Activation of tumor cell integrin α_vβ₃ controls angiogenesis and metastatic growth in the brain

Abstract

The incidence of brain metastasis is rising and poses a severe clinical problem, as we lack effective therapies and knowledge of mechanisms that control metastatic growth in the brain. Here we demonstrate a crucial role for high-affinity tumor cell integrin α_vβ₃ in brain metastatic growth and recruitment of blood vessels. Although α_vβ₃ is frequently up-regulated in primary brain tumors and metastatic lesions of brain homing cancers, we show that it is the α_vβ₃ activation state that is critical for brain lesion growth. Activated, but not non-activated, tumor cell α_vβ₃ supports efficient brain metastatic growth through continuous up-regulation of vascular endothelial growth factor (VEGF) protein under normoxic conditions. In metastatic brain lesions carrying activated α_vβ₃, VEGF expression is controlled at the post-transcriptional level and involves phosphorylation and inhibition of translational respressor 4E-binding protein (4E-BP1). In contrast, tumor cells with non-activated α_vβ₃ depend on hypoxia for VEGF induction, resulting in reduced angiogenesis, tumor cell apoptosis, and inefficient intracranial growth. Importantly, the microenvironment critically influences the effects that activated tumor cell α_vβ₃ exerts on tumor cell growth. Although it strongly promoted intracranial growth, the activation state of the receptor did not influence tumor growth in the mammary fat pad as a primary site. Thus, we identified a mechanism by which metastatic cells thrive in the brain microenvironment and use the high-affinity form of an adhesion receptor to grow and secure host support for proliferation. Targeting this molecular mechanism could prove valuable for the inhibition of brain metastasis.

Brain metastases are diagnosed in 10% to 40% of patients with progressing cancer, and the incidence is rising as patients live longer and extracranial metastases respond to improved treatments. However, brain metastases still cannot be treated effectively, and mechanisms controlling brain metastatic growth are largely unknown (1–3).

Here, we demonstrate that the high-affinity state of tumor cell adhesion receptor integrin α_vβ₃ critically promotes metastatic growth and recruitment of supporting blood vessels within the brain microenvironment. Integrins are cell surface receptors composed of non–covalently linked α and β subunits that mediate cell–matrix and cell–cell interactions and transduce signals that have impacts on cell survival, proliferation, adhesion, migration, and invasion. Integrin signals can also originate inside cells, affect receptor affinity, and thereby control ligand binding, cross talk with other receptors, and alter cell adhesion and proliferation (4–6). Integrin α_vβ₃ also plays a role on sprouting endothelial cells and contributes to angiogenesis (7). In several tumor types, including glioma, breast cancer, and melanoma, expression of α_vβ₃ supports invasion and metastasis (8–11). Notably, these tumors either originate in the brain or frequently spread to the brain. Because of this correlation and our previous findings that expression of α_vβ₃ by tumor cells, particularly in its high-affinity form, promotes metastatic dissemination (12), we asked whether α_vβ₃ and the activation state of the receptor also play a role in metastatic growth within the brain microenvironment. To address this question directly, we generated and validated a human tumor cell model in which variants share a genetic background and express integrin α_vβ₃ either in a constitutively activated or non-activated functional form. Our results show that tumor cell α_vβ₃ activation promotes tumor cell growth in the brain but not in the mammary fat pad (MFP) as a primary site, pointing to a critical influence of the tissue environment on the ability of high-affinity α_vβ₃ to enhance tumor growth. Our study suggests that the mechanism through which activated α_vβ₃ supports brain metastatic growth is based on elevated expression of vascular endothelial growth factor (VEGF) because of inhibition of translational repressor 4E-BP1, resulting in efficient tumor angiogenesis under normoxic conditions. This function prevents development of hypoxia, associated tumor cell apoptosis, and retardation of lesion growth. Thus, as an initiator of this process, activated integrin α_vβ₃ could potentially serve as a target against brain metastasis.

