Signal amplification through nucleotide extension and excision on a dendritic DNA platform

Abstract

Techniques that provide strong signal amplification are useful in diagnostic applications, especially in detecting low concentrations of non-amplifiable target molecules. A versatile and strong signal amplification method based on activities of a DNA polymerase to generate high concentrations of pyrophosphate (PPi) is described. The generation of PPi is catalyzed by nucleotide extension and excision activities of a DNA polymerase on an oligonucleotide cassette. The signal is generated upon enzymatic conversion of PPi to ATP and ATP levels subsequently detected with firefly luciferase. Bioluminesence produced by an oligonucleotide cassette consisting of just two polymerase reaction sites is sufficient to detect them at low attomole levels. The attachment of a large number of these oligonucleotide cassettes to DNA dendrimers enabled the detection of such polyvalent substrate molecules at low zeptomole (10^–21 mol) concentrations. The extent of signal amplification obtained with dendrimer substrates is comparable to exponential target amplifications provided by nucleic acid amplification methods. The attachment of such PPi-generating dendritic DNA platforms to ligands that mediate target recognition would potentially permit detection of extremely low concentrations of analytes in diagnostic assays.

INTRODUCTION

Molecular detection and quantification play an important role in basic research and have become a necessity in clinical practice. In clinical applications, identification of molecules that are associated with pathological conditions in the form of either direct causative agents or surrogate markers is useful for diagnosis of a disease and monitoring its progression. The invention of new drug therapies that are efficacious for certain viral infections such as human immunodeficiency virus (HIV) has increased the demand for measuring low levels of viral loads. Detection of extremely low viral loads became feasible with the development of nucleic acid amplification technologies. Due to their ability to be amplified using different enzymatic approaches (1–5), nucleic acid molecules have been detected at extremely low concentrations, as low as a single molecule (6). Nucleic acid sequences at sparingly low concentrations are pre-amplified to enrich their concentrations, allowing them to be detected by less sensitive methods. The nature of strong target amplification methods, like PCR, increases the chances of sample contamination through amplicon carryover, leading to false positive results. As a result, techniques to circumvent carryover contamination in PCR have been developed (7,8). In spite of such efforts, other approaches that do not amplify target nucleic acid sequences but rely on strong signal amplification for detection have also been developed (9,10). Branched DNA (bDNA) represents a signal amplification strategy that has been developed as an alternative method to PCR. Amplification of the signal in the bDNA approach is based on the attachment of hundreds of alkaline phosphatase molecules to a tree-like or comb-like DNA dendrimer structure. Several such trees are linked to a single molecular recognition event to harness the catalytic power of multiple alkaline phosphatase molecules.

In contrast to the assays based on target amplification methods, the frequency of occurrence of false positives is low among assays that rely on signal amplification. In fact, strong signal amplification methods are critical for detecting low concentrations of molecules other than nucleic acids that are not amenable to pre-amplification before detection. Hence, detection of targets such as proteins, lipids, carbohydrates and small molecules at low concentrations requires extremely sensitive methods for their detection. As a result, development of novel approaches for detecting extremely low concentrations of analytes would benefit many applications.

Most diagnostic assays are based on a molecular recognition event coupled to a signal generation event. Molecular recognition is mediated by a ligand such as an antibody, an aptamer (11–13) or any other molecule that can bind the target of interest with high affinity and specificity. In certain assay formats, the signal can be a measurable physical property of the target or the ligand that changes upon association (e.g. fluorescence polarization; 14). In most other assay formats, the signal generation event is provided by a secondary ligand that carries a suitable reporter molecule. The secondary ligand binds to the target already bound to the primary ligand, generating a sandwich.

One of the ways to achieve signal amplification is through the use of the catalytic power of enzymes (15–17). It has become a common practice to conjugate enzymes that catalyze signal generation to the secondary ligand to boost the sensitivity of detection. The overall sensitivity enhancement is usually dependent upon the type of signal and the number of enzyme molecules conjugated to the ligand. The number of enzyme molecules that can be conjugated to a ligand (e.g. to an antibody) is generally limited. Although enzymes coupled to antibodies are the most versatile and popular reagents used in diagnostic assays, enzyme conjugation has its own drawbacks. Some enzymes are difficult to conjugate without compromising activity. Certain other enzymes lose their activity soon after conjugation. Furthermore, challenges also exist in coupling and purification of enzyme–ligand conjugates. Hence, approaches that could eliminate enzyme conjugation would be attractive.

Here we describe a general approach for signal amplification that does not require conjugation of an enzyme to the ligand that provides molecular recognition in diagnostic assays. In this approach, instead of direct conjugation of the enzyme to a ligand, the enzyme is recruited to a ligand that has a substrate, or platform, to carry out catalysis. Signal amplification is facilitated by designing an oligonucleotide cassette that serves as a template for the DNA polymerase isolated from phage T7 (T7 DNAP) to perform catalysis to generate a vast quantity of inorganic pyrophosphate (PPi), the primary reporter molecule of the detection scheme. The PPi generated is subsequently converted to ATP and the resulting ATP concentration is detected and quantified with firefly luciferase as described previously (18,19). The method described here allows the detection of oligonucleotide cassette molecules with two polymerase reaction sites at low attomole levels. Attachment of oligonucleotide cassettes onto a DNA-based polymeric support called a 3D DNA dendrimer (20,21) further enhanced the sensitivity. These modified dendrimers represent molecular hubs consisting of thousands of reactive centers that accommodate a DNA polymerase to generate PPi through catalytic amplification. Using this approach, we have been able to detect as few as 5 zmol of such dendritic molecules. The potential attachment of DNA dendritic molecules carrying a large number of oligonucleotide cassettes to antibodies and other ligands that mediate molecular recognition could be useful in detecting target molecules at extremely low concentrations in diagnostic assays.

MATERIALS AND METHODS

Materials

ATP Bioluminescence Assay Kit CLS II and alkaline phosphatase were bought from Boehringer Mannheim (Indianapolis, IN). ATP sulfurylase, adenosine 5′-phosphosulfate (APS) and 4,5′,8-trimethylpsoralen (Trioxalen) were purchased from Sigma Chemical Co., (St Louis, MO). T7 DNA polymerase was from New England Biolabs (Beverly, MA). All DNA oligonucleotides synthesized using standard cyanoethyl phosphoramidite chemistry and purified by high pressure liquid chromatography were purchased from Operon Technologies Inc. (Alameda, CA). The 4-, 6- and 8-layer DNA dendrimers and their derivatives were purchased from Genisphere Inc. (Philadelphia, PA).

Signal amplification cassettes (SACs)

Equimolar concentrations of the two oligonucleotide strands with complementary regions that form the dual-track SAC were heated to 80°C in a buffer consisting of 10 mM Tris–HCl, pH 8.0, 50 mM KCl, 2.5 mM MgCl₂ for 3 min and slowly cooled to room temperature to help annealing. A single-stranded oligonucleotide with the ability to form a hairpin structure to generate the single-track SAC was treated in the same way. After annealing oligonucleotide strands in SAC constructs, they were photo-crosslinked with Trioxalen. Briefly, a saturated solution of Trioxalen in ethanol was added to 100 µg/ml DNA solution at 1:15 (v/v) (Trioxalen:DNA) and incubated for 30 min at 37°C. Photo-crosslinking was carried out in a Stratalinker UV Crosslinker 2400 equipped with a 365 nm light source for 45 min at 1 mJ/min on ice. These conditions, yielding 70–80% photo-crosslinked products, were chosen by monitoring the extent of crosslinking of radiolabeled oligonucleotides on denaturing polyacrylamide gels. After photo-crosslinking, unreacted Trioxalen was removed by chloroform extraction and SAC constructs were recovered upon ethanol precipitation.

