NANO LETTERS
Ultradense Synthetic Gene Brushes on a Chip
2009 Vol. 9, No. 2 909-913
Amnon Buxboim,† Shirley S. Daube,‡ and Roy Bar-Ziv*,† Department of Materials and Interfaces and Chemical Research Support, The Weizmann Institute of Science, RehoVot 76100, Israel Received December 27, 2008; Revised Manuscript Received January 8, 2009
ABSTRACT Dense brushes of linear DNA polymers are assembled on a biochip with ∼30 nm between anchorage points, amounting to a few mega-basepairs/µm3. In bulk solution, a barrier incurs to conjugate more than two end-functionalized DNAs. However, such doublets bind the surface with almost equal efficiency to singlets, suggesting that extended brush buildup reduces the barrier. On-chip transcription reveals that doublets are roughly 2-fold inefficient compared to singlets, a manifestation of the interaction of the enzymatic machinery with the dense brush. Synthetic gene brushes made of DNA conjugates provide simple means to regulate expression on a chip.
Cell-free gene expression opens possibilities to develop synthetic systems and biochemical materials at the nanoscale,1-17 reviewed in refs 18-20. Such systems could be conveniently realized on a surface by patterning and immobilizing genetic information. Biosynthetic reactions are then compartmentalized, allowing for spatial propagation of products in the solution bathing the surface.3,16,17 Successful implementation of spatial gene expression systems on a surface requires understanding the elementary unit of expression, namely, a dense region of immobilized DNA and its interaction with the transcription machinery at distances of ∼30 nm between chains. Recently, we investigated such synthetic gene brushes and observed a structure-function relationship of expression on a chip.16,17 Immobilized genesslinear DNA polymers typically 2000 base pairs longsform extended brushes at close packing below the DNA persistence length (∼50 nm). At intergene distances of 30-100 nm, we found that transcription is highly sensitive to brush density, gene orientation, and DNA composition. The sensitivity of transcription to brush structure provides primitive forms of regulation and control on the chip. Here we extend the study to focus on the very dense regime where the local DNA concentration is of the order of a few mega base pairs per µm3, which is comparable to the mean density of the bacterial chromosome, and is 10-fold denser than DNA polymer brushes assembled on microbeads.21,22 We do so by introducing a new variable in the dense brush, namely, the number of DNA polymers conjugated together at their ends. We first investigated the propensity to form DNA conjugates in bulk solution. For DNA polymers of several persistence lengths, we observed † ‡
Department of Materials and Interfaces. Chemical Research Support.
10.1021/nl8039124 CCC: $40.75 Published on Web 01/26/2009
2009 American Chemical Society
a large barrier for conjugating more than two, consistent with a lower barrier observed previously by atomic force microscopy (AFM) imaging of short DNA (150 bp) conjugated to streptavidin (SA).23 Surprisingly, a study of the self-assembly of dense brushes made of mixtures of singly and doubly conjugated DNA polymers revealed that both pack on the surface almost equally well. That is, the barrier to form triplets in solution did not exist upon surface binding, when the third group to bind was a surface biotin. Finally, we carried out on-chip transcription from these brushes and observe that high density is sensed by the enzymatic machinery with a reduction in efficiency due to doubly conjugated DNA. To assemble synthetic gene brushes, we use a photolithographic biochip platform recently developed.16 Briefly, a photolithographic biocompatible monolayer is formed on a silicon dioxide surface in a single step. The monolayer comprises a hybrid molecule (we termed “daisy”) that has a poly(ethylene glycol) backbone with a silicon dioxide binding group and a protected amine at its two ends. UV light deprotects the amines, to which biotins are then chemically attached for subsequent localization of SA conjugated linear dsDNA polymers (SA-DNA). The number of deprotected amines and hence surface biotins can be controlled by the UV flux up to the limit of ∼1.6 nm per amine, thereby setting the limit of surface binding capacity. The distance between SA-DNA reported here ∼30 nm is larger than the size of 5 nm SA and is hence dictated by DNA polymer packing rather than by surface binding capacity (Figure 1). Prior to assembly of DNA brushes, we first explored the spectrum of conjugation states of SA and linear DNA polymers in bulk solution. Linear dsDNA of lengths 420, 1200, and 2100 bp were prepared by standard PCR with a
Figure 1. DNA conjugation and gene brush schemes (not to scale). (Left) Singlet, doublet, triplet, and quadruplet formed by conjugating end-functionalized (red circles) DNA polymers to SA (green). (Right) Brush assembly by conjugating SA-DNA to surface biotins on a biocompatible surface (yellow). Transcription in the dense brush carried out by RNA polymerase (blue) synthesizing nascent mRNA (red lines). Typical dimensions: SA and T7 RNA polymerase ∼5-10 nm; DNA end-to-end ∼150 nm; SA conjugation distance on the surface ∼30 nm.
