We have examined the effect of adenoviruses expressing soluble transforming growth factor receptorII-Fc (sTGFβRIIFc) in a 4T1 mouse mammary tumor bone metastasis model using syngeneic BALB/c mice. Infection of 4T1 cells with a non-replicating adenovirus, Ad(E1−).sTβRFc, or with two oncolytic adenoviruses, Ad.sTβRFc and TAd.sTβRFc, expressing sTGFβRIIFc (the human TERT promoter drives viral replication in TAd.sTβRFc) produced sTGFβRIIFc protein. Oncolytic adenoviruses produced viral replication and induced cytotoxicity in 4T1 cells. 4T1 cells were resistant to the cytotoxic effects of TGFβ-1 (up to 10 ng ml−1). However, TGFβ-1 induced the phosphorylation of SMAD2 and SMAD3, which were inhibited by co-incubation with sTGFβRIIFc protein. TGFβ-1 also induced interleukin-11, a well-known osteolytic factor. Intracardiac injection of 4T1-luc2 cells produced bone metastases by day 4. Intravenous injection of Ad.sTβRFc (on days 5 and 7) followed by bioluminescence imaging (BLI) of mice on days 7, 11 and 14 in tumor-bearing mice indicated inhibition of bone metastasis progression (P<0.05). X-ray radiography of mice on day 14 showed a significant reduction of the lesion size by Ad.sTβRFc (P<0.01) and TAd.sTβRFc (P<0.05). Replication-deficient virus Ad(E1−).sTβRFc expressing sTGFβRIIFc showed some inhibition of bone metastasis, whereas Ad(E1−).Null was not effective in inhibiting bone metastases. Thus, systemic administration of Ad.sTβRFc and TAd.sTβRFc can inhibit bone metastasis in the 4T1 mouse mammary tumor model, and can be developed as potential anti-tumor agents for breast cancer.
In the United States alone, of the nearly 209,000 women diagnosed with breast cancer each year, about 43,000 die.1 A majority of the women develop bone metastases, tumor-induced bone destruction, hypercalcemia and spinal cord compression during the advanced stages of breast cancer, thus seriously compromising the lifestyle of the affected patients.2 Currently, there are only limited therapies for bone metastases. Although the two types of drugs—bisphosphonates and denosumab, an antibody against RANKL (receptor activator of nuclear factor kappa-B ligand)—can inhibit bone resorption, their ability to cure bone metastases remains to be established.3 Thus, development of novel therapies to treat breast cancer bone metastasis is a major unmet need in medicine.
In the recent years, oncolytic adenoviruses have shown some potential in the treatment of cancer.4, 5, 6, 7, 8, 9, 10, 11, 12 In an attempt to develop novel therapeutic approaches for bone metastases, our laboratory has developed oncolytic adenoviruses that would kill the cancer cells and simultaneously express a soluble form of transforming growth factor beta (TGFβ) receptorII-Fc (sTGFβRIIFc) that can target TGFβ-induced signaling pathways.10, 11, 12 We chose to target the TGFβ pathway because high levels of circulating TGFβ protein is a poor prognostic marker in breast cancer patients.13, 14 Furthermore, aberrant TGFβ signaling at the bone metastasis site has been postulated to be a key factor in the progression of breast cancer bone metastases.14, 15, 16, 17, 18, 19 Therefore, there is a growing interest in developing inhibitors of TGFβ signaling for the treatment of various cancer metastases.20, 21, 22, 23, 24 Using an MDA-MB-231 human breast cancer bone metastasis model in immunodeficient mice, we have recently shown that intravenous delivery of oncolytic adenoviruses, Ad.sTβRFc and TAd.sTβRFc, in tumor-bearing mice are effective in inhibiting the established bone metastases.11, 12 However, before initiating a clinical trial in breast cancer patients, it is important to examine the efficacy of these oncolytic adenoviruses in an immunocompetent animal model because they have the ability to limit adenoviral replication and thus its efficacy. Keeping that in mind, we have now conducted in vitro and in vivo studies using a mouse mammary 4T1 tumor cell model. We report here that infection of 4T1 cells with recombinant adenoviruses produced transgene expression, and 4T1 cells supported adenoviral replication and were killed by oncolytic adenoviruses. Although 4T1 cells were resistant to TGFβ-1-induced cytotoxicity, TGFβ was able to activate signaling. More importantly, intracardiac inoculation of 4T1 cells in BALB/c mice produced bone metastases and osteolytic lesions, and thus is an appropriate pre-clinical model for our purpose. We report here that intravenous injections of Ad.sTβRFc and TAd.sTβRFc inhibited the progression of skeletal metastases in BALB/c mice. Based on our findings, we believe that oncolytic adenoviruses targeting TGFβ pathways can be developed for treating breast cancer bone metastases.
