Rational design, synthesis, and evaluation of tetrahydroxamic acid chelators for stable complexation of zirconium(IV)

Article abstract of DOI:10.1002/chem.201304115

Metals of interest for biomedical applications often need to be complexed and associated in a stable manner with a targeting agent before use. Whereas the fundamentals of most transition-metal complexation processes have been thoroughly studied, the compl

Full text of DOI:10.1002/chem.201304115

DOI: 10.1002/chem.201304115
Full Paper
&
Imaging Agents
Rational Design, Synthesis, and Evaluation of Tetrahydroxamic
Acid Chelators for Stable Complexation of Zirconium(IV)
FranÅois Guꢀrard,*[a] Yong-Sok Lee,[b] and Martin W. Brechbiel[a]
Chem. Eur. J. 2014, 20, 5584 – 5591
5584
ꢁ 2014 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim

Full Paper
Abstract: Metals of interest for biomedical applications
often need to be complexed and associated in a stable
manner with a targeting agent before use. Whereas the fun-
damentals of most transition-metal complexation processes
have been thoroughly studied, the complexation of ZrIV has
been somewhat neglected. This metal has received growing
attention in recent years, especially in nuclear medicine,
with the use of 89Zr, which a b+-emitter with near ideal char-
acteristics for cancer imaging. However, the best chelating
agent known for this radionuclide is the trishydroxamate
desferrioxamine B (DFB), the ZrIV complex of which exhibits
suboptimal stability, resulting in the progressive release of
89Zr in vivo. Based on a recent report demonstrating the
higher thermodynamic stability of the tetrahydroxamate
complexes of ZrIV compared with the trishydroxamate com-
plexes analogues to DFB, we designed a series of tetrahy-
droxamic acids of varying geometries for improved complex-
ation of this metal. Three macrocycles differing in their
cavity size (28 to 36-membered rings) were synthesized by
using a ring-closing metathesis strategy, as well as their acy-
clic analogues. A solution study with 89Zr showed the com-
plexation to be more effective with increasing cavity size.
Evaluation of the kinetic inertness of these new complexes
in ethylenediaminetetraacetic acid (EDTA) solution showed
significantly improved stabilities of the larger chelates com-
pared with 89Zr-DFB, whereas the smaller complexes suffered
from insufficient stabilities. These results were rationalized
by a quantum chemical study. The lower stability of the
smaller chelates was attributed to ring strain, whereas the
better stability of the larger cyclic complexes was explained
by the macrocyclic effect and by the structural rigidity. Over-
all, these new chelating agents open new perspectives for
the safe and efficient use of 89Zr in nuclear imaging, with the
best chelators providing dramatically improved stabilities
compared with the reference DFB.
Introduction
Positron emitted tomography (PET) is a powerful imaging tech-
nique that is used for the detection of various pathologies, in-
cluding cancers. 18F (t1/2 =1.8 h) is currently the most common-
ly used b+-emitter in nuclear imaging, and progress is continu-
ously being reported for the chemistry and applications of this
radionuclide.[1] In parallel, a growing number of investigations
on the chemistry of longer-lived radionuclides such as 64Cu (t1/
2 =12.7 h), 86Y (t1/2 =14.7 h), and 124I (t1/2 =100.2 h) is ongoing,
expanding the possible applications of PET imaging.[2] Among
these promising b+-emitters, 89Zr (t1/2 =78.4 h) is particularly in-
teresting when associated with higher molecular weight tar-
geting agents such as cancer targeting antibodies.[3] Indeed, its
half-life ideally matches the blood kinetics of full antibodies,
providing highly sensitive imaging of tumors for several days
after injection.
Figure 1. Structure of DFB and schematic representation of the tetrahy-
droxamic acid chelators C5, C6, and C7 considered in this study, with each
hydroxamic acid linked by a chain of five, six or seven methylene groups,
respectively.
through a conjugated trishydroxamate-based chelating agent,
the natural siderophore desferrioxamine B (DFB) (Figure 1).[5,6]
Although DFB has been a useful tool with which to demon-
strate the high potential of 89Zr for PET-applications,[7] many
animal studies have shown this chelator to be far from optimal
to adequately sequester ZrIV for use in vivo, forming a complex
that lacks kinetic inertness. This instability is visible in the pro-
gressive uptake of 89Zr in the bone as the metallic radionuclide
is released from DFB over a few days.[8] Possible consequences
of this instability are toxicity to the bone marrow with poten-
tially high doses of radioactivity that can be deposited due to
the relatively long half-life of 89Zr associated with its 909 keV g-
emission, as well as the loss of image resolution due to lower
signal/noise ratio, especially when the tumors are in the vicini-
ty of bones. This instability issue stems from the lack of chela-
tion chemistry studies directed specifically toward ZrIV, leaving
an important area of research ill-defined compared with other
transition metals. For instance, a review of recent literature on
89Zr labeling of proteins shows that studies focused mostly on
the conjugation moiety and not on the ligand itself.[6,9,10] Only
Hydroxamic acids are bidentate ligands that are known to
bind strongly to hard cations such as Zr4+ through their
oxygen atoms,[4] and 89Zr is usually bound to antibodies
[a] Dr. F. Guꢀrard, Dr. M. W. Brechbiel
Radioimmune & Inorganic Chemistry Section
Radiation Oncology Branch
National Cancer Institute
National Institutes of Health
Bethesda, Maryland 20892 (USA)
Fax: (+1) 301-402-1923
E-mail: francois.guerard@univ-nantes.fr
[b] Dr. Y.-S. Lee
Center for Molecular Modeling
Division of Computational Bioscience
Center for Information Technology
National Institutes of Health
Bethesda, Maryland 20892 (USA)
Supporting information for this article is available on the WWW under
http://dx.doi.org/10.1002/chem.201304115.
