With lipid nanoparticles (LNPs) gaining more and more popularity in research and therapy, microfluidic high-throughput production platforms have received considerable attention. Microfluidic devices enable a scalable, precise and reproducible production of LNPs or liposomes compared to bulk production (Shepherd et al., 2021). Several key factors have to be considered during the design of a device suitable for LNP production (Roces et al., 2020). Due to the randomness of LNP formation, bulk production is not suitable. This is mirrored by the fact that bulk production experiments show low reproducibility, large LNPs (> 100 nm) and a poor loading efficiency (Shepherd et al., 2021). Like that, microfluidics promise to be a great alternative, as it relies on very small diameter channels in so called microfluidics devices to enhance the LNP formation rates. An aqueous phase and a lipid phase flow through said channels and using the turbulence created inside, LNPs form at a much higher rate and more consistent size, than when using bulk production methods.
A problem with microfluidic approaches is the existence of laminar flow, which makes the mixing of the two phases only possible through diffusion at the interfacial area. To overcome this problem micromixers have been developed. These can be classified as either active or passive micromixers. Active micromixers need external force (e.g. pressure, temperature etc.) to create turbulence which enhances LNP formation. The creation of these structures in microfluidics is often complicated and time consuming. Contrary to that, passive micromixers are built by the creation of structures inside the microfluidic devices which cause turbulence formation (Nady et al., 2021). One specific structure, the staggered herringbone micromixer (SHM), has proven to be a very effective way of LNP production and has been further refined in the last years (Nady et al., 2021; Roces et al., 2020; Shepherd et al., 2021; Whulanza et al., 2018). Next to the SHM, other passive micromixers were also proposed by Shepherd et al. (2021), including micro ring mixers.
Polydimethylsiloxane (PDMS) has proven to be one of the most widespread used materials for realisation of such devices due to its low cost, durability and compatibility with the used reagents (Nady et al., 2021; Shepherd et al., 2021; Torino et al., 2018; Whulanza et al., 2018).
Our aim was to develop different types of PDMS based microfluidic designs, including a single channel SHM device, for LNP production and to compare them. In the end the best design was supposed to replace the batch production of liposomes in the wetlab. The produced LNPs needed to be smaller than 100 nm in diameter, as that would ensure a good transfection efficiency. Furthermore the desired mixing capability of the device needed to be at least 90 %. For the analysis and comparison of LNPs an eGFP approach could be used. For this, a light source in the range of 470 ± 20 nm would also be necessary, as well as a photo sensor to measure the light emission from the LNPs and thus, their size.
All needed components would be fixed on a Polymethylmethacrylate plate. Afterwards they could be used for LNP production and the sensor could be connected to the Arduino chip in order to read the eGFP intensities and LNP sizes.
For this purpose, we designed different prototypes of passive micromixers, which can be seen in Figures 1 and 2. To do so, Autodesks Fusion360 (Version 2.0.14113) was used. The dimensions of our devices were adapted from Zhigaltsev et al. (2012), featuring 200 μm wide and 79 μm high mixing channels, in the case of a SHM with herringbone structures formed by 31 μm high and 50 μm thick features on the roof of the channel (Figure 2). Additionally, a micro ring mixer device was designed. It was adapted from Ripoll et al. (2022) with 150 µm wide and 80µm high channels (Figure 1).
In order to build such a device, there are several steps that need to be taken care of. In the first step a glass petri dish is coated with a UV sensitive photoresist by spin-coating. A stencil is then hovered close to the surface and the petri dish is subjected to UV radiation. This leads to all of the photoresist directly hit by the radiation to harden, while the parts that are covered by the stencil remain removeable. After being dipped into a developing solution and a baking step the remaining non-hardened photoresist is washed off. This leaves the channels of the device as a three dimensional coat on the petri dish. This stage is also referred to as the stencil, as it can be reused multiple times in order to create the same microfluidic device (Scott et al., 2021).
Our stencil for a micro ring mixer design was made for us by Dr. Sadaf Pashapour, an on-campus scientist specialised in microfluidics. Production of a SHM stencil was outside of our capacities, as it was not possible to create a multi-thickness photoresist layer on the same chip with the equipment provided.
In our case the device would have been made out of PDMS as discussed before. In order to create such a device PDMS is poured over the stencil and a thermal annealing process creates a firm seal between PDMS and the surface. After hardening the device is cut out and the inlets and outlets are fabricated using a hole puncher. The cutout is then activated using a O2-Laser and fixed on a glass slide using pressure bonding (Scott et al., 2021). Due to a tight schedule and problems during the acquisition of PDMS we were not able to produce microfluidic chips and test them.
Regarding the automation of LNP production using microfluidics, we created an early stage design for a tabletop device, which can be used with self made microfluidics chips. It features a rubber base to securely place the microfluidic chip onto. The materials needed for the LNP production can be filled into falcon tubes and placed into the boreholes, as well as a falcon tube for the final LNP solution. Three peristaltic pumps are used to move the liquids through the microfluidic chip. They are supposed to be controlled by an Arduino board via the included software. This way the flow rate can be controlled. As a safety feature this board only starts the production with the lid closed, to prevent contamination of the LNPs as well as spillage onto any body parts of the user as a result of insufficiently fitted tubing. An additional photo-sensor (not included in the render) is able to register the size and amount of LNPs produced by the implemented eGFP fluorescence. Using this information the flowrate can be corrected in order to ensure a certain LNP size, even if initially the flow rate was set too high or too low. The maintenance cabinet below the main work area is used to store a waste container and cleaning solution which can be used to clean the tubing and microfluidic chip in a retrograde fashion. This could enable the reuse of chips and maybe reduce the PDMS usage, as well as making the chips safer for disposal.
