Biofechnol, Prog. 1995, 11, 393-396
393
RNA Transcription *om Immobilized DNA Templates H. Anthony Marble*and Robert H. Davis*,? Departments of Chemical Engineering and Chemistry and Biochemistry, University of Colorado, Boulder, Colorado 80309-0424 We describe a n RNA transcription protocol based on the multiple reuse of solid-phase synthetic DNA templates. The templates are assembled onto streptavidin-coated agarose beads via a single 5’-terminal biotin located on the noncoding template strand. Transcription occurs in a n aqueous buffered suspension containing solid-phase DNA, dissolved enzyme (T7 RNA polymerase), and nucleoside triphosphate substrates (NTPs). A direct comparison of solution and solid-phase templates under standard transcription conditions reveals similar initial reaction rates and overall yields. Immobilized templates store stably for periods of several months and are easily recovered by mild centrifugation. We demonstrate the successive reuse of these templates throughout 15 rounds of transcription. The templates remained active, although an incremental decay in transcription was observed beyond five rounds. Template activity was partially restored by supplementing the support-bound oligonucleotide with fresh coding-strand DNA, These findings indicate that multiple reuse of template is a viable strategy for reducing the amount of DNA template required in RNA transcription.
Introduction Recent scientific discoveries have demonstrated that ribonucleic acid (RNA) molecules have potential commercial value due to their catalytic and selective ligandbinding properties (Cech, 1987; Altman et al., 1988; Tuerk and Gold, 1990). For example, potential chemotherapeutic applications of RNA include the treatment of hepatitis, immunodeficiency, and other viral diseases (Wu et al., 1989; Sarver et al., 1990; Tuerk et al., 1992). This potential, along with a heightened interest in obtaining suitable amounts of RNA for structural studies, has created the need for engineering strategies to produce RNA in larger quantities than previously possible. RNA molecules may be produced in vitro by organic chemical synthesis (Vinayak et al., 1992)or by enzymatic transcription from synthetic or plasmid DNA templates in solution (Milligan and Uhlenbeck, 1989). Both methods are usually limited to a few milligrams of product, or less, with synthesis generally being most efficient for short molecules and transcription being most efficient for long molecules (Davis, 1995). The current work focuses on transcription. A major limitation of solution-phase transcription methodologies is that the DNA templates are discarded after use in a single batch reaction. Since the DNA templates are not consumed, their disposal is unnecessary. Considering the high material costa associated with DNA preparation, template recovery and subsequent reuse may provide an economical alternative for transcriptional applications. One approach is to immobilize the promoter-containing DNA onto a solid support matrix and conduct transcription in a buffered, aqueous suspension containing enzyme and substrates. Arias and Dynan (1989) immobilized biotinylated DNA templates to streptavidin-agarose beads and showed that they retained transcription activity using RNA polymerase 11. Recently, Fujita and Silver (1993)reported that immobilized Department of Chemical Engineering. of Chemistry and Biochemistry. To whom correspondence should be addressed.
i Department
DNA templates containing bacteriophage T3 or T7 promoter sequences sustained transcription in the presence of their respective RNA polymerase. Prior to this, Elov et al. (1991) showed T7 RNA polymerase-mediated transcription from solid-phase DNA templates using a flowthrough, column reactor. Despite the limited success obtained by these investigators, transcription applications utilizing solid-phase DNA templates remain largely unexplored. In this report, we demonstrate the multiple reuse of immobilized DNA using a repeated-batch transcription protocol designed to maximize RNA yield and minimize template-related costs.
