N.K. and D.D. contributed equally to this work
Cancer Cell Biology
Integrin-dependent response to laminin-511 regulates breast tumor cell invasion and metastasis
Article first published online: 5 MAY 2011
Copyright © 2011 UICC
International Journal of Cancer
Volume 130, Issue 3, pages 555–566, 1 February 2012
How to Cite
Kusuma, N., Denoyer, D., Eble, J. A., Redvers, R. P., Parker, B. S., Pelzer, R., Anderson, R. L. and Pouliot, N. (2012), Integrin-dependent response to laminin-511 regulates breast tumor cell invasion and metastasis. Int. J. Cancer, 130: 555–566. doi: 10.1002/ijc.26018
- Issue published online: 24 NOV 2011
- Article first published online: 5 MAY 2011
- Accepted manuscript online: 8 MAR 2011 10:43AM EST
- Manuscript Accepted: 9 FEB 2011
- Manuscript Received: 11 NOV 2010
- Australian National Health & Medical Research Council Research. Grant Number: 454564
- National Breast Cancer Foundation fellowship
- Deutsche Forschungsgemeinschaft
- breast cancer metastasis;
The basement membrane protein, laminin (LM)-511, is a potent adhesive and migratory substrate for metastatic breast tumor cells in vitro. Its expression correlates with tumor grade and metastatic potential in vivo. These observations suggest that responsiveness to autocrine or paracrine-derived LM-511 may be an important property regulating breast cancer metastasis in vivo. To address this, we compared the metastatic potential of 4T1 mammary carcinoma cells to that of 4T1 variants isolated by repeated chemotactic migration toward LM-511 in vitro (4T1LMF4) followed by serial injection into the mammary gland and recovery of spontaneous metastases from bone (4T1BM2). Variant subpopulations exhibited a distinct morphology on LM-511 and increased expression of β1 and β4 integrins compared to parental 4T1 cells. Importantly, mice inoculated with 4T1LMF4 and 4T1BM2 variants showed a 2.5- to 4-fold increase in the incidence of spontaneous metastasis to bone compared to 4T1 tumor-bearing mice. Functionally, 4T1BM2 variants were more adherent and more invasive toward LM-511 than parental 4T1 cells. Treatment of 4T1BM2 cells with lebein-1, a disintegrin with selectivity toward LM-type integrin receptors, potently inhibited their migration and invasion toward LM-511. Similarly, α3β1 integrin-dependent migration and invasion of human MDA-MB-231 breast carcinoma cells toward LM-511 were significantly inhibited by lebein-1. Taken together, these results provide strong evidence that LM-511 contributes to the metastasis of breast tumors and suggest that targeting integrin-LM-511 interactions with lebein-1 or other inhibitors of LM-511 receptors may have therapeutic potential for patients with advanced breast cancer.
Metastasis from primary breast tumors is the leading cause of cancer related death in women. In these patients, metastases develop most commonly in bone, lung, liver and brain. In particular, bone metastasis that occurs in the majority of patients with advanced breast cancer inevitably leads to various debilitating skeletal complications.1 Yet despite significant advances, the mechanisms involved in the development of bone or soft tissue metastases remain elusive. The current evidence indicates that organ-specific tropism is regulated in part by stromal factors (chemokines, growth factors and extracellular matrix proteins) produced at metastatic sites.2, 3 In turn, metastatic colonization depends on intrinsic properties of tumor cells and on the acquisition of specific attributes that allow them to respond appropriately to the signals emanating from the metastatic microenvironment. Thus, complex interactions between intrinsic and stromal-derived signals cooperate to facilitate homing, survival and growth of metastatic breast tumors in distant organs.4–6
However, not all cells within a given tumor have the same ability to respond to external signals and to metastasize. The concepts of tumor heterogeneity and rare metastatic variants are now firmly established7, 8 and are best exemplified by the heterogeneity in metastatic ability of genetically related tumor lines of the 4T1 mammary cancer metastasis model.9 In this model, several tumor lines were derived from the same spontaneous mammary tumor and found to be either non-metastatic (67NR), weakly metastatic to lung (66cl4) or metastatic to multiple sites (4T1).9 Further work from our laboratory led to the isolation of clonal sublines derived from 4T1 cells (4T1.2 and 4T1.13) that are significantly more metastatic to soft tissues and bone than 67NR, 66cl4 or parental 4T1 tumors from which they were derived.10, 11 Gene expression profiling of weakly metastatic and bone metastatic tumors of the model revealed a high proportion of extracellular matrix-related genes expressed at higher levels in aggressive bone metastatic tumors compared to weakly or non-metastatic tumors.10 In particular, the expression of the basement membrane protein laminin (LM)-511 (previously LM-10) was found to correlate with metastatic potential in this mouse model, with bone metastatic tumors showing widespread expression in both primary tumors and in metastases.12 These findings are consistent with elevated epithelial and vascular expression of LM-511 often observed in advanced human breast tumors and metastases.12, 13 Functionally, metastatic mouse and human breast tumor lines exhibit greater adhesion and motility on this substrate than non-metastatic lines.12 These findings led us to hypothesize that the ability to respond to LM-511, either through autocrine or paracrine mechanisms, may contribute to the acquisition of a more aggressive and metastatic phenotype.
