Development of a Cell-Based High Throughput Luciferase Enzyme Fragment Complementation Assay to Identify Nuclear-Factor-E2-Related Transcription Factor 2 Activators

Abstract

Nuclear-factor-E2-related transcription factor 2 (Nrf2) regulates a large panel of Phase II genes and plays an important role in cell survival. Nrf2 activation has been shown as preventing cigarette smoke-induced alveolar enlargement in mice. Therefore, activation of the Nrf2 protein by small-molecule activators represents an attractive therapeutic strategy that is used for chronic obstructive pulmonary disease. In this article, we describe a cell-based luciferase enzyme fragment complementation assay that identifies Nrf2 activators. This assay is based on the interaction of Nrf2 with its nuclear partner MafK or runt-related transcription factor 2 (RunX2) and is dependent on the reconstitution of a “split” luciferase. Firefly luciferase is split into two fragments, which are genetically fused to Nrf2 and MafK or RunX2, respectively. BacMam technology was used to deliver the fusion constructs into cells for expression of the tagged proteins. When the BacMam-transduced cells were treated with Nrf2 activators, the Nrf2 protein was stabilized and translocated into the nucleus where it interacted with MafK or RunX2. The interaction of Nrf2 and MafK or RunX2 brought together the two luciferase fragments that form an active luciferase. The assay was developed in a 384-well format and was optimized by titrating the BacMam concentration, transduction time, cell density, and fetal bovine serum concentration. It was further validated with known Nrf2 activators. Our data show that this assay is robust, sensitive, and amenable to high throughput screening of a large compound collection for the identification of novel Nrf2 activators.

Introduction

Cells are constantly exposed to many types of stress that are generated internally or externally. In a normally functional and healthy mammalian cell system, complex defensive mechanisms are developed and tightly regulated to counteract the toxicants and carcinogens to which cells are exposed.¹ Many types of diseases, such as chronic obstructive pulmonary disease (COPD), cancer, and inflammation, are related to the dysfunction of these protective systems.² The antioxidant response induced by the nuclear-factor-E2-related factor 2 (Nrf2) is one of the major pathways that protects cells from oxidative damage.^1,3
–5 This system is tightly regulated through Nrf2's negative regulator Kelch-like ECH-associated protein 1 (Keap1) and the ubiquitin E3 ligase Cullin3.^6

–9 Under normal conditions, Nrf2 is located in the cytoplasm and bound to Keap1,^10,11 which, in turn, promotes the ubiquitination of Nrf2 and the degradation of the Nrf2 protein.⁷ Under stressed conditions or on induction by exogenous Nrf2 activators, the Cul3/Keap1/Nrf2 complex is disrupted, and the ubiquitination of Nrf2 is impaired. Consequently, the Nrf2 protein is stabilized and is allowed to translocate into the nucleus where it interacts with nuclear factors such as MafK or runt-related transcription factor 2 (RunX2),^12,13 to form heterodimers.¹⁴ Within the nucleus, the Nrf2/MafK heterodimer, together with other partners such as CBP/p300, binds to the antioxidant responsive element (ARE) and induces Phase II gene expression.² It is well established that regulation of Nrf2/Keap1 signaling pathway is critical for cell protection from oxidative stress, while its deregulation is linked with several diseases. Subsequently, Nrf2 activators are considered a potential novel therapeutic approach for several diseases, such as COPD and neurodegenerative disease.^15,16 An assay that is sensitive, robust, and high throughput is required for full diversity screening of a large compound collection for Nrf2 activators. Here, we describe an enzyme fragment complementation (EFC) method that is developed to check Nrf2 activation by monitoring the interaction of Nrf2/MafK and Nrf2/RunX2.

EFC assays are based on the ability to express an enzyme as two separate fragments whose enzymatic activity can be reconstituted by bringing the fragments into close proximity. In these assays, two separate enzyme fragments are genetically fused to two different protein-binding partners. The interaction between the two binding partners, therefore, brings together the two enzyme fragments that form an active reporter enzyme. This approach has been exploited to characterize protein-protein interactions.^17,18 Common reporter genes or enzymes that have successfully been used in EFC assays in mammalian cells include β-galactosidase,^19,20 β-lactamase,²¹ green fluorescent protein, and luciferase.^22,23 In the current study, protein pairs of Nrf2/MafK and Nrf2/RunX2 were tagged with amino and carboxyl luciferase fragments (e.g., Nrf2-Luc2N and Luc2C-MafK). The cDNAs encoding these paired fusion proteins were co-transduced into cells via BacMam technology.²⁴ On treatment with Nrf2 activators, the Nrf2-Luc2N protein would translocate into the nucleus where it would interact with Luc2C-MafK or Luc2C-RunX2. As illustrated in Figure 1, the binding of Nrf2 and MafK brings together Luc2N and Luc2C to form active luciferase. Our results show that luciferase activity is dependent on the co-expression of two fused luciferase fragments within the cells. We optimized various aspects of the experimental protocol in order to maximize the sensitivity of the assay. In addition, we used several known small-molecule Nrf2 activator compounds to validate the pharmacology of the assay.

Fig. 1.

Schematic illustration of EFC analysis for Nrf2 activator-induced luciferase activity. Under basal conditions, Keap1 binds to Nrf2 and targets it for ubiquitylation and proteasomal degradation via the Cul3-based E3-ubiquitin ligase complex. On treatment with Nrf2 activators (goal of the screen), Nrf2-Luc2N would translocate into the nucleus where it would interact with Luc2C-MafK (or Luc2C-RunX2, not shown) and form active luciferase. EFC, enzyme fragment complementation; Keap1, Kelch-like ECH-associated protein 1; Nrf2, nuclear factor E2 related transcription factor 2; RunX2, runt-related transcription factor 2.

Materials And Methods

Reagents

Dimethyl sulfoxide (DMSO) was purchased from Sigma (St Louis, MO). Steady Glo luciferase activity detection kit was purchased from Promega (Madison, WI). DMEM/F12 cell culture medium was prepared in-house from CellGro powders. Fetal bovine serum (FBS) was purchased from Life Technology (Grand Island, NY). 2-HBA [(1E, 4E)-1,5-bis (2-hydroxyphenyl)-1,4-pentadien-3-one], the 2-HBA analog [(2E, 5E)-2,5-bis[(2-hydroxyphenyl)methylidene]cyclopentanone], and GSK compound was synthesized in-house. Restriction enzymes were purchased from New England Biolabs (Ipswich, MA). L-sulforaphane and tert-butylhydroquinone (tBHQ) were from Sigma Aldrich (Saint Louis, MO). Prostaglandin J2, 15-deoxy-delta 12 14 (15d-PGJ2) was from EMD Chemicals (Gibbstown, NJ).

HEK-293 MSRII cells (HEK-293 cells expressing the macrophage scavenger receptor, which facilitates the adherence of the cells to a plastic substrate) were cultured in DMEM/F12 supplemented with 10% FBS. The cells were allowed to grow ∼70% confluence, trypsinized, washed, and frozen in a solution of 90% FBS, 10% DMSO. Freshly thawed HEK-293 MSRII cells were used for transduction, and treatment was performed as described next.

BacMam Vector and EFC Constructs Generation

BacMam vectors containing a cytomegalovirus promoter were generated as previously described.²⁴ Two modified BacMam vectors were used. The Luc2N contains a multiple cloning site (BamHI, NruI, MluI, Acc651, KpnI, XhoI, and NheI) followed by a flexible Gly/Ser linker and residues 2-416 of firefly luciferase. The Luc2C contains amino acids 398–550 of firefly luciferase followed by a flexible Gly/Ser linker and a multiple cloning site (BamHI, NruI, PspOMI, ApaI, XhoI, NheI, and BstEII) (Fig. 2). For the proteins described next, Luc2C is tagged to the N terminus, while Luc2N is tagged to the C terminus of the target protein.

Fig. 2.

Scheme of BacMam vectors illustrating the cloning sites and firefly luciferase fragment positions. (A) Depicts the Luc2N vector where the gene of interest is cloned into the multiple cloning site without the addition of a stop codon, followed by a glycine-serine linker and then the first 416 amino acids of firefly luciferase (Luc2N) with a stop codon inserted after residue 416. (B) Illustrates the Luc2C vector where the carboxy terminus of luciferase (amino acids 398–550) is followed by a glycine-serine linker, then the multiple cloning site. When cloning into this vector, a stop codon should be inserted after the gene of interest. Please note that “N” and “C” refer to the part of the luciferase gene that is expressed, instead of the N or C term of the target gene.

To prepare the fusion constructs, the polymerase chain reaction products of full-length Nrf2, MafK, and RunX2 were digested with BamHI and XhoI, purified, and inserted in-frame into BamHI/XhoI-digested Luc2N or Luc2C vectors. The nucleotide sequences of all the final constructs were verified by sequence analysis. BacMam viruses for these constructs were generated as previously described.²⁴ Briefly, bacmid DNA was prepared from DH10Bac competent cells, which were transformed with plasmid DNA. Sf9 cells were then transfected with bacmid DNA to generate P0 virus in the supernatant. The P1 virus was generated by a further amplification of the P0 virus. The virus was titered using the Baculotiter Assay Kit from Invitrogen (Carlsbad, CA) according to the manufacturer's recommendations. The virus solutions were stored at 4°C in the dark.

The following six BacMam viruses were used for this study (with titer values): Luc2C tagged Nrf2 (5.95×10⁸ pfu/mL), Luc2N tagged Nrf2 (2.27×10⁸ pfu/mL), Luc2C tagged MafK (1.81×10⁸ pfu/mL), Luc2N tagged MafK (2.19×10⁸ pfu/mL), Luc2C tagged RunX2 (2.33×10⁸ pfu/mL), and Luc2N tagged RunX2 (2.39×10⁸ pfu/mL). These constructs formed four EFC pairs: (1) Luc2C-Nrf2 with MafK-Luc2N; (2) Nrf2-Luc2N with Luc2C-MafK; (3) Luc2C-Nrf2 with RunX2-Luc2N; and (4) Nrf2-Luc2N with Luc2C-RunX2. As described next, Luc2C and Luc2N paired BacMam viruses were used for the assay development.

Cell Transduction with BacMam Viruses and Compound Treatment

BacMam transductions were performed on HEK-293 MSRII cells as previously described.²⁵ Briefly, frozen cells were thawed and washed with DMEM/F12 (no Phenol red). The cells were diluted to the desired cell density in the same medium. BacMam pairs were added to the cells at the desired concentration and ratio. The cells that had been mixed with BacMam viruses were subsequently dispensed into 384-well plates which contained pre-dispensed compounds (see next). The total volume was 40 μL/well. The plates were incubated at 37°C with 5% CO₂ at varying time points as described in the results section. The luciferase activity was measured as described next.

When co-titrating the BacMams of each pair of the EFC constructs, we used volume ratios (v/v) instead of virus particles per cell (multiplicity of transduction [MOT]) for simplicity. The actual MOT can be calculated from the virus titers and cell density provided.

Measurement of Reconstituted Luciferase Activity

Luciferase activity was measured with the Steady Glo luciferase assay kit and was used according to the manufacturer's instructions. Cell plates were allowed to equilibrate to room temperature for ∼20 min after removal from the 37°C incubator. Then, 20 μL/well of luciferase substrate solution was dispensed to the cell plates using a small Combi dispenser cartridge. The plates were incubated at room temperature for 20 min and read with a ViewLux plate reader using a luminescence protocol (clear filter, 30 s exposure with a luminescence label).

Compound Addition and pEC₅₀ Measurement on Automated Platforms

Stock concentrations of compounds were prepared at 10 mM in neat DMSO. The HP digital dispenser (Hewlett Packard, Corvallis, Oregon) was used to dispense compound stocks directly into the cell plates. For dose-response experiments, the compounds were prepared in 384-well plates following a threefold dilution scheme with DMSO starting from 10 mM, and then, 200 nL of compound solutions were transferred to the assay plate using an Echo Dispenser (Labcyte, Sunnyvale, CA). This gave 11 doses for each compound with a final concentration starting from 50 μM. For a single-concentration screen, 400 nL of the compound at 1 mM stock was added instead to give a final concentration of 10 μM. To generate controls for dose-response curves, 400 nL of DMSO was added to column 6 as the low control, while 400 nL of an in-house compound at 1.5 mM was added to column 18 as the high control. The assay plates were incubated for 22 h at 37°C with 5% CO₂. The luciferase signal was then measured as just described.

