Akademisk avhandling som med tillstånd av Kungl Tekniska Högskolan i Stockholm framlägges till offentlig granskning för avläggande av teknologie doktorsexamen i kemi med inriktning mot organisk kemi fredagen den 14 september kl 10.00 i sal F3, KTH, Lindstedtsvägen 26, Stockholm. Avhandlingen försvaras på engelska. Opponent är Professor Pher G. Andersson, Uppsala universitet.

Stereoselective Synthesis of Amino Alcohols:

Applications to Natural Product Synthesis

Staffan Torssell

Doctoral Thesis

Stockholm 2007

ISBN 978-91-7178-734-7 ISSN 1654-1081 TRITA-CHE-Report 2007:52 © Staffan Torssell, 2007 Universitetsservice US AB, Stockholm

Staffan Torssell, 2007: ”Stereoselective Synthesis of Amino Alcohols: Applications to Natural Product Synthesis” Organic Chemistry, KTH Chemical Science and Engineering, Royal Institute of Technology, SE-100 44 Stockholm, Sweden.

Abstract This thesis is divided into four separate parts with amino alcohols as the common feature. The first part of the thesis describes the development of an efficient three-component approach to the synthesis of α-hydroxy-β-amino esters. Utilizing a highly diastereoselective Rh(II)-catalyzed 1,3-dipolar cycloaddition of carbonyl ylides to various aldimines, syn-α-hydroxy-β-amino esters are formed in high yields and excellent diastereoselectivities. An asymmetric version was also developed by employing chiral α-methylbenzyl imines as dipolarophiles yielding enantiomerically pure syn-α-hydroxy-β-amino esters. This methodology was also applied on a short asymmetric synthesis of the paclitaxel side-chain as well as in an asymmetric synthetic approach towards the proteasome inhibitor omuralide. Furthermore, the use of chiral Rh(II) carboxylates furnishes the syn-α-hydroxy-β-amino esters in moderate enantioselectivity (er up to 82:18), which indicates that the reaction proceeds via a metal-associated carbonyl ylide. The second part describes the development of a 1,3-dipolar cycloaddition reaction of azomethine ylides to aldehydes for the synthesis of α-amino-β-hydroxy esters. Different methods for the generation of the ylides, including Vedejs’ oxazole methology and an Ag(I)/phosphine-catalyzed approach have been evaluated. The best results were obtained with the Ag(I)/phosphine approach, which yielded the desired α-amino-β-hydroxy ester in 68% yield and 3.4:1 syn:anti-selectivity. The last two parts deals with the total synthesis of the amino alcohol-containing natural products D-erythro-sphingosine and (−)-stemoamide. The key transformation in the sphingosine synthesis is a cross-metathesis reaction for the assembly of the polar head group and the aliphatic chain. In the stemoamide synthesis, the key feature is an iodoboration/Negishi/RCM-sequence for the construction of the β,γ-unsaturated azepine core of stemoamide followed by a stereoselective bromolactonization/1,4-reduction strategy for the installation of the requisite C8-C9 trans-stereochemistry. Keywords: Amino alcohol, asymmetric 1,3-dipolar cycloaddition, azomethine ylide, carbenoid, carbonyl ylide, cross-metathesis, omuralide, oxazolidine, rhodium, sphingosine, stemoamide, stereoselective synthesis, total synthesis.

Abbreviations

1,3-DC 1,3-dipolar cycloaddition 9-BBN 9-borabicyclo[3.3.1]nonane B-I-9-BBN 9-iodo-9-borabicyclo[3.3.1]nonane BINOL 1,1-bi-2-naphthol BOP-Cl bis(2-oxo-3-oxazolidinyl)phosphinic chloride BOX bis(oxazoline) CM cross-metathesis Cy cyclohexyl DBB 4,4’-di-tert-butylbiphenyl DCE 1,2-dichloroethane (DHQ)2PHAL hydroquinine 1,4-phtalazinediyl diether (–)-DIPT (–)-diisopropyl tartrate DMAD dimethyl acetylenedicarboxylate DMI 1,2-dimethylimidazole DMPU 1,3-dimethyl-3,4,5,6-tetrahydro-2(1H)-pyrimidinone DOS diversity-oriented synthesis EDA ethyl diazoacetate FMO frontier molecular orbital KHMDS potassium hexamethyldisilazane LiHMDS lithium hexamethyldisilazane NaHMDS sodium hexamethyldisilazane N.A. not available n.d. not determined n.r. no reaction p-TSA p-toluene sulfonic acid PyBOX pyridine bis(oxazoline) quant. quantitative yield QUINAP 1-(2-diphenylphosphino-1-naphthyl)isoquinoline RCM ring-closing metathesis Rh2(pfb)4 rhodium perfluorobutyrate Rh2(S-biTISP)2 dirhodium bis[bridged-di(S-(N-2,4,6-triisopropyl phenylsulfonyl)prolinate) Rh2(S-DOSP)4 dirhodium tetrakis(S-(N-dodecyl benzenesulfonyl)prolinate) SAE Sharpless asymmetric epoxidation XC chiral auxiliary Δ heating (typically to reflux)

List of Publications

This thesis is based on the following papers, referred to in the text by their Roman numerals I-V:

I. 1,3-Dipolar Cycloadditions of Carbonyl Ylides to Aldimines: A Three Component Approach to syn-α-Hydroxy-β-amino Esters Staffan Torssell, Marcel Kienle and Peter Somfai Angew. Chem. Int. Ed. 2005, 44, 3096-3099 Angew. Chem. 2005, 117, 3156-3159

II. 1,3-Dipolar Cycloadditions of Carbonyl Ylides to Aldimines: Scope, Limitations and Asymmetric Cycloadditions Staffan Torssell and Peter Somfai Adv. Synth. Catal. 2006, 348, 2421-2430

III. 1,3-Dipolar Cycloadditions of Azomethine Ylides to Aldehydes: Synthesis of α-Amino-β-hydroxy Esters Staffan Torssell and Peter Somfai Preliminary Manuscript

IV. A Practical Synthesis of D-erythro-Sphingosine Using a Cross-Metathesis Approach Staffan Torssell and Peter Somfai Org. Biomol. Chem. 2004, 2, 1643-1646

V. Total Synthesis of (−)-Stemoamide Staffan Torssell, Emil Wanngren and Peter Somfai J. Org. Chem. 2007, 72, 4246-4249

Table of Contents

Abstract Abbreviations List of publications 1. Introduction .............................................................................................1

1.1. Stereoselective Synthesis .......................................................................... 3 1.2. β-Amino Alcohols ........................................................................................ 4 1.3. Synthesis of β-Amino Alcohols................................................................... 5

1.3.1. Amino Alcohols from a Pre-Existing Carbon Skeleton .....................................5 1.3.2. Amino Alcohols from C-C Bond Forming Reactions.........................................7

1.4. 1,3-Dipolar Cycloaddition Reactions.......................................................... 9 1.4.1. 1,3-Dipoles .......................................................................................................10 1.4.2. Dipolarophiles...................................................................................................10 1.4.3. Regio- and Stereoselectivity in 1,3-Dipolar Cycloadditions............................11

1.5. Aim of This Thesis ....................................................................................13 2. 1,3-Dipolar Cycloadditions of Carbonyl Ylides to Aldimines:

Synthesis of syn-α-Hydroxy-β-amino Esters ......................................15 2.1. Introduction................................................................................................15

2.1.1. 1,3-Dipolar Cycloadditions of Carbonyl Ylides................................................15 2.1.2. Aim of the Study and Synthetic Strategy.........................................................17

2.2. 1,3-Dipolar Cycloadditions .......................................................................18 2.3. Mechanistic Aspects of the 1,3-Dipolar Cycloaddition............................23 2.4. Asymmetric 1,3-Dipolar Cycloadditions...................................................25

2.4.1. Chiral Rh(II) Catalysts......................................................................................25 2.4.2. Chiral Diazo Esters ..........................................................................................26 2.4.3. Chiral Imines ....................................................................................................27 2.4.4. Chiral Lewis Acids............................................................................................29

2.5. Synthetic Applications: Paclitaxel Side-Chain and Towards the Total Synthesis of Omuralide ............................................................................31

2.5.1. Paclitaxel Side-Chain.......................................................................................31 2.5.2. Omuralide: Introduction....................................................................................32 2.5.3. Retrosynthetic Analysis of Omuralide .............................................................34 2.5.4. Synthetic Efforts Towards Omuralide..............................................................35

2.6. Conclusions...............................................................................................38 3. 1,3-Dipolar Cycloadditions of Azomethine Ylides to Aldehydes:

Synthesis of syn-α-Amino-β-hydroxy Esters ......................................39 3.1. Introduction................................................................................................39

3.1.1. Generation of Azomethine Ylides....................................................................39 3.1.2. Aim of the Study and Synthetic Strategy.........................................................41

3.2. 1,3-Dipolar Cycloadditions .......................................................................42 3.2.1. 1,3-Dipolar Cycloadditions of Oxazoline-Derived Azomethine Ylides ...........42 3.2.2. 1,3-Dipolar Cycloadditions of Metallo-Azomethine Ylides..............................44 3.2.3. Mechanistic Aspects ........................................................................................45

3.3. Conclusions and Outlook .........................................................................47 4. Total Synthesis of D-erythro-Sphingosine .........................................49

4.1. Introduction................................................................................................49 4.2. Retrosynthetic Analysis of D-erythro-Sphingosine ..................................50 4.3. Synthesis of D-erythro-Sphingosine.........................................................51

4.3.1. Oxazolidinone Synthesis .................................................................................51 4.3.2. Cross-Metathesis and Completion ..................................................................52

4.4. Conclusions...............................................................................................54 5. Total Synthesis of (−)-Stemoamide ...................................................55

5.1. Introduction................................................................................................55 5.2. Retrosynthetic Analysis of (−)-Stemoamide ............................................56 5.3. Synthesis of (−)-Stemoamide...................................................................57

5.3.1. Construction of the Pyrrolo[1,2-a]azepine Core..............................................57 5.3.2. Lactonization and Completion .........................................................................58

5.4. Conclusions...............................................................................................61 6. Concluding Remarks............................................................................63 Acknowledgements Appendices

1

1. Introduction

During the last decades, organic chemists have achieved spectacular progress in the synthesis of complex molecules. The synthesis of brevetoxin A and B by K. C. Nicolaou and co-workers, a project that took 16 years to finish, still stands as a milestone in total synthesis (Figure 1).1

O

OO

O

O

O

O

OO

O O

CHO

O

Me

H H HH

H H Me

H

H

H

H Me HH

Me Me HH

H

Me

HO

Me Figure 1. Brevetoxin B

Although synthetic organic chemists have clearly shown, by their amazing work, that they are capable of synthesizing almost any compound imaginable, there is a constant, growing demand for new discoveries in the field of organic chemistry since the synthetic methods existing today are still far from satisfactory. The challenge in organic chemistry today, lies not so much in the synthesis of monstrous natural products, which can evidently be achieved with enough manpower, time and money, as in the development of new, straightforward methodologies for the construction of complicated targets and sub-structures. For example, accurate control of the individual functional groups within a complex molecule (chemoselectivity) still remains an unanswered question. Traditionally, this challenge has been solved by the use of protecting group to allow the functional groups to react on an individual basis. Ideally, this strategy overcomes many of the chemoselectivity problems encountered in the synthesis of complex molecules but it often adds an additional layer of synthetic complexity. This not only adds to the amount of chemical operations within a synthesis, but also lowers the overall efficiency of the synthesis. Two interesting solutions to the chemoselectivity issue in the field of natural product synthesis are the four-step enzymatic synthesis of the anti-cancer agent epothilone C by Khosla and co-workers2 and the protecting 1 (a) Nicolaou, K. C.; Tiebes, J.; Theodorakis, E. A.; Rutjes, F. P. J. T.; Koide, K.; Sato, M.;

Untersteller, E. J. Am. Chem. Soc. 1994, 116, 9371-9372. (b) Nicolaou, K. C.; Yang, Z.; Shi, G.-S.; Gunzner, J. L.; Agrios, K. A.; Gärtner, P. Nature 1998, 392, 264-269. (c) Nicolaou, K. C. Tetrahedron 2003, 59, 6683-6738.

2 Boddy, C. N.; Hotta, K.; Tse, M. L.; Watts, R. E.; Khosla, C. J. Am. Chem. Soc. 2004, 126, 7436-7437.

2

group-free synthesis of the marine alkaloids ambiguine H, hapalindole U, welwitindolinone A and fischerindole I by the Baran group.3 Another new area, which has grown exponentially during the last decade, is organocatalysis, pioneered by the MacMillan, List and Barbas groups.4 In organocatalysis, small organic molecules act as the catalytic species with the advantage of being metal-free, usually non-toxic, readily available and often very robust. Due to the absence of transition metals, organocatalysis appear to be an attractive complement to more traditional methods in areas where metal contaminations are not tolerated, e.g. the pharmaceutical industry. These are just a few examples where the chemists are recognizing what could be looked upon as simple, but still brilliant solutions to many of the problems existing in modern organic chemistry, such as long, linear synthetic sequences, hazardous byproducts, the need for expensive and sensitive metal-catalysts. Other areas like diversity-oriented synthesis5 and chemical genetics6,7 described by Schreiber and co-workers, have opened new opportunities for chemists to use small molecules to perturb and gain new understanding of functions required in living organisms. The use of small molecules to dissect biological pathways in the same manner as in genetics, where random mutations are used for screening of specific cellular functions, has led to considerable efforts in constructing planning algorithms for the synthesis of complex and diverse collections of compounds.8 From a drug discovery point of view this approach would lead to a simultaneous identification of target proteins and small molecules that can act as protein modulators. These are just a few of the new revolutionizing areas within the organic chemistry community that will set the standard for future developments.

3 Baran, P. S.; Maimone, T. J.; Richter, J. M. Nature 2007, 446, 404-408. It should be noted that

this is not the first example of a protecting group-free total synthesis, for example see Robinson’s tropinone synthesis: Robinson, R. J. Chem. Soc. 1917, 111, 762-768. However, due to the complexity of the molecular framework in the current example it must be considered to be a landmark within natural product synthesis.

4 For excellent reviews, see: (a) Dalko, P. I.; Moisan, L. Angew. Chem., Int. Ed. 2001, 40, 3726-3748. (b) Dalko, P. I.; Moisan, L. Angew. Chem., Int. Ed. 2004, 43, 5138-5175. (c) Berkessel, A.; Gröger, H. Asymmetric Organocatalysis: From Biomimetic Concepts to Applications in Asymmetric Synthesis; Wiley-VCH Verlag GmbH: Weinheim, 2005. (d) Seayad, J.; List, B. Org. Biomol. Chem. 2005, 3, 719-724; (e) Lelais, G.; MacMillan, D. W. C. Aldrich. Acta 2006, 39, 79-87.

5 Schreiber, S. L. Science 2000, 287, 1964-1969. 6 Chemical genetics, def: Biological investigation by modulating protein functions with small

molecules. 7 Schreiber, S. L. Bioorg. Med. Chem. 1998, 6, 1127-1152. 8 For a review on “A planning strategy for diversity-oriented synthesis”, see: Burke, M. D.;

Schreiber, S. L. Angew. Chem. Int. Ed. 2004, 43, 46-58.

3

1.1. Stereoselective Synthesis

The world around us is chiral9 and ever since the discovery that two enantiomers of a chiral compound can have different pharmacological effect in living systems asymmetric synthesis have been one of the most expansive fields in organic chemistry. The birth of stereoselective synthesis dates back to 1890, when Emil Fischer realized that addition of sodium cyanide to chiral carbohydrate aldehydes gave an uneven distribution of the two possible diastereomers.10 In order to synthesize an optically active compound one has not only to consider the connectivity of the atoms (constitution) within the molecule but also the three-dimensional structure (stereoisomerism). The spatial distribution of the substituent in a chiral molecule can have a dramatic impact on the interaction with other chiral compounds, e.g. a chiral drug and a biological receptor and there are numerous examples of chiral compounds where the two enantiomers have fundamentally different effects in biological systems.11 The solution to these problems lies within stereoselective synthesis, which allows for the formation of one stereoisomer in preference to others. Different strategies can be employed in order to access enantiomerically enriched compounds including resolution of a racemic mixture, the chiral-pool approach or asymmetric synthesis. Stereoselective or asymmetric synthesis can be further divided into the following subgroups:12

1) Reagent control: The formation of a new stereogenic center is governed by a chiral reagent or catalyst not covalently bound to the substrate (enantioselective or diastereoselective).

2) Substrate control: The formation of a new stereogenic center is controlled by the inherent chirality of the substrate (diastereoselective).

3) Auxiliary control: The formation of a new stereogenic center is controlled by a stoichiometric amount of a chiral auxiliary covalently bound to the substrate but not part of the final structure (diastereoselective).

9 Chiral, greek; handed. Body with non-superimposable mirror images. 10 Fischer, E. Chem. Ber. 1894, 27, 3189-3232. 11 Lin, G.-Q.; Li, Y.-M.; Chan, A. S. C. Principles and Applications of Asymmetric Synthesis,

Wiley-Interscience: New York, 2001, pp. 4-7. 12 For further discussion, see: (a) Nógrádi, M. Stereoselective Synthesis, VCH Verlagsgesellschaft

mbH: Weinheim, 1995. (b) Eliel, E. L.; Wilen, S. H. Stereochemistry of Organic Compounds, Wiley-Interscience: New York, 1994.

4

1.2. β-Amino Alcohols

The β-amino alcohol and α-hydroxy-β-amino acid moieties are found in a large variety of biologically important compounds, e.g. natural products and peptides, as well as in a growing number of ligands and chiral auxiliaries for asymmetric synthesis. Amino alcohols are mainly divided into three general subgroups:13

1) Naturally occurring amino alcohols. 2) Synthetic, pharmacologically active amino alcohols. 3) Chiral catalysts and auxiliaries containing the amino alcohol motif.

Hydroxy amino acids are the most common class of naturally occurring compounds containing the β-amino alcohol subunit. For example, the vancomycin14 class of antibiotics contains an arylserine moiety (Figure 2) and the antifungal agent sphingofungin (Figure 12, Chapter 4) contains a hydroxy amino acid moiety in the polar head group.

N

N

HHOMeO

H

H

N NH

HN

O

OH

Ph

Quinine

Saquinavir

NHO

OHHHOOH

(+)-Castanospermine

HN

O

O

O

NH

HOCl

O

NH

OH

O

Vancomycin

N

NH

OH

MeO2C

H

Me

Vinblastine

N

N

OAc

OH

H

CO2Me

Me

MeO

Me

O

OH

NH

Cl

NH

O HN

O

O

NH

Me

NH2

O

OH

HO

HO2C

O

OHHO

OH

O

O

HOMe

MeH2N

O

N

H

H

Ot-BuHN

O

NH2

Me

Me

Figure 2. Biologically Active β-Amino Alcohols (β-AA in Bold Face Letters)

Another large group of biologically active natural products, which includes numerous alkaloids, is the cyclic amino alcohols. Pertinent examples being vinblastine,15 used in cancer therapy, and quinine that is used in malaria treatment. Also belonging to the group of cyclic amino alcohols is the polyhydroxylated alkaloids, also known as aza-sugars, e.g. (+)-castanospermine,16 which is a potent inhibitor of α- and β-glucosidases.

13 Bergmeier, S. C. Tetrahedron 2000, 56, 2561-2576. 14 Nicolaou, K. C.; Snyder, S. A. Classics in Total Synthesis II, Wiley-VCH Verlag GmbH:

Weinheim, 2003, pp 239-300, and references cited therein. 15 Nicolaou, K. C.; Snyder, S. A. Classics in Total Synthesis II, Wiley-VCH Verlag GmbH:

Weinheim, 2003, pp 505-531, and references cited therein. 16 (a) Michael, J. P. Nat. Prod. Rep. 1999, 16, 675-696. (b) Michael, J. P. Nat. Prod. Rep. 2001,

18, 520-542.

5

Peptidomimetics constitute the most important group of synthetically produced pharmacologically active amino alcohols, most commonly HIV-1 protease inhibitors, for example Saquinavir (Figure 2).13,17 β-Amino alcohols also play an important role as chiral ligands and chiral auxiliaries in asymmetric catalysis, most commonly derived from natural sources. The amino alcohols are generally derivatized to improve their chelating ability or to increase their steric directing effect.18 Figure 3 depicts β-amino alcohol derivatives used in asymmetric synthesis.19

N

HO

OHN

R1 R2

ON

O

N N

O

(S,S)-i-Pr-PyBOX Evan's Oxazolidinone(+)-N-methyl ephedrine Figure 3. Examples of β-Amino Alcohol-containing Chiral Ligands and Auxiliaries

1.3. Synthesis of β-Amino Alcohols

Synthetic routes toward enantiomerically pure β-amino alcohols have traditionally relied on derivatization of the chiral pool of amino acids, with the inherent limitation of accessible targets.20 To gain access to compounds with other functionalities, considerable efforts have been made to develop asymmetric routes to β-amino alcohols, which to date can be divided into two strategically different approaches:

1) Introduction of the β-amino alcohol moiety on a pre-existing carbon skeleton.

