Steric Effects and Mass-Transfer Limitations Surrounding Amplification

May 6, 2009 - In this paper, we focus on the in vitro transcription reaction (IVT) of both a long and short strand of H5 influenza A RNA (1777 and 465...
0 downloads 0 Views 2MB Size
Download PDF
pubs.acs.org/Langmuir © 2009 American Chemical Society

Steric Effects and Mass-Transfer Limitations Surrounding Amplification Reactions on Immobilized Long and Clinically Relevant DNA Templates Stephanie E. McCalla, Alexander L. Luryi, and Anubhav Tripathi* Biomedical Engineering Program Division of Engineering and Medical Sciences, Brown University, Providence, Rhode Island 02912 Received December 16, 2008. Revised Manuscript Received February 18, 2009 DNA and RNA are commonly captured on solid substrates during purification and isolation, where they can be transferred to downstream amplification and transcription reactions. When compared to the solution phase, however, immobilized DNA- and RNA-directed reactions are less efficient because of a variety of complex factors. Steric inhibition because of the bead surface and neighboring biological polymers, a change in solution chemistry because of the high local concentration of template molecules, and mass transfer to the bead surface could all affect the overall reaction kinetics. Furthermore, these effects may be particularly evident when working with long clinically relevant molecules, such as mRNA, viral RNA, and cDNA. In this paper, we focus on the in vitro transcription reaction (IVT) of both a long and short strand of H5 influenza A RNA (1777 and 465 nt) on both free and immobilized DNA templates to study these phenomena. We found that transcription was less efficient on immobilized beads than in solution, but that it can be dramatically increased with optimal solution chemistry. Using high ribonucleotide concentrations (>6 mM total rNTP), the RNA yield from long immobilized cDNA templates was boosted to 60% of solution control. Surprisingly, we found that steric effects because of surrounding immobilized molecules were only significant when the DNA molecules were short enough to achieve a high density (9  10-4 μm2/molecule) on the silica substrate, such that the gap between molecules is on the order of the polymerase diameter. Eventually, these findings can be exploited in an automated microreactor, where isolation, purification, amplification, and detection of nucleic acids can be unified into one portable device.

Introduction Clinical specimens frequently contain small concentrations of RNA contaminated with unwanted genetic material, pathogens, and debri. These conditions require purification and amplification to obtain sufficient material for expression analysis or detection. For instance, a single cell is estimated to contain between 0.1 and 1 pg of mRNA, which is too small for current analysis techniques.1 Many diagnostic applications require a high-throughput, reproducible RNA amplification strategy with minimal skewing of gene expression. In vitro transcription (IVT) fills these requirements. IVT is a linear amplification process carried out at a constant temperature, eliminating both exponential bias seen in polymerase chain reaction (PCR) and the need for a thermal cycler.2 Additionally, any errors made by the polymerase would not be exponentially amplified as they would be during PCR.1 During IVT reactions (Figure 1), a DNA-dependent RNA phage polymerase repeatedly creates RNA from a DNA template. The polymerase binds to a promoter site, synthesizes RNA using the bottom cDNA strand as a template, and terminates at the end of the DNA chain, allowing for the release of the new RNA transcript.3 During RNA elongation, the addition of each new magnesium-bound rNTP results in the following reaction:4 RNAj þ MgNTP2 - fRNAj þ1 þ MgP2 O7 2 - þ H þ

ð1Þ

*To whom correspondence should be addressed. E-mail: anubhav_tripathi@ brown.edu. (1) (2) (3) (4)

Phillips, J.; Eberwine, J. H. Methods 1996, 10, 283–288. Moll, P. R.; Duschl, J.; Richter, K. Anal. Biochem. 2004, 334(1), 164. Sousa, R.; Mukherjee, S. Prog. Nucleic Acid Res. Mol. Biol. 2003, 73, 1–41. Kern, J. A.; Davis, R. H. Biotechnol. Prog. 1999, 15(2), 174–184.