Results

Tumor Cell Integrin α_vβ₃ Activation Enhances Metastatic Growth in the Brain.

We chose MDA-MB-435 human cancer cells for our analysis because they are highly metastatic, can spread to the brain, and express α_vβ₃, and because the activation state of the receptor strongly affects metastasis to extracranial sites (12). To generate cell variants that express α_vβ₃ in defined states of activation, endogenous α_vβ₃ was knocked-down, targeting the 3′UTR of the β₃ subunit gene and then reconstituted either with non-activated β₃WT or constitutively activated mutant β₃D723R (12, 13). In MDA-MB-435 cells, α_vβ₃ is the only β₃ integrin. Thus, targeting β₃ by stable transduction with lentiviral shRNA resulted in 80% reduction of total β₃ protein (Western blot, not shown) and 70% reduction in α_vβ₃ surface expression (Fig. 1A), without affecting other integrins. This was confirmed by fluorescence activated cell sorter (FACS) analysis of a panel of integrins, including α_vβ₅ and α_vβ₆ (not shown). Importantly, β₃ down-regulation remained stable in tumor xenografts [supporting information (SI) Fig. S1]. The cDNA constructs used for reconstitution with β₃WT or β₃D723R lack 3′-UTRs and are therefore not affected by the β₃ shRNA. α_vβ₃ surface expression in β₃D723R cells resembled that of the parental cells, whereas β₃WT cells expressed twice as much α_vβ₃ (Fig. 1A). Notably, α_vβ₃ activation by mutant β₃D723R was stable after tumor cell implantation into the brain or MFP.

Fig. 1.

Impact of tumor cell integrin α_vβ₃ expression and activation on lesion growth in the brain and mammary fat pad (MFP). (A) Analysis of integrin α_vβ₃ expression by flow cytometry. Parent, MDA-MB-435 transduced with control shRNA ScrB; β₃kd, MDA-MB-435 transduced with β₃ shRNA; WT, MDA-MB-435/β₃kd reconstituted with wild-type β₃ (non-activated α_vβ₃); D723R, MDA-MB-435/β₃kd reconstituted with β₃D723R (activated α_vβ₃). (B) Lesion growth in the brain (days 0–21) or MFP (days 0–34) based on bioluminescence signal of F-luc–tagged tumor cells by non-invasive in vivo imaging. (C) Lesion growth in the brain (days 0–21) or MFP (days 0–36) of MDA-MB 435/β₃- cells selected from parental cells with saporin-anti-β₃ antibody and reconstituted by transfection with β₃WT of β₃D723R based on in vivo imaging (brain) or caliper measurement (MFP). Each dot represents one animal and red lines the mean values per group; group sizes in (B): brain: parent and β₃kd, n = 6; β₃WT and β₃D723R, n = 11; MFP: parent, n = 10; β₃kd, n = 8; β₃ and β₃D723R, n = 6; (C), all groups, n = 7. P values obtained by a two-tailed Student's t test with unequal variance.

When implanted into the forebrain (striatum), tumor cells expressing activated β₃D723R exhibited a significant growth advantage over β₃kd cells, cells expressing non-activated β₃WT, and parental cells transduced with scrambled shRNA used as a control (Figs. 1B, 2A). In contrast, in the MFP, mimicking a primary tumor site, expression of activated β₃D723R did not accelerate tumor growth, which was generally slower than in the brain (Fig. 1B). Notably, β₃WT and β₃D723R cells displayed identical growth rates in vitro (Fig. S2A).

Fig. 2.

Tumor cell integrin α_vβ₃ activation enhances proliferation in the brain. (A) Non-invasive bioluminescence imaging 21 days after intracranial implantation of 1 × 10⁴ β₃WT- (Left) or β₃D723R- (Right) expressing MDA-MB-435 tumor cells. (B) Growth of β₃WT and β₃D723R-expressing tumor cells in the brain. Each dot represents one animal; horizontal lines represent mean values per group. (C and D) Intracranial proliferation of β₃WT- and β₃D723R-expressing cells in brain lesions by in vivo BrdU uptake; n = 5 per group, P values obtained by a two-tailed Student's t test with unequal variance. (E) Apoptosis in β₃WT and β₃D723R brain lesions by TUNEL staining (brown).

