Microscopic acid-base equilibria of a synthetic hydroxamate siderophore analog, piperazine-1,4-bis(N-methylacetohydroxamic acid)

Article abstract of DOI:10.1039/a702138k

The protonation behavior of the cyclic diaminodihydroxamate ligand, piperazine-1,4-bis(N-methyl-acetohydroxamic acid) (H₂L¹), has been studied at both the macroscopic and the microscopic level. Potentiometric and ¹H NMR te

Full text of DOI:10.1039/a702138k

View Article Online / Journal Homepage / Table of Contents for this issue
Microscopic acid–base equilibria of a synthetic hydroxamate
siderophore analog, piperazine-1,4-bis(N-methylacetohydroxamic
acid)
M. Amélia Santos,a M. Alexandra Esteves,a M. Cândida Vaz,a
J. J. R. Fraústo da Silva,a Bela Noszalb and Etelka Farkasc
a
Centro de Química Estrutural, Complexo 1, Instituto Superior Técnico, 1096 Lisboa-codex,
Portugal
b Semmelweis Medical University, Budapest, Hungary
c L. Kossuth University, Debrecen, Hungary
The protonation behavior of the cyclic diaminodihydroxamate ligand, piperazine-1,4-bis(N-methyl-
acetohydroxamic acid) (H 2L1), has been studied at both the macroscopic and the microscopic level.
1
Potentiometric and H N M R techniques have been used for the study of this ligand as well as several
model compounds: N-methylchloroacetohydroxamic acid, glycinehydroxamic acid and piperidino(N-
methylacetohydroxamic acid). M olecular modeling calculations have also been performed to predict the
most stable conformations and to estimate relevant contributions to the overall protonation process.
The results of the protonation microconstants show that the N-donors in H 2L1 are much less basic than
the O-donors. The protonated amine moieties release most of their protons in the acid region while the
deprotonation of the hydroxamate moieties starts only above pH 5. The theoretical modeling calculations
show the effect of electrostatic interactions and internal hydrogen bonds on the interactivity of the basic
sites throughout the protonation process.
group. Thus, these studies afford a qualitative picture of the
Introduction
protonation sequence. Furthermore, they enable the calcu-
lation of parameters characterizing the acid–base properties,
which can be used for the construction of the complete pro-
tonation scheme of this molecule and determination of the
protonation microconstants, as well as the corresponding
microspeciation.
Thus, to obtain further insight into the protonation sequence
for H2L1, 1H NMR titrations were carried out for this ligand as
well as for several model compounds: N-methylchloroaceto-
hydroxamic acid (HL2), glycinehydroxamic acid (HL3) and
piperidino(N-methylacetohydroxamic acid) (HL4). The model
compounds are simpler derivatives of the parent ligand and
proved to be useful in establishing the individual basicity and
NMR–pH profile of each proton binding site.8
Siderophores are a class of naturally occurring iron-chelating
compounds which are produced by microorganisms to trans-
port iron from the environment into cells. The development of
siderophore analogs has been the object of much interest,
given their potential use as pharmaceutical drugs, in particular
for chelation therapy.1,2
We have been involved recently in the study of polyazapoly-
hydroxamate ligands as analogs of siderophores.3,4 The ligand
piperazine-1,4-bis(N-methylacetohydroxamic acid) (H2L1)5,6 is
one example of this new family of biomimetic binders which
has been proven to be a reasonable model of rhodotorulic acid,
a naturally occurring dihydroxamate siderophore produced by
rhodotorula pilimane.7 This siderophore analog has four pro-
tonation sites (two hydroxamate and two amino groups) and
thus, depending on pH, H2L1 can exist in solution as a variety
of protonated species. The corresponding protonation macro-
constants were determined by potentiometry in a previous
work.5 However, macroconstants only give quantitative infor-
mation about the basicity of proton-binding sites when proton-
ation occurs independently. The fact that this ligand has some
protonation sites of similar basicity indicates that overlapping
of protonation processes may occur and so, for a certain pH, it
is possible to have protonation isomers presenting quite differ-
ent properties in terms of metal ion coordination and corres-
ponding biological processes.
Molecular modeling calculations were also performed on the
neutral and protonated forms of H2L1, HL4 and some related
species, to help the understanding of the protonation behavior
at both the macroscopic and the microscopic level.
Results
The stepwise protonation constants for H2L1 and related model
ligands, calculated from data of potentiometric titrations
(T = 25.0 ± 0.1 ЊC; I = 0.1  KNO3) or obtained from the liter-
ature, are summarized in Table 1. H2L1 has a cyclic diamine
Therefore, since the protonation macroconstants calculated
for H2L1 do not characterize sufficiently the acid–base behavior
of the individual proton binding sites of this ligand,8,9 its pro-
tonation process has been studied further. The present paper
reports the results obtained.
It is well known that the protonation of a basic site leads to
electronic deshielding effects on the adjacent methylene (or
methyl) protons, the magnitude of such effects being dependent
on the type of basic center.10 The average chemical shift of the
nearby carbon-bonded protons, as a function of pH, is
expected to reflect the fractional protonation of each basic
HO
O
N
C
H3C
N
N
CH3
OH
C
N
O
H2L1
moiety (piperazine) connecting two hydroxamic acid arms.
