Documenti di Didattica
Documenti di Professioni
Documenti di Cultura
In the search for novel tools to combat cancer, nanoparticles (NPs) have attracted a lot of attention.
Recently, the controlled release of cancer-cell-killing metal ions from doped NPs has shown promise, but
fine tuning of dissolution kinetics is required to ensure specificity and minimize undesirable toxic side-
effects. Theoretical tools to help in reaching a proper understanding and finally be able to control the dis-
solution kinetics by NP design have not been available until now. Here, we present a novel set of true
nanodescriptors to analyze the charge distribution, the effect of doping and surface coating of whole
Received 28th June 2018, metal oxide NP structures. The polarizable model of oxygen atoms enables light to be shed on the
Accepted 27th October 2018
charge distribution on the NP surface, allowing the in detail study of the factors influencing the release of
DOI: 10.1039/c8nr05220d metal ions from NPs. The descriptors and their capabilities are demonstrated on a Fe-doped ZnO nano-
rsc.li/nanoscale particle system, a system with practical outlook and available experimental data.
Introduction been demonstrated for example with ZnO NPs, where dis-
solution of the NPs and the associated release of Zn2+ ions
While nanoparticles (NPs) are already used in medicine as bio- have been shown to specifically affect p53-mutated cancer
sensors and drug delivery agents,1,2 their biomedical use in, cells.20 Various studies have demonstrated that the rate of
for example, the treatment of cancer remains challenging.3 ZnO dissolution can be affected, for example by coating the
Aside from many beneficial properties of NPs, their high bioac- ZnO NPs with silica or by doping them with Fe2+ ions.21,22 As
tivity is often associated with toxic side effects, which have demonstrated, the Fe-doping reduces the toxicity to normal
been broadly described elsewhere.4–12 cells and lowers the dissolution rates. Clearly, the line
NPs are also used as drug delivery systems due to their between curing cancer and killing healthy cells is a thin one;
improved solubility and pharmacokinetics.13,14 One of the therefore, the factors influencing the dissolution kinetics of
studied delivery media is mesoporous silica as it can be modi- metal oxide NPs need to be well understood and accurately
fied in multiple ways.15,16 Huang et al.16 focused mainly on the tunable in order to make this approach valid for real-world
shape effects of nano-sized mesoporous silica. They found that application.
the length of the rod determines where the drug will end up It would be most beneficial if theoretical analysis could
and also its clearance rate. For example, short-rods were easily provide the aforementioned understanding, and to allow
trapped in the liver.17–19 precise tuning of the dissolution/toxicity of doped metal oxide
(Cyto)toxicity, if delivered in a controlled and selective NPs. First and foremost, atomistic description of large sections
manner, can be a potent tool for killing cancer cells. This has of NPs is required to handle doping, and numeric energy/inter-
action description is required to assess activity – potency for
dissolution of surface atoms/ions. Unfortunately, most of the
a
Institute of Chemistry, University of Tartu, Ravila 14a, Tartu 50411, Estonia. published nano-QSARs contain descriptors that are inherently
E-mail: karu@ut.ee incapable of such subtlety in NP description.
b
NanoHealth and Optical Imaging Group, Department of Imaging and Pathology, KU
In 2010, Riviere et al.23 proposed biological surface adsorp-
Leuven, B3000 Leuven, Belgium
c
MoSAIC, KU Leuven, B3000 Leuven, Belgium tion indices (BSAI descriptors) to characterize the interactions
d
Aarhus University, Dept Bioscience, Vejlsøvej 25, PO Box 314, 8600 Silkeborg, between chemical groups on the NP surfaces and the amino-
Denmar acid residues of the proteins by quantifying the competitive
e
Institute of Technology, University of Tartu, Nooruse 1, Tartu 50411, Estonia adsorption of a set of small molecule probes onto the nano-
† Electronic supplementary information (ESI) available: Examples of newly calcu-
particles. From there on, several experimental and compu-
lated descriptors for three NPs (0%Fe, 6%Fe and 10%Fe) and the descriptor
matrix for all 133 NPs calculated in this work can be downloaded from http:// tational studies have been carried out for modeling the NP
www.ut.ee/cc/nanodescriptors. See DOI: 10.1039/c8nr05220d surface interactions in biological systems.24–28
This journal is © The Royal Society of Chemistry 2018 Nanoscale, 2018, 10, 21985–21993 | 21985
View Article Online
Paper Nanoscale
While theoretical descriptors for small organic molecules an approach for the NP-specific definition of the shell and
can be readily calculated with various levels of theory to rep- core area of the NPs, as well as surface structure optimization.