2) Concomitant formation of a new carbon-carbon bond and one or two of the vicinal stereogenic centers in one single step.

1.3.1. Amino Alcohols from a Pre-Existing Carbon Skeleton The most common approach to stereoselective synthesis of β-amino alcohols starts from a pre-existing carbon skeleton. Alkene derivatives are frequently used as substrates for these transformations that are stereospecific. A major drawback with these reactions is often the low regioselectivity, which can generally be circumvented if the substrate contains a regio-directing group.

17 Baker, W. R.; Condon, S. L. J. Org. Chem. 1993, 58, 3277-3284. 18 Ager, D. J.; Prakash, I.; Schaad, D. R. Chem. Rev. 1996, 96, 835-875. 19 (a) Frantz, D. E.; Fässler, R.; Carriera, E. M. J. Am. Chem. Soc. 2000, 122, 1806-1807.

(b) Kobayashi, S.; Sugiura, M.; Kitagawa, H.; Lam, W. W.-L. Chem. Rev. 2002, 102, 2227-2302. (c) Ager, D. J.; Prakash, I.; Schaad, D. R. Aldrichim. Acta 1997, 30, 3-12.

20 Reetz, M. T. Angew. Chem. Int. Ed. 1991, 30, 1531-1546.

6

One of the most investigated routes towards enantiomerically pure β-amino alcohols is the opening of epoxides with nitrogen nucleophiles. Since both cis- and trans-epoxides are available in high enantiomeric purity this approach can be utilized in the synthesis of both syn- and anti-β-amino alcohols. The regioselectivity of epoxide openings is often poor but can be controlled by introduction of regio-directing groups, most commonly phenyl or vinyl. The attack of hard nucleophiles usually proceeds at the activated benzylic or allylic carbon (Scheme 1).21,22 Scheme 1. Aminolysis of Vinylepoxides

R1

O

NH4OH R1

OH

R2

R3

R4

R3

R4

R2 NH2

78 - 100%regioselectivity 2:1 - 100:0

!

β-Amino alcohols can also be obtained in an analogous fashion through ring opening of other cyclic substrates such as aziridines,22,23 sulfates24 and carbonates.25 The most direct approach toward enantioselective synthesis of β-amino alcohols is the Sharpless asymmetric aminohydroxylation of alkenes in which the same chiral catalyst as in the asymmetric dihydroxylation reaction is utilized. α,β-Unsaturated esters and phosphonates have proven to be the most suitable substrates for this reaction (Scheme 2).26 Although this transformation is an attractive approach to the direct enantioselective synthesis of amino alcohols, the yields are often moderate due to regioselectivity problems. Scheme 2. Sharpless Asymmetric Aminohydroxylation

R OMe

O

R OMe

OMsHN

OH

MsClNNa(DHQ)2PHAL

K2OsO2(OH)4

55 - 90%ee 90 - 99%

R OMe

OOH

NHMs

21 Jaime, C.; Ortuno, R., M.; Font, J. J. Org. Chem. 1988, 53, 139-141. 22 Olofsson, B.; Somfai, P. J. Org. Chem. 2002, 67, 8574-8583. 23 Hwang, G.-I.; Chung, J.-H.; Lee, W. K. J. Org. Chem. 1996, 61, 6183-6188. 24 (a) Lohray, B. B.; Gao, Y.; Sharpless, K. B. Tetrahedron Lett. 1989, 30, 2623-2626. (b) Chang,

H.-T.; Sharpless, B. Tetrahedron Lett. 1996, 37, 3219-3222. 25 Cho, G. Y.; Ko, S. Y. J. Org. Chem. 1999, 64, 8745-8747. 26 Li, G.; Chang, H.-T.; Sharpless, B. K. Angew. Chem. Int. Ed. 1996, 35, 451-454.

7

1.3.2. Amino Alcohols from C-C Bond Forming Reactions The amino alcohol moiety can also be constructed by coupling two fragments, one containing the oxygen functionality and one containing the nitrogen functionality, with a concomitant formation of a new carbon-carbon bond and two vicinal stereogenic centers. In order for this approach to be efficient both enantio- and diastereocontrol is required. However, this approach is limited to certain types of substrates. There are mainly two different strategies employed for enantioselective synthesis of amino alcohols based on C-C bond forming reactions; Mannich-type reactions27 and addition of glycine derived enolates to aldehydes,28 both of which will be discussed below. One elegant example of a highly stereoselective Mannich-type reaction is based on a nucleophilic addition of α-alkoxy enolates to imines affording amino alcohols with high to excellent enantioselectivity (Scheme 3).29 Depending on the choice of enolate, both syn (R2 = TBS) and anti (R2 = Bn) amino alcohols can be formed with high diastereoselectivity. Scheme 3. Enantioselective Mannich-type Approach

NAr

HR1R2O

OR3

OTMSZr(Ot-Bu)4

(R)-Br2-BINOL

DMI R1 OR3

ArHN

OR2

O

R1 OR3

ArHN

OR2

O

41 - 100%syn:anti 99:1 - 6:94

ee 76 - 98% Amino alcohols can also be synthesized by Lewis acid-catalyzed aldol reactions. Zirconium/BINOL-catalyzed reactions of glycine derived silyl ketene acetals to aldehydes furnish anti-α-amino-β-hydroxy acids in excellent yields and enantioselectivities (Scheme 4).30

27 (a) List, B.; Pojarliev, P.; Biller, W. T.; Martin, H. J. J. Am. Chem. Soc. 2002, 124, 827-833.

(b) Córdova, A.; Notz, W.; Zhong, G.; Betancort, J. M.; Barbas III, C. F. J. Am. Chem. Soc. 2002, 124, 1842-1843. (c) Yoshida, T.; Morimoto, H.; Kumagai, N.; Matsunaga, S.; Shibasaki, M. Angew. Chem. Int. Ed. 2005, 44, 3470-3474. (d) Trost, B. M.; Jaratjaroonphong, J.; Reutrakul, V. J. Am. Chem. Soc. 2006, 128, 2778-2779.

28 (a) Horikawa, M.; Busch-Petersen, J.; Corey, E. J. Tetrahedron Lett. 1999, 40, 3843-3846. (b) Yoshikawa, N.; Shibasaki, M. Tetrahedron 2002, 58, 8289-8298. (c) Ooi, T.; Kameda, M.; Taniguchi, M.; Maruoka, K. J. Am. Chem. Soc. 2004, 126, 9685-9694.

29 Kobayashi, S.; Ishitani, H.; Ueno, M. J. Am. Chem. Soc. 1998, 120, 431-432. 30 Kobayashi, J.; Nakamura, M.; Mori, Y.; Yamashita, Y.; Kobayashi, S. J. Am. Chem. Soc. 2004,

126, 9192-9193.

8

Scheme 4. Enantioselective Aldol Approach O

HRN

OMe

OTMS Zr(Ot-Bu)4(R)-I2-BINOL

PrOHR OMe

OH

NHCOCF3

O

R OMe

OH

NHCOCF3

O

71 - 93%anti:syn up to 92:8

ee 85 - 97%

F3C

OTMS

One of the most direct approaches for the synthesis of β-amino alcohols would be to couple carbonyl functionalities with imine functionalities in a pinacol-type cross coupling. However, achieving both good chemoselectivity and stereoselectivity is a significant synthetic challenge. Recently, this situation has been addressed by the development of a highly diastereoselective pinacol-type cross-coupling between enantiomerically pure N-tert-butylsulfinyl imines and aliphatic aldehydes leading to enantiomerically pure anti-β-amino alcohols in excellent yields (Scheme 5).31 Scheme 5. Diastereoselective SmI2-mediated Pinacol-type Cross Coupling

R1 H

NS

O

R2 H

OSmI2

t-BuOH

THF

-78 °C

R2

R1

OH

HNS

O

HCl

R2R1

OH

NH2

70 - 95%

dr 82:12 - 99:1

ee 95 - 99%

Another approach would be to utilize the stereodirecting effect of a pre-existing stereogenic center. This could be done by nucleophilic additions to α-amino aldehydes, which usually proceed with good diastereoselectivity. A recent example on a divergent protocol for substrate-controlled diastereoselective synthesis of aminodiols based on nucleophilic Mukaiyama aldol additions to α-amino-β-silyloxy aldehydes is depicted in Scheme 6.32 The stereochemical outcome of these reactions is dependent both on the inherent stereochemistry of the aldehyde, i.e. syn or anti, and the substitution pattern on the nitrogen, i.e. chelation- or Cornforth control. Careful analysis of this reaction shows that the outcome cannot be rationalized by the polar Felkin-Anh model. Instead, the revised Cornforth model seems to be more adequate. 31 Zhong, Y.-W.; Dong, Y.-Z.; Fang, K.; Izumi, K.; Xu, M.-H.; Lin, G.-Q. J. Am. Chem. Soc.

2005, 127, 11956-11957. 32 Restorp, P.; Somfai, P. Org. Lett. 2005, 7, 893-895.

9

Scheme 6. Aminodiols from Diastereoselective Mukaiyama Aldol Additions

R1

H

NTsR2

TBSO O

R1

H

NTsR2

TBSO O

Nu:

BF3.Et2O

R1

Nu

NHTs

TBSO OH

R1

Nu

NTsBn

TBSO OH

R1

Nu

NHTs

TBSO OH

R1

Nu

NTsBn

TBSO OH

Chelation-control Felkin-Anh control

R2 = H or Bn

91 - 94%

dr 92:8 - 98:2

81 - 92%

dr >98:2

88 - 89%

dr 88:12 - 98:2

49 - 91%

dr 44:56 - 47:53 The above-mentioned methods for the stereoselective synthesis of β-amino alcohol are just a survey of the recent development in the area but it highlights the importance of this compound class in organic synthesis.

1.4. 1,3-Dipolar Cycloaddition Reactions

The 1,3-dipolar cycloaddition reaction is a five-atom equivalent of the Diels-Alder reaction for the formation of five-membered rings (Scheme 7). The reacting components in the 1,3-dipolar cycloaddition are a 4π-electron component (1,3-dipole or ylide 1) and a 2π-electron component (dipolarophile 2), which react in a [4πs+2πs]-fashion to form five-membered compound 3. The reaction is mechanistically related to the Diels-Alder reaction, in which a four-atom, 4π-electron component (diene 4) and a 2π-electron component (dienophile 2) also react in a [4πs+2πs]-fashion to yield six-membered compound 5 (Scheme 7). The 1,3-dipolar cycloaddition reaction is a powerful method to generate complex five-membered heterocycles containing multiple stereogenic centers from simple starting materials with high level of stereocontrol. Consequently, the reaction is often included as key step in the synthesis of many complex natural products and biologically active compounds.33 Scheme 7. 1,3-Dipolar Cycloaddition and the Diels-Alder Reaction

AB

CC

BA 1,3-Dipolar Cycloaddition

Diels-Alder Reaction

1 2 3

4 2 5 33 For reviews, see: (a) Synthetic Applications of 1,3-Dipolar Cycloaddition Chemistry Toward

Heterocycles and Natural Products; Padwa, A., Pearson , W. H., Eds.; John Wiley & Sons, Inc.: New York, 2002. (b) Chiu, P. Pure Appl. Chem. 2005, 77, 1183-1189. (c) Gothelf, K. V.; Jørgensen, K. A. Chem. Rev. 1998, 98, 863-909.

10

The asymmetric 1,3-dipolar cycloaddition of various 1,3-dipoles and dipolarophiles using different chiral catalysts and auxiliaries has emerged as a relatively new area of research and during the last decade considerable progress has been made, resulting in novel routes to enantiomerically enriched heterocyclic systems.34 Although these results have provided valuable synthetic tools there are still unexplored aspects of 1,3-dipolar cycloaddition reactions.

1.4.1. 1,3-Dipoles The 1,3-dipole, or ylide, is a three-atom, 4π-electron system consisting of two filled and one empty orbital. 1,3-Dipoles can be divided into two main groups with the common structural feature that both have at least one zwitterionic resonance structure where the charges are situated in a 1,3-fashion (Figure 4):35 Propargyl anion-type: 1,3-dipoles in which one of the resonance structures has a double bond on the sextet atom and the other resonance structure has a triple bond on that atom. Allyl anion-type: 1,3-dipoles where one of the resonance structures has a single bond on the sextet atom and the other structure has a double bond.

N NNR N NNR

N NR2C N NR2C

N CR'NR N CR'NR

N CRO N CRO

Azide

Diazoalkane

Nitrile imine

Nitrile oxide

Azomethine imine

Azomethine ylide

Nitrone

Carbonyl ylide

N NR'R2C N NR'R2C

R'' R''

N CR'2R2C N CR'2R2C

R'' R''

N CR2O N CR2O

R' R'

R2C O CR'2 R2C O CR'2

Propargyl Anion-Type Allyl Anion-Type

Figure 4. Common 1,3-Dipoles

1.4.2. Dipolarophiles The dipolarophile is a 2π-electron moiety that readily reacts with 1,3-dipoles in 1,3-dipolar cycloadditions. Most commonly, electron deficient alkenes are used as dipolarophiles, e.g. α,β-unsaturated carbonyl functionalities, but allylic alcohols, allylic halides, vinylic ethers and alkynes are also frequently employed (Figure 5). However, other 2π-electron species such as carbonyl

34 For a recent review on asymmetric 1,3-dipolar cycloadditions, see: Pellissier, H. Tetrahedron

2007, 63, 3235-3285. 35 Smith, M. B.; March, J. March's Advanced Organic Chemistry; 5 ed.; John Wiley & Sons, Inc.:

New York, 2001.

11

groups and imines have also shown to be excellent dipolarophiles when combined with certain 1,3-dipoles.36

R1

O

O

R1R1

OHOMe

R2 R3

O

R2 R3

NR4

X

s-cis

s-trans Figure 5. Various C-C Dipolarophiles Used in 1,3-Dipolar Cycloadditions

1.4.3. Regio- and Stereoselectivity in 1,3-Dipolar Cycloadditions Generally, 1,3-dipolar cycloadditions are regarded as concerted, but not synchronous processes.37 As a consequence, when 1,3-dipoles react with 1,2-substituted alkenes, the two newly formed stereogenic centers are formed in a stereospecific manner due to the syn-attack of the dipole, i.e. a trans-alkene gives rise to a trans-substituted five-membered heterocycle (Scheme 8, path a). In the case of a non-concerted, step-wise mechanism one terminus of the dipole adds first to the trans-alkene dipolarophile, which results in an intermediate where a rotation around the C-C bond in the dipolarophile moiety of the intermediate would yield the isomeric cis-five-membered heterocycle (Scheme 8, path b). Scheme 8. Concerted vs. Non-Concerted 1,3-Dipolar Cycloadditions

AB

C

CB

A

CB

A CB

A CB

A

Rotation around

C-C bond

a

b

a) Concerted mechanism b) Non-concerted mechanism

The transition state in 1,3-dipolar cycloadditions can be rationalized with the FMO-theory. The reaction can either be controlled by HOMOdipole-LUMOdipolarphile interactions or LUMOdipole-HOMOdipolarphile interactions. The 36 (a) Padwa, A.; Weingarten, M. D. Chem. Rev. 1996, 96, 223-269. (b) Mehta, G.; Muthusamy, S.

Tetrahedron 2002, 58, 9477-9504. 37 Huisgen, R. In 1,3-Dipolar Cycloaddition Chemistry; Padwa, A., Ed.; John Wiley & Sons, Inc:

New York, 1984; Vol. 1, pp 1-176.

12

1,3-dipolar cycloaddition reaction has been categorized into three types based on the relative FMO energies of the dipoles and the dipolarophiles.38 For Type I reactions, the dominant interaction is between the HOMO of the dipole and the LUMO of the dipolarophile, e.g. reactions with azomethine ylides and carbonyl ylides. For Type II reactions, the reactants have similar FMO energies and consequently the HOMO-LUMO interactions of both dipole-dipolarophile and dipolarophile-dipole are important. For Type III reactions, the dominant interaction is between the LUMO of the dipole and the HOMO of the dipolarophile. This classification of the 1,3-dipolar cycloaddition reaction is also dependent on the dipolarophile. Substituents on the dipole and the dipolarophile influence and perturb the FMO-energies, which can have a dramatic effect on the reactivity, and the regio- and diastereoselectivity of the reaction. The regioselectivity in the reaction is governed by both steric and electronic effects and addition of dipoles to terminal olefins usually proceeds by addition of the most sterically encumbered dipole functionality to the terminal olefin carbon.39 However, these steric effects can be overruled by strong electronic effects, for example reactions with olefins substituted with a strong electron-withdrawing or electron-donating group. When the electronic properties of the substrates are dictating the selectivity, the atom with the largest FMO coefficient in the dipole will interact with the atom with the largest FMO coefficient in the dipolarophile (Scheme 9).39 Scheme 9. Regioselectivity in the 1,3-Dipolar Cycloaddition

ABC

CB

A

R

R

EDG

C

A

B

C

A

B

EDG

EWGEWG

LUMO HOMO HOMO LUMO

CB

A

R

a)

b) c)

When an allyl anion-type dipole reacts with a dipolarophile the reaction could either go through an endo- or an exo-transition state producing diastereomeric products. In contrast to the Diels-Alder reaction, where the endo-TS can be stabilized by secondary π-orbital interactions, the interaction between the LUMOdipole and the HOMOdipolarophile in the endo-TS is usually small (Type II and III). In the case of a Type I reaction, the secondary π-orbital interaction is

38 (a) Sustmann, R. Tetrahedron Lett. 1971, 12, 2717-2720. (b) Sustmann, R. Pure Appl. Chem.

1974, 40, 569-593. 39 Houk, K. N.; Yamaguchi, K. In 1,3-Dipolar Cycloaddition Chemistry; Padwa, A., Ed.; John

Wiley & Sons, Inc: New York, 1984; Vol. 2, pp 407-450.

13

absent due to the node in the dipole molecular orbital (Scheme 9c). The diastereoselectivity is consequently primarily controlled by steric interactions of the reactants (Scheme 10).40 Scheme 10. Diastereoselectivity in the 1,3-Dipolar Cycloaddition

AB

C

endo-TS

O

AB

C

OHC

AB

C

exo-TS

O

AB

C

OHC

small overlap

CB

A OA

BC

OHC

AB

C

OHC

trans cis

trans cis

1.5. Aim of This Thesis

The aim of this work was to develop new synthetic methods for the synthesis of β-amino alcohols and ultimately apply these methodologies in the total synthesis of amino alcohol-containing natural products (Figure 6). By choosing the 1,3-dipolar cycloaddition as the key transformation both α-hydroxy-β-amino (Chapter 2) and α-amino-β-hydroxy esters (Chapter 3) can be accessed by changing the dipole and dipolarophile components (Scheme 11).

HO C13H27

NH2

OH

D-erythro-Sphingosine (Chapter 4)

N

O

O

O H

H

(-)-Stemoamide (Chapter 5)

NH

O

O

OOH

Omuralide (Chapter 2)

H

Figure 6. Targets in Natural Product Synthesis

Scheme 11. 1,3-DC as Key Transformation for the Synthesis of β-Amino Alcohols

1,3-DCCarbonylYlide

Imine

AzomethineYlide

AldehydeR3O R2

O

OH

NHR1

OR3R2

O

NHR1

OH

Chapter 2Chapter 3

40 (a) Gothelf, K. V.; Hazell, R.; Jørgensen, K. A. J. Org. Chem. 1996, 61, 346-355. (b) Gothelf,

K. V. In Cycloaddition Reaction in Organic Chemistry; Kobayashi, S., Jørgensen, K. A., Eds.; Wiley-VCH Verlag GmbH: Weinheim, 2001, pp 211-247.

14

15

2. 1,3-Dipolar Cycloadditions

of Carbonyl Ylides to Aldimines: Synthesis of syn-α-Hydroxy-β-amino Esters

(Papers I-II)

2.1. Introduction

As the structural complexity of synthetic targets originating from both natural and non-natural sources increases, there is a growing demand for new methodologies for their construction. As such, there is a constant development of new procedures and refinement of already established methodologies to meet these demands. Cycloaddition chemistry has for a long time been one of the key methodologies in modern synthetic organic chemistry for the construction of mono- and polycyclic systems, since it results in increased molecular complexity with high level of stereocontrol from often relatively simple and readily available precursors.

