Spatially Resolved DNA Brushes on a Chip: Gene ... - ACS Publications

Oct 16, 2009 - Department of Materials and Interfaces, The Weizmann Institute of Science, ... In constructing this cascade, we investigated enzymatic ...
0 downloads 0 Views 3MB Size
NANO LETTERS

Spatially Resolved DNA Brushes on a Chip: Gene Activation by Enzymatic Cascade

2009 Vol. 9, No. 12 4462-4466

Maya Bar and Roy H. Bar-Ziv* Department of Materials and Interfaces, The Weizmann Institute of Science, RehoVot 76100, Israel Received August 23, 2009; Revised Manuscript Received October 5, 2009

ABSTRACT We assemble on a chip spatially resolved DNA polymer brushes with density that can be controlled along continuous gradients, from dilute to dense packing, with 20-30 nm between DNA molecules. Investigation of DNA digestion in a ∼1 kb DNA brush showed that endonucleases can digest within the brush, yet their activity is impeded at high DNA density and for restriction sites near the bottom. Local gene activationsa form of switchswas then demonstrated by a digestion-ligation cascade between two spatially resolved DNA brushes, leading to the in situ expression of green fluorescent protein upon a diffusion-controlled DNA swapping.

Cell-free transcription/translation circuits open exciting possibilities for the bottom-up construction of synthetic biochemical systems with simplified regulation and metabolism, which could advance our understanding of complex functions and networks in a cell.1-7 Such systems may ultimately be combined with DNA nanotechnology to create new functional devices (reviewed in refs 8-11). The basic element of a cell-free circuit is the cascaded propagation of signals.1,12,13 However, there is limited control over the cascading species if genes are expressed from DNA and biosynthetic machinery in bulk solution or in homogeneous gels.14 To improve control, one can localize biosynthesis on a surface at the micrometer scale,7,15 in which case cascading molecules will diffuse between spatially resolved locations.7 A DNA brush, which is comprised of double-stranded DNA polymers attached to a solid support, is an elementary unit for protein expression on the surface.16,17 The density of surface DNA can be controlled during brush assembly, forming a condensed phase of genetic material that mimics the dense environment of the cell, with orientation order induced by the collective stretch of the polymers, at typical distances of 20-30 nm between adjacent polymers, and a local concentration as high as mega-base-pairs per micrometer cubed.16,17 The ability to change the DNA density makes the brush a unique system to examine interactions of DNA with DNAassociated proteins, such as transcription and digestion, under dilute or dense conditions, where the relevant lengths are in the range of 10-100 nm. We have previously demonstrated a two-stage transcription/translation cascade on a chip between two DNA brushes * Corresponding author, [email protected]. 10.1021/nl902748g CCC: $40.75 Published on Web 10/16/2009

 2009 American Chemical Society

that were predetermined to be spatially resolved: A protein synthesized from a gene at one location was the input to drive the expression of a second gene at a different location.7 Here we designed a dual-brush cascade that involves a local enzymatic DNA rearrangement. The cascade is triggered by DNA digestion, which leads to the release and diffusion of DNA fragments from one brush to the other, followed by ligation. Once the correct swapping occurs, the newly ligated DNA brush encodes a complete sequence of the green fluorescent protein (GFP) under a promoter suitable for cellfree transcription/translation. Hence, the result of this diffusion-controlled DNA rearrangement cascade is a localized expression of GFP in a form of a switch. In constructing this cascade, we investigated enzymatic DNA digestion in the brush. For this purpose, DNA brushes with spatially varying density were assembled and then digested as a function of time, in changing DNA length and location of enzyme recognition site along the sequence. To construct a biochip with localized DNA brushes, we use our single-step biocompatible photolithographic interface (described in the Supporting Information).7,16,17 Briefly (Scheme 1), a glass support is coated with a monolayer of a synthesized hybrid molecule (termed daisy) comprised of a chain of poly(ethylene glycol) (PEG) backbone with a silicone dioxide binding group at one end and a photolabileprotected amine at the other end.7 Daisy monolayer is packed with 3 nm2 per PEG molecule.7 The amines are locally deprotected by exposing the daisy-coated surface to 365 nm UV light through an objective lens at designated areas using a small circular aperture placed at the field stop of a fluorescent microscope. Subsequently, biotin is bound to the