To address the relevance of these findings in vivo and to provide further evidence supporting the role of LM-511 in breast cancer progression, we asked whether selection of tumor cell subsets based on their rapid migratory response to LM-511 in vitro would be sufficient to enrich for more aggressive tumor variants in vivo. We report that LM-511-dependent selection significantly enriches for tumor variants with increased bone metastatic potential. Functional characterization of these variants indicates that integrin-dependent interactions of metastatic breast tumors with LM-511 are likely to contribute to their invasion and metastasis and suggests that interfering with these interactions may provide a useful strategy to prevent or delay the development of metastatic lesions in patients with advanced breast cancer.
Material and Methods
Cell culture and reagents
The 4T1 mouse mammary tumor line was obtained from Dr. F. Miller (Karmanos Cancer Institute, Detroit, MI) and maintained in α-minimal essential medium (α-MEM) supplemented with 5% fetal calf serum (FCS) and 1% penicillin-streptomycin (Invitrogen Australia Pty, Victoria, Australia). MDA-MB-231 cells were cultured in α-MEM, 10% FCS, sodium pyruvate (1 mM), glutamine (2 mM) and 1% penicillin-streptomycin. Unless otherwise indicated, all in vitro assays were performed in serum-free medium consisting of α-MEM supplemented with bovine serum albumin (BSA, 0.05% w/v), L-glutamate (2 mM) sodium pyruvate (1 mM) and 1% penicillin-streptomycin. Recombinant human LM-511 was purified as described previously.14 Lebein-1 was isolated from crude vipera lebetina venom as described previously.15 Function-blocking antibodies directed against the α3 integrin subunit (P1B5) were purchased from Millipore Australia (NSW, Australia) and used at 10 μg/ml.
Establishment of 4T1 tumor variants
Chemotactic assays were performed in Transwell migration chambers (Corning, Life sciences, Acton, MA) as described previously,16 with minor modifications. Briefly, parental 4T1 tumor cells (2 × 105/100 μl/well) were seeded into the upper well in serum-free medium and allowed to migrate toward soluble recombinant human LM-511 (5 μg/ml) in the lower chamber for 3 hr at 37°C. Migrated cells were recovered from the lower wells, expanded in culture and the selection process repeated three times. The resulting variant population was designated 4T1LMF4.
After in vitro selection, aggressive bone metastatic variants were derived from 4T1LMF4 cells using an in vivo spontaneous metastasis assay (see below and Ref.10). The mice were culled 25–28 days after inoculation of 4T1LMF4 cells into the mammary gland, and bone macrometastases were extracted from femurs by mincing with a scalpel blade followed by digestion with collagenase type II (Invitrogen) for 1 hr at 37°C. Recovered tumor cells were expanded in culture, subjected to one more round of selection in vivo and designated 4T1BM2.
Spontaneous metastasis assay and histology
All mice used were 6–8 week old female Balb/c (Walter & Eliza Hall Institute, Melbourne, Australia) and were maintained in a specific pathogen-free environment and fed ad libitum. All procedures involving mice conformed to National Health and Medical Research Council animal ethics guidelines. The 4T1 spontaneous metastasis model has been described in detail previously.10, 11 The mice were sacrificed on day 25–28 and primary tumors weighed. Lungs, spines and femurs were fixed in 10% buffered formalin, processed for paraffin embedding, sectioned and stained with H&E. Bones were decalcified in 5% nitric acid overnight before embedding. The presence of lung and bone metastases was confirmed by microscopic observations of three sections per bone/lung (100 μm apart) and H&E images photographed using an Olympus BX-51 light microscope.
