We previously reported the discovery of a novel ribosomal S6 kinase 2 (RSK2) inhibitor, (R)-5-Methyl-1-oxo-2,3,4,5-tetrahydro-1H-[1,4]diazepino[1,2-a] indole-8-carboxylic acid [1-(3-dimethylamino-propyl)-1H-benzoimidazol-2-yl]-amide (BIX 02565), with high potency (IC50 = 1.1 nM) targeted for the treatment of heart failure. In the present study, we report that despite nanomolar potency at the target, BIX 02565 elicits off-target binding at multiple adrenergic receptor subtypes that are important in the control of vascular tone and cardiac function. To elucidate in vivo the functional consequence of receptor binding, we characterized the cardiovascular (CV) profile of the compound in an anesthetized rat CV screen and telemetry-instrumented conscious rats. Infusion of BIX 02565 (1, 3, and 10 mg/kg) in the rat CV screen resulted in a precipitous decrease in both mean arterial pressure (MAP; to -65 ± 6 mm Hg below baseline) and heart rate (−93 ± 13 beats/min). In telemetry-instrumented rats, BIX 02565 (30, 100, and 300 mg/kg p.o. QD for 4 days) elicited concentration-dependent decreases in MAP after each dose (to −39 ± 4 mm Hg on day 4 at Tmax); analysis by Demming regression demonstrated strong correlation independent of route of administration and influence of anesthesia. Because of pronounced off-target effects of BIX 02565 on cardiovascular function, a high-throughput selectivity screen at adrenergic α1A and α2A was performed for 30 additional RSK2 inhibitors in a novel chemical series; a wide range of adrenergic binding was achieved (0–92% inhibition), allowing for differentiation within the series. Eleven lead compounds with differential binding were advanced to the rat CV screen for in vivo profiling. This led to the identification of potent RSK2 inhibitors (cellular IC50 <0.14 nM) without relevant α1A and α2A inhibition and no adverse cardiovascular effects in vivo.
Myocardial ischemia elicits a series of pathological processes, including intracellular acidosis that arises from anaerobic glycolytic metabolism. Although the cardiac Na+/H+ exchanger (NHE) plays a critical role in the regulation of cellular pH postischemia by removing protons while concomitantly internalizing Na+, what begins as a protective response later transforms into a maladaptive process; elevated cellular Na+ decreases Ca2+ efflux, ultimately leading to Ca2+ overload and cardiac hypertrophy (reviewed in Karmazyn et al., 2005). Inhibitors of NHE have been shown to reverse this sequence of events and have demonstrated efficacy in preclinical models of ischemia-reperfusion injury (Karmazyn et al., 2001) and heart failure (Karmazyn, 2001). Ribosomal S6 kinase (RSK) is an NHE-activating factor and, therefore, a potential point for therapeutic intervention in ischemia-mediated cardiac damage (Avkiran et al., 2008). Indeed, we have previously reported the discovery of (R)-5-Methyl-1-oxo-2,3,4,5-tetrahydro-1H-[1,4]diazepino[1,2-a] indole-8-carboxylic acid [1-(3-dimethylamino-propyl)-1H-benzoimidazol-2-yl]-amide (BIX 02565), a potent RSK2 inhibitor (IC50 = 1.1 nM) targeted for the treatment of heart failure secondary to myocardial infarction through indirect NHE inhibition (Boyer et al., 2011; Kirrane et al., 2011).
Although no off-target effects of BIX 02565 were discovered in early-lead optimization efforts (Kirrane et al., 2011), off-target inhibition of binding at multiple adrenergic receptors (α1A, α1B, α1D, α2A, and β2) was identified when the selectivity of the compound was assessed in a panel of 68 receptors, ion channels, and protein targets. The high degree of adrenergic-specific binding suggested a potential for cardiovascular liabilities due to their importance in the control of vascular tone and cardiac function (Langer and Schoemaker, 1989; McDevitt, 1989). Indeed, both α- and β-adrenergic receptors are established mediators of the direct cardiac and vascular effects of endogenous catecholamines as well as classic pharmacological agents, including phenylephrine and prazosin (α1), clonidine (α2), dobutamine (β1), and terbutaline (β2) (Adams, 1984). Thus, for the present class of compounds, principally aimed for treatment in already cardiovascular-compromised patients, it was deemed essential to rigorously assess off-target adrenergic inhibition during lead optimization, including follow-up assays to assess translation to functional effects in vivo.
