New derivatives of ascomycin with modifications in the amino acid region - Synthesis and biological activities, and X-ray crystal structure of 5,6-dehydroascomycin

Article abstract of DOI:10.1002/hlca.200800436

The immunomodulatory macrolide ascomycin (1a) inhibits T-cell activation via binding to macrophilin-12 and inhibition of the phosphatase calcineurin. Its structural analogs pimecrolimus and tacrolimus have recently become available as the first novel topical treatments of atopic dermatitis since the introduction of topical corticosteroids in the 1950s. This stimulated the search for novel derivatives with an improved biological profile. Though several derivatives of 1a are known, only a few derivatives with modifications on the amino acid moiety are available because of the chemical inaccessibility of this region. To this end, we present here a new approach using a photochemical reaction as the key step. Thus, irradiation of ascomycin (1a) led to mixtures of the methoxy products 2a and 8, the cleavage product 4a, the but-1-enyl derivative 7, and the oxazolidinone 9 depending on the solvent. The selectivity of the reaction was improved to furnish 2a or 9 in preparatively useful yields. The mechanism and scope of the reaction were investigated. Starting from 2a, several analogs featuring novel modifications on the amino acid moiety, which are not easily accessible through routine methods, were synthesized in a few steps. Further, using the photoreaction key intermediates with potential for broader modifications on the amino acid moiety were synthesized, and their utility was exemplified by the synthesis of vinylpipecolic acid and vinylproline analogs. An interesting photochemical cleavage of the amide bond in the derivatives of ascomycin (1a) is presented. The structural and conformational features of the new analogs together with the X-ray crystal structure of 5,6-dehydroascomycin (6a) are presented, and their biological activities are discussed. Of all the derivatives, 6a showed the best activities in in vitro and in vivo models of allergic contact dermatitis whilst showing a lower risk of immunosuppression.

Full text of DOI:10.1002/hlca.200800436

Helvetica Chimica Acta – Vol. 92 (2009)
839
New Derivatives of Ascomycin with Modifications in the Amino Acid Region
Synthesis and Biological Activities, and X-Ray Crystal Structure of
,6-Dehydroascomycin
–
5
1
2
by Murty A. R. C. Bulusu* ), Peter Waldstꢀtten, Thomas Tricotet, Christophe Rochais, Andrea Steck ),
2
Markus Bacher ), Gerhard Schulz, and Josef G. Meingassner
Novartis Institutes for BioMedical Research GmbH & Co KG, Brunnerstrasse 59, A-1235 Vienna
and Peter Hiestand, Gerhard Zenke, Walter Schuler, and Trixie Wagner
Novartis Institutes for BioMedical Research, Lichtstrasse 35, CH-4002 Basle
Dedicated to Professor Dr. Andrea Vasella on the occasion of his 65th birthday
The immunomodulatory macrolide ascomycin (1a) inhibits T-cell activation via binding to
macrophilin-12 and inhibition of the phosphatase calcineurin. Its structural analogs pimecrolimus and
tacrolimus have recently become available as the first novel topical treatments of atopic dermatitis since
the introduction of topical corticosteroids in the 1950s. This stimulated the search for novel derivatives
with an improved biological profile. Though several derivatives of 1a are known, only a few derivatives
with modifications on the amino acid moiety are available because of the chemical inaccessibility of this
region. To this end, we present here a new approach using a photochemical reaction as the key step. Thus,
irradiation of ascomycin (1a) led to mixtures of the methoxy products 2a and 8, the cleavage product 4a,
the but-1-enyl derivative 7, and the oxazolidinone 9 depending on the solvent. The selectivity of the
reaction was improved to furnish 2a or 9 in preparatively useful yields. The mechanism and scope of the
reaction were investigated. Starting from 2a, several analogs featuring novel modifications on the amino
acid moiety, which are not easily accessible through routine methods, were synthesized in a few steps.
Further, using the photoreaction key intermediates with potential for broader modifications on the
amino acid moiety were synthesized, and their utility was exemplified by the synthesis of vinylpipecolic
acid and vinylproline analogs. An interesting photochemical cleavage of the amide bond in the
derivatives of ascomycin (1a) is presented. The structural and conformational features of the new analogs
together with the X-ray crystal structure of 5,6-dehydroascomycin (6a) are presented, and their
biological activities are discussed. Of all the derivatives, 6a showed the best activities in in vitro and in
vivo models of allergic contact dermatitis whilst showing a lower risk of immunosuppression.
1. Introduction. – Inflammatory skin diseases such as atopic dermatitis (AD) and
psoriasis are chronic and genetically predisposed conditions that impair patientsꢀ
quality of life [1][2]. AD is the most common skin disease of young children, with a
prevalence of up to 20%, and psoriasis is mostly a disease of adulthood, affecting 2 – 3%
of the Western population. The macrolactam ascomycin (1a) is a fermentation product
1)
2)
Current address: Nabriva Therapeutics AG, Leberstrasse 20, A-1112 Vienna (phone:
þ 43-1-740931213; fax: þ 43-1-74093 1900; e-mail: murty.bulusu@nabriva.com)
Current address: Bruker Analytical Services Austria (Bruker Austria GmbH), Leberstrasse 20, A-
1112 Vienna.
ꢁ
2009 Verlag Helvetica Chimica Acta AG, Zꢂrich

8
40
Helvetica Chimica Acta – Vol. 92 (2009)
from Streptomyces hygroscopicus var. ascomyceticus, originally isolated due to its
ꢃ
antifungal activities [3]. Recently, pimecrolimus (Elidel , SDZ ASM 981; 1b), which is
ꢃ
a Cl analog of 1a, and tacrolimus (Protopic , FK 506; 1c), both possessing a common
core structure, have become available as the first novel topical treatments of AD since
the introduction of the corticosteroids more than 40 years ago [4]. The clinical results
with 1b stimulated the search for novel derivatives and their biological evaluation
[
4][5].
Although the pathophysiologies of AD and psoriasis are different, and are still a
matter of ongoing intensive research, there is general agreement that T cells play a key
role in both diseases. The initial step in the immunomodulatory activity of the
macrolides 1a – 1c is binding of the macrolide to macrophilin-12 (FKBP 12), which is
then followed by binding to calcineurin A and B, thus inhibiting the activity of the
phosphatase [6], an enzyme required for the dephosphorylation of the cytosolic form of
the nuclear factor of activated T cells (NF-AT). This leads to the inhibition of the
release of cytokines such as IFN-g, and IL-2, -4, -5, -10, and TNF-a, thus targeting T-cell
activation and proliferation. The region C(1)-C(14) comprising the pipecolic acid
(¼ piperidine-2-carboxylic acid) and the three carbonyl moieties is involved in binding
to macrophilin and hence is called the binding domain. The region C(14)-C(22) is
involved in the subsequent steps of the mechanism of action and hence termed the
effector domain. The analogs studied so far are limited with respect to the structural
variations in the binding and effector domains because of the inaccessibility, in
particular, of the pipecolic acid and C(14)-C(22) regions, for derivatization through
routine organic methods. Some of our efforts leading to new derivatives of 1a – 1c with
modifications primarily in the binding and also in effector domains have been
communicated earlier [7]. Here, we give the full details including further work and the
biological activities.
2
. Results and Discussion. – 2.1. Photochemistry of Ascomycin (1a) and Its Analogs,
and Synthesis of New Analogs. a-Keto amides upon irradiation lead to several types of
products arising primarily through intramolecular abstraction of g-H-atom by the C¼O
group. Allylic C¼O compounds undergo a variety of reactions such as oxadi-p-
methane rearrangement, 1,3-sigmatropic shifts, and Norrish type-I cleavage [8]. The
macrolide ascomycin (1a) features a novel tricarbonyl system, connected through an

Helvetica Chimica Acta – Vol. 92 (2009)
841
amide bond at one end and masked as a hemiketal at the other end, in the binding
domain, and an allylic C¼O moiety in the effector domain. Furthermore, because of
the cyclic structure, the molecule possesses a relatively rigid conformation. These
features prompted us to study the photochemistry of 1a with a view to discovering new
pathways to novel synthetic derivatives, with modifications in the amino acid and
effector domains, which are less accessible for routine derivatization.
Irradiation of 1a in MeOH for 2 h at room temperature using Pyrex-filtered UV
light (>280 nm) led to 6-b-methoxy-9-dihydroascomycin (2a; 22%), the lactone 4a
(
(
10%), and the but-1-enyl derivatives 7 (40%) and 8 (4%); in addition, unchanged 1a
19%) was recovered (Scheme 1 and Table 1). The reaction is stereoselective giving
only the 6-b-MeO isomer 2a and only one of the two possible C(19)-diastereoisomers
of each of 7 and 8 with the but-1-enyl chain exclusively with (E)-configuration. On the
other hand, irradiation of 1a in MeCN/H O 4 :1 for 2 h at room temperature led to the
2
lactone 4a (10%), the but-1-enyl compound 7 (30%), and the oxazolidinone 9 (11%);
in addition, unchanged 1a (26%) was recovered.
The formation of 2a could be rationalized through the intermediacy of the
zwitterion Z formed by electron transfer from the amide N-atom to the C(9)¼O
1
group, followed by H-transfer [8]. The alternative conformer Z is expected to be of
2
higher energy than Z because of the 1,2-steric repulsions between the C(1)(O)ꢀO and
1
the N-acyl groups. Attack of MeOH from the b-face of Z (via a chair-like transition
1
state, resulting in 2a) is favored over that from the a-face (boat-like transition state).
The transformation of 1a to 7 could be rationalized as a photochemically allowed [1,3]-
sigmatropic shift. Molecular models and X-ray crystal structure [6c – f] of 1a indicate an
2
S
2
S
ideal alignment of the bonds involved for a s þ p mode of interaction, which is
expected to result in C(19)-b-Me and (E)-but-1-enyl configurations. The formation of
only one isomer 7 with (E)-but-1-enyl chain led us to tentatively assign the
configuration of the Me group at C(19) as b. No epimerization at C(21) in the
recovered starting material could be detected, thus indicating a concerted migration
pathway for the formation of 7. The acyclic derivative 4a could be arising through
decarbonylative fragmentation. The formation of the oxazolidinone 9 in the non-
nucleophilic solvent MeCN involves intramolecular a-attack of the O-atom at C(9)
(
i.e., Z ) through the higher-energy boat-like transition state. That no oxazolidinone 9 is
1
formed upon irradiation of 1a in MeOH would imply that b-attack by MeOH on Z1
(
(
through chair-like transition state) is more favored than intramolecular a-attack
through boat-like transition state). Finally, the derivative 8, featuring modifications
both in the binding and effector regions, could be formed via a consecutive
photoreaction of 2a or 7.
The photoproduct 2a features a versatile MeO group in the amino acid moiety and
was of interest for further synthetic modifications. Hence, we focused on improving the
selectivity of the photoreaction. Interestingly, selective excitation of the tricarbonyl
chromophore of 1a in MeOH using filtered UV light (> 360 nm, aq. NaBr-Pb(NO3)2
filter) at 58 for 9 h resulted in total consumption of the starting material leading to 2a
(
75%) and 4a (15%) as the only products isolated (Table 1). Analysis of the crude
reaction mixture by HPLC also gave similar yields, and no other products could be
detected. The reaction could be scaled up conveniently (40 g), and the required 2a
could be freed of 4a by a short chromatographic filtration, followed by crystallization.

8
42
Helvetica Chimica Acta – Vol. 92 (2009)
Scheme 1. Irradiation of the Ascomycin Analogs 1a – 1d
Table 1. Yields of the Isolated Products [%] upon Irradiation of 1a under Different Conditions
Reaction conditions
2a
4a
7
8
9
1aa)
>
>
>
280 nm, MeOH, r.t., 2 h
22
–
75
10
10
15
40
30
–
4
–
–
–
11
–
19
26
–
280 nm, MeCN/H O 4 : 1, r.t., 2 h
2
360 nm, MeOH, 58, 9 h
a) Unchanged recovered starting material.