Parental 4T1 and its variants, 4T1LMF4 and 4T1BM2, were seeded in the presence of 0.5% serum in culture flasks precoated with LM-511 (2 μg/ml). After 6 hr incubation at 37°C, adherent cells were rinsed twice with PBS and serum-starved overnight in serum-free medium. The cells were detached in PBS containing 0.01% EDTA and cell surface integrin expression analyzed by flow cytometry on a Canto flow cytometer (Becton Dickinson, San Jose, CA) as described previously.16 Anti-integrin primary antibodies used were α2β1 (BMA2.1, 5 μg/ml, Chemicon), α3 (clone 42, 5 μg/ml; BD Pharmingen), α6 (NKI-GoH3, 5 μg/ml; Chemicon), α7 (334908, 5 μg/ml, R&D Systems), αv (RMV-7, 2.5 μg/ml; BD Pharmingen), β1 (MB1.2, 5 μg/ml; Chemicon), β3 (2C9.G2, 2.5 μg/ml; BD Pharmingen), β4 (346-11A, 5 μg/ml; BD Pharmingen) and β6 (268622, 2.5 μg/ml; R&D Systems). Control isotype-matched rat IgG1 (Jackson ImmunoResearch, Australia) and IgG2a (R35-95, BD Pharmingen), hamster IgG1 (A19-3, BD Pharmingen) and mouse IgG1 and IgG2a (1B5 and ID4.5, respectively, kindly provided by Dr. P. Simmons, University of Texas Health Science Center, USA) were all used at 10 μg/ml.
Protein extracts from cells cultured as described for flow cytometric analysis were separated by SDS-PAGE under nonreducing conditions and integrin expression analyzed by standard immunoblotting as previously described.17 Integrin antibodies used were β1 (2 μg/ml, clone MB1.2, Millipore) and β4 (10 μg/ml, clone 346-11A, BD Pharmingen). Blots were developed using the enhanced chemoluminescence kit (ECL, Amersham Biosciences), according to the manufacturer's instructions. Integrin expression levels were normalized against actin levels and quantified using Image J software (NIH).
Morphological examination and actin staining
4T1 cells and 4T1LMF4 and 4T1BM2 variants (1 × 105) were seeded in serum-free medium in 24-well plates precoated with LM-511 (2 μg/ml) or type IV collagen (20 μg/ml) and photographed after 2 hr on a Leica DM IRB inverted microscope. F-actin staining was performed on cells cultured under the same conditions in glass chamber slides for 2 hr, fixed 1 hr in 10% buffered formalin and permeabilized for 5 min with 0.1% Triton X-100 prior to standard staining with rhodamine-phalloidin (Invitrogen) and nuclear counter staining with OliGreen fluorescent stain (Invitrogen) and photographed after 2 hr on an Olympus BX-51 fluorescence microscope.
Short-term (30 min) adhesion assays were performed in the absence of serum using a calcein-labeling method as described previously.12 The adhesion assay was completed three times and the data are presented as the means ± standard deviation of a representative experiment performed in triplicate wells.
Cell proliferation assay
In vitro cell proliferation assays were performed using the Sulforhodamine B (SRB) colorimetric assay.16 Proliferation of 4T1 variants was measured over 5 days in complete α-MEM medium with an initial cell density of 500/well and five replicate wells per time point. Proliferation assays were completed three times and the data are presented as the means ± standard deviation of a representative experiment with five replicate wells.
Migration and invasion assays
Migration and invasion assays were performed in triplicate Transwell migration chambers (8-μm pore size, Corning) as described previously.12, 16 Cells were allowed to migrate for 3 hr (4T1BM2) or 4 hr (4T1.2 and MDA-MB-231) at 37°C. Invasion assays were stopped after 18 hr (4T1BM2 cells) or 24 hr (MDA-MB-231 cells). For inhibition of migration and invasion, the cells were preincubated for 30 min on ice in serum-free medium containing lebein-1 (50 and 200 nM) before seeding together into the assay wells. The results from a representative experiment (n = 3) are shown and expressed as the mean number of cells/field ± standard deviation of nine fields of view per condition.
Isolation of aggressive metastatic breast tumor variants by LM-511-dependent chemotaxis in vitro and in vivo selection
We and others have reported that breast cancer progression and metastasis are associated with increased vascular and epithelial expression of LM-511.12, 13 Moreover, our earlier study using a syngeneic model of spontaneous breast cancer metastasis demonstrated that aggressive bone-metastatic 4T1.2 mammary carcinoma cells adhere and migrate more effectively than weakly (66cl4) or nonmetastatic (67NR) lines in response to LM-511.12 Additionally, when 4T1.2 cells were tested for their ability to migrate toward various extracellular matrix proteins, LM-511 was the most potent substrate promoting haptotactic migration, even at a concentration as low as 2 μg/ml (Fig. 1a). Collectively, these observations suggest that enhanced ability to respond to LM-511, through autocrine or paracrine mechanisms, is likely to contribute to the aggressive phenotype of 4T1.2 tumors.
To further demonstrate the relevance of tumor cell interactions with LM-511 to metastatic potential in vivo, we tested whether selection of cells that migrate rapidly toward LM-511 in vitro would be sufficient to enrich for tumor variants with enhanced metastatic potential. Parental 4T1 mammary carcinoma cells used for these experiments were chosen for their heterogeneity and weak bone metastatic capacity compared to 4T1.2 and 4T1.13 clonal sublines.10, 11 The selection protocol involved four rounds of in vitro chemotactic migration toward soluble LM-511 (Fig. 1b). Cells seeded in the upper wells of Transwell chambers were allowed to migrate for 3 hr and cells that had migrated to the bottom wells were recovered and expanded in culture. This process was repeated three times to generate the 4T1LMF4 variant population. 4T1LMF4 cells were further selected in vivo by inoculation into the mammary fatpad and recovery of metastatic nodules. A bone metastasis was chosen as the incidence of these was visibly higher in 4T1LMF4 tumor-bearing mice than in the 4T1 group whereas both groups of mice developed lung metastasis with similar frequency (see below). The in vivo selection was repeated and the resulting variant population called 4T1BM2 (Fig. 1b).
Consistent with a differential responsiveness to LM-511, parental 4T1 and variant populations showed distinct morphologies when cultured on a LM-511 matrix (Fig. 1c). Whereas parental 4T1 cells had a predominantly spread morphology, 4T1BM2 cells were mostly rounded. 4T1LMF4 showed a mixed population of spread and round cells consistent with a partial enrichment for a small subpopulation from parental 4T1 cells. In contrast to their differential response to LM-511, parental and variant populations displayed the same rounded morphology on a collagen-IV matrix (Fig. 1c) indicating that morphological changes are specific to LM-511. This was further confirmed by staining of the actin cytoskeleton (Fig. 1d). Parental 4T1 cells formed numerous actin fibers on LM-511 whereas 4T1LMF4 and 4T1BM2 formed predominantly cortical actin. All populations showed the same cortical actin distribution when cultured on collagen-IV. Taken together, these observations support the concept that the selection protocol enriches for a small subpopulation of cells with a distinct responsiveness to LM-511 compared to that of parental 4T1 cells from which they were derived.
LM-511 responsiveness enhances metastatic potential in vivo
Tumor growth and spontaneous metastatic ability of parental 4T1 and the 4T1LMF4 and 4T1BM2 variants were compared after orthotopic inoculation into Balb/c mice (Fig. 2). The 4T1LMF4 and 4T1BM2 tumors grew slightly smaller than parental 4T1 tumors (p < 0.01 and p < 0.05, respectively) (Fig. 2a). Smaller tumor size in the variant populations was reflected by a small but significant reduction in primary tumor weight at harvest compared to parental 4T1 tumors (p < 0.05) (Fig. 2b). As expected, all groups consistently developed lung metastases (Fig. 2c). However, the number and size of metastatic nodules in lung were visibly greater in variants than in parental 4T1 cells (Fig. 2c). The presence of bone macrometastases was also clearly evident by histology in mice inoculated with 4T1LMF4 (not shown) and 4T1BM2 cells (Fig. 2c, bottom right panel). Importantly, quantitation of the incidence of mice developing bone metastases revealed that despite the smaller size of 4T1LMF4 and 4T1BM2 primary tumors compared to parental 4T1 tumors, the selection process resulted in an enrichment for bone metastatic cells (Fig. 2d). Although the size of lesions (data not shown) and incidence of mice with bone metastases detected in the parental 4T1 group was low (21%, n = 24), the number of mice with bone macrometastases increased significantly in the LM-511-selected 4T1LMF4 variant group (55%, n = 20) and further increased in mice inoculated with the in vivo selected bone metastatic 4T1BM2 variant (85%, n = 20). Taken together, the results implicate LM-511 in the metastasis of breast tumors and indicate that the ability to respond efficiently to LM-511 confers a greater metastatic potential to multiple organs, particularly to bone.
