The unexpected conversion of 1,5,8,12-tetraazadodecane-glyoxal bisaminal into its amidinium salt in an acidic medium

Article abstract of DOI:10.1002/ejoc.200390026

When refluxed in aqueous hydrochloric acid solution, 1,5,8,12-tetraazadodecane-glyoxal bisaminal is quantitatively converted into its amidinium salt, which is a lactam precursor. The mechanism of this unexpected reaction is discussed. Wiley-VCH Verlag GmbH & Co, KGaA, 69451 Weinheim, Germany, 2003.

Full text of DOI:10.1002/ejoc.200390026

SHORT COMMUNICATION
The Unexpected Conversion of 1,5,8,12-Tetraazadodecane-Glyoxal Bisaminal
into Its Amidinium Salt in an Acidic Medium
[a]
[a]
[a]
[a]
´ `
Helene Bernard, Eugenia Del Rio Garcia, Sonia Ferec, Nelly Kervarec, and
Henri Handel*[a]
Keywords: Bisaminals / Amines / Nitrogen heterocycles / Amides
When refluxed in aqueous hydrochloric acid solution,
1,5,8,12-tetraazadodecane-glyoxal bisaminal is quantitat-
ively converted into its amidinium salt, which is a lactam pre-
cursor. The mechanism of this unexpected reaction is discus-
sed.
( Wiley-VCH Verlag GmbH & Co. KGaA, 69451 Weinheim,
Germany, 2003)
Introduction
usually reversible and a simple acidic hydrolysis allows the
recovery of the starting material.[7] However, we observed
that a lactam was also isolated when the bisaminal 2 was
treated with aqueous hydrochloric acid (Figure 2).
In this paper we describe the isolation and characteriza-
tion of new intermediates involved in the formation of the
lactam resulting from 2; the observation of the reversibility
of these reactions allowed us to propose a mechanism for
this sequence.
Owing to its recognised applications in medical imaging,
1,4,7,10-tetraazacyclododecane (cyclen) has gained a pre-
ponderant position among tetraazamacrocycles.[1] The
complex of its 1,4,7,10-tetraacetate derivative (DOTA) with
the gadolinium() ion (DOTAREM) is a well-known
MRI contrast agent and was one of the first compounds
available for this purpose. The recently proposed new gen-
eration ligands of this type have different functionalities on
the nitrogen atoms, therefore the mono N-substitution of
cyclen is an important step for the synthesis of these chelat-
ing agents.[2]
Results and Discussion
Compound 2 is easily obtained by the condensation of
aqueous glyoxal with N,NЈ-bis(3-aminopropyl)ethylenedia-
mine in ethanol solution as described previously.[8] When 2
was hydrolysed in a strongly acidic medium (11  HCl) the
reaction led to a unique product, 3, which is stable under
these conditions. The 13C NMR spectrum of this com-
pound is significantly different from the one expected for
the deprotected linear tetraamine. In fact, ten different
peaks are observed and, among them, one corresponds to
an sp2 carbon. The whole structure was established by
Bisaminals of cyclic tetraamines have proven to be effici-
ent protection tools for the synthesis of mono N-alkylated
ligands: when the reaction is applied to cyclen, it consists
of an alkylation of cyclen-glyoxal 1a followed by treatment
of the resulting quaternary ammonium salt to release the
free mono N-alkylated cyclen.[3,4] Hydrolysis by treatment
with an aqueous sodium hydroxide solution or refluxing in
hydrazine hydrate usually gives good results. Recently, an
unexpected reaction involving the glyoxal-protected cyclen
1a was described in a Bracco SPA patent:[5] the hydrolysis
of this compound between pH 5 and 9 led to the lactam 1b,
which was easily converted into cyclen monoacetate 1c (Fig-
ure 1).
1
1
NMR correlation sequences (HMBC H-15N, HMBC H-
1
1
13C, TOSCY H-1H, HMQC H-13C).
The first significant structural information was gained
from the HMBC 1H-15N correlation (Table 1, compound
3), which shows two nitrogen atoms of similar nature and
possessing a marked sp2 character. They appear at δ ϭ
Ϫ267 and Ϫ273 ppm relative to nitromethane as external
reference. Furthermore, the other two nitrogen atoms res-
onate at δ ϭ Ϫ350 and Ϫ335 ppm, which indicates their sp3
character. The single signal at δ ϭ 4.41 ppm, integrating for
two protons, provided us with further important informa-
tion as it correlates with all the nitrogen atoms except the
Bisaminals of linear tetraamines have been used success-
fully for the synthesis of cyclic adducts.[6] The formation of
the intermediate is easily achieved by direct reaction of the
tetraamine with a dicarbonyl compound. This reaction is
[a]
University of Bretagne Occidentale, UMR-CNRS 6521,
6 avenue Le Gorgeu B.P. 809, 29285 Brest
Fax: (internat.) ϩ33-(0)2/9801-6594
E-mail: henri.handel@univ-brest.fr
Eur. J. Org. Chem. 2003, 255Ϫ258  2003 WILEY-VCH Verlag GmbH & Co. KGaA, 69451 Weinheim 1434Ϫ193X/03/0102Ϫ0255 $ 20.00ϩ.50/0
255