Data Processing and Analysis

The potency of compounds was expressed as pEC₅₀ and calculated as follows: The direct luminescence signal or normalized percentage of activation by DMSO (low control) and the high control was used for pEC50 calculation with a GraphPad Prism nonlinear regression activator curve fitting (4-parameter logistic fit equation, www.graphpad.com). As an indicator of assay robustness, the Z′ value was calculated as given next using the luminescence signal: \documentclass{aastex}\usepackage{amsbsy}\usepackage{amsfonts}\usepackage{amssymb}\usepackage{bm}\usepackage{mathrsfs}\usepackage{pifont}\usepackage{stmaryrd}\usepackage{textcomp}\usepackage{portland,xspace}\usepackage{amsmath,amsxtra}\pagestyle{empty}\DeclareMathSizes{10}{9}{7}{6} \begin{document} \begin{align*}Z^\prime = 1-\frac {3 ( \hbox{St. dev. of low control} + \hbox {St. dev. of high control)}} {\hbox{Avg. of high control} - \hbox {Avg. of low control}} \end{align*} \end{document}

Results

Detection of Nrf2 and MafK Interaction Through EFC

When Nrf2 translocates to the nucleus, it interacts with the DNA binding protein MafK and the Nrf2-MafK complex, consequently activating Phase II gene expression. EFC technology that directly measures the Nrf2-MafK interaction provides an easy method for monitoring Nrf2 protein activity. BacMams were generated to express two pairs of fusion constructs for this interaction: Luc2C-Nrf2 and MafK-Luc2N; Nrf2-Luc2N and Luc2C-MafK. Each BacMam pair was transduced into HEK-293 MSRII cells at different concentrations as indicated next. Theoretically, on the treatment with Nrf2 activators, the Nrf2 protein would be allowed to accumulate and translocate into the nucleus where increased local concentrations of the tagged Nrf2 protein would enhance its interaction with tagged MafK, resulting in higher luciferase activity. Hence, when developed with a luciferase substrate, the intensity of luminescence would reflect the increased Nrf2 protein level within the nucleus. We carried out the following experiments to optimize the assay.

Titration of total BacMam concentration and titration of the ratio of the two components

BacMam technology^24
–26 was used to deliver the EFC fusion constructs into cells. It is assumed that the total BacMam virus concentration of the two components will determine the protein level and, consequently, the assay signal. In addition, since the reaction involves two proteins, the ratio between the two components will also affect the assay signal. Hence, we performed a grid titration experiment where the total BacMam concentration as well as the ratio of the two components was varied. The total concentration of Luc2C-Nrf2 and MafK-Luc2N BacMams was varied between 0.25%, 0.5%, 1%, and 2%, v/v. At each concentration, the volume ratio of tagged Nrf2/tagged MafK was varied from 100%/0% to 0%/100% in small increments. Figure 3 summarizes the signal-to-background ratio (S/B) under these conditions on treatment with a 2-HBA analog (a phase 2 gene inducer²⁷) at 9.5 μM. The background signal was obtained by treating each sample with DMSO only. For both pairs, the luciferase signal increased in proportion to the total BacMam concentrations to which the cells were exposed (Data not shown). However, higher total BacMam concentrations had higher background luminescent signals (DMSO treatment), resulting in lower S/B ratios. This background problem was likely caused by the over-expression of both components at higher BacMam concentrations. As shown in Figure 3, of the concentrations tested, 0.5% and 0.25% of total BacMam concentrations had the highest S/B ratio, whereas 2% of total BacMam had a much lower S/B ratio. Four percent of total BacMam concentration showed no induction at all for either pair (data not shown).

Fig. 3.

Titration of total BacMam concentration and the ratio of the paired two BacMams (A) Luc2C-Nrf2 and MafK-Luc2N EFC pair (B) Nrf2-Luc2N and Luc2C-MafK EFC pair. HEK-293 MSRII cells, 10,000 cells/well/40 μL, were co-transduced with the paired BacMams at varying concentrations and ratios. The ratio of tagged Nrf2/tagged MafK was expressed as (v/v). Transduced cells were treated with 9.5 μM 2-HBA analog for 26 h at 37°C with 5% CO₂. Luciferase activity was measured. The data presented are means±SEM of duplicate wells. 2-HBA analog, (2E, 5E)-2,5-bis[(2-hydroxyphenyl)methylidene]cyclopentanone; SEM, standard error of the mean.

Figure 3 also clearly shows that the cells transduced with tagged Nrf2 BacMam or tagged MafK BacMam alone displayed no luciferase activity (Fig. 3A, B; 100%/0% and 0%/100% samples). In contrast, the luminescence signal dramatically increased in the presence of both components on compound treatment. The fact that luciferase activity is only induced in the presence of two components strongly suggests that the luciferase signal is dependent on the reconstituted luciferase activity generated by the complementation of the two fusion proteins. The ratio of the two BacMams also affected the S/B ratio. For instance, for the Luc2C-Nrf2/MafK-Luc2N pair, a total concentration of 0.5% BacMam gave the highest S/B ratio with the volume ratio of 20%/80%. This volume ratio corresponds to an MOT of 2.4/3.4. Similarly, a total concentration of 0.5% BacMam demonstrated the highest S/B with a volume ratio of 40%/60% (v/v) for Nrf2-Luc2N/Luc2C-MafK. This corresponds to an MOT of 1.8/2.2. Since the S/B ratio was not dramatically different between 0.25% and 0.5% total BacMam, but the luminescence signal was higher with 0.5% total BacMam, we chose the latter condition for further optimization. At a 0.5% total BacMam concentration, the ratio of tagged Nrf2/tagged MafK at 40%/60% (v/v) showed a good response under most of the tested conditions for both pairs. We, therefore, chose this ratio for further optimization. In parallel, a few known Nrf2 activators, including 2-HBA, L-sulforaphane, tBHQ, and 15d-PGJ2, were also tested in the experiment. Consistent data were obtained with these tool compounds. The final assay conditions were chosen primarily based on the response of the 2-HBA analog, although the responses of other tool compounds were also considered.

