A versatile antibody capture system drives specific in vivo delivery of mRNA-loaded lipid nanoparticles

Materials

mRNA synthesis

All in vitro transcription mRNA were synthesized from a PCR template containing a T7 promoter upstream followed by the codon-optimized open reading frame. All constructs contain a 5’ UTR, 3’ UTR and 125 polyA tail. All mRNA was transcribed using the HiScribe T7 High Yield RNA Synthesis Kit (New England Biolabs). Capping was cotranscriptionally performed using CleanCap Reagent AG (TriLink Biotechnologies). All uridine was replaced with N1-methylpseudouridine (TriLink Biotechnologies). In vitro transcription reaction was treated with DNase to eliminate the template, and dsRNA was removed by cellulose clean-up methods³⁸. The final product was purified by sodium acetate precipitation.

LNP formulation

LNPs were formulated as previously reported with modification⁸. A lipid mixture consisting of Dlin-MC3-DMA or SM102, DSPC (Avanti Polar Lipids), cholesterol (Sigma) and DMG-PEG₂₀₀₀ or DSPE-PEG₂₀₀₀ (Avanti Polar Lipids) was prepared in ethanol as a 20-mM stock. The molar composition used was 50:10:38.5:1.5 molar ratio. The lipid solution was mixed by flowing through a microfluidic mixing device Nanoassemblr (Precision Nanosystems) with an aqueous mRNA solution in 10-mM citrate buffer (pH 4) at a 1:3 organic to an aqueous volume ratio at a total flow rate of 4 ml per min. The resulting LNPs were then diluted twice with PBS (pH 7.4) immediately and further dialysed overnight. Then, the next day, LNPs were filtered through a 0.22-µm filter.

LNP characterization

The size distribution, particle number per millilitre and mode size of LNP was measured using NanoSight NS300 (Malvern Panalytical). The zeta potential of LNPs was measured using a Zetasizer Nano ZS (Malvern Panalytical). Total mRNA content and encapsulation efficiency were determined by performing a standard RiboGreen (Thermo Fisher) assay.

Molecular cloning of nanobody

Genes encoding the TP1107 sequence were synthesized as a gene fragment (Integrated DNA Technologies) for cloning into a pET His6 TEV LIC cloning vector²³. Plasmids will be deposited to the Addgene repository.

Single-domain antibody expression and purification

pET-TP1107 was co-transformed alongside pEVOl-pAzF into B-95.ΔA E. coli, which expresses the orthogonal machinery for the incorporation of azPhe in recognition of UAG codon during protein translation²⁵. The B-95.ΔA E. coli strain is a unique expression vector in which 95 of its original UAG codons have been replaced along with the elimination of release factor 1 to facilitate the improved incorporation efficiency of azPhe²⁶.

An overnight culture was inoculated into fresh Terrific Broth media with appropriate antibiotics and grown at 37 °C and shaking until the optical density, OD₆₀₀, reached 0.7–1.0. The single-domain antibody (sdAb) expression was induced by the addition of IPTG (1 mM), l-arabinose (0.02%) and azPhe amino acid (2 mM). Protein expression was continued for a further 12–14 h at 30 °C before harvesting the bacteria by centrifugation. Bacterial pellets were harvested by centrifugation (4,000g, 20 min) and resuspended in Ni-NTA wash buffer followed by cell lysis using a high-pressure homogenizer (Avestin Emulsiflex C5).

On lysis, cell debris was centrifuged (12,000g, 30 min) and the supernatant was collected for purification via an immobilized metal affinity chromatography column. Additional size exclusion chromatography was used to remove non-specifically bound proteins using Superdex 75 10/300 GL gel filtration column (GE Healthcare). The sdAb concentration was determined using Nanodrop (Thermo) spectrophotometer at 280 nm.

