Base Editing Pinpoints Exon 7 Knockout for High-Yield CHO Antibody Production

Chinese hamster ovary (CHO) cells dominate the manufacturing of recombinant therapeutic proteins, and the glutamine synthetase (GS) selection system is one of the most widely used tools for building production cell lines. Yet two commercial GS-deficient CHO platforms, ATCC CHO-K1 and Merck’s CHOZN, behave very differently in the hands of cell-line engineers, and researchers have now traced part of that divergence to a surprisingly simple variable: exactly where along the GS gene a premature stop codon lands. A new study published in Bioresources and Bioprocessing systematically introduced stop mutations into each of the six coding exons of the Glul gene using cytosine base editing and found that the position of the premature termination codon (PTC) strongly determines whether the knockout is complete, how the cells respond to stress, and how much antibody they ultimately produce.

The two commercial lines carry different frameshift mutations in Glul. ATCC CHO-K1, the first subclone derived from parental CHO-ori cells, harbors a 26-base-pair deletion in exon 2, while CHOZN cells were generated with zinc finger nucleases that created a 10-base-pair deletion and a 1-base-pair substitution in exon 6. Both mutations introduce frameshifts, and both lines show little or no detectable GS protein by Western blot. But when the researchers measured GS messenger RNA, they found substantial residual transcripts in both lines, showing that the frameshifts do not abolish transcription. The apparent loss of GS protein reflects a post-transcriptional block rather than transcript elimination, a distinction the authors argue should caution anyone treating residual mRNA as a proxy for residual enzyme activity.

The behavioral contrast between the two platforms was striking. In glutamine-free medium, CHOZN cells were markedly more sensitive to glutamine deprivation, while ATCC CHO-K1 cells sustained relatively active proliferation even without the amino acid. Under methionine sulfoximine (MSX) selection, the inhibitor used in GS-based screening, both lines proliferated more slowly. Profiling of stress-response markers, including Atf4, Ddit3, and Ppp1r15a, revealed that CHOZN mounted a substantially amplified integrated stress response under deprivation, co-inducing residual GS transcripts with the stress genes, whereas ATCC CHO-K1 was essentially unresponsive, with neither GS transcript nor stress markers changing appreciably with glutamine status. The authors suggest this divergent stress engagement, rather than residual transcript abundance itself, may contribute to the different glutamine sensitivities.

To test whether PTC position matters in a controlled setting, the team turned to a cytosine base editor, CBE4max-SpRY, which converts cytosine to thymine without creating double-strand DNA breaks. This matters because conventional nuclease editing with zinc finger nucleases, TALENs, or CRISPR/Cas9 relies on error-prone repair of double-strand breaks, producing heterogeneous mixtures of frameshift and in-frame alleles. In benchmarking experiments at exon 6, CRISPR/Cas9 achieved indel frequencies of up to 64 percent, three of seven TALEN pairs showed detectable activity with a maximum of 25.7 percent, and the ZFN pair reached 9.4 percent, but all of these double-strand-break strategies also generated in-frame indels at frequencies of 0.6 to 15.3 percent, edits that may fail to disrupt the reading frame at all. Base editing, by contrast, installs the intended stop codon directly, so the measured editing frequency reflects the desired outcome.

Using sgRNAs designed against each coding exon, the researchers generated monoclonal lines carrying PTCs at six positions: Q27X in exon 2, W60X in exon 3, Q135X in exon 4, Q201X in exon 5, Q205X in exon 6, and R286X in exon 7. Residual GS transcript levels relative to the wild-type EK1 control ranged widely, from roughly 0.7-fold for the exon 2 line down to 0.03-fold for exon 4 and 0.04-fold for exon 7, with the exon 6 line at 0.39-fold. When the team depleted UPF1, a core component of the nonsense-mediated mRNA decay pathway, GS transcripts in the exon 4 and exon 7 lines rebounded dramatically, by approximately 19.2-fold and 12.8-fold respectively, providing functional evidence that UPF1-dependent decay is a major driver of the near-complete transcript depletion in those lines. Responses in the other lines were far weaker or absent.

Protein measurements added a further layer of nuance. GS protein was undetectable in every PTC-edited line except the exon 6 line, despite substantial residual transcripts in several lines, particularly those edited at exons 2 and 3. The exon 6 line, uniquely, retained detectable GS protein and was the only edited line besides the wild-type parent that could proliferate in glutamine-free medium during serial-passage cultivation across a glutamine gradient of 0, 1, 2, and 6 millimolar. The authors conclude that transcript abundance alone is insufficient to predict knockout completeness, and that GS knockout cannot be regarded as a uniform binary state.

