University of Groningen Total synthesis of mycolic acids and site-selective functionalization of aminoglycoside antibiotics Tahiri, Nabil

(1)

University of Groningen

Total synthesis of mycolic acids and site-selective functionalization of aminoglycoside

antibiotics

Tahiri, Nabil

IMPORTANT NOTE: You are advised to consult the publisher's version (publisher's PDF) if you wish to cite from it. Please check the document version below.

Document Version

Publisher's PDF, also known as Version of record

Publication date: 2019

Link to publication in University of Groningen/UMCG research database

Citation for published version (APA):

Tahiri, N. (2019). Total synthesis of mycolic acids and site-selective functionalization of aminoglycoside antibiotics. University of Groningen.

Copyright

Other than for strictly personal use, it is not permitted to download or to forward/distribute the text or part of it without the consent of the author(s) and/or copyright holder(s), unless the work is under an open content license (like Creative Commons).

Take-down policy

If you believe that this document breaches copyright please contact us providing details, and we will remove access to the work immediately and investigate your claim.

Downloaded from the University of Groningen/UMCG research database (Pure): http://www.rug.nl/research/portal. For technical reasons the number of authors shown on this cover page is limited to 10 maximum.

(2)

531024-L-sub01-bw-Tahiri 531024-L-sub01-bw-Tahiri 531024-L-sub01-bw-Tahiri 531024-L-sub01-bw-Tahiri Processed on: 3-5-2019 Processed on: 3-5-2019 Processed on: 3-5-2019

Processed on: 3-5-2019 PDF page: 1PDF page: 1PDF page: 1PDF page: 1

Total Synthesis of Mycolic Acids and

Site-Selective Functionalization of

Aminoglycoside Antibiotics

Nabil Tahiri

(3)

The work described in this thesis was executed at the Stratingh Institute for Chemistry, University of Groningen, The Netherlands.

The authors of this thesis wish to thank the Nederlands Organization for Scientific Research for funding.

Cover design by Leonardo Miluccio (website: www.leonardomiluccio.myportfolio.com email: leonardomiluccio@gmail.com)

Printed by Ipskamp Printing BV, Enschede, The Netherlands ISBN: 978-94-034-1684-7 (print)

(4)

Total Synthesis of Mycolic Acids and Site-Selective

Functionalization of Aminoglycoside Antibiotics

Proefschrift

ter verkrijging van de graad van doctor aan de Rijksuniversiteit Groningen

op gezag van de

rector magnificus prof. dr. E. Sterken en volgens besluit van het College voor Promoties

De openbare verdediging zal plaatsvinden op vrijdag 7 juni 2019 om 16:15 uur

door

Nabil Tahiri

geboren op 13 juli 1989 te Breda

(5)

Promotores Prof. dr. A. J. Minnaard Prof. dr. M. D. Witte Beoordelingscommissie Prof. dr. B. L. Feringa Prof. dr. S. R. Harutyunyan Prof. dr. H. Hiemstra

(6)

“People should continue to work hard at organic synthesis – it’s hard,

exacting, frustrating, but always fascinating….

synthesis can create whole new worlds.”

(7)

(8)

Chapter 1:

Mycolic Acids from Mycobacterium

tuberculosis

(13)

2

1.1 Mycobacterium tuberculosis

1.1.1 Basic facts

The Mycobacterium genus, covering over 150 recognized species,[1]_{is mostly known}

for its members that are pathogenic to humans. Of these environmental bacteria,

Mycobacterium tuberculosis (M. tb), which causes tuberculosis (TB), is by far the most

lethal member of the family. Other well-known pathogenic members include M.

ulcerans, which is responsible for Buruli ulcer and M. leprae, also known as Hansen’s

bacillus spirally, which causes leprosy. Although these are the three main causative agents for mycobacterium infections worldwide, the amount of newly reported infections for M. leprea (about 250k yearly)[2]_{and M. ulcerands (5-6k yearly),}[3]_{are in}

stark contrast with the number of M. tuberculosis infections. On its own, M.

tuberculosis already constituted for 10 million estimated infections in 2017.[4]_Two

thirds of these infections were from eight countries: India (27%), China (9%), Indonesia (8%), the Philippines (6%), Pakistan (5%), Nigeria (4%), Bangladesh (4%) and South Africa (3%). Although the severity of TB infections varies on a national level, clearly, TB is a major burden on most developing countries (Figure 1). Especially in regions where HIV infection rates are relatively high, such as in African nations, TB infection results in higher mortality rates, due to the patient’s compromised immune response.

Figure 1. Estimated TB incidence rate by WHO in 2017.

Although the total number of people infected with M. tb is estimated by the WHO at around 1.7 billion (23% of the world population!), in most patients TB manifests itself in the form of a latent infection. Especially in developed countries, a latent infection usually goes unnoticed. These patients do not show the typical symptoms (vide infra) associated with TB (Figure 2a), because the tuberculosis bacilli in their body are not active and are fortunately not able to spread this highly contagious disease (in Dutch:

(14)

3

gesloten tbc). However, since 10% of the patients with latent TB develop the active form, which is very contagious (in Dutch: open tbc), treatment should start as soon as TB has been diagnosed. Typical symptoms (Figure 2a) include persistent coughing (with blood), fevers, night sweats, and weight loss (as a result of reduced appetite). Unfortunately, the treatment is not straightforward, as is reflected by the high mortality rate of 1.6 million in 2017 alone, making it the leading cause of death by a single infectious agent worldwide (more than HIV/AIDS).[4]_{For patients with latent TB,}

several factors (Figure 2b) such as a HIV infection, cancers, diabetes, alcohol consumption and smoking can significantly increase the likelihood to develop active TB. The latter has been demonstrated to double the mortality rate by active TB in men in India.[5]

Since the physiology of M. tuberculosis highly depends on oxygen, the mammalian respiratory is often its primary target. However, upon spreading to other organs (extrapulmonary tuberculosis), M. tuberculosis can nestle in the pleura (in tuberculous pleurisy), the central nervous system (in tuberculous meningitis), the lymphatic system (in scrofula of the neck), the genitourinary system (in urogenital tuberculosis), and the bones and joints (in Pott disease of the spine).[6]

1.1.2 Treatment and detection

If diagnosed in time, the treatment of latent TB is usually executed with a typical regimen consisting of either isoniazid, or isoniazid combined with rifapentine (first-line drugs) over several months. In the case of active TB, treatment is less straightforward, and usually requires the use of the more toxic drugs pyrazinamide and/or ethambutol (also first-line drugs) in combination with the aforementioned ones.[7]_{Unfortunately, as}

for most bacteria, the rise of the antibiotic resistance also resulted in reduced activity of the aforementioned first-line antibiotics against multi- and extensively drug resistant TB over the past decades. As a consequence, harsher treatments are applied for combatting these pathogens, and this involves the consecutive addition of more toxic antibiotics to the treatment stack. In a typical regimen against resistant TB, a combination of at least three additional drugs (second-line drugs), on top of the first-line drugs, are applied. This combination contains at least one member of the fluoroquinones[8]_{(inhibits DNA}

