Microfluidic devices for scaling nanoparticle production

Manufacturing of RNA lipid nanoparticles (LNP), such as those used in COVID-19 vaccines, involves mixing RNA in an aqueous buffer with lipids dissolved in ethanol to trigger LNP self-assembly. The mixing conditions of RNA and lipids are crucial, as they ultimately affect product quality.

During fluid mixing, fluid flow occurs on a continuum from turbulent to laminar conditions. At the molecular level, turbulent conditions are highly random and difficult to control, like mixing tea with a spoon, for example. The random nature of turbulent mixing may not matter in the kitchen, but sophisticated and innovative drug delivery nanoparticles, subject to regulatory scrutiny, require sophisticated production controls.

The microfluidic mixing process involves complex intermolecular interactions between molecules of solvent, buffer, and at least five chemical types that affect the resulting particles’ physical characteristics. These characteristics, such as particle size, affect the molecules’ pathway in the body and the resulting immune response. Therefore, well-defined particle characteristics are critical to ensuring the safety and efficacy of drug products.

Microfluidic mixing provides access to non-turbulent conditions that enable rapid, controlled, and reproducible mixing of fluids in milliseconds for the continuous self-assembly of drug-delivery nanoparticles. Fluids are manipulated at the micro-scale and often flow through channels of different lengths and geometries.

The following advantages of microfluidic mixing have made it a go-to technology for the development of RNA vaccines [1], and other drug delivery technologies:

• Reduced reagent consumption
• Controlled non-turbulent mixing
• Tunable process parameters to optimize particle characteristics
• Small footprint to minimize cleanroom space requirements
• Batch-to-batch reproducibility

The history and evolution of nanoparticle manufacturing for therapeutic delivery

Both process and chemistry influence nanoparticle fabrication. Early lipid-based nanoparticles for small molecule drug delivery, such as those used for the liposomal cancer drug doxorubicin, were created through a complex multi-step workflow. This process included thin-film hydration followed by high-pressure homogenization and sonication or extrusion through a nanoporous membrane. The resulting liposomes consisted of an aqueous compartment within a lipid bilayer.

Liposomes have been used historically to deliver gene therapies in preclinical research. Initially, these molecules were made by first forming liposomes containing cationic lipids and then mixing them with nucleic acids to promote electrostatic complexation. After discovering that cationic lipids are cytotoxic, researchers began using pH-sensitive ionizable lipids that are uncharged at physiological pH to improve tolerability.

At the same time, a second-generation bottom-up self-assembly method was developed, in which lipids dissolved in ethanol were combined with nucleic acids dissolved in an aqueous buffer. Particle formation and nucleic acid complexation were accomplished in a single step, simplifying production and allowing for a more scalable continuous flow format.

By encapsulating the nucleic acid inside a condensed-core particle, the resulting lipid nanoparticles differed in structure from liposomes [2]. LNPs were discovered to outperform first-generation liposomes [3], however, this method is dependent on the creation of turbulence, which is accomplished by combining streams of aqueous and ethanol reagents at high speeds, such as in a T-junction mixer. Random molecular collisions govern mixing, which necessitates large volumes and high flow rates. Therefore, scaling the process up or down is challenging and requires considerable process redevelopment.

Benefits of microfluidics for nanoparticle production over traditional methods

A third-generation technique for nanoparticle production includes a microfluidic staggered herringbone micromixer (SHM) mixer, which maintains the advantages of continuous flow manufacturing and additionally allows greater control over the mixing environment. Non-turbulent fluid mixing ensures consistent conditions for each volume of liquid passing through the mixer, imparting reproducibility within a batch and between batches.

In addition, process parameters, such as the flow rate ratio (FRR) and total flow rate (TFR) [4], can be tuned to dial in the physicochemical characteristics of the produced particles, influencing the performance of LNP drug products. There’s a range of microfluidic channel geometries and architectures that have been employed in nanomedicine production, including the following:

• NanoAssemblr™ classic mixer: This design produces size-controlled nanoparticles by altering flow rate and total flow rate ratios. The classic mixer has been used extensively in preclinical development, with citations in >500 peer-reviewed publications. However, it has some limitations. This mixer needs consistent fabricating of herringbone structures at the microscale, which leads to complicated and expensive processes [5]. The original staggered herringbone micromixer mixers described by Belliveau et al. were designed to operate optimally at 5 mL/min, already an order of magnitude faster than other microfluidic options [6]. The SHM design was improved with the classic mixers that operate up to 20 mL/min, the highest known throughput that can be achieved in an SHM while maintaining flow characteristics. This capacity is enough to encapsulate ~ 850 mg (gross) of mRNA ― enough for over 28 000 vaccine doses ― in a 4 h run. The primary bottleneck in the process is downstream tangential flow filtration. Still, if desired, scale-out by parallelization can be employed to achieve higher throughput as the application demands.

• NanoAssemblr™ NxGen™ mixer: This new mixer was designed to simplify single-mixer scale-up. By using circular structures within the flow path, it offers comparable mixing efficiencies under non-turbulent conditions as well as a higher single-mixer flow rate. NxGen™ technology produces particles with the same critical quality attributes as the classic mixer [5] while enabling single-mixer flow rates from 1 to 200 mL/min, which is higher than flow rates frequently reported for T-junction mixers [7]. At 200 mL/min, a single NxGen™ mixer can nominally encapsulate 8.5 g of mRNA (> 283 000 vaccine doses) in a 4 h run ― 10 times the capacity of the classic mixer. For additional perspective, Pfizer-BioNTech’s phase III study for their COVID-19 vaccine required 44 000 doses ― or ~1.32 g of encapsulated mRNA [8] *. The wide range of flow rates available with NxGen™ technology [9] is well-suited for preclinical development and translating to clinical studies without changing technologies, which minimizes process redevelopment and risk.

