Synthesis and dopamine receptor pharmacological evaluations on ring C ortho halogenated 1-phenyl-benzazepines

Abstract

A series of 1-phenyl-benzazepines containing bromine or chlorine substituents at the ortho position of the appended phenyl ring (2’-monosubstituted or 2’,6’- disubstituted patterns) were synthesized and evaluated for affinity towards dopamine D₁R, D₂R and D₅R. As is typical of the 1-phenyl-benzazepine scaffold, the compounds displayed selectivity towards D₁R and D₅R; analogs generally lacked affinity for D₂R. Interestingly, 2’,6’-dichloro substituted analogs showed modest D₅R versus D₁R selectivity whereas this selectivity was reversed in compounds with a 2’-halo substitution pattern. Compound 10a was identified as a D₁R antagonist (K_i = 14 nM; IC₅₀ = 9.4 nM).

Graphical Abstract

The neurotransmitter dopamine is implicated in a number of physiological functions in both the periphery and central nervous system (CNS) such as locomotion, blood pressure regulation, cognition and emotion.^1–7 Perturbations in dopaminergic neurotransmission underlie some CNS disorders such as Parkinson’s disease, schizophrenia and drug abuse.^8–11 Therefore normalization of dopaminergic neurotransmission with pharmacological agents has been explored as a means to treat these conditions.

Dopamine exerts its pharmacological actions via agonist activity at 5 dopamine receptors (D₁R – D₅R). Dopamine receptors are classified as “D₁-like” (constituted by D₁ and D₅ receptor sub-types – D₁R and D₅R) and “D₂-like” (comprising D₂R, D₃R and D₄R) based on the structure and function of the receptors and pharmacological studies.^12–15 The discovery of ligands that are highly selective for either the D₁R or D₅R sub-types has proved challenging due to the close transmembrane structural similarity between D₁R and D₅R (>80% homology in transmembrane regions). Thus, commercially available D₁R ligands (i.e. “D₁R-like” ligands) usually display similar affinity at D₅R. It is only quite recently that D₁R subtype selective ligands have been disclosed.^{16, 17} The availability of highly selective ligands for either D₁R or D₅R is of current interest as such compounds would be useful tools to unravel the individual roles of D₁R and D₅R in various physiological processes and serve as lead molecules for related CNS disorders.

The 1-phenylbenzazepine framework is a classical template for D₁R-like ligands and numerous compounds along a continuum of functional activity (i.e. full agonists, partial agonists, antagonists) have been identified with this scaffold.^18–23 A number of these compounds are used as research tools in pharmacological studies. For example, SCH 23390 (1) is a widely used D₁R-like antagonist tool; it displays very high affinity for D₁R and D₅R (0.2 and 0.3 nM respectively).²⁴ SKF 38393 (2) is a widely used D₁R-like agonist with strong affinity for D₁R and D₅R (1.0 and 0.5 nM respectively).²⁵ Fenoldopam (3), a peripherally restricted D₁R-like partial agonist is currently the only compound from this class that is in use clinically (it is used as a fast-acting anti-hypertensive drug).²⁶

There have been several structure-activity relationship (SAR) studies on 1-phenylbenzazepines as D₁R-like receptor ligands and these studies have established that the aryl substituent groups as well as the nitrogen substituent can significantly impact D₁R-like affinity, D₁R-like selectivity versus “D₂-like” receptors as well as functional activity.^27–32 Although several SAR studies have been performed on the scaffold, it has not been determined how halogen substituents in the ortho position of ring C in the 1-phenyl-benzazepine framework impacts affinity and selectivity for dopamine receptors; there is no data available concerning the D₁R versus D₅R affinity of the compounds. Given this gap in the SAR of 1-phenyl-benzazepines, we set out to examine the role of ring C ortho halo substituents on D₁R/D₅R affinity in this scaffold. We hypothesized that such substituents might cause differences in receptor interactions with the substituents themselves and/or lead to modified conformations of the molecule as a whole that could directly influence D₁R and D₅R affinity and selectivity. Thus, we set out to synthesize a set of 1-phenylbenzazepine derivatives with variations in the ring A moiety (either a catechol or protected catechol motif), nitrogen alkyl group substituent and ring C ortho halogenated motif.

