Analytical Biochemistry 394 (2009) 7–12

Contents lists available at ScienceDirect

Analytical Biochemistry

journal homepage: www.elsevier .com/locate /yabio

Characterization of cluster-assembled nanostructured titanium oxide coatingsas substrates for protein arrays

Roberta Carbone a,*, Marzia De Marni a, Andrea Zanardi a, Simone Vinati a, Emanuele Barborini a,Lorenzo Fornasari b, Paolo Milani c

a Tethis, Via Russoli 3, 20143 Milano, Italyb Congenia–Genextra Group, Piazzetta Bossi 4, 20121 Milano, Italyc CIMAINA and Dipartimento di Fisica, Università di Milano, Via Celoria 16, 20133 Milan, Italy

a r t i c l e i n f o a b s t r a c t

Article history:Received 23 March 2009Available online 7 July 2009

Keywords:Protein microarraysNanobiotechnologyTitanium oxide (TiOx)Biochips

0003-2697/$ - see front matter � 2009 Elsevier Inc. Adoi:10.1016/j.ab.2009.07.005

* Corresponding author. Fax: +39 0236569183.E-mail address: roberta.carbone@tethis-lab.com (R

1 Abbreviations used: ELISA, enzyme-linked immunostructured titanium oxide; SCBD, supersonic cluster bforce microscopy; PBS, phosphate-buffered saline; NCmultiplier tube; IgG, immunoglobulin G; S/N, signavariation; M-CV, mean intraslide CV; W-CV, overall imean interslide repeatability; B-CV, general estimate o

Protein microarray technologies are rapidly expanding to fulfill current needs of proteome discovery fordisease management. Nanostructured materials have been shown to present interesting features whenused in biological settings: nanostructured titanium oxide film (ns-TiOx), synthesized by supersonic clus-ter beam deposition (SCBD), has recently emerged as a biocompatible substrate in different biologicalassays. The ns-TiOx surface is characterized by a morphology at the nanoscale that can be tuned to mod-ulate specific biomolecule–material interactions. Here we present a systematic characterization of ns-TiOx coatings as protein binding surfaces, comparing their performances with those of most commoncommercial substrates in protein and antibody microarray assays. Through a robust statistical evaluationof repeatability in terms of coefficient of variation (CV) analysis, we demonstrate that ns-TiOx can be usedas reliable substrate for biochips in analytical protein microarray application.

� 2009 Elsevier Inc. All rights reserved.

The complexity of the human proteome has encouraged thedevelopment of sophisticated multiplexed technologies for moreappropriate methods of analysis. As a consequence, the trend ofdetection technologies has moved from low-throughput analysis(e.g., enzyme-linked immunosorbent assay [ELISA],1 Western blots,mass spec) to ‘‘high-content/high-throughput” approaches relyingeither on planar- or ‘‘bead”-based solutions [1–4]. In this context,the development of novel miniaturized devices, such as biochipsand microarrays, can offer a valid approach to interrogate the prote-ome by multiplexing the information at a reasonable cost.

The successful development of such devices relies on the optimi-zation of the complex interactions that occur between ‘‘proteins”and the ‘‘surface” that provides the means for immobilization. Themodalities of protein adsorption are quite different, according tothe variety of the substrates (e.g., two-dimensional, three-dimen-sional, covalent binding, electrostatic, diffusion), and deeply influ-ence the performance of the assay. Moreover, substratefluorescence background, assay reproducibility, and affinity toward

ll rights reserved.

. Carbone).sorbent assay; ns-TiOx, nano-eam deposition; AFM, atomic, nitrocellulose; PMT, photo-l-to-noise; CV, coefficient ofntraslide repeatability; A-CV,f interslide repeatability.

specific targets (which influence limit of detection and linearity) aredistinctive features of protein microarray assays, which require pre-cise selection of optimal conditions [5–7]. In this regard, the avail-ability of novel surfaces for protein immobilization represents anopportunity for the setting up of microarray-based devices for opti-mized protein profiling [8–12].

