DOI: 10.1002/chem.201203003

A Top-Down Synthesis Route to Ultrasmall Multifunctional Gd-Based SilicaNanoparticles for Theranostic Applications

Anna Mignot,[a, b, e] Charles Truillet,[a] FranÅois Lux,*[a] Lucie Sancey,[a] C�dric Louis,[b]

Franck Denat,[c] Fr�d�ric Boschetti,[d] Laura Bocher,[e] Alexandre Gloter,[f]

Odile St�phan,[f] Rodolphe Antoine,[g] Philippe Dugourd,[g] Dominique Luneau,[h]

Ghenadie Novitchi,[h] Leandro Carlos Figueiredo,[i] Paulo Cesar de Morais,[i, r]

Laurent Bonneviot,[j] Belen Albela,[j] FranÅois Ribot,[k, l] Luk Van Lokeren,[l]

Isabelle D�champs-Olivier,[m] FranÅoise Chuburu,[m] Gilles Lemercier,[m]

Christian Villiers,[n, o] Patrice N. Marche,[n, o] G�raldine Le Duc,[p] St�phane Roux,[q]

Olivier Tillement,[a] and Pascal Perriat[e]

Introduction

During the past decade, intense research has been carriedout in the domain of multifunctional nanoparticles for thera-nostic applications. These nanoparticles present the opportu-nity to combine several different features in a single item.They can improve disease diagnosis, enable new treatments,and monitor therapeutic responses. They offer a largenumber of possible combinations between 1) imaging tech-ACHTUNGTRENNUNGniques such as optical imaging, scintigraphy, magnetic reso-nance imaging (MRI), ultrasound-based imaging and X-raycomputed tomography and 2) therapeutic techniques, suchas drug and gene delivery, hyperthermia, photodynamic andneutron therapy or sensitising for radiotherapy.[1–5]

Nanoparticles for theranostic applications need to be bio-compatible, nontoxic, tuneable in size and suitable for effi-cient clearance from the body. Silica nanoparticles are oneof the most popular probes described in the literature.[6]

Multifunctional silica nanoparticles have been reported ascontrast agents for several combinations of imaging modali-ties: optical imaging and positron emission tomography

(PET),[7] optical imaging and MRI,[8–10] and optical imagingwith both MRI and computed tomography.[11] Some studiescombine imaging and therapy: MRI and gene delivery,[12]

optical imaging and drug delivery,[13] and optical imagingand photodynamic therapy.[14,15]

The two most common approaches described in literaturefor silica particle synthesis are the Stçber method and thereverse microemulsion process. The reverse microemulsionprocess provides particles with sizes ranging from 10 nm to1.5 mm with good monodispersity.[16,17] However, only lowyields are obtained, and the nanoparticles prepared by thismethod require long purification steps to remove all the sur-factants before any biological application. On the otherhand, the Stçber method yields particles with sizes rangingfrom 10 nm to 2 mm with rather good monodispersity.[18–20]

With modified Stçber-type syntheses, sizes down to 3–7 nmcan be reached.[21]

Grafting functional groups on the particle’s surface alwaysleads to a size increase. The more functionalities that areadded, the bigger the particles will be. Nanoparticle size is amajor parameter in toxicity concerns. First, particle size

Abstract: New, ultrasmall nanoparticleswith sizes below 5 nm have been ob-tained. These small rigid platforms(SRP) are composed of a polysiloxanematrix with DOTAGA (1,4,7,10-tetra-ACHTUNGTRENNUNGazacyclododecane-1-glutaric anhydride-4,7,10-triacetic acid)–Gd3+ chelates ontheir surface. They have been synthe-ACHTUNGTRENNUNGsised by an original top-down process:1) formation of a gadolinium oxideGd2O3 core, 2) encapsulation in a poly-siloxane shell grafted with DOTAGAligands, 3) dissolution of the gadolini-um oxide core due to chelation ofGd3+ by DOTAGA ligands and4) polyACHTUNGTRENNUNGsiloxane fragmentation. These

nanoparticles have been fully charac-terised using photon correlation spec-troscopy (PCS), transmission electronmicroscopy (TEM), a superconductingquantum interference device (SQUID)and electron paramagnetic resonance(EPR) to demonstrate the dissolutionof the oxide core and by inductivelycoupled plasma mass spectrometry(ICP-MS), mass spectrometry, fluores-cence spectroscopy, 29Si solid-stateNMR, 1H NMR and diffusion ordered

spectroscopy (DOSY) to determine thenanoparticle composition. Relaxivitymeasurements gave a longitudinal re-laxivity r1 of 11.9 s�1 mm

�1 per Gd at60 MHz. Finally, potentiometric titra-tions showed that Gd3+ is strongly che-lated to DOTAGA (complexation con-stant logb110 = 24.78) and cellular testsconfirmed the that nanoconstructs hada very low toxicity. Moreover, SRPsare excreted from the body by renalclearance. Their efficiency as contrastagents for MRI has been proved andthey are promising candidates as sensi-tising agents for image-guided radio-therapy.

Keywords: cancer · gadolinium ·nanoparticles · silica · theranostic

Chem. Eur. J. 2013, 00, 0 – 0 � 2013 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim

These are not the final page numbers! ��&1&

FULL PAPER

needs to be kept below 50 nm to avoid undesirable macro-phage uptake. Second, particle size also plays a key role inthe elimination route from the body. Effective renal clear-ance, which seems to be the most reproducible and most ef-ficient route of excretion, is achieved for very small sizesonly. Silica particles with hydrodynamic diameters between3 and 7 nm combine reasonable circulation lifetimes and ef-ficient renal clearance from the body.[22] Similar studies withquantum dots set the upper limit for renal clearance at a hy-drodynamic diameter of 5.5 nm.[23] On the other hand, largerparticles accumulate in the liver and intestines, with pro-longed retention in the body, and may cause toxicity con-cerns due to longer exposures.

To reach such small sizes while conserving multimodality,we attempted to reduce the size of nanoparticles previouslydescribed elsewhere.[24,25] These nanoparticles are made upof a gadolinium oxide core with a polysiloxane shell and aregrafted with poly(ethylene glycol) (PEG) ligands. They pres-ent an average initial size ranging from 11 to 14 nm. Here,our approach was to reduce the polysiloxane shell thickness,in order to decrease the size and to increase the ratio be-tween “active” species for imaging and therapeutic purposes(Gd) and inert matter (polysiloxane). To prevent the leak-

age of any Gd3+ ions from the oxide core, DOTAGA(1,4,7,10-tetraazacycloACHTUNGTRENNUNGdoACHTUNGTRENNUNGdec ACHTUNGTRENNUNGane-1-glutaric anhydride-4,7,10-triacetic acid) was grafted onto the particle surface to collectany dissolved Gd3+ .

Indeed, free Gd3+ is toxic and can be responsible fornephrogenic systemic fibrosis (NSF), a systemic fibrosingdisorder that principally affects the skin, but can also in-volve any tissue in the human body and even result indeath. Cases of NSF have been reported after injection ofGd-containing contrast agents into patients, namely gado-diamide, gadopentetate or gadoversetamide, which arelinear DTPA derivatives.[26] No case of NSF from macrocy-clic DOTA-derivatives (DOTA= 1,4,7,10-tetraazacycloACHTUNGTRENNUNGdo-ACHTUNGTRENNUNGdec ACHTUNGTRENNUNGane-1,4,7,10-tetraacetic acid) has been reported to ourknowledge, thanks to their improved complexation con-stants and slower dissociation kinetics.[27] In the presentstudy, DOTAGA was used instead of the classical DOTAbecause it has five carboxylic acid groups instead of onlyfour. Thanks to this extra group, eight coordination sites willstill remain available for complexing Gd3+ ions after graft-ing onto the nanoparticle surface.

An unexpected top-down phenomenon occurred aftergrafting the DOTAGA; nanoparticles with a hydrodynamic

[a] Dr. A. Mignot, C. Truillet, Dr. F. Lux, Dr. L. Sancey,Prof. O. TillementLaboratoire de Physico-Chimie des Mat�riaux LuminescentsUMR 5620 CNRS-Universit� Claude Bernard Lyon 1Universit� de Lyon, 69622 Villeurbanne Cedex (France)E-mail : Francois.Lux@univ-lyon1.fr

[b] Dr. A. Mignot, Dr. C. LouisNano-H SAS, 38070 Saint Quentin Fallavier (France)

[c] Prof. F. DenatInstitut de Chimie Mol�culaire de l�Universit� de BourgogneUMR CNRS 5260, Universit� de Bourgogne21078 Dijon Cedex (France)

[d] Dr. F. BoschettiCheMatech, 21000 Dijon (France)

[e] Dr. A. Mignot, Dr. L. Bocher, Prof. P. PerriatMat�riaux Ing�nierie et Science, INSA LyonUMR 5510 CNRS, Universit� de Lyon69621 Villeurbanne Cedex (France)

[f] Dr. A. Gloter, Prof. O. St�phanLaboratoire de Physique des Solides, UMR 8502 CNRSUniversit� Paris Sud, 91405 Orsay Cedex (France)

[g] Dr. R. Antoine, Dr. P. DugourdLaboratoire de Spectrom�trie Ionique et Mol�culaireUMR 5579 CNRS, Universit� Claude Bernard Lyon 169622 Villeurbanne Cedex (France)

[h] Prof. D. Luneau, Dr. G. NovitchiLaboratoire des Multimat�riaux et InterfacesUMR 5615 CNRS, Universit� Claude Bernard Lyon 169622 Villeurbanne Cedex (France)

[i] Dr. L. C. Figueiredo, Prof. P. C. de MoraisInstituto de Fisica, Nfflcleo de F�sica AplicadaUniversidade de Brasilia, 70919-970, Bras�lia-DF (Brazil)

[j] Prof. L. Bonneviot, Dr. B. AlbelaLaboratoire de Chimie, UMR 5182 CNRSEcole Normale Sup�rieure, 69384 Lyon Cedex 04 (France)

[k] Dr. F. RibotLaboratoire de Chimie de la Mati�re Condens�e de ParisUMR 7574 CNRS, Coll�ge de France, 75231 Paris Cedex 05 (France)

[l] Dr. F. Ribot, Dr. L. Van LokerenUPMC Universit� Paris 06, UMR 7574Chimie de la Mati�re Condens�e de Paris (France)