The analogs were synthesized as shown in Scheme 1. The epoxides 9 were available commercially or could be readily synthesized from the corresponding styrenes 8. We initially attempted reaction of amine 4 and epoxides 9 to form amino alcohols 5 without the use of any additives/catalysts, as is typically done for the synthesis of 1-phenyl-benzazepine syntheses,^33–35 but found the reactions to be low yielding. Presumably this less than favorable outcome was due to the steric hindrance in the epoxides used. Lewis acids or lithium salts have often been used to promote ring opening of epoxides with amine nucleophiles.^36–38 As LiNTf₂ was reported to provide high yields in such reactions,³⁹ we examined the use of this reagent and were happy to find that the reactions of 4 and 9 to form 5 proceeded in reasonable yield. Acid-catalyzed cyclization of 5 afforded the benzazepine framework of compounds 6a-d. Compounds 6a-d served as key intermediates from which the synthesis diverged to prepare other analogs with variations in the phenolic moiety and/or N-substituent.

Thus, treatment of compounds 6b with BBr₃ gave the catecholic compound 7. The secondary amine in compounds 6b-d was methylated via reductive amination to give analogs 10a-c respectively. Treatment of compounds 6b-d with allyl bromide allowed for the preparation of N-allylated analogs 11a-c. Compounds 10c and 11c in turn were demethylated by reaction with BBr₃ affording catechols 12 and 13 respectively.

The binding affinity of compounds 6a-d, 7, 10a-c, 11a-c, 12 and 13 were assessed at dopamine D₁, D₂ and D₅ receptors. Data for these assessments are presented in Table 1 as K_i values in nM. Compounds 6a-d had D₁R affinities ranging from 72 to 147 nM. Affinities at D₅R for this sub-group of compounds were slightly lower overall (ranging from 82 to 483 nM), so that on a whole, compounds 6a-d were slightly more selective for D₁R over D₅R. Compound 6c is interesting as it is the only compound of the 6a-6d subset that was D₅R selective over D₁R (3-fold). Compounds 6a, 6b and 6d displayed modest selectivity for D₁R over D₅R (up to 3-fold).

Table 1.

Binding affinity of analogs at D₁R, D₂R and D₅R

					Ki(nM)a
Cmpd #	R	R1	X	X1	D1b	D2C	D5d
6a	Me	H	H	H	126.4 ± 9.1	nae	309.9 ± 23
6b	Me	H	H	Cl	147.4 ± 4.1	na	482.8 ± 29.4
6c	Me	H	Cl	Cl	76.1 ± 3.6	na	25.4 ± 2.6
6d	Me	H	H	Br	72 ± 5.6	na	82 ± 6.2
7	H	H	H	Br	26 ± 3.1	na	67 ±7.1
10a	Me	Me	H	C1	14 ± 2.3	na	46.4 ± 3.8
10b	Me	Me	Cl	Cl	144.6 ± 9.0	na	49.1 ± 3.7
10c	Me	Me	H	Br	16 ± 1.4	na	47 ± 3.2
11a	Me	Allyl	H	Cl	48.3 ± 8.2	na	264.4 ± 13.6
11b	Me	Allyl	Cl	Cl	1044 ± 59.2	1507.8 ± 89.2	479.1 ± 39.2
11c	Me	Allyl	H	Br	41 ± 2.8	na	501 ±64
12	H	Me	H	Br	59.4 ± 4.9	na	223.2 ± 18.6
13	H	Allyl	H	Br	132.9 ± 8.9	na	442.2 ± 56
(+)−					4.04 ± 0.2
Butaclamol
Haloperidol						5.58 ± 0.3
SKF 83566							3.95 ± 0.2

^aExperiments carried out in triplicate;

^b[3H]SCH23390 used as radioligand;

^c[3H]N-methylspiperone used as radioligand;

^d[3H]SCH23390 used as radioligand;

^ena – not active (<50% inhibition in a primary assay when tested at 10μM).

Compound 7 with a bromo substituent, showed strong D₁R and D₅R affinity. In comparing phenol 7 to its C7 methoxy analog 6d, it is apparent that the effect of cleavage of the C7 methoxy group to give the catechol 7, results in a roughly 3-fold increase in D₁R affinity; improvement in affinity at D₅R was more modest.

Compounds 10a-c are the N-methylated analogs of 6b-d respectively; compounds 10a-c as a group displayed stronger D₁R affinity than their N-des-methyl counterparts. Compounds 10a and 10c had the strongest D₁R affinity of any compound evaluated for this study (K_i of 14 and 16 nM for 10a and 10c respectively). The compounds in this sub-group with mono-halogen substituents (10a and 10c) showed modest D₁R selectivity over D₅R (3-fold and 4-fold respectively), but this selectivity was reversed in the dichloro substituted compound 10b.

Similar trends as for the N-methylated analogs 10a-c were seen for the N-allylated analogs 11a-c. In that regard, both the mono-halo substituted analogs 11a and 11c were D₁R selective whereas the dichloro substituted analog was D₅R selective. In general, the N-allyl analogs displayed lower affinity for D₁R and D₅R than their N-methyl congeners.

