High-affinity fluorescent ligands for the 5-HT(3) receptor.

The synthesis, photophysical and biological characterization of a small library of fluorescent 5-HT(3) receptor ligands is described. Several of these novel granisetron conjugates have high quantum yields and show high affinity for the human 5-HT(3)AR.

Fluorescence is a useful tool in cell biological studies, and fluorescent labeling of target proteins have enabled numerous in vivo studies of protein function. Whilst it has become common practice to genetically fuse a fluorescent protein, such as GFP, to a protein of interest, the large size of such fluorescent proteins (e.g., GFP: 238 aa, 27 kDa) can affect the structure and function of the target protein. Ion channels and transmembrane receptors typically contain several α-helices which are connected via short peptide loops and, given that these proteins undergo large conformational changes, they offer few fusion sites for large fluorescent proteins. An alternative to fusion is to use a low-molecular weight ligand conjugated to a fluorophore. High-affinity fluorescent ligands can be used to visualize the receptor of interest in cells, and also have potential as tracer compounds in fluorescence polarization, and flow cytometry applications that target specific receptors and ion channels.

5-HT3Rs are members of the Cys-loop family of ligand-gated ion channels which also includes nACh, GABAA, and glycine receptors. These transmembrane proteins enable rapid synaptic transmission in the central and peripheral nervous system and are composed of five pseudosymmetrically arranged subunits surrounding a central ion-conducting pore. The neurotransmitter binding sites are located in the extracellular N-terminal domains at the interface of two adjacent subunits. Five 5-HT3R subunits have been discovered to date (5-HT3A–5-HT3E) which led to the conclusion that 5-HT3R populations most likely comprise several subtypes characterized by distinct functional properties; thus human 5-HT3 signaling is more complex than originally anticipated. To date, only homomeric 5-HT3A and heteromeric 5-HT3AB receptors have been extensively characterized in heterologous systems. Recent mutagenesis and cysteine modification studies indicate that agonists and antagonists bind to an A–A interface both in human homomeric 5-HT3A and heteromeric 5-HT3AB receptors which is consistent with their observed identical competitive pharmacologies. Nonetheless, the discovery of subtype-selective molecular tools for the study of 5-HT3R populations in native cells and tissue remains an important goal.

Antagonists of 5-HT3Rs are used in the clinic to prevent chemotherapy- and radiotherapy-induced nausea and vomiting, post-operative nausea and vomiting, for the treatment of irritable bowel syndrome, and 5-HT3R antagonists might be beneficial for the treatment of psychiatric and neurological disorders, such as anxiety, drug dependence and bulimia nervosa. They have also been shown to reduce pain in certain conditions including rheumatoid arthritis, fibromyalgia and migraine.

As a complementary approach to conventional biological methods such as site-directed mutagenesis, radioligand binding and electrophysiology, we are developing biophysical small-molecular probes to investigate the structure and function of ligand-gated ion channels. Previously, we undertook a SAR study of the high-affinity competitive 5-HT3R antagonist granisetron (1, Fig. 1, Ki = 1.45 nM), and identified positions on the granisetron core which were tolerant to substitution. In this initial study we discovered that methoxy-substituted granisetrons 2–5 are quite fluorescent and some of them bound with high affinity to the 5-HT3R. However, their quantum yields were poor (Table 1). We subsequently conjugated a commercial fluorophore, BODIPY FL, to the N1-position of granisetron and obtained high-affinity probe 6, which had much higher fluorescence intensity, and was used to visualize recombinant 5-HT3ARs in mammalian cells. Unfortunately, 6 gave high fluorescence background in gut preparations and primary neurons that could not be washed out. These limitations prompted us to generate a small library of fluorescent 5-HT3R ligands with improved properties. Herein, we describe the synthesis, photophysical and biological characterization of these novel granisetron–fluorophore conjugates.

Figure 1

Reference compound and fluorescent granisetron derivatives.

