Identification of the hypertension drug niflumic acid as a glycine receptor inhibitor

Glycine is one of the major neurotransmitters in the brainstem and the spinal cord. Glycine binds to and activates glycine receptors (GlyRs), increasing Cl− conductance at postsynaptic sites. This glycinergic synaptic transmission contributes to the generation of respiratory rhythm and motor patterns. Strychnine inhibits GlyR by binding to glycine-binding site, while picrotoxin blocks GlyR by binding to the channel pore. We have previously reported that bath application of strychnine to zebrafish embryos causes bilateral muscle contractions in response to tactile stimulation. To explore the drug-mediated inhibition of GlyRs, we screened a chemical library of ~ 1,000 approved drugs and pharmacologically active molecules by observing touch-evoked response of zebrafish embryos in the presence of drugs. We found that exposure of zebrafish embryos to nifedipine (an inhibitor of voltage-gated calcium channel) or niflumic acid (an inhibitor of cyclooxygenase 2) caused bilateral muscle contractions just like strychnine-treated embryos showed. We then assayed strychnine, picrotoxin, nifedipine, and niflumic acid for concentration-dependent inhibition of glycine-mediated currents of GlyRs in oocytes and calculated IC50s. The results indicate that all of them concentration-dependently inhibit GlyR in the order of strychnine > picrotoxin > nifedipine > niflumic acid.

www.nature.com/scientificreports/ To search for new chemical compounds that block GlyRs, we screened a chemical library of ~ 1,000 approved drugs and pharmacologically active molecules through their ability to cause touch-evoked bilateral muscle contractions in zebrafish embryos. Strychnine served as a positive control. The screening identified nifedipine and niflumic acids as candidates of GlyR inhibitors. We also found that picrotoxin also affects zebrafish behavior when applied at high concentration. Our electrophysiological recordings using Xenopus oocytes revealed that all of the strychnine, picrotoxin, nifedipine and niflumic acids showed concentration-dependent blockade of glycine-gated currents in both homomeric and heteromeric GlyRs. In both human and zebrafish GlyR cases, the half-maximal inhibitory concentration (IC 50 ) was strychnine < picrotoxin < nifedipine < niflumic acid.

Results
Bath application of nifedipine or niflumic acid affects tactile response of zebrafish. We have previously reported that zebrafish embryos swim when touched at 24 and 48 h postfertilization (hpf), whereas strychnine-treated embryos show obvious shrinkage of the body without swimming due to the loss of GlyRmediated reciprocal inhibition 13,19 . We assumed that in vivo screening of chemical compounds that cause touchevoked body shrinkage of zebrafish is a promising way to search for new GlyR inhibitors. To this end, three zebrafish embryos were treated with a drug (10 ~ 100 μM) in a well of 96-well plate at the onset of 24 hpf and then assayed for touch responses at 27 and 48 hpf, when zebrafish embryos show tactile-induced short-distance forward movement and escape swimming, respectively 20 . Among ~ 1,000 drugs that have some known physiological activities, about 79% of the drugs did not affect tactile response, while 10% caused immotile and 10% killed embryos at both stages in our first screening performed in a blind manner. We found 10 drugs that caused touchevoked body shrinkage or weird uncoordinated response, the latter with contractions of small amplitude and/ or compromised rhythmicity (Table 1). We then purchased these drugs, performed second round of screening at several different concentrations and identified three drugs that caused touch-evoked body shrinkage: strychnine, nifedipine and niflumic acid, which are known as specific GlyR inhibitor, voltage-gated calcium channel inhibitor and cyclooxygenase inhibitor, respectively. Successful identification of strychnine in our assay indicates that our screening works well. Nifedipine was previously suggested as a GlyR inhibitor 21 . Although picrotoxin, the other known GlyR inhibitor, was included in our chemical library, we failed to identify it in our screening.
We then detailed tactile response of zebrafish embryos at 48 hpf for four drugs: strychnine, picrotoxin, nifedipine and niflumic acid. Upon touch at the tail, control (1% DMSO) zebrafish embryos showed escape swimming by rhythmic side-to-side trunk contractions ( Fig. 1a; Supplementary video 1). The shape of the notochord was straight during swimming. Embryos treated with 70 μM strychnine looked normal before touch. However, when touched, they initially exhibited bilateral trunk contractions resulting in the shrinkage of the body and eventually recovered to the normal state ( Fig. 1b; Supplementary video 2). The shape of the notochord became zigzag due to the abnormal body compression along the anterior-posterior axis. Bath application of embryos with picrotoxin at 5 mM but not at lower concentrations caused shrinkage of the body following tactile stimulation. Their notochord was corrugated during abnormal touch response ( Fig. 1c; Supplementary video 3). Embryos treated with nifedipine at 200 μM displayed bilateral muscle contractions, and as a consequence dorsally bent of the body just like strychnine-treated embryos showed ( Fig. 1d; Supplementary video 4). The shape of the notochord was distorted during the response. Likewise, exposure of embryos with 500 μM niflumic acid induced touch-evoked shrinkage of the body and zigzag notochord ( Fig. 1e; Supplementary video 5). These abnormal touch responses in the presence of either four drugs were indistinguishable from touch-evoked behaviors of zebrafish GlyR α1 and GlyRβb mutants 13,14 . These results indicate that our in vivo chemical screening using zebrafish embryos was sufficient to identify potential GlyR inhibitors. The order of drug concentration efficient for inducing abnormal touch response was strychnine (70 μM) < nifedipine (200 μM) < niflumic acid (500 μM) < picrotoxin (5 mM).
Homomeric and heteromeric GlyRs of zebrafish and human origins were formed. To assess the drug sensitivity of GlyRs, we employed electrophysiology using Xenopus oocytes. We first checked glycine-medi- Table 1. List of chemicals and touch-evoked responses of zebrafish embryos. In the first screening, application of either of ten chemicals to zebrafish embryos caused touch-evoked body shrinkage or weird body contraction. The second screening identified thee chemicals as for further analysis. www.nature.com/scientificreports/ ated gating of α1 and β subunits of GlyR of zebrafish and human origins, as these subunits were the major GlyR components in vertebrates 9,10,13,14 . We injected cRNAs encoding zebrafish α1 subunit into oocytes, recorded glycine-dependent currents by two-electrode voltage-clamp and detected a concentration-dependent increase of currents for the calculation of half-maximally activated receptor (EC 50 ) and hill coefficients (Fig. 2a,b). Expression of zebrafish α1 subunit with βb subunit also yielded glycine currents, and the cumulative concentrationcurve was slightly right-shifted compared with the homomeric zebrafish α1 GlyR case. Similarly, injection of human α1 subunit only and α1 subunit with β subunit in oocytes generated functional homomeric and heteromeric GlyRs, respectively. These zebrafish and human GlyRs were activated by micromolar amounts of glycine and their EC 50 s were comparable to those in a previous study 22 (Table 2). Injection of zebrafish βb subunit only or human β subunit only into oocytes did not yield currents (data not shown). These data confirm that zebrafish homomeric α1 and heteromeric α1βb GlyRs, as well as human homomeric α1 and heteromeric α1β GlyRs, were properly formed in oocytes.

Niflumic acid blocked both homomeric and heteromeric zebrafish GlyRs by millimolar amounts.
Since niflumic acid-treated zebrafish embryos as well as strychnine-, picrotoxin-and nifedipinetreated embryos exhibited abnormal touch response just like zebrafish GlyR mutants showed, niflumic acid likely affect GlyRs as strychnine, picrotoxin and nifedipine do. To address whether these drugs inhibit GlyR, we applied these drugs at different concentrations along with 200 μM glycine to Xenopus oocytes that express zebrafish GlyR α1, recorded currents and calculated IC 50 and hill coefficients. In addition to strychnine, picrotoxin and nifedipine, niflumic acid showed concentration-dependent inhibition on homomeric zebrafish α1 GlyRs, suggesting that niflumic acid acts on homomeric α1 GlyRs as an inhibitor (Fig. 3a,b). Similarly, niflumic  www.nature.com/scientificreports/ acid and the other three GlyR inhibitors caused concentration-dependent inhibition of heteromeric zebrafish α1βb GlyRs (Fig. 3c,d). The IC 50 s and hill coefficients of strychnine, picrotoxin and nifedipine were comparable between against homomeric α1 and against heteromeric α1βb GlyRs, while the IC 50 of niflumic acid against homomeric α1 GlyRs were higher than that against heteromeric α1βb GlyRs, indicating that niflumic acid efficiently blocks heteromeric zebrafish GlyRs than homomeric zebrafish GlyRs (Table 3). Although the efficient drug concentrations that induce abnormal touch response in zebrafish embryos was strychnine < nifedipine < niflumic acid < picrotoxin, our electrophysiology suggested the order of IC 50 s as strychnine < picrotoxin < nifedipine < niflumic acid.

Niflumic acid blocked both homomeric and heteromeric human GlyRs at millimolar amounts.
We also investigated whether niflumic acid blocks human GlyRs. Exposure of niflumic acid at different concentrations with 200 μM glycine to Xenopus oocytes that express human GlyR α1 revealed that niflumic acid blocked GlyR in a concentration-dependent manner (Fig. 4a,b). Our electrophysiology confirmed that  www.nature.com/scientificreports/ strychnine, picrotoxin and nifedipine also blocked homomeric human α1 GlyRs. Likewise, niflumic acid and the other three GlyR inhibitors showed concentration-dependent inhibition on heteromeric human α1β GlyRs (Fig. 4c,d). The IC 50 s and hill coefficients of strychnine, nifedipine and niflumic acid were comparable between against homomeric α1 and against heteromeric α1β GlyRs. As reported previously, the IC 50 of picrotoxin against heteromeric human α1β GlyRs were significantly higher than that against homomeric human α1 GlyRs 18 . The order of IC 50 s was strychnine < picrotoxin < nifedipine < niflumic acid. Taken together, these results indicate that niflumic acid is an inhibitor of GlyR.

Discussion
In this study, we employed chemical screening using zebrafish embryos to search for new GlyR inhibitors. Among ~ 1,000 approved drugs and pharmacologically active molecules, niflumic acid was identified as a candidate. We then performed Xenopus oocyte physiology and calculated IC 50 s and hill coefficients of niflumic acid against zebrafish and human GlyRs. Interestingly, the sufficient drug concentrations that induce abnormal touch Table 3. Summary of drug-mediated inhibition properties of homomeric and heteromeric zGlyR and hGlyR. Values represent the mean ± SEM from 10 oocytes each. Where indicated **p < 0.01, ***p < 0.001 reflect difference between homomeric and heteromeric GlyRs.  www.nature.com/scientificreports/ response in zebrafish embryos were strychnine < nifedipine < niflumic acid < picrotoxin, the values of IC 50 of GlyR was strychnine < picrotoxin < nifedipine < niflumic acid.
In vivo chemical screening to identify new GlyR inhibitors. Although many zebrafish-based chemical screenings use transgenic or mutant zebrafish and thus need to select transgene-positive or mutation-carrying embryos/larvae, our screening uses wild-type zebrafish embryos at 48 hpf without any selection, enabling easy sample collection for the screening. In our chemical library, drugs were initially solved in DMSO and eventually diluted to 1/100 in each well of a 96-well plate for treatment. Thus, zebrafish embryos are exposed to drugs in the presence of 1% DMSO. This DMSO concentration facilitates the penetration of drugs into zebrafish body and is harmless. With these advantages, our in vivo drug screening tested ~ 50 drugs in a week and identified strychnine, nifedipine and niflumic acid as potential GlyR inhibitors. Interestingly, we failed to identify picrotoxin in this screening. But this was reasonable because 5 mM concentration of picrotoxin was necessary for inducing abnormal touch response. Since we succeeded in identifying three GlyR inhibitors from ~ 1,000 drugs, we will be able to find many novel ones if we expand the screening scale. While inhibition of GlyR impairs glycinergic transmission that results in the body shrinkage in zebrafish embryos, inhibition of glial glycine transporter 1 (GlyT1) causes synaptic accumulation of glycine and thus potentiates inhibitory glycinergic transmission, leading to reduction of touch response 23 . Although sarcosine, which is GlyT1 inhibitor, was not included in our chemical library, if it was, body contractions with small amplitude could be seen.

Niflumic acid as a GlyR inhibitor. An anti-inflammation drug niflumic acid blocks cyclooxygenase-2 and
is clinically used to reduce joint and muscular pain in rheumatoid 24 . It is also known as a blocker of calciumactivated chloride channels 25 , voltage-gated chloride channels 26 , NMDA receptors 27 and GABA A receptors 28 .
Here we found that niflumic acid blocks both homomeric and heteromeric human GlyRs. Recent studies have investigated that niflumic acid affects human GlyRs expressed in CHO cells with higher blocking potency in homomeric α2 and α3 GlyRs than in homomeric α1 GlyRs 29 . Our electrophysiology results assayed in Xenopus oocytes were consistent with their human GlyR results. Since we could also reveal niflumic acid-sensitivity of zebrafish GlyRs and niflumic acid-induced motor deficits in zebrafish embryos, niflumic acid-mediated GlyR inhibition is assumed to be conserved among vertebrates.

Effective drug concentration for blocking zebrafish GlyRs in Xenopus oocytes and zebrafish embryos.
Our electrophysiology in Xenopus oocytes and touch-evoked response in zebrafish embryos revealed the effective concentration of drugs in vitro and in vivo, respectively. In the case of strychnine and picrotoxin, the effective concentrations for GlyR inhibition in zebrafish embryos (strychnine: 70 μM; picrotoxin: 5 mM) were 1,000 folds higher than the IC 50 values of GlyRs expressed in Xenopus oocytes (strychnine: 70 nM; picrotoxin: 6 μM). This 1,000 fold difference is likely due to the low penetration of drugs through the skin. On the other hand for nifedipine and niflumic acid, the effective concentrations to induce abnormal touch response in zebrafish embryos (nifedipine: 200 μM; niflumic acid: 500 μM) were comparable to IC 50 s of GlyRs in oocyte physiology (nifedipine: 200 μM; niflumic acid: 400 μM). Since both of these drugs are insoluble to water and thus hydrophobic, they can be highly permeable to the skin.

Material and methods
Reagents. All  Video recording of zebrafish touch response. Tactile-evoked zebrafish movements in the presence of control 1% DMSO or drug were video recorded at 48 hpf using a dissection microscope. Touch responses elicited by tactile stimulation delivered to the tail with forceps were captured with a high-speed CCD camera at 200 frames per second (HAS-220, Ditect) as described previously 31 . Embryos were exposed to 70 µM strychnine, 5 mM picrotoxin, 200 µM nifedipine, 500 µM niflumic acid for 30 min before assay.