Interaction between the BAG1S isoform and HSP70 mediates the stability of anti-apoptotic proteins and the survival of osteosarcoma cells expressing oncogenic MYC

The human cell line U2OS MYC-ER was generously provided by Dr. Martin Eilers (University of Würzburg, Würzburg, Germany). Cells were cultured in Dulbecco’s modified Eagle’s medium (DMEM, Mediatech) supplemented with 10% fetal calf serum (FBS, Gemini Bio-Products) at 37 °C in 5% CO₂.

BAG1L (50 kDa), BAG1M (46 kDa), and BAG1S (36 kDa) pcDNA3 expression plasmids were kindly provided by Dr. Graham Packham (University of Southampton, Southampton, UK). BAG1S C204A mutant was constructed by site-directed mutagenesis using QuikChange (Agilent Technologies). The ectopic BAG1 cell lines were generated by co-transfection of BAG1 plasmids and .25 fold of eGFP plasmid (Addgene). Transfection was performed using Continuum™ reagents according to the manufacturer’s protocol (Gemini Bio-Products). Cells were sorted by the Core Flow Cytometry Facility (Thomas Jefferson University, Philadelphia, PA) and cultured normally.

As indicated, cells were infected with lentiviral shRNA plasmids corresponding to BAG1 (NM_004323.4-1189s21c1) directed against the 3’UTR and control luciferase shRNA (SHC007) that were obtained from the TRC collection (Sigma-Aldrich). Lentiviral packaging plasmids (pCMV-R8.2 and pCMV-VSV-G) were cotransfected with shRNA vectors into 293 T cells (ATCC). Viral supernatants were collected, filtered, and added directly to target cells in the presence of 8 μg/ml polybrene (Sigma Aldrich). Cells were selected with with 8 μg/mL puromycin (Sigma-Aldrich) 24 h after infection.

The MYC-ER fusion protein was activated by adding 4-hydroxytamoxifen (4-OHT) (Sigma Aldrich) at a final concentration of 100 nM for indicated times. The HSP70 inhibitor MKT-077 (Sigma-Aldrich) was used at a final concentration of 10uM for 48 h.

To quantify apoptotic cell death, cells were collected by trypsinization and stained using the Annexin V PE-7AAD apoptosis detection kit (BD Pharmingen). Fluorescence was detected by flow cytometry CytoFlex LX Flow Cytometer (Beckman Coulter). 10,000 total events were collected and subsequently analyzed for the percentage of Annexin V-PE positive cells.

Quantitative RT-PCR (RT-PCR) was performed by real-time analysis using the Step One Plus detection system (Applied Biosystems) and FAST SYBR GREEN PCR Master Mix kit (Applied Biosystems). RNA was isolated and purified using the Trizol extraction method (Invitrogen). cDNA was then generated using the High-Capacity cDNA Reverse Transcription Kit (Applied Biosystems).

Cells were harvested and lysed in E1A whole cell lysate buffer supplemented with protease inhibitor cocktail (Sigma-Aldrich). Immunoblotting concentration in lysates was determined using the bicinchoninic acid (BCA) assay and analyzed by SDS-PAGE using antibodies against BAG1 (generously provided by Dr. Graham Packham), HSP70 (Abcam, #ab2787), PARP (Cell Signaling, #9532), CASPASE-3 (Cell Signaling, #9662), RAF1 (R&D Systems, #MAB3585), XIAP (Cell Signaling, #2042), GCR (Cell Signaling, #3660), SLC7A6 (ThermoFisher, #PA5–30575), POLR1D (ThermoFisher, #PA5–30575), ACTIN (Santa Cruz, #sc-47,778), GAPDH (Abcam, #ab9485).

For protein-protein interaction studies, ∼750 μg of whole cell lysates (WCL) was used for immunoprecipitation (IP). BAG1 IPs were performed by incubating lysates with 10 μL Anti-BAG1 antibody (Abcam, #ab32109) or Rabbit IgG control (Abcam, #ab172730) for 16 h at 4 °C and precipitates captured using protein A/G beads (Santa Cruz).