Titration of serum concentration in assay cell-culture medium

Serum concentration affects multiple cellular processes, such as transduction efficiency and protein expression, and the presence of serum proteins can affect compound potency.²⁸ Therefore, it is very important to optimize the assay serum concentration to maximize sensitivity and response. For the Luc2C-Nrf2/MafK-Luc2N pair, FBS concentration in the cell culture medium was titrated from 0% to 10%. Figure 4A shows the dose response of cells cultured at varying concentrations of FBS (0%–10%) after stimulation with varying concentrations of the 2-HBA analog. The signal induction was much higher when cells were cultured in the presence of 1.25% and 2.5% FBS than in the presence of any other FBS concentrations. Cells cultured in 10% FBS exhibited the lowest signal of any of the serum concentrations tested. This is possibly due to nonspecific binding of the 2-HBA analog to serum proteins, resulting in a lower effective concentration of the compound and, therefore, a lower induction of Nrf2 protein expression. The conditions with 0% and 5% serum had moderate signal induction. Since there was a minimal difference in the signals obtained from the cells cultured in 1.25% versus 2.5% FBS, we chose a concentration of 2% FBS in the cell-culture medium for further assay optimization in order to ensure cell health as well as to reduce the nonspecific protein binding of compounds.

Fig. 4.

Optimization of serum concentration and cell density. Total BacMam concentration was maintained at 0.5%, and the ratio of Luc2C-Nrf2 and MafK-Luc2N was at 40%/60% (v/v). Cells were treated with varying doses of compound for 26 h at 37°C 5% CO₂. (A) Titration of serum concentration. HEK-293 MSRII cells were resuspended in DMEM/F12 medium supplemented with a varying FBS concentration. Twenty thousand cells/well/40 μL was used for transduction and treatment. (B) Titration of cell density. HEK-293 MSRII cells were resuspended in DMEM/F12 medium supplemented with 2% FBS. Varied cell numbers were used for transduction and treatment. The data presented are means±SEM of duplicate wells. DMEM, Dulbecco's modified Eagle medium; FBS, fetal bovine serum.

Cell density titration

Using the optimized conditions just mentioned for the BacMam and serum concentrations within the assay, we next titrated the cell density per well to gauge the effect on compound potency and S/B ratio. In this experiment, a regular Greiner 384-well plate (Greiner 781075) was used. In order to prevent potential evaporation and save reagents, 40 μL/well culture volume was used. Again, using the pair of Luc2C-Nrf2/MafK-Luc2N, we tested varying concentrations of the 2-HBA analog versus different cell densities ranging from 5,000 to 25,000 cells/well. As shown in Figure 4B, 5000 cells/well had very low signals, possibly due to the low levels of protein and low tolerance for compound cytotoxicity. Higher cell densities (10,000/well to 20,000/well) had a much higher S/B ratio. Data show a trend of higher signal induction correlating from 5,000 to 20,000 cells/well. The highest S/B ratio was obtained with 20,000 cells/well. By contrast, the condition with 25,000 cells/well had a relatively low signal induction, as well as a lower potency for the tool compound. This could be due to the negative effect of high cell confluence on the transduction efficiency. Though the potency of the 2-HBA analog was slightly lower with 20,000 cells/well, the signal induction was significantly higher than with any other condition tested. Considering the signal induction level and compound sensitivity, we selected 20,000 cells/well/40 μL assay volume as our final cell density condition.

Time course of BacMam transduction and compound treatment

This assay detects the interaction of Nrf2 and MafK proteins, and the data suggest that the signal intensity is affected by the relative amounts of tagged Nrf2 and MafK expressed. Since the assay relies on BacMam-dependent transient expression, the timing of transduction and compound treatment are considered important parameters for optimization. To obtain a significant interaction of tagged Nrf2 and MafK, yet minimize the exposure of cells to overexpressed exogenous proteins, a time course experiment was performed in the presence of 2-HBA analog from 8 to 40 h. Figure 5A shows the activation of Luc2C-Nrf2/MafK-Luc2N with 10 μM of the 2-HBA analog at varying incubation times. The data showed that 8 and 17 h treatments were insufficient to generate a high S/B ratio, whereas a 24 h treatment had a much higher S/B ratio. The S/B ratio obtained with a 24 h treatment was acceptable for a high throughput screening (HTS) assay. Though the S/B ratio was even higher at a 32 and 40 h treatment, these treatment times were not selected due to the concern that longer exposure of cells to exogenous proteins and compound treatment would induce toxicity and result in unexpected biological consequences. Longer treatment time (e.g., >40 h) possibly increased the cytotoxicity; therefore, it resulted in slightly lower signal induction. We subsequently chose the 24 h treatment as the final assay condition.

Fig. 5.

Optimization of compound treatment time and DMSO concentration. HEK-293 MSRII cells were plated at 20,000 cells/well/40 μL in DMEM/F12 supplemented with 2% FBS, transduced with Luc2C-Nrf2 and MafK-Luc2N BacMams at a total concentration of 0.5% with 40%/60% ratio (v/v). (A) Time course of transduction and treatment. Cells were transduced and treated with 10 μM 2-HBA analog for different time periods at 37°C with 5% CO_2. (B) DMSO titration. Cells were transduced and treated with different doses of 2-HBA analog under different DMSO concentrations for 24 h at 37°C with 5% CO_2. The data presented are means±SEM of duplicate wells. DMSO, dimethyl sulfoxide.

DMSO titration

The addition of a compound necessarily introduces DMSO into a cell culture, which potentially affects transduction efficiency, cell health, and other biological processes. To evaluate the DMSO tolerance limit, we titrated DMSO concentration in the culture medium while also testing the dose responses of tool compounds under different final DMSO concentrations. The minimum level of DMSO from the compound was at 0.5%. Additional DMSO was added to achieve desired DMSO concentrations. As shown in Figure 5B, the S/B ratio was the highest with 0.5% DMSO closely followed by 1% DMSO. At 1.5% DMSO, the S/B ratio was slightly decreased. DMSO at 2.5% significantly decreased the signal induction, while 4.5% DMSO totally abolished the response. These data demonstrated that the response to compound treatment was quite sensitive to the DMSO concentration. The optimal DMSO concentration for this assay should be kept at or below 1%.

Detection of Nrf2 and RunX2 Interaction Through EFC

It was reported that RunX2 interacts with Nrf2 in the nucleus.¹² The interaction of Nrf2 and RunX2 should also be able to reflect the stabilization of the Nrf2 protein. We generated split luciferase tagged Nrf2 and RunX2 BacMams and tested for protein interaction through EFC technology. The experimental conditions were evaluated in a manner that was similar to the interaction of Nrf2 and MafK. For the titration of cell density and serum concentration, very similar results were obtained for both pairs (Nrf2/RunX2 and Nrf2/MafK).

