CRISPR/Cas9-mediated targeted knock-in of human erythropoietin at the β-casein locus results in lactogenic hormone-responsive expression in HC11 mammary epithelial cells

Abstract

Precise genomic integration strategies are essential for achieving stable and regulated transgene expression. In this study, we established a CRISPR/Cas9-mediated targeted knock-in system to integrate the human erythropoietin (hEPO) gene into the endogenous regulatory region of the mouse β-casein locus in HC11 mammary epithelial cells. A donor vector carrying hEPO was designed for homology-directed repair and successfully introduced into the β-casein locus. A heterozygous single-cell knock-in clone was isolated and validated by genomic analysis. Upon stimulation with lactogenic hormones, the integrated hEPO transgene exhibited hormone-responsive expression, resulting in an approximately 20-fold increase in protein levels compared to non-induced conditions. To enable downstream purification and activity assessment, a GST-tagged hEPO construct was employed. The fusion protein was purified using glutathione affinity chromatography, followed by proteolytic cleavage to obtain recombinant hEPO. The purified protein displayed measurable biological activity with a specific activity of 53.4 mIU/μg. These findings demonstrate that targeted integration at the β-casein locus results in lactogenic hormone-responsive expression driven by endogenous regulatory elements in mammary epithelial cells. This approach provides a controlled gene expression platform that may be applicable to hormone-inducible expression systems for therapeutic protein studies.

Supplementary Information

The online version contains supplementary material available at 10.1038/s41598-026-45234-5.

Introduction

Establishing an optimal expression system for the target gene through precise gene targeting is important in producing recombinant proteins for therapeutic purposes^1,2. In knock-in gene targeting, a transgene is introduced into the specific site of the endogenous gene, leading to the knock-out of the endogenous gene and the integration of a transgene. Subsequently, the transgene is stably expressed by the endogenous gene regulatory region³.

Previous reports have suggested that creating transgenic animals to produce recombinant proteins is a cost-effective and high-quality way to produce proteins⁴. Several prior studies have aimed to obtain recombinant proteins through transgenic animals. Indeed, previous studies have produced transgenic animals expressing human erythropoietin (hEPO) as a protein fused to bovine β-lactoglobulin in milk, reporting hEPO levels of up to 0.3 mg/mL in mice and up to 0.5 mg/mL in rabbits. However, non-specific expressions were observed in various tissues due to random transgene integration⁵. In addition, transgenic mice that produce hEPO in the urine have been generated, with protein levels of 290 pg to 6 ng/mL detected. However, some transgenic lines have been reported to have problems with ectopic expression due to random integration⁶. Several studies have shown that most transgenic animals producing EPO are susceptible to unexpected diseases such as transcriptional and protein leakage⁶. EPO is a growth hormone that is particularly difficult to produce in transgenic animals due to its structural integrity⁷. As such, the problem of positional effects due to random integration of transgenes, gene silencing, and activation of exogenous promoters remains a major hurdle to the production of transgenic animals.

Therefore, targeted integration methods using HR and the CRISPR/Cas9 systems are being utilized to integrate transgenes into specific gene loci. However, intracellular homologous recombination efficiency is low, with HR-mediated knock-in efficiencies of less than 10–20% reported^8,9. In homology-directed repair (HDR), Rad51 and Dmc1 are key recombinases in the HR pathway that facilitate homologous DNA pairing and strand exchange¹⁰. Dmc1 is primarily a recombinase during meiotic recombination, whereas Rad51 functions as the major recombinase in both mitotic and non-mitotic homologous recombination processes¹¹. A previous study reported that a 2- to threefold overexpression of the Rad51 gene in Chinese hamster ovary (CHO) cells led to a 20-fold increase in HR efficiency. Thus, these improvements in efficient gene targeting systems are considered to lead to accurate and targeted integration¹². Despite this, limited research has been conducted on developing a system for stable expression of foreign genes using the CRISPR/Cas9 system to produce recombinant protein⁴.