Negative-stain TEM

Negative-stain TEM was carried out by applying 3 µl of a 0.05 mg ml⁻¹ solution onto a continuous carbon TEM grid (EMS 300 mesh), which was pretreated in a plasma chamber (30 s, 15-mA plasma current) followed by multiple applications of uranyl formate (0.01% w/v). Imaging was performed on a Thermo L120C TEM device at a magnification of 92k, yielding a physical pixel size of 1.55 Å per pixel. Here 41 images were recorded on a Ceta direct electron detector. Single-particle analysis was carried out in the RELION v. 3.1.2 software package²⁴. Briefly, images had their CTF parameters estimated followed by automated particle picking and successive rounds of two-dimensional classification to homogenize the particle stack. This yielded 19k particles for ab initio three-dimensional (3D) model generation, which was carried out in cryoSPARC³⁹ and further 3D refinements were finalized in RELION, resulting in an ~16-Å 3D reconstruction. A Protein Data Bank (PDB) model was generated by initially performing rigid-body fitting of the mouse IgG (PDB ID 1IGY (ref. ⁴⁰)) using UCSF Chimera⁴¹. This fitted model then underwent a molecular dynamics flexible fitting refinement using UCSF ChimeraX/ISOLDE^42,43 to fit the antibody into the 3D volume. The resultant PDB was then used as a template for HADDOCK docking^{44,45,46,47,48} of the nanobody. The docking clusters that best fit the experimental density were then considered for a further round of molecular dynamics flexible fitting with tight torsion and distance restraints based on the original 1IGY model and the starting model for the nanobody.

Conjugation of TP1107 to DBCO-PEG₂₀₀₀-DSPE

Azide-incorporated TP1107_optimal can be directly conjugated onto DBCO-PEG₂₀₀₀-DSPE through Strain-Promoted Alkyne-Azide Cycloaddition (SPAAC) chemistry. The conjugation mixture was prepared at a DBCO:azide molar ratio of 2:1. Meanwhile, to illustrate the effect of randomly oriented sdAbs, 2 molar excess of NHS-azide (198 Da, Thermo) was initially conjugated onto TP1107 sdAb. Excess unconjugated NHS-azide linkers were removed using a 7 K MWCO Zeba desalting column (Thermo). The azide-modified/azide-incorporated sdAbs were mixed with DSPE-PEG₂₀₀₀-DBCO at a 0.5 molar excess and left for 24 h at 37 °C as TP1107_random. No further purification was required.

Post-insertion of active targeting module into LNPs and functionalized mAb-TP1107_{optimal/random} LNPs

The post-insertion of DSPE-PEG₂₀₀₀-TP1107 was performed by adding lipidated nanobody to the LNP solution at 0.5% w/w, followed by brief mixing before incubating at 4 °C for 48 h unmixed. The free TP1107 or unreacted DSPE-PEG₂₀₀₀-DBCO was removed via Amicon 100 kDa MWCO (Merck) ultrafiltration. A total of five washes (2,000 rpm, 10 min) were completed. Functionalized LNP was prepared by mixing antibodies with TP1107_{optimal/random} LNP at the selected ratio and incubated 4 °C overnight.

Calculation of TP1107 per LNP

The sdAb number of each LNP calculated as below:

where the concentration of sdAb was calculated from western blot using JESS Simple Western (Bio-Techne) and the number of LNPs was measured by NanoSight NS300 (Malvern Panalytical).

Targeting antibody

Targeting antibody used for in vitro and ex vivo studies is anti-hTfR (OKT9, purchased from WEHI), mouse anti-hCD3 antibody (UCHT1, Thermo Fisher), mouse anti-hCD4 (SK3, BioLegend), mouse anti-hCD5 (UCHT2, Thermo Fisher), mouse anti-hCD7 (124-1D1, Thermo Fisher), mouse anti-hCD22 (eBio4KB128 (4KB128), Thermo Fisher) and mouse IgG1 kappa isotype control (P3.6.2.8.1, Thermo Fisher). Targeting antibody used for the in vivo study is mouse anti-mouse CD3ε (QA17A05, BioLegend).