Functional characterization of 4T1 variants in vitro
To explore the potential mechanisms underlying the increase in metastatic capacity of 4T1 variants, we initially compared their functional characteristics to that of parental 4T1 cells in vitro (Fig. 3). Parental 4T1 cells and variants proliferated at the same rate in response to serum stimulation in vitro (Fig. 3a). This observation and the smaller size of 4T1LMF4 and 4T1BM2 tumors (Fig. 2a) suggest that tumor growth in vivo is likely to be subject to more complex regulation and that changes in proliferative capacity are unlikely to account for the enhanced metastatic potential of 4T1 variants (Figs. 2c and 2d). In contrast, while parental 4T1 cells adhered well to LM-511 (∼ 60%), adhesion of 4T1LMF4 and 4T1BM2 cells to LM-511 was increased further in this short-term assay, resulting in ∼ 80% of 4T1BM2 cells adhering within 30 min (p < 0.01) (Fig. 3b). Adhesion of all cell populations to control BSA was negligible (<10%). We reported previously that LM-511 is a potent promigratory/invasive factor for aggressive bone metastatic breast tumor cells.12 Therefore, we used standard Matrigel invasion assays to test whether 4T1 variants were more invasive than parental 4T1 cells when LM-511 was used as chemoattractant (Fig. 3c). As expected, no significant invasion was observed in the absence of LM-511. Importantly, 4T1 variants showed a gradual increase in invasion toward LM-511, with almost a 2-fold increase for 4T1BM2 compared to parental 4T1 cells (P < 0.005) (Fig. 3c).
Because changes in morphology, actin cytoskeleton and migratory/invasive properties of tumor cells are intimately linked to adhesion receptors, alterations in the expression of integrins were investigated. Flow cytometric analysis revealed a small but reproducible increase in β1 integrin subunit after in vitro selection with LM-511 (4T1LMF4 and 4T1BM2) (Fig. 3d, top panels), whereas 4T1BM2 cells also displayed a substantial increase in β4 integrin expression (Fig. 3d bottom panels). These changes in integrin expression were further confirmed by immunoblot analysis showing a 1.78-fold and 1.90-fold increase in β1 expression in 4T1LMF4 and 4T1BM2 variants, respectively, and a 1.91-fold increase in β4 expression in 4T1BM2 cells (Fig. 3e). No significant changes were observed in the expression of α2 (low), α3 and αv (intermediate) and α5, α6 and β3 integrins (high) and none of the populations expressed α7 integrin (data not shown).
The disintegrin lebein-1 inhibits metastatic breast tumor cell migration and invasion toward LM-511
The results above indicate that 4T1LMF4 and 4T1BM2 variants contain a subpopulation of cells derived from parental 4T1 that express higher levels of LM-511 receptors and respond more efficiently to this substrate. Thus, LM-511 may contribute functionally to metastasis in part by promoting integrin-dependent adhesion and invasion of breast tumor cells. Multiple integrins bind LM-511 including α3β1, α6β1 and α6β4.18–20 Our previous investigation using function-blocking antibodies demonstrated that integrin α3β1 is the primary LM-511 receptor mediating MDA-MB-231 cell migration.12 A similar analysis in the highly metastatic murine 4T1.2 clonal line showed that β1-type receptors are involved in promoting LM-511-dependent migration and invasion but could not define a specific role for α3β1 integrin due to the lack of function-blocking antibodies against the murine α3 integrin subunit.12
As an alternative we made use of lebein-1, a heterodimeric disintegrin derived from snake venom Vipera lebetina.15 Lebein-1 has the unusual characteristic of being selective for LM-type integrins, inhibiting α3β1, α6β1 and α7β1 integrins (but not α1β1 and α2β1 collagen receptors) in an RGD-independent manner.15, 21 However, its effect on LM-511-mediated migration and invasion has never been tested.
