Diastereoselective One-Pot Synthesis of 7- and 8-Substituted 5-Phenylmorphans

Abstract

Novel 7- and 8-alkyl and aryl substituted 5-phenylmorphans were synthesized from substituted allyl halides and N-benzyl-4-aryl-1,2,3,6-tetrahydropyridine by a highly efficient and diastereoselective reaction series, “one-pot” alkylation and ene-imine cyclization followed by sodium borohydride reduction. Mild cyclization conditions gave the desired substituted 5-phenylmorphans in good yield as a single diastereomer.

The molecular structure of the 5-phenylmorphans (1, Fig. 1) was conceptualized¹ as a structurally simplified fragment of morphine or heroin (Fig. 1), and some N-substituted 5-phenylmorphans were found to have morphine-like activity.² Recently Hiebel et al.,³ synthesized a C9βOH N-phenethyl-5-phenylmorphan (the 1R,5R,9S-enantiomer of 2, Fig. 1) that had extremely high affinity for the μ-opioid receptor and was far more potent than morphine in vivo; its epimer (1R,5R,9R) had 230 fold less affinity. This is a remarkable effect of the stereochemistry at a single OH group. In order to determine what pharmacological profile would be conferred by substituents at the C-7 or C-8 positions we needed to find a synthetic path to these less accessible compounds. Only 7-amino⁴ and 6,7- and 7,8-fused indole derivatives⁵ synthesized from 7-keto-5-phenylmorphan, and C3 and C7-alkyl or alkenyl 5-phenylmorphans have been reported thus far, the latter by Zimmerman⁶ who used a phosphoric acid/formic acid mixture and did not assign the stereochemistry of his products.

Figure 1.

5-Phenylmorphan fragment ±-1 of morphine or heroin, and potent analogue ±-2

Although synthetic strategies have been developed for 5-phenylmorphans,^1,2,7,8 we hoped to find a concise synthesis of the desired target molecules. We thought that a 7-keto derivative might be used as an intermediate, but found that the 7-keto group had unexpectedly low reactivity toward C-C bond forming reactions such as Wittig olefination.⁹ Among available strategies, a 3-step synthesis using allyl bromide reported by Evans et al., brought our attention to the possibility of applying that methodology in a novel way to prepare 6-, 7-, and 8- substituted 5-phenylmorphans.^10,11 However, after the original report,¹⁰ the reaction scope and mechanism were not studied. We decided to investigate the methodology of Evans et al., to see whether it could be applied to the synthesis of the C-7 or C-8 substituted 5-phenylmorphans.

In the original report¹⁰ the N-methyl tetrahydropiperidine 3i was alkylated using allyl bromide 5a to generate an all-carbon quaternary center (6a, Scheme 1). A mixture of neat acids (1:1 HCO₂H and H₃PO₄) was used for the cyclization of the crude alkylated product 6a (where R₁=R₂=R₃=H). The cyclized enamine 7a (R₁=R₂=R₃=H) of Evans, et al., was then isolated from strong acid and reduced. Although the cyclization worked well, the reaction was slow (66 h at rt), there were problems isolating the cyclized product from the highly acidic media, and no substituted allyl bromides were tested.

Scheme 1.

Substituted 5-phenylmorphans via series of alkylation, cyclization, and (NaBH₄) reduction reactions

For our functionalized allylic substrates we explored more practical and efficient cyclization conditions and examined milder acids in organic solvents. To test the scope of the reaction, β-phenyl-substituted allyl bromide 5b was used. The original conditions were initially applied, and the desired product 8b was obtained in moderate yield (entry 1, Table 1). We then screened organic acids for the cyclization. p-Toluenesulfonic acid (p-TsOH) in refluxing toluene cyclized the ene-enamine 6b giving a single diastereomer in higher yields within a shorter time than with the mixed acid conditions (entry 2, Table 1). In contrast to the acidic conditions of Evans et al., the crude cyclized enamine 7b under our modified conditions could be directly reduced with sodium borohydride to give the desired amine 8b.⁸ Thus, the use of p-TsOH eliminated the necessity of removing the acid before the reduction. Not only did the use of a stoichiometric amount of the reagent at elevated temperature work for the reaction, but a catalytic amount of p-TsOH at room temperature also gave the desired product with only a small reduction in yield (entry 3, Table 1). Other, stronger, acids also gave the desired product at room temperature, albeit with lower yields (entries 4-5, Table 1).

Table 1.

Alternative cyclization conditions

entries	cyclization conditions	temperature (°C), time (in days)	yield (%)a
1	HCO2H/H3PO4 (1:1)	rt, 7	48
2	p-T s O H ( 2 equiv), toluene	reflux, 2b	56
3	p-T s O H ( 0 . 2 equiv), CH2Cl2	rt, 3	41
4	T f O H ( 0 . 2 equiv), CH2Cl2	rt, 3	38
5	3 0 % T F A i n CH2Cl2	rt, 3	44

^aIsolated yields of 8b (3 steps overall).;

^bProlonged reaction time did not increase the yield.

Using our optimized conditions, other substituted allyl bromides and chlorides were tested (5c-h, Scheme 1). The unsubstituted phenyl moiety 5b and the bromo- and chloro-compounds 5c,d all underwent the desired reaction in good yields (entries 1-3, Table 2). The simple methyl substituted 5e also worked well, providing the desired 7-methyl 5-phenylmorphan 8e (entry 4, Table 2).

Table 2.

Synthesis of 5-phenylmorphans 8b-j using substituted allyl halides 5b-i

entry	allyl halide	yield (%)	product
1		56
2		49
3		41
4		65
5		64
6		52
7		39a,b
8		60 (0)c
9		36d (31)e

^aDiastereomeric mixture.;

^b20% of 8f was also obtained.;

^cWhen the original reaction conditions (HCO₂H/H₃PO₄, 1:1) were used, the desired product was not obtained;

^dNot isolated - NMR and MS indicated product;

^eYield of the cyclized enamine 7i;

^fYield of 8j after reduction of 7j.

All of these R₂-(β)-substituted allyl bromides gave only single diastereomers 8b-e. Reaction with the R₁-(γ)-methyl substituted (E)-1-bromobut-2-ene, 5f, provided 8-methyl substituted 5-phenylmorphan 8f as a single diastereomer in good yield (entry 5, Table 2). Moreover, cyclohexyl-fused 5-phenylmorphan 8g was synthesized as a single compound (entry 6, Table 2).

However, the R₃-(α)-methyl substituted compound (5h) gave 8h as an inseparable diasteromeric mixture of the C6-methyl isomers (entry 7, Table 2) with 20% of the diastereomerically pure C8-methyl isomer 8f arising from γ-alkylation. When the R₁-(γ)-disubstituted compound (1-bromo-3-methylbut-2-ene) 5i was used, a 5-membered cyclization product 8i formed. Under the original conditions of Evans, et al., a mixture of cyclization adducts were obtained from 5i. Interestingly, the reaction of 2,5-dimethoxy substituted compound 3iii with 3-bromo-2-methylprop-1-ene (5e) gave enamine 7j (entry 9, Table 2), structurally similar to an intermediate in the synthesis of a para-a oxide-bridged phenylmorphan.¹² The enamine was further reduced to obtain the amine 8j (entry 9, Table 2).

Allyl bromides such as cinnamyl bromide 5k, bromocyclohex-2-ene 5l, and O-TBDPS protected substrate 5m did not undergo the cyclization. Moreover, N-carbethoxy protected substrate 3iv instead of N-Bn, did not undergo the desired cyclization reaction. In order to examine the relative stereochemistry of the substituents in the cyclized products, representative products 8b,f,g were converted into their phenolic relatives 9b,f,g using known procedures.¹³ Crystalline HBr salts of 9b,f,g were obtained for X-ray crystallographic structure analyses to determine the relative stereochemistry of substituents at C-7 and C-8 with regard to the piperidine ring (Fig. 2). For clarity, if a substituent is on the same side of the cyclohexane ring as the piperidine ring, it is called cis, otherwise it is trans. X-Ray analyses showed that substituents at C-7 were trans-oriented and those at C-8 were cis. Substituents such as a methyl group did not show selectivity at C-6. These selectivities provided information that enabled us to postulate a possible mechanism for this highly selective cyclization.

Figure 2.

Ortep plots of 9b, f, g (one enantiomer was drawn for 9f)

In the report of Evans, et al.,¹⁰ it was briefly noted that this cyclization might occur by ene-imine cyclization followed by a hydride shift. Although this appeared to reasonably explain the formation of the obtained products, there was no evidence for this hypothesis nor can it explain why this reaction was highly diastereoselective for β-and γ-substituted allylic groups with ene-enamine substrates, which are not α-substituted.

We hypothesized that the exceptional diastereoselectivities were due to (1) alkylation on the Re-face of the olefin that occurred relatively fast because of the more favored Zimmerman-Traxler chair-like TS.¹⁰ It would have to be the most rapid alkylation to obtain, as was found, a single stereoisomer at the C-8-position, as in 8f-g; (2) the hydride shift after cyclization must only occur intramolecularly to obtain the stereochemistry found at the 7-position, i.e., the trans products 8b-e, 8g. If the hydride came from an intermolecular source, the opposite stereochemistry would have been found; moreover, the intramolecular hydride shift is the most likely way to generate an imine after the cyclization; (3) if the chiral center at C-6 was generated via a non-selective alkylation, it would give the diastereomeric mixture 8h (Scheme 2).

Scheme 2.

Possible mechanism of the highly stereoselective cyclization

In summary, our improved series of reactions, alkylation, cyclization, then reduction, were successfully applied to the synthesis of new 7- and 8-substituted 5-phenylmorphans in good yields as single diastereomers. The stereochemistry of the products was confirmed by X-ray crystallographic analyses. A possible mechanism, intramolecular hydride shift as key aspects of the highly diastereoselective cyclization that gave trans selectivity at C-7 and cis selectivity at C-8. Further evidence to support our hypothesis can come from the use of more diversely substituted-allyl bromides, and these will be tested in future work. Also, pharmacological data for structure activity relationship (SAR) studies will be carried out and reported in subsequent publications using compounds 8 and 9 with N-substituents other than N-benzyl or N-methyl.