Morphinan Alkaloids and Their Transformations: A Historical Perspective of a Century of Opioid Research in Hungary ^†

1. Introduction

The word opium has its origin in the ancient Greek ὄπιον (ópion), which the referred to the juice of any plant, belying the profound importance of specifically the juice of Papaver somniferum in ancient and contemporary medicine. The German apothecary and pioneer of alkaloid chemistry Friedrich Wilhelm Adam Sertürner (1783–1841) began working on opium extraction in 1804, first describing a crude preparation that he designated principium somniferum and later morphium, after Morphius, the Greek god of dreams. His isolation of pure morphine (1, Figure 1.) from raw opium was the first of any alkaloid [1]. Throughout history, opium, morphine, and its semisynthetic derivatives have been a double-edged sword, being indispensable in medicine, while also bringing crises of dependency and overdose. The touting of oxycodone (14-hydroxydihydrocodeinone) as an analgesic with supposedly superior safety profile was surely a factor in the exacerbation of the opioid crisis in some countries [2], now made worse by the availability of synthetic compounds such of astonishing potency as fentanyl and the even stronger nitazines.

Alkaloida Chemical Company (Alkaloida Vegyészeti Gyár) was founded by the Kabay brothers János, Péter and József in Bűdszentmihály (former Tiszabűd and Szentmihály; since 1952 Tiszavasvári) in Hungary [3,4], with signing of the foundation agreement on September 10, 1926 [5]. The principal founder, pharmacist János Kabay (1896-1936) is notable for his 1925 patent [6] on the innovative «green method» for extracting directly opiate alkaloids from parts of the poppy plant, thus circumventing the inefficient and labour-intensive process of opium harvesting by resin collection collection of the resin [6]. Further improvements in 1931 enabled the commercially viable extraction of the key alkaloids, morphine (1a) and an alkaloid mixture of mainly codeine (2a) and narcotine/noscapine (8) from the previously worthless straw and dried capsules of the poppy plant («dry method») [6]. The foundation of the Hungarian morphine alkaloid-industry, Kabay’s process was adopted the world over [7,8].

Kossuth Lajos University (Kossuth Lajos Tudományegyetem, KLTE; today University of Debrecen, Hungary) established its Institute of Organic Chemistry in 1947. The early 1950s saw a strong initiative by Ministries of the Hungarian Heavy Industries and Education to form regionally important connections between new industrial centres and nearby universities. The long-standing collaboration between the Institute of Organic Chemistry of KLTE and Alkaloida arose as a direct result of these efforts. Rezső Bognár (1913-1990) [9], a former student of the Emil Fischer (1852–1919) associate Géza Zemplén (1883–1956), received in 1950 the appointment as head of the Institute of Organic Chemistry. Early in his tenure, Bognár placed special emphasis on joint efforts with Alkaloida specialists to optimize the Kabay-technology, aiming to obtain higher yields of the valuable opiate alkaloids codeine, thebaine, narceine, neopine, and narcotoline [10,11,12,13,14]. During 1979–1995, Sándor Makleit (1930–2012) became head of the Department of Organic Chemistry and its morphine alkaloid research group, in which capacity he coordinated interactions between KLTE and Alkaloida. Upon Makleit’s retirement, Sándor Berényi (1947–2019) redirected research towards an emphasis on aporphine-chemistry via the rearrangement of morphinans. The contributions of László Szilágyi (NMR), Gyula Batta (NMR), and Zoltán Dinya (IR, MS) were essential in the field of structural analysis of the new synthetic morphinan derivatives. Sándor Hosztafi (1952–), formerly of the Alkaloida Chemical Company, has contributed enormously to the field of semi-synthetic morphinans, aporphine derivatives and other poppy alkaloids, and remains active at the Department of Pharmaceutical Chemistry of Semmelweis University, Budapest.

In this account, we cannot omit mentioning the long-standing cooperation between the Debrecen morphine alkaloid research group and Semmelweis University (Department of Pharmacology, Budapest) for the pharmacological analysis of newly synthesized opioid receptor (OR) ligands. These investigations were first directed by József Knoll (1925–2018) and Zsuzsanna Fürst (1939–). The biochemical characterization of new semi synthetic alkaloids was performed under guidance of Anna Borsodi, Sándor Benyhe and Géza Tóth at the Biological Research Center of the Hungarian Academy of Sciences (MTA Szegedi Biológiai Központ, Biokémiai Intézet, Szeged), and at Gedeon Richter Pharmaceuticals (Richter Gedeon Vegyészeti Gyár, Budapest).

Given this historical and personal background, we now present a comprehensive account of opiate alkaloid chemistry, placing our focus on the extensive contributions in recent decades of the Hungarian research group in morphine alkaloids in Debrecen and at the Alkaloida Chemical Company in the field of morphinan syntheses, as part of a concerted search for novel structures and synthetic approaches. We begin with a brief presentation of opiate nomenclature, stereochemistry, and the biosynthesis of morphinan alkaloids.

2. Chemistry

2.1. Poppy Alkaloids

The diverse alkaloids of Papaver somniferum are categorized based on their chemical structures [15,16,17,18,19]. The well-known of these are benzylisoquinoline- (I), morphinan- (II), aporphine- (III), benzo[c]phenantridine- (IV), papaverrubine- (V), narceine-typ- (VI), protoberberine- (VII), protopine- (VIII), phthalideisoquinoline (IX), with ground scaffolds presented in Figure 2.

2.2. The Stereochemistry of Morphinans

There are two theoretical sterical arrangements in the case of decalin and morphinans (Figure 3), depending on the ring-junction of the A/B-ring and B/C-ring, respectively [17]. For cis-decalins, the H-1 and H-6 hydrogens and for cis-morphinans, the H-14 and the C¹³ – C¹⁵ bond are on the same face. In trans-decalins the H-1 and H-6 hydrogens and in B/C-trans-morphinans, the H-14 and the C¹³-C¹⁵ bond are on the opposite face. In naturally- occurring morphinan alkaloids, the junction of the B- and C-rings is a B/C-cis fusion analogous to cis-decalin as shown in structure IIa of Figure 3, where IIb presents the structure of a B/C-trans-morphinan. It is not possible to convert a morphinan (B/C-cis-morphinan) into an isomorphinan (B/C-trans-morphinan) by rotation of molecular parts without breaking chemical bonds. The absolute configurations of the C-14 chiral carbon differ for the isomers: 14R (morphinan) and 14S (isomorphinan).

Morphine (1a, Figure 4) and a significant number of its derivatives have a T-shaped three-dimensional molecular geometry. The isomers of morphine (1a, Figure 4, Table 1) and codeine (2a) have five chiral carbons: positions 5, 6, 9, 13, and 14. In 1a and 2a, the carbon-carbon double bond is situated in position C7-C8 (Δ^7,8), the ring-C has a flattened boat conformation, and the allylic hydroxyl group in position C-6 is pseudo-equatorial. The C6 in 1a and 2a has absolute configuration S. In the C-6 epimer molecules isomorphine (1b, α-isomorphine, Figure 4) and isocodeine (2b), the 6-hydroxyl group is located above the plane of the ring-C in a pseudo-axial position. The C-6 in 1b and 2b has absolute configuration R.

Further Δ^6,7 isomers for morphine (1a) and codeine (2a) are known as allopseudomorphine (1c, also known as β-isomorphine), γ-isomorphine (1d), allopseudocodeine (2c) and pseudocodeine (2d). In these compounds, the alcoholic hydroxyl group connects to the position-8 of the ring-C. Of the ring-C carbons, C-5, C-6, C-7, C-8 and C-13 are in the same plane, while C-14 is located above them. The ring-C of the compounds has an envelope conformation. In allopseudomorphine (1c) and allopseudocodeine (2c) the 8-OH is pseudo-axial and the configuration of C-8 is R. The two isomers with Δ^6,7 double bond and pseudo-equatorial 8-hydroxyl group, γ-isomorphine (1d) and pseudocodeine (2d), have 8S absolute configuration. Figure 4 and Table 1 provide an overview of the structure and the absolute configuration of the parent compounds.

The neopine (3a), isoneopine (3b), neomorphine (3c) and neoisomorphine (3d) derivatives also have a Δ^8,14 unsaturated bond in ring-C. These compounds have only four asymmetric carbon atoms (5, 6, 9 and 13), since the Δ^8,14 double bond C-14 is no longer asymmetric. The C-ring is rigid and of half chair conformation. In the cases of neopine (3a) and neomorphine (3c), the 6-hydroxy group is in pseudo-axial position and the carbon-6 has absolute configuration S. In compounds of the 6-beta series, like isoneopine (3b) and isoneomorphine (3d), the 6-OH is pseudo-equatorial and the carbon-6 has configuration R. The poppy alkaloids with a conjugated 6,8-dien system in ring-C: thebaine (4) and oripavine (5), have only three asymmetric carbons, with the absolute configurations 5R, 9R, and 13S.

2.3. Biosynthesis of Morphinan Alkaloids

A suite of five O-methyltransferases and two N-methyltransferase enzymes contribute to benzyisoquioline alkaloid (BIA) biosynthesis [20] in five genera of flowering plants, including Papaveraceae (poppy), and Ranunculaceae (buttercup). The phytochemical synthesis of morphine alkaloids begins with L-tyrosine (10) via a branching pathway that yields on one side tyramine (11) through the action of L-tyrosine decarboxylase, which is a dimeric pyridoxal-phosphate dependent enzyme [21]. A phenolase enzyme then transforms the tyramine product (11) to dopamine (13). In the right-hand branch, L-tyrosine (10) gives rise to the keto acid 4-hydroxyphenylpyruvate (12) via L-tyrosine deaminase [22], and then the corresponding acetaldehyde (14) via a specific decarboxylase. Condensation of the two products (dopamine (13) and 4-hydroxyphenylacetaldehyde (14)) to form (S)-norclaurine (15) proceeds via an asymmetric Pictet−Spengler condensation, which is catalysed by the enzyme norclaurine synthase [23]. Mechanistic studies of the norcoclaurine synthase enzyme derived from Thalictum flavium (yellow meadow-rue) indicated a two-step cyclization of the putative iminium ion intermediate, which constitutes the first committed step in the formation of poppy BIAs.

Once formed, (S)-norclaurine (15, Figure 5.) undergoes a series of enzymatic methylations, first by norcoclaurin-6-O-methyltransferase in the presence of the co-substrate S-adenosylmethionine (SAM), which represents a key rate-limiting step in benzyisoquioline synthesis [24], followed by tetrahydrobenzylizochinoline-N-methyltransferase. The product (S)-N-methyl-coclaurine (17) undergoes ring-hydroxylation by a phenolase enzyme or a higher affinity P450-dependent monooxygenase (CYP80B1) in poppy [25], and then a further SAM-dependent O-methyltransferase reaction to give (S)-reticulin ((S)-19). Interestingly, reticulin in the traditional food Annona muricata (soursop) may be responsible for dopamine neuron degeneration resulting in a form of parkinsonism that is endemic in the island of Guadeloupe [26].

A specific NADPH-dependent reductase isomerizes (S)-reticuline ((S)-19) to (R)-reticuline ((R)-19). After ring closure of (R)-reticuline ((R)-19, Figure 6.) via an NADPH/O₂-dependent enzyme yielding salutidarine (7), an NADPH/NADP^⊕-dependent oxidoreductase yields (7S)-salutaridinol (21), which is then ring-acetylated by an acetyl coenzyme A-dependent enzyme; spontaneous de-acetylation gives thebaine (4, paramorphine), which gains its common name from the anciet Greco-Egyptian city of Thebes. In rat brain preparations, thebaine (4) has only weak binding affinity for ORs [27], and unlike the synthetic enantiomer (+)-thebaine [28], has little analgesic potency. Having been long-regarded as a useless side-product of poppy-extraction, thebaine (4) now has an annual global market value of USD 1 billion, serving as starting material in the industrial production of opiate agonists (e.g., oxycodone, oxymorphone), antagonists (Nal-compounds: naloxone, naltrexone, nalbuphine), and Diels-Alder type Bentley-compounds (buprenorphine, etorphine, diprenorphine). The ORs in brain present important targets for molecular imaging by positron emission tomography (PET), as presented in an extensive review [29]. In particular, thebaine (4) is a precursor for the radiosynthesis of various PET tracers, for OR imaging (see section 2.12.). It has been known since the 1970s that thebaine (4) predominates in Papaver bracteatum (the Armenian poppy), which contains only a small amount of other alkaloids [30,31]. Alternately, Zenk and researchers at Tasmanian Alkaloids obtained the morphine (1a) free Papaver somniferum plant designated as top1 (top: thebaine - oripavine – poppy) through genetic modification [32]. The genetic cultivar contains predominantly thebaine (4, 1.65 %) and oripavine (5, 0.43 %) but no morphine (1a, 0 %) or codeine (2a, 0 %).

Thebaine (4, Figure 7) yields morphine (1a) by two parallel pathways in the opium poppy. In one pathway, an enol ether hydrolase enzyme forms neopinone (23), which equilibrates to codeinone (24), an alkaloid with about one-third the analgesic potency of codeine (2a). An NADPH-dependent reductase converts codeinone (24) to codeine (2a), which is demethylated to yield the still more potently analgesic morphine (1a). In the other pathway, 3-O-demethylation of thebaine (4) yields oripavine (5), which is also a potent analgesic, albeit with high toxicity. Oripavine (5) is important industrially as the precursor for synthetic opiates (i.e., etorphine and buprenorphine); in the poppy plant, oripavine (5) undergoes enolether hydrolysis to yield morphinone (25), which has similar analgesic potency to that of codeine (2a). Conversion of morphinone (25) to morphine (1a) entails an NADPH-dependent codeinone reductase, various isoforms of which also catalyze the reduction of certain other intermediates of morphine (1a) biosynthesis [33]. Interestingly, morphine biosynthesis in the poppy plant entails compartmentation in several cell types [34,35]; the initial steps occur in sieve elements of the phloem, whereas the conversion of salutaridine (7) to thebaine (4) predominantly takes place in adjacent laticifers, where the main morphine accumulation occurs.

Raw opium is the dried latex from seedpods of the opium poppy, which naturally contains a mixture of thebaine alkaloids and intermediates, the composite of which contribute to its somnorific/analgesic effects and toxicity. The principle alkaloid in opium samples from south East Asia was morphine (11-23%), with lesser amounts of codeine (2-4%) and thebaine (1-3%) [36]. Poppy straw from various Papaver somniferum cultivars used in oil seed production in the Czech Republic had relatively invariant alkaloid concentrations across three consecutive harvests [37]. Cluster analysis distinguished cultivars with high morphine concentration (1.6%) and relatively low levels of other alkaloids, versus cultivars with the oppositive relationships. Comestible poppy seeds also contain various alkaloids, extending over a >100-fold concentration range, depending on the source [38]. Baking substantially reduces the alkaloid content of poppy seeds, such that it would be practically impossible to obtain intoxication from Bejgli, a traditional Hungarian Christmas cake, although there are case reports of opioid-dependent individuals consuming poppy seeds by the kilogram [39]. On the other hand, there are reports of lethal codeine (2a) /morphine (1a) overdose among poppy seed tea drinkers [40], and in drinkers of tea made with «home grown» dried seed pods [41]. Thebaine (4) is a pro-convulsant at high doses and poppy seeds with a particularly high content of thebaine (4, Figure 8.) caused serious neuromuscular toxicity poisoning in tea drinkers in Australia [42]. Thus, the toxicity of poppy products depends upon factors such as the particular cultivar, growing conditions, and mode of preparation.

The opioid receptors (ORs) [43] regulate numerous (patho)physiological processes being involved in pain modulation, euphoria, reward behaviours, and substance abuse, with involvement in the pathophysiology of various psychiatric disorders, epilepsy, and neurodegenerative conditions such Alzheimer’s disease (AD) [44]. Endogenous opioid peptide ligands and exogenous opioids activate several types of G-protein coupled (GPCR) receptors (μ-OR, δ-OR κ-OR, NOP). The concept of «biased agonism» underlies the search for OR ligands elicting therapeutic effects, but with lesser risk of toxicity and dependence [45,46]. In general OR agonists inhibit the enzyme adenyl cyclase (AC), thus decreasing the intracellular cAMP level, and thereby inhibiting the voltage-gated Ca^2⊕ ion channels while activating K^⊕ efflux. The result hyperpolarisation inhibits the presynaptic release of neurotransmitters (e.g., dopamine (DA), norepinephrine (NE), acetylcholine (ACh), γ-aminobutyric acid (GABA)), which produce a variety of effects including sedation, central analgesia, and euphoria.

Morphine (1a) yields a number of metabolites in humans by various metabolic pathways [47]. The most important morphine metabolites are M3G (26, Figure 9), and M6G (27), which form by glucuronidation catalysed by hepatic uridine 5’-diphosphoglucuronyltransferase (UGT). The sulfation of 1a by hepatic sulfotransferase (SULT1A3) results in morphine-3-sulfate (30, M3S), and morphine-6-sulfate (28, M6S). The N¹⁷-demethylation of 1a by cytochrome P450 (CYP450) gives normorphine (29), and the oxidative dimerization lead to pseudomorphine (6, 2,2’-bimorphine). The SAM-dependent 3-O-methylation of morphine (1a) to codeine (2a) can occur in liver, lung or kidney.

The chemical structures of some phase I and phase II metabolites of morphine are depicted in Figure 9.

2.4. The Makleit-Bognár Nomenclature

In 1968, Makleit and Bodnár [48] proposed the introduction of a new nomenclature for morphine derivatives. According to this nomenclature, naming of all ring-C substituted semisynthetic compounds derives from the isomers of codeine (2a-d, Figure 4, Figure 10). The Makleit-Bognár nomenclature requires only prior knowledge of names and stereochemistry of the basic four codeine isomers: codeine (2a), isocodeine (2b), allopseudocodeine (2c) and pseudocodeine (2d).

Makleit and Bognár suggested abandoning the previously used names for morphine isomers, α- (1b), β- (1c) and γ-isomorphine (1d), in favour of isomorphine (1b), allopseudomorphine (1c), and pseudomorphine (1d), all in analogy to the codeine isomer nomenclature. They also proposed adoption of the rational name “2,2’-bimorphine” to replace the old name of the poppy alkaloid bis-derivative pseudomorphine (6, Figure 1). This simplified nomenclature is generalizable for all in ring-C substituted derivatives, and faithfully reflects their structure and stereochemistry. Ergo, chemists can ascertain quickly and precisely the name and exact structure of any new derivatives, knowing the names and structures of the four codeine isomers (2a-d, Figure 4, Figure 10). We note that the position of a substituent below and above the plane ring-C differs for C-6 and C-8; if a C-6 substituent is pseudo-axial, it lies above of the plane of the ring-C, while a C-8 a pseudo-axial substituent is below the ring-C.

The term “deoxy-X” is a proposed term for all in ring-C modified derivatives, rendering redundant the descriptions (6α, 6β, 8α, 8β), as the name already specifies the exact structure of the compound. For interpretation of this nomenclature, we present numerous examples in Figure 11. Like the Maat-nomenclature system [49] in the field of the Diels-Alder adducts of morphinan-6,8 dienes [50], chemists have not universally adopted the rational Makleit-Bognár nomenclature [48], despite its considerable advantages.

2.5. Early Syntheses of Morphine Derivatives with Pharmaceutical Importance

Opiates are plant products formed by a series of enzymatic reactions as outlined above, whereas chemistry has built upon the rather sparse selection of poppy alkaloid scaffolds to generate countless derivatives. After elaborating the Kabay-Bognár technology for the isolation of morphine (1a) and its accompanying alkaloids from dry poppy straw and heads [10], the transformation of morphine (1a) to other active pharmaceutical ingredients came into focus at Alkaloida and the University of Debrecen.

