A genetically encoded BRET-based SARS-CoV-2 Mpro protease activity sensor

COVID-19, caused by SARS-CoV-2, remains a major global health threat. Viral replication depends on proteolytic processing of the pp1a/pp1ab polyproteins by the main protease (Mpro, 3CLpro), a cysteine protease that functions as a homodimer and cleaves at 11 sites with strict specificity for sequences containing a critical Gln. As Mpro cleavage sequences are not recognized by human proteases, Mpro is an attractive antiviral target. Sensitive live-cell reporters are needed to monitor Mpro activity for mechanistic studies and drug discovery. The study aims to develop and validate genetically encoded BRET-based sensors to monitor Mpro proteolytic activity in live cells and in vitro, evaluate their specificity and kinetics, compare performance to existing FlipGFP assays, and assess pharmacological inhibition including the impact of molecular crowding.

Multiple assay formats have been used to monitor coronavirus proteases, including FRET-based peptide assays and split-luciferase reporters for high-throughput screening. FRET-based in vitro assays with Mpro cleavage sequences have identified inhibitors such as Boceprevir, GC376, and calpain inhibitors. FlipGFP-based constructs using the Mpro N-terminal autocleavage site enable live-cell detection by converting non-fluorescent to fluorescent GFP upon cleavage; combinations of FlipGFP and luciferase assays have been used for inhibitor discovery. BRET offers advantages of high signal-to-noise and dynamic range, relying on spectral overlap and proximity/orientation of donor and acceptor. The mNeonGreen (mNG) and NanoLuc (NLuc) pair is widely used due to excellent spectral overlap and brightness, including in proteolytic sensors. Prior work also notes that peptide substrate flexibility and secondary structure formation can modulate Mpro cleavage efficiency, and peptide binding can allosterically affect Mpro dimerization/activation.

Sensor design: Two genetically encoded BRET sensors were engineered by inserting SARS-CoV-2 Mpro N-terminal autocleavage sequences between mNeonGreen (acceptor) and NanoLuc (donor): a short peptide (AVLQSGFR) and a long peptide (KTSAVLQSGFRKME), creating single-chain fusion constructs (mNG–peptide–NLuc). A His-tag enabled detection of fragments by Western blot. Sequences were cloned into mammalian expression vectors; the short sensor was also subcloned into pET-28b(+) for bacterial expression. Computational analysis: Conservation of the N-terminal autocleavage site was assessed across 1,984 SARS-CoV-2 pp1a sequences (NCBI Virus; MAFFT alignment), showing invariance. Structural models of short and long peptides (including linkers) were built using Modeller with SARS-CoV Mpro H41A–substrate complex (PDB 2Q6G) as template. All-atom explicit-solvent Gaussian accelerated MD (GaMD) simulations (NAMD 2.13, CHARMM36m, TIP3P water, 0.15 M NaCl) were performed for 1 μs per peptide to evaluate flexibility, radius of gyration, RMSD/RMSF, and secondary structure propensity. Cell culture and transfection: HEK 293T cells were cultured in DMEM + 10% FBS + 1% pen/strep at 37°C, 5% CO2. Transfections used PEI. Sensors were co-expressed with either WT Mpro (pLVX-EF1alpha-SARS-CoV-2-nsp5-2xStrep) or catalytically inactive C145A mutant (Addgene #141370 and #141371). Sensor:protease DNA ratio was typically 1:5. For dose-response, varying Mpro plasmid amounts (0 to 125 ng/well) were co-transfected keeping total DNA constant. Live-cell BRET assays: At indicated times (4–48 h, typically 16–48 h) after transfection in 96-well white plates, furimazine substrate was added (1:200) and emission spectra (380–664 nm, 400 ms/wavelength) or dual-wavelength readings (467 nm NLuc; 533 nm mNG) were acquired (Tecan Spark). BRET ratio was calculated as 533/467 nm. Total mNG fluorescence (Ex 480/Em 530 nm) was used to assess expression. Western blot: Lysates from co-transfected cells were prepared in 2× Laemmli buffer, separated by SDS-PAGE, transferred to PVDF, and probed with anti-His (sensor fragment) and anti-Strep (Mpro) antibodies to confirm cleavage and Mpro expression. Live-cell inhibitor assays: Cells co-expressing sensor and WT or C145A Mpro were treated concomitantly with GC376 (range 3.33 μM to 333 μM; 10 mM stock in 50% DMSO). After 16 h, BRET was measured; percentage activity was normalized to no-Mpro controls. IC50 values were derived from sigmoidal dose–response fits. FlipGFP comparison: HEK 293T cells were co-transfected with FlipGFP(Mpro) T2A mCherry and WT or C145A Mpro. mCherry marked transfected cells; GFP conversion indicated cleavage. Epifluorescence imaging at 24–48 h quantified %GFP+ among transfected cells. In vitro assays: Lysates from cells expressing sensors were incubated with recombinant SARS-CoV Mpro (NR-700) at 37°C at various concentrations (50 nM–5 μM; detailed kinetics at 200 or 500 nM). BRET was monitored over time. Enzyme kinetics with the short sensor at 200 nM Mpro provided apparent parameters (k′, Hill coefficient). For inhibition, 500 nM Mpro was preincubated with GC376 (10⁻⁴–10⁻⁹ M) for 30 min at 37°C, then mixed with sensor-containing lysate; BRET decrease rates yielded IC50. Molecular crowding was modeled by including 25% (w/v) PEG 8000; effects on cleavage kinetics and GC376 potency were evaluated by comparing IC50 with and without PEG. Data analysis used GraphPad Prism; experiments were performed in triplicates and repeated as indicated.

• Sensor performance and specificity in live cells: Both short and long BRET sensors showed two emission peaks (NLuc 467 nm; mNG 533 nm). Basal BRET ratios: short 2.17 ± 0.04; long 1.71 ± 0.20 (mean ± SD, N=3). Co-expression with WT Mpro caused marked BRET decrease for both sensors; no significant change with C145A mutant. Western blots showed loss of full-length sensor and appearance of ~30 kDa His-tagged mNG fragment only with WT Mpro, confirming proteolytic cleavage.
• Expression dependence: BRET decreases scaled with Mpro DNA dose (0–125 ng/well). Detectable decreases occurred at ≥1.25 ng/well. EC50 for Mpro plasmid DNA: 1.77 ± 1.07 ng/well (short) and 1.85 ± 1.01 ng/well (long).
• Time course: In no-Mpro controls, BRET increased from 4 h and plateaued by ~16 h, likely reflecting mNG maturation versus NLuc. With WT Mpro, significant BRET decrease was evident by 8 h and continued to 48 h; no decrease with C145A. Proteolysis half-lives: 9.76 ± 3.06 h (short) and 6.53 ± 2.88 h (long).
• Comparison to FlipGFP: FlipGFP responded later (GFP+ cells 67 ± 7% at 24 h; 84 ± 2% at 48 h with WT Mpro) and showed background activation with C145A (11 ± 6% GFP+ at 48 h), contrasting with high specificity of BRET sensors.
• Live-cell pharmacology: GC376 increased BRET ratio (inhibition of cleavage) dose-dependently for WT Mpro; no effect with C145A. IC50 values: 9.80 ± 3.84 μM (short) and 17.86 ± 2.14 μM (long). FlipGFP assay yielded IC50 5.453 ± 1.03 μM (for comparison).
• In vitro proteolysis: Recombinant Mpro cleaved sensors in a concentration-dependent manner; ≥500 nM required for discernible cleavage, with faster kinetics at 500 nM versus 50 nM. Kinetic analysis (short sensor, 200 nM Mpro): apparent k′ = 3.09 ± 0.02 μM and Hill coefficient 1.58, decreasing to 1.16 at higher Mpro concentration, consistent with dimerization/cooperativity.
• In vitro inhibition: GC376 inhibited cleavage with IC50 73.1 ± 7.4 nM (short) and 86.9 ± 11.0 nM (long) in the absence of crowding.
• Molecular crowding: 25% (w/v) PEG 8000 markedly increased proteolysis rate, consistent with enhanced effective Mpro dimerization via excluded volume effects. Under crowding, GC376 potency decreased substantially: IC50 shifted to 2623 ± 760 nM (short) and 10,260 ± 3280 nM (long) after 2 h, aligning with reduced inhibitor efficacy observed in live-cell context.
• Sequence conservation and peptide flexibility: The N-terminal autocleavage motif (AVLQSGFR) was invariant across 1,984 SARS-CoV-2 pp1a sequences. GaMD simulations showed both peptides are flexible with turns; the short peptide displayed occasional α-helical propensity in central residues, potentially influencing cleavage efficiency. DLS suggested a compact sensor architecture supporting strong basal BRET.