resent most molecular features, the size of NPs (not to The previous set of nanodescriptors36 was based on atom
mention the scaling with radius) is the obvious limiting factor. counts, potential energies, and coordination numbers. NP
Therefore, all approaches have had to choose between hand- structures were generated by replicating the unit cells of the
ling full NPs at simpler levels of theory or consider just a small thermodynamically most stable crystal structures and then
fraction of the NPs and use quantum chemical methods. In deleting atoms outside the set radius of NPs. Electroneutrality
2011, Puzyn et al.29 studied the cytotoxicity of 17 metal oxide was achieved by adding an appropriate number of oxygen or
Published on 29 October 2018. Downloaded by University of New England on 1/20/2019 9:06:26 PM.
NPs to the bacteria Escherichia coli, making a presumption metal atoms onto the surface. The drawback of this approach
that the toxicity is independent of the size of the NPs. A corre- was that the lattice parameters of metal oxides were not taken
lation was established between the cytotoxicity and the heat of into account for the added atoms. Thus, two atoms could have
formation for metal cations. Jagiello et al.30 have applied size- ended up too close to each other, or even overlap. This, in
dependent electronic properties as descriptors for nano-QSAR. turn, can adversely affect the calculation of potential energy
However, these parameters were extrapolated from molecular descriptors, particularly for the smallest sized NPs. In the
representation and had no relation to the actual size of the current approach, the necessary amount of oxygen or metal
whole NP, not to mention the surface structure. atoms (calculated from the number of metal and oxygen atoms
Toropov et al.31 used “optimal” descriptors – something and their respective charges) was removed from the surface of
considered as intermediates between classical and nanode- NPs to achieve electroneutrality. Such an approach mitigates
scriptors, which were derived from both the structure of NPs the possibility of two neighboring atoms having a too short or
(represented as SMILES codes) as well as experimental data long bond length or overlap due to random placement of new
related to the substance and the experimental conditions. atoms.
There have been other attempts making use of the SMILES Secondly, the Lennard-Jones potential37 with polarizable
codes in NP modeling.32–34 As NPs are often coated to reduce oxygen atoms was used here for the calculation of energies
agglomeration or to promote transport into cells, studies of instead of the Buckingham potential.38 Main reasons for chan-
NPs coated with organic molecules or metal atoms have been ging the potential were: (a) the possibility of using the same
carried out. QSAR35 representations of those particles have parametrization for multiple metal oxides39 and (b) employing
used the presumption that all the properties of NPs depend polarizable potentials to include descriptors based on oxygen
solely on the structure of the coating, ignoring the underlying polarization and possible extension of this methodology to
NP. It appears that the principal feature of nanomaterials – the include coating molecules, as the Lennard-Jones potentials are
high concentration of atoms with high potential energy – has available for a large number of atoms and atom pairs.
not really been adequately described in any of these studies. The NP structures were optimized using the Polak–Ribiere
version40 of the conjugate gradient algorithm. This is an
important advantage over bulk crystal structure-based geome-
try, as the differences in the bond length in the core and shell
Results and discussion area of the NP are now properly represented. All the above-
mentioned modifications enable modeling of more complex
Background of true nanodescriptors (e.g. coated, doped) NPs and also developing new nanodescrip-
Previously, we have reported a set of full particle nanodescrip- tors based on atom polarization.
tors,36 calculated directly and solely from the atomic structure
of the NPs, not requiring external information apart from the Derivation of nanodescriptors
crystal structure of the respective bulk materials. These nano- In the present approach, the generation of the NP structure for
descriptors were based on a force-field calculation of the computational modeling has been improved. Two novel
potential energies of whole NPs and the core and shell layers, classes of nanodescriptors are introduced based on (i) the
and have been proved effective for modeling, grouping and polarized oxygen atoms and (ii) the force vectors of all atoms.