Chem. Eur. J. 2014, 20, 5584 – 5591
www.chemeurj.org
5585
ꢁ 2014 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim

Full Paper
one article reports the use of alternative chelators with a series
of phosphonate-based ligands, however, results showed that
these new chelators were less efficient than DFB for 89Zr com-
plexation.[11]
alkyl chains and bearing an alkene group at both ends for
macrocyclization by ring-closing metathesis (RCM). First, O-
benzyl protected hydroxylamine building blocks 2a–c and 3a–
c were prepared by N-alkylation of N-Boc-O-benzylhydroxyla-
mine with bromoalkyl esters of suitable chain lengths to form
compounds 1a–c, followed by Boc removal or ester hydrolysis
(Scheme 1). As depicted in Scheme 2, the O-benzylated tetra-
To increase the stability of the complexation of ZrIV, our ap-
proach was to improve rationally the characteristics of DFB to
fit the requirements of the cation. The preference of ZrIV to
form octacoordinated complexes with four hydroxa-
mate ligands was confirmed only recently,[12] suggest-
ing that the chelating properties of DFB, which con-
sists of only three hydroxamate subunits, could be
improved by the inclusion of a fourth hydroxamate
in a suitably preorganized chelator. Whereas ade-
quate preorganization is essential for the formation
of thermodynamically stable chelates,[13] macrocyclic
structures can also favor higher kinetic inertness of
their metallic complex, a point clearly illustrated in
Liu and Edwards’ review.[14] This factor is crucial when
considering in vivo applications because the com-
plexes, which are highly diluted in their use, are ex-
posed to endogenous competitive cations and natu-
Scheme 1. Preparation of the O-benzyl-protected hydroxylamine building blocks
2a–c and 3a–c (with 1a: n=5, 1b: n=6, and 1c: n=7). Bn=benzyl; Boc=tert-butoxy-
carbonyl; TFA=trifluoroacetic acid.
ral chelators that may challenge the stability of the metal che-
late.
hydroxamates 8a–c were obtained by appropriate sequence of
conjugation via mixed anhydride intermediates formed by
using ethyl chloroformate, as reported previously,[15] and ester
or Boc deprotection. Macrocyclization was performed by using
the 2nd generation Grubbs catalyst in CH2Cl2. A minimum of 0.3
equivalent of catalyst was necessary to ensure complete con-
version, and a maximum diene concentration of 1.2 mm was
found to be optimal to avoid detectable formation of dimers
(from the reaction mixture analyzed by mass spectrometry).
After purification, the protected macrocyclic tetrahydroxamates
9a–c were obtained as Z/E mixtures with yields of 52–58%
(Scheme 3). Finally, hydrogenation allowed the simultaneous
reduction of the resulting alkene and benzyl removal, affording
the macrocyclic chelators C5, C6, and C7. In parallel, com-
pounds 8a–c were directly hydrogenated to form the open
chain tetrahydroxamic acid analogues L5, L6, and L7; their
evaluation being essential to assess the impact of the macrocy-
clic effect on the stabilization of the ZrIV complexes. Detailed
synthetic procedures are available in the Supporting Informa-
tion.
For these reasons, we focused on the development of a mac-
rocyclic tetrahydroxamic acid ligand of appropriate geometry
for ZrIV. To probe the optimal ring size, three macrocycles com-
posed of 28, 32, and 36 atoms were designed, all of them sym-
metrical, with a 5-, 6- and 7-carbon alkyl chain between each
hydroxamic acid; the cyclic compounds are denoted C5, C6,
and C7, respectively (Figure 1). These ring sizes were deemed
to be in the appropriate range to preserve the spatial orienta-
tion of the four hydroxamate ligands as observed in the X-ray
structure of tetrakis[N-(hydroxyl)-N-methylacetamidato]zirco-
nium(IV).[12] Accordingly, a comparison of these chelators was
undertaken to establish an optimal ring size that imparted
both thermodynamic stability and kinetic inertness to the ZrIV
complexes.
Results and Discussion
Preparation of tetrahydroxamic acid chelators
Little data is available on the aqueous stability or structural
characterization of chelators complexed to ZrIV. For instance,
even in the case of the Zr-DFB complex widely used in nuclear
The general synthetic approach was based on the formation of
linear chains of four O-benzylated hydroxamates connected by
Scheme 2. Preparation of the O-benzyl tetrahydroxamate precursors (NMM=N-methylmorpholine).