[m] Dr. I. D�champs-Olivier, Prof. F. Chuburu, Prof. G. LemercierInstitut de Chimie Mol�culaire de Reims, UMR 7312 CNRSUniversit� de Reims Champagne-Ardenne51687 Reims Cedex 2 (France)

[n] Dr. C. Villiers, Prof. P. N. MarcheInstitut Albert Bonniot, INSERM U823, BP17038042 Grenoble Cedex 9 (France)

[o] Dr. C. Villiers, Prof. P. N. MarcheUniversit� Joseph Fourier, 38042 Grenoble Cedex 9 (France)

[p] Dr. G. Le DucEuropean Synchrotron Radiation FacilityID 17 Biomedical Beamline, BP 220, 38043 Grenoble Cedex (France)

[q] Prof. S. RouxInstitut UTINAM, UMR 6213 CNRSUniversit� de Franche-Comt�, 25030 BesanÅon Cedex (France)

[r] Prof. P. C. de MoraisDepartment of Control Science and EngineeringHuazhong University of Science and TechnologyWuhan 430074 (P.R. China)

[+] European Synchrotron Radiation FacilityID 17 Biomedical Beamline, BP 220, 38043 Grenoble Cedex (France)

[$] Institut UTINAM, UMR 6213 CNRSUniversit� de Franche-Comt�, 25030 BesanÅon Cedex (France)

Supporting information for this article is available on the WWWunder http://dx.doi.org/10.1002/chem.201203003.

www.chemeurj.org � 2013 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim Chem. Eur. J. 0000, 00, 0 – 0

�� These are not the final page numbers!&2&

diameter of 3 nm were obtained.[28] This phenomenon corre-sponds to Gd2O3 core dissolution followed by chelation ofGd3+ ions by DOTAGA ligands and simultaneous fragmen-tation of the polysiloxane shell. These small rigid platforms(SRP) showed no toxicity after injection in mice and effi-cient renal clearance. They have proven their efficiency ascontrast agents for in vivo X-ray and fluorescence imaging,scintigraphy and magnetic resonance imaging.[29] Here, wespecifically investigate the top-down process that occursduring the particle synthesis. Therefore, we characterisedthe SRPs using several complementary techniques. Electronspectroscopy and microscopy, photon correlation spectrosco-py, electron paramagnetic resonance and magnetic suscepti-bility measurements were used to prove the oxide core dis-solution. Mass spectrometry, 29Si solid-state NMR and1H NMR spectroscopy and inductively coupled plasma massspectrometry were performed to determine the nanoparticlecomposition. Finally, potentiometric titrations showed thatno Gd3+ leaching occurs after chelation by DOTAGA, andcellular tests confirmed a very low toxicity of the SRPs ob-tained. Such promising results lead us to test our systems inin vivo therapeutic applications. After injection in gliosarco-ma-bearing rats and treatment with radiotherapy, the surviv-al of the rats was monitored in comparison with, on onehand, untreated rats and, on the other hand, rats treatedwith radiotherapy only.

Results and Discussion

Nanoparticle synthesis

General description : Gadolinium oxide cores (Figure 1 a),formed according to the equation: 2 GdCl3 + 6 NaOH!Gd2O3 +3 H2O+6 NaCl, were synthesised in diethyleneglycol (DEG).[30] DEG was chosen for its high viscosity,which prevents the agglomeration of the growing nanoparti-cles. These cores present an average diameter of 3.5 nm, asconfirmed by photon correlation spectroscopy (PCS) mea-ACHTUNGTRENNUNGsure ACHTUNGTRENNUNGments.

Then, a polysiloxane shell was grown from these cores(Figure 1 b) using hydrolysis–condensation of aminopropyltriethoxysilane (APTES) and tetraethyl orthosilicate(TEOS), leading to a global diameter of 4.5 nm, as deter-mined by PCS measurements (Figure 2 a).

The expected composition of the polysiloxane shell wasSi1O1.7C1.8N0.6H4.8 per Si atom, with regard to the quantitiesintroduced. A ratio of only 2 Si per Gd was chosen here toreduce the thickness of the polysiloxane shell and furtherimprove renal elimination in medical applications (using aratio of 6 Si per Gd leads to a total size between 11 and14 nm).[25] The construct possesses a diameter of 4.5 nm(mean standard deviation (MSD) of 1.8 nm), correspondingto a 50 % yield in the polysiloxane synthesis, if we considera polysiloxane density of 2. The Gd2O3 cores encapsulatedin the polysiloxane shell remain well crystallised with a sub-5 nm diameter, as revealed by the high-resolution transmis-

Figure 1. Scheme of SRP synthesis: a) core synthesis, b) polysiloxane shell synthesis, c) DOTAGA grafting, d) transfer to water, e) core dissolution,f) polysiloxane fragmentation.

Chem. Eur. J. 2013, 00, 0 – 0 � 2013 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim www.chemeurj.org

These are not the final page numbers! ��&3&

FULL PAPERUltrasmall Silica Nanoparticles

sion electron microscopy (HRTEM) image in Figure 2 d. Forinstance, lattice fringes can be clearly assigned to the (222)and (400) planes with inter-reticular distances of approxi-mately 3.1 and 2.8 �, respectively. The fast Fourier trans-form of the imaged area (see inset Figure 2 d) confirms thecrystalline nature of the Gd2O3 cores, corresponding to theexpected cubic crystal structure (space group Ia3). Finally,the polysiloxane shell displayed amine groups on its surfacefor further functionalisation.

In a third step, DOTAGA was grafted to the amine func-tional groups of the core–shell particles, while the nanoparti-cles were still dispersed in DEG (Figure 1 c). Instead ofPEG or diethylene triamine pentaacetic acid (DTPA), theDOTAGA ligands were specifically selected to permit effi-cient chelation of Gd3+ in case of accidental release fromthe core. Moreover, thanks to their carboxylic acid groups,

they make the core–shell particles hydrophilic, conferringthem good colloidal stability in water and biological buffers.

Finally the particles were transferred into water, purifiedand freeze-dried for storage. They can be dispersed again inwater before use; the hydrodynamic diameter obtained afterdispersion was equal to the diameter before freeze-drying,with less than 5 % difference. Colloidal stability was provid-ed by electrostatic repulsion between negatively chargednanoparticles, as reflected by the negative value of the zetapotential, namely �17 mV at pH 7.4 (see Supporting Infor-mation).

Evidence of a top-down process : In order to optimise thenanoparticles� colloidal stability, different amounts ofDOTAGA were introduced in the particle synthesis route,ranging from 0.5 to 3 DOTAGA ligands per Gd atom. Thefinal sizes of the obtained nanoparticles after precipitation,dispersion in water and two days purification were deter-mined by PCS measurements (Figure 3). Surprisingly, the

particle size decreased with increasing amounts ofDOTAGA. For the lowest amount of DOTAGA(0.5 DOTAGA per Gd), the synthesis was non-reproducibleand lead to some agglomeration, most probably because theparticles were not hydrophilic enough. With 0.75 DOTAGA,an average size of 8 nm was obtained (Figure 2 b), which isconsistent with the expected core–shell structure. Indeed,similar syntheses with short PEG chains (PEG250-(COOH)2) instead of DOTAGA gave hydrodynamic diame-ters between 6 and 8 nm. With higher DOTAGA amounts,sizes of around 3 nm (smaller than the initial Gd2O3 cores)were obtained, suggesting a possible new top-down mecha-nism. Based on HRTEM investigations, the addition of0.75 DOTAGA per Gd still preserved the crystalline natureof the Gd2O3 core, as observed in Figure 2 e, in which (400)lattice planes can be distinguished, as well as a size of ap-proximately 3.5 nm in diameter. On the other hand, thehybrid systems obtained with 2 DOTAGA per Gd do notpresent any crystallinity, as shown in Figure 2 f, despite highGd concentrations detected by energy dispersive X-ray spec-

Figure 2. PCS measurements and high-resolution TEM images of theGd2O3-polysilodane core–shell particles a), d) before DOTAGA addition,b), e) after addition of 0.75 DOTAGA/Gd2O3, and c) f) after addition of2DOTAGA/Gd.

Figure 3. Final particle size variation depending on the number of intro-duced ligands: DOTAGA, PEG and DTPA.

www.chemeurj.org � 2013 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim Chem. Eur. J. 0000, 00, 0 – 0

�� These are not the final page numbers!&4&

F. Lux et al.

troscopy (EDX) measurements. Apparently, the size de-crease of the particles is directly correlated with the disap-pearance of the crystalline Gd2O3 core.

The same experiments were also realised by replacing theDOTAGA by DTPA, which is also a chelator for Gd3+ , anda similar decrease in size was observed with increasingamounts of DTPA. On the other hand, when the same ex-periments were realised with PEG, which is not a chelatorfor Gd3+ , no significant evolution in size was observed(Figure 3), and the core crystallinity was not affected (TEMimages not shown). Therefore, it seems that specific chelat-ing ligands for Gd3+ , such as DOTAGA or DTPA, played akey role in this process.

All this experimental evidence leads to the hypothesisthat, in aqueous solutions, the DOTAGA ligands stronglyaccelerated the Gd3+ ion dissolution from the Gd2O3 core.Since the polysiloxane layer was optimised to be rather thin,the presence of possible defects in this layer might favourthe migration of Gd3+ out of the core and these ions beingchelated by the ligands on the particle surface. The thinpoly ACHTUNGTRENNUNGsiloxane shell, bearing no more internal support, mighttherefore collapse on itself and break up into several frag-ments. Small rigid platforms (SRP), with average sizes of3 nm, were hence obtained with no remains of the initialcore–shell structure. The bottom-up synthesis (consisting ofcore formation, encapsulation and functionalisation) turnsinto a top-down route (Figure 1 d–f).

SRP�s visualisation and localisation by electron spectro-mi-croscopy techniques : To confirm the SRP�s structure, theywere also studied by scanning transmission electron micros-ACHTUNGTRENNUNGcopy (STEM) by means of high-angle annular dark-field(HAADF) images combined with electron energy loss spec-troscopy (EELS) using a spectrum imaging approach.HAADF imaging is known for its high sensitivity to compo-sition, since the image intensity varies approximately as Z1.7,giving rise to a “Z-contrast” image. Figure 4 presentsHAADF images of the SRPs, which can be clearly observedat the present resolution. The agglomerated SRPs form athin film either on the carbon layer (Figure 4 a) or self-sup-ported in the sub-micrometer hole of the holey carbon grid(Figure 4 b). High-magnification HAADF imaging makes itpossible to distinguish individual bright contrasts in thenano-size range that arise mainly from the Gd signal contri-bution (Figure 4 c).