read-across of metal oxide NP properties and biological activi- To distinguish between the released Zn2+ ions and those
ties.22 While these first true nanodescriptors were a step embodied in the structure of NPs, henceforth, the latter are
forward, they still have shortcomings from the theoretical per- called Zn and O atoms for the sake of clarity. The parametriza-
spective which also affected their potential applications. tion of these properties is crucial as they specifically describe
Firstly, the fixed core and shell region definition (1 nm the potential energy (stability, activity) of the atoms on the
from surface) was rather arbitrary and not entirely adequate surface of NPs.
for the development of high precision descriptors allowing to The polarization of oxygen atoms is calculated using the
distinguish the energies between the surface and non-surface adiabatic core–shell method where a shell oxygen atom has
atoms. Secondly, the atomic surface structure was still carved 1.5 times the charge of the oxygen atoms in the non-polarized
out of the bulk material, again not providing enough accuracy model and the core oxygen has −0.5 of the non-polarized
at the atomic layer level. The present study takes the approach oxygen charge.37 The pairwise dipole moment is calculated
of whole-particle nanodescriptors a step further by specifying by defining the distance between the corresponding core–
21986 | Nanoscale, 2018, 10, 21985–21993 This journal is © The Royal Society of Chemistry 2018
View Article Online
Nanoscale Paper
Fig. 2 Surface region parameter dependence on the size of the surface region (a) potential energy, (b) coordination number, (c) oxygen dipole
moment and (d) force vector.
This journal is © The Royal Society of Chemistry 2018 Nanoscale, 2018, 10, 21985–21993 | 21987
View Article Online
Paper Nanoscale
21988 | Nanoscale, 2018, 10, 21985–21993 This journal is © The Royal Society of Chemistry 2018
View Article Online
Nanoscale Paper
achieve the necessary doping level. This simple approach is while the increase in the Fe content causes the opposite
possible due to the fact that Zn and Fe have the same oxi- effect.
dation state, and adding Fe atoms did not cause a change in To study the systematic behavior of lattice energy depend-
the overall charge of the NP. In the case of different oxidation ing on the size and doping level of NPs, descriptor values were
states or different lattice structure, special care must be taken calculated for 19 different sizes and 7 different doping percen-
for constructing doped NPs. Plots of selected descriptor values tages (133 combinations in total) of ZnO NPs. Fig. 5 displays
vs. the Fe doping percentage are presented in Fig. 4. the size and doping level dependence of the lattice energy
As seen in Fig. 4a and b, potential energy descriptors descriptor.
Published on 29 October 2018. Downloaded by University of New England on 1/20/2019 9:06:26 PM.
exhibit a nearly linear correlation with the Fe content in the Fig. 5 demonstrates that the destabilizing effect, caused by
NP. This is due to the difference in lattice parameters and the the increasing number of doping Fe atoms (or overall NP size),
interatomic potential of the respective oxides. The Fe–O bond is similar for all the studied NPs. This outcome enables the
length in the wurtzite structure is 2.167 Å,42 while the Zn–O development of a two-parameter linear model, which can
bond length in the wurtzite structure is 1.98 Å.43 Therefore, be used to apply the novel nanodescriptors in nano-QSARs
the insertion of Fe atoms into the lattice of ZnO causes local without explicitly performing the force-field calculations on
distortions in the crystal structure and decreases the potential particular NPs. The linear model allows the interpolation of
energy – the lattice energy becomes less negative (Fig. 4a descriptor values based on the radius and amount of Fe
and b). (weight %) in the NP:
Combining the size and doping dependencies of nanode-
scriptors, one can see that the size and Fe content may have 1
D ¼ a þ b Feðwt%Þ þ c ; ð1Þ
a very different effect on their values. In the case of lattice R
energy, the increase of the NP radius stabilizes the NP,
where D is the value of the predicted descriptor, a, b, and c are
equation coefficients, Fe(wt%) is the doping level (weight %)
Fig. 4 Plot of selected descriptor values vs. the Fe doping percentage; Fig. 5 Effect of doping (weight percent) and size of ZnO NPs on the
(A) lattice energy of NPs and (B) average potential energy of oxygen calculated lattice energy and average coordination number of all atom
atoms in the shell region. descriptors.
