The heavy and light chains genes were cloned into separate pcDNA3.4 TOPO expression vectors by GeneArt (ThermoFisher GeneArt, Regensburg, Germany). Gene synthesis and cloning were performed by GeneArt using ThermoFisher’s recommended Kozak consensus sequence and codon optimization for human cells, as determined by the GeneArt codon optimization tool. The commercially available pcDNA3.4 vector is designed for high-level expression in mammalian systems and includes the human cytomegalovirus (CMV) promoter to drive strong transcriptional activity. Additionally, the vector contains the woodchuck hepatitis virus posttranscriptional regulatory element (WPRE) to enhance transcript stability and expression efficiency.

Both heavy and light chain constructs included their native signal peptides to ensure proper secretion of the expressed antibody into the culture medium.

Plasmid DNA was amplified in Escherichia coli One Shot Top™ TOP10 chemically competent cells (ThermoFisher Scientific, Waltham, MA, cat. no. K204040) according to the manufacturer’s protocol. Following bacterial culture and plasmid amplification, plasmid DNA was purified using GeneJET endo-free plasmid maxiprep kit (ThermoFisher, cat. no. K0861) following the kit instructions.

Expi293F cells (Thermo Fisher Scientific, cat. no. A14527) were thawed and cultured in BalanCD HEK293 medium (Fujifilm Irvine Scientific, cat. no. 94155), a more affordable alternative to the Expi293 Expression Medium recommended by the manufacturer (Expi293 Expression System, 2025). A direct medium exchange was performed without compromising cell viability, as previously reported (Napoleone et al., 2021). According to the supplier’s protocol, Expi293F cells are typically cultured in an 8% CO₂ atmosphere to maintain pH in the proprietary medium. However, in this protocol, cells were adapted to standard 5% CO₂ conditions without affecting viability or productivity, as BalanCD HEK293 medium is compatible with lower CO₂ levels.

Cryopreserved cells were stored at −150 °C and thawed rapidly in a 37 °C water bath until dissolved. The vial was decontaminated with 70% ethanol before opening in a laminar flow hood. Cells were gently transferred to 5 mL of pre-warmed BalanCD HEK293 medium, centrifuged at 200 g for 5 min to remove residual DMSO, and resuspended in 1 mL of fresh pre-warmed medium. The suspension was transferred to a 125 mL sterile, vented Erlenmeyer flask containing 30 mL of BalanCD HEK293 medium and incubated at 37 °C, ≥80% humidity, and 5% CO₂ on an orbital shaker. Shaking speed was adjusted based on the platform diameter: 125 rpm (19 mm), 120 rpm (25 mm), or 95 rpm (50 mm). Cell viability was monitored post-thaw and typically reached ≥90% within 3–4 days. Cells were subcultured when they reached 1 – 3 million viable cells/mL (mvc/mL).

For routine maintenance, cells were cultured at 37 °C with 5% CO₂ in 30 mL volumes in 125 mL Erlenmeyer plastic flasks with ventilated caps (VWR, cat. no. 734–1832). Cultures were maintained between 0.3 and 3 – 5 mvc/mL and passaged every 3–4 days. Cell viability and viable cell density were assessed using trypan blue exclusion and measured with a Countess 3 automated cell counter (Invitrogen).

Expi293F cells were cryopreserved at a density of 1 × 10⁷ viable cells/mL in BalanCD HEK293 medium supplemented with 10% DMSO. Cells were pelleted, resuspended in ice-cold freeze medium, and frozen using a controlled-rate freezing method (−1 °C/min), followed by transfer to −150 °C for long-term storage.

To maximize transfection efficiency, Expi293F cells were split the day before transfection to remove depleted medium and accumulated secreted factors, and to ensure the culture was in a healthy logarithmic growth phase. Cells were harvested by centrifugation at 200 g for 10 min at room temperature. The spent medium was carefully removed, and cells were resuspended in pre-warmed fresh BalanCD HEK293 medium at a density of 1 mvc/mL.

