(-)-3.27 in the reaction of aziridine 2.2 with metal enolate of dibenzoyl methane is reported in the following Scheme 3.11.

(-)-3.27 in the reaction of aziridine 2.2 with metal enolate of dibenzoyl methane is reported in the following Scheme 3.11.

In this rationalization, evidence is given to the observation that corresponding syn- and/or anti-1,4-addition products 3.30 and 3.30 and dibenzoylated glycal-derived anti-1,2-addition product (+)-3.26 and 2,5-dihydropyrrole (+)-3.28 (the primary reaction products) are not isolated under the t-BuOK/t-BuOLi protocol.

In conclusion, enantiopure cis-2,5-disubstituted-N-mesyl-2,5-dihydropyrroles (+)-3.9, (+)-3.15, (+)-3.21, (+)-3.24 and (–)-3.29 have been obtained, in a unique and completely stereoselective fashion, by addition of metal enolates of -dicarbonyl compounds (C-nucleophiles) to glycal-derived vinyl N-mesyl aziridines 2.2 and 2.2. A reaction mechanism involving a base-catalyzed rearrangement of the corresponding 1,4- addition products, with ring contraction, is congruent and experimental evidence was obtained. The presence of two highly functionalized side chains and of the endocyclic double bond, make the obtained 2,5-disubstituted-2,5-dihydropyrroles particularly useful for further elaboration on the route to alkaloids and furanosidic azasugars.

3.3. Preliminary study on the use of other N-protective-activating group than mesyl group in aziridine 2.2

The results obtained in the rearrangement process occurring in the ring opening additions of appropriate C-nucleophiles to aziridines 2.2 and 2.2 are certainly valuable considering the synthetic interest toward simple molecules, as the obtained 2,5- dihydropyrroles are.

However, its value could be considerably higher if the N-protecting/activating group, present in the final 2,5-dihydropyrroles, could be easily removed in order to have corresponding 2,5-dihydropyrroles bearing a free –NH- group. Actually, a free-NH-group makes possible further elaborations, as N-alkylation and/or N-acylation, in order to have product of biological interest, as generically shown in Scheme 3.12.

In this sense the N-mesyl group, which, on one side, due to its particular stability

and activating properties, often offers the possibility of the realization of reactions

otherwise not possible, on the other side is not actually the best choice to have a free amino group by deprotection procedures. As a consequence, it appeared necessary to introduce on the starting aziridines 2.2 and 2.2 a different N-activating group which could easily be removed after the rearrangement process to 2,5-dihydropyrroles had taken place.

Considering that the simple N-acetyl group could not be used because the corresponding N-acetyl aziridine had proved to be not sufficiently reactive (see Chapter 2), our natural choice fell on aziridines 2.3 and 2.3 which bear the N-(o- nitrobenzensulphonyl) [N-(nosyl)] protecting/activating group, which had previously studied and advantageously used in our laboratory in the regio- and stereoselective glycosylation of alcohols (see Chapter 2).

Actually the N-nosyl group present in the corresponding N-nosylamino pseudoglycals was found to be easily removed by the simple PhSH/K

CO

protocol, in accordance with a S

Ar reaction mechanism (Scheme 3.13), to give corresponding free – NH

-containing products, as desired in good yield and in the absence of any side, undesiderable reactions (see Scheme 2.26, Chapter 2).

Scheme 3.12. Glycal-derived - and -aziridines bearing a removable N-protecting activating group.

BnO O

BnO O N

N PG

PG

N PG

CH(COY)₂ BnO

OH

N PG

CH(COY)2 BnO

OH

N H

CH(COY)2 BnO

OH

N H

CH(COY)₂ BnO

OH deprotection

deprotection

products from N-alkylation, acylation, etc.

products from N-alkylation, acylation, etc.

aziridine

(PG other than -Ms)

aziridine 

(PG other than -Ms)

2,5-disubstituted, 2,5-dihydropyrroles (Y= OMe. Ph)

Figure 3.1. N-nosyl aziridines 2.3a and 2.3b.

2.3 2.3

BnO O O

BnO

N N

Ns Ns

Ns =

NO₂ SO₂

However, even if the choice for the type of the new N-protecting group had been, practically, already made in favor of the N-nosyl group, we took this occasion also for examining, briefly, and only in the glycosylation of alcohols, the behavior of other two possible N-protecting, easily removable, group, of aziridine 2.2a, taken for convenience as the model, as N-benzyloxycarbonyl (N-Cbz) and N-trifluoroacetamide (N-TFA) groups.