DNA dendrimers

The 4-layer (mol. wt 1.2 × 10⁷ g/mol), 6-layer (mol. wt 1.2 × 10⁸ g/mol) and 8-layer (mol. wt 1.2 × 10⁹ g/mol) DNA dendrimers were prepared at Genisphere Inc. as previously described (20,21). Briefly, DNA dendrimers were assembled from seven oligonucleotides, strands 1–7. The seven strands were pairwise hybridized in TNE200 buffer (50 mM Tris, pH 8.0, 10 mM EDTA, 200 mM NaCl) to form five building block ‘monomers’: A, B′, B′′, C′ and C′′. Strand 1 and strand 2 were hybridized to form the A monomer. Similarly, strand 3 was hybridized to strand 4 and strand 4 was hybridized to strand 5 to produce the B′ and B′′ monomers, respectively. The C′ monomer was prepared by hybridizing strand 2 and strand 6 and the C′′ monomer was prepared by hybridizing strand 2 and strand 7. Each monomer is composed of four 31 nt single-stranded arms and one central 50 base waist. From these building blocks the core dendritic structure was assembled. At each layer of assembly the dendritic structure was photo-crosslinked with Trioxalen. The SACs were introduced by hybridization followed by photo-crosslinking of an oligonucleotide that is complementary to the 3′-arms of the DNA dendrimers. This oligonucleotide provided a 30-nt long adenosine template for the T7 DNA polymerase. After completion of all assembly, DNA dendrimers were finally purified using ultracentrifugation on 10–50% sucrose gradients containing 50% deionized formamide run at 40°C. Gradient fractions containing DNA dendrimers were pooled, ethanol precipitated and resuspended in a buffer consisting of 50 mM Tris–HCl, pH 8.0, and 10 mM EDTA.

The nucleotide extension and excision-coupled signal amplification (NEESA) reaction

Signal amplification reactions were carried out in 96-well U-bottom microtiter plates (solid white) from Corning Costar (Cambridge, MA) in duplicate. SACs in various configurations (monomeric, dimeric and multimeric on DNA dendrimers) were incubated at 37°C for 1 h in 20 µl of reaction buffer consisting of 20 mM Tris–acetate, pH 7.9, 10 mM Mg(OAc)₂, 50 mM KOAc, 1 mM dithiothreitol, 250 mM TTP, 2.5 µg bovine serum albumin, 1 U T7 DNAP, 50 pmol APS and 125 mU ATP sulfurylase. The amount of ATP generated in each reaction was measured with ATP Bioluminescence Assay Kit CLS II from Boehringer Mannheim according to the manufacturer’s instructions. Briefly, 45 µl of luciferase reagent was added to each well and incubated for 10 min in the dark. The intensity of bioluminescence was measured using an EGG Berthold LB 96P luminometer. The average value of bioluminescence obtained from duplicate reactions was used for further analysis. In decoupled assays, PPi was generated in the same buffer, but in the absence of APS and ATP sulfurylase, for 1 h at 37°C. Subsequently, APS and ATP sulfurylase were added to the reaction and the incubation was carried out for an additional 30 min to convert PPi to ATP.

Alkaline phosphatase catalyzed signal amplification

Varying concentrations of alkaline phosphatase (mol. wt 140 000 g/mol) in 10 µl of 100 mM diethanolamine buffer, pH 10, was added to a 100 µl solution consisting of 100 mM diethanolamine, pH 10, 10% (v/v) Sapphire enhancer solution (Tropix, Bedford, MA), 17 µl/ml CSPD (a 1,2-dioxetane substrate from Tropix), 1 mM MgCl₂ and 0.02% sodium azide. After 20 min incubation in the dark, the intensity of the chemiluminescence signal was measured using an EGG Berthold LB 96P luminometer.

