Total Synthesis of (+/-)-Roserine

Article abstract of DOI:10.1021/jo971641o

Electrophilic cyclization of amino ketals 14a and 14b with 6 M HCl at 23 deg C, followed by oxidation with air in 6M HCl at 100 deg C, gave tetrahydrophenanthridines 16a and 16b, respectively, which were transformed into the pyrrolophenanthridinium salts

Full text of DOI:10.1021/jo971641o

J . Org. Chem. 1997, 62, 9305-9313
9305
Tota l Syn th esis of (()-Roser in e†
Aaron E. Bunnell, Lee A. Flippin,* and Yanzhou Liu
Department of Medicinal Chemistry, Neurobiology Unit, Roche Bioscience, 3401 Hillview Avenue,
Palo Alto, California 94304-1397
Received September 3, 1997X
Electrophilic cyclization of amino ketals 14a and 14b with 6 M HCl at 23 °C, followed by oxidation
with air in 6 M HCl at 100 °C, gave tetrahydrophenanthridines 16a and 16b, respectively, which
were transformed into the pyrrolophenanthridinium salts 1a and 1b. Compound 1b was shown to
be the correct structure of the Narcissus pallidulus constituent, roserine.
In tr od u ction
Roserine is a novel pyrrolophenanthridinium alkaloid
recently isolated in small quantity from Narcissus pal-
lidulus by Codina et al.1a Structure 1a was originally
proposed for roserine on the basis of its high-resolution
1
mass spectrum, routine H and 13C NMR spectra, and a
1H-1H COSY spectrum; however, three additional iso-
meric pyrrolophenanthridinium structures 1b-d , all of
which are compatible with much of the spectroscopic
data, were implicitly rejected “on the basis of the H(6)
and H(7) correlation observed in a complete COSY
experiment” (Figure 1).1b Although the structure pro-
posed for roserine contains an asymmetric center at C(4),
there is currently no evidence for an absolute configura-
tion of the natural material or, indeed, whether the
alkaloid is even optically active.
Two structural features of 1a appear to be unusual
upon comparison with related natural products. With
the single exception of roserine, all 16 or so pyrrolo-
phenanthridinium alkaloids currently known from plant
sources2a contain a fully aromatic C-ring. It is also
notable that within the broader class of Amaryllidaceae
phenanthridine alkaloids, which includes a variety of
pyrrolophenanthridines and pyrrolophenanthridones struc-
turally related to the known pyrrolophenanthridinium
salts, several members exhibit a pattern of three contigu-
ous oxidized carbon atoms in the aromatic A-ring;
however, all such examples appear to be 7,8,9-trisubsti-
tuted in contrast to the 8,9,10-trimethoxy substitution
pattern that has been proposed for roserine.2b
F igu r e 1. Numbering scheme1 and isomeric candidates (1a -
d ) for the structure of roserine.
A number of pyrrolophenanthridinium alkaloids were
investigated for their potential utility as antineoplastic
agents by Cheng et al.3 A notable example from this
structural class, ungeremine, showed low toxicity coupled
with remarkable efficacy during preclinical testing.3a,4
Ungeremine was advanced to human clinical trials in the
People’s Republic of China where it reportedly showed
good overall efficacy against ovarian and uterine cervix
carcinomas but relatively poor activity against gastric
and “miscellaneous” carcinomas.5 Thus, a clear rationale
exists for the continued preclinical SAR (structure-
activity relationship) study of novel and structurally
diverse pyrrolophenanthridinium alkaloids.
The novelty of roserine’s reported structure, its appar-
ent scarcity from a single natural source, and the need
for an assessment of its biological activity make total
synthesis an attractive route for the procurement of
useful amounts of this natural product. However, the
† Dedicated to Kenneth O. Hartley on the occasion of his 83rd
birthday.
* To whom correspondence should be addressed. tel: 415-354-7847.
fax: 415-354-7363. e-mail: lee.flippin@roche.com.
(1) (a) Bastida, J .; Codina, C.; Viladomat, F.; Rubiralta, M.; Quirion,
J .-C.; Weniger, B. J . Nat. Prod. 1992, 55, 134. (b) The numbering
system used by Codina et al., was adopted in this paper to facilitate
discussion and comparison of our results. A widely accepted alternative
numbering scheme for the pyrrolo[3,2,1-de]phenanthridinium ring
system may be found in The Ring Systems Handbook; American
Chemical Society: Columbus, OH (1993). coden: RSHAEE 1629RSF-
3262RSF (1993).
(2) (a) CAS Online substructure search. (b) The Dictionary of
Natural Products (CD-ROM version); Chapman & Hall: London, 1996.
(3) (a) Zee-Cheng, R. K.-Y.; Yan, S.-J .; Cheng, C. C. J . Med. Chem.
1978, 21, 199. (b) Samoilova, M. Ye.; Akhvlediani, R. N.; Frolova, Ye.
P.; Kurkovskaya, L. N.; Yershova, Yu. A.; Safonova, T. S.; Korovin, B.
V.; Suvorov, N. N. Khim.-Farm. Zh. 1994, 28, 10.
(5) (a) Xu, B.; Wang, X.-W. Drugs Future 1997, 22, 123. (b) Fei, C.;
Liu, G. Z.; Dong, M. G.; Hua, Z. D.; Wu, Q. F.; Li, S. J .; Hu, L. Y. New
Drugs Clin. Rem. (China) 1985, 4, 1.
(4) Lu, D. Y.; Xu, J .; Xu, B.; Zhou, A. M. Chin. Biochem. J . 1993, 9,
626.
S0022-3263(97)01641-1 CCC: $14.00 © 1997 American Chemical Society

9306 J . Org. Chem., Vol. 62, No. 26, 1997
Bunnell et al.
Sch em e 1a
nonaromatic C-ring of roserine presents a synthetic
challenge that is not readily amenable to any of the
currently available methods for construction of the pyr-
rolophenanthridinium skeleton.6 In this report, we de-
scribe (1) a total synthesis of structure 1a , (2) correction
of the structure proposed by Codina et al. for roserine,
and (3) the first total synthesis of structure 1b and its
spectroscopic correlation with natural roserine.
carried out in 6 M HCl starting at 100 °C in the presence
of air, tetrahydrophenanthridine 6 was the major product
but significant amounts of mono-demethylated tetrahy-
drophenanthridines 7 and 8 were also formed. Limited
attempts to selectively aromatize the heterocyclic ring
of compound 5 did not give satisfactory results. Thus,
isolated 5 reacted with 20% Pd-C (decalin, reflux, 4 h)
to give fully aromatized 8,9,10-trimethoxyphenanthri-
dine, 9, in 74% yield, while addition of 1 equiv of DDQ
to crude 5 in aqueous 6 M HCl gave a complex product
mixture containing only a small amount of compound 6.
Ultimately, a two-stage reaction sequence performed
without isolation of intermediate 5 was found to be
optimal for the preparation of 6. Thus, compound 4 was
allowed to stand in 6 M HCl at room temperature under
a nitrogen atmosphere for 20 h; the nitrogen atmosphere
was replaced with air, and the rapidly stirred mixture
was heated to 100 °C for 2 h to give compound 6 in 48%
yield. These reaction conditions, which minimize expo-
sure of the final product to hot 6 M HCl, appeared to
completely avoid detectable formation of side-products 7
and 8.
Resu lts a n d Discu ssion
Our initial approach to 1a was predicated on the
assumption that 8,9,10-trimethoxy-1,2,3,4-tetrahydro-
phenanthridine, 6, might be induced to undergo directed
lateral metalation at C(1) and subsequent functionaliza-
tion with a two-carbon D-ring fragment.7 The prepara-
tion of intermediate 6 is shown in Scheme 1. Amino ketal
28 was used in a two-step reductive amination of 3,4,5-
trimethoxybenzaldehyde to give amino ketal 4 in 80%
overall yield. Aqueous 6 M HCl at room temperature,
which has previously been described for the conversion
of electron-rich N-benzyl aminoacetaldehyde acetals to
4-hydroxy-1,2,3,4-tetrahydroisoquinolines,9 promoted the
cyclization of amino ketal 4 to hexahydrophenanthridine
5 in 56% yield. Tetrahydrophenanthridine 6, probably
formed by reaction of 5 with adventitious traces of
oxygen, was also isolated in 7% yield from this reaction
mixture. When the cyclization-aromatization of 4 was
Tetrahydrophenanthridine 6 requires only two ad-
ditional carbon atoms to complete the skeleton of target
1a ; however, we were unable to laterally metalate and
alkylate 6 at C(4) despite several attempts to do so.10
(10) Tetrahydrophenanthridine 6 proved highly resistant toward
metalation with a variety of strong, nonnucleophilic bases. Thus, THF
solutions of LDA or LTMP and ether solutions of LTMP-KOtBu or
mesityllithium all failed to generate a colored anion from 6 over 1-3
h between -78 °C and +23 °C. After quenching these reaction mixtures
(6) (a) Rosa, A. M.; Lobo, A. M.; Branco, P. S.; Prabhakar, S.; Sa-
da-Costa, M. Tetrahedron 1997, 53, 299. (b) Xiong, Y.; Moore, H. J .
