A pH-dependent cyanate reactivity model: Application to preparative N-carbamoylation of amino acids

Article abstract of DOI:10.1039/b005856o

Recent developments in peptide synthesis have underlined the importance of optimising, on a preparative scale, the N-carbamoylation of amino acids by aqueous cyanate. To this purpose, a theoretical model of aqueous cyanate reactivity was designed. The parameters of the model were evaluated, for various pH and temperatures, from a critical survey of the literature, together with additional experimental data. Computer-simulated kinetics based on this model showed the reaction efficiency to be significantly dependent on pH, and suggested optimum conditions to be moderate temperatures and pH 8.5-9. Discussion of the practical convenience of these theoretical results led us to prefer 40-50 °C and a pH range of 7-8 as reaction conditions, thus maintaining reaction times within a few hours. Various N-carbamoyl amino acids (ureido derivatives of glycine, L-valine, L-alanine, L-leucine, DL-methionine, N^ε-trifluoroacetyl-L-lysine, β-alanine) were thus successfully synthesised on the gram to kilogram scales.

Full text of DOI:10.1039/b005856o

View Article Online / Journal Homepage / Table of Contents for this issue
A pH-dependent cyanate reactivity model: application
to preparative N-carbamoylation of amino acids
Jacques Taillades,* Laurent Boiteau, Isabelle Beuzelin, Olivier Lagrille, Jean-Philippe Biron,
Willy Vayaboury, Odile Vandenabeele-Trambouze, Olivia Giani and Auguste Commeyras
2
Organisation Moléculaire – Evolution et Matériaux Fluorés, Chemistry Department, University
of Montpellier 2, CC 017 – Place E. Bataillon – 34095 Montpellier cedex 5, France.
E-mail: lboiteau@univ-montp2.fr
Received (in Cambridge, UK) 20th July 2000, Accepted 30th April 2001
First published as an Advance Article on the web 7th June 2001
Recent developments in peptide synthesis have underlined the importance of optimising, on a preparative scale,
the N-carbamoylation of amino acids by aqueous cyanate. To this purpose, a theoretical model of aqueous cyanate
reactivity was designed. The parameters of the model were evaluated, for various pH and temperatures, from a
critical survey of the literature, together with additional experimental data. Computer-simulated kinetics based on
this model showed the reaction efficiency to be significantly dependent on pH, and suggested optimum conditions
to be moderate temperatures and pH 8.5–9. Discussion of the practical convenience of these theoretical results led
us to prefer 40–50 ЊC and a pH range of 7–8 as reaction conditions, thus maintaining reaction times within a few
hours. Various N-carbamoyl amino acids (ureido derivatives of glycine, -valine, -alanine, -leucine, -methionine,
Nε-trifluoroacetyl--lysine, β-alanine) were thus successfully synthesised on the gram to kilogram scales.
Though other organic synthetic methods exist, such as trans-
Introduction
carbamoylation from urea degradation,3 hydantoin hydrolysis
In recent decades the N-carbamoylation reaction, especially
by strong bases4 or enzymatic systems,5 the simplest and
of amino acids, has been considered to be mostly relevant to
cheapest way to prepare CAAs is by reacting the free amino
biochemistry and metabolic pathways. Interest in this reaction
acid with buffered aqueous mineral cyanate.6,7 Most authors
has, however, been renewed since the recent discovery of a new
have worked at pH around 5–6, which appears not to be the
synthetic route to amino acid N-carboxyanhydride (NCA),
optimum conditions, since the N-carbamoylation reaction is in
competition with cyanate hydrolysis and urea formation. While
starting from N-carbamoyl amino acids (CAAs)1 (Scheme 1).
cyanate hydrolysis only requires the use of a larger excess of
cyanate, urea formation (a consequence of cyanate hydrolysis)
can be a major disadvantage for further product purification,
especially when the CAA is very soluble in water. All reaction
rates are highly pH-dependent, which led us to reconsider their
kinetics and to draw up a global cyanate reactivity model,
which was examined by means of a computer-simulated time
evolution.
Cyanate reactivity kinetics
It was not our intention to carry out a complete kinetic study of
cyanate reactivity in aqueous solution, the kinetics and mech-
anism of which have been studied by various authors during
the past decades, though under rather different conditions. Our
objective was rather to obtain a workable kinetic model over a
large pH range for temperatures above 40–50 ЊC, a temperature
range that appeared suitable for amino acid N-carbamoylation
after a series of preliminary experiments.
Scheme 1
Applied to N-carbamoyl peptides or to monoalkylureas, the
same reaction leads to the free amino function,2 thus offering
possibilities of using carbamoylation as an N-protective group
strategy, whereas the CAAs and alkylureas were considered as
very stable compounds, the hydrolysis of which requires drastic
basic conditions or the use of enzymes. Both strategies require
mastering N-carbamoylation on a preparative scale. However
in the abundant literature on the subject (mostly biochemical
studies), the synthetic procedures described provide little detail
concerning the efficiency and selectivity of the reaction, or the
purification of the products, hence the need to reconsider the
subject.
It is now accepted that the reactivity of cyanate with water, as
well as ammonia and amines, actually involves a protic or
nucleophilic attack of isocyanic acid, forming carbamates or
ureas. This slow-rate step is followed by fast decomposition of
the carbamates into ammonia and carbon dioxide, whereas
ureas are much more stable. All reactions have been found to be
first-order with respect to each of their reactants. In addition,
2Ϫ
the carbonate anion CO3 strongly catalyses HNCO acid
hydrolysis,8 the rate increment also being first-order with
2Ϫ
respect to HNCO and CO3
.
In the scope of this study, the possible reversal of reactions
(1)–(6) will be neglected: since ammonia and carbon dioxide
DOI: 10.1039/b005856o
J. Chem. Soc., Perkin Trans. 2, 2001, 1247–1254
This journal is © The Royal Society of Chemistry 2001
1247

View Article Online
concentrations remain low, carbamate formation due to them
remains limited; (alkyl)urea hydrolysis is very slow, and is
negligible below 50–60 ЊC for reaction times not exceeding a
few hours. The general rate law can thus be written as eqn. (7),
with the rate constants k1–k6 relating to reactions (1)–(6)
respectively.
