TETRAHEDRON: ASYMMETRY REPORT NUMBER 63 Synthetic...

Tetrahedron:

Tetrahedron: Asymmetry 15 (2004) 1203–1237

Asymmetry

TETRAHEDRON: ASYMMETRY REPORT NUMBER 63

Synthetic pathways to salsolidine

Teodoro S. Kaufman*

Instituto de Qu�ımica Org�anica de S�ıntesis (CONICET-UNR) and Facultad de Ciencias Bioqu�ımicas y Farmac�euticas,Universidad Nacional de Rosario, Suipacha 531, S2002LRK Rosario, Argentina

Received 5 January 2004; accepted 6 February 2004

Dedicated to Professor Edmundo A. R�uveda on occasion of his 70th birthday

Abstract—Salsolidine is a simple 1-substituted tetrahydroisoquinoline isolated from many natural sources as the racemate and in itsenantiomeric modifications. The wide variety of synthetic strategies leading to this natural product, mainly in its optically activeforms, is discussed.� 2004 Elsevier Ltd. All rights reserved.

Contents

* Tel./fax:

0957-4166/

doi:10.101

1. Introduction 1204

2. Isolation, natural occurrence, and interaction of salsolidine with biomolecules and biological

systems 1205

3. Synthetic strategies leading to salsolidine in racemic form 1207

4. Syntheses of the salsolidines in their optically active forms 1210

4

+54

$ - s

6/j.te

4.1. Resolution of racemates and separation of diastereomeric mixtures 1210

4.2. Hydride reduction and catalytic hydrogenation of C@N and C@C double bonds 1211

4.2.1. General 1211

4.2.2. Enantioselective reduction of 3,4-dihydroisoquinolines. Chiral borohydrides 1212

4.2.3. Diastereoselective reduction of chiral activated azomethines

(chiral iminium ions) 1213

4.2.4. Enantioselective borane reduction of azomethines activated

with chiral Lewis acids 1214

4.2.5. Catalytic enantioselective reduction of 1-alkyl-3,4-dihydroisoquinolines 1216

4.2.6. Diastereoselective catalytic hydrogenation or hydride reduction

of chiral enamides 1219

4.2.7. Enantioselective reduction of enamides with chiral catalysts 1220

4.3. Metallation of tetrahydroisoquinoline derivatives 1222

4.3.1. Diastereoselective alkylation of metallated chiral tetrahydroisoquinoline

derivatives 1222

4.3.2. Enantioselective protonation of metallated tetrahydroisoquinoline derivatives 1223

.4. Alkylation of azomethines 1223
4
4.4.1. Enantioselective alkylation of azomethines in the presence of

chiral auxiliaries 1223

4.4.2. Diastereoselective alkylation of chiral activated azomethines

(chiral iminium ions) 1226

.5. Syntheses employing organochalcogen derivatives 1232

5. Conclusion 1233

Acknowledgements 1234

References and notes 1234

-341-4370477; e-mail: [email protected]

ee front matter � 2004 Elsevier Ltd. All rights reserved.tasy.2004.02.021

mail to: [email protected]

mail to: [email protected]

1204 T. S. Kaufman / Tetrahedron: Asymmetry 15 (2004) 1203–1237

1. Introduction

Isoquinolines comprise of the largest family of naturallyoccurring alkaloids, being found abundantly but notexclusively in the plant kingdom. The 1-substitutedtetrahydroisoquinolines constitute an important groupamong the isoquinolines; while many of them have beenisolated as both of their enantiomers from independentsources or just as the racemate, most of these naturalproducts have been characterized as only one enantio-mer with either the (1S)-absolute configuration orthe opposite (1R)-geometry. The construction of the1,2,3,4-tetrahydroisoquinoline ring system has been apopular area of research in natural products chemistrysince the early 1900s, with activity in this field almost asold as the discovery of the isoquinoline system itself.

Despite its simplicity, the elaboration of the isoquinolinering has been approached in various ways. In a sys-tematic classification of the methods for the synthesis ofthe isoquinoline ring system, Kametani1 described fivedifferent types, according to the mode of formation ofthe pyridine ring (Fig. 1). The first type involves the ringclosure between the carbon atom, which constitutes theC-1 position of the isoquinoline and the aromatic ring,while in the second type the heterocycle is formed be-tween C-1 and the nitrogen. Type 3 necessitates C–Nbonding between the nitrogen and the atom, whichforms the C-3 position in the resulting isoquinoline.

N N NN N

Type 1 Type 3 Type 5Type 2 Type 4

Figure 1. Classification of the synthetic strategies toward the iso-

quinoline ring system, according to Kametani.N

MeO

MeOMe

HN

MeO

MeOMe

H

1S-(-)-Salsolidine, (-)-1 1R-(+)-Salsolidine, (+)-1

1 2

345

6

78

N

MeO

MeOMe

HN

MeO

MeOMe

H

1S-(-)-Salsolidine, (-)-1 1R-(+)-Salsolidine, (+)-1

1 2

345

6

78

Figure 2.

Likewise, the fourth type involves cyclization at the C-3and C-4 level while the fifth type is meant to describe theprocess in which the isoquinoline ring is concluded bythe generation of a C–C bond between the aromaticmoiety and the C-4 position. Although the literaturerecords examples of all of the five types, synthetic pro-tocols of types 1, 2, and 5 have been demonstrated to bethe most widely used for the elaboration of tetrahy-droisoquinolines.

Among the protocols of the currently available meth-odological arsenal, some are only suitable for thesynthesis of 1-substituted tetrahydroisoquinolines asracemic compounds, while others, in spite of havingbeen tested for only one enantiomer, can provide eachone of the constituents of the enantiomeric pair of 1-substituted tetrahydroisoquinolines, in some cases withtheir absolute stereochemistry known in advance.

The enantioselective synthesis of simple isoquinolinealkaloids was pioneered by Brossi et al.,2 Kametaniet al.,3 Yamada et al.,4 and others in the 1970s andgained strong interest as a consequence of their initialbreakthroughs, while some simple alternatives for

accessing tetrahydroisoquinolines in enantiomericallypure forms, such as the resolution of racemates, can betraced back to an earlier time. Many novel strategieshave been explored with a range of approaches becom-ing useful over the past few years. Most importantly, aconsiderable group of the newly developed syntheticprotocols are general and have already found use in theelaboration of other related alkaloids and even morecomplex structures. A recent review by Rozwadowska5

shows the many different strategies devised for theenantioselective synthesis of simple 1-substituted tetra-hydroisoquinoline alkaloids.

Strikingly, as early as in 1987, Huber and Seebach6

pointed out that all possible methods of synthesizingenantiomerically pure compounds have been applied tothe elaboration of homochiral 1-substituted tetrahy-droisoquinolines, including resolution, catalytic, andstoichiometric enantioselective reactions as well as theincorporation of components from the pool of chiralbuilding blocks.

Salsolidine 1 (1-methyl-6,7-dimethoxy-1,2,3,4-tetrahy-dro-isoquinoline, CAS 493-48-1) is a simple tetrahy-droisoquinoline alkaloid (Fig. 2), which can be isolatedfrom different natural sources, mainly Cactaceae andChenopodiaceae and has been the subject of numeroustotal syntheses, both as racemate and in its homochiralforms up to the point that nowadays, the observationmade by Huber and Seebach also holds true for theenantioselective synthesis of salsolidine itself, as thisreview will make evident.

On the other hand, there are no examples recording theelaboration of salsolidine by Kametani’s type 4 synthe-sis; only one publication discloses the elaboration of thenatural product following a type 3 strategy while manyreports describe the syntheses of salsolidine in which thestarting material is already an isoquinoline derivative,built according to one of the five above mentioned types.

The variety of syntheses of salsolidine reported to date,especially those leading to the natural product in itsoptically active form, can be accounted for in manyways. First, the cardinal significance of chiral amines inthe pharmaceutical and agrochemical industries, as wellas their increasing importance among flavors and fra-gances, which have placed great demand and strongpressure for the search of new strategies for their effi-cient and cost-effective syntheses. Thus, in light of theresulting developments over the past 25 years and sincethe salsolidine enantiomers have often been used astargets for testing newly developed synthetic methods, it

T. S. Kaufman / Tetrahedron: Asymmetry 15 (2004) 1203–1237 1205

is not surprising that the enantioselective preparation ofthe salsolidines has been thoroughly studied and,therefore, experienced an enormous growth in this time.Methods for the asymmetric preparation of amines haverecently been reviewed by Johansson.7

Secondly, the continuous expansion of the arsenal ofreagents and reactions at the disposal of the modernsynthetic organic chemist, coupled to the structuralfeatures and simplicity of salsolidine, constitute anadditional reason for chemists being inclined to preparethe natural product, especially in its homochiral forms.

On the other hand, 1-methyl-6,7-dimethoxy-3,4-dihy-droisoquinoline 2, which is readily accessible by theBischler–Napieralski reaction,8a–e;9a as well as by oxi-dation of salsolidine itself (81% yield) with 3 equiv ofdiphenylselenium bis(trifluoroacetate),8f is a represen-tative cyclic imine, which has been repeatedly employedfor testing the efficiency, scope, and limitations of manynew hydride reducing agents and hydrogenation cata-lysts, with salsolidine being the expected reactionproduct of these transformations.

Moreover, the related and easily available N-acetyl en-amine 3a9a (Fig. 3) has been employed as a representa-tive enamide, having been often used as a startingmaterial for the elaboration of the natural product,while enamides 3b and 3c9b;c have also found some useas starting materials for the synthesis of 1.

N

MeO

MeOMe

N

MeO

MeOCH2

2 R= Me3b R= H3c R= OMe

O

a

R

3

Figure 3.

Additionally, there is general interest in enantiopuretetrahydroisoquinoline systems, since many isoquinolinealkaloids exhibit valuable physiological activities orunique structures.10 Such a situation has continuouslyattracted the attention of the synthetic organic chemists.

Finally, multiple syntheses of salsolidine can also beconsidered in part by the use of salsolidine itself andother related simple tetrahydroisoquinolines as startingmaterials for the elaboration of bioactive substances11a–c

and in biological studies,11d mostly associated to thepathogenesis of Parkinson’s disease and other patho-genic processes of the central nervous system.

Most of the enantioselective syntheses of salsolidinereported to date are based on more or less subtle butgenerally highly interesting modifications of a handful ofdifferent basic strategic schemes, including the resolutionof diastereomeric mixtures, stoichiometric chiralitytransfer through hydride reduction of optically activea-alkylbenzylamine derivatives and structurally related

compounds (chiral iminium ions), diastereoselectivehydrogenation of chiral enamides, enantioselectivehydrogenation of imines and enamides with chiral cat-alysts, catalytic asymmetric reduction of imines, diaste-reoselective addition of organometallic reagents tochiral iminium compounds, addition of organometallicsto iminium compounds in the presence of chiral ligands,and the addition of chiral organometallic reagents toiminium derivatives, as shown in Scheme 1. Otherstrategies include the intramolecular addition of aminesto chiral vinyl sulfoxides, the asymmetric Pictet–Spen-gler reaction, and the reduction of 1-alkyl-3,4-dihydro-isoquinolines by chiral sodium(triacyloxy)borohydrides.

The aim of this review is to consider the literature andmost relevant advances achieved over the past 25 yearsconcerned with the synthesis of salsolidine, emphasizingthe enantioselective preparation of the natural product.Accordingly, synthetic strategies are divided into pro-tocols leading to racemic salsolidine and those furnish-ing this product in its homochiral form. Among thelatter, the methods covered are divided into five maincategories: (1) resolution of racemates and separation ofdiastereomeric mixtures, (2) reduction of C@N andC@C double bonds, (3) metallation of isoquinolinederivatives with or without formation of new C–Cbonds, (4) alkylation of azomethines, and (5) the use oforganochalcogen derivatives. A brief comment on theoccurrence of salsolidine in nature, as well as a summaryof the reported effects of salsolidine, mainly on enzy-matic systems are also reported herein.

2. Isolation, natural occurrence, and interaction ofsalsolidine with biomolecules and biological systems

Salsolidine was first isolated by Proskurnina and Or�e-khov from Salsola richteri (Chenopodiaceae) as thelevorotatory enantiomer (�)-1 [(S)-salsolidine];12 acouple of years later, the same group reported theoccurrence of this natural product in Salsola arbuscula.13

The originally proposed structure for the natural prod-uct was confirmed by chemical tests and a series ofspecific rotation measurements and mixed melting pointdeterminations with semi-synthetic (S)-salsolidine, pre-pared via (þ)-tartaric acid-assisted resolution of racemicsalsoline 4, followed by methylation of the free levoro-tatory base with ethereal diazomethane. Additionalevidence of identity was gained from mixed meltingpoint measurements of the hydrochloride, picrate andpicrolonate of both, natural and semi-synthetic (�)-1,which showed no melting point depression.

Both enantiomers of the natural product, as well as theracemate, have been isolated from natural sources.14

Thus, (R)-(þ)-salsolidine was found in the leaf, fruit,and stem of Genista pungens (Leguminosae),15 while theracemate (or the natural product without disclosure ofthe characteristics of its stereogenic center) wasfound, among others, in Carnegiea gigantea (the firstreport of its occurrence in cacti)16 and Pachicereus

N

MeO

MeOMe

H*N

MeO

MeO R*

N

MeO

MeOMe

N

MeO

MeOCH3

R

N

MeO

MeO

N+

MeO

MeOMe

R*

N

MeO

MeOCH2

R*

N

MeO

MeOCH2

R

N

MeO

MeOMe

(H, R*)

Resolution

R*M or RM/L*

Na(OAc)3BH

L*M/H2

BuLi/MeI

R'M/L*

H2,catalyst

L*M/H2LA*, BH3

NaBR*3H

NaBH4

Scheme 1.


pectenaboriginum (Cactaceae) from Mexico and USA,16

Bienertia cycloptera,17 Corispermum leptopyrum,18

Salsola kali,19 S. soda,20 and S. ruthenica20 from Poland,Hamada articulata ssp. Scoparia,21 Haloxylon articula-tum,22 and H. scoparium (Chenopodiaceae), as well as inthe stem of Alhagi pseudalhagi from India,23 and Cal-ycotome spinosa,24 Desmodium cephalotes,25 and D. tili-aefolium15 among the Leguminosae. The recentdetection of trace amounts of salsolidine and othersimple tetrahydroisoquinolines in Neobuxbaumia mul-tiareolata, N. scoparia, and N. tetetzo (Cactaceae)employing GC–MS techniques is a finding with highchemotaxonomic significance; it suggests that genusCarnegiea is different from Neobuxbaumia and is prob-ably derived from it.26

Relevant physical properties and spectral data of thenatural product in racemic and homochiral formshave been recently compiled by Shamma et al.27a ItsNMR spectra were interpreted27a–c as well as its massspectrum27d and other relevant data being disclosed.27e

Additionally, its chiroptical properties were examined aspart of the development of a new semiempirical quad-rant rule based on one-electron theory, useful forpredicting the configuration of 1-methyl tetrahydroiso-quinolines.28 Finally, X-ray diffraction data of thehydrochloride hydrate as well as the hydrochloridedihydrate have also been published.29

Experiments using tissue-cultured cells from Calluspallida var. tenuis and Callus incisa fed with (�)-[1-D

and 1-methyl-CD3]-salsolinol (salsolinol-D4), allowedthe HPLC, mass spectra, and 1H NMR detection ofminor amounts of salsolidine among the resultingmetabolites; it was also shown that (�)-salsolinol 5metabolized to produce salsolidine, among other relatedtetrahydroisoquinolines, in the tissue-cultured cells ofCorydalis ochotensis var. raddeana, Corydalis ophio-carpa, and Macleaya cordata R. Br. Moreover, it wasfound that (�)-5 metabolized in live whole plants withresults similar to those observed in tissue-culturedcells.30a

Furthermore, by using an LC/API-MS system, it wasshown that 3,4-dihydroxy-phenethylamine (dopamine)condenses with acetaldehyde to give salsolinol 5, whichis further metabolized to produce 6-O-methylsalsolinol6a (isosalsoline), which in turn, is O- and N-methylatedto provide salsolidine 1 and N-methylisosalsoline 6b,respectively, in several plant tissue cultures of Papaver-aceae (Fig. 4).30b

The earliest preparations of 1 in homochiral form weremade by the chemical manipulation of products ac-cessed by classical Bischler–Napieralski8a–e as well asPictet–Spengler31 condensations, which furnished theracemate, followed by resolution with (þ)-tartaricacid.12 It was Battersby and Edwards in the early 1960swho unequivocally determined that natural (�)-salsol-idine had an (S)-configuration, by its degradation to2-carboxyethyl-LL-alanine and comparison with thisaminoacid derivative.32

N

MeH

HO

MeON

MeH

HO

HO

N

MeMe

MeO

MeON

MeR

MeO

HO

4 5

S-(-)-7R= H, R-(+)-6aR= Me, R-(+)-6b

Figure 4.


Some efforts toward the measurement of salsolidine asan analyte have already been published. Interestingly, aspectrophotometric assay of 1 based on its reaction withammoniacal copper sulfate in carbon disulfide, leadingto the production of a copper–dithiocarbamate complexsoluble in benzene, has been reported. The complex canbe extracted from the reaction medium and measured atits absorbance maximum of 448 nm.33a An alternativespectrophotometric procedure for the quantification of 1and 4 was disclosed earlier by Russian authors33b withthe fluorometric determination of salsolidine as animpurity in the preparations of 4 also being reported.33c

Interestingly, the HPLC separations of the enantiomersof salsolidine as N-a-naphthoyl (a ¼ 1:21)34a and N-b-naphthoyl (a ¼ 1:83)34b derivatives, employing 3,5-dini-trobenzoyl phenylglycine and (S,S)-[N-(3,5-dinitro-benzoyl)]-amino-3-methyl-1,2,3,4-tetrahydrophenanth-rene-based chiral stationary phases, respectively, wererecently reported. However, the chiral HPLC with b-cyclodextrin as chiral selector was unable to separatesalsolidine enantiomers as previously reported byStammel et al.34c Additionally, the production of highlyspecific antibodies, with a high affinity against (�)-sal-solidine (K ¼ 1:5� 109 M�1) by immunization of rabbitswith bovine serum albumin–salsolidine conjugates, wasdemonstrated.34d These antibodies were employed todevelop a radioimmunoassay for the alkaloid and thusmay find use in the determination of trace amounts oftetrahydroisoquinolines in organic tissues and biologicalfluids.

Several biological activities have been ascribed to (orinvolved) the natural product. A study by Oxenkrugdemonstrated that 1 and other alkaloids such as 4 (Fig.4) and papaverine, inhibit the uptake of 5-hydroxytryptamine by human blood platelets,35 showing a 30–60% inhibition at a concentration of 10�4 M. The ste-reospecific N-methylation of (S)-1 [to (S)-(�)-carnegine,7] and of the endogenous alkaloid (R)-isosalsoline (R)-6a by the enzyme amine N-methyl transferase frombovine liver has been shown by Bahnmaier et al. Thisgroup found the enantiomers (S)-1 and (R)-6a to bethose preferentially methylated.36 These authors alsorevealed that certain tetrahydroisoquinolines are capa-ble of mimicking direct but not receptor-mediatedinhibitory effects of estrogens and phytoestrogens on

testicular endocrine function; salsolidine, however,was ineffective in this assay.37 Moreover, Brossi et al.studied the effect of simple isoquinoline alkaloids onmonoamine oxidase (MAO) inhibition, finding that 3,4-dihydroisoquinolines were more potent than isoquino-lines and their 1,2,3,4-tetrahydro- congeners in MAO Ainhibition, while only a few heterocycles inhibited MAOB. Among the dihydro- and tetrahydroisoquinolines, astereoselective competitive inhibition of MAO A causedby the (R)-enantiomers was observed. These results weresupportive of the view that the topographies of MAO Aand MAO B inhibitor binding sites are different.38

Salsolidine was also reported to inhibit competitively themethylation of the catecholamine metabolite 3,4-di-hydroxybenzoic acid by an enriched rat liver prepara-tion of the enzyme catechol-O-methyltransferase, withKi ¼ 0:19mM.39 Dostert et al.40 reviewed the role ofdopamine-derived alkaloids in alcoholism and Hun-tington’s disease and the group of Airaksinen, afterstudying the binding of b-carbolines and tetrahydro-isoquinolines to opiate receptors of the d-type, con-cluded that tetrahydroisoquinolines, like salsolidine,with Ki values higher than 100 lM, were less potent thanb-carbolines and that opiate receptors do not appear tobe the major sites of action of simple tetrahydroiso-quinolines.41a The neurotoxicity of several tetrahydro-isoquinoline derivatives against SH-SY5Y cells has alsorecently been studied; salsolidine proved to be lessneurotoxic than the dihydroxylated compound 5, con-trasting with the tendency observed among the mono-substituted 1-methyl tetrahydroisoquinolines.41b On theother hand, a trimethoprim analogue made by theincorporation of a 2,4-diaminopyrimidine unit tothe salsolidine nucleus provided an active dihydrofolatereductase inhibitor41c and phenylpropionylamido-diph-enylalkyl-tetrahydroisoquinolines, including the salsoli-dine moiety, have been reported as luteinizing hormonereleasing hormone (LHRH) antagonists.41d Dibenzo-18-crown-6 derivatives of salsolidine, prepared by conden-sation of the crown acid dichloride with the naturalproduct, have also been described.41e

3. Synthetic strategies leading to salsolidine inracemic form

Racemic salsolidine has been elaborated in several ways,including the reduction of isoquinolines and their ben-zopyrylium precursors, metallation–alkylation of tetra-hydroisoquinoline derivatives and modifications of theclassical Bischler–Napieralski, Pictet–Spengler, andPomeranz–Fritsch isoquinoline syntheses.

Despite that racemic salsolidine (�)-1 being the result ofmany published syntheses, some of these can be modi-fied in order to become synthetic protocols suitable forthe elaboration of salsolidine in optically active form,transforming this into a highly attractive test field fornovel synthetic methodology.

Contemporary with the Minter synthesis of N-carb-oxymethyl salsolidine 8 from 1-methyl-6,7-dimethoxy

N

MeCO2Me

MeO

MeON

Me

MeO

MeO

N

CNCO2Et

MeO

MeON

H

MeO

MeO

8

10 11OH

9

Figure 5.

O+

MeO

MeOMe

N+

MeO

MeOMe

Bn

N

MeO

MeOMe

Bn

1. BnNH22. HClO4

NaBH4

H2, Pd/C

(±)-1

12 13

14

N

MeO

MeOMe

H

ClO4-

Scheme 2.


isoquinoline 9 by DIBAL-H mediated reduction of itsisoquinoline–borane complex and subsequent reactionwith methyl chloroformate (Fig. 5),42 Rozwadowskaand Br�ozda described the synthesis of (�)-1 by inter-mediacy of 6,7-dimethoxy-2-ethoxycarbonyl-1,2-di-hydroisoquinaldonitrile 10, a Reissert compound,43

which proved to be a convenient starting material forthe elaboration of related 1-substituted tetrahydroiso-quinoline natural products, such as the ubiquitous caly-cotomine 11 and carnegine 7.

The reaction of 2-benzopyrylium salts with amines hasbeen reported by Russian scientists as an alternativeapproach to (�)-1. In their sequence, submission of6,7-dimethoxy-1-methyl-2-benzopyrylium salt 12 to areaction with benzylamine formed a mixture of thecorresponding N-benzyl isoquinolinium salt 13, isolatedas the perchlorate, and the related naphthylamine(Scheme 2); reduction of the former with sodium boro-

HN

MeO

MeOMe

S N+

MeO

MeOS

MeO

N

MeO

MeOCH2 O

Me

MeO

MeOM

N

MeO

MeOO

t-Bu

MeO

MeOM

15 17

Raney Nickel

BrCH2COCl

3a 1

20 21a R=21b R

No

(98%)

(71%)

(96%, de

1. t-BuLi, TMEDA, THF, -40ºC

2. MeI

(99%)

Scheme 3.

hydride to N-benzylsalsolidine 14,44a followed by hy-drogenolytic cleavage of the N-benzyl group affordedthe natural product, albeit in a moderate overall yield.44b

As an example of the usefulness of thio N-acyliminiumions as electrophilic partners in cationic p-cyclizations,Padwa et al.45 have recently shown that the reactiveN-acyliminium ions, formed from thioamides like 15and bromoacetyl chloride, can react with activated p-nucleophiles in the form of an appropriately substitutedaromatic ring tethered on the nitrogen of a thioamide.Upon cyclization, the resulting N,S-acetals can be fur-ther manipulated leading to different products; amongthem 1-substituted tetrahydroisoquinolines.

A Kametani type 1 formal synthesis of salsolidine,shown in Scheme 3, was employed to demonstrate this

N

MeO

MeOR

O

Me

N

eH

N

MeO

MeOMe O

N

e

N

MeO

MeOO

Ot-BuR

O

16 R= S18 R= S-O

Raney Nickel

H2O2

19NaOH, EtOH, reflux, 5 days

t-BuO= t-Bu

22

aAlH4, THF (76%)r TFA

= 25%)

(55%)

(95%)

10b

1. t-BuLi, TMEDA,2. MeI

(70%)

N

MeO

MeO H

N

MeO

MeO

N

OMe

NMe2

PhMe, 110ºC, 36 h(99%)

N

MeO

N

MeO

MeON

MeO

Li

N

MeO

MeON

MeO

Me

N

MeO

MeO HMe

NO2

OMe

23

24 25

1. H2-Pd2. Me2NCH(OMe)2

(90%)

2627

28 (±)-1

n-BuLi-100ºC

MeI (>72%)

N2H4(84%)

Scheme 4.


principle. To this end, thioamide 15 was converted intothe related N,S-acetal 16 by way of an iminium ionderivative 17. In turn, 16 was transformed into theknown N-acetyl enamine 3a with Raney nickel, in 71%yield. As mentioned above, this enamide has beenrepeatedly employed as a key precursor of salsolidine,including the enantioselective synthesis of the naturalproduct.46

Recently, a Polish team led by Rozwadowska et al.47

disclosed a variant of Padwa�s group strategy in one oftheir syntheses of (�)-salsolidine, using the thiazo-lino[2,3-a]isoquinoline (S)-oxide 18, easily prepared inhigh yield and 25% de (in favor of the anti-diastereomerwith an a-oriented sulfoxide oxygen and a b-position forthe 10b methyl substituent) by a 30% hydrogen perox-ide-mediated oxidation of 16, as intermediate. Raneynickel treatment of sulfoxide 18, furnished 55% of N-acetyl salsolidine 19 instead of the related enamide 3a,which afforded the natural product in an almost quan-titative yield upon basic hydrolysis.

As alternative strategies, the elaboration of salsolidinederivatives by alkylation of a-sulfoxide carbanions hasalso been reported47 and Indian researchers havedescribed an AcOH-promoted Pictet–Spengler typesynthesis of (�)-1 in 60% yield from 3,4-dimethoxy-phenethylamine, employing a perhydro-oxazine as thecarbonyl equivalent.48 A different approach, consistingof a heteroatom-facilitated metallation process followedby alkylation, was taken by Coppola49a and more re-cently by Simpkins et al.49b In the former case, N-Boc-6,7-dimethoxy-1,2,3,4-tetrahydroisoquinoline 22 waslithiated with the t-BuLi–TMEDA complex, leadingto N-Boc salsolidine 21a (yield �70%), which affordedthe natural product upon a facile TFA-mediateddeprotection.

Analogously, in the latter synthesis racemic N-pivaloylsalsolidine 21b was obtained as part of a sequenceleading to (�)-1, by MeI alkylation of the 1-lithioderivative of N-pivaloyl-6,7-dimethoxy-tetrahydroiso-quinoline 20. The pivaloyl protecting group was reduc-tively removed in high yield, by treatment with sodiumaluminum hydride.

In a series of elegant studies, Meyers et al. experimentedduring the 1980s and the beginning of the 1990s with thealkylation of a-amino carbanions derived from form-amidines, the repeated formation of salsolidine, amongother relevant targets.50 Thus, transformamidination oftetrahydroisoquinoline 23 with formamidine 25, readilyaccessible from 2-methoxymethyl nitrobenzene 24, fur-nished almost quantitative yield of formamidine 26.After metallation with n-BuLi at �100 �C and alkylationwith MeI, 26 afforded the methylated derivative 28,probably through intermediacy of lithiated species 27, inover 72% yield; final hydrazinolysis of 28 provided 84%of racemic salsolidine, as shown in Scheme 4.50a Thechiral version of their developments in the area ofdiastereoselective carbon–carbon bond forming strate-gies employing chiral formamidines, is discussedbelow.

As an example of their proposed modification of theoriginal Pomerantz–Fritsch isoquinoline synthesis lead-ing to tetrahydroisoquinolines, Bobbitt et al.51 describeda preparation of (�)-1 starting from veratraldehyde 29,involving methyllithium addition to the intermediateSchiff base 30 formed by condensation of 29 withaminoacetal, to furnish 31 (Scheme 5). Salsolidine wasisolated in good yield as its hydrochloride salt after Pd/C-mediated hydrogenolysis of the epimeric tetrahydro-isoquinolin-4-ols 32, the overall sequence involving aKametani type 5 synthesis.

The Pummerer sulfoxide rearrangement, coupled to anelectrophilic aromatic cyclization reaction, devised byJapanese scientists, constitute the key steps of anotherexample of a Kametani type 5 tetrahydroisoquinolinesynthetic strategy. The general protocol has beendeveloped only recently, as a special case of the Bobbittstrategy and thus has been explored not as extensivelyas other classical cyclizations leading to tetrahydro-isoquinoline derivatives. Initial work was done byTakano et al. and Sano et al.,52 inspired in other sulf-oxide-mediated electrophilic reactions.53

For the synthesis of (�)-1, the required N-acylsulfoxide35 was prepared (Scheme 6) in 87% overall yield from3,4-dimethoxy acetophenone 33. Thus, reductive

MeO

MeOMe

O

MeO

MeOMe

N

SPh

H

MeO

MeOMe

N

SPh

CHO

O

MeO

MeOMe

N

+SHPh

CHO

MeO

MeOMe

N

SPh

CHOMe

NH

R(±)-1

1. PhS(CH2)2NH2,(i-PrO)4Ti

2. NaBH4

1. HCOOH, Ac2O2. NaIO4, MeOH/H2O

TFAA,PhH

1. NiCl2,NaBH4

2. NaOH

33 34

35

36 37 R= H38 R= Me

(96%)

(80%)

(87%from 33)

Scheme 6.

CHO

MeO

MeO

MeO

MeON

OEt

OEt

MeO

MeON

OEt

OEt

MeH

MeO

MeON

MeH

OH

PhH, ∆(95%)

NH2OEt

OEt

MeMgBr,Et2O

29 30

31

4N HCl14-18 h

5% Pd/C

(66%)

(±)-1.HCl32

MeO

MeON

MeHHCl

Scheme 5.


amination of the latter with 2-phenylthioethylamineunder titanium isopropoxide promotion54 in an EtOH–AcOH medium furnished 34, which in turn was acylatedwith mixed formyl–acetyl anhydride and then oxidized tothe diastereomeric mixture of sulfoxides 35 with NaIO4in aqueous MeOH. Treatment of 35 with TFAA inbenzene at room temperature for 18 h, gave 96% of the

cyclized product 36, which was reductively desulfurizedin 88% yield with the sodium borohydride–nickel chlo-ride reagent and then subsequently deformylated to thenatural product in 91% yield by alkaline hydrolysis.55

In a systematic study,56 it was demonstrated that thistype of cyclization is sensitive to the solvent and thenature of the N-acyl substituent. In CH2Cl2, a complexmixture of products was obtained with the formylmoiety as an N-protecting group apparently playing animportant role in facilitating the intramolecular cycli-zation to take place. Despite that only the elaboration ofracemic salsolidine has been reported following thisroute to date, the strategy has been adapted to largescale preparation of both enantiomers of 1-methyl tet-rahydroisoquinoline 3757a as well as the four stereo-isomers of 1,3-dimethyl-1,2,3,4-tetrahydroisoquinoline38,57b which suggests its potential applicability to anenantioselective synthesis of salsolidine. This routeseems to be very attractive for preparing substrates foruse in biological studies, since isotope labeling at the C-4position is possible by reductive elimination of thephenylthio group.

Finally, the clean indium metal-mediated reduction of3,4-dihydroisoquinoline 2 to salsolidine in a quantitativeyield has recently been reported by Moody et al.58a

Contrary to the outcome of related heterocyclic ringsystems when zinc metal was employed as reducingagent, no evidence of any dimeric products formed bycoupling of the heterocyclic rings was found by theseauthors.58b

4. Syntheses of the salsolidines in their opticallyactive forms

Stereoselective synthesis is one of the most relevantadvances in synthetic organic chemistry over the last fewdecades. In addition to the resolution of racematesthrough the formation of diastereomeric salts with chiralacids and the separation of diastereomeric mixturesprepared with the aid of chiral derivatizing agents, ac-cess to salsolidines in their homochiral forms has beenrepeatedly achieved by diastereo- and enantioselectivesyntheses, exploiting strategies such as C@N and C@Cdouble bond reduction (hydrogenation, hydrosilylation,and hydride addition), C@N double bond alkylation,diastereoselective alkylation of chiral tetrahydroiso-quinoline derivatives, and enantioselective protonation.Other alternatives involving enantioselective C@Oreduction and chiral aminoselenenylation have alsosuccessfully yielded chiral modifications of the naturalproduct.

4.1. Resolution of racemates and separation of diaste-reomeric mixtures

The resolution of racemic salsolidine or related simpletetrahydroisoquinolines by the formation of tartratesalts, was the first strategy aimed at the acquisition ofthe natural product in optically active form.12

N

MeH

MeO

MeO

N

MeH

MeO

MeO

N

MeH

MeO

MeO(R) N

Me

MeO

MeONH

(R)

O Me

NaOBu,BuOH

NaOBu,BuOH

OCN (R) Ph

Crystallizationi-Pr2O-CH2Cl2

(5:1)

41, Non-crystalline (44%)

40, Crystalline (46%)(-)-1

(+)-1

39

(±)-1

(74%)(S) N

Me

MeO

MeONH

(R)

O Me

Scheme 7.

N

R1

RN

R1

RSiHPh2

H+

Ph2SiH2 *

Cat.


A modern example of the resolution of (�)-1 by theformation of diastereomeric mixtures was provided byBrossi et al.59 Their strategy was based on the prepara-tion, separation, and decomposition of diastereomericureas by a reaction of (�)-1 with an optically activeisocyanate.

Reaction of (�)-1, prepared by the Bischler–Napieralskiroute involving the related 3,4-dihydroisoquinoline 2,8;9

with (R-(þ)-a-phenethylisocyanate 39, furnished amixture of diastereomeric products (Scheme 7), whichonce submitted to crystallization in an isopropyl ether–methylene chloride (5:1) solvent mixture, afforded thecrystalline urea 40, while its diastereomer 41 remained inthe mother liquors; basic hydrolysis of the purifieddiastereomers with 2M BuONa in BuOH, providedboth enantiomers of salsolidine in around 35% overallyields each, and enantiomeric excesses comparable tothose obtained by the tartaric acid resolution procedure.

The same group prepared diastereomeric p-nitrophenylurea derivatives of (�)-1 employing the known (R)-[1-(p-nitro-phenyl)ethyl]amine as the derivatizing agent;60a

these diastereomeric compounds were easily separatedby HPLC. Interestingly, the same principle was alsoemployed in the opposite sense; Russian scientists re-ported the use of (S)-salsolidine as a reagent for theconfigurational determination of isocyanates such as a-phenethylisocyanate, by kinetic resolution. This strategycan also be applied to alcohols with stereogenic cen-ters.60b

N

R1

RH

PRh

P

Solvent (PhH)ClO

O

Ph

Ph

Ph

Ph

H

H *Cat.=

Scheme 8.

4.2. Hydride reduction and catalytic hydrogenation ofC@N and C@C double bonds

4.2.1. General. Being complementary to the C@O reduc-tion, the asymmetric reduction of heterotopic C@N

double bonds to diastereo- or enantio-enriched aminefunctionalities is an important approach for accessingchiral amines, as is the enantioselective reduction of theC@C double bond of enamine derivatives.

Hydride reduction of the C@N bond of 3,4-dihydro-isoquinolines and 3,4-dihydroisoquinolinium derivativesconstitutes one of the first strategies, which have beenevaluated for the enantioselective elaboration of chiral1-substituted tetrahydroisoquinolines. As it has madestriking progress over the last two decades, driven by theubiquity of amine groups in natural products, pharma-ceuticals, herbicides, and other bioactive materials, itstill remains an actively explored strategy and an activearea of research. This topic has recently been re-viewed.61a–c

Nowadays being a core technology, catalytic enantio-selective processes have also been explored.61d In thefield of isoquinoline alkaloids, several examples of cat-alytic enantioselective syntheses have appeared in theliterature by which a large quantity of nonracemiccompounds can be secured using small amounts of achiral catalyst. The performances of the catalysts basedon Ti, Ru, Rh, and Ir were tested through enantiose-lective syntheses of salsolidine. Most of them were usedto carry out C@N hydrogenations, while others such asthose based on Rh and Ru have been employed for thesomewhat less explored enantioselective reduction of theC@C bond of enamides.

Although less successful, other processes have also beeninvestigated for the enantioselective reduction of C@Ndouble bonds. Cho and Chun were the first to report theasymmetric reduction of N-substituted ketimine deriv-atives with stoichiometric amounts of chiral oxazabor-olidines and borane; however, their procedure proved tobe effective only for aromatic ketimines, since N-substituted alkyl ketimines gave poor results (9% ee with2-butanone-phenylimine).62

Furthermore (although it has not been applied to theelaboration of salsolidine itself) Kagan provided the firstexample toward catalytic enantioselective synthesis of 1-substituted tetrahydroisoquinolines by assembling anenantioselective hydrosilylation of 1-alkyl-3,4-dihydro-isoquinolines catalyzed by a DIOP–Rh(I) complex(Scheme 8). The transformation, however, proceeded atbest with a modest enantioselectivity of 39%.63


Interestingly however, Kutney et al. disclosed that1-alkyl-3,4-dihydroisoquinolines were nonreducible with(R)-(þ)-Cycphos-Rh, a reagent which proved useful as acatalyst for the asymmetric hydrogenation of prochiralnoncyclic Schiff bases, furnishing amines with up to 91%ee (enantiomeric excess).64

N

Me

MeO

MeO

N

MeO

MeO B ON

O

Me H

O

OR H

ZnClCl

N

Me

MeO

MeO H

(-)-1

N

Me

MeO

MeO H

(+)-1

2

Re-face attack (favored)

N

OO

HOR H O

42a R= Me42b R= Bn42c R= CH2CHMe2

MeO

MeOZn

ON

O

Me

O

OR HCl Cl

NH

B

Si-face attack (disfavored)

Scheme 9.

4.2.2. Enantioselective reduction of 3,4-dihydroisoquino-lines. Chiral borohydrides. While in the seventies theasymmetric reduction of prochiral ketones with chirallymodified metal hydrides was capable of deliveringoptically active alcohols with reasonable enantiomericexcess, the asymmetric reduction of cyclic imines stillremained unsuccessful.65

In 1971, Grundon et al. reported that the reduction of3,4-dihydropapaverine with lithium hydro(methyl)-dipinan-3-a-ylborate followed by a reaction with methyliodide provided (�)-laudanosine methiodide in 8.9%ee.66 Sometime later and based on the reports of Gribbleand Ferguson, Brown and Rao, Liberatore and Morr-acci,67a–c on the synthesis and isolation of triacyloxy-borohydrides, Iwakuma et al. disclosed the asymmetricreduction of cyclic imines with isolated chiral sodiumtriacyloxy borohydrides, easily obtained from the reac-tion of NaBH4 and 3 equiv of N-acyl derivatives ofoptically active a-aminoacids.68 These authors improvedthe enantiomeric excess of chiral laudanosine, reportingthe synthesis of the (þ)-enantiomer in 71% ee with thehelp of an N-benzyloxycarbonyl-proline derived catalystin methylene chloride at room temperature. Resorting tothe same strategy, (S)-salsolidine was acquired in 85%yield and 70% ee, with the chiral auxiliary being recov-ered nearly quantitatively.

By analogy with earlier observations of Reetz69 alongwith results of the analysis of the reaction products be-tween picoline and trifluoroacetoxyborohydrides, theformation of an intermediate borane–amine complexupon reaction of the reductant with the starting 3,4-di-hydroisoquinoline 2 was postulated as a first step forthis transformation. In this proposal, products aregenerated in a second stage by inter- or intra-molecularhydride reduction of the imine.

Evidence favoring this proposed mechanism was gainedfrom the reaction of trifluoroacetoxyborohydride with1-methyl-6,7-dimethoxy-3,4-dihydroisoquinoline 2 inTHF at 0 �C, which afforded 80% yield of the imine–

Table 1. Reduction of 2 with chiral triacyloxyborohydride 42c, derived from

Entry no Solvent Time (h) Ad

1 THF 5 ––

2 THF 8 ––

3 Et2O 12 ––

4 CH2Cl2 144 ––

5 DME 123 ––

6 ClCH2CH2Cl 40 ––

7 THF 8 Zn

8 –– 1 Al

borane complex. Upon reflux in THF for 48 h, thisproduced racemic salsolidine in only 30% yield.70

A recent and highly improved variation of this strategywas contributed by Hajipour and Hantehzadeh.71 TheseIranian authors prepared bulky triacyloxyborohydridesderived from N,N-phthaloyl aminoacids 42a–c,72 whichinitially provided up to 78% of salsolidine in only 71% eestarting from dihydroisoquinoline 2, when the reactionwas run in THF at room temperature (Table 1).

Important solvent and additive effects were observed(entries 1–7), with the transformations run in THF beingthe quickest and more selective ones; moreover, in thepresence of ZnCl2, the selectivity of the catalyst showeda slight improvement, leading to salsolidine in 70% yieldand 79% ee. Interestingly, however, when this asym-metric reduction was carried out with the reagentsimpregnated on alumina, under solid state conditions(entry 8), 90% salsolidine was produced, exhibiting aremarkable 100% ee. In all reductions, the (S)-enantio-mer was predominant; Scheme 9 provides a mechanistic

N,N-phthaloyl valine

ditive Yield (%) Ee (%)

78 71

79 75

45 52

50 56

55 58

65 62

Cl2 (1.2 equiv) 72 80

2O3 (solid state) 90 100

MeO MeO RNH2

X Me


explanation accounting for the enhanced selectivity ob-served when the transformation was carried out in thepresence of ZnCl2.

COR

MeO MeON Ar

HH

Me

MeO

MeON Ar

HMe

Me

O

MeO

MeON+ Ar

HMe

Me

Cl2PO2-

MeO

MeON Ar

HMe

Me

MeO

MeON

MeH

Ac2OTEA

44a X=Y= H44b X= H, Y= Cl44c X=Y= Cl

Y

43a R= H43b R= OH

45a-c R= O46a-c R= H, H

47a-c

POCl3, PhH90ºC

NaBH4, EtOH,-78ºC(97-99%)

Pd/C, EtOH,10% HCl

48a-c

49a-c (-)-1

(97-99%)

(83-90%)

BH3.THF,BF3.Et2O

(97-99%)

Scheme 10.

4.2.3. Diastereoselective reduction of chiral activatedazomethines (chiral iminium ions). The stereoselectivenucleophilic addition of hydride ion to the endocyclicC@N bond of chiral iminium ions derived from opti-cally active amines constitutes a powerful tool for theelaboration of saturated nitrogen heterocycles with highdiastereomeric excess. The mechanism accounting forthat success was most probably the stereocontrolledformation of an iminium ion–borohydride ion pair priorto reduction, followed by a diastereoselective attack of ahydride ion to the activated iminium moiety, with con-comitant generation of a new stereogenic center. Facileremoval of chiral auxiliaries was critical for the successof this strategy, leading to products with excellent levelsof enantiomeric excess.

The numerous stereoselective syntheses of enantio-enriched salsolidine and related alkaloids accomplishedover the past 25 years with the aid of hydride reductionof optically active iminium ions, attest to how thisstrategy has also evolved into a chirally efficient process,making use of various chiral auxiliaries and reagents.The use of such a strategy can be traced back to a seriesof studies on the syntheses of heterocyclic compoundsperformed by Kametani et al.

This Japanese group discovered that when N-alkyl-3,4-dihydroisoquinolinium compounds, prepared from3,4-dimethoxyphenyl acetaldehyde 43a by the POCl3-assisted Bischler–Napieralski closure of their N-acyl-b-phenethylamine derivatives, were reduced with NaBH4at 0 �C or hydrogenolyzed over 10% palladiumhydroxide on charcoal, the corresponding tetrahydro-isoquinolines were obtained in moderate yields;73 mostinterestingly, however, was the discovery that whenthe N-alkyl groups were chiral, diastereomeric mixturesof optically active compounds were produced. In thisway, after removal of the chiral auxiliary, (R)- and (S)-salsolidine were accessed in 15–44% ee employing chiraldihydroisoquinolinium precursors 48 derived from (R)-(þ)-a-phenethylamine, (S)-(�)-a-phenethylamine 44a,(S)-(�)-a-ethylbenzylamine, and (S)-(�)-1-(1-naphthyl)-ethylamine, being the products arising from the chirala-methylbenzylamines those obtained in better enan-tiomeric excess (36–44% ee). In spite of the excellentchemical yields attained, the performance of the overallsequence in terms of diastereoselection, however, wasrather poor.

Sometime later, while studying the conformationalpreference of these iminium ions whose asymmetryoriginates from a stereogenic center appended to thenitrogen atom of the iminium ion moiety, Polniaszekand McKee74a described an interesting improvement toKametani’s original procedure (Scheme 10). Carryingout the reduction of the 3,4-dihydroisoquinolinium saltderived from 44a with sodium borohydride at �78 �C,(1S)-49a was obtained in 77% yield and 86% de. To

complete the synthesis, the N-benzyl moiety was effi-ciently removed by palladium on carbon-mediated hy-drogenolysis.

In a further improvement, Polniaszek prepared ‘secondgeneration’ chiral amines such as (S)-(�)-1-(2-chloro-phenyl)ethylamine 44b and (S)-(�)-1-(2,6-dichloro-phenyl) ethylamine 44c via directed metallation of silylderivatives of commercially available (S)-(�)-a-phen-ethylamine.75 For comparative purposes, the chiralinductors 44a–c, which possess enhanced steric differ-ences between the aryl and methyl groups, were incor-porated into dihydroisoquinolinium type iminium ions48a–c.

To that end, a carbonyldiimidazole-mediated reaction of44a–c with 3,4-dimethoxy phenylacetic acid 43b wasfollowed by a reduction of the resulting amides 45a–c tothe corresponding amines 46a–c with a BH3ÆTHF/BF3ÆEt2O reagent. These were conveniently acylatedwith acetic anhydride furnishing 47a–c and then sub-jected to a POCl3-mediated cyclization to the corre-sponding iminium salts 48a–c, which upon NaBH4reduction at �78 �C afforded chiral 1-substituted tetra-hydroisoquinolines 49a–c. Conventional catalytichydrogenolysis of their N-benzyl group provided (S)-(�)-salsolidine, showing that the reduction proceededwith great diastereoselection. Diastereomeric ratios (1S/1R) of salsolidine derivatives were 91:9 for 44a, 100:0 for44b and 98.4:1.6 for 44c, representing efficient examplesof 1,3-asymmetric induction. Interestingly enough, when


bulkier substituents were placed at the C-1 position ofthe resulting tetrahydroisoquinoline, the 2,6-dichloro-phenyl derivative was at least as efficient as its mono-chloro congener.

In this monotonic series of 2,6-diH, 2-Cl, and 2,6-diClderivatives, it was seen that the degree of diastereoselec-tion observed in the reduction of iminium ions 48a–cseemed to be governed by steric factors; apparently, thesingle stereogenic center appended to the nitrogen atomof the iminium ion moiety creates different stericenvironments at the two iminium ion diastereofaces in thetransition state of this nucleophilic addition reaction.

This transformation is ionic in nature, in which a rea-sonable first step may be an ion metathesis, with theformation of an iminium ion–borohydride ion pair priorto the second step, consisting in a hydride addition tothe iminium. Since these compounds possess a C@Ndouble bond embedded in a rather rigid six-memberedring, it appears that only two conformational degrees offreedom are accessible to these structures: (1) rotationabout the C–N bond linking the N to the stereogeniccenter and (2) rotation about the C–C bond linking thestereocenter to the aromatic moiety.

The data provided by these researchers supports theview that iminium ions prefer transition state confor-mations in which the si face is more hindered to nucle-ophilic approach (Scheme 25). Substitution withchlorine increases the net steric shielding of the si dia-stereoface, thus increasing the diastereoselection.

A chirally complementary, modification of the strategydepicted in the above sequence was reported by Ki-bayashi et al.76 This consisted of the synthesis andreduction of isoquinolinium salts, such as 51, bearing anhydrazonium moiety derived from natural proline.

With different precursors, easily elaborated by the Bis-chler–Napieralski cyclization of conveniently substi-

Table 2. Synthesis of salsolidine derivatives 52 by reduction of 51

N

MeO

MeO NR

Me

O

50

N+

MeO

MeO NR

Me

51

POCl3, PhH,reflux

Hydride,Solvent

Entry R Hydride Solvent

1 Me NaBH4 MeOH

2 CH(CH3)2 NaBH4 MeOH

3 CH2OBn NaBH4 MeOH

4 CH2OBn NaBH4 MeOH

5 CH2OBn NaBH4 MeOH

6 CH2OBn LiBEt3 H THF

7 CH2OBn Vitride THF

8 CH2OBn DIBAL-H THF

9 CH2OBn K-Selectride THF

tuted b-phenethylamides 50, salsolidine derivatives 52carrying a (1R)-configuration were isolated in 42–88%yield and 84–96% de. As shown in Table 2, solvent,reducing agent, and reduction temperature were moreinfluential in the recovery yields than in the enantio-meric excesses recorded for the isolated products.Removal and recovery of the proline derived chiralauxili-aries were carried out by refluxing first withBH3ÆTHF, followed by the destruction of the resultingtetrahydroisoquinoline–borane complexes with refluxingHCl.

The preferential formation of one of the enantiomers ofthe natural product was rationalized on the basis of thepyramidal stability of the trivalent nitrogen of the chiralpyrrolidine ring, constituting an efficient asymmetryinducing a stereogenic center, and the existence ofan energetically favored conformer, which can beattacked from its less hindered face.77 An alternativestrategy involving hydrazonium ions, which is discussedbelow, was reported to lead to the opposite enantiomerof the natural product; this entails the nucleo-philic addition of carbon nucleophiles to hydrazoniumsalts.

4.2.4. Enantioselective borane reduction of azomethinesactivated with chiral Lewis acids. A fundamentally dif-ferent approach to that consisting of reduction of chiralactivated azomethines was reported by Kang et al.78

Their protocol employed a borane-mediated reductionof 1-alkyl-3,4-dihydroisoquinoline derivatives usingenantioface-selective coordination on the amine nitro-gen assisted by chiral Lewis acids, which producesactivated iminium species. These authors tested variousLewis acids, such as thiazazincolidine complexes 53aand 53b,79 TADDOL–Ti complex 54,80 Ohno’s catalyst5581 and LL-DPMPM–Zn complex 56 (Fig. 6),82 andfound that the zinc complex 53a gave the optimalenantioselectivity.

N

MeO

MeO HMe

N

MeO

MeO NR

Me

(+)-152

N RH

+

1. BH3.THF,2. HCl

Temperature (�C) De 52 (%) Yield 52 (%)

�50 84 73�50 86 71�10 90 88�50 92 84�90 94 81�50 92 68�50 92 63�50 92 42�50 96 44

NS

Zn

PhMe

R

EtR O

TiOO

O O iPr

iPrOPhPh

Ph Ph

N

N

O iPr

iPrOTf

Tf

NZn O Ph

Ph

MeEt

53a R, R= -(CH2)5-53b R, R= Me, Me

54

55 56

Figure 6.


The borane source proved to be influential in the reac-tion’s outcome. Good chemical yields (60–81%) andenantiomeric excesses ranging from 62% to 86% wereachieved when BH3.THF was employed; other boranesexamined, such as borane–dimethylsulfide complex,bis(2,6-dimethylphenoxy)borane (BDMPB), and pina-col borane furnished enantiomeric excesses in the range4–45%. Interestingly, the reaction provided a mixture ofsalsolidine and a second compound, presumably a bor-ane complex of dihydroisoquinoline 2, which could notbe forced to undergo reduction but reverted to thestarting material upon aqueous work-up.

Selectivity [(R)-salsolidine was obtained in all cases] wasexplained (Scheme 11) on the basis of the coordination

N

MeH

MeO

MeON

MeH

MeO

MeO

(-)-1 (+)-1

OMeMeO

SN

HH

Ph Me

+N

MeOMeO

Zn

Me

S EtN

HH

Ph Me

R

R

R

R

-Favored+N

Zn

Me

Et-

Disfavored

Scheme 11.

Table 3. Reduction of 2 with boranes in toluene, in the presence of thiazazi

Entry no Borane (equiv) Temperature (�C)

1 BH3ÆTHF (2) 102 BH3ÆTHF (2) 03 BH3ÆTHF (2) �54 BDMPB (3) �105 BH3ÆSMe2 (3) �206 Pinacolborane (2) �5

of the chiral Lewis acid to the nitrogen lone pair of thedihydroisoquinoline, which can occur in two ways;the preferred one is that in which the unfavorable A1;3

strain between the bulky methyl group and the ethylgroup on zinc is minimized. In addition to this stericreason, the anti-relationship between the C@N bondin the dihydroisoquinoline and the C–Zn bond in thecatalyst, seems to make the complex leading to (þ)-1 thelower energy species, presumably due to electronicreasons.78

This model also explains the lower enantioselectivityfound when coordinating solvents such as dimethylsulfide or THF are used, on the basis of interference ofthe coordination of the chiral catalyst to the substrate inthe transition state.

The enantiomeric excess (Table 3) was found to improveas the reaction temperature was lowered from 10 to�5 �C (entries 1–3); conversely, chemical yields droppedfrom 81% to 65% as a result of this change in reactionconditions. BH3ÆTHF seemed to be the most efficientreducing reagent within this system.

Cho et al. found that 3,4-dihydroisoquinoline 2 wasinert to Itsuno’s reagent 57a;83a interestingly, however,oxazaborolidine-type catalysts have been employed forthe elaboration of 1-substituted tetrahydroisoquinolinealkaloids.83b;c More recently, Bolm and Felder reportedthat 2 was refractory to the b-hydroxy sulfoximine-cat-alyzed enantioselective reduction with borane, when 57bwas employed as catalyst (Fig. 7);84 it is believed thatunder the reaction conditions, cyclic imines like 2 formborane adducts, which prevent hydride reduction.

ncolidine catalyst 53a

Time (h) Yield (%) Ee, % (config.)

12 81 62 (R)

12 75 79 (R)

12 65 86 (R)

15 32 87 (R)

12 4 60 (R)

12 45 82 (R)

N+BH

Ph

PhH

H3B

HNS

OH

PhPh

OPh

OMeOMeO

BO

H

Me Me

MeMe

57a 57b 57c

N

Me

MeO

MeO

R

RO

BOH

HH

N

Me

MeO

MeO

Si-face attack (favored)

R

R

OB O

H

HH

Re-face attack (disfavored)

Figure 7.

RArSO2


Interestingly, chiral oxaborolidines of the dialkoxy-borane type, like 57c available from the commercialdimethyl LL-tartrate, were observed to reduce cyclic andacyclic azomethines.85 However, 57c is a stoichiometricreductant, which has to be used in approximately five-fold excess in the presence of 1.2 equiv of a mild Lewisacid, such as MgBr2ÆOEt2 for the reaction to be morestereoselective. In this way, (þ)-1 (isolated as theN-carboxymethyl derivative, 8) was accessed in verymodest 50% yield and 28% ee.

The si-face directing effect leading to the preferentialformation of the (1R)-enantiomer can be explained (Fig.7) by considering a transition state, which avoids non-bonding interactions between the methylene protons (H-3) of 2 and the bulky substituent of the dialkoxyboranering.

N

MeO

MeO

Me

NRu

N

Cl

n

H H59a η6-arene= p-cymene; Ar= 4-MeC6H459b η6-arene= p-cymene; Ar= 2,4,6-Me3C6H259c η6-arene= benzene; Ar= 2,4,6-Me3C6H259d η6-arene= benzene; Ar= 1-naphthyl(S,S)-59

(S,S)-59"H2"

+

Si-face attack

(R,R)-59 "H2"

Re-face attack

4.2.5. Catalytic enantioselective reduction of 1-alkyl-3,4-dihydroisoquinolines. The catalytic enantioselectivereduction of 1-alkyl-3,4-dihydroisoquinolines has beenachieved with different degrees of success employingtitanium, ruthenium, and iridium-based catalysts. Whilestudying the viability of titanium catalysts for the cat-alytic reduction of unsaturated organic compounds,Willoughby and Buchwald discovered a titanocene-derived catalytic system (Scheme 12) valuable for theasymmetric hydrogenation of imines.86 Their catalyst58b was generated in situ by metallation with n-BuLi ofthe 1,10-binaphth-2,20-diolate derivative 58a, a previ-ously described ansa-titanocene precatalyst,87 and asubsequent reaction of the metallated species with phe-nyl silane under a hydrogen atmosphere.

58a 58b

1. BuLi (2 equiv.)2. PhSiH3 (2.5 equiv.)O

OTi Ti H

Scheme 12.

Figure 8.

This catalytic system does not require a coordinatinggroup on the substrate and impressive levels of selec-tivity are achieved, taking into account that the catalystoperates by discrimination on the basis of the shape ofthe substrate. In the reaction cycle, the titanium(III)hydride form of the catalyst 58b reacts with the imineproducing a 1,2-insertion and forming a titanium aminecomplex, which is then hydrogenolyzed via a sigmabond metathesis reaction to regenerate the catalyst,releasing the amine product.

Reactions were typically run at 65 �C and, interestingly,enantiomeric excesses achieved by the use of this catalystwere higher with cyclic imines than with their acycliccounterparts; they also required less pressure and thehydrogenation outcome was less sensitive to changes inhydrogen pressure. In the case of the Z-imine 1-methyl-

6,7-dimethoxy-3,4-dihydroisoquinoline 2, however, asmall pressure effect was observed; at 2000 psig theobserved ee of the product was 98%, while at 80 psig theee obtained was 95%. This substrate, which has a syngeometry, upon hydrogenation with the (R,R,R)-cata-lyst gave (�)-1, which is consistent with the generalreaction mechanism proposed for the catalyst.

In recent publications, Noyori et al.88a;b studied theasymmetric transfer hydrogenation of 1-substituted-3,4-dihydroisoquinolines with different preformed chiralRu(II) catalysts of general structure 59 (Fig. 8),employing formic acid–triethylamine media.

Transfer hydrogenation with stable organic hydrogendonors is highly attractive because of its operationalsimplicity, high cost performance and the less hazardousproperties of the reducing agents. Screening experimentsrevealed that the transformation was best effected with a5:2 formic acid–triethylamine azeotropic mixture inacetonitrile at 28 �C, a substrate concentration of 0.5M,a substrate/catalyst ratio of 200 and 6 equiv of HCO2H;these conditions provided a stereodirected synthesis of(þ)-1 in remarkable 99% yield and 95% ee (Table 4,entries 1 and 2).

Interestingly, the reaction did not take place without theaddition of triethylamine or in alcoholic and etherealmedia with the rate and enantioselectivity of the reac-tion being influenced by the g6-arene and 1,2-diamineligands. It is also known that the alkyl substituents onthe p-cymene ligand as well as the free amine and tosylgroup play crucial roles with regards to the reactivity.The general sense of asymmetric induction with thiscatalytic system is illustrated in Figure 8; it has beenproposed that in the stereodetermining hydrogen-transfer step, the chiral Ru species (probably a hydride)formally discriminates the enantiofaces at the sp2

nitrogen atom of the imine, generating a stereogenic sp3

carbon (a-control according to the latent trigonal centerrule).88c

Table 4. Reaction of 2 and 60 with catalyst 59a in MeCN to enantioselectively afford salsolidine 1

N

Me

MeO

MeON

Me

MeO

MeO H

Catalyst

HCO2H-Et3N

21

O

Me

MeO

MeO

NHBoc

1. HCO2H, 9 vol. 16 h2. Et3N (to give 5:2 ratio with HCO2H)3. Catalyst

60

*

Entry no Substrate Catalyst S/C ratio Time (h) Yield (%) Ee, % (config.)

1 2 (S,S)-59a 200 3 >99 95 (R)

2 2 (S,S)-59a 1000 12 97 94 (R)

3 2 (R,R)-59a 400 120 95 88 (S)

4 60 (R,R)-59a 400 20 85 88 (S)


Very recently, Wills et al. reported a one-pot process forthe enantioselective synthesis of salsolidine from theBoc–phenethylamine derivative 60,9c employing areductive amination strategy, under transfer hydroge-nation conditions.88d

These authors were able to demonstrate that formicacid-mediated removal of the Boc protecting group of60, followed by the addition of enough triethylamine togive a 5:2 ratio with the formic acid, and submissionof the mixture to an enantioselective transfer hydro-genation reaction in acetonitrile, in the presenceof the ruthenium(II)-based catalyst (R,R)-59a (pre-pared in situ) provided the natural product withthe same enantiomeric excess obtained when 2 washydrogenated under similar conditions (Table 4,entries 3 and 4). For the synthesis of certain amines,this is a potentially valuable procedure, since imineformation and imine reduction are carried out in onesynthetic step, avoiding the isolation of the imineintermediate.

Cobley et al.89 disclosed another series of ruthenium-based complexes involving chiral diphosphines and di-amines as ligands, as robust and highly useful tools forthe catalytic asymmetric hydrogenation of imines. Oneof these optically active complexes was considered thepreferred one for the enantioselective hydrogenation of2; this transformation employed (R,R)-1,2-diamino-

Table 5. Catalytic hydrogenation of 2 leading to salsolidine with the (R,R)-

N

Me

MeO

MeO

MeO

MeO

H2, 15 BarCatalyst, KtBuO,

2-PrOH

S/C/B= 100:1:20

2 1

Catalyst Temperature (�C)

(R,R)-Et–DuPHOS–RuCl2 (R,R)-DACH 65

(R,R)-Et–DuPHOS–RuCl2 (R,R)-DACH 80

cyclohexane [(R,R)-DACH, 61a] as a diamine, while(R,R)-Et–DuPHOS 61b, a phosphethane type auxiliary,was used as chiral diphosphine; unlike others, this cat-alytic system requires a strong base, such as potassiumtert-butoxide. In this work, conversions were deter-mined by 1H NMR, while the percentages of ee wereobtained by chiral GC analysis of volatile derivatives,with the results shown in Table 5.

Using for asymmetric catalytic hydrogenation catalystssimilar to those employed in the pioneering hydrosily-lation work of Kagan et al. (Scheme 8),63 James et al.disclosed64 that when imine 2 was treated with a rho-dium catalyst, including DIOP 62a as ligand in MeOH,the reaction did not proceed to salsolidine, evidencingthe formation of 63. Under different reduction condi-tions (H2, 1 atm, CH2Cl2), a related complex, whichexisted as a couple of isomers 64a and 64b in solutionstate, was also observed (Fig. 9). These authors alsofound that no reduction product was obtained when(þ)-cycphos 62b (1mol%, PhH–MeOH, 1:1, H2, 1000–1500 psig) was employed as chiral ligand.

For acyclic imines, these researchers64b suggested thatthe coordination of the nitrogen atom of the imine tothe rhodium center is the first step of the reaction. Dueto the importance of the solvent system in the reactionsstudied (PhH–MeOH, 1:1), the existence of a five-coordinated Rh(I) intermediate and its transformation

Et–DuPHOS–RuCl2-(R,R)-DACH catalytic system

NH2

NH2

61a

PEtEt

P

Et

Et

61b

N

MeH*

Time (h) Conversion (%) Ee (%)

68.5 79.5 79

68.5 96 76

N

MeMeO

Rh

MeO

MeO

63

PPh2

PPh2

62b

P

PPhPh

Ph Ph

N

MeCl

Rh

MeO

MeO

64a

P

PPhPh

Ph Ph

N

Me

ClRh

MeOOMe

64b

P

PPhPh

Ph PhO

O

O

O

O

O

O

O

PPh2PPh2

62a

H

Figure 9.


into a monohydride species was assumed, as shownin Scheme 13. In the case of cyclic imines suchas 2, the reaction would not proceed beyond the firststep.

Rh(I)P

P S

S+

Rh(III)P

P H

S

+O

H

HMe

Rh(III)P

P H

N

+O

H

HMe R1

R3

R2

Rh(III)P

P H

N

+O

H

HMe R1

R3

R2

RhP

P S

N

+O

H

HMe R1

R3

R2H

RhP

P S

NO

S

HMe R1

R3

R2H

H

H2

R3

N

R2

R1

H-Transfer

+

H-Transfer

H R2

NR1H

R3 H2

S= SolventP= Tertiary Phosphine

Solvent

Solvent

MeOH(Solvent)

Scheme 13.

The late transition metal complexes in which a coordi-nating group is often necessary for the catalytic asym-metric hydrogenation of cyclic imines, thought time agoto be somehow less successful, have also been employedfor the enantioselective synthesis of salsolidine, as ademonstration of the remarkable progress made in thisarea.

Iridium–iodine based catalytic systems like Ir(I)–MOD–DIOP–TBAI and Ir(I)–BCPM–bismuth(III) iodide,which are efficient for the asymmetric hydrogenation ofcertain cyclic ketimines leading to enantiomeric excessesup to 91%, were found unable to hydrogenate 1-alkyl-3,4-dihydroisoquinolines with enantiomeric excesseshigher than 20%.90a;b

These systems seem to require fine tuning; it has beenrecorded that addition of phthalimide 65 (4mol%) to aniridium(I)–(R,R)-BICP 66a system gave high conversionto salsolidine but with very poor enantiomeric excess(94.6% and ee¼ 4.0%), while addition of benzylamine(5%) to the same catalytic system provided the naturalproduct in no more than 41.3% ee,91a while the use of aniridium(I)–BINAP-phthalimide complex for the suc-cessful elaboration of the related (S)-calycotomine (S)-11 in 86% ee has been recorded91b and a modified DIOPligand (R,R)-MOD-DIOP, 62d was used as part of aneutral iridium complex, which in the presence of TBAIgave 91% of (R)-1, in 27.5 ee (S/C¼ 100, H2 at100 atm).91d

On the other hand, Morimoto et al. recently demon-strated that the addition of imides as co-catalysts al-lowed a remarkable improvement in the asymmetrichydrogenation of 1-alkyl-3,4-dihydroisoquinolines whenthe BCPM 66b type of chiral diphosphine–iridium(I)complex catalysts was employed (Table 6).90c–e Theseimprovements have found uses in other, related sys-tems.91c

Enantiomeric excesses of (�)-1 of up to 93% (entry 6)were obtained with 4 equiv of phthalimide as additive;the reaction was run in toluene at 2–5 �C, under aninitial hydrogen pressure of 100 atm and 1mol%of catalyst, conveniently prepared in situ from[Ir(COD)Cl]2 and (2S,4S)-BCPM (Scheme 14).

N

Me

MeO

MeON

Me

MeO

MeO H

2 (+)-1

N H

O

O

[Ir(COD)Cl]2, 65

N

(c-C6H11)2P

PPh2O O

tBu

66b65

66a or 66b, H2

H

H

Ph2P

PPh2

66a

Scheme 14.

Table 6. Asymmetric hydrogenation of 2 with the diphosphine–iridium(I) catalyst system, employing (2S,4S)-BCPM 66b as the chiral ligand

Entry no Additive Solvent Temperature (�C) Time (h) Conversion (%) Ee, % (config.)

1 None PhH–MeOH 20 24 90 18 (R)

2 None PhMe 20 24 22 14 (S)

3 Succinimide PhH–MeOH �10 72 94 67 (S)4 Phthalimide PhH–MeOH �10 48 98 76 (S)5 Phthalimide PhH 20 24 96 70 (S)

6 Phthalimide PhMe 2–5 24 95 85–93 (S)

7 Phthalimide THF 20 20 95 41 (S)

8 Phthalimide CH2Cl2 20 20 94 70 (S)

9 Succinimide PhH–MeOH �10 72 94 67 (S)10 Saccharin PhH–MeOH 20 45 82 4 (R)

11 Hydantoin PhH–MeOH 20 30 96 49 (S)

12 4-Cl-Phthalimide PhMe 20 20 97 81 (S)

13 4-Cl-Phthalimide PhH–MeOH 20 30 95 56 (S)

14 4,4-Cl2-Phthalimide PhMe 20 20 95 76 (S)

NEtN

Et

N

NEt

Et O

Me

OHH

PhCH2CH2


Clear solvent effects were observed, with less polarsolvents showing higher enantioselectivities; the eeobserved in toluene (>85%) dropped to 41% when THFwas employed, under similar conditions. Five memberedcyclic imides proved to be the best co-catalysts, while theuse of benzamide as co-catalyst provided (þ)-1, in only25% ee.

As mentioned above, the hydrosilylation reactions forthe enantioselective elaboration of 1-substituted tetra-hydroisoquinolines were pioneered by Kagan et al. withlimited success and so this strategy was not applied tothe elaboration of salsolidine.63

More recently, however, a two-step synthesis of (S)-salsolidine by means of the asymmetric hydrosilylationof the precursor nitrone 67 using the Ru2Cl4[S-(�)-p-tolbinap]2–Et3N–Ph2SiH2 catalyst system, has beenreported.92 The intermediate hydroxylamine (2-hydroxysalsolidine, 68), obtained in 99% chemical yield and 56%ee after a 2 h reaction period at 35 �C, was finallyreduced with zinc–HCl to the natural product (Scheme15).

N+

Me

MeO

MeO

N

Me

MeO

MeO H

67

(-)-1

N

Me

MeO

MeO OH

S-68

Ru2Cl4[(S)-(-)-p-tolbinap]2,Et3N, CH2Cl2, 35ºC,

Ph2SiH2

O-

Zn, HCl

PP

Tol

TolTolTol

(S)-(-)-p-Tolbinap

(99%, ee= 56%)

Scheme 15.

NHN CH2OH

N

NH

HOH2C OCH2Ph

Me

OCH2PhH

H

HH

Et

74

69

NEtN

Et

N

NEt

EtO

Me

OH

PhCH2CH2

N

O

MeMeO

MeO

69a

H

Mg+2

Figure 10.

Production of N-hydroxylamine (S)-68 was rationalizedby assuming a mechanism, which involved insertion ofthe Ru–phosphine complex into the Si–H bond of the

silane to give a silylhydridoruthenium species. In turn,this underwent enantioselective insertion into thenitrone, depending on the configuration of the latter.Both, the silane and the catalyst had an influence on theenantiomeric excess of the product, with Ph2SiH2 beingthe most appropriate silane.

The biomimetic reduction of 2 with the NADH ana-logue 69 (Fig. 10) was achieved, yielding salsolidine in36% yield.93 Surprisingly, in spite of the availability ofchiral NADH models, its enantioselective reduction wasnot reported in this work. Finally, it is worth mention-ing that the reduction of 2 with baker’s yeast has alsobeen attempted, but proved to be unsuccessful.94

4.2.6. Diastereoselective catalytic hydrogenation or hy-dride reduction of chiral enamides. A series of reports by


Czarnocki et al.95 indicated the achievement of theenantiodivergent syntheses of (R)- and (S)-N-acetylsalsolidine 19 and its related compounds by catalytichydrogenation or borohydride reduction of enamides 70and 71. The starting enamides were prepared in 50–80%overall yield by acylation of 2, which is readily accessibleby the Bischler–Napieralski reaction,8;9 with acetonidederivatives of the natural (R,R)-tartaric acid as chiralauxiliaries (Scheme 16).

Hydrogenation of the enamides over Adams catalystprovided a mixture of diastereomers 72 and 73 enrichedin the C-1 (R)-isomers (ee¼ 9.7–42.4% after hydrazi-nolysis and N-acylation), while a mixture enriched in theC-1 (S)-isomer (81% chemical yield, ee¼ 15%) wasrecovered from enamides bearing unprotected hydroxylgroups on the 2-tartaroyl moiety, upon reduction withNaBH4 in acidic EtOH at �5 �C.

4.2.7. Enantioselective reduction of enamides with chiralcatalysts. The enantioselective reduction of enamideswith chiral catalysts was accomplished (in several caseswith a high degree of enantioselectivity) with NADHmimics as well as by catalytic hydrogenation employingchiral rhodium(I) and ruthenium(II) complexes as cat-alysts.

N

MeO

MeO

O

NO

O

OOMe

OMe

N

MeO

MeO

O

OEtO

O

O

R1 R2

N

MeO

MeO

N

MeO

MeOMe O

PtO2, H2, 1

70

71a R1, R2= -(CH2)5-71b R1, R2= Me

PtO2, H2,1 atm.

73a R1, R273b R1, R2

1. NaBH4, acidic EtOH2. N2H4, KOH, (CH2OH)2, ∆3. Ac2O, KOH

S-19

MeO

MeO

3a

N

Ph2P

CH2PPh2Piv

66d, Rh+, H2(92%, ee= 45%)

66d

Scheme 16.

The hydrogenation of enamide 3a was repeatedly em-ployed as the model reaction, furnishing N-acetyl sal-solidine 19, the transformation of which into salsolidinein high chemical yield and without loss of stereochemi-cal integrity, is known (Scheme 3).

Bourguignon et al. studied the reduction of enamideswith NADH models as hydride donors. The chiralNADH mimic 74 (Fig. 10), derived from (S)-phenyla-laninol, reduces C@O and C@N bonds in the presenceof a large excess (8 equiv) of Mg(ClO4)2; in this way,enamide 3a was reduced to give 95% of (1R)-N-acetyl-salsolidine (R)-19 in up to 87% ee. Addition of excessMg(ClO4)2 to the reaction medium was critical, since theuse of 1 equiv of the salt provided the product in only32% ee;93 however, the related methyl carbamate 3c wasreduced to furnish 95% of (R)-8, but in only 26% ee.Curiously, inversion of selectivity, which was dependentupon the amount of Mg(ClO4)2 employed, occurredduring the reduction of methyl benzoyl formate to me-thyl mandelate; however, no such inversion was ob-served when the reduction of 3,4-dihydroisoquinolineenamides was carried out.

The use of nonchiral NADH models, such as 69 toprovide racemic 19 and its conversion to racemic sal-solidine, have also been described by this group in ear-

N

MeO

MeO

O

NO

O

OOMe

OMe

O

OEtO

O

O

R1 R2

N

MeO

MeOMe O

atm.

72

= -(CH2)5-= Me

1. Acidic EtOH2. N2H4, KOH, (CH2OH)2,∆

3. Ac2O, KOH

R-19

N

O

N

Ph2P

CH2PPh2Boc

66c, Rh+, H2(82%, ee= 34%)

1. N2H4, KOH,(CH2OH)2, ∆

2. Ac2O, KOH(28%, overall)

(75-85%)

66c


lier publications.96 A ternary complex 69a between thesubstrate, the NADH mimic 69, and Mg2þ was assumedto be responsible for the hydride transfer; in the case ofthe chiral NADH mimic 74, this model correctly ex-plains the preferential si-face hydrogen transfer to 3a,leading to enantioenriched (R)-19.

Rhodium and ruthenium-based chiral catalysts havebeen employed for the catalytic enantioselective hydro-genation of N-acyl enamines. In a pioneering commu-nication by Achiwa97 on the hydrogenation of enamide3a catalyzed by chiral bisphosphine–Rh complexes, theenantioenriched elaboration of both enantiomers ofsalsolidine by simple modifications in reaction condi-tions was reported (Scheme 16).