Org. Chem. 1996, 61, 9168. (c) Hutchings, R. H.; Meyers, A. I. J . Org.
Chem. 1996, 61, 1004. (d) Parnes, J . S.; Carter, D. S.; Kurz, L. J .;
Flippin, L. A. J . Org. Chem. 1994, 59, 3497 and references therein. (e)
Lauk, U.; Durst, D.; Fischer, W. Tetrahedron Lett. 1991, 32, 65. (f)
See also, Banwell, M. G.; Wu, A. J . Chem. Soc., Perkin Trans. 1 1994,
2671.
(7) For a general discussion of this topic, see Clark, R. D.; J ahangir,
A. Org. Reac. 1995, 47, 30-33.
(8) (a) Driguez, H.; Vermes, J . P.; Lessard, J . Can. J . Chem. 1978,
56, 119. (b) We thank Mr. David Repke for the large-scale preparation
of amino ketal 2 by an unpublished method. Thus, 2-chlorocyclohex-
with
a two-carbon fragment (ethylene oxide) or methyl iodide to
determine possible sites of metalation, the only detectable tetrahy-
drophenanthridine in each crude product mixture was unreacted 6.
Treatment of 6 with n-BuLi in THF followed by a methyl iodide quench
gave a very complex mixture comprised in part from direct addition of
an n-butyl group to the CdN bond; however, no recognizable methyl
group incorporation, aside from N-methylation, could be detected from
the 1H NMR spectrum of the crude reaction mixture. The metalation
of 3-methylisoquinoline was briefly examined as a potential model
system for the metalation of 6; however, in distinct contrast to 6,
3-methylisoquinoline readily gave a deep red-colored anion with LTMP
(THF, -78 °C). This anion was quenched with methyl iodide (-78 °C)
to give 3,4-dimethylisoquinoline rather than the expected product,
3-ethylisoquinoline (see also, Gisby, G. P.; Sammes, P. G.; Watt, R. A.
J . Chem. Soc., Perkin Trans. 1 1982, 249).
anone was converted to
2 in 85% yield via azide substitution,
ketalization, and reduction of the azido group.
(9) (a) Oppolzer, W.; Robbiani, C. Helv. Chim. Acta 1983, 66, 1119.
(b) Bobbit, J . M.; Sih, J . C. J . Org. Chem. 1968, 33, 856. (c) Euerby,
M. R.; Waigh, R. D. J . Chem. Res., Synop. 1987, 2, 36.

Total Synthesis of (()-Roserine
J . Org. Chem., Vol. 62, No. 26, 1997 9307
Sch em e 2a
We next investigated an approach to structure 1a that
would potentially allow incorporation of the C(11)-C(12)
D-ring fragment at an earlier stage in the synthesis via
elaboration of ketone 1011 (Scheme 2). The mild reaction
conditions used for the preparation of aldimine 3 gave
no apparent reaction between ketone 10 and 3,4,5-
trimethoxybenzylamine, while several attempts to gener-
ate an imine from these reactants under Dean-Stark
conditions (toluene, reflux, molecular sieves or catalytic
p-TsOH) gave complex product mixtures which were not
investigated further. However, when ketone 10 and
3,4,5-trimethoxybenzylamine were allowed to react ac-
cording to the method of Bhattacharyya12 in neat Ti-
(OiPr)4 followed by reduction with methanolic NaBH4, the
resulting crude mixture appeared to be composed pri-
marily of reductive amination products. After careful
analysis this product mixture was shown to consist of
ethyl esters 12a and 13a (4:1 ratio) and isopropyl esters
13a and 15a (4:1 ratio); the ratio of total ethyl esters
(12a +13a ) to isopropyl esters (14a +15a ) was approxi-
mately 3:2. The partial transesterification of 10 during
imine formation could be avoided by replacing Ti(OiPr)4
with Ti(OEt)4 in Bhattacharyya’s procedure; however, the
use of Ti(OEt)4 caused the reaction mixture to solidify
after a short time, and, following reduction with NaBH4-
EtOH, the isolated yield of diastereomeric esters 12a and
13a (4:1 ratio) was poor. A single attempt to conduct the
imine formation using 0.29 M Ti(OEt)4 in abs EtOH,
followed by reduction with ethanolic NaBH4, also gave
12a and 13a in poor yield. Finally, 2,2-dimethoxycyclo-
hexanone was alkylated with isopropyl iodoacetate13 to
give 1,2-cyclohexanedione monoketal 11 in 68% yield.
Treatment of 13 with 3,4,5-trimethoxybenzylamine in
neat Ti(OiPr)4 followed by addition of ethanolic NaBH4
afforded the cis and trans diastereomers 14a and 15a in
a 4:1 ratio. However, when the reduction step of the
procedure was carried out with a slurry of NaBH4 in
2-propanol, the 14a :15a diastereomer ratio improved to
10:1. Employing the latter method, pure cis diastereomer
14a was isolated in 59% yield and used for all subsequent
work toward the synthesis of target structure 1a . Treat-
ment of 14a with 6 M HCl at room temperature under a
nitrogen atmosphere for 14 h, followed by introduction
of an air atmosphere and warming the reaction mixture
to 100 °C for 1 h, gave tetrahydrophenanthridine 16a in
60% yield. Reduction of 16a with borane-methyl sulfide
complex (BMS) in THF at room temperature for 16-20
h gave the desired primary alcohol 17a in 40-55% yield;
however, this problematic transformation was accompa-
nied by formation of an incompletely characterized over-
reduction product when sufficient BMS was used to
ensure complete consumption of the starting material.
Several attempts to optimize the lithium aluminum
hydride reduction of 16a in ether also gave 17a in 40-
50% yield. Eventually it was found that careful treat-
ment of 16a with 1.05 equiv of BMS in THF minimized
over-reduction while reproducibly giving 55% of alcohol
17a and 35% of a recyclable ethyl ester, 18a , after
workup of the reaction mixture with ethanolic HCl.
Reaction of 17a with a large excess of PBr3 in refluxing
benzene for 2.5 h14 gave pyrrolophenanthridinium bro-
mide 1a (Br-) in fair yield; however, this procedure was
significantly improved by the use of 1 equiv of either PCl3
or PBr3 in refluxing benzene for 30 min to give 1a (Cl-)
or 1a (Br-) in 76 and 87% yield, respectively.
(11) Compound 10 has been previously mentioned as
reaction product without physical characterization: Langschwager, W.;
Hoffmann, H. M. R. Liebigs Ann 1995, 797. Note: There is
a minor
a
typographical error in the experimental section of this paper that
matches the compound number for ethyl (3,3-dimethoxy-2-oxocyclo-
hexyl)acetate with physical properties of ethyl (3,3-diethoxy-2-oxocy-
clohexyl)acetate.
(12) (a) Bhattacharyya, S. Tetrahedron Lett. 1994, 35, 2401. (b) See
also Seebach, D.; Hungerbuhler, E.; Naef, R.; Schnurrenberger, P.;
Weidmann, B.; Zuger, M. Synthesis 1982, 138.
(13) Chuchani, G.; Hernandez, J . A.; Yepez, M.; Alonso, M. E. Int.
J . Chem. Kinet. 1977, 9, 811.
(14) These reaction conditions were previously described for
related D-ring closure in a synthesis of vasconine: See ref 6a.
a

9308 J . Org. Chem., Vol. 62, No. 26, 1997
Ta ble 1. Effect of Solven t a n d Cou n ter ion on 13C NMR Sh ifts of 1a a
Bunnell et al.
1a (Br-)
1a (Cl-)
carbon
number
1:1
1:1
1a
∆ (ppm)
1a m, lit.e
CDCl3-CD3ODc
CD3ODc
CDCl3
CDCl3-CD3OD
CD3OD
mean(s)d
b
CDCl3
C(8)
C(9)
155.3
150.4
149.2
144.3
142.0
129.0
128.9
125.3
106.6
61.7
61.3
57.3
57.0
40.9
31.8
26.5
26.0
22.8
156.7
151.7
150.4
145.7
141.2
130.7
130.2
126.3
106.0
62.4
62.0
57.9
57.1
41.9
32.4
27.2
27.0
23.5
157.3
152.2
150.1
146.7
142.2
131.0
130.6
126.9
106.5
62.5
62.0
58.5
57.2
42.5
32.8
27.7
27.5
24.1
155.6
150.8
149.4
144.2
143.3
129.0
128.8
125.7
107.1
61.9
61.6
57.2
57.1
41.2
32.0
26.8
26.3
23.1
156.8
151.8
150.5
145.8
141.5
130.7
130.2
126.4
106.1
62.4
62.0
58.0
57.1
42.0
32.4
27.3
27.1
23.6
157.3
152.2
151.1
146.8
142.2
131.0
130.6
126.9
106.6
62.5
62.0
58.4
57.2
42.5
32.7
27.7
27.5
24.1
156.5(0.9)
151.5(0.8)
150.1(0.7)
145.6(1.1)
142.1(0.7)
130.2(1.0)
129.9(0.8)
126.3(0.6)
106.5(0.4)
62.2(0.3)
61.8(0.3)
57.9(0.5)
57.1(0.1)
41.8(0.7)
32.4(0.4)
27.2(0.5)
26.9(0.6)
23.5(0.5)
-5.9
1.0
3.3
3.6
3.6
-5.3
2.8
7.0
9.7
-0.3
-0.6
0.5
0.3
1.8
0.7
0.4
3.9
1.7
C(10)
C(4a)
C(6)
C(10a)
C(10b)
C(6a)
C(7)
10-MeO
9-MeO
C(12)
8-MeO
C(4)
C(11)
C(3)
C(1)
C(2)
a
b
Solute concentration was 0.04 M. Referenced to the center peak of CDCl3 (δ ) 77.2). c Referenced to the center peak of CD3OD
(δ ) 49.2). Mean (standard deviation); N ) 6. e Roserine values from reference 1.
d
A comprehensive H and 13C NMR characterization of
structure 1a was carried out using 1a (Br-) in CDCl3
solution. The position and relative configuration of the
aliphatic protons of rings C and D were assigned on the
basis of 1H-1H coupling and NOE correlations observed
with 2D COSY and NOESY experiments.15 The NOESY
spectrum of 1a (Br-) also showed, as expected, strong
NOE’s for H(6) with H(7) and H(12R,â); however, the
COSY spectrum did not show a detectable correlation
between H(6) and H(7). Proton H(7) also showed an NOE
with the singlet at δ 4.09 allowing assignment of the
8-methoxy group, while an NOE between both H(1)
protons and the singlet at δ 3.97 was used to establish
the 10-methoxy group. The 13C NMR spectrum of 1a
(Br-) showed a discreet resonance for each carbon atom
while an APT experiment gave the number of protons
attached to each. The protonated carbons were assigned
by their chemical shift values and one-bond heteronuclear
(13C-1H) correlations were obtained from a 2D HETCOR
experiment. The unambiguous assignment of all carbon
atoms of 1a (Br-), including the quaternary centers, was
carried out by analysis of long-range 2D COLOC and 2D
HMBC experiments optimized for three bonds. Solvent
effects on the 1D proton and carbon NMR spectra of 1a
(Cl-) and 1a (Br-) were studied using CDCl3, CD3OD,
and 1:1 CDCl3-CD3OD solutions. The 13C NMR spectra
of 1a (Table 1) showed little solvent-dependence or
counterion-dependence; thus, the standard deviation, s,
for every 13C resonance of 1a (2 counterions × 3 solvents;
N ) 6 for each resonance line) was e1.1 ppm. In
to be reliable diagnostic signals for a partial comparison
with the published spectral data for roserine. The
absence of significant solvent-induced shifts in the 13C
NMR spectra of 1a suggested that all of these data would
be crucial for an unambiguous synthetic confirmation of
roserine’s identity. Direct comparison of the solvent-
independent portions of the 1H NMR spectra of 1a , as well
as the entire set of 13C NMR data for 1a , with the
corresponding published spectral data for roserine clearly
revealed that they are isomeric structures.
1
Several lines of evidence suggested the likely identity
1
of roserine with structure 1b. The 1D H and 13C NMR
data reported for roserine are consistent with the general
skeletal features of all four structures 1a -d , whereas the
8,9,10-trimethoxy pattern of structure 1a was specifically
proposed for roserine on the basis of a 1H-1H COSY
experiment which suggested coupling between proton
H(6) at δ 9.50 and the proton at δ 6.95 (assigned as H(7)).
Codina et al. apparently did not attempt to confirm the
1
H(7) assignment with an NOE experiment.17 A H-1H
COSY spectrum of authentic 1a showed no detectable
correlation between H(6) and H(7) (vide supra); however,
this result is not very surprising in light of the corre-
sponding 1H-1H coupling information known for iso-
quinoline18 (Figure 2).
There is no apparent coupling between H(1) and H(6),
H(7), or H(8) in isoquinoline: The only readily observable
1H-1H coupling constant for H(1) with an A-ring proton
appears to be J H(1)-H(5) ) 0.9 Hz. An additional example
of transannular 1H-1H coupling that is especially rel-
evant to the isoquinolinium moiety of roserine is provided
by the related pyrrolophenanthridinium alkaloid vasco-
nine6d (Figure 2). Protons H(7) and H(10) of vasconine
were unambiguously assigned from a NOESY experiment
but, while the 1H-1H COSY spectrum shows a clear
H(6)-H(10) correlation, there is no evidence of an H(6)-
H(7) correlation in this alkaloid. We inferred from these
data that, of the four possible isomers 1a -d , only
structure 1b might be confidently expected to exhibit a
coupling relationship between H(6) and an A-ring proton;
therefore, 1b is an appropriate candidate for the correct
1
contrast, the H NMR signals assigned to H(6), H(7), and
H(12R) of structure 1a were observed to be strongly
solvent-dependent.16 Halide counterion effects on the 1H
NMR spectra of 1a were negligible over the range of
solvents examined.
Aside from the three important signals mentioned
above, the 1H NMR spectra of 1a were essentially
constant over the range of solvents studied; thus, the
isolated multiplets assigned to H(1R,â), H(3R), H(4),
H(11R,â), and the three methoxy singlets seemed likely
(15) The relative configuration of protons in structures 1a and 1b
were determined experimentally relative to H(4). Proton H(4) in 1a
and 1b is arbitrarily assigned the â-configuration in Figure 2.
(16) Complete details are available in the Supporting Information.
(17) Bastida, J . Personal communication: May 28, 1997.
(18) Kook, A. M.; Smith, S. L.; Brown, E. V. Org. Magn. Reson. 1984,
22, 730.

Total Synthesis of (()-Roserine
Ta ble 2. Effect of Solven t a n d Cou n ter ion on 13C NMR Sh ifts of 1ba
1b (Br-) 1b (Cl-)
J . Org. Chem., Vol. 62, No. 26, 1997 9309
carbon
number
1:1
1:1
1b
∆ (ppm)
1bm, lit.e
CDCl3-CD3ODc
CD3ODc
CDCl3
CDCl3-CD3OD
CD3OD
mean(s)d
b
CDCl3
C(9)
C(7)
C(4a)
C(8)
C(6)
C(10a)
C(10b)
C(6a)
C(10)
7-MeO
8-MeO
C(12)
9-MeO
C(4)
162.3
150.6
146.7
142.1
138.7
135.5
127.0
119.3
97.0
62.9
61.5
57.8
57.1
163.6
150.8
147.4
142.6
137.9
136.4
128.4
119.9
97.8
62.7
62.0
57.8
57.5
164.2
151.6
148.4
143.4
138.9
137.2
129.0
120.5
98.8
63.0
62.1
58.4
57.7
162.2
150.5
146.6
142.1
140.0
135.4
126.8
119.4
96.8
62.8
61.6
57.7
56.9
163.7
150.8
147.4
142.6
138.0
136.5
128.4
119.9
97.9
62.7
62.1
57.8
57.5
164.2
151.5
148.3
143.4
138.9
137.2
129.0
120.5
98.8
63.0
62.1
58.3
57.7
163.4(0.9)
151.0(0.5)
147.5(0.8)
142.7(0.6)
138.7(0.8)
136.4(0.8)
128.1(1.0)
119.9(0.5)
97.9(0.9)
62.9(0.1)
61.9(0.3)
58.0(0.3)
57.4(0.3)
41.0(0.7)
32.5(0.4)
27.8(0.5)
23.9(0.4)
22.7(0.6)
1.0
0.5
0.7
0.7
0.2
0.9
1.0
0.6
1.1
0.4
-0.5
0.6
0.6
1.0
0.8
1.0
0.9
0.9
40.3
32.0
27.2
23.4
41.0
32.5
27.7
23.9
41.7
33.0
28.3
24.3
40.2
32.0
27.2
23.3
41.1
32.5
27.8
23.9
41.7
32.9
28.3
24.3
C(11)
C(3)
C(1)
C(2)
22.1
22.6
23.4
22.1
22.7
23.4
a
b
Solute concentration was 0.04 M. Referenced to the center peak of CDCl3 (δ ) 77.2). c Referenced to the center peak of CD3OD
(δ ) 49.2). Mean (standard deviation); N ) 6. e Roserine values from reference 1.
d
solution, revealed that the 13C NMR spectra of 1b are
essentially independent of counterion or solvent effects
(Table 2). The standard deviation for each 13C resonance
of 1b (2 counterions × 3 solvents; N ) 6 for each line)
was s e 1.0 ppm. We were therefore gratified to observe
that the mean value for each 13C spectral line of 1b
precisely matched a corresponding spectral line reported
for natural roserine (|δ1b(mean) - δroserine| e 1.1 ppm for
each line). When the 13C NMR spectral lines reported
for natural roserine in CDCl3 were included in a statisti-
cal calculation with the corresponding values for 1b (Cl-
and Br-) recorded only in CDCl3 (N ) 3 for each of the
18 lines), the standard deviation was s e 0.8 ppm for each
line, while the mean of all the standard deviations (smean
)
was only 0.2 ppm.