Similarly, HEK-293 MSRII cells were transduced with tagged Nrf2 BacMam and the pairing RunX2 BacMam at different total concentrations (10%, 5%, and 2.5%), respectively. At each total BacMam concentration, the ratio of tagged Nrf2/tagged RunX2 varied at 100%/0%, 70%/30%, 50%/50%, 30%/70%, and 0%/100%. The cells were then treated with 2-HBA analog at different doses for 24 h at 37°C and 5% CO₂. After compound treatment, luciferase activity was measured as described earlier. Figure 6 shows the data under the treatment of 10 μM 2-HBA analog for both Luc2C-Nrf2/RunX2-Luc2N and Nrf2-Luc2N/Luc2C-RunX2. When tagged Nrf2 or tagged RunX2 was used alone, there was no luciferase activity. In the presence of both components of Nrf2 and RunX2, the luciferase activity was dramatically increased. The titration of total BacMam concentration shows that a 10% and 5% BacMam concentration generated a much higher luciferase activity than does a 2.5% total BacMam concentration. This is likely due to the BacMam dose-dependent expression of proteins. Overall, 10% of total BacMam concentration had better dose-response curves for tested tool compounds (data not shown). We, therefore, chose 10% of total BacMam concentration for this interaction. The data again strongly suggest that the luminescence signal was the result of the specific interaction of tagged Nrf2 and tagged RunX2. This interaction enabled the two fragments Luc2N and Luc2C to be in close proximity and form active luciferase.

Fig. 6.

Titration of total BacMam concentration and the ratio of the two paired BacMams (A) Luc2C-Nrf2 and RunX2-Luc2N EFC pair (B) Nrf2-Luc2N and Luc2C-RunX2 EFC pair. HEK-293 MSRII cells were co-transduced with the paired BacMams at varying concentrations and ratios (expressed as v/v). Cells were resuspended in DMEM/F12 medium supplemented with 2.5% FBS. Twenty thousand cells/well/40 μL were transduced and treated with 10 μM 2-HBA analog for 26 h at 37°C with 5% CO₂. Luciferase activity was measured. The data presented are means±SEM of triplicate wells.

The ratio of the two BacMams expressing tagged Nrf2 and RunX2 was also titrated. As shown in Figure 6, the highest S/B ratio was obtained with Nrf2/RunX2 at 50%/50% (v/v). At an equal volume ratio, the MOT ratio was about 119/48 for Luc2C-Nrf2/RunX2-Luc2N at 10% total. The MOT for Nrf2-Luc2N/Luc2C-RunX2 was about 45/46 at 10% total BacMam.

Evaluation of Nrf2/MafK and Nrf2/RunX2 EFC with Nrf2 Activators

To further characterize the assay, several known Nrf2 activators (L-sulfroaphane, 2-HBA, 2-HBA analog, tBHQ, and 15d-PGJ2) were evaluated with the following two EFC pairs: Luc2C-Nrf2/MafK-Luc2N and Luc2C-Nrf2/RunX2-Luc2N. Compounds were tested at different concentrations under the optimized conditions of both EFC pairs. For instance, Figure 7 shows the dose-response curves obtained from cells transduced with the split-luciferace EFC pairs after treatment with varying concentrations of these five Nrf2 activators. The 2-HBA analog showed significant activation in both EFC pairs with a similar pEC50 value around 5.5. As can be seen from Figure 7, the maximal signal induction of 2-HBA analog was around 20-fold in both pairs at ∼10 μM, though the maximal signal of induction was often higher with the Luc2C-Nrf2/RunX2 pair than that with the Luc2C-Nrf2/MafK-Luc2N pair. 2-HBA also showed around sevenfold maximal signal induction in both pairs. tBHQ showed very good efficacy (above 10-fold signal induction), but the potency was not as high (maximal induction at 50 μM). Though 15d-PGJ2 had lower signal induction (<10-fold), its potency seemed higher (maximal induction at around 7 μM). L-sulforaphane showed only maximal around fivefold induction at ∼20 μM in both pairs. Overall, all these five known Nrf2 activators showed expected induction in the assay. The response level was comparable in both Nrf2/MafK and Nrf2/RunX2 pairs.

Fig. 7.

Evaluation of known Nrf2 activators in EFC with two different paired BacMams: (A) HEK-293 MSRII cells were transduced with Luc2C-Nrf2/MafK-Luc2N BacMam at total concentration of 0.5% with 40%/60% volume to volume ratio; (B) HEK-293 MSRII cells were transduced with Luc2C-Nrf2/RunX2-Luc2N BacMam at total 10% with 50%/50% volume-to-volume ratio. Cells were plated at 20,000 cells/well/40 μL in DMEM/F12 supplemented with 2% FBS, transduced with defined BacMam concentration, and treated with compounds at different doses for 24 h. Luciferase activity was then measured. The data presented are means±SEM of duplicate wells. 15d-PGJ2, prostaglandin J2; 2-HBA, (1E,4E)-1,5-bis(2-hydroxyphenyl)-1,4-pentadien-3-one; tBHQ, tert-butylhydroquinone.

Notably, these compounds had a dramatic decrease at higher concentrations, which gave rise to a bell-shaped curve in full-dose response experiments. These bell-shaped curves were likely due to the cytotoxicity resulting from the robust activation of Nrf2. Due to this, it was difficult to obtain a normal dose-response curve for pEC50 fitting. To accurately evaluate the activity of compounds, we used two parameters from the experiment data: the potency generated from the dose curve and the maximal signal induction compared with DMSO.

As shown in Figure 7, it was difficult to achieve a plateau in the dose-response curves. Hence, it was difficult to choose a compound that had a high response at a plateau concentration. High variability, which contributes to the poor robustness feature, was observed using the concentration of a compound at a slope range as a high control for the assay. This was due to the fact that the induction level was very sensitive to the variation of the compound concentration in the slope range resulting from liquid handling error. After testing a few compounds at a varied concentration as a high control for the assay, we chose an in-house compound as a high control to test the robustness features of Luc2C-Nrf2/RunX2-Luc2N. This compound had a relatively broader plateau concentration, though the highest S/B was only around ninefold. Using this compound as the high control and DMSO as the low control, we ran a pilot screen of 1,408 compounds at 10 μM in the final assay condition to evaluate the robustness feature and hit rate of the screen. Since both EFC pairs showed similar responses to tested tool compounds, we chose Luc2C-Nrf2/RunX2-Luc2N pair to illustrate the point. As shown in Table 1, all of the plates had good Z′ values with a S/B ratio around 9. The two copies of the plates, tested on different days, had an excellent correlation (correlation coeff. at 0.93, data not shown). The histogram of hits showed even distribution centered around 0% activation, indicating that the assay was not biased for activation (data not shown). Moreover, the hit rate was ∼1% when the cut-off value was set at 45%. This was deemed acceptable for a cell-based HTS assay, particularly considering that the high control only had a modest activation (around ninefold).