A) B)

(15)

4

gyrase, an enzyme necessary for the separation of replicating DNA), aminoglycosides[9]

(target bacterial ribosome, thereby hampering protein synthesis) and oxazolidinones[10]

(also targets bacterial ribosome, resulting in disruption of the protein translation process) and are administered over an extended period up to two years. Bedaquiline was approved by the FDA in 2012,[11]_{as part of the accelerated approval program, and is}

used nowadays as a last resort antibiotic against MDR-TB and XDR-TB in patients for whom the abovementioned drugs are not effective. Needless to say, a combination of these drugs, which on their own can have horrible side effects such as: vomiting, diarrhea, tendon ruptures, serotonin syndrome, psychosis and hearing problems, can affect the patient’s living standard as such, that in some cases continuation of the treatment is irresponsible.

As is the case for the treatment, the detection and confirmation of M. tuberculosis can be just as cumbersome and time consuming. When a pneumonia-like illness persist for more than three weeks, a TB infection should be considered as one of the major suspects. Unfortunately, the only method allowing for TB detection with high certainty, is via culturing of a sputum sample and direct observation by microscopy. Because of

M. tuberculosis’ extremely slow growth rate, this process can take from 4 up to 8

weeks. Recently, the microscopic observation drug susceptibility (MODS) assay has been proven to be a faster and more sensitive alternative that current culture based test, and additionally allowing for a better resistance type determination.[12]_{Part of the}

shortened diagnosis time, is due to the microscopic observation of TB directly from sputum. Since M. tuberculosis can either show up as G+ as G- with gram staining, the acid fast staining is a more widely applied technique. Although not impeccable, the Mantoux test and chest X-rays have been used in conjunction with other tests.

1.2 The cell envelope

Parts of the difficulty in the diagnosis and treatment of TB can be attributed to the unusually complex build-up of the cell envelope of M. tuberculosis. The influx of drugs and the natural host defense mechanisms can be greatly influenced by this heavily fortified cell wall. The envelope can be subdivided in three major segments: the inner membrane (also known as plasma membrane), the inner core and the outer membrane. Due to decades of extensive research a more complete picture (as shown in Figure 3) of this complex multiple layered architecture has been constructed thanks to the evolution of modern experimental techniques.

At the basis of the envelope lies the inner membrane, which resembles the typical bacterial membrane, and mainly consists of an abundance of glycerophospholipids which are aggregated into a bilayer. This bilayer is made up of mainly phosphatidylinositol (PI), phosphatidylglycerol, phosphatidylserine (PS), phosphatidylethanolamine (PE), cardiolipin (CL), and mannosylated forms of PI (PIMs).[13]_{Next to these glycerophospholipids, the inner membrane consists of minor}

(16)

5

amounts of glycolipids such as lipomannan (LM) and ManLAM. The latter is non-covalently integrated into the bilayer with its lipid moiety, but its polysaccharide segment penetrates deeply into the inner core, bridging both the peptidoglycan (PG) and arabinogalactan (AG) layers, reaching all the way up to the outer membrane. These complex inner plasma lipids fulfill a critical role in the selective permeability of the cell, ATP regulation, DNA replication and electron transport. The inner membrane also consists of a variety of isoprenoids, which serve as carriers during the biosynthesis of some of the most critical components of the cell envelope.[14]

Figure 3. Schematic overview of the cell envelope of M. tuberculosis. Light blue symbols

represent arabinose residues, red symbols represent galactose residues, brown symbols represent mannose residues, and black circles represent glucose residues. D-arabino-D-mannan, D-glucan, and D-mannan are capsular polysaccharides. Mycolic acid chains are shown in dark green. LM, lipomannan; ManLAM, mannose- capped lipoarabinomannan; AG, arabinogalactan; PG, peptidoglycan.[13]

On top of the inner membrane lies the non-covalently bound insoluble inner core, which is subdivided in two segments: the peptidoglycan (PG) and the arabinogalactan (AG). Closest to the inner membrane lies the peptidoglycan (PG) layer. This layer is made up of a polysaccharide backbone consisting of N-acetyl-α-D-glucosamine (GlcNAc) and

(17)

6

modified muramic acid (Mur). The backbone is heavily cross-linked with peptide linkers, resulting in extra rigidity. Compared to, for example, E. coli, the degree in crosslinks in Mycobacteria is 50% higher. Its main function is to provide structural rigidity and the regulation of the osmotic pressure. Two opposing theories regarding the three-dimensional structure of the PG layer have been proposed in literature, in which the PG layer is present in either a parallel (as shown in Figure 3) or perpendicular arrangement with respect to the inner membrane. Although the latter seems to have gained the most support,[15]_{the parallel theory has recently attracted more acceptance}

after three-dimensional NMR studies in solution.[16]

The AG is connected to PG segment with the intermediacy of a single disaccharide linker consisting of a rhamnosyl residue attached to an N-acetylglucosaminosyl-1-phosphate residue. The galactan backbone consists of repeating dimeric galacto-furanose units that are sequentially connected at C5 and C6. At the 8th_{, 10}th_{and 12}th

positions, the backbone is branched with an arabino-furanose polysaccharide, containing a length of 31 monosaccharides. These branches span towards the outer membrane, and are covalently connected to the outer membrane.

Covalently attached to the AG layer via ester bonds, are the α-alkyl, β-hydroxy long-chain (C70–C90) fatty acids, known as the mycolic acids. These lipids are at the center

stage of the bilayered outer membrane, and fulfill a prominent role in the formation of this highly apolar and impenetrable layer. Because of the high abundance of mycolic acids, the outer membrane is often referred to as the mycomembrane. Seminal work by Daffé et al.[17]_{demonstrated that a single mycolic acid molecule is able to fold in order}

to form a more compact zipper-like structure (Figure 4). This particular arrangement of mycolic acids results in a higher lipid density in the outer membrane compared to, for example, corynomycolic acids in Corynebacteria. Compared to mycolic acids, corynomycolic acids lack both cyclopropyl rings and contain considerably shorter tails. However, both fatty acids contain the same α-alkyl β-hydroxycarboxylic acid segment. Mycolic acids make up the major part of the outer membrane, and can be found covalently bound to arabino-furanose residues or as non-covalently bound trehalose monomycolate (TMM), trehalose dimycolate (TDM), glycerol monomycolate (GroMM) and glucose monomycolate (GMM).

(18)

7

Figure 4. Zipper model of the OM of Mycobacteria. (A) Mycobacteria. (B) Corynebacteria.

Hydrocarbon chains of the lipids are drawn to scale. Black, mycolic acid; dark blue, phospholipids (16- to 18-carbon-long chains); dark gray, peptidoglycan-arabinogalactan; light blue, GPL; light gray, porin; orange, trehalose dimycolate; red, glucose monomycolate. Mycolic acids and trehalose mycolates are folded. An unfolded mycolic acid (in black) is shown in panel A. It is too large to be accommodated in the OM.[17]

The outer membrane contains, besides mycolic acids, other lipids such as phospholipids, diacyltrehaloses (DAT) and polyacyltrehaloses (PAT). The latter two trehaloses are esterified to iterative methyl branched fatty acids. Unlike mycolic acids, these glycolipids are not essential for the pathogens survival. However, DAT and PAT are only found in virulent strains. The final layer, on top of the outer membrane is called the capsule. This layer is loosely bound to the outer membrane and consists of polysaccharides, protein and only 3% lipid.

1.3 Mycolic acids

Lipids play an important role in the survival and pathogenesis of M. tuberculosis. A plethora of lipids are applied in a dynamic interplay with the host during different stages of infection. A systematic review of these cell wall lipids and their precise role is outside the scope of this work but has been provided before by others.[18–21]_{In this}

section, the different classes, biosynthesis, and stereochemical elucidation of mycolic acids will be discussed.

1.3.1 Structures and classes

It is almost a century ago since a variety of lipids from M. tuberculosis was isolated by Anderson.[22]_{One class of compounds present in these isolates comprised the mycolic}

acids, one of the most important lipids in the physiology of M. tuberculosis. These long chain α-alkyl β-hydroxy fatty acids, as assigned by Asselineau and Lederer,[23]_{are not}

limited to a single molecular structure, and coexist as a mixture of different classes depending on the species. Pyrolysis of these lipids provides a meroaldehyde and a fatty acid. Both vary in length, and the ratio of homologues is a characteristic for the strain by which they are produced. The meroaldehyde contains two functionalities. These functionalities are referred to as the proximal (closest to the carboxylic acid) and distal (furthest away from carboxylic acid) positions. Based on the substitution pattern on the

(19)

8

proximal and distal positions, mycolic acids are categorized in three classes: the α-, the methoxy- and the ketomycolic acids (Scheme 1). These are separable by chromatography, but the homologues within each class have very similar physical properties, and are therefore extremely difficult to separate. The separation of homologues of α-mycolates was first established by Qureshi and Takayama on analytical scale in 1978.[24]

The α-mycolic acids are the highest in abundance, and make up more than 70% of the isolatable mycolic acids. They contain two cyclopropyl groups, of which both are exclusively found with cis stereochemistry (Scheme 1). The size of the α-mycolic acids ranges between 74 and 80 carbons in total. The methoxy- and ketomycolic acids make up the rest of the amount, and contain an α-methyl methoxy and α-methyl keto moiety at the distal position, respectively.[25]_{The cyclopropyl functionality is positioned at the}

proximal position in both classes, and can be either cis or trans. In the case of a trans cyclopropanation, an α-methyl group is present. These mycolic acids are slightly bigger compared to the α-mycolic acids, and contain around 84 to 88 carbon atoms in total.[26]

1.3.2 Biosynthesis

Since mycolic acids are essential for the survival of M. tuberculosis, inhibition of the biosynthesis of these lipids has shown to be a viable strategy in order to combat this pathogen.[27]_{Therefore, the elucidation of the key enzymes, involved in the biosynthesis}

of mycolic acids, has been the topic of many studies. This allowed for a better understanding of their role in the physiology of M. tuberculosis, but moreover, might also lead to the identification of novel drug targets. As of today, the whole biosynthetic pathway has largely been elucidated, and will be concisely summarized in this section. An excellent review regarding the mycolic acid biosynthesis was published in 2005 by Takayama.[28]

The key-players involved in the biosynthesis, are the FAS-I and FAS-II enzymes (Fatty Acid Synthases). The FAS-I enzyme shows great resemblance with the eukaryotic variant and is responsible for de novo synthesis of fatty acids using acetyl-CoA and malonyl-CoA. This multifunctional enzyme is involved in the synthesis of fatty acids in a bimodal fashion. However, unlike in humans where this enzyme results in palmitic acid (C16H32O2), the FAS-I enzyme in mycobacteria can provide elongated fatty acids

ranging from 20 up to 26 carbons.[28]_{The biosynthesis starts off with the condensation}

of acetyl-CoA and malonyl-CoA in the FAS-I cycle (Scheme 2). Both acetyl and malonyl moieties are transferred to the enzyme. Then, after activation of the malonyl, resulting in the loss of CO2, both parts are combined and form a β-ketoester. The ketone

is removed, via reduction to the corresponding alcohol, elimination of water and subsequent reduction of the double bond. This results in the overall elongation of the chain by two carbons. This process is then repeated until the desired length is

(20)

Sc h em e 1. Ov er vi ew o f th e d iff er en t my co lic acid class es isolated f ro m M. tu b ercu lo sis. Note t h at o n ly o n e h o m o lo gu e is sho w n , b u t n atu ral sa m p les are obt ai ne d a s a hom ol og ue s m ix tur e, v ar yi n g i n va lu e f o r a ,b,c a nd d. T h e py ro ly si s o f m ycol ic a ci ds l ea ds t o a m er o al dehy de a nd a f att y acid .

(21)

10

achieved and will result in varying mixtures of homologues depending on the strain (only one length is shown in Scheme 2). These products can be used as substrates in the FAS-II system (Step 1a), resulting in elongation into the meromycolic moiety, or will form the basis of the α-branch (vide infra) (Step 1b).[28]

The eicosanoic-S-CoA (C20H40O2 fatty acid), prepared by the FAS-I (Step 1a) is then

transferred to the FAS-II system with the intermediacy of mtFabH (a β-ketoacyl-ACP synthase III), the key enzyme which links both systems together.[29,30]_{By the action of}

this enzyme, eicosanoic-CoA is reacted with malonyl-SACP, resulting in a β-ketoacyl intermediate, extended with two carbons. When the cis double bonds, which are necessary for introducing the cyclopropane and oxygenations, are exactly introduced in the FAS-II sequence is unknown. Takayama proposed that after the first two carbon extension by mtFabH, the β-ketoacyl intermediate is reduced to the trans isomer with the use of β- hydroxyacyl-ACP reductase and β-hydroxyacyl-ACP dehydrase. At this point the trans double bond is isomerized to a cis double bond, and moved down the chain to the required location of the distal functionality. The module responsible for this process was called FAS-IIA by Takayama. After a single cycle through the FAS-IIA module, the cis-Δ3_-C

22:1-S-ACP product is carried through an iterative two carbon

elongation by the FAS-II module (Step 2), in a similar fashion to the FAS-I module. This involves five cycles for the α-mero-acid and eight cycles for the methoxy- and ketomero acids, resulting in cis-Δ13_-C

32:1 -SACP (x = 10 in Scheme 2) and cis-Δ19-C38:1

-SACP (x ൌ 16 in Scheme 2), respectively. At this point another cis double bond, which will form the basis of the proximal functionality, is installed in a similar way to the distal olefin (Step 3). First, C2 elongation is achieved by the formation of a β-ketoacyl intermediate, which is reduced to the β-hydroxyacyl, followed by dehydration and isomerization to the cis isomer. During this process the olefin moves one carbon towards the distal direction. This second C2 elongation/isomerization process is performed by FAS-IIB, as proposed by Takayama, but no evidence for two independently operating enzymes was supplied.[28]_{The product at this stage re-enters}

the FAS-II module (Step 4), and elongation of the lipids proceeds until the required length is achieved. This results in cis,cis-Δ19,31_-C

50:2-SACP (FAS-II product x ൌ 10 and

y ൌ 17) in the case of the α-mero acid, cis,cis-Δ19,37_-C

56:2-SACP (x ൌ 16 and y ൌ 17) for

the methoxymero acid, and cis,cis-Δ21,39_-C

58:2-SACP (x ൌ 16 and y ൌ 19) for the

ketomero acid. KasA and KasB are the enzymes involved in the formation of the β-ketoacyl species in de FAS-II cycle. Since they show different chain-length specificities, it is believed that KasA is mainly responsible for the first part of the iterative FAS-II process,[31]_{and KasB for the second FAS-II process.}[32]

(22)

Sc h em e 2. B ioc he m ica l sy nt he si s of m et h oxym yc o li c a ci d. T h e p ar t s h ow n i n b lu e o rigin ates f ro m th e FA S -I, an d in b lac k f ro m th e FA S -II p ath w ay . T h e sc h em e is adap ted f ro m Tak ay am a [28] a nd Ma rr ak chi . [26] α -: x ൌ 10 a nd y ൌ 17 , m et h oxy -: x ൌ 16 a nd y ൌ 1 7 , ket o -: x ൌ 16 and y ൌ 19.

(23)

12

It is known from experimental evidence that at this point the cyclopropane rings are introduced in the meromycolic lipid (Step 5). This is done by a class of methyl transferases which use SAM as a cofactor. The distal and proximal cyclopropanes in α-mycolates (1) are introduced by MmaA2 and PcaA, respectively (Scheme 3).[33][34]_The

oxygenation at the distal position in methoxy- and ketomycolic acids, is a result of syn methyl hydroxylation of the cis olefin. The resulting secondary alcohol 2, the common intermediate for the methoxy- and ketomycolic acids, can either be oxidized to give keto-mycolic acids (3), or methylated to result in methoxymycolic acids (4). The proximal cis cyclopropyl is most probably introduced by the action of MmaA2 for methoxymycolic acids. Although MmaA2 null mutants did show to accumulate unsaturated methoxymycolic acids, some intact methoxymycolic acids (with cyclopropyl incorporated) were observed. NMR showed that the ratio of cis/trans cyclopropyl in methoxymycolic acid reduced by half compared to the wild-type (from 5:1 cis/trans to 10:1 cis/trans), but the ratio in ketomycolic acid was unaffected.[33]_This

suggests that probably multiple enzymes can introduce the proximal cyclopropyl group in methoxymycolic acids. CmaA2 might be able to assist in the cyclopropanation of the proximal cis olefin.[35]_{Finally, the trans cyclopropyl at the proximal position (in}

methoxy- and ketomycolic acids) is introduced via the intermediacy of methylated trans olefin 5 formed by MmaA1. This trans olefin is in turn converted to the trans cyclopropyl 6 by CmaA2.

Scheme 3. The enzymes involved in the cyclopropanation and oxygenation of mycolic acids.

The final step in the biosynthesis is the combination of the α-branch (from Step 1b) and the meromycolic acid (from Step 5), synthesized by the FAS-I and FAS-II systems, respectively. This is achieved via a Claisen-type condensation. In order to enhance the nucleophilic properties, the α-branch is activated by the installation of a carboxylate by AccD4 and AccD5 (Step 7). Then, the meroacyl-S-ACP is transferred to a meroacyl- AMP by the FadD32 enzyme (Step 6). The Claisen condensation takes place by the Pks13 enzyme, resulting in the mycolic motif (Step 8). The final step is the reduction of the β-ketone to the corresponding alcohol by a reductase, which still has to be defined.

(24)

13

1.3.3 Stereochemical elucidation

For the elucidation of the absolute stereochemistry of the mycolic acids, optical rotation has mostly been the method of choice. Because of the remote distance between the stereocenters, they do not influence each other’s optical activity, and can therefore be subtracted and added, in order to determine the relative contribution of each moiety to the overall specific optical rotation. This method dates back to a hypothesis of Van’t Hoff and is known as the principle of optical superposition.[36]_{Therefore, the absolute}

stereochemistry of the segments can be deduced by comparison of the optical rotation of the natural product with those of model compounds with predetermined absolute stereochemistry. A challenge in applying this strategy is that the specific rotation normally is determined in degrees milliliter gram-1_dm-1_{. Since the specific rotation for a}

segment of a compound depends on the number of molecules in the sample rather than the mass concentration, the optical rotation of the natural product and model compounds that differ widely in molecular weights often cannot be compared directly. The specific rotations have to be calculated in mol/100 ml instead of the usual g/100 ml (see formula in Scheme 4) for a reliable comparison, which is known as the specific molar rotation ([Φ]D). Using this strategy, the relative contribution to the optical rotation has been

determined for the α-alkyl β-hydroxycarboxylic acid and the α-methyl methoxy segments.

The α-alkyl β-hydroxyl segment was determined to be R,R by Asselineau et al.[37]_in

1970 for α-mycolic acid. Previously, the same authors determined the absolute stereochemistry of the closely related corynomycolic acids to be R,R. Because the cyclopropane rings and the length of the aliphatic chains did not influence the optical rotation, and both lipids showed a comparable specific molar rotation, the authors concluded that mycolic acids possessed the same absolute stereochemistry (Scheme 4). Furthermore, the change in specific molar rotation of the diol (obtained by reduction of the carboxylic acid), the methyl ester and chemically epimerized α-mycolic acid was identical to their corynomycolic acid counterparts. More evidence was provided by Moody and coworkers,[38]_{who showed that the R,R stereochemistry in mycolic acids}

was critical for T cell recognition and the generation of an immune response.

The specific molar rotation for the α-alkyl β-hydroxyl segment in α-mycolic acid was determined by Asselineau et al.[37]_{to be + 40. The specific rotation, and therefore also}

the specific molar rotation of natural methoxymycolic acid methyl ester was determined by Minnikin and Polgar[39]_{and Asselineau et al.}[37]_{to be 0 (Scheme 4). Since the}

proximal cyclopropyl ring does not significantly contribute to the overall specific rotation, the specific molar rotation of the αmethyl methoxy moiety has to be around -40, in order to achieve an overall specific molar rotation of 0. The authors were able to assign the absolute stereochemistry for the α-methyl methoxy moiety by preparing the model structures 7 and 8 shown in Scheme 4, which contain either a syn or anti configuration for the α-methyl methoxy, respectively. The specific molar rotation of 7 was + 48, while the value obtained for 8 was + 4.8. Because the amplitude of the optical

(25)

14

rotation of 7 was in the expected range, but with the opposite sign, the authors concluded that the stereochemistry at this position must be S,S. Based on these values, an anti stereochemistry would have had negligible contribution to the specific molar rotation, since the experimentally determined value for 8 was only + 4.8.

Scheme 4. Schematic overview of the strategy used by Asselineau et al. for the determination of

the absolute stereochemistry. The relative stereochemistry of the cyclopropyl function is cis. [Φ]D

= the specific molar rotation which is calculated in mol / 100 ml.

Because the isolated cyclopropyl rings do not contribute in a significant degree to the specific molar rotation, the α-methyl methoxy moiety could be determined by simply subtracting the contribution of the α-alkyl β-hydroxyl segment from the overall specific molar rotation. Clearly, as the optical rotation of the cyclopropyl ring is negligible and can therefore not be used, its absolute stereochemistry still remains to be determined. However, by NMR the relative stereochemistry was determined to be cis.

(26)

15

It needs to be pointed out that although the evidence for the stereochemistry of the α-methyl methoxy moiety seems convincing, the principle of optical superposition is nowadays no longer used. Optical rotations are sensitive to a number of factors (temperature, solvent, concentration) and a requirement for optical superposition is, as mentioned before, that the stereocenters do not influence each other. Although the stereocenters in mycolic acids are separated by long spacers, the influence of the conformation of the molecule on the optical rotation is not known.

1.3.4 Previous total synthesis

The total synthesis of each member of the three mycolic acid classes (i.e. α-[40,41]

methoxy-[42,43]_{and ketomycolic acids}[43,44]_{) has been achieved earlier by Baird and}

coworkers, and has been reported in a series of papers. Since the methoxymycolic acids are the main subject of this work, and the reported total syntheses of all three classes are quite similar, only the methoxymycolic acids will be discussed in this section.

In 2007, Baird and coworkers published the synthesis of several cis methoxymycolic acid diastereomers.[42]_{Although the authors started with the synthesis of both}

enantiomers of the cyclopropyl and α-methyl methoxy moieties, which should therefore result in four diastereomers, they decided to complete only two out of the four intended diastereomers. Surprisingly, the authors did not specify why the synthesis for the other diastereomers was not completed. However, these late stage intermediates were used to determine the optical rotations.

The synthesis started with compound 9, a known literature compound easily prepared from mannitol in three steps (Scheme 5). The synthesis of the enantiomer of 9 was accomplished by starting from L-gulonolactone. The methyl branch in 10 was installed by applying a highly diastereoselective conjugate addition of MeLi to 9 at -78 °C. With the absolute stereochemistry set for both enantiomers of 10, the most distal lipid part of the meromycolic moiety was installed using a modified Julia-Kocienski olefination, and a subsequent catalytic hydrogenation of the obtained mixture of cis and trans olefins. The transformation from 11 to 12 was realized in four steps, which established the incorporation of the first half of the long aliphatic C16 linker between the α-methyl

methoxy motif and the cyclopropyl ring. In order to connect the remaining part, a second Julia-Kocienski olefination was necessary, resulting in 14. At this point, another four steps were necessary to result in compound 15.

As mentioned before, the stereochemistry of the cyclopropyl in the natural product is still undefined. Therefore, it makes sense to prepare both enantiomers in order to aid in the structure elucidation. Furthermore, by making both diastereomers of the cyclopropyl one could investigate whether the stereochemistry has any effect on the biological activity, and verify its negligible contribution to the specific rotation. The authors started with both enantiomers of 17, which were reported earlier.[45]_{Aldehyde 17 was}

(27)

16

PCC oxidation of the remaining alcohol. The aldehyde was then elongated with another Julia-Kocienski olefination, followed by hydrogenation of the newly formed double bond. The combination of both enantiomers of 15 and 18 should eventually result in four diastereomers. However, only three diastereomers (20a-c) were prepared, for reasons not specified by the authors. It is highly remarkable that only one enantiomer of the cyclopropyl building block 18 was carried through the synthesis. Especially, since it is the stereochemistry of the cyclopropyl moiety that has not been determined. Therefore, this fragment of the molecule would benefit the most from a total synthesis. Of these three enantiomers, again only two (20a and 20c) were connected to the protected α-alkyl β-hydroxy motif 21. Since the stereochemistry at these two centers has been determined to be present in the R,R configuration in the natural material,[37]_only

one enantiomer of this fragment was prepared. Fragment 21 was reported previously in the synthesis of α-mycolic acid,[41,46]_{and required 14 steps in total, starting from}

(R)-aspartic acid (synthesis not shown in Scheme 5). Having established 22a and 22c, the authors did not continue with the synthesis, which would require two additional steps. Instead, they compared the optical rotations of these derivatives with a sample of the methyl ester of the natural material. Remarkably, the paper referred to by Baird and coworkers for the reference optical rotations, did not contain any determined optical rotations of structures 22a and 22c.

The third and remaining diastereomer 20b was not connected to the α-alkyl β-hydroxy motif using 21. Instead, TBS ether 27 was used, which according to the authors could be deprotected in a more selective manner. However, the synthesis of this fragment (the synthesis is shown in Scheme 5) was more complex compared to its predecessor 21, and required a total of 16 steps to prepare.[47]_{The stereochemistry was installed by}

subjecting α,β-unsaturated ester 23, which could be obtained in six steps, to a Sharpless asymmetric dihydroxylation, yielding diol 24. The undesired alcohol was removed in two steps, and allylation at the same position resulted in the installation of the α-branch. From 26 it took six subsequent steps to arrive at aldehyde 27. Surprisingly, the authors only coupled fragments 20a and 20b to 27, in which only the stereochemistry at the α-methyl methoxy fragment varied, resulting in the late stage protected mycolic acids 28a and 28b. TBS deprotection and hydrolysis of the methyl ester resulted in mycolic acids

29a and 29b, the only two completed mycolic acids in this synthesis. The optical

rotation of 29b corresponded to that of the natural product. However, by not preparing both diastereomers of the cyclopropyl, in which the S,S α-methyl methoxy fragment was kept constant, the authors risked eventually not having achieved the synthesis of the natural isomer.

(28)

17

Scheme 5. A concise overview of the asymmetric total synthesis of methoxymycolic acid

(29)

18

1.4 Outline of the thesis

Unmistakably, the eradication of pathogenic bacteria, such as M. tuberculosis, is one of the greatest challenges that humanity is facing. This, of course, is closely linked to the rising antibiotic resistance, which as a result, will cause an increase of deaths from bacterial infections in the coming decades. As is evident from this chapter, mycolic acids play a prominent role in the physiology of M. tb. These molecules have been isolated and have had their molecular structure partially characterized since a few decades ago. Furthermore, extensive research by biochemists has illuminated the mechanisms by which these molecules are prepared by M. tb and their contribution to the rigidity of the cell envelope. However, we still lack the knowledge about critical details regarding the role of mycolic acids in host-pathogen interactions.

Since mycolic acids are antigenic lipids, their exploitation in vaccine development and improved diagnostic tools are serious options. Although the molecular structures of mycolic acids have been largely known since the late 1960s,[37,39]_{and these molecules}

have been synthesized before (section 1.5), the absolute stereochemistry of the cyclopropyl rings has unfortunately still not been unambiguously determined. Clearly, a total synthesis of these lipids could push the field of TB research forward on multiple aspects: 1) synthetically pure samples of mycolic acid and its esters would allow biologists to study their role in the human-pathogen interaction in more detail. This could eventually lead to novel vaccines and diagnostic assays. These synthetic products are more advantageous over natural samples, the latter of which are isolated in varying ratios of homologues. Synthetic samples can be designed and prepared on-demand with predetermined chain lengths and free of trace contaminants. 2) Synthesis of all possible diastereomers at the undefined stereocenters might boost the elucidation of the absolute stereochemistry. This will simplify future synthesis and product developments, since chemist will know which isomer to synthesize.

The research described in this thesis is twofold. The first part is aimed at the total synthesis of mycolic acids, along with (preliminary) attempts to elucidate the absolute stereochemistry. The second part aims to provide novel aminoglycoside antibiotics that can potentially be used against bacteria such as M. tuberculosis, in which our in-house developed regioselective oxidation fulfills a prominent role.

Chapter 2, describes the synthesis of all the fragments required for the synthesis of

methoxymycolic acid. In order to elucidate the structure, we varied the absolute stereochemistry at the α-methyl methoxy and cyclopropyl segments, while keeping the α-alkyl β-hydroxy moiety as the R,R isomer. Overall, this resulted in the synthesis of 5 fragments on multi-gram scale. In Chapter 3 these fragments were connected, which resulted in four mycolic acid diastereomers in gram quantities. Furthermore, the obtained mycolic acids were derivatized as the glucose and glycerol esters for biological studies.

(30)

19

In chapter 4 we focused on the optimization of the aerobic regioselective oxidation of carbohydrates. In order to improve the catalyst’s resistance towards oxidative decomposition, a deuteration strategy not applied before in catalysis was exploited. In

chapter 5 the concise synthesis of neomycin B amphiphiles is described. By using our

regioselective oxidation we could introduce the amphiphilic moiety at a position not described before in a straightforward manner. Finally, in Chapter 6 a summary and future perspectives are given.

1.5 References

[1] H. C. King, T. Khera-Butler, P. James, B. B. Oakley, G. Erenso, A. Aseffa, R. Knight, E. M. Wellington, O. Courtenay, PLoS One 2017, 12, 1–15.

[2] L. C. Rodrigues, D. N. J. Lockwood, Lancet Infect. Dis. 2011, 11, 464–470.

[3] D. Zingue, A. Bouam, R. B. D. Tian, M. Drancourt, Clin. Microbiol. Rev. 2018, 31, 1– 52.

[4] World Health Organization, Global Tuberculosis Report, 2018.

[5] V. Gajalakshmi, R. Peto, T. S. Kanaka, P. Jha, Lancet 2003, 362, 507–515. [6] M. P. Golden, H. R. Vikram, Am. Fam. Physician 2005, 72, 1761–1768.

[7] World Health Organization, The Selection and Use of Essential Medicines. Report of the

WHO Expert Committee, 2015 (Including the 19th WHO Model List of Essential Medicines and the 5th WHO Model List of Essential Medicines for Children), 2015.

[8] M. I. Andersson, J. Antimicrob. Chemother. 2003, 51, 1–11.

[9] M.-P. Mingeot-Leclercq, Y. Glupczynski, P. M. Tulkens, Antimicrob. Agents

Chemother. 1999, 43, 727–737.

[10] C. Roger, J. A. Roberts, L. Muller, Clin. Pharmacokinet. 2018, 57, 559–575. [11] S. Deoghare, Indian J Pharmacol 2013, 45, 536–537.

[12] T. W. Yang, H. O. Park, H. N. Jang, J. H. Yang, S. H. Kim, S. H. Moon, J. H. Byun, C. E. Lee, J. W. Kim, D. H. Kang, Medicine 2017, 96,

[13] M. Jackson, Cold Spring Harb. Perspect. Med. 2014, 4, 1–22.

[14] D. Kaur, M. E. Guerin, H. Škovierová, P. J. Brennan, M. Jackson, Adv. Appl. Microbiol.

2009, 69, 23–78.

[15] W. Vollmer, J. V. Höltje, J. Bacteriol. 2004, 186, 5978–5987.

[16] S. O. Meroueh, K. Z. Bencze, D. Hesek, M. Lee, J. F. Fisher, T. L. Stemmler, S. Mobashery, Proc. Natl. Acad. Sci. 2006, 103, 4404–4409.

[17] B. Zuber, M. Chami, C. Houssin, J. Dubochet, G. Griffiths, M. Daffé, J. Bacteriol. 2008,

190, 5672–5680.

[18] G. Ciamak, J Res Med Sci 2018, 23, 633.

[19] A. Queiroz, L. W. Riley, Rev Soc Bras Medi Trop 2017, 50, 9–18. [20] Rajni, N. Rao, L. S. Meena, Biotechnol. Res. Int. 2011, 1–7.

[21] D. E. Minnikin, L. Kremer, L. G. Dover, G. S. Besra, Chem. Biol. 2002, 9, 545–553. [22] N. Kresge, R. Simoni, R. Hill, J. Biol. Chem. 2008, 283, 5–7.

[23] J. Asselineau, E. Lederer, Nature 1950, 166, 782–783.

[24] N. Qureshi, K. Takayama, H. Jordi, H. Schnoes, J. Biol. Chem. 1978, 253, 5411–5417. [25] D. E. Minnikin, N. Polgar, Chem. Commun. 1967, 1172–1174.

[26] H. Marrakchi, M. A. Lanéelle, M. Daffé, Chem. Biol. 2014, 21, 67–85. [27] K. Takayama, L. A. Davidson, Trends Biochem. Sci. 1979, 4, 280–282. [28] K. Takayama, C. Wang, G. S. Besra, Clin. Microbiol. Rev. 2005, 18, 81–101.

[29] K. H. Choi, L. Kremer, G. S. Besra, C. O. Rock, J. Biol. Chem. 2000, 275, 28201–28207. [30] J. N. Scarsdale, G. Kazanina, X. He, K. A. Reynolds, H. T. Wright, J. Biol. Chem. 2001,

276, 20516–20522.

[31] L. Kremer, L. G. Dover, S. Carreere, K. M. Nampoothiri, S. Lesjean, A. K. Brown, P. J. Brennan, D. E. Minnikin, C. Locht, G. S. Besra, Biochem. J. 2002, 364, 423–430.

(31)

20

[32] R. A. Slayden, C. E. Barry, Tuberculosis 2002, 82, 149–160. [33] M. S. Glickman, J. Biol. Chem. 2003, 278, 7844–7849. [34] M. Glickman, J. Cox, W. Jacobs, Mol. Cell 2000, 5, 717–727.

[35] Y. Yuan, R. E. Lee, G. S. Besra, J. T. Belisle, C. E. Barry, J. Biol. Chem. 1995, 45, 27292–27298.

[36] Van ’t Hoff, Bull. Soc. Chim. [2] 1875, 23, 298.

[37] C. Asselineau, G. Tocanne, J.-F. Tocanne, Bull. Soc. Chim. Fr. 1970, 1455–1459. [38] D. B. Moody, M. R. Guy, E. Grant, T. Y. Cheng, M. B. Brenner, G. S. Besra, S. A.

Porcelli, J. Exp. Med. 2000, 192, 965–976.

[39] D. E. Minnikin, N. Polgar, Tetrahedron Lett. 1966, 2643–2647.

[40] J. R. Al Dulayymi, M. S. Baird, E. Roberts, Tetrahedron 2005, 61, 11939–11951. [41] J. R. Al Dulayymi, M. S. Baird, E. Roberts, Chem. Commun. 2003, 228–229.

[42] J. R. Al Dulayymi, M. S. Baird, E. Roberts, M. Deysel, J. Verschoor, Tetrahedron 2007,

63, 2571–2592.

[43] G. Koza, M. Muzael, R. R. Schubert-Rowles, C. Theunissen, J. R. Al Dulayymi, M. S. Baird, Tetrahedron 2013, 69, 6285–6296.

[44] G. Koza, C. Theunissen, J. R. Al Dulayymi, M. S. Baird, Tetrahedron 2009, 65, 10214– 10229.

[45] G. D. Coxon, S. Knobl, R. Evan, M. S. Baird, J. R. Al Dulayymi, G. S. Besra, P. J. Brennan, D. E. Minnikin, Tetrahedron Lett. 1999, 40, 6689–6692.

[46] J. A. Frick, J. B. Klassen, A. Bathe, J. M. Abramson, H. Rappoport, Synthesis 1992, 621. [47] G. Toschi, M. S. Baird, Tetrahedron 2006, 62, 3221–3227.

(32)

Chapter 2:

Synthesis of the Methoxymycolic Acid

Fragments

Part of this chapter will be submitted for publication:

N. Tahiri, P. Fodran, D. Jayaraman, J. Buter, T. A. Ocampo, I. Van Rhijn, D. B. Moody, M. D. Witte, A. J. Minnaard, manuscript in preparation

(33)

22

2.1 Introduction

Although the stereochemistry in the α-alkyl β-hydroxy segment seems established to be

R,R, by optical rotation[1]_{and in vitro testing of natural 2R,3R and racemic glucose}

monomycolate stereoisomers,[2]_{the remaining functionalities, in particular the}

cyclopropyl moiety, seem yet to be defined regarding their absolute stereochemistry. Therefore, four diastereomers of methoxymycolic acid were synthesized with all possible diastereomers in the syn α-methyl methoxy and cis cyclopropyl functionalities. This chapter describes the synthesis of all three fragments required for the convergent total synthesis of this mycolic acid.

In order to allow the efficient synthesis of all four desired methoxymycolic acid diastereomers, we envisioned that a convergent synthesis would be the most optimal approach. Dividing the molecule in three parts of roughly equal stereochemical complexity would allow construction of any diastereomer, by proper combination of the individual fragments A, B and C (Scheme 1). The specific disconnections between fragment A and B, and fragment B and C, were based on the fact that cis-configured cyclopropyl moieties are readily introduced on allylic alcohols via Charettes method, and that the required allylic alcohol can be prepared from readily available building blocks (for detailed description see section 2.3). The choice for the specific disconnections was further based on the availability of commercial starting materials that are required for the introduction of the other stereocenters and correct linker lengths. In order to vary the stereochemistry in the α-methyl methoxy and cyclopropyl moieties, both enantiomers of fragments B and C were synthesized whereas fragment A was synthesized only as the R,R isomer.

Scheme 1. Retrosynthetic analysis of mycolic acid.

Combining the fragments by utilizing a sp3_-sp3 _{cross-coupling such as the Suzuki-Fu}

(34)

23

required. We therefore applied this chemistry for the combination of fragments B and

C. However, disconnecting next to the cyclopropyl moiety excludes the application of a

Suzuki-Fu coupling, since insertion of palladium will result in cyclopropyl cleavage. We therefore turned to a modified Julia-Kocienski olefination, which requires an additional reduction step of the newly formed double bond.

2.2 Fragment A

2.2.1 Retrosynthetic analysis

The most challenging aspect in the synthesis of fragment A is the construction of the α-alkyl, β-hydroxyl acid moiety. Intuitively, construction of both stereocenters seemed to be best achieved by means of an aldol reaction. Although the body of literature on the asymmetric aldol reaction is huge, methods on its asymmetric anti-selective segment is much less represented. From the available options, the methodology developed by Masamune[3] _{seemed to be matching best with our desires (Scheme 2): easy access to}

the auxiliary and high stereoselectivity with linear aliphatic aldehydes. Therefore, fragment A is best represented as intermediate 1 which contains an ephedrine based auxiliary, readily installed by esterification, to allow for the aldol reaction to proceed with high stereocontrol. Moreover, this auxiliary can be maintained throughout the synthesis in order to provide enhanced solubility while also acting as a protection group for the carboxylic acid functionality. Disconnection between the α- and β-carbons of the ester, yields aldehyde 5 (which had to be prepared), and further installment of the α-branch can be achieved by means of chain elongation of the bromoester of 3 and 2 with 1-hexadecene (4) using a Suzuki-Fu cross-coupling. Compounds 3 and 4 are low-cost commercially available materials, whereas 2 is affordable.

