Access to the aeruginosin serine protease inhibitors through the nucleophilic opening of an oxabicyclo[2.2.1]heptane: Total synthesis of microcin SF608

Article abstract of DOI:10.1002/chem.201400046

Serine proteases play key roles in many biological processes and are associated with several human diseases such as thrombosis or cancer. During the search for selective inhibitors of serine proteases, a family of linear peptides named the aeruginosins was discovered in marine cyanobacteria. We herein report an entry route into the synthetically challenging core fragment of these natural products. Starting from the common oxabicyclic building block 11, we accessed the octahydroindole core of the aeruginosins, exemplified by the total synthesis of microcin SF608 (2). Key to the synthetic strategy is a highly efficient nucleophilic opening of an oxabicyclo[2.2.1]heptane producing the hydroindole motif of microcin SF608. Moreover, during the synthetic efforts we have observed an unusual regioselective epoxide reduction. Detailed experimental studies of this reaction led us to propose a mechanistic rationale involving intramolecular hydrogen atom delivery by a carbamate NH group to control the regioselectivity of the homolytic epoxide cleavage. Expect the unexpected! An entry route to the aeruginosin protease inhibitors is reported and showcased on the total synthesis of microcin SF608 (see scheme). Detailed experimental studies of an unusual regioselective epoxide reduction observed during this synthesis suggests a mechanistic rationale for this transformation involving intramolecular hydrogen atom delivery by a carbamate NH to direct the regioselectivity of the homolytic epoxide cleavage.

Full text of DOI:10.1002/chem.201400046

DOI: 10.1002/chem.201400046
Full Paper
&
Natural Products
Access to the Aeruginosin Serine Protease Inhibitors through the
Nucleophilic Opening of an Oxabicyclo[2.2.1]heptane: Total
Synthesis of Microcin SF608
Stefan Diethelm, Corinna S. Schindler, and Erick M. Carreira*[a]
Abstract: Serine proteases play key roles in many biological
processes and are associated with several human diseases
such as thrombosis or cancer. During the search for selective
inhibitors of serine proteases, a family of linear peptides
named the aeruginosins was discovered in marine cyanobac-
teria. We herein report an entry route into the synthetically
challenging core fragment of these natural products. Start-
ing from the common oxabicyclic building block 11, we ac-
cessed the octahydroindole core of the aeruginosins, exem-
plified by the total synthesis of microcin SF608 (2). Key to
the synthetic strategy is a highly efficient nucleophilic open-
ing of an oxabicyclo[2.2.1]heptane producing the hydroin-
dole motif of microcin SF608. Moreover, during the synthetic
efforts we have observed an unusual regioselective epoxide
reduction. Detailed experimental studies of this reaction led
us to propose a mechanistic rationale involving intramolecu-
lar hydrogen atom delivery by a carbamate NH group to
control the regioselectivity of the homolytic epoxide cleav-
age.
hibitors was discovered, represented by banyaside B (5).[7] The
key characteristic of these novel alkaloids is a densely packed
glycosylated azabicyclononane (Abn) core 6 substituting for
Introduction
Proteolytic reactions are some of the most important enzyme-
catalysed transformations in biological systems. Proteases are
involved in a diverse set of physiological processes such as
regulation of enzyme activity, signalling events or processing
of biomolecules.[1] Moreover, these enzymes are associated
with a number of human diseases such as cancer, thrombosis
or neurological disorders.[2] In particular, the serine proteases
have been the subject of extensive studies over the past dec-
ades.[3] The quest for selective and potent inhibitors of this
class of peptidases is of key importance for the control of their
associated biochemical processes and the study and treatment
of human diseases. In a screening program for serine protease
inhibitors initiated in the early 1990s, Murakami and co-work-
ers isolated from the marine cyanobacterial strain Microcystis
aeuriginosa the linear peptide aeurginosin 298A (1) (Figure 1).[4]
Within a few years, an impressive array of closely related natu-
ral products, such as microcin SF608 (2) or dysinosin A (3),
were isolated from marine sources.[5] Structure elucidation of
the aeruginosins revealed a common core fragment compris-
ing of a carboxy hydroxyl octahydroindole (Choi) ring 4. More-
over, a terminal guanidine functionality represents a key fea-
ture of these oligopeptides.[6] Soon after the isolation of the
first aeruginosins, a new subclass of this family of protease in-
[a] S. Diethelm, C. S. Schindler, Prof. Dr. E. M. Carreira
Laboratorium fꢀr Organische Chemie
Eidgençssische Technische Hochschule Zꢀrich
Vladimir-Prelog-Weg 3, 8093 Zꢀrich (Switzerland)
E-mail: erickm.carreira@org.chem.ethz.ch
Figure 1. Aeruginosin family of serine protease inhibitors (top) and diver-
gent access to the Abn and Choi core of the aeruginosins starting from allyl-
ic alcohol 10 (bottom).
Supporting information for this article is available on the WWW under
http://dx.doi.org/10.1002/chem.201400046.
Chem. Eur. J. 2014, 20, 1 – 11
1
ꢀ 2014 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
&
&
These are not the final page numbers! ÞÞ

Full Paper
the hydroindole moiety of the previously reported aerugino-
sins. Extensive biological investigation by in vitro assays of this
natural product family revealed that the aeruginosins are selec-
tive inhibitors of important peptidase classes.[5a] For example,
aeruginosin 298A (1) inhibits thrombin, a key enzyme in the
human blood coagulation cascade, with an IC50 value of
a direct opening of the hydrofuran by attack of the nitrogen
nucleophile. The product of this transformation shares its un-
derlying carboxy hydroindole ring system with the Choi core
of many of the aeruginosin peptides. This unexpected finding
opened the possibility of accessing the core element of both
aeruginosin subclasses from one common oxabicyclic precur-
sor 10 (Figure 1, bottom). To implement such a divergent strat-
egy to the synthesis of the whole family of the aeruginosin
serine protease inhibitors, we opted for the total synthesis of
microcin SF608 (2) employing the nucleophilic opening of an
oxabicyclo[2.2.1]heptane.
0.3 mgmLÀ1 [4]
whereas microcin SF608 (2) was reported to be
,
a selective inhibitor of trypsin (IC50 0.5 mgmLÀ1).[5b]
The promising biological activity, as well as the unprece-
dented core structure of the aeruginosin peptides, has attract-
ed the interest of many synthetic chemists. A number of re-
search groups have reported innovative strategies to access
the carboxy hydroindole core found in most aeruginosins.[8,5a]
Our group has recently reported the synthesis of the Abn core
of the banyaside subclass[9] and the total synthesis of nominal
banyaside B along with structural reassignment of this natural
product.[10] Key to the synthetic strategy is the stepwise SN2’
opening of oxabicyclo[2.2.1]heptane 7 to produce tricyclic scaf-
fold 8 (Scheme 1, top, path a). In the course of these synthetic
studies, we discovered an alternative reaction path that
oxabicyclo[2.2.1]heptanes, such as 7, can follow (Scheme 1,
top, path b).[11] When 7 was treated with Lewis acids, such as
TMSOTf, secondary alcohol 9 was produced resulting from
As depicted in Scheme 1, bottom, the synthetic strategy we
pursued towards microcin SF608 leads back to allylic alcohol
11, the enantiomer (10) of which has been previously em-
ployed for the synthesis of the banyaside Abn core and thus
serves as common precursor to both of the aeruginosin central
amino acids.[9] Diol 12 was envisioned as a late-stage inter-
mediate, whereas the C(5) alcohol stemming from the hydro-
furan opening has to be removed to arrive at an intermediate
incorporating the substitution pattern of the natural product.