Milk has been reported as the most suitable source of recombinant proteins for therapeutic purposes in producing transgenic animals^1,13,14 because of the high β-casein protein content and the high expression efficiency of the β-casein promoter^15,16. Most importantly, loss of the β-casein locus has not been reported to affect offspring production, reproduction, or lactation, making it the most suitable locus for gene targeting¹⁷. Additionally, HC11 cells, a mouse mammary gland cell line, have been reported to be an appropriate model system for targeting or modulating the expression of the β-casein locus¹⁸. In addition, a previous study reported that GST-fused CD95 protein was successfully purified using the GST tag system¹⁹. However, the study has not been reported on a system that stably expresses a GST-tagged hEPO transgene at the β-casein gene locus by knock-in gene targeting using the CRISPR/Cas9 system to produce therapeutic proteins.

Therefore, in this study, we aimed to establish a CRISPR/Cas9-mediated targeted knock-in system to integrate a therapeutic gene into the endogenous regulatory region of the mouse β-casein locus in HC11 mammary epithelial cells. First, we generated a heterozygous HC11 mammary epithelial cell line in which a GST-tagged hEPO transgene was inserted into the β-casein gene locus by knock-in mediated the CRISPR/Cas9 system. Using these heterozygous knock-in cells, we analyzed whether the transgene RNA and protein were stably expressed. In addition, after inducing the cells with lactogenic hormones, we verified whether the recombinant hEPO protein expressed under the control of the endogenous regulatory elements of the β-casein gene could be purified using a GST SpinTrap column and whether the fusion protein could be cleaved by the GST tag cleavage system.

Results

The knock-in vector expressing the hEPO gene in the mouse β-casein gene locus

First, we designed a knock-in vector to insert a transgene into a specific gene locus to promote the stable expression of the transgene under the control of the endogenous gene regulatory region to overcome the instability of gene expression caused by random transgene insertion in the cells. The knock-in vector was constructed to express the GST-hEPO fusion protein at the mouse β-casein gene locus, as illustrated in the schematic diagram and nucleotide sequence (Fig. 1 and Fig. S1). The vector consists of a 5’ homology arm (1.024 kb), GST-hEPO fusion gene, SV40 polyadenylation signal (poly-A), PGK-Neo selection marker, and 3’ homology arm (1.015 kb). The poly-A signal was included to stabilize the foreign gene expression. The Neo gene, which lacks a polyadenylation signal, was used as a positive selection marker to facilitate the selection of successfully transfected cells. The knock-in vector was designed with a 3’ homology arm corresponding to the exon 3 region of the mouse β-casein gene locus, aligning precisely with the cleavage site generated by the CRISPR/Cas9 system. If HR occurs at exon 3 in the mouse β-casein gene locus, then the GST-hEPO gene will be expressed under the control of the endogenous β-casein promoter and regulatory sequences. Additionally, the GST-hEPO fusion protein contains a PreScission protease cleavage site, allowing efficient removal of the GST-tag and purification of recombinant hEPO protein.

Fig. 1.

The knock-in strategy in the mouse β-casein gene locus. (a) Genome structure of the β-casein gene locus. (b) Schematic diagram of the knock-in vector used in this study. (c) Schematic diagram of the knock-in vector inserted into the β-casein gene locus by homologous recombination. The knock-in vector is constructed so that when inserted into the β-casein locus, RNA can be expressed from the endogenous gene regulatory DNA sequence in β-casein, and the synthesized hEPO polypeptide is cleaved by Prescission protease.

