Fluorescent probes of the Isoxazole-Dihydropyridine Scaffold: MDR-1 binding and homology model

Abstract

Isoxazole-1,4-Dihydropyridines (IDHPs) were tethered to Fluorescent moieties using double activation via a lanthanide assisted Weibreb amidation. IDHP-fluorophore conjugate 3c exhibits the highest binding to date for IDHPs at the multidrug-resistance transporter (MDR-1), and IDHP-fluorophore conjugates 3c and 7 distribute selectively in SH-SY5Y cells. A homology model for IDHP binding at MDR-1 is presented which represents our current working hypothesis.

For three decades 4-Aryl-1,4-Dihydropyridines (DHPs) have been used as antihypertensive agents which function as antagonists of the L-type voltage gated calcium channel (VGCC).¹ Fluorescent probes of DHPs have been applied to the visualization of the distribution of L-type calcium channels. ²

The DHP scaffold has also been recognized as a privileged structure and used as for identifying lead hits for several molecular targets.^3,4 In this regard the well established structure activity relationship (SAR) of DHP antagonists of the VGCC⁵ can be potentially utilized to design against antihypertensive activity, and hence, selectivity for non-calcium channel targets.⁶

The observation that DHPs nicardipine⁷ and niguldipine⁸ exhibit activity as inhibitors of the multidrug-resistance transporter (MDR-1, also known as p-glycoprotein or P-gp) efflux pump,⁹ has stimulated renewed interest in DHPs.¹⁰ Pharmacophore modelling for MDR-1 indicates several common features for inhibitors,¹¹ notably at least two aryl pockets, and a tether of approximately 3–5 methylenes, leading to another functional group which possesses H-bond donor and lipophilic groups.¹² In contrast, substrates for MDR-1 usually contain a group located near the latter distal end of the tether which is protonated at physiological pH. Of interest to us are the recent observations by the groups of Andrus, and Chmielewski and Hrycyna that bivalent molecules can have significant activity as inhibitors, and that consideration of the tether connection is critical.¹³

Our continuing interest in the bioisosteric 4-isoxazolyl-DHPs^5b,14 (IDHPs) led us to examine a common pharmacophore for DHP MDR-1 inhibitors (Figure 1), in light of the similarity of the x-ray structure of Isoxazolyl-DHP 1¹⁵ with the general MDR-1 inhibitor model proposed by Ecker,¹² the most striking similarity being the anchor-like conformation of the DHP relative to the C(3) aryl moiety. Recently, we found that IDHPs exemplified by structure 1 are in fact robust ligands of MDR-1,²⁴ in studies conducted in collaboration with the Psychoactive Drug Screening Program (PDSP), and exhibit an SAR distinct from their activity at the VGCC. In the present study we wished to test the hypothesis as to whether bivalent IDHPs would have enhanced activity at MDR-1.

Figure 1.

SYBYL common pharmacophore model of nicardipine (cyan) niguldipine (pink), and 3-[4,N-dansyl-1',4'-butanediamino-]carboxamido-4-Isoxazolyl-DHP 3a (green).

Double activation preparation of bivalent fluorescent-labeled Isoxazolyl-DHPs (IDHPs)

The Isoxazolyl-DHPs (1) can be prepared easily on a multi-gram scale using methods previously reported by our lab,^5b,14–16 however, the direct reaction of the esters is sluggish given their delocalized vinylogous urethane nature. We have found that modest yields can be obtained for 2 using our double activation method,¹⁷ which uses a lanthanide catalyst in the presence of a Weinreb activated amine.¹⁸ The resulting tethered amines can be transformed to fluorophore-labeled analogs quite effectively. Thus, a series of probes with differing tether lengths (3a–d), and diverse fluorophores (3–7) were readily available. The method, of course, is not limited to isoxazolyl-dihydropyridines, and the o-, m-, and p- bromophenyl DHPs were also converted to dansyl analogs (Supplementary Material).¹⁹

The fluorophores 4–7 were selected to study a range of reactivities in the presence of the IDHP moiety. The 9-Anthryl adduct 4 was prepared by reductive amination with sodium cyanoborohydride from amine 2 and anthracene-9-carboxaldehyde.²⁵ Pyrene-2-carboxylic acid did not give synthetically useful yields with dicyclohexy carbodiimide (DCC), likely owing to steric hindrance, however the corresponding Schotten-Bauman reaction with the derived acid chloride proceeded to the amide 5 quite effectively. Fluorescein isothiocyanate (FITC) gave thiourea adduct 6. Finally sulforhodamine B, similar to the danzyls, proceeded virtually quantitatively to the adduct 7.

From a microscopist’s perspective, an important consideration in synthesis of a library of fluorophore-conjugated compounds is the extinction coefficient of the fluorophore groups. The extinction coefficient defines how strongly a substance absorbs light at a particular wavelength. Hence, a fluorophore with a high extinction coefficient will enable a microscopist to image a fluorescently labeled compound at a lower, more physiologically relevant concentration than a compound labeled with a fluorophore with a low extinction coefficient. Extinction coefficients of the fluorophores listed in Table 1, span a range from 4400 for dansyl through 85,000 for fluorescein and approximately 120,000 for the sulforhodamines.

Table 1.

Bivalent fluorescent IDHPs prepared for the current study 3–7.

Compds	Fluorophore	Tether length	Empirical Formula	Mass Spectra (m/Z)
3a	Dansyl	4	C37H43O6N5S	685[M]+
3b	Dansyl	6	C39H47O6N5S	713[M]+
3c	Dansyl	8	C41H51O6N5S	741[M]+
3d	Dansyl	12	C45H59O6N5S	797[M]+
4	9-Anthryl	6	C42H46N4O4	669[M-1]+
5	2-Pyryl-carboxy	6	C44H44N4O5	708[M]+
6	Fluorescein	6	C48H47N5O9S	869[M+1]+
7	SulforhodamineB	6	C54H64N6O10S	1021 [M]+

^a3a to 3d were obtained by FAB MS on a JEOL-JMS-AX505HA.

^b4–7 were obtained by HR-ESMS, Micromass MS / Waters HPLC.

Conformational Dynamics of the Fluorescent Probes

The conformational dynamics of IDHPs dramatically effect the topology presented to a biomolecular target.^{5b,14–16,24} The danzyl probes, most notably 3c, in addition to the expected set of signals for the isoxazolyl-DHP, also exhibited a second set of signals with pronounced magnetic anisotropy in typical organic solvents (i.e., CDCl₃), and of equal intensity. Nuclear Overhauser spectroscopy evidenced interactions between the DHP C(2) and C(6) methyls with the Isoxazolyl C(3) phenyl, and the Isoxazolyl C(3) phenyl with the dimethyl amino groups of the dansyl moiety, indicative of a folded conformation. Raising the temperature resulted in a conformational reorganization to a single set of chemical shifts which were consistent with an extended conformation.¹⁹ The interconversion was not reversible in wet DMSO. Computation indicated that the folded conformation was lower in energy, in the gas phase, consistent with its predominant presence in less polar solvent. The VT and computation are shown in the Supplementary Material.

Multidrug-Resistance Inhibition (MDR-1)

MDR-1 inhibition was evaluated in the Psychoactive Drug Screening Program (PDSP) assay, the details of which are given in the Supplementary Material, and described on the PDSP web site http://pdsp.med.unc.edu/pdspw/MDR.php. Compound 3c exhibited 63% inhibition, verses 0% for untreated cells and 100% for the MDR inhibitor cyclosporin.^21,22 This represents a significant increase in MDR-1 binding compared to IDHP-1, which exhibited a value of 48.9% in the PDSP assay.^23,24

Homology Modeling

In the absence of a homo sapien X-ray crystal structure of MDR-1 a homology mode was constructed using Mus musculus MDR-1, PDB:3G5U,²⁶ as a template for threading of the human gene sequence (P08183). ^27–29 The sequence alignment was generated using Clustal W,³⁰ the alignment of and the subsequent threading indicated high sequence homology to the human sequence with a sequence identify of 82.34%. To visualize the hypothetical interactions between 3c and MDR-1 molecular modeling was conducted. Ligand structures were drawn and energy minimized (Powell method, 0.01 kcal mol-1 Å-1 gradient termination, MMFF94s force field, MMFF94 charges, 1000 maximum iterations) using the SYBYL software package (Tripos, St. Louis, MO). Virtual dockings of energy-minimized ligand to the MDR-1 homology model were performed using the GOLD software package (Cambridge Crystallographic Data Center, Cambridge, UK)³¹ and scored using Chem PLP with default settings. Docking algorithms were performed with the constraint of limiting the allowed binding area to a 6Ǻ radius around the ARG905 residue. Ligand-receptor ensemble structures were obtained by merging the highest-ranked output ligand orientation structures with the input MDR-1 homology model structure using PyMOL. An MDR-1 human homology model protein active site analysis was performed using Q-site finder.³³ This program binds hydrophobic probes to the MDR-1 homology model and ranks sites with the most favorable binding energy. These clusters are then placed in rank order of the likelihood of being a binding site according to the sum total binding energies for each cluster. The first binding cluster was bound in the rhodamine binding site or R-site, the second and third sites were located proximal to the nucleotide binding domains (NBDs). The overlap of the dexniguldipine binding site in MDR-1 via photoaffinity labeling (shown in red in Figure 2),³² combined with the output of the Q-site finder gave a consensus binding site for IDHP. ³³

Figure 2.

Homology model of IDHP 3c (gold) docked into MDR-1 based on pdb accession number 3G5U. The red area is the dexniguldipine photoaffinity site.

Interestingly, the predicted binding location of the IDHP resides near the conserved V907 of Coupling Helix 2, and an essential Mg²⁺ binding residue Q475 of the Walker domain required for ATP hydrolysis. A ligand binding in this location would be expected to limit conformational reorganization necessary for xenobiotic efflux. This presents a testable working hypothesis for future studies, and target design.

Fluorescent Microscopy

SH-SY5Y human neuroblastoma cells (American Type Cell Culture Cat no. CRL-2266) lines were cultured in the recommended media including 10% fetal bovine serum and antibiotics. Cells were plated on coverslips in 12-well plates at 1X10⁵/well and allowed to adhere for 48 h. Cells were incubated with DHP-sulfonorhodamine diluted in dimethyl sulfoxide (DMSO) for a prescribed period of time (30 min to 21 h) at 37°C/5% CO₂. After incubation, media was removed, cells were washed twice with phosphate-buffered saline (PBS), and fixed in 4% paraformaldehyde in PBS (20 min, RT). Fluorescent images were taken with a Nikon Eclipse E800 fluorescent microscope and images were processed with Nuance 1.6.2.368 alpha software (Cambridge Research Institute).

Representative results are shown in Figure 4. Dansyl isoxazole 3c and Sulforhodamine isoxazole 7 possess fluorophores with radically different physical characteristics, yet they localize in SH-SY5Y cells in virtually identical locations, whereas controls for the fluorophores not conjugated to the IDHP showed nonselective binding throughout the cell. This suggests that it is the IDHP moiety which is the driving force for the localization. The characteristics of the fluorophores were effected by the tether. For example, the sulforhodamine control alone had higher intensity (albeit no selectivity in cellular distribution) compared to the conjugate 7, with a tether of six methylene units. In contrast the fluorophore conjugate 3c, with eight methylenes had a higher intensity than the corresponding control (dansylcadaverine), which is consistent with the former being inside of the Forster radius and exhibits quenching relative to fluorophore alone, whereas the latter with the longer tether is beyond this radius 4-Aryl-DHPs tethered via an ester linkage to the fluorophore BODIPY localized in the L-type calcium channel in a functional, extracellular heparin-binding site.² It has previously been reported that expression of MDR-1 is reported to be localized at several subcellular sites, including the nucleus, mitochondria, and Golgi apparatus of rat brain cells and suggests a role in multidrug resistance at subcellular sites.²⁰ The current study suggests an analogous distribution in SH-SY5Y cells, and will be the subject of future studies.

Figure 4.

Confocal fluorescence Microscopy of SH-SY5Y cells with: (Left), 10 μM Dansyl Isoxazolyl-DHP 3c (green); and (Right) 1 μM Sulforhodamine B-isoxazole DHP 7 (orange), imaged at 60×, with a 90 minute incubation time. Scale bar indicates a distance of 10 μm.

Summary

Isoxazolyl-DHPs can be readily conjugated to fluorophores, once tethering moieties are connected using a further application of our double activation methodology. A wide variety of fluorophores can be conjugated to the IDHP using this approach. The localization of the conjugates in SH-SY5Y cells appears to be driven by the IDHP moiety. The distribution of the present conjugates are not consistent with previous reports of L-type calcium channels,² and appear to concentrated in the nuclear envelope and the Golgi.²⁰ We have developed a homology model of MDR-1, which combined with our experimental observations, represents our current working hypothesis. Current studies are focused on whether the IDHPs may represent useful tools for the study of MDR-1, and ultimately lead to clinically useful inhibitors. Our progress will be reported in due course.

Figure 3.

Close up of IDHP 3c docked into MDR-1 The fluorophore interacts at the edge of the red dexniguldipine photoaffinity site, the DHP moiety of 3c (gold) is mainly located near the NBD, as predicted by Q-site.

Scheme 1.

Synthesis of Isoxazole-DHP fluorophore conjugates 3–7.

Scheme 2.

Summary of close contacts between 3c and MDR-1 from docking the homology model.