As shown in Figure 4, treatment of 4T1BM2 cells with lebein-1 potently and dose-dependently inhibited migration (Fig. 4a) and Matrigel invasion (Fig. 4b) in response to LM-511. At 200 nM, lebein-1 inhibited migration and invasion by greater than 70% and 85%, respectively. The inhibitory effect of lebein-1 on 4T1BM2 migration and invasion was not due to toxicity because the cells exhibited the same colony forming efficiency when treated with lebein-1 at 200 nM (69.2% ± 5.2) compared to control PBS-treated cells (77.2% ± 3.8) (Fig. 4c). In separate experiments, we also confirmed that lebein-1 (200 nM) partially inhibited the migration and invasion of parental 4T1 (Fig. 4d) and 4T1LMF4 variant (Fig. 4e) from which 4T1BM2 cells were derived.
To gain some insight into the specific LM-511 receptor(s) targeted by lebein-1, its inhibitory effect on invasion/migration was also tested against the metastatic human MDA-MB-231 breast tumor line known to attach and migrate on LM-511 primarily via α3β1 integrin.12 Consistent with our previous study,12 LM-511-induced migration of MDA-MB-231 cells was potently inhibited by anti-α3 (˜85% inhibition) but not by isotype control antibodies (Fig. 5a). Importantly, lebein-1 also significantly inhibited LM-511-dependent migration of MDA-MB-231 cells, resulting in greater than 60 and 80% inhibition when used at 50 and 200 nM, respectively. Matrigel invasion was inhibited partially by anti-α3 antibodies (˜45% inhibition) but not by a control isotype antibody. Similarly, lebein-1 mediated partial inhibition of invasion in this assay when used at 200 nM (˜33% inhibition) but not at lower concentrations (Fig. 5b). Taken together, these results indicate lebein-1 efficiently inhibits the migration and invasion responses of MDA-MB-231 cells to LM-511 and that these responses are mediated primarily via α3β1 integrin.
Increased vascular and epithelial expression of LM-511 correlates with cancer progression and metastasis in human and murine mammary tumors.12, 13 Moreover, highly metastatic breast cancer lines adhere and migrate better than nonmetastatic tumor lines on LM-511 in vitro, leading us to propose that interaction with LM-511 through either autocrine or paracrine mechanisms may be an important property contributing to breast cancer progression. Here, we sought to address the relevance of these findings to the metastatic properties of breast tumors in vivo. The isolation of highly metastatic variants from parental 4T1 cells described herein relied in part on the ability of tumor cells to migrate rapidly toward LM-511. The significant enhancement in metastatic ability of 4T1LMF4 and 4T1BM2 variants, particularly to bone, compared to parental 4T1 cells is consistent with the reported metastatic heterogeneity of breast tumors7, 22 and further supports the role of LM-511 in metastatic dissemination.
However, responsiveness to LM-511 is unlikely to be sufficient alone to dictate specificity toward bone because metastasis to other sites, notably to lung where LM-511 is abundant,23, 24 was also clearly increased in mice inoculated with 4T1LMF4 or 4T1BM2 (see Fig. 2c). Thus, the expression pattern of LM-511 in breast tumors and metastases12, 13 and modulation of metastatic abilities demonstrated in our study defines LM-511 as a “metastasis progression gene” or as a “non-site-specific modulator of metastasis.”4, 6 The ability to respond to bone-derived factors is likely to be a more critical determinant regulating bone-specific tropism.2, 3 Nevertheless, our finding that LM-511 selection in vitro was sufficient alone to increase the incidence of spontaneous metastasis from the mammary gland to bone by greater than 2.5-fold (54%) is particularly remarkable. Other groups who have reported on the isolation of human25, 26 and murine27 bone metastatic variants typically required multiple rounds of selection in vivo and direct inoculation of tumor cells into the vasculature for variants to metastasize consistently to bone. In the murine study, 4T1 variants derived from repeated intravenous injections and recovery from bone led to an increase from 14% (parental 4T1) to 77% (4T1E/M3 variants) in the incidence of mice developing bone metastases.27 However, in the same study the incidence of 4T1E/M3 spontaneous bone metastases was significantly lower (20%).