The catecholic analogs 12 and 13 are the O-demethylated analogs of 10c and 11c respectively; both 12 and 13 showed lower D₁R and D₅R affinity than their methylated precursors and both also had modest D₁R selectivity versus D₅R (in the 3- to 4-fold range).

In analysis of the effect of monohalogenated versus dihalogenated substitutions in the pendant aryl ring, interesting observations emerge. In the case of the 6b/6c pair, the dihalogenated compound 6c showed higher D₁R and D₅R affinities than the monohalogenated congener 6b. However, a similar change in the 10a/10b pair and the 11a/11b pair resulted in diminished D₁R and D₅R affinities for the corresponding dihalogenated analogues. This result indicates that the presence of an N-alkyl substituent is more favorable for binding of the monohalogenated versus their dihalogenated congeners at D₁R/D₅R, whereas absence of such a substituent leads to a stronger preference towards binding of the dihalogenated versus monohalogenated variants.

Comparison of data for the 6b/6d, 10a/10c and 11a/11c compound pairs enabled an analysis of the effect of monochloro versus monobromo substitution in the analogue series. In the case of the 6b/6d pair, the bromo analogue 6d had higher D₁R and D₅R affinity than the chloro analogue 6b. However, in the case of the 10a/10c and 11a/11c pairs, changing from a chloro group to a bromo group did not result in a similar increase in affinity of the brominated analogues for the D₁R as was seen for 6b/6d; affinities for the chloro and bromo variants were similar (e.g. 14 nM and 16 nM for 10a and 10c respectively at D₁R). Meanwhile, at the D₅R, affinities of the bromo and chloro analogues were similar (for 10a/10c) or were worse for the bromo analogue (for 11a/11c). Therefore, it appears that in this series, the presence of an N-alkyl substituent group does not lead to a strong preference for binding of the monobromo versus monochloro variants at D₁R; however, the absence of an N-alkyl substituent leads to stronger binding of the monobromo versus monochloro congeners at D₁R.

We selected the compound with the highest D₁-like receptor affinity for further evaluation of functional activity. Thus, compound 10a was evaluated for agonist and antagonist activity in D₁R assays that measured cAMP modulation by Eurofins Lead Hunter Discovery Services. As expected, (based on structural similarity to 3) 10a displayed strong antagonist activity in these assays (IC₅₀ = 9.4 nM for 10a; IC₅₀ of positive control SCH 39166 = 1.5 nM). No agonist activity was detected for 10a.

In order to provide insights into the important receptor-ligand interactions between the ortho halogen substituted 1-phenyl-benzapines and the D₁R and D₅R, computational docking studies were conducted for the series of analogs in Table 1. In this context, we explored the docked ligand poses and identified key interactions that have a significant impact on binding to the dopamine receptors for this ligand series. These efforts focused mainly on the compounds 7, 10a and 10c, which displayed the best experimental binding affinities to D₁R.

Homology models of D₁R and D₅R were generated and utilized in the docking studies. The D₁R homology model was constructed from the high-resolution crystal structure of the human β₂-adrenergic G protein-coupled receptor (GPCR) with pdb code 2RH1 followed by induced fit docking with several halogenated 1-phenyl-benzazepine analogs.⁴⁰ In a similar manner, the D₅R homology model was created from the high-resolution crystal structure of the β₁-adrenergic GPCR with pdb code 6H7J followed by induced fit docking with the benzazepine analogs.⁴¹ Models of appropriate amino acid backbone and side-chain orientations in the ligand binding site. The homology model building procedure involved application of the Schrödinger Prime Structure Prediction, Induced Fit Docking and Glide software tools in conjunction with manual intervention to support the formation of known key receptor-ligand interactions. The docking runs of the 1-phenyl-benzazepine analogs into the D₁R and D₅R binding sites utilized the Schrödinger Glide methodology in Standard Precision (SP) mode. Using this approach, the Glidescore scoring function provided an estimate of the ligand binding affinities for the highest ranked poses of the ligand series in the D₁R and D₅R targets. The binding poses for the compounds 7, 10a and 10c (docked as the R enantiomers), which gave the best D₁R experimental affinities, are depicted in Figures 2A and 2B.

Figure 2.

Docked poses of compounds 7 (blue carbon atoms), 10a (green carbon atoms) and 10c (pink carbon atoms) in A - the D₁R target and B - the D₅R target. The receptor targets are depicted by secondary structure elements and grey carbon atoms for select residues. Key quaternary N – Asp salt bridges are depicted by the pink dashed lines, H-bonding interactions by the yellow dashed lines, aromatic H-bonding by the turquoise dashed lines and π-π stacking by the blue dashed lines. Docking studies were performed with the R enantiomers.