(1)
(2)
(3)
Fig. 1 Apparent rates at 50 ЊC, as a function of pH, of glycine N-
carbamoylation k6a (solid line; squares: experimental values from Table
5), and of other cyanate reactions (dotted lines, calculated using data
a
from Tables 3 and 4): cyanate hydrolysis k0 (A), carbonate catalytic
increment k4a (B), urea formation k5a (C).
(9)
a
k0 thus representing the apparent rate of cyanate hydrolysis
(including acid- and base-catalysis). The apparent rate con-
(4)
(5)
(6)
stants kia are related to the ki and kiЈ rate constants, and to the
respective dissociation constants KCy, Kw, K1Car, K2Car, KAm
,
K1AA, K2AA (of cyanic acid, water, 1st and 2nd dissociation
of carbonic acid, ammonium, and 1st and 2nd dissociation
of amino acid respectively). With h standing for the H3Oϩ con-
a
a
centration, k0 –k6 are respectively expressed according to
eqn. (10)–(13), in which the relationships between ki and the
respective kiЈ are also made explicit. The pH-dependent shapes
of these apparent rates are exemplified in Fig. 1.
(10)
v = [HNCO] × (k1 × [H3Oϩ] ϩ k2 ϩ k3 × [HOϪ] ϩ k4 ×
[CO32Ϫ] ϩ k5 × [NH3] ϩ k6 × [RNH2]) (7)
Considering proton transfer in acid–base equilibria to
be much faster than reactions (1)–(6), the rate law can be
equivalently rewritten as eqn. (8), in terms of cyanate anion
concentration and related rate constants k2Ј–k5Ј.
v = [HNCO] × k1 ϩ [NCOϪ] × (kЈ2 × [H3Oϩ] ϩ kЈ3 ϩ kЈ4 ×
[HCO3Ϫ] ϩ kЈ5 × [NH4ϩ] ϩ kЈ6 × [RNH3ϩ]) (8)
(11)
(12)
(13)
The different rate constants have been determined at various
temperatures by different authors, who presented them either as
ki or as kiЈ. These data and references are compiled in Table 1.
However due to the scarcity of data within the temperature
range we are interested in, an extrapolation was necessary, as
well as a critical analysis of literature data, since the two series
are not equivalent in practice. We performed some comple-
mentary measurements at pH 5–6.5 (Table 2), a pH range where
reaction (2) is highly preponderant; our results proved to be in
agreement with the data from the literature.
Apparent rates
So that it can be more easily used in a pH-dependent kinetic
model, the rate law is best expressed as a function of the total
species concentrations (acid ϩ base forms), and of the pH-
dependent apparent rate constants kia. Thus, with NCOt, CO3 ,
t
NH3t and AAt standing for the total concentrations of cyanate,
carbonate, ammonia and amino acid respectively, the rate law is
expressed as follows in eqn. (9),
1248
J. Chem. Soc., Perkin Trans. 2, 2001, 1247–1254

Table 1 Kinetic data from the literature for the cyanate degradation model (the water activity is aggregated into the constants)
T /ЊC
log(k1/L molϪ1 sϪ1
)
log(k2Ј/L molϪ1 sϪ1
)
log(k2/sϪ1
)
log(k3Ј/sϪ1
)
log(k3/L molϪ1 sϪ1
)
log(k4Ј/L molϪ1 sϪ1
)
log(k4/L molϪ1 sϪ1
)
log(k5Ј/L molϪ1 sϪ1
)
log(k5/L molϪ1 sϪ1
)
pKCy
Ref.
0
0
0
(Ϫ0.045)bc
Ϫ3.745
3.70a
3.65
3.92
9
10
11
1.5
4
Ϫ1.845
Ϫ1.785
Ϫ1.545
9
12
12
9
9
12
7
13
(Ϫ0.761)bc
Ϫ0.042b
Ϫ4.301
Ϫ3.582
3.54a
3.54a
3.68
10
10
13.1
18
25
25
25.5
27
30
30
50
60
65
70
80
80.9
82
94
100
100
Ϫ1.390
Ϫ1.220
(0.434b
Ϫ3.106
Ϫ1.102
Ϫ5.000
3.54
3.29
3.47
(2.188)bc
Ϫ7.712
Ϫ7.535
2.991
Ϫ4.452
1.748
Ϫ0.907
9
9
8
3.73
(Ϫ0.364)c
1.319b
0.68
1.56
Ϫ2.478
Ϫ2.78
Ϫ1.89
Ϫ1.569
Ϫ4.039
Ϫ3.535
Ϫ2.925
2.214
2.590
3.170
3.806
3.46a
3.45a
3.70a
Our work
Our work
14
9
12
14
9
12
12
8
1.826
Ϫ5.975
Ϫ5.745
Ϫ5.548
Ϫ5.062
Ϫ5.046
Ϫ5.022
Ϫ4.545
Ϫ4.499
Ϫ4.298
3.342
3.838
Ϫ2.903
Ϫ2.035
1.997
2.417
2.367
Ϫ1.004
3.70a
9
a Extrapolated by author (not measured at this temperature). b Italic data (kiЈ) are recalculated from the respective ki. c Data in brackets were excluded from our model fit.

View Article Online
between the ki, the latter (usually calculated from measured kia)
being highly affected by the uncertainty concerning KCy.
Most measurements of k2–k5 were taken above pH 4, where the
cyanate anion, rather than cyanic acid, is the major species; and
where the kia : kiЈ ratio does not involve KCyЈ, unlike the kia : ki
ratio. Furthermore, when kiЈ values not provided by the liter-
ature were recalculated from the respective ki (using the author’s
values for dissociation constants: data in italics in Table 1), they
correlated very well with those directly mentioned in the
literature.
