Enzymatic Beacons for Specific Sensing of Dilute Nucleic Acid

Abstract

Enzymatic beacons, or E-beacons, are 1:1 bioconjugates of the nanoluciferase enzyme linked covalently at its C-terminus to hairpin forming ssDNA equipped with a dark quencher. We prepared E-beacons biocatalytically using HhC, the promiscuous Hedgehog C-terminal protein-cholesterol ligase. HhC attached nanoluciferase site-specifically to mono-sterylated hairpin oligonucleotides, called steramers. Three E-beacon dark quenchers were evaluated: Iowa Black, Onyx-A, and dabcyl. Each quencher enabled sensitive, sequence-specific nucleic acid detection through enhanced E-beacon bioluminescence upon target hybridization. We assembled prototype dabcyl-quenched E-beacons specific for SARS-CoV-2. Targeting the E484 codon of the virus Spike protein, E-beacons (80 × 10⁻¹² M) reported wild-type SARS-CoV-2 nucleic acid at ≥1 × 10⁻⁹ M by increased bioluminescence of 8-fold. E-beacon prepared for the SARS-CoV-2 E484K variant functioned with similar sensitivity. Both E-beacons could discriminate their target from the E484Q mutation of the SARS-CoV-2 Kappa variant. Along with mismatch specificity, E-beacons are two to three orders of magnitude more sensitive than synthetic molecular beacons.

Graphical Abstract

Insert text for Table of Contents here. Enzymatic beacons, or E-beacons, have been prepared by hedgehog catalysed bioconjugation and shown to function as turn-on biosensors with single-base pair mismatch specificity using SARS-CoV-2 as a sample target.

Introduction

Reagents for the specific detection of dilute nucleic acid are fundamental to molecular and cellular genetics and various clinical diagnostics, including tests for viral pathogens like SARS-CoV-2.^[1–6] Molecular beacons, the fluorogenic hairpin-forming oligonucleotide-based sensors, have provided a standard detection tool.^{[7, 8]} Modified with fluorophore and quencher at opposite ends, the oligonucleotide fluorescence is suppressed in the hairpin, or off state. Molecular beacons switch on by hybridizing to complementary nucleic acid, which separates the fluorophore and quencher, increasing radiative emission. Improvements to molecular beacon technology have been driven mainly through chemical synthesis with the introduction of more efficient quenchers and new fluorophore/quencher pairs.^[9]

Here we describe an alternative detection strategy where biocatalysis is employed in the preparation of the sensor and for the sensor output. In enzymatic beacons or E-beacons the fluorophore of a molecular beacon is replaced by the compact, ATP-independent bioluminescent enzyme, nanoluciferase (Nluc).^[10–13] This substitution provides an internal, amplifiable light source. Nluc with the engineered substrate furimazine produces light that is sufficiently bright for measurement using a portable luminometer.^[14] With detection signal enzymatically amplified, we also expected that nucleic acid detection assays would consume less reagent while improving sensitivity relative to synthetic molecular beacons.

To connect Nluc to a hairpin-forming oligonucleotide as a potential E-beacon, we applied the protein-nucleic acid bioconjugation activity found in hedgehog precursor proteins (Figure 1A).^[15] Hedgehog precursor proteins harbor a promiscuous C-terminal autoprocessing domain, HhC, that catalyzes two linked activities: self-cleavage from an adjacent N-terminal protein, and site-specific sterol ligation to that departing N-terminal protein. These reactions occur without cofactor or NTP requirements. We have found that the Drosophila melanogaster HhC has broad substrate tolerance, maintaining robust bioconjugation activity toward heterologous N-terminal substrate proteins and catalyzing ligation with sterols of varying structure and chemical appendages, including oligonucleotides. ^{[15, 16]} For E-beacons, we fused this promiscuous HhC to the C-terminus of the Nluc enzyme (Supporting Methods).

Figure 1. Concept and biocatalytic preparation of prototype E-beacon.

(A) Scheme showing HhC catalyzed bioconjugation of Nluc (grey) to a hairpin-forming mono-sterylated oligonucleotide (steramer) modified with a 3’ dark quencher (Q). Conjugation involves formation of an internal thioester (step 1), followed by steramer binding and acyl transfer to the sterol hydroxyl group (step 2). (B) General mechanism for turn-on nucleic acid detection by E-beacon, whereby hybridization of the Nluc conjugated hairpin oligonucleotide with complementary nucleic acid displaces the quencher (Q) from Nluc, enhancing bioluminescence signal. (C) E-beacon preparation monitored by SDS-PAGE. A fusion of Nluc with HhC is purified first by Ni-NTA chromatography, followed by size exclusion chromatography (SEC), then reacted with steramer (conjugation). E-beacon is isolated by agarose gel extraction. Nucleic acid was visualized by UV with GelRed stain.

To serve as an HhC substrate for ligation to Nluc, the hairpin oligonucleotide component required mono-sterylation.^[15] Our initial E-beacon preparations used a ssDNA oligo with the sequence: (sterol)-5’-CGCTCCCAAAAAAAAAAACCGAGCG-(3’-IBQ). The underlined regions self-anneal to form the hairpin stem; the italicized 15 nucleotides represent the probe or loop region for target nucleic acid hybridization. The same sequence was used by Kramer, Tyagi et al. for their early biophysical studies on molecular beacons.^[17] We started off using the Iowa Black (IBQ) dark quencher at the (3’) to suppress Nluc bioluminescence at the (5’) in the hairpin (off) state (Figure 1B). Mono-sterylation (5’) was achieved by coupling a (5’) alkyl amine modifier of the oligo to the carboxyl group of 23, 24-bisnor 5-cholenic acid-3β-ol (Figure 1A, right and Supporting Methods).^[18]

We prepared prototype E-beacon, Eb.1, by combining the Nluc-HhC precursor protein with mono-sterylated hairpin-forming oligodeoxynucleotide in vitro.^[15] His-tagged Nluc-HhC precursor was expressed in E. coli and purified in soluble form by Ni-NTA resin and size exclusion chromatography (Figure 1C, lanes 1–2). Site-specific bioconjugation of Nluc to the sterylated oligo was carried out as described by Zhang et al.^[15] In a typical reaction, we incubated Nluc-HhC precursor at 2 μM (final) with 25–50-fold excess of mono-sterylated oligo in HhC buffer, overnight on the benchtop, then assessed reaction progress by SDS-PAGE (Figure 1C, see conjugation). Agarose gel extraction provided a convenient means of purifying the conjugate, Eb.1 (Figure 1C, last two lanes).

We observed encouraging nucleic acid sensing with Eb.1. As an initial test, bioluminescence readings from samples of Eb.1 with added oligonucleotide complementary to the hairpin probe region (signal; GGTTTTTTTTTTTGG) were collected and compared with bioluminescence readings from Eb.1 mixed with non-complementary oligonucleotide (noise; CTGGTCTTCGGGCTA). In these experiments, Eb.1 was present at 2 × 10⁻⁹ M and oligonucleotide was added to 25 × 10⁻⁹ M. Samples were incubated in hybridization buffer (KCl 100 mM, MgCl₂ 1 mM, in 10 mM Tris buffer, pH 8) at 25 °C for 30 min followed by addition of the Nluc substrate, furimazine, according to the manufacturer’s instructions (Promega).^[10] Representative data from a 96-well experiment, with 48 signal and 48 noise samples, are summarized with the histogram in Figure 2A. The signal/noise was 2.5–3-fold, comparing the ratio of Eb.1 samples with complementary oligo to Eb.1 samples with random oligo. The increased Eb.1 bioluminescence accords with the general unquenching mechanism:^{[7, 8]} target/probe hybridization displaces the 3’-quencher from the light source, here Nluc, permitting brighter signal.^[19]

Figure 2. E-beacon proof of concept and sequence specific nucleic acid detection*.