6168

DOI: 10.1021/la804144s

The rate-limiting step of the reaction is product release rather than elongation of the RNA product;5 therefore, it is not surprising that detailed kinetic studies using long DNA templates show that efficient transcription of long templates can indeed be achieved.6 An optimized IVT reaction can synthesize full-length RNA at a maximum rate with minimal abortive transcript production, meaning that short (2-14 nucleotides) RNA segments are not prematurely released from the DNA/polymerase complex.3,4,6,7 The efficiency of these reactions is dictated by a variety of factors, including solution pH, the specific DNA template, and the concentrations of magnesium, enzyme, reactants, and products. The removal of products is particularly important for reaction efficiency, because RNA, pyrophosphate (P2O47 ), and proton molecules produced during the reaction inhibit further RNA synthesis.8 These adverse solution conditions make it advantageous to immobilize rare or expensive DNA templates because solution conditions could be carefully controlled in a reactor while recycling the template, resulting in higher full-length RNA yield and lower transcript aborts.4,8-11 Furthermore, immobilized DNA and RNA can be removed from contaminated clinical samples and even reused for subsequent reactions.10,12-14 (5) Kuzmine, I. M. J. Mol. Biol. 2001, 305, 559–566. (6) Arnold, S. Biotechnol. Bioeng. 2001, 72(5), 548–561. (7) Jia, Y. P.; Patel, S. S. Biochemistry 1997, 36(14), 4223–4232. (8) Kern, J. A.; Davis, R. H. Biotechnol. Prog. 1997, 13(6), 747–756. (9) Breckenridge, N. C.; Davis, R. H. Biotechnol. Bioeng. 2000, 69(6), 679–687. (10) Young, J. S.; Ramirez, W. F.; Davis, R. H. Biotechnol. Bioeng. 1997, 56(2), 210–220. (11) Gallifuoco, A.; Alfani, F.; Cantarella, M. Biotechnol. Bioeng. 2002, 79(6), 641–646. (12) Davis, R. H.; Breckenridge, N. C. J. Biotechnol. 1999, 71(1-3), 25–37. (13) Marble, H. A.; Davis, R. H. Biotechnol. Prog. 1995, 11(4), 393–396. (14) Andreadis, J. D.; Chrisey, L. A. Nucleic Acids Res. 2000, 28(2), E5–E5.

Published on Web 5/6/2009

Langmuir 2009, 25(11), 6168–6175

McCalla et al.

Article

Figure 1. Biotin-functionalized H5 cDNA was anchored to 4.82 μm silica beads coated with streptavidin. A hydrophobic polymer spacer and extra DNA bases were added before the promoter site (in bold) to prevent steric inhibition of enzyme binding from the bead surface. DNA was added at different concentrations to binding reactions to create different bound densities on the bead surface. Rs represents the average distance between each DNA molecule.

Immobilizing biological molecules on solid surfaces is common during isolation and purification of rare samples of DNA and RNA from large, contaminated sources, such as cell homogenate or clinical samples. As a result, current literature addresses many ways to immobilize biological polymers.15-18 A simple and cost-effective method of nucleotide capture is biotinylated complementary oligonucleotides in the form of hybridization probes and PCR primers that specifically hybridize DNA and RNA. These probes are immobilized on a solid substrate, such as streptavidin-coated beads.13,17-20 Once immobilized, these molecules can be used in amplification, reverse transcription, or transcription reactions. Basic studies on DNA-directed enzyme reactions, such as immobilized PCR,14,15,21,22 rolling circle amplification,23 and in vitro transcription9,10,12-15,19,20,24,25 provide a proof of concept for template-immobilized polymerase reactions. Most literature focuses on short (10-26 bp) model DNA templates,9,10,12,13,19 with minimal details on template-immobilized RNA production efficiency. Detailed kinetics of long molecules (0.5-2 kbp) in solution have been extensively investigated,6 but studies of transcription using long, immobilized templates lack kinetic analysis and have minimal comparison to solution control efficiency.14,15,20,24,25 Researchers reported both a drop in RNA production for long immobilized DNA templates,14,25 a slight drop in RNA production for immobilized DNA templates producing 26 nt RNA, and no difference between solution-phase and immobilized transcription for DNA templates producing 12 nt RNA,13 implying that transcript length affects immobilized transcription kinetics. Clinically relevant RNA, such as mRNAs and viral RNAs, are often long; for example, influenza A RNA can be up to 2320 nt in length depending upon the segment of (15) Rege, K.; Viswanathan, G.; Zhu, G.; Vijayaraghavan, A.; Ajayan, P. M.; Dordick, J. S. Small 2006, 2(6), 718–722. (16) Shchepinov, M. S.; Case-Green, S. C.; Southern, E. M. Nucleic Acids Res. 1997, 25(6), 1155–1161. (17) Du, Q. Top. Curr. Chem. 2006, 261, 45–61. (18) Huang, S. C.; Swerdlow, H.; Caldwell, K. D. Anal. Biochem. 1994, 222(2), 441–449. (19) Fujita, K.; Silver, J. BioTechniques 1993, 14(4), 608–617. (20) Liu, M. P. Promega Notes 1997, No.64, 21. (21) Carmon, A; Mitchell, S. E.; Thannhauser, T. W.; Muller, U.; Kresovich, S. BioTechniques 2002, 32(2), 410–420. (22) Kohsaka, H.; Carson, D. A. J. Clin. Lab. Anal. 1994, 8(6), 452–455. (23) McCarthy, E. L. Anal. Bioanal. Chem. 2006, 386(7), 1975–1984. (24) Steffen, J.; Nickisch-Rosenegk, M.; Bier, F. F. Lab Chip 2005, 5(6), 665–668. (25) Ghosh, D. J. Biochem. Biophys. Methods 2005, 62(1), 51–62.

