Anal Bioanal Chem DOI 10.1007/s00216-015-8694-2
RESEARCH PAPER
Identification of known drugs targeting the endoplasmic reticulum stress response Kun Bi 1 & Kana Nishihara 2,3 & Thomas Machleidt 4 & Spencer Hermanson 4 & Jun Wang 4 & Srilatha Sakamuru 2 & Ruili Huang 2 & Menghang Xia 2
Received: 24 February 2015 / Revised: 3 April 2015 / Accepted: 10 April 2015 # Springer-Verlag Berlin Heidelberg 2015
Abstract The endoplasmic reticulum (ER), a multifunctional organelle, plays a central role in cellular signaling, development, and stress response. Dysregulation of ER homeostasis has been associated with human diseases, such as cancer, inflammation, and diabetes. A broad spectrum of stressful stimuli including hypoxia as well as a variety of pharmacological agents can lead to the ER stress response. In this study, we have developed a stable ER stress reporter cell line that stably expresses a β-lactamase reporter gene under the control of the ER stress response element (ESRE) present in the glucoseregulated protein, 78 kDa (GRP78) gene promoter. This assay has been optimized and miniaturized into a 1536-well plate format. In order to identify clinically used drugs that induce ER stress response, we screened approximately 2800 drugs from the NIH Chemical Genomics Center Pharmaceutical Collection (NPC library) using a quantitative highthroughput screening (qHTS) platform. From this study, we have identified several known ER stress inducers, such as 17AAG (via HSP90 inhibition), as well as several novel ER stress inducers such as AMI-193 and spiperone. The confirmed drugs were further studied for their effects on the phosphorylation of eukaryotic initiation factor 2α (eIF2α), the Xbox-binding protein (XBP1) splicing, and GRP78 gene * Menghang Xia [email protected] 1
Life Science Solutions Group, Thermo Fisher Scientific, Rockford, IL 61105, USA
2
National Center for Advancing Translational Sciences, National Institutes of Health, Bethesda, MD 20892, USA
3
Radiation Genetics, Graduate School of Medicine, Kyoto University, Yoshidakonoe, Sakyo, Kyoto 606-8501, Japan
4
Promega Corporation, Fitchburg, WI 53711, USA
expression. These results suggest that the ER stress inducers identified from the NPC library using the qHTS approach could shed new lights on the potential therapeutic targets of these drugs. Keywords ATF6 . ER stress pathways . NPC library . qHTS . XBP1
Introduction The endoplasmic reticulum (ER) is a multifunctional organelle where protein folding/maturation, lipid biosynthesis, calcium storage, and release occur. Disruption of ER homeostasis which triggers ER stress response has been linked to many human diseases, such as cancer and inflammation [1–3]. ER stress leads to the activation of three signaling pathways through ER membrane protein sensors including protein kinase-like ER kinase (PERK), inositol-requiring enzyme (IRE1α), and transcription factor 6 (ATF6) [4]. In the absence of ER stress inducers, these ER proteins are sequestered by an ER chaperone protein called glucose-regulated protein, 78 kDa (GRP78; also known as binding immunoglobulin protein (BiP)). Upon ER stress, the unfolded/misfolded proteins accumulate in the ER and bind to GRP78, releasing and activating the three ER stress sensor proteins. Activated PERK, a serine/threonine kinase, phosphorylates and inhibits the translation initiation factor eIF2α, which then shuts down general protein translation and activates transcription factor ATF4 leading to the gene expression of ER chaperones and factors involved in glutathione and amino acid biosynthesis and antioxidative stress response. IRE1α protein has both the serine/ threonine kinase domain and the endoribonuclease domain. Once activated, IRE1α gains the endoribonuclease activity and processes an intron from the X-box-binding protein
K. Bi et al.
(XBP1) resulting in the alternate splicing product, which then translates into a XBP1 protein, a transcription factor that activates the expression of genes involved in protein degradation and restoring protein folding. The GRP78-released ATF6 upon ER stress translocates to the Golgi apparatus, where it is cleaved to produce the active ATF6 fragment, which then translocates back to the nucleus to activate the expression of downstream ER stress genes such as GRP78 and GRP94. All these signaling events aim to decrease the accumulation of unfolded proteins in the ER. However, when ER stress is severe and prolonged, it can lead to cell death [5]. Due to its significant involvement in the development of various diseases [6], ER stress pathway and its components have been targeted for drug discovery. Several ER stress inducers for therapeutic intervention have been identified. For example, bortezomib, a proteasome inhibitor that can induce severe ER stress leading to apoptosis, has been clinically applied to treat multiple myeloma patients [7]. Salubrinal, a phosphatase inhibitor that protects cells against ER stress induced apoptosis by agonizing PERK-eIF2α pathway, has the therapeutic potential for treating neurodegenerative diseases [8]. Since many environmental stress signals lead to ER stress response, ER stress induced apoptosis could also be the mechanism for the therapeutic effect of other cancer drugs such as Sorafenib [9]. To investigate whether some of the clinically used drugs can induce ER stress response and to identify novel ER stress inducers, we have developed a high-throughput compatible ATF6 reporter gene assay and screened the NIH Chemical Genomics Center Pharmaceutical Collection (NPC library) [10] consisting of 2816 small molecule drugs approved for human or animal use in the USA, Canada, Japan, and Europe and compounds registered for testing in clinical trials. Each compound was tested at 15 different concentrations in a quantitative high-throughput screening (qHTS) format [11]. We then performed follow-up studies to determine the effects of the confirmed hit compounds on eIF2α phophosphorylation, XBP1 alternate splicing, and ER chaperone GRP78 expression. Several known ER stress inducers, such as 17-AAG and cyclosporine A were identified, as well as several novel ER stress inducers such as AMI-193 and spiperone.