However, interpreting such contrasts as individual SRPsonly based on HAADF images remains rather speculative.Combining Z-contrast images with an EELS spectrum imag-ing approach allows us to chemically map the different ele-ments and their relative positions. Here, EELS spectra werecollected at every position of the probe while scanning theregion of interest, typically 20 20 nm. Figure 5 a presentsthe HAADF intensity profile map acquired simultaneouslyduring the experiment. Carbon, silicon, and gadolinium in-tensity maps were obtained by extracting the C-K, Si-L2,3,and Gd-N4,5 edges, respectively (Figure 5 b–d). The Si-L2,3

map shows almost spherical contrast areas the diameters of

which typically span from 1.5 to 2.5 nm, corresponding tothe fragmented polysiloxane. In the upper part of Figure 5,the sample is thin enough so that individual polysiloxanecores can be observed, as confirmed by the Si-L2,3 intensity

Figure 4. STEM observation of SRPs in dark field mode.

Figure 5. a) HAADF signal intensity of the region of interest acquiredsimul ACHTUNGTRENNUNGtaneously with the EELS measurements. b), c) and d) Intensitymaps of the C-K, Si L2,3, and Gd N4,5 edges, respectively. The blue circlesgive an estimation of the fragmented polysiloxane sizes and positions.The red circles indicate the position of individual Gd atoms. The combi-nation of blue (polysiloxane) and red (Gd) circles have also been report-ed in the upper left quarter of the C-K map. Scale bars are 5 nm.

Chem. Eur. J. 2013, 00, 0 – 0 � 2013 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim www.chemeurj.org

These are not the final page numbers! ��&5&

FULL PAPERUltrasmall Silica Nanoparticles

probed between two spheres going to zero (see example ofEELS spectra in Supporting Information). The Gd-N4,5 mappresents very localised peaks, which are correlated with thebright spots of the STEM-HAADF image, as expected,since Gd is the heaviest element present in this system. Thepositions of individual Gd atoms were statistically estimatedand indicated in Figure 5 d and the Supporting Information.Based on the EELS experiment, we systemically found thatevery polysiloxane core is observed with 3 Gd atoms (in theupper part of Figure 5 b, in which Si particles can be separat-ed unambiguously, around 117 Gd atoms are counted in thevicinity of ca. 40 Si particles). This value is lower than theone obtained by other techniques (see section on the deter-mination of the SRP molecular structure), but we presumethat some Gd atoms might be missed while scanning theregion of interest with EELS. The positions of some polysi-loxane cores (blue circles) and of some Gd atoms (red cir-cles) have been indicated over the C-K intensity map in Fig-ure 5 b. The C and Si distributions are clearly anti-correlatedand some Gd atoms can be observed in the C between theparticles, confirming the overall structure of the SRPs as de-picted in Figure 11 a (see below). The Supporting Informa-tion presents an example of the EELS-based elemental lo-calisation of a single SRP by combining the Si-L2,3, Gd-N4,5,and C-K intensity maps displayed in Figure 11 b–d, respec-tively. A fragmented polysiloxane core is surrounded by fiveindividual Gd atoms, while the C is located in the interstitialspace. This clearly illustrates the 2D elemental structure of aSRP as proposed in the present study.

To go now further in the characterisation of the mecha-nisms involved in the top-down synthesis, several characteri-sation experiments have been performed: magnetic suscepti-bility measurements (performed with a superconductingquantum interference device, SQUID) and electron para-magnetic resonance (EPR) measurements will be presentedbelow, to investigate the disappearance of the oxide core.Nanoparticles with 2 DOTAGA per Gd were chosen for allthe following experiments, because they are right on the pla-teau in the size measurements. This should provide bothgood core dissolution and good reproducibility in the syn-theses.

Characterisation of the top-down mechanism

Evidence of core disappearance by susceptibility measure-ments : Magnetic susceptibility measurements were per-formed using a SQUID to validate the absence of any gado-linium oxide core in the SRPs. Two control samples wereused: 3.5 nm sized gadolinium oxide stabilised with PEGand paramagnetic Gd3+–DOTA chelates. All samples wereintroduced to the SQUID in the solid state (freeze-dried so-ACHTUNGTRENNUNGlutions). On one hand, gadolinium oxide showed paramag-netic behaviour at room temperature (RT) and antiferro-magnetic behaviour below the Neel temperature (1.72 K).This low-temperature transition, compared with that ob-tained in bulk oxide (17–18 K), is entirely explained by sizeeffects (mainly reduction of magnetic interactions). On the

other hand, both SRPs and Gd3+–DOTA chelates, the Neeltemperatures of which were 0.11 and 0.13 K, had a behav-iour characteristic of non-interacting Gd3+ ions isolated bydiamagnetic matrices. Only negligible amounts of oxide coreremain then in the SRPs (less than 5 % of the startingamount),[31] confirming TEM observations, in which most ofthe Gd was located in Gd3+–DOTAGA chelates.

Evidence of core disappearance by EPR spectroscopy : Toconfirm the absence of cores, EPR experiments were per-formed on different Gd species: SRPs, Gd3+–DOTA com-plexes and, this time, bulk Gd2O3 (Figure 6). Bulk oxide ex-

hibits a negligible signal as compared to Gd3+–DOTA com-plexes (Figure 6, inset) which should make it possible to dis-tinguish Gd-based cores from SRPs bearing Gd complexesunambiguously. Gd3+–DOTA complexes are characterisedby a g-value of 1.99, a field distance between the two spec-trum extrema, DHpp, of 62 G (Supporting Information) anda slightly asymmetric signal. These values are similar tothose reported in literature, with DHpp depending on themetal-ion concentration and temperature.[32,33] The SRPspresent a similar EPR spectrum as the Gd3+–DOTA com-plexes, meaning that similar species are present, but with alarger DHpp of 100 G. This small difference between the twoEPR spectra indicates the presence of weak magnetic inter-actions between the Gd atoms, which is consistent with theSRP structure in which neighbouring chelates lie on thepoly ACHTUNGTRENNUNGsiloxane surface. The broadening of the EPR line mightbe also related to some dispersion in the surface density ofthe Gd3+–DOTAGA groups, with such a dispersion inducinga distribution of interaction strengths between the neigh-bouring chelates.

Figure 6. X-band EPR of SRP measured at 150 K after ageing in H2O for3 and 20 h, 2 and 7 days. Inset: EPR signal of the Gd3+–DOTA complexand a suspension of micrometric Gd2O3 particles in H2O. Total Gd con-centration was 2.36 mm for each sample.

www.chemeurj.org � 2013 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim Chem. Eur. J. 0000, 00, 0 – 0

�� These are not the final page numbers!&6&

F. Lux et al.

In addition, the evolution of the particles during the puri-fication step (ageing) in water was monitored by EPR spec-troscopy in order to better characterise the disappearance ofthe Gd-based cores. At the beginning of the purification (t=

3 h), the SRP signal increased quickly, indicating a rapid for-mation of Gd3+–DOTAGA chelates after core dissolution.This evolution slowed down significantly after a few hours,and the EPR signal reached a maximum at 48 h (Figure 6).If we consider that the system does not evolve any moreafter 7 days, we find that 13.4 % Gd2O3 is remaining after3 h, 5.5 % Gd2O3 after 20 h, and that the core dissolution iscomplete after 2 days. These results are consistent with bothTEM and magnetic susceptibility measurements showing theabsence of Gd2O3 core after this duration. Finally, theyhelped us to determine and fix a purification time of 2 daysfor our standard particle synthesis.

At this stage, we have demonstrated that the SRPs dis-cussed are likely composed of a polysiloxane matrix, carry-ing, a priori, both DOTAGA and DOTAGA–Gd3+ chelateson their surface. Any crystalline Gd2O3 has disappearedfrom the nanoconstruct so obtained.

Determination of the SRP molecular structure

Global chemical formula of the SRPs obtained by inductivelycoupled plasma mass spectrometry (ICP-MS): To investigatethe SRP chemical composition, chemical analysis was per-formed by ICP-MS. The obtained compositions in Gd, Si, Nand C, in atomic and mass ratios, are given in Table 1.Before functionalisation of the core–shell particle, the mea-ACHTUNGTRENNUNGsure of the Si/N ratio (after purification in DEG againstethanol) showed that the composition of the polysiloxanewas not modified during the hydrolysis–condensation of itsprecursors (60 % APTES, 40 % TEOS before reaction, 60 %R-APTES, 40 % R-TEOS after reaction, the letter R indi-cating that the precursors have reacted so that their chemi-cal formulas have changed (Supporting Information); tosimplify the notations, this letter will be omitted in the fol-lowing). Assuming that the APTES/TEOS ratio will alsoremain constant during the further functionalisation and pu-rification steps (which will be verified later by other tech-ACHTUNGTRENNUNGniques), one easily finds that the SRPs contain around1.5 DOTAGA ligands per Gd (Table 2). This implies thattwo types of DOTAGA are attached on the surface: thosethat are complexed by Gd3+ and the others that are notcomplexed by any ion. Calculations also show that some re-sidual DEG remains in low amounts in the final product(0.65 DEG molecule (C4O3H10) per Gd). HPLC measure-

ments indicated a high purity of the colloidal solution; theresidual DEG is then attached to the SRPs surface. Finally,the analyses performed make it possible to obtain the syn-thesis yield for each group of species: it varies from 7 % forDOTAGA to 19 % for APTES and TEOS (Table 2). In con-clusion, an SRP average composition can be deduced fromall the data (Table 2): Gd1ACHTUNGTRENNUNGAPTES2.3ACHTUNGTRENNUNGTEOS1.5 ACHTUNGTRENNUNGDOTAGA1.5.

Chemical formula of a unique SRP obtained by mass spec-trometry (MS): To obtain the chemical formula of a singleconstruct knowing the global formula given by Tables 1 and2, an exact estimation of its size, or better, its mass is re-quired. Since PCS provides only a hydrodynamic size andTEM observations are not sufficient to delineate ultrasmallhybrid particles with enough precision, we tried to get themolecular weight of a single SRP by electrospray mass spec-trometry measurements. Macromolecules that contain multi-ple groups capable of being ionised (e.g., Gd–DOTAGAcomplexes) can sustain multiple charges so that distributionsof charge states are often observed in the mass-to-chargespectrum. This is observed for SRPs (Figure 7 a), for which aclear distribution of charges is observed in the m/z range be-tween 1000 and 4000 Th. The results obtained can be ex-ploited since the “envelope” of the peaks is stable and therelative abundance of the different peaks is found to be onlyslightly dependent on the electrospray ionization (ESI) MSexperimental conditions. A multiplicative correlation algo-rithm (MCA)[34] was then used to estimate the mass of thenanoparticles (see Figure 7 b, the mass spectrum generated).This spectrum is characterised by a main population at8.5 kDa (MSD of 0.5 kDa) associated to a broader onepeaking at 9.7 kDa (MSD of 0.5 kDa). This mass is consis-tent with all the morphology information obtained untilnow, including the mass deduced from centrifugal filtrationusing membranes (Vivaspin) with different pore sizes (datanot shown).

The result can be further refined (in particular to get in-formation about the internal structure of the SRPs) by usingMS/MS experiments.[35] The technique consists in a first stepby selecting a species in the first m/z spectrum before pro-voking its dissociation by collision with He atoms. The disso-ciated elements are then studied by a second mass analysisgiving a second m/z spectrum. Here, the species correspond-ing to the broad peak centred at m/z=1405 in the spectrumof Figure 7 a (MS spectrum) was chosen: as seen above, it isthe SRP itself with a 6+ charge due to the electrospraytechnique in the positive mode. The second m/z spectrum(MS/MS spectrum) corresponding to the dissociated SRPs isgiven in Figure 7 c. We observed the peak at m/z=1405 cor-responding to the 8.5 kDa SRPs with a charge state of + 6.

Table 1. Chemical composition of the nanoparticles deduced fromICP-MS.

Gd Si N C

measured quantity [mol] 1.0 3.8 8.0 36.6measured quantity [g] 1.0 0.68 0.71 2.79

Table 2. Relative abundance of the elements building up the particles.

APTES TEOS DOTAGA Gd

introduced quantity [mmol] 12 8 20 10measured quantity [mmol] 2.3 1.5 1.5 1synthesis yield [mol %] 19 19 7 10

Chem. Eur. J. 2013, 00, 0 – 0 � 2013 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim www.chemeurj.org

These are not the final page numbers! ��&7&

FULL PAPERUltrasmall Silica Nanoparticles

The peak is not isotopically resolved but presents abundantneutral losses on its left, characteristic of the differentgroups that have been cleaved from the SRP�s surface. Thisfragmentation of “surface groups” has already been ob-served in the case of gold nanoclusters.[36–39] The MS/MSspectrum is also characterised by 1) low-mass singly chargedDOTAGA ligands, complexed or not to Gd3+ , and polysi-loxane fragments made up of APTES and TEOS, and2) two intense doubly and triply charged ions between 1500and 2000 Th (Figure 7 c). Interestingly, these peaks corre-spond to molecular masses of about 3520 and 4910 Da thesum of which is roughly equal to that of the initial SRP, ap-proximately 8500 Da. The two corresponding sub-particlescertainly result from the cleavage of the SRP in two parts,as already observed for CdS nanoclusters.[40,41] Moreover,the sum of the charges of the two sub-particles is equal tothe 6+ charge hypothesised for the SRP if we add onesingly charged molecular degradation product (low-masssingly charged DOTAGA ligand). These two parts cannotbe assigned directly by MS/MS experiments due to a lack ofmass accuracy and resolution of the instrument. HoweverMS/MS/MS experiments performed by successively choosingthese two parts to obtain a third m/z spectrum make it pos-sible to propose a composition of these two parts: Gd3-ACHTUNGTRENNUNGAPTES7ACHTUNGTRENNUNGTEOS5ACHTUNGTRENNUNGDOTAGA4 and Gd4ACHTUNGTRENNUNGAPTES9 ACHTUNGTRENNUNGTEOS6-

ACHTUNGTRENNUNGDOTAGA6. The formula of a single SRP is then Gd7-ACHTUNGTRENNUNGAPTES16ACHTUNGTRENNUNGTEOS11ACHTUNGTRENNUNGDOTAGA10, a formula in perfect agree-ment with both the average mass of 8500 kDa and theglobal chemical formula Gd1ACHTUNGTRENNUNGAPTES2.3ACHTUNGTRENNUNGTEOS1.5ACHTUNGTRENNUNGDOTAGA1.5

(or Gd7 ACHTUNGTRENNUNGAPTES16ACHTUNGTRENNUNGTEOS10.5ACHTUNGTRENNUNGDOTAGA10.5).

Confirmation of two kinds of DOTAGA—complexed ornon-complexed to Gd3+ , measurement by fluorescence titra-tion : The presence of both types of ligands (complexed ornot to Gd3+) is suggested by the formula of the SRP itself,since, according to it, the nanoconstruct contains more li-gands (10) than Gd cations (7). This observation can be alsodirectly verified by the two peaks present in the low m/zrange of the MS spectrum (Figure 8 a), since they corre-spond to two singly charged APTES–DOTAGA species:one non-complexed (594.30 Th) and the other bearing aGd3+ ion (749.22 Th).

To obtain the quantitative proportion of the complexedDOTAGA [Eq. (1)] and compare it to the formula deter-

Figure 8. a) Low m/z part of the mass spectrum recorded on electro-sprayed SRP. b) Quantification of the ratio of free DOTAGA ligands byresolved time luminescence. The intensities of 5D0–

7F1 (squares) and5D0–

7F2 (triangles) transition peaks are plotted as a function of increasingamounts of Eu3+ .

Figure 7. a) Full m/z spectrum obtained after electrospraying a SRP solu-tion, b) spectrum generated after deconvolution with the multiplicativecorrelation algorithm, c) ESI-MS/MS analysis of the peak at m/z (1405�25) Th. Collision-induced dissociation performed in the ion trap with Heas a collision gas and with a normalised collision energy of 50% and30 ms activation time. The inset displays isotopic patterns of singly charg-ed fragment ions.

www.chemeurj.org � 2013 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim Chem. Eur. J. 0000, 00, 0 – 0

�� These are not the final page numbers!&8&

F. Lux et al.

mined by mass spectrometry (p=0.66), resolved time lumi-nescence experiments were performed after the addition ofdifferent amounts of Eu3+ ions to a solution containingSRPs. It is possible to differentiate the non-complexed li-gands from the complexed ones, since only the first are ableto chelate additional Eu3+ ions. This chelation can be easilyfollowed by the correlative change in the Eu luminescence.Indeed, free Eu ions have a luminescence almost completelyquenched by the OH� oscillators present in water, whereasthe complexed Eu3+ ions are prevented from quenching bythe protecting surrounding ligand. Several sets were pre-pared by the addition of different amounts of Eu3+ ions tothe nanoparticles, ranging from 0 to 2 times the Gd content(noted 0 to 200 % on Figure 8 b). Gadolinium content wasmeasured by inductively coupled plasma atomic emissionspectroscopy (ICP-AES). Namely, europium chloride(EuCl3·6 H2O) was dissolved in water, and added to the solu-tions containing the nanoparticles. Because of the H+ re-lease during rare-earth complexation, the pH of the solu-tions was adjusted to pH 6, which is high enough for fastEu3+ complexation, but simultaneously low enough to pre-vent hydroxide formation. The mixtures were stirred for48 h, and the effective complexation of Eu3+ ions byDOTAGA was controlled by MS (see Supporting Informa-tion). Luminescence measurements were made in solutionscontaining 0.1 mm Gd. Eu3+ ions in water were excited at395 nm and that resulted in two emission peaks: 5D0–

7F1 at592 nm and 5D0–

7F2 at 616 nm.

p ¼ ½complexed DOTAGA�½complexed DOTAGA� þ ½non-complexed DOTAGA�

ð1Þ

For small amounts of added Eu3+ (<60 % per Gd), the in-tensities of both emission peaks increase linearly with theamount of Eu3+ , demonstrating the formation of increasingquantities of luminescent DOTAGA–Eu3+ chelates. As ex-pected, for large amounts of added Eu3+ (>80 %), the lumi-nescence intensity reached a plateau (Figure 8 b); the addi-tional Eu3+ remained free (and non-luminescent) in solu-tion. Therefore, we can determine the ratio of freeDOTAGA ligands to be (70 �10)% of the amount of Gdpresent in the particles. In other words, the proportion ofcomplexed DOTAGA is found by this method to be equalto (59 �5) %, in rather good agreement with the value de-duced from ICP-MS ((66�10) %).

Quantitative evaluation of two kinds of DOTAGA—com-plexed or non-complexed to Gd3+ by potentiometric titration :Another way to determine the proportion of complexed li-gands anchored at the periphery of the nanoparticles con-sists of performing potentiometric titrations of these acid–base entities. Direct pH titrations of the nanoparticles, fol-lowed by data modelling with PROTAF software, can leadto the number of protons released during the course of thetitration and then to the number of free DOTAGA ligandson the surface. Indeed, from pH 3 to 12, the complexed li-

gands do not release any H+ , whereas the non-complexedones release 3 H+ . The model was based on the knowledgeof the acidity constants of the non-complexed ligands, whichwere assumed to be similar to the DOTA acidity con-stants.[49] The presence of supplementary ionisable protonsat the surface of the nanoparticle (Si�OH and Si�NH2 pro-tons) has to be taken into account in the calculation. Conse-quently, at the end of the optimisation process, a freeDOTAGA concentration of 0.519 mm was calculated for aglobal Gd3+ concentration of 0.8 mm. These values corre-spond to 65 % of non-complexed DOTAGA per Gd andthen to a proportion p= 61 % of complexed DOTAGA. Thisresult is in a good agreement with both the values deter-mined from luminescence experiments (p=66 %) and ICP-AES (p=59 %).