Compounds 7, 10a and 10c give very similar docked poses in the D₁R binding pocket as shown in Figure 2A with binding energies in the range −7.8 kcal/mol to −8.2 kcal/mol. The docked poses display the quaternary N - Asp103 salt bridge, H-bonding interactions of the ligand hydroxyl group to the Asn292 side chain, and for compound 7 with an additional hydroxyl group in the catechol moiety there is also a H-bond to the Ser198 sidechain, as well as an aromatic H-bond involving the pendant phenyl group and Ser188.

In the D₅R binding site, the main receptor-ligand interactions for the docked poses of compounds 7, 10a and 10c comprise the quaternary N - Asp120 salt bridge, hydrogen bonding interactions of the ligand hydroxyl group to the Asn316 or Ser229 sidechain, and again for compound 7 there is another hydrogen bond with its second catechol hydroxyl group to Ser233, as well as π-π hydrophobic interactions involving the ligand aromatic rings with Phe312 and Trp116. Compounds 10a and 10c form docked poses with binding energies of −8.2 kcal/mol and −8.1 kcal/mol, respectively, which are very similar to those in the D₁R binding pocket.

Overall, the computationally predicted binding energies for the docked series of halogen substituted 1-phenyl-benzapine derivatives in Table 1 are similar in both the D₁R and D₅R structures or a little better in D₅R as a consequence of the slightly stronger hydrogen bonding and π-π hydrophobic interactions. In this context, the docking scores do not align with the selectivity trends derived from the experimental binding affinities. This is at least partially attributable to the modest nature of the observed D₁R/D₅R experimental selectivities of the ring C halogenated analogs. Furthermore, D₁R and D₅R are very similar structurally in the ligand binding pocket, which provides justification for the close computational binding energies for most of the compounds in these two target sites. The docking outcomes for the compounds in Table 1 involved the R enantiomers whereas the affinity data were obtained with racemic mixtures and this could also have an impact on the match between the experimental and computational results. Docking simulations were investigated with the S enantiomers, however, they generated similar trends compared to the R enantiomers with, in general, slightly worse predicted binding energies in both the D₁R and D₅R targets.

In conclusion, this study extends the available SAR information on 1-phenyl-benzazepines as D₁R-like ligands with regards to the effect of ring C ortho halogen substituents. As is evident from examination of the data, the compounds in this study maintain selectivity for D₁-like receptors over D₂R, with modest selectivity for either D₁R or D₅R. As compared to known 1-phenyl-benzapeine D₁R-like tools such as 1 and 2, it is apparent that the ortho halogen group does not significantly improve D₁R or D₅R affinity. However, one of the findings from this work is that compounds with di-ortho-halo substituents (i.e. C2’/C6’ substitution) favor binding to D₅R, whereas compounds with a mono-ortho-halo (C2’) substituent favor D₁R binding over D₅R. In addition, the SAR data suggests that the most favorable outcome for good D₁R affinity is to have either mono-ortho-halogenation in tandem with N-alkyl substitution or di-orthohalogenation without N-alkyl substitution.

Evaluation of the functional activity of 10a reaffirms the idea that an 8-hydroxy-7-methoxy moiety favors antagonist rather than agonist activity. This result is in line with the generally accepted view that a catechol motif is required for agonist activity in the 1-phenyl-benzazepine scaffold.

Our molecular docking studies revealed interactions that were relevant for affinity of the molecules at D₁R and D₅R, but were unable to resolve interactions necessary for the observed modest D₁R sub-type selectivity of 7, 10a and 10c. Examination of larger sets of compounds with ortho halogenated patterns in future, including enantiopure analogs, may provide a larger body of data to aid in the challenging optimization of these ligands towards D₁R or D₅R potency and sub-type selectivity.

Figure 1.

Structures of typical 1-phenylbenzazepine D₁R-like ligands – Fenoldopam, SCH 23390 and SKF 38393.

Reagents and Conditions:

(a) 1. mCPBA, DCM, rt, 12h; 2. NaOH, rt, (88 – 96%); (b) 9, LiNTf₂, THF, reflux, 24 h, (56 −76)%; (c) 1. TFA, H₂SO , rt, 5h; 2. NaOAc, (43 – 54%); (d) BBr₃ , DCM, 0 °C, 4 h, (56 – 80%); (e) HCHO, Na(OAc)₃BH, ACN, rt, 12 h, (23 – 48%); (f) Allyl Bromide, TEA, ACN, rt, 16 h, (56 – 64%).