The site-specific growth phenotype seen in β₃WT and β₃D723R cells was also observed with MDA-MB-435 β₃-/WT and β₃-/D723R cells (Fig. 1C). The previously described β₃- cells (12) were obtained through selection with a saporin-conjugated anti-β₃ antibody and used to generate β₃WT and β₃D723R re-expressing variants by plasmid transfection, and antibiotic selection (12). The organ-dependent growth behavior observed in both independently generated cell models confirmed that the phenotype is indeed due to the α_vβ₃ activation state. Thus, activation of integrin α_vβ₃ critically enhances tumor cell growth in the brain but is not required for growth in the MFP, implying that α_vβ₃ activation promotes metastatic growth in a microenvironment-dependent manner.

To investigate the mechanism involved, we followed tumor cell growth in the brain over time and realized initially no difference between cells expressing activated versus non-activated α_vβ₃. However, 10 days after implantation of 10⁴ tumor cells, a significant growth difference became evident and then increased continuously (Fig. 2B). This difference was caused by an enhancement in tumor cell proliferation. On day 21, in vivo BrdU incorporation was 25% higher in β₃D723R than in β₃WT implants (Figs. 2C, 2D). Both lesion types showed low and comparable levels of apoptosis (<1%) by terminal deoxynucleotidyl transferase dUTP nick end labeling (TUNEL) staining (Fig. 2E). Thus, once lesions in the brain reach an initial critical size, tumor cell integrin α_vβ₃ activation strongly enhances proliferation, without affecting cell survival.

Tumor Cell Integrin α_vβ₃ Activation Promotes Angiogenesis and Prevents Development of Hypoxia in Brain Lesions.

Having observed that the growth advantage of brain implants expressing activated β₃D723R became apparent once the lesions reached an initial critical size, we asked whether a host response contributes to the accelerated tumor growth. In vivo labeling and detailed immunohistochemistry revealed two major points. First, the tumor area covered by blood vessels (CD34 stain) was significantly larger in lesions expressing activated β₃D723R than in those expressing non-activated β₃WT (Fig. 3A). This included all tumors, regardless of size. Notably, on day 21 post-implantation, five of five lesions in the β₃D723R group were large and measured 2 to 3 mm in diameter, whereas only one in five in the β₃WT group reached a threshold size of 1.5 mm. Second, whereas microvessels were distributed evenly throughout the large tumor areas in the β₃D723R group with no or minimal hypoxic regions (<1%), vessels in the β₃WT group were located mainly along the lesion margins. The only lager tumor in this group had developed extensive hypoxic regions in the poorly vascularized central area (Fig. 3B).

Fig. 3.

Tumor cell integrin α_vβ₃ activation increases angiogenesis and reduces hypoxia in brain lesions. (A) Quantification of the total vessel area in β₃WT and β₃D723R brain lesions 21 days after tumor cell implantion. Vessels were visualized by CD34 staining. (B) Immunohistochemistry of tumor vessels (brown) and hypoxic regions (gray-blue) in brain lesions on day 21. Borders between tumor area and brain parenchyma are outlined showing representative small (≤1.5 mm) and large (≥ 1.5 mm) β₃WT lesions. All β₃D723R lesions were >2 mm. (C) (Left) In vivo signal of β₃WT (black) and β₃D723R (red) brain lesions on day 21, and on day 29 for β₃WT (gray) to obtain larger lesions for this group. Only lesions with a diameter ≥1.5 mm (≈5 × 10⁷ bioluminescence photons in vivo) were included in the quantification of hypoxic tumor regions (marked with a bracket as large tumors). (Right) Percentage of hypoxic area in brain lesions >1.5 mm in diameter; n = 5 per group; P values based on two-tailed Student's t test with unequal variance. (D) Large β₃WT lesions (> 1.5 mm) show increased apoptosis (Left; TUNEL staining, brown; nuclei, green) within hypoxic regions (Right; Hypoxyprobe-1 staining, gray-blue; nuclei, green) in adjacent sections. (E) Formation of the VEGF:VEGFR2 complex on tumor vessels in β₃D723R brain lesions (Gv39M staining, brown).