1
Not surprisingly, the H NMR spectra of 1b showed
marked solvent-dependent behavior for H(6), H(10), and
H(12R), while the chemical shifts for all other proton
resonances were virtually constant over the range of our
solvent study.16 Aside from the three highly solvent-
variable signals, the chemical shift and multiplicity (or
signal bandwidth in the case of unresolved multiplets)
of every resonance in the 1H NMR spectrum of 1b
correlated with a matching signal reported for roserine.19
F igu r e 2. Selected 1H-1H coupling and NOE relationships:
See text for discussion.
structure of roserine. Structures 1a , 1c, and 1d also
share the structural motif of a C(10) methoxy substituent
proximal to H(1R) and H(1â) of the saturated C-ring. In
structure 1a this substitution pattern leads to large,
solvent-independent downfield shifts in H(1â), ∆δ ) 0.6
ppm, and H(1R), ∆δ ) 0.35 ppm, relative to the H(1R,â)
signals reported for roserine. It seemed likely that the
corresponding methylene protons of structures 1c and 1d
would be similarly deshielded, whereas the H(1) meth-
ylene protons of 1b would be shielded relative to 1a .
Fortunately, the optimized synthetic approach that
afforded the bromide and chloride salt forms of compound
1a also proved entirely satisfactory for a synthesis of both
salt forms of structure 1b. Thus, reductive amination
of ketone 11 with 2,3,4-trimethoxybenzylamine gave cis
diastereomer 14b in 73% isolated yield (crude mixture
cis-14b:trans-15b ) 13:1), and the three-step transfor-
mation of 14b into pyrrolophenanthridinium salts 1b
(Cl-) and 1b (Br-) proceeded without incident.
We were initially puzzled by our inability to precisely
1
match the entire set of H NMR spectral data reported
for natural roserine using CDCl3 solutions of compound
1b; furthermore, it was disappointing to learn that we
could not directly compare 1b with roserine because no
authentic sample of the natural product had been re-
tained by the Codina group.20 However, inspection of
(19) The solvent-independent signals for 1b (Cl-) and 1b (Br-) in
CDCl3 were uniformly offset downfield from the corresponding reported
roserine signals by 0.11 ( 0.03 ppm. Differences of this magnitude
have been ascribed to minor variances in solute concentration (ref 6f)
or solvent composition (ref 6d) in previous 1H NMR studies of
pyrrolophenanthridinium alkaloids. The proton signals of greatest
diagnostic value in a comparison of 1b with the reported data for
roserine were H(1â): δm ) 3.27 (br dd, J m ) 17, 6 Hz) [lit.1 δ ) 3.13
(dd, J ) 16, 6 Hz)], H(1R): δm ) 2.95 (symmetrical multiplet) [lit.1
δ
) 2.85 (ddd, J ) 16, 6, 1.5 Hz)], and H(10): δm ) 7.06 ( 0.11 (singlet)
[lit.1(assigned as H(7)) δ ) 6.95 (singlet)]. When corrected for the global
0.11 ppm offest (vide supra), all other proton NMR signals reported
for roserine were statistically indistinguishable from the corresponding
signals recorded for either 1a or 1b. See the Supporting Information
for a complete tabulated solvent-effect and counterion-effect study of
1a and 1b in three solvents.
The structure of 1b was fully assigned15 by analysis of
2D NOESY, COSY, HETCOR, HMBC, and COLOC
experiments recorded using a CDCl3 solution of 1b (Br-).
A carbon NMR study of 1b (Br-) and 1b (Cl-), each
examined in CDCl3, 1:1 CDCl3-CD3OD, and CD3OD

9310 J . Org. Chem., Vol. 62, No. 26, 1997
Bunnell et al.
1
Codina’s original H-1H COSY spectrum for roserine in
was concentrated with a rotary evaporator to give a crude
yellow solid. Recrystallization of the crude material from
hexanes gave 490 mg of 3 as white needles (80% yield): mp
CDCl3, which includes the 1-D proton spectrum on the x
and y axes, revealed the presence of a previously unre-
ported large, broad signal (δ 3.6-3.9) that is not cor-
1
112.5-113 °C; H NMR δ 8.21 (s, 1 H), 7.03 (s, 2 H), 3.91 (s,
6 H), 3.87 (s, 3 H), 3.55 (br m, 1 H), 3.24 (s, 3 H), 3.15 (s, 3 H),
2.08 (ddd, J ) 13.5, 10.8, 3.9 Hz), 1.95-1.35 (m, 7 H); 13C NMR
159.2, 153.4, 140.2, 132.4, 105.3, 100.9, 70.3, 60.9, 56.2, 47.7,
47.4, 32.4, 29.0, 22.9, 20.9; EI-MS m/z 337 (M, 5%), 323 (20%),
322 (100%), 306 (8%), 222 (16%), 196 (12%), 181 (8%), 176 (8%),
149 (25%). Anal. Calcd for C18H27HNO5: C, 64.07; H, 8.05;
N, 4.36. Found: C, 64.35, H, 8.05; N, 4.36.
1
related with any H nucleus from the natural product
structure. The sum of the 1H and 13C NMR signals
reported for roserine, plus the four heteroatoms necessary
for the simplest rationalization of the chemical shift
information, requires an empirical formula of C18H22NO3:
Since a high-resolution mass spectrum of the natural
product satisfactorily verified the NMR-predicted for-
(2,2-Dim et h oxycycloh exyl)(3,4,5-t r im et h oxyb en zyl)-
a m in e (4). A mixture of imine 3 (500 mg, 1.5 mmol) and 34
mg of PtO2 in 10 mL of abs EtOH was hydrogenated (23 °C, 1
atm H2) for 6 h. The reaction mixture was filtered through
Celite and concentrated under vacuum to give 490 mg of 4
(96%) as a clear, colorless oil: 1H NMR δ 6.59 (s, 2 H), 3.86 (s,
6 H), 3.83 (s, 3 H), 3.83 (d, J ) 13.4 Hz, 1 H), 3.61 (d, J ) 13.4
Hz, 1 H), 3.18 (s, 3 H), 3.10 (s, 3 H), 2.88 (br t, J ) 3 Hz),
1.8-1.25 (m, 8 H); 13C NMR δ 153.2, 137.0, 136.7, 105.2, 101.4,
60.9, 56.1, 55.6, 51.7, 47.7, 47.2, 27.9, 25.3, 22.5, 19.4; EI-MS
m/z 339 (M, 8%), 324 (12%), 308 (5%), 196 (12%), 181 (100%);
LSIMS-HRMS Calcd for C18H30NO5 (M + H): 340.2124.
Found: 340.2122. Anal. Calcd for C18H29NO5: C, 63.69; H,
8.61; N, 4.13. Found: C, 63.28; H, 8.65; N, 4.12.
mula1 it is clear that the newly discovered H NMR signal
is due to an unrelated impurity in Codina’s sample. We
have made several unsuccessful attempts to identify this
1
1
substance with the aim of evaluating its effect on the H
NMR chemical shifts of 1b; however, without recourse
to the original roserine sample or further information
about the exact nature of its contaminant, it is unlikely
that the solvent-dependent proton signals of 1b and
roserine can be precisely matched until the natural
product is isolated again.21
Proton H(4) of 1b (Cl-) did not exchange at a measur-
able rate in neat D2O (neutral pH, ambient temperature)
over the course of three weeks; therefore it may be
reasonable to expect that the absolute configuration of
roserine, if it is not a natural racemate, can eventually
be determined following reisolation from the plant source.
8,9,10-Tr im eth oxy-2,3,4,4a ,5,6-h exa h yd r op h en a n th r i-
d in e (5). Amino ketal 4 (100 mg, 0.3 mmol) was dissolved in
3 mL of aqueous 6 M HCl under a nitrogen atmosphere. The
reaction mixture was allowed to stir at 23 °C for 16 h; TLC
analysis (silica gel, 95:5 CH2Cl2-CH3OH) showed complete
consumption of 4 and the presence of a major product 5 (Rf )
0.26) and a minor product 6 (Rf ) 0.56). The reaction mixture
was made strongly basic with 3 M NaOH and extracted with
3 × 15 mL of ethyl acetate. The combined organic layers were
Con clu sion s
This work demonstrates the structure correction and
first total synthesis of the pyrrolophenanthridinium
alkaloid, (()-roserine. The optimized synthetic route
appears to be general and practical; thus, substantial
amounts of synthetic roserine, 1b (Cl-), and an isomer,
1a (Cl-), are currently available for biological assessment.
The biological activity of these compounds will be re-
ported in due course.
dried (MgSO4), filtered, and concentrated with
a rotary
evaporator to give 71 mg of a crude, pink oil. Preparative TLC
(silica gel; 95:5 CH2Cl2-CH3OH) afforded 46 mg of 5 (56%)
and 6 mg of 6 (7%). Compound 5 was a colorless oil: 1H NMR
δ 6.88 (br t, J ) 4 Hz, 1 H), 6.33 (s, 1 H), 4.11 (d, J ) 16.4 Hz,
1 H), 3.94 (d, J ) 16.4 Hz, 1 H), 3.84 (s, 3 H), 3.83 (s, 3 H),
3.80 (s, 3 H), 3.37-3.28 (m, 1 H), 2.30-2.21 (m, 2 H), 2.21-
2.12 (m, 1 H), 1.86-1.74 (m, 1 H), 1.69-1.53 (m, 1 H), 1.49-
1.34 (m, 1 H); 13C NMR δ 152.8, 151.9, 141.6, 131.8, 130.5,
125.7, 120.3, 104.8, 61.0, 59.9, 55.9, 53.8, 49.0, 31.6, 27.0, 20.9;
EI-MS m/z 275 (M, 100%), 274 (88%), 259 (10%), 258 (10%),
245 (43%), 246 (45%), 232 (12%), 216 (10%); LSIMS-HRMS
Calcd for C16H22NO3 (M + H): 276.1600. Found: 276.1605.