Morphine (1a) was converted to codeine (2a) by methylation of the phenolic 3-hydroxyl group by aryl trimethylammonium hydroxides (Rodionov’s method [51,52]). Methods were elaborated for the synthesis of benzylmorphine (32, 7,8-didehydro-4,5α-epoxy-6α-hydroxy-17-methyl-3-phenylmethoxy-morphinan), ethylmorphine (33, 7,8-didehydro-4,5α-epoxy-6α-hydroxy-17-methyl-3-ethoxy-morphinan) and pholcodine (34, 7,8-didehydro-4,5-epoxy-6α-hydroxy-17-methyl-3-(2-(4-morpholinyl)ethoxy)-morphinan) [53] on an industrial scale starting from morphine (1a, Figure 12). The latter compound (34) became clinically significant because of its lesser toxicity and proclivity to induce respiratory depression in comparison to morphine (1a), and due to its more favourable sedative effect as compared to codeine (2a) or other morphine derivatives. Application of pholcodine (34) as an antitussive agent in the treatment of dry non-productive cough for young patients and adults has been widespread, and 34 was for a long time the second-most important active pharmaceutical ingredient (API) manufactured from morphine (1a) [8]. Neuromuscular blocking agents (NMBAs) are among the leading causes life-threatening drug-induced perioperative anaphylaxis. Novel studies performed by Mertes et al. [54] confirmed a remarkable relationship between application of pholcodine (34) and other quaternary ammonium derivatives and the risk for NMBA-related perioperative anaphylaxis mediated by immunoglobulin E (IgE). This association lead to the withdrawal of pholcodine-containing antitussive-syrups and cold medicines from the EU and UK markets [55].

Dihydromorphine (35, paramorphan) was synthesized by heterogeneous catalytic hydrogenation of the Δ^7,8 double bond of morphine (1a) in dilute acetic acid. Dihydromorphinone (36, dilaudid) was prepared either by catalytic rearrangement of 1 or by Oppenauer oxidation of dihydromorphine (35). Szabó and Bognár from the KLTE developed a procedure for the preparation and purification [56] of the OR antagonist N¹⁷-allyl-normorphine (37, nalorphine). The synthesis starts from morphine (1a, route [57]: 1 → diacetylmorphine → diacetyl-N¹⁷-cyano-normorphine → normorphine → 37) using (among other approaches) the von Braun method [58] for the N-demethylation step. Apomorphine (39), a dopamine receptor agonist that, can relieve motor symptoms of Parkinson’s disease, was synthesized from morphine (1a) with concentrated hydrochloric acid. Berényi et al. [59,60] later investigated in detail the mechanism of the morphine (1a) → apomorphine (39) rearrangement. For details of the desomorphine (38) synthesis [61], see section 2.7 and Figure 16.

2.6. Reduction of Thebaine

Thebaine (4) comprises a 6,8-diene and an enol ether function in ring-C, which together correspond to a 1-methoxy-1,3-cyclohexadiene sub-structural unit [16,17]. The B and C-rings together form a trans-3,4,8,9-tetraline system, while the C and D-rings constitute a trans-1,2,3,4,5,10-hexahydro-isoquinoline subunit. An easily attacked part of the molecule is the C⁵-O bond of the dihydrofuran-ring-E, which is a part of the C⁴-O-C⁵-C⁶=C⁷ allylether system. The characteristics of these sub-structures enable a large variety of reactions for thebaine (4) in manner sensitive to reaction conditions. This is particularly apparent its reduction with various reagents or during catalytic hydrogenation under different reaction conditions, which results in a diverse array of products (Figure 13).

Indeed, the targeted reduction of thebaine is a very challenging process [62,63]. Neither application of chemical reducing agents nor the heterogeneous catalytic hydrogenation result in a sole product. In the first case, reduction in basic- or acidic- media, yields products containing a phenolic hydroxyl group due to the opening of the E-ring. Treatment of thebaine (4) with sodium in hot ethanol results in dihydrothebaine-Φ (43, Δ^5,8-phenolic dihydrothebaine) as main product [64]. Bentley et al. [65,66] found that the application of sodium – liquid ammonia gave dihydrothebaine-Φ (43) in 90 % yield and a small amount of β-dihydrothebaine [67], along with approximately 10 % Δ^6,8-phenolic-dihydrothebaine (40). Razdan et al. [67] reported that treatment of thebaine (4) with an excess of potassium in liquid ammonia gave a 1:1 mixture of 40 and 43 in 95 % yield, from which 40 could be isolated in 34 % overall crystalline yield. Treatment of 43 with K/liquid ammonia in the presence of Fe(NO₃)₃ produced a 1:1 mixture of 40 and 43 in 79 % yield. When performing the reduction with LiAlH₄, in benzene – diethyl ether (or THF) [68,69], β-dihydrothebanie (40) was obtained in low (22–27 %) yield. The third isomer Δ^5,7-phenolic dihydrothebaine (45) can be prepared from codeinemethylether (44) by treatment with sodium ethylate [62]. Reduction of thebaine (4) with stannous chloride (SnCl₂) gave thebainone-A (41), or metathebainone (42) via rearrangement of the carbon scaffold. The yield of the reduction of thebaine (4) to 40 with the intact conjugated 6,8-diene system was greatly improved by the finding of Linders et al. [70], whereby reduction with zinc under alkaline conditions resulted in almost quantitative conversion to dihydrothebaine (40).

For the formation of metathebainone (46) and thebainone-A (47), Robinson [71] suggested the following reaction mechanism (Figure 14): (i) the enolether group of the hydrochloric salt of thebaine undergoes hydrolysis and an unstable addition product (X) is formed via proton attack on the oxygen of the E-ring; (ii) rearrangement: heterolysis of the C⁵-O bond, electron system shift, anionotrop rearrangement of the ethanamine side-chain from C¹³ to C¹⁴ (XI); (iii) intermediate (XI) stabilizes via proton losing/transmission to (XII); (iv) formation of metathebainone (46) via saturation of the easily accessible Δ^7,8-double bond. Experimental conditions leading to the formation of thebainone-A (47) were: (v) the unstable intermediate (X) transforms to (XIII) via heterolysis of the C⁵-O bond; (vi) formation of (XIV) through uptake of H^⊕ and 2e^y; (vii) ketonizaion of (XIV) through a 1,4-mechanism to thebainone-A (47).

As noted above, thebaine (4) is a non-analgesic neuromuscular toxin, long regarded as a nuisance side-product of poppy extraction. However, its conversion to pharmacologically useful semisynthetic opiates has since become the objective of concerted efforts. Catalytic hydrogenation of thebaine (4) results in a complex mixture of products depending on the reaction conditions (e.g., pH of the solution, type of catalyst) [72]. The main driving force of these investigations in mid-20^th century Hungary was to study the possibilities for transformation of thebaine (4) via dihydrothebaine (48) to dihydrocodeinone (49, hydrocodone), an API (active pharmaceutical ingredient) used for the treatment of moderate to serve pain and as a cough suppressant. Szabó and Bognár [73,74] performed detailed research into the catalytic reduction of thebaine (4) as a function of reaction conditions (e.g., type of catalyst, pressure, concentration, temperature, pH, rate of hydrogen uptake). They isolated from the product mixtures a large number of morphinan derivatives e.g., dihydrothebaine (48), dihydrocodeinone (49), dihydrothebainone (52), β-dihydrothebainone, etc., (see also Figure 15) [73,74]. For characterization of the obtained products, Szabó and Bognár undertook a careful analysis of their infrared spectra and photometry combined with paper chromatography. This enabled them to detect more than twenty morphinans in the product mixture. They also performed their syntheses through independent reaction routes, and reported in detail the mechanisms of product formation [73,74].

The occurence of parallel and consecutive side-reactions during the heterogenous catalytic hydrogenation of thebaine (4) is explicable by the following processes (Figure 15):

1,2-Addition of hydrogen to the Δ^7,8-bond bond of thebaine (4). This results in dihydrothebaine (48), which is transformable to dihydrocodeinone (49) through hydrolysis.

1,4-Addition of hydrogen to the conjugated system of thebaine (4) in ring-C leads to codeine methyl ether (44), which is reducible to tetrahydrothebaine (50).

1,4-Addition of hydrogen to the oxygen atom of the E-ring and to C-7. This constitutes a hydrogenolysis of the allyl ether group, which results in the opening of the ether bridge. In the first step Δ^5,8-phenolic-dihydrothebaine (43, dihydrothebaine-Φ) is formed, which can be transformed to dihydrothebainone-Δ⁵-methylenolate (51). The enol ether hydrolysis of this compound results in dihydrothebainone (52). Subsequent hydrogenation leads to dihydrothebainol-6-methyl ether (53).

1,6-Addition of hydrogen to the oxygen atom of the E-ring and to C-14. This process results in Δ^5,7-phenolic-dihydrothebaine (45). Following/further hydrogenation of the latter compound forms dihydrothebainone-Δ⁵-methyl enolate (51) and dihydrothebainol-6-methyl ether (53).

2.7. Synthesis of Desomorphine

Very recently, the chemistry, synthesis, and pharmacology of desomorphine (38) was reviewed in considerable detail [76,77,78]. One can deduce the structure of 6-desoxymorphine derivatives from C-6 hydroxy compounds by formal elimination of the hydroxy group or from 6-ketomorphinans by reduction. Due to their increased lipophilicity (and consequently greater brain penetration), these derivatives have higher analgesic potency compared to OR ligands with hydroxy and keto groups in position-6. This suggests that a substitution in position-6 is not essential for a high OR affinity. Indeed, Rapoport and Bonner reported in 1951 [79] that Δ⁷-desoxymorphine (57, 6-desoxymorphine, desomorphine-E, for binding affinity see [80]: K_i (μOR) = 2.9 nM, K_i (δOR) = 11.8 nM, K_i (κOR) = 44.5 nM, δ/μ = 4, κ/μ = 16), prepared from Δ⁷-desoxycodeine (for binding affinity see [80]: K_i (μOR) = 305 nM, K_i (δOR) = 4520 nM, K_i (κOR) = 3090 nM, δ/μ = 15, κ/μ = 10) by 3-O-demethylation with pyridine hydrochloride at 220 °C (6 min, N₂ atmosphere), had ~ 8.5-fold higher analgesic potency than morphine (1a), consistent with the much higher μ-OR affinity. Comparing the analgesic properties of various morphine derivatives (mouse hot-plate test) Eddy et al. [81,82] found the following order of relative potencies (in brackets): desomorphine (11.7) > dihydromorphinone (7.0) > dihydromorphine (1.2) > morphine (1.0) > dihydrodesoxycodeine (0.72) > dihydrocodeinone (0.66) > dihydrocodeine (0.17).

6-Desoxydihydromorphine (38, 4,5α-epoxy-17-methylmorphinan-3-ol, also known as dihydrodesoxymorphine-D, and desomorphine) differs structurally from morphine (1a) with respect to the absence of the alcoholic hydroxyl group in position C-6α and the saturation of the Δ^7,8-double bond in ring-C. Desomorphine (38) and morphine (1a) both have a T-shaped 3D structure, although in in 38 the ring-C has a chair conformation, while in 1a has a boat conformation. Desomorphine (38) is about 8-10 times more potent than morphine (1a) in the mouse hot-plate test [83] (ED₅₀ (38) = 0.14 μmol/kg, ED₅₀ (1a) = 3.3 μmol/kg).

Furthermore, desomorphine (38) has a rapid onset and shorter duration of analgesic action, and is accompanied with higher toxicity (rat LD₅₀ (38) = 27 mg/kg, LD₅₀ (1a) = 226–318 mg/kg [84]). Desomorphine (38) is suggested to be a high affinity μ-OR agonist, as evident by its potent analgesic effects [85], but its subtype selectivity is unknown. The synthesis of 38 was already described in the 1930s by Small and colleagues in the course of their search for new analgesics with a lesser addictive pharmacological profile compared to morphine (1a) [86,87]. In the 1950s, desomorphine (38) served as an API under the trade names Desomorphine^® and Permonid^® (Hoffmann-La Roche) in the USA and in Switzerland. The early syntheses of 38 were performed from morphine (1a), codeine (2a), dihydromorphine (35), or desoxymorphine-C (a Δ^6,7-didehydro compound: 17-methyl-4,5α-epoxy-3-hydroxy-6,7-didehydro-morphinan). The 6β-chloro derivatives α-chlorocodide (59b, Figure 16), α-chloromorphide, or α-chlorodihydromorphide were prepared with thionyl chloride (SOCl₂) from the corresponding starting morphinan (1a, 2a, 35) (note: α- and β- descripts are not sterical but positional: α: 6-substituted and β: 8-substituted). Thereafter the respective halogenocodides (e.g., α-chlorocodide (59b) or β-chlorocodide (65a), Figure 17) or halogenomorphides were subjected to catalytic hydrogenation [87] for reduction of the double bond in ring-C and dehalogenation. These reactions were performed in methanol or in diluted hydrochloric acid in the presence of Pd/BaSO₄. Depending upon the character of the substrate and the applied reaction and solvent conditions, these reactions gave desomorphine (38) in a yield of ~ 30-83 %.

In 1956, Bognár and Makleit [61] developed four-step procedures for the preparation of dihydro-6-desoxymorphine (38) from morphine (1a, Figure 16, route A: 1a → 35 → 54 → 58 → 38; route B: 1a → 55 → 54 → 58 → 38). They first synthesized 3-O-acetyl-dihydromorphine (54) from morphine (1a) via dihydromorphine (35) or via 3-O-acetylmorphine (55). They performed the partial acetylation by the method of Welsh [88] with acetic anhydride in the presence of aqueous NaHCO₃ solution. The Δ^7,8-double bond of 3-O-acetylmorphine (55) was saturated (H₂, Pd-C) in aqueous acidic media (HCl, pH 6-6.5). 3-O-Acetyl-dihydromorphine (54) was converted with p-toluene-sulfonyl chloride in pyridine to 3-O-acetyl-dihydromorphine tosylate (58), which was then treated with LiAlH₄ according to the method of Schmid and Karrer [89] in THF. This hydrogenolysed (hydrogen substituted) the tosyloxy group in position-6, in concert with 3-O-deacetylation, giving the desired 6-desoxydihydromorphine (38) in an overall yield of 7.7%.

In 1975, Makleit et al. [90] published a novel route for the preparation of desomorphine (38) starting from morphine (route: 1a → 3-O-acetylmorphine (55) → 3-O-acetylmorphine tosylate (56) → desoxymorphine-E (57) → 38). The authors concluded that with lithium aluminium hydride reduction of 3-O-acetylmorphine tosylate (56), the presence of the allylic system in ring-C facilitates the elimination reaction [90], thus giving higher yield compared to the analogous reaction of 3-O-acetyldihydromorphine tosylate (58) [61]. Desoxymorphine-E (57, Δ^7,8-didehydro) was converted with diazomethane to desoxycodeine-E (61, Δ^7,8-didehydro) or by catalytic reduction (H₂, PtO₂, MeOH, 1 h) quantitatively to desoxydihydromorphine-D (38).

In 2012, the research group of Chen [91] described a novel protocol for the synthesis of desomorphine (38) from codeine (2a, route: 2a → 60 → 61 → 62 → 38). Codeine mesylate (60) was treated with 2 equiv. of LiAlH₄ in THF (0 °C, 30 min and RT for 1 h) to yield desoxycodeine-E (61) in 93 % yield. The latter compound (61) was hydrogenated (H₂, PtO₂, MeOH, 4 Bar, 1 h) to dihydodesoxycodeine-D (62) in quantitative yield. 3-O-Demethylation of 62 was achieved by application of boron tribromide in dichloromethane to yield 38 in 43 % yield. The overall yield of the synthesis of 38 was 38 % starting from codeine (2a). The authors also performed the synthesis of a deuterium labelled version of desomorphine staring from dihydrodesoxycodeine-D in a three-step procedure.

Tragically, desomorphine (38) is the main active component [92] of the illicit designer drug krokodil (street name, from Russian word for crocodile: крoкoдил, also known as Russian magic, krok/crok, flesh-eating heroin, poor man’s heroin), which has spread as a cheap substitute of heroin in the former Soviet countries in the past two decades [84,91,93,94]. The illicit synthesis of desomorphine (38), utilizes codeine phosphate or codeine sulphate tablets sold at pharmacies as a cough suppressant. Operators reportedly use gasoline/petrol to extract the free base of 2a liberated from the tablets by addition a strong alkali (KOH, NaOH). According to one street method (route: codeine (2a) → iodocodide (59c) → dihydrodesoxycodeine-D (62) → 38), the extracted codeine (2a) is treated with iodine (iodine tincture), hydrochloric acid, and red phosphorus (from match heads) for dehalogenation. Heating of the mixture gives a very complex product mixture containing 38 as the main component, along with perhaps 50 different morphinan derivatives [95], with carryover of residual phosphorus, heavy metals, and highly toxic by-products. Naloxone serves as an antidote for acute krokodil toxicity, but is naturally ineffective against somatic injury arising from subcutaneous or intravenous injection of such a terrible concoction. Indeed, the street name krokodil refers to the necrotic skin changes associated with its use. Habitual krokodil users also suffer from horrendous wounds including deep tissue destruction, limb amputation auto-imputation of fingers, and osteonecrosis of facial bones and teeth.

2.8. Nucleophilic Substitution Reactions in the Morphine-Series

The nucleophilic substitution reaction of arylsulfonyl- or alkysulfonyl- esters (tosyl, mesyl, nosyl, brosyl, mesityl) leaving groups is a fundamental transformation [96,97] for the exchange of primary and secondary hydroxyl groups to other substituents in biology organic chemistry, and radiochemistry for the production of PET tracers [98]. In 1956, Stork and Clarke [99] studied the reactions of codeine tosylate (63a) with Cl^y, Br^y, I^y, and piperidine nucleophiles. The reaction of 63a with chloride anion (LiCl, acetone, reflux, 4 h) gave the kinetic product α-chlorocodide (59b, 6β-chloro-Δ^7,8-deoxycodeine, 6-chloro-6-desoxyisocodeine) in quantitative yield according to S_N2 mechanism (substitution with inversion). α-Chlorocodide (59b) is unstable and isomerizes readily to its thermodynamic isomer β-chlorocodide (65a, 8β-chloro-Δ^6,7-deoxycodeine, 8-chloro-8-desoxypseudocodeine, S_Ni’). Heating codeine tosylate (63a) in benzene for 36 h with piperidine 6-piperidocodide (64) was obtained in 87 % yield. In the reaction of codeine tosylate (63a) with bromide (LiBr, acetone, reflux, 2.5 h) or iodide (NaI, acetone, reflux, 2.5 h) anions, there was formation of the corresponding β-halocodide derivatives (pseudocodeine series, S_N2’ (S_N2 + S_Ni’)) (65b, bromide, 98 % and 65c, iodide, 43 %).

Bognár et al. [100] synthesized 6-deoxy-6-fluoroisocodeine (59a) from codeine tosylate (63a) with tetrabutylammonium fluoride in acetonitrile in 48 % yield, and determined its configuration of the C-6 carbon by NMR spectroscopy. They found that the 6β-fluoro compound (59a) would not isomerize to the corresponding 8β-fluoro-derivative (β-fluorocodide, pseudocodeine series). In 1976, Makleit et al. [101] studied the substitution reaction of codeine tosylate (63a) with Cl^y nucleophile (LiCl) in N,N-dimethylformamide. They found that reaction temperature had a significant influence on the course of the reaction; at lower temperatures (40 °C, 5 h) there was formation of the kinetic product α-chlorocodide (59b, S_N2, 76 %) while at higher temperatures (e. g., 120 °C, 10 h) the thermodynamic isomer β-chlorocodide (65a, S_N2 + S_Ni’, 33 %) was the predominant product. They also detected high temperature conversion of the 6β-chloro (59b) to the 8β-chloro (65a) isomer (120 °C, 10 h, 41 %).