To investigate whether the development of hypoxic regions in the β₃WT group is a function of lesion size, we allowed implants of β₃WT-expressing cells to grow longer and become larger, making it possible to compare them to β₃D723R-expressing lesions. An association between hypoxia development and tumor size in lesions expressing non-activated β₃WT would further point to angiogenesis as a growth limiting factor. On day 29 post-implantation, four of five β₃WT lesions were larger than 1.5 mm, and they were used for comparison with β₃D723R lesions of the same or larger size (Fig. 3C). Although β₃D723R lesions developed no or only minimal hypoxia, β₃WT lesions harbored large hypoxic regions (Fig. 3C) with increased tumor cell apoptosis (Fig. 3D). Thus, expression of activated α_vβ₃ by the tumor cells is associated with their enhanced ability to attract new blood vessels, thereby avoiding hypoxia and protecting the growing tumor cells from apoptosis. De novo angiogenesis, rather than tumor cell growth around exiting vessels (vessel co-opting), was confirmed by expression of VEGF:VEGFR2 complex on the tumor vessels (Fig. 3E), identifying the angiogenic state of the endothelial cells (14).

Tumor Cell Integrin α_vβ₃ Activation Promotes VEGF Expression at the Post-Transcriptional Level in Normoxic Brain Lesions.

To identify a mechanism by which activated tumor cell integrin α_vβ₃ promotes angiogenesis in the brain, we focused on VEGF, a major promoter of endothelial proliferation (15). As hypoxia potently induces VEGF via transcription factor HIF-1α (16) or via induction of cap-independent internal ribosome entry site IRES-mediated translation (17), we analyzed VEGF expression in hypoxic versus normoxic tumor regions, identifying hypoxia by staining for Hypoxyprobe-1 and probing for HIF-1α. In hypoxic regions, VEGF protein expression was strong and associated with increased HIF-1α signal regardless of the α_vβ₃ activation state of the tumor lesions (Fig. 4A), indicating VEGF up-regulation at the transcriptional level. In contrast, in normoxic tumor regions devoid of Hypoxyprobe-1 and HIF-1α signal, VEGF was detected only in β₃D723R but not in β₃WT lesions (Fig. 4A).

Fig. 4.

Tumor cell integrin α_vβ₃ activation increases VEGF expression in the brain post-transcriptionally by 4E-BP phosphorylation. (A) VEGF protein expression in normoxic (upper panels) versus hypoxic areas (lower panels) of β₃WT and β₃D723R brain lesions by immunohistochemistry. Hypoxic regions defined by positive staining for Hypoxyprobe-1 (Left) and HIF-1α (Middle). β₃WT and β₃D723R lesions express VEGF in hypoxic regions. In normoxic regions, only β₃D723R lesions express VEGF (Right; enlarged on far right). (B) VEGF protein levels in microdissected β₃WT and β₃D723R brain lesions 21 days after tumor cell implantation (immunoprecipitation [IP], VEGF; Western blot [WB], VEGF). (C) VEGF mRNA levels in microdissected β₃WT and β₃D723R brain lesions on day 21 post-implantation (RT-PCR). Lesions from three different mice per group. (D) Phospho-4E-BP1 signal in microdissected β₃WT and β₃D723R brain lesions 21 days after tumor cell implantation (Western blot with anti-phospho-4E-BP1 and anti-tubulin antibodies). (E) Down-regulation of 4E-BP1 in β₃WT cells by shRNA results in a strong increase of VEGF protein level under normoxia in vitro. VEGF was immunoprecipitated from cell culture supernatants before Western blot analysis. (F) Detection of mTOR and Akt activation levels with phospho-specific antibodies in microdissected β₃WT and β₃D723R brain lesions 21 days after tumor cell implantation.

