Furan- and Thiophene-2-Carbonyl Amino Acid Derivatives Activate Hypoxia-Inducible Factor via Inhibition of Factor Inhibiting Hypoxia-Inducible Factor-1

Induction of a series of anti-hypoxic proteins protects cells during exposure to hypoxic conditions. Hypoxia-inducible factor-α (HIF-α) is a major transcription factor that orchestrates this protective effect. To activate HIF exogenously, without exposing cells to hypoxic conditions, many small-molecule inhibitors targeting prolyl hydroxylase domain-containing protein have been developed. In addition, suppression of factor inhibiting HIF-1 (FIH-1) has also been shown to have the potential to activate HIF-α. However, few small-molecule inhibitors of FIH-1 have been developed. In this study, we synthesized a series of furan- and thiophene-2-carbonyl amino acid derivatives having the potential to inhibit FIH-1. The inhibitory activities of these compounds were evaluated in SK-N-BE(2)c cells by measuring HIF response element (HRE) promoter activity. Several furan- and thiophene-2-carbonyl amino acid derivatives inhibited FIH-1 based on correlations among the docking score of the FIH-1 active site, the chemical structure of the compounds, and biological HIF-α/HRE transcriptional activity.

Accordingly, in this study, we aimed to develop novel small-molecule inhibitors of FIH-1.

Design and Synthesis of New FIH-1 Inhibitors
FIH-1 has an Fe atom in its active center, and this Fe is held in place by two histidine and aspartic acid residues. 2OG is a required cofactor that mediates FIH-1 activity. Based on X-ray crystallography data (PBD ID: 1MZF), 2OG binds the Fe atom bidentately, and forms a hydrogen bond with Lys214 and Tyr145 in human HIF-1 (Figure 2a) [37]. In this study, furan-and thiophene-2-carbonyl amino acid derivatives were designed to antagonize 2OG binding [38] based on the X-ray crystallography structure. The designed structure had a bidentate site for Fe atom binding and a carboxylic acid that could form a hydrogen bond with Lys214 and/or Tyr145 residues (Figure 2b). The R 1 moiety contained an H, Me, phenyl, 4 -phenoxyphenyl, or 4 -biphenyl. The R 2 moiety was prepared with either an H or phenyl group. The R 3 moiety corresponded to an amino acid residue (glycine, alanine, phenylalanine, tryptophan, or tyrosine). A Me or Et group was introduced as the R 4 moiety to increase the lipophilicity and cell membrane permeability. The R 4 moiety was thought to be converted to H by hydrolysis in the cell. Many inhibitors have been reported to act as antagonists of 2OG [15][16][17]39]; however, the use of furan-and thiophene-2-carbonyl amino acid derivatives as antagonists of 2OG has not been evaluated. The synthetic strategy to design the compounds is summarized in Scheme 1. Furan-and thiophene-2-carbonyl amino acid derivatives were synthesized using condensation reactions of thiophene-2-carboxylic acids with amino acid salts. Nonsubstituted furan-or thiophene-2-carboxylic acids and benzofuran-or benzothiophene-2-carboxylic acids were obtained from commercial supplies. The other furan-or thiophene-2-carboxylic acids were synthesized as follows. 2-([1,1 -Biphenyl]-4-yl)thiophene and 2-(4-phenoxyphenyl)thiophene were prepared in moderate yields via a Negishi coupling reaction using the zinc salt of thiophene and aryl bromide (Scheme 2). Next, 5-substituted thiophene-2-carboxylic acids and 5-methylfuran-2-carboxylic acid were synthesized by carboxylating the corresponding furan or thiophene (Scheme 3).
Since the starting material was commercially available, the method for the synthesis of 4-phenylthiophene-2-carboxylic acid differed from that of 5-phenylfuran-2-carboxylic acid (Scheme 5). Suzuki-Miyaura cross-coupling of 4-bromo-2-thiophenecarbaldehyde and phenylboronic acid was used to obtain the corresponding coupling product, which yielded the corresponding carboxylic acid upon subsequent oxidation. For the condensation reaction, 1-ethyl-3-(3-dimethylaminopropyl)carbodiimide hydrochloride (EDC) was used as a condensation agent in the presence of 1-hydroxybenzotriazole (HOBt) (Scheme 6). The use of EDC/HOBt produced the condensation products in good yields and suppressed racemization of amide units.

