Synthesis of potential drug metabolites by a modified Udenfriend reaction

Article abstract of DOI:10.1016/j.tetlet.2010.12.014

The scope and the limitations of a modified Udenfriend reaction for the one-step synthesis of potential drug metabolites were explored. Several drugs (clozapine, chlorpromazine, imipramine, buspirone, diltiazem, and propranolol) were subjected to modified Udenfriend conditions (Fe²⁺/Mn ²⁺/EDTA/ascorbic acid/O₂). From each reaction, one to four oxidation products were obtained in 1-8% overall yield. Many of these products (9 out of 14) have been reported to be metabolites of the parent drugs in vivo. The products resulted mainly from aromatic hydroxylation, and are not readily accessible by conventional synthesis. Thus, the described reaction may be useful in drug discovery whenever a facile synthetic access is more important than high yields (e.g., for a fast derivatisation of compounds or the preparation of metabolites). Poorly water-soluble compounds cannot be converted, which is an important limitation of this method. 2010 American Chemical Society.

Full text of DOI:10.1016/j.tetlet.2010.12.014

Tetrahedron Letters 52 (2011) 749–752
Contents lists available at ScienceDirect
Tetrahedron Letters
journal homepage: www.elsevier.com/locate/tetlet
Synthesis of potential drug metabolites by a modified Udenfriend reaction
Roger Slavik , Jens-Uwe Peters , Rudolf Giger à, Markus Bürkler, Eric Bald
⇑
Discovery Chemistry, F. Hoffmann-La Roche Ltd, CH-4070 Basel, Switzerland
a r t i c l e i n f o
a b s t r a c t
Article history:
Several drugs (clozapine, chlorpromazine, imipramine, buspirone, diltiazem, and propranolol) were sub-
jected to modified Udenfriend conditions (Fe2+/Mn2+/EDTA/ascorbic acid/O2). From each reaction, one to
four oxidation products were obtained in 1–8% overall yield. Many of these products (9 out of 14) have
been reported to be metabolites of the parent drugs in vivo. The products resulted mainly from aromatic
hydroxylation, and are not readily accessible by conventional synthesis. Thus, the described reaction may
be useful in drug discovery whenever a facile synthetic access is more important than high yields (e.g., for
a fast derivatisation of compounds or the preparation of metabolites). Poorly water-soluble compounds
cannot be converted, which is an important limitation of this method.
Received 15 September 2010
Revised 29 November 2010
Accepted 3 December 2010
Available online 9 December 2010
Keywords:
Metabolites
Biomimetic oxidation
Aromatic hydroxylation
Drug discovery
Ó 2010 Elsevier Ltd. All rights reserved.
Oxidation reagents and reaction systems, which mimic Cyto-
chrom-P450 (CYP) mediated phase I metabolism,1 have been an
area of active research.2 Such reagents are often based on a ‘single
oxygen donor’ (such as H2O2), or molecular oxygen as an oxidative
reagent, a transition-metal catalyst (e.g., Fe2+), and frequently, a
metal-complexing ligand; electrochemical oxidations are also
known.3 Unless an intentionally simple model compound is used,
these reagents typically give rise to a range of oxidation products,
which are reminiscent of the oxidation patterns observed in CYP
metabolism. For instance, the oxidation of phencyclidine4 or
cimetidine5 with Fenton’s reagent (Fe2+/H2O2) yields several
metabolite-like products. Predictably selective reagents have also
been discovered recently.6 Similarly, several (enantio-) selective
methods for the transition-metal catalysed epoxidation of alkenes
are established synthetic tools in organic synthesis.7
Potentially, biomimetic oxidations providing a reasonable num-
ber of oxidation products from single substrates would be highly
desirable for drug discovery, because they might provide a fast
access to metabolites, but would also be useful for a fast derivati-
sation of compounds in the early hit-to-lead phase of a discovery
project. Low yields would be acceptable as long as such reactions
provide a fast access to small amounts (ꢀ5 mg) of target com-
pounds for pharmacological characterisation or as references for
PK studies. Nevertheless, biomimetic oxidations have not yet found
widespread use in industrial research, possibly because a straight-
forward, general use for preparative purposes may not be obvious
from the literature: many papers discuss the study of specific oxi-
dations with simple model compounds, the use of optimised con-
ditions for the synthesis of specific metabolites, or oxidations
without a (preparative) isolation of the products (e.g., for in vitro
toxicity tests, or for spectroscopic identifications).3 Instead, indus-
trial researchers frequently rely on liver preparations or recombi-
nantly expressed CYPs for the synthesis of metabolites.8 Liver
preparations, for example, microsomes, are, however, costly and
available only in small quantities, which limits the amount of
obtained metabolites to the microgram range. The synthesis via
recombinant CYPs gives larger amounts of metabolites, but is
labour-intensive and time-consuming; moreover, the relevant
CYPs have to be identified beforehand in metabolite identification
studies.