Table 1

Photophysical properties of fluorescent granisetron probes and their binding affinities for the human 5-HT3AR

Probe	λmax Abs (nm)	ε (M−1 cm−1)	λmax Em (nm)	Φf	Ki (nM)
					mean ± SEM
1	—	—	—	—	1.45 ± 0.13a
2	298b	10,900b	388b	0.01b	26 ± 7c
3	326b	—	383b	0.02b	5,300 ± 200c
4	300b	—	375b	0.01b	3,000 ± 1,000c
5	302b	7,800b	413b	0.03b	71 ± 8c
6	504b	89,200b	511b	0.61b	2.8 ± 0.7c
14	536b	8,700b	573b	0.48b	0.9 ± 0.3
15	432d	37,100b	476d	0.04d	199 ± 39
16	495d	13,500b	518d	0.66d	1.6 ± 0.3
17	432d	78,000b	471d	0.05d	7.3 ± 2.5
18	493d	18,600b	519d	0.69d	1.1 ± 0.2
19	465b	22,600b	536b	0.13b	NB
20	431d	71,900b (29,000d)	480d	0.04d	142 ± 5
21	498d	17,700b (70,700d)	524d	0.68d	6,300 ± 1,600
22	557d	53,100d	584d	0.24d	384 ± 68
26	465b	20,400b	540b	0.10b	8.7 ± 5.8
27	350b	11,500b	524b	0.29b	1.2 ± 0.6
28	497b	78,800b	505b	0.44b	1.6 ± 0.8
36	430d	40,100b	479d	0.02d	1.9 ± 0.9
37	430d	40,600b	478d	0.05d	157 ± 9
38	437d	21,800d	484d	0.06d	1,300 ± 300
39	498d	44,200d	519d	0.19d	208 ± 55

From Ref. 16.

In MeOH.

From Ref. 13.

In phosphate buffer pH 7; —, not attempted; NB, no binding.

The synthesis of the N1-conjugated compound series is depicted in Schemes 1–3. Amide 7 was first N-alkylated with either aliphatic aminopropyl-, aminobutyl- or more polar PEG-linker building blocks 8–10. Subsequent Boc-deprotection liberated the primary amino group that was used to couple various fluorophores F1–F5 to the granisetron-linker constructs 11–13 (Scheme 2). For the synthesis of probes 26–28 it was more advantageous to couple the linker to the fluorophore first, and then perform the N1-alkylation in the second step (Scheme 3). Unexpectedly, when we condensed 4-bromobutanoyl chloride with 2,4-dimethylpyrrole following a literature protocol we obtained BODIPY dye 25 as the sole product where two moles of acyl chloride have reacted with two moles of the pyrrole. The structure of 25 was confirmed by full spectroscopic characterization and a crystal structure.

Scheme 2

Synthesis of granisetron probes 14–22. Reagents and conditions: (a) for 14: 11, i-Pr2EtN, BODIPY TMR-X succinimidyl ester, DMF, rt. For 15, 17 and 20: 11–13, DCC, HOBt, Et3N, DMF/CH2Cl2; 7-(diethylamino)coumarin-3-carboxylic acid, rt. For 16, 18 and 21: 11–13, FITC, Et3N, DMF, rt. For 19: 12, Na2CO3, NBD–Cl, DMF/THF/H2O, 45 °C. For 22: 13, rhodamine B 5(6)-isothiocyanate, Et3N, DMF, rt.

Scheme 3

Synthesis of granisetron probes 26–28. Reagents and conditions: (a) for 23: NBD–Cl, Et3N, THF/H2O 4:1, rt. For 24: Dns–Cl, Et3N, CH2Cl2, rt; (b) CH2Cl2, 50 °C; i-Pr2EtN, CH2Cl2/toluene, rt; BF3·Et2O, 50 °C; (c) t-BuOK, THF/DMF 5:1, 0 °C; 23–25, 0 °C to rt.

Other than the N1-position on granisetron, we identified the C7-position as tolerant to substitution in our initial SAR study. Therefore we synthesized probes 36–39, where the fluorophores were conjugated to this latter position via different linkers. The synthesis of 7-hydroxy granisetron 31 was described previously; we have, however, found a more practical and scalable route to access this intermediate (Scheme 4). The indazole-3-carboxylate 29 was synthesized using a 1,3-dipolar cycloaddition of in situ generated aryne with a diazo ester. Only the C7-substituted regioisomer was obtained in this reaction. This was followed by ester hydrolysis, amide formation with bicyclic amine 30, selective N1-methylation and methyl ether cleavage. It was crucial to follow the above order of steps since N-methylation of the indazole ester 29 also yields small amounts of N2-methylated side product which is extremely difficult to separate from the desired N1-isomer. The hydroxyl group of 31 was alkylated with protected spacers 8, 10 or 32 and the Boc-group was subsequently cleaved. Finally, 7-(diethylamino)-coumarin-3-carboxylic acid or FITC was coupled to the spacers to yield probes 36–39.