In the present study, BIX 02565 was tested in an anesthetized rat CV screen (intravenous infusion) to assess the potential for changes in mean arterial pressure and heart rate and to establish dose response; these studies were followed by assessment of the compound in telemetry-instrumented conscious rats (oral administration) to establish parity between the models and assess the potential influence of route of administration and presence of anesthesia. Additional RSK2 inhibitors were tested in a directed selectivity assay at α1A- and α2A-receptors toward the development of structure-activity relationships (SAR) around adrenergic binding within a novel chemical series; lead compounds were subsequently tested in vivo for the effects on cardiovascular function. We demonstrate a strong correlation between adrenergic binding in vitro and decreases in blood pressure in vivo independent of activity at the target. Results also demonstrate that the focused integration of a small, directed selectivity screen to address the risk for off-target binding, with follow-up studies in a high-throughput in vivo cardiovascular model (anesthetized rat), led to the identification of potent (<1 nM) and selective (no off-target binding) RSK2 inhibitors with no adverse cardiovascular effects in vivo.
Materials and Methods
All experiments were performed under protocols approved by the Boehringer-Ingelheim Institutional Animal Care and Use Committee and according to the United States Animal Welfare Act. All compounds tested were synthesized at Boehringer-Ingelheim Pharmaceuticals, Inc.
Radioligand binding studies were performed at MDS Pharma Services (Taipei, Taiwan). Mean percentage inhibition of specific binding or activity is shown for each assay tested, and in selected assays (for BIX 02565) when inhibition of adrenergic binding generally exceeded 50%, an IC50 was determined by a nonlinear least-squares regression analysis (MathIQ; ID Business Solutions Ltd., UK). The methods for the radioligand binding studies, specific to each assay performed (listed by assay number in Supplemental Table 1S) can be found at: http://pharmacology.ricerca.com/.
Determination of IC50 at RSK2.
Compounds were assessed for their ability to inhibit the phosphorylation of a substrate peptide by RSK2 as described previously (Boyer et al., 2011). In brief, human RSK2 protein (Invitrogen, Carlsbad, CA) was used to measure kinase activity utilizing Kinase GloPlus (Promega, Madison, WI) that uses a luciferin-luciferase based detection reagent to quantify residual ATP. The relative light unit signal was measured on an LJL Analyst (Molecular Devices, Sunnyvale, CA) in luminescence mode using 384 aperture; relative light unit signals were converted to percentage of control; the IC50 was fitted to a standard four-parameter logistic equation.
Determination of IC50 in Cellular Assay of RSK2 Activity.
Compounds were assessed for their ability to inhibit the phosphorylation of the transcription factor cAMP response element-binding protein (CREB) by RSK2 in cells as described previously (Boyer et al., 2011). A monolayer of exponentially growing HLR-CREB cells (PathDetect HeLa Luciferase Reporter CREB cells; Stratagene, La Jolla, CA) was prepared and transfected using Effectene (QIAGEN, Valencia, CA) with RSK2. Cells were plated into 96-well culture plates, and compounds were added 20 to 24 h after transfection. Luciferase expression (48 h) and activity (5 min) were determined using Steady-Glo (Promega) per manufacturer's instructions, and results were represented as the percentage luciferase activity relative to the control measured in the absence of inhibitors (percentage of control); the IC50 was fitted to a standard four-parameter logistic equation. Each data point represents an average of triplicate observations.
Aortic Ring Studies.