Hydroindole 12 could be accessed by the nucleophilic opening
of oxabicyclo[2.2.1]heptene 13. Moreover, selective introduc-
tion of the C(6) hydroxyl group would be required either
before or after the key cyclization step. Finally, 13 might be
synthesized from oxabicyclic building block 11 by amination in
the a-position to the ester and removal of the C(3a) hydroxyl.
Herein we present a full account of the implementation of
this strategy towards the total synthesis of microcin SF608.[12,13]
In the course of the synthetic studies conducted toward this
end, we have gained detailed insight into the nucleophilic
opening of oxabicycles employed as a key step in the synthe-
sis. In addition, we have discovered a highly regioselective ep-
oxide reduction. Based on experimental studies we propose
a detailed mechanistic rationale for this transformation involv-
ing intramolecular hydrogen atom donation by a neighbouring
NH group. Accordingly, we have established oxabicyclic build-
ing block 11 as a common precursor to the entire family of the
aeruginosin serine protease inhibitors.
Results and Discussion
Prospecting studies
According to the proposed synthetic plan, we initially opted
for the construction of a suitable cyclization precursor to evalu-
ate the feasibility of the strategy in a model system as outlined
in Scheme 2. To this end, introduction of an amine at C(2) of
building block 11 was required. To facilitate the diastereoselec-
tive installation of the amine, we envisioned lactone 14 as
a suitable substrate. The rigidity of the tricyclic scaffold 14
could help to differentiate the two diastereotopic faces of the
corresponding ester enolate. Accordingly, lactone 14 was syn-
thesized by conjugate reduction of a,b-unsaturated ester 11
mediated by an N-heterocyclic carbene (NHC) copper catalyst
derived from 15[14] as recently described by Buchwald.[15]
Proper adjustment of reaction conditions allowed for concomi-
tant cyclization of the intermediate tertiary alcohol to the cor-
Scheme 1. Differential opening of oxabicycle 7 (top) and retrosynthetic strat-
egy for microcin SF608 (2) (bottom).
&
&
Chem. Eur. J. 2014, 20, 1 – 11
www.chemeurj.org
2
ꢀ 2014 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
ÝÝ These are not the final page numbers!

Full Paper
Scheme 2. Prospecting studies: a) 15 (3 mol%), CuCl2·2H2O (3 mol%), KOtBu (0.3 equiv), PMHS (3.0 equiv), tBuOH (0.5 equiv), THF, 08C, 10 min, 62%; b)
LiHMDS (1.5 equiv), À788C, 30 min; then TrisN3 (1.3 equiv), À458C, 2 min; then AcOH/KOAc (4.6 equiv), CH2Cl2, 73% (81% brsm); c) BnNH2 (3.0 equiv), HCl
(0.2 equiv), CH2Cl2, 258C, 2 h, 79%; d) PMe3 (2.0 equiv), MeCN/THF (5:1), 258C, 30 min; then H2O, 73%; e) Bz2O (1.1 equiv), CH2Cl2, 258C, 1 h, 76%; f) NaBH4
(2.0 equiv), MeOH, 08C, 30 min, 67%; g) TBDPSCl (1.5 equiv), NEt3 (1.5 equiv), DMAP (cat.), CH2Cl2, 258C, 1 h, 94%; h) PMe3 (2.0 equiv), MeCN/THF (5:1), 258C,
30 min; then H2O, 71%; i) Bz2O (1.1 equiv), DMAP (cat.), CH2Cl2, 258C, 1 h, 98%. PHMS=Polymethylhydrosiloxane; LiHMDS=lithium hexamethyldisilazane;
TrisN3 =2,4,6-triisopropylbenzenesulfonyl azide; Bz=benzoyl; TBDPS=tert-butyldiphenylsilyl; DMAP=N,N-dimethyl-4-aminopyridine.
responding lactone 14, isolated in 62% yield. Interestingly, the
Table 1. Azidation of lactone 14.
endocyclic double bond in 14 also proved reactive under the
conditions originally reported by Buchwald. Only upon careful
optimization of the reaction parameters (base, additive, tem-
perature, reaction time), complete selectivity for 1,4-reduction
could be achieved (see the Supporting Information for further
details). We speculate that the high reactivity of the endocyclic
Entry[a]
Azidation reagent
d.r. (17a/17b)[b]
Yield [%]
double bond likely follows from strain induced by the oxabicy-
clic scaffold.
1[c]
1.2:1
81
With 14 in hand, we explored its functionalization in the a-
position by enolate formation and reaction with various elec-
trophilic amination reagents. However, the poor solubility of
14 in ethereal solvents, such as Et2O or THF, precluded efficient
enolate formation with a variety of bases tested (lithium diiso-
propylamide (LDA), KHMDS, NaHMDS, LiHMDS, LiNH2, KH).
Gratifyingly, slow addition of a dilute solution of 14 in CH2Cl2
to a solution of LiHMDS at À788C effectively produced the
lithium enolate. Introduction of an azido group in lactone 14
was examined based on a protocol reported by Evans.[16] The
lithium enolate derived from 14 was therefore treated with tri-
isopropylbenzenesulfonyl azide (TrisN3) 16 at À788C followed
by quenching of the reaction with AcOH. No product forma-
tion was observed under these conditions, but instead starting
material was recovered unchanged. However, when after eno-
late formation at À788C the reaction temperature was in-
creased to À458C, the desired azidolactone 17 was isolated in
good yield (73%), albeit with low diastereoselectivity (1.2:1)
slightly favouring the desired isomer.[17] To avoid undesired
side reactions, such as formation of the diazolactone, it proved
essential to carefully control the reaction time by quenching
the reaction after 2 min, following addition of the sulfonyl
azide. We reasoned that improved diastereoselectivity might
be achieved by fine-tuning the properties of the azidation re-
agent. Accordingly, a number of sulfonyl azides were screened
as outlined in Table 1.[18] Neither electron-rich nor -deficient ar-
ylsulfonyl azides gave improved diastereoselectivity (Table 1,
2
3
4
1.2:1
1.2:1
1:1
55
75
68
5
6
1:2
58
60
1.1:1
[a] General conditions: LiHMDS (1.5 equiv), CH2Cl2, À788C; then reagent
(1.3 equiv), À158C; then AcOH/KOAc. [b] Determined by crude NMR spec-
troscopic analysis. [c] Reagent added at À458C.
entries 2 and 3). Alkyl-substituted reagents, such as benzylsul-
fonyl azide, also did not change the reaction outcome
(entry 4). However, when (+)-camphorsulfonyl azide was em-
ployed, the product was obtained with a diastereomeric ratio
(d.r.) of 2:1 favouring the undesired isomer (entry 5). Interest-
ingly, the enantiomeric reagent derived from (À)-camphor re-
sulted in a 1:1 selectivity, which indicated a mismatch pairing
of reagent and substrate (entry 6). To the best of our knowl-
edge, the former is the first example of a diastereoselective
enolate azidation based on reagent control.
