An expeditious route to fluorinated rapamycin analogues by utilising mutasynthesis

Article abstract of DOI:10.1002/cbic.200900723

Rapamycin is a drug with several important clinical uses. Its complex structure means that total synthesis of this natural product and its analogues is demanding and lengthy. A more expeditious approach is to utilise biosynthesis to enable the generation

Full text of DOI:10.1002/cbic.200900723

DOI: 10.1002/cbic.200900723
An Expeditious Route to Fluorinated Rapamycin
Analogues by Utilising Mutasynthesis
Rebecca J. M. Goss,*[a] Simon Lanceron,[a] Abhijeet Deb Roy,[a] Simon Sprague,[b]
Mohammed Nur-e-Alam,[b] David L. Hughes,[a] Barrie Wilkinson,[b] and Steven J. Moss*[b]
Rapamycin is a drug with several important clinical uses. Its
complex structure means that total synthesis of this natural
product and its analogues is demanding and lengthy. A more
expeditious approach is to utilise biosynthesis to enable the
generation of otherwise synthetically intractable analogues. In
order to achieve this, rules governing biosynthetic precursor
substrate preference must be established. Through determin-
ing these rules and synthesising and administering suitable
substrate precursors, we demonstrate the first generation of
fluorinated rapamycin analogues. Here we report the genera-
tion of six new fluororapamycins.
Introduction
Rapamycin is a polyketide natural product that selectively in-
hibits the kinase mammalian target of rapamycin (mTOR), a
central regulator of the cell cycle and proliferation.[1] Inhibition
of mTOR has significant pharmacological value. This fact is un-
derscored by rapamycin’s clinical application as an immuno-
suppressant therapy (Rapamuneꢀ) following organ transplanta-
tion, and the recent discovery that inhibition of mTOR signal-
ling by rapamycin extends lifespan in genetically heterogene-
ous mice, even when administered late in life.[2] Recently the
semisynthetic analogues temsirolimus (Toriselꢀ) and everolimus
(Afinitorꢀ) were approved as mTOR inhibitors for the treatment
of advanced renal cell carcinoma, and further analogues are in
late stage clinical trials.[3] These semisynthetic analogues are
similarly modified at the C40 secondary hydroxyl group on the
highly substituted cyclohexane moiety, attesting to the limita-
tions of this approach. Moreover, although a number of total
syntheses of this challenging molecule and its analogues have
been described,[4] the complexity and length of these synthe-
ses preclude their utility in lead optimisation.
rapamycins have demonstrated a lack of antiproliferative activi-
ty against cancer cell lines and in the mixed lymphocyte reac-
tion (a measure of immunosuppressive activity) as reported in
the Supporting Information. The advantage of using this strain
background as a model system to access substrate specificity
is that a single rapalogue is produced if a suitable starter unit
is fed. Rules governing the incorporation of the starter unit
may be readily determined by using this model system and
then applied to the production of more advanced compounds.
The presence of a hydrogen-bond acceptor on the PKS start-
er acid has been shown to be crucial for the incorporation of
the cyclohexanecarboxylic acid into rapamycin.[6–8] Exogenously
added cyclohexane-1-carboxylic acid and cyclohex-1-enecar-
boxylic acid are utilised by the PKS to initiate biosynthesis only
after they are first hydroxylated at the 4- and 3-positions, re-
spectively.
