DNA-based communication in populations of synthetic protocells

Abstract

Developing molecular communication platforms based on orthogonal communication channels is a crucial step towards engineering artificial multicellular systems. Here, we present a general and scalable platform entitled 'biomolecular implementation of protocellular communication' (BIO-PC) to engineer distributed multichannel molecular communication between populations of non-lipid semipermeable microcapsules. Our method leverages the modularity and scalability of enzyme-free DNA strand-displacement circuits to develop protocellular consortia that can sense, process and respond to DNA-based messages. We engineer a rich variety of biochemical communication devices capable of cascaded amplification, bidirectional communication and distributed computational operations. Encapsulating DNA strand-displacement circuits further allows their use in concentrated serum where non-compartmentalized DNA circuits cannot operate. BIO-PC enables reliable execution of distributed DNA-based molecular programs in biologically relevant environments and opens new directions in DNA computing and minimal cell technology.

Figure 1. Design elements for biomolecular implementation of protocellular communication (BIO-PC).

a, General strategy of the BIO-PC platform. Protocells with encapsulated DNA gate complexes are localized on a 2D spatial grid and can sense, process and secrete short ssDNA-based signals. The system is initiated by adding of ssDNA inputs and the response dynamics associated with the compartmentalized DSD reactions for each protocell are followed by confocal microscopy. b, Individual protocells can be configured to perform various tasks ranging from signal detection to Boolean logic operations. Individual modules can be combined to implement more complex population behaviours such as cascaded signalling, bidirectional communication and distributed computing. c, Procedure for preparing DNA-encapsulating protocells. Streptavidin-containing proteinosomes are assembled by covalently crosslinking BSA-NH₂/PNIPAAm nanoconjugates at the interface of water-in-oil emulsion droplets. The crosslinked microcapsules are then phase transferred into water and biotinylated DNA strands are localized in the interior via biotin-streptavidin interactions. d, A mechanistic model for toehold-mediated DSD reactions inside protocells. The input strand A diffuses through the semipermeable membrane at a rate governed by permeability constant P (μm min^-1) and then activates a fluorescent (Cy5) DNA gate complex F:Q (F = fluorophore/gate; Q = quencher/output strand) via a DSD reaction described by a bimolecular rate constant k (nM^-1 min^-1). e, CAD drawing of a microfluidic protocell trap array with computer-rendered trapped protocells shown in red. f, Confocal micrographs of eight trapped proteinosomes showing time-dependent increase in Cy5 fluorescence associated with the activation of an encapsulated DNA gate complex. Scale bar 50 μm. g, Fluorescence traces and model fittings of DSD reactions in protocells with high (red) and low (blue) membrane permeability. Reactions were carried out with 100 nM of A for high-P and 1000 nM for low-P proteinosomes. To maintain a constant input A concentration in the medium, the input strand solution was slowly flowed through the microfluidic chamber at a rate of approximately 0.1 μl/min throughout the experiment (Methods). Concentrations of the activated DNA gate complex (F:A complex) are determined from time-dependent fluorescence measurements on individual protocells trapped within the microfluidic array device. h, Estimated permeability constants for the two protocell populations. i, Estimated bimolecular rate constants for the DSD reactions inside high-P and low-P proteinosomes compared to the estimated rate constant for a reference DSD reaction under batch conditions.

Figure 2. Signalling cascade between protocell populations using non-enzymatic DNA signal amplification.

a, Abstract diagram of a two-layer signalling cascade between two protocell populations. Proteinosomes of the first population sense the externally added input strand and respond by activating a Cy5 fluorescent DNA gate complex and secreting a signal strand to the external environment. The second population can sense the signal and produces a fluorescent response. Non-enzymatic DNA catalysis is used to recycle the signal strand by consuming the abundant fuel strand, thereby producing an amplified response. b Molecular reaction diagram of the signalling cascade. Coloration of graphics represents state of fluorescent probes: non-activated (grey), Cy5 activated (red), Alexa546 activated (orange). It is important to note that while the signal strand Q₁ is shown to diffuse out of the population 2 proteinosome for illustrative purposes, it can also trigger the next reaction inside the same proteinosome by reacting with another inactive DNA gate complex. The sequences for this system are given in Supplementary Table 2. c, Mean and standard deviations of the fluorescent traces of two high-P proteinosome populations in the absence of a fuel strand. [Input] = 100 nM, the fractions of the two populations were χ₁=0.55, χ₁=0.45. Concentrations of the activated DNA gate complexes are determined from time-dependent Cy5 (population 1, red profile) or Alexa546 (population 2, orange profile) fluorescence measurements on individual protocells trapped within the microfluidic array device. d, Mean and standard deviations the fluorescent traces of two high-P proteinosome populations in the presence of 1 μM of fuel strand. [Input] = 100 nM, χ₁=0.54, χ₁=0.46. Note the large increase in the fluorescent output of population 2 (orange profile) due to signal amplification. e, Mean and standard deviations of 2D simulation data using parameters of the system in c (red and orange curves, population 1 and 2, respectively). f, Simulation data using parameters of the system in d. g, Confocal micrographs (Exp.) and simulation data (Sim.) of a group of six proteinosomes comprising three members of population 1 and three members from population 2. The delay between activation of population 1 and 2 during the initial minutes of the reaction is shown.

Figure 3. Three-layer amplified signalling cascade.

a, Abstract diagram of three-layer signalling cascade with each stage providing fuel-mediated signal amplification upon activation by its cognate input. b, Epifluorescence micrograph of three-color barcoded high-P proteinosomes overlaid on a bright-field micrograph of the microfluidic trapping array. The barcoding is realized by membrane labelling of proteinosome populations 1, 2 and 3 using FITC (green), DyLight405 (dark blue) and a 1:1 mixture of the two fluorophores (light blue), respectively. c, Cy5 fluorescence traces of proteinosome populations 1, 2 and 3 of the three-layer signalling cascade showing progressive levels of signal amplification. A larger fraction of the third population, χ₃ relative to the first two populations (χ₁ and χ₂) was used to ensure that the stronger mean response of the third population can only occur due to signal amplification. Reaction was performed using 10 nM of input and 500 nM of each fuel strand. d, Cy5 fluorescence traces showing the response of the system consisting only of populations 1 and 3. At time point 1, the three fuel strands were added to the protocell community followed by both fuel and input at time point 2. The reaction was performed using 10 nM of input and 500 nM of each of the three fuel strands. Both proteinosome populations were high-P and were assembled with 4 μM of streptavidin.

Figure 4. Protocellular negative feedback loop.

a, Abstract diagram of a negative feedback loop between two protocell populations engaged in compartmentalized DSD reactions. The system is triggered by introducing the ssDNA input strand into the external medium, which then activates the Cy5-labeled DNA gate complex of population 1 and induces the release of a strand that acts as an activator for population 2, switching on their Cy3-labeled DNA gate complex. As a consequence, an inhibitor strand is released into the medium and protocells in population 1 are deactivated by switching off their fluorescent DNA gate complexes. Protocells of population 2 also implement an amplification module, allowing them to recycle the input strands. b, Molecular reaction diagram of the negative feedback network between two proteinosome populations. Coloration of graphics represents state of fluorescent probes: non-activated (grey), Cy5 activated (red), Cy3 activated (orange). The sequences for this system are given in Supplementary Table 4. c, Time-dependent activation levels for two high-P proteinosome populations implementing the negative feedback loop. The activation and deactivation of population 1 (red curve, Cy5 fluorescence) is consistent with negative feedback signalling in the binary population. The progressive increase in the fluorescence of population 2 (orange curve, Cy3) is due to fuel-mediated signal amplification. The reaction was initiated with 200 nM of input and 1 μM of fuel strands to run the amplification reaction in population 2. The fractions of the two populations were χ₁=0.49, χ₂=0.51. d, Activation levels with reduced levels of input (100 nM) and fuel (300 nM), χ₁=0.53, χ₁=0.47. e, Response of population 1 alone (high-P prepared using 4 μM of streptavidin) showing time-dependent changes in concentration (fluorescence) due to activation of the streptavidin-anchored DNA gate complex (F1:A) in the absence of a released inhibitor strand. Reaction was performed with 200 nM of the input and 1 μM of the fuel strand; both were added at the time point indicated by the dashed line.