The role of the surface nanostructure is the object of increasinginterest so as to understand how this affects the surface–proteininteraction in view of the nanoscale engineering of substrates toobtain better performing protein microarrays. Despite the wide ef-fort in addressing the mechanism and role of surface topography inprotein–biomaterial interaction, this topic is still a matter of con-troversial debate, probably due to the complexity of models thatare evaluated in different studies [13–15]. We recently character-ized cluster-assembled nanostructured titanium oxide (ns-TiOx)thin films produced by supersonic cluster beam deposition (SCBD)[16,17] as a biocompatible [18] and optimal substrate for cellmicroarray-based assays [19]. Furthermore, we also showed thatcluster-assembled ns-TiOx can efficiently bind streptavidin byadsorption [20], underlying a putative role of surface nanoscaleand chemistry, thereby suggesting that cluster-assembled ns-TiOxsurfaces can be of interest for protein microarray applications.

Here we present a systematic evaluation of cluster-assembledns-TiOx slides on protein microarray assays in comparison withcommercial slides. We characterized the spot morphology regular-ity, and we compared the protein adsorption efficiency, intraslide

8 Nanostructured TiOx coatings as substrates for protein arrays / R. Carbone et al. / Anal. Biochem. 394 (2009) 7–12

repeatability, and interslide repeatability to assess the use of ns-TiOx slides as a reliable protein microarray substrate.

Fig. 1. AFM surface topography of a typical 50-nm-thick ns-TiOx coating depositedon glass coverslip (500 � 500 nm2, vertical scale 15 nm).

Materials and methods

ns-TiOx slide deposition by SCBD

ns-TiOx coatings (50 nm thickness) were prepared on standardglass slides (Nexterion Glass D, clean room cleaned, Schott, Mainz,Germany) by depositing under high vacuum a supersonic seededbeam of TiOx clusters produced by a pulsed microplasma clustersource, as described in detail in Refs. [16,17]. After ns-TiOx deposi-tion, slides were immediately annealed at 250 �C overnight in aclean oven, as described previously [20]. Coated slides were thenpackaged in vacuum and stored at room temperature in the darkuntil use.

The surface morphology of cluster-assembled ns-TiOx coatingswas characterized by atomic force microscopy (AFM), employinga Digital Instruments Nanoscope multimode IV atomic force micro-scope in tapping mode.

Protein and antibody array assays

For protein arrays, streptavidin (Cy3 conjugated, Amersham–GEHealthcare, UK) was reconstituted and stored in phosphate-buf-fered saline (PBS, pH 7.4, Lonza Group) at 2 mg/ml at �20 �C andthen diluted in spotting buffer with PBS and 5% (w/v) glycerol(BDH Biosciences) to a final concentration of 0.5–500 lg/ml andspotted using a BioDot AD1500 equipped with Axsys software.

Commercial epoxy-coated slides (SuperEpoxy slides, TelechemInternational, USA) and nitrocellulose (NC) slides (FAST slides,Whatman, GE Healthcare, UK) were treated according to manufac-turer specifications. Slides were spotted at 50% humidity at 40 nl/spot with 2 mm pitch and 20 replicates per concentration.

After 3 h of incubation in 75% humidity, ns-TiOx slides wereblocked with 5% (w/v) milk in 0.5% PBST (0.5% PBS/Tween 20)(BDH Biosciences) for 1 h. After blocking, slides were washed twicewith 0.5% PBST, then with PBS, and finally with deionized H2O toremove salts. Slides were then analyzed with a GenePix 4000Bscanner at a 532 nm wavelength with 33% laser power and anappropriate photomultiplier tube (PMT) gain for each substrate(epoxy: PMT 750; ns-TiOx: PMT 750; NC: PMT 300) at 10 lm/pixelof resolution.

For antibody array experiments, mouse immunoglobulin G(IgG) antibodies (Jackson ImmunoResearch, USA) were diluted intwo different spotting buffers containing PBS/5% (w/v) glycerol orPBS/1.5 M betaine (Sigma–Aldrich, USA) at final concentrations of0.5–1000 lg/ml.

Slides were spotted at 75% humidity at 40 nl/spot with 2 mmpitch and 20 replicates per concentration. Epoxy-coated and NCslides were processed for incubation, blocking, and detection fol-lowing manufacturer instructions; ns-TiOx slides were incubatedfor 5 h in 75% humidity and then blocked for 1 h in 0.5% PBST in5% milk. At the end of blocking, slides were washed twice with0.5% PBST and then incubated overnight with 1 lg/ml goat anti-mouse IgG (Cy3 conjugated, Jackson ImmunoResearch) in 1% (w/v) milk in 0.5% PBST.

