DR5 disulfide bonding functions as a sensor and effector of protein folding stress

Abstract

New agents are needed that selectively kill cancer cells without harming normal tissues. The TRAIL ligand and its receptors, DR5 and DR4, exhibit cancer-selective toxicity. TRAIL analogs or agonistic antibodies targeting these receptors are available but have not yet received FDA approval for cancer therapy. Small molecules for activating DR5 or DR4 independently of protein ligands may activate TRAIL receptors as a monotherapy or potentiate the efficacy of TRAIL analogs and agonistic antibodies. Previously described Disulfide bond Disrupting Agents (DDAs) activate DR5 by altering its disulfide bonding through inhibition of the Protein Disulfide Isomerases (PDIs) ERp44, AGR2, and PDIA1. Work presented here extends these findings by showing that disruption of single DR5 disulfide bonds causes high-level DR5 expression, disulfide-mediated clustering, and activation of Caspase 8-Caspase 3 mediated pro-apoptotic signaling. Recognition of the extracellular domain of DR5 by various antibodies is strongly influenced by the pattern of DR5 disulfide bonding, which has important implications for the use of agonistic DR5 antibodies for cancer therapy and as research tools. Importantly, other ER stressors, including Thapsigargin and Tunicamycin also alter DR5 disulfide bonding in various cancer cell lines and in some instances, DR5 mis-disulfide bonding is potentiated by overriding the Integrated Stress Response (ISR) with inhibitors of the PERK kinase or the ISR inhibitor ISRIB. These observations indicate that the pattern of DR5 disulfide bonding functions as a sensor of ER stress and serves as an effector of proteotoxic stress by driving extrinsic apoptosis independently of extracellular ligands.

Introduction

Cancer remains one of the most lethal diseases, making the identification of safer and more effective therapies urgent. Identification of cancer drug targets that are required by malignant cells, but not normal cells, is key. Targeting proteins involved in the folding and maturation of oncoproteins, but not “house-keeping” proteins, holds great promise. Protein Disulfide Isomerases (PDIs) comprise a family of 22 human enzymes that play essential roles in the folding of secreted and membrane proteins [1]. Previous work showed that PDIs may be favorable targets for anticancer agents (e.g., [2, 3]). However, much of this work focused on canonical PDIs with CXXC active site motifs and little is known about non-canonical PDIs that possess CXXS active site trapping motifs that lack the second, resolving cysteine. Previous work indicated that bicyclic thiosulfonate compounds termed Disulfide bond Disrupting Agents (DDA) bind to the PDIs PDIA1, ERp44, AGR2, and AGR3 through their active site Cys residues [4]. DDAs block client binding to PDIA1 and ERp44 and prevent disulfide-mediated AGR2 dimerization. Further, mutation of the active site Cys residues of ERp44 and AGR2 ablate binding to biotinylated DDAs. Collectively, these results suggest that DDAs inhibit the catalytic activity of PDIA1, ERp44, AGR2, and AGR3 by covalently modifying their active site Cys residues.

Importantly, DDAs show significant activity against breast tumors and metastatic lesions in animal models without affecting surrounding stromal cells or normal tissues [5, 6]. Tumor cell death occurred through apoptosis, and DDA-driven apoptosis was associated with downregulation of the HER-family oncoproteins EGFR, HER2, and HER3 and upregulation and activation of DR5, a receptor for the pro-apoptotic ligand TRAIL. However, significant questions remain regarding DDA modes of anticancer action, determinants of cancer responsiveness to DDAs, and the features controlling DDA safety and metabolic stability. The work presented here was designed to address these questions. The results reveal that DR5 plays a central role in the cancer-specific, pro-apoptotic effects of the DDAs, that DR5 levels and signaling activity through the Caspase 8-Caspase 3 axis are controlled by the state of DR5 disulfide bonding, and that multiple inducers of endoplasmic reticulum protein folding stress alter DR5 disulfide bonding. These observations suggest that DR5 functions as a sensor of proper disulfide bond formation in proteostasis and executes an apoptotic program in cancer cells in response to excessive protein misfolding in the secretory pathway.

Materials and Methods

Cell culture, preparation of cell extracts, and immunoblot analysis

MCF10A cells were cultured in a humidified incubator set at 37°C containing 5% CO₂ as described previously [7]. All other cell lines were grown in Dulbecco’s Modified Eagle’s Medium (GE Healthcare Life Sciences, Logan, UT, USA) with 10% fetal bovine serum (10% FBS–DMEM). A431, AsPC1, HaCaT, HCC1937, HepG2, HMEC, MCF10A, MDA-MB-468, SH-SY5Y, and T47D were purchased from American Type Culture Collection (ATCC) (Manassas, VA, USA). SUM149pt was purchased from Applied Biological Materials, Inc. (Richmond, BC, Canada). ERp44 and PERK knockout HepG2 cell lines were described previously [8]. WM793 cells were kindly provided by Dr. W. Douglas Cress, Moffitt Cancer Center. Generation of the MCF10A/Vector, MCF10A/EGFR and MCF10A/MYC cell lines is described in previous work [5]. Derivation of the HCI-012/LVM2/LR10 cell line is described in previous work [5, 6]. Generation of DR5 knock out MDA-MB-468 cells and MDA-MB-468 cells stably expressing DR5 using the Tet-ON system is previously described [5]. Generation of the MDA-MB-468 cells stably expressing CDCP1 or vector control is described in previous work [9]. Cell lines were used within the first 15 passages and routinely tested for mycoplasma by Hoechst staining.

Cell lysates were prepared as described in a previous publication [10]. Immunoblot analysis was performed employing the following antibodies purchased from Cell Signaling Technology (Beverly, MA, USA) [Akt, #4691; P-Akt[T308], #13038; ATF4, #11815; Caspase 8, #4790; CDCP1, #13794; Cleaved Caspase 3, #9664; Cleaved Caspase 8, #9496; Cyclophilin B, #43603; DCR2, #8049; DR3, #20772; DR4, #42533; DR5, #8074; eE2F, #2332; P-eE2F[T56], #2331; ERp57, #2881; FADD, #2782; FAS, #4272; GRP78, #3177; HER3, #4754; LC3, #3868; LRP5, #2506; MET, #3127; PARP, #9532; PCSK9, #55728; PDIA1, #3501; PERK, #5683; TNFR1, #3736; and XBP1s, #12782], from Santa Cruz Biotechnology (Santa Cruz, CA, USA) [Actin, sc-47778; AGR2/3, sc-376653; AGR3, sc-390940; c-Myc (9E10), sc-40; DR5, sc-166624; EGFR, sc-373746; ERAP1, sc-271823; ERp44, sc-393687; PDIA6, sc-365260; and pY99, sc-7020], and from Rockland Immunochemicals, Inc. (Limerick, PA, USA), Streptavidin-Alkaline Phosphatase conjugated (SA-AP), S000–05.

Quantitative Analysis of Immunoblot Results

Protein levels in immunoblots were quantified using Adobe Photoshop 2023 (Berkeley, CA, USA) and ImageJ 1.54k 15 September 2024 (NIH, Bethesda, MD), as previously described [11], followed by normalization to Actin as a loading control.

Materials

Reagents were purchased from the following companies: Tunicamycin and Chloroquine: Sigma-Aldrich (St. Louis, MO); Thapsigargin: AdipoGen (San Diego, CA, USA); Lapatinib: Selleck Chemicals (Houston, TX); Doxycycline: Enzo Life Science (Farmingdale, NY, USA); TORIN1, and dithiothreitol (DTT): TOCRIS Bioscience (Minneapolis, MN); Cyclosporine A (CsA): Biorbyt (Duran, NC, USA); Rapamycin and PERK Inhibitor I (GSK2656157): Calbiochem (Burlington, MA, USA); Bafilomycin A1, Gefitinib, ISRIB, MK2206, and Q-VD-OPH: Cayman Chemical (Ann Arbor, MI, USA); TNF, (PHC3016); N-ethylmaleimide (NEM): Thermo Fisher Scientific (Grand Island, NY, USA); eEF2K inhibitor (A-484954), Ceapin-A7 (HY-108434), Nelfinavir (HY-15287), KIRA-6 (HY-19708), MG132 (HY-13259), AMG PERK 44 (HY-12661A), CCF642 (HY-100430), Securinine (HY-N2079), LOC14 (HY-100432), 16F16 (HY-124866), Alisporivir (HY-12559), and VPS34 inhibitor Vps34-IN-1: MedChemExpress (Monmouth Junction, NY, USA); FK506: InvivoGen (San Diego, CA, USA); b-AP15: MedKoo Biosciences (Chapel Hill, NC, USA).

Tumor studies and histochemical analysis

012/LVM2/LR10 xenograft tumor studies were carried out in adult female NOD-SCID-γ (NSG) mice, as described in a previous publication [6]. After the development of tumors (approximately 4 mm³), mice were randomly assigned to two treatment groups: DMSO (Vehicle) and 10 mg/kg dMtcyDTDO). Mice were treated every weekday for twenty days by intraperitoneal injection, administering 50 μL per injection. At the end of the twenty-day period, tissue samples were collected and fixed in 4% paraformaldehyde/Phosphate-Buffered Saline (PBS), followed by paraffin-embedding, sectioning and staining with hematoxylin and eosin (H&E) by the University of Florida Molecular Pathology Core (https://molecular.pathology.ufl.edu/).

Prior to endpoint, peripheral blood was collected by facial vein puncture into EDTA-treated tubes; complete blood cell counts (CBCs) were obtained using an Element HT5 fully automated hematology analyzer (Heska, Loveland, CO, USA).

Disulfide Bond-mediated Oligomerization

Disulfide bond-mediated oligomerization under non-reducing conditions was analyzed as described in previous work [4].

Vector Construction

In order to construct the Tet-DR4 expression vector, DR4 (Addgene plasmid #61382) was amplified using the following primers: 5′-TTTTATCGATCACCATGGCGCCCGTCGCCGTCTGG-3′ and 5′-TTTTGGATCCTCACTCCAAGGACACGGCAG-3′ and cloned into the pRetroX-TetOne-Puro vector with a modified cloning site that incorporates Not I, Bcl I, and Cla I sites 5′ to the BamH I site (Clontech, Mountain View, CA, USA).

The initial mutations of C81S, C119S and C160S in DR5 were performed in pcDNA3 with QuikChange mutagenesis and the following primers, respectively: 5′-CCAGCCCCTCAGAGGGATTGAGTCCACCTGGACACCATATC-3- and 5′- GATATGGTGTCCAGGTGGACTCAATCCCTCTGAGGGGCTGG-3′, 5′- GCTTGCGCTGCACCAGGAGTGATTCAGGTGAAGTGG-3′ and 5′- CCACTTCACCTGAATCACTCCTGGTGCAGCGCAAGC −3′ and 5′- CGGAAGTGCCGCACAGGGAGTCCCAGAGGGATGGTCAAGG −3′ and 5′- CCTTGACCATCCCTCTGGGACTCCCTGTGCGGCACTTCCG −3′. Mutations were verified by sequencing. The following primers were used to add a 5′-EcoRI and a 3′-BamHI site to C81S, C119S and C160S DR5 by PCR amplification: 5′- TTTTGAATTCCACCATGGAACAACGGGGACAGAAC-3′ and 5′-TTTTGGATCCTTAATGATGATGATGATGATGGGACATGGCAGAGTCTGC-3′. The C119S and C160S DR5 mutants were subsequently cloned into the EcoRI and BamHI sites of pRetroX-TetOne-Puro vector (Clontech, Mountain View, CA, USA). Mutation of the DR5 C94 was produced in the DR5 C81S construct to produce the DR5 C81S/C94S using QuikChange mutagenesis and the following primers: 5′-CATATCTCAGAAGACGGTAGAGATAGCATCTCCTGCAAATATGGACAGG-3′ and 5′- CCTGTCCATATTTGCAGGAGATGCTATCTCTACCGTCTTCTGAGATATG-3′. Mutation of the DR5 C137S was introduced into the DR5 C119S construct to produce the DR5 C119S/C137S using QuikChange mutagenesis and the following primers: 5′-CCACGACCAGAAACACAGTGAGTCAGTGCGAAGAAGGCACCTTC −3′ and 5′-GAAGGTGCCTTCTTCGCACTGACTCACTGTGTTTCTGGTCGTGG −3′. The C178S mutation of DR5 was introduced into the DR5 C160S construct to produce the DR5 C160S/C178S using QuikChange mutagenesis and the following primers: 5- CACCCTGGAGTGACATCGAAAGTGTCCACAAAGAATCAGGTAC −3′ and 5′-GTACCTGATTCTTTGTGGACACTTTCGATGTCACTCCAGGGTG-3′. The C153S and C156S mutations of DR5 were produced using QuikChange mutagenesis and the following primers, respectively: 5′-GAAGAAGATTCTCCTGAGATGAGCCGGAAGTGCCGCACAGGG-3′ and 5′- CCCTGTGCGGCACTTCCGGCTCATCTCAGGAGAATCTTCTTC-3′ and 5′-CTCCTGAGATGTGCCGGAAGAGCCGCACAGGGTGTCCCAGAGGG-3′ and 5′-CCCTCTGGGACACCCTGTGCGGCTCTTCCGGCACATCTCAGGAG-3′. The K245R DR5 mutation was produced by QuikChange Mutagenesis and the following primers: 5′-GTCCTTCCTTACCTGCGAGGCATCTGCTCAGGT-3′ and 5′-ACCTGAGCAGATGCCTCGCAGGTAAGGAAGGAC-3′. Tet-DR5 [ΔC41] was produced by amplifying DR5 by PCR, adding 5′-EcoRI and 3′-BamHI sites using the following primers: 5′-TTTTGAATTCCACCATGGAACAACGGGGACAGAAC-3′ and 5′-TTTTGGATCCTTACTTGTCGTCATCGTCTTTGTAGTCGACAGAGGCATCTCGCCCGG-3′ followed by cloning into the EcoRI and BamHI sites of the pRetroX-TetOne-Puro vector. The Tet-DR4[ΔC43] was produced by amplifying DR4 by PCR, adding 5′-ClaI and 3′-BamHI sites using the following primers: 5′-TTTTATCGATCACCATGGCGCCCGTCGCCGTCTGG-3′ and 5′-TTTTGGATCCTTACTTGTCGTCATCGTCTTTGTAGTCGATCGAGGCGTTCCGTCCAGTTTTG-3′ followed by cloning into the ClaI and BamHI sites of the pRetroX-TetOne-Puro vector.

In order to clone ERp44, total RNA from T47D cells was extracted with TRIzol Reagent (Invitrogen, Waltham, MA USA) according to the manufacturer’s protocol. Total cellular RNA was reverse transcribed to synthesize first-strand cDNA using the PCR conditions listed: 25 °C for 10 min, 42 °C for 30 min, and 95 °C for 5 min. DNA encoding ERp44 was subsequently amplified using the following primers: 5′- TTTTGGATCCCACCATGCATCCTGCCGTCTTCC-3′ and 5′-TTTTCTCGAGTTAAAGCTCATCTCGATCCCTC-3′. The PCR fragment encoding ERp44 was cloned into the 5′ BamHI and 3′ XhoI sites of the pMXs-IRES-Blasticidin retroviral vector (RTV-016) (Cell Biolabs, Inc., San Diego, CA USA). The following primers were used to produce the C29S mutation of ERp44 using QuikChange mutagenesis: 5′- CTTTAGTAAATTTTTATGCTGACTGGAGTCGTTTCAGTCAGATGTTG-3′ and 5′-CAACATCTGACTGAAACGACTCCAGTCAGCATAAAAATTTACTAAAG-3.′ All mutations were verified by sequencing.

MTT Cell Viability Assays

In order to evaluate cell viability, cells were plated at 7,500 cells/well in 96-well plates and incubated at 37°C for 24h. Cells were subsequently treated with various compounds for 72 h at 37°C. Following removal of the cell media, cells were incubated with 0.5 mg/ml MTT (3-(4,5-dimethylthiazol-2-yl)-2,5-diphenyltetrazolium bromide) (Biomatik, Wilmington, DE, USA) in PBS at 37°C for 1 h. The MTT solution was subsequently removed and the MTT formazan product was dissolved in 100 μl of DMSO, followed by measurement of MTT formazan absorbance (570 and 690 nm) in a plate reader.

Protein Synthesis Assays

Leucine incorporation into proteins was assayed using ³H-Leucine (cat. # NET460001MC) obtained from Perkin Elmer (Waltham, MA, USA), as described in a previous publication [12].

Chemical Synthesis of DDAs

The DDAs presented in Fig. 1A. were prepared based on existing literature procedures from our team and others. RBF3, D5DO, and D7DO were obtained according to the methods described by Field and colleagues [13, 14]. DTDO was synthesized as we described previously [15], as well as tcyDTDO [6], dMtcyDTDO and dFtcyDTDO [16], and Bio-Pyr-DTDO [4].

Fig. 1:

DDA compounds that selectively inhibit AGR2, PDIA1, and ERp44 block the maturation of select transmembrane and secreted proteins, but strongly upregulate DR5. A. Chemical structures of representative DDAs used herein. B. Demonstration of the selectivity of biotinylated DDA binding to the target proteins PDIA1, ERp44, and AGR2. Extracts from T47D cells were incubated with the indicated competitors for 2h and then incubated for 1 h after the addition of biotinylated-DDA, followed by sample analysis by gel electrophoresis and blotting with Streptavidin-Alkaline Phosphatase detection. C. Non-reducing immunoblot analysis of the effect of 24 h treatment of the indicated cells with the specified DDAs at 2.5 μM each. M represents monomeric DR5 isoforms and O represents disulfide-bonded DR5 oligomeric complexes. S and L refer to the short and long forms of DR5 and S′ and L′ refer to the same DR5 isoforms with altered electrophoretic mobility caused by DDA treatment. Actin serves as a loading control. D. Left panel, reducing immunoblot analysis of MDA-MB-468 cells treated with increasing dFtcyDTDO concentrations showing higher expression of XBP1s and decreased levels of the mature forms and increased relative levels of the pro- forms of MET and PCSK9. Right panel, non-reducing immunoblot analysis using the indicated antibodies. Red arrows represent oligomeric ERp44 isoforms lost upon dFtcyDTDO treatment and the green arrow represents high molecular mass ERp44 isoforms elevated by dFtcyDTDO treatment. E. MDA-MB-468 cells treated for 24 h as indicated and analyzed by immunoblot under non-reducing conditions. F. MDA-MB-468 cells treated for 24 h with vehicle or 2.5 μM dMtcyDTDO, with or without 10 ng/ml TRAIL or 10 ng/ml TNF as indicated, and analyzed by immunoblot under non-reducing conditions. O and M represent the Oligomeric and Monomeric protein isoforms in panels C-F.