Male rats (275–350 g) were euthanized under isoflurane anesthesia. A 1- to 2-inch segment of the thoracic aorta was excised, dissected from connective tissue, and cut into 1- to 3-mm length rings that were mounted onto wire myographs at a resting tension of 2.5 g in 95% O2/5% CO2 Krebs' buffer at 37°C. Rings were allowed to equilibrate for 60 min; rings were constricted with phenylephrine (0.3 μM), and a concentration-dependent dilation was determined with BIX 02565. Force recordings from the myographs during the experiment were acquired using PowerLab and Chart software (100 samples/s; ADInstruments Ltd., Chalgrove, Oxfordshire, UK); values were extracted into Excel and analyzed using XLfit.
Anesthetized Rat Cardiovascular Screen.
The purpose of the rat CV screen was toward rapid screening to establish parity between adrenergic binding and CV effects in vivo to drive compound selection, and thus, only a limited number of animals were tested for each compound (12 compounds, n = 3–4/group), and no statistical analysis was performed. Doses were selected to achieve within half-log unit of drug plasma concentrations of that tested in the adrenergic binding assay (10 μM; target for in vivo studies = 3–30 μM). In brief, male Sprague-Dawley rats (Charles River Laboratories, Inc., Wilmington, MA) were anesthetized with inactin (110 mg/kg i.p.), and a tracheotomy was performed. The femoral artery and veins were catheterized for measurement of MAP and heart rate, for collection of blood samples at 20-min intervals, and for compound infusion/hydration as described previously (Kym et al., 2006; Banfor et al., 2009; Franklin et al., 2009). For each compound tested, vehicle and doses for each infusion period (milligrams per kilograms per 20 min) are listed in Results.
Cardiovascular Profile in Telemetry-Instrumented Conscious Rats.
Mean arterial pressure was assessed in conscious, freely moving male Sprague-Dawley rats (n = 6/group; Charles River Laboratories, Inc.) instrumented with telemetry transmitters (DSI-Ponemah) similar to that described previously (Honore et al., 2009). BIX 02565 (30, 100, and 300 mg/kg p.o. QD) was administered as a solution (10 ml/kg) in a 20% hydroxy-propyl-β-cyclodextran vehicle as described previously (Prakash et al., 2008). Mean arterial pressure was reported from 2 h before (baseline) and 90 h after the first dose; compound was administered at 0, 24, 48, and 72 h. A blood sample was collected from satellite rats (n = 3/group) at 1-h after dose (Tmax) on days 1 and 4 for analysis of plasma drug concentrations by mass spectrometry.
BIX 02565 (also referred to as compound 1) was tested in a panel of 68 receptors, ion channels, and protein targets for off-target inhibition of radioligand binding (Table 1; Supplemental Table 1S). The compound elicited >50% inhibition at adrenergic α1A-, α1B-, α1D-, α2A-, β2-, and imidazoline I2 receptors (IC50 values ranged between 0.052 and 1.820 μM).
BIX 02565 produced concentration-dependent relaxation ex vivo in the phenylephrine-constricted rat aortic ring at concentrations above 0.03 μM with a calculated EC50 of 3.1 μM (Fig. 1). Subsequently, BIX 02565 was infused in the anesthetized rat in a low-dose (0.1, 0.3, and 1.0 mg/kg per 20 min) and high-dose (1.0, 3.0, and 10.0 mg/kg per 20 min) series of continuous infusions to test the effect of compound on hemodynamics in vivo. Plasma concentrations reached up to 0.9 or 21.8 μM, respectively; concentrations at the end of each infusion period are shown within Fig. 2. In the low-dose series, BIX 02565 had no effect on mean arterial pressure or heart rate. In the high-dose series, BIX 02565 elicited decreases in mean arterial pressure (to −18 ± 3, −48 ± 6, and −65 ± 6 mm Hg below baseline) and heart rate (to −15 ± 7, −71 ± 13, and −93 ± 13 beats/min); during the post-treatment period (60–120 min), values remained below those of vehicle controls (Fig. 2, A and B). BIX 02565 was also tested in telemetry-instrumented conscious rats (30, 100, and 300 mg/kg p.o. QD); plasma concentrations of the compound tended to accumulate over the course of the study (Cmax values in the 300 mg/kg dose group increased from 1.7 μM on day 1 to 8.2 μM on day 4 at Tmax). BIX 02565 elicited dose-dependent decreases in MAP after each daily oral dose; on day 4, values reached −2 ± 2, −21 ± 3, and −39 ± 4 mm Hg below baseline at the approximate Tmax at 3 h after dose (Fig. 2C); the compound also elicited decreases in heart rate up to −83 ± 22 beats/min in the 300 mg/kg dose group at the same time point (Supplemental Fig. 1S).