Chem. Eur. J. 2014, 20, 1 – 11
www.chemeurj.org
3
ꢀ 2014 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
&
&
These are not the final page numbers! ÞÞ

Full Paper
Synthesis of the projected cyclization precursor then in-
volved introduction of the peptide appendage found in the
natural product. However, for these model studies, we decided
to introduce simplified side chains incorporating fewer func-
tionalities to facilitate spectroscopic analysis of the products.[19]
Accordingly, the lactone ring in 17 was first opened with ben-
zylamine (79% yield) as a mimic for the agmatidine side chain
of the natural product. Subsequent reduction of the azide
under Staudinger conditions delivered the corresponding pri-
mary amine (73%). Treatment of this intermediate
with Bz2O afforded bisamide 18 in 76% yield. Both
amide nitrogen atoms in 18 could now serve as po-
tential nucleophiles in the cyclization event, thus
opening the possibility of forming two cylization
products, namely, hydroindole versus hydroquinoline.
We surmised, however, that formation of a five-mem-
bered ring by cyclization of the C(2) nitrogen atom
might be kinetically preferred over six-ring formation.
When bisamide 18 was treated with TBSOTf as an ac-
tivator, indeed, only one product was obtained. Anal-
ysis by 2D NMR spectroscopic experiments of the
TBS-protected derivative 20 (Scheme 2, box, only di-
agnostic NMR correlations shown) revealed the pres-
(9-BBN), (+)- or (À)-diisopinocampheylborane (-Ipc2BH)) were
unsuccessful. An alternative approach to hydration of the en-
docyclic double bond in 11 would proceed by initial olefin ep-
oxidation followed by selective reductive opening of the epox-
ide at C(7). Accordingly, diene 11 was subjected to epoxidation
by using mCPBA (93% yield) followed by palladium-catalysed
hydrogenation of the a,b-unsaturated ester and concomitant
base promoted cyclization to produce epoxylactone 24 in 78%
yield (Scheme 3). Introduction of the C(2) amine was again
Scheme 3. Synthesis of epoxide 28: a) mCPBA (2.0 equiv), CH2Cl2, 258C, 6 h, 93%; b) H2
(1 atm.), Pd/C (10 wt%), 1 h; then K2CO3 (5 mol%), MeOH, 258C, 78%; c) LiHMDS
(1.5 equiv), À788C, 30 min; then TrisN3 (1.3 equiv), À458C, 2 min; then AcOH/KOAc
(4.6 equiv), CH2Cl2, 58% (92% brsm), d.r. 1.2:1; d) O-TBDPS-4-amino-1-butanol (27)
(1.1 equiv), CH2Cl2, 258C, 12 h, 94%; e) H2 (1 atm.), Pd/C, MeOH, 258C, 1 h; then CbzCl
(1.2 equiv), NaHCO3 (5.0 equiv), EtOAc/H2O (1:1), 258C, 12 h, 88%; mCPBA=m-chloroper-
benzoic acid; Cbz=benzyl carbonoyl.
ence of a free C(2) amide NH confirmed by HMBC
and COSY signals. This clearly suggested the exclu-
sive formation of a hydroquinoline system 19 instead
of the desired hydroindole derivative. A possible ex-
planation for the complete selectivity for 6-exo attack
of the benzylamide on the hydrofuran ring is the re-
duced steric hindrance of the N-benzylamide com-
pared to the more congested C(2) counterpart. To cir-
cumvent the intrinsic preference for 6-exo cyclization of the
bisamide substrate, we decided to synthesize a cyclization pre-
cursor lacking the C(1) amide. To this end, azidolactone 17 was
reduced with NaBH4 to provide an unstable diol product in
67% yield.[20,21] Immediate protection with TBDPSCl followed
by conversion of the azide into the corresponding benzoate
proceeded smoothly under the previously employed condi-
tions to give 21. When substrate 21 was subjected to opti-
mized cyclization conditions by using TMSOTf/NEt3, only the
desired hydroindoline product 22 was obtained in 66% yield.
The successful generation of this hydroindole moiety validated
the synthetic strategy towards the aeruginosin Choi core. We
next envisioned introduction of a secondary alcohol at C(6) to
complement the substitution pattern of the natural product.
However, a number of attempts including hydroboration,
metal-mediated hydration,[22] reductive epoxide opening or re-
ductive double bond transposition[23] followed by olefin oxida-
tion failed to give the desired products, such as 23. For this
reason we decided to investigate introduction of the C(6) hy-
droxyl prior to the cyclization step.
achieved as described previously[16] to give 25 in 79% yield
with a d.r. of 1.2:1. The two diastereomers could be separated
by chromatography on silica gel, and the configuration of the
major isomer was confirmed by X-ray crystallographic analysis
of derivative 26 obtained by lactone opening with 2-phenyl-
ethylamine (Scheme 3, box).[24] Interestingly, when (+)-cam-
phorsulfonyl azide was used as an azidation reagent for this
substrate, the product was formed in a diastereomeric ratio of
1:5, again favouring the undesired isomer as observed before
(data not shown, for further details see the Supporting Infor-
mation). Next, introduction of the agmatidine side chain of mi-
crocin SF608 was envisioned by opening of the lactone ring by
an amine nucleophile. Model studies had indicated before that
a protected guanidine derivative[25] would not be tolerated
under the reaction conditions required for the nucleophilic
opening of the oxabicycle (TMSOTf/NEt3). We therefore envi-
sioned the introduction of a surrogate of the agmatidine side
chain. To this end, azidolactone 25 was opened smoothly with
O-TBDPS-protected 4-amino-1-butanol 27 in 94% yield. The
protected alcohol could later serve as a handle for the intro-
duction of the guanidine. The azide was subsequently hydro-
genated and the resulting free amine was protected as a Cbz
carbamate under biphasic conditions to deliver 28 (88%). With
epoxide 28 in hand, we now focused on the installation of the
C(6) hydroxyl group by reduction of the C(7)ÀO bond.
C(6) oxidation by epoxidation
Initial experiments for selective hydroboration/oxidation of the
double bond in oxabicyclic substrates, such as 14 or 18, under
a variety of conditions (BH3·SMe2, 9-borabicyclo[3.3.1]nonane
&
&
Chem. Eur. J. 2014, 20, 1 – 11
www.chemeurj.org
4
ꢀ 2014 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
ÝÝ These are not the final page numbers!

Full Paper
Reduction of epoxide 28
The reductive opening of epoxides is a widely used transfor-
mation for the synthesis of alcohols. Generally, the reductions
involve harsh reaction conditions or highly reactive reagents,
such as LiAlH4, Birch conditions (Li/NH3) or LiBEt3H (super hy-
dride).[26] We reasoned that the dense substitution pattern of
epoxide 28 including the TBDPS ether, Cbz carbamate, tertiary
alcohol as well as the strained ether bridge would not allow
for the use of these commonly employed reagents. However,
RajanBabu and Nugent reported in 1990 the use of
[TiIIICl(Cp)2][27] in the presence of hydrogen donors to effect re-
ductive cleavage of highly functionalized oxiranes.[28] The high
functional-group tolerance of the conditions described
prompted us to study this approach for substrate 28 as out-
lined in Scheme 4.
Scheme 5. Proposed equilibrium between radical intermediates 31 and 32.
ductive tranformation, while regioisomer 31 has no compara-
ble low-energy pathway available.
Selective reductive quenching of only one radical intermedi-
ate is difficult to explain on the basis of steric or electronic fac-
tors, as the only differentiating stereocenter at C(3a) seems too
remote from the reactive site to impact the reaction.
We therefore reasoned that a directing effect, possi-
bly exerted by the C(3a) alkyl substituent, might de-
termine the choice for one reaction path over the
other (Scheme 6). In particular, this side chain could
act as an internal quencher of radical 32 by intramo-
lecular hydrogen atom delivery. Two RÀH bonds can
thereby be considered as potential hydrogen atom
sources: 1) the C(2)ÀH or, alternatively, 2) the NH of
the proximal Cbz carbamate. Although OH and NH
groups are known to be very poor hydrogen atom
donors due to the heteroatom–H bond strength (85–
Scheme 4. Reductive epoxide opening of 28. Conditions: [TiCl(Cp)2] (2.0 equiv), THF,
258C, 6 h, 70% (combined yield of 29a and 29b). Cp=cyclopentadienyl.
To our surprise, when 28 was treated with freshly prepared
[TiCl(Cp)2] in the absence of a hydrogen atom donor, formation
of a single product was observed by TLC analysis and LCMS
monitoring. Spectroscopic analysis revealed that only the de-
sired C(6) alcohol was formed. However, the stereocenter at
C(2) had epimerized completely to produce a chromatographi-
cally inseparable 1:1 mixture of diastereomers 29a and 29b in
a combined yield of 70%. The regioisomeric product 30 was
only isolated in a trace amount (4% yield) but without any epi-
merization at C(2).[29] This unexpected and puzzling result
caught our interest and a more detailed analysis of this reac-
tion was initiated.
Mechanistic studies on the epoxide opening
Scheme 6. Trapping of radical 32 by intramolecular hydrogen atom delivery.
Based on the generally accepted mechanism for this transfor-
mation,[30] it can be assumed that after initial association of
[TiCl(Cp)2] to the oxirane oxygen reductive cleavage of one
CÀO bond in 28 occurs to produce either radical 31 or 32 as
indicated in Scheme 5. An equilibrium between these two spe-
cies likely exists, as has been postulated for 1,2-disubstituted
epoxide substrates by Gansꢁuer et al. based on computational
and experimental studies.[30b] It is, however, not obvious that
formation of one of the two radical intermediates is preferred
over the other. To account for the exclusive formation of prod-
uct 29, we therefore hypothesize that a Curtin–Hammett sce-
nario is operating, whereby only one of the two rapidly equili-
brating radical intermediates, namely 32, undergoes further re-
110 kcalmolÀ1 for NÀH), a number of recent reports have dem-
onstrated that this bond strength can be considerably attenu-
[31]
ated upon coordination of a Lewis acid, such as BEt3
or
[TiClCp2],[32] to the heteroatom. Thereby, these groups can be
turned into potent hydrogen atom donors. Initial molecular
modelling of radical intermediate 32 indicated that C(2)ÀH is
not within good reach for the C(7) radical as a consequence of
the constraints induced by the bicyclic system. However, the
carbamate NH can be brought into alignment with the SOMO
of a carbon centered radical at C(7) in 32. To test this hypothe-
sis, a number of simple deuteration experiments were carried
out (data not shown, see Supporting Information for more de-
Chem. Eur. J. 2014, 20, 1 – 11
www.chemeurj.org
5
ꢀ 2014 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
&
&
These are not the final page numbers! ÞÞ

Full Paper
tails). Deuterium incorporation into the C(7) methylene group
of the product could indeed be observed when the carbamate
NH proton of the substrate had been exchanged by deuteri-
um. A mechanistic rationale based on hydrogen abstraction at
the nitrogen does not yet explain epimerization at C(2) of the
product. However, a number of theoretical and experimental
studies on nitrogen-centered radicals suggests that these spe-
cies readily undergo a [1,2]-hydrogen migration to produce
a more stable carbon centered radical.[33] If such a reaction
path is operating in our system, the observed epimerization at
C(2) could be readily explained by formation of an intermedi-
ate radical at C(2) through such an [1,2]-hydrogen shift in inter-
mediate 33 to the adjacent nitrogen-centered radical produc-
ing 34 (Scheme 6).
To gain further insight into the mechanistic possibilities so
far discussed, a number of experiments were carried out as de-
picted in Scheme 7. Initially, we tested epoxide substrate 35
with R configuration at C(2) under the reaction conditions pre-
viously employed for the reductive opening of epimer 28
(Scheme 7, Eq. (1)). Interestingly, although the regioselectivity
remained high, only partial epimerization at C(2) had occurred
in the product 29b (d.r. 2:1). Thus, only the epimerization
event seems to depend on the C(2) configuration, whereas the
regioselectivity determining step does not. This finding renders
a direct abstraction of the C(2)ÀH by an intermediate C(7) radi-
cal unlikely. Assuming that the postulated [1,2]-hydrogen shift
is indeed operating, substitution of the Cbz carbonyl group on
nitrogen by an alkyl substituent would open the possibility of
an alternative [1,2]-migration path involving this alkyl substitu-
ent. This should result in reduced epimerization at C(2).
Indeed, when benzyl amine 36 was treated with [TiCl(Cp)2],
the expected product 37 was obtained with excellent regiose-
lectivity with a d.r. of 2.5:1 (Scheme 7, Eq. (2)). This result is
consistent with partial quenching of the nitrogen-centered rad-
ical through [1,2]-hydrogen migration from the benzyl sub-
stituent. We next envisioned to probe a possible influence of
the C(3a) hydroxyl group on the reaction outcome. According-
ly, deoxygenated substrate 38 was synthesized[34] and subject-
ed to [TiCl(Cp)2] (Scheme 7, Eq. (3)). A mixture of regioisomeric
alcohol products 39 and 40 was obtained in a ratio of 5:1. The
reduced selectivity observed for this substrate indicates some
minor influence of the C(3a) hydroxyl group. This can either be
explained by direct electronic control of the reaction or by in-
fluence of the substitution at C(3a) on the conformational flexi-
bility of the alkyl side chain harbouring the NH group. Finally,
extensive optimization of the reaction conditions revealed that
addition of a large excess of the hydrogen atom donor 1,4-cy-
clohexadiene (1,4-CHD) along with a lowering of the reaction
temperature to 08C led to reduced epimerization to produce
the desired product with a d.r. of 3:1 (Scheme 7, Eq. (4)). Alter-
natively, water can be used as a hydrogen atom source,[32] al-
though the product is obtained in lower yield (37%, 71%
brsm). Once again, under these conditions we observed high
regioselectivity. This result indicates that initial regioselective
quenching of the C(7) radical is fast and thus is not intercepted
by an external hydrogen atom donor. In contrast, the epimeri-
zation event can be inhibited by an external reductant. In addi-
tion, we have observed that performing the reductive opening
of diastereomer 35 at reflux temperature produces a fully epi-
merized product ((Scheme 7, Eq. (1)). It is noteworthy, that this
strategy allows for partial recovery of the undesired C(2) dia-
stereomer initially produced during the azidation reaction of
lactone 24.
In summary, Scheme 8 depicts a detailed mechanistic ex-
planation of this unusual regioselective epoxide reduction
based on the aforementioned experimental data. We hypothe-
size that treatment of epoxide 28 with [TiCl(Cp)2] initially gen-
erates two rapidly equilibrating radical intermediates 31 and
32 by homolytic cleavage of one CÀO bond of the oxirane. In
accordance with the Curtin–Hammett principle, only the C(7)
radical isomer 32 can undergo rapid follow-up reaction while
isomer 31 only slowly undergoes intermolecular reductive
quenching by the solvent THF. Radical 32 is trapped through
intramolecular hydrogen-atom donation by the carbamate NH,
enabled by coordination of titanium to the Cbz carbonyl
group. It is useful to consider the resulting nitrogen-centered
Scheme 7. Investigation of factors influencing the epoxide reduction.
&
&
Chem. Eur. J. 2014, 20, 1 – 11
www.chemeurj.org
6
ꢀ 2014 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
ÝÝ These are not the final page numbers!

Full Paper
protocols based on acid-mediated activation of the tertiary al-
cohol (TFA, BF3·OEt2, ZnI2[35]) to generate a carbocation inter-
mediate followed by reductive quenching were investigated.
Under a variety of conditions tested, complex product mixtures
resulted. This can likely be attributed to rearrangement events
of the intermediate oxabicyclic carbocation. To circumvent
problems associated with carbocation formation, we turned
our attention to radical-based deoxygenation strategies.[36] In-
terestingly though, functionalization of the C(3a) hydroxyl
group proved challenging. In particular, conversion into con-
ventional precursors prescribed for deoxygenation, such as
xanthates or thioester derivatives, failed, leading to either re-
covery of starting material or complete decomposition when
strong bases were employed (e.g. KH). We then turned our at-
tention to a scarcely used photochemical deoxygenation pro-
tocol originally reported by Saito.[37] This procedure relies on
conversion of the alcohol into an activated m-CF3 benzoate de-
rivative. Irradiation of this species with UV light in the presence
of N-methylcarbazole initiates an electron transfer from the
electron-rich carbazole to the p-system of the benzoate. The
resulting aryl radical anion then undergoes protonation fol-
lowed by fragmentation, ultimately leading to cleavage of the
CÀO bond. To test this approach, benzoate 43 was synthesized
by selective acetylation of the secondary alcohol in 29 with
Ac2O (99%) followed by treatment of the tertiary alcohol with
3-trifluoromethylbenzoyl chloride, affording ester 43 in 91%
yield (Scheme 9). Irradiation of 43 with UV light in the pres-
ence of N-methylcarbazole and 1,4-cyclohexadiene as a hydro-
gen-atom donor produced the desired deoxygenated product
44 in an excellent yield of 68% (96% brsm) via intermediates
45 and 46. It is noteworthy that tertiary radical 46 is trapped
exclusively from the exo-face producing only the observed
product 44. Moreover, the clean outcome of this reaction is in
stark contrast with previous observations reporting that terti-
ary alcohol substrates are prone to disproportionation, gener-
ating significant amounts of an alkene side product.[37a] We did
not observe such side reactions.
Scheme 8. Mechanistic rationale based on the experimental data.
radical 33 in its mesomeric form 41. In the case of a Cbz-pro-
tected amine, this allows for two alternative pathways for ab-
straction of the C(2)ÀH. Thus, either of two pathways may be
envisioned, involving a [1,2]-H-shift from intermediate 33 or
a 1,4-hydrogen abstraction from C to O from radical 41 as de-
picted. Both pathways lead to intermediate 34 with concomi-
tant loss of the C(2) stereocenter. Finally, generation of the ob-
served product can either occur through radical quenching by
solvent or by titanium enolate 42 and subsequent protonation
during workup. In the presence of an excess of a reactive H-
atom donor, the nitrogen-centered radical 33 may be inter-
cepted before hydrogen migration can occur, thus
leading to product 29a that has not suffered epimer-
ization. The efficiency of the [1,2]-shift likely depends
on the conformation of the C(2)ÀN bond. Only when
an optimal orbital overlap between the SOMO of the
nitrogen and the CÀH bond is achieved can the mi-
gration occur. The conformational preference and
flexibility of the C(2)ÀN bond is influenced by
a number of factors including the configuration at
C(2), the reaction temperature as well as the substitu-
ents on the nitrogen atom.
Excision of the C(3a) hydroxyl group
After successful installation of the C(6) hydroxyl
group, we now focused on accessing a cyclization
Scheme 9. Synthesis of cyclization precursor 47: a) Ac2O (1.5 equiv), NEt3 (4.0 equiv),
DMAP (cat.), CH2Cl2, 258C, 30 min, 99%; b) m-CF3-C6H4COCl (1.5 equiv), NEt3 (3.0 equiv),
DMAP (cat.), CH2Cl2, 258C, 2 h, 91%; c) hn (UV), N-methylcarbazole (1.5 equiv), 1,4-cyclo-
hexadiene (100 equiv), THF/H2O (1:1), 258C, 12 h, 68% (96% brsm); d) H2 (1 atm.), Pd/C
precursor. To address the functionalization pattern of
the Choi core of the aeruginosins, deoxygenation of
the C(3a) alcohol is needed. To this end, a number of
deoxygenation procedures were evaluated. Initially, (10 wt.%), MeOH, 258C, 1 h, 93%.
Chem. Eur. J. 2014, 20, 1 – 11
www.chemeurj.org
7
ꢀ 2014 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
&
&
These are not the final page numbers! ÞÞ

Full Paper
Nucleophilic opening of oxabicycle 47
acetate to the C(5) alcohol were observed. After extensive
screening of conditions (see the Supporting Information for
details), we found that hexafluorosilic acid in the presence of
triethylamine cleanly afforded a mixture of fully deprotected
hydroindole 49 along with some partially deprotected prod-
uct.[38] Subjecting this compound again to the same conditions
gave aminoalcohol 49 in a total of 81% yield after three
rounds of deprotection.
We now turned our attention to the key hydroindole ring clo-
sure. As observed before, the presence of two amide substitu-
ents in a cyclization precursor led to the exclusive formation of
a hydroquinoline product (vide supra, Scheme 2). However, we
reasoned that the presence of an unprotected amine at C(2)
might override this inherent preference for 6-exo cyclization
based on the higher nucleophilicity of a primary amine com-
pared to an amide nitrogen. Accordingly, the Cbz protecting
group in 44 was removed under hydrogenolytic conditions to
provide amine 47 in excellent yield. Upon exposure of 47 to
TMSOTf/NEt3 formation of a single product 48 was observed
as monitored by TLC and LCMS (Scheme 10). Isolation and
With the aeruginosin core fragment 49 in hand, only a few
steps were required to access the natural product microcin
SF608. Introduction of the dipeptide side chain 50 was studied
first. Initial experiments revealed that the C(2’) stereocenter is
prone to epimerization during the peptide coupling event. A
number of coupling reagents (N’-(3-dimethylaminopropyl)-N-
ethylcarbodiimide (EDC), (benzotriazol-1-yloxy)tripyr-
rolidinophosphonium (PyBOP),[39] 3-(diethoxyphos-
phorloxy)-1,2,3-benzotriazin-4(3H)-one (DEPBT),[40] (1-
[bis(dimethylamino)methylene]-1H-1,2,3-triazolo[4,5-
b]pyridinium 3-oxid hexafluorophosphate) (HATU))
and conditions were evaluated to avoid epimeriza-
tion (see the Supporting Information for more de-
tails).[41] Finally, we found HATU to be the most effec-
tive coupling agent. Under these conditions amide
coupling of hindered secondary amine 49 with acid
50 proceeded in 77% yield producing peptide 51
without any observable epimerization.
The hydroindole core in 51 still incorporates a hy-
droxyl group at C(5), as a remnant of the nucleophilic
opening of the oxabicyclic framework. To access mi-
crocin SF608, this alcohol needed to be removed as
the natural product harbours a methylene at this po-
sition. However, it is important to note that a sub-
group of aeruginosin protease inhibitors, namely the
dysinosins, incorporate
a S-configured hydroxyl
group at C(5) of the Choi core (e.g. dysinosin A (3),
Scheme 2).[5a,c] Thus, the C(5) alcohol in 51 can serve
as a handle to access both Choi substructures from
one common intermediate. To this end, inversion of
the C(5) hydroxyl was tested. Under standard Mitsu-
nobu conditions (diisopropyl azodicarboxylate
(DIAD), PPh3, 4-nitrobenzoic acid) no product was ob-
tained. This is likely due to the inaccessibility of the
concave face of the hydroindole ring, thus blocking
SN2 attack on C(5). However, a sequence including
oxidation of the alcohol to the corresponding ketone
(91%) followed by NaBH4 reduction delivered the de-
sired C(5)-S-configured alcohol 52 in 63% yield. Nota-
bly, ketone reduction occurred exclusively from the
Scheme 10. Synthesis of Choi derivatives 52 and 53: a) TMSOTf (6.0 equiv), NEt3
(8.0 equiv), CH2Cl2, 258C, 1 h, 81%; f) H2SiF6 (1.0 equiv), NEt3 (1.5 equiv), MeCN, 658C,
12 h, 81% (after three rounds of deprotection); c) 50 (1.1 equiv), HATU (1.2 equiv),
iPr2NEt (1.2 equiv), CH2Cl2, 258C, 12 h, 77%; d) TPAP (0.1 equiv), NMO (3.0 equiv), CH2Cl2,
258C, 30 min, 91%; e) NaBH4 (2.0 equiv), THF, 08C, 30 min, 63%; f) m-CF3-C6H4COCl
(1.5 equiv), NEt3 (3.0 equiv), DMAP (cat.), CH2Cl2, 258C, 2 h, 59%; g) hn (UV), N-methylcar-
bazole (1.0 equiv), 1,4-cyclohexadiene (200 equiv), THF/H2O (1:1), 258C, 12 h, 37% (75%
brsm). TMS=trimethylsilyl; Tf=trifluoromethanesulfonyl; HATU=O-(7-azabenzotriazol-1-
yl)-N,N,N’,N’-tetramethyluronium hexafluorophosphate; DMF=dimethylformamide.
characterization of the product revealed that the desired hy-
droindole system is formed exclusively, and no product result-
ing from attack of the C(1) amide nitrogen atom could be de-
tected. The product was obtained as a bistrimethylsilyl deriva-
tive. Removal of the silyl groups proved unexpectedly difficult.
Using mild deprotection protocols no reaction was observed
and starting material was reisolated unchanged. In contrast,
when employing harsher conditions, undesired side reactions
including loss of the TBDPS group and migration of the C(6)
convex face of the hydroindole ring. The microcin Choi core in
turn was accessed using the deoxygenation protocol used pre-
viously.[37] Thus alcohol 51 was converted into an intermediate
m-CF3 benzoate in 59% yield. Irradiation with UV light pro-
duced 53 in 37% yield (75% brsm). During the photochemical
step some hydrolysis of the benzoate ester was observed. This
side product could be easily recovered and was subjected
again to the deoxygenation conditions.
&
&
Chem. Eur. J. 2014, 20, 1 – 11
www.chemeurj.org
8
ꢀ 2014 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
ÝÝ These are not the final page numbers!

Full Paper
Scheme 11. Completion of the total synthesis of microcin SF608: a) TAS-F (2.0 equiv), MeCN, 258C, 2 h, 88%; b) MsCl (1.2 equiv), NEt3 (1.5 equiv), CH2Cl2, 258C,
1 h, 96%; c) NaN3 (2.0 equiv), DMF, 508C, 12 h, 85%; d) LiOH (2.0 equiv), THF/H2O (5:1), 08C, 66% (70% brsm); e) PMe3 (2.0 equiv), MeCN/THF (5:1), 258C,
30 min; then H2O; then 56 (2.0 equiv), NEt3 (2.0 equiv), CH2Cl2, 258C, 2 h, 85%; f) H2 (1 atm.), Pd/C (10 wt.%), MeOH, 258C, 6 h, 87%; TAS-F=tris(dimethylami-
no)sulfonium difluorotrimethylsilicate.
tion was discovered in the course of the total synthesis. De-
tailed experimental studies prompted us to propose a mecha-
nistic scheme for this transformation. Initial homolytic CÀO
bond cleavage promoted by titanium(III) is followed by intra-
molcular hydrogen atom delivery form a proximal NH group
trapping the intermediate carbon centered radical. We further
postulate that the generated nitrogen-centered radical can
then undergo a [1,2]-hydrogen shift. This results in epimeriza-
tion of the adjacent stereocenter in the epoxide substrate.
These findings might open new possibilities for selective epox-
ide reduction in complex settings.
Completion of the total synthesis of microcin SF608
To complete the total synthesis of 2, the task of introducing
the guanidine functionality on the agmatidine side chain re-
mained. To this end, the TBDPS protecting group was removed
using TAS-F and the primary alcohol was converted into mesy-
late 54 by using MsCl (Scheme 11). When mesylate 54 was
treated with NaN3 clean conversion to azide 55 was observed
(85% yield). After cleavage of the acetate protecting group
with LiOH (66%), the azide was reduced under Staudinger con-
ditions followed by treatment with Goodman’s reagent 56[42]
to afford Cbz-protected guanidine 57 (85%). Finally, all the re-
maining protecting groups could be globally removed by hy-
drogenation using Pd/C to afford microcin SF608 in 87% yield.
It is noteworthy, that this protecting group strategy allows for
easy purification of the natural product by simple filtration
without the need of HPLC purification. The spectroscopic data
obtained with synthetic 2 matched the reported data in all re-
spects.[5b,12]
Acknowledgements
We are grateful to Dr. B. Schweizer and M. Solar for X-ray crys-
tallographic analysis of 26, and to Dr. M.-O. Ebert, R. Arnold, R.
Frankenstein and P. Zumbrunnen for NMR spectroscopic mea-
surements. S.D. acknowledges the Stipendienfonds Schweizeri-
sche Chemische Industrie (SSCI) and Novartis for a predoctoral
fellowship. C.S.S. thanks the Roche Research Foundation for
a predoctoral fellowship.
Conclusion
We have developed a general entry route to the carboxy hy-
droxyl octahydroindole derivatives of the aeruginosin peptides
relying on the nucleophilic opening of oxabicyclo-
[2.2.1]heptanes. The strategy we document enables access to
all the core structures of the aeruginosin serine proteases from
a common oxabicyclic building block 11. We have implement-
ed this strategy in the total synthesis of microcin SF608, which
was accessed in 21 synthetic steps from 11. During these syn-
thetic studies we have gained additional insight into the key
nucleophilic opening of bicyclic hydrofuran derivatives. More-
over, a highly regioselective titanium mediated epoxide reduc-
Keywords: aeruginosins · epoxide reduction · hydroindoles ·
natural products · serine protease inhibitors
[1] M. A. Phillips, R. J. Fletterick, Curr. Biol. Curr. Opin. Sturct. Biol. 1992, 2,
713–720.
[2] Proteases in Health and Disease in Progress in Molecular Biology and
Translational Science Vol. 99 (Ed: E. Di Cerca), Elsevier Academic Press,
Burlington, 2011, pp. 1–313.
[3] a) D. Voet, J. G. Voet in Biochemistry 4th edition, John Wiley & Sons,
Hoboken, 2011, 525–538; on the mechanism of serine proteases, see
also: b) C. S. Craik, S. Roczniak, C. Largman, W. R. Rutter, Science 1987,
237, 909–913.
Chem. Eur. J. 2014, 20, 1 – 11
www.chemeurj.org
9
ꢀ 2014 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
&
&
These are not the final page numbers! ÞÞ

Full Paper
[4] M. Murakami, Y. Okita, H. Matsuda, T. Okino, K. Yamaguchi, Tetrahedron
Lett. 1994, 35, 3129–3132.
[24] CCDC-785240 (26) contains the supplementary crystallographic data for
this paper. These data can be obtained free of charge from The Cam-
bridge Crystallographic Data Centre via www.ccdc.cam.ac.uk/data_re-
quest/cif.
[25] tert-Butoxycarbonyl (Boc)- or Cbz-protected amine and guanidine deriv-
atives proved incompatible with the cyclization conditions (TMSOTf/
NEt3). In the case of Boc-protected nitrogen functionalities, loss of the
Boc group was observed.
[5] For a comprehensive review covering the chemistry and biology of the
aeruginosin serine protease inhibitors, see: a) K. Ersmark, J. R. Del Valle,
S. Hanessian, Angew. Chem. 2008, 120, 1220–1242; Angew. Chem. Int.
Ed. 2008, 47, 1202–1223; b) for the isolation of microcin SF608, see: R.
Banker, S. Carmeli, Tetrahedron 1999, 55, 10835–10844; c) for the isola-
tion of dysinosin A, see: A. R. Carroll, G. K. Pierens, G. Fechner, P. deAl-
meida Leone, A. Ngo, M. Simpson, E. Hyde, J. N. A. Hooper, S.-L. Bos-
trçm, D. Musil, R. J. Quinn, J. Am. Chem. Soc. 2002, 124, 13340–13341.
[6] The guanidine group has been shown to play a key role in binding of
the aeruginosin to the active site of serine proteases. For crystal struc-
tures of serine proteases in complex with aeruginosin peptides, see
ref. [5a] and in particular: J. L. Rios Steiner, M. Murakami, A. Tulinsky, J.
Am. Chem. Soc. 1998, 120, 597–598.
[7] For the isolation of banyaside B, see: a) A. Pluotno, S. Carmeli, Tetrahe-
dron 2005, 61, 575–583; b) for the isolation of related natural products,
see: K. Fujii, K. Sivonen, K. Adachi, K. Noguchi, Y. Shimizu, H. Sano, K.
Hirayama, M. Suzuki, K. Harada, Tetrahedron Lett. 1997, 38, 5529–5532.
[8] For a few selected examples, see: a) N. Valls, M. Lꢂpez-Canet, M. Vallri-
bera, J. Bonjoch, J. Am. Chem. Soc. 2000, 122, 11248–11249; b) P. Wipf,
J.-L. Methot, Org. Lett. 2000, 2, 4213–4216; c) Y. Fukuta, T. Ohshima, V.
Gnanadesikan, T. Shibuguchi, T. Nemoto, T. Kisugi, T. Okino, M. Shibasa-
ki, Proc. Natl. Acad. Sci. USA 2004, 101, 5433–5438; d) S. Hanessian, R.
Margarita, A. Hall, S. Johnstone, M. Tremblay, L. Parlanti, J. Am. Chem.
Soc. 2002, 124, 13342–13343; e) S. Hanessian, M. Tremblay, J. F. W. Pe-
tersen, J. Am. Chem. Soc. 2004, 126, 6064–6071.
[9] C. S. Schindler, C. R. J. Stephenson, E. M. Carreira, Angew. Chem. 2008,
120, 8984–8987; Angew. Chem. Int. Ed. 2008, 47, 8852–8855.
[10] C. S. Schindler, L. Bertschi, E. M. Carreira, Angew. Chem. Int. Ed. 2010, 49,
9229–9232.
[11] C. S. Schindler, S. Diethelm, E. M. Carreira, Angew. Chem. 2009, 121,
6414–6417; Angew. Chem. Int. Ed. 2009, 48, 6296–6299.
[12] For a previous total synthesis of microcin SF608, see: N. Valls, M. Vallri-
bera, M. Lꢂpez-Canet, J. Bonjoch, J. Org. Chem. 2002, 67, 4945–4950.
[13] S. Diethelm, C. S. Schindler, E. M. Carreira, Org. Lett. 2010, 12, 3950–
3953.
[14] NHC ligand 15 was synthesized according to: A. J. Arduengo III, R.
Krafczyk, R. Schmutzler, H. A. Craig, J. R. Goerlich, W. J. Marshall, M. Un-
verzagt, Tetrahedron 1999, 55, 14523–14534.
[15] V. Jurkauskas, J. P. Sadighi, S. L. Buchwald, Org. Lett. 2003, 5, 2417–
2420.
[16] D. A. Evans, T. C. Britton, J. A. Ellman, R. L. Dorow, J. Am. Chem. Soc.
1990, 112, 4011–4030.
[17] The two diastereomers of 17 were not separable by column chroma-
tography. The relative configuration was assigned in analogy to the re-
action of epoxide 25 (vide infra), which could be characterized by X-ray
crystallographic analysis of derivative 26.
[18] For a more detailed screening of reagents and their preparation, see
the Supporting Information, Table S2, page S6.
[19] We observed that many of these Choi containing peptides appear as
rotamers, thus making spectroscopic analysis more difficult.
