David H. Bing², Christian Boles¹, Farah N. Rehman¹, Mark Audeh¹, Michael Belmarsh¹, Brian Kelley²,
and Christopher P. Adams^1
1Mosaic Technologies, Inc., 255 Broadway, Winter Hill, MA 02145
²CBR Laboratories, Center for Blood Research, 800 Huntington Ave., Boston, MA 02115

× Ø × Ø × Ø × Ø × Ø × Ø × Ø × Ø × Ø × Ø × Ø × Ø × Ø × Ø × Ø

The essential features of this novel technology are shown schematically in Figure 1. Unlike solid phase capture PCR typing systems where amplification occurs with fluid phase primers, and amplicons are subsequently captured by hybridization to solid phase probes, all amplification in Bridge occurs on the solid surface, and all amplified product remains covalently bound in specific pixels of an ordered array. Sets of PCR primer pairs are immobilized via their 5' ends in discrete areas (pixels) on the surface of an appropriate solid support (Figure 1, lower). Each of the pixels has multiple copies of a single PCR primer pair (negative and positive). Solution phase target DNA anneals to one of the immobilized primers (arbitrarily chosen as positive in Figure 1) and is extended by polymerase during the first cycle. The newly synthesized strand is covalently attached to the support via the primer linkage. During the second cycle, the immobilized first cycle product provides the template for extension by an immobilized primer of opposite polarity (negative in Figure 1). This double-stranded product "bridges" the two primers, and both strands are covalently attached to the support. In the third cycle, both immobilized strands can serve as templates for new primer extension, thereby providing a mechanism for exponential signal growth. Double-stranded Bridge products can be labeled for fluorescence detection using dye-labeled dNTPs or DNA-specific stains. The identity of positive amplification reactions is determined from the position of the fluorescent signal in the array.

Bridge amplification provides four advantages for high-throughput multiplex applications. First, a single Bridge assay can detect tens to hundreds of different genetic markers (human, viral, bacterial) in a single sample. Second, interference from primer artifacts and unequal amplification efficiency between different primer sets is eliminated. Third, cross contamination of samples with amplified DNA is eliminated because the amplified DNA remains bound to the solid phase during and after the amplification reaction. Fourth, by using optical detection methods, amplification progress can be monitored in real time, thereby reducing assay time and providing a reliable method for quantifying input target copy number.

In this report we demonstrate exponential amplification on supports in the absence of solution phase amplification. We have used primer sets covalently immobilized on polystyrene bead supports and have demonstrated a 1.2 to 1.3 fold product increase per cycle for Bridge amplification. While this increase is low compared to the increase per cycle for solution phase PCR (1.5-1.9), it is sufficient to achieve specific amplification of target sequences from 10⁴ copies of genomic DNA. In addition, we present data demonstrating that the Bridge process can be used to type single nucleotide polymorphisms. Thus, as presently configured, Bridge amplification can be used to type human single copy genes in DNA isolated from peripheral blood samples.

Primer preparation: All oligonucleotides were purchased commercially (Ransom Hill Bioscience, Ramona, CA). Primers used for Bridge amplification carried 5' terminal primary amine groups, linked via a six carbon spacer arm. Deprotected, lyophilized primers were dissolved in sterile water, extracted twice with water-saturated isobutanol, brought to 0.2M NaCl, and precipitated with 95% ethanol (3 volumes). The pellet was rinsed in 95% ethanol, and resuspended in water. Concentration was determined from absorbance at 260 nm, assuming that a 33 µg per ml solution has an OD of 1. All concentrations reported refer to primer strands. All oligonucleotide solutions were stored frozen (-20°C).

