N7metalCSD InorgChimActs 2016

Full text

Binding of metals to purine N7 nitrogen atoms and implications
for nucleic acids: A CSD survey
Filip Leonarski
a,b
, Luigi D’Ascenzo
a
, Pascal Auffinger
a,
⇑
a
Architecture et Réactivité de l’ARN, Université de Strasbourg, Institut de Biologie Moléculaire et Cellulaire du CNRS, 67084 Strasbourg, France
b
Faculty of Chemistry, University of Warsaw, Pasteura 1, 02-093 Warsaw, Poland
article info
Article history:
Received 21 February 2016
Received in revised form 31 March 2016
Accepted 1 April 2016
Available online 9 April 2016
Keywords:
Imine nitrogen
Imidazolate
Chemical balance
Metal ions
Metal substitution
abstract
Understanding the structure and function of RNA and DNA systems depends partly on our comprehen-
sion of the binding features of metal ions to nucleobases. Such knowledge is important for an unambigu-
ous assignment of ionic species to solvent electronic densities in crystallographic structures. Since the
purine N7 atom is considered to be the best nucleobase metal binding site, we focus herein on describing
the occurrence and coordination geometries of direct binding of alkali, alkaline earth and biologically
relevant transition metals to this site. Further, we compare binding of such metals to purine N7 atoms,
as well as imine sites occurring in small molecules such as imidazolates and water molecules. We analyze
also the structure of the coordination shell of penta- and tetrahydrated metal ions bound to one or two
purine N7 atoms. These structures can be used to validate proposed Mg
2+
and other metal binding sites in
large PDB structures where such assignments are often difficult to make. This survey suggests that Mg
2+
ions bind with weak affinities to nucleobase N7 atoms. Although Mg
2+
ions are essential to nucleic acid
systems, purine N7 binding sites are, in most contexts, probably not of primary importance in RNA and
DNA.
Ó2016 Elsevier B.V. All rights reserved.
1. Introduction
The binding of metal ions to nucleic acids and proteins, despite
the large number of studies and books devoted to the subject, is
still a very active and important field of research [1,2]. It is well
appreciated that nucleotides and amino acids interact with all
biologically relevant alkali, alkaline earth and transition metals
participating in the metallome [3,4] through nucleobases, sugar-
phosphate backbones, amino acid side chains and peptide back-
bones. Their binding affinities depend on the type of the involved
metals and binding site atoms that are sometimes categorized as
hard and soft [5,6]. It has been proposed to separate metal ions into
oxygen seeking, sulfur/nitrogen seeking and borderline or interme-
diate classes [7]. As such, it is assumed that hard metals such as
alkaline and alkaline earth ions (including Mg
2+
) are likely to asso-
ciate with the anionic oxygen atoms from phosphate (nucleic
acids) or carboxylate groups (proteins) while softer metals (Mn
2+
,
Ni
2+
,Cu
2+
,Zn
2+
or Cd
2+
as well as Ag
+
and Tl
+
) prefer to interact
with histidine/nucleobase imine nitrogen atoms [8]. Accordingly,
it has been suggested to group Na
+
,Mg
2+
,K
+
and Ca
2+
as an oxygen
class, Mn
2+
,Fe
2+
and Co
2+
as an imidazole class and Cu
2+
,Ni
2+
and
Zn
2+
as a sulfur class [9]. Although useful, such classifications have
to be considered with caution. Indeed, the hard Mg
2+
ion is found
in chlorophyll where it interacts in a pentacoordinated manner
with exactly five nitrogen atoms, four belonging to the chlorin
group and one to an additional histidine ring, instead of its ordi-
nary oxygen atom ligands. However, such an ‘‘out-law” complex
requires assistance of chelatase enzymes for its formation [3].
In the general case, the stabilities of complexes formed by
divalent metal ions in biologically relevant conditions has been
predicted to follow the order [3,10]:
Mg
2þ
<Mn
2þ
<Fe
2þ
<Co
2þ
<Ni
2þ
<Cu
2þ
>Zn
2þ
That is similar to the covalent contribution of metals [8]:
K
þ
<Na
þ
<Ca
2þ
<Mg
2þ
<Mn
2þ
<Fe
2þ
<Co
2þ
Ni
2þ
Cu
2þ
>Zn
2þ
In this respect, it can be recalled that the concentrations of
unbound metal ions in the cytosol range from millimolar (Na
+
,
K
+
,Mg
2+
) to micro- (Mn
2+
,Fe
2+
,Ca
2+
), nano- (Co
2+
,Ni
2+
), femto-
(Zn
2+
) and attomolar (Cu
+
/Cu
2+
)[11]. While proteins are exposed
to almost all kinds of biogenic metal ions including those
considered as toxic in higher organisms (such as Cd
2+
,[12]), nucleic
http://dx.doi.org/10.1016/j.ica.2016.04.005
0020-1693/Ó2016 Elsevier B.V. All rights reserved.
⇑
Corresponding author. Tel.: +33 388 41 70 49; fax: +33 388 60 22 18.
E-mail address: p.auffinger@ibmc-cnrs.unistra.fr (P. Auffinger).
Inorganica Chimica Acta 452 (2016) 82–89
Contents lists available at ScienceDirect
Inorganica Chimica Acta
journal homepage: www.elsevier.com/locate/ica

acids are in vivo almost exclusively surrounded by K
+
and Mg
2+
and
possibly by Na
+
and Ca
2+
ions [13,14]. When other ions are found in
the vicinity of nucleic acids, they are usually chelated like Zn
2+
in
zinc finger motifs.
For nucleobases, it has become common knowledge that the
best direct metal binding site is the purine N7 nitrogen [15,16]
and that direct binding to imine N1/N3 nitrogen atoms is much less
frequent and occurs only under specific conditions. The stabilities
of single nucleoside/metal complexes, determined in solution by
the affinity of N7 atoms, are the weakest for Mg
2+
and Ca
2+
and
highest for ions such as Mn
2+
,Zn
2+
,Cd
2+
and Cu
2+
in that order.
Further, adenine has weaker affinities for these metals than gua-
nine and even negative affinities for Ca
2+
and Mg
2+
. The same order
of affinities is reported for nucleotide macrochelate formation and
all other possible mono-, di- and triphosphate combinations as
well as for some dinucleotides [1]. It has also been reported that
the binding affinities of Mg
2+
and Mn
2+
to thiophosphates is not
very different and much weaker than that of Zn
2+
and Cd
2+
.
In crystallographic structures deposited in the PDB, numerous
examples of direct binding of transition metals to N7 atoms have
been reported [13,14] next to a very large number of N7-Mg
2+
binding events. The latter are supposed to play a role, for instance,
in the catalytic mechanism of the hammerhead ribozyme [17,18].