H. Bernard, E. Del Rio Garcia, S. Ferec, N. Kervarec, H. Handel
SHORT COMMUNICATION
Figure 1. Hydrolysis of glyoxal-protected cyclen to cyclen monoacetate
Figure 2. Hydrolysis sequence of compound 2
Figure 3. Atom numbering of compounds 3Ϫ5
1
one at δ ϭ Ϫ350 ppm; these observations are consistent
This structure was corroborated by the TOCSY H-1H
with the lack of a bond between the C14 and N1 atoms. The sequence (Table 2), which highlighted the interaction of the
other correlations are consistent with the amidinium ion 3 protons borne by the carbon atoms C14 and C11. This result
depicted in Figure 2.
implies a bond between C13 and N12.
256
Eur. J. Org. Chem. 2003, 255Ϫ258

Unexpected Conversion of 1,5,8,12-Tetraazadodecane-Glyoxal Bisaminal
SHORT COMMUNICATION
1
Table 1. HMBC H-15N spectroscopic data for compounds 3Ϫ5
3
4
5
δ1 ppm (C)
δ15 ppm (N)
δ1 ppm (C)
δ15 ppm (N)
δ1 ppm (C)
δ15 ppm (N)
H
N
H
N
H
N
2.05 (10)
2.12 (3)
3.07 (2)
3.26 (4)
3.40 (11)
3.53 (9)
3.55 (6)
3.75 (7)
4.41 (14)
Ϫ267 (8); Ϫ273 (12)
Ϫ335 (5); Ϫ350 (1)
Ϫ273
1.53 (3)
1.77 (10)
2.32 (4)
2.52 (6)
2.65 (2)
2.94 (14)
2.99 (7)
3.01 (9)
3.23 (11)
Ϫ340 (5); Ϫ360 (1)
Ϫ190 (12); Ϫ300 (8)
Ϫ300
1.26 (3)
1.32 (10)
2.08 (4)
2.30 (11, 14)
2.37 (2)
2.73 (6)
Ϫ361 (1); Ϫ343 (5)
Ϫ362 (12); Ϫ265 (8)
Ϫ362; Ϫ265
Ϫ361
Ϫ343; Ϫ265
Ϫ343; Ϫ265
Ϫ265
Ϫ267
Ϫ360
Ϫ267; Ϫ335
Ϫ267; Ϫ335
Ϫ267; Ϫ273; Ϫ335
Ϫ190; Ϫ300; Ϫ340
Ϫ300; Ϫ340
Ϫ300
2.94 (13)
3.09 (9)
1
Table 2. TOCSY H-1H spectroscopic data for compound 3
which readily evolved to another product 5 (subsequently
identified as an amide). The analysis of the NMR sequences
is in agreement with the structures 4 and 5 (Table 1 3 4).
The free amidine 4 can also be prepared quantitatively
by dehydration of the amide by stirring and refluxing a mix-
ture of 5 and alumina in toluene in a DeanϪStark appar-
atus. Confirmation of the structures of 4 and 5 was ob-
tained upon their reduction by BH3·SMe2, which led to the
well-known and commercially available piperazine derivat-
ive 6 (Figure 2).[9]
3
δ1 ppm (C)
δ1 ppm
H
H
2.05 (10)
2.12 (3)
3.07 (2)
3.26 (4)
3.40 (11)
3.53 (9)
3.55 (6)
3.75 (7)
4.41 (14)
3.4; 3.53; 4.41(w)
3.07; 3.26
2.12; 3.26
2.12; 3.07
2.05; 3.53; 4.41
2.05; 3.40
3.75; 4.41
3.55; 4.41
A probable mechanism for the formation of the amide
2.05(w); 3.40; 3.55; 3.75 during the hydrolysis sequence is proposed in Figure 2.
Normally, complete hydrolysis of the bisaminal to the free
tetraamine results from the first acid-catalysed attack of the
Passing 3 through an anionic exchange resin-packed col- two aminal carbons by two molecules of water. The forma-
umn gave the expected dehydrochlorinated amidine 4, tion of the amidinium salt implies the breaking of a single
Table 3. HMQC correlation data (1J,1H-13C) spectroscopic data for compounds 3Ϫ5
3
4
5
δ1 ppm
δ13 ppm (C)
δ1 ppm
δ13 ppm (C)
δ1 ppm
δ13 ppm (C)
H
C
H
C
H
C
2.05
2.12
3.07
3.26
3.40
3.53
3.55
3.75
4.41
20.7 (10)
24.6 (3)
39.2 (2)
56.6 (4)
41.1 (11)
50.3 (9)
50.8 (6)
48 (7)
1.53
1.77
2.32
2.52
2.65
2.94
2.99
3.01
3.23
30.4 (3)
21.2 (10)
55.5 (4)
50.9 (6)
40.3 (2)
57.9 (14)
49.1 (7)
46.9 (9)
43.3 (11)
152.6 (13)
1.26
1.32
2.08
2.30
2.30
2.37
2.73
2.94
3.09
30.3 (3)
30.3 (10)
55.3 (4)
38.9 (11)
49.8 (14)
40.2 (2)
57.4 (6)
46.3 (13)
43.3 (9)
167.1 (7)
51.3 (14)
155.3 (13)
1
Table 4. HMBC H-13C spectroscopic data for compounds 3Ϫ5
3
4
C
5
C
δ1 ppm
δ13 ppm
δ1 ppm
δ13 ppm
δ1 ppm
δ13 ppm
H
C
H
H
2.05
2.12
3.07
3.26
3.40
3.53
3.55
3.75
4.41
41.1; 50.3
39.2; 56.6
24.6; 56.6
24.6; 39.2; 50.8; 51.3
20.7; 50.3
20.7; 41.1; 155
48
50.8; 155
1.53
1.77
2.32
2.52
2.65
2.94
2.99
3.01
3.23
40.3; 55.5
43.3; 46.9
30.4; 40.3; 50.9; 57.9
49.1; 55.5; 57.9
30.4; 55.5
50.9; 55.5; 152.7
46.9; 50.9; 152.7
21.2; 43.3; 49.1; 152.7
21.2; 46.9; 152.7
1.26
1.32
2.08
2.30
2.37
2.73
2.94
3.09
40.3; 55.3
38.9; 43.3
30.3; 40.2; 49.8; 57.3
30.3; 43.3; 46.3; 55.3; 57.3
30.3; 55.3;
49.8; 55.3; 167.1
49.8; 167.1
30.3; 38.9; 46.3; 167.1
50.8; 56.6; 155
Eur. J. Org. Chem. 2003, 255Ϫ258
257

