Design of Phthalazinone Amide Histamine H₁ Receptor Antagonists for Use in Rhinitis

Abstract

The synthesis of potent amide-containing phthalazinone H₁ histamine receptor antagonists is described. Three analogues 3e, 3g, and 9g were equipotent with azelastine and were longer-acting in vitro. Amide 3g had low oral bioavailability, low brain-penetration, high metabolic clearance, and long duration of action in vivo, and it was suitable for once-daily dosing intranasally, with a predicted dose for humans of approximately 0.5 mg per day.

Allergic rhinitis affects globally around 500 million people, with high prevalence in the industrialized world. Furthermore, it has been steadily increasing with a near quadrupling of primary care consultations over the last 50 years.^1,2 Rhinitis symptoms include nasal itching (pruritus), irritation and repetitive sneezing, rhinorrhea, nasal congestion, headache, irritation of the throat, watering of the eyes (epiphora), and edema. Nasal congestion may become more pronounced leading occasionally to breathing through the mouth and snoring³ and loss of smell (hyposmia).⁴

Small molecule pharmacotherapy for allergic rhinitis includes antihistamines, corticosteroids, leukotriene antagonists, decongestants, mast cell stabilizers, and anticolinergics.^2,3,5 The most commonly used medicines for the treatment of rhinitis are antihistamines, which are H₁ receptor antagonists (inverse agonists).^6,7 These were introduced several decades ago, and although they were effective, they caused sleepiness due to their ability to interact with brain H₁ receptors responsible for wakefulness. Oral second-generation antihistamines possessing polar carboxylic acid groups (fexofenadine and cetirizine, Chart 1) have reduced brain-penetration and a corresponding reduction in their side-effects. Nasal congestion is one symptom that second generation antihistamines are not treating effectively; therefore, they are frequently used in combination with α-adrenergic agonist decongestants, for example, pseudoephedrine. However, these decongestants have their own side-effects, such as hypertension. A further advancement in treating rhinitis was the development of topical treatments, such as azelastine and olopatadine hydrochlorides. The advantage of intranasal treatments is the attainment of a higher concentration of drug directly to the nasal cavity with reduced systemic side-effects due to the lower dose used. In order to avoid absorption of the portion of the dose that is swallowed, topical treatments must exhibit low oral absorption. Azelastine hydrochloride dosed intranasally is used in allergic rhinitis patients not responding well to fexofenadine and loratadine, and is more effective than cetirizine.^8,9 Olopatadine hydrochloride nasal spray has been approved in the US and has comparable efficacy to azelastine, same duration of action (12 h), and slower onset of action (30 min vs 15 min), but also suffers from the same side-effects, dysgeusia (bitter taste), headache, nose-bleeds (epistaxis), and drowsiness (somnolence).¹⁰

Chart 1. Representative Oral and Intranasal H₁ Receptor Antagonists Used Clinically, Single-Ligand Histamine H₁H₃ Dual Receptor Antagonist 1, and Selective H₁ Receptor Antagonists 2a–c Based on a Phthalazinone Scaffold.

We have previously disclosed our work in optimizing histamine H₁ receptor antagonists containing a phthalazinone scaffold and also the development of single-compound dual H₁ and H₃ receptor antagonists, leading on to our clinical candidate 1 (Chart 1).¹¹ We have also published on phthalazinones 2a–c and their more hydrophilic aza- and diaza-analogues.¹² Rhinitis is closely linked to other allergic diseases, such as asthma, rhinosinusitis, and conjuctivitis. One disadvantage of allergic rhinitis treatment using antihistamine monotherapy is the neglect of other comorbidities. In general, the most potent anti-inflammatory agents are the glucocorticoids, such as fluticasone furoate, which is approved for the once-daily treatment of allergic rhinitis. In this publication we describe our efforts in identifying a long-acting, potent human H₁ receptor antagonist, which could either be used in combination with an H₃ receptor antagonist or with fluticasone furoate.^13,14

We have previously reported on the identification of a channel in the 7TM bundle,¹¹ which was approached from the 2a pyrrolidine nitrogen. We have exploited this finding by appending the H₃ fragment and the butyl linker to construct the H₁H₃ dual receptor antagonist 1.¹¹ Highly potent H₁ antagonists were obtained with simple straight chain alkyl substituents on the pyrrolidine nitrogen of 2a, such as 2b and 2c. The ether oxygen of 2c was thought to have formed an H-bond with a Tyr of TM7.