Materials and Methods Template Design and Architecture. The template that we have focused on (Figure 1) consists of two complementary DNA molecules which hybridize to form a unique 33-base-paired duplex tailed by a singlestranded overhang, containing the desired RNA coding sequence. The chosen template for the majority of experiments reported herein codes for a 28-nucleotide pseudoknot RNA which specifically binds to the human immunodeficiency virus reverse transcriptase (HIV-RT) primer binding site (Tuerk et al., 1992). The sequence transcribes poorly in solution, providing a demanding challenge for the immobilized template construct. A second sequence, coding for a dodecamer (GGCGCUUGCGUC) which transcribes very well from solutionphase DNA, was also used. The duplex is composed of a class I11 T7 consensus promoter sequence (18 bp) downstream of a short leader region (15 bp). Its noncoding strand protrudes a trinucleotide overhang, which is capped with a 5’-terminal biotin residue t o immobilize the noncoding template strand onto the surface of streptavidin-coated agarose beads. The equilibrium dissociation constant of the biotin-streptavidin complex is M (Pierce Immunochemicals, Rockford, IL), only indicating that the noncovalent binding has very high affinity. However, Fujita and Silver (1993) measured the equilibrium dissociation constant of DNA-biotin-streptavidin complexes and reported values several orders of magnitude larger (10-11-10-12 M). The leader region
8756-7938/95/3011-0393$09.00/00 1995 American Chemical Society and American Institute of Chemical Engineers
Biotechnol. Prog., 1995, Vol. 11, No. 4
394 STREPTAVIDIN PROTEIN
-
T7 PROMOTER
OVERHANG
A
&
TCTGATTCAGCTAGTCCA TAATACGACTCACTATAG CTAAGTCGATCAGGT ATTATGCTGAGTGATATC CCAGGCTTCACGTTGCCCTTTTACGTGA
SPACER
AGAROSEBEAD (not to scale)
CODING SEQUENCE
BIOTIN
Figure 1. Schematic of the template immobilization strategy and the sequence of bases for the immobilized DNA coding for a 28-mer RNA pseudoknot.
increases helical stability and acts to bridge the biotinylated template from the support. Although the noncoding strand is immobilized, the coding strand remains dissociable under helix-destabilizing conditions; this modularity may promote the potential reuse of the support with different coding sequences. DNA Immobilization. Synthetic oligonucleotides (5'biotin and unmodified)were prepared using conventional phosphoramidite chemistry. The noncoding template strand (7.5 nmol of purified 5'-biotin DNA 36-mer) was added to a 1.0 mL solution containing 50% v/v streptavidin-coated agarose beads (Pierce Immunochemicals) and incubated overnight at 4 "C with gentle mixing. The supernatant was removed, and the beads were washed successively (4 x 0.5 mL) with sterile, deionized water. Unbound DNA present in the original supernatant and in the subsequent washes was quantified spectrophotometrically at 260 nm in order to determine the amount of 5'-biotin DNA remaining on the support. Typically, 5-7 nmol of DNA remained bound, corresponding to a DNA concentration of 10-14 pM. Double-stranded DNA templates were formed by incubating (overnight at 4 "C with gentle mixing) the aforementioned support-bound oligonucleotide with a 50%molar excess of the purified RNA-coding strand (61mer). Following DNA hybridization, the support was washed (3 x 0.5 mL sterile, deionized water) and quantified for the presence of unbound DNA. The beads were then resuspended in sterile, deionized water to a final volume of 1.0 mL, yielding a 50% vlv stock mixture. Solid-phase templates were stored at 4 "C with 0.02% sodium azide (preservative) until further use.