Evaluation of HIF Activation by FIH-1 Inhibition
To evaluate HIF transactivation following FIH-1 inhibition in vivo, we employed a luciferase reporter assay system in the human neuroblastoma cell line SK-N-BE(2)c. The cells were stably transfected with a reporter vector possessing secretion-type luciferase (Metridia longa luciferase, MLuc) under control of the 7-time-repeat human VEGFA regulatory sequence (40 bp length containing the HRE) and mini TATA promoter. Hereafter these HRE reporter cells were designated SKN:HRE-MLuc cells [40]. To evaluate the inhibitory activity of FIH-1, SKN:HRE-MLuc cells were cultured under mild hypoxic conditions (3% O 2 ) based on previous studies showing that under normoxic conditions (21% O 2 ), FIH-1 inhibition does not significantly affect for HIF activation, whereas under 3% O 2 conditions, FIH-1 inhibition elevates HIF transcriptional activity [1,30,41].
To confirm that our proposed system could be utilized to evaluate FIH-1 inhibitory activity, FIH-1 was transiently silenced by transfecting SKN:HRE-MLuc cells for 72 h with siRNA. After 24 h culture with fresh medium under normoxic conditions, FIH-1 protein was analyzed by immunoblotting. FIH-1 protein levels in FIH-1 siRNA transfectants were significantly reduced compared with that in untreated or scrambled siRNA-transfected control cells (Figure 3a). Using this FIH-1 siRNA system, we compared the efficacy of dimethyloxalyl glycine (DMOG) or FibroGen compound (FG4592) under normoxic or hypoxic conditions (3% O 2 ) for 24 h (Figure 3b). Under hypoxic conditions, PHD proteins were mostly inactivated; therefore, the HIF-HRE top-up transcriptional activity under these conditions can be measured as FIH-1 inhibitory activity [30].
According to the experimental design indicated in Figure 3b, SKN:HRE-MLuc cells were transfected with the indicated siRNAs, after 72 h, the transfectants were treated with normoxic or hypoxic conditions (3% O 2 ) for 24 h, and the luciferase activities were determined (Figure 3c). HIF transcriptional activity on SKN:HRE-MLuc cells was significantly elevated during hypoxia treatment (compare A with B). Moreover, HIF was not stabilized in FIH-1 knockdown cells under normoxic conditions (compare A with C). In contrast, under hypoxic conditions, HIF transcriptional activity was enhanced in FIH-1 knockdown cells, supporting the inhibitory activity of FIH-1 (compare B with D). Treatment with DMOG, which inhibits both PHDs and FIH-1, resulted in higher HIF stabilization activity (compare A with E, F, G, or H). On the other hand, treatment with FG4592, which is a selective inhibitor of PHD [36,42], stabilized HIF compared with vehicle-treated cells (compare A with I). The difference between E and I was supported by changes in FIH-1 activity. Therefore, FG4592 treatment under hypoxic conditions only slightly stabilized HIF (compare I with J). Importantly, FG4592 treatment did not affect FIH-1 inhibitory activity. Additionally, SKN:HRE-MLuc cells were significantly activated following FIH-1 knockdown treatment under mild hypoxic conditions, even in the presence of FG4592 (compare J with L). Taken together, these results suggested that measuring HIF-HRE transcriptional activity with continuous mild hypoxia may reflect FIH-1 activity.
Next, to confirm that the proposed system could be used to evaluate FIH-1 inhibition, we treated the cells with DMOG, which can inhibit both PHD and FIH as a positive control, or dimethyl N-oxalyl-D-phenylalanine (DM-NOFD), which is a prodrug of N-oxalyl-D-phenylalanine, a reported FIH-1-selective inhibitor (Figure 4b) [31,34]. Treatment with DM-NOFD (100 µM) significantly enhanced HIF activity (Figure 4a). This result indicated that our FIH-1 evaluation system could be utilized for quantification of further optimized small compounds derived from DM-NOFD.  . The degree of induction is presented as relative luciferase units, with the value from DMSO-treated cells set as 1 for each treatment. Experiments were performed in triplicate. Data are means ± SEMs (n = 3). The statistical significance of results compared with data from the control group was calculated using one-way ANOVA with Newman-Keuls multiple-comparison tests. * p = 0.05 to 0.01; *** p < 0.001; (b) Chemical structure of DM-NOFD.
a 25% activity represents the lower dose limit showing 25% of the RLU intensity of the positive control DMOG; b 100 µM activity represents the RLU intensity at 100 µM stimulation compared with the positive control DMOG; c The values were calculated using the MTS assay results. ND = not determined.