PEI-coated polystyrene supports: Polystyrene microbeads, 0.87 µm in diameter, modified with primary aliphatic amines (245 µeq/g beads), were purchased commercially (Bangs Labs, Carmel, IN). Prior to use, beads were washed by centrifugation (12,000 x g, 2 minutes) and resuspension in 0.1M KPO4 buffer (pH 6.8). Beads were reacted with 5% glutaraldehyde (EM grade, Polysciences) in 0.1M KPO₄ buffer for 1 hour at room temperature with vigorous shaking on a vortex mixer. The beads were washed twice with 0.1M KPO₄ buffer and once with 0.1M SBB. The beads were then reacted with 3% polyethyleneimine (PEI, 2000 dalton average MW, Aldrich) in 0.1M SBB for 1 hour at room temperature, again with vigorous shaking. Ethanolamine was added to 0.5 M (pH 8.0), and shaking was continued for another hour. The beads were washed extensively with 0.1M SBB and stored at 4°C until use.

Measurement of primer surface density: Primer density was measured by hybridizing ³²P-labeled complementary oligonucleotides to primer-modified supports. In general, probes were designed to be 20 nucleotides in length, and complementary to the 3' terminal region of the immobilized primer. Probes were labeled with terminal transferase (USB) and alpha-³²P-dCTP at 3000 Ci/mmoles (NEN). Unincorporated nucleotides were removed using a silica adsorption method (QiaAmp, Qiagen) and centrifugal ultrafiltration on a Microcon 3 (Amicon). Surface area was calculated using the nominal diameter specified by the supplier.

Amplification using Bridge supports: Reactions (usually 10-50 µl) were carried out in a standard thermal cycler (Perkin Elmer 2400) using a suspension of Bridge supports in place of solution phase primers. Usually, 0.1-0.25 mg of bead support were used in each reaction. Reaction conditions were 10 mM Tris-HCl (pH 8.3 at 25°C), 50 mM KCl, 2.5 mM MgCl2, 200 µM each dNTP, 100 µg/ml BSA, 3.3 µM alpha-³²P-dCTP at 3000 Ci/mmole (NEN), 0.5% Tween 20, 0.05 U/µl AmpliTaq DNA polymerase (Perkin Elmer). Following amplification, beads were washed exhaustively in TE/Tw, either by centrifugation and resuspension, or by vacuum filtration (Multiscreen 0.2 µm filters, Millipore), to remove unincorporated nucleotides. Product incorporation was determined by Cerenkov counting in a liquid scintillation counter (Beckman LS 3801).

Genomic DNA: Genomic DNA was purified from Ficoll/Hypaque-purified peripheral blood lymphocytes by digestion with proteinase K (10 mM Tris-HCl, pH 8, 50 mM NaCl, 10 mM EDTA, 0.5% SDS, 100 µg/ml proteinase K, 12 hours, 37°C), followed by sequential phenol and butanol extractions. The purified DNA was sheared by passing the solution through 16 and 20 gauge hypodermic needles, ethanol precipitated, and resuspended in TE buffer. DNA concentration was determined spectrophotometrically, assuming that a 50 µg/ml solution of DNA has an OD₂₆₀ of 1.

HIV plasmids: Expression plasmids containing wild-type (wt, pKRTwt) and AZT-resistant mutant (pKRT41/215/219) versions of the HIV-1 reverse transcriptase (RT) protein were generously provided by Dr. Rich D’Aquila. The relevant sequences of the two plasmids for codons 40-42 are: wt -- GAG ATG GAA; codon 41 mutant -- GAG CTG GAA. For use as PCR targets, plasmids were restricted with HindIII, and diluted to the indicated copy number with TE containing 10 µg/ml human genomic DNA.

Demonstration of Bridge product formation: The proposed mechanism of Bridge amplification predicts that the amplification products should be attached to the surface by both primers, in an inverted U conformation (see Figure 1, end of cycles 2 and 3). To test this prediction, Bridge amplification was performed with primers containing restriction sites for enzymes which do not cleave within the amplification target. The primers were specific for a 545 base pair (bp) human dystrophin gene fragment (primers D-45-F and D-45-R, Table 1). The forward primer (D-45-F), contained a restriction site for XbaI (see underlined sequence, Table 1) and the reverse primer (D-45-R) carried a restriction site for Cla I. The primers were coupled to epoxysilane-treated silica beads and used for Bridge amplification. A dystrophin PCR product, produced using the same primers in a solution phase PCR reaction, was used as the initial target for the Bridge reaction. Amplification was carried out for 30 cycles using alpha-³²P-labeled dCTP to label products. Following the amplification, the beads were washed to remove unincorporated label, and split into equal portions for restriction analysis of the immobilized products as shown in Figure 2.

