Metal-catalyzed copolymerization of imines and CO: A non-amino acid route to polypeptides

Article abstract of DOI:10.1002/anie.200700646

(Chemical Equation Presented) The path less taken: Copolymerization of imines and carbon monoxide has long been highly desirable as a route to synthesize polypeptides, but has been difficult to implement because of the lack of appropriate catalysts. This reaction has now been realized for the first time by using a simple cobalt catalyst (see scheme).

Full text of DOI:10.1002/anie.200700646

Communications
DOI: 10.1002/anie.200700646
Polypeptide Synthesis
Metal-Catalyzed Copolymerization of Imines and CO: A
Non-Amino Acid Route to Polypeptides**
Huailin Sun,* Jian Zhang, Qiuhua Liu, Lei Yu, and Jiangyu Zhao
Angewandte
Chemie
6068
ꢀ 2007 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Angew. Chem. Int. Ed. 2007, 46, 6068 –6072

Angewandte
Chemie
Polypeptides are an exceptionally significant class of biopo-
lymers that not only are responsible for both the structure and
function of most living things but also have broad applications
in materials, catalysis, and pharmaceuticals. For a century,
almost all studies on polypeptide synthesis have been based
on the use of amino acids as starting materials, which requires
tedious procedures for presynthesis of amino acids and
subsequent activation of the highly stable carboxyl groups
using stoichiometric amounts of special reagents to form
peptide bonds.[1–3] Herein, we report a shortcut method that
does not use amino acids but instead employs readily
available imines and carbon monoxide (CO) as monomers
that undergo metal-catalyzed alternating copolymerization to
directly form polypeptides. We find that a simple acylcobalt
complex can effectively catalyze this reaction to produce
polypeptides of high molecular weights with low polydispers-
ity. This efficient metal-catalyzed synthesis of polypeptides
from inexpensive and plentiful starting materials makes large-
scale production of the polypeptide materials possible in a
fashion similar to Ziegler–Natta polymerization reactions.
Construction of polymers from small organic monomers
represents an important transformation. Beginning with
Ziegler and Nattaꢀs discoveries, metal catalysis has been
recognized as an extremely efficient tool for such purposes.
Recent breakthroughs achieved in homogeneous catalysis
using well-defined metal catalysts have resulted in rapid
progress in the study of various types of polymerization
reactions, such as the insertion and the metathesis polymer-
ization of alkenes and the copolymerization of olefins and
CO.[4] Most of these reactions, however, involve the use of
alkenes as monomers. In sharp contrast, similar reactions of
imines are largely unexplored, even though imines constitute
a huge class of organic compounds readily available from
various aldehydes (or ketones) and appropriate amines. This
phenomenon is usually attributed to the structural character
of imines, that is, the lone pair of electrons on nitrogen that
favors s coordination to metals and hence prohibits the
p complexation required for activation of the carbon–nitro-
gen double bond. Of particular note is the copolymerization
of imines and CO, which has long been suggested to be a
potentially attractive route to polypeptides but is difficult to
realize owing to lack of appropriate catalysts.[5–8]
achievement constitutes a critical step toward metal-catalyzed
copolymerization of imines and CO. However, desired
subsequent insertion of CO into the resulting carbon–metal
bond failed to occur. The adjacent amide carbonyl strongly
coordinates to the metal center to form a stable five-
membered metallacycle that prevents coordination of incom-
ing CO. Tuning the ligands on palladium could partly solve
the problem, resulting in CO insertion to produce a single
amino acid unit.[7,9] However, facile elimination of metal
moieties from the organic scaffold took place. As a result,
even a simple dipeptide has never been obtained. Other
metals, such as nickel[10] and manganese,[11] have also been
tested, but none exhibited desired catalytic activity. There-
fore, a catalyst that allows continuous insertion of imines and
CO is still unknown.
