On‐Chip De Novo Production of mRNA Vaccine in Lipid Nanoparticles

Abstract

RNA‐based therapeutics have revolutionized precision medicine due to their unprecedented potency, specificity, and adaptability. However, the inherent limited stability of RNA, including mRNA used in vaccines, is a major obstacle to the full realization of their potential. This instability, coupled with the centralized nature of vaccine production, currently limits the generation of RNA therapeutics at the point of care, which will otherwise fully harness the potential of these agents. Here, a microfluidic platform is presented for on‐demand, personalized synthesis of modified mRNA stabilized by lipid nanoparticles. The design includes trapped biotinylated DNA, tagged T7 RNA polymerase, and a Tesla mixer, allowing the on‐chip synthesis, purification, and encapsulation of mRNA in uniform lipid nanoparticles (LNPs), all conducted seamlessly on the same microfluidic device. This on‐chip microfluidic synthesis approach is found to match standardized mRNA production yields, yet surpasses typical purification methods. Furthermore, as a proof‐of‐concept, the versatility and efficacy of the platform are demonstrated by generating diverse RNA sequences and structures, exhibiting functionality in human cell lines and mouse models. Moreover, an active SARS‐CoV‐2 vaccine is successfully engineered, highlighting the platform's potential for personalized vaccination strategies and offering a promising avenue for high throughput, decentralized vaccine delivery, reduced cold chain dependence, and even advancing current personalized medicine approaches through custom RNA therapeutics.

A microfluidic platform integrates mRNA synthesis, purification, and LNPs formulation in a continuous, streamlined process. DNA templates are transcribed on‐chip, producing mRNA that is immediately purified and encapsulated into LNPs. This approach enables rapid, scalable production of mRNA‐LNP formulations for research or therapeutic applications.

1. Introduction

RNA‐based therapeutics recently emerged at the forefront of precision medicine with unprecedented specificity and potency.^{[ 1} , ² , ³ , ⁴ , ^{5 ]} Nevertheless, RNA molecules are notoriously unstable and prone to biochemical and physical degradation. The global COVID‐19 pandemic has underscored both the potential and the limitations of mRNA‐based vaccines, particularly when there is a requirement for bulk and rapid administration of vaccines.^{[ 6} , ⁷ , ⁸ , ⁹ , ^{10 ]} Achieving widespread vaccine coverage requires addressing distribution challenges and implementing strategies to significantly increase the manufacturing capacity of vaccines on a global scale.^{[ 11 ]} Predictably, the lowest vaccination rates are observed in low‐income countries, prompting a historic global initiative named COVAX. This initiative aims to ensure fair and equitable access to COVID‐19 vaccines, acknowledging the need for a coordinated and inclusive approach to address global immunization challenges.^{[ 12 ]}

A range of approaches have recently been explored to enhance the stability of mRNA encapsulated in LNPs. These strategies include increasing mRNA stability in vivo by using self‐amplifying RNA or circular RNA and increasing formulation stability by lyophilizing the entire mRNA‐LNP composite.^{[ 13} , ¹⁴ , ¹⁵ , ^{16 ]} Each technique aims to mitigate the challenges associated with mRNA degradation, contributing to the overall improvement of vaccine stability and efficacy.^{[ 17} , ^{18 ]}

A complementary strategy is decentralizing vaccine manufacturing, striving for on‐site and on‐demand production. In June 2022, BioNTech initiated the construction of an mRNA manufacturing facility in Rwanda, Africa, marking a notable step forward. Africa, the second most populated continent, currently imports over 95% of its medicines and vaccines. Establishing a local manufacturing facility aims to address this dependence, fostering greater self‐sufficiency in vaccine production within the region.^{[ 19} , ²⁰ , ²¹ , ²² , ^{23 ]}

To address these challenges, our work focuses on developing a modular, high‐throughput microfluidic‐based platform for on‐site and on‐demand generation of LNPs containing specific mRNA. Microfluidic‐based platforms have proven to be particularly useful for the generation of nano‐compartment, LNPs and for the spatiotemporal separation of biochemical reactions.^{[ 24} , ²⁵ , ²⁶ , ²⁷ , ²⁸ , ²⁹ , ^{30 ]} Thus, we engineered compartmentalized biochemical reactions within different sections of the microfluidic platform to streamline the encapsulation of resulting mRNA, while allowing for high flexibility in the sequences synthesized. Along with demonstrating the product yield exceeds that of bulk techniques, we characterize the stability of the resulting nanoemulsions and their capacity to act as an active vaccine in vivo. The presented platform overcomes the limitations associated with current methods, offering an efficient and adaptable solution for the production of mRNA‐containing LNPs on‐site.

