Debenzylation using catalytic hydrogenolysis in trifluoroethanol, and the total synthesis of (-)-raumacline

Article abstract of DOI:10.1016/j.tetlet.2008.01.104

N-Debenzylations using catalytic hydrogenolysis often fail to take place unless alcoholic solvents are used, but this can lead to N-alkylation as a side reaction; using trifluoroethanol as the solvent overcomes this problem, and leads to highly reliable hydrogenolyses for a wide range of substrates, including the final deprotection step in our total synthesis of (-)-raumacline.

Full text of DOI:10.1016/j.tetlet.2008.01.104

Available online at www.sciencedirect.com
Tetrahedron Letters 49 (2008) 2150–2153
Debenzylation using catalytic hydrogenolysis in trifluoroethanol,
and the total synthesis of (À)-raumacline
Patrick D. Bailey *, Mark A. Beard, Hoa P. T. Dang, Theresa R. Phillips,
Richard A. Price, James H. Whittaker
School of Chemistry, University of Manchester, Oxford Road, Manchester M13 9PL, United Kingdom
Received 7 January 2008; accepted 23 January 2008
Available online 30 January 2008
Abstract
N-Debenzylations using catalytic hydrogenolysis often fail to take place unless alcoholic solvents are used, but this can lead to N-
alkylation as a side reaction; using trifluoroethanol as the solvent overcomes this problem, and leads to highly reliable hydrogenolyses
for a wide range of substrates, including the final deprotection step in our total synthesis of (À)-raumacline.
Ó 2008 Elsevier Ltd. All rights reserved.
We have recently completed the total synthesis of (À)-
raumacline (Scheme 1).1 During this work, we encountered
an unexpected N-alkylation by the alcoholic solvent when
utilizing catalytic hydrogenolysis for the final N-debenzyl-
ation step. We believe this observation may be more wide-
spread than is generally acknowledged, because the spectra
of alkylated and non-alkylated derivatives are so similar,
and the former may be present to the extent of only a
few %. We report herein our solution to this problem,
our exploration of its wider applicability, and some impor-
tant observations when applied to our synthesis of (À)-rau-
macline and related alkaloids.
4 steps
CN
NH2
L-tryptophan
N
H
1
Ref. 2
2
Cis -selective
Pictet-Spengler
Note 3
CN
Bn
5 steps
CN
NH
N
N
Me
N
H
3
4
Et
CO2Me
OSiPh2But
H
LiNEt2
NC
The synthetic problem related to the final step in our
synthesis of (À)-raumacline (7?8 in Scheme 1), in which
we simply followed the deprotection procedure published
by Cook and Fu.5,6 All of the data for our intermediates
in the synthesis of raumacline were completely consistent
with our reported structures, up to and including the pro-
tected precursor 7. The final deprotection failed to take
place in non-alcoholic solvents (e.g., ethyl acetate, dichlo-
romethane, THF); Cook had reported that hydrogenolyses
O
H
a) LiBH4
b) H+ / heat
N
O
N
N
Me H
Bn
N
Me
Bn
CO2Me
H
5
Et
6
Et
DIBAL-H
H
H
OH
H
OH
H
See text
(Ref. 4)
O
NH
N
O
N
Me H
N
Me H
Bn
H
H
Et
Et
8
7
Scheme 1. Our total synthesis of (À)-raumacline.1 (See above-mentioned
*
Corresponding author. Tel.: +44 1782 584583.
E-mail address: p.bailey@natsci.keele.ac.uk (P. D. Bailey).
references for further information.)
0040-4039/$ - see front matter Ó 2008 Elsevier Ltd. All rights reserved.
doi:10.1016/j.tetlet.2008.01.104