(A) Bioluminescence from Eb.1 increases in the presence of complementary oligonucleotide (green, n=48) compared to E-beacon samples mixed with noncomplementary oligonucleotide (red, n=48). (B) Results of “blinded test” of Eb.1 specificity indicate that Eb.1 distinguishes complementary oligonucleotide (green) from oligonucleotides containing 1–3 base mismatches. See Table 1 for sequence information. (C) Comparison of E-beacons with different quenchers: IowaBlack (Eb.1), dabcyl (Eb.2), and ONYX-A (Eb.3). Bioluminescence was measured after 10-minute incubation with complementary oligonucleotide (green, n=48) or noncomplementary oligonucleotide (red, n=48). *In A, B, and C, the E. beacon was present at 2 × 10⁻⁹ M final; oligonucleotide at 25 × 10⁻⁹ M; temperature, 25 °C; substrate, furimazine.

Next, we more stringently evaluated the specificity of Eb.1 as a nucleic acid reporter by measuring bioluminescence output when combined with test oligos carrying only one to three mismatches with the hairpin probe. Oligo sequences are listed in Table 1. As above, each oligo was added to 25 × 10⁻⁹ M in hybridization buffer containing Eb.1 at 2 × 10⁻⁹ M. These experiments were carried out “blinded”: one lab member distributed complementary and mismatched oligonucleotides into sample wells of a 96-well plate in a semi-random pattern; another lab member without knowledge of the plate layout added Eb.1, recorded bioluminescence measurements and carried out the data analysis. The results of a representative E-beacon specificity test are summarized in Figure 2B. The largest enhancement in Eb.1 bioluminescence is apparent in samples mixed with perfectly complementary target (signal/noise, 3-fold). No increase in bioluminescence was apparent in wells containing oligos with central mismatches of two and three base pairs (G2 and G3). The oligo with one mismatch (G1) in the center produced 10% unquenching of the E-beacon. Unquenching of 15–20% of Eb.1 was observed when mixed with oligo G2F, carrying (T-to-G) mismatches with the probe region at the 3’ and the 5’ flanks. Nonetheless, samples of Eb.1 with G1 and G2F, compared with samples of Eb.1 plus the fully complementary oligo, were still easily distinguished. The fidelity of Eb.1 observed here suggested that this prototype’s specificity, like a synthetic molecular beacon,^{[7, 8]} was sufficiently strong for detecting subtle sequence variations.

Table 1.

Sequences of hairpin oligodeoxynucleotides for E-beacons and ssDNA oligonucleotides used for evaluating E-beacon function. Except where indicated otherwise, sequences are shows from 5’ to 3’. Red = mismatch; Bold = codon 484 of SARS-CoV-2 spike protein; [S] sterol; [Q] quencher

Name	Sequence
Eb.1–3	[S]-CGCTC CCAAAAAAAAAAACC GAGCG-[Q]
EB.1 Target:	GGTTTTTTTTTTTGG
Random:	CTGGTCTTCGGGCTA
G1:	GGTTTTGTTTTTTGG
G2:	GGTTTTGGTTTTTGG
G3:	GGTTTTGGGTTTTGG
G2F:	GGGTTTTTTTTTGGG
C3:	GGTTTTCCCTTTTGG
Eb.19(WT)	[S]-CGCTC TGGTGTTGAAGGTTT GAGCG-[Q]
E484 Target	(3’) ACCACAACTTCCAAA (5’)
Eb.19(E484K)	[S]-CGCTC TGGTGTTAAAGGTTT GAGCG-[Q]
E484K Target	(3’) ACCACAATTTCCAAA (5’)
E484Q Target	(3’) ACCACAAGTTCCAAA (5’)

