Registered charity number: 207890Registered charity number: 207890

Cre

dit

on

ly if

req

uir

ed

Showcasing joint research from Kraków (Jagiellonian

University, AGH Univ. Sci. Techn.) and Athens (INRaSTES,

IAMPPNM/ NCSR Demokritos).

Imaging and spectroscopic comparison of multi-step methods

to form DNA arrays based on the biotin-streptavidin system

Multi-step methods using biotin-streptavidin system to form DNA

arrays are characterized with surface science techniques after

each deposition/reaction step. Molecular distributions of DNA

arrays reveal features characteristic for each protocol. Optimal

immobilization procedure provides eff ective gene mutation

detection.

Analystwww.rsc.org/analyst

ISSN 0003-2654

PAPERChristopher J. Easley et al.A microfl uidic interface for the culture and sampling of adiponectin from primary adipocytes

Volume 140 Number 4 21 February 2015 Pages 947–1360

As featured in:

See Katarzyna Gajos et al.,Analyst, 2015, 140, 1127.

www.rsc.org/analyst

Cre

dit

on

ly if

req

uir

ed

Analyst

PAPER

Cite this: Analyst, 2015, 140, 1127

Received 22nd May 2014,Accepted 9th December 2014

DOI: 10.1039/c4an00929k

www.rsc.org/analyst

Imaging and spectroscopic comparison ofmulti-step methods to form DNA arrays based onthe biotin–streptavidin system†

Katarzyna Gajos,*a Panagiota Petrou,b Andrzej Budkowski,a Kamil Awsiuk,a

Andrzej Bernasik,c Konstantinos Misiakos,d Jakub Rysz,a Ioannis Raptisd andSotirios Kakabakosb

Three multi-step multi-molecular approaches using the biotin–streptavidin system to contact-print DNA

arrays on SiO2 surfaces modified with (3-glycidoxypropyl)trimethoxysilane are examined after each depo-

sition/reaction step by atomic force microscopy, X-ray photoelectron spectroscopy and time of flight sec-

ondary ion mass spectrometry. Surface modification involves the spotting of preformed conjugates of

biotinylated oligonucleotides with streptavidin onto surfaces coated with biotinylated bovine serum

albumin b-BSA (approach I) or the spotting of biotinylated oligonucleotides onto a streptavidin coating,

the latter prepared through a reaction with immobilized b-BSA (approach II) or direct adsorption

(approach III). AFM micrographs, quantified by autocorrelation and height histogram parameters (e.g.

roughness), reveal uniform coverage after each modification step with distinct nanostructures after the

reaction of biotinylated BSA with streptavidin or of a streptavidin conjugate with biotinylated oligonucleo-

tides. XPS relates the immobilization of biomolecules with covalent binding to the epoxy-silanized

surface. Protein coverage, estimated from photoelectron attenuation, shows that regarding streptavidin

the highest and the lowest immobilization efficiency is achieved by following approaches I and III,

respectively, as confirmed by TOF-SIMS microanalysis. The size of the DNA spot reflects the contact

radius of the printed droplet and increases with protein coverage (and roughness) prior to the spotting, as

epoxy-silanized surfaces are hardly hydrophilic. Representative TOF-SIMS images show sub-millimeter

spots: uniform for approach I, doughnut-like (with a small non-zero minimum) for approach II, both with

coffee-rings or peak-shaped for approach III. Spot features, originating from pinned contact lines and

DNA surface binding and revealed by complementary molecular distributions (all material, DNA, strepta-

vidin, BSA, epoxy, SiO2), indicate two modes of droplet evaporation depending on the details of each

applied approach.

1. Introduction

The fabrication of DNA microarrays plays a crucial role in bio-sensors and biomedical research. Different techniques have

been developed to prepare such arrays, including printing1–3

and photolithography4 as the most popular approaches. Sensi-tivity and reliability of DNA microarrays depend critically onthe immobilized DNA density, spot uniformity and repeatabil-ity. Therefore, any research aiming at optimized DNA immobil-ization and array preparation is intertwined with thedevelopment of methods for spot quality determination.

Many spectroscopic techniques have been recently appliedto characterize surfaces with immobilized oligonucleotides orDNA arrays. Especially two methods with unique specificityand sensitivity namely X-ray Photoelectron Spectroscopy (XPS)and Time-of-Flight Secondary Ion Mass Spectrometry(TOF-SIMS) have been widely applied to examine DNAspots.1–3,5,6 Both techniques rely on surface-sensitive character-istic signals7 that reveal the chemical state and atomic concen-tration (XPS) or the molecular composition (TOF-SIMS).

†Electronic supplementary information (ESI) available: Cross-sectional analysisof AFM images, representative high resolution XPS spectra, TOF-SIMS micro-analysis using C11H8NO

+ ions, TOF-SIMS imaging of DNA spots with negativeand positive ions. See DOI: 10.1039/c4an00929k

aM. Smoluchowski Institute of Physics, Jagiellonian University, Łojasiewicza 11,

30-348 Kraków, Poland. E-mail: kasia.fornal@uj.edu.plbInstitute of Nuclear & Radiological Science & Technology, Energy & Safety, NCSR

Demokritos, 15310 Aghia Paraskevi, GreececAcademic Centre for Materials and Nanotechnology, AGH-University of Science

and Technology, Al. Mickiewicza 30, 30-059 Kraków, PolanddInstitute of Nanoscience and Nanotechnology, NCSR Demokritos, 15310 Aghia Paraskevi,

Greece

This journal is © The Royal Society of Chemistry 2015 Analyst, 2015, 140, 1127–1139 | 1127

Publ

ishe

d on

09

Dec

embe

r 20

14. D

ownl

oade

d by

UN

IWE

RSY

TE

T J

AG

IEL

LO

NSK

I on

02/

02/2

015

16:1

6:04

.

View Article OnlineView Journal | View Issue

Moreover, by employing imaging modes of these techniquesone can examine spot uniformity and DNA spatial distribution.Such information is important because DNA spots are oftennot homogeneous (e.g. with features such as ‘coffee rings’ ontheir perimeter). However, studies reported so far on spectro-scopic chemical imaging have been limited to DNA spots andarrays prepared on commercial amine-reactive polymer-coatedslides,1,5,6 silane-treated glass slides or silicon supports.2,3,8

The recent quest for hybridization-efficient strategy of oligo-nucleotide immobilization has revealed very promisingmethods by utilizing the biotin–streptavidin system.4,9–13 Miret al. reported that the streptavidin-formed bridge between thesurface and biotin-labelled oligonucleotides minimizes non-specific binding and imposes anti-fouling characteristics incomparison with thiolated DNA brushes.10 However, the appli-cation of proteins as a bridge to immobilize DNA probesinvolves multiple surface-modification stages. When multi-stepprocedures are applied, surface examination after each individ-ual modification stage is required but hardly provided. In linewith this demand, we have recently applied complementaryspectroscopic (XPS, TOF-SIMS) but also microscopic tech-niques (Atomic Force Microscopy, AFM) to characterize, step-by-step, the surfaces in the course of different multi-step pro-cedures employing the biotin–streptavidin system to immobi-lize uniformly oligonucleotides on silicon substrates modifiedwith (3-aminopropyl)triethoxysilane (APTES).9

This study is an extension of both immobilization protocolsusing biotin–streptavidin and step-by-step characterizationmethods to spatially-resolved oligonucleotide micro-array spotsfabricated on silicon surfaces modified with (3-glycidoxy-propyl)trimethoxysilane (GOPS). A surface analysis of differentmulti-step DNA immobilization methods reveals the nano-structure (AFM), chemistry (XPS) and molecular coverage (XPSand TOF-SIMS) both within and outside the DNA spots todetermine the most and the least effective immobilizationapproaches with respect to streptavidin.

Apart from biochemistry, oligonucleotide arrays are asubject of intensive research of soft matter physics. The salientfeatures of DNA spot formation have been revealed14–18 andapplied to describe observations made for the arrays on rela-tively simple substrates.1–3,5,6,8 This work extends such studiesto DNA arrays formed through solid pin contact printing ontothe surfaces covered with proteins. Oligonucleotide spots areimaged with TOF-SIMS to reveal their characteristic features,specific for each immobilization approach. Multi-moleculardistributions are analyzed within spot formation models todiscuss the size and the morphology of DNA spots obtainedwithin each of the three immobilization protocols.

2. Experimental2.1 Materials

Silicon wafers were purchased from Montco Silicon Techno-logies, Inc. (Spring City, PA, USA). 3-Glycidoxypropyl(tri-methoxysilane) (GOPS), N-[2-hydroxyethyl]piperazine-N′2-

ethanesulfonic acid] (HEPES) and bovine serum albumin (BSA;Cohn fraction V, RIA grade) were purchased from SigmaChemical Co., St Louis, MO, USA. 6-((Biotinoyl)amino)hexa-noic acid sulfosuccinimidyl ester (sulfo-NHS-LC-biotin) andstreptavidin were from Pierce (Rockford, IL, USA). Ethylene-diaminetetracetic acid (EDTA) was from Fluka Chemie (Buchs,Switzerland). The biotinylated oligonucleotide probe corres-ponding to the wild-type sequence of the 3099delT mutationin the BRCA1 gene (biotin-5′-TTAAAACTAAATGTAA-GAAAAATC-3′), a dually labelled probe with the same sequence(5′-end biotin and 3′-end AlexaFluor 546), as well as a fluores-cently-labelled complementary oligonucleotide (5′-AAAAA-GCAGATTTTTCTTACATTTAGTTTTAACAA-3′, with AlexaFluor546 attached to the 5′-end) was obtained from VBC Biotech(Vienna, Austria). Biotinylated BSA (b-BSA) was preparedaccording to a published method.19,20

2.2 Epoxy-silanized surfaces for protein adsorption

Silicon substrates on which a 1000 nm thick silicon dioxide(SiO2) layer was thermally grown by low pressure chemicalvapor deposition were cleaned and hydrophilized by immer-sion in the Piranha solution (1 : 1 (v/v) H2O2–H2SO4) for20 min, followed by washing with distilled water and dryingunder nitrogen stream. Next, the hydrophilized SiO2 surfaceswere modified by immersion in a 1% (v/v) GOPS solution in anhy-drous toluene for 18 hours, then washed 3 times sequentiallywith toluene and ethanol under sonication and dried with N2.

The epoxy-silane layer formation on SiO2 surfacesimmersed in a GOPS solution is determined by the compe-tition between hydrolysis and polymerization of methoxysilanegroups at the hydrated surface and in solution. Mono-mole-cular layers (with 0.95 nm thickness) and aggregates of mole-cules packed in multi-layers are formed preferentially at higherand lower GOPS concentrations, respectively. Protein immobil-ization to the epoxy-modified surface takes place through atwo-step mechanism. First, physical adsorption of the proteinis promoted. Second, the epoxy-activated surface reacts withdifferent (amino, thiol, hydroxyl) moieties on the surface ofthe adsorbed protein to form strong bonds with minimalchemical modification of the protein.

2.3 Preparation of oligonucleotide arrays

Oligonucleotides were printed onto GOPS modified surfacesusing a BioOdyssey Calligrapher Miniarrayer (BioRad Labo-ratories Inc.) equipped with solid pins with a diameter of375 microns (Arrayit Corporation, Sunnyvale, CA, USA).

Three different approaches to contact-print DNA arrays onepoxy-silanized SiO2 surfaces (A), shown schematically inFig. 1, have been implemented to prepare spatially-resolvedareas of oligonucleotide-modified surfaces. The first approachstarted with the adsorption of b-BSA from a 100 μg mL−1 solu-tion in a 50 mM phosphate coating buffer, pH 7.0, at roomtemperature (RT) (B). Next, blocking of the free surface siteswas performed by immersion in a 10 mg mL−1 BSA solution incoating buffer for 1 h at RT (C), followed by washing anddrying. Finally, spots of pre-formed streptavidin–biotinylated

Paper Analyst

1128 | Analyst, 2015, 140, 1127–1139 This journal is © The Royal Society of Chemistry 2015

Publ

ishe

d on

09

Dec

embe

r 20

14. D

ownl

oade

d by

UN

IWE

RSY

TE

T J

AG

IEL

LO

NSK

I on

02/

02/2

015

16:1

6:04

. View Article Online

oligonucleotide conjugates (D) were printed. The conjugateswere prepared through the reaction of an 88 μg mL−1 strepta-vidin solution (1.67 μM) with 5 μM of a biotinylated oligonucleo-tide solution, both diluted in coating buffer for 10 min at RTprior to application onto the surface. In the second approach,the last step was replaced by two steps. First, the reaction ofb-BSA coated surfaces with a 100 μg mL−1 streptavidin solutionwas performed in coating buffer for 1 h at RT (E). Then the sur-faces were washed and spots of biotinylated oligonucleotideswere printed from a 5 μM solution in coating buffer (F). Thethird approach consisted of the direct adsorption of streptavidinon the silanized substrates (G) from a 25 μg mL−1 solution incoating buffer for 1 h at RT, followed by washing with coatingbuffer and distilled water. Next, blocking with BSA (H) andprinting spots of biotinylated oligonucleotides (I) was per-formed as described above for the second approach.

All spotted surfaces were kept in a humidity chamber (65%humidity) for 1 h at RT and then were rinsed with coatingbuffer and distilled water, and dried by N2. To enable themicroscopic localization of DNA spot areas hybridization withthe fluorescently labeled complementary probe was performedusing a 500 nM oligonucleotide solution in 1× HEN buffer

(200 mM HEPES, 100 mM NaCl, 2 mM EDTA, pH 8.0) for30 min at RT. The surfaces were then washed with 1× HEN anddistilled water cooled at 4 °C, and dried with N2.

2.4 Preparation of fluorescently-labeled PCR products andhybridization protocol

Fluorescently labeled at the 5′-end PCR products for themutation 3099delT were obtained using a 5′-end reverseprimer labeled with AlexaFluor 647 dye (5′-ACGGCTAA-TTGTGCTCACTGT-3′) and a non-labeled forward primer(5′-GCAACGAAACTGGACTCATTACTC-3′). The PCR protocol wascarried out on a Bio-Rad thermal-cycler (Bio-Rad, Hercules,CA) using a 50 μL total reaction volume containing 200 nggenomic DNA, 0.4 μM of each primer, 1× PCR buffer (20 mMTris-HCl, 5 mM KCl, pH 8.4), 1.5 mM MgCl2, 0.2 mM dNTPs,and 1.25 U iTaq DNA polymerase (Bio-Rad, Hercules, CA). Theprotocol involved an initial step at 94 °C for 5 min, 20 thermalcycles (94 °C for 45 s, annealing at 55 °C for 30 s decreased by0.5 °C at every subsequent cycle, 72 °C for 45 s, followed by 15thermal cycles: 94 °C for 45 s, 48 °C for 30 s and 72 °C for45 s) and a final extension step at 72 °C for 7 min. The PCRproducts were denatured at 95 °C for 5 min, cooled on ice for1 min and immediately diluted 1 : 1 with 2× HEN buffer(200 mM HEPES, 100 mM NaCl, 2 mM EDTA, pH 8.0) contain-ing 0.5% (w/v) SDS. For the hybridization, 50 μL of thismixture was loaded onto the spotted slides. After incubationfor 30 min at 30 °C, the slides were sequentially washed withserially diluted HEN solutions (from 1× HEN up to 0.063×HEN) and finally with distilled water cooled at 4 °C. The slideswere dried under nitrogen stream and scanned using a PerkinElmer Gx microarray scanner (Perkin-Elmer, Norwalk, CT). Thespot fluorescence intensity was determined using the ScanArray Express System Software.

2.5 AFM surface characterization

Topography and phase contrast images of the modified SiO2

substrates, after each step of every oligonucleotide immobiliz-ation approach (with samples’ names corresponding to thesteps defined in Fig. 1) were collected under ambient con-ditions (in air and at room temperature) with an Atomic ForceMicroscope (Agilent 5500, Agilent Technologies) working in anon-contact mode. AFM probes with a spring constant ofabout 2 N m−1, small tip radius (7 nm) and resonant frequen-cies of about 70 kHz were used. The set point and all gainswere adjusted to obtain minimal noise and a clear image ofthe examined surface. AFM topographic micrographs weretaken at several different areas on each sample’s surface. ForDNA arrays these areas were located within printed oligo-nucleotide spots. The localization of DNA spots under a lightmicroscope was possible only for sample D. Measurements ofthe samples F and I were performed after additional hybridiz-ation of the samples with complementary targets labeled fluor-escently to enable the visualization of DNA array spots underfluorescence microscopy. Such samples after hybridization aredenoted as F* and I*, respectively. The localization of DNAspots with respect to characteristic surface marks enabled the

Fig. 1 Schematic representation of three approaches for oligonucleo-tide immobilization onto epoxy-silanized SiO2 surfaces (A) based on thestreptavidin–biotin system: I (A–B–C–D), II (A–B–C–E–F) and III (A–G–H–I). The first method (I) involves the adsorption of biotinylated BSA (B),blocking of free surface sites with BSA (C), and printing spots of pre-formed streptavidin–biotinylated oligonucleotide conjugate (D). Alterna-tively, in the second approach (II) the spotting of the biotinylated oligo-nucleotide (F) is preceded by the reaction of surface immobilizedbiotinylated BSA with streptavidin (E). Finally, the third protocol (III) con-sists of the adsorption of streptavidin (G) and blocking with BSA (H), fol-lowed by the printing of biotinylated oligonucleotide spots (I).

Analyst Paper

This journal is © The Royal Society of Chemistry 2015 Analyst, 2015, 140, 1127–1139 | 1129

Publ

ishe

d on

09

Dec

embe

r 20

14. D

ownl

oade

d by

UN

IWE

RSY

TE

T J

AG

IEL

LO

NSK

I on

02/

02/2

015

16:1

6:04

. View Article Online

examination of spot areas using AFM. Performed measure-ments showed that the used AFM setup does not resolvehybridization effects.

Surface vertical nanostructures were specified using theparameters of height distribution in a topographic AFMimage, such as the average height (<h>), the root-mean squareroughness (RMS) and the skewness. In addition, lateral nano-structure parameters were provided based on autocorrelationanalysis of topographic images by the doubled width-at-half-maximum (2whm) of the radial averaged autocorrelation func-tion. AFM micrographs were analyzed with the WSxM softwareprovided by Nanotec Electronica S.L. (downloadable at http://www.nanotec.es).21

2.6 XPS surface characterization

X-ray photoelectron spectroscopy measurements were per-formed using a Versa Probe II (PHI) spectrometer equippedwith an Al Kα radiation source (1486.6 eV). A photoelectrontake-off angle Θ equal to 45° was applied for all measure-ments. A dual-beam charge neutralizer was used to compen-sate the charge-up effect. High resolution spectra wererecorded with an analyzer pass energy of 23.5 eV. The operat-ing pressure in the analytical chamber was less than 5 × 10−7

Pa. For uniform samples (A–C, E, G–H) XPS spectra wereacquired from two non-overlapping 1 mm × 1 mm areas. Forthe selected DNA array with spots visible under ambient light(sample D), XPS spectra were acquired from 300 μm × 300 μmareas located both inside and outside the oligonucleotidespots. All XPS peaks were charge referenced to the neutral(C–C) carbon C 1s peak at 284.8 eV. The spectrum backgroundwas subtracted using the Shirley method.

2.7 TOF-SIMS surface characterization

After each step of the immobilization procedure the samplesurface (samples A–I) was analyzed using the TOF.SIMS 5(ION-TOF GmbH) instrument, equipped with a 30 keVbismuth liquid metal ion gun. The Bi3

+ clusters were used asthe primary ions with the ion dose density lower than 2 × 1011

ion cm−2 to ensure static mode conditions. For micro-analysis,positive and negative ion static TOF-SIMS spectra wereacquired from several non-overlapping 100 μm × 100 μm areas.In the case of DNA arrays, the sampling areas were locatedboth inside and outside the spots visible in the surveyTOF-SIMS maps. In turn, TOF-SIMS images were obtained in astage scan mode by scanning: 1 mm × 1 mm area with resolu-tion 1024 × 1024 points (images of single spots) and 2.5 mm ×2.5 mm area with resolution 1280 × 1280 points (maps ofoligonucleotide arrays). A low energy electron flood gun wasused for charge compensation. High mass resolution spectra,with minimal (m/Δm) > 5300 at C4H5

+ and C4H− peaks, were

acquired. Mass calibration was performed with H+, H2+, CH+,

C2H2+ and C4H5

+ peaks for the positive secondary ion spectra,and with CH−, C2H

−, C3− and C4H

− peaks for the negativeones. All TOF-SIMS data were normalized by total ionintensity.

3. Results and discussion3.1. Surface analysis of multi-step oligonucleotideimmobilization methods

The nanostructure, chemistry and coverage of epoxy-silanizedsurfaces on which oligonucleotide spots were created followingthree multi-step methods involving the biotin–streptavidinsystem (Fig. 1) are analyzed step-by-step using complementarymicroscopic and spectroscopic techniques. AFM topographicmicrographs reveal the nanostructure of biomolecular layersand follow its evolution upon consecutive surface-modificationstages. High resolution XPS spectra give insight into thechemical state of both biomolecules and the epoxy-silanizedsurface for three different immobilization methods, whilecomplete XPS data yield an estimation of the biomolecularsurface density. Finally, a TOF-SIMS micro-analysis of immobi-lized streptavidin confirms the XPS conclusion on the effective-ness of the three methods to immobilize oligonucleotides.

3.1.1 AFM analysis of the nanostructured molecular over-layer. Representative AFM topographic micrographs (Fig. 2),recorded from epoxy-silanized surfaces after successive stepsof the three multi-molecular immobilization approaches(Fig. 1), reveal in all cases lateral uniformity of a nanostruc-tured molecular overlayer, described as a random set ofsurface features. A cross-sectional analysis of the AFM images,together with that of bare SiO2, is presented in Fig. ESI-1 of

Fig. 2 Representative topographic AFM micrographs recorded fromepoxy-silanized SiO2 surfaces after the successive steps of the differentapproaches followed for the immobilization of capture oligonucleotidesin spots (cf. Fig. 1 for used notation). For the samples F* and I* measure-ments were performed after the additional hybridization with the fluor-escently labeled complementary target oligonucleotide to localize theDNA spots under a fluorescence microscope. To maximize the contrastdifferent height ranges were applied with Δz = 5 nm (A–C, G–I*) and10 nm (D, E–F*).

Paper Analyst

1130 | Analyst, 2015, 140, 1127–1139 This journal is © The Royal Society of Chemistry 2015

Publ

ishe

d on

09

Dec

embe

r 20

14. D

ownl

oade

d by

UN

IWE

RSY

TE

T J

AG

IEL

LO

NSK

I on

02/

02/2

015

16:1

6:04

. View Article Online

ESI.† The AFM inspection of the surfaces reveals several para-meters of the nanostructures created on the surfaces modifiedstep-by-step with different biomolecules (Fig. 3). In particular,the lateral nanostructure is characterized by the mean size ofsurface features 2whm, provided by the doubled width-at-half-maximum of a radial averaged autocorrelation function com-puted for a given AFM image. In turn, the vertical nanostruc-ture is specified using the parameters of height distribution inan AFM image, such as average height (<h>, the distribution’smean), root mean square roughness (RMS, the distribution’sspread) and surface skewness (the distribution’s asymmetry).

The GOPS modified SiO2 substrate (sample A) is character-ized by surface feature’s size of 2whm = 17 nm, that accordswith literature reports of structural grains of the GOPS filmwith a size of about 10–20 nm.22 Its vertical nanostructure isdescribed by positive surface skewness, surface roughnessRMS = 0.5(±0.1) nm that accords with our previous report[0.65(±0.02) nm]23 and average height <h> = 1.3 (±0.1) nm thatis in the literature range (<h> = 1.1 nm23 and thickness of0.75 nm,22 based on AFM22,23 and up to 1.4 nm from ellipso-metry measurements).24

The adsorption of biotinylated BSA (sample B) or strepta-vidin (sample G) and the subsequent blocking with BSA(samples C and H) results in the creation of a biomolecularoverlayer with vertical nanostructural parameters distinctively

larger than those determined for a bare GOPS film (Fig. 3).The average height <h> and surface roughness RMS increasewith Δ<h> ∼ 0.6 nm and ΔRMS ∼ 0.2 nm, respectively. Positivesurface skewness reflects a coarse surface dominated by peaks.The size of the determined surface features, 2whm ∼18–20 nm, is comparable to 13–17 nm and 17–21 nm predictedas the apparent size of adsorbed streptavidin (4.5 × 4.5 ×5 nm3) and BSA (8 × 8 × 3 nm3), respectively. The predictedvalues are due to broadening caused by the AFM tip (with7 nm radius) for half-spherical25 and spherical26 shapes ofimmobilized molecules.

Striking changes in the biomolecular overlayer structure areobserved when the adsorbed biotinylated BSA reacts withstreptavidin (sample E) or its conjugate with a biotinylated oli-gonucleotide (sample D). A huge increase in average AFMheight <h> to 4.4–5.4 nm and surface roughness RMS to1.3–2.2 nm (Fig. 3) reflects the formation of complexesbetween biotinylated BSA and streptavidin on the studiedsurface. Similar observations were reported for amino-silanized surfaces (with somewhat smaller values of Δ<h> ∼2.4 nm and ΔRMS ∼ 0.3 nm).9 Such changes in the biomole-cular layer roughness are accompanied by a drastic reduction(0.8 nm) of surface skewness, down to values close to zero,indicating the coexistence of peaks and valleys on the samesurface. The lateral nanostructure is also modified, with the

Fig. 3 Lateral (a) and vertical (b–d) nanostructure parameters of SiO2 surfaces corresponding to the successive steps of the three oligonucleotideimmobilization approaches (cf. Fig. 1). (a) Mean size of surface features 2whm determined from topographic AFM images. (b) Mean values of averageheight <h>, (c) root mean square roughness RMS and (d) skewness, as determined from AFM height histograms. Error bars are standard deviations,each determined from 4 to 6 AFM images or height histograms of the same surface.

Analyst Paper

This journal is © The Royal Society of Chemistry 2015 Analyst, 2015, 140, 1127–1139 | 1131

Publ

ishe

d on

09

Dec

embe

r 20

14. D

ownl

oade

d by

UN

IWE

RSY

TE

T J

AG

IEL

LO

NSK

I on

02/

02/2

015

16:1

6:04

. View Article Online

size of surface features increasing up to 21–24 nm as a resultof streptavidin–biotinylated BSA complex formation. Inaddition, a somewhat smaller surface roughness RMS and anaverage height <h> are observed for sample D in comparisonwith E (Fig. 3). This together with the respective AFM images(cf. Fig. 2D and 2E) suggests the creation of a more denselypacked immobilized molecule layer when a streptavidin conju-gate with a biotinylated oligonucleotide is deposited bycontact-printing rather than the whole surface covered with astreptavidin solution.

Finally, the surfaces prior to (samples E and H) and after(samples F* and I*) the printing of biotinylated oligonucleo-tide spots show similar parameter values with respect to bio-molecular layer nanostructures (Fig. 3). It should be noted thatthe reaction of surface immobilized streptavidin with themuch smaller biotinylated oligonucleotides is not resolvedusing AFM and the same applies to the hybridization of theoligonucleotides immobilized onto the samples F and I withfluorescently labelled complementary sequences, resulting insamples F* and I*, respectively.

3.1.2 Surface chemical state analyzed with XPS. A detailedanalysis of the high resolution C 1s and O 1s core-level XPSspectra yields information about various functional groupspresent at the surface. The C 1s envelope can be resolved intofive contributions (see Fig. 4 and Fig. ESI-2†) referred to theneutral carbon (C–C) at 284.8 eV and four carbons withdifferent electron-deficient environments specific for thegroups: C–O (and C–N) at 286.3 eV, epoxy group at 287.3 eV,CvO (and NCvO) at 288 eV, and O–CvO at 288.8 eV. Sincethe GOPS-modified silicon substrate is expected to formcovalent bonds with biomolecules, it is of interest to analyzethe components of the C 1s spectrum observed at 287.3 eV and288 eV, as they reflect the non-reacted fraction of the epoxygroups of the silanized surface and the CvO (and NCvO)groups characteristic for immobilized biomolecules, respecti-vely. Efficient surface modification with GOPS is indicated bythe estimated abundance (sample A, Fig. 4) of C 1s in theepoxy group equal to 12 (±3)% (cf. ∼6% reported elsewhere).24

This value is however lower than 33.3% expected for an idealepoxy-silane monolayer,24 indicating an incorporated adventi-tious carbon27,28 and the ring-opening reaction of some epoxygroups accompanied by oxidation (contributing to the C–Opeak at 286.3 eV) but not by isomerization to carbonyl groups(the absence of the line centered at 288 eV). Therefore, thesignal from C 1s in CvO (and NCvO) can be taken as charac-teristic for biomolecules. Similar information can be indepen-dently obtained from oxygen in carbonyl groups (CvO*, allbiomolecules) unresolved from that in phosphate (DNA,sample D) groups, that contribute to the O 1s envelope withthe peak at 531.5 eV. The remaining contributions to the O 1score-level XPS spectrum reflect (not shown) oxygen in SiO2

(532.5 eV) and in ether groups (C–O*–C, 534.1 eV).A complete set of XPS data was used in Fig. 5a to character-

ize the surface after each successive reaction step in terms ofthe atomic concentrations of carbon (from C 1s in the epoxygroup, gray columns) – characteristic for the non-reacted

epoxy-groups of the silanized surface (upper and right scales)as well as carbon (from C 1s in CvO and NCvO, dashedcolumns) and oxygen (from O 1s in CvO and the phosphategroup, black columns) – both specific for the immobilized bio-molecules (lower and left scales). The atomic concentration ofcarbonyl (and amide) carbon is comparable within error barsfor all surfaces with that of oxygen in the same group, withnegligible contribution from the phosphate group oxygen ofthe DNA molecules for sample D.

The concentration of carbon corresponding to the chemicalstate that is specific for biomolecules follows that of oxygen inthe same sample and they are anti-correlated for all thesamples (Fig. 5a, cf. Fig. 4 and ESI-2†) with the atomic concen-tration of carbon in the epoxy-groups of the surface. Thisreflects a progressive coverage of the epoxy-silanized surfacewith proteins, accompanied by their covalent binding (stepsA–B–C and A–G–H), and followed by complex formation withstreptavidin that attenuates the C 1s photoelectrons emittedfrom the substrate [steps C–D (sample D inside the DNA spot)and C–E]. The data for sample D are shown in Fig. 5a for twosurface regions: outside (background) and inside (spot) thecontact-printed DNA spots. The similar atomic concentrationsdetermined for sample C and the background of sample D

Fig. 4 High resolution XPS C 1s core-level surface spectra recordedprior to (sample A) and after (sample D) the first immobilizationapproach (cf. Fig. 1). Contributions to the C 1s envelope are shown,referred to various carbon environments (see the legend). Note that theintensities of signals from carbon in the carbonyl OvC (and amideOvCN) groups, specific for biomolecules, and of the epoxy-groups ofthe silane layer (non-reacted) are anti-correlated.

Paper Analyst

1132 | Analyst, 2015, 140, 1127–1139 This journal is © The Royal Society of Chemistry 2015

Publ

ishe

d on

09

Dec

embe

r 20

14. D

ownl

oade

d by

UN

IWE

RSY

TE

T J

AG

IEL

LO

NSK

I on

02/

02/2

015

16:1

6:04

. View Article Online

indicate that the spotting procedure does not affect the surfaceexcept for the printed spot area. The comparable XPS datadetermined for samples G and A also suggest a very weakdirect streptavidin immobilization following immobilizationapproach III. In addition, the results shown in Fig. 5a indicatethat immobilization approach II is less effective than approachI, in terms of surface coverage with biomolecules as a result ofstreptavidin reaction (cf. the steps C–E and C–D).

3.1.3 Biomolecular surface coverage determined from XPS.To evaluate the coverage of epoxy-silanized surfaces with bio-molecules the XPS data were analyzed using the bilayer modelof a biomolecular overlayer on the silane film (Fig. 1),described in detail in our previous publications.9,23,29 As a pre-requisite, the thickness d of the GOPS film was evaluated fromthe attenuation of the Si 2p photoelectrons characteristicfor the substrate. The value obtained, d = 0.95(±0.1) nm,accords with our previous result (0.97(±0.05) nm)23 and other

reports.22,30 The bilayer model allows us to relate the effectiveXPS thickness D of the biomolecular layer:

XN=ZN

XSi=ZSi¼ 1� exp �D=λN cosΘð Þ

exp � Dþ dð Þ=λSi cosθð Þ ð1Þ

with the overlayer-to-substrate composition ratio, i.e. with theratio of XPS intensities XN/XSi (N 1s/Si 2p) recalculated withsensitivity factors Si = λiσi (specified by photoionization cross-section σi

31 and attenuation length λi32). The ratio XN/XSi is

normalized with respect to the molar fraction Zi of the atomsemitting N 1s and Si 2p photoelectrons. For the value of themolar fraction Zi we take the atomic concentrations deter-mined from an independent XPS analysis of the bulk proteinmaterial (ZN = 15%)33 and the non-silanized SiO2 surface (ZSi =34.5 (±0.5)%). The effective XPS thickness D, that justifiesphotoelectron attenuation, allows for the estimation of thesurface coverage with different biomolecules, using their den-sities as the scaling factor (1.37 g cm−3 and 1.39 g cm−3 forproteins and streptavidin–biotinylated oligonucleotide conju-gate, respectively).9

The biomolecular layer surface density evaluated by employ-ing the bilayer model is plotted in Fig. 5b (dashed columns,right scale) along with the input data of the overlayer-to-sub-strate composition ratio N 1s/Si 2p (solid columns, left scale).The values of biomolecular surface density determined for thedifferent samples correlate well with those from the surfacechemical state analysis (cf. Fig. 5a). Moreover, the determinedbiomolecular coverage values resulting from multi-step immo-bilization approaches onto epoxy-silanized silicon surfaces(Fig. 5b, right scale) are comparable with the results obtainedrecently for amino-silanized substrates.9 For the surfacesmodified with GOPS, the subsequent coating with biotinylatedBSA (sample B), blocking with BSA (sample C) and reactionwith streptavidin (sample E) provide values of 0.12 (±0.01),0.24 (±0.01), and 0.76 (±0.5) mg m−2, respectively. A similardeposition/reaction on an APTES-silanized substrate yieldedvalues of 0.43 (±0.11), 0.66 (±0.12), and 2.23 (±0.24) mg m−2,respectively.9 Based on these results, a lower protein coveragewith b-BSA is observed for silicon surfaces modified withGOPS rather than APTES, which accords with our previousobservations made for the gamma globulin molecule adsorp-tion on these two substrates.23 The reduced surface density ofb-BSA molecules and the consequent reduction in the avail-able biotin binding sites onto the surface is expected to lead toa decreased surface coverage with biomolecules after the reac-tion with streptavidin. Such a relation is broken, however,when contact-printing is applied. This is due to the increasedratio of the surface to deposited solution volume achievedcompared with the whole-surface coverage with protein solu-tion. This is the most probable reason why successive surfaceexposures to a streptavidin conjugate with biotinylated oligo(sample D) yield a density of 2.17 (±0.12) mg m−2, aftercontact-printing onto an expoxy-silanized surface, as comparedto 1.31 (±0.22) mg m−2 after the exposure of the whole amino-modified substrate to the same conjugate solution.9

Fig. 5 Surface chemical states (a) and overlayer-to-substrate compo-sition ratio (b) determined with XPS after successive surface reactionsteps (cf. Fig. 1) for uniform surfaces (samples A–C, E, G–H), and forDNA array (sample D): inside (s – spot) and outside the spots (b – back-ground). (a) Atomic concentrations of carbon (from C 1s spectra) in theepoxy group – characteristic for the non-reacted groups of the silanizedsurface (gray upper columns, right scale) and in the carbonyl CvO (andamide COvN) groups (dashed lower columns, left scale) – specific forthe immobilized biomolecules. Atomic concentration of oxygen (from O1s spectra) in the CvO (biomolecules) and in phosphate (DNA) groups(solid lower columns, left scale). (b) Normalized (with sensitivity factors)intensities of the photoelectrons N 1s/Si 2p (solid columns, left scale),and the estimated biomolecular surface density (dashed columns, rightscale).

Analyst Paper

This journal is © The Royal Society of Chemistry 2015 Analyst, 2015, 140, 1127–1139 | 1133

Publ

ishe

d on

09

Dec

embe

r 20

14. D

ownl

oade

d by

UN

IWE

RSY

TE

T J

AG

IEL

LO

NSK

I on

02/

02/2

015

16:1

6:04

. View Article Online

As for surface chemistry analysis, the surface coverage isconcluded to be similar for sample C (0.24 (±0.01) mg m−2)and the background of sample D (0.24 (±0.01) mg m−2). Thisindicates that the spotting procedure does not affect thesurface around the printed spot area. Also, the direct adsorp-tion of streptavidin onto the GOPS-modified substrate follow-ing approach III (sample G) provides an almost negligible (asfor sample A) protein surface density equal to 0.10 (±0.08) mgm−2. This result, together with those for samples D(2.17 (±0.12) mg m−2) and E (0.76 (±0.5) mg m−2) as well as forsample C prior to the biotin–streptavidin reaction(0.24 (±0.01)) indicates that in terms of streptavidin the mostand the least effective immobilization approaches areapproaches I and III, respectively.

3.1.4 TOF-SIMS micro-analysis of immobilized streptavidin.The TOF-SIMS micro-analysis was also applied to evaluate theeffectiveness of the different immobilization approaches interms of surface coverage with streptavidin in order to confirmthe results obtained by XPS. In this case, the surface analysiswas extended to the (background and inner) regions of DNAspots not resolved by XPS (samples F and I). The presence ofstreptavidin at the surface after successive deposition/reactionsteps is revealed through detection of the characteristicTOF-SIMS signals, which correspond to the ion fragmentsC9H8N

+ (Fig. 6) and C11H8NO+ (Fig. ESI-3†) of tryptophan, the

amino acid with the highest abundance in streptavidin mole-cule.9,34 Since the TOF-SIMS intensities characteristic for bio-molecules might be affected by contaminations (e.g. buffers),matrix substrate-induced effects, and even variation of mole-cular distribution within DNA spots, in addition to the notfully resolved mechanism of ion formation, the conclusionsdrawn from an empirical comparison of single TOF-SIMSsignals cannot be without reservations.35 Therefore we restrict

ourselves to a qualitative rather than a quantitative sample-to-sample analysis of the results shown in Fig. 6 and ESI-3.†Thus, the intensity of both TOF-SIMS signals characteristic forstreptavidin can be ranked as follows: the highest one isobtained from the area inside the DNA spot of sample D, lowersignals are observed for samples E–F (both inside and outsidethe spot), very low signals are received for samples G–H–I(inside the spot and in the background), and finally, zerosignal is obtained for samples A–B–C as well as for the back-ground area of the DNA spot of sample D.

These TOF-SIMS results, although qualitative, confirm theconclusions drawn from the XPS analysis on the efficiency ofthe different immobilization protocols in terms of the surfacecoverage with streptavidin and extend them also to samples Fand I. To further confirm these findings, fluorescence imagesof the oligonucleotide spots prepared following the threedifferent approaches were analyzed. The mean fluorescenceintensity values obtained from the spots (prepared on glassslides with fluorescently labeled biotinylated oligonucleotides)are presented in ESI (Fig. ESI-4†). The results show thatapproach I provided almost 4 and 1.7 times higher spot fluo-rescence intensity values compared to approaches III and II,respectively. These results reflect to a great extent the findingsof the TOF-SIMS and XPS analyses of the SiO2 surfaces andconfirm the superiority of immobilization approach I overapproaches II and III.

3.2. Detection of PCR products with DNA arrays preparedaccording to optimal protocol

Oligonucleotide arrays prepared on glass slides following theoptimal immobilization protocol were used for the detectionof 3099delT mutation in PCR products obtained from the DNAof individuals that are not carriers of the mutation (wild-type).In Fig. 7 a representative fluorescence image of an array con-sisting of two 3 × 3 spot grids is provided, the left corres-ponding to the immobilized mutant-type sequence and theright to the wild-type sequence of the 3099delT mutation, after

Fig. 6 TOF-SIMS microanalysis of streptavidin immobilized at thesurface after the successive steps of deposition/reaction involved in thethree immobilization approaches (see Fig. 1 for the notation used). Sec-ondary ion intensities (normalized per total ion intensity) characteristicfor streptavidin (C9H8N

+ from tryptophan) determined for uniform sur-faces (samples A–C, E and G–H) as well as inside (s – spot) and outside(b – background) the oligonucleotide spots (samples D, F and I). Errorbars are standard deviation values determined from repetitive (3–8)measurements of the same surface.

Fig. 7 Fluorescence image of an array on a glass slide consisting of two3 × 3 spot grids, the left corresponding to the immobilized mutant-typesequence and the right to the wild-type sequence of 3099delTmutation, after hybridization with PCR products obtained from the DNAof a wild-type individual and sequential washings as described in theExperimental section.

Paper Analyst

1134 | Analyst, 2015, 140, 1127–1139 This journal is © The Royal Society of Chemistry 2015

Publ

ishe

d on

09

Dec

embe

r 20

14. D

ownl

oade

d by

UN

IWE

RSY

TE

T J

AG

IEL

LO

NSK

I on

02/

02/2

015

16:1

6:04

. View Article Online

hybridization with PCR products obtained from the DNA of awild-type individual. The ratio of the fluorescence signal inten-sities of the wild-type oligonucleotide spots to the mutant-typeoligonucleotides was approx. 100, demonstrating the ability ofthe proposed approach for microarray fabrication to clearlydiscriminate between the mutant and wild-type sequences.

3.3. Molecular distribution in oligonucleotide spots imagedwith TOF-SIMS

3.3.1 Repeatability of characteristic features within arraysof spots. Spot-to-spot variations in molecular distribution canhamper potential applications of DNA spot arrays for quanti-tative determination. To examine this issue in our study, sub-millimeter spot arrays of immobilized oligonucleotides wereanalyzed with TOF-SIMS imaging. The signals SUM DNA−, con-sisting of the ion fragments C4H2N3O

− (m/z = 108) andC5H4N5O

− (m/z = 150) characteristic for guanine, C4N3− (m/z =

90) and C5H4N5− (m/z = 134) specific for adenine, C4H2N3

−

(m/z = 92) for cytosine and C5H5N2O2− (m/z = 125) for

thymine1,7 were used. Characteristic TOF-SIMS chemical mapsof 2.5 mm × 2.5 mm surface areas corresponding to the finalsteps of the three immobilization protocols (D, F, I), recordedwith oligonucleotide-derived negative ions SUM DNA−, areshown in Fig. 8. They show a good repeatability of size, shapeand DNA density for the spots in the arrays fabricated witheach one of the studied immobilization method.

Some features of oligonucleotide distribution across thespots are characteristic for each of the spot arrays shown inFig. 8: DNA distribution with rings around the spot periphery(samples D, F) and around an inner domain with lower DNAdensity (sample F), as well as central enrichments (sample I)characterize the final steps (D, F, I) of the three differentimmobilization approaches used to form oligonucleotidearrays. In addition, the spot-size of biotinylated oligonucleo-tides is considerably smaller, when contact-printed onto anepoxy-silanized surface with streptavidin immobilized directly(sample I, ∼270 µm) rather than bonded to a pre-adsorbed bio-tinylated-BSA (sample D, ∼550 µm, and sample F, ∼570 µm).

3.3.2 Molecular distribution within single spots. Toexamine the specific spot features, TOF-SIMS imaging was

applied to visualize the molecular distribution within singleoligonucleotide spots after the final steps (D, F, I) of eachimmobilization approach: maps of total negative ion intensi-ties, reflecting cumulative material distribution (Fig. ESI-5,†total (−)), were used to normalize (for each pixel) the intensi-ties of DNA-derived negative ions and to yield the chemicalmaps of oligonucleotide composition within a single spot(Fig. ESI-5,† PO3

− and SUM DNA−). Moreover, the maps oftotal positive intensities, pointing to material distribution(Fig. 9, total (+)), were applied to normalize the intensities ofpositive fragment ions characteristic for different moleculesand to provide, for each DNA spot, the complementary maps(Fig. 9) of streptavidin (signal C9H8N

+, characteristic for trypto-phan),36,37 BSA (signal C5H10N

+ specific for lysine),36,38,39

epoxy-terminated silane GOPS (signal CH3O+)23 and silicon

substrate (signal Si+). The signals characteristic for the two pro-teins were chosen based on the amino acid abundance in therespective protein molecules (9.25% for tryptophan and 3.96%for lysine in streptavidin; 0.30% for tryptophan and 11.35% forlysine in BSA).34 Maps of negative and positive ion fragments,similar to those provided in Fig. 9 and ESI-5† and reflectingalso the cumulative material and molecular distribution butacquired with a higher magnification from different spots, arepresented in ESI (Fig. ESI-6 and ESI-7†).

In addition, for both the negatively- and positively-chargedions and for each type of axisymmetric DNA spot (samples D,F, I), we provide radial-averaged secondary ion intensitiesplotted against distance from a spot center. These plots(Fig. 10) represent the complementary radial profiles ofTOF-SIMS signals characteristic for DNA, streptavidin, BSA andepoxy-terminated silane.

3.3.3 Spot features and patterns resulting from solid pinprinting. To analyze the above introduced TOF-SIMS mapsand radial profiles, we will first briefly recall the models thatcould explain DNA spot morphologies formed by solid pincontact printing.

Due to the minute amount of liquid deposited onto thesurface (far less than 1 μL), the binding of oligonucleotides tothe surface reactive sites (here employing the biotin–strepta-vidin system) is greatly influenced by the evaporation of the

Fig. 8 TOF-SIMS chemical imaging of DNA arrays (2.5 mm × 2.5 mm area, oligonucleotide-derived negative ions SUM DNA−) prepared on SiO2 sur-faces: repeatability of size, shape, intensity and specific features of oligonucleotide distribution on the spots of the same array is demonstrated.

Analyst Paper

This journal is © The Royal Society of Chemistry 2015 Analyst, 2015, 140, 1127–1139 | 1135

Publ

ishe

d on

09

Dec

embe

r 20

14. D

ownl

oade

d by

UN

IWE

RSY

TE

T J

AG

IEL

LO

NSK

I on

02/

02/2

015

16:1

6:04

. View Article Online

oligonucleotide solution droplet delivered onto the surface bythe pin.14 The solvent evaporation leads to a higher concen-tration of oligonucleotides in the droplet, and hence increasesthe kinetics of its binding to the reactive surface sites.16,18 Inaddition, the spotting pin can introduce reproducible irregula-rities in the initial oligonucleotide distribution inside thedeposited droplet thus affecting the homogeneity (e.g. axisym-metry) of the final spots.15,17 Nevertheless, the surface energythat determines the contact radius of the spotted droplet (forits constant volume) can influence the spot size as reflected inthe dried spots.15,16

Droplet evaporation follows two modes that can beexchanged in the course of the process:18 constant contactarea but decreasing contact angle at the pinned contact linecharacterizes the first evaporation mode; while constantcontact angle, de-pinned contact line and diminishing contactarea are indicative of the second mode.16,18 Surface depositionscenarios as well as final spot features and patterns are influ-

enced by the dominant droplet evaporation mode. Flatteningdroplets with a constant area result in homogeneous depo-sition, often accompanied by an outward flow to the pinnedcontact line, replenishing the evaporating liquid.40,41 Theyyield, thus, uniform spots with or without ring stains at theirperiphery.18,41 In turn, the deposition from droplets with aretracting contact line, and hence an increasing solute concen-tration, would commonly result in spots with solute maximumat the center.16,18 When the retracting contact line is arrestedtemporally, deposits from such pinning events form concentricring patterns in the dried spots, sometimes with a solute-depleted center.40,42 In order to describe the finally observedoligonucleotide spot morphology, their specific binding to thefunctionalized surface must be considered in addition todroplet evaporation modes.16 The relative impact of both evap-oration modes to spot features and patterns is further speci-fied by parameters such as the areal density of reactive surfacesites, the affinity to surface binding sites, the contact angle ofthe oligonucleotide solution drops and the interactionbetween oligonucleotide molecules.16

3.3.4 Size of oligonucleotide spots obtained with the threeimmobilization approaches. The spotting for each one of thethree different oligonucleotide immobilization approaches wasperformed under different conditions and hence the final spotformation also followed different scenarios. The spots of bio-tinylated oligonucleotides contact-printed onto the epoxy-sila-nized surface with streptavidin immobilized directly onto the

Fig. 9 Molecular distribution of the whole material (row ‘total (+)’) andcomplementary composition maps of streptavidin (row ‘Str’, corres-ponding to the signal of ion C9H8N

+ characteristic for tryptophan), BSA(row ‘BSA’, corresponding to the signal of ion C5H10N

+ specific forlysine), epoxy-terminated silane GOPS (row ‘epoxy’, corresponding tothe signal of ion CH3O

+) and silicon substrate (row ‘SiO2’, correspondingto the signal of ion Si+), obtained by TOF-SIMS imaging of the DNAspots created following each one of the three immobilization protocols(columns D, F and I, respectively, cf. Fig. 1). The TOF-SIMS intensitymaps of all positive ion fragments (row ‘total (+)’) were used to normal-ize for each pixel the intensities of positive fragment ions characteristicfor the different molecules (shown after normalization at the rest of therows).

Fig. 10 Specific features of oligonucleotide distribution across therepresentative spots presented as radial-averaged secondary ion inten-sity vs. distance from the spot center: (a) DNA-derived negative ions(SUM DNA−), (b) positive ions characteristic for: streptavidin (signalC9H8N

+ characteristic for tryptophan), BSA (signal C5H10N+ character-

istic for lysine) and epoxy-terminated silane GOPS (signal CH3O+).

Paper Analyst

1136 | Analyst, 2015, 140, 1127–1139 This journal is © The Royal Society of Chemistry 2015

Publ

ishe

d on

09

Dec

embe

r 20

14. D

ownl

oade

d by

UN

IWE

RSY

TE

T J

AG

IEL

LO

NSK

I on

02/

02/2

015

16:1

6:04

. View Article Online

surface following approach III are smaller (sample I, ∼270 µmin diameter) than those formed for streptavidin bonded to thepre-adsorbed biotinylated-BSA (sample D, ∼550 µm, andsample F, ∼570 µm) following the two other immobilizationmethods (cf. Fig. 8). The spotting solution was identical forapproaches II and III, pointing to the decisive impact ofsurface free energy on the spot’s final size. A water contactangle value Θ of ∼71°, characteristic for the GOPS-modifiedoxidized silicon,43 indicates a hardly hydrophilic surface (dueto the hydrophobic alkane core and the non-ionizable terminalepoxy group).43 Its wettability can be increased by the adsorp-tion of proteins (with polar groups),44 since dried BSA layersexhibit the Θ value of ∼20°–30°.44 Also, the presence of biotinon the surface (due to polar ureido- and tetrahydrothiophene-rings)45 should result in a decrease of the contact angle,45 andit has been reported to be further slightly reduced after thereaction with streptavidin.45,46 The chemical heterogeneity ofthe surface is much lower for immobilization method III ascompared with the other two protocols: this is indicated by thehigher surface fraction of the non-reacted epoxy groups (cf. thedata for sample H vs. samples C and E, in Fig. 5a) and thelower protein surface coverage (cf. the data in Fig. 5b). There-fore the effective (Cassie–Baxter) contact angle is expected tobe larger for approach III as compared to the other two, result-ing in both the smallest initial droplet contact area and thefinal spot size.15,16

Finally, the surface used for spotting following approach II(sample E) resembles that of approach I (sample C), with thedifference that the streptavidin is also bonded to the surfaceprior to spotting (Fig. 5b) that induces a higher roughness tothe respective surfaces (Fig. 3c). Hence, a smaller effective(Cassie–Baxter and Wenzel) contact angle and a somewhatlarger spot size are expected.15 Also, the viscosity of thespotted droplet16 is lower for the biotinylated oligonucleotidesolution rather than for the streptavidin–biotinylated oligo-nucleotide conjugate solution used for spotting in approach IIvs. approach I. This difference promotes a further increase inspot size of sample F that is actually slightly larger (∼570 µm)than that of sample D (∼550 µm).

3.3.5 Spot morphologies for the three oligonucleotideimmobilization approaches. The creation of oligonucleotidespots following approach III is dominated by the hydrophobi-city of the surface, which is indicated by a smaller dried spot’ssize as compared to the ones of the two other approaches(Fig. 8). The initially pinned contact line of the spherical-likedrop is reflected by the ring of a dried drop material on thespot periphery (sample I), clearly visible on the TOF-SIMSmaps of total intensity of all positive (Fig. 9) and all negative(Fig. ESI-5†) ions. TOF-SIMS is a surface-sensitive techniquewith the sampling depth limited to 1–2 nm. Therefore, thisstrong peripheral material enhancement is visible not only asa bright ring of all the ions emitted from the top surface butalso as a dark ring of ions emitted from the substrate includ-ing both the silicon and epoxy-silane layer (Fig. 9). Since thesurface density of reactive streptavidin sites is low (see Fig. 6and ESI-3,† as confirmed by Fig. 5b and 5a for sample H), only

a limited amount of spotted biotinylated oligonucleotide canbind during the first mode of droplet evaporation.16 Then,during the second mode of evaporation, the binding of bioti-nylated oligonucleotides takes place from a retracting(probably still spherical) droplet with continually increasingoligonucleotide concentration. This results in maximal oligo-nucleotide concentration at a spot center, as indeed shown bythe pixel-to-pixel normalized chemical maps of oligonucleo-tide composition (Fig. 8 and ESI-5,† PO3

− and SUM DNA−).Since the relative oligonucleotide composition is enhanced ata spot center (brighter spike), it is accompanied by a depletion(darker domain) of another spot component at the same area,e.g. BSA, as shown by the respective normalized maps (cf.maps in Fig. ESI-5 and 9, Fig. ESI-6 and ESI-7,† cf. profiles inFig. 10). This result is not an effect of the surface-limitedsampling depth as the regions at the center and outside thespots have the same intensity of the substrate-derived ions inthe respective TOF-SIMS maps (silicon, epoxy-silane, cf. Fig. 9,10 and ESI-7†).

The contact-printing of oligonucleotide arrays followingimmobilization approach II is performed on a surface (sampleE) with the wettability enhanced by chemical and topographicheterogeneity, thus yielding the largest dried spots obtained bythe three studied approaches (Fig. 8). In this case, the initiallypinned contact line of a drop with a flat spherical cap-likeshape results also in the deposit of the dried drop material atthe periphery of the spot. This is visualized in sample F as anouter bright ring of total TOF-SIMS intensity of all positive(Fig. 9) and all negative (Fig. ESI-5†) ions, echoed (due tosurface-limited TOF-SIMS sampling) by an outer dark ring atthe images corresponding to the substrate-derived ions(silicon and expoxy-silane, Fig. 9). These ring-like features arenot related with any damages of the organic overlayer (e.g. dueto printing process),1 the latter manifested as domains with anenhanced rather than diminished intensity of silicon ions (cf.TOF-SIMS maps in Fig. 9 and ESI-7† for sample F). The overallspot morphology is affected considerably by the first dropletevaporation mode (constant contact area, decreasing contactangle). This is a consequence of the high surface wettabilityand relatively high surface density of the reactive sites (strepta-vidin, see Fig. 6 and ESI-3†).16 Thus, flattening droplets yield auniform and effective binding of oligonucleotides, slightlyenhanced at their periphery18,41 (see the data acquired withSUM DNA− and PO3

− for sample F in Fig. 8, ESI-5 and ESI-6,†the profile in Fig. 10). Finally, the contact line of flattened dro-plets is retracted within the second mode, with the depositedoligonucleotide to bind more strongly onto the dried out spotsurface rather than onto the area covered by the contractingdrop.40,42 At the end, the contact line is arrested at the spotcenter resulting in an enhanced ring-like material deposit.This is visualized as a brighter inner ring in the TOF-SIMSintensity images of all the negative ions (sample F, Fig. ESI-5and ESI-6†), matched by a dark inner ring of the epoxy-derivedions (maps in Fig. 9 and ESI-7,† the profile in Fig. 10). Theabove presented scenario explains the intensity of the oligo-nucleotide-derived ions: lower but non-zero inside the inner

Analyst Paper

This journal is © The Royal Society of Chemistry 2015 Analyst, 2015, 140, 1127–1139 | 1137

Publ

ishe

d on

09

Dec

embe

r 20

14. D

ownl

oade

d by

UN

IWE

RSY

TE

T J

AG

IEL

LO

NSK

I on

02/

02/2

015

16:1

6:04

. View Article Online

domain, higher at the remaining spot area and enhancedaround the inner domain and at the spot periphery (Fig. 8 and10, Fig. ESI-6 and ESI-7†). As it was discussed above, theenhancement in the relative composition of the one spot com-ponent, e.g. oligonucleotide, in the region inside the spot butoutside the inner domain is accompanied by the depletion inanother constituent, e.g. streptavidin (see the profiles SUMDNA− and STR in Fig. 10). The non-zero intensity of the oligo-nucleotide-derived ions inside the inner domain is an argu-ment against the scenario,17,47 where opening of a hole ratherthan droplet evaporation leads to the doughnut-like spot mor-phology48 of Fig. 8.

When immobilization approach I is applied, the spotting ofthe oligonucleotide solution is performed onto a surface(sample C) with wettability only slightly lower than that of thesurface used following approach II (sample E). Also, in thiscase the surface density of the biotin reactive sites of sample Cis the same as the one of sample E (see the scheme in Fig. 1).Therefore, the spot morphology is dominated by the firstdrying mode of the oligonucleotide solution droplet,16 similarto the scenario of approach II. As the oligonucleotide solutionnow contains the high molecular weight conjugate of biotiny-lated-oligonucleotide with streptavidin, the dried spots have avery high contrast on the TOF-SIMS maps taken with all posi-tive (Fig. 9) or all negative (Fig. ESI-5†) ions and no traces ofthe original contact lines are visible. The enhanced at spot per-iphery uniform and effective binding of oligonucleotides, thatcharacterizes the first droplet evaporation mode, is nowdemonstrated not only by the high intensity of the negativeoligonucleotide-derived ions (sample D, Fig. 8, Fig. ESI-5 andESI-6,† the profile in Fig. 10) but also by the positive ions ofstreptavidin fragments (Fig. 9 and ESI-7,† the profile inFig. 10). Both signal types show a high normalized intensity ofoligonucleotide and streptavidin (bright spots) and their per-ipheral enhancement (bright rings), accompanied by a lowintensity of the substrate-derived ions (e.g. silicon and epoxy-silane fragments, see dark spots) as well as the peripheraldepletion of other overlayer components (e.g. BSA, see the darkring). Although the impact of the second mode of droplet evap-oration on the spot morphology is much weaker, its character-istic patterns of singular concentric rings, indicating atemporally arrested contact line of the retracting droplet, arestill visible in the TOF-SIMS maps of both oligonucleotidesand streptavidin-derived ions (Fig. 9, ESI-5 and ESI-6†).

4. Conclusions

The nanostructure, chemistry and coverage of surfaces,obtained within multi-step protocols to form DNA arrays, arefor the first time characterized after each deposition/reactionstep (cf. ref. 1–3, 5, 6, 49–51).

The protocols, compared with the imaging and spectroscopictechniques, employ a streptavidin-formed bridge between thesurface and biotin-labelled oligonucleotides.4,10–13,52,53 Thehighest immobilization efficiency is achieved following a novel

extension of this strategy, when streptavidin conjugates withbiotinylated oligonucleotides are spotted on the surfaces coatedwith biotinylated BSA. This protocol is likely to prove useful inthe future development of microarrays for gene mutation detec-tion with unique sensitivity and specificity.

Multi-molecular distributions of DNA spots (maps of SiO2,epoxy-silane, oligonucleotide, BSA, streptavidin) reveal repeata-ble features (size, morphology), related with the earlier-determined substrate properties (protein coverage, roughness)characteristic for each immobilization protocol.

Minor inferences: protein immobilization is improved whenspotting is used rather than uniform surface coverage with aprotein solution (C–D vs. C–E in Fig. 5 and 6). An analysis ofAFM height histograms distinguishes between non-specificbinding, e.g. blocking with BSA, and specific protein–proteinbinding, due to biotin–streptavidin interactions (Fig. 3b–d).

Acknowledgements

This work was supported by the EU funded project “FOODS-NIFFER” (FP7-ICT8-318319). The equipment used was pur-chased thanks to the financial support of the EuropeanRegional Development Fund (POIG.02.01.00-12-023/08). One ofus (K.G.) is grateful for financial support from theM. Smoluchowski Kraków Research Consortium (in the frame-work of the KNOW scholarship).

Notes and references

1 C. Y. Lee, G. M. Harbers, D. W. Grainger, L. J. Gamble andD. G. Castner, J. Am. Chem. Soc., 2007, 129, 9429.

2 L. E. Cheran, D. Vukovich and M. Thompson, Analyst, 2003,128, 126.

3 N. Graf, T. Gross, T. Wirth, W. Weigel and W. Unger, Anal.Bioanal. Chem., 2009, 393, 1907.

4 A. Malainou, P. Petrou, S. Kakabakos, E. Gogolides andA. Tserepi, Biosens. Bioelectron., 2012, 34, 273.

5 A. N. Rao, N. Vandencasteele, L. J. Gamble andD. W. Grainger, Anal. Chem., 2012, 84, 10628.

6 P. Gong, G. M. Harbers and D. W. Grainger, Anal. Chem.,2006, 78, 2342.

7 C. J. May, H. E. Canavan and D. G. Castner, Anal. Chem.,2004, 76, 1114.

8 X. Fang, B. Li, E. Petersen, Y.-S. Seo, V. A. Samuilov,Y. Chen, J. C. Sokolov, C.-Y. Shew and M. H. Rafailovich,Langmuir, 2006, 22, 6308.

9 K. Awsiuk, J. Rysz, P. Petrou, A. Budkowski, A. Bernasik,S. Kakabakos, M. M. Marzec and I. Raptis, Appl. Surf. Sci.,2014, 290, 199.

10 M. Mir, P. J. Cameron, X. Zhong, O. Azzaroni, M. Álvarezand W. Knoll, Talanta, 2009, 78, 1102.

11 K. Tsougeni, G. Koukouvinos, P. Petrou, A. Tserepi,S. Kakabakos and E. Gogolides, Anal. Bioanal. Chem., 2012,403, 2757.

Paper Analyst

1138 | Analyst, 2015, 140, 1127–1139 This journal is © The Royal Society of Chemistry 2015

Publ

ishe

d on

09

Dec

embe

r 20

14. D

ownl

oade

d by

UN

IWE

RSY

TE

T J

AG

IEL

LO

NSK

I on

02/

02/2

015

16:1

6:04

. View Article Online

12 J. Escorihuela, M. J. Banuls, R. Puchades and A. Maquieira,Chem. Commun., 2012, 48, 2116.

13 X. Su, Y. J. Wu, R. Robelek and W. Knoll, Langmuir, 2005,21, 348.

14 M. C. Pirrung, Angew. Chem., Int. Ed., 2002, 41, 1276.15 G. McHale, Analyst, 2007, 132, 192.16 H. Y. Kim, S. E. Lee, M. J. Kim, J. I. Han, B. K. Kim,

Y. S. Lee, Y. S. Lee and J. H. Kim, BMC Bioinf., 2007, 8, 485.17 R. Blossey and A. Bosio, Langmuir, 2002, 18, 2952.18 V. Dugas, J. Broutin and E. Souteyrand, Langmuir, 2005, 21,

9130.19 S. E. Kakabakos, T. K. Christopoulos and E. P. Diamandis,

Clin. Chem., 1992, 38, 338.20 P. S. Petrou, M. Chatzichristidi, A. M. Douvas, P. Argitis,

K. Misiakos and S. E. Kakabakos, Biosens. Bioelectron.,2007, 22, 1994.

21 I. Horcas, R. Fernandez, J. M. Gomez-Rodriguez,J. Colchero, J. Gomez-Herrero and A. M. Baro, Rev. Sci.Instrum., 2007, 78, 013705.

22 I. Luzinov, D. Julthongpiput, A. Liebmann-Vinson,T. Cregger, M. D. Fosterand and V. V. Tsukruk, Langmuir,2000, 16, 504.

23 K. Awsiuk, A. Budkowski, A. Psarouli, P. Petrou,A. Bernasik, S. Kakabakos, J. Rysz and I. Raptis, ColloidsSurf., B, 2013, 110, 217.

24 F. Libang, G. Guangxin, C. Min and W. Limin, Macromol.Mater. Eng., 2007, 292, 754.

25 Z. J. Lv, J. H. Wang, G. P. Chen and L. H. Deng, Int. J. Biol.Macromol., 2010, 47, 661.

26 O. Ouerghi, A. Touhami, A. Othmane, H. Ben Ouada,C. Martelet, C. Fretigny and N. Jaffrezic-Renault, Biomol.Eng., 2002, 19, 183.

27 J. A. Howarter and J. P. Youngblood, Langmuir, 2006, 22,11142.

28 E. Strein and D. Alfred, Thin Solid Films, 2008, 517, 1011.29 K. Awsiuk, A. Budkowski, P. Petrou, A. Bernasik,

M. M. Marzec, S. Kakabakos, J. Rysz and I. Raptis, ColloidsSurf., B, 2013, 103, 253.

30 J. Kim, P. Seidler, L. S. Wan and C. Fill, J. Colloid InterfaceSci., 2009, 329, 114.

31 J. H. Scofield, J. Electron Spectrosc. Relat. Phenom., 1976, 8,129.

32 D. Briggs, Surface Analysis of Polymers by XPS and StaticSIMS, Cambridg University Press, Cambridge, 1998.

33 R. L. Meyer, A. Arpanaei, S. Pillai, N. Bernbom,J. J. Enghild, Y. Y. Ng, L. Gram, F. Besenbacher andP. Kingshott, Colloids Surf., B, 2012, 102, 504.

34 M. Dubey, K. Emoto, H. Takahashi, D. G. Castner andD. W. Grainger, Adv. Funct. Mater., 2009, 19, 3046.

35 S. Ferrari and B. D. Ratner, Surf. Interface Anal., 2000, 29,837.

36 M. S. Wagner and D. G. Castner, Langmuir, 2001, 17,4649.

37 Y. P. Kim, M. Y. Hong, H. K. Shon, D. W. Moon, H. K. Kimand T. G. Lee, Appl. Surf. Sci., 2006, 252, 6801.

38 O. Iguerb and P. Bertrand, Surf. Interface Anal., 2008, 40,386.

39 M. Wilhelmi, C. Müller, C. Ziegler and M. Kopnarski, Anal.Bioanal. Chem., 2011, 400, 697.

40 R. D. Deegan, Phys. Rev. E: Stat. Phys., Plasmas, Fluids,Relat. Interdiscip. Top., 2000, 61, 475.

41 R. D. Deegan, O. Bakajin, T. F. Dupont, G. Huber,S. R. Nagel and T. A. Witten, Phys. Rev. E: Stat. Phys.,Plasmas, Fluids, Relat. Interdiscip. Top., 2000, 62, 754.

42 L. Shmuylovich, A. Q. Shen and H. A. Stone, Langmuir,2002, 18, 3441.

43 J. N. Goddard and D. Erickson, Anal. Bioanal. Chem., 2009,394, 469.

44 T. Białopiotrowicz and B. Jańczuk, J. Surfactants Deterg.,2001, 4, 287.

45 E. H. Williams, A. V. Davydov, A. Motayed,S. G. Sundaresan, P. Bocchini, L. J. Richter, G. Stan,K. Steffens, R. Zangmeister, J. A. Schreifels and M. V. Rao,Appl. Surf. Sci., 2012, 258, 6056.

46 F. Rios and S. Smirnov, ACS Appl. Mater. Interfaces, 2009, 1,768.

47 P. C. Ohara and W. M. Gelbart, Langmuir, 1998, 14, 3418.48 M. Beier and J. D. Heheisel, Nucleic Acids Res., 1999, 27,

1970.49 H. F. Arlinghaus, Appl. Surf. Sci., 2008, 255, 1058.50 S. Hellweg, A. Jacob, J. D. Hoheisel, T. Grehl and

A. F. Arlighaus, Appl. Surf. Sci., 2006, 252, 6742.51 H. Hashimoto, K. Nakamura, H. Takase, T. Okamoto and

N. Yamamoto, Appl. Surf. Sci., 2004, 231–232, 385.52 A. Bonanni, M. I. Pividori and M. del Valle, Anal. Bioanal.

Chem., 2007, 389, 851.53 L.-Q. Chu, W. Knoll and R. Foerch, Biosens. Bioelectron.,

2009, 25, 519.

Analyst Paper

This journal is © The Royal Society of Chemistry 2015 Analyst, 2015, 140, 1127–1139 | 1139

Publ

ishe

d on

09

Dec

embe

r 20

14. D

ownl

oade

d by

UN

IWE

RSY

TE

T J

AG

IEL

LO

NSK

I on

02/

02/2

015

16:1

6:04

. View Article Online