Cleavage with either enzyme alone should only liberate extension products attached by one primer (see end of cycle 1, Figure 1). Simultaneous cleavage with both enzymes should liberate singly-bound extension product and doubly-bound Bridge products (see end of cycle 2, Figure 1). As seen in Figure 2, double digestion with Cla I and Xba I liberated at least 3-fold more product (Lane XC, Figure 2) than either single digestion. Release of product by single digestion is expected (Lanes X and C, Figure 2), since the solution phase target DNA is the same size as the Bridge product, and the target was present throughout the amplification reaction. These data demonstrate that approximately two thirds (63%) of the products are bound by both primer ends. Since no agglutination of beads occurred during the amplification reaction, the data suggest that the doubly-bound product is the Bridge product.

Exponential kinetics of Bridge amplification: The Bridge model presented in Figure 1 predicts an exponential signal increase of 2 per cycle, as expected for solution phase PCR. To experimentally check the kinetics of Bridge amplification, the time course shown in Figure 3 was carried out. Bridge primers specific for a 270 bp human beta globin gene fragment (Bglo(+) and Bglo(-), Table 1) were coupled to polyethyleneimine-coated polystyrene microbeads (PEI-PS supports), and used in Bridge amplification reactions. The initial target was a double-stranded beta-globin fragment (Figure 4b) produced using solution phase PCR primers (Bglo3up and Bglo2D, Table 1) and human genomic DNA. After the first cycle, the beads were washed to remove solution phase target, and placed in fresh PCR mix lacking target DNA. The increase in bead-bound product was determined from the incorporation of ³²P-labeled nucleotides. The time course of the reaction through 10 cycles, presented in Figure. 3, shows that bead-bound product increases with an approximate 1.2-fold gain per cycle for cycles 4 through 10, consistent with exponential amplification.

In this experiment, an artifactual Bridge signal could result if primers lost from the support establish a solution phase PCR reaction that feeds solution-phase PCR product back to the beads. To rule out this possibility, we analyzed the DNA products generated in the solution and solid phases of the experiment shown in Figure 3. The restriction map of the initial target DNA, and the predicted extension products are shown in Figures 4b-4e. At the end of 1, 7 and 10 cycles, beads were pelleted by centrifugation, and the solution phases were removed, ethanol precipitated, and restricted with Sty I. The beads were washed, counted to produce the data of Figure 3, and then also restricted with Sty I. The restriction digests were analyzed on a nondenaturing acrylamide gel, shown in Figure 4a. The results show that there is only a small amount of product in the solution phase, and that product does not increase between cycles 7 and 10. There is at least 10-fold more product on the beads, and more importantly, the bead-bound product increases markedly throughout the amplification reaction. These results demonstrate that the signal increase shown in Figure 3 results from a true solid phase amplification process that operates in the absence of a solution phase component.

Bridge amplification of human genomic DNA targets: The PEI-PS Bridge supports used in Figures 3 and 4, were also capable of amplifying the expected beta-globin gene fragment from a human genomic DNA target, as shown in Figure 5. Bridge amplification reactions (20 µl) containing 0.1 µg of purified human genomic DNA were cycled 80 times (3 hours total cycling), and DNA products present in the solution phase (S lanes) and bead phase (B) were analyzed by restriction and gel electrophoresis as described for Figure 4a. The predicted globin amplification product is found only on the beads -- no specific solution phase products were found. These results demonstrate that Bridge amplification has sufficient sensitivity and specificity to perform amplification of human genomic targets at 10⁴ initial input copies.

Allele-specific Bridge amplification: We have performed two experiments to determine whether the Bridge process can be used to type single nucleotide sequence polymorphisms. First, a two cycle Bridge amplification was performed to assess the specificity of priming during Bridge product formation -- the step equivalent to the second cycle shown in Figure 1. This test involved typing HLA DQA1 alleles. Second, a 20 cycle Bridge amplification was carried out using allele-specific primers for wt and mutant HIV-1 reverse transcriptase genes.