While encouraged by these results, we viewed the Eb.1 signal/noise of ~3 and the assay Z’ factor ^[20] of 0.46 as narrow and in need of improvement. The Z’ factor indicates assay robustness by the degree of separation between the values of positive (signal) and negative (noise) samples; the ideal Z’ factor is 1. A Z’ factor of 0.5 suggests that the assay reliability is borderline. This analysis led us to consider other quencher groups for the E-beacon. The peak bioluminescence from the Nluc/furimazine reaction occurs at 460 nm. The maximum absorbance of the Iowa Black quencher is 531 nm. Better overlap was sought between the Nluc bioluminescence and the absorbance spectrum of the quencher. As alternative dark quenchers, we evaluated dabcyl and ONYX-A (Sigma). The dabcyl max absorbance is 453 nm and the ONYX-A absorbance peak is 515 nm. For meaningful comparison with Eb.1, we prepared new E-beacons from mono-sterylated hairpin oligos carrying the same sequence as Eb.1 and we applied the same biocatalytic approach for site-specific attachment to the C-terminus of Nluc. The dabcyl containing E-beacon, Eb.2, and the ONYX-A E-beacon, Eb.3, were likewise isolated by agarose gel extraction.

Changing the E-beacon quencher improved signal/noise and assay quality, in the order ONYX-A> dabcyl> IBQ. Figure 2C compares Eb.1 Eb.2 and Eb.3 readings with E-beacons at 2 × 10⁻⁹ M and complementary or random oligo at 25 × 10⁻⁹ M. Eb.3 with the ONYX-A quencher showed the highest single/noise of 8.7-fold and Z’ of 0.79. With the dabcyl quencher in Eb.2, the signal/noise was 6, better than IBQ, however the scatter in readings dropped the Z’ to 0.47. Because the hairpin sequences in EB.1-.3 were the same, the changes in signal/noise are difficult to explain in terms of intra-oligo interactions, solution properties of the E-beacons, or Nluc activity; they seem more likely the result of specific photochemical characteristics (e.g. quenching efficiency). Structural information on SIGMA’s ONYX quencher which might have helped interpret these results is not yet available.

Lastly and with a view toward application, we designed and tested prototype E-beacons for SARS-CoV-2 detection. Reliable assays for rapid identification of patients infected with SARS-CoV-2, particularly those shedding viable virus, remain crucial to pandemic management. In considering an E-beacon configuration to facilitate scale-up for high-volume testing, we settled on using the dabcyl rather than the ONYX-A or Iowa Black quenchers. Dabcyl modified oligos are widely available and, unlike ONYX-A, dabcyl phosphoramidites can be purchased for solid phase oligo synthesis. With dabcyl, we were able to improve our oligonucleotide sterylation protocol from microscale EDC-based solution coupling to a more robust solid phase coupling method with a DNA synthesizer. This approach routinely provided >90% yield for oligonucleotides carrying 5’ sterol and 3’ dabcyl modifications (Supporting Methods). Because of the high yield and on-column washing steps, sterylated oligos prepared this way did not require additional HPLC purification prior to bioconjugation of Nluc by HhC. As an additional benefit, we found that E-beacon incorporating solid-phase synthesized oligos proved superior over our initial dabcyl-quenched E-beacon, Eb.2, as measured by S/N and by Z’ factor. Compared to Eb.2, newly prepared dabcyl-quenched E-beacons showed an average S/N of 8–9, compared with 6 for Eb.2, and the assay results were more reliable, with a Z’ factor of 0.8 compared with 0.47 for Eb.2 (see Supporting Figure 1 and Supporting Methods).

We selected the Spike coding region of SARS-CoV-2 as target for this new set of E-beacons. We focused on the glutamate 484 codon of Spike as changes at this codon can help distinguish wild-type SARS-CoV-2 from emerging variants. The E-beacon intended for wild-type, Eb.19(WT), was prepared by HhC catalyzed ligation of Nluc to the sterylated dabcyl-modified hairpin oligo: (sterol)-5’-CGCTCTGGTGTTGAAGGTTTGAGCG-3’-(dabcyl). A second E-beacon was prepared for detecting the E484K mutation, an “escape mutation” which is present in certain α along with β, γ, ι, η and μ variants of SARS-CoV-2. ^[21] This Eb.19(E484K) incorporated the oligo: (sterol)-5’-CGCTCTGGTGTTAAAGGTTTGAGCG-3’-(dabcyl). The bold text represents the spike 484 codon.

