Metal-dependent DNAzymes for Quantitative Detection of Metal Ions in Living Cells: Recent Progress, Current Challenges, and Latest Results on FRET Ratiometric Sensors

Abstract

Many different metal ions are involved in various biological functions including metallomics and trafficking, and yet there are currently effective sensors for only a few metal ions, despite the first report of metal sensors for calcium more than forty years ago. To expand upon the number of metal ions that can be probed in biological systems, we and other labs employ the in vitro selection method to obtain metal-specific DNAzymes with high specificity for a metal ion, and then convert these DNAzymes into fluorescent sensors for these metal ions using a catalytic beacon approach. In this forum article, we summarize recent progress made in developing these DNAzyme sensors to probe metal ions in living cells and in vivo, including several challenges that we were able to overcome for this application, such as DNAzyme delivery, spatiotemporal control, and signal amplification. Furthermore, we have identified a key remaining challenge for quantitative detection of metal ions in living cells and present a new design and results of a FRET-based DNAzyme sensor for ratiometric quantification of Zn²⁺ in HeLa cells. By converting existing DNAzyme sensors into a ratiometric readout without compromising the fundamental catalytic function of the DNAzymes, this FRET-based ratiometric DNAzyme design can readily be applied to other DNAzyme sensors as a major advance in the field to develop much more quantitative metal ion probes for biological systems.

Graphical Abstract

We summarize recent progress in developing metal-specific DNAzyme sensors to probe metal ions in biological systems, while highlighting advances in meeting major challenges in DNAzyme delivery, spatiotemporal control, and signal amplification in living cells and in vivo. Furthermore, we present our latest results of a FRET-based DNAzyme sensor for ratiometric quantification of Zn²⁺ in cells. This FRET-based ratiometric method can be applied to other DNAzyme sensors to enable quantitative imaging of different metal ions in many biological systems.

Introduction

Metal ions are involved in many important functions in biology and are thus actively studied for their roles in processes such as signal transduction and enzymatic activity. Of particular interest is the role of labile metal ions, i.e., metal ions that are not tightly bound to biomolecules and are available for biological functions such as cellular signaling, redox reactions, and metal ion trafficking for further incorporation into proteins and other biomolecules.^1–5 Furthermore, it has been found that metal toxicity and disruption of metal homeostasis has been linked to various diseases including different types of cancer,^6–8 neurodegenerative diseases,^9–10 and nutritional immunity/bacterial infection.^11–13 For instance, misregulation of iron homeostasis, particularly related to increases in the amount of labile iron, has been implicated in many different diseases including various cancers and neurodegenerative diseases, because of the ability of labile iron to generate peroxide radicals, which are highly cytotoxic.^14–15 Other redox-active metal ions such as manganese and copper have similar affects toward various diseases and also play key roles as cofactors in many important proteins.^16–17 On the other hand, non-redox-active metal ions are often involved in other processes such as cell signaling pathways, both directly and indirectly, as has been found for calcium,^{16, 18} zinc,¹⁹ copper,²⁰ magnesium,^21–22 and potassium.²³ As such, their misregulation can potentially be a useful indicator of abnormal cellular processes, which could be better monitored with sensors for these metal ions. Furthermore, many other heavy metal ions, such as mercury, lead, and uranium, are not present endogenously in biological systems, but their potential contamination has led to an increasing number of studies on their harmful effects in biological systems.²⁴

These wide variety of functions from metallomics to signaling, carried out in whole or in part by metal ions, have led to significant interest in developing sensors to probe the location and distribution of these metal ions in cells.^25–32 Toward this goal, many fluorescent sensors have been developed for several of these metal ions, and are most frequently based on small organic molecules or proteins, which have been leveraged to enable detection of changes in cellular metal ion concentration.^{26, 33–42} While these studies have aptly demonstrated that such sensors can enrich our knowledge on the biological functions of metal ions, current sensors have been limited to a relatively few number of metal ions, most notably Ca²⁺, Zn²⁺, Hg²⁺, Fe²⁺, and Cu^2+/+, with slow progress being made on developing sensors for novel metals, despite the first report of a Ca²⁺ sensor more than 40 years ago. Understanding the role of other metal ions will require the continued development of new sensors for many other metal targets.

The development of new organic molecule- and protein-based sensors has been focused on rational design methods,⁴³ but successfully designing sensors with high selectivity for specific metal ions with little to no interference from other competing ions remains a significant challenge for rational design. Furthermore, success in design for one type of metal may not readily translate to success with designing sensors for other metal ions, so the process must be started again for each sensor and each metal. An alternative approach to rational design which can potentially help to avoid many of these problems is combinatorial selection, also called in vitro selection, which does not rely on prior knowledge of metal-binding, and for which sensor selectivity and affinity can be improved by adjusting the stringency of selection conditions.^44–47 While there have been some successful publications for in vitro selection and library screening of proteins and peptides that selectively bind metal ions,^48–49 in vitro selection of metal-specific nucleic acids have emerged as the most promising approach. One major advantage for nucleic acids is that polymerase chain reaction (PCR) allows for amplification of a very small population of nucleic acids obtained during the selection step, which allows the pool to be regenerated during each selection step while maintaining its increased enrichment of sequences with the desired metal-binding activity. Similar amplification techniques are not readily available for any other class of molecules, which limits their ability to be used for similar in vitro selection techniques. A particularly active application of in vitro selection is the identification of deoxyribozymes, also called DNAzymes, that can catalyze enzymatic reactions, often in the presence of specific metal ions.⁵⁰ One major advantage of this technique is that the selection of metal-specific DNAzymes does not require a metal affinity column for the selection step, which could hamper the selection of molecules that can bind metal ions deeper in the binding pocket – an often essential component of tight and selective binding. After identification of these metal-dependent DNAzymes, their coupling of metal-binding with enzymatic activity allows for facile signal transduction and amplification. As a result, many DNAzyme-based sensors for a wide range of metal ions with high selectivity have been obtained. In this forum article, we first summarize recent progress that has been made in the development of DNAzyme-based sensors for intracellular detection of metal ions, followed by identification of a key remaining challenge: the difficulty in obtaining quantitative signal readouts with this class of metal sensors in cellular imaging, and end with presentation of new results to meet this challenge by developing a ratiometric DNAzyme sensor for quantitation of metal ions in living cells.

