An engineered T7 RNA polymerase that produces mRNA free of immunostimulatory byproducts

Materials

Synthetic oligonucleotides were obtained from Integrated DNA Technologies. NTP solutions were acquired from Hongene. α-32P-GTP and α-32P-CTP were purchased from Perkin Elmer. Pyrophosphatase was purchased from New England Biolabs. Plasmids were transformed into BL21 chemically competent E. coli cells. WT and mutant T7 RNAP were overexpressed as N-terminal hexahistidine-tagged variants and purified by Ni-NTA affinity chromatography. C-terminal foot substitutions were generated by PCR using NNK degenerate primers. A total of 96 bacterial colonies were Sanger-sequenced to identify 17 of the 20 desired substitutions.

Computational structure preparation

Crystal structures of T7 RNAP in the IC (PDB: 1CEZ, 2PI4, 1QLN), the EC (PDB: 1MSW, 1H38, 1S76) and an intermediate state (PDB: 3E2E) were computationally prepared by removing water, ions and nucleic acids and then rebuilding missing loops and sidechains using the comparative modeling protocol RosettaCM (flags in Supplementary Tables 14)47. DNA and RNA were reintroduced into the rebuilt protein structures by aligning the protein in the original complex with the rebuilt lone protein and then copying the coordinates of the nucleic acids into the rebuilt structure. The resulting DNA–(RNA)–protein complexes were subsequently minimized with heavy-atom crystallographic constraints using the Rosetta constrained minimization protocol (flags in Supplementary Table 5). These minimized models were assessed for quality by checking root mean square deviation from the reference crystal structures and filtering out models with any steric clashes or other residue-level abnormalities highlighted by the Rosetta score function.

Computational modeling of folding free energy change

Fixed-backbone folding free energies of mutation changes (∆∆Emut; in Rosetta Energy Units and distinguishable from ∆∆Gmut) were calculated using the Rosetta macromolecular modeling suite35,48. Structures were first prepared using the structure preparation protocol discussed earlier. Next, all possible NTD single substitutions in the prepared IC and EC structures were scanned using the Rosetta ddg_monomer protocol (flags in Supplementary Table 6, with example table provided as Supplementary Table 7)48. Results were tabulated and summarized using the Python Pandas package. Secondary structures of IC and EC structures were determined from the PDB secondary structure records and by DSSP49.

Computational modeling of protein cavities

Changes in protein void volume due to C-terminal foot insertions were calculated using the ProteinVolume software tool36. First, insertion mutants were modeled by building structures with the Ala insertion (884A) using RosettaCM (as described in ‘Computational structure preparation’ but with an 884-mer insertion sequence instead of the 883-mer native sequence) and then using Rosetta’s ddg_monomer protocol to introduce the other 19 residue types at position 884. Twenty-four 884A models were generated from a set of five parental T7 RNAP crystal structures representing the IC (PDB: 1CEZ, 2PI4), the EC (PDB:, 1MSW, 1S76) and an intermediate state (PDB: 3E2E). These 24 884A models served as starting points for a second ddg_monomer stage, in which the ddg_monomer protocol yielded five structure models for each insertion mutant (including A884A), resulting in a total of 2,400 structure models (24 starting structures × 20 mutants × 5 structures per mutant). The solvent-excluded, van der Waals and void volumes for each of these 2,400 structures were calculated using ProteinVolume v.1.3 with default parameters (starting probe size 0.8, ending probe size 0.2 and surface minimum distance 0.1).

Statistics

The Pearson correlation coefficient was calculated using the pearsonr function from the Python package scipy.stats. This correlation coefficient is calculated as:

$$r = \frac\nolimits_^n \left( \right)\left( \right)}}\nolimits_^n \left( \right)^2\mathop \nolimits_^n \left( \right)^2} }}$$

where mx is the mean of vector \(}}} = \\), my is the mean of vector \(}}} = \\) and n is the sample size.

mRNA synthesis and characterization

mRNA was synthesized in vitro by T7 RNAP-mediated transcription at 37 °C using 100% substituted N1-methylpseudouridine-triphosphate and a linearized DNA template, which incorporated the 5′ and 3′ untranslated regions and a polyadenosine tail. Reactions with WT and mutant enzymes were treated similarly. NTPs were included at equimolar concentrations. After transcription, the Cap 1 structure was added to the 5′ end using Vaccinia capping enzyme (New England Biolabs) and Vaccinia 2′O-methyltransferase (New England Biolabs). The mRNA was purified by oligo-dT affinity purification. For mRNAs described as ‘with RP’, ion-paired reversed-phase (RP) chromatography was subsequently used for purification. All mRNAs were buffer exchanged by tangential flow filtration into sodium citrate, pH 6.5, and sterile filtered. The mRNA was kept frozen at –20 °C until further use.