We used the CRISPR/Cas9 system to precisely insert the knock-in vector into the mouse β-casein gene locus; sgRNAs were designed, and their activity was validated as below. The mouse β-casein gene consists of nine exons and eight introns, with the start codon located in exon 2 and the stop codon in exon 8 (Fig. S2). To evaluate the activity of the designed sgRNAs, HC11 cells were transfected with pGuide-it-ZsGreen1 sgRNA expression vector, which was designed to target each exon of the mouse β-casein gene (Table S1). A T7 endonuclease I assay was conducted following transfection to detect indel mutations. The results demonstrated that sgRNAs targeting exons 3, 4, and 5 induced indel formation in HC11 cells (Fig. S3 and Table S2). Based on these findings, the sgRNA targeting exon 3 in the mouse β-casein gene, which exhibited efficient genome editing activity, was selected for this study.

Production of GST-hEPO knock-in single-cell clones in HC11 cells by CRISPR/Cas9

To establish a single-cell clone expressing the GST-hEPO transgene at the mouse β-casein locus, HC11 cells were transfected with the knock-in vector, followed by G418 selection to isolate targeted clones. To confirm the integration of the knock-in construct, 5’ arm PCR was performed on G418-resistant colonies, revealing eight positive clones with successful vector insertion (Fig. S4). Among these, genomic DNA from colonies No.67, No.68, and No.139 was further analyzed using 3’ arm PCR, which confirmed a 1.4 kb positive band, indicating precise integration at the target locus (Table S3). A schematic diagram illustrating the primers used for screening PCR is provided in Fig. S5. Both targeted and non-targeted PCR analyses were performed to distinguish homologous recombination from random integration. In colonies No.67, No.68, and No.139, a 4.6 kb band was detected, confirming successful homologous recombination at one allele, while a 1.9 kb band was observed, corresponding to the non-recombined allele. Based on these results, colonies No.67, No.68, and No.139 were identified as heterozygous knock-in clones.

Stable expression of hEPO mRNA and protein in the GST-hEPO knock-in cells following lactogenic hormone induction

To confirm the expression of GST-hEPO mRNA in knock-in single-cell clones by the endogenous gene regulatory region, we analyzed the mRNA expression using RT-PCR. The GST-hEPO mRNA expression was detected in colonies No.18, No.23, No.41, No.63, No.67, No.68, and No.139 (Fig. S6A). Furthermore, to evaluate the relative expression levels of the hEPO gene in response to lactogenic hormone induction, qRT-PCR was performed on colonies No.23, No.67, and No.68. The results showed that upon induction, GST-hEPO mRNA expression increased approximately 1.8-fold in colony No.23, 2.5-fold in colony No.67, and 2.6-fold in colony No.68 compared to the non-induced controls (Fig. S6B).

To confirm the expression of hEPO protein, western blot analysis was performed using total protein from knock-in single-cell clones following lactogenic hormone induction in serum-containing media. As shown in Fig. 2A, hEPO protein expression was dramatically increased in both colonies No.67 and No.68 upon lactogenic hormonal induction. Next, we analyzed the total protein from cells and culture supernatant following lactogenic hormone induction in serum-free media. The hEPO protein was not detected in the total protein. However, the expression of hEPO protein in the total protein from culture supernatant was dramatically increased in both colonies No.67 and No.68 (Fig. 2B). The western blot analysis of β-actin protein was performed on proteins extracted from cells, as it is an intracytoplasmic protein, and not on medium samples from cell cultures.

Fig. 2.

GST-hEPO protein was highly expressed upon lactogenic hormone induction in the human EPO knock-in cell lines. (a) GST-hEPO protein expression in total protein extracts from colonies No. 67 and 68 cultured in serum-containing medium. (b) GST-hEPO protein expression in total protein extracted from the cells and culture medium of colonies No. 67 and 68 grown in serum-free conditions. (c) Human EPO ELISA analysis of the from media recovered from colonies No. 67 and 68 before and after lactogenic hormone induction. NS, no statistical difference; *, P < 0.05; **, P < 0.01; ***, P < 0.0001. Original blots are presented in Supplementary Fig. S7.