Dissociation constants
Temperature-dependent expressions of the dissociation con-
stants of the species involved were found in the literature15 and
are summarised in Table 3, as second-order polynomials
that usually fit experimental data very well (a third-order poly-
nomial was necessary for K1Car).
The dissociation constant of HNCO has never been accur-
ately measured above 40 ЊC, due to its intrinsic instability in
water. Even below this temperature, the data found in the
literature (Table 1) show a serious discrepancy: neither a
polynomial nor a homographic function of temperature fitting
these data yields realistic values when extrapolated above 40 ЊC.
While the expression listed in Table 3 is satisfactory below
30–35 ЊC, we assumed pKCy to be roughly 3.45 above 35–40 ЊC,
and found a compromise to minimise the consequences of this
approximation.
This determined our preference for using kiЈ-based kinetics in
our model, despite the closer physical realism of ki: inaccuracies
in KCy therefore affect no significant terms above pH 4 (CAA
synthesis is inefficient below pH 4, anyway). Extrapolated
temperature-dependent expressions of rate constants kiЈ in
the form log(k) = a Ϫ b/T were thus evaluated using least-
square fitting, and are shown in Table 4. Some isolated data
were however excluded when notoriously inaccurate, k2Ј values
found at low temperature by Lister9 and Jensen,12 or obviously
inconsistent with other data, k2Ј value found by Williams
Rate constants of cyanate degradation, choice of the model
The constants ki and kiЈ cannot be considered as equivalent: as
first pointed out by Kemp and Kohnstam,14 data for kiЈ values
are probably much more reliable than ki values.This can be seen
in Table 1 in the much better correlation between the kiЈ than
8
and Jencks,7 k1Ј value found by Vogels et al. (in brackets in
Table 1).
Rate constant of amino acid N-carbamoylation (k6Ј)
Table 2 Experimental cyanate decomposition rate at various pH and
temperatures; [NCOϪ]0 = 1.6 mM
The rate of N-carbamoylation of various amines and amino
acids by cyanate was first studied by Williams and Jencks7 at
25 ЊC in the scope of a mechanistic study, with no evaluation
of the activation energy. Since we needed to work at higher
temperatures we performed a new series of measurements on
several amino acids. To minimise the influence of cyanate degra-
dation, initial reaction rates only were evaluated at various
pHs. The results are summarised in Table 5 and in Fig. 1 for
30 ЊC
50 ЊC
pH
5
5.5
Ϫ4.58
0.92
6
6
6.5
Ϫ4.82
1.68
a
log(k0 /sϪ1
)
Ϫ4.45
0.55
0.68
Ϫ5.42
0.58
Ϫ4.57
1.43
1.56
log(k2Ј/L molϪ1 sϪ1
Mean log(k2Ј)
)
Table 3 Temperature (ЊC) dependent expression of dissociation constants pK = A Ϫ BT ϩ CT2 Ϫ DT3 calculated from literature data by least-
square fitting
pKw
pKAm
pK1Car
pK2Car
pKCy
pK1Gly
pK2Gly
A
14.9394
4.2044
1.6899
—
10.0820
3.6123
1.0536
—
6.5787
1.3345
1.9039
0.81746
10.3190
1.4713
0.99487
—
3.9208
2.8700
4.3014
—
2.4403
4.7818
0.48047
—
10.487
3.071
0.95328
—
102B
104C
106D
Table 4 Calculated temperature (ЊC) dependent expressions of rate constants log(kiЈ) = a Ϫ (b/T ), with respective correlation coefficient r2 and
activation energies EA/kcal molϪ1 = b × ln(10) × R/kcal molϪ1 KϪ1. Units for kiЈ are the same as in Table 1
log(k1/L molϪ1 sϪ1
)
log(k2Ј/L molϪ1 sϪ1
)
log(k3Ј/sϪ1
)
log(k4Ј/L molϪ1 sϪ1
)
log(k5Ј/L molϪ1 sϪ1
)
a
9.9542
3249.2
0.99613
14.9
11.7720
3301.5
0.94579
15.1
9.1574
5041.1
0.99734
23.1
5.8098
3082.7
0.93061
14.1
10.9340
4592.5
0.99859
21.0
b
r2
EA/kcal molϪ1
Table 5 Experimental conditions and results for amino acid N-carbamoylation rate measurement. Constants k6Ј and k6a refer to eqn. (13)
a
[AA]0/
[KOCN]0/
mmol LϪ1
106k6 /
log(k6Ј/
mean
log(k6Ј)
Amino acid
Gly
T /ЊC
pK2AA
pH
mmol LϪ1
molϪ1 L sϪ1
molϪ1 L sϪ1
)
30
9.65215
5
5
5
5
6
7
8
9
56
56
56
56
56
56
56
56
11.2
11.2
300
60
70
50
140
280
420
140
100
140
100
100
15.9
15.9
373
102
77
340
319
326
Ϫ3.476
Ϫ3.504
Ϫ3.495
Ϫ2.825
Ϫ2.757
Ϫ2.760
Ϫ2.741
Ϫ2.738
Ϫ2.925
Ϫ2.887
Ϫ2.887
Ϫ2.85
Ϫ3.49
Gly
50
9.19015
1420
1700
1680
1630
1080
1190
1300
1300
1420
1364
3530
Ϫ2.78
Ϫ2.90
Val
50
9.14215
6
6.5
6.5
8
8
8
β-Ala
Lys(TFA)
Thr
50
50
50
9.60515
8.9
Ϫ2.87
Ϫ2.45
8.5516
100
1250
J. Chem. Soc., Perkin Trans. 2, 2001, 1247–1254

View Article Online
a
those concerning glycine, measured apparent rates k6 being
converted into k6Ј according to eqn. (13). The first-order
kinetics were confirmed for glycine and valine by the linear
dependence of rate on initial cyanate concentration at constant
pH. Since N-carbamoylation of glycine was initially intended to
serve as a model for theoretical calculations, this case was more
thoroughly investigated. The rate constants obtained for glycine
at 30 and 50 ЊC are consistent with that found by Williams and
Jencks at 25 ЊC, and correspond to an activation energy of 17.2
kcal molϪ1 for k6Ј (7.1 kcal molϪ1 for k6, assuming pKCy to be
3.45 at 50 ЊC).