Materials and methods
4T1 (ATCC, Manassas, VA) mouse mammary tumor cells, 4T1-luc2 (Caliper life sciences, Hopkinton, MA), MV1Lu (ATCC) mink epithelial cells, and HEK 293 (ATCC, Manassas, VA) human embryonic kidney cells were grown in Dulbecco's modified Eagle's medium containing 10% bovine calf serum (Invitrogen, Grand Island, NY).
Adenoviral vectors used in these studies are: Ad(E1−).Null, an E1 minus replication-deficient adenovirus containing no foreign gene; Ad(E1−).GFP, a replication-deficient adenovirus expressing EGFP protein; Ad(E1−).sTβRFc, a replication-deficient adenovirus expressing sTGFβRIIFc gene; Ad.sTβRFc, an oncolytic adenovirus expressing sTGFβRIIFc gene (constructed using dl01/07 mutant of Ad5, containing two deletions in E1A region as previously described)10 and TAd.sTβRFc, an oncolytic adenovirus expressing sTGFβRIIFc gene with the human TERT promoter driving the adenoviral replication as published.25 Adenoviral vectors were grown in HEK 293 cells and purified by double CsCl2 gradient as described.26 Viral particle (VP) numbers were determined by measuring OD260 (optical density 260) of the sodium dodecyl sulfate-treated adenoviral solutions.
Adenoviral replication assay
4T1 cells were plated in 24-well dishes (2 × 105 cells/well). The following day, cells were infected with viral vectors (5 × 103 VPs/cell) and the incubations were continued for 3 days. Cells were then subjected to immunohistochemistry for adenoviral hexon staining using an Adenoviral Titer commercial kit (Clontech, Mountain view, CA) as described earlier.25, 27 Positive hexon expressing brown cells were photographed, and counted under the microscope to quantify viral replication.
To measure TGFβ-1-induced cytotoxicity, cells were plated in 96-well plates 103 cells/well. The following day, cells were infected with various concentrations of TGFβ-1 (0.001–10 ng ml−1) (Sigma, St Louis, MO), and the incubations were continued for 7 days. Cells were washed, fixed and stained with sulforhodamine B (Sigma), and the A564 (absorbance at 564 nm) measured as previously described.26 Untreated control cells were considered to have 100% survival. To examine viral-induced cytotoxicity, the same protocol was used except that 4T1cells were incubated with various doses of adenoviral vectors for 7 days before sulforhodamine B staining.
4T1 cells were plated in 6-well dishes (4 × 105 cells/well). The following day, cells were infected with Ad(E1−).GFP (2.5 × 104 VPs/cell) and incubated for 48 h. Cells were photographed using a fluorescent microscope ( × 200).
To examine adenoviral vector-mediated sTGFβRIIFc expression, 4T1 cells were plated in 6-well dishes (4 × 105 cells/well). The following day, cells were infected with various viral vectors (2.5 × 104 VPs/cell). After 24 h, media was changed to serum-free media, and the incubations continued for another 24 h. sTGFβRIIFc expression in the media and the cell lysates were examined by western blot analyses as previously described.10 sTGFβRIIFc protein amounts in the media were measured by enzyme-linked immunosorbent assay using antibodies against the human IgG Fc fragment (Jackson ImmunoResearch, West Grove, PA, USA) as previously described.11
4T1 cells were plated in 6-well plates (4 × 105 cells/well). The following day, cells were serum starved for 6 h, and then treated with TGFβ-1 (1 ng ml−1) in the absence or presence of sTGFβRIIFc (250 ng ml−1) for 1 h. Cells were analyzed for p-SMAD2, p-SMAD3 and for total SMAD2/3 using western blots as previously described.28 The blots were visualized by enhanced chemiluminescence substrate (Amersham Biosciences, Piscataway, NJ).