Metabolic stability using rat and human liver and intestinal microsomes

To understand the rate of metabolism of compounds across species the in vitro metabolic stability of each compound was performed using liver and intestinal microsomes from rats and humans in triplicate. Verapamil was used as a positive control to check the activity of the microsomes. The incubation mixtures consisted of liver or intestinal microsomes (1 mg/ml protein for liver microsomes and 0.5 mg/ml protein for intestinal microsomes), substrate (10 μM), and NADPH (1 mM) in a total volume of 0.2 ml potassium phosphate buffer (50 mM, pH 7.4). Reactions were initiated with the addition of NADPH and kept in an incubator shaker at 37°C. Aliquots of 20 μl were collected at 0, 5, 10, 15, 30, 45, and 60 min and mixed with 100 μl of acetonitrile with formic acid (0.1% v/v) containing phenacetin (50 ng/ml; internal standard) for the termination of the reaction. The samples were then vortex mixed and filtered through a 0.45 μm PTFE Solvinert membrane filtration plate under centrifugation at a speed of 2000 ×g for 5 min at 4°C. The filtrates were subjected to UPLC-MS/MS analysis.

The intrinsic clearance of the compounds was calculated using a half-life employing the ‘substrate depletion’ approach. The apparent half-life was calculated from the pseudo-first-order rate constants obtained by linear regression of log (concentration) and time plots. The in vitro intrinsic clearance for compound was estimated using the formula:

$$t_{1/2}=\mathit{\text{ln}}(2)/k$$

$${Cl}_{int}(\mu L/\mathit{\text{min}}/mg\mathit{\text{protein}})=\frac{0.693}{t_{1/2}}\text{X}\frac{V_{\text{incubation}\text{in}\mu L}}{\text{Protein}\text{Concentration}\text{in}\text{mg}}$$

Where k is the slope of the line obtained by plotting the natural logarithmic of the percentage of parent remaining versus time and V is the volume of incubation.

The in vitro intrinsic clearance from rat and human liver microsomes were scaled to whole-organ (hepatic) in vivo intrinsic CL (CL_{int, H}) using the scaling factors available in the literature using equation [17]:

$${Cl}_{\text{int},\text{H}}\left(\text{mL/min*kg}\right)=\frac{0.693}{t\text{1/2}\left(\text{min}\right)}\text{X}\frac{\text{Volume}\text{of}\text{incubation}\left(\text{mL}\right)}{\text{Protein}\text{in}\text{incubation}\left(\text{mg}\right)}\text{X}\frac{\text{Liver}\text{weight}\left(g\right)}{\text{Body}\text{weight}\left(\text{kg}\right)}\text{X}\text{SF}$$

The scaling factor used for the rat was 45 (45 mg microsomal protein/g liver) and liver weight (g) per kg body weight was 40 g/kg while for human scaling factors were 29 (29 mg microsomal protein/g liver) and liver weight (g) per kg body weight was 24 g/kg [18, 19].

LC-MS/MS analysis:

UPLC-MS/MS analysis was carried out using a Waters Acquity Class I Plus UPLC coupled with a Waters Xevo TQ-S Micro triple quadrupole mass spectrometer. The chromatographic separation was achieved using Acquity UPLC CSH C18 column (2.1 mm × 50 mm, 1.7 μm) using the mobile phase consisting of 0.1% formic acid (A) – methanol (B) with a gradient program of 80 % A held for 0.5 min, then decreased A to 65% reaching 1.0 min and further decreased to 40 % A by 2.5 min and held at 40 % until 3.0 min, then sharply decreased back to the initial conditions by 3.1 min and maintained until 3.5 min. The column and autosampler temperatures were kept at 50 °C and 4 °C, respectively. The mobile phase was delivered at a flow rate of 0.35 mL/min and the injection volume was set to 2 μL. The MassLnyx software version 4.2 was used for instrument control and TargetLynx for data analysis. The mass spectrometer was operated in positive ion mode and detection of the ions was performed in the multiple reaction monitoring (MRM) mode. The monitored ion transitions (m/z) and instrument conditions can be seen in Supplementary Table 1. Each compound was monitored using two precursor-to-daughter ion transition pairs, one as a quantifier and another as a qualifier to get better selectivity for each compound. The ion spray voltage was set at 3000 V, the desolvation temperature was 400 °C, the desolvation gas flow was 850 L/h, and the cone gas flow was 50 L/h.

Flow Cytometry Analysis

Cells were lifted from plates using cell scrapers and washed in ice cold PBS. Single cell suspensions were prepared, counted, and diluted to 1 × 10⁶ cells/100 μL. Subsequently, cells were stained for DR4 (DJR1-APC, Cat: 307208, Biolegend, San Diego, CA) and DR5 (DJR2–4-PE, Cat:307406, Biolegend, San Diego, CA) markers for 30 min at 4 °C. Cells were then washed twice in ice-cold PBS and stained with viability dye (violet fluorescent reactive dye, Cat: L34955, Invitrogen, Waltham, MA, USA). FACS Buffer (1% FBS, 0.5 mM EDTA in PBS (400 μL)) was subsequently added. Cells were not fixed or permeabilized. Stained samples were analyzed using single-color compensation and FMO controls on a Sony SP6800 spectral analyzer and quantified using FlowJo V10.8.1 (BD Biosciences, San Jose, CA). Cells were gated in the following sequence: SSC-A × FSC-A, FSC-H × FSC-A, SSC-H × SSC-A, and Live Cells, to determine Mean Fluorescence Intensity (MFI) of DR5 or DR4.

Confocal Immunofluorescence Microscopy

Cells were treated for 24 h as indicated, fixed with 4% paraformaldehyde and stained without permeabilization using the DR5 monoclonal sc-166624 primary antibody and the Alexa 488-conjugated anti-mouse secondary antibody (ThermoFisher Scientific (Grand Island, NY, USA) and mounted using DAPI-containing medium. Image acquisition was performed with a Leica Application Suite X software on a Leica Stellaris5 confocal system using HC PL APO CS2 40X oil-immersion objective (numerical aperture: 1.3). Comparative images were stained, imaged, and processed simultaneously under identical conditions.

DISC Assays

DISC assays were performed essentially as described [20]. Briefly, cells (7.5 × 10⁶/sample) were treated for 16 h with vehicle or 2.5 μM dMtcyDTDO, incubated for 1 h at 37°C with 10 μg/sample Biotinylated TRAIL (Biotinylated-TRAIL (U1110): UBPBio (Dallas, TX), and complexes isolated using Streptavidin-Agarose beads (20353, ThermoFisher. Beads were washed and eluted under non-reducing conditions, however N-ethylmaleimide was not employed to preserve disulfide bonding status as it was observed to inactivate the Streptavidin-Agarose resin. The complexes were analyzed by immunoblot.

Statistical Analysis

Protein detection by immunoblot, MTT viability assays, and protein synthesis assays were performed as described in a previous publication [12]. Results of at least quadruplicate technical replicates and three biological replicates are presented. Results are presented as the average ± standard deviation. Statistical significance was considered p ≤ 0.05 using Student’s unpaired t-test.

Data Availability

The data generated in this study are available within the article and its supplementary data files.

Results

DDA-triggered selective ER retention (sERr) is associated with elevated DR5 levels and signaling:

The DDAs used herein are presented in Fig. 1A. We proposed that DDAs exhibit rapidly reversible covalent bonding to protein thiols by disulfide bond formation, except for the target PDIs that form stable disulfide bonds with DDAs [16]. In further support of this premise, we incubated T47D cell extracts with the biotinylated DDA probe Bio-Pyr-DTDO alone or combined with a 100-fold excess of the unlabeled DDA competitors shown in Fig. 1B. Bands recognized by Bio-Pyr-DTDO were identified as PDIA1, ERp44, and AGR2 by mass spectrometry and immunoblot as reported previously [4]. Endogenous biotinylated proteins are observed in the absence of Bio-Pyr-DTDO treatment (asterisks). As expected, Bio-Pyr-DTDO binding was blocked by the more reactive, less selective DDAs DTDO, D5DO, D7DO, and RBF3. In contrast, the less reactive, more selective DDA tcyDTDO did not affect Bio-Pyr-DTDO binding, nor did the thiol-reactive deubiquitinase inhibitor b-AP15 [21]. The thiol-reactive compound N-ethylmaleimide prevented Bio-Pyr-DTDO binding to DDA targets. These observations support the selectivity of bicyclic DDAs against a subset of PDIs.