Characterisation of the polysiloxane structure by means ofnuclear magnetic resonance (NMR) spectroscopy : In orderto better characterise the polysiloxane structure, solid state29Si NMR experiments were performed on the SRPs(Figure 9). To avoid the perturbing contribution of Gd3+ ,

Gd3+was replaced by Lu3+ , which is diamagnetic instead ofparamagnetic. This substitution should not modify thenature of the nanoconstruct obtained (Lu-SRP), since allthe lanthanides present similar properties and reactivity. In-terestingly, Lu3+ and Gd3+ have also similar complexationconstants with DOTA. This explains why all the characteri-sations performed on Lu-SRPs give the same results asthose obtained with (Gd) SRPs (size, zeta potential, etc.).The 29Si NMR signal can be de-convoluted into six contribu-tions that correspond to six different Si environments. Theyare of two main types: CSi ACHTUNGTRENNUNG(OSi)nO3�n and SiACHTUNGTRENNUNG(OSi)mO4�m,commonly labelled Tn and Qm, respectively. Tn species areformed from APTES and Qm from TEOS. The de-convolu-tion gives a T/Q ratio of 62:38, in perfect agreement withthe initial APTES/TEOS ratio of 60/40. Moreover, the con-

Figure 9. Solid-state 29Si NMR spectrum of the Lu-SRP.

Chem. Eur. J. 2013, 00, 0 – 0 � 2013 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim www.chemeurj.org

These are not the final page numbers! ��&9&

FULL PAPERUltrasmall Silica Nanoparticles

densation degree of both species, (SxQm*m/4)/(Sx) and(SyTn*n/3)/(Sy) are both equal to 0.82, suggesting again asimilar reactivity for both precursors under the synthesisconditions used.

Lu-SRPs were also characterised by liquid-state 1H NMRspectroscopy (Figure 10 a). The singlet at d=4.68 ppm corre-sponds to HOD and the multiplets at d=3.63 and 3.54 ppmto residual DEG. According to the results reported for anaminopropyl-based particle functionalised with eightDOTAGA macrocyclic ligands,[42] the following assignmentscan be made for the other broad resonances. The signal atd= 0.65 ppm corresponds to the a-CH2 (directly linked tothe Si) of APTES. The signal at d= 1.55 ppm corresponds tothe b-CH2 of APTES, while the one at d= 1.68 ppm toDOTAGA-conjugated APTES. The shoulder at d=

1.84 ppm is related to one peculiar proton of the glutaricspacer (CHHCHN) between the SRPs and the ligands. Themany overlapping resonances, which appear between d= 2

and 4 ppm, correspond to all the CH2 of the tetraazacyclo-decane unit, all the CH2 of the acetates, the g-CH2 of theAPTES (functionalised or not), and the remaining protonsof the glutaric spacer. The signal at d=8 ppm might be asso-ciated to the amide proton (NHCO) formed during reactionbetween aminopropyl and DOTAGA. Therefore, 1H NMRspectra show that a large part of the APTES has been func-tionalised by DOTAGA, which is in good agreement withthe chemical formula of the SRP: 10.5 DOTAGA for16 APTES.

To go deeper into the characterisation of the SRPs, Lu-SRPs were also characterised by 1H diffusion ordered spec-troscopy (DOSY) NMR spectroscopy. In order to amplifythe signal of free species and appreciate better the propensi-ty of the species to be released, a purification of only oneday (instead of two) was voluntarily performed. The pulsed-field gradient echo experiments gave access to the diffusioncoefficients for the different species, which makes it possibleto obtain the sizes of the objects attached to them.[43,44] The2D 1HDOSY NMR spectrum of Lu-SRPs (Figure 10 b)showed that, except for DEG, all the resonances were asso-ciated to the same diffusion coefficient of around1.10�10 m2 s�1. On the other hand, DEG is associated to twodiffusion coefficients: one equal to that of the other species,namely 1.10�10 m2 s�1, and another one of 7.10�10 m2 s�1.[45]

This means that all the species (except a portion of theDEG) are attached to the SRPs. Indeed, 1.10�10 m2 s�1 corre-sponds to a hydrodynamic diameter of 3.8 nm, very close tothe SRP size obtained by PCS. Also, the observed spread ofthe hydrodynamic size covers values ranging from 2.4 to6.4 nm, again in very good agreement with the size distribu-tion estimated from PCS (Figure 2 c). Concerning DEG, aportion attaches to SRPs later, whereas an approximatelyequivalent portion remains free in solution, indicating thatDEG is the only species needing more than one day of puri-fication. The other species belonging to the SRPs seem, incontrast, to be very strongly attached to the nanoconstruct.This indicates that the SRPs keep their integrity in solution,which is of great interest for biomedical applications.

Finally, taking into account all the results obtained untilnow (and particularly the proportions of each Tn and Qm

species) a global molecular scheme can be proposed for asingle SRP (Figure 11 a). It appears that the average size of3 nm is in very good agreement with the average mass of8500 kDa found by MS. Moreover, the scheme correspondsalmost perfectly to an SRP imaged by EELS cartography(Figure 11 b–d), the SRP observed presenting five ligandsbearing Gd.

Biomedical applications

Gd enrichment of SRPs : In vivo radiotherapy and magneticresonance imaging (MRI) applications require that theagent used (here the SRP) possesses the highest possiblecontent of Gd. To achieve this, the 40 % non-complexed li-gands were chelated with additional Gd3+ ions by adding anexcess of Gd3+ and stirring for 24 h at room temperature at

Figure 10. a) 1H NMR spectrum of Lu-SRP dispersed in D2O.b) 2D 1H DOSY NMR spectrum of Lu-SRP dispersed in D2O (D=

100 ms and d =5.5 ms).

www.chemeurj.org � 2013 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim Chem. Eur. J. 0000, 00, 0 – 0

�� These are not the final page numbers!&10&

F. Lux et al.

pH 6 just before purification. The verification that all thenon-complexed ligands were effectively chelated was madeby relaxometry and fluorescence titration experiments.Moreover, ICP-MS analyses showed that, during the stirringstep under an excess of Gd3+ ions, the quantity of ligandscomplexed by Gd3+ varies effectively from 60 (Gd/Si=

25 %) to �100 % (Gd/Si =39 %). In all the following sec-tions, enriched SRPs will be used except for potentiometrictitration, for which the presence of non-complexed ligands isrequired.

Preliminary toxicity studies : To demonstrate the absence oftoxicity from such SRPs, several types of experiments wereperformed. With the aim to prevent any Gd3+ release in invivo applications, we verified first that the DOTAGA graft-ed on SRPs has approximately the same complexation abili-ty as free DOTAs, which are considered the most complex-ing ligands for Gd3+ .

The complexation constants of different complexes[logb110, Eq. (2) in the experimental section] were obtainedby means of potentiometric measurements. To take into ac-count that the complexation process of macrocyclic ligandsat low pH is slow, the kinetic process was accelerated by in-cubating at 37 8C for six weeks. Data modelling using aseries of 96 points between pH 2.4 and 4.5 (Figure 12 a) ledto logb110 =24.78 for the SRPs. This value indicated that thecomplexation of Gd3+ by the SRP ligands is, for pH>3,almost as efficient as by DOTA (logb110 = 25.58).[37] For com-parison, b110 is several orders of magnitude higher than foracyclic commercial contrast agents such as Magnevist (Gd–DTPA, logb110 =22.1) or Omniscan (Gd–DTPA–BMA,

logb110 = 16.9). To give a more complete picture of the com-plexation efficiency of the DOTAGA grafted to the SRPsurface, the free-Gd/coordinated-Gd ratio is given over thewhole pH range, for grafted DOTAGA and DOTA (Fig-ure 12 b). As expected, this ratio is very low and less thanone order of magnitude lower for SRPs than for DOTA.This means that the structure of the SRPs ensures strongcomplexation of Gd3+ and should then prevent any releaseof Gd3+ in the body after in vivo injection.

Immunotoxicity was also evaluated using dendritic cellsthat are responsible for the immune response of the bodyand, for this reason, need to survive with all their functional-ities after SRPs internalisation. Dendritic cells were incubat-ed with nanoparticles for 24 h and then labelled using 7-AAD for necrosis detection. Figure 13 a shows that incuba-tion with SRPs up to a Gd concentration of 1 mm has strictlyno effect on the viability of the cells and that necrosis be-comes significant only for the highest concentrations, with aLD50 (quantity required to kill 50% of cells) close to 5 mm

Gd. Moreover, to determine if nanoparticles perturb the

Figure 11. a) Molecular structure of the Gd-based SRP. On average, 27 Siatoms (in yellow) build up the polysiloxane structure. This structureholds 10 DOTAGA ligands, and 7 Gd3+ ions (in green). Some DOTAGAligands remain free from any ion. Carbon atoms are represented in cyan,N in blue, and O in red. b), c) and d) Intensity maps of the Si-L2,3, Gd-N4,5, and C-K edges, respectively. The blue circle highlights one polysilox-ane core. The red circles indicate the position of individual Gd atoms.

Figure 12. a) Speciation diagram of Gd3+/{DOTAGA, Si-RH} system as afunction of pH under the synthesis conditions: [Gd3+]=0.80 mm,[DOTAGA]=1.32 mm and [Si-RH] =5.92 mm. (L stands for theDOTAGA ligands grafted on the SRP. SiRH stands for other surfaceprotons). b) Comparison of Gd3+ affinities for, on the one hand, free(i.e., not attached to particles) DOTA and, on the other hand, DOTAGAligands grafted to SRPs, as a function of pH.

Chem. Eur. J. 2013, 00, 0 – 0 � 2013 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim www.chemeurj.org

These are not the final page numbers! ��&11&

FULL PAPERUltrasmall Silica Nanoparticles

complement system that is part of the innate immunesystem, particles (at 10 mm Gd) were incubated with humanserum at 37 8C for the different times indicated in Fig-ure 13 b. The figure shows that the complement activity issimilar in all the samples regardless of the presence or ab-sence of nanoparticles. This indicates a total absence ofcomplement activation in presence of nanoparticles anddemonstrates that the complement system is absolutely notmodified, even for large doses of particles.

First theranostic applications : The ability of SRPs to behaveas efficient positive contrast agents has been exploited forthe first theranostic application involving SRPs. SRPs wereinjected intravenously ([Gd]3+ =80 mm, 1.4 mL) in gliosar-coma (9LGS)-bearing rats and their accumulation in thetumour, due to the well-known enhanced permeability andretention (EPR) effect, was followed by MRI, as shown inFigure 14 a, b.