Docking Simulations Using Furan-and Thiophene-2-Carbonyl Amino Acid Derivatives with FIH-1
To understand the inhibitory tendencies observed in HRE-MLuc reporter assays on SKN:HRE-MLuc cells, docking simulations with FIH-1 were performed with four compounds (7, 9, 10, and 16); among the compounds that stabilized HIF activity (i.e., compounds 7, 9-11, 15-18, 28, 30, and 38), these four compounds had characteristic scaffolds. Docking simulations were run on Molegro Virtual Docker 6.0.0 [43]. The corresponding hydrolyzed compounds were used in the docking simulations (compounds 7, 9, and 10). The results of the FIH-1 docking studies are shown in Table 3. For compound 7, a furan derivative with a bulky group (such as Ph-CH 2 -) in the R 3 position, and compound 9, a furan derivative with a 3-indolyl-CH 2 group in the R 3 position, both dockings had good scores (entries 1 and 2). The predicted binding models of compounds 7 and 9 within the active center of FIH-1 are shown in Figure 5a,b. When compound 9, which had an L-tryptophan scaffold, was compared with compound 10, which had a D-tryptophan scaffold, good docking scores were obtained (Table 3, entries 2 and 3). The predicted binding models of compounds 9 and 10 showed that both 3-indolyl-CH 2 groups in the R 3 position fit within the space, although compounds 9 and 10 were enantiomers to each other (Figure 5b,c). The thiophene compound 16, which was related to compound 9, also docked within FIH-1; however, it had a lower score (Table 3, entry 4). The predicted binding model of compound 16 showed a similar conformation (Figure 5d). These results supported the notion that compounds with bulky groups in the R 1 position did not show the activity because there was not sufficient space ahead of the R 1 position. In contrast, compound 1 did not selectively inhibit FIH-1 and could inhibit other enzymes that used 2OG as a cofactor because of its scaffold.

Inactibation of HIF by FIH-1 Inhibitors under Normoxic Conditions
We next measured HIF transactivity of FIH-1 inhibitors under normoxic conditions to confirm whether these inhibitors stabilized HIF via inhibiting PHD. Under normoxic conditions, inhibition of PHD dramatically activated HRE reporter. In contrast, when PHD was inhibited under hypoxic conditions, the rate of activation of the HRE reporter was low because hypoxia already partially activated the HRE reporter. The compounds showing HIF activation under hypoxic conditions (compounds 7, 9-11, 15-18, 28, 30, and 38) were evaluated by the luciferase assays in SKN:HRE-MLuc cells under normoxic conditions ( Figure 6). Treatment with 100 µM FG4592, a commonly used selective inhibitor of PHD, was used as a positive control and showed significant HRE reporter activity. In contrast, compounds 7, 9-11, 15-18, 28, 30, and 38 did not elicit any HIF transactivation. This result indicated that these compounds selectively inhibited FIH-1. At the same time, these compounds (7, 9-11, 15-18, 28, 30, and 38) did not fit into the catalytic domain of the PHD2 (data not shown). . HIF activation using the identified compounds (7, 9-11, 15-18, 28, 30, and 38) under normoxic conditions. The degree of induction is presented as relative luciferase units, with the value from DMSO treatment of cells set as 1 for each treatment. Experiments were performed in triplicate. Data are means ± SEMs (n = 3). The statistical significance of the results compared with data from the control group was calculated using one-way ANOVA with Newman-Keuls multiple-comparison tests. ns, p > 0.05; *** p < 0.001.

mRNA Expression in SK-N-BE(2) Cells
To confirm transactivation of HIF by our optimized compounds, the expression of HIF target genes was analyzed with quantitative polymerase chain reaction (PCR) in SK-N-BE(2)c cells. We attempted to analyze several HIF target genes including CA9, VEGF-α, EPO, and PHD3, under 3% O 2 conditions. Among these genes, as described previously [30], CA9 was significantly elevated when treated with compound 9 (Figure 7a). Notably, this induction of CA9 did not occur via HIF-1α mRNA transactivation (Figure 7b). In addition, de novo translation of HIF-1α mRNA was slightly suppressed by DMOG treatment but not by treatment with compound 9. . Gene induction is presented as relative mRNA levels, with the qRT-PCR values from DMSO-treated cells set as 1 for each treatment. Experiments were performed in triplicate. Data are means ± SEMs (n = 3). The statistical significance of the results compared with data from the control group was calculated using one-way ANOVA with Newman-Keuls multiple-comparison tests. ns, p > 0.05; * p < 0.05-0.01; *** p < 0.001.

