|Home | About | Journals | Submit | Contact Us | Français|
Optical inteference (OI) coated slides with unique optical properties were utilized in microarray analyses, demonstrating their enhanced detection sensitivity over traditional microarray substrates. The OI coating is comprised of a proprietary multilayered, dielectric, thin-film interference coating located beneath the functional coating (aminosilane or epoxysilane). It is designed to enhance the fluorescence in the Cy3 and Cy5 channel by increasing the light absorption of the dyes by about 6-fold and by redirecting emitted fluorescence into the detector during scanning, resulting in a theoretical limit of about 12-fold signal amplification. Two-color DNA microarray experiments conducted on the OI slides showed over 8-fold signal amplification, conservation of gene expression ratios, and increased signal-to-noise ratio when compared to control slides, indicating enhanced detection sensitivity. Protein microarray assays also exhibited over 8-fold signal amplification at three different target concentrations, demonstrating the versatility of the OI slides for different microarray applications. Further, the DNA and protein assays performed on the OI slides exhibited excellent detection sensitivity even at the low target amounts essential for diagnostic applications. The OI slides are compatible with commonly used protocols, printers, scanners and other microarray equipment. Therefore, the OI slides offer an attractive alternative to traditional microarray substrates, where enhanced detection sensitivity is desired.
Technological advances in the field of genomics have resulted in the sequencing of genomes from several organisms, including the human genome. The emergence of DNA and protein microarray technologies has made it possible to monitor gene expression of thousand of genes and understand their function simultaneously, in a manner not possible with traditional assays.1–3 Microarray technology is accepted as a preferred gene expression tool, being employed routinely for studies on genotyping,4 metabolic pathways,5 cancer research,6 toxicology,7 and functional genomics.8,9 This technology is now being applied for many diagnostic applications, as it offers several important advantages over the standard polymerase chain reaction– (PCR)10,11 and ELISA-based detection methods.12 The major advantage is the utilization of hundreds of DNA/antibody probes, representing the genome or proteome of several organisms for simultaneous detection and identification of organisms.
Although an intensive amount of research and commercial effort has been invested in the microarray field over the last 10 years, the technology is struggling with several limitations, due to the lack of standardization. The problem areas are related to the quality of microarray subcomponents along with inadequately standardized preparation and usage, which result in poor reproducibility and sensitivity of the microarray data, making it difficult to separate the experimentally induced variation from the true biological results.13 Attempts have been made to assess sources of variability in microarray gene expression results,14,15 and methods such as sound experimental design and replication have been suggested to minimize the variation that will allow measurement of low-expression changes with high precision.13,15
A typical DNA microarray experiment involves the labeling of 2–5 μg of reference and test mRNA with different fluorophores, followed by mixing in equal proportion and hybridization on a microarray surface. The ratio of corrected fluorescence signal (raw signal – local background) from the test and reference is used to evaluate the over- or under-expression of the genes. However, in many instances the amount of RNA is limiting and the activity of low-expressing genes such as regulatory transcription factors may not be clearly quantifiable. Similarly, the availability of protein targets is limiting in most instances. These reasons have prompted scientists to investigate methods that can increase the sensitivity of the microarray analyses. The approaches include the utilization of (i) novel substrates involving planar waveguides16 and evanescent resonator platforms;17 (ii) substrates with increased surface area;18 (iii) protocol modifications by means of reagents such as sodium borohydride;19 (iv) alternate labeling involving the use of DNA dendrimers,20 quantum dots,21 silver particles,22 and resonance light scattering of gold particles;23 and (v) signal amplification through the use of tyramide precipitation24 and rolling circle amplification.25 Although these methods show varying improvements in sensitivity over traditional microarray technologies, the high cost and/or use of cumbersome/ non-routine procedures or detection technologies limit their widespread usage.
Alternatives such as optical interference (OI) coating technology are required to increase the detection sensitivity in microarray analyses. The OI coating was developed (patent pending) based on the expertise acquired while developing traditional coatings used for optical applications as well as the comprehensive information available in the literature on dielectric and interference phenomena.26–30 The OI coating is comprised of a multilayered, dielectric, thin-film interference coating on a glass surface, and is intended to increase microarray sensitivity by maximizing photoabsorption of dye molecules in the vicinity of the surface and by redirecting the portion of the fluorescence that would normally be lost through the substrate towards the detector within a scanner via reflection. Briefly, the design of the thin-film dielectric coating is based on the phenomenon of optical interference, which can be either constructive or destructive. By design, the interference is made constructive to enhance the electric field strength (E) at the surface or in the vicinity (within about 60 nm of the substrate) of the microarray surface, and also to reflect and redirect the microarray fluorescence signals toward the detector during scanning.