Chem. Eur. J. 2014, 20, 5584 – 5591 www.chemeurj.org
5586
ꢁ 2014 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim

Full Paper
Scheme 3. Formation of the cyclic and acyclic tetrahydroxamic acid chela-
tors.
medicine, no thermodynamic or crystallographic data is avail-
able. The difficulty in obtaining such data is often related to
solubility issues, and the complexes of the present study were
no exception. When preparation of the complexes was at-
tempted by ligand exchange from ZrIV acetylacetonate with
the newly synthesized chelators, it was not possible to dissolve
the product of the reaction after isolation, precluding solution
analysis or crystallization. Nonetheless, mass spectrometry anal-
yses of the reaction solutions before isolation of the complexes
were performed. It was thereby possible to confirm the forma-
tion of the expected ZrIV complexes with C7, L7, and L6. The
inability to detect complexes with C6, C5, and L5 may be relat-
ed to the lower stability of these smaller complexes under the
conditions of the analysis. However, an FTIR study showed the
disappearance of NO-H stretching of the free ligands at ca.
3170 cmꢀ1 upon complexation as well as a redshift of the car-
bonyl stretching of the free ligands from ca. 1600 to ca.
1585 cmꢀ1 in all the complexes (see for example, C7 in
Figure 2). Additionally, ICP-EOS analysis of the isolated com-
plexes confirmed the expected ratio of zirconium in the com-
plexes.
Figure 2. FTIR spectra of free ligand C7 and its ZrIV complex.
by chromatography, with DFB included for comparison. DFB
appeared to be the most rapid for complexing 89Zr, with
a quantitative radiochemical yield (RCY) after 30 min at 208C.
Although exhibiting slightly slower kinetics, the larger ligands
C7, L7, and L6 demonstrated excellent complexation abilities
(more than 99% complexation after 120 min at 208C). In con-
trast, the smaller ligands appeared to be less suitable with de-
creasing cavity size, with higher temperatures required to
obtain high complexation yields, and the smaller macrocycle
C5 reaching only 29% complexation even at 808C (Table 1). In
Due to the high lipophilicity of the Zr complexes, precipita-
tion issues prevented determination of the stability constants
in aqueous solution or water/organic mixtures by using con-
ventional methodologies such as potentiometry or spectro-
photometry. Consequently, the comparison of their relative sta-
bility was performed at trace concentrations with 89Zr as well
as with quantum chemical calculations, as discussed in the
next sections.
Table 1. Influence of time and temperature on the complexation yield of
89Zr by L5, L6, L7, C5, C6, and C7, and DFB.
Ligand
208C
508C
808C
30 min
Yield [%]
120 min
Yield [%]
30 min
Yield [%]
Complexation of 89Zr and stability studies
L5
C5
L6
C6
L7
C7
DFB
60
17
82
77
88
71
22
95
73
23
96
87
29
>99
92
>99
>99
>99
To assess the complexation ability of these new ligands at
trace concentration of cation, a radiolabeling study was per-
formed with 89Zr in aqueous solution. For this, the ligands
were incubated with [89Zr] ZrIV oxalate at pH 7, at various tem-
peratures for a range of times, and the samples were analyzed
79
92
>99
>99
>99
>99
>99
>99
92
>99
Chem. Eur. J. 2014, 20, 5584 – 5591
www.chemeurj.org
5587
ꢁ 2014 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim

Full Paper
the case of the smaller ligands, the actual RCY may be under-
estimated because of possible decomplexation during the con-
ditions of analysis, which involved chromatography in the pres-
ence of 50 mm ethylenediaminetetraacetic acid (EDTA) solution
as eluent. In this system, EDTA is present in large excess and
complexes rapidly the free 89Zr, forming a complex that is sep-
arated from the other 89Zr chelates. However, transchelation
can also occur with weak complexes during the elution, result-
ing in an underestimated RCY (see details in the Experimental
Section). Overall, L6, L7, and C7 appear to be the best ligands
in the context of radiolabeling proteins, with nearly quantita-
tive complexation yields being obtained within 2 h at 208C.
The lower kinetics of complexation observed with the tetrahy-
droxamate ligands compared with DFB would probably be ap-
propriate when considering labeling of proteins given the rela-
tively long half-life of 89Zr, which permits a convenient time-
scale for conducting radiochemical procedures.