8,9,10-Tr im eth oxy-1,2,3,4-tetr ah ydr oph en an th r idin e (6).
Meth od A. Amino ketal 4 (878 mg, 2.6 mmol) was allowed
to react in aqueous 6 M HCl at 23 °C for 20 h under a nitrogen
atmosphere. An air atmosphere was applied to the vessel, and
the rapidly stirred reaction mixture was warmed to 100 °C
for 2 h. The reaction mixture was cooled to 0 °C, made strongly
basic with 30% NaOH, and extracted with 3 × 20 mL of
CH2Cl2. The combined organic layers were dried (MgSO4),
filtered, and concentrated with a rotary evaporator to give a
clear yellow oil. The crude product was purified by flash
chromatography (silica gel, 40:1 CH2Cl2-CH3OH) to afford 340
mg of 6 (48%) as a colorless oil: 1H NMR δ 8.82 (s, 1 H), 7.00
(s, 1 H), 3.99 (s, 3 H), 3.98 (s, 3 H), 3.92 (s, 3 H), 3.38 (br t, J
) 6 Hz, 2 H), 3.05 (br t, J ) 6 Hz, 2 H), 1.96-1.80 (m, 4 H);
13C NMR δ 152.0, 149.4, 148.6, 147.8, 145.5, 126.5, 124.7,
124.4, 102.5, 60.9, 60.6, 55.5, 33.1, 27.1, 23.1, 22.2; EI-MS m/z
273 (M, 100%), 258 (50%), 243 (20%), 242 (20%). Anal. Calcd
for C16H19NO3: C, 70.31; H, 7.01; N, 5.12. Found: C, 70.40;
H, 7.00; N, 4.79.
Exp er im en ta l Section
Routine 1H and 13C spectra were recorded at 300.13 and
75.40 MHz, respectively, using CDCl3 solutions with internal
tetramethylsilane as the reference unless otherwise noted.
Routine operating procedures and solvent purification methods
used in this work have been characterized elsewhere.22 El-
emental analyses were performed by Analytical Services,
Central Research and Development, of Roche Bioscience.
(2,2-Dim et h oxycycloh exyl)(3,4,5-t r im et h oxyb en zyli-
d en e)a m in e (3). 3,4,5-Trimethoxybenzaldehyde (372 mg, 1.9
mmol) and amino ketal 28 (300 mg, 1.9 mmol) were combined
in 50 mL of anhydrous MeOH, and the mixture was brought
to reflux under a nitrogen atmosphere. The reaction mixture
was periodically concentrated by distillation (1 atm), and the
solvent was replaced with fresh MeOH; after 4 h the mixture
(20) Bastida, J .; Codina, C. Personal communication: J une 18, 1997.
(21) For example, pure 1b (Cl-) was subjected to the unusual
preparative TLC conditions outlined in ref 1; however, the 1H NMR
spectrum (CDCl3) of 1b isolated from this experiment was unchanged.
Mixtures of CD3OD, DMSO-d6, D2O, or H2O with CDCl3 caused
variable, pronounced shifts in the solvent-dependent proton signals
of 1b (Cl-), but none of these experiments provided a precise match to
the original roserine data. Interestingly, a proton NMR spectrum of
1b (Cl-) in neat D2O did provide a nearly exact match to all of the
data reported for roserine (see Supporting Information). A fresh
analysis of all the original spectral data for roserine might prove useful
in probing the identity of the impurity peak; however, beyond the 1H-
1H COSY spectrum, we have been unable to obtain any further original
physical data or information concerning the isolation of natural
roserine from the Codina group.
Meth od B. Amino ketal 4 (400 mg, 1.2 mmol) was dissolved
in 12 mL of aqueous 6 M HCl, and the solution was heated to
reflux for 3.5 h under an air atmosphere. The reaction mixture
was made strongly basic with NaOH and extracted with
CH2Cl2 to give a dark crude product. Flash chromatography
(silica gel, 97:3 CH2Cl2-CH3OH) gave 129 mg of 6 (39%) and
two phenolic products, 7 (55 mg, 18%) and 8 (41 mg, 13%).
1
Routine H NMR spectra of 7 and 8 indicated that they were
isomeric mono-demethylated derivatives of structure 6; their
exact structures were assigned from a series of proton NOE
(22) Flippin, L. A.; Berger, J .; Parnes, J . S.; Gudiksen, M. S. J . Org.
Chem. 1996, 61, 4812.

Total Synthesis of (()-Roserine
J . Org. Chem., Vol. 62, No. 26, 1997 9311
experiments. 9-Hyd r oxy-8,10-d im eth oxy-1,2,3,4-tetr a h y-
d r op h en a n th r id in e (7). H NMR δ 8.78 (s, 1 H), 7.13 (s, 1
revealed 14a and 15a in a 4:1 ratio. B. Na BH4-2-P r op a n ol
Meth od . Under conditions identical with those described
above, keto ester 11 and 3,4,5-trimethoxybenzylamine were
allowed to react in titanium isopropoxide. A slurry of NaBH4
(148 mg, 3.9 mmol) in 10 mL of abs 2-propanol was added,
and, after 8 h, the reaction mixture was worked up as above
to give 1.10 g of a crude product mixture. 1H NMR analysis
of the crude material showed 14a and 15a in a 10:1 ratio. Flash
chromatography (silica gel, 130:1 CH2Cl2-CH3OH) afforded
490 mg (59%) of cis diastereomer 14a as a colorless oil, 40 mg
of a 14a /15a mixture, and 5 mg (6%) of slightly impure trans
diastereomer 15a .
1
H), 4.07 (s, 3 H), 3.89 (s, 3 H), 3.38 (br t, J ) 6 Hz, 2 H), 3.07
(br t, J ) 6 Hz, 2 H), 1.96-1.80 (m, 4 H). 8-Hyd r oxy-9,10-
dim eth oxy-1,2,3,4-tetr ah ydr oph en aan th r idin e (8). 1H NMR
δ 8.70 (s, 1 H), 7.20 (s, 1 H), 3.92 (s, 3 H), 3.81 (s, 3 H), 3.33
(br t, J ) 6 Hz, 2 H), 3.00 (br t, J ) 6 Hz, 2 H), 1.89-1.73 (m,
4 H).
Deh yd r ogen a tion of 5 w ith 20% P d /C. 8,9,10-Tr i-
m eth oxyp h en a n th r id in e (9). Compound 5 (20 mg, 0.07
mmol) and 40 mg of 20% Pd/C were mixed in 2 mL of decalin.
The mixture was allowed to reflux for 4 h under a nitrogen
atmosphere; TLC (silica gel, 95:5 CH2Cl2-CH3OH) showed
complete consumption of 5 and the formation of a single new
product. The reaction mixture was diluted with 10 mL of
benzene and extracted with 5 mL of aqueous 4 M HCl. The
aqueous layer was made strongly basic with NaOH and
extracted with ethyl acetate to give 14 mg (74%) of 9: Rf )
14a : 1H NMR δ 6.64 (s, 1 H), 5.02 (heptet, J ) 6.2 Hz, 1 H),
3.90 (d, J ) 13 Hz, 1 H), 3.88 (s, 3 H), 3.83 (s, 3 H), 3.69 (d, J
) 13 Hz, 1 H), 3.22 (s, 6 H), 2.92 (m, bandwidth ) 7.3 Hz, 1
H), 2.52-2.47 (m, 1 H), 2.31-2.18 (m, 2 H), 1.80 (br d, J ) 10
Hz, 1 H), 1.56-1.28 (m, 6 H), 1.229 (d, J ) 6.2 Hz, 3 H), 1.224
(d, J ) 6.2 Hz, 3 H); 13C NMR δ 173.0, 153.5, 141.5, 105.5,
103.0, 67.8, 61.2, 59.7, 56.5, 55.9, 48.2, 47.9, 38.9, 28.0, 26.2,
22.3, 21.9; EI-MS m/z 439 (M, 2%), 424 (10%), 181 (100%).
Anal. Calcd for C23H37NO7 (0.5 mol H2O): C, 61.59; H, 8.54;
N, 3.12. Found: C, 61.37; H, 8.54; N, 3.29.
1
0.6; H NMR δ 9.33 (dd, J ) 7.8, 2.0 Hz, 1 H), 9.13 (s, 1 H),
8.16 (dd, J ) 7.8, 2.0 Hz, 1 H), 7.70 (td, J ) 7.1, 1.7 Hz, 1 H),
7.65 (td, J ) 7.1, 1.7 Hz, 1 H), 7.25 (s, 1 H), 4.08 (s, 3 H), 4.06
(s, 6 H); 13C NMR δ 153.4, 152.4, 151.5, 146.3, 144.2, 129.9,
127.6, 127.2, 126.2, 124.2, 123.6, 121.6, 105.1, 61.3, 60.5, 56.1;
EI-MS m/z 269 (M, 100%), 254 (53%), 239 (16%), 226 (44%),
211 (80%), 196 (20%), 183 (20%), 140 (57%); EI-HRMS Calcd
for C16H15NO3: 269.1052. Found: 269.1052. An analytical
sample was obtained by recrystallization of the picrate deriva-
tive of 9 from methanol: mp (picrate) 201-202 °C. Anal. Calcd
for C22H18N4O10: C, 53.02; H, 3.64; N, 11.24. Found: C, 52.93;
H, 3.46; N, 11.16.