Metal-specific DNAzymes as Important Sensors for Metal Ions

DNAzymes are a class of DNA molecules, first discovered in 1994, that display enzymatic activity^{46, 51–54} such as RNA cleavage,^{51, 55} porphyrin metalation,⁵⁶ the Diels-Alder reaction,⁵⁷ and various amino acid modifications.⁵⁸ Unlike most other types of metal sensors that usually involve a rational design process requiring a priori knowledge about metal-binding and signal transduction and often resulting in a lot of trial-and-error, metal-specific DNAzymes are obtained through in vitro selection, which uses a large randomized DNA library of up to 10¹⁵ different sequences.^59–60 These random DNA sequences are then subjected to a series of different selection steps to isolate DNA sequences that can catalyze a specific chemical reaction (e.g., phosphodiester cleavage) in the presence of a metal ion under pre-defined conditions (e.g., at a certain pH). After several rounds of selection, typically between 5–20 rounds with PCR amplification steps in between each round, DNA sequences with high affinity and selectivity for the target metal ions are gradually enriched.^{46–47, 51–53, 61} By increasing the selection stringency (e.g. shorter reaction time to improve catalytic rates and/or lower metal concentration to improve metal affinity) over subsequent selection rounds, DNAzymes with specific binding affinity and sensitivity can be obtained.^{51, 59, 62–63} Similarly, negative or counter selection steps can be used to remove sequences that react with other competing metal ions to improve selectivity of the obtained DNAzymes for the desired target.⁶⁴ After the selection pool reaches a desired metal-dependent activity, the pool is sequenced and individual sequences are tested to discover the best sequences. Using this strategy, we and other labs have developed metal-dependent DNAzymes which have high sensitivity and specificity for many different types of metal ions, including: Pb²⁺,⁵¹ Zn²⁺,⁶³ Co²⁺,^64–65 Cu²⁺,^66–67 UO₂²⁺,⁶⁸ Mg²⁺,⁶⁹ Cd²⁺,⁷⁰ Hg²⁺,^{62, 71} Na⁺,^72–73 Ag⁺,⁷⁴ Cr³⁺,⁷⁵ and lanthanide metals.^76–79

Since many of these DNAzymes are dependent on the metal ions present during the selection to perform their catalytic function and are often selective toward those specific metal ions, a growing field of study has been the application of these metal-dependent DNAzymes for use as sensors for their respective metal ions.^{44–45, 61–62, 80–87} Though DNAzymes for many different types of chemical reactions have been selected,^{54, 88} the most common reaction for metal-specific DNAzymes is the RNA-cleavage transesterification reaction, as this is a relatively easy reaction to perform with various metal cofactors and thus results in relatively fast reaction rates, making the speed of sensor response rapid.⁵⁵ Because these DNAzymes catalyze a common reaction, almost all of these DNAzymes share a similar secondary structure, as depicted in Figure 1a: an enzyme (E) strand (blue) containing the “catalytic core” and a substrate (S) strand (black) containing the ribonucleotide cleavage site, which are hybridized together with two binding arms that flank the catalytic core (purple) and cleavage site (orange). The DNAzyme catalytic function typically depends only on the sequence of the catalytic core, so the sequence of most of the binding arms can be changed as desired for different applications.

Figure 1.

a) Scheme and secondary structure of a catalytic beacon based on the 8–17 DNAzyme reported previously;^{63, 80} A fluorophore (F) is attached to the substrate (S) strand (black) while quenchers (Q) are attached to the enzyme (E) strand (blue) and the opposite end of the substrate strand. The DNAzyme catalytic core is depicted in purple, across from the ribonucleotide cleavage site (rA) in orange. Before substrate cleavage, the quenchers minimize the fluorescent signal from the fluorophore, but after Zn²⁺-dependent cleavage of the substrate, the fluorophore labeled fragment can dissociate, producing significantly enhanced fluorescence due to the presence of Zn²⁺. b) Scheme and secondary structure of the FRET ratiometric 8–17 DNAzyme probe outlined in this work. The FRET donor (D) is attached to the substrate (S) strand (black) while the FRET acceptor (A) is attached to the enzyme (E) strand (blue). The DNAzyme catalytic core is depicted in purple, across from the ribonucleotide cleavage site (rA) in orange. The nitrobenzyl group “cages” the 2ʹ-OH of the rA cleavage site, preventing DNAzyme activity, until it is removed with 365nm light irradiation, which restores the 2ʹ-OH. After this decaging, the active DNAzyme is excited with the donor excitation wavelength for imaging. In the non-cleaved conformation, FRET between the D and A produces high A emission and low D emission. After Zn²⁺-induced DNAzyme cleavage, the D-modified strand can dissociate, allowing for high D emission but low A emission.

With the continued development of solid-phase oligonucleotide synthesis, many different functional groups can be readily incorporated into DNAzyme sequences, such as fluorophores, quenchers, and various conjugation groups for attachment to other moieties such as gold nanoparticles or fluorophores, thus allowing metal-dependent catalytic activity to be turned into a sensor readout.^89–96 The authors’ group has pioneered the development of these metal-dependent DNAzymes as fluorescent sensors using a modular design strategy called the catalytic beacon (Figure 1a).^{80, 85} With the catalytic beacon strategy, a fluorophore (F) is attached to the substrate strand (black) while quenchers (Q) are attached to both the substrate and enzyme (blue) strands. This design allows the complex to have relatively low fluorescent output in the initial, non-cleaved state, due to quenching of the fluorophore. The sequence and length of the two binding arms can be designed such that the melting temperature of the DNAzyme is higher than ambient temperature, so the substrate strand will remain tightly bound to the enzyme strand. However, upon metal-dependent cleavage at the indicated ribonucleotide site (orange), the two halves of the cleaved substrate strand have much lower melting temperatures to their respective binding arms, which allows for the fluorophore-labeled substrate half to dehybridize from the complex with quenchers, resulting in a significantly increased fluorescent signal which is dependent on presence of the metal ion. This strategy has been used for the detection of various metal ions in a variety of different samples for in vitro detection,^{45, 60, 86, 88, 97–102} which has further lead to the development of commercialized products, such as by ANDalyze, Inc., for the detection of heavy metals in environmental samples.