This study also aimed to demonstrate a dependence between
the amino acid N-carbamoylation rate and the amino group
dissociation constant, in order to allow more accurate com-
puterised kinetic calculations for various amino acids. In their
kinetic study,7 Williams and Jencks found a good correlation at
25 ЊC between amine basicity (pKA) and N-carbamoylation rate
k6, with an approximate law log(k6) = 0.15 ϩ 0.3 × pKA, which
may also be rewritten, according to eqn. (13), as log(k6Ј) =
3.44 – 0.7 × pKA; this correlation was interpreted as an effect of
increasing nucleophilicity. Practically, this correlation means
that the maximum apparent N-carbamoylation rate (according
a
to eqn. (13), k6 max = k6Ј) must be expected to decrease with
increasing amine basicity.
Unfortunately, although following a similar trend as at 30 ЊC,
the experimental rate constants k6Ј we measured for different
amino acids at 50 ЊC exhibited a very bad correlation to amino
group pKA; therefore no realistic linear correlation could be
assigned from our measurements. For theoretical calculations
we can therefore at least assume that, for common amino acids,
the rate constant k6Ј will vary by at most one order of mag-
nitude from the value found for glycine.
Fig. 2 Computer-simulated N-carbamoylation of glycine at 50 ЊC,
with starting conditions [Gly]0 = 0.5 M, [NCO]0 = 0.5 M (top), 0.65 M
(bottom); simulations performed at constant pH (curves) or with free
pH variations (isolated marks, plotted at final pH), stopped once 99%
of initial cyanate is consumed. Product yields (mol% relative to initial
glycine) are plotted as a function of pH: N-carbamoylglycine (A; ᭜),
urea (B; ᭛), unreacted cyanate (C; ᭝); reaction time (D; ϩ). Split lines
(B) and (D), and duplicate marks in the bottom graph correspond to
consumption of either 99% initial cyanate (upper) or 99.5% initial
glycine (lower) at the end of reaction.
N-Carbamoylation of amino acids
Competition between different reactions
The pH-to-apparent rate profiles of cyanate reactivity (amino
acid is exemplified by glycine) are shown in Fig. 1. The N-
carbamoylation apparent rate constant k6 is maximum and
cyanate was introduced. Examples of simulation results are
displayed in Fig. 2.
a
roughly constant within a pH range between pKCy and pK2AA
,
Validity limits of the model. Due to uncertainty concerning
pKCy, and therefore HNCO/NCOϪ concentrations below pH 5,
the results of our model are not very accurate in this pH range,
where cyanate hydrolysis is anyway too fast for amino acid
N-carbamoylation to be efficient. The practical consequences
are thus of minor importance. Above 60–70 ЊC, as well as for
long reaction times, reverse reactions of (1)–(6), not taken into
account in our model, are no longer negligible. Calculated
reaction times and CAA yields are thus probably below those
that could be physically observed; however we still consider our
simulation results to be qualitatively pertinent.
a
where k6a = k6Ј. The cyanate hydrolysis rate (k0 ) shows
continuous decay as pH increases, then remains constant above
pH 8. The highest selectivity of the former reaction against the
latter thus occurs around pH 8.5.
a
The urea formation apparent rate constant k5 and
a
carbonate-catalysis rate increment k4 have similar profiles to
a
a
k6 , with maximum plateaux between pKCy and pKAm for k5
a
( = k5Ј), and between pK1Car and pK2Car for k4 ( = k4Ј). These
maximum apparent rate constants are situated no more than
a
one order of magnitude lower than k6 [log(k5Ј) = Ϫ3.28,
log(k4Ј) = Ϫ3.73 at 50 ЊC]; no larger difference in apparent rates
can be expected with other amino acids and at any pH. There-
fore a good selectivity of reaction (6) versus the others requires
that the concentrations of ammonia and carbonate should
remain very low. Since both species are by-products of cyanate
hydrolysis [reactions (1)–(3)], this confirms the necessity of
limiting the latter reaction as much as possible, by operating
above pH 6 and as close as possible to pH 8.5.
Simulations results. These showed that no good yield of CAA
can be obtained at pH < 6 (except by using a considerable
excess of cyanate, which would be very inconvenient on a large
scale). At 0.5 M glycine concentration, and using only 1 equiv-
alent cyanate, 95% conversion (at most) is attained after ca. 30
hours at 50 ЊC and pH 9. Using excess cyanate (1.3 equivalent),
full glycine conversion can be obtained above pH 8 before
complete cyanate hydrolysis (in ca. 10 h). Starting with 0.2 M
glycine (this corresponds to the solubility limit of the least
polar amino acids at 50 ЊC), and using 1 equivalent cyanate,
95% conversion (at most) is reached at 50 ЊC and pH 9.4, but
after ca. 70 hours. Full conversion may then be reached above
pH 8.2 within ca. 10–20 h, when using 1.5 equivalent cyanate.
From these results it obviously appears that pH 5–6, as
proposed in the literature,6 is not the best reaction condition. If
the reaction is stopped once 99.5% of initial glycine has been
converted, the minimum urea formation occurs at pH 9–10,
which is higher than the value (pH 8.5) predicted by the appar-
Computer simulation
To refine these conclusions, we designed and performed a
computer simulation of the evolution of the system, based on
the (temperature-dependent) kinetic and dissociation constants
discussed above, using glycine as a model amino acid. A step-
wise integration algorithm with variable time increments
was implemented to integrate eqn. (7), pH and temperature
remaining constant throughout the simulation. The simulation
was run until consumption of either glycine or cyanate was
(almost) complete, the latter case occurring when molar excess
J. Chem. Soc., Perkin Trans. 2, 2001, 1247–1254
1251

View Article Online
a
a
ent rate ratio k6 /k0 . This, however, corresponds to the shortest
times required to attain 99.5% conversion. Complete excess
cyanate consumption leads to an increase in urea production,
the maximum of which (ca. 30–35 mol% of the initial excess
cyanate) occurs around pH 8.5 (Fig. 2). Whatever the pH,
CAA/urea selectivity decreases as the temperature increases.
