Introduction

Hydrogel-based three-dimensional (3D) microarrays are an increasingly attractive format for biological assays [1–9]. Hydrogels offer many benefits over two-dimensional (2D) surfaces used in conventional immobilization protocols for proteins. Many proteins immobilized on 2D surfaces are prone to denaturation in the interface between the solution and solid support [10–12]. Hydrogels, however, more closely resemble a solution providing an environment conducive for maintaining proteins in their native form. In addition, 3D hydrogels provide an increased capacity for the immobilization of proteins, and a stable support for ligand binding allowing access of immobilized proteins from all sides.

Many reports focus on 3D microarrays within polyacrylamide-based gels [1, 4, 5, 7, 8]. In addition, 3D hydrogels such as polyethylene glycol-based hydrogels have been reported and utilized to construct 3D microarrays for high throughput analysis of DNA and proteins [9]. Many of these hydrogels have been successfully used in immunoassays showing that they have a pore size at least large enough to pass an immunoglobulin (IgG) (estimated radius of 5–7 nm [13]). However, diffusion may still be problematic in these types of hydrogel system [14]. We describe an alternate hydrogel type with accessible carbohydrate moieties that has the distinct advantage of very large pores [2]. Several chemistries have been used to cross-link proteins within this sugar hydrogel containing an amine functional or carboxy functional group.

As the sugar polyacrylate hydrogels are known to have a large pore size, we interrogated the gels with quantum dot-based fluororeagents. Quantum dots (QDs) are inorganic nanocrystal fluorophores that offer several advantages over conventional dyes. The color of light that the QD emits is dependent on the particle size; the nanocrystal synthesis is a tightly controlled process that results in materials with narrow and symmetric emission spectra [15]. These fluorophores exhibit high photostability, and have excitation and emission properties that make them ideal for multiplexing. Since illuminating in the UV can excite all of the sizes (colors) of QD reagents [16–18], a single excitation source can be used for QDs emitting at multiple wavelengths.

In this report we examined the direct binding of staphylococcal enterotoxin B (SEB) and cholera toxin (CT) immobilized within hydrogels to biotinylated antibodies that were then detected with streptavidin-coated QDs. Confocal microscopy was performed to confirm that the toxins were immobilized within the gel and that the QD-based detection reagents were able to diffuse freely within the gel. Two types of sugar hydrogel (amine-terminated and carboxylic acid-terminated) and a commercial hydrogel were examined in these studies.

Experimental

Preparation of hydrogel thin films

Hydrogel thin films were formed on glass slides (2.5×7.5 cm) that had been cleaned and functionalized with acrylic groups 3-(trimethoxysilyl)propyl methacrylate (MTPTS) as described by Charles et al. [7]. The galactose acrylate monomer, 6-acryloyl-β-O-methyl galactopyranoside, prepared according to Martin et al. [2] was dissolved in deionized water at a concentration of 22% (w/v), along with N,N methylene bis-acrylamide cross-linker at 0.5% (w/w) of the monomer concentration and N-(3-aminopropyl)methacrylamide at 5% (w/w) of the sugar acrylate monomer concentration. The polymerization was accomplished via a free radical polymerization upon addition of the initiators sodium persulfate (1.0 mg, 3.6 μmol) and TEMED (2 μL, 17 nmol). A 150 μL droplet of the hydrogel solution was “sandwiched” between the slide treated with MTPTS and a second slide treated with dichlorodimethylsilane (DCDM); Teflon tape (thickness approximately 100 μm) was used as a spacer. Slides were placed in a nitrogen atmosphere overnight to promote polymerization. After polymerization, the DCDM-treated slide was separated from the MTPTS-treated slide revealing a thin film of highly cross-linked sugar hydrogel covalently bound to the slide surface containing an amine-terminated moiety. Carboxylic acid-functionalized hydrogels were prepared in similar fashion to the amine-terminated hydrogel except that 2-acrylamidoglycolic acid monohydrate at 5% (w/w) was used in place of the (3-aminopropyl) methacrylamide. The estimated thickness of the hydrogel films was approximately 100 μm.

Covalent cross-linking of toxins in 3D hydrogel

The amine-terminated hydrogel films were treated with the homobifunctional cross-linker, bis(sulfosuccinimidyl)suberate (BS3) dissolved in a 10 mmol L−1 Na-phosphate buffer, pH 6.0 to a final concentration of 2.5 mmol L−1. Covalent attachment of BS3 to the polymerized hydrogel was achieved through binding of the NHS-ester moiety of the cross-linker to the amino functionality localized within the internal network of the sugar hydrogel. The pendant NHS-ester moiety on the cross-linker provided the necessary functional group for covalent attachment of the SEB or CT protein. The optimum concentration of BS3 (2.5 mmol L−1) used for hydrogel modification was determined from previously reported literature [14]. Covalent attachment of proteins to the carboxy-modified hydrogels was accomplished by treatment of the gels with a 10 mmol L−1 solution of 1-ethyl-3-[3-dimethylaminopropyl]carbodiimide (EDC) under acidic conditions (10 mmol L−1 Na-phosphate, pH 6.0). The reaction between the carboxylate in the hydrogel with the amine-containing protein resulted in the formation of a stable amide bond. Commercially available 3D hydrogel films (obtained from Packard Biosciences, Meriden, CT, USA) were processed according to the manufacturer’s procedure prior to microarray printing of proteins.

Printing of toxins

Staphylococcal enterotoxin B (Toxin Technology, Sarasota, FL, USA) and CT (Calbiochem, San Diego, CA, USA) were printed on to the hydrogel films using the noncontact piezotip dispensing Packard Biochip I Microarrayer (Meriden, CT, USA). Each printed element was deposited in replicates of 7–15 with a print volume of 1.8 nL. Microarrays were printed with unlabeled SEB and unlabeled CT, each spanning concentrations from 1.56 to 200 μg mL−1. Separate microarrays of dye-labeled toxins were also printed on the hydrogels. SEB was labeled with Cy5 bisfunctional NHS ester and CT with Cy3 monofunctional NHS ester according to the manufacturer’s instructions (Amersham Biosciences, Piscataway, NJ, USA), except that 3 mg protein was labeled for each packet of dye rather than 1 mg. Concentrations of Cy5-labeled SEB spanned 0.8–100 μg mL−1 while concentrations of Cy3-labeled CT spanned 0.54–70 μg mL−1.

Three-dimensional microarrays in which Cy5- and Cy3-labeled proteins were immobilized were imaged with the ScanArray Lite microarray scanner (Packard BioChip Technologies, Billerica, MA, USA) equipped with a laser containing selectable pre-programmed settings for the detection of fluorescence emission from the Cy3 and Cy5 dyes. Microarray spot pixel resolution was set for 10 μm. The fluorescence intensity of each microarrayed element was determined using QuantArray microarray analysis software program (Packard BioChip Technologies). Standard curves to correlate fluorescence intensity with the concentration of Cy3- or Cy5-labeled proteins were constructed by scanning printed microarrays prior to blocking and washing steps.

Binding of antibody and QD reagent

Printed hydrogels were blocked with a solution of 2% bovine serum albumin (BSA) in phosphate-buffered saline (PBS), pH 7.4, by layering approximately 1 mL of the blocking solution on top of the gel, and incubating at room temperature for 90 min. After blocking, the slides containing the hydrogels were washed twice by immersion in PBS and dried at room temperature.

Rabbit anti-SEB was purchased from Toxin Technology (Sarasota, FL, USA), and rabbit anti-cholera toxin was purchased from Biogenesis (Kingston, NH, USA). Antibodies were biotinylated using EZ-link biotin-LC-NHS ester (Pierce, Rockland, IL, USA) as previously described [19]. Biotinylated antibody, diluted to 15 μg mL−1 in PBS, was layered over the appropriate array (about 75 μL/array area) and allowed to incubate at room temperature for 1–2 h. Excess liquid was removed from the slide, followed by three successive 1 min wash steps in PBS. Streptavidin-coated QDs (585 nm emission) obtained from Qdot Corporation (Hayward, CA, USA) were diluted to a final concentration of 50 nmol L−1 in the supplied QD dilution buffer, layered over each array, and incubated for 1 h at room temperature. After blotting of excess QD reagent, the slides were washed by immersing three times into PBS for 1 min, and then air dried.

Confocal microscopy

Three-dimensional reconstructions of hydrogels were constructed by collecting a series of scans using an inverted Leica TCS SP confocal microscope equipped with Leica TCS NT version 2.00 software (Leica Microsystems, Deerfield, IL, USA). Printed toxins localized in spots of approximately 150 μm diameter were visualized using a 10× objective. Scans of air-dried samples were performed in xy mode for a series of z scans (format 512×512, medium speed). A typical scan consisted of a xy area of 1 mm2 and 100 μm (2 μm per step) in the z direction. Sample was positioned with the glass slide facing the objective and scans in z direction performed from the glass to the gel.

The hydrogel itself was visualized using a scattering technique, observing back-reflected light in the first channel. Fluorophores within the gel were visualized by collecting fluorescence in the second channel. Confocal settings utilized to visualize the gel (scattering), Cy3-labeled cholera toxin (CT-Cy3), Cy5-labeled SEB (SEB-Cy5), cholera toxin detected with the QD reagent (CT-QD), and SEB detected with the QD reagent (SEB-QD), are reported in Table 1.

Table 1 Confocal setting for visualization of individual fluorophores
Full size table

Fluorescence microscopy

Images of the bound QD reagent were visualized on a Nikon eclipse E800 microscope using a filter cube with a 20 nm band-pass filter centered at 585 nm designed for the 585 nm emitting QDs (Chroma Technology, Rockingham, VT, USA). Images were recorded using a Sony DKC-5000 CCD digital camera. Both the 10× and 4× objectives were used in visualizing the arrays, for a total magnification of 100× and 40×, respectively. The total fluorescence intensities within spots were determined by analyzing the TIFF files generated by the camera with a custom function written for the data analysis program Igor Pro (Wavemetrics, Lake Oswego, OR, USA). This function totaled the pixel values for a fixed diameter circle around the center of each spot to generate the total fluorescence detected. To correct for background fluorescence, the total fluorescence from a nearby control region of equal size was subtracted from the spot value. Constraints of the system limited our analysis to four representative spots quantified for their fluorescence intensity.

Results and discussion

Hydrogel film preparation

Protein microarray technology is developing rapidly allowing researchers to simultaneously examine a large number of specific protein-binding events [20, 21]. An important aspect of the design and fabrication of protein microarrays is the ability to achieve an efficient means of protein immobilization that will maintain protein activity and lead to high sensitivity. We prepared 3D sugar hydrogel films for protein immobilization. These materials should provide similar stability for immobilized biomolecules as in solution while overcoming the diffusional problems encountered in our previous generation of polyacrylamide-based hydrogels [14]. Our cross-linking strategy involved using amine-terminated hydrogels in conjunction with the BS3 cross-linking agent and coupling to carboxy-terminated gels using EDC chemistry, both of which allowed protein immobilization through lysine groups.

Confocal microscopy

Confocal microscopy was used to examine sugar hydrogels containing immobilized Cy5- and Cy3-labeled toxins, and those containing unlabeled toxins immobilized and treated with biotinylated anti-toxin antibodies and streptavidin-coated QD reagent. The purpose of this study was to assess diffusion of various biological species in the sugar-derivatized gels. Biomolecules of different molecular weights and sizes were used in this study: SEB (28 kDa), CT (84 kDa), and anti-toxin antibodies (155 kDa). Furthermore, the size of the streptavidin-coated QD reagent is comparable with a large protein (10–15 nm [15, 22]). Although the diameter of the streptavidin-coated QDs was approximately the same length as the long axis of a standard IgG, the QD occupies a much larger volume than an IgG due to its spherical shape. It was therefore important to ensure that it was able to diffuse within the hydrogel.

Apparent hydrogel thicknesses of 70–90 μm were determined by imaging their light scattering, collecting the back-reflected light, in confocal reflection mode. Fluorescence from immobilized toxins was observed within the gel with the maximum signal occupying a depth of about 35 μm from the air–gel interface. Side views of 3D reconstructions of immobilized CT-Cy3 within an amine-functionalized gel are shown in Fig. 1A. Channel 1 (1) shows an image of the gel as visualized by scattering. Channel 2 (2) shows the fluorescence from the CT-Cy3, and an overlay of both channels is shown at the far right of the figure. On this particular image (Fig. 1A) the gel is shown in such a way that the center of the gel is transparent to facilitate the visualization of fluorescent Cy3 spots sandwiched in the gel. Data from the images in Fig. 1A were extracted and are shown graphically in panels Figs. 1B and 1C. Figure 1B shows the scattering recorded from the gel at two separate spots where CT-Cy3 was immobilized (shown in Fig. 1A), and is used to determine the thickness and uniformity of the gel. The intensity of scattering is plotted on the y axis with the distance in microns corresponding to the z scan as it moved away from the slide surface plotted on the x axis. Both curves are essentially identical, indicating the overall uniformity of the gel. The sharper transition starting near 10 μm corresponds to the glass–gel interface. The gel–air interface gives a gradual transition (50–90 μm) due to the heterogeneity of this interface. Figure 1C shows a plot of fluorescence intensity from CT-Cy3 immobilized within the same locations. Plotted are Cy3 fluorescence on the y axis and distance of the z scan on the x axis. The maximum fluorescent intensities of the spots are located at 36 and 38 μm for spots from 35 and 70 μg mL−1 applied CT-Cy3, respectively. If we assume that the thickness of the hydrogel falls between 10 and 50–80 μm, as shown in Fig. 1B, this shows that the Cy3-labeled toxin was successfully immobilized in the middle of the gel. The difference in fluorescence intensity from the two areas correlates to the difference in concentration of applied CT-Cy3 in each spot. These results are consistent with a higher concentration of CT-Cy3 in the spot corresponding to 70 μg mL−1 applied CT-Cy3.

Fig. 1
figure 1

Images of CT-Cy3 within an amine-functionalized gel. A Side views of CT-Cy3 immobilized within the gel (center of gel depicted as transparent for visual purposes). The left image shows channel 1, the gel visualized by light scattering. The middle image shows channel 2, the corresponding Cy3 fluorescence from the slice of the gel, and the right image shows an overlay of channels 1 and 2. The spots correspond to applied concentrations of 35 μg mL−1 (top) and 70 μg mL−1 (bottom) CT-Cy3. B Light-scattering signal intensity versus distance into the gel, determined from the image in A at points corresponding regions of applied CT-Cy3: 35 μg mL−1 (blue line) and 70 μg mL−1 (red line). C Fluorescence intensity versus distance profile of the gel, determined from the image in A, at points corresponding to 35 μg mL−1 (blue line) and 70 μg mL−1 (red line) applied CT-Cy3.

Full size image

Figure 2 shows a 3D visualization of an analogous experiment using streptavidin-coated QDs and biotinylated anti-CT antibodies to detect immobilized unlabeled CT within an amine-functionalized gel. This figure shows a side image of the gel as imaged by light scattering (1), the fluorescence from the QD reagent (2), and an overlay of both channels. These experiments confirm that the large QD reagents were able to get into the gel and bind to protein immobilized within the hydrogel. Control experiments showed that when nonbiotinylated antibodies were incubated with the immobilized toxin, we did not observe arrays of the streptavidin-QD reagent.

Fig. 2
figure 2

Side views of unlabeled CT within an amine-functionalized gel detected using QD reagents. The left image shows channel 1, the gel visualized by light scattering. The middle image shows channel 2, the corresponding QD fluorescence from the slice of the gel (false color to distinguish from Cy3), and the right image shows an overlay of channels 1 and 2. The spots correspond to applied concentrations of 35 μg mL−1 (top) and 70 μg mL−1 (bottom) CT.

Full size image

In parallel experiments SEB-Cy5 immobilized within an amine gel, as well as SEB-Cy5 and CT-Cy3 immobilized within carboxy-functionalized gels showed that labeled toxins were localized in the middle of the hydrogel. SEB immobilized within an amine-functionalized gel was detected with QD reagents which were able to penetrate and bind within the hydrogel.

Fluorescence microscopy

Unlabeled SEB and CT immobilized within sugar-hydrogels and detected using QD-based detection reagents were visualized on a fluorescence microscope and images captured for quantification. In addition to the two cross-linking chemistries used within the sugar-hydrogel, a commercially available gel substrate was examined with the QD reagent.

The three different hydrogels showed distinct differences in both spot size uniformity and in the amount of fluorescent background signal. Figure 3 shows representative spots for each type of gel. In the amine-functionalized hydrogels (Fig. 3A), the spot size was fairly consistent until the lowest concentrations where it decreased noticeably. The carboxy-functionalized gels (Fig. 3B) showed the most dramatic decrease in spot size with decreasing concentrations. This is evident from the significant decrease in diameter observed even at the two highest concentrations used. This could be related to diffusion within the hydrogel on spotting of the protein prior to cross-linking within the gel. A decrease in the spot diameter was also observed in images of the dye-labeled toxins immobilized within these gels. In the commercially available gel (Fig. 3C) the spot size was the most uniform from the highest to lowest toxin dilutions.

Fig. 3
figure 3

Representative spots from the three types of hydrogel. QD reagent used in the detection of CT. Toxin concentrations of 200 and 100 μg mL−1. A Amine-functionalized gel, B carboxy-functionalized gel, C commercially available gel.

Full size image

When analyzing the spot intensity, a constant diameter was used for each measurement. This approach was chosen over varying the diameter to fit the spot size for each dilution as it more closely mirrors the procedure used in commercial QuantArray microarray analysis software program used to quantify microarrays. The total fluorescence was integrated for each spot rather than the mean fluorescence. Thus this method would presumably give a more accurate representation of smaller, more highly fluorescent spots than if an average signal was used. Experiments examining the signal from fluorescently labeled toxins pre- and post-block/wash indicated that the amount of bound toxin may only be 20–30% of the applied toxins for all three types of gel examined. This immobilization efficiency is similar to results reported previously by our group for polyacylamide gels [14] and on the low end of that reported by Rubina et al. [1].

Direct binding of CT

Arrays of unlabeled CT with concentrations from 1.56 to 200 μg mL−1 (corresponding to between 2.81 and 360 pg applied protein per spot) were spotted on to each type of hydrogel. Unlabeled CT was detected after incubating first with a biotinylated anti-CT antibody, and subsequently with streptavidin derivatized QDs. Although we could visually observe fluorescent signals from Cy3-labeled CT at concentrations as low as 6.25 μg mL−1, only the four highest concentrations of the unlabeled toxin could be detected using the QD reagent. This difference may be because of how the fluorescence signals were measured (ScanArray scanner versus the microscope). It could also be a function of having a two-step detection process where first a biotinylated antibody and then the streptavidin-coated QDs need to diffuse into the hydrogel and bind to their targets.

Binding data are shown in Fig. 4. All three types of gel performed nearly identically with fluorescent signals clearly above background at spotted concentrations of >20 μg mL−1. The fluorescence signal for all three was linear with CT concentration, with an r2 of 0.9878, 0.9999, and 0.9988 for the amine-functionalized, carboxy-functionalized and commercially available hydrogels, respectively. Analysis of the covariance for these linear regions showed that there is no significant difference between the three types of gel (P>0.25). The lowest detected level of CT was seen at an applied concentration of 25 μg mL−1 (45 pg total applied CT).

Fig. 4
figure 4

Direct binding of CT visualized by QD reagent. Data represent the average of the fluorescence intensities of at least four spots and error bars represent the standard deviation.

Full size image

Direct binding of SEB

Unlabeled SEB was arrayed onto each type of gel at concentrations from 1.56 to 50 μg mL−1 (corresponding to between 2.81 and 90 pg applied protein per spot) and visualized with QD reagent. All six concentrations of SEB could be visualized through the direct binding assay. Previously we had reported the detection of 30 ng mL−1 applied SEB using a Cy5 anti-SEB antibody reagent [14]. It is possible that QD visualization would also be possible at lower applied SEB concentrations.

Figure 5 shows the results from the direct binding of SEB detected using QD reagent. In contrast to the QD-mediated detection of CT, the SEB binding curves were not linear; they rose steeply at the lowest concentrations and plateaued at the higher concentrations of applied SEB. Using the linear portion of the curves (r2>0.95) for comparison, the dose–response curves showed significant differences as determined by ANCOVA (P<0.005). The amine-functionalized hydrogel provided the most intense signal over almost the entire concentration range. A factor that contributed significantly to the performance of the carboxy-functionalized hydrogel was the background signal, which was highest in these gels. Although the absolute spot intensity was comparable in the amine- and carboxy-functionalized hydrogels, the background subtracted for the carboxy-functionalized hydrogels was one and a half times higher than for the amine-functionalized gels. Although they had low backgrounds similar to those of the amine-functionalized gels, the commercially available gels performed the worst, showing the lowest fluorescence intensity at all concentrations except the highest (50 μg mL−1).

Fig. 5
figure 5

Direct binding of SEB visualized by QD reagent. Data represent the average of the fluorescence intensities of at least four spots and error bars represent the standard deviation.

Full size image

Conclusions

Three-dimensional hydrogels as substrates for protein immobilization have the potential to provide a number of advantages over traditional 2D substrates. They create an environment that is stable, biocompatible and simulates a solution phase system, in contrast to the limited binding kinetics of the solid–liquid phase interaction evident on a 2D surface. In addition, hydrogels provide an increased capacity for the immobilization of proteins, an environment conducive for optimum ligand binding, and minimal auto-fluorescence [1, 4, 23, 24]. Proteins immobilized within hydrogels should have a free range of motion, potentially resulting in increased interactions with their target ligands.

In the present study we have demonstrated the potential of a galactoside-derivatized polyacrylate hydrogel for protein immobilization. The large pore size of this sugar hydrogel should help to overcome diffusional limitations that can be present in polyacrylamide-based hydrogels [14]. About 30% of applied protein was cross-linked within these gels. Optimizing protein cross-linking conditions may result in a larger amount of retained capture protein immobilization. This optimization could include altering parameters such as the pH of the reaction and cross-linker concentration.

Quantum dot reagents provided a detection reagent for visualizing proteins immobilized within hydrogels. This is the first demonstration that QDs can be used in systems utilizing this new matrix. This study shows that these large reagents approximately three times larger in molecular volume than IgGs can diffuse into the hydrogels and be used to detect bound target. With the proper QD reagents and filter sets these gels provide the possibility to visualize 4–6 binding events in each spotted element [18].