Helvetica Chimica Acta – Vol. 92 (2009)
843
It should be mentioned here that, in a semiquantitative study, O did not show any
2
effect on the reaction, thus indicating that the triplet state is not involved.
The reaction of 1a in CD CN using light of l > 360 nm was investigated by NMR
3
methods. Thus, irradiation of 1a in CD CN at 08 for 5 h led to a 3 :2 mixture consisting
3
of 9i, which is the C(10)¼O isomer of 9, and 1a, in addition to small amounts of 4a
(
Scheme 2). Irradiation for 10 h led to a mixture consisting mainly of 9i, and only minor
amounts of 1a and 4a. Heating this sample at 608 for 4 h led to the disappearance of 9i
and formation of the oxazolidinone 9. This confirms that the initial photoproduct is 9i
which is converted to 9 slowly upon warming. In a preparative experiment, 2 g of 1a in
MeCN (1 l) was irradiated for 16 h at 08, and an aliquot was purified by semipreparative
HPLC (cyclohexane/i-PrOH 85 :15, all operations carried out at 08) to give an enriched
1
13
sample of 9i for spectral analysis ( H, C, and C,H correlation). Characteristic of 9i are
13
the C-NMR signals at d 212.5 (C(22)), 210.6 (C(10)), 168.4 (C(1)), 165.4 (C(8)), 85.8
(
C(6) and cross-peak for HꢀC(6) at d 4.97 (ddd, J ¼ 9.7, 3.7, 1.9 Hz)), 81.7 (C(9) and
cross-peak for HꢀC(9) at d 5.17 (d, J ¼ 1.9 Hz)), 51.9 (C(2) and cross-peak for HꢀC(2)
at d 4.67 (d, J ¼ 6.0 Hz)), 38.6 (C(11)).
Scheme 2. Irradiation of 1a in CD CN
3
The primary photoproduct 9i could be formed through Z , which is formed either
1k
directly through excitation of the trioxo form of 1a, or indirectly through excitation of
a in the hemiketal form leading to Z , followed by its equilibration to Z . The
1
1h
1k
macrolide 1a does exist as a mixture of the interconvertible C(10)-hemiketal and trioxo
forms [5b]. The trioxo chromophore imparts bright yellow color to the compounds and

8
44
Helvetica Chimica Acta – Vol. 92 (2009)
is expected to absorb more strongly in the l > 360-nm region, and, therefore, it could be
the chromophore being excited leading to the Z . However, our photochemical amide
1k
cleavage reactions with a locked hemiketal structure at C(10) demonstrated that the
dicarbonyl (hemiketal) forms also show photoreactivity under the irradiation
conditions [7d]. Therefore, no conclusion could be drawn as to whether the dioxo or
trioxo forms alone, or both, are actually involved in these photoreactions. However, the
more flexible Z form appears to be necessary for attaining the required higher energy
1k
boat-like transition state involved in the cyclization leading to 9i. Similar intermediates
could be involved in the irradiation of 1a in MeOH; this was not investigated in detail.
Larger alkoxy groups could also be introduced at C(6) using the photochemical
method. Thus, irradiation of 1a in EtOH and PrOH led to the (6S)-ethoxy derivative 2e
(
59%) and the (6S)-propoxy derivative 2f (10%), respectively. Irradiations in higher
alcohols such as BuOH and i-BuOH led to complex mixtures which were not pursued
further. The lower yields of the alkoxy products with higher alcohols is probably due to
their lower nucleophilicity which allows unspecific side reactions to predominate.
The intact tricarbonyl moiety is crucial for the biological activity of 1a – 1c [5a].
Thus, 2a was oxidized [9] selectively employing 5 mol-% Cu(OAc) or cupric 2-
2
ethylhexanoate and 5 mol-% pyridine/O /4-ꢄ molecular sieves/CH Cl /16 h at room
2
2
2
temperature to give 6-b-methoxyascomycin (3a) in 96% yield (Scheme 3). A control
run employing 5 mol-% cupric 2-ethylhexanoate and omitting pyridine resulted in ca.
30% yield after 20 d, thus indicating the role of pyridine in the high catalytic turnover,
possibly through its ligation or depolymerization of the catalyst. To our knowledge, this
represents a very mild and efficient variation compared to the reported methods for the
oxidation of a-hydroxy ketones using copper salts [9]. The analogs 3b – 3f, comprising
modifications at C(21), C(33), and C(6), were prepared analogously through
irradiation of 1a – 1d in appropriate alcohol to give the photoproducts 2b – 2f (in
addition to 4b – 4d), followed by their oxidation.
2
The pipecolic acid ring in ascomycin 1a possesses a C conformation in solution,
5
solid state, and in its binding complex with macrophilin [6]. This feature, combined with
the tricarbonyl moiety, is expected to play a pivotal role in deciding the conformation of
the cyclic structure which is important for its biological action. We, therefore, envisaged
transforming the photoproducts 2a – 2d to the corresponding enamides 5a – 5d,
incorporating dehydropipecolic acid unit as a close bioisoster of pipecolic acid, with
potential for a good macrophilin-12 affinity. Thus, heating the derivatives 2a – 2d with
NH Cl in DMFat 788 under reduced pressure in a rotary evaporator, to ensure removal
4
of the MeOH formed in the reaction, afforded the enamides 5a – 5d (80 – 93%) [10].
Again, oxidation of the OH group at C(9) of 5a – 5d, as described above using
Cu(OAc) , afforded the analogs 6a – 6d (83 – 95%) featuring the 5,6-dehydropipecolic
2
acid unit. It is noteworthy that, in contrast to the facile elimination of MeOH from 2a –
2
d featuring C(9)ꢀOH, 3a, featuring C(9)¼O, was recovered (95%) unchanged under
identical reaction conditions. It appears that the highly electronegative C(9)¼O in 3a
disfavors the formation of a cationic site adjacent to the amide N-atom, hence
disfavoring the acid-catalyzed elimination reaction. It is also likely that stereoelectronic
effects resulting from a change in the conformation (2a vs. 3a), or the conformational
flexibility, also play a role in this reaction. The three-step strategy for the synthesis of
5,6-dehydroascomycin (6a) features high yields, inexpensive reagents, simple oper-

Helvetica Chimica Acta – Vol. 92 (2009)
845
Scheme 3. Transformation of the Photoproducts 2a – 2f to Derivatives with Modifications on the
Pipecolic Acid Moiety of 1a – 1d
ations, and is devoid of protecting groups on a highly functionalized and sensitive
molecule. It could successfully be employed for preparing 6a on a 100-g scale in the
laboratory and multi-kg scale in a pilot plant.
We further studied the applicability of the above chemistry to the proline analog 10
which is available as a side-product in the fermentation of ascomycin (Scheme 4).

8
46
Helvetica Chimica Acta – Vol. 92 (2009)
Irradiation of 10 in MeOH gave rise to both the (5R)- and (5S)-MeO derivatives 11
8%) and 12 (24%), respectively. The lower stereoselectivity observed in the photo-
(
reaction of 10, as opposed to the formation of a single isomer with the pipecolic acid
analogs 1a–1d, is probably due to the higher flexibility of the five-membered enaminium
intermediate Z (n ¼ 0; Scheme 1). It may be noted that b-attack by MeOH leading to 12
1
is still the more favored pathway. The photoproducts 11 and 12 could be transformed to
the methoxy-proline and dehydroproline analogs 14, 15, and 16 as described previously.
The yields in this series were not optimized because of the limited availability of 10.
Irradiation of 22-dihydro-33-TBS-ascomycin (17) in MeOH at 0 – 58 for 10 h using
Pyrex-filtered light resulted in the 6-MeO derivatives 17a (37%) and 17b (9%) as the
only products; the lower yields are due to losses during repeated chromatographic
purification. Clearly, the transformation of the trigonal C(22) to a tetrahedral center
made the macrolactam ring more flexible, thus reducing the stereoselectivity of the
attack by MeOH. Nevertheless, b-attack is still predominant over a-attack. Further-
more, due to the lack of allylic carbonyl chromophore, the photoreaction is selective
even with Pyrex-filtered light. Selective oxidation of 17a and 17b as before, followed by
desilylation with aqueous HCl/MeCN, afforded the analogs 18a and 18b, respectively, in
good yields. Selective hydrogenation (Pd/C/H /AcOEt) of the but-1-enyl double bond
2
II
in 7 gave 19 (75%). Cu xidation of the OH at C(9) of 8 afforded the analog 20 (35%).
2
.2. Transformation of the Photoproduct 2a to New Analogs. The photoproduct 2a
features a versatile hetero-acetal functionality which, as expected, could be hydrolyzed
under mild acidic conditions affording the aldehyde 21 in a 89% yield (Scheme 5). In
21, which is easily available in multigram quantities in two simple steps staring from 1a,
the original reactive C(9)¼O group is now formally protected as C(9)ꢀOH, and
C(22)¼O being a keto group, the high reactivity of the aldehyde group could be used
for selective modifications on the amino acid side chain without employing protecting
groups. To this end, 21 was oxidized selectively with aqueous NaClO to the acid 22 in a
2
good yield [11]. Selective oxidation of the OH group at C(9) of 22 with Cu(OAc) led
2
to the homoglutamic acid analog 24a in a 64% yield. Attempted esterification of 24a
with CH N resulted in a complex mixture because of the competing reaction of the
2
2
reactive C(9)¼O with CH N [12]. This problem, however, could be overcome by
2
2
reversing the reaction sequence. Thus, esterification of 22 with CH N gave 23 (85%),
2
2
which, after oxidation with Cu(OAc) , led to the homoglutamic acid ester analog 24b
2
(
72%). Wittig olefination of the aldehyde 21 with 1.4 equiv. Ph PCHCO Bn gave the
3 2
unsaturated benzyl ester 27a (49%), which, after Cu(OAc) oxidation of the OH group
2
at C(9), afforded 28a in a 75% yield. Catalytic hydrogenation of 28a resulted in
saturation of the enone C¼C bond and cleavage of the Bn group affording the acid
analog 29a. The methyl ester analog 29b was prepared similarly via 27b and 28b. The
analogs 24a, 24b, 28b, and 29a exist as a mixture of their six-membered and seven-
membered ring hemiketals (see Structural and Conformational Aspects).
The aldehyde 21 is also a good intermediate for the synthesis of bridged derivatives
with rigid conformations in the binding region. Thus, oxidation of 21 with Cu(OAc)2,
followed by careful neutral workup, afforded the analog 25 (90%; crude product). The
1
keto-aldehyde 25 exists as the free aldehyde as evidenced by H-NMR (CDCl ), and
3
not as aminal. However, 25 upon standing in CDCl for a few days, or on FC (SiO ), is
3
2
transformed to an isomer mixture 26a/26b. Alternatively, 3a, upon mild acidic

Helvetica Chimica Acta – Vol. 92 (2009)
847

8
48
Helvetica Chimica Acta – Vol. 92 (2009)
hydrolysis, followed by FC, afforded 26a (16%), 26b (25%), and unchanged 3a (48%).
1
The H-NMR spectrum of 26a (2 :1 mixture of C(9)-isomers) showed signals at d 4.90
(
d, J(2,3a) ¼ 6.3 Hz, HꢀC(2)) for the major component and at d 4.58 (d, J(2,3a) ¼ 7.5,
HꢀC(2)) for the minor one; at d 5.09 (dd, J(5,6) ¼ 3.5, 10, HꢀC(6)) for the major
component, and d 5.37 (dd, J(5,6) ¼ 4, 10, HꢀC(6)) for the minor one. These data
indicate the (6R)-configuration with chair conformation for the amino acid ring with
C(1) in the axial and C(6)ꢀO in the equatorial position for both components of 26a.
1
On the other hand, the H-NMR spectrum of 26b (one major isomer) showed signals at
d 3.82 (dd, J(2,3) ¼ 4.5, 10, HꢀC(2)) and d 4.85 (dd, J(5,6) ¼ 3.5, 10, HꢀC(6)),
indicating (6S)-configuration with the flipped chair with C(1) and C(6)ꢀO placed
equatorially.
2
.3. Transformation of the Enamide 6a to New Rigid Analogs. We further explored
the transformation of the enamide 6a to further rigid analogs through cyclopropanation
Scheme 6). However, the enamide C¼C bond was much less reactive than C(9)¼O.
Thus, 6a, upon treatment with CH N , gave a mixture of the spiro-oxiranes 33 (C(9)-
(
2
2