Table 1.

Statistics of Luc2C-Nrf2/RunX2-Luc2N Enzyme Fragment Complementation Assay

	Experiment 1		Experiment 2
Plate no.	Z′	S/B	Z′	S/B
1	0.72	6.0	0.59	8.9
2	0.70	6.2	0.61	7.5
3	0.75	5.7	0.55	9.3
4	0.70	6.0	0.57	8.9
5	0.67	5.3	0.49	7.8

HEK293 MSRII cells were plated at 20,000 cells/well/40 μL, transduced with Luc2C-Nrf2 and RunX2-Luc2N BacMam at total 10% with equal volume, treated with dimethyl sulfoxide for Column 6 as the low control, or with a GSK compound for column 18 as the high control. S/B=Avg. signal of column 18/Avg. signal of column 6. Z′ is calculated as described in the Methods section.

Discussion

Protein–protein interactions play a pivotal role in regulating biological processes, and their importance in drug discovery is well documented.²⁹ EFC is a powerful tool that detects protein–protein interactions in living cells. Different reporter systems have been used to build EFC assays and to monitor protein dimerization,³⁰ protein degradation,³¹ and protein nuclear translocation.³² A β-galactosidase EFC assay has been reported for HTS.³³ In this article, we successfully developed a cell-based luciferase EFC assay in a 384-well format to detect Nrf2 activation. Nrf2 is degraded through the ubiquitin-proteasome system and has a very short half life.^34,35 On activation, Nrf2 gets stabilized, translocates into the nucleus where it interacts with other transcription factors such as MafK and RunX2, and binds ARE elements, resulting in increased Phase II gene expression. We designed two pairs of fusion constructs based on the interactions just mentioned: Nrf2/MafK and Nrf2/RunX2. The first line of evidence that confirmed the specific interaction of these protein–protein pairs came from the grid-titration experiments, where a higher than basal signal was produced only in the presence of both tagged Nrf2 and tagged MafK or tagged Nrf2 and tagged RunX2 (Figs. 3 and 6). We then set out to optimize each assay so that it could be used to screen a large compound collection.

The optimization of the assay revealed several factors that potentially affected assay performance. Since the assay signal was directly related to the amount of the two binding partners, the expression levels of the two components were critical to the signal intensity and background level. For Nrf2 and MafK interaction, high BacMam concentration resulted in relatively high background signals (Fig. 3). This was likely due to some level of Nrf2 translocation into the nucleus under conditions where Nrf2 was over-expressed at high BacMam virus concentrations. In fact, it has been reported that overexpressed Nrf2 protein is able to transfer to the nucleus under unstressed conditions.³⁶ On the one hand, it was concerned that the exogenous Nrf2, MafK or RunX2 would interfere with biological processes, resulting in nonphysiological responses. On the other hand, a certain level of expression was required to generate enough interaction between the two components so that luciferase activity was high enough to be distinguished from the background. Clearly, the proper expression level of exogenous Nrf2 protein was critical to achieve the best sensitivity and S/B ratio. Our assay condition was, hence, intensively titrated against the total BacMam concentration and the ratio of the two components. Instead of choosing the condition with the highest signal induction (30 to 40 h induction), we selected a 24 h induction time that showed an acceptable signal induction with minimized exposure of cells to these exogenous proteins. Meanwhile, we also checked cell viability post BacMam transduction under the selected conditions. It was shown that no significant cytotoxicity was observed (data not shown). Since our primary goal is to develop an assay which is able to monitor Nrf2 activation, it is logical to assume that the best assay would be developed by calibrating Nrf2 protein levels to be low, relative to its binding partners such as MafK or RunX2. The addition of compounds would stabilize the Nrf2 protein via the disruption of the Nrf2-Keap1-Cul3 complex and allow free Nrf2 to interact with MafK or RunX2 when it enters the nucleus. Our results suggest that the expression level of the two binding partners should be carefully titrated to obtain the highest sensitivity and S/B ratio, and to minimize the interference of exogenous protein expression.

Serum concentration is another critical factor for the assay performance. It can potentially affect cell status, expression level, and compound potency. Our data demonstrate that either high or low serum concentrations (such as 10% or 0%) do not produce the best S/B ratio or dose-response curves (Fig. 4). Ten percent FBS could decrease the potency of compounds via serum protein binding to the compounds. On the other hand, without serum, cell proliferation and viability could be compromised, resulting in low protein expression and low S/B ratio. It is clearly important to optimize serum concentration in a cell-based assay and for most of our cell based assays, we typically find that the optimal serum concentration is 1%–2%.

In this study, we analyzed two interactions: Nrf2 with MafK and Nrf2 with RunX2. For each interaction, there were two different tagged fusions (Luc2C or Luc2N interchangeable). Titration data showed that there was no dramatic difference in compound-induced luciferase activity between swapped fusions (e.g., N-term vs. C-term fusion) for the same interaction (compare Fig. 3 to Fig. 6). For example, the optimized assay of each pair (Luc2C-Nrf2/MafK-Luc2N and Nrf2-Luc2N/Luc2C-MafK) had a total BacMam concentration of 0.5%, with a similar serum concentration and cell density. The BacMam transduction time was also comparable for these two pairs. The same is true when comparing Luc2C-Nrf2/RunX2-Luc2N with Nrf2-Luc2N/Luc2C-RunX2 pairs. However, a different interaction pair had a different requirement for the total BacMam concentration. The maximal signal for Nrf2/MafK interaction was reached at 0.5% total BacMam concentration, whereas the maximal signal for Nrf2/RunX2 was between 5% and 10%. Though the tested compounds had similar potencies, the induction level was slightly different between the two interactions. The majority of the tested compounds showed slightly higher responses in Nrf2/MafK than in Nrf2/RunX2 interaction. This was possibly due to the difference of the Nrf2 affinity to MafK and RunX2. Taken together, our data indicate that the tag position had little effect on the complementation for the EFC pairs tested, while the characterization of the assay may depend mainly on the interaction of the target proteins.