Interleukin (IL)-11 assays
4T1 cells were plated in 6-well plates (2 × 105 cells per well). The following day, cells were serum starved over night, and then exposed to TGFβ-1 (0.1, 1 or 5 ng ml−1) for 48 h. Media were analyzed for IL-11 levels by enzyme-linked immunosorbent assay using the previously described method.28
All animal experiments were conducted using the animal protocols approved by the IACUC committee of the NorthShore University HealthSystem. To establish bone metastases, 4T1-luc2 cells were injected in the left heart ventricle (day 0) of 4-week-old BALB/c mice (Charles River laboratories, Wilmington, MA). On day 4, the mice were subjected to bioluminescence imaging (BLI) in dorsal and ventral positions using Xenogen IVIS Spectrum imaging equipment (Caliper life sciences, Hopkinton, MA). Photon signals were quantified using living image software 3.0 (Caliper life sciences, Hopkinton, MA) as previously described.12 Mice were divided into various groups, with statistically indistinguishable BLI signals among each group. Various viral vectors were administered via tail vein on days 5 and 7 (5 × 1010 VPs per injection/mouse, each injection in a 0.1-ml volume). The control group of mice was administered the buffer alone.
Mice were imaged in dorsal and ventral positions on days 7, 11 and 14 using the IVIS Spectrum imaging system (Caliper Life Sciences). Whole-body BLI signals were used to quantify the metastasis as previously described.12 Signals in the hind limbs were separately quantified to measure the skeletal metastases as described.12
On day 14 following tumor cell injections, mice were also subjected to X-ray radiography in prone position using Faxitron (Faxitron X-ray Corporation, Wheeling, IL). Skeletal lesion sizes were measured in the femur and tibia of both the hind limbs using Image J software as described earlier.11, 12
All statistical analyses were performed using GraphPad Prizm 5 (GraphPad software, San Diego, CA). Data are presented as mean±s.e.m. To analyze BLI signal progression, a two-way repeated-measure analysis of variance followed by Bonferroni post-tests was used. For multiple groups, statistical significance was analyzed using one-way analysis of variancefollowed by Bonferroni post-tests. P<0.05 was considered a statistically significant difference.
4T1 cells can be infected with human adenoviral vectors
Experiments were conducted to examine the infectability of 4T1 cells with replication-deficient and replication-competent adenoviral vectors. 4T1 cells were infected with Ad(E1−).GFP, a non-replicating adenovirus, for 48 h, and the cells were visualized under a fluorescent microscope. The vast majority of cells produced a strong GFP signal (Figure 1a). In another experiment, cells were infected with Ad(E1−).sTβRFc, a replication-deficient adenovirus, and two oncolytic adenoviruses—Ad.sTβRFc and TAd.sTβRFc. Cell lysates and the extracellular media were subjected to western blot analyses for sTGFβRIIFc expression. Infection of 4T1 cells with Ad(E1−).sTβRFc, Ad.sTβRFc and TAd.sTβRFc resulted in sTGFβRIIFc protein production, which could be detected in the cell lysates as well as the extracellular media (Figure 1b). The amounts of sTGFβRIIFc were quantified in the media using enzyme-linked immunosorbent assay, and were found to be in the range of 6.21–15.48 μg ml−1 of media (Figure 1c). These results indicate that 4T1 cells can be infected with human adenoviruses and that infection with Ad(E1−).sTβRFc, Ad.sTβRFc or TAd.sTβRFc leads to the production of sTGFβRIIFc protein, which is secreted into the media.
