Structural Analysis and Optimization of NK 1 Receptor Antagonists through
`Modulation of Atropisomer Interconversion Properties
`Jeffrey S. Albert,* Cyrus Ohnmacht, Peter R. Bernstein, William L. Rumsey, David Aharony, Yun Alelyunas,
`Daniel J. Russell, William Potts, Scott A. Sherwood, Lihong Shen, Robert F. Dedinas, William E. Palmer, and
`Keith Russell
`CNS Discovery Research, AstraZeneca Pharmaceuticals LP, 1800 Concord Pike, P.O. Box 15437,
`Wilmington, Delaware 19850-5437, and EST, AstraZeneca Pharmaceuticals LP, Waltham, Massachusetts 02451
`Received April 24, 2003
`We have previously described a series of antagonists that showed high potency and selectivity
`for the NK1 receptor. However, these compounds also had the undesirable property of existing
`as a mixture of interconverting rotational isomers. Here we show that alteration of the
`2-naphthyl substituent can modulate the rate of isomer exchange. Comparisons of the NK
`1
`receptor affinity for the various conformational forms has facilitated the development of a
`detailed NK
`1 pharmacophore model.
`Introduction
`The tachykinins are a family of three mammalian
`neuropeptides: substance P (SP), neurokinin A (NKA),
`and neurokinin B (NKB). The preferred receptors for
`these are termed NK
`1,N K2, and NK3, respectively. The
`NK1 receptor is widely distributed in the CNS and
`peripheral tissue; SP acts as a neurotransmitter or
`neuromodulatory agent. The NK
`1 receptor may be
`involved in several pathophysiological conditions includ-
`ing asthma, emesis, anxiety, depression, and pain.
`1,2
`The areas of neurokinin antagonist development have
`been extensively reviewed.
`3-6
`We have described the identification of the orally
`active, dual NK1/NK2 receptor antagonist ZD6021 (1).7,8
`In subsequent work, it was found that, for compounds
`related to 1, substitutions at the naphthyl 2-position
`altered the NK
`1/NK2 receptor selectivity. This led to the
`identification of ZD4974 (2) as a potent, orally available,
`NK1-selective antagonist.9
`In compound 2, the naphthalene 2-methoxy substitu-
`ent was necessary for NK 1 selectivity; however, it also
`caused the compound to exist in solution as a mixture
`of atropisomers (equilibrating conformational isomers).
`Due to the restricted rotations at the amide and carbo-
`nyl-aryl bonds (Figure 1), a total of four atropisomers
`were evident by high-pressure liquid chromatography
`(HPLC) and NMR spectroscopy.
`Despite its oral potency, progression of 2 as a poten-
`tial drug would be complicated because of the existence
`of multiple conformational forms; these could present
`significant challenges due to potential safety, analytical,
`and manufacturing concerns.
`10,11 To address this, we
`have analyzed the kinetic properties and structure -
`activity relationships (SARs) for 2 and for related
`naphthyl-substituted analogues. These efforts have led
`to the identification of3, which had biological properties
`similar to those of 2, but had improved kinetic proper-
`ties and could be isolated in a single conformational
`form. In this paper, we describe the structural and
`biological properties of 3, and we propose an NK
`1
`receptor pharmacophore model for its activity.
`Chemistry
`Compounds 3-8 could be accessed from the trisub-
`stituted naphthalene 9, which was readily available
`from prior work.9 Synthetic procedures are illustrated
`for the preparation of 3. Selective demethylation of the
`methyl ether in 9 was carried out using magnesium
`iodide (Scheme 1). 12 The resulting naphthol 10 was
`converted to triflate 11, which was coupled with tribu-
`tylvinyltin in the presence of lithium chloride, tetraki-
`striphenylphosphine palladium, and 2,6-di-tert-butyl-4-
`methylphenol to afford12 in good yield. The vinyl group
`was reduced using hydrogen in the presence of pal-
`ladium on carbon to afford 13. Using trimethylsilyl
`iodide, the methyl ester of13 was converted to carboxy-
`lic acid 14. This material was converted to the acid
`chloride and then reacted with N-[(S)-2-(3,4-dichlo-
`rophenyl)-4-[4-[(S)-2-methylsulfinylphenyl]-1-piperidi-
`nyl]butyl]-N-methylamine
`7 to give 3. Using analogous
`approaches, 4-8 were prepared from 9.