Also consistent with previous work, the bicyclic DDAs dFtcyDTDO and dMtcyDTDO increased the levels of DR5, and immunoblot analysis under non-reducing conditions showed an electrophoretic mobility shift of monomeric DR5 and an increase in disulfide bonded oligomeric forms of DR5 (Fig. 1C). A previous study from the Tirosh laboratory showed that under Endoplasmic Reticulum (ER) stress conditions, trafficking of some mis-disulfide bonded transmembrane receptor tyrosine kinases became arrested in the ER through the formation of large disulfide bonded complexes involving ERp44 [8]. This mechanism was termed selective ER retention, or sERr. Since DDAs induce ER stress and inhibit ERp44 client binding [4], we examined whether DDA treatment activates or inhibits sERr. Increasing dFtcyDTDO blocked maturation of MET, an established marker of sERr and prevented maturation of PCSK9 through its auto-cleavage (Fig. 1D). This was associated with increased expression of the ER stress marker XBP1s. Analysis of the same samples under non-reducing conditions showed that increasing dFtcyDTDO concentrations increased EGFR oligomerization, and elevated levels of monomeric and oligomeric DR5. dFtcyDTDO had a modest effect on PDIA1 client binding. In contrast, dFtcyDTDO blocked the formation of lower molecular mass ERp44 disulfide bonded complexes with clients (red arrows), while very high mass ERp44 complexes (green arrow) were elevated. This observation suggests that DDAs caused sERr. However, the observation that high levels of mis-disulfide bonded monomeric DR5 accumulated suggests that DR5 may evade sERr. Comparison of DDA biochemical effects with those of reported PDI inhibitors, including CCF-642 [22], the natural product/therapeutic Securinine [23], LOC14 [24], and 16F16 [25] showed that several of these compounds downregulated the essential Wnt co-receptor LRP5, EGFR, and increased DR5 levels (Fig. 1E). Only DDA treatment 1) increased DR5 oligomerization at 25 μM and altered disulfide bonding of monomeric DR5 as indicated by decreased electrophoretic mobility, 2) caused strong upregulation of the ER stress marker GRP78 at both 2.5 and 25 μM, and 3) showed no increase in PDIA6 oligomerization. These observations suggest that the PDI selectivity pattern of DDAs results in unique effects on the client binding profiles of ERp44, PDIA1, and PDIA6. Notably, the ERp44/PDIA6 ratio was recently shown to control the kinetics of the sERr response [26].

DR4 and DR6 are members of a large family of Death Receptors with conserved extracellular Cysteine-rich ligand binding domains, therefore we examined DDA effects on other Death Receptors. Across several cell lines DDA treatment upregulated DR5 and TNFR1 but had little effect on levels of DR4 or FAS, and DR3 was not detected in these lines (Supplementary Figure S1). Since DDAs upregulate both TNFR1 and DR5, we examined which of these receptors might mediate ligand-induced apoptosis. As observed previously [5], the combination of DDA and TRAIL treatment markedly increased Caspase 8 cleavage (Fig. 1F), while TNF treatment had little effect on Caspase 8 cleavage. Based on these results, we focused primarily on how DDAs upregulate DR5 and influence its pro-apoptotic signaling, but also monitored TNFR1 levels since shared mechanisms may control DR5 and TNFR1 upregulation by DDAs.

The DDA dMtcyDTDO also induced sERr as indicated by near complete blockade of MET maturation (Fig. 2A). This effect was not altered by signaling inhibitors of mTORC1 (rapamycin), Akt (MK2206), EGFR (Gefitinib), or EGFR and HER2 (Lapatinib). The PERK kinase suppresses protein synthesis under ER stress conditions in part through phosphorylation of eIF2α [8], and previous work [8] showed that sERr is strongly potentiated by PERK inhibition. In MDA-MB-468 cells, Thapsigargin induced a partial block of MET processing that was strongly potentiated by the PERK inhibitor GSK2606414 (hereafter called PERKi) as expected. Since DDAs activate sERr, and sERr is potentiated by PERKi, cell viability studies were performed to examine the effect of DDA/PERKi combination treatment. While PERKi alone had little effect on cell viability, PERKi strongly potentiated dFtcyDTDO cytotoxicity in MDA-MB-468 breast cancer cells (2B, left panel) and WM793 melanoma cells (2B, right panel). sERr was initially investigated primarily in HepG2 hepatoma cells [8], so we compared the combinatorial effects of dMtcyDTDO and PERKi in HepG2 and MDA-MB-468 cells. In both lines PERKi alone had no effect on MET or PCSK9 processing (Fig. 2C). However, PERKi increased the levels of unprocessed MET in both lines and elevated levels of unprocessed PCSK9 in HepG2 cells. This is consistent with PERKi permitting continued synthesis of nascent MET and PCSK9 under ER stress conditions. Combinatorial DDA/PERKi treatment was associated with increased Caspase 8 cleavage, consistent with the enhanced toxicity to cancer cells. PERKi did not strongly potentiate DDA upregulation of DR5, but potentiated Caspase 8 cleavage/activation (Fig. 2D). DR5 knockout partially blunted Caspase 8 cleavage. The compound ISRIB [27] negates the integrated stress response (ISR) by overcoming the effects of eIF2α phosphorylation [28]. When combined with dMtcyDTDO, ISRIB and PERKi produced similar enhancements in the levels of unprocessed MET and PCSK9 (Fig. 2E). This is consistent with both agents overcoming ISR triggered by DDA treatment. Combinatorial DDA/PERKi activation of Caspase 8 in MDA-MB-468 cells was not altered by forced CDCP1 expression, which disrupts cell-cell adhesion and blocks suspension-induced apoptosis [9], (Supplementary Figure S2A). Further, PERKi actions were not mimicked by inhibition of eEF2K [29] that controls translation initiation (Supplementary Figure S2B). Together, these observations suggest that PERKi enhances DDA toxicity to cancer cells by potentiating DR5 pro-apoptotic signaling rather than upregulating DR5 expression. Previous work showed that the prolyl isomerase inhibitor Cyclosporine A (CsA) potentiates DDA toxicity against cancer cells in association with CsA induction of Cyclophilin B loss from cells via secretion [6]. CsA treatment mimicked the effects of PERK inhibition in potentiating DDA upregulation of monomeric and oligomeric DR5 (Fig. 2F). The CsA effect was mimicked by Alisporivir, a non-immunosuppressive CsA analog developed for antiviral therapy [30, 31].

Fig. 2:

PERK inhibition amplifies the pro-apoptotic effects of DDAs on cancer cell lines. A. Reducing immunoblot analysis of MDA-MB-468 cells treated for 24 h with the indicated combinations of dMtcyDTDO (2.5 μM), Rapamycin (100 nM), TORIN1 (100 nM), MK2206 (5 μM), Gefitinib (10 μM), Lapatinib (10 μM), Thapsigargin (400 nM), and PERKi (1μM). Red arrows denote pro- or mature protein isoforms. B. MTT cell viability assays of MDA-MB-468 cells (left panel) or WM793 cells (right panel) treated for 72 h as indicated. Data are plotted as the average (N = 6), with error bars representing standard deviation. C. Reducing immunoblot analysis of HepG2 or MDA-MB-468 cells treated for 24 h as indicated with 2.5 μM dMtcyDTDO or 1 μM PERKi. Red arrows denote pro- or mature protein isoforms. D. Reducing immunoblot analysis of the indicated cell lines treated as specified for 24 h with dMtcyDTDO (2.5 μM) or PERKi (1 μM). E. Reducing immunoblot analysis of MDA-MB-468 cells or SUM149pt cells treated for 24 h as indicated with dMtcyDTDO (2.5 μM), ISRIB (200 nM), or PERKi (1μM). F. MDA-MB-468 cells treated for 24 h with vehicle or 2.5 μM dFtcyDTDO, with or without PERKi, Cyclosporine A, or Alisporivir at the indicated concentrations. Culture medium was collected to analyze the secreted (sec) or cell (internal (int or unspecified)) fractions. Samples were analyzed by immunoblot under non-reducing conditions.

Multiple ER stress inducers alter DR5 disulfide bonding:

Since ERp44 is a DDA target, we examined if ER stress alters DR5 disulfide bonding in the absence of ERp44, or in HepG2 ERp44 knockout cells in which wild type or catalytically null (C29S) versions of ERp44 were reintroduced. The most notable effect was that the ER stressor Thapsigargin and PERKi had little effect alone, but irrespective of the presence or absence of ERp44, Thapsigargin + PERKi decreased DR5 electrophoretic mobility and increased its levels (Fig. 3A). This suggests that while DDAs are sufficient to alter DR5 disulfide bonding alone, other ER stressors may perturb DR5 disulfide bonding, particularly if combined with agents that override the ISR. Consistent with this, treatment of MDA-MB-468 or HepG2 cells with Thapsigargin + PERK inhibitor AMG44 [32] exhibited altered monomeric DR5 electrophoretic migration (Fig. 3B), which was not observed with the other ER stress inducers tested, Cyclosporine A and the HIV drug Nelfinavir. Further, treatment of MDA-MB-468 cells with dFtcyDTDO altered DR5 levels and disulfide bonding, while Tunicamycin increased DR5 levels without altering its mobility, and PERKi alone had no discernable effect (Fig. 3C). Tunicamycin + PERKi induced a partial shift in DR5 mobility, similar to that seen with dFtcyDTDO treatment, and this shift was associated with higher Caspase 8 cleavage and more numerous Caspase 3 cleavage products. Unlike DR5, DR4 is N-glycosylated. DR4 levels were not changed under any of the conditions, but DR4 mobility was increased by Tunicamycin, presumably due to its deglycosylation.