2. Design and Characterization of On‐Chip RNA Synthesis and Encapsulation

The standard in vitro transcription (IVT) reaction, widely employed in manufacturing mRNA and its subsequent applications, involves a linear DNA template, RNA polymerase, and ribonucleotide triphosphates (NTPs). Following transcription, mRNA purification is carried out using bead‐based, affinity, or other chromatography and filtration techniques.^{[ 31} , ³² , ³³ , ^{34 ]} In forming LNPs, purified mRNA is combined with lipids through various microfluidic‐based methods.^{[ 35} , ³⁶ , ³⁷ , ³⁸ , ^{39 ]} While considered state‐of‐the‐art, this approach currently holds several limitations, such as the lack of flexibility and adaptability, and the use of multiple open systems in sequence, making it more susceptible to nuclease contaminations. Another limitation is the inherent scalability bottleneck associated with bulk‐scale techniques. Large‐scale production facilities require substantial funding to set up and maintain, limiting the accessibility of mRNA therapeutics, particularly in resource‐limited settings.^{[ 40 ]} The complex logistics of transporting and storing these temperature‐sensitive therapies pose a significant challenge.^{[ 41 ]}

In addressing challenges associated with existing methods, we aimed to develop a comprehensive and generalized platform that seamlessly integrates the synthesis, purification, and encapsulation of mRNA in line. This platform maintains the necessary versatility to accommodate diverse coding sequences (CDS) through standard cloning methods. Thus, we first devised and optimized a protocol for generating the linearized DNA template, starting from the 5′ end, which includes a T7 promoter (T7P), CDS, and T7 terminator (T7T). The template is then amplified via polymerase chain reaction (PCR) using universal biotinylated primers complementary to the T7P and T7T (Figure 1a) (Experimental Section). The system's modularity is highlighted by the ability to substitute the CDS with standard cloning methods while still employing the same biotinylated primers, allowing DNA immobilization following the transcription reaction (Figure S1, Supporting Information).

Figure 1.

Design and characterization of on‐chip RNA synthesis. a) Schematics of universal template preparation. The CDS is located downstream of the T7P sequence and upstream of a T7T sequence. The template is then amplified using biotin‐tagged primers complementary to the T7P and T7T sequences. Following the PCR purification step, the DNA template is tagged with biotin both at the 5′ end and at the 3′ end. b) Design of the microfluidic device with the T junction motif for droplets encapsulating RNA. c) Illustration of the on‐chip IVT reaction, RNA purification, and encapsulation. i) SA and Ni‐NTA beads are trapped in the reaction compartment. Biotinylated PCR product is introduced to the device alongside HIS‐tagged T7RP and IVT reaction components. Following incubation at 37 °C, the IVT reaction components are allowed to flow through the device. The DNA template and HIS‐tagged T7RP are then immobilized onto the beads. ii, iii) brightfield images of the beads trapped in the compartment. The purified mRNA is allowed to flow toward the T junction motif, whereas iv) the oil phase intersects the aqueous mRNA phase, resulting in the formation of water‐in‐oil droplets. vi) Fluorescence microscopy image of DNA template tagged with Cy3, immobilized onto SA beads in the main compartment. Scale bar 200 µm. vii) Fluorescence microscopy image of the on‐chip produced droplets encapsulating Mango RNA aptamer. Scale bar 200 µm. d) Gel electrophoresis of RBD‐hFc mRNA and SARS‐CoV‐2 full spike protein mRNA produced on‐chip. e) On‐chip or column‐based purification yields of both non‐modified and modified SARS‐CoV‐2 full spike protein mRNA.  ^∗ p < 0.05, ^∗∗ p < 0.01 (two‐tailed Student's t test, n  = 3). f) On‐chip generated mRNA stability at different pH and with RNAse for up to 24 h.

After optimizing amplification, we developed a streamlined platform integrating on‐chip in vitro transcription (IVT), RNA purification, and lipid droplet encapsulation. This unified microfluidic system enables controlled conditions and minimal manual handling, with potential applications in RNA production, RNA‐protein interaction studies, and RNA‐based delivery systems. (Figure 1b,c). We thus incorporated size‐excluding pillars in the first part of the device to trap streptavidin (SA) and Nickle Nitrilotriacetic acid (Ni‐NTA) affinity beads to which the specific biotinylated DNA and HIS‐tagged T7 RNA polymerase (T7RP) could be introduced and bound (Figure 1c; Figure S1, Supporting Information). This microfluidic platform allowed for trapping the SA and Ni‐NTA beads mixture in a single compartment within the microfluidic device, followed by adding IVT reaction components. These include the biotinylated DNA template and HIS tagged T7RP, and allowed the reaction to initiate. Following 2–4 h of incubation, the reaction mixture was washed and allowed to flow through the device, while both the biotinylated DNA input and T7RP were retained within the compartment (Figure 1c). Western blot analysis of the eluted volume revealed that the T7RP was successfully retained in the compartment (Figure S2, Supporting Information). The RNA produced in the solution passed the beads compartment, thus completing the mRNA on‐chip production and purification steps (Figure 1c).

