Diagnostic nanoparticle targeting of the EGF-receptor in complex biological conditions using single-domain antibodies

Figure S1. (A) Quantification of EGFR level by RT-qPCR. FaDu cells were treated with two siRNA (siEGF1 and siEGF2) and neg siRNA. The RNA was extracted after 48 h post-transfection. Transcription of EGFR mRNA was analysed using RT-qPCR assay. The relative mRNA amount was calculated in relation to negative silencing cells (n =4, mean ± S.D.). (B) Uptake of Alexa Fluor® 488EGF in FaDu cells by flow cytometry. FaDu cells were treated with siEGFR and neg-siRNA for 48 h. The cells were then exposed to fluorescently labelled EGF (200 ng/mL at 37°C for 4 h. Shown is the median cell fluorescence intensity (MFI) obtained by flow cytometry of FaDu cells. (C & D) Confocal microscopy images of FaDu cells exposed to fluorescently labelled EGF protein (Alexa Fluor® 488conjugated). Cells were exposed to 200 ng/mL of protein, for 2 h, then fixed and the nuclei was stained with DAPI. Control cells (non-EGFR receptor silenced) show a high EGF protein uptake (C), while silenced cells show significantly low uptake of protein (D).


Introduction
Precise delivery of therapeutics, diagnostics or theranostics to specic tissues represents one of the major challenges in cancer imaging and therapy.2][13][14][15][16][17] For active targeting, biorecognition molecules (ligands) directed against selected tumour biomarkers are graed to the nanoparticle surface to increase and specify their delivery through specic ligand-biomarker interactions.][20] However, regarding clinical translation, while the limited success of current nanoparticle formulations in achieving highly effective biorecognition can be attributed to various reasons, it is currently incompletely understood. 21,22The fact that actively targeted nanoparticles oen fail to show benet at the (pre-)clinical stage can originate in difficulties these objects encounter in nding their target cells in vivo. 235][26] Immediately upon exposure of nanoparticle-based agents to a biological environment, macromolecules, such as proteins and lipids, tend to adsorb to their surface and a biomolecular "corona" is formed. 27,28These non-specic binding processes can have a major inuence on cellular nanoparticle uptake 29,30 as well as on the biorecognition and interaction of surface-graed targeting moieties with their corresponding receptors. 25,26We stress that this loss of specicity in targeting capacity need not necessarily diminish the overall uptake into cells.This would lead to an inability to discriminate between non-cancerous cells and tumour cells based on receptor proles.This issue is signicant, since avoiding deposition in non-targeted tissues and organs is particularly critical for radiolabelled nanoparticlebased diagnostic agents and other potentially toxic drugs.Different ligands may be affected in different ways by the biological environment, ranging from complete loss of specicity to partial loss.Here we stress the value of preliminary targeting studies in realistic milieu, prior to more extended (for example, in vivo) studies.
We begin by comparing the targeting capabilities of two peptides and a small single-domain antibody exemplied by the epidermal growth factor receptor (EGFR).This 170 kDa transmembrane glycoprotein is involved in critical cellular processes such as proliferation, differentiation and apoptosis. 31,327][38][39] The current successful approaches include inhibitory antibodies such as Cetuximab and Panitumumab, which prevent EGFR ligands from interacting and activating the receptor as well as receptor-ligand internalisation. 40However, the large size and long half-life of full monoclonal antibodies represent serious disadvantages for the application of monoclonal antibodies in imaging and therapy.They are taken up by various normal tissues, especially accumulating in the liver, and are cleared relatively slowly from the blood stream.Additionally, the diffusion through and penetration into solid tumours is rather poor. 41The optimal probe for multimodal imaging is characterised by fast tissue penetration and rapid circulation clearance as well as high tumour and low liver uptake.Ultrasmall nanoparticles have been proposed to comply with these requirements and thus represent promising next-generation tumour-targeting nanotracers.To maintain their small size, targeting moieties with low spatial dimensions such as peptides, aptamers and antibody fragments are needed.
In the present investigation, the preparation of EGFR-targeted uorescent silica nanoparticles by conjugation of specic peptides or single-domain antibodies, respectively, is reported.The latter targeting moieties are antagonistic camelid-derived variable domains binding the extracellular domain of the receptor. 42,43Both peptide ligands, GE11 (GYHWY-GYTPQNVI) [44][45][46][47] and D4 (LARLLT), 47,48 have been recently reported to bind EGFR-positive cells in vitro and in vivo, GE11 interacts with the EGF binding pocket whereas D4 binds to a distant epitope of the extracellular domain.