mate ligands (ca. 1750 fold excess of EDTA). This test was
chosen to simulate the transchelation that might occur with
ZrIV in vivo when its complex is used in high dilution and ex-
posed to endogenous natural competing chelators. Over
a week, varying levels of decomplexation were observed. Al-
though fast release of 89Zr occurred with C5 and L5 within the
first minutes of incubation, the release was significantly slower
for the other ligands (Figure 3). Stability was higher with the
The kinetic inertness of these complexes was then assessed
in various media. The mechanisms by which 89Zr is released
from DFB in vivo have, to our knowledge, never been de-
scribed and remain hypothetical; therefore, we chose to assess
the stability of these complexes in various conditions that sim-
ulate environments likely to occur in vivo or during their prep-
aration and storage. When incubated for seven days in phos-
phate buffer (pH 7.4) at 208C, or in human serum at 378C, sim-
ilar to 89Zr-DFB, virtually no release of 89Zr was detected for
any of the complexes (stability more than 99%). However,
when incubation was performed in phosphate buffer at pH 6.5
(a pH value that is representative of many tumoral environ-
ments) and 378C, significant decomplexation was observed for
the smaller ligands L5, C5, and C6, whereas the larger ligands,
including DFB, showed very limited decomplexation, with
more than 97% of complex remaining intact after seven days
(Table 2). Consequently, a more challenging test was necessary
Figure 3. Stability of the 89Zr-complexes in 50 mm EDTA, at pH 7 and 378C,
over seven days (n=4).
larger tetrahydroxamates, and better results were obtained
with the cyclic ligand C7 compared with its linear analogue L7.
The inverse tendency was observed for the L5/C5 and L6/C6
pairs, illustrating a detrimental macrocyclic effect for the small-
er rings, whereas it becomes beneficial for the 36-member ring
C7. Most importantly, greater stabilities than DFB were ob-
tained with two of the new ligands (L7 and C7), with the high-
est kinetic inertness obtained for C7 compared with DFB (87ꢁ
3 vs. 47ꢁ3% of intact complex, respectively, after seven days),
confirming the expected superiority of tetrahydroxamate li-
gands with optimal geometric preorganization.
Table 2. Stability of the 89Zr-complexes in 0.1m sodium phosphate buffer
at pH 6.5, over seven days.
Fraction of intact complex
Ligand
1 day
4 days
7 days
Yield [%]
Yield [%]
Yield [%]
Quantum chemical studies
L5
C5
L6
C6
L7
C7
DFB
93
69
97
89
99
89
46
94
79
98
99
98
77
42
94
61
98
99
98
To provide a structural basis for the observed stabilities, quan-
tum chemical calculations were carried out with density func-
tional theory utilizing the M06L functional[17] with the pseudo-
potential LanL2DZ[18] for the Zr atom and the 6-31+G* basis
set for the rest of the atoms, as implemented in Gaussian 09
software.[19] To this end, we first calculated DG for the complex-
ation reaction of Zr(OH)4 by Me-AHA (N-methyl-acetohydroxa-
mic acid), to give Zr(Me-AHA)4, the X-ray structure and stability
constant in H2O of which (log b=45.98 at 258C corresponding
to a DG of ꢀ62.7 kcalmolꢀ1) was previously determined.[12] The
calculated DG for the complexation [see Eq. (1)] in the reaction
field of water was ꢀ56.1 kcalmolꢀ1, underestimating the exper-
imental DG by 6.6 kcalmolꢀ1.[20] The calculated DH and TDS at
298.15 K were ꢀ68.9 and ꢀ12.8 kcalmolꢀ1, respectively, indicat-
ing that the reaction is entirely enthalpy driven.
>99
99
to observe significant release of 89Zr from the most stable com-
plexes and allow their respective stabilities to be distinguished.
EDTA has been reported to form thermodynamically stable
complexes with ZrIV with Log K=29,[16] but it exhibits insuffi-
cient kinetic inertness in vivo for medical use. For this reason,
a 50 mm EDTA solution at pH 7 was used to incubate the com-
plexes at 378C; this concentration corresponded to a large
excess of competing chelator compared with the tetrahydroxa-
Chem. Eur. J. 2014, 20, 5584 – 5591
www.chemeurj.org
5588
ꢁ 2014 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim

Full Paper
is that the ring-strain energy going from L7 to L6 is only
2.6 kcalmolꢀ1 as compared with the C7 to C6 transition being
8.8 kcalmolꢀ1. Whereas ring opening alleviates the ring strain
of the L6 complex, it does not overcome the ring strain in L5
(14.8 kcalmolꢀ1 for the L6 to L5 transition).
ZrðOHÞ4 þ 4 MeAHA ! ZrðMeAHAÞ4 þ 4 H2O
ð1Þ
Having captured ca. 90% of DG for the formation of Zr(Me-
AHA)4, we used the same approach to further calculate the re-
action energetic of [Eq. (2)] with cyclic and acyclic ligands.