15a : 1H NMR δ 6.62 (s, 2 H), 5.02 (heptet, J ) 6.3 Hz, 1
H), 3.87 (s, 6 H), 3.83 (s, 3 H), 3.74 (d, J ) 13 Hz, 1 H), 3.17
(s, 3 H), 3.05 (s, 3 H), 2.66-2.54 (m, 2 H), 2.45-2.35 (m, 2 H),
1.73-1.58 (m, 4 H), 1.50-1.39 (m, 1 H), 1.24 (d, J ) 6.3 Hz, 3
H), 1.22 (d, J ) 6.3 Hz, 3 H).
Isop r op yl cis-[3,3-Dim eth oxy-2-[(2,3,4-tr im eth oxyben -
zyl)a m in o)cycloh exyl]a ceta te (14b) a n d Isop r op yl tr a n s-
[3,3-Dim eth oxy-2-[(2,3,4-tr im eth oxyben zyl)a m in o]cyclo-
h exyl]a ceta te (15b). Keto ester 11 (2.23 g, 8.6 mmol) and
2,3,4-trimethoxybenzylamine23 (3.4 g, 17.2 mmol) were dis-
solved in 3.94 g (17.3 mmol) of titanium isopropoxide at 23 °C
under a nitrogen atmosphere. The reaction mixture was
allowed to stand at room temperature for 24 h, and a solution
of 0.653 g (17.3 mmol) of NaBH4 in 45 mL of 2-propanol was
added. The reaction mixture was allowed to stand at room
temperature for 12 h, diluted with 200 mL of CH2Cl2 and 40
mL of aqueous 2 M NH4OH, and filtered through Celite. The
filter cake was washed with additional CH2Cl2, and the organic
layers were combined, dried (MgSO4), and concentrated to give
5.1 g of crude material. 1H NMR analysis of the crude mixture
revealed a 13:1 mixture 14b and 15b, respectively. The crude
product was purified by flash chromatography (silica gel, 98:2
CH2Cl2-CH3OH) to give 2.40 g (64%) of 14b and 0.14 g (4%)
of 15b, both as colorless oils.
Eth yl (3,3-Dim eth oxy-2-oxocycloh exyl)a ceta te (10).11
To a solution of 1.00 g (6.3 mmol) of 2,2-dimethoxycyclohex-
anone in 5 mL of THF at 0 °C under a nitrogen atmosphere
was added 7.6 mL (7.6 mmol) of a 1 M KOtBu in THF. The
reaction mixture was allowed to warm to 23 °C for 1 h, giving
a deep burgundy-colored solution. The reaction mixture was
cooled to -78 °C, and ethyl iodoacetate (1.35 g, 6.3 mmol) was
added dropwise. After 2 h at -78 °C, the reaction mixture
was quenched by addition of satd aqueous NH4Cl and ex-
tracted with ether. The ether layers were dried (MgSO4),
filtered, and concentrated with a rotary evaporator to give 1.15
g of a crude oily product. Flash chromatography (silica gel,
4:1 hexanes-ethyl acetate) of the crude material afforded 940
mg (65%) of 10 as a colorless oil: 1H NMR δ 4.18-4.05 (m, 2
H), 3.37-3.24 (m, 1 H), 3.31 (s, 3 H), 3.24 (s, 3 H), 2.77 (dd, J
) 16.8, 8.5 Hz, 1 H), 2.43 (dq, J ) 13.8, 3 Hz, 1 H), 2.20 (dd,
J ) 16.8, 5.3 Hz, 1 H), 2.12-2.00 (m, 1 H), 1.91 (qd, J ) 12.5,
3.9 Hz, 1 H), 1.76-1.66 (m, 1 H), 1.48 (td, J ) 13.8, 4.5 Hz, 1
H), 1.35 (qd, J ) 13.1, 4.0 Hz, 1 H), 1.25 (t, J ) 7.1 Hz, 3).
Similarly, alkylation of 2,2-dimethoxycyclohexanone with iso-
propyl iodoacetate gave keto ester 11 (68%) as a crystalline
solid: mp 50-51 °C; 1H NMR δ 4.99 (heptet, J ) 6.3 Hz, 1 H),
3.37-3.23 (m, 1 H), 3.31 (s, 3 H), 3.24 (s, 3 H), 2.73 (dd, J )
16.7, 8.7 Hz, 1 H), 2.43 (dq, J ) 13.8, 3 Hz, 1 H), 2.18 (dd, J
) 16.7, 5.3 Hz, 1 H), 2.11-2.00 (m, 1 H), 1.92 (qd, J ) 13.4,
3.9 Hz, 1 H), 1.76-1.66 (m, 1 H), 1.48 (td, J ) 13.7, 4.5 Hz, 1
H), 1.35 (qd, J ) 13.0, 4.0 Hz, 1 H), 1.24 (d, J ) 6.3 Hz, 3),
1.22 (d, J ) 6.3 Hz, 3 H); 13C NMR δ 207.1, 171.8, 101.1, 67.9,
49.9, 48.9, 44.7, 35.6, 34.5, 34.0, 21.8, 21.2. Anal. Calcd for
C13H22O5: C, 60.45; H, 8.58. Found: C, 60.59; H, 8.48.
1
14b: H NMR δ 7.00 (d, J ) 8.4 Hz, 1 H), 6.63 (d, J ) 8.4
Hz, 1 H), 5.02 (heptet, J ) 6.2 Hz, 1 H), 3.93 (s, 3 H), 3.87 (s,
3 H), 3.85 (s, 3 H), 3.84 (d, J ) 12 Hz, 1 H), 3.62 (d, J ) 12 Hz,
1 H), 3.20 (s, 3 H), 3.19 (s, 3 H), 2.88 (m, bandwidth ) 7.6 Hz,
1 H), 2.50-2.38 (m, 1 H), 2.31-2.15 (m, 2 H), 1.80-1.73 (m, 1
H), 1.55-1.36 (m, 5 H), 1.236 (d, J ) 6.2 Hz, 3 H), 1.233 (d, J
) 6.2 Hz, 3 H); 13C NMR δ 172.9, 153.0, 152.2, 142.3, 124.3,
110.1, 107.1, 102.6, 67.4, 61.2, 60.8, 59.2, 56.0, 50.4, 47.7, 47.5,
38.4, 36.1, 37.5, 25.7, 21.93, 21.91, 21.5; EI-MS m/z 439 (M,
12%), 424 (80%), 409 (10%), 408 (10%), 181 (100%), 166 (20%);
LSIMS-HRMS Calcd for C23H38NO7 (M + H): 440.2648.
Found: 440.2645.
1
15b: H NMR δ 6.98 (d, J ) 8.4 Hz, 1 H), 6.61 (d, J ) 8.4
Hz, 1 H), 5.04 (heptet, J ) 6.3 Hz, 1 H), 3.93 (s 3 H), 3.92-
3.80 (m, 3 H), 3.85 (s, 3 H), 3.84 (s, 3 H), 3.65 (br d, J ) 13 Hz,
1 H), 3.14 (s, 3 H), 2.99 (s, 3 H), 2.64-2.40 (m, 4 H), 2.00-
1.84 (m, 1 H), 1.72-1.55 (m, 2 H), 1.48-1.35 (m, 2 H), 1.32-
1.26 (m, 1 H), 1.254 (d, J ) 6.3 Hz, 3 H), 1.242 (d, J ) 6.3 Hz,
3 H); 13C NMR δ 173.7, 153.0, 152.3, 142.2, 124.6, 106.9, 101.7,
77.3, 67.2, 61.0, 60.7, 59.5, 56.0, 47.7, 47.1, 46.9, 37.4, 32.9,
28.0, 24.8, 22.0, 21.9, 18.3; LSIMS-HRMS Calcd for C23H38NO7
(M + H): 440.2648. Found: 440.2651.
Isop r op yl cis-[3,3-Dim eth oxy-2-[(3,4,5-tr im eth oxyben -
zyl)a m in o]cycloh exyl]a ceta te (14a ) a n d Isop r op yl tr a n s-
[3,3-Dim eth oxy-2-[(3,4,5-tr im eth oxyben zyl)a m in o]cyclo-
h exyl]a ceta te (15a ). A. Na BH4-Eth a n ol Meth od . Keto
ester 11 (500 mg, 1.9 mmol) and 3,4,5-trimethoxybenzylamine
(764 mg, 3.9 mmol) were dissolved in 1.11 g (3.9 mmol) of
titanium isopropoxide at 23 °C under a nitrogen atmosphere.