Experiment was performed in triplicate and all samples collectively subjected to proteomics analysis. Cells stably expressing vector control, BAG1S, or BAG1ΔS were infected with shRNA directed against BAG1 or luciferase (Vector-shLUC, Vector-shBAG1. BAG1S-shBAG1, BAG1ΔS-shBAG1). Five days post-infection, conditions were treated 100 nM 4-OHT for 12 h. Cellular extracts were standardized for protein quantity and run 0.5 cm into a 10% TRIS-Glycine gel (Invitrogen Novex). The sample lanes and a gel control were cut out and subsequently digested with trypsin. The digests were analyzed by LC-MS/MS on a Q Exactive HF mass spectrometer using an extended LC method. MS/MS spectra generated from the LC-MS/MS runs were searched using full tryptic specificity against the UniProt human database (www.​uniprot.​org; 10/01/2017) using the MaxQuant 1.6.1.0 program. “Match between runs” feature was used to help transfer identifications across experiments to minimize missing values. Protein quantification was performed using unique+razor peptides. Razor peptides are shared (non-unique) peptides assigned to the protein group with the most other peptides (Occam’s razor principle). False discovery rates for protein, and peptide identifications were set at 1%.

The output from MaxQuant identified 6225 total proteins. Common contaminants (e.g. keratins) and proteins identified by a single peptide (low-confidence identifications) were removed yielding 4915 remaining proteins. The abundance of a protein in a sample can be determined from the intensity i.e. the sum of the peptide MS intensities for the protein. To account for larger proteins generating more peptides, the intensity values were adjusted by dividing with the number of theoretical peptides for each protein (iBAQ intensity). In addition, the intensity values are also normalized (LFQ Intensity) to take into account the potential differences in sample loading. Normalization is done using the MaxLFQ algorithm [19]. The LFQ intensities were log2 transformed and then used for quantitative comparison. Student’s t-test (p-value) and fold change were calculated for binary comparison.

Conditions Vector-shBAG1, BAG1S-shBAG1, BAG1ΔS-shBAG1 were all divided by Vector-shLUC to normalize fold-change assessments to endogenous BAG1 levels. These standardized fold-change values were annotated knockdown (KD), BAG1S (S), and BAG1ΔS (ΔS) respectively. Proteins were sorted and hits with a p-value≤0.05 across all conditions were retained (309 proteins). To increase the stringency, proteins that did not have a ≥ |1.5| fold-change in the KD condition were discarded (153 proteins). Proteins were classified based on biological process characterized by UniProt. To visualize changes a heatmap was constructed using software described in Nucleic Acids Research (Babicki et al., Nucleic Acids Re, 2016).

The fold-changes of proteins in KD were compared with S and ΔS values to determine if reintroduction of either plasmid generated a partial rescue. Partial rescue was defined by an increase of ≥10% compared to KD.

Data collected from at least three independent experiments are presented as mean ± standard deviation. Statistical testing was performed using SPSS with differences between two groups determined by a Student’s t-test. Significance is denoted in the figures as: ***P < 0.005; **P < 0.02; *P < 0.05.

To study MYC influence in malignant transformation, a system employing a conditional MYC-ER allele has been used extensively to mimic oncogenic levels of MYC function [20, 21]. Treatment with the estrogen analog 4-hydroxytamoxifen (4-OHT) activates this MYC-ER fusion protein and allows for induction of MYC activity (Additional file 1: Figure S1). Depending on the cellular context, MYC-ER cells display a robust MYC-dependent surge in either proliferation or apoptosis [22]. In human osteosarcoma U2OS MYC-ER cells where BAG1 induction was blocked, activation of MYC resulted in substantial apoptotic cell death [5]. As we reported previously, depletion of BAG1 followed by MYC activation via treatment with 4-OHT, resulted in a significant increase in apoptosis, as measured by either Annexin V staining or immunblotting for PARP cleavage (Fig. 1a, b and c). To verify that loss of BAG1 is the direct cause of the cell death observed with this shRNA construct, U2OS MYC-ER cells were generated to ectopically express the three major isoforms of the BAG1 protein. This exogenous BAG1 was resistant to knockdown by shRNA directed against the 3’UTR, yielding a system where all isoforms of endogenous BAG1 can be eliminated and the cells then rescued with individual isoforms (Fig. 1d). The combined ectopic expression of all three BAG1 isoforms significantly rescued the cell death marked by loss of endogenous BAG1 as observed by flow cytometry and apoptotic proteins to assess apoptosis (Fig. 1e, f and g). This established a platform for further dissecting the causal link between BAG1 and the suppression of MYC-driven apoptosis.

