7442
C. Kelso et al. / Tetrahedron Letters 48 (2007) 7440–7443
(EIMS and +ve and ꢀve ion ESMS) for the formation
of the desired product 11 (expected m/z = 384.1). ESMS
(ꢀve ion) analysis for the crude reaction mixtures typi-
cally displayed a prominent peak at m/z = 365.2, corre-
sponding to an MꢀH species 18 mass units lower than
for 11. Lack of success with Walsh and Olsen’s condi-
tions led us to explore the reduction using other combi-
nations of hydride reagents (NaCNBH3, NaBH4 and
NaBH(OAc)3), acids (CH3CO2H, TsOH and PPTS)
and solvents (MeOH, CH3CO2H and THF) at various
temperatures, however, none of these reactions provided
any trace of 11 and similarly showed multiple products
and unreacted 10 by TLC analysis, as well as ꢀve ion
ESMS spectra in which a peak at m/z = 365.2 (as be-
fore) was apparent.
NOEs
H C
H
3
H
O
O
OH
OH
H
H
OMe
OH
H
H
Figure 1. NOEs observed in the NOESY spectrum (500 MHz, 600 ms
mixing time) of 7.
protonate to create a good benzylic leaving group,
which loses H2O following attack by C13 to yield 7.
Although reasonable, this mechanism is incomplete
since it cannot account for the fact that only the
7S,9R,13S stereoisomer of 7 was formed in the reaction.
The 7S,9R stereochemistry would be predicted in 7 from
the above mechanism based on the 7S,9S stereochemis-
try of daunomycinÆHCl 8a precursor. However, unless
hydride reduction of benzenesulfonylhydrazone 10 is
stereospecific, with a hydride ion attacking the proton-
ated benzenesulfonylhydrazonium ion from just one
face, some of the 7S,9R,13R-epimer of 7 should also
be observed.
Using Maryanoff’s conditions for the reduction of tosyl-
hydrazones to methylenes (i.e., NaCNBH3, TsOH, 1:1
DMF–sulfolane, 110 °C),15 we noted that (i) 10 was
completely consumed within 15 min and that a single
new spot of higher Rf than 10 was visible by TLC; (ii)
the reaction mixtures prominently displayed the pre-
viously observed m/z = 365.2 peak in their ꢀve ion ESMS
spectra. After purification by silica gel chromatography
and comprehensive 2D-NMR characterization (DQF-
COSY, HMQC, HMBC and NOESY), it was clear that
the product was not 11 but novel anthracyclinone 7.
The A-ring of daunomycinone 9 is known to prefer a
conformation in which the two hydroxyl groups at C7
and C9 adopt a pseudo-diaxial orientation stabilized
by an intramolecular 6-membered hydrogen bond (H-
bond) ring (Scheme 2, 9).17 Similar conformations can
be envisaged for the cis- and trans-isomers of benzene-
sulfonylhydrazone 10 (Scheme 2, cis-10a and trans-
10a). However, 10 presents additional H-bond donor
and acceptor groups in its benzenesulfonylhydrazone
moiety, which could potentially compete with the C7–
OH for H-bonding to C9–OH by forming 5- and
6-membered H-bond rings. These alternative H-bonded
conformers could either retain the pseudo-diaxial orien-
tation of the C7 and C9 alcohols (Scheme 2, not shown)
or, alternatively, a ring-flip could occur to reposition the
two alcohols in a pseudo-diequatorial orientation
(Scheme 2, cis-10b and trans-10b). As mentioned previ-
The yields of 7 under these conditions were consistently
around 30%, which seemed low considering the clean
conversions that were observed from TLC analysis. It
was subsequently apparent that the remaining mass cor-
responded to highly polar (presumably polymerized)
material that could only be eluted using MeOH. At-
tempts to optimize the yield of 7 by varying the reaction
conditions met with limited success; although it was
determined that sulfolane could be omitted from the
reaction without affecting the yield. The 7S,9R,13S
absolute stereochemistry for 7 was unambiguously as-
signed based on the 7S,9S stereochemistry of its dauno-
mycin precursor and from NOEs (Fig. 1).
The generally accepted mechanism for the reduction of
tosylhydrazones to methylenes under acidic conditions
invokes an initial hydride attack on the sp2 carbon of
a protonated tosylhydrazonium ion to generate the re-
duced tosylhydrazine derivative. Elimination of p-tolu-
enesulfinic acid from the tosylhydrazine subsequently
gives rise to a diazene intermediate, which decomposes
by losing N2 to afford the corresponding methylene
compound, Eq. 1.16
ously, H and 13C NMR spectra for 10 were extremely
1
complex supporting the notion that 10 probably exists
(in CDCl3) as an equilibrium mixture of a number of
these isomers/conformers, which interconvert slowly
on the NMR timescale.
Protonation of cis-10b and trans-10b under acidic condi-
tions would result in formation of the cis- and trans-benz-
Ts-
N2
BH3CN-
ð1Þ
H+
+
R2C=NHNHTs
R2CHN=NH
R2C=NNHTs
R2CHNHNHTs
R2CH2
An analogous mechanism for the formation of 7 from 10
can be postulated wherein the developing negative
charge at C13 resulting from loss of N2 in the final step
does not protonate to yield 11 but instead attacks C7.
Under the acidic reaction conditions, the C7–OH could
enesulfonylhydrazonium ions cis-10c and trans-10c.
Models indicate that the Re faces of cis-10c and trans-
10c are both significantly shielded from approaching
nucleophiles by the axial hydrogen at C7. If cis-10c
and trans-10c represent the predominant species present