The study addresses the need for robust, specific, and sensitive live-cell reporters of SARS-CoV-2 Mpro activity. The mNeonGreen–NanoLuc BRET sensors directly report proteolytic cleavage with high dynamic range and minimal background, as evidenced by strong BRET decreases with WT Mpro and no cleavage with the C145A active-site mutant. Compared to FlipGFP, the BRET sensors detect activity earlier (as soon as 8 h post-transfection) and exhibit higher specificity with negligible activation by inactive protease. The dose- and time-dependent data quantify Mpro functional potency in cells and provide half-lives of cleavage for both sensor designs. Pharmacological experiments demonstrate that the sensors can quantify inhibitor efficacy in live cells and in vitro, capturing differences across conditions. The marked reduction in GC376 potency under molecular crowding indicates that intracellular-like environments can enhance Mpro activity, likely by promoting dimerization, and can diminish inhibitor effectiveness. This has implications for translating in vitro inhibitor potencies to cellular and physiological contexts. The kinetic signatures (apparent cooperativity decreasing at higher enzyme levels) further support a role for Mpro oligomeric state in catalytic function. The invariance of the autocleavage site among SARS-CoV-2 isolates supports the broad applicability of the sensors for current and emerging variants. Overall, the sensors are well-suited for live-cell antiviral screening, mechanistic enzymology of Mpro, and evaluating the impact of sequence variations.

Genetically encoded BRET-based sensors using mNeonGreen and NanoLuc flanking Mpro N-terminal autocleavage sequences enable sensitive, specific detection of SARS-CoV-2 Mpro proteolytic activity in live cells and in vitro. The sensors show rapid, robust response to WT Mpro with no cleavage by the C145A mutant, outperforming FlipGFP in response time and specificity. They effectively quantify pharmacological inhibition by GC376 and reveal that molecular crowding enhances Mpro activity while reducing inhibitor potency, underscoring the importance of cellular context. These sensors can be deployed for antiviral drug discovery targeting Mpro and for functional genomics to assess effects of Mpro sequence variation. Future work could apply the sensors to diverse cell types and infection models, expand inhibitor profiling, and refine sensor designs for multiplexed or subcellularly targeted readouts.

• Detection kinetics depend on intracellular expression and maturation of the reporter components; the authors note that timing may vary with actual protease expression in host cells. mNeonGreen maturation appears slower relative to NLuc, influencing early BRET dynamics.
• Donor–acceptor fusion architecture could, in principle, affect peptide accessibility and cleavage efficiency; a longer flanking sequence was included to mitigate this, but residual effects cannot be excluded.
• In vitro modeling of crowding used PEG 8000, which approximates but may not fully recapitulate cellular crowding complexity; effects on inhibitor potency in vivo could differ by cell type and conditions.
• Recombinant enzyme assays required relatively high Mpro concentrations (≥500 nM) for clear cleavage under the assay conditions, which may not reflect all physiological scenarios.
• Pharmacological validation focused primarily on GC376 as proof-of-principle; broader inhibitor panels would strengthen generalizability.