Intracellular Imaging of Metal Ions with DNAzyme Catalytic Beacons

In recent years, a major focus for the further development of DNAzyme catalytic beacon sensors has been towards their application in biological systems. DNAzymes have several key advantages over other types of fluorescent probes for this application, including high biocompatibility and stability, relatively low cost and easy synthesis, a modular system such that improvements in the design of one DNAzyme sensor can easily be applied to others, and being able to take advantage of other advances in DNA nanotechnology which are constantly finding new applications and techniques in biological systems. As such, the advancement of DNAzyme catalytic beacon sensors from in vitro to biological applications can provide tools to answer many different biological questions regarding labile metal ions that cannot be answered with the currently limited number of existing metal sensors.

The first application of DNAzymes for intracellular metal sensing was achieved in 2013 by our group with the uranyl-dependent DNAzyme 39E which was attached to 13nm gold nanoparticles (AuNPs) through a thiol modification on the DNAzyme, allowing for efficient uptake in cells through endocytosis and thus localization in lysosomes (Figure 2a).¹⁰³ The DNAzymes were able to detect exogenously added uranyl, simulating exposure to uranium contamination, in comparison to non-treated cells or a negative control in which the RNA cleavage site of the substrate is replaced with a DNA base to block DNAzyme activity. In addition to detection of UO₂²⁺,^103–104 other DNAzymes since this initial study have also been used for the intracellular detection of Zn²⁺,^105–112 Mg²⁺,¹¹³ Pb²⁺,^{104, 108, 113–115} Cu²⁺,^{106, 111, 115–116} Na⁺,^{72, 117} and histidine,¹¹⁸ using a variety of different detection methods and delivery agents.

Figure 2.

Important advancements for DNAzyme-based metal sensors in cells. (A) First cellular detection of metal ions with a DNAzyme catalytic beacon probe (39E) by conjugation to AuNPs for imaging of UO₂²⁺ in cellular lysosomes. Reproduced with permission from reference 103. Copyright 2013 American Chemical Society. (B) A photocaging strategy for the catalytic beacon to inactivate DNAzyme cleavage by “caging” the RNA 2ʹ-OH with a nitrobenzyl group. 365nm light irradiation (hν) is used to restore the 2ʹ-OH of the rA cleavage site. The DNAzyme is then able to cleave its substrate in the presence of its specific metal ion (M²⁺), producing the fluorescence signal. Reproduced with permission from reference 105. Copyright 2014 Wiley-VCH. (C) Amplification strategy to increase the DNAzyme probe sensitivity using Catalytic Hairpin Assembly (CHA). Release of the cleaved substrate strand initiator (I, red) generated by metal-dependent DNAzyme activity allows for toehold-mediated strand displacement opening of hairpin 1 (H1, blue) which can further open the reporter hairpin 2 (H2, green), separating the fluorophore (F) and quencher (Q) to produce the fluorescent signal. Reproduced with permission from reference 117. Copyright 2017 Wiley-VCH.

In addition to intracellular sensing of metals and metabolites, DNAzymes have also been used for sensing and regulation of RNA in cells (e.g. mRNA, miRNA, etc.) to control gene expression by cleaving only specific RNA sequences.¹¹⁹ Similarly, DNAzymes have been used as an intracellular amplification strategy for nucleic acid targets, such as for miRNA, synthetic DNA circuits as used in DNA computers, or readouts from other functional DNA sensors such as aptamers.^{101, 120} Though techniques developed for these other nucleic acid therapies can likely be applied in future for the further development of intracellular DNAzyme sensors, in the following sections we will focus on the techniques currently used for intracellular metal sensing by DNAzymes, with a focus on strategies developed specifically to address several key issues facing practical applications of this new class of metal ion sensors.

Intracellular Delivery Methods of DNAzyme Sensors for Increased Cellular Uptake and Controlled Localization.

Metal Nanoparticles.

Since DNAzymes are highly negatively charged and are not readily uptaken by cells, the use of different transfection agents for the delivery of DNAzyme catalytic beacon probes is crucial to obtain sufficient intracellular delivery yield, as well as to control the final localization of the probes. As already noted in the above paragraph, conjugation of DNA to gold nanoparticles (AuNPs) allows cellular uptake via endocytosis and thus delivery to endosomes and lysosomes, and has been demonstrated for delivery of a number of DNAzymes in different mammalian cells.^{103, 106, 110, 116} This method is suitable for detection of labile metal in lysosomes and for DNAzymes which can work at this relatively low pH (pH ≈5–6), but other transfection reagents must be used to avoid or escape endocytosis if other cellular localization is desired. We have also demonstrated that gold nanoshells (AuNS) can be used for DNAzyme delivery, and it was found that the localized temperature increase produced by NIR light irradiation of the AuNS was sufficient to allow for endolysosomal escape and release of DNAzymes into the cytoplasm.¹⁰⁷ Other nanomaterials have also been used for DNAzyme delivery, most of which also utilize endocytosis-mediated transfection. For instance, graphene oxide (GO) nanosheets have been used for transfection of DNAzymes in MCF-7 cells, the major advantage being that DNA can adsorb onto the GO without any chemical conjugation methods.¹¹¹

Cationic Polymers.

In addition to these nanoparticle-mediated delivery methods, another commonly used transfection method makes use of cationic polymers, which can associate with or encapsulate the negatively charged DNA and are often reported to enable “endosomal escape”, allowing for cytosolic and nuclear delivery. One of the most widely used methods for DNAzyme transfection is the commercially available Lipofectamine, which utilizes positively charged lipids for cytoplasmic delivery.^{105, 115, 117} Similarly, cationic polypeptides, such as those containing guanidinium moieties, have also been used as a novel transfection method for delivery of DNAzyme probes to the cytoplasm.⁷² Similar methods using other polymers as transfection agents include dendritic polyethylene coupled to cationic poly(phenylene ethynylene) for cytosolic/nucleolar localization in HepG2 cells,¹¹⁴ encapsulation of DNAzyme probes in semipermeable polymethacrylic acid nanocapsules to deliver to MCF-7 cells,¹⁰⁸ and encapsulation of a DNA-based (G-quadruplex) hybrid catalyst in molecularly imprinted acrylamide nanogels for HeLa cell delivery.¹²¹

DNA Nanostructures.

