BAG-1 is a Novel Cytoplasmic Binding Partner of the Membrane Form of Heparin-binding EGF-like Growth Factor: A Unique Role for proHB-EGF in Cell Survival Regulation

Several cell functions related to growth and survival regulation have been attributed specifically to the membrane form of heparin-binding EGF-like growth factor (proHB-EGF), rather than to the diffusible, processed HB-EGF isoform. These findings suggest the existence of a functional binding partner specifically for the membrane form of the growth factor. In this study we have identified the prosurvival cochaperone, BAG-1, as a protein that interacts with the cytoplasmic tail domain of proHB-EGF. Interaction between BAG-1 and the 24-amino acid proHB-EGF cytoplasmic tail was initially identified in a yeast two-hybrid screen and was confirmed in mammalian cells. The proHB-EGF tail bound BAG-1 in an hsp70-independent manner and within a 97 amino acid segment that includes the ubiquitin homology domain in BAG-1 but does not include the hsp70 binding site. Effects of BAG-1 and proHB-EGF co-expression were demonstrated in cell adhesion and cell survival assays and in quantitative assays of regulated secretion of soluble HB-EGF. Because the BAG-1 binding site is not present on the mature, diffusible form of the growth factor, these findings suggest a new mechanism by which proHB-EGF, in isolation from the diffusible form, can mediate cell signaling events. In addition, because effects of BAG-1 on regulated secretion of soluble HB-EGF were also identified, this interaction has the potential to alter the signaling capabilities of both the membrane-anchored and the diffusible forms of the growth factor. In this study we report that the cytoplasmic tail of proHB-EGF interacts with BAG-1, a multifunctional protein first identified as a binding partner of the anti-apoptotic protein Bcl-2 (11). BAG-1 associates with several signaling molecules and is capable of suppressing apoptosis. Our findings suggest a novel mechanism through which proHB-EGF might mediate physiological processes related to growth, adhesion and cell survival. MC2-proHB-EGF-AP cells were either treated or untreated by 0.4 mM etoposide for 24 h and cells were fixed with incubation with 2% paraformaldehyde for 1 h at room temperature. Cell permeabilization was performed with 0.1% Triton-X-100 in PBS for 3 min on ice. Cells were washed with PBS, blocked for 30 min with PBS/0.1% BSA/0.075% glycine (blocking buffer) and incubated with anti-AP monoclonal Ab (8B6, Sigma) (1:250) and rabbit polyclonal anti-BAG-1 antibody (1:1000) (N20, Santa Cruz) diluted in blocking buffer for 1 h at room temperature. After washing with blocking buffer, cells were incubated with Texas-red-conjugated donkey anti-mouse IgG and fluorescein (FITC)-conjugated donkey anti-rabbit IgG (Jackson ImmunoResearch, Inc. West Grove, PA) for 45 min. Slides were washed extensively with blocking buffer prior to mounting and were viewed using a BioRad 1024 Laser Scanning Confocal Imaging System. Up to forty serial optical sections (approximately 0.5 m m section thickness) were collected on informative cells. Individual channels of double labeled cells were collected as two separate series and merged in Confocal Assistant (written by Todd Breljie).

Cell culture and transfection. LNCaP and PC-3 cells were grown in RPMI-1640/10% FBS. CHO-K1 cells were cultured in F12K/10%FBS. NRK52E and COS7 cells were grown in DMEM/10% FBS. MC2 cells were cultured in T medium as described (13). All cells were maintained in a humidified atmosphere of 95% air/5% CO 2 at 37 o C.
MTT and DNA fragmentation assays were performed as described HB-EGF secretion assay. Secretion of HB-EGF was measured by determining levels of alkaline phosphatase (AP) in the medium using cells expressing proHB-EGF-AP fusion proteins as described previously (13). Briefly, 40,000 cells/well were seeded in 24-well plates and 24 h later, cells were stimulated by etoposide at the different concentrations under serum free conditions. Medium was collected 24 h later and AP activity was measured spectrophotometrically.
Statistical Analysis. Data were compared using a paired Student t-test. P values less than 0.05 were considered significant.