As part of the microfluidic device design, downstream to the transcription and RNA purification compartment, we introduced a microfluidic T junction motif, allowing the generation of microdroplets in a high‐throughput manner, where the RNA‐containing aqueous solutions are encapsulated as emulsions by an immiscible oil phase. As a proof‐of‐concept, we demonstrate the transcription of the light‐up RNA aptamer, Mango.^{[ 42} , ^{43 ]} The aptamer DNA sequence was bound to the beads and allowed the production of an RNA aptamer in solution, which was able to fold to a correct secondary structure and bind to its ligand that was supplemented to the IVT reaction. Following their production, a fluorescent signal could be detected inside the droplets using a fluorescence microscope (Figure 1b,c; Figure S3, Supporting Information).

The inclusion of mRNA modifications, such as 5′ capping and pseudouridine, is key to the successful expression of mRNA‐based therapies.^{[ 44} , ⁴⁵ , ^{46 ]} Our subsequent objective was to investigate the system's capability to produce modified mRNAs of varying lengths and to assess the synthesis and purification yields. Employing a similar approach to that outlined for DNA template generation, we introduced a DNA template encoding varying mRNA lengths (Table S1, Supporting Information), including their 5′ untranslated region (UTR) and 3′ UTR, such as the hFc‐conjugated receptor‐binding domain (RBF‐hFc) (1705 nt), and SARS‐CoV‐2 spike mRNA (3888 nt) to an array of devices. The output of purified modified mRNA products was monitored using electrophoresis gel and showed a comparable yield to that achieved by conventional bulk approaches (above 100ug per reaction), it should be noted that we used equal amounts of IVT reagents for both on‐chip mRNA synthesis and standard IVT (Figure 1d).

Furthermore, we measured and compared the purification yields of modified and non‐modified spike mRNA with a commercial purification kit (see Experimental Section). The electrophoresis gel run results demonstrate that the purification of both modified and non‐modified mRNA is significantly higher using the microfluidic platform (Figure 1e). Subsequently, we assessed the stability of the mRNA produced under two pH conditions (pH 4 and 8) and in the presence of RNAse for up to 24 h at room temperature (Figure 1f). The concentration of intact mRNA remained similar at both pH conditions and was rapidly degraded when incubated with the RNAse.

3. On‐Chip De Novo Generation of mRNA in LNPs

Encapsulating RNA therapeutics within LNPs offers many advantages, such as enhanced stability, efficient and targeted delivery, scalability, and versatility. A microfluidic‐based platform can enable the large‐scale production of RNA‐LNPs, enabling the creation of uniform RNA‐LNPs.

Recognizing the flexibility of microfluidic platforms, we decided to integrate the encapsulation of RNA in LNPs into our device design. Building on the initial design presented in Figure 1, we modified the introduction of the lipid solution (ionizable lipid (EA2 or EA‐405), DSPC, cholesterol, and DMG‐PEG (40:10.5:47.5:2 mol ratio)) to enhance the functionality of the device (Figure 2a,b). This design allows for the co‐flow of the aqueous mRNA and lipid solution, dissolved in EtOH, through the Tesla mixer, resulting in the production of LNPs encapsulating the aqueous phase via shear flow (Figure 2b,c; Figure S4, Supporting Information). To facilitate on‐chip turbulent mixing of the aqueous mRNA and lipid solutions, a Tesla‐mixer element was incorporated.^{[ 47} , ⁴⁸ , ⁴⁹ , ⁵⁰ , ^{51 ]} This integrated approach ensures the synthesis and encapsulation occur in line, offering the advantage of a streamlined process.

Figure 2.

On‐chip de novo production of mRNA in lipid nanoparticles. a) Fluorescence microscopy image of the microfluidic device with the Tesla mixer for LNP generation using a 0.01% (V/V) fluoresceine solution. Scale bar 5 mm. b) Fluorescence microscopy image of the intersection between the main compartment and the lipid mix using 0.01% (V/V) fluoresceine solution and EtOH. Scale bar 100 µm. c) Illustration of the on‐chip RNA‐LNP generation. i) mRNA is synthesized and purified as described in Figure 1c. ii) lipid mixture is introduced from a second inlet. iii) the mRNA aqueous solution encounters the lipid solution and is mixed in the Tesla mixer to form LNP (iv). d) DLS characterization of the on‐chip produces LNPs pre‐ and post‐cryopreservation treatment, including size distribution and polydispersity index (PDI) e) Particle size (back) and PDI (green) of on‐chip produced over time. f) A representative Cryo‐EM image of on‐chip produced LNPs. Scale bar 100 nm. g) Agarose electrophoresis gel of naked mRNA and on‐chip produced LNP with or without RNAse treatment for up to 24 h.

The mRNA encoding for the luciferase enzyme was synthesized following the previously outlined procedures. Following a 2 h incubation in the mRNA synthesis and purification compartment, the RNA was eluted with an acidic acetate buffer (Figure 2c, Experimental Section) and then flowed alongside the lipid solution at a 3:1 ratio, respectively. The resulting mRNA‐containing LNPs demonstrate a uniform size distribution, with a 99.2 nm hydrodynamic radius and low polydispersity indexes (PDIs), measured through dynamic light scattering (DLS), and are comparable to values measured when prepared using conventional IVT and LNP encapsulation (60–130 nm for hydrodynamic radius and lower than 0.3 PDI) (Figure 2d).^{[ 47} , ⁵² , ^{53 ]} To further characterize the LNP charge distribution, zeta potential measurements were used, showing a ζ value of 0.907, consistent with previously published data (Figure 2d).^{[ 54 ]} Modulating the aqueous‐to‐lipid flow rate ratio during microfluidic mixing resulted in LNPs with distinct and controllable size distributions and high mRNA encapsulation efficiency (Figure S5, Supporting Information). While initially intended for on‐site and on‐demand generation, we were intrigued to explore the characteristics of RNA‐LNPs subjected to cryogenic treatment. Following production, the RNA‐LNPs underwent cryogenic treatment, as outlined in the materials and methods, and were subsequently analyzed following a freeze‐thaw cycle. DLS analysis indicated that the cryogenic treatment had no discernible impact on the hydrodynamic radius or uniform distribution of the LNPs (Figure 2d). For comparison, we analyzed RNA‐LNPs prepared using a standard method with commercially purchased RNA and a state‐of‐the‐art LNPs formulation system (NanoAssemblr), which showed comparable size and uniformity, further confirming the stability of our platform (Figure S5, Supporting Information). Next, we aimed to evaluate the stability of the RNA‐LNPs at 4 °C over an extended time. We measured the hydrodynamic radius and the polydispersity index values at various time points over 28 days, and these values remained consistent, demonstrating high stability of the mRNA‐LNPs (Figure 2e). Cryogenic electron microscopy (Cryo‐EM) analysis aligned with the DLS data, illustrating small, uniform particles (Figure 2f).

Along with characterizing the stability of the formed LNPs at a range of physical conditions, we further explored their stability under biological conditions. Thus, LNPs' stability and resistance to nuclease‐mediated degradation were determined by their incubation in the presence of RNAse A for up to 24 h (Figure 2g). The mRNA‐LNPs exhibited high resistance to nucleases, as evidenced by the electrophoresis gel, in stark contrast to naked mRNA samples, which underwent rapid degradation.

4. On‐Chip Production of a Functional SARS‐CoV‐2 Vaccine

Following on‐chip‐produced mRNA‐LNPs characterization, we sought to assess their activity both in vitro and in vivo. mRNA luciferase‐encapsulating LNPs (Luc‐LNPs) were generated using the described system. HeLa cells were incubated with increasing concentrations of Luc‐LNPs, and luminescence was measured 24 h post‐transfection. A dose‐dependent response with heightened luminescence signals at higher Luc‐LNP concentrations could be determined (Figure 3a). Similar results were observed when transfecting cells with mRNA‐Luc after a freeze‐thaw cycle, indicating sustained activity of LNPs following cryopreservation (Figure 3a).

Figure 3.

On‐chip production of an effective SARS‐COV‐2 vaccine a) in vitro luciferase assay of increased concentrations of on‐chip produced LNPs per or post cryopreservation treatment. b,c) in vivo biodistribution study of the on‐chip produced LNPs encapsulating luciferase mRNA at the whole animal (b) and organ (c) levels. d) Western blot analysis of RBD‐hFc expression in HEK293FT cells treated with on‐chip produced LNPs encapsulating RBD‐hFc mRNA. e) igG titer of mice treated with LNPs produced on chip encapsulating luciferase (Luc) mRNA or the RBD‐hFc mRNA, pre‐ and post‐booster shots.

To further evaluate the on‐chip‐produced Luc‐LNPs’ performance in vivo, a biodistribution study was conducted. Luc‐LNPs were injected into wild‐type C57BL/6 mice via the tail vein, and mRNA expression was analyzed at both systemic and organ levels (Figure 3b,c). The administration of luciferase mRNA by the LNPs resulted in measurable in vivo luciferase activity using IVIS spectrum imaging. At 6 h post‐injection, the predominant luciferase signal was detected in the liver and spleen, representing the main organs known to accumulate LNPs and to express the encapsulated mRNA. This observed outcome aligns with expectations and underscores the biodistribution characteristics of the on‐chip‐produced LNPs in vivo.

Next, we investigated the feasibility of producing a functional vaccine using our microfluidic‐based manufacturing platform. We previously reported on a SARS‐CoV‐2 mRNA‐LNP vaccine encoding for the RBD‐hFc.^{[ 55} , ^{56 ]} We first generated RBD‐hFc‐LNPs using our platform and tested them in HeLa cells. Following 24 h of incubation, we detected RBD‐hFc protein in the culture medium and the intracellular fraction of treated cells (Figure 3d). Next, we evaluated the in vivo efficacy of these microfluidics‐generated RBD‐hFc‐LNPs in mice. BALB/c mice received RBD‐hFc‐LNPs intramuscular immunizations. The mice followed a prime‐boost vaccination regimen, where the mice received an initial immunization (priming) on day 0 and a follow‐up injection (boost) 25 days later (Figure 3e). To assess the antibody response, blood samples were collected from the two groups at two‐time points: at day 23, before the booster dose (pre‐boost), and at day 39, following the booster dose (post‐boost). Serum samples were then analyzed for the presence of antibodies against the SARS‐CoV‐2 spike protein. At the pre‐booster shot timepoint, anti‐spike antibody titers of 1 × 10²−1 × 10³ were recorded in vaccinated animals, compared to < 1 × 10¹ in the control group. Notably, following the booster vaccination, the RBD‐hFc‐LNP group developed a robust antibody response ranging between 1 × 10³ and 1 × 10⁵, while the control group remained unresponsive. The titer values from the mice treated with the on‐chip produced VLPs are comparable with the titer values of mice treated with RBF‐hFc LNPs produced using traditional methods.^{[ 55 ]} This suggests that the RBD‐hFc‐LNP vaccine triggered a spike‐specific immune response, potentially indicating its protective efficacy against the virus.

5. Discussion

The microfluidic platform we have developed offers a novel approach for on‐demand, point‐of‐care synthesis of LNPs containing customized mRNA sequences. This integrated system streamlines the mRNA encapsulation process by compartmentalizing biochemical reactions within the device while maintaining flexibility for incorporating diverse coding sequences. The platform achieves yields comparable to bulk‐scale techniques in a comparable time frame and surpasses typical purification methods. In addition, the microfluidic approach presented allows for testing large libraries of mRNA‐based vaccines, as it is low‐cost and time‐efficient. Amidst the rising global demand for RNA‐based therapeutics, considerable attention is dedicated to improving RNA stability and extending its shelf life. Nevertheless, we argue for an additional emphasis on decentralizing therapeutic production, facilitating small‐scale manufacturing directly in point‐of‐care settings. This innovative technology marks the inaugural step toward realizing this vision. While further characterization and optimization are necessary to ensure the platform's robustness and meet stringent quality standards for human use, this study represents a significant first step toward on‐site manufacturing of mRNA‐based therapies.

6. Experimental Section

Biochip Preparation—Biochip Design

Microfluidic devices were designed using AutoCAD software (Autodesk AutoCAD), based on the previous design.^{[ 24 ]} The initial device used for oil‐in‐water mRNA encapsulation consisted of a main compartment with four rows of pillars with dimensions of 25 µm × 25 µm to immobilize beads. The total channel length and width of the compartment were ≈85 and 1 mm, respectively. A microfluidic T‐junction was introduced following this compartment, with channel dimensions of 50 µm × 50 µm. The height of the entire device was set to 50 or 100 µm. Similarly, for LNP generation, the main compartment design was as above, yet a side channel with dimensions of 50 µm × 50 µm for lipid flow was introduced following the compartment. The device further included a Tesla‐ Mixer element with an overall channel width of 95 µm, constrictions of 25 µm with an overall length of 34.5 mm consisting of 63 repeats. The height of the entire device was set to 50 or 100 µm.

Biochip Preparation—Biochip Fabrication

The microfluidic biochips were fabricated from polydimethylsiloxane (PDMS, Dow Corning) using SU8 on silicon masters and standard soft lithography techniques. Inlets and outlets were punched and PDMS was then plasma bonded to glass slides to create a sealed biochip.

On‐Chip Production and Purification of RNA and mRNA

Linear DNA fragments were produced by polymerase chain reaction (PCR) using Phusion High‐Fidelity PCR Master Mix (NEB). For biotinylated products, PCR was carried out with modified primers with biotin attached at the 5′‐end (IDT). The forward biotin‐conjugated primer was designed as the T7P sequence with an initiating sequence of 5′ AG 3′ for enabling co‐transcriptional capping. The reverse primer was designed as the 3′ end of the T7T sequence (see Table S1, Supporting Information). PCR products were then purified using phenol:chloroform extraction. High‐performance Ni‐NTA beads (GE Healthcare) and streptavidin resins (Genscript) were mixed in a 1:1 ratio and washed three times with ultra‐pure water following three additional washes with transcription buffer; 200 mm Tris‐HCl, 120 mm MgCl₂, 10 mm spermidine, 50 mm NaCl, 1 mm DTT pH 7.9. Following washes, 3 µL of the bead's mixture was inserted manually to the biochip. Following trapping of Ni‐NTA and SA beads, in vitro transcription components. Transcription mix (1× transcription buffer 200 mm Tris‐HCl, 120 mm MgCl₂, 10 mm spermidine, 50 mm NaCl, 1 mm DTT pH 7.9 and 10 mm ATP, 7.5 mm GTP, 10 mm CTP (Promega), 10 mm N1‐Me‐Pseudo UTP 8 mm m7GpppG (TriLINK), 6 µL of HIS tagged T7RP (was purified according to previously published protocol), and 3000 ng PCR product) was flown into the transcription compartment at a rate of 300 µL h⁻¹. The biochip was placed onto a 37 °C platform (NBT) for 2–4 h. Following incubation, the compartment was then washed with a 50 µL transcription buffer at a rate of 800 µL h⁻¹ and the purified mRNA was collected.