The Udenfriend reaction system (Fe2+, EDTA, ascorbic acid, O2)9
has been explored as a model of hepatic metabolism,10 particularly
for aromatic hydroxylation (for mechanistic studies and
discussion, see Ref.11). To investigate whether such a biomimetic
oxidation might be an alternative to the above-mentioned biotech-
nology-based oxidations, we subjected marketed drugs to a Fe2+
/
Mn2+ variant of the Udenfriend reaction system (for details see
note12).
Clozapine, chlorpromazine, imipramine, buspirone, diltiazem,
and propranolol were oxidised under these conditions until analyt-
ical HPLC indicated a complete consumption of the substrate
(Scheme 1). The HPLC chromatogram suggested clean reactions,
with no remaining starting material and a reasonably small num-
ber of product peaks. However, most of the starting material seems
to have been converted to a black, tarry product in all of these reac-
tions. This insoluble product is easily removed by filtration,
⇑
Corresponding author. Tel.: +41 61 68 82636.
E-mail addresses: slavikr@student.ethz.ch (R. Slavik), jens-uwe.peters@roche.
com (J.-U. Peters), giger.mcc@gmail.com (R. Giger).
Present address: Steinbock Apotheke, Quaderstrasse 16, CH-7000 Chur,
Switzerland.
à
Present address: GIGER MedChem Consulting, Eichenweg 20, CH-4132 Muttenz,
Switzerland.
0040-4039/$ - see front matter Ó 2010 Elsevier Ltd. All rights reserved.
doi:10.1016/j.tetlet.2010.12.014

750
R. Slavik et al. / Tetrahedron Letters 52 (2011) 749–752
N
N
N
N
O
N
O
N
5.2%
N
N
N
(99%
purity)
9
N
N
Cl
HO
7.9%
Cl
buspirone
HO
N
N
7
H
N
N
H
N
N
O
N
O
clozapine
2 products, 1a,b
(table 1)
4
R1
R2
N
N
OMe
1
HO
7
S
N
S
N
S
Cl
N
Cl
2.6%
4.4%
OH
O
5
O
3
N
N
S
O
O
chlorpromazine
4 products, 2a-d
N
(table 2)
3 products, 5a-c
diltiazem
(table 4)
N
N
4
N
H
O
N
N
H
O
OH
2
N
7.5%
OH
OH
OH
1.4%
(75%
purity)
10
imipramine
3 products, 3a-c
(table 3)
propranolol
6
Scheme 1. Reagents and conditions: O2 (ambient pressure), FeSO4, Mn(OAc)2, EDTA, ascorbic acid, aqueous buffer (pH 4), 2 h 45 °C. The given yields are overall yields of
isolated products; yields are purity-corrected.19
whereas the desired oxidation products are readily separated from
the other components of the reaction mixture by an adjustment of
the pH to ꢀ8.5, and extraction with organic solvent. The oxidation
products were then isolated by preparative reverse-phase HPLC
with a standardised solvent gradient. Compounds were character-
ised by a variety of NMR methods13 and MS. The results for each
compound (Scheme 1) are discussed below:
Table 2
Chlorpromazine oxidation products
Compd
Yield (%)
Purity19 (%)
2a
2b
2c
2d
1-OH
3-OH
(5) S@O
7-OH
0.40
0.84
0.75
0.58
86
80
100
81
Clozapine (Table 1): the two isolated oxidation products, 1a and
1b, are also observed as (relatively minor) human metabolites.14
The major human metabolites, the N-demethylation and N-oxida-
tion products of clozapine,15 were not formed in our reaction sys-
tem; furthermore, the de-chlorination/hydroxylation product,
which is another minor human metabolite,16 was not found. Prod-
ucts 1a and 1b may be formed in vivo via a clozapine nitrenium
ion as a reactive metabolite.17,18
Table 3
Imipramine oxidation products
Compd
Yield (%)
Purity19 (%)
3a
3b
3c
2-OH
4-OH
10-OH
1.46
3.47
2.56
84
96
89
Chlorpromazine (Table 2): four oxidation products, 2a–d, were
isolated. Three of the products are hydroxylated in an activated
aromatic position, whereas the fourth product is a sulfoxide.
