The Ras Signaling Inhibitor LOX-PP Interacts with Hsp70 and c-Raf To Reduce Erk Activation and Transformed Phenotype of Breast Cancer Cells

Abstract

The lysyl oxidase gene (LOX) inhibits Ras signaling in transformed fibroblasts and breast cancer cells. Its activity was mapped to the 162-amino-acid propeptide domain (LOX-PP) of the lysyl oxidase precursor protein. LOX-PP inhibits Erk signaling, motility, and tumor formation in a breast cancer xenograft model; however, its mechanism of action is largely unknown. Here, a copurification-mass spectrometry approach was taken using ectopically expressed LOX-PP in HEK293T cells and the heat shock/chaperone protein Hsp70 identified. Hsp70 interaction with LOX-PP was confirmed using coimmunoprecipitation of intracellularly and bacterially expressed and endogenous proteins. The interaction was mapped to the Hsp70 peptide-binding domain and to LOX-PP amino acids 26 to 100. LOX-PP association reduced Hsp70 chaperone activities of protein refolding and survival after heat shock. LOX-PP interacted with the Hsp70 chaperoned protein c-Raf. With the use of ectopic expression of LOX-PP wild-type and deletion proteins, small interfering RNA (siRNA) knockdown, and Lox^−/− mouse embryo fibroblasts, LOX-PP interaction with c-Raf was shown to decrease downstream activation of MEK and NF-κB, migration, and anchorage-independent growth and reduce its mitochondrial localization. Thus, the interaction of LOX-PP with Hsp70 and c-Raf inhibits a critical intermediate in Ras-induced MEK signaling and plays an important role in the function of this tumor suppressor.

INTRODUCTION

Lysyl oxidase (LOX) (protein-6-oxidase; EC 1.4.3.13) is the key extracellular enzyme that controls collagen and elastin cross-linking, which is required for the biosynthesis of functional extracellular matrices. The LOX gene was isolated as the Ras recision gene (rrg) with the ability to inhibit the transforming activity of the H-Ras oncogene in NIH 3T3 fibroblasts (10, 35). Reduced LOX expression has been reported to occur in many carcinomas (5, 21, 23, 33, 60, 63, 73). Ectopic LOX gene expression in gastric cancer cells resulted in reduced tumor formation in nude mice (33). Lysyl oxidase is synthesized and secreted as a 50-kDa inactive proenzyme (Pro–LOX), which is processed by proteolytic cleavage to a functional 32-kDa enzyme (LOX) and an 18-kDa propeptide (LOX-PP). Expression of Pro–LOX in Ras-transformed NIH 3T3 fibroblasts inhibited the activities of the Erk1/2 and Akt kinases and transcription factor NF-κB (30). Subsequently, LOX-PP was identified as the inhibitor of Ras signaling and transformed phenotype in NIH 3T3 fibroblasts (51), NF639 breast cancer cells (44), and H1299 lung cancer and PANC-1 pancreatic cancer cells, with mutant RAS and TP53 genes (73). In NF639 cells, driven by Her-2/neu which signals via Ras, LOX-PP expression decreased Her-2/neu-mediated signaling and Erk activation in exponentially growing cells or following serum stimulation (44). LOX-PP also reduced mesenchymal phenotype in vitro, as judged by induction of epithelial and reduction of mesenchymal markers and invasive colony formation in Matrigel and tumor xenograft formation in a nude mouse model (44). Furthermore, LOX-PP attenuated fibronectin-stimulated integrin signaling and migration in breast cancer cells (75). Together, these studies suggest a role for LOX-PP in inhibiting the invasive phenotype of carcinomas, but the exact mechanisms by which it functions have not been elucidated.

Hsp70s are a family of stress response proteins that act as molecular chaperones that prevent protein aggregation and also refold denatured or unfolded proteins via their ATPase activity, catalyzed by ATP-hydrolysis (22, 32). The human Hsp70 family includes at least 8 distinct genes that code for Hsp70 isoforms, located on several different chromosomes. Hsp70 family proteins are structurally conserved, and include an ATPase domain in the amino-terminal region, a peptide binding domain in the carboxy-terminal region, and an acidic motif (EEVD) in the extreme carboxy-terminal region. Hsp70 family proteins localize to the nucleus, mitochondria, endoplasmic reticulum, Golgi apparatus, cytosol, and extracellular cell surface. The two major cytoplasmic isoforms are Hsc70 and Hsp70/Hsp72, which is also termed Hsp70. Generally, Hsc70 is abundantly and ubiquitously expressed in nontumor tissues, whereas Hsp70/Hsp72 expression is induced by various signals. Constitutively expressed Hsp70/Hsp72 proteins have been detected in several tumors. In breast cancer, Hsp70 overexpression has been associated with shorter disease-free survival, a more invasive phenotype, and poorer prognosis (9). The exact roles of the Hsp70/Hsp72 family in cancer are not fully understood; however, these proteins can contribute to tumor cell survival and tumorigenesis via multiple antiapoptotic functions and through their role as a cochaperone for Hsp90 (47, 66, 76). Interestingly, it is reported that dual silencing of Hsc70 and Hsp70 induces tumor specific apoptosis (57). In the present study, we identified Hsp70 protein as a novel binding partner of LOX-PP, mapped the domains of interaction, and demonstrated LOX-PP can also associate with c-Raf, an Hsp70 client. Functionally, the interaction with LOX-PP reduced the chaperone and survival functions of Hsp70 upon stress and decreased the activation of c-Raf that promotes a more transformed phenotype of breast cancer cells.

MATERIALS AND METHODS

Plasmid construction.

For construction of N-terminally glutathione S-transferase (GST)-tagged LOX-PP and its deletion mutants, the cDNA encoding amino acids (aa) 1 to 162 (full length), 1 to 115 (ΔC1), 1 to 100 (ΔC2), 1 to 77 (ΔC3), 1 to 61 (ΔC4), 1 to 25 (ΔC5), 26 to 162 (ΔN1), 62 to 162 (ΔN2), 78 to 162 (ΔN3), 101 to 162 (ΔN4), 116 to 162 (ΔN5), and 26 to 100 (M1) and deletions of aa 26 to 61 (ΔM1), 26 to 77 (ΔM2), and 26 to 100 (ΔM3) of LOX-PP was amplified from full-length cDNA (44) and inserted into the BamHI/ClaI site of pEBG-GST mammalian expression vector, a generous gift from Bruce Mayer (University of Connecticut Health Center, Farmington, CT). For construction of C-terminally GST-tagged LOX-PP, the cDNAs encoding GST and LOX-PP were amplified and inserted into pcDNA3.1(+) (Invitrogen, Carlsbad, CA). For construction of C-terminally V5/His-tagged proteins, the cDNAs encoding LOX-PP, LOX, or Pro–LOX were inserted into pcDNA4/V5-His vector (Invitrogen). For preparation of recombinant rat LOX-PP, recombinant LOX-PP (rLOX-PP)–myc-His protein, in which the signal peptide of Pro–LOX was replaced with the one from osteonectin (BM-40) in the pcDNA4/TO/myc-His vector, was used as described previously (70). pCX_bsr-Hsp70 and pAD-c-Raf were kindly supplied by Michael Sherman and Vladimir Gabai (Boston University School of Medicine, Boston, MA) and Linda Van Aelst (Cold Spring Harbor Laboratories, Cold Spring Harbor, NY), respectively. The cDNA encoding the Hsp70 wild type (WT) was inserted into pEBG, pGEX4T-3 (GE Healthcare, Uppsala, Sweden), and pFLAG-CMV-2 (Sigma-Aldrich, St. Louis, MO). The cDNAs encoding Hsp70 ΔSmaI and ΔBglII were digested with SmaI and BglII, respectively, as described by Milarski and Morimoto (43), and inserted into pFLAG-CMV2. The cDNA encoding c-Raf WT was subcloned into pEGFP-C1 (Clontech, Palo Alto, CA) and pGEX4T-3 vector. The cDNA encoding LOX-PP WT and ΔM3 was subcloned into pGEX4T-3 vector. All constructs with deletions were generated by PCR and verified by DNA sequencing.