To determine whether α_vβ₃ activation in normoxic regions regulates VEGF expression at the transcriptional or post-transcriptional level (18, 19), proteins and total RNA were extracted from microdissected β₃WT- and β₃D723R-expressing lesions of mouse brains 21 days after tumor cell implantation. All analyzed β₃WT lesions were smaller than 1.5 mm and still without hypoxic areas. β₃D723R lesions, although larger, also showed no or minimal hypoxia (<1%) (Fig. 3). Western blot analysis of VEGF immunoprecipitated from brain tumor lysates confirmed increased VEGF protein levels in β₃D723R-expressing lesions (Fig. 4B). Reverse transcription–polymerase chain reaction (RT- PCR) with VEGF-specific primers yielded four distinct PCR products (230, 310, 365, and 430 bp) (Fig. 4C), corresponding to splice variants VEGF₁₂₁, VEGF₁₄₅, VEGF₁₆₅, and VEGF₁₈₉ (20), with no differences in mRNA levels or isoform patterns between β₃WT- and β₃D723R-expressing lesions (Fig. 4C). Thus, under normoxic conditions, tumor cell integrin α_vβ₃ activation regulates VEGF expression at the post-transcriptional level.

To investigate the mechanism involved, we focused on VEGF mRNA regulation. VEGF mRNA is “weakly” transcribed because of its highly structured 5′UTR, and its translation under normoxia strongly depends on the availability of eukaryotic initiation factor eIF4E (19, 21). eIF4E is bound and inhibited by eIF4E binding protein 1 (4E-BP1), which, when phosphorylated, no longer binds eIF4E and makes it available for translation of weak RNAs (22). 4E-BP1 was significantly stronger phosphorylated in β₃D723R- than β₃WT-expressing brain lesions (Fig. 4D), suggesting that increased VEGF expression is due to phosphorylation and inactivation of 4E-BP1. A causal relationship between 4E-BP1 inhibition and VEGF production under normoxia was confirmed by the finding that shRNA knock-down of 4E-BP1 in β₃WT cells strongly increased VEGF protein levels in vitro (Fig. 4E). Similar to β₃WT and β₃D723R cells, 4E-BP1 knock-down and control cells also showed identical growth rates in culture (Fig. S2B).

We further investigated mTOR and Akt as major constituents of pathways controlling the 4E-BP1 protein (22). The mTOR protein is large and prone to degradation during tissue processing. We therefore compared the ratio between phosphorylated mTOR(Ser-2481) and total mTOR and found no difference between β₃WT- and β₃D723R-expressing lesions (Fig. 4F). Similarly, Akt activation measured by phosphorylation on Ser-308 and Thr-437 was also not altered because of α_Vβ₃ activation (Fig. 4F). Thus, enhanced phosphorylation of 4E-BP1 associated with α_Vβ₃ activation of the tumor cells is apparently not controlled by mTOR or Akt activation.

Tumor Cell Integrin α_vβ₃ Activation Does Not Prevent Hypoxia in Mammary Fat Pad Tumors.

Although tumor cell integrin α_vβ₃ activation significantly increased growth of MDA-MB-435 cell implants in the brain, it had no impact on tumor growth in the MFP (Fig. 1B, 1C). There, β₃kd as well as β₃WT- and β₃D723R-expressing cells grew more slowly than in the brain, and α_vβ₃ activation made no difference. In fact, MFP lesions expressing β₃WT or β₃D723R showed much lower blood vessel densities than brain lesions and developed extensive hypoxic regions with strong apoptosis and necrosis (Fig. 5). Hypoxic areas and some of the surrounding tissue strongly expressed VEGF in β₃WT as well as β₃D723R derived MFP lesions (Fig. S3). Because of the extensive expansion of hypoxic regions in MFP tumors, seen even in early lesions still smaller than 1 mm (not shown), it was not possible to define clearly normoxic areas and to analyze VEGF expression there. Thus, in the MFP, tumor cell α_vβ₃ activation is not sufficient to enhance recruitment of tumor blood vessels, prevent hypoxia, and accelerate tumor cell growth. Therefore, tumor cell integrin α_vβ₃ activation selectively enhances cancer growth in a microenvironment-dependent manner, strongly supporting lesion growth in the brain, a particularly feared metastatic site.