`Results
`(a) Influence of the 2-Naphthyl Substituent on
`Naphthyl-Carbonyl Bond Rotation.For compounds
`containing the naphthamide group, rotation is restricted
`at the amide and the naphthyl-carbonyl bonds (Figure
`1) due to resonance and steric effects. In such systems,
`four atropisomers can potentially result since each bond
`* To whom correspondence should be addressed at CNS Discovery
`Research. Phone: (302) 886-4771. Fax: (302) 886-5382. E-mail:
`jeffrey.albert@astrazeneca.com.
`Figure 1. Structures of 1-3. Arrows indicate bonds with
`restricted rotation.
`519J. Med. Chem. 2004, 47, 519-529
`10.1021/jm030197g CCC: $27.50 © 2004 American Chemical Society
`Published on Web 12/23/2003
`Downloaded via 216.241.254.30 on August 28, 2025 at 17:51:36 (UTC).
`See https://pubs.acs.org/sharingguidelines for options on how to legitimately share published articles.
`HELSINN EXHIBIT 2045
`Azurity Pharmaceuticals, Inc. v. Helsinn Healthcare S.A.
`IPR2025-00945
`Page 1 of 11
`
`
`
`
`
`
`
`can assume two orientations (s-cis and s-trans for the
`amide bond, axial-R and axial-S for the naphthyl-
`carbonyl bond). 13 For 1 and analogues where R ) H
`(Figure 1), NMR spectra at room temperature were
`broadened due to the presence of multiple conforma-
`tional forms. Interconversion between the forms ap-
`peared to be relatively rapid since (1) the conformers
`could not be resolved by HPLC and (2) moderate heating
`(50-60 °C) caused coalescence of the NMR resonances
`from the individual conformers.
`In such naphthamide systems, it is known that the
`rate of interconversion is sensitive to the nature of
`substitution at the naphthamide 2- and 8-positions.
`14,15
`Introduction of a methoxy group to the naphthyl ring
`(to afford 2) substantially increased the bond rotational
`barrier; four distinct components could be observed by
`HPLC, shown in Figure 2, and NMR spectroscopy.
`Samples that were enriched in the major atropisomer
`of 2 (2c in Figure 2) were obtained by preparative HPLC
`of the atropisomer mixture. Using analytical HPLC, we
`monitored the reequilibration of the atropisomers and
`estimated the interconversion half-life to be 2-4 h (pH
`7.4, 100 mM phosphate, 37 °C). Because this rate is near
`the expected biological lifetime of the compound, it
`would be likely that exposure would occur to all atro-
`pisomers even if the compound was administered in a
`single conformational form. Therefore, it was judged to
`be unacceptable for further drug development despite
`its potent and selective activity as an NK
`1 receptor
`antagonist.
`Systematic studies of bond rotation rates have been
`carried out by Clayden for closely related 2-substituted
`naphthamides.14,16 These studies indicated that the
`bond rotation rate was substantially influenced by the
`nature of the 2-substituent. This prompted us to exam-
`ine a series of analogues of 2 with the goal of either
`accelerating bond rotation (to reach the fast exchange
`regime) or retarding bond rotation (to enable isolation
`and storage of the active component).
`(b) 2-Substituted Naphthamide Analogues of 2.
`A series of 2-substituted naphthamide derivatives were
`investigated to identify potent NK
`1 receptor antagonists
`which minimized complications due to multiple atropi-
`someric forms. In addition, we sought to find a replace-
`ment for the naphthol ether group in 2 to reduce the
`potential for metabolic liabilities.
`As expected from prior studies,
`9 the 2-naphthyl-
`substituted compounds 3-7 showed reduced receptor
`binding affinity for NK 2 in comparison to 1 (Table 1).
`Additionally, all compounds existed as a mixture of four
`conformational forms as evidenced by HPLC and NMR.
`Replacement of the methoxy group of 2 with methyl (4)
`led to slightly reduced affinity at NK
`1. Replacement
`with ethyl (3) resulted in a retention of high NK 1
`receptor affinity. Replacement with larger groups ( 5-
`7) led to substantially reduced NK 1 receptor affinity.
`The most potent among these (3) was selected for
`further evaluation in biological and structural studies.
`(c) Rate of Atropisomer Interconversion for 3.
`The four atropisomeric components in the equilibrium
`mixture of 3 were separated using preparative HPLC
`in a manner similar to that for 2. The individual
`components were designated as 3a, 3b, 3c, and 3d in
`order of increasing HPLC retention times (these desig-
`nations are analogous to those for compounds 2a, 2b,
`2c, and 2d in Figure 2). Atropisomer interconversion
`was monitored by analytical HPLC starting from each
`of the isolated isomers in aqueous solution. Results were
`independent of conditions of pH between 4 and 8 and
`temperature between 50 and 70 °C. Interconversion
`occurred predominately between pair3a/3d and pair3b/
`3c. Interconversion among all other atropisomers was
`minimal under the time scale studied (up to 18 h at 70
`°C). The rate for the disappearance and appearance of
`Scheme 1a
`a Reagents: (i) Mg, I 2 (73%); (ii) trifluoromethanesulfonic
`anhydride, triethylamine (97%); (iii) tributylvinyltin, lithium
`chloride, tetrakis(triphenylphosphine)palladium, and 2,6-di- tert-
`butyl-4-methylphenol (91%); (iv) H 2 (g), 5% palladium on carbon
`(89%); (v) (TMS)I (95%).
`Figure 2. High-pressure liquid chromatogram of 2 showing
`the resolution of the atropisomeric forms (2a-d). Conditions:
`column Phenomonex Luna C18(2) (3 ím, 4.6  75 mm), 53%
`methanol, 47% water (0.1% TFA), 1.5 mL/min, UV detection
`at 220 nM.
`Table 1. Receptor Affinity (pKB) for Substituted Aryl Sulfoxide
`Piperidine Analogues of 2a
`compd R NK1b NK2c
`1 H 9.00 ( 0.11 8.30 ( 0.12
`2 OMe 9.50 ( 0.11 7.50 ( 0.03
`3 Et 9.56 ( 0.04 7.31 ( 0.28
`4 Me 9.06 ( 0.27 7.10 ( 0.01
`5 CH2CH2CH3 7.80 ( 0.46 6.72 ( 0.33
`6 CH2CH(CH3)2 7.23 ( 0.18 6.35 ( 0.09
`7 Ph 7.39 ( 0.30 7.27 ( 0.16
`a pKB determinations using rabbit pulmonary artery tissue ( n
`) 2-6). b Agonist ASMSP ((Ac-(Arg6,Sar9,Met(O2)11)SP6-11). c Ago-
`nist BANK (â-Ala8NKA(4-10)).
`520 Journal of Medicinal Chemistry, 2004, Vol. 47, No. 3 Albert et al.
`Page 2 of 11
`
`
`
`
`
`
`
`the corresponding isomer obeyed first-order kinetics and
`was pH independent between pH 4 and pH 8. The
`activation energy for atropisomer interconversion was
`determined to be 24( 1 kcal/mol by Arrhenius extrapo-
`lation of rate constants obtained at different tempera-
`tures. This corresponds to a half-life of about 1.8 days
`at 37 °C. The observed interconversion rate is consistent
`with rates seen for related systems.
`14,16 Solid samples
`of 3d (as the citrate salt) showed no changes in atropi-
`somer distribution upon storage at room temperature
`for four weeks; this indicated that any reequilibration
`in the solid state must be very slow. Compared to the
`expected biological half-life of the compound (discussed
`below), the extrapolated half-life at 37 °C is slow (1.8
`days); hence, 3 was judged to be acceptable for further
`investigation.
`(d) Comparison of Atropisomeric Forms of 3.
`Human Receptor Binding Affinity.Receptor binding
`affinity was determined for each of the isolated atropi-
`somers of 3 (Table 2). By keeping the individual atro-
`pisomers chilled during the purification and sample
`preparation, we minimized reequilibration. Prior to
`initiating studies, we verified that each atropisomer had
`a purity level of>95%. An exception was for atropisomer
`3a; some batches had a purity level of only >90%.
`We found NK
`1 receptor binding affinity varied over
`approximately a 17-fold range from 0.40 nM (3b) to 6.94
`nM (3d). For each, receptor affinity was higher for NK1
`than for NK 2. However, selectivity was weak; NK 1
`receptor affinity was greater than NK2 affinity by only
`about 5-20-fold. NK 3 receptor affinity was somewhat
`weaker (Ki ) 97 nM) for the equilibrium mixture (3);
`NK3 receptor affinities for the individual atropisomers
`were not determined.
