WP4 - DNA computation of container addresses
This report summarizes the overall advancement made in the MatchIT project for work package 4.
The overall objective of this WP was to achieve coherent labelling system for Chemtainers that enables control of specific interactions between Chemtainers and between the Chemtainers and a solid surface. The experimental achievements and overall design is being disseminated as a paper (Gil et al, 2013). The generic programmable information chemistry was formulated in a calculus and is being disseminated by WP6 (Fellermann & Cardelli, 2013).
Initially, molecular computation knowledge and know-how was utilized to assemble a robust, programmable and reliable labelling system. This system was designed to facilitate specific interactions between chemtainers and with the microfluidic system's surface. The work-package contained consecutive steps to achieve its goals:
Step 1: Design a computation system for labelling and relabeling containers. A ssDNA based design was established to enable differential labelling of each component of the system, i.e. Chemtainers and matrix specific locations. The labels were shown to enable specific interactions and several logic and technical operations were designed. For example, changing the label (relabeling) after an interaction had occurred was designed and examined in
Step 2: A modeling based demonstration of the systems computational capabilities. A computer simulation has been developed to demonstrate how the specific interactions perform computation through integrated information processing and material production processes, MatchIT automaton, collaborative work with WP6.
Step 3: Relabeling mechanism demonstration in solution (proof of principle). A relabeling mechanism based on strand migration (displacement) was designed. Since working with Phospholipid Vesicles (PV) was proven to be technically easier and more significant in the scope of this research – we chose perform the proof of principle demonstration directly with PV (see results below).
Step 4: Relabeling signal examination. To control relabeling timing and location, a triggering mechanism is needed. Several molecular, chemical and physical signals were investigated as triggering agents, or “relabeling signals”. Due to the fact that the labelling mechanism is based on DNA, the most intuitive (and researched, in the field of molecular computing) relabeling signal are ssDNA or RNA molecules. Moreover, these signals might enable versatile interactions capabilities for the final, integrated system. ssDNA triggered displacement process was utilized and shown to perform readdressing on DNA labeled PV. Iterative nature was also shown by changing an address more than once (Fig. 4.1).
Step 5: Relabeling mechanism demonstration on surface and containers. Relabeling of surface addresses was performed in collaboration with WP2 and WP1. More details are given below and the results are described in the WP2 report.
Step 6: Partial integration – with the microfluidic system. WP2 obtained self-assembled DNA monolayers (SAM) on gold surface in the presence of mercaptohexanol (MCH). Thiol-capped DNA oligonucleotides were used for the modification of gold surface. Former studies suggest that the fluorescence is enhanced when a fluorophor is placed in a distance at least of 10 nm above the surface. For this purpose 24nt oligonucleotide containing mercaptohexanol at its 5’-end was used. When hybridized with reporter strand it gives ca. 9.7 nm distance between a dye and the surface. The thiol-group was reduced with use of TCEP (Na3PO4, pH 7.0) and DNA was purified then in accordance to the standard procedure using Waters Oasis extraction cartridge.
Step 7: Integration with containers. Integrating chemtainers’ labelling system with microfluidics was done by applying labeled PV on labeled surface with matching or non-matching addresses. Only the matching address enabled vesicle docking, followed by pH mediated content release – executed by microfluidics.
Step 8: Autonomous relabeling mechanism. Autonomous operation could be demonstrated by inducing a relabeling operation by the product (or by-product) of a chemical reaction. We have simulated ATP cargo release, envisioned as a reasonable side of a main reaction to be carried out in fused vesicles, as an example. The ATP was coupled to a relabeling mechanism via aptamer mediated strand displacement. This way, ATP was shown to activate the operation AtoB which was otherwise inactive.
Novel design was constructed, presented in figure 4.1, in which each tag is comprised of 15-nt long ssDNA molecule representing the address, and two regulatory regions (5-nt long each), each on each side of the address (Fig. 4.1).
A genetic algorithm that generated a library of unique addresses with minimal interactions between them was created. The best addresses were chosen after empirical tests for self-dimerization and non-specific association with other addresses. Additionally, a pool of regulatory regions was generated by the same algorithm. The system's design ensures that non-specific interactions between regulatory regions and addresses are inconsequential.
Figure 4.1: Schematic representation of addresses attached to a vesicle.
Strand migration (displacement) reactions were designed to perform the requisite operations on addresses. The following operations were blueprinted:
- Readdressing (i.e. Change one address to another).
- Blocking (eliminating a specific address).
- Logic OR and logic AND operations.
- Reducing/increasing addresses number per Chemtainer.
- Predefined changing of addresses.
