Evaluation of 4-hydroxy-6-methyl-3-pyridinecarboxylic acid and 2,6-dimethyl-4-hydroxy-3-pyridinecarboxylic acid as chelating agents for iron and aluminium

Article abstract of DOI:10.1016/j.ica.2011.04.009

4-Hydroxy-6-methyl-3-pyridinecarboxylic acid (DQ6) and the new compound 2,6-dimethyl-4-hydroxy-3-pyridinecarboxylic acid (DQ726) were evaluated for possible application for iron (Fe) and aluminium (Al) chelation therapy. Metal/ligand solution chemistry, cytotoxicity, octanol/water partitioning (D_o/w), and chelation efficiency were studied. The solution chemistry of the two ligands with Fe(III) and Al(III) was investigated in aqueous 0.6 m (Na)Cl at 25 °C by means of potentiometric titrations, UV-Vis spectrophotometry, and ¹H NMR spectroscopy. DQ6 exhibited a high coordination efficiency towards Al(III). Fe(III)/DQ6, Al(III)/DQ726, and Fe(III)/DQ726 complexes were less stable. These results were confirmed by chelation efficiency measurements performed in an octanol/aqueous solution. Accordingly, the effects of the substitution at various ring positions of 4-hydroxy-3-pyridinecarboxylic acid were rationalised. Partitioning experiments at pH 7.4 showed both DQ6 and DQ726, and their Fe(III) and Al(III) complexes, to be hydrophilic. The toxicity of DQ6 and of DQ726 was investigated with human cancer cell lines and normal human primary cells: no cytotoxic effects were observed up to 0.1 mM, following a 3 days exposure. According to our results, DQ6 has the favourable properties to be a chelating agent for Al.

Full text of DOI:10.1016/j.ica.2011.04.009

Inorganica Chimica Acta 373 (2011) 179–186
Contents lists available at ScienceDirect
Inorganica Chimica Acta
journal homepage: www.elsevier.com/locate/ica
Evaluation of 4-hydroxy-6-methyl-3-pyridinecarboxylic acid
and 2,6-dimethyl-4-hydroxy-3-pyridinecarboxylic acid as chelating
agents for iron and aluminium
Annalisa Dean a, Maria Grazia Ferlin b, Paola Brun c, Ignazio Castagliuolo c, Robert A. Yokel d, Alfonso Venzo e,
G. Giorgio Bombi a, Valerio B. Di Marco a,
⇑
a Dipartimento di Scienze Chimiche, Università di Padova, via Marzolo 1, 35131 Padova, Italy
b Dipartimento di Scienze Farmaceutiche, Università of Padova, via Marzolo 5, 35131 Padova, Italy
c Dipartimento di Istologia, Microbiologia e Biotecnologie Mediche, via Gabelli 63, 35121 Padova, Italy
d Pharmaceutical Sciences Department, 511C Multidisciplinary Science Building, University of Kentucky Academic Medical Center, Rose Street, Lexington, KY 40536-0082, USA
e CNR, Istituto di Scienze e Tecnologie Molecolari, via Marzolo 1, 35131 Padova, Italy
a r t i c l e i n f o
a b s t r a c t
Article history:
4-Hydroxy-6-methyl-3-pyridinecarboxylic acid (DQ6) and the new compound 2,6-dimethyl-4-hydroxy-
3-pyridinecarboxylic acid (DQ726) were evaluated for possible application for iron (Fe) and aluminium
(Al) chelation therapy. Metal/ligand solution chemistry, cytotoxicity, octanol/water partitioning (Do/w),
and chelation efficiency were studied. The solution chemistry of the two ligands with Fe(III) and Al(III)
was investigated in aqueous 0.6 m (Na)Cl at 25 °C by means of potentiometric titrations, UV–Vis spectro-
photometry, and 1H NMR spectroscopy. DQ6 exhibited a high coordination efficiency towards Al(III).
Fe(III)/DQ6, Al(III)/DQ726, and Fe(III)/DQ726 complexes were less stable. These results were confirmed
by chelation efficiency measurements performed in an octanol/aqueous solution. Accordingly, the effects
of the substitution at various ring positions of 4-hydroxy-3-pyridinecarboxylic acid were rationalised.
Partitioning experiments at pH 7.4 showed both DQ6 and DQ726, and their Fe(III) and Al(III) complexes,
to be hydrophilic. The toxicity of DQ6 and of DQ726 was investigated with human cancer cell lines and
normal human primary cells: no cytotoxic effects were observed up to 0.1 mM, following a 3 days expo-
sure. According to our results, DQ6 has the favourable properties to be a chelating agent for Al.
Ó 2011 Elsevier B.V. All rights reserved.
Received 21 December 2010
Received in revised form 4 April 2011
Accepted 8 April 2011
Available online 19 April 2011
Keywords:
Iron
Aluminium
Chelation therapy
Potentiometry
NMR
In vitro toxicity
1. Introduction
The simplest HPs, the unsubstituted DT0 and DQ0, have a dis-
tinct disadvantage: although their affinity towards Fe(III) and
Hydroxypyridinecarboxylic acids (HPs, Table 1) are new poten-
tial chelating agents for iron (Fe) and aluminium (Al) because they
display a number of favourable properties. They form strong com-
plexes with both Fe(III) and Al(III) [1–6], and have a very low affin-
ity towards Zn(II) [7,8], which suggests the absence of essential
metal decorporation in vivo [9]. They have a low molecular weight,
which is a prerequisite for oral activity [10]. Toxic side effects in-
duced by redox activity are unlikely for both the free ligands and
the Fe(III)/ligand complexes [3,6]. The HPs investigated so far
(Table 1) display negligible toxic effects (IC50 > 0.1 mM) to cancer
cell lines and primary human cells, following a 3 days exposure
[3,6]. DT0 is non-toxic towards animals, and it was proposed as
an aspirin-like drug [11,12]. An analogue, 4-pyridoxic acid (3-
hydroxy-5-hydroxymethyl-2-methyl-4-pyridinecarboxylic acid),
the main metabolite of vitamin B6, is also non-toxic [13].
Al(III) is very high, it is still much lower than that of chelators
available presently, such as deferiprone (L1). This affinity was sig-
nificantly increased by methyl substitutions at the pyridinic ring.