`(e) Comparison of Atropisomeric Forms of 3. In
`Vitro Activity. Throughout our tachykinin research
`program, for compounds in this series, we have continu-
`ally observed that when receptor binding affinity (to
`human receptors expressed in mouse erythroleukemia
`(MEL) cells) is in the nanomolar range, there tends to
`be a poor correlation between these data and measures
`of in vivo potency when analyzed in various guinea pig,
`rabbit, and gerbil assays. We often observed differences
`of 20-50-fold, or even greater, in animal models among
`compounds that showed similar, and high, human
`receptor binding affinity. This occurs despite the high
`homology among human, guinea pig, rabbit, and gerbil
`tachykinin receptors. This apparent discrepancy be-
`comes more pronounced for compounds of higher affin-
`ity; in these cases the human receptor binding affinity
`appears to plateau, and greater differences are seen in
`the in vivo and tissue models. As a consequence of this,
`and the generally good correlation between potency
`measures in multiple animal models, we base our
`compound evaluation and SAR analysis on results from
`animal models and isolated tissue response data rather
`than binding data to human receptors expressed in
`mouse erythroleukemia cells.
`Pharmacological activity and selectivity were assessed
`in vitro using pulmonary artery isolated from rabbit
`(RPA). As noted above, rabbit neurokinin receptors are
`homologous to human neurokinin receptors. Using RPA,
`NK
`2 receptor potency (pKB) of the individual atropiso-
`mers of 3 (Table 3) varied by about 10-fold; 3b was the
`most potent (pKB ) 7.79), while 3d was the least (pKB
`) 6.25). For the NK1 receptor, 3 induced a suppression
`of the maximal response to added agonist. Such behav-
`ior is often associated with noncompetitive or partially
`competitive antagonism; in these cases, p K
`B cannot be
`determined. However, the p KB value of 3 could be
`estimated using very low antagonist concentration to
`minimize effects on the maximal relaxation response.
`To compare the NK
`1 receptor antagonist activities
`among the individual atropisomers (3a-d at 100 nM),
`the magnitude of the inhibition of the maximum ago-
`nist-mediated tissue relaxation response in the presence
`of antagonist was calculated in comparison to the
`agonist-mediated relaxation response in the absence of
`drug. Greater antagonist activity was indicated by
`increased percent inhibition (Table 3).
`17 Accordingly,
`NK1 receptor activity varied in the following rank
`order: 3d > 3a > 3b > 3c (affinity for 3a is likely to be
`even weaker than indicated because the tested sample
`contained 10% of the most active compound 3d as a
`contaminant). At lower antagonist concentration (10
`nM) it was clearly demonstrated that antagonist activity
`was greater for 3d than for 3a.
`It is notable that the ranking of activity for com-
`pounds 3a-d is different between human receptor
`binding affinity (Table 2) and RPA receptor antagonist
`activity (Table 3). As explained above, the apparent
`inconsistency between the results in these models has
`been previously observed for structurally related com-
`pounds; many have shown similar and high receptor
`binding affinity in cloned MEL cells although they have
`quite different activities in functional models.
`9 Reasons
`for this difference have not been thoroughly investi-
`gated, but could be due to factors including differences
`in receptor expression or compound exposure in the
`Table 2. Antagonist Binding to Human Tachykinin Receptors
`[Ki (nM)]a
`compd NK1b NK2c NK3d
`3 1.24 ( 0.15 37 ( 69 7 ( 11
`3a 1.71 ( 0.43 23 ( 3n d e
`3b 0.40 ( 0.10 35 ( 4n d e
`3c 2.94 ( 0.57 22 ( 6n d e
`3d 6.94 ( 0.85 33 ( 3n d e
`a Human tachykinin receptor expressed in mouse erythroleu-
`kemia cells, n g 2 determinations. b Against [ 3H]SP. c Against
`[125I]NKA. d Against [125I]MePheNKB. e Not determined.
`Table 3. In Vitro Receptor Activity (p KB) Using Rabbit
`Pulmonary Arterya
`NK1
`NK2
`compd pK Bb
`% inhibition at
`[antagonist] )
`100 nMd
`% inhibition at
`[antagonist] )
`10 nMd pKBe
`3 9.56 ( 0.04 62 nd c 7.31 ( 0.28
`3a ndc 89 <10 7.11( 0.08
`3b ndc 51 ndc 7.79 ( 0.21
`3c ndc 22 ndc 6.83 ( 0.11
`3d ndc 98 73 6.25 ( 0.06
`a pKB determinations using rabbit pulmonary artery tissue ( n
`) 2-6). b Agonist ASMSP, antagonist concentration 1 nM. c Not
`determined. d Because these compounds displayed nonpurely com-
`petitive antagonism, activity is evaluated by monitoring the
`magnitude of the maximum tissue relaxation response (control
`response, (5%); values are expressed as percent inhibition, which
`is calculated as 100- percentage of control response; larger values
`indicate greater potency.
`17 Agonist ASMSP, antagonist concentra-
`tion 100 nM. e Agonist BANK, antagonist concentration 1 íM.
`Atropisomer Modulation in NK 1 Receptor Antagonists Journal of Medicinal Chemistry, 2004, Vol. 47, No. 3 521
`Page 3 of 11
`
`
`
`
`
`
`
`models. Because of this apparent discrepancy, we rely
`on a combination of in vivo and tissue-based models to
`evaluate compound potency, as explained above.
`(f) Comparison of Atropisomeric Forms of 3. In
`Vivo CNS Activity and Pharmacokinetic Analysis.
`Central administration of ASMSP to gerbils induces a
`foot-tapping response which can be attenuated by prior
`dosing with an orally available CNS-penetrant, NK
`1
`receptor antagonist. This can provide a convenient way
`to assess activity of NK
`1 receptor antagonists.18 Gerbils
`were orally treated with antagonist at 5 ímol/kg 6 h
`prior to administration of agonist. The CNS activity was
`then monitored by comparing the degree of foot-tapping
`response with control animals that were treated with
`antagonist vehicle only. Among the four atropisomers,
`3d showed the greatest NK
`1 receptor antagonism in the
`foot-tapping model (Table 4); this compound also had
`the greatest effect in RPA tissue (Table 3).
`We note that although3d showed the highest potency
`in the rabbit pulmonary artery tissue functional assay
`(Table 3) and the highest potency in the gerbil foot-
`tapping assay (Table 4), it did not show the highest
`affinity among the four atropisomers to human recep-
`tors expressed in MEL cells (Table 2). As explained
`above, throughout our tachykinin program, compounds
`in this chemical series tend to show better consistency
`among rabbit tissue models, guinea pig in vivo models,
`and gerbil in vivo models than for binding to human
`receptors expressed in MEL cells. For this reason, we
`regard the tissue models and in vivo studies (particu-
`larly in combination) as a more reliable measure of
`compound potency than binding to human receptors
`expressed in MEL cells.
`To determine which was the NK
`1-preferring atropi-
`somer, we compared the potency of each individual
`purified component in the rabbit pulmonary tissue
`functional model and the gerbil foot-tapping model. At
`the time each experiment was initiated, we confirmed
`that the atropisomer sample under study was present
`at a purity of >95% (with the exception of 3a, which
`had a purity level of >90%, containing primarily 3d as
`the minor, contaminating species). On the basis of the
`potency results from the rabbit pulmonary tissue func-
`tional model and the gerbil foot-tapping model, we
`assign 3d as the preferred NK
`1 atropisomer.
`However, it is possible that the atropisomer distribu-
`tion could change during the course of the experiments.
`This is a particular concern for testing in the gerbil foot-
`tapping model because the atropisomer ratios could
`potentially be influenced by, for example, the presence
`of plasma proteins or endogenous rotamases during the
`long time course of the experiments (up to 6 h). To
`confirm that the potent activity of3d in the gerbil model
`was indeed due to 3d alone (and not due to other
`atropisomers which could potentially have accumulated
`from interconversion during the experimental time
`course), we analyzed the atropisomer distribution in
`gerbil plasma samples. Using identical protocols for the
`gerbil foot-tapping model, 3d was orally administered
`to gerbils (n ) 6). At times of 1 and 6 h following
`administration, plasma samples were analyzed by LC-
`MS to determine the atropisomer distribution. We found
`that the distribution of 3d remained consistent at 95 (
`1% from the initial point of dosage, at the 1 h interval,
`and at the 6 h interval. The distribution of the major
`contaminating atropisomer (3c) also remained constant
`at about 4.1%. These results confirm that the suppres-
`sion of the foot-tapping response is due to 3d and not
`due to other forms.
`Pharmacokinetic analysis indicated that 3 had 37%
`oral bioavailability with a biological half-life of about 6
`h in dog (Table 5). Since the atropisomer interconversion
`rate (approximately 1.8 days) was about 7-fold slower
`than the biological elimination rate in dog, it is possible
`that dosing with the single atropisomer3d would result
`in minimal exposure to the three other atropisomers.
`Unfortunately, we were unable to analyze the atropi-
`somer distribution of samples from animal studies (in
`biological matrixes) due to insufficient chromatographic
`resolution in dog. Pharmacokinetic parameters are
`summarized in Table 5.
`(g) Structural Analysis of 3. The atropisomers of
`3 could be individually analyzed by NMR because of the
`relatively slow interconversion among them. The chemi-
`cal shifts from the amide N-methyl protons of 3a and
`3b were shifted downfield relative to those of 3c and
`3d (Table 6). Such a downfield shift is expected when
`the methyl group is oriented near the carbonyl oxygen;
`this is possible only in thecis-amide configuration. The
`corresponding effect is observed for the H16 protons; in
`this case the H16 protons in 3c and 3d are shifted
`downfield relative to those of 3a and 3b. Again, this
`downfield shift is expected in the trans-amide configu-
`ration due to the proximity of those protons to the
`amide-carbonyl bond. In this manner, we provisionally
`assigned trans-amide configurations to 3c and 3d, and
`cis-amide configurations to 3a and 3b.
`It is also notable that the chemical shifts for H8 of
`3b and (particularly) 3d are significantly upfield in
`comparison to the more typical shifts for H8 in 3a and
`3c. The upfield shift of the H8 proton in 3d suggests
`that it may be located proximal to the shielding region
`of the dichloroaryl ring. NMR data for H4 are included
`Table 4. Inhibition of ASMSP-Induced Foot-Tapping Response
`in Gerbila
`compd
`% inhibition
`of response compd
`% inhibition
`of response
`3 33 ( 17 3c 0.3 ( 0.1
`3a 1.4 ( 0.8 3d 91 ( 9
`3b 13 ( 11
`a Determined 6 h after oral dosing of antagonist at 5 umol/kg
`and initiated by CNS administration of ASMSP (100 pmol); greater
`values indicate higher compound potency. n ) 4-8.
`Table 5. Pharmcokinetic Analysis of 3 in Dog (n ) 3)
`pharmacokinetic param result pharmacokinetic param result
`oral dose (ímol/kg) 10 AUC-PO(0-i) [(ng.h)/mL] 6134 ( 4507
`formulation 75% PEG/saline extrapolated AUC (%) 7 ( 2
`Cmax (ng/mL) 662 ( 438 bioavailability (%) 37.3 ( 6.5
`Tmax (h) 4 ( 1 t1/2 (h) 5.9 ( 0.6
`522 Journal of Medicinal Chemistry, 2004, Vol. 47, No. 3 Albert et al.
`Page 4 of 11
`
`
`
`
`
`
`
`for reference; the chemical shifts are nearly equivalent
`for all atropisomers.
`NMR had suggested that 3d (and 3c) was oriented
`with a trans-amide (vide supra), but the carbonyl-aryl
`orientation remained uncertain. We assumed that the
`upfield shifting of the H8 proton resulted from some
`type of aryl-aryl interaction of the naphthalene with
`the dichloroaryl ring. Using molecular mechanics mod-
`eling, we compared the low-energy conformers for the
`axial-R and axial-S carbonyl structures (each with the
`trans-amide). Low-energy conformers with the axial- S
`configuration directed H8 toward the dichloroaryl ring
`(consistent with NMR data), whereas low-energy con-
`formers with the axial-R configuration had H8 directed
`away from the dichloroaryl ring. On the basis of this,
`we associated 3d with S stereochemistry of the carbo-
`nyl-aryl bond. NMR had also suggested that 3b con-
`tained a cis-amide; analogous modeling studies com-
`paring the axial-R and axial-S carbonyl structures for
`the cis-amide-containing 3b allowed us to associate 3b
`with S stereochemistry of the carbonyl -aryl bond.
`Taken together, we propose the structural assignment
`for each of the atropisomers as shown in Figure 3.
`Although these data all refer to studies conducted in
`CDCl
`3, similar chemical shift changes were observed for
`each atropisomer in deuterated methanol and deuter-
`ated dimethyl sulfoxide. Unfortunately, we were unable
`to observe nuclear Overhauser effects between H4 or
`H8 and the dichloro ring in any of the atropisomers
`studied.