Fig. 3:

A variety of ER stressors alter DR5 disulfide bonding. A, upper panel. ERp44-deficient HepG2 cells into which vector, wild type or catalytically null ERp44 were reintroduced were treated as indicated for 24 h and analyzed by non-reducing immunoblot. A, lower panel. Expanded region of the DR5 immunoblot showing altered DR5 disulfide bonding in the Thapsigargin/PERKi (GSK2606414) combination treatment. B, upper panel. MDA-MB-468 or HepG2 cells were treated for 24 h as indicated and analyzed by non-reducing immunoblot unless otherwise indicated. B, lower panel. Expanded region of the DR5 immunoblot showing altered DR5 disulfide bonding in the Thapsigargin/PERKi (AMG-44) combination treatment. Asterisks denote up-shifted bands. C. Non-reducing immunoblot of MDA-MB-468 cells treated as indicated for 24 h. D. MDA-MB-468 cells were treated as indicated for 24 h and subjected to non-reducing (DR5, Cleaved Caspase 3, DR4, Actin) or reducing (MET) immunoblot analysis. E. Protein synthesis assays of cells pre-treated for 24 h as indicated before protein synthesis was measured by ³H-Leucine incorporation over a 2 h pulse. Data are plotted as the average (N = 6), with error bars representing standard deviation. F. HepG2 cells were treated as indicated for 24 h and subjected to non-reducing (DR5, Cleaved Caspase 3, DR4, Actin) or reducing (MET) immunoblot analysis. Unless otherwise specified, the following concentrations of compounds were used in A-F above: dFtcyDTDO (2.5 μM), Thapsigargin (400 nM), Tunicamycin (500 ng/ml), Cyclosporine A (10 μM), Dithiothreitol (DTT; (2.5 mM)), ISRIB (200 nM), Nelfinavir (20 μM), PERKi AMG-44 (10 μM) or PERKi GSK2606414 (10 μM). G. Control, PERK knockout or IRE1α knockout HepG2 cells were treated for 24 h with vehicle or 7.5 μM dMtcyDTDO and analyzed by non-reducing immunoblot. H. MDA-MB-468 cells were treated for 24 h as indicated with 2.5 μM dMtcyDTDO, 20 μM CEAPIN-A7, 10 μM KIRA-6, and 10 μM AMG-44 and analyzed by non-reducing immunoblot. O and M represent Oligomeric and Monomeric protein isoforms in panels A, B, D, F, G, and H.

Analysis of the effects of other ER stressors on MDA-MB-468 cells indicated that while PERKi alone had little effect on Caspase 8 cleavage, PERKi potentiated induction of Caspase 8 cleavage by Thapsigargin, Tunicamycin, Cyclosporine A, and Dithiothreitol (Fig. 3D). Increased Caspase 8 cleavage correlated with DR5 oligomerization, and in some cases, reduced mobility of monomeric DR5. Immunoblot of MET under reducing conditions showed that dFtcyDTDO and Thapsigargin induced sERr. Tunicamycin + PERKi induction of sERr is difficult to assess given the potentially offsetting effects on electrophoretic mobility of deglycosylation and lack of MET proteolytic processing. PERKi caused greater accumulation of unprocessed MET in combination with Thapsigargin than with dFtcyDTDO. Consistent with this observation, protein synthesis assays showed that PERKi increased protein synthesis in the presence of Tunicamycin, but further decreased protein synthesis in the presence of dFtcyDTDO (Fig. 3E). Analysis of the effects of ER stressors on HepG2 cells showed little effect of PERKi or dFtcyDTDO on DR5 levels, while PERKi caused DR5 mobility shifts when combined with Thapsigargin or Tunicamycin (Fig. 3F). Thapsigargin + PERKi caused sERr as assessed by MET processing. Levels of the PERK downstream effector ATF4 increased in response to Thapsigargin, Tunicamycin, and Cyclosporine A. In each case ATF4 upregulation was partially reversed by PERKi.

We next compared ER stress responses observed in PERK knockout and control HepG2 cells. In control cells Thapsigargin increased monomeric and oligomeric DR5 levels and PERKi-cotreatment caused upshifting of the long DR5 isoform and increased DR5 oligomerization (Fig. 3F). As expected, PERKi had no discernable effect on the levels or electrophoretic mobility of DR5 in the PERK knockout cells. We consistently observed a smaller band recognized by the PERK antibody in the presence of PERKi + ER stressors. This likely results from Caspase cleavage of PERK since it is decreased by Caspase inhibitor Q-VD-OPH. Due to differences in the responses of MDA-MB-468 and HepG2 cells to the various ER stressors, we examined ER stress-induced changes in DR5 electrophoretic mobility under non-reducing conditions in neuroblastoma (SH-SY5Y), cervical carcinoma (A431), and human mammary epithelial (HMEC) cells. In SH-SY5Y cells, dFtcyDTDO induced an upward DR5 shift. dFtcyDTDO + PERKi did not further slow DR5 mobility, but increased DR5 oligomerization (Supplementary Figure S2C). DR5 mobility and oligomerization were most strongly affected by PERKi combined with Thapsigargin or Tunicamycin in the A431 cells. ER stressors showed the smallest effect on monomeric DR5 levels in HMECs, which also exhibited a low level of DR5 oligomerization when ER stressors were combined with PERKi. Similar analyses in WM793 melanoma cells showed that dFtcyDTDO alone induced DR5 shifts under non-reducing conditions that were not further potentiated by PERKi or ISRIB (Supplementary Figure S2D). Thapsigargin, but not Tunicamycin, induced partial DR5 shifts that were accentuated by PERKi. Together, the findings in Figs. 1 and 3 reveal that changes in DR5 disulfide bonding caused by ER stressors differ among various cancer and non-transformed cell lines and that in some cases ER stress alone is sufficient to alter monomeric DR5 mobility and induce its oligomerization, while in other cases, these effects are potentiated by PERKi.

We next performed experiments in PERK or IRE1α knockout HepG2 cells or employed pharmacological PERK, IRE1α, or ATF6 inhibitors to examine the roles of these ER stress sensors in DDA-induced DR5 upregulation and oligomerization. PERK, but not IRE1α knockout, was associated with elevated DDA-induced DR5 oligomerization (Fig. 3G). Similarly, PERK inhibition increased Thapsigargin-induced DR5 upregulation and oligomerization in control cells, but not PERK knockout HepG2 cells in which Thapsigargin-induced DR5 upregulation and oligomerization was maximal without PERK inhibition (Supplementary Figure S2E). As expected, PERK inhibitor AMG44 [32] potentiated DDA-induced DR5 upregulation and oligomerization (Fig. 3H). The ATF6 inhibitor CEAPIN-A7 [33] was without effect, while the IRE1 inhibitor KIRA-6 [34] negated DR5 and TNFR1 upregulation by DDA treatment, prevented DDA-induced GRP78 upregulation, and decreased DR4 expression.

EGFR overexpression elevates DDA-induced accumulation of mis-disulfide bonded, monomeric DR5:

Given previous observations that DDAs downregulate HER-family proteins [15], and that EGFR overexpression sensitizes cells to DDA cytotoxic effects [35], we examined if DDAs differentially perturb DR5 disulfide bonding and levels in various cancer lines versus non-transformed cells. dFtcyDTDO, but not dMtcyDTDO, increased DR5 levels in the T47D luminal breast cancer cell line. dMtcyDTDO and dFtcyDTDO did not increase DR5 levels in non-transformed MCF10A mammary epithelial cells or in HaCaT human keratinocytes (Fig. 4A). In contrast, the DDAs induced robust increases in DR5 expression in the MDA-MB-468 and HCC1937 triple-negative breast cancer cell lines. Similarly, while dFtcyDTDO, dFtcyDTDO + PERKi, and Tunicamycin + PERKi decreased DR5 mobility in MDA-MB-468 cells (Fig. 4B), only Tunicamycin + PERKi increased DR5 levels and decreased its mobility in HMECs. Consistent with this result MDA-MB-468 cells are more sensitive than HMECs to DDA reduction of cell viability, while the colon cancer line LS513 exhibits intermediate sensitivity (Fig. 4C). Since MDA-MB-468 cells express high EGFR levels [36], we examined if EGFR overexpression is sufficient to confer sensitivity of DR5 to DDA-induced changes in disulfide bonding in MCF10A cells. dFtcyDTDO decreased DR5 mobility in the EGFR overexpressing cells, but not the vector control cells (Fig. 4D). The ER stressor Cyclosporine A did not induce this effect. Analysis of the disulfide bonding status of the DDA targets AGR2, ERp44, and PDIA1 showed that EGFR overexpression increased levels of disulfide-bonded oligomers of these PDIs as observed previously [4]. AGR2 is secreted by some cells [37–40], so we examined if dFtcyDTDO or Cyclosporine A caused AGR2 secretion. As expected [41], Cyclosporine A caused Cyclophilin B secretion, but AGR2 secretion was not observed under these conditions.

Fig. 4:

DDA upregulation of DR5 occurs in breast cancer cells or mammary epithelial cells overexpressing MYC or EGFR. A. The indicated cell lines were treated for 24 h with 2.5 μM dMtcyDTDO or dFtcyDTDO and analyzed by immunoblot under reducing conditions. B. MDA-MB-468 cells or Human Mammary Epithelial Cells (HMEC) were treated for 24 h as indicated and subjected to non-reducing immunoblot. Band “D” may represent PERK degradation products produced by Caspases. C. MTT viability assay in which the indicated cells were treated with the specified dMtcyDTDO concentrations for 72 h and viability evaluated spectroscopically. Results are presented as the average of eight replicates ± standard deviation. D. MCF10A cells engineered to overexpress EGFR or the corresponding vector control line were treated as indicated for 24 h with 2.5 μM dFtcyDTDO, 10 μM Cyclosporine A, or vehicle. The medium was collected and concentrated for analysis of secreted proteins, and the cell extracts were analyzed for internal proteins. Immunoblot analysis was performed under non-reducing conditions. E. MCF10A cells engineered to overexpress EGFR or MYC were treated for 24 h with 2.5 μM dFtcyDTDO or vehicle and cells were analyzed by non-reducing immunoblot. Bands shown represent monomeric protein isoforms. F. MCF10A cells engineered to overexpress EGFR or MYC, or MDA-MB-468 cells were treated for 24 h with 2.5 μM dMtcyDTDO or vehicle and cells were analyzed by non-reducing immunoblot. O and M represent Oligomeric and Monomeric protein isoforms in panels B., D. and F.