Each of the operations is performed by adding ssDNA/dsDNA molecules to the system. This addition could be done manually, automatically (by the microfluidic system) and theoretically also autonomously, by content release from Chemtainers.
A formal calculus able to capture the complexity of compartmentalized reaction systems such as colonies of possibly nested vesicular compartments was developed by WP6. Compartments contain molecular cargo as well as surface markers in the form of DNA single strands. These markers serve as compartment addresses and allow for their targeted transport and fusion, thereby enabling reactions of previously separated chemicals. The overall system organization allows for the setup of programmable chemistry in microuidic or other automated environments. This work demonstrates a simple sequential programming language based on state-of-the-art microfluidic technology and DNA addresses. Since the provided proofs are constructive, they can be utilized to automatically infer control programs for the construction of target structures from a limited set of resource molecules.
To evaluate the feasibility of this system the readdressing operation was experimentally examined. Fluorescence-labeling of 6 DNA addresses was used to monitor the addresses' interactions. Experiments done in collaboration with WP2 show specific association of addressed vesicles (Fig. 4.2). First, vesicles labeled with Cy3 fluorophore (green) attached to address A (vA) were mixed with vesicles labeled with Cy5 fluorophore (red) attached to address A' (vA', Fig. 4.2, a). In this experiment, non-specific associations, which could be estimated by vA-vA association (green vesicles associating with green vesicles) or vA'-vA' association, are significantly less abundant than specific associations (vA-vA'). Similarly, when vC (green) are mixed with vC' (red) specific associations (vC-vC') are significantly more abundant than non-specific associations (fig. 4.2, b). Minimal or no interactions occurred when vA were mixed with vC' (fig. 4.2, c) or when vA' were mixed with vC (fig. 4.2, d).
Figure 4.2: Specific phospholipid vesicles association. a) Association could be detected when vesicles labeled with address "A" (vA, green) were mixed with vesicles labeled with address vA' (red). b) Similarly, when vC (green) are mixed with vC' (red) significant association could be observed. c) Minimal or no interaction occurred when vA (green) were mixed with vC' (red). d) Minimal or no interaction occurred when vA' (red) were mixed with vC (green). e) Significant association could be observed when vA (green) are readdressed to C (with AtoC operation), in the presence of vC' (red). f) The reciprocal experiment (negative control) where vA' (red) are mixed with vC (green) in the presence of AtoC, show no association.
Next, the readdressing operation that changes address A to Address C' (AtoC') was examined. When vA were mixed with vC' in the presence of the DNA molecule that realizes the readdressing operation AtoC' significant association could be observed (fig. 4.2, e). Moreover, the association rate seems to be higher than the association observed for vA-vA' or vC-vC'. To validate that the interaction is indeed specific, the "reciprocal" experiment was performed as a negative control. Namely, vA' were mixed with vC in the presence of the same DNA molecule (AtoC). In this experiment no association was detected (fig. 4.2, f).
To be able to perform address mediated fusion that would result in a labeled Chemtainer that could be further used, the addresses of the original chemtainers should be "recycled". Therefore a technical process, named "peel", was added to remove any mediator molecules. Peeling was evaluated by a set of experiments in which an XtoY operation was "reversed" by Peeling (Fig. 4.3). The figure demonstrated that association is only observed when the proper operation (BtoC') is performed without peeling. Whenever peeling was used (or when the addresses are not matching) – no association could be observed. This also demonstrated that the association of fusion is due to (reversible) DNA-hydrogen bonds.
Figure 4.3: Demonstration of the "peel" operation. After DNA addresses mediated vesicles' fusion, the DNA addresses are sometimes still attached via a connecting dsDNA molecule (that realized the address change operation). A "peel" operation that removes this strand could be realized by adding a ssDNA molecule that is complementary to one of the strands comprising the connecting molecule. For example, when vB and vC are connected through BtoC, the operation rtoC removes the ssDNA that represents the new address C in the molecule AtoC. A "double peel" could be performed by adding 2 ssDNA molecules that are complementary to both strands comprising the XtoY operation, as deomnstrated in the lower right image for BtoC.
This operation enabled to perform a logic AND operation on fused Chemtainers. The AND operation was designed as illustrated in Fig. 4.4. Several A&BàC molecules have been designed and constructed with different arms' lengths (the arms complementary to A and B). Due to technical difficulties only initial test were performed to empirically find which length of arms is enough to enable the gate operation when A and B are on the same PV, but not when each of them is on a different Chemtainer.