Oncolytic adenoviruses replicate and induce cytotoxicity in 4T1 cells
Next, we examined the replication potential of adenoviral vectors in 4T1 cells. Cells were incubated with various adenoviral vectors (5 × 103 VPs/cell) at 37 °C for 72 h, and viral titer was determined by hexon staining. Figure 2a shows typical hexon staining of 4T1 cells exposed to various viral vectors. There were very few hexon expressing brown cells in Ad(E1−).Null- or Ad(E1−).sTβRFc-treated samples (Figure 2a). However, a large number of 4T1 cells infected with Ad.sTβRFc or TAd.sTβRFc were hexon positive (Figure 2a). Quantification of hexon-positive cells indicated that Ad.sTβRFc produced viral titers (Figures 2b and c), which were about 257-times higher than the non-replicating adenovirus Ad(E1−).Null (P<0.001). TAd.sTβRFc produced 175-times higher viral titer than Ad(E1−).Null (P<0.01) (Figures 2b and c). However, viral titers in Ad(E1−).sTβRFc-infected cells were similar to those in Ad(E1−).Null-treated cells (Figures 2b and c). From these results, we conclude that continuous incubation of 4T1 cells with oncolytic adenoviruses can produce viral replication.
To examine whether viral replication can result in cytotoxicity, 4T1 cells were incubated with various adenoviral vectors for 7 days, and the cytotoxicity assays were performed. Both the oncolytic adenoviruses, Ad.sTβRFc and TAd.sTβRFc, produced a dose-dependent cytotoxicity in 4T1 cells (Figure 2d). Based on the IC50 values, Ad.sTβRFc and TAd.sTβRFc were about 34.2-fold and 24.0-fold, respectively, more toxic than the non-replicating virus Ad(E1−).Null (Figure 2e). By contrast, a non-replicating virus Ad(E1−).sTβRFc produced toxicity that was comparable with Ad(E1−).Null (Figures 2d and e).
4T1 cells are resistant to killing by TGFβ, but retain TGFβ-mediated signaling pathways
Next, we investigated the killing effect of TGFβ in 4T1 cells. 4T1 cells were exposed to various concentrations of TGFβ-1, and 7 days later cytotoxicity was measured. As a positive control, another rodent cell-type MV1Lu, known to be sensitive to TGFβ, was used. As shown in Figure 3a, there was little to no cytotoxic effect of TGFβ-1 even at the highest concentration used (10 ng ml−1) in 4T1 cells. However, MV1Lu cells were killed even by a very low concentration of TGFβ-1, with an IC50 of <0.1 ng ml−1.
To examine whether TGFβ could induce signaling in 4T1 cells, we exposed 4T1 cells to TGFβ-1 and analyzed the cell lysates for SMAD2 and SMAD3 phosphorylation. Figure 3b shows that TGFβ-1 induced SMAD2 and SMAD3 phosphorylation in 4T1 cells. Co-incubation of sTGFβRIIFc with TGFβ-1 inhibited TGFβ-1-dependent SMAD2 and SMAD3 phosphorylation (Figure 3b). We also examined the effect of TGFβ-1 on IL-11 production, a known osteolytic factor in human breast cancer cells.28, 29 TGFβ-1 induced IL-11 protein production in a dose-dependent manner (Figure 3c). These results indicate that 4T1 cells respond to TGFβ-1 and undergo activation of signaling pathways that are known to favor bone metastases in human breast cancer cells.18, 28, 30 Importantly, sTGFβRIIFc is able to abolish the TGFβ signaling.
Oncolytic adenoviral-mediated inhibition of 4T1-induced metastases: BLI analyses
Next, we examined the effect of systemic administration of adenoviral vectors expressing sTGFRIIFc in a 4T1 bone metastasis model. 4T1-luc2 cells were inoculated into the left heart ventricles of BALB/c mice. After 4 days, mice were subjected to whole-body BLI in both dorsal and ventral positions. Mice were split into multiple groups, with nearly equal BLI signal within each group. Two doses of adenoviral vectors were given via the tail vein—the first dose on day 5 (5 × 1010 VPs/mouse) and a second dose on day 7 (5 × 1010 VPs/mouse). Mice were subjected to BLI on days 7, 11 and 14 following tumor-cell injection. A representative mouse showing BLI signal from each treatment group is shown in Figure 4a. Whole-body BLI signals were quantified and are shown in Figure 4b. There was a time-dependent increase in the whole-body BLI signal to 0.88 × 1011 photons sec−1 in the control group of mice that received buffer alone (Figure 4b). There was no significant inhibition of BLI signal in the Ad(E1−).Null, Ad(E1−).sTβRFc or TAd.sTβRFc treatment groups (Figure 4b). However, Ad.sTβRFc induced a significant inhibition (P<0.05) of the whole-body BLI. As 4T1 cells also established bone metastasis in the hind limbs (Figure 4a), the effect of viral vectors on the BLI signal in the hind limbs was quantified. In the control group of mice, the BLI signal in the hind limbs reached to 0.86 × 1010 photons sec−1, and there was a significant inhibition of BLI signal accumulation in the hind limbs in the Ad.sTβRFc-treated group (P<0.05). However, Ad(E1−).Null, Ad(E1−).sTβRFc and TAd.sTβRFc treatments had no significant effect on the BLI signal intensity in the hind limbs (Figure 4c).