However, reactions carried out at temperatures lower than
40–50 ЊC and/or pH higher than 9 require a considerable time
to reach completion (more than 100 h).
excess cyanate, as well as to precipitate the CAA, which is
recovered by filtration.
Concerning more polar amino acids (which yield very water-
soluble CAAs), by-product salts may in theory be removed
by the use of cation-exchange resins. However these must be
carefully handled since the CAA readily undergoes irreversible
cyclisation into the corresponding hydantoin under acidic
conditions (pH < 2); such hydantoin formation seems to be
enhanced by the presence of hydrochloric acid. However we
obtained a good result for C-Gly preparation by carrying out
the reaction at 50 ЊC in a 0.5 M glycine–0.65M cyanate
solution, without pH regulation, obtaining ca. 95% C-Gly and
less than 10% urea after 10 h reaction. These results are in
accordance with our computer simulations for these con-
ditions (Fig. 2). Most of the product was recovered pure,
by simple filtration of the crude reaction mixture through
a cation-exchange column. Developments concerning this
and applications to other polar amino acids will be dealt with in
a forthcoming paper.
Additional calculations were performed to examine the
reaction without pH regulation. Considering acid–base
equilibria to be much faster than reactions of cyanate, the
simulation algorithm was modified so that the pH was
corrected after each integration time step, using the first-order
differential of the electrical neutrality equation (Σ[anions] =
Σ[cations]) against [Hϩ] concentration. The initial pH of
the reaction mixture was calculated using the same equation
with initial species concentrations. In accordance with
experimental data, calculations show a continuous increase in
pH, very fast at the beginning of the reaction, then slower
towards an asymptotic value around 9–10, this value mostly
depending on the initial stoichiometric excess of cyanate/
glycine.
Results in terms of CAA yield, urea formation and reaction
time are similar to those obtained using the constant-pH
algorithm with the pH fixed at the asymptotic value (isolated
marks in Fig. 2). We only observed a slight increase in reaction
time and urea formation, the latter mostly occurring at the
beginning of the reaction, when pH is still below 8. However,
these results must be considered as merely qualitative since this
algorithm provides less accurate results than the constant-pH
algorithm. For instance, the calculated asymptotic pH (after
10–15 h reaction at 50 ЊC) from an initial solution of 0.5 M
glycine and 0.65 M cyanate is 9.53, while the experimental value
under the same conditions is pH 9.98.
Conclusion
In this paper we have reported efficient conditions for prep-
arative amino acid N-carbamoylation by aqueous cyanate, and
have highlighted the important influence of medium to weakly
basic pH. While our computer simulations indicate pH 8.5–9 as
an optimum, they also exhibit the robustness of the reaction,
which may be achieved over a wider pH range (7–10) with
moderate excess of cyanate. In many cases, however, pH 7–8
was preferred as the most convenient on a preparative scale.
Investigations are in progress for improving polar, high-
functional amino acid N-carbamoylation.
Amino acid N-carbamoylation may have had important
consequences in the molecular origins of life, as part of the
emergence of prebiotic peptides.17 These preliminary results are
also necessary for a further study of this reaction within this
scope, especially in a CO2 atmosphere, the prebiotic relevance
of which is uncontested. This subject will be discussed in more
detail in a future paper.
Application to preparative synthesis
According to our computer simulations, optimum CAA yield
and CAA/urea selectivity (a critical requirement for purifi-
cation when the CAA is very soluble in water) should be
obtained by reacting 1.2–1.5 equivalent cyanate, at moderate
temperatures and with pH regulated to 8.5–9, stopping the
reaction once ca. 99% amino acid is converted. The remaining
excess cyanate can be conveniently hydrolysed without pro-
ducing more urea by rapidly acidifying the medium to pH
2–3. Practically speaking however, these optimum conditions
are not very convenient for preparative purposes, since low
temperatures and basic pH induce long reaction times (over 100
h). Buffering the reaction at medium to basic pH results
in a significant increase in salt content, the separation of
which from water-soluble CAA is as difficult as from urea.
Compromises had thus to be found.
Concerning low-polarity CAAs, weakly soluble in acidified
water (pH < 3), neither urea nor salts are a major problem,
provided they are only present in moderate amounts. For prac-
tical convenience on a laboratory scale, more cyanate was used
(1.5–2 equiv.) than theoretically necessary, and a pH around 7.5
and temperatures of 50 ЊC were applied, so that the reaction
time remained within 10–24 h; when racemization on α-carbon
is not a problem, higher reaction temperatures can be used.
Under these conditions, the N-carbamoyl derivatives C-Ala,
C-Val, C-Leu, C-Phe, C-Met, C-Lys(TFA) were prepared and
isolated with good yields through a straightforward procedure.
In a typical experiment, the warm solution of amino acid
was mixed with cyanate, then adjusted if necessary to the
desired pH, which is maintained constant by further controlled
addition of acid (e.g. HCl or HNO3). The reaction can easily
be followed by HPLC. After completion the medium is chilled,
rapidly acidified to pH 2–3 in order to hydrolyse the unreacted
Experimental
Materials and methods
Melting points were measured using a Buchi 520 apparatus.
1H-NMR spectra were recorded on a Varian EM360 (60 MHz)
or a Bruker AC250 spectrometer (250 MHz) in DMSO-d6 solu-
tion. pH regulations were performed using a Methrom Titrino
719S autotitrator. HPLC analyses were carried out using a
Varian apparatus set up with a Varian 2510 pump, a Varian
2550 UV detector and a Varian 4290 integrator. Nucleosil
RP-C18, 5 µm, 250 × 4.1 mm id, analytic columns (obtained
from Shandon, France) were used at room temperature under
isocratic conditions (detailed conditions are provided below).