Since we previously observed that as with EGFR, MYC overexpression sensitizes cells to DDA-driven apoptosis [4], we examined the effects of MYC on DR5 disulfide bonding. dFtcyDTDO shifted DR5 mobility in MYC overexpressing cells, albeit not to the extent observed with EGFR overexpression (Fig. 4E). GRP78 immunoblot indicated that dFtcyDTDO induced a stronger ER stress response in the EGFR and MYC overexpressing cells as compared with the vector control. We previously showed that DDAs selectively upregulate DR5 in a subset of cancer cells and oncogene transformed epithelial lines [4, 5]. Comparison of DDA responses observed between the MCF10A stable cell lines and MDA-MB-468 breast cancer cells shows that vector control MCF10A cells exhibit little upregulation of DR5 or TNFR1 in response to DDA treatment, while upregulation of both Death Receptors is observed in MCF10A cells overexpressing EGFR or MYC and in the MDA-MB-468 cells (Fig. 4F). Immunoblot for EGFR and MYC shows that MYC is highly expressed in the MCF10A/MYC line compared with MDA-MB-468, while EGFR is much more highly expressed in MDA-MB-468 cells as compared with the MCF10A/EGFR line. The results in Fig. 4 show that this DR5 upregulation by ER stressors is associated with changes in the disulfide bonding of the monomeric forms of DR5, and in some cases is associated with disulfide-mediated DR5 oligomerization.

DR5 levels and oligomerization states are altered by perturbation of DR5 auto-inhibitory domain disulfides:

A recent study showed that the disulfide bond-rich extracellular domain of DR5 serves to prevent receptor oligomerization and pro-apoptotic signaling in the absence of its ligand TRAIL [42]. A more recent article narrowed down the DR5 autoinhibitory domain to a positive patch involving residues R154, K155, and R157 [43]. Based on a crystal structure [44], these basic residues share a common orientation due to two disulfide bonds, C156-C170 and C139–153 (Fig. 5A). We hypothesized that loss of these two disulfide bonds may disrupt the auto-inhibitory domain, culminating in DR5 clustering and activation of Caspase 8-Caspase 3-mediated apoptosis independently of TRAIL or DDA treatment. This hypothesis was tested by doxycycline-inducible expression of wild type and disulfide-defective DR5 point mutants. High-level inducible expression of the long form of wild type DR5 required induction by doxycycline combined with DDA treatment as described previously [5], while mutation of one or both of the Cys residues of the C160-C178 disulfide bond conferred high level DR5 expression in the absence of DDA treatment, as did the C81S mutation (Fig. 5B). Interestingly, DDA treatment still caused an upward shift in these mutants under reducing conditions suggesting that DDAs may disrupt multiple DR5 disulfide bonds. The DR5 disulfide-defective mutants, but not wild type DR5, also caused formation of high molecular weight DR4 oligomers in the absence of DDA treatment suggesting that endogenous DR4 may co-aggregate with ectopic, mis-disulfide bonded DR5 oligomers.

Fig. 5:

Genetic disruption of multiple DR5 disulfide bonds induces its stabilization and pro-apoptotic signaling. A. Structural model of DR5 showing its disulfide bonds, and the positive patch autoinhibitory domain described in the literature. B. Non-reducing immunoblot analysis of MDA-MB-468 cells engineered with doxycycline-inducible expression of wild type (WT) DR5 or the indicated Cys to Ser disulfide bond mutants. Cells were treated as indicated for 24 h with 1 μg/ml doxycycline and 2.5 μM dMtcyDTDO. The red arrow denotes DR4 oligomers that coincide with DR5 oligomerization. C. Reducing immunoblot analysis of the indicated MDA-MB-468 stable cell lines. Cells were treated for 24 h as specified with 1 μg/ml doxycycline or doxycycline + 10 μM Q-VD-OPH. The catalog numbers of DR5 and DR4 antibodies are shown. D. Non-reducing immunoblot analysis of the indicated MDA-MB-468 cell lines with doxycycline-inducible expression of wild type DR4 and DR5, and DR4 and DR5 C-terminal deletion constructs defective in apoptotic signaling. Cells were treated for 24 h as specified with 1 μg/ml doxycycline or doxycycline + 2.5 μM dMtcyDTDO. E. Non-reducing immunoblot analysis of the indicated MDA-MB-468 cell lines with doxycycline-inducible expression of wild type and apoptosis-defective DR5. Cells were treated for 24 h as specified with 1 μg/ml doxycycline or doxycycline + 2.5 μM dMtcyDTDO. F. Non-reducing immunoblot analysis of the indicated MDA-MB-468 doxycycline-inducible stable cell lines. Cells were treated for 24 h as indicated. G. Non-reducing immunoblot analysis of the indicated MDA-MB-468 cell lines with doxycycline-inducible expression of wild type and apoptosis-defective DR5. Cells were treated for 24 h as specified with 1 μg/ml doxycycline or doxycycline + 2.5 μM dMtcyDTDO. O and M represent Oligomeric and Monomeric protein isoforms in panels B and D-G.

We next examined if Caspase activation limits expression of DR5[C153S], DR4, or the murine TRAIL receptor (mDR5) by inhibiting Caspases with Q-VD-OPH [45]. Q-VD-OPH increased the inducible expression of all three receptors and prevented formation of the p18 fragment of Caspase 8, but not the p41/p43 fragment (Fig. 5C). This likely indicates that Q-VD-OPH does not inhibit receptor-driven Caspase 8 autocleavage but inhibits the previously described [46] Caspase 3 cleavage of Caspase 8. We further examined the relationship between DR5 and DR4 oligomerization and Caspase activation using C-terminal DR5 or DR4 deletion constructs (DR5 (ΔC41) and DR4 (ΔC43)) since similar mutants were previously shown incapable of coupling to Caspase 8 activation [47]. Doxycycline and doxycycline + dMtcyDTDO produced similar effects on the levels of the wild type and mutant DR5 and DR4, although the mutants were incapable of activating the Caspase 8-Caspase 3 cascade (Fig. 5D). Since DDAs cause ER stress, we examined if ER stress is independent of DR5-mediated Caspase activation. dMtcyDTDO upregulated ER stress markers, decreased AKT phosphorylation, and increased disulfide-mediated EGFR oligomerization irrespective of Caspase activation (Fig. 5E).

DR5 mutants lacking the disulfide bonds that form the positive patch exhibit high expression and oligomerization in the absence of DDA treatment, unlike wild type DR5 (Fig. 5F). Upregulation of DR5 by disruption of positive patch Cys residues was observed with two antibodies that recognize the C-terminal (#8074) or N-terminal (sc-166624) regions of DR5, although the latter antibody exhibited a strong binding preference for oligomeric DR5 isoforms over monomeric DR5 (Fig. 5G). The PhosphoSite database (Phosphosite.org) lists K245 as a major site of DR5 ubiquitination. Mutation of this site to Arg modestly increased receptor levels in the doxycycline and doxycycline + DDA treated samples, but did not mimic the ability of the C-S mutations to exhibit high level expression in the absence of DDA treatment. In summary, the results in Fig. 5 indicate that individual mutation of several different DR5 disulfides, including the positive patch disulfides, is sufficient for high level expression of DR5 and activation of Caspases independent of DDA treatment or ER stress. Further, stabilization of mis-disulfide bonded DR5 does not require Caspase activation, but activation of Caspase 8 by mis-disulfide bonded DR5 requires its C-terminus that is necessary for DISC formation [47].

DDAs activate autophagy and autophagy inhibitors potentiate DDA-induced DR5 accumulation:

Proteasomal degradation (ERAD) and autophagy are important modes for the disposal of misfolded proteins. However, the fate of mis-disulfide bonded, aggregated proteins in the secretory pathway is underexplored. Since DDAs induce ER stress [35], and ER stress frequently activates autophagy [48], we examined if DDAs stimulate autophagy. dFtcyDTDO treatment induced an upward DR5 shift in a concentration-dependent manner. GPR78 expression and the autophagy marker LC3 lipidation were both increased to maximal levels at the lowest dFtcyDTDO concentration tested (Supplementary Figure S3A). Treatment with the autophagy/lysosome inhibitor Bafilomycin A1 increased levels of monomeric and oligomeric DR5 isoforms in the absence of DDA treatment (Supplementary Figure S3B). Similar studies employing the autophagy/lysosome inhibitor Chloroquine showed increased DR5 levels and accumulation of oligomeric EGFR compared with vehicle treatment. Combining dFtcyDTDO with Chloroquine or PERKi increased DR5 levels and Caspase 8 cleavage over that observed with dFtcyDTDO alone (Supplementary Figure S3B). Viability assays on cells treated as in Supplementary Figure S3C showed that combining dFtcyDTDO with either PERKi or Chloroquine reduced viability more than dFtcyDTDO and combining the three agents decreased viability the most (Supplementary Figure S3D). An inhibitor of the autophagy PI3-kinase VPS34 (VPS34i, [49]) increased DR5 expression to a similar extent as Bafilomycin, and combining dFtcyDTDO with either VPS34i or Bafilomycin increased expression of monomeric DR5 more than each individual treatment (Supplementary Figure S3E). Of these treatments, only dFtcyDTDO or the dFtcyDTDO-containing treatments upshifted monomeric DR5. The strong accumulation of upshifted monomeric DR5 caused by autophagy inhibitors suggests that autophagy plays a role in degrading mis-disulfide bonded DR5. We also examined if combined proteasome inhibition with MG132 and autophagy/lysosome inhibition with Bafilomycin mimicked the ability of DDAs + PERK inhibition to upregulate DR5 or TNFR1 (Supplementary Figure S3F). Bafilomycin A1 alone caused marked accumulation of oligomeric DR5, DR4, and TNFR1, suggesting that lysosomal degradation may be the primary disposal route for aberrantly disulfide bonded Death Receptor oligomers. DDA + PERK inhibition upregulated oligomeric DR5 and TNFR1, but not DR4. Further increases of monomeric and oligomeric Death Receptors were not observed upon combining DDAs with PERKi AMG44, MG132, and Bafilomycin.