Materials and methods Generation of ESRE-bla HeLa reporter gene cell line HeLa cells were transduced with lenti-virus created from the ViraPower™ Lenti-viral Expression System (Life Technologies, Carlsbad, CA) pLenti6 ESRE-bsd/bla containing the −130 to −40 regions of the GRP78 promoter and selected for blasticidin resistance. The blasticidin-resistant pools were stimulated with tunicamycin for 4 h. Responding cells were
collected into single-cell clones onto 96-well plates using FACS. The clones that displayed the highest fold signal change with tunicamycin compared with a dimethylsulfoxide (DMSO) control were selected and expanded, and the best responding clone was chosen for assay development. ER stress beta-lactamase reporter gene assay and RNAi experiment The ESRE-bla HeLa cells were cultured in Dulbecco’s modified Eagle’s medium (DMEM) containing 10 % dialyzed fetal bovine serum (FBS), 0.1 mM nonessential amino acids (NEAA), 1 mM sodium pyruvate, 25 mM 4-(2hydroxyethyl)-1-piperazineethanesulfonic acid (HEPES), 100 U/mL penicillin, and 100 μg/mL streptomycin (All cell culture reagents are from Life Technologies). Assays were performed in an assay medium containing phenol-red-free OPTI-MEM, 0.5 % dialyzed FBS, 0.1 mM NEAA, 1 mM sodium pyruvate, 100 U/mL penicillin, and 100 μg/mL streptomycin. Cells were incubated with indicated ER stressinducing agents for 5 to 6 h. β-lactamase substrate LiveBLAzer™ FRET B/G detection mix (Life Technologies) was added to the cells and incubated at room temperature for 2 h. Fluorescence intensity at 406 nm excitation and 460 and 530 nm emission was measured using TECAN Safire2 (Tecan Group Ltd, Maennedorf, Switzerland) with optimal gain settings determined by the instrument. After subtracting fluorescence background from cell-free wells, the ratio of fluorescence intensity at 460 vs 530 nm (designated as 460/ 530 nm) was calculated. Response ratio is a measurement of the assay window and is calculated as the 460/530-nm emission ratio of the stimulated wells divided by the 460/530-nm emission ratio of the unstimulated wells. Response ratios were plotted against test ligand concentration in log scale and then analyzed using Prism software (GraphPad Software, Inc. San Diego, CA). Sigmoidal concentration-response equation with varying slope was used to fit the data and generate concentrations of half-maximal stimulation (EC50) value. Z′-factor values were calculated as: Z′-factor=1−((3×stdevunstim +3× stdevmaxstim)/(avgmaxstim −avgunstim)). Assays were performed in quadruplicates. For RNAi experiments, ESRE-bla HeLa cells were plated in antibiotic-free growth medium in 96-well assay format at 7500 cells/well and transfected using Lipofectamine™ RNAiMAX and 20 nM of the following panel of RNAi duplexes/control conditions: mock transfected (RNAiMAX reagent only, no RNAi duplex), Stealth™ RNAi Negative Control Med GC, beta-lactamase RNAi-positive control duplex (sense strand sequence, GCGCAAACUAUUAACUGG CGAACUA), ATF4#1 (GCAGCCACUAGGUACCGCCA GAAGA), ATF4#2 (UAUCAAAUCUUUCAGGUACUGG AUC), ATF6#1 (CCCUAGUGUGGGACCUGCAAAU CAA), ATF6#2 (CCGUAUUCUUCAGGGUGUUGUG
Profiling of drugs targeting the endoplasmic reticulum stress pathway
GAA). At 60 h posttransfection, cells were stimulated with tunicamycin and the beta-lactamase reporter gene assay was performed and data was analyzed as described above.
same protocol as described above except 24 point titrations were within one 1536-well plate. Data analysis
NCGC pharmaceutical collection and compounds used in the study The NIH Chemical Genomics Center Pharmaceutical Collection consists of 2816 small molecule compounds which include drugs approved for human or animal use in the USA, Canada, Japan, and Europe and compounds registered for testing in clinical trials. The compound plate preparation was described elsewhere [10, 11]. Briefly, all the compounds were dissolved in dimethylsulfoxide (DMSO), prepared in a 1536well plate format, and stored desiccated at room temperature for as long as 6 months when in use or heat sealed and stored at −80 °C for long-term storage. For follow-up studies, primaquine, spiperone, cyclosporine A, 6-thioguanine, amlodipine, and telmisartan, were purchased from Sigma-Aldrich (St. Louis, MO). AMI-193 and roxindole were purchased from Tocris Bioscience (Ellisville, MO). Bortezomib was purchased from LC laboratories (Boston, MA). Quantitative high-throughput screening The ESRE-bla HeLa cells were suspended in assay medium. The cells were dispensed at 1500 cells/6 μL/well in 1536-well black wall/clear bottom plates (Greiner Bio-One North America, NC) using a Flying Reagent Dispenser (FRD; Aurora Discovery, CA) or Thermo Scientific Multidrop Combi (Thermo Fisher Scientific Inc., Waltham, MA). After the cells were incubated at 37 °C overnight, 23 nL of compounds at 15 concentrations from the NPC library was transferred to the assay plate by a pin tool (Kalypsys, San