Inhibition of firefly luciferase activity by a HIF prolyl hydroxylase inhibitor.

The three hypoxia-inducible factor (HIF) prolyl-4-hydroxylase domain (PHD) 1-3 enzymes confer oxygen sensitivity to the HIF pathway and are novel therapeutic targets for treatment of renal anemia. Inhibition of the PHDs may further be beneficial in other hypoxia-associated diseases, including ischemia and chronic inflammation. Several pharmacologic PHD inhibitors (PHIs) are available, but our understanding of their selectivity and its chemical basis is limited. We here report that the PHI JNJ-42041935 (JNJ-1935) is structurally similar to the firefly luciferase substrate D-luciferin. Our results demonstrate that JNJ-1935 is a novel inhibitor of firefly luciferase enzymatic activity. In contrast, the PHIs FG-4592 (roxadustat) and FG-2216 (ICA, BIQ, IOX3, YM 311) did not affect firefly luciferase. The JNJ-1935 mode of inhibition is competitive with a Ki of 1.36 μM. D-luciferin did not inhibit the PHDs, despite its structural similarity to JNJ-1935. This study provides insights into a previously unknown JNJ-1935 off-target effect as well as into the chemical requirements for firefly luciferase and PHD inhibitors and may inform the development of novel compounds targeting these enzymes.


Introduction
The transcription factor hypoxia-inducible factor (HIF) is the master regulator of the cellular transcriptional response to oxygen deprivation (hypoxia) [1]. HIF is a heterodimer formed by one of three α (HIF-1α, HIF-2α, HIF-3α) and one β (HIF-β) subunit [2]. HIF-α subunits are regulated in an oxygen-dependent manner [1]. In normoxia, three prolyl-4-hydroxylase domain (PHD1-3) enzymes hydroxylate two prolyl residues within the oxygen-dependent degradation (ODD) domain of HIF-α, leading to an increased affinity of hydroxylated HIF-α to the von Hippel-Lindau (VHL) protein [3]. VHL is an adaptor protein of an E3 ubiquitin ligase complex that catalyzes the poly-ubiquitination of HIF-α, marking it for rapid proteasomal degradation [4]. Inhibition of the PHDs, for example by hypoxia or pharmacologic agents, stabilizes HIF-α, leading to enhanced HIF-dependent gene expression [3]. One of the most prominent HIF target genes is the hormone erythropoietin (Epo), which is produced in the adult kidney [5]. In chronic kidney disease (CKD), Epo production is no longer sufficient to maintain a normal hematocrit, leading to anemia [6]. Several different pharmacologic HIF PHD inhibitors (PHIs) are in advanced clinical trials for treatment of renal anemia [6,7] and the first compound (FG-4592/ roxadustat) has recently been approved for treatment of patients in China [8]. Interestingly, pre-clinical analyses show that PHIs may also be beneficial in other hypoxia-associated diseases, including chronic inflammation, fibrosis, ischemia and possibly even cancer [6,[9][10][11][12].
Firefly luciferase is widely used as research tool for the analysis of transcription factor activity, cellular ATP content, kinase activity and more [13]. The transcription factor HIF is extensively investigated for its essential role in the cellular response to hypoxia [14]. Therefore, reporter gene assays employing HIF-dependent firefly luciferase reporter constructs are part of the basic repertoire to study this central pathway. Furthermore, one of the currently best methods to analyze PHD activity in vivo is a transgenic mouse model carrying a firefly luciferase fused to the ODD domain of HIF-1α [15][16][17][18][19].
We previously compared the selectivity of different pharmacologic HIF hydroxylase inhibitors for the HIF hydroxylases, and analyzed their effect on HIF stabilization and activity as well as on cellular energy metabolism [20]. Surprisingly, the compound JNJ-1935 did not increase HIF-dependent firefly luciferase bioluminescence, although it strongly induced Epo in the same samples [20]. Therefore, we investigated the underlying mechanism.
In this study, we demonstrate that JNJ-1935 is a potent firefly luciferase inhibitor in cellulo, in cell lysates ex vivo as well as of the recombinant enzyme in vitro. Importantly, a strong competitive inhibition occurred in cellulo at JNJ-1935 concentrations that are commonly used to study the HIF pathway [20,21]. Although JNJ-1935 and the firefly luciferase substrate D-luciferin are structurally similar, D-luciferin had no effect on the PHD-dependent HIF-1α and HIF-2α destabilization.