After 12 h NaBH4 (148 mg, 3.9 mmol) in ca. 20 mL of abs EtOH
was added, and the reaction mixture was allowed to stand
overnight at 23 °C. The reaction mixture was diluted with 50
mL of CH2Cl2 and quenched with 10 mL of 2 M NH4OH. The
mixture was filtered through Celite and extracted with CH2Cl2,
the combined organic layers were dried (MgSO4) and filtered,
and the solution was concentrated to give 1.21 g of a crude
product mixture. 1H NMR analysis of the crude material
(8,9,10-Tr im eth oxy-1,2,3,4-tetr a h yd r op h en a n th r id in -
4-yl)a cetic Acid (16a ). Amino ketal 14a (500 mg, 1.1 mmol)
was allowed to react in 10 mL of aqueous 6 M HCl at 23 °C
under a nitrogen atmosphere for 12 h. The nitrogen atmo-
sphere was replaced by air, and the reaction mixture was
(23) Gutsche, C. D.; J ohnson, H. E. J . Am. Chem. Soc. 1954, 76,
1776.

9312 J . Org. Chem., Vol. 62, No. 26, 1997
Bunnell et al.
stirred rapidly and heated to 100 °C for 1 h. The mixture was
cooled to room temperature, brought to pH ) 12 using aqueous
NaOH, and extracted with 3 × 50 mL of CH2Cl2. These CH2Cl2
layers were discarded. The strongly basic aqueous layer was
neutralized to pH ) 7 using 4 M HCl and extracted with 3 ×
50 mL of CH2Cl2. The combined organic layers were dried
(MgSO4), filtered, and concentrated with a rotary evaporator
to give 271 mg of a crude product. Purification of the crude
material by flash chromatography (silica gel, 95:5 CH2Cl2-
CH3OH) afforded 228 mg (60%) of 16a as a white crystalline
with BMS under conditions identical with those described
above gave 220 mg (56%) of 17b and 114 mg (26%) of ester
18b as colorless oils.
17b: 1H NMR δ 9.17 (s, 1 H), 6.89 (s, 1 H), 4.09 (s, 3 H),
4.01 (s, 3 H), 3.96 (s, 3 H), 3.86-3.75 (m, 2 H), 3.28-3.18 (m,
1 H), 2.99 (br t, J ) 6 Hz, 2 H), 2.10-1.70 (m, 6 H); 13C NMR
δ 157.0, 152.0, 149.2, 143.3, 140.4, 133.7, 123.9, 118.0, 96.8,
62.2, 61.9, 61.3, 56.1, 40.5, 38.9, 31.1, 25.6, 20.6; LSIMS-
HRMS Calcd for C18H24NO4 (M + H): 318.1705. Found:
318.1700. Anal. Calcd for C18H23NO4 (0.5 mol H2O): C, 66.24;
H, 7.41; N, 4.29. Found: 66.39; H, 7.24; N, 4.46.
1
solid: mp 152-153 °C; H NMR δ 8.79 (s, 1 H), 7.06 (s, 1 H),
18b: 1H NMR δ 9.21 (s, 1 H), 6.89 (s, 1 H), 4.25-4.13 (m, 2
H), 4.09 (s, 3 H), 4.01 (s, 3 H), 3.96 (s, 3 H), 3.62-3.49 (m, 1
H), 3.20 (dd, J ) 15.5, 4.4 Hz, 1 H), 2.98 (br t, J ) 5.9 Hz, 1
H), 2.53 (dd, J ) 15.5, 9.7 Hz, 1 H), 2.19-1.68 (m, 4 H), 1.28
(t, J ) 7.1 Hz, 3 H); 13C NMR δ 173.3, 156.7, 151.2, 149.2,
144.2, 140.4, 133.3, 118.1, 107.1, 96.9, 61.9, 61.3, 60.2, 56.0,
40.1, 37.9, 28.2, 25.5, 20.0, 14.3.
4.03 (s, 3 H), 4.00 (s, 3 H), 3.94 (s, 3 H), 3.58 (br dt, J ) 18.3,
4 Hz, 1 H), 3.5-3.37 (m, 1 H), 3.32-3.18 (m, 1 H), 2.85 (dd, J
) 15.6, 10.0 Hz, 1 H), 2.68 (dd, J ) 15.6, 1.2 Hz, 1 H), 2.21-
2.12 (m, 1 H), 2.10-2.02 (m, 1 H), 1.81-1.54 (m, 2 H); 13C NMR
δ 176.9, 155.9, 152.2, 150.9, 149.5, 147.8, 130.0, 129.4, 127.1,
105.2, 63.5, 63.2, 58.2, 46.0, 39.4, 33.4, 30.1, 24.4; LSIMS m/z
332 (M + H, 100%), 314 (5%), 286 (5%), 272 (5%). Anal. Calcd
for C18H21NO5: C, 65.24; H, 6.39; N, 4.23. Found: C, 65.11;
H, 6.38; N, 4.38.
(7,8,9-Tr im eth oxy-1,2,3,4-tetr a h yd r op h en a n th r id in -4-
yl)a cetic Acid (16b). Compound 14b (1.00 g, 2.3 mmol) was
allowed to react in 100 mL of aqueous 6 M HCl at 23 °C under
a nitrogen atmosphere for 36 h. The reaction mixture was
then exposed to air and heated to 110 °C for 1.5 h. The
reaction mixture was cooled to room temperature, adjusted to
pH ) 7 with aqueous 3 M NaOH, and extracted with CH2Cl2.
The combined organic layers were dried (MgSO4), filtered, and
concentrated to give 600 mg of a crude residue. The crude
material was purified using flash chromatography (silica gel,
95:5 CH2Cl2-CH3OH) to afford 300 mg (40%) of 16b as a white
crystalline solid: mp 139-141 °C; 1H NMR δ 9.16 (s, 1 H),
6.91 (s, 1 H), 4.13 (s, 3 H), 4.05 (s, 3 H), 3.98 (s, 3 H), 3.50-
3.39 (m, 1 H), 3.16-3.05 (m, 1 H), 2.99-2.76 (m, 1 H), 2.84
(dd, J ) 15, 10 Hz, 1 H), 2.67 (dd, J ) 15, 1 Hz, 1 H), 2.22-
2.10 (m, 2 H), 1.93-1.78 (m, 1 H), 1.67-1.55 (m, 1 H); 13C NMR
δ 174.9, 158.6, 149.7, 149.3, 141.8, 140.9, 134.3, 125.3, 118.2,
96.9, 61.9, 61.4, 56.2, 43.5, 36.7, 31.7, 25.4, 21.7; EI-MS m/z
331 (M, 35%), 287 (65%), 286 (100%), 272 (30%), 258 (10%),
256 (15%), 242 (20%), 228 (20%). Anal. Calcd for C18H21NO5:
C, 65.24; H, 6.39; N, 4.23. Found: C, 65.15; H, 6.32; N, 4.36.
2-(8,9,10-Tr im eth oxy-1,2,3,4-tetr a h yd r op h en a n th r id in -
4-yl)eth a n ol (17a ). To a solution of 16a (200 mg, 0.6 mmol)
in 10 mL of THF at 23 °C was added 0.32 mL (0.64 mmol) of
neat BMS over 1 h. After 4 h the solvent was removed with
a rotary evaporator, and the residue was dissolved in 15 mL
of 2 M ethanolic HCl. The reaction mixture was allowed to
reflux for 45 min and concentrated to a solid residue under
vacuum. The residue was dissolved in ca. 5 mL of water,
adjusted to pH ) 12 with NaOH, and extracted with CH2Cl2
to give 190 mg of a crude product. The crude material was
purified using flash chromatography (silica gel, 98:2 CH2Cl2-
CH3OH) to give 105 mg (55%) of 17a and 64 mg (30%) of ester
18a as colorless oils.
17a : 1H NMR δ 8.79 (s, 1 H), 7.00 (s, 1 H), 4.00 (s, 3 H),
3.98 (s, 3 H), 3.92 (s, 3 H), 3.88-3.72 (m, 2 H), 3.39 (br t, J )
6 Hz, 2 H), 3.27-3.17 (m, 1 H), 2.09-1.64 (m, 6 H); 13C NMR
δ 153.3, 151.3, 150.5, 147.7, 146.7, 127.6, 126.2, 125.4, 103.3,
62.6, 61.7, 61.4, 56.3, 41.2, 39.7, 31.0, 28.4, 21.5. EI-MS m/z
317 (M, 7%), 286 (7%), 273 (100%), 258 (30%), 242 (12%), 228
(5%). An analytically pure derivative was obtained by recrys-
tallization of the picrate salt of 17a from abs ethanol: mp
(picrate) 123-124 °C. Anal. Calcd for C24H26N4O11: C, 52.75;
H, 4.80; N, 10.25. Found: C, 52.78; H, 4.78; N, 10.20.