Yet, these N7-Mg
2+
binding events are rather surprising given the
above mentioned preference of these ions for anionic oxygen
atoms belonging to phosphate groups [7,8]. To get a better view
on the solvent structure of these large systems in experimental
studies, various metal ions are used as substitutes in the identifica-
tion of biologically relevant binding sites. For example, Tl
+
,Rb
+
and
Cs
+
have been used as heavy atom replacements for detecting Na
+
and K
+
binding sites, while Mn
2+
,Zn
2+
or Cd
2+
are used as
substitutes for Mg
2+
ions [19]. Some other metals, like Cd
2+
, are
also used as probes to study the effect of metal ions on nucleic
acids [20–23].
Hence, to clarify issues related to the structural characterization
and the role of metal ions in nucleic acid structures from the PDB, it
is important to gather reliable data and statistics on the preferred
coordination modes of these metals to nucleobases and especially
the purine N7 sites. For that purpose, we surveyed the Cambridge
Structural Database (CSD) for metal binding to similar sites. We
concentrated on the binding of hexacoordinated metal ions that
are probably the most biologically relevant. Present data comple-
ment those already reported for protein systems [8,9,24–26] and
represent an addition to existing web services providing access
to structural databases for metal binding sites [27,28].
2. Material and methods
The Cambridge Structural Database (CSD Version 5.37, February
2016) [29–31] was searched to characterize metal to nitrogen
atom coordination distances. We considered purine nucleobase-
like fragments and an imidazole ring fragment as found in both
purines or histidine amino acid side chains (Fig. 1). Note that for
imidazole and more specifically histidine rings, the two nitrogen
atoms are often reported as equivalent [8]. We considered also
an imine fragment common to the above-mentioned motifs where
the nitrogen is strictly bound to two carbon atoms. Besides, we
extracted metal–water coordination distances from the CSD. We
integrated in our search the following transition metals from the
first and second row (Mn, Fe, Co, Ni, Cu, Zn, Cd) as well as alkali
and alkaline earth metals. We further included Tl that is sometimes
used as a K
+
ion substitute [13,14]. However, we did not consider
beryllium (Be) since it is not present in its ionic forms in the
PDB. A metal to nitrogen coordination distance was selected based
on the existence of a coordination bond defined by the CSD. To
eliminate non-specific coordination, we excluded metals that are
located at more than ±1.0 Å from the plane defined by the
nitrogen-containing fragment. All searches were performed with
the ConQuest [32] software using filters so that disordered and
error-containing structures were excluded. The searches were
restricted to structures with R-factor values 60.05 unless
otherwise stated. The Mercury program was used for analyses
and visualization of all these structures [33].
Unfortunately structures deposited to the CSD, even those of
very high resolution, are not free of errors that persist despite sig-
nificant structure validation efforts [8,24,34]. Such structures are
difficult to eliminate from a search ensemble. Some of those, asso-
ciated with unreasonably short or long coordination distances
were eliminated after visual inspection that unavoidably includes
a certain level of arbitrariness. On the other hand, differences in
coordination lengths might be attributable to specific solid-state
interactions involving particular ligands. It can be noted that some
of the metal to nitrogen coordination distance histograms show
more than one peak that are most probably associated with
different metal oxidation states or Jahn–Teller effects as in the
case of Cu
2+
, low-spin Co
2+
,Ni
3+
, high-spin Cr
2+
and Mn
3+
[35].
Moreover, CSD oxidation states of transition metals are sometimes
ill-defined presumably due to typographical or other mistakes
[36,37]. Thus, associating an oxidation state with a given metal is
generally not straightforward. When possible, we present for each
element arguments that could lead to such an assignment.
As always, a critical eye is required even when working with
high-resolution crystallographic data.
3. Results and discussion
3.1. Statistical and structural overview of metal binding to imine
nitrogen atoms and water
As expected, among all metals the coordination distances with
N/O atoms vary the most for the alkali and alkaline earth ions
indicating that these distances can be used as a part of the ion
identification process (Table 1). The differences appear less signif-
icant for the investigated transition metals. Cd is the largest metal
ion with distances around 2.3 Å followed by Mn with distances
around 2.2 Å and Mg with coordination distances below 2.1 Å.
As such, the Mn coordination distances exceed systematically
those of Mg by 0.1 Å.
There is no significant difference in coordination distances of
investigated metals to N/O atoms associated with fragments III
and IV, respectively. Those differences lie within statistical uncer-
tainties. Thus, given the precision of the collected data associated
Fig. 1. CSD search fragments used for characterizing metal binding to various
purine (I), imidazole ring (II), imine like fragments (III) and water (IV). The black
dashed lines indicate that any bond type, as defined by the CSD, can be considered.
The red dashed lines indicate that we searched for a direct coordination between
the metal and the N/O atoms as defined in the CSD structures. The ‘‘*” next to a
nitrogen indicates that only sp
2
atoms are taken into account. All fragments are
planar. Water hydrogen atoms where explicitly included in fragment IV. (For
interpretation of the references to color in this figure legend, the reader is referred
to the web version of this article.)
F. Leonarski et al. / Inorganica Chimica Acta 452 (2016) 82–89 83

with the large diversity of structural fragments considered here, it
is difficult to infer simple rules regarding N/O coordination
distances.
Non-biologically-relevant contexts or environments occurring
frequently in the CSD may affect our statistics as well. Even very
precise quantum mechanical calculations provide context depen-
dent coordination distances. For instance, the Mg
2+
...Ow coordina-
tion distance of a single water molecule has been calculated by
quantum mechanical techniques to be in the 1.92–1.96 Å range
while for Mg[H
2
O]
6
2+
the same distance lie in the 2.08–2.10 Å range
and therefore closer to crystallographic derived values [38].
Some metals display more than one optimal coordination
distance. Thallium, which is considered as a good mimic for K
+
in
crystallographic investigations [39], displays a 2.7 Å coordination
distance to water and a 2.5 Å coordination distance to imine nitro-
gen atoms. Yet, Tl binds poorly to oxygen atoms, therefore the
statistics for Tl–O are not very reliable. Two coordination peaks
appear in the Tl–N histograms that could have as origin a different
thallium oxidation state (2.3 Å for Tl
3+
and 2.7 Å for Tl
+
). For Mn,
two peaks at 2.0 and 2.2 Å are distinguishable in the imine
nitrogen histograms. The short coordination distance could be
related to Mn atoms in a rare +3 oxidation state and associated with
a Jahn–Teller effect [35]. The +2 oxidation state is certainly related
to the more common 2.2–2.3 Å coordination distances. Note that
the Mn
2+
–water coordination distance (2.19 Å) is larger that the
Mg
2+
–water coordination distance (2.06 Å). For Fe, Co and Ni,
the two peaks related to the imine containing fragments are
separated by 0.1–0.2 Å. We assume that the shortest and longest
coordination distances can be attributed to the +3 and +2 oxidation
states, respectively. For Zn, two peaks associated with a large
spread are also observed. All transition metal to water coordination
distance histograms are single peaked except the one related to Cu
that is the result of a well-documented Jahn–Teller effect [24]. This
effect is not observed when nitrogen ligands are involved.
3.2. Low Mg
2+
binding occurrence to N7 atoms correlates with
nitrogen metal affinities
The occurrence of Mg
2+
binding events to imine nitrogen atoms
is relatively low in the CSD, especially for nucleobases. Only one
instance of a Mg
2+
ion bound to the N7 atoms of two in-plane
theophylline like purines [40] and another showing a Mg
2+
binding
to two stacked guanine N7 atoms (high R-factor) have been
reported [41]. This observation correlates with the low binding
level of Mg
2+
ions to histidines in proteins [8] and has to be
compared with the higher occurrence of other transition metals
next to N7 atoms (Table 1). Only one occurrence of Mn
2+
binding
to N7 (high R-factor) has been reported [42] along with two
pentahydrated Mn
2+
ion containing structures for which no
coordinates were deposited [43]. These examples will be described
below. It has to be noted that the apparent low occurrence of com-
plexes with Mg
2+
and Mn
2+
should be viewed with caution since it
may perhaps only reflect the fact that these compounds crystallize
less easily.
Regarding other transition metals, their higher affinity for nitro-
gen ligands correlates with a much larger number of contacts to
Table 1
Average metal to nitrogen coordination distances derived from the CSD (version 5.37, update February 2016), obtained by analyzing biologically relevant fragments (Fig. 1). The
number of hits is given in brackets. Standard deviations are provided when the number of hits is above ten. For some elements, more than one peak could be identified in distance
histograms and average values with standard deviations are given for each of them. The searches were restricted to structures with R-factor values 60.05 unless otherwise stated.
Disordered, error containing, polymeric and powder structures were excluded from the search.
Metals
a
Fragment IFragment II Fragment III Fragment IV
Purine: N7
b
Imidazole N
c
All imine N
d
Water
e
Alkali metals and thallium
f
Li-Lithium (Li
+
) 2.10 [3]
g
2.09 ± 0.03 [20] 2.05 ± 0.07 [682] 1.96 ± 0.06 [562]
Na-Sodium (Na
+
) 2.60 [3]
g
2.44 [3] 2.45 ± 0.05 [135] 2.41 ± 0.08 [3342]
K-Potassium (K
+
)–
g
2.82 [68]
g
2.85 ± 0.06 [81] 2.86 ± 0.13 [2222]
Rb-Rubidium (Rb
+
)–
g
2.94 [1]
g
3.04 ± 0.12 [10] 3.04 ± 0.17 [107]
Cs-Cesium (Cs
+
)–
g
–
g
3.11 [2] 3.24 ± 0.14 [326]
Tl-Thallium (I,III) –
g
2.49 [8] 2.42 ± 0.23 [221] 2.88 ± 0.38 [31]
(Thallium peaks) –
g
(2.32, 2.78) (2.27 ± 0.08, 2.68 ± 0.14)
Alkali earth metals
Mg-Magnesium (Mg
2+
) 2.23 [2] 2.19 ± 0.02 [37] 2.10 ± 0.07 [825] 2.06 ± 0.03 [1362]
Ca-Calcium (Ca
2+
)–
g
2.52 ± 0.07 [29] 2.48 ± 0.11 [328] 2.41 ± 0.06 [855]
Sr-Strontium (Sr
2+
)–
g
2.66 [9] 2.65 ± 0.11 [141] 2.61 ± 0.06 [293]
Ba-Barium (Ba
2+
) 2.85 [1]
g
2.9 [5] 2.89 ± 0.09 [134] 2.83 ± 0.09 [553]
Transition metals
Mn-Manganese (I–V) 2.32 [2]
g
2.23 ± 0.04 [345] 2.19 ± 0.11 [6294] 2.19 ± 0.05 [2524]
(Manganese peaks) (2.03 ± 0.05, 2.27 ± 0.05)
Fe-Iron (I–V) 2.16 [3] 2.09 ± 0.09 [749] 2.07 ± 0.11 [9337] 2.10 ± 0.05 [995]
(Iron peaks) (1.98 ± 0.03, 2.15 ± 0.05) (1.97 ± 0.04, 2.15 ± 0.07)
Co-Cobalt (I–IV) 2.05 ± 0.09 [11] 2.05 ± 0.08 [1317] 2.05 ± 0.11 [9710] 2.10 ± 0.03 [3946]
(Cobalt peaks) (1.96, 2.10) (1.93 ± 0.02, 2.13 ± 0.04) (1.94 ± 0.04, 2.13 ± 0.05)
Ni-Nickel (I–IV) 2.10 ± 0.05 [10] 2.07 ± 0.07 [1122] 2.01 ± 0.10 [12198] 2.08 ± 0.04 [4199]
(Nickel peaks) (1.91 ± 0.02, 2.07 ± 0.07) (1.89 ± 0.04, 2.07 ± 0.05)
Cu-Copper (I–III) 2.00 ± 0.03 [43] 1.99 ± 0.03 [2238] 2.00 ± 0.05 [26031] 2.24 ± 0.23 [5315]
(Copper peaks) (1.97 ± 0.02, 2.37 ± 0.17)
Zn-Zinc (I, II) 2.05 ± 0.5 [27] 2.05 ± 0.06 [927] 2.08 ± 0.07 [11509] 2.09 ± 0.05 [2715]
Cd-Cadmium (Cd
2+
) 2.33 ± 0.05 [25] 2.29 ± 0.05 [780] 2.34 ± 0.06 [4666] 2.31 ± 0.04 [1458]
a
When appropriate, oxidation states as mentioned in the CSD are given in parenthesis.
b
Statistics for the imine purine N7 atoms of fragment I.
c
Cumulated statistics for the two imidazole nitrogen atoms found in fragment II. These statistics include those related to fragment I.
d
Cumulated statistics for the imine atom found in fragment III. These statistics include those related to fragments Iand II.
e
Statistics for fragment IV.
f
Thallium in its Tl
+
form is often considered as a K
+
substitute and as such has been added to this table.
g
No restrictions were applied to these searches.
84 F. Leonarski et al. / Inorganica Chimica Acta 452 (2016) 82–89

fragment III compared to fragment IV (water). The reverse is
observed for alkali and alkali-earth metals including Mg
2+
where
the more frequent binder is oxygen. In that respect, it has been
reported from quantum mechanical investigations that for Mg
2+
the O6 inner shell binding mode is favored over the N7 binding
mode [44]. It is worth mentioning that Tl
+
shows an opposite trend
and is, like transition metals, more frequently bound to nitrogen
atoms (Table 1).