Oncolytic adenoviral vectors’-mediated inhibition of 4T1-induced metastases: X-ray analyses
To further examine the effects of vectors’ administration on bone metastases, mice were subjected to X-ray radiography on day 14. A representative example of X-ray radiographs from each group is shown in Figure 5a. The long bones in buffer-treated and Ad(E1−).Null-treated mice had large osteolytic lesions, as indicated by red arrows. The lesion sizes were relatively smaller in the other treatment groups. The tumor lesions in the hind limbs were quantified using the Image J program and are shown in Figure 5b. Tumor sizes in the buffer group were 6.92±1.27 mm2. Tumor sizes in Ad(E1−).Null, Ad(E1−).sTβRFc, Ad.sTβRFc and TAd.sTβRFc treatment groups were 5.79±0.86, 3.56±0.72, 2.30±0.69 and 2.94±0.57 mm2, respectively. These results indicate that although Ad(E1−).Null and Ad(E1−).sTβRFc had no significant effect on the tumor sizes, a significant inhibition of the tumor growth was observed in the Ad.sTβRFc (P<0.01) and TAd.sTβRFc (P<0.05) treatment groups (Figure 5b).
Although we were able to monitor BLI and X-ray until day 14, by day 18 a number of animal deaths were observed in each of the treatment groups. The following ratios indicate the number of mice that died between days 14 and 18 over the initial number of mice in each group: buffer, 3:9; Ad(E1−).Null, 4:9; Ad(E1−).sTβRFc, 3:9; Ad.sTβRFc, 2:11; and TAd.sTβRFc, 3:11. To confirm Ad(E1−).sTβRFc-, Ad.sTβRFc- and TAd.sTβRFc-mediated sTGFβRIIFc expression in the blood, samples were collected from the remaining mice on day 18 and the sTGFβRIIFc production was analyzed by enzyme-linked immunosorbent assay. The results indicated high levels: 2.43±1.67, 6.53±16.31 and 15.41±24.86 μg ml−1 of sTGFβRIIFc in the blood samples from the Ad(E1−).sTβRFc, Ad.sTβRFc and TAd.sTβRFc treatment groups, respectively. Thus, although the intravenous injection of Ad(E1−).sTβRFc, Ad.sTβRFc and TAd.sTβRFc resulted in sTGFβRIIFc production, it appears that the replicating viruses expressing sTGFβRIIFc were the most effective in inhibiting bone metastases.
The key finding here is that intravenous delivery of oncolytic virus Ad.sTβRFc expressing sTGFβRIIFc can inhibit bone metastasis in the 4T1 mouse mammary tumor bone metastasis model in a syngeneic host as revealed by BLI studies. X-ray radiographic analyses showed inhibition of tumor growth by Ad.sTβRFc and TAd.sTβRFc, though Ad.sTβRFc was superior to TAd.sTβRFc. A non-replicating virus, Ad(E1−).sTβRFc, expressing sTGFβRIIFc showed some inhibition of bone metastasis in X-ray analyses; Ad(E1−).Null was not effective in either BLI or X-ray analyses.