Slides were then washed twice with 0.5% PBST, then with PBS,and finally with deionized H2O. Slides were then analyzed with aGenePix 4000B scanner at a 532 nm wavelength with 33% laserpower and appropriate PMT gain for each substrate (epoxy: PMT300; ns-TiOx: PMT 300; NC: PMT 150) at 10 lm/pixel of resolution.

Data were exported to Microsoft Excel software for processing.The fluorescence at every spot was evaluated as the median of allmeasured pixels so as to reduce the influence of possible outlayers.

The estimate of the fluorescence given at every concentration re-sulted from the average of 18 to 20 replicates (i.e., 18–20 differentspots in the same experimental condition) according to the micro-array assay. Signal-to-noise (S/N) ratio was calculated as the ratiobetween the mean spot intensity and the standard deviation ofbackground intensity.

Statistical analysis of microarray data

According to recent work on quality control of microarrayexperiments [21,22], we evaluated the repeatability of both micro-array assays by means of the coefficient of variation (CV) because itexpresses the ‘‘dispersal of data around the average” (the standarddeviation) weighted by the average value (so as to make compara-ble the amounts of dispersal at low and high concentrations). CVvalues were first calculated, for each slide, by taking into consider-ation the (median) fluorescence values measured at all spots sepa-rately for each given concentration.

Intraslide repeatabilityBy gathering the results of each set of four to six experiments (2

slides/experiment) related to the different assays (mouse IgG andstreptavidin–Cy3 arrays) and conditions (spotting buffers), it ispossible to average among slides the ‘‘internal” CVs at every con-centration, obtaining a series of mean intraslide CV values (hereaf-ter M-CV) ± 1.96 standard error (SE). The sequence of these M-CVsallows us to draw the pattern of variation of intraslide repeatabilityaccording to the different spotted protein concentrations. More-over, we can obtain a single parameter representing the overallintraslide repeatability (hereafter W-CV, where W represents‘‘within”) by averaging all of the M-CV values, indifferently fromthe concentration given.

Interslide repeatabilityIf we take into consideration the average median fluorescence

per concentration in any slide (same experimental conditions)and then we calculate the CV of those values among slides, we ob-tain a reliable estimate of interslide repeatability (A-CV, a singlevalue at any concentration). By averaging the values referring toall A-CV values at different concentrations, we can also obtain a

Fig. 2. (A) Mouse IgG spot morphology and background on ns-TiOx slide. Theexcellent morphology is demonstrated by surface plot (B) and plot profile (C) of aselected 3 � 3 array of mouse IgG spots detected by anti-mouse Cy3 antibody. (D)ns-TiOx slide background in comparison with epoxy and NC slides. A total of 500evenly spaced circles (1 mm diameter) were analyzed at 100% laser power and PMT400 for each slide in duplicate. In the inset graph, background values of ns-TiOx andepoxy slides are shown with a different scale of mean intensity. AU, arbitrary units.

Nanostructured TiOx coatings as substrates for protein arrays / R. Carbone et al. / Anal. Biochem. 394 (2009) 7–12 9

general estimate of the interslide CV (B-CV ± 1.96 SE, where B rep-resents ‘‘between”).

Results and discussion

Morphology and properties of ns-TiOx-coated slides

The random stacking of nanoparticles on substrates resultingfrom SCBD produces films with a homogeneous nanoscale porosityand roughness [23]. A typical ns-TiOx surface morphology ob-tained by AFM is shown in Fig. 1; root mean square roughness ofapproximately 10 nm was observed. Due to both small thicknessand electronic structure of ns-TiOx (which has a bandgap largerthan 3 eV), the coating appears to be perfectly transparent for vis-

Fig. 3. Protein and antibody array experiments: analysis of mean spot intensity. (A) Stvalues of replicate spots (±SD) are shown according to spotted protein concentrations foracquisition, NC slide data are presented in a separate graph. AU, arbitrary units.

ible wavelength, according to the requirements of fluorescence-based characterization of protein array [24].

Therefore, the ns-TiOx coating was tested for spot morphologyto evaluate its performance in protein array immunoassays. Re-cently, Derwinska and coworkers [25] raised the issue of on-chipimmunoassay reproducibility through extensive evaluation ofqualitative parameters such as spot morphology. ns-TiOx coatingwas carefully tested by spotting 500 lg/ml mouse IgG antibodiesdetected with Cy3-labeled anti-mouse IgG antibodies. As shownin Fig. 2, the spot morphology is uniform, devoid of ‘‘coffee ring”-or ‘‘donut”-like shapes that are often present, particularly at highprotein concentration. From both the surface plot (Fig. 2B) and plotprofile (Fig. 2C) of a selected 3 � 3 array, the analyzed spots pres-ent excellent quality and homogeneity, providing the ideal basisfor further evaluation of the coating in protein array applications.In addition, the ns-TiOx-coated slide presented a very low back-ground when analyzed with a 532 nm laser scanner in comparisonwith epoxy-coated and NC slides (Fig. 2D).