To further quantify the secreted hEPO protein levels, ELISA analysis was performed on the culture supernatant following lactogenic hormone induction. The hEPO levels in the non-induced group were very low at 8.8 mIU/mL in colony No.67; however, in the induced group, hEPO levels increased 20-fold to 171.9 mIU/mL (Fig. 2C). Similarly, in colony No.68, the hEPO levels were 11.6 mIU/mL in the non-induced condition. In contrast, lactogenic hormone increased the expression level to 85.9 mIU/mL, representing a tenfold increase (Fig. 2C).

Purification and specific activity of recombinant hEPO protein in the knock-in cell clone

To determine whether GST-tagged hEPO protein could be purified, a GST affinity column was employed to purify GST-tagged hEPO protein secreted from a GST-hEPO knock-in HC11 cell clone, and the GST tag was cleaved to obtain hEPO protein. As shown in Table 1, the specific activity of GST-cleaved hEPO protein was 53.4 mIU/μg and yielded approximately 10%. GST-cleaved hEPO protein was analyzed by Western blotting to confirm the cleavage of the GST tag, and GST-tagged hEPO protein was used as a control for accuracy (Fig. 3A and B). As a result of Western blotting using hEPO antibody, GST-tagged hEPO protein was observed at 55–75 kDa in the GST-hEPO sample, and the GST-cleaved hEPO protein was observed at 35–40 kDa in the cleaved hEPO sample (Fig. 3A). To confirm GST tag cleavage, GST-tagged hEPO protein was used as a control. The cleaved GST tag was detected by Western blotting. As a result of Western blotting using the GST antibody, the GST-tagged hEPO protein was observed at 55–75 kDa in the GST-hEPO sample; meanwhile, the cleaved GST from the GST-hEPO protein was not observed at the 26 kDa (Fig. 3B). However, GST protein in the cleaved hEPO sample was detected at 42 kDa, which is similar to the PreScission protease molecular weight (Fig. 3B). The specific activity of pre-purification GST-tagged hEPO and purified GST-cleaved hEPO protein, measured by ELISA (Table 1), were then converted from mIU/μg units to pg/μg units and presented in a graph (Fig. 3C). As a result, 412.4 pg hEPO protein per μg was valued (Fig. 3C).

Table 1.

Purification and specific activity of recombinant hEPO protein by GST Spin Trap column and PreScission protease.

Purification step	Volume (mL)	Total protein (μg)	Total activity (mIU)	Specific activity (mIU/μg)	Purification (fold)	Yield (%)
Crude lysate	1.5	547.1	1189.6	2.2	1	100
GST-purification and cleavage	0.2	2.2	117.5	53.4	24.6	9.9

Fig. 3.

Analysis of purified GST-hEPO and purified hEPO cleaved by Prescission protease using Western blot and ELISA. (a) Western immunoblotting using hEPO antibody. (b) Western immunoblotting using GST antibody. (c) Graph showing conversion of mIU/ug to pg/μg by ELISA assay. The hEPO concentration is expressed using a conversion factor of 1 mIU = 7.86 pg (Thermo Fisher Scientific). Original blots are presented in Supplementary Fig. S8.

Discussion

This study establishes a stable expression system for transgenes by controlling the endogenous gene regulatory region of the β-casein gene, and further demonstrates that the recombinant hEPO protein produced from this system can be efficiently purified using a GST SpinTrap column followed by GST tag cleavage.

As mentioned in the introduction, an optimal transgene expression system that ensures stable and high-yield protein expression is required for efficient recombinant protein production for therapeutic purposes. Traditional transgenic animal models have demonstrated the feasibility of producing recombinant hEPO; however, issues such as ectopic expression, low glycosylation efficiency, and unpredictable insertion effects are major challenges^5,6. Therefore, precise gene targeting methods, such as CRISPR/Cas9-mediated knock-in, provide a promising alternative to random transgene integration, allowing for controlled expression under endogenous gene regulatory regions.