The polymorphic second exon of the HLA DQA1 gene was amplified from two individuals using solution phase PCR. One individual was DQA1 type 0103/0501 and the other was 0103/0102. Equal amounts of each amplified product were added to Bridge reactions specific for the 0103 allele or the 0501 allele. The beads were cycled once, washed in NaOH to remove noncovalently bound product, and cycled once more in the presence of radioactive label. After the second cycle, the beads were washed and radioactive incorporation was measured.

The second allele typing experiment was carried out in the context of a full Bridge amplification reaction using 20 amplification cycles. Four amplifications were carried out in parallel, using primer sets specific for mutant and wt states of HIV-1 reverse transcriptase codon 41. Plasmids carrying the two forms of the gene were used as input target (see Table 1 for primers and Materials and Methods for the relevant codon 41 sequences). The data are shown in Figure 7. Both primer sets were capable of discriminating the two sequences. The difference between signals obtained from homologous and heterologous primer/target combinations ranged from 8-fold (mutant primer set) to 100-fold (wt primer set). Similar levels of discrimination were obtained using these same primer sets in solution phase PCR reactions, suggesting that the immobilization reaction does not alter primer specificity. Taken together with the HLA data, these data demonstrate the feasibility of using Bridge amplification to type allelic polymorphisms.

The results presented in this paper demonstrate that Bridge amplification has many properties desirable for a high-throughput genetic typing method. From a practical standpoint, Bridge’s two major advantages are its high multiplex capacity, and the elimination of the potential for carryover contamination. At its current stage of development, Bridge amplification can amplify genomic single-copy targets at levels of sensitivity and specificity needed for many genetic screening applications. With additional increases in amplification efficiency, Bridge amplification will also find applications in infectious disease diagnostics and forensic testing. There are many potential variables which will should affect amplification efficiency including choice of support material and geometry, primer surface density, target length, and primer attachment chemistry. Optimization of several of these variables will probably be necessary to achieve the highest sensitivity.

Figure 8 shows another strategy for improving the sensitivity of Bridge assays. In all solid phase assays, it is important to ensure efficient interaction between solution phase target and the support. In the bead cartridge device shown in Figure 8, electrophoretic mixing is performed to allow capture of denatured target DNA on the amplification support prior to addition of amplification reagents. A linear array of primer-modified beads (0.1-1mm diameter) serves as the support. Each bead is modified with different primer set. Following the target capture step, amplification reagents are added the cartridge, and thermocycling is performed to amplify target. Detection is performed by optically scanning the bead array, and positive reactions are identified from their position in the array.

In addition to the commercial applications in genetic screening and infectious disease diagnostics, we feel that Bridge amplification will have broad research applications for analysis of gene expression, genetic mapping, and many other areas of genomic research.

The authors acknowledge Mike Boss for dystrophin primer sequence, Keith Crawford for the gift of purified human WBC’s, Rich D’Aquila for HIV-RT plasmids, the helpful suggestions of Janice Williamson, and Bob Mandle for material and intellectual support. This work was supported by NIH SBIR grant R43 HL54402.

REFERENCES

Bglo(-)-mT 5’-NH₂-(CH₂)₆-T₈ GAAGAGCCAAGGACAGGTAC-3’

Bglo(+)-mT 5’-NH₂-(CH₂)₆-T₈ CAACTTCATCCACGTTCACC-3’

Bglo3up 5’-CG ATCTTCAATATGCTTACCAAGCTGTGATTCCA-3’

Bglo2D 5’-AAGGACTCAAAGAACCTCTGGGTCCAAGGGTAG-3’

D-45-F 5’-NH₂-(CH₂)₆-T₄-CTCTAGAAAACATGGAACATCCTTGTGGGGAC-3’

D-45-R 5’-NH₂-(CH₂)₆-T₄-CATCGATCATTCCTATTAGATCTGTCGCCCTAC-3’

A301 5’-NH₂-(CH2)₆-T₈-TATGATGTTCAAGTTGTGTTTTGC-3’