Strong highlighting of the tumour could be easily ob-served in the right hemisphere only a few minutes after theinjection of SRPs. The evolution of the MRI signal has alsobeen compared in Figure 14 b over time for the left (healthyzone) and right (tumour) hemispheres. These temporal evo-lutions showed only small values for the Gd content in theleft hemisphere a few minutes after the injection. In the

right hemisphere, the signal was approximately one order ofmagnitude greater, and maximal at about 5 min after the in-jection, with a very small decrease observed 5 min later. Theabsence of retention in the healthy zone confirmed previousexperiments performed using different imaging tech-ACHTUNGTRENNUNGniques.[29] These experiments also showed that, due to theultra-small size of the SRPs, the removal of the SRPs isalmost entirely ensured via renal excretion.

After having determined by MRI studies the time evolu-tion of the Gd in each hemisphere, we decided to proceedwith a first irradiation test 1 h after the injection of the par-ticles, in order to have a very high ratio between the Gdcontent in the right hemisphere (with tumour) and in thecontralateral hemisphere (without tumour). Rats bearing9LGS were then exposed to microbeam radiation therapy(MRT) at ESRF 1 h after the injection and their survival

Figure 13. a) Dendritic cells were incubated at 37 8C in the presence ofSRP (0.1 to 10 mm Gd). Then, after washing, cell death was determinedby the incorporation of 7-AAD, as measured by flow cytometry. Thisfigure is representative of three independent experiments. b) Comple-ment activity was measured on human serum incubated with SRP (10 mm

Gd concentration), for various amounts of time. Complement inactiva-tion was also controlled by incubation at 56 8C for 30 min (conditionsknown to completely inactive ate the complement).

Figure 14. a) T1 weighted image of the brain of a 9 LGS-bearing rat5 min after the injection of SRP. Red area: localisation in the healthyzone, green area: localisation in the tumoral zone. b) Temporal evolutionof the MRI signal in tumour and in normal tissue in the left hemisphere.c) Comparison of survival curves obtained on 9LGS-bearing rats withouttreatment, only treated by MRT and treated by MRT 1 hour after SRPinjection during the 175 days following tumour implantation. The MRTirradiation was done in cross-fired mode, using 50 mm wide microbeams,with a spacing of 211 mm between the beams. The skin entrance dose wasset up at 400 Gy for the peak and 20 Gy for the valley.

www.chemeurj.org � 2013 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim Chem. Eur. J. 0000, 00, 0 – 0

�� These are not the final page numbers!&12&

F. Lux et al.

was compared to that of rats that were either not treated ortreated with MRT only. The untreated rats (n=5) had amedian survival time of 20 days and the rats treated onlywith MRT (n= 7) presented a median survival time of48 days. On the other hand, the median survival time wasextended to 58 days (n=7) when MRT was combined withSRP injection (Figure 14 c). This experiment suggests an ef-ficient radiosensitising effect of the SRPs, in good agree-ment with other studies (in vitro and in vivo) performed onGd-based particles.[46,47] The nanoparticles have proven theireffect at even very small Gd concentrations (less than 1 ppmin the tumour). Finally, the possibility to guide the therapyby MRI is essential here to determine the best conditionsfor launching the X-rays. Indeed, we confirmed here that itis necessary to have the lowest dose as possible in thehealthy region, while simultaneously keeping the highest Gdcontent in the tumour zone to obtain the best therapeuticresults. This compromise, obtained by imaging, makes it pos-sible to optimise the therapy conditions.

Conclusion

This study describes the synthesis and characterisation ofultra-small polysiloxane nanoparticles displayingDOTAGA–Gd3+ groups on their surface. From previouswork, it was already emphasised that these platforms werethe first multifunctional silica-based particles that, in spiteof some kidney retention, are sufficiently small to escapehepatic clearance and enable animal imaging by four com-plementary techniques. Here, it is carefully demonstratedthat these nanoconstructs are obtained by a new top-downprocess; this method makes it possible to decrease the parti-cle size and simultaneously conserve multimodality. Theprocess consists of 1) formation of a gadolinium oxideGd2O3 core, 2) encapsulation in a polysiloxane shell graftedwith DOTAGA ligands, 3) dissolution of the gadoliniumoxide core by chelation of the Gd3+ by the DOTAGA li-gands and 4) polysiloxane fragmentation. We were able tooptimise the synthesis to obtain a narrow size distributionand complete core dissolution. Also, the SRPs can be rein-forced with rare-earth cations or radionuclides by furtherchelation from surface DOTAGA to enhance their contrastproperties. After detailed characterisation by several com-plementary techniques, we were able to set up the precisemolecular structure of SRPs. Potentiometric titrationsshowed that no leaching of Gd3+ could occur (complexationconstant logb110 =24.78) and cellular tests confirmed a verylow toxicity of the nanoconstructs. Finally, in vivo experi-ments with rats showed that the particles can accumulate intumours by the EPR effect, allowing for efficient radiothera-py guided by MRI. For all these reasons, SRPs certainlyconstitute a major advance for more selective therapy andpersonalised medicine.

Experimental Section

Chemicals : Gadolinium chloride hexahydrate (GdCl3·6H2O, 99 %), euro-pium chloride hexahydrate (EuCl3·6H2O, 99 %), lutetium chloride hexa-hydrate (LuCl3·6H2O, 99 %), sodium hydroxide (NaOH, 99.99 %), tet-raethyl orthosilicate (SiACHTUNGTRENNUNG(OC2H5)4, TEOS, 98%), aminopropyltriethoxysi-lane (H2N ACHTUNGTRENNUNG(CH2)3-Si ACHTUNGTRENNUNG(OC2H5)3, APTES, 99%), triethylamine (TEA,99.5 %), and anhydrous dimethyl sulfoxide (DMSO, 99.5 %) were pur-chased from Aldrich Chemicals (France). Diethylene glycol (DEG, 99%)was purchased from SDS Carlo Erba (France). Acetone (reagent grade)was purchased from Sodipro (France) and was used as received. 1,4,7,10-Tetraazacyclododecane-1-glutaric anhydride-4,7,10-triacetic acid(DOTAGA) was provided by Chematech (France). For the preparationof aqueous nanoparticle suspensions, milli-Q water (1>18 MW) wasused.

Synthesis of gadolinium oxide cores embedded in a polysiloxane shell :These particles were obtained by a two-step route. First, gadoliniumoxide cores were synthesised by addition of soda on gadolinium salts, andthen polysiloxane shell growth was induced by hydrolysis–condensationof convenient silane precursors.

Preparation of gadolinium oxide cores : A first solution was prepared bydissolving GdCl3·6 H2O (5.58 g, 15 mmol) in DEG (500 mL) at room tem-perature (RT), leading to a Gd concentration of 30 mm. A second solu-tion was prepared by adding 4.95 mL of a 10 m aqueous solution ofNaOH (49.5 mmol) to 500 mL DEG. The second solution was progres-sively added to the first one, at RT, for 24 h. A transparent colloid ofgad ACHTUNGTRENNUNGolinium oxide nanoparticles in DEG was obtained. The Gd concen-tration of this colloidal solution was 15 mm.

Encapsulation of Gd2O3 cores by polysiloxane shell : Polysiloxane shellswere grown around the Gd cores, by a sol–gel route, using two silane pre-cursors: APTES and TEOS in a 60:40 molar ratio (TEOS leads to SiO2

and APTES to SiO1.5ACHTUNGTRENNUNG(CH2)3ACHTUNGTRENNUNG(NH2)). The reaction was performed in DEGat 40 8C by several consecutive additions of APTES and TEOS, and wascatalysed by TEA. Namely, APTES (1050 mL, 4.5 mmol) and TEOS(670 mL, 3 mmol) were added to the previously prepared solution ofGd2O3 (1 L) in DEG under stirring at 40 8C. After 1 h, a DEG solution(2550 mL, 0.1 m of TEA, 10m of water) was added. The whole coatingprocedure was repeated three more times, every 24 h. The final mixturewas stirred for 48 h at 40 8C. At this stage, 1.2 APTES per Gd and0.8TEOS per Gd were added in total. The obtained solution could bestored at RT for weeks without alteration. The Gd concentration of thiscolloidal solution stayed approximately equal to 15 mm.

Covalent grafting of DOTAGA on hybrid nanoparticles : Finally, a largeexcess of anhydrous DOTAGA was added to the core–shell particles forcovalent grafting, with a ratio of 2 DOTAGA per Gd atom. Namely,DOTAGA (13.76 g, 30 mmol) were dispersed in DMSO (200 mL). Thesuspension of nanoparticles was added to this DOTAGA solution andthe resulting mixture was stirred for 48 h at RT.

Purification : The nanoparticles were then precipitated in acetone (9 L).The acetone was removed; the nanoparticles were washed two moretimes in acetone (3 L) and precipitated again by centrifugation. Finally,the nanoparticles were dispersed in water (200 mL), the remaining ace-tone was removed with a rotary evaporator at RT and the solution wasstirred for one night at RT. Purification of the nanoparticles was per-formed by filtration through Vivaspin membranes (MWCO=5 kDa) pur-chased from Sartorius Stedim Biotech (France). The colloidal solutionwas introduced into 20 mL Vivaspin tubes, and centrifuged. This step wasrepeated several times, by filling the tubes with water and centrifugingagain, until the desired purification rate was reached (at least 1000).Then, the solution was filtered through 0.2 mm syringe filters to removethe largest impurities. The solution was freeze-dried for storage, using aChrist Alpha 1-2 lyophilisator.

Electron spectroscopy and microscopy techniques : Transmission electronmicroscope (TEM) investigations were carried out using a JEOL 2010microscope operating at 200 kV to study the morphologies of the syn-thesised hybrid systems. The spatial resolution of the microscope is 1.7 �.Studies of the SRPs were also performed on a scanning transmission

Chem. Eur. J. 2013, 00, 0 – 0 � 2013 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim www.chemeurj.org

These are not the final page numbers! ��&13&

FULL PAPERUltrasmall Silica Nanoparticles

electron microscope (STEM) using a VG-HB501 STEM operated at100 kV with an electron probe size of approximately 1 nm. Core-levelelectron energy-loss spectroscopy (EELS) experiments were performedsimultaneous to imaging the SRPs to chemically map the elements of in-terest. The EEL spectrometer provides an energy resolution of 400 meVin the core-loss region. Elemental maps were performed with an energydispersion of 0.3 eV per channel and an acquisition time of 1 ms perspectrum. The sample holder was maintained at 150 K using a liquid-ni-trogen-cooled cryo-stage mainly to reduce carbon contamination andbeam damage on the organic layer.