Alternatively, cells can be resuspended at 2.5 – 3 mvc/mL to reduce medium usage without compromising transfection efficiency or protein yield. Following resuspension, cells were returned to standard culture conditions (37 °C, 5% CO₂, 125 rpm) and incubated overnight prior to transfection.

On the day of transfection, Expi293F cell density and viability were measured. Cells were harvested by centrifugation at 200 g for 10 min at room temperature and resuspended in pre-warmed BalanCD HEK293 medium to a final density of 20 mvc/mL.

Transfections were performed in 20 mL cultures (at 20 mvc/mL) by adding a total of 1.25 μg DNA per mvc, using a 1:5 ratio of plasmid DNA to salmon sperm DNA. Polyethyleneimine (PEI MAX, linear 40 kDa, 1 mg/mL) was added at 3.75 µL per mvc, corresponding to an optimized 3:1 PEI-to-DNA ratio (de los Milagros Bassani Molinas et al., 2014). Transfections were carried out in 125 mL Erlenmeyer plastic flasks with ventilated caps (VWR, cat. no. 734–1832) and incubated for 3 h under standard culture conditions (37 °C, 5% CO₂, 125 rpm).

After 3 h, transfected cells were diluted to final density of 1 mvc/mL in 400 mL of BalanCD HEK293 cell medium supplemented with valproic acid (VPA) to a final concentration of 3.5 mM. Cultures were transferred to 1 L Erlenmeyer plastic flasks with ventilated caps (VWR, cat. no. 734–1889) and incubated for 6–7 days under standard shaking conditions.

For harvesting, Celpure® P300 (Sigma/Merck, cat. no. 525243) was added directly to the cell culture at 1% (w/v). The suspension was briefly mixed and then vacuum-filtered through a 0.22 μm filter to remove cells and debris prior to purification.

The purification strategy was dependent on the specific design of the protein construct and the presence or absence of affinity tags, therefore, protocols were tailored accordingly and are not described in full detail here.

However, for constructs containing Fc regions, we recommend using HiScreen Fibro PrismA columns (Cytiva, cat. no. 17549816) as an efficient alternative to traditional HiTrap Protein-G columns (Cytiva, cat. no. 17040501). HiScreen Fibro PrismA columns support flow rates up to 30 mL/min, compared to 5 mL/min for Protein-G columns, enabling up to an 80% reduction in purification time without compromising yield or purity. Elution was performed using 0.7% (v/v) acetic acid, a milder elution condition that minimizes antibody denaturation while maintaining efficient recovery. Eluted fractions were immediately neutralized by the addition of with 1 M Tris, pH 8.0, at a ratio of 1:10 (v/v) to ensure pH stability and maintain antibody integrity.

Equal volumes of representative bispecific antibody samples, collected directly from cell culture supernatant or from purified protein preparations, depending on specific experiment, were mixed with 4x LDS sample buffer (ThermoFisher, cat. no. B0007) and loaded onto 4%–12% Bis-Tris protein gels (Invitrogen, cat. no. NW04125BOX) covered in MES buffer (ThermoFisher, cat. no. NP0002). The samples were separated on SDS-PAGE with 80 V for approximately 2 h, then transferred to a methanol activated PVDF membrane (ThermoFisher, cat. no. 88520) at 100 V for 2 h. After transfer, membranes were dried, reactivated in methanol, and blocked in 5% (w/v) skimmed milk in TBS-T overnight. The membrane was incubated with a goat-anti-mouse conjugated to horse-radish peroxidase (HRP) (Sigma, cat. no. 12–349). The membrane was washed thrice with TBS-T. HRP chemiluminescent substrate (Invitrogen, cat. no. WP20005) was added and imaged using the Odyssey Fc (LI-COR Biosciences).

Expi293F cells used in this study are HEK293 cells adapted to grow in suspension at high density (>5 mvc/mL), enabling significantly higher recombinant protein yields per culture volume (Expi293 Expression System, 2025). In our previously published TGE protocol (Fang et al., 2017), we used the recommended Expi293F expression medium, which is specifically formulated for these cells but is relatively expensive. To reduce costs, we replaced it with BalanCD HEK293 medium, which decreased the medium cost from $358 to $42 per litre, a reduction by nearly eightfold. BalanCD HEK293 medium was selected as a replacement because it is specifically designed to support high-density, suspension-cultured HEK293 cells and enhance TGE performance (Gao et al., 2022).