A504 5’-NH₂-(CH₂)₆-T₈-TACGGTCCCTCTGGCCAGTT-3’

A502 5’-NH₂-(CH2)₆-T₈-TACGGTCCCTCTGGCCAGTA-3’

A301 5’-NH₂-(CH2)₆-T₈-TATGATGTTCAAGTTGTGTTTTGC-3’

GH26 5’-GTGCTGCAGGTGTA AAC TTG TACCAG-3’

GH27-PO₄ 5’-PO₄-CACGGATCCGGTAGCAGCGGTAGAGTTG-3’

codon 41 wt 5’-NH₂-(CH₂)₆-T₈-TAAGCATTAGTAGAAATTTGTACAGACA-3’

codon 41 mut 5’-NH₂-(CH₂)₆-T₈-TAAGCATTA GTAGAAATTTGTACAGACC-3’

1. reverse primer5’-NH₂-(CH₂)₆-T₁₁-GTCATGCTACTTTGGAATA-3’

Figure 1: Schematic representation of Bridge Amplification. Bridge amplification is a solid phase method for amplifying target nucleic acid sequences. The technology is based on the covalent immobilization of multiple sets of primer pairs on a derivatized solid support. A unique feature is the ability to detect target sequences without using solution based primer sets, hybridization, or electrophoresis. Shown in this general scheme at the bottom of the figure is an array of 100 pixels on a flat surface. In fact, the support could be a bead, chip or any other suitable solid phase format (see Figure 8). Each pixel contains primer pairs (negative and positive) at a density of approximately 10⁹ primers per mm². Each pixel is specific for one target DNA sequence. Shown in the top left of the figure is an isolated pair of 5' immobilized primers (positive and negative) and a specific target DNA strand. The solution above the array contains amplification buffer, target DNA, polymerase, and labeled dNTPs. The temperature is raised to 94°C for target denaturation. When the temperature is lowered, the target DNA hybridizes to the primer. The temperature is raised to 72^oC and polymerase extension occurs. When the temperature is raised to start the second cycle (94^oC), the target is released to participate in other hybridization reactions, while the extension product remains covalently linked to the primer. The temperature is lowered and the 3' end of the extended product hybridizes to the complementary primer (negative). During extension, a double-stranded bridge is formed with each strand covalently linked to one of the primer pair. Denaturation for the start of the third cycle leaves both extension products free to participate in further amplification cycles. All amplified DNA remains covalently attached to a specific pixel on the array. Detection of incorporated label in a pixel indicates the presence of a specific target DNA sequence in the sample.

Figure 2. Restriction analysis of the confirmation of Bridge amplification products. Bridge amplification was carried out using primers specific for a 545 bp human dystrophin gene fragment (D-45-F, D-45-R, Table 1). The forward primer contained a restriction site for XbaI and the reverse primer carried a restriction site for Cla I. Neither enzyme cleaves within the amplified product. Following amplification the beads were washed to remove unincorporated label, and split into four aliquots. One aliquot was incubated in restriction buffer without enzyme

(-), one was cleaved with Xba I (X lane), one was cleaved with Cla I (C lane), and the last was cleaved with both enzymes (XC lane). Densitometric analysis of the gel is shown at right. The relative intensities of the band are 0.86:0.39:3.34 (C:X:XC). These data demonstrate that the majority of the Bridge products (63%) are only released by double digestion, and therefore are bound by both primer ends in the predicted Bridge conformation
(see Figure 1).

Figure 3. Exponential kinetics of Bridge amplification. 10¹¹ copies of a 600 bp double-stranded beta-globin PCR product (generated with primers Bglo3up, Bglo2D, see Table 1 and Figure 4b) were incubated with primer-modified PEI-PS beads (0.60 mg) in reaction mix containing ³²P-labeled nucleotides (see Materials and Methods for conditions). One amplification cycle was performed (94°C, 10 sec.; 60°C, 1 min.), and the beads were washed to remove solution phase target. The washed beads were divided into four equal portions. The first was counted to determine first cycle incorporation. The remaining three portions were returned to separate PCR mixtures (without additional target) and cycled for 3, 6, or 9 additional cycles. Following cycling, the beads were spun down, and the reaction supernatants were removed and purified by ethanol precipitation for gel analysis (see Figure 4). The beads were washed to remove unincorporated label and product was determined by Cerenkov counting.