Makleit [102,103] thoroughly investigated the synthesis of tosyl and mesyl derivatives in the so-called morphine series (Δ^7,8-unsaturated derivatives: morphine (1a), codeine (2a), 14-hydroxycodeine (66) and their 7,8-dihydro derivatives: dihydromorphine (35), dihydrocodeine (67), 14-hydroxydihydrocodeine (68), Figure 18). His objective was to study their nucleophilic substitution reactions, and to clarify the reaction mechanism, through production of numerous semisynthetic morphinan derivatives [102,103]. The Makleit group further showed that the C-ring of the Δ^7,8-unsaturated derivatives (1a, 2a, 66, Figure 18) has a flattened boat conformation with the C-6 hydroxyl group having an allylic alcohol character in a pseudo equatorial position. Conversely, the C-ring in the 7,8-dihydro series (35, 67, 68, Figure 18) has the chair conformation, with the C6-OH in an axial position. In both cases, the hydroxyl group connects to an asymmetric carbon in position-6. They prepared sulfonate esters of codeine (2a) and dihydrocodeine (67) by application of previously well-established methods resulting in codeine tosylate (63a, 58 %, Figure 19), codeine mesylate (63b, 16 %), dihydrocodeine tosylate (71a, 75 %), and dihydrocodeine mesylate (71b, 23 %) respectively [103,104].

2.8.1. Reactions of 7,8-dihydro Compounds

The reactions of the 7,8-dihydro-6α-sulfonate esters (71–73) with the nucleophile anions H^y, N₃^y, CH₃COS^y, and CH₃COO^y in polar aprotic solvents occur according to the S_N2 mechanism via Walden-inversion, yielding products of the iso-series. Dihydrocodeine tosylate (71a) reacted with N₃^y nucleophile resulted in 6-desoxy-6-azido-dihydroisocodeine (74a, Figure 20) [48,104,105], while 14-hydroxydihydrocodeine tosylate (73a) under identical conditions (10 equiv., NaN₃, DMF, 100 °C, 24 h) with the same reagent gave 6-desoxy-6-azido-14-hydroxydihydroisocodeine (74b, 31 %) [106]. Reduction of the 6-azido compounds (74a–b) with LiAlH₄ [104] or by catalytic hydrogenation [106] led to the corresponding 6-amino (6-beta-amino or 6-amino-dihydroisocodeine) derivatives (75a–b).

Makleit and Bognár [107] accomplished the epimerization of the C-6 chiral centres of codeine (2a) and dihydrocodeine (67). The acetolysis of codeine tosylate (63a) with 10 % acetic acid (reflux, 4 h) gave isocodeine (2b) in 16 % yield [107]. In the 1990s, the Makleit group developed protocols resulting in higher yields for the derivatives of the iso-series via application of the Mitsunobu reaction (see section 2.11.). 6-O-Acetyl-dihydroisocodeine (76) was prepared starting from dihydrocodeine mesylate (71b) with acetic acid, acetic anhydride, and sodium acetate (reflux, 31 h). Subsequently, 6-O-deacetylation of 76 was achieved by alkaline hydrolysis to yield dihydroisocodeine (77), with overall 21 % yield from dihydrocodeine mesylate (71b)). When dihydrocodeine tosylate (71a) was treated with CH₃COS^y nucleophile (KSCOCH₃, DMF) 6-acetylthio-dihydroisocodeine (78) was obtained [108,109]. This latter compound was deacetylated with sodium methylate to 6-desoxy-6-thiol-dihydroisocodeine (79). Upon reacting 3-O-acetyl-dihydromorphine tosylate (72a) with hydride anion (LiAlH₄), there was elimination of the tosyloxy group to yield 6-desoxydihydromorphine (38) [61].

2.8.2. Reactions of Δ^7,8-Unsaturated Derivatives

Sulfonate esters (6α-O-tosyl and 6α-O-mesyl) of Δ^7,8-unsaturated derivatives react with different nucleophiles either according to an S_N2 mechanism to give 6β-substituted derivatives, or via an S_N2 reaction accompanied by allylic rearrangement (an S_N2 + S_Ni’ mechanism) to result in 8-substituted-8-desoxypseudo compounds (supposed mechanism see later in section 2.9.1., Figure 39). Both the character of the nucleophile and the stereochemical properties of the substrates are of crucial importance in determining the reaction mechanism. For example, codeine tosylate (63a) with LiCl gave α-chlorocodide (59b) [99], whereas the reaction of isocodeine tosylate (81) under identical conditions with the same reagent resulted in β-chlorocodide (65a) [106].

Codeine tosylate (63a) with tetrabutylammonium fluoride in acetonitrile [100] led to the formation of 6-fluoro-6-desoxyisocodeine (59a, Figure 21). The isolated 6α-fluoro-type compound (59a, α-fluorocodide) cannot isomerize to the 8β-fluoro-type (β-fluorocodide) derivative. Compound 59a was also prepared from pseudocodeine tosylate (see structure 97 later in section 2.8.3., Figure 25) with tetrabutylammonium fluoride (5.4 equiv., CH₃CN, reflux, reflux, 29 h) in poor yield (10 %) [110]. 6-Fluoro-6-desoxy-14-hydroxyisocodeine (80) was prepared from 14-hydroxycodeine tosylate (70a) under identical conditions [106]. When codeine tosylate (63a) was treated with 10 % acetic acid (reflux, 4 h), the main direction of the reaction was towards formation of isocodeine (2b, 16 % isolated yield) [107] without allylic rearrangement. This method was also applicable for the synthesis of isomorphine (1b, 16 %) from 3-O-acetylmorphine tosylate (69a) with subsequent preparation of. dihydroisomorphine (82) through catalytic reduction of isomorphine (1b) [111].

Codeine tosylate (63a) with azide anion (NaN₃) in N,N-dimethylformamide gave 8-azido-8-desoxypseudocodeine (83, 8-azidocodide, Figure 22), which was converted to 8-amino-8-desoxypseudocodeine (84, 8-aminocodide) with lithium aluminum hydride [48,104]. Azidolysis of 14-hydroxycodeine tosylate (70a) with 1.25 equiv. of sodium azide gave 6-azido-6-desoxy-14-hydroxyisocodeine (87, 38 %) when the reaction mixture was heated at 100 °C for 4 h [106]. While the application of the same 0.25 % NaN₃ excess and longer reaction time (100 °C, 8h) led to 8-azido-8-desoxy-14-hydoxypseudocodeine (85, 46 %). 6-Azido-6-desoxy-14-hydoxyisocodeine (87) was isomerized by heating in DMF (100 °C, 6 h) 83 %) to 85 in 83 % yield [106]. The reduction of 8-azido-8-desoxy-14-hydoxypseudocodeine (85) with LiAlH₄ gave 8-amino-8-desoxy-14-hydoxypseudocodeine (86), although the catalytic hydrogenation led to 8-amino-8-desoxy-14-hydoxydihydropseudocodeine (88) [106].

The treatment of codeine tosylate (63a) with CH₃COS^y nucleophile (KSCOCH₃, DMF) [109] led to 8-acetylthio-8-desoxypseudocodeine (89, Figure 23), which was deacetylated with sodium methylate to give 8-thiol-8-desoxypseudocodeine (90). Oxidation of the latter compound gave an 8,8’-didesoxy-8,8’-pseudocodeinyl disulphide type compound. Surprisingly, the reaction of codeine tosylate (63a) with thiocyanate anion (^yS=C=N, KSCN, in acetone) gave (in a kinetically second order reaction) the product 8-isothiocyanato-8-desoxypseudocodeine (92, -NCS) instead of the expected 8-thiocyanato-8-desoxyspeudocodeine (91, -SCN) [109].

Analogous reaction of 3-O-acetylmorphine tosylate (69a, Figure 24) with CH₃COS^y led to 3-O-acetyl-8-acetylthio-8-desoxypseudomorphine (93), which was deacetylated with sodium methylate to 8-thiol-8-desoxypseudomorphine (94) [109]. he treatment of 69a with potassium thiocyanate resulted in 3-O-acetyl-8-isothiocyanato-8-desoxypseudomorphine (95) [112]. Reaction of 69a with N₃^y anion gave 8-azido-8-desoxypseudomorphine (96, Makleit-Bognár nomenclature, 8-azido-8-desoxy-γ-isomorphine).

2.8.3. Reactions of pseudocodeine tosylate.

In the early 1970s, Makleit et al. [104] extended their studies on the nucleophilic substitution reaction of sulfonate esters to the Δ^6,7-double bond containing pseudocodeine tosylate (97, Figure 25). Pseudocodeine (2d) was prepared from α-chlorocodide (59b) by hydrolysis according to the method of Small and Lutz [113]. The tosyl ester (97) was prepared from 2d in 76 % yield. They investigated the substitution reactions by application of a large variety of nucleophilic reagents. Using the reagents type-X (F^y, Cl^y or piperidine) isocodeine derivatives (99) were formed either by an S_N1’ or S_N2’ mechanism in a yield of range 10–65 %. Using type-Y nucleophiles (Br^y, I^y, NCS^y, N₃^y, CH₃COS^y) pseudocodeine derivatives (98) were isolated in poor yields (10-52 %). In these latter cases the mechanism of the reaction was either S_N1 accompanied by retention or a combination of S_N2’ and S_Ni’.

The corresponding products were isolated in poorer yields despite more vigorous reaction conditions, and longer reaction times in comparison to the analogous reactions of codeine tosylate (63a). The authors undertook comparative solvolysis of pseudocodeine tosylate (97) and codeine tosylate (63a) in an acetic acid - potassium acetate system, finding that the solvolysis of 8β-O-tosylate (97) solvolysis occurred slower than that of the 6α-O-Tos (63a) following in both cases a first-order reaction. They investigated more thoroughly the reaction of pseudocodeine tosylate (97) with piperidine, definitively showing formation of 6β-(1-piperidinyl) derivative (64, 6-piperidocodide) by an S_N1’ reaction mechanism.

2.8.4. Neopine Derivatives

The neopine derivatives (e.g., isoneopine (3b), Figure 26.) are critical starting materials for the synthesis of B/C-trans-morphinans, which can be very difficult to produce [114,115]. In the 1970s, Berényi, Makleit, and Bognár worked out numerous syntheses for the preparation of neopine (3a) [116] and isoneopine (3b) [117]. Furthermore, Berényi, Makleit, and Dobány developed an improved method for the isolation of natural neopine (3a) from poppy straw [118], without interfering with the original Kabay-process. In 1977, Makleit et al. [116] reported an efficient method for the preparation of neopine (3a) from thebaine (4) via 14-chlorocodeine (structure see later: 124a in Figure 37). The reduction of the latter compound (124a) was achieved with sodium bis(2-methoxyethoxy)aluminium hydride. The overall yield for neopine (3a) from thebaine (4) was 72 %. In 1980, Berényi and Makleit [117] published a novel method for the preparation of isoneopine (3b), wherein azidolysis of neopine mesylate (100) resulted in 6-deoxy-6-azido-isoneopine (101) in 75 % yield. The latter compound was reduced with Zn/NaI/DMF to 6-deoxy-6-amino-isoneopine (102, 65 %). Treatment of the amino compound (102) with NaNO₂ in acetic acid gave 6-O-acetylisoneopine (103, 90 %) and finally the alkaline hydrolysis of the 6-O-acetyl group resulted quantitively in isoneopine (3b). The overall yield from neopine (3a) was 44 %.

Taking advantage of the availability of neopine (3a) in large quantities, Berényi et al. [119] prepared 6-O-mesylneopine (100, Figure 27.) for the investigation of its substitution reactions with nucleophile partners (F^y, Cl^y, Br^y, I^y, N₃^y). When 6-O-mesylneopine (100) was reacted with tetrabutylammonium fluoride in acetonitrile (reflux, 2 h), 6-demethoxythebaine (105, 77 %) was obtained. In the reaction of neopine mesylate (100) with iodide anions (10 equiv. NaI, DMF, 120 °C, 30 h), 6-demethoxythebaine (105, 21 %) was isolated, with 36 % recovery of the starting material (100).

Applying chloride anions (10 equiv. LiCl, DMF, 120 °C, 6 h), 6-deoxy-6-chloro-isoneopine (104a) and 6-demethoxythebaine (105) were formed in a ratio of 6:4 [119]. Using bromide anions as nucleophile (10 equiv. LiBr, DMF, 120 °C, 11h), 6-deoxy-6-bromo-isoneopine (104b, 32 %) and 6-demethoxythebaine (105, 31 %) were isolated, with 10 % recovery of starting material (100). The 6-β position of the halogen substituent in compounds 104a and 104b (Cl, Br) was substantiated by NMR spectroscopy, in which value of the geminal coupling of the H-5 and H-6 protons (³J_5,6 = 9.5–10 Hz) was characteristic for both compounds. The azidolysis of neopine mesylate (100), with 10 equivalents of NaN₃ in DMF (120 °C, 24 h) resulted in 6-deoxy-6-azido-isoneopine (101, 55 %). The latter compound (101) was converted to 6-deoxy-6-amino-isoneopine (102) by heterogenous catalytic reduction (H₂, 50 % acetic acid, platinum oxide). The authors concluded that the first step of these reactions is an S_N2 nucleophile substitution resulting in the corresponding 6-desoxy-6-haloisoneopines (104a–b) followed by a second elimination step resulting in 6-demethoxythebaine (105). 6-Desoxy-6-bromo-isoneopine (104b) was converted to 6-demethoxythebaine (105, reflux, 4 h, 72 %) with tetrabutylammonium fluoride, while 6-deoxy-6-chloro-isoneopine (104a) gave according to 50% conversion a 1:1 mixture of 6-demethoxythebaine (105) and the starting material (104a). Based on these observations, Hutchins et al., and Beyerman, Maat, and Crabbendam reported improved preparations of 6-demethoxythebaine (105) in high yiels either from neopine (3a) or from codeine (2a) via isocodeine (2b) or β-bromocodide (65b) [120]. The reaction of thebaine (4) with peracids, peracetic acid (CH₃CO-OOH) or performic acid (HCO-OOH) [121] resulted in 14-hydroxycodeinone (structure see later 132 in Figure 40) by 1,4-addition of hydroxyl groups to the 6,8-conjugated system followed by methanol-elimination from the 6-hemiacetal intermediate.

The analogous treatment of 6-demethoxythebaine (105, Figure 27) with formic acid and hydrogen peroxide [119] gave 14-hydroxy-allopseudocodeine (106, 71 %). This latter compound was also prepared previously by Currie et al. [122] from 6-O-tosyl-14-hydroxycodeine (70a, Figure 28) in an S_N2’ reaction (70 % AcOH) through 8-O-acetyl-14-hydroxy-allopseudocodeine (107) and subsequent alkaline hydrolysis. Seki et al. [123] prepared 106 from 14-hydroxy-α-chlorocodide (108) via acetolysis with 10 % AcOH and subsequent hydrolysis.

2.8.5. Alkyl Mesylate and Allyl Halide Structural Units in the Same Molecule

Studying the relative reactivity of alkyl mesylates (C⁸-C⁷- C⁶-OMs) and allyl halides (C¹⁴=C⁸-C⁷-X) Simon et al. [124] investigated the nucleophilic substitution reactions of 6-O-mesyl-7α-haloneopine derivatives (112a–b, Figure 30) containing both structural units in the same molecule. The starting materials 7α-chloroneopine (111a) and 7α-bromoneopine (111b) were prepared in three steps from thebaine (4) by the method of Abe et al. [125] (Figure 29). In brief, thebaine (4) was reacted with N-bromosuccinimide to give 14-bromocodeinone (109) [126]. This latter compound was stereoselectively reduced with NaBH₄ to 14-bromocodeine (110). The treatment of 110 with concentrated hydrochloric acid or 48% hydrobromic acid resulted in the corresponding 7α-haloneopines (111a–b).

Nucleophilic substitution reactions of 6-O-mesyl-7α-chloro-neopine (112a) and 6-O-mesyl-7α-bromoneopine (112b) were studied with the following nucleophiles: Cl^y, Br^y, I^y, N₃^y [124] (Figure 30). When 6-O-mesyl-7α-chloroneopine (112a, Figure 30) was reacted with azido anions (5 equiv. NaN₃), there was formation of a complex product mixture. The column chromatography isolated the main product 7-chloro-6-demethoxythebaine (113a, 31 %) and the side products 6-O-mesyl-7β-azidoneopine (114, 10 %) and 6-O-mesyl-6-demethoxythebaine (115, 5 %).

Reacting 6-O-mesyl-7α-bromoneopine (112b, Figure 30) with 5 equiv. of sodium azide in N,N-dimethylformamide (addition of water was necessary to dissolve NaN₃), the main product was 6-O-mesyl-7β-azidoneopine (114, 48 %). The isolated side products were 6-O-mesyl-6-demethoxythebaine (115, 14%) and 9α-azido-6-O-mesyl-indolinocodeine (116, 3 %).

Studying the reactions of 6-O-mesyl-7α-chloroneopine (112a, Figure 31) with 5 equivalents of reagent (LiCl, LiBr, KI) [124], the main product with chloride anions (5 equiv. LiCl, DMF, 100 °C, 24 h) was 7-chloro-6-demethoxythebaine (113a, 39 %), with isolation of 6-chloro-6-demethoxythebaine (118a, 18 %) and 6-O-mesyl-7β-chloroneopine (117a, 3 %) side products. Using LiBr as the source of nucleophile under otherwise identical conditions, the main reaction was methanesulfonic acid elimination, which resulted in 7-chloro-6-demethoxythebaine (113a, 31 %) and 6-bromo-6-demethoxythebaine (118b, 18 %), as well as 6-O-mesyl-7β-bromoneopine (117b, 13 %). In the presence of iodine anions (5 equiv. KI, DMF, 100 °C, 48 h), the main product was 6-demethoxythebaine (105, 34 %).

Shorter reaction times (20–40 min) were applied in studying the reactions of 7α-bromoneopine mesylate (112b, Figure 32) with Cl^y, Br^y, and I^y nucleophiles. With LiCl, the main product was 6-O-mesyl-7β-chloroneopine (117a, 19 %), with additional isolation of 6-O-mesyl-6-demethoxythebaine (115, 7 %). When 112b was reacted with LiBr, the products were 6-O-mesyl-7β-bromoneopine (117b, 24%), 7-bromo-6-demethoxythebaine (113b, 3 %), and 6-O-mesyl-6-demethoxythebaine (115, 6 %). Applying potassium iodide, 6-demethoxythebaine (105, 72 %) was the single isolated product. The study [124] also presented an elegant and efficient method for the preparation of 7-halo-6-demethoxythebaines (113a–b).

When boiling 6-O-mesyl-7α-chloroneopine (112a, Figure 33.) or 6-O-mesyl-7α-bromoneopine (112b) in the presence of potassium tert-butylate in methanol, the main reaction was methanesulfonic acid elimination (E_1,2 (-CH₃SO₂OH)). 7-Chloro-6-demethoxythebaine (113a) was prepared as the sole product in 95 % yield from 112a, and 7-bromo-6-demethoxythebaine (113b) was the main product in 50 % from 112b.