On‐Chip Production and Encapsulation of RNA and mRNA

IVT was performed as described above. Following 2–4 h of incubation, the IVT reaction was mixed with acetate buffer (pH 4). The resulting solution was introduced into a microfluidic mixer at a flow rate of 900 µL h⁻¹ through the aqueous inlet. Simultaneously, a lipid mixture consisting of ionizable lipid (EA2, EA‐405, or SM‐102), DSPC, cholesterol, and DMG‐PEG (at a molar ratio of 40:10.5:47.5:2) was introduced through the second inlet. Depending on the formulation, the lipid phase was flowed at either 900 µL h⁻¹ (FRR = 1:1), 1029 µL h⁻¹ (FRR = 0.75:1), or 450 µL h⁻¹ (FRR = 4:1), resulting in total flow rates (TFRs) of 1800, 1800, and 2250 µL h⁻¹, respectively. The resultant mixture was dialyzed against phosphate‐buffered saline (PBS; pH 7.4, Hy‐Labs) for 16 h to remove ethanol.

Fluorescence Microscopy—Microfluidics Devices Visualization

A 1XPBS (Hy‐Labs) solution containing 0.01% (W/V) of fluorescein (Sigma–Aldrich) was flown to the devices at a rate of 900 µL h⁻¹ using Cetoni GmbH neMESYS Syringe Pumps (Korbussen, Germany). In the Tesla mixer device, from the lipid mix inlet, 70% (V/V) of EtOH was flown at a rate of 300 µL h⁻1. Images of the devices were obtained using a Nikon Eclipse Ti‐E inverted microscope with 470/40 excitation filter and 525/50 emission filter.

Fluorescence Microscopy—DNA Retention inside the Microfluidic Device

The device was loaded with 3 uL of High‐performance Ni‐NTA beads and streptavidin resins as specified above. Linear DNA fragments were produced by polymerase chain reaction (PCR) using Phusion High‐Fidelity PCR Master Mix. The forward primer was tagged with Cy‐3 in its 5′ end and the reverse primer was tagged with biotin in its 5′ end (see Table S1, Supporting Information). PCR products were then purified using phenol:chloroform extraction. 1000 ng DNA was diluted in 100ul of transcription buffer and was flown in the device at a rate of 300 µL h⁻¹. The device was then washed with 50 uL of transcription buffer at a flown rate of 800 µL h⁻¹ _. Images of the beads trapped in the device were obtained using a Nikon Eclipse Ti‐E inverted microscope with 535/20 excitation filter and 590/50 emission filter.

Fluorescence Microscopy—Mango Aptamer Encapsulation in Droplets

On‐chip in vitro transcription was performed as mentioned above with the addition of Thiazol orange (Sigma–Aldrich) to the transcription buffer at a final concentration of 5 nM. Following 2–4 h, transcription buffer was flown to the compartment at a flown rate of 800 µL h⁻¹. The purified Mango aptamer was allowed to pass to the T junction motif. In the second inlet the oil phase, which consists of fluorinated oil (Fluorinert FC‐40, Sigma–Aldrich) and 2% w/w fluorosurfactant (RAN biotechnologies) was flown at a flown rate of 800 µL h⁻¹. At the T junction, the oil phase intersects the aqueous phase resulting in water‐in‐oil droplets formation. The droplets were collected and imaged using a Nikon Eclipse Ti‐E inverted microscope with 470/40 excitation filter and 525/50 emission filter.

Standard In Vitro Transcription Purification

Transcription mix was placed in a PCR tube and incubated in a PCR machine for 2–4 h at 37 °C. Following incubation, the mRNA was purified using MEGAclear Transcription Clean‐Up Kit (invitrogen) according to the manufacturer instructions.

mRNA Concentration Measurements

Purified mRNA (from both on‐chip production and purification and standard IVT and purification) were diluted 1:10‐1:90 and measured using a Qubit high‐sensitive RNA assay kit (Invitrogen). For calculating purification yields, 2 µL of the transcription mix before the purification step was also diluted similarly and measured using the Qubit high‐sensitive RNA assay kit.

For mRNA stability assays, on‐chip purified mRNA concentration was measured using an RNA assay kit (Invitrogen) and was diluted to 50 ng µL⁻¹ in either an acetate buffer or 1x Tris‐borate‐EDTA (TBE) buffer pH 8 or TBE buffer pH 8 supplemented with 1ul of RNASe A (10 mg mL⁻¹) (ThermoFisher Scientific) to a final volume of 200 µL. At each time point, a 2 µL aliquot was collected and was immediately subjected to liquid nitrogen and put in −80 °C. Following 24 h all samples were measured using Qubit RNA assay kit.