Conjugation of mTfR to DSPE-PEG₂₀₀₀-DBCO

A 5 molar excess of NHS-azide was initially conjugated to purified mAb_TfR (OKT9), a kind gift from J. Mintern’s group. Excess unconjugated NHS-azide was removed by a 7K MWCO Zeba desalting column (Thermo Fisher) following the manufacturer’s protocol. The azide-mTfR was incubated with DSPE-PEG₂₀₀₀-DBCO at 37 °C overnight at a DBCO:azide ratio of 2:1. No further purification is required.

Preparation of mAb_lysine LNPs

Roughly 0.05% w/w of mAb-PEG2000-DSPE mixture was added to the formulated LNPs. The reaction was incubated at 4 °C for 48 h. The post-inserted LNP was then concentrated by an ultrafiltration system and slowly applied to a 90-cm-bed-length gravity-flow size exclusion column prepared with Sepharose-CL4B gel^12,13. The mobile phase was PBS. The fractions that contained LNP were collected and concentrated by the ultrafiltration system. The mRNA concentration was measured by a RiboGreen assay, as described before. The particle size was determined by NTA.

Cell culture maintenance

Jurkat cells were maintained with RPMI media (Gibco) supplied with 10% foetal bovine serum (FBS) and penicillin/streptomycin (100 U ml⁻¹). Cells were cultured at 37 °C in a humidified incubator with 5% atmospheric CO₂ along with routine testing or mycoplasma contamination.

Mouse models

B6.Cg-Gt(ROSA)26Sortm14(CAG-tdTomato)Hze/J (IMSR_JAX: 007914) mice were purchased from The Jackson Laboratory and maintained locally at the Monash animal research platform, and mice were ordered and shipped for the experiment on request. C57BL/6J mice were obtained from the Monash animal research platform. Male and female mice aged 6–14 weeks were used in the experiments. Ai14 mice were used for the data shown in Fig. 6 and Supplementary Fig. 14. C57BL/6J mice were used for the data shown in Supplementary Fig. 14. All the experimental procedures followed the protocols approved by the Institutional Animal Care and Use Committee at Monash University, and the experimental plan was approved by the Monash Office of Research Ethics and Integrity Committee under ethics approval nos. 37404 and 41587. All the mice groups were randomized and gender balanced.

Human PBMC collection and purification

Healthy donors aged between 18 years and 50 years old in both sexes were recruited voluntarily after the invitation to participate. Ethics is approved by the Monash University Human Research Ethics Committee, application ID 37405. Human blood was collected as per the experimental plan. Here 10–30 ml of human blood was collected and diluted with PBS before carefully layered on Ficoll-Paque PLUS density gradient media at 1:1 v/v. A PBMC layer was collected after 400g, 40-min spin and washed with prewarmed RPMI media twice. PBMCs were either used for experiments or frozen in cell-freezing media at –80 °C.

Antibody-capturing LNP safety assessment and cytokine measurement

C57BL/6 mice were acquired under ethics approval no. 37404. The mice received intravenous injections of 0.1 mg kg⁻¹ of various LNP formulations or control groups, vehicle, unmodified SM102/DSPE LNP, unmodified SM102/DSPE LNP and mCD3 LNP with SM102 and DSPE-PEG₂₀₀₀. Blood samples were collected at 6 h and 24 h post-injection to evaluate the impact on liver enzymes (alanine transaminase and aspartate transaminase) and cytokine release. After 24 h, the animals were euthanized for liver and spleen harvesting and subsequent histological analysis. The harvested liver and spleen tissues were fixed in 10% neutral-buffered formalin for at least 48 h. Two mice from each group were assessed by a veterinary pathologist for further evaluation. Plasma was obtained by centrifuging blood samples at 1,000g for 5 min at 4 °C and stored as single aliquots. The levels of mouse cytokines IL-1α, IL-1β, IL-10, IL-6, MIP-1α, MCP-1, IL-2, TNF-α, IFN-γ and IL-4 were measured using the BD Cytometric Bead Array Mouse Flex Set according to the manufacturer’s protocol. Data collection was performed using a Stratedigm S1000EXi flow cytometer.