Helvetica Chimica Acta – Vol. 92 (2009)
849
Scheme 5. Transformation of the Photoproduct 2a to New Noncyclic Amino Acid Analogs

8
50
Helvetica Chimica Acta – Vol. 92 (2009)
isomers) in excellent yield, from which the individual isomers could be isolated in 19
and 41% yields [12]. On the other hand, treatment of 6a with a large excess of CH N in
2
2
the presence of [Cu(acac) ], followed by partial purification by HPLC, led to a 1:4 :3
2
mixture 30a/31a/unreacted 6a in a 48% yield. After repeated FC, pure samples of 30a
and 31a could be obtained and characterized. On the other hand, selective reduction of
the C(9)¼O of 6a with H S led to the deoxo derivative 32 in good yield [13].
2
Cyclopropanation of 32 with CH N as described above afforded the 9-deoxo-
2
2
derivatives 30b (9%) and 31b (15%) in low yields.
1
The derivatives 30a and 30b showed H-NMR signals at d 4.36 (dd, J ¼ 6.1, 2.7) and
4
3
.34 (dd, J ¼ 6.5, 1.6), respectively, which were assigned to HꢀC(2). In the derivatives
1a and 31b, HꢀC(2) appeared at d 4.41 (dd, J ¼ 8.6, 6.2) and d 4.47 (t, J ¼ 6.5),
respectively. We believe that 30a and 30b contain the cyclopropane ring on the b-face of
the pipecolic acid ring, and the resulting 1,3-diaxial interactions distort the half-chair,
thus leading to a small and a large coupling constant between HꢀC(2) and HꢀC(3),
respectively, as observed above. In support of this, for 30b a weak NOESY cross-peak
was observed between HꢀC(2) and HꢀC(6). On the other hand, 31a and 31b possess
the cyclopropane ring on the a-face of the pipecolic acid ring, hence, the ring can adopt
an undistorted half-chair conformation leading to two coupling constants of the same
order between HꢀC(2) and HꢀC(3). Because of the overlapping signals, it was not
possible to further confirm the configurations of C(5) and C(6).
The rigid derivative 34 featuring C(6), C(10) bridging was obtained through hydrolysis
1
of the enamide 32 with aqueous HCl in a 37% yield (Scheme 6). The H-NMR spectrum of
3
4 indicated a single isomer and showed signals at d 4.65 (dd, J¼ 7.1, 1.8, HꢀC(2)), 5.19
(dd, J¼ 9.0, 5.3, HꢀC(6)), 2.95 (d, J¼ 15.9, H’ꢀC(9)), 2.50 (d, J¼ 16.2, HꢀC(9)). In
addition, NOEs were observed from HꢀC(6) to HꢀC(9); HꢀC(9) to MeꢀC(11);
H’ꢀC(9) to HꢀC(12), and, H’ꢀC(9) to HꢀC(14). These data suggest a structure
featuring a chair conformation for the amino acid ring with C(1) in the axial position and
C(6)ꢀO in the equatorial position, a chair conformation for the pyran ring with the
substituents on C(11), C(13), and C(14) placed equatorially, a boat conformation for the
lactam ring, and an inversion of configuration at C(10) as depicted by formula 34.
2
.4. Synthesis of the Protected Derivatives 36 and 38 as Key Intermediates through
Photoreactions. The amino acid analogs described so far were synthesized in a few steps
from ascomycin (1a) without using any protecting groups and hence were also limited
in the extent of modification on the amino acid moiety. To establish new synthetic
strategies for synthesizing analogs with broader modifications on the amino acid,
appropriate protecting groups are required. Hence, we studied the photochemistry of
i
the Pr Si-protected derivative 35 [14] (Scheme 7), which was prepared from 1a through
2
monosilylation of the OH group at C(33), Evansꢀ reduction of C(22)¼O (1.1 equiv.
þ
ꢀ
Me N (OAc) HB , MeCN/AcOH 100 :35, ꢀ 5 to 38, 12 h, 70%), followed by silylation
4
3
i
(
3.0 equiv. (Cl Pr Si) O, 3.0 equiv. 1H-imidazole, DMF, r.t., 3 d, 70%).
2 2
Irradiation of 35 in MeOH for 8 h afforded the (6S)-9-hydroxy-6-methoxy
derivative 36 in good yield, which, after oxidation with Cu(OAc) , gave the (6S)-6-
2
methoxy-9-oxo derivative 37 in excellent yield. The derivative 37 still possesses a H-
atom at C(6) in the equatorial position and can, in principle, undergo a similar
photoreaction which would initially lead to the intermediate Z , the fate of which would
3
be interesting. In fact, irradiation of 37 in MeOH for 6 h, followed by chromatography,

Helvetica Chimica Acta – Vol. 92 (2009)
851
Scheme 6. Transformation of the Enamide 6a to New Rigid Analogs

8
52
Helvetica Chimica Acta – Vol. 92 (2009)
Scheme 7. Synthesis of the Key Intermediates 36 and 38 through Photoreactions

Helvetica Chimica Acta – Vol. 92 (2009)
853
led to the amide-leavage product 38 in a 44% yield. It appears that attack of MeOH at
C(6), or intramolecular nucleophilic attack by C(9)ꢀO at C(6), which have been the
observed pathways so far, are indeed not favored for Z because of the resulting steric
3
congestion. Instead, Z undergoes attack by MeOH on the activated electrophilic
3
amide C¼O group, leading to the amide cleavage product 38 [8]. Alternatively, the
zwitterion Z could undergo cleavage leading to a hydroxy ketene which is then trapped
3
by MeOH to give 38. Other derivatives of ascomycin (1a), especially the unprotected
3a, underwent a similar cleavage reaction [7d]. This represents a simple and mild
method of cleaving a strong amide bond in a highly sensitive molecule such as 3a not
containing any protecting groups.
2
.5. Synthesis of 6-Vinyl- and 5-Vinylproline Analogs of Ascomycin (1a) from 36.
The two photoproducts 36 and 38, bearing the appropriate protecting groups, are
available in multigram scale. We studied the scope of 36 for synthesizing the 6-vinyl
analog 46 (Scheme 8) as well as the 5-vinylproline analogs 52 and 53 (cf. Scheme 9).
t
Acidic hydrolysis of the hemiacetal 36, followed by reintroduction of BuMe Si
2
(
TBS) on C(33)ꢀOH afforded the aldehyde 39. Addition of vinylmagnesium bromide
to 39 gave the allyl alcohol 40. Selective oxidation of the OH group at C(9) of 40 with
Cu(OAc) , followed by selective activation of the secondary allylic alcohol as methyl
2
carbonate, afforded the secondary carbonate 41 in 42% yield over two steps. The key
step, namely the reinstallation of the C(e)ꢀN bond, was achieved through Pd-catalyzed
cyclization of 41 leading to the protected vinyl analog 42 in good yield [15]. An
alternative route was also explored for the transformation of 39 to 42. Thus, Wittig
olefination of 39 with Ph PCHCO Me afforded the corresponding unsaturated ester,
3
2
i
which, upon reduction with Bu AlH (DIBAH), led to the primary allylic alcohol 43.
2
Transformation of 43 to 44 was effected as described above through Cu(OAc)2-
catalyzed selective oxidation of the OH group at C(9), followed by selective activation
of the primary allylic alcohol as methyl carbonate derivative. Again, similar to the
transformation of 41 to 42, Pd-catalyzed cyclization of 44 afforded 42 in 63% yield. The
1
H-NMR spectra of the samples of 42, obtained from either 44 or 41, are identical to
each other, and indicated the presence of three inseparable components (C(6)-epimers
or conformers in 0.7:1:1.9 ratio). Desilylation of 42, followed by selective oxidation of
1
the OH group at C(22), afforded 6-vinylascomycin (46) in good yield. The H- and
1
3
C-NMR spectra of 46 showed it to consist of two major (ca. 1:1 ratio) and two minor
1
components. No coalescence of the signals was observed in the H-NMR in (D )DMSO
6
at 808, indicating it to be a mixture of C(6)-epimers, each as a pair of rotamers. Because
of overlapping of the signals, no information could be obtained on the configuration of
C(6) or conformation of the C(6)-epimers. The basic structure of 46, however, was
established using C,H correlation spectra.
It should be mentioned here that the derivative 39 could be transformed to 35
through reduction of the HꢀC(6)¼O with NaBH , followed by oxidation of OH at
4
C(9) with Cu(OAc) , and subsequently by ring-closure according to the Mitsumobu
2
method. This offers yet another route for synthesizing derivatives which was not
explored further.
The transformation of the aldehyde 39 to the 5-vinylproline analogs 52 and 53 is
depicted in Scheme 9. Addition of a C -unit to 39 through a Grignard reaction gave the
1
vinyl ether 47, which was first converted to the a,b-unsaturated aldehyde and then

8
54
Helvetica Chimica Acta – Vol. 92 (2009)

Helvetica Chimica Acta – Vol. 92 (2009)
855

8
56
Helvetica Chimica Acta – Vol. 92 (2009)

Helvetica Chimica Acta – Vol. 92 (2009)
857

8
58
Helvetica Chimica Acta – Vol. 92 (2009)
reduced with DIBAH to the primary allylic alcohol 48. Cu(OAc) Oxidation of the OH
2
group at C(9), followed by methoxycarbonylation, afforded the primary allyl carbonate
49. Pd-Catalyzed cyclization of 49, followed by chromatographic separation of the
epimers, led to the protected (5S)-vinylproline analog 50a (27%) and the (5R)-proline
analog 51a (30%). Noteworthy here are the longer reaction time and the higher
temperature required for the cyclization of 49, compared to those for 41 and 44.
Desilylation of 50a and 51a gave 50b and 51b, which, after Dess – Martin oxidation,
afforded 53 and 52, respectively.
1
The H-NMR spectra of the series of proline derivatives 50a, 50b, and 53 showed
signals at d 4.51 (dd, J ¼ 8.5, 2.1), 4.70 (dd, J ¼ 8.5, 2.3), and 4.46 (dd, J ¼ 8.6, 2.1),
respectively, which are assigned to HꢀC(2) based on C,H correlation spectra. These
splittings are similar to those in (5R)-methoxyproline analog 14, and indicate (5S)-
configuration for 50a, 50b, and 53, with C(1) placed pseudoaxially as shown in 53
1
(
Scheme 9). On the other hand, the H-NMR spectra of the other series 51a, 51b, and
5
2 displayed signals at d 4.49 (t, J ¼ 8.6), 4.55 (t, J ¼ 8.3), and 4.41 (t, J ¼ 8.4),
respectively, which are assigned to HꢀC(2). Again, these splittings are analogous to
those in (5S)-methoxyproline analog 15, indicating (5R)-configuration for 51a, 51b,
and 52, with C(1) oriented pseudoequatorially as shown in 52 (Scheme 9).
2
.6. Structural and Conformational Aspects. The conformation of the macrolides
1a – 1c in the binding domain containing the pipecolic acid moiety is expected to play a
crucial role in its binding to macrophilin (FKBP 12), and hence in its further
downstream activities. Here, we summarize some of the structural and likely
1
conformational features of the new derivatives deduced from the H-NMR coupling
constants in CDCl3.
2
.6.1. Derivatives Containing Modified Pipecolic Acids. The macrolide 1c exists as a
:1 mixture of s-cis and s-trans amide-bond rotamers; both rotamers possess chair
conformation for the pipecolic acid with the ester group containing C(1) placed axially
6]. 6-b-Methoxyascomycin (3a) consisted of two isomers in a 2 :1 ratio; the major
isomer showed coupling constants representative of a chair conformation (J(2,3) ¼ 1.8,
.3; J(5,6) ¼ 2.7, 2.7), whereas the minor isomer that of a boat-like conformation
2
[
5
(
J(2,3) ¼ 6.5, 7.1; J(5,6) ¼ 1.4, 4.0). Coalescence of the signals was observed upon
heating in DMSO at 380 K, indicating their interconvertibility. The 6b-EtO and the 6b-
PrO analogs, 3e and 3f, respectively, consisted of similar conformers, but the
distribution was slightly shifted towards the boat conformation (chair/boat 3 :2). This
deviation is most probably due to the 1,3-diaxial interactions between the ester group
containing C(1) and the RO groups at C(6).
3
The X-ray crystal structure ) of 6a is given in Fig. 1, and the side-on view of the
amino acid region in Fig. 2. Due to the presence of the C¼C bond, the conformation of
the piperidine ring differs significantly from the chair found in ascomycin (1a) [6e]. It is
best described by a least-squares plane through the atoms C(2), C(4), N(5), C(6), and
N(7) (maximum deviation from the plane 0.0777(9) ꢄ for N(7), and the distance
3)
CCDC 710864 contains the supplementary crystallographic data for this work. These data can be
obtained free of charge via http://www.ccdc.cam.ac.uk/data_request/cif (or from the Cambridge
Crystallographic Data Centre, 12 Union Road, Cambridge CB2 1EZ, UK; fax: þ 441223336033;
e-mail: deposit@ccdc.cam.ac.uk).