Scheme 4

Synthesis of granisetron probes 36–39. Reagents and conditions: (a) N2CHCO2Et, TBAF, THF, −78 °C to rt; (b) 2 M NaOH, MeOH, rt; (c) DCC, HOBt, DMF/CH2Cl2; 30, rt; (d) t-BuOK, THF/DMF 4:1, 0 °C; MeI, 0 °C to rt; (e) 1 M BBr3 in CH2Cl2, CH2Cl2, rt; (f) K2CO3, n-Bu4NI, 8, 10 or 32, DMF, 60 °C; (g) 4 M HCl in 1,4-dioxane, rt; (h) for 36–38: DCC, HOBt, DMF/CH2Cl2; 7-(diethylamino)coumarin-3-carboxylic acid, rt. For 39: FITC, Et3N, DMF, rt.

The photophysical properties of fluorescent granisetron derivatives were measured in MeOH and/or phosphate buffer at pH 7 (Table 1). The granisetron probes that have fluorescein (16, 18 and 21) or BODIPY dyes (6, 14 and 28) appended to N1 show the highest quantum yields (Φf). The quantum yield of the coumarin-containing probes 15, 17, 20, 36–38 was low in pH 7 buffer, but was substantially increased in less polar solvents (e.g., for 15: Φf (CH2Cl2) = 0.55). The binding affinities of the fluorescent granisetron probes for the human 5-HT3AR were determined by competition binding studies with [3H]granisetron. Probes 14, 16, 18, 27, 28 and 36 exhibited affinities similar to the parent compound granisetron (Table 1) and thus could be useful as tracer ligands for fluorescence-based binding assays and tools for imaging. In terms of probe design it appears that optimal binding is obtained if the fluorophore is conjugated to N1 of granisetron via a short (butyl) aliphatic spacer.

The utility of granisetron probes 16–18, 26–28 and 36 to fluorescently label the receptor was studied using live cell imaging of COS-7 cells transiently transfected with human 5-HT3A receptors. Only probes 16, 18 and 28 gave detectable staining. Probes 16 and 18 produced selective staining for 5-HT3R as demonstrated by fluorescence at the periphery (plasma membrane) of transfected cells, whilst this fluorescence was absent in mock transfected cells (Fig. 2). Furthermore, 5-HT3R staining with these probes was inhibited by co-incubation with the 5-HT3R antagonist ondansetron (OND, 10 μM) (Fig. 2). In contrast, probe 28 produced intense fluorescence in both 5-HT3A- and mock-transfected cells, and was not blocked by ondansetron (Fig. 2). These data suggest nonspecific interactions of this probe with the cells.

Figure 2

Fluorescent labeling of human 5-HT3AR in live COS-7 cells. Cells were either transfected with human 5HT3A cDNA (left and middle panels), or mock transfected (right panels). 24 h later they were incubated with 100 nM of probes 16, 18 or 28 in HBS buffer for 1 h at room temperature in the dark. The cells were imaged using a fluorescence microscope set to the appropriate absorption/emission wavelengths (Table 1). Some cells were also co-incubated with 10 μM ondansetron (OND, middle panels) to block 5-HT3 receptors. Scale bar represents 50 μm.

In summary, we have designed and synthesized a small library of fluorescent 5-HT3 receptor ligands. Most notably, novel granisetron conjugates 14, 16, 18, 27 and 28 have similar binding affinities for the human 5-HT3AR as measured for the parent compound 1. This is in agreement with previous studies which showed that bulky fluorophores can be appended via short aliphatic linker to the N1-position of granisetron. Moreover, these high affinity probes exhibit high quantum yields and emission maxima above 500 nm in polar media. Probes 16 and 18, in particular, show specific fluorescent labeling of the human 5-HT3AR in live cells. We are currently focusing our efforts on evaluating our probes in fluorescence polarization and flow cytometry applications.