Expi293F cells adapted to BalanCD HEK293 medium without requiring a stepwise adaptation process and performed comparably to cells cultured in Expi293 expression medium with respect to viability, cell density, growth rate, and recombinant protein yield, as previously shown (Napoleone et al., 2021). Based on these results, BalanCD HEK293 medium was adopted as the expression medium in the updated protocol. Furthermore, because BalanCD HEK293 medium already contains 0.1% poloxamer (a surfactant similar to Pluronic F-68), the previously required Pluronic F-68 supplementation was no longer necessary (Vasu et al., 2025).

Plasmid DNA production is both time-consuming and expensive, making its reduction a valuable improvement. By maintaining the total DNA concentration while diluting the plasmid DNA with inexpensive salmon sperm filler DNA, plasmid usage was reduced by up to 80%. As shown in Figure 2, two transfection conditions using the same total DNA concentration, one with and one without salmon sperm DNA, produced comparable antibody yields, demonstrating that filler DNA does not negatively impact expression efficacy. Additionally, we included a representative Western blot of purified IgG, confirming that there was no difference in antibody quality or integrity after purification between the conditions (Figure 2).

In our previous TGE protocol, the harvesting procedure required a 1-h centrifugation step to pellet cells and debris, followed by a 1-h filtration process. Despite centrifugation, 0.22 µm filters frequently clogged and had to be replaced 3–5 times due to residual cell material. To overcome this bottleneck, Celpure® P300, a refined diatomaceous earth, was incorporated into the updated TGE protocol. Celpure® is a natural siliceous mineral compound with high absorptive capacity for water and oils (Özen et al., 2015), enabling it to effectively sequester cells and cell debris. This allows for direct filtration of the culture supernatant without prior centrifugation (Buyel et al., 2015).

After mixing the harvested medium with Celpure®, filtration proceeds rapidly and without clogging, eliminating the need to replace filters. This modification reduced filtration time sixfold, from 60 min to just 10 min, and shortened the total harvest time from approximately 2.5 h to just 15 min.

Importantly, the updated harvesting procedure had no negative impact on antibody yield. As shown in Figure 3A, Western blot analysis confirmed that Celpure® treatment does not interfere with antibody expression or recovery. Under non-reducing conditions, multiple bands were observed, corresponding to different disulfide-bonded conformations, such as intact antibodies, dimers, or partially assembled species. Upon DTT reduction, these resolved into individual heavy and light chains. The detection antibody, specific for full-length antibodies or heavy chains of native mouse IgG, recognised only the heavy chain, resulting in a single band (Figure 3A). To further validate purification, an SDS-PAGE image of the purified IgG was included (Figure 3B). Additionally, total antibody levels remained consistent between Celpure®-treated and untreated samples, confirming that the treatment does not affect antibody yield (Figure 3C).

To evaluate the impact of antibody format on production efficiency, we compared antibody yields across different antibody types purified using HiTrap Protein-G chromatography. As shown in Figure 4A, bispecific antibodies consistently yielded lower amounts of purified antibody compared to conventional IgG constructs, suggesting that antibody design may influence expression levels or purification efficiency.

To address this limitation, we evaluated an alternative purification method using the high-flow HiScreen Fibro PrismA column. As shown in Figure 4B, the bispecific antibody yielded even higher amounts of purified antibody when processed using the HiScreen Fibro PrismA column compared to traditional HiTrap Protein-G. This suggests that the sixfold increased flow rate and improved binding capacity of HiScreen Fibro PrismA can enhance purification efficiency, particularly for more complex antibody formats.

Collectively, these modifications to both upstream and downstream workflows have made the TGE protocol (Figure 1) significantly more resource-efficient, streamlined, and faster, while maintaining high antibody yield, integrity, and transfection efficiency.