Data for the model compounds, reported in Table 1, show that
the basicity of piperazine (L5 ) nitrogens (log K1 = 9.84; log
J. Chem. Soc., Perkin Trans. 2, 1997
1977

View Article Online
Table 1 Stepwise protonation macroconstants for H2L1, HL4, L5, HL2
and HL3
Compound
log K1
log K2
log K3
log K4
H2L1 a
HL4 b
L5 c
9.53(2)
9.79(1)
9.84(2)
7.98(1)
9.12(6)
8.45(2)
7.44(1)
5.65(1)
6.67(2)
2.44(6)
HL2 b
HL3 d
7.37(3)
a
From ref. 5. b This work. c From ref. 11. d From ref. 12.
Fig. 2 1H NMR spectrum and titration curves (δ as a function of
pD*), obtained for the ligands (A) HL2 and (B) HL3 (CL = 2 × 10Ϫ2 mol
dmϪ3
)
Fig. 1 1H NMR spectrum of H2L1 at pD* 7.0 and titration curves,
δ as a function of pD* (CL = 2 × 10Ϫ2 mol dmϪ3
)
K2 = 5.65)11 is not too different from that expected for the
hydroxamate groups (log K = 8–9).12 So, a series of interactions
may occur when H2L1 is protonated without the calculated
macroconstants reflecting what happens at the level of each
individual protonation site.
To obtain further information on the protonation sequence
of H2L1, we have performed a 1H NMR titration of this ligand;
the set of curves obtained (Fig. 1) confirms that the protonation
of each basic group produces an electronic deshielding effect
on the nearby methylene (or methyl) protons with concomitant
downfield shifts of the corresponding resonance peaks. The
quantitative evaluation of this effect can be achieved using the
so-called Sudmeier and Reilley13 approach in which it is
assumed that, associated to each methylene proton (i) adjacent
to a basic center (j) being protonated, a mean protonation shift
constant or shielding constant (Cij) can be calculated; on the
other hand, the contributions from the protonation of several
different basic sites, in the vicinity of the methylene group, are
additive and so ∆δ = Σni= 1 Cij fj, where fj is the percentage of pro-
tonation. These Cij values depend on several factors such as the
type of basic center and the distance between the methylene
group and the basic center considered. For a series of linear
polyaminocarboxylates, Sudmeier and Reilley measured the
shielding constants for methylene groups in the following
Fig. 3 1H NMR spectrum of HL4 at pD* 8.2 and corresponding
titration curves, δ as a function of pD* (CL = 2 × 10Ϫ2 mol dmϪ3
)
models had to be chosen which give slightly different values for
the CN shielding constants.
Although Sudmeiers shielding constants could be used for
qualitative interpretation of the protonation sequence, suitable
models were also used for quantitative determination of con-
moieties: CH2COOϪ (CO = 0.20 ppm), CH2NR2 (CN = 0.75
α
1
ppm) and CH2CH2NR2 (CN = 0.35 ppm). However, more
stants in the present work. Therefore, H NMR titrations were
β
recent studies8,14 demonstrated that these values did not
show self-consistency in the interpretation of the protonation
of cyclic polyaminopolycarboxylates and appropriate cyclic
also performed for the following model compounds: HL2, HL3
(Fig. 2) and HL4 (Fig. 3). Table 2 summarizes all the chemical
shifts calculated for the ligands.
1978
J. Chem. Soc., Perkin Trans. 2, 1997

View Article Online
Table 2 Chemical shifts (∆δ) due to stepwise protonation of the amine (∆N ) and hydroxamate (∆H ) n basic sites, roughly evaluated for several atom
types from the 1H NMR titration
∆δ
H
a
N
a
N
b
N
cα
N
cβ
N
cγ
Molecule
Proton type
n
∆
∆
∆
∆
∆
∆
HL2
HL3
a, b
1
1
2
1
2
1
2
3
4
0.12
—
0.08
—
—
—
—
—
—
—



a
0.64
0.77
—
0.22
0.66
—
—
0.09
0.70
—
0.28
—
0.19
—
HL4
a, b, c
a, b, c



H2L1
0.10
—
0.73
0.88
—
—
—
—
—
0.68
Titration curves of HL2 [Fig. 2A] show the effect of
hydroxamate protonation on methylene (a) and N-methyl type
(b) protons. As expected, there is only one inflexion, which is
equivalent to that obtained by potentiometric titration.
and Reilley,13 for amine protonation of linear polyaminocarb-
oxylates.
Therefore, based on results of model compounds, a general
picture can be constructed for the protonation of H2L1: while
the last protonation stage, corresponding to the inflexion of the
c proton curve at ca. pD* 2, can be clearly attributed to proton-
ation of an amine group (log K4 = 2.44), the protonation of the
other three sites are not independent processes. A detailed
analysis of the titration curves corresponding to b and a pro-
tons shows the existence of small inflexions around pD* 8,
suggesting that, despite the interactivity in these protonation
processes, the first stage of the protonation mostly involves a
hydroxamate site.
Interactions of the protonation processes are also indicated
by the small inflexion observed in the b proton curve at ca. pD*
2. This may be due to some disruption of hydrogen bonds
involving the nitrogen atoms of amino groups and the nearby
hydroxamate hydroxylic protons in a six-membered ring inter-
mediate (see below for the diprotonated species).