This journal is © The Royal Society of Chemistry 2018 Nanoscale, 2018, 10, 21985–21993 | 21989
View Article Online
Paper Nanoscale
Table 1 Parameters for interpolation of descriptor values based on the size and composition of NPs
Descriptor a b c R2 R2cv
and R is the radius of NPs in angstroms. The coefficient values As the proposed mechanism of toxicity of Fe doped ZnO
Published on 29 October 2018. Downloaded by University of New England on 1/20/2019 9:06:26 PM.
for three descriptors: the lattice energy of the NP, average nanoparticles is the release of Zn2+ ions from the surface of a
coordination numbers of all atoms and average length of the NP,46 the amount of available Zn on the surface can be
force vector surface normal component for all Zn atoms are expected to correlate with the cell death, membrane damage
presented in Table 1. and mitochondrial ROS. Statistical parameters for these one-
The statistical fit between the calculated and extrapolated parameter equations are presented in Table 2. As visible from
descriptor values is very good, R2 varies between 0.934 and Table 2, the slope for the equations relating the number of Zn
0.998. Leave one out cross-validation (R2cv, Table 1) demon- atoms to biological endpoints is positive, indicating a logical
strating the predictive power of the models was performed for relationship between the descriptor and the measured prop-
these multilinear equations. All of these descriptors depend erty – increasing the number of available Zn atoms on the
on the reciprocal value of the NP radius. This is caused by the surface of NPs also increases the toxicity of the NPs. It is inter-
spherical shape of the NPs, as the surface area to volume ratio esting to note that the statistical parameters for mitochondrial
of spheres is also a function of 1/R. The relationship between ROS are slightly higher than those for cell death. One possible
the NP size and the calculated descriptors is in agreement explanation is that the generation of reactive oxygen species is
with the previously published results.44 It must be noted that the primary outcome from the release of Zn2+ ions from the
some properties in the case of smaller NPs (diameter below surface of NPs, while cell death is the sum of several factors
≈30 nm) show rather abrupt changes as a function of NP and mechanisms resulting from the interaction with NPs.
radius. Therefore, it is crucial to always take the size of NPs While the number of Zn atoms on the surface of the NP
into account when calculating nanodescriptors and using was shown to be important to describe the toxic effects of
these to predict the properties of NPs. doped NPs, the count itself cannot possibly be the only factor
as surface atoms on larger NPs should be relatively less active,
Application of nanodescriptors: Fe-doped ZnO NPs and cancer also, doping may change surface atom activity. Therefore, one
cells also has to take into account the chemical surroundings of the
The developed descriptors were applied in the analysis of the Zn atoms in the NP, e.g. the potential energy of the Zn atoms.
biological effect of doped NPs on the KLN205 and HeLa cell One possible approach to numerically characterize the poten-
lines. The biological data for different parameters of these cell tial energy in a crystalline material is via lattice-energy. To that
lines were obtained from a study by Manshian et al.22 There, end, one-parameter equations for the lattice energy of NP/
high-content based multiparametric screening had been Surface area of NP versus selected biological endpoints were
applied to investigate the biological effects of the Fe-doped generated (Table 3). The inclusion of the surface area in the
ZnO NPs. The cells were exposed to different concentrations of descriptor reduces the role of the size of the NP, and therefore,
the various Fe-doped ZnO NPs for 8 h and then stained and amplifies the effect of doping with other elements, like Fe in
analyzed using a previously validated high-content imaging our case.
methodology optimized for use in bio-nano interaction Again, the mitochondrial ROS correlates better with the
studies.45 In the current study, the following endpoints were lattice energy of the NP/Surface area of the NP than the cell
selected: (i) occurrence of cell death, (ii) membrane damage death or the membrane damage.