In summary, through the titration of several critical components in the assay, a cell-based high throughput EFC assay was developed that identifies Nrf2 activators (Table 2). The assay was optimized to be sensitive, robust (Z′≥0.5), and amenable for HTS. In this assay, the readout directly monitors the translocation and interaction of the Nrf2 protein with its functional binding partners in the nucleus. This assay, hence, provides information directly on the target protein level and its function. This is a clear advantage over the classical promoter-driven luciferase reporter assay. The sensitivity of the developed assay has been demonstrated with several known Nrf2 activators. The obtained potency and efficacy are consistent with those from other assays in our laboratory such as a luciferase reporter assay (data not shown). The robustness features and the hit rate of the assay were also evaluated by screening a small compound collection, which demonstrated that this assay is suitable for a HTS. The EFC technology described here should be applicable for studying other types of protein–protein interactions.

Table 2.

High Throughput Screening Assay Protocol Table

Step	Parameter	Value	Description
1	Compound library	400 nL	10 μM final from 1 mM stock solution for single shot. 1:3-fold dilution of 10 mM stock for dose-response
2	Controls	400 nL	Dimethyl sulfoxide in column 6 and 1.5 mM in-house compound in column 18
3	Plating of cell BacMam solution	40 μL	20,000 cells/well of mixture of BacMam and cells
4	Incubation time	22–24 h	Plate was incubated at room temperature for 30 min before incubation at 37°C and 5% CO₂
5	Reporter reagent	20 μL	Steady-Glo luciferase Substrate solution was added to each well
6	Incubation time	20 min	Plates were kept in the dark at an ambient temperature
7	Assay readout	clear	CCD imager, luminescence mode

Step Notes

1. Compounds were stamped by Echo Dispenser.

2. Compounds were stamped by Echo Dispenser.

3. Use the combi standard tube to dispense 40 μL/well of Cell BacMam solution to the well.

4. Plate lidded until read.

5. The frozen components of Steady-Glo Luciferase Substrate and Buffer were equilibrated and mixed. They were kept in the dark at room temperature and were prepared on the same day of the experiment.

6. Plate lidded until read.

7. Clear filter, 30 s exposure with luminescence label.

Footnotes

Acknowledgments

The authors thank the Sample Management Technology Group for their help with compound management; Bill Rumsey, Jim Callahan, Yolanda Sanchez, and Josh Cottom for their critical reading of the article; and Gordon McIntyre and Thomas Meek for their support of carrying out this work.

Disclosure Statement

No competing financial interests exist.

Abbreviations

References

Kensler

, Wakabayashi

, Biswal

. Cell survival responses to environmental stresses via the Keap1-Nrf2-ARE pathway. Annu Rev Pharmacol Toxicol, 2007; 47:89–116.

Kobayashi

, Yamamoto

. Molecular mechanisms activating the Nrf2-Keap1 pathway of antioxidant gene regulation. Antioxid Redox Signal, 2005; 7:385–394.

Kim

, Cha

, Surh

. A protective role of nuclear factor-erythroid 2-related factor-2 (Nrf2) in inflammatory disorders. Mutat Res Fundam Mol Mech Mutagen, 2010; 690:12–23.

Kobayashi

, Li

, Iwamoto

, Nakajima-Takagi

, Kaneko

, Nakayama

, Eguchi

, Wada

, Kumagai

, Yamamoto

. The antioxidant defense system Keap1-Nrf2 comprises a multiple sensing mechanism for responding to a wide range of chemical compounds. Mol Cell Biol, 2009; 29:493–502.

Singh

, Vrishni

, Singh

, Rahman

, Kakkar

. Nrf2-ARE stress response mechanism: A control point in oxidative stress-mediated dysfunctions and chronic inflammatory diseases. Free Radic Res, 2010; 44:1267–1288.

Cullinan

, Gordan

, Jin

, Harper

, Diehl

. The Keap1-BTB protein is an adaptor that bridges Nrf2 to a Cul3-based E3 ligase: oxidative stress sensing by a Cul3-Keap1 ligase. Mol Cell Biol, 2004; 24:8477–8486.

Villeneuve

, Lau

, Zhang

. Regulation of the Nrf2-Keap1 antioxidant response by the ubiquitin proteasome system: an insight into cullin-ring ubiquitin ligases. Antioxid Redox Signal, 2010; 13:1699–1712.

Zhang

, Lo

, Cross

, Templeton

, Hannink

. Keap1 is a redox-regulated substrate adaptor protein for a Cul3-dependent ubiquitin ligase complex. Mol Cell Biol, 2004; 24:10941–10953.

Zhang

. Mechanistic studies of the Nrf2-Keap1 signaling pathway. Drug Metab Rev, 2006; 38:769–789.

10.

Itoh

, Wakabayashi

, Katoh

, Ishii

, Igarashi

, Engel

, Yamamoto

. Keap1 represses nuclear activation of antioxidant responsive elements by Nrf2 through binding to the amino-terminal Neh2 domain. Genes Dev, 1999; 13:76–86.

11.

Itoh

, Mimura

, Yamamoto

. Discovery of the negative regulator of Nrf2, Keap1: a historical overview. Antioxid Redox Signal, 2010; 13:1665–1678.

12.

Hinoi

, Fujimori

, Wang

, Hojo

, Uno

, Yoneda

. Nrf2 negatively regulates osteoblast differentiation via interfering with Runx2-dependent transcriptional activation. J Biol Chem, 2006; 281:18015–18024.

13.

Itoh

, Chiba

, Takahashi

, Ishii

, Igarashi

, Katoh

, Oyake

, Hayashi

, Satoh

, Hatayama

, Yamamoto

, Nabeshima

. An Nrf2/small Maf heterodimer mediates the induction of phase II detoxifying enzyme genes through antioxidant response elements. Biochem Biophys Res Commun, 1997; 236:313–322.

14.

Itoh

, Igarashi

, Hayashi

, Nishizawa

, Yamamoto

. Cloning and characterization of a novel erythroid cell-derived CNC family transcription factor heterodimerizing with the small Maf family proteins. Mol Cell Biol, 1995; 15:4184–4193.