2.1.1. 1,3-Dipolar Cycloadditions of Carbonyl Ylides Carbonyl ylides are highly reactive dipoles that have been used as key intermediates in a variety of reactions since the 1960’s. The structural analysis of a carbonyl ylide reveals that it is a 1,3-dipole (6, Scheme 12) and can therefore undergo different reaction pathways; internal cyclization reactions to form epoxides (7) or concerted rearrangements and internal proton transfers to form substituted ethers (8). However, the synthetically most useful transformation is the 1,3-dipolar cycloaddition reactions where the carbonyl ylides readily react with alkenes or alkynes yielding the corresponding five-membered oxacycles (9).41 Scheme 12. Reactions of Carbonyl Ylides

R1 O R3

R2

R4

O

R4

R3

R2

R1 R1 O R3

R2

R4

X Y

B

O

A

R1

R2

R3

R4

A B

7 6 8

9

X-Y

41 McMills, M. C.; Wright, D. In Synthetic Applications of 1,3-Dipolar Cycloaddition Chemistry

Towards Heterocycles and Natural Products; Padwa, A., Pearson , W. H., Eds.; John Wiley & Sons, Inc.: New York, 2002, pp 253-314.

16

Transition metal-catalyzed diazo decomposition in the presence of aldehydes is an attractive method for the formation of stabilized carbonyl ylides (Scheme 13). Decomposition of diazo compounds leads to the formation of highly electrophilic metallocarbenoids. These electrophilic species can then interact with the nonbonding electrons of a Lewis base. If the Lewis base is an uncharged compound, this interaction leads to the formation of an ylide, which reacts readily with nucleophilic species, such as ethers, amines and halides, as well as aldehydes, esters, ketones and imines.42 Scheme 13. Metal-catalyzed Diazo Decomposition and Carbonyl Ylide Formation

MLn

R2C N2R2C

N2

MLnN2

R2C MLn R2C MLn

O PhO PhR2C

Intramolecular 1,3-dipolar cycloaddition reactions of diazo-derived carbonyl ylides have been used as key-steps in the synthesis of complex natural products by Padwa and others.36 In contrast, the intermolecular 1,3-dipolar cycloadditions of carbonyl ylides have been scarce.43,44 Doyle and co-workers have recently investigated the reactivity of carbonyl ylides and thereby increased the understanding of substituent effects on the product outcome.43d Suga and co-workers have developed the first example of a highly enantioselective 1,3-dipolar cycloaddition of a carbonyl ylide with various dipolarophiles catalyzed by chiral Lewis acids (Scheme 14).44

42 Padwa, A.; Hornbuckle, S. F. Chem. Rev. 1991, 91, 263-309. 43 (a) Huisgen, R.; de March, R. J. Am. Chem. Soc. 1982, 104, 4953-4954. (b) Jiang, B.; Zhang,

X.; Luo, Z. Org. Lett. 2002, 4, 2453-2455. (c) Lu, C.-D.; Chen, Z.-Y.; Liu, H.; Hu, W.-H.; Mi, A.-Q. Org. Lett. 2004, 6, 3071-3074. (d) Russell, A. E.; Brekan, J.; Gronenberg, L.; Doyle, M. P. J. Org. Chem. 2004, 69, 5269-5274.

44 (a) Suga, H.; Inoue, K.; Inoue, S.; Kakehi, A. J. Am. Chem. Soc. 2002, 124, 14836-14837. (b) Suga, H.; Inoue, K.; Kakehi, A.; Shiro, M. J. Org. Chem. 2005, 70, 47-56.

17

Scheme 14. Lanthanide/Pybox-catalyzed Asymmetric Carbonyl Ylide Cycloaddition

OMe

CHN2

O

O

O

OMe

O

O

OMe

O

O

OCH2ArH

Pybox

M(OTf)3O

OMe

OO

OCH2Ar

O

OMe

OO

OCH2Ar

endo exo

Rh(II)

53 - 97%up to 93% ee

2.1.2. Aim of the Study and Synthetic Strategy During the development of a diversity-oriented library synthesis protocol based on β-amino alcohols as scaffolds,45 we became interested in utilizing the intramolecular 1,3-dipolar cycloaddition reactions of azomethine ylides to olefins as a key complexity-generating step (Scheme 15). However, this particular strategy failed and instead we decided to develop the 1,3-dipolar cycloaddition for the synthesis of β-amino alcohols. Scheme 15. Failed DOS-approach based on 1,3-Dipolar Cycloaddition Reactions

R

NHTs

OH

R

TsN

OH

RCM R

TsN

OH

R

TsN

O

O

N R'

n n n

TsN

O NH

RH

OH

n

R'

1,3-DC

Multi-component 1,3-dipolar cycloaddition reactions involving carbonyl43 and azomethine ylides46 as well as ammonium47 and oxonium ylides48 derived from transiently formed carbenoid intermediates have recently received considerable attention since they constitute a powerful tool for the construction of five-membered heterocycles. Surprisingly, to our knowledge very few examples of cycloadditions involving carbonyl ylides and imines have been reported.49,50

45 Torssell, S. Amino Alcohols: Stereoselective Synthesis and Applications in Diversity-Oriented

Synthesis, Licentiate thesis, KTH, Stockholm, 2005, pp 27-34. 46 (a) Hansen, K. B.; Finney, N. S.; Jacobsen, E. N. Angew. Chem. Int. Ed. 1995, 34, 676-678. (b)

Jørgensen, K. A.; Rasmussen, K. G. J. Chem. Soc., Perkin Trans. 1 1997, 1287-1291. (c) Li, G.-Y.; Chen, J.; Yu, W.-Y.; Hong, W.; Che, C.-M. Org. Lett. 2003, 5, 2153-2156. (d) Galliford, C. V.; Beenan, M. A.; Nguyen, S. T.; Scheidt, K. A. Org. Lett. 2003, 5, 3487-3490. (e) Yan, M.; Jacobsen, N.; Hu, W.; Gronenberg, L. S.; Doyle, M. P.; Colyer, J. T.; Bykowski, D. Angew. Chem. Int. Ed. 2004, 43, 6713-6716.

47 Wang, Y.; Zhu, Y.; Chen, Z.; Mi, A.; Hu, W.; Doyle, M. P. Org. Lett. 2003, 5, 3923-3926. 48 (a) Lu, C.-D.; Liu, H.; Chen, Z.-Y.; Hu, W.-H.; Mi, A.-Q. Org. Lett. 2005, 7, 83-86. (b) Huang,

H.; Guo, X.; Hu, W. Angew. Chem. Int. Ed. 2007, 46, 1337-1339. 49 Padwa, A.; Precedo, L.; Semones, M. A. J. Org. Chem. 1999, 64, 4079-4088.

18

With this in mind, we envisioned that amino alcohol 10 could be obtained by hydrolysis of the corresponding oxazolidine 11, which could be accessed by a 1,3-dipolar cycloaddition reaction of carbonyl ylide 12 to imine 13 (Scheme 16). The carbonyl ylide, in turn, is derived from a carbene insertion of 14, generated from diazoester 15 and a metal catalyst, into the lone-pair of aldehyde 16. Scheme 16. Synthetic Strategy

R3O R2

O

OH

NHR1

NR1O

R3O2C R2

R4

R3O

O

M

R2 NR1

R4O

10 11

12

13

16

O R4 14

H3O+ 1,3-DC

carbene

insertion

R3O

O

N2carbene

generation 15

CO2R3

2.2. 1,3-Dipolar Cycloadditions

In our initial investigation, commercially available benzylidene benzylamine (17a) and benzaldehyde (18a) were chosen as the imine and aldehyde components together with Rh2(OAc)4 as catalyst (Scheme 17). Slow addition of ethyl diazoacetate (EDA, 19, 1h at rt) afforded the desired oxazolidine 20a, which upon hydrolysis yielded syn-amino alcohol 21a (dr 93:7) in 82% yield. Scheme 17. 1,3-Dipolar Cycloaddition of Carbonyl Ylides to Imines

N2 PhO

O

EtO Rh2(OAc)4

CH2Cl2NBnO

Ph

EtO2C Ph

p-TSA

Ph

OH

NHBn

17a

MeOHH2OPh NBn

18a

82%dr 93:7

19

20a syn-21a

EtO

O

The relative stereochemistry of β-amino alcohol was confirmed by converting 21a into the corresponding oxazolidinone (22) using standard conditions22 followed by 1H NMR analysis of the vicinal C4-C5 coupling constant (Scheme 18). The coupling constant was J = 5.1 Hz, which is consistent with a 4,5-trans stereochemistry.51

50 After the initial publication of this project (Paper I) two examples of 1,3-DCs of carbonyl ylides

to imines have been reported, see: (a) Muthusamy, S.; Krishnamurthi, J.; Suresh, E. Synlett 2005, 3002-3004. (b) Suga, H.; Ebiura, Y.; Fukushima, K.; Kakehi, A.; Baba, T. J. Org. Chem. 2005, 70, 10782-10791.

51 (a) Bergmeier, S. C.; Stanchina, D. M. Tetrahedron Lett. 1995, 36, 4533-4536. (b) Barrett, A. G. M.; Seefeld, M. A.; White, J. P.; Williams, D. J. J. Org. Chem. 1996, 61, 2677-2685, (c) Lindström, U. M.; Somfai, P. Synthesis, 1998, 109-117, and references cited therein.

19

Scheme 18. Determination of Relative Stereochemistry

NBnO

O

PhEtO2C

Triphosgene

H5

H4i-Pr2NEtJ4,5 = 5.1 Hz

22

Ph

OH

NHBn

syn-21a

EtO

O

With working reaction conditions at hand, it was of interest to optimize the key components of the reaction (Table 1). First, different metal catalysts were evaluated. Rh2(pfb)4, which has previously been used in a variety of carbenoid reactions, gave the desired product albeit with diminished yield and diastereoselectivity (Table 1, entry 2). Cu(OTf)2, commonly used as a suitable catalyst for the decomposition of EDA (19), only led to recovery of 17a (entry 3), the reason probably being coordination of the metal to the basic imine nitrogen with concomitant carbenoid formation inhibition.46a,d,52 In order to investigate if Lewis acid activation of the imine would influence the reaction outcome, addition of Cu(OTf)2 and Yb(OTf)3 to the reaction mixture were screened (entries 4 and 5). In both cases the desired product was obtained in good yield but with diminished diastereoselectivity. This change in reaction outcome indicated that a complexation between the Lewis acid and imine had indeed occurred. Next, a variety of different N-substituted imines were screened. With imine 17b, derived from aniline, together with Rh2(OAc)4 as catalyst, a complex reaction mixture was obtained with no trace of the desired product (entry 6). Attempts with Cu(OTf)2 resulted in the formation of aziridine 23 (2:1 cis:trans selectivity, entry 7) through either an intramolecular ring-closure of the corresponding azomethine ylide (Scheme 19, path a) or a Lewis acid activation of the imine followed by a nucleophilic addition of the diazo ester (path b) with SN2 displacement of N2 in an aza-Darzen-type reaction.46a,52 Attempts with imine 17c only gave recovered starting material when using either Rh2(OAc)4 or Cu(OTf)2 (entries 8 and 9). Scheme 19. Aziridine Formation from Intramolecular Ring-Closure

N2 PhO

O

EtO

17b

Ph NPh

18a19

PhN Ph

EtO

O

Cu(OTf)2

THF

N

Ph

EtO2C Ph

N

Ph

EtO2C Ph

23 (cis:trans 2:1)

NPh

Ph

Cu

O

OEt

N2

Ph

NPh

N2

O

OEt

Cu -N2

Path a

Path b 52 For comparison of Cu(I)- and Rh(II)-catalyzed diazo decomposition in the presence of an imine,

see; Doyle, M. P.; Yan, M.; Hu, W.; Gronenberg, L. S. J. Am. Chem. Soc. 2003, 125, 4692-4693.

20

Table 1. Optimization of 1,3-Dipolar Cycloadditiona

1) Cat.

2) p-TSAPh

OH

NHAr1O

EtO Ph

OH

NHAr1O

EtO

syn-21 anti-21

Ph NAr119

1817

O Ar2

Entry 17 (Ar1) 18 (Ar2) Cat. dr (syn:anti)b Yield 21 (%)c

1 a (Bn) a (Ph) Rh2(OAc)4 93:7 a 82d

2 a (Bn) a (Ph) Rh2(pfb)4 88:12 a 13d

3 a (Bn) a (Ph) Cu(OTf)2 N.A. a 0e

4f a (Bn) a (Ph) Rh2(OAc)4 67:33 a 67 5g a (Bn) a (Ph) Rh2(OAc)4 64:36 a 70 6 b (Ph) a (Ph) Rh2(OAc)4 N.A. b 0h

7 b (Ph) a (Ph) Cu(OTf)2 N.A. b 0i

8 c (4-MeOPh) a (Ph) Rh2(OAc)4 N.A. c 0e

9 c (4-MeOPh) a (Ph) Cu(OTf)2 N.A. c 0e

10 d (Ts) a (Ph) Rh2(OAc)4 n.d. d 19j 11 a (Bn) b (4-MeOPh) Rh2(OAc)4 80:20 a 67

12 a (Bn) c (4-NO2Ph) Rh2(OAc)4 85:15 a 47

a The reaction was carried out with imine 17 (1.0 equiv), aldehyde 18 (1.5 equiv), Rh2(OAc)4 (2.0 mol%) or Cu(OTf)2 (5 mol%) and powdered 4Å MS in CH2Cl2 (Rh-cat.) or THF (Cu-cat.) at rt with addition of ethyl diazoacetate 19 (1.5 equiv) over 1 h. Hydrolysis was performed with p-TSA (2 equiv) in MeOH:H2O (95:5). b Determined by 1H NMR analysis of the crude reaction mixture after hydrolysis. c Isolated yields (syn+anti). d Syn-isomer. e Recovered starting material. f Cu(OTf)2 (5 mol%) added. g Yb(OTf)3 (5 mol%) added. h Complex reaction mixture. i Aziridine 23 formed (cis:trans 2:1). j Ratio 21d:24a = 40:60.

Heteroatom-substituents, e.g. oxime ethers, hydrazones and N-(trimethylsilyl) imines, led only to the formation of the corresponding 1,3-dioxolane 24a via a 1,3-dipolar cycloaddition in which the aldehyde also served as dipolarophile (Eq. 1).43 Apart from N-benzyl substituted imines, only N-tosyl imine 17d afforded the desired product, although only in minute quantities along with substantial amounts of dioxolane 24a (entry 10).

OO

Ph

Ph CO2Et

24a (cis:trans)

Rh2(OAc)4

17

Ph N 1918a

CH2Cl2

X

X = OBn, NPh2, TMS

17

Ph NX

(1)

21

Replacing benzaldehyde (18a) for other aromatic aldehydes (4-MeO-Ph and 4-NO2-Ph) did not improve the reaction outcome (entries 11 and 12).

Rh2(OAc)417a 19

18CH2Cl2

(2)OO

Ar

Ar CO2Et

24 (cis:trans)

Ar O

Ar = Ph, 4-MeO-Ph 20:24 >98:2

Ar = 4-NO2-Ph 20:24 60:40

NBnO

Ar

EtO2C Ph

20

Interestingly, when 4-NO2-benzaldehyde was used as the aldehyde component significant amounts of dioxolane 24 were formed (entry 12, Eq. 2). In this case the electron-deficient 4-NO2-benzaldehyde also served as dipolarophile. This side product was not detected in the case of benzaldehyde or 4-MeO-benzaldehyde and consequently the remaining experiments were performed with aldehyde 18a and benzyl-substituted imines.

The experimental outcome using different N-benzyl imines under the optimized reaction conditions is summarized in Table 2. In all cases, the reactions proceeded cleanly to provide the desired syn-α-hydroxy-β-amino esters in high yield and excellent diastereoselectivity.53 Several benzylidene benzylamine derivatives containing electron withdrawing (entries 2-5) or donating substituents (entries 6-8) in meta- or para-position gave comparable yields and diastereoselectivities irrespective of the steric and electronic properties of the aryl substituent. Also the furfural-derived imine 17l afforded the corresponding product 21l in high yield and diastereoselectivity (entry 9), which is of interest since the furan moiety can be readily derivatized into several useful functional groups (Scheme 20).54,55 The reaction of ethyl glyoxalate imine 17m gave syn-β-hydroxyaspartate 21m, a potent blocker of the glutamate transporters,56 in high yield and good diastereoselectivity (entry 10).

53 The syn- and anti-diastereomers where readily separated with flash chromatography for all

substrates. 54 For examples of oxidative cleavage of the furan moiety, see: Aggarwal, V. K.; Vasse, J.-L. Org.

Lett. 2003, 5, 3987-3990 and references therein. 55 For examples of the aza-Achmatowicz reaction, see: Haukaas, M. H.; O'Doherty, G. A. Org.

Lett. 2001, 3, 401-404 and references therein. 56 Shimamoto, K.; Shigeri, Y.; Yasuda-Kamatani, Y.; Lebrun, B.; Yumoto, N.; Nakajima, T.

Bioorg. Med. Chem. Lett. 2000, 10, 2407-2410.

22

Table 2. 1,3-Dipolar Cycloadditions to Various Aldiminesa

1) Rh2(OAc)4

2) p-TSAR

OH

NHBnO

EtO R

OH

NHBnO

EtO

syn-21 anti-21

R NBn18a 19

17

Entry 17 (R) dr (syn:anti)b Yield syn-21 (%)c

1 a (Ph) 93:7 a 82 2d e (4-NO2C6H4) 91:9 e 61 3 f (4-ClC6H4) 98:2 f 75 4 g (4-FC6H4) 97:3 g 78 5 h (3-MeOC6H4) 98:2 h 77 6 i (4-MeC6H4) 97:3 i 87 7d j (4-MeOC6H4) 94:6 j 78 8 k (2-Naphthyl) 98:2 k 83 9d,e l (2-Furyl) 92:8 l 75

10d,e m (CO2Et) 83:17 m 64 a The reaction was carried out with imine 17 (1.0 equiv), benzaldehyde 18a (1.5 equiv), Rh2(OAc)4 (2.0 mol%) and powdered 4Å MS in CH2Cl2 at rt with addition of ethyl diazoacetate 19 (1.5 equiv) over 1 h. Hydrolysis was performed with p-TSA (2 equiv) in MeOH:H2O (95:5). b Determined by 1H NMR analysis of the crude reaction mixture after hydrolysis, syn- and anti-diastereomers separable with flash chromatography. c Isolated yield. d 10 h addition time. e Reaction performed at 0 °C.

Scheme 20. Possible Transformations of Furan-substituted Amino Alcohol

O

NHPg

OPg

O

EtO

PgN

OPg

EtO

O

O

OH

H

NHPg

OPg

O

EtOOH

O

Aza-AchmatowiczOxidative cleavage

m-CPBARuCl3 (cat), NaIO4

syn-21l

During the last couple of years it has been recognized that the diazoacetate structure has a large impact on the reaction outcome.57 Most pronounced effect was found for diazoacetates functionalized with electron-donating groups, e.g. vinyl and aryl. It has been shown independently by the groups of Davies58 and Doyle59 that Rh-catalyzed decomposition of donor/acceptor diazoacetates in

57 Doyle, M. P.; McKervey, M. A.; Ye. T. Modern Catalytic Methods for Organic Synthesis with

Diazo Compounds, Wiley-Interscience, New York, 1998. 58 Davies, H. M. L.; DeMeese, J. Tedrahedron Lett. 2001, 42, 6803-6805. 59 (a) Doyle, M. P.; Hu, W.; Timmons, D. J. Org. Lett. 2001, 3, 933-935. (b) Doyle, M. P.; Hu,

W.; Timmons, D. J. Org. Lett. 2001, 3, 3741-3744.

23

the presence of aldehydes leads to the formation of epoxides and dihydrofurans through intramolecular ring closure of the transient carbonyl ylide. This is in contrast to the behavior of EDA, which forms 1,3-dioxolanes through 1,3-dipolar cycloadditions. The reaction between donor/acceptor diazoacetates 25 and 26 with imine 17a and aldehyde 18a gave, unfortunately, no traces of the desired products (Eq. 3 and 4); instead epoxide 27 (from 25) and a mixture of epoxide 28 and dihydrofuran 29 (from 26) were isolated in comparable yields and selectivities as previously reported.58,59

17a 18a

Ph CO2Me

N2

25

O

Ph CO2Me

Ph

27

17a 18a

CO2Me

N2

26

O

CO2Me28

Ph Ph

Ph

OPh

Ph

CO2Me

29

dr 89:11

dr >98:2 dr >98:2

Rh2(OAc)4

CH2Cl2

Rh2(OAc)4

CH2Cl271%28:2970:30

75%

(3)

(4)

2.3. Mechanistic Aspects of the 1,3-Dipolar Cycloaddition

Mechanistically, the reaction is believed to proceed through a chemoselective insertion of the metallocarbene into the lone-pair of benzaldehyde (18a) forming either a metal-free43,60a (30a) or a metal-associated ylide60 (30b, Scheme 21). Scheme 21. Metal-Free and Metal-Associated Ylides

Ph O CO2Et

HMLn

OPh

CO2Et

S-shaped metal-free ylide 30a

metal-associatedylide 30b

LnMOEt

O

OPh and/or

18a metallocarbene

If the metal-free ylide 30a is formed, it has been suggested that the reaction could occur via the thermodynamically most stable S-shaped ylide,60a which then reacts with the imine in an endo-selective cycloaddition favoring the

60 For the discussion about metal-associated ylides, see: (a) Doyle, M. P.; Forbes, D. C.;

Protopopova, M. N.; Stanley, S. A.; Vasbinder, M. M.; Xavier, K. R. J. Org. Chem. 1997, 62, 7210-7215. (b) Kitagaki, S.; Anada, M.; Kataoka, O.; Matsuno, K.; Umeda, C.; Watanabe, N.; Hashimoto, S.-I. J. Am. Chem. Soc. 1999, 121, 1417-1418. (c) Hodgson, D. M.; Pierard, F. Y. T. M.; Stupple, P. A. Chem. Soc. Rev. 2001, 30, 50-61. (d) Hodgson, D. M.; Labande, A. H.; Pierard, F. Y. T. M.; Expósito-Castro, M. Á. J. Org. Chem. 2003, 68, 6153-6159.

24

formation of trans-oxazolidine 20 (Scheme 22).61 The endo-transition state is favored over the exo-transition state due to the steric interactions between the ester functionality on the ylide and the benzylidene substituent on the imine. Scheme 22. Proposed Stereochemical Model for 1,3-DCs of Metal-Free Ylides

endo-TS (Favored)

BnN

OPh

R

EtO2C NBnO

Ph

EtO2C R

trans-20

OPh

R

BnNEtO2C

exo-TS (Disfavored)

BnN

OPh

EtO2C NBnO

Ph

EtO2C R

cis-20

OPh

BnNEtO2C

R

StericInteraction

R

Major

Minor

In the case of metal-associated ylide 30b, the choice of catalyst influences the stereochemical outcome of the reaction, since the catalyst is associated in the transition state and is in close proximity to the forming C-C bond. The only proposed mechanistic pathway for 1,3-dipolar cycloadditions of metal-associated carbonyl ylides deals with their reaction with aldehydes (Scheme 23).60a,c A stepwise mechanism is suggested where a nucleophilic attack of the aldehyde lone-pair on ylide 30b is followed by orientation of the aldehyde in such way that the steric interaction between the aldehyde substituent (Ar) and the carbenic proton is minimized. Catalyst decomplexation and ring-closure then affords cis-dioxolane 24 as the major product. Scheme 23. Proposed Mechanism for Metal-associated Carbonyl Ylides

Ar O CO2Et

HMLn

Ar O CO2Et

HMLn

O

Ar

HO

O

Ar

CO2EtAr

30b cis-24

ArCHO

To address whether the carbonyl ylide involved in this reaction is metal-associated or not, its generation and subsequent reaction with imine 17a was performed using different Rh(II)-catalysts (vide infra).

61 It should be noted that it is not the difference in free-energy between the ylide conformers (ΔG)

that dictates which conformer that will react. It is the difference in transition state energies (ΔG*), according to the Curtin-Hammett principle.

25

2.4. Asymmetric 1,3-Dipolar Cycloadditions

2.4.1. Chiral Rh(II) Catalysts In order to determine if the cycloaddition proceeds via metal-free ylide 30a or metal-associated ylide 30b several different Rh(II) catalysts were screened. If 30b is an intermediate it can be expected that different catalysts would result in different stereochemical outcomes. Also, the employment of chiral Rh(II) catalysts provides the opportunity to investigate the catalytic enantioselective 1,3-dipolar cycloaddition.

Rh

O

Rh

O

N

4

O

O O O

Rh

Rh

OO

O OO O

N

N

H H

H H

SO2Ar

SO2Ar

N

NArO2SSO2Ar

HH

H H

33

Rh2(S-biTISP)2Ar = 2,4,6-triiPrC6H2

Rh

O

Rh

O

NArO2SH

4

31

Rh2(S-DOSP)4Ar = C6H4C11H23

32

Rh2(S-BPTV)4

Figure 7. Chiral Rh(II)-Catalysts

When exchanging Rh2(OAc)4 for Rh2(pfb)4, 21a was obtained in only 13% yield with decreased diastereoselectivity (syn:anti 88:12, compare Table 2, entries 1 and 2). Employing the commercially available chiral Rh2(S-DOSP)4 catalyst (31, Figure 7) under standard reaction conditions, amino alcohol 21a was isolated in good yield with excellent diastereoselectivity and modest enantioselectivity (Table 3, entry 1). Catalyst 31 has been reported to give best results in hydrocarbon solvents and when the reaction was performed in hexane instead of CH2Cl2 the enantiomeric ratio increased (82:18) although at the expense of diminished yield and diastereoselectivity (entry 2).62 The same observations were made when Et2O was used as solvent (entry 3). Unfortunately, both the Hashimoto catalyst60b (Rh2(S-BPTV)4, 32) and the second generation Davies catalyst63 (Rh2(S-biTISP)2, 33), which have been reported to give excellent results in asymmetric cyclopropanations conducted in CH2Cl2, failed to give any improved results (entries 4 and 5).

62 Davies, H. M. L.; Bruzinski, P. R.; Lake, D. H.; Kong, N.; Fall, M. J. J. Am. Chem. Soc. 1996,

118, 6898-6907. 63 (a) Davies, H. M. L.; Panaro, S. A. Tetrahedron Lett. 1999, 40, 5287, 5290. (b) Davies, H. M.

L. Eu. J. Org. Chem. 1999, 2459-2469.

26

Table 3. 1,3-Dipolar cycloadditions using chiral Rh(II) catalystsa

1) Rh(II) cat

2) p-TSAPh

OH

NHBnO

EtO17a 18a 19 Ph

OH

NHBnO

EtO

syn-21a anti-21a

Entry Cat. Solvent Cond.b

[°C / h]

drc

(syn:anti)

erd

(syn)

Yielde

syn-21a (%)

1 31 CH2Cl2 rt / 1 94:6 62:38 62 2 31 Hexane rt / 1 71:29 82:18 17 3 31 Et2O rt / 1 78:22 81:19 14 4 32 CH2Cl2 0 / 10 90:10 52:48 56 5 33 CH2Cl2 rt / 10 n.d. n.d. <5g

a The reaction was carried out with imine 17a (1.0 equiv), benzaldehyde 18a (1.5 equiv), chiral Rh(II) cat (2.0 mol%) and powdered 4Å MS in CH2Cl2 with addition of ethyl diazoacetate 19 (1.5 equiv). b Reaction temperature and addition time. c Determined by 1H NMR analysis of the crude reaction mixture. d Determined by chiral HPLC (Chiracel OD). e Isolated yield. g 24a major product.

Although it has been suggested that carbonyl ylides derived from benzaldehyde and EDA using Rh(II) catalysts would generate a metal-free ylide (30a),60a these results clearly supports the formation of a metal-associated ylide (30b). This is also the first example of an enantioselective intermolecular 1,3-dipolar cycloaddition of carbonyl ylides to imines catalyzed by chiral Rh(II) carboxylates.

2.4.2. Chiral Diazo Esters It was also of interest to develop an asymmetric protocol based on a chiral auxiliary for the synthesis of enantiomerically pure syn-α-hydroxy-β-amino esters and ultimately apply it to the synthesis of biologically active natural products. Several different approaches based on chiral auxiliaries can be envisioned but the use of chiral diazoesters has the advantage of allowing the use of the same chiral substrate for different imines. Also, the chiral auxiliary is easily cleaved off and recovered after the reaction by either ester hydrolysis/reduction or a transesterification reaction. (−)-8-Phenylmenthyl diazoacetate (34) has successfully been employed as the carbene precursor in asymmetric cyclopropanation reactions.64 Unfortunately, the reaction of 34 with 17a and 18a only gave unidentified byproducts (Scheme 24).

64 Bedekar, A. V.; Koroleva, E. B.; Andersson, P. G. J. Org. Chem. 1997, 62, 2518-2526.

27

Scheme 24. Asymmetric 1,3-Dipolar Cycloaddition Using 8-Ph-Menthyl Diazoester

O

Ph

N2

O

34

17a 18aRh2(OAc)4

CH2Cl2

No desiredproduct

The chiral diazoacetate 35, derived from (R)-pantolactone65 has been reported to give good results when used in asymmetric cyclopropanations66 and aziridinations.67 When subjected to the standard reaction conditions only a modest asymmetric induction was obtained (Scheme 25). The desired amino alcohol 36 was isolated in high yield as a 60:40 mixture of two diastereomers (absolute configuration was not determined).68 In order to determine the relative stereochemistry of the product, the mixture was subjected to a Ti(OiPr)4-catalyzed transesterification protocol,69 which gave the known amino alcohol syn-21a as single diastereomer indicating that the relative stereochemistry of 36 was syn. The enantiomeric ratio of syn-21a (er 60:40)70 was also in good agreement with the diastereomeric ratio of 36 (dr 60:40). Scheme 25. Asymmetric 1,3-Dipolar Cycloadditions Using Pantolactone Diazo Ester

O

ON2

OO

O

O

OO

Ph

OH

NHBn

36

dr 60:40

O

EtO Ph

OH

NHBn

17a 18a

syn-21aer 60:40

35

1) Rh2(OAc)4 CH2Cl2, 0 °C

2) p-TSA MeOH:H2O 82%

Ti(OiPr)4, EtOH

PhMe, 110 °C60%

This result indicates that the inherently high diastereoselectivity (syn:anti) in the reaction with the achiral imines was maintained, but the discrimination between the enantiotopic faces of the imine was low, most likely due to the distal location of the pantolactone moiety.

2.4.3. Chiral Imines Since the use of chiral diazoesters only met with limited success we became interested in investigating the approach based on using chiral imines, a strategy that has previously been applied in cycloaddition chemistry.71 Initially we

65 Corey, E. J.; Myers, A. G. Tetrahedron Lett. 1984, 25, 3559-3562. 66 Doyle, M. P.; Protopopova, M. N.; Brandes, B. D.; Davies, H. M. L.; Huby, N. J. S.; Whitesell,

J. K. Synlett 1993, 151-153. 67 Akiyama, T.; Ogi, S.; Fuchibe, K. Tetrahedron Lett. 2003, 44, 4011-4013. 68 Only two out of four possible diastereomers were formed; syn:anti >95:5 (no anti-isomer

detected by 1H NMR analysis of crude reaction mixture) 69 Barrett, A. G. M.; Lebold, S. A. J. Org. Chem. 1991, 56, 4875-4884. 70 Determined by chiral HPLC, Chiracel OD column, 15% isopropanol in hexanes. 71 Jørgensen, K. A. Angew. Chem. Int. Ed. 2000, 39, 3558-3588.

28

decided to use (+)-α-methylbenzylamine, which has the advantage of being cheap, commercially available in both enantiomeric forms and easily removable under standard hydrogenolytic conditions. Consequently, chiral imine 37a, derived from (+)-α-methylbenzylamine, was reacted using standard conditions, which led to the formation of amino alcohol 38a in high yield and diastereoselectivity (8.2:1:1:0) (Table 4, entry 1). The major diastereomer was readily separated from the two minor diastereomers using flash chromatography. The relative stereochemistry of the major product was determined to be syn by conversion into the corresponding oxazolidinone followed by 1H NMR analysis of the relevant coupling constants.51

Table 4. Optimization of 1,3-Dipolar Cycloadditions Using Chiral Iminesa

1) Rh2(OAc)4

2) p-TSAPh

OH

HNO

EtOPh N18aXC diastereomers

XC

19

37 38

Entry 37 (XC) Lewis acid drb Yield 38 (%)c

1 a none 8.2:1:1:0 a (77)

2 b none n.d. b (traces)

3 c none N.A. c (0)d

4 d none N.A. c (0)d

5 d Yb(OTf)3 3.6:3:1.7:1 d (75)e

6 d Zn(OTf)2 3:2.7:1.9:1 d (93)e a The reaction was carried out with imine 37 (1.0 equiv), benzaldehyde 18a (1.5 equiv), Rh2(OAc)4 (2.0 mol%) and powdered 4Å MS in CH2Cl2 at 0 °C with addition of ethyl diazoacetate 19 (1.5 equiv) over 10 h. b Determined by 1H NMR analysis of the crude reaction mixture. c Isolated yield (all isomers). d 24a major product. e Isomers not separable with flash chromatography.

Several other commonly used chiral imines were also screened in order to improve the results, but this unfortunately met with no success (Table 4). When chiral imine 37d, derived from phenylalanine, was employed only dioxolane 24a could be detected with no trace of the desired product. Interestingly, when a Lewis acid was added to the reaction mixture, amino alcohol 38d was isolated as the sole product in high yield although as an inseparable mixture of all four possible diastereomers (entries 5 and 6) presumably due to an activation of the imine.

Ph

Nap

StBu

O

Ph

OMe

Ph

OMe

Ph

OMe

29

The outcome of different imines derived from (+)-α-methylbenzylamine is summarized in Table 5. For all aromatic imines the reaction proceeded cleanly to give the desired syn-product in high yield and diastereoselectivity (entries 1-6), and in all cases the major isomer was separable from the two minor isomers using flash chromatography. It should be noted that the reaction also proceeded cleanly when conducted on multi gram-scale (2-4 g of imine ent-37f, entry 4). Furthermore, chiral imine 37j, derived from ethyl glyoxalate, gave amino alcohol diester 38j in good yield and diastereoselectivity (entry 8). However, 2-furylimine 37i gave the corresponding cycloaddition products in low selectivity (entry 7).

Table 5. 1,3-Dipolar Cycloaddition to (+)-α-Methylbenzyl Iminesa

1) Rh2(OAc)4

2) p-TSAR

OH

HNO

EtOR N18a Ph diastereomers

Ph

19

37 38

Entry 37 (R) drb Yield 38 (%)c

1 a (Ph) 8.2:1:1:0 a (77) 2 e (4-MeO-Ph) 5.4:1:1:0 e (86) 3 f (3-MeO-Ph) 7.1:1:1:0 f (87) 4d ent-f (3-MeO-Ph) 8.3:1:1:0 ent-f (82) 5 g (4-F-Ph) 4.5:1:1:0 g (71) 6 h (4-Br-Ph) 8.4:1:1:0 h (62) 7e i (2-furyl) 1.8:1.6:1:0 i (67) 8e j (CO2Et) 9.3:1.8:1:0 j (58)

a The reaction was carried out with imine 37 (1.0 equiv), benzaldehyde 18a (1.5 equiv), Rh2(OAc)4 (2.0 mol%) and powdered 4Å MS in CH2Cl2 at 0 °C with addition of ethyl diazoacetate 19 (1.5 equiv) over 10 h. b Determined by 1H NMR analysis of the crude reaction mixture. c Isolated yield (all isomers), major isomer separable from minor isomers with flash chromatography. d Reaction performed using 2.0 g of (-)-imine. e Reaction performed at -10 °C.

2.4.4. Chiral Lewis Acids Since it was noted that Lewis acid activation of the imine-component had a significant impact on the stereochemical outcome of the cycloaddition reactions using both achiral and chiral imines (Table 1, entries 4 and 5; Table 4, entries 5 and 6) a natural continuation of our development of asymmetric 1,3-dipolar cycloaddition reactions would consequently involve screening of chiral Lewis acids.

30

The metal salts used as Lewis acids were chosen due to their properties as being active and aldimine selective (Cu(OTf)2, Sc(OTf)3 and Yb(OTf)3).72 Zn(OTf)2 is reported as being more neutral in its character but the basic OTf−-ligands is believed to increase the aldimine-selectivity.72 The chiral ligands used in this screen were chosen since they are most commonly used in conjunction with the above-mentioned metal-salts and the chiral complexes were prepared according to published procedures (Figure 8, 39-Cu(II),73a -Zn(II);73b 40-Yb(III),44b -Sc(III)44b)

N

N

OO

NN

OO

N

39

(R,R)-t-Bu-BOX40

(R,R)-i-Pr-PyBOX Figure 8. Chiral Ligands

Very disappointingly, in all cases syn-amino alcohol 21a was isolated in low yield and diastereoselectivity along with no significant enantioselectivity (Table 6, entries 1-5). Table 6. 1,3-Dipolar Cycloadditions using Chiral Lewis Acidsa

1) Rh2(OAc)4

2) p-TSAPh

OH

NHBnO

EtO17a 18a Ph

OH

NHBnO

EtO

syn-21a anti-21a

chiral LA19

Entry Lewis acid Cond.b

[°C / h]

drc

(syn:anti)

erd

(syn)

erd

(anti)

Yielde

syn-21a (%)

1 39/Cu(OTf)2 20 / 1 80:20 51:49 53:47 57 2 39/Cu(OTf)2 -10 / 1 90:10 51:49 n.d. 52 3 39/Zn(OTf)2 20 / 1 91:9 52:48 n.d. 68 4 40/Yb(OTf)3 -15 / 1 82:18 rac 54:46 36 5 40/Sc(OTf)3 -15 / 1 78:22 rac 53:47 34 a The reaction was carried out with imine 17a (1.0 equiv), benzaldehyde 18a (1.5 equiv), Rh2(OAc)4 (2.0 mol%), chiral Lewis acid (10 mol%) and powdered 4Å MS in CH2Cl2 with addition of ethyl diazoacetate 19 (1.5 equiv). b Reaction temperature and addition time. c Determined by 1H NMR analysis of the crude reaction mixture. d Determined by chiral HPLC (Chiracel OD). e Isolated yield.

72 Kobayashi, S.; Busujima, T.; Nagayama, S. Chem. Eur. J. 2000, 6, 3491-3494. 73 (a) Evans, D. A.; Peterson, G. S.; Johnson, J. S.; Barnes, D. M.; Campos, K. R.; Woerpel, K. A.

J. Org. Chem. 1998, 63, 4541-4544. (b) Gothelf, A. S.; Gothelf, K. V.; Hazell, R. G.; Jørgensen, K. A. Angew. Chem. Int. Ed. 2002, 41, 4236-4238.

31

2.5. Synthetic Applications: Paclitaxel Side-Chain and Towards the Total Synthesis of Omuralide

Having developed a synthetically viable asymmetric 1,3-dipolar cycloaddition of carbonyl ylides to chiral imines for the synthesis of enantiomerically pure syn-α-hydroxy-β-amino esters the ultimate goal was to apply this methodology to the total synthesis of biologically active natural products.

2.5.1. Paclitaxel Side-Chain The utility of the present methodology was first exemplified by a short asymmetric synthesis of the C12 side-chain of paclitaxel (41), which has previously been shown to be an important pharmacophore for the anti-tumor activity of paclitaxel (Figure 9).74

12

O

HO

OH

O

O

BzOH

AcO

Ph

OH

ONHPh

O

OAc

Figure 9. Paclitaxel (Taxol®)

Compound 38a, readily isolated from the two minor isomers, was subjected to standard hydrogenolysis conditions followed by a N-selective benzoylation to give amide 42 (77% yield, 2 steps, Scheme 26). Finally, ester hydrolysis using LiOH yielded the paclitaxel side-chain 41 as a white solid in 42% overall yield over four steps. Analytic data of 41 were in all aspects identical to those previously reported.26,54,75 This also confirmed the absolute stereochemistry of 38a, obtained from (R)-imine 37a, to be (2R, 3S).

Scheme 26. Asymmetric Synthesis of C13 Side-Chain of Paclitaxel

EtO Ph

OH

HNO Ph

EtO Ph

OH

HNO Ph

O

HO Ph

OH

HNO Ph

O

38a 42 41

1) H2, Pd(OH)2/C

EtOH, HCl

2) PhCOCl, NaHCO3 EtOAc, 0 °C 77% (2 steps)

LiOH.H2O

THF:MeOH:H2O89%

74 (a) Guenard, D.; Gueritte-Voegelein, F.; Potier, P. Acc. Chem. Res. 1993, 26, 160-167. (b)

Kingston, D. G. I. Pure Appl. Chem. 1998, 70, 331-334. (c) Kingston, D. G. I. J. Nat. Prod. 2000, 63, 726-734.

75 For the shortest synthesis of 41 see ref 26 as well as: (a) Deng, L.; Jacobsen, E. N. J. Org. Chem. 1992, 57, 4320-4323. (b) Gou, D.-M.; Liu, Y.-C.; Chen, C.-S. J. Org. Chem. 1993, 58, 1287-1289.