Results
The HB-EGF tail domain interacts with the prosurvival protein, BAG-

1
The 24-residue cytoplasmic tail of proHB-EGF exhibits a high degree of inter-species sequence conservation (95% amino acid identity between mouse and human), suggesting an important functional role for this region of the protein. In a previous study, proHB-EGF was shown to protect NRK52E renal epithelial cells from apoptosis induced by H 2 O 2 or etoposide treatment (9).
Soluble HB-EGF was not able to replicate this cytoprotective effect. In order to determine if the proHB-EGF tail is involved in this process, NRK52E cells were transfected with intact proHB-EGF or proHB-EGF tail-deleted expression constructs. Cells expressing the two forms of the protein were then challenged with etoposide or H 2 O 2 . NRK52E cells expressing the proHB-EGF construct exhibited less apoptosis than vector-only control cells ( Fig. 1), consistent with findings reported by Takemura et al. (9). In contrast, cells expressing the taildeleted construct exhibited a similar level of apoptosis to the vector-only cells, suggesting a role for the tail domain in cytoprotection from apoptosis inducers.
These observations led us to search for proteins that interact with the proHB-EGF cytoplasmic domain. We screened a yeast two-hybrid expression library, constructed from a human prostate (LNCaP) xenograft tumor growing in a mouse host, with the proHB-EGF tail domain (a.a. 185-208), using a LexAbased system. Approximately 10 6 independent clones were screened. From the 6 strongest potential interactors identified in the screen, two clones, pJG45-B11 and pJG45-E68 ( Fig. 2A), each contained the entire open reading frame of the short form of the mouse protein, BAG-1 (219 amino acids, expected MW 24.5 kDa) (11). These two clones are identical but were derived from two independent yeast transformations. In follow-up experiments, BAG-1 and the HB-EGF tail also interacted in the yeast two-hybrid system when the cDNAs were switched into the opposite (prey<-->bait) plasmid vectors.
To confirm the association between BAG-1 and HB-EGF, a GST fusion protein containing the HB-EGF cytoplasmic tail (GST-HB-EGF  ) was constructed and used in pull-down assays. COS7 cells were transiently transfected with expression plasmids encoding mouse BAG-1. Lysates from these cells or from LNCaP cells (to test for binding with the native/endogenous, human form of BAG-1) were incubated with purified GST-HB-EGF  . A complex was formed between GST-HB-EGF (185-208) and BAG-1 but not between BAG-1 and GST alone (Fig. 2B). The converse experiment was also performed with a GST-BAG-1 fusion protein, using lysates from cells expressing AP-tagged proHB-EGF. In these experiments, a complex was formed between proHB-EGF and GST-BAG-1, but not between proHB-EGF and GST, or between GST-BAG-1 and proHB-EGF in which the tail domain was deleted (Fig. 2C). Complex formation between BAG-1 and the proHB-EGF tail was also demonstrated by Far-Western blot, in which the HB-EGF tail was immobilized and the interaction occurred on blotting membranes instead of in solution (Fig. 2D).
The 219 amino acid form of BAG-1 identified in the screen contains a ubiquitin homology domain (residues 37-73) and a central region (residues 90-172) that binds to Bcl-2 (11). Its carboxyl-terminal domain is required for direct interaction with the ATPase domain of hsp70 heat shock protein (16). We generated a GST-BAG-1(∆C) construct (residues 1-97), containing the ubiquitin homology region, and GST-BAG-1 (∆N) (residues 100-219), which carries binding sites for most of the known BAG-1 interactors. Complex formation with proHB-EGF was observed with GST-BAG-1 and GST-BAG-1(∆C), but not with 1(∆N) and hsp70 (Fig. 2E), demonstrating the capability of GST-BAG-1(∆N) to bind to a known BAG-1 binding protein despite its failure to bind to proHB-EGF.
This result also indicates that the BAG-1 interaction with proHB-EGF is not mediated by hsp70 and it rules out the possibility that aberrant folding of GST-BAG-1(∆N) is the reason for the absence of binding to HB-EGF. In a reciprocal experiment, GST-HB-EGF  was also able to form a complex with endogenous BAG-1 (MW 33-35 kDa) and endogenous hsp70 from human (LNCaP) cells (Fig. 2F).
We also investigated the dynamics of the BAG-1/HB-EGF tail interaction in cells induced to undergo apoptosis. LNCaP cells were treated with wortmannin, a PI3-kinase inhibitor that rapidly induces apoptosis in this cell line (15), and lysates were used in pull-down experiments. Interestingly, complex formation between the HB-EGF tail and endogenous BAG-1 diminished in a time-dependent manner following wortmannin treatment (Fig. 2G). Similar results were obtained when wortmannin-insensitive PC-3 cells were induced to undergo apoptosis by treatment with staurosporine, a protein kinase inhibitor (data not shown). These results suggest that the BAG-1/proHB-EGF interaction is not favored when cells undergo programmed cell death.
Taken together, these experiments 1) demonstrated a direct interaction between BAG-1 and the proHB-EGF cytoplasmic domain, 2) localized the HB-EGF interacting domain to within residues 1-97 of BAG-1 and 3) also revealed that BAG-1 can form a ternary complex with both proHB-EGF and hsp70 through interactions with these proteins at distinct binding sites.