Reporter Gene Assays
Firefly and Renilla luciferase bioluminescence were measured using the Dual-Luciferase Reporter Assay (Promega, Madison, Wisconsin, USA). For the assessment of PHI-dependent effects on firefly reporter gene expression ( Supplementary Fig. S1), HRB5rl cells were treated with PHIs for 20 h, the supernatant was removed, the cells were washed with PBS, Passive Lysis Buffer (Promega) was added for 5-10 min at RT Fig. 1. Characterization of the effect of JNJ-1935 in a HIF-dependent firefly luciferase reporter gene assay. HRB5rl cells were treated with the indicated PHIs for 20 h and the same cells lysates were analyzed for A, HIF-dependent firefly luciferase bioluminescence, B, control Renilla luciferase bioluminescence, C, protein concentrations and D, HIF-1α and HIF-2α protein levels by immunoblotting. Numbers above columns indicate induction factors relative to the corresponding vehicle control. FFL, firefly luciferase; RLL, Renilla luciferase; Conc., concentration; exp., exposure; RLU, relative light units. *, p < 0.05; **, p < 0.01; and ***, p < 0.001 compared to the corresponding vehicle control. § §, p < 0.01; and § § §, p < 0.001 compared to the single treatment with 100 μM FG-4592. All statistical analyses were performed by Student's t-test. A-C, Data are shown as mean + SEM from four independent experiments or D, are representative for two independent experiments.
before one freeze-thaw cycle was performed and the lysates were mixed with an equal volume of the Luciferase Assay Reagent II (Promega), containing the firefly luciferase substrate. Firefly luciferase bioluminescence was assessed by a microplate luminometer (Berthold Technologies, Bad Wildbach, Germany). Thereafter, freshly mixed Stop & Glo reagent (Promega; containing the Renilla luciferase substrate and quenching the firefly luciferase bioluminescence) was added and the emitted light from the Renilla luciferase was measured. For the investigation of the effect of JNJ-1935 on firefly luciferase in cell lysates, HRB5rl cells were incubated for 22 h under normoxic or hypoxic conditions to induce HIF-dependent firefly luciferase expression. Subsequently, the supernatant was removed, the cells were washed with PBS, the cells were lysed with Passive Lysis Buffer, PHIs or the corresponding solvent control were added to the cell lysates and firefly luciferase bioluminescence was determined immediately.

Analysis of Recombinant Firefly Luciferase Activity
Using a serial dilution of recombinant firefly luciferase (QuantiLum; Promega), we determined its optimal concentration for our assay as 242 pM (data not shown). The EC 50 of D-luciferin (Cayman) was analyzed by mixing 242 pM recombinant firefly luciferase with assay buffer (100 mM Tris-HCl pH 7.

Statistical Analysis
Student's t-test was applied for the statistical analysis of significance.

JNJ-1935 Selectively Affects Firefly Luciferase
PHIs are of increasing interest for the treatment of patients and are extensively studied as modulators of hypoxia signalling [11]. However, many compounds remain poorly characterized. We compared the effects of the PHIs JNJ-42041935 (JNJ-1935), FG-4592 and FG-2216 (alternative names: ICA, BIQ, IOX3, YM 311) on HIF-1α and HIF-2α stabilization and HIF activity ( Supplementary Fig. S1). In a HIF-dependent reporter gene assay FG-4592 and FG-2216 significantly increased firefly luciferase bioluminescence by 11.2 or 7.5 fold, respectively (Fig. 1A). JNJ-1935 in turn did not lead to any significant difference in firefly luciferase bioluminescence compared to vehicle control (Fig. 1A). Combined treatment with both FG-4592 and JNJ-1935 significantly decreased the emitted light from firefly luciferase compared to FG-4592 treatment alone in a JNJ-1935 concentration dependent manner (Fig. 1A). Renilla luciferase bioluminescence showed some significant but not biologically relevant (0.9-1.5-fold) differences following the various compound treatments in the same samples (Fig. 1B). These results indicated that the PHIs did not affect Renilla luciferase activity, because the expression of Renilla luciferase was constitutive and not controlled by HIF or, hence, the PHIs (Supplementary Fig. S1). Total protein concentrations were comparable (Fig. 1C), indicating that there was no major differential effect by any of the PHIs on cell growth, proliferation or death during the course of this experiment. Of note, in the same samples all PHIs, including JNJ-1935, stabilized HIF-1α and HIF-2α to a similar extent (Fig. 1D). HIF-1α and HIF-2α were also strongly stabilized following the combined treatment with both FG-4592 and JNJ-1935, with no obvious difference in comparison to the separate treatments with either JNJ-1935 or FG-4592 (Fig. 1D). This demonstrated that all PHIs inhibited the PHDs, but only FG-4592 and FG-2216 led to a HIF-dependent increase of firefly luciferase bioluminescence. Because Renilla luciferase bioluminescence was not affected by JNJ-1935, these results indicated that JNJ-1935 Firefly luciferase bioluminescence was determined in the cell lysates with or without the addition of the indicated PHIs or the corresponding vehicle control. All samples were normalized to the corresponding bioluminescence in untreated normoxic cell lysates (first column). Numbers above columns indicate reduction factors relative to the corresponding vehicle control. FFL, firefly luciferase; Conc., concentration; #, p < 0.05; and ###, p < 0.001 compared to normoxia untreated; §, p < 0.05; and §, p < 0.01 compared to hypoxia untreated; ***, p < 0.001 compared to the corresponding vehicle control. All statistical analyses were performed by Student's t-test. Data are shown as mean + SEM from three independent experiments.
selectively affected the firefly luciferase.