Diego, CA). The final concentration of the compounds in the 6-μL assay volume ranged from 0.5 nM to 38 μM. The positive control plate format was as follows: Column 1, 17-AAG, ranged from 23 pM to 7.6 μM; column 2, 3.8 μM 17-AAG (EC100); column 3, DMSO only; and column 4, 1.9 μM 17-AAG; columns 5 to 48, compounds from NPC library. The plates were incubated for 6 h at 37 °C. One microliter of LiveBLAzer™ B/G FRET substrate (Life Technologies) detection mix was added and the plates incubated at room temperature for 2 h. Fluorescence intensity (405 nm excitation, 460 and 530 nm emissions) was measured using an EnVision plate reader (Perkin Elmer, Shelton, CT). Data was expressed as the ratio of 460/530 nm emissions (without background subtraction). Use of ratio metric readout decreases the well-to-well variation of the cell numbers in the assay plate [12]. In the confirmation study, selected active compounds were re-tested in 24 point titrations with concentration ranging from 5 pM to 38 μM in the ESRE-bla HeLa reporter assay using the
Data normalization and curve fitting was performed as previously described [13]. Briefly, raw plate reads for each titration point were first normalized relative to 17-AAG control (3.8 μM, 100 %) and DMSO only wells (basal, 0 %), and then corrected by applying a pattern correction algorithm using compound-free control plates (DMSO plates). Concentration-response titration points for each compound were fitted to the Hill equation yielding EC50 and maximal response (efficacy) values. Compounds from qHTS were classified into four major curve classes using the set of criteria listed in previous studies [14, 15]. From the primary screen, eight compounds were selected and purchased from commercial vendors for further extensive studies based on potency and quality of concentration-response curve and novelty of mechanism. LanthaScreen® cellular eIF2α phosphorylation assay GripTite™ 293 MSR cells stably expressing GFP-eIF2α (Life Technologies) were plated onto a white tissue culture treated 384-well assay plate (Corning Incorporated, Tewksbury, MA) in growth medium (DMEM with GlutaMax, 10 % dialyzed FBS, 25 mM HEPES, 0.1 mM NEAA, 100 U/mL penicillin, and 100 μg/mL streptomycin). Cells were left untreated or treated with indicated ER stress agents (such as tunicamycin) or hit compounds for 2 h before cell lysis and assay equilibration with complete lysis buffer including 2 nM terbiumlabeled anti-eIF2α pSer52 antibody (Life Technologies). TR-FRET emission ratios were determined on a BMG PHERAstar fluorescence plate reader (BMG LABTECH, Cary, NC; excitation at 340 nm, emission 520 and 490 nm as described previously [16]. RT-PCR and XBP1 alternate splicing ESRE-bla HeLa cells were left untreated or treated with 20 μM indicated hit compounds for 16 h. Cells were then pelleted and total RNA was isolated using Ambion® RiboPure™ RNA Purification Kit (Life Technologies) and converted to complementary DNA (cDNA) using SuperScript®III reverse transcriptase (Life Technologies). The cDNA for XBP1 was then amplified using polymerase chain reaction using Platinum® Taq DNA Polymerase (Life Technologies). The nucleotide sequences of the XBP1 primer pairs (5′ to 3′) were TTACGAGAGAAAACTCATGGCC and GGGTCCAAGTTGTCCAGAATGC. PCR products were separated by electrophoresis on 1.2 % agarose gels and visualized with SYBR® Safe DNA gel stain (Life Technologies).
K. Bi et al.
Fig. 1 CellSensor® ESRE-bla HeLa cell line validation. a ER stress response to known ER stress inducers in ESRE-bla HeLa cell. A23187 (EC50 =17.8 nM), bortezomib (EC50 =40.0 nM), ionomycin (EC50 = 44.8 nM), thapsigargin (EC 50 = 5.6 nM), tunicamycin (EC 50 = 301.7 nM), and 17-AAG (EC50 = 397.6 nM). b Validation of the involvement of ATF6 in the response of the ESRE-bla HeLa cell. The 460/530-nm emission ratio was plotted in y-axis for each RNAi oligo.
Med-GC and β-lactamase are the controls. c Concentration-response curve of 17-AAG (EC50 =590 nM) in 1536-well plate. Error bars represent SD from at least three independent experiments. The statistical significance of differences was evaluated by the Student’s t test. *p100
Yes Yes Yes Yes No NA No No No
NT 9.5 4.4 17 6.1 32 2.3 2.1 4.7
NA not available NT not tested
robust response to a number of other known ER stressinducing compounds including calcium ionophore A23187 (EC50 =17.8 nM) and ionomycin (EC50 =44.8 nM) which deplete calcium in the ER, thapsigargin (EC50 =5.6 nM) which inhibits the ER calcium ATPase and decreases calcium ions from the ER and 17-AAG (EC50 =397.6 nM), a HSP90 inhibitor that increases unfolded proteins in the ER [17]. In addition to the EC50 value differences, these compounds generated different levels of response in this reporter line. Tunicamycin and 17-AAG generated the highest response ratio of 5, followed by thapsigargin with a response ratio of 4 and A23187 and ionomycin of 3 (Fig. 1a). Interestingly, this reporter line shows little response to bortezomib, a proteasomal inhibitor that causes protein aggregation in the ER and induces severe ER stress response leading to apoptosis [18]. To validate the involvement of ATF6 in the response of the ESRE-bla HeLa cells to tunicamycin, ATF6-specific Stealth™ RNAi oligos were transfected into the reporter cells prior to the stimulation with tunicamycin. Both ATF6 RNAi oligos significantly knocked down the ER stress response (p100 μM), and spiperone (EC50 =85.4 μM)
DM SO ,c on tro DM l SO ,c on tro Te l lm isa rta n Sp i pe ro n e Cy clo sp o ri ne 6 -T A hi o gu an i ne Pr im aq ui n e Bo rte zo mi b Am l od i pi ne AM I-1 93 Ro xin do le
(bp) 400 300
Ma rke r
Profiling of drugs targeting the endoplasmic reticulum stress pathway
U S
200
Fig. 4 XBP1 mRNA splicing monitored by RT-PCR. HeLa cells were treated with indicated compounds at 20 μM for 16 h. Cells were harvested, and RT-PCR were run with the extracted total RNA using
XBP1-specific primers. The unspliced band runs about 326 bp and spliced band runs about 304 bp. U unspliced band, S spliced band. Arrows indicate the compounds that induced XBP1 mRNA splicing
was tested in each plate to evaluate the plate-to-plate consistency. As shown in Fig. 2a, the 17-AAG concentration-response curves reproduced well in all 34 plates with an average EC50 of 1.6±0.4 μM. The average signal to background ratio was 2.4±0.2 and Z′ averaged 0.71±0.05 for the entire screen (Fig. 2b). After the primary screening, eight compounds (Table 1) were selected for further study based on their potency (≤10 μM) and novelty. These powder samples were purchased from commercial vendors. Of the eight purchased compounds, the activities of six compounds were confirmed in ESRE-bla HeLa reporter assay, suggesting that those six compounds activate ATF6 pathway following the induction of ER stress. Among the six confirmed compounds, 6-thioguanine was the most potent with EC50 of 3.45 μM, followed by primaquine (EC50 =4.32 μM), roxindole (EC50 = 9.2 μM), AMI-193 (EC50 =14.8 μM), and spiperone (EC50 = 18.6 μM). Cyclosporine A and telmisartan were not confirmed in the follow-up study. These compounds had lower efficacy of 20 and 24 %, respectively, in the primary screening.
46 nM and a maximal fold increase of 3.8 (Fig. 3a). Five out of eight hit compounds tested increased the phosphorylation of eIF2α over 2-fold after 2 h of incubation with cells (Fig. 3b). Immunosuppressant drug, cyclosporine A, reported to induce ER stress and cell death [19, 20], induced concentration-dependent eIF2α phosphorylation with an EC50 value of 595 nM and a maximal fold of increase of 2.4 (Fig. 3b). Two structurally similar anti-psychotic drugs [21], spiperone and AMI-193 (a.k.a., spiramide), increased eIF2α phosphorylation 4.8- and 3.1-fold at the highest concentration (100 μM) tested, respectively (Fig. 3b). Amlodipine and roxindole both induced eIF2α phosphorylation just over 2-fold (Fig. 3b). Amlodipine is a calcium channel blocker and used to treatment for hypertension [22]. Roxindole is an agonist for dopamine receptor and serotonin receptor and used for the treatment for depression [23]. Amlodipine at high concentrations of 33 and 100 μM caused significant amount of cell death. Three other compounds, 6-thioguanine, primaquine, and telmisartan, tested had minimal effect on eIF2α phosphorylation under the same treatment condition (Data not shown).
Effect of active compounds on other branches of the ER stress pathways To further understand the mechanisms of action, we investigated the effects of these compounds on the other components of ER stress response pathways in addition to the ATF6 activation (Table 2). A high-throughput homogenous immunoassay, LanthaScreen® Cellular Assay [16], was used to study the effect of compounds on the phosphorylation of eIF2α at serine 52. In this assay, GripTite™ 293MSR cells stably expressing GFP-eIF2α were plated in a 384-well plate and incubated with DMSO or compounds for 2 h prior to cell lysis, terbiumlabeled anti-eIF2α pSer52 incubation, and TR-FRET detection in the same well. Treatment with control compound tunicamycin resulted in a concentration-dependent increase of eIF2α phosphorylation with an EC50 value of 200 nM and a maximal 2.2-fold increase compared with DMSOtreated cells (Fig. 3a). Despite little activity in the ATF6 reporter cell (ESRE-bla HeLa) assay, bortezomib, a second control compound here induced eIF2α phosphorylation in a concentration-dependent manner with an EC50 value of
Fig. 5 The effect of compound treatment on the GRP78 expression. HeLa cells were treated with indicated compounds at 20 μM for 16 h and monitored the increased expression of GRP78 by RT-PCR analysis. RQ relative quantification. The statistical significance of differences was evaluated by the Student’s t test. *p < 0.05 was considered to be significantly different from the DMSO control