In this paper, the mechanism of signal amplification of the OI coated slides is explained, their utilization for DNA and protein microarray applications is demonstrated, and data are presented showing the enhancement of signal generation from the hybridized spots, leading to increased detection sensitivity. Such increased sensitivity is not only critical for quantifying expression of low-expressor genes/proteins, but also for diagnostic applications where cell numbers or target proteins are limiting in many samples. In addition, this work illustrates compatibility of the OI coated slides with standard microarray protocols and detection systems.
The OI coating was designed with the aid of commercially available software (TFCalc 3.5, 2002, Software Spectra, Inc., Portland, OR), with the intent of (i) maximizing photoabsorption or the E of dye molecules in the vicinity of the surface (~60 nm) at excitation wavelengths of 535 and 635 nm and (ii) redirecting the maximum intensity of fluorescence signal towards the detector in the 500–720 nm spectral range via reflection to extract maximum fluorescence signals from routinely used Cy3- and Cy5-labeled targets. Borosilicate glass was coated with alternate high- and low-refractive-index coatings to form a multilayer stack using a proprietary deposition process. The increase in E2 was calculated at the surface of the OI coating using TFCalc software. The percent spectral transmission (T) of the OI coating was measured from 400 to 800 nm on a Hitachi Spectrophotometer (Model U-2000, Tokyo, Japan) and was converted to reflection (R) data using the equation R = 100 – T.
The OI coated slides and control slides (without OI coating) were cleaned using a specially developed washing protocol, followed by subsequent coating with aminosilane or epoxysilane for DNA or protein microarray analyses, respectively.
DNA microarrays intended for expression analysis were prepared by printing 50-mer, amine-modified rat oligos (MWG-Biotech, Inc., High Point, NC) at 20 μM concentration in 25% DMSO in a 5 × 5 subarray format, two subarrays per slide on three each of the OI and control slides coated with aminosilane chemistry. The standard Schott Nexterion protocol for Slide A was used to process the DNA microarrays.31 One microgram of rat liver and kidney mRNA (Stratagene, La Jolla, CA) was converted into cDNAs labeled with Cy5 and Cy3 respectively, using aminoallyl chemistry (Amersham Biosciences/GE Healthcare, Piscataway, NJ). DNA micro-arrays were hybridized at 42°C for 16 h under lifter cover slips (Erie Scientific Company, Portsmouth, NH) with 30 pmol of each label. After washing, the microarrays were scanned using an Axon 4000-B laser scanner (Axon Instruments/Molecular Devices, Corp., Union City, CA) at a photomultiplier tube (PMT) gain of 600 and 750 V for Cy3 and Cy5, respectively, and the data were analyzed using GenePix Pro 4.0 software (Axon Instruments/Molecular Devices, Corp.). Data were not filtered or normalized to any standard spot. Average corrected median signals (median signal – local background) from all microarrays were computed for two each randomly selected red, green, and yellow spots representing six genes. These average corrected median signals were then normalized to the corresponding values from the control DNA microarrays. In addition, Cy5 to Cy3 ratios for selected genes were calculated and normalized to ratios derived from the control DNA microarrays. Signal-to-noise ratio (SNR) was calculated by subtracting the local background from the median signal and dividing the resultant number by the standard deviation of background. When calculating SNR, signal of individual spots was corrected by subtracting the average signal from the ARA and ACT1 spots, which represented non-specific hybridization signal.
The DNA microarrays intended for diagnostic analysis were printed with two Brucella species-specific 60-mer probes on the OI and control slides, along with several nonspecific DNA control probes. A Bioprime Array CGH Genomic Labeling System kit (Invitrogen Corporation, Carlsbad, CA) was utilized to label the genomic DNA from Brucella melitensis. Briefly, the DNA was digested with DPNI and labeled directly with Cy3-dCTP (Amersham Biosciences/GE Healthcare) using random primers. Various target amounts (1, 2.5, 5, and 25 pmol/slide) were used for overnight hybridization, and the DNA microarrays were processed as described above. Finally, the DNA microarrays were washed, scanned at a PMT gain of 600 V in the Cy3 channel, and the background-corrected median signal data on one Brucella probe was analyzed at different target concentrations.
