesions as confirmed by histological assessment (Fig. 3).
To evaluate presence of DCIS-like lesions, mammary glands were assessed for myoepithelial layer integrity, which defines whether the tumor has progressed from in situ to invasive disease. We stained for expression of myoepithelial markers p63, calponin, and SMA, which clinically are used to delineate DCIS from IDC. We found that intraductal injection of all four BALB/c mammary tumor cell lines primarily resulted in invasive disease, with DCIS rarely detected. We observed multifocal, invasive lesions associated with disrupted ductal integrity as measured by loss of p63, SMA, and calponin staining (Fig. 3A,B). Two cell lines (4T1 and EMT6), when injected at lower concentrations, displayed DCIS-like lesions as a rare event (Fig. 3B; 1/10 and 1/11 injections, respectively), and these DCIS-like lesions were coincident with IDC lesions (Fig. 3C; Supplementary Fig. 1; Supplementary Table 2). These findings suggest that these BALB/c mammary tumor cell lines, when delivered intraductally into immune-competent hosts, do not form stable DCIS-like lesions nor progress through an in situ DCIS stage prior to invasion.
One mechanism that may contribute to the low DCIS incidence in these immune-competent BALB/c mice is the presence of a Th2-skewed immune system. In the BALB/c host, the Th2-skewed immune system is characterized by an abundance of regulatory T cells, and reduced abundance of cytotoxic CD8 + T cells (29). This immune milieu is tumor promotional in other settings, and thus may accelerate progression from in situ to invasive disease. We assessed for CD45 (common lymphocyte antigen) expression and found an accumulation of CD45 + cells around these intraductally-derived tumors (Fig. 3A); an observation consistent with tumor cell injection stimulating an immune response, despite tumor cells being derived from the same BALB/c background.
To further explore if the host immune skew might influence DCIS formation, we next investigated a C57BL/6 immune-competent MIND model for its ability to support DCIS development. The C57BL/6 mouse strain is associated with a Th1-skewed immune milieu, characterized by an increased abundance of cytotoxic CD8 + T cells (29, 30). As this Th1-skewed immune milieu is suggested to be tumor inhibitory (31), it is possible that the immune milieu of C57BL/6 mice may better facilitate DCIS development.
Intraductal injection of murine mammary cell lines into C57BL/6 immune-competent hosts does not recapitulate human DCIS.
C57BL/6 tumor cell lines Py230 and EO771 were injected intraductally into mammary glands of C57BL/6 mice. Based on preliminary experiments designed to find optimal cell numbers and time to overt tumor development, Py230 cells were intraductally injected at 1x104 (10K) and 5x104 (50K) cells per gland, and EO771 cells were injected at 5x103 (5K) or 5x104 (50K) cells per gland. For Py230 intraductal injections, mammary glands were collected 14 days post-injection. For EO771 intraductal injections, we extended the EO771 model to 35 days post-tumor cell injection. These time-points were prior to the development of overt lesions as assessed by palpation, but concurrent with micro-lesions as confirmed by histological assessment (Fig. 4). Mammary glands were collected and tumor incidence, multiplicity and progression to invasive disease assessed. To assess prevalence of DCIS, tumors were stained for myoepithelial p63, SMA and calponin expression.
Similar to our observations in the BALB/c model, we found that intraductal tumor cell injection resulted in the establishment of multifocal invasive lesions within the mammary gland (Fig. 4A, B). At 35 days post injection, intraductal injection of EO771 cells resulted in only invasive tumors, at both high and low cell numbers (Fig. 4C,D). The majority of Py230 lesions were locally invasive (57/63; 90.5% IDC) and a minority were DCIS-like (6/63; 9.5% DCIS). Of note, DCIS lesions formed only in a single Py230 injected gland (1/11) (Fig. 4C), with invasive disease also observed in this gland. It is noteworthy to highlight that injecting fewer cells did not increase the incidence of DCIS for any cell line (Fig. 4C; Supplementary Fig. 2; Supplementary Table 2). Similar to the BALB/c model, when assessed for CD45 expression, there was an accumulation of CD45 + cells around the intraductally-derived tumors (Fig. 4A), consistent with tumor cell injection stimulating an immune response in this C57BL/6 mouse model.
Murine mammary tumor cell lines do not recapitulate in situ disease when intraductally injected into immunocompromised hosts.
The influx of CD45 + cells into intraductally-derived tumors in both BALB/c and C57BL/6 mice, together with the low incidence of DCIS, suggests that a host immune response might impair the ability of DCIS lesions to form in syngeneic, immune-competent mouse models. Previous studies have demonstrated that human breast cancer cell lines injected intraductally into immunocompromised mice progress through a long-lived (months) DCIS stage prior to invasion (16, 18, 19). Therefore, we next sought to determine whether intraductal injection of murine mammary cancer cell lines into immunocompromised hosts similarly permits progress through a pre-invasive, in situ stage.
First, using human breast cancer cell lines HCC70 and MCF-7, we confirmed DCIS-like lesion formation in SCID mice when using our surgery-free intraductal delivery model. For both cell lines, intraductal injection resulted in the development comparatively large of tumors that histologically resembled DCIS, and were confirmed to be in situ lesions through H&E analysis and by intact calponin staining in the surrounding myoepithelium (Fig. 5). Next, murine mammary tumor cell lines Py230 and EO771 were injected into the mammary glands of SCID/C57BL/6 (SCID) mice. In parallel, tumor cells were injected into wild-type C57BL/6 (WT) mice as controls. Intraductal delivery of Py230 and EO771 cell lines into immunocompromised mice again did not result in the development of DCIS-like lesions. All Py230 and EO771 lesions that formed in the immunocompromised mice were invasive lesions as determined by the loss of myoepithelial markers SMA and calponin (Fig. 6). Intriguingly, while tumor incidence did not differ between SCID and WT mice, tumor multiplicity and average tumor size per gland was significantly increased in immunocompromised mice (Supplementary Fig. 3). Thus, while the use of immunocompromised mice did not increase incidence of DCIS, the attenuated immune response in SCID mice led to more successful establishment and growth of lesions. (Supplementary Fig. 3, Supplementary Table 3).