Scheme 1. A Schematic Illustration of DNA Brush Assemblya

a (a) A daisy molecule conjugated on a glass slide is illuminated by 365 nm UV light and (b) surface amines are exposed; (c) the amines are linked to biotin-NHS and then (d) to biotinylated DNA through streptavidin.

exposed amines through N-hydroxysuccinimide ester (NHS) coupling. Biotinylated DNA fragments (PCR products generated with one biotinylated primer and another that is fluorescently labeled) are then conjugated to surface biotins through streptavidin. The DNA density can be controlled by tuning the UV flux to deprotect only a fraction of the amines.7 A continuous DNA surface density gradient, illustrated in Figure 1a, can then be generated using a computerized interface that controls the microscope stage position and the UV light shutter. By moving the stage in small steps at a changing rate with open shutter, we can progressively increase or decrease the exposure time of UV through the circular aperture, gradually deprotecting the surface amines along a predetermined trajectory and gradient. Figure 1b shows a fluorescent image of a brush with linearly changed density increasing gradually from zero to the maximum possible value along a 350 µm line. The use of a circular aperture leads to a parabolic density profile in the perpendicular direction, and we hence refer hereafter to its maximal value averaged over 10 µm (Figure 1c). We investigated the effect of brush density on the activity of restriction endonucleases and analyzed the digestion efficiency as a function of the position of the recognition site and the DNA length. We hypothesized that the distance between the DNA molecules in the dense areas of the gradient might impede the enzyme penetration and that this Nano Lett., Vol. 9, No. 12, 2009

Figure 1. (a) A schematic representation of a biochip with gradual distribution of DNA. The red ovals illustrate restriction sites located close to the bottom of the brush. (b) A fluorescent image of a gradient brush with linearly changed DNA density (350 µm length, 33 µm width). The fluorescence is normalized to the maximum value along the line. An illustration of the exposure time used along the chip is presented above. (c) Density gradient of the chip presented in (b). The DNA density is determined by the fluorescence along the brush in the direction of the gradient. Each dot on the graph represents the average of one line of pixels across the brush width over ∼10 µm.

effect can be stronger when the enzyme recognition sites are located deeper in the brush, close to the surface. We prepared a series of DNA brushes of 1130 base pairs (bp) containing recognition sequences located either close to the bottom of the brush (“bottom”) or close to the solution (“top”) (Figure 2a). To verify that our results are not specific to a certain endonuclease and not dependent on the DNA overhangs after digestion, we used two potent and robust type II restriction enzymes with specific six base-pair recognition sites, EcoRV and BamH1, forming blunt-ended DNA segments or four bp overhangs on the digested products, respectively. Both EcoRV and BamHI operate as dimers and depend on Mg2+ cofactors for activity. Furthermore, these enzymes are similar in size (28.7 and 24.6 kDa as monomers, respectively)18,19 and they occupy a similar area around the DNA helix in their bound states (less than 4 nm perpendicular to the DNA helix). A comparison between the activities of these two enzymes in the brush is, therefore, more likely to imply on effects related to the DNA structure rather than on specific differences between the enzymes. To analyze the effect of DNA length, we prepared a set of graded 4463

Figure 2. Digestion of graded-density DNA brushes. (a) A scheme of representative DNA fragments used. The 200-bp brush and the two 1130 bp sequences with restriction sites located close to the solution (“top”) or close to the bottom of the brush (“bottom”) are presented. The number of bp between the restriction sites and the top of the brush are depicted. The star represents the Alexa-fluor-647 fluorophore. (b) Fluorescent images of a 1130 bp brush with restriction site at the top are shown before (top), after 5 min (middle), and after 10 min of digestion (bottom) with EcoRV. (c-e). EcoRV digestion of various DNA brushes. The fraction of cleaved DNA in the course of reaction is presented as a function of DNA density before digestion of a 200 bp brush with bottom restriction site (c), and 1130 bp brushes with top (d) or bottom (e) restriction sites.