Cytokine measurement for the whole blood stimulated with targeted LNP

Healthy donor’s blood was collected on the day of the experiment in heparin-coated collection tubes. Different targeted LNPs were added to each well at a final concentration of 2 ng μl⁻¹ and incubated for 24 h. The blood samples were then centrifuged to collect plasma for cytokine measurement. Human cytokines IFN-γ, IL-1α, IL-1β and TNF were quantified using the BD Cytometric Bead Array Human Flex Set according to the manufacturer’s protocol. Data collection was performed using a Stratedigm S1000EXi flow cytometer.

Cell association and transfection assay with functionalized LNPs

To assess the binding and transfection efficiencies of functionalized LNPs in Jurkat cells, approximately 50,000 or 100,000 cells were added to individual wells in a 96-well plate. A final concentration of 0.5 ng µl⁻¹ or 1 ng µl⁻¹ of mRNA was added to the cells and incubated at 37 °C for varied time periods. Subsequently, cells were washed three times with 2% FBS/PBS following centrifugation at 400g for 5 min. Cells were then resuspended in 80 µl of 2% FBS/PBS, and the MFI was quantified using a Stratedigm S1000EXi flow cytometer. eGFP and Cy5 fluorescence were excited at 488 nm and 642 nm, respectively, with fluorescence emission collected at 520/20 nm and 676/29 nm.

Human PBMC association and transfection assay with functionalized LNPs

To assess the binding and transfection efficiencies of the functionalized LNPs, approximately 500,000 PBMCs were added to individual wells in a 96-well plate with functionalized LNPs at a final concentration of 1 ng µl⁻¹. Then, cells were incubated at 37 °C for 24 h. Then, PBMCs were washed thrice with 2% FBS/PBS after centrifugation at 400g for 5 min. To phenotype the subpopulations, cells were stained against αCD3-PE mAb (clone OKT3, BioLegend), αCD4-BV510 mAb (clone OKT4, BioLegend), αCD8-BV786 mAb (clone SK1, BioLegend), αCD19-BV421 mAb (clone HIB19, BioLegend), αCD14-Alexa Fluor 700 mAb (clone HCD14, BioLegend), αCD56-BV605 mAb (clone 5.1H11, BioLegend) and viability dye (eBioscience Fixable Viability Dye eFluor 780, Thermo Fisher) on ice for 30 min. All the antibodies were used at 1:200 dilutions with Human TruStain FcX (BioLegend) as per the manufacturer’s protocol. After washing away the excessive antibody, cells were resuspended with 100 µl of 2% FBS/PBS for the flow analysis (Stratedigm S1000EXi). Cells were identified by a combination of surface markers: CD4⁺ T cells (CD3⁺ T cells and CD4⁺ T cells), CD8⁺ T cells (CD3⁺ T cells and CD8⁺ T cells), monocytes (CD3⁻, CD19⁻, CD56⁻ and CD14⁺), NK cells (CD3⁻, CD19⁻, CD14⁻ and CD56⁺) and B cells (CD3⁻ and CD19⁺). eGFP and Cy5 fluorescence was excited at 488 nm and 642 nm with fluorescence emission collected at 520/20 nm and 676/29 nm, respectively. Data in Supplementary Fig. 8 were obtained via a Cytek Aurora five-laser full spectrum cytometer.