1
The H NMR titration curve of HL3 [Fig. 2B] exhibits one
main inflexion (∆δ = 0.64) in the entire pD* range encompass-
ing both the amine and hydroxamate protonations, thus sug-
gesting that these processes are overlapping. Comparison of the
inflexion magnitudes of HL2 and HL3 (Fig. 2A and B), clearly
indicates that the deshielding effect produced in this a methyl-
ene proton by the N-amine protonation (∆aN ) is much higher
than that of the hydroxamate protonation (∆aH ). A more detailed
analysis of the HL3 a proton curve also reveals the existence of
an apparently minor discontinuity at the first stage of the pro-
tonation process that can be taken as indicative of a slightly
higher acidity of the NH3ϩ group relative to the NHOH group.
Furthermore, protonation microconstants calculated for α-
alaninehydroxamic acid from 13C NMR experimental data 15
support that acidity order.
Unfortunately, the use of HL3 as a model to estimate the
amine-protonation shift on the H2L1 methylene a protons is
limited due to the different substituents on the hydroxamate
nitrogen in these compounds. Therefore, another model (HL4,
Fig. 3) has been selected, to improve the evaluation of the
shielding effects on each non-labile proton.
O
O
H
N
N
N
N
H
O
O
1
The H NMR spectrum of HL4 and the corresponding pH
(pD*) dependent chemical shifts are shown in Fig. 3. The
curves corresponding to the a, c, d and e protons exhibit well-
defined inflexions centered at ca. pD* 10, while those for both
the a and b protons present a small inflexion around pD* 8.
Thus, this set of titration curves clearly shows that for this mol-
ecule the protonation of the hydroxamate group mostly occurs
after the protonation of the amine group. It also gives an indi-
cation of the average chemical shifts produced by amine and
hydroxamate protonation on each type of nearby non-labile
protons. In effect, Table 2 shows that the amine protonation
produces a deshielding effect on the a methylene protons
(∆δ = 0.66) which is about three to five times the effect of the
hydroxamate protonation. Thus, the chemical shift induced in
this methylene group provides useful information to help eluci-
date the sequence of protonation. Although the chemical shift
produced in the N-methyl protons by the hydroxamate proton-
ation is smaller than that involved in the methylene protons, it
can still be used as a good probe for the protonation of this
group. Table 2 shows the average chemical shift produced by the
amine protonation on the cyclic methylene c protons which are
α-, β- and γ-positioned relative to the amine group. The cα
methylene protons present an average chemical shift (∆δ = 0.70)
close to those of the a protons. As the number of bonds
between the basic amino site and non-labile proton increases,
the shift decreases dramatically (∆δβ/∆δα = 40%; ∆δγ/∆δα ≈
27%). The chemical shifts calculated for the cα and cβ methylene
protons (∆δNc = 0.70 and ∆δNc = 0.28) are comparable to the
Such an effect could explain the fact that the ammonium
protons of this ligand are more acidic than those in the corres-
ponding cyclic amine (see Table 1), although the inductive
effect of the hydroxamate groups should also be considered.
The very easy dissociation of the first proton from the fully
protonated ligand species can also be explained in terms of
coulombic repulsive effects between the positive sites of the
N,N-diprotonated species.
Providing an approximate qualitative characterization of the
protonation sequence of H2L1 is available, its quantitative
evaluation is also possible using the Studmeier and Reilley
approach, as has been carried out previously for the amino-
carboxylic analog.10 However, that method is limited, and is
very dependent on the pH and the adequacy of the shielding
constants used.14
For comparison, we have simulated the structure of some
neutral and protonated species and tried to relate their relative
stability and charge distribution to the microprotonation pro-
cess. We began by modeling the neutral species of the amino-
hydroxamic acids HL4 and H2L1, with the hydroxamate moi-
eties protonated. Both the global minima have the cyclic ring
in a chair conformation and the hydroxamic acid ‘arms’ in
equatorial positions. Fig. 4 shows the optimized geometries
for all the models studied. The effect of different hydrogen
bonding schemes on the heat of formation was first analysed
for the monohydroxamate model HL4. It was shown that stab-
ilizing effects which could result from internal NOH ؒ ؒ ؒ O᎐C
᎐
α
β
shielding constants referred to above, calculated by Sudmeier
hydrogen bonding (five-membered ring intermediate), expected
J. Chem. Soc., Perkin Trans. 2, 1997
1979

View Article Online
Table 3 Heats of formation (∆fH), steric energy (E), Mulliken charges and H-bond length or H-contact distances (O ؒ ؒ ؒ H or N ؒ ؒ ؒ H) calculated
for several species related to H2L1 and HL4
Distance/Å (AM1)
Mulliken charges (AM1)
∆fH/kcal molϪ1
E/kcal molϪ1
N ؒ ؒ ؒ HO
NЈ ؒ ؒ ؒ HOЈ
Compound a
(AM1)
(MM/AM1)
O
N
NЈ
OЈ
(NϩH ؒ ؒ ؒ O)
(NЈϩH ؒ ؒ ؒ O)
[L1]2Ϫ
Ϫ63.19
Ϫ66.22
85.30
320.49
Ϫ4.41b
Ϫ59.61
Ϫ52.07
96.79
116.99
38.36
43.95
80.23
27.07
42.97
29.18
40.65
12.26
17.42
15.98
Ϫ0.698
Ϫ0.247
Ϫ0.246
Ϫ0.281
—
Ϫ0.684
Ϫ0.245
Ϫ0.290
—
Ϫ0.225
Ϫ0.259
Ϫ0.286
Ϫ0.003
Ϫ0.275
Ϫ0.233
Ϫ0.265
ϩ0.006
Ϫ0.298
Ϫ0.269
—
Ϫ0.266
Ϫ0.254
0.004
Ϫ0.001
—
—
—
—
—
Ϫ0.709
Ϫ0.246
Ϫ0.289
Ϫ0.312
—
—
—
—
—
—
—
2.79
(2.24)
(2.25)
—
—
—
—
—
[H2L1
]
2.73
(4.00
(2.18)
—
[H3L1OOЈ
]
ϩ
[H4L1ON OЈ
]
2ϩ
ON N ЈOЈ
L5
[L4]Ϫ
[HL4
[H2L4
L6
—
]
2.67
(2.31)
—
—
—
O
ϩ
]
ON
Ϫ19.02
Ϫ31.10b
Ϫ80.90
L7
—
Ϫ0.243
—
—
—
—
—
—
HL8
a
[HnLiX ؒ ؒ ؒ Z]p means an n-protonated species of the ligand Li, the n protons being localized on the n following atoms. b Geometry is not at stationary
point.