We explored potential assay conditions for the SARS-CoV-2 E-beacons in two stages. First, we sought to determine the minimum working concentration of E-beacon. In these experiments, the concentration of target oligonucleotide, which is a DNA analog of the virus RNA, was held constant and in excess at 1 × 10⁻⁷ M, while the E-beacon was titrated from 1 × 10⁻⁸ M to 1 × 10⁻¹⁴ M. Bioluminescence readings following addition of substrate furimazine were recorded as before. A positive result was defined arbitrarily as having a signal/noise of >5 fold. With the S/N threshold at 5, the lowest possible operating concentration of Eb.19 (WT) was 3 × 10⁻¹³ M (Supporting Figure 2). Measurements at this low concentration of E-beacon required the luminometer gain setting at the maximum level.

In stage two, we held the E-beacon concentration constant and titrated the target oligos to find the detection threshold. We chose 80 × 10⁻¹² M for the E-beacon. Although higher than the minimum Eb.19 (WT) concentration defined above, we selected 80 pM to avoid pitfalls associated with assaying biomolecules in the extremely dilute regime, such as slow association kinetics and idiosyncratic adsorption effects, i.e., “death by dilution”. The assay parameter we sought here was the EC₅₀ or the concentration of target oligo that could produce 50% unquenching of Eb.19 (WT). We also explored the effect of E-beacon/target oligo incubation time on sensor output. In Figure 3A, we show bioluminescence readings of samples with Eb.19 (WT) plotted as a function of increasing target oligo concentration after 10 min, 1 hr and 3 hr incubations. Hyperbolic binding isotherms were fit to the experimental data to derive EC₅₀ values. For Eb.19 (WT) the EC₅₀ was 1.64 × 10⁻⁹ M for the 10 min incubation; with the extended 1 hr and 3 hr incubations, the Eb.19(WT) appeared ~10-fold more sensitive (Figure 3A, inset). Similar behavior was observed with Eb.19 (E484K) (Supporting Figure 3). The improved sensitivity (lower EC₅₀) with a 1–3 hr incubation of Eb19 (WT) and target oligo indicates that slow onset of E-beacon/target hybridization is a factor to consider with very dilute nucleic acid.

Figure 3.

E-beacons for SARS-CoV-2 (A) Bioluminescence readings of samples with Eb.19 (WT) after 10 min. 1 hr or 3 hr incubation with increasing concentration of complementary target oligonucleotide. The curves show hyperbolic binding isotherms with calculated half maximum emission (EC₅₀) at 2.54 × 10⁻⁹ M of oligonucleotide using the 10 min incubation, 0.2 × 10⁻⁹ M for the 1 hr incubation and 0.11 × 10⁻⁹ M for the 3 hr incubation. (B) Single base-pair mismatch discrimination by E-beacon, Eb.19 (WT). Bioluminescence from Eb.19(WT), at 80 × 10⁻¹² M, is plotted as a function of increasing oligonucleotide concentration. E484 is the complementary target, E484K is single base pair mismatch, “random” is non-complementary (see Table 1). Samples were incubated for 3 hrs followed by bioluminescence mesurent. (C) Bioluminescence spectrum measured by RLU/sec showing the selective detection of E484K oligonucleotide by E-beacon, Eb.19 (E484K). Oligonucleotides E484 and E484Q represent single base pair variants of wild type SARS-CoV-2 and Kappa SARS-CoV-2, respectively.

To assess specificity, bioluminescent signal from samples of Eb.19 (WT) mixed with target oligo was compared with signal from samples of Eb.19 (WT) mixed with the random oligo and the SARS-CoV-2 E484K variant oligo (single base-pair mismatch, see Table 1). As with Eb.1, the random oligo did not unquench Eb.19 (WT) over the concentration range tested (10⁻¹² to 10⁻⁷ M) (Figure 3B, lowest trace). Unquenching of Eb.19 (WT) did become apparent in samples containing the E484K oligo, although this “noise” required 1000-fold higher concentration of the E484K oligo over the target oligo (Figure 3B, compare squares to circles). In Figure 3C, we present selectivity data for Eb.19 (E484K) in the form of comparative bioluminescence spectra. Here Eb.19 (E484K) was mixed with either E484K target oligo (Figure 3C, top trace), with random oligonucleotide, E484 (WT) or an oligo corresponding to E484Q of the SARS-CoV-2 Kappa variant. Although some unquenching is apparent with the single base variants, signal of Eb.19 (E484K) was greatest with the fully complementary target, E484K (6–8 fold). Together, the results in Figure 3 A–C are promising and show that with 80 pM E-beacon, detection is sensitive to 1 nM target nucleic acid and base pair specific.