3.3. Binding characteristics of metals and water to purine N7 atoms
The CSD embeds a large diversity of chemical compounds
where tetra- and pentacoordinated metals coordinate to various
atoms such as sulfur, chloride, other metals,... Since such coordi-
nation patterns are rare in nucleic acids, we restrict our survey to
the more biologically relevant hexacoordinated metals. We found
that, when binding to purine N7 atoms, metal ions are mostly in
plane with the base and at a 3.7 ± 0.1 Å distance from the purine
N6/O6 atoms (Fig. 2). This distance along with the coordination
distance to the N7 atom could be used to distinguish these metals
from water molecules or NH
4
+
ions in lower resolution structures
from the PDB.
Indeed, the hydrogen bond distance for water molecules to N7
atoms is, as expected, close to 2.8 ± 0.1 Å. For 6-aminopurines, a
water molecule hydrogen bonded to both N6/N7 can be observed
with N6...Ow distances of 3.1 ± 0.1 Å. Besides, other in plane water
molecules are observed with N6...Ow distances in the 3.5–4.5 Å
range. For 6-oxopurines, the average O6...Ow distance histogram
has a first peak at 3.6 Å and a second close to 4.0 Å. No ‘‘in plane”
water molecules at hydrogen bond distance of both N7/O6 atoms
are observed.
Metal–N6/O6 distances around 3.0 Å were only observed for
tetracoordinated transition metals such as Cu or Zn and might only
be observed in structures of the PDB under specific crystallization
conditions. Such coordination schemes seem unlikely for Mg
2+
or
Mn
2+
ions for which direct coordination to both N7/O6 atoms has
not been reported (see below).
Metals can also bind to the N7 atom of 6-aminopurines with sim-
ilar distances as in 6-oxopurines. It is probable that this N7 purine
site will more difficultly accommodate larger ions such as Ca
2+
,
Sr
2+
or even K
+
. Thus, this 6-aminopurine site is probably more selec-
tive for smaller ions and it has even been argued if small ions can
bind to it. Gas phase quantum mechanical calculations (performed
in the absence of water) suggest that Mg
2+
might not bind to the
N7 atom of adenine [45]. Calculations taking into account first-shell
water molecules reach opposite conclusions [16,46].
3.4. Pentahydrated metal ions binding to purine N7 atoms
Eight structures of pentahydrated metals (Co, Ni and Cd) bound
to the N7 atom of purine like fragments were deposited to the CSD
as well as four other fragments (with Mn, Ni and Fe) for which no
coordinates were archived (Table 2). Seven of these metals are
bound to guanine or inosine fragments and one to adenine. The
metal ion positions are consistent with the ion coordination dis-
tances noted in Table 1. In all instances, the metal–N6/O6 distance
is 3.7 ± 0.1 Å (Fig. 2). Thus, from this limited set of examples, diva-
lent transition metals appear to bind similarly to 6-aminopurines
and 6-oxopurines. Hence, this metal–N6/O6 distance can be
considered as a reliable criterion for characterizing metal binding
to N7 atoms in the lower resolution structures of the PDB.
The placement of the five coordinated water molecules is simi-
lar in all structures suggesting a regular hydration pattern for ions
bound to purine N7 atoms (Fig. 3). Overall, the five water mole-
cules form along with the N7 atom an octahedral coordination
scheme. The closest water molecule to the N6/O6 atoms is at a
2.8 ± 0.1 Å hydrogen bond distance. The second closest water
molecule to the N6/O6 atoms is at a 3.4 ± 0.2 Å non-hydrogen
bonding distance indicating an asymmetrical arrangement with
respect to the purine plane. Both these distances are in agreement
with N6–Ow distances close to 2.8 and 3.3 Å derived from quan-
tum mechanical calculations [16] The water closest to the N6/O6
atoms is either a hydrogen bond acceptor or donor. Hence, in this
case also, these two N6/O6-water distances can be considered as
useful criteria for validating metal binding to N7 atoms in PDB
structures.
3.5. Water orientation in the metal coordination sphere is adaptable
The positions of the hydrogen atoms belonging to coordinated
water molecules are also in agreement with quantum mechanical
calculations [16,38] and first principle molecular dynamics calcula-
tions of the hydration of Mg
2+
ions in aqueous solution [47–49].
These calculations as well as high-resolution CSD data, suggest that
water molecules tend to asymmetrically coordinate Mg
2+
ions
through one of the oxygen atom lone pairs, an outcome that could
probably not have been derived from simple electrostatic gas phase
considerations. In the pentahydrated metal bound purines (Table 2),
the angle associated with the metal ion and the bisector of the coor-
dinated water molecules is 146 ± 17°. These angles are 152 ± 14°
(CSD code: YOHJAI) and 171 ± 2°(CSD code: CIRVAA01) in two
neutron diffraction structures of hexahydrated Mg
2+
ions. However,
in DFT calculations of hexahydrated Mg
2+
gas phase clusters, the
water molecules coordinate rather symmetrically to Mg
2+
[19,38,48]. These differences suggest a strong influence of the
environment on the orientation of the water molecules. Such an
unexpected adaptability of the water molecule orientation seems
Fig. 2. Definition of two characteristic distances for metal binding to purine N7
atoms. (Left) The average d2 distance is derived from an ensemble of hexacoor-
dinated metal ions binding to the N7 atom. (Right) d1 and d2 distances for water
molecules hydrogen bonded to both N6/N7 atoms of 6-aminopurines.
Table 2
CSD structures of pentahydrated ions bound to a purine N7 atom (Fig. 3).
Purine Ion d1
a
d2
b
Hydrogens R-factor [%] CSD code
Pentahydrated metal
G Cd 2.37 3.83 No 6.0 AGOPCD
Inosine Ni 2.11 3.74 Yes 7.5 ANIMPH01
G Co 2.13 3.71 Yes 3.4 BIPVIF01
Inosine Co 2.15 3.70 Yes 4.3 DIDSOY
Inosine Co 2.12 3.68 Yes 2.8 FIZHUR
Inosine Ni 2.06 3.64 Yes 3.2 FIZJAZ10
Inosine Co 2.16 3.79 Yes 5.1 IMPCOH
A Ni 2.07 3.74 Yes 2.4 ZZZAAF01
G
c
Mn FAMNIQ01
G
c
Mn QQQGLY
G
c
Ni GUOSNI
G
c
Fe FAMNEM
a
Distance between the metal ion and the N7 atom (Fig. 2).
b
Distance between the N7 and the N6/O6 atoms (Fig. 2).
c
No coordinates were deposited to the CSD.
F. Leonarski et al. / Inorganica Chimica Acta 452 (2016) 82–89 85

required to accommodate metal binding to both 6-aminopurines
and 6-oxopurines.
In the adenine/Ni complex, the purine amino group remains
essentially planar despite the presence of the hydrated metal ion,
in opposition to quantum mechanical calculations that advocate
its strong pyramidalisation [16]. Yet, one has to exert caution
regarding the hydrogen positions inferred from crystallographic
structures that might be sometimes affected by refinement options
and might not always be reliable [34]. In this respect, it can be
noted that guanine amino groups that are distant from metal ions
are sometimes non planar in CSD structures. Such hydrogen atom
positions are certainly also very sensitive to their environment.
As a word of caution, it should be considered that in biomolec-
ular systems metals could bear other ligands than water to com-
plete their solvation shell such as for example Cl coordinated to
Zn as observed in a Z-DNA structure [50]. This is also important
for metals like Pt, Pd or Ag. Ligands like OH

are also probable
and were considered as bridging compounds in bimetallic com-
plexes [51,52].
3.6. Hydrated metal ions coordinated to two purine N7 atoms
Besides the above-mentioned pentahydrated metal ions coordi-
nating to a single N7 atom, we found only two binding patterns in
the CSD that involve tetrahydrated metals and two purine N7
atoms. The first comprises a planar and the second a stacked
arrangement of the two purines (Fig. 4). In one instance, an amino
group is found in position 6 (Table 3). All these structures are sim-
ilar and the purines are organized in a head-to-tail manner. In
biomolecular systems, such arrangements would correspond to
metal mediated base pairs. As for pentahydrated metals (Fig. 2),
the orientation of the metal-bound water molecules differs. Metal
coordinated water molecule close to position 6 are involved in a
hydrogen bond with O6 but not N6 atoms. In the latter, the closest
hydrogen atoms of water are pointing away from the amino group.
The second pattern involves stacked head-to-tail purines that
occur in the CSD structure of the cyclic diguanylic acid or cyclic
d-GMP complex with Mg
2+
[41], a molecule that is recognized as
a second messenger used in signal transduction in a wide variety
of bacteria [42]. In these patterns, the two purines are highly tilted
(37 ± 5°;Table 4). They display large R-factor values (8.8 ± 1.9) that
Fig. 3. CSD structures of pentahydrated metals binding to purine N7 atoms and N6/O6–Ow distances. For clarity, the sugar carbon bound hydrogens are not shown. (a)
Structure derived from a complex showing a Ni ion bound to an adenine (CSD code: ZZZAAF01). (b) Structure derived from a complex showing a Co ion bound to a guanine
(CSD code: BIPVIF01). The average distances are derived from the structures listed in Table 2.
Fig. 4. CSD structures of tetrahydrated metals binding to two purine N7 atoms. (a) This structure is derived from a complex showing a Mg
2+
ion bound to two theophyllines in
a planar arrangement (CSD code: CUCZEH). (b) This structure is derived from a complex showing a Mg
2+
ion bound to two guanines in a stacked arrangement (CSD code:
SUKHUB).
Table 3
CSD structures of tetrahydrated ions bound to two N7 atoms in a planar purine
arrangement (Fig. 4a).
Purine Ion d1
a
d3
b
Hydrogens R-factor [%] CSD code
Tetrahydrated metal –planar arrangement
A Cu 2.0/2.0 4.0 Yes 3.1 AMADCU
G Cu 2.0/2.1 4.1 Yes 4.8 BAHMAY
Theophylline Mg 2.2/2.2 4.6 Yes 4.9 CUCZEH
Theophylline Cd 2.3/2.3 4.6 Yes 4.4 DIFRAL
G Ni 2.2/2.2 4.3 Yes 3.8 HOPBOD
2-Amino purine Co 2.1/2.1 4.3 Yes 9.5 HOZBEF
Xanthine Zn 2.2/2.2 4.4 No 5.6 JIXFEB
Xanthine Ni 2.2/2.2 4.3 Yes 3.0 LIZMOW
G Cd 2.3/2.3 4.6 No 3.5 NARREE
G Cd 2.3/2.3 4.6 No 4.7 NARRII
Inosine Cu 2.0/2.0 4.5 Yes 4.0 TAHYPC
a
Distance between the metal ion and the N7 atoms (Fig. 2).
b
Distance between the two N7 atoms.
86 F. Leonarski et al. / Inorganica Chimica Acta 452 (2016) 82–89

suggest the occurrence of structural stress in the crystals. As
expected, the two metal–N7 distances are close. Besides, short
N7–N7 distances (2.9 ± 0.1 Å) are observed (the latter are similar
to the distance between nearby water molecules in the first metal
hydration shell). These distances are also much shorter than the
stacking distance between two nucleobases (3.4 Å). Such
distances and angles should be regarded as characteristic for metal
coordination to two N7 atoms belonging to stacked purines.
Indeed, such tilts are rare in biomolecular structures and might
be characteristic of metal binding if they are associated with a
short N7–N7 distance. In these parallel and stacked arrangements
as well as in the binding of pentahydrated ions to N7 atoms, all
metals occupy the same binding spots and suggest their ability
to replace each other in larger structures.
Interestingly, for the stacked arrangement (Fig. 4b), the N7
binding site can also be occupied by a Na
+
ion (CSD code:
GUOPNA12) [53]. The tilt of the two purines (36°) is similar to
those observed in other complexes (Table 4). All other features of
this arrangement are similar too, with the exception of the dissym-
metric Na
+
–N7 distances (2.4 and 2.6 Å). The N7–N7 distance is
also larger (3.3 Å) and closer to normal stacking distances. This
represents an example where a monovalent ion can occupy a site
that is generally attributed to divalent ions in biomolecular sys-
tems, a feature that should be kept in mind during the refinement
and analysis of the solvent structure of the large nucleic acid
systems deposited to the PDB.
3.7. Simultaneous metal binding to N6/O6 and N7 atoms – the case of
(imid)azolates
The simultaneous binding of Mg
2+
or a transition metal to
purine O6/N7 atoms is sometimes considered in quantum mechan-
ical calculations [54] and it has marginally been inferred that bind-
ing to O6 over N7 atoms is preferred [44,55,56]. In the CSD, such
events are not observed for purines but only for the closely related
imidazolate and azolate compounds (over 500 occurrences with 13
different metals including Na
+
,Ca
2+
,Sr
2+
and Ba
2+
). The main dif-
ference between purines and (imid)azolates relates to the presence
of a carboxylate group in the latter and a slightly different binding
site geometry that leads to a 0.4 Å change between the N/O coor-
dinating atoms (Fig. 5). In (imid)azolates, the two N/O atoms are at
the appropriate 2.7 Å distance for completing the coordination
sphere of a transition metal ion such as Mg
2+
whereas purines with
a3.1 Å distance and a different orientation of the coordinating
groups are not. This 2.7 Å distance correlates with the coordination
distance d(Ow...Ow) of water molecules in the first hydration shell
of first row transition metals that is around 2.9 Å (Table 5).
However, this (imid)azolate site can also accommodate larger ions
such as Na
+
,Ca
2+
and Cd
2+
for which the d(Ow...Ow) distance
extends to 3.2 Å for Cd
2+
and even 3.5 for Na
+
. Thus, another
plausible explanation is related to the fact that the carboxyl group
has a higher affinity than the carbonyl group for these metals. Very
likely, we observe here a combination of both effects.
Larger ions such as Ca
2+
[57] and Ba
2+
[58] were found to
coordinate simultaneously to N7/O6 atoms in PDB structures.
Therefore, it is also likely that ions such as Na
+
or K
+
(that has, like
Ba
2+
,a2.8 Å coordination distance) could bind to the N7/O6
atoms of a guanine in large nucleic acid structures. Yet, it seems
that O6 atoms are better binding sites for alkali ions as they are
involved in maintaining guanine quartet structures occurring for
instance in telomeres. Alkali ions were also reported to interact
with thymine O2 atoms in a structure of d(ApT) minihelix [59].
Note that for 6-aminopurines, no tautomeric forms involving the
deprotonation of the adenine and associated with a direct metal–
N6 contact, as reported elsewhere for metals such as platinum
[15,60], were observed.
Table 4
CSD structures of tetrahydrated ions bound to two N7 atoms in a stacked purine
arrangement (Fig. 4b).
Purine Ion d1
a
d3
b
Angle
c
Hydrogens R-factor [%] CSD code
Tetrahydrated metal – stacked arrangement
Inosine Co 2.2/2.2 2.9 41 No 9.0 BEXRAX10
G Co 2.2/2.2 2.9 42 No 10.0 BEXREB10
G Zn 2.2/2.2 3.0 36 No 6.7 DAZTED
G Zn 2.1/2.1 3.0 33 No 6.1 DAZTIH
G Cu 2.0/2.0 2.8 44 Yes 4.7 ESIWOT
Inosine Cu 2.0/2.0 2.8 42 No 9.3 GANXOI
G Ni 2.1/2.1 2.9 39 No 13.1 GAVDIQ
G Na 2.4/2.6 3.3 36 Yes 7.9 GUOPNA12
G Mn 2.3/2.4 3.0 30 No 9.6 QOCVIP
G Co 2.2/2.3 2.9 35 No 11.2 SIWWIE10
G Mg 2.3/2.3 2.9 33 No 9.4 SUKHUB
a
Distance between the metal ion and the N7 atoms (Fig. 2).
b
Distance between the two N7 atoms.
c
Tilt angle between the two purine planes.
Fig. 5. Difference between 6-oxopurine and imidazolate like fragments. (a and b) Comparison between the imide nitrogen and carbonyl oxygen atom distances in inosine
(CSD code: FIZHUR) and imidazolate (CSD code: DEZNIG). (c) Complex between an imidazolate and a Mg
2+
ion (CSD code: DEZNIG). (d) Neutron diffraction structure of a
hexahydrated Mg
2+
complex (CSD code: YOHJAI).
F. Leonarski et al. / Inorganica Chimica Acta 452 (2016) 82–89 87

3.8. Are (imid)azolates a reliable N/O metal affinity balance?
As described above, imidazolates and the related azolates are
specific classes of molecular fragments where an imine nitrogen
and an anionic carboxylate oxygen atom bind simultaneously to
a metal ion in a close to perfect geometry. Thus, we thought that
these compounds could reflect the difference in affinity of a metal
for the N versus O atoms [61] and could provide information sim-
ilar to those provided by a large family of ‘‘molecular balances”
that involve, among others, the use of rotameric folding molecules
to quantify non-covalent interactions [62]. We are calling this
tentatively a ‘‘metal affinity balance”. Hence, we compared the
metal–N to metal–O coordination distances in these compounds
(Table 6). Despite the shortage of data leading to poor statistics,
the hard alkaline earth metals including Mg
2+
seem to prefer bind-
ing to oxygen over nitrogen. On the other hand, the softer cations
such as Cu, Zn or Cd are associated with shorter metal–N distances
reflecting a higher affinity for nitrogen. This last result is somewhat
surprising since the imine nitrogen atom is in competition with an
‘‘anionic” carboxylate oxygen atom. The apparent preference of
Mg
2+
for oxygen atoms is probably also at play in large nucleic acid
systems suggesting that N7 atoms are at best secondary interaction
sites populated only under specific conditions and that contact
distances of Mg
2+
to imine nitrogen atoms can be stretched from
the optimal 2.1 Å coordination distance to a less frequent 2.2 Å
or even larger coordination distance in specific contexts.
3.9. Metal ion substitutions in small and large structures
We described here several metal binding patterns associated
with a large diversity of metals. These binding patterns suggest that
transition metal ions, that produce easily recognizable electron
densities, can be used as a probe for inferring the binding of ions
such as Mg
2+
,Na
+
or K
+
that display weak and/or non-characteristic
electron densities. However, such an assertion should be taken with
some caution especially regarding their relevance for in vitro as well
as crystallographic studies. For instance, the crystal structures of
the Mg
2+
and Mn
2+
cyclic d-GMP complexes exhibit both the same
arrangement of stacked purines linked to the metal through their
N7 atoms [42]. However, their spectroscopic properties were found
to be very different in solution. While the Mg
2+
ion did not produce
any signal change with respect to the metal free conditions, the
Mn
2+
ion affected significantly the spectroscopic properties of this
molecule. Similarly, very small spectroscopic effects of Mg
2+
ions
over Zn
2+
or Cd
2+
ions on the hammerhead ribozyme were reported
[63]. On the crystallographic side, a systematic study conducted on
the binding of metals to a RNA duplex, revealed strong differences
in the association of eleven monoatomic ions and two hexamines to
the structure. This study raised interrogations related to the use of
metal substitutions in large structures [64]. Indeed, a conforma-
tional change induced by the presence of a Mn
2+
ion was also
reported for a signal recognition particle (SRP) where Mg
2+
was
substituted by Mn
2+
[65]. In these structures, Mn
2+
changed the
conformation of a nucleotide by binding to a site that allowed to
form inner sphere coordination with a N7 atom and a non bridging
phosphate oxygen atom from a neighboring residue. Here also, the
exact position of a metal is strongly dependent upon the nature of
the metal and its environment.