Helvetica Chimica Acta – Vol. 92 (2009)
859
Fig. 1. Structure of 6a in the crystal [16a]. Atomic displacement ellipsoids are drawn at the 50%
probability level, and H-atoms are drawn as spheres of arbitrary radius. For clarity, only atoms discussed
in the text are labeled. For the enantiomer shown the Flack ꢁ parameter refined to 0.00(8) [16d].
Fig. 2. Side-on view of the amino acid region of 6a [16a]. All atomic radii are arbitrary.
between this plane and C(3), which is 0.634(2) ꢄ). The geometry at the ring atom N(7)
is planar, the bond angles around this atom sum up to 359.97(19)8. The entire amide
moiety is also planar which is reflected in a torsion angle C(6)ꢀN(7)ꢀC(8)ꢀO(55) of

8
60
Helvetica Chimica Acta – Vol. 92 (2009)
0
8
.4(2)8. The angle between the planes of the C(8)¼O(55) and C(9)¼O(54) group is
9.90(1)8. The overall conformation of 6a, however, is close to that of ascomycin (1a).
The derivatives 30a and 30b bearing the cyclopropane ring on the b-face of the ring
adopt a half-chair conformation similar to 6a, but probably slightly distorted to
accommodate the 1,3-diaxial interaction. On the other hand, 31a and 31b featuring the
cyclopropane ring on the a-face of the ring adopt an undistorted half-chair
conformation with the ester group placed axially.
The ester group containing C(1) is oriented pseudoaxially in the a-methoxyproline
analog 14 and the a-vinyl-proline analog 53 (thus coming closer to the conformation of
1a), and pseudoequatorially in the b-methoxyproline analog 15 and the b-vinylproline
analog 52 (thus deviating considerably from the conformation of 1a). Both of the C(9)-
isomers of the oxazolidinone 9 showed a small (ca. 4 Hz) and a large (ca. 10 Hz) J
between HꢀC(6) and HꢀC(5); HꢀC(2) appeared as a doublet (J ꢂ 6). This points to
2
a C -conformation for the pyrrolidine ring with the ester group placed axially and the
5
C(6)ꢀO placed equatorially.
2
.6.2. Derivatives Containing Noncyclic Amino Acids. The tricarbonyl moiety of
ascomycin derivatives is known to exist predominantly as six-membered-ring hemi-
ketal; the amount of seven-membered-ring hemiketal is low and is dependent on the
structure and the solvent [6]. The derivatives 41 and 44 existed exclusively as the seven-
membered-ring hemiketals, and 49 existed as 95 :5 mixture of the seven-membered-
and six-membered-ring hemiketals. Freshly prepared CDCl solutions of the deriva-
3
tives 24a, 24b, 28b, and 29a contained the isomers in varying proportions, which
equilibrated (24 h) to mixtures containing the six-membered-ring as the major
component as well as a considerable amount of the seven-membered-ring hemiketal
as the minor component; other derivatives in this series were not investigated. All the
derivatives featuring cyclic amino acids discussed before existed mainly as six-
membered-ring hemiketals. The six- and seven-membered-ring structures are distin-
1
13
guished by H, C-long range-correlation spectroscopy (HMBC), generally optimized
for three-bond couplings. In the seven-membered-ring hemiketal, the MeꢀC(11) H-
atoms showed, as expected, a cross-peak to the ketone C-atom (C(10), at ca. 210 ppm),
whereas, in the case of the six-membered-ring hemiketal MeꢀC(11) H-atoms showed a
cross-peak to a ketal C-atom (C(10), at ca. 100 ppm). Moreover, due to the spatial
vicinity to the ketone C¼O group, the MeꢀC(11) H-atoms of the seven-membered-
ring hemiketal are shifted to lower field (ca. 1.2 ppm) compared to those of the six-
1
membered-ring isomer (ca. 1.0 ppm), allowing the deduction of the H chemical shifts
of MeꢀC(11) from these spectra. Hence, the only Me group resonating at ca. 1.2 ppm is
the MeꢀC(11) of the seven-membered-ring isomer, and all other Me groups of these
derivatives, except those located on the C¼C bonds, resonate in the region 1.0 –
0.8 ppm.
3
. Biological Activities. – Selected derivatives are grouped into structural types, and
their in vitro and in vivo activities along with those of the reference compounds 1c and
a are collected in Table 2. For easy understanding of the SARs, the structural
modifications are mnemonically abbreviated.
.1. In vitro Activities (MBA, RGA, and MLR). The first step in the mode of action
of these compounds is binding to the cytosolic receptor macrophilin-12 [6]; this is
1
3

Helvetica Chimica Acta – Vol. 92 (2009)
861
measured by macrophilin binding assay (MBA). The binary complex then inhibits the
Ca-dependent phosphatase calcineurin, an enzyme required for the dephosphorylation
of the cytosolic form of the nuclear factor of activated T cells (NF-AT), thus inhibiting
the release of cytokines such as IFN-g, IL-2, -4, -5, -10, and TNF-a, thus targeting T-cell
activation and proliferation. These events are assayed by IL-2 reporter gene assay
(
(
RGA) and mixed lymphocyte reaction (MLR). It should be noted that binding
MBA) is necessary but not sufficient for the elicitation of the T-cell modulatory
activities (RGA and MLR). Whereas the 6-MeO derivative 3a showed weak binding,
and correspondingly weak RGA and MLR activities, the 6-EtO and 6-PrO, 3e and 3f,
respectively, showed negligible activities. This is probably due to the change in the
conformation of these molecules as discussed above, or a misfit of the alkoxy groups at
C(6) into the binding pocket. It is known that the pipecolinyl ring of 1c is deeply
embedded in its complex with FKBP 12, and, hence, binding is sensitive to
modifications in this region [6g]. The 5,6-dehydroascomycin (6a) is very close to 1c
both in conformation and space-filling (X-ray structure) and, as expected, showed very
good activities; the corresponding 9-deoxo analog 32 showed somewhat weaker
activities, reflecting the importance of the C(9)¼O group for the activity. Interestingly,
the a-cyclopropano derivatives 31a and 31b, which are conformationally very close to
6
a but with the a-face of the ring filled with CH group, showed reasonably good
2
activities. On the other hand, their b-counterparts 30a and 30b were significantly less
active; the reason for this could lie in their deviation from the chair conformation and
filling of the space on the b-face of the ring (as in 3a, 3e, and 3f). Note that the 9-deoxo
analogs 32, 31b, and 30b are weaker in their activities than their corresponding C(9)¼O
analogs 6a, 31a, and 30a. The 6-vinyl derivative 46 is much less active, similar to 3e.
Neither of the isomers of the epoxide 33 showed any appreciable activity, pointing to
the importance of the C(9)¼O group for binding.
Among the proline derivatives, neither of the vinyl analogs 52 and 53 showed any
useful activity. Interestingly, the 5a-MeO analog 14, in spite of its apparent closeness in
space-filling to the biologically inactive 5a-vinyl analog 53, showed useful activities
(analogous to 5a-cyclopropano analog 31a). This is probably due to the higher
rotational freedom of the MeO group allowing it to orient itself better in the binding
pocket, compared to the spacially more rigid vinyl group. Unfortunately, no data are
available on the 5b-MeO analog 15. The dehydroproline 16 showed good activity,
analogously to the dehydropipecolic 6a.
The analogs 24a and 24b, featuring the noncyclic amino acid, were not active.
Because of the lack of the rigid cyclic amino acid, these molecules adopt a totally
different inactive conformation.
The rigid bicyclic structures 9 and 26a, in spite of their ideal piperidine chair
conformation with the ester group at C(2) in the axial position, and 26b, featuring the
piperidine ring in the chair conformation with the ester group at C(2) in the equatorial
position, are inactive. Apparently, some flexibility in this region is needed for good
binding.
The ring-contracted derivatives 7 and 19, which feature unaltered binding domains,
showed excellent binding to macrophilin. However, their weak activities in the RGA
and MLR indicate that these modifications in the effector domain do not favor further
interactions of the initially formed FKBP-binding complex with calcineurin A and

8
62
Helvetica Chimica Acta – Vol. 92 (2009)
Table 2. Biological Activities of the Derivatives of Ascomycin
PRIVATE} Abbreviated structural feature MBAa) RGAb)
rIC50 rIC50
MLR ) ) ACD
c d
ACD
{
e
f
g
g h
rIC50
mouse ) ) ) pig ) )
Reference compounds
1
1
c
a
Tacrolimus (FK 506)
Ascomycin
1.0
0.6
1.0
1.8
1.0 57*
2.1 44*
68*
58*
Derivatives containing modified pipecolic acid
3
3
3
6
3
3
3
3
3
4
3
3
a
e
f
a
6b-MeO-1a
6b-EtO-1a
6b-OPr-1a
5,6-Dehydro-1a
9-Deoxo-6a
5,6-a-Cyclopropyl-1a
9-Deoxo-31a
5,6-b-Cyclopropyl-1a
9-Deoxo-30a
6-Vinyl-1a
Minor epoxide-1a
Major epoxide-1a
70
14500
240
> 2600
> 15305
470
9 ns
36*
19 ns
n.a.
27*
51*
34*
2.0
15
8.2
2.0
47
15.0
40
41.0
24.0 53*
170 26 ns
22.0 41*
2
1a
1b
0a
0b
6
22*
535
49*
31*
4600
3
3
2000
250
24*
21*
Derivatives containing modified proline
14
16
52
53
5a-OMe-10
4,5-Dehydro-10
5b-Vinyl-10
5a-Vinyl-10
14
1.7
19.4
4.0
3 ns
5300
n.a.
Derivatives containing noncyclic amino acid
2
2
4a
4b
C-6-Acid-1a
C-6-Acid Me ester-1a
n.a.
n.a.
Derivatives containing bicyclic structures in the binding domain
9
2
2
Oxazolidinone-1a
6b-O-Ketal-1a
6a-O-Ketal-1a
290
6000
1900
3150
> 3300
> 3300
> 1000
n.a.
n.a
26 ns
3*
26*
6a
6b
Derivatives with modifications in the effector domain
7
1
2
But-1-enyl-1a
Butyl-1a
6b-MeO-7
1.7
0.7
1700
425
2200
2700
> 1000
> 1000
n.a.
24 ns
23 ns
8 ns
9
0
a) Macrophilin binding assay: IC50 value relative to 1c; IC (1c) 1.1 nm (S. D. 0.4, n ¼ 35). Test method:
50
1
c coupled to BSA is used to coat microtiter plate wells. Biotinylated recombinant human macrophilin-12
is allowed to bind to the immobilized 1c in the absence (as a control) and presence of a test sample.
Bound biotinylated macrophilin-12 is assessed by incubation with a streptavidin-alkaline phosphatase
conjugate, followed by incubation with p-nitrophenyl phosphate as a substrate. Readout is the OD at
4
05 nm. Binding of a test sample to the biotinylated macrophilin-12 results in a decrease in the amount of
macrophilin available for binding to the immobilized 1c and, thus, in a decrease in the OD 405 compared
to the control. A dilution series of the test sample (in duplicate) allows the calculation of the
concentration which results in 50% inhibition of the macrophilin/immobilized 1c-binding (IC ). The
50
inhibitory ability of the test sample is expressed as relative IC (rIC ), i.e., the ratio IC (test sample) to
50
50
50
b
IC (free 1c). ) Inhibition of production of IL2 reporter gene assay: IC50 value relative to 1c. IC (1c)
50
50
0
.27 nm (S.D. 0.09, n ¼ 3). Test method: the E. coli gene lacZ (reporter gene), which encodes the enzyme
b-galactosidase, has been placed under transcriptional control of the human IL-2 promoter and stably
transfected into the human T cell JURKAT. The IL-2 promoter-driven b-galactosidase expression, which

Helvetica Chimica Acta – Vol. 92 (2009)
863
calcineurin B, thus not resulting in any T-cell modulatory activities. The derivative 20
featuring modifications both in the binding and effector domains showed very weak
binding, and hence also very weak RGA and MLR activities.
3
.2. In vivo Activities. Selected derivatives were tested topically in ACD (allergic
contact dermatitis) mouse model as well as a few selected ones in ACD pig model. Of
the two ACD models, the mouse model is less stringent than the pig model. Since the
murine skin is more permeable than the human skin, the activities in mice are only
indicative and not predictive for the human situation. The data marked as ꢅnsꢀ (not
statistically significant) can be considered as no activity. The pig model is more
predictive, because pig skin and human skin are closely related in their drug
Table 2 (cont.)
is quantified by measuring the fluorescence of its cleaved substrate, correlates with IL-2 expression and is
thus a direct readout for IL-2 gene transcription. The results are expressed as relative IC (rIC ), i.e., the
50
50
ratio of the IC50 (nm) of the test compound to the IC50 (nm) of 1c. c) Inhibition of mixed lymphocyte
5
reaction: IC50 value relative to 1c. IC (1c) 0.15 nm (S.D. 0.04, n ¼ 6). Test method: 1 ꢁ 10 spleen cells of
50
each BALB/c and CBA mice are cultivated in flat-bottom microtiter plates for 4 d in serum-free tissue
culture medium. [ H]Thymidine (1 mCi/well) is then added. The cells are harvested after another 16-h
cultivation period and the [ H]thymidine incorporation into the DNA is subsequently measured. The
extent of [ H]thymidine incorporation reflects the number of cells which were dividing during the last
3
3
3
1
6 h of cultivation and is thus representative for cell proliferation. The MLR is performed in the absence
or presence of a serially diluted compound and 1c as a standard. The concentration required for 50%
inhibition of cell proliferation (IC ) is calculated; results are expressed as relative IC (rIC ), i.e., the
5
0
50
50
d
e
ratio of the IC50(test sample) to IC50(1c). ) n.a.: Not active (> 8000). ) Inhibition [%] of allergic contact
dermatitis in mouse after topical application of 0.01% drug soln. Test method: female NMRI mice were
sensitized on the abdomen with oxazolone (2% in acetone) and, after 7 d, challenged with 2% oxazolone
(
for topical testing) or 0.5% oxazolone (for systemic testing) on the inner surface of the right ears (8
animals per group in each experiment). The unchallenged left ears served as internal controls, and
evaluation was made from the difference in auricular weights as a measure of inflammatory swelling 24 h
after the challenge. Topical treatment with ethanolic solns. (10 ml) of the test compound or the reference
compound 1c, and dexamethasone was performed once, 30 min after elicitation of ACD. Groups of
controls were treated similarly with the vehicle topically. Auricular weight differences in test and control
groups were evaluated with one way ANOVA followed by Dunnet test (parametric) or by Kruskal –
Wallis and Dunnꢀs test (non-parametric) using the SigmaStatꢃ program. Significance was taken as p <
0
.05. The activity was calculated as the % inhibition of inflammatory swelling (weight increase) relative
f
g
to vehicle treated animals. ) ns: Not statistically significant. ) *: Statistically significant (p < 0.05).
h
)
Inhibition [%] of allergic contact dermatitis in domestic pig after topical application of 0.13% drug
soln. Test method: young domestic hybrid pigs (Landrasse ꢁ Deutsches Edelschwein) were sensitized
with 400 ml 2,4-dinitrofluorobenzene (10% in DMSO/acetone/olive oil) applied to four areas at both
pinnae and groins. Challenge reactions were elicited 12 d later with 15 ml DNFB (1%) at test sites
arranged at craniocaudal lines on the dorsolateral back. The test sites were treated twice with 20 ml soln.
of test compounds or vehicle, 30 min and 6 h post challenge. For testing formulated compounds, ca.
1
00 mg of creames (in doses) were applied. One day after challenge, each test site was visually evaluated
for intensity, and expanse of erythema and induration (gross lesion). Each sign was scored on a scale
from 0 (normal) to 4 (severe) allowing a combined maximal score of 12 per designated site. The
percentage inhibition was calculated using the equation: % Inhibition ¼ {(M ꢀ M ) ꢁ 100}/MVS; M:
VS
DS
mean; VS: vehicle-treated sites; DS: drug-treated sites. For statistical analysis of the differences between
scores of contralateral test sites, Studentꢀs t-test (two tailed paired) was applied. The significance level was
set at p < 0.05.

8
64
Helvetica Chimica Acta – Vol. 92 (2009)
penetration properties. The derivatives 6a and 31a which showed good in vitro activities
also showed good activities in these two animal models.
The derivative 6a with its established practical synthesis and good biological
activities was evaluated further, and the data are summarized here. In the murine ACD
model after subcutaneous or oral applications, 6a showed good activity. In the pig
model, upon topical application, 6a was comparable to the ultrapotent corticosteroid
clobetasol-17-propionate and fluticasone propionate. Furthermore, 6a was less potent
by a factor of ꢃ 10 than FK 506 in the localized graft vs. host model in rats when given
subcutaneously. Compound 6a was also by a factor of ꢃ 15 less potent than FK 506 in
preventing the allogeneic kidney transplant rejection in rats after oral application.
These results indicate a good potential for 6a in treating ACD and other T cell-
mediated dermatitis disorders with substantially lower risk of systemic immunosup-
pression.
4
. Conclusions. – A novel photoreaction on ascomycin (1a) leading to methox-
ylation in the e-position of the pipecolic acid moiety was described. Starting from this
methoxy photoproduct, several derivatives featuring novel modifications on the amino
acid, which are not easily accessible through routine methods, could be synthesized in
fewer steps. Using the photoreaction, key intermediates with potential for broader
modifications in the binding domain were synthesized, and their utility was exemplified
through synthesis of vinylpipecolic acid and vinylproline analogs. An interesting
photochemical cleavage of the amide bond in the derivatives of ascomycin (1a) was
discussed. The structural and conformational features of the derivatives together with
the X-ray crystal structure of the biologically most active 5,6-dehydroascomycin (6a)
were presented, and their biological activities were discussed.
Experimental Part
General. Solvents (THF, DMF, CH Cl , toluene) were dried over 4-ꢄ molecular sieves. MeOH was
2
2
distilled over Na. Flash chromatography (FC): silica gel 60 (40 – 63 mm). SiO -2.5% NaHCO was
2
3
prepared by mixing 400 g of SiO2 with 1 l of aq. 1% NaHCO , evaporating to dryness on a rotary
3
evaporator under vacuum, followed by drying under vacuum at 808 for 3 h. M.p.: uncorrected. NMR
1
Spectra: either on a Bruker AMX 500 or a Bruker DRX-500 spectrometer (500.13 MHz for H and
1
3
1
25.77 MHz for C); unless stated otherwise, the solvent was CDCl ; chemical shifts in ppm, referenced
3
1
13
to residual solvent signals (7.26 for H, 77.0 for C in CDCl , 3.31 and 49.0 in CD OD). The assignments
were based on diagnostically relevant NMR experiments such as homonuclear H, H-COSY, TOCSY,
NOESY, heteronuclear H, C-correlation, HSQC, and long-range HMBC spectra, or, comparison with
analogous compounds and are highly reliable. In case of isomers/rotamers, where possible, the data of the
minor isomer are marked with an asterisk (*). The system as indicated in the atom-numbering scheme
3
3
1
1
1
13
(
Fig. 3) with slight adaptation for other derivatives is used for all the compounds for assigning the signals.
HR-MS: 9.4T APEX-III FT-MS (Bruker Daltonics).
CA Index Name of 1a: 15,19-Epoxy-3H-pyrido[2,1-c][1,4]oxaazacyclotricosine-1,7,20,21(4H,23H)-
tetrone, 8-ethyl-5,6,8,11,12,13,14,15,16,17,18,19,24,25,26,26a-hexadecahydro-5,19-dihydroxy-3-[(1E)-2-
[
(1R,3R,4R)-4-hydroxy-3-methoxycyclohexyl]-1-methylethenyl]-14,16-dimethoxy-4,10,12,18-tetrameth-
yl-, (3S,4R,5S,8R,9E,12S,14S,15R,16S,18R,19R,26aS)-; other names: ascomycin; changchuanmycin; FK
5
20; FR 520; FR 900520; immunomycin; L 683590.
General Procedure A (GP A): A soln. of the compound (5 – 10 g) in a dry solvent (1 l) in a Pyrex tube
was degassed with He and irradiated externally using a medium-pressure Hg lamp (Hanovia, TQ-150)

Helvetica Chimica Acta – Vol. 92 (2009)
865
Fig. 3. Atom-numbering scheme
fitted with a Pyrex filter. The jacket containing the light source was placed in a large H O bath (58 or r.t.),
2
and the reaction vessels were placed all around in the same bath.
The reaction was monitored using TLC or anal. HPLC (Polygosyl 10 CN column, cyclohexane/i-
PrOH 9 :1 or 85 :15, isocratic). After completion of the reaction (2 – 9 h), the solvent was removed, and
the residue was purified by FC on SiO or prep HPLC (Polygosyl CN column, cyclohexane/i-PrOH 9 :1
2
or 85 :15; isocratic).
General Procedure B (GP B). Similar to GP A, but employing a TQ 718 lamp fitted with a Duran
filter in a well-type reactor. The well (which usually contains the reactant) was filled with an aq. filter
soln. (transparent to l > 360 nm) prepared by dissolving 650 g of NaBr· 2 H O and 3.0 g of Pb(NO ) in
2
3 2
doubly distilled H O and made up to 1 l. The well-reactor containing the light source and the filter soln.
2
was placed in a large H O bath, and the reaction vessels were placed all around in the same bath.
2
General Procedure C (GP C). A mixture of the sample (0.2 g) and 1 – 2 equiv. of Cu(OAc) or
2
cupric-2-ethylhexanoate in CH Cl or THF (10 ml) was stirred under O balloon at r.t. or reflux temp.
2
2
2
until the reaction was complete (1 – 3 d; TLC or anal. HPLC). The solvent was removed, and the residue
was partitioned between aq. NaHCO and AcOEt. The org. phase was separated and washed repeatedly
3
with aq. NaHCO , until the washings were colorless, and finally with brine. The org. extract was dried,
3
concentrated, and the residue was purified by FC or prep. HPLC.
General Procedure D (GP D). A mixture of the sample (10 g), Cu(OAc) , or cupric 2-ethyl-
2
2 2 2
hexanoate (5 – 100 mol-%), py (10 – 20 mol-%), and 4-ꢄ mol. sieves was stirred in CH Cl under O
balloon at r.t. until completion (1 – 2 d). The mixture was passed through a bed of Celite, and the filtrate
was concentrated. The residue was processed as in GP C.
General Procedure E (GP E). A mixture of the sample (1 g) and dry NH Cl (0.5 – 1.0 equiv.) in DMF
4
(
100 ml) was reacted in a rotary evaporator under reduced pressure at 788 such that the DMF slowly
distilled off. After completion of the reaction (4 h), the remaining DMF was removed under high
vacuum, and the residue was dried and partitioned between aq. NaHCO and AcOEt. The org. layer was
3
separated, and the aq. layer was re-extracted with AcOEt. The combined org. extracts were washed with
brine, dried, and concentrated. The residue was purified by FC or prep. HPLC.
Compounds 2a, 4a, 7, and 8. According to GP A, 1a (0.2 g, 0.252 mmol) in MeOH (50 ml) was
irradiated at r.t. for 2 h. 25 Such batches (total 5 g, 6.31 mmol) were combined and purified by prep.
HPLC to give 2a (1.09 g, 22%), 7 (1.983 g, 40%), 8 (0.206 g, 4%), 4a (0.482 g, 10%), and unchanged
starting material 1a (0.955 g, 19%). Alternatively, according to GP B, 1a (10 g, 12.63 mmol) in MeOH
(
2
1 l) was irradiated at 58 for 9 h, and 1/10 of the resulting residue was purified as described above to give
a (0.78 g, 75%) and 4a (0.145 g, 15%); the crude photolysate from this step could also be taken without
purification to the next step.
Data of 2a. Crystallized from AcOEt. White solid. M.p. 130 – 1378; 1H-NMR (ca. 3 :1 mixture
isomers/rotamers): major isomer: 5.84 (br. s, HꢀC(6)); 5.45 (d, J ¼ 5.5, HꢀC(2)); 5.22 (d, J ¼ 8.9,
HꢀC(20/29)); 5.07 (br. s, HꢀC(26)); 4.88 (d, J ¼ 10.4, HꢀC(20/29)); 4.49 (s, HOꢀC(10)); 4.42 (s,
HꢀC(9), HOꢀC(9)); 3.78 (dd, J ¼ 9.6, 1.2, HꢀC(14)); 3.60 (m, HꢀC(15), HꢀC(24)); 3.40 (s,

8
66
Helvetica Chimica Acta – Vol. 92 (2009)
MeOꢀC(32)); 3.40 (m, HꢀC(13/33)); 3.38 (s, MeOꢀC(13)); 3.28 (s, MeOꢀC(15)); 3.37 (m, HꢀC(13),
HꢀC(33)); 3.20 (m, HꢀC(21)); 3.10 (s, MeOꢀC(6)); 3.00 (ddd, J ¼ 11.2, 7.8, 4.2, HꢀC(32)); 2.84 (br. s,
OH); 2.81 (dd, J ¼ 18.1, 3.5, 1 HꢀC(23)); 2.39 (dd, J ¼ 18.1, 5.8, H ꢀC(23)); 2.4 – 0.8 (overl. ms); minor
b
isomer: 6.14 (s, HOꢀC(10)); 5.55 (br. s, HꢀC(6)); 5.04 (d, J ¼ 9.1, HꢀC(20/29)); 5.02 (br. s, HꢀC(26));
4
.82 (d, J ¼ 9.7, HꢀC(20/29)); 4.68 (dd, J ¼ 6.6, 4.4, HꢀC(2)); 4.47 (br. d, HꢀC(9)/HOꢀC(9)); 4.37 (br.
d, J ¼ 8.6, HꢀC(9)/HOꢀC(9)); 3.90 (m, HꢀC(24)); 3.85 (dd, J ¼ 9.5, 2.1, HꢀC(14)); 3.75 (d, J ¼ 2.3,
OH)); 3.60 (m, HꢀC(15)); 3.40 (s, MeOꢀC(32)); 3.35 (s, MeOꢀC(13)); 3.27 (s, MeOꢀC(15)); 3.45 –
3
2
1
7
3
.25 (m, HꢀC(13), HꢀC(33), HꢀC(21)); 3.00 (ddd, J ¼ 11.2, 7.8, 4.2, HꢀC(32)); 2.59 (dd, J ¼ 16.6,
13
.0, H ꢀC(23)); 2.4 – 0.7 (overl. ms). C-NMR: major isomer: 214.0 (C(22)); 173.3 (C(8)); 170.7 (C(1));
a
39.7 (C(19)); 131.1 (C(27)); 129.7 (C(29)); 123.9 (C(20)); 97.0 (C(10)); 84.2 (C(32)); 80.5 (C(6)); 77.8;
5.2; 74.8; 73.5; 73.4; 72.6; 71.8; 57.1; 56.5; 56.3; 54.2; 52.1; 49.3; 43.7; 40.4; 36.1; 35.0; 34.8; 33.9; 33.5;
1.3; 30.5; 30.2; 26.1; 25.7; 22.6; 20.8; 16.0; 15.4; 15.0; 13.9; 11.3; 11.1; minor isomer: 213.2 (C(22)); 174.3
(
C(8)); 168.4 (C(1)); 140.6 (C(19)); 129.0 (C(29)); 122.7 (C(20)); 99.8 (C(10)); 84.3 (C(6)); 76.9; 73.8;
7
2
0.8; 70.1; 69.5; 57.6; 56.0; 55.8; 55.0; 52.3; 48.7; 44.1; 41.1; 36.6; 34.93; 34.87; 32.9; 32.2; 30.3; 29.9; 25.4;
5.0; 23.4; 18.0; 16.21; 16.16; 15.4; 14.2; 11.5; 9.9. Anal calc. for C H N O : C 64.13, H 8.93, N 1.70;
44
73
1
13
þ
found: C 63.75, H 8.47, N 1.53. HR-MS: 846.4977 ([M þ Na] ; calc. 846.4980).
1
Data of 4a. White foam, H-NMR (ca. 12 :5 mixture of rotamers): major rotamer: 8.08 (s, HC¼O);
5
.35 (d, J ¼ 9.0, HꢀC(20/29)); 5.21 (d, J ¼ 8.4, HꢀC(26)); 5.13 (d, J ¼ 5.2, HꢀC(2)); 4.94 (br. dd, J ¼ 9.9,
1
.2, HꢀC(20/29)); 4.07 (dd, J ¼ 7.8, 1.6); 3.97 (m); 3.70 (m); 3.60 – 3.30 (overl. ms); 3.22 (m, HꢀC(21));
3
0
.00 (m, HꢀC(32)); 2.92 (br. s, OH); 2.64 (m, 2 HꢀC(23)); 2.80 (br. s, OH); 2.48 (m, HꢀC(11)); 2.40 –
.80 (overl. ms); minor rotamer: 8.03 (s, HC¼O); 5.35 (d, J ¼ 9.0, HꢀC(20/29)); 5.23 (d, J ¼ 8.5,
HꢀC(26)); 4.94 (br. dd, J ¼ 9.9, 1.2, HꢀC(20/29)); 4.32 (br. d); 4.26 (d, J ¼ 4.9, HꢀC(2)); 4.07 (dd, J ¼
7.8, 1.6); 3.97 (m); 3.70 (m); 3.60 – 3.30 (overl. ms); 3.22 (m, HꢀC(21)); 3.00 (m, HꢀC(32)); 2.80 (m,
13
HꢀC(6)); 2.64 (m, 2 H, HꢀC(23)); 2.48 (m, HꢀC(11)); 2.40 – 0.80 (overl. ms). C-NMR of both
rotamers: 212.1; 212.0*; 173.6; 169.8*; 169.5; 162.3*; 161.8; 138.42*; 138.35; 134.1*; 133.8; 130.8; 130.7*;
1
5
3
24.2; 84.11; 84.07*; 82.9*; 82.52; 82.47*; 82.3; 77.3; 77.2*; 73.6; 73.5*; 73.4; 67.0; 66.9*; 57.5; 57.0*; 56.7;
6.3; 54.4; 50.7; 48.5*; 48.4; 45.9*; 45.8; 44.0; 38.8; 38.7*; 38.2*; 34.9; 34.8; 34.7*; 34.4; 33.3; 32.2; 31.2;
0.2; 27.2*; 26.9; 26.4; 25.5; 24.2; 24.1; 21.5; 21.4*; 19.6*; 19.5; 17.0; 16.4; 12.22; 12.18*; 11.6; 8.90; 8.86*.
þ
HR-MS: 786.4765 ([M þ Na] ; calc. 786.4728).
1
Data of 7. White foam. H-NMR: 5.63 (dt, J ¼ 15.7, 6.4, HꢀC(21)); 5.46 (br. d, J ¼ 3.2, HꢀC(26));
5
1
.32 (dt, J ¼ 15.7, 1.6, HꢀC(20)); 4.94 (dm, J ¼ 9.0, HꢀC(29)); 4.56 (d, J ¼ 1.7, HOꢀC(10)); 4.38 (dm, J ¼
4.0, 1 HꢀC(6)); 4.35 (dd, J ¼ 6.0, 2.0, HꢀC(2)); 3.81 (m, HꢀC(24)); 3.58 (d, J ¼ 1.9, HOꢀC(24)); 3.57
(
dd, J ¼ 9.6, 1.3, HꢀC(14)); 3.49 (ddd, J ¼ 10.3, 3.4, 1.3, HꢀC(15)); 3.41 (s, MeO); 3.38 (m, HꢀC(13);
HꢀC(33)); 3.35 (s, MeO); 3.30 (s, MeO); 3.05 – 2.95 (m, 1 HꢀC(6), 1 HꢀC(23), HꢀC(32)); 2.71 (br. s,
HOꢀC(33)); 2.40 – 0.70 (overl. ms); 1.64 (d, J ¼ 1.4, MeꢀC(27)); 1.15 (s, MeꢀC(19)). Anal calc. for
þ
C
43
H
69
N
1
O
12: C 65.21, H 8.78, N 1.77; found: C 64.9, H 8.6, N 1.7. HR-MS: 812.4926 ([M þ Na] ; calc.
8
12.4925).
1
Data of 8. White foam. H-NMR: 5.83 (t, J ¼ 2.3, HꢀC(6)); 5.48 (dt, J ¼ 15.6, 6.4, HꢀC(21)); 5.42
(
dd, J ¼ 5.7, 2.1, HꢀC(2)); 5.20 (m, HꢀC(20), HꢀC(26), HꢀC(29)); 4.49 (d, J ¼ 1.6, OH); 4.36 (d, J ¼
7
.8, HOꢀC(9)/HꢀC(9)); 4.19 (d, J ¼ 7.8, HOꢀC(9)/HꢀC(9)); 4.07 (s, OH); 3.72 (dd, J ¼ 9.5, 1.4,
HꢀC(14)); 3.65 (br. t, J ¼ 9.7); 3.52 (m); 3.41 (s, MeO); 3.35 – 3.20 (overl. ms); 3.33 (s, MeO); 3.27 (s,
MeO); 3.11 (s, MeO); 3.01 (m, HꢀC(32)); 2.78 (br. s, OH); 2.64 (d, J ¼ 17.8, H ꢀC(23)); 2.40 – 0.80
a
13
(
(
(
overl. ms); 1.62 (s, MeꢀC(27)); 1.35 (s, MeꢀC(19)). C-NMR: 216.1 (C(22)); 173.8 (C(1/8)); 171.0
C(1/8)); 134.2 (C(20)); 131.7 (C(21)); 129.8 (C(27)); 129.0 (C(29)); 122.4 (C(20)); 97.3 (C(10)); 84.3
C(32)); 80.4 (C(6)); 76.8; 75.5; 74.8; 73.5; 73.3; 72.4; 71.0; 56.9; 56.6; 56.5; 56.2; 52.5; 52.3; 45.7; 40.4;
3
8.7; 35.9; 35.0; 34.94; 34.90; 33.6; 31.3; 30.5; 30.1; 25.7; 25.3; 25.0; 20.0; 18.9; 16.4; 15.5; 14.3; 13.5; 11.3.
Anal calc. for C H N O : C 64.13, H 8.93, N 1.70; found: C 64.8, H 9.2, N 1.6.
44
73
1
13
Compounds 2b and 4b. Prepared analogously to 2a starting from 1b according to GP B.
1
Data of 2b. White foam. Yield: 66%. H-NMR (ca. 4 :1 mixture of rotamers): major rotamer: 5.83 (t,
J ¼ 2.6, HꢀC(6)); 5.45 (dd, J ¼ 5.4, 1.9, HꢀC(2)); 5.29 (d, J ¼ 8.7, HꢀC(20/29)); 5.07 (br. s, HꢀC(26));
4
3
.89 (d, J ¼ 10.5, HꢀC(20/29)); 4.55 (m, HꢀC(33)); 4.49 (s, HOꢀC(10)); 4.42 (s, HꢀC(9), HOꢀC(9));
.78 (dd, J ¼ 9.5, 1.4, HꢀC(14)); 3.60 (m, HꢀC(15), HꢀC(24)); 3.50 – 3.10 (overl. ms); 3.10 (s,
MeOꢀC(6)); 2.80 (dd, J ¼ 18.1, 3.5, 1 HꢀC(23)); 2.40 (dd, J ¼ 18.1, 5.9, 1 HꢀC(23)); 2.40 – 0.80 (overl.

Helvetica Chimica Acta – Vol. 92 (2009)
867
ms); minor rotamer: 6.14 (s, HOꢀC(10)); 5.53 (br. s, HꢀC(6)); 5.11 (d, J ¼ 8.8, HꢀC(20/29)); 5.04 (br. s,
HꢀC(26)); 4.83 (d, J ¼ 9.7, HꢀC(20/29)); 4.69 (dd, J ¼ 6.7, 4.2, HꢀC(2)); 4.55 (m, HꢀC(33)); 4.38 (br. s,
HꢀC(9)/HOꢀC(9)); 4.26 (br. s, HꢀC(9)/HOꢀC(9)); 3.90 (m, HꢀC(24)); 3.85 (dd, J ¼ 9.5, 2.1,
HꢀC(14)); 3.73 (s, OH); 3.60 (m, HꢀC(15)); 3.50 – 3.10 (overl. ms); 2.59 (dd, J ¼ 16.6, 2.0, 1 HꢀC(23));
13
2
.40 – 0.70 (overl. ms). C-NMR: major rotamer: 214.0 (C(22)); 173.3 (C(8)); 170.7 (C(1)); 139.7
(
C(19)); 131.0 (C(27)); 129.4 (C(29)); 123.8 (C(20)); 97.0 (C(10)); 80.5 (C(6)); 79.3; 77.6; 75.2; 74.7; 73.4;
7
2
1
2.5; 71.8; 59.3; 57.1; 56.29; 56.27; 55.6; 54.2; 52.1; 49.2; 43.7; 40.2; 36.0; 34.9; 33.8; 33.4; 31.9; 31.8; 30.5;
6.1; 25.7; 25.1; 22.6; 20.1; 16.0; 15.3; 15.0; 14.8; 14.0; 11.3; 11.0. C-NMR: minor rotamer: 213.2 (C(22));
74.2 (C(8)); 168.4 (C(1)); 140.5 (C(19)); 131.2 (C(27)); 129.0 (C(29)); 122.7 (C(20)); 99.7 (C(10)); 84.4
13
(
3
8
C(6)); 76.9; 76.7; 73.4; 70.8; 70.1; 69.4; 59.3; 57.5; 56.0; 55.8; 54.9; 52.3; 48.6; 43.9; 41.0; 36.5; 34.8; 32.9;
þ
2.2; 29.8; 26.9; 25.4; 25.2; 23.4; 18.0; 16.2; 16.1; 15.0; 14.1; 11.5; 9.8. HR-MS: 864.4649 ([M þ Na] ; calc.
64.4641).
1
Data of 4b. White foam. Yield: 10%. H-NMR (ca. 5 :2 mixture of rotamers): major rotamer: 8.08 (s,
HC¼O); 5.39 (d, J ¼ 8.7, HꢀC(20/29)); 5.21 (d, J ¼ 8.1, HꢀC(26)); 5.13 (d, J ¼ 5.4, HꢀC(2)); 4.94 (d, J ¼
9
.9, HꢀC(20/29)); 4.53 (br. d, J ¼ 2.7, HꢀC(33)); 4.06 (dd, J ¼ 7.8, 1.6); 3.98 (m); 3.69 (m); 3.60 – 3.30
(
overl. ms); 3.22 (m, HꢀC(21)); 2.88 (br. s, OH); 2.64 (m, 2 HꢀC(23)); 2.48 (m, HꢀC(11)); 2.40 – 0.80
(
4
overl. ms); minor rotamer: 8.02 (s, HC¼O); 5.39 (d, J ¼ 8.7, HꢀC(20/29)); 5.23 (d, J ¼ 7.6, HꢀC(26));
.94 (d, J ¼ 9.9, HꢀC(20/29)); 4.53 (br. d, J ¼ 2.7, HꢀC(33)); 4.32 (br. d, J ¼ 12.9, H ꢀC(6)); 4.25 (d, J ¼
a
5
.2, HꢀC(2)); 4.06 (dd, J ¼ 7.8, 1.6); 3.98 (m); 3.69 (m); 3.60 – 3.30 (overl. ms); 3.22 (m, HꢀC(21)); 2.80
13
(
m, H ꢀC(6)); 2.64 (m, 2 HꢀC(23)); 2.48 (m, HꢀC(11)); 2.40 – 0.80 (overl. ms). C-NMR: both
b
rotamers: 212.1; 212.0*; 173.6; 169.8*; 169.5; 162.3*; 161.9; 138.5*; 138.4; 133.7*; 133.6; 131.2; 131.0*;
1
4
2
24.2; 82.6; 82.5; 82.0; 79.4; 79.3*; 77.2; 73.64; 73.56*; 67.1; 59.31; 59.26*; 57.6; 57.0*; 56.8; 55.8; 54.5; 50.7;
8.5*; 48.4; 45.9*; 45.7; 44.0; 39.0; 38.9*; 38.2*; 35.0; 34.9; 34.8*; 33.3; 32.2; 31.8; 31.5; 27.3*; 26.9; 26.5;
5.5; 25.3; 24.3; 24.2*; 21.43; 21.39*; 19.6*; 19.5; 17.0; 16.4; 12.4; 11.6; 8.94; 8.87*.
Compound 2c. Prepared analogously to 2a starting from 1c according to GP B. White foam. Yield:
0%. Anal. HPLC of the crude photolysate indicated a yield of ca. 70%. No attempt was made to isolate
3
4
1
c. H-NMR (ca. 5 :2 mixture of rotamers): major rotamer: 5.83 (br. t, J ¼ 2.8, HꢀC(6)); 5.70 (m,
HꢀC(37)); 5.45 (dd, J ¼ 5.5, 1.9, HꢀC(2)); 5.22 (d, J ¼ 8.9, HꢀC(20/29)); 5.08 (br. s, HꢀC(26)); 5.03
(
dm, J ¼ 17.0, 1 HꢀC(38)); 4.97 (dm, J ¼ 11.2, 1 HꢀC(38)); 4.91 (d, J ¼ 10.5, HꢀC(20/29)); 4.50 (s,
HOꢀC(10)); 4.42, 4.40 (2d as ABq, HꢀC(9), HOꢀC(9)); 3.78 (dd, J ¼ 9.6, 1.4, HꢀC(14)); 3.60 (m,
HꢀC(15), HꢀC(24)); 3.50 – 3.20 (overl. ms); 3.10 (s, MeOꢀC(6)); 3.00 (m, HꢀC(32)); 2.84 (br. s, OH);
2
5
.81 (dd, J ¼ 18.1, 3.5, 1 HꢀC(23)); 2.55 – 0.70 (overl. ms); minor rotamer: 6.12 (d, J ¼ 1.6, HOꢀC(10));
.70 (m, HꢀC(37)); 5.56 (br. s, HꢀC(6)); 5.05 – 4.95 (m, HꢀC(20/29), HꢀC(26), HꢀC(38)); 4.86 (d,
J ¼ 9.7, HꢀC(20/29)); 4.68 (dd, J ¼ 6.8, 4.4, HꢀC(2)); 4.58 (br. s, HOꢀC(9)); 4.37 (br. s, HꢀC(9)); 3.91
(
m, HꢀC(24)); 3.84 (dd, J ¼ 9.5, 2.2, HꢀC(14)); 3.65 (s, OH); 3.60 (m, HꢀC(15)); 3.50 – 3.20 (overl.
ms); 3.00 (m, HꢀC(32)); 2.84 (br. s, OH); 2.59 (dd, J ¼ 16.6, 2.0, 1 HꢀC(23)); 2.55 – 0.70 (overl. ms).
1
3
C-NMR: major rotamer: 213.2 (C(22)); 173.3 (C(8)); 170.7 (C(1)); 139.9 (C(19)); 135.7 (C(37)); 131.2
C(27)); 129.7 (C(29)); 123.3 (C(20)); 116.5 (C(38)); 97.0 (C(10)); 84.3 (C(32)); 80.5 (C(6)); 77.7
C(26)); 75.3 (C(9)); 74.8 (C(15)); 73.5 (C(33)); 73.4 (C(13)); 72.6 (C(14)); 71.8 (C(24)); 57.1 (MeO);
6.5 (MeO); 56.3 (2 MeO); 52.6 (C(21)); 52.1 (C(2)); 49.2 (C(18)); 43.6 (C(23)); 40.4 (C(25)); 36.0
C(11)); 35.0; 34.8; 34.0; 33.8; 33.5; 31.4; 30.5; 30.2; 26.2; 25.7; 25.1; 20.1; 16.1; 15.4; 14.9; 13.9; 11.1; minor
rotamer: 212.3 (C(22)); 174.4 (C(8)); 168.4 (C(1)); 140.7 (C(19)); 135.6 (C(37)); 131.2 (C(27)); 129.1
(
(
5
(
(
(
(
C(29)); 122.1 (C(20)); 116.4 (C(38)); 99.8 (C(10)); 84.4 (C(6)); 84.3 (C(32)); 77.0 (C(15/26)); 76.9
C(15/26)); 73.9 (C(13)); 73.5 (C(33)); 70.8 (C(14)); 70.1 (C(24)); 69.6 (C(9)); 57.6; 56.0; 55.8; 52.9
C(21)); 52.3 (C(2)); 48.6 (C(18)); 44.1 (C(23)); 41.2 (C(25); 36.6; 35.0; 34.9; 34.6; 32.9; 32.2; 30.3; 29.9;
þ
2
5.4; 25.0; 23.4; 18.0; 16.20; 16.17; 15.1; 14.2; 9.9. HR-MS: 858.4985 ([M þ Na] ; calc. 858.4980).
Compound 2d. Prepared analogously to 2a starting from 1d according to GP B. 1d (1.0 g, 1.29 mmol)
gave crude 2d (1.04 g, 100%) which was taken to the next step without purification.
Compound 2e. Prepared analogously to 2a starting from 1a/EtOH according to GP B. White foam.
1
Yield: 34% H-NMR (ca. 8 :3 mixture of rotamers): major rotamer: 5.92 (t, J ¼ 2.7, HꢀC(6)); 5.45 (dd,
J ¼ 5.5, 1.9, HꢀC(2)); 5.28 (d, J ¼ 2.0, HꢀC(26)); 5.25 (d, J ¼ 9.0, HꢀC(20/29)); 4.86 (d, J ¼ 9.2,
HꢀC(20/29)); 4.52 (s, HOꢀC(10)); 4.47 (d, J ¼ 7.4, HꢀC(9)/HOꢀC(9)); 4.34 (d, J ¼ 7.4, HꢀC(9)/
HOꢀC(9)); 3.76 (dd, J ¼ 9.6, 1.4, HꢀC(14)); 3.72 (m, HꢀC(24)); 3.58 (m, HꢀC(15)); 3.45 – 3.15 (overl.

