Heterochiral Nucleic Acid Circuits

Abstract

The programmability of DNA/RNA-based molecular circuits provides numerous opportunities in the field of synthetic biology. However, the stability of nucleic acids remains a major concern when performing complex computations in biological environments. Our solution to this problem is L-(deoxy)ribose nucleic acids (L-DNA/RNA), which are mirror images (i.e. enantiomers) of natural D-nucleotides. L-oligonucleotides have the same physical and chemical properties as their natural counterparts, yet they are completely invisible to the stereospecific environment of biology. We recently reported a novel strand-displacement methodology for transferring sequence information between oligonucleotide enantiomers (which are incapable of base pairing with each other), enabling bio-orthogonal L-DNA/RNA circuits to be easily interfaced with living systems. In this perspective, we summarize these so-called “heterochiral” circuits, provide a viewpoint on their potential applications in synthetic biology, and discuss key problems that must be solved before achieving the ultimate goal of engineering complex and reliable functionality.

The advent of DNA nanotechnology, whereby Watson–Crick (WC) base pairing is used to generate complex static and dynamic systems, promises to have a profound impact on the field of synthetic biology.^1,2 In particular, DNA/RNA-based molecular circuits, whose operation is based on strand-displacement, provide a straightforward approach for designing synthetic systems capable of autonomous decision-making and complex logic functions.^3,4 Importantly, DNA/RNA-based circuits can be easily interfaced with living organisms via WC base pairing to the endogenous biopolymer, making them a highly attractive platform for engaging and/or manipulating biological processes in a programable and controlled manner. For example, the ability of DNA/RNA-based circuits to detect specific combinations of nucleic acid biomarkers (e.g. mRNAs and microRNAs) provides the basis for a number of exciting diagnostic technologies and provides a starting point for the development of “smart therapeutics” that only become activated within diseased environments.^5–7 Furthermore, the marriage of strand-displacement circuits with gene knock-in and knock-out tools, such as RNA interference (RNAi) and CRISPR-Cas9, provide a promising approach for precise, conditional control of gene expression.^8,9

Despite their potential, if DNA/RNA-based circuits are to be routinely employed for complex tasks in live cells, two key limitations must be overcome: (1) rapid nuclease degradation and (2) unintended interactions with bio-macromolecules, both of which adversely affect the performance of the device. The most common approach for stabilizing circuit components is the use of chemically modified nucleotides, including 2′-O-methyl ribonucleotides,^9,10 locked nucleic acids (LNAs),^11,12 and/or phosphorothioate (PS) linkages. Although these modifications can dramatically increase the intracellular stability of nucleic acids ¹³, they also alter the thermodynamic properties of circuit components in an unpredictable manner. Indeed, DNA strand-displacement reactions involving chemically modified nucleic acids are poorly characterized, making the design of corresponding circuits extremely challenging. Moreover, modified nucleotides can be toxic and tend to have adverse effects on cell viability. Due to these issues, modification-independent approaches for stabilizing DNA circuit components have also been explored, including the ligation of vulnerable free DNA ends and the use of nuclease-resistant structural elements, such as hairpins and nanostructures.^2,14,15 However, these methods have found only limited success. Importantly, none of the above approaches address potential off-target interactions of DNA/RNA-based circuits with abundant cellular nucleic acids, which can lead to leak (i.e. the undesired triggering of strand displacement reactions) and/or sequestration of circuit components, thereby eroding performance. Thus, there remains a need for novel approaches to construct biocompatible DNA/RNA-based circuitry.

A promising, yet rarely considered alternative to the use of modified nucleotides is a simple inversion of stereochemistry, i.e. the use of L-(deoxy)ribose-based nucleic acids (L-DNA/RNA). L-oligonucleotides are mirror images (or enantiomers) of native D-oligonucleotides and are completely resistant to nuclease degradation, nontoxic, and nonimmunogenic.^16–20 Importantly, as enantiomers, D- and L-DNA/RNA have identical physical properties in terms of solubility, hybridization kinetics, and duplex thermal stability.^16,17 Thus, well-established principles for designing D-DNA/RNA-based circuits can be directly applied to L-DNA/RNA without further optimization. In this way, L-oligonucleotides can be considered the ideal nucleic acid analogue. Furthermore, L-oligonucleotides avoid off-target interactions with cellular nucleic acids because oligonucleotides of opposite chirality (D versus L) are incapable of forming contiguous Watson–Crick (WC) base pairs with each other.^21–23 Unfortunately, this property also represents the key limitation of L-DNA/RNA because it precludes their direct interfacing with endogenous nucleic acids.