Chlorpromazine forms numerous metabolites in man; 2c and 2d
are major metabolites which may contribute to the therapeutic
efficacy of chlorpromazine.20
higher than that of 3a (2-OH); 3b is also a major metabolite in
humans.21 Additionally, a benzylic methylene unit was oxidised
(3c), which is the only example for an aliphatic hydroxylation
among the investigated compounds (apart from demethylation,
which may involve CH3 hydroxylation as a mechanistic step). Com-
pound 3c is also found as a human metabolite.22
Imipramine (Table 3): aromatic hydroxylation took place in the
activated 2- and 4-position. The yield of 3b (4-OH) was somewhat
Buspirone: the aromatic hydroxylation product 4 was obtained
as the only product in relatively good (5.2%) yield and high (99%)
purity. Compound 4 is one of the numerous buspirone metabolites,
which are observed in vivo.23 The important buspirone metabolite
1-pyrimidinylpiperazine, which is believed to contribute to buspi-
rone’s anxiolytic effects and its duration of action,24 was not found
in the reaction mixture.
Table 1
Clozapine oxidation products
Compd
Yield (%)
Purity19 (%)
1a
1b
7-OH
9-OH
4.93
2.98
84
71

R. Slavik et al. / Tetrahedron Letters 52 (2011) 749–752
751
Diltiazem (Table 4): the ester motif of the parent compound was
hydrolysed to the alcohol in all three isolated products. Whereas
5a was merely the hydrolysis product (i.e., no oxidation occured),
5b and 5c were additionally hydroxylated or O-de-methylated,
respectively. Compound 5a25 and 5c26 are also human metabolites
of diltiazem.
metabolism plays also no significant role in vivo, and amantadine
is mainly excreted unchanged renally.28 Verapamil gave a mixture
of regioisomeric O-de-methylation products, which was unsepara-
ble under our standard HPLC conditions (13.2% isolated yield). O-
de-methylation is also the major metabolic pathway of verapamil
in vivo.29
Propranolol: the ortho-hydroxylated 6 was the only isolated
product; the main in vivo metabolite, resulting from a hydroxyl-
ation in the para-position of the N substituent,27 was not found.
Moreover, the formation of 6 was relatively inefficient (1.4% yield,
75% purity). The observed regioselectivity may be explained by a
metal-chelating and thus ortho-directing effect of the propranolol
side chain.
All products were isolated by a standardised reverse-phase
HPLC protocol in a purity of >70% (analytical HPLC with UV detec-
tion at k = 254 nM), a purity which may be regarded as sufficient
for a preliminary in vitro pharmacological characterisation. All
experiments were performed only once, and no attempts were
made to improve the yields or purities.
The majority of the isolated oxidation products (10 out of 14)
are aromatic hydroxylation products. This preference for aromatic
hydroxylation may add to the attractiveness of this methodology,
because such hydroxylation products are usually difficult to obtain
via the established synthetic methods for the parent compounds.
In contrast, N-oxides and N-desalkylation metabolites (which are
usually easy to synthesize via established routes) were not ob-
served, likely because the protonation of basic alkyl-amines under
the acidic reaction conditions protects such motifs from N-oxida-
tion and oxidative desalkylation.
Other compounds (Fig. 1): the attempted oxidation of diazepam
and gemfibrozil failed due to the poor solubility of these com-
pounds in the acidic, aqueous reaction medium. Pre-dissolution
in acetone, or the addition of acetone or methanol as a co-solvent
brought no improvement. Amantadine gave no oxidation products,
despite good solubility; HPLC analysis indicated that the starting
material remained unchanged. Interestingly, phase I (oxidative)
Additional experiments were performed to compare the
modified Udenfriend conditions with the Fenton conditions, and
the original Udenfriend conditions. Oxidation of clozapine and pro-
pranolol under both Fenton conditions described in Ref.30 yielded
numerous products, but only traces of mono-hydroxylation prod-
ucts, as indicated by HPLC/MS. Oxidation of clozapine under the
original Udenfriend conditions9b gave an isolated yield of 3.0%
(94% purity) of 1b, but no 1a, whereas propranolol was not con-
verted at all under these conditions within 4.5 h. Thus, at least
for clozapine and propranolol, the modified Udenfriend conditions
gave superior results in the preparation of metabolite-like
products.