DNA nanostructures have also been used as transfection agents, such that the enzyme and substrate strands of the DNAzymes are directly incorporated into the DNA nanostructure, without significantly affecting its catalytic activity, allowing for efficient delivery into cells while still maintaining their activity as metal-dependent probes. The main benefit of these systems is that no conjugation steps to attach other transfection agents to the DNAzymes are required, because the DNAzyme sequences are synthesized as part of the DNA nanostructure, which can then be assembled through hybridization of the different strands as normal. This strategy was first used with branching, Y-shaped DNA dendrimers for delivery of a histidine-dependent DNAzyme with relatively diffuse cytosolic delivery.¹¹⁸ Additionally, a larger three-dimensional DNA tetrahedron nanostructure has also been used for delivery of both UO₂²⁺ and Pb²⁺-dependent DNAzymes simultaneously, with punctate localization which wasn’t further characterized.¹⁰⁴

Other Methods.

As another unique transfection method, diacyllipid tails have been attached to DNA during solid phase synthesis as a way to localize DNAzymes to the cell membrane.¹¹³ This technique enables the immediate extracellular environment of the cells to be monitored to detect, for example, efflux of metal ions from the cells, and was as such demonstrated for detecting cellular efflux of both Mg²⁺ and Pb²⁺. Additionally, another recent paper utilized magnetic particles for delivery mediated by application of an external magnetic field.¹¹² The DNAzyme probes and reporter DNA sequences were conjugated to Fe₃O₄ particles which were further coated in polydopamine to increase their biocompatibility. Application of the magnetic field allowed the particles to move across the cell membrane and escape endocytosis, though the final localization of the probes was not examined in detail.

In addition to the techniques described here for delivery of DNAzymes as intracellular probes, techniques which have been tested for delivery of DNAzymes for gene regulation, as reviewed elsewhere,^122–123 could also be applied for DNAzyme probes. Similarly, because DNAzymes are chemically and structurally very similar to other nucleic acids, cellular delivery techniques which have been developed for other types of nucleic acid therapies, such as aptamers,¹²⁴ siRNA, and antisense oligonucleotides,^125–127 could also potentially be applied to the delivery of DNAzyme probes. The variety of transfection techniques used for DNAzymes highlights the necessity of determining the optimal transfection method for the desired application, especially regarding the final localization of the probe. As such, further development of these techniques to deliver probes to new cellular locations is a major ongoing area of research, with a short discussion on this topic in the Future Directions section.

Spatiotemporal Control of DNAzyme Sensors in Living Cells and In Vivo.

Because all metal-dependent RNA-cleaving DNAzymes are reactivity-based probes which rely on catalysis of the RNA cleavage reaction to generate a signal output (e.g. “high” fluorescence state), the signal generation is irreversible, so once activated, the sensors cannot return to the pre-cleaved state (e.g. “low” fluorescence state).¹²⁸ Because of this issue, it is highly desirable to be able to block or “cage” the reaction so that it cannot be initiated until the DNAzyme probe is delivered into the desired cellular location and is ready to be used. This caging strategy is especially important for intracellular detection because transfection times can range from 2–16 hours, so it is important to keep the DNAzymes from activating during this time. To address these issues, several caging strategies have been developed that will block RNA cleavage activity until the probes are decaged by some external stimuli, as described below.

Near-UV Light-based Activation.

The first such decaging strategy utilizes chemical modification of adenosine phosphoramidite with a 2ʹ-nitrobenzyl group,¹²⁹ which can then be incorporated into standard solid-phase oligonucleotide synthesis at the typical adenosine ribonucleotide cleavage site (Figure 2b).¹⁰⁵ The nitrobenzyl group effectively blocks the 2ʹ-hydroxyl of the RNA cleavage site which is required for cleavage activity. Upon irradiation with 365nm light, however, the nitrobenzyl group is released and the 2ʹ-hydroxyl group is restored or “decaged”, allowing for catalytic cleavage of the substrate strand by the DNAzyme.

While this strategy works well for both in vitro and intracellular detection of Zn²⁺, the synthesis of this phosphoramidite is non-trivial and currently not commercially available. Additionally, this method has so far only been demonstrated to photocage adenosine, whereas independent syntheses must be undertaken to obtain photocaged versions of the other ribonucleosides.¹³⁰ Because of this issue, other more general approaches to photocaging have also been developed. One such alternative method involves the post-synthetic modification of the DNA backbone using a photolabile thioether-enol phosphate modification.¹³¹ Its use as a photocage was demonstrated on the most commonly used DNAzymes, 8–17 and 10–23, by incorporating phosphorothioate bonds at three crucial nucleotides within the catalytic cores, such that when the DNAzyme strands were modified at these sites, they blocked the DNAzyme activity, most likely through steric effects that prevented the DNAzymes from folding properly. The photocaging groups were removed with 365nm light, restoring the phosphodiester bond, thus re-activating the DNAzymes. This method was successfully demonstrated to detect Zn²⁺ ions both in vitro and in HeLa cells. Although this post-synthetic modification method was easier than the pre-synthetic modification method and didn’t rely on different synthesis strategies for different nucleotide bases, it is also less general because the most effective blocking sites must be determined individually for each DNAzyme. Furthermore, it is not guaranteed to completely inactivate the DNAzyme. Other post-synthetic modifications which have been used for caging of RNA¹³² could also have potential for caging of the RNA site of DNAzymes.

Temperature-based Activation.