Figure 4. Gel analysis of reaction beta-globin PCR products. Part 4a: After determining bead incorporation (Figure 3), bead pellets and purified reaction supernatants were digested with Sty I. Products from half of each reaction supernatant were loaded in each lane (left part of gel). An unrestricted sample of the cycle 10 supernatant is shown as a control for Sty I digestion. Bead products from the entire bead pellet were loaded in each lane (right side of gel). As expected from the restriction map of the products (Figure 4e), the predominant product fragments released from the beads are the 135 and 56 bp species. Even allowing for the 2-fold underloading of the supernatant samples, the bead products from cycles 7 and 10 are present in great excess over the solution phase products. In addition, no increase in solution phase product is seen going from cycle 7 to 10, whereas the bead products show clear cycle-dependent increases. Part 4b: Restriction map of the input target DNA. The input target was a PCR fragment generated from genomic human DNA with primers Bglo2D and 3up (Table 1). The position of the Bridge amplification primers (Bglo(+) and (-), Table 1) within the input target is indicated by the half arrows above the map. Sty I cleavage sites are indicated by hash marks across DNA strands. Parts 4c-4e: Restriction maps of Bridge products predicted after cycles 1-3. Only one of two Bridge products expected after cycle 2 as illustrated in part 4d. The crosshatched line at the bottom in each drawing is the support and the Bridge primers are shown as tethered half arrows.

Figure 5. Bridge amplification of human genomic DNA. Bridge amplifications were carried out using human beta-globin primers (Bglo(+) and Bglo(-) primers, Table 1) on PEI-PS bead supports as described in Figures 3 and 4, except that 80 cycles were performed. 10⁴ copies of human genomic DNA ("Genomic" lanes) or TE buffer ("TE" lanes) were used as initial target. Following amplification, gel analyses of bead products were carried out as described for Figure 4a. The analysis confirms that only the bead fraction contains the diagnostic 135 bp Sty I restriction product. No solution phase product was formed.

Figure 6. HLA typing using a two cycle Bridge process. Bridge supports with primer sets specific for HLA DQA1 alleles 0103 and 0501 were used to type DQA1 second exon solution phase PCR products generated from individuals of known DQA1 type. The experiment is described in detail in Materials and Methods. The signal reflects the efficiency of the Bridge formation step shown in the second cycle of Figure 1. High signals were seen for both Bridge reactions using PCR product from the individual with genotype 0103/0501. In contrast, only the 0103-specific support showed high signal using the product from the individual with genotype 0102/0103. These data demonstrate that the priming reaction that generates the Bridge product is specific enough for allele-specific typing.

Figure 7. Typing single point mutations with Bridge amplification. Primers sets specific for mutant and wild-type sequences at codon 41 of the HIV reverse transcriptase gene (see Table 1) were attached to PEI-PS microbeads and used in 20 cycle Bridge amplifications. Input target was 10⁹ HIV-RT plasmid copies. High amplification signals were obtained only when the target and primer type matched. Target discrimination varied between 8-fold (mutant primers) and 100-fold (wt primers) for the primer sets used in these experiments.

Figure 8. Proposed Bridge amplification design for high sensitivity applications. Bead Bridge amplification supports (0.1-1mm in diameter) are arranged in a linear array within a tubular cartridge. Each bead is modified with a different primer set, and therefore comprises a distinct amplification test site. Denatured sample is introduced into the cartridge, and the single-stranded DNA is electrophoretically mixed throughout the cartridge to allow efficient target capture on the amplification support. Following capture, amplification reagents are introduced into the cartridge, and thermocycling is initiated. Amplified products are detected on the beads using optical methods, in this case using a fluorescent DNA label or stain. The identity of a positive reaction is determined from the position of the bead within the array.