We recently reported a novel methodology for sequence-specifically interfacing D- and L-oligonucleotides, enabling (for the first time) development of DNA/RNA-based circuitry having fully-interfaced D- and L-oligonucleotide components (referred to as “heterochiral” circuitry).²⁴ Our approach takes advantage of peptide nucleic acids (PNA), which unlike native DNA and RNA, has no inherent chirality (Figure 1a). As a result, PNA hybridizes to DNA and RNA irrespective of chirality.²¹ On the basis of this property, we conceived two novel toehold-mediated strand-displacement reactions that exploit DNA/PNA heteroduplexes in order to interface the two enantiomers of DNA. In reaction A (Figure 1b), a DNA toehold domain (t*) can hybridize to an input (D/L-IN) of the same chirality (via t/t* domains), which eventually leads to the displacement of the incumbent PNA strand. By transforming an input signal of defined chirality into an achiral output signal, this reaction effectively decoupled the sequence information from the stereochemical information in the polymer. Accordingly, we refer to these DNA/PNA heteroduplexes (D/L-A₁) as “racemization” gates. The achiral PNA strand can serve as an input for downstream circuit components of either chirality. In contrast, Reaction B (Figure 1c) places the toehold domain (t*) on the PNA itself, allowing the strand-displacement reactions to be initiated by an input strand of either chirality. For example, a D-input strand (D-IN) can be used to displace an incumbent L-DNA strand from a L-DNA/PNA heteroduplex (L-A₂). We refer to this DNA/PNA heteroduplex (L-A₂) as an “inversion” gate because it effectively inverts the stereochemistry of the input from D to L in a sequence specific manner. Although we explain both reactions in the context of DNA, they are also compatible with input and output stands made of RNA.

Figure 1.

Heterochiral DNA strand-displacement reactions. (a) The three types of nucleic acids discussed herein. D-DNA/RNA (black), L-DNA/RNA (blue), and PNA (green) are distinguished by color. (b,c) Schematic illustrations of the heterochiral strand-displacement mechanism for reaction A (b) and reaction B (c). Nucleic acids are depicted as lines with half arrows denoting the 3′ end and an asterisk indicating complementarity between sequence domains. (d) A biocompatible, microRNA-specific sensor based on the L-RNA version of the fluorogenic aptamer Mango III. When properly folded, Mango III binds thiazole orange (TO) resulting in strong fluorescent activation. Activation of the sensor is achieved through heterochiral strand-displacement of a PNA blocking strand from the L-Mango III aptamer by the D-microRNA target.

We believe these rather simple reactions now provide the foundation for interfacing L-DNA/RNA-based circuits with endogenous D-nucleic acids within living cells and organisms. Given the bio-orthogonal nature of L-oligonucleotides, we expect that L-DNA/RNA-based circuits will have dramatically improved intracellular performance, reliability, and utility compared to traditional approaches, opening the door to a number of exciting applications. For example, logic-driven L-DNA/RNA-based circuits could be programmed to generate optical and/or chemical “outputs” in response to specific combinations (or patterns) of nucleic acid “inputs” within living cells, providing opportunities for “intelligent” disease diagnosis and treatment. As a proof-of-principle, we recently developed a fluorescent L-RNA-based biosensor for the detection of microRNAs inputs, which we used to demonstrate, for the first time, sequence-specific interfacing of oligonucleotide enantiomers in living cells (Figure 1d).²⁵ Furthermore, by utilizing chemical outputs, such as antisense oligonucleotides or enzyme inhibitors, L-DNA/RNA circuits could be used to regulate gene expression and modulate protein activity in response to both endogenous or exogenous nucleic acid input signals. It is worth noting that metal ions and/or small molecule metabolites could also be interfaced with L-DNA/RNA circuits by simply replacing inversions gates with mirror-image aptamers (i.e. Spiegelmers) or (deoxy)ribozymes that have been engineered to release L-DNA/RNA output strands in response to their target analyte.²⁶ Lastly, beyond L-DNA/RNA circuits, one can imagine the exciting possibility of using heterochiral methods to create self-replicating systems and/or synthetic cells having both D-DNA and L-DNA genomes. In addition to possible applications in synthetic biology, such systems may provide fundamental insights into the origin of biological homochirality.