Chromatographic silica gel (60 mesh) was obtained from
Merck.
Pure water (50 MΩ) was obtained using a milli-Q (Millipore)
system after column-exchange deionization. Methanol, aceto-
nitrile and hexane were obtained from Baekrout. Glycine, -
alanine, -valine, -leucine, -phenylalanine, fluorenylmethyl
chloroformate (FMOC) and trifluoroacetic acid were obtained
from Aldrich. β-Alanine was obtained from Fluka. Methionine
was obtained from Aventis Animal Nutrition. ε-Trifluoroacetyl-
-lysine and -threonine were obtained from Degussa. Sodium
hydroxide and silver nitrate were obtained from Prolabo.
Hydrochloric acid (36%), nitric acid (70%), acetic acid and
triethylamine, were obtained from Carlo Erba. All reagents
and solvents were used as received except potassium cyanate,
which was washed with methanol then dried in vacuo prior to
use.
1252
J. Chem. Soc., Perkin Trans. 2, 2001, 1247–1254

View Article Online
Kinetic measurements
taken, diluted if necessary in pure water, and analysed by
HPLC, eluent water ϩ 0.05% TFA; tR/min = 4.5 (C-Thr), 5.0
Isocyanic acid degradation. A solution of 78 mg (0.96 mmol)
potassium cyanate in 60 mL water was stirred in a thermostated
reactor; the pH was regulated (5 to 6.5) using 0.2 M HNO3 by
means of the autotitrator. The reaction was followed by reverse
silver titration: at regular time intervals, 5 mL aliquots were
taken, cooled and added to 1 mL 0.1 M AgNO3. The AgOCN
precipitate was filtered off, washed with water, and the remain-
ing Agϩ in the filtrate was titrated with 0.01 M HCl, using silver
electrode potentiometry.
(C-ꢂ-Ala), 13.5 (α,ω-dicarbamoyllysine) at 1.0 mL minϪ1
.
Preparation of CAAs
N-Carbamoyl-L-valine C-Val. To a warm (50 ЊC) stirred solu-
tion of 175.5 g (1.5 mol) -valine in 1 L water was added 184.5 g
(2.25 mol) potassium cyanate. The pH was then regulated to 7.5
using the autotitrator to control continuous 6 M HCl addition.
Reaction completion was checked by HPLC, eluent water–
acetonitrile (95 : 5) ϩ 0.05% TFA; tR/min = 4.15 at flow rate 1
mL minϪ1. After completion (about 15 h), the reaction mixture
was cooled by means of an ice bath and acidified by adding 6 M
HCl until reaching pH 2. After filtration, the crude product was
recrystallised from boiling water and dried in a heating
desiccator, yield 84%. Mp 207–209 ЊC. δH 0.78 (3H, d, J = 7 Hz,
CH3), 0.82 (3H, d, J = 7 Hz, CЈH3), 1.93 (1H, m, Hβ), 3.95 (1H,
dd, Hα, Jαβ = 5 Hz, JαNH = 9 Hz), 5.55 (2H, s, NЈH2), 6.11 (NH,
d, J = 9 Hz), 12.45 (1H, s, COOH). δC 19.89 (Cγ), 21.05 (CγЈ),
29.90 (Cβ), 63.83 (Cα), 162.94 (NCONЈ), 179.69 (COOH).
Glycine N-carbamoylation. 20 ml of a 0.1–0.42 mM potas-
sium cyanate solution in water was stirred in a thermostated
reactor and the pH was adjusted with 4 M nitric acid. A solu-
tion of 126 mg (1.68 mmol) glycine in 10 mL water and
adjusted to the same pH was immediately added. The pH was
regulated throughout the reaction using 4 M nitric acid. At
regular time intervals, aliquots were taken and submitted to
FMOC derivatisation prior to HPLC analysis.
FMOC-derivatisation. (Adapted from Clapp et al.18): a 0.5
mL aliquot was taken, cooled and added to 0.5 mL of 0.2 M
borate buffer (pH 7.7) containing 1 mM -alanine (internal
reference). To half of this mixture 0.5 mL of a fluorenylmethyl
chloroformate solution (15 mM) in acetone was added and the
mixture was stirred for 30 s, then the solution was washed 3
times with 2 mL hexane. 0.1 mL of the aqueous layer was then
analysed by HPLC (eluent MeOH–MeCN–acetate buffer:
20 : 30 : 50. The acetate buffer was made from 3 mL acetic acid,
1 mL triethylamine in 0.9 L water, adjusted to pH 4.2 with
sodium hydroxide then made up to 1 L with water); tR/min = 8.5
N-Carbamoyl-L-alanine C-Ala. Same procedure as C-Val,
using 135 g (1.5 mol) -alanine in 1 L water and 184.5 g (2.25
mol) potassium cyanate, reaction at 50 ЊC, pH 7.5 for 15 h.
Eluent for HPLC analysis: water–acetonitrile (95 : 5) ϩ 0.05 %
TFA; tR/min = 8.4 at 0.8 mL minϪ1. The crude product was
recrystallised from boiling water to yield 240 g (82%) pure
material. Mp 165–167 ЊC. δH 1.22 (3H, d, Hβ, Jαβ = 7.24 Hz),
4.04 (1H, dq, Hα, JHα-NH = Jαβ = 7.35 Hz), 5.60 (2H, s, NЈH2),
6.23 (1H, d, NH, JNH-Hα = 7.7 Hz), 12.47 (1H, s, COOH).
δC 19.05 (Cβ), 48.79 (Cα), 159.06 (NCONЈ), 176.28 (COOH).
(FMOC-Gly), 11.1 (FMOC-Ala) at 0.9 mL minϪ1
.