Another photocaging strategy relies on the localized temperature increase produced by irradiation of gold nanoshells (AuNS).¹⁰⁷ The DNAzyme probes are attached to the AuNS as a three-stranded DNA precursor, such that the third “linker” DNA strand both conjugates the DNAzyme to the AuNS and also hybridizes to the enzyme strand in such a way that it prevents full hybridization of the substrate strand, thus blocking DNAzyme activity. However, upon irradiation with NIR light, the AuNS emits the absorbed energy as heat to significantly increase the localized temperature. This temperature increase is enough to dehybridize the enzyme and linker strands, allowing the enzyme/substrate complex to break away from the AuNS and fully hybridize together, thus effectively reactivating the DNAzyme. The utilization of NIR light for photoactivation is a major advantage because it is both less cytotoxic and has better penetration in biological systems.

Near-IR Light-based Activation for In Vivo Systems.

The use of NIR light for photocaging has further been adapted for use with the nitrobenzyl-caged DNAzymes mentioned previously via conjugation to lanthanide-doped upconversion nanoparticles (UCNPs).¹⁰⁹ Upon irradiation with 980nm NIR light, these UCNPs are able to upconvert the irradiated NIR light into 365nm UV light which can decage the nitrobenzyl group on the DNAzymes. Furthermore, the UCNPs allow for efficient uptake by cells and can also act as a delivery agent for in vivo systems. As demonstration of this concept, this UCNP photocaging system was delivered to zebrafish embryo and larvae for imaging of Zn²⁺, representing a major advancement in the field as the first demonstration of DNAzyme sensors in vivo.

Increasing the Sensitivity of DNAzyme Sensors in Cells by using Amplification Strategies.

It has been shown by several groups that the most commonly used DNAzymes, 10–23 (Mg²⁺) and 8–17 (Zn²⁺), are not sensitive enough to detect endogenous levels of their respective labile metal ions in cells.^133–135 Thus, if DNAzyme sensors are truly to be used to detect biologically-relevant levels of metal ions, an improvement in sensitivity of these probes is required. Utilizing recent developments in the ever-growing field of DNA nanotechnology, several DNA amplification techniques have been demonstrated with DNAzyme probes based on protein-enzyme-free DNA circuits.^136–138 These methods typically rely on toehold-mediated strand displacement (TMSD) of separate hairpin strands and reporter strands, which contain the signal-generating fluorophores and quenchers. These strategies allow for a single DNA molecule to trigger the opening of multiple hairpin sequences, which in turn can trigger further opening of other reporter sequences to generate the fluorescent signal.

Even though these methods have been demonstrated extensively in test tubes and for the detection of RNA, their use with intracellular DNAzyme probes for metabolites has only been demonstrated within the past two years. In 2017, Wu et al. reported the use of one such amplification method called Catalytic Hairpin Assembly (CHA)¹³⁹ to amplify the fluorescent signal of a Na⁺ DNAzyme for improved intracellular sensing (Figure 2c).¹¹⁷ Cleavage of the DNAzyme substrate strand in the presence of Na⁺ allowed the shorter substrate fragment to dehybridize and act as the initiator (I) sequence to open a hairpin sequence (H1). Opening of H1 produces a large single-stranded overhang which can further open a second hairpin sequence (H2). H2 acts as a molecular beacon because it is modified on both ends with a fluorophore (F) and quencher (Q) which are adjacent to each other in the hairpin structure but are separated upon hairpin opening, thus generating the fluorescent signal. Furthermore, binding of H1 to H2 causes release of I, which can go on to open more H1 sequences, resulting in multiple H1 and H2 openings triggered by a single I molecule, which is the mechanism of signal amplification. With this CHA amplification strategy, the researchers were able to image endogenous Na⁺ levels in HeLa cells, which is the first reported detection of endogenous levels of any metal by a DNAzyme sensor.¹¹⁷

Around the same time, a strand displacement amplification technique was also reported to improve the sensitivity of a Pb²⁺ DNAzyme probe in MB-231 triple-negative breast cancer cells and in A-549 adenocarcinoma alveolar basal epithelial cells.¹⁴⁰ Another version of this strand displacement technique called Hybridization Chain Reaction (HCR)¹⁴¹ was also recently reported for the simultaneous detection of Zn²⁺ and Cu²⁺ in MCF-7 cells, with HCR amplification able to improve the detection limit by as much as two orders of magnitude.¹¹¹ HCR uses a similar mechanism as CHA, with DNAzyme-triggered release of an initiator sequence to open hairpins which produce the fluorescent signal, with the main difference being that instead of single H1/H2 duplexes, HCR incorporates overhangs in H1 and H2 sequences such that they form long linear multiplexes of multiple H1 and H2 sequences. Another amplification strategy which had previously been developed for other nucleic acid-based techniques and has recently been applied to DNAzymes is the use of toehold mediated strand displacement (TMSD) on so-called DNA machines.¹¹² In short, DNAzymes were attached to magnetic nanoparticles that were also conjugated with fluorophore-labeled reporter sequences, which were quenched by the magnetic nanoparticles. DNAzyme-triggered release of the substrate strand causes a cascade of up to three TMSD reactions all on the same magnetic bead, thus significantly amplifying the signal, allowing for increased sensitivity toward Zn²⁺ sensing. These and other methods which will likely be employed in future allow for significant increases in sensitivity which cannot be replicated with other small molecule or protein-based sensors, demonstrating another key advantage of DNAzyme sensors.

Important Considerations and Limitations of Intracellular Metal Imaging by DNAzymes

Labile Metal Ion Pools.

An important consideration for all recognition- and dynamic reactivity-based probes is to understand the metal ion pools that the sensors will be able to access. Unlike methods like ICP-MS or X-ray fluorescence microscopy, which measure the total metal content or concentration within a cell, fluorescence-based probes will only be able to detect the “labile” metal ion pool, or those metal ions that are less tightly bound by proteins or other molecules within the cell and have a high enough affinity for the added sensing probe.³⁷ Despite this limitation, reactivity-based probes can be beneficial in understanding labile metal ions in metal trafficking and homeostasis within the cell, which is often directly involved in the propagation of different disease states, such as in the production of harmful reactive oxygen species by labile iron.^14–15 It should also be noted that the labile metal ion pool will only significantly differ from the total concentration of a given metal species if it is highly sequestered in proteins, as is the case for most metal ion cofactors like iron, manganese, copper, and zinc. Because these metal ions can be beneficial or harmful, depending on factors such as their concentration and location, cells often need to tightly regulate these labile pools, making probing of these labile pools even more desirable. In contrast, other metal ions like sodium, potassium, and calcium do not typically strongly bind to proteins or other molecules and so their “labile” pool is very close to their total cellular content. As such, these fluorescent sensors can detect fluctuations in the overall concentrations of these ions, which can also be important to monitor because they are often related to different cell signaling pathways, as discussed in the Introduction.

