Fabrication of 2D-protein arrays using biotinylated thiols: results from fluorescence microscopy and atomic force microscopy

The coupling of a fluorescently labelled protein, streptavidin, to biotinylated organic surfaces is investigated using fluorescence microscopy (FM) and atomic force microscopy (AFM).

In order to study the importance of non-specific adsorption investigations were carried out using two-dimensionally structured self-assembled monolayers (SAMs), which were fabricated using micro-contact printing.

The relative amount of specific vs. non-specific adsorption could be readily determined by comparing the amount of streptavidin adsorbed on adjacent regions consisting of biotinylated organothiolates and of protein-resistant oligoethyleneglycol (OEG)-thiols.


An important strategy in molecular architecture used for non-covalent coupling of molecular units is based on the strong binding of biotin to the protein avidin.1

Biotin is a fairly small molecule which can be easily attached to surfaces and to other molecules.

Streptavidin will bind to surface-anchored biotin with high selectivity and a large binding energy.

Since it has four binding pockets for biotin a subsequent attachment of further biotinylated moieties is possible thus providing an elegant way of coupling e.g. biotinylated proteins to appropriately functionalized surfaces.

A major problem, however, with regard to this type of molecular architecture is the non-specific binding of streptavidin, i.e. the adhesion to organic surfaces which are not biotinylated.

Since the non-specific adsorption of proteins to surfaces is a topic which currently attracts a large amount of interest,2 we have carried out a systematic study of specific vs. non-specific binding in case of streptavidin by using microscopic techniques (atomic force microscopy, or AFM; and fluorescence microscopy).

Streptavidin and avidin belong to the same super family of biotin-binding proteins but they differ in origin: Streptavidin is extracted from a bacterium called Streptomyces avidinii while avidin is found in egg-white and oviduct of birds, amphibians and reptiles.

Each of both proteins consists of four identical subunits and each subunit is capable of binding one D-biotin in its binding pocket.3

Upon binding the D-biotin is deeply buried in the binding pocket.

Thus the avidin-biotin system is frequently considered a general model for key-lock-interactions to study molecular recognition processes.

Streptavidin and avidin are famous for their high affinities towards biotin.

The tight binding of the biotin results in a very high affinity constant of KA ∼ 1015 M−1(1) which implies that the binding of biotin towards streptavidin or avidin is nearly irreversible under physiological conditions.

For this reason the avidin–biotin system is occasionally referred to as a biological glue.

Many strategies for fabricating biosensor arrays take advantage of these properties of the avidins.4

Frequently fluorescently labelled avidins have been used as molecular markers that are bound to biotinylated targets.

Other approaches make use of the tetramer structure of avidin: avidins are used as building blocks for a sandwich layer which is bound on one side to a biotinylated support and on the other side to biotinylated target molecules.5

Avidin is not only used as biotin-binding interface but also as a signal amplifier.6

If an avidin is attached towards a biotinylated sample there are still up to three free remaining binding pockets to which dyes, fluorescence labels or enzymes can be linked to.

Since the introduction of micro-contact printing (μCP) by Whitesides a fast and cheap method for creation of laterally structured metal surfaces is available.7–9

Micro-contact printing (μCP) using PDMS-stamps has been used to create laterally structured devices for array biosensors which should finally be used for parallel detection of different biological molecules.8,10–12

Recently it has been proposed to employ light sensitive molecules for the lateral structuring of SAMs.13,14

One group reported about a caged biotin which becomes “uncaged” if exposed to a laser beam of a certain wavelength.14

2D-protein patterns have been studied not only with fluorescence microscopy but also with microscopic techniques as atomic force microscopy and SPR-microscopy.15,16

Our own work focuses on laterally structured biotinylated SAMs and the evaluation of binding properties of streptavidin using AFM and fluorescence microscopy.

Experimental section

Substrate preparation

Polished silicon wafers with a 100 surface termination were purchased from Prolog Semicor LTD, Ukraine.

The wafers were cut into 4 mm × 8 mm pieces and stored in pure acetone.

Before usage they were rinsed with pure acetone and absolute ethanol (EtOH).

After drying in a stream of nitrogen the substrates were installed in a Leybold Inficon XTC/2 metal evaporator.

First a layer of 80 Å titanium @ 2 Å s−1 was deposited to improve the adhesion of the subsequently deposited gold layer.

Gold has been evaporated @ 15 Å s−1 to a final thickness of approximately 1250 Å.

Evaporation has been done at room temperature and a pressure of approximately 10−7 mbar.

The evaporated silicon wafers were stamped via micro contact-printing (μCP) which is well established in literature.8

As a master we use a block of glass with a periodic pattern of deepened square areas (40 μm × 40 μm) that are surrounded by 2.5 μm raised and 10 μm wide bridges.

An inverse elastic replicate (stamp) is created with polydimethylsiloxane (PDMS).

The stamp is loaded by incubation with a drop (30 μl applied from a pipette) of an 1.37 mM ethanolic solution of an OEG6-thiol (for chemical structure see Fig. 1) for 90 s then the stamp is dried in a stream of nitrogen and pressed gently towards the gold surface for 90 s.

After printing the stamp is removed and the samples were immersed into an ethanolic solution containing a mixture of two thiols: 1-mercaptoundecanol (OH-thiol) and a biotinylated thiol (Biotin-thiol) (for chemical structures see Fig. 1).

The total thiol concentration in the mixtures (3:1 unless otherwise noted) was kept constant at ∼50 μM.

After immersion the laterally structured self-assembled monolayers (SAMs) were rinsed with absolute ethanol.

Incubation conditions

All samples were incubated at room temperature with a streptavidin-Alexa-Fluor488® conjugate (Molecular Probes).

The attached fluorescence label (Alexa-Fluor488-group) exhibits spectroscopic properties which are similar to those of fluorescein (λex ∼ 488 nm and λem ∼ 520 nm) but with an enhanced stability towards bleaching.

In an initial set of experiments incubation times and concentrations were varied so as to minimize non-specific binding of streptavidin towards OEG6-covered regions.

Incubation times (pure contact times) of 5 s together with streptavidin concentration of 200 nM were found to yield good results.

At higher incubation concentrations and/or times a significantly larger amount of aggregated, randomly distributed streptavidin clusters was observed.

Immediately after streptavidin incubation the substrates were intensively rinsed with pure water obtained from a MilliQ-system.

The rinsed samples were stored in pure water until they were investigated with fluorescence or atomic force microscopy.

Streptavidin-Alexa-Fluor488® was used without further purification in a buffer containing the following components: 150 mM NaCl (Baker), 10 mM Hepes (Gerbu) and 5 mM MgCl2 (Baker).

The pH was adjusted to 7.4 with NaOH and HCl.

The incubation solutions were used multiple times and stored at +4 °C in between the experiments.

The concentrations of the incubation solutions have been checked in regular intervals using a fluorometer (Spex II).

For simplicity in the following Streptavidin-Alexa-Fluor488® is referred to as streptavidin.

The ethylene glycol groups in the OEG6 thiol are rather sensitive to oxidation.

Even when storing them at +4 °C a noticeable degradation (as evidenced by the onset of protein adsorption on SAMs made from the thiol) was observed already after a period of 4 weeks.

For this reason the pure, freshly synthesised OEG6 thiol was stored at −15 °C under a nitrogen atmosphere.

Confocal fluorescence microscopy (confocal-FM)

Confocal fluorescence microscopy was carried out using a confocal laser scanning microscope MRC 1024 (Biorad, Hemel Hempstead, UK).

The instrument is also equipped with an Ar-Ion laser (the line at 488 nm was used for the excitation of the fluorophor) and an inverted microscope (Nikon, Eclipse TE-300 DV, infinity corrected optics) with a 60× water immersion objective (NA 1.2, collar rim correction, Nikon).

In front of the photomultiplier serving as a detector a bandpass filter (520 nm ± 20 nm, half-bandwidth) is mounted for selective fluorescence detection.

During fluorescence imaging the samples are fixed in a home-made sample holder to be able to perform fluorescence microscopy in a “contact free” manner (see Fig. 2).

Special care was taken not to mechanically damage the substrate surface before and during fluorescence imaging since we used the confocal fluorescence microscopy for a pre-screening of substrates before they were investigated with the AFM.

Fluorescence images were recorded using a laser power of 1 mW, in the focal plane of the objective lens.

All images were obtained by accumulation of four single images (∼1 s per image).

Bleaching experiments were conducted according to the following procedure: first, the microscope is switched from the accumulate mode to the direct mode (permanent scanning) and the electronic zoom is turned to maximum (i.e. 10×).

Then the zoomed area is irradiated for 3 min at full laser power (1 mW).

After this bleaching period the microscope is switched to the accumulate mode and the zoom to its initial value.

Then imaging is performed as described before.

Atomic force microscopy (AFM)

The atomic force micrographs were recorded using a Nanoscope IIIa (Digital Instruments) AFM.

The microscope was operated in two different modes: in contact- and tapping- mode.

During contact-AFM silicon nitrite cantilevers with force constants from 0.06 N m−1 to 0.58 N m−1 and in tapping-AFM TESP cantilevers with force constants of 20 and 100 N m−1 were employed.

Topographic images were generated in both modes from the height information of the samples.

In contact mode additionally the lateral force microscopy (LFM) mode has been used to image local friction differences of the sample.

Substrates were investigated before and after protein incubation with contact-AFM and tapping-AFM.

Before investigation all samples were taken out of distilled water and dried in a stream of nitrogen.

All AFM measurements were taken in air.


Contact-AFM and tapping-AFM with 2D-SAM structures

In Fig. 3 we show typical AFM-micrographs obtained for the laterally structured SAMs using different imaging modes.

As expected, the topographic images (b and d) reveal a poor contrast only, but when using the friction-mode (a) the lateral pattern produced by the μCP-process is clearly visible.

The regions appearing as bright squares correspond to the part of the Au-substrate where the OEG6-thiol was stamped on the surface, the darker stripes correspond to the SAM formed during the subsequent immersion into the 3:1 mixture of the OH-terminated and the biotinylated thiol.

In the upper row on the left a LFM-mode image and on the right a phase-image is presented.

The bottom row shows topographic (height) images.

Confocal FM after incubation with fluorescently labelled streptavidin

Fig. 4 shows two micrographs obtained with the fluorescence microscope for patterned samples after incubation in the streptavidin solution.

The high contrast in both images reveals that despite the quenching due to the close proximity of the Au-layer the pattern produced by the μCP-process can be imaged with high contrast.

The edges of the patterns are sharp and well defined.

A low density of randomly distributed bright spots resulting from non-specifically bound clusters of aggregated protein is observed.

Samples investigated before incubation revealed a non-textured, weak background signal (data not shown).

This apparent residual fluorescence did not show any photo bleaching and results from the noise of the photo detector unit.

Intentionally photo-bleached areas of streptavidin incubated samples revealed the same residual signal and were used for background subtraction.

See Table 1 for detailed information on the contributions of different signal sources.

Under imaging conditions the protein substrates were exposed to a laser power of approximately 1 mW for about 4 s.

Significant bleaching was only observed after much longer exposition times (>20 s at 1 mW).

The results shown in Fig. 4 and Table 1 clearly demonstrate that the streptavidin adsorbs preferentially on the stripes corresponding to the regions containing the biotinylated thiol.

For the squares a much weaker fluorescence is observed.

In case of the 4 week old OEG6-thiols (see squares in Fig. 4a) the fluorescence of the squares can be photo-bleached.

This observation reveals that a significant amount of streptavidin has adsorbed to the 4 week old OEG6-thiols, indicating that the protein resistance of the aged OEG6-thiol SAM is significantly reduced.

The total amount of streptavidin adsorbed on the OEG-terminated SAM surface can be determined quantitatively by correcting for the background (noise of the photomultiplier).

As shown in Table 1 the total amount of adsorbed streptavidin corresponds to 13% of the total streptavidin adsorption on biotinylated surfaces.

In contrast no bleaching is observed if the squares are prepared from fresh OEG6-solutions.

In Fig. 4b we have included a part of a stripe into the photo bleaching to indicate where the bleaching was performed.

These results promote the view that squares prepared from fresh OEG6-solutions almost completely resist streptavidin adsorption.

Contact-AFM/tapping-AFM after incubation with streptavidin

In Fig. 5 we show results obtained by AFM in the contact (Fig. 5a) and tapping (Fig. 5b) mode for the patterned SAMs incubated for 5 s with a 200 nM streptavidin solution at room temperature.

Contact mode images were taken at contact forces as low as possible in order to prevent mechanical deformation and/or disruption of the protein pattern.

In both AFM imaging modes the 2D-pattern fabricated by the micro-contact printing process can be observed.

A closer analysis, however, reveals a different height contrast between stripes and squares.

The contact mode image reveals a significantly larger height difference than those obtained in the tapping mode.

This is illustrated in Fig. 6 where line profiles (cross sections) corresponding to the white dashed lines in Fig. 5 are displayed.

The line profile (filled circles in Fig. 6) from the contact AFM-micrograph reveals a height difference of 22.4 ± 3 Å between the top of the stripes and the top of the squares.

This value has to be compared with the height of streptavidin of 45 Å as deduced from the streptavidin bulk structure5.


Contact-AFM and tapping-AFM for the patterned SAMs

When the 2D pattern generated by the micro-contact printing process is imaged prior to incubation the largest height difference seen in contact AFM mode between the squares (OEG6-SAM) and the stripes (3:1 OH- and biotinylated SAM) amounts to about 5 Å.

Before comparing this value to that expected from the difference in length between that of the biotinylated thiol (37 Å) and the OEG-thiol we have to consider that the biotinylated thiols are diluted in the OH-terminated thiols (Fig. 7, left side).

During scanning the AFM-tip is expected to push the biotinylated thiols aside, thus effectively imaging the top surface of the considerably shorter OH-terminated thiols.

From the difference in length between the OH-terminated (15.4 Å) and the OEG6-thiol (38 Å) we thus expect a height difference of 22.6 Å between the squares and the stripes.

The situation is further complicated, however, because when imaging the SAM-surfaces in air at a relative moisture of between 45 and 65% a thin layer of water molecules is present on the SAM surfaces, and especially the hydrophilic ethylene-glycol-rich part of the OEG6-thiols will attract water molecules and swell upon hydration.

The fact that the fairly small height differences actually vanish when the tapping mode is employed further indicates that the height difference seen in contact mode also results to a major extent from differences in compressibility.

A quantitative analysis of all of these effects that contribute to the effective height difference seen in the AFM is impossible.

When the lateral friction is used for imaging a better contrast is observed, the corresponding image (Fig. 3a) clearly shows the square and stripe pattern produced by the μCP-process.

The friction is larger on the squares, which is attributed to a slightly stronger interaction between the tip and the OEG6-thiols.

The friction contrast during imaging suggest that the AFM tip penetrates into the SAM (especially into the OEG6-SAM) during contact mode scanning (compared with discussion of SAM compression above).

Altogether the AFM results demonstrate that the pattern has successfully transferred from the PDMS stamp to the substrate.

Confocal FM with 2D-protein structures

After incubation with fluorescently labelled streptavidin the data from fluorescence microscopy clearly reveal bright stripes and dark squares, indicating that the streptavidin has in fact been adsorbed preferentially on the stripes containing the biotinylated thiols.

The total intensity observed on the stripes is consistent with the presence of one monolayer (see AFM measurements below) of intact streptavidin bound to the biotinylated surface of the biotinylated SAM.

Our results for patterned SAMs prepared from fresh thiol solutions are fully consistent with previous published surface plasmon resonance studies.5

Surprisingly, the same experiments carried out with a 4 week old OEG-solution reveal already a significant amount of non-specific streptavidin adsorption.

We attribute the loss of protein resistance to oxidative particularly degradation of the OEG6-thiol.

Contact-AFM and tapping-AFM for 2D-protein structures

At first sight the data obtained by AFM after incubation of the patterned SAMs appear not to be consistent with the results from fluorescence microscopy, where streptavidin was found to preferentially adsorb on the stripes.

Naively, one would assume that this process will increase the thickness of the stripes by 45 Å, the dimension of streptavidin in its bulk, cuboid-like structure 45 × 45 × 53 Å3.5

In the tapping mode AFM micrographs, therefore, the stripes, which showed no height difference to the squares before incubation, should now appear 45 Å higher.

In the AFM data, however, virtually no difference in height is seen between stripes and squares in tapping mode.

In the contact mode AFM data, where significantly larger forces act between tip and sample, a height-difference of 22.4 ± 3 Å between the top of the stripes and the top of the squares is seen, with the stripes appearing higher.

The stripes appearing higher is expected, since before incubation the stripes and squares had approximately the same height.

But, again, the height difference is almost a factor two smaller than that expected from the bulk dimension of the streptavidin (see above).

Considering the results obtained from AFM before and after incubation and as well the information gained from fluorescence microscopy, we conclude that not more than a full monolayer of streptavidin can be present on the stripes, resulting in the scenario depicted in Fig. 7.

After removal from the water the streptavidin is partially dehydrated and reduced in dimension, resulting in a difference in height between the stripes and the squares which is smaller than the resolution in AFM tapping mode.

Similar reductions in size have been frequently discussed in previous work on imaging proteins with AFM in air.17

When the incubated, patterned SAMs are imaged in AFM contact mode, the stripes appear about 22 Å above the squares.

This is not a true height difference but results form the larger compressibility of the OEG6-thiol compared to the already partially dehydrated and thus more compact streptavidin.

This scenario presented in Fig. 7 is then fully consistent with the results from fluorescence microscopy and the AFM results recorded prior to incubation.


We have investigated the interaction of fluorescently labelled streptavidin with SAMs exhibiting a pattern of biotinylated regions using fluorescence microscopy and AFM.

In agreement with previous results we find specific binding of the streptavidin to the biotinylated parts of the organic surface, yielding a 2D pattern of streptavidin monolayers on the surface.

The combination of AFM and fluorescence microscopy with μCP is demonstrated to be extremely useful with regard to a characterisation of the binding of streptavidin to biotinylated surfaces.

In contrast to previous SPR studies,5,18 specific and non-specific binding events can be detected, distinguished and quantified on the same sample exposing different types of organic surfaces at the same time.

In particular all different types of organic surfaces investigated share the same history and incubation conditions.

In the work presented here we were able to identity different origins of non-specific adsorption, namely non-specific adsorption on degraded OEG6-thiols and nucleation resulting in the formation of 3D-aggregates.

The results from fluorescence microscopy were essential with regard to the interpretation of the AFM data, in particular for a quantitative analysis of the height differences seen in the AFM micrographs.

The present approach demonstrates the advantages of using patterned organic surfaces in protein binding studies and suggests that obtaining stable protein-resistant surfaces by using oligoethylene terminated thiols is more complicated than anticipated previously.

In particular the possibility of degradation of the protein resistance has to be considered, a problem which only recently has started to attract attention19.