DDAs upregulate DR5 paralog DcR2:

The pro-apoptotic TRAIL receptors DR4 and DR5 share conserved disulfide-rich domains with the TRAIL decoy receptors DcR1 and DcR2 that bind TRAIL but cannot activate Caspases. Specifically, the DR5 autoinhibitory domain is largely conserved with DR4, DcR1, and DcR2 (Supplementary Figure S4A). We considered that DDAs might stabilize DcR1 or DcR2 in a similar manner as DR5. Since we previously found that the prolyl isomerase inhibitor Cyclosporine A potentiated DDA cytotoxic effects [6], we examined Cyclosporine A effects on the levels of decoy receptors. We did not detect DcR1 in the cell lines examined but observed upregulation of DcR2 after dFtcyDTDO treatment (Supplementary Figure S4B). We also observed that Cyclosporine A, but not FK606, which inhibits a different family of prolyl isomerases, decreased DDA upregulation of DcR2, but not DR5. dFtcyDTDO increased DcR2 levels more at low concentrations than high concentrations in some experiments, but irrespective of the pattern of DcR2 upregulation by dFtcyDTDO, it was blocked by co-treatment with Cyclosporine A (Supplementary Figure S4B–D). Quantitation of band intensities revealed that Cyclosporine A treatment potentiated the effects of low (310 nM) dFtcyDTDO concentration on DR5 oligomerization (Supplementary Figure S4E, left panel) and decreased DcR2 upregulation by dFtcyDTDO (Supplementary Figure S4E, right panel).

DR5 bonding status influences antibody recognition, but not trafficking to the cell surface:

Since sERr prevents some transmembrane receptors from reaching the cell surface [8] and DR5 was shown to be activated in the Golgi by binding to aggregated proteins [50], we examined DR4 and DR5 cell surface labeling by flow cytometry. Using the Clone DJR2–4 (7–8) antibody for flow cytometry analyses, Doxycycline induction of the wild type, C153S, and C156S mutants showed increased DR5 cell surface labeling, however co-treatment with dFtcyDTDO decreased apparent DR5 surface localization (Fig. 6A, upper panel). DR4 flow cytometry studies showed that doxycycline increased DR4 surface levels and this was not altered by dFtcyDTDO co-treatment. Since recognition of DR5 by the flow cytometry antibody could be hindered by changes in DR5 disulfide bonding, we examined levels of DR5 and its disulfide bonding mutants using three different commercially available antibodies (Fig. 6B). The #8074 antibody directed against the cytoplasmic, C-terminal portion of DR5 recognized all of the DR5 proteins, including the monomeric and oligomeric forms of DR5. As shown (in Fig. 5 and elsewhere [5]), wild type DR5 was only maximally expressed in cells induced with doxycycline and treated with DDAs. The sc-166624 DR5 antibody directed against the N-terminal cysteine-rich region preferentially recognized the oligomeric forms of DR. In contrast, Clone DJR2–4 (7–8) directed toward the N-terminal cysteine-rich portion of DR5 did not recognize the C119S/C137S DR5 mutant, but bound the C153S, C156S, C160S, and C160S/C178S DR5 mutants. However, dFtcyDTDO treatment ablated DR5 recognition by this antibody. These observations suggest that binding of Clone DJR2–4 (7–8) antibody is sensitive to DR5 disulfide bonding, which is altered by DDA treatment. Since this antibody is commonly used for DR5 labeling in flow cytometry studies and multiple ER stressors, including DDAs, alter DR5 disulfide bonding, lack of signal with this antibody may be indicative of changes in DR5 disulfide bonding rather than DR5 downregulation or internalization.

Fig. 6:

Effects of altered disulfide bonding on DR5 cell surface localization and antibody recognition. A. Flow cytometry analysis of the indicated doxycycline-inducible MDA-MB-468 stable cell lines with an antibody to DR5 (Clone DJR2–4 (7–8)) (top panel) or DR4 (bottom panel). Prior to analysis, cells were treated for 24 h as indicated with 10 μM Q-VD-OPH, 1 μg/ml doxycycline, or 2.5 μM dFtcyDTDO. Dots represent the average values from three independent biological replicates performed in triplicate. * Represents p < 0.05, **** represents p < 0.0001, and ns represents not significant (p > 0.05). B. Non-reducing immunoblot analysis of the indicated MDA-MB-468 stable cell lines treated for 24 h as specified. Note the alternate staining patterns observed with different DR5 antibodies. O and M represent Oligomeric and Monomeric protein isoforms. C. The indicated MDA-MB-468 stable cell lines were treated for 24 h, as indicated with 1 μg/ml doxycycline and 2.5 μM dFtcyDTDO and subjected to cell surface protein biotin labeling. Cell surface proteins (External; Ext.) were affinity purified with Streptavidin-agarose, and the unlabeled flow-through (Internal; Int.) proteins were also collected. Both fractions were analyzed by non-reducing immunoblot using the indicated antibodies. D. Cell surface protein labeling experiment as in panel C except that cell lines were treated with the indicated combinations of 1 μg/ml doxycycline, 2.5 μM dFtcyDTDO, and 10 μM Cyclosporine A. E. Immunofluorescence microscopy of MDA-MB-468 cells stably expressing tetracycline-inducible wild type DR5. Cells were treated as indicated for 24 h with 1 μg/ml doxycycline, 2.5 μM dMtcyDTDO, or both doxycycline and dMtcyDTDO, and stained with the sc-166624 monoclonal antibody and counterstained with DAPI. Lower panels show expanded images taken from the boxed regions of the upper panels. F. MDA-MB-468 cells were treated for 24 h with vehicle or 2.5 μM dMtcyDTDO, followed by addition of 10 μg/sample of Biotinylated TRAIL where indicated. Biotin-TRAIL containing complexes were collected with Streptavidin-Agarose and the complexes analyzed by non-reducing immunoblot.

Previous work showed that DDA treatment increased surface localization of DR5 as measured by biotin labeling [4]. The same approach was used to examine the localization of the DR5 mutants to the cell surface. The results showed that the C119S/C137S, C153S, and C156S mutants trafficked to the cell surface, particularly in the context of DDA treatment (Fig. 6C). Under these conditions, expression of DR5 disulfide bonding mutants did not elicit an ER stress response as indicated by the markers GRP78, XBP1s, and PERK activation (phosphorylation). However, dFtcyDTDO activated all these indicators of ER stress. A similar cell surface biotinylation experiment showed trafficking of wild type and mutant DR5 to the cell surface. Surface localization of the C119S/C137S and C153S DR5 mutants occurred in both the presence or absence of dFtcyDTDO or the ER stress inducer Cyclosporine A (Fig. 6D). Expression of mis-disulfide bonded DR5 mutants did not upregulate the ER stress or autophagy markers GRP78, or LC3, respectively, but DDA treatment upregulated both markers. Together, these results indicate that under conditions where DR5 disulfide bonding is perturbed by either mutagenesis or DDA treatment, DR5 traffics to the cell surface. This is consistent with previous work showing that DDA treatment increases cancer cell sensitivity to the DR4/5 ligand TRAIL [5]. Immunofluorescence microscopy analysis of unpermeabilized MDA-MB-468 with the sc-166624 DR5 antibody was performed to further confirm the DR5 cell surface localization indicated by cell surface biotinylation. This study showed that, similar to immunoblot results, DDA treatment and doxycycline strongly increased immunostaining of inducibly expressed wild type DR5 (Fig. 6E). Death Inducing Signaling Complex (DISC) assays were carried out to determine if DDA treatment alters the composition of Death Receptor complexes isolated using biotinylated TRAIL. The results showed that biotinylated TRAIL pulled down similar levels of DR5, DR4, full length and cleaved Caspase 8, and the adaptor protein FADD (Fig. 6F). This observation could indicate that different pools of DR5 exist upon DDA treatment, a pool competent to bind TRAIL and activate Caspase 8, and a separate DR5 pool that activates DR5 independently of TRAIL. This interpretation is consistent with the increased Caspase 8 cleavage observed upon DDA/TRAIL co-treatment and TRAIL-independent DDA induction of Caspase 8 cleavage.