Interestingly, another study on the isolation of bone metastatic 4T1 variants using in vivo selection alone reported an incidence of spontaneous bone metastasis as high as 47% in parental 4T1 tumors.28 The reason for the discrepancy between their finding and the relatively low incidence of spontaneous bone metastasis detected in mice inoculated with parental 4T1 cells in our study and that of Takahashi et al.27 is unclear but may be due to differences in the methodology used and time allowed for the development of bone metastases. Unlike our model where bone macrometastases were clearly visible as early as 28 days after tumor cell inoculation, the model developed by Rose et al. required removal of the primary tumor and 44–49 days for the detection of bone metastases in parental and variant 4T1 populations. Thus, in vitro selection using LM-511 could provide a simple and more rapid approach to isolate aggressive metastatic variants. Used in combination with standard in vivo selection, this methodology may be particularly useful for the isolation of site-specific metastatic variants, particularly to sites of metastasis where current animal models of spontaneous metastasis lack robustness, such as bone, liver or brain. The isolation of metastatic variants from other human or mouse tumor models, particularly for tumors where LM-511 has been implicated (breast, lung and prostate cancer), would be supportive of our findings in the 4T1 breast cancer model and provide a more rigorous proof of the role of LM-511 in the metastatic process.
Whether the selection of metastatic breast tumor variants is a property specific to LM-511 or is shared by other LM isoforms remains unclear. Recently, Hughes et al.29 described a similar strategy for the isolation of isogenic sublines of Hs578T human breast cancer cells based on their ability to invade LM-111-rich Matrigel matrix in vitro. Although the metastatic ability of these variants to bone and other sites was not investigated in this study, the authors found that Hs578Ts(i)8, derived from eight sequential passages through Matrigel, were more tumorigenic and invasive in vivo than the parental line. However, the clinical relevance of findings from in vitro studies of tumor-matrix interactions and the purported significance of LM should be evaluated against which specific LM isoforms are present in the tumor microenvironment and at metastatic sites. We have shown recently that aggressive bone metastatic 4T1.2 and 4T1.13 mouse mammary carcinoma lines express significantly higher levels of LM-511 than the genetically related nonmetastatic 67NR and weakly metastatic 66cl4 lines. None of these tumors express detectable LM-111 or LM-332 in vivo.12 These observations are consistent with immunohistochemical findings in advanced human breast tumors where LM-511 expression is often elevated whereas LM-111 and −332 are down-regulated.12, 13, 30–33 Moreover, in contrast to LM-111 and LM-332, LM-511 is abundant in the vasculature (including the sinusoids of the bone) where it could facilitate extravasation and establishment of metastatic foci. Thus, although we cannot completely rule out a similar role for other LM isoforms, the pattern of LM isoform expression in vivo suggests that LM-511 may be a more relevant substrate driving the metastasis of breast tumors to bone.
Up-regulation of the β4 integrin subunit in 4T1BM2 cells isolated from a bone lesion suggests a particular role for α6β4 receptor (the only known β4 integrin dimer) in regulating metastasis to bone. Although we have yet to demonstrate a direct functional contribution for integrin α6β4 in promoting bone metastasis through interaction with LM-511, its role is supported by several lines of evidence. Integrin α6β4 contributes to the adhesion and migration of several tumor cell types on LM-511 in vitro.17, 18 In agreement with our data, the expression of α6 and/or β4 integrin subunits is often elevated in bone metastatic tumor cells isolated from the marrow of patients with breast cancer34 and enhanced α6β4 expression is associated with a poor prognosis in patients with breast cancer, particularly for those with LM-producing breast tumors.35 More recently, α6β4 expression was shown to associate most significantly with the aggressive “basal-like” subtype of breast tumors.36 Although its precise role in vivo has not been completely characterized, α6β4 has been proposed to facilitate survival of tumor cells at the primary and metastatic sites.37, 38 Additionally, α6β4 may play a role in homing of breast tumor cells to bone because exogenous expression of α6β4 in bone metastatic MDA-MB-435 cells leads to demethylation and re-expression of genes regulating cell motility, invasion and metastasis.39 Conversely, down-regulation of α6 or β4 integrin subunits by siRNA decreases MDA-MB-231 cell migration and invasion.37 Together, these studies indicate that in addition to its prognostic value, α6β4 may represent a relevant target for therapeutic interventions in patients with advanced breast cancer. Interfering with LM-511/α6β4 integrin interactions would be expected to inhibit several steps required for successful metastasis to bone. This is currently being tested in vivo using the 4T1BM2 model.