Protein and antibody array on ns-TiOx-coated slides: comparison withcommercial slides

To evaluate protein binding, we set up a microarray experimentby spotting a range of different protein concentrations and com-paring ns-TiOx coating with commercial slides. We ran our testsin parallel using an NC slide substrate, which is known to bind pro-teins through adsorption and hydrophobic interactions, and epoxy-coated slides, which are characterized by a mechanism of covalentbinding with primary amino residues on the target proteins.

To evaluate our substrate for protein binding, different concen-trations of a Cy3-labeled model protein (streptavidin–Cy3) werespotted and incubated on ns-TiOx-coated slides as described inMaterials and methods. Finally, slides were washed with 0.5% PBSTto remove the excess and were analyzed by scanner at a 532-nmwavelength.

In Fig. 3A, we report the mean intensities versus concentrationsof epoxy- and ns-TiOx-coated slides analyzed at the same PMT va-

reptavidin–Cy3 microarray. (B) Mouse IgG microarray. Averages of mean intensityns-TiOx, epoxy, and NC slide substrates. Due to the different PMT values for image

Fig. 4. S/N ratios of microarray assays. Graphs represent the calculations of S/Nratios for streptavidin–Cy3 (A) and mouse IgG (B) microarray assays for eachsubstrate and buffer condition.

Fig. 5. Analysis of repeatability of ns-TiOx, epoxy, and NC slides in protein and antibodyevaluated according to assay and substrate buffer condition in a function of spotted pro

10 Nanostructured TiOx coatings as substrates for protein arrays / R. Carbone et al. / Anal. Biochem. 394 (2009) 7–12

lue in the same graph (see Materials and methods), and in a sepa-rate graph we present data from the streptavidin–Cy3 binding as-say in NC slides that were analyzed with a different PMT value. Theresults show that ns-TiOx-coated slides presented significantlyhigher binding features compared with epoxy-coated slides andshowed no saturation at high protein concentration if comparedwith the binding curve of NC slides, thereby demonstrating thecapability of the substrate to stably bind proteins with directdependence on protein adsorption with concentrations up to0.5 mg/ml.

The mechanism of protein binding on ns-TiOx coating is farfrom fully described and understood, although we speculate thatproteins could bind to ns-TiOx substrate with similar mechanismscompared with other more conventional TiO2-based surfaces[26,27], that is, mainly through adsorption. Recently, we thor-oughly investigated the mechanism of streptavidin/ns-TiOx bind-ing, showing that carboxyl-terminated protein residues, exposedat the surface of streptavidin molecules, may play a major role inthe stable adsorption of streptavidin to ns-TiOx surfaces [20]. Spe-cific features, such as surface nanotopography/roughness andchemical composition/hydrophilicity, can also strongly influenceprotein adsorption properties [13–15,28–30]. We are currently

array assays. Specific patterns of intraslide M-CVs (A) and interslide A-CVs (B) aretein concentration.

Fig. 6. Overall intraslide W-CVs and interslide B-CVs of ns-TiOx, epoxy, and NCslides. Graphs present the average values of all M-CVs and A-CVs at differentprotein concentrations ± 1.96 (SE), resulting in a general estimate on repeatabilityand variation per substrate/condition.

Nanostructured TiOx coatings as substrates for protein arrays / R. Carbone et al. / Anal. Biochem. 394 (2009) 7–12 11

investigating the role of these parameters in detail, and our obser-vations will be described in a separate study.

To further characterize ns-TiOx coating, we then performedmouse IgG antibody array assay followed by detection with Cy3-la-beled anti-mouse IgG. We again compared the performance of ns-TiOx coating with that of NC slides and epoxy-coated slides. MouseIgG antibodies were spotted (20 replicates/concentration) in twodifferent spotting buffers (5% glycerol and 1.5 M betaine) at differ-ent concentrations on each substrate and were incubated up to 5 hon ns-TiOx slides to allow efficient binding. For commercial sub-strates, we followed specific manufacturer instructions.