The calculated DG difference between the relative stability
of Zr-C7 over Zr-L7 observed after a week is only 0.4 kcal
molꢀ1, suggesting that this stability difference may be related
to kinetics. As shown in Figure 4, the two open alkyl chains in
ZrðOHÞ4 þ H4L ! ZrL þ 4 H2O
ð2Þ
Cyclic (C5–C7) and acyclic (L5–L7) complexes of ZrIV were
built by using the previously reported X-ray structure of PuIV
complexed with the cyclic trishydroxamate ligand desferriox-
amine E as the starting point.[21] Neither the energy-minimized
ZrIV complexes nor the ligands may represent a global mini-
mum for the system, given the flexibility of each ligand. None-
theless, the calculated DG values in the reaction field of water
(Table 3) were all in good agreement with the relative stabili-
Table 3. Gibbs free energy, enthalpy, and entropy contribution (kcal
molꢀ1) for the complexation of Zr(OH)4 by the tetrahydroxamic acid lig-
ands [Eq. (2)] at 298.15 K calculated at the M06L level with the pseudopo-
tential LanL2DZ for the Zr atom, and the 6-31+G* basis set for the rest
of the atoms[a]
Ligand
DG
DH
TDS
C7
L7
C6
L6
C5
L5
ꢀ66.4 (ꢀ71.0)
ꢀ67.4 (ꢀ70.6)
ꢀ60.0 (ꢀ62.2)
ꢀ66.4 (ꢀ68.0)
ꢀ44.7 (ꢀ49.2)
ꢀ47.5 (ꢀ53.2)
ꢀ42.3 (ꢀ45.2)
ꢀ43.7 (ꢀ47.3)
ꢀ37.1 (ꢀ36.7)
ꢀ43.3 (ꢀ44.9)
ꢀ22.4 (ꢀ23.8)
ꢀ24.1 (ꢀ28.9)
24.1 (25.8)
23.7 (23.3)
22.9 (25.5)
23.1 (23.1)
22.3 (25.4)
23.4 (24.3)
Figure 4. Geometry-optimized cyclic Zr-C7 and acyclic Zr-L7 complexes. The
arrows indicate the two open alkyl chains. Carbon in green, oxygen in red,
nitrogen in blue, and central zirconium in grey (hydrogen atoms not
shown). Average NOꢀZr bond length is 2.169 ꢂ in Zr-L7 and 2.167 ꢂ in Zr-
C7, and the average COꢀZr bond length is 2.289 ꢂ in Zr-L7 and 2.296 ꢂ in
Zr-C7. These calculated bond lengths are comparable to those determined
by X-ray diffraction of the crystalline Zr(Me-AHA)4 (NOꢀZr=2.183 ꢂ and
COꢀZr=2.204 ꢂ). The coordinates of the complexes and ligands discussed
in this paper are available in the Supporting Information.
[a] Numbers in parentheses represent the values calculated in the reac-
tion field of water with the polarizable continuum model (PCM) based on
the gas-phase optimized geometries.
ties of the 89Zr complexes observed in Figure 3. It is also noted
that all cyclic and acyclic complex formations are both enthal-
py and entropy driven. For example, Zr-C7 is 8.8 and 21.8 kcal
molꢀ1 more stable than Zr-C6 and Zr-C5, respectively, in terms
of DG calculated in the reaction field of water. This energy dif-
ference correlates well with the observation that 87ꢁ3% of
89Zr remains complexed to C7 after seven days, whereas only
32ꢁ4% is still complexed to C6, and nearly none by C5. In
terms of kinetics, the release of ca. 50% of the 89Zr from C5
and C6 occurred within the first minutes and a single day, re-
spectively. This suggests that ZrIV complexes exhibiting better
thermodynamic stabilities also have higher energy barriers for
ZrIV release. The stability differences among the three cyclic
complexes are primarily due to ring strain, as manifested by
their DH values; the entropy contribution at 298.15 K for the
three complexes are all comparable (ca. 25 kcalmolꢀ1). The cal-
culated ring-strain energy going from C7 to C6 is 8.8 kcal
molꢀ1, and further increases to 13.0 kcalmolꢀ1 when going
from C6 to C5, suggesting that reduction of the size of the
macrocycle cavity is detrimental to the stability of the resulting
ZrIV complex.
Zr-L7 can fluctuate more than the alkyl chains in the more
rigid Zr-C7. This fluctuation is likely to lead to weakening of
the ZrꢀO bonds, resulting in a lower energy barrier for ZrIV re-
lease from L7. On the other hand, because the L6 complex is
more stable than the C6 analogue by 5.8 kcalmolꢀ1, the ob-
served higher stability of Zr-L6 over Zr-C6 can be attributed to
thermodynamics. Whereas better thermodynamic stability
does not necessarily correspond to higher kinetic inertness,
the present thermochemical calculations indicate that such
a correlation exists in the smaller cyclic and acyclic ligands
complexed with ZrIV studied here. On the other hand, when
similar thermodynamic stabilities are achieved with L7 and C7,
a beneficial macrocyclic effect on the kinetic inertness of C7
was observed.
Overall, these calculations as well as the results of the stabili-
ty studies, illustrate well the fact that the best preorganization
is reached when chains of seven carbon atoms separate the
hydroxamate functions. At this optimal size, kinetic inertness
was significantly enhanced with the macrocycle C7 compared
with its acyclic analogue L7.
The observed stability with the acyclic complexes also corre-
lates well with their calculated DG values. A notable difference
Chem. Eur. J. 2014, 20, 5584 – 5591
www.chemeurj.org
5589
ꢁ 2014 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim

Full Paper
Complex Zr-C6: Yield=82%; not detected by mass spectrometry
(ESI, +); Zr calcd (%) 13.82; found: 14.14.
Conclusion
Complex Zr-L6: Yield=96%; MS (ESI, +): m/z: 689.2 [M+H]+; Zr
calcd (%) 13.22; found: 15.11.
Complex Zr-C7: Yield=78%; MS (ESI, +): m/z: 715.3 [M+H]+; Zr
calcd (%) 12.74; found: 12.59.
The rational design of a chelating agent fitted to the ZrIV
cation was proposed by using a suitable set of ligands with
varying degrees of preorganization. A kinetic inertness evalua-
tion performed under relatively harsh conditions by using
EDTA as a strong competing agent provided a classification of
these ligands and proved that the use of four hydroxamic
acids in a single chelator can significantly increase the kinetic
inertness of the resulting complex, which is essential for in
vivo applications of radionuclides. The lack of stability data in
aqueous solutions inherent to this kind of complex of low sol-
ubility was circumvented by conducting quantum chemical
calculations using the crystallographic structures of ZrIV and
PuIV complexed with hydroxamates as a template for the con-
struction of the present ZrIV complexes. The relative stabilities
of the investigated complexes were rationalized in terms of
structures and energetics, and the good agreement seen in
the present work indicate that quantum chemical calculations
can provide an insight into the design of metal–ligand com-
plexes of interest, particularly when aqueous solution studies
are not applicable.
89Zr Production
89Zr was produced and purified at the National Institutes of Health,
Bethesda, MD, USA, by the following procedure: Pressed pellets of
yttrium metal (200 mg, 99.99% purity; American Elements, USA)
were irradiated with a proton beam of 15 MeV and a current of
20 mA for 2–4 h with a GE PET trace cyclotron. 89Zr was separated
from the yttrium target material by the use of hydroxamate resin
as described by Holland et al.[22] Briefly, the target material was dis-
solved in fractions of 6m HCl (4ꢃ0.5 mL). After 1 h, the undis-
solved solid residue was separated by filtration, the resulting solu-
tion was diluted to 5 mL with deionized water and then loaded
onto the hydroxamate resin column. The column was then washed
with 2m HCl (4ꢃ2.5 mL) and deionized water (4ꢃ2.5 mL). After the
solution was removed from the column, the 89Zr was eluted with
successive portions of 1m oxalic acid. The first 0.4 mL fraction was
discarded and the next 0.7 mL fraction was collected for further
use.
Finally, to our knowledge, this is the first time complexes of
ZrIV have been reported that exhibit significantly higher stabili-
ties than the reference DFB used in nuclear medicine for the
complexation of 89Zr. Modification of the most promising lig-
ands of this study for conjugation to antibodies and in vivo ap-
plications are in progress, especially with L7 and C7, which ex-
hibited the best stabilities under all the conditions tested, as
well as efficient complexation abilities of 89Zr at 208C.
Complexation studies with 89Zr
All solutions described below were prepared with deionized water
purified through a chelex column prior to use. Stock solutions of
89Zr at pH 7 were prepared as follows: To the 89Zr solution in oxalic
acid (450 mL) was added 1m Na2CO3 solution (450 mL) and the pH
was adjusted to pH 7 by addition of small aliquots of 0.1m Na2CO3
or 0.1m HCl. Water (450 mL) containing 3% bovine serum albumin
was then added. To 45 mL of stock solution (ca. 1.3 MBq of 89Zr in
a typical experiment) were added 7.5 nmol (5 mL) L5, L6, L7, C5,
C6 or C7 in DMSO or desferrioxamine B (DFB) mesylate in deion-
ized water. These solutions were incubated at the desired tempera-
tures and durations and analyzed by ITLC-SG using a 50 mm EDTA
solution adjusted to pH 7 in deionized water as eluent. Analyses of
the TLCs were performed with a Typhoon 8600 scanner (GE Health-
care) in phosphorimaging mode. The percentage of the activity
bound to the ligand after TLC analysis was calculated by convert-
ing the TLC scan into a chromatogram and integrating the peak
corresponding to the spot at the bottom of the TLC. Activity ratios
were averaged out of a minimum of two TLC plates for each condi-
tion. A selection of representative TLC plates is provided in the
Supporting Information. For the stability studies described below,
higher activities were used, with ca. 50 MBq 89Zr reacted with
38 nmol of ligand. The specific activity achieved was 0.96 MBq/
nmol with L5, 0.30 MBq/nmol with C5, 1.26 MBq/nmol with L6,
1.21 MBq/nmol with C6, and 1.31 MBq/nmol with L7, C7, and DFB.
Experimental Section
Preparation of cold ZrIV complexes
Typical procedure with Zr-L7: Ligand L7 (113 mg, 171 mmol) dis-
solved in isopropyl alcohol (10 mL) was added to zirconium(IV) ace-
tylacetonate (81 mg, 162 mmol) dissolved in anhydrous methanol
(10 mL) under a nitrogen atmosphere. The solution was heated for
14 h at 708C, resulting in a milky solution with the complex partial-
ly precipitated. Analysis of the solution by mass spectrometry (ESI,
+) confirmed the formation of a 1:1 metal/ligand complex (m/z:
745.3 [M+H]+ and Zr isotopic distribution peaks). The solution
was evaporated, resulting in the formation of a pale-yellow
powder. Residual zirconium(IV) acetylacetonate and ligand were re-
moved by heating to reflux and filtering the solid successively in
methanol and isopropyl alcohol. After drying in vacuum, a pale-
yellow solid (91 mg, 76%) was obtained. A sample was analyzed
by ICP-OES for Zr content (Zr calcd (%) 12.23; found: 13.29; experi-
mental error limits of the analysis: 10%).