BAG1 protein exists as multiple isoforms which originate from a single transcript of the human BAG1 locus on chromosome 9 [9, 10]. A selection of different translation start sites generate the major BAG1 isoforms: BAG1L, BAG1M, and BAG1S [23]. The isoforms share a common carboxyl terminus, which includes ubiquitin-like and BAG domains [24]. However, the isoforms differ in the length of their amino termini. BAG1L and BAG1M contain 10 hexapeptide motif (TRSEEX) repeats, whereas BAG1S possess only four repeats. In addition, the extended amino terminus of BAG1L holds a nuclear localization signals (NLS) supporting its predominate localization to the nucleus [11]. Conversely, BAG1M and BAG1S are mainly detected in the cytosol [5, 9, 11, 23‐25, 7‐10] (Fig. 2a). Moreover, the different isoforms of BAG1 are linked to different effector functions. To assess the influence that different BAG1 isoforms have on MYC-dependent survival, U2OS MYC-ER cells were generated to exogenously express the individual isoforms. Depletion of endogenous BAG1 in each cell line demonstrated discrete rescue of either BAG1L, M, or S isoform expression when compared to cells transfected with vector control (Fig. 2b). As expected, apoptosis assays (both Annexin V staining and immunoblotting for and PARP cleavage) using the control cells demonstrated a substantial increase in cell death under conditions of MYC activation and endogenous BAG1 depletion. Neither of the larger isoforms, BAG1L or BAG1M, were sufficient to rescue survival in cells with oncogenic levels of MYC. However, under these conditions reintroduction of BAG1S significantly reduced the level of apoptosis (Fig. 2c, d and e). The selective ability of BAG1S in MYC-induced cells to repress death strongly suggested that the small isoform of BAG1 plays the key role in pro-survival function of BAG1 in this context.

Molecular chaperones of the 70kDA HSP70 family are involved in a multitude of cellular processes, including protein folding, protein translocation, and protein degradation [26‐28]. To mediate these diverse functions, HSP70 cooperates with a variety of chaperone cofactors [29]. Specific cofactors either modulate the ATPase and peptide binding cycle of the chaperone or help direct HSP70 to specific proteins and subcellular compartments [30‐32]. BAG1 is a component of the chaperone system that binds to the ATPase domain of HSP70 [33]. By making use of the ATP-binding domain rather than the substrate-binding domain, various other proteins can attach to the chaperone complex [15, 32, 34‐36]. Previous studies indicated that HSP70 interacts with each of the BAG1 isoforms via the evolutionarily conserved BAG domain located in the carboxyl terminus and that these interactions enable different biochemical and biological activities of target proteins [37, 38].

The HSP70 chaperone functions in part to eliminate mis-folded or mis-targeted proteins [39, 40]. Removal of corrupted proteins is crucial for cell viability because these proteins accumulate as non-specific aggregates which can become toxic to the cell [41, 42]. To determine if the anti-apoptotic function of BAG1S required the HSP70 chaperone complex, a mutant BAG1S was generated that prevented binding to HSP70 via a previously reported single amino acid change from a cysteine (Cys) to alanine (Ala) at residue 204 (C204A) (Fig. 3a) [43]. Stable U2OS MYC-ER cells expressing shRNA-resistant BAG1S mutant (BAG1ΔS), BAG1S, and vector control were developed and expression verified by depleting the cells of endogenous BAG1 (Fig. 3b). Immunoprecipitation (IP) of BAG1 confirmed that BAG1S, but not BAG1ΔS, was able to bind HSP70 (Additional file 2: Figure S2).

Apoptosis assays (both Annexin V staining and immunoblotting for PARP cleavage) demonstrated that the MYC-dependent cell death observed with loss of BAG1 is rescued by reintroduction of BAG1S but not BAG1ΔS (Fig. 3d, e and f). These findings suggest that BAG1S provides its pro-survival function by serving as a cofactor for the HSP70 chaperone complex.