single biotin attached at their 5′-end. Increasing amounts of SA were conjugated to DNA for several hours at a fixed DNA concentration (120 nM) with molar ratios of DNA to SA ranging from 4/1 down to 1/5. The conjugated mixtures were analyzed using gel electrophoresis. With each SA capable of binding four biotins, we indeed resolved the free dsDNA from the four conjugation states with one to four DNAs bound to SA, henceforth termed singlet, doublet, triplet, and quadruplet conjugates (Figure 2a). At each [DNA]/[SA] ratio a variety of configurations was obtained and the relative intensity of the gel bands reflected the conjugate populations, which changed with the ratio. Expectedly, in excess of SA over DNA, only free DNA, singlets, and doublets formed with the fraction of singlets gradually decreasing with increase of [DNA]/[SA] ratio. In excess of DNA, bands of triplets and even quadruplets appeared (Figure 2a), although at very low amounts, suggesting that binding up to four DNA polymers to a single SA was physically possible yet very inefficient. We measured the band intensity of each DNA conjugate and, knowing the amount of SA used, deduced the population of the conjugates including the free unmarked SA in each sample. To check whether all the possible configurations were realized with equal probability, or whether joining the ends of DNA polymers presents a barrier, we compared the data against a calculation of the binomial distribution. Since the binding of SA to biotinylated DNA is irreversible, the probability of a single binding is just p ) [DNA]/4[SA] < 1. We assume that binding an additional DNA occurs with equal probability, that is, a single DNA polymer bound to SA does not inhibit or enhance binding of an additional DNA, and so forth. Therefore, accounting for the number of different ways of distributing k DNAs in four binding sites, the probability for SA with k occupied biotin sites and (4 - k) free biotin sites per SA is just: pk ) 4!/k!(4 - k)!·pk·(1 - p)4-k. The experimental conjugate and binomial distributions are both plotted in Figure 2b for 1200 bp DNA with [DNA]/[SA] ) 3 and 1/3; all the sample distributions for all three DNA lengths are given in the Supporting Information. As exemplified for [DNA]/[SA] < 1/3, when SA is in excess (Figure 2b), the data are well represented by the binomial distribution 910
Figure 2. DNA-SA conjugation in solution. (a) Gel electrophoresis analysis of 1200 base pairs DNA conjugated to SA at varying ratios of [DNA]/[SA]. The SA-free DNA, singlet, doublet, triplet, and quadruplet appear as distinct bands. (b) The distribution of conjugates as deduced from the gel and the calculation of the binomial distribution for [DNA]/[SA] ) 1/3 and 3. (c) The mean conjugation of 420, 1200, and 2100 bp DNA and the binomial calculation as a function of [DNA]/[SA] ratio.
