Genetic Code Expansion in Mammalian Cells: a Plasmid System Comparison

Abstract

Genetic code expansion with unnatural amino acids (UAAs) has significantly broadened the chemical repertoire of proteins. Applications of this method in mammalian cells include probing of molecular interactions, conditional control of biological processes, and new strategies for therapeutics and vaccines. A number of methods have been developed for transient UAA mutagenesis in mammalian cells, each with unique features and advantages. All have in common a need to deliver genes encoding additional protein biosynthetic machinery (an orthogonal tRNA/tRNA synthetase pair) and a gene for the protein of interest. In this study, we present a comparative evaluation of select plasmid-based genetic code expansion systems and a detailed analysis of suppression efficiency with different UAAs and in different cell lines.

1. Introduction

Proteins carry out the essential processes of life through their diverse structures and functions. These properties arise from the 20 natural amino acids and their extensive post-translational modifications (PTMs), such as methylation, glycosylation, and phosphorylation. Despite the diversity and variability of natural proteins’ functions, the natural amino acids have limited chemical functionalities and reactivity compared to the repertoire of synthetic organic chemistry. These limitations have been overcome through pioneering work by Schultz and others on expanding the genetic code for the incorporation of customizable and chemically unique unnatural amino acids (UAAs).^{1, 2} UAA mutagenesis in mammalian cells has facilitated the study of biological events at the molecular level, such as proximity-based protein-protein interactions in living organisms;^3–5 activation of protein function using small molecules^6–8 and light;^9–14 and introduction of various PTMs, including phosphorylation,^{15, 16} acetylation,¹⁷ crotonylation,¹⁸ and SUMOylation.⁸ Genetic code expansion has also been achieved in animals.^{19, 20} Apart from providing a better understanding of the molecular mechanisms of protein functions, this technique has also been applied to the development of therapeutics, including site-specific antibody-drug conjugates (ADCs),^21–25 controllable Chimeric Antigen Receptor T (CAR-T) cells,^26–28 stabilized^{29, 30} and spatiotemporally controllable³¹ adeno-associated viruses, and attenuated live vaccines.^{32, 33}

UAA mutagenesis employs orthogonal aminoacyl-tRNA synthetase (aaRS)-tRNA_CUA pairs to suppress amber stop codons (UAG), resulting in the incorporation of the UAA. This strategy requires the delivery of the following components into host cells: (i) the UAA, (ii) the gene encoding an engineered aaRS mutant for the desired UAA, (iii) the gene for the tRNA_CUA, and (iv) the gene for the target protein carrying the amber codon at the desired position.

While the production of sufficient protein-of-interest (POI) with the UAA incorporated is crucial, the efficiency of UAA incorporation is often limited in mammalian cells³⁴ and animals,²⁰ frequently requiring optimization. Numerous factors influence the efficiency of UAA mutagenesis, some of which often cannot be changed, such as the position of UAA insertion and the chemical structure of the UAA. However, other parameters can be optimized to improve protein yields, such as UAA concentration, aaRS and tRNA_CUA expression levels, POI mRNA transcription levels, and the aaRS’s catalytic efficiency. It should also be noted that the sequence context of the nonsense codon can be altered within the limits of silent mutations for improved suppression efficiency, which is out of the scope of this work.³⁵

Among the aaRS-tRNA pairs used for UAA mutagenesis, the pyrrolysyl-tRNA synthetase/tRNA_CUA (PylRS/PylT) pair has risen to prominence as the most widely applied.³⁶ This system is advantageous because the pair is orthogonal in both mammalian cells and E. coli hosts, allowing rapid testing and optimization in E. coli before moving into higher cells and organisms. Further, PylRS is able to recognize chemically diverse UAAs, and it does not have a natural substrate in bacterial and mammalian cells.³⁷ Among these substrates, we have chosen to focus on Nϵ-allyloxycarbonyl-lysine (AOK) and hydroxycoumarin lysine (HCK). The former is a well-validated positive control for several PylRS variants (including wild-type) as well as handle for bioconjugation via photoinitiated thiol-ene reactions,³⁸ Pd-catalyzed deallylation, and inverse electron-demand Diels-Alder reactions. The latter is a photocaged lysine that has been incorporated into several proteins in mammalian cells and zebrafish, for example enabling the activation of signaling pathways and DNA manipulation using light.^39–41 The conclusions drawn from our work are expected to generalize to many other UAAs.

We utilized an established mCherry-TAG-EGFP fusion protein as a reporter to gauge the incorporation efficiency¹³ of both AOK and HCK by HCKRS (the Y271A-L274M mutant of PylRS)⁴⁰ using different vectors and cell types. When cells translate this protein, mCherry is first produced, then the ribosome reaches the UAG codon. Here, tRNA_CUA-AOK or tRNA_CUA-HCK adds the UAA to the growing polypeptide chain, allowing for subsequent EGFP translation. Our results with this reporter demonstrate that any of the aforementioned factors (concentrations of UAA, expression levels of aaRS, copy numbers of tRNA_CUA cassettes, promoter of the POI, and aaRS variants) can be the limiting factor under different circumstances.

2. Results and discussion

Here, we conducted a comparative study of four commonly used plasmid systems for UAA incorporation in mammalian cells (Figure 1): 1) the pAG dual-plasmid system originally developed by the Chin Lab¹³ and re-engineered by us ( ① ),⁴⁰ 2) a new pcDNA dual-plasmid system developed by our lab( ② ), 3) the pE323/363 dual-plasmid system developed by the Chin Lab ( ③ and ④ ),³⁴ and 4) the pEA single plasmid system developed by the Arbely Lab (⑤-⑦).⁴²

Figure 1.

Different systems used for UAA mutagenesis in mammalian cells. A) pAG_HCKRS_4xPylT and pmCherry-TAG-EGFP. CMVenh = CMV immediate early enhancer; B) pcDNA_HCKRS_2xPylT and pcDNA_mCherry-TAG-EGFP_2xPylT. C) pE323_HCKRS_4xPylT and pE363_mCherry-TAG-EGFP_4xPylT. D) pEA_HCKRS_mCherry-TAG-EGFP_4xPylT.

We tested multiple versions of these systems to dissect the effects of the following: 1) different promoters, 2) copy number of PylT expression cassettes, 3) use of transcription enhancing elements, 4) number of plasmids introduced to cells, and 5) different PylRS mutants and homologs, including MmPylRS (from Methanosarcina mazei),⁴³ MbPylRS (from Methanosarcina barkeri),⁴⁴ an Mb-Mm chimeric PylRS,⁴⁵ and a PylRS derived from MbPylRS with an N-terminal nuclear export signal (NES).⁴⁶ In all cases, an N-terminal FLAG tag was appended to the PylRS for detection by western blot. Furthermore, the difference between two UAAs, different concentrations of the same UAA, and different cell types were also compared.

2.1. UAA concentration

The UAA concentration in cell culture media usually ranges from 0.1 mM to 2 mM, depending on the toxicity, solubility, and the ease of synthesis of the specific UAA. Increasing the concentration of UAAs typically enhances suppression efficiency until saturating the system, as noted when comparing 1 mM and 0.25 mM of AOK for constructs ②-⑦ in Figure 2A and 3A.

Figure 2.

Evaluation of suppression efficiency with different UAA mutagenesis systems in NIH3T3 cells. A-B) Suppression efficiency represented by integrated EGFP fluorescence divided by integrated mCherry fluorescence of each well in the presence of 0.25 or 1 mM of AOK, or 0.25 mM of HCK. C) Individual mean mCherry and EGFP fluorescence intensity values for each cell (n = 500, randomly sampled for each system) with 1 mM of AOK.

Figure 3.

Evaluation of suppression efficiency with different UAA mutagenesis systems in HEK293T cells. A-B) Suppression efficiency represented by integrated EGFP fluorescence divided by integrated mCherry fluorescence of each well with 0.25 or 1 mM of AOK, or 0.25 mM of HCK. C) Suppression efficiency represented by mean mCherry and EGFP fluorescence intensity of each cell (n = 500 for each system) with 1 mM of AOK.