Furthermore, obstructing chaperone function with the small molecule HSP70 inhibitor MKT-077 in combination with MYC activation resulted in robust cell death (Fig. 3g). Collectively, these data suggest that suppressing HSP70/BAG1 chaperone function by repressing expression/activity of either member separately or disrupting the interaction, rescues MYC-driven apoptosis. Stated more directly, MYC activation triggers the cell to become dependent on the cooperation and function of the HSP70/BAG1S chaperone complex for survival.

BAG1 can modulate the ATP-driven activity of HSP70 and this regulation of chaperone function may provide a common mechanism to explain the diverse effects of BAG1 [44]. Although the protein refolding activity of HSP70 has received the greatest attention, chaperones can participate in a multitude of processes including protein translocation, stabilization, and degradation [26, 34]. BAG1 isoforms contain a ubiquitin-like domain (Fig. 2a) and can directly bind components of the ubiquitylation machinery and regulate client protein stability [12, 45, 46]. Glucocorticoid Receptor (GCR), X-linked Inhibitor of Apoptosis Protein (XIAP), and Rapidly Accelerating Fibrosarcoma 1 (RAF1) have been previously described as clients of HSP70 chaperone activity that function in part to activate pro-survival pathways [13, 43, 47]. To assess the dependence of MYC and BAG1S on these established HSP70 client proteins, we activated MYC for 12 or 24 h (as indicated) in context of endogenous BAG1, loss of BAG1 or singularly expressed BAG1S, both unmodified and mutated to prevent interaction with HSP70 (BAG1ΔS) (Fig. 4a). With 12 h of oncogenic MYC activity, chaperone targets were not substantially affected. However, after 24 h of MYC induction, client proteins were stabilized in conditions where the HSP70/BAG1S chaperone complex remained intact (Fig. 4a). Importantly, the stabilization of pro-survival proteins GCR, XIAP, and RAF1 by sustained HSP70/BAG1S expression provides a potential mechanisms by which BAG1S-dependent survival can be controlled.

In an effort to evaluate proteins that could contribute to BAG1S-driven pro-survival with early oncogenic MYC activation, an unbiased, comprehensive proteomics screen was performed (Fig. 4c). Three experimental replicates were subject to LC-MS/MS analysis to identify consistent and significant protein alterations. A visual representation of the proteomic analysis scheme is provided in Fig. 4b, and the procedure is further described in the Methods. In brief, levels of individual proteins from conditions with loss of BAG1 (KD), ectopic BAG1S (S), and ectopic BAG1ΔS (ΔS) were compared to levels in cells expressing endogenous BAG1 (shLUC control). This allowed normalization of protein fold-changes. Proteins that had a significant p-value (≤0.05) across all conditions and a change in expression of ≥|1.5| in knockdown (KD) were further assessed. The output proteins were classified based on biological process characterized by UniProt and differential expression of proteins in each condition was visualized via heatmap (Fig. 4d). Of the 153 proteins that whose levels were controlled by BAG1, 56 were at least partially rescued by reintroduction of BAG1S, and 30 of those were specific to BAG1S and not HSP70-interaction deficient mutant BAG1ΔS (Fig. 4e). Several of the proteins rescued by BAG1S and not BAG1ΔS are presented in Table 1. Empirically assessing protein expression for two of the BAG1S-specific rescued proteins, Solute Carrier Family 7 Member 6 (SLC7A6) and RNA Polymerases I Subunit D (POLR1D), verified the results of the screen (Fig. 4g). Of the 30 proteins whose regulation by BAG1S correlates with protection from MYC-induced apoptosis, nearly half have been previously associated with survival [48‐62]. Furthermore, the majority of identified proteins were formerly linked with MYC (Table 1) [52, 58, 63‐70]. BAG1 expression is a biomarker of poor prognosis, presumably because it activates a pro-survival pathway that tumor cells need in order to proliferate. By characterizing additional members of this pro-survival process we are closer to identifying methods to restore MYC-driven apoptosis. Importantly, several of the proteins identified are candidates for therapeutic targeting based on existing drugs [71‐74]. Collectively, this adds to the existing knowledge of BAG1-mediated activity, which is summed and graphically represented in Fig. 5.