In Situ Genetically Cascaded Amplification for Imaging RNA Subcellular Locations

Abstract

In situ amplification methods, such as hybridization chain reaction, are valuable tools for mapping the spatial distribution and subcellular location of target analytes. However, the live-cell applications of these methods are still limited due to challenges in the probe delivery, degradation and cytotoxicity. Herein, we report a novel genetically encoded in situ amplification method to non-invasively image the subcellular location of RNA targets in living cells. In our system, a fluorogenic RNA reporter, Broccoli, was split into two non-fluorescent fragments and conjugated to the end of two RNA hairpin strands. The binding of one target RNA can then trigger a cascaded hybridization between these hairpin pairs and thus activate multiple Broccoli fluorescence signal. We have shown that such in situ amplified strategy can be used for the sensitive detection and location imaging of various RNA targets in living bacterial and mammalian cells. This new design principle provides an effective and versatile platform for tracking various intracellular analytes.

Graphical Abstract

INTRODUCTION

With high sensitivity and low detection limit, in situ amplification methods have emerged as a powerful technique for measuring the cellular concentration and subcellular location of target biomolecules.^1–4 For example, fluorescence in situ hybridization (FISH) is one of the gold standard techniques to reveal the cellular distributions of RNA and DNA molecules.^5–8 By conjugating with in situ amplification methods, such as hybridization chain reaction (HCR),^9–12 signal amplification by exchange reaction (SABER),^13,14 and click chemistry-based amplification (clampFISH),¹⁵ the sensitivity and efficiency of FISH has been dramatically improved. However, these techniques require the delivery and multi-step washing of fluorophore-modified synthetic oligonucleotides. As a result, most of these methods are currently only applicable for fixed-cell imaging.^8,16,17

In situ DNA-based amplification methods have also been used for live-cell imaging.^18–20 However, these synthetic DNA probes are still suffering from challenges including enzymatic degradations, difficulties in the probe delivery, and potential biosafety risks.⁴ In contrast, we believe RNA-based amplification methods will be beneficial for live-cell applications. These RNA-based tools can be genetically encoded and directly transcribed inside living cells, which may be potentially useful for in situ signal amplification.

Fluorogenic RNA aptamers, such as Spinach or Broccoli, are specific RNA strands that can selectively bind and activate the fluorescence signal of nontoxic, cell-permeable small-molecule dyes.^21–23 Based on these fluorogenic RNAs, a series of biocompatible and genetically encoded sensors have been developed for non-invasive live-cell imaging of RNAs, proteins, small molecules, and metal ions.^24–32 However, without tagging multiple aptamers or achieving signal amplification, it remains a challenge for imaging low-abundance targets, especially in mammalian cells.^4,33,34

In order to improve the sensitivity of fluorogenic RNA-based sensors, we have recently developed a signal amplification circuit based on RNA catalytic hairpin assembly, termed CHARGE, for low-abundance RNA imaging in living bacterial cells.³⁵ In this system, RNA target functioned as a catalyst and triggered the assembly of multiple RNA hairpins. However, after the reaction, target RNA was disconnected from the reporters. As a result, CHARGE cannot be used to track the subcellular location of the targets.

In this study, we developed an IN SItu Genetic Hybridization Amplification Technique, termed INSIGHT, for the sensitive imaging of the subcellular distributions and locations of RNA targets in living cells. Our approach is based on a combination of fluorogenic RNA reporters with RNA-based hybridization chain reaction (HCR). HCR is an enzyme-free and isothermal in situ amplification technique,^3,10 which has been applied for detection of a wide variety of analytes, including proteins, nucleic acids, and small molecules.^36–38

In our INSIGHT design, in the absence of initiator RNAs, two types of RNA hairpins (H1 and H2) coexisted under a metastable state (Figure 1). Upon the addition of an initiator RNA, the H1 hairpin will be opened and expose a sequence that can invade and hybridize with the H2 hairpin. A cascaded hybridization between these two hairpins was then initiated and generated a double-stranded concatemeric nanostructure.

Figure 1.

Schematic of the genetically encoded INSIGHT system. After the cellular transcription of the H1 and H2 strands (blue) from a delivered vector, the binding of initiator RNA (red) triggers the cascaded H1/H2 hybridization. As a result, the reassembly of Broccoli (green) can be used for imaging the subcellular location of the initiator strand.