All complexes were prepared by using the same procedure. Only
L6, L7, and C7 formed a complex that could be detected by mass
spectrometry (ESI, +). All complexes had the expected Zr content
within the error range of the ICP-OES analysis method, except for
Zr-L6.
In vitro stability of the [89Zr]-zirconium(IV) complexes
Each 89Zr complex solution (10 mL), prepared as described above,
were incubated for seven days in phosphate-buffered saline (40 mL,
pH 7.4, 10% DMSO) at 208C, sodium phosphate buffer (40 mL,
0.1m, pH 6.5, 10% DMSO) at 378C, human serum (40 mL) at 378C,
or 50 mm EDTA (40 mL, pH 7, 10% DMSO) at 378C, respectively,
and analyzed by using the chromatographic system described in
the above section after 1 h, 3 h, 1, 2, 3, 4, 5, 6, and 7 days.
Complex Zr-C5: Yield=92%; not detected in MS (ESI, +); Zr calcd
(%) 15.11; found: 14.7.
Complex Zr-L5: Yield=94%; not detected by mass spectrometry
(ESI, +); Zr calcd (%) 14.39; found: 13.38.
Chem. Eur. J. 2014, 20, 5584 – 5591
www.chemeurj.org
5590
ꢁ 2014 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim

Full Paper
[4] R. Codd, Coord. Chem. Rev. 2008, 252, 1387–1408.
Quantum chemical studies
[5] M. J. W. D. Vosjan, L. R. Perk, G. W. M. Visser, M. Budde, P. Jurek, G. E.
Kiefer, G. A. M. S. van Dongen, Nat. Protoc. 2010, 5, 739–743.
[6] B. M. Zeglis, P. Mohindra, G. I. Weissmann, V. Divilov, S. A. Hilderbrand, R.
Weissleder, J. S. Lewis, Bioconjugate Chem. 2011, 22, 2048–2059.
[7] a) M. A. Deri, B. M. Zeglis, L. C. Francesconi, J. S. Lewis, Nucl. Med. Biol.
2013, 40, 3–14; b) J. P. Holland, M. J. Evans, S. L. Rice, J. Wongvipat, C. L.
Sawyers, J. S. Lewis, Nat. Med. 2012, 18, 1586–1591.
[8] a) T. K. Nayak, K. Garmestani, D. E. Milenic, M. W. Brechbiel, J. Nucl. Med.
2012, 53, 113–120; b) D. S. Abou, T. Ku, P. M. Smith-Jones, Nucl. Med.
Biol. 2011, 38, 675–681.
[9] L. R. Perk, M. J. W. D. Vosjan, G. W. M. Visser, M. Budde, P. Jurek, G. E.
Kiefer, G. A. M. S. van Dongen, Eur. J. Nucl. Med. Mol. Imaging 2010, 37,
250–259.
[10] J. N. Tinianow, H. S. Gill, A. Ogasawara, J. E. Flores, A. N. Vanderbilt, E.
Luis, R. Vandlen, M. Darwish, J. R. Junutula, S.-P. Williams, J. Marik, Nucl.
Med. Biol. 2010, 37, 289–297.
[11] E. W. Price, B. M. Zeglis, J. S. Lewis, M. J. Adam, C. Orvig, Dalton Trans.
2014, 43, 119–131.
[12] F. Guꢀrard, Y.-S. Lee, R. Tripier, L. P. Szajek, J. R. Deschamps, M. W. Brech-
biel, Chem. Commun. 2013, 49, 1002–1004.
[13] D. J. Cram, Angew. Chem. 1986, 98, 1041–1060; Angew. Chem. Int. Ed.
Engl. 1986, 25, 1039–1057.
[14] S. Liu, D. S. Edwards, Bioconjugate Chem. 2001, 12, 7–34.
[15] A. S. Reddy, M. S. Kumar, G. R. Reddy, Tetrahedron Lett. 2000, 41, 6285–
6288.
The X-ray structure of PuIV complexed with DFE reported previous-
ly[21] was used as a template to build the ZrIV-C7 complex. After Zr
substitution for Pu, each amide group in the ring was converted
into (-CH2CH2-) and then a fourth hydroxamate group with a
9-carbon alkyl chain on each end was inserted. After geometry op-
timization at the level of B3LYP/LanL2DZ, a CH2CH2 moiety in each
ring were deleted, and the resulting ZrIV-C7 complex was further
optimized in the gaseous phase at the M06L level with the pseu-
dopotential LanL2DZ for the Zr atom and the 6–31+G* basis set
for the rest of the atoms. The acyclic ZrIV-L7 complex was con-
structed by modifying the optimized ZrIV-C7 complex, and then
energy minimized. Starting with these two structures, the rest of
the cyclic and acyclic complexes were built successively by delet-
ing a CH2 moiety in each ring in their respective geometry-opti-
mized complex. To calculate the DG and DH for complexation reac-
tion, each ligand was constructed from its corresponding geome-
try-optimized complex by deleting ZrIV, and then protonating each
of the four hydroxamates. The geometries of these ligands were
optimized at the level of M06L/6–31+G*. To approximate the sol-
vent effect onto the energetics, single-point energy calculations
were also done in the reaction field of H2O with the PCM model as
implemented in Gausssian 09 software.[19] All calculations utilized
the five-component d functions, and the coordinates of the ligands
and complexes are listed in the Supporting Information.