1a (Br -). Meth od A.14 Compound 17a (58 mg, 0.18 mmol)
was dissolved in 10 mL of benzene. To this solution was added
0.2 mL of PBr3 in 2 mL of 1:1 ether-hexane, and the reaction
mixture was allowed to reflux under a nitrogen atmosphere
for 2.5 h. Methanol (2 mL) was added, and the reaction
mixture was concentrated with a rotary evaporator. The crude
residue was purified by flash chromatography (silica gel, 95:5
CH2Cl2-CH3OH) to give 30 mg (44%) of 1a (Br-): 1H NMR δ
10.45 (s, 1 H), 7.89 (s, 1 H), 5.34 (br dd, J ) 13.3, 8.3 Hz, 1 H),
4.88 (ddd, J ) 13.3, 13.3, 6.3 Hz, 1 H), 4.09 (s, 3 H), 4.07 (s, 3
H), 3.97 (s, 3 H), 3.73 (br dd, J ) 18.7, 6.3 Hz, 1 H), 3.51-3.36
(m, 1 H), 3.27-3.12 (m, 1 H), 2.74 (br ddd, J ) 12.5, 6.3, 6.3
Hz, 1 H), 2.44-2.26 (m, 2 H), 2.10 (dddd, J ) 12.2, 12.2, 12.2,
8.3 Hz, 1 H), 2.02-1.82 (m, 1 H), 1.45 (dddd, J ) 11.8, 11.8,
11.8, 2.8 Hz, 1 H); 13C NMR δ 155.5, 150.6, 149.2, 144.1, 142.5,
129.0, 128.9, 125.4, 106.6, 61.7, 61.4, 57.3, 57.0, 41.0, 31.8, 26.6,
26.1, 22.9. See the Supporting Information for proton assign-
ments and detailed results of a 1H NMR solvent-effect study
using three solvents. EI-MS m/ s 300 (M, 2%), 286 (24%), 285
(100%),284 (19%), 271 (22%), 270 (92%), 267 (32%), 266 (52%),
256 (60%), 254 (40%), 242 (86%), 239 (21%), 239 (8%), 238
(48%), 228 (8%), 227 (16%), 226 (15%), 225 (8%), 212 (12%),
199 (33%), 183 (8%), 171 (30%), 170 (12%), 167 (9%), 154 (12%),
143 (15%), 142 (20%), 128 (8%), 127 (6%), 115 (14%), 102 (4%),
89 (4%), 77 (5%); LSIMS-HRMS Calcd for C18H22NO3:
300.1600. Found: 300.1607. Meth od B. Compound 17a (39
mg, 0.12 mmol) was dissolved in 10 mL of benzene. To this
solution was added 33 mg (0.12 mmol) of PBr3 in 1 mL of 1:1
ether-hexane, and the reaction mixture was allowed to reflux
for 10 min. The reaction mixture was concentrated under
vacuum, and the residue was purified by flash chromatography
(silica gel, 94:6 CH2Cl2-CH3OH) to afford 37 mg (76%) of 1a
(Br-).
1a (Cl-). To a solution of compound 17a (120 mg, 0.38
mmol) in 10 mL of benzene was added 52 mg (0.38 mmol) of
PCl3 in 1 mL of benzene. The reaction mixture was allowed
to reflux for 20 min under a nitrogen atmosphere, 2 mL of
methanol was added, and the mixture was concentrated with
a rotary evaporator. The crude residue was purified by flash
chrmatography (silica gel, 92:8 CH2Cl2-CH3OH) to give 127
mg (87%) of 1a (Cl-): darkens 92 °C; mp 92.5-93 °C; See the
Supporting Information for proton assignments and detailed
1
results of a H NMR solvent-effect study using three solvents;
EI-MS m/z 300 (M, 2%), 286 (20%), 285 (100%), 284 (20%),
271 (16%), 270 (86%), 267 (29%), 266 (46%), 256 (60%), 254
(36%), 242 (60%), 239 (16%), 238 (36%), 228 (4%), 227 (13%),
226 (10%), 225 (4%), 212 (9%), 199 (20%), 183 (6%), 171 (20%),
170 (10%), 167 (7%), 154 (6%), 143 (11%), 142 (10%), 128 (4%),
127 (6%), 115 (10%), 102 (2%), 89 (6%), 77 (6%).
18a : 1H NMR δ 8.83 (s, 1 H), 7.00 (s, 1 H), 4.27-4.14 (m,
2H), 3.99 (s, 3 H), 3.97 (s, 3 H), 3.92 (s, 3 H), 3.61-3.49 (m, 1
H), 3.43-3.34 (m, 1 H), 3.18 (dd, J ) 15.3, 4.3 Hz, 1 H), 2.52
(dd, J ) 15.3, 9.8 Hz, 1 H), 2.14-1.67 (m, 4 H), 1.28 (t, J )
7.2 Hz, 3 H); 13C NMR δ 173.4, 152.8, 150.2, 150.1, 148.4,
146.0, 126.7, 125.4, 125.2, 102.8, 61.3, 61.0, 60.2, 55.9, 40.7,
38.6, 27.88, 27.86, 20.7, 14.3; EI-MS m/z 359 (M, 64%), 344
(6%), 314 (11%), 298 (4%), 286 (100%); EI-HRMS Calcd for
1b (Br -). To a solution of 17b (14.0 mg, 0.044 mmol) in 4
mL of benzene was added 1 drop of PBr3. The solution was
allowed to reflux for 20 min under a nitrogen atmosphere,
cooled to room temperature, diluted with 1 mL of methanol,
and concentrated with a rotary evaporator. The crude product
was purified by flash chromatography (silica gel, 96:4 CH2Cl2-
CH3OH) to give 10.0 mg (60%) of 1b (Br-): 1H NMR δ 9.93 (s,
1 H), 6.94 (s, 1 H), 5.37 (br dd, J ) 12.5, 8.3 Hz, 1 H), 5.07
(ddd, J ) 12.5, 12.5, 6.3 Hz, 1 H), 4.27 (s, 3 H), 4.13 (s, 3 H),
C
20H25NO5: 359.1733. Found: 359.1732.
2-(7,8,9-Tr im eth oxy-1,2,3,4-tetr a h yd r op h en a n th r id in -
4-yl)eth a n ol (17b). Reduction of 411 mg (1.24 mmol) of 16b

Total Synthesis of (()-Roserine
J . Org. Chem., Vol. 62, No. 26, 1997 9313
3.98 (s, 3 H), 3.63-3.48 (m, 1 H), 3.23 (br dd, J ) 17.8, 6.3 Hz,
1 H), 2.99-2.84 (m, 1 H), 2.79 (br ddd, J ) 12.5, 6.3, 6.3 Hz,
1 H), 2.48-2.32 (m, 2 H), 2.18 (dddd, J ) 12.5, 12.5, 12.5, 8.3
Hz, 1 H), 2.09-1.91 (m, 1 H), 1.47 (dddd, J ) 12.6, 12.6, 12.6,
2.7 Hz, 1 H); 13C NMR δ 162.3, 150.6, 146.7, 142.1, 138.6,
135.5, 127.0, 119.3, 97.0, 62.9, 61.6, 57.9, 57.1, 40.3, 32.0, 27.2,
23.4, 22.1. See the Supporting Information for proton assign-
melting; mp 126-128 °C dec. See the Supporting Information
for proton assignments and detailed results of a 1H NMR
solvent-effect study using three solvents. EI-MS m/z 300 (M,
2%), 286 (12%), 285 (55%), 284 (28%), 271 (19%), 270 (100%),
256 (20%), 254 (13%), 242 (43%), 239 (37%), 238 (4%), 228 (3%),
226 (10%), 225 (8%), 214 (5%), 212 (6%), 199 (24%), 183 (5%),
171 (15%), 170 (8%), 154 (4%), 143 (6%), 142 (6%), 115 (6%),
1
ments and details of a H NMR solvent-effect study using three
102 (2%), 83 (3%), 77 (4%); LSIMS-HRMS Calcd for C18H22-
NO3: 300.1600. Found: 300.1603.
solvents. EI-MS m/z 300 (M, 2%), 286 (12%), 285 (52%), 284
(26%), 271 (20%), 270 (100%), 256 (16%), 254 (13%), 242 (28%),
239 (26%), 228 (4%), 226 (8%), 225 (7%), 214 (4%), 212 (4%),
199 (12%), 183 (3%), 171 (8%), 154 (3%), 143 (3%), 142 (4%),
115 (4%), 102 (2%), 83 (4%), 77 (5%); LSIMS-HRMS Calcd
for C18H22NO3: 300.1600. Found 300.1602.
Su p p or tin g In for m a tion Ava ila ble: Tabulated results of
1H NMR study of 1a (Br-), 1a (Cl-), 1b (Br-), and 1b (Cl--
a
)
using CDCl3, 1:1 CDCl3-CD3OD, and CD3OD solutions, and
1
copies of representative H and 13C NMR spectra of 1a (Br-),
1b (Cl-). To a solution of 17b (120 mg, 0.38 mmol) in 10
mL of benzene was added 52 mg (0.38 mmol) of PCl3 in 1 mL
of benzene. The reaction mixture was allowed to reflux for
20 min and worked up as described above, and the crude
product was purified by flash chromatography (silica gel, 92:8
CH2Cl2-CH3OH) to give 111 mg (87%) of 1b (Cl-) as a white,
hygroscopic crystalline solid: turns red 105-120 °C prior to
1b (Cl-), and 1b (Br-) in various solvents (13 pages). This
material is contained in libraries on microfiche, immediately
follows this article in the microfilm version of the journal, and
can be ordered from the ACS; see any current masthead page
for ordering information.
J O971641O