Photon correlation spectroscopy (PCS) size measurement : Direct mea-ACHTUNGTRENNUNGsurements of the size distribution of the nanoparticles suspended in anymedium were performed with a Zetasizer Nano-S PCS (633 nm He-Nelaser) from Malvern Instruments (resolution 0.5 nm). Measurementswere directly taken on the colloid after core synthesis, shell synthesis andafter surface modification with DOTAGA and purification. The Gd con-centration in the measured solution was always 15 mm.

Inductively coupled plasma-atomic emission spectrometry (ICP-AES)analysis : Determination of the Gd content in a sample was performed byICP-AES analysis (with a Varian 710-ES spectrometer). Before measur-ing the Gd concentration, samples of the colloidal suspension were dis-solved in concentrated nitric acid, heated for 3 h at 80 8C, and left onenight at RT. The samples were then diluted with water until the nitricacid concentration reached 5%. Chemical analyses were also performedon the as-prepared samples at the Service Central d�Analyses du CNRS(Solaize, France) using ICP-MS, and enabled determination of the Gd, C,N and Si contents to a precision of 0.3%.

Fluorescence measurements : Fluorescence measurements were carriedout using a Varian Cary Eclipse fluorescence spectrophotometer, in theresolved time mode. The parameters used were: 395 nm excitation wave-length, which is a characteristic excitation for Eu3+ ions, excitation slit20 nm, emission slit 10 nm, delay time 0.1 ms and gate time 5 ms.

Mass spectrometry : Full scan mass experiments were performed using alinear quadrupole ion trap mass spectrometer (LTQ, Thermo Fisher Sci-entific, San Jose, CA) with enlargement for the high 2000–4000 Th range.The nanoparticle suspension was electrosprayed at a flow rate of5 mLmin�1 in positive ion mode. For collision-induced dissociation (CID)experiments, He was used as collision gas. The m/z isolation window was�25 Th. Isotopic distributions of fragment ions were recorded using thezoom scan mode of the LTQ ion trap mass spectrometer.

Solid-state 29Si nuclear magnetic resonance (NMR) spectroscopy : Solidstate 29Si NMR experiments were performed with a Bruker Avance500WB spectrometer, with a MAS 4 mm N-P/F-H probe, at a MAS rateof 5 kHz, a 2 s repetition delay and a flip angle of 308. Spectra were col-lected for 4 days. To avoid magnetic interactions, Lu replaced Gd in thecore synthesis.1H NMR and diffusion ordered spectroscopy (DOSY): All experimentswere performed at 298 K, without spinning, on a Bruker AvanceIII 300spectrometer equipped with a 5 mm BBFO probe, the z-gradient coil ofwhich is able to provide a maximum gradient strength of 49.8 Gcm�1.30 mg of lyophilised SRP, prepared with Lu instead of Gd to avoid para-magnetic effects, were dispersed in 1 mL of D2O. Standard 1H NMR ex-periments were acquired with a presaturation sequence (zgpr) to mini-mise the residual signal of HOD (d =4.68 ppm). 1H DOSY experimentswere acquired with a stimulated echo sequence that includes a longitudi-nal eddy current delay (Te=5 ms), bipolar sine-shaped gradient pulses ofamplitude G, two sine-shaped spoil gradients and presaturation of theHOD signal (ledbpgppr2s). 32 increments, each with 128 transients, weretaken between 1 and 45 Gcm�1 (linear spacing). The diffusion delay (D)and gradient pulse length (d/2) were chosen to achieve at least a fivefoldattenuation. The following values (D, d/2), both in ms, were used: (100,2.75) and (200, 2.00). 1H DOSY data were processed with the Giga-Maxent software interfaced with Topspin.[48] The reported hydrodynamicdiameters (DH) are simply derived from the diffusion coefficients (D)with the well-known Stokes–Einstein formula: DH =kBT/3phD, in whichkB is the Boltzmann constant, T the absolute temperature, and h the vis-cosity of the solvent (1.13 cP for D2O at 298 K).

Potentiometric measurements : Potentiometric titrations were carried outwith an automatic titrator composed of a microprocessor burette Met-rohm dosimat 665 and a Metrohm 713 pH meter connected to a comput-er. All measurements were performed within a thermo-regulated cell at(25.0�0.1) 8C under an Ar stream to prevent dissolution of the CO2. Theionic strength was adjusted to 0.1 with NMe4Cl. The combined Type “U”glass Metrohm electrode used had a very low alkaline error. The proce-dures and apparatus used for protometric measurements have been pre-viously described.[49] The ionic product of water was determined by titra-tion of acetic acid with a CO2-free NMe4OH solution (pKw =13.78 at(25.0�0.1) 8C in 0.1m of NMe4Cl). The determination of the complexa-tion ability of the SRP towards Gd3+ was performed according to the“batch method”.[49] The mother solutions of SRP (0.4 mm in HCl) andGd3+ (5 mm), were mixed according to different Gd3+/SRP stoichiomet-ric ratios (I=0.1). For each ratio, a series of 24 stopped flasks was pre-pared. Each flask corresponded to a pH value, obtained by micro-addi-tion of NMe4OH (50 mm, I=0.1). All the flasks were stored under Arand put for six weeks in a thermo-regulated enclosure at 37 8C. All thecolloidal suspensions were stable for this specified period. Before pHmeasurements, these solutions were finally allowed to reach equilibriumtemperature (25 8C) for 48 h. The protometric data were processed byusing the PROTAF program,[50 ]in order to obtain the best chemicalmodel fit and refined overall constants bmlh [Eq. (2)]

mMþ lLþ hHþ ÐMmLlHh

bmlh ¼MmLlHh½ �

M½ �m L½ �l Hþ½ �hð2Þ

Electron paramagnetic resonance (EPR) spectroscopy : EPR measure-ments were carried out in H2O–DEG (95:5, vol %) at 150 K using aBruker Elexsys 500 spectrometer with X-band microwave (n�9.4 GHz).The modulation amplitude was 5 G and the microwave power 0.2 mW.The EPR line width DHpp was determined as the peak–peak distance be-tween the extreme points of the first derivative of the EPR signal.

Superconducting quantum interference device (SQUID): Magnetic sus-ceptibility measurements were performed using a SQUID MPMS-XL5Quantum Design magnetometer, with a temperature scale from 1.9 K to350 K.

Dendritic cells : Dendritic cells were generated from bone marrow ex-tracted from C57BL/6 mice (Janvier, Le Genest St Isle, France) as previ-ously described.[51] Briefly, bone marrow cells were isolated by flushingfrom the femurs. Erythrocytes and GR1 positive cells were removed bymagnetic cell sorting and the remaining negatively sorted cells were re-suspended at 5 105 cells mL�1 in complete Iscove�s modified Dubelcco�smedium (IMDM) supplemented with growth factors (GM-CSF, FLT-3Land IL-6) and cultured at 37 8C in the presence of 5% CO2. The transfor-mation of the progenitors into fully active dendritic cells occurs during10 days culture.

Incubation with nanoparticles : After washing, dendritic cells(106 cells mL�1) were incubated in culture medium in 24 well plates for24 h; then nanoparticles were added and cells further incubated at 37 8C.The impact on dendritic cells was tested for various parameters, such asnecrosis or activation. Necrosis was analyzed by flow cytometry after24 h incubation with nanoparticles; cells were stained in the presence of7-amino-actinomycin D (7AAD). The analyses were performed on aFACS LSRII (BD Bioscience) and the results recalculated using FCS Ex-press V3 (De Novo Software). Each experiment was performed at leastthree times with cells issued from different cultures.

Measurement of classical pathway complement activity : The classicalpathway complement activity was measured as previously described.[52]

Briefly, sheep erythrocytes were sensitised with rabbit antiserums tosheep red blood cells. Human serum was incubated at 37 8C in the pres-ence of nanoparticles as indicated in the text, and then 20 mL of thisserum were added to 3 mL of the medium containing the erythrocytes(4 106 cells mL�1). The lysis of red blood cells by complement at 37 8Cwas monitored at 660 nm using a spectrophotometer (Evolution201,Thermo Scientific). The time required for 50 % hemolysis (TH50) wasused for the determination of complement activity.

www.chemeurj.org � 2013 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim Chem. Eur. J. 0000, 00, 0 – 0

�� These are not the final page numbers!&14&

F. Lux et al.

Magnetic resonance imaging (MRI) and microbeam radiation therapy(MRT) experiments : The materials and methods related to MRI andMRT experiments are given in Supporting Information. All operativeprocedures related to animal care strictly conformed to the Guidelines ofthe French Government with licenses 380325 and A3818510002.

Acknowledgements

Thanks are due to CMST COST Action TD 1004: “Theragnostics imag-ing and therapy: an action to develop novel nanosized systems for imag-ing-guided drug delivery”; CMST COST Action D38 “ Metal-Based Sys-tems for Molecular Imaging Applications.” CAPES-COFECUB for con-tract no. Ph 645/09; ANR TheraGuIma “Nanoparticules multifonction-nelles pour la radioth�rapie guid�e par imagerie”. Thanks are also due toCLYM for access to microscope 2010F.

[1] R. A. Petros, J. Desimone, Nat. Rev. Drug Discovery 2010, 9, 615 –627.

[2] A. G. Cuenca, H. Jiang, S. N. Hochwald, M. Delano, W. G. Cance,S. R. Grobmyer, Cancer 2006, 107, 459.

[3] A. Louie, Chem. Rev. 2010, 110, 3146 – 3195.[4] H.-C. Huang, S. Barua, G. Sharma, S. K. Dey, K. Rege, J. Controlled

Release 2011, 155, 344.[5] J. Kim, Y. Piao, T. Hyeon, Chem. Soc. Rev. 2009, 38, 372 – 390.[6] L. Wang, W. Zhao, W. Tan, Nano Res. 2008, 1, 99.[7] R. Kumar, I. Roy, T. Y. Ohulchanskky, L. A. Vathy, E. J. Bergey, M.

Sajjad, P. N. Prasad, ACS Nano 2010, 4, 699.[8] J.-K. Hsiao, C.-P. Tsai, T.-H. Chung, Y. Hung, M. Yao, H.-M. Liu,

C.-Y. Mou, C.-S. Yang, Y.-C. Chen, D.-M. Huang, Small 2008, 4,1445.

[9] W. J. Rieter, J. S. Kim, K. M. L. Taylor, H. An, W. Lin, T. Tarrant,W. Lin, Angew. Chem. 2007, 119, 3754 –3756; Angew. Chem. Int. Ed.2007, 46, 3680.

[10] J. S. Kim, W. J. Rieter, K. M. L. Taylor, H. An, W. Lin, W. Lin, J.Am. Chem. Soc. 2007, 129, 8962.

[11] S. Santra, R. P. Bagwe, D. Dutta, J. T. Stanley, G. A. Walter, W. Tan,B. M. Moudgil, R. A. Mericle, Adv. Mater. 2005, 17, 2165.

[12] G. Bhakta, R. K. Sharma, N. Gupta, S. Cool, V. Nurcombe, A.Maitra, Nanomedicine 2011, 7, 472.

[13] J.-H. Park, L. Gu, G. Von Maltzahn, E. Ruoslahti, S. N. Bhatia, M. J.Sailor, Nat. Mater. 2009, 8, 331.

[14] X. He, X. Wu, K. Wang, B. Shi, L. Hai, Biomaterials 2009, 30, 5601.[15] S. Kim, T. Y. Ohulchanskyy, H. E. Pudavar, R. K. Pandey, P. N.

Prasad, J. Am. Chem. Soc. 2007, 129, 2669.[16] K. Osseo-Asare, F. J. Arriagada, Colloids Surf. 1990, 50, 321.[17] H. Yamauchi, T. Ishikawa, S. Kondo, Colloids Surf. 1989, 37, 71.[18] W. Stçber, A. Fink, E. Bohn, J. Colloid Interface Sci. 1968, 26, 62.[19] G. H. Bogush, M. A. Tracy, C. F. Zukoski IV, J. Non-Cryst. Solids

1988, 104, 95.[20] D. L. Green, J. S. Lin, Y.-F. Lam, M. Z.-C. Hu, D. W. Schaefer, M. T.

Harris, J. Colloid Interface Sci. 2003, 266, 346.[21] M. Benezra, O. Penate-Medina, P. B. Zanzonico, D. Schaer, H. Ow,

A. Burns, E. Destanchina, V. Longo, E. Herz, S. Iyer, J. Wolchok,S. M. Larson, U. Wiesner, M. S. Bradbury, J. Clin. Invest. 2011, 121,2768.

[22] A. A. Burns, J. Vider, H. Ow, E. Herz, O. Penate-Medina, M. Baum-gart, S. M. Larson, U. Wiesner, M. Bradbury, Nano Lett. 2009, 9,442.

[23] H. S. Choi, W. Liu, P. Misra, E. Tanaka, J. P. Zimmer, B. I. Ipe,M. G. Bawendi, J. V. Frangioni, Nat. Biotechnol. 2007, 25, 1165.

[24] J.-L. Bridot, A.-C. Faure, S. Laurent, C. Riviere, C. Billotey, B.Hiba, M. Janier, V. Josserand, J.-L. Coll, L. Vander Elst, R. Muller,S. Roux, P. Perriat, O. Tillement, J. Am. Chem. Soc. 2007, 129, 5076.

[25] A.-C. Faure, S. Dufort, V. Josserand, P. Perriat, J.-L. Coll, S. Roux,O. Tillement, Small 2009, 5, 2565.

[26] D. R. Broome, Eur. J. Radiol. 2008, 66, 230 –234.[27] A. E. Merbach, E. Toth, The Chemistry of Contrast Agents in Medi-

cal Magnetic Resonance Imaging, Wiley, Chichester, 2001.[28] F. Lux, O. Tillement, M. Saint Jean, P. Mowat, P. Perriat, S. Roux,

A. Mignot, Ultrafine nanoparticles comprising a functionalised poly-organosiloxane matrix and including metal complexes; method forobtaining same and uses thereof in medical imaging and/or therapy,WO/2011/135101.

[29] F. Lux, A. Mignot, P. Mowat, C. Louis, S. Dufort, C. Bernhard, F.Denat, F. Boschetti, C. Brunet, R. Antoine, P. Dugourd, S. Laurent,L. Vander Elst, R. Muller, L. Sancey, V. Josserand, J.-L. Coll, V.Stupar, E. Barbier, C. Remy, A. Broisat, C. Ghezzi, G. Le Duc, S.Roux, P. Perriat, O. Tillement, Angew. Chem. 2011, 123, 12507;Angew. Chem. Int. Ed. 2011, 50, 12299.

[30] C. Louis, R. Bazzi, C. A. Marquette, J-L. Bridot, S. Roux, G.Ledoux, B. Mercier, L. Blum, P. Perriat, O. Tillement, Chem. Mater.2005, 17, 1673.

[31] B. Mutelet, N. Keller, S. Roux, M. A. Flores-Gonzales, F. Lux, M.Martini, O. Tillement, C. Billotey, M. Janier, C. Villiers, G. Novitchi,D. Luneau, P. Perriat, Appl. Phys. A 2011, 105, 215.

[32] P. Caravan, J. J. Ellison, T. J. McMurry, R. B. Lauffer, Chem. Rev.1999, 99, 2293.

[33] D. H. Powell, O. M. NiDhubhghaill, D. Pubanz, L. Helm, Y. S. Lebe-dev, W. Schlaepfer, A. E. Merbach, J. Am. Chem. Soc. 1996, 118,9333.

[34] J. J. Hagen, C. A. Monnig, Anal. Chem. 1994, 66, 1877.[35] K. L. Busch, G. L. Glish, S. A. McLuckey, Mass Spectrometry: Tech-

niques and Applications of Tandem Mass Spectrometry, Wiley-VCH,Weinheim, 1988.

[36] R. Hamouda, B. Bellina, F. Bertorelle, I. Compagnon, R. Antoine,M. Broyer, D. Rayane, P. Dugourd, J. Phys. Chem. Lett. 2010, 1,3189.

[37] K. M. Harkness, D. E. Cliffel, J. A. McLean, Analyst 2010, 135, 868.[38] K. M. Harkness, L. S. Fenn, D. E. Cliffel, J. A. McLean, Anal. Chem.

2010, 82, 3061.[39] T. G. Schaaff, G. Knight, M. N. Shafigullin, R. F. Borkman, R. L.

Whetten, J. Phys. Chem. B 1998, 102, 10643.[40] A. Eichhçfer, O. Hampe, Chem. Phys. Lett. 2005, 407, 186.[41] J. J. Gaumet, G. A. Khitrov, G. F. Strouse, Nano Lett. 2002, 2, 375.[42] J. Henig, E. Toth, J. Engelmann, S. Gottschalk, H. A. Mayer, Inorg.

Chem. 2010, 49, 6124.[43] C. S. Johnson, Prog. Nucl. Magn. Reson. Spectrosc. 1999, 34, 203.[44] L. Van Lokeren, G. Maheut, F. Ribot, V. Escax, I. Verbruggen, C.

Sanchez, J. C. Martins, M. Biesemans, R. Willem, Chem. Eur. J.2007, 13, 6957.

[45] A diffusion coefficient of 7.5 m2 s�1 has been measured for DEGalone in D2O at 298 K (7 mLmL�1).

[46] P. Mowat, A. Mignot, W. Rima, F. Lux, O. Tillement, C. Roulin, M.Dutreix, D. Bechet, S. Huger, L. Humbert, M. Barberi-Heyob, M. T.Aloy, E. Armandy, C. Rodriguez-Lafrasse, G. Le Duc, S. Roux, P.Perriat, J. Nanosci. Nanotechnol. 2011, 11, 7833.

[47] G. Le Duc, I. Miladi, C. Alric, P. Mowat, E. Bruer-Krisch, A. Bou-chet, E. Khalil, C. Billotey, M. Janier, F. Lux, P. Perriat, S. Roux, O.Tillement, ACS Nano 2011, 5, 9566.

[48] M. A. Delsuc, T. E. Malliavin, Anal. Chem. 1998, 70, 2146 –2148.[49] J. Moreau, E. Guillon, J.-C. Pierrard, J. Rimbault, M. Port, M.

Aplincourt, Chem. Eur. J. 2004, 10, 5218.[50] R. Fournaise, C. Petitfaux, Talanta 1987, 34, 385.[51] M. Faure, C. L. Villiers, P. N. Marche, Cell. Immunol. 2004, 228, 8.[52] M.-B. Villiers, A. Reboul, N. M. Thielens, M. G. Colomb, FEBS

Lett. 1981, 132, 49.

Received: August 23, 2012Published online: && &&, 0000

Chem. Eur. J. 2013, 00, 0 – 0 � 2013 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim www.chemeurj.org

These are not the final page numbers! ��&15&

FULL PAPERUltrasmall Silica Nanoparticles

Theranostic Agents

A. Mignot, C. Truillet, F. Lux,*L. Sancey, C. Louis, F. Denat,F. Boschetti, L. Bocher, A. Gloter,O. St�phan, R. Antoine, P. Dugourd,D. Luneau, G. Novitchi,L. C. Figueiredo, P. C. de Morais,L. Bonneviot, B. Albela, F. Ribot,L. Van Lokeren, I. D�champs-Olivier,F. Chuburu, G. Lemercier, C. Villiers,P. N. Marche, G. Le Duc, S. Roux,O. Tillement, P. Perriat . . . . . . &&&&—&&&&

A Top-Down Synthesis Route toUltrasmall Multifunctional Gd-BasedSilica Nanoparticles for TheranosticApplications

From the top down: Ultrasmall nano-particles with a sub-5 nm size havebeen synthesised by using a top-downroute. The nanoparticles are composedof a polysiloxane matrix holding Gd–DOTA chelates (DOTA=1,4,7,10-tet-

raazacyclododecane-1,4,7,10-tetraaceticacid). In rats, they have shown inter-esting properties as theranostic agentsfor radiosensitisation guided by mag-netic resonance imaging.

www.chemeurj.org � 2013 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim Chem. Eur. J. 0000, 00, 0 – 0

�� These are not the final page numbers!&16&

F. Lux et al.