the backbone molecule. However, the calculated hydrogen bond
lengths for HL4 (dHN ؒ ؒ ؒ H O = 2.67 Å) and hydrogen bond contacts
for H2L1 (dHN ؒ ؒ ؒ H O = 2.73, 2.79 Å) may be overestimated by the
AM1 method.17 The fact that larger distances have been
obtained for the diaminodihydroxamate ligand should be
attributed to coulombic repulsions between the concomitant
positive-induced N-dipoles. So, it became apparent that the
existence of one of the hydrogen bonding interactions prevents
the formation of the second one. This model supports the exist-
ence of interactivity in the protonation process.
Although there are limitations to the application of semi-
empirical methods to charged species,18 we have also simu-
lated the N-protonated species of HL4 and H2L1. As expected,
due to electrostatic effects, a strong interaction between the
ammonium proton and the hydroxamate oxygen atom
(dN H ؒ ؒ ؒ OH = 2.24 Å, H2L1) was observed. The large increase in
energy associated with the second N-protonation, as compared
with the first one, derives from the contribution of electrostatic
terms. It gives support to the low value found for the last N-
protonation constant for this diazadihydroxamate compound.
On the other hand, analysis of the AM1 Mulliken charges,
calculated for the global minimum energy conformations cor-
responding to neutral species of HL4 and H2L1 (see Table 3),
shows that the charge on the amine-nitrogen atoms is more
negative in the mono-amine (Ϫ0.265) than in the corresponding
diamine compound (Ϫ0.256), thus supporting an easier N-
protonation of the former compound when compared with the
latter one. Similar charge differences were found for the corres-
ponding unsubstituted cyclic amines [piperazine (L5), piperi-
dine (L6)], which are also in agreement with differences in the
corresponding basicities (log K1 = 9.84 for L5; log K1 = 11.01 for
L6).11
Fig. 4 Stereoviews of global minimum energy structures calculated for
Regarding the charges on the hydroxamate oxygens, it was
shown that the corresponding anions have a higher charge in
the dihydroxamate (Ϫ0.704) than in the monohydroxamate
compound (Ϫ0.684). These differences may also be related to
differences in the corresponding basicities. Finally, comparison
of the charge distribution on the monoaminohydroxamic acid
(HL4) with the distribution calculated for related models
[N-butylpiperidine (L7) and N-methylcyclohexanylacetohydrox-
amic acid (HL8)] suggests that the electron-withdrawing effect
produced by the hydroxamate moiety on the amine group
is more relevant than the corresponding reverse effect (see
Table 1).
several species of H2L1, HL4 and related model molecules, by the AM1
method. O-Atoms are shaded, and N-atoms are double shaded. H-
Bond (or H-bond type contacts) are shown as dotted lines. The desig-
nation [HnLiX ؒ ؒ ؒ Z]p means an n-protonated species of the ligand Li, the
n protons being localized on the n atoms X and Z.
for the Z form (cis) of the hydroxamate moiety,16 are over-
come by the effect of another hydrogen bonding interaction
NOH ؒ ؒ ؒ N, having a less constrained six-membered ring
structure (involving the hydroxamate proton and the amine
nitrogen atom), and also by the dipole–dipole stabilizing effects
inherent in the adopted hydroxamate E form (trans).
Similarly, modeling the neutral diaminodihydroxamic acid
(H2L1) leads to the global minimum energy conformation hav-
ing the hydroxamate moieties in E (trans) conformation, thus
leaving the hydroxamate protons free for interaction with the
amine-nitrogen atoms, via hydrogen-bonded six-membered ring
structures, tilted ‘upward’ or ‘downward’ from the chair ring of
This molecular modeling approach allowed, therefore, a
comparison of the preferred conformations and the acid–base
behavior of the diaminodihydroxamate ligand (H2L1) and
related models. It was shown that this cyclic α-amino-
hydroxamic acid should have the cyclic diamine backbone
in a chair conformation and the hydroxamate moieties of
1980
J. Chem. Soc., Perkin Trans. 2, 1997

View Article Online
Scheme 1 Protonation scheme of H2L1
the side chain interacting with the nearby amine groups via
OO'
NN'
hydrogen bonded six-membered ring structures, tilted ‘upward’
and ‘downward’. The effect of electronic (or electrostatic)
interactions as well as of the internal hydrogen bonds on the
experimentally observed interactivity of basic sites along the
protonation processes were also demonstrated.