K. Bi et al.
Next, we examined the effect of these compounds on XBP1 messenger (mRNA) splicing mediated by IRE1, the third sensor for ER stress. HeLa cells were treated with the eight compounds or control compound bortezomib for 16 h and harvested for RNA extraction. XBP1 RT-PCR results showed that bortezomib (control), spiperone, cyclosporine A, and AMI-193 induced XBP1 mRNA splicing (Fig. 4). Treatment of cells with 20 μM amlodipine for 16 h caused severe cell death, therefore, little RNA was extracted to generate RT-PCR products. Other compounds tested under the same condition did not induce XBP1 splicing in HeLa cells when administered at 20 μM (Fig. 4). Both ATF6 and XBP1 activation can lead to increased expression of genes for ER chaperones, such as GRP78. To examine the effect of compound treatment on the endogenous GRP78 gene expression, we performed real-time RT-PCR analysis on HeLa cells treated with DMSO or indicated compounds at 20 μM for 16 h. Shown in Fig. 5, all eight compounds resulted in different levels of induction (Fig. 5). Compared with DMSO treatment, amlodipine increased GRP78 expression about 32-fold, followed by AMI-193 17-fold, spiperone 10-fold, 6-thioguanine 6-fold, and cyclosporine A and roxindole about 5-fold, respectively. This data suggests that these six compounds (p
RESEARCH PAPER
Identification of known drugs targeting the endoplasmic reticulum stress response Kun Bi 1 & Kana Nishihara 2,3 & Thomas Machleidt 4 & Spencer Hermanson 4 & Jun Wang 4 & Srilatha Sakamuru 2 & Ruili Huang 2 & Menghang Xia 2
Received: 24 February 2015 / Revised: 3 April 2015 / Accepted: 10 April 2015 # Springer-Verlag Berlin Heidelberg 2015
Abstract The endoplasmic reticulum (ER), a multifunctional organelle, plays a central role in cellular signaling, development, and stress response. Dysregulation of ER homeostasis has been associated with human diseases, such as cancer, inflammation, and diabetes. A broad spectrum of stressful stimuli including hypoxia as well as a variety of pharmacological agents can lead to the ER stress response. In this study, we have developed a stable ER stress reporter cell line that stably expresses a β-lactamase reporter gene under the control of the ER stress response element (ESRE) present in the glucoseregulated protein, 78 kDa (GRP78) gene promoter. This assay has been optimized and miniaturized into a 1536-well plate format. In order to identify clinically used drugs that induce ER stress response, we screened approximately 2800 drugs from the NIH Chemical Genomics Center Pharmaceutical Collection (NPC library) using a quantitative highthroughput screening (qHTS) platform. From this study, we have identified several known ER stress inducers, such as 17AAG (via HSP90 inhibition), as well as several novel ER stress inducers such as AMI-193 and spiperone. The confirmed drugs were further studied for their effects on the phosphorylation of eukaryotic initiation factor 2α (eIF2α), the Xbox-binding protein (XBP1) splicing, and GRP78 gene * Menghang Xia [email protected] 1
Life Science Solutions Group, Thermo Fisher Scientific, Rockford, IL 61105, USA
2
National Center for Advancing Translational Sciences, National Institutes of Health, Bethesda, MD 20892, USA
3
Radiation Genetics, Graduate School of Medicine, Kyoto University, Yoshidakonoe, Sakyo, Kyoto 606-8501, Japan
4
Promega Corporation, Fitchburg, WI 53711, USA
expression. These results suggest that the ER stress inducers identified from the NPC library using the qHTS approach could shed new lights on the potential therapeutic targets of these drugs. Keywords ATF6 . ER stress pathways . NPC library . qHTS . XBP1
Introduction The endoplasmic reticulum (ER) is a multifunctional organelle where protein folding/maturation, lipid biosynthesis, calcium storage, and release occur. Disruption of ER homeostasis which triggers ER stress response has been linked to many human diseases, such as cancer and inflammation [1–3]. ER stress leads to the activation of three signaling pathways through ER membrane protein sensors including protein kinase-like ER kinase (PERK), inositol-requiring enzyme (IRE1α), and transcription factor 6 (ATF6) [4]. In the absence of ER stress inducers, these ER proteins are sequestered by an ER chaperone protein called glucose-regulated protein, 78 kDa (GRP78; also known as binding immunoglobulin protein (BiP)). Upon ER stress, the unfolded/misfolded proteins accumulate in the ER and bind to GRP78, releasing and activating the three ER stress sensor proteins. Activated PERK, a serine/threonine kinase, phosphorylates and inhibits the translation initiation factor eIF2α, which then shuts down general protein translation and activates transcription factor ATF4 leading to the gene expression of ER chaperones and factors involved in glutathione and amino acid biosynthesis and antioxidative stress response. IRE1α protein has both the serine/ threonine kinase domain and the endoribonuclease domain. Once activated, IRE1α gains the endoribonuclease activity and processes an intron from the X-box-binding protein
K. Bi et al.
(XBP1) resulting in the alternate splicing product, which then translates into a XBP1 protein, a transcription factor that activates the expression of genes involved in protein degradation and restoring protein folding. The GRP78-released ATF6 upon ER stress translocates to the Golgi apparatus, where it is cleaved to produce the active ATF6 fragment, which then translocates back to the nucleus to activate the expression of downstream ER stress genes such as GRP78 and GRP94. All these signaling events aim to decrease the accumulation of unfolded proteins in the ER. However, when ER stress is severe and prolonged, it can lead to cell death [5]. Due to its significant involvement in the development of various diseases [6], ER stress pathway and its components have been targeted for drug discovery. Several ER stress inducers for therapeutic intervention have been identified. For example, bortezomib, a proteasome inhibitor that can induce severe ER stress leading to apoptosis, has been clinically applied to treat multiple myeloma patients [7]. Salubrinal, a phosphatase inhibitor that protects cells against ER stress induced apoptosis by agonizing PERK-eIF2α pathway, has the therapeutic potential for treating neurodegenerative diseases [8]. Since many environmental stress signals lead to ER stress response, ER stress induced apoptosis could also be the mechanism for the therapeutic effect of other cancer drugs such as Sorafenib [9]. To investigate whether some of the clinically used drugs can induce ER stress response and to identify novel ER stress inducers, we have developed a high-throughput compatible ATF6 reporter gene assay and screened the NIH Chemical Genomics Center Pharmaceutical Collection (NPC library) [10] consisting of 2816 small molecule drugs approved for human or animal use in the USA, Canada, Japan, and Europe and compounds registered for testing in clinical trials. Each compound was tested at 15 different concentrations in a quantitative high-throughput screening (qHTS) format [11]. We then performed follow-up studies to determine the effects of the confirmed hit compounds on eIF2α phophosphorylation, XBP1 alternate splicing, and ER chaperone GRP78 expression. Several known ER stress inducers, such as 17-AAG and cyclosporine A were identified, as well as several novel ER stress inducers such as AMI-193 and spiperone.