The exon 7-edited line, designated E7, emerged as the standout. It showed complete functional GS deficiency, no detectable GS protein, and sustained high viability, retaining more than 90 percent viability throughout cultivation in glutamine-replete medium even though it grew more slowly and reached a lower peak viable cell density than the other edited lines. Because the exon 7 sgRNA sequence is fully conserved in the homologous GS1 locus on chromosome 1, the researchers also checked whether GS1 editing might explain the phenotype. Among three independent E7 clones, GS1 was homozygously edited in one, heterozygously edited in another, and unedited in the third, yet GS protein was undetectable in all three, and the clone with unedited GS1 showed growth and viability profiles comparable to its siblings. GS1, currently annotated as a noncoding pseudogene, showed no compensatory upregulation in most lines, leading the authors to conclude that GS1 editing is unlikely to be the principal determinant of the E7 phenotype, though they note a subtle contribution cannot be completely excluded.

The production test used an anti-PD-1 monoclonal antibody in 16-day fed-batch cultures of stable pools. Without MSX selection, the E7 pool yielded a final titer of 0.74 grams per liter at a cell-specific productivity of 3.7 picograms per cell per day, more than three times the productivity of the CHOZN control pool, which reached 0.39 grams per liter at 1.0 picogram per cell per day. With 5 micromolar MSX, the E7 titer rose 38 percent to 1.02 grams per liter with a productivity of 4.9 picograms per cell per day, 4.45-fold higher than the MSX-treated CHOZN pool. MSX produced no comparable improvement in the EK1 or CHOZN pools, which the authors attribute to host-dependent differences in selection stringency and pool enrichment rather than a productivity ceiling. All pools maintained high viability between 86 and 93 percent, with glucose between 1.06 and 5.02 grams per liter and lactate at basal levels of 0.00 to 1.09 grams per liter, indicating balanced metabolic profiles across the engineered lines.

Transcriptomic analysis of the non-producing cell lines revealed graded remodeling: the ATCC CHO-K1 line showed the most extensive differential expression relative to the wild-type control, with 1,536 differentially expressed genes, while CHOZN was closest to the control with 304, and E7 was intermediate with 608. E7 showed marked downregulation of glutamate-ammonia ligase activity genes and nitrogen-assimilation pathways, a molecular hallmark of complete GS ablation that neither commercial line displayed. Compared with the exon 4 line, E7 also showed lower expression of genes involved in glutamine transport, nitrogen metabolism, proliferation, and basal ER-stress markers. The authors are careful to frame these as associations: the data do not establish that the expression changes were caused by the R286X position or that they drove the productivity advantage, and mechanisms such as transgene copy number and clonal composition were not directly examined.

The study’s limitations are clearly stated. Most exon-specific comparisons used representative monoclonal lines, several growth and fed-batch experiments were single standardized runs on an industrial CDMO platform rather than biological replicates, and the production results require validation through independent fed-batch and clone-level studies. Targeted off-target sequencing of 13 predicted sites found editing only at the GS1 locus, RNA-seq-based variant analysis showed no significant increase in expressed SNPs or indels, and PCR confirmed elimination of residual editor plasmid sequences from all clonal lines, but the authors acknowledge that whole-genome sequencing would be needed to exclude rare variants. Within those bounds, the work delivers a defined, exon-resolved strategy: for laboratories building GS-deficient CHO hosts, the position of the stop codon is not a detail but a design parameter, and exon 7 stands out as a target that combines complete knockout with the manufacturing phenotype that cell-line developers prize.

Subject of Research: Exon-specific base editing of glutamine synthetase in CHO cells to create a high-productivity antibody manufacturing cell line

Article Title: Base editing-mediated exon-specific knockout of glutamine synthetase generates a high-productivity CHO cell line for antibody manufacturing

Article References: Pu, T., Wang, C., Lou, H., Mi, Y., Zhang, S., Liu, S., Xiao, Z., Wang, Q., Liu, Q., & Liu, Y. (2026). Base editing-mediated exon-specific knockout of glutamine synthetase generates a high-productivity CHO cell line for antibody manufacturing. Bioresources and Bioprocessing, 13(1), Article 156. https://doi.org/10.1186/s40643-026-01150-0

Image Credits: AI Generated

DOI: 10.1186/s40643-026-01150-0

Keywords: CHO cells, glutamine synthetase, base editing, premature termination codon, cell line development, monoclonal antibody, fed-batch culture, nonsense-mediated mRNA decay, glutamine auxotrophy, recombinant protein production, MSX selection, Bioresources and Bioprocessing