RESULTS AND DISCUSSION

The nucleotide extension reaction catalyzed by the polymerase activity of DNA polymerase (DNAP) is accompanied by the release of a PPi molecule for each nucleotide incorporated into the growing strand of DNA. Detection of PPi generated upon single nucleotide extension has been previously used as a tool for DNA sequencing and mutation scanning (22,23). In the present study we intended to use PPi generation catalyzed by a DNA polymerase as a method for signal amplification. This was accomplished by identifying conditions that favor the catalytic amplification of PPi production. In addition to the main activity of nucleotide incorporation in a template-dependent manner (nucleotide extension or polymerase activity), certain DNA polymerases also possess exonuclease activities that catalyze nucleotide excision from either the 5′→3′ or 3′→5 direction, or both (24). The 3′→5′ exonuclease activity, also known as proof-reading activity, of DNA polymerases exists to ensure faithful replication and repair of the genome. Some DNA polymerases, such as T7 DNA polymerase (T7 DNAP) and Escherichia coli DNA polymerase I, are known to possess a very strong 3′→5′ exo-nuclease activity (25).

The signal amplification approach described here is directly dependent upon the amplification of PPi synthesis. In order to amplify PPi synthesis, DNA oligonucleotide cassettes that could be used as substrates for T7 DNA polymerase were synthesized (Fig. 1A). The recessive 3′-ends of these cassettes could be extended on a template strand consisting of a contiguous row of adenines. Although the template strand could contain any combination of all four nucleotides, the addition of dATP to the reaction could interfere with the downstream detection step catalyzed by firefly luciferase. It is known that dATP, a closely related analog of ATP, has the ability to non-specifically activate firefly luciferase, thereby generating high background levels of signal (22,23). In our preliminary studies we observed that dGTP and dCTP also exhibited background signal in the assay. This could be due to either possible contamination of dGTP and dCTP with a substance that activates luciferase or an inherent ability to activate firefly luciferase. Hence, we limited dNTPs exclusively to TTP by using a template strand consisting of only adenines. A strong polymerase activity coupled to a strong 3′→5′ exonuclease activity of T7 DNAP allows it to engage in repeated nucleotide extension and excision on the oligonucleotide cassette, as schematically illustrated in Figure 1B. These two activities, occurring in synchrony, are expected to continue until either the TTP is completely consumed or the polymerase becomes inactive. The overall result of the two catalytic activities of T7 DNAP is the liberation of a vast quantity of PPi at the expense of TTP from a single template molecule. PPi accumulated in this manner is converted to ATP by the enzyme ATP sulfurylase and APS. The amount of ATP generated from PPi is then quantified by the enzyme firefly luciferase (Fig. 1B). To accommodate repeated nucleotide extension and excision without damaging the substrate, a psoralen crosslinking site was included near the 3′-end of the oligonucleotide cassette. The crosslinked site prevents potential complete digestion of DNA strands by the 3′→5′ exonuclease activity of T7 DNAP. We refer to the oligonucleotide constructs shown in Figure 1A and their derivatives as SACs.

Figure 1.

(A) Two examples of SACs used in the study. The single-track SAC (top) is a hairpin molecule with a single template strand consisting of 15 residues of adenines. The dual-track SAC (bottom) has two such template strands. (B) A schematic representation of the principle behind the NEESA approach. The polymerase activity of T7 DNAP extends (fill-in reaction) the recessive 3′-end of a SAC molecule with TTP. For each TTP molecule incorporated a PPi molecule is liberated to the medium. Once extension is completed, the template is regenerated by the 3′→5′ exonuclease activity for another round of polymerase activity. The end result of these two reactions is the accumulation of a high concentration of PPi. The PPi produced in this manner will be enzymatically converted to ATP by ATP sulfurylase and APS. The amount of ATP generated in the reaction can be detected with firefly luciferase and luciferin. (C) Bioluminescence signal produced by SAC molecules. Closed circles indicate the signal generated by the dual-track SAC molecule shown in (A), whereas the open circles show the level of signal derived from a single-track SAC molecule. Squares show the control experiment in which a DNA strand (5′-ATGCCTAAGTTTCGAACGCGG CTAGCCAGCTTTTGCTGGCTAGCCGCGT-3′) that could fold into a hairpin structure was used as the substrate. This hairpin substrate allows extension and excision of a single thymidine residue. These experiments were carried out under decoupled conditions.

