Jiyong
Lee‡
a,
M. Muralidhar
Reddy
a and
Thomas
Kodadek
*b
aDivision of Translational Research, Departments of Internal Medicine and Molecular Biology, University of Texas Southwestern Medical Center, 5323 Harry Hines Blvd, Dallas, TX 75390, USA
bDepartments of Chemistry & Cancer Biology, The Scripps Research Institute, Scripps Florida, 130 Scripps Way, Jupiter, FL 33458, USA. E-mail: Kodadek@scripps.edu; Tel: +1-561-228-2461
First published on 12th May 2010
The orexin neurohormones control a variety of important physiological processes by signaling through two related G protein-coupled receptors, including appetite and feeding, wakefulness and energy homeostasis. Pharmacological manipulation of orexin signaling is an important goal. Here we describe the isolation of orexin receptor ligands from a library of microarray-displayed peptoids via a novel two-color, cell-based screen. Functional analysis of derivatives of these “hits” resulted in the development of moderate potency, low molecular weight receptor antagonists. Moreover, further optimization efforts resulted in the fortuitous discovery of a compound that positively potentiates the activity of the receptor. This compound is the first small molecule reported to up-regulate orexin signaling.
Not surprisingly, there has been considerable interest in the development of compounds with which to manipulate orexin signaling. Non-peptide antagonists of OXRs have been reported by a number of groups (for a review, see ref. 8) and show promise in the treatment of insomnia.9 They may also be useful for the treatment of panic anxiety attacks5 and drug addiction,10 though no clinical trials for these indications have been carried out. On the other hand, activation of orexin signaling will be required to treat narcolepsy. The majority of human narcoleptics lack orexin-producing neurons, possibly due to an autoimmune attack on these cells,11 so in this case a full agonist will be required as a therapeutic lead.12 Based on the recent findings of Funato et al.7 an OXR2 receptor agonist might be useful in treating diet-induced obesity and diabetes. Since diabetics presumably produce normal physiological levels of orexin, a positive allosteric potentiator of receptor signaling might be interesting as well for this indication since it would hyperactivate orexin signaling in response to the native hormone.
To the best of our knowledge there have been no published reports of non-peptidic small molecules capable of stimulating orexin signaling. In this edge article, we describe the discovery of the first positive allosteric potentiator of the orexin receptors. This molecule resulted from the analysis of derivatives of modest potency orexin receptor antagonists that were isolated from a novel microarray-based screen.
Fig. 1 A microarray-based, two-color, cell-binding screen to isolate ligands of human OXR1. (A) Light microscopic images of cells binding to a peptoid microarray. Scale bar = 250 μm. (B) Schematic illustration of the screening procedure. (C) Representative superimposed images (red and green) of cells on a microarray after washing. (D) Ratiometric analysis of microarray images. The ratio of red to green fluorescence on 99 spots that displayed above background signal is shown. The unfilled bar at the far right of the graph represents the mean of the 99 spots. Bars marked with an asterisk represent peptoids that were subjected to sequence analysis by tandem mass spectrometry. Unambiguous sequences were obtained for peptoids marked OBP1–OBP5 (see ESI†). F635 and F532 are mean fluorescence emission intensities of spots with excitation wavelengths of 635 nm and 532 nm, respectively. B635 and B532 are background fluorescence emission intensities with 635 nm and 532 nm, respectively. |
To effect this strategy, we first examined if microarray-displayed peptoids17 are capable of binding to cells stably expressing human OXR1 (HEK293/hOXR1).18 The cells were incubated with a peptoid array (Fig. S1†) for one hour after pre-blocking the array with 3% BSA in DMEM to inhibit non-specific binding. The slide was then washed with PBS, and the binding of the cells to the array was examined under a microscope. As shown in Fig. 1A, circular patterns of cell monolayers were observed at certain points on the array and the diameters of the monolayers were similar to those of the printed peptoid spots (200–300 μm), suggesting binding of cells to peptoids spotted at these positions. This experiment suggested that the hybridization conditions were appropriate to assay for peptoid–receptor interactions.