[20] We have previously observed that oxabicyclic system harboring an en-
docyclic double bond and an azide functionality in its proximity are
prone to decomposition. This could be explained by reaction of the
azide with the strained olefin followed by further decomposition.
[21] When the reduction of 17 was performed in MeOH instead of THF, ex-
clusive formation of an enamine product was observed: The structure
[26] S. Murai, T. Murai, S. Kato in Comprehensive Organic Synthesis Vol. 8
(Eds. B. M. Trost, I. Fleming), Pergamon, Oxford, 1991, pp. 871–893.
[27] For all reactions described [TiIIICl(Cp)2] was freshly prepared from
[TiIVCl2(Cp)2] and Zn powder (see the Supporting Information).
[28] a) T. V. RajanBabu, W. A. Nugent, M. S. Beattie, J. Am. Chem. Soc. 1990,
112, 6408–6409; b) T. V. RajanBabu, W. A. Nugent, J. Am. Chem. Soc.
1994, 116, 986–997; c) for a review on titanium mediated epoxide re-
duction, see: A. Gansꢁuer, T. Lauterbach, S. Narayan, Angew. Chem.
2003, 115, 5714–5731; Angew. Chem. Int. Ed. 2003, 42, 5556–5573.
[29] In all reactions performed, re-isolated starting material and all side
products including regioisomeric alcohol and alkene retained the con-
figuration at C(2) (see the Supporting Information). This strongly sug-
gests that epimerization is directly coupled to the formation of the de-
sired regioisomer 29.
[30] For early mechanistic investigations, see ref. [28b]; for further experi-
mental and computational studies, see also: a) K. Daasbjerg, H. Svith, S.
Grimme, M. Gerenkamp, C. Mꢃck-Lichtenfeld, A. Gansꢁuer, A. Barchuk, F.
Keller, Angew. Chem. 2006, 118, 2095–2098; Angew. Chem. Int. Ed. 2006,
45, 2041–2044; b) A. Gansꢁuer, A. Barchuk, F. Keller, M. Schmitt, S.
Grimme, M. Gerenkamp, C. Mꢃck-Lichtenfeld, K. Daasbjerg, H. Svith, J.
Am. Chem. Soc. 2007, 129, 1359–1371.
[31] a) D. A. Spiegel, K. B. Wiberg, L. N. Schacherer, M. R. Medeiros, J. L.
Wood, J. Am. Chem. Soc. 2005, 127, 12513–12515; b) J. Jin, M. New-
comb, J. Org. Chem. 2007, 72, 5098–5103; for a related example, see:
c) D. Pozzi, E. M. Scanlan, P. Renaud, J. Am. Chem. Soc. 2005, 127,
14204–14205; for computational studies, see: d) W. Tantaway, H. Zipse,
Eur. J. Org. Chem. 2007, 5817–5820.
[32] a) J. M. Cuerva, A. G. CampaÇa, J. Justicia, A. Rosales, J. L. Oller-Lꢂpez, R.
Robles, D. J. Cꢄrdenas, E. BuÇuel, J. E. Oltra, Angew. Chem. 2006, 118,
5648–5652; Angew. Chem. Int. Ed. 2006, 45, 5522–5526; b) J. Jin, M.
Newcomb, J. Org. Chem. 2008, 73, 7901–7905; c) M. Paradas, A. G. Cam-
paÇa, T. Jimꢅnez, R. Robles, J. E. Oltra, E. BuÇuel, J. Justicia, D. J. Cꢄrde-
nas, J. M. Cuerva, J. Am. Chem. Soc. 2010, 132, 12748–12756.
[33] a) B. F. Yates, L. Radom, J. Am. Chem. Soc. 1987, 109, 2910–2915; b) V.
ˇ
Vrcek, H. Zipse, J. Org. Chem. 2009, 74, 2947–2957.
[34] Compound 38 was synthesized by using the same deoxygenation pro-
tocol as used for alcohol 29 later in the synthesis.
[35] C. K. Lau, C. Dufresne, P. C. Bꢅlanger, S. Piꢅtrꢅ, J. Scheigetz, J. Org. Chem.
1986, 51, 3038–3043.
[36] For a review covering radical deoxygenation of alcohols, see: W. Hart-
wig, Tetrahedron 1983, 39, 2609–2645.
[37] a) I. Saito, H. Ikehira, R. Kasatani, M. Watanabe, T. Matsuura, J. Am. Chem.
Soc. 1986, 108, 3115–3117; for an application of this methodology in
the context of natural product synthesis, see also: b) A. G. Myers, K. R.
Condroski, J. Am. Chem. Soc. 1995, 117, 3057–3083.
[38] a) R. F. Newton, D. P. Reynolds, M. A. W. Finch, D. R. Kelly, S. M. Roberts,
Tetrahedron Lett. 1979, 20, 3981–3982; b) A. S. Pilcher, D. K. Hill, S. J.
Shimshock, R. E. Waltermire, P. DeShong, J. Org. Chem. 1992, 57, 2492–
2495.
[39] J. Coste, D. Le-Nguyen, B. Castro, Tetrahedron Lett. 1990, 31, 205–208.
[40] H. Li, X. Jiang, Y.-h. Ye, C. Fan, T. Romoff, M. Goodman, Org. Lett. 1999,
1, 91–93.
[41] For a review covering peptide coupling reagents, see: E. Valeur, M.
Bradley, Chem. Soc. Rev. 2009, 38, 606–631.
[42] a) K. Feichtinger, C. Zapf, H. L. Sings, M. Goodman, J. Org. Chem. 1998,
63, 3804–3805; b) K. Feichtinger, H. L. Sings, T. Baker, K. Matthews, M.
Goodman, J. Org. Chem. 1998, 63, 8432–8439.
was confirmed by 2D NMR spectroscopic analysis (see the Supporting
Information).
[22] P. Magnus, A. H. Payne, M. J. Waring, D. A. Scott, V. Lynch, Tetrahedron
Lett. 2000, 41, 9725–9730.
Received: January 6, 2014
[23] A. G. Myers, B. Zheng, Tetrahedron Lett. 1996, 37, 4841–4844.
Published online on && &&, 0000
&
&
Chem. Eur. J. 2014, 20, 1 – 11
www.chemeurj.org
10
ꢀ 2014 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
ÝÝ These are not the final page numbers!

Full Paper
FULL PAPER
&
Natural Products
S. Diethelm, C. S. Schindler, E. M. Carreira*
&& – &&
Access to the Aeruginosin Serine
Protease Inhibitors through the
Nucleophilic Opening of an
Oxabicyclo[2.2.1]heptane: Total
Synthesis of Microcin SF608
Expect the unexpected! An entry route
to the aeruginosin protease inhibitors is
reported and showcased on the total
synthesis of microcin SF608 (see
led us to propose a mechanistic ratio-
nale for this transformation involving in-
tramolecular hydrogen atom delivery by
a carbamate NH to direct the regiose-
lectivity of the homolytic epoxide cleav-
age.
scheme). Detailed experimental studies
of an unusual regioselective epoxide re-
duction observed during this synthesis
Chem. Eur. J. 2014, 20, 1 – 11
www.chemeurj.org
11
ꢀ 2014 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
&
&
These are not the final page numbers! ÞÞ