In summary, we report the preparation and in vitro characterization of E-beacons, enzymatic bioluminescent counterparts to synthetic fluorogenic molecular beacons. The general concept was introduced by Deo and her associates who assembled a (Renilla) luciferase-hairpin oligonucleotide-(3’-dabcyl) conjugate for microRNA detection,^[12] and subsequently created a (Gaussia) luciferase-hairpin oligonucleotide-(3’- dabcyl) conjugate for HIV detection.^[11] The Deo sensors were prepared by spontaneous coupling of hairpin-forming oligos onto reactive residues of those luciferase enzymes. Spontaneous coupling is relatively inexpensive, but the technique produces heterogeneity in enzyme: oligonucleotide stoichiometry (1:1, 1:2, 1:3 etc) and the site(s) of attachment can vary between batches. We prepare E-beacons by site-specific C-terminal coupling of nanoluciferase using HhC as the bioconjugation catalyst.^[15] This approach yields 1:1 enzyme: oligonucleotide stoichiometry. Protein-nucleic acid bioconjugation with HhC requires mono-sterylated oligonucleotides (steramers). Although not yet available commercially, we show here that steramers can be obtained in >90% yield from commercial reagents on a DNA synthesizer. E-beacons incorporating steramers prepared by solid phase for SARS-CoV-2 showed sensitive and sequence specific nucleic acid detection. For comparison, synthetic molecular beacons are used at ~ 1 × 10⁻⁷ M; the Deo sensors were reported at 2 × 10⁻⁹ M; E-beacon signaling with the ultrabright Nluc/furimazine reaction are effective in the mid to low pM range. With so little E-beacon needed for each assay, a microgram-scale E-beacon preparation yields ample reagent for thousands of tests. Sensitivity is on par with the turn-off, two-component BRET-beacons.^[22] In the BRET beacon, an Nluc-oligo conjugate, generated by spontaneous chemical coupling, is hybridized to a second target-binding detector oligo. In this non-covalent assembly, BRET beacons report fully complementary target DNA by loss-of-fluorescence, rather than the gain-of-luminescence in the E-beacon. Specificity, the influence of base pair mismatches on fluorescence output with the BRET beacon, was not reported.

It is worthwhile to consider the potential challenges to translating E-beacons into a viable clinical diagnostic for pathogens like SARS-CoV-2. Amplification of the patient specimen viral RNA, which can be in the attomolar range,^[23] would certainly be required to generate sufficient target nucleic acid to turn-on E-beacon signal. For point-of-care testing, we favor isothermal amplification methods which are completed in 30 to 60 minutes and carried out at ≤ 63 °C.^{[24, 25]} Preliminary experiments suggest that E-beacons are sufficiently thermostable to include directly in an RT-LAMP isothermal amplification reaction (Supporting Figure 5). After 30 min. of heating at 63 °C, template vs. non-template discrimination (S/N) was 10-fold, although the overall E-beacon bioluminescence was diminished. Alternative isothermal methods, such as NASBA and RPA, are carried out at lower temperatures, ~40 °C and may prove more suitable. A second obstacle to the practical application of E-beacons involves finding a means to access the specimen, post-amplification, to allow addition of Nluc substrate. Accessing the sample must be done in a manner that avoids well-to-well contamination. For multi-well E-beacon assays, an adhesive plate seal that is amenable to both sample luminescence measurement and mechanical perforation (for Nluc substrate addition) would be required.