The oxidation products 1a–4, 5b, and 6 should be more suscep-
tible to oxidation than their parent compounds, due to the
increased electron-density of the hydroxylated aromatic rings.
The formation of black side products may be an indication of sub-
strate over-oxidation and subsequent polymerisation.31 Thus, reac-
tion times might be optimised to minimise over-oxidation, and
thus to maximise the yield of the desired mono-oxidation prod-
ucts. We monitored the oxidation of some substrates by analytical
HPLC in time-course experiments, and indeed found optimal
reaction times with a maximal relative peak area of the desired
product(s) in the HPLC chromatogram (clozapine: 1 h; chlorprom-
azine: 1.5–2.5 h; imipramine: 2 h; buspirone: 1–2.5 h).32 We also
noted that the reaction rate is very dependent on the set-up of
the reaction. For instance, passing oxygen through a glass frit into
the reaction mixture leads to much faster reaction times than the
use of a capillary. Thus, the optimal reaction time depends on
the substrate and the reaction set-up and can be determined
case-by-case.
In conclusion, the described modified Udenfriend reaction may
be useful to quickly access oxidation products, in particular
hydroxylation products, of drug-like compounds. The method
may be used in situations were a facile synthetic access is more
important than high product yields, for example, for fast derivati-
sation in the early stages of drug discovery, or for providing small
amounts of metabolites with no established or difficult synthetic
access. The described conditions are only applicable to compounds
which are sufficiently soluble in the aqueous acidic reaction med-
ium. This may constitute a limitation for a broad application in
drug discovery, as contemporary discovery projects often deal with
poorly water-soluble compounds. Similar oxidation systems,
which use organic solvents, appear, therefore, worthy of further
investigations.
Table 4
Diltiazem oxidation/hydrolysation products
Compd
R1
R2
Yield (%)
Purity19 (%)
5a
5b
5c
OMe
OMe
OH
H
OH
H
1.75
1.8
0.81
77
98
82
O
N
NH2
O
COOH
N
Cl
References and notes
1. For an overview on typical phase I metabolism reactions, mechanisms, and
significance for drug discovery, see: (a) Guengerich, F. P. Chem. Res. Toxicol.
2001, 14, 611–650; (b) Guengerich, F. P. J. Biochem. Mol. Toxicol. 2007, 21, 163–
168; (c) Smith, D. A.; Obach, R. S. Drug Metab. Dispos. 2005, 33, 1409–1417; (d)
Smith, D. A.; Obach, R. S. Chem. Res. Toxicol. 2009, 22, 267–279; (e) Smith, D. A.;
Obach, R. S. Bioanalysis 2010, 2, 1223–1233.
2. (a)Biomimetic Oxidations Catalyzed by Transition Metal Complexes; Meunier, B.,
Ed.; Imperial College Press, 1999; (b) Murahashi, S.-I.; Zhang, D. Chem. Soc. Rev.
2008, 37, 1490–1501; (c) Berkessel, A. Adv. Inorg. Chem. 2006, 58, 1–28; (d)
Bernadou, J.; Meunier, B. Adv. Synth. Catal. 2004, 346, 171–184; (e) Rocha
Gonsalves, A. M. d’A.; Pereira, M. M J. Mol. Catal. A: Chem. 1996, 113, 209–221;
(f) Johansson, T.; Weidolf, L.; Jurva, U. Rapid Commun. Mass Spectrom. 2007, 21,
2323–2331; (g) Che, C.-M.; Huang, J.-S. Chem. Commun. 2009, 27, 3996–4015;
(h) Piera, J.; Bäckvall, J.-E. Angew. Chem., Int. Ed. 2008, 47, 3506–3523.
3. Lohmann, W.; Karst, U. Anal. Bioanal. Chem. 2008, 391, 79–96.
4. Masumoto, H.; Takeuchi, K.; Ohta, S.; Hirobe, M. Chem. Pharm. Bull. 1989, 37,
1788–1794.
diazepam
gemfibrozil
amantadine
N
OMe
OMe
N
MeO
MeO
verapamil
Figure 1. Other drugs subjected to the described conditions. Diazepam and
gemfibrazil were not sufficiently soluble, amantadine was inert to the oxidation
conditions, verapamil gave unseparable products (see text).