Materials and methods Generation of ESRE-bla HeLa reporter gene cell line HeLa cells were transduced with lenti-virus created from the ViraPower™ Lenti-viral Expression System (Life Technologies, Carlsbad, CA) pLenti6 ESRE-bsd/bla containing the −130 to −40 regions of the GRP78 promoter and selected for blasticidin resistance. The blasticidin-resistant pools were stimulated with tunicamycin for 4 h. Responding cells were
collected into single-cell clones onto 96-well plates using FACS. The clones that displayed the highest fold signal change with tunicamycin compared with a dimethylsulfoxide (DMSO) control were selected and expanded, and the best responding clone was chosen for assay development. ER stress beta-lactamase reporter gene assay and RNAi experiment The ESRE-bla HeLa cells were cultured in Dulbecco’s modified Eagle’s medium (DMEM) containing 10 % dialyzed fetal bovine serum (FBS), 0.1 mM nonessential amino acids (NEAA), 1 mM sodium pyruvate, 25 mM 4-(2hydroxyethyl)-1-piperazineethanesulfonic acid (HEPES), 100 U/mL penicillin, and 100 μg/mL streptomycin (All cell culture reagents are from Life Technologies). Assays were performed in an assay medium containing phenol-red-free OPTI-MEM, 0.5 % dialyzed FBS, 0.1 mM NEAA, 1 mM sodium pyruvate, 100 U/mL penicillin, and 100 μg/mL streptomycin. Cells were incubated with indicated ER stressinducing agents for 5 to 6 h. β-lactamase substrate LiveBLAzer™ FRET B/G detection mix (Life Technologies) was added to the cells and incubated at room temperature for 2 h. Fluorescence intensity at 406 nm excitation and 460 and 530 nm emission was measured using TECAN Safire2 (Tecan Group Ltd, Maennedorf, Switzerland) with optimal gain settings determined by the instrument. After subtracting fluorescence background from cell-free wells, the ratio of fluorescence intensity at 460 vs 530 nm (designated as 460/ 530 nm) was calculated. Response ratio is a measurement of the assay window and is calculated as the 460/530-nm emission ratio of the stimulated wells divided by the 460/530-nm emission ratio of the unstimulated wells. Response ratios were plotted against test ligand concentration in log scale and then analyzed using Prism software (GraphPad Software, Inc. San Diego, CA). Sigmoidal concentration-response equation with varying slope was used to fit the data and generate concentrations of half-maximal stimulation (EC50) value. Z′-factor values were calculated as: Z′-factor=1−((3×stdevunstim +3× stdevmaxstim)/(avgmaxstim −avgunstim)). Assays were performed in quadruplicates. For RNAi experiments, ESRE-bla HeLa cells were plated in antibiotic-free growth medium in 96-well assay format at 7500 cells/well and transfected using Lipofectamine™ RNAiMAX and 20 nM of the following panel of RNAi duplexes/control conditions: mock transfected (RNAiMAX reagent only, no RNAi duplex), Stealth™ RNAi Negative Control Med GC, beta-lactamase RNAi-positive control duplex (sense strand sequence, GCGCAAACUAUUAACUGG CGAACUA), ATF4#1 (GCAGCCACUAGGUACCGCCA GAAGA), ATF4#2 (UAUCAAAUCUUUCAGGUACUGG AUC), ATF6#1 (CCCUAGUGUGGGACCUGCAAAU CAA), ATF6#2 (CCGUAUUCUUCAGGGUGUUGUG
Profiling of drugs targeting the endoplasmic reticulum stress pathway
GAA). At 60 h posttransfection, cells were stimulated with tunicamycin and the beta-lactamase reporter gene assay was performed and data was analyzed as described above.
same protocol as described above except 24 point titrations were within one 1536-well plate. Data analysis
NCGC pharmaceutical collection and compounds used in the study The NIH Chemical Genomics Center Pharmaceutical Collection consists of 2816 small molecule compounds which include drugs approved for human or animal use in the USA, Canada, Japan, and Europe and compounds registered for testing in clinical trials. The compound plate preparation was described elsewhere [10, 11]. Briefly, all the compounds were dissolved in dimethylsulfoxide (DMSO), prepared in a 1536well plate format, and stored desiccated at room temperature for as long as 6 months when in use or heat sealed and stored at −80 °C for long-term storage. For follow-up studies, primaquine, spiperone, cyclosporine A, 6-thioguanine, amlodipine, and telmisartan, were purchased from Sigma-Aldrich (St. Louis, MO). AMI-193 and roxindole were purchased from Tocris Bioscience (Ellisville, MO). Bortezomib was purchased from LC laboratories (Boston, MA). Quantitative high-throughput screening The ESRE-bla HeLa cells were suspended in assay medium. The cells were dispensed at 1500 cells/6 μL/well in 1536-well black wall/clear bottom plates (Greiner Bio-One North America, NC) using a Flying Reagent Dispenser (FRD; Aurora Discovery, CA) or Thermo Scientific Multidrop Combi (Thermo Fisher Scientific Inc., Waltham, MA). After the cells were incubated at 37 °C overnight, 23 nL of compounds at 15 concentrations from the NPC library was transferred to the assay plate by a pin tool (Kalypsys, San Diego, CA). The final concentration of the compounds in the 6-μL assay volume ranged from 0.5 nM to 38 μM. The positive control plate format was as follows: Column 1, 17-AAG, ranged from 23 pM to 7.6 μM; column 2, 3.8 μM 17-AAG (EC100); column 3, DMSO only; and column 4, 1.9 μM 17-AAG; columns 5 to 48, compounds from NPC library. The plates were incubated for 6 h at 37 °C. One microliter of LiveBLAzer™ B/G FRET substrate (Life Technologies) detection mix was added and the plates incubated at room temperature for 2 h. Fluorescence intensity (405 nm excitation, 460 and 530 nm emissions) was measured using an EnVision plate reader (Perkin Elmer, Shelton, CT). Data was expressed as the ratio of 460/530 nm emissions (without background subtraction). Use of ratio metric readout decreases the well-to-well variation of the cell numbers in the assay plate [12]. In the confirmation study, selected active compounds were re-tested in 24 point titrations with concentration ranging from 5 pM to 38 μM in the ESRE-bla HeLa reporter assay using the
Data normalization and curve fitting was performed as previously described [13]. Briefly, raw plate reads for each titration point were first normalized relative to 17-AAG control (3.8 μM, 100 %) and DMSO only wells (basal, 0 %), and then corrected by applying a pattern correction algorithm using compound-free control plates (DMSO plates). Concentration-response titration points for each compound were fitted to the Hill equation yielding EC50 and maximal response (efficacy) values. Compounds from qHTS were classified into four major curve classes using the set of criteria listed in previous studies [14, 15]. From the primary screen, eight compounds were selected and purchased from commercial vendors for further extensive studies based on potency and quality of concentration-response curve and novelty of mechanism. LanthaScreen® cellular eIF2α phosphorylation assay GripTite™ 293 MSR cells stably expressing GFP-eIF2α (Life Technologies) were plated onto a white tissue culture treated 384-well assay plate (Corning Incorporated, Tewksbury, MA) in growth medium (DMEM with GlutaMax, 10 % dialyzed FBS, 25 mM HEPES, 0.1 mM NEAA, 100 U/mL penicillin, and 100 μg/mL streptomycin). Cells were left untreated or treated with indicated ER stress agents (such as tunicamycin) or hit compounds for 2 h before cell lysis and assay equilibration with complete lysis buffer including 2 nM terbiumlabeled anti-eIF2α pSer52 antibody (Life Technologies). TR-FRET emission ratios were determined on a BMG PHERAstar fluorescence plate reader (BMG LABTECH, Cary, NC; excitation at 340 nm, emission 520 and 490 nm as described previously [16]. RT-PCR and XBP1 alternate splicing ESRE-bla HeLa cells were left untreated or treated with 20 μM indicated hit compounds for 16 h. Cells were then pelleted and total RNA was isolated using Ambion® RiboPure™ RNA Purification Kit (Life Technologies) and converted to complementary DNA (cDNA) using SuperScript®III reverse transcriptase (Life Technologies). The cDNA for XBP1 was then amplified using polymerase chain reaction using Platinum® Taq DNA Polymerase (Life Technologies). The nucleotide sequences of the XBP1 primer pairs (5′ to 3′) were TTACGAGAGAAAACTCATGGCC and GGGTCCAAGTTGTCCAGAATGC. PCR products were separated by electrophoresis on 1.2 % agarose gels and visualized with SYBR® Safe DNA gel stain (Life Technologies).
K. Bi et al.
Fig. 1 CellSensor® ESRE-bla HeLa cell line validation. a ER stress response to known ER stress inducers in ESRE-bla HeLa cell. A23187 (EC50 =17.8 nM), bortezomib (EC50 =40.0 nM), ionomycin (EC50 = 44.8 nM), thapsigargin (EC 50 = 5.6 nM), tunicamycin (EC 50 = 301.7 nM), and 17-AAG (EC50 = 397.6 nM). b Validation of the involvement of ATF6 in the response of the ESRE-bla HeLa cell. The 460/530-nm emission ratio was plotted in y-axis for each RNAi oligo.
Med-GC and β-lactamase are the controls. c Concentration-response curve of 17-AAG (EC50 =590 nM) in 1536-well plate. Error bars represent SD from at least three independent experiments. The statistical significance of differences was evaluated by the Student’s t test. *p100
Yes Yes Yes Yes No NA No No No
NT 9.5 4.4 17 6.1 32 2.3 2.1 4.7
NA not available NT not tested
robust response to a number of other known ER stressinducing compounds including calcium ionophore A23187 (EC50 =17.8 nM) and ionomycin (EC50 =44.8 nM) which deplete calcium in the ER, thapsigargin (EC50 =5.6 nM) which inhibits the ER calcium ATPase and decreases calcium ions from the ER and 17-AAG (EC50 =397.6 nM), a HSP90 inhibitor that increases unfolded proteins in the ER [17]. In addition to the EC50 value differences, these compounds generated different levels of response in this reporter line. Tunicamycin and 17-AAG generated the highest response ratio of 5, followed by thapsigargin with a response ratio of 4 and A23187 and ionomycin of 3 (Fig. 1a). Interestingly, this reporter line shows little response to bortezomib, a proteasomal inhibitor that causes protein aggregation in the ER and induces severe ER stress response leading to apoptosis [18]. To validate the involvement of ATF6 in the response of the ESRE-bla HeLa cells to tunicamycin, ATF6-specific Stealth™ RNAi oligos were transfected into the reporter cells prior to the stimulation with tunicamycin. Both ATF6 RNAi oligos significantly knocked down the ER stress response (p100 μM), and spiperone (EC50 =85.4 μM)
DM SO ,c on tro DM l SO ,c on tro Te l lm isa rta n Sp i pe ro n e Cy clo sp o ri ne 6 -T A hi o gu an i ne Pr im aq ui n e Bo rte zo mi b Am l od i pi ne AM I-1 93 Ro xin do le
(bp) 400 300
Ma rke r
Profiling of drugs targeting the endoplasmic reticulum stress pathway
U S
200