Table 1. Comparison of the NanoAssemblr™ classic mixer (SHM), NxGen™ mixer (toroidal mixer (TrM)), and T-junction mixer. †Based on typical mRNA formulation parameters and vaccine dose of 30 µg mRNA per dose. Does not account for process yield.

Non-turbulent mixing offers advantages over traditional formulation methods, with the physics happening at the micro-scales helping to promote:

• Uniform-sized nanoparticles: With non-turbulent mixing, nanoparticles can be synthesized with high uniformity and a suitable size for nano applications (around 50 to 300 nm), making it possible to control the flow and mixing conditions [11].

• Better drug-loading efficiencies and storage: Although hydrophobic drugs can be entrapped in liposomes, the encapsulation efficiency is typically low because of entrapment at the bilayer interface. In this regard, Kastner et al [12] created liposomes using a chaotic advection micromixer device to solubilize propofol, a poorly water-soluble drug.
• The result were liposomes that:
  • encapsulated significantly more drugs than sonicated liposomes
  • demonstrated good stability and remained unaffected after eight weeks of storage at 4°C and 25°C

• Promising delivery results for clinical applications: Microfluidics protect nucleic acids to help maintain their stability. In a study using microfluidic-based core-shell nanoparticles, researchers created a novel lipid/polymer hybrid nano assembly composed of small interfering RNA (siRNA) complexed in the inner hydrophilic core of reverse poly ɛ-caprolactone-polyethyleneimine (PCL-PEI) micelles before coating a neutral lipid membrane [11]. Compared to bulk-mixed lipid/micelle/siRNA nanoparticles, the core-shell nanostructure produced by microfluidics provided more robust protection of siRNA locked in the core and better stability in circulation. Furthermore, these nanoparticles inhibited tumor growth by significantly downregulating epidermal growth factor receptor (EGFR) mRNA and protein expression levels in vitro and in vivo.

• Translation of liposomes from the bench to the clinic: With the aim of scalability, Forbes et al [5] used a NanoAssemblr™ classic mixer to produce liposomes while incorporating insulin, bovine serum albumin (BSA), or ovalbumin (OVA) along with SHM device in-line purification and in-line particle size monitoring for in-process control.

• Key results:
  • The liposome formulations offered a high protein loading (20% to 35%) compared to sonication or extrusion methods (<5%).
  • Optimized production led to the generation of highly loaded liposomes with size control (60 to 100 nm) and a polydispersity index (PDI) < 0.2.

Scaling nanoparticle production for clinical or industrial use

The widespread clinical use of RNA-based therapeutics will require moving beyond the laboratory to large-scale manufacturing. Microfluidic technology allows production to be scaled from batch sizes suitable for pre-clinical studies to clinical and commercial production using the same process with minimal redevelopment.

As detailed above, a single NanoAssemblr™ classic mixer operating at 20 mL/min can produce 28 000 vaccine doses in a 4 h shift. For larger batches, manufacturers can scale out by employing multiple identical microfluidic mixers in parallel. Additionally, NxGen™ technology allows facile scale-up, where mixer dimensions are enlarged to drive flow rates through a single mixer that is an order of magnitude higher while preserving the underlying physics of mixing and promoting well-controlled, reproducible conditions. Therefore, NxGen™ technology enables the scale-up of nanoparticle production from pre-clinical to industrial scales using the same technology, therefore minimizing process redevelopment and mitigating risk.

NxGen™ technology offers scale-independent production because the systems can be run from a small laboratory-scale to continuous production via scale-up rather than scale-out. In a study by Roces et al. [4], GenVoy-ILM™ lipid mix composition, which contains an ionizable lipid similar to MC3 with an apparent pKa value of 6.0, was chosen. The NanoAssemblr™ classic mixer and NxGen™ technology were then used to generate blank and polyadenylic acid (PolyA)-loaded GenVoy-ILM™ ionizable LNP at 12 (classic system and NxGen™ technologies), 60 (NxGen™ technology only), and 200 mL/min TFR (NxGen™ technology and GMP system). The blank formulation was significantly smaller in size (55 nm) than the PolyA-loaded counterpart (78 nm) (p < 0.05). Both LNP had comparable sizes at all speeds tested. In addition, the nanoparticles showed equivalent sizes with high polyA encapsulation efficiencies (larger than 95%) [4].

The new NxGen™ microfluidic mixer design allows seamless scale-up production from bench-scale (12 mL/min) to GMP production requirements of over 20 L/h [5]. Furthermore, with tangential flow filtration, it is possible to achieve scalable downstream processing to support the microfluidic production of nanomedicines with a high yield. Finally, the study confirmed the manufacturing possibility of the nanoparticles quickly and reproducibly using a scale-independent manufacturing process, reducing risk when moving from bench production to manufacturing approved commercial products.

Microfluidic platforms with high throughput and flow rate capacity, such as the NanoAssemblr™ systems [4], are already available on the market. The NanoAssemblr™ manufacturing platform allows the incorporation of nucleic acid and lipid formulations for mRNA drugs into one end of a device small enough to fit on a lab workbench.

LNP made with the NanoAssemblr™ platform are uniform and homogeneous in structure. Furthermore, because NxGen™ technology is scalable from lab production to commercial batch sizes with the same mixer design, LNP production volumes can be easily increased, saving several months on the development time of drug candidates in the pipeline, leading to cost savings. This workflow dramatically simplifies mRNA-LNP development, allowing researchers with a deep understanding of disease to explore the clinical potential of genomic drugs.

* 22 000 subjects x 2 doses x 30 ug dose = 1.32 g

‡Any other third-party trademarks are the property of their respective owners.