32

2.5.2. Omuralide: Introduction Lactacystin (43, Figure 10) was isolated by Omura and co-workers in 1991 from the cultured broth of Streptomyces I sp. OM-6519, and the structure and absolute configuration were determined by NMR and X-ray crystallographic analyses.76 The biological effects of lactacystin consist of its ability to induce differentiation in cultured neuronal cells due to inhibition of mammalian 20S and 26S proteosomes.76 The 20S proteasome is a proteinase complex that constitutes the catalytic core of the 26S proteasome. It is essential for the normal turnover of cellular proteins and for removal of damaged and misfolded proteins. It has also an important role in controlling cell growth.77 Examples of selective proteasome inhibitors are depicted in Figure 10.

32

NH4

O

O

O

6

OH

NH

O

OHHOSO

NHAc

CO2HLactacystin (43) Omuralide (44)

NH

O

O

OOH

Salinosporamide A (47)

H

Cl

Figure 10. Potent and Selective Proteasome Inhibitors

Important work by Corey, Schreiber and co-worker has shown that lactacystin acts as a proinhibitor of the proteasome. An in vivo elimination of N-acetyl-L-cysteine forms omuralide (clasto-lactacystin β-lactone, 44, Scheme 27), which is capable of permeating cell membranes. Inhibition then occurs through acylation of the N-terminus threonine residues of the proteasome.78 Scheme 27. 20S Proteasome Inhibition

NH

O

O

OOH

NH

O

OH

NH

O

OHHO HOSO OO

NHAc

CO2H

Inactivated proteasome

43 44

N-acetylcysteine

HN

O NH

NH

O

NH

OH

Thr 1

76 (a) Omura, S.; Fujimoto, T.; Otoguro, K.; Matsuzaki, K.; Moriguchi, R.; Tanaka, H.; Sasaki, Y.

J. Antibiot. 1991, 44, 113-116. (b) Omura, S.; Matsuzaki, K.; Fujimoto, T.; Kosuge, K.; Furuya, T.; Fujita, S.; Nakagawa, A. J. Antibiot. 1991, 44, 117-118.

77 DeMartino, G. N.; Slaughter, C. A. J. Biol. Chem. 1999, 274, 22123-22126. 78 (a) Fenteany, G.; Standaert, R. F.; Lane, W. S.; Choi, S.; Corey, E. J.; Schreiber, S. L. Science

1995, 268, 726-731. (b) Groll, M.; Ditzel, L; Löwe, J.; Stock, D.; Bochtler, M.; Bartunik, H. D.; Huber, R. Nature, 1997, 386, 463-471.

33

Due to their interesting biological properties and complex structural features many total syntheses of lactacystin and omuralide have been disclosed to present date and the first total synthesis of lactacystin was achieved in 1992 by E. J. Corey and co-workers.79 The Donohoe-group has reported an elegant synthesis of racemic omuralide where they utilized their newly developed reductive aldol methodology.80 Scheme 28. Reductive Aldol: Keystep in the Donohoe Synthesis of Omuralide

NBoc

CO2Et

1) Li, DBB, THF

2) MgBr2

3) O

H

74%one pot NBoc

HOEtO2C

46 dr > 20:1

O

H

N

OEtO

O

OtBu

Mg

NPg

HO2C

O

OHHO

NPg

RO2C

O

OPgHO

MeONH

2

44

45 The keystep in Donohoe’s synthesis was the reductive aldol reaction where treatment of N-Boc pyrrole 45 with Li/DBB followed by addition of MgBr2 and isobutyraldehyde gave the desired anti-aldol product 46 in excellent selectivity (Scheme 28). The reaction is believed to proceed via a Z-enolate by complexation of the metal-enolate to the N-Boc group. Structure-activity relationship (SAR) studies have shown that the requirements for successful 20S proteasome inhibition are rather strict.81 The β-lactone moiety is an absolute requirement and the stereochemistry at the C2, C3 and C6 position cannot be altered without losing biological activity (44, Figure 9). The isopropyl group attached at the C6 position is important but can for example be exchanged for a cyclohexenyl group (salinosporamid A, 47, Figure 10).82 The methyl group at C4 is the only completely replaceable group and most synthetic approaches have introduced the C4 methyl in an early stage.83 With this in mind and on the basis of the Donohoe reductive anti-aldol reaction we constructed a synthetic route to address these issues; a late stage introduction of both the C4 methyl and the isopropyl residue at C6. 79 Corey, E. J.; Reichard, G. A. J. Am. Chem. Soc. 1992, 114, 10677-10678. 80 Donohoe, T. J.; Sintim, H. O.; Sisangia, L.; Harling, J. D. Angew. Chem. Int. Ed. 2004, 43,

2293-2296; Donohoe, T. J.; Sintim, H. O.; Sisangia, L.; Ace, K. W.; Guyo, P. M.; Cowley, A.; Harling, J. D. Chem, Eur. J. 2005, 11, 4227-4238.

81 (a) Corey, E. J.; Li, W.-D. Z. Tetrahedron Lett. 1998, 39, 7475-7478. (b) Corey, E. J.; Li, W.-D. Z. Tetrahedron Lett. 1998, 39, 8043-8046. (c) Corey, E. J.; Li, W.-D. Z.; Nagamitsu, T.; Fenteany, G. Tetrahedron, 1999, 55, 3305-3316.

82 (a) Feling, R. H.; Buchanan, G. O.; Mincer, T. J.; Kauffman, C. A.; Jensen, P. R.; Fenical, W. Angew. Chem. Int. Ed. 2003, 42, 355-357; (b) Drahl, C.; Cravatt, B. F.; Sorensen, E. J. Angew. Chem. Int. Ed. 2005, 44, 5788-5809.

83 Soucy, F.; Grenier, L.; Behnke, M. L.; Destree, A. T.; McCormack, T. A.; Adams, J.; Plamondon, L. J. Am. Chem. Soc. 1999, 121, 9967-9976.

34

2.5.3. Retrosynthetic Analysis of Omuralide We envisioned that omuralide (44) could be assembled using our asymmetric 1,3-dipolar cycloaddition of carbonyl ylides and α-methylbenzyl imines providing syn-amino alcohol ent-38f (Scheme 29). A Birch reduction of the 3-methoxyphenyl-moiety followed by ozonolysis would give β-keto ester 48, which then be transformed into γ-lactam 49 by a stereoselective ketone reduction/lactamization sequence. The β-lactone moiety (50) can be constructed using known procedures, which sets the stage for a late stage introduction of both the C4 methyl stereochemistry and the C6 isopropyl residue using a stereoselective aldol reaction. Scheme 29. Retrosynthetic Analysis of Omuralide

OEtMeO

NH

OH

O

O O

MeO

NHPg

NPg

O

HO

O

OH

NH

O

O

OOH

NPg

O

O

O

Stereoselective reduction

Aldol

Olefination

LactonizationOx. cleavage

Lactamization

Reduction

OH

OH

BirchOzonolysis 1,3-DC

N2 PhO

O

EtO

MeON Ph

Ph

44 50 49

48 ent-38f Another important precedence that supports the late stage aldol strategy in our retrosynthetic analysis is the report on diastereoselective aldol reactions of β-lactone enolates to aldehydes, which proceeds in high yield and diastereoselectivity (Scheme 30).84 Scheme 30. Diastereoselective Aldol Additions of β-Lactone Enolates to Aldehydes

R OR'

OR''

Li

OH

OO

R'R''

R

OH

Yield: 85-95%dr > 85:15

O

R'' R'

O1) LDA

2) RCHO

Blocks additionfrom top-face

A potential side reaction when treating a β-lactone with a strong base would be the elimination to the corresponding acrylic acid but it has been shown that β-lactone enolates are surprisingly stable at low temperature and the

84 Mulzer, J.; Chucholowski, A. Angew. Chem. 1982, 94, 787-788.

35

elimination occurs only slowly at room temperature. This observation has been explained by the rigidity of the β-lactone enolate, which forces the orbitals α and β in a complete orthogonal orientation to each other that suppresses elimination (Figure 11, compare with anti- and syn-elimination).85

!

"

!

O

orthogonal

"

!

O X

"

X

anti syn

Figure 11. Orthogonal Orbital Orientation in the β-Lactone Enolate

2.5.4. Synthetic Efforts Towards Omuralide Initially, we set out to perform the Birch reduction of the 3-methoxy-phenyl functionality with a concomitant removal of the α-methylbenzyl auxiliary (Scheme 31). Reduction of the ethyl ester in ent-38f using LiBH4 cleanly gave the corresponding alcohol 51 in high yield. Treatment of compound 51 with Li in liquid ammonia led to a successful reduction of the 3-methoxy-phenyl group but along with an unwanted reduction of the α-methylbenzyl auxiliary to produce tetraene 52 in quantitative yield. Scheme 31. Failed Birch Attempts

OEtMeO

NH

OH

OPh

THF95%

OHMeO

NH

OH

Ph

Li/NH3

t-BuOHTHFquant.

OHMeO

NH

OH

ent-38f 51 52

LiBH4

To circumvent this unfortunate setback, the chiral auxiliary was removed under hydrogenolytic condition in the presence of Boc-anhydride to furnish carbamate 53 in quantitative yield (Scheme 32). Protection of the diol as an acetonide (54) followed by a Birch reduction gave the desired diene (55) in high yield. Oxidative cleavage of the diene to the corresponding β-keto ester 56 using ozone followed by DMS quench of the residual ozonide proceeded in good yield.

85 Mulzer, J.; Kerkmann, T. J. Am. Chem. Soc. 1980, 102, 3620-3622.

36

Scheme 32. Birch/Ozonolysis Sequence

OHMeO

NH

OH

Ph

51

H2, Pd(OH)2(Boc)2O

MeOHquant.

OHMeO

NHBoc

OH

53

OMe

MeO

NHBoc

54

O

O

MeO

NHBoc

55

O

O

O O

MeO

NHBoc

56

O

O

p-TSA

DMF76%

Na/NH3

EtOHEt2O, -78 °C

93%

O3CH2Cl2:MeOH

-78 °C

thenDMS

-78 °C - rt63%

No further studies on the synthesis towards omuralide have yet been performed due to lack of time, but the remaining steps will be discussed below. In order to complete the synthesis of omuralide, several issues have to be addressed. First of all, the order of the lactamization/stereoselective ketone reduction leading to 57 has to be worked out (Scheme 34). In the case of performing the ketone reduction first, the desired 1,2-syn-amino alcohol functionality 58 could be obtained if the reduction is conducted under polar Felkin-Anh control (Scheme 33a). The α-NHBoc moiety in 59 will then act as the polar substituent for electronic reasons and exert the major stereodirecting effect. Scheme 33. Stereoselective Ketone Reductions

H

H

R'

BocHN

O

RH

LA

H

BocHN

R'HO

H

R

RR'

OH

NHBoc

RR'

O

NHBoc LAH

a) Felkin-Anh

c) 1,3-Syn reduction

B

O

O

Et

Et

H

R'H

BocHNR

H

OO

B

BocHN

HR

H

R'

Et

EtH R R'

OH

NHBoc

OH

R R'

O

NHBoc

OH

Et2BOMeNaBH4

b) Chelation control

RR'

O

NHBoc LAH

NH

OLA

O

R R'

OtBu

H

Backside attack

H

H RR'

OH

NHBoc

59

59

62

58

60

61

37

It has been reported that stereoselective reductions of β-keto esters containing an α-NHBoc functionality, such as 59, using TiCl4 and BH3⋅pyridine at low temperature proceed to yield the desired 1,2-syn-amino alcohol in high selectivity.86 One potential drawback with the Felkin-Anh approach is the report on nucleophilic additions to syn-α-amino-β-hydroxy aldehydes that proceeds with low selectivity (Scheme 6, Chapter 1) since the syn-α-amino-β-hydroxy structural motif is present in 56.32,87 It should also be noted that when using a chelating Lewis acid, such as TiCl4, there is a possibility for the formation of a 7-membered chelate between the ketone and the carbamate carbonyl instead of between the ketone and the ester carbonyl, which would result in the formation of the 1,2-anti-amino alcohol (60) under Cram control (Scheme 33b).88 Another possibility for a successful ketone reduction would be to utilize the 1,3-syn-hydroxy structural motif in the product (61) and carry out a 1,3-syn selective reduction of 62 with diethylmethoxyborane as the complexing agent and sodium borohydride as the hydride source (Scheme 33c).89 Here, the R’-group is oriented in a pseudo-equatorial position in the transition state with the hydride attacking for stereoelectronic reasons from the lower face resulting in a 1,3-equatorial orientation of both the R- and the R’-group in the six-membered chair conformation.90 Scheme 34. Proposed Completion of the Synthesis of Omuralide

NPg

O

HO

NH

O

O

OOH

NPg

O

O

O

O O

MeO

NHBoc

56

O

OLactamization

Reduction

OO

OlefinationOx. Cleavage

Lactonization

Aldol

Reduction/Deprotection

57 50 44

Once that is solved, the exo-methylene group in 50 can be introduced using Eschenmoser’s salt and a suitable base (Scheme 34).91 The formation of the 86 (a) Meloni, M. M.; Taddei, M. Org. Lett. 2001, 3, 337-340. (b) Marcantoni, E.; Alessandrini, S.;

Malavolta, M.; Bartoli, G.; Bellucci, M. C.; Sambri, L.; Dalpozzo, R. J. Org. Chem. 1999, 64, 1986-1992.

87 For further discussions, see: (a) Restorp, P. Stereoselective Nucleophilic Additions to α-Amino Aldehydes: Applications to Natural Product Synthesis, Doctoral thesis, KTH, Stockholm, 2006, pp 5-17. (b) Evans, D. A.; Cee, V. J.; Siska, S. J. J. Am. Chem. Soc. 2006, 128, 9433-9441.

88 For bidentate chelation between aldehydes and α-NHCbz using TiCl4, see: Gryko, D.; Prokopowicz, P.; Jurczak, J. Tetrahedron: Asymmetry 2002, 13, 1103-1113.

89 Chen, K.-M.; Hardtmann, G. E.; Prasad, K.; Repic, Shapiro, M. J. Tedrahedron Lett. 1987, 28, 155-158.

90 (a) Eliel, E. L.; Nader, F. J. Am. Chem. Soc. 1969, 91, 536-538. (b) Lewis, M. D.; Cha, J. K.; Kishi, Y. J. Am. Chem. Soc. 1982, 104, 4976-4978.

91 For examples of α-olefinations of lactams using Eschenmoser’s salt, see: (a) Tsujimoto, T.; Ishihara, J.; Horie, M.; Murai, A. Synlett 2002, 399-402. (b) Ishihara, J.; Horie, M.; Shimada, Y.; Tojo, S.; Murai, A. Synlett 2002, 403-406. (c) Brimble, M. A.; Crimmins, D.; Trzoss, M. Arkivoc 2005, 39-52.

38

β-lactone functionality should proceed without problems using well precedented methodologies by first an oxidative cleavage the 1,2-diol followed by oxidation of the transient aldehyde to the corresponding carboxylic acid. Treatment of the β-hydroxy carboxylic acid with BOP-Cl and triethylamine would then lead to the formation of β-lactone 50 as previously reported.92 This sets the stage for the late stage aldol reaction (vide supra) followed by a stereoselective olefin hydrogenation from the sterically least congested bottom face. Ideally, this step should also include the deprotection NH-moiety to complete the synthesis of omuralide (44).

2.6. Conclusions

We have developed a three-component approach for the synthesis of syn-α-hydroxy-β-amino esters based on a highly diastereoselective Rh(II)-catalyzed 1,3-dipolar cycloaddition of carbonyl ylides to imines. By using chiral α-methylbenzyl imines as dipolarophiles an asymmetric version of the cycloaddition was realized yielding enantiomerically pure syn-α-hydroxy-β-amino esters, and this methodology was applied on a short asymmetric synthesis of the paclitaxel side-chain as well as in an asymmetric synthetic approach towards the proteasome inhibitor omuralide. Furthermore, the use of chiral Rh(II) carboxylates produce the syn-α-hydroxy-β-amino esters in moderate enantioselectivity (er up to 82:18), which indicates that the reaction proceeds via a metal-associated carbonyl ylide.

92 Corey, E. J.; Li, W.; Reichard, G. A. J. Am. Chem. Soc. 1998, 120, 2330-2336.

39

3. 1,3-Dipolar Cycloadditions

of Azomethine Ylides to Aldehydes: Synthesis of syn-α-Amino-β-hydroxy Esters

(Paper III)

3.1. Introduction

Since their discovery and subsequent reaction with electron-deficient olefins in 1965 by Heine and Peavy,93 azomethine ylides have been extensively studied. The reaction belongs to one of the most important classes of 1,3-dipolar cycloadditions since it represents the conceptually most simple and efficient method for the construction of saturated, nitrogen-containing, five-membered heterocycles. Ylide generation, often performed in situ, followed by cycloaddition with suitable dipolarophiles furnishes pyrrolidines and pyrrolines in only one step from simple starting materials.

3.1.1. Generation of Azomethine Ylides The reaction first studied was the thermolysis of the corresponding aziridines in the presence of electron-deficient alkenes or alkynes. Huisgen and co-workers have systematically shown that thermolysis of 1-phenyl-2,3-dicarbomethoxyaziridines (63) involves a conrotatory ring opening to the stabilized S- or W-ylides (Scheme 35).94 The S-dipole, generated from the cis-aziridine, reacts with most dipolarophiles to form heterocycle trans-64 without loss of dipole geometry, while the W-dipole, from the trans-aziridine, also generates products corresponding to the S-dipole when reacted with less reactive dipolarophiles due to dipole interconversion.94a Scheme 35. Aziridine Thermolysis/1,3-Dipolar Cycloaddition

N

Ph

N

Ph

MeO2C N

CO2Me

Ph

MeO2C N

Ph

CO2Me

XX

XX

N

X X

MeO2C CO2Me

Ph

N

X X

MeO2C CO2Me

Ph

S-ylide

W-ylide

!

!

MeO2C CO2Me

MeO2C CO2Me

trans-63

cis-63

cis-64

trans-64

93 Heine, H. W.; Peavy, R. E. Tetrahedron Lett. 1965, 6, 3123-3126. 94 (a) Huisgen, R.; Mäder, H. J. Am. Chem. Soc. 1971, 93, 1777-1779. (b) Huisgen, R.; Sheer, W.;

Huber, H. J. Am. Chem. Soc. 1967, 89, 1753-1755.

40

Other routes to stabilized azomethine ylides comprise thermolysis of imines derived from aromatic aldehydes and α-amino ester, and the thermal N-alkylamino ester/aldehyde condensation.95 Although many of these methods have met some synthetic success in the intramolecular version, most suffers from limitations. Since the ylides are generated at elevated temperature they are prone to undergo proton transfer, intramolecular cyclization or ylide interconversion reactions. As a consequence, a low-temperature protocol for the generation of stabilized azomethine ylides would be a significant improvement. Reduction of N-methyl oxazolium salts (65), generated from oxazole 66, to the corresponding 4-oxazolines (67) constitutes one alternative since the highly unstable 4-oxazoline can undergo ring opening to azomethine ylide 68 (Scheme 36). In the absence of a dipolarophile the ylide is further reduced to tertiary amine 69 but in the presence of a suitable dipolarophile substituted pyrrolidines (70) or pyrrolines (71) can be isolated in high yields.96 Scheme 36. Vedejs’ Oxazoline Route to Azomethine Ylides

N

O

R

R'

MeOTf

MeCN

N

O

R

R'

MePhSiH3CsF

N

O

R

R'

Me

R N

Me

R'O

CO2Me

NR

MeO2C

R N

Me

R'O

NR

MeO2C CO2Me

Me Me

R' R'

O O

DDQN

R

MeO2C CO2Me

Me

R'

O

PhSiH3CsF

DMAD64-95%47-87%

656667 68 69

70 71 Another attractive mild method for azomethine ylide formation is the combination of a metallocarbene and an imine.46 The 1,3-dipole intermediate generated in situ could then be reacted with a variety of dipolarophiles, most commonly electron-deficient alkene or alkyne moieties (Scheme 37). In the case of a chiral metal catalyst, this methodology constitutes an elegant approach for the development of a catalytic, enantioselective 1,3-dipolar cycloaddition reaction.46a,e

Scheme 37. Metal-catalyzed Three-Component Formation of Pyrrolidines

N

HR

ArX Y

N2OR

O

N

Ar

RO2C R

X Y

[M]

95 For a review, see: Coldham, I.; Hufton, R. Chem. Rev. 2005, 105, 2765-2809. 96 (a) Vedejs, E.; Grissom, J. W. J. Am. Chem. Soc. 1986, 108, 6433-6434. (b) Vedejs, E.;

Grissom, J. W. J. Am. Chem. Soc. 1988, 110, 3238-3246.

41

From a synthetic point of view, the most attractive approach for the generation of azomethine ylides is to use a metal-cation to form a quaternary N-metallated salt, rather than a covalently bound N-substituent, followed by a deprotonation of the acidic α-proton since this would lead to a NH-product after quenching the reaction. Ultimately, if using a weak, reversible base in situ protonation of the formed N-metal species would release the metal-cation and regenerate the base and thereby opening the possibility for the development of a catalytic reaction. Since the azomethine ylide is inherently achiral, this approach would also serve as an excellent system for the synthesis of enantiomerically enriched products when using chiral metal catalysts. Schreiber and co-workers have recently developed an elegant approach using a silver(I)acetate/QUINAP-catalyzed [3+2] azomethine ylide cycloaddition of α-imino esters and acrylates furnishing substituted pyrrolidines in excellent yields and enantioselectivities (Scheme 38).97,98 The azomethine ylides are formed by a bidentate complexation of the chiral catalyst to the imine nitrogen and the carbonyl oxygen followed by deprotonation in the α-position. Scheme 38. Ag(I)-catalyzed Asymmetric Azomethine Ylide Cycloaddition

N

HAr

OMeO

OtBu

O

NH

Ar COOMe

tBuOOCi-Pr2NEt

cat. AgOAc/(S)-QUINAP

THF, -45 °C20h

89 - 95%89 - 96% ee

N

PPh2

(S)-QUINAP

3.1.2. Aim of the Study and Synthetic Strategy The aim of this study was to develop a complementary methodology to our previously discussed 1,3-dipolar cycloaddition strategy for the synthesis of the α-amino-β-hydroxy esters from carbonyl ylides and imines. To our surprise, to date only two reports on 1,3-dipolar cycloaddition of azomethine ylides to aldehydes99 have been disclosed where one example was based on stabilized ylides derived from thermolysis of 2-aroylaziridines99a (Scheme 39). In this investigation the authors state that only the S-ylide reacts with the dipolarophile since the thermal conrotatory opening of both cis- and trans-

97 (a) Chen, C.; Li, X.; Schreiber, S. L. J. Am. Chem. Soc. 2003, 125, 10174-10175. For other

recent examples using chiral Zn(II)-, Ag(I)- or Cu(II)-catalysts, see: (b) Gothelf, A. S.; Gothelf, K. V.; Hazell, R. G.; Jørgensen, K. A. Angew. Chem. Int. Ed. 2002, 41, 4236-4238. (c) Longmire, J. M.; Wang, B.; Zhang, X. J. Am. Chem. Soc. 2002, 124, 13400-13401. (d) Oderaotoshi, Y.; Cheng, W.; Fujitomi, S.; Kasano, Y.; Minakata, S.; Komatsu, M. Org. Lett. 2003, 5, 5043-5046.

98 For excellent reviews on asymmetric 1,3-dipolar cycloadditions of metallo-azomethine ylides, see: (a) Husinec, S.; Savic, V. Tetrahedron: Asymmetry, 2005, 16, 2047-2061. (b) Pandey, G.; Banerjee, P.; Gadre, S. R. Chem. Rev. 2006, 106, 4484-4517.

99 (a) Dallas, J.; Lown, J. W.; Moser, J. P. J. Chem. Soc. Chem. Commun. 1970, 278-279. (b) Padwa, A.; Chen, Y.-Y.; Chiacchio, U.; Dent, W. Tetrahedron, 1985, 41, 3539-3535.

42

aziridines (72) leads to the 2,4-trans relative stereochemistry in the product (73). This is presumably due to the fact that the ylide interconversion is faster than the cycloaddition at the elevated reaction temperature. This trend has also been observed for cycloadditions of aziridine-derived azomethine ylides to olefins.94a Scheme 39. Azomethine Ylide Generation and Subsequent Cycloaddition

N

N

iPr

Ar COPh

Ar COPh

iPr

NAr

COPh

NAr COPh

iPr

iPr!

!

Ar'CHONiPrO

Ar

COPhAr'

NiPrO

Ar

COPhAr'

48-71%

cis-72

trans-72 W-ylide

S-ylidetrans-73 cis-73

Based on this literature precedence we decided to start our investigation of 1,3-dipolar cycloadditions of azomethine ylides to aldehydes based on Vedejs’ oxazole methodology since this approach offers a convenient way of generating the desired ylides even at low temperatures. In analogy with our previous approach, hydrolysis of oxazolidine 74 under acidic conditions would lead to the desired α-amino-β-hydroxy ester 75 (Scheme 40). The oxazolidines would be available from a 1,3-dipolar cycloaddition of azomethine ylide 76, generated by reduction of the corresponding oxazolium salt 77, and aldehyde 16. The desired N-alkyl oxazolium saltsare readily available from suitably 2-substituted 5-alkoxy oxazoles (78) by N-alkylation. Scheme 40. Synthetic Strategy

R3O R2

O

NHR1

OH

OR1N

R3O2C R2

R4

R2 O

75 74

76

16

N R4

H3O+ 1,3-DC

ylide

generation

77

R1

N

OR4

R3O

R1

N

OR4

R3O

78

R1OTf

CO2R3

OTf

3.2. 1,3-Dipolar Cycloadditions

3.2.1. 1,3-Dipolar Cycloadditions of Oxazoline-Derived Azomethine Ylides

Preparation of the desired oxazoles started by a N-benzoylation of glycine methyl ester hydrochloride 79 under standard biphasic condition to give amide 80 in high yield (Scheme 41). Dehydration of N-benzoyl glycine 80 using P2O5

43

in refluxing chloroform100 furnished the desired 5-methoxy-2-phenyloxazole 81 in moderate yield, which was followed by a N-methylation reaction by treatment of the oxazole with methyltriflate to give the N-methyl oxazolium salt 82 as a crystalline solid in quantitative yield. Scheme 41. Synthesis of N-Methyl Oxazolium Salt

H3NOMe

O

ClPhCOCl

EtOAcNaHCO387%

PhHN

OMe

O

O

N

O

Ph

OMe

P2O5

CHCl3!40%

MeOTf

CH2Cl2quant.

N

O

Ph

OMe

MeOTf

79 80 81 82 Since the azomethine ylide, formed from the tautomerization of the reduced oxazolium salts, is a reactive species the dipolarophile needs to be present throughout the reaction. Consequently, a chemoselective reduction of the oxazolium salts to the corresponding 4-oxazolines in the presence of the dipolarophile is needed without overreduction of the azomethine ylide to form tertiary amine 69. The combination of cesium fluoride and phenylsilane fulfills these criteria and has proved to be the reducing agent of choice in the reactions of azomethine ylides and acrylates.101 Treatment of anhydrous cesium fluoride in acetonitrile with a solution of oxazolium salt 82, benzaldehyde (18a) and phenylsilane at room temperature led to the formation of the desired oxazolidine 85 in good yield although along with substantial amounts of amine 86 (Scheme 42). Hydrolysis of oxazolidine 85 under standard conditions gave amino alcohol 87 as a 2:1 mixture of syn:anti diastereomers.102 Scheme 42. 1,3-Dipolar Cycloaddition of Oxazole-Derived Azomethine Ylides

N

OPh

OMe

Me

OPh

82 18a

PhSiH3CsF

MeCN

N

OPh

OMe

MeN PhSiH3

CsF

MeCN

ON

Ph

Me

MeO2C Ph

p-TSA

MeOHH2O58%

MeO Ph

O

NHMe

OH

8587

dr 2:1

83 84 86

87:86 65:35 - 50:50

OTf

Me

O OMe

Ph N

Me

O OMe

Ph

100 L’abbe, G.; Ilisiu, A. M.; Dehaen, W.; Toppet, S. J. Chem. Soc., Perkin Trans.

1 1993, 2259-2261. 101 The use of sodium borohydride and sodium cyanoborohydride led only to overreduction to the

corresponding tertiary amine, see: Ref 94. 102 For determination of the relative stereochemistry, see: Ref 51.

44

All attempts to suppress the undesired overreduction of the azomethine ylide to tertiary amine 86, by either lowering the reaction temperature or by slow addition of phenylsilane failed. Although disappointing, the successful formation of α-amino-β-hydroxy ester 87 at least proves the potential of strategy as an efficient entry to β-amino alcohols.

3.2.2. 1,3-Dipolar Cycloadditions of Metallo-Azomethine Ylides Unable to improve the results from the oxazole-derived azomethine ylide cycloaddition, we were forced to reconsider our approach for the generation of the ylides. Consequently, the Ag(I)/phosphine catalyst system was chosen as the next method since it has proved successful in reactions using other dipolarophiles.97a,c This methodology also has the advantage of generating unsubstituted amines, which is not the case in the above-mentioned approach. Treatment of glycinate-derived imine 88a with 5 mol% of a preformed Ag(I)/PPh3 catalyst in the presence of Hünig’s base in THF is believed to generate Ag-azomethine ylide 89. Addition of 89 to benzaldehyde (18a) yields oxazolidine 90a with 56% conversion of 88a (Scheme 43). Acidic hydrolysis of the N,O-acetal under standard conditions led to syn-amino alcohol 91a in a 1.6:1 diastereoselectivity.102 Scheme 43. AgOAc/PPh3-Catalyzed 1,3-Dipolar Cycloadditions

Ph NOEt

O

AgOAcPPh3

i-Pr2NEt

THF56% conv.

OPhOHN

EtO2C

Ph

Ph

p-TSA

MeOHH2O

EtO Ph

O

NH2

OH

91a

dr 1.6:1

88a 18a

90a

Ph NOEt

O

89

Ag

PPh3Ph3P

Although encouraging, both the conversion and diastereoselectivity has to be improved in order for this approach to be a synthetically viable entry to syn-α-amino-β-hydroxy esters. First, the benzylidene substituent was varied in order to evaluate how this would influence the reaction outcome. Consequently, several imines having 4-F-, 4-CN- and 4-CF3-phenyl substituents were screened (Table 7, entries 2-5). When using the more electron-poor substrates and increasing the catalyst loading to 10 mol% the conversion was improved to above 90% (entry 4 and 5) with a slightly increased diastereoselectivity (compare entry 1 and 5). Next, the ethyl ester was exchanged for the bulkier tert-butyl ester as the increased steric bulky might influence the stereochemical outcome of the reaction. Unfortunately, only a slight decreased conversion and diastereoselectivity was observed (entries 6 and 7). Other operations, such as changing solvent (toluene, CH2Cl2) or lowering the reaction temperature (0 or -20 °C) gave no improvements.

45

Table 7. Optimization of AgOAc/PPh3-Catalyzed 1,3-Dipolar Cycloadditionsa

Ar NOR

O

AgOAcPPh3i-Pr2NEt OHN

RO2C

Ar

Ph

p-TSA

MeOHH2O

RO Ph

O

NH2

OH

88

18a

90 syn-91

THFRO Ph

O

NH2

OH

anti-91

Entry Cat. (mol%) 88 (Ar / R) Conv. 1,3-DC (%)b dr 91 (syn:anti)c

1 5 a (Ph / Et) 56 a 1.6:1 2 5 b (4-F-Ph / Et) 63 a 1.7:1 3 10 b (4-F-Ph / Et) 82 a 1.7:1 4 10 c (4-CN-Ph / Et) 90 a 2.3:1 5 10 d (4-CF3-Ph / Et) 92 a 2.4:1 6 10 e (4-CF3-Ph / t-Bu) 87 b 1.9:1 7 10 f (4-CN-Ph / t-Bu) 91 b 1.8:1

a The reaction was carried out over night with imine 88 (1.0 equiv), aldehyde 18a (2.0 equiv), i-Pr2NEt (0.2 equiv) and AgOAc/PPh3 (0.1 M in THF, 1:2.4) in THF (c = 0.1 M) at rt. Hydrolysis was performed with p-TSA (2 equiv) in MeOH:H2O (95:5). b Determined by 1H NMR analysis of the crude reaction mixture before hydrolysis. c Determined by 1H NMR analysis of the crude reaction mixture after hydrolysis.

Benzophenone-derived tert-butyl glycinate imines have been reported to be excellent substrates in various direct aldol reactions with aldehydes for the formation of α-amino-β-hydroxy esters.28 Therefore, commercially available benzophenone-derived imine 92 was used as imine component under the above-mentioned reaction conditions to afford, after hydrolysis, amino alcohol 91b in 68% yield with a slightly increased diastereoselectivity (syn:anti 3.4:1, Scheme 44). Scheme 44. 1,3-Dipolar Cycloaddition Using Benzophenone-Derived Imine

Ph NOt-Bu

O 1) AgOAc/PPh3 i-Pr2NEt, THF

t-BuO Ph

O

NH2

OH

92

18a

Ph

91b

syn:anti 3.4:1

2) HCl (1 M) MeOH:H2O 68%

No further studies on the 1,3-dipolar cycloaddition of azomethine ylides to aldehydes have yet been performed due to lack of time (for further discussions see section 3.3).

3.2.3. Mechanistic Aspects The low diastereoselectivity in the Ag(I)-catalyzed reaction compared to the Ag(I)-catalyzed 1,3-dipolar cycloaddition of azomethine ylides to acrylates

46

(Scheme 38) and the 1,3-dipolar cycloaddition of carbonyl ylides to imines (Chapter 2) is somewhat puzzling. Since it is know that 1,3-dipolar cycloadditions of azomethine ylides belongs to the Type I reactions (section 1.4.3.) where the important interaction is between the HOMO of the dipole and the LUMO of the dipolarophile, the introduction of more electron-withdrawing benzylidene-substituents is expected to lower the HOMOdipole and consequently retard the reaction. This in sharp contrast with the results in Table 7 where the more electron-poor imines led to faster reactions and higher conversions and might suggest that alternative reaction pathways are involved. One possibility would be to invoke a step-wise pathway where the first step would be an aldol addition of Ag-enolate 93 to aldehyde 18a forming Ag-alkoxide 94 (Scheme 45). The alkoxide could then undergo either an internal cyclization forming oxazolidine 95 or an acid-base reaction to form the linear product 96. In the case with the benzaldehyde-derived imines (88) no traces of the linear product could be found but with imine 92 the 13C NMR of the crude reaction mixture (prior to hydrolysis) revealed what is believed to be a mixture of products 95 and 96. This result supports the step-wise reaction pathway and is in line with the reports on the asymmetric aldol reactions of glycine-derived enolates under either phase-transfer conditions28a,c or under Lewis acid catalysis28b where mixtures of 95 and 96 were formed when using imine 92. Scheme 45. Possible Reaction Pathways of Benzophenone Derived Imines

Ph NOt-Bu

O

AgOAcPPh3i-Pr2NEt

OPhOHN

t-BuO2C Ph92 18a

95

Ph NOt-Bu

O

93

Ag

PPh3Ph3P

THF

1,3-DC

Aldol

PhPh

Ph Ph

Ot-Bu

O

94

N

Ph

O

Ph

Ph

Ag

Cyclization OHN

t-BuO2C Ph95

Ph Ph

ProtonationOt-Bu

O

96

N

Ph

OH

Ph

Ph

Further experimental work has to be done in order to elucidate the reaction pathway and thereby gain more understanding in order to improve the reaction outcome.

47

3.3. Conclusions and Outlook

We have investigated a 1,3-dipolar cycloaddition approach of azomethine ylides to aldehydes as a potential entry to α-amino-β-hydroxy esters. Two strategically different approaches for the generation of the azomethine ylide have been evaluated, the Vedejs’ oxazole-methodology and Ag(I)/PPh3-catalyzed reaction. The best results where obtained by using the Ag(I) catalyst in which the amino alcohol was isolated in 68% yield with 3.4:1 syn-selectivity. Preliminary data supports a step-wise aldol/cyclization pathway rather than a concerted 1,3-dipolar cycloaddition reaction. Further improvements have to be made in order for this to be a synthetically viable route to this important compound class. First, more work has to be done in order to prove the reaction pathway in order to gain understanding of the reaction and to improve the diastereoselectivity. Secondly, with optimized reaction conditions at hand, the substrate scope and limitations has to be screened using a large variety of aldehydes. Last, an asymmetric version of this transformation must be developed for the synthesis of optically enriched products. This could either be done by utilizing a chiral Ag(I) catalyst (compared with Scheme 38) or by using chiral α-imino esters (e.g. 8-phenylmenthyl ester, compare with Scheme 24 and 25, Chapter 2).

48

49

4. Total Synthesis of

D-erythro-Sphingosine (Paper IV)

4.1. Introduction

Glycosphingolipids are ubiquitous constituents of cell membranes, located abundantly in all plasma membranes as well as in intracellular organelles in all eukaryotic cells. The recent discovery of their biologically active metabolites e.g. sphingosine, ceramide and lysosphingolipids has generated interest in the physiological role of these molecules. The backbone of the sphingolipids consists of a long aliphatic chain connected to a polar 2-amino-1,3-diol head group. D-erythro-Sphingosine (97, Figure 12) and related compounds have been shown to inhibit protein kinase C (PKC), which affects cell regulation and signal transduction, and exhibit anti-tumor promoter activities in various mammalian cells (Scheme 46). In addition, these compounds may function as modulators of cell function and possibly also as secondary messengers.103 Scheme 46. Metabolic Pathway of Sphingolipids

OH

C13H27O

HN C15H31

O

P

O

OMe3N

OH

C13H27HO

HN C15H31

OP

O

OMe3N OH

OH

C13H27HO

NH2

OH

C13H27O

NH2

PHO

OH

O

Sphingomyelin Ceramide Sphingosine (97) Sphingosine-1-phosphate

apoptosiscell differentiation

inhibition of PKC activation of PKCcell differentiation

cell growth Due to the biological importance of 97 and its derivatives many synthetic routes to enantiopure sphingolipids have been described, most commonly utilizing starting materials from the chiral pool, e.g. L-serine or carbohydrates.104 Other investigations have employed asymmetric reactions such as Sharpless asymmetric epoxidations or aldol reactions with chiral

103 (a) Cremesti, A. E.; Goni, F. M.; Kolesnick, R. FEBS Lett. 2002, 531, 47-53. (b) Karlsson, K.-

A. Trends Pharm. Sci. 1991, 12, 265-272. (c) Hannun, Y.; Bell, R. M. Science 1989, 243, 500-507. (d) Hannun, Y. Science 1996, 274, 1855-1859.

104 (a) Koskinen, P. M.; Koskinen, A. M. P. Synthesis 1998, 1075-1091. (b) Gargano, J. M.; Lees, W. J. Tetrahedron Lett. 2001, 42, 5845-5847. (c) Nakamura, T.; Shiozaki, M. Tetrahedron 2001, 57, 9087-9092. (d) van der Berg, R. J. B. H. N.; Korevaar, C. G. N.; van der Marel, G. A.; Overkleeft, H. S.; van Boom, J. H. Tetrahedron Lett. 2002, 43, 8409-8412. (e) Lee, J.-M.; Lim, H.-S.; Chung, S.-K. Tetrahedron: Asymmetry 2002, 13, 343-347.

50

auxiliaries to install the two vicinal stereogenic centers.105,106 Recently Kobayashi and co-workers developed an elegant approach to the synthesis of L-erythro-Sphingosine based on a highly enantioselective zirconium catalyzed aldol reaction of a glycine-derived silicon enolate with aldehydes (Scheme 4, Chapter 1).30

HO

HO

HO

NH2

NH2

OH

OH

OH

OH

OH

OH

OH

NH2

O

D-erythro-Sphingosine (97)

Phytosphingosine

Sphingofungin B Figure 12. Various Sphingoid Bases

4.2. Retrosynthetic Analysis of D-erythro-Sphingosine

The formation of the E-olefin moiety of 97 has traditionally been accomplished by either an E-selective C-C double bond forming reaction, e.g. Wittig-Schlosser, Julia-Kocienski or Horner-Wadsworth-Emmons-type olefination, or stereoselective reductions of alkynes. We became interested in assembling 97 by using an alkene cross-metathesis reaction to introduce the required E-olefin moiety (Scheme 47). Such a strategy would be convenient compared to alternatives using more traditional C-C double bond forming reactions, since the required catalyst is commercially available and the metathesis usually proceeds with low amount of byproduct formation.107 This approach would also allow for a divergent route towards sphingosine analogs by simply changing the cross-metathesis partners.108 The polar head group 98 should be available from epoxide 99, which, in turn, is readily prepared from divinylcarbinol (100) using a Sharpless asymmetric epoxidation/Payne

105 (a) Nugent, T. C.; Hudlicky, T. J. Org. Chem. 1998, 63, 510-520. (b) He, L. L.; Byun, H. S.;

Bittman, R. J. Org. Chem. 2000, 65, 7627-7633. (c) Corey, E. J.; Choi, S. Tetrahedron Lett. 2000, 41, 2765-2768.