density DNA brushes of lengths 200, 440, 1130, and 2160 bp with a bottom recognition sequence for EcoRV. The DNA fragments were labeled with Alexa-647, and the digestion reactions were followed by loss of fluorescence (Figure 2b). The digestion of the 200 bp brush (Figure 2c) was completed within 10 min of reaction and there was no clear effect of density on the enzyme efficiency in a time resolution of 5 min, as evident from the linear dependency between the digested fraction and brush density. This implies that there is little or no inhibition on enzyme penetration into the bottom of the brush for this DNA length. However, when the 1130 bp brushes were digested, the brush density and the position of the restriction sites along the DNA sequences had an apparent effect on endonuclease activity (Figure 2d,e). When the EcoRV recognition site was located at the top (Figure 2d), 10 min of reaction was sufficient to complete the digestion. However, when the restriction site was located at the bottom of the brush, about 30 min was necessary for complete digestion (Figure 2e). The curved dependency between the amount of cleaved DNA and the brush density observed at short incubation times of both top and bottom digestion experiments indicate on a brush density effect. At the dilute areas of the brush, the DNA was completely cleaved within the first 5 min, but above a certain density, the uncleaved/cleaved DNA ratio increased concomitantly with increased brush density. In either case, after approximately 30 min of reaction the DNA was almost completely digested. Similar results were obtained using BamH1 (not shown), implying that the digestion type (blunt or staggered) did not have a considerable effect on the digestion efficiency. The fact that the digestion was complete within ∼30 min regardless of brush density implies that the endonucleases 4464

can penetrate these brushes at all densities. Previous analysis of the effect of DNA density on the efficiency of digestion using 44 bp DNA patches prepared by AFM nanografting showed that below a certain threshold of density digestion does not occur.20 The authors stated that DNA molecules in the dense patches are too close to allow a restriction enzyme to squeeze in.20 It is likely that the maximum density obtained in our brushes is lower than the densities of those patches due to the differences in assembly method and to the 25fold differences in the DNA lengths. Our brushes of 1130 bp sequences are equivalent to 7-8 persistence lengths, with an approximate distance of 20-30 nm between adjacent DNA molecules at the most dense areas.16,17 This distance should be sufficient for a dimer of BamHI or EcoRV to enter.21,22 Nevertheless, the high density of long DNA brushes introduces some hindrance on the enzyme to reach its binding site, and the reaction is slowed down. The observation that digestion is faster at the top of the brush as compared to the bottom suggests that lengthy sequences above the recognition site hamper its accessibility, but not entirely, since with time the entire brush is cleaved even at the bottom. A similar effect of inhibition was observed for brushes of different sequence lengths with an equivalent location of restriction site relative to the bottom of the brush. The digestion of a 200 bp sequence was faster than the digestion of 1130 bp brushes (Figure 2c-e), and the digestion of a 2160 bp brush was slower (not shown). In a study of the effects of the location of a promoter sequence on the efficiency of transcription in a brush, it was shown that when the promoter was located at the top of the brush the product yield was 5 times larger relative to the yield of a brush with a bottom promoter.16 It was also shown that the reactions with bottom promoters reached a maximum Nano Lett., Vol. 9, No. 12, 2009

yield at a density of brush where the distance between adjacent DNA molecules was ∼65 nm.16 Saturation was not observed in our digestion experiments although DNA density reached a distance of ∼30 nm between DNA molecules. We reason that this can be due to the smaller volume necessary for a restriction endonuclease to penetrate and cleave the DNA relative to a transcription process (RNA polymerase is ∼2-fold bigger than a dimer of our restriction enzymes and an extra space is necessary for the nascent mRNA chain). In essence, our findings indicate that although dense brushes and buried restriction sites inhibit endonuclease activity to some extent, the endonucleases can penetrate brushes of all densities down to their very bottom. Having proven that digestion can take place in a dense brush, and since ligation within surface-bound DNA was previously shown on thiolated surfaces,23 we constructed the digestion-ligation cascade. We assembled two separate DNA brushes labeled with different fluorophores (Scheme 2a). We used this two-brush chip to construct a regulatory gene switch in which in situ DNA shuffling transforms the switch from “off” state to “on” state. The first brush was assembled of A-C, a 1045 bp fragment containing the sequence coding for GFP and T7 terminator with a SpeI restriction site upstream, followed by a short random sequence (segment C). The second brush contained fragment B-D (443 bp), consisting of a 343 bp of random sequence (segment B) followed by an XbaI restriction site and a T7promoter (segment D, Scheme 2b). A cut-and-paste of the DNA using XbaI and SpeI endonucleases and ligase could form the mixed gene versions A-D and B-C (Scheme 2c,d) due to the equivalent overhangs of SpeI and XbaI digested products. Although A-C and B-D can be regenerated in this route, formation of A-D and B-C should be favored because of the loss of the XbaI and SpeI restriction sites (Scheme 2d). The generation of the shuffled A-D and B-C sequences is a diffusion-controlled process that depends on the distance between the two sequences on the chip. The switch is activated when an A-D fragment is formed, containing a T7 promoter upstream of the GFP gene from which GFP can be expressed. B-C would be a byproduct. We immobilized a dense A-C brush on a 10 µm radius circle and a dense B-D brush on a ring with 55 and 12 µm external and internal radii, respectively, surrounding the A-C circle over a 30-fold larger surface area (Figure 3a and Supporting Information). The distance between the two genes was approximately 2-3 µm at their closest location. In this design we hoped to maximize the formation of A-D product upon digestion, as the A fragments, which are immobilized to the surface after digestion, would be surrounded by many more free D fragments than C fragments, increasing the chances for an A site to trap a D fragment before it diffuses away from the surface into the bulk. The switch was turned to the “on” state in a single step of digestion-ligation reaction. To conduct this process we added the enzymes SpeI, XbaI, and T4 ligase to the solution on a chip placed on the microscope stage and followed the fluorescence changes directly. After 100 min of reaction the chip was washed and an E. coli cell-free transcription/translation reaction mix was Nano Lett., Vol. 9, No. 12, 2009

Scheme 2. The Gene Switch Turn Ona

a (a) An illustration of the biochip switch with the two DNA brushes before (top) and after (bottom) digestion. The A-C brush (left, 1045 bp) has a recognition site for SpeI (dark gray ovals) and is labeled with Alexa488 (green). The B-D brush (right, 443 bp) has a XbaI restriction site (light gray ovals) and is labeled with Alexa-647 (red). After digestionligation, the A-D segment is red-labeled and ready for transcriptiontranslation of GFP. (b-d) The enzymatic cascade. (b) The “off” state with the inactive genes A-C and B-D. The T7 terminator is marked by T and the T7 promoter is marked by P. (c) DNA translocation in a single step: C and D fragments are digested with SpeI and XbaI, respectively, and they can be ligated to either A or B sequences. (d). The “on” state. Once the chimera gene A-D is formed, the restriction sites are lost. The inactive B-C can also be generated (not shown).

added without moving the chip from the stage. The expression of GFP was followed by increase in fluorescence (see Supporting Information). Figure 3b presents the fluorescence of the two brushes during the digestion-ligation and Figure 3c shows expression of GFP. The digestion of both A-C and B-D was clearly observed by the loss of the signals at 488 and 647 nm, respectively. The synthesis of GFP was observed as a significant increase in the green fluorescence signal (1000-fold), indicating that the A-D gene was successfully formed. The expression reaction reached a maximum after 130 min possibly due to deficiency of raw materials in the cell extract. The ligation of either A-D or B-C did not show up as a gain of fluorescent DNA signal 4465

multiple genes in close proximity. The spatial control demonstrated here provides the opportunity to generate more complex synthetic gene circuits, including cascades and directional transport of information. Gene chips that involve several genes allow generation of feedback and feed forward circuits for protein synthesis and degradation that are regulated at the DNA level. Such constructions introduce fascinating prospects for investigating principles of metabolism and regulation, and maybe even differentiation. Acknowledgment. This work was funded by the Israel Science Foundation and the Minerva Foundation. We thank Yael Heyman, Tsevi Beatus, Gaby Shemer, Yahel Atsmon, and Dan Bracha for their help with planning the DNA gradient chips and the daisy coating. Supporting Information Available: A description of the methods and procedures used in this work. This material is available free of charge via the Internet at http://pubs.acs.org. References

Figure 3. (a) Fluorescence images of the biochip switch before digestion: left, fluorescence of the A-C brush at 488 nm; middle, fluorescence of the B-D brush at 647 nm; right, merged images with the 647 nm signal presented in red and the 488 nm signal in green. (b) Digestion-ligation reaction over time. Initial time is determined as the time when enzymes were introduced to the solution. Most of the A-C sequences were digested during the first 3 min of reaction (green triangles, solid line). The B-D fragments were slowly cleaved (red circles, solid line). There was a decrease in signal at 647 nm also at the A-C site (red triangles, dashed line), and the generation of A-D fragments could not be detected. (c) On-chip expression of GFP before the digestion-ligation (red circles) and after (blue squares), demonstrating that after the biochip was turned to the “on” state, the GFP gene was active. Initial time is when cell extracts were added. A description of GFP concentration calculation is given in the Supporting Information.