Employing the rhodium complex of (2S,4S)-N-Boc-4-diphenyl phosphino-2-diphenyl phosphinomethyl-pyr-rolidine 66c (BPPM), (S)-salsolidine was prepared in34% ee. However, when the same catalytic cycle wasperformed with PPPM 66d, the enantiomeric (R)-sal-solidine was obtained in 45% ee after deacetylation; inrelated work, the catalytic enantioselective hydrogena-tion of 3a with the Rh–BICP 66a complex acting as acatalyst to (R)-19 in quantitative yield and ee¼ 77.8%,was also reported,91c as well as the elaboration of (R)-1and (S)-1 by asymmetric hydrogenation of 3a with dif-ferent Rh(I) complexes based on DIOP 62a,c,d and (2R,3S)-MOCBP, 75, and diarylphosphino-pyrrolidine 66eand 66f motifs, as shown in Table 7.91d

A cationic rhodium(I) complex was found to exhibitbetter enantioselectivity than the related neutral one

Table 7. Asymmetric hydrogenation of enamide 3a to N-acetyl salsolidine 1

O

O PAr2PAr2

62a (R,R)-DIOP, Ar= C6H562c (R,R)-p-MeO-DIOP, Ar= 4-MeO-C6H462d (R,R)-MOD-DIOP, Ar= 3,3'-Me2-4MeO-C6H2

PPh2

MeOMeO PPh2

75

N

MeO

MeO

3a

O

EtOH, 50

H2, Cat

Entry no Ligand Rh source S/C ratio

1 62b Rhþ(COD)BF�4 1000

2 62c Rhþ(COD)BF�4 1000

3 62c Rhþ(COD)BF�4 200

4 62c [Rhþ(COD)Cl]2 200

5 62d Rhþ(COD)BF�4 200

6 66e Rhþ(NBD)ClO�4 200

7 66f Rhþ(NBD)ClO�4 200

8 75 Rhþ(COD)BF�4 200

9 75 Rhþ(COD)BF�4 500

10 75 Rhþ(COD)BF�4 200

(entries 3 and 4) and pressure (entries 2 and 3) as well asligand (entries 2, 5, and 6) effects on enantiomericexcesses were also observed.

In several contributions, Noyori et al. reported thehighly enantioselective synthesis of 1-substituted tetra-hydroisoquinolines by the BINAP–Ru(II)-catalyzedhydrogenation (Fig. 11) of N-acyl-1-alkylidene-tetrahy-droisoquinolines. In this way, (�)-1 was quantitativelyaccessed from the precursor N-acetyl enamide 3a, in97% ee employing D-(S)-76, while use of the related N-formyl enamide 3b as starting material furnished theproduct in quantitative yield and 96% ee.98 A mnemonicpicture for the prediction of the sense of the enantio-selective hydrogenation of N-acyl-1-benzylidene-1,2,3,4-tetrahydroisoquinolines with BINAP–Ru(II) catalystswas derived from experimental data; catalytic hydroge-nation of acyl-enamines such as 3a seem to obey thesame principles.46

Hydrogenation conditions are rather mild and verypractical, employing 0.5–1% of catalyst and hydrogen(4 atm) at room temperature in EtOH–CH2Cl2 (5:1)mixtures. Since both enantiomers of the catalyst arereadily available, the synthesis is stereochemically flexi-ble with either of the enantiomers of the natural productcan be potentially synthesized with equal ease, bychoosing the appropriate handedness of the catalyst.

The synthetic scheme is also general, with other tetra-hydroisoquinolines being obtained in high chemicalyield and enantiomeric excess following analogoussequences. Interestingly however, with a given

9 catalyzed by bis-phosphine–Rh(I) complexes

N

PPh2Ph2P

COPhN

Ph2P

CO2t-Bu

P(4-MeO-C6H4)2

66e 66f

N

Me

MeO

MeO

19

O

ºC, 20h

alyst

H2 (atm) Conversion (%) Ee, % (config.)

5 72 40.5 (S)

5 100 51.6 (S)

1 100 62.4 (S)

1 100 48.4 (S)

1 100 29.1 (S)

1 100 25.6 (S)

1 98 46.2 (R)

1 100 80.6 (R)

1 100 80.6 (R)

5 100 76.6 (R)

P

PRu

O

OMe

Me

OO

Ph

PhPh

Ph

P

PRh+

L

L

Ph

PhPh

Ph

L= MeOH

ClO4-

∆-(S)-76 R-77

(R)-BINAP-Ru(II)

N

MeO

MeO

HH Me

O

(S)-BINAP-Ru(II)

Figure 11.

NH

MeO

MeON

MeO

MeON

Me3CO

N

MeO

MeON

Me3CO

Me

N

MeO

N

MeO

MeOOMe

N

MeO

1. t-BuLi2. MeI

Me3CO

N

Me2N

23

79

80

1. Boc2O2. tBuLi-TMEDA3. MeI (99%)

21b

N2H4, AcOH

NaAlH4(76%)

PhHN Ph

BuLi,TMEDA

(94%,ee= 86%)

81

(96%)

(60-63%overall)

78


handedness, the related rhodium-based BINAP catalystssuch as 77 deliver the opposite configuration at theC-1 position through hydrogenation; therefore, notsurprisingly, hydrogenation of the same N-acetylderivative 3a with the rhodium catalyst (R)-77 gave (S)-N-acetylsalsolidine (S)-19 in 82% chemical yield and60% ee.46

It was also found that the N-acyl function is crucial forthe appropriate reaction with Rh(I) catalysts becauseit acts like a tether, binding the heterocycle to thecatalytic metal center. Another interesting differencebetween Ru(II) and Rh(I) catalysts is that when 77is employed, the N-formyl substrates, such as 3b, arenot hydrogenated under conditions where 3a is easilyreduced.

HMeOMe

MeOOMe

S-21b (-)-1

Scheme 17.

4.3. Metallation of tetrahydroisoquinoline derivatives

4.3.1. Diastereoselective alkylation of metallated chiraltetrahydroisoquinoline derivatives. Substituents can beintroduced at the a-position of secondary amines viametallation, followed by treatment of the metallatedspecies with the appropriate electrophile, by alkylationof azomethines as well as their synthetic equivalents oractivated forms, or by catalytic oxidation and sub-sequent treatment of the oxidized intermediates withappropriate nucleophiles.99;100

Throughout their studies on the alkylation of a-aminocarbanions derived from chiral formamidines, carriedout during the 1980s and the beginning of the 1990s,Meyers et al. were among the first to study the enan-tioselective formation of new C–C bonds adjacent tonitrogen atoms.101 As a result, several model tetrahy-droisoquinolines were enantioselectively alkylated withvarious alkyl halides in 50–70% yields and excellentenantiomeric excesses, employing different chiral form-

amidines, and many natural products were synthesizedin this way.102

The most consistent results were observed with thechiral auxiliary derived from (S)-valinol tert-butyl ether,while others gave products in 80–99% ee. For instance,(S)-1 was obtained by this procedure in 60–63% yieldand >97% ee, by way of the formamidine derivative 79,easily available from the known 6,7-dimethoxytetra-hydroisoquinoline 23 and formamidine 78 (Scheme17).103a

Removal of the chiral auxiliary from the resultingalkylated derivative 80 was conveniently carried out byhydrazinolysis. Formamidines derived from (S)-salsoli-dine itself were also employed for the elaboration ofasymmetric 1,1-disubstituted tetrahydroisoquino-lines.103b

By analogy with models proposed for other opticallyactive formamidines,101c the high stereoselectivity ob-served for the (S)-valinol tert-butyl ether chiral auxiliaryin 80 may be attributed to the different conformationalpreferences of the diastereomeric lithium salts 82a and82b; in the former, the chiral auxiliary appeared to lieover the plane of the isoquinoline ring, while in thelatter, the auxiliary was placed out and away from the

NEtO


isoquinoline, as a consequence of the (S)-configurationof the stereogenic center (Fig. 12).

N

H

(S)

N

MeO

MeO

MeO

MeOLi

N

(S)

MeI

Me3CO

H

82a 82b

Solvent

SolventH

N

Me3CO

Li

Figure 12.

NH

MeO

MeON

MeO

MeON

1. t-BuLi, -78ºC2. MeI, -100ºC

(93%)

O

23 84

85(+)-1

O

N

MeO

MeON

O

Me

NH

MeO

MeOMe

N2H4(100%)

83PhH, TsOH (cat.)

Scheme 18.

Steric factors disfavored the production of 82a, whilethe steric blockade of the bottom side of the preferreddiastereomer 82b, resulted in alkylation occurring fromthe top side, furnishing the observed (S)-configurationof the product. The role of the chiral auxiliary versusthe configurational stability of the C–Li bond wasstudied. Interestingly, metallation and alkylation ofoptically active salsolidine carrying an achiral form-amidine moiety afforded completely racemic mate-rial.103c

Resorting to an analogous sequence of transformations,Gawley et al. reported the synthesis of (R)-salsolidine bythe use of oxazolines as chiral auxiliaries for the asym-metric alkylation of amines,104 thus providing a chirallycomplementary sequence to that disclosed by Meyerset al. for the synthesis of the natural product.

To that end, tetrahydroisoquinoline 23 was transformedinto 84 via reaction with (S)-ethoxyoxazoline 83, derivedfrom (S)-valine. In turn, this was metallated with t-BuLiat �78 �C and the resulting species quenched withMeI at �100 �C, to furnish 93% of a chromato-graphically separable mixture (9:1) of diastereomericoxazolines 85. Hydrazinolysis of 85 caused removalof the oxazoline moiety, providing (þ)-1 quantita-tively, as shown in Scheme 18, in excellent enantio-meric purity.34a Similarly, Quirion et al.104c were ableto prepare (+)-1 in 53% yield and up to 93% eefrom the tetrahydroisoquinoline 23 through the dia-stereoselective alkylation of amides derived fromgulonic acid.

4.3.2. Enantioselective protonation of metallated tetrahy-droisoquinoline derivatives. Recently, Simpkins et al.disclosed another interesting and complementary strat-egy toward chiral salsolidine (Scheme 17), involving theenantioselective protonation of metallated tetrahydro-isoquinoline derivatives.50 Thus, metallation of N-piva-loyl salsolidine 21b with the t-BuLi–TMEDA complexat �40 �C in the presence of 81, a chiral amine derivedfrom a-phenethylamine provided 94% of N-pivaloyl-(S)-salsolidine (S)-21b in 86% ee, an enantiomeric excess,

which was raised to 96% upon crystallization frompetroleum ether. Reductive deprotection of the N-piva-loyl moiety with sodium aluminum hydride afforded(�)-1 in 76% yield. The asymmetric reactions oforganolithium reagents under control of chiral ligandshave recently been reviewed.105

4.4. Alkylation of azomethines

4.4.1. Enantio

TETRAHEDRON: ASYMMETRY REPORT NUMBER 63 Synthetic...

Documents

Transcript of TETRAHEDRON: ASYMMETRY REPORT NUMBER 63 Synthetic...