In order to get a clearer characterization of the protonation
scheme of this molecule and an exact picture of its acid–base
properties, we have calculated the protonation microconstants.
These microscopic protonation constants were calculated as
follows.
O–
H3C
O–
C
O
N
N
C
O
N
CH3
ON
O'N
The complete protonation scheme of H2L1 can be seen in
Scheme 1 which shows that as many as 32 microconstants can
be derived for this ligand. Considering, however, the symmetry
relations existing in this molecule, identical basicity can be
assumed for twin donors(e.g. kO = kOЈ, kN = kN Ј, kON = kON Ј). On the
other hand, it is known that protonation at one basic site modi-
fies the basicities of other sites in the molecule. This modific-
ation can be quantified in terms of interactivity parameters
(∆log k).19,20
N, N'
N, O
∆log kN—N' = log kN-log kNN' = log kN '-log kNN'
∆log kN—O = log kN-log kON = log kO- log kON
= log kN'-log k = log kO'-log kON'
N'
O'
N, O'
O, O'
∆log kN—O' = log kN-log kNO' = log kO'-log kON'
= log kN'-log kON' = log kO-log kON'
∆log kO—O' = log kO-log kOO' = log kO'-log kOO'
Scheme 2 Possibilities of interactivities in H2L1
kON = kN × 10∆log k
(2)
(3)
N ᎐O
∆log kxϪy = log kyx Ϫ log k x = log kxy Ϫ log ky
(1)
OЈ
= kO × 10∆log k × 10∆log k
N ᎐O
N ᎐OЈ
k
N,N Ј
Interactivity parameters are assumed to have two important
properties: (i) they have identical values in the same moieties of
different molecules and (ii) they are largely independent of the
protonation stage of sites other than x and y. Thus, the value of
the interactivity parameter can be transferred from one mol-
ecule to another, provided the moiety considered is the same.
Taking the above principles together to quantify the acid–base
properties of H2L1, one has to know independent microcon-
stants and interactivity parameters. For this ligand, the various
possibilities of interactivity are shown in Scheme 2.
From the analysis of the protonation and interactivity
schemes for H2L1 (Schemes 1 and 2), one can conclude that in
addition to the macroconstants, knowledge of the kO and kN
microconstants and four independent interactivity parameters,
∆log kN ᎐N Ј, ∆log kN ᎐O, ∆log kN ᎐OЈ and ∆log kO᎐OЈ, are necessary
for the adequate acid–base characterization of H2L1. Combin-
ing kO and kN microconstants plus the appropriate set of inter-
activity parameters, any microconstant can be obtained [e.g.
eqns. (2) and (3)]
The resulting relationships between the cumulative macro-
constants (β values), the microconstants and interactivity
parameters are shown in eqns. (4)–(7).
β1 = kO ϩ kN ϩ kN Ј ϩ kOЈ = 2kN ϩ 2kO
(4)
O
β2 = kOkN × 10∆log k ϩ kOЈkN × 10∆log k ϩ k kOЈ × 10∆log k
ϩ
N ᎐O
N ᎐OЈ
O᎐OЈ
N
N Ј
kN kN Ј × 10∆log k ϩ k kOЈ × 10∆log k ϩ k kOЈ × 10∆log k
=
N ᎐N Ј
N ᎐OЈ
N Ј᎐OЈ
2kOkN (10∆log k ϩ 10∆log k ) ϩ
N ᎐O
N ᎐OЈ
O
2
∆log kO᎐OЈ
(kN )2 × 10∆log k ϩ (k ) × 10
(5)
N ᎐N Ј
N
∆log kN ᎐N Ј
β3 = kOkN × 10∆log k k × 10
× 10∆log kN ᎐OЈ
ϩ
N ᎐O
O
∆log kN ᎐OЈ
∆log kN ᎐N Ј
kOkN × 10∆log k k × 10
× 10∆log kO᎐OЈ
× 10∆log kN ᎐OЈ
× 10∆logkN ᎐OЈ
ϩ
N ᎐O
N
kOЈkN Ј × 10∆log k k × 10
ϩ
N ᎐O
O
∆log kO᎐OЈ
kOЈkN Ј × 10∆logk k × 10
=
N ᎐O
J. Chem. Soc., Perkin Trans. 2, 1997
1981

View Article Online
Table 4 Microscopic protonation constants of H2L1 in log units (for
assignment of microconstants see Scheme 2)
∆log kN ᎐N Ј
∆log kO᎐OЈ
2kOk2N × 10∆log k × 10∆log k × 10
ϩ
N ᎐O
N ᎐OЈ
2kN k2O × 10∆log k × 10∆log k × 10
=
N ᎐O
N ᎐OЈ
kN = kN Ј = 8.5
kO = kOЈ = 9.1
OЈ
2kOkN × 10∆logk
×
N ᎐O
kON = k = 7.4
kON = k = 8.0
N Ј
OЈ
N Ј
kON Ј = kNOЈ = 7.6
kNOЈ = kON Ј = 8.2
10∆log k (k × 10
ϩ kO × 10∆log kO᎐OЈ
)
(6)
(7)
N
∆log kN ᎐N Ј
N ᎐OЈ
kOOЈ = kOOЈ = 8.8
kNN Ј = kNNNЈ =Ј 4.8
kON N Ј = kON ЈN = 7.0
kNON Ј = kN OЈ = 3.7
β4 = kOkN × 10∆log k k × 10
× 10∆log kN ᎐N Ј
×
N
∆log kN ᎐OЈ
N ᎐O
N Ј
OЈ
k
k
k
= kNN ЈOЈ = 3.9
= kON OЈ = 7.7
= kNON ЈOЈ = 2.8
k
= kOOЈN Ј = 7.8
ON
ON
kO × 10∆log k × 10
× 10∆log k
=
∆log kN ᎐O
OЈ
kNOOЈ = kNOЈO = 6.5
O᎐OЈ
N ᎐OЈ
ON Ј
N Ј
ON OЈ
OЈ
N N ЈO
k
= kON N ЈO = 6.7
2
∆log kN ᎐OЈ
(kO)2(kN )2 × (10∆log k ) × (10
)2 ×
N ᎐O
∆log kN ᎐N Ј
10∆log k × 10
O᎐OЈ
The protonation macroconstants (β values) were reported in
a previous paper.5 Values of log β1, log β2, log β3 and log β4 at
25 ЊC and 0.1  KNO3 are 9.53, 17.99, 214.66 and 27.10,
respectively. The estimations of the four interactivity param-
eters were based on values for model compounds available in
the literature. For example, ∆log kN ᎐N Ј was assumed to be the
same in H2L1 as in piperazine, a symmetrical molecule with an
analogous structural moiety and two binding sites for which
this interactivity parameter is given by eqn. (8). The value
∆log kN ᎐N Ј = log K1 Ϫ log K2 ϩ 0.6
(8)
obtained is ca. Ϫ3.7 to Ϫ4.0 depending on the conditions (ionic
strength).12
Given the large number of bonds separating the two
hydroxamate groups in H2L1, one might assume that no inter-
action exists between these two sites. This means that ∆log
kO᎐OЈ = 0. However the protonation constants of long-chain α–
ω dicarboxylic acids (e.g. adipic, pimelic, azelaic)12 indicate that
log K1 Ϫ log K2 differs somewhat from 0.6 (being 0.73–0.8).
Thus, a small interaction between the two hydroxamates may
also occur in H2L1.
The ∆log kN ᎐O value was obtained from the analog α-
alaninehydroxamic acid.15 The nitrogen and hydroxamate
groups are separated by the same number of bonds in H2L1 and
α-alaninehydroxamic acid. The ∆log kO᎐N value for the latter
compound is Ϫ1.12 and Ϫ1.09 for I = 0.2  (KCl) and I = 1.0 
(KCl), respectively.15
Fig. 5 Concentration distribution curves, calculated for individual
species of H2L1 at different pH (CL = 2 × 10Ϫ2 mol dmϪ3). 1, H4L1; 2,
H3L1ON OЈ; 3, H3L1ON ЈO; 4, H3L1ON N Ј; 5, H3L1N N ЈOЈ; 6, H2L1OOЈ; 7, H2L1
;
;
N OЈ
8, H2L1N ЈO; 9, H2L1ON ; 10, H2L1N N Ј; 11, H2L1N ЈO; 12, HL1O; 13, HL1
OЈ
14, HL1N ; 15, HL1
.
N Ј
Using the results above, all the values of the microconstants
could be calculated and are listed in Table 4 (as logarithms).
Based on the set of values presented in Table 4, microspeciation
curves were obtained and are shown in Fig. 5. From the micro-
constant values one can conclude that the N-donors in this
molecule are much less basic than the O-donors. With increas-
ing pH, the protonated N moieties release most protons in the
acid region, while deprotonation of an OH moiety starts only
above pH 5.
Finally, calculations for the best fit of macroconstants (less
than 0.01 log unit difference for each of the calculated and
experimental log β values), gave the following results.
log kN = 8.52,
∆log kN ᎐N Ј = Ϫ3.70,
∆log kN ᎐OЈ = Ϫ0.95,
log kO = 9.10
Experimental
∆log kN ᎐O = Ϫ1.12
∆log kO᎐OЈ = Ϫ0.32
Reagents
All the hydroxamate ligands were prepared in our laboratory
except the glycinehydroxamic acid, which is commercially avail-
able. All the reactions were monitored by TLC. Analytical
reagents were used as supplied. The solvents were dried accord-
ing to standard methods.21 All reaction products were charac-
terized by the usual methods of analysis: mp, IR, NMR and
mass spectra.
Hence, a first conclusion is that there is no great difference
between the basicities of the hydroxamic moiety and the
piperazine nitrogen. This result is in full agreement with that
expected from model studies.12,15
Comparison of the calculated interactivity parameters with
the corresponding estimated ones shows very good agreement
in the case of ∆log kN ᎐N Ј and ∆log kN ᎐O. However, both ∆log
kO᎐OЈ and ∆log kN ᎐OЈ are somewhat higher than expected. The
value expected for ∆log kO᎐OЈ was based on data from single-
chain model compounds (different dicarboxylic acids) and the
value for ∆log kN ᎐OЈ was also based on an estimation that took
into consideration only the number of bonds separating the
donors concerned. However, if we take into account that the
central piperazine ring transmits the effects via two ethylene
moieties, the difference may be accounted for. In effect, a sig-
nificant difference exists between the corresponding interactiv-
ity parameters on the ‘double connecting ethylene’ piperazine
and the ‘single connecting ethylene’ ethylenediamine (Ϫ3.8 and
Ϫ2.2, respectively). This clearly shows that the effect is greatly
increased by the ‘two way’ interaction manner. This may be the
reason why somewhat higher ∆log kN ᎐OЈ and ∆log kO᎐OЈ values
were obtained for H2L1.