Langmuir 2009, 25(11), 6168–6175

interest.26 Many factors that have previously been overlooked need to be considered when immobilizing long DNA templates; parameters such as DNA radius of gyration, ratio of DNA molecules to bead surface area, and solution chemistry may affect the results of DNA/bead binding and subsequent RNA transcription. Because the DNA flexibility and radius of gyration may contribute to the steric effects, the effect of excess magnesium could also change between solution-phase and bead-immobilized IVT for long DNA segments.27,28 We present a model system to examine the effect of solution chemistry, DNA size, and DNA packing density on immobilized transcription kinetics. The results can inform solution chemistry and binding density of the immobilized transcription reaction for optimal production of long clinically relevant RNA, which is vital for the development of microreactors that employ beadimmobilized templates for RNA amplification. These results also inform potential mechanisms for the drop in immobilized transcription efficiency seen in both this study and the literature.

Materials and Methods Synthesis of H5 Template DNA. A wild-type humanisolate influenza A H5 DNA sequence [A/Hanoi/30408/2005 (H5N1) NCBI-AB239125] was synthesized (DNA 2.0, Menlo Park, CA) as an IVT template, inserted into a pJ10 Escherichia coli propagation plasmid, and sequenced. The template encodes for the hemagglutinin (HA) segment of the influenza virus. Plasmids were linearized using BspEI (New England Biolabs, Ipswich, MA) and amplified using PCR. Custom primers were ordered from Integrated DNA Technologies (Coralville, IA). Biotinylated H5 DNA was created using a primer linked to a biotin molecule and a TEG polymer spacer: 50 -biotin-TEGGGAGGCCGGAGAATTGTACGATTTAGGTGACACTA TAGAGTAGAAACAAGG-30 . The underlined section represents the SP6 polymerase promoter region, while the bold section corresponds to nucleotides from the H5 viral insert. The 50 end of the sequence was chosen from the SP6 primers in the literature,2 which contain spacer sequences preceding the SP6 priming site. The second primer for the long DNA segments (1814 bp), 50 -AGCAAAAGCAGGGGTTCAATCTGTC-30 , is specific (26) Fundamental Virology, 4th ed.; Knipe, D. M., Howley, P. M., Eds.; Lippincott Williams and Wilkins: Philadelphia, PA, 2001; p 1395. (27) Ahmad, R.; Arakawa, H.; Tajmir-Riahi, H. A. Biophys. J. 2003, 84(4), 2460–2466. (28) Draper, D. E. RNA 2004, 10(3), 335–343.

DOI: 10.1021/la804144s

6169

Article

McCalla et al.