(35)

24

2.2.2 Synthesis of the aldehyde spacer

The synthesis of fragment A started with the synthesis of linker 5, which was required for the aldol reaction. Commercially available pentadecanolide (6) was reduced cleanly using DIBAL to its lactol counterpart (Scheme 3). On small scale the workup was eventless, but on a 14 gram scale excessive extractions with DCM were required in order to achieve decent recovery of the product due to its low solubility. Lactol 7 was subjected to a Horner-Wadsworth-Emmons reaction,[4]_{resulting in the α,β-unsaturated}

thioester 8 in 70% overall yield over two steps. In the first instance, we planned a step-by-step conversion of 8 to 5 by means of hydrogenation of the double bond, silyl protection of the alcohol and subsequent DIBAL reduction of the thioester to the aldehyde. However, because hydrogenation over Pd/C was unsuccessful, probably because of traces of thiols resulting in catalyst poisoning, we realized that a Fukuyama reduction might actually achieve all three conversions in one single step. Indeed, treatment of 8 with triethylsilane in acetone in the presence of Pd/C resulted in the hydrogenated TES protected aldehyde 5 in excellent purity after column chromatography. Although three transformations were in this way efficiently achieved in one step, the yield turned out to differ considerably from batch to batch. In this reaction, seven intermediates can be formed varying considerably in polarity, and this might suggest full conversion of the starting material while actually the reaction is not complete. Furthermore, yields were generally lower when the reaction was performed on multi-gram scale compared to sub-gram scale.

Scheme 3. Synthesis of aldehyde 5 required for the aldol reaction.

Since later in the Masamune aldol reaction the TES protection group proved to be too labile, we opted for the installment of a more stable protection (such as a TBS group) at this stage. Therefore, instead of TESH, TBSH was used in the Fukuyama reduction (Scheme 4). Unfortunately, no noticeable conversion of 8 was observed by TLC. Alternatively, the Fukuyama reduction of TBS protected thioester 10 could provide a quick entry to TBS protected aldehyde 5b. TBS protection of 8 under standard conditions went smoothly in quantitative yield, but surprisingly, Fukuyama reduction of

(36)

25

Scheme 4. Attempted synthesis of aldehyde 5b.

In an attempt to obtain TBS protected aldehyde 5b, we revisited the step-by-step approach in which first the thioester 10 was reduced to the alcohol, followed by hydrogenation of the double bond, and subsequent oxidation of the alcohol to the aldehyde (Scheme 5). Reduction of TBS thioester 10 with LiAlH4 went in quantitative

yield, but resulted in an inseparable mixture of saturated alcohol 11 and allylic alcohol

12 in a 2:1 ratio, respectively. Reduction of the double bond with Pd/C under H2

atmosphere was unsuccessful, and reduction using our in-house developed diimide reduction[5]_{resulted in complete hydrogenation, but concomitant cleavage of the TBS}

group led to the corresponding diol. We therefore decided to continue with the Fukuyama reduction as depicted in Scheme 3 and address the problems associated with the TES group at a later stage in the synthesis.

Scheme 5. Attempted synthesis of 11 from 10. 2.2.3 Completion of fragment A

With aldehyde 5 in hand, the synthesis of fragment A continued by the in situ generation of the acyl bromide of 3 using oxalyl bromide, and subsequent esterification with 2 in quantitative yield. This was to prevent scrambling of the primary bromide with chloride, which occurs when preparing the acid chloride using oxalyl chloride. Masamune aldol[3]_{reaction of chiral ester 13 with aldehyde 5 yielded the desired}

β-hydroxy ester in a 9:1 diastereomeric ratio. Unfortunately, upon workup the TES group was partially cleaved, and it was therefore decided to completely remove the TES group by stirring the crude reaction mixture in acidic THF. Although no separation of the diastereomers was observed on TLC, a substantial separation was achieved by applying a slow gradient on column chromatography. By careful analysis of the collected fractions with 1_{H-NMR, and by combining all fractions with a dr better than 95:5, the}

(37)

26

very high dr of 97:3. Although this was a very laborious and time consuming process, scaling up the reaction to 10 g of chiral ester did not cause any problems and resulted in more than 8 g of diol 14. This amount was sufficient to complete the synthesis of all four diastereomers. The remaining steps in the synthesis of fragment A proceeded with relative ease, and consisted of TBS protection of both alcohols in 14, resulting in compound 15 in good yield. Compound 16 was obtained by chain elongation of 15, by applying a sp3_-sp3_Suzuki-Fu[6]_{cross-coupling. Then, the selective deprotection of the}

primary alcohol went in excellent yield, resulting in 17. We decided to perform this deprotection under acidic conditions to prevent any possible loss of the β-silylether via an E1cb elimination. Finally, alcohol 17 was oxidized in high yield to aldehyde 1 by applying a Dess-Martin periodinane oxidation. The crude aldehyde could be purified by multiple acetonitrile/pentane extractions, followed by a simple column purification. This aldehyde was stored under nitrogen at -20 °C for at least two months without any noticeable degradation. Overall, fragment A (1) was synthesized in nine steps (longest linear sequence) from commercially available pentadecanolide (6) in 24% overall yield and 97:3 dr.

Scheme 6. Completion of fragment A.

2.3 Fragment B

2.3.1 Retrosynthetic analysis

Fragment B, the middle segment of the molecule containing the cis cyclopropyl functionality, can be derived from alcohol 18 after a few simple functional group interconversions (Scheme 7). The alcohol 18 can be obtained from allylic alcohol 19 by

(38)

27

means of a Charette asymmetric cyclopropanation.[7,8]_{Since the stereochemistry in 19 is}

transferred to the cyclopropanation product 18, it is of utmost importance to perform the cyclopropanation with an extremely high cis purity in 19, the key intermediate in the synthesis of fragment B. The allylic alcohol is necessary for coordination with the chiral promotor in order to establish efficient face discrimination in the cyclopropanation reaction. Because the synthesis of the cis alkene was challenging, we aimed to synthesize this compound via a Z-selective Horner-Wadsworth-Emmons reaction (Scheme 7, Route 1) of aldehyde 21, or by addition of 22 (Route 2) or 23 (Route 3) to paraformaldehyde.

Scheme 7. Retrosynthetic analysis of fragment B. 2.3.2 Route 1

In order to perform the modified Horner-Wardsworth-Emmons reaction of 21 and phosphonate 27, the starting materials had to be synthesized first. Phosphonate 27 was prepared on a 35 gram scale following a straightforward procedure by Touchard[9]

which started by the reaction of ethyl dichlorophosphite and 2-tert-butylphenol, followed by an Arbusov reaction with ethyl bromoacetate (Scheme 8). Although the procedure reported that the product of the Arbusov reaction could be used without any purification, we noticed that a simple trituration of the crude material in heptane resulted in analytically pure material in excellent yield. Because of the big scale, we decided to obtain aldehyde 21 using a Swern oxidation, and oxidation of 21 gram of alcohol 28 went smoothly. Next, Horner-Wardsworth-Emmons reaction of aldehyde 21 and phosphonate 27 at ˗78 °C, resulted in the α,β-unsaturated ester as a single isomer in 90% yield. In order to prevent any conjugate reduction, we subjected 29 to a DIBAL

University of Groningen Total synthesis of mycolic acids and site-selective functionalization of aminoglycoside antibiotics Tahiri, Nabil