(see also Supporting Information). However, when DNA is in excess, for example [DNA]/[SA] ) 3, triplets and quadruplets appear with significantly reduced likelihood than would have been expected if binding of DNA polymers to SA was independent (binomial distribution). The mean SA conjugation of each distribution represents the propensity for conjugating DNA polymers to SA, shown in Figure 2c for all three lengths as a function of [DNA]/ [SA] ratio. The quantitative results support the qualitative gel picture, showing that for DNA polymers, the formation of triplets and quadruplets, while physically permitted, is strongly disfavored with the mean conjugation saturating at a value of 2 for all lengths. An attempt to overcome this barrier with up to 10-fold excess of DNA incubated for 3 days still did not lead to increased formation of triplets. Furthermore, reducing DNA and SA concentration while Nano Lett., Vol. 9, No. 2, 2009
Figure 4. Transcription in a DNA brush. Transcription rate as a function of mean conjugation in solution (a) and on the chip (b).
Figure 3. Singlet and doublet DNA brush assembly. (a) Lanes represent mixtures of mean conjugation (A through H) ranging between 1.15 and 1.85 as in (c) and (d). (b) Fluorescent image of DNA chip (18 × 18 mm2) brush; mostly singlet solution labeled by Alexa 647; mostly doublets by Alexa 488. (c) Surface density of singlets, doublets, total DNA, and SA on the chips in molecules per µm2. (d) Mean conjugation on the chip as a function of mean conjugation in solution.
retaining their ratio, or imposing crowding conditions with 0.2 mg/mL circular DNA or 10% poly(ethylene glycol) did not improve triplet/quadruplet formation. In addition, the possibility that some of the SA molecules were physically unable to bind three and four polymers was ruled out by conjugating the SA with a short unstructured single-stranded DNA oligmer (poly-Thymine-10), which showed perfect agreement with the binomial distribution (Supporting InforNano Lett., Vol. 9, No. 2, 2009
mation). To conclude, it appears that joining ends of more than two DNA polymers at a single point in bulk solution is possible but incurs a significant barrier. Presumably, this is a manifestation of SA being a dimer of dimers with four biotin binding sites with similar affinities. However, two binding sites are further apart, leading to conjugation in a trans configuration, while the other two are in close proximity, leading to cis configuration of conjugates and possible steric hindrance.24,25 Specifically for SA-DNA conjugates, it has been previously shown by AFM imaging of 150 bp long DNA:SA doublets that the majority of conjugates obtained had an obtuse angle between the DNA rods, implying a preferential trans binding.23 Apparently, the DNA molecules used hereswith a radius of gyration of 150-200 nmsimpose a greater barrier for binding in cis. The barrier for triplet formation in solution suggested that within a mixture of singlets and doublets only the singlets might assemble into brushes on a surface. Previously, brushes were assembled from bulk solution at ratio [DNA]/[SA] ) 1.5 and without removing possible residues of free SA;16,17 the relative amounts of singlets and doublets on the surface were unknown. In order to directly determine the distribution of singlets and doublets on the surface, we prepared two separate mixtures of SA-DNA conjugates. In one solution we conjugated mostly singlets at [DNA]/[SA] < 1/2 with DNA end-labeled with Alexa-647 and in the other we used [DNA]/[SA] ) 2 ratio where the DNA was end-labeled with Alexa-488 and obtained mostly doublets. To eliminate possible blocking of the surface upon grafting, free SA was filtered out from the first conjugates solution (see Methods). Eight mixtures of SA-DNA conjugates with varying mean 911