JNJ-1935 Inhibits Firefly Luciferase Enzymatic Activity
Our previous analyses did not reveal whether JNJ-1935 affected firefly luciferase gene transcription, mRNA stability, protein levels or enzymatic activity. A comparison of the chemical structures of JNJ-1935, D-luciferin (the firefly luciferase substrate) and coelenterazine (the Renilla luciferase substrate) showed an intriguing structural similarity between JNJ-1935 and D-luciferin but not between JNJ-1935 and coelenterazine ( Fig. 2A). FG-4592 and FG-2216 were structurally different to JNJ-1935, D-luciferin and coelenterazine ( Fig. 2A). This was in agreement with our observation that firefly but not Renilla luciferase bioluminescence was specifically affected by JNJ-1935 (Figs. 1A, B).
To test whether JNJ-1935 directly inhibits firefly luciferase enzymatic activity, we first exposed HRB5rl cells to hypoxic conditions (0.2% O 2 for 22 h), leading to a more than 20-fold increase in firefly luciferase bioluminescence in the corresponding cell lysate (Fig. 2B). Following incubation of the same cell lysates with 100 μM FG-4592 or 500 μM FG-2216, a modest reduction of the emitted light from firefly luciferase was observed by 1.3 to 1.8-fold (Fig. 2B). Incubation with JNJ-1935, however, led to a concentration-dependent decrease of firefly luciferase bioluminescence in both normoxic and hypoxic cell lysates (Fig. 2B). The lowest JNJ-1935 concentration of 6.25 μM decreased the emitted light from firefly luciferase by 7 and 8-fold, whereas the highest JNJ-1935 concentration of 100 μM diminished the bioluminescence by 186.5 and 317-fold in cell lysates from normoxic and hypoxic cells, respectively (Fig. 2B). In summary, in the utilized samples firefly luciferase protein production occurred independent of the PHIs. Nonetheless, JNJ-1935 abrogated firefly luciferase bioluminescence, suggesting that JNJ-1935 directly affected specific firefly luciferase enzymatic activity.

JNJ-1935 Is a Potent Competitive Firefly Luciferase Inhibitor
For a detailed analysis of the observed inhibition of firefly luciferase by JNJ-1935, recombinant purified firefly luciferase was used. First, we determined the half maximal effective concentration (EC 50 ) and the K M value of D-luciferin in our experimental setup. The firefly luciferase K M value for D-luciferin has previously been reported as 2-10 μM [29]. In our experiments, the EC 50 for D-luciferin was determined as 2.05 μM (Fig. 3A) and the K M value as 2.59 μM (Supplementary Fig. 2SA). Subsequently, the effective range of JNJ-1935 was investigated using Dluciferin at its estimated EC 50 together with 0.01-200 μM JNJ-1935 (Fig. 3B). The IC 50 of JNJ-1935 was calculated as 2.63 μM. Next, the JNJ-1935 MOI of firefly luciferase was investigated. Dixon and Lineweaver-Burk plots demonstrated linear correlations that were characteristic for a competitive inhibitor (Figs. 3C, D). The K i of JNJ-1935  for firefly luciferase was determined as 1.36 μM. To analyze the inhibition of firefly luciferase by JNJ-1935 further, eight different concentrations with varying D-luciferin concentrations ranging from 0.2-20 μM were compared with each other (Supplementary Fig. S2B). Interestingly, we observed an uneven distribution of the curves with a larger gap between 3 μM (10 -5.5 M) and 10 μM (10 −5 M; Supplementary  Fig. S2B). These results indicate that JNJ-1935 may not be a purely competitive inhibitor at higher concentrations.
D-luciferin analogues commonly show a K i ranging from 0.1 to 3.4 μM and an IC 50 from 0.2-15 μM [30]. Hence, the K i and IC 50 of JNJ-1935 fit well within the known range for D-luciferin analogues. At the structural core of JNJ-1935 is a benzimidazole moiety [31]. Benzimidazole-based moieties have previously been identified as a common chemical core structure of firefly luciferase inhibitors [32,33]. Compounds with a benzimidazole chemotype are generally thought to be competitors for D-luciferin [32]. In agreement with this, we found that the JNJ-1935 MOI was competitive. A competitive MOI suggests that JNJ-1935 inhibits free firefly luciferase by binding to the D-luciferin binding site [34].

