The structure elucidation and total synthesis of β-lipomycin

Article abstract of DOI:10.1002/anie.201402259

Here we describe the synthesis of β-lipomycin, a secondary metabolite isolated from the fermentation broth of Corallococcus coralloides. The synthesis relies on the structural assignment made by a statistical method, the so-called profile hidden Markov model. Using this protocol, not only the configuration of the secondary alcohol, but also of the adjacent methyl branch could be deduced. The synthesis therefore not only provides access to this natural product but also confirms the validity of this approach for configurational assignment at methyl branches of modular polyketides.

Full text of DOI:10.1002/anie.201402259

Angewandte
Chemie
DOI: 10.1002/anie.201402259
Natural Product Structures
The Structure Elucidation and Total Synthesis of b-Lipomycin**
Olaf Hartmann and Markus Kalesse*
Abstract: Here we describe the synthesis of b-lipomycin,
a secondary metabolite isolated from the fermentation broth of
Corallococcus coralloides. The synthesis relies on the structural
assignment made by a statistical method, the so-called profile
hidden Markov model. Using this protocol, not only the
configuration of the secondary alcohol, but also of the adjacent
methyl branch could be deduced. The synthesis therefore not
only provides access to this natural product but also confirms
the validity of this approach for configurational assignment at
methyl branches of modular polyketides.
T
he lipomycins were first isolated in 1972 by the group of
Axel Zeeck from a strain of Streptomyces aureofaciens.[1] The
orange–red compounds were shown to inhibit the growth of
several Gram-positive bacteria (with MICs ranging from 0.3–
3 mgmLꢀ1) and to have no effect on fungi and yeasts. The
name of these natural products arose from the fact that their
antibiotic activity was antagonized by naturally occurring
lipids such as lecithin and other sterols. Whereas the sugar of
a-lipomycin (1) was quickly identified as d-digitoxose, the
structure elucidation of the aglycon b-lipomycin (2) took
a further year and was accomplished by chemical degradation,
mass spectrometry, and NMR spectroscopy.[2] However, apart
from the (S)-glutamate stereocenter (C5’) within the acyltetr-
amic acid[3] ring system, the configuration of the C12 and C13
stereocenters remained unknown (Figure 1).
Figure 1. Structures of the lipomycins as published by Schabacher/
Zeeck and Steyn (numbering according to Schabacher/Zeeck).[2]
Gene cluster analysis has contributed to the structure
elucidation of many polyketides when classical methods like
chemical degradation, NMR, MS, and X-ray methods were
not sufficient.[8] McDaniel[9] and Caffrey[10] independently
found conserved amino acid residues within the ketoreduc-
tases (KR) and were able to predict the configuration of the
formed alcohols. Thus, a pivotal LDD motif at position 93
(Caffreyꢀs numbering) leads to the formation of a d-config-
ured alcohol, whereas a tryptophan (W) in position 141 gives
an l alcohol. Leadlay et al.[11] published a gene analysis of
enoylreductases (ER) that revealed the configuration of
isolated methyl groups with the help of a pivotal tyrosin (Y)
residue in position 52. With these two simple tools in hand,
the prediction of several configurations within polyketides
can be done with high accuracy even though there are rare
examples for which this analysis does not provide the correct
assignment. Additionally, a reliable tool to predict the
configuration of methyl branches following secondary alco-
hols, as observed for lipomycin, is missing as well. In the
course of our work on the synthesis of complex natural
products, we have developed a profile hidden Markov model
(HMM)[12,13] that allows the accurate assignment of secondary
alcohols as well as methyl branches following l-configured
alcohols (Figure 2). The advantage of using the hidden
Markov model for predicting the configuration of modular
polyketides is the fact that this model uses all amino acids (M)
for the configurational prediction and not only selected
residues. Additionally, it accounts for insert (I) and deleted
(D) states which correspond to additional or missing amino
acids. That increases the reliability of the prediction and
makes previous sequence alignment superfluous. The term
Additionally, although Schabacher and Zeeck proposed
the N1’-methyl-4’-hydroxy-D3’ pyrrolidin-2’-one form (as in 1)
for the cyclic ring system, Steyn and co-workers predicted the
lipomcyins to rather exist as a tautomer with a Z-configured
exocyclic double bond (C1–C3’ within 2) after NMR and X-
ray studies of related compounds.[4]
Structural relationships exists with the polyenoyl tetramic
acids oleficin (3)[5] and altamycin (4),[6] both of which were
also isolated from Actinomycetales and differ only by the
number of the double bonds in their aliphatic tail. In contrast
to almost all other naturally occurring tetramic acids, the
slime mold pigment fuligorubin (5)[7] shows an R configu-
ration at the glutamate stereocenter.