2.8.6. Substrates Containing Double Allylic System

Substrates with Allyl Halide and Allyl Tosylate Sub-Units

Berényi et al. [127] was the first to study the nucleophilic substitution reactions of 6-O-tosyl-14-halocodeine substrates (119a–b), which possess a double allylic system: allyl halide (C⁷=C⁸-X¹⁴) and allyl tosylate (C⁸=C⁷-C⁶-OTos) in the same molecule. They investigated reactions of 14-chlorocodeine tosylate (119a, Figure 34) [127] and 14-bromocodeine tosylate (119b) [126] with Cl^y, Br^y, N₃^y nucleophiles. Surprisingly, 6-chloro-6-demethoxythebaine (118a, 68 %) as well as 6-bromo-6-demethoxythebaine (180b, 74 %) was isolated when the more stable 14-chlorocodeine tosylate (119a) was reacted with Cl^y, Br^y anions (LiX, X = Cl. Br, DMF, 100 °C, 24 h).

Interestingly, 6-O-tosyl-14-bromocodeine (119b) gave 6-O-tosyl-6-demethoxythebaine (120) under similar conditions. Heating 6-O-tosyl-14-bromocodeine (119b) in DMF (100 °C, 2 h) without any nucleophilic reagent gave the latter compound (120, 26 %). A further surprising result was that 6-O-tosyl-6-demethoxythebaine (120) did not react with Cl^y or Br^y anions.

The authors [127] explained the complicated formation of 6-halo-6-demethoxythebaine derivatives (118a–b) by the following elementary steps (Figure 35). (i): S_N2’ reaction: nucleophilic attack at position-7, resulting in 6-O-tosyl-7-substituted-neopine (A); (ii): S_N2 reaction: further nucleophile attack on the 6-tosyloxy group, resulting in 6-deoxy-6β,7β-disubstituted neopine (B). (iii): 1,4-elimination reaction.

The azidolysis [127] of 14-chlorocodeine tosylate (119a, Figure 36) with 2.5 equiv. NaN₃ resulted in 6-deoxy-6β,7α-diazido-neopine (121, 27 %) as main product and 6-deoxy-6β,7β-diazido-neopine (122, 7 %) as a minor by-product. Carrying out the azidolysis of 119a with 2.5 equiv. trimethylsilyl azide in DMF at 100 °C, yielded 6-O-tosyl-6-demethoxythebaine (120) and 6-deoxy-6β,7β-diazido-neopine (122) in a ratio of approximately 1:1, with a significant amount of starting material (119a) remaining in the product mixture after 24 h reaction time. When 6-O-tosyl-14-bromocodeine (119b) was reacted with azido anions (2.5 equiv. NaN₃), the major product was 6-O-tosyl-7β-azidoneopine (123, 25 %), with lesser amount of 6-O-tosyl-6-demethoxythebaine (120, 16 %) in 16 % yield [127]. The reactivity sequence of the leaving groups was as follows: allyl bromide > allyl tosylate > allyl chloride.

Substrates Containing Double Allyl Halide Sub-Structural Units

In 1988, Berényi et al. [128] reported the substitution reactions of 6β,14-dichlorodesoxycodeine (125a) and 6β-chloro-14-bromodesoxycodeine (125b, Figure 37), containing double allyl halide (C⁷=C⁸-C¹⁴-X and C⁸=C⁷-C⁶-Cl) sub-structural units with the nucleophiles Cl^y, Br^y, and N₃^y applied as LiCl and LiBr and NaN₃ respectively. They also studied reactions in hot DMF in absence of a nucleophilic partner. The authors claimed that only leaving groups (Cl^y, Br^y) linked to the tertiary C¹⁴ carbon (C⁷=C⁸-C¹⁴-X) took part in nucleophilic substitution reactions among the investigated 6β-chloro-14-halodesoxycodeines (125a–b).

The starting materials 125a and 125b for these investigations were prepared from 14-chlorocodeine (124a) [126] and 14-bromocodeine (124b, thebaine (4) → 14-bromocodeinone (109) → 1b), respectively [129]. The reaction of 124a with 2 equiv. phosphorus pentachloride in chloroform gave 125a in 66 % yield, and the analogous reaction of 14-bromocodeine (124b) gave 125b in 53 % yield. They also prepared 125a from 14-chlorocodeine tosylate (119a) with LiCl.

6β-Chloro-14-bromodeoxycodeine (125b, C⁷=C⁸-C¹⁴-Br and C⁸=C⁷-C⁶-Cl) reacted readily with Cl^y anion (10 equiv. LiCl) in DMF, giving 6β-chloro-7β-bromodesoxyneopine (126a) as main product in 42 % yield, with further isolation of 6α-chloro-7β-bromodesoxyneopine (126b) in 4 % yield by column chromatographic separation of the mother liquor. Application of Br^y nucleophile (5 equiv. LiBr, DMF) for 125b, gave 6-β-chloro-7β-bromodesoxyneopine (126c) in 16 % yield. Using N₃^y anions (5 equiv. NaN₃, DMF, 100 °C, 1 h) gave 6β-chloro-7α-azidodesoxyneopine (126d, 44 %) and 6β-chloro-9α-azidodesoxyindolinocodeine (127, 8 %). After heating of 125b in DMF at 100 °C for 30 min in the absence of Nu^y unreacted starting material (125b), 6β-chloro-7β-bromodesoxyneopine (126c), and 6-chloro-6-demethoxythebaine (118a) were separated from the product mixture by column chromatography. Extending the reaction time (1 h) gave 6-chloro-6-demethoxythebaine (118a) in 62 % yield.

The reaction of 6β,14-dichlorodesoxycodeine (125a, C⁷=C⁸-C¹⁴-Cl and C⁸=C⁷-C⁶-Cl) with Br^y anions (5 equiv. LiBr, DMF, 100 °C, 8 h) led to 6-chloro-6-demethoxythebaine (118a) in 72 % yield. Azidolysis of 125a under identical conditions as described above for 125b resulted similarly in 6β-chloro-7β-bromodesoxyneopine (126c, 48 %) and 6-chloro-6-demethoxythebaine (118a, 7 %). After heating of 125a in DMF at 110 °C without Nu^y reagent for 30 min, 6-chloro-6-demethoxythebaine (118a, 60 %), 6β-chloro-7β-chlorodesoxyneopine (126a, 4 %) and 6β-chloro-7α-chlorodesoxyneopine (126b, 1 %) were isolated. Upon continued the heating for 24 h, 6-chloro-6-demethoxythebaine (118a) was obtained in 72 % yield. The differences in reactivity of the allyl chloride (C⁷=C⁸-C¹⁴-Cl and C⁸=C⁷-C⁶-Cl) and allyl bromide (C⁷=C⁸-C¹⁴-Br) subunits of the molecules manifested in the differing times required for whole conversion via an elimination reaction (E_1,2 type) in the absence of a Nu^y reagents. Thus, conversion of 6β-chloro-14-bromodesoxycodeine (125b) to 6-chloro-6-demethoxythebaine (118a) took one hour, whereas conversion of 6β,14-dichlorodesoxycodeine (125a) required 24 h.

2.9. Azidomorphinans

Organic azides [130,131,132] are suitable building blocks for the synthesis of a broad range of nitrogen-containing scaffolds. Due to the high electrophilicity of the azido group (mesomeric structures: -N=N-N: ↔ -N=N^⊕=N^y ↔ -N^y-N^⊕≡N ↔ -N^y-N=N^⊕), organic azides are prone to react with various nucleophiles [133,134], and furthermore serve as starting materials for a large variety of transformations (e.g., reactions with electrophiles, cycloaddition, 1,3-cycloaddition with alkynes = click reaction, reduction to amines, etc.). Incorporation of the azido pharmacophore into a bioactive scaffold can be advantageous in providing new molecules with desirable pharmacological profiles, including increased potency, and other biochemical and pharmacological characteristics [135].

Cocker et al. [136] and later Willner et al. [137] thoroughly investigated the displacement by N₃^y and other nucleophile of N-acetylated 7-aminocephalosporanic acid (7-ACA) derivatives. The prepared 3-azidomethyl-3-cephem analogues were effective in inhibiting the growth of gram-negative bacteria [136,137].

Azidomazenil (8-azido-5,6-dihydro-5-methyl-6-oxo-4H-imidazo-1,4-benzodiazepine-3-carboxylic acid ethyl ester, Ro-15,4513, Hoffmann-La-Roche), the azido analogue of the 1,4-benzodiazepine derivative flumazenil (Ro-15,1788, Hoffmann-La-Roche), a partial inverse GABA_A receptor (GABA_AR) agonist, blocks the electrophysiological and behavioural effects of low to moderate ethanol doses [138]. Indeed, GABA_ARs may mediate the rewarding effects of ethanol, which require binding at the benzodiazepine-binding site (cBZR) of the α5-subunit of the GABA_AR [139]. In 1992, the ¹¹C-labelled version of Ro-15,4513 was developed by Halldin et al. [140] for PET imaging of benzodiazepine receptors in living human brain [139].

The azido analogue of levorphan and its tritiated version N¹⁷-(4-azidophenyl-CH₂C(³H)₂)-norlevorphan were synthesized by Winter et al. [141] in 1972. N¹⁷-Desmethyl-levorphan was reacted with β-(4-azidophenyl)ethyl bromide in methanol (K₂CO₃, 42 °C, 90 h) to yield the target compound, which was 4-fold more effective (ED₅₀ = 0.78 mg/kg) than levorphan as an analgesic in the mouse hot-plate test, perhaps due to its greater lipophilicity and brain penetration.

2.9.1. Azidomorphine Analogues

6-Azido substituted morphine derivatives have a conspicuous analgesic effect [142,143], having a 50–300 fold increase in analgesic potency relative to morphine (1a) for the case of a 6-azido substitution of dihydroisomorphine (82, Figure 21) [144]. Spanning several decades, the research group of Makleit paid particular attention to investigating the nucleophilic substitution reaction of the sulfonate esters of the morphine series (see Figure 18 and Figure 19). They studied extensively the azidolysis of 6-O-tosyl and 6-O-mesyl esters of different morphinans and the subsequent transformations to the corresponding amino derivatives. In 1968, Bognár and Makleit investigated the syntheses of aminocodide and aminomorphide derivatives (Figure 38) [48], at a time when only a few in ring-C amino substituted morphinan derivatives were known. Furthermore, the absolute stereochemistry of these compounds, 6-aminocodide, 8-aminocodide (84) and 8-aminodihydrocodide (75a) where uncertain at that times. As starting points in the Δ^7,8 unsaturated series, they used the sulfonate esters of codeine (2a) and 3-O-acetylmorphine (55). These materials readily reacted with azide anion (N₃^y), giving the 8β-azides azidocodide (83) as well 8-azidomorphide (128) i.e., products belonging to the to the pseudocodeine or γ-isomorphine series (pseudomorphine derivatives, according to the Makleit-Bognár nomenclature).

The nucleophile substitution of codeine tosylate (63a) with azide anion (N₃^y) can occur according to an S_N2’ reaction mechanism (Figure 39), whereby the azide anion attacks the Δ^7,8-double bond at C-8 position to give the 8β-azide (XV, Nu = N₃, 83) directly. Another possible mechanism is an S_N2 reaction followed by allylic rearrangement ([3,3]-suprafacial sigmatropic rearrangement, S_Ni’ mechanism), where the nucleophile (N₃^y) attacks the C-6 position. The hypothesized 6β-azide (XVI, Nu = N₃^y, allylic azide) intermediate undergoes an allylic rearrangement, which shifts the new substituent from C-6 into the C-8 position to give 8-azidocodide (XV, Nu = N₃^y, 83).

The analogous reaction of 6-O-sulfonate esters belonging to the 7,8-dihydro series (e.g., 71a, 72a) [104] with the N₃^y nucleophile gave the corresponding 6β-azides: 6-desoxy-6-azidodihydroisocodeine (74a, “azidocodeine”) [105] and 6-desoxy-6-azidodihydroisomorpine (130, “azidomorphine”) [105], with inversion of the configuration in a stereospecific nucleophile substitution reaction (S_N2). The azides (83,128 and 74a,130) were reduced with LiAlH₄ in THF to their corresponding 8β-amino (84,129) or 6β-amino (75a,131) derivatives.

In a rat hotplate test of analgesia, azidomorphine (130, 6-AM, 6-desoxy-6-azidodihydroisomorphine; 3-hydroxy-4,5-α-epoxy-6β-azido-7,8-dihydro-17-methyl- morphinan) [105] was 270-300 times more potent than morphine (ED₅₀ (130) = 0.016 mg/kg, versus ED₅₀ (1a) = 4.7 mg/kg), whereas azidocodeine (74a, 6-desoxy-6-azidodihydroisocodeine, ED₅₀ (74a) = 0.36 mg/kg;) was only ten times more potent than morphine (1a) [145]. Nonetheless, mice showed lesser propensity for developing physical dependence for azidomorphine (130) compared to morphine (1a). Interestingly, there was a consistent dissociation between the analgesic activity and physical dependence liability in monkeys, rats, and mice [145]. Intensive investigations by Bulaev et al. of the pharmacological characteristics in vivo of azidomorphine (130) in different animal models (rats, rabbits and cats) consistently showed 20–100 times greater analgesic activity compared to morphine (1a) [146,147]. Knoll et al. [148] demonstrated that azidomorphine (130) is metabolized in rats by N-demethylation to norazidomorphine and by conjugation with glucuronic acid. In human studies, azidomorphine (130) was 40–300 times more potent than morphine (1a) [149], relieving severe pain in 0.5–1 mg doses (sc. or. iv.), but with less constipation, vomiting, and euphoria than with equianalgesic doses of morphine (1a). When giving a composition containing azidomorphine (130, 0.5 mg), the non-narcotic analgesic rimazolium mesylate (1,6-dimethyl-3-carbethoxy-4-oxo-6,7,8,9-tetrahydro-homopyrimidazol, probon, MZ-144; 150 mg) potentiated the analgesic effects, while antagonizing the respiratory depressant effects of 130 [144]. In a human subject, azidomorphine (130) was metabolized by conjugation (6-AM-3G) and by reduction to 6-desoxy-6-aminoisomorphine [150].

In 1974, Sasvári et al. [151] reported the X-ray structural analysis of azidomorphine (130). When exploring the crystal structure of 14-hydroxyazidomorphine (138) Kálmán et al. [152], found that the substitution of the H-14 of azidomorphine (130) with a hydroxy group did not alter the C- and D-ring chair conformation, but did alter the orientation of the 6β-azido group. Tamás et al. studied mass spectroscopic fragmentation pathway of azidomorphine (130) and its derivatives [153]. Under electron impact (70 eV, source temperature 100–150 °C) azidomorphine analogues gave a remarkably different fragmentation pattern compared to that of the parent morphine alkaloids; the main decomposition route involved the loss of a nitrogen (N₂) molecule in the 6-desoxy-6-azidoisomorphine series (74a, 130, 138, 144). Among 8-azido-substituted compounds containing a Δ^6,7-double bond in ring C. (e.g., azidomorphide (128, 8-desoxy-8-azido-γ-isomorphine) or azidocodide (83)), there was loss of the loss of the N₃ radical (^.N=N^⊕=N^y) due to the allylic effect. Dinya et al. [154,155] performed infrared spectroscopic investigations and quantum chemical analysis of azidomorphinans.

Aiming to improve upon the toxicity of azidomorphine (130, LD₅₀ (130) = 8.1 mg/kg; LD₅₀ (1a) = 320 mg/kg, i.v. in the rat) [145], Makleit et al. developed 14-hydroxyazidomorphine” (138, 6-desoxy-6-azido-14-hydroxydihydroisomorphine, 3,14β-dihydroxy-4,5α-epoxy-6β-azido-7,8-dihydro-17-methylmorphinan, Figure 40) in 1972 [106,156,157]. The synthesis of 138 from thebaine (4) followed seven consecutive steps. In brief, 14-hydroxycodeinone (132) was obtained from 4 with hydrogen peroxide and formic acid [123], and the Δ^7,8 double bond of 132 was then saturated under heterogenous catalytic conditions (H₂, 10% Pd-C, AcOH) to give oxycodone (133). Oxymorphone (134) was prepared from 133 by 3-O-demethylation with 48% hydrobromic acid. The C-6 keto group of 134 was reduced by sodium borohydride in a stereospecific reaction to 14-hydroxy-dihydromorphine (135). The phenolic 3-hydroxyl group of 135 was protected by partial acetylation using the method of Welsh [88] with acetic anhydride (NaHCO₃, H₂O). The 6α-hydroxyl group of compound 136 was transformed to sulfonate esters (137a–b) with the corresponding reagent (MsCl or TosCl, pyridine, 24 h, RT). Finally, the tosylester (137a) was subjected to azidolysis (sodium azide, DMF, H₂O, 100 °C, 24 h) to give the desired azido derivative (138) [106,156]. 6-Desoxy-6-azido-14-hydroxydihydroisomorphine (138) had similar analgesic potency in the mouse hot plate test as azidomorphine (130): ED₅₀ (130) = 0.024 mg/kg, ED₅₀ (138) = 0.029 mg/kg) [149], although with lower toxicity. The introduction of the hydroxy group in position-14 of azidomorphine (130) conspicuously reduced the toxicity; azidomorphine (130) was 6.5 times more toxic in rats and 11.6 times more toxic in mice than 14-hydroxyazidomorphine (138) [149]. Human studies indicated 40-fold greater analgesic activity for azidomorphine (130) and 14-hydroxyazidomorphine (138) compared to morphine (1a) [158].

Antitussive activity of 138 was determined and compared with the reference-antitussive codeine (2a) using the Gosswald citric acid aerosol-induced cough test in Wistar rats [157]. 14-Hydroxyazidomorphine (138) had 1000-fold higher anti-tussive activity compared to codeine (2a, AtD₅₀ (2a, s.c.) = 19.0 mg/kg, AtD₅₀ (138, s.c.) = 0.021 mg/kg, ratio 2a/138 = 904) [157].

In a subsequent study, the same research group [159] performed structure optimization-investigations aiming to identify azidomorphine analogues with a more advantageous pharmacological profile. For synthesis of 3-desoxyazidomorphine (141, Figure 41), they first produced the starting material 3-desoxydihydromorphine (139) by removal of the 3-hydroxyl group of morphine via hydrogenation of morphine-3-(1-phenyltetrazolyl)- or morphine-3-(pyrimidyl)-ether [160]. Under these conditions, the Δ^7,8-double bond was also saturated. In the next steps, the tosylester (140) was prepared and subjected to azidolysis (sodium azide, DMF, H₂O, 100 °C, 24 h) and transformed to 3-desoxyazidomorphine (141) in 20 % yield.

Analogously, “azidoethylmorphine“ (144, “3-O-ethylazidomorphine”, azidodionine, 6-azido-6-desoxy-ethyldihydroisomorphine) [159] was prepared from ethylmorphine (33) in three steps (Figure 41). Compound 33 was reduced to 3-O-ethyl-dihydromorphine (142) [161], which was converted to the esters ethyldihydromorphine tosylate (143a) or mesylate (143b) [161]. Subsequent azidolysis of the esters (143a–b) gave azidoethylmorphine (144). Antitussive activity by the Gosswald method in rats was 60-fold higher for azidodionine (144) than for codeine (2a) (AtD₅₀ (2a, oral) = 100.0 mg/kg, AtD₅₀ (144, oral) = 1.67 mg/kg, ratio 2a/144 = 59.88) [159,162,163].

3-O-Morpholinylethylazidomorphine (145, azidopholcodine, Figure 41) [159,162] has also been prepared from morphine (1a) in five steps (morphine (1a) → ethylmorphine (33) → ethyldihydromorphine → 3-O-morpholinylethyldihydromorphine → 3-O-morpholinylethyldihydromorphine mesylate → 145), and directly from azidomorphine (130) by alkylation with morpholinylethylchloride thanks to the stability of the azido group of 130 against the strong base NaOEt. Azidopholcodine (145) had four-fold greater antitussive activity in rats AtD₅₀ (145, s.c.) = 4.5 mg/kg) as compared to codeine (2a) (AtD₅₀ (2a, s.c.) = 19.0 mg/kg, ratio 2a/145 = 4.22) [163].