Characterization of synthesised nanoparticles
Fluorescently labelled silica nanoparticles (SiO 2 ) were successfully functionalized with EGFR-specic peptides D4 (SiO 2 -D4) and GE11 (SiO 2 -GE11) as well as with the single-domain antibody 7C12 (SiO 2 -sdAb).The initial amine functionalized nanoparticles consistently displayed a surface density of 6 NH 2 per nm 2 , measured by ninhydrin assay, while bifunctional PEG linkers attached with a density of around 1 SMPEG per nm 2 , according to thermogravimetric analysis.Bioconjugation was then conrmed, following extensive centrifugal cleaning, by micro BCA protein assay against PEG controls.Characterization of nanoparticle conjugates by dynamic light scattering (DLS) and differential centrifugal sedimentation (DCS) showed a shi in apparent particle size aer functionalisation with targeting moieties (Table 1 and Fig. 1).The increase in the hydrodynamic diameter upon peptide/protein conjugation without substantial alteration in the polydispersity indices indicated the formation of relatively monodisperse nanoparticle conjugates.

Binding and uptake of uorescent nanoparticles
In order to investigate EGFR-specic targeting of functionalized nanoparticles, we analysed binding and uptake in the epithelial cell line FaDu originating from a squamous cell carcinoma of the hypopharynx. 50These human head and neck tumour cells express approximately 7 Â 10 5 EGFR molecules per cell, which represents a moderate expression level. 51,52Moreover, RNA interference (RNAi) was used to knockdown the expression of the receptor in these cells to determine the effect of the targeting moieties on nanoparticle uptake.It has been shown recently that the absolute uptake level does not simply give information on the specicity of the targeting moiety on nanoparticles to relevant receptors; however, the difference of particle uptake in silenced and non-silenced cells can be used to indicate the relative contribution made by the specic pathway. 26Two validated small interfering RNA (siRNA) duplexes referred to as siEGFR-1 and siEGFR-2, both targeting different regions of the target mRNA, were separately introduced into FaDu cells.The efficiency of the gene silencing was determined by measuring the reduction of EGFR-encoding mRNA using quantitative real time PCR (Fig. S1A †).Furthermore, the uptake of uorescently labelled EGF by silenced and non-silenced FaDu cells was analysed by ow cytometry (Fig. S1B †) and confocal microscopy (Fig. S1C and D †).Successful knockdown of EGFR was observed using either of the siRNA duplexes as seen by the reduction of about 90% of mRNA aer 48 h post-transfection (Fig. S1A †).In addition, reduction of cell surface located EGFR was conrmed by a decrease in Alexa Fluor® 488-EGF binding by siEGFR-2 silenced FaDu cells from both ow cytometry and confocal microscopy.
Cell binding and uptake of peptide functionalized nanoparticles SiO 2 -D4 and SiO 2 -GE11 were determined by ow cytometry in EGFR-positive FaDu cells as well as in EGFRnegative MDA-MB-435S cells originally isolated from a ductal adenocarcinoma of the breast (Fig. 2). 53r both types of peptide functionalized nanoparticles, a high cellular uptake into EGFR-positive and EGFR-negative cells was observed.This, together with the fact that uptake rates are  almost equal between silenced and non-silenced FaDu cells provides evidence that SiO 2 -D4 as well as SiO 2 -GE11 were largely not taken up by EGFR-specic pathway.5][46] However, Ongarora et al. observed only poor uptake of phthalocyanine-GE11 conjugates, whereas phthalocyanine-D4 derivatives accumulated in different tumour cell lines. 47These partially contradictory outcomes illustrate that the chemical nature of the conjugates and their characteristics such as charge and polarity may have a substantial inuence on their specic tumour targeting abilities.Since both peptides, D4 as well as GE11, appeared to be incompatible with the herein utilised nanoparticle platform, single-domain antibodies (sdAbs) representing alternative EGFR-specic targeting moieties were attached to the surface of silica nanoparticles (SiO 2 -sdAb).Exposure of silenced and non-silenced FaDu cells to sdAb-conjugated nanoparticles reveals substantial disparities in the level of uptake between both cell populations (Fig. 3).
Knockdown of EGFR expression leads to a reduction of uptake of about 65% suggesting a predominant receptor dependent binding and internalisation of SiO 2 -sdAb (Fig. 3A).Moreover, confocal imaging of EGFR-positive FaDu cells shows co-localization of sdAb-conjugated nanoparticles with EGFR aer 30 min exposure and internalisation as well as accumulation in the lysosomes aer 6 h.Almost no interaction of SiO 2 -sdAb was observed by confocal microscopy of silenced FaDu cells even aer 6 h of exposure (Fig. 3B).Similar results were obtained for the epidermoid carcinoma cell line A431 (Fig. S2 †), which is characterised by strong overexpression of EGFR with 1-3 Â 10 6 receptors per cell. 54,55Although these results prove EGFR-specic binding and uptake of sdAb-functionalized silica nanoparticles in buffer or serum-free medium, efficient targeting in more realistic biological environments is an essential prerequisite for later in vivo application.It has been shown recently, that the transfer of nanoparticles into a complex biological environment, e.g.serum, leads to the formation of a dynamic protein corona on the surface of nanoparticles. 56,57hese corona components may block the interactions of targeting moieties conjugated to the nanoparticle surface with their putative target and cause a loss of targeting speci-city. 25,26,58In order to verify SiO 2 -sdAb targeting to EGFR of FaDu cells in a biological milieu, we investigated their cellular binding and uptake in presence of different concentration of both human serum and foetal calf serum (Fig. 4).Increasing concentrations of human serum interfere with overall SiO 2 -sdAb uptake (Fig. 4A), however, the fraction of uptake via EGFR does not decrease substantially (Fig. 4B).In the presence of foetal calf serum, FaDu cells internalise sdAb-functionalized silica nanoparticles to a greater extent compared to human serum.For both sera, the reduction of overall uptake levels can be related the formation of a protein corona. 29To further investigate this, nanoparticles were exposed to 50 mg mL À1 of human serum and the associated biomolecular corona was isolated as described previously. 59As shown in Fig. 5, graing of a PEG linker interlayer and sdAbs on the surface of nanoparticles obviously reduces the non-specic adsorption of serum proteins.Such a functionalisation strategy has been shown recently to largely but not completely suppress serum protein adsorption. 26he observed differences in cellular internalisation between human and foetal calf serum in spite of similar protein concentrations might be caused by characteristic components of the particular serum.These include soluble, serum-resident forms of EGFR, 60 which bind and block the antigen binding regions of the sdAbs conjugated to silica nanoparticles.Such EGFR analogs lack the cytoplasmic and transmembrane domains of the receptor and originate either from alternative splicing of primary mRNAs or from proteolytic cleavage of fulllength EGFR isoforms. 61Also human EGF representing an endogenous competitor for sdAb-mediated EGFR binding of nanoparticle conjugates may contribute to the identied effect, that FaDu cells internalise SiO 2 -sdAb to a lesser extent in human compared to foetal calf serum.

Characterisation of radiolabelled nanoparticles
The sdAb-functionalized silica nanoparticles were further modied with 1,4,7-triazacyclononane-triacetic acid (NOTA) in order to achieve the attachment of a 64 Cu radiolabel for positron emission tomographic (PET) imaging. 62,63Graing this bifunctional chelator did not affect the biorecognition of EGFR-targeted nanoparticles by FaDu cells, as shown in Fig. 6, where following NOTA conjugation to the corresponding batch uptake behaviour remains unchanged.Moreover, NOTA-functionalisation of SiO 2 -sdAb has no inuence on the formation of the biomolecular corona (Fig. 5).NOTA-conjugated nanoparticles were radiolabelled by incubation with [ 64 Cu]CuCl 2 solution at room temperature for up to 1 h.Within this time period, a radiochemical yield of >98% (as analysed by radio-TLC) was obtained and longer incubation times did not improve the radiochemical yield (Fig. 7A).
In order to investigate the competition of free human EGF with radiolabelled SiO 2 -sdAb-NOTA for EGFR binding, we analysed nanoparticle binding to FaDu cells in the presence of an excess of this endogenous ligand (Fig. 7B).Upon incubation of FaDu cells with free human EGF, targeting of SiO 2 -sdAb-NOTA to EGFR is lost.Furthermore, the therapeutic antibody Cetuximab competes for the binding to EGFR, suggesting that sdAb-functionalized nanoparticles bind epitopes overlapping with or in close proximity to EGF and Cetuximab binding sites.To investigate EGF competition in more detail, we determined cellular binding of radiolabelled SiO 2 -sdAb-NOTA to FaDu in the presence of increasing EGF concentrations (Fig. 8).
No reduction of nanoparticle binding was observed up to 200 pM EGF, whereas higher concentrations of the endogenous EGFR ligand substantially decrease receptor-specic nanoparticle interaction.An EGF concentration of 500 nM completely blocks the corresponding receptor and remaining nanoparticle binding occurs by EGFR non-specic nanoparticle-cell interaction.However, at physiological EGF serum concentrations ranging from 10 pM to 190 pM, 64,65 no impairment of SiO 2 -sdAb-NOTA binding to their molecular target was observed.Concentration of EGF in the human serum used here was determined by either dilution of serum (1280 pg mL À1 ) or by serum spiking (1145 pg mL À1 ).These values correspond to $180 to 200 pM and are in good agreement with EGF levels of other commercially available pooled serum samples (Fig. S3 †).
Overall, the presented results clearly illustrate the strong inuence of the corresponding biological context on the efficiency of receptor-specic nanoparticle targeting.Recently we have shown that targeting specicity of transferrin-conjugated nanoparticles is lost upon transfer to a complex biological environment.Furthermore, we found that proteins in the cell culture media restrain NP surface bound transferrin from interacting with its receptor. 26The results presented herein    7 Radiolabelling and cellular binding of SiO 2 -sdAb-NOTA.After modification with 1,4,7-triazacyclononane-triacetic acid (NOTA), sdAb-functionalized silica nanoparticles were labelled with 64 Cu until a radiochemical purity of >98% was obtained as analysed by radio-TLC (A).A 1 mM excess of human epidermal growth factor (EGF) or of the EGFR-inhibitory antibody Cetuximab (C225), respectively, blocks binding of radiolabelled sdAb-functionalized silica nanoparticles to EGFR-presenting FaDu cells.Binding data are expressed as % of injected dose per mg protein (%ID per mg protein).Each point represents the mean AE SD of three samples.conrm these ndings, since in both cases we observed that the efficiency of receptor-specic nanoparticle targeting is affected by the biological context.However, for the sdAb-EGFR ligandreceptor pair we see that the specicity is reduced, but not obscured completely.These observations clearly illustrate, that results obtained in biologically irrelevant conditions (e.g.simple buffer systems, serum-free conditions) are not very meaningful.As a minimal prerequisite we suggest to carry out cellular binding and uptake studies in the biological uids in which the particles will be applied.However, currently no prediction can be made as to if a certain ligand-nanoparticle conjugate maintains its specicity in complex biological context.This means that targeting ability has to be checked for every single ligand-receptor pair.