HEK293 cells, which do not detectably express orexin receptors (as determined by Western blotting and quantitative RT-PCR; data not shown), were stained with SYTO 85 (green), and HEK293/hOXR1 cells were stained with SYTO 60 (red). After mixing the cells in an approximately 1:1 ratio, they were applied to a peptoid microarray displaying 5760 different 9-mers and incubated for 1 h at 37 °C. After washing followed by fixation, the slide was scanned at 635 nm to visualize HEK293/hOXR1 binding and 532 nm to visualize HEK293 binding. The two images were then superimposed (Fig. 1B). We hypothesized that spots showing a high ratio of red over green fluorescence display peptoids that bind to OXR1 specifically. After measuring the level of fluorescence in each channel (532 nm and 635 nm, respectively) using a standard microarray scanner, we identified 99 spots displaying above background fluorescence in one or both channels (Fig. S2†). Some representative images of these spots are shown in Fig. 1C. Calculated ratios of the fluorescence intensities in the red and green channels are illustrated in Fig. 1D. The mean fluorescence ratio (635/532) of the 99 spots was 5.9. The fact that this ratio was not 1.0 was due to the optics of the scanner. We measured the fluorescence emissions ratio with 635 and 532 nm excitation wavelengths of a known 1:1 mixture of the differentially labeled cells and measured a ratio of 6.1. Thus, we concluded that the ratio of 5.9 reflects equal binding of the two cells on most of the features of the microarray. The twelve peptoid features displaying the highest red/green ratio above the mean were chosen as possible OXR1 ligands. Tandem MALDI mass spectrometry (using compound from the solution used to spot the microarray) provided unambiguous structures for five of these peptoids, which we named orexin receptor binding peptoids (OBPs; see Fig. S3†).
As shown in Fig. S5A,† none of the OBPs showed agonist activity, but some weak antagonist activity was observed (Fig. S5B†). For example, OBP1 showed an approximately 30% inhibition of orexin A-induced cAMP elevation at 300 μM. The activity of OBP2 was not quantified since it was cytotoxic to both HEK293 and HEK293/hOXR1 cell lines. The weak activity of OBP1 as an antagonist shows that the microarray-based binding assay is capable of registering even modest affinity receptor-binding molecules, probably because of avidity effects.
To further address the selectivity of OBP1 binding to OXR1, we utilized a different assay. OBP1 on Tentagel beads (TG-OBP1, Fig. S6†) were pre-incubated with 3% BSA in DMEM and then incubated with SYTO 60 stained-HEK293/hOXR1 cells or SYTO 60 stained-HEK293 cells, washed, and visualized by fluorescence microscopy. As shown in Fig. S7,† HEK293/hOXR1 cells showed binding to TG-OBP1 while HEK293 cells did not. This indicates that the peptoid does not bind stably to any other molecule on the surface of the HEK293 cells. We cannot rule out potential functional effects on other receptors without further study.
To determine if the peptoid recognizes the hormone-binding site of the receptor or some other surface of the extracellular region of the protein, we carried out a competition experiment. Whereas binding of HEK293/hOXR1 cells to TG-OBP1 was abolished by free OBP1, the addition of excess orexin A did not block binding of the cells to the immobilized peptoid (Fig. S7†). This suggests that OBP1 binds to a site distinct from that recognized by orexin itself and that it acts via an allosteric mechanism.
The low potency of OBP1 clearly must be improved for it to be a useful reagent. As a first step towards this goal, we sought to identify the minimal pharmacophore in the peptoid. Derivatives of OBP1 were synthesized in which each side chain (R) was replaced, in turn, with a methyl group (Fig. 2A and Fig. S8†).21 As shown in Fig. 2B, placement of the methyl group at the fifth and sixth positions of the peptoid (compounds OBP1-5 and OBP1-6) resulted in a complete loss of antagonist activity. Likewise, Tentagel beads displaying a peptoid in which both the Nmba and Npip residues of OBP1 were replaced with methyl groups did not show any binding to HEK293/hOXR1 cells (Fig. S7†). Finally, a truncated form of OBP1 (OBPt) containing only these two residues was found to exhibit antagonist activity equal to or even slightly better than the parent OBP1 peptoid (Fig. 2C). These results argue that the Nmba and Npip residues in the OBP1 hit comprise the minimal pharmacophore.