and (iii) the induction of reactive oxygen species (30 µg ml−1,
8 h exposure). These three parameters were chosen as they rep-
resent the proposed toxic mechanism of ZnO NPs (induction Table 2 Statistical parameters for the correlations of the three end-
of reactive oxygen species) and the measured effect on the points of the two cell lines with the log number of Zn atoms in the shell
region of the NP
vitality of cells (occurrence of cell death and membrane
damage). While statistically significant linear correlations
Cell line Endpoint (unitless) Slope Intercept R2 R2cv
between the calculated descriptor values and the measured
parameters (7 NPs with different doping weight percent) were HeLa Cell death 1.60 −4.69 0.947 0.837
HeLa Membrane damage 2.44 −7.55 0.969 0.919
established, the small number of data-points does not really HeLa Mitochondrial ROS 3.81 −11.79 0.977 0.962
support focusing on the correlation statistics too much, as is KLN205 Cell death 1.86 −5.63 0.950 0.856
so often the case with NP bioactivity. Therefore, more attention KLN205 Membrane damage 2.62 −8.17 0.965 0.907
KLN205 Mitochondrial ROS 4.01 −12.49 0.965 0.938
should be paid to the mechanistic relationships and descriptor
value trends. Respective plots are presented in ESI, Fig. S1.
21990 | Nanoscale, 2018, 10, 21985–21993 This journal is © The Royal Society of Chemistry 2018
View Article Online
Nanoscale Paper
Table 3 Statistical parameters for the correlations of the three end- lular toxicity correlated with the level of free Zn2+ in solution,22
points of the two cell lines with the descriptor lattice energy of the NP/ it was not possible to draw any correlations on the level of Zn2+
Surface area of the NP
ions still present at the NP surface at different time points.
Until now, both in the literature and in our own work, toxicity
Cell line Endpoint (unitless) Slope Intercept R2 R2cv
has mainly been associated with free Zn2+ ions. Considering
HeLa Cell death 1008 2.44 0.914 0.759 the strong effect of Fe-doping on NP toxicity even at low levels
HeLa Membrane damage 1552 3.34 0.951 0.868
HeLa Mitochondrial ROS 2439 5.21 0.974 0.948 (1–2%), strong correlation with dissolution kinetics,46 and
KLN205 Cell death 1173 2.66 0.917 0.776 assuming a homogeneous distribution of Fe within the ZnO
Published on 29 October 2018. Downloaded by University of New England on 1/20/2019 9:06:26 PM.
KLN205 Membrane damage 1661 3.52 0.941 0.841 NPs, we conclude that the difference in the level of surface-
KLN205 Mitochondrial ROS 2570 5.41 0.963 0.919
associated Zn2+ ions for these conditions can only play a
Respective plots are presented in ESI, Fig. S2. minor role in the overall differences in cellular toxicity
observed for these NP formulations.
For cell-free systems, the concentration of dissolved ions
If we want to represent both the size of the NP and the can be measured as a function of time, and the solubility con-
chemical nature of the surface of metal oxide NPs as a single stant can be determined, but this is not straightforward in cel-
stable descriptor, we should take a look at the species that is lular environments. Here, any predicted solubility constant
always present in any metal oxide NP – oxygen. The biological values would not be very insightful due to differences between
endpoints were correlated against the descriptor average force cells in view of the subcellular compartmentalization of the
vector surface normal component of oxygen atoms in the shell NPs and any associated ions that have been released. This
region (Table 4), as this descriptor characterizes the stability of results in local differences in concentrations of ions and NPs
oxygen atoms on the surface of the NP, (somewhat indirectly) within the same cell and between different cells in a single
describing the activity of all atoms on the surface of the NP. population. Furthermore, these differences in final biological
While the statistical parameters for the correlations with compartmentalization will further influence the kinetics and
Average force vector surface normal component of oxygen extent of ion release due to differences in local pH, and ionic
atoms in the shell region appear slightly lower than those of strength of the microenvironment. Additionally, as zinc is an
the descriptors considered above, the parameters are remark- essential element for cellular homeostasis, zinc distribution is
ably even, suggesting stable performance. The slightly lower linked to the active transport mechanisms,47 which will affect
statistical parameters are likely related to the fact that zero the solubility constant and will, depending on the local con-
doping ( pure ZnO) particles are included in the dataset. centrations, change in time. Therefore, the present models can
only address the Zn release on the NP side without attempting
Interpretation of the model to introduce direct predictions of solubility or concentrations
While the present treatment looks at the toxicity of released in various biological environments due to the lack of experi-
Zn2+, it cannot possibly differentiate between the impact of the mental data that can be obtained to support such predictions.