[*] Dr. O. Hartmann, Prof. Dr. M. Kalesse
Institute for Organic Chemistry and Centre of Biomolecular Drug
Research (BMWZ), Leibniz Universitꢀt Hannover
Schneiderberg 1B, 30167 Hannover (Germany)
and
Helmholtz Centre for Infection Research (HZI)
Inhoffenstrasse 7, 38124 Braunschweig (Germany)
E-mail: Markus.Kalesse@oci.uni-hannover.de
[**] We thank Dr. M. Hofferberth and Prof. Dr. R. Brꢁckner for spectra of
their authentic and synthetic lipomycin samples.
Supporting information for this article is available on the WWW
under http://dx.doi.org/10.1002/anie.201402259.
Angew. Chem. Int. Ed. 2014, 53, 7335 –7338
ꢀ 2014 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
7335

.
Angewandte
Communications
Figure 2. Architecture of the profile HMM with five match states. Blue
squares indicate match states, pink diamonds insert states, and
orange circles delete states. B and E refer to the beginning and end of
an amino acid sequence, respectively.
“profile” refers to different emission and transition propabil-
ities of individual positions. Moreover, the absolute value of
the ScoreDiff value derived from the HMM provides
a measure of reliability of the predicted configuration.
When the lipomycin biosynthetic gene cluster was pub-
lished in 2006,[14] Bechthold et al. found W141 and the
absence of the LDD motif within the KR to be responsible
for the formation of the C12 and C13 stereocenters. There-
fore, they predicted the C13 alcohol to be l configured, but
they were unable to assign the C12 methyl configuration.
Using our profile HMM approach all amino acids of the
specific subset of each ketoreductase were taken into account
and they were allocated to one of the two families of
secondary alcohols. Additionally, we used Viterby scores[15] to
quantify the reliability of our predictions. In the case of the
C13 alcohol of the lipomycins we found a negative ScoreDiff
value (ꢀ53.96) that is consistent with the l configuration and
therefore supports the configurational assignment made by
the Bechthold group. For the C12 methyl group we found
a ScoreDiff of ꢀ39.37 and thust identified the stereocenter to
be l configured as well. This led to the structure of b-
lipomycin (2) as depicted in Figure 3.
Scheme 1. Retrosynthetic analysis of b-lipomycin (2).
condensation, as it bears substituents on the C5’ and on the
N1’ positions, both of which are important for high yields in
these type of reactions.[18] The Dieckmann process was
supposed to build up the five-membered tetramic acid ring
system in analogy to the related step in the synthesis of
fuligorubin (5) reported by Ley and co-workers[19] (Figure 1).
The C8’-carboxylic acid was to be protected as the methyl
ester in order to circumvent difficulties that might arise when
with more hindered esters such as tert-butyl.[20] The desired
cyclization giving the C4’-methyl ester rather than the C8’-
methyl ester was to be controlled by the ring size (five-
membered ring vs. seven-membered ring). Connections
between fragments 7 and 8,[21] as well as between 8 and 9
were achieved by HWE reactions as Ley had reported low
yields and poor E/Z selectivities during corresponding Wittig
transformations.[22] Compound 9 on the other hand should be
easily accessible by means of a Kobayashi vinyloguos
Mukaiyama aldol reaction (VMAR).[23]
In the synthetic direction, the western fragment 6 is easily
accessible from commercially available H-Glu(OMe)-OMe
(10) (Scheme 2). Boc protection, N-methylation, and subse-
quent Boc deprotection quickly furnished the desired pre-
cursor 6 in high yields. Compounds 7 and 8 were synthesized
from Meldrumꢀs acid (11) and ethyl 4-bromocrotonate (12),
respectively, according to literature procedures.[16,21]
The synthesis of eastern fragment
9 started from
Figure 3. The proposed absolute configuration of b-lipomycin (2).
commercially available 2-methyl-pent-2-enoic acid (13)
(Scheme 3). Introduction of the d-valine-derived (R)-Evans
auxiliary via the pivaloic anhydride and subsequent conver-
sion to the TBS ketene acetal 15 utilizing sodium hexame-
thyldisilylamide (NaHMDS) and TBS chloride was achieved
in good yields. The following Kobayashi VMAR reaction was
the key step in the synthesis of 9. With one equivalent of
titanium tetrachloride and freshly distilled isobutyraldehyde
the desired product 16 was obtained as a single diastereomer
in 76% yield after reaction at ꢀ408C for 16 h. TBS
protection, followed by reductive cleavage of the auxiliary
and allylic oxidation with activated manganese dioxide
furnished a,b-unsaturated aldehyde 7 in a good yield of
70% over three steps.