LNP production and characterization

The mRNA was encapsulated in a lipid nanoparticle through a modified ethanol-drop nanoprecipitation process described previously50. Briefly, ionizable, structural, helper and PEG lipids were mixed with mRNA in acetate buffer, pH 5.0, at a ratio of 3:1 (lipids/mRNA). The mixture was neutralized with Tris-Cl, pH 7.5, sucrose was added as a cryoprotectant and the final solution was sterile filtered. Vials were filled with formulated LNP and stored frozen at –70 °C until further use. The final drug product underwent analytical characterization, which included the determination of particle size and polydispersity, encapsulation, mRNA purity, osmolality, pH, endotoxin and bioburden, and the material was deemed acceptable for in vivo study.

RNA yields

Yields of purified RNA were measured by ultraviolet spectrophotometry at 260 nm. A standard extinction coefficient for RNA of 40 ng-cm µl−1 was used to determine concentration.

Radioactive-gel-based analysis of IVT products

The formation of short RNAs or abortive transcripts and dsRNA products was assessed using IVT in combination with 32P-labeled NTPs. In vitro transcription reactions were performed in a total volume of 50 µl containing 400 nM WT or mutant T7 RNAP; 7.5 mM each of ATP, CTP, GTP and UTP; 0.5 µCi µl−1 α-32P-GTP (to label abortive species) or α-32P-CTP (to label reverse-complement species); 100 U ml−1 pyrophosphatase and 1.8 µM dsDNA template in 1× IVT buffer (New England Biolabs). The dsDNA template was generated by annealing synthetic DNA sequences containing the T7 promoter (5′-TAATACGACTCACTATAGGGAAATAAGAGAGAAAAGAAGAGTAAGAAGAAATATAAGAGCCACCAAAAAAAAAAAAAAAAAAAATCTAG-3′) in 1× phosphate-buffered saline (PBS). This template has a G-rich initiation sequence of 42 nucleotides, enhancing T7 RNAP EC transition, before the first incorporation of CTP at position 43. Thus, reactions using α-32P-GTP selectively label abortive species, full-length RNA, run-on transcripts and large RNA species generated from loopback transcription. By contrast, reactions incorporating α-32P-CTP label selectively labeled reverse complements, full-length RNA, run-on and loopback products. IVT reactions were incubated for 2 h at 37 °C and quenched by addition of 80 mM EDTA. RNA was isolated by ethanol precipitation, and pellets were resuspended in 1× RNA loading dye. Samples were briefly heated (70 °C for 1 min) and loaded onto 20% acrylamide, 6 M urea gels and electrophoresed for 30 min at 20 W followed by 2 h at 40 W. Gels were exposed to phosphor screens (1 h to overnight) and imaged using a Typhoon FLA9500 Biomolecular Imager. The gels were viewed under two contrast levels: low contrast levels were used to compare full-length mRNA yields, and high contrast levels were used to determine the impurity profile (original gels are shown in Supplementary Fig. 3). The radioactive assay offers the advantage of high sensitivity and the ability to label all IVT impurities.

RNase T1 digest assay

To allow for high-throughput, quantitative analysis of run-on and loopback transcripts, an RNase T1-based assay was used as described previously by Jiang et al.37. RNase T1 selectively cleaves single-stranded RNA after guanosine residues, which facilitates the identification of run-on transcripts when using DNA templates encoding terminal 3′ guanosine. RNase T1 treatment does not affect run-off transcripts, allowing them to maintain their 3′-hydroxide; however, in the presence of run-on transcripts, cleavage of the 3′ end results in a 3′-monophosphate scar. For RNase T1 digestion, 40 μl mRNA (1 mg ml−1) was mixed with 60 μl urea (8 M; Sigma-Aldrich), 12 μl Tris-HCl (1 M at pH 7.0; Invitrogen) and 0.8 μl EDTA (0.5 M; Invitrogen, Thermo Fisher). The samples were then denatured at 90 °C for 10 min and cooled at room temperature. RNase T1 (20 μl of 1,000 U μl−1, Thermo Fisher Scientific) was added, and the samples were incubated at 37 °C for 15 min. Treated samples were separated and analyzed via reversed-phase ion-pairing liquid chromatography using an Agilent 1290 UPLC and Agilent 6530 quadrupole time-of-flight mass spectrometer.