Cell culture and treatment conditions.

ERα-positive MCF-7 and ZR-75 and ERα-negative Hs578T breast cancer cells, which contain a mutated constitutively active H-Ras, were purchased from the American Type Culture Collection (ATCC; Manassas, VA). Human embryonic kidney HEK293T cells and Bosc23 cells were obtained from the ATCC. Cells were maintained in Dulbecco's minimal essential medium (DMEM) supplemented with 10% heat-inactivated fetal bovine serum (FBS), 2 mM glutamine, and penicillin-streptomycin, as recommended by the ATCC. Mouse embryo fibroblasts (MEFs) from Lox^−/− or control C57BL/6 mice were prepared as described previously (39) and cultured in high-glucose DMEM supplemented with 20% FBS in DMEM with l-glutamine, nonessential amino acids, bicarbonate, and antibiotics. T-Rex293 cells were cultured as described previously (70). NIH 3T3 cells were kindly provided by Amitha Palamakumbura (Boston University Goldman School of Dental Medicine, Boston, MA) and cultured as described previously (30). The NF639 cell line, kindly provided by P. Leder (Harvard Medical School, Boston, MA) was derived from a mammary gland tumor in a mammary tumor virus (MMTV)-ERBB2 transgenic mouse and cultured as described previously (18). For heat shock treatment, cells were transfected with GST-tagged WT or ΔM3 LOX-PP or GST using Lipofectamine 2000 (Invitrogen). After 24 h, cells were submerged in a 45°C water bath for 10 min and allowed to recover at 37°C for 24 h. Cells were then assayed using 0.4% trypan blue uptake (Invitrogen) for the percentage of cell death. Assays were performed three independent times, each in duplicate, and the means ± standard deviations (SD) are presented. The RNA duplexes used for targeting mouse LOX (oligonucleotide A, 5′-TAGGGCGGATGTCAGAGACTA-3′; oligonucleotide B, 5′-AACGATCCTTTCAAATTATAA-3′) were purchased from Qiagen (Germantown, MD) and transfected at a final concentration of 20 nM using Lipofectamine RNAiMAX (Invitrogen).

Antibodies and immunoblot analysis.

Antibodies against Hsp70 (SPA-810), Hsp70/Hsc70 (SPA-820), and Hsp90 (SPA-830) were purchased from Stressgen (Victoria, BC, Canada). Antibodies against α-tubulin (DM1A), β-tubulin (TUB 2.1), γ-tubulin (GTU-88), β-actin (AC-15), and FLAG (M2) were from Sigma-Aldrich. Antibodies against Erk1/2 (no. 9102), p-Erk1/2 (phospho-Thr202/Tyr204; no. 9101), Akt (no. 9272), MEK1/2 (L38C12; no. 4694), p-MEK1/2 (phopho-Ser217/221; no. 9121), and EGFR (no. 2232) were purchased from Cell Signaling (Danvers, MA). Antibodies from Santa Cruz Biotechnology (Santa Cruz, CA) included anti-GST (B-14) and anti-B-Raf (F-7). Antibodies against c-Raf (clone 53) and Apaf-1 (A92820) were from BD Transduction (Franklin Lakes, NJ). Monoclonal antibodies against V5 (R960-25) and COX-1 (COX 111) and polyclonal antibody against green fluorescent protein (GFP) (A-6455) were from Invitrogen. Rabbit polyclonal antibody against V5 (E14) from Delta BioLabs (Gilroy, CA) was used for immunoprecipitation and immunofluorescence microscopy. Antibody against His tag (120-003-812) was from Macs Miltenyi Biotec (Germany). Rabbit polyclonal antibodies against LOX-PP were either prepared as described previously (27) or purchased from Novus Biologicals (NBP1-30327) (Littleton, CO) and react with mouse or rat or human or rat, respectively.

For preparation of whole-cell lysates, cells were solubilized in phosphate-buffered saline (PBS) with 0.5% SDS, 1 mM Na₃VO₄, and 10 mM NaF. Lysates were separated by SDS-PAGE, transferred to polyvinylidene difluoride (PVDF) membranes, and subjected to immunoblotting using the appropriate primary antibodies and horseradish peroxidase (HRP)-conjugated secondary antibodies (goat anti-mouse IgG-HRP [Santa Cruz; no. sc-2005] or anti-rabbit IgG-HRP [Bio-Rad; no. 170-6515]), as described previously (30).

Affinity isolation of LOX-PP-interacting proteins.

HEK293T cells plated on six 35-mm dishes were transfected with 2 μg of pEBG or pEBG-mLOX-PP expression plasmid per dish using Lipofectamine 2000 reagent according to the manufacturer's protocol. After 24 h, cells were lysed in 300 μl of buffer A (25 mM HEPES-KOH [pH 7.2], 150 mM KCl, 2 mM EDTA, 1 mM phenylmethylsulfonyl fluoride (PMSF), 1 mM dithiothreitol, 0.5 μg/ml leupeptin, 2 μM pepstatin A, 1 μg/ml aprotinin, and 1% Triton X-100) per dish. The lysates were centrifuged in a microcentrifuge for 10 min at 13,000 rpm at 4°C to remove insoluble material. The supernatant (about 2 mg) was incubated with 20 μl glutathione-Sepharose 4B (GE Healthcare) for 2 h at 4°C. The resin was washed four times with lysis buffer, and proteins were eluted with SDS-PAGE loading buffer. Following separation by 10% SDS-PAGE, proteins were visualized by Coomassie blue staining and bands corresponding to 70 kDa and 52 kDa were isolated. In-gel proteolytic digestion and mass spectrometry (liquid chromatography-tandem mass spectrometry [LC-MS-MS]) was performed by the Taplin Biological Mass Spectrometry Facility (Boston, MA).

Effects of ATP on the association of Hsp70 and LOX-PP.

HEK293T cells, plated on 35-mm dishes, were transfected with the indicated expression plasmids. After 24 h, cell lysates were prepared in 300 μl of lysis buffer A or in the same buffer with 1 mM ATP and 2 mM MgCl₂ but without EDTA. GST pulldown assays were performed as described above. The beads were washed with lysis buffer, and the precipitated proteins were detected by immunoblotting.

Luciferase refolding assay.

The cDNA of cytoplasm-localized luciferase (Luc-Cyto) in which the C-terminal peroxisomal localization signal has been destroyed by a Leu-Val mutation of the amino acid at position 550 was made by site-direct mutagenesis as described by Michels et al. (42). HEK293T cells were transfected with cytomegalovirus (CMV) promoter-driven Luc-Cyto expression plasmid and GST-tagged WT or ΔM3 LOX-PP or GST using Lipofectamine 2000 according to the manufacturer's protocol. After 24 h, cultures were treated with 20 μg/ml cycloheximide for 30 min at 37°C. To denature the expressed luciferase protein, cells were incubated at 45°C for 30 min in a water bath. The cells were then incubated for the indicated periods of time at 37°C to allow for refolding of the luciferase and lysed, and protein concentrations measured using a protein assay kit (Bio-Rad). The luciferase activities in equal amounts of protein lysates were measured using the luciferase assay system (Promega, Madison, WI) according to the manufacturer's protocol.

Preparation of recombinant proteins.