The presence of fluorine in a molecule can confer beneficial
physical and medicinal properties such as increased stability
and improved bioavailability.[9] Although a CÀF bond is almost
the same length as a CÀOH bond, the corresponding cyclohex-
anecarboxylic acids fluorinated at the 4-, 3- and 2-positions
were not incorporated. This reinforces the idea that a hydro-
gen bond acceptor is required in order for the rapamycin PKS
to utilise a carboxylic acid to initiate biosynthesis. In light of
this we decided to investigate the ability of the PKS to utilise
substrates with both a fluorine atom and a hydroxyl group,
An alternative and expeditious route through which to gen-
erate new rapamycin analogues (rapalogues) is the exploita-
tion of its biosynthetic pathway. Mutasynthetic approaches
have previously been successfully employed to generate ana-
logues of a number of natural products.[5] Notably, the afore-
mentioned cyclohexane moiety of rapamycin can be targeted
through the application of precursor-directed feeding and mu-
tasynthesis methods.[6,7] For example, administration of suitable
cyclohexanecarboxylic and cycloheptanecarboxylic acids to a
mutant (MG2-10, assigned BIOT-3016) of the producing organ-
ism, Streptomyces hygroscopicus, which cannot produce the
natural starter acid required to initiate polyketide synthesis by
the polyketide synthase (PKS), enables the selective production
of prerapamycin analogues.[7] Prerapamycin (1b, Scheme 2,
below) is the untailored polyketide that lacks all post-poly-
ketide synthase modifications, that is, oxidation at C9 and C27,
and O-methylation at C16, C27 and C39. Studies with the pre-
[a] Dr. R. J. M. Goss, Dr. S. Lanceron, Dr. A. Deb Roy, Dr. D. L. Hughes
School of Chemical Sciences, University of East Anglia
Norwich, NR4 7TJ (UK)
E-mail: r.goss@uea.ac.uk
[b] S. Sprague, Dr. M. Nur-e-Alam, Dr. B. Wilkinson, Dr. S. J. Moss
Biotica Technology Ltd., Chesterford Research Park
Cambridge, CB10 1XL (UK)
steven.moss@biotica.com
Supporting information for this article is available on the WWW under
http://dx.doi.org/10.1002/cbic.200900723.
698
ꢁ 2010 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
ChemBioChem 2010, 11, 698 – 702

Rapamycin
mediated lactone opening.[10]
The subsequent displacement of
iodide by the released hydroxyl
afforded the two epoxides, 7
and 8 in a 7:1 ratio, which were
separated chromatographically
and their relative stereochemis-
try determined from their NOESY
spectra (Supporting Informa-
tion). Synthesis of 11 and 13 was
achieved by ring opening epox-
ide 7 with Et3N:HF to yield ethyl
(1R*,3R*,4R*)-4-fluoro-3-hydroxy-
cyclohexanecarboxylate 11 and
ethyl (1S*,3R*,4R*)-3-fluoro-4-hy-
droxycyclohexanecarboxylate 13
in a 4:1 ratio. In order to sepa-
rate these regioisomers chroma-
tographically, they were protect-
ed as their respective THP
ethers, 9 and 10. Following chro-
matographic separation, removal
of the THP group under acidic
conditions furnished compounds
11 and 13. The fluorination of
epoxide 8 gave fluorohydrin 12
as a single compound. In order
to access ethyl (1R*,3R*,4R*)-3-
fluoro-4-hydroxycyclohexanecar-
boxylate 18, the THP ether of
ethyl (1S*,3R*,4R*)-3-fluoro-4-hy-
droxycylclohexanecarboxylate
9
was epimerised by refluxing with
Scheme 1. Synthesis of ethyl (1S*,3R*,4R*)-, (1S*,3R*,4S*)-, (1R*,3R*,4R*)- and (1R*,3R*,4S*)-3-fluoro-4-hydroxy cyclo-
hexanecarboxylate 13, 16, 18, 21 and (1R*,3R*,4R*)- and (1R*,3S*,4S*)- 4-fluoro-3-hydroxycyclohexanoates 11 and
12 (boxed). a) I2, KI, NaHCO3, H2O, RT, 16 h, 96%; b) EtONa, EtOH, THF, 08C, 16 h, 48% 7, 7% 8; c) Et3N:HF, RT, 16 h,
76% as a mixture of 11 and 13; d) DHP, H2NSO3H, 4 d, 52% 9, 15% 10; e) p-TsOH, EtOH, RT, 2 h, 76%; f) Et3N:HF,
RT, 76%; g) p-TsOH, EtOH, RT, 2 h, 99%; h) triflic anhydride, pyridine, DCM, 08C, 2 h, 76%; i) (Bu)4N+BzOÀ, toluene,
RT, 16 h, 65%; j) Na, EtOH, RT, 16 h, 57% 16, 33% 21; k) tBuOK , toluene, reflux, 40 min, 53%; l) p-TsOH, MeOH, RT,
2 h, 92%; m) triflic anhydride, pyridine, DCM, 08C, 2 h, 67%; n) (Bu)4N+BzOÀ, toluene, RT, 16 h, 61%; o) Na, EtOH,
RT, 16 h, 88% 21, 4% 16.
tBuOK prior to deprotection of
the
THP
ether.
Ethyl
(1S*,3R*,4S*)-, and (1R*,3R*,4S*)-
3-fluoro-4-hydroxycyclohexane-
carboxylate 16, and 21 were ac-
cessed by inverting the hydrox-
yls of 13 and 18, respectively.
This was achieved by transform-
and thus we synthesised a series of appropriately substituted
fluorohydrins (Scheme 1).
ing the fluorohydrins 13 and 18 to their corresponding triflates
followed by benzoylation using tetrabutylammonium benzoate
to yield 15 and 20. Hydrolysis of the benzoate esters under
basic condition yielded the corresponding fluorohydrins 16
and 21 along with a small amount of epimerised product
(Scheme 2). The relative stereochemistry of compounds 12, 13,
and 18 was confirmed from the crystal structures of their PNB
derivatives (Figure 1). Derivatives of 11, 16 and 21 did not
yield well to crystallisation; the relative stereochemistry of
these compounds was clearly deduced from their NOESY spec-
tra (Supporting Information).
Results and Discussion
In order to determine the importance of relative stereochemis-
try of substituents on the cyclohexane ring for incorporation
into the polyketide we synthesised the ethyl (1S*,3R*,4R*)-,
(1S*,3R*,4S*)-, (1R*,3R*,4R*)- and (1R*,3R*,4S*)-3-fluoro-4-hy-
droxycyclohexanecarboxylates 13, 16, 18 and 21, respectively.
The issue of regiochemistry was also addressed by the synthe-
sis and feeding of ethyl (1R*,3R*,4R*)- and (1R*,3S*,4S*)-4-
fluoro-3-hydroxycyclohexanecarboxylates 11 and 12.
Fluorohydroxycyclohexanecarboxylic acids and their ester
derivatives were administered to S. hygroscopicus MG2-10
(BIOT-3016) at a final concentration of 2 mm. Using this strain,
rapalogues are only generated if a suitable starter unit is ad-
ministered as a carboxylic acid or carboxylic acid derivative.[7]
In our general synthetic procedure (Scheme 1), relative ste-
reochemistry was established through the iodolactonisation of
cyclohex-3-ene carboxylic acid, followed by sodium ethoxide-
ChemBioChem 2010, 11, 698 – 702
ꢁ 2010 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.chembiochem.org
699

R. J. M. Goss, S. J. Moss et al.
necarboxylic acid by this strain). When the six fluorohydrins 11,
12, 13, 16, 18 and 21 were administered to S. hygroscopicus
MG2-10 (BIOT-3016), we observed the production of six new
prerapamycin analogues (Table 1). Careful analysis by LC-MS/
MS showed these were consistent with rapalogues carrying a
fluorine atom in the cyclohexane moiety. Notably, the regio-
chemistry and relative stereochemistry of the fluorohydrins sig-
nificantly affected the new rapalogue production titre. Our
studies revealed a significant preference for location of the hy-
droxyl group at the 4-position, as suggested by previous stud-
ies.[6–8] For instance, the rapalogue resulting from administering
18 is six times more abundant than the equivalent rapalogue
resulting from feeding 11. Preference for a trans relationship
between the hydroxyl and carboxylate groups was also evi-
dent, with the administration of 18 resulting in three times the
titre of product compared to that resulting from administration
of 13 to the cultures. With a 1,4-trans relationship in place, the
presence of fluorine is well tolerated with little difference in
preference for it occupying the axial or equatorial position.