The protein microarrays were prepared by printing polyclonal Anti-human IgG (Sigma Aldrich, St. Louis, MO) in PBS buffer at 200 μg/mL concentration on the OI and control slides using a Biochip arrayer (PerkinElmer, Wellesley, MA). After printing, the proteins were immobilized for 1 h at ~70% relative humidity, followed by blocking with 300 mM phosphate buffer solution (pH 8) containing 1 mg/mL bovine serum albumin (BSA) (Sigma Aldrich). The microarrays were incubated with human IgG (Sigma Aldrich) diluted in human serum (Pierce, Rockford, IL) as a target under cover slips for 60 min. A monoclonal anti-human IgG (Sigma Aldrich) labeled with Cy5 (Amersham Biosciences/GE Health-care) was added on the microarray at 1 μg/mL concentration and incubated for 60 min to detect the captured target for this sandwich assay. The slides were washed for 10 min in phosphate buffered saline (PBS) with 0.5% Tween-20 three times, followed by a single wash in PBS and scanned using an Axon laser scanner at a PMT gain of 600 V. The Cy5 dye was exclusively used for the protein microarrays because a number of proteins fluoresce under the Cy3 excitation and emission conditions, complicating detection of protein probe-target interaction from intrinsic fluorescence of printed proteins.
The OI and control slides coated with aminosilane and epoxysilane chemistries prior to printing and hybridization were scanned using an Axon laser scanner at a PMT gain of 600 and 750 V in Cy3 and Cy5 channels, respectively, to determine the contribution of the OI coating to the background fluorescence. A 10 × 10 grid (100 μM spots, 150 μM pitch) was placed on the scanned image, and the background data were analyzed using GenePix Pro 4.0 software.
The physical location of the multilayered OI coating on the glass surface is schematically represented in Figure1AFigure1A.. Also included in Figure 1A1A is a plot demonstrating the value of E2 predicted in the vicinity of the OI coating during scanning. Calculations based on the TFCalc software predicted theoretical enhancement in the E2 at the surface of the OI slides by about 6.35-fold (4 for the OI slides vs. 0.63 for the control slides) (Figure 1B1B).). In addition, the OI coating was designed to redirect the fluorescence signal into the detector of a scanner via reflection of light from 500 to 720 nm. The theoretical and measured reflection curves depicted in Figure 1C1C showed excellent correlation in designed and observed values for reflection for the OI coated slides.
DNA and protein microarray experiments were performed to quantify the enhanced sensitivity of microarrays prepared using the OI coated slides. Scanned images of two-color DNA microarray expression analysis for the control and the OI slides are displayed in Figure 2A2A.. A total of 13 spots (see circled spots, Figure 2A2A)) were detected on the control slides, whereas 2 additional spots (a3 and b4) were detected on the OI slides after correcting for the non-specific signal. All spots on the OI slides exhibited signal amplification. Approximately eightfold signal amplification was observed on the OI slides after comparing the normalized Cy3 and Cy5 signals from 6 randomly selected spots to the corresponding data from the control slides (Figure 2B2B).). Figure 2C2C displays the normalized ratios of selected spots on both slide types, demonstrating the conservation of ratios after amplification of signal on the OI slides. These results prove similar levels of signal amplification in the Cy3 and Cy5 channel. Based on the SNR data, at least four low-expressor genes (Table 11 and Figure 2A2A,, spots a3, b4, d1, and c5) were positively detected on the OI slides as they exhibited ≥3 SNR value. Further, the protein microarray results (Figure 2D2D)) also displayed signal amplification of over eightfold in the Cy5 channel by using a sandwich detection assay. Since similar signal amplification was observed on the aminosilane- and epoxysilane-coated OI slides for DNA and protein micro-arrays, respectively, the signal enhancement characteristic of the OI coating is independent of the organic functionalities used for bimolecular attachment.
In diagnostic arrays, detection and identification is solely based on the presence of detectable signal at a given spot after correcting for the nonspecific signal. Figure 3A3A demonstrates overall detection superiority of the OI slides over the control slides for DNA-based diagnostic arrays, with sensitivity (ability to distinguish true hybridized signal from the background) exceeding 1 pmol of the target (which corresponds to 6 ng of the labeled DNA based on experimental data). Likewise, the protein sandwich assays (Figure 3B3B)) also displayed over eightfold signal enhancement over 0.1 to 10 μg/mL of target concentrations. Therefore, the OI slides are suitable substrates for DNA- as well as protein-based diagnostic applications, as the slides provide increased signal intensity at all target concentrations. In addition, the OI slides provide a detection limit that is ≥10X higher than that achievable using traditional microarray slides (see Figure 3A3A,, values for 2.5 and 25 pmol, and Figure 3B3B).
The normalized backgrounds of the OI and control slides coated with aminosilane and epoxysilane prior to printing and hybridization are displayed in Figure 44.. The background of the OI slides prior to hybridization is about 3.5-fold higher than the control slides, regardless of the functional coating. Further, Table 11 shows that the background measured after hybridization is increased by three- to sixfold on the OI slides (includes intrinsic background and the one due to nonspecific target binding). However, the effect of increased background on the OI slides has a negligible impact on the overall detection sensitivity, because even after correcting for the hybridized background, eightfold amplification is observed in the signal (Table 11,, column 5, and Figure 2B2B).
One goal of this paper is to explain the signal amplification mechanism of the OI slides and to demonstrate its utilization for DNA and protein microarray applications with increased sensitivity. Increased sensitivity will not only allow detection of low-expressor genes/proteins with a high level of confidence, but can be critical in diagnostic application(s), where targets present in samples are often limiting.
The OI coating is comprised of a proprietary multilayered, dielectric, thin-film interference coating (Figure 1A1A),), which is located beneath a chemically functional coating, such as aminosilane or epoxysilane. Each layer is optically non-absorbing and, at each interface (air-film, film-film, film-substrate), the incident light, i.e., the excitation source in a laser scanner, is split into reflection and transmission. The splitting ratio is determined by the refractive indices of the adjacent materials at each interface. In OI coatings, this process is repeated at each interface, resulting in an infinite number of internal reflections and transmissions. To understand the outcome of this multitude of light rays inside the OI coating, light has to be considered as an electromagnetic wave with an oscillating electric and magnetic field.32 Both fields oscillate perpendicular to each other and with respect to the direction of propagation. Since the wave is a periodic structure, the thickness of each film is designed to be on the length scale of the wavelengths intended for modulation. As the light penetrates into the OI coating, the reflected and transmitted light rays overlap and the local electromagnetic fields undergo interference phenomena. Depending on the mutual orientation and phase of the local fields at the point of interference between the dielectric layers, the resulting field may be larger than the individual fields (constructive interference) or smaller (destructive interference). In the case of the current OI coating, the interference was designed to be constructive to enhance E2 locally at the surface or in the vicinity of the microarray surface (Figure 11).
The signal generated as a result of hybridization in the vicinity (within ~60 nm) of the surface of the OI slides is enhanced through both coating-enhanced photo-absorption of dye molecules and redirection (reflection) of the fluorescence toward the detector within the scanner. The fluorescence associated with the labeled target occurs in two steps. In the first step, the excitation energy is absorbed at a specific wavelength, while in the second step, energy is emitted or released at another wavelength. The conversion of the absorbed energy into fluorescence is a dye-specific property. However, the OI coating influences (increases) the amount of energy available for absorption. By constructive interference, the electric fields of the incident and reflected light add up and form an E maximum in the vicinity of the outermost surface. The electric field is about twice as large as the incident field in the vicinity of the OI coated surface. Further, absorption varies with the square of E, so the absorption in the vicinity of the OI coated surface is enhanced by four times, i.e., (2E)2 = 4E2 compared to free space (E). On a traditional glass slide (i.e., the control slide), the fluorescing material is positioned on a standard glass surface; due to the out-of-phase reflection of the incident light at the air-glass interface, a weak destructive interference occurs, and the electric field at the glass surface is slightly reduced (0.79E) as compared to the free space electric field. Therefore, the squared electric field is about (0.79E)2 = 0.63E2 on a control slide. The photoabsorption of a dye molecule on the OI-slides is ideally increased by the ratio of the squares of the electric fields, i.e., by a factor of 4E2/0.63E2 = ~6.3, when compared to that on a control slide. Furthermore, in the case of a control glass slide, the fluorescence emitted towards the glass surface after the labeled target material absorbs the incident light is mainly lost due to transmission through the ~96% transparent glass. However, with the OI slides, the fluorescence is almost completely reflected and redirected towards the detector in the scanner. This could lead to about a twofold additional increase in the optical performance of the OI coating as compared to the control slide. The combined enhancement due to both increased dye photoabsorption and maximized redirection of fluorescence (reflection) into the detector results in a theoretical limit of 6.3 × 2 = ~12.6 for signal amplification, and in the OI slides this beneficial effect has been engineered such that it occurs at the excitation and emission wavelengths of the Cy3 and Cy5 dyes typically used for microarray analysis.
Both DNA and protein microarrays prepared on the OI slides have shown consistent amplification of signal and higher detection sensitivity. The experimentally observed signal amplification of about eightfold (Figure 2B and 2D2D)) is substantial for microarray applications, as it provides users with nearly an order of magnitude increase in sensitivity. However, this observed amplification is ~1.5X lower than the theoretically achievable limit of ~12.6-fold signal amplification. There could be several reasons for this minute deviation: (1) the increased E2 may not have a proportional effect on increasing fluorescence due to dye saturation; (2) heating (materials fluoresce less at higher temperatures); (3) and dye photo-degradation (operating at 4E2 is equivalent to exposing the dyes at 6X higher laser intensity). The overall influence of increased background fluorescence, originating from the functional and/or OI coating (Figure 44)) and the amplification of nonspecifically bound target resulting in increased hybridized background (Table 11)) is negligible, because the corrected signal is still enhanced by about eightfold (after correcting for total background) using a complex hybridization assay.
For DNA microarray analyses, Cy5/Cy3 expression ratios are vital for scoring the spots as low or high expressors. It should be noted that the expression ratios obtained on the control glass slides are retained when using the OI coated slides (Figure 2C2C);); this is important for maintaining the validity of the microarray data. SNR is another quantitative measure to detect true signal from the background, and an SNR value of 3 is commonly considered as the lower limit of detection. Increasing the signal or reducing the background can increase the SNR, and hence significantly higher SNR is expected on the OI slides as consistently higher corrected signal (at least eightfold) is observed. The standard deviation of the background pixels is controlled by the hybridization efficiency at a particular spot and is independent of the substrate type; hence it is not likely to affect the SNR. The results (Table 11)) clearly show that at least four spots (a3, b4, d1, and c5) can be detected with a high level of confidence (SNR ≥ 3), further verifying the increased detection sensitivity of the OI slides. However, the utilization of real-time PCR to confirm expression of selected genes is still recommended. The increased sensitivity provides the opportunity to investigate target concentrations that are ~10X lower than those observable with traditional slides (Figure 33).). Therefore, the OI slides can enable the detection of low-expressor genes or proteins with a high level of confidence over a wide range of microarraying applications.
The high sensitivity offered by the OI slides makes them suitable for microarraybased DNA and protein diagnostic applications. The data in Figure 3A and 3B3B clearly show the high level of detection sensitivity when a series of decreasing target amounts were used for hybridization. The detectable sensitivity of <1 pmol target for DNA diagnostic application corresponds to <6 ng of labeled DNA, or DNA isolated from <3 million cells, assuming that 2 fg of DNA is extractable from 1 bacterial cell. This sensitivity could be further augmented if the OI slides are used in conjunction with whole-genome amplification33 or other signal amplification mechanisms, such as tyramide and rolling signal amplification,24,25 to reach the sensitivity achieved with PCR-based assays, which are the gold standard.
The spot morphologies achieved on the control and the OI slides (Figure 2A2A)) are identical because the functional coating (i.e., aminosilane) used to immobilize the microarray probes is the same on both slide types. Also, the signal amplification is not dependent on the functional coating, as evidenced by the fact that different functional coatings were used to obtain the same ~8X increase in sensitivity (Figure 2 B, DD)) using both DNA and protein microarrays. Therefore, the OI coated slides can be prepared with a multitude of different functional coatings for biomolecule attachment, such as aminosilane for long oligos and cDNAs, epoxysilane for modified oligos and proteins, aldehyde for peptides, or hydrogel for protein/antibody attachment. It should be noted that for achieving high signal enhancements, the functional coating should be as thin as possible in order for this effect to be realized using the current OI design. However, for functional coatings thicker than 60 nm, a new OI design could easily be optimized for a specific application. In addition, the multilayer design of the OI coating could be easily optimized for fluorescent dyes of any other wavelength, or even more than two wavelengths. Such design freedom is considered a major advantage of the OI coating as compared to simple, metallic-based reflectors. Further, the OI coated slides can be used in place of traditional slides without any modification of protocol. The OI slides have been tested in their current form using contact and noncontact microarray printers as well as laser and CCD-based scanners, and hence are compatible with routinely used microarray equipment. The only caveat is that the signal on the OI slides will not be detected if slides are scanned through the glass as recommended by some scanner manufacturers (due to the fact that they are highly reflective). Thus, researchers are now in a position to incorporate the OI slides into their experiments to address important functional genomics questions where increased sensitivity will be a decisive factor.
Technical assistance from Gary Bolus, Michael Wotring, and JoAnn Stankowski in conducting microarray experiments and critical comments from Dr. Katrin Steinmetzer are greatly appreciated. Part of this study was supported by the Office of Naval Research (Contract # N0001403C0149). The authors have no competing financial, professional, or personal interests in this work.