PK/PD analysis and Demming regression suggest a robust correlation between the two cardiovascular models based on change in mean arterial pressure versus plasma concentrations achieved (Fig. 3), demonstrating that the effects were independent of route of administration (intravenously or orally) and independent of the influence of anesthesia. Taken together, these results suggest that the higher-throughput anesthetized rat CV model could be used as an effective screening tool for off-target pharmacology for subsequent compounds.
Thus, 30 RSK2 inhibitors within a novel chemical series were tested in a directed biochemical selectivity screen for inhibition of specific binding at α1A- and α2A-adrenergic receptors (Table 2; Supplemental Table 2S); IC50 values at RSK2 and in a cellular RSK2 activity assay were also determined. Results demonstrate no correlation between binding at α1A and α2A versus in vitro or cellular potency at the target (Fig. 4).
Eleven additional compounds were advanced to the anesthetized rat cardiovascular screen to establish correlation between in vitro binding and effects on cardiovascular function in vivo (Table 3). Plasma concentrations achieved appeared dose-linear for the RSK2 inhibitors tested, and results demonstrate that targeted concentrations were achieved in vivo (within half-log unit of the concentration tested in the in vitro binding assays; Supplemental Fig. 2S). Detailed results from the anesthetized rat cardiovascular screen are shown for four selected compounds (compounds 5, 7, 8, and 12) that span a wide range of percentage inhibition values at α1A- and α2A-receptors (Fig. 5).
The PK/PD relationship for reductions in mean arterial pressure versus plasma concentrations, or plasma concentrations expressed as a function of inhibition of radioligand binding at α1A, was assessed for all RSK2 inhibitors tested in the rat cardiovascular screen. Greater inhibition of radioligand binding at α1A-adrenergic receptors resulted in enhanced decreases in mean arterial pressure in vivo (Fig. 6).
In the present study, we demonstrate that a lead RSK2 inhibitor, BIX 02565 (Kirrane et al., 2011), elicits inhibition of radioligand binding at multiple adrenergic receptor subtypes, an effect that translated into acute hypotension and bradycardia in vivo in both the anesthetized and conscious rat. Follow-up studies demonstrate that the integration of a directed selectivity screen at α1A and α2A led to SAR trends within a novel chemical series and, when tested in vivo, resulted in the identification of novel and potent RSK2 inhibitors without off-target adrenergic receptor binding and without in vivo cardiovascular liabilities.
We previously reported the discovery of BIX 02565, a potent RSK2 inhibitor (IC50 = 1.1 nM) targeted for the treatment of heart failure (Boyer et al., 2011; Kirrane et al., 2011). However, in the present study, when the selectivity of BIX 02565 was assessed in a broad biochemical selectivity screen (68 targets, including receptors, ion channels, and G protein-coupled receptors; Table 1), the compound also elicited off-target inhibition of radioligand binding at multiple adrenergic receptor subtypes (α1A, α1B, α1D, α2A, and β2) and at the imidazoline I2 receptor (IC50 values between 0.052 and 1.820 μM), many of which are important in the regulation of cardiac function and vascular tone. Indeed, we demonstrate that when the cardiovascular effects of the compound were assessed in the anesthetized and conscious rat, BIX 02565 elicited precipitous and dose-dependent decreases in mean arterial pressure concomitant with marked bradycardia. It is noteworthy that we demonstrate a robust correlation between the cardiovascular models at overlapping plasma concentrations of BIX 02565 independent of the route of administration and influence of anesthesia.