752
R. Slavik et al. / Tetrahedron Letters 52 (2011) 749–752
5. Zbaida, S.; Kariv, R.; Fischer, P. Arch. Biochem. Biophys. 1988, 261, 12–15.
6. (a) Chen, M. S.; White, M. C. Science 2007, 318, 783–787; (b) Chen, M. S.; White,
M. C. Science 2010, 327, 566–571.
m), 1.61–1.63 (4H, m), 2.27 (2H, t), 2.35–2.37 (4H, m), 2.61 (4H, s), 3.52–3.56
(4H, m), 3.64 (2H, t), 8.00 (2H, s), 9.25 (1H, br s).
13. Compounds were characterised by 1H NMR (all), HSQC (all), 13C NMR (2c, 3a,b,
5a–c), HMBC (2c, 3a,b, 5a–c), 15N HMBC (2c), NOESY (3a,b).
14. Dain, J. G.; Nicoletti, J.; Ballard, F. Drug Metab. Dispos. 1997, 25, 603.
15. Fadiran, E. O.; Leslie, J.; Fossler, M.; Young, D. J. Pharm. Biomed. Anal. 1995, 13,
185–190.
7. (a) Gelalcha, F. G.; Bitterlich, B.; Anilkumar, G.; Tse, M. K.; Beller, M. Angew.
Chem., Int. Ed. 2007, 46, 7293–7296; (b) Katsuki, T. In Comprehensive
Asymmetric Catalysis; Jacobsen, E. N., Pfaltz, A., Yamamoto, H., Eds.; Springer,
1999; Vol. 2, pp 621–648; (c) Jacobsen, E. N.; Wu, M. H. In Comprehensive
Asymmetric Catalysis; Jacobsen, E. N., Pfaltz, A., Yamamoto, H., Eds.; Springer,
1999; Vol. 2, pp 649–677.
8. (a) Schroer, K.; Kittelmann, M.; Lutz, S. Biotechnol. Bioeng. 2010, 106, 699–706;
(b) Bernhardt, R. J. Biotechnol. 2006, 124, 128–145; (c) Hanlon, S. P.; Friedberg,
T.; Wolf, C. R.; Ghisalba, O.; Kittelmann, M. In Modern Biooxidation; Schmid, R.
D., Urlacher, V., Eds.; Wiley-VCH, 2007; pp 233–252; (d) Ceccarelli, S. M.;
Schlotterbeck, G.; Boissin, P.; Binder, M.; Buettelmann, B.; Hanlon, S.; Jaeschke,
G.; Kolczewski, S.; Kupfer, E.; Peters, J.-U.; Porter, R. H. P.; Prinssen, E. P.;
Rueher, M.; Ruf, I.; Spooren, W.; Stampfli, A.; Vieira, E. ChemMedChem 2008, 3,
136–144.
16. Stock, B.; Spiteller, G.; Heipertz, R. Arzneimittelforschung 1977, 27, 982–990.
17. Maggs, J. L.; Williams, D.; Pirmohamed, M.; Park, B. K. J. Pharmacol. Exp. Ther.
1995, 275, 1463–1475.
18. Dragovic, S.; Boerma, J. S.; van Bergen, L.; Vermeulen, N. P. E.; Commandeur, J.
N. M. Chem. Res. Toxicol. 2010, 23, 1467–1476.
19. The purities were assessed by analytical HPLC with UV detection (k = 254 nm).
The yields were corrected for purity (i.e., the impurities do not contribute to
the given yield).
20. (a) Daniel, W. A.; Syrek, M.; Ryłko, Z.; Kot, M. Pol. J. Pharmacol. 2001, 53, 615–
621; (b) Chetty, M.; Pillay, V. L.; Moodley, S. V.; Miller, R. Eur.
Neuropsychopharmacol. 1996, 6, 85–91.
9. (a) Udenfriend, S.; Clark, C. T.; Axelrod, J.; Brodie, B. B. J. Biol. Chem. 1954, 208,
731–739; (b) Brodie, B. B.; Axelrod, J.; Shore, P. A.; Udenfriend, S. J. Biol. Chem.