above background, possibly due to the presence of some A-C sequences at the location of B-D and vice versa, which reduced the signal-to-noise ratio. To conclude, we demonstrated spatially resolved DNA brushes on a chip, with precise control over DNA location and density with micrometer-scale resolution. Density graded DNA brushes are useful in unraveling the characteristics of dilute and crowded DNA transactions, which is important for the design of gene circuits with DNA gradients and

4466

(1) Noireaux, V.; Bar-Ziv, R.; Libchaber, A. Proc. Natl. Acad. Sci. U.S.A. 2003, 100 (22), 12672–12677. (2) Noireaux, V.; Libchaber, A. Proc. Natl. Acad. Sci. U.S.A. 2004, 101 (51), 17669–17674. (3) Isalan, M.; Lemerle, C.; Serrano, L. Plos Biology 2005, 3 (3), 488– 496. (4) Dittmer, W. U.; Simmel, F. C. Nano Lett. 2004, 4 (4), 689–691. (5) Dittmer, W. U.; Kempter, S.; Radler, J. O.; Simmel, F. C. Small 2005, 1 (7), 709–712. (6) Kim, J.; White, K. S.; Winfree, E. Mol. Syst. Biol. 2006, 10.1038/ ms4100099. (7) Buxboim, A.; Bar-Dagan, M.; Frydman, V.; Zbaida, D.; Morpurgo, M.; Bar-Ziv, R. Small 2007, 3 (3), 500–510. (8) Jungmann, R.; Renner, S.; Simmel, F. C. HFSP J. 2008, 2 (2), 99– 109. (9) Seeman, N. C. Trends Biochem. Sci. 2005, 30 (3), 119–125. (10) Simpson, M. L. Mol. Syst. Biol. 2006, 2. (11) Bath, J.; Turberfield, A. J. Nat. Nanotechnol. 2007, 2 (5), 275–284. (12) Ishikawa, K.; Sato, K.; Shima, Y.; Urabe, I.; Yomo, T. FEBS Lett. 2004, 576 (3), 387–390. (13) Zhang, D. Y.; Turberfield, A. J.; Yurke, B.; Winfree, E. Science 2007, 318 (5853), 1121–1125. (14) Park, N.; Um, S. H.; Funabashi, H.; Xu, J. F.; Luo, D. Nat. Mater. 2009, 8 (5), 432–437. (15) Shivashankar, G. V.; Liu, S.; Libchaber, A. Appl. Phys. Lett. 2000, 76 (24), 3638–3640. (16) Buxboim, A.; Daube, S.; Bar-Ziv, R. Mol. Syst. Biol. 2008, 4. (17) Buxboim, A.; Daube, S. S.; Bar-Ziv, R. Nano Lett. 2009, 9 (2), 909– 913. (18) Horton, N. C.; Perona, J. J. Proc. Natl. Acad. Sci. U.S.A. 2000, 97 (11), 5729–34. (19) Viadiu, H.; Kucera, R.; Schildkraut, I.; Aggarwal, A. K. J. Struct. Biol. 2000, 130 (1), 81–5. (20) Castronovo, M.; Radovic, S.; Grunwald, C.; Casalis, L.; Morgante, M.; Scoles, G. Nano Lett. 2008, 8 (12), 4140–4145. (21) Newman, M.; Strzelecka, T.; Dorner, L. F.; Schildkraut, I.; Aggarwal, A. K. Structure 1994, 2 (5), 439–452. (22) Winkler, F. K.; Banner, D. W.; Oefner, C.; Tsernoglou, D.; Brown, R. S.; Heathman, S. P.; Bryan, R. K.; Martin, P. D.; Petratos, K.; Wilson, K. S. EMBO J. 1993, 12 (5), 1781–1795. (23) Aqua, T.; Naaman, R.; Daube, S. S. Langmuir 2003, 19 (25), 10573–10580.

NL902748G

Nano Lett., Vol. 9, No. 12, 2009