3.3.1. Vinyl N-Cbz and N-TFA aziridines 3.31a and 3.33a in methanolysis reactions

The new two aziridines, the vinyl N-Cbz 3.31a and N-TFA aziridine 3.33a, as the corresponding N-mesyl and N-nosyl ones, were supposed to be not stable and could be obtained only in situ and only if the corresponding stable precursor, the trans O-mesyl-N- Cbz amino 3.30a and trans O-mesyl-N-TFA amino derivative 3.32a, respectively, were able to cyclization under alkaline conditions (usually MeONa or t-BuOK) (Scheme 3.14).

The synthesis of trans O-mesyl-N-Cbz amino- 3.30a and trans O-mesyl-N-TFA amino derivative 3.32a starts from trans amino alcohol 2.45a. In the case of 3.30a, trans amino alcohol 2.45a is initially reacted with Cbz-Cl to give trans N-Cbz-amino alcohol

Scheme 3.13. SNAr mechanism of deprotection of –NHNs protecting group.

Scheme 3.14. N-Cbz and N-TFA aziridines 3.31 and 3.33 and corresponding stable precursors 3.30

and 3.32.

3.31 3.33

BnO O O

BnO

N N

BnO₂C F₃COC BnO O

NHCO₂Bn MsO

t-BuOK

BnO O

NHCOCF₃ MsO

t-BuOK

3.30 3.32

NO2

SO₂ NHR PhSH K2CO3

SPh N

SO2NHR - SPh

-O O

+ N

SO2 NHR SPh

-O O^- +

RNH^-+ SO² +

NO2

SPh RNH2

MeCN

(free amino derivative)

3.34a which was O-mesylated by MsCl/Py protocol to give the desired trans O-mesyl-N- Cbz amino derivative 3.30a (Scheme 3.15).

In the case of 3.32a, the trans amino alcohol 2.45a was left to react with ethyl triflouroacetate

to give the corresponding trans N-TFA alcohol 3.35a, then mesylated to the desired trans O-mesyl-N-TFA amino derivative 3.32a (Scheme 3.16).

MeOH was chosen as the simple alcohol for checking glycosylation by these new aziridines and in order to evaluate their reactivity under these conditions and eventually the corresponding regio- and stereoselectivity. Protocol A and protocol B reaction conditions were tried.

The reaction of trans N-Cbz amino-O-mesylate 3.30a in MeOH, as the solvent (protocol A) in the presence of t-BuOK leads to the formation of an 84:16 mixture of the corresponding methyl - 3.36 (syn-1,4-addition product) and -O-glycoside 3.36 (anti- 1,4-addition product) after slight long time (5 days) and repeated addition of t-BuOK (6 equiv) (Scheme 3.17).

Scheme 3.15. Synthesis of trans O-mesyl-N-Cbz amino derivative 3.30a.

3.34

BnO O BnO O

NH2

HO

Cbz-Cl

2.45

NHCbz HO

3.30

BnO O MsCl

NHCbz Py MsO

Scheme 3.16. Synthesis of trans O-mesyl-N-TFA amino derivative 3.32a.

3.35

BnO O BnO O

NH₂ HO

CF3CO2Et

2.45

NHCOCF₃ HO

3.32

BnO O MsCl

NHCOCF₃ Py MsO

This result demonstrates that the corresponding aziridine 3.31a was formed and that the usual preferred 1,4-addition occurred, in a completely regioselective fashion. As for the stereoselectivity, a consistent preference (86%) for syn-1,4-addition product with the formation of that anomer (, the coordination product) having the same configuration of the starting aziridine () is still present also in this case, as in the previously examined N-mesyl aziridine 2.2a.

Unexpectedly and advantageously, under protocol B reaction conditions, the reaction of trans N-Cbz-O-mesylate 3.30a in MeCN with MeOH (3 equiv) in the presence of t-BuOK (4 equiv) turned out to be decidedly faster and completed in 1 h with the practically exclusive formation of the a-anomer, the coordination product, with the corresponding -anomer 3.36, the non-coordination product, if present, in an amount less than 1%. In other words, the reaction is still completely regioselective and shows a syn- stereoselectivity decidedly superior (>99%) to that found under protocol A reaction conditions (86%). All this means that the N-Cbz functionalization of aziridine 3.31a can be useful for reactions carried out in the presence of a reduced amount of nucleophile and, potentially, useful in the rearrangement process we have realized with aziridine 2.2a.