`These structural assignments are further consistent
`with the kinetic results; as noted above, we observed
`that the fastest interconversion occurred between pair
`3a/3d and pair 3b/3c. Very slow interconversion oc-
`curred between all other pairs. According to our struc-
`tural assignment, the fastest interconversions would be
`due to the simultaneous inversion of both the amide
`bond and the carbonyl-aryl bond. Independent rotation
`of either the amide or carbonyl -aryl bonds (leading to
`direct interconversion between pair 3d/3b, 3b/3a, 3a/
`3c,o r 3c/3d) would be much slower according to these
`observations.
`This type of simultaneous, paired interconversion has
`been observed for similar naphthamide-containing sys-
`tems.
`16 In these cases, it is understood that inversion
`of the carbonyl-aryl bond causes steric clashes with the
`amide which are relieved by distortion (and concomitant
`rotation) of the amide bond. Amide rotation would
`similarly be facilitated by distortion/rotation of the
`carbonyl-aryl bond. Such interdependence is referred
`to as geared rotation.
`19
`Unfortunately, crystals of 3d were unsuitable for
`crystallographic analysis. However, useful crystals were
`obtained for the truncated analogue 8 in which the
`piperidine group was removed (Figure 4). In solution,
`8 also exists as a mixture of atropisomers, but can be
`easily crystallized and isolated in a single form. NMR
`spectral properties of the crystallized form of 8 are
`similar to those of 3d; the H8 and the NMe signals are
`shifted upfield (ä 6.35 and 2.62, respectively). As for3d,
`these results suggest a conformation involving thetrans-
`amide and an (S)-carbonyl-aryl bond. Furthermore, 8
`does show NK
`1 receptor antagonist activity.20 For 8, the
`trans-amide, (S)-carbonyl amide assignment was con-
`firmed by crystallographic analysis (Figure 5). In addi-
`tion, the crystal structure shows that the naphthyl and
`dichloroaryl rings of 8 are positioned to allow an edge-
`to-face aryl-aryl stacking interaction as originally
`proposed for 3d on the basis of NMR data. The plane of
`the naphthalene is nearly orthogonal to the plane of the
`dichloroaryl ring, and the naphthalene H8 is located 3.4
`Å from the dichloroaryl ring centroid (2.9 Å from the
`nearest carbon of the dichloroaryl ring).
`(h) Development of a Pharmacophore Model for
`3d. Evidence from NMR is consistent with a structural
`model for 3d where the dichloroaryl and naphthalene
`Table 6. 1H NMR Chemical Shifts of Selected Resonances in
`Atropisomers of 3a
`H4 H8 H16 NMe
`3a 8.53 7.67 3.68 3.30
`3b 8.49 6.76 3.62 3.32
`3c 8.48 7.63 4.88 2.66
`3d 8.44 6.19 4.82 2.64
`a Chemical shifts expressed in parts per million for the midpoint
`of the spin system, relative to the peak for tetramethylsilane.
`Spectra were acquired at room temperature in CDCl 3.
`Figure 3. Structural assignments for atropisomers 3a-d.
`Figure 4. Structure of 8.
`Atropisomer Modulation in NK 1 Receptor Antagonists Journal of Medicinal Chemistry, 2004, Vol. 47, No. 3 523
`Page 5 of 11
`
`
`
`
`
`
`
`assume orientations analogous to those seen in the
`crystal structure of 8. These results suggest a binding
`model involving an edge-to-face stacking interaction
`between the dichloroaryl and the naphthalene rings.
`An NK
`1 receptor pharmacophore model involving
`stacked aryl rings has been implicated in numerous
`structural studies on the basis of crystallographic and
`solution experiments
`21-30 as well as modeling. 31-33
`Additionally, many NK 1 pharmacophore models have
`suggested the requirement for a hydrogen bond acceptor
`in the vicinity of the aryl stacking region. Consistent
`with this, we found that NK
`1 receptor activity was lost
`for the analogue containing the reverse amide, or
`replacement of the amide with a sulfonamide (data not
`shown). Another example where a hydrogen-bonding
`group seems necessary can be found in the development
`of MK-869
`21,34 (15; Figure 6). This is a potent and
`selective NK1 receptor antagonist which showed anti-
`depressant activity in clinical evaluation.35 The design
`of this compound included a ketal oxygen to serve as a
`key hydrogen bond acceptor. By building upon this
`model, a subsequent family of compounds was developed
`which contained a highly preorganized spirocyclic core
`and which maintained a ketal oxygen as a critical
`hydrogen-bonding acceptor.