Conclusions
In this study, we used SKN:HRE-MLuc cells cultured under mild hypoxic conditions to reflect FIH-1 activity even in the presence of PHD proteins. To evaluate FIH-1 function more specifically, N to C transactivation domain of the HIF-1α (amino acids 531-826) and GAL4 fusion protein with a GAL4-DNA binding domain-driven reporter system would be a powerful alternative evaluation method and high-throughput screening system [44]. Additionally, we designed and synthesized novel FIH-1 inhibitors having furan-or thiophene-2-carbonyl amino acids as the main scaffold. Several compounds showed HIF activation potential using SKN:HRE-MLuc cells under mild hypoxic conditions. These compounds tended to have bulky lipophilic moieties in the R 3 position. The results of docking simulations and luciferase assays under normoxic conditions supported the idea that the identified compounds activated HIF transcription by FIH-1 inhibition. Selective FIH-1 inhibitors are still rare; therefore, the identified compounds may provide alternative HIF activation tools.

General Information
DMSO, DMOG, and FG4592 were obtained from Merck (Darmstadt, Germany), and other general reagents were purchased from Nacalai Tesque (Kyoto, Japan). DM-NOFD was synthesized according to a literature method [45].

Preparation of 2-([1,1 -biphenyl]-4-yl)thiophene and 2-(4-phenoxyphenyl)thiophene [46]
Under an N 2 atmosphere, thiophene (1.5 g, 18 mmol) was placed in a three-necked flask with distilled tetrahydrofuran (THF, 30 mL). The flask was cooled at 0 • C, and n-butyllithium (ca. 1.6 M in hexane, 10 mL, 16 mmol) was added dropwise. After addition, the mixture was stirred at room temperature for 30 min. The flask was cooled at 0 • C again, and dried zinc chloride (2.0 g, 15 mmol) suspended in THF (10 mL) was added dropwise. After addition, the mixture was stirred at room temperature for 30 min., 4-bromobiphenyl or 4-bromodiphenyl ether (15 mmol) and Pd(PPh 3 ) 4 (0.69 g, 0.6 mmol) in THF (10 mL) were added, and the mixture was stirred at 80 • C. After stirring for 16 h, the mixture was cooled to room temperature and quenched with water. The resulting mixture was extracted with diethyl ether and dried with MgSO 4 . After filtration, the solvent was removed under reduced pressure. The desired pure thiophene was obtained after purification by column chromatography on silica gel. ((1,1 -biphenyl)-4-yl)thiophene-2-carboxylic acid, 2-(4-phenoxyphenyl)thiophene-2-carboxylic acid, 5-phenylthiophene-2-carboxylic acid, and 5-methylfuran-2-carboxylic acid [47] 2-Methylfuran was obtained from commercial supplies. 2-Phenylthiophene was prepared according to a previously described method [48]. Under an N 2 atmosphere, 2-([1,1 -biphenyl]-4-yl) thiophene, 2-(4-phenoxyphenyl)thiophene, or 2-methylfuran (1.0 equiv.) was placed in a three-necked flask with distilled THF or Et 2 O. The flask was cooled at 0 • C, and n-butyllithium (1.1 equiv.) was added dropwise over a period of 15 min. The mixture was then stirred for 30 min at room temperature. After the mixture was cooled at −78 • C with a dry ice/acetone bath, CO 2 gas was introduced under vigorous stirring and cooling. After stirring for 2 h, the mixture was warmed to room temperature and quenched with water. The ether layer was separated and washed with water. The combined solution was acidified with a 3 N HCl solution at 0 • C, and then extracted with EtOAc. The organic layer was washed with brine, dried with MgSO 4 , and concentrated under reduced pressure. The desired products were obtained without further purification.
Compounds 39 [52], 41 [53] were previously characterized. The analytical data of the final products are described below.

Evaluation of HIF Activity under Hypoxia Using a Luciferase Assay
SKN:HRE-MLuc (4 × 10 4 ) were incubated for 24 h on a 96 well culture plate. The medium was replaced with complete media and the test compounds (100, 25, 6.3 and 1.6 µM). DMSO (0.1%) was used as a vehicle control, DMOG (100 µM) and FG4592 (100 µM) were used as a PHDs or FIH-1 inhibitor. The cells were incubated for 24 h under a humidified atmosphere of 5% CO 2 and 3% O 2 at 37 • C. After incubation, the MLuc activity in culture media was measured with luciferase substrate (Takara, Shiga, Japan) by LB960 luminometer (Berthold, Wildbad, Germany). The HIF activity was calculated as the MLuc intensity, based on the vehicle and positive controls. The lower limit dose out of 100 µM, 25 µM, 6.3 µM, 1.6 µM was listed in Tables 1 and 2, where 25% RLU intensity was measured using DMOG as a positive control.

MTS Assay
Cellular toxicity was measured using the CellTiter 96 ® Aqueous One Solution Cell Proliferation Assay kit (Promega, Madison, WI, USA). The cells were stimulated with each compound and treated with the CellTiter solution. After incubation for 3 h, the absorbance of the solutions at 492 nm was measured by Multiskan FC (Thermo Fisher Scientific, Waltham, MS, USA). As a negative control, wells with culture media were used. Wells treated with DMSO (0.1%) were used as a positive control. The IC 50 values were determined using these controls.

Evaluation of HIF Activity under Normoxia Using a Luciferase Assay
SKN:HRE-MLuc (4 × 10 4 ) were incubated for 24 h on a 96 well culture plate. The medium was replaced with complete media and the tested compounds (100 µM). DMSO (0.1%) was used as a negative control, and DMOG (100 µM) was used as a positive control. The cells were incubated for 24 h under a humidified atmosphere of 5% CO 2 at 37 • C. After incubation, the MLuc activity in culture media was measured with the method descried above.

Silence of FIH-1 by Small Interfering RNA
The siRNA reagents, including the transfection reagents, were obtained from Thermo Fisher Scientific. Sequence: forward 5 -GAA ACA UUG AGA AGA UGC UUG GAG A-3 and reverse 5 -UCU CCA AGC AUC UUC UCA AUG UUU C-3 . SKN: HRE-MLuc were transfected with siRNA transiently, according to the protocol for Lipofectamine 2000 (Thermo Fisher Scientific) under normoxic conditions. The siRNA-transfectants were incubated for 72 h. To confirm the FIH-1 knockdown effect in cellular lysate was evaluated by immunoblotting after an additional 24 h culture under normoxic condition. Briefly, cell lysate from siRNA transfectants were separated by SDS polyacrylamide gels, transferred to polyvinylidene difluoride membranes. The blots were blocked with 5% skim milk (Nacalai Tesque) in TBS-Tween, and treated with anti-human FIH-1 polyclonal goat antibody (sc-26219, Santa Cruz Biotechnology, Dallas, TX, USA) or anti β-actin mouse monoclonal antibody (Merck) antibody, subsequently the blots were treated with horseradish peroxidase-conjugated anti-goat or -mouse secondary antibody. Immunoreactivity was visualized with LAS400 image analyzer (GE healthcare, Little Chalfont, UK) using the Western Lightning Chemiluminescence Reagent Plus kit (PerkinElmer, Waltham, MS, USA).

Gene Expression Analysis
Total RNA was extracted from cells using ISOGEN (Nippon Gene, Tokyo, Japan). cDNA syntheses were performed using random hexamer primers and SuperScript II Reverse Transcriptase (Thermo Fisher Scientific) following the manufacturer's instructions. Quantitative reverse transcription polymerase-chain reaction (RT-PCR) analysis was performed as follows: amplification was done in an initial denaturing step at 96 • C for 2 min, followed by 35 cycles of PCR (96 • C for 1 min, 57 • C for 30 s, 72 • C for 30 s) using recombinant Taq polymerase (Takara, Otsu, Japan) with the respective primer pairs. For human carbonic anhydrase IX (CA9), 5 -CCT GGC TGC TGG TGA CAT CC-3 and 5 -AAG GAA GTG GCA TAA TGA GC-3 . Real time RT-PCR was performed on a LightCycler480 thermal cycler system using a LightCycler480 SYBR Green I Master (Roche Applied Science, Indianapolis, IN) according to the manufacturer's instructions. Data were analyzed using the comparative Ct method as means of relative quantification, and were normalized to an endogenous reference (β-actin) and relative to a calibrator (normalized Ct value obtained from cells treated with vehicle only). Data were expressed as 2 −∆∆Ct .

Statistical Analyses
All data were analyzed using one-way ANOVA followed by Newman-Keuls Multiple Comparison Test using GraphPad Prism 6 software (La Jolla, CA, USA).