H. Bernard, E. Del Rio Garcia, S. Ferec, N. Kervarec, H. Handel
SHORT COMMUNICATION
until a hygroscopic white solid was isolated. The solid was then
filtered off and washed with small amounts of absolute ethanol;
compound 3 (1.14 g, 3.75 mmol) was isolated in 75% yield. NMR
spectroscopic data are gathered in Table 1Ϫ4
bis-aminal bridge with production of a cation and migra-
tion of a hydrogen aminal atom. The amidinium salt 3 is
remarkably stable in acidic medium; however, after dehy-
drochlorination the resulting amidine 4 is rapidly hydro-
lysed into the six-membered lactam 5. This reaction is re-
versible and the amidinium salt 3 is easily regenerated in
acidic medium. However, the expected corresponding am-
ino acid sodium salt was detected after treatment of 5 with
alkaline medium. This end compound, which is difficult to
isolate, rapidly converts into the lactam 5 and the amidin-
ium salt 3 in acidic medium.
Procedure for the Preparation of 5: Compound 3, dissolved in a
minimum of water, was passed through a column packed with an-
ionic exchange resin Amberlyst A-26 (3 g of resin for 1 g of chlo-
rohydrate). A mixture of free amidine 4 and amide 5 was obtained,
which converted into amide 5, isolated in 93% yield, after evapora-
tion of the water. NMR spectroscopic data are gathered in Tables
1, 3 and 4. C10H22N4O (214): calcd. C 56.07, H 10.28, N 26.17;
found C 56.00, H 10.27, N 26.27.
When hydrolysis of 2 was performed in dilute acidic solu-
tion, we noticed that the reaction was slower; formation of
the free tetraamine was never observed. To the best of our
knowledge this behaviour is unique among the other bisam-
inals formed by the condensation of linear tetraamines and
glyoxal. Although the corresponding amidinium salt was
not identified, the previously mentioned hydrolysis of
cyclen-glyoxal is probably the consequence of a similar
mechanism.
Procedure for the Preparation of 4: Compound 5 (0.214 g, 1 mmol)
was dissolved in a minimum of water. Alumina (10 mmol) and tolu-
ene (50 mL) were then added. The mixture was refluxed for 48 h
and a DeanϪStark trap was used to collect the water. After cool-
ing, the suspension was filtered and the solid was repeatedly washed
with a mixture of dichloromethane and methanol (9:1). The solv-
ents were evaporated to give 4 (0.141 g, 0.72 mmol) in 72% yield.
NMR spectroscopic data are gathered in Tables 1, 3 and 4.
Procedure for the Preparation of 6 from 5: An excess of BH3·SMe2
(2 mmol) was added to a solution of 5 (1 mmol) in 20 mL of THF.
The mixture was refluxed under a nitrogen atmosphere for 48 h.
After cooling, the unchanged BH3·SMe2 was destroyed by slow ad-
dition of methanol, and the solvents then evaporated to yield a
white solid. This solid was taken up in 10% aqueous HCl (20 mL)
and refluxed overnight. After cooling, the pH was raised to 14 with
NaOH pellets and the product extracted with CH2Cl2 (3 ϫ 20 mL).