released Zn2+ ions and that of the surface-associated Zn atoms. The use of parameters associated with Zn2+ release is, at
This limitation largely stems from the lack of experimental present, the best way forward for such cellular environments.
data, as the influence of the two components is very difficult, For numerical comparison between the current nanode-
if not impossible to separate. While some studies replace the scriptors and the previous set,22 one-parameter models of the
NP-released Zn2+ by exogenous Zn2+ ions, this will neither biological endpoints with the respective descriptors of
result in the same exposure kinetics nor in the same distri- different generations were analyzed. The descriptors “the log
bution of Zn2+ ions within the cells (for NP-released Zn2+, number of Zn atoms in the shell region of the NP” and “the
there will be a local high accumulation of ions in the lysoso- lattice energy of the NP/Surface area of the NP” had improved,
mal compartments, which is not the case for added Zn2+ ions). R2 by an average of 0.007 and 0.037 and R2cv by = 0.0150 and
While in previous work, we found that oxidative stress and cel- 0.078, respectively. While the statistical performance improve-
ment for “the log number of Zn atoms in the shell region of
the NP” is solely affected by a more accurate representation of
Table 4 Statistical parameters for the correlations of the three end- the core and shell regions, “the lattice energy of the NP/
points of the two cell lines with the descriptor: average force vector Surface area of the NP” additionally benefited from the surface
surface normal component of oxygen atoms in the shell region
geometry optimization as well as by the replaced interatomic
potential.
Cell line Endpoint (unitless) Slope Intercept R2 R2cv
This journal is © The Royal Society of Chemistry 2018 Nanoscale, 2018, 10, 21985–21993 | 21991
View Article Online
Paper Nanoscale
metal oxide NPs of different sizes. In the present work, a 7 A. Manke, L. Wang and Y. Rojanasakul, BioMed Res. Int.,
number of important improvements to the previous set of full- 2013, 2013, 942916.
particle all-atom theoretical nanodescriptors have been 8 A. Kahru and H.-C. Dubourguier, Toxicology, 2010, 269,
achieved, which, for the first time, allow theoretical insight 105.
into NP characteristics of such detail. Firstly, a methodology 9 I. Passagne, M. Morille, M. Rousset, I. Pujalte and
for calculating the optimal thickness of the (highly active) B. L’Azou, Toxicology, 2012, 299, 112.
surface layer of NPs was described, together with surface atom 10 A. Avalos, A. Isabel Haza, D. Mateo and P. Morales, J. Appl.
position optimization, which enable derivation of new descrip- Toxicol., 2014, 34, 413.
Published on 29 October 2018. Downloaded by University of New England on 1/20/2019 9:06:26 PM.
tors as well as fine-tune those of the previous set to increase 11 M. Misawa and J. Takahashi, Nanomedicine, 2011, 7, 604.
consistency and general applicability. Secondly, a set of new 12 M. Horie, K. Fujita, H. Kato, S. Endoh, K. Nishio,
nanodescriptors was developed, based on the force vectors of L. K. Komaba, A. Nakamura, A. Miyauchi, S. Kinugasa,
all atoms and the “dipole moments” of oxygen atoms, which Y. Hagihara, E. Niki, Y. Yoshida and H. Iwahashi,
allow distinguishing between the regular (bulk) NP atoms and Metallomics, 2012, 4, 350.
those carrying the special properties of NPs – high surface 13 Q. A. Pankhurst, J. Connolly, S. K. Jones and J. Dobson,
activity. J. Phys. D: Appl. Phys., 2003, 36, R167–R181.
To prove the adequacy of these developments, the calcu- 14 R. Gupta and B. Rai, Nanoscale, 2018, 10, 4940.
lated descriptors were used to model the cell death, membrane 15 N. Z. Knezevic and J. O. Durand, Nanoscale, 2015, 7,
damage and mitochondrial ROS of Fe-doped ZnO nano- 2199.
particles for two cancerous cell lines. The new and improved 16 X. L. Huang, L. L. Li, T. L. Liu, N. J. Hao, H. Y. Liu, D. Chen
nanodescriptors were shown to correlate well with the end- and F. Q. Tang, ACS Nano, 2011, 5, 5390.
points, as well as to allow analysis of the mechanisms govern- 17 X. T. Pan, L. X. Bai, H. Wang, Q. Y. Wu, H. Y. Wang, S. Liu,
ing the endpoints for doped metal oxide NPs of different sizes. B. L. Xu, X. H. Shi and H. Y. Liu, Adv. Mater., 2018, 30,
While a Fe(II) doped ZnO system has been studied here, the e1800180.
nanodescriptors and the overall approach are applicable to any 18 T. X. Liang, L. G. Xiao, C. Liu, K. Gao, H. M. Qin, Y. Cao
metal oxide NPs with various dopings. and X. B. Peng, Org. Electron., 2016, 29, 127.
19 J. Bogdan, J. Plawinska-Czarnak and J. Zarzynska,
Nanoscale Res. Lett., 2017, 12, 225.
Conflicts of interest 20 M. I. Setyawati, C. Y. Tay and D. T. Leong, Biomaterials,
2013, 34, 10133.
There are no conflicts to declare. 21 S. L. Chia and D. T. Leong, Heliyon, 2016, 2, 18.
22 B. B. Manshian, S. Pokhrel, U. Himmelreich, K. Tamm,
L. Sikk, A. Fernandez, R. Rallo, T. Tamm, L. Madler and
Acknowledgements S. J. Soenen, Adv. Healthcare Mater., 2017, 6, 11.
23 X. R. Xia, N. A. Monteiro-Riviere and J. E. Riviere, Nat.
This material is based upon work supported by the EU Nanotechnol., 2010, 5, 671.
Commission (MODERN, Contract no. 309314), Horizon 2020 24 X. R. Xia, N. A. Monteiro-Riviere, S. Mathur, X. F. Song,
ERC to SJS (ERC StG no. 757398), by the institutional research L. S. Xiao, S. J. Oldenberg, B. Fadeel and J. E. Riviere, ACS
funding IUT (IUT20-15) of the Estonian Ministry of Education Nano, 2011, 5, 9074.
and Research and by the Fonds Voor Wetenschappelijk 25 R. Chen, Y. Zhang, F. D. Sahneh, C. M. Scoglio,
Onderzoek (FWO-Vlaanderen; KAN to BM and SJS). W. Wohlleben, A. Haase, N. A. Monteiro-Riviere and
J. E. Riviere, ACS Nano, 2014, 8, 9446.
26 W. Y. Wang, A. Sedykh, H. N. Sun, L. L. Zhao, D. P. Russo,
Notes and references H. Y. Zhou, B. Yan and H. Zhu, ACS Nano, 2017, 11, 12641.
27 M. Gonzalez-Durruthy, L. C. Alberic, C. Curti, Z. Naal,
1 S. S. Agasti, S. Rana, M.-H. Park, C. K. Kim, C.-C. You and D. T. Atique-Sawazaki, J. M. Vazquez-Naya, H. Gonzalez-Diaz
V. M. Rotello, Adv. Drug Delivery Rev., 2010, 62, 316. and C. R. Munteanu, J. Chem. Inf. Model., 2017, 57, 1029.
2 W. H. De Jong and P. J. A. Borm, Int. J. Nanomed., 2008, 3, 28 M. Gonzalez-Durruthy, A. V. Werhli, L. Cornetet,
133. K. S. Machado, H. Gonzalez-Diaz, W. Wasiliesky,
3 H. S. Huang and J. F. Hainfeld, Int. J. Nanomed., 2013, 8, C. P. Ruas, M. A. Gelesky and J. M. Monserrat, RSC Adv.,
2521. 2016, 6, 58680.