We tested two oligonucleotide cassettes containing either one or two template strands (Fig. 1C, circles). As expected, the signal generated by the cassette containing two template strands (dual-track SAC) was twice as high as that produced by the cassette with one template strand (single-track SAC). A hairpin template that could only be extended by a single thymidine residue was used as a control (Fig. 1C, squares). The signal generated by this control substrate was very low, indicating that PPi production was not substantially amplified when a single nucleotide was subjected to repeated extension and excision. We observed that an extension of at least 10 nt was necessary to obtain satisfactory signal amplification by this approach. Moreover, the use of DNA polymerases that lack 3′→5′ proof-reading exonuclease activity was not as effective in generating the same level of signal observed with T7 DNAP (data not shown). These enzymes included the DNA polymerase isolated from Thermus aquaticus (Taq DNA polymerase) and the Klenow fragment of E.coli DNA polymerase I. These results support the proposed mechanism (Fig. 1B) for the catalytic amplification of PPi production that requires both polymerase and 3′→5′ exonuclease activities of a DNA polymerase. We call this signal amplification method nucleotide extension and excision-coupled signal amplification (NEESA). Using this approach, we detected low attomole concentrations of SAC molecules containing either one or two substrate strands or polymerase reaction sites (Fig. 1C).

One of the most sensitive detection methods commonly used in immunoassays is the generation of a chemiluminescent signal upon hydrolysis of an adamantyl 1,2-dioxetane aryl phosphate substrate catalyzed by alkaline phosphatase (17,26). We compared the signal produced by the NEESA method, which generates bioluminescence, to that of chemiluminescence produced by alkaline phosphatase (Fig. 2A). At the operational level, the NEESA technique requires a longer incubation period than the alkaline phosphatase reaction (70 versus 10 min). Except for this, the NEESA procedure is simple to use and straightforward. All reagents required for the NEESA technique are stable when stored under the recommended conditions. The graph of the bioluminescent signal generated by NEESA (Fig. 2A, closed circles) has a steeper slope than that of the chemiluminescent signal produced by alkaline phosphatase (open circles). This indicates that the change in signal per SAC concentration change is greater than the corresponding change in signal per alkaline phosphatase concentration change. However, the notable difference between the two techniques is the background signal associated with the NEESA method. The low end sensitivity of the NEESA approach suffers from a high background signal. Using the cut-off value of twice the observed background signal, we estimated that 20 amol is the limit of detection of the assay employing dual-track SAC molecules. This is approximately four times higher than the observed detection limit with alkaline phosphatase (Fig. 2B and C).

Figure 2.

(A) A comparison of the bioluminescent signal of the NEESA method with the chemiluminescent signal generated by alkaline phosphatase. Closed circles show the bioluminescent signal generated by the dual-track SAC molecule according to the NEESA approach carried out under coupled reaction conditions, whereas the open circles indicate the chemiluminescent signal produced by alkaline phosphatase. (B) The bioluminescent signal derived from the NEESA reaction with a low concentration of the dual-track SAC molecules, showing the low end sensitivity of the method. (C) The chemiluminescent signal generated a by low concentration of alkaline phosphatase. In all three panels each data point represents an average value of duplicate readings.

In addition to ATP, its preferred cofactor, other NTPs and dNTPs also activate firefly luciferase to some degree in a non-specific manner. Most effective in this regard is dATP (22,23). Hence, the observed background of bioluminescence is likely to be due to a low level of non-specific activation of luciferase by one or more potential contaminants such as dATP, ATP and PPi. It may be possible to eliminate or decrease the background signal by using ultrapure reagents that are free of such contaminants.