Because vasodilators belong to one of the few classes of drugs used in treatment of heart failure shown to reduce mortality from the disease (Wasson et al., 2005; Ghali et al., 2007; Gerc and Buksa, 2010; Cole et al., 2011), BIX 02565 may have indeed been efficacious to reduce progression to heart failure in vivo due to the vasodilatory properties of the compound alone. However, since the off-target pharmacology of BIX 02565 rendered it potentially difficult to dissociate efficacy as a result of off-target vasodilation as opposed to inhibition of RSK2, advancement of BIX 02565 was discontinued.
Parity was established between the in vivo cardiovascular models, and thus, the higher-throughput anesthetized rat CV screen, which necessitated much less compound than the telemetry model (∼30 mg versus 1.3 g for the conscious rat telemetry study), was integrated into the lead optimization efforts within a novel chemical series of RSK2 inhibitors. In addition, to facilitate the rapid identification of novel RSK2 inhibitors without in vitro adrenergic binding, a directed selectivity screen was implemented to assess specific binding at α1A- and α2A-receptors; in total, 31 compounds were tested (see Table 2 and Supplemental Table 3S). The assessment of binding at adrenergic α1A- and α2A-receptors was selected principally because of the finding that another lead RSK inhibitor (compound 8) that was tested in the same 68 binding assays demonstrated significant inhibition at both receptor subtypes but not at α1B-, α1D-, or β2-receptors (full radioligand binding results not shown). Results from the directed selectivity screen identified structurally diverse compounds with a range of inhibition values at α1A and α2A (Table 2; Supplemental Table 2S) that, in concert with in vivo cardiovascular results (summarized in Table 3), led to the development of SAR trends within the chemical series. We identified that affinity for α1A and α2A was largely independent of core substitution, as demonstrated by the unsubstituted compound 2, the geminal dimethyl substituted compound 3, and several related analogs. Furthermore, changes to benzimidazole (compound 4) and azaindole (compound 5) cores did not affect the adrenergic binding or cardiovascular liabilities identified in vivo. However, removal of the basic amine side chain from the benzimidazole amide modestly reduced inhibition of the tested adrenergic receptors but did not completely eliminate the effect in vivo as seen in the ethyl analogs with indole (compound 6) and benzimidazole cores (compound 7) and isopropyl compound 8. Compound 9 contains the only substitution that seemed to successfully modulate inhibition of adrenergic binding, presumably because of the electronic effect of the trifluoroethyl chain, although the compound was not advanced for in vivo translation. Complete benzimidazole replacements, such as azabenzimidazole (compound 10) and the azaindazole (compound 11), were more effective in eliminating nonspecific binding at adrenergic α1A- and α2A-receptors, an effect that was also observed in the benzyl substituted pyrazole analogs (compounds 12 and 13), and the isoxazole analog (compound 14), which translated to a more favorable cardiovascular profile in vivo. These results suggest that the benzimidazole amide acts as a recognition element for binding at adrenergic receptors and was used to guide medicinal chemistry optimization that ultimately led to compound 15, which demonstrated excellent RSK potency (RSK2 IC50 = 0.2 nM; cellular potency = 0.32 nM) without the potential for cardiovascular liability.
In the present study and in both anesthetized and conscious rats, decreases in arterial pressure produced by BIX 02565 may have been expected to elicit reflex tachycardia (via nodal β1 stimulation); however, no relevant tachycardia was observed despite no binding of the compounds at β1-receptors, and in fact, some of the compounds that elicited decreases in blood pressure also produced bradycardia that appeared to correlate well with adrenergic binding. Although the lack of a reflex tachycardia could be at least partially explained in anesthetized rats based on blunted baroreceptor sensitivity secondary to anesthesia, this does not explain a lack of tachycardia in conscious rats, leaving open the possibility that RSK2 itself may directly modulate heart rate. However, since potent and selective RKS2 inhibitors were identified without relevant effects on heart rate at high plasma concentrations, the possibility of a direct RSK2-mediated effect on CV function seems unlikely, although these results do not rule out the possibility of other off-target effects within the chemical series.