When the UAA shows some toxicity to cells or is hard to dissolve in aqueous media, its concentration can be lowered to a point where cell survival and UAA incorporation are both sustained.

2.2. Cell types

Studies of different physiological processes call for the use of different cell types, some of which are easier to transfect than others. Thus, we tested the performance of these systems in HEK293T cells, an easy-to-transfect human neuroendocrine cell line,⁴⁷ and in NIH3T3, a hard-to-transfect murine fibroblast cell line.⁴⁸ For both cell lines, plasmids developed by the Chin lab ( ③ and ④ ) have the highest incorporation yield (Figure 2A, 2B, 3A, and 3B). However, some differences still exist between the two different cell lines, primarily due to differences in transfection efficiency and plasmid replication. When the single-plasmid systems are used for UAA mutagenesis (⑤-⑦), the amber codon suppression efficiency and the UAA to no UAA ratio is higher in the more easy-to-transfect HEK293T cells (Figure 3A and 3B) than in NIH3T3 cells (Figure 2A and 2B). We attribute this effect to the SV40 origin of replication on the pEA single-plasmid system functioning effectively in HEK293T cells, but not in NIH3T3 cells.⁴⁹

It is also noteworthy that in both cell lines, ③ and ④ provided the highest target protein yields not only for the entire cell population, but also in individual cells. In Figure 2C and 3C, each dot represents the mean fluorescent intensity of mCherry and EGFP in a single cell, with cells transfected with ③ and ④ showing the highest EGFP intensity of all systems tested. This finding suggests that these systems are preferred for both bulk assays (e.g., western blot) and single cell analyses (e.g., fluorescence microscopy, cytometry, or single-cell sequencing).

2.3. Copy number and origin of tRNA_CUA

Previous reports show that high concentrations of tRNA_CUA are vital for optimal UAA incorporation efficiency.^{34, 50–52} As observed in Figure 2A and 2B, the amber codon suppression efficiency of ⑤ exceeds that of ② in NIH3T3 cells, which has a similar level of PylRS expression as ⑦ (Figure 4B), but only 2 copies of tRNA_CUA expression cassettes on each plasmid (Figure 1B). Hence, we conclude that a higher copy number of tRNA_CUA expression cassettes per plasmid enhances UAA incorporation in cells. For certain applications when background signal from excess PylT-UAA trapped in the cell can interfere (e.g., fluorescent labeling of UAAs), the number of copies of PylT should be reduced.⁵³

Figure 4.

Evaluation of transfection efficiency and aaRS expression levels with different UAA mutagenesis systems in NIH3T3 cells. A) Expression levels of the mCherry-TAG-EGFP construct were determined by fluorescence microscopy by normalizing integrated mCherry fluorescence to integrated Hoechst 33342 fluorescence (for nuclear staining) in each well. All values are presented as fold changes relative to that of ①. B) Expression levels of PylRS in each system, represented by the ratios between the density of anti-FLAG-PylRS and anti-GAPDH bands, both quantified from western blots (representative western blot shown on the right). Error bars represent standard deviations from three independent experiments. UAAs were not added to the media in these experiments.

Each of these systems uses different homologs or different combinations of tRNA_CUA. The pAG system (①) encodes 4 copies of MmPylT. The pcDNA system (②), by contrast, encodes one copy of MbPylT and one copy of Dh (Desulfitobacterium hafniense) PylT;⁵⁰ two different homologs are used to reduce loss of repetitive sequences due to homologous recombination during bacterial amplification of the plasmid.⁵⁴ The pE323/363 system encodes MmPylT with a U25C mutation,⁵⁵ and the pEA system uses two copies of the MmPylT-U25C PylT and two copies of DhPylT. MmPylT and MbPylT only differ by one nucleotide (T42 in the former and C42 in the latter) and can both be readily acylated by PylRS.⁵⁶ MmPylT’s U25C mutation only mildly increases incorporation efficiency,³⁴ DhPylT does not perform better than MmPylT,⁵⁷ and these homologs perform similarly and are interchangeable in E. coli.⁵⁸ Though U6 promoters and H1 promoters are used exclusively or in combination in different systems, it is believed that these commonly used Pol III promoters do not differ significantly in the RNA expression levels they generate.⁵⁹

Taken together, these observations suggest that the choice of PylT homolog does not significantly affect suppression efficiency in our study. It should also be noted that most of the plasmids used in this study range from 7 to 10 kb in length (see experimental section for details), with the exception of pmCherry-TAG-EGFP (5.5 kb) in ①. Although transfection efficiency decreases with increasing plasmid size below 5 kb,⁶⁰ plasmid sizes ranging from 7 to 10 kb do not significantly alter the efficiency of lipid-mediated transfection.⁶¹

2.4. Promoters for protein of interest and PylRS transcription

The consequences of using different promoters and enhancer elements in a combinatorial fashion has been widely discussed.^62–64 While more complex combinations of promoters are conceivable, we limited our evaluation to the most widely-used UAA incorporation systems, i.e., the CMV (cytomegalovirus) promoter and the EF1α (human elongation factor 1 alpha) promoter. We observed inhibitory effects when multiple CMV enhancer elements are used for POI and PylRS transcription. CMV enhancers in system ① led to attenuated expression of both genes (Figure 5A) and PylRS mRNA (Figure 5B). We reason that this effect is due to the competition for common transcriptional elements by six CMV enhancers in system ①.⁶² This effect is more pronounced in HEK293T cells, as they are more efficiently transfected and have the ability to replicate plasmids with SV40 origins (①, ②, and ⑤-⑦), presumably leading to higher copy numbers of plasmid ① in cells.

Figure 5.

Evaluation of transfection efficiency and aaRS expression levels with different UAA mutagenesis systems in HEK293T cells. A) Expression levels of the mCherry-TAG-EGFP construct were determined by fluorescence microscopy by normalizing integrated mCherry fluorescence to integrated Hoechst 33342 fluorescence (for nuclear staining) in each well. All values are presented as fold changes relative to that of ①. B) Expression levels of PylRS in each system, represented by the ratios between the density of anti-FLAG-PylRS and anti-GAPDH bands, both quantified from western blots (representative western blot shown on the right). Error bars represent standard deviations from three independent experiments. UAAs were not added to the media in these experiments.

In addition to competition between the same promoters and enhancers, competition between mixed promoters should be considered when optimizing expression systems. In comparing vectors ③ and ⑦, we observed that the EF1α promoter drives much higher PylRS expression than the CMV promoter (Figure 5B), while EF1α-driven mCherry fluorescence remained unchanged (Figure 5A), suggesting the EF1α promoter’s dominance over the CMV promoter in HEK293T cells. This difference could be attributed to the widely observed silencing of CMV promoter in mammalian cells due to DNA methylation, though this phenomenon is mainly reported in stable cell lines over time^65–67 and evidence for non-integrating vectors is still scarce.⁶⁸ Additionally, topological changes such as negative supercoiling upstream of the CMV promoter induced by the stronger EF1α promoter may contribute.⁶²

These potential mechanisms merit consideration when using single-plasmid UAA mutagenesis systems, especially those with the CMV promoter, as promoter silencing and plasmid supercoiling may lead to inferior UAA mutagenesis efficiency as observed for in ⑤, ⑥, and ⑦. For optimal UAA mutagenesis, the competition appears to be relieved using a dual-plasmid system with two EF1α promoters, such as ③ and ④ (Figure 3A and 3B).

2.5. Variants of PylRS

Even though different PylRS homologs have similar sequences and structures, they differ in their catalytic efficiency for certain substrates.^{69, 70} In our screening with AOK and HCK, MmPylRS had suppression efficiency equal to or greater than that of MbPylRS in mammalian cells (comparing ⑤ and ⑦ in Figure 2A, 2B, 3A, and 3B). The former has ~3-fold higher catalytic efficiency for UAA activation, while the latter has ~3-fold-higher catalytic efficiency for PylT aminoacylation.⁷¹ Thus, the catalytic efficiency overall is comparable, though biochemically measured enzyme kinetics do not always correlate with suppression efficiency.

Recent progress in the field was made by evolving an MmPylRS and MbPylRS chimera. One mutant had an up to 9.7-fold higher incorporation efficiency than its native PylRS counterpart, without sacrificing amino acid specificity. This optimized enzyme had four mutations in the N-terminal domain (“IPYE” — V30I, T56P, H62Y, and A100E), enhancing its catalytic efficiency.⁴⁵ In our testing, this chimeric construct indeed exhibited higher suppression efficiency than MmPylRS under challenging conditions (i.e., in NIH3T3 cells and at lower concentrations of UAA (Figure 2A and 2B)).

In addition, localization of the aaRS to the cytoplasm has also been demonstrated to improve the efficiency of UAA incorporation, either using a nuclear export sequence (NES)⁴⁶ or the recently discovered, predominately cytosolic Methanomethylophilus alvus homolog of the aaRS.⁷² In our experiments, NES-MbHCKRS facilitated slightly higher suppression efficiency than its unmodified counterpart (comparing ⑤ and ⑥ in Figure 2B). Other reports also show that appending an NES to PylRS confers higher amber codon suppression efficiency for low concentrations of BocK (0.05 mM) with PylRS_AF, but appears to have minimal influence when applied to higher concentrations of UAA.⁵⁷

Intriguingly, the PylRS homologs had varying levels of expression from the same plasmid backbone. The MmPylRS had the highest expression level, while the MbPylRS showed significantly reduced expression (comparing ⑤ and ⑦ in Figure 4B and 5B), with the chimeric HCKRS (comparing ③ and ④ in Figure 4B and 5B) and NES-MbHCKRS (comparing ⑤-⑦ in Figure 4B and 5B) in the middle. Though MbPylRS (⑤) and NES-MbPylRS (⑥) share almost identical gene sequences and are likely expressed in similar amounts, the latter’s cytoplasmic enrichment contributed to its higher signal on western blot because the mammalian protein extraction buffer (GE) used in this study is less efficient in the extraction of nuclear proteins. It is noteworthy that even with relatively low expression levels, chimeric HCKRS still demonstrated excellent amber suppression, showing that PylRS concentration is not a limiting factor when its catalytic capability is optimized. It is interesting to notice that, though expressed at a relatively lower level, the chimeric IPYE-PylRS in ④ exhibited higher suppression efficiency than MmPylRS in ③.

3. Conclusion

Overall, of these commonly used platforms for UAA incorporation in mammalian cells, we conclude that the pE323/pE363 two-plasmid system ( ③ and ④ ) delivers the highest incorporation efficiency across different UAA species and cell types, in particular when using a PylRS chimera with an optimized N-terminal domain ( ④ ). In this system, one plasmid contains a (U6-PylT)₄ cassette and P_EF1α-PylRS, and the other contains the P_EF1α-POI and an additional (U6-PylT)₄ cassette. The PylRS-PylT plasmid from this system has been conveniently and effectively used in double-transfections with previously tested and established POI vectors.⁸ If an experiment requires co-expression of several transgenes from multiple plasmids (e.g., reporters for live cell imaging), however, single plasmid systems ⑤ - ⑦ delivering the POI, PylT, and PylRS may be more convenient.

In addition to these general considerations, in order to obtain high expression levels of the UAA-containing POI, it is beneficial to have a high UAA concentration, have a high copy number of PylT expression cassettes, limit the use of repetitive CMV enhancers and promoters, and distribute the genes for the aaRS and the POI onto different plasmids to avoid transcriptional interference. In particular, the use of the EF1α promoter for PylRS and the POI is recommended as it did not show significant competition or silencing effects when used in both plasmids. Also, the chimeric IPYE-HCKRS mutant’s incorporation efficiency was equal to, or better, than the Mm homolog, with superior performance in challenging conditions (e.g., low UAA concentrations in difficult-to-transfect NIH3T3 cells). These parameters should be adjusted to the demands of the specific cellular processes under study (e.g., the desire for the POI to be expressed at a similar level as its endogenous variant). We hope that the results of this study will help investigators apply and adapt UAA mutagenesis systems in future research that requires efficient genetic code expansion in mammalian cells.

4. Experimental

4.1. Mammalian cell culture and transfection

NIH3T3 and HEK293T cells were acquired from ATCC and maintained in DMEM (Fisher) with 10% FBS (Fisher/USB) without antibiotics and incubated at 37 °C, 5% CO₂. All experiments were performed in triplicate.

For imaging purposes, cells were seeded in poly-D-lysine (70,000–150,000 kDa, MP Biomedicals)-treated 96-well plates (Greiner Bio-One) at 20,000 cells/well. When cell density reached ~90%, growth media was changed to 90 μL fresh DMEM with 10% FBS. To prepare the transfection reagent, 100 ng total plasmid DNA (50 ng of each plasmid for dual-plasmid systems) was mixed with 0.3 μL Lipofectamine 3000 and 0.2 μL P3000 (Fisher) in 10 μL Opti-MEM (Fisher) and added to each well. AOK or HCK was added to designated wells at this time from 100 mM stock solutions (in DMSO). DMSO (1 μL/well) was also added to no UAA (no amino acid) wells as a negative control. Protein expression was carried out at 37 °C, 5% CO₂ for 36 hours before nuclear staining and fluorescence microscopy (see 4.3).

For western blotting, cells were seeded in 24-well plates with 50,000 cells/well. When cell density reached ~90%, growth media was changed to 450 μL fresh DMEM with 10% FBS. To prepare the transfection reagent, 500 ng total plasmid DNA (250 ng of each plasmid for transfecting dual-plasmid systems) was mixed with 1.5 μL Lipofectamine 3000 and 1 μL P3000 (Fisher) in 50 μL Opti-MEM (Fisher) and added to each well. AOK or HCK was added to designated wells at this time from 100 mM stock solutions (in DMSO). DMSO (5 μL/well) was also added to no UAA wells as negative control. Protein expression was carried out at 37 °C, 5% CO₂ for 36 hours before cell lysis and western blotting.

4.2. Plasmids

All DNA fragments amplified by PCR were purified using a PCR Cleanup Kit (Omega Bio-tek). After restriction digestion, all plasmid backbones were treated with Antarctic Phosphatase (NEB) and purified by agarose gel extraction (Omega Bio-tek). All constructed plasmids were confirmed by Sanger sequencing (Genewiz).

The pmCherry-TAG-EGFP plasmid (5.5 kb) was constructed by the Chin lab.¹³ The pAG_HCKRS_4xPylT (9.7 kb) plasmid was previously constructed by Chungjung Chou of our lab.⁴⁰

The pcDNA_HCKRS_2xPylT (7.2 kb) and pcDNA_mCherry-TAG-EGFP_2xPylT (7.5 kb) plasmids were constructed as follows. The U6-PylT-H1-PylT cassette was ordered as a gBlock (IDT, Table S1), digested by BglII and MfeI (New England Biolabs, NEB), and ligated with T4 DNA Ligase (NEB) into a pcDNA3 (Invitrogen) backbone after the backbone was digested with the same restriction enzymes. This plasmid was then digested with EcoRI-HF and BamHI-HF (NEB) for T4 ligation with either HCKRS (Mb) or mCherry-TAG-EGFP fragments digested with the same restriction enzymes. HCKRS (Mb) was PCR amplified using Mb_pcDNA_F/R primers (Table S2, same for all other primers) and Phusion High-Fidelity Polymerase (Fisher). mCherry-TAG-EGFP was PCR amplified using reporter_pcDNA_F/R primers with pmCherry-TAG-EGFP as the template.

The pE323_wildtype MmPylRS_4xPylT (10.0 kb) and pE363_mCherry-TAG-EGFP_4xPylT (9.1 kb) plasmids were kindly provided by the Chin lab. The Y271A and L274M mutations were introduced to wildtype MmPylRS on pE323 plasmid to convert it to HCKRS (Mm) using the MmPylRS_Y271A_L274M_F/R primers. The construction of pE323_Mb-Mm chimeric HCKRS and its assembly into the pE323 backbone have been described previously.⁸

The pEA_4xPylT plasmid was kindly provided by the Arbely Lab. It was digested with NotI-HF, KpnI-HF, HindIII-HF, and BamHI-HF (NEB) for Gibson assembly with the mCherry-TAG-EGFP fragment and one of the aaRS fragments discribed below.⁷³ The mCherry-TAG-EGFP fragment was PCR amplified using reporter_pEA_F/R primers, with pmCherry-TAG-EGFP as template. The HCKRS (Mb) fragment was PCR amplified from pAG_HCKRS_4xPylT using FLAG-Mb_pEA_F and Mb_pEA_R primers. The NES-HCKRS (Mb) fragment was PCR amplified using FLAG-NES-Mb_pEA_F and Mb_pEA_R primers, with pAG_HCKRS_4xPylT as the template. The HCKRS (Mm) fragment was PCR amplified using FLAG-Mm_pEA_F and Mm_pEA_R primers, with pE323_HCKRS (Mm)_4xPylT as the template. The final pEA_4xPylT_HCKRS_mCherry-TAG-EGFP plasmid is ~8.4 kb.

4.3. Fluorescence microscopy

Cells in 96-well plates were washed carefully with PBS once. Cells were then incubated in 1 μM Hoechst 33342 dye in PBS at 37 °C, 5% CO₂ for 15 minutes. The cells were then carefully washed once with PBS. After washing, the buffer was replaced with Live Cell Imaging Solution (Fisher), and imaged in tiled grids using a Zeiss Observer Z1 microscope (10x objective, NA 0.4 plan-apochromat) linked to an HBO 100 high-pressure mercury plasma arc-discharge lamp. Fluorescence microscopy was performed using the following filter cubes (Zeiss): EGFP (38 HE; ex: BP470/40; em: BP525/50), DsRed (43 HE; ex: BP550/25; em: BP605/70), and Hoechst filter (49, Ex. G 365; Em. BP 445/50). For each well, the tiles were stitched, and the fluorescence intensity in each channel (mCherry, EGFP, and Hoechst 33342) was quantified using Zen 2 software. For background subtraction, the maximum fluorescence intensity of an area without any transfected cells was set as the lower threshold for integration of total fluorescence intensity of the entire well.

Imaging cytometry was analyzed using ImageJ (including DnsAnalysisCreator and DnsMacroExample macros) developed by De Novo Software. First, positively transfected cells were identified in the mCherry channel, individually labelled, and used to create a mask using the DnsAnalysisCreator macro. Second, mean mCherry and EGFP fluorescence for each cell was quantified and exported by the DnsMacroExample macro. Approximately 1,000–2,000 cells were analyzed for each condition, and 200 cells were randomly selected for display in Figure 2C and 3C.

4.4. Western blotting

Cells in 24-well plates were washed carefully with ice-cold PBS buffer once, and then lysed with mammalian protein extraction buffer (GE) supplemented with HALT protease inhibitor cocktail (Fisher) with orbital shaking on ice for 20 min. The lysate was clarified by centrifugation at 21,000 rcf for 20 minutes at 4 °C. After boiling at 95 °C for 10 minutes in Laemmli buffer, the supernatants were run on 10% SDS-PAGE gels and transferred to a PVDF membrane (Millipore) following standard protocols. The membranes were blocked in 5% (w/v) non-fat milk (LabScientific) dissolved in TBS-T at room temperature for 1 hour. Anti-FLAG primary antibody (Proteintech, 20543–1-AP) and anti-GAPDH primary antibody (Proteintech, 10494–1-AP) were used at a 1:1,000 dilution in TBS-T and incubated at 4 °C overnight. The membranes were incubated in HRP-conjugated anti-rabbit IgG (CST, 7074S) at a 1:10,000 dilution in TBS-T for at room temperature for 1 hour. Membranes were then incubated with SuperSignal West Pico Chemiluminescent Substrate (Fisher) and visualized with a Bio-Rad ChemiDoc™ MP Imaging System. Quantification was performed by integrating the density of each band using ImageLab 5.0 software.