Transcriptionwith Solution-Phase and Immobilized Templates. Transcription was assayed in vitro by measuring the incorporation of radioactively labeled uridine triphosphate ([a-32PlUTP,800 Cilmmol, DuPont/ NEN, Wilmington, DE) into RNA products at 37 "C. Unless otherwise specified, each reaction (0.1 mL) contained 40 mM Tris (pH Sal),20 mM MgC12, 5 mM DTT, 1 mM spermidine hydrochloride, 0.01% triton X-100, 1 mM ATP, 1 mM GTP, 1 mM CTP, 0.1 mM UTP, 0.125 pM [a-32PlUTP,1 pM DNA template (10 pL of DNAstreptavidin-bead complex), and 0.75 pM (75 pg/mL) T7 RNA polymerase. Otherwise identical reactions were carried out with 1 pM nonbiotinylated solution-phase DNA template; these reactions included beads without immobilized DNA. Note that the UTP concentration was chosen to be lower than that of the other NTPs in order t o improve the fractional incorporation and, hence, the signal-to-noise ratio of the radiolabel. Solid-phase templates were kept suspended in the transcription reaction by employing a rotary mixer (10 rpm) to avoid settling. An initial sample (5 pL) was removed from each reaction prior to the addition of enzyme. Time points (5 pL reaction aliquots) were removed at successive intervals, quenched on ice with deionized formamide, and loaded
onto 20% (19:l) denaturing (8 M urea) polyacrylamide gels for electrophoresis (600 V, 15 mA for 3 h). Gelfractionated RNA products were visualized, and the radioactivity was quantified using a phosphorescent imager system (Molecular Dynamics, Sunnyvale, CA). Template Reusability Study. Solid-phase DNA templates were reused throughout the course of 15 successive in vitro transcription reactions (0.1 mL), using the same conditions described above. An initial time point was removed from each reaction prior to the addition of enzyme. Upon addition of T7 RNA polymerase (75 pg/mL, 0.75 pM), samples (5 pL) were removed at defined intervals for electrophoresis and subsequent analysis. After each 2-h transcription cycle, the agarose beads were recovered from the reaction mixture by mild centrifugation (3000g) and careful decanting. The beads were washed (2 x 0.5 mL, decanting after each wash) with sterile, deionized water and sedimented for reuse in subsequent rounds of transcription. Fresh enzyme, NTPs, and buffer were added in the concentrations cited above at the beginning of each transcription cycle. After 15 cycles of in vitro transcription, the templates were washed twice with sterile, deionized water (2 x 0.5 mL) and decanted. The beads were resuspended (0.5 mL of HzO) and heated at 90 "C for 5 min in order to denature the coding strand from the solid-phase biotinylated oligomer. The support (approximately 10 pL of beads) was recovered by centrifugation, decanted, and mixed with 100 pmol(10 pL) of fresh, noncoding DNA (61-mer). The mixture was incubated overnight at 4 "C, encouraging hybridization to the support-bound oligonucleotide. These recharged templates were used for a sixteenth and final round of in vitro transcription.
Results and Discussion Comparison of Solution and Immobilized Templates. A comparison of solution-phase and solid-phase templates for the 28-mer is presented in the autoradiogram in Figure 2. The gel reveals a heterogeneous product distribution. Only a fraction (roughly 10%) of the total UTP incorporated is destined toward the fulllength RNA product. Most of the UTP is consumed in the synthesis of short abortive transcripts, ranging from 2 to 10 nucleotides in length. These products form when nascent transcripts dissociate prematurely from the active ternary complex (Sousa et al., 1992). The length of the desired RNA product is verified by comigration with size-specific markers. The 28-mer product band in Figure 2 is lighter for the immobilized templates than for the solution-phase templates. Figure 3 shows that the initial UTP incorporation rate and the final total yield using the immobilized templates are similar to those observed with templates free in solution. However, the specific RNA product yield (28-mer) obtained with the solid-phase DNA is slightly
Biotechnol. Prog., 1995, Vol. 11, No. 4
I
I
S
S
90'
60'
90'
60' 27
395
28
27
I
S
I
S
120'
150'
I
I
S
Figure 4. Autoradiogram of an electrophoresis gel showing the RNA product distribution from solution-phase (S)and immobilized (I) DNA templates coding for the 12-mer after 60150 min from the start of transcrbtion: the arrow marks the product band. A
0.8
Figure 2. Autoradiogram of an electrophoresis gel showing the RNA product distribution from solution-phase (S)and immobilized (I) DNA templates coding for the 28-mer at 60 and 90 min from the start of transcription; the arrow marks the product band. Two size markers, a 27-mer (27) and a 28-mer (28), are included for reference. Each marker was derived by differential enzymatic labeling of a chemically synthesized RNA 27-mer (either by 5'-32P-terminal phosphorylation or by 3'terminal 32pCpligation) which is identical to the transcription target except for the first two bases on the 5' end. The markers were run in lanes 1,6, and 7, and they show some heterogeneity due to nonperfect coupling during RNA synthesis.
I
I I
0
0 U
z
E c 3 W
2
0.4
0
0.3 -
0
A
30
60 90 TIME (MINUTES)
120
1 so
Figure 5. Fractional incorporation of labeled UTP into all RNA transcripts versus time for successive rounds of batch transcription using immobilized DNA templates; results are shown for rounds 1(0),5 (0),10 (O), 15 (A), and 16 (x).