for the conserved 50 end of influenza A RNA. The second primer for the short (502 bp) H5 cDNA fragment, 50 -ATGGAAGACGGGTTCCTAGATG-30 , targeted a sequence within the H5 cDNA. Solution conditions for the PCR reaction were 0.02 unit/μL Taq DNA polymerase (New England Biolabs, Ipswich, MA), 0.2 mM dNTP, 0.5 μM each primer, 0.075 ng/μL linearized H5 plasmid, 10 mM Tris-HCl, 50 mM KCl, and 1.5 mM MgCl2 for a total volume of 100 μL. PCR was performed on a BioRad MyCycler (Hercules, CA) operated at maximum ramp rates with an initial denature of 95 °C for 2 min, then 36 cycles consisting of a 95 °C denaturing step for 30 s, a 56 °C annealing step for 30 s, and a 72 °C extension step for 2 min. All PCR constructs were oriented to produce native viral RNA beginning with the sequence “G AGU”, with the first G being from the last base in the promoter sequence above. PCR products were then purified using the Wizard SV gel and PCR cleanup system (Promega, Madison, WI) and quantified by measuring the absorbance at 260 nm with an extinction coefficient of 40 ng cm μL-1 using an ND-1000 spectrophotometer (Nanodrop Technologies, Wilmington, DE). Immobilization of the cDNA Template. Nonporous streptavidin-coated silica beads with a diameter of 4.82 μm and 1.35  6 10 biotin-FITC binding sites per bead (Bangs Laboratories, Fishers, IN) were washed 3 times in DI water and twice in binding buffer to condition the beads [10 mM Tris, 0.2% Tween 20, 1 M NaCl, and 1 mM EDTA at pH 8]. Note that the specific number of biotin sites per bead does not significantly effect binding because even in the most densely bound conditions DNA occupied e6% of available biotin sites. Earlier experiments using beads with an order of magnitude more binding sites yielded similar results for DNA binding and RNA transcription (data not shown). Biotinylated DNA was incubated with streptavidin beads for 30 min in binding buffer at ambient temperature, during which the bead/DNA mixture was agitated at 1200 rpm to prevent sedimentation. Control tubes with DNA but without beads were simultaneously incubated to rule out nonspecific adsorption to the tube walls. A separate control was run during one 1814 bp DNA-binding reaction and one 502 bp DNA-binding reaction to rule out nonspecific DNA adsorption to the beads. DNA identical to the cDNA used in the binding reactions but without the 50 biotin molecule was incubated with the beads. The supernatant was analyzed from both the control tubes and the tubes containing beads using Sybrgreen (Molecular Probes, Eugene, OR) for low concentrations of DNA and the Nanodrop ND-1000 spectrophotometer (Nanodrop Technologies, Wilmington, DE) for high concentrations of DNA. The amount of DNA remaining in solution was subtracted from the amount of DNA added to find the amount of DNA bound to the bead surface, and DNA density was calculated assuming that the DNA is uniformly bound to the beads. There was no difference between the DNA concentration in the supernatant from the tube containing nonbiotinylated DNA with beads and the solution from the tube containing DNA only. The amount of DNA in these controls matched the amount of DNA originally added to these tubes, ruling out nonspecific adsorption to either the tube walls or the bead surface. Beads were then washed twice in an excess of 10 mM Tris buffer at pH 8.5. Subsequent supernatant did not have measurable amounts of DNA. Experiments using 1814 bp cDNA were run on several different bead stocks to rule out false trends because of beads lost during wash steps. Bead-immobilized DNA was stored at 4 °C and used within 1 month of preparation. In Vitro Transcription. In vitro transcription was performed using the SP6 RNA polymerase, buffer, and reagents from New England Biolabs (Ipswich, MA) unless otherwise specified. Reactions with volumes of 10-20 μL were prepared as specified by the company protocol and terminated by rapid freezing at -80 °C, followed by 20 min of incubation with 0.1 unit/μL of Turbo DNase (Ambion, Austin, TX) at 37 °C. Solution conditions are as follows, unless otherwise specified: 42.5 mM

6170

DOI: 10.1021/la804144s

Tris, [rNTP] + 4 mM MgCl2, 10 mM dithiothreitol, 2 mM spermidine, 200 μg/mL ultrapure BSA (Ambion, Austin, TX), 4 units/μL pyrophosphatase, 1 unit/μL RNase inhibitor, and 1 unit/μL SP6 polymerase. The rNTP concentration varied from 0.5 to 2.5 mM each of ATP, UTP, GTP, and CTP (2-10 mM total rNTPs). IVT reactions were run at 40 °C while agitated at 1200 rpm to prevent bead sedimentation. Solution controls were run under these same conditions for consistency. High yields of RNA (>0.1 ng/μL DNA) were quantified using the RNA Nano 2100 electrophoresis bioanalyzer (Agilent, Santa Clara, CA). Low RNA yields were quantified using calibrated Ribogreen fluorescence (Molecular Probes, Eugene, OR), and RNA size and full-length peak ratios were verified on representative samples using the RNA Nano 2100 electrophoresis bioanalyzer. The fraction rNTPs used in each reaction was calculated by the following equation: ½rNTP -½RNAnt ½rNTP

ð2Þ

where [rNTP] is the number of available rNTP molecules, [RNA] is the number of RNA molecules produced in the transcription reaction, and nt is the number of nucleotides in the RNA. Note that the short RNA aborts will not significantly factor into the total RNA produced, because short aborts are at maximum of 0.7% of the full-length RNA mass.3

Results To better understand the steric effects caused by the interaction between neighboring immobilized molecules and the interaction between the DNA and the enzyme, we introduce the following dimensionless parameters: β ¼

Rs Rp

ð3Þ

γ ¼

Rs 2Rg

ð4Þ

The parameter β represents the relative space between DNA molecules available for the RNA polymerase and is calculated by normalizing the average distance between occupied DNA-binding sites (Rs, see Figure 1) to the polymerase radius (Rp ≈ 8 nm).29 The parameter γ represents the half-distance between bound cDNA molecules normalized to the radius of gyration of the cDNA, and it roughly measures the level of possible interaction between neighboring DNA molecules: γ < 1 implies that neighboring molecules can interact, while γ > 1 implies that neighboring cDNA molecules cannot interact. The DNA used in these studies can be treated as a wormlike chain, which has an approximate radius of gyration sffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi ε -a -ae-L=a Þ 2aðLNeff Rg ¼ 6

ð5Þ

where a = 50 nm is the persistence length of DNA, L is the end-end distance of the DNA molecule, Neff = L/2a is the number of statistical segments in the DNA, and ε = 0.1 is a correction factor for an excluded volume.30 It should be noted that the DNA radius of gyration is a dynamic property that (29) Sousa, R. Nature 1993, 364(6438), 593–599. (30) Bloomfield, V. A.; Crothers, D. M.; Tinoco, I. Nucleic Acids: Structures, Properties, and Functions; University Science Books: Herndon, VA, 2000.