Gel Electrophoresis—mRNA

Following mRNA on‐chip production and purification, 2 µL of the mRNA sample was denaturized by incubation at 65 °C for 10 min. Then, 2 µL of RNA loading dye (NEB) was added to the sample. For LNPs, RNA loading dye was added directly to the sample. The samples were loaded onto a 1.3% agarose gel.

Gel Electrophoresis—LNPs

LNPs of 2 µL were mixed with 2 uL of RNA loading dye. For triton treatment, triton (Sigma–Aldrich) was added to the samples to a final concentration of 0.5% (V/V). The samples were immediately loaded onto a 1.3% agarose gel.

Immunoblots—RBD‐hFc Expression

LNPs‐encapsulated RBD‐hFc mRNA (1 µg mL⁻¹) were added to 5 × 10⁵ HeLa cells seeded in a six‐well plate with 3 mL DMEM‐based growth medium for HeLa cells (10% FBS, 1% Penstrep and 1% Sodium pyruvate (all from Biological Industries)). Expression cells were lysed on ice in RIPA buffer (Merck) at 24 h after transfection. Lysates were incubated at 4 °C for 10 min, then centrifuged at 20 000 × g for 10 min at 4 °C.

Immunoblots—HIS‐T7RP On‐Chip Retention

Purified HIS‐tagged T7RP was used as a control. 6 µL of an IVT reaction (as described above) was used as the input measure of HIS‐tagged T7RP. Following on‐chip IVT the IVT reaction was flown at a rate of 900 µL h⁻¹ and 6 µL from the output was used as the output measure of the HIS‐tagged T7RP on‐chip retention.

Acrylamide gel was prepared in the following fashion, running gel was created by mixing DDW, 30% acrylamide mix, 1.5 m Tris pH 8.8, 10% SDS, 10% ammonium persulfate and TEMED, then pouring the mixture between two tightly held glass squares and leaving it for half an hour to polymerize in oxygen‐poor conditions. Stacking gels were created by mixing DDW, 30% acrylamide mix, 1 m Tris pH 6.8, 10% SDS, 10% ammonium persulfate, and TEMED, then poured on top of the running gel and left to polymerize for an hour. Gels were then placed into a running buffer containing 1x Tris‐Glycine (TG)‐SDS and DDW and were run at a voltage of 150 mV until the appropriate phase was separated from the bulk of the proteins in the sample. Proteins were then transferred from the gel to a membrane in a transfer buffer containing 10% methanol, 1x TG, and DDW and a transfer reaction was run for 90 min at a current of 300 mA.

The membrane was then blocked for 1 h in blocking solution (5% BSA (Difco), 0.02% sodium azide (Sigma–Aldrich) in 1XTTBS (0.1% Tween‐20 in 1XTBS (Sigma–Aldrich)), and then incubated with the primary antibody diluted in blocking solution. The membrane was washed three times for 15 min each in TTBS, incubated for 1 h in the secondary antibody, and washed three times for 10 min each in TTBS. The membrane was developed using Clarity Western ECL (BIO‐RAD), according to the manufacturer's instructions. Images were obtained using ChemiDoc Touch Imaging System (BIO‐RAD). For the His‐T7RP membrane, goat Anti‐6XHis tag antibody‐ChiP grade primary antibody (Abcam) and Goat Anti‐Rabbit IgG H&L (HRP) secondary antibody (Abcam) were used. For RBD‐hFC membranes, mouse anti‐SARS‐CoV‐2 Spike Antibody (SinoBiological) and Goat anti‐Mouse (HRP) secondary antibody (Abcam) were used.

Dynamic Light Scattering

Size distribution and ζ‐potential of LNPs were measured by dynamic light scattering (DLS) using a Malvern nano ZS ζ‐sizer (Malvern Instruments). For size distribution, LNPs were diluted in PBS (1:40, volume ratio) and measured within a Polystyrol 10 mm × 4 mm × 45 mm disposable cuvette. For ζ‐potential measurements, LNPs were diluted in DDW (1:100, volume ratio) and measured within a sample cell folding capillary cuvette (DTS1060, Malvern)

Cryopreservation of mRNA‐LNPs

mRNA‐LNPs were concentrated to a final concentration of 500 µg mL⁻¹ using 100KD amicon tubes (MERCK) at 3500 X G. Filter‐sterilized 50% (V/V) sucrose (Sigma–Aldrich) was then added to a final concentration of 10% (V/V). The mRNA‐LNPs were then aliquoted and placed at −80 °C. Following thawing, the mRNA‐LNPs were diluted to a desired concentration using 1XPBS.

Quantification of mRNA Encapsulation

LNP encapsulation efficiency was quantified according to the manufacturer's protocol using the Quant‐iT RiboGreen RNA assay (Thermo Fisher Scientific, CA, USA).