0
200
100
300
TIME (MINUTES)
Figure 3. Fractional incorporation of labeled UTP into all RNA transcripts (upper curves, with 0 from solution-phase templates and 0 from solid-phase templates) and into the 28-mer RNA product (lower curves, with A from solution-phase templates and v from solid-phase templates) versus time since initiation of batch transcription.
reduced when compared to that for templates free in solution. The data plotted in Figure 3 are the average of two separate experiments. The typical error bars shown in this and subsequent graphs give the ranges spanned by the sets of two measurements. More recent studies in our laboratory with the 12-mer RNA target show no significant differences in the product yield from solution-phase and immobilized templates. As seen in Figure 4, the specific product bands are pronounced relative to the abort bands. This experiment was carried out using optimized NTP concentrations in
Table 1. Total RNA Yield after 1 Hour of Transcription using Immobilized Templates Prepared on 9/29/92 date of reaction yield 0.326 12/07/92 0.238 1/21/93 0.367 3120193 0.331 4/09/93
stoichiometric ratios (8 mM GTP, 6.4 mM CTP, and 4.8 mM UTP), with 40 mM MgC12, 1pM DNA template, and 0.50pM T7 RNA polymerase. The average 12-mer yield of two experiments was 2.2 mg/mL (0.35 fractional UTP incorporation) for both the solution-phase and immobilized templates. Template Reusability. A single batch of solid-phase templates coding for the 28-mer was stored a t 4 "C for several months, with samples being withdrawn periodically for use in transcription reactions having 1pM DNA template and 0.23 pM T7 RNA polymerase. Table 1 shows that the RNA yields (fractional UTP incorporation) from these samples exhibit some scatter but no systematic decrease. This indicates that the immobilized DNA templates remain stable when stored for long periods, without significant loss of transcriptional activity. The immobilized templates used in the studies reported in Table 1were synthesized by Genosys Biotech-
Biotechnol. Prog., 1995, Vol. 11, No. 4
396 3
6 TRANSCRIPTION ROUND NUMBER Figure 6. Cumulative yield of 28-mer RNA product for successive rounds of batch transcription from immobilized DNA templates. nologies (The Woodlands, TX)and contain a five-atom hydrophobic spacer joining the biotin and DNA. New immobilized templates used in all other studies reported in this paper were synthesized by Macromolecular Resources (Fort Collins, CO) and contain a 15-atom amphipathic spacer which is thought to project the biotin further from the DNA and thereby improve immobilization and transcription. Reaction time courses obtained from the reusability study with the immobilized templates containing the amphipathic spacer are depicted in Figure 5. No significant change in transcriptional activity is observed throughout the first five rounds of reaction. The absence of product at time zero in each reaction round confirms that RNA is synthesized de nouo and not carried over from previous rounds. Both the initial rate and the overall yield per round (at 2 h) decrease incrementally after five rounds of reaction. Nonetheless, transcription is robust and the overall RNA yield remains significant. The transcriptional rate does not plateau in the later rounds, suggesting that the yield may be increased by extending the reaction time. Template activity is partially restored by the addition of fresh noncoding strand (round 16). Specific incorporation into the desired product species is presented in Figure 6. This plot portrays the cumulative specific product yield (expressed as moles of RNA 28-mer produced per moles of DNA template) obtained with solid-phase templates throughout successive transcription cycles. The yield for each reaction is determined using the fractional specific incorporation rate measured at 2 h. The rate of product accumulation is constant during the first five rounds and then declines gradually, until round 16 when it is bolstered by the addition of fresh noncoding strand. The cumulative yield of 28-mer after 16 rounds is 13 mol of RNA/mol of DNA, which is substantially greater than the average yield of 1.9 mol of RNA/mol of DNA for solution-phase reactions. One possible source for the decrease in transcription rate observed in the later rounds of reaction is template instability. Since the coding strand is not immobilized, it can dissociate from the support-bound oligomer. Once in solution, the coding strand would be lost during sampling and in subsequent washes. Alternatively, the loss in template activity may simply reflect the ensuing loss of DNA-containing beads throughout the course of the study. While the restoration of template activity is consistent with the former, both factors may contribute to the observed loss in template activity. It is unlikely that the activity loss is related to dissociation of biotinylated DNA from streptavidin-coated support, since the dissociation constant is extremely small. Furthermore, template restoration is not consistent with such an inactivation mechanism.