Production of PPi and its conversion to ATP could take place simultaneously (coupled conditions) or sequentially (decoupled conditions). Coupling of the two reactions has the advantage of eliminating potential product inhibition of the polymerase activity by high concentrations of PPi. Contrary to what was expected, however, the signal produced by reactions carried out under decoupled conditions was somewhat higher than that generated under coupled conditions (compare Fig. 1C to Fig. 2A). The reason for this observation is not clear and warrants further investigation. The practical convenience of the assay run by coupling the two reactions outweighs the small sensitivity gain observed under decoupled conditions. Hence, further studies were carried out under conditions that coupled the production and conversion of PPi.

As shown in Figure 2, the sensitivity of the NEESA method observed with a simple oligonucleotide cassette was not sufficient to measure concentrations at very low levels (<10⁶ molecules). Hence, we investigated possible approaches to improve the sensitivity of the signal produced by the NEESA approach. These focused on further enhancement of PPi production. One way of achieving this goal is by increasing the number of polymerase-reactive sites on a single substrate molecule, which will lead to stoichiometric amplification of the signal. Polymeric matrices allow the attachment of multiple SACs, making them an attractive platform for generating polyvalent substrate molecules. Different types of polymeric matrices comprised of DNA itself have been described (20,27,28). These polymeric or dendritic DNA structures have been used to amplify radioactive or fluorescent signals by direct incorporation of label in hybridization-based detection (28–30). In the bDNA approach, DNA-based polymeric matrices have been used as a platform to attach a significant number of alkaline phosphatase molecules that provided amplification of the chemiluminescent signal for detecting very low viral copy numbers (9). Dendritic DNA structures have also been used to amplify the signal in biosensors (21,31). Although polymeric supports made up of monomers other than DNA are also available and may be useful to concentrate large numbers of SACs, an advantage of DNA-based polymers is the presence of natural DNA 3′- and 5′-ends. These ends could easily be modified for facile introduction of SAC molecules. To further amplify PPi production, we chose one of the dendritic DNA matrices called 3D DNA dendrimers (20) to attach a multiple number of SAC molecules.

3D DNA dendrimers are macromolecular assemblies constructed from single-stranded DNA molecules (20,21). Sequential addition of dendritic building block reagents allows the controlled assembly of dendrimers. Each level of assembly is defined as a layer in a dendrimer. The available number of 3′- and 5′-ends (or arms) increases exponentially as the number of layers increases in a 3D DNA dendrimer molecule. The calculated number of 3′-ends on 4-, 6- and 8-layer dendrimer molecules are 162, 1457 and 13 122, respectively. We tested all three dendrimer assemblies, 4-, 6- and 8-layer. SACs were introduced into each dendrimer assembly by hybridization and psoralen photo-crosslinking of a single-stranded DNA template containing a stretch of 30 adenines and a region complementary to the 3′-arm (Fig. 3A, top). This attachment of single-stranded DNA templates converted the 3′-ends of each dendrimer to reaction sites for T7 DNA polymerase to carry out nucleotide extension and excision, thereby amplifying PPi production. The number of reaction sites in a modified dendrimer molecule increases with an increase in the number of layers. As a result, the amount of PPi generated per DNA dendrimer molecule is dependent on the number of 3′-ends in each assembly and is expected to increase with an increase in the number of layers.

Figure 3.

(A) Schematic representation of a DNA dendrimer modified to contain a SAC (top). A 61 nt oligonucleotide containing 31 residues complementary to the C(+) arm of the dendrimer and a 30 nt stretch of an oligoadenosine track was hybridized and covalently crosslinked to the outer surface of the dendrimer. This modification enabled the 3′-end of the C(+) arm to be extended by T7 DNAP. The graph shows the detection of 4- (closed circles), 6- (open circles) and 8-layer (squares) DNA dendrimers consisting of multiple numbers of SACs using the NEESA approach. Each data point represents the average of duplicate readings. (B) Bioluminescent signals obtained by the NEESA method on low concentrations of 4- (top), 6- (middle) and 8-layer (bottom) DNA dendrimers equipped with SACs. Each data point represents the average of duplicate measurements.