At the outset we were interested in a very potent, homochiral H₁ antagonist suitable for a low-dose, small-volume intranasal delivery with a 24 h duration of action. We have previously shown that the (R)-enantiomer of 2a was the higher affinity enantiomer.¹¹ We envisioned replacing the ether oxygen of 2c with an amide group, which might provide an even tighter H-bond leading to increased affinity and/or duration, but also introduce metabolic instability leading on to increased clearance, a requirement for topical administration. Furthermore, introducing an electron withdrawing group, such as the amide functionality, was expected to lower the pK_a of the pyrrolidine nitrogen, thereby increasing selectivity against hERG.^12,15,16

Chemistry

The target amides 3 were prepared by the three-step synthetic route outlined in Scheme 1. The (R)-enantiomer of 4(11) was alkylated with N-(2-bromoethyl)phthalimide to give 5 in 87% yield, deprotected using hydrazine to give amine 6 (77% yield), and then acylated with a variety of carboxylic acids using TBTU to provide amides 3d,f,g,h or with the corresponding acid chloride to give 3a,b,c,e. Alkylation of 4 with 2,2,2-trifluoro-N-(2-iodoethyl)acetamide gave the trifluoroacetamide 7, which was alkylated with iodomethane and deprotected to provide the secondary amine 8. The latter was acylated with 4-methoxybutanoic acid and TBTU to give the tertiary amide 9g.

Scheme 1. Synthesis of Amides 3 and 9.

Reagents and conditions: (a) N-(2-bromoethyl)phthalimide, K₂CO₃, 2-butanone, 80 °C, 18 h, 87%; )b) NH₂NH₂·H₂O, EtOH, 80 °C, 1.25 h, 77%; (c) RCO₂H, TBTU, Et₃N, DMF, 2 h, or RCOCl, Et₃N, DCM; (d) 2,2,2-trifluoro-N-(2-iodoethyl)acetamide, DIPEA, 2-butanone, 100%; (e) (i) MeI, K₂CO₃, DMF; (ii) K₂CO₃, H₂O, MeOH, 29% (two steps).

In addition to amides 3a–h and 9g, alternative amides 10b,h where the acyl part of the amide group is reversed (closer to the pyrrolidine basic nitrogen) were prepared and their synthesis is outlined in Scheme 2. These compounds were synthesized by alkylation of pyrrolidine 4 with ethyl 4-bromobutyrate to give ester 11, which was converted to carboxylic acid 12 by base hydrolysis and finally converted to amides 10b and 10f.

Scheme 2. Synthesis of Amides 10b and 10f.

Reagents and conditions: (a) BrCH₂CH₂CH₂CO₂Et, K₂CO₃, DMF, 150 °C, microwave, 110 min, 51%; (b) 2 M NaOH, H₂O, MeOH, 20 °C, 1 h, 91%; (c) nPrNH₂ for 10b or MeOCH₂CH₂NH₂ for 10f, TBTU, Et₃N, DMF, 47–68%.

Results and Discussion

The biological screens used to assay the compounds were reported previously¹¹ and include H₁ receptor affinity using recombinant human histamine H₁ receptor,¹⁷ adrenergic α_1A and α_1B receptor affinity,¹⁸ and hERG activity,¹⁸ and the data are summarized in Table 1. Azelastine, which was used as a standard, exhibited high affinity for both the H₁ receptor (pK_i = 8.9) and the hERG channel (dofetilide binding assay pIC₅₀ = 7.0).¹¹ The straight chain amides 3a and 3b were equipotent to azelastine; however, they were less selective for the α_1A and α_1B receptors and were rejected. The branched amides 3c and 3d were less potent than azelastine, but also suffered from reduced selectivity for the α_1A and α_1B receptors and were rejected too. The introduction of an oxygen atom in the unbranched amide chain was used to probe for additional H-bond interactions. The tetrahydropyran analogue 3h, like the other branched amides, was less potent and was rejected. The reverse amide 10b had high hERG affinity and was also rejected.

Table 1. Antagonist Affinity^a at the Human H₁ Receptor (FLIPR Assay), the α_1A and α_1B Receptors (in Fibroblast Cells by Means of Plate-Based Calcium Imaging), Measured logD at pH 7.4, and hERG Binding Affinity.

compd	H1 pKi (n)	α1A pKi (n)	α1B pKi (n)	logD7.4	hERG pIC50
Azelastine	8.9 ± 0.0 (364)	7.3 ± 0.0 (145)	7.3 ± 0.0 (97)	2.3	7.0 (116)
3a·TFA	9.2 ± 0.2 (4)	7.8 ± 0.4 (2)	7.8 ± 0.1 (2)	3.4	6.7 (1)
3b·TFA	9.1 ± 0.1 (4)	7.5 ± 0.1 (2)	7.1 ± 0.4 (2)	3.7	7.2 ± 0.3 (2)
3c·TFA	8.4 ± 0.1 (10)	8.4 ± 0.1 (5)	7.2 ± 0.3 (2)	3.2	6.4 ± 0.0 (2)
3d·TFA	8.4 ± 0.1 (6)	7.1 ± 0.1 (3)	7.1 (1)	3.9	7.3 ± 0.1 (2)
3e·TFA	8.9 ± 0.4 (2)	7.3 (1)	7.5 (1)	3.1	6.7 (1)
3f·HCO2H	9.1 ± 0.1 (4)	8.1 ± 0.1 (2)	7.2 ± 0.1 (2)	3.2	6.4 (1)
3g	8.5 ± 0.0 (4)	7.3 ± 0.1 (9)	7.0 ± 0.2 (8)	3.5	6.4 ± 0.0 (4)
3h·TFA	8.0 ± 0.1 (6)	6.7 ± 0.0 (3)		3.3	6.1 ± 0.0 (2)
9g·TFA	8.5 ± 0.1 (4)	7.5 ± 0.1 (2)	6.9 (1)	3.5	6.4 ± 0.1 (4)
10b·HCl	8.3 ± 0.1 (5)	8.0 ± 0.3 (2)	7.5 ± 0.2 (3)	3.3	8.8 (1)
10f·HCl	9.1 ± 0.1 (12)	8.2 ± 0.1 (6)	8.0 ± 0.1 (6)	2.5	7.4 ± 0.1 (3)

^aMean ± SEM of estimated functional pK_i for n < 3; the SEM is the SD; n = number of experiments

In addition to the above assays, another lower-throughput version of the H₁ assay was used, which provided more precise pA₂ values and an indication of the analogues’ duration of action.¹¹ The data for phthalazinone analogues 3a–h, 9g, 10b and 10f are summarized in Table 2 and compared to azelastine (pA₂ 9.7). Their duration of action in vitro is expressed as faster, slower, or no-difference wash-out time relative to azelastine. Slower wash-out equated to longer duration of action than azelastine, whereas faster wash-out to shorter duration. Analogues 3f and 10f were rejected based on their lower affinity than azelastine in the secondary assay (Table 2). The remaining compounds 3e, 3g, and 9g were equipotent with azelastine, exhibited a longer duration of action in vitro, and were therefore of interest for further examination. Analogue 3g was historically made before the other two compounds and hence was investigated first.

Table 2. Antagonist Affinity at H₁ Receptor (pA₂) and Duration of Action in Vitro.

compd	pA2 ± SEMa	n	wash-out
Azelastine	9.7 ± 0.1	19	reference
3a·TFA	9.6 ± 0.2	6	no difference
3b·TFA	9.2 ± 0.3	2	slower
3c·TFA	9.88 ± 0.15	8	slower
3d·TFA	9.5 ± 0.1	13	slower
3e·TFA	9.7 ± 0.1	14	slower
3f·HCO2H	9.3 ± 0.1	10	slower
3g	9.74 ± 0.09	20	slower
3h·TFA	9.1 ± 0.1	13	slower
9g·TFA	10.0 ± 0.3	12	slower
10b·HCl	9.8 ± 0.2	9	slower
10f·HCl	9.3 ± 0.1	8	slower

^aAll pA₂ values calculated from curve shifts generated after 30 min incubation time at 100 nM antagonist concentration.

