Process Robustness in Lipid Nanoparticle Production: A Comparison of Microfluidic and Turbulent Jet Mixing

The recent clinical and commercial success of lipid nanoparticles (LNPs) for nucleic acid delivery has incentivized the development of new technologies to manufacture LNPs. As new technologies emerge, researchers must determine which technologies to assess and how to perform comparative evaluations. In this article, we used a quality-by-design approach to systematically investigate how mixing technology influences the physicochemical and structural properties of LNPs. Specifically, a coaxial jet in co-flow turbulent mixer was systematically compared to a staggered herringbone microfluidic mixer via matched formulation and process conditions. A full-factorial design-of-experiments study with three factors and three levels was executed for each mixer to compare process robustness in the production of antisense oligonucleotide LNPs. LNPs generated with the coaxial turbulent jet mixer were consistently smaller, more uniform, and had greater oligonucleotide encapsulation as compared to the microfluidic mixer, but had a greater variation in internal structure and less ordered cores. A subset of the study was replicated for mRNA – LNPs, with comparable trends in particle size and encapsulation, but more frequent bleb features for LNPs produced by the coaxial turbulent jet mixer. The study design used here provides a roadmap for how researchers may compare different mixing technologies (or process changes more broadly), and how such studies can inform process robustness and manufacturing control strategies.

SAXS data were collected in the high throughput mode (HT-SAXS) using the Advanced Light Source SIBYLS beamline 12.3.1 at the Lawrence Berkeley National Laboratory (Berkeley, CA). The X-ray wavelength was set at λ = 1.216 Å, and the sample-to-detector distance was 2070 mm, resulting in a scattering vector, q, ranging from 0.01 Å–1 to 0.45 Å–1. The scattering vector is defined as q = 4π sin θ/λ, where 2θ is the scattering angle. Experiments were performed at 20 °C. Briefly, the sample was exposed for 10 s with the detector framing at 0.3 s to maximize the signal while merging the SAXS signal using the SAXS FrameSlice application (https://bl1231.als.lbl.gov/ran). No radiation damage was observed during the 10 s exposure, and all of the collected frames were merged. The merged SAXS profile was further processed using BioXTAS RAW (https://bioxtas-raw.readthedocs.io/en/latest/index.html) and OriginPro 2022b from OriginLab Corporation (Northampton, MA). The SAXS scattering profiles were set to a consistent baseline to account for scattering profiles with intensities that decreased dramatically at high q. The profiles were then processed using Batch Peak Analysis on OriginPro to identify the position of the first observable peak. The baseline for the peak search was determined by a straight line that connects the data at q = 0.05 and 0.3. Subsequently, the baseline was subtracted, the data was rescaled and smoothed using the Savitzky–Golay method with a polynomial order of 2 and points of window of 20, and a peak search algorithm was applied to identify the q position of the q0 peak via the first derivative of the data and height threshold of 60%. Our study restricted analysis of the SAXS data to quantification of a single peak position. Literature reports show that multiple types of order may be present in samples, which can lead to broad or asymmetric peaks that may not be fully described by a single peak position. Deeper analysis may provide additional insight. Twenty-seven SAXS measurements were made for each mixer, one for each combination of FRR, N/P ratio, and MC3 content, for a total of 54 measurements. All SAXS measurements were performed on samples after dilution, concentration, and solution exchange. Samples were analyzed after storage at 2–8 °C for up to 1 week but most commonly within 2 days. Samples were loaded into a 96-well plate without further manipulation for measurement.

All data in the ZIP file contain final merged SAXS scattering profiles from lipid nanoparticles loaded with antisense oligonucleotides (ASO-LNPs). Total lipid concentration of 10 mM. Three variables were involved: flow rate ratio (FRR), N/P ratio, and MC3 lipid composition ratio. Formulation flow rate was varied to achieve FRR of 3:1, 4.5:1, or 6:1. N/P ratios were varied from 1, 2, or 6. Lipid compositions were varied to be 32.7:13.5:51.8:2.1, 50:10:38.5:1.5, or 59.3:8.2:31.3:1.2 MC3/DSPC/Cholesterol/DMG-PEG2000. These combinations are indicated in the file names as "111" or "112", where "111" = (-1,-1,-1) or lowest parameter values for all three categories, and "333" = (1,1,1) or highest parameter values for all three categories, in the order of FRR, N/P, and MC3.

2023-06-27

2023-11-03

HT-SAXS

Advanced Light Source

SIBYLS BL12.3.1

1.216 Å

2.07 m

Sample:

• Macromolecule 1: ASO-LNP
• Sample Full Name: Lipid nanoparticles loaded with antisense oligonucleotides
• Sample Type: Lipid Nanoparticle
• Source Organism:
• Source Organism NCBI Taxonomy ID:
• Expression System:
• Expression NCBI Taxonomy ID:
• Uniprot ID:
• Sequence or Chemical Formula: