|Home | About | Journals | Submit | Contact Us | Français|
A series of novel high affinity opioid receptor ligands have been made whereby the phenolic-OH group of nalbuphine, naltrexone methiodide, 6-desoxonaltrexone, hydromorphone and naltrindole was replaced by a carboxamido group and the furan ring was opened to the corresponding 4-OH derivatives. These furan ring “open” derivatives display very high affinity for μ and κ receptors and much less affinity for δ. The observation that these target compounds have much higher receptor affinity than the corresponding ring “closed” carboxamides significantly strengthens our underlying pharmacophore hypothesis concerning the bioactive conformation of the carboxamide group.
We recently reported the synthesis and exceedingly high μ opioid receptor binding affinity (Ki = 0.052 nM) of a novel derivative 1 of 3-desoxy-3-carboxamidonaltrexone 2.1 The design of 1 was based a strategy whereby the 3-carboxamido group was stabilized in the putative bioactive conformation 1a via an intramolecular H-bond to the adjacent 4-OH donor. The rationale behind this pharmacophore hypothesis arose from the observation that carboxamide derivative 2 had much lower binding affinity than naltrexone (3).2 This result was in conflict with our other SAR studies where the OH → CONH2 switch on non-4,5α-epoxymorphinan core opioid structures (e.g., 2,6-methano-3-benzazocines and morphinans) resulted in sustained or enhanced binding affinity.1-3 We recently reported studies where the OH → CONH2 switch was performed on fifteen additional diverse opioid core structures; results were entirely consistent with our earlier work.4 These SAR data coupled with the observation that the proton NMR (CDCl3) spectrum of 2 revealed a strong H-bond between the carboxamide NH (as donor) and the neighboring ether oxygen (i.e., 2a) led us to reason that the putative carboxamide bioactive conformation was that as shown in 2 rather than 2a and the compound must pay an energy penalty to adopt the putative bioactive conformation 2 resulting in lower affinity.1 We also provided strong proton NMR evidence that a) the intramolecular H-bond of 1a was a consequence of the carboxamide acting as acceptor and not as donor (i.e., 1b) and b) the benefit of the 4-OH of 1 was to stabilize the putative bioactive conformation 1a and not via direct contact with the receptor.1
X-ray crystal structures were recently obtained for compounds 1 (as the HCl salt; CCDC-710249) and 2 (CCDC-710250).5 Stick representations are shown in Figure 1. The conformations of the two compounds in the solid state are very similar to those we previously proposed in CDCl3 solution, namely a) the presence of an intramolecular H-bond in 1 between the carboxamide O and a donor H on the 4-hydroxyl group and b) the presence of an intramolecular H-bond in 2 between the carboxamide NH and the ether O of the furan ring bridge. These combined NMR and X-ray physical data strongly corroborate our bioactive conformation hypothesis as outlined above.
We now report the syntheses of additional examples of 3-desoxy-3-carboxamido-4-hydroxy opioids related to 1. The main goal of this study is to confirm the pharmacophore hypothesis that was based on analysis of limited opioid receptor binding previously reported.1 The phenolic-OH-containing opioids nalbuphine (6),6 naltrexone methiodide (10),7,8 6-desoxonaltrexone (13),9 hydromorphone (16)10 and naltrindole (19)11 were chosen as substrates for modification to the corresponding 3-desoxy-3-carboxamido-4-hydroxy derivatives related to 1 (i.e., furan ring “open”) and the 3-desoxy-3-carboxamido derivatives analogous to 2 (i.e., furan ring “closed”).
Target compounds related to nalbuphine (6)6 were made using the methodology described in Schemes Schemes11 and and2.2. Compound 4 having the furan ring “open” form, was made in 17% yield by the reduction of keto derivative 201 with NaBH4 in MeOH (Scheme 1). This reaction also provided the 6-β-ol analogue 7 (35%); the two isomeric alcohols were easily separated by silica gel flash chromatography. As shown in Scheme 2, carboxamide target compound 5 with the furan ring “closed” was made in 79% yield via the partial hydrolysis of nitrile intermediate 211 using KOH in refluxing t-BuOH.
Target compound 8, the “open” carboxamide having the naltrexone methiodide core structure 10 (see Table 1 for structure), was made in 60% yield by heating 1 at 70 °C in a sealed tube for 4 d with 10 equivalents of CH3I in acetone (Scheme 3). The stereochemistry of the quaternary nitrogen center of 8 was assigned (R)-using 2D NOESY NMR (DMSO-d6, 500MHz, mixing time = 0.6 sec, relax delay = 0.9 sec). A cross peak was observed between the proton of 14-OH group and the protons of CH3 group indicating the CH3 group occupies the axial conformation with respect the 6-membered piperidine ring. This stereochemical assignment is consistent with recent synthetic studies centered around the N-methylation of derivatives of naltrexone.7 In similar fashion, compounds 2 and 3 (naltrexone) were converted to target compounds 9 (43%) and 10 (41%), respectively; stereochemistry of the newly introduced center of chirality was also assigned as (R)- using 2D NOESY NMR. The FDA-approved drug methylnaltrexone,7,8 is the N-methyl quaternary bromide salt of naltrexone. It is unclear from the literature whether the quaternized nitrogen is (S)- or (R)- or if it is a mixture of the two diastereomers.7,8 Nevertheless we prepared naltrexone methiodide 10 and used it as a comparator to carboxamide target compounds 8 and 9 (see Table 1 for structures) since all three share the same stereochemistry.
Target compounds 11 and 12 related to 6-desoxonaltrexone (13)9 were made using methodologies shown in Schemes Schemes44 and and5.5. The known 3-cyano derivative 2212 of naltrexone was treated with zinc dust, 37% HCl, in refluxing acetic acid to give desired target compound 11 in 22% yield along with 1 (Scheme 4). The corresponding ring “closed” carboxamide 12 was prepared by a multi-step procedure shown in Scheme 5. The 6-keto group of naltrexone (3) was reduced to the known 6-desoxonaltrexone 139 in 62% yield using standard Wolff-Kishner conditions of hydrazine hydrate, diethylene glycol and KOH at reflux. Triflate ester 23 was then made in 98% yield by treating 13 with PhN(SO2CF3)2 and triethylamine in CH2Cl2. Compound 23 was converted to nitrile 24 in 95% yield using Zn(CN)2, Pd(PPh3)4, in DMF and partial hydrolysis of 24 using KOH in refluxing t-BuOH provided target compound 12 in 95% yield.
Carboxamide target compounds 14 and 15 related to hydromorphone 1610 were made using the procedure outlined in Scheme 6. The 3-triflate ester 2513 of morphine was converted to nitrile 2614 in 66% yield using Zn(CN)2, Pd(PPh3)4, in DMF. Selective reduction of the double bond of 26 to dihydro derivative 27 was accomplished in quantitative yield using H2, 10% Pd/C in MeOH. Oxidation of 27 using standard Swern conditions provided 28 in 92% yield. Nitrile derivative 28 was then converted to target compound 14 in 63% yield using Zinc dust, NH4Cl in EtOH and to target compound 15 in 85% yield using KOH in t-BuOH at reflux.
The naltrindole-based carboxamide target compound 17 was made in 31% yield by treating 1 with PhNHNH2, p-TsOH in refluxing EtOH (Scheme 7). This procedure is similar to that used to make naltrindole from naltrexone.11 The carboxamide variant 18 was prepared from naltrindole using procedures similar to those previously described.1-3 Naltrindole (19)11 was first converted to its triflate ester 29 in 56% yield using PhN(SO2CF3)2, Et3N in CH2Cl2. Compound 29 was then subjected to CO, NH3, Pd(OAc)2 and DPPF in DMSO to provide target compound 18 in 27% yield.
Both furan ring “open” and “closed” carboxamide target compounds as well as their phenolic-OH counterparts were evaluated for their affinity for μ, δ and κ opioid receptors. Binding data are detailed in Table 1. Membrane protein from CHO cells that stably expressed one type of the human opioid receptor was used.15,16 The objectives of this study are to compare the binding affinities of a) the “open” to the “closed” form in each carboxamide pair, and secondarily b) the “closed” carboxamide target compounds to the corresponding parent phenolic-OH opioids.
Consistent with our earlier observations for naltrexone derivatives 1 and 2,1 there is a convincing trend in the SAR that the ring “open” carboxamide partner has much higher affinity for μ, δ and κ opioid receptors than the corresponding ring “closed” carboxamide. As shown in Table 1, against μ, the “open” derivatives 4 (nalbuphine core), 8 (naltrexone methiodide core), 11 (6-desoxonaltrexone core), 14 (hydromorphone core) and 17 (naltrindole core) have 7-, 28-, 16-, 4- and 90-fold higher affinity, respectively, than the corresponding furan ring “closed” carboxamides, 5, 9, 12, 15, and 18. Doing the identical comparison for δ, the increase in potency is 2-, >35-, 150-, 31- and 12-fold. For the κ receptor, the increase in binding affinity was 27-, 55- 8- and 58-fold for “open” derivatives 8, 11, 14 and 17; however, for the pair with a nalbuphine core, the “open” analogue 4 had 20-fold lower affinity than the “closed” form 5. For 14 of 15 comparisons of ring “open” to “closed” carboxamides (5 novel pairs against 3 receptor subtypes), the observation that the “open” analogues have much higher affinity strongly corroborates our pharmacophore hypothesis regarding carboxamide bioactive conformation of naltrexone as outlined earlier in this letter. For the single exception, we do not have an explanation as to why the nalbuphine-based pair 4 and 5 against κ does not follow this SAR trend. Absolute affinity of ring “open” target compounds 4, 8, 11, 14, and 17 for μ is very high; Ki values were all subnanomolar (0.16 to 0.52 nM) except for compound 8, a derivative of naltrexone methiodide which was slightly lower (Ki = 1.3 nM). This latter result is not surprising since the affinity of parent OH compound 10 is also in the single digit range (Ki = 2.0 nM). Compound 7, obtained as a byproduct, is a diastereomer of nalbuphine-derived target compound 4 and displays outstanding affinity for μ (Ki = 0.072 nM).
With the exception of naltrindole-derived target compound 17, ring “open” carboxamides 4, 7, 8, 11, and 14 have much lower affinity for δ than μ where Ki values ranged between 3.9 and 280 nM. However for target compound 17, not only is affinity for δ extremely high (Ki = 0.025 nM), it has 6-fold higher affinity than the well known δ-selective parent, naltrindole (19).11 Three ring “open” carboxamides, 7, 11 and 17, have very high affinity for the κ receptor (Ki = 0.34, 0.29 nM and 0.81 nM, respectively); the other carboxamides in this class (4, 8 and 14) have somewhat lower affinity with Ki values in the 2.3-9.0 nM range.
In addressing the secondary objective of this study, we find that novel ring “closed” carboxamides 9, 12, 15 and 18 display lower affinities for the three receptors than their corresponding phenolic-OH counterparts 10, 13, 16 and 19. These reductions in binding affinity are between 4- and 150-fold except for the naltrindole-based carboxamide 18 which has nearly comparable (2-fold lower) affinity for δ as naltrindole itself. These findings are consistent with our previous SAR studies2,4 and add strong support for our underlying pharmacophore hypothesis. There is an important exception, however. In the nalbuphine case, the “closed” carboxamide 5 has much higher affinity for δ and κ receptors than nalbuphine (6) by factors of 4-fold and 6-fold, respectively. Against μ, the two have comparable affinity. We have no rationale why the 3-desoxy-3-carboxamido analogue 5 of nalbuphine (6) does not follow this trend in SARs.
Intrinsic opioid-receptor mediated activity for high affinity (Ki values < 5 nM) carboxamide-containing ligands was determined using [35S]GTPγS binding assays at μ, δ and/or κ opioid receptors; results are shown in Table 2. In cases where the parent phenol had high affinity for a particular receptor, they were also assayed. Procedures similar to those previously reported were used.18 Like naltrexone (3), a μ antagonist with a small amount of μ agonist effects in the [35S]GTPγS binding assay,19 its two carboxamide derivatives 1 and 2 were also found to be potent antagonists at μ; antagonist potency at μ correlated reasonably well with binding affinities. Naltrexone and its “closed” carboxamide derivative 2 produced a weak stimulation of [35S]GTPγS binding mediated by the μ receptor. The higher affinity of the “open” ring carboxamide derivative 1 of naltrexone for the μ receptor may account for its lack of agonist effect in the [35S]GTPγS binding assay. At δ and κ receptors, both 1 and naltrexone were found to be mixed agonists/antagonists. For compounds 4-7 having the nalbuphine core, 4, 6 and 7 were mixed agonists/antagonists at μ of varying potency while 5 was a weak antagonist. Compound 7 had very high affinity in the receptor binding assay, and it was primarily an antagonist at the μ receptor. At the κ receptor, 4 and 7 were weak agonists/antagonists while 5 and 6 were agonists; potencies in this case correlated reasonable well with binding affinities. Naltrexone methiodide (10) and its “open” carboxamide analogue 8 were found to be antagonists at both μ and κ; potencies for the two were similar at μ and somewhat divergent at κ. 6-Desoxonaltrexone (13) and both of its carboxamido analogues 11 and 12, were all antagonists at μ of varying potencies. Compound 11 was an antagonist at δ and a mixed agonist/antagonist at the κ receptor. Hydromorphone (16), a prototypic μ agonist, shares this agonist profile with its two carboxamide analogues, 14 and 15 and potencies in the GTPγS binding assay correlated well with binding affinities at μ. At the κ receptor, both 14 and 16 were also agonists having similar potency that correlated well with binding affinity. Comparing the known δ-selective antagonist naltrindole (19)11 to carboxamide analogues 17 and 18, the data show that all three were antagonists at δ in our assays. Good correlation was observed between GTPγS and binding potencies at δ. Due to its relatively high affinity to μ and κ compared to 18 and 19, “open” carboxamide derivative 17 was evaluated at these receptors where it displayed an antagonist profile.
A series of novel carboxamido-substituted, furan ring “open” derivatives of nalbuphine, naltrexone methiodide, 6-desoxonaltrexone, hydromorphone and naltrindole generally display very high affinity for opioid receptors and are much more potent than the corresponding ring “closed” carboxamides. These data significantly strengthen our underlying pharmacophore hypothesis that the bioactive conformation of the carboxamide group of, for example, 1 and 2 is that represented by structures 1a and 2. Further support of the hypothesis was gained by our observation that binding affinities of the furan ring “closed” carboxamides 9, 12, 15, and 18 are lower than their phenolic-OH counterparts 10, 13, 16 and 19. It is interesting to note that that the only significant exceptions seen in both SAR trends (“open” carboxamide is preferred over “closed” and phenolic-OH is preferred over “closed” carboxamide) are for the nalbuphine core structures 4-6. Compound 4 against κ, but not at μ or δ, is the only “open” carboxamide that displays lower affinity than the corresponding “closed” form 5. Additionally, compound 5 against δ and κ is the only “closed” form to have higher binding affinity than the corresponding phenolic-OH parent 6. At this point, we can not offer any explanation for this divergence in SAR. For those analogues studied in [35S]GTPγS binding assays, a trend was observed where the “open” and “closed” carboxamide analogues of a particular phenolic-OH opioid displayed nearly the same functional activity as the OH counterpart. This trend was the strongest when studying the receptor type for which the parent OH opioid had the highest binding affinity (e.g., hydromorphone core compounds at μ and naltrindole core derivatives at δ). For certain cores at certain receptors (e.g., hydromorphone core compounds at μ and naltrindole core derivatives at δ), potency in the GTPγS binding assay correlated nicely with binding affinity. However, for other cores (e.g., nalbuphine) a poor correlation was observed. The value of the SAR data generated in this study is not only the strengthening of our underlying pharmacophore hypothesis, but also in the identification of a number of novel high affinity opioid receptor ligands. Additional research in this area is ongoing in our laboratories and will be the subject of future communications.
We gratefully acknowledge the contributions of Rensselaer’s mass spectroscopist Dr. Dmitri Zagorevski and the technical assistance provided by Brian I. Knapp of the University of Rochester. We thank Dr. Douglas M. Ho of Harvard University for the X-ray crystallographic data. Funding of this research was from NIDA (DA12180 and KO5-DA00360) and the NSF (Agilent 1100 series LC/MSD system). We also acknowledge financial support from AMRI.
Publisher's Disclaimer: This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final citable form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.