The observed bioluminescence generated by the three types of SAC-modified DNA dendrimers (4-, 6- and 8-layer) is shown in Figure 3A (bottom). As expected, the signal increased with an increase in the number of dendrimer layers. Hence, the sensitivity of detection increased in the order 4-layer < 6-layer < 8-layer. In each case, the signal increased linearly with an increase in the amount of DNA dendrimer and subsequently reached a plateau. The saturation of signal could be due to a limitation of one or more ingredients crucial for production and/or conversion of PPi. Out of the three DNA dendrimers tested, the 4-layer assembly was the least sensitive. The detection limit of the 4-layer dendrimer was 1 amol (the signal was greater than twice the observed background value) (Fig. 3B, top). The sensitivity of detection was further improved with the 6-layer dendrimer, to as low as 100 zmol (Fig. 3B, middle). The 8-layer assembly with 13 122 modifiable 3′-ends gave the most sensitive signal, detecting as few as 5 zmol (Fig. 3B, bottom). This is ~3500 molecules for the 8-layer dendrimer assembly.

This level of detection of dendrimer molecules at low zeptomole levels was achieved by combining two approaches to amplify PPi production: (i) allowing the DNA polymerase to repeatedly incorporate several nucleotides, 30 thymidines, on a single template molecule (catalytic amplification); (ii) attaching a great many, hundreds to thousands, of these template molecules to a single polymeric molecule (stoichiometric amplification). As discussed above, a substrate that allowed repeated extension and excision of a single nucleotide was not optimal for amplification of PPi synthesis (Fig. 1C, squares). The attachment of a template that allowed for single nucleotide extension and excision rather than multiple nucleotide incorporation and excision on dendrimer molecules did not produce sensitivity comparable to that observed in Figure 3 (data not shown). Hence, optimized catalytic amplification combined with stoichiometric amplification was required for highly sensitive detection.

A standard curve was generated using known standards of PPi, after conversion to ATP, under identical conditions to those used to test SAC molecules in the assay. The amount of PPi produced by monomeric (single-track SAC), dimeric (dual-track SAC) and multimeric (SACs on 3D DNA dendrimers) substrates was determined using the standard curve. The amount of PPi produced with each substrate was then used to derive the extent of signal amplification (Table 1). We observed ~10⁴-fold amplification of the signal when either monomeric or dimeric SAC molecules were used. The extent of signal amplification obtained with the dual-track SAC substrate was ~2-fold higher than that produced by the single-track SAC molecule. This is in agreement with the mechanism of PPi production. Multimerization of SACs on DNA dendrimers substantially increased the extent of amplification, giving rise to 10⁶- to 10⁸-fold signal amplification. The number of 3′-ends exponentially increases with an increase in the number of layers in DNA dendrimers. The extent of signal amplification observed with DNA dendrimers parallels the exponential increase in available 3′-ends. Based on the observed value of signal amplification for the dual-track SAC substrate containing 30 extendable nucleotide positions per molecule, we have calculated the expected extent of amplifications for dendrimer substrates (Table 1). The calculated values of signal amplification are in good agreement with those obtained experimentally.

Table 1. Extent of signal amplification observed in the NEESA method using SAC molecules in different configurations.

Substrate	Signal amplification
	Observed	Calculated
Single-track SAC	2.1 × 104
Dual-track SAC	5.4 × 104
SAC on 4-layer dendrimer	6.6 × 106	8.7 × 106
SAC on 6-layer dendrimer	6.6 × 107	7.8 × 107
SAC on 8-layer dendrimer	8.0 × 108	7.0 × 108

In each case, the observed signal amplification was derived from a standard curve obtained with PPi standards that were converted to ATP and detected by luciferase. For example, 40 pmol PPi generated 500 000 RLU in the standard curve. The same amount of signal was produced by 6 amol of the 6-layer dendrimer. Hence signal amplification of the 6-layer dendrimer = 40 000 000 amol/6 amol = 6.6 × 10⁶. The calculated signal amplification for each dendrimer substrate was based on the observed value for the dual-track SAC substrate containing 30 extendable nucleotides and the expected number of 3′-ends in the dendrimer. For example, the calculated value for the 4-layer dendrimer substrate with 162 3′-ends = 5.4 × 10⁴ × 162 = 8.7 × 10⁶. A template DNA strand consisting of 30 contiguous adenines was attached to the 3′-ends of each DNA dendrimer.

In PCR, the magnitude of exponential target amplification is dictated by the number of thermal cycles, and varies from ~10⁵ to 10⁸ between 20 and 30 cycles (32). In isothermal target amplification methods, such as the strand displacement assay and transcription-mediated amplification, the magnitude of amplification is dependent on the incubation time of the reaction. The isothermal methods have produced 10⁶- to 10⁸-fold target amplification after 1–2 h incubation (3,4). The magnitude of signal amplification obtained by the NEESA approach employing DNA dendrimer substrates falls well within the range of amplification obtained with exponential target amplification methods. The power of exponential target amplification of PCR has been harnessed to detect targets other than nucleic acid sequences at low concentrations. This has been accomplished by attaching a DNA strand to an antibody in an approach called immuno-PCR (33,34). The antibody conjugated to DNA is used as the secondary antibody in an immunoassay and the presence of DNA is detected by PCR. The immuno-PCR approach provides 10⁵ times more sensitivity than the conventional immunoassay (35). Since immuno-PCR relies on target amplification, the technique is also prone to the contamination associated with PCR. On the other hand, the NEESA approach, providing powerful amplification of the signal rather than the target, would be less affected by contamination.

The 3D DNA dendrimers used in the present study contain two types of arms, 3′ and 5′, in each layer. While these arms in internal layers are occupied by their complementary strands, those in the outer layer remain single-stranded. In the present study, only 3′-arms in the outer layer were modified to introduce SAC molecules. This leaves the 5′-arms free for attachment of the modified dendrimers to ligands that perform molecular recognition in various diagnostic assays. These ligands may include antibodies, aptamers and oligonucleotide sequences that could be used to bind to a wide array of target molecules. A reagent of this nature with the ability to bind to a given target and to report its presence at extremely low concentrations would be of potential use in diagnostic applications. Dendrimers conjugated to specific oligonucleotide sequences have been used to detect the presence of complementary sequences in a biosensor format and also in a hybridization-based assay (21,29). Attachment of DNA dendrimers to antibodies would not be difficult either. In fact, as discussed above, DNA strands have been coupled to antibodies to enhance the sensitivity of detection in immuno-PCR applications (33,34).

The bDNA approach is based on the attachment of thousands of alkaline phosphatase molecules to a target nucleic acid sequence to be detected. This is accomplished by hybridizing a large number of alkaline phosphatase-conjugated oligonucleotides to a ‘tree-like’ structure made up of DNA strands. The bDNA method has given detection of <100 viral copies/ml (9). As demonstrated here, the NEESA approach has the sensitivity to detect DNA dendrimers at extremely low concentrations and does not require enzyme–oligonucleotide conjugates. Based on this observation, it may be possible to develop assays with sensitivities even higher than those based on the bDNA approach using the NEESA technique that employs DNA dendrimer substrates containing hundreds to thousands of SACs.

The NEESA approach described here for signal amplification is simple to use, practical and does not require enzyme–ligand conjugates. Instead of conjugating an enzyme to ligands that mediate binding to an analyte, the current method is based on the recruitment of the enzyme to a molecular platform that could potentially be attached to a wide range of ligands. All reagents used in the assay are stable for long-term storage. The nature of the substrate that one may use in the NEESA approach would depend on the sensitivity of the assay. While dendrimers are the substrate of choice for assays that require extremely high sensitivity, simple substrates like the dual-track SAC substrate may suffice for assays that require only moderate sensitivity. The DNA-based dendrimer platforms described here are manufactured on the industrial scale with accuracy and reproducibility and are commercially available. They are stable for long-term storage, easy to conjugate to a wide array of ligands and provide extremely low detection sensitivity when combined with the NEESA technique.