Dye conjugate solution
N-1-(3-Trimethoxysilylpropyl)-N 0 -uoresceyl thiourea (FITC-APTMS) or (RITC-APTMS) conjugate solutions were prepared by dissolving 4 mg of reactive dye in 2 mL of anhydrous ethanol.Twenty mL of APTMS (about 11Â molar excess) was then added immediately to this solution, with the mixture then shaken at room temperature in darkness for 4 h.The reaction time course was initially monitored by 1

Nanoparticle preparation
To 25 mL of EtOH (99.9%) was added 0.91 g of aq.ammonia (28.0-30.0%NH 3 basis) in a polypropylene container.To this mixture, under rapid stirring, was added 500 mL of the prepared conjugate solution.The reaction was stirred for 15 min, upon which TEOS (940 mL) was added.The reaction was then stirred at 600 rpm at 25 C for further 20 h in darkness.The resulting nanoparticle suspension was centrifuged down at 14 000 rpm for 20 min, with the pellet then resuspended in fresh EtOH aided by bath sonication.This washing procedure was repeated twice more, followed by three water washes and a nal resuspension in water at a total volume of 12 mL.

Surface amination
The FITC-SiO 2 particles were suspended in water at a concentration of 10 mg mL À1 and to this suspension APTES was added to a nal concentration of 1 vol%.The reaction which proceeded with gradual agglomeration visible, was shaken at 600 rpm for 2 h at room temperature followed by incubation at 90 C for 1 h.The particles were cleaned by centrifugation and resuspension in water four times, giving a nal clear suspension.The number of amines presented at the NP surface was measured by ninhydrin assay.Following centrifugal washing of NPs into pure ethanol (Â3) they were then incubated with ninhydrin reagent (0.7 mg mL À1 ) in absolute ethanol at 60 C for 30 minutes and measured against APTES standard curves.

Protein conjugation to pegylated nanoparticles
To 0.12 mmol of protein (per 10 mg nanoparticles) dissolved at a concentration of 2 mg mL À1 in PBS (pH 7.4) was added SAT(PEG) 4 dissolved in dimethyl sulfoxide (DMSO) (76 mL of 1 mg mL À1 , 0.18 mmol).Aer 30 min shaking slowly at room temperature, 100 mL (mL À1 reaction) of deacetylation buffer composed of 0.5 M hydroxylamine and 25 mM ethylenediaminetetraacetic acid (EDTA) in PBS, pH 7.4 was added.The reaction was allowed to continue for 2 h, followed by cleaning on a Sephadex G25 column with exchange into deoxygenated 20 mM HEPES buffer (pH 7.4).The collected protein fraction was then incubated for ve minutes with tris-(2-carboxyethyl)phosphine (TCEP) (0.24 mmol) before mixing with PEG modied NPs.