Increased expression of β1 integrin subunits in 4TLMF4 and 4T1BM2 variants is also consistent with the importance and potential of LM receptors as antimetastatic targets.40 In particular, a role for integrin α3β1 in metastasis to lung has been proposed based on immunofluorescence detection of LM-332 subunits in exposed regions of pulmonary vascular basement membranes after direct intravenous inoculation of various human tumor cell lines in rats and inhibition of tumor cell arrest in those regions after treatment with blocking antibodies against subunits of LM-332 or integrin α3β1.41 Antibodies against either subunit of α3β1 integrin, the primary receptor for LM-332, similarly inhibited experimental metastasis of MDA-MB-231 breast tumor cells to lung in mice.41 These findings indicate that, at least in some breast tumors, LM-332 may contribute in part to their homing and establishment in lung via interactions with α3β1 integrin receptors. It should be noted, however, that α3β1 integrin is also the primary receptor for LM-511 in many tumor cell types including breast (see Fig. 5 and Refs.12, 18). Although its role was not investigated by Wang et al.,41 LM-511 would also be expected to be abundant in vascular BMs of the lung and to bind α3β1 integrin with high affinity in metastatic breast tumor cells.12
Engagement of α3β1 integrin upon binding to its substrates could directly facilitate tumor cell invasion into metastatic sites as demonstrated for extravasation and invasion of MDA-MB-231 breast cancer cells into the lung parenchyma.42 The potent invasive and migratory responses induced by LM-511 in 4T1BM2 cells and their inhibition by lebein-1 (Figs. 4a and 4b) are consistent with this possibility. Collectively, results from the above studies and the extensive metastasis of 4T1 tumors to lung (see Fig. 2c) with a concomitant increase in β1 integrin expression in 4T1 variants raise the possibility that LM-332 and LM-511 may have overlapping functions in promoting metastasis of breast tumors to lung (and possibly other sites) via α3β1 integrin. Another interesting possibility is that autocrine-derived LM-332/511 and LM-332/511 produced at metastatic sites may have different roles in dictating organ-specific metastasis. Further work will be required to address this possibility.
Lebein-1 binds multiple LM receptors including integrin α3β1, α6β1 and α7β1.15 The absence of α7β1 receptors in 4T1-derived lines (data not shown), however, excludes a role for this receptor in mediating migration of 4T1BM2 cells to LM-511. In contrast, almost complete inhibition of LM-511-dependent migration and invasion of 4T1BM2 (Fig. 4a) and migration of MDA-MB-231 cells (Fig. 5a) by lebein-1 or anti-α3 integrin antibodies indicate that lebein-1 inhibits these responses primarily by blocking α3β1 integrin. To our knowledge, this is the first demonstration that lebein-1 can impair the ability of metastatic breast tumor cells to respond efficiently to LM-511. However, a role for other receptors in regulating breast tumor cell invasion cannot be excluded because anti-α3 antibodies only partially (˜45%) inhibited MDA-MB-231 invasion. Collectively, these results suggest that interfering with LM-511/α3β1 integrin interactions may provide a useful strategy to prevent or delay the metastatic spread of breast and possibly other tumor types. In support of this, suppression of α3β1 integrin with short hairpin RNAs was reported to inhibit MDA-MB-231 invasion, tumorigenesis and cross-talk with endothelial cells.43 The strong inhibitory effect of lebein-1 seen in our in vitro study and the recent observation that lebein-1 can interfere with the attachment of HepG2 hepatocellular carcinoma cells to sinusoids in an experimental model of liver metastasis44 warrant further in vivo investigations on the efficacy of lebein-1 in clinically relevant models of spontaneous breast cancer metastasis such as the 4T1BM2 model.
This work was supported by an Australian National Health & Medical Research Council Research Grant #454564 to N.P. and R.L.A. and National Breast Cancer Foundation fellowship to R.L.A. The authors wish to thank Ralph Rossi for advice and support with flow cytometry analyses and Laura Kirby for technical assistance with immunoblots. We are grateful to BioStratum for providing the LM-511-overexpressing clone. JAE was supported by Deutsche Forschungsgemeinschaft through grant SFB/TR23 project A8.