Synthesis
The synthesis of the ligand H2L1 involved the previous prepar-
ation of N-methylchloroacetohydroxamic acid (HL2) which was
then used for N-substitution of the corresponding cyclic
diamine, piperazine (L5), in an alkaline medium, according to a
method described in the literature.5 The synthesis of the model
piperidino(N-methylacetohydroxamic acid) (HL4) employs a
multi-step approach whereby an O-protected hydroxamic acid,
N-methyl-O-benzylchloroacetohydroxamic acid, is synthesized
by acyl condensation with N-methyl-O-benzylhydroxylamine
which is achieved using previously reported methods.4,22 The
hydroxamic acid is then coupled with piperidine, followed by
O-deprotection as the final step.
N-Methyl-O-benzylchloroacetohydroxamic acid. A suspen-
sion of N-methyl-O-benzylhydroxylamine (1.55 g, 8.9 mmol) in
tetrahydrofuran (75 ml) was cooled in an ice bath and potas-
1982
J. Chem. Soc., Perkin Trans. 2, 1997

View Article Online
sium carbonate (2.46 g, 17.8 mmol) was added and left stirring
under N2 for 0.5 h. Chloroacetyl chloride (1.05 g, 9.3 mmol)
was dissolved in the same solvent (10 ml), added dropwise to
the reaction mixture and stirred. After 6 h the reaction was
finished, the reaction mixture was filtered and the filtrate evap-
orated under reduced pressure. The oil residue was dissolved in
ethyl acetate (100 ml) and this solution was washed with water
(50 ml), 0.5  citric acid (2 × 50 ml), 0.6  sodium bicarbonate
(2 × 50 ml) and brine (50 ml). The organic extract was dried
over Na2SO4 and evaporated in vacuo to give the hydroxamic
acid as a colorless oil in 79% yield (1.501 g, 7.0 mmol). ν(neat)/
ing to the N-substituted compounds were then located by
molecular dynamics/molecular mechanics (MD/MM) methods,
using the BIOSYM program.24 A final refinement with full
geometry optimization was then carried out, employing the
PRECISE option of the AM1 semi-empirical SCF-Method,
included in the MOPAC program that is contained in the same
package of software. The AM1 method was chosen because it is
thought to provide an adequate description of hydrogen bond-
ed conformations.25 Although our AM1 calculations neglect
solvation effects on the hydrogen bond interactions, some
compensation of this effect was taken into account in the final
MM/AM1 calculations via the distance dependence option for
the electrostatic interactions (Eelec = Σ qiqj/ε qij, summed over all
pairs of i and j atoms, where qi and qj are the charges on indi-
vidual atoms within each pair and rij is the distance between
them).
cmϪ1 (C᎐O); δ (CDCl ) 3.27 (s, 3H, CH ), 4.12 (s, 2H, CH Cl),
᎐
H
3
3
2
4.89 (s, 2H, ArCH2), 7.41 (m, 5H, ArH); (Anal. Calc. for
C10H12NO2Cl: C, 56.21; H, 5.62, N, 6.56%. Found: C, 56.59; H,
5.83; N, 6.95%).
Piperidino(N-methyl-O-benzylacetohydroxamic acid). A solu-
tion of piperidine (0.24 ml, 2.4 mmol) in dry dimethylform-
amide (30 ml) was cooled in an ice bath under N2. Sodium
hydride (0.06 g, 2.6 mmol) was then added and the mixture
stirred. After 15 min, a solution of N-methyl-O-benzyl-
chloroacetohydroxamic acid (0.51 g, 2.4 mmol) in the same
solvent (10 ml) was added dropwise and the mixture was stirred
for 7 h at room temperature. The reaction mixture was ex-
tracted in ethyl acetate (80 ml) and washed with brine (3 × 50
ml). The organic extract was dried over Na2SO4 and the solvent
was evaporated under reduced pressure to give the reaction
product as an oil in 75% yield (0.47 g; 1.8 mmol). ν(neat)/cmϪ1
1660; δH (CDCl3) 1.60 (m, 6H, NCH2CH2 CH2CH2), 2.47 (m,
4H, NCH2), 3.21 (s, 3H, CH3), 3.23 (s, 2H, CH2CO), 4.89 (s,
2H, ArCH2), 7.41 (m, 5H, ArH).
Acknowledgements
The authors thank the Junta Nacional de Investigação
Científica (JNICT) for financial assistance: B. N. for the grant
OTKA T 175070; E. F. for grant OTKA T 023612.
References
1 B. F. Matzanke, G. Mueller and K. N. Raymond, in Iron Carriers
and Iron Proteins, ed. T. M. Loehr, VCH, New York, 1989, ch. 1;
K. N. Raymond, Pure Appl. Chem., 1994, 66, 773.
2 A. L. Crumblis, in Handbook of Microbial Iron Chelates,
ed. G. Winkelmann, CRC Press, Boca Raton, FL, 1991, ch. 7.
3 M. A. Santos, M. A. Esteves, M. C. T. Vaz and M. L. S. Gonçalves,
Inorg. Chim. Acta, 1993, 214, 47.