15.

Eftekharzadeh

, Maghsoudi

, Khodagholi

. Stabilization of transcription factor Nrf2 by tBHQ prevents oxidative stress-induced amyloid beta formation in NT2N neurons. Biochimie, 2010; 92:245–253.

16.

Ramsey

, Glass

, Montgomery

, Lindl

, Ritson

, Chia

, Hamilton

, Chu

, Jordan-Sciutto

. Expression of Nrf2 in neurodegenerative diseases. J Neuropathol Exp Neurol, 2007; 66:75–85.

17.

Luker

, Piwnica-Worms

. Optimizing luciferase protein fragment complementation for bioluminescent imaging of protein-protein interactions in live cells, animals. Methods in Enzymology Imaging in Biological Research, Part A, 385 Conn

. 349–360. Academic Press: St. Louis, MO, 2004.

18.

Michnick

. Protein fragment complementation strategies for biochemical network mapping. Curr Opin Biotechnol, 2003; 14:610–617.

19.

Eglen

, Singh

. Beta galactosidase enzyme fragment complementation as a novel technology for high throughput screening. Comb Chem High Throughput Screen, 2003; 6:381–387.

20.

Naqvi

, Lim

, Rouhani

, Singh

, Eglen

. Beta galactosidase enzyme fragment complementation as a high-throughput screening protease technology. J Biomol Screen, 2004; 9:398–408.

21.

Galarneau

, Primeau

, Trudeau

, Michnick

. Beta-lactamase protein fragment complementation assays as in vivo and in vitro sensors of protein protein interactions. Nat Biotechnol, 2002; 20:619–622.

22.

Paulmurugan

, Gambhir

. Novel fusion protein approach for efficient high-throughput screening of small molecule-mediating protein-protein interactions in cells and living animals. Cancer Res, 2005; 65:7413–7420.

23.

Stefan

, Aquin

, Berger

, Landry

, Nyfeler

, Bouvier

, Michnick

. Quantification of dynamic protein complexes using Renilla luciferase fragment complementation applied to protein kinase A activities in vivo. Proc Natl Acad Sci USA, 2007; 104:16916–16921.

24.

Fornwald

, Lu

, Wang

, Ames

. Gene expression in mammalian cells using BacMam, a modified baculovirus system. Methods Mol Biol, 2007; 388:95–114.

25.

Xie

, Tang

, Lu

, Ames

, Ratcliffe

, Li

. Development of a high throughput cell-based assay for soluble epoxide hydrolase using BacMam technology. Mol Biotechnol, 2010; 45:207–217.

26.

Kost

, Condreay

, Ames

, Rees

, Romanos

. Implementation of BacMam virus gene delivery technology in a drug discovery setting. Drug Discov Today, 2007; 12:396–403.

27.

Gao

, Talalay

. Induction of phase 2 genes by sulforaphane protects retinal pigment epithelial cells against photooxidative damage. Proc Natl Acad Sci USA, 2004; 101:10446–10451.

28.

Seibert

, Morchel

, Gulden

. Factors influencing nominal effective concentrations of chemical compounds in vitro: medium protein concentration. Toxicol In Vitro, 2002; 16:289–297.

29.

Luker

, Smith

, Luker

, Gammon

, Piwnica-Worms

. Kinetics of regulated protein-protein interactions revealed with firefly luciferase complementation imaging in cells and living animals. Proc Natl Acad Sci USA, 2004; 101:12288–12293.

30.

Remy

, Wilson

, Michnick

. Erythropoietin receptor activation by a ligand-induced conformation change. Science, 1999; 283:990–993.

31.

Zhao

, Vainshtein

, Gellibolian

, Shu

, Dotimas

, Wang

, Fung

, Horecka

, Bosano

, Eglen

. Homogeneous assays for cellular protein degradation using beta-galactosidase complementation: NF-kappaB/IkappaB pathway signaling. Assay Drug Dev Technol, 2003; 1:823–833.

32.

Fung

, Peng

, Kobel

, Dotimas

, Kauffman

, Olson

, Eglen

. A homogeneous cell-based assay to measure nuclear translocation using beta-galactosidase enzyme fragment complementation. Assay Drug Dev Technol, 2006; 4:263–272.

33.

Graham

, Bevan

, Lowe

, Palmer

, Rees

. Application of beta-galactosidase enzyme complementation technology as a high throughput screening format for antagonists of the epidermal growth factor receptor. J Biomol Screen, 2001; 6:401–411.

34.

McMahon

, Itoh

, Yamamoto

, Hayes

. Keap1-dependent proteasomal degradation of transcription factor Nrf2 contributes to the negative regulation of antioxidant response element-driven gene expression. J Biol Chem, 2003; 278:21592–21600.

35.

Taguchi

, Motohashi

, Yamamoto

. Molecular mechanisms of the Keap1-Nrf2 pathway in stress response and cancer evolution. Genes Cells, 2011; 16:123–140.

36.

Kweon

, Adhami

, Lee

, Mukhtar

. Constitutive overexpression of Nrf2-dependent heme oxygenase-1 in A549 cells contributes to resistance to apoptosis induced by epigallocatechin 3-gallate. J Biol Chem, 2006; 281:33761–33772.

Development of a Cell-Based High Throughput Luciferase Enzyme Fragment Complementation Assay to Identify Nuclear-Factor-E2-Related Transcription Factor 2 Activators

Abstract

Introduction

Materials And Methods

Reagents

BacMam Vector and EFC Constructs Generation

Cell Transduction with BacMam Viruses and Compound Treatment

Measurement of Reconstituted Luciferase Activity

Compound Addition and pEC50 Measurement on Automated Platforms

Data Processing and Analysis

Results

Detection of Nrf2 and MafK Interaction Through EFC

Titration of total BacMam concentration and titration of the ratio of the two components

Titration of serum concentration in assay cell-culture medium

Cell density titration

Time course of BacMam transduction and compound treatment

DMSO titration

Detection of Nrf2 and RunX2 Interaction Through EFC

Evaluation of Nrf2/MafK and Nrf2/RunX2 EFC with Nrf2 Activators

Discussion

Footnotes

Acknowledgments

Disclosure Statement

Abbreviations

References

Compound Addition and pEC₅₀ Measurement on Automated Platforms