Deregulated HIF-1α is critical for breast cancer cell migration and invasion. (A) Western blot analysis of the HIF-1α levels in nontransformed breast epithelial cells (HBL-100; a, b) and breast cancer cells (MDA-MB-231; c, d) under either normoxia (21% O₂, lane 1) or hypoxia (1% O₂, lanes 2–5) over the indicated time points. Note: Equal loadings of all samples and all procedures side by side. (B) The efficiency of FG-12 lentiviral infection in MDA-MB-231 cells, as indicated by expression of an in-cis CMV-driven GFP gene, followed by FACS analyses. The same field was shown with either phase contrast (left) or fluorescence lens (right). (C) Specific down-regulation of HIF-1α (a) or HIF-1β (d) proteins by FG-12-delivered shRNA, as indicated by Western blot analyses. (D) Twelve-well tissue culture plates were precoated with type I collagen (20 μg/ml, 2 h). Serum-starved cells were plated (250,000 cells/well) in serum-free medium, and >90% of the cells attached within 2 h. The wound closure at 16 h was photographed and quantified as average gap (AG; Li et al., 2004). n = 3, p < 0.05. (E) Down-regulation of HIF-1α or HIF-1β inhibited MDA-MB-231 cell invasion through a Matrigel barrier (b and c vs. a), according to manufacturer's protocol. Note: OD reading (Bio-Rad Protein Assay at 590 nm) on the penetrated cells only. The data are expressed as means ± SD (n = 4, p < 0.05).

Deregulated HIF-1α uses secreted Hsp90α for migration and invasion. (A) Serum-free conditioned medium (CM; 25 μl of 10× concentrated) of HBL-100 (lanes 1 and 2) or MDA-MB-231 (lanes 3 and 4) cells incubated under normoxia (N) or hypoxia (H) for 14 h was analyzed for the presence of Hsp90α proteins by Western blotting. (B) CM of HIF-1α– or HIF-1β–down-regulated cells was analyzed for the presence of Hsp90α (a, lanes 2, 3 vs. lane 1). A 1.0-ml amount of 1× CM was concentrated 20 times and subjected to zymography gel analysis (Materials and Methods; b, lanes 1–3). (C) Approximately 400 ng each of recombinant Hsp90α, Hsp90β, and GRP94 were resolved on duplicate SDS gels and subjected to either Coomassie blue staining (top) or Western blot using the same anit-Hsp90α antibody (bottom). This antibody is specific against Hsp90α. (D) Reintroduction of wt and CA mutant (a, lanes 2, 3), but not dominant-negative mutant (b, lanes 4), of HIF-1α rescued Hsp90α secretion in HIF-1α–down-regulated MDA-MB-231 cells (c, lanes 2 and 3 vs. lanes 1 and 4). (E) Twelve-well tissue culture plates were coated with type I collagen (40 μg/ml, 2 h). Unattached collagens were removed by washing with Hank's balanced salt solution buffer. Serum-starved (18 h) MDA-MB-231 cells were plated (250,000 cells/well) under serum-free medium so that the cell density reached 90% confluence within 2 h. The “wounds” were made with a p-200 pipette tip. The wound closure was photographed and quantitation (AG) carried out as described (Li et al., 2004). The results shown here were reproducible in three independent experiments (n = 3, p < 0.05).

Secreted Hsp90α is essential for migration and invasion of MDA-MB-231 cells under serum-free conditions. (A) Colloidal gold migration assays show that anti-Hsp90α neutralizing antibodies blocked individual MDA-MB-231 cell motility in a dose-dependent manner (c–e vs. a, b). The addition of excess amount of recombinant Hsp90α reversed the inhibition by the antibodies (f). MI, migration index (%) (Li et al., 2004). (B) Anti-Hsp90α neutralizing antibodies blocked MDA-MB-231 cell invasion through a Matrigel in a dose-dependent manner (c–e vs. a, b). The addition of excess amount of recombinant Hsp90α reversed the inhibition of invasion by the antibodies (f). The OD reading is the same as in Figure 1.

F-5 epitope in secreted Hsp90α mediates tumor cell migration and invasion. (A) A summary of truncated peptides of Hsp90α that retain either a full or partial promotility activity of full-length Hsp90α. Cell motility data are summary (%) of colloidal gold motility and “scratch” assays combined (n = 3, each assay, p < 0.05). (B) FPLC-purified, full-length, F-2, F-5, and F-6 were visualized in SDS gel stained with Coomassie brilliant blue (lanes 1–4), with indicated amounts of BSA as controls (lanes 5–7). F-8 is a synthetic peptide. (C) The five peptides with their optimized concentrations were tested for rescuing invasion defect of HIF-1α–down-regulated MDA-MB-231 cells. (D) Quantitation of the invasion data (n = 3, p < 0.05) in C. (E) Lysates of MDA-MB-231 cells infected with vector alone (lane 1) or lentivirus carrying shRNA against LRP-1 receptor (lane 2) were analyzed by Western blotting with anti–LRP-1 antibody. (F) LRP-1–down-regulated cells were unable to invade as the control cells (b vs. a), and Hsp90α was unable to rescue the invasion defect of the LRP-1–down-regulated cells (n = 4, p < 0.05).

Hsp90α signaling is essential for MDA-MB-231 cell lung colonization and tumor formation in vivo. (A) Lentiviral system, FG-12, mediated shRNA-LRP-1 delivery and down-regulation of endogenous LRP-1 in MDAMB-231 cells (lane 2 vs. lane 1). (B) Western blot screening of normal and cancer cell lines for expression of LRP-1 receptor. (C) Constitutive expression of HIF-1α in MDA-MB-468 cells under either normoxia (N) or hypoxia (H; a). Constitutive secretion of Hsp90α by MDA-MB-468 cells (c). (D) Approximately 1 × 10⁶ luciferase-engineered MDAMB-231 cells infected with either vector only (a) or vector carrying shRNA-LRP-1 (b) or LRP-1^−/− MDAMB-468 cells (c) were injected into the tail vein of SCID mice (n = 7 per group). Whole-body bioluminescence imaging of the mice was performed once a week. Representative images (two per group) of the mice show lung colonization of the injected cells at days 1, 14, 28, 56, and 70. (E) Between days 70 and 75 (note: four to five of the seven mice injected with the control MDAMB-231 cells died of tumors), mice were put to sleep and the entire lung removed and sectioned for H&E staining analyses. Blue and green arrows indicate invading tumor nodule and normal lung tissue, respectively. L, lung parenchyma; T, tumor nodule; scale bars as indicated. (F) One-way analysis of variance analysis of the staining shown in E was carried out to compare the number of tumors in the lungs of mice carrying MDA-MB-231-vector, MDA-MB-231-LRP-1RNAi, and MDA-MB-468 cells. *p < 0.05, **p < 0.003.

A model of secreted Hsp90α as a potential target for HIF-1α–positive cancers. The severe hypoxia often found at the center of a tumor causes constitutive accumulation of HIF-1α. The deregulated HIF-1α triggers secretion of Hsp90α via exosomes. The secreted Hsp90α binds, via F-5 epitope, to cell-surface LRP-1 receptor and promotes motility and invasion of tumor cells in an autocrine manner. Whereas current clinical trials focus on intracellular HIF-1α, we propose that targeting the F-5 epitope of secreted Hsp90α would be more effective and safer in the treatment of cancer patients.