Chemogenetic ON and OFF switches for RNA virus replication

The study addresses the challenge of externally controlling replication of therapeutic RNA viruses to improve safety and manage toxicity in applications such as oncolytic virotherapy, gene delivery, and vaccination. While transcriptional control systems for DNA viruses are established, effective, externally tunable control of RNA virus replication remains limited. Previous RNA-based controls (aptazymes, SMASh tags, and photo-responsive elements) offered partial or impractical control and did not completely block replication. Vesicular stomatitis virus (VSV), a negative-strand RNA virus with therapeutic promise but potential neurotoxicity, provides a relevant model. The authors propose a chemogenetic control mechanism by linking essential VSV replication proteins to the autocatalytic HIV protease dimer, whose activity can be modulated with clinically approved protease inhibitors (PIs), enabling ON or OFF control of viral replication.

Prior approaches to regulate RNA virus activity include: (1) RNA aptazymes fused to viral genes achieving <100-fold replication control and up to ~30-fold transgene control, but not complete shutdown; (2) SMASh-tags for an OFF switch in measles virus via C-terminal tagging of P protein; and (3) light-responsive elements enabling ON control but impractical clinically. VSV is widely studied as vaccine vector, oncolytic agent, and tracer, but translational use is limited by neurotoxicity and potential spread among livestock. HIV protease is a homodimeric aspartyl protease that autocatalytically processes HIV polyproteins and is targetable by clinical PIs (e.g., amprenavir, saquinavir, indinavir, lopinavir). The authors position their approach as an advance providing robust, tunable replication control using approved small molecules at clinically relevant doses, potentially enhancing safety compared to existing methods.

Design and construction: The HIV protease (PR) linked dimer, flanked by cognate cleavage sites and flexible GGSG linkers, was inserted into essential VSV replication proteins to create switches. ON-switch designs included intramolecular insertion of PR dimer into: (a) the P protein at amino acid 196 (Pprot), and (b) the L protein at amino acid 1620 (Lprot). An OFF-switch design replaced the intergenic region to create an N-terminal fusion of GFP–PR dimer–L (Prot-OFF), where active PR releases functional L in absence of PI; PI presence maintains a nonfunctional GFP–PR–L fusion.

Cloning and rescue: Constructs were assembled via Gibson assembly and restriction cloning into recombinant VSV backbones expressing GFP or luciferase reporters. Virus rescue used helper-virus-free calcium phosphate transfection in 293T cells expressing VSV replication proteins, followed by amplification on BHK-21 cells. Variants included VSV-Pprot-GFP, VSV-Lprot-GFP, VSV-Pprot-Luc, a dual P+L PR insertion (VSV-P-Lprot-GFP), and VSV-Prot-OFF-GFP.

In vitro assays: Mini-genome assay using a P-deficient VSV variant encoding DsRed evaluated Pprot function ± amprenavir (APV). Dose–response assays infected BHK cells at MOI 1 with increasing PI concentrations (APV, SQV, LPV, IDV). Single-step growth kinetics at MOI 3 (±10 µM APV for ON; no PI for OFF) were measured by TCID50. Western blotting against HIV PR validated proteolysis: PR dimer band (~22 kDa) in absence of PI versus intact fusion sizes (Pprot ~54 kDa; Lprot ~266 kDa; Prot-OFF fusion ~290 kDa) in presence of PI. Genetic stability was tested by 20 serial passages under optimal (10 µM) and suboptimal (1 µM) PI for ON variants and without PI for OFF, followed by functional PI-dependence assays, PCR across insert regions, and Sanger sequencing.

In vivo ON-switch studies: Athymic nude mice with subcutaneous U87 xenografts received intratumoral VSV-Pprot-Luc (2×10^6 TCID50) with or without intraperitoneal PI (APV + ritonavir). Bioluminescence imaging monitored replication. Separate cohorts received VSV-Lprot-GFP intratumorally ± PI to assess tumor growth and survival.

Neurotoxicity models: BALB/c mice received stereotactic intracranial injections (2 µL; 2×10^5 TCID50) of VSV-DsRed (wild-type backbone control), VSV-Pprot-GFP ± APV/RTV, or VSV-Lprot-GFP with PI regimens including APV/RTV or indinavir/RTV, and APV addition to injectate. Weight, clinical neurotoxicity scores, and histology of brain sections were assessed.

In vivo OFF-switch studies: NOD-SCID mice with G62 glioma xenografts received intratumoral VSV-GFP or VSV-Prot-OFF-GFP (2×10^7 TCID50) on days 0 and 7. Upon emergence of neurotoxicity signs in the Prot-OFF group (~day 15), mice were randomized to continued vehicle or initiation of PI cocktail (SQV + RTV, 3× daily) to activate OFF-switch. Tumor growth, survival, and neurotoxicity were monitored. A parallel histology cohort initiated PI 3 days after a single virus injection to assess intratumoral spread via anti-VSV-N immunofluorescence.

Statistics: TCID50, t tests, ANOVA with Tukey correction, and log-rank (Mantel–Cox) tests were used; p<0.05 deemed significant.

• ON-switch functionality: Insertion of HIV PR dimer into P (Pprot) or L (Lprot) produced VSV variants that replicated only in the presence of HIV protease inhibitors. APV enabled dose-dependent replication control approximately from 300 nM to 100 µM. Additional PIs (saquinavir 10 µM, lopinavir 10 µM, indinavir 10 µM) also regulated replication.
• Molecular validation: Western blots showed intact fusion proteins with PI (Pprot ~54 kDa PR-reactive band; Lprot ~266 kDa with PR-reactive band) and a ~22 kDa PR dimer band without PI, confirming autoproteolysis in the absence of inhibitor.
• Genetic stability: After 20 in vitro passages under optimal and suboptimal PI concentrations, ON-switch viruses remained PI-dependent with no escape variants detected. One mutation (PR2 G23A; R8K) in Pprot did not affect control; Lprot exhibited secondary L mutations during rescue (consistent with prior L insertion studies).
• Replication kinetics: In 10 µM APV, VSV-Lprot-GFP replicated comparably to parental VSV-GFP, whereas VSV-Pprot-GFP showed mild attenuation (~1–1.5 log lower titers). A dual P+L insertion ON-switch was functional but markedly attenuated (>2 log reduction), so not pursued further.
• In vivo ON control: Intratumoral VSV-Pprot-Luc in U87 xenografts showed initial BLI (likely from residual APV in stocks), which declined without continued PI but plateaued for 17 days with PI (APV + RTV), with associated tumor control. VSV-Lprot-GFP treated tumors had delayed growth and improved survival with PI versus without (log-rank p=0.0052). Viruses recovered from tumors retained PI dependence and insert integrity.
• Neurotoxicity mitigation: Intracranial wild-type-based VSV-DsRed was lethal within 4 days, but VSV-Pprot-GFP caused no neurotoxicity over 9 days with or without systemic PI; GFP expression remained confined to injection track regardless of PI, consistent with poor CNS penetration of PIs. VSV-Lprot-GFP with higher CNS-penetrant PI regimen (indinavir/RTV) and APV inclusion in injectate also showed no neurotoxicity and limited spread.
• OFF-switch functionality: The GFP–PR–L fusion (Prot-OFF) yielded replication that was inversely dependent on PI dose; SQV and APV blocked replication in a dose-dependent manner. In vitro, Prot-OFF displayed wild-type-like single-step kinetics without PI. Rescue selected secondary mutations at PR residues 85/86 (not associated with PI resistance), potentially enhancing PR activity in the fusion context; constructs without these changes could not be rescued. After 20 passages without PI, OFF-switch responsiveness persisted.
• In vivo OFF control: In NOD-SCID G62 xenografts, VSV-GFP induced early neurotoxicity. VSV-Prot-OFF maintained tumor control but some neurotoxicity. Initiation of PI (SQV+RTV) after signs appeared prevented further neurotoxicity but reduced tumor control leading to relapse. Starting PI 3 days after a single VSV-Prot-OFF injection curtailed intratumoral spread to isolated regions, whereas VSV-GFP spread widely.

The work demonstrates a chemogenetic proteolysis-based system to externally control RNA virus replication with approved small molecules. By embedding the HIV protease dimer within essential replication proteins (P or L), inhibitors convert a potentially active autoprotease into a controllable switch: PI presence maintains protein integrity (ON), while absence triggers cleavage and inactivation. Conversely, relocating the protease between proteins yields an OFF switch in which PI maintains a nonfunctional fusion. Targeting early replication/transcription steps improves control stringency compared with approaches acting at later stages (e.g., fusion). Compared to aptazyme systems, the presented switches achieve robust, tunable control in the low micromolar PI range using clinically established drugs and can completely block replication with sufficient PI. The approach enhances safety: ON-switch viruses require drug presence to replicate, lowering environmental risks; limited CNS PI penetration inherently restricts activation in the brain, adding a safety layer against neurotoxicity. The OFF switch offers a pharmacological safety valve for therapeutic vectors, enabling replication halt upon toxicity. The strategy may generalize to other mononegaviruses and RNA viruses if essential replication proteins tolerate intramolecular insertions (for ON) or large terminal fusions (for OFF). Combining RNA- and protein-based regulation could further diversify control (e.g., independent regulation of replication and transgene expression).

This study introduces two complementary, drug-controllable switches for RNA virus replication using a conditional HIV protease system in VSV: ON switches (protease inserted within P or L) that require PI for activity, and an OFF switch (N-terminal GFP–protease–L fusion) that is disabled by PI. The system enables dose-dependent, robust control from complete inhibition to near wild-type replication, is genetically stable over multiple passages, functions in vivo to sustain antitumor activity or to halt replication and prevent neurotoxicity, and uses clinically approved protease inhibitors. The approach provides a platform for safer oncolytic virotherapies and vaccine vectors. Future work should extend validation to additional RNA viruses, optimize insertion sites and linker designs to balance fitness and control, evaluate long-term in vivo safety and pharmacology (including CNS contexts), and explore combinations with orthogonal regulatory modalities for multilayered control.

• CNS activation constraint: Limited CNS penetration of many HIV protease inhibitors prevents effective ON-switch activation in the brain, restricting potential applications that require CNS replication but simultaneously enhancing safety against neurotoxicity.
• Attenuation: P-insert ON variant shows mild attenuation; dual P+L insertion is strongly attenuated (>2 logs), limiting multi-switch strategies. Prot-OFF appears attenuated in vivo relative to wild-type VSV in spread.
• PI dose considerations: Effective control required micromolar PI concentrations higher than reported HIV EC50s in vitro; translation to optimal human dosing for viral control requires further pharmacological study.
• Generalizability: Successful application to other RNA viruses depends on availability of permissive insertion sites in essential replication proteins (ON) and tolerance of large terminal tags (OFF); these conditions are virus-specific and not yet broadly validated.
• Genetic changes: Rescue of Prot-OFF selected mutations in PR (aa 85–86), suggesting functional constraints; implications for long-term stability and resistance warrant further study.
• Study scope: Environmental containment benefits of ON-switch were not directly tested; in vivo studies were limited in sample size and performed unblinded; whole-genome sequencing was not reported (insert regions sequenced). Potential immunogenicity of xenogenic proteins (HIV PR) in clinical settings was not assessed.