High-Resolution Mapping of Redox-Immunomarked Proteins Using

ARTICLE pubs.acs.org/ac

High-Resolution Mapping of Redox-Immunomarked Proteins Using Electrochemical Atomic Force Microscopy in Molecule Touching Mode Agn
es Anne, Arnaud Chovin, Christophe Demaille,* and Manon Lafouresse Laboratoire d’Electrochimie Moleculaire, UMR 7591 CNRS, Universite Paris Diderot, Sorbonne Paris Cite, 15 rue Jean-Antoine de Baïf, F-75205 Paris Cedex 13, France

bS Supporting Information ABSTRACT: We explore the possibility of using molecule touching atomic force electrochemical microcopy (Mt/AFM SECM) for high-resolution mapping of proteins on conducting surfaces. The proposed imaging strategy relies on making surface-immobilized proteins electrochemically “visible” via redox-immunomarking by specific antibodies conjugated to poly(ethylene glycol) (PEG) chains terminated by redox ferrocene (Fc) heads. The flexibility and length of the PEG chains are such that, upon approaching a combined AFM SECM microelectrode tip toward the surface, the Fc moieties can efficiently shuttle electrons from the surface to the tip. The so-generated SECM positive feedback tip current allows the specific localized detection of the sought protein molecules on the surface. This new electrochemical imaging scheme is validated experimentally on the basis of a model system consisting of mouse IgGs adsorbed onto electrode surfaces and recognized by Fc PEG-labeled antimouse antibodies. In order to estimate the resolution of Mt/AFM SECM for protein imaging, regular arrays of submicrometer-sized spots of mouse IgGs are fabricated onto gold electrode surfaces using particle lithography. The Fc PEG-immunomarked mouse IgG spots are imaged by Mt/AFM SECM operated in tapping mode. Both an electrochemical image, reflecting the surface distribution of the redox-labeled IgGs, and a topography image are then simultaneously and independently acquired, with a demonstrated resolution in the ∼100 nm range. The strength of Mt/AFM SECM imaging is to combine the nanometric resolution of AFM with the selectivity of the electrochemical detection, potentially allowing individual target proteins to be identified amidst similarly sized “nano objects” present on a conducting surface.

M

any advanced (bio)sensors use surface-attached proteins, such as redox enzymes, as transducing elements. However, the actual two-dimensional (2D) distribution of the transducing molecules on the sensor surface is generally unknown. Yet, the sensitivity of the sensor can be drastically different if the proteins are homogeneously or heterogeneously distributed on the surface.1 Being able to map the actual surface distribution of transducing proteins on a given sensor surface is therefore most useful since it allows the analytical performances of the device to be related with the true molecular arrangement of the sensing layer. To address this issue, local probe imaging techniques, such as atomic force microscopy (AFM) or scanning electrochemical microscopy (SECM),2 4 have notably been used to image biosensing surfaces. Each of these microscopy techniques displays specific advantages for characterizing bioanalytical platforms.1 AFM offers a very high, nanometric, resolution which, in principle, allows individual protein molecules to be located on sufficiently flat surfaces. However, because regular AFM is simply based on the physical interaction of an inert probe with the imaged surface, it is largely unspecific: only the overall surface topography is imaged, and proteins of interest can be hard to distinguish from similarly sized-objects also present on the r 2011 American Chemical Society

surface. At the opposite, because it uses a microelectrode as a local probe, SECM has the advantage of featuring the inherent selectivity of electrochemical detection techniques. SECM has in particular become an important tool for characterizing the distribution of enzymes on biochemical sensors.5 9 In the most commonly used detection mode (so-called tip-collecting mode) the redox-active product of surface-bound enzymes is collected at the microelectrode (tip) placed a few micrometers above the surface. Hence, measurement of a tip current specifically indicates the presence of the sought enzyme in the surface region located immediately below the tip. This particularly attractive SECM detection scheme can be extended for mapping any protein, including non-redox-active ones, via enzyme labeling of the protein of interest. An interesting example of this enzymatic labeling strategy for protein mapping by SECM is given by the works of Wittstock and co-workers,10,11 Matsue and coworkers,12 15 and others,16 who developed miniaturized versions of enzyme-linked immunosorbent assays (ELISA). In these Received: July 27, 2011 Accepted: September 13, 2011 Published: September 13, 2011 7924

dx.doi.org/10.1021/ac201907v | Anal. Chem. 2011, 83, 7924–7932

Analytical Chemistry works ∼100 μm sized spots of capture antibody/analyte complexes, or simply adsorbed analyte spots, were enzymatically immunomarked using analyte-specific antibody enzyme conjugates and “read” by SECM imaging. However, the major limitation of SECM for imaging the surface distribution of enzymes, or enzyme-tagged proteins, is the relatively poor resolution of this technique, as compared to AFM. Indeed SECM resolution is primarily limited by the size of the microelectrode probe, and because up to now SECM probes larger than a few micrometers in diameter were used for enzyme distribution imaging, a resolution in the 10 μm range has typically been reported in the literature.3,5 Kranz and co-workers,17,18 and Hirata et al.,19 succeeded in improving this resolution by combining SECM with AFM, via the fabrication and use of submicrometer-sized AFM SECM probes, and were able to resolve micrometer-sized enzyme-filled pores on model biosensor surfaces.18 Yet, ultimately, no matter the size of the microelectrode, or the sophistication of the combined SECM-based imaging technique used, SECM mapping of enzymes and enzyme-tagged proteins suffers from a fundamental limitation which renders attainment of submicrometer resolution difficult to achieve. This limitation is rooted in the fact that the enzymatic product has to diffuse from the enzyme to the probe in order to be collected and detected, a process which is only efficient if a sufficiently large collecting microelectrode probe is used.3,5,20 Therefore, new strategies have to be developed in order to use SECM-derived techniques for high-resolution mapping of proteins immobilized on conducting surfaces. In that context we investigate here the feasibility of using Mt (molecule touching)/AFM SECM, an original SECM microscopy we introduced recently,21 which is free of the diffusional limitations of classical SECM and which combines the nanometer resolution of AFM with the selectivity of electrochemistry, to map submicrometer-sized domains of redox-labeled proteins on surfaces. The principle of Mt/AFM SECM is that a forcesensing combined AFM SECM microelectrode probe is used to directly electrochemically contact surface-grafted, redox-tagged macromolecules. We initially demonstrated that such a configuration permitted us to map the distribution of nanometer-sized poly(ethylene glycol) chains end-grafted onto gold electrode arrays.21 In the present work we intend to illustrate that Mt/AFM SECM can also efficiently be used for high-resolution mapping of much more fragile biomacromolecules onto electrode surfaces.

’ EXPERIMENTAL SECTION Chemicals. The heterobifunctional poly(ethylene glycol) derivative NHS PEG3400 Fc (average number of OCH2CH2 monomer units 79), containing a redox ferrocene ethyl unit (Fc) at one end and an amine-reactive N-hydroxysuccinimide (NHS) ester at the other end for Fc PEGylation of the antibodies, was synthesized and characterized as previously described.22 The linear methoxy-terminated PEG2000 disulfide molecule used for particle lithography was prepared by coupling a commercial methoxyPEG2000 NHS ester (Creative PEGWorks, MW 2000) with 2aminoethyl disulfide (cystamine) (Fluka, dihydrochloride salt product) following the same method that was detailed previously for a Fc PEG 3400 disulfide derivative. 23 All chemical and solvents were analytical grade and used without further purification. All aqueous solutions were made with Milli-Q purified water (Millipore). Unless otherwise specified, all the protein-containing solutions were prepared in phosphate-buffered saline (PBS)

ARTICLE

pH 7.4 buffer. The mouse IgG (immunoglobulin-G) used as an antigen, the goat IgG used to evaluate nonspecific binding (ChromPure IgGs, whole molecules), and the antimouse goat IgG (Affinipure antibody) were from Jackson ImmunoResearch Laboratories. The polystyrene latex beads were purchased from Sigma-Aldrich (Ø = 0.46 μm) and Fluka (Ø = 2 μm). Preparation of the Fc PEG-Labeled IgG. Attachment of Fc PEG chains to goat antimouse IgGs was carried out by reacting the NHS activated ester of home-synthesized NHS PEG3400 Fc chains with the amino groups of the IgG species, following a slightly modified previously described procedure24 (see the Supporting Information). Preparation of the Substrate Surfaces. Highly ordered pyrolytic graphite (HOPG) and gold were used as substrate materials. The HOPG surface (Carbone Lorraine, France), was peeled off with tape before each experiment to reveal a fresh surface. Flat gold surfaces were produced by template-stripping of a 200 nm thick gold layer deposited on mica,25 as previously described.21 Assembly of a Saturated Layer of Antigen/IgG PEG Fc Immunocomplexes on HOPG. The freshly peeled HOPG surface was mounted into a liquid cell and covered with 100 μL of a solution of 500 μg/mL mouse IgG (antigen) in PBS buffer. After 2 h the antigen-bearing HOPG surface was thoroughly washed with PBS buffer. A solution of 20 μg/mL Fc PEGlabeled antimouse goat IgG in PBS buffer was then placed on the surface and left to react with the mouse IgG antigen overnight (∼16 h). After washing with PBS buffer, followed by 20 min of desorption, the HOPG surface, then bearing a saturated layer of antigen/antibody complexes, was characterized by cyclic voltammetry in an aqueous phosphate buffer pH 8, 0.1 M ionic strength. Fabrication of the Antigen/IgG PEG Fc Immunocomplex Dot Arrays on Gold Surfaces Using Particle Lithography. The particle lithography technique used to form a regular array of mouse IgG (antigen) dots on a gold surface was adapted from the procedure described by Schmidtke and co-workers to form protein dots on glass.26,27 Full details are given in the Supporting Information. Fabrication of Combined AFM SECM Tips. The tips were hand-fabricated according to a procedure adapted from literature,28 and largely detailed elsewhere.29 Briefly, a 60 μm diameter gold wire is flattened, and its extremity is successively bent and etched, so as to obtain a flexible cantilever (spring constant in the 0.5 3 N/m)23 bearing a conical tip with a spherical apex ∼100 nm in radius. The tip is fully insulated by deposition of an electrophoretic paint and glued onto an AFM chip. The apex is selectively exposed in order to act as a current-sensing nanoelectrode. AFM and Combined AFM SECM Experiments. In situ AFM images were acquired using a JPK microscope operated in tapping mode. The AFM SECM experiments were carried out with a Molecular Imaging PICOSPM I AFM microscope (Agilent), which was modified as described in previous contributions.21,23 Cyclic and Differential Pulse Voltammetry. Cyclic voltammetry (CV) characterization of saturated antigen/IgG PEG Fc layers on HOPG and gold surfaces was carried out using the “substrate” channel of the homemade AFM SECM bipotentiostat described in the Supporting Information. The patterned gold surfaces, bearing antigen/IgG PEG Fc immunocomplex dots, were characterized using CV and differential pulse voltammetry (DPV); see the Supporting Information. A CHI600 electrochemical workstation (CH instrument, Austin, Texas) was used for DPV. 7925

dx.doi.org/10.1021/ac201907v |Anal. Chem. 2011, 83, 7924–7932

Analytical Chemistry

ARTICLE

Figure 1. Working principle of Mt/AFM SECM detection of proteins immobilized onto a conducting surface. The target protein is recognized by a specific antibody (IgG) labeled by flexible PEG chains bearing redox ferrocene (Fc) heads. A combined AFM SECM probe is brought in molecular contact with the antigen/antibody complex, and provided the PEG chain is long enough and the standard potential of the Fc heads falls between the tip and substrate potentials, tip-to-substrate redox cycling of the IgG-borne Fc heads occurs, generating a specific SECM positive feedback current.

Figure 2. Assembly and cyclic voltammetry characterization of a saturated mouse IgG/antimouse IgG PEG Fc immunocomplex layer on an HOPG electrode surface. (Top) Depiction of the immunological redox marking of the mouse IgG antigen layer by the Fc PEGylated antibody. (Bottom) Cyclic voltamogram recorded at the HOPG electrode bearing a saturated layer of adsorbed antigens (mouse IgG), before (dotted green trace) and after (solid blue trace) immunological recognition of the antigens by the Fc PEG-labeled antimouse antibody (IgG PEG Fc). Aqueous phosphate buffer pH 8, 0.1 M ionic strength. T = 25 °C. Scan rate, 2 V/s. Surface concentration in Fc PEG = 1.7  10 11 mol/cm2. HOPG electrode surface area, 0.7 cm2.

’ RESULTS AND DISCUSSION The minimal requirement for visualizing protein molecules on a surface using Mt/AFM SECM is that the molecules to be imaged bear a redox label. As a versatile way of redox labeling target protein molecules immobilized on a substrate (electrode) surface we propose to make use of specific antibodies (IgGs), raised against the target protein to map, and redox-labeled by ferrocene (Fc) moieties attached to flexible poly(ethylene glycol) (PEG) chains (Figure 1). The Fc PEGylated antibody recognizes the target protein, forming an immunocomplex, and provided the PEG tether is long enough and the tip and substrate potentials flank the standard potential of the Fc heads, Fc moieties can then shuttle electrons from the surface to the tip. Measurement of the resulting SECM positive feedback tip current hence allows the specific detection and mapping of the sought protein molecules on the surface. The redox immunomarking strategy we propose here is similar to the one currently employed in immunohistochemistry where fluorescence labeled antibodies are used to visualize target proteins. Its interest is threefold: (i) It is largely applicable since antibodies can be produced against a large variety of antigens. (ii) It affords a very high specificity, since the antigen antibody immunological recognition reaction is highly specific. (iii) The protein to be imaged does not have to be chemically modified to bear a label, and its native conformation is thus preserved. For the proof-of-concept experiments we report here, model protein-bearing electrode surfaces were fabricated by adsorbing a mouse immunoglobulin-G (IgG), playing the role of the antigenic target protein to map, onto HOPG and gold conducting substrates. Consequently, Fc PEG chains were covalently linked to antimouse antibodies (goat IgGs). Fc PEG chains of 3400 molecular weight were specifically selected since we previously demonstrated that these chains are sufficiently long and

flexible to shuttle electrons over a few tens of nanometers within multilayered IgG PEG Fc assemblies immunologically constructed onto vitreous carbon electrode surfaces.24,30 Fc PEGylation of the antimouse IgG was carried out through the reaction of the accessible amino groups of the goat immunoglobulin with the NHS-activated ester function borne by custom-synthesized NHS PEG3400 Fc linear heterobifunctional chains. As described in earlier works Fc PEG3400-functionalized IgGs antibody bear from 5 to 10 Fc PEG chains.24 It is worth noting that many other antibodies labeled by ferrocene,31 35 or by other redox moieties,36,37 have also been described in literature and used as markers in electrochemical immunoassays. However, because their redox labels were directly linked to the surface of the IgGs, i.e., not via a flexible chain, these redox-IgGs may not allow electrons to be transported over sufficiently long distances to be used as immunomarkers for Mt/AFM SECM. Redox Immunomarking of a Layer of Mouse IgG Antigens by Fc PEG-Labeled Antimouse Antibodies. In order to assess the efficiency of redox immunomarking by the Fc PEG-labeled antimouse antibody, a layer of mouse IgG adsorbed on HOPG was exposed overnight to an IgG PEG Fc-containing solution (Figure 2). HOPG was selected here as the electrode material due to it great smoothness and ease of use. Formation of the antigen/IgG PEG Fc immunocomplex layer was checked by CV as presented in Figure 2. After exposure to the IgG PEG Fc solution, the CV recorded at the antigen-bearing HOPG surface changed from being purely capacitive (Figure 2, dotted trace) into the typical reversible peak-shaped signal due to the surfaceconfined Fc heads undergoing fast (Nernstian) electron transfer (Figure 2, solid trace).38 The peak current was proportional to the scan rate, the peakto-peak separation was small (∼5 mV at 2 V/s), while the (almost) common value of the forward and backward peak 7926

dx.doi.org/10.1021/ac201907v |Anal. Chem. 2011, 83, 7924–7932

Analytical Chemistry potential was of ∼0.15 V/SCE. This later value coincides with the value previously reported for the standard potential E°Fc/Fc+ of the PEG-borne Fc/Fc+ redox couple.39 Such a behavior, that we reported previously and confirmed here, indicates that the IgG PEG Fc is able to recognize its antigen on the surface, forming an antigen/antibody immunological redox-active complex as sketched in Figure 2. Thanks to the flexibility of the PEG chains, the Fc heads can reach the substrate surface and exchange electrons with it. Integration of the cathodic (or anodic) peak of the background-corrected signal obtained after overnight exposure of the antigen surface to the Fc PEG-labeled antibody yielded a maximum surface coverage in Fc heads of ΓFc ∼ 1 2  10 11 mol/cm2 (see the Experimental Section). Considering that 5 10 Fc PEG chains decorate each IgG PEG Fc,24 the antibody surface concentration is then of ΓIgG ∼ (1 4)  10 12 mol/cm2. This coverage value translates into an average distance separating two neighboring Fc PEG IgGs of 1/(N ΓIgG)1/2 ∼ 7 15 nm, which is comparable with the known size of an IgG molecule (∼5 15 nm depending on its orientation, see below).40 This later result indicates that a saturated layer of antigen/antibody immunocomplexes is formed on the HOPG surface. Importantly, if a goat IgG, i.e., an antigen for which the antimouse IgG PEG Fc has little or no affinity, is adsorbed on the surface instead of the mouse IgG, a surface concentration in Fc PEG chains (i.e., IgG PEG Fc) representing ∼5% of the saturating coverage is recorded (ΓFc < 0.5  10 12 mol/cm2). This result underlines that the antimouse IgG PEG Fc specifically binds to its adsorbed antigen and not to some other location on the surface, i.e., less than 5% nonspecific binding is observed. Contacting Fc PEG-Labeled IgGs with an Incoming Oscillating Microelectrode: Tapping Mode Mt/AFM SECM. Mapping the distribution of the adsorbed target mouse antigen in Mt/AFM SECM implies being able to establish and maintain a delicate physical and electrochemical contact between the metal tip and the Fc PEG-labeled antibody molecules. In order to keep this contact as gentle as possible, Mt/AFM SECM was operated in tapping mode. In this particular AFM mode, designed to minimize tip substrate interactions, the tip is mechanically oscillated and the amplitude of the tip oscillation is used as the input of the distance regulating feedback loop. Hence, before going further, one needs to understand the physical interactions and electrochemical behavior of an oscillating tip approaching a redox-labeled protein layer; i.e., one needs to record and analyze tapping mode Mt/AFM SECM approach curves. In a typical experiment, a combined AFM SECM probe was acoustically excited at its fundamental flexural resonance frequency (∼2 3 kHz) so that it oscillated with a free amplitude of A0 ∼ 10 12 nm. The oscillating probe (biased at Etip = +0.30 V/SCE) was approached in situ toward the HOPG surface bearing a layer of the antigen/IgG PEG Fc complexes (biased at Esub = 0.05 V/SCE). The tip oscillation amplitude, A, and tip current itip were simultaneously recorded and plotted as a function of the piezoelongation Z in characteristic approach curves, such as the ones presented in Figure 3. One can see in Figure 3a that upon approaching the probe toward the surface the tip oscillation is progressively damped, as a result of tip surface interactions. Concomitantly, a tip current is recorded (Figure 3b), which smoothly increases as the probe is pushed toward the substrate. After some critical piezo elongation, taken as the origin of the Z axis in Figure 3 (i.e., Z = 0), the probe oscillation is fully damped (A = 0) and the tip is then in permanent contact with the surface. As seen in Figure 3b, pushing

ARTICLE

Figure 3. Mt/AFM SECM tapping mode approach curves recorded upon approaching a combined AFM SECM probe toward an HOPG surface bearing a saturated layer of mouse IgG/antimouse IgG PEG Fc immunocomplex. The oscillation amplitude of the probe A and the tip current itip are plotted as function of the piezo elongation Z, respectively, in curves a and b. The origin of the Z axis is taken as the point where the tip oscillation is fully damped (A = 0). Hence, for Z > 0 the time-averaged tip-to-substrate distance is d = Z. The green star symbol denotes the amplitude set point for imaging and corresponding current. The inset in panel a shows a schematic view of the oscillating tip approaching the immunocomplex layer; the time-averaged tip-to-substrate distance d is represented. The inset in panel b shows the substrate potential (Esub) dependence of the tip current (itip) measured for a fixed tip-to-substrate distance d ∼ 11 nm corresponding to the imaging set point. Aqueous phosphate buffer pH 8, 0.1 M ionic strength. Tip and substrate potentials: Etip = +0.30 V/SCE, Esub = 0.05 V/SCE. The probe is oscillated at its fundamental flexural frequency = 2.157 kHz. Free amplitude is ∼11 nm.

the tip toward the surface beyond this point results in a slight current increase followed by a current plateau. In this contactmode-like regime, the protein layer is compressed by the tip and the tip is deflected upward. However, for Z > 0, i.e., in the actual tapping mode regime, we observed that the overall (timeaveraged) vertical deflection of the tip was zero, meaning that one can define an average tip-to-substrate distance d as d = Z. As already explained, in tapping mode operation, the amplitude of the tip oscillation is the input of the distance regulating feedback loop; hence setting a constant amplitude of, for example, ∼8 nm, allowed an average tip substrate separation d of ∼11 nm to be indefinitely maintained by the feedback loop, i.e., the tip was held at the position marked by a green start symbol in Figure 3a. The faradaic nature of the corresponding ∼0.4 pA tip current (green start symbol in Figure 3b) could then be unambiguously demonstrated by cycling the substrate potential from its initial value of Esub = 0.05 V/SCE down to +0.30 7927

dx.doi.org/10.1021/ac201907v |Anal. Chem. 2011, 83, 7924–7932

Analytical Chemistry

ARTICLE

Figure 4. Fabrication of antigen/IgG PEG Fc immunocomplex dot arrays onto gold electrode surfaces via particle lithography. The inset in the lower right part of the figure shows the differential pulse voltammogram (DPV) of the gold surface bearing the mouse IgG dot array (fabricated using 2 μm diameter beads) after redox immunomarking with the Fc PEGylated antimouse IgG. DPV parameters and the corresponding cyclic voltammogram (CV) are given in the Supporting Information.

V/SCE and back. The resulting itip versus Esub variation, shown in the inset in Figure 3b, is a typical voltammogram centered at a potential close to the E° of the Fc heads. This result provides conclusive evidence that the electrochemical current recorded at the oscillating probe is indeed due to the tip electrochemically contacting the Fc heads borne by the Fc PEG-labeled IgGs, as depicted in Figure 1. The Fc heads are oxidized when colliding with the tip and are subsequently reduced back upon contacting the substrate surface. This SECM positive feedback cycle generates a distinctive tip current whose intensity is controlled by the motional dynamics of the PEG chains and is also proportional to the amount of chains under the tip, i.e., to the local surface density in redox immunocomplex. The plateau current observed in the “contact” regime of the current approach curve (i.e., for Z < 0) indicates hindered diffusional motion of the confined Fc PEG chain.41 The working principle of Mt/AFM SECM detection of redox-immunomarked surface-immobilized proteins is therefore now established. Beyond this important result, analyzing the Mt/AFM SECM tapping mode approach curves can also yield insights regarding the structure of the antigen/antibody immunocomplex layer on the HOPG surface. However, this first requires that the expected overall thickness of the immunolayer is estimated from the molecular sizes and possible orientations of the IgGs on the surface, as follows. The height of the adsorbed mouse IgG can range from 5 to 15 nm depending on whether the IgG lies flat on the electrode or is in an upright position.40 The orientation of the IgG PEG Fc antibody is better defined: At least one of the Fab subunits of the IgG molecule, bearing the antigen binding sites represented by black rectangles in Figure 1, is necessarily oriented toward the adsorbed antigen surface. However, since the subunits of an IgG are connected via flexible hinges, various orientations can still be envisioned for the IgG PEG Fc molecules. Yet, within the saturated antigen/ antibody layer, steric repulsion between neighboring IgGs probably favors an upright orientation of the IgG PEG Fc. As a result, it seems reasonable to consider that the Fc PEG-labeled antibody layer is ∼15 nm thick. Lastly we know from previous experiments where surface end-grafted Fc PEG3400 were characterized by Mt/AFM SECM, that Fc PEG3400 chains can most efficiently transport electrons over ∼10 nm.21 Hence, the effective thickness of the mouse IgG/antimouse IgG PEG Fc

layer, as probed by the tip, should be in the order of 30 40 nm. With this figure in mind one can now go back to the current approach curves shown in Figure 3b. One can see that a current starts to be detected from a threshold d value of ∼40 50 nm. Recalling that, in tapping mode, the instantaneous tip-to-substrate distance actually varies from d + A down to d A, the observed threshold distance is fully compatible with the 30 40 nm range expected for the effective thickness of the redoximmunocomplex layer. This later result confirms that the antigen/antibody immunocomplexes form a well-defined singlemolecular layer on the electrode surface. The next issue to address is to demonstrate the ability of Mt/AFM SECM to resolve the 2D distribution of redoxlabeled IgGs heterogeneously distributed onto a surface. For an unambiguous demonstration one has to be able to image a test surface bearing antigen/IgG PEG Fc complexes arranged in an heterogeneous but predefined manner. Among the available techniques allowing regular protein arrays to be assembled on surfaces, we selected an easily implemented benchtop method, particle lithography, in order to fabricate the required antigenpatterned surfaces. In the present case, this assembly technique made necessary that the substrate material was changed from HOPG to gold. Resolving the 2D Distribution of Proteins on Surfaces: Mt/AFM SECM Imaging of Antigen/IgG PEG Fc Immunocomplex Dot Arrays. Recently, Schmidtke and co-workers reported a technique combining particle lithography with PEG silane chemistry to assemble hexagonal protein dot arrays on glass.26,27 In the present work, this technique was adapted in order to assemble regular patterns of adsorbed mouse IgG (antigen) onto gold surfaces, which were subsequently redox-immunomarked by Fc PEGylated antimouse antibodies, as schematically represented in Figure 4. An aqueous solution containing 2 μm diameter polystyrene beads was allowed to dry onto a flat template-stripped (TS) gold surface. Upon solvent evaporation the beads were observed to form multilayers of hexagonally packed assemblies. Bead adhesion to the gold surface was then reinforced by curing the surface at 80 °C for 60 min. After the surface had cooled down to room temperature it was immersed in a 1 mM PEG2000 disulfide aqueous solution and allowed to react overnight under argon. As depicted in Figure 4, and as anticipated from the work of 7928

dx.doi.org/10.1021/ac201907v |Anal. Chem. 2011, 83, 7924–7932

Analytical Chemistry Schmidtke and co-workers, the beads served as a lithographic mask, restricting PEG grafting to the gold surface regions surrounding the beads. The beads were then removed by ultrasonication, leaving on the surface an array of “holes” where bare gold was exposed, the rest of the gold surface bearing a self-assembled monolayer of protein-repellent PEG2000 chains. The surface was then immersed in a PBS buffer solution containing the antigen (mouse IgG) which selectively adsorbed in the “holes”, leading to the formation of a periodic array of small antigen domains (dots), as revealed by regular AFM imaging of the antigen patterned surface; see parts B and C of Figure S1 in the Supporting Information. In the present case, as a final step, the antigen domains were “revealed” (i.e., immunologically redox labeled) by immersing the surface into a solution containing the Fc PEGylated antimouse IgG. Successful redox immunomarking of the antigen dots by the Fc PEG IgG antibody was ascertained by recording a differential pulse voltammogram at the patterned gold surface showing a single peak in the 0.10 0.15 V/SCE region (see the inset in Figure 4). DPV was preferred to CV due to its higher sensitivity for detecting the very small amount of IgG PEG Fc molecules present on the patterned surface (see Supporting Information Figure S2). The surface was then imaged using Mt/AFM SECM in tapping mode in phosphate buffer pH 8. Typical topography and current images of the gold surface bearing the redox-immunocomplex dot array are presented, respectively, in parts a and b of Figure 5. The tip and substrate potentials were, respectively, set to Etip = +0.30 V/SCE, Esub = 0.05 V/SCE, the image scan rate was 0.4 Hz, and tip damping was set to ∼20 30%. As exemplified in Figure 5a an array of antigen/antibody immunocomplex dots, forming a slightly distorted hexagonal motif, was observed in the topography image. The spacing between dots was of ∼2 μm, in good agreement with the diameter of the beads used for lithography. The diameter of the dots was in the order of 0.3 0.4 μm, even though in some occasions a few smaller dots, down to ∼100 nm in diameter, were also seen (see top of Figure 5a). The height of the dots, as can be measured from the cross section of the topography image presented below Figure 5a, was in the order of ∼10 15 nm. We note that this apparent height is smaller than the 15 30 nm height expected from the size of the antigen/antibody complex, as estimated above. This result may indicate that the immunocomplex is slightly squeezed by the tip during imaging, as is often observed in tapping mode imaging of soft matter and biomolecules,42,43 or that the adsorbed antigen is partly deformed. Yet what is remarkable is that, due to the extreme efficiency of the protein-repellant nature of the PEG2000 layer, no IgG molecule was seen between the dots from tapping mode topography images acquired either in Mt/AFM SECM or in regular AFM microscopy (see parts B and C of Supporting Information Figure S1). Most importantly, one can see that almost all of the dots imaged in topography also appear in the current image as rounded current spots, compare parts a and b of Figure 5. Only a few dots, such as the small one seen in the top part of the topography image in Figure 5a, did not give rise to a measurable current spot (see Figure 5b). This later result is important since it demonstrates that Mt/AFM SECM can discriminate protein dots recognized by the Fc PEG-labeled antibody from unreacted dots similar in size. Incidentally, the same result also shows that no cross talk exists between the topography and current signals. Yet, in most cases, when the tip was scanned over

ARTICLE

Figure 5. Mt/AFM SECM tapping mode imaging of a gold surface bearing an ordered array of adsorbed mouse IgG (antigen) dots fabricated using bead lithography (2 μm beads) and subsequently recognized by a Fc PEG-labeled antimouse IgG: (a) topography image; (b) simultaneously acquired tip current image. The almost hexagonal symmetry of the dot array is underlined. Cross sections of the topography and current images along the short white line, passing through the center of an antigen/antibody immunocomplex dot, are shown. Aqueous phosphate buffer pH 8, 0.1 M ionic strength. Tip and substrate potentials: Etip = +0.30 V/SCE, Esub = 0.05 V/SCE. The probe is oscillated at its fundamental flexural frequency = 2.82 kHz; ∼25% damping; imaging rate, 0.3 Hz.

a dot a bell-shaped current signal, characterized by an ∼0.3 0.5 pA peak intensity and an ∼0.3 0.4 μm width at mid height, was indeed recorded (see the cross section of a current spot shown below Figure 5b). It is worth noticing that we observed no major changes in current or in topography upon prolonged imaging of the patterned surface. Hence, one can conclude that, as a benefit of using tapping mode, the fragile protein dots were not damaged by the tip during imaging. In order to ascertain the origin of the recorded current spots, the dependence of the current contrast on the tip and substrate potentials was studied. To this aim a 6 μm  6 μm region of the antigen/antibody patterned gold surface was repeatedly imaged three times. The pairs of topography and current images successively acquired during the initial, second, and third scans are presented, respectively, in parts a and b, parts c and d, and parts e and f of Figure 6. The first scan (Figure 6, parts a and b) was meant to acquire reference images and was consequently carried out at fixed tip and substrate potentials of Etip = +0.30 V/SCE, Esub = 0.05 V/SCE. The second scan (Figure 6, parts c and d) served to specifically probe the effects of the substrate potential, whereas the third scan (Figure 6, parts e and f) was acquired in order to evidence tip potential effects. Considering first the topography images of the three scans (Figure 6, parts a, c, and e) one can see that, in spite of an apparent upward and leftward drift of the imaged area from scan to scan, some of the protein dots can be identified on all of the sequentially acquired images. In particular, the two dots enclosed in a circle for one (upper dot) and a square symbol for the other (lower dot) are seen in each of the three topography images presented in Figure 6, parts a, c, and e. The shape and separation 7929

dx.doi.org/10.1021/ac201907v |Anal. Chem. 2011, 83, 7924–7932

Analytical Chemistry

Figure 6. Mt/AFM SECM tapping mode imaging of a gold surface bearing an ordered array of antigen/IgG PEG Fc immunocomplex. Illustration of the effects of the tip potential Etip and substrate potential Esub on the topography and current contrasts. The very same location of the substrate was scanned three times successively. Pairs of topography and current images (a and b), (c and d), (e and f) were acquired, respectively, during the first, second, and third scans. From top to bottom: for the first scan (1) the tip and substrate potentials were kept at constant values of Etip = +0.30 V/SCE, Esub = 0.05 V/ SCE. During the second scan (2) the substrate potential was swept to +0.30 V/SCE and back to 0.05 V/SCE in the region delimited by the two horizontal dotted lines seen in the current image (d). During the third scan (3) the tip potential was changed to 0.05 V/SCE in the middle of the image, as indicated in the current image (f). Aqueous phosphate buffer pH 8, 0.1 M ionic strength. The probe is oscillated at its fundamental flexural frequency = 2.79 kHz; ∼30% damping; imaging rate, 0.2 Hz.

between some of the dots is seen to vary slightly from scan to scan, as a result of drift and hysteresis of the piezotube of the AFM scanner, and to tip asymmetry. These effects are common artifacts in AFM imaging44 and do not hamper visualization of the dots of interest here. Considering now specifically the second scan (Figure 6, parts c and d): in the image region delineated by the horizontal dotted lines shown in the current image Figure 6d, the substrate potential was swept at 5 mV/s from 0.05 to +0.30 V/SCE and back. One can see that, whereas this Esub sweep had no

ARTICLE

consequence on the topography image, the spot circled in a dotted line, imaged at an Esub potential too anodic for the reduction of the ferricinium Fc+ heads to take place, disappeared from the current image. Similarly, during the third scan (Figure 6, parts e and f) the tip potential was changed from +0.30 to 0.05 V/SCE, a potential such that the tip can no longer oxidize the Fc heads, with no consequences on the topography image but with the effect of making the circled spot disappear from the current image. In summary, it was observed that the dots gave rise to current spots only when the tip and substrate potentials were such that redox cycling of the Fc heads, schematized in Figure 1, could occur. This result illustrates the selectivity of the current detection of the immunocomplex. Indeed, the possibility of scanning (or at least changing) the tip and substrate potentials while acquiring meaningful topography data is of utmost importance since it opens the possibility of using the Esub or Etip dependence of the tip current contrast to ascertain that features seen in the current image solely reflects the distribution of the sought antigen/ antibody complex on the surface. This feature also makes Mt/AFM SECM able to differentiate and colocalize IgGs tagged by redox labels differing in standard potentials and directed against different antigens. Hence, the relative distribution of several antigenic molecules of interest could in principle be mapped using Mt/AFM SECM. The possibility of mapping the 2D distribution of a target antigen on the surface of an electrode, using immunological recognition by a redox-labeled antibody and Mt/AFM SECM imaging, is at this stage fully established. It is also worth reporting that Mt/AFM SECM images of antigen/Fc PEG-labeled antibody dot arrays, similar to the ones presented in Figure 5, could also be acquired with the gold substrate unbiased (i.e., in open circuit conditions, see Supporting Information Figure S3). In such a case, the Fc heads borne by all of the dots on the quite large substrate surface are effectively interconnected via the gold substrate material and provide a redox buffer capacity sufficiently large to sustain the very low imaging current. This result extends the applicability of Mt/AFM SECM to the imaging of redoxlabeled biomacromolecules immobilized onto any conducting unbiased surface. Estimating the Resolution Attainable in Mt/AFM SECM Mapping of Antigen/Antibody Immunocomplex-Bearing Surfaces. Estimating the lateral resolution of Mt/AFM SECM imaging of redox-labeled biomacromolecules comes down to the question of the smallest antigen/IgG PEG Fc immunocomplex dot which can be resolved both in the topography and in the current images. One way of fabricating smaller protein dots using the particle lithography method described above is to use smaller diameter beads.26,27 We thus replaced the 2 μm beads we used previously by 0.5 μm diameter beads and followed the abovedescribed protocol to fabricate an antigen/antibody dot array onto a gold substrate. Mt/AFM SECM images of such a patterned substrate are shown in Figure 7. The topography image (Figure 7a) reveals the presence of a slightly distorted hexagonally packed array of protein dots. The dot spacing is ∼0.5 μm, in good agreement with the size of the beads, whereas the dot height is, as previously, in the 10 15 nm range. The diameter of the immunomarked dots falls in the 100 150 nm range (see cross sections below the topography image in Figure 7) and was as expected reduced by using smaller beads for particle lithography. By comparing the topography image Figure 7a with the current image Figure 7b one can see that 7930

dx.doi.org/10.1021/ac201907v |Anal. Chem. 2011, 83, 7924–7932

Analytical Chemistry

ARTICLE

such a small number of biomolecules, and probably even less, can be specifically detected in Mt/AFM SECM.

Figure 7. Mt/AFM SECM tapping mode imaging of a gold surface bearing an ordered array of adsorbed mouse IgG (antigen) dots fabricated using bead lithography (0.5 μm beads) and subsequently recognized by the Fc PEG antimouse IgG: simultaneously acquired (a) topography image and (b) tip current image. Cross sections of the topography and current images along the short white line, passing through the center of an antigen/antibody immunocomplex dot, are shown. Aqueous phosphate buffer pH 8, 0.1 M ionic strength. Tip and substrate potentials: Etip = +0.30 V/SCE, Esub = 0.05 V/SCE. The probe is oscillated at its fundamental flexural frequency = 1.89 kHz; ∼20% damping; imaging rate, 0.3 Hz.

almost every antigen/antibody immunocomplex dot visible in the topography image also appears as a spot in the current image. The resolution attained here, defined as the smallest dot appearing both in the topography and in the current image, is in the order of ∼100 nm. This resolution was limited by the tip size (the probes had ∼100 nm tip radius) and by the size of the smallest protein dots we could fabricate using the particle lithography technique. Limitation of the resolution due to the tip size (i.e., tip convolution effects) is evidenced by the fact that the dots appear broader in the current than in the topography images shown in Figure 7b. Indeed, we previously observed for other systems imaged in tapping mode Mt/AFM SECM20 that the topography image tends to be better resolved than the current image. This phenomenon is due to the fact that, whereas the resolution in current is governed by the area of the tip end accessible to the ∼10 nm long flexible PEG chains, resolution in topography is dictated by much shorter-ranged tip sample physical interactions. Hence, resolution could probably be significantly improved to a few tens of nanometers by using much smaller-sized etched tips. The use of polystyrene beads smaller than ∼0.5 μm could also in principle allow regular arrays of proteins dots even smaller than the ones reported here to be formed on gold electrodes. Yet we observed that the simple drying procedure we used to assemble the polystyrene beads into ordered arrays tended to fail for beads smaller than 0.5 μm in diameter. This problem could be solved by using more sophisticated techniques to assemble bead arrays such as Langmuir Blodgett-like techniques.45 Yet it is interesting to note that within each of the ∼100 nm diameter dots imaged in Figure 7 only ∼50 redox-labeled IgGs are at most present, assuming a saturating coverage in IgG of ∼10 12 mol/cm2 (as estimated above). It is remarkable that

’ CONCLUSION It was demonstrated here that Mt/AFM SECM imaging, combined with immunological labeling by Fc PEGylated antibodies, can be used to map the surface distribution of submicrometer-sized domains of IgGs adsorbed onto a gold electrode substrate, with an effective resolution in the order of ∼100 nm. These results can conceptually be transposed to the Mt/ AFM SECM imaging of any target protein immobilized onto a surface provided that (i) the surface is an electrode or simply conducting, (ii) antibodies against the protein to map are available, and (iii) the length of the PEG chain is tuned as a function of the size of the protein. The interest of the Mt/ AFM SECM configuration is that it combines the nanometric resolution of AFM with the selectivity of the electrochemical detection. As a result target proteins can be identified amidst similarly sized “objects” present on the surface, based on their specific recognition by redox-labeled IgGs. Moreover, the use of antibodies differing both in their specificity and redox labels could allow the colocalization of multiple target antigens on a surface simply by varying the tip and/or substrate potential. Finally, since the tip-to-substrate cycling of the redox label is a very efficient electrochemical amplification mechanism,46 49 single protein molecule imaging by Mt/AFM SECM can reasonably be contemplated. ’ ASSOCIATED CONTENT

bS

Supporting Information. Experimental details and additional figures. This material is available free of charge via the Internet at http://pubs.acs.org.

’ AUTHOR INFORMATION Corresponding Author

*E-mail: [email protected].

’ ACKNOWLEDGMENT This work has been supported by the Region Ile-de-France in the framework of C’Nano IdF. C’Nano IdF is the nanoscience competence center of Paris Region, supported by CNRS, CEA, MESR, and Region Ile-de-France. ’ REFERENCES (1) Casero, E.; Vazquez, L.; Parra-Alfambra, A. M.; Lorenzo, E. Analyst 2010, 135, 1878–1903. (2) Bard, A. J. In Scanning Electochemical Microscopy; Bard, A. J., Mirkin, M. V., Eds.; Marcel Dekker: New York, 2001; pp 1 15. (3) Wittstock, G.; Burchardt, M.; Pust, E. S.; Shen, Y.; Zhao, C. Angew. Chem., Int. Ed. 2007, 46, 1584–1617. (4) Sun, P.; Laforge, F. O.; Mirkin, M. V. Phys. Chem. Chem. Phys. 2007, 9, 802–823. (5) Pierce, D. T.; Bard, A. J. Anal. Chem. 1993, 65, 3598–3604. (6) Horrocks, B. R.; Wittstock, G. In Scanning Electrochemical Microscopy; Bard, A. J., Mirkin, M. V., Eds.; Marcel Dekker: New York, 2001; pp 445 519. (7) Kranz, C.; Wittstock, G.; Wohlschl€ager, H.; Schuhmann, W. Electrochim. Acta 1997, 42, 3105–3111. (8) Maciejewska, M.; Sch€afer, D.; Schuhmann, W. Electrochem. Commun. 2006, 8, 1119–1124. 7931

dx.doi.org/10.1021/ac201907v |Anal. Chem. 2011, 83, 7924–7932

Analytical Chemistry (9) Hussien, E. M.; Erichsen, T.; Schuhmann, W.; Maciejewska, M. Anal. Bioanal. Chem. 2008, 391, 1773–1782. (10) Wittstock, G.; Yu, K.-J.; Halsall, H. B.; Ridgway, T. H.; Heineman, W. R. Anal. Chem. 1995, 67, 3578–3582. (11) Wijayawardhana, C. A.; Wittstock, G.; Halsall, H. B.; Heineman, W. R. Electoanalysis 2000, 12, 640–644. (12) Shiku, H.; Matsue, T.; Uchida, I. Anal. Chem. 1996, 68, 1276– 1278. (13) Shiku, H.; Hara, Y.; Matsue, T.; Uchida, I.; Yamauchi, T. J. Electroanal. Chem. 1997, 438, 187–190. (14) Kasai, S.; Yokota, A.; Zhou, H.; Nishizawa, M.; Niwa, K.; Onouchi, T.; Matsue, T. Anal. Chem. 2000, 72, 5761–5765. (15) Yasukawa, T.; Hirano, Y.; Motochi, N.; Shiku, H.; Matsue, T. Biosens. Bioelectron. 2007, 22, 3099–3104. (16) Zhang, X.; Peng, X.; Jin, W. Anal. Chim. Acta 2006, 558, 110–114. (17) Kueng, A.; Kranz, C.; Lugstein, A.; Bertagnolli, E.; Mizaikoff, B. Angew. Chem., Int. Ed. 2003, 42, 3238–3240. (18) Kranz, C.; Kueng, A.; Lugstein, A.; Bertagnolli, E.; Mizaikoff, B. Ultramicroscopy 2004, 100, 127–134. (19) Hirata, Y.; Yabuki, S.; Mizutani, F. Bioelectrochemistry 2004, 63, 217–224. (20) Anne, A.; Cambril, E.; Chovin, A.; Demaille, C.; Goyer, C. ACS Nano 2009, 3, 2927–2940. (21) Anne, A.; Cambril, E.; Chovin, A.; Demaille, C. Anal. Chem. 2010, 82, 6353–6362. (22) Anne, A.; Moiroux, J. Macromolecules 1999, 32, 5829–5835. (23) Anne, A.; Demaille, C.; Goyer, C. ACS Nano 2009, 3, 819–827. (24) Anne, A.; Demaille, C.; Moiroux, J. J. Am. Chem. Soc. 1999, 121, 10379–10388. (25) Hegner, M.; Wagner, P.; Semenza, G. Surf. Sci. 1993, 291, 39–46. (26) Taylor, Z. R.; Patel, K.; Spain, T. G.; Keay, J. C.; Jernigen, J. D.; Sanchez, E. S.; Grady, B. P.; Johnson, M. B.; Schmidtke, D. W. Langmuir 2009, 25, 10932–10938. (27) Taylor, Z. R.; Sanchez, E. S.; Keay, J. C.; Johnson, M. B.; Schmidtke, D. W. Langmuir 2010, 26, 18938–18944. (28) Macpherson, J. V.; Unwin, P. R. Anal. Chem. 2000, 72, 276–285. (29) Abbou, J.; Demaille, C.; Druet, M.; Moiroux, J. Anal. Chem. 2002, 74, 6355–6363. (30) Anne, A.; Demaille, C.; Moiroux, J. J. Am. Chem. Soc. 2001, 123, 4817–4825. (31) Akram, M.; Stuart, M. C.; Wong, D. K. Y. Electroanalysis 2006, 18, 237–246. (32) Okochi, M.; Ohta, H.; Tanaka, T.; Matsunaga, T. Biotechnol. Bioeng. 2005, 90, 14–19. (33) Lim, T. K.; Ohta, H.; Matsunaga, T. Anal. Chem. 2003, 75, 3316–3321. (34) Wang, J.; Iba~ nez, A.; Chatrathi, M. P. Electrophoresis 2002, 23, 3744–3749. (35) Kossek, S.; Padeste, C.; Tiefenauer, L. J. Mol. Recognit. 1996, 9, 485–487. (36) Wei, M.-Y.; Wen, S. D.; Yang, X.-Q.; Guo, L. H. Biosens. Bioelectron. 2009, 24, 2909–2914. (37) Wei, M.-Y.; Wen, S. D.; Yang, X.-Q.; Guo, L. H. Anal. Chim. Acta 2009, 632, 15–20. (38) Laviron, E. Voltammetric Methods for the Study of Adsorbed Species. In Electroanalytical Chemistry; Bard, A. J., Ed. ; Marcel Dekker: New York, 1982; Vol. 12, pp 53 157. (39) Anne, A.; Demaille, C.; Moiroux, J. Macromolecules 2002, 35, 5578–5586. (40) Lamy, J.; Lamy, J.; Billiald, P.; Sizaret, P. Y.; Cave, G.; Frank, J.; Motta, G. Biochemistry 1985, 24, 5532–5542. (41) Abbou, J.; Anne, A.; Demaille, C. J. Phys. Chem. B 2006, 110, 22664–22675. (42) Rodriguez, R. D.; Anne, A.; Cambril, E.; Demaille, C. Ultramicroscopy 2011, 111, 973–981. (43) Yang, C.-W.; Hwang, I.-S.; Chen, Y. F.; Chang, C. S.; Tsai, D. P. Nanotechnology 2007, 18, 084009 (8 pp).

ARTICLE

(44) Atomic Force Microscopy; Eaton, P., West, P., Eds.; Oxford University Press: New York, 2010; pp 121 135. (45) Marquez, M.; Grady, B. P. Langmuir 2004, 20, 10998–11004. (46) Fan, F.-R. F.; Bard, A. J. Science 1995, 267, 871–874. (47) Fan, F.-R. F.; Juhyoun, K.; Bard, A. J. J. Am. Chem. Soc. 1996, 118, 9669–9675. (48) Sun, P.; Mirkin, M. V. J. Am. Chem. Soc. 2008, 130, 8241–8250. (49) Zevenbergen, M. A. G.; Wolfrum, B. L.; Goluch, E. D.; Singh, P. S.; Lemay, S. G. J. J. Am. Chem. Soc. 2009, 131, 11471–11477.

7932

dx.doi.org/10.1021/ac201907v |Anal. Chem. 2011, 83, 7924–7932