Figure 5. Compartmentalized DNA-based Boolean logic circuits.

a, Schematic representations of combinatorial sensing-and-processing protocell network with either AND or OR logic gates based on the synergistic activity of two transducer populations and a single receiver population. b, Molecular reaction diagram of the protocell-based three-population AND network. The sensor proteinosomes (populations A and B) contain DNA gate complexes that initially consist of a streptavidin-anchored fluorescently active strand (F₁ or F₂) and signalling strands (A or B), respectively. F₁ and F₂ become quenched upon binding of the cognate input strands, which in turn releases signalling strands A and B from the two different sensor populations. The population of AND gate proteinosomes contain a DNA gate complex consisting of a deactivated strand (F₃) bound to two complementary oligonucleotides that function either as a blocker or quencher/output strand during processing. Activation of the complex requires the binding of both the signal strands such that F₃ acts as a fluorescent probe of the AND operation. Signal A binds to the initially free toehold of F₃, displaces the blocker strand and thereby reveals a second toehold where signal B can bind and displace the output (quencher) strand. Non-activated protocells (grey); Cy5-activated protocells (red). The DNA sequences for this system are given in Supplementary Table 5. The reaction diagram of the OR circuit is given in Supplementary Figure S11. c, Experimental data of the AND and OR devices based on a protocell network consisting of two sensor proteinosome populations in combination with an AND or OR gated population. The plots show time-dependent changes in the concentration of the activated DNA gate complex in the AND or OR proteinosomes. The color-coded traces correspond to the configurations indicated in the truth table, with the first two columns indicating the presence (1) or absence (0) of the input strands A and B respectively and third column indicating the presence (+) or absence (-) of the sensor populations. All populations comprise high-P proteinosomes with equal input strand concentrations (100 nM) where applicable; In experiments with all three populations, their respective fractions were within the range of χ_A=0.40-0.49, χ_B=0.41-0.48 χ_AND=0.06-0.16 for the AND gate and χ_A=0.37-0.47, χ_B=0.45-0.54 χ_OR=0.08-0.15 for the OR gate. The complete data with standard deviation for each input configuration and the activation of the individual sensor modules are shown on Supplementary Fig. S11. d, Schematic representation of two high-P proteinosome populations collectively implementing a serially gated logic circuit. Population 1 calculates logic OR from inputs B and C, and the protocells of population 2 calculate logic AND of input A and the output of population 1. See Supplementary Figure S12. e, Experimental data of the two-gate, three-input logic circuit. The color-coded traces correspond to the input configurations shown in the truth table with 1 indicating the presence and 0 the absence of the respective input strand. Experiments were performed with 500 nM of each input strand, χ_OR=0.48-0.58, χ_AND=0.42-0.52. The complete data sets with standard deviation for each input are shown in Supplementary Fig. S12.

Figure 6. Compartmentalized DNA reaction networks in 50% FBS.

a, Experimental procedure for validating the stability of a streptavidin-bound quenched DNA gate complex (F:Q, 100 nM; streptavidin, 100 nM) in 50% FBS under batch conditions. The stability of the complex was determined by measuring the fluorescence associated with DSD activity after addition of input strand A (200 nM). b, Fluorescence traces for the batch DSD reaction in 0% (buffer only, blue trace) and 50% (red) FBS showing loss of activity in 50% FBS. c, Validation procedure for a compartmentalized version of the DSD reaction shown in a. d, Mean and standard deviation of protocell activation levels in 0% (buffer, blue trace) and 50% (red) FBS. The experiments were performed by incubating the DNA gate-containing proteinosomes for 48h in 50% FBS or 0% FBS, followed by adding 2 μM of input strand A to initiate the DSD reaction. High-P proteinosomes containing free BSA (60 mg/mL) were used to minimize osmotic collapse. e, Schematic representation of a simple non-catalytic transducer-receiver signalling cascade between two protocell populations incubated for 48 h in 50% FBS. Proteinosomes of population 1 sense the externally added input strand and respond by activating a Cy5 fluorescent DNA gate complex and secreting a signal strand that is sensed by population 2 to produce a Alexa546 fluorescent response. f, Mean and standard deviations of protocell activation levels for populations 1 and 2 of the 2-stage signalling cascade showing fluorescent activation in both populations. The reaction was started after 48h incubation in 50% FBS by adding 2 μM of the input strand. High-P proteinosomes containing free BSA (60 mg/mL) were used to minimize osmotic collapse.