Nanoparticle pegylation
The aminated particles were washed twice with 20 mM HEPES buffer (pH 7.4) by centrifugation, before resuspension in the same buffer at a concentration of 10 mg mL À1 .They were added to an equal volume solution of freshly diluted 5 mg mL À1 SM-PEG 8 -Mal, which corresponds to around 10Â close packed monolayer in 20 mM HEPES (pH 7.4), with mixing.The clear suspension reaction was shaken for 2 h followed by centrifugation at 14 000 rpm and two washes with 20 mM HEPES buffer (pH 7.4) and then nally resuspended in deoxygenated 20 mM HEPES buffer (pH 7.4) to a nal concentration of 10 mg mL À1 nanoparticles.The work was timed so that the modied protein solution and modied particle dispersion would be ready simultaneously and were then combined in a ratio of 0.12 mmol proteins per 10 mg particles with a nanoparticle reaction concentration of 5 mg mL À1 and shaken gently together for 2 h at RT before incubating at 4 C overnight.The solution was then cleaned of unreacted protein by centrifugation and resuspension three times in ltered 20 mM HEPES (pH 7.4).The number of bound proteins was measured by micro BCA assay against their corresponding preserved PEG control samples.

Chelator conjugation to nanoparticles
Five mg (8.9 mmol) of SCN-Bn-NOTA was dissolved in DMSO (1000 mL).Seven mL (60 nmol) of this solution was then added to 0.5 mL of NP suspension (5 mg NP, 12.6 nmol sdAb) giving a reaction ratio of approx.5 : 1 (reactive macrocycle: sdAb), with immediate mixing by inversion.The dispersion was then slowly shaken for 30 min followed by washing by three cycles of centrifugation (12 000 rpm for 15 min) and resuspension in mM HEPES (pH 7.4).

Differential centrifugal sedimentation (DCS) and dynamic light scattering (DLS)
Nanoparticle dispersion was measured by DLS performed on a Malvern Nanosizer ZS.Particles were suspended at a concentration of 100 mg mL À1 in the relevant buffer.Size measurements were averaged results from 3 Â 11 runs.DCS experiments were performed with a CPS Disc Centrifuge DC24000 (CPS Instruments).Particles were injected at a concentration of 500 mg mL À1 into a 24-8% sucrose-suspension medium (water or PBS) gradient spinning at 20 000 rpm.

Radiolabelling and instant thin-layer chromatography
The production of 64 Cu was performed at Cyclone® 18/9 (Helmholtz-Zentrum Dresden-Rossendorf) in a 64 Ni(p, n) 64 Cu nuclear reaction with specic activities of 150-250 GBq mmol À1 Cu diluted in HCl (10 mM). 66To 100 mg of SiO 2 -sdAb-NOTA nanoparticles in 100 mL 10 mM MES, pH 6.0, 1 MBq [ 64 Cu]CuCl 2 was added and incubated at room temperature for 60 min.A 5 mL aliquot of the reaction was combined with 2 nmol EDTA, pH 7.0 and the labelling process of the nanoparticles (R f ¼ 0) was monitored by radio-TLC using ITLC-SA plates (Merck Millipore) in combination with a mobile phase of 0.9% NaCl in dH 2 O.As control, separate radio-TLC analysis of [ 64 Cu]Cu-EDTA (R f ¼ 1) was performed in the same mobile phase.Evaluation of radio-TLC was carried out using a radioactivity thin layer analyser (Rita Star, Raytest).

Heterologous expression and purication of sdAb
Single-domain antibodies were expressed and puried as described recently. 49ll culture Tissue culture reagents were purchased from Biochrom AG and GIBCO Invitrogen Corporation/Life Technologies Life Sciences unless otherwise specied.The adherent human tumour cell lines A431 (ATCC® number: CRL-1555), FaDu (ATCC® number: HTB-43) and MDA-MB 435S (ATCC® number: HTB-129) were maintained as monolayer cultures in DMEM supplemented with 10% foetal calf serum (FCS), respectively, and incubated in a humidied atmosphere of 95% air/5% CO 2 at 37 C.All cell lines were conrmed to be mycoplasma negative using the LookOut mycoplasma PCR detection kit (Sigma-Aldrich) and were tested monthly.