After blocking and washing, the unbound slides were then incu-bated overnight with 1 lg/ml Cy3-labeled anti-mouse IgG anti-body. After a series of washing in 0.5% PBST and finally indeionized H2O, slides were dried under a hood and then scannedat 532 nm at 33% laser power with an opportune PMT accordingto the specific surface (see Materials and methods for details).Fig. 3B shows that ns-TiOx slides presented a stronger bindingcapacity compared with epoxy-coated slides, in particular for anti-body concentration commonly used in antibody or ELISA-based ar-ray assays (i.e., from 100 lg to 1 mg/ml).

The overall comparative performance was analyzed by evaluat-ing the S/N ratio. As shown in Fig. 4A, the S/N ratio of ns-TiOx slidesin streptavidin–Cy3 array assay is extremely high due to the verylow value of standard deviations on background spots. Mostremarkably, the S/N ratio calculated for each protein concentrationfor mouse IgG microarray experiments (Fig. 4B) is consistentlysimilar or higher for the ns-TiOx coating for all of the tested spot-ting buffers, in particular for the glycerol buffer, assessing the com-petitive performance of our slides. This is a relevant result giventhat the S/N ratio represents one of the most important featuresfor efficient protein detection whenever sensitivity and specificityare required.

Repeatability of ns-TiOx coating in protein array assay: analysis of CV

To fully characterize the performance of ns-TiOx coating, weanalyzed the intraslide and interslide CVs according to the specificmicroarray assay (streptavidin–Cy3 and mouse IgG arrays), buffer(glycerol and betaine in the case of mouse IgG arrays), and proteinconcentrations (see Materials and methods for details).

The patterns of variation of the intraslides and interslidesaccording to protein concentration CV are shown in Fig. 5A andB, respectively. The analysis of the intraslide M-CVs showed, forall of the tested substrates and experimental conditions, goodrepeatability (CVs <25%) except for the epoxy substrate in the caseof the mouse IgG assay with glycerol buffer. Moreover, the ns-TiOxsubstrate showed behavior comparable to that of the commercialNC slides.

The interslide A-CV analysis (Fig. 5B) showed different resultsaccording to the specific substrate, assay, and condition. However,the ns-TiOx presented the best repeatability in two of the threeconditions analyzed. In particular, we found that the use of glycerolbuffer conferred the more consistent and regular performance tothe ns-TiOx coating in the antibody array experiment, suggestingthat the choice of the optimal combination of substrate–buffers–assay condition is definitely a mandatory requisite to correctly ap-proach specific protein microarray assays.

Finally, we provided an overall comparison of W-CV and B-CV(Fig. 6) of the different substrates that further demonstrated thesimilar repeatability of ns-TiOx coating compared with the com-mercial slides analyzed. Altogether, these data suggest that clus-ter-assembled ns-TiOx slides present excellent performance inmodel protein microarray experiments, suggesting that this coat-ing is a potential novel substrate for specific microarrayapplications.

Conclusion

We have performed a systematic characterization of cluster-assembled ns-TiOx coating as protein binding surface, comparingits performance with that of most common commercial substratesin protein and antibody microarray assays.

Cluster-assembled ns-TiOx showed remarkable properties inboth assays, in terms of autofluorescence background, proteinadsorption, and S/N ratio, suggesting its possible use in differentprotein microarray applications. Moreover, the reproducibility ofour results, as demonstrated by the accurate statistical intraslide/interslide data provided, further strengthens the relevance of ourfindings for future exploitation of ns-TiOx coatings in specificmicroarray settings.

The performance of our coating depends on the peculiar mor-phology at the nanoscale resulting from the random stacking ofhighly defected nanoparticles deposited on the substrate, as dis-cussed in detail in Refs. [20,31].

Finally, we note that our SCDB deposition technology [32] isfully compatible with miniaturization and microfabrication on avariety of supports (e.g., glass, quartz, silicon) [33], suggesting thatcluster-assembled ns-TiOx can represent a novel high-perfor-mance coating for immunodetection on miniaturized biochipsand biosensors.

Acknowledgments

The authors thank Loris De Cecco (IFOM–IEO campus) for sup-port in microarray analysis, Samuele Venturini and Aldo Novali(Tethis) for technical assistance, and Alessandro Podestà and Si-mone Bovio (CIMAINA) for atomic force microscopycharacterization.

References

[1] P. Bertone, M. Snyder, Advances in functional protein microarray technology,FEBS J. 272 (2005) 5400–5411.