Figure 4.4: Design of a logic operation (AND). After DNA addresses mediated vesicles' fusion (illustrated above), the DNA addresses are still attached. A "peel" operation is therefore needed (see Figure 4.1), after which the vesicle is decorated by A and B addresses. A logic AND operation is designed to be realized by a triplex-DNA molecule containing the complementary addresses to A and B only when in close prozimity but not to each of them alone. Upon hybridization the address C is exposed, labelling the newly fused vesicle.
To demonstrate the integration of chemtainers and MEMs, DNA-directed immobilization of giant unilamellar vesicles (GUVs) on gold electrodes was shown. Initially, it was shown that immobilization of thiolated single stranded DNA (ssDNA) was restricted to the gold electrodes of the microfluidic chip. The Chip was loaded with GUVs functionalized with ssDNA either complementary or non-complementary (as a control) to the immobilized ssDNA. Although the vesicle immobilization was found to be sequence specific, it was found to be not restricted to the gold electrodes. This finding was unexpected and indicated that the ssDNA was immobilized not only on the gold electrodes but also on the remaining SiOx surfaces enabling the complementary GUVs to bind to the entire surface. The annealing of fluorescently labeled ssDNA complementary to the immobilized ssDNA did result in a fluorescence signal only for the ssDNA oriented upward, while the signal was quenched for the ssDNA laying flat. Although not detectable by the fluorophore the ssDNA on the SiOx surfaces was still available for immobilizing vesicles. This issue was solved by “blocking” the chip by pre-incubating it with randomized DNA followed by the immobilization of thiolated ssDNA in the presence of 1 % BSA. A good evidence of sequence-specificity was the release of vesicles from those electrodes containing pH-switchable self-assembled monolayers upon pH jump caused electronically in quinhydrone system. In other words, sequence specific reversible binding were shown to occur between immobilized DNA triplexes and address-labeled vesicles. A paper describing our findings is under preparation in collaboration with WP1, WP2 and WP5 (Minero et al., 2013).
ATP mediated address change was shown by activating the AtoB molecule (implementing readdressing of vA to vB) by ATP. Fig. 4.5 shows the design of a conditional-AtoB molecule (Fig. 4.5a) and the interaction level between AtoB and address B’ in the presence or absence of ATP (Fig. 4.5b). In these experiments a strand complementary to the ATP aptamer was used as a positive control. Since the molecule representing address B is fluorescently labeled and the AtoB molecule contains a quencher (Q) - decreased fluorescence indicates interaction. Initial variations in fluorescence (in the first ~100 min) results from reaction components addition to a solution containing only the buffer and the labeled address B molecule. Although the positive control (ssDNA strand complementary to the aptamer) showed the best yield, it is well demonstrated that ATP and not ATP-buffer (ddw, double distilled water) activates the AtoB molecule. As this reaction performed best at 37°C, which is not optimal for vesicles, different temperatures were examined with different buffers. Even though temperature decreasing below 32°C reduced the reaction yield dramatically, our past experience with aptamers shows that further research could result in practical yields at R.T..
Figure 4.5: ATP activation of AtoB operation. (a) design of conditional AtoB molecule ATP aptamer blocks the molecule activity until removed by ATP binding. (b) While B' is labelled, active AtoB reduces the fluorescence by bringing the quencher close to the fluorosphore. Addition of MITs05, which is complementary to the aptamer, or ATP - displaces the aptamer, thus enabling binding to B' which results in decrease in fluorescence.
The aptamer based mechanism, described above, could be integrated with the rest of MatchIT approach by designing a system in which the interaction between A and B is dependent on the completion of a reaction within a (fussed) vesicle. For example, a scenario could be designed in which vA should interact (associate of fuse) with vB only after a (bio)chemical reaction had been completed inside vA (which could be the product of fusion between vC and vD). In this case, ATP (or any other small molecule) would be designed to be a byproduct of the (bio)chemical reaction. As the reaction proceeds, ATP is being accumulated in the vesicle. At a certain concentration it will be released to the medium and activate the ATP-dependent-AtoB complex (Fig 4.5a) which was presence ab-initio. Upon activation, the AtoB molecule could change the address on vA to B, thus enabling new interaction between vesicles after the (bio)chemical reaction was completed.
Binyamin Gil, Maik Hadorn, Uri Shabi, Ehud Shapiro & Martin M Hanczyc, "DNA programming of mesoscale vesicles" 2013, in preparation
Harold Fellermann & Luca Cardelli, "Programmatic manipulation of nanoscale bioreactors", 2013, in preparation
A. Minero, P. Wagler, M. Hadron, E. Bönzli, B. Gil, J. McCaskill, S Rasmussen and M. M Hanczyc (Order of authors not final) "Programmable reversible isothermal DNA-addressing of vesicles and beads to surfaces via local pH-switching microelectrodes." 2013, in preparation