DDA safety and identification of a metabolically stable DDA analog:

Previous work with the DDAs RBF3 [15] and tcyDTDO [5] did not reveal evidence of toxicity under conditions in which they induce the death of primary and metastatic breast cancer cells in mouse models. We also examined the metabolism of the DDA tcyDTDO by liver microsomes [6], but did not present analyses of the stability of dMtcyDTDO and dFtcyDTDO toward metabolism in liver and intestinal microsomes. Recent work demonstrated the activity of both dMtcyDTDO and dFtcyDTDO in mouse models of breast cancer [12], but did not examine the effects of these compounds on normal tissues such as the liver or hematopoietic cells. Examination of breast tumor tissue from mice treated with vehicle or 10 mg/kg dMtcyDTDO showed extensive death of tumor tissue in the dMtcyDTDO-treated, but not the vehicle-treated mice (Fig. 7A, upper panels). Liver tissues from vehicle or dMtcyDTDO-treated mice were indistinguishable (Fig. 7A, lower panels). Analysis of complete blood cell counts from tumor-bearing mice treated with vehicle or 10 mg/ml dMtcyDTDO were in the normal range for healthy mice (Supplementary Figure S5). The apparent decrease in platelets across the samples was likely due to partial clotting prior to analysis.

Fig. 7:

Microsome stability of select DDAs and lack of DDA effects on liver morphology. A. Hematoxylin and Eosin stained breast tumor (upper panels) and liver tissue samples (lower panels) from mice bearing 012/LVM2/LR10 tumors after treatment with vehicle (peanut oil) or 10 mg/kg dMtcyDTDO by oral gavage for 20 days. B. Stability of tcyDTDO, dMtcyDTDO, or dFtcyDTDO metabolism in rat or human liver microsomes in the presence or absence of 1 mM NADPH. Verapamil serves as a positive control. C. Stability of tcyDTDO, dMtcyDTDO or dFtcyDTDO metabolism in rat or human intestinal microsomes in the presence or absence of 1 mM NADPH. Verapamil serves as a positive control. Data points are plotted as the average (N = 3), with error bars representing standard deviation.

Stability studies in human liver microsomes supplemented with NADPH showed that dFtcyDTDO was metabolically stable (t_1/2 > 60 min), while tcyDTDO, and dMtcyDTDO were metabolized by phase I enzymes with half-lives of 11.9, and 47.9 min, respectively (Fig. 7B). Similar studies employing human intestinal microsomes showed that the half-life of all three DDA compounds exceeded 60 min (Fig. 7C). Although more thorough analysis of these DDAs is needed, studies performed to date indicate that dFtcyDTDO has a more favorable metabolic stability profile than tcyDTDO or dMtcyDTDO.

Discussion

Previous work showed that the DDA compounds induce ER stress, which is associated with high-level DR5 expression and disulfide-mediated oligomerization [5, 12, 35]. Further, mutational disruption of a subset of DR5 disulfide bonds was demonstrated to stabilize DR5 and trigger disulfide-mediated DR5 oligomerization. However, the relationship between the DDA-induced ER stress response and DDA effects on DR5 were unexplored. The results presented here show that individual mutational disruption of a total of five of the seven DR5 disulfide bonds tested each caused similar degrees of DR5 stabilization and oligomerization. This includes the disulfides present within the previously described auto-inhibitory domain [42, 43]. The observation that DDAs still induce a mobility shift of monomeric disulfide point mutants of DR5 suggests that while loss of individual disulfide bonds is sufficient to stabilize DR5 and promote pro-apoptotic signaling, DDAs disrupt multiple DR5 disulfide bonds.

Studies have demonstrated transcriptional regulation of DR5 through a PERK-ATF4-CHOP-DR5 pathway [51, 52–56]. In this model, PERK inhibition is predicted to block DR5 upregulation by ER stress. Our previous studies in breast cancer lines found little effect of knocking out or overexpressing CHOP on DR5 levels, suggesting that other DR5 regulatory mechanisms exist [5]. Additional work is needed to fully elucidate the mechanism by which in the context of ER stress, PERKi treatment alters DR5 levels, disulfide bonding status, and intracellular localization to promote extrinsic apoptosis. We found here that PERK inhibition potentiated DR5 pro-apoptotic signaling, but only modestly increased DR5 levels. A report showed that DR5 is activated by binding to misfolded proteins in the Golgi [50, 57], so PERKi might elevate DR5 oligomerization through this mechanism. Alternatively, overriding ISR may activate DR5 by increasing protein folding flux under ER stress conditions and surpass the ability of the DDA targets ERp44, PDIA1, and AGR2 to catalyze disulfide bond formation or their functions in protein folding checkpoints. A recent report showed that breast cancer metastases exhibit elevated ER stress and are responsive to a new, more selective PERK inhibitor [58]. This, along with our previous study showing DDA activity against breast cancer metastases [5], provides a rationale for DDA/PERKi combination therapy for the treatment of metastatic breast cancer. Likewise, the observation that Cyclosporine A blocks DDA-induced upregulation of DcR2 further supports the previous contention [6] that DDA/Cyclosporine co-treatment may exhibit enhanced efficacy against breast malignancies.

Overexpression of the EGFR or MYC oncoproteins was shown to sensitize cells to DDA cytotoxic effects, but the underlying mechanisms were not investigated [5]. Results presented here show that EGFR or MYC overexpression permits DDA perturbation of DR5 disulfide bonding that is not observed in vector control non-transformed MCF10A mammary epithelial cells. This may partially explain the ability of DDAs to mimic the cancer-specific cytotoxic effects of the DR5/4 ligand TRAIL. Interestingly, expression of disulfide bonding mutants of DR5 does not trigger an ER stress response or activate autophagy. Further, mis-disulfide bonded DR5 traffics to the cell surface, unlike sERr clients [8], consistent with the previous observation that DDAs synergize with TRAIL to kill cancer cells [5]. The present results extend previous work on the relationship between ER stress and DR5 activation by showing that in addition to DDAs, other ER stressors, including Tunicamycin and Thapsigargin, can alter DR5 disulfide bonding in a manner that is potentiated in some cases by PERKi co-treatment. A subject of ongoing investigation is why ER stressors that alter DR5 disulfide bonding do not have the same effect on its paralog DR4. Together, our work [4, 5, 35] and that of others [50, 56], suggest that DR5 has evolved as a direct sensor and effector of ER stress/protein misfolding and that ER stress can activate DR5 through transcriptional mechanisms and at the protein level through altered DR5 disulfide bonding and DR5 binding to misfolded proteins. Importantly, DR5 exhibits a TRAIL-independent gain of function under these conditions that inactivate a wide variety of other transmembrane oncoproteins, including EGFR and MET. While our results point to a central role for the extrinsic apoptosis pathway in mediating DDA-induced cancer cell death, it is likely that the parallel reduction of mitotic/survival signaling caused by induction of selective ER retention (sERr) also promotes execution of apoptosis. This is consistent with previous results showing DDA-induced downregulation of EGFR, HER2, and HER3 and decreased Akt and ERK phosphorylation [5, 6, 15, 35].

DDA studies have been performed largely in breast cancer cell lines, therefore it will be critical to determine the importance of these DDA-driven effects in non-transformed cells and across other tumor types. This DDA cancer selectivity is further supported by the Broad Institute’s Dependency Map (DepMap; depmap.org/portal/). Of the 22 human disulfide isomerases, only four, ERp44, AGR2, AGR3, and TMX1, are considered as “strong dependencies” in the DepMap, and our studies have shown DDAs to inhibit three of these, ERp44, AGR2, and AGR3. It is possible that ERp44, AGR2, and AGR3 client proteins vary with tumor type. As an example, out of 17,347 genes in the DepMap CRISPR screen, the colon cancer lines C80, COLO205, LS513, and SNUC4 rank AGR2 as the first, fourth, third, and fifth most important gene for viability, respectively. Interestingly, ERN2 encodes the IRE1α homolog IRE1β whose expression is restricted to Goblet cells. The DepMap lists IRE1β as the top predictor of AGR2 dependence. Two recent reports show that AGR2 functions as an inhibitor of IRE1β that overcomes the cytotoxic effects of this enzyme [59], and that the active site Cys81 of AGR2 is required for IRE1β inhibition [60, 61]. It will be important to determine if DDAs exhibit anticancer activity against tumor lines in which AGR2 is required to prevent IRE1β-mediated cancer cell death. This is an area of active investigation by our team.

As an inhibitor of ERp44, DDAs may override two key protein disulfide bonding checkpoints, the retrograde Golgi-ER recycling of secretory proteins discovered by Sitia and Colleagues [62], and the selective ER retention mechanism of some receptor tyrosine kinases discovered by Tirosh and Colleagues [8]. Based on these new mechanistic insights and molecular and pharmacological tools, the stage is set to investigate the molecular and biological functions of the non-canonical PDIs, and strongly selective cancer dependencies, ERp44, AGR2, and AGR3. The results in Fig. 7 showing that dFtcyDTDO exhibits the best metabolic stability against liver microsomes, while having similar potency to dMtcyDTDO nominates dFtcyDTDO as the basis for ongoing optimization efforts. As no cell permeable, selective ERp44 and AGR2 catalytic site inhibitors are available, future efforts will focus on modifications to carbon atoms adjacent to the DDA thiosulfonate group in efforts to generate ERp44- and AGR2-selective inhibitors for tool compounds and potential anticancer therapeutics.

Implications Statement.

Extreme endoplasmic reticulum stress triggers triage of transmembrane receptor production, whereby mitogenic receptors are downregulated and death receptors are simultaneously elevated.