Although L-oligonucleotides and heterochiral strand-displacement address a number of concerns regarding the use of nucleic acid-based circuits as synthetic biology tools, numerous challenges must be overcome before achieving the ultimate goal of engineering complex and reliable functionality for synthetic biology applications. First, the thermodynamic and kinetic properties of heterochiral-strand displacement reactions must be thoroughly characterized in order to enable precise engineering of related L-DNA-based circuits. Such studies should also address potential limitations of PNA, including poor water solubility that places limits on its length and sequence.²⁷ Second, we must develop a better understanding of the intracellular behavior of L-oligonucleotide including their bio-distribution, trafficking, and toxicity. For example, intracellular environments are highly heterogeneous and the distribution of biomolecules is restrictive in both time and space, making it difficult to localize circuit components to the cellular compartment where they will function. For this reason, L-DNA/RNA-based heterochiral circuits that functions well in the test tube may not be effective in cells. Additionally, intracellular delivery of nucleic acids (regardless of stereochemistry) is extremely challenging, and in the case of L-oligonucleotides, relatively unexplored. Nevertheless, several groups have demonstrated that L-oligonucleotides can be efficiently delivered into cells using traditional approaches, such as cationic liposomes (e.g. Lipofectamine) and cholesterol conjugates.^25,28 Because these challenges parallel those faced by therapeutic oligonucleotides, more effective solutions may already be emerging from that field.²⁹ For example, GalNAc (N-acetylgalactosamine) conjugates³⁰ and cell-penetrating peptides³¹, both of which promote efficient cellular uptake and endosomal escape of D-oligonucleotide-based drugs, may also prove effective for the delivery of L-oligonucleotide-based circuit components. It is worth noting that the superior biostability of L-oligonucleotides may facilitate the development of completely novel delivery technologies. Finally, it is not currently possible to synthesize and/or replicate L-oligonucleotides using the cellular machinery. While this challenge may seem insurmountable, several recent studies suggest that a solution may be closer than expected. For example, Sczepanski et al reported an RNA enzyme (or ribozyme) that catalyzes the ligation and polymerization of RNA of the opposite chirality.³² In particular, the D-RNA version of this “cross-chiral” ribozyme was shown to catalyze the ligation of two or more L-RNAs. More recently, the ligation of L-DNA using a protein ligase comprised of D-amino acids was also demonstrated.³³ Furthermore, two research groups independently synthesized the D-amino acid version of DNA polymerase IV (D-Dpo4) and demonstrated mirror-image transcription and PCR amplification of L-RNA and L-DNA, respectively, using synthetic L-DNA templates.^34–36 Such enzymes may one day enable intracellular assembly of complex L-DNA/RNA-based circuits from readily deliverable nucleoside building blocks. Given recent advances in polymerase engineering ³⁷, these studies raise the exciting possibility of creating novel polymerases capable of directly synthesizing L-DNA and/or L-RNA using a D-DNA template, which may ultimately allow L-oligonucleotides to be genetically encoded. Given the progress that has already been made, it seems very likely that these challenges will be overcome, advancing L-oligonucleotides to the forefront of synthetic biology tools for manipulating and controlling biological information.

Summary Points.

• Heterochiral circuits enable sequence-specific interfacing between the otherwise orthogonal enantiomers of DNA/RNA (D versus L).

• Because mirror image L-DNA/RNA are orthogonal to the stereospecific environment of biology, heterochiral circuits constructed from L-DNA/RNA overcome the primary barriers to engineering complex and reliable functionality for applications in synthetic biology.

• Despite the advantages of L-oligonucleotides, a number of challenges must be overcome before L-DNA/RNA-based circuits can take their place at the forefront of synthetic biology.