8
68
Helvetica Chimica Acta – Vol. 92 (2009)
ms); 2.99 (m, HꢀC(32)); 2.76 (dd, J ¼ 18.0, 3.1, 1 HꢀC(23)); 2.74 (br. s, OH); 2.50 – 0.70 (overl. ms);
minor rotamer: 6.19 (s, HOꢀC(10)); 5.60 (br. s, HꢀC(6)); 5.07 (m, HꢀC(20/29), HꢀC(26)); 4.85 (d,
HꢀC(20/29)); 4.65 (dd, J ¼ 6.6, 4.4, HꢀC(2)); 4.32 (br. s, HꢀC(9)); 4.07 (br. s, HOꢀC(9)); 3.89 (m,
HꢀC(24)); 3.85 (dd, J ¼ 9.5, 2.2, HꢀC(14)); 3.65 (s, OH); 3.50 – 3.20 (overl. ms); 2.99 (m, HꢀC(32));
1
3
2
2
.74 (br. s, OH); 2.58 (dd, J ¼ 16.5, 2.0, 1 HꢀC(23)); 2.50 – 0.70 (overl. ms). C-NMR: major rotamer:
14.0 (C(22)); 173.1 (C(8)); 170.3 (C(1)); 139.7 (C(19)); 132.6 (C(27)); 130.1 (C(29)); 123.9 (C(20)); 96.9
(
C(10)); 84.2 (C(32)); 78.2 (C(6)); 76.9 (C(26)); 75.3 (C(9)); 74.9 (C(15)); 73.52; 73.46; 72.5 (C(14));
7
4
2
1
7
1.1 (C(24)); 63.7 (MeCH O); 57.2 (MeO)); 56.4 (MeO); 56.3 (MeO); 54.0 (C(2/21)); 52.2(C(2/21));
2
9.1 (C(18)); 43.2 (C(23)); 41.6 (C(25)); 36.1 (C(11)); 34.9; 34.7; 33.8; 33.0; 31.3; 30.9; 30.3; 26.1; 25.8;
2.9; 20.1; 16.0; 15.4; 14.9; 14.7; 13.9; 11.4; 11.1; minor rotamer: 212.4 (C(22)); 174.3 (C(8)); 168.5 (C(1));
40.6 (C(19)); 132.2 (C(27)); 129.2 (C(29)); 122.6 (C(20)); 99.8 (C(10)); 84.2 (C(32)); 82.3 (C(6)); 76.8;
3.9 (C(13)); 73.5 (C(33)); 70.8 (C(14)); 70.2 (C(24)); 69.3 (C(9)); 63.4 (MeCH O); 57.6; 56.4; 56.0; 54.9
2
(
2
C(21)); 52.4 (C(2)); 48.7 (C(18)); 43.8 (C(23)); 41.8 (C(25)); 36.6; 34.8; 32.9; 32.3; 30.5; 30.2; 25.5; 25.2;
þ
3.7; 18.0; 16.23; 16.18; 15.2; 14.8; 14.2; 11.5; 10.0. HR-MS: 860.5138 ([M þ Na] ; calc. 860.5136).
Compound 2f. Prepared analogously to 2a starting from 1a/PrOH according to GP B. White foam.
1
Yield: 9%. H-NMR (ca. 3 :1 mixture of rotamers): major rotamer: 5.93 (t, J ¼ 2.6, HꢀC(6)); 5.47 (dd,
J ¼ 5.5, 1.9, HꢀC(2)); 5.28 (d, J ¼ 1.6, HꢀC(26)); 5.25 (d, J ¼ 8.9, HꢀC(20/29)); 4.88 (d, J ¼ 10.4,
HꢀC(20/29)); 4.53 (s, HOꢀC(10)); 4.49 (d, J ¼ 7.2, HꢀC(9)/HOꢀC(9)); 4.35 (d, J ¼ 7.2, HꢀC(9)/
HOꢀC(9)); 3.78 (dd, J ¼ 9.6, 1.3, HꢀC(14)); 3.74 (m, HꢀC(24)); 3.59 (m, HꢀC(15), OH); 3.45 – 3.15
(
overl. ms); 3.00 (m, HꢀC(32)); 2.77 (dd, J ¼ 18.0, 3.0, H ꢀC(23)); 2.70 (br. s, OH); 2.50 – 0.70 (overl.
a
ms); minor rotamer: 6.19 (s, HOꢀC(10)); 5.62 (br. s, HꢀC(6)); 5.07 (m, HꢀC(20/29), HꢀC(26)); 4.88
(
3
d, J ꢂ 10, HꢀC(20/29)); 4.67 (dd, J ¼ 6.8, 4.3, HꢀC(2)); 4.35 (br. s, HꢀC(9)); 4.06 (br. s, HOꢀC(9));
.90 (m, HꢀC(24)); 3.86 (dd, J ¼ 9.5, 2.2, HꢀC(14)); 3.69 (m, MeCH O); 3.50 – 3.20 (overl. ms); 2.99 (m,
2
HꢀC(32)); 2.70 (br. s, OH); 2.59 (dd, J ¼ 16.5, 2.0, H ꢀC(23)); 2.50 – 0.70 (overl. ms). HR-MS: 874.5303
a
þ
(
(
[M þ Na] ; calc. 874.5293).
Compound 3a. According to GP D, 2a (12.4 g, 15.05 mmol), CH Cl (108 ml), Cu(OAc) · H O
2
2
2
2
138 mg, 0.7 mmol), pyridine (123 ml, 1.5 mmol), 4-ꢄ mol. sieves (11.5 g), r.t., 18 h. FC (SiO -
2
1
2
.5%NaHCO ) gave 3a (11.9 g, 96%). White foam. H-NMR (DMSO; ca. 2 :1 mixture of rotamers):
3
major rotamer: 5.68 (t, J ¼ 2.9, HꢀC(6)); 5.20 (d, J ¼ 8.9, HꢀC(20/29)); 5.10 (br. s, HꢀC(26)); 4.91 (d,
J ¼ 10.1, HꢀC(20/29)); 4.47 (br. d, J ¼ 5.2, HꢀC(2)); 4.20 (d, J ¼ 1.2, HOꢀC(10)); 3.73 (m, HꢀC(24));
3
.72 (dd, J ¼ 9.6, 1.2, HꢀC(14)); 3.65 – 3.05 (overl. ms); 3.02 (m, HꢀC(32)); 2.75 – 2.60 (m, 1 HꢀC(23),
OH); 2.40 – 0.80 (overl. ms); minor rotamer: 5.31 (s, HOꢀC(10)); 5.18 (s, HꢀC(26)); 5.06 (m, HꢀC(6));
5
.02 (d, J ¼ 9.0, HꢀC(20/29)); 5.00 (d, J ¼ 10.0, HꢀC(20/29)); 4.67 (t, J ¼ 6.5, HꢀC(2)); 3.98 (m,
HꢀC(24)); 3.92 (dd, J ¼ 9.5, 2.0, HꢀC(14)); 3.65 – 3.05 (overl. ms); 3.02 (m, HꢀC(32)); 2.75 – 2.60 (m,
13
1
1
7
5
3
1
HꢀC(23), OH); 2.40 – 0.80 (overl. ms). C-NMR (both rotamers): 214.1*; 213.6; 195.4; 190.4*; 169.1;
67.7*; 167.3; 166.9*; 139.6*; 138.8; 131.2; 131.0*; 129.9; 129.3*; 123.7; 123.1*; 98.5*; 96.9; 84.3; 83.4*; 79.4;
8.0; 77.2*; 76.0*; 74.7; 73.62; 73.56; 73.47*; 72.7; 71.6*; 71.3; 69.3*; 57.3; 56.9; 56.5; 56.4; 56.3; 55.7*; 54.8*;
4.5; 54.2; 53.0*; 49.0*; 48.8; 44.1*; 43.6; 40.8; 40.7*; 36.2*; 35.1; 34.9*; 34.8; 33.2*; 32.94; 32.90*; 32.8;
1.31; 31.27*; 30.4; 30.2; 29.4*; 26.4; 25.7; 25.5*; 24.4; 23.65*; 23.59; 20.2; 18.8; 16.2; 16.1; 16.0; 15.9; 15.2*;
þ
5.0; 14.2*; 14.0; 11.6*; 11.0; 10.7; 10.0*. HR-MS: 844.4827 ([M þ Na] ; calc. 844.4823).
Compound 3b. Prepared analogously to 3a starting from 2b according to GP D. White foam. Yield:
1
1
0%. H-NMR (ca. 2 :1 mixture of rotamers): major rotamer: 5.68 (t, J ¼ 3.0, HꢀC(6)); 5.27 (d, J ¼ 8.6,
HꢀC(20/29)); 5.09 (br. s, HꢀC(26)); 4.92 (d, J ¼ 10.0, HꢀC(20/29)); 4.55 (m, HꢀC(33)); 4.47 (br. d, J ¼
5
3
.5, HꢀC(2)); 4.18 (d, J ¼ 1.2, HOꢀC(10)); 3.73 (m, HꢀC(24)); 3.71 (dd, J ¼ 9.5, 1.1, HꢀC(14)); 3.65 –
.05 (overl. ms); 2.67 (dd, J ¼ 15.5, 3.3, 1 HꢀC(23)); 2.40 – 0.80 (overl. ms); minor rotamer: 5.36 (s,
HOꢀC(10)); 5.19 (s, HꢀC(26)); 5.07 (overl. d, HꢀC(20/29)); 5.06 (br. s, HꢀC(6)); 5.02 (d, J ¼ 9.2,
HꢀC(20/29)); 4.65 (t, J ¼ 6.6, HꢀC(2)); 4.55 (m, HꢀC(33)); 3.98 (m, HꢀC(24)); 3.92 (dd, J ¼ 9.5, 2.4,
HꢀC(14)); 3.65 – 3.05 (overl. ms); 2.70 (br. d, J ꢂ 17, 1 HꢀC(23)); 2.40 – 0.80 (overl. ms). HR-MS:
þ
8
62.4487 ([M þ Na] ; calc. 862.4484).
Compound 3c. Prepared analogously to 3a starting from 2c according to GP D. White foam. Yield:
1
6
8
3%. H-NMR (ca. 2 :1 mixture of rotamers): major rotamer: 5.70 (m, HꢀC(6), HꢀC(37)); 5.21 (d, J ¼
.9, HꢀC(20/29)); 5.11 (br. s, HꢀC(26)); 5.08 – 4.96 (m, HꢀC(38)); 4.93 (d, J ¼ 10.1, HꢀC(20/29)); 4.45
(dd, J ¼ 5.2, 1.8, HꢀC(2)); 4.20 (d, J ¼ 1.2, HOꢀC(10)); 3.75 (m, HꢀC(24)); 3.72 (dd, J ¼ 9.6, 1.3,