The completely syn-stereoselective result obtained with N-Cbz aziridine 3.31a is in accordance with the typical behavior of these heterocyclic systems under protocol B reaction conditions which favor the occurrence of an aziridine nitrogen-nucleophile coordination as shown in structure 3.37 of the following Scheme 3.18. In this way, the

Scheme 3.17. Methanolysis of N-Cbz aziridine 3.31a under protocol A and protocol B reaction conditions.

syn-1,4-addition pr oduct

O OMe BnO

CbzHN

3.36

BnO O O OMe

BnO CbzHN N

+

3.31

BnO O

NHCbz MsO

3.30 Cbz

MeOH

P rotocol A: MeOH (solvent) t-BuOK (6 equiv) 86% 14%

P rotocol B: MeOH (3 equiv) t-BuOK (4 equiv)/MeCN >99% <1%

anti-1,4-addition product

t-BuOK

3.36

nucleophile is appropriately disposed for an entropically favored attack on vinyl carbon C(1) from the same side as aziridine nitrogen with completely selective formation of the 1,4-syn adduct as observed (Scheme 3.18).

The behavior, under basic conditions, of trans N-TFA-O-mesylate 3.32a, the possible precursor of aziridine 3.33a, is decidedly different from the other precursors of glycal-derived epoxides and aziridines ever examined in our laboratory. Actually, the reaction of 3.32a in MeCN in the presence of MeOH (3 equiv) and t-BuOK (2 equiv) (protocol B) led in 1 h to only one product which turned out not to be an addition product, because not containing the -OMe group (

H NMR). An appropriate and accurate NMR study carried out in a collaboration with Prof.ssa Gloria Uccello Barretta and Dott.ssa Federica Balzano of the Dipartimento di Chimica e Chimica Industriale of our University indicated for this compound a structure corresponding to cis hydroxy trifluoroacetylamino glycal derivative 3.41, a diastereoisomer of trans hydroxy trifluoroacetylamino glycal derivative 3.35a (Figure 3.2), that is the precursor of trans N-TFA-O-mesylate 3.32a (Scheme 3.16). The same result was obtained also when the same reaction was carried out in MeOH, as the solvent and nucleophile (protocol A).

The result obtained clearly indicated that a) no cyclyzation of trans N-TFA-O- mesylate 3.32a to corresponding N-TFA aziridine 3.33a, as desired, had occurred, b) a

Scheme 3.18. Rationalization of the stereoselective result observed in the glycosylation of MeOH by N-Cbz aziridine 3.31.

O OMe

BnO CbzHN

3.36

H O Me

3.37 O

N

OBn

O N

BnO

a

route a

coor dination pr oduct

syn-1,4-addition pr oduct

BnO O O OMe

BnO CbzHN t-BuOK N

(4 equiv)

3.31 3.36

(less than 1%)

r oute a = pseudoaxial attack BnO O

NHCbz MsO

3.30

MeOH (3 equiv)

Cbz

Cbz

Cbz

S

2 type substitution of –OMs group at C(4) had occurred and c) the configuration at C(3) and the structural integrity of the linked N-trifluoroacetyl group had been maintained.

A tentative rationalization of the unusual behavior observed with trans N-TFA- O-mesylate 3.32a is shown in the following Scheme 3.19.

This rationalization starts from the behavior of t-BuOK which, due to the enhanced reactivity of the carbonyl group of the N-trifluorocetyl substituent, behaves as O-nucleophile instead as a base, as usual. The nucleophilic addition of t-BuOK to the electrophilic center generates an sp

alcoholate species 3.38 (addition step of a common nucleophilic addition-elimination substitution at a carboxylic acid derivative) which, in an entropically-favored intramolecular fashion, attacks the adjacent C(4) carbon, bearing the good leaving group –OMs, with inversion of configuration (intramolecular S

2 process)

Scheme 3.19 Product from the reaction of trans N-TFA-O-mesylate 3.32 in the presence of MeOH under protocol A and B reaction conditions.

Figure 3.2. The cis-3.41, a diastereoisomer of trans-glycal derivative 3.35a.

BnO O

NHCOCF₃ HO

BnO O

NHCOCF₃ HO

3.35 3.41

BnO O

NH MsO

t-BuOK

CF₃ O

BnO O

NH MsO

F₃C O

t-BuO^-K⁺

O-t-Bu -

BnO O

NH F₃C

O

O - MsO

t-Bu MeOH

BnO O

NH HO

C CF₃ O+ C(CH₃)₃ BnO O

NHCOCF₃ HO

+ CH₂

CH3

H3C and/or

MeOH

3.32 3.38 3.39

3.40 (CH₃)₃COMe 3.41

MeOH -H⁺

to give the cyclic sp

intermediate 3.39. Subsequent elimination step, as shown in 3.39, leads to the insertion of an –OH group (after protonation of the formed alcoholate by MeOH, present in solution) at C(4) with inversion of configuration and re-formation of the trifluoroacetylamino group “protonated” at the carbonyl oxygen by t-butyl group, as shown in 3.40. Subsequent, spontaneous elimination from 3.40 of the sufficiently stable t- butyl cation, with supposed, even if reasonable, formation of isobutylene and/or methyl t- butyl ether (not checked), leads to the cis hydroxy trifluoroacetyl amino derivative 3.41, as the only found reaction product.

Clearly, this unexpected behavior inhibits the use of trans N- trifluoroacetylamino-O-mesylate 3.32a as a precursor of the desired corresponding aziridine 3.33a, because the evidently fast nucleophilic addition-elimination substitution at the trifluoroacetylamino group and tandem intramolecular substitution of the –OMs group at C(4) inhibits the typical behavior as a base of t-BuOK, necessary for the formation of the intermediate aziridine.

3.3.2. N-Nosyl aziridine 2.3a in reactions with metal enolates derived from dibenzoylmethane

The behavior of protecting-activating N-nosyl group of the new aziridine 2.3a was examined under the same reaction conditions which turned out to be appropriate for the occurence of the rearragement process found with aziridine 2.2a and leading to 2,5- dihydropyrroles. As for the necessary C-nucleophile, the metal enolate derived from dibenzoyl methane was exclusively used because considered as an appropriate model for this tranformation, at least in this first approach to the use of an N-activating group of the aziridine, susceptible of a subsequent removal.

The reaction of aziridine 2.3a with the K-enolate of dibenzoyl methane (t-BuOK,

as the only base used) in anhydrous toluene afforded a crude reaction mixture consisting

of the corresponding anti-1,2-addition product, the trans dibenzoylmethyl-N-mesylamino

derived glycal (-)-3.42 (28%), the 1,4-addition product as a mixture of - and -adducts

3.43 and 3.43 (32%) and the cis-2,5-disubstituted-2,5-dihydropyrrole (+)-3.44 (40%)

which were sepatated by TLC and appropriately characterized. The mixture of - and - anomers 3.43 and 3.43 was not furtherly examined (Scheme 3.20).

Because the reaction carried out in anhydrous toluene had shown the formation of a reaction mixture as a consistent suspension, we thought necessary to change the solvent by using a solvent able to favor the solubility of all the reacting species. As a consequence, the same reaction was repeated under the same conditions (t-BuOK as the only base) by using anhydrous THF as the solvent. Under these conditions, a similar result was obtained (presence of corresponding anti-1,2-addition product, 1,4-addition product and 2,5-dihydropyrrole) with the difference that all the obtained products, anti 1,2-addition product 3.45 (20%), 1,4-addition product 3.46b (45%) and 2,5- dihydropyrrole 3.47 (35%) turned out to be mono-debenzoylated and the 1,4-addition product portion turned out to correspond to the only anomer 3.46 and not to a mixture of both anomers. In other words, changing the solvent from toluene to THF had not effect on the regioselectivity of the reaction, but, the more solubilizing properties of THF decidedly had favored the occurence of the haloform-like reaction on all the obtained product.

However, the particular behavior of these reaction conditions has some undesired consequences on the obtainment of the 2,5-dihydropyrrole system. Actually, as we know

Scheme 3.21. Reaction of N-nosyl aziridine 2.3 with K-enolate of dibenzoylmethane in dry THF.

Scheme 3.20. Reaction of N-nosyl aziridine 2.3 with K-enolate of dibenzoylmethane in dry toluene.

BnO O O

BnO

(–)-3.42 (28%) 3.43 + 3.43 (32%)

Ph Ph

O O

t-BuOK toluene

N Ns

CH(COPh)₂ NsNH

CH(COPh)₂

CH(COPh)₂

NsNH

BnO OH 2.3 +

(+)-3.44 (40%) +

BnO O O

BnO

(–)-3.45 (20%) 3.46 (45%)

Ph Ph

O O t-BuOK

N Ns

CH₂COPh NsNH

CH₂COPh

CH₂COPh

NsNH

BnO OH +

2.3

(+)-3.47 (35%) +

THF

that the 2,5-dihydropyrrole system formed in this reaction derives from the dibenzoylmethyl substituted 1,4-addition product, if this product loses one benzoyl group, the rearrangement process as described in Scheme 3.22, cannot occur (for this transformation it is necessary the presence of a sufficiently acidic C-H bond which only a double -CO- substitution can guarantee, see Schemes 3.7 and 3.11). Accordingly, with THF as the solvent, the amount of 1,4-addition product (45%) is consistently superior to that found in the corresponding reaction carried out in toluene (32%).

All this means that in a prosecution of the study of this reaction, new conditions should be found (solvent, temperature, reaction time) in order to avoid the occurrence of the debenzoylation process and, as a consequence, to increase the yield of the 2,5- dihydropyrrole of our interest. The observation that only the b- anomer 3.46b is recovered in the final reaction mixture could only indicate that, probably, the corresponding a- anomer is faster in its transformation into pyrrole 3.47 and/or that the b-anomer is faster in the haloform-like debenzoylation process: both possibilities can contribute to the obtainment of the present result (Scheme 3.22).

Actually the amounts of dihydropyrrole obtained in the reactions carried out using only t-BuOK as the necessary base (Schemes 3.20 and 3.21) were still

Scheme 3.22. Different behavior under basic conditions of 1,4-addition product 3.43 from the reaction of N-nosyl aziridine 2.3a with K-enolate of dibenzoyl methane.

HC

3.43()

t-BuOK

HO C

O-t-Bu - Ph

3.46

O Ph

Ph O

t-BuO-

(as a nucleophile)

O Ph

t-BuOK

stable no rearrangement to 2,5-dihydropyrrole can occour BnO O

NsNH

BnO O

NsNH

CH₂COPh BnO O

NsNH

t-BuOK

r oute b b

t-BuO- (as a base)

r oute a (Schemes 3.7 and 3.11)

N Ns

CH₂COPh BnO

OH

(+)-3.47 (35%) a

not sufficiently acid

unsatisfactory. For these reason, the conditions based on the use of THF as the solvent and the double base-protocol (t-BuOK in association with t-BuOLi) which exceptional results had given with the N-mesyl aziridine 2.2a, were tried. Unfortunately and also unexpectedly, a very complex reaction mixture was obtained under these conditions to the point that no other attempts were carried out. Probably the N-nosyl group is not so stable to alkaline conditions and in particular to the presence of C- (enolates) and O- nucleophiles (alcoholates) as N-mesyl group is, and undesired side reactions can occur.

Actually, it is possible to find that reactions, which proceed very well with N-mesyl group in the molecule, don’t show the same qualitative and quantitative result when the N-nosyl protection is alternatively used.

3.3.3. Deprotection of N-nosyl anti-1,2-glycal derivatives 3.42, 3.45 and 2,5-dihydropyrrole 3.47

The dibenzoylmethyl anti-1,2-glycal derivative 3.42 was initially used in order to check the possibility of deprotecting the N-nosyl functionality of these carbonyl products with the obtainment of the corresponding glycal derivative bearing a free –NH

moiety on C(4). The known PhSH/K

CO

/MeCN protocol was used as previously done in 4-N- nosylamino substituted pseudoglycal chemistry (see Chapter 2). The reaction, carried out at room temperature, was followed by TLC which clearly indicated that the starting material was rapidly consumed with the formation of only one product which was isolated. Unexpectedly, accurate

H and

C NMR analysis indicated for this product a structure largely different from that reasonably expected. Actually, the deprotection of the –NHNs group had occurred, but with the contemporary mono-debenzoylation of the dibenzoylmethyl group and formation of 4-N-benzoyl-3-benzoylmethyl glycal derivative 3.48 (Scheme 3.23).

The formation of this compounds is reasonably subsequent to the deprotection of

the –NHNs group by PhSH/K

CO

protocol and due to the 1,2-relationship between the so

formed –NH

group and one of the close –COPh groups. An appropriate rationalization of

the formation of 3.48 is given in the Scheme 3.23.

The transformation of the –NHNs group into an amide group as –NHCOPh constituted certainly an improvement of the quality of the obtained product, but at the same time it required a second step (the hydrolysis of the amide functionality) in order to have a free–NH

which could require conditions deleterious for other functionalities present in the molecule. We have not tried, at least for the moment, the hydrolysis of the amide 3.48, because we have preferred to carry out the PhSH/K

CO

protocol on N-nosyl derivative 3.45, the monodebenzoylated analogue of 3.42. Actually, we thought that the occurrence of the mechanism described in Scheme 3.23 was due to the presence of two benzoyl groups on the adjacent position and it could be avoided if only one group of this type was present. The result confirmed our supposition and the treatment of monobenzoyl derivative 3.45 by the usual PhSH/K

CO

protocol yielded the corresponding 4-amino derivative 3.49 by a clean reaction and in a satisfactory yield. Subsequently, the free amino derivative 3.49 was acetylated by Ac

O/Py protocol to give N-acetyl derivative 3.49-Ac for further, complete characterization of these compounds (Scheme 3.24).

Scheme 3.23. Rationalization of the formation of N-benzoyl derivative 3.48.

O

NsNH

CH(COPh)₂

PhSH K₂CO₃

O

3.48

H₂N CH BnO

BnO

O Ph

O Ph

O

HN CH BnO

O Ph OH

Ph

O

PhCOHN HC BnO

OH Ph O

PhCOHN

CH₂COPh BnO

(–)-3.42

Scheme 3.24. Deprotection of N-nosyl group and formation of 4-amino glycal derivative 3.49.

O

NsNH

CH₂COPh PhSH K₂CO₃

BnO O

H₂N

CH₂COPh

BnO Ac₂O O

AcHN

CH₂COPh BnO

Py

(-)-3.45 3.49 3.49-Ac

On the basis of this preliminary results obtained with a glycal-derived system, we have transferred the PhSH/K

CO

protocol on the 2,5-dihyropyrrole system bearing only one benzoyl group as the 2-benzoylmethyl derivative 3.47. In this way we wanted to avoid useless complications due to the presence of two benzoyl groups as previously found. The treatment of a solution in MeCN of 2,5-dihydropyrrole 3.47 with PhSH (2 equiv) and K

CO

(2 equiv) showed (TLC) the disappearance of the starting material in 4 h with the contemporary formation of a new product characterized by a low R

value.

Usual work-up afforded 2,5-dihydropyrrole 3.50, with the desired free secondary nitrogen group, through a decidedly clean reaction and in a satisfactory yield. (Scheme 3.25).

In conclusion, this second and final part of my experimental thesis has demonstrated that the N-nosyl protecting group of the intermediate aziridine 2.3a is a sufficiently activating group for the rearrangement process to 2,5-dihydropyrrole to occur and, at the same time, a protecting group easily to introduce and simple to remove in order to have corresponding N-deprotected reaction products and, in particular, N-deprotected-2,5-dihydropyrroles of our interest.

On the basis of all these results, future studies on this subject will be certainly focused on finding reaction conditions leading to more satisfactory yields of N-nosyl-2,5- dihydropyrroles, through the described rearrangement process, which will be deprotected to corresponding 2,5-dihydropyrroles, as free bases. These highly substituted 2,5- dihydropyrroles will be subjected to alkylation or acylation procedure of the free secondary –NH- functionality in order to have N-substituted furanosidic azasugars of possible pharmacological interest. The appropriate transformation of the several functional groups present in the synthesized 2,5-dihydropyrroles (carbonyl and hydroxyl groups and double bond) will be certainly taken into consideration.

Scheme 3.25. Deprotection of N-nosyl–2,5-dihydropyrrole 3.47 and formation of 2,5-dihydropyrrole 3.50, as a free base.

PhSH K₂CO₃ N

Ns

CH2COPh BnO

OH N

H

CH2COPh BnO

OH

(+)-3.47 (-)-3.50