Langmuir 2009, 25(11), 6168–6175

McCalla et al.

changes based on local environment and solution conditions, such that the radius of gyration is merely an estimation of the space occupied by dilute DNA in solution.30-32 Transcription Kinetics of Long Immobilized DNA. Transcription reactions with identical solution conditions and mass of DNA template were run simultaneously in solution and on beads to investigate the effect of immobilization on reaction efficiency. To assess the effect of DNA immobilization on RNA transcription, 1.9 ng/μL of biotin-functionalized full-length H5 cDNA (lane 2 in Figure 2a) was added to a solution with streptavidin-coated silica beads without a wash step and run for 4 h with 2 mM total rNTP, such that DNA was available both immobilized on the bead surface and free in solution. A solution control (free cDNA template) reaction was run simultaneously for comparison. Transcription in the presence of the beads yielded 75 ( 9% of total RNA production when compared to the solution control. RNA degradation was ruled out as a cause of the decreased yield because of the clear electropherogram peak visible in both samples (Figure 2b). Nonspecific adsorption of RNA to the bead surface was unlikely because of the presence of streptavidin and cDNA on the bead surface. To further investigate these effects, transcription products from beads with γ = 0.15 and β = 4 were evaluated from 10 μL reactions incubated from 0 to 4 h (Figure 3a). The RNA yield begins to drop for both solution control and bead-immobilized templates after 30 min because of both a buildup of reaction products and a loss of rNTPs; during the first 30 min of transcription, 29% of rNTPs were incorporated in immobilized reactions and 39% of rNTPs were incorporated in solutionphase reactions, causing a drop in transcription from 83 to 50 nucleotides template-1 s-1 between 15 and 45 min (Figure 3b). A second set of experiments with a lower DNA concentration of [DNA] = 0.1 ng/μL and packing density of γ = 0.46 and β = 12 gave linear kinetics by increasing the ratio of rNTPs to transcribed DNA bases from 130 to 47 000 (Figure 3c), thus removing the reagent-limiting conditions and boosting kinetics to 237 ( 20 and 152 ( 15 nuleotides template-1 s-1 for solutionphase and immobilized transcription, respectively (Figure 3d). Even when reagents are not a limiting factor, immobilized transcription was significantly lower than solution-phase transcription in both high and low [DNA] regimes. The following experiments investigate potential mechanisms for this phenomenon, including a local decrease in the ratio between reagents and template molecules, mass transport limitations of reagents to the bead surface, and steric effects from surrounding molecules. DNA = 0.1 ng/μL was chosen for subsequent experiments to ensure linear kinetics. Effect of DNA-Packing Density and DNA Length on Transcription. Two different lengths of cDNA bound in a range of γ = 0.32-1.7 and β = 3.7-30 were used as templates in 45 min transcription reactions with total [rNTP] = 7 mM. DNA gel shown in Figure 2a confirms the sizing; lane 1 contains an H5 fragment (502 bp, ∼47 nm), and lane 2 contains the full-length H5 cDNA (1814 bp, ∼107 nm). Results of these experiments are shown in Figure 4. Full-length immobilized DNA templates produce 61 ( 8% RNA yield of that obtained with their solution-phase counterparts, regardless of the value of γ (Figure 4b); a Fisher analysis of variation (ANOVA) test showed no significant difference ( p < 0.51) in normalized transcription efficiency over the range of γ = 0.32-1.1 and (31) Robertson, R. M.; Laib, S.; Smith, D. E. Proc. Natl. Acad. Sci. U.S.A. 2006, 103(19), 7310–7314. (32) Truskey, G. A.; Yuan, F.; Katz, D. F. Transport Phenomena in Biological Systems; Pearson/Prentice Hall: Upper Saddle River, NJ, 2003.

Langmuir 2009, 25(11), 6168–6175

Article

Figure 2. Gel electrophoresis to confirm DNA and RNA sizing and integrity. (a) Agilent DNA 7500 pseudo-gel of the H5 cDNA used in this study: lane 1, 502 bp H5 cDNA fragment; lane 2, 1814 bp fulllength H5 cDNA. (b) Representative Agilent RNA 6000 pseudogel of RNA product from [DNA] > 0.1 ng/μL experiments: lane 1, solution control (1776 nt); lane 2, immobilized transcription (1776 nt). (c) Representative Agilent RNA 6000 pseudo-gel of RNA product from [DNA] = 0.1 ng/μL experiments: lane 1, short solution phase (464 nt); lane 2, short immobilized (464 nt); lane 3, long solution phase (1776 nt); lane 4, long immobilized (1776 nt). DOI: 10.1021/la804144s

6171

Article

McCalla et al.

Figure 3. IVT kinetics. (a) Kinetics in NTP-limiting conditions. The rate of RNA production drops after 30 min when using 2 mM of NTP, 6.2 ng/μL of full-length H5 cDNA, and a high packing density of γ = 0.15 and β = 4. The bead and solution-phase kinetics follow the same trend, but the rate of reaction for bead-immobilized IVT is consistently lower. (b) Same kinetics shown in a, graphed as a nucleotide incorporation rate over time per template. This graph shows the steep decrease transcription rate between 15 and 45 min in nucleotide-limiting conditions. (c) Linear transcription kinetics are achieved in both the bead and solution phase when NTP is at an excess over DNA. Beads had a packing density of γ = 0.46 and β = 12. (d) Same kinetics shown in c, graphed as nucleotide incorporation rate over time per template.

β = 8.5-30. Short 502 bp cDNA fragments do not follow this trend, however; an ANOVA test shows significant differences over varying γ ( p < 0.011) (Figure 4c). Data points corresponding to tightly packed fragments (γ = 0.25 and 0.32, β = 3.0 and 3.7) have a significantly lower RNA yield when compared to the average yield of more sparsely bound templates (γ = 0.54-1.7, β = 6.4-20) ( p < 0.0012, Bonferroni two-tailed t test). When γ g 0.54 and β g 6.4, no significant difference was found between transcription efficiency of short immobilized cDNA fragments (65 ( 10%) and full-length cDNA (students t test, p < 0.1). This result implies that steric effects due to crowding of neighboring molecules do not play a significant role in the decrease of immobilized transcription efficiency unless molecules are small enough to reach a critical packing density (β e 3.7). Transcription reactions using full-length H5 cDNA were also run using total [rNTP] = 2 mM of each nucleotide (Figure 4a). Immobilized transcription efficiency is still independent of the cDNA packing density as represented by γ and β ( p < 0.49), but the overall efficiency has dropped significantly to 40 ( 10% of the solution control (two-tailed student t test, p < 0.000 01). 6172

DOI: 10.1021/la804144s

These combined results imply that the drop in immobilized efficiency is largely substrate-dependent. Substrate Dependence of Immobilized Transcription. Full-length H5 cDNA immobilized with a packing density of γ = 0.46 and β = 12 was transcribed concurrently with solutionphase H5 cDNA under varied [rNTP] for 45 min to investigate the substrate dependence of immobilized transcription reactions (Figure 5). Both solution- and bead-phase transcription showed characteristics of Michaelis-Menten kinetics, where the reaction rate increases with [rNTP] until substrate concentrations reached a critical level of [rNTP] ≈ 6 mM or a ratio of rNTP to transcribed DNA bases of 41 000 (Figure 5a). The template immobilized reaction yield relative to the solution control yield increases with increasing [rNTP], reaching a stable 57 ( 6% above 4 mM rNTP (Figure 5b), confirming the results seen in Figure 4. Immobilized Transcription is a Reaction-Limited Process. Substrate dependence of the immobilized transcription reaction seen in parts a and b of Figure 4 and Figure 5b could be a mass-transport limitation caused by diffusion-limited delivery Langmuir 2009, 25(11), 6168–6175

McCalla et al.

Article

of the substrate to the highly concentrated DNA on the bead surface. The mass transport limitation can be investigated using the Thiele modulus, defined as the ratio between the reaction rate (1/k[DNA]) in the absence of mass-transfer limitations to the rate of substrate diffusion to the bead surface (L2/D) or φ ¼

rffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi L2 k½DNA D

ð6Þ

where L is the diffusion length, k is the rate constant in the absence of mass transport limitations, and D is the diffusion coefficient.32,33 Thus, when the Thiele modulus is small, the rate of substrate transport to the bead surface is large when compared to the rate of the substrate that is depleted near the surface because of the reaction. In this case, the reaction is not hindered by insufficient reagents at the bead surface because of reagent transport limitations. Conversely, when the Thiele modulus is large, the rNTPs are incorporated into the growing RNA chain faster than they can diffuse to the enzyme; therefore, the mass transfer would be rate-limiting. Transcription kinetics of solution-phase cDNA template (Figure 5a) between 0 and 6 mM rNTP can be assumed linear to acquire an apparent rate constant per DNA template in the absence of mass-transfer limitations 1 d½RNA ¼ k½rNTP ½DNA dt

ð7Þ

The apparent rate constant was found to be k = 44 000 [rNTP] [DNA]-1 s-1, and the diffusion coefficient of rNTP was taken as D = 75 μm2/s.34 The length is set to the maximum possible length that the rNTPs would need to travel, which is the full length of the H5 cDNA L = 617 nm. The DNA concentration was found by taking the average moles of DNA found on a single bead and dividing by the volume occupied by the DNA, which gave [DNA] = 1.3  10-7 M for γ = 0.32. Given these parameters, the Thiele modulus was calculated to be φ = 0.005. Because the Thiele modulus is significantly less than one, the rate that the rNTPs reach the bead surface is much greater than the rate that the rNTPs are consumed in the reaction, implying that the substrate dependence of immobilized transcription is not due to mass-transfer limitations.