In vivo assessment of CD3 targeting of LNPs to T cells across multiple organs

Ai14 mice were injected intravenously with unmodified LNPs, CD3-targeted LNPs or isotype control LNPs loaded with Cre mRNA. After 24 h, blood was collected via cardiac puncture, and the mice underwent transcardiac perfusion with PBS to remove circulating blood. Red blood cells were lysed using ammonium–chloride–potassium buffer (Thermo Fisher) at a 1:10 (v/v) ratio twice, followed by washing with 2% FBS/PBS. The liver, spleen and lymph nodes (inguinal, iliac and cervical) were collected and processed as follows.

The liver was minced and digested using a gentleMACS dissociator with 2.8 mg ml⁻¹ of collagenase H and 0.28 mg ml⁻¹ of DNase. The digested mixture was filtered to remove the undigested material and subjected to a slow spin at 60g. The supernatant was collected and spun down to collect the pellet. The pellet was resuspended in 30% Percoll media and spun to remove hepatocytes, followed by resuspension with ammonium–chloride–potassium lysis buffer and washing with 2% FBS/Hanks’ balanced salt solution before antibody staining.

The spleen was minced with 1 mg ml⁻¹ of collagenase III and 0.28 mg ml⁻¹ of DNase and digested by constant, gentle mixing until fully digested. Cells were filtered and red blood cells were lysed using an ammonium–chloride–potassium buffer.

Lymph nodes were collected and homogenized by passing through a 0.45-µm filter. Dissociated cells were collected and washed with media.

All the immune cell pellets were stained with a flow cytometry panel containing the following antibodies: αCD3e-BV650 mAb (clone 145-2C11, BD Biosciences), αCD90.2-BV650 mAb (clone 53-2.1, BD Biosciences), αCD4-APC-Cy7 mAb (clone GK1.5, BioLegend), αCD8-BV711 mAb (clone 53-6.7, BD Biosciences), αCD19-BV786 mAb (clone 1D3, BD Biosciences), αCD11b-BV421 mAb (clone M1/70, BioLegend), αLy6C-BUV661 mAb (clone HK1.4.rMAb, BD Biosciences), αLy6G-BV605 mAb (clone 1A8, BD Biosciences), αCD45-Pacific Blue mAb (clone S18009F, BioLegend), αI-A/I-E-BV510 mAb (clone M5/114.15.2, BioLegend), αF4/80-PE/Dazzle mAb (clone BM8, BioLegend) and αCD11c-Alexa Fluor 700 mAb (clone N418, BioLegend). Additionally, Mouse BD Fc Block and viability dye (LIVE/DEAD Fixable Blue Dead Cell Stain Kit, Thermo Fisher) were included. Samples were incubated on ice for 30 min, followed by washing to remove excess antibody.

Flow cytometry was performed using a Cytek Aurora five-laser cytometer, and data were analysed using FlowJo v10.10.0 (BD Biosciences). Leucocyte phenotyping was conducted using the following markers: CD4⁺ T cells (CD45⁺, CD11b⁻, CD3e⁺ or CD4⁺); CD8⁺ T cells (CD45⁺, CD11b⁻, CD3e⁺ or CD8⁺); dendritic cells (CD45⁺, CD3e⁻, CD19⁻, CD11c⁺ or MHCII⁺); monocytes (CD45⁺, CD11b⁺, Ly6C⁺ or Ly6G⁻); neutrophils (CD45⁺ CD11b⁺, Ly6C⁺ or Ly6G⁺); macrophages (CD45⁺ CD11b⁺, Ly6C low, Ly6G⁻, F4/80⁺ or SSA low); and CD19⁺ B cells (CD45⁺, CD11b⁻, CD3⁻, CD19⁺ or MHCII⁺).

Statistics and reproducibility

Data are presented as mean ± standard deviation (s.d.) based on the data obtained from at least n = 3 independent experiments, wells or mice. Statistical significance was determined using GraphPad Prism 9.0 and stated in each figure legend.

Reporting summary

Further information on research design is available in the Nature Portfolio Reporting Summary linked to this article.