Fig. 4 XBP1 mRNA splicing monitored by RT-PCR. HeLa cells were treated with indicated compounds at 20 μM for 16 h. Cells were harvested, and RT-PCR were run with the extracted total RNA using
XBP1-specific primers. The unspliced band runs about 326 bp and spliced band runs about 304 bp. U unspliced band, S spliced band. Arrows indicate the compounds that induced XBP1 mRNA splicing
was tested in each plate to evaluate the plate-to-plate consistency. As shown in Fig. 2a, the 17-AAG concentration-response curves reproduced well in all 34 plates with an average EC50 of 1.6±0.4 μM. The average signal to background ratio was 2.4±0.2 and Z′ averaged 0.71±0.05 for the entire screen (Fig. 2b). After the primary screening, eight compounds (Table 1) were selected for further study based on their potency (≤10 μM) and novelty. These powder samples were purchased from commercial vendors. Of the eight purchased compounds, the activities of six compounds were confirmed in ESRE-bla HeLa reporter assay, suggesting that those six compounds activate ATF6 pathway following the induction of ER stress. Among the six confirmed compounds, 6-thioguanine was the most potent with EC50 of 3.45 μM, followed by primaquine (EC50 =4.32 μM), roxindole (EC50 = 9.2 μM), AMI-193 (EC50 =14.8 μM), and spiperone (EC50 = 18.6 μM). Cyclosporine A and telmisartan were not confirmed in the follow-up study. These compounds had lower efficacy of 20 and 24 %, respectively, in the primary screening.
46 nM and a maximal fold increase of 3.8 (Fig. 3a). Five out of eight hit compounds tested increased the phosphorylation of eIF2α over 2-fold after 2 h of incubation with cells (Fig. 3b). Immunosuppressant drug, cyclosporine A, reported to induce ER stress and cell death [19, 20], induced concentration-dependent eIF2α phosphorylation with an EC50 value of 595 nM and a maximal fold of increase of 2.4 (Fig. 3b). Two structurally similar anti-psychotic drugs [21], spiperone and AMI-193 (a.k.a., spiramide), increased eIF2α phosphorylation 4.8- and 3.1-fold at the highest concentration (100 μM) tested, respectively (Fig. 3b). Amlodipine and roxindole both induced eIF2α phosphorylation just over 2-fold (Fig. 3b). Amlodipine is a calcium channel blocker and used to treatment for hypertension [22]. Roxindole is an agonist for dopamine receptor and serotonin receptor and used for the treatment for depression [23]. Amlodipine at high concentrations of 33 and 100 μM caused significant amount of cell death. Three other compounds, 6-thioguanine, primaquine, and telmisartan, tested had minimal effect on eIF2α phosphorylation under the same treatment condition (Data not shown).
Effect of active compounds on other branches of the ER stress pathways To further understand the mechanisms of action, we investigated the effects of these compounds on the other components of ER stress response pathways in addition to the ATF6 activation (Table 2). A high-throughput homogenous immunoassay, LanthaScreen® Cellular Assay [16], was used to study the effect of compounds on the phosphorylation of eIF2α at serine 52. In this assay, GripTite™ 293MSR cells stably expressing GFP-eIF2α were plated in a 384-well plate and incubated with DMSO or compounds for 2 h prior to cell lysis, terbiumlabeled anti-eIF2α pSer52 incubation, and TR-FRET detection in the same well. Treatment with control compound tunicamycin resulted in a concentration-dependent increase of eIF2α phosphorylation with an EC50 value of 200 nM and a maximal 2.2-fold increase compared with DMSOtreated cells (Fig. 3a). Despite little activity in the ATF6 reporter cell (ESRE-bla HeLa) assay, bortezomib, a second control compound here induced eIF2α phosphorylation in a concentration-dependent manner with an EC50 value of
Fig. 5 The effect of compound treatment on the GRP78 expression. HeLa cells were treated with indicated compounds at 20 μM for 16 h and monitored the increased expression of GRP78 by RT-PCR analysis. RQ relative quantification. The statistical significance of differences was evaluated by the Student’s t test. *p < 0.05 was considered to be significantly different from the DMSO control
K. Bi et al.
Next, we examined the effect of these compounds on XBP1 messenger (mRNA) splicing mediated by IRE1, the third sensor for ER stress. HeLa cells were treated with the eight compounds or control compound bortezomib for 16 h and harvested for RNA extraction. XBP1 RT-PCR results showed that bortezomib (control), spiperone, cyclosporine A, and AMI-193 induced XBP1 mRNA splicing (Fig. 4). Treatment of cells with 20 μM amlodipine for 16 h caused severe cell death, therefore, little RNA was extracted to generate RT-PCR products. Other compounds tested under the same condition did not induce XBP1 splicing in HeLa cells when administered at 20 μM (Fig. 4). Both ATF6 and XBP1 activation can lead to increased expression of genes for ER chaperones, such as GRP78. To examine the effect of compound treatment on the endogenous GRP78 gene expression, we performed real-time RT-PCR analysis on HeLa cells treated with DMSO or indicated compounds at 20 μM for 16 h. Shown in Fig. 5, all eight compounds resulted in different levels of induction (Fig. 5). Compared with DMSO treatment, amlodipine increased GRP78 expression about 32-fold, followed by AMI-193 17-fold, spiperone 10-fold, 6-thioguanine 6-fold, and cyclosporine A and roxindole about 5-fold, respectively. This data suggests that these six compounds (p