106 Olofsson, B.; Somfai, P. J. Org. Chem. 2003, 68, 2514-2517. 107 (a) Connon, S. J.; Blechert, S. Angew. Chem. Int. Ed. 2003, 42, 1900-1923. (b) Chatterjee, A.;

Choi, T.-L.; Sanders, D. P.; Grubbs, R. H. J. Am. Chem. Soc. 2003, 125, 11360-11370. 108 After the publication of this project several sphingosine and sphingosine-analog syntheses

using a cross-metathesis have appeared in the literature, see: (a) Rai, A. N.; Basu, A. Org. Lett. 2004, 6, 2861-2863. (b) Rai, A. N.; Basu, A. J. Org. Chem. 2005, 70, 8228-8230. (c) Chaudhari, V. D.; Ajish Kumar, K.S.; Dhavale, D. D. Org. Lett. 2005, 7, 5805-5807. (d) Yamamoto, T.; Hasegawa, H.; Hakogi, T.; Katsumura, S. Org. Lett. 2006, 8, 5569-5572. (e) Peeters, C.; Billich, A.; Ghobrial, M.; Högenauer, K.; Ullrich, T.; Nussbaumer, P. J. Org. Chem. 2007, 72, 1842-1845.

51

rearrangement sequence. The realization of this strategy yields a short synthesis of 97 as outlined below. Scheme 47. Retrosynthetic Analysis of D-erythro-Sphingosine

HO C13H27

OH

NH2

ONBn

OH

O

C13H27

HOO

SAEOH

CM carbamate tetheredepoxide opening

97 98

99100

4.3. Synthesis of D-erythro-Sphingosine

4.3.1. Oxazolidinone Synthesis Sharpless asymmetric epoxidation of commercially available divinylcarbinol 100 afforded epoxide 101, which was followed by a base-induced Payne rearrangement to give 99 in good yield (65% over 2 steps) and excellent enantiomeric purity (>99% ee), as previously reported (Scheme 48).109 It was expected that intermolecular reaction of 99 with various nitrogen nucleophiles would, for electronic reasons, result in a preferential ring opening in the allylic position.21,110 To circumvent this, the C2 amino functionality was introduced through a two-step procedure using an intramolecular epoxide-opening via a tethered carbamate. Alkyl isocyanates have previously been used as nitrogen nucleophiles to give regioselective attack at the C2 position of epoxy alcohols.111 Consequently, when epoxy alcohol 99 was treated with benzyl isocyanate and Et3N in a sealed tube, N-benzyl carbamate 102 was obtained in excellent yield. The subsequent deprotonation of 102 using NaH followed by intramolecular epoxide opening proceeded sluggishly and gave a poor yield of oxazolidinone 98.111 Gratifyingly, when using NaHMDS in THF at low temperature 98 was obtained in 88% yield.112 As expected the product derived from the favored 5-exo-tet cyclization was the only detected product; the

109 (a) Schreiber, S. L.; Schreiber, T. S.; Smith, D. B. J. Am. Chem. Soc. 1987, 109, 1525-1529.

(b) Jäger, V.; Schröter, D.; Koppenhoefer, B. Tetrahedron 1991, 47, 2195-2210. (c) Romero, A.; Wong, C.-H. J. Org. Chem. 2000, 65, 8264-8268.

110 Olofsson, B.; Khamrai, U.; Somfai, P. Org. Lett. 2000, 2, 4087-4089. 111 (a) Roush, W. R.; Adam, W. A. J. Org. Chem. 1985, 50, 3752-3757. (b) Jäger, V.; Hummer,

W.; Stahl, U.; Gracza, T. Synthesis 1991, 769-776. (c) Jäger, V.; Stahl, U.; Hummer, W. Synthesis 1991, 776-782.

112 Martin, R.; Moyano, A.; Pericas, M. A.; Riera, A. Org. Lett. 2000, 2, 93-95.

52

product derived from the 6-endo-tet cyclization being disfavored according to the Baldwin rules.113 Scheme 48. Synthesis of Polar Head Group

OH

HOO

BnHN OO

O

O

NBn

O

OH

OH

O

100 101 99

102 98

(-)-DIPT, Ti(Oi-Pr)4cumene-OOH

CH2Cl2, -35 °C

77%, >99% ee

NaOH

85%

BnNCOEt3N

Et2O, 60 °C98%

NaHMDS

THF, -15 °C88%

4.3.2. Cross-Metathesis and Completion Due to the lipid chain and the hydrophilic head group, sphingosine exhibits surfactant-like properties, e.g. low solubility in organic solvents at low temperature, and problematic flash chromatography due to aggregation of the material on the silica gel column.106 The late introduction of the aliphatic chain by a cross-metathesis reaction overcomes many of these problems and has advantages over several of the existing strategies towards 97, in which the lipid chain is introduced at an early stage of the synthesis.

Table 8. Cross-Metathesis

ONBn

OH

O

ONBn

OH

O

C13H27 ONBn

OH

O

O

BnN

O

OH

cat.

98 103 107

C13H27

Entry Cat. mol % Conditionsa Product (%)b E:Zc

1 104 5 40 °C, 43 h n.r. N.A. 2 105 5 rt, 3 days 103 (35) 14:1 3 105 5 40 °C, 24 h 103 (49) 5:1 4d 105 10 50 °C, 1.5 h 103 (7) / 107 (57) 12:1 / E 5e 105 10 40 °C, 0.75 h 103 (52) 16:1 6 106 10 40 °C, 1.5 h 103 (59) 17:1

a 1-pentadecene (2 equiv.) in CH2Cl2. b Isolated yields. c Determined by 1H NMR analysis of the crude reaction mixture. d 1-pentadecene (2 equiv.) in PhMe. e Ti(Oi-Pr)4 (2 equiv.) was added prior to the catalyst.

113 (a) Baldwin, J. E. J. Chem. Soc, Chem. Commun. 1976, 734-736. (b) Baldwin, J. E.; Cutting, J.;

Dupont, W.; Kruse, L.; Silberman, L.; Thomas, R. C. J. Chem. Soc, Chem. Commun. 1976, 736-738.

53

Compound 103 was synthesized by an olefin cross-metathesis reaction of allylic alcohol 98 and 1-pentadecene in the presence of Ru-catalysts 104-106 (Figure 13 and Table 8).

RuCl

ClRu

PCy3

Cl

Cl

NMesMesN

104 105

Ru

N

NMesMesNCl

Br

N

Br

Cl

106

PhPCy3

PCy3

Ph Ph

Figure 13. Ru-catalyst used in cross-metathesis reaction

The initial coupling attempts under standard cross-metathesis conditions,107b using the commercially available Grubbs’ ruthenium carbene catalysts 104 and 105 (Figure 13), resulted in either no reaction (Table 8, entry 1) or required prolonged reaction time, giving 103 in moderate yield (entry 2) or low E:Z selectivity (entry 3). When the reaction was conducted in toluene instead of CH2Cl2 homodimer 107 precipitated from the reaction mixture as a white solid, thereby shifting the equilibrium towards this undesired dimerization (entry 4). The low chemical yield encountered with catalyst 104 or catalyst 105 in CH2Cl2 at room temperature, or the low E:Z selectivity encountered with catalyst 105 in refluxing CH2Cl2, could be circumvented by addition of 2 equivalents of Ti(Oi-Pr)4 to the reaction mixture and 103 was isolated in 52% yield (78% based on recovered starting material) with 16:1 E:Z selectivity (entry 5). The E:Z isomers were readily separated by flash chromatography. The need for Ti(Oi-Pr)4 could be explained by the high heteroatom density in 98, since Ru-catalyst 105 have been reported to form inactive chelates with Lewis-basic heteroatoms in the substrate.114 The presence of Ti(Oi-Pr)4 decreases this chelation and increases the amount of active catalyst, which translates into higher yields.114,115 When employing Grubbs’ phosphine-free 3-bromopyridine catalyst 106116a (Figure 13), both the yield and the E:Z selectivity were increased to 59% (82% based on recovered starting material) and 17:1 respectively (entry 9). Catalyst 106 was readily available in one step from commercially available 105 and has shown increased activity in various cross-metathesis reactions (Scheme 49).116,117 The increased activity of catalyst 106 is due to the absence of phosphine ligands. The 3-bromopyridine ligands lead to a faster ligand-dissociation from the precatalysts and a decreased 114 Fürstner, A.; Langemann, K. J. Am. Chem. Soc. 1997, 119, 9130-9136. 115 (a) Ghosh, A. K.; Cappiello, J.; Shin, D. Tetrahedron Lett. 1998, 39, 4651-4654. (b) Rodríguez,

S.; Castillo, E.; Carda, M.; Marco, J. A. Tetrahedron 2002, 58, 1185-1192. 116 (a) Love, J. A.; Morgan, J. P.; Trnka, T. M.; Grubbs, R. H. Angew. Chem. Int. Ed. 2002, 41,

4035-4037. (b) Kang, B.; Kim, D.-H.; Do, Y.; Chang, S. Org. Lett. 2003, 5, 3041-3043. (c) Kanemitsu, T.; Seeberger, P. H. Org. Lett. 2003, 5, 4541-4544.

117 Commercially available from Sigma-Aldrich (from 2007).

54

tendency for ligand re-association. Consequently, the concentration of the active catalytic-species in the reaction mixture is higher and thereby increasing the activity. Scheme 49. Synthesis of Bispyridine Catalyst 106

Ru

PCy3

Cl

Cl

NMesMesN

105

Ru

N

NMesMesNCl

Br

N

Br

Cl

106 (100%)

Ph Ph

excess3-Br-C5H4N

-PCy3

Removal of the benzyl group in 103 with sodium in liquid ammonia proceeded cleanly to afford NH-oxazolidinone 108, which was then hydrolyzed without further purification with 1M KOH in refluxing EtOH:H2O to yield the desired D-erythro-sphingosine (97) as a white solid in 33% overall yield over seven steps from commercially available 100 (Scheme 50). Analytical data of 97 were in all aspects identical to those previously reported.104,105,106 Scheme 50. Deprotection of D-erythro-Sphingosine

HO C13H27

NH2

OH

ONBn

OH

O

C13H27 ONH

OH

O

C13H27

103 108 97

Na/NH3

Et2O, -78 °Cquant.

1M KOH

H2O:EtOH100 °Cquant.

4.4. Conclusions

We have developed a short, flexible synthetic route for the gram-scale synthesis of D-erythro-sphingosine (97). The key feature of the synthesis is the assembly of the polar head group with the long aliphatic chain using an E-selective cross-metathesis reaction. This strategy also opens up for the possibility of a rapid synthesis of sphingosine and ceramide analogs by merely changing cross-metathesis partners, which was later exemplified by Basu and others using different head groups (Scheme 51).108a,d,e

Scheme 51. Basu’s Synthesis of Sphingosine Analogs using the CM-Approach

OPMB OPMB

Rcat.RTBDPSO TBDPSO

NHFmoc NHFmoc

R = C13H27

C7H15

C13H26STr

55

5. Total Synthesis of

(−)-Stemoamide (Paper V)

5.1. Introduction

The Stemona class of natural products, isolated from the root extracts of Stemona tuberosa, comprises 42 isolated compounds that represents a family of polycyclic alkaloids structurally characterized by the presence of a pyrrolo[1,2-a]azepine core. Several members also contain a α-methyl-γ-butyrolactone subunit (Figure 14).118 Traditionally, the root extracts have been employed in Chinese and Japanese folk medicine for the treatment of respiratory disorders but also as an antihelminthic.118 Apart from (−)-stemoamide (109),119 the Stemona class also contains (−)-stenine,120 (+)-croomine,121 (−)-tuberostemonine122 and (−)-stemonine123 all of which have been targets for several total syntheses during the last decade due to their intricate structural features as well as their interesting biological properties. (−)-Stemoamide (109), isolated in 1992 by Xu and coworkers,118a is structurally the simplest member of the Stemona class containing a tricyclic core and four contiguous stereogenic centers. The first total synthesis was reported by Williams and coworkers in 1994.119a

118 (a) Lin, W.-H.; Ye, Y.; Xu, R.-S. J. Nat. Prod. 1992, 55, 571-576. (b) Pilli, R. A.; Ferreira de

Oliveira, M. C. Nat. Prod. Rep. 2000, 17, 117-127, and references cited therein. 119 Total and formal syntheses of (−)-stemoamide: (a) Williams, D. R.; Reddy, J. P.; Amato, G. S.

Tetrahedron Lett. 1994, 35, 6417-6420. (b) Kinoshita, A.; Mori, M. J. Org. Chem. 1996, 61, 8356-8357. (c) Jacobi, P. A.; Lee, K. J. Am. Chem. Soc. 2000, 122, 4295-4303. (d) Gurjar, M. K.; Reddy, D. S. Tetrahedron Lett. 2002, 43, 295-298. (e) Sibi, M. P.; Subramanian, T. Synlett 2004, 1211-1214. (f) Olivo, H. F.; Tovar-Miranda, R.; Barragán, E. J. Org. Chem. 2006, 71, 3287-3290.

120 Total syntheses of (−)-stenine: (a) Wipf, P.; Kim, Y.; Goldstein, D. M. J. Am. Chem. Soc. 1995, 117, 11106-11112. (b) Morimoto, Y.; Iwahashi, M.; Nishida, K.; Hayashi, Y.; Shirahama, H. Angew. Chem. Int. Ed. 1996, 35, 904-906. (c) Morimoto, Y.; Iwahashi, M.; Kinoshita, T.; Nishida, K. Chem. Eur. J. 2001, 7, 4107-4116.

121 Total syntheses of (+)-croomine: (a) Williams, D. R.; Brown, D. L.; Benbow, J. W. J. Am. Chem. Soc. 1989, 111, 1923-1925. (b) Martin, S. F.; Barr, K. J. J. Am. Chem. Soc. 1996, 118, 3299-3300. (c) Martin, S. F.; Barr, K. J.; Smith, D. W.; Bur, S. K. J. Am. Chem. Soc. 1999, 121, 6990-6997.

122 Total syntheses of (−)-tuberostemonine: (a) Wipf, P.; Rector, S. R.; Takahashi, H. J. Am. Chem. Soc. 2002, 124, 14848-14849. (b) Wipf, P.; Rector, S. R. J. Am. Chem. Soc. 2005, 127, 225-235.

123 Total synthesis of (−)-stemonine: Williams, D. R.; Shamim, K. S.; Reddy, J. P.; Amato, G. S.; Shaw, S. M. Org. Lett. 2003, 5, 3361-3364.

56

N

98

N9a

10

O

O

O

H

H

H

(-)-Stemoamide (109)

NH

NH

O

OH

H

O O

O

H

O

H

(-)-Stenine (+)-Croomine

NH

O

H

O

H

(-)-Tuberostemonine

H

NH

OH

H

O

(-)-Stemonine

O

H

H

O

Pyrrolo[1,2-a]azepine

OH O

Figure 14. Stemona Alkaloids

In 2000, Jacobi and co-workers disclosed the shortest asymmetric synthesis of (−)-stemoamide to date (Scheme 52).119c Their synthesis relies on the newly developed intramolecular Diels-Alder/retro-Diels Alder sequence of alkyne oxazole 110 for the formation of butenolide 111. The highly reactive alkyne oxazole undergoes a [4+2]-cycloaddition to form Diels-Alder adduct 112 which then rapidly expels MeCN through a retro-Diels Alder reaction giving 113. Finally acidic hydrolysis yields the desired butenolide 111 in good yield as a single detectable stereoisomer. Scheme 52. Jacobi’s Asymmetric 10 Step Synthesis of (−)-Stemoamide

NH

O

H

N

O

H

N

O

H

112

O

OEt

N

O

Cl

MeO

O

N

O

OEt

MeO

NaH N

O

H

110

O

NMeO

N

OMeO

-MeCN

N

O

H

113

O

MeO

N

O

H

111

O

O

109

N

O

H

O

O

H

H H

H+NiCl2

NaBH4

(2 steps)(3 steps)

(4 steps) !

!

5.2. Retrosynthetic Analysis of (−)-Stemoamide

Retrosynthetically, we set out to install the γ-butyrolactone functionality with the correct C8-C9 trans-stereochemistry by a stereoselective hydroboration proceeding from the most accessible top-face of β,γ-unsaturated azepine 114 (Scheme 53). The last step, α-methylation of the γ-butyrolactone to afford the

57

correct stereochemistry, is well precedented in the literature.119e,f,124 The pyrrolo[1,2-a]azepine core was envisioned to be derived from vinyliodide 115 through a palladium(0)-catalyzed Negishi-coupling followed by a ring-closing metathesis reaction. In a previous study by the Mori group, a similar azepine structure containing the trisubstituted olefin was constructed by an enyne metathesis approach.119b Vinyliodide 115 should be available from enyne 116 by chemoselective iodoboration of the alkyne in the presence of the terminal alkene. Enyne 116 is accessible from commercially available (S)-pyroglutaminol 117 using well-established methodology. Scheme 53. Retrosynthetic Analysis of (−)-Stemoamide

N

O

O

O

H

H

H

N

HO

OH

H

H

N

O

HI N

O

H

Methylation

Lactonization

OEtO

Hydroboration

N

O

H

OEtO

RCM

Negishicoupling

IodoborationOxidation/alkynylation Alkylation

NH

O

HOH

(-)-Stemoamide (109)114

117

116115

5.3. Synthesis of (−)-Stemoamide

5.3.1. Construction of the Pyrrolo[1,2-a]azepine Core Our synthesis began with protection of primary alcohol 117 as the corresponding TBS-ether 118 (Scheme 54). Subsequent N-alkylation using NaHMDS at low temperature followed by removal of the TBS-group proceeded as expected to cleanly give alcohol 119 in excellent yield. Conversion of alcohol 119 into alkyne 116 was realized by a Swern oxidation followed by treatment of the crude aldehyde with Ohira-Bestmann diazophosphonate125 120 and K2CO3 in MeOH to afford the desired alkyne in excellent yield over two steps. The vinyliodide moiety in 115 was installed by a chemoselective iodoboration of the alkyne functionality in the presence of a primary alkene using B-I-9-BBN at low temperature126 to give 115 in good yield. The subsequent Pd(0)-catalyzed sp2-sp3-coupling to install the

124 Kohno, Y.; Narasaka, K. Bull. Chem. Soc. Jpn. 1996, 69, 2063-2070. 125 (a) Ohira, S. Synth. Commun. 1989, 19, 561-564. (b) Müller, S.; Liepold, B.; Roth, G. J.;

Bestmann, H. J. Synlett 1996, 521-522. 126 (a) Hara, S.; Dojo, H.; Takinami, S.; Suzuki, A. Tetrahedron Lett. 1983, 24, 731-734. (b)

Hodgson, D. M.; Foley, A. M.; Lovell, P. J. Synlett 1999, 744-746.

58

β,γ-unsaturated ester required some optimization. Subjecting a solution of vinyliodide 115 and Pd(PPh3)4 in THF/DMPU to Reformatsky reagent 121, derived from ethyl α-bromoacetate,127 gave compound 122 in high yield. The use of 15-25 vol% DMPU as co-solvent proved to be crucial for successful coupling results and when DMPU was omitted the reaction proceeded slowly and sluggishly. With diene 122 in hand, we were pleased to find that the azepine ring could be formed by a ring-closing metathesis reaction using Grubbs’ 2nd generation catalyst128 (105, Figure 13, Chapter 4) in refluxing CH2Cl2 at high dilution (c = 0.005 M), which afforded compound 114 in excellent yield. When the reaction was run at higher concentrations (c > 0.01 M) the desired product was isolated along with inseparable dimerization by-products. Scheme 54. Construction of the Pyrrolo[1,2-a]azepine Core

NH

O

HOH

TBSCl, Im

CH2Cl2, rt98%

NH

O

HOTBS

BrNaHMDSDMF, -15 °C

N

O

HOH

2) TBAF THF, 0 °C92% (2 steps)

N

O

H

1) Swern2) 120, K2CO3 MeOH, rt92% (2 steps)

CH2Cl2/Hex-20 °C, 72%

N

O

HI

121

Pd(PPh3)4N

O

H

OEtO

(MeO)2P BrZnOEt

O O O

N2

B-I-9-BBN

105

CH2Cl240 °C, 92%

N

O

H

OEtO

120 121

1)

THF/DMPU50 °C, 78%

117 118 119

116115122

114

5.3.2. Lactonization and Completion The next key step in the synthesis was the stereoselective hydroboration of the trisubstituted olefin. To our disappointment, all attempts to install the C8-hydroxy group along with the correct C8-C9 trans-stereochemistry by treating β,γ-unsaturated ester 114 with a variety of hydroborating agents failed (Scheme 55). When using 9-BBN in THF at reflux, conditions used for selective hydroboration of β,γ-unsaturated amides,129 no reaction occurred. Treatment of 114 with BH3⋅THF at low temperature (T < −25 °C) gave no 127 Roberts, C. D.; Schütz, R.; Leumann, C. J. Synlett 1999, 819-821. 128 Fürstner, A.; Thiel, O. R.; Ackermann, L.; Schanz, H.-J.; Nolan, S. P. J. Org. Chem. 2000, 65,

2204-2207. 129 Vedejs, E.; Kruger, A. W. J. Org. Chem. 1999, 64, 4790-4794.