P. D. Bailey et al. / Tetrahedron Letters 49 (2008) 2150–2153
2151
Pd
H
R
H
OH
R
R
H
H
c
a
t.
7/ H2/
Pd-C/ ROH
N
O
O
O
N
R
H
Me H
1
2
H
N
R R NH
Et
8: R = H
OH
H
+ 9: R = Me (MeOH)
+10: R = Et (EtOH)
O
N
Bn
R
H / Pd-C
2
Me H
H
H
N
N
Et
7
OEt
R1
R2
R1
R2
H
H
.
7 HCl salt/
H2/ Pd-C/ EtOH
N
O
N
Me H
Bn
Scheme 3. Mechanism for alkylation.
Et
11
HO
HO
Scheme 2. Attempted deprotections of 11.
a)
N
NH
Et
N
N
OH
H
H
Me
Et
Me
OH
in methanol could be accompanied by N-methylation,6,7
and indeed, hydrogenolysis of 7 in methanol yielded the
N-methyl derivative 9 (Scheme 2), and we also observed
N-methylation in similar debenzylations.8
ajmaline (14)
4,21-secoajmaline (15)
b)
H
OH
Cook apparently overcame the problem by using
ethanol as the solvent6,7 but, although this reduced the
amount of alkylation, we nevertheless observed significant
amounts (ca. 25%) of N-ethyl derivative 10. We also tried
hydrogenolysis of the HCl salt of 7 in ethanol (the condi-
tions reported by Cook and Fu6,7), but this generated the
ethoxy acetal 11 in our hands, without achieving debenzyl-
ation (Scheme 2).
H
O
NH
N
Me
H
H
Et
raumacline (8)
Scheme 4. Semi-synthesis of (À)-raumacline. Reagents and conditions: (a)
NaBH4, (26.8 equiv), MeOH, citrate/NaOH buffer, pH 6.0, 4 °C, 24 h,
100%; (b) (À)-riboflavin, hm, MeOH, acetone, citrate/NaOH buffer pH 6.0,
4 °C, 24 h, 78% (see Ref. 10).
The explanation suggested by Cook for the N-methyl-
ation in methanol seems a reasonable one,6,7 in which
the catalyst initially oxidizes the alcohol to the aldehyde,
which can then be reductively trapped by the free amine
(Scheme 3).
One possible solution, we felt, would be to conduct the
hydrogenolysis in an alcoholic solvent for which the corre-
sponding aldehyde was less stable, hopefully raising the
activation barrier for dehydrogenation of the alcohol,
and thereby blocking the alkylation pathway. We reasoned
that trifluoroethanol (TFE) might meet our needs, with the
fluorinated aldehyde 13 destabilized by the ÀI effect of the
CF3 group, relative to non-fluorinated ethanal 12.
As can be seen in Figure 1, the 1H NMR spectra of semi-
synthetic raumacline, and our totally synthetic material,
were very different. However, it transpired that traces of
acid accounted for the differences, as a base wash led to
superimposable spectra. Although acids are well known
to modify the spectra of basic compounds such as alkal-
oids, such a dramatic effect is rare.
H
H
δ+
δ+
>
H3C
<
F3C
O
δ−
O
δ−
12
13
This approach was completely successful, providing (À)-
raumacline 8 from 7 in 100% yield, with no hint of N-
alkylation.
Although the final outcome was the successful con-
version of 7?8, the 1H NMR spectrum of synthetic 8
was initially so different from that reported for the natural
product6,7,9 that we assumed a completely different com-
pound had been formed. We, therefore, set about obtaining
(À)-raumacline from natural sources, for comparison; we
chose to prepare semi-synthetic 8 from ajmaline, using a
slightly modified procedure to that employed by Sto¨ckigt’s
group,10 as shown in Scheme 4.
Fig. 1. 1H NMR spectra of (À)-raumacline 8: (a) our synthetic 8 after base
wash; (b) semi-synthetic 8 (Sto¨ckigt method, Scheme 4); (c) our synthetic 8
before base wash.