Cell silencing and ow cytometry
A total of 30 000 cells were seeded in 24 well plates (Greiner), and incubated for 24 h before silencing of the gene coding for epidermal growth factor receptor (EGFR).Cells were then transfected with 15 pmol of Silencer Select siRNA siEGFR-1 (#s563) or siEGFR-2 (#s564) using Oligofectamine™ according to the manufacturer's instructions (Life Technologies).Neg1 silencer was used as a negative control.Cells were transfected with siRNAs in all experiments 48 h before exposure to nanoparticles or labelled EGF.Aer 48 h silencing, cells were washed for 10 min in serum-free DMEM.The medium was then replaced by the nanoparticle dispersions, freshly prepared by diluting the nanoparticle stock in serum-free DMEM, or medium supplemented with different concentration of FCS or human serum, for different times, depending on the experiment.Similar experiments were performed by exposing cells to 200 ng mL À1 Alexa Fluor® 488-labelled human EGF in serumfree DMEM.For ow cytometry, cells were washed once with DMEM supplement with 10% FCS and twice with PBS and harvested with trypsin.Cell pellets were then xed at room temperature with 4% formalin (Sigma-Aldrich) for 20 min, and resuspended in PBS before cell-associated uorescence (15 000 cells per sample) was measured using an Accuri C6 reader (BD Accuri Cytometers).The results are reported as the median of the distribution of cell uorescence intensity, averaged over two to three independent replicates.Error bars represent the standard deviation between replicates.Each experiment was performed at least three times.

Confocal microscopy
For confocal microscopy, 10 4 cells were seeded onto 35 mm plates with 15 mm diameter glass coverslips and grown for 24 h prior to silencing.Aer 48 h silencing, both silenced cells and non-silenced cells (controls) were exposed to uorescently labelled EGF protein (Alexa Fluor® 488-conjugated, at a concentration of 200 ng mL À1 for 2 h) and to SiO 2 -sdAb nanoparticles at a concentration of 10 mg mL À1 for 30 min and for 6 h.For organelle and protein staining, samples were then washed three times with 1 mL PBS, xed for 20 min with 1 mL of 4% formalin at room temperature.The cell-membrane was permeabilised using 1 mL of 0.1% saponin (Sigma Aldrich) solution for 5 min at room temperature and cell were then incubated for 30 min at room temperature with a blocking solution of 1% bovine serum albumin fraction V (Sigma Aldrich) in PBS-Tween to prevent antibody non-specic binding.Samples were then incubated for 1 h at room temperature with a primary antibody 1 : 200 rabbit polyclonal to LAMP-1 (Abcam) and with a primary antibody 1 : 200 mouse monoclonal antibody to EGFR (Abcam), washed three times with 1 mL PBS, and then incubated at room temperature for 1 h with 1 : 400 dilution of Alexa Fluor® 488 goat anti-rabbit IgG and with 1 : 400 dilution of Alexa Fluor® 647 goat anti-mouse IgG as secondary antibodies (Molecular Probes, Life Technologies).Samples were washed three times with 1 mL PBS and incubated for 5 min with DAPI (Sigma Aldrich) before mounting with MOWIOL (Polysciences Inc.) on slides for imaging.The cells were observed using a Carl Zeiss LSM 510 Meta laser scanning confocal microscope with lasers at 364 nm and long pass lter LP 385 nm (DAPI), 488 nm and band pass lter 505-530 nm (uorescently labelled EGF protein and LAMP-1 antibody), 543 nm and band pass lter 558-612 nm (nanoparticles) and 633 nm and band pass lter 644-719 nm lter (EGFR antibody).