GST, GST–LOX-PP WT and ΔM3, GST-Hsp70, and GST-c-Raf were expressed in Escherichia coli BL21(DE3) pLysS (Invitrogen). The bacteria were grown in Luria broth at 30°C to an A₆₀₀ of 0.5, and protein expression was induced with 0.1 mM isopropyl β-d-thiogalactoside. At 3 h for GST, GST–LOX-PP WT and ΔM3, and GST-Hsp70 or 1 h for GST-c-Raf postinduction, the bacteria were harvested by centrifugation (3,700 × g for 10 min at 4°C), resuspended in lysis buffer (1% Triton X-100, 50 mM Tris, 150 mM NaCl, 1 mM EDTA, and 1 mM PMSF [pH 8.0]), and lysed by sonication. The lysate was centrifuged at 13,000 rpm for 10 min. The supernatant was loaded on to a glutathione-Sepharose 4B column, washed extensively with lysis buffer containing 300 mM NaCl and 0.1% Triton X-100, and eluted with phosphate buffer (36 mM Na₂HPO₄ and 14 mM NaH₂PO₄ [pH 7.2]) containing 100 mM NaCl, 0.1% Triton X-100, and 30 mM glutathione. Purified rLOX-PP–myc-His protein, with the BM-40 signal peptide (70), and GST fusion proteins (0.5 μM) were used in binding assays conducted in phosphate buffer containing 100 mM NaCl and 0.1% Triton X-100. Purified binding partners, at the indicated concentrations, were incubated for 3 h at 4°C. Glutathione-Sepharose 4B was then added and allowed to incubate for 1.5 h at 4°C with gentle rotation. The beads were sedimented in a microcentrifuge. After extensive washing, the proteins bound to the beads were solubilized in sample buffer and subjected to SDS-PAGE followed by immunoblotting.

Immunoprecipitation analysis.

NF639, ZR-75, or NIH 3T3 cells were lysed with buffer A, as described above. Either rabbit anti–LOX-PP (reference 27 or NBP1-30327 [Novus Biologicals]), mouse anti-Hsp70/Hsc70, or rabbit anti-V5 tag (2 μg) was added to 500 μg cell lysate, followed by overnight incubation at 4°C. Protein G-Sepharose beads (Invitrogen) were then added to the mixture, followed by incubation at 4°C for 2 h with gentle shaking. The beads were washed four times with buffer A. The immune complexes were eluted from the Sepharose beads with SDS-PAGE sample buffer and the precipitated proteins analyzed by Western blotting.

Migration, invasion, colony, and NF-κB activity assays.

Suspensions of 1 × 10⁴ NF639 cells or MEFs were layered, in triplicate, in the upper compartments of a Transwell (Costar, Cambridge, MA) on a 6.5-mm-diameter polycarbonate filter (8-μm pore size) and incubated at 37°C for the indicated times. Migration of the cells to the lower side of the filter was evaluated by staining with crystal violet and quantified by spectrometric determination at A₅₇₀ as described previously (75). Assays were performed three times independently, each time in triplicate, and the means ± SD are presented. Suspensions of 1 × 10⁴ NF639 cells were layered in the upper compartment of a Transwell on a 6.5-mm-diameter polycarbonate filter (8-μm pore size) precoated with 10 μg of Matrigel and incubated at 37°C for 6 h. Migration of cells was quantified as described above for cell migration. Soft agar assays (30) and NF-κB element-driven luciferase assays (73) were performed as we have described previously.

Immunofluorescence microscopy.

NF639 cells plated on coverslips were transfected with small interfering RNA (siRNA) for 48 h. Cells were incubated for 30 min in 250 nM MitoTracker (Invitrogen), washed with PBS, and fixed with methanol at −20°C for 5 min. Following washing with PBS, cells were blocked by incubation in 2% (wt/vol) bovine serum albumin (A3059; Sigma) in PBS (buffer 1) and incubated for 1 h in primary antibody diluted in buffer 1. Following washing with PBS, the coverslips were incubated for 1 h in appropriate secondary antibodies conjugated with Alexa Fluor 488 (Invitrogen) diluted in buffer 1. Subsequently, the coverslips were mounted in Slowfade Gold antifade reagent (Invitrogen) with Hoechst 33342 (Invitrogen) and observed under a Nikon Eclipse E400 microscope. The images are representative of two independent experiments performed in duplicate.

Isolation of mitochondrial membranes.

NF639 cells, cultured on 100-mm dishes and transfected with 20 nM control siRNA or siLOX oligonucleotide B RNA for 48 h or wild-type or Lox^−/− MEFs cultured on 150-mm dishes, were harvested with a cell scraper, followed by centrifugation. Further steps were carried out on ice or at 4°C. Cells were washed with 1× PBS and then homogenized with 20 strokes in 3 volumes of homogenization buffer (10 mM Tris-HCl, pH 7.5, 0.25 M sucrose, 5 mM MgCl₂, 10 mM KCl, and 1 mM PMSF) in a stainless steel homogenizer. After centrifugation at 800 × g for 10 min, the postnuclear supernatant was centrifuged at 3,000 × g for 10 min and the resulting mitochondrial membrane preparation washed once with homogenization buffer.

RESULTS

Mass spectrometry identifies Hsp70 as a LOX-PP associated protein.

To identify proteins that can associate with LOX-PP, the murine propeptide was cloned into the mammalian pEBG vector, which expresses inserted cDNAs as GST fusion proteins (41). HEK293T cells were transfected with pEBG-mLOX-PP LOX-PP expression plasmid or a pEBG empty-vector (EV) control. After 24 h, extracts were prepared and incubated with glutathione-Sepharose 4B resin and bound proteins washed extensively and eluted with SDS-PAGE loading buffer. Following 10% SDS-PAGE, proteins were visualized by Coomassie blue staining (Fig. 1 A). Bands of approximately 70 kDa and 52 kDa were seen in GST–LOX-PP precipitates and not in the control GST lane. These were excised and subjected to in-gel proteolytic digestion and mass spectrometry (LC-MS-MS). The bands were identified as the 70-kDa heat shock protein (Hsp70) and α/β-tubulin, respectively (Table 1 ). To confirm the mass spectrometry analysis, the GST–LOX-PP and GST proteins were purified from HEK293T cells in a small-scale GST pulldown assay and subjected to Western blotting for Hsp70, for α-tubulin, β-tubulin, and γ-tubulin, and for β-actin and GST as loading controls. Hsp70 specifically copurified with GST–LOX-PP, as did all three of the tubulin proteins (Fig. 1B), confirming the associations identified by mass spectrometry. Hsp90, which frequently functions as a cochaperone with Hsp70, was somewhat unexpectedly not detected in the proteins that purified with LOX-PP on the resin column as visualized by Coomassie blue staining in Fig. 1A. Consistently, when we tested directly for its presence by immunoblotting, Hsp90 was not detectable among the proteins that coprecipitated with GST–LOX-PP (Fig. 1B). The association of ectopically expressed LOX-PP with endogenous Hsp70 proteins was also readily seen in several breast cancer cell lines, including ZR-75 and Hs578T (Fig. 1C) and MCF-7 (not shown), following transient GST–LOX-PP-versus-GST expression. Similarly, endogenous LOX-PP and Hsp70 proteins were found to coimmunoprecipitate in extracts from ZR-75 cells (Fig. 1D), which express low levels of LOX-PP, and from NIH 3T3 (Fig. 1E) and NF639 cells (see Fig. 6B).

Fig. 1.

Identification of Hsp70/72 and tubulin as LOX-PP-interacting proteins. (A) HEK293T cells were transfected with vectors expressing either GST or GST–LOX-PP. Proteins were precipitated with glutathione-Sepharose 4B beads and resolved by SDS-PAGE, and gels were stained with Coomassie brilliant blue R-250. The positions of the GST and GST–LOX-PP bands are indicated. The bands at approximately 52 kDa and 70 kDa were excised, subjected to in-gel digestion, and then analyzed by LC-MS-MS and identified as α/β-tubulin (α/β-tub.) and Hsp70/72, respectively (Table 1). Stars represent bands present in both lanes, presumably nonspecific associated proteins. Positions of molecular mass markers are given in the left lane. (B) Lysates of HEK293T cells expressing either GST or GST–LOX-PP plated on a 35-mm dish were purified as described for panel A using a GST pulldown assay and subjected to Western blotting with the indicated antibodies. For estimation of the amounts of expressed proteins, 4% of each of the lysates was separated and immunoblotted (Input). (C) ZR-75 (upper panels) and Hs578T (lower panels) breast cancer cells were transfected with expression plasmids for GST or GST–LOX-PP and the resulting lysates subjected to GST pulldown and Western blotting as described for panel B. (D) Triton X-100 extracts of ZR-75 cells were immunoprecipitated (IP) with rabbit anti-IgG or Novus LOX-PP antibodies, as indicated. The precipitated proteins were analyzed by Western blotting with antibodies against Hsp70 and LOX-PP. As the band of precipitated LOX-PP migrated close to that of rabbit IgG light chain, protein A-conjugated HRP was used as a secondary “antibody” to detect immunoprecipitated LOX-PP. (E) Triton X-100 extracts of NIH 3T3 fibroblasts were immunoprecipitated with rabbit IgG or LOX-PP antibodies (upper panels) or mouse IgG or Hsp70 antibodies (lower panels) and subjected to Western blotting with the indicated antibodies as described for panel D.

Table 1.

Summary of proteins identified by mass spectrometry^a

Molecular mass (kDa)	GenBank accession no.	Name	No. of peptides	Protein coverage by amino acid count (%)
70	P11142	Heat shock 70-kDa protein 8 (HSC70)	30	255/646 (39.5)
	P08107	Heat shock 70-kDa protein 1 (Hsp70)	24	164/641 (25.6)
52	Q6LC01	β-Tubulin	22	166/437 (38.0)
	Q13748	α-Tubulin	19	190/450 (42.2)

^aA summary of the analysis of the results of mass spectrometry of the two bands excised as described for Fig. 1A is shown.

Fig. 6.

Hsp70-interacting protein c-Raf associates with LOX-PP. (A) HEK293T cells were transfected with expression plasmids for GST or GST–LOX-PP (G–LOX-PP) and expressed proteins purified on glutathione-Sepharose 4B beads. Bound proteins were analyzed by Western blotting for the presence of Hsp70 and proteins known to interact with Hsp70, including c-Raf, Apaf-1, Akt, B-Raf, EGFR, Erk1/2, and MEK1/2, and for GST as the control (pulldown). For estimation of the amounts of expressed proteins, 4% of each of the lysates was immunoblotted (Input). (B and C) Triton X-100 extracts of NF639 cells (B) and NIH 3T3 fibroblasts (C) were immunoprecipitated with the antibodies indicated at the top. The precipitated proteins were analyzed by Western blotting with antibodies against Hsp70, c-Raf, and LOX-PP. As the band of precipitated LOX-PP and Pro–LOX migrated close to that of rabbit IgG light and heavy chains, respectively, protein A-conjugated HRP was used as a secondary “antibody” to detect immunoprecipitated LOX-PP and precursor Pro–LOX. (D) NF639 cells were transfected with expression plasmid for empty vector (EV) or V5-tagged LOX-PP, LOX, or Pro–LOX proteins. Triton X-100 extracts were immunoprecipitated with V5 antibodies and the precipitated proteins detected with antibodies against the c-Raf or V5 tag. (E) rLOX-PP–myc-His (0.5 μM) was incubated with GST (0.5 μM) or GST-c-Raf (G-c-Raf; 0.5 μM) for 3 h and then with glutathione-Sepharose 4B beads for 90 min. The proteins bound to the resin were separated by SDS-PAGE and visualized by immunoblotting with His tag antibody (upper panel) or Coomassie brilliant blue staining (lower panel). For estimation of the amount of rLOX-PP–myc-His protein present, 5% of the mixture was separated and immunoblotted (Input). (Inset) Coomassie blue-stained gel of bacterial expressed and purified ∼100-kDa GST-c-Raf (1 μg) is shown. Stars denote putative products of degradation or incomplete synthesis. The same molecular weight markers were used, and their positions are indicated on the left.

Next, we asked whether Hsp70 can interact with LOX-PP using purified preparations of myc-His-tagged recombinant rat rLOX-PP (Fig. 2A) and GST-Hsp70 protein (Fig. 2B). GST-Hsp70 protein brought down recombinant LOX-PP, confirming the direct association between these two proteins (Fig. 2C). Furthermore, LOX-PP was coprecipitated with GST-Hsp70 in HEK293T cells (Fig. 2D). Interestingly, α-tubulin was not detected in the copurified proteins, indicating that tubulin is not an intermediate in the Hsp70–LOX-PP protein complex (Fig. 2D). The association with α-tubulin is consistent with our previous observations in differentiating MC3T3-E1 cells, where LOX-PP was found to interact with the microtubule network (19). Thus, LOX-PP associates with Hsp70 and α/β/γ-tubulin. Furthermore, our findings suggest that LOX-PP is a client of Hsp70 but not Hsp90.

Fig. 2.

The interaction between LOX-PP and Hsp70 is direct and is reduced in an ATP-dependent manner. (A) T-Rex293 cells were transfected with a pcDNA4/TO/LOX-PP/myc-His vector expressing rLOX-PP–myc-His protein. rLOX-PP–myc-His (LOX-PP–myc-His) protein was purified on a nickel affinity column, a sample (1 μg) was subjected to SDS-PAGE, and the gel was stained with Coomassie blue. (B) E. coli-expressed GST and GST-tagged Hsp70 (G-Hsp70) proteins were purified on glutathione-Sepharose 4B beads, and 1 μg of each protein was resolved by gel electrophoresis. Left panel, Coomassie blue-stained gel; right panel, Western blotting (WB) with a GST antibody. (C) rLOX-PP–myc-His (0.5 μM), shown in panel A, was incubated with GST (0.5 μM) or GST-Hsp70 (G-Hsp70; 0.5 μM) for 3 h, following incubation with glutathione-Sepharose 4B beads for 90 min. The precipitated proteins and a sample of the input were analyzed by immunoblotting with antibodies against His tag and GST. Molecular mass markers are indicated on the left. (D) HEK293T cells were transfected with expression plasmids for GST or GST-Hsp70 (G-Hsp70) with LOX-PP-V5, and GST pulldown assays were performed as described for Fig. 1B and subjected to Western blotting with the indicated antibodies. (E) HEK293T cells were transfected with GST, N-terminal GST-tagged LOX-PP (G–LOX-PP), or C-terminal GST-tagged LOX-PP (LOX-PP-G). After 24 h, cell lysates were subjected to GST pulldown with glutathione-Sepharose 4B beads in the presence (+) or absence (−) of 1 mM ATP. The coprecipitated proteins were examined by Western blotting with antibodies against Hsp70, α-tubulin (α-tub.), and GST. For estimation of the amounts of expressed proteins, 4% of each of the lysates was separated and immunoblotted (Input).

Interaction of LOX-PP with Hsp70 is inhibited upon ATP binding.

Previous investigations of the Hsp70 family have shown that ATP binding induces a conformational change and that the interaction of Hsp70 proteins with their binding partners is frequently sensitive to the nucleotide-bound state (20). Thus, we next tested whether the interaction of LOX-PP with Hsp70 is ATP dependent. In the presence of 1 mM ATP, the interaction of LOX-PP and Hsp70, but not α-tubulin, was almost completely eliminated (Fig. 2E). These findings suggest a conformational dependence for the interaction with Hsp70 and regulation through an ATP-binding/hydrolysis cycle.

Mapping the domains of Hsp70 mediating interaction with LOX-PP.

Hsp70 has both ATPase and peptide binding domains (22, 32). To map the regions interacting with LOX-PP, deletion mutants of these domains in Hsp70 were constructed as described by Milarski and Morimoto (43) (Fig. 3A). Hsp70 ΔBglII lacks amino acids 121 to 427 in the BglII-BglII fragment, which codes for the ATPase domain, and Hsp70 ΔSmaI lacks amino acids 439 to 617 in the SmaI-SmaI fragment, which codes for the peptide binding domain. These Hsp70 constructs were coexpressed with GST–LOX-PP or GST, as a control, in HEK293T cells. Following pulldown of the GST proteins, bound proteins were identified by Western blotting using a Flag antibody (Fig. 3B, top panel). GST–LOX-PP brought down Hsp70 ΔBglII as well as Hsp70 WT but not Hsp70 ΔSmaI (Fig. 3B, GST pulldown lanes). GST failed to bring down any proteins, as expected. Furthermore, analysis of aliquots of each of the original lysates confirmed essentially equal loading of expressed proteins and efficient pulldown of all of the GST proteins (Fig. 3B, lower panel). Thus, the peptide binding domain of Hsp70 mediates interaction with LOX-PP.

Fig. 3.

Mapping of the binding site for LOX-PP on Hsp70. (A) A schematic representation of the Hsp70 mutants used in this study is shown. Full-length (WT) or deletion mutant Hsp70 cDNAs (ΔBglII and ΔSmaI) were inserted into the pFLAG-CMV-2 vector. Positions of the ATPase and peptide binding (bdg.) domains and the acidic motif EEVD are as indicated. (B) GST or GST–LOX-PP was coexpressed with full-length FLAG-Hsp70 (WT) or mutant FLAG-Hsp70 ΔBglII or FLAG-Hsp70 ΔSmaI in HEK293T cells. GST proteins were precipitated with glutathione-Sepharose 4B beads and bound proteins subjected to Western blotting with anti-FLAG and anti-GST antibodies. For estimation of the amounts of expressed proteins, 4% of each of the lysates was separated and immunoblotted (Input panels).

Mapping the domains of LOX-PP mediating interaction with Hsp70.

Structural prediction and circular dichroism analyses of the propeptide region of LOX-PP indicated that it assembles as an intrinsically disordered protein (IDP) (44, 70), which occurs in proteins that take on structures once assembled in a complex. To begin to map the region(s) mediating binding with Hsp70, a series of progressive C-terminal (ΔC1 to ΔC5) and N-terminal (ΔN1 to ΔN5) deletion mutants of LOX-PP were constructed (Fig. 4A). Expression of the deletion proteins was confirmed following transfection into HEK293T cells (Fig. 4B, bottom panel). GST or GST-tagged proteins were pulled down with glutathione-Sepharose 4B beads, and the binding of Hsp70 and α-tubulin assessed by Western blotting. Hsp70 binding was reduced when the region between aa 78 and 100 was deleted from the C terminus or when aa 26 to 62 were removed from the N terminus (Fig. 4B, top panel). To further map the LOX-PP-interacting domain, additional deletion constructs within the region of aa 25 to aa 101 were prepared and analyzed (Fig. 4C). While ΔM1 and ΔM2 retained a low level of interaction with Hsp70 and α-tubulin, the ΔM3 LOX-PP variant, with a deletion of aa 26 to 100, failed to interact with either protein (Fig. 4D and data not shown). Finally, to test the ability of this region for interaction, the M1 construct, which expresses a peptide consisting of this region (aa 26 to aa 100), was prepared. The M1 peptide interacted with both Hsp70 and α-tubulin (Fig. 4E). Thus, the domain of LOX-PP that interacts with Hsp70 and α-tubulin is mapping internally to aa 26 to aa 100.

Fig. 4.

LOX-PP binding to Hsp70 maps to aa 26 to 100 of LOX-PP. (A) Schematic representation of the initial LOX-PP mutants used in this study is shown. SP, signal peptide. (B) HEK293T cells were transfected with plasmids expressing the deletion mutants indicated in panel A. GST constructs were pulled down with glutathione-Sepharose 4B beads, and the bound proteins were detected by Western blotting with antibodies against Hsp70, α-tubulin (α-tub.), and GST (upper panels). For estimation of the amounts of expressed proteins, 4% of each of the lysates was separated and immunoblotted (Input). (C) Schematic representation of additional deletion constructs of LOX-PP is shown. (D and E) HEK293T cells were transfected with EV or plasmids expressing WT, ΔM1, ΔM2, and ΔM3 (D) or EV, WT, ΔM3, and M1 (E). GST constructs were pulled down and bound proteins detected by Western blotting as described for panel B. (Input) Ten percent of each of the lysates was separated and immunoblotted.

LOX-PP reduces Hsp70-mediated protein folding and cell survival.

To elucidate the functional effects of LOX-PP on Hsp70, we first measured whether LOX-PP alters Hsp70 expression. No change in either its level or its rate of decay was seen (not shown). Next, the effects of LOX-PP on the ability of Hsp70 to promote correct refolding were assessed using a luciferase protein refolding assay. A cytoplasm-localized luciferase (Luc-Cyto), in which the C-terminal peroxisomal localization signal has been destroyed as described in Materials and Methods, was employed. Transfected Luc-Cyto was inactivated by incubating HEK293T-EV, HEK293T–LOX-PP WT, or HEK293T–LOX-PP ΔM3 cells at 45°C for 30 min. After a subsequent 1-, 2-, or 3-h recovery period at 37°C, luciferase activity was assayed using a luminometer. A decreased ability to promote luciferase refolding was seen over the time course in the HEK293T–LOX-PP WT versus HEK293T-EV cells (Fig. 5 A). In contrast, the LOX-PP-ΔM3 deletion, which cannot interact with Hsp70, had no effect on luciferase refolding compared to what was found for the cells with the EV DNA. Thus, LOX-PP decreases the ability of Hsp70 to promote protein folding, and this requires the region of association.

Fig. 5.

LOX-PP reduces Hsp70 chaperone and cell survival functions. (A) HEK293T cells were transfected with a cytoplasm-localized firefly luciferase expression plasmid and either GST–LOX-PP WT (WT), GST–LOX-PP-ΔM3 (ΔM3), or GST (EV) using Lipofectamine 2000. After 24 h, 20 μg/ml cycloheximide was added to prevent further protein synthesis. The cultures were incubated for 30 min at 37°C and then shifted to 45°C for 30 min to denature the luciferase protein. The cells were then incubated at 37°C to recover the folding of luciferase. At the indicated time points, the cells were lysed, protein concentrations were measured, and luciferase activities of equal amounts of protein were determined using a luciferase assay system (upper panels). (Lower panels) Western blotting was performed for GST and α-tubulin, which confirmed efficient synthesis of the GST proteins and equal loading, respectively. (B) ZR-75 and NF639 cells (left and right panels, respectively) were transfected with vectors expressing GST–LOX-PP WT (WT), GST–LOX-PP-ΔM3 (ΔM3), or GST (EV). After 24 h, cultures were incubated for 10 min at 45°C and then returned to 37°C for 24 h. (Upper panels) ZR-75 and NF639 cells were stained with 0.4% trypan blue and positive (dead) cells counted. The results from 3 individual experiments (means ± SD) are shown. (Lower panels) WCEs were subjected to immunoblotting for GST and α-tubulin, as described for panel A.

As another measure of the effects of LOX-PP on Hsp70 function, the ability of Hsp70 to protect cells from heat shock-induced cell death was assessed. Specifically, we tested whether survival of ZR-75 breast cancer cells is affected by expression of either the LOX-PP WT or the ΔM3 mutant following heat shock at 45°C for 10 min and recovery for 24 h. Cell death was measured by uptake of trypan blue (Fig. 5B). While only ∼5% of ZR-75 cells died with EV DNA, an approximately 5-fold increase in cell death was seen upon expression of LOX-PP WT protein. In contrast, the LOX-PP-ΔM3 mutant had no detectable effect on survival of heat-shocked ZR-75 cells (Fig. 5B, left panels). Western blotting was performed for GST and α-tubulin, which confirmed efficient synthesis of the GST proteins and equal loading, respectively. Similarly, NF639 breast cancer cells were found to be substantially more sensitive to heat-shock induced cell death following ectopic expression of LOX-PP WT protein but not LOX-PP-ΔM3 mutant (Fig. 5B, right panels). Thus, LOX-PP enhanced the sensitivity of ZR-75 and NF639 breast cancer cells to heat shock. Together, these studies indicate that LOX-PP has the ability to decrease the chaperone function of Hsp70. The inability of the ΔM3 mutant to either prevent refolding or induce cell death suggests that these effects are mediated by interaction of Hsp70 with the propeptide.

LOX-PP associates with c-Raf, an Hsp70-interacting protein.

Hsp70 has been found to interact with a diverse set of proteins in mediating its various functions. To begin to address the functional role of the LOX-PP interaction with Hsp70, we tested for association of LOX-PP with Hsp70-interacting proteins that are mediators of signaling pathways known to be affected by LOX-PP in breast cancer cells (Fig. 6 A). Notably GST–LOX-PP brought down c-Raf and Apaf-1 in addition to Hsp70, whereas no association was detected with the Hsp70-interacting proteins Akt, B-Raf, epidermal growth factor receptor (EGFR), Erk1/2, and MEK1/2 (Fig. 6A) and cytochrome c, JNK1, Src, or poly(ADP-ribose) polymerase (PARP) (data not shown). Interestingly, we have noted that ectopic LOX-PP expression reduced Erk1/2 activation by the Ras/Raf pathway in NF639 breast cancer cells, which are driven by a Her-2/neu to Ras signaling cascade, and in Ras-transformed NIH 3T3 fibroblasts (30, 44). Thus, we sought to test the ability of endogenous c-Raf and LOX-PP to associate in these lines. Triton X-100-soluble lysates from NF639 and NIH 3T3 cells were immunoprecipitated with an antibody against either LOX-PP or Hsp70 or its isogenically matched antibody, rabbit IgG or mouse IgG, respectively (Fig. 6B and C). The antibody against LOX-PP brought down Hsp70 and c-Raf; interestingly, in both lines, the pulldown of c-Raf appears somewhat more effective than that of Hsp70 (Fig. 6B and C). The LOX-PP antibody precipitated the precursor Pro–LOX, which contains the peptide on its amino-terminal domain, and the processed LOX-PP peptide. The antibody against Hsp70 brought down both LOX-PP and Pro–LOX as well as c-Raf, consistent with interaction via the propeptide domain. Notably, the LOX-PP produced extracellularly by proteolytic cleavage of Pro–LOX, which was detected at only low levels in the cell extracts, as judged by the input samples, was quite readily detected in association with Hsp70 presumably in the intracellular compartment. To further assess the domains of interaction, NF639 cells were transfected with vectors expressing either LOX-PP, Pro–LOX, which contains LOX-PP and the LOX enzyme domains, or the LOX enzyme alone. The interaction with c-Raf was observed with LOX-PP and Pro–LOX but not the enzyme (Fig. 6D). Thus, LOX-PP associated quite effectively with c-Raf, as well as Hsp70, in these breast cancer cells, which is consistent with our previous observations that ectopic LOX-PP expression in the NF639 and NIH 3T3 cells inhibits the activity of Erk1/2 kinases (30, 44), which are downstream of c-Raf.

We next tested for direct interaction of LOX-PP and c-Raf, using a GST-tagged protein (Fig. 6E, inset). Recombinant LOX-PP–myc-His was incubated with either GST or GST-c-Raf for 3 h, and then GST-tagged and associated proteins were purified using glutathione-Sepharose 4B beads. The protein bound to the resin was visualized by immunoblotting with a His tag antibody or Coomassie blue staining (Fig. 6E, upper or lower panel, respectively). A band at the position of full-length LOX-PP was seen with GST-Raf but not GST alone, indicating that LOX-PP can associate directly with c-Raf, although this interaction was weaker than expected based on the coimmunoprecipitation seen in cell extracts as described above (Fig. 6B and C). This might be due to the lack of a modification(s) on the bacterially expressed c-Raf necessary to promote its direct interaction with the propeptide or the possibility that Hsp70 might stabilize the interaction of c-Raf and LOX-PP.

Interaction with LOX-PP reduces Raf signaling via Erk.

The finding that LOX-PP associates with c-Raf led us to further assess the effects of the propeptide on signaling downstream of this kinase. To test whether LOX-PP reduces activation of MEK1/2 by c-Raf, plasmids expressing GFP-tagged c-Raf or control GFP were transfected into HEK293T cells in the absence or presence of GST–LOX-PP or GST (Fig. 7 A). Resulting cell lysates from GST–LOX-PP- and c-Raf-expressing cells displayed reduced MEK1/2 activity compared to those from the GST alone, consistent with inhibition of c-Raf activity by LOX-PP. The ability of LOX-PP to inhibit Erk1/2 activity in two additional breast cancer lines, Hs578T and ZR-75, was assessed next. Ectopic LOX-PP expression reduced Erk1/2 activity in both lines without affecting total Erk1/2 levels (Fig. 7B). Next, we tested whether the region comprising aa 26 to 100 of LOX-PP is necessary for its interaction with c-Raf as well as with Hsp70. Lysates were prepared from HEK293T cells ectopically expressing either GST, GST–LOX-PP WT, GST–LOX-PP ΔM3 (Δ26-100), or GST–LOX-PP M1 (26-100) and subjected to glutathione-Sepharose 4B beads. Binding of c-Raf to full-length LOX-PP and to LOX-PP M1 containing only aa 26 to 100 was readily detected, whereas no binding was seen with the GST–LOX-PP ΔM3 protein, which is missing this region, or with the GST control protein (Fig. 7C). The mutants were next compared to WT LOX-PP for their effects on activation of MEK1/2 by c-Raf, as described above and shown in Fig. 7A. The GST–LOX-PP ΔM3 protein was unable to reduce c-Raf activity, as judged by levels of phospho-MEK measured, whereas the WT and M1 proteins were effective. The data from three independent experiments were quantified using ImageJ software and the average level (± SD) relative to the level for EV, set at 100%, for the LOX-PP ΔM3 protein was 108% ± 13%, whereas values of 74% ± 7% and 63% ± 10% were obtained for the LOX-PP WT and M1 proteins, respectively (data not shown).

Fig. 7.

The region comprising aa 26 to 100 of LOX-PP that interacts with Hsp70 is necessary for inhibition of c-Raf kinase and the transformed phenotype. (A) HEK293T cells were transfected with the plasmids expressing GFP-tagged c-Raf (Raf) or the GFP (G) control in the presence of GST–LOX-PP (PP) or the GST (G) control, as indicated. Cell lysates (20 μg) were analyzed for the effects on the activity of the c-Raf downstream mediator MEK1/2 by Western blotting using anti-phospho-MEK1/2 (Ser-217/221) and for total MEK1/2 as well as for GFP, GST, and β-actin as loading controls. (B) Hs578T and ZR-75 cells were transfected with plasmids expressing V5-tagged LOX-PP or EV DNA. Samples of WCEs (20 μg protein) were subjected to immunoblotting using antibodies against phospho-Erk1/2 (Thr-202/Tyr-204), total Erk1/2, the V5 tag, and β-actin. (C) HEK293T cells were transfected with the expression plasmids for the indicated GST-tagged LOX-PP proteins (WT, ΔM3, or M1) or GST alone (EV). Proteins were purified using glutathione-Sepharose 4B beads and subjected to Western blotting for c-Raf and GST. For estimation of the amounts of expressed proteins, 4% of each of the lysates (Input) was immunoblotted. (D) Bacterially expressed GST, GST–LOX-PP WT, or GST–LOX-PP ΔM3 proteins were purified (bottom panel) and their effects on the ability of NF639 cells to form colonies after 2 weeks of growth in soft agar assessed. Plates were photographed at ×4 magnification. (E) GST (EV), GST–LOX-PP WT, or GST–LOX-PP ΔM3 was ectopically expressed in NF639 breast cancer cells. After 48 h, cells were subjected to a migration assay for 16 h in triplicate, and cells that migrated to the lower side of the filter were stained with crystal violet and quantified by spectrometric determination at A₅₇₀. Averages ± SD are given. P values were calculated using Student's t test. *, P < 0.01.

To test whether the interaction with c-Raf and Hsp70 is necessary for the observed inhibition of the transformed phenotype by LOX-PP (44, 75), the effects of LOX-PP WT versus ΔM3 deletion on anchorage-independent growth and migration of NF639 cells were compared. Purified bacterially expressed GST, full-length GST–LOX-PP WT, or GST–LOX-PP ΔM3 proteins (Fig. 7D, bottom panel) were tested for their effect on soft agar colony formation by NF639 cells. While LOX-PP WT protein robustly inhibited growth of NF639 cells in soft agar, the GST–LOX-PP ΔM3 protein had no effect compared to the GST protein alone (Fig. 7D). Migration assays were performed 48 h after ectopic expression of GST, full-length GST–LOX-PP WT, or GST–LOX-PP ΔM3 in NF639 breast cancer cells (Fig. 7E). Consistent with our previous studies, an ∼40% reduction was seen with expression of full-length LOX-PP. In contrast, no reduction in NF639 cell migration was seen with the LOX-PP ΔM3 mutant (Fig. 7E). Thus, the domain comprising aa 26 to 100, which mediates the interaction of LOX-PP with c-Raf, is required for inhibition of c-Raf activity and reduction in the transformed phenotype by the propeptide.

LOX gene knockdown or knockout induces a more transformed phenotype.

To determine the effects of knockdown of the endogenous LOX-PP protein on Erk activation, two siLOX RNAs (oligonucleotide A or oligonucleotide B) and a scrambled control siRNA were employed. NF639 and NIH 3T3 cells were transfected with the siRNA at 20 nM (final concentration) and cultures incubated for 48 h. Media and isolated whole cell extracts (WCEs) were subjected to immunoblot analysis for LOX proteins using an antibody that recognizes the propeptide domain (Fig. 8 A). The data confirmed the ability of both siLOX RNAs to effectively knock down secreted and intracellular LOX-PP and its precursor Pro–LOX, as expected. Notably, the decreases in LOX-PP levels were accompanied by substantial increases in active phospho-Erk in both NF639 and NIH 3T3 cells, whereas no changes were observed in total Erk1/2 (Fig. 8A). We next tested the effects of LOX knockdown on the ability of NF639 cells to migrate. A robust increase in the ability of NF639 cells to migrate was noted upon treatment with the two siLOX RNAs compared to the level for the control siRNA, with oligonucleotide B being somewhat more effective (Fig. 8B, left panel). Next the effects of the siRNAs were tested on invasion by NF639 cells. The ability of NF639 cells to invade through Matrigel was significantly increased upon treatment with the two siLOX RNAs compared to the level for the control siRNA, with oligonucleotide B causing a larger increase (Fig. 8B, right panel).

Fig. 8.

Knockdown of LOX gene expression enhances Erk signaling and cell migration and invasion. (A) NF639 and NIH 3T3 cells were transfected with 20 nM (each) siLOX RNAs oligonucleotide A or oligonucleotide B or scrambled control siRNA for 48 h. Samples of medium (20 μl) and WCEs (20 μg) were subjected to immunoblot analysis for LOX using an antibody that recognizes the propeptide domain, pErk1/2, total Erk1/2, and β-actin. (B) NF639 cells were transfected as described for panel A. (Left panel) After 48 h, cells were subjected to a migration assay for 6 h in triplicate as described for Fig. 7E. Averages ± SD are given. P values were calculated using Student's t test. *, P < 0.01. (Right panel) After 48 h, cells were subjected to an invasion assay for 6 h in triplicate as described in Materials and Methods. Averages ± SD are given. P values were calculated using Student's t test. *, P < 0.01. (C) WT or Lox^−/− MEFs were subjected to a migration assay for 16 h in triplicate and the data quantified as described for Fig. 7E. Averages ± SD are given. P values were calculated using Student's t test. *, P < 0.01.

MEFs were derived from mice null for the LOX gene, which die perinatally (38), and from WT C57BL/6 mice and assayed for their ability to migrate. MEFs from Lox^−/− mice displayed increased migration compared to those from the wild-type mice (Fig. 8C). Previously, we demonstrated that LOX-PP inhibits NF-κB activity in Ras-transformed NIH 3T3 cells (30). Consistently, a 4.28 ± 0.94-fold increase in the activity of an NF-κB element-driven reporter was observed in Lox^−/− MEFs compared to the level for MEFs from wild-type mice (not shown). Thus, decreased LOX gene expression is accompanied by a more transformed phenotype, as judged by increases in Erk and NF-κB activities and in the ability of cells to migrate and invade through Matrigel.

LOX gene knockdown leads to a more mitochondrial localization of c-Raf.

Rapp and coworkers have observed that activation of its downstream targets Erk1/2 by c-Raf is accompanied by a more mitochondrial localization of c-Raf (15). This led us to test the effects of LOX gene knockdown on c-Raf localization to the mitochondria. NF639 cells were treated with 20 nM (final concentration) of siLOX RNA oligonucleotide B or with scrambled control siRNA. After 48 h of incubation, mitochondrial proteins were isolated and subjected to immunoblotting for c-Raf (Fig. 9A, upper panel). Knockdown of LOX-PP was associated with an increased localization of c-Raf to the mitochondria. Western blotting for the mitochondrial protein COX-1 confirmed equal loading. Similarly, the amount of c-Raf localized in the mitochondrial fraction from the Lox^−/− MEFs was substantially higher than that from MEFs from wild-type C57BL/6 mice (Fig. 9A, lower panel). Mitotracker and immunofluorescence analysis were next used to visualize how the localization of c-Raf to the mitochondria was altered by knockdown of LOX-PP in NF639 cells (Fig. 9B). In cells treated with control siRNA, c-Raf displayed both a cytoplasmic and a mitochondrial localization. A substantial increase in c-Raf staining in the mitochondria was detected with siLOX RNA oligonucleotide B treatment of NF639 cells (Fig. 9B). Furthermore, the mitochondria appeared more clustered and perinuclear than filamentous, reminiscent of the changes observed with induction of a constitutive active c-Raf protein in NIH 3T3 cells by Rapp and coworkers (15). Similar data were obtained with the WT and Lox^−/− MEFs (data not shown). Thus, LOX-PP appears to reduce the localization of c-Raf to the mitochondria and to decrease Erk signaling.

Fig. 9.

Knockdown of LOX gene expression enhances targeting of c-Raf to the mitochondria. (A) NF639 cells treated with control siRNA or siLOX oligonucleotide B RNA for 48 h (upper panels). The mitochondrial fractions (10 μg) were subjected to 10% SDS-PAGE and analyzed by immunoblotting for c-Raf and COX-1. The mitochondrial fraction (10 μg) of WT or Lox^−/− MEFs was subjected to immunoblotting for c-Raf and COX-1, as described above (lower panels). (B) NF639 cells were transfected as described for panel A. After 48 h, cells were incubated with 250 nM Mitotracker for 30 min, fixed with methanol, and processed for immunofluorescence analysis as described in Materials and Methods. Bars: 5 μm.

DISCUSSION

Here, a novel mechanism of action of the Ras inhibitor protein LOX-PP that occurs via its association with Hsp70 and c-Raf, leading to reduced Hsp70 function and c-Raf-mediated signaling and a transformed phenotype in breast cancer cells, was elucidated. Hsp70 was identified as a binding partner of LOX-PP using copurification-mass spectrometry, and this interaction was shown to reduce Hsp70 functional ability to promote refolding and survival from heat shock-induced death. The domains of interaction were mapped to the peptide binding region of Hsp70 and to aa 26 to aa 100 of LOX-PP. Testing of the known Hsp70 clients revealed that LOX-PP is also associated with c-Raf. This identification of c-Raf is consistent with previous observations showing that LOX-PP reduces signaling downstream of c-Raf in breast, lung, pancreatic, and prostate cancer cells, including Erk1/2 activation in growing cells or upon stimulation with serum or basic fibroblast growth factor (bFGF) (30, 44, 52, 73) and migration and the transformed phenotype (30, 44, 45, 73). Furthermore, we now show that knockdown or knockout of LOX-PP has the inverse effects, i.e., leads to increased cell migration and invasion, and Erk1/2 and NF-κB activities. Knockdown of LOX-PP levels also led to increased localization of c-Raf to the mitochondria, which correlated with the activation of Erk1/2. These findings are consistent with the report by Rapp and coworkers (15) showing that active c-Raf localizes to the mitochondria and that this localization plays an important role in activation of Erk1/2. Thus, LOX-PP forms complexes comprising Hsp70 and c-Raf, which reduces the signaling mediated by these two factors, which have been shown to contribute to the progression of breast and other cancer cells (2, 4, 49, 72). Overall, our findings suggest the potential use of this peptide in treatment of cancers driven by signaling via these pathways.

LOX-PP appears to associate directly with Hsp70 in breast cancer cells, as judged by in vitro binding assays and coimmunoprecipitation analysis. Elevated Hsp70 has been detected in many cancers, including breast cancers (6, 48), and the expression level of Hsp70 correlates with metastasis, resistance to anticancer drugs, and poor prognosis in many human cancers (reviewed in reference 8). Hsp70 prevents caspase-dependent and -independent cell death triggered by apoptotic stimuli, such as heat shock, tumor necrosis factor, serum withdrawal, and chemotherapeutic agents. Furthermore, Hsp70 has been shown to downregulate Jun N-terminal protein kinase (JNK), p38 mitogen-activated protein kinase (MAPK), and caspase and to stabilize lysosomal membrane integrity (14). Notably, the association with LOX-PP functionally reduced the ability of Hsp70 to promote correct folding of luciferase protein and resulted in substantially increased cell death upon heat shock of breast cancer cells. These findings lead us to hypothesize that the interaction with LOX-PP compromises Hsp70 functions essential for growth and survival of tumor cells.

Hsp70 family proteins are highly conserved and have three regions mediating interaction with various proteins, including an ATPase domain in the N-terminal region, a peptide binding domain in the C-terminal region, and an acidic motif (EEVD) at the C terminus. The peptide binding domain of Hsp70 has been reported to bind to PKCβII, p53, Rictor, apoptosis-inducing factor (AIF), JNK1, TRAF2, TRAF6, Ku70, and MstI (7, 11, 16, 29, 37, 40, 54, 58, 59). Meanwhile, Hsp70 interacts with Hip, Bag-1, Bax, hYVH1, PARP-1, CD40, and Ask1 via its ATPase domain (3, 17, 25, 26, 36, 53, 64) and with Hop (Hsp70/Hsp90-organizing protein) and CHIP via the EEVD motif (1, 12). In this study, mutational analysis demonstrated that the peptide binding domain of Hsp70 is necessary for the interaction with LOX-PP. Interestingly LOX-PP does not interact with the common Hsp70 cochaperone Hsp90 in either breast cancer or HEK293T cells. While it is somewhat rarer to be a sole client of Hsp70, other examples include Rb, CD40, and TRAF6 (3, 7, 24, 28). Importantly, LOX-PP interacted with c-Raf, an Hsp70-associating protein (65), to inhibit the signaling via Erk1/2 that promotes a migratory phenotype in breast cancer cells. It should be noted that we did not observe interaction of LOX-PP with MEK1/2, Erk1/2 or B-Raf kinase, or H-Ras (unpublished results). These results suggest that our previous data showing that LOX-PP reduces Ras-mediated Erk1/2 activation (30, 44, 45, 73) is due to the ability of LOX-PP to reduce the activity of its upstream kinase c-Raf. Since the LOX gene was originally identified as the Ras recision gene with ability to suppress Ras-mediated transformation (10, 35), an active LOX enzyme which is lacking the propeptide region was found to promote an invasive phenotype in breast cancer cells (13, 55, 56). These paradoxical results can be explained, in part, by our findings that LOX-PP, but not the LOX enzyme, can interact with c-Raf and reduce its activity.

Previous structure prediction studies of LOX-PP using DISOPRED, GlobPlot and DisProt, and circular dichroism (CD) analysis have indicated that the propeptide assembles as an intrinsically disordered protein (44, 70). Here, the domains of LOX-PP mediating binding with Hsp70 were mapped to aa 26 to 100. This sequence contains a proline-rich region between aa 26 and 34 and an arginine-rich region between aa 62 and 72. While prolines within a peptide sequence disrupt ordered secondary structures, proline-rich sequences are important for mediating interactions with several protein-protein interaction domains, including the Src homology 3 (SH3), Ena/VASP homology 1 (EVH1), and WW domains (74). Indeed, in more recent LOX-PP copurification-mass spectrometry analysis of ZR-75 human breast cancer cells, an adaptor protein containing an SH3 domain was identified as a novel LOX-PP-interacting protein (S. Sato, unpublished observations), and work is in progress to test for their association. Arginine is known to be an electrically positively charged amino acid and prefers to be on the outside of the proteins. Consistently, hydropathy plot analysis predicts that this region of LOX-PP is on the surface. Lastly, the carboxy-terminally truncated construct of LOX-PP appeared to exhibit stronger binding to Hsp70 and tubulin than full-length LOX-PP. This finding suggests that aa 116 to 162 in the carboxy-terminal region may contain a domain that suppresses the interaction with Hsp70 and tubulin. Further study is required to address this possibility.

Previous work implicated the carboxy-terminal region of LOX-PP in productive secretion of Pro–LOX. Specifically, Kagan and coworkers (31) established that the amino terminus of the propeptide region of Pro–LOX contains a signal sequence essential for normal secretion. Sommer and Mecham and their collaborators found that Pro–LOX mutants that lack aa 23 to 157 were not secreted (62, 67). These findings suggest that the carboxy-terminal region of the propeptide plays a critical role in Pro–LOX secretion in addition to the signal peptide. To test this possibility, we analyzed the localization of WT and M1 LOX-PP in breast cancer cells by immunofluorescence analysis. While WT LOX-PP localized to the Golgi apparatus in Hs578T cells, as judged by colocalization with Golgi marker protein, consistent with previous work using osteoblasts (19), the M1 LOX-PP mutant localized throughout the cytosol and failed to associate with the Golgi apparatus (data not shown). Taken together, these findings suggest that the c-Raf suppressor domain of LOX-PP maps to the region of aa 26 to 100 and that the carboxy-terminal region is required for proper localization and correct processing. In this regard, it is known that glycosylation occurs within the Golgi apparatus during secretion (50), and it has been reported that LOX-PP is posttranslationally modified by N-glycosylation and O-glycosylation (68, 70).

Here, tubulins were identified as additional binding proteins of LOX-PP in breast cancer and HEK293T cells. Notably, tubulin was not an intermediate protein for the association of Hsp70 with LOX-PP. This finding suggests that the tubulin–LOX-PP and Hsp70–LOX-PP complexes are independent of each other. Interestingly, the colocalization and association of LOX-PP with tubulin in the microtubules of differentiated osteoblasts were observed previously using confocal microscopy and overlay binding analysis (19). The microtubule network plays important roles in the control of cell cycle progression, cell movement, vesicle transport, and signal transduction (34, 69, 71). It will be important to determine the effects of LOX-PP association on these diverse functions.

Ectopic LOX-PP expression was seen to enhance cell death of breast cancer cells upon heat shock. Recently, we reported that LOX-PP sensitizes pancreatic and breast cancer cells to doxorubicin-induced apoptosis, which is caspase dependent (46). Interestingly, we observed that Apaf-1 was also coprecipitated with LOX-PP. Hsp70 has been shown to block Apaf-1/cytochrome c-mediated caspase activation (61), suggesting that LOX-PP may promote apoptosis via overriding the inhibition of Apaf-1 protease activity mediated by its interaction with Hsp70 protein. In summary, LOX-PP has been shown to interact with Hsp70 and c-Raf and thereby to reduce chaperone function and signaling via Erk1/2 that promote survival and a more transformed phenotype in breast cancer cells. It will be important to test whether this association can be extended to other cancers driven by Ras signaling.