DNA probes on chip surfaces studied by scanning force microscopy using specific binding of colloidal gold

Abstract

Single-stranded DNA was covalently bound on chip surfaces using two different silanization procedures. The resulting surfaces were characterized by fluorescence and scanning force microscopy using sequence-complementary DNA molecules with labels. Colloidal gold (30 nm) was used as the topographic label. Scanning force microscopy revealed the individual labels on the surface and their distribution. Steps of silane layers or DNA-modified surfaces prepared using an elastomeric mask provided internal controls for comparison of modified with unmodified surfaces.

INTRODUCTION

DNA analytic is an emerging field with a broad range of applications. Microarray and DNA-chip technology provide the means for a highly paralleled detection of DNA molecules, enabling applications like expression studies (covering some thousand genes in parallel) or sequencing by hybridization (1,2).

Crucial points for DNA-chips are the density and accessibility of the surface-immobilized capture DNA. These parameters were studied by autoradiography and fluorescence experiments (3), but the results lack the high lateral resolution in the lower micrometer and nanometer range needed for further miniaturization of the chips. A microscopical method with a higher lateral resolution is scanning force microscopy [SFM; also known as atomic force microscopy (AFM)] (4). This technique allows in principle the direct and label-free detection of individual DNA-hybridization events on surfaces, but its resolution is typically lower than the surface roughness. Therefore, only integral studies (aiming at micrometer-range areas of hybridized molecules) were reported using topographic (5,6) or interaction (7) SFM contrast. However, the understanding of the hybridization process on solid substrates, which is crucial for optimization of DNA-chip technology, requires the study of the behavior of individual or small ensembles of molecules. An SFM method addressing this problem based on nanoparticle labeling of DNA molecules was demonstrated on gold surfaces (A.Csaki, R.Möller, W.Straube, J.M.Köhler and W.Fritzsche, submitted). This technique is in principle able to detect interactions with a resolution limited by the contact area of the labels, which is in the lower square nanometer range. In this paper, the technique is used to compare different surface modification methods for immobilization of DNA-oligonucleotides on wafer surfaces, using gold nanoparticles with a diameter of 30 nm.

MATERIALS AND METHODS

Chip activation

Thermally oxidized wafers were cut in chips with dimensions of ~10 × 10 mm. The chips were first dry etched by oxygen plasma. Just before silanization they were activated by sonication for 10 min each in nitric acid, hydrogen peroxide and di-ionized water. Afterwards they were dried for 5 min at 80°C and used immediately.

Surface modification

Mercapto-silanized chips were prepared by incubation in a 1% 3′-mercaptopropyltrimethoxysilane in freshly distilled toluene for 6–8 h at 80°C. They were then washed thoroughly with toluene, methanol and di-ionized water and used immediately. In another set of experiments, the toluene used for incubation and washing was dried with sodium and distilled from benzophenone radical before incubation.

For substrate modification with 3′-aminopropyltrimethoxysilane (APTES), a modified protocol from the literature was used (8). The chips were immersed for 15 min in 1% APTES solution in 95% acetone/water. Afterwards, the chips were washed five times (5 min each) with acetone and dried for 45 min at 110°C. They were then incubated for 2 h with a solution of 0.2% 1,4-phenylenediisothiocyanate in 10% pyridine/dimethyl formamide and washed with methanol and acetone. The activated chips may be stored in a vacuum dessicator containing anhydrous calcium chloride for a longer time without discernible loss of activity.

For substrate modification with 3′-glycidoxypropyltrimethoxysilane (GOPS) the slides were suspended in dry toluene containing 1% silane at 80°C for 4–6 h, using a modified procedure from the literature (9). Then they were washed thoroughly with ethylacetate and used immediately.

Immobilization of DNA-oligonucleotides

5′-Amino-modified DNA-oligonucleotides (BioTeZ, Berlin, Germany) with a length of 12 bases were used as capture probes C1 and C2 (Fig. 1).

Figure 1.

DNA used in the study. Two oligonucleotides C1 and C2 with different sequences were used for immobilization on the chip surface as capture DNA. They were covalently bound to silanized surfaces using amino-chemistry. The presence of these molecules on the surface was probed by target DNA labeled with a fluorescent dye FITC (TF) or colloidal gold (TG).

For binding on APTES-surfaces, the amino-oligonucleotide was dissolved at a concentration of 2 mM in 100 mM sodium carbonate/bicarbonate buffer (pH 9.0). Droplets of the solution were applied directly to the activated chip in the desired pattern. The slides were incubated at 37°C in a covered Petri dish containing a small amount of water for 3–4 h, removed, washed once with 1% ammonium hydroxide, three times with water and dried at room temperature.

For binding on GOPS-surfaces, the amino-oligonucleotides were dissolved at a concentration of 50 µM in 0.1 M KOH. Droplets of the solution were applied to the epoxy-silanized surface of the chips and incubated at 37°C for 6 h in a covered Petri dish containing a small amount of water. After removal of the droplets, the chips were washed with di-ionized water at 50°C for 15 min with constant shaking, prior to air-drying.

Probe molecules

A DNA-oligonucleotide (BioTeZ) with a sequence complementary to C1 (cf. Fig. 1) and labeled with FITC at the 5′-end was used as fluorescently labeled target DNA TF (Fig. 1).

The activity of mercapto-silanized surfaces was probed by incubation with the stock solution of gold nanoparticles (30 nm mean diameter; Plano, Wetzlar, Germany) for 6 h at room temperature. After incubation, samples were extensively washed in water and air-dried.

For preparation of gold–DNA complexes, 3′-alkylthiolated oligonucleotides (BioTeZ) were cleaved from CPG support and preincubated with gold nanoparticle solution (30 nm mean diameter, Plano) for 16 h at room temperature in a ratio of 0.33 nM gold and 200 nM DNA (10). Another incubation was performed after adjusting the solution to 0.1 M NaCl/10 mM sodium phosphate buffer at pH 7.0 for 40 h at room temperature. The DNA–nanoparticle complexes were washed with buffer and redispersed in 0.3 M NaCl /10 mM sodium phosphate buffer at pH 7.0. Prior to hybridization on the chip, a linker oligonucleotide with a partly complementary sequence (12 bases, BioTeZ) was hybridized to the thiol-oligonucleotide, resulting in the final DNA–gold complex shown as TG in Figure 1. Therefore, gold-labeled DNA was incubated with an excess of linker for 1 h at 66°C and cooled to room temperature. The unbound linker was removed by centrifugation, and the final complex was redissolved in 5× SSPE and 0.5% SDS.

Probing immobilized DNA by labeled target DNA

Fluorescence labeling was by application of droplets of the fluorescently labeled DNA TF (1 µM in 5× SSPE, 0.5% SDS) to each chip, followed by an incubation for 3 h at 30°C in a closed Petri dish containing a small amount of water. After hybridization, the chips were washed twice with washing buffer (2× SSPE, 0.1% SDS) and covered with a microscope cover glass for fluorescence microscopy.

Hybridization of gold-labeled DNA with the DNA–gold complex TG was either by immersion of the chips in solution or by application of droplets on the chips. After incubation at 65°C for 10 min, the samples were cooled to room temperature over the course of several hours. After washing twice with washing buffer (see above) they were washed thoroughly with di-ionized water prior to air-drying.

Microscopy

A fluorescence microscope Axiotech (Carl Zeiss, Jena, Germany) equipped with a CCD camera Sensicam (PCO Computer optics, Kehlheim, Germany) was used.

SFM (also known AFM) was conducted with a Dimension 3100 (Digital Instruments, Santa Barbara, CA) in tapping mode in air.

RESULTS AND DISCUSSION

Comparison of two silanization techniques

A technique was needed to covalently bind oligonucleotides on oxidized wafer surfaces. A variety of approaches have been described in the literature (11–14). The technique should be reliable and based on simple chemicals and reactions. Two methods using silanization of surfaces for binding of amine-modified oligonucleotides were adapted from the literature, based on APTES (8) and GOPS (9).

The APTES silanization method is fast and uses aqueous solutions, but includes an additional activation step with phenylendiisothiocyanate (PDC). Disadvantages of the GOPS method are the duration of the silanization reaction, but also the needed exclusion of water.

Aiming at a smooth silane surface

A typical problem with the preparation of silane surfaces is an increased roughness. This effect can be attributed to polymerization reactions in the liquid phase. For standard autoradiographic or fluorescent experiments, high surface roughness has no negative effect or is even needed for an enhanced surface area. However, increased roughness is a serious hindrance for microstructured surfaces by introducing failures (gaps or bridges) into the final structure. A simple method for removal of the unwanted material is mechanical wiping over the surface using clean room tissue. To test this procedure, surfaces modified with mercaptosilane were used. The thiol groups of this compound serve as a model system for active chemical groups on the wafer substrate. By binding colloidal gold to such thiolated surfaces using the high affinity of thiol groups to gold (15), active surface regions should be specifically labeled. SFM was used for characterization of the roughness of the modified surface. Figure 2a shows a topographic image of an unlabeled mercaptosilane surface revealing large junks (bright areas) of material covering the surface, resulting in a high surface roughness in the range of several tens of nanometers (cross-section in Fig. 2a). Wiping cleaned such surfaces; SFM imaging reveals that the surface usually appears much smoother after this procedure (Fig. 2b), confirmed by the decrease in surface roughness <5 nm (cross-section). The distribution of active thiol groups on the surface was probed by labeling with gold nanoparticles. The resulting SFM image (Fig. 2b) shows the traces of the wipes visualized by the distribution of the gold particles. Individual gold particles are clearly resolved (inset). The surface roughness is improved, but the particle distribution points to non-homogeneity in the distribution of active sites.

Figure 2.

Surface roughness and distribution of active groups. Scanning force micrographs. Scale bars in insets are 1 µm (lateral) and 10 nm (height). (a) A conventional mercaptosilane surface exhibits a surface roughness with peaks reaching 50 nm or more (see left inset for cross-section). (b) After wiping, the surface roughness decreases significantly, as shown in the cross-section (inset bottom-left). A solution of colloidal gold was used to visualize the distribution of active SH-groups, resulting in a striped pattern. The individual gold particles can be clearly resolved (inset top-right). (c) A comparable low surface roughness, but this time connected with a homogeneous distribution of the gold labels, was obtained using dried toluene for the silanization procedure.

Although the wiping procedure significantly decreases the surface roughness, the observed non-homogeneous surface coverage by active groups makes this procedure not applicable for a defined coverage of microstructured surfaces. Another possibility for smoother surfaces is the optimization of the silanization procedure. As mentioned above, the observed material on the silanized surface (Fig. 2a) is probably due to polymerization occurring in the liquid phase during silanization. Suppressing this unwanted process by minimizing the proportion of water in the reaction resulted in an improved surface roughness (cross-section in Fig. 2c; cf. Fig. 2a), which is suitable for microstructures without further processing (e.g. wiping). The distribution of the active groups was tested by gold labeling and was found to be homogeneous (Fig. 2c).

APTES-based surface modification

In one series of reactions, the surfaces were modified with APTES, activated with PDC and incubated with amino-modified DNA-oligonucleotides. At first, the APTES layer was characterized after creation of a step between APTES-modified surface regions (left in SFM micrograph in Fig. 3a) and the silicon oxide surface (right in Fig. 3a). This step was induced by covering parts of an APTES-surface by an elastomeric mask as described elsewhere (16). After treatment with oxygen plasma and removal of the mask, the surface was characterized by SFM (Fig. 3a, right). A height difference of ~1.8 nm was found, which can be attributed to the thickness of the APTES layer.

Figure 3.

DNA immobilization using APTES-surfaces. (a) A step in an APTES-surface was fabricated using the shown procedure. Parts of the silanized surface were covered by an elastomeric mask and thereby protected against damage by oxygen plasma. The resulting step of ~1.8 nm is visible in the scanning force micrograph and the cross-section. The high structure along the step is probably due to edge effects during the plasma etching or is built up by remains of the mask material. (b) Capture DNA C1 was immobilized onto APTES-surfaces and probed using the fluorescently labeled target DNA TF. These areas yielded a high fluorescence signal, compared to a low signal on the gold structures. (c) The signal observed in (b) is due to a specific DNA–DNA interaction, because neither capture DNA C2 with a non-complementary sequence (c) nor APTES-surfaces without capture DNA (d) yielded any significant fluorescence. (e) Pure APTES-surfaces were also probed with gold-labeled target DNA TG. Now some surface-bound particles are visible in the SFM. (f) A significantly larger amount of gold particles binds in the case of surfaces with immobilized capture DNA C1 (complementary to TG).

APTES-modified surfaces were activated with PDC and incubated with the amino-DNA C1. The distribution of these oligonucleotides was visualized using complementary fluorescence-labeled DNA (TF), which bind specifically to the immobilized DNA probe C1. The corresponding fluorescence image (Fig. 3b) shows a strong signal on the silicon oxide surfaces, compared to a minimal fluorescence on the gold structures. This result points to a high specificity of the capture DNA immobilization. The binding of the labeled DNA to the immobilized capture DNA is sequence specific; control experiments demonstrated that the labeled DNA TF does not significantly bind to capture DNA C2 with a non-complementary sequence (Fig. 3c) or to an unmodified silicon oxide (Fig. 3d) surface. However, a more quantitative study of the unspecific binding based on fluorescence detection is difficult, especially due to reflection problems caused by the gold structures on the sample surface.

SFM was used for a more detailed study of this problem. Unspecific binding of DNA-oligonucleotides was investigated by incubation of PDC-activated APTES-surfaces with gold-labeled oligonucleotides. The SFM reveals a surface coverage of about four labels per square micrometer (Fig. 3e). Similar surfaces were used for binding of amino-modified oligonucleotides, prior to probing by incubation with gold-labeled oligonucleotides with a sequence complementary to the surface-bound DNA. The SFM-image of such specific binding (Fig. 3f) yields a surface coverage of about 25 labels per square micrometer, which is significantly higher than the value for unspecific binding.

Immobilization of DNA using GOPS

In another set of experiments, silicon oxide surfaces containing gold structures were modified with GOPS prior to immobilization of amino-modified oligonucleotide C1 as capture probe (first scheme in Fig. 4a). To investigate the binding of the fluorescence-target TF onto different surfaces (like complementary capture probe C1, silicon oxide background and gold), parts of the surface were covered by an elastomeric mask prior to oxygen plasma etching (second scheme). This etching removes the silane as well as the immobilized oligonucleotides at the unprotected surface regions, resulting in differentiated regions on one chip (third scheme). Incubation with the target oligonucleotide TF should result in binding of this fluorescence marker only in the silanized regions protected by the mask against dry etching (last scheme). This expected pattern could be proved in the experiment, a fluorescence micrograph (Fig. 4b) shows low fluorescence signal in the silicon oxide region (left), compared to a high signal at the GOPS regions. This surface area is covered by the capture DNA C1 with a sequence complementary to the fluorescence target TF, so the signal points to a specific binding of the target DNA. The unspecific binding of target DNA onto silicon oxide is low (dark regions in Fig. 4b). The specificity of binding onto GOPS-surfaces was studied using gold-labeled target TG. GOPS-surfaces with and without capture DNA were incubated with TG, and the extent of binding was characterized using SFM. Pure GOPS-surfaces revealed no bound gold particle, excluding unspecific gold or DNA binding onto these kinds of surfaces (Fig. 4c). A high degree of gold binding was observed on GOPS-modified surfaces with immobilized capture DNA, which is complementary to the DNA on the gold label (Fig. 4d).

Figure 4.

DNA immobilization by GOPS-modification. (a) Scheme for obtaining structured layers of capture DNA C1 on the chip surface using an elastomeric mask, prior to probing with fluorescently labeled target-DNA TF. (b) The result of the procedure described in (a) is shown in fluorescence contrast. The last step of the scheme describes the structure along the dashed line in the micrograph. (c) Control experiment, demonstrating the low unspecific binding of gold-labeled DNA TG to GOPS-surfaces without capture DNA. (d) The incubation of surfaces of complementary capture DNA C1 with gold-labeled TG results in a surface coverage with gold colloids due to specific DNA–DNA interactions.

CONCLUSIONS

Chip surfaces modified by silanization provide a reliable substrate for immobilization of DNA-oligonucleotides with a large potential for applications in biotechnology and nanotechniques. The silanization yields molecular layers with smooth surfaces, resulting in a roughness in the lower nanometer range. Two different techniques were used to bind amino-modified oligonucleotides onto silanized chips. The activity and distribution of the immobilized DNA molecules were probed using complementary DNA with fluorescence or topographic labels. Fluorescence microscopy yields fast and easy overviews of the chip areas with DNA. The lateral distribution in the sub-micrometer range could be revealed using colloidal gold as the topographic marker for SFM. The demonstrated sequence-specificity of the DNA binding and the low unspecific binding to pure or modified chip surfaces shows the potential of these techniques for the detection of specific molecular binding events on biochips.