Here we report on the total synthesis of b-lipomycin to
provide support for the correct structural assignment made by
the profile hidden Markov model. In a retrosynthetic analysis
b-lipomycin (2) can be divided into the four segments 6, 7, 8,
and 9 (Scheme 1).
Although the lipomycins were reported to be more stable
than other polyenes[1] to light and air, we planned to build up
the pentane system in the endgame of the synthesis. Our plan
was to join the glutamate-derived fragment 6 and the
bifunctional fragment 7[16] through a silver(I)-promoted
coupling reaction[17] to give the corresponding amide. This
amide should be a suitable substrate for a Dieckmann
7336
www.angewandte.org
ꢀ 2014 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Angew. Chem. Int. Ed. 2014, 53, 7335 –7338

Angewandte
Chemie
Scheme 2. Synthesis of fragments 6, 7, and 8. a) NaHCO3, Boc2O,
100%; b) NaH, MeI, 75%; c) TFA, 100%; d) bromoacetyl bromide,
pyridine; e) tert-butyl thiol, 41% over 2 steps; f) Na, diethyl phosphite,
=
55%; g) DiBAl-H, 80%, h) MnO2, 77%, i) Ph3P CHCO2Et, toluene,
35%; j) P(OEt)3, neat, 75%. Boc=tert-butyloxycarbonyl; TFA=tri-
fluoroacetic acid; DiBAl-H=diisobutylaluminum hydride.
Scheme 4. Completion of the total synthesis of b-lipomycin (2). a) Ag-
(O2CCF3), NEt3, 71%; b) LiHMDS, 82%; c) DiBAl-H, 99%; d) MnO2,
95%, e) NaH, 19, then 21, 66%; f) LiOH, MeOH, H2O, 92%; g) 25%
aq HF, CH3CN, 85%.
dioxide gave the a,b-unsaturated aldehyde 21. In contrast to
20, compound 21 and its corresponding alcohol were found to
be significantly more unstable.
Subsequent to the Dieckmann/HWE one-pot protocol
(vide supra) the tetramic acid moiety including pentaene 22
(if the enol double bond is not regarded as an olefin) was
subjected to a smoothly proceeding saponification using
aqueous lithium hydroxide. Removal of the remaining TBS
group at C13 in the resulting acid was the remaining
transformation before completion of the total synthesis.
However, this step proved to be troublesome and needed
intensive optimization. First attempts with HF-pyridine with
or without additional pyridine[25] and with tris(dimethylami-
no)ꢀsulfonium difluorotrimethylsilicate) (TAS-F, with or
without additional water) did not lead to useful conversion.
With TBAF, although deprotections in similar systems were
reported at room temperature,[26] only little conversion was
observed even at elevated temperatures[27] (608C). As we
were concerned about racemization at the C5’ stereocenter
under these conditions we switched to aqueous hydrogen
fluoride in acetonitrile.[28] Gratifyingly, these conditions
cleanly led to formation of the desired natural product b-
lipomycin (2) in good yield (85%).
Scheme 3. Synthesis of aldehyde 9. a) (R)-Evans auxiliary, PivCl, NEt3,
LiCl, 84%; b) NaHMDS, TBSCl, 96%; c) TiCl4, isobutyraldehyde, 76%,
d.r.>95:5; d) TBSOTf, 2,6-lutidine, 95%; e) LiBH4, MeOH, 77%;
f) MnO2, 95%. (R)-Evans auxiliary=(R)-4-isopropyloxazolidin-2-one;
PivCl=trimethylacetyl chloride, HMDS=hexamethyldisilylamide;
TBS=tert-butyldimethylsilyl; OTf=trifluoromethylsulfonyl.
With all the fragments in hand, we coupled glutamate-
derived 6 and fragment 7 under mild conditions using silver
trifluoroacetate at room temperature in 71% yield
(Scheme 4). We had intended to cyclize the resulting product
19 in a Dieckmann process, but under several sets of
conditions we were unable to isolate the C3’–C4’ cyclized
product, probably due to high water solubility even under
acidic conditions. Gratifyingly, without isolation of such an
intermediate, we could obtain the Dieckmann-cyclized and
HWE-coupled product 22 (66%, E/Z > 95:5 as determined by
1H NMR analysis) in a one-pot process when aldehyde 21 was
added.[24] The required aldehyde 21 was obtained by HWE
reaction between phosphonate 8 and VMAR-derived alde-
hyde 9 using LiHMDS as a base to yield tetraene 20 (E/Z >
95:5). We found this compound to be relatively stable, even
on silica gel. DiBAl-H reduction yielded the corresponding
allylic alcohol and subsequent reoxidation with manganese
The spectroscopic data of the synthetic material match
that of both nature-identical (obtained by acidic hydrolysis of
a-lipomycin) and synthetic material provided by Hofferberth
and Brꢁckner, who confirmed the configuration of the
lipomycins by an independent synthesis (see the preceding
Communication).[29] Moreover, the fact that the optical
rotation is close to the value reported by Zeeck and co-
20
workers[1] (synthetic ½aꢁD ¼ꢀ165.0, c = 0.02, MeOH; authentic
20
½aꢁD ¼ꢀ176, c = 0.09, MeOH) indicates that no racemization
at the C5’ position had occurred.
Angew. Chem. Int. Ed. 2014, 53, 7335 –7338
ꢀ 2014 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.angewandte.org
7337

.
Angewandte
Communications
[12] Our methodology also addresses alcohol stereocenters. A.
Kitsche, M. Kalesse, ChemBioChem 2013, 14, 851 – 861. For
applications to uncharacterized sequences a web interface is
Synthetic access to the polyenoyl tetramic acid b-lipo-
mycin (2) was achieved in a longest linear sequence of 12
steps starting from commercially available substances with
a yield of 17.0%. This synthesis and its preceding structural
assignment solely rely on the statistical analysis of the pivotal
ketoreductase and confirm the validity and practicability of
this analysis.
provided
under
http://glimmer.rstudio.com/kitsche/
profileHMM/ where the ScoreDiff value is calculated according
to the profile HMMs.
[13] For details on the theory of profile HMMs and their application
to biological sequence analysis see: a) S. R. Eddy, Bioinformatics
1998, 14, 755 – 763; b) N. E. Davey, R. J. Edwards, D. C. Shields,
BMC Bioinf. 2010, 11, 14.
[14] C. Bihlmaier, E. Welle, C. Hofmann, K. Welzel, A. Vente, E.
Breitling, M. Mꢁller, S. Glaser, A. Bechthold, Antimicrob.
Agents Chemother. 2006, 50, 2113 – 2121.
[15] The Viterby score is the result of the most probable alignment
path of a KR sequence to a profile HMM. From these Viterby
scores we calculated the log-difference-of-odds-scores (Score-
Diff) to discriminate between d and l series. Negative values
correspond to l, positive values predict d configuration. High
deviations from zero (> 15) indicate a certain prediction. For
details see Ref. [12].
Received: February 12, 2014
Published online: May 21, 2014
Keywords: lipomycins · profile hidden Markov model ·
.
pentaenes · structure elucidation ·
vinylogous Mukaiyama aldol reaction
[1] B. Kunze, K. Schabacher, H. Zꢂhner, A. Zeeck, Arch. Mikrobiol.
1972, 86, 147 – 174.
[2] K. Schabacher, A. Zeeck, Tetrahedron Lett. 1973, 14, 2691 –
2994.
[3] For a review on tetramic acids, see: B. J. L. Royles, Chem. Rev.
1995, 95, 1981 – 2001.
[16] S. V. Ley, P. R. Woodward, Tetrahedron Lett. 1987, 28, 345 – 346.
[17] S. V. Ley, P. R. Woodward, Tetrahedron Lett. 1987, 28, 3019 –
3020.
[18] R. N. Lacey, J. Chem. Soc. 1954, 850 – 854.
[4] M. J. Nolte, P. S. Steyn, P. L. Wesseis, J. Chem. Soc. Perkin Trans.
1 1980, 1057 – 1065.
[19] a) S. V. Ley, S. C. Smith, P. R. Woodward, Tetrahedron Lett. 1988,
29, 5829 – 5832; b) S. V. Ley, S. C. Smith, P. R. Woodward,
Tetrahedron 1992, 48, 1145 – 1174.
[20] Racemization was reported by the Ley group under strongly
basic conditions and elevated temperature, see Ref. [19b].
[21] a) M. Kinoshita, H. Takami, M. Taniguchi, T. Tamai, Bull. Chem.
Soc. Jpn. 1987, 60, 2151 – 2161; b) G. Wang, Z. Huang, E.-I.