Indeed, additional off-target activity may be supported for some early compounds in the series since, as indicated, heart rate actually decreased concomitantly with reductions in mean arterial pressure (e.g., compounds 1, 2, 3, and 5) and also was well correlated with adrenergic binding. However, this effect is presumed to be independent of α1-blocking activity because a prototypical α1-antagonist (e.g., prazosin) would have been expected elicit no effect on heart rate (Khwanchuea et al., 2008) or produce reflex-mediated tachycardia (Luft et al., 1986). In an important publication by Luft et al. (1986) investigating the central and peripheral actions of sympatholytic antihypertensive agents, the authors demonstrated that prazosin administered to rats at doses that produced decreases in MAP had either no-effect on heart rate (2 μg i.v. or i.c.v.) or elicited reflex-tachycardia at higher doses (5 μg i.v.). Although another α1-blocker, urapidil, did elicit bradycardia when administered intracerebroventricularly (but not after intravenous infusion), the effect subsequently was deemed a result of enhanced parasympathetic tone in the animals versus a direct effect at α1-receptors. Thus, although bradycardia was observed by some compounds tested within the chemical series and correlated well with adrenergic binding, the effect is believed to be unrelated to binding at α1-receptors and more likely attributed to one of the many potential regulators of SA nodal pacemaker current (reviewed in Lakatta et al., 2010) at the level of the surface membrane (e.g., membrane clock) or through modulation of intracellular Ca2+ handing (e.g., Ca2+ clock).
In the present study, the compounds were also tested for potency at RSK2 in vitro and in a cellular assay. Results demonstrate no correlation for inhibition of adrenergic binding versus activity at the target, suggesting that the potency of the compounds at RSK2 is independent of off-target adrenergic binding. A subset of the compounds with a range of adrenergic inhibitory activity was advanced to the anesthetized rat cardiovascular model to assess in vivo the functional consequence of differential adrenergic binding on mean arterial pressure and heart rate. Strong correlation was observed and subsequent analyses suggest that potency to inhibit adrenergic binding, but not potency or activity at RSK2, was responsible for the hypotension observed in vivo.
Therefore, in the present study, we demonstrate a focused approach aligning in vitro and in vivo pharmacology and medicinal chemistry efforts toward the identification of off-target effects of a lead RSK2 inhibitor that led to hypotension and bradycardia in vivo. Put into practice, a directed screen at selected adrenergic receptors in parallel with efforts to optimize the potency of the compounds at the target resulted in development of SAR around off-target receptor binding within a novel chemical series while retaining high potency at the target. These efforts resulted in the identification of potent RSK2 inhibitors (RSK2 IC50 and cellular IC50 values <1 nM) without relevant binding at α1A- and α2A-receptors and with no adverse cardiovascular effects in vivo at high plasma concentrations.
Participated in research design: Fryer, Madwed, and Boyer.
Conducted experiments: Muthukumarana, Chen, Smith, Mazurek, Harrington, Dinallo, O'Neill, and Harrison.
Contributed to chemical synthesis: Burke, DiCapua, Guo, Kirrane, Snow, Zhang, Soleymanzadeh, Kashem, and Kugler.
Wrote or contributed to writing of the manuscript: Fryer, Reinhart, and Boyer.
Article, publication date, and citation information can be found at http://jpet.aspetjournals.org.
- Na+/H+ exchanger
- BIX 02565
- compound 1
- ribosomal S6 kinase
- structure-activity relationships
- mean arterial pressure
- once per day
- cAMP response element-binding protein.
- Received October 21, 2011.
- Accepted November 23, 2011.
- Copyright © 2012 by The American Society for Pharmacology and Experimental Therapeutics