Pharmacokinetics Amide 3g

TFA had low in vivo CNS penetration in the rat (brain–blood ratio of 0.18 and mean brain concentration of 121 ng/g, n = 3) assessed in samples taken 5 min after an intravenous bolus dose of 1 mg/kg. High in vitro plasma protein binding was observed in four species (rat, guinea pig, dog, and human), and blood cell association of 3g·HCl was low to moderate in the same species (Table 3, in SI section). The membrane permeability of 3g·HCl across MDCK cells in the presence and absence of a P-glycoprotein (PgP) inhibitor was 220 nm/s at pH 7.4 when incubated at a concentration of 2.5 μg/mL, indicating that 3g is not a PgP substrate and that low brain exposure would be maintained. The trifluoroacetate and hydrochloride salts of 3g were progressed to the male CD Sprague–Dawley rat and Beagle dog pharmacokinetic studies in vivo. The compounds were dosed as solutions in H₂O–PEG200–DMSO (50:45:5). Each compound was dosed iv at 1 mg/kg (both species) and po at 3 mg/kg for the rat and 2 mg/kg for the dog. In addition one dog was also dosed subcutaneously (Table 4, in SI section). Following intravenous administration of 3g blood clearance (Clb) was high compared to liver blood flow (LBF) (means of 66 and 31 mL/min/kg in rat and dog, respectively) and the steady-state volume of distribution (V_ss) was low to moderate (means of 0.7 and 1.1 L/kg in rat and dog, respectively). The resultant elimination half-life was short (means of 0.2 and 0.5 h in rat and dog, respectively). Elimination of parent compound via urine was low following oral, subcutaneous, and intravenous administration of 3g·HCl to the dog (nondetectable following oral administration and <3% following subcutaneous and intravenous administration). Following oral administration of 3g bioavailability was low (<3% for rat and <8% for dog). Oral bioavailability in the rat was assessed at 3 mg/kg (negligible bioavailability) and 300 mg/kg (<3% bioavailability). At the higher dose in addition to low bioavailability the blood concentrations were highly variable between rats. Absorption of 3g into the hepatic portal vein following oral administration to the rat was also shown to be low (3 to 7% using a 3 mg/kg dose). Following subcutaneous administration of 3g to the dog bioavailability appears complete. These data suggest that the low in vivo bioavailability may be due to high first-pass metabolism (both intestinal and hepatic) rather than poor intestinal permeability. The bioavailability of azelastine in the rat and dog¹¹ was higher than that of 3g, whereas in humans the reported azelastine bioavailability is 40%.¹⁹ The in vitro rate of metabolism was high for 3g·HCl using liver microsomes from mouse (50 mL/min/g tissue), rat (50 mL/min/g), dog (12 mL/min/g), and human (25 mL/min/g). The high in vitro hepatic clearance data is in agreement with the high in vivo clearance data for rat and dog. The in vitro rate of metabolism for 3g·HCl was also significant using intestinal microsomes from rat (12 mL/min/g), monkey (>50 mL/min/g), and human (11 mL/min/g). These data are in agreement with the low bioavailability/absorption observed following oral administration to the rat, despite 3g showing good cell permeability. Incubation of 3g·HCl (10 μM initial concentration) with cryopreserved human hepatocytes for 3 h resulted in extensive metabolism of the compound, detecting at least 16 metabolites. Of the metabolites, labeled M1 to M16 according to their retention time on the LC–MS/MS chromatogram, M1 was the first compound to elute off the column. The metabolites ranged from oxidation of the pyrrolidine ring (lactam formation, dehydrogenation of pyrrolidine ring), amide hydrolysis, N- and O-dealkylation, and glucuronidation. The structures of these metabolites are shown in Chart 2 in the SI section together with their relative abundance. The levels of the two acyl glucuronide metabolites were low compared to the overall metabolic profile. No evidence was found for the formation of N-oxide or glutathione conjugated metabolites. Some second-generation antihistamines are substrates and modulators of CYP450, in particular the subtype CYP3A4. The inhibitory potential of 3g was assessed for the five major human CYP450 enzymes (CYP1A2, >100 μM; CYP2C9, 6.7 μM; CYP2C19, 27 μM; CYP2D6, 1.7 μM; CYP3A4, 1.3 μM). Compound 3g did not cause time-dependent inhibition with CYP2D6, but a significant change in IC₅₀ with time was observed using CYP3A4 (mean fold change of 5). In the Cypex cDNA screen Azelastine also shows time-dependent inhibition of CYP3A4.

Effect of 3g on Histamine-Induced Nasal Congestion in Vivo

The standard Buxco whole body plethysmography technique was used to investigate the effect of intranasally dosed 3g on histamine-induced nasal congestion in conscious, unrestrained guinea pigs, which were previously sensitized with ovalbumin and aluminum hydroxide intranasally over a 3-week period prior to the study. Recording PenH (enhanced pause) AUC over 40 min following bilateral histamine challenge (15 or 10 mM, 25 μL/nostril, under light isoflurane anesthesia) allows for the assessment of efficacy and duration of action of intranasally dosed antihistamines.²⁰

Compound 3g showed significant inhibition of histamine-induced congestion, compared to a vehicle-pretreated/histamine challenged control group, at 1 and 24 h after administration of a 1 mg/mL solution (25 μL/nostril). In contrast, azelastine did not demonstrate a similar duration of action at the same concentration (Figure 1). Further studies indicated that duration of action is less than 48 h following administration of a 1 mg/mL solution of 3g (data not shown). Neither 3g nor azelastine produced a procongestant response at any time point studied (effect on baseline PenH, data not shown). Arterial blood samples taken immediately following evaluation of nasal congestion (4 or 25 h after intranasal dosing) do not show any systemic exposure from 1 mg/mL of 3g in these studies, which indicates a low bioavailability by this route and a local mechanism of action.

Figure 1.

Duration of action of azelastine and 3g in the guinea pig model of nasal congestion. Animals were exposed to histamine at 1 and 24 h after an intranasal dose of 1 mg/mL of 3g·TFA, 3g·HCl, or azelastine. Mean ± SEM (n = 16–27 per group). *p < 0.05 compared to histamine control group; ^#p < 0.05 compared to vehicle/PBS control group. Bar indicates p < 0.05 individual comparison as indicated. ANOVA with posthoc Hochberg analysis.

In order to determine the cross species activity between human and guinea pig,²⁰ the effect of 3g on the contractile response to histamine in guinea pig isolated trachea (i.e., upper respiratory tract and closest available tissue assay to the nasal cavity for investigation) and human bronchus was determined (guinea pig pA₂ 8.7 and human bronchus 8.2) demonstrating that the potency was comparable between species.

Predicted Human Dose

The clinical dose of 3g is expected to be low (approximately 0.5 mg per day) based on the clinical dose used for azelastine and a comparison of efficacy data from animal models for 3g and azelastine. Based on high blood clearance, low oral rat and dog bioavailability, and a low clinical dose, the systemic exposure to 3g in humans is expected to be very low. The concentration of 3g was less than the limits of quantification for the assays used in all assayed blood samples taken from guinea pigs showing a pharmacological response in a histamine induced nasal congestion model following intranasal administration of 3g. In order to appreciate toxicology risks/overage the pharmacokinetic profiles in preclinical species following intravenous administration were used to give estimates of the maximum systemic exposure in humans if most of the intranasal dose was swallowed and bioavailability was complete. For the predictions a standard absorption rate was assumed; the volume of distribution was kept constant and the elimination rate was scaled based on liver blood flow. Based on these simulations the maximum estimates for exposure in humans following a 0.5 mg dose are a total drug C_max of 1.4 and 1.6 ng/mL (based on rat and dog data, respectively) and an AUC_0–24 of 4.1 and 4.8 h.ng/mL (based on rat and dog data, respectively).