DNAzyme Stability.

Because the cellular environment is known to contain many types of nucleases, stability of DNAzymes over the course of the imaging timeframe is a potential concern. Unlike linear or circular ssDNA and dsDNA, DNAzymes have been shown to form well-defined 3D structures like globular proteins and are thus much more stable and less subjected to degradation by endonucleases. Furthermore, since DNAzyme sensors are often modified on both the 5ʹ and 3ʹ ends by fluorophores and quenchers, these modifications can block degradation by most exonucleases. Therefore, DNAzymes have been shown to have much higher stability in biological systems than most other types of DNA molecules. To increase DNAzyme stability further, DNAzymes can be modified in such a way that they can no longer be recognized by cellular nucleases. For example, in one study, an all L-DNA version of the 8–17 DNAzyme was used, which significantly increased the DNAzyme stability since cellular nucleases are only able to recognize D-DNA.¹¹⁵ In addition, the use of 3ʹ-inverted nucleotides at the ends of DNA,¹⁴² and incorporation of three or more phosphorothioate linkages at the ends of DNA,¹⁴³ have also been shown to be effective at decreasing degradation by nucleases. Sometimes the transfection agents used for delivery of the DNAzyme sensors can also increase their DNAzyme cargo’s cellular stability, too. For instance, it has been demonstrated that when DNA is conjugated to AuNPs as described in the above section, they are at least partially protected from nuclease digestion by the localized high salt concentration resulting from the highly negatively charged particle surface.¹⁴⁴

DNAzyme Negative Controls.

One key advantage of DNAzyme-based probes is the ability to incorporate relatively simple negative control sensors into the experiment. Negative controls are especially important when interference from various factors which could potentially impact the fluorescence of the fluorophore could be an issue, such as pH and temperature fluctuations. The most common negative control is the non-cleavable substrate, which replaces the ribonucleotide cleavage site of the substrate strand with its respective deoxyribonucleotide, so that it can no longer be cleaved by the DNAzyme. By comparing fluorescent changes of the active probe with the non-cleavable substrate control, one can ensure that fluorescent changes are due to cleavage or dehybridization of the substrate strand, rather than fluorescent fluctuations due to any other phenomena. Additionally, an inactive enzyme negative control can also be employed if unwanted metal-independent cleavage of the substrate is a possible concern. This works by using an inactive form of the enzyme, which is often found from mutagenesis studies of DNAzyme sequences where just one or two specific base mutations will completely inactivate the DNAzyme activity. A normal cleavable substrate can then be used with the inactive enzyme, such that even in the presence of the metal ion target, it shouldn’t be cleaved due to the DNAzyme. This strategy enables the researcher to determine if other factors in the cellular environment are cleaving the substrate strand independent of DNAzyme activity. Furthermore, when a caging strategy is used, as described previously, samples which are not subjected to the decaging method (e.g. not irradiated with 365 nm light) can also be used as a negative control.

Single Fluorophore Intensity-Based Probes.

Although catalytic beacon sensors based on DNAzymes have been demonstrated for the selective fluorescent detection of metal ions in cells,^{72, 103, 105} as discussed in the previous sections, all such examples reported to date have been based on the fluorescent signal of a single fluorophore. While these sensors can have excellent dynamic range and have thus been widely used for the study of cellular metal ions,^{33–34, 36, 145} the use of a single fluorophore readout can make obtaining quantitative information about metal ions in cells difficult for two reasons.¹⁴⁵ First, many species in cells and tissues, such as tryptophan and nicotinamide adenine dinucleotide (NAD), display intrinsic fluorescent signals and this resulting autofluorescence can fluctuate depending on time and location. As a result, the fluorescent signals from the above DNAzyme sensors are vulnerable to interference from this background fluorescence. Second, cellular uptake of any sensors, including DNAzyme sensors, cannot be assumed to be uniform between cells, or even between different locations of the same cell. As a result, any observed change in fluorescence may reflect either changes in concentration of the target metal ion or uneven loading of the sensor. This limitation is more pronounced when cellular uptake involves active processes, as is usually the case for DNAzyme delivery.^{45, 92, 146}

Overcoming a Major Challenge in Cellular Imaging of Metal Ions: Development and Demonstration of Ratiometric DNAzyme Sensors for Quantitative Metal Ion Sensing

Ratiometric FRET-Based Sensors.

To overcome the limitations of single fluorophore-based probes in order to attain quantifiability of the target analyte, a common method has been the use of ratiometric sensors that incorporate a second signaling moiety that remains invariant in the presence of the target analyte.¹⁴⁵ A particularly effective ratiometric strategy for cellular imaging utilizes Förster Resonance Energy Transfer (FRET),¹⁴⁷ which has been widely used for sensors based on fluorescent proteins.^148–150 With FRET-based sensors, energy transfer from one fluorophore (the FRET donor) to another (the FRET acceptor) occurs in a distance-dependent manner.¹⁴⁷ As a result, either the FRET efficiency¹⁵¹ or a simpler FRET ratio (e.g., the ratio of acceptor fluorescence to donor fluorescence with consistent excitation of the donor only) can be calculated. Since the FRET ratio is not vulnerable to cellular autofluorescence or variations in sensor concentrations due to different uptake of sensors by different cells, FRET sensors are considered quantitative.^{147, 152}

A major advantage of DNAzyme sensors is that different fluorophores and other modifications can be readily conjugated to the ends of the DNAzymes during the process of DNA synthesis and, because these modifications are almost always located away from the metal-binding sites, they do not usually interfere with either metal binding or DNAzyme activity.⁴⁴ In this work, we present the design and demonstration of a ratiometric FRET DNAzyme for quantitative detection of metal ions by attaching two fluorophores to the DNAzymes, and the use of such a system for cellular detection and quantification of labile metal ions.

An overview of the ratiometric DNAzyme approach is shown in Figure 1b, using the 8–17 DNAzyme as an example. Previous work with DNAzyme-based sensors have typically used the catalytic beacon design, shown in Figure 1a. In this design, a fluorophore (F) on the RNA-containing substrate strand is quenched by proximity to a quencher such as Dabcyl or Black Hole Quencher (Q). In the absence of the target metal ion, the uncleaved substrate strand remains hybridized to the enzyme strand at two binding arms flanking the catalytic core of the DNAzyme, because the melting temperature of the hybridization is higher than the intended ambient temperature (usually either room temperature or 37°C). This hybridization results in the quenchers being close to the fluorophore, thus producing minimal fluorescent signal. The presence of target metal ions induces substrate cleavage by the DNAzyme, and since each of the two fragments of the cleaved product has a lower melting temperature, the two fragments will dehybridize from the DNAzyme. This dehybridization separates the fluorophore-containing fragment from the quenchers to produce a turn-on fluorescence signal, signaling presence of the target metal ion.

In the ratiometric approach (Figure 1b), two fluorophores comprising a FRET donor (D) and acceptor (A) are placed adjacent to each other on opposite strands. As FRET efficiency is highly dependent on the distance between the two (i.e., the FRET efficiency is proportional to 1/r⁶, where r is the distance between the two fluorophores), when the enzyme and substrate strands are annealed together, excitation of the FRET donor will allow energy transfer, producing a fluorescence output from the FRET acceptor. In the presence of target metal ion, however, the DNAzyme cleaves the substrate strand into two product fragments, resulting in lower melting temperatures of the two fragments to the enzyme strand. The dehybridization of the substrate strand after metal-induced cleavage should be reflected by a concomitant loss of FRET, i.e., significantly reduced fluorescence from the acceptor and significantly increased fluorescence from the donor. Importantly, while the intensity of fluorescence from either fluorophore alone will necessarily depend on sensor concentration, the FRET ratio will not. This design allows ratiometric DNAzymes to be used for quantification even when the exact sensor concentration is not known, as is typically the case within cells.

In Vitro Activity Assays and Zn²⁺ Quantification.

The photophysical properties of the ratiometric DNAzyme sensor are shown in Figure 3a. When the Cy3 and Cy5 dyes are used as a FRET pair on the Zn²⁺-responsive 8–17 DNAzyme (Figure 1b), upon excitation of Cy3 at 490 nm, emission of Cy5 at 666 nm is observed, suggesting FRET has occurred between the Cy3 donor and Cy5 acceptor (Figure 3a, black trace). After addition of Zn²⁺, a decrease of Cy5 emission was observed, with concomitant increase in Cy3 emission at 570 nm under the same excitation conditions (Figure 3a), indicating that the FRET has changed. A time-course analysis of the fluorescence changes at these two wavelengths (Figure 3b) shows that the increase in Cy3 emission and decrease in FRET-induced Cy5 emission occur simultaneously and are complete within 5 minutes upon addition of 50 μM Zn²⁺, in agreement with typical results using the catalytic beacon form of the 8–17 DNAzyme. The initial rate of the signal change was calculated using established reaction kinetics,¹⁵³ and was plotted at multiple Zn²⁺ concentrations (Figure 3c) with a linear relationship over this Zn²⁺ concentration range. Based on the linear fit of this relationship, a limit of detection (LOD) of 12 μM Zn²⁺ was obtained, based on the common LOD definition: 3 × σ_blank.

Figure 3.

Photophysical properties and FRET response of ratiometric FRET sensor. (A) Fluorescence emission of 8–17 DNAzyme containing proximal Cy3 and Cy5, under excitation of Cy3 at 490 nm. Initial fluorescence (black trace) shows both Cy3 emission (at 570 nm) and Cy5 emission via FRET (at 666 nm). After addition of 100 μM Zn²⁺ (red traces), the Cy3 peak increases in intensity while the Cy5 peak decreases. (B) Time-course of direct Cy3 emission (black) and FRET-induced Cy5 emission (red) upon addition of 50 μM Zn²⁺. (C) Calibration curve between calculated initial rate of change (k_obs) and Zn²⁺ concentration.

Similar work was carried out using the FRET pair fluorescein (FAM) and tetramethylrhodamine (TAMRA) (Supplemental Figure S1). FRET was again observed by emission of the FRET acceptor TAMRA upon excitation of the FRET donor FAM. However, while quantitation was again possible using the ratio of two fluorophore emissions, significant overall quenching was observed when the dyes were in close proximity. This quenching is most likely due to formation of excitonic states caused by collisional contact of donor and acceptor, which is a known property of xanthene-based dyes such as fluorescein and rhodamine.¹⁵⁴ An alternative strategy to minimize collisional quenching was further investigated (Supplemental Figure S2), in which the substrate strand was modified with both donor and acceptor dyes on opposite fragments of the substrate, using an internally-modified dT-TAMRA base. This construct showed reduced FRET efficiency, but no collisional quenching, and could also be used for quantification.

Intracellular Detection of Zn²⁺.