[2] M.M. Ling, C. Ricks, P. Lea, Multiplexing molecular diagnostics andimmunoassays using emerging microarray technologies, Expert Rev. Mol.Diagn. 7 (2007) 87–98.

12 Nanostructured TiOx coatings as substrates for protein arrays / R. Carbone et al. / Anal. Biochem. 394 (2009) 7–12

[3] D. Saerens, G.H. Ghassabeh, S. Muyldermans, Antibody technology inproteomics, Brief Funct. Genomic Proteomic 7 (2008) 275–282.

[4] J. Sobek, K. Bartscherer, A. Jacob, J.D. Hoheisel, P. Angenendt, Microarraytechnology as a universal tool for high-throughput analysis of biologicalsystems, Comb. Chem. High Throughput Screen. 9 (2006) 365–380.

[5] P. Angenendt, J. Glokler, D. Murphy, H. Lehrach, D.J. Cahill, Toward optimizedantibody microarrays: a comparison of current microarray support materials,Anal. Biochem. 309 (2002) 253–260.

[6] C. Schnack, B. Hengerer, F. Gillardon, Identification of novel substrates for Cdk5and new targets for Cdk5 inhibitors using high-density protein microarrays,Proteomics 8 (2008) 1980–1986.

[7] S.L. Seurynck-Servoss, A.M. White, C.L. Baird, K.D. Rodland, R.C. Zangar,Evaluation of surface chemistries for antibody microarrays, Anal. Biochem.371 (2007) 105–115.

[8] M. Cretich, G. Pirri, F. Damin, I. Solinas, M. Chiari, A new polymeric coating forprotein microarrays, Anal. Biochem. 332 (2004) 67–74.

[9] A.J. Nijdam, M. Ming-Cheng Cheng, D.H. Geho, R. Fedele, P. Herrmannb, K.Killianb, V. Espinab, E.F. Petricoin, L.A. Liotta, M. Ferrari, Physicochemicallymodified silicon as a substrate for protein microarrays, Biomaterials 28 (2007)550–558.

[10] M. Reck, F. Stahl, J.G. Walter, M. Hollas, D. Melzner, T. Scheper, Optimization ofa microarray sandwich–ELISA against hINF-c on a modified nitrocellulosemembrane, Biotechnol. Prog. 23 (2007) 1498–1505.

[11] A. Ressine, S. Ekstrom, G. Marko-Varga, T. Laurell, Macro-/nanoporous siliconas a support for high-performance protein microarrays, Anal. Chem. 75 (2003)6968–6974.

[12] X. Zhou, J. Zhou, Protein microarrays on hybrid polymeric thin films preparedby self-assembly of polyelectrolytes for multiple-protein immunoassays,Proteomics 6 (2006) 1415–1426.

[13] M. Han, A. Sethuraman, R.S. Kane, G. Belfort, Nanometer-scale roughnesshaving little effect on the amount or structure of adsorbed protein, Langmuir19 (2003) 9868–9872.

[14] K. Rechendorff, M.B. Hovgaard, M. Foss, V.P. Zhdanov, F. Besenbacher,Enhancement of protein adsorption induced by surface roughness, Langmuir22 (2006) 10885–10888.

[15] A. Dolatshahi-Pirouz, K. Rechendorff, M.B. Hovgaard, M. Foss, J. Chevallier, F.Besembacher, Bovine serum albumin adsorption on nano-rough platinumsurfaces studied by QCM-D, Colloids Surf. B 66 (2008) 53–59.

[16] E. Barborini, P. Piseri, P. Milani, A pulsed microplasma source of high intensitysupersonic carbon cluster beams, J. Phys. D: Appl. Phys. 32 (1999) L105–L109.

[17] I.N. Kholmanov, E. Barborini, S. Vinati, P. Piseri, A. Podestà, C. Ducati, C. Lenardi,P. Milani, The influence of the precursor clusters on the structural andmorphological evolution of nanostructured TiO2 under thermal annealing,Nanotechnology 14 (2003) 1168–1173.

[18] R. Carbone, I. Marangi, A. Zanardi, L. Giorgetti, E. Chierici, G. Berlanda, A.Podestà, F. Fiorentini, G. Buongiorno, P. Piseri, P.G. Pelicci, P. Milani,Biocompatibility of cluster-assembled nanostructured TiO2 with primary andcancer cells, Biomaterials 27 (2006) 3221–3229.