We have further split RNA Broccoli into two non-fluorescent fragments and separately conjugated them to each end of the H1 and H2 hairpins (Figure S1). These two fragments cannot be spontaneously hybridized to activate Broccoli fluorescence (Figure S2).²⁴ In the presence of an initiator RNA, the cascaded H1/H2 hybridization induced the reassembly of a chain of Broccoli and thus activated an amplified fluorescence signal (Figure 1). Importantly, the generated chain of Broccoli will be associated with the initiator strand, which can be used for the cellular tracking. Herein, we demonstrated that such an in situ amplification method can be genetically encoded and used for the detection and location imaging of various RNA targets in both bacterial and mammalian cells.

RESULTS

Design and Optimization of the INSIGHT System.

We first asked how we can split the Broccoli RNA to display weak fluorescence signal when separately labeled at the 5’- and 3’- end of the hairpin, and meanwhile, upon the initiator-triggered H1/H2 hybridization, reassemble and activate the fluorescence of a small-molecule dye, 3,5-difluoro-4-hydroxybenzylidene-1-trifluoroethyl-imidazolinone (DFHBI-1T) (Figure 1).

We started by testing a design of Broccoli splits similar as that was used in our recent development of CHARGE (Figure S1a). Two Broccoli fragments were then conjugated to each end of an H1/H2 hairpin pair. This hairpin pair has been previously used in developing HCR.³⁹ After we in vitro synthesized these RNA strands (termed D1), however, a high background signal was observed, even in the solution containing only H1 or H2 (Figure S1b). We realized that such background signal was likely due to the spontaneous assembly of Broccoli within each hairpin. To reduce this background, we decided to decrease the number of base pairs between two Broccoli fragments (D2). Indeed, the background was dramatically reduced in D2 but limited signal enhancement was observed after adding the initiator (Figure 2a and S1c).

Figure 2.

Optimization and in vitro characterization of the INSIGHT system. (a) The fluorescence response of different sequence designs was measured after 4 h incubation in a solution containing 0 or 50 nM initiator RNA, 250 nM H1 and H2 (or Broccoli), and 5.0 μM DFHBI-1T. (b) Calibration curve for the initiator RNA detection. The horizontal line indicated the fluorescence intensity for estimating the limit of detection of the system. Shown are mean and SEM values of three independent replicates.

To increase the signal-to-background (S/B) ratio, we came up with a new sensor design principle by blocking two (D3) or four (D4) nucleotides of Broccoli within the stem region of H1 and H2 hairpins (Figure S1d and S1e). In the absence of initiator, these blocked nucleotides were not able to spontaneously assemble into Broccoli, i.e., a low background signal. While upon the addition of initiator RNA, the cascaded assembly of H1 and H2 triggered the release of these blocked nucleotides and exhibited high fluorescence activation. Indeed, compared with D1 (1.5-fold) and D2 (1.2-fold), an increased fluorescence enhancement was observed for D3 (1.8-fold) and D4 (2.1-fold) (Figure 2a).

We realized that even though the background fluorescence has been effectively reduced in solution containing either H1 or H2, after mixing these two hairpin strands, an obvious increase in the fluorescence signal was shown (Figure S1). One potential reason for such undesired signal should be the spontaneous unfolding of H1 or H2 hairpin. To further improve the stability of both hairpins, we tried to increase the stem lengths of H1 and H2. Indeed, an optimal design of RNA hairpin pairs (D5) was achieved in this way, with low background and large signal enhancement (3.0-fold) in the presence of initiator (Figure 2a). D5 was thus selected to conduct the subsequent experiments.

In Vitro RNA Detection with INSIGHT.

After optimizing the sequences of INSIGHT, we next tested the performance of this in situ amplification approach under different environmental conditions. We first studied the effect of different buffers on the fluorescence signal, including HEPES, Tris-buffered saline, and phosphate-buffered saline (Figure S3a). INSIGHT can be used for initiator RNA detection in all these buffers. The largest S/B ratio was observed in the HEPES buffer, which was used in the subsequent in vitro tests. Magnesium ion (Mg²⁺) concentration is another important parameter affecting RNA folding and the efficiency of hybridization chain reactions. An enhanced INSIGHT signal was observed after increasing Mg²⁺ concentration in the range of 1–10 mM (Figure S3b). However, the background signal was also enhanced at the same time. A similar S/B ratio was observed in the presence of 1 mM (2.2-fold), 5 mM (2.3-fold), or 10 mM (2.3-fold) Mg²⁺ ions (Figure S3b), indicating that INSIGHT can function well in the presence of physiological concentrations of Mg²⁺ ions (~1–2 mM).^40,41

We have further studied the effect of temperature and incubation time on the INSIGHT system. At both 22°C and 37°C, the initiator-induced INSIGHT signal gradually increased over the 5 h incubation time, with 90% of maximum fluorescence observed after ~2.5 h incubation (Figure S3c and S3d). The optimal S/B ratio was achieved after 4 h incubation at 22°C (3.1-fold) or 3 h incubation at 37°C (2.9-fold). These results together indicated that INSIGHT can be used for detection of initiator RNA under different conditions.

We next asked if INSIGHT can be used for detecting low-abundance initiator RNA. As shown in Figure 2b, after incubating various concentrations of initiator, 2–250 nM, with 250 nM H1/H2 (D5) and 5.0 μM DFHBI-1T, a dose-response curve was generated. A clear initiator-induced fluorescence activation was observed, with a detection limit of 1.9 nM. As a control, under the same condition, the minimal detectable amount of Broccoli was 30.2 nM (Figure S4), a 15-fold difference from that of INSIGHT. We have further used a native gel electrophoresis assay to characterize the product of hybridization chain reactions (Figure S4). Indeed, long RNA product bands could be visualized only in the presence of initiator RNA, even when the initiator concentration was 50-fold lower than that of the hairpins (Figure S5). Indeed, INSIGHT can be used to detect low-abundance RNAs.

RNA Imaging in Bacterial Cells with INSIGHT.

We next applied INSIGHT for RNA imaging in live E. coli cells. We first cloned the H1 and H2 sequences into a dual-expression pETDuet vector, and the initiator sequence in a pCDFDuet vector. After transformation and isopropyl β–ᴅ−1-thiogalactopyranoside (IPTG) induction, an intense cellular fluorescence signal was observed (Figure 3). This signal was comparable with that of Broccoli-expressing cells. As a control, cells that were transformed with only H1 or H2, or the mixture of H1 and H2 (H1/H2), displayed minimal fluorescence intensity (Figure 3). We have further determined the cellular fluorescence levels from 300 individual E. coli cells in each case. On average, compared with cells expressing H1/H2, a 3.0-fold signal enhancement was observed in the presence of initiator RNA (H1/H2/I). Indeed, INSIGHT can be activated for the cellular imaging of RNAs.

Figure 3.

Confocal fluorescence imaging of BL21 (DE3)* E. coli cells expressing, respectively, pETDuet-H1 (H1), pETDuet-H2 (H2), pETDuet-H1/H2 (H1/H2), pCDFDuet-initiator + pETDuet-H1/H2 (H1/H2/I), and pETDuet-Broccoli (Broccoli). Scale bar, 20 μm. Distribution of cellular fluorescence levels was measured from a total of 300 cells in each case. The dashed line indicated the averaged background signal as measured in cells expressing H1/H2.

We also wanted to study the kinetics of INSIGHT in living cells. After IPTG induction, we imaged the cells expressing H1/H2 or H1/H2/I every one hour during a total of 4 h incubation (Figure S6). In the presence of initiator RNA, INSIGHT signal can be visualized after only 1 h incubation, while the maximum cellular fluorescence was shown in 3 h. Minimal H1/H2 fluorescence was observed in the same period. While longer incubation time will indeed enhance the background signal of the spontaneous hybridization between H1 and H2 (Figure S6). As a result, ~3 h incubation was optimal with enhanced S/B ratio.

We next asked if we can develop a modular INSIGHT system to image different RNA sequences of interest. For this purpose, an RNA molecular beacon was introduced into the system (Figure 4a). The loop sequence of the molecular beacon can be changed to hybridize with different RNA targets. These molecular beacons also contained a conserved stem region. Part of the initiator strand was blocked in this stem. Target binding induced the dehybridization of the stem, activated the initiator RNA, which further induced a cascaded hybridization between H1 and H2. As a result, the same sequence of H1 and H2 can be used to detect various RNA targets.

Figure 4.