Fig. 5.

Tumor cell integrin α_vβ₃ activation does not prevent hypoxia in mammary fat pad tumors. Hypoxia (Hypoxyprobe-1 in gray-blue, Upper), apoptosis/necrosis (TUNEL staining in brown, Middle) and blood vessel denstity (CD34 in brown, Lower) in MFP tumors 34 days after implanting 1 × 10⁴ β₃WT- or β₃D723R-expressing tumor cells. Quantifications are included (n = 5 per group). P values are based on two-tailed Student's t test with unequal variance. Note: All 34-day old MFP tumors were smaller than 21-day-old β₃D723R brain lesions.

Discussion

We analyzed a role of integrin α_vβ₃ activation in tumor cell growth in the brain and mammary fat pad to represent clinically highly relevant metastatic and primary sites. We focused explicitly on α_vβ₃ in tumor cells, as it is largely unknown how the activation state of this receptor affects cancer progression whereas the endothelial receptor has been widely studied in angiogenesis. We show that tumor cell integrin α_vβ₃ activation strongly promotes metastatic growth in the brain by enabling tumor cells to attract blood vessels independently of hypoxia and thereby avoiding hypoxia-related growth stagnation and apoptosis. In contrast, in the mammary fat pad, tumor cell α_vβ₃ had no effect on lesions growth, which was generally slower than in the brain, poorly vascularized, extensively hypoxic, and associated with apoptosis and necrosis. Expression of activated tumor cell integrin α_vβ₃ could not overcome this limitation as it did in the brain. Therefore, the specific ability of activated tumor cell integrin α_vβ₃ to enhance angiogenesis clearly depends on the tissue microenvironment. Organ-dependent differences in growth factors, chemokines, cytokines, and matrix proteins, as well as stromal and immune components, may distinctly affect tumor cell growth and endothelial behavior in the mammary fat pad versus the brain (23) and may synergize with (brain) or antagonize (MFP) growth-promoting functions of activated tumor cell α_vβ₃.

Intracranial MDA-MB-435 lesions expressed four different VEGF isoforms, including freely diffusible VEGF₁₂₁ and partially diffusible VEGF₁₆₅, the most frequent isoform in VEGF-secreting cells (15, 19). Because of the strong hypoxia in the MFP environment, we were unable to reliably define normoxic MFP tumor areas and determine VEGF expression there. However, as previously shown, over-expression of VEGF₁₂₁ in glioma cells increased angiogenesis in the brain, but not in the subcutaneous environment (24). This suggests that even though MDA-MB-435 β₃D723R tumors might express VEGF in normoxic regions in the MFP, this might not be sufficient to promote angiogenesis. Similarly, HIF-1α knockout in transformed astrocytes reduced their growth in the subcutis but promoted it in the brain (25). Thus, the inability of the mammary fat pad to support continuous, hypoxia-independent angiogenesis may be related to the nature and responsiveness of vascular development in the mammary fat pad compared with the brain.