To study the antagonistic properties in the azidomorphine series, Makleit et al. synthesized the N¹⁷-substituted analogues (149a–b, cyclopropylmethyl, allyl, Figure 42) of azidomorphine (130), azidocodeine (74a), 14-hydroxyazidomorphine (138) and 14-hydroxyazidocodeine [164]. N¹⁷-Cyclopropylmethyl-norazidomorphine (149a, CAM) and N¹⁷-allyl-norazidomorphine (149b, AAM) were synthesized via two different routes. In the first approach, acetylation of the phenolic hydroxy group in position-3 of azidomorphine (130) resulted in 3-O-acetylazidomorphine (146). The von Braun reaction with cyanogen bromide in the presence of the 6-azido group gave high yields (e.g., 146 → 147 in 90 % yield). Alkaline hydrolysis of the cyanamide (147) and subsequent acid hydrolysis gave azidonormorphine (148) in 47 % yield. N¹⁷-alkylation of 148 with allyl bromide or cyclopropylmethyl bromide resulted in the corresponding N¹⁷-substituted-norazidomorphines (149a–b). On the other hand, the partial acetylation of N¹⁷-cyclopropylmethyl-nordihydromorphine (150a) or N¹⁷-allyl-nordihydromorphine (150b) resulted in the corresponding 3-O-acetyl-nordihydromorphines (151a–b). The latter compounds were converted into their sulfonate esters (tosyl, mesyl, 152a–d) for subsequent azidolysis, resulting in the desired analogues (149a–b).

In 1977, Knoll et al. [165] reported on their pharmacological investigations of numerous N¹⁷-substituted-azidomorphine analogues. They compared the OR opioid receptor antagonistic properties of N¹⁷-cyclopropylmethyl-norazidomorphine (149a, CAM, Figure 42), N¹⁷-allyl-norazidomorphine (149b, NAM) and their 14-hydroxy analogues: N¹⁷-cyclopropylmethyl-14-hydroxynorazidomorphine (COAM) and N¹⁷-allyl-14-hydroxynorazidomorphine (NOAM) relative to reference antagonist (naltrexone, see structure later 199, Figure 52) in different tissues. Surprisingly, N¹⁷-substituted-azidomorphine analogues were potent antagonists in the isolated cat nictitating membrane (CNM) assay, but were extremely potent pure agonists in the guinea-pig ileum (GPI), and mouse vas deferens (MVD) tests. CAM (149a) [164] stimulated the A-type receptors, which are involved in behavioural effects of OR ligands (GPI and MVD tests; cholinergic neurotransmission). On the other hand, CAM (149a) had greater antagonist potency than naloxone in the CNM test (B-type receptor: responsible for the analgesic, antitussive, and respiratory depressant effects of OR ligands; catechol adrenergic neurotransmission) [165]. Subsequently, Fürst et al. tested the antagonistic potencies of 149a–b in the oxymorphone righting test [166]. The N¹⁷-substituted azidomorphines (149a–b) were more potent antagonists than nalorphine or pentazocine, but were equipotent with naloxone. N¹⁷-Cyclopropylmethyl-norazidomorphine (149a, CAM) was suggested to be a κOR subtype agonist based on its agonistic activity in rabbit vas deferens (RVD) assay, and given the low potency of naloxone to reverse its agonistic effect in GPI and in MVS tests [166,167]. CAM (149a) has a different pharmacological profile from bremazocine or ethylketazocine (EK), which are agonists in the CNM assay.

Inspired by the novel pharmacological profile of the N¹⁷-substituted-norazidomorphines (149a–b) [165], the Alkaloida research group prepared N¹⁷-substituted-norazidoethylmorphine derivatives (155a–c, R = CPM, allyl, n-propyl, Figure 43) [168] through two independent routes from azidoethylmorphine (144) as well from ethyldihydromorphine (142).

In brief, in the first reaction sequence [168], azidoethylmorphine (144, Figure 43) was N-demethylated using the von Braun method (BrCN) via N-cyano-azidoethylmorphine (153). The cyanamide (153) was hydrolysed with 6 % hydrochloric acid; N-demethylation of azidoethylmorphine (144) was also achieved easily and in good yields with methyl- or phenyl chloroformate. The formed N¹⁷-phenyl- or N¹⁷-methyl-oxycarbonyl-N¹⁷-desmethyl-azidoethylmorphine was converted to norazido-ethylmorphine (154) with 25% KOH in ethanol. Subsequent alkylation of N¹⁷-desmethyl-azidoethylmorphine (154) with cyclopropylmethyl bromide or allyl bromide resulted in N¹⁷-cyclopropylmethyl-norazidoethylmorphine (155a, ECAM, Figure 43) and N¹⁷-allyl-norazidoethylmorphine (155b) in yields of 46 and 54 %, respectively.

In a second approach [168], the synthesis of N¹⁷-substituted-norazido-ethylmorphines (155a–c, Figure 43) was performed in six consecutive transformations starting from ethyldihydromorphine (142). The alcoholic hydroxyl group in position-6 of 142 was protected with an acetyl group by treatment with acetic anhydride, resulting in 6-O-acetyl-ethyldihydromorphine (156, 94 %). The von Braun N-demethylation of 156 with cyanogen bromide and subsequent hydrolysis of the cyanamide (157) gave ethyldihydronormorphine (158). Other routes to N¹⁷-desmethyl-dihydromorphine (158) entailed treatment of 156 with diethyl azodicarboxylate (DEAD) or with methyl- or phenyl-chloroformate. The N-demethylation without protection of the 6-OH function starting directly from ethyldihydromorphine (142) was also feasible. Compound 158 was alkylated with the appropriate reagent (cyclopropylmethyl bromide, allyl bromide, n-propyl bromide) to N¹⁷-substituted-ethyldihydronormorphines (159a–c). The compounds 159a–c were converted into their mesylesters (160a–c) and the subsequent azidolysis of these compounds gave target N¹⁷-substituted-norazidoethylmorphines (155a–c).

The agonist-antagonist characteristics of the synthesized analogues were determined by the hot-plate method in rats. A ten-fold higher morphine (1a) dose was necessary to achieve the same analgesic effect in conjunction with 5 mg/kg s.c. ECAM (155a) compared to the controls without antagonist co-treatment [168]. With oral application of 155a (25–30 mg/kg), a 2–3-fold higher morphine (1a) doses was required for an equianalgesic effect. Similarly to ECAM (155a), the application of N¹⁷-allyl-norazidoethylmorphine (155b) at 50 mg/kg (s.c.) raised by tenfold the necessary morphine (1a) dose. Interestingly, ECAM (155a) showed an anorexic effect in food-deprived rats.

2.9.2. 14-Hydroxy-8-azido-8-desoxyallopseudocodeine Derivatives

In 1988, Gulyás et al. [169] synthesized a series of 6- and 7-halo substituted 14-hydroxy-8-azido-8-desoxyallopseudocodeine derivatives (Figure 44). The starting materials for these investigations, 6-demethoxythebaine (105), 6-chloro- and 6-bromo-6-demethoxythebaine (118a–b), 7-chloro- and 7-bromo-6-demethoxythebaine (113a–b) were prepared according to earlier protocols of the Makleit group [119,124,170] (see also sections 2.8.4.–2.8.6.). 6-Demethoxythebaine (105) was obtained from neopine mesylate (100) with tetrabutylammonium fluoride [119], 6-chloro- and 6-bromo-6-demethoxythebaine (118a–b), from 14-chloro-codeine tosylate (119a) with LiCl or LiBr [170], and finally 7-chloro- and 7-bromo-6-demethoxythebaine (113a–b) from 7α-halo-neopine mesylates (112a–b) by boiling with potassium tert-butylate in methanol [124]. 6-Demethoxythebaine derivatives (105, 113a–b, 118a–b) were converted to the corresponding 8β,14β-epoxides (161a–e, 8β,14β-epoxycodides) with formic acid-hydrogen peroxide (low yields: 22–42 %), or alternatively with disuccinylperoxide or m-chloroperbenzoic acid in 10 % formic acid. The epoxides (161a–e) were subjected to azidolysis (5 equiv. NaN₃, dioxane, H₂O, 100 °C, 10 h) to give 6- and 7-substituted 14-hydroxy-8-azido-8-desoxyallopseudocodeine derivatives (162a–e) [169] in 28–63 % yields.

The latter compounds were transformed to the corresponding 8α-amines (163a–e) with a yield of 28–63 %. The acetolysis of the epoxides (161a–e) was performed with 10 % acetic acid at 70 °C (1.5 h), which gave 8α-acetoxy-14-hydroxy derivatives (164a–e, 36–83 %). The hydrolysis of the acetates with 10 % KOH in ethanol resulted in 6- and 7-substituted 14-hydroxyallopseudocodeine derivatives (165a–e) [169] in 36–90 % yield.

In a subsequent study, Gulyás and Makleit [171] investigated the reactions of 8β,14β-epoxycodide (161a, Figure 45) containing a vinyl-oxirane sub-structural unit. The starting material (161a) [169] was prepared from 6-demethoxytebaine (105). The reduction of 161a with lithium aluminium hydride in diethyl ether gave 14-hydroxydesoxycodeine-C (166) in 50 % yield. When performing the azidolysis of 8β,14β-epoxycodide (161a) with azidotrimethylsilane in the presence of zinc (II) iodide the sole isolated product was 14-trimethylsilyloxy-8-azido-8-desoxyallopseudocodeine (167, 60%). Removing the TMS protecting group with sodium fluoride in methanol yielded the same 14-hydroxy-8-azido-8-desoxyallopseudocodeine (165a, 67 %), which was prepared earlier from 161a with sodium azide (dioxane-H₂O, 100 °C, 10 h) [171]. When treating 161a with Cl^y anions in aqueous-acidic acetonitrile (LiCl, 10 % HCl – CH₃CN) [171], there was formation of 6-chloro-14-hydroxyisocodeine (168, 14-hydroxy-α-chlorocodide, 20 %), 14-hydroxy-isocodeine (169), and 14-hydroxy-allopseudocodeine (106).

Due to the difficulties encountered during the separation of the solvolysis products, the 6β,14β-dihydroxy (169, iso)/8α,14β-dihydroxy (106, allopseudo) product ratio was determined by proton NMR spectrum analysis of the mixture, which gave a ratio of 5:6. The analogous reaction of 8β,14β-epoxycodide (161a) with Br^y ions in aqueous-hydrobromic acid – acetonitrile mixture (LiBr, 10 % HBr – CH₃CN) resulted in 8-bromo-8-desoxy-14-hydroxy-pseudocodeine (170, 14-hydroxy-β-bromocodide, 31 %) as the main product, and a mixture of by-products 169/106 (ratio: 3:4) [171]. The formation of the main products with different structures, i.e., 6β-chloro (168) and 8β-bromo (170), is explicable by steric-, electronic- and stability- factors, and is in agreement with the earlier results of Bognár et al. [172] concerning the nucleophilic substitutions of 14-hydroxydihydrocodeine tosylate (73a). The reaction of 73a with F^y or Cl^y nucleophiles gave the corresponding 6-substituted-6-desoxy-14-hydroxydihydroisocodeine derivatives, whereas using Br^y anions resulted in 8-bromo-8-desoxy-14-hydroxypseudocodeine (170) [172]. The treatment of 8β,14β-epoxycodide (161a) with the organocuprate reagent lithium dibutylcuprate (LiCu(C₄H₉)₂, prepared in situ from copper (I) iodide and butyllithium in diethyl ether -hexane at – 78 °C), resulted in the 6β-butyl derivative (171) in 21 % yield [171].

2.9.3. Azido Derivatives of 6,14-ethenomorphinans

Berényi et al. investigated the azidolysis of tosylesters (174a–g) derived from different in ring C bridged morphinans (Bentley compounds, or 6,14-ethenomorphinans [173,174,175,176], 173a–g, Figure 46) [177,178]. The morphinandienes 4, 105, and 118a were transformed with methyl vinyl ketone or acrolein in a diastereoselective Diels-Alder reaction to the corresponding in ring-C bridged 7-acetyl- (172a,c,d) or 7-formyl- (172b) adducts. Next, the keto group (172a–d) was reduced with NaBH₄ in methanol to a mixture of diastereomeric secondary alcohols (173a, b, d–g) [177] or to the appropriate primary alcohol (173c) [178], which were tosylated (TosCl, pyridine, room temperature, 1–6 days) to yield 174a–g. After separation by column chromatography of the diasteromerically pure tosylates (174a–b, 174d–g), they were subjected to azidolysis (NaN₃, 100 °C, 1.5 h). A novel class of morphinans containing a 4-azatetracyclo-[4.4.0.0^2.4.0^3.8]decane ring system (176a–g) was prepared by heating the azides (175a–g) in N,N-dimethylformamide at 100 °C for 24 h.

When using longer reaction times for the azidolysis of the tosylates (174a–g), substitution and subsequent cyclization by intramolecular addition of the azide to the Δ^18,19-double bond resulted directly in the aziridine derivatives 176a–g. The authors found a significant difference in the course of the reactions for 20S- and 20R- tosylesters. After azidolysis of the 20S-tosylates (174), 7-ethylidene derivatives were isolated as major products resulting from an elimination reaction. However, the azidolysis of the 20R-tosylates (174) under identical conditions lead to the formation the 20S-azides (175), by inversion of the configuration. Despite detection of the products by TLC, their isolation and characterization was possible in only a few cases (e.g. 20S-175g, and 175c) due to their generally low stability. When tosyl ester (174c) of the primary alcohol (173c) was subjected to azidolysis (5 equiv., NaN₃, DMF, H₂O, 100 °C, 1.5 h), the 7-azidomethyl derivative (175c) and the aziridine (176c) were formed. After separation by column chromatography, the product ratio 175c/176c was 9:1. The azide 175c was converted to the aziridine (176c, 59 %) by heating in N,N-dimethylformamide at 100 ° for 24 h. The structure of 175c was also proven by chemical reaction; its reduction (hydrazine hydrate, Raney-nickel, EtOH, reflux, 30 min) resulted in the primary amine 7α-aminomethyl derivative in 69 % yield. In 1992, Batta et al. [179] reported the detailed NMR analysis of the 20R- and 20S diastereomeric alcohols (173a–b, 173d–g), their tosylates (174a–b, 174d–g), and also the 4-azatetracyclo[4.4.0^2,4.0^3,8] derivatives (176a–f). The authors [177,178,179] further discussed the differing behaviours of the 20S and 20R diastereoisomers, explaining the greater predisposition of the 20S-tosylates for elimination by the more advantageous, nearly antiperiplanar position of the 20-OTos group to the 7β-H.

The primary and secondary azides (175a–g, Figure 46) developed by Berényi et al. [177,178,179] proved to be unstable due to the intramolecular addition of the azido group into the Δ^18,19-double bond. They degraded into azatetracyclodecane derivatives (176a–g, Figure 46) with nitrogen loss via heating or spontaneously at room temperature, and were thus unsuitable as ligands for pharmacological investigations. Therefore, the Makleit research group extended their investigations to the 20-tertiary azides of 18,19-dihydro-6,14-ethenomorphinans. In 1993, Sepsi et al. [180] described the synthesis of a series of tertiary azides with morphinan scaffold.

The first strategy, which was synthesis of the target azides via nucleophilic substitution of aryl or alkylsulfonate esters, failed due to conversion of the tertiary alcohols (e.g., 20-methylthevinol, 173a, Figure 47) into the corresponding 20-O-tosylate or mesylate (e.g., 177). However, the use of hydrazoic acid and BF₃-etherate [181] turned out to be practicable. 20-Methylthevinol (173a, Figure 47) thereby converted into the tertiary azide (178) in 62 % yield, and its reduction to the corresponding tertiary amine (179) was successfully performed (83 %) with hydrazine hydrate/Raney-Ni in ethanol (reflux, 30 min). As expected, the thermal treatment of 178 (DMF, 100 °C, 48 h) gave the aziridine (180) as a result of the addition of the azido group into the 6,14-etheno bridge.

To avoid the aforementioned ring closure reactions, representatives of the 6,14-ethano series, tertiary alcohols (182a–e, Figure 48) prepared from dihydrothevinone (181) in Grignard reactions (RMgX, R = Me, Et, nPr, Ph, tBu), were chosen as starting materials for the synthesis of the target tertiary azides (183a–e). The replacement of the 20-OH tertiary hydroxyl group of 182b–d (20R-182b, 20R-182c, 20S-182d, and 20S-182e) for an azido group with hydrazoic acid and boron trifluoride etherate resulted in a diastereomeric mixture of 20R and 20S of azides (183b–d, 35–44 %) as major products, but also gave elimination products (184b–d, 5–13 %). The diastereomeric azides ratios were determined from the ¹H-NMR spectra based on the integrals of characteristic signals (20-CH₃, 6-OCH₃, 5β-H): 183b (4:1), 183c (3:1), and 183d (1:1).

Efforts to split up the diastereomeric azides (183b–d) were unsuccessful, but there was separation of the diastereomeric mixture of amines (185a–b), prepared from 20R-propyldihyrothevinol (183c) by reduction with hydrazine hydrate and Raney-Ni in boiling ethanol. The conversion of 20-tert-butyldihydrothevinol (182e) to the corresponding tertiary azide (183e) failed. Instead of 183e, the furanocodide derivative (190) was isolated as the major product and the anhydro/olefin compound (191) as minor by-product. Furanocodides are known acid-catalyzed rearrangement products of thevinols [182,183]. For preparation of the tertiary-azide with orvinol structure (189), 20-methydihydrothevinol (182a) was 3-O-demethylated to 20-methyldihydroorvinol (186). The phenolic hydroxyl group of 186 was protected by application of the Welsh method [88]. 3-O-Acetyl-dihydroorvinol (187) was reacted with hydrazoic acid and BF₃ ^. Et₂O to give the azide 188 (60 %), which was deprotected with hydroxylamine hydrochloride to yield 189 (79 %). The tert-azido-orvinol derivative (189) [180] as lead compound was tested for OR agonist properties in the GPI in vitro and the mouse hot plate tests (HP) in vivo, proving to be an μOR agonist. In the mouse HP tests, 189 and 186 showed 10–18 times higher potency relative to morphine (ED₅₀ [mg/kg] = 0.59 (189), 0.32 (186), 6.0 (1a). However, in comparison with the native 20-methyldihydroorvinol (186), the azido compound 189 did not show higher relative activity [180].

2.9.4. 6-Azido-6-demethoxythebaine

In 1997, Csutorás et al. [184] reported the synthesis of 6-azido-6-demethoxythebaine (193, Figure 49) in six consecutive transformations starting from thebaine (4). 6-Azido-6-desoxy-14-hydroxyisocodeine (87) was prepared from 4 as described earlier by Makleit et al. (4 → 132 → 66 → 70a–b → 87, Figure 49) [106]. The sulfonate esters (70a–b) were converted to 6-azido-6-desoxy-14-hydroxyisocodeine (87) by treatment with NaN₃ in DMF/H₂O at 100 °C for 4 h. A more prolonged reaction time under the same conditions gave 8-azido-8-desoxy-14-hydroxypseudocodeine (85) through rearrangement. Otherwise, the reaction of 87 with two equivalent PBr₃ resulted in 7β-bromo-6β-azidodesoxyneopine (192) in 77 % yield. In that product (192), the trans-diaxial orientation of the 7β-bromo substituent relative to the 6α-hydrogen favours hydrogen bromide elimination, which is readily obtained by treatment with potassium tert-butoxide in ethanol (room temperature, 30 min, 75 %). The structure of 6-azido-6-demethoxythebaine (193) was confirmed by IR and NMR spectra and by chemical reaction. It was stable at – 20 °C, but decomposed within a few days at room temperature due to its vinylazide structural element. In a search for derivatives showing higher stability, the reaction of the azidodiene (193) with carbon disulphide in the presence of triphenylphosphine (reflux, 2 h) gave 6-isothiocyanato-6-demethoxythebaine (194) in 51 % yield. The hetero Diels-Alder reaction of 6-azido-6,8-morphinandiene (193) with 4-phenyl-4H-1,2,4-triazoline-3,5-dione (PTAD) at 0 °C readily gave a stable cycloadduct (195) in excellent yield.