Fig. 2 Pharmacophore identification via “sarcosine scanning”. (A) Each side chain (R) of the Nth residue was replaced, in turn, with a methyl group to afford OBP1-N (where N = 1–9), sarcosine containing peptoids. (B) Effects of sarcosine replacements on the antagonist activity of OBP1. The y axis shows the measured increase in cAMP concentrations in the cells relative to cells not treated with orexin (i.e., treatment with 0.5 μM orexin results in a four-fold stimulation of cAMP production). (C) Chemical structure of truncated OBP1 (OBPt) and its antagonist activity. Error bars represent the standard deviation of the mean of triplicate experiments. |
We noticed some structural similarity between OBPt and ACT-078573 (also known as Almorexant; see Fig. 3A), an orexin receptor antagonist developed by Actelion Pharmaceuticals.9 Based on this purported similarity, we hypothesized that appending a hydrophobic unit to the N-terminus of OBPt would place this group in approximately the same region of space as the CF3-substituted aryl ring in ACT-078573 and improve binding. The N-terminal secondary amine of OBPt was benzylated or benzoylated to afford OBPt-1 and OBPt-2, respectively (Fig. 3A). OBPt-1 was found to block orexin-A-induced cAMP elevation much more efficiently (IC50 = 20 μM; Fig. 3B) than OBPt. OBPt-2 showed a weaker, but still improved, activity (IC50 = 55 μM). Similar results were obtained by monitoring ERK phosphorylation, a known downstream mediator of OXR1 (Fig. 3C).19
Fig. 3 Effect of the introduction of hydrophobic groups at the N-terminus of OBPt on activity. (A) Comparison of the structures of ACT-098573, a potent antagonist of OXR1 and OXR2, and OBPt. The structures are oriented to illustrate the hypothesis that addition of an aryl substituent on the N-terminal nitrogen of the peptoid (marked with a filled arrow) would fill space that is occupied by the trifluoromethylphenyl-containing side chain in ACT-098573. (i) Benzaldehyde, BAP; (ii) benzoic acid, DIC, HOAt. (B) Antagonist activities of OBPt-1 and OBPt-2. Increasing concentrations (2, 20, 30, 50, and 75 μM) of peptoids were used. CON is an N-benzylated control peptoid (see Fig. S9†). Error bars represent the standard deviation of the mean from triplicate experiments. (C) Inhibition of orexin-mediated ERK phosphorylation by OBPt-1. A representative figure from three different experiments is shown. |
To identify even better compounds, a small library of additional derivatives was synthesized in which positions R1 through R4 (Fig. 4A) were varied. Detailed information about library construction and determination of the structure–activity relationships will be reported elsewhere. We found that OBPt-3 and OBPt-4 were more potent OXR1 antagonists than OBPt-1, with IC50 values of 4.5 μM and 15.1 μM, respectively (Fig. S10†). We then synthesized a derivative combining the two substitutions in OBPt-3 and OBPt-4 that distinguished them from OBPt-1 to afford OBPt-5. As shown in Fig. 4B, OBPt-5 showed increased potency (IC50 = 1.7 μM). The potency of the enantiomer OBPt-6 was almost identical (IC50 = 2.9 μM). Finally, we found that OBPt-5 also antagonized OXR2, which is 64% identical to OXR1 (Fig. 4C). OBPt-5 did not directly interfere with forskolin (adenyl cyclase activator)-induced cAMP production (Fig. S12†), which does not depend on orexin signaling. This argues that the antagonist effect of OBPt-5 was not due to some receptor-independent activity.