We address this issue by employing cobalt rather than
palladium or other previously used metals. Cobalt is chosen
because it is one of the most frequently used catalysts for
carbonylation reactions,[4] and it has recently been success-
fully applied to catalyze copolymerization of aziridines and
CO.[12] It is now well-established that the active species in such
catalysis is acylcobalt, the chemistry of which has been
thoroughly studied.[13] Evidence also indicates that cobalt is
less prone to coordinate to an adjacent carbonyl group, which
will be critically important for the present reaction. Although
a phosphine-substituted acylcobalt complex has failed to
catalyze this reaction,[8] we find now that the simple acylcobalt
complex 1 without a phosphine ligand can effectively catalyze
the copolymerization of imines and CO under suitable
conditions, giving rise to the desired polypeptides as shown
in Equation (1).
Catalyst 1 was synthesized according to the reported
procedures.[12,13] Its ability to catalyze the copolymerization
In 1998, Sen and co-workers[5] and Arndsten and co-
workers[6] independently reported the first observation of
imine insertion into acyl carbon–palladium bonds. This
was first probed with the stable imine MeN CHC6H5 (2a),
=
which is available either commercially or from condensation
of benzaldehyde and methylamine. The polymerization was
performed in dioxane under 800 psi of CO pressure at 508C.
After removing the solvent and washing with hexane, the
corresponding polypeptide 3a was obtained as a solid product
(Table 1, entries 1 and 2). Analysis by 1H NMR spectroscopy
showed the presence of phenyl, N-methyl, and methine
protons at the expected positions,[14–16] which were confirmed
by 13C and HSQC NMR spectroscopy. The methine proton
exhibited a rather high chemical shift at about 6.50 ppm,
which is in accord with previous observations.[15] The amide
[*] Prof. H. Sun, J. Zhang, Q. Liu, L. Yu, J. Zhao
Department of Chemistry, Nankai University
94 Weijin Road, Tianjin 300071 (China)
Fax: (+86)22-2349-8132
E-mail: sunhl@nankai.edu.cn
[**] We thank Prof. J.-P. Cheng, Prof. J.-G. Tian, Prof. Q.-C. Wang, and
Prof. Y.-X. Che for their help. The authors also thank Dr. D. J.
Siegward and Prof. K. Kavallieratos for reviewing the manuscript,
and Prof. X.-Q. Zhu for valuable discussion. This work was
supported by NSFC (nos. 20372036 and 20421202), the Natural
Science Foundation of Tianjin (no. 05YFJMJC06800) and the
Ministry of Education of China (no. 03406).
1
carbonyl that did not appear in the H NMR spectrum was
observable in 13C NMR and IR spectra. The polypeptide
structure was finally established by MALDI-TOF MS anal-
ysis. The mass spectrum exhibited a series of peaks with the
same interval between them, which was exactly equal to the
Supporting information for this article is available on the WWW
under http://www.angewandte.org or from the author.
Angew. Chem. Int. Ed. 2007, 46, 6068 –6072
ꢀ 2007 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.angewandte.org 6069

Communications
Table 1: Cobalt-catalyzed copolymerization of imines and CO to produce
polypeptides.
Entry Imines[a] M/C[b]
t
Polypeptides Yield Mn ”10À3[c] PDI[d]
[h]
[%]
1
2
3
4
2a
2a
2a
2a
11:1
6
73
38
45
48
2.4
2.4
1.9
1.7
1.16
22:1 12
1.16
11:1[e]
11:1[f]
6
6
[g]
[g]
5
6
2b
2b
11:1
22:1 12
6
100
54
3.2
3.6
1.23
1.39
7
8
2c
2c
11:1 12
22:1 12
100
59
4.1
4.1
1.28
1.22
Figure 1. MALDI-TOF MS analysis of polypeptides. a) The spectrum of
3a (Table 1, entry 1) obtained in the reflection mode with 2,5-dihydrox-
ybenzoic acid (DHB) as the matrix. The inset is an expansion of the
m/z region from 770 to 940. b) Tentative assignment of the end group
structures.