Having demonstrated that the Cy3/Cy5 FRET pair in the 8–17 DNAzyme can be used to quantify Zn²⁺ in buffer, we investigated this ratiometric DNAzyme sensor for quantification of Zn²⁺ in HeLa cells. The cells were transfected with the DNAzyme sensor using Lipofectamine 3000, treated with different concentrations of Zn²⁺ by incubating cells with zinc chloride and zinc pyrithione in a 2:1 ratio, following a protocol reported previously,¹⁰⁹ and imaged via confocal laser-scanning microscopy (CLSM). Prior to imaging, the cells were irradiated with 365nm light for 30 minutes to decage the DNAzymes. As shown in Figure 4a, excitation at 514 nm produced emissions in channels corresponding to both Cy3 and Cy5, indicating a good FRET efficiency within the cells and the successful delivery of intact 8–17 DNAzymes. To avoid confusion, the channel with Cy3 excitation and Cy5 emission is called the “FRET” channel, to distinguish it from Cy5 excitation with Cy5 emission, which is called the “Cy5” channel. For samples prepared with different concentrations of Zn²⁺, increasing Zn²⁺ levels corresponded with a decrease of FRET signal and an increase of Cy3 signal (Figure 4a). The fluorescence intensities were quantified by calculating the FRET ratio (defined as FRET/Cy3) and a linear relationship between the FRET ratio and added [Zn²⁺] was observed, indicating that the ratiometric DNAzyme has been used to more accurately quantify metal ions in living cells for the first time. Furthermore, a time-course kinetic assay could be conducted with cells by monitoring the same cells both before and after adding Zn²⁺. With 50μM added Zn²⁺, the calculated FRET ratio gradually decreased over time up to 50 minutes after adding Zn²⁺ (Supplementary Figure S4), further demonstrating that the ratiometric DNAzyme can be used to probe both the presence of metal ions and their concentration changes intracellularly.

Figure 4.

Intracellular imaging of zinc with a ratiometric DNAzyme sensor. (A) HeLa cells were treated with the indicated concentrations of Zn²⁺, transfected with the ratiometric Zn²⁺ DNAzyme sensor via Lipofectamine 3000, and then decaged with 365nm irradiation directly before imaging. The Cy3 (green) donor channel increases with increasingly added Zn²⁺ while the FRET (red) channel decreases with increasingly added Zn²⁺. Scale bars 50 μm. (B) Fluorescence intensities from confocal images were quantified to calculate the FRET ratio (FRET/Cy3) from at least three representative images for each condition and plotted (black) in comparison to replicate samples which were not decaged with 365 light (red) as a negative control. Confocal images of the non-decaged samples and comparisons with the Cy5 channels are in Supplemental Figure S3.

Conclusion and Future Directions

In this forum article, we have summarized recent progress in developing DNAzyme-based sensors for probing metal ions in living cells and in vivo, such as in zebrafish. These recent advances were made due to our ability to overcome a few major barriers to the application of DNAzymes as intracellular metal sensors, including delivery of DNAzyme sensors intracellularly and in vivo, spatiotemporal control of DNAzyme activity, and signal amplification for detection of endogenous metal ions in biological systems. Furthermore, we have presented a new design and results of a FRET-based catalytic beacon sensor that can convert the single intensity-based readout of existing DNAzyme sensors into a ratiometric readout without compromising the fundamental catalytic function of the DNAzyme. This FRET-based ratiometric DNAzyme design can be readily applied to other DNAzyme sensors as a major advance in the field to enable a much more quantitative readout for intracellular metal sensing by DNAzyme sensors.

Though the field of intracellular DNAzyme sensors has advanced substantially since its initial demonstration six years ago, there are still significant advancements that can be made to address several more key issues. For example, while this FRET ratiometric method offers significant improvements over single intensity-based measurements, which can suffer from high interference and variable transfection efficiencies in biological systems, it can still suffer from other limitations of intensity-based imaging. For instance, the filters and laser configuration on a typical microscope set-up only allow for imaging of 4 or 5 distinct fluorophores at a time with intensity-based imaging, but the number of simultaneous sensors can be increased significantly with lifetime-based fluorescence imaging. Additionally, anisotropy-based fluorescence imaging is also an interesting potential technique which could help alleviate issues with non-specific quenching by several transition metals¹⁵⁵ which has made their detection with fluorescent probes difficult. Another improvement to standard fluorescent imaging techniques involves the use of near-IR (NIR) fluorophores, which have already been used in place of more traditional visible light fluorophores for one DNAzyme probe because of the increased penetration depth of NIR light in biological tissue.¹¹⁰ Similarly, non-fluorescent based sensing modalities could also prove useful in future studies, as demonstrated in the recent example of a plasmon resonance energy transfer probe based on DNAzymes conjugated to AuNPs for detection of Cu²⁺ in cells.¹¹⁶ Imaging modalities that have been developed for other nucleic acid-based probes could also be adapted for use with DNAzyme sensors. One prominent example is the development of photoacoustic aptamer probes,¹⁵⁶ which could further help to improve the utilization of DNAzymes with in vivo systems due to the much higher penetration depth of photoacoustic signals in comparison to fluorescence.

One of the major remaining issues of using DNAzymes for metal sensing in live cells is that, since they inherently rely on reactivity to generate their signal, they are essentially irreversible and so cannot return to an “off” state after being turned “on”.¹²⁸ Though photoactivation as described previously can help to alleviate this issue, the typical decaging time of up to 30 minutes with the current nitrobenzyl group presents issues as well, since it is very difficult to detect fluctuations faster than this amount of time, though new caging strategies could potentially shorten this time. Another purported strategy is to turn metal-dependent DNAzymes into metal-binding aptamers which can thus be used as reversible sensors similar to molecular beacons, as has been demonstrated for Ag^{+ 157} and Na⁺.¹⁵⁸ This strategy relies on removing the catalytic function of the DNAzyme to produce a signal purely based on metal binding, and although it has worked well for monovalent-dependent DNAzymes for which metal binding usually produces conformational changes of the DNAzyme to induce activity, it is unclear if this strategy could be translated to redox metal-dependent DNAzymes in which the metals likely play a role in the catalytic reaction itself to induce activity and may not produce much change in the overall DNAzyme structure.

One final key advancement in the field will be toward improved control over the final localization of DNAzyme probes in intracellular compartments. Since probes obviously won’t be able to image metals in locations that they themselves cannot reach, it is crucial to develop new strategies to deliver probes to new locations in the cell, such as to mitochondria, the Golgi apparatus, or endoplasmic reticulum, in order to enable these sensors to answer more complicated biological questions involving these cellular compartments. This goal can potentially be accomplished by utilizing targeting reagents which have been developed for targeted sub-cellular delivery of other nucleic acid agents, such as conjugation to naturally-derived targeting peptides^159–160 or synthetically-developed aptamers.¹⁶¹ All told, these and other further advancements in the field of DNAzyme sensors can provide significant improvements in the detection of metal ions in cells, which can help to further elucidate the complex role of metal ions in biological systems.