Serum characterisation
Human serum (Biochrom AG) was tested for total protein content using a bicinchoninic acid (BCA) protein assay (Thermo Scientic).The amount of EGF present in human serum was quantied using a Human EGF ELISA Kit (Invitrogen).The ELISA assay was carried out according to manufacturer's spec-ications.The absorbance at 450 nm was read using a Spec-traMAX 190 plate reader.Two approaches were used and compared in order to determine the concentration of EGF.The rst method was carried out by serially diluting serum and examining the levels of EGF quantied for each of the diluted samples.The second approach involved spiking a sample of serum with known amounts of EGF and measuring the response observed in the assay.
In vitro binding and uptake studies of radiolabelled SiO 2 -sdAb-NOTA A total of 50 000 cells were seeded in 24 well plates (Greiner) and cultivated for 24 h before exposure to nanoparticles.Aer 24 h, cells were washed for twice with warm PBS.The buffer was then replaced by the nanoparticle dispersions, freshly prepared by diluting the radiolabelled nanoparticle stock in serum-free DMEM, or medium supplemented with different concentration of FCS or human serum, for different times, depending on the experiment.Following treatment with radiolabelled nanoparticles for certain time periods, cells were washed twice with PBS in order to ensure removal of loosely attached nanoparticles from the cellular membrane.Finally, cell lysis was achieved by the addition of 1% SDS in 0.1 M NaOH and incubation for 30 min at room temperature with vigorous shaking.The radioactivity in the cell extracts was quantied using an automated gamma counter (PerkinElmer Life and Analytical Sciences).Total protein concentration in cell extracts was determined colorimetrically with the DC Protein Assay (Bio-Rad Laboratories) according to the manufacture's microplate assay protocol using bovine serum albumin as protein standard.

Competition assay
A total of 15 000 FaDu cells were seeded in 48 well plates (Greiner) and cultivated for 24 h before exposure to nanoparticles.Aer 24 h, cells were washed twice with ice-cold PBS and incubated on ice for 30 min.Subsequently, different concentrations of human EGF ranging from 1 pM up to 1 mM as well as 10 mg mL À1 radiolabelled SiO 2 -sdAb-NOTA were added.Aer further incubation on ice for 2 h, cells were washed twice with ice-cold PBS, lysed by addition of 1% SDS in 0.1 M NaOH and incubated for 30 min at room temperature with vigorous shaking.The radioactivity in the cell extracts was quantied using an automated gamma counter (PerkinElmer Life and Analytical Sciences).

Isolation and characterisation of nanoparticle-protein complexes
Biomolecular corona forming on silica nanoparticles was isolated as described recently with slight modications. 59Briey, samples containing 100 mg mL À1 of SiO 2 , SiO 2 -sdAb or SiO 2 -sdAb-NOTA, respectively, were incubated with 50 mg mL À1 of "off the clot" human serum (Biochrom AG) diluted with dH 2 O for 1 h at 37 C in protein LoBind vials (Eppendorf) with signicantly reduced protein-to-surface binding.Aer incubation in serum, samples were centrifuged for 20 min at 10 000 Â g at 4 C to pellet the nanoparticle-protein complexes and to remove the supernatant serum.The pellet was then washed three times with 1 mL dH 2 O and centrifuged again for 20 min at 10 000 Â g at 4 C to remove proteins with low affinity for the nanoparticle surface.Before the last centrifugation step, the nanoparticle dispersions were transferred into new vials in order to discard proteins bound to the inner surface of the vials.The nanoparticle-protein pellet was resuspended in Laemmli sample buffer (Bio-Rad Laboratories) immediately aer the last centrifugation step and incubated for 5 min at 100 C to denature the proteins.Aer cooling to room temperature, the samples were nally loaded on a 12% polyacrylamide gel and subjected to electrophoresis until the bromophenol blue dye of the sample buffer reached the end of the gel.On each gel, one lane was used to separate a molecular weight ladder standard, the PageRuler pre-stained protein ladder (Thermo Fisher Scientic).Aer electrophoresis, proteins were stained with PageBlue protein staining solution (Thermo Fisher Scientic) according to the manufacturer's instructions.

Conclusions
In conclusion, sufficient specic recognition of targeting ligands graed to the surface of nanoparticles by their corresponding receptors depends on a variety of factors.These include the binding affinity of the ligated nanoparticle to its molecular target as well as the endogenous competitor concentration, and both factors inuence the residence time for a ligand at its receptor binding site.The dissociation constant, which describes how tightly a particular ligand binds to its corresponding target, differs by one order of magnitude between the investigated peptide GE11 and the sdAb 7C12.It is not surprising, then, that the fraction of specic EGFR-mediated cellular uptake is substantially increased for sdAb-functionalized nanoparticles compared to their peptide-conjugated counterparts.However in this case, as for all nanoparticle-cell interaction studies there are a range of variables at play such as colloidal stability related to peptide pI, NP surface self-adsorption effects, etc. precluding direct comparison based on dissociation constants.In this study sdAb functionalized platforms were shown to function well in terms of biological recognition specic interactions.We observed a serum species type dependence in overall NP uptake where matching cell and serum protein for species resulted in the greatest diminution of overall nanoparticle uptake, suggesting the possibility of loss of specicity in situ.Our investigations using EGF competition studies suggest that it may not result mainly from endogenous EGF competition.Nevertheless, the sdAb-functionalized nanoparticles retain sufficient efficiency to remain credible candidates for further consideration.
We stress here the key overarching point.There is considerable potential for particles in situ to lose, or at least modulate, their specicity, compared to expectations in simple buffers.Even the differences between human and bovine serum, may be signicant and clearly demonstrates the need to choose carefully appropriate experimental conditions and combinations in drawing conclusions from in vitro data.While we are not yet in a position to predict which ligands, and which ligation chemistries and nanoparticles lead to modulation of targeting efficiency, we believe that studies such as those presented here should be a basic prerequisite screen prior to more in depth consideration and in vivo study.