Conclusions Solid-phase templates sustain T7 RNA polymerase transcription to an extent similar to that of DNA templates in free solution, when compared under standard reaction conditions. Unlike solution-phase DNA, immobilized DNA is easily recovered from the crude transcription reaction mixture for subsequent reuse. DNA removal simplifies the composition of the reaction mixture, ultimately facilitating its purification. Most importantly, support-bound templates show only gradual reduction in activity through multiple cycles of transcription, thereby increasing the overall RNA yield from the templates. Transcriptional activity after many rounds may be partially restored by the addition of fresh codingstrand DNA. Together, these results suggest that immobilized templates are particularly useful for large-scale transcription applications, including continuous-flow and semibatch bioreactor designs.
Acknowledgment This work was supported in part by grants from the National Science Foundation, the Whitaker Foundation, and the Colorado RNA Center. We wish to thank Dr. Steve Schultz and Dr. Art Pardi for providing us with the purified T7 RNA polymerase used in these studies. We also thank Dr. Michael Lochrie (NeXstar Pharmaceuticals, Inc.) for the chemically synthesized RNA 27mer marker.
Literature Cited Altman, S.;Baer, M.; Guerrier-Takada,C.; Vioque, A. Enzymatic cleavage of RNA by RNA. Trends Biochem. Sci. 1988,11, 515-518. Arias, J. A.; Dynan, W. S. Promoter dependent transcription by RNA polymerase I1 using immobilized enzyme complexes. J. Biol. Chem. 1989,264,3223-3229. Cech, T. R. The chemistry of self-splicing RNA and RNA enzymes. Science 1987,236,1532-1539. Davis, R. H. Large-scale oligoribonucleotideproduction. Curr. Opin. Biotech. 1995,6 , 213-217. Elov, A. A.; Volkov, E. M.; Shabarova, Z. A. Synthesis of RNA with the aid of T7 RNA polymerase and immobilized DNA in a reactor of the flow-through type. Bioorg. Khim. 1991,17, 789-794.
Fujita, K.;Silver, J. Surprising lability of biotin-streptavidin bond during transcription of biotinylated DNA bound to paramagnetic streptavidin beads. Biotechniques 1993, 14, 608-617. Milligan, J. F.; Uhlenbeck, 0. C. Synthesis of small RNAs using T7 RNA polymerase. Methods Enzymol. 1989 180,51-62. Sarver, N.;Cantin, E. M.; Chang, P. S.; Zaia, J. A.; Ladue, P. A.; Stephens, D. A.; Russ, J. J. Ribozymes as potential antiHIV-1 therapeutic agents. Science 1990,247,1222-1225. Sousa, R.; Patra, D.; Lafer, E. M. Model for the mechanism of bacteriophage T7 RNAP transcription initiation and termination. J . Mol. Biol. 1992,224,319-334. Tuerk, C.; Gold, L. Systematic evolution of ligands by exponential enrichment: RNA ligands to bacteriophage T4 DNA polymerase. Science 1990,249,505-510. Tuerk, C.; MacDougal, S.; Gold L. RNA pseudoknots that inhibit human immunodeficiency virus type 1reverse transcriptase. Proc. Natl. Acad. Sci. U S A . 1992,89,6988-6992. Vinayak, R.; Anderson, P.; McCollum, C.; Hampel, A. Chemical synthesis of RNA using fast oligonucleotide deprotection chemistry. Nucleic Acids Res. 1990,20,1265-1629. Wu, H.-N.;Lin, Y .-J.;Lin, F.-P.; Makina, S.; Chang, M.-F.;Lai, M. M. C. Human hepatitis 6 virus RNA subfragments contain an autocleavage activity. Proc. Natl. Acad. Sci. U.S.A. 1989, 86,1831-1835.
Accepted May 26, 1995.@ BP9500268 @Abstractpublished in Advance ACS Abstracts, July 1, 1995.