ˇ
ˇ
[16] J. Prꢅsilovꢅ, J. Havlꢆcek, J. Inorg. Nucl. Chem. 1970, 32, 953–960.
[17] Y. Zhao, D. G. Truhlar, J. Chem. Phys. 2006, 125, 194101–194118.
[18] W. R. Wadt, P. J. Hay, J. Chem. Phys. 1985, 82, 284–298.
[19] Gaussian 09 (Revision A.02), M. J. Frisch G. W. Trucks, H. B. Schlegel,
G. E. Scuseria, M. A. Robb, J. R. Cheeseman, G. Scalmani, V. Barone, B.
Mennucci, G. A. Petersson, H. Nakatsuji, M. Caricato, X. Li, H. P. Hratch-
ian, A. F. Izmaylov, J. Bloino, G. Zheng, J. L. Sonnenberg, M. Hada, M.
Ehara, K. Toyota, R. Fukuda, J. Hasegawa, M. Ishida, T. Nakajima, Y.
Honda, O. Kitao, H. Nakai, T. Vreven, J. A. Montgomery, Jr., J. E. Peralta, F.
Ogliaro, M. Bearpark, J. J. Heyd, E. Brothers, K. N. Kudin, V. N. Staroverov,
R. Kobayashi, J. Normand, K. Raghavachari, A. Rendell, J. C. Burant, S. S.
Iyengar, J. Tomasi, M. Cossi, N. Rega, J. M. Millam, M. Klene, J. E. Knox,
J. B. Cross, V. Bakken, C. Adamo, J. Jaramillo, R. Gomperts, R. E. Strat-
mann, O. Yazyev, A. J. Austin, R. Cammi, C. Pomelli, J. W. Ochterski, R. L.
Martin, K. Morokuma, V. G. Zakrzewski, G. A. Voth, P. Salvador, J. J. Dan-
nenberg, S. Dapprich, A. D. Daniels, ꢇ. Farkas, J. B. Foresman, J. V. Ortiz,
J. Cioslowski, D. J. Fox, Gaussian Inc., Wallingford CT, 2009.
Acknowledgements
This work was supported by the Intramural Research Program
of the NIH, National Cancer Institute, Center for Cancer Re-
search and the Center for Information Technology. The quan-
tum chemical study utilized PC/LINUX clusters at the Center
for Molecular Modeling of the NIH (http://cit.nih.gov). The au-
thors are grateful to Dr. L.P. Szajek at the PET department for
providing 89Zr.
[20] In this model, Zr(OH)4 can be considered as the predominant form of
ZrIV in water at pH 7 in the absence of strong ligand at high dilution, as
reported previously. See: A. Veyland, L. Dupont, J. Pierrard, J. Rimbault,
M. Aplincourt, Eur. J. Inorg. Chem. 1998, 1765–1770.
Keywords: chelates
quantum chemistry · zirconium
· imaging agents · ligand design ·
[1] For reviews, see: a) M. Tredwell, V. Gouverneur, Angew. Chem. 2012,
124, 11590–11602; Angew. Chem. Int. Ed. 2012, 51, 11426–11437; b) Z.
Li, P. S. Conti, Adv. Drug Delivery Rev. 2010, 62, 1031–1051.
[21] M. P. Neu, J. H. Matonic, C. E. Ruggiero, B. L. Scott, Angew. Chem. 2000,
112, 1501–1503; Angew. Chem. Int. Ed. 2000, 39, 1442–1444.
[22] J. P. Holland, Y. Sheh, J. S. Lewis, Nucl. Med. Biol. 2009, 36, 729–739.
[2] For reviews, see: a) T. J. Wadas, E. H. Wong, G. R. Weisman, C. J. Ander-
son, Chem. Rev. 2010, 110, 2858–2902; b) Y. Zhou, K. E. Baidoo, M. W.
Brechbiel, Adv. Drug Delivery Rev. 2013, 65, 1098–1111; c) T. K. Nayak,
M. W. Brechbiel, Bioconjugate Chem. 2009, 20, 825–841.
[3] G. Fischer, U. Seibold, R. Schirrmacher, B. Wꢄngler, C. Wꢄngler, Molecules
2013, 18, 6469–6490.
Received: October 21, 2013
Revised: February 20, 2014
Published online on April 16, 2014
Chem. Eur. J. 2014, 20, 5584 – 5591
www.chemeurj.org
5591
ꢁ 2014 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim