A pan-orthoebolavirus neutralizing antibody encoded by mRNA effectively prevents virus infection

ABSTRACT

Orthoebolavirus is a genus of hazardous pathogens that has caused over 30 outbreaks. However, currently approved therapies are limited in scope, as they are only effective against the Ebola virus and lack cross-protection against other orthoebolaviruses. Here, we demonstrate that a previously isolated human-derived antibody, 2G1, can recognize the glycoprotein (GP) of every orthoebolavirus species. The cryo-electron microscopy structure of 2G1 Fab in complex with the GPΔMucin trimer reveals that 2G1 binds a quaternary pocket formed by three subunits from two GP protomers. 2G1 recognizes highly conserved epitopes among filoviruses and achieves neutralization by blocking GP proteolysis. We designed an efficient mRNA module capable of producing test antibodies at expression levels exceeding 1500 ng/mL in vitro. The lipid nanoparticle (LNP)-encapsulated mRNA-2G1 exhibited potent neutralizing activities against the HIV-pseudotyped Ebola and Sudan viruses that were 19.8 and 12.5 times that of IgG format, respectively. In mice, the antibodies encoded by the mRNA-2G1-LNP peaked within 24 h, effectively blocking the invasion of pseudoviruses with no apparent liver toxicity. This study suggests that the 2G1 antibody and its mRNA formulation represent promising candidate interventions for orthoebolavirus disease, and it provides an efficient mRNA framework applicable to antibody-based therapies.

Introduction

Orthoebolavirus is a highly dangerous genus of the family Filoviridae that can cause severe hemorrhagic fever in humans and non-human primates. To date, six orthoebolaviruses have been identified: Ebola virus (EBOV), Sudan virus (SUDV), Reston virus (RESTV), Bundibugyo virus (BDBV), Taï Forest virus (TAFV), and Bombali virus (BOMV). Four of these viruses (EBOV, SUDV, BDBV, and TAFV) are pathogenic to humans and have triggered more than 40 outbreaks over the past four decades with a high case fatality rate of 25%–90% [1]. EBOV is responsible for the two largest orthoebolavirus disease outbreaks in history, causing 11,325 deaths in West Africa (2014–2016) and 2,287 deaths in the Democratic Republic of the Congo (2018–2020) [2]. Notably, the most recent orthoebolavirus outbreak (ended on January 11, 2023) was caused by SUDV, which had not appeared for a decade and resulted in 164 cases and 77 deaths in Uganda [3].

The glycoprotein (GP) on the surface of the viral envelope is the primary target for the development of prophylactic and therapeutic drugs against orthoebolaviruses. Following the West African outbreak, four vaccines, namely, Ervebo, Zabdeno/Mvabea, Ad5-EBOV, and GamEvac-Combi, were approved by different regulatory agencies [4]. In terms of therapeutic drugs, in vivo data indicate that monoclonal antibodies (mAbs) significantly reduce the mortality rate of patients [5]. Therefore, two antibody-based therapies, Inmazeb [6] and Ebanga [7], have therefore been approved. These intervention measures are essential in addressing such serious public health threats. However, the differences in GP amino acid sequences across the orthoebolaviruses is up to 59% [8]; this poses a challenge to the development of broad-spectrum drugs. Furthermore, a recent study has revealed that pre-existing immunity to EBOV has limited protective effects against SUDV [9]. The two approved antibody drugs only target EBOV and cannot cross-neutralize other species. The multiple species and escape variants of orthoebolaviruses [10], the uncertainty of the pathogen in future outbreaks, the inapplicability of vaccines to specific populations, and the long development cycles of antibody drugs underscore the importance of exploring broad- or rapid-acting countermeasures.

Broad-spectrum antibodies are an efficient means of combating the orthoebolavirus. The advantage of broadly neutralizing antibodies (bnAbs) lies in the reduced economic and time costs of developing antibodies specific to different species that are suitable for emergency prevention and treatment when the pathogen has not yet been identified. Several mAbs, with cross- or pan-neutralizing capabilities against pathogenic species, such as 6D6 [11], CA45 [12], ADI-15742 [13], ADI-15878 [13], rEBOV-515 [14], rEBOV-442 [14], and FVM04 [15], have been reported. Pan-neutralizing antibodies typically target the unobscured, highly conserved GP2 subunit and the Base region of the GP1 subunit, achieving neutralization by blocking GP cleavage and conformational rearrangement that occur post-receptor binding [12,13]. Based on the candidates above, there are three antibody cocktails, rEBOV-515/442 [16], CA45/FVM04 [17], and MBP134 [18,19], which have exhibited ideal protective efficacy in larger animal models (ferrets or non-human primates) against EBOV, SUDV, and BDBV. However, the lack of direct comparison among these bnAbs and the possibility of viral escape mutations following antibody treatments [10] highlight the need to expand the limited library of candidate antibodies to address the threat posed by these dangerous viruses.

Another promising strategy is the in vivo expression of antibodies using messenger ribonucleic acid (mRNA) technology. The advantages of mRNA antibodies include cell-free production, consistent manufacturing processes, rapid response, and the ability to target previously undruggable intracellular targets. In recent years, technological advancements in chemical modification and delivery strategies have substantially addressed the challenges of mRNA stability and translation efficiency, demonstrating the immense potential of mRNA-based therapies. Particularly, during the COVID-19 pandemic, several mRNA vaccines have achieved great success, confirming the safety, efficacy, and practical value of these drugs. Simultaneously, mRNA-based passive therapies have made significant progress in preclinical studies on several diseases [20–25]. Among these, mRNA-1944, used against the chikungunya virus, has completed Phase I clinical trials (NCT03829384) and exhibited good safety and efficacy [25]. However, no study has yet evaluated the potential application of mRNA antibodies in prevention and treatment of the orthoebolavirus disease, which has a short latent period.

In this study, we characterize a previously identified human protective mAb, 2G1 [26]. 2G1 recognizes all orthoebolavirus species and strongly neutralizes pseudotyped pathogenic species. 2G1 binds to the quaternary epitope of the GP trimer, with critical residues that are highly conserved among filoviruses. In this study, we designed an efficient universal mRNA antibody expression framework. The mRNA-2G1-LNP formulation demonstrated cross-neutralizing capabilities at the cellular level that were more than ten times that of the IgG form, and it blocked the invasion of EBOV and SUDV pseudoviruses in mice at a low dose. This study highlights the potential of 2G1 and its mRNA formulation for the prevention and treatment of orthoebolavirus disease. The mRNA module can also serve as a reference for other antibody therapies.

Materials and methods

Ethics statement

Female BALB/c mice (female, 20 g, 5–6 weeks old) were purchased from Sino Animal Science and Technology Development Co., Ltd (Beijing, China) and acclimatized for at least one week. The animal experiments strictly adhered to the ethical guidelines of the National Regulations for the Administration of Laboratory Animals and were approved by the Institutional Animal Care and Use Committee of the Laboratory Animal Centre of the Beijing Institute of Biotechnology (approval no. IACUC-SWGCYJS-2022-001). Throughout the study, the animals had access to an appropriate living environment comprising clean cages, optimal temperature/humidity levels, and sufficient food and water. All the mice were euthanized at the end of the experiment.

Cells and viruses

293 T cells (ATCC, CRL-3216) were maintained in Dulbecco’s modified Eagle’s medium (DMEM) (Gibco, Billings, MT) supplemented with 100 μg/mL streptomycin, 100 U/mL penicillin, and 10% fetal bovine serum (FBS) (Gibco) at 37°C and 5% CO₂. Expi293F^TM cells (Thermo Fisher Scientific, Waltham, MA) were cultivated in Expi293^TM Expression Medium (Thermo Fisher Scientific) at 37°C, 85% relative humidity, and 8% CO₂ in a 125-rpm incubator. The pseudotyped rHIV-orthoebolaviruses were packaged, stored, and used in a Biosafety Level-2 (BSL-2) facility at the Beijing Institute of Biotechnology.

Plasmid construction

The full-length sequences of EBOV Makona strain (GenBank: KJ660346.2), EBOV Mayinga strain (GenBank: NC_002549.1), BDBV (GenBank: NC_014373.1), SUDV (GenBank: NC_006432.1), TAFV (GenBank: NC_014372.1), RESTV (GenBank: NC_004161.1), and BOMV (GenBank: NC_039345.1) GP were codon optimized for Homo sapiens, synthesized (Sangon Biotech, Shanghai, China), and cloned into the pcDNA3.1 vector. The coding sequence of GPΔMuc-GCN4-His-SA (containing 1–308; 490–656 residues of the EBOV strain Makona GP) was constructed as previously described [27] and cloned into the pcDNA3.1 vector for expression. The cDNA sequence containing the variable regions of FVM04 (patent no. PCT/US2016/022141), mAb114 (patent no. US16190676A1), and MR72 (patent no. PCT/US2016/019644) were codons optimized for Homo sapiens and added to the human IgG1 constant region before cloning into the pcDNA3.1 vector. The in vitro transcription (IVT) template sequence containing the T7 promoter, 5’ UTR, antibody H/L coding sequence, 3’ UTR, and poly (A) tail were synthesized and cloned into the pUC57 vector. Alanine-scanning mutants of EBOV GP or mAb 2G1 were constructed using a Q5 Site-Directed Mutagenesis Kit (New England Biolabs, Ipswich, MA). The cDNA sequence of domain C (374–620 aa) of human Niemann-Pick C1 (NPC1-C; GenBank: NM_000271.4) was fused with an N-terminal tPA signal peptide and a C-terminal His₆ tag and cloned into pcDNA3.1 vector. Fab and IgG expression plasmids of mAbs 2G1, 4F1, and 5E9 had been constructed previously [26].

Protein generation

Antibodies, Fabs, NPC1-C, and GPΔMuc were expressed in Expi293F cells using the ExpiFectamine 293 Transfection Kit (Thermo Fisher Scientific) following the manufacturer's instructions. Antibodies were purified using HiTrap rProteinA columns (Cytiva, Marlborough, MA), and Fabs, NPC1-C, and GPΔMuc were purified using HisTrap Excel columns (Cytiva).

Recombinant EBOV GPΔTM was purchased from Sino Biological Inc. (Beijing, China; Cat No. 40442-V08H4). The EBOV GPcl was prepared based on published methods with minor modifications [28]. Briefly, 2 mg/mL of GPΔMuc in PBS was incubated with 0.5 mg/mL of thermolysin (Sigma Aldrich, St. Louis, MO) for 1 h at 37°C. The reaction was stopped by 0.5 mM phosphoramidon (Sigma Aldrich), followed by buffer exchange and concentration using a 30 kDa Centrifugal Filter Unit (Milliporee, Burlington, MA). After passing through a 0.22 µm syringe filter (Millipore), GPcl was purified using a Superdex 200 Increase 10/300 column (GE Healthcare, Chicago, IL).

Flow cytometric analysis

To determine the capacity of 2G1 to bind to surface-anchored orthoebolavirus GPs, 3 × 10⁶ 293 T cells were seeded in T25 flasks 24 h before transfection. When the confluency reached 80%, the cells were transfected with 6.5 μg of GP expression plasmids using the Lipofectamine3000 transfection reagent (Invitrogen, Waltham, MA). After cultivation for 36 h, the cells were harvested by digestion with 2 mL of PBS containing 0.02% (w/v) ethylene diamine tetraacetic acid. The cells were washed with PBS and resuspended in 1 mL PBS containing 2% (v/v) FBS (Solarbio, Beijing, China). The cells were filtered through a 40 μm strainer, counted, and diluted to 5 × 10⁶ cells/mL. A cell volume of 100 μL was assigned to each tube and incubated with 2 μg/mL of fluorescein isothiocyanate (FITC, Sigma Aldrich)-labeled 2G1 or an irrelevant control for 1 h at 25°C. The cells were washed using 3 mL PBS and resuspended in 100 μL PBS containing 2% (v/v) FBS. A total of 20,000 events were recorded for each tube on a FACSCanto II flow cytometer (BD Biosciences, Franklin Lakes, NJ) and analyzed using FlowJo software (v10.6.2).

To identify the critical residues of GP for 2G1 binding, 1 μg of plasmid bearing the full-length of EBOV GP wild-type or mutants was transfected into 293 T cells in 24-well plates. The cells were harvested and washed as described above and then stained with 2 μg/mL of FITC-labeled 2G1 and Alexa Fluor 647 (Thermo Fisher Scientific)-labeled 4F1.

Surface plasmon resonance

The binding kinetics of Ag-Ab were determined on a Biacore T200 instrument (GE Healthcare), and all procedures were performed at 25°C. The antibody was diluted to 1 μg/mL using HBS-EP buffer (Cytiva) and captured to a Sensor Chip Protein A (Cytiva) for 120 s at a 10 μL/min flow rate. EBOV GPΔTM or GPΔMuc in serial dilutions (starting at 200 nM in a 2-fold gradient) were loaded onto the chip to associate for 120 s at 30 μL/min, followed by dissociation for 900 s in the HBS-EP buffer. Five representative curves were subtracted from the baseline and fitted to a 1:1 binding model using the Biacore T200 Evaluation software (v3.2) to calculate the kinetic constants.

ELISA assay

To assess the binding abilities of antibodies under various pH conditions, EBOV GPΔTM, GPΔMuc, or GPcl was coated onto plates at 100 ng/well overnight at 4°C. The plates were washed using a 405 LS microplate washer (BioTek Instruments, Winooski, VT) and blocked with PBS containing 2% (w/v) bovine serum albumin (Sigma Aldrich) for 1 h at 37°C. Antibodies diluted in PBST at pH 7.5, 6.5, or 5.5 were added and incubated for 1 h at 37°C. The plates were rewashed, and 100 µL of the horseradish peroxidase (HRP)-conjugated anti-human IgG antibody (Abcam, Cambridge, UK) at a 1:10,000 dilution was added to each well, followed by incubation for 1 h at 37°C. After a final wash, 100 µL of 3,3′,5,5′-tetramethylbenzidine solution (Solarbio) was applied to each well for 6 min at 25°C, followed by the addition of 50 µL of termination buffer (Solarbio). The absorbance was measured at 450/630 nm using a SpectraMax ABS microplate reader (Molecular Devices, San Jose, CA).

EBOV GPΔMuc was used to identify the critical paratopes of 2G1 and to evaluate the expression level and dynamics, in vitro and in vivo, of mRNA-2G1. For each quantification test, a standard curve was generated using purified 2G1, and the cell supernatants or mouse sera were serially diluted to obtain an appropriate value in the linear range of the standard curve, in order to accurately compute the concentration of 2G1.

Pseudovirus packaging and neutralization test

For rHIV-orthoebolavirus packaging, the pcDNA3.1-orthoebolaviruses GP (2 µg) and pNL4-3.Luc. R^-E^- (16 µg) plasmids were co-transfected into cells cultured in T75 flasks using Lipofectamine3000 transfection reagent (Invitrogen). The medium was replaced with 18 mL fresh medium 6 h post-transfection, and the cells were maintained at 37°C for an additional 42 h. The supernatants were harvested by centrifugation at 4°C, 800 × g for 15 min and filtered through 0.45 μm syringe filters. The rHIV-orthoebolaviruses were stored in aliquots at −80°C, and the 50% tissue culture infectious dose (TCID₅₀) was determined using the Reed-Muench method after the solution was frozen. Each tube containing pseudovirus was thawed only once.

For antibody neutralization assays, 50 µL antibodies in three-fold serial dilutions starting at 50 µg/mL were added to 96-well cell culture plates andincubated with 50 µL rHIV-orthoebolavirus (∼2×10⁴ TCID₅₀/mL) at 37°C for 1 h. Antibodies and viruses were all diluted with DMEM. In each plate, six wells were set up in which the antibody solution was replaced by an equal volume of medium; this was to serve as a reference for calculating the neutralization rate. Approximately 3 × 10⁴ cells in 100 µL DMEM were added to each well and cultured at 37°C for 48 h. A volume of 100 µL medium was removed, after which 100 µL Bright-Lite™ luciferase assay reagent (Vazyme, Nanjing, China) was added to each well. After incubation at 25°C for 2 min, 150 µL solution was pipetted on to 96-well white flat bottom plates (Corning Inc., Corning, NY), and the luminescence was detected using a GloMax Navigator microplate luminometer (Promega, Madison, WI).

For mRNA-2G1-LNP neutralization assays, 3 × 10⁴ cells in 100 µL DMEM were seeded into 96-well plates. After cultivation for 18 h, 50 µL mRNA-2G1-LNP in 3-fold serial dilutions starting at 3 µg/mL were added, followed by the addition of 50 µL rHIV-EBOV or -SUDV. Luminescence was detected as previously described.

Cryo-electron microscopy

The GPΔMuc was incubated with the 2G1 Fab at a molar ratio of 1:1.5 for 1 h. Subsequently, the GPΔMuc/Fab complexes underwent purification through size exclusion chromatography employing a Superdex Increase 200 10/300 column (GE Healthcare) with 1× PBS as the elution buffer. Following purification, the complexes were concentrated to 3 mg/mL using a 30 kDa MWCO spin column (Millipore).

The protein complex (1.0 mg/mL) was dispensed in a volume of 4 μL onto a glow-discharged holey carbon grid (R 1.2/1.3, 300 mesh; Quantifoil Micro Tools GmbH, Jena, Germany). The grid was pre-treated by glow-discharging for 30 s at a medium level using a Harrick Plasma apparatus (PDC-32G-2) following a 3-min evacuation. Vitrification was performed using a Vitrobot Mark IV (Thermo Fisher Scientific) set to 8°C in an atmosphere of 100% humidity. After allowing the grids with the applied samples to equilibrate for 30 s, both sides of the grid were subjected to -1 force blotting for 3.5 s using filter paper. The sample quality was assessed using a Talos Arctica 200 kV electron microscope fitted with a K2 camera (Gatan, Pleasanton, CA). The grids exhibiting optimal ice thickness and particle density were moved to a Titan Krios electron microscope (Thermo Fisher Scientific) operating at 300 kV. Data collection was performed using a Thermo Fisher Krios G4 instrument operated at 300 kV, equipped with a Falcon4 detector. Micrographs were collected at a calibrated magnification of 75,000×.

Data collection comprised 4463 images of the Ag-Ab complex, with a pixel size of 1.1. Each image stack received an approximate total dose of 50 e-/Ų. The data collection process was fully automated and facilitated using AutoEMation and EPU (Thermo Fisher Scientific) software. For EM image processing, motion correction was performed using MotionCor2 (v1.4.7), and Ctffind (v4.1.14) was employed for CTF estimation. Further, CTF estimation was performed using CryoSPARC41. Micrographs with a CTF fitting resolution < 6 Å were excluded. All the subsequent processing steps were performed using CryoSPARC. After blob selection and inspection, 1,061,609 particles were selected. After undergoing bin4 extraction and 2D classification, 326,558 particles were selected to generate the ab initio models. These particles were used to create two ab initio models with C3 symmetry. After homo-refinement and NU refinement, the best map served as the reference for template selection across the entire dataset. Subsequently, NU refinement in C3 symmetry and center shifts were applied, particles were bin1 extracted with a box size of 384, and NU refinement was performed, resulting in a map with a resolution of 2.98 Å. Chimera [29] was used to dock G protein (PDB ID: 7KEJ) to the cryo-EM map. Structural domains were constructed de novo using EMBuilder [30]. Manual adjustments of the models were performed using the COOT [31]. Subsequent structural refinements were performed in real space using PHENIX [32].

Receptor binding inhibition assay

Microplates were coated with 100 µL of GPcl at 1 μg/mL overnight at 4°C. After blocking, plates were incubated with antibodies in serial dilutions starting at 50 μg/mL for 30 min at 37°C. Plates were washed and incubated with 100 µL of biotinylated NPC1-C at 5 μg/mL for 30 min at 37°C. Bound NPC1-C was detected using HRP-conjugated streptavidin (Thermo Fisher Scientific). Inhibition of NPC1-C binding was calculated by comparing the optical density value in the presence of antibodies to that of an irrelevant control. MR72, a previously reported mAb that competes with NPC1-C [13,14], was used as the positive control.

Enzyme cleavage block experiments

For GP cleavage inhibition by IgGs, 2 µg of EBOV GPΔMuc was incubated with 0.5, 5, or 50 µg of antibody in 20 µL of PBS for 30 min at 37°C, followed by another 30-minute incubation with the addition of 1 µL of 1 mg/mL thermolysin to a final concentration of 50 µg/mL. The reaction system was mixed with 5.25 µL of non-reducing loading buffer (5×) and boiled for 10 min at 99°C. One microliter of the reaction containing approximately 80 ng of GPΔMuc substrate was diluted with 4 µL of loading buffer (1×) and loaded onto an 8%–16% SurePAGE gel (GenScript, Nanjing, China) for migration at 150 V for 50 min. The proteins were transferred onto a nitrocellulose membrane using an eBlot™ L1 instrument (GenScript) and then blocked with PBS containing 5% (w/v) skim milk (Solarbio) for 1 h at 25°C. The membrane was incubated five times using PBS containing 0.1% (v/v) Tween-20 and incubated with a 1:5,000 dilution of HRP conjugated Anti-His tag antibody (Abcam) for 1 h at 25°C. After a final wash, Immobilon Western chemiluminescent HRP substrate (Millipore) was added to the membrane, which was then imaged using an iBright FL1500 imaging system (Invitrogen).

Similarly, cleavage was inhibited by Fabs. Essentially, 2 µg of EBOV GPΔMuc was incubated with 0.2, 2, or 20 µg of Fabs and digested by thermolysin under the same conditions. For Western blot analysis, vaccinee sera were used as primary antibodies and detected using a 1:10,000 dilution of goat anti-human IgG Fc-HRP (Abcam).

mRNA preparation and LNP formulation

pUC57 plasmids bearing mRNA modules of antibody heavy or light chains were linearized using Hind III-HF endonuclease (New England Biolabs) and extracted using a gel extraction kit (Omega Biotek, Norcross, GA). IVT reactions were conducted using a T7 High Yield RNA Transcription Kit (Vazyme) following the manufacturer's instructions. The IVT products were purified using VAHTS RNA clean beads (Vazyme), and the cap structure was added using vaccinia capping enzyme (Vazyme) and mRNA Cap 2'-O-methyltransferase (Vazyme). The capped mRNA products were purified, quantified, and analyzed using the Qsep100 capillary electrophoresis system (Bioptic).

To encapsulate mRNA-2G1-LNP, mRNA-2G1H and -2G1L were mixed at a molar ratio of 1: 1.1 in sodium acetate buffer, and SM-102, DSPC, cholesterol, and PEG-lipid were mixed at a previously reported molar ratio of 50: 10: 38.5: 1.5 in ethanol [33]. The two mixtures were mixed in a ratio of 1:3 and passed through an INano L nanomedicine preparation system (Micro&Nano Biologics, Shanghai, China) at a flow rate of 12 mL/min to form mRNA-2G1-LNP particles. The size distribution was determined using a Zetasizer Ultra instrument (Malvern Panalytical, Malvern, UK) and imaged using a transmission electron microscope (Hitachi, Tokyo, Japan). After the determination of the encapsulation rate and effective concentration using the Quant-iT RiboGreen RNA reagent and kit (Invitrogen), the mRNA-2G1-LNP was stored at 4°C.

In vitro expression of mRNA-2G1

Approximately 3 × 10⁵ 293 T cells in 1 mL of DMEM were seeded into each well of 24-well plates and cultured to ∼80% confluence. Cells were transfected with a total amount of 1 µg naked mRNA-2G1H/L mixtures using the Lipofectamine3000 transfection reagent (Invitrogen) or directly added with 1 µg mRNA-2G1-LNP. After incubation for 6 h at 37°C, the medium was replaced with 1 mL of fresh Opti-MEM I reduced serum medium (Gibco). The supernatant was collected at various time points according to the experimental design, and the antibody concentration was quantified by ELISA, as described above.

Detection of intact IgG and free light or heavy chain in mRNA-2G1-LNP expression supernatant

The supernatant was mixed with reducing or non-reducing buffer and then loaded onto a 4%–12% SurePAGE gel (GenScript) for migration at 150 V for 50 min. A 1:10,000 dilution of goat anti-human IgG Fc-HRP (Abcam) and a 1:1,000 dilution of rabbit monoclonal [EPR5367-8] to kappa light chain-HRP (Abcam) were used for heavy and light chain detection, respectively. The remaining steps were performed as described previously.

Prophylactic efficacy of mRNA-2G1-LNP in mice

The BALB/c mice were randomly divided into five groups (n = 8/group for rHIV-EBOV and n = 5/group for rHIV-SUDV). Animals were administered 1, 0.5, 0.25, or 0.125 mg/kg mRNA-2G1-LNP intravenously (i.v.). The control group was given an equal volume (100 μL) of PBS. Twelve hours later, each mouse was intraperitoneally injected (i.p.) with 2 × 10⁵ TCID₅₀ of the rHIV-EBOV or rHIV-SUDV pseudovirus. Four days post-infection, the mice were anesthetized for 10 min in a 2% isoflurane chamber and intraperitoneally injected with 150 mg/kg of XenoLight D-luciferin (PerkinElmer, Waltham, MA). The bioluminescence signal was detected for 5 min using an IVIS Spectrum system (PerkinElmer) and the normalized bioluminescence intensity (p/s/cm²/sr) of the imaging area was calculated.

Expression dynamics of mRNA-2G1-LNP in mice

The BALB/c mice were randomly divided into five groups (n = 5/group). Animals were intravenously administered with either 1 or 0.5 mg/kg of mRNA-2G1-LNP, 5 or 2.5 mg/kg of 2G1 antibody, or an equal volume (100 μL) of empty LNP particles. Body weight was monitored for 14 days. Blood samples (30–50 μL) were collected at 0.25, 1, 3, 7, and 14 days after administration and centrifuged at 5000 × g, 4°C for 10 min to isolate serum. The sera were stored at −80°C. After the last collection, an equal volume of serum was collected and serially diluted for antibody concentration quantification using ELISA as described above.

Biochemical criteria detection

The BALB/c mice were randomly divided into two groups (n = 5/group). Animals were administered either 1 mg/kg of mRNA-2G1-LNP or an equal volume (100 μL) of PBS via the i.v. route. Blood was collected three days after administration by removing the eyeballs. Four biochemical criteria, ALT, AST, LDH, and CR, were detected using the corresponding assay kits (BeijingBJ•XinChuangYuan BIOTECH) on a TBA-40FR automatic biochemical analyzer (Toshiba, Tokyo, Japan).

Sequence and structure analysis

The germline genes of 2G1 were obtained using IMGT/V-QUEST [34]. Kabat numbering of the VH or VL amino acids was annotated using abYsis [35]. SnapGene (DotMatics, Boston, MA) was used for general sequence analysis, while multiple sequence alignment was performed using Clustal W [36]. Aligned sequences were displayed using ESPript (v3.0) [37]. Structural analysis and superimposition were performed using ChimeraX (v1.6.1) [38].

Statistical analysis

In the ELISA and neutralization assays, each data point represents the mean ± SD of three replicates from a representative experiment, and the dose–response curves were fitted with a four-parameter logistic model to compute the EC₅₀ or IC₅₀ values. Dunnett's multiple comparisons test was used for comparisons involving more than two groups, and the unpaired t-test was used for comparisons between two groups. All statistical analyses were performed using GraphPad Prism 9.5.0, and the significance was defined as ns, p ≥0.05; *, p < 0.05; **, p < 0.01; ***, p < 0.001; *****, p < 0.0001.

Results

The mAb 2G1 exhibited a broad-spectrum binding and neutralizing capability against the orthoebolaviruses

A previous study has shown that 2G1 can cross-react with the recombinant GP ectodomains of EBOV, SUDV, BDBV, and RESTV [26]. To further determine the binding width, we conducted an analysis using flow cytometry to assess the ability of 2G1 to recognize the full-length GPs of the EBOV Makona and Mayinga strains, and that of the remaining five orthoebolaviruses, displayed on the surface of 293 T cells. 2G1 bound to the natural GP structures of all six orthoebolaviruses (Figure 1A). We characterized the avidity of 2G1 to the recombinant extracellular forms of EBOV GP in the presence (GPΔTM) or absence (GPΔMuc) of the mucin domain, revealing comparable kinetic constants of 5.25 nM for GPΔTM and 5.44 nM for GPΔMuc (Figure 1B). Because the primary receptor for orthoebolaviruses is intracellular, antibody neutralization predominantly occurs within late endosomes under acidic conditions [39,40]. Therefore, we investigated the impact of varying the pH on the binding activity of 2G1 to GPΔTM, GPΔMuc, and cleaved GP (GPcl). We found that a low pH had a minimal effect on the binding of 2G1 to GPΔTM and GPΔMuc, with half-maximal effective concentration (EC₅₀) values only increasing by 1.56 (11.58 vs 18.10 ng/mL) and 1.65 times (5.67 vs 9.33 ng/mL), respectively, at pH 5.5 compared to neutral conditions; a rather modest effect (Figure 1C). Considering the relatively low binding capacity of 2G1 for the receptor-bound GPcl, we postulated that 2G1 may not achieve neutralization by blocking the receptor. Next, we determined the neutralizing breadth of 2G1 using recombinant human immunodeficiency virus (rHIV) skeleton pseudoviruses carrying each of the orthoebolavirus GPs, except for that of BOMV, owing to its low packaging titer. 2G1 demonstrated potent and broad neutralization against all the tested orthoebolaviruses, with half-maximal inhibitory concentration (IC₅₀) values ranging from 0.02 to 0.44 μg/mL, outperforming the control antibodies mAb114 and FVM04 (Figure 1D).

Figure 1.

The mAb 2G1 exhibited a broad anti-orthoebolaviruses activity. (A) Recognition of 2G1 to the full length of orthoebolavirus GPs displayed on the cell surface. Data represent mean ± SD of three replicates from one representative experiment. An FITC-labeled irrelevant mAb was used as the control. (B) Binding kinetics of 2G1 to EBOV GPΔTM or GPΔMuc determined by surface plasmon resonance. Five representative curves from one representative test were fitted to compute the kinetic constants. (C) Binding activities of 2G1 to EBOV GPΔTM, GPΔMuc, and GPcl under various pH conditions tested by ELISA. (D) Neutralization capacities of 2G1, FVM04, and mAb114 against pseudotyped rHIV-EBOV (Makona), – EBOV (Mayinga), – SUDV, – BDBV, – TAFV, and – RESTV. Data are presented as the mean ± SD of three replicates from one representative experiment. (E) The amino acid divergence between the original and inferred genes of 2G1. (F) The binding abilities of the wild-type and inferred germline cross-pairing forms of 2G1 to EBOV GPΔMuc, tested by ELISA. (G) The binding curves of HCDR mutants of 2G1 to EBOV GPΔMuc.

To dissect the contributions of the heavy and light chains to binding, we prepared wild-type and inferred germline (IGL) cross-paired forms of 2G1. Focusing on the variable regions of the light chains (VL) and heavy chains (VH), a comparison revealed a greater amino acid divergence between VL-IGL and VL-WT (9/106) than between VH-IGL and VH-WT (6/124) (Figure 1E). The binding activities of 2G1-H_IGL/K_WT and 2G1-IGL to GP were significantly reduced (91.4-fold and 55.4-fold lower, respectively) compared to that of 2G1-WT (Figure 1F). In contrast, the binding ability of 2G1-H_WT/K_IGL was only 4.4-fold lower than that of 2G1-WT, underscoring the pivotal role of the heavy chain in binding. Further analysis through alanine scanning mutagenesis of the complementarity-determining regions (CDRs) of the heavy chain identified one critical residue in CDR2 (Y56), and three critical residues in CDR3 (C99, G100B, and C100D) (Figure 1G). Mutations at these positions resulted in a substantial decrease in the binding capacity of 2G1, with EC₅₀ values ranging from 59.9 to 133.9 times higher than that of the wild-type.

2G1 recognizes conserved epitopes of the GP-trimer and neutralizes orthoebolavirus by impeding the proteolysis of the GP

To elucidate the structural underpinning of the binding and neutralizing mechanism of 2G1, we determined the structure of 2G1-Fab in complex with EBOV GPΔMuc at a resolution of 2.98 Å using cryo-electron microscopy (cryo-EM) (Figure S1 and Table S1). The central connection of the 2G1 light and heavy chains was aligned nearly parallel to the main body of the GP1 subunit, with the capacity of three Fabs to engage with each GP trimer (Figure 2A). We superimposed the antigen–antibody (Ag-Ab) complex structure with those of KZ52 (PDB ID: 5HJ3) and ADI-15878 (PDB ID: 6EA7) from the Protein Data Bank (PDB), both of which also bind to GP Base [8,41]. The binding pattern of 2G1 was strikingly similar to that of ADI-15878, encompassing both the targeted region and the orientation angle of the Fab (Figure 2B). Like ADI-15878 [8], 2G1 recognized the pocket beneath the N-terminal tail of GP2, a region that bridges the GP1 and GP2 subunits from one GP protomer to the GP2 subunit from the other. However, some differences in the binding details between the two human bnAbs were observed. The epitope of ADI-15878 spanned the pocket and was almost evenly shared by two adjacent GP2 subunits, whereas 2G1 appeared to target the center of the pocket more directly (Figure 2C). 2G1, through the CDRs HCDR1-3, LCDR1, and LCDR3, buries an area of approximately 902 Ų on the GP surface. At the contact interface, 13 antibody residues form a robust interaction network with 15 antigen residues, including six hydrogen bonds (H-bonds) and abundant hydrophobic effects (Figure 2D). The epitope of 2G1 consists of four domains from GP [10]. Six of its residues are from the tip of the internal fusion loop (IFL), five are from HR1, three are from the core of GP1, and one is from the N-terminal region of GP2. According to previous analyses (Figure 1F), the heavy chain predominates in binding, accounting for most (11/13) of the amino acids in contact with GP. Notably, two residues contribute significantly to the Ag-Ab binding through the formation of more than one H-bond: the Oγ and mainchain O atoms of S100C^HCDR3 form H-bonds with the N atom of G528^GP and mainchain O atom of A526 ^GP, respectively; the Oϵ2 and mainchain O atoms of E564^GP form hydrogen bonds with the Oγ atom of S53^HCDR2 and the Oη atom of Y56^HCDR3, respectively (Figure 2D).

Figure 2.

2G1 recognized the hydrophobic pocket beneath the N-terminal tail of GP2 and inhibited GP proteolysis. (A) Front and top views of the cryo-EM structure of the EBOV GPΔMuc/2G1 Fab complex. Molecules are shown as surface representations. The GP1/2 subunits of three protomers A – C of GPΔMuc are colored in pale/blue violet, light/dodger blue, and light/hot pink, respectively. The VH of 2G1 is colored in lime green, and the VL is colored in goldenrod. (B) Superimposition of the 2G1, KZ52 (PDB ID: 5HJ3), and ADI-15878 (PDB ID: 6EA7) variable regions onto EBOV GPΔMuc. Molecules are shown as surface representations. 2G1, KZ52, and ADI-15878 were colored goldenrod, light sea green, and lime green, respectively. (C) Footprints of 2G1 or ADI-15878 on the EBOV GPΔMuc. The buried areas of 2G1 and ADI-15878 are indicated by goldenrod and lime green dotted lines, respectively. The paratope residues of 2G1 are shown as sticks. (D) Magnified view of the interface between 2G1 and EBOV GPΔMuc. EBOV GPΔMuc is represented as a molecular surface, the interface residues are shown as stick representations, and H-bonds are indicated by dotted black lines. GP1 (protomer A), GP2 (protomer A), GP2 (protomer B), 2G1 HCDR, 2G1 LCDR, oxygen atoms, and nitrogen atoms are colored pale violet, blue violet, dodger blue, lime green, goldenrod, red, and blue, respectively. (E) Relative binding percentage (%WT) of non-competing 2G1 and 4F1 to surface-anchored EBOV GP mutants. Data represent the means of three replicates of one representative experiment. Minor and pivotal epitopes of 2G1 are indicated in orange and dodger blue, respectively. (F) Conservation of critical 2G1 epitopes among filovirus GPs. (G) Antibodies blocking the binding of NPC1-C to GPcl. (H) Ability of antibodies or Fabs to inhibit the cleavage of EBOV GPΔMuc by thermolysin. THL, thermolysin; IMF, intermediate mass fragment.

To identify the critical residues targeted by 2G1, we generated single-point mutants of the full-length GP by mutating the residues on the interface as well as one flanking residue on each side. We used flow cytometry to identify the critical sites for 2G1 and a non-competing control antibody, 4F1 [26]. A562, T565, and L569 were identified as pivotal epitopes for 2G1, and mutations at these sites led to a substantial decrease in the relative binding activity of 2G1, ranging from 38.8 to 158.8 times lower than that of 4F1 (Figure 2E). Residues I33, P34, and G528 were identified as minor epitopes, with mutations resulting in a 1.8 to 2.1-fold reduction in the relative binding capacity. Notably, all critical residues, as well as G528, were completely conserved across filovirus GPs, and I33 and P34 exhibited a high degree of similarity (Figure 2F). Although the epitope of 2G1 is close to the N-linked glycan of N563 (Figure 2C), it does not participate in or impair 2G1 binding because 2G1 has comparable binding abilities with the N563A mutant of GP to the wild type (Figure 2E). Interestingly, two of the four critical amino acids of the antibody, C99 and G100B, and all three critical amino acids of the antigen are adjacent to the contact residues (Figures 1G, 2D, and E), suggesting that they might play an important structural role in Ag-Ab binding.

Given that the orthoebolaviruses enter target cells via non-specific adhesion and macropinocytosis [40], antibodies specific to GP usually exert neutralizing effects by obstructing GP proteolysis, receptor binding, or intracellular membrane fusion events. The control antibody MR72 effectively inhibited the binding of GPcl to its receptor, the domain C of the endosomal entry receptor Niemann-Pick C1 (NPC1-C) [13], whereas 2G1 did not demonstrate obvious blocking effect (Figure 2G), corroborating the earlier observation of the low binding activity of 2G1 to GPcl (Figure 1C). To ascertain whether 2G1 could impede the generation of activated GPcl, we used thermolysin to treat GPΔMuc pre-incubated with varying concentrations of 2G1 IgG or Fab. A significant reduction in the production of GPcl with increasing concentrations of 2G1 was observed, accompanied by an accumulation of GPΔMuc substrate and its cleavage intermediate forms (Figure 2H). The presence of Fab yielded similar outcomes, suggesting that 2G1 neutralizes orthoebolavirus by obstructing the proteolysis of GP.

The designed mRNA framework can efficiently and stably encode antibodies

We attempted to design an efficient mRNA framework for carrying 2G1. A typical mRNA module comprises elements such as the 5’ cap structure, a 5’ untranslated region (UTR), the coding sequence, a 3’ UTR, and a poly (A) tail. Based on a previously reported mRNA construct [23], we constructed an initial mRNA antibody module containing a Cap1 structure, a 5’ UTR (TEV), the coding sequence of 2G1 heavy or light chain, a 3’ UTR (F-I), a 100-nucleotide poly (A) tail, and interelement linkers (Figure 3A). Linear DNA templates are required for IVT of mRNA. We compared the effects of template preparation methods on the IVT products. In capillary electrophoresis, linearized plasmids yielded mRNA with superior uniformity compared to PCR-derived templates (Figure 3B). Using transfection reagents, we transfected a mixture of heavy and light mRNAs into HEK293 T cells and harvested the expression supernatants at multiple time points to determine the optimal detection time for the evaluation system. The concentration of functional 2G1 peaked at a limited maximum level of 74.4 ng/mL at 48 h post-transfection (Figure 3C).

Figure 3.

The optimized mRNA antibody module demonstrated an efficient translation ability. (A) The schematic of the initial mRNA module. Colored shapes represent different elements as indicated. (B) The uniformity of IVT products yielded using linearized plasmids or PCR-derived templates analyzed by capillary electrophoresis. (C) Temporal expression levels of unoptimized mRNA-2G1 (molar ratio of H: L = 1: 1) in the supernatant of 293T cells. (D–J) Optimized mRNA module design. The factors investigated include the cap1 or cap0 structure (D), with or without the intermediate linkers (E), UTR combinations (F and G), UTP or ψTP usage (H), poly (A) tail forms (I), and the original or optimized coding sequences (J). Data represent mean ± SD of three replicates from one representative experiment. (K) Validation of translation efficiency of the optimized mRNA module using two antibodies, 1E5 and 1B6, against henipavirus. (L) Determination of optimal molar ratio of heavy and light chains. (M) Size distribution of the LNP-encapsulated mRNA-2G1 nanoparticles. (N) Images of nanoparticles obtained under an electron microscope. (O) Expression levels of mRNA-2G1-LNP and naked mRNA-2G1 in 293T cell supernatants. (P) Detection of intact IgG and free light or heavy chains in the mRNA-2G1-LNP expression supernatant. (Q) The encapsulation rate and effective concentration of mRNA-2G1-LNP after 36 days of storage at 4°C. (R) Variation curve of in vitro translation efficiency of mRNA-2G1-LNP along with storage time.

We optimized the constituent elements of mRNA to improve its translational efficiency. First, we compared Cap1 and Cap 0, common in mammals, and observed no statistically significant divergence in their impact on antibody yield (Figure 3D). To streamline component replacement and optimization, we introduced linkers between the components of the initial module; however, these linkers could potentially perturb the secondary structure of the regulatory elements. Upon removal, a marked increase in the expression of the coding antibodies was observed (Figure 3E). Subsequently, we focused on identifying the optimal UTR components. Employing the 5’ and 3’ UTRs of human α1-globin (HBA1) and β-globin (HBB) alongside four UTR combinations previously reported [23,42–44], we determined that the natural HBA1 and HBB UTRs exhibited the highest translational efficiency (Figure 3F). We further exchanged the UTRs of HBB and HBA1 and introduced a new combination of 5’ UTR of the Xenopus laevis Beta globin (XBB) and 3'-HBA1 [20]. The mRNA antibodies (HBA1-HBB and HBA1) utilizing the 5’ UTR of HBA1 demonstrated the optimal translational efficacy, with antibody concentrations surpassing 1500 ng/mL (Figure 3G). Nucleotide modification contributes to reducing mRNA immunogenicity and avoids translation interruption in vivo [45]. However, substituting uridine triphosphate (UTP) for the N1-methyl-pseudouridylate (ψTP) had no significant effect on the efficiency of mRNA antibody translation at the cellular level (Figure 3H). Similarly, there was no significant difference in the antibody expression levels of mRNA-2G1 bearing two different forms of the poly (A) tail, A₃₀-linker-A₇₀ and A₇₀C₃₀ (Figure 3I). Optimization of the coding sequence moderately augmented antibody production, albeit without a significant difference from the non-optimized cohort (Figure 3J). Ultimately, we substantiated the universal applicability of the module, as the expression levels of mRNAs encoding two neutralizing antibodies, 1B6 and 1E5 [46], against henipaviruses reached 1835 and 5533 ng/mL, respectively (Figure 3K).

We mixed the heavy and light chains of mRNA antibodies into the cells at five different ratios and determined that the optimal molar ratio was 1:1.1 (Figure 3L). Next, we formulated LNP-encapsulated mRNA-2G1 (mRNA-2G1-LNP). The encapsulation efficiency of mRNA-2G1-LNP was approximately 89.7%, with an average particle diameter of approximately 84.6 nm (Figure 3M), exhibiting a spherical morphology under electron microscopy (Figure 3N). In 293 T cells, the expression of antibodies from mRNA-2G1-LNP was 36.6% higher than that of naked mRNA-2G1 transfected with lipofectamine (p < 0.01) (Figure 3O). In the expression supernatant, mRNA-2G1-LNP produced complete IgGs with a few free light chains (Figure 3P). To ascertain the stability of the LNP formulation, we stored mRNA-2G1-LNP at 4°C for eight weeks, monitoring changes in encapsulation efficiency, effective concentration, and cellular-level expression. On the 36th day, neither the encapsulation rate nor the effective concentration had changed significantly (Figure 3Q). The cellular-level expression of antibodies from mRNA-2G1-LNP at the storage endpoint remained at 52.4% of that observed in the second week (Figure 3R).

mRNA-2G1-LNP exhibited potent in vitro neutralization and in vivo protection against pseudotyped EBOV and SUDV

Next, we sought to determine whether mRNA-2G1-LNP could rapidly produce sufficient antibodies for neutralization or protection. First, we evaluated the neutralization efficacy of mRNA-2G1-LNP against rHIV-EBOV in vitro. A concentration of 0.1 μg/mL of mRNA-2G1-LNP, when added 6 h before infection, was capable of neutralizing 90% of the pseudovirus (Figure 4A). The IC₅₀ value of mRNA-2G1-LNP antibody against rHIV-EBOV was 16.4 ng/mL, and the neutralization capacity was 19.8 times that of IgG (Figures 1D and 4A). Next, we evaluated the in vivo protective activity of the mRNA-2G1-LNP in a bioluminescent-imaging-based BALB/c mouse infection model [47]. Mice were given mRNA-2G1-LNP via tail vein injection 12 h before the challenge, and all intervention groups demonstrated good protection efficacy (Figure 4B). The total luminescence value of 0.125 mg/kg in the lowest-dose group was only 11.9% of that in the PBS group (p < 0.0001), and no luminescence signal was detected in the high-dose groups (Figure 4C). Similarly, mRNA-2G1-LNP effectively neutralized rHIV-SUDV with an IC₅₀ of 32.6 ng/mL, which was 12.5 times the neutralizing activity of the IgG format (Figures 1D and 4D). In the mouse challenge model of the SUDV pseudovirus, prophylactic administration of mRNA-2G1-LNP provided sufficient protection, with the minimal dosage group of 0.125 mg/kg demonstrating a total luminescence intensity of only 6.72% of that of the control group (p < 0.0001), while the high-dose group completely blocked rHIV-SUDV infection (Figures 4E and F).

Figure 4.

mRNA-2G1-LNP potently inhibited the pseudotyped EBOV and SUDV infection. (A) The neutralization ability of mRNA-2G1-LNP against rHIV-EBOV. (B and C) The in vivo protection efficacy of mRNA-2G1-LNP against rHIV-EBOV in mice. The normalized bioluminescence intensity (p/s/cm2/sr) of images captured four days post-infection (B) was calculated and compared to the PBS group (C). (D – F) In vitro neutralization ability (D) and in vivo protective efficacy (E and F) of mRNA-2G1-LNP against rHIV-SUDV. (G) Serum antibody concentration. BALB/c mice (n = 5) were administered a single dose of mRNA-2G1-LNP, 2G1 antibody, or an equal volume of empty LNP. Antibody concentrations at the indicated times were determined using quantitative ELISA. (I – L) Comparison of biochemical criteria in the serum three days post-administration between the mRNA-2G1-LNP (1 mg/kg) and PBS groups. ALT, alanine aminotransferase; AST, aspartate aminotransferase; LDH, lactate dehydrogenase; CR, creatinine.

To investigate the expression kinetics and metabolic profile of mRNA-2G1-LNP in mice, we administered mRNA-2G1-LNP, 2G1 antibody, or unencapsulated LNP particles via tail vein injection. After the administration of mRNA-2G1-LNP, the serum antibody concentration increased rapidly (Figure 4G). At 6 h, the average serum antibody levels in the 1 and 0.5 mg/kg groups were 25.80 and 8.45 μg/mL, respectively. At 24 h, the serum antibodies reached their peak values: 31.73 and 11.36 μg/mL, respectively. On day 7, serum antibody concentrations decreased to 14.19 and 6.67 μg/mL, respectively. Notably, these concentrations were higher than the IC₉₀ value of 2G1 in vitro against authentic EBOV (3.2 μg/mL) [26]. The observed short half-life of human 2G1 in mice, estimated to be approximately three days (Figure 4G), may limit the potential for further escalation and prolongation of mRNA-encoded antibody levels. To assess the liver toxicity of mRNA-2G1-LNP in mice, we monitored body weight and serum biochemical criteria indicative of liver function post-administration, including alanine aminotransferase (ALT), aspartate aminotransferase (AST), lactate dehydrogenase (LDH), and creatinine (CR). The body weights of the mice in all of the groups exhibited an initial decline followed by recovery, then culminating in equivalent weights by day 14 (Figure 4H). There was no significant difference in body weight between those in the mRNA-2G1-LNP and IgG groups (p > 0.05). Early fluctuations in weight gain may have been related to the intensive blood collection efforts at that stage. A dose of 1 mg/kg mRNA-2G1-LNP did not significantly affect liver function in mice, and there was no statistical difference in the four biochemical criteria compared to that in the PBS group (Figure 4I–L).

Discussion

Orthoebolavirus disease, which is characterized by its severity, recurrent outbreaks, multiple pathogenic species, and cyclic variation during epidemics, poses great challenges to the development of drugs that prevent and treat it. The efficacy of vaccines in combating EBOV is well established, and several vaccines have received regulatory approval [4]. Nonetheless, it is difficult to roll out a vaccination in the absence of a pandemic. Studies involving non-human primates and human patients have indicated that mAbs are highly efficacious treatments for orthoebolavirus disease [5,10]. Despite the identification of numerous antibodies following the 2014–2016 West African outbreak, a limited number of mAbs exhibit broad-spectrum neutralizing activity against diverse orthoebolaviruses, and even fewer bnAbs demonstrate robust protective efficacy (≥60%) against EBOV in mice [48]. One murine-derived antibody, 6D6, and three antibodies, ADI-15878/15742/23774, isolated from survivors have demonstrated pan-binding and -neutralizing capabilities against five orthoebolaviruses (unknown against BOMV) [11,13]. Here, we delineated the properties of a cross-reactive mAb, 2G1, previously identified in vaccine-immune individuals [26]. To the best of our knowledge, 2G1 is the first reported antibody capable of binding to natural GP structures of all six orthoebolaviruses. Remarkably, 2G1 exhibited comparable or even better neutralizing activity against the remaining orthoebolaviruses than against EBOV, against which protective efficacy was validated in mice [26]. Despite the lack of data, previous studies have shown that in vitro neutralization is highly correlated with in vivo protection [48], and it is reasonable to speculate that 2G1 exerts broad anti-Ebola activity in vivo.

A systematic analysis of 168 antibodies elucidated the epitopes on the GP trimer (consisting of GP1-GP2 heterodimers), including glycan cap (GC), Head, Mucin, GP1 core, GP1/2, Base, IFL, and HR2 [48]. Of these epitopes, the more spatially accessible GC, Head, and Mucin, which are distant from the viral membrane, have been shown to possess higher immunodominance during vaccination [49]. However, these regions are hotspots of interspecies sequence divergence and natural evolutionary mutations [10]. Antibodies targeting the IFL, GP1 core, and Head demonstrated high cross-reactivity frequencies, with IFL being most strongly associated with pan-orthoebolaviruses activity. The epitopes of 6D6 [11,50], CA45 [12], ADI-23774 [13,51,52], ADI-15878 [8,13], ADI-15742 [13], and 1C11 [53], which are known to neutralize at least four orthoebolaviruses, are located (fully or partially) in the IFL domain. Notably, 2G1 that was derived from vaccinees, as well as ADI-15878 that was obtained from survivors, recognized the hydrophobic pocket situated beneath the N-terminus of GP2 (Figure 2C) [8], indicating the significance of this region as a target for the development of pan-orthoebolavirus drugs or vaccines. Although 2G1 shares a binding pattern with ADI-15878, there are visible distinctions between the two antibodies. 2G1 recognizes a quaternary epitope composed of the IFLtip, HR1, GP1 core, and GP2 N-terminus, which combines the epitopes of the most promising broad-spectrum cocktail components against orthoebolavirus disease: ADI-15878 (IFL + HR1) and ADI-23774 (3₁₀ pocket + IFL base + GP2 N-terminus) [8,51]. In addition to differences in epitopes, ADI-15878 exhibits robust binding to GPcl and is hypothesized to impede the fusion-triggering process after receptor engagement [13]. Conversely, 2G1 demonstrated weak binding to GPcl and neutralized the virus by obstructing the upstream cleavage of GP. In vivo, a single dose of 300 μg ADI-15878 administered two days post-challenge provided 80% (8/10) protection in a mouse model with a 95% (19/20) mortality rate [13], whereas a single dose of 100 μg 2G1 administrated one-day post-challenge conferred 100% (10/10) protection in a 100% (10/10) mortality mouse model [26]. Although a direct comparison is not available to determine whether ADI-15878 or 2G1 is superior, the potential of 2G1 as a candidate for therapeutic intervention or inclusion in a pan-orthoebolavirus cocktail is credible.

In recent years, mRNA technology has rapidly developed under the influence of the novel coronavirus epidemic, and its advantages of rapid construction and cell-free manufacturing have made it a promising modality for biomedical development. In this study, we constructed an mRNA module capable of encoding the pan-orthoebolavirus antibody 2G1. Consistent with previous findings [54,55], UTR, in particular the 5’ UTR, is pivotal in influencing translational efficiency. However, the UTR elements employed in other mRNA-based vaccines or antibodies have not been optimal for encoding 2G1. Our results revealed that the UTR combination derived from human HBA1 yielded the highest functional 2G1 expression levels in vitro. Although the 5’ and 3’ UTRs of HBA have both been utilized separately in mRNA modules [20,42], their combined use in an unedited form has not been reported. Other factors, such as cap structure, chemical modification, codon optimization, and poly (A) tail configuration, did not significantly affect mRNA antibody expression. Ultimately, we developed an mRNA antibody framework comprising the Cap1, the 5’ UTR (HBA1), the antibody sequence, the 3’ UTR (HBA1), and a poly A₁₀₀ tail. This framework demonstrated good compatibility, with two additional tested macaque-human chimeric antibodies showing even higher functional antibody expression levels. The in vitro neutralizing activity of the mRNA-2G1-LNP exceeded that of the IgG format by more than an order of magnitude. In vivo, a 0.25 mg/kg preventative dose completely inhibited pseudoviral infection, underscoring its potential in addressing the orthoebolavirus threat. Whether the expression level of mRNA-2G1-LNP in vivo is sufficient to provide adequate protection requires further validation using live virus models. We noticed that the half-life of the 2G1 protein in mice was comparable to that of HB27, a human antibody against SARS-CoV-2, and the transient expression levels of naked mRNA for both were similar in 293 T cells [20]. However, mRNA-HB27-LNP exhibited a significantly prolonged circulating half-life, which may be attributed to the quality of the LNP formulation, given that mRNA-HB27-LNP utilized a clinical-grade encapsulation platform and process.

Preclinical studies on mRNA passive immunotherapy have been extended to various diseases, including infectious diseases [20–22,56–58], tumors [22,23,59], toxins [22], and neurodegenerative diseases [24]. Although a phase I clinical trial for an mRNA antibody against the Chikungunya virus has been completed, a gap exists before mRNA passive therapy can be applied. Besides the long-term risks associated with mRNA that are yet to be comprehensively understood, safety concerns related to the delivery vehicle [60], particularly at high doses, for example in antibody therapies, is a key issue that demands further exploration. As for the present study, two additional limitations are noteworthy. Firstly, the lack of BSL-4 laboratory resources prevented us from obtaining data confirming the protective effect of mRNA-2G1-LNP against live viruses. Secondly, the potential reasons for the short half-life of 2G1 and its mRNA formulation observed in mice need to be identified to prevent restrictions of its future application.

In summary, this study confirmed the pan-orthoebolavirus reactivity of 2G1 and elucidated the structural underpinnings and neutralization mechanisms of its activity, thereby providing a candidate component and design rationale for the development of broad-spectrum drugs against orthoebolaviruses. The construction of a highly effective mRNA-2G1 provides a potential emergency prevention measure for orthoebolavirus disease, and can serve as a reference for the module design of other mRNA-based biological therapies.

Funding Statement

This work was supported by National Natural Science Foundation of China [grant number: 82204260].

Disclosure statement

W.C., C.Y., P.F., and T.F. are inventors of one CN patent entitled “A monoclonal antibody 2G1 against the GP2 subunit of Ebola virus glycoprotein and its application” (Patent No. ZL201811309856.1). This patent characterized the antibody described in the manuscript and does not impose any restrictions on the publication of the data. The remaining authors declare no competing interests.

Data availability statement

The cryo-EM structure of 2G1-Fab complexed with the EBOV GPΔMuc has been deposited in the Electron Microscopy Data Bank (EMDB) and Protein Data Bank with the accession codes EMDB-38899 and PDB ID 8Y3U. This study also used the structures 5HJ3, 6EA7, and 7KEJ from the Protein Data Bank. Further requests for materials and reagents should be directed to the corresponding authors.

Author contributions

W.C. and C.Y. conceived the research; P.F. conducted structure and mechanism studies; B.S. and Z.L. performed neutralization assays and mRNA experiments; T.F., Z.S., and J.L. prepared proteins and constructed mutants of GP or antibodies; Y.R. executed kinetic analyses; X.Z. and Y.L.Y. encapsulated mRNA-LNP particles; P.F. wrote the manuscript; all authors read and approved the manuscript.