dsRNA ELISA assay

A complementary method to the RNase T1 digest assay is the dsRNA sandwich enzyme-linked immunosorbent assay (ELISA), which selectively identifies dsRNA species that are at least ~40 bp in length, as described previously by Nelson et al.6 K1 mouse monoclonal antibodies (SCICONS) were immobilized on 96-well Nunc Immuno plates (Thermo Fisher Scientific) for 3 h and blocked with 10% nonfat dry milk in PBS overnight at 4 °C. mRNA samples were added to the plates and incubated for 2 h at room temperature, washed and exposed to K2 mouse monoclonal antibody solution (SCICONS) at room temperature for 1 h. Following the incubation steps, plates were exposed to horseradish peroxidase-conjugated goat anti-mouse immunoglobulin G detection antibody (Thermo Fisher Scientific) for 1 h following a final wash. Signal detection was performed at 450 nm on a Synergy H1 plate reader (BioTek). Determination of dsRNA concentration in total RNA was calculated using a standard curve generated with 400-bp dsRNA prepared using 100% N1-methyl-pseudouridine chemistry.

IFN-β response in BJ fibroblasts

IFN-β response in BJ fibroblasts is used to provide a qualitative measurement of innate immune response. As previously described by Nelson et al.6, BJ fibroblasts acquired from the American Type Culture Collection (ATCC) were cultured in complete media comprising Eagle’s minimal essential medium with l-glutamine (ATCC) supplemented with 10% heat-inactivated fetal bovine serum (HI-FBS) (Life Technologies). Cells were seeded in 96-well cell culture plates (Corning Inc.) at 20,000 cells per well and maintained for 24 h before transfection. Cells were transfected with mRNA (250 ng per well) or poly(I:C) (10 ng μl−1; InvivoGen) using Lipofectamine 2000 (Thermo Fisher Scientific). At 48 h after transfection, supernatants were harvested and analyzed, and IFN-β protein levels in culture supernatants and mouse sera were determined via Ella microfluidic ELISA (ProteinSimple) per the manufacturer’s protocol.

Culture of human MDMs

As previously described by Nelson et al.6, frozen aliquots of human monocytes were thawed and suspended in complete media comprising Roswell Park Memorial Institute (RPMI) 1640 growth medium (Life Technologies) supplemented with 10% HI-FBS (Life Technologies) and human recombinant macrophage colony-stimulating factor (M-CSF; 20 ng ml−1; Invitrogen). The resuspended cells were seeded in 96-well flat-bottomed culture plates (Corning) at 150,000 cells per well, followed by incubation for 4 days to facilitate macrophage differentiation. At 24 h before transfection, the growth medium was replaced with fresh RPMI 1640 supplemented with 10% HI-FBS. Thereafter, MDMs were transfected with mRNA (250 ng per well). Following 5 h of incubation, the supernatant was discarded, and the cells were lysed with branched DNA lysis buffer.

IP-10 branched DNA assay

As previously described by Nelson et al.6, all reagents used for the IP-10 branched DNA assay formed part of the QuantiGene Singleplex Assay Kit (Thermo Fisher Scientific). A working probe set was prepared, including a capture extender, label extender, blocking probe and IP-10 probe per the instructions, and added to the capture plate. The prepared MDM cell lysates were incubated with the working probe solution for 16 to 22 h at 55 °C to facilitate hybridization. Reaction amplification was performed by the addition of a preamplifier, amplifier and label probe, with intermittent incubation and wash steps. The signal was detected on a Synergy H1 luminometer (BioTek) via chemiluminescent substrate.

In vivo mouse models

All animal procedures and experiments were approved by the Institutional Animal Care and Use Committee at Moderna and conducted in accordance to ARRIVE guidelines. Female C57BL/6 mice (n = 4) of ~8 weeks of age were obtained from Charles River Laboratories. Mice were injected intravenously via the tail vein with lipid nanoparticle formulations containing hEPO mRNA at 0.5 mg kg−1 and euthanized 6 h after injection. Blood was collected by cardiac puncture after euthanasia and processed for serum.

Mouse ProcartaPlex immunoassay

Cytokine levels in mouse sera were evaluated per the manufacturer’s instructions using a bead-based ProcartaPlex immunoassay kit (Thermo Fisher Scientific) consisting of granulocyte CSF, IFN-α, IFN-γ, IL-12p70, IL-6, CXCL10, CCL2, CCL4, CCL7 and TNFα.

hEPO ELISA

hEPO expression levels were measured from serum collected 6 h after dosing in mice by an Ella microfluidic ELISA (ProteinSimple) per the manufacturer’s recommendations.

Reporting summary

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

留言 (0)

沒有登入
gif