Discussion

Figure 4. Effect of immobilized DNA density on transcription. (a) Transcription of immobilized 1814 bp DNA relative to solution control, [rNTP] = 2 mM total. Transcription of immobilized samples is 40 ( 10% of the solution control, independent of β or γ ( p < 0.49, ANOVA). (b) Transcription of immobilized 1814 bp DNA relative to the solution control, [rNTP] = 7 mM total. Again, transcription is independent of β or γ ( p < 0.51, ANOVA) at 61 ( 8% of solution control. (c) Transcription of immobilized 502 bp DNA relative to solution control, [rNTP] = 7 mM total. Transcription is independent of β if β > 3.7 but drops significantly at high packing densities ( p < 0.0012, Bonferroni two-tailed t test). The parameter γ does not appear to predict the presence of steric effects, because both the 1814 and 502 bp DNA molecules have a tightly packed γ = 0.32, but only the 502 bp DNA has a drop in transcription efficiency. Langmuir 2009, 25(11), 6168–6175

Immobilization of DNA and RNA onto solid substrates is becoming a commonplace element in biological protocols because of a demand for amplification of rare transcripts, often from contaminated clinical samples. Although many researchers have reported successful protocols for amplification reactions using immobilized DNA, there has been very little study of the effect of immobilization on reaction efficiency. Andreadis and Chrisey14 found that immobilized transcription efficiency was ∼80% of the solution control under optimized conditions using 1 kbp DNA templates, which increased as the amount of DNA in the reactions increased. In the above study, transcription using a high DNA concentration was also larger when a greater total amount of DNA was used in the reactions (Figures 3a and 5c); immobilized transcription was 76% of the solution control for reactions using 3.73 ng/μL DNA compared to 40% of solution control for reactions using 0.1 ng/μL DNA when all other solution (33) Tischer, W.; Kasche, V. Trends Biotechnol. 1999, 17(8), 326–335. (34) Batada, N. N. Proc. Natl. Acad. Sci. U.S.A. 2004, 101(50), 17361–17364.

DOI: 10.1021/la804144s

6173

Article

McCalla et al.

Figure 5. Substrate dependence of transcription. (a) Substrate dependence on the transcription rate of full-length (1814 bp) H5 cDNA free in solution. In the range of [rNTP] = 0-6 mM, k = 44 000 [rNTP] [DNA]-1 s-1. (b) Transcription of the beadimmobilized template relative to solution control. These results confirm that immobilized transcription is substrate-dependent, as found in parts a and b of Figure 4. The line is for illustrative purposes only.

conditions remained the same. We argue that this effect is possibly due to the nonlinear transcription kinetics in reagent-limiting (high [DNA]) conditions seen in Figure 3b; the solution-phase transcription rate slows more quickly, allowing immobilized transcription reactions to approach solution-phase yield. Ghosh et al.25 also reported a large drop in immobilized transcription efficiency when using 39.34 kbp DNA, which they credited to two different types of steric effects based on the immobilization scheme: interaction of the DNA molecule with the immobilized surface for single DNA molecules covalently linked to beads and overcrowding around the promoter site for DNA tightly packed on glass slides. Although our DNA is shorter and may exhibit different characteristics, we investigated similar steric effects as a main cause in the decrease of immobilized transcription efficiency for long, clinically relevant viral cDNA molecules. There are several different types of steric effects under investigation in our study. The first type of steric inhibition is enzyme interaction with the bead surface, which was ruled out by the polymer spacer and nucleotides upstream of the promoter6174