`36
`Computational evaluation of the pharmacophore model
`was carried out by analysis of a set of compounds related
`to 3 (which spanned a wide range of receptor binding
`affinity) along with all the structural information known
`to us. The software program CATALYST
`37 was used to
`generate and rank possible pharmacophore models. The
`appropriateness of a given model was assessed by (1)
`correlations between predicted and measured affinity
`of the various antagonists and (2) the involvement of
`an aryl-aryl stacking interaction and hydrogen bond
`acceptor group.
`Our proposed model for 3d shares key features with
`models for chemically diverse NK
`1 receptor antagonists
`from Merck21 (15) and Takeda29 (16), shown in Figure
`6. Structural studies of compounds related to 15 and
`16 suggest that the aryl rings may adopt either an edge-
`to-face or face-to-face interaction. Our results tend to
`favor an edge-to-face alignment of the aryl rings. Figure
`7 shows the predicted conformations of 3d along with
`the reference compounds 15 and 16 when fit using our
`optimized pharmacophore model. This overlay suggests
`that each compound can adopt an edge-to-face aryl
`stacking interaction, and each contains a group which
`could serve as a hydrogen bond acceptor (ketal oxygen
`for 15 and amide carbonyl for 16).
`In summary, our prior studies had identified 2 as a
`potent, orally available NK
`1 receptor antagonist but
`which had the undesirable structural property of exist-
`ing as a mixture of interconverting atropisomers.
`9 By
`altering the substitution at the 2-position of the naph-
`thalene, it was possible to slow the rate of exchange
`while improving the NK
`1 receptor affinity. For the
`resulting compound ( 3), the individual atropisomers
`were separated and individually analyzed. Structural
`analysis of the most potent atropisomer (3d) is consis-
`tent with an NK
`1 receptor binding model that involves
`an edge-to-face stacking interaction between the dichlo-
`roaryl and naphthalene rings, with the naphthalene
`amide carbonyl serving as a hydrogen bond acceptor.
`This model is consistent with models for other chemi-
`cally diverse, high-affinity NK
`1 receptor antagonists.
`Experimental Section
`Biological Studies. The cloning, heterologous expression,
`and scale-up growth of MEL cells transfected with the NK 1,
`NK2,o rN K3 receptor were conducted as previously described
`for the human NK2 receptor.38-41 The human NK1 receptor was
`identical to that reported previously,42,43 whereas the human
`NK3 receptor differed from the genomic sequence at AA439
`(Cys vs Phe).44,45 Ligand binding assays were conducted with
`[3H]SP (for NK1), [3H]NKA (for NK2), and [125I]MePheNKB (for
`NK3). Cloning of NK1,N K2, and NK3 receptors was conducted
`as published.46 Isolated tissue responses (pKB) and pulmonary
`mechanics studies were carried out as previously described. 8
`For pKB determinations, different antagonist concentrations
`were used according to the affinity of the compound under
`study; concentrations ranged from 10 nM (for the highest
`affinity antagonists) to 10 íM (for lower affinity antagonists).
`In the gerbil foot tap model, gerbils were orally treated with
`antagonist at 5 mmol/kg 6 h prior to administration of ASMSP
`agonist (100 pmol). The CNS potency was then monitored by
`comparing the degree of foot tapping response with control
`animals that were treated with antagonist vehicle only.
`18
`Bioavailability Analysis. Compounds were administered
`to dog (n ) 3) at 1-10 ímol/kg by iv bolus injection or at 10-
`100 ímol/kg orally as a solution in 75% polyethylene glycol
`400 in normal saline. Blood samples were taken via surgically
`implanted cannula or by venipuncture over a 24 h period, and
`plasma was analyzed for unchanged compound by high-
`pressure liquid chromatography-mass spectroscopy (LC-MS).
`Kinetic Analysis. Solutions of each isolated atropisomer
`(at approximately 1.4 uM) were incubated in 10 mM buffer
`solution in the presence of 20% 2-methoxyethanol. The buffer
`solution was prepared using sodium phosphate or sodium
`acetate with ionic strength adjusted to 100 mM using