[19] R. Carbone, L. Giorgetti, A. Zanardi, I. Marangi, E. Chierici, G. Buongiorno, F.Fiorentini, M. Faretta, P. Piseri, P.G. Pelicci, P. Milani, Retroviral microarray-based platform on nanostructured TiO2 for functional genomics and drugdiscovery, Biomaterials 28 (2007) 2244–2253.

[20] L. Giorgetti, G. Bongiorno, A. Podestà, G. Berlanda, P.M. Scopelliti, R. Carbone, P.Milani, Adsorption and stability of streptavidin on cluster-assemblednanostructured TiOx films, Langmuir 24 (2008) 11637–11644.

[21] B. Guilleaume, A. Buness, C. Schmidt, F. Klimek, G. Moldenhauer, W. Huber, D.Arlt, U. Korf, S. Wiemann, A. Poustka, Systematic comparison of surfacecoatings for protein microarrays, Proteomics 5 (2005) 4705–4712.

[22] U. Sauer, L. Bodrossy, C. Preininger, Evaluation of substrate performance for amicrobial diagnostic microarray using a four parameter ranking, Anal. Chim.Acta 632 (2009) 240–246.

[23] P. Milani, S. Iannotta, Cluster Beam Synthesis of Nanostructured Materials,Springer, Berlin, 1999.

[24] E. Barborini, A.M. Conti, I.N. Kholmanov, P. Piseri, A. Podestà, P. Milani, C.Cepek, O. Sakho, R. Macovez, M. Sancrotti, Nanostructured TiO2 films with 2 eVoptical gap, Adv. Mater. 17 (2005) 1842–1846.

[25] K. Derwinska, U. Sauer, C. Preininger, Reproducibility of hydrogel slides in on-chip immunoassays with respect to scanning mode, spot circularity, and datafiltering, Anal. Biochem. 370 (2007) 38–46.

[26] S.R. Sousa, M.M. Bras, P. Moradas-Ferreira, M.A. Barbosa, Dynamics offibronectin adsorption on TiO2 surfaces, Langmuir 23 (2007) 7046–7054.

[27] A. Klinger, D. Steinberg, D. Kohavi, M.N. Sela, Mechanism of adsorption ofhuman albumin to titanium in vitro, J. Biomed. Mater. Res. 36 (1997) 387–392.

[28] D. Khang, S.Y. Kim, P. Liu-Snyder, G.T. Palmore, S.M. Durbin, T.J. Webster,Enhanced fibronectin adsorption on carbon nanotube/poly(carbonate)urethane: independent role of surface nano-roughness and associatedsurface energy, Biomaterials 28 (2007) 4756–4768.

[29] L.T. Allen, M. Tosetto, I.S. Miller, D.P. O’Connor, S.C. Penney, I. Lynch, A.K.Keenan, S.R. Pennington, K.A. Dawson, W.M. Gallagher, Surface-inducedchanges in protein adsorption and implications for cellular phenotypicresponses to surface interaction, Biomaterials 27 (2006) 3096–3108.

[30] M.E. Nagassa, A.E. Daw, W.G. Rowe, A. Carley, D.W. Thomas, R. Moseley,Optimisation of the hydrogen peroxide pre-treatment of titanium: surfacecharacterisation and protein adsorption, Clin. Oral Implants Res. 19 (2008)1317–1326.

[31] T. Caruso, C. Lenardi, R.G. Agostino, M. Amati, G. Buongiorno, T. Mazza, A.Policicchio, V. Formoso, E. Maccallini, E. Colavita, G. Chiarello, P. Finetti, F.Sutara, T. Skala, P. Piseri, K.C. Prince, P. Milani, Electronic structure of clusterassembled nanostructured TiO2 by resonant photoemission at the Ti L2,3 edge,J. Chem. Phys. 128 (2008) 094704.

[32] K. Wegner, P. Piseri, H. Vahedi Tafreshi, P. Milani, Cluster beam deposition: atool for nanoscale science and technology, J. Phys. D: Appl. Phys. 39 (2006)R439–R459.

[33] E. Barborini, S. Vinati, M. Leccardi, P. Repetto, G. Bertolini, O. Rorato, L.Lorenzelli, M. Decarli, V. Guarnieri, C. Ducati, P. Milani, Batch fabrication ofmetal oxide sensors on micro-hotplates, J. Micromech. Microeng. 18 (2008)055015.