DOI: 10.1021/la804144s

binding site (Figure 1). Additional nucleotides upstream of the promoter-binding site did not yield more efficient transcription (data not shown). The second type of steric inhibition is polymerase interaction, with a large network of bound DNA molecules blocking the access to the bead surface or hindering polymerase extension, which can be measured by γ. Remarkably, neither a variable free magnesium concentration (see the Supporting Information) nor γ had a significant effect on immobilized transcription efficiency for long DNA templates. The radius of gyration of DNA is strongly affected by neighboring DNA molecules because of electrostatic repulsions between polyanionic backbones.30,31 This may cause DNA to lengthen into “brush mode” when tightly packed on bead surfaces (γ < 1), where the DNA stretches out to avoid contact with neighboring molecules. Conversely, when γ > 1 DNA may begin the transition into “mushroom mode”, where the DNA will spread out to its full radius of gyration.18 The parameter γ did not appear to be a good predictor of steric inhibition, implying both that DNA did not fully collapse to a configuration that blocked its promoter at low binding densities and that neighboring molecules in brush formation stretched far enough to prevent enzyme interaction with neighboring DNA/enzyme molecules at most of the high binding densities. The final type of steric inhibition investigated is due to close DNA spacing that hinders enzyme access to the promoter site and enzyme progress along the DNA strand regardless of the configuration of the DNA; an effect that is measured by the parameter β. When packing density reaches below a critical level around β < 3.7, steric effects can be seen; this is most likely due to enzyme interaction with neighboring strands and enzymes as well as decreased access of the enzymes to the promoter site due to the close DNA molecular spacing. This packing density was only attained using shorter cDNA molecules. These results imply that the drop in reaction efficiency is not due to steric effects for packing densities of β > 3.7. Even when steric effects from surrounding molecules do not affect the relative immobilized transcription yield, immobilized transcription reactions produce ∼60% of solution control reactions under optimal substrate conditions for both DNA molecules used in this study. When DNA is immobilized to a solid substrate, the resulting reaction has much higher local DNA concentrations. The low calculated Thiele modulus rules out mass-transfer limitations because of a high local substrate sink; the diffusivity of reagents through the DNA layer would have to be decreased by ∼105 for the rate of substrate diffusion to be on the order of the reaction rate. The lack of a significant change in relative immobilized transcription under most conditions with varying magnesium and γ does not support such a drastic change in substrate diffusivity through the DNA layer. One possible reason for this change in immobilized transcription efficiency is a large change in the kinetics of rNTP incorporation because of the high local apparent DNA concentration around the bead surface. The DNA surrounding the beads would sequester free enzyme, creating a higher local enzyme concentration. We found that solutionphase transcription is independent of the enzyme concentration for the concentration range used here (data not shown) and is instead dependent upon the rNTP concentration (Figure 5a). This implies that rNTPs are the limiting reagent, such that a decrease in the ratio between rNTP molecules and enzyme molecules could slow the reaction. In solution-phase reactions, the higher [rNTP]/ [enzyme] ratio would drive faster rNTP binding to the enzyme/ DNA complex. Immobilizing DNA to a solid substrate changes the local concentration of DNA by 2 orders of magnitude, while the various binding densities used only span 1 order of magnitude. Longer DNA molecules can have multiple enzyme molecules Langmuir 2009, 25(11), 6168–6175

McCalla et al.

Article

transcribing the DNA at a time, which may explain why the drop in immobilized transcription efficiency is larger when using longer DNA molecules. Another possible mechanism of the drop in immobilized transcription efficiency is the decrease in the probability of reagent interaction by limiting the pathways available to the enzyme and rNTP. Substrate and enzyme can approach a DNA molecule free in solution from a full 360°, while reagents approaching an immobilized molecule are limited to at most a 180° approach. This could also contribute to the observed substrate-dependent drop in immobilized transcription efficiency.

transfer, both often cited as problematic in immobilized reactions, are not the main cause of decreased immobilized transcription efficiency, although the relative space between DNA molecules available for the RNA polymerase can be an important factor for the efficiency of immobilized amplification reactions. For maximal transcription efficiency, binding densities of β > 3.7 and high substrate/DNA ratios should be used. These observations can inform the design of assays and reactors that employ immobilized templates during RNA and DNA amplification by allowing researchers to choose proper binding densities and substrate concentrations for different DNA lengths.

Conclusions Immobilized transcription reactions are subject to a substratedependent drop in reaction efficiency. Surprisingly, unless the DNA molecules can become very tightly packed and the radius surrounding the DNA molecules is on the order of the polymerase diameter, the packing density and flexibility of DNA do not appear to effect transcription efficiency. When these beads are dispersed in solution, mass transfer is also not limiting the reaction and reagents such as rNTPs and polymerase molecules can thus diffuse rapidly through a dense layer of bound DNA molecules under most conditions. Steric hindrance and mass

Langmuir 2009, 25(11), 6168–6175

Acknowledgment. We thank Dr. Matthew Kirby for his guidance and insight. The authors acknowledge the financial support of the National Science Foundation (Grant BES0555874) and a NASA Rhode Island Space Grant for this research. Supporting Information Available: Full experimental details of the effect of the magnesium concentration on immobilized transcription. This material is available free of charge via the Internet at http://pubs.acs.org.

DOI: 10.1021/la804144s

6175