[g]
[g]
9
10
2d
2d
11:1 12
22:1 12
75
31
1.9
1.8
11
12
13
14
15
16
2e
2e
2e
2e
2e
2e
22:1 12
44:1 12
66:1 12
100:1 24
100:1 12
100
100
100
100
87
8.2
11.8
18.5
28.8
22.1
11.0
1.29
1.22
1.18
1.18
1.20
1.27
weight polypeptides. Since it was initially thought that the
reaction might proceed via formation of N-acyliminium
intermediates from reaction of acylcobalt with imines,[8]
chain termination might occur through the well-known
nucleophilic addition of adjacent phenyl groups to the
cationic carbon atom of N-acyliminium.[17] However, the
fact that the copolymerization could proceed in benzene
suggests that this explanation is not the correct one. A
plausible pathway that does not involve N-acyliminium
intermediates is the so-called coordination mechanism, in
which an imine is first coordinated to cobalt and then
undergoes concerted intramolecular insertion into an acyl
carbon–cobalt bond via a four-centered transient state,
resulting in a five-membered metallacycle. A similar mech-
anism has been suggested by theoretical calculations for imine
insertion into acyl carbon–palladium bonds.[18] However, the
five-membered cobalt metallacycle must be less stable than
that of palladium, just as expected, and hence allow CO
coordination and insertion to produce the next generation of
acylcobalt intermediates. These intermediates have a vacant
coordination site at cobalt, which would be easily coordinated
by imines to begin the next round of chain propagation
(Scheme 1a).
The chain termination mechanism has been examined by
means of end-group analysis. 1H and 13C NMR spectra
showed the presence of the phenylacetyl end group inherited
from the catalyst, but the second end groups were not
detected. According to MALDI-TOF MS analysis, they were
certainly not carboxyl groups derived from hydrolysis of the
acylcobalt species[12] but were most likely consistent with the
imidazoline derivative shown in Figure 1b. Attempts to
isolate such end groups through hydrolysis of the polypep-
tides were not successful, but a small-molecule imidazoline 4
was obtained and completely characterized.[16] This com-
pound is obviously a product of 1,3-dipolar cycloaddition of
100:1
6
36
17
18
19
20
2 f
2 f
2 f
2 f
22:1 12
44:1 24
80:1 36
100:1 48
100
100
100
100
4.3
8.0
13.7
19.4
1.18
1.15
1.14
1.17
[a] Reactions were performed in dioxane at 508C under 800 psi of CO
pressure. [b] Monomer-to-catalyst molecular ratio. [c] Number- and
weight-average molecular weights (Mn and Mw, respectively) were
determined by GPC relative to polystyrene standards in THF. [d] Poly-
dispersity index, Mn/Mw. [e] Reaction in DME. [f] Reaction in benzene.
[g] Not determined due to closeness to lower molecular-weight limits.
mass of the repeating unit (Figure 1a). Hydrolysis of the
polypeptide quantitatively produced N-methyl phenylglycine,
which further confirms the structure of the polypeptide.[16]
The successful characterization of the polypeptide dem-
onstrates that repeated insertion of imines and CO has
occurred. Gel permeation chromatography (GPC) analysis
indicated a number-average molecular weight (Mn) of about
2000 dalton. The reaction conditions were optimized. It was
found that varying CO pressures from 600 to 1000 psi did not
significantly change the yield of the polypeptide. Raising the
temperature to 808C lead to reduced yields presumably owing
to catalyst decomposition, and reaction at room temperature
also resulted in decreased yields probably caused by chain
initiation problems. Other solvents, such as dimethoxyethane
(DME) and benzene, were unable to improve the reaction
(Table 1, entries 3and 4). It seems that less-polar solvents
were not favored by the copolymerization.
These results indicated that rapid chain termination is a
serious problem preventing formation of high-molecular-
6070 www.angewandte.org
ꢀ 2007 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Angew. Chem. Int. Ed. 2007, 46, 6068 –6072

Angewandte
Chemie
Mn > 4000 dalton (Table 1, entries 7 and 8). Interestingly,
=
even the very crowded mesityl imine MeN CHC6H2-2,4,6-
Me3 (2d) provided polypeptide 3d, albeit in lower molecular
weights and yields (Table 1, entries 9 and 10).