D-Luciferin Does Not Inhibit the PHDs
JNJ-1935 and D-luciferin are structurally similar ( Fig. 2A) and JNJ-1935 inhibits firefly luciferase enzymatic activity. Therefore, we tested whether D-luciferin can in turn inhibit the PHDs. HEK293 cells were treated with 25-200 μM JNJ-1935 or D-luciferin, respectively, as well as with 100 μM FG-4592 or 500 μM FG-2216. The three known PHIs stabilized HIF-1α and HIF-2α protein (Fig. 4). D-luciferin, however, had no effect at the same concentrations as JNJ-1935 (Fig. 4), demonstrating that D-luciferin is not a PHD inhibitor at the tested concentrations in HEK293 cells.
It has previously been demonstrated that the acidic group in JNJ-1935 as well as the electron pairs of the nitrogen atoms on the benzimidazole and pyrazole moieties are needed for its interaction with PHD2 [31]. These structural components are also present in D-luciferin. But D-luciferin has only one aromatic ring, whereas JNJ-1935 has two. This chemical difference may impede the binding of D-luciferin to PHD2 residue R383. Another interaction of JNJ-1935 is formed with PHD2 Y303 via the JNJ-1935 benzimidazole NH group (and a water molecule) [31]. However, D-luciferin contains a benzothiazole instead of a benzimidazole and hence misses the NH group ( Fig. 2A). The interaction with PHD2 Y303 contributes to potency differences in the binding of molecules to PHD2 [35]. Overall, it seems unlikely that Dluciferin interacts with PHD2.

Conclusions
Understanding the selectivity and molecular principles of pharmacologic PHIs is of utmost importance for their use in basic research as well as for their translation to the clinics for the treatment of diseases such as ischemia and chronic inflammation. We here identified a previously unknown off-target effect of JNJ-1935 that is important to be considered when JNJ-1935 is analyzed for its effect on HIF-dependent reporter gene assays or other firefly luciferase-based experiments, including drug screens. Furthermore, our results provide insights into the chemical characteristics of firefly luciferase and PHD inhibition and may contribute to the development of novel pharmacologic compounds targeting the PHDs/firefly luciferase.

Declaration of Competing Interest
The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper. Figure S1. The HIF pathway and the principle of a HIF reporter gene assay. The transcription factor hypoxia-inducible factor (HIF) consists of the oxygen-dependently regulated α and the constitutively expressed β subunit. In normoxia, the prolyl-4-hydroxylase domain 1-3 (PHD1-3) proteins use molecular oxygen as co-substrate to hydroxylate HIF-α on two different prolyl residues. The von Hippel-Lindau (VHL) protein binds to prolyl-4-hydroxylated HIF-α and recruits an E3 ligase complex that poly-ubiquitinates HIF-α, targeting it for proteasomal degradation. In hypoxia, O2 is no longer available for the PHDs and the hydroxylation of HIF-α cannot occur. HIF-α escapes its proteasomal degradation, translocates into the nucleus, dimerizes with HIF-β, recruits the transcriptional co-activators p300/CBP and enhances HIF target gene expression via binding to hypoxia-response elements (HREs). In stably transfected HRB5rl cells, the genome of each cell contains separate genes that either encode a firefly (FFL) or a Renilla luciferase (RLL). The firefly luciferase gene is under the control of a SV40 promoter combined with HRE enhancers. When the heterodimeric HIF is present, it binds to the HREs and enhances firefly luciferase gene expression. The Renilla luciferase gene is under the control of the SV40 promoter only, no HREs are present. Therefore, the Renilla luciferase gene expression is constitutive and not regulated by HIF. The Renilla luciferase bioluminescence is used as control to normalize for biological variabilities, such as differences in cell number or cell death.