4. Summary and perspectives
Present data extracted from the CSD should help warrant
correct identification of metal binding sites in nucleic acids and
other biopolymers, a process that is often complicated by the lower
resolution of the structures deposited to the PDB. This is especially
true for ions like Mg
2+
that are isoelectronic with Na
+
/NH
4
+
ions and
water molecules [66]. For instance, we showed that N7–ion dis-
tances could be used for preliminary metal identification but also
for distinguishing metals from water molecules or hydrogen bond-
ing ions such as NH
4
+
. The N6/O6–ion distances around 3.7 Å and
the metal bound water molecule orientation are likewise charac-
teristic of metal binding to purine N7 atoms. Although simultane-
ous binding of transition metals and Mg
2+
ions to N7/O6 atoms can
be excluded, larger ions such as K
+
or Ba
2+
are able to coordinate to
both atoms. However in 6-aminopurines, only the smaller divalent
metals seem to be able to bind to N7 atoms.
Such data should also help to better conduct and understand
biochemical substitution experiments whose results are sometimes
difficult to interpret. Besides, this survey provides data allowing to
improve parameterization of classical and polarizable molecular
dynamics force fields, the accuracy of which being essential for
providing a good understanding of structure–dynamics–function
relationships of biomolecular systems. Designing a correct parame-
terization for these force fields is still very challenging and depends
largely on our ability to interpret experimental data [18,67].
Acknowledgements
P.A. wishes to thank Prof. Eric Westhof for ongoing support as
well as Dr Quentin Vicens for careful reading of the manuscript
and for helpful discussions. L.D. received a PhD fellowship
from the French ‘‘Ministère de la recherche et de l’enseignement”.
Table 5
Distance d(OwOw) between two close water molecules in the
first hydration shell of hexahydrated metal ions derived from a
CSD search (Fig. 5d). The number of hits is given in brackets.
Standard deviations are provided when the number of hits is
above ten. The searches were restricted to structures with R-factor
values 60.05. Disordered, error containing, polymeric and powder
structures were excluded from the search.
Metals d(OwOw)
Na 3.47 [2]
Mg 2.92 ± 0.06 [132]
Ca 3.22 [5]
Mn 3.08 ± 0.08 [53]
Fe 2.99 ± 0.06 [29]
Co 2.95 ± 0.06 [138]
Ni 2.91 ± 0.06 [130]
Cu 2.87 [3]
Zn 2.95 ± 0.07 [61]
Cd 3.21 ± 0.10 [21]
Table 6
Metal–N and metal–O coordination distances in imidazolate (Fig. 5) and azolate
compounds. A negative and positive ‘‘delta” value suggest a higher affinity of the
metal for oxygen and nitrogen, respectively. The number of hits is given in brackets.
Standard deviations are provided when the number of hits is above ten. No
restrictions were applied to this search.
Metals d(metal...N) d(metal...O) Delta
Alkali earth metals
Mg 2.20 [2] 2.09 [2] 0.11
Ca 2.54 [2] 2.44 [2] 0.10
Sr 2.74 ± 0.05 [15] 2.66 ± 0.06 [15] 0.08
Ba 2.97 ± 0.13 [18] 2.86 ± 0.09 [18] 0.11
Transition metals
Mn 2.23 ± 0.08 [63] 2.21 ± 0.08 [63] 0.02
Fe 2.16 ± 0.06 [22] 2.15 ± 0.07 [22] 0.01
Co 2.03 ± 0.12 [190] 2.02 ± 0.11 [190] 0.01
Ni 2.06 ± 0.05 [196] 2.10 ± 0.06 [196] +0.04
Cu 1.98 ± 0.08 [87] 2.12 ± 0.21 [87] +0.14
Zn 2.08 ± 0.06 [130] 2.18 ± 0.08 [130] +0.10
Cd 2.27 ± 0.04 [263] 2.40 ± 0.07 [263] +0.13
88 F. Leonarski et al. / Inorganica Chimica Acta 452 (2016) 82–89

F.L. gratefully acknowledges financial support from the Polish Min-
istry of Higher Education and Science (Mobility Plus programme,
Project No. 1103/MOB/2013/0).
References
[1] R.K.O. Sigel, H. Sigel, Metal–ion interactions with nucleic acids and their
constituents, in: J. Reedjik, K. Poeppelmeier (Eds.), Comprehensive Inorganic
Chemistry II, Elsevier, Oxford, 2013, pp. 623–660.
[2] T. Dudev, C. Lim, Chem. Rev. 114 (2014) 538.
[3] R.J.P. Williams, Coord. Chem. Rev. 216 (2001) 583.
[4] H.A. Hill, P.J. Sadler, J. Biol. Inorg. Chem. 21 (2016) 5.
[5] D.R. Garmer, N. Gresh, J. Am. Chem. Soc. 116 (1994) 3556.
[6] T. Dudev, Y.L. Lin, M. Dudev, C. Lim, J. Am. Chem. Soc. 125 (2003) 3168.
[7] E. Nieboer, D.H.S. Richardson, Environ. Pollut. Ser. B 1 (1980) 3.
[8] M.J. Harding, M.W. Nowicki, M.D. Walkinshaw, Cryst. Rev. 16 (2010) 247.
[9] H. Zheng, M. Chruszcz, P. Lasota, L. Lebioda, W. Minor, J. Inorg. Biochem. 102
(2008) 1765.
[10] H. Irving, R.J.P. Williams, Nature 162 (1948) 746.
[11] A.W. Foster, D. Osman, N.J. Robinson, J. Biol. Chem. 289 (2014) 28095.
[12] T.W. Lane, M.A. Saito, G.N. George, I.J. Pickering, R.C. Prince, F.M.M. Morel,
Nature 435 (2015) 42.
[13] P. Auffinger, N. Grover, E. Westhof, Met. Ions Life Sci. 9 (2011) 1.
[14] P. Auffinger, L. D’Ascenzo, E. Ennifar, Met. Ions Life Sci. 16 (2016) 167.
[15] B. Lippert, Coord. Chem. Rev. 200–202 (2000) 487.
[16] J. Sponer, M. Sabat, L. Gorb, J. Leszczynski, B. Lippert, P. Hozba, J. Chem. Phys. B
104 (2000) 7535.
[17] A. Mir, J. Chen, K. Robinson, E. Lendy, J. Goodman, D. Neau, B.L. Golden,
Biochemistry 54 (2015) 6369.
[18] M.T. Panteva, G.M. Giambasu, D.M. York, J. Phys. Chem. B 119 (2015) 15460.
[19] C.W. Bock, A.K. Katz, G.D. Markham, J.P. Glusker, J. Am. Chem. Soc. 121 (1999)
7360.