4 M. A. Esteves, M. C. T. Vaz, M. L. S. Gonçalves, E. Farkas and
M. A. Santos, J. Chem. Soc., Dalton Trans., 1995, 2565.
5 M. A. Santos, M. A. Esteves, M. C. T. Vaz and M. L. S. Gonçalves,
J. Chem. Soc., Dalton Trans., 1993, 927.
6 M. A. Santos, M. A. Esteves and J. M. G. Martinho, J. Chem. Soc.,
Dalton Trans., 1993, 3123.
7 C. J. Carrano and K. N. Raymond, J. Am. Chem. Soc., 1978, 100,
5371.
8 J. R. Ascenso, M. A. Santos, J. J. R. Fraústo da Silva, M. Cândida,
M. C. T. Vaz and M. G. B. Drew, J. Chem. Soc., Perkin Trans. 2,
1990, 2215.
9 T. L. Sayer and D. L. Rabenstein, Can. J. Chem., 1976, 54, 3392.
10 B. Kurzak, H. Kozlowski and E. Farkas, Coord. Chem. Rev., 1992,
114, 169.
Piperidino(N-methylacetohydroxamic acid), HL4. A solution
of piperidino(N-methyl-O-benzylacetohydroxamic acid) (0.46
g, 1.8 mmol) in methanol (30 ml) was treated with Pd on carbon
(10%, 0.15 g) under H2 (1 atm) and the mixture was stirred for
4 h at room temperature. The reaction mixture was filtered
and the solvent was evaporated. The residue was then chrom-
atographed on a silica gel column, using methanol–dichloro-
methane (2:5) as eluent. The desired product was collected
and recrystallized from ethanol–diethyl ether to yield the
final product as white crystals (157 mg, 51%); mp 86–88 ЊC.
ν(KBr)/cmϪ1 1650 (C᎐O); δ (D O; pD = 8.58) 1.64 (m, 2H,
᎐
H
2
NCH2CH2CH2), 1.86 (m, 4H, NCH2CH2), 3.21 (m, 7H,
NCH3 ϩ NCH2), 3.97 (s, 2H, CH2CO); m/z 172 [Mϩ], 170
[M Ϫ 2H]ϩ, 155 [M Ϫ OH]ϩ; (Anal. Calc. for C8H16N2O2ؒ
0.5HCl: C, 50.46; H, 8.67; N, 14.72%. Found: C, 50.51; H, 8.98;
N, 14.34%).
11 H. Irving and L. D. Pettit, J. Chem. Soc., 1963, 3051.
12 A. E. Martell and R. M. Smith, Critical Stability Constants, Plenum
Press, New York, 1975, vol. 6.
13 J. L. Sudmeier and C. N. Reilley, Anal. Chem., 1964, 36, 1699; 1707.
14 J. F. Desreux, E. Merciny and M. F. Loncin, Inorg. Chem., 1981, 20,
987.
NMR measurements
The dissociation macroconstants were determined by potentio-
metric titrations as described previously.5 The 1H NMR spectra
were recorded on a Varian Unity 300 spectrometer at 25 ЊC.
Solutions of the ligands (2 × 10Ϫ2 mol dmϪ3) for NMR pH
titrations were made in D2O with sodium 3-(trimethylsilyl)-
[2,2,3,3-2H4]propionate as reference. The pD* of the solution
(operational pD, since the pH meter was standardized with
conventional buffers at pH 4 and 7) was adjusted with DCl and
KOD, and was measured with a 420A Orion pH meter,
equipped with a combined Ingold U402-M3-S7/200 micro-
electrode.
15 E. Farkas, T. Kiss and B. Kurzak, J. Chem. Soc., Perkin Trans. 2,
1990, 1255.
16 M. A. Santos, A. M. Lobo and S. Prabhakar, Rev. Port. Quím.,
1989, 31, 48.
17 J. Rodrigues, J. Comput. Chem., 1994, 15, 183.
18 B. H. Besler, K. M. Merz and P. Roldman, J. Comput. Chem., 1990,
11, 431.
19 B. Noszál, in Biocoordination Chemistry, ed. K. Burger, Ellis Hor-
wood, New York, 1990, p. 18.
20 B. Noszál and R. Kassai-Táncsos, Talanta, 1991, 38, 1439.
21 P. D. Perrin, W. L. F. Armarego and D. R. Perrin, Purification of
Laboratory Chemicals, Pergamon Press, London, 1965.
22 B. H. Lee, G. F. Gerfen and M. J. Miller, J. Org. Chem., 1984, 49,
248.
23 U. Burkert and N. L. Allinger, ACS Monograph 177, Washington
DC, 1982.
24 INSIGHT/DISCOVER: Insight II, version 95.0, Biosym Inc., San
Diego, CA, 1995.
Calculation of microconstants
Microconstants and interactivity parameters were calculated by
using the ‘Scientist’ non-linear curve-fitting software adjusted
to the relationships in eqns. (4)–(7).
25 M. J. S. Dewar, E. G. Zoebisch, E. F. Healy and J. J. P. Stewart,
J. Am. Chem. Soc., 1985, 107, 3902.
Molecular modeling simulations
To perform the molecular modeling of H2L1, HL4 and related
species, we started with both the most stable cyclic backbone
conformations, chair and twist-boat, for piperazine (L5) and
piperidine (L6), as derived by Burkert and Allinger 23 for six-
membered rings. The minimum energy geometries correspond-
Paper 7/02138K
Received 23rd February 1997
Accepted 29th May 1997
J. Chem. Soc., Perkin Trans. 2, 1997
1983