1954, 208, 741–750.
21. Potter, W. Z.; Calil, H. M.; Sutfin, T. A.; Zavadil, A. P., III; Jusko, W. J.; Rapoport, J.;
Goodwin, F. K. Clin. Pharmacol. Ther. 1982, 31, 393–401.
10. (a) Tamagaki, S.; Sasaki, M.; Tagaki, W. Bull. Chem. Soc. Jpn. 1989, 62, 159–163;
(b) Clark, C. T.; Downing, R. D.; Martin, J. B., Jr. J. Org. Chem. 1962, 27, 4698–
4701.
22. Crammer, J. L.; Scott, B. Psychopharmacologia 1966, 8, 461–468.
23. Jajoo, H. K.; Mayol, R.; LaBudde, J. A.; Blair, I. A. Drug Metab. Dispos. 1989, 17,
634–640.
11. (a) Grinstead, R. R. J. Am. Chem. Soc. 1960, 82, 3472–3476; (b) Rahhal, S.;
Richter, H. W. J. Am. Chem. Soc. 1988, 110, 3126–3133; (c) Belanzoni, P.;
Bernasconi, L.; Baerends, E. J. J. Phys. Chem. A 2009, 113, 11926–11937; (d)
Klinker, E. J.; Shaik, S.; Hirao, H.; Que, L., Jr. Angew. Chem., Int. Ed. 2009, 48,
1291–1295.
24. Mahmood, I.; Sahajwalla, C. Clin. Pharmacokinet. 1999, 36, 277–287.
25. Smith, M. S.; Verghese, C. P.; Shand, D. G.; Pritchett, E. L. Am. J. Cardiol. 1983, 51,
1369–1374.
26. Meshi, T.; Sugihara, J.; Sato, Y. Chem. Pharm. Bull. 1971, 19, 1546–1556.
27. Bond, P. A. Nature 1967, 213, 721.
12. Representative procedure—oxidation of buspirone: Buspirone (822 mg,
2.13 mmol), EDTA trisodium salt (382 mg, 1.1 mmol), Mn(II)-acetate
tetrahydrate (784 mg, 3.2 mmol) and ascorbic acid (2.63 g, 14.9 mmol) were
dissolved in water (10.0 ml) and acetate buffer 2 M, pH 4 (12.0 ml) to give a
colourless solution. A second solution containing EDTA trisodium salt (382 mg,
1.1 mmol) and iron(II) sulfate heptahydrate (593 mg, 2.13 mmol) in water
(10.0 ml) was added to the mixture to give a brown solution. The mixture was
well-stirred and heated to 45 °C. Oxygen was bubbled into the mixture through
a glass frit for 2 h. The reaction mixture was then adjusted to pH ꢀ8.5 using
concd NaOH, and extracted with ethyl acetate. The combined organic layers
were dried over Na2SO4. Solvents were removed under reduced pressure, and
28. Goralski, K. B.; Smyth, D. D.; Sitar, D. S. J. Pharmacol. Exp. Ther. 1999, 290, 496–
504.
29. Nelson, W. L.; Olsen, L. D.; Beitner, D. B.; Pallow, R. J., Jr. Drug Metab. Dispos.
1988, 16, 184–188.
30. Zbaida, S.; Kariv, R.; Fischer, P.; Gilhar, D. Xenobiotica 1987, 17, 617–621.
31. Black, tarry products were also observed in the attempted oxidation of
propranolol under the original Udenfriend conditions, which gave no
conversion of propranolol; thus, the observed polymeric products may
originate at least in part from other components of the reaction system, for
example, ascorbic acid.
32. We also attempted to minimize over-oxidation by using the described
modified Udenfriend procedure, however with lower amounts (one third) of
reagents, and a limited reaction time (2 h). Under these conditions, clozapine
gave, after incomplete conversion, 1a in 1.5% yield (62% purity), and 1b in 1.0%
yield (100% purity), whereas no oxidation was observed for propranolol.
oxidation product
4 (44.6 mg, 5.16%) was isolated from the residue by
automated, preparative HPLC (Phenomenex Gemini C18 column 75 Â 20 mm,
solvent gradient 20–98% MeOH in 0.1% Et3N (aq) over 13.0 min, flow rate
40 ml/min, UV detection [k = 254 nm]). 1H NMR (DMSO-d6): d = 1.39–1.43 (8H,