conjugation, ranging between 1.15 and 1.85, were prepared by mixing the two conjugate solutions at different ratios (Figure 3a). From these singlets/doublets mixtures, eight corresponding DNA brushes were assembled (Figure 3b),16,17 and their density composition was quantified against a calibration curve (not shown) by scanning the fluorescence of bound DNA in the two color channels after washing off unbound DNA (Figure 3a,b). As a function of the bulk solution mean conjugation, the density of surface-bound SA remained roughly constant (Figure 3c), the mean inter-SA distance varying between 31 and 36 nm. However, the total DNA density increased from 800 to 1900 polymers/µm2, by a gradual replacement of doublets by singlets (Figure 3c). In fact, the value of the mean conjugation of the chip was only slightly reduced compared to that in bulk solution (Figure 3d), indicating that singlets were only mildly more reactive to surface binding than doublets. This is in sharp contrast to the solution conjugation result (Figure 2) where the third biotin binding site was not accessible. Control experiments in which SA-DNA conjugates were incubated with daisy surfaces lacking biotins resulted in very low nonspecific adsorption,16,17 suggesting that indeed doublets were bound to the surface via biotin. We conclude that upon buildup of a dense DNA brush, with 30 nm between SA anchoring points, the loss of DNA polymer degrees of freedom exposes the third SA biotin site of a doublet and reduces the barrier incurred for triplet formation in solution. The picture emerging from our previous results on brush buildup where DNA extension follows DNA grafting on the surface provides a possible mechanism for alleviating the barrier for surface binding of doublets. We propose that on the surface the DNA is extended, thus exposing the third binding pocket of SA to freely bind the biotin on the surface. Finally, we probed the effect of doublets on the transcription rate at this highly dense brush regime. The DNA polymers included a sequence of T7 RNA polymerase promoter, located 60 bases from the conjugated SA end, followed by 900 bases of transcript, with transcription directed outward from the surface. Transcription was carried out by immersing the chips with a reaction mix in an open unbound configuration.17 We monitored the on-chip transcription by measuring the radioactive nucleotide incorporation into nascent RNA released to solution. This was carried out on the same DNA brushes used to probe density. In contrast to homogeneous solution reactions (Figure 4a), the transcription rate per gene decreased linearly with the mean conjugation in the dense brushes (Figure 4b). These data extend our previous finding where the transcription rate was observed to decrease at high density in brushes of fixed mean conjugation (1.2). To summarize, we studied the assembly of very dense gene brushes using SA-conjugated DNA polymers. We find that joining the ends of more than two linear polymers at a single point in bulk solution incurs a barrier, likely due to polymer degrees of freedom. However, upon brush buildup on the surface this barrier is alleviated, providing simple means to increase the surface DNA density. Concomitantly with the high DNA density transcription efficiency is reduced. This 912
results in a synthetic DNA brush with a density of ∼4 Mbp/ µm3 and may imply that transcription efficiency is hampered under crowding conditions of condensed and orderly aligned DNA in a cell. At high densities, despite the increase in the number of promoters, the limiting factor of transcription may be the concentration of RNA polymerase that has been shown to be confined within layers in the brush,17 thus leading to an observed reduction in transcription efficiency. In addition, the close proximity of promoters in the doublet configuration that dominates the ultradense region may lead to