proHB-EGF and BAG-1 functionally cooperate in vivo
To explore the possibility of a functional interaction between proHB-EGF and BAG-1, CHO cell populations were engineered sequentially to stably express either BAG-1, proHB-EGF, or both proteins. BAG-1+proHB-EGF-expressing cells exhibited a more epithelial-like cellular morphology, in comparison to cells expressing either BAG-1 or proHB-EGF alone or control vectors, or the parent cell, all of which exhibited a more fibroblastic appearance (Fig. 3). These data suggest that coexpression of both proteins confers functional properties on transfected cells that are not seen when each protein is expressed separately.
A similar requirement for proHB-EGF and BAG-1 co-expression to change a cellular phenotype was observed in other assays. BAG-1+proHB-EGF-expressing cells exhibited quantitatively reduced cell adhesion, as measured by sensitivity to trypsin/EDTA treatment, in comparison to cells expressing either BAG-1 or proHB-EGF or control plasmids (Fig. 4A). The presence of BAG-1 with proHB-EGF in CHO cell transformants also affected the sensitivity of these cells to certain apoptotic stimuli. BAG-1+proHB-EGF cells demonstrated increased resistance to apoptosis induced by etoposide, a topoisomerase inhibitor, in comparison to cells expressing either BAG-1 or proHB-EGF alone (Fig.4B). This resistance to apoptosis induction appeared to be confined to specific survival pathways, however, because BAG-1+proHB-EGF cells did not show synergistic protective effects when apoptosis was induced by staurosporine (data not shown).
We also compared secretion of soluble HB-EGF in response to apoptotic stimuli in proHB-EGF-and proHB-EGF(∆tail)-expressing MC2 cells transfected with BAG-1 and vector only. Etoposide treatment induced rapid secretion of soluble HB-EGF in cells expressing transfected BAG-1 but not in cells transfected with an empty vector (Fig. 5A). In contrast, in cells expressing a tail-deleted form of proHB-EGF, transfection with BAG-1 did not alter the secretion response to etoposide. EGFR levels in proHB-EGF and proHB-EGF(∆tail) expressing cells were equivalent (Fig. 5B), indicating that differential capture of the soluble HB-EGF ligand by the EGFR cannot account for the observed differences in the secretion response. Similar results were observed in the CHO cell background (data not shown). These data suggest that the proHB-EGF tail is involved in regulated processing of the cell-associated form of the growth factor to the soluble form and that BAG-1 is capable of modulating this process in a manner that is dependent on the presence of the tail domain.
Immunofluorescence confocal microscopy indicated that proHB-EGF and BAG-1 co-localized within cytoplasm vesicles and at the plasma membrane, consistent with the possibility that BAG-1 can affect trafficking and maturation of soluble HB-EGF (Fig. 6). BAG-1 and proHB-EGF colocalized sites diminished in frequency when cells were treated with etoposide.

Discussion
The results of this study demonstrate that the membrane form of HB-EGF interacts with the anti-apoptotic protein, BAG-1, and that this interaction is likely to have functional significance. The BAG-1/HB-EGF tail interaction was demonstrated in a number of independent assays, including yeast two-hybrid, GST-pull-down, and Far-Western blot methods. Evidence for colocalization of BAG-1 and proHB-EGF was also obtained with immunofluorescence confocal microscopy. In addition, cooperative effects of BAG-1 and proHB-EGF expression were observed in assays of cell adhesion, apoptosis, and growth factor secretion. Consistent with these findings, we found that deletion of the proHB- Importantly, our results suggest a novel mechanism whereby the membrane-anchored form of HB-EGF might alter cell function independently of the soluble form of the molecule. BAG-1 was originally identified as a Bcl-2binding protein but is now known to interact with and regulate a number of signaling proteins. Because the proHB-EGF tail and membrane-anchoring domains are removed from the mature growth factor by proteolytic cleavage, these results provide the first unambiguous mechanism whereby the HB-EGF pro-form could mediate processes distinct from those conferred by processed HB-EGF.
Several previous studies have identified bioactivities solely attributable to proHB-EGF, although the reason for the distinction between the precursor and secreted forms is not clear because both proteins presumably function principally by activating identical high-affinity receptor tyrosine kinases.
However, BAG-1 is the first protein to be identified that interacts with the proHB-EGF cytoplasmic domain in an ectodomain-independent manner. Because CHO cells express low or negligible EGFR levels (18) -2 (11,26). From these and other data, a critical role for BAG-1 in cell signaling related to cell survival mechanisms, but also to other processes, can be inferred despite uncertainty as to its precise mechanism of action.
The heat shock protein, hsp70, has been reported to be a favored BAG-1 interactor. In this case, BAG-1 appears to function as a regulator of protein folding and/or trafficking by acting as a competitive antagonist of the cochaperone, Hip (27). Although a number of proteins are thought to associate with BAG-1 because of its ability to bind hsp70 (16), we demonstrate in this study that proHB-EGF binds to a C-terminal deletion mutant of BAG-1 which does not bind hsp70 and that the HB-EGF interaction site on BAG-1 is distinct from the hsp70 interaction site. Furthermore, the HB-EGF tail-BAG-1 interaction occurs in yeast and yeast hsp70 is not a BAG-1 binding partner (16). Taken together, these data indicate that proHB-EGF and BAG-1 interact directly and in an hsp70-independent manner. This finding suggests that interaction of BAG-1 with proHB-EGF, and possibly with other regulatory proteins, may be functionally distinct from its role as a cochaperone in mechanisms of protein folding and/or stabilization.