DT1 and DT2 showed a higher coordination strength than that of
DT0. The complexation strength of DQ1 towards Fe(III) was slightly
higher than that of DQ0; that towards Al(III) did not change. The 2-
methyl substitution of DQ0 (which gives DQ2) decreased signifi-
cantly the coordination strength towards Fe(III) and Al(III). The
complexes of DQ716 with both Fe(III) and Al(III) are much more
stable than those of the other HPs examined so far [6], so that this
compound was proposed as a chelating agent for Fe and Al.
More than one HP having the proper chemical requirements
would be needed, in order to perform a pharmacological screening
(e.g. evaluation of chelation efficiency and toxicity in vivo) and
eventually individuate at least one compound to be used as drug.
Moreover, the effects of the methyl substitution at different ring
positions on the metal stability of the complexes are still not clear.
The metal–ligand solution chemistry of other derivatives should be
⇑
Corresponding author. Tel.: +39 0498275219; fax: +39 0498275271.
E-mail address: valerio.dimarco@unipd.it (V.B. Di Marco).
0020-1693/$ - see front matter Ó 2011 Elsevier B.V. All rights reserved.
doi:10.1016/j.ica.2011.04.009

180
A. Dean et al. / Inorganica Chimica Acta 373 (2011) 179–186
Table 1
250.00). Starting materials as well as solvents were purchased
from Sigma (Milan, Italy).
Hydroxypyridinecarboxylic acids (HPs) examined so far.
Name (IUPAC)
Acronym
References
2.1.1. 3-(Dimethylaminomethylene)-4-oxo-6-methyl-2-pyrone (1)
About 10 mL (d = 0.89 mg/mL, 74 mmol) N,N-dimethylformam-
ide dimethyl acetale were slowly added to a stirred suspension of
4-hydroxy-6-methyl-2-pyrone (5 g, 40 mmol) in 10 mL dioxane:
the starting material dissolved and the solution became brown.
The reaction ran at a temperature of 15 °C for 2 h, when a precip-
itate formed. The precipitate was collected, washed with cold diox-
ane and acetone, and dried in vacuo. Yield 60% (literature [14]
72%); mp 148–150 °C (literature [14] 152–154 °C); HRMS (ESI) cal-
culated for [M+H]+ m/z 179.169, found 180.174. 1H NMR (DMSO-
d2) d 1.10 (t, 3H, C-CH3), 2.35 (q, 2H, C-CH3), 3.22 (s, 3H, N-CH3),
3.48 (s, 3H, N-CH3), 5.55 (s, 1H, olefinic proton); 8.22 (s, 1H, olefinic
proton).
3-Hydroxy-4-pyridinecarboxylic acid
3-Hydroxy-1-methyl-4-pyridinecarboxylic
acid
3-Hydroxy-2-methyl-4-pyridinecarboxylic
acid
4-Hydroxy-3-pyridinecarboxylic acid
4-Hydroxy-1-methyl-3-pyridinecarboxylic
DT0 (3H4P)
DT1
(1M3H4P)
DT2
[1–3]
[3–5]
[3,6]
DQ0 (4H3P)
DQ1
(1M4H3P)
DQ2
[1,2,6]
[4–6]
acid
4-Hydroxy-2-methyl-3-pyridinecarboxylic
acid
1,6-Dimethyl-4-hydroxy-3-pyridinecarboxylic
acid
[6]
[6]
DQ716
2.1.2. 4-Hydroxy-6-methylpyridin-3-carboxyl acid (2)
About 1 g (5.58 mmol) of pyrone derivative 1 was suspended in
30% aqueous ammonia (20 mL) and 1 mL NH(CH3)2. After stirring
for 30 min at room temperature, the solution was evaporated un-
der reduced pressure to about 1/3 of its volume and the remaining
solution cooled (ice-bath) and acidified to pH 3 with HCl 1 M. The
formed precipitate was collected and dried yielding a solid product
which was re-crystallized from water to give pure product. Yield
64% (literature [14] 49%); mp 264–266 °C (literature [14] 267–
268 °C); HRMS (ESI) calculated for [M+H]+ m/z 153.125, found
154.258. 1H NMR (D2O + NaOD) d 2.15 (s, 3H, C-CH3), 7.22 (s,
1H), 8.52 (s, 1H). Anal. Calc. for C8NO3H9: C, 57.48; H, 5.43; N,
8.38. Found: C, 51.30; H, 5.60; N, 7.40%.
Fig. 1. 4-Hydroxy-6-methyl-3-pyridinecarboxylic acid (DQ6), and 2,6-dimethyl-4-
hydroxy-3-pyridinecarboxylic acid (DQ726) shown in their most protonated forms
(H3L+).
studied, so that the best strategy can be developed for the identifi-
cation of the strongest Fe(III) and Al(III) chelators.
2.1.3. Ethyl 4-hydroxy-2,6-dimethylpyridin-3-carboxylate (3)
About 5 g (35 mmol) of 2,2,6-trimethyl-4(1H)-1,3-dioxin-4-one
and 2.5 g (19 mmol) of ethyl-(Z)-3-aminobut-2-enoate (ethyl-
3-crotonate) were heated in a flask at 120–130 °C for 1 h, until
no more water formed (Dean–Stark apparatus). On cooling, diethyl
ether (20 mL) was added to the reaction mixture and a precipitate
formed, which was collected and washed with a small amount of
diethyl ether and dried to give solid product; yield 45% (literature
[15] 40%); mp 162 °C (literature [15] 168 °C); HRMS (ESI) calcu-
lated for [M+H]+ m/z 167.160, found 168.178. 1H NMR (D2O +
NaOH) d 1.30 (t, 3H, C-CH3), 2.55 (s, 6H, CH3), 4.29 (q, 2H, C-
CH2), 6.81 (s, 1H).
This paper describes our evaluation of 4-hydroxy-6-methyl-
3-pyridinecarboxylic acid (DQ6) and 2,6-dimethyl-4-hydroxy-
3-pyridinecarboxylic acid (DQ726) as possible chelating agents
for Fe and Al (Fig. 1). DQ6 has been synthesised and characterised
previously [14], but its use as chelating agent for iron and alumin-
ium was never explored. According to our knowledge, DQ726 is a
new compound and it has never been prepared yet. Both DQ6
and DQ726 were synthesised, and their coordination properties to-
wards Fe(III) and Al(III) were studied by means of potentiometric,
UV–Vis, and (in the case of Al(III)) 1H NMR measurements. Their
octanol/water partitioning coefficient (Do/w), as well as their effi-
ciencies in chelating Fe(III) and Al(III) at physiological pH, were
determined in vitro. Their cytotoxicity was assessed on human can-
cer cell lines and primary cultures of human cells.
2.1.4. 4-Hydroxy-2,6-dimethylpyridin-3-carboxyl acid (4)
About 1 g (5.98 mmol) of pyridin-carboxylate derivative 3 was
suspended in NaOH 0.5 M (20 mL) and refluxed for 4 h. Then, the
resulting solution was cooled, acidified with aqueous HCl 2 M to
pH 6 and extracted with chloroform to remove the unreacted ester.
Further acidification of aqueous solution to pH 3 and cooling (ice-
bath) gave a white solid which was re-crystallized from water to
yield pure crystalline product. Yield 90%; mp = 320–325 °C
(decomp.); HRMS (ESI) calculated for [M+H]+ m/z 167.160, found
168.178; 1H NMR (DMSO-d6) d 2.33 (s, 3H, CH3), 2.71 (s, 3H,
CH3), 6.66 (s, 1H, H-5), 12.82 (s, 1H, NH), 16.15 (s, 1H, COOH). Anal.
Calc. for C8NO3H9: C, 57.33; H, 5.42; N, 8.36. Found: C, 54.40; H,
5.05; N, 7.80%.
2. Experimental
2.1. Synthesis
Melting points were determined on a Gallenkamp MFB 595
010M/B capillary melting point apparatus, and are uncorrected.
Infrared (IR) spectra were measured on a Perkin–Elmer 1760 FT-
IR spectrometer using potassium bromide pressed disks. Values
are expressed in cmÀ1 1H NMR spectra were recorded on Varian
.
Gemini (200 MHz) and Bruker (300 MHz) spectrometers, using
the indicated solvents. NMR data are reported as d values (ppm)
relative to tetramethylsilane as an internal standard. Elemental
analyses were performed in the Microanalytical Laboratory,
Department of Pharmaceutical Sciences, University of Padova,
using a Perkin–Elmer elemental analyser model 240B; results fell
in the range of calculated values 0.4%. Mass spectra were obtained
with a Mat 112 Varian Mat Bremen (70Ev) mass spectrometer and
Applied Biosystems Mariner System 5220 LC/MS (nozzle potential
2.2. Thermodynamic study
All potentiometric measurements were performed using a Radi-
ometer ABU93 tri-burette apparatus. UV–Vis and 1H NMR spectra
were recorded using a Perkin–Elmer Lambda 20 spectrophotome-
ter and a Bruker DRX-400 spectrometer operating at 400.13 MHz,
respectively. All analyte concentrations were expressed in the
molality scale (mol/kg of water). For potentiometric and UV–Vis

A. Dean et al. / Inorganica Chimica Acta 373 (2011) 179–186
181
Table 3
measurements, working solutions of HCl (0.13 m), NaOH (0.13 m),
Acidic properties of DQ6 and DQ726. For the definition of L see caption of Fig. 1.
FeCl3 (0.045 m, containing HCl 0.33 m), and AlCl3 (0.11 m, contain-
ing HCl 0.40 m), were prepared and standardised as described pre-
viously [16,17]. Both ligands were used as synthesised to prepare
0.0051 m (DQ6) and 0.0041 m (DQ726) working solutions. The io-
nic strength of all solutions was adjusted to 0.6 m (0.594 M) (Na)Cl
[16]. Solutions for 1H NMR measurements were prepared by dis-
solving weighed amounts of ligand and AlCl3 (Carlo Erba, 98%
min) in D2O (Aldrich, 99.9% atom D). The internal reference was
Me3SiCH2CH2COOH (TSP, Aldrich 99%+).
Potentiometric measurements were carried out at 25.0 0.1 °C;
duplicate potentiometric measurements were performed using
two glass electrodes (VWR 662-1792) and an Ag/AgCl/0.6 m NaCl
reference electrode [18,19] with a J-shaped junction. In addition
to glass electrode calibration, base standardisation, and ligand
standardisation experiments [16], titrations of metal/ligand mix-
tures were performed. Metal ion to ligand ratios were from 1:1
to 1:9, Fe(III) concentrations ranged from 1.8 Â 10À4 to
1.1 Â 10À3 m, Al(III) concentrations ranged from 2.7 Â 10À4 to
7.6 Â 10À4 m. Experimental points were regarded as non-equilib-
rium points and rejected in subsequent data analysis if the poten-
tial did not stabilize within 6 min after each addition of titrant
(maximum allowed rate of e.m.f. change: 0.05 mV/min). In the
acidic pH range of Al(III)/DQ6 titrations, the measured e.m.f. drifted
and reached a constant value only after ca. 2 h, suggesting a low
complex formation rate. This finding is in agreement with previous
results with other HPs [1,3,4,6,20]. Experimental details regarding
the handling of the slow kinetics during the titrations are reported
[20].
Species
DQ6
DQ726
pKa
n
pKa
n
H3L+
H2L
0.40 0.01a
6.32 0.01
11.20 0.04
1
10
10
0.59 0.02a
5.32 0.02
11.83 0.01
1
12
12
HLÀ
n = Number of UV–Vis wavelengths or number of potentiometric titrations used for
data elaboration. Uncertainty is given by the fitting algorithm and represents an
estimate of the reproducibility of the given value. In the case of UV–Vis data, it is a
‘‘within sample’’ and ‘‘within wavelength’’ reproducibility.
a
Values obtained by UV–Vis.
Table 4
Stability constants for metal/ligand complexes, at 25 °C in aqueous (Na)Cl 0.6 m
(reactions: m M3+ + l L2À + h H+ ¢ MmLlHh3mÀ2l+h). For the definition of n see footnote
of Table 3.