Most of the previously published studies using oncolytic adenoviruses have been conducted in human xenografts established in immunodeficient nude mice, mainly because mouse tumor cells are not considered good targets for the human adenoviruses. However, it is critical that we continue to explore the animal models in which oncolytic adenoviruses can be examined in immunocompetent syngeneic hosts as described here. It is quite interesting that 4T1 cells can be infected with human adenoviruses resulting in high levels of transgene expression, indicating the presence of adenoviral receptors even in mouse 4T1 cells. Moreover, continuous exposure of 4T1 cells to oncolytic adenoviral vectors can produce a viral titer. This indicates that human adenoviruses can result in virus entry and replication, clearly demonstrating that human adenoviruses can indeed infect and replicate in 4T1 mouse tumor cells, which is consistent with a previous report.31 This infection could be via the previously known adenoviral receptor, by an unknown adenoviral receptor or even by other pathways, including clathrin-independent mechanisms such as macropinocytosis, phagocytosis or trans-endocytosis.32 Once the VPs are internalized by the cells, however, viral replication proceeds as in human breast cancer cells. Although the exact relationship of the Ad.sTβRFc- and TAd.sTβRFc-induced replication resulting in the cytotoxicity of the 4T1 tumor model remains to be examined, it is tempting to speculate that the viral replication resulting in the cytotoxic effects of the adenoviral vectors in the mouse tumor cells could have a role in mediating the in vivo anti-tumor responses reported here.
Another important observation is the inability of TGFβ to kill 4T1 cells and yet induce the TGFβ signaling pathway (SMAD-phosphorylation), and the production of IL-11 (a well-known osteolytic factor in human breast cancer bone metastasis). Thus, in this regard, 4T1 is an appropriate tumor model for examining the role of TGFβ signaling in bone metastases. In the radiographic analyses, a non-replicating adenovirus expressing sTGFβRIIFc showed some inhibition of bone metastasis, albeit weaker than oncolytic adenovirus Ad.sTβRFc. Again, these studies suggest that the expression of sTGFβRIIFc, coupled with viral replication and cytotoxicity, is potentially having a role in mediating the inhibition of bone metastases.
Intravenous delivery of adenoviruses will result in their uptake mainly in the liver, and in smaller amounts in other tissues and the skeletal tumors.11, 12, 27, 33 We believe that the infection of tumor cells in vivo will result in the viral replication in the tumor cells causing cell killing and partial tumor destruction. Infection of tumor cells and other mouse organs will result in the production of sTGFβRIIFc that will be released in the blood. The sTGFβRIIFc production resulting in the inhibition of TGFβ signaling at the tumor/bone site will also contribute towards the inhibition of bone metastases. Among the three vectors expressing sTGFβRIIFc, the most effective vector is the Ad.sTβRFc; TAd.sTβRFc is slightly weaker than Ad.sTβRFc, and the least effective is the Ad(E1−).sTβRFc. As all the three vectors produce nearly equal amounts of sTGFβRIIFc, Ad.sTβRFc is the most effective, which is perhaps due to its higher replication potential in the tumor cells. TAd.sTβRFc can also replicate in the tumor cells, but its replication potential is slightly lower than Ad.sTβRFc; and Ad(E1−).sTβRFc is replication-deficient. Based on these results, we believe that both viral replication and the sTGFβRIIFc expression have an important role in the inhibition of bone metastases.
In addition to understanding the role of viral replication and the inhibition of TGFβ signaling at the tumor–bone microenvironment, the future availability of a 4T1 bone metastasis model will also allow us to further explore the role of the adenoviral vector-induced innate and humoral immune responses,34, 35, 36 the role of TGFβ in suppressing the immune system14, 15, 37 and how that can be reversed by the oncolytic adenoviruses expressing sTGFβRIIFc. These questions can be addressed only in fully immunocompetent animal models.
In conclusion, our work described here shows that oncolytic adenoviruses targeting the TGFβ pathway can inhibit breast cancer bone metastases in a mouse mammary tumor model established in a syngeneic immunocompetent host and represents an important step in developing oncolytic adenoviruses for the treatment of breast cancer bone metastases. This animal model will now allow us to investigate the underlying molecular mechanism of action of the oncolytic adenoviruses, which may help in refining this method of treatment.
The research described here was funded by a grant from the National Cancer Institutes Grant #R01CA127380 (PS). We are thankful to Janardan Khandekar, Theodore Mazzone, Bruce Brockstein and an anonymous source for their generous support.
About this article
Calcified Tissue International (2014)