Fig. 2
Fig. 2 Uptake of peptide-functionalized nanoparticles by different cancer cell lines.EGFR-positive FaDu and EGFR-negative MDA-MB 435S cells were silenced for 48 h with negative silencer control (neg siRNA) and siRNA for EGFR (siEGFR-2) prior to exposure to 100 mg mL À1 SiO 2 -D4 (A) or SiO 2 -GE11 (B).Median cell fluorescence intensity was measured by flow cytometry, showing that the uptake is not reduced in cells silenced for EGFR.

Fig. 3
Fig. 3 Uptake of sdAb-functionalized nanoparticles by FaDu cells.Median cell fluorescence intensity determined by flow cytometry of FaDu cells exposed to 10 mg mL À1 of SiO 2 -sdAb showing that the uptake is strongly affected by EGFR knockdown (A).Confocal microscopy images of non-silenced and silenced FaDu cells exposed to SiO 2 -sdAb nanoparticles for 30 min and 6 h in serum free DMEM (B).Nanoparticles in red, LAMP-1 in green and EGFR in white.Scale bars of 10 mm for the main images and 2 mm for the zoomed images.

Fig. 4
Fig. 4 Uptake of SiO 2 -sdAb in different concentration of human (A/B) and foetal calf (C/D) serum.Median cell fluorescence intensity measured by flow cytometry of silenced (-E) and non-silenced (-N) FaDu cells exposed to 10 mg mL À1 of SiO 2 -sdAb in serum-free medium (SF) and medium supplemented with human (A) or foetal calf (C) serum, respectively, showing that the uptake is strongly dependent on the present concentration of serum.The EGFR-dependent fractions were calculated using the difference in fluorescence between non-silenced (neg siRNA) and silenced (siEGFR-2) cells divided by the fluorescence of non-silenced cells from the uptake curves in (A) or (C), e.g.((non-silencedsilenced)/non-silenced).This allows quantifying that, in spite of increasing serum concentrations, the fraction of uptake depending on EGFR remains high (B/D).

Fig. 5
Fig. 5 SDS-PAGE analysis of protein corona composition on SiO 2 nanoparticles upon incubation in 50 mg mL À1 of human serum.Nanoparticle surface associated proteins were isolated after incubation of SiO 2 (lane 1), SiO 2 -sdAb (lane 2) or SiO 2 -sdAb-NOTA (lane 3) with 50 mg mL À1 of "off the clot" human serum.Attachment of sdAbs on the surface of nanoparticles obviously reduces the unspecific adsorption of serum proteins, whereas further functionalisation with the copper-64 chelator 1,4,7-triazacyclononane-triacetic acid (NOTA) shows minimal influence on corona composition.

Fig.
Fig.7Radiolabelling and cellular binding of SiO 2 -sdAb-NOTA.After modification with 1,4,7-triazacyclononane-triacetic acid (NOTA), sdAb-functionalized silica nanoparticles were labelled with64 Cu until a radiochemical purity of >98% was obtained as analysed by radio-TLC (A).A 1 mM excess of human epidermal growth factor (EGF) or of the EGFR-inhibitory antibody Cetuximab (C225), respectively, blocks binding of radiolabelled sdAb-functionalized silica nanoparticles to EGFR-presenting FaDu cells.Binding data are expressed as % of injected dose per mg protein (%ID per mg protein).Each point represents the mean AE SD of three samples.

Fig. 8
Fig. 8 Competition curves of human epidermal growth factor versus [ 64 Cu]Cu-SiO 2 -sdAb-NOTA using FaDu cells.Binding of radiolabelled sdAb-functionalized silica nanoparticles to EGFR-presenting FaDu cells was investigated in the presence of increasing concentrations of EGF.Percentage of bound activity was calculated in the way that the mean counts of a triplicate data point were related to the counts of data points without competitor.All counts were decay corrected.Each point represents the mean AE SD of three samples.

Table 1
Characteristics of nanoparticle conjugates and corresponding targeting ligands