In order to verify the identity of one new rapalogue, and
thus validate our findings, we fed fluorohydrin 18 to a larger
culture of S. hygroscopicus MG2-10 (BIOT-3016). From 1.5 L of
fermentation broth we isolated 16.4 mg of the new com-
pound 1 (Scheme 2) by using standard procedures as de-
scribed previously.[6,7] The structure of 1 was determined by a
combination of high-resolution
Scheme 2. Directing biosynthesis to generate 39-desmethoxy-39-fluoro-9-
deoxo-16-O-desmethyl-27-desmethoxyrapamycin (1).
Previous attempts to generate fluorinated rapamycin ana-
logues by administering 4-, 3- and 2-fluorocyclohexanecarbox-
ylic acids to the strain had been unsuccessful, and we postulat-
ed that this was due to the requirement of a hydrogen-bond
acceptor at the 3- or 4-position.[8] Hydroxylation of the fluoro-
cyclohexane carboxylates by the strain was prevented by the
presence of fluorine (cf. the facile hydroxylation of cyclohexa-
MS and multidimensional NMR,
(Supporting Information) with
comparison to the structural
data for closely related struc-
tures described previously.[5,6] As
expected, 1 exists as two inter-
converting forms, and we as-
signed the signals corresponding
to each form if possible. It is be-
lieved that these two forms cor-
respond to isomers which result
from cis–trans isomerism of the
amide bond.
Conclusions
In summary, we report the gen-
eration of six fluorinated ana-
logues of rapamycin. Through
these experiments we have re-
vealed that variation in both the
regiochemistry and relative ste-
reochemical orientation of the
substitution on a substituted cy-
clohexanecarboxylic acid are tol-
erated by the PKS in cases in
Figure 1. Crystal structures of A) 12-PNB, showing the ethyl ester adopting the equatorial position and 4-fluoro
and 3-O-PNB groups adopting the axial positions, B) 13-PNB, showing the ethyl ester adopting the equatorial po-
sition and the 3-fluoro and 4-O-PNB groups adopting the axial positions, and C) 18-PNB showing all groups equa-
torial. Hydrogen atoms have been omitted for clarity. Interestingly, a switch in conformation was observed for 12
and 13, upon their conversion to the PNB derivative, placing the O-PNB group into the axial position. This change
in conformation was not observed for 18.
which they are utilised as starter
acids to initiate polyketide bio-
synthesis. A preference was ob-
served for 1,4-hydroxyacids with
a trans relationship. These results
700
www.chembiochem.org
ꢁ 2010 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
ChemBioChem 2010, 11, 698 – 702

Rapamycin
(55%) to mobile phase B (95%) over 10 min followed by an isocrat-
ic hold at mobile phase B (95%) for 2 min with a flow rate of
1 mLminÀ1. Mobile phase A was 1:9 acetonitrile/water, containing
ammonium acetate (10 mm) and trifluoroacetic acid (0.1%), mobile
phase B was 9:1 acetonitrile/water, containing ammonium acetate
(10 mm) and trifluoroacetic acid (0.1%). Rapamycin analogues were
identified by the presence of the characteristic rapamycin triene,
centred on l=278 nm. Samples were quantified based on a rapa-
mycin calibration curve, measuring peak area at l=280 nm.
Table 1. Incorporation of synthetic starter acids into rapamycin.
Starter acid
analogue
Rapamycin analogue
(Scheme 2)
Rapamycin analogue
produced [mgLÀ1
]
27Æ2
29Æ2
5Æ1
Preparation of 39-desmethoxy-39-fluoro-9-deoxo-16-O-desmeth-
yl-27-desmethoxyrapamycin (1): Under sterile conditions MD6
media (2 L), d-fructose aqueous solution (40% (w/v), 100 mL) and
filter sterilised l-lysine aqueous solution (14% (w/v), 28 mL) were
mixed together to homogeneity. S. hygroscopicus seed cultures
(0.5 mL) were used to inoculate portions (300ꢂ7.45 mL) of the
MD6-fructose-l-lysine medium in 50 mL falcon tubes. The falcon
tubes were sealed with foam bungs and were shaken at 300 rpm
(2.5 cm throw) at 268C for 24 h.