Transmission Electron Cryomicroscopy Samples Preparation and Imaging

For the cryo‐TEM technique, a drop of 2.5 µL of the sample was placed on a carbon lacey film supported on a 300 mesh Cu grid (Ted Pella Ltd). The excess liquid was blotted, and the specimen was vitrified via a rapid plunging into liquid ethane precooled with liquid nitrogen in a controlled environment automatic vitrification system (Leica EM GP) where the temperature and the relative humidity were controlled. The samples were examined at −178  °C using ThermoFisher Scientific (former FEI) Talos F200C transmission electron microscope operating at 200 kV and equipped with a Gatan 626 cold stage. The images were taken with Ceta 16 m CMOS camera.

In Vitro Luciferase Assay

Increasing concentrations of LNPs‐encapsulated Luc mRNA (3 × 10⁴ cells mL⁻¹) were added to 2 × 10⁴ HeLa cells seeded in a 96‐well plate with 100 µL DMEM‐based growth medium for HeLa cells (10% FBS, 1% Penstrep and 1% Sodium pyruvate (all from Biological Industries)). At 24 h post LNPs treatment, cells were lysed and luciferase activity was measured using the Promega Luciferase assay kit according to the manufacturer's protocol.

Animal Experiments

This study was carried out in strict accordance with the recommendations for the Care and Use of Laboratory Animals of the National Institutes of Health. All animal experiments were performed in accordance with Israeli law and were approved by the Ethics Committee for animal experiments at Tel Aviv University (TAU—LS—IL—2403 – 107 – 3) per current regulations and standards of the Israel Ministry of Health.

In Vivo Luciferase Assay and Organ Distribution Study

LNPs‐encapsulated Luc mRNA (mRNA dose: 0.5 mg kg⁻¹) were intravenously injected into 8–10 weeks‐old female C57BL/6 mice (Envigo, Rehovot, Israel) At 6 h post‐injection, mice were intraperitoneally injected with D‐Luciferin (150 mg kg⁻¹) and major organs were harvested for imaging using IVIS‐spectrum‐CT (Perkin Elmer Inc).

Animal Vaccination Assay

Groups of 6–8 weeks of female BALB/c mice were administered intramuscularly (100 µL) with either 5ug RBD‐hFc mRNA or luciferase mRNA encapsulated in LNPs. Animals were boosted at day 25 with the same dose as day 0. Serum was collected on days 23 and 49.

ELISA

MaxiSORP ELISA plates were precoated with recombinant spike protein (2ug mL⁻¹) in carbonate buffer at 4 °C overnight. Then, the plates were washed three times with PBST 0.05% Tween 20 and blocked with 2% BSA (Sigma–Aldrich, #A8022) in PBST at 37 °C for 1 h. The plates were washed three times with PBST and then incubated with serial dilutions of mouse sera in PBST/BSA at 37 °C for 1 h. Following washing, goat anti‐mouse alkaline phosphatase‐conjugated IgG (Jackson Immuno Research Laboratory, No. 115‐055‐003) was added at 37 °C for 1 h. Following incubation, the plates were washed three times with PBST and the reactions were developed with p‐nitrophenyl phosphate substrate (PNPP; Sigma Aldrich, N2765). Plates were read at 405 nm absorbance, and antibody titers were calculated as the highest serum dilution with an OD value above two times the averaged OD of the negative control.

DNA Constructs

All DNA sequences are provided in Table S1 (Supporting Information). Plasmid encoding for mango light‐up RNA aptamer was purchased from Genscript. The plasmid encoding SARS‐CoV‐2 spike protein (pGBW‐m4046828) was a gift from Ginkgo Bioworks & Benjie Chen (Addgene plasmid # 145742; http://n2t.net/addgene:145742; RRID:Addgene_145742). All other plasmids were constructed using Gibson Assembly Master Mix (NEB) according to the standard Gibson assembly method.^{[ 57 ]}

Statistical Analysis

All data were pre‐processed to ensure quality, including checks for consistency and outliers. Data were presented as mean ± standard deviation (SD), with a sample size of n  = 3 for each group. Statistical analyses were conducted using GraphPad Prism version 9.5 (GraphPad Software). Two‐tailed unpaired Student's t‐tests were used to evaluate differences between groups. A P‐value of less than 0.05 was considered statistically significant.

Conflict of Interest

The authors declare no conflict of interest.

Author Contributions

S.Z.T. and E.G. conceived the project. S.Z.T., A.L., A.E, U.E, G.F.‐Z., T.K., L.G., M.G., E.K., and Y.N. designed and performed the experiments with input from O.C., T.PJ.K., D.P., and S.Z.T. E.G. wrote the manuscript with input from all authors.

Supporting information

Supporting Information

Zilberzwige‐Tal S., Levin A., Ezra A., Elia U., Finkelstein‐Zuta G., Kreiser T., Gershon L., Goldsmith M., Kon E., Navon Y., Cohen O., Knowles T. P., Peer D., Gazit E., On‐Chip De Novo Production of mRNA Vaccine in Lipid Nanoparticles. Small 2025, 21, 2500114. 10.1002/smll.202500114

Data Availability Statement

The data that support the findings of this study are available from the corresponding author upon reasonable request.