It was previously shown that VEGF can promote breast cancer cell survival in an autocrine manner (26). However, in our model, tumor cell survival was not limiting, and we found increased proliferation but no evidence for enhanced survival of tumor cells upon α_vβ₃ activation in normoxic brain lesions. Integrins can signal through different growth promoting pathways, including Src and MAPK (27–30). However, we detected no significant differences in pSrc and pERK levels between β₃WT- and β₃D723R-expressing brain lesions (not shown). In fact, all of our data point to increased vascularization as the main mechanism through which activated tumor cell integrin α_vβ₃D723R accelerates lesion growth in the brain: increased vascular density, strongly reduced hypoxia, and formation of VEGF:VEGFR2 complex indicating an angiogenic switch. Furthermore, divergence in growth rates between β₃WT- and β₃D723R-expressing brain lesions became obvious only after the tumors reached a size at which blood supply became limiting in β₃WT tumors.

Although lesions expressing activated or non-activated α_vβ₃ up-regulated VEGF protein in hypoxic regions, only lesions with the activated receptor expressed elevated VEGF under normoxia. This was caused by α_vβ₃ activation–dependent control of VEGF at the post-transcriptional level. Our data indicate that activated α_vβ₃ promotes VEGF protein expression through inhibition of 4E-BP1, a repressor of mRNA translation. α_vβ₃ Activation in brain lesions induced phosphorylation and thereby inactivation of 4E-BP1, which is known to result in the release of eIF4E and increased translation of VEGF mRNA (21). A causal relationship between VEGF increase and 4E-BP1 inhibition in our model was provided by knock-down of 4E-BP1 in β₃WT cells, as this resulted in increased VEGF production under normoxic conditions. Notably, under hypoxia, 4E-BP1 might exert an opposite effect on VEGF, as its over-expression has been shown to increase IRES-dependent VEGF translation (17). In the current study, however, we focused on normoxia, because increased VEGF expression resulting from α_vβ₃ activation occurred in normoxic brain lesions. Thus, our findings establish a link between the activation state of integrin α_vβ₃, regulation of translation, and angiogenesis in vivo.

Several integrins, including α₆β₄, α_IIbβ₃, and α₂β₁, can promote translation of weak mRNAs through the PI3K-Akt-mTOR pathway (31, 32). In contrast, activation of integrin α_vβ₃ did not result in enhanced activation of mTOR or Akt in brain lesions, suggesting that a different mechanism is involved. Notably, mTOR-independent phosphorylation of 4E-BP1 in breast cancer cells has been previously documented (33).

The incidence of brain metastases in tumor patients is rising. However, although therapies for primary tumors and extracranial metastases are improving, effective treatments for brain metastases are still missing (1–3). Our data define the activated, high-affinity conformer of tumor cell integrin α_vβ₃ as a potential new therapeutic target in brain metastasis. Moreover, our study points to the importance of the tumor microenvironment when considering different treatment options for cancer. Beyond drug accessibility, brain metastases apparently require specific targeting. Thus, information on mechanisms that promote metastatic growth in the brain as presented here can be a basis for the development of novel, effective therapies.

Materials and Methods

Tumor Cell Model.

MDA-MB-435 adenocarcinoma cells were from J.E. Price (M. D. Anderson), cultured as previously described (12), and stably tagged with Firefly luciferase (F-luc) using a lentiviral transduction system (34). For generation of MDA-MB-435 β₃kd, β₃WT, β₃D723R, β₃-/WT, β₃-/D723R, and 4E-BP1 knock-down cells, see SI Methods.

Animal Experiments.

Female 6- to 8-week-old CB17/SCID mice were injected with 1 × 10⁴ F-luc–tagged cancer cells either stereotactically into the striatum (2.5 mm right from the midline, 2.5 mm anterior from bregma, and 3 mm deep) or into the axillary mammary fat pad and imaged once (mfp) or twice (brain) weekly using an IVIS200 bioluminescence imaging system (Xenogen). BrdU (Sigma) (1 ml of 1 mg/ml) was injected intraperitoneally at 24 hours, and Hypoxyprobe-1 (NPI) (150 μl of 10 mg/ml) intraperitoneally 60 minutes before tissue harvest. All animal work complied with National Institutes of Health and institutional guidelines (TSRI is AAALAC accredited).

For detailed protocols of immunohistochemistry, RNA, and protein expression, see SI Methods.