A modular INSIGHT system for imaging SgrS. (a) Schematic of a general INSIGHT platform for target RNA detection. The target RNA hybridizes with the molecular beacon and releases an initiator strand (red) to trigger the hybridization chain reaction between H1 and H2. (b) Optimizing the number of stem base pairs in the molecular beacon. Fluorescence signal was measured after 4 h incubation in a solution containing 0 or 20 nM SgrS (or initiator for H1/H2), 250 nM H1 and H2 (or Broccoli), 250 nM molecular beacon, and 5.0 μM DFHBI-1T. Shown are mean and SEM values of three independent replicates. (c) Confocal fluorescence imaging of BL21 (DE3)* cells expressing pETDuet-H1/H2 and pCDFDuet-SgrSMB-7B (SgrS-targeting molecular beacon). Images were taken after 3 h incubation in the presence of 0, 1 g/L, or 20 g/L glucose. Scale bar, 10 μm. Distribution of cellular fluorescence levels was measured from a total of 300 cells in each case. The dashed line indicated the averaged background signal as measured in cells without glucose induction.

To test this idea, we first designed a molecular beacon to detect SgrS, a small RNA involved in bacterial glucose transportation.⁴² Three molecular beacons containing 7, 8, or 10 base pairs in the stem region were in vitro synthesized, and then the corresponding INSIGHT signal was measured in the presence or absence of 20 nM SgrS (Figure 4b). Based on the large S/B ratio, an optimal molecular beacon structure with a 7-base-pair stem (7B) was identified for further studies.

We next asked if we can image SgrS RNA in living E. coli cells. The 7B molecular beacon was cloned into a pCDFDuet vector and then co-transformed with the pETDuet-H1/H2 vector into BL21 (DE3)* cells. It is known that the cellular expression of SgrS can be induced by adding glucose.⁴³ Indeed, in the presence of 0–20 g/L glucose, an enhanced cellular fluorescence was observed with the increasing amount of glucose (Figure 4c). As a control, after adding the same concentration of glucose, there is almost no change in the fluorescence signal from cells expressing Broccoli or H1/H2/I (Figure S7).

We have further used real-time PCR to quantify the cellular level of SgrS RNA after adding glucose. Consistent with some previous studies,⁴⁴ our results indicated that 24–70 nM concentration of SgrS was generated in the presence of 0–20 g/L glucose (Figure S8). Based on the corresponding cellular fluorescence (Figure 4), INSIGHT can be used for the sensitive imaging of low-abundance (~30 nM) target RNA in living cells.

INSIGHT-based Imaging of the Subcellular Location of Target RNAs.

Our next goal is to study if we can use INSIGHT to study the subcellular distribution of target RNAs. We chose SgrS and a bglF mRNA as the targets, since the cellular location of both of these two RNAs have been well characterized.^44,45 By simply changing the loop region of the molecular beacon, an INSIGHT system was developed to detect bglF RNA using the same H1/H2 hairpins (Figure S9).

After expressing the SgrS- or bglF-targeting INSIGHT system in BL21 (DE3)* cell, a 4’,6-diamidino-2-phenylindole (DAPI) dye was also added to stain the location of the nucleoid. Consistent with previously reported FISH-based images,^44,45 the SgrS RNA predominantly located in the cytoplasm (71.1%) in E. coli cells, whereas the transcripts encoding bglF were detected mainly (82.5%) nearby the cell membranes (Figure 5). Both fluorescence signals exhibited little overlap with that of DAPI (Figure S10). As another control, the fluorescence signal of similarly transcribed Broccoli was homogeneously distributed throughout the whole cell region (Figure 5 and S10). All these results suggested that INSIGHT can indeed be used to study the spatial distribution and subcellular location of target RNAs.

Figure 5.

Imaging the subcellular location of SgrS and bglF in BL21 (DE3)* cells. Confocal fluorescence imaging of cells expressing pETDuet-H1/H2 + pCDFDuet-SgrSMB-7B or pCDFDuet-bglFMB-7B (molecular beacon that targets SgrS or bglF), or pETDuet-Broccoli. DAPI was used to stain the nucleoid of E. coli. Scale bar, 5 μm.

RNA Imaging in Mammalian Cells with INSIGHT.

After demonstrating the performance of INSIGHT in bacterial cells, we next asked if this system can also be used for RNA imaging in mammalian cells. We first cloned the H1, H2 and initiator sequences into an AIO-Puro vector (H1/H2/I), which contains three U6 promoter regions. After transfecting this vector into HeLa cells, obvious cellular fluorescence can be visualized (Figure S11). On average, cells expressing H1/H2/I (activated INSIGHT) exhibited 2.4-fold higher fluorescence signal than those expressing H1/H2, and 2.2-fold brighter than those expressed H1/H2/random sequence (Figure S11). Indeed, INSIGHT can be potentially used for RNA imaging in mammalian cells.