Species
Fe(III)/DQ6
Al(III)/DQ6
Log b
n
Log b
n
MLH2+
ML+
20.51 0.03a
1
18.65 0.01
18.76 0.09a
–
36.24 0.06
36.9 0.1a
30.1 0.1
21.75 0.05
52.8 0.1
–
10
2
17.00 0.03
38.93 0.04
4
8
+
ML2H2
10
2
4
4
6
ML2ÀH
ML2
34.31 0.03
–
55.80 0.04
47.99 0.07
8
ML3H3À
8
8
ML3H2
Fe(III)/DQ726
Al(III)/DQ726
Log b
n
Log b
n
All stability constants were calculated using the computer pro-
gram PITMAP [21] by simultaneous fitting of all data sets. The pro-
gram minimises the sum of the squares of the differences between
experimental and calculated e.m.f. values. Optimisation is per-
formed using pitmapping [22] or simplex [23] as nonlinear least
squares algorithms. Mass balance equations were solved, i.e. species
concentrations at equilibrium were obtained, by means of the New-
ton–Raphson method [23]. The stability constants for metal/hydro-
xo complexes have been taken from the literature: log bFeOH = À2.87,
MLH2+
ML+
19.19 0.05a
1
17.70 0.01
6
1
17.49 0.05a
16.51 0.08
36.84 0.06
32.9 0.1
26.98 0.05
53.12 0.04
47.6 0.1
2
4
4
4
4
3
–
+
ML2H2
34.37 0.03
27.8 0.1
–
49.47 0.06
–
6
4
ML2ÀH
ML2
ML3H3À
ML3H2
4
a
Values obtained by UV–Vis.
log bFe(OH)2 = À6.16 [17], log bFe(OH)3 = À12.16, log bFe(OH)4
À22.16, log bFe2(OH)2 = À2.9, log bFe3(OH)4 = À6.3, log bFe12(OH)34
=
=
values for the proton signals were obtained by a pre-scan delay of
10 s. The assignment of the proton resonances was performed by
standard chemical shift correlations and NOESY measurements
when necessary. Spectra were recorded in D2O solutions contain-
ing free ligand, and in solutions containing Al(III) and the ligand.
The pH was measured with a Crison 5014 combined glass electrode
previously calibrated in buffered aqueous solutions at pH = 4 and
7. The values of pD were computed by adding 0.41 pH units to
the pH meter readings [26] in order to correct for isotopic and sol-
vent effects due to the use of D2O instead of H2O.
À48.9 [24], log bAlOH = À5.52, log bAl(OH)2 = À11.3, log
bAl(OH)3 = À17.3, log bAl(OH)4 = À23.46, log bAl3(OH)4 = À13.57, log
bAl13(OH)32 = À109.2 [25].
The UV–Vis measurements are summarised in Table 2. The pH
was measured with the same electrodes and procedures as for
potentiometric titrations. pH values below 2 were computed from
the stoichiometric concentration of HCl, because the [H+] modifica-
tions produced by the other species were negligible under these
conditions. In other cases the pKa1 for the ligand (Table 3), the
log b values for some metal/ligand complexes (Table 4) and the val-
ues of
e (molal absorbivity coefficient, Table 2) at the given wave-
2.3. n-Octanol/water distribution and chelation efficacy
lengths were computed by the program PITMAP.
1H NMR spectra were obtained at 25 °C. Chemical shift values
are given in d units with reference to internal TSP. Suitable integral
The methods were described in Ref. [27] for Al(III) and in Ref. [2]
for Fe(III). DQ6 and DQ726 efficiencies were determined at pH = 7.4
Table 2
Experimental details for UV–Vis measurements. In the last column, the optimised
e
values for some species are reported.
Solutions
CM Â 103 (m)
–
0.537
1.172
0.595
CL Â 103 (m)
0.118
0.682
0.861
0.876
pH
k (nm)
Path cell (cm)
e
 10À3 (molÀ1  kg cmÀ1
)
DQ6
0.26–1.17
0.30–1.27
2.29–2.94
2.74–3.94
227, 250, 280
400
244, 249
244, 249
1.0
1.0
0.1
0.1
2.48 0.02 (H2L+)5.51 0.01 (HL) (at 250 nm)
Fe3+ + DQ6
Al3+ + DQ6
Al3+ + DQ6
1.60 0.07 (FeLH2+
)
6.47 0.07 (AlLH2+) (at 249 nm)
14.9 0.1 (AlL2H2+) (at 249 nm)
DQ726
–
0.303
0.646
0.508
0.25–1.25
0.28–1.43
2.58–4.02
280
400
244
1.0
1.0
0.2
0.19 0.02 (H3L+)1.13 0.02 (H2L)
Fe3+ + DQ726
Al3+ + DQ726
0.650
0.518
0.80 0.04 (FeLH2+
)
4.43 0.04 (AlLH2+
)

182
A. Dean et al. / Inorganica Chimica Acta 373 (2011) 179–186
in a system containing 2 mL n-octanol, 2 mL of an aqueous solution
containing 1 Â 10À3 mol/L ligand, and a form of Al or Fe that has
very limited solubility (aluminium oxide or ferric acetate basic)
added in excess of the ability of the ligand to bind the metal if
all ligand associated with metal. Approximately 5 mg of alumin-
ium oxide or ferric acetate basic was added, introducing ꢀ50 or
25 Â 10À6 moles of Al or Fe; ꢀ75 and 40-fold more Al or Fe, respec-
tively, than the 2 mL of 1 Â 10À3 mol/L ligand (2 Â 10À6 moles)
could complex, considering a 1:3 metal:ligand stoichiometry.
Efficiency was calculated as the sum of the increased molar con-
centration of metal in the presence minus the absence of the li-
gand, divided by the molar concentration of the ligand  100%,
assuming a 1:3 metal–ligand complex for these bidentate ligands.
When the stoichiometry was different from 1:3, the efficiency was
corrected. Four replicate trials, each condition with duplicate
observations, were conducted.
starting from commercial 4-hydroxy-6-methyl-2-pyrone [14],
which was reacted with N,N-dimethylformamide dimethyl acetale
to give the 3-dimethylaminomethylene derivative 1 (60% yield).
The latter, by treatment with aqueous ammonia (NH4OH 30%),
dimethylamine as catalyst and then with HCl 1 M, provided the de-
sired acid 2 in a 60% yield. 1H NMR, melting point and elemental
analysis data were in agreement with reported values.
For the synthesis of the new compound DQ726 (4), a two steps
procedure was adopted similarly as previously reported [15]
(Scheme 2). The reaction between commercial 2,2,6-trimethyl-
4(1H)-1,3-dioxin-4-one and ethyl-(Z)-3-aminobut-2-enoate (ethyl
3-aminocrotonate) in a 1:1 ratio yielded the known ethyl ester 3
in a 40% yield. The latter was then transformed into the new acid
4 by treatment with NaOH 0.5 M followed by acidification with
HCl 2 M at pH 3 (90% yield).
3.2. Acidity constants of DQ6 and DQ726
2.4. Cytotoxicity assays
Potentiometric titrations of each ligand allowed the determina-
tion of some pKa values, which are reported in Table 3. Accurate
acidity constants at pH values lower than ca. 1.5–2 could not be ob-
tained from potentiometric measurements, because in these condi-
tions the pH modification due to the acid–base equilibria was
negligible. For each ligand, the pKa value for the most protonated
form (H3L+) was determined by means of UV–Vis measurements.
All pKa values are in agreement with those previously observed
for other HPs [1–6].
An unequivocal assignment of the pKa values of DQ6 and DQ726
is not possible, because 4-hydroxypyridine derivatives can adopt a
chinoid electronic configuration in tautomeric equilibrium with
the corresponding aromatic form. However, the formation of a
strong intramolecular bond between the deprotonated COOÀ and
the protonated phenolic OH is expected to shift the tautomeric
equilibrium towards the aromatic form. Therefore, the pKa1
(H3L+ ? H2L), pKa2 (H2L ? HLÀ), and pKa3 (HLÀ ? L2À) of both li-
gands can be confidently assigned to the carboxylic COOH, to the
pyridinic NH, and to the phenolic OH, respectively.
Tests were carried out with DQ6 and DQ726, following the pro-
cedure previously described [3]. Briefly, the cytotoxic activity was
determined using
a standard 3-[4,5-dimethylthiazol-2-yl]-2,5-
diphenyltetrazodium bromide (MTT)-based colorimetric assay
(Sigma). Quadruplicate cultures were employed for each treatment
using human cancer cell lines (OVCAR, OE33, A549, and HeLa; ob-
tained from American Type Culture Collection, ATCC) and normal
human cells (keratinocytes, fibroblasts). The cytotoxic effect of
each tested compound was evaluated by the ratio between the
number of living cells present in the sample and in a blank treated
with the solvent only, following a 3 days exposure.
3. Results and discussion
3.1. Synthesis of DQ6 and DQ726
Following a previously described route depicted in Scheme 1
and 4-hydroxy-6-methylnicotinic acid DQ6 (2) was synthesised
Scheme 1. Synthesis of DQ6 (2) [14].
Scheme 2. Synthesis of DQ726 (4).

A. Dean et al. / Inorganica Chimica Acta 373 (2011) 179–186
183
3.3. Metal/ligand complexes: potentiometric results
products of the Fe(III) complexes are probably hydroxo species, i.e.
the deprotonation occurs at the Fe(III)-coordinated water mole-
cules. The pKa values of the complexes FeLH and FeL2H2, which
can be easily computed from data in Table 4, are similar to
the pKa of free Fe(III) (2.87) (e.g. for Fe(III)/DQ6 pKa(FeLH) = 3.51,
pKa(FeL2H2) = 4.62), whereas for the species FeL3H3, which likely
has no coordinated water, pKa = 7.81. Thus, for example, the com-
plex FeL2H should be better referred to as FeL2H2(OH). In the case
of Al(III) complexes, the intrinsic acidity of the coordinated water
molecules (pKa for Al(H2O)6 = 5.52) is comparable to that of the
pyridinic nitrogen of DQ6 and DQ726. Therefore, the deprotonated
Al(III) complexes are likely mixtures of non-hydroxo and hydroxo
species, e.g. AlL2H + AlL2H2(OH).
Potentiometric data were elaborated in a sequential manner, as
described previously [1], to avoid the difficulties produced by the
strong correlation between the parameters to be optimised in the
presence of a large number of metal/ligand species. The stoichiom-
etry and the values of the stability constants of the metal/ligand
complexes identified in solution are shown in Table 4. The distribu-
tion diagrams for metal/ligand solutions, computed for solutions
containing 5 Â 10À4 m metal ion and 2 Â 10À3 m ligand, are re-
ported in Fig. 2 (for DQ6) and in Fig. S1 of Supplementary material
(for DQ726).
In the case of DQ6, the predominant species for both metal ions
are MLH, ML2H2 and ML3H3 (for simplicity, charges will be gener-
ally omitted in the formulae, except in Figures and Tables). The
deprotonation products of the main metal/DQ6 complexes can be
also detected. For DQ726, the species ML3H3 is less important for
both metal ions, and a number of deprotonation products become
significant if not predominant (e.g. FeL2H). Some of them (e.g. FeL2)
were observed only at very large ligand-to-metal ratios, when the
pH of Fe(OH)3 precipitation was shifted towards more basic values.
In all cases, apart for Al(III)/DQ6, the precipitation of the metal
hydroxide begins below physiological pH.
3.4. Metal/ligand complexes: UV–Vis results
UV–Vis spectra for Fe(III)/DQ6 and Al(III)/DQ6 solutions at sev-
eral pH values are shown in Fig. S2 in the Supplementary material.
All spectra show the typical absorption of substituted pyridinic
rings. A weak charge-transfer peak appears at around 400 nm
when Fe(III) coordinates to each ligand. Al(III) complexation pro-
duces an increase of the absorption coefficients at almost all UV
wavelengths.
According to the hard-soft rule and due to the chelation princi-
ple, the metal coordination occurs most likely via the carboxylate
and the deprotonated hydroxylic groups, whereas the nitrogen
atom remains protonated in the MLH-type species. Deprotonation
UV–Vis data allowed the determination of the stability con-
stants of some complexes (Table 4). The log b values of the MLH
complexes formed in Fe(III)/DQ6 and Fe(III)/DQ726 solutions can-
not be obtained by potentiometric titrations because these species
are already formed at very acidic pH values (Figs. 2a and S1a). In
these cases, the UV–Vis technique was necessary to compute this
value, which was kept constant during the elaboration of the
potentiometric data. In the case of Al(III), UV–Vis log b values were
obtained to confirm the potentiometric results. The agreement of
the stability constant values determined by potentiometry and
UV–Vis for AlLH2+ (Table 4) is very good. For AlL2H2+ the difference
is larger and should be attributed to the UV–Vis uncertainty in
measuring Al(III)/ligand absorption spectra. Different Al(III) com-
plexes display very similar UV–Vis spectra as the absorption is
due to the ligand only. Therefore, UV–Vis data regarding Al(III) be-
come less reliable when several species coexist in solution.
3.5. Al(III)/ligand complexes: 1H NMR results
The 1H NMR spectra of each free ligand show only singlets due
to the absence of resolvable H–H coupling constants. For DQ6 at
pD = 1.5, signals were observed at d 2.56 (3H, CH3(6)), 6.98 (1H,
H(5)), and 8.70 (1H, H(2)). For pyridinic ring numbering see
Fig. 1. For DQ726 at pD = 2.6, signals were at d 2.48 (3H, CH3(6)),
2.79 (3H, CH3(2)), and 6.78 (1H, H(5)). Signals of DQ6 and DQ726
moved upfield by increasing the pD value due to the deprotonation
of the pyridinic nitrogen at pD ca. 6 (Table 3): for example, at
pD = 5.8 CH3(6), CH3(2), and H(5) protons of DQ726 resonate at d
2.41, 2.53, and 6.51, respectively. The aromatic proton spectra for
free DQ6 solutions at various pD are shown in Fig. 3a.
In the presence of Al(III), the resonances of the free ligands were
still observed, together with new peaks due to the complex forma-
tion at all the investigated pD values. Fig. 3b displays the spectra
(aromatic proton region only) of Al(III)/DQ6 solutions.
At pD = 1.5 one new set of narrow signals appeared, which be-
longs clearly to a single species. On the basis of the potentiometric
data this complex is AlLH. The H(5) signal of AlLH is well resolved
from the corresponding signal of the free ligand.
At pD = 2.8 new signals appear together with those of the free
ligand and those of AlLH. The H(5) peaks are again the most inter-
esting and they show the sharp peak of the free ligand (d 6.93), that
of AlLH (d 6.85), a broad band centred at d 6.77, and four small
peaks at d 6.64–6.72 (see inset in Fig. 3b). According to the
Fig. 2. Distribution diagrams of the most important Fe(III) (a) and Al(III) (b) species
in the presence of DQ6 in aqueous (Na)Cl 0.6 m, T = 25 °C; CM = 5 Â 10À4 m,
CL = 2 Â 10À3 m. Dashed lines show the theoretical starting pH for hydroxide
precipitation.

184
A. Dean et al. / Inorganica Chimica Acta 373 (2011) 179–186
Fig. 3. 1H NMR spectra (aromatic zone) at various pD values of D2O solutions containing: (a) only DQ6 (CDQ6 = 8.9 Â 10À3 m); (b) Al(III) (CAl = 3.0 Â 10À3 m) and DQ6
(CDQ6 = 8.9 Â 10À3 m).
potentiometric results, AlL2H2 is the main species at this pD, so the
broad band should be assigned to AlL2H2 and the four peaks to
AlL3H3 (see below). The broadness of the AlL2H2 signal can be due
both to slight differences in the chemical shifts of several isomers
(up to eight diastereoisomers can exist in solution for this complex),
and to the interconversions of the diastereoisomers at a rate
comparable with the NMR time scale. Broad NMR signals for Al:L
1:2 species are commonly observed for Al(III)/HPs complexes
[1,3,4], and slow kinetics of interconversion was confirmed [3,6].
At pD 5.9 the signals of AlLH and of the free ligand become very
small. Together with the signals of AlL2H2, four new signals appear
for H(5) at d 6.65–6.73 (see inset in Fig. 3b), which can be assigned
to AlL3H3. DQ6 is asymmetric, thus it may chelate the metal, form-
ing AlL3H3, in two different spatial configurations. If the three li-
gands of AlL3H3 chelate the metal ion in the same configurations,
a symmetrical diastereoisomer forms, and the three protons are
chemically equivalent. If one ligand chelates differently from the
other two, an unsymmetrical diastereoisomer forms, and the three
protons should give rise to different signals in the NMR spectrum.
As a consequence, four signal are expected (and observed in this
case) in the spectrum. These signals show the same integral: this
can be explained if the formation microconstant of the symmetric
and of the unsymmetric diastereoisomer are very similar, so that
their ratio in solution is 1:3. A similar 1H NMR pattern was ob-
served for other Al(III)/HP solutions [3,4,6].
The 1H NMR spectra for Al(III)/DQ726 are shown in Fig. S3 (Sup-
plementary material). At pD = 2.6 three sets of signals were ob-
served: one can be assigned to the free ligand, the other two sets
to AlLH (narrow, more intense) and AlL2H2 (broad, low intensity).
At pD 4.2 the narrow signals of AlLH vanish, and a number of
new broad signals appear. In the aromatic zone, for example, four
separated signals were observed at d 6.42, 6.50, 6.55, and 6.65 to-
gether with the peak of the free ligand. In agreement with previous
findings, these new signals are due to different structural isomers
of AlL2H2 and AlL3H3. The pattern changes when pD increases from
4.2 to 5.8, allowing the provisional assignment of some signals to
AlL2H2 (which predominates at pD 4.2) and the others to AlL3H3
(which predominates at pD 5.8).
Semiquantitative data can be obtained from NMR spectra by
calculating the relative integrals of the signals of Al(III)/DQ6 and
Al(III)/DQ726 complexes and those of the free ligand. These values
can be compared with those calculated on the basis of the specia-
tion data reported in Tables 3 and 4 computed at the same Al(III)
and ligand concentrations and at the same pH values. Table 5 sum-
marises the results for Al(III)/DQ6, where the relative amounts of
AlLH, AlL2H2 and AlL3H3 were computed at most pD values. Differ-
ences can be explained not only by the uncertainty of the NMR
integration values and the broadness of some signals, but mainly
by the presence of isotopic effects due to the use of D2O as NMR
solvent. The conditional stability constants of the complexes in
D2O are lower than in H2O, because the deuterated ligand is less
acidic than the protonated one (primary isotopic effect), i.e. D+ is
a better metal competitor for the ligand than H+ [28]. A significant
At pD 8.7 the signal sets shift upfield, in agreement with the
deprotonation detected for the complexes at this pD by potentio-
metric titrations.

A. Dean et al. / Inorganica Chimica Acta 373 (2011) 179–186
185
Table 5
Relative amounts of free DQ6 and Al(III)/DQ6 complexes (with respect to total ligand
concentration) at various pD values according to NMR (integration values of H(5)) and
theoretical speciation. At pD 5.9, the notation ‘‘AlL2Hx’’ denotes AlL2H2 + AlL2H + AlL2.
pD
values
Relative amounts according
to NMR data
Relative amounts according to
theoretical speciation
1.5
2.8
Free ligand: 72%
AlLH: 28%
AlL2H2: 0%
Free ligand: 53%
AlLH: 15%
Free ligand: 70%
AlLH: 21%
AlL2H2: 9%
Free ligand: 37%
AlLH: 8%
AlL2H2: 29%
AlL2H2: 41%
AlL3H3: 3%
AlL3H3: 14%
5.9
8.7
Free ligand: 8%
AlLH: 4%
AlL2Hx: 28%
AlL3H3: 60%
Free ligand: 14%
Complexed ligand: 86%
Free ligand: 2%
AlLH: 0%
AlL2Hx: 7%
AlL3H3: 91%
Free ligand: 53%
3AlL3H3 + 2AlL2Hx: 47%
larger amount of free ligand is expected in D2O than in H2O. For
Al(III)/DQ726 complexes, only the relative amount of total free li-
gand was computed, and it was 72%, 45%, and 30% at pD = 2.6,
4.2, and 6.8, respectively. Corresponding theoretic values are 69%,
36%, and 21%. Here, too, the differences can be explained by the
primary isotopic effect.
3.6. Metal/ligand complexes: discussion
In order to evaluate the effects of the 2- and 6-methyl substitu-
tions to DQ0, pM versus pH (pM = Àlog[M]) was computed for
DQ716, DQ726, DQ6, DQ2, and DQ0 (Fig. 4): higher pM values indi-
cate stronger metal/ligand complexes. Computations were per-
formed at metal and ligand stoichiometric concentrations [29]
which are representative of physiologic conditions (10À6 m metal
ion and 5 Â 10À5 m ligand). The pFe and pAl plots for L1 at the
same conditions (speciation data from Clarke and Martell [30])
are reported for comparison. Fig. 4 demonstrates that HPs are
weaker Fe(III) and Al(III) chelators than L1. It is also important to
note that Al(III)/DQ6 complexes are as strong as those formed by
DQ716 (this derivative forms the most stable complexes among
the HPs synthesised so far [6]). However, Fe(III)/DQ6 complexes
are ca. 1 order of magnitude less stable than Fe(III)/DQ716 com-
plexes. DQ726 and DQ2 have practically the same affinity, and they
form the weakest Fe(III) and Al(III) complexes among all HPs.
The high stability of Al(III)/DQ6 complexes confirms that the 6-
methyl substitution of the pyridinic ring increases the Lewis basi-
city of DQ0. This can be explained by the electron-donating effect
of the methyl group, which has a positive effect on the HP’s affinity
towards hard metal ions. However, Fe(III)/DQ6 complexes are less
stable than Fe(III)/DQ716 complexes, indicating that the 6-methyl
effect is less important for Fe(III) than for Al(III). For Fe(III), the
maximum positive effect on the metal stability of the ligands ap-
pears to be due to the 1-methyl substitution.
Fig. 4. pFe (a) and pAl (b) plots for several ligands; Cmetal = 10À6 m,
Cligand = 5 Â 10À5 m. pFe plots stop at the theoretical starting pH for hydroxide
precipitation.
pH 7.4. The two sets for DQ6 data gave Do/w = 0.030 0.003 and
0.021 0.002, the two for DQ726 gave Do/w = 0.0024 0.0012 and
0.0014 0.0003. Both ligands are markedly hydrophilic at physio-
logic pH. Metal/ligand complexes retain hydrophilicity: Do/
w = 0.022 0.002 for Fe(III)/DQ6, 0.015 0.001 for Al(III)/DQ6,
0.0032 0.0011 for Fe(III)/DQ726, 0.0019 0.006 for Al(III)/
DQ726. In the presence of the Al ion, DQ6 is significantly more
hydrophilic than in the absence of it (unpaired two-tailed t-test):
this is a favourable property for metal chelators as they should en-
hance metal renal clearance.
The chelation efficiency measurements showed different behav-
iours of the two ligands. DQ6 chelated a significant amount of
Fe(III) (20% 3%) and especially of Al(III) (47% 7%) at pH 7.4.
The chelation efficiency of DQ726 was negligible for both metal
ions (efficiency < 1%). In fact, Do/w values of DQ726 in the absence
and in the presence of metal ion are statistically equivalent. pM
values calculated from thermodynamic data (Fig. 4) are in agree-
ment with these chelation efficiency results.
The stability decrease of the Fe(III) and Al(III) complexes due to
the 2-methyl substitution on DQ0 was observed previously for DQ2
[6], and confirmed here for DQ726. The most probable reason is an
ortho effect: the steric hindrance of the methyl group in the 2-
position does not allow the bulky carboxyl group in 3 to be copla-
nar with the pyridine ring. This causes a distortion of the chelate
ring and thus a decrease of the complex stability.
3.8. In vitro cytotoxic activity
Analyzing the percentage of living cells with respect to cells ex-
posed following 3 days exposure, it was apparent that DQ6 or
DQ726, when used at concentrations up to 0.1 mM, shows no cyto-
toxic activity on human cancer cell lines (OVCAR, OE33, A549, and
HeLa) and also on primary keratinocytes and fibroblasts. Concen-
trations higher than 0.1 mM could not be investigated because of
the toxic effects due to the pH variation in the culture medium
3.7. n-Octanol/water distribution and chelation efficacy
The octanol/water distribution coefficients (Do/w) of free DQ6
and DQ726 were measured in both the Al and Fe distribution sys-
tems (each by four replicate trials with duplicate observations) at

186
A. Dean et al. / Inorganica Chimica Acta 373 (2011) 179–186
(produced by solvent additions). Therefore no IC50 value (concen-
tration of chemical resulting in 50% inhibition of cell viability)
could be calculated. The IC50 was larger than 0.1 mM for all the
investigated compounds.
Appendix A. Supplementary material
Supplementary data associated with this article can be found, in
the online version, at doi:10.1016/j.ica.2011.04.009.
References
4. Conclusions
[1] V.B. Di Marco, R.A. Yokel, M.G. Ferlin, A. Tapparo, G.G. Bombi, Eur. J. Inorg.
Chem. (2002) 2648.