Fig. 4 Attempted optimization of OBPt-1. (A) Chemical structures of some of the OBPt-1 derivatives examined. (B) Antagonist activities of OBPt-5 and OBPt-6. (C) Effect of OBPt-5 on orexin A- or orexin B-induced OXR2 activation of HEK293 cells expressing human OXR2. SB408124 is an OXR1 selective antagonist.20 Proline bis-amide is an OXR1/2 dual antagonist22 (Fig. S11†). |
Fig. 5 Discovery of a positive allosteric potentiator of the orexin receptors. (A) Chemical structures of tested compounds (OBPt-7, OBPt-8, and OBPt-9). (B) Effects of the compounds on the response (cAMP elevation) of OXR1-expressing cells to an EC20 concentration of orexin A. The level of cAMP elevation by 0.3 μM orexin A (the EC100 concentration) is also shown for comparison. (C) Concentration–response curves of OBPt-9 and CON on cAMP elevation of OXR1-expressing cells in the presence of an EC20 concentration of orexin A. (D) Concentration–response curves of orexin A on cAMP elevation of OXR1-expressing cells in the presence or absence of OBPt-9. |
We next determined the effect of OBPt-9 on the potency and efficacy of orexin A. Cells ware pre-incubated with OBPt-9 or DMSO (vehicle) and subsequently stimulated with increasing concentrations of orexin A. As shown in Fig. 5D, OBPt-9 induced a leftward and upward shift of the orexin A response curve. The EC50 value for orexin A in the presence of vehicle was 41 nM, whereas, the EC50 values were 27 and 12 nM in the presence of 0.1 μM and 5 μM of OBPt-9, respectively. Moreover, the maximal response to orexin A was about 2-fold higher in the presence of OBPt-9 (5 μM) than with vehicle alone.
Next, we examined if OBPt-9 can also potentiate the response of OXR2 to orexin A. After transient expression of OXR2, HEK293 cells also carrying a receptor-driven reporter gene were treated with OBPt-9 and then with increasing concentrations of orexin A. As depicted in Fig. S14,† OBPt-9 showed a similar potentiation pattern as was observed with the OXR1-containing cells. The orexin A EC50 was 60 nM in the presence of vehicle and 22 nM in the presence of OBPt-9. Moreover, the maximum level of reporter gene activation in the presence of OBPt-9 was almost twice that observed in the presence of vehicle. We also found that OBPt-9 did not affect ATP-dependent, endogenous P2 receptor-mediated ERK phosphorylation in HEK293 (Fig. S15†). This suggests the orexin receptor potentiation activity of OBPt-9 is not due to some non-specific cell surface receptor activation.
Recently, we reported a similar two-color, cell-based screen carried out on Tentagel beads that allows up to several million peptoids to be screened simultaneously.15 While the microarray platform is limited to a few thousand compounds, it has the advantage of allowing facile and quantitative comparisons of the binding properties of all of the compounds on the array.23 Thus, while we demonstrated here the utility of microarrays for screening primary, naïve libraries, its most useful application may be in evaluating libraries of derivatives of primary hits in the search for improved ligands. Experiments along these lines are underway.
Importantly, we have discovered the first small-molecule allosteric potentiator of the orexin receptor. Allosteric potentiators bind to a site on the receptor distinct from that of the native ligand and accentuate the response of the receptor to that ligand, but cannot stimulate receptor function independently. It has been suggested that allosteric potentiators might have advantages over classical orthosteric agonists from the therapeutic point of view. For example, allosteric potentiators would not drive chronic receptor activation, but rather accentuate natural cycles of activation of the receptor.24,25 Animal experiments to test the utility of these compounds in vivo are underway.
Footnotes |
† Electronic supplementary information (ESI) available. Supplementary tables and figures. See DOI: 10.1039/c0sc00197j |
‡ Present address: Department of Chemistry, The Scripps Research Institute, La Jolla, CA 92037, USA. |
This journal is © The Royal Society of Chemistry 2010 |