2.10. Fluorinated Morphinans

Halogenated derivatives of morphine alkaloids have been the subject of continuous interest in organic chemistry [16,18]. Brominated morphinans (e.g. 1-bromo and 1,7-dibromo derivatives of dihydrothebainone and β-dihydrothebainone, 1-bromo-nordihydrothebainone) were valuable intermediates for the first morphine total synthesis accomplished by Gates and Tschudi [185], and were later successfully used for the preparation of B/C-trans-fused 1-bromodihydrocodeinone [186] and for the synthesis of codeine from dihydrothebainone [187,188]. Iodinated and brominated morphinans are precursors for the synthesis of tritiated radioligands [189,190], which are valuable research tools for OR pharmacological investigations. The effects of fluorine incorporation in biologically active derivatives has been thoroughly investigated [191], and are the subject of a very recent comprehensive survey of fluorine-containing morphinans by Sandulenko et al. [192].

2.10.1. Ring-C Fluorinated Morphinans

The morphine alkaloid research group at the University of Debrecen has paid particular attention to the synthesis of in ring-C halogenated morphinan derivatives [18,193]. 6-Fluoro-6-deoxyisocodeine (59a, Figure 50) was obtained via the S_N2 reaction of codeine tosylate (63a) with tetrabutylammonium-fluoride (TBAF) [100] in 48 % yield, and separately from pseudocodeine tosylate (97) [110] in an S_N1’ reaction with the same nucleophile in acetonitrile, albeit in poor yield (10 %). In 1972, Bognár et al. [172] described the synthesis of 6-fluoro-6-deoxy-14-hydroxyisocodeine (80) from 14-hydroxycodeine tosylate (70a) with TBAF. In an extension of earlier studies of the group [104,106], Somogyi et al. [194] investigated the S_N2 type reactions of dihydrocodeine mesylate (71b) with halide anions (F^y, Cl^y, Br^y, I^y).

6-Fluoro-6-deoxydihydroisocodeine (196a, 6-fluorodihydrocodide, α-fluorodihydrocodide) was prepared from dihydrocodeine mesylate (71b) [194] with tetrabutylammonium fluoride in acetonitrile in low yield. 6-Chloro-6-deoxydihydroisocodeine (196b, 6-chlorodihydrocodide, α-chlorodihydrocodide) was also prepared by heating of 71b for 1 h with ten equiv. of lithium chloride in 75 % yield. When 71b was treated with LiBr in DMF at 100 °C for 43 h, 6-bromo-6-deoxydihydroisocodeine (196c, 6-bromodihydrocodide, α-bromodihydrocodide) was isolated in 26 % yield. Performing this latter reaction at higher temperatures (reflux for 4 h) desoxycodeine-C (197, Δ⁷-desoxycodeine) was formed in 63 % yield. When treating of 71b with 10 equiv. of sodium iodide the corresponding 6-iodo derivative could not be isolated, instead 197 was obtained in 29 % yield.

In 1980, Boswell and Henderson [195] reported the fluorination of morphinan-6-ones (e.g., dihydrocodeinone (49), and 14-hydroxydihydrocodeinone) with diethylaminosulfur trifluoride (DAST). 6,6-Difluoro compounds were isolated as main products, and 6-monofluoro-Δ^6,7 unsaturated derivatives as by-products. In 1984, Ganti [196] synthesized numerous 14-fluoro-analogues of oxymorphone (198a, Figure 51), naltrexone (198b), and nalbuphine (198c) from 6-O-protected 14-hydroxy-dihydromorphinones with DAST.

In 1984, the research group of Kenner C. Rice [197,198,199,200,201,202] developed 6β-fluoro-oxymorphol (204a, foxy), and 6β-fluoro-naltrexol (204c, cyclofoxy) analogues (Figure 52). They performed the synthesis of 6-fluoro-6-desoxy-14-hydroxydihydroisomorphine derivatives (204a–d) in two steps from the corresponding 3-O-acetyl-6-trifluromethansulfonyloxy-14-hydroxydihydromorphines (203a–d).

For instance, the Rice group prepared cyclofoxy (204c, cyF) by introducing the fluorine substituent was in the 6β-position via an S_N2 reaction by displacement of the 6α-OTf leaving group of 205c with fluoride ions (potassium fluoride) in the presence of 18-crown-6 in acetonitrile. They then removed the acetyl-protecting group by treatment the fluorinated intermediate (203c) with NH₄OH in methanol. Labelled foxy and cyclofoxy derivatives see in section 2.12.

Woudenberg and Maat prepared chlorine-containing etorphine analogues from the 7-chloro-morphinan-6,8-dienes (7-chloro-6-demethoxythebaine and 7-chloro-5β-methyl-6-demethoxythebaine) in multistep-syntheses [203], and subsequently evaluated their pharmacological characteristics, showing generally high affinity for μOR and variable selectivity for κOR subtypes [204]. The Makleit group investigated Diels-Alder (DA) cycloaddition of the highly reactive dienophile 4-phenyl-4H-1,2,4-triazoline-3,5-dione (PTAD) to 7-halomorphinandienes (Cl, Br) [205]. In 1990, studying the DA reaction of thebaine with 2-fluoroacrolein, Jeong et al. observed the formation of a diastereomeric mixture of 7-fluoro-7-formyl-DA adducts (207a–b, 7α/7β, Figure 53) and a hetero-DA adduct [206]. The same research group reported the reactions of thebaine with trifluoromethyl substituted acetylenic dienophiles (trifluoropropyne, hexafluoro-2-butyne) and with perfluoroaldehydes [207]. Hetero Diels-Alder (HDA) reaction of thebaine and 5-methyl-thebaine with trifluoroacetaldehyde resulted in 14-(trifluoro-2-hydroxy-ethyl)-5-methylcodeinone derivatives [208].

In 1992, Berényi and colleagues elaborated a new methodology for the synthesis of 6-fluoro-6-demethoxythebaine derivatives (208a–c, N¹⁷-substituent: CH₃, H, n-propyl) from 14-hydroxy-6-fluoro-6-deoxyisocodeine [209]. They used a microwave-promoted synthesis for acid-catalyzed rearrangement of morphinans and fluoromorphinans (208a–c) [210] to yield the corresponding aporphine derivatives. Subsequently, they reported the cycloaddition reaction of 6-fluoro-6-demethoxythebaine with methyl-vinyl ketone and the synthesis and biological evaluation of 6-fluoro-6-demethoxy-20-methylorvinol (209) [211,212]. In 2013, Magnus performed the synthesis of 7β-fluorodihydrocodeine (210, Figure 53) [213]. The six-step transformation from codeine (2a) involved the fluorination of a 6,7-epoxyde intermediate with a hydrogen-fluoride-pyridine complex.

2-Haloaporphines are dopamine agonists that can be synthesized from halomorphinans by acid-catalyzed rearrangement. 2-Fluoro-N-n-propylnorapomorphine (2F-NPA) [209,214] is the most important member of this class, with extremely high affinity (K_i = 12 pM) and D₂ subtype selectivity (D₂/D₁ > 50,000) [215]. In 1994, Hosztafi and Makleit [216] reported the synthesis of numerous C-1 halogenated (1-chloro and 1-bromo) derivatives of codeine, morphine and their 7,8 dihydro analogues, and subsequently the rearrangement of 1-halogenated-codeines to the corresponding in ring-D substituted 8-halogen-apomorphines [217]. The synthesis of [8,9-³H]apomorphine was accomplished by bromination of apomorphine hydrochloride in trifluoroacetic acid (TFA) with molecular bromine [218] to 8,9-dibromo-apomorphine and following tritium dehalogenation (³H₂, 10% Pd/C, ethanol). Filer [219] presented the radiosynthesis of 8,9-di-[¹⁸F]fluoroapomorphine by fluorodestannylation of 8,9-di-(trimethylstannyl)-apomorphine. 8,9-Dibromoapomorphine and/or 8,9-diiodoapomorphine were applied as intermediates for the preparation of the labelling tin precursor.

2.10.2. 1-Fluoro-Substituted Morphinans

In 1974, Lousberg and Weiss [220] reported the first synthesis of 1-fluorocodeine (216a) in 1974. Subsequently, Makleit and Dubina [221] published an improved method for its preparation through pyrolysis of diazonium fluoroborate (215a) (Balz-Schiemann reaction). 1-Fluorodihydrocodeine (216b) was prepared from 1-fluorocodeine (216a) by heterogenous catalytic reduction [221].

In 2016, Hosztafi and Marton [222] synthesized 1-fluoro-substituted codeine derivatives (220a–b, N¹⁷-allyl, N¹⁷-n-propyl, Figure 54) and their 7,8-dihydro derivatives (220c–d). They first prepared 1-fluoromorphine (217a) and 1-fluorodihydromorphine (217b) from their respective codeine precursors via 3-O-demethylation with boron tribromide in chloroform. 1-Fluorocodeine (216a) and N¹⁷-n-propyl-1-fluorocodeine (220a) were subjected to acid-catalyzed rearrangement using methansulfonic acid (100 C°, 45 min) to give respectively 8-fluoroapocodeine (222a) and N⁶-propyl-8-fluoroapocodeine (222b), respectively with low yields (20–24 %). 3,6-di-O-Acetyl-1-fluorodihydromorphine (221) was synthesized by acetylation of 1-fluorodihydromorphine (218b) with acetic anhydride (100 °C, 1 h, 55 %). In 2018, Hosztafi et al. reported the synthesis of 1-iodo-substituted codeine and dihydrocodeine derivatives [223].

2.10.3. Fluorinated 6,14-ethenomorphinans

In the past decade, the Moiseev research group expended considerable efforts towards the synthesis of fluorinated 6,14-ethenomorphinan derivatives [192]. In 2016, they reported the synthesis of the key starting material for their investigations of the 7α-trifluoroacetyl analogue of thevinone (172a): 21,21,21-trifluorothevinone (224, Figure 55), starting with thebaine (4) [224]. Due to the low stability of the dienophile trifluoromethyl vinyl ketone, the direct conversion of 4 to 224 in a Diels-Alder reaction failed. This led them to adopt a three-step procedure in which thebaine was first converted in a [4+2] cycloaddition reaction with acrolein to the 7α-formyl cycloadduct (172b, thevinal, Figure 55). This was reacted with the nucleophilic trifluoromethylating Ruppert-Prakash reagent (TMSCF₃) to yield the diastereomeric mixture of the secondary alcohols 20R- and 20S-223, with generation of trifluoromethyl anion (CF₃^y) in situ from TMSCF₃ in the presence of tetrabutylammonium fluoride (TBAF). The mixture of the epimeric alcohols (223) was transformed to the desired 21,21,21-trifluorothevinone (224) by Swern oxidation. Subsequently Zelentsova et al. [225] prepared 21,21-difluorothevinone (225) and 21-fluorothevinone (226) via defluorination of 21,21,21-trifluorothevinone (224) with magnesium and trimethylsilyl chloride and following acidic hydrolysis of the TMS intermediate. Grignard reaction of 224 with phenylmagnesium bromide gave a mixture of fluorinated 20R- and 20S-phenylthevinols (227) [224,226,227].

In 2023, Sandulenko et al. [228] synthesized numerous 20-methyltrifluoro-6-O-desmethylorvinols (229, 232, 235, Figure 56) starting from thevinone (172a) and N¹⁷-allyl-dihydronorthevinone (233). Notable, structure-activity relationships are so far reported for only a few 6-O-desmethylorvinols [229]. The addition of Me₃SiCF₃ to thevinone (172a) resulted in 20R-228. Boiling the latter compound with 48 % hydrobromic acid for 30 min led to 20-trifluoromethyl-6-O-desmethyl-orvinol (229, 60 %) via an unusual simultaneous 3-O- and 6-O-demethylation. For preparation of 20-trifluoromethyl-6-O-desmethyl-cyprenorphine (232), 20R-228 was N-demethylated with ethyl azodicarboxylate to 20-trifluoromethyl-northevinol (230). Next, the secondary amine was acylated with cyclopropanecarbonyl chloride, whereupon the resulting N¹⁷-acyl compound reduced with lithium aluminium hydride to the N¹⁷-cyclopropylmethyl derivative (231). O-Demethylation of 231 with boron tribromide in dichloromethane at – 78 °C gave 229 in 76 % yield. Analogously, N¹⁷-allyl-dihydronothevione (233) [230] with Me₃SiCF₃ gave the 20R-234, which was O-demethylated to N¹⁷-allyl-20-trifluormethyl-6-O-desmethyl-dihydronororvinol (235, 53 %). In general, thevinols with tertiary hydroxyl groups in position-20 are acid-sensitive compounds, thus prone to undergo by acid-catalyzed dehydration an enol ether hydrolysis and rearrangement to 14-alkenylcodeinones, anhydro-20-alkylthevinols, or to 5,14-bridged thebainones. The authors attributed the suitability of BBr₃ for O-demethylation of fluorinated thevinols to the presence of the EWG CF₃ group in the molecule [226]. The 20-CF₃ group can prevent the formation of a carbocation from the C-20 tertiary alcohols, and accordingly avoid the intramolecular rearrangements mentioned above. The target compounds (229, 232, and 235) administered s.c. were evaluated for analgesic activity in rodent tail-flick tests in comparison to morphine (1a). Their results showed that the introduction of fluorine substituent in position-20 of the orvinol scaffold did not abolish the analgesic properties; 20-trifluoromethyl-6-O-desmethyl-orvinol (229) showed analgesic activity comparable to that of morphine (1a). 20-Trifluoromethyl-6-O-desmethyl-cyprenorophine (232) with N¹⁷-cyclopropylmethyl substituent proved to be a partial agonist with weak analgesic activity and N¹⁷-allyl-20-trifluoromethyl-6-O-desmethyl-dihydronororvinol (235) displayed no analgesic activity.

In a subsequent study, the same research group placed their attention on the stereochemistry of the C-20 chiral centre of fluorine containing thevinols [231]. They thoroughly investigated the reactions of 21,21,21-trifluorothevinone (224, 7α-trifluoroacetyl, Figure 57) with various alkylmagnesium halides- (MeMgI, EtMgBr, nPrMgBr, iPrMgBr, tBuMgCl) and alkyllithium reagents (tBuLi, iPrLi) and also the addition of TMSCF₃ to thevinone (172a) and its analogues (7α-acetyl). These studies proved to be fundamental importance in elucidating the structure-activity relationships thevinols and orvinols with different C-20 configuration. The reaction of 7-acyl-type 6,14-ethenomorphinans with Grignard reagents results in complex product mixtures [50,232,233] due to various processes: Grignard addition, reduction of the carbonyl function, and base catalyzed-rearrangement. Sandulenko et al. [231] found that the reaction of thevinone (172a) and its 7α-acyl analogues (7α-propionyl, 7α-butanoyl) with trimethyl-(trifluoromethyl)-silane in the presence of CsF in THF at room temperature gave a mixture of the tertiary alcohols 20R-236 and 20S-236, with predominance of the 20R isomer. The reaction of 21,21,21-trifluorothevinone (224) with two equivalents of methylmagnesium iodide in diethyl ether at room temperature resulted in a 100:67 mixture of the 20R-236 and 20S-236 epimeric tertiary alcohols. In the reaction of 224 with MeMgI, they also studied effects of addition of metal salts (MgX₂, X = Cl, I; ZnCl₂, MeOMgI, tBuOK) on the product ratio, based on the earlier approach of Zelentsova et al. using one-dimensional ¹⁹F NMR spectra [227]. When 224 (Figure 57) was reacted with 1.2 equivalent of methyllithium in THF, there formed a mixture of isomeric tertiary alcohols 20R-236 and 20S-236 (R = Me) with predominance of the 20S epimer (e.g., at -78 °C, ratio 20R/20S = 25:100). Reacting 21,21,21-trifluorothevinone (224) with RMgX reagents other than MeMgI (R = Et, nBu, iPr, tBu) gave predominant formation of the secondary alcohols 237.

In 2023, Zelentsova et al. [225] reported the synthesis of 21,21-difluorothevinone (225) and 21-fluorothevinone (226) by defluorination of 21,21,21-trifluorothevinone (224, Figure 55 and Figure 57). Prominent members of the 6,14-ethenomorphinan series buprenorphine, diprenorphine and dihydroetorphine contain a saturated 6,14-ethano bridged C-ring [50,176,234]. Therefore, the next logical step was to try to synthesize the 18,19-dihydro analogues of 224, 225 and 226. The direct conversion of 224 to 18,19-dihydro-21,21,21-trifluorothevinone (238) by catalytic hydrogenation of the Δ^19,19 double bond failed. Even harsh reaction conditions at 60 Bar hydrogen pressure (AcOH, 10 % Pd-C, 55–60 °C, 45 h) gave only 15 % conversion rate, with isolation of the rearranged product 239 in 11 % yield instead of the desired compound (238), which is the 7α-trifluoacetyl analogue of dihydrothevinone (181). As an alternative route for the synthesis of N¹⁷-substituted-20-difluoromethyl-tevinols (242a–c), dihydrothevinone (181) was refluxed with the pronucleophile (difluoromethyl)trimethylsilane (Me₃SiCHF₂) in HMPA/THF in the presence of CsF to give 20R-240 in 35 % yield. The secondary amine 20R-241 was prepared from 20R-240 by N-demethylation with diethyl azodicarboxylate (DEAD) in acetonitrile. Subsequently the target tertiary amines (242a–c) were prepared by functionalization with the corresponding alky- or acyl halides with yields in a range of 48–52 %. 20R-242c was also prepared directly from N¹⁷-allyl-dihydronorthevinone (233, Figure 58) with Me₃SiCHF₂ in 23 % yield.

Aiming mind to synthesize the conjugates of 6,14-ethenomorphinans with bioactive molecules, the same research group prepared 7α- and 7β-carboxylic acid esters of thevinone, so called thevinoic acid fluoroalkyl esters, from thebaine (4) with β-fluoroalkyl acrylates [235].

Very recently, Finke et al. [236] reported a procedure for the preparation of 6-trifluoromethyl substituted morphinans from morphinan-6-ones. 6-Ketomorphinans (e.g., 14-hydroxycodeinone (132), 4-O-methylsinomenine, 1-iodo-4-O-methylsinomenine) were converted to the corresponding 6-trifluoromethyl compounds with (trifluoromethyl)trimethylsilane (Me₃SiCF₃, Ruppert-Prakash reagent) in the presence of Bu₄NF or tBuOK.

2.11. Application of the Mitsunobu Reaction in the Morphine Series

The Mitsunobu reaction [237,238] continues to find wide use for the functionalization of alcoholic hydroxyl groups [239], and in the field of alkaloid synthesis [240]. During this reaction, a primary or secondary alcohol (R-OH, e.g., 2a, Figure 59) substrate, reacts with a pronucleophile (H-Nu (243), a compound containing a dissociable proton) in the presence of an azodicarboxylic acid ester (e.g. DEAD, DIAD, ADDP or DTBAD, 244) and triaryl or trialkylphosphine (245, e.g. TPP or TBP, Figure 59). Following a complex reaction sequence, the alcohol (R-OH, 2a) alkylates the conjugated base (Nu^y) of the pronucleophile (H-Nu, 243), forming R-Nu (246) as main product, and side products hydrazinecarboxylic acid alkyl ester and the corresponding phosphine oxide (e.g. Ph₃P=O, 248).

The Mitsunobu reaction is an indispensable alternative to classical S_N2 type reactions. Taking place under mild (0-25 °C) and neutral conditions, the reaction operates in dipolar aprotic solvents (e.g. THF, benzene, DMF, acetonitrile). The reaction time (τ) is commonly between 0.5 and 12 hours, with rare instances of more prolonged τ-s. The Mitsunobu reaction is chemoselective: only the alcoholic hydroxyl groups react under such mild conditions. Typically, primary alcohols react more quickly than secondary alcohols, which enables selective acylation of substrates when primary and secondary hydroxyl groups are present in the same molecule. The Mitsunobu reaction is usually stereoselective and yields products with opposite configuration (Walden inversion). Side reactions (elimination or allylic rearrangement) rarely occur. As pronucleophiles, molecules with dissociable protons (pK_a <11) can be used and examples of weak-acidity nucleophiles with higher pK_a value (<15) are also known. The mechanism of the Mitsunobu reaction is depicted in Figure 60.

In the first elementary step of the process, a reactive betaine (XVII) is formed from the azodicarboxylic acid ester (244) and trialkyl or triarylphosphine (245) in an irreversible reaction.

In the second step, the reactive zwitterionic adduct (XVII) reacts with the acidic pronucleophile (H-Nu, 243) to produce the nucleophile ( Nu^y).

The alcoholic hydroxyl group (2a) attacks the protonated betaine (XVIII) to form an alkoxyphosphonium intermediate (XIX). During this process, 1,2-hydrazinecarboxylic acid dialkyl ester (247) arises.

The conjugated base form (Nu^y) of the pronucleophile (H-Nu, 243) reacts with the alkylphosphonium intermediate (XIX), from which emerges the product (R-Nu, 246) as well as triaryl or trialkylphosphine oxide (248). Applying a chiral secondary alcohol as starting substrate a change of the configuration of the chiral centre is to be expected (Walden inversion).

2.11.1. The First Application

Robson and Kosterlitz [241] postulated that denaturation of the OR binding sites in guinea-pig brain homogenates could be achieved by alkylation of the protein with phenoxybenzamine (PBZ). Indeed, incubation of brain preparations with PBZ reduced the specific binding of [³H]dihydromorphine (μ-ORs) and [³H]D-Ala²-D-Leu⁵]enkephalin (δ-ORs). Conversely, the presence of OR ligands (peptides or opiates) protected against the inactivation due to alkylating agents. Smith and Simon [242] found that treatment with the thiol (SH) alkylating reagent N-ethylmaleimide (NEM) inhibited the stereospecific binding of tritiated OR ligands (e.g., [³H]D-Ala²-D-Leu⁵]enkephalin). Bowen et al. [243] postulated that disulphide bonds in the ORs are essential for ligand binding, proposing a three-state allosteric model (μ-agonist, μ-antagonist and δ-agonist-preferring states) under regulation by a “SH” – “S-S” (thiol – disulphide) exchange mechanism.

Based in that hypothesis, Fujii et al. [244] investigated in 1988 the possibility of introducing an SH group into the C-6 position of the morphine (1a) skeleton. In a first attempt, morphine (1a) was reacted with thioacetic acid in the presence of 1,1-dineopentyloxytriethylamine (toluene, 80 °C) to yield the 3-O-acetyl-6β-thioester (250, Figure 61, 59%) and the 6β-thioester (249, 30%). In a second method, morphine (1a) was reacted with thioacetic acid under Mitsunobu conditions (diisopropyl azodicarboxylate (DIAD), triphenylphosphine (TPP)). This approach gave predominant formation of 6-S-acetyl-6-desoxy-isomorphine (249, 73%) and its 3-O-acetyl derivative (250, 22%). The absolute stereochemistry of the C-6 stereo centre was proven by NMR spectroscopic methods. In the ¹H-NMR spectra of compounds 249 and 250, the coupling constant ³J_5β,6α was 0.5 Hz, which implies a 6β-orientation of the thioacetyl group [245]. Fujii et al. [244] went on the hydrolyze S-acetyl-6-desoxy-isomorphine (249) by treatment with 0.2 M potassium hydroxide in ethanol to give 6β-thiomorphine (251, 66%). Interestingly, the analogous reaction of dihydromorphine (35) with thioacetic acid (DIAD, TPP) resulted in 3-O-acetylmorphine (254) as main product (93%) and the corresponding 3-O-acetyl-6β-thioacetyl derivative (253) as minor product (7%). When dihydrocodeine (67) was subjected to the Mitsunobu reaction with thioacetic acid under identical conditions, there was formation of the 6β-thioester (255) in almost quantitative yield. 3-O-Demethylation with boron tribromide in CHCl₃ gave 256 in 86% yield. The latter compound was hydrolyzed with 0.2 M KOH solution in ethanol to afford 6βthio-7,8-dihydro-morphine (257). The reaction of 6β-thiomorphine (251) and 6β-7,8-dihydro-thiomorphine (257) with 2-nitrobenzenesulfenyl chloride (acetonitrile, 0 °C) gave the corresponding disulphides (252,258).

The analgesic activities of 6-S-acetyl-6-desoxy-isomorphine (249, Figure 61) and the disulphide (252) derivative were tested in guinea pig ileum (GPI) [244]. Compound 249 proved to be twice as potent as morphine (1a), whereas the disulphide (252) was quasi-equipotent with morphine (1a).

In 1990, Kanematsu et al. [246] synthesized S-activated 6-sulfhydryl-dihydoisomorphine (253) starting from dihydromorphine (35) using Mitsunobu conditions by a modification of the Fujii et al. [247] approach (5 eq. AcSH, 5 eq. DIAD and 5 eq. TPP). 6β-7,8-Dihydro-thiomorphine (257) was reacted with 5-nitro-2-pyridinesulfenyl chloride (CH₂Cl₂, 0 °C) giving the corresponding disulphide (259, 84%). Analgesic activity of the synthesized 6β-(5’-nitro-2’-pyridyldithio)deoxydihydromorphine (259) was tested in guinea-pig ileum (GPI) and mouse vas deferens (MVD) preparations giving IC₅₀ values of 9.3 and 76 nM, respectively. In mice (MVD), compound 259 had approximately three times higher potency than morphine. They also investigated the thiol-disulphide exchange reaction of the 5’-nitro-2’-pyridyldithio derivative (259) with L-cysteine methyl ester in DMF.

2.11.2. Preparation of Isomorphine and Isocodeine Derivatives

At the beginning of the 1990s, there were only a few known morphine derivatives with 6-beta configuration (iso-series, 6β-OH). Their preparation in low yield procedures was only possible by application of complicated multistep syntheses, with the further disadvantage of numerous side-products. In 1969, Makleit and Bognár [107] prepared isocodeine (2b) in low yield (16 %) by acetolysis of codeine tosylate (63a) with 10 % acetic acid. Subsequently, there was a report of synthesis of isomorphine (1b) from 3-O-acetylmorphine tosylate (69a), and preparation of dihydroisomorphine (82) from isomorphine (1b) by catalytic reduction [111]. Fleischhacker [248] found higher yields when performing the solvolysis with 70 % acetic acid. In 1971, Kirby and Massey [249] reported the synthesis of isocodeine (2b) from codeine (2a) in a three-step procedure: codeine (2a) → codeine tosylate (63a) → isocodeine acetate → isocodeine (2b). Codeine tosylate (63a) was converted into isocodeine acetate with hexadecyltrimethylammonium acetate, and the alkaline hydrolysis of the resulting 6β-acetate led to isocodeine (2b, 68 % overall yield from 2a). Simon [250] was the first from the Makleit group to investigate extensively the application of the Mitsunobu reaction in the field of morphine alkaloids.

In 1991, Simon et al. [251] thereby synthesized numerous isocodeine (262a–c, Figure 62) and isomorphine (262g–j) derivatives via Mitsunobu reaction for the preparation of 6-benzoate esters (261a–j) and their subsequent hydrolysis.

This elegant two-step method [251] made available for the first time novel isocodeine- isomorphine- derivatives and previously known compounds of the iso series with acceptable yields. In the first step, codeine (2a), N¹⁷-substituted codeine derivatives (260b, 260c, nPr, allyl), 3-O-ethylmorphine (260d), 3-O-benzylmorphine (260e), 3-O-morpholinylethyl-morphine (260f), 3-O-acetylmorphine (55) or N¹⁷-substituted-3-O-acetylmorphine derivatives (260h–j, nPr, allyl, cyclopropylmethyl) were reacted with benzoic acid as the acidic pronucleophile (H-Nu, 243). The Mitsunobu reactions were undertaken in the presence of diethyl azodicarboxylate (DEAD) and triphenylphosphine (TPP, 245), resulting in the corresponding 6-O-benzoyl-isocodeine (261a–j) derivatives. The latter compounds were treated with a 10% aqueous potassium hydroxide solution to give the isomorphine and isocodeine derivatives (262a–j) with free 6β-hydroxyl group.

In 1994, Friedmann et al. [252] investigated the influence of C-6 configuration on the OR affinity of a large number of N¹⁷-substituted-morphine derivatives (C-6α-hydroxy, N¹⁷-substituent: CH₃ (1a), n-propyl (260h), allyl (260i), cyclopropylmethyl (260j) N¹⁷-substituted-isomorphine (C-6β-hydroxy, 1b, 262h–j) and their 7,8-dihydro analogues [251]. They used in vivo rat tail-flick, mouse hot plate and in vitro isolated GPI assays for the pharmacological characterization [252]. The OR agonist activity of the N¹⁷-methyl derivatives (morphine (1a), isomorphine (1b), dihydromorphine (35) and dihydroisomorphine (82)) were determined by mouse hot plate, rat hot plate and rat tail flick tests, in comparison with morphine (1a). The antinociceptive actions of the C-6α and C-6β derivatives did not differ significantly, having relative potencies 0.6–1.9 fold that of morphine (1a). In the isolated electrically stimulated GPI preparation, isomorphine (1b) was twice as potent as morphine (1a), but dihydroisomorphine (82) showed only 1.18-fold potency of dihydromorphine (35) relative to normorphine. The relative potencies to normorphine were in the range of 1.1–4.4. The epimerization of C-6α to C-6β derivatives had only slight effects on OR activities in vitro, but influence the antagonistic effect of the N¹⁷-cyclopropylmethyl and N¹⁷-allyl compounds. N¹⁷-Allyl-dihydronorisomorphine (AD₅₀ = 4.0 mg/kg, (s.c) (3.2–7.3), showed a 10-fold increase in antagonist potency compared to the parent N¹⁷-allyl-normorphine (260i, 0.48 (0.4–0.57)) in the rat tail-flick test vs. morphine.

2.11.3. Reactions of Codeine Isomers and Neopine

In the previously presented experiments, the starting secondary alcohols carried the hydroxyl group at position 6α- of the C-ring of the morphinan skeleton. In 1993, Simon [250] investigated the Mitsunobu reaction of codeine isomers (isocodeine (2b), allopseudocodeine (2c) and pseudocodeine (2d)) with various pronucleophiles (H-Nu compounds) by application of diethyl azodicarboxylate (DEAD) and triphenylphosphine (TPP). These reactions were performed under exactly the same conditions as described above for 6α-hydroxyl derivatives [251].

Upon subjecting isocodeine (2b, 6β-OH, Figure 63) to Mitsunobu reaction with benzoic acid or phthalimide, there was no conversion. A previous investigation of the nucleophile substitution reactions of isocodeine tosylate (81) by Makleit [102] found formation of pseudocodeine derivatives due to allylic rearrangement. In the Mitsunobu reaction of pseudocodeine (2d, 8β-OH) with 4-nitrobenzoic acid or phthalimide, 6β-substituted derivatives (isocodeine series) were formed due to [3,3-sigmatropic] rearrangement. Presumably, the attack of the corresponding nucleophile from the α-side is sterically hindered in an S_N2 type reaction. Makleit, Somogyi and Bognár [110] had similar experiences during investigations of the nucleophile substitution reactions of pseudocodeine tosylate (97). They found formation of 6-deoxy-6-substituted-isocodeines (99) or 8-deoxy-8-substituted-pseudocodeine derivatives (98), depending on the nature of the nucleophilic anion [110].

There was no observable reaction when allopseudocodeine (2c, 8α-OH) was treated with benzoic acid or 4-nitrobenzoic acid under Mitsunobu conditions [250]. Interestingly, when phthalimide was applied as pronucleophile, there was formation of the 8β- phthalimido derivative (98, Nu = NPht, pseudocodeine series), a result corresponding to the Mitsunobu reaction mechanism (8α-hydroxyl (2c) → 8-phthalimido-pseudocodeine (98)). The authors proved the structure of the synthesized phthalimido compound (98) by NMR spectroscopy and by chemical reaction. The reaction of the 8β-phtalimido-compound (98, Nu = NPht) with hydrazine hydrate gave the same known 8β-amine (8-amino-pseudocodeine (84)) as prepared earlier by Bognár and Makleit [104] via another route (codeine (2a) → codeine tosylate (63a) (or mesylate, 63b) → 8-azidocodide (83) → 8-aminocodide (84)).

When choosing neopine (3a) as starting material, the replacement of the C-6α secondary alcohol using benzoic acid or phthalimide as Mitsunobu pronucleophile failed [246], although in both cases there was 6-demethoxythebaine (105) present in the product mixture. Likewise, when the reaction was performed in the absence of the pronucleophile (H-Nu), only the presence of the elimination product, 6-demethoxythebaine (105), could be detected. In the neopine (3a) molecule, the 6α-hydroxyl group occupies a position pseudo axial to the 7β-proton, which represents a favourable steric condition for elimination. The elimination reactions of 6-O-mesyl-neopine (100) derivatives are well known from the literature [119,124,253].

2.11.4. Synthesis of 6β-Aminomorphinans

Conversion of codeine 6-sulfonesters (63a, 63b) to 6β-halo substituted derivatives by nucleophilic substitution (S_N2) was achievable only for F^y and Cl^y anions [99,100,101]. The reaction of the sulfonesters with Br^y or I^y nucleophiles occured via a complex mechanism (S_N2 + S_Ni’), resulting in 8β-substituted products due to a [3,3] sigmatriopic rearrangement.

We note that for the same reasons, application of N₃^y nucleophile and the subsequent reduction of the formed azide gave 8β-amino derivatives. Morphinan-6-ones were employed earlier for production of 6-amino-4,5-epoxymorphinans via reductive amination, which resulted in 6α/6β-amino epimer mixtures [254,255,256]. The Makleit research group developed a stereoselective method for the preparation of 6β-aminomorphinans [257]. In 1992, they extended their investigations [257] to a synthesis numerous N¹⁷-substituted-6β-aminocodeine (266a–d) and N¹⁷-substituted-6β-aminomorphine (266e–h, Figure 64) derivatives. The carefully-selected starting morphinan derivatives (263a–h) all contained a Δ^7,8-double bond in ring-C. By treatment of codeine (2a), N¹⁷-substituted-codeine (263b–d), 3-O-acetylmorphine (55) and 3-O-acetyl-N¹⁷-substituted-morphine (263f–h) derivatives with phthalimide under Mitsunobu conditions (DEAD, TPP, benzene), the corresponding 6β-phthalimido-codeines (264a–d) (68-92 %) and 6β-phthalimido-3-O-acetylmorphines (264e–h) (39-56 %) were formed with inversion of the configuration. Side reactions, e.g., allylic shifts, were not expected. 6β-Phthalimido-morphines (265a–d) were prepared by 3-O-deacetylation of the compounds 264e–h with hydroxylamine hydrochloride in aqueous ethanol (50 °C, 10 min). The 6β-phthalimido compounds (264a–h) were subjected to hydrazinolysis (see also Gabriel synthesis) to give the primary amine 6β-amino derivatives (266a–h) in 50-98 % yield. For preparation of 6β-amino-7,8-dihydro derivatives (267a–f), the selected Δ^7,8 compounds (266a,b,d,e,f,h) were saturated under heterogenous catalytic conditions (H₂, 10% Pd-C, EtOH). 6β-Amino-dihydrocodeines (267a–c) and 6β-amino-dihydromorphines (267d–f) were obtained in 37-75 % yield.

The Makleit group [258] accomplished the Mitsunobu reaction of N¹⁷-demethyl-N¹⁷-substituted-14-hydroxycodeines (268a–d, CH₃, nPr, allyl, cyclopropylmethyl) as well as N¹⁷-demethyl-N¹⁷-substituted-3-O-acetyl-14-hydroxymorphines (268e–h, Figure 65) with phthalimide.

The 6β-phthalimido derivatives (269a–h) were prepared with yields in a range of 20-65 %, without any observed allyl-migration. The treatment of these compounds with hydrazine hydrate in ethanol yielded 6β-amino-14-hydroxycodeine (271a–d) and 14-hydroxymorphine (271e–h) derivatives. Most of these 6β-amino derivatives were previously unavailable via the sulphonate ester → azide → amine route [48] or by reductive amination of morphinan-6-ones [254,255].

Continuing their systematic investigations, the Makleit group extended the epimerization studies for compounds saturated in ring-C (7,8-dihydro compounds): dihydromorphines, dihydrocodeines and their 14-hydroxy analogues [259]. Applying benzoic acid as pronucleophile in the Mitsunobu reaction of dihydrocodeine (67), there was incomplete conversion. Interestingly, by application of p-nitrobenzoic acid instead, the corresponding dihydroisocodeine p-nitro benzoate was isolated in a yield of 90 %. Accordingly, 14-hydroxydihydrocodeine (68) and N¹⁷-substituted-14-hydroxydihydrocodeine derivatives (273b–c, Figure 66) were reacted under Mitsunobu conditions (TPP, DEAD) with 4-nitrobenzoic acid in anhydrous benzene to yield the corresponding p-nitrobenzoic esters (274a–h). Ester cleavage of these compounds was performed by alkaline hydrolysis, giving the desired 6β epimers (275a–h) with yields in the range of 37–90 %. In this manner, the authors achieved a new stereoselective synthesis of the human metabolites of naloxone (β-naloxol (275g)) and naltrexone (199, β-naltrexol (275h)).

In a second series of experiments [259], dihydrocodeine- (67), 3-O-acetyldihydromorphine- (54), 14-hydroxydihydrocodeine- (68, 276b–d, X = OH) and 3-O-acetyl-14-hydroxydihydromorphine-derivatives (276e–h, X = OH, Figure 67) were treated with phthalimide in benzene in the presence of triphenylphosphine and diethyl azodicarboxylate (DEAD) to yield the corresponding 6β-phthalimido compounds (277a–h). For hydrolysis of the phenol esters (277e–h), they used hydroxylamine hydrochloride in aqueous ethanol giving the corresponding 6β-phthalimido-14-hydroxydihydomorphine derivatives (278a–d) with free phenolic hydroxyl group in 58–70 % yield. Treatment of the 6β-phthalimido derivatives (277a–h) with hydrazine hydrate led stereoselectively to the 6β-amino compounds (279a–h) with 51–90 % yield. At the time, these investigations were considered as novel stereoselective syntheses of 6β-oxymorphamine (279e, R₂ = CH₃), 6β-naloxamine (279g, R₂ = allyl) and 6β-naltrexamine (279h), R₂ = cyclopropylmethyl). Szilágyi et al. presented detailed NMR analysis of the prepared new compounds [260], and Fürst et al. subsequently undertook their pharmacological characterizations [252,261].

2.11.5. Synthesis of 6β-succinimido Derivatives

Upon reacting codeine (2a) with the succinimide pronucleophile (HSu) [257] under identical Mitsunobu conditions as above (DEAD, TPP, benzene), 6β-succinimido-codeine (280a, Figure 68) was obtained in 80 % yield. In the case of 3-O-acetylmorphine (55), only the isolation of the deacetylated 6β-succinimido-morphine (281b) was feasible. Saturation of the Δ^7,8 double bond under heterogenous catalytic conditios (H₂, 10 % Pd-C, RT, atmospheric pressure) was only achievable for 6β-succinimido derivatives (280a, 73 %, 280b, 55 %). In the case of the appropriate 6β-phthalimido compounds (264a–h), the reduction of the Δ^7,8 double bond was not realizable under atmospheric pressure.

2.11.6. Reaction of 14-halogenocodeines

14-Halogenocodeines (282a–b, Figure 69) contain both an allyl alcohol (C⁸=C⁷-C⁶-OH) and an allyl halide (C⁷=C⁸-C¹⁴-X) sub-structural unit within the same molecule. Thebaine (4) reacted with N-halosuccinimides (NCS, NBS) in acetic acid or acetone-water (2:1 (v/v)) gave 14-halocodeinones [126].

The reduction of codeinone derivatives (e.g., codeinone (24), 14-hydroxycodeinone (132), 1-bromocodeinone, and 14-bromocodeinone (109)) with sodium borohydride led stereoselectively to the corresponding codeines (e.g., 282a–b, 6α-hydroxy compounds). 14-Chlorocodeine (282a) [116] and 14-bromocodeine (282b) [126] are useful starting materials for the synthesis of neopine derivatives. The reduction of neopinone (23) with NaBH₄ results in a mixture of neopine (3a) and isoneopine (3b). Okuda et al. [262] found that the NaBH₄ reduction of 14-bromocodeinone (109) in methanol-water gave three products: the main product neopine (3a, 46 %), isoneopine (3b, 6 %) and indolinocodeine (28 %) were isolated.

In 1993, Simon et al. [263] investigated the Mitsunobu reaction of 14-halogenocodeines (282a–b, Figure 69) with phthalimide as pronucleophile in the presence of triphenylphosphine (TPP) and dialkyl azodicarboxylates (DEAD or DIAD) or diphenyl azidophosphate (DPPA). The reaction of 14-chlorocodeine (282a) with phthalimide (TPP, DEAD, benzene) resulted in three products. The main product, a Δ^6,8-conjugated diene (285a) containing a hydrazine dicarboxylic diethyl ester substituent in position-6, was obtained in 36 % yield. Further products, 6β-phthalimido-14-chloro-6-deoxycodeine (283, 12 %) formed by inversion and 6β,14-dichloro-6-deoxycodeine (284a, 13 %), were isolated by column chromatography in a ratio of about 1:1. Interestingly, when 14-bromocodeine (282b) was reacted with phthalimide under identical conditions (TPP, DEAD, benzene, RT, 1 h), there was no observed formation of a 6β-phthalimido derivative, presumably due to the steric hindrance by the bulky 14-bromo substituent. In this case, the main product were the thebaine analogue (285a, 34 %) followed by 6β,14-dibromo-6-deoxycodeine (284b, X = Br, 25 %). To prove the structure of the Δ^6,8-diene (285a) with hydrazine dicarboxylic diethyl ester substituent in position-6, it was boiled with methyl vinyl ketone to yield the corresponding DA adduct thevinone analogue (286a). In a second approach, 14-halogenocodeines (282a–b) were reacted under the above-mentioned conditions (TPP, DEAD) in the absence of a pronucleophile (H-Nu). Formation of 6β,14-dichloro-6-deoxycodeine (284a, 32 %) and 6β,14-dibromo-6-deoxycodeine (284b, 25 %) as well the Δ^6,8-diene (285a, 32 %, from 282a and 25 % from 282b respectively) were observed. The authors [263] proved the structure of 6β,14-dihalo-6-deoxycodeines (284a–b) by spectroscopy and chemical reactions. When these derivatives (284a–b) were allowed to react in N,N-dimethylformamide in the absence of a nucleophile [128], the same 6-halogeno-6-demethoxythebaine (118a–b) derivatives were obtained, as had been prepared earlier in the nucleophilic substitution reaction of 14-chlorocodeine tosylate (119a) with Cl^y and Br^y ions, respectively (see section 2.8.6.) [127]. The formation of the 6β,14-dihalo-6-deoxycodeines (284a–b) and the Δ^6,8-dienes (118a–b) in this reaction was explained by the in-situ liberation of hydrogen halides (HCl, HBr), which could then act as pronucleophiles in the Mitsunobu reaction. To substantiate the presence of the in situ generated H-Nu compound, the hydrochloride addition salt of codeine (2a) was allowed to react under Mitsunobu conditions (TPP, DEAD), which gave α-chlorocodide (59b, Figure 69) in 70 % yield. In a further approach, 14-halogenocodeines (282a–b) were reacted with diphenylphosphorylazide (DPPA) under Mitsunobu conditions, giving the 6β-azido-14-halogeno-6-deoxycodeine derivatives (287a, X = Cl, 37 %, 287b, X = Br, 25 %).

The reaction of morphine (1a) or codeine (2a) with thionyl chloride resulted in α-chloromorphide (288a) and α-chlorocodide (59b), respectively [99]. Notably, the reactions of codeine (2a) with thionyl bromide [99] or codeine tosylate (63a) with lithium bromide [99] both resulted in β-bromocodide (65b, Figure 17), the product of thermodynamic control. When hydrogen halide salts (hydrochloride or hydrobromide) of codeine (2a) or morphine (1a) was allowed to react under Mitsunobu conditions (DEAD, TPP, toluene) in the absence of other nucleophile [264], 6β-halogen substituted derivatives were isolated (e.g., from codeine hydrochloride → α-chlorocodide (59b), from codeine hydrobromide → α-bromocodide (288c), Figure 70). The importance of the above-mentioned results of Simon et al. [264] have to be emphasized, as this was the first isolation of α-bromocodide (288c), the product of kinetic control. Analogue reactions starting from the hydrogen halide (HCl, HBr) salts of 14-hydroxy derivatives (14-hydroxycodeine (66), 14-hydroxymorphine (289a), 14-hydroxydihydrocodeine (68) and 14-hydroxydihydromorphine (289b)) led to the corresponding 6β-halogen (Cl, Br) substituted compounds (290a–d). The reactions of Δ^7,8 unsaturated derivatives (1a, 2a, 66, 289a) were complete within 1 h, but the conversion proceeded much slower (3–4 h) in the 7,8-dihydro series (68, 289b).

2.11.7. Novel Applications

In this section, we present a few novel examples highlighting potential applications of the Mitsunobu reactions in the field of semisynthetic morphinans, as informed by results of the Makleit group in the 1990s. In 2014, Fujimura et al. [265] reported the pharmacological characteristics of the OR ligand TRK-130 (278d, naltalimide, N-[(5R,6R,14S)-17-cyclopropylmethyl-4,5-epoxy-3,14-dihydroxymorphinan-6-yl]phthalimide). The compound TRK-130 (278d), which proved to be a selective partial μOR agonist (K_i for μOR = 0.268, δOR = 121, and κOR = 8.97 nM, respectively, δ/μ = 451, κ/μ = 33.5) [265] in radioligand-binding assays with human ORs. Naltalimide (278d) is used as a prophylactic-agent against urinary-incontinence, i.e., overactive bladder (OAB) [265]. The synthesis of naltalimide (278d) was achieved by the Simon et al. method [259] from 3-O-acetyl-N¹⁷-cyclopropylmethyl-14-hydroxydihydromorphine (3-O-acetyl-α-naltrexol, 276h, route: 276h → 277h → 278d, X = OH, see also Figure 68) in a Mitsunobu reaction with phthalimide. Izumimoto et al. [266] developed an alternative synthesis route for naltalimide (278d) and its N¹⁷-allyl analogue (278c) from β-naltrexamine (279h) [259] and β-naloxamine (279g) [259], respectively. The corresponding 6β-amine (279g–h) was reacted with phthalic anhydride in DMF in the presence of triethylamine at 140 °C [266] to yield the 6β-phthalimido derivatives (278c–d) with yields of 34–58 %.

For potentially treating substance-dependence, the research group of Kenner C. Rice developed heroin vaccine haptens [267,268] (Heroin Hapten (MorHap, 291, Figure 72)). One of the chosen hapten scaffolds was 6β-amino-6-desoxymorphine (266e). For its preparation, they applied the method described earlier by the Makleit group [257]. In brief, they selectively acetylated morphine to 3-O-acetylmorphine (55) by the Welsh method [88]. They treated that product with phthalimide under Mitsunobu conditions (TPP, DIAD, toluene, RT, 2 h) to yield 6β-phtalimido-3-O-acetylmorphine (264e, 88 %). The stereochemistry of the phthalimido derivative (264e) was substantiated by X-ray crystallographic analysis. The cleavage of the phthalimido protecting group with hydrazine hydrate in ethanol (55 °C, 90min) led to 6β-amino-6-desoxymorphine (266e) in 81 % yield. Subsequently, the free 6β-amine was acylated with the activated ester 2,5-dioxopyrrolidin-1-yl 3-(tritylthio)propanoate to the desired hapten (291, MorHap). Finaly, the hapten (291, MorHap) was conjugated to the carrier protein (tetanus toxoid (TT) or cross-reactive material 197 (CRM₁₉₇)). Antibodies formed due to the immunisation effect of the TT-MorHap conjugate led to nearly complete protection against heroin dependence in mouse experiments.

Scammels et al. [269] prepared fluorescently labelled morphinan-sulfo-Cy5 conjugates for visualization of ORs in living cells. For these bioimaging investigations, they developed a fluorescent partial OR agonist ligand. For the synthesis of C-6 amide pre-congeners first 3-O-(4-methoxybenzyl)-6-amino-6-desoxymorphine (296, Figure 73), was synthesized from morphine (1a) in five steps. The phenolic hydroxyl group of 1a was protected as PMB (p-methoxy-benzyl) ether. Next, 3-O-PMB-morphine (292) was reacted with benzoic acid under Mitsunobu conditions (DIAD, TPP, toluene) to form the benzoate ester (293). The latter compound (293) was converted into the 3-O-PMB-isomorphine (294) by saponification with 1M potassium hydroxide in aqueous ethanol solution. The 6-beta-hydroxyl function of 294 was transformed to a 6α-amino group via a second Mitsunobu reaction and subsequent phthalimide cleavage. 3-O-PMB-isomorphine (294) was reacted with phthalimide (DIAD, TPP, toluene) to give 6α-phthalimido-3-O-PMB-morphine (295). The treatment of this latter compound with hydrazine hydrate in ethanol resulted in 3-O-PMB-6-amino-6-desoxymorphine (296).

In 1998, the research group of Nagase [270] developed nalfurafine (300, TRK-820, 17-cyclopropylmethyl-3,14-dihydroxy-4,5-epoxy-6β-[N-methyl-trans-3-(3-furyl)-acrylamido]morphinan, Remith^®, Figure 74) a 4,5-epoxymorphinan-type OR ligand. TRK-820 (300) was selective full, very high affinity agonist for κ-ORs, and a partial agonist for μ-ORs and δ-ORs (K_i (κ-OR) = 75 pM, K_i (μ-OR) = 5.2 nM, K_i (δ-OR) = 161 nM, μ/κ = 69.3, δ/κ = 2146) [271]. TRK-820 (300) is an approved pharmaceutical for treatment of haemodialysis-related uremic pruritus and patients with chronic liver diseases. Nalfurafine (300) was prepared from naltrexone (297) via N-methyl-β-naltrexamide (299). The secondary amine (299) was acylated with 3-(3-furyl)acryloylchloride to yield 300. Recently, Suzuki et al. [272] synthesized nalfurafine (300) and its 10α-hydroxy analogues from 3-O-acetyl-α-naltrexol (276h) via a modified synthesis route by application of the Mitsunobu reaction.

In the last decade, 6β-acylaminomorphinans [273] and 6β-pyridinyl amidomorphinans were synthesized [274] starting from precursors arising from earlier work by the Makleit group [257].

2.12. Poppy Alkaloids as Starting Materials for Molecular Imaging

Molecular imaging by positron emission tomography (PET) or single-photon emission computed tomography (SPECT) enables the detection of one or more subtypes of ORs in healthy brain and in pathologies involving opioid neurotransmission [29,275,276,277]. OR PET radiochemistry development has benefited from the foundational work on opioid chemistry described above, which presented important lead compounds for radiolabeling. As with the lead compounds, some classes of OR PET ligands have incomplete sub-type selectivity, and differ with respect to their agonist/antagonist binding profile. Furthermore, the extreme potency of some opioid agonists calls for particular attention to the molar activity of OR agonist tracers.

OR PET imaging began with development of the μ-OR selective agonist 4-anilidopiperidine derivative carfentanil ([¹¹C]caf) [278] and the μ/κ OR selective antagonist morphinan foxy ([¹⁸F]204a, [¹⁸F]foxy) and cyclofoxy derivatives ([¹⁸F]203a, acetyl-[¹⁸F]cyclofoxy; [¹⁸F]204c, [¹⁸F]cyclofoxy ([¹⁸F]FcyF); [¹¹C]204c, [¹¹C]cyclofoxy ([¹¹C]cyF)) [197,198,199,200,201,202] (Figure 75). The synthesis of the [¹⁸F]foxy and [¹⁸F]cyclofoxy derivatives can start from the poppy alkaloid thebaine (4) [279], by conversion to 14-hydroxycodeinone and further synthetic manipulation of the C-ring to introduce the 6β-fluoro group by nucleophilic substitution [198] (see Figure 52).

Following upon the synthesis of the δ-OR selective antagonist N1′-([¹¹C]methyl)naltrindole ([¹¹C]MeNTI, (K_i (μ) = 14 nM, K_i (δ) = 0.02 nM, K_i (κ) = 65 nM,) [280] and subsequently N1’-(2-[¹⁸F]fluoroethyl)naltrindole ([¹⁸F]FE-NTI, BU97001, affinities) were prepared [281].

Researchers have developed a range of carbon-11 and fluorine-18 labelled cyclofoxy derivatives (Figure 75), namely [¹¹C]cyclofoxy ([¹¹C]cyF [201]), [¹⁸F]foxy [197], [¹⁸F]acetylcyclofoxy [199,200] and [¹⁸F]cyclofoxy ([¹⁸F]FcyF) [197,202]. Among these, the μOR/κOR antagonist [¹⁸F]FcyF has served in a study of OR-availability in methadone-treated former opioid addicts [202], revealing lower OR availability of the striatum as compared to healthy volunteers, in measure of the occupancy by methadone.

Notably, Pasternak et al. reported on the synthesis of radioiodinated iodobenzoyl derivatives of 279e, 279g, and 279h 6β-oxymorphamine (¹²⁵I-BOxyA), 6β-naloxamine (¹²⁵IBNalA), and 6β-naltrexamine (¹²⁵IBNtxA) in 2011 [282], with potential application for SPECT imaging.

6,14-Ethenomorphinans (orvinols / Bentley compounds) [50,234,283] are in ring-C bridged semisynthetic derivatives of the poppy alkaloid thebaine [279] and/or oripavine [284,285,286]. Their radiolabelled derivatives are the most frequently used tracers for OR molecular imaging, e.g., the nonselective OR antagonist ¹¹C-diprenorphine ([¹¹C]DPN, 301a, (K_i (μ) = 0.07 nM, K_i (δ) = 0.23 nM, K_i (κ) = 0.02 nM, Figure 75) [287,288], the partial muOR agonist and kappa antagonist ¹¹C-buprenorphine ([¹¹C]BPN, 301b) [288,289] (K_i (μ) = 1.5 nM, K_i (δ) = 6.1 nM, K_i (κ) = 2.5 nM, K_i (NOP) = 77.4).

Following upon development of the agonist ligand ¹¹C-phenethyl-orvinol (302a, [¹¹C]PEO) (K_i (μ) = 0.18 nM, K_i (δ) = 5.1 nM, K_i (κ) = 0.12 nM) [290], Schoultz et al. reported its ¹⁸F-fluorine labelled version, ¹⁸F-phenethyl-orvinol (302b, [¹⁸F]FE-PEO, (K_i (μ) = 0.10 nM, K_i (δ) = 0.49 nM, K_i (κ) = 0.08 nM) [291,292,293]. In 2000, Wester et al. reported the radiosynthesis of the antagonist ligand 6-O-(2-[¹⁸F]fluoroethyl)-6-O-desmethyl-diprenorphine ([¹⁸F]FE-DPN, 301c, K_i (μ) = 0.24 nM, K_i (δ) = 8.00 nM, K_i (κ) = 0.20 nM, Figure 76) [294] thorough an indirect [¹⁸F]fluoroethylation procedure. Subsequently, Schoultz et al. [295] elaborated a more efficient method for the synthesis of [¹⁸F]FE-DPN (301c) and [¹⁸F]FEOTos and they extended their method also for the radiosynthesis of 6-O-(2-[¹⁸F]-fluoroethyl)-6-O-desmethyl-buprenorphine (301d, [¹⁸F]FE-BPN, K_i (μ) = 0.24 nM, K_i (δ) = 2.10 nM, K_i (κ) = 0.12 nM) [295]. More recently, Marton et al. [296] developed a novel precursor molecule, 6-O-(2-tosyloxyethyl)-6-O-desmethyl-3-O-trityl-diprenorphine (TE-TDDPN, «Henriksen precursor»), for the one-pot, two step nucleophilic radiosynthesis of [¹⁸F]FE-DPN ([¹⁸F]301c). In 2023, the research group of Mikecz [297] optimized the radiosynthesis of [¹⁸F]FE-DPN ([¹⁸F]301c) from the new precursor. Despite its incomplete OR subtype selectivity in vitro, 6-O-(2-fluoroethyl)-6-O-desmethyl-diprenorphine (FE-DPN, 301c) preferentially binds to the μ-ORs in living brain [298].

2.13. Other Semi-Synthetic Derivatives

In the main part of the review, we presented some selected research topics from the huge body of work from more than five decades of scientific activity of the alkaloid research groups of the University Debrecen and the Alkaloida Chemical Company. In this final section, we seek to highlight some additional important studies. The investigation of Horváth and Makleit [299] is noteworthy due to its importance in the field of the morphine total synthesis. The authors developed an efficient method for the conversion of 6-O-demethyl-salutaridine to salutaridine. In the early 1990s, the Makleit research group turned with great interest towards the Bentley-compounds [176,234]. The main driving force was the finding of Lewis et al. [300] that buprenorphine, a 6,14-ethenomorphinan type partial μOR agonist and κOR antagonist, serves as an alternate to methadone for the treatment of heroin/diamorphine dependence. Makleit et al. developed numerous novel synthesis routes for the efficient preparation of buprenorphine, diprenorphine and their N¹⁷-substituted analogues. [230,301,302]. Very recently, we provided an overview of their research in the field of C-ring bridged morphinans together with the Diels-Alder adducts of morphinan-6,8-dienes and their transformations dating back almost a century [50]. We cannot omit mention of the investigations of Berényi, who established aporphine-chemistry as a new direction for the Debrecen alkaloid research group. They expended enormous efforts over the years for the synthesis of morphinan-6,8-dienes [50] and their rearrangements to potent dopaminergic aporphine derivatives [279,303]. We emphasize the efforts extending over many years by Hosztafi in methodological research for the N-demethylation of morphine alkaloids [304,305,306]. We also note the thorough investigations carried out by Seller et al. [307] for the refinement of the poppy alkaloid extraction technology with the help of modern technical equipment and separation methods.