59

reaction, while increased reaction temperature gave a complex mixture of products. It has been shown that hydroboration of similar β,γ-unsaturated ester systems using BH3⋅THF can lead to reduction of the ester functionality, furnishing the corresponding primary alcohol.130 Therefore, in order to screen hydroboration conditions without the interfering ester reduction, 114 was reduced to alcohol 123 using LiBH4 in high yield. Alcohol 123 was then subjected to the same hydroboration conditions as before, as well as Evans’ catecholborane/SmI3 procedure.131 In these cases, either no reaction occurred or the amide was reduced to the corresponding amine along with the desired hydroboration of the olefin. It has been reported by others that hydroboration of hindered trisubstituted olefins in the presence of amides can be troublesome.132 Scheme 55. Failed Hydroboration Attempts

N

O

H

OEtO

Conditions

see textN

O

H

O

O

H

H

N

O

H

HOConditions

see textN

O

H

HO

Ox

HO

H

LiBH4THF, rt93%

114

123 In light of the failed hydroboration approach we were forced to modify our synthetic strategy. Instead we decided to install the butyrolactone moiety using a bromolactonization reaction (Scheme 56), a strategy that was successfully used by Mori and co-workers in their efforts towards 109.119b Elimination of the transient bromide to afford butenolide 124 followed by a stereoselective 1,4-reduction would intercept our originally planned late stage α-methylation. Scheme 56. Revised Retrosynthetic Analysis

N

O

O

O

H

H

H

N

O

H

H

1,4-Reduction/Methylation

Bromolactonization

N

O

H

OEtO

OO

(-)-Stemoamide (109)124 114

130 Simila, S. T. M.; Reichelt, A.; Martin, S. F. Tetrahedron Lett. 2006, 47, 2933-2936. 131 Evans, D. A.; Muci, A. R.; Stürmer, R. J. Org. Chem. 1993, 58, 5307-5309. 132 Enders, D.; Lenzen, A.; Backes, M.; Janeck, C.; Catlin, K.; Lannou, M.-I.; Runsink, J.; Raabe,

G. J. Org. Chem. 2005, 70, 10538-10551.

60

Classical reagents used in halolactonization reaction, such as I2/KI, Br2/CHCl3 or NBS/THF/AcOH, have shown to give poor results when subjected to β,γ-unsaturated acids and instead CuBr2 supported on alumina is generally the reagent of choice.133,134 When heated in a non-polar media, CuBr2 thermally decomposes into CuBr and Br2, which when activated by the alumina serves as a highly reactive and selective source of electrophilic bromine. Thus hydrolysis of ethyl ester 114 using LiOH gave acid 125, which upon exposure to CuBr2 on alumina at 65 °C in CHCl3 successfully gave the desired 5-endo-trig cyclization product (Scheme 57). In situ treatment of the transient alkyl bromide with triethylamine at room temperature effected elimination to butenolide 124 as a single diastereomer (65% yield from ester 114).119b The formation of the C8 β-epimer in the bromolactonization was gratifying, although it was not easily predicted from molecular models. Mori and co-workers speculate that the observed reaction outcome is due to the product stability.119b Scheme 57. Bromolactonization and Endgame

THF:MeOH:H2Ort

N

O

H

HO

MeI LiHMDS

THF -78 °C78%

N

O

H

OEtO

LiOH.H2O

N

O

H

OO

H

N

O

H

OO

H

H

O CuBr2 on Al2O3CHCl3, 65 °C

then Et3N, rt

65% (2 steps)

NiCl2 NaBH4

MeOH -30 °C90%

N

O

H

OO

H

H

114 125

109 126 124 The C8-C9 trans-stereochemistry required for 109 was then introduced via a stereoselective 1,4-reduction of butenolide 124 proceeding by hydride attack from the sterically most accessible β-face using NiCl2⋅6H2O and NaBH4,119b,c,e,135 which gave the known lactone 126 in high yield.119e,f,124 Finally, α-methylation of lactone 126 using literature conditions finished the synthesis of (−)-stemoamide.119e,f,124 Analytical data for 109 were in all aspects identical to those reported in the literature.119

133 Rood, G. A.; DeHaan, J. M.; Zibuck, R. Tetrahedron Lett. 1996, 37, 157-158. 134 For the preparation of CuBr2 on Al2O3, see: Kodomari, M.; Satoh, H.; Yoshitomi, S. Bull.

Chem. Soc. Jpn. 1988, 61, 4149-4150. 135 Satoh, T.; Nanba, K.; Suzuki, S. Chem. Pharm. Bull. 1971, 19, 817.

61

5.4. Conclusions

We have developed a 12-step, stereoselective total synthesis of (−)-stemoamide (109) in 20% overall yield starting from commercially available (S)-pyroglutaminol (117). The key features of the synthesis are (I) an iodoboration/Negishi/RCM-sequence for the construction of β,γ-unsaturated azepine derivative 114 and (II) a stereoselective bromolactonization/1,4-reduction strategy for the installation of the requisite C8-C9 trans-stereochemistry.

62

63

6. Concluding Remarks

This thesis deals with the development of new methods for the stereoselective synthesis of β-amino alcohols and the total synthesis of amino alcohol-containing natural products. Specifically, a highly diastereoselective Rh(II)-catalyzed 1,3-dipolar cycloaddition of carbonyl ylides to aldimines for the synthesis of syn-α-hydroxy-β-amino esters have been developed. An asymmetric version has also been realized by employing chiral α-methylbenzyl imines as dipolarophile yielding enantiomerically pure syn-α-hydroxy-β-amino esters. Furthermore, the use of chiral Rh(II) carboxylates furnishes the products in moderate enantioselectivity, which indicates that the reaction proceeds via a metal-associated carbonyl ylide. In addition, a 1,3-dipolar cycloaddition of azomethine ylides to aldehydes for the synthesis of the regioisomeric syn-α-amino-β-hydroxy ester has also been studied. Two different methods for the in situ generation of the ylides have been evaluated and the best results were obtain using an Ag(I)/PPh3 catalyst together with α-imino esters. Further studies on this topic have to be made in order to improve the methodology as a synthetically viable entry to this important compound class. Also, the asymmetric total syntheses of the amino alcohol-containing natural products D-erythro-sphingosine and (−)-stemoamide have been conducted. To conclude, there is a constant growing demand for new discoveries in the field of organic chemistry. In order to advance the field of complex molecule synthesis for the discovery and synthesis of for example new therapeutic drugs, the development of efficient and selective synthetic methods is of outmost importance and will hold a prominent position in organic chemical research.

64

65

Acknowledgements

Då var det hela över… men det hade inte gått utan hjälp och stöd från ett stort antal personer som ska ha ett stort TACK! Först och främst skulle jag vilja tacka min handledare, Peter Somfai, för att du antog mig som doktorand och för att delar med dig av din kunskap och ditt engagemang. Det har varit fem mycket lärorika år. Johan, Ki, Per och Peter, för er hjälp med kommentarer och korrekturläsning av avhandlingen. Daniel, Per och Janne – Philly/NYC. Per och Mari – San Fran → San Diego. Alla gamla och nuvarande medlemmar i PS-gruppen för en mycket trevlig och stimulerande arbetsmiljö. Per och Daniel, för er hjälp och värdefulla kommentarer på de olika projekten. Det var väldigt lärorikt som junior i gruppen. Tessie, för din hjälp med att bekämpa Europe på labbet! Pavel, Sebastian och Martina, för alla fikastunder. Daniel, Samir och Martin – “Crazy guys” i Licheng-gruppen. Lena Skowron, Ann Ekqvist och Henry Challis, för er hjälp med alla praktiska detaljer på avdelningen. Ulla Jacobsson, för ditt oändliga engagemang med allt fixande på avdelningen och för din hjälp med NMR:n. Aulin-Erdtman stiftelsen, Knut och Alice Wallenbergs stiftelsen och Kungliga Vetenskapsakademien, för resestipendier till konferenserna i Philadelphia och San Francisco. Prof. Tobias Rein och Doc. Jan-Erik Nyström, för möjligheten att jobba på AstraZeneca, Södertälje, under flytten av labbet. Examensarbetarna Marie Högstedt och Marcel Kienle, för er hjälp med 1,3-DC projektet. Emil Wanngren, för ditt arbete med stemoamide-syntesen. Polymergänget, Danne, Pelle och Andreas, kaffe på kurslab (priceless) och en och annan trivselöl (som alltid spårade ur!). Olle, Danne och Martin – gamla KTH-maffian, nu e det bara en kvar… Min familj, som alltid stöttat och ställt upp under alla år. Lisa, min älskling, det tog ”bara” 9 år men nu e jag äntligen klar på KTH!

66

67

Appendix A

The following is a description of my contribution to Publications I to V, as requested by KTH. Paper I: I contributed to the formulation of the research problems, performed the majority of the experimental work and supervised diploma worker Marcel Kienle, who prepared and reacted substrate 17k. I wrote the manuscript. Paper II: I contributed to the formulation of the research problems, performed the experimental work and wrote the manuscript. Paper III: I contributed to the formulation of the research problems, performed the experimental work and wrote the manuscript. Paper IV: I contributed to the formulation of the research problems, performed the experimental work and wrote the manuscript. Paper V: I contributed to the formulation of the research problems, performed the majority of the experimental work and supervised diploma worker Emil Wanngren. I wrote the manuscript.

68

69

Appendix B

This appendix contains experimental procedures and analytical data of compounds mentioned in the thesis but not reported in publications I-V. Chapter 2: (4R*,5S*)-Ethyl 3-benzyl-2-oxo-4-phenyloxazolidine-5-carboxylate (22). To a solution of amino alcohol syn-21a (3.8 mg, 13 µmol) in CH2Cl2 (0.5 mL) was added triphosgene (5.7 mg, 19 µmol) followed by i-Pr2NEt (4.5 µL, 26 µmol). The reaction mixture was stirred for 15 h and quenched with H2O (0.5 mL). Extrelut NT3®-workup followed by evaporation under reduced pressure gave the analytically pure oxazolidinone 22 (3.6 mg, 88%) as a clear oil: 1H NMR (400 MHz, CDCl3) δ = 7.42 (m, 3H), 7.31-7.22 (m, 5H), 7.13 (m, 2H), 4.91 (d, JAB = 15.1 Hz, 1H), 4.70 (d, J = 5.1 Hz, 1H), 4.51 (d, J = 5.1 Hz, 1H), 4.26-4.21 (q, J = 7.1 Hz, 2H), 3.69 (d, JAB = 15.1 Hz, 1H), 1.25 (t, J = 7.1 Hz, 3H); 13C NMR (100 MHz, CDCl3) δ = 168.4, 156.8, 136.9, 135.1, 129.62, 129.58, 129.0, 128.4, 128.2, 127.1, 78.0, 62.4, 61.8, 46.1, 14.2. (2S,3R)-3-((S)-1-Phenylethylamino)-3-(3-methoxyphenyl)propane-1,2-diol (51). To a solution of ent-38f (856 mg, 2.49 mmol) in THF (70 mL) was added LiBH4 (162 mg, 7.48 mmol) in one portion at 0 °C. The temperature was allowed to slowly reach rt and the mixture was stirred over night and was quenched by addition of NH4Cl (10 mL). The mixture was extracted five times with EtOAc, dried (MgSO4) and concentrated under reduced pressure. Flash chromatography (EtOAc + 2% MeOH + 1% NH4OH) of the residue gave 51 (711 mg, 95%) as a clear oil: [α]20

D = -148 (c 1.3, CHCl3); 1H NMR (500 MHz, CDCl3) δ = 7.36-7.31 (m, 2H), 7.30-7.21 (m, 4H), 6.85 (m, 1H), 6.76 (m, 2H), 3.81 (s, 3H), 3.63 (br s, 1H), 3.59 (q, J = 6.6 Hz, 1H), 3.51 (dd, J = 11.5, 2.5 Hz, 1H, 3.34 (m, 2H), 1.30 (d, J 6.6 Hz, 3H); 13C NMR (125 MHz, CDCl3) δ = 159.9, 144.2, 141.9, 129.8, 128.6, 127.3, 127.0, 119.7, 113.3, 112.8, 74.3, 64.1, 62.1, 55.2, 54.7, 25.2. tert-Butyl (1R,2S)-2,3-dihydroxy-1-(3-methoxyphenyl)propylcarbamate (53). A suspension of 51 (2.306 g, 7.65 mmol), (Boc)2O (3.520 mL, 15.3 mmol) and Pd(OH)2 (230 mg, 10% on C) in MeOH (40 mL) was stirred under a H2-atmosphere (balloon). After 21.5 h, the mixture was filtered through celite and concentrated under reduced pressure. Flash chromatography (CH2Cl2 + 5 → 10% MeOH) gave 53 (2.266g, 99.6%) as a white solid: 1H NMR (500 MHz, CDCl3) δ = 7.30-7.24 (m, 1H), 6.92-6.80 (m, 3H), 5.34 (br s, 1H), 4.79 (br s, 1H), 3.94 (br s, 1H), 3.80 (s, 3H), 3.56 (t, J = 5.7 Hz, 2H), 2.79 (br s, 1H), 2.77 (d, J = 3.7 Hz, 1H, 1.44 (s, 9H); 13C NMR (125 MHz, CDCl3) δ = 160.0, 156.7, 141.1, 129.9, 118.9, 113.0, 112.7, 80.3, 75.1, 63.7, 55.9, 55.3, 28.3.

70

tert-Butyl (R)-(3-methoxyphenyl)((S)-2,2-dimethyl-1,3-dioxolan-4yl)methyl carbamate (54). To a solution of 53 (2.261 g, 7.60 mmol) in DMF (70 mL) was added 2-methoxypropene (2.860 mL, 30.4 mmol) followed by p-TSA (72 mg, 0.38 mmol). The resultant mixture was stirred for 1.5 h at rt and then poured onto Et2O and washed with H2O. The aqueous layer was extracted once with Et2O and the combined organic layers were washed with brine, dried (MgSO4) and concentrated under reduced pressure. Flash chromatography (heptane:EtOAc 4:1) of the residue gave 54 (1.946 g, 76%) as a clear oil: 1H NMR (500 MHz, CDCl3) δ = 7.27-7.23 (m, 1H), 6.94-6.87 (m, 2H), 6.80 (dd, J = 8.0, 2.3 Hz, 1H), 5.26 (br s, 1H), 4.65 (br s, 1H), 4.34 (br s, 1H), 3.99 (t, J = 7.3 Hz, 1H), 3.80 (s, 3H), 3.78 (m, 1H), 1.48 (s, 3H), 1.41 (br s, 9H), 1.33 (s, 3H); 13C NMR (125 MHz, CDCl3) δ = 160.0, 155.6, 142.2, 129.6, 119.2, 112.8, 112.7, 109.8, 79.7, 78.2, 66.9, 55.4, 55.2, 28.3, 26.4, 25.1. tert-Butyl (R)-(5-methoxycyclohexa-1,4-dienyl)((S)-2,2-dimethyl-1,3- dioxolan-4-yl)methyl carbamate (55). To a solution of 54 (1.882 g, 5.58 mmol) in freshly distilled NH3 (30 mL) and Et2O (60 mL) was added EtOH (10 mL) followed by addition of small pieces of Na (1.280 g, 55.8 mmol) over 20 min at -78 °C. The resultant deep-blue mixture was stirred at -78 °C for 2 h and was then carefully quenched by addition of sat. NH4Cl (6 mL). The NH3 was carefully evaporated by allowing the mixture to reach rt and the residue was washed with brine, dried (MgSO4) and concentrated under reduced pressure. Flash chromatography (heptane:EtOAc 6:1 → 4:1) of the residue gave 55 (1.752 g, 93%) as a clear oil: 1H NMR (500 MHz, CDCl3) δ = 5.70 (s, 1H), 4.96 (br s, 1H), 4.60 (t, J = 2.9 Hz, 1H), 4.27 (t, J = 6.1 Hz, 1H), 4.08 (br s, 1H), 4.02 (dd, J = 8.2, 6.6 Hz, 1H), 3.70 (dd, J = 8.1, 7.2 Hz, 1H), 3.55 (s, 3H), 2.83 (m, 2H), 2.72 (t, J = 7.7 Hz, 2H), 1.43 (s, 12H), 1.33 (s, 3H); 13C NMR (125 MHz, CDCl3) δ = 155.7, 152.6, 133.0, 121.3, 109.6, 90.1, 79.5, 76.4, 66.9, 55.6, 54.0, 28.9, 28.4, 26.6, 26.3, 25.1. tert-Butyl (S)-3-(methoxycarbonyl)-1-((S)-2,2-dimethyl-1,3-dioxolan-4-yl)-2-oxopropylcarbamate (56). A solution of 55 (1.608 g, 4.74 mmol) in CH2Cl2 (60 mL) and MeOH (15 mL) was bubbled through with O3 using a fritted glass rod for 1.5 h at -78 °C. The reaction mixture was quenched by addition of dimethylsulfide (6.96 mL, 94.7 mmol) and the resultant mixture was slowly allowed to reach rt over night and then concentrated under reduced pressure. Flash chromatography (heptane:EtOAc 4:1 → 2:1) of the residue gave 56 (986 mg, 63%) as a clear oil: 1H NMR (500 MHz, CDCl3) δ = 5.28 (d, J = 8.0 Hz, 1H), 4.61 (dt, J = 6.4, 3.1 Hz, 1H), 4.40 (dd, J = 8.3, 2.8 Hz, 1H), 4.10 (dd, J = 8.6, 6.7 Hz, 1H), 3.74 (s, 3H), 3.62 (d, J = 4.1 Hz, 2H), 1.46 (br s, 9H), 1.42 (s, 3H), 1.32 (s, 3H).

71

Chapter 5: (S,Z)-1,2,6,7-Tetrahydro-9-(2-hydroxyethyl)-5H-pyrrolo[1,2-a]azepin-3(9aH)-one (123) To a solution of ethyl ester 114 (42.0 mg, 0.177 mmol) in THF (1.5 mL) was added LiBH4 (11.6 mg, 0.531 mmol) at 0 °C. After 1 h at 0 °C, the temperature was allowed to reach rt and the mixture was stirred over night. The reaction mixture was quenched by addition of NH4Cl (1.5 mL, sat. aq.) and extracted five times with EtOAc, dried (MgSO4) and concentrated under reduced pressure. Flash chromatography (5 → 10% MeOH in EtOAc) of the residue gave 123 (32.2 mg, 93%) as a clear oil: 1H NMR (500 MHz, CDCl3) δ = 5.57 (dd, J = 8.3, 5.2 Hz, 1H), 4.11 (t, J = 7.5 Hz, 1H), 4.02 (ddd, J = 13.8, 8.2, 3.2 Hz, 1H), 3.68 (t, J = 6.6 Hz, 2H), 2.89 (dt, J = 14.1, 8.2 Hz, 1H), 244-2.12 (m, 7H), 2.00-1.82 (m, 2H), 1.75-1.59 (m, 1H); 13C NMR (125 MHz, CDCl3) δ = 174.7, 138.0, 125.7, 64.1, 61.2, 39.3, 37.9, 30.4, 26.0, 24.8, 22.4; HRMS (FAB+) calcd for C11H17NO2 [M+H]+: 196.1332, found: 196.1332. 2-((S,Z)-2,3,5,6,7,9a-Hexahydro-3-oxo-1H-pyrrolo[1,2-a]azepin-9-yl)acetic acid (125) To a solution of ethyl ester 114 (110.0 mg, 0.463 mmol) in THF:MeOH:H2O (18 mL, 2:1:1) was added LiOH⋅H2O (39.0 mg, 0.927 mmol) at rt and the resultant mixture was stirred for 2 h. The reaction mixture was quenched by addition of HCl (5.5 mL, 1 M) and extracted five times with EtOAc to give 125 (106 mg, quant.) as a white solid: 1H NMR (500 MHz, CDCl3) δ = 5.73 (dd, J = 8.3, 5.4 Hz, 1H), 4.28 (t, J = 7.1 Hz, 1H), 4.08 (ddd, J = 13.7, 7.9, 3.7 Hz, 1H), 3.09-2.98 (m, 3H), 2.52-2.38 (m, 2H), 2.36-2.26 (m, 2H), 2.05-1.95 (m, 2H), 1.94-1.86 (m, 1H), 1.80-1.70 (m, 1H). 1H NMR showed that the sample contained traced of AcOH.

72