7Æ1
10Æ1
Compound 18 was prepared in methanol to a concentration of
0.32 molLÀ1 and was fed (50 mL, 16 mmol) to 300 previously pre-
pared S. hygroscopicus MG2–10 (BIOT-3016) cultures to reach a final
concentration of 2 mmolLÀ1. The cultures were sealed with foam
bungs and were shaken at 300 rpm (2.5 cm throw) at 268C for a
further five days. The 300 cultures were combined, the cells har-
vested by centrifugation (25 min at 3480g), and the supernatant
was discarded. Acetonitrile (2.75 L) was added to the cells; they
were stirred for one hour and decanted. The cells were extracted a
second time with acetonitrile (2.4 L). The combined organics were
reduced in vacuo to an aqueous slurry that was extracted with two
volume equivalents of ethyl acetate. The solvent was then re-
moved under reduced pressure to yield a crude extract (3.5 g).
26Æ2
28Æ1
4Æ1
The crude extract was dissolved in MeOH (50 mL) and silica (5 g)
added. The solvent was removed in vacuo, and the resultant
powder was loaded onto a column (40 g) of silica that was precon-
ditioned with a mixture of ethyl acetate and hexanes (1:1). The
column was then eluted with ethyl acetate/hexanes mixtures (1 L
of 1:1; then 1 L 55:45; then 4 L of 60:40) and 250 mL fractions
were collected. Fractions 5–20 were found to contain the target
compound and were combined and the solvent removed in vacuo.
The enriched extract was then dissolved in acetonitrile (1.8 mL)
and purified by preparative HPLC (Column-Waters Xterra MS C18,
10 micron, 250 mm ꢂ 19 mm diameter; flow rate 20 mLminÀ1; sol-
vent A=water, solvent B=acetonitrile; isocratic 45% B, tR of target
compound=21 min). The solvent was removed in vacuo to reveal
the target compound as an off-white amorphous solid (16.4 mg).
suggest that the fluorinated cyclohexanecarboxylic acids previ-
ously prepared and fed to the same organism[8] did not result
in fluorinated rapamycins because the strain was unable to
hydroxylate the substrate due to the presence of a fluorine
substituent.
Experimental Section
Full synthetic details and compound characterisation are described
within the Supporting Information.
Feeding experiments: Compounds 11–13, 16, 18, 21–23 were
prepared in methanol (0.32 molLÀ1) and were individually fed in
quadruplicate to 24-hour-old S. hygroscopicus MG2–10 (BIOT-3016)
cultures (50 mL to 7 mL culture, giving a final concentration of
2 mmolLÀ1), except for compound 16, which was fed in triplicate.
The falcon tubes were shaken at 300 rpm (2.5 cm throw) at 268C
for a further 5 days. A single sample of uninoculated media and
two unfed S. hygroscopicus cultures were used as negative con-
trols.
CCDC 723198 (18-PNB) 723199 (12-PNB) and 723200 (13-PNB) con-
tain the supplementary crystallographic data for this paper. These
data can be obtained free of charge from The Cambridge Crystallo-
graphic Data Centre via www.ccdc.cam.ac.uk/data_request/cif
Acknowledgements
Analysis of fluorinated rapamycins: An aliquot of whole culture
broth (0.9 mL) was added to methanol (0.9 mL) in a 2 mL Eppen-
dorf, and then shaken for 30 min. The sample was centrifuged
(10 min, 16000g) and the supernatant (150 mL) was transferred to
an HPLC vial for analysis by HPLC with diode array detection. The
HPLC system comprised an Agilent HP1100 equipped with a Hy-
perclone 3 mm BDS C18 130A column 150ꢂ4.6 mm (Phenomenex)
heated to 508C. The gradient elution was from mobile phase B
The authors are grateful to the Royal Society and to B.P. for a
research fellowship (to R.J.M.G.) and to EPSRC and Biotica for a
CASE studentship (S.L.). The National Mass Spectrometry Labora-
tory at Swansea is gratefully acknowledged. We also thank Dr.
Alan Haines for fruitful discussions and for providing tetrabutyl-
ammonium benzoate.