[2] V.B. Di Marco, R.A. Yokel, H. Li, A. Tapparo, G.G. Bombi, Inorg. Chim. Acta 357
(2004) 3753.
[3] A. Dean, M.G. Ferlin, P. Brun, I. Castagliuolo, D. Badocco, P. Pastore, A. Venzo,
G.G. Bombi, V.B. Di Marco, Dalton Trans. (2008) 1689.
[4] V.B. Di Marco, A. Dean, M.G. Ferlin, R.A. Yokel, H. Li, A. Venzo, G.G. Bombi, Eur. J.
Inorg. Chem. (2006) 1284.
[5] V.B. Di Marco, A. Dean, R.A. Yokel, H. Li, G.G. Bombi, Polyhedron 26 (2007)
3227.
[6] A. Dean, M.G. Ferlin, P. Brun, I. Castagliuolo, R.A. Yokel, D. Badocco, P. Pastore,
A. Venzo, G.G. Bombi, V.B. Di Marco, Dalton Trans. (2009) 1815.
[7] R.C. Hider, T. Zhou, Ann. NY Acad. Sci. 1054 (2005) 141.
[8] É. Sija, A. Dean, T. Jakusch, V.B. Di Marco, A. Venzo, T. Kiss, Monatsh. Chem. 142
(2011) 399.
[9] Z.D. Liu, R.C. Hider, Med. Res. Rev. 22 (2002) 26.
[10] V.B. Di Marco, A. Tapparo, A. Dolmella, G.G. Bombi, Inorg. Chim. Acta 357
(2004) 135.
[11] H. Tendeloo, F. Broekman, V. Stroer, J. Siemelink, Neth. Appl. 70 (1972) 11749.
[12] W.O. Foye, J.M. Kauffman, Chim. Ther. 2 (1967) 462.
[13] H.G. Kraft, L. Fiebig, R. Hotovy, Arzneimittelforschung (Drug Res.) 11 (1961)
922.
[14] H. Yamada, H. Tobiki, K. Jimpo, K. Gooda, Y. Takeuchi, S. Ueda, T. Komatsu, T.
Okuda, H. Noguchi, K. Irie, T. Nakagome, J. Antibiot. 36 (1983) 532.
[15] M. Sato, H. Ogasawara, K. Kato, M. Sakai, T. Kato, Chem. Pharm. Bull. 31 (1983)
4300.
[16] V.B. Di Marco, A. Tapparo, G.G. Bombi, Ann. Chim. (Rome) 89 (1999) 397.
[17] V.B. Di Marco, A. Tapparo, G.G. Bombi, Ann. Chim. (Rome) 91 (2001) 595.
[18] G.J. Janz, Silver–silver halide electrodes, in: D.J.G. Ives, G.J. Janz (Eds.),
Reference Electrodes: Theory and Practice, Academic Press, New York,
London, 1961.
[19] G. Biedermann, L.G. Sillén, Ark. Kemi 5 (1953) 425.
[20] V.B. Di Marco, A. Tapparo, G.G. Bombi, Ann. Chim. (Rome) 89 (1999) 535.
[21] V.B. Di Marco, Ph.D. Thesis, University of Padova, 1998.
[22] L.G. Sillén, Acta Chem. Scand. 18 (1964) 1085.
Both DQ6 and DQ726 form strong Fe(III) and Al(III) complexes
in aqueous solution having 1:1, 1:2, and 1:3 metal:ligand stoichi-
ometry. The free ligands and their metal complexes are hydrophilic
at physiologic pH. None of the ligands was cytotoxic in the inves-
tigated concentration range (IC50 > 0.1 mM). Both ligands form
weaker Fe(III) and Al(III) complexes than L1: this means, in princi-
ple, that a similar metal depletion efficiency can be obtained if
higher doses of ligand are administered in the chelation therapy.
This should not be a problem if the absence of toxicity in vitro dis-
played by all HPs will be confirmed in vivo.
The DQ726 complexes with Fe(III) and Al(III) are weaker than
those formed by the other HPs, as confirmed by the very low che-
lation efficiencies measured for metal/DQ726 solutions at physio-
logic conditions simulated in vitro. The coordination strength is
very similar to that of DQ2. The decrease of the metal ion binding
ability is attributed to an ortho effect. It is likely that a similar ef-
fect would be observed with any functional group in position 2
other than the methyl. Therefore, any substitution of DQ0 at the
position 2 should be avoided to let the ligand be effective as a me-
tal ion chelator in vivo.
The Fe(III)/DQ6 complexes are more stable than those of other
HPs, but less stable than those of DQ716. On the other hand, the
Al(III)/DQ6 complexes have practically the same stability as those
of DQ716. The substitution at position 1 appears to be much more
important in enhancing the Fe(III) coordination strength, the sub-
stitution at 6 appears to be more important for Al(III). The 1-substi-
tuted derivatives of DQ0 are good candidates as chelating agents
for Fe. The 6-substituted derivatives of DQ0 can represent a family
of ligands which are relatively more effective for Al(III) than for
Fe(III), i.e. these chelators would not deplete Fe during Al chelation
therapy. DQ6 itself can be proposed as a chelating agent for Al.
[23] W.H. Press, B.P. Flannery, S.A. Teukolsky, W.T. Vetterling, Numerical Recipes:
The Art of Scientific Computing, Cambridge University Press, Cambridge, 1986.
[24] C.M. Flynn Jr., Chem. Rev. 84 (1984) 31.
[25] L.O. Öhman, Inorg. Chem. 27 (1988) 2565.
[26] A.K. Covington, M. Paabo, R.A. Robinson, R.G. Bates, Anal. Chem. 40 (1968) 700.
[27] R.A. Yokel, H.B. Kostenbauder, Toxicol. Appl. Pharmacol. 91 (1987) 281.
[28] C.A. Bunton, V.J. Shiner Jr., J. Am. Chem. Soc. 83 (1961) 42.
[29] T. Kiss, E. Farkas, Perspect. Bioinorg. Chem. 3 (1996) 199.
[30] E.T. Clarke, A.E. Martell, Inorg. Chim. Acta 191 (1992) 57.
Acknowledgements
This work was supported by University of Padova, ‘‘Progetto
Ateneo 2008’’. We thank Sara Simonato and Daniele Lena (Univer-
sity of Padova) for their excellent work.