We then applied INSIGHT for imaging endogenous mRNAs in mammalian cells. We first developed a molecular beacon probe (termed SMB) to target survivin mRNA (Table S2). The resultant survivin-targeting INSIGHT exhibited a similar level of fluorescence enhancement as that induced by the initiator strand (Figure 6a). In contrast, when using a control molecular beacon of random loop sequence (C-SMB), little fluorescence enhancement was observed after adding survivin RNA.

Figure 6.

(a) Fluorescence signals were measured in a solution containing 0 (blank) or 20 nM survivin RNA (or initiator), 250 nM H1 and H2 (or Broccoli), 250 nM molecular beacon, and 5.0 μM DFHBI-1T. Shown are mean and SEM from three replicated experiments. (b) Confocal fluorescence imaging in HeLa cells at 24 h after Lipofectamine® 3000-based transfection. Images were taken in the presence of 40 μM DFHBI-1T. 5 nM YM155 (inhibitor) was added to inhibit the generation of survivin mRNA. Scale bar, 10 μm. (c) Mean cellular fluorescence intensities in the experiments as shown in panel (b). Shown are mean and SEM from ten individual cells.

We next cloned SMB, together with H1 and H2, into the above-mentioned AIO-Puro vector. After transfecting into HeLa cells, fluorescence activations could be clearly observed in the cytoplasm (Figure 6b and S12). This result is consistent with some previous studies in HeLa cells for the subcellular location of survivin mRNA.^46,47 As a control, negligible fluorescence signal was observed in cells expressing H1/H2 or H1/H2/C-SMB. To further confirm that the observed fluorescence signals indeed indicated the cellular expression of survivin mRNA, we treated the cells with YM155, a survivin inhibitor that is known to suppress the expression of survivin mRNA in HeLa cells.⁴⁸ As expected, after adding YM155, the cellular fluorescence signals disappeared even in cells expressing H1/H2/SMB (Figure 6b, 6c, and S12). Indeed, these results indicated that INSIGHT can be used for target mRNA imaging in mammalian cells.

DISCUSSION

DNA-based in situ amplification methods have been used to visualize and quantify the subcellular distributions of nucleic acids, proteins and small molecules. However, the live-cell applications of these methods are still limited, mainly due to the challenges in the probe delivery and degradation. Here, we report a novel genetically encoded RNA-based in situ amplification approach, termed INSIGHT. By conjugating fluorogenic RNA aptamers with hybridization chain reactions, INSIGHT can be used to image target RNAs in living bacterial and mammalian cells.

The INSIGHT system has several important features for cellular imaging. (1) The INSIGHT probes can be genetically encoded and directly transcribed inside cells, obviating the invasive delivery process. (2) Target-triggered hybridization chain reaction amplifies the signal and improves the sensitivity of the approach, which is especially important for low-abundance target imaging. (3) As an in situ amplification method, INSIGHT can be used for imaging the spatiotemporal distributions of target analytes. In addition, the fluorescence signals were concentrated in several localized spots, and as a result, it can be brighter and easier to observe than those signals dispersed in the homogeneous solution. (4) By simply changing the loop region of a molecular beacon, the INSIGHT system can be conveniently designed to detect different target RNAs. As a result, INSIGHT is a modular and universal platform for the cellular imaging of various target analytes.

The concatemeric nanostructure generated by INSIGHT activation may also potentially improve the intracellular folding and/or enzymatic resistance of fluorogenic RNAs. By replacing Broccoli with other fluorogenic RNA aptamers such as DNB,⁴⁹ Corn,^50,51 Mango,⁵² and Pepper,³⁴ it should be possible to extend the potential applications of INSIGHT for the multi-color imaging of cellular analytes.

In addition to their applications in the cellular imaging, these genetically encoded hybridization chain reactions may be potentially used to build up intracellular RNA scaffolds for the programmable regulation of cellular functions.⁵³ Indeed, synthetic RNA scaffolds have been applied to spatially organize multi-enzyme complexes.^54,55 By changing Broccoli into protein-binding aptamers, INSIGHT can be potentially used to build target-responsive artificial cellular network.

In sum, we envision this INSIGHT system can be broadly applied for sensing intracellular molecules of interest. INSIGHT can be used for real-time imaging of the subcellular locations and interactions of target analytes. The rational design of INSIGHT can also be potentially useful in regulating metabolic pathways and cellular functions.