ChemBioChem 2010, 11, 698 – 702
ꢁ 2010 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.chembiochem.org
701

R. J. M. Goss, S. J. Moss et al.
[5] a) A. Kirschning, F. Taft, T. Knobloch, Org. Biomol. Chem. 2007, 5, 3245–
3259; b) S. Weist, R. D. Suessmuth, Appl. Microbiol. Biotechnol. 2005, 68,
141–150; c) R. Thiericke, J. Rohr, Nat. Prod. Rep. 1993, 10, 265–289.
[6] P. A. Lowden, G. A. Bçhm, S. Metcalfe, J. Staunton, P. F. Leadlay, Chem-
BioChem 2004, 5, 535–538.
Keywords: antitumor agents
immunosuppressants · natural products
· biosynthesis · fluorine ·
[7] M. A. Gregory, H. Petkovic, R. E. Lill, S. J. Moss, B. Wilkinson, S. Gaisser,
P. F. Leadlay, R. M. Sheridan, Angew. Chem. 2005, 117, 4835–4838;
Angew. Chem. Int. Ed. 2005, 44, 4757–4760.
[8] R. J. M. Goss, S. E. Lanceron, N. J. Wise, S. J. Moss, Org. Biomol. Chem.
2006, 4, 4071–4073.
[9] a) “Biomedical Frontiers of Fluorine Chemistry”, K. L. Kirk, R. Filler, ACS
Symp. Ser. 1996, 639, 1–24; b) S. Weist, B. Bister, O. Puk, D. Bischoff, S.
Pelzer, G. J. Nicholson, W. Wohlleben, G. Jung, R. D. Sꢃssmuth, Angew.
Chem. 2002, 114, 3531–3534; Angew. Chem. Int. Ed. 2002, 41, 3383–
3385; c) Z. Hojati, C. Milne, B. Harvey, L Gordon, M. Borg, F. Flett, B. Wil-
kinson, P. J. Sidebottom, B. A. M. Rudd, M. A. Hayes, C. P. Smith, J. Mick-
lefield, Chem. Biol. 2002, 9, 1175–1188.
[1] a) L. M. Ballou, R. Z. Lin, J. Chem. Biol. 2008, 1, 27–36; b) R. T. Abraham,
Cell 2002, 111, 9–12.
[2] D. E. Harrison, R. Strong, Z. D. Sharp, J. F. Nelson, C. M. Astle, K. Flurkey,
N. L. Nadon, J. E. Wilkinson, K. Frenkel, C. S. Carter, M. Pahor, M. A.
Javors, E. Fernandez, R. A. Miller, Nature 2009, 460, 392–395.
[3] I. R. Konings, J. Verweij, E. A. Wiemer, S. Sleijfer, Curr. Cancer Drug Targets
2009, 9, 439–450.
[4] For example see: a) K. C. Nicolaou, T. K. Chakraborty, A. D. Piscopio, N.
Minowa, P. Bertinato, J. Am. Chem. Soc. 1993, 115, 9345; b) D. Romo,
S. D. Meyer, D. D. Johnson, S. L. Schreiber, J. Am. Chem. Soc. 1993, 115,
7906; c) C. M. Hayward, D. Yohannes, S. J. Danishefsky, J. Am. Chem. Soc.
1993, 115, 9345; d) A. B. Smith, S. M. Condon, J. A. McCauley, J. L.
Leazer, J. W. Leahy, R. E. Maleczka, J. Am. Chem. Soc. 1995, 117, 5407;
e) M. L. Maddess, M. N. Tackett, H. Watanabe, P. E. Brennan, C. D. Spilling,
J. S. Scott, D. P. Osborn, S. V. Ley, Angew. Chem. 2007, 119, 597–603;
Angew. Chem. Int. Ed. 2007, 46, 591–597.
[10] K. Narasaka, H. Tanaka, F. Kana, Bull. Chem. Soc. Jpn. 1991, 64, 387–391.
Received: November 26, 2009
Published online on February 23, 2010
702
www.chembiochem.org
ꢁ 2010 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
ChemBioChem 2010, 11, 698 – 702