4 A. Nel, T. Xia, L. Madler and N. Li, Science, 2006, 311, 622. 29 T. Puzyn, B. Rasulev, A. Gajewicz, X. Hu, T. P. Dasari,
5 M. Horie, H. Kato, K. Fujita, S. Endoh and H. Iwahashi, A. Michalkova, H.-M. Hwang, A. Toropov, D. Leszczynska
Chem. Res. Toxicol., 2012, 25, 605. and J. Leszczynski, Nat. Nanotechnol., 2011, 6, 175.
6 T. Xia, M. Kovochich, M. Liong, L. Maedler, B. Gilbert, 30 K. Jagiello, B. Chomicz, A. Avramopoulos, A. Gajewicz,
H. Shi, J. I. Yeh, J. I. Zink and A. E. Nel, ACS Nano, 2008, 2, A. Mikolajczyk, P. Bonifassi, M. G. Papadopoulos,
2121. J. Leszczynski and T. Puzyn, Struct. Chem., 2017, 28, 635.
21992 | Nanoscale, 2018, 10, 21985–21993 This journal is © The Royal Society of Chemistry 2018
View Article Online
Nanoscale Paper
31 A. P. Toropova, A. A. Toropov, R. Rallo, D. Leszczynska and 40 E. Polak and G. Ribière, Rev. Francaise Informat. Recherche
J. Leszczynski, Ecotoxicol. Environ. Saf., 2015, 112, 39. Operationelle, 1969, 16, 35.
32 A. A. Toropov, A. P. Toropova, T. Puzyn, E. Benfenati, 41 V. Satopaa, J. Albrecht, D. Irwin and B. Raghavan, 31st
G. Gini, D. Leszczynska and J. Leszczynski, Chemosphere, International Conference on Distributed Computing Systems
2013, 92, 31. Workshops, Minneapolis, MN, 2011, p. 166.
33 T. Puzyn, D. Leszczynska and J. Leszczynski, Small, 2009, 5, 42 M. Boiocchi, F. Caucia, M. Merli, D. Prella and
2494. L. Ungaretti, Eur. J. Mineral., 2001, 13, 871.
34 A. A. Toropov, A. P. Toropova, E. Benfenati, G. Gini, 43 Y. N. Xu and W. Y. Ching, Phys. Rev. B: Condens. Matter
Published on 29 October 2018. Downloaded by University of New England on 1/20/2019 9:06:26 PM.
T. Puzyn, D. Leszczynska and J. Leszczynski, Chemosphere, Mater. Phys., 1993, 48, 4335–4351.
2012, 89, 1098. 44 J. L. Ying, T. Zhang and M. Tang, Nanomaterials, 2015, 5,
35 S. Kar, A. Gajewicz, T. Puzyn and K. Roy, Toxicol. in Vitro, 1620.
2014, 28, 600. 45 B. B. Manshian, U. Himmelreich and S. J. Soenen, Chem.
36 K. Tämm, L. Sikk, J. Burk, R. Rallo, S. Pokhrel, L. Madler, Res. Toxicol., 2017, 30, 595.
J. J. Scott-Fordsmand, P. Burk and T. Tamm, Nanoscale, 46 T. A. Xia, Y. Zhao, T. Sager, S. George, S. Pokhrel, N. Li,
2016, 8, 16243. D. Schoenfeld, H. A. Meng, S. J. Lin, X. Wang, M. Y. Wang,
37 J. E. Jones, Journal, 1924, 463–477. Z. X. Ji, J. I. Zink, L. Madler, V. Castranova, S. Lin and
38 R. A. Buckingham, Proc. R. Soc. London, Ser. A, 1938, 168, 264. A. E. Nel, ACS Nano, 2011, 5, 1223.
39 L. F. Zhao, L. C. Liu and H. Sun, J. Phys. Chem. C, 2007, 47 Q. Lu, H. Haragopal, K. G. Slepchenko, C. Stork and
111, 10610. Y. V. Li, Int. J. Physiol., Pathophysiol. Pharmacol., 2016, 8, 35.
This journal is © The Royal Society of Chemistry 2018 Nanoscale, 2018, 10, 21985–21993 | 21993