Negishi, Tetrahedron Lett. 2009, 50, 3220 – 3223.
[5] a) J. Gyimesi, I. Ott, I. Horvꢃth, I. Koczka, K. Magyar, J.
Antibiot. 1971, 24, 277 – 282; b) G. Horvꢃth, J. Gyimesi, Z.
Mꢄhesfalvi-Vajna, Tetrahedron Lett. 1973, 14, 3643 – 3648.
[6] a) N. P. Barashkova, Y. D. Shenin, L. G. Myasnikova, Antibiotiki
1976, 21, 487 – 491; b) Y. D. Shenin, Antibiot. Med. Biotekhnol.
1986, 31, 835 – 841.
[7] I. Casser, B. Steffan, W. Steglich, Angew. Chem. Int. Ed. Engl.
1987, 26, 586 – 587; Angew. Chem. 1987, 99, 597 – 598.
[8] For examples see: a) D. Janssen, D. Albert, R. Jansen, R. Mꢁller,
M. Kalesse, Angew. Chem. Int. Ed. 2007, 46, 4898 – 4901; Angew.
Chem. 2007, 119, 4985 – 4988; b) M. Bock, K. Buntin, R. Mꢁller,
A. Kirschning, Angew. Chem. Int. Ed. 2008, 47, 2308 – 2311;
Angew. Chem. 2008, 120, 2341 – 2344; c) D. Menche, F. Arikan,
O. Perlova, N. Horstmann, W. Ahlbrecht, S. C. Wenzel, R.
Jansen, H. Irschik, R. Mꢁller, J. Am. Chem. Soc. 2008, 130,
14234 – 14243; d) C. Jahns, T. Hoffmann, S. Mꢁller, K. Gerth, P.
Washausen, G. Hçfle, H. Reichenbach, M. Kalesse, R. Mꢁller,
Angew. Chem. Int. Ed. 2012, 51, 5239 – 5243; Angew. Chem.
2012, 124, 5330 – 5334.
[9] R. Reid, M. Piagentini, E. Rodriguez, G. Ashley, N. Viswana-
than, J. Carney, D. V. Santi, C. R. Hutchinson, R. McDaniel,
Biochemistry 2003, 42, 72 – 79.
[10] P. Caffrey, ChemBioChem 2003, 4, 654 – 657.
[11] a) D. H. Kwan, Y. Sun, F. Schulz, H. Hong, B. Popovic, J. C. C.
Sim-Stark, S. F. Haydock, P. F. Leadlay, Chem. Biol. 2008, 15,
1231 – 1240; b) D. H. Kwan, P. F. Leadlay, Chem. Biol. 2010, 5,
829 – 838.
[22] However, they isomerized to the desired E-configured com-
pounds when a catalytic amount of iodine was added, see
Ref. [19b].
[23] S. Shirokawa, M. Kamiyama, T. Nakamura, M. Okada, A.
Nakazaki, S. Hosokawa, S. Kobayashi, J. Am. Chem. Soc. 2004,
126, 13604 – 13605.
[24] A similar Dieckmann – olefination one-pot process was pre-
sented in the total synthesis of streptolydigin, see: S. V. Pronin,
S. A. Kozmin, J. Am. Chem. Soc. 2010, 132, 14394 – 14396.
[25] O. Hartmann, M. Kalesse, Org. Lett. 2012, 14, 3064 – 3067.
[26] a) J. A. Marshall, N. D. Adams, J. Org. Chem. 2002, 67, 733 – 740;
b) K. A. Scheidt, T. D. Bannister, A. Tasaka, M. D. Wendt, B. M.
Savall, G. J. Fegley, W. R. Roush, J. Am. Chem. Soc. 2002, 124,
6981 – 6990.
[27] J. Hoecker, K. Gademann, Org. Lett. 2013, 15, 670 – 673.
[28] A. B. Smith III, G. K. Friestad, J. Barbosa, E. Bertounesque, J. J.-
W. Duan, K. G. Hull, M. Iwashima, Y. Qiu, P. Grant Spoors,
B. A. Salvatore, J. Am. Chem. Soc. 1999, 121, 10478 – 10486.
[29] M. L. Hofferberth, R. Brꢁckner, Angew. Chem. 2014, 126, DOI:
10.1002/ange.201402255; Angew. Chem. 2014, 53, DOI: 10.1002/
anie.201402255.
7338
www.angewandte.org
ꢀ 2014 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Angew. Chem. Int. Ed. 2014, 53, 7335 –7338