[20] D. Donghi, J. Schnabl, Met. Ions Life Sci. 9 (2011) 197.
[21] M.C. Erat, R.K. Sigel, Met. Ions Life Sci. 9 (2011) 37.
[22] M. Pechlaner, R.K. Sigel, Met. Ions Life Sci. 10 (2012) 1.
[23] R.K. Sigel, M. Skilandat, A. Sigel, B.P. Operschall, H. Sigel, Met. Ions Life Sci. 11
(2013) 191.
[24] M.H. Harding, Acta Crystallogr., D 55 (1999) 1432.
[25] M.H. Harding, Acta Crystallogr., D 57 (2001) 401.
[26] M.H. Harding, Acta Crystallogr., D 56 (2000) 857.
[27] H. Zheng, M.D. Chordia, D.R. Cooper, M. Chruszcz, P. Muller, G.M. Sheldrick, W.
Minor, Nat. Protoc. 9 (2014) 156.
[28] H. Zheng, I.G. Shabalin, K.B. Handing, J.M. Bujnicki, W. Minor, Nucleic Acids
Res. 43 (2015) 3789.
[29] F.H. Allen, Acta Crystallogr., B 58 (2002) 380.
[30] J. Chisholm, E. Pidcock, J. van de Streek, L. Infantes, S. Motherwell, F.H. Allen,
CrystEngComm 8 (2006) 11.
[31] C.R. Groom, F.H. Allen, Angew. Chem. Int. Ed. Engl. 53 (2014) 662.
[32] I.J. Bruno, J.C. Cole, P.R. Edgington, M. Kessler, C.F. Macrae, P. McCabe, J.
Pearson, R. Taylor, Acta Crystallogr., B 58 (2002) 389.
[33] C.F. Macrae, I.J. Bruno, J.A. Chisholm, P.R. Edgington, P. McCabe, E. Pidcock, L.
Rodriguez-Monge, R. Taylor, J. van de Streek, P.A. Wood, J. Appl. Crystallogr. 41
(2008) 466.
[34] A.L. Spek, Acta Crystallogr., D 65 (2009) 148.
[35] C. Murray, B. Gildea, H. Muller-Bunz, C.J. Harding, G.G. Morgan, Dalton Trans.
41 (2012) 14487.
[36] G.P. Shields, P.R. Raithby, F.H. Allen, W.D. Motherwell, Acta Crystallogr., B 56
(2000) 455.
[37] S.E. Harris, A.G. Orpen, J.J. Bruno, R. Taylor, J. Chem. Inf. Model. 45 (2005) 1727.
[38] G.D. Markham, J.P. Glusker, C.W. Bock, J. Phys. Chem. B 106 (2002) 5118.
[39] S. Basu, S.A. Strobel, Methods 23 (2001) 264.
[40] Y. Shi, B. Lou, Acta Crystallogr., E 71 (2015) 321.
[41] M. Egli, R.V. Gessner, L.D. Williams, G.J. Quigley, G.A. Vandermarel, J.H.
Vanboom, A. Rich, C.A. Frederick, Proc. Natl. Acad. Sci. U.S.A. 87 (1990) 3235.
[42] V. Stelitano, A. Brandt, S. Fernicola, S. Franceschini, G. Giardina, A. Pica, S.
Rinaldo, F. Sica, F. Cutruzzola, Nucleic Acids Res. 41 (2013) e79.
[43] P. De Meester, D.M. Goodgame, T.J. Jones, A.C. Skapski, Biochem. J. 139 (1974)
791.
[44] J.E. Sponer, V. Sychrovsky, P. Hobza, J. Sponer, Phys. Chem. Chem. Phys. 6
(2004) 2772.
[45] N. Russo, M. Toscano, A. Grand, J. Phys. Chem. A 107 (2003) 11533.
[46] L. Rulisek, J. Sponer, J. Phys. Chem. B 107 (2003) 1913.
[47] J.M. Martinez, R.R. Pappalardo, E.S. Marcos, J. Am. Chem. Soc. 121 (1999) 3175.
[48] F.C. Lightstone, E. Schwegler, R.Q. Hood, F. Gygi, G. Galli, Chem. Phys. Lett.
(2001) 549.
[49] A. Bhattacharjee, A.B. Pribil, B.R. Randolf, B.M. Rode, T.S. Hofer, Chem. Phys.
Lett. 536 (2012) 39.
[50] P. Drozdzal, M. Gilski, R. Kierzek, L. Lomozik, M. Jaskolski, Acta Crystallogr., D
69 (2013) 1180.
[51] J.P. Glusker, A.K. Katz, C.W. Bock, Struct. Chem. 12 (2001) 323.
[52] C. Grauffel, C. Lim, J. Phys. Chem. B (2015).
[53] M. Dracinsky, M. Sala, P. Hodgkinson, CrystEngComm 16 (2014) 6756.
[54] A.S. Petrov, G. Lamm, G.R. Pack, J. Phys. Chem. B 106 (2002) 3294.
[55] A. Famulari, F. Moroni, M. Sironi, M. Raimondi, Comput. Chem. 24 (2000) 341.
[56] I. Solt, I. Simon, A.G. Csaszar, M. Fuxreiter, J. Phys. Chem. B 111 (2007) 6272.
[57] T.K. Chiu, R.E. Dickerson, J. Mol. Biol. 301 (2000) 915.
[58] M.M. Perbandt, M.M. Vallazza, C.C. Lippmann, C.C. Betzel, V.V. Erdmann, Acta
Crystallogr., D 57 (2001) 219.
[59] N.C. Seeman, J.M. Rosenberg, F.L. Suddath, J.J. Kim, A. Rich, J. Mol. Biol. 104
(1976) 109.
[60] M. Roitzsch, B. Lippert, J. Am. Chem. Soc. 126 (2004) 2421.
[61] A. Nimmermark, L. Ohrstrom, J. Reedijk, Z. Kristallogr. 228 (2013) 311.
[62] I.K. Mati, S.L. Cockroft, Chem. Soc. Rev. 39 (2010) 4195.
[63] G. Wang, B.L. Gaffney, R.A. Jones, J. Am. Chem. Soc. 126 (2004) 8908.
[64] E. Ennifar, P. Walter, P. Dumas, Nucleic Acids Res. 31 (2003) 2671.
[65] R.T. Batey, J.A. Doudna, Biochemistry 41 (2002) 11703.
[66] L.D. Williams, Top. Curr. Chem. 253 (2005) 77.
[67] R.P.P. Neves, S.F. Sousa, P.A. Fernandes, M.J. Ramos, J. Chem. Theory Comput. 9
(2013) 2718.
F. Leonarski et al. / Inorganica Chimica Acta 452 (2016) 82–89 89