Lateral flow microarrays: a novel platform for rapid nucleic acid detection based on miniaturized lateral flow chromatography

Abstract

Widely used nucleic acid assays are poorly suited for field deployment where access to laboratory instrumentation is limited or unavailable. The need for field deployable nucleic acid detection demands inexpensive, facile systems without sacrificing information capacity or sensitivity. Here we describe a novel microarray platform capable of rapid, sensitive nucleic acid detection without specialized instrumentation. The approach is based on a miniaturized lateral flow device that makes use of hybridization-mediated target capture. The miniaturization of lateral flow nucleic acid detection provides multiple advantages over traditional lateral flow devices. Ten-microliter sample volumes reduce reagent consumption and yield analyte detection times, excluding sample preparation and amplification, of <120 s while providing sub-femtomole sensitivity. Moreover, the use of microarray technology increases the potential information capacity of lateral flow. Coupled with a hybridization-based detection scheme, the lateral flow microarray (LFM) enables sequence-specific detection, opening the door to highly multiplexed implementations for broad-range assays well suited for point-of-care and other field applications. The LFM system is demonstrated using an isothermal amplification strategy for detection of Bacillus anthracis, the etiologic agent of anthrax. RNA from as few as two B. anthracis cells was detected without thermocycling hardware or fluorescence detection systems.

Figure 1.

(A) NASBA primer-binding sites are shown in the relevant region of the predicted B. anthracis plcR mRNA sequence based on GenBank accession number AY265698. The terminal 3′ base of plc-P1 is complementary to the U of the ochre stop codon, indicated with an arrowhead, diagnostic for B. anthracis. (B) The predicted nucleotide sequence plcR mRNA in the region represented by synthetic target dnaR89. The binding sites of detection probe R-57-76-3TN, as well as capture probes R-77-96, R36-55 and R-24-43 are indicated.

Figure 2.

(A) A compact plastic housing was designed to a carry conjugate release pad and a LFM membrane. A small port is used to introduce the 10-μl sample volume and a rectangular window allows direct visualization of the microarray capture features. The device is 39 × 5 mm. (B) A schematic representation of the hybridization sandwich assay used for LFM-based nucleic acid detection. Carboxyl-polystyrene dyed microspheres are linked to amine-modified detection oligonucleotide R-57-76-3TN. The microsphere/analyte complex forms by hybridization as sample solution liberates dried microspheres from the conjugate release pad. This complex is captured from solution by hybridization to immobilized capture probes as capillary flow transports the sample/bead solution through the large-pore nitrocellulose matrix. The resulting increase in local microsphere concentration, at capture features complementary to the target analyte, rapidly produces a colorimetric signal visible to the naked eye and easily detected at low concentrations using widely available flatbed scanners. The hybridization based nature of the assay render it well suited for multiplexed detection.

Figure 3.

(A) LFM substrates patterned with different concentrations of capture oligonucleotides R-77-96, R-36-55, and R-24-43 were used to detect dnaR89 with R-57-76-3TN microspheres. Signals generated at microarray capture features printed at 200, 400 and 800 μM were quantified following lateral flow of samples containing 5, 10 and 20 fmol dnaR89. Signals were normalized for each capture probe and target concentration. Average signal intensities were calculated and presented in this bar graph. 400 μM printing concentrations consistently provided the strongest signal independent of capture sequence or dnaR89 concentration. Error bars are the 95% confidence interval (two tailed, n = 12). (B) Scatter plot of normalized signal intensity versus SSC concentration. LFM running buffer was optimized for SSC concentration using R-57-76-3TN to detect dnaR89 (circles) or plcRivt (squares). (C) Line plot of normalized signal intensity versus formamide concentration. Formamide concentrations between 0 and 20% in LFM running buffer based on 4× SSC were evaluated for dnaR89 (circles) and plcRivt (squares). Five percent formamide provided near optimal detection of both dnaR89 and plcRivt. (D) Line plot of normalized signal intensity versus the R-57-76-3TN to microsphere ratio. Oligonucleotides/bead of 2.2 × 10⁴ in coupling reactions provided the best performing conjugated microsphere populations as judged by hybridization sandwich assay signal intensity. For parts B–D error bars are the 95% confidence interval (two tailed, n = 4).

Figure 4.

Representative LFMs are shown following detection of the indicated amounts of dnaR89. The microarray physical layout is provided in the color legend. The panel labeled ‘Ponceau S’ is an LFM prior to sample addition. Ponceau S allows visualization of successful oligonucleotide deposition but migrates away from the capture zone during sample transport across the substrate. Contrast was adjusted using the Auto Contrast function in Photoshop CS2 to increase reproduction contrast. Auto Contrast adjustment was not used for images subjected to quantification. The bar is 600 µm for all LFM panels.

Figure 5.

(A) The relative performance of three different capture oligonucleotides (R-77-96, circle/solid line; R-36-55, square/solid line; R-24-43, diamond/dashed line) was determined using varying amounts of dnaR89 from 0 to 200 fmol. The capture probe R-77-96 provides significantly more sensitive detection than the other capture sequences evaluated using R-57-76-3TN coupled microspheres. (B) R-77-96 signal intensity versus amol dnaR89 from 0 to 2500 amol is plotted with a linear regression line (R² = 0.989). (C) R-24-43 signal intensity versus fmol dnaR89 from 2.5 to 100 fmol plotted with a linear regression line (R² = 0.968). For all parts error bars are the 95% confidence interval (one tailed, n = 6).

Figure 6.

Time course of LFM detection: 10-μl samples containing either 1000 fmol (circle), 100 fmol (square) or 10 fmol (diamond) dnaR89 were run on appropriately patterned LFMs. Video data were collected and colorimetric signal intensity measured from video frames at R-77-96 capture features. Capillary transport of the 10-μl sample was complete by 120 s. Lines represent logarithmic curve fits to the data.

Figure 7.

(A) Indicated amounts of total cellular RNA from B. anthracis Sterne strain 7702 or, as a negative control, 2 ng B. thuringiensis strain HD 621 RNA (0 fg panel) were introduced to 1 μg of total human cellular RNA isolated from HeLa S3 cells. RNA mixtures were subjected to NASBA amplification for 60 min after which 2 μl aliquots of the NASBA reactions were mixed with 8 μl of LFM running buffer and introduced to LFMs. Enlarged LFM sub-regions are shown following cropping, grayscale conversion and Auto Contrast adjustment in Photoshop. The legend indicates microarray element identities: (+) dnaR89 as a positive hybridization control, (–) R-57-76-3N as negative hybridization control, (24–43) capture probe R-24-43, (36–55) capture probe R-36–55, (77–96) capture probe R-77-96. (B) Graph of quantified signals from B. anthracis and B. thuringiensis challenged LFMs with linear regression line (R² = 0.970). 0 fg B. anthracis total cellular RNA data point contains 2 ng B. thuringiensis total cellular RNA in addition to 1 μg human total cellular RNA. Error bars depict measurement SD (three determinations).