After drying (MgSO4) and solvent evaporation, the well-known
and commercially available piperazine derivative 6 was isolated in
80% yield. 13C NMR (100.61 MHz, CDCl3, 298 K): δ ϭ 56.2, 53.0;
40.5 (8 CαϪN), 30.2 (2 CβϪN) ppm.
Further investigations concerning the extension of this
reaction to bisaminals formed from other α-dicarbonyl
compounds are in progress.
Experimental Section
General: All reagents were of commercial quality and solvents were
dried using standard procedures. Elemental analyses were per-
formed at the Service de Microanalyse, CNRS, 91198 Gif sur
Yvette, France and at the Service Central d’Analyse, CNRS, B. P.
22, 69390 Vernaison, France.
NMR Spectral Studies
Acknowledgments
´
We are grateful to the Region Bretagne for financial support.
2D NMR spectra were recorded in CDCl3 (compounds 4 and 5)
or D2O (compound 3) at 298 K on a DRX Avance 500 Bruker
spectrometer equipped with an indirect triple TBI 1H{BB}13C
5 mm probehead operating at 500.13 MHz for 1H, 125.77 MHz for
13C and 50.68 MHz for 15N. Heteronuclear multiple quantum co-
herence (HMQC), heteronuclear multiple band coherence (HMBC;
60 ms mixing time) 1H-13C and heteronuclear multiple band coher-
[1]
R. B. Lauffer, Chem. Rev. 1987, 87, 901Ϫ927.
P. Caravan, J. J. Ellison, T. J. McMurry, R. B. Lauffer, Chem.
Rev. 1999, 99, 2293Ϫ2352.
J. Kotek, P. Hermann, P. Vojtisek, J. Rohovec, I. Lukes, Collect.
Czech. Chem. Commun. 2000, 65, 243Ϫ266.
M. Le Baccon, F. Chuburu, L. Toupet, H. Handel, M. Soibi-
net, I. Deschamps-Olivier, J. P. Barbier, M. Aplincourt, New J.
Chem. 2001, 25, 1168Ϫ1174.
Bracco S. P. A., International Patent, WO 99/05145, 1999.
G. Herve, H. Bernard, N. Le Bris, M. Le Baccon, J. J. Yaouanc,
H. Handel, Tetrahedron Lett. 1999, 40, 2517Ϫ2520.
L. Duhamel, in The Chemistry of Functional Groups, supple-
ment F (Ed.: S. Patai), Wiley, New York, 1982, p. 849
[2]
[3]
[4]
1
ence (HMBC; mixing time of 60 and 100 ms) H-15N spectra were
recorded using standard pulse sequences. For example, in a typical
HMBC experiment the raw data set consisted of 512(F2) ϫ 233(F1)
complex data points zero-filled to 512 in the F1 dimension prior to
Fourier transform, with a spectral width of 19490 and 1498 Hz in
the F1 and F2 dimensions, respectively.[10] The atom labelling of
compounds 3Ϫ5 is shown in Figure 3.
[5]
[6]
´
[7]
[8]
´
G. Herve, N. Le Bris, H. Bernard, J. J. Yaouanc, H. des Ab-
bayes, H. Handel, J. Organomet. Chem. 1999, 585, 259Ϫ265.
R. C. Northrop, Jr., P. L. Russ, J. Org. Chem. 1975, 40,
558Ϫ559.
The spectra are available from the authors on request.
Received September 12, 2002
Procedure for the Synthesis of 3: Compound 2 (0.98 g, 5 mmol) was
added to 10 mL of 11  hydrochloric acid and the reaction mixture
was stirred for 2 h. After evaporation of the solvent a brown oil
was obtained. Absolute ethanol (10 mL) was then added and re-
moved under vacuum. This operation was repeated several times
[9]
[10]
[O02510]
258
Eur. J. Org. Chem. 2003, 255Ϫ258