reduction of transcription. Finally, the gradual density increase obtained by exchanging singlets by doublets may emulate structural changes in protein-induced packaging of DNA in vivo. Such fine-tuning of packaging is known to regulate gene expression by allowing certain regions to “breathe” and enable transcription factors to interact with the DNA. High-density DNA brush build-up could be tailored to different purposes: use only singlets when high transcription levels are required; alternatively, use doublets for those purposes where extremely high densities are required but transcription does not play a significant role. Methods. End-functionalized DNA was prepared by standard PCR protocols using forward and reverse primers that were 5′-biotinylated and 5′-alexa fluorophore endfunctionalized (Integrated DNA Technologies, USA). Conjugation of SA to biotinylated DNA was as described in ref 17. Excess SA was removed by using 100 kDa cutoff centrifugal filters (YM-100 Microcone, Millipore) at 1000g. Brush assembly and transcription were done as described in ref 17. Acknowledgment. We wish to thank D. Bracha, O. Krichevsky, S. Safran, and G. Shemer for useful discussions. This work was supported by the Israel Science Foundation and by the Minerva Foundation. Supporting Information Available: Gel electrophoresis analysis of DNA conjugated to SA at varying ratios of [DNA]/[SA]. This material is available free of charge via the Internet at http://pubs.acs.org. References (1) Szostak, J. W.; Bartel, D. P.; Luisi, P. L. Nature 2001, 409 (6818), 387–390. (2) Shimizu, Y.; Inoue, A.; Tomari, Y.; Suzuki, T.; Yokogawa, T.; Nishikawa, K.; Ueda, T. Nat. Biotechnol. 2001, 19 (8), 751–755. (3) Shivashankar, G. V.; Liu, S.; Libchaber, A. Appl. Phys. Lett. 2000, 76 (24), 3638–3640. (4) Pohorille, A.; Deamer, D. Trends Biotechnol. 2002, 20 (3), 123–128. (5) Hanczyc, M. M.; Fujikawa, S. M.; Szostak, J. W. Science 2003, 302 (5645), 618–622. (6) Noireaux, V.; Bar-Ziv, R.; Libchaber, A. Proc. Natl. Acad. Sci. U.S.A. 2003, 100 (22), 12672–12677. (7) Nomura, S.; Tsumoto, K.; Hamada, T.; Akiyoshi, K.; Nakatani, Y.; Yoshikawa, K. ChemBioChem 2003, 4 (11), 1172–1175. (8) Dittmer, W. U.; Simmel, F. C. Nano Lett. 2004, 4 (4), 689–691. (9) Kim, J.; Winfree, E.; Hopfield, J. AdVances in Neural Information Processing Systems (NIPS) 2004, 681–688. (10) Noireaux, V.; Libchaber, A. Proc. Natl. Acad. Sci. U.S.A. 2004, 101 (51), 17669–17674. (11) Dittmer, W. U.; Kempter, S.; Radler, J. O.; Simmel, F. C. Small 2005, 1 (7), 709–712. (12) Isalan, M.; Lemerle, C.; Serrano, L. PLoS Biol. 2005, 3 (3), 488–496. (13) Noireaux, V.; Bar-Ziv, R.; Godefroy, J.; Salman, H.; Libchaber, A. Phys. Biol. 2005, 2 (3), P1–8. Nano Lett., Vol. 9, No. 2, 2009
(14) Kim, J.; White, K. S.; Winfree, E. Mol. Syst. Biol. 2006, 2, DOI 10.1038/msb4100099. (15) Daube, S. S.; Arad, T.; Bar-Ziv, R. Nano Lett. 2007, 7 (3), 638–641. (16) Buxboim, A.; Bar-Dagan, M.; Frydman, V.; Zbaida, D.; Morpurgo, M.; Bar-Ziv, R. Small 2007, 3 (3), 500–510. (17) Buxboim, A.; Daube, S. S.; Bar-Ziv, R. Mol. Syst. Biol. 2008, 4. (18) Jungmann, R.; Renner, S.; Simmel, F. C. HFSP Journal 2008, 2 (2), 99–109. (19) Doktycz, M. J.; Simpson, M. L. Mol. Syst. Biol. 2007, 3, DOI 10.1038/ msb4100165. (20) Simpson, M. L. Mol. Syst. Biol. 2006, 2, DOI 10.1038/msb4100104.
Nano Lett., Vol. 9, No. 2, 2009
(21) Kegler, K.; Salomo, M.; Kremer, F. Phys. ReV. Lett. 2007, 98, 058304. (22) Kegler, K.; Konieczny, M.; Dominguez-Espinosa, G.; Gutsche, C.; Salomo, M.; Kremer, F.; Likos, C. N. Phys. ReV. Lett. 2008, 100, 118302. (23) Neish, C. S.; Martin, I. L.; Henderson, R. M.; Edwardson, J. M. Br. J. Pharmacol. 2002, 135 (8), 1943–1950. (24) Green, N. M. Methods Enzymol. 1990, 184, 51–67. (25) Kaiser, K.; Marek, M.; Haselgrubler, T.; Schindler, H.; Gruber, H. J. Bioconjugate Chem. 1997, 8 (4), 545–551.
NL8039124
913