3-Ureidopropionic acid C-ꢂ-Ala. Same procedure as C-Val,
using 13.35 g (150 mmol) β-alanine in 0.5 L water and 13.36 g
(165 mmol) potassium cyanate. pH regulated to 7.5, reaction
time 18 h. Eluent for HPLC analysis: water–acetonitrile
(80 : 20) ϩ 0.05% TFA; tR/min = 3.6 at 1.0 mL minϪ1. The
crude product was recrystallised from boiling water to yield
11.9 g (60%) pure material. δH 2.37 (2H, t, Hα, J = 6.0 Hz), 3.18
(2H, dt, Hβ, Jd = Jt = 5.9 Hz), 5.40 (2H, s, NЈH2), 6.00 (1H, t,
NH, Jt = 5.4 Hz), 12.20 (1H, s, COOH). δC 35.0 (Cα), 35.7 (Cβ),
159.0 (NCONЈ), 174.0 (COOH).
L-Valine N-carbamoylation. The same protocol was used as
for glycine, except that FMOC derivatisation was not used: 60
mL of a 15.9–37.3 mM potassium cyanate solution in water
was stirred in a thermostated reactor. To this a solution of 126
mg (1.08 mmol) valine in 10 mL water and adjusted to the same
pH was immediately added. At regular time intervals, aliquots
were taken and analysed by HPLC, after dilution to 1 : 10 in
HPLC eluent: water–acetonitrile (90 : 10) ϩ 0.05% TFA; tR/
min = 4.8 (Val), 10.1 (C-Val) at 0.9 mL minϪ1
.
Nꢀ-Carbamoylation of Nꢁ-trifluoroacetyl-L-lysine. Into
a
N-Carbamoyl-L-leucine C-Leu. Same procedure as C-Val,
using 30 g (0.23 mol) -leucine in 160 mL water and 30 g (0.37
mol) potassium cyanate, at 50 ЊC, pH 7.5 for 18 h. The crude
product was recrystallised from boiling water to afford 30.45 g
(76%) pure material. Mp 216–218 ЊC. δH 0.89 (6H, dd, Hδ,
J1 = J2 = 6.0), 1.45 (2H, m, Hβ), 1.65 (1H, m, Hγ), 4.08 (1Hd, td,
Hα, JHα-NH = 8.4 Hz, Jαβ = 6.3 Hz), 5.55 (2H, s, NЈH2), 6.18 (1H,
d, NH, J = 8.4 Hz), 12.44 (1H, s, COOH). δC 22.42 (CЈδ), 23.71
(Cδ), 25.16 (Cγ), 41.89 (Cβ), 51.56 (Cα), 159.19 (NCONЈ), 176.12
(COOH).
stirred solution of 2.55 g (10.5 mmol) Nε-trifluoroacetyl--
lysine in 150 mL water, warmed to 50 ЊC and adjusted to pH 8
with solid NaOH, was dissolved 946 mg (11.5 mmol) potassium
cyanate at time 0. The pH was regulated to 8 throughout the
reaction with 6 M HCl using the autotitrator. At regular time
intervals, 0.1 mL aliquots were taken, diluted in pure water and
analysed by HPLC, eluent: water–acetonitrile (90 : 10) ϩ 0.05%
TFA; tR/min = 5.4 (Lys(TFA), used as internal reference), 9.7
(C-Lys(TFA)) at 1.0 mL minϪ1
.
pKA of Nꢁ-trifluoroacetyl-L-lysine. 25 mL of a 0.02 M solution
of Nε-trifluoroacetyl--lysine in pure water and warmed to
50 ЊC was titrated with 0.1 M sodium hydroxide, the pH
being measured by the autotitrator. The pKA (8.9 at 50 ЊC)
was directly read from the titration curve (pH vs. NaOH mol.
added). This value is consistent with that found in the liter-
ature19 at 25 ЊC.
N-Carbamoyl-L-phenylalanine C-Phe. Same procedure as for
C-Val, using 11 g (67 mmol) phenylalanine and 20 g (246 mmol)
potassium cyanate in 500 mL water; at 50 ЊC, pH 7 for 15 h.
10.8 g (78%) pure material was obtained after recrystallisation
from water. Mp 200 ЊC. δH 2.8 (2H, dd, Hβ), 4.35 (1H, m, Hα),
5.10 (2H, s, NЈH2), 6.25 (1H, d, NH), 7.25 (5H, m, HAr), 11.80
(1H, s, COOH).
ꢂ-Alanine and L-threonine N-carbamoylation. Into a stirred
solution of amino acid and α,ω-dicarbamoyllysine (8 mM,
internal reference) in 150 mL water, warmed to 50 ЊC and
adjusted to pH 8 with solid sodium hydroxide (for Thr) or 6 M
hydrochloric acid (for β-Ala), was dissolved potassium cyanate
at time 0. The pH was regulated to 8 with 4 M HCl using the
autotitrator. At regular time intervals, 0.1 mL aliquots were
Nꢀ-Carbamoyl-Nꢁ-trifluoroacetyl-L-lysine C-Lys(TFA). Same
procedure as for C-Val, using 20 g (82.6 mmol) Nε-trifluoro-
acetyl--lysine and 10 g (123.5 mmol) potassium cyanate in 150
mL water at 50 ЊC; the pH was regulated to 7.5 during the
reaction. After stirring at 50 ЊC for 1 h 30 min, another 3.35 g
(41.4 mmol) potassium cyanate was added to the mixture,
which was again stirred at regulated pH for 1 h 30 min. The
J. Chem. Soc., Perkin Trans. 2, 2001, 1247–1254
1253

View Article Online
chilled mixture was then acidified to pH 2 using concentrated
HCl, and the precipitated product was recovered by filtration
and dried in vacuo to yield 17.37 g (60.9 mmol, 73.8%)
C-Lys(TFA) (11.78 g (50%) after recrystallisation from water).