2152
P. D. Bailey et al. / Tetrahedron Letters 49 (2008) 2150–2153
The key observations are that debenzylation of anilines
H
OH
H
17
H17
H15
H6a
usually proceed smoothly in any alcoholic solvent, whilst
tetrahydroquinolines and tetrahydro-b-carbolines can be
more problematic, and may depend on the activity of the
specific batch of catalyst being used. For these latter sys-
tems, not only does TFE provide a faster and more reliable
deprotection, but it also avoids the potential alkylation of
the nitrogen—a problem that may be occurring (perhaps
only to the extent of a few %) without people being aware
of it (see Ref. 8). Some surprising examples of selectivity
were also observed. For example, mono-debenzylation of
20 occurred very efficiently. Moreover, the modified condi-
tions surprisingly allowed a highly efficient N-debenzyla-
tion in the presence of an O-benzyl group (entry h,
22?23); this selectivity may be of more widespread use,
and we are exploiting it in the final stages of the synthesis
of ajmaline-related alkaloids. Finally, of course, our condi-
tions have allowed the clean and efficient deprotection of 7,
thereby allowing us to obtain completely pure (À)-rauma-
cline at the end of our total synthesis.
H6b
HO
H21b
H21a
5
NH
O
16
15
3
N
21
N
O
Me
20
14
H
H
19
H5
HN
18
19
18
H14b
H20
H3
H16
H14a
[α]D23 −28.8 (c 1.25, CHCl3)
[α]D20 −26 (c 0.1, CHCl3)6,7
Fig. 2. Key data for our synthetic (À)-raumacline 8.
To absolutely confirm the stereochemistry and optical
integrity of our synthetic (À)-raumacline, extensive 2D
NMR experiments were run, and key NOE and optical
rotation results are shown in Figure 2.
Finally, we carried out a limited study on the wider
scope/limitations of conducting catalytic hydrogenolysis
in a range of alcoholic solvents, including trifluoroethanol,
and these results are summarized in Table 1.
Table 1
Catalytic hydrogenolyses over 10% Pd–C for a range of N-benzylated compounds; for entries 16–21, the time for complete deprotection (confirmed by 1H
NMR) is given in each of three solvents (X = no reaction)
Entry
OMe
MeOH
EtOH
TFE
>2 ha
30 min
10 min
16
17
NHBn
CO2Me
30 min
15 min
24 h
NHBn
O
b
70% after 48 h
24 h
2 h
18
N
Bn
X
X
X
X
X
24 h
19
20
NBn
3 h monoc
NBn2
NBn
N
H
21
X
5 h
N
H
H
N
H
N
H2/ Pd-C/ TFE
(85% isolated yield)
CN
CN
22
23
OBn
N
Me
Bn
N
Me
H
H
OBn
H
a
Notes: 70% hydrogenolysis was seen after 2 h.
b
100% C@C bond reduction after 24 h in MeOH, after 2 h in EtOH and after 1 h in TFE.
Presence of fully deprotected product (ca. 5%) after 3 h, increasing to ca. 20% after 24 h (identified by multiplet at d 2.8).
c

P. D. Bailey et al. / Tetrahedron Letters 49 (2008) 2150–2153
7. Fu, X.; Cook, J. M. J. Org. Chem. 1993, 58, 661–672.
2153
Acknowledgement
8. We have observed N-alkylation in other cases, as exemplified by the
deprotection of the Weinreb amide X, for which mass spectrometric
evidence indicated the presence of about 15% of the N-methyl
by-product:
We thank the EPSRC for funding.
References and notes
H
1. Bailey, P. D.; Clingan, P. D.; Mills, T. J.; Price, R. A.; Pritchard, R.
G. Chem. Commun. 2003, 2800–2801.
2. Kutney, J. P.; Eigendorf, G. K.; Matsue, H.; Murai, A.; Tanaka, K.;
Sung, W. L.; Wada, K.; Worth, B. R. J. Am. Chem. Soc. 1978, 100,
938–943.
3. Alberch, L.; Bailey, P. D.; Clingan, P. D.; Mills, T. J.; Price, R. A.;
Pritchard, R. G. Eur. J. Org. Chem. 2004, 1887–1890.
4. Bailey, P. D.; Hollinshead, S. P.; McLay, N. R.; Morgan, K.; Palmer,
S. J.; Prince, S. N.; Reynolds, C. D.; Wood, S. D. J. Chem. Soc.,
Perkin Trans. 1 1993, 431–439.
O
H
OMe
Me
N
N
N
R
Me H
H
OH
X: R = Bn
H2/
Pd-C/
MeOH
R = H [M+H]+ 358 (100%)
R = Me [M+H]+ 372 ( 16%)
5. When we completed our synthesis of (À)-raumacline,1 we assumed
that the final deprotection step following the procedure of Fu and
Cook6,7 would cleanly yield the desired product 8. Only on
subsequent scaleup did we identify the problem of alkylation as a
side reaction.
9. Polz, L.; Sto¨ckigt, J.; Takayama, H.; Uchida, N.; Aimi, N.; Sakai, S.
Tetrahedron Lett. 1990, 31, 6693–6696.
10. Endress, S.; Sto¨ckigt, J. Helv. Chim. Acta 1993, 76, 2544–2546.
6. Fu, X.; Cook, J. M. J. Am. Chem. Soc. 1992, 114, 6910–6912.