High-molecular-weight
polypeptides
were
finally
obtained using alkyl in place of aryl imines. Owing to the
low sensitivity of the reaction with respect to steric factors of
=
the substituent, pivalaldehyde imine MeN CH-tBu (2e),
which is very stable during prolonged storage, was used. The
reaction was performed under the same conditions as for aryl
imines. After removing the solvent, the desired polypeptide
3e was obtained as a solid product in quantitative yield.
Surprisingly, the quantitative conversion was maintained even
at high monomer-to-catalyst (M/C) molecular ratios (Table 1,
entries 11–14). The products were characterized by various
spectroscopic methods.[16] Analysis by 1H NMR spectroscopy
showed that the methine protons had much smaller chemical
shifts (less than 5.5 ppm) than in the aryl-substituent cases,
indicative of dramatic decrease of the acidity of the hydrogen
atom. This decrease must be the reason why chain termi-
nation has been largely suppressed. MALDI-TOF MS of the
low-molecular-weight samples gave a series of polypeptide
molecular-ion peaks (see Figure S35 in the Supporting
Information) whose exact masses were consistent with
Münchnone end groups rather than the imidazoline deriva-
tives observed for the aryl polypeptides. This finding could be
rationalized by the fact that the chain termination might occur
after complete consumption of the imine, thus making 1,3-
dipolar addition of the imine to Münchnone impossible.
GPC analysis revealed a low polydispersity index (PDI)
for the polypeptides. The molecular weights of the polypep-
tides increased roughly linearly with the increase of M/C
molecular ratios (see Figure S44 in the Supporting Informa-
tion); meanwhile, the PDIs improved gradually. Similar
changes of the PDIs were also observed during copolymer-
ization. This result is consistent with the slow chain initiation
(Scheme 1a), which is incompatible with the relatively fast
chain propagation. Replacing one of the methyl groups of the
Scheme 1. Possible mechanism of cobalt-catalyzed alternating copoly-
merization of imines and CO. a) Proposed mechanism for chain
initiation by CO dissociation from catalyst 1, which is a well-estab-
lished slow process occurring on a time scale of hours,[12,13] and of
chain propagation through imine coordination to cobalt followed by
intramolecular migratory insertion. b) Proposed chain-termination
pathway, as exemplified by the formation of imidazoline 4. Further
transformation of imidazolines to downstream derivatives might occur
under appropriate conditions.
imines and mesoionic Münchnones.[7] The latter should have
been formed through b-hydrogen elimination of the acylco-
balt intermediate and subsequent cyclization of the resulting
ketene (Scheme 1b). Similar b-hydrogen elimination to form
ketenes has been found to occur extensively in palladium
systems.[7,9] In the case of cobalt, however, this was the first
time such a reaction has been observed. Nevertheless, the
reverse process (addition of hydridocobalt to ketenes) has
been reported.[19] Although further work is required for
complete clarification of the chemistry involved here, it seems
to be of little doubt that the second end group should be
derived from the imidazolines after loss of carbon dioxide and
hydrogen as a result of oxidation upon exposure to air.[16]
As b-hydrogen elimination is the key step of chain
termination, the high acidity of the methine hydrogen atom,
revealed by the large chemical shift of the proton, must be
responsible for the facile chain termination during polymer-
ization. Thus, when a methyl group was introduced onto the
=
tert-butyl substituent in 2e with an ethyl group in imine MeN
CHC(Me)2Et (2 f) reduced the rate of polymerization,
probably for steric reasons, while the quantitative trans-
formation to polypeptide 3 f was also observed (Table 1,
entries 17–19). Linear change of the molecular weights with
the M/C ratios was demonstrated again (see Figure S45 in the
Supporting Information). The PDIs of the polymers were
obviously lower than before the introduction of the ethyl
group. This finding may be explained by the lowered chain
propagation rate that becomes more compatible with the slow
chain initiation. Noting the behaviors of the copolymeriza-
tion, including quantitative conversion, low PDIs, and linear
increase of molecular weight, the tertiary alkyl imine systems
seemingly meet most of the criteria for a living polymeri-
zation process, although attempts to synthesize block copoly-
mers by adding a second imine monomer were not successful,
probably because of the instability of the active catalytic
species after consumption of the monomers.[20,21]
=
phenyl ring in the imine MeN CHC6H4-p-Me (2b), a sudden
increase in the yield and the molecular weight of polypeptide
3b was observed (Table 1, entries 5 and 6). This improvement
is obviously a result of the electron-donating effect of the
methyl group, which retards chain termination by lowering
the acidity of the hydrogen atom. Further increase of the
electron-releasing ability using the p-methoxy group in imine
The success in modifying the tertiary alkyl group leaves
room for further introduction of functional groups. Interest-
ingly, the N- and C-substituted polypeptides obtained above
=
MeN CHC6H4-p-OMe (2c) resulted in polypeptide 3c with
Angew. Chem. Int. Ed. 2007, 46, 6068 –6072
ꢀ 2007 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
www.angewandte.org
6071

Communications
The resulting solution was transferred to a flask, and the solvent was
removed under vacuum to afford the crude product polypeptide.
cannot, for steric reasons, be accessed by other means,
including the most frequently used ring-opening polymeri-
zation of amino acid N-carboxyanhydrides.[22–25] These poly-
peptides, unlike the N-unsubstituted ones, are soluble in
common organic solvents, such as THF and chloroform. One
of the unique properties of these polymers is that they can be
facilely degraded by trifluoroacetic acid (TFA).[16] Although
TFA cleavage of small peptides has been reported
recently,[26,27] polypeptide degradation by TFA has never
been observed. The ultimate products were found to be a
mixture of amino acid 5 and dipeptide 6, for the case of the
tert-butyl polypeptide 3e, with regiospecific deuteration on
the a-carbons when the degradation was performed in
[D1]TFA. This result is consistent with the mechanism
involving cleavage of peptide bonds by adjacent carbonyl
Received: February 12, 2007
Revised: April 16, 2007
Published online: July 3, 2007
Keywords: carbon monoxide · copolymerization ·
.
homogeneous catalysis · imines · polypeptides
[1] N. Sewald, H.-D. Jakubke, Peptides: Chemistry and Biology,
Wiley-VCH, Weinheim, 2002.
[2] M. Bodanszky, A. Bodanszky, Principles of Peptide Synthesis,
2nd ed., Springer, New York, 1994.
[3] T. J. Deming, Adv. Mater. 1997, 9, 299 – 311.
[4] Applied Homogeneous Catalysis with OrganometallicCom-
pounds, 2nd ed. (Eds.: B. Cornils, W. A. Herrmann), Wiley-
VCH, Weinheim, 2002.
oxygen atoms to form Münchnone intermediates (see
[28]
Scheme S3in the Supporting Information).
Such unique
degradation properties would be useful for applications of
special functional materials.
[5] S. Kacker, J. S. Kim, A. Sen, Angew. Chem. 1998, 110, 13 3 5 –
1337; Angew. Chem. Int. Ed. 1998, 37, 1251 – 1253.
[6] R. D. Dghaym, K. J. Yaccato, B. A. Arndtsen, Organometallics
1998, 17, 4 – 6.
[7] R. D. Dghaym, R. Dhawan, B. A. Arndtsen, Angew. Chem. 2001,
113, 3328 – 3330; Angew. Chem. Int. Ed. 2001, 40, 3228 – 3230.
[8] J. K. Funk, H. Yennawar, A. Sen, Helv. Chim. Acta 2006, 89,
1687 – 1695.
[9] R. Dhawan, R. D. Dghaym, B. A. Arndtsen, J. Am. Chem. Soc.
2003, 125, 1474 – 1475.
[10] J. L. Davis, B. A. Arndtsen, Organometallics 2000, 19, 4657 –
4659.
[11] D. Lafrance, J. L. Davis, R. Dhawan, B. A. Arndtsen, Organo-
metallics 2001, 20, 1128 – 1136.
[12] L. Jia, H. Sun, J. T. Shay, A. M. Allgeier, S. D. Hanton, J. Am.
Chem. Soc. 2002, 124, 7282 – 7283.
[13] V. Galamb, G. Palyi, F. Ungvary, L. Marko, R. Boese, G. Schmid,
J. Am. Chem. Soc. 1986, 108, 3344 – 3351.
[14] M-X. Wang, J. Liu, D-X. Wang, Q-Y. Zheng, Tetrahedron:
Asymmetry 2005, 16, 2409 – 2416.
[15] H. O. Bayer, R. Huisgen, R. Knorr, F. C. Schaefer, Chem. Ber.
1970, 103, 2581 – 2597.
[16] See the Supporting Information.
[17] W. N. Speckamp, M. J. Moolenaar, Tetrahedron 2000, 56, 3817 –
3856.
[18] L. Cavallo, J. Am. Chem. Soc. 1999, 121, 4238 – 4241.
[19] F. Ungvary, J. Organomet. Chem. 1986, 303, 251 – 255.
[20] G. W. Coates, P. D. Hustad, S. Reinartz, Angew. Chem. 2002, 114,
2340 – 2361; Angew. Chem. Int. Ed. 2002, 41, 2236 – 2257.
[21] R. R. Schrock, Acc. Chem. Res. 1990, 23, 158 – 165.
[22] H. R. Kricheldorf, a-Aminoacid-N-Carboxyanhydrides and
Related Materials, Springer, New York, 1987.
We have shown that the highly desirable copolymerization
of imines and CO has been realized through proper choice of
a simple cobalt catalyst, demonstrating once again the ability
of metal catalysis to manipulate organic transformations.[29]
Of particular note are the ready availability and low cost of
the starting materials as well as the atom-economic feature of
the reaction process (see Scheme S1 in the Supporting
Information), which render the reaction well-suited for
large-scale production of the polypeptide materials. In fact,
this method is the shortest possible route for chemical (or
abiotic) synthesis of polypeptides. Whether such a copoly-
merization strategy might have been adopted by nature for
the prebiotic origin of polypeptides might deserve further
investigation.[30] Although the present study has been con-
fined to the use of stable imines, their successful copolymer-
ization with CO has indicated that this transformation can be
both thermodynamically and kinetically viable. Further
expandsion of the scope of this reaction and full elucidation
of the reaction mechanism are the subject of current research.
[23] A. Cosani, M. Palumbo, M. Terbojevich, E. Peggion, Macro-
molecules 1978, 11, 1041 – 1045.
[24] D. G. H. Ballard, C. H. Bamford, J. Chem. Soc. 1958, 355 – 360.
[25] T. J. Deming, Nature 1997, 390, 386 – 389.
[26] J. Urban, T. Vaisar, R. Shen, M. S. Lee, Int. J. Pept. Protein Res.
1996, 47, 182 – 189.
[27] M. Goodman, C. J. Creighton, US patent 6,417,397, 2002.
[28] M. Teixido, F. Albericio, E. Giralt, J. Pept. Res. 2005, 65, 153–
166.
Experimental Section
Imine (about 1 mL) was added to a Par pressure reactor containing a
suitable amount of catalyst 1 in dioxane (50 mL), which had been pre-
pressurized under 800 psi of CO overnight. After the addition, the
pressure of CO was returned to 800 psi, and the reactor was heated in
an oil bath at 508C while the reaction mixture was magnetically
stirred for the period of time specified in Table 1. After cooling to
room temperature, the pressure was released and the reactor opened.
[29] T. M. Trnka, R. H. Grubbs, Acc. Chem. Res. 2001, 34, 18 – 29.
[30] L. Leman, L. Orgel, M. R. Ghadiri, Science 2004, 306, 283– 286.
6072 www.angewandte.org
ꢀ 2007 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
Angew. Chem. Int. Ed. 2007, 46, 6068 –6072