Mp 169–170 ЊC. δH 1.30 (2H, m, Hγ), 1.58 (4H, m, Hβ ϩ Hδ),
3.17 (2H, dt, Hε, Jd = Jt = 6.5 Hz), 4.04 (1H, m, Hα), 5.59 (2H, s,
NЈH2), 6.22 (1H, d, NαH, Jd = 8.2 Hz), 9.43 (1H, t, NεH, Jt = 6.5
Hz), 12.51 (1H, s, COOH). δC : 23.3 (s, Cγ), 28.7 (s, Cβ), 32.5 (s,
Cδ), 40.0 (s, Cα), 52.9 (s, Cα), 116.8 (q, CF3, Jq = 288.3 Hz), 157.0
(q, COCF3, Jq = 35.8 Hz), 159.2 (s, CON2), 175.5 (s, COOH).
m/z (FABϩ/GT) : 571 (2M + Hϩ), 324 (M + Kϩ), 286 (M + Hϩ).
νmax(cmϪ1) (film) 1160, 1185, 1213 (C–F), 1555 (N–H), 1642
identical physical characteristics to an authentic sample
supplied by Aventis Animal Nutrition.
Acknowledgements
We are grateful to the CNRS, the French Ministries of
Education and Research, and Degussa-Hüls AG (Germany)
for their financial support, and to Adam Lindsey-Clark for
his help in revising the English expression of this article.
References
1 H. Collet, C. Bied, J. Taillades, L. Mion and A. Commeyras,
Tetrahedron Lett., 1996, 37, 9043.
2 H. Collet, L. Boiteau, J. Taillades and A. Commeyras, Tetrahedron
Lett., 1999, 40, 3355.
(C᎐O ), 1699 (C᎐Oacid), 3273, 3329, 3444 (N–H). [α]D = ϩ10Њ
᎐
᎐
urea
(EtOH at 0.5 g/100 mL, 20 ЊC, wavelength = 589 nm).
N-Carbamoylglycine C-Gly. To a warm (50 ЊC), stirred solu-
tion of 3.75 g (50 mmol) glycine in 50 mL water was added 5.27
g (65 mmol) potassium cyanate. The reaction was followed by
HPLC analysis (eluent: aqueous KH2PO4 (10 mM) ϩ sodium
hexanesulfonate (5 mM), adjusted to pH 2.5 with H3PO4, UV
detection at 195 nm). After 10 h stirring at 50 ЊC (the pH
meanwhile rose from 7.12 at the beginning to 10.0 at the end of
the reaction), the mixture was chilled, and filtered through a
column containing AG-50-X2 cation-exchange resin (Hϩ form;
further elution with water). After water evaporation in vacuo,
then drying in a heating desiccator, the major fraction afforded
4.88 g (41.3 mmol, 83%) pure C-Gly (hydantoic acid: mp
165 ЊC. δH 3.66 (2H, d, J = 5.6 Hz, Hα), 5.64 (2H, s, NH2), 6.14
(1H, t, J = 5.6 Hz, NH), identical physical data to an authentic
sample from Sigma). A further fraction (824 mg) contained ca.
90 mol% C-Gly (6.6 mmol) and 10 mol% urea (NMR). Overall
yield C-Gly 95%.
3 V. F. Gonko, N. D. Shustova, G. M. Anoshina, T. E. Zubova and
L. B. Radina, Pharm. Chem. J., 1978, 12, 601.
4 A. Rousset, M. Lasperas, J. Taillades and A. Commeyras,
Tetrahedron, 1980, 36, 2649.
5 A. Yamashiro, K. Kobuta and K. Yokozeki, Agric. Biol. Chem.,
1988, 52, 2857; O. Keil, M. P. Schneider and J. P. Rasor, Tetrahedron:
Asymmetry, 1995, 6, 1257.
6 D. G. Smyth, J. Biol. Chem., 1967, 242, 1579.
7 A. Williams and W. P. Jencks, J. Chem. Soc., Perkin Trans. 2, 1974,
1753.
8 G. D. Vogels, L. Uffink and C. Van der Drift, Recl. Trav. Chim.
Pays-Bas, 1970, 89, 500.
9 M. W. Lister, Can. J. Chem., 1955, 33, 426.
10 K. Taufel, C. Wagner and H. Dunwald, Z. Elektrochem., 1928, 34,
115.
11 R. Naumann, Z. Elektrochem., 1910, 16, 773.
12 M. B. Jensen, Acta Chem. Scand., 1958, 12, 1657.
13 R. Caramazza, Gazz. Chim. Ital., 1958, 88, 308.
14 I. A. Kemp and G. Kohnstam, J. Chem. Soc., 1956, 900.
15 G. Kortüm, W. Vogel and K. Andrussow, Dissociation Constants of
Organic acids in Aqueous Solution, Butterworths, London, 1961.
16 P. K. Smith, A. T. Gorham and E. R. B. Smith, J. Biol. Chem, 1942,
144, 737.
17 J. Taillades, I. Beuzelin, L. Garrel, V. Tabacik, C. Bied and
A. Commeyras, Orig. Life Evol. Biosphere, 1998, 28, 61; J. Taillades,
H. Collet, L. Garrel, I. Beuzelin, L. Boiteau, H. Choukroun and
A. Commeyras, J. Mol. Evol., 1999, 48, 638.
N-Carbamoyl-DL-methionine C-Met. 447 g (3.0 mol) -
methionine and 323 g (4.0 mol) potassium cyanate were
suspended in 3.5 L water, then stirred and heated to reflux
for 30 min (dissolution was completed during heating). The
solution was then cooled to 5–10 ЊC and 400 mL 12 M HCl was
slowly added, while keeping the temperature at 10 ЊC. The
mixture was then stirred for another 1 h, then filtered. The solid
was washed 4 times with 100 mL water, then dried in vacuo, to
yield 447 g (2.44 mol, 81%) pure material, which presented
18 C. H. Clapp, J. S. Swan and J. L. Poetchmann, J. Chem. Educ., 1992,
69, 122.
19 E. E. Schallenberg and M. Calvin, J. Am. Chem. Soc., 1955, 77,
2779.
1254
J. Chem. Soc., Perkin Trans. 2, 2001, 1247–1254