Specificity Profiling of 3g

The selectivity of 3g for the human H₁ receptor over the other three histamine receptors (H₂, H₃, and H₄) was evaluated in binding assays on recombinantly expressed histamine receptors, using measurements of inhibition of histamine stimulated cAMP in HEK-293 cell membranes (for H₂), inhibition of histamine stimulated GTPγS binding in CHO cell membranes (for H₃), and inhibition of radiolabeled histamine binding in HEK-293 cell membranes (for H₄) and found to show >1000-fold specificity. Compound 3g had a pA₂ for H₁ = 9.4 (n = 54), pK_i for H₂ 5.1 (n = 2), pK_i for H₃ < 6.2 (n = 29), and pK_i for H₄ < 6.4 (n = 2). The selectivity of 3g and azelastine against 50 different receptors, ion channels, enzymes, and transporters was evaluated in the CEREP Specificity Screen in agonist or antagonist mode as appropriate. Compound 3g (at 1 μM) caused >50% inhibition at 6 targets in this screen (α₁, D₁, D_2S, 5-HT_2A, 5-HT₆, and 5-HT₇), indicative of some off-target activity; a similar profile was seen with azelastine, which caused >50% inhibition at 10 targets, four of which overlapped with 3g. Further work was performed to quantify and compare the specificity profile of 3g relative to H₁ receptor (pA₂ 9.4), which showed >100-fold specificity, apart from the human 5-HT_2B receptor (pK_i 7.8 in antagonist mode; 40-fold selectivity for H₁). This is not considered to be an issue since azelastine, which is used clinically, had the same pK_i value in the same assay. In addition 3g has lower bioavailability than azelastine suggesting that systemic exposure after intranasal administration is likely to be negligible. Compound 3g also showed significant affinity (pIC₅₀ = 6.4) in the hERG binding assay, however, this was lower than azelastine (pIC₅₀ = 7.0). Furthermore, 3g was clean in the rabbit ventricular wedge assay (up to 10 μM), suggesting that the risk of 3g-induced QT prolongation is low at the predicted exposure levels.

The mutagenic and clastogenic potential of 3g was assessed in the AMES test, Mouse Lymphoma screen and DEREK structure–activity relationship in silico analysis. No evidence of mutagenic potential was detected in the mini-AMES assay up to the maximum concentrations tested (which were limited by precipitation and/or toxicity as appropriate in each strain) in the presence or absence of an in vitro metabolic activation system (S9-mix). No evidence of clastogenic potential was detected in the mouse lymphoma screen up to maximum concentrations of 80 μg/mL for 3 h in the presence of S9-mix and 40 μg/mL for 24 h in the absence of S9-mix (both maximum concentrations were limited by cytotoxicity). The analysis of the molecular structure of 3g using DEREK in silico analysis showed no evidence of overt mutagenic or toxic potential.

Conclusion

A series of pyrrolidine amide human H₁ histamine receptor antagonists based on the phthalazinone scaffold were synthesized. Amides 3e, 3g, and 9g were equipotent with azelastine, the clinical gold-standard (pA₂ 9.9, 9.7, and 10.0, respectively, vs 9.7 for azelastine), and had a longer duration of action than azelastine in vitro. Amide 3g was selective for the H₁ receptor with >1000-fold selectivity over the other three histamine receptors. Furthermore, 3g had significantly longer duration of action in vivo than azelastine in a nasal congestion model. The bioavailability of 3g was low, which minimizes the potential for side-effects from the swallowed fraction of the intranasal dose. Mean brain concentration following iv dosing in the rat was low, which coupled with the low bioavailability, should limit the potential for any CNS related side-effects. The in vitro rate of metabolism for 3g was high using human liver and intestinal microsomes, and in hepatocytes. A number of metabolites in human hepatocytes were detected by LC–MS/MS and includes metabolites from oxidation of the pyrrolidine ring (lactam formation), amide hydrolysis, N- and O-dealkylation, and glucuronidation. High in vivo clearance was observed in the rat and dog. Amide 3g had a lower affinity than azelastine for the hERG channel, and high human plasma-protein binding. CYP3A4 was inhibited by 3g in a time-dependent manner; however, azelastine behaves similarly in the Cypex cDNA screen. As the bioavailability of 3g is low, coadministration with other drugs is unlikely to cause any significant drug–drug interactions in the clinic. In summary, 3g is suitable for progression as an intranasal candidate for the treatment for allergic rhinitis, either as a monotherapy or in combination with an anti-inflammatory steroid, such as fluticasone furoate.

Supporting Information Available

The Supporting Information is available free of charge on the ACS Publications website at DOI: 10.1021/acsmedchemlett.7b00112.

• Experimental details on all compounds and characterizing data, Tables 3 and 4, and Chart 2 (PDF)

The authors declare no competing financial interest.