Previous studies have shown that random transgene integration often causes variable expression due to position effects and can lead to transcriptional repression, overexpression, or ectopic expression, sometimes resulting in physiological abnormalities in transgenic animals^20,21. For example, Park et al. reported high hEPO levels in the milk of transgenic pigs generated by microinjection²², and Kim et al. observed EPO secretion in transgenic mice⁷; however, both studies relied on exogenous promoters and random insertion. In contrast, in this study, the GST-hEPO transgene was precisely inserted into the β-casein locus, allowing its expression to be driven by endogenous β-casein regulatory elements. Lactogenic hormone induction further confirmed that the inserted transgene responds appropriately to β-casein gene regulatory sequence. We believe that because the β-casein signal sequence is conserved in-frame, the resulting GST-hEPO protein is efficiently directed to the ER-Golgi pathway, where essential post-translational modifications occur. These results suggest that this knock-in system can support more stable and predictable transgene expression than conventional random integration and may provide a useful platform for producing therapeutic proteins.

Nevertheless, previous studies have demonstrated that CRISPR/Cas9-mediated knock-in events typically yield a higher frequency of heterozygous clones than homozygous clones, with the latter being particularly difficult to isolate^23,24. Accordingly, the primary objective of this study was to verify whether the inserted transgene could be effectively expressed through the endogenous gene regulatory sequence within the knock-in locus. In light of the challenges associated with generating homozygous knock-in clones, subsequent analyses including RNA and protein expression profiling as well as protein purification were performed using heterozygous knock-in cells. Also, CHO and HEK293 cells are widely recognized as the dominant host systems for the industrial-scale production of therapeutic proteins, largely due to their stability and performance in long-term and high-density bioreactor cultures^25–27. These cell lines provide well-established manufacturing platforms with predictable growth kinetics, robust productivity, and regulatory precedence. In the present study, however, HC11 mammary epithelial cells were selected to evaluate the stability of transgene expression following CRISPR/Cas9-mediated knock-in. Unlike CHO or HEK293 cells, HC11 cells possess a hormone-responsive differentiation program that can be precisely induced by lactogenic hormones. This unique biological property allows transgenes inserted into the β-casein locus to be regulated by endogenous mammary gland–specific promoter and enhancer elements. Thus, HC11 cells serve as an advantageous model for validating hormonally controlled, locus-specific expression systems that more closely mimic the regulatory environment of the mammary gland in vivo. Although murine mammary epithelial cell lines, including HC11, may exhibit glycosylation patterns that differ from human or CHO cells raising potential concerns regarding therapeutic protein quality this limitation should be considered within the broader context of intended applications. The knock-in strategy demonstrated here is directly relevant to the development of transgenic livestock engineered for the secretion of recombinant proteins in milk, an approach that has been successfully utilized for large-scale, cost-effective protein production. Therefore, the HC11-based system provides not only a mechanistic model for studying β-casein locus regulation but also a translational framework that can support the future generation of high-yield in vivo bioreactors.

Also, during this study, three heterozygous knock-in clones (No. 67, No. 68, and No. 139) were initially established. Notably, clone #139 was excluded from subsequent analyses due to pronounced variability in RNA recovery across replicate experiments. This observation likely reflects underlying clonal heterogeneity, highlighting that even among clones derived from the same knock-in event, differences in transcriptional activity or cellular state can influence experimental reproducibility. Such variability underscores the importance of evaluating multiple independent clones when assessing the stability and consistency of transgene expression. Additionally, future research is needed to elucidate the mechanisms underlying the differences in gene expression due to clonal heterogeneity in knock-in cells.

ELISA analysis was performed in this study to quantify recombinant hEPO secretion in knock-in cells after lactogenic hormone induction, revealing 171.9 mIU/mL in colony No.67 and 85.9 mIU in colony No.68. Given that 1 mIU/mL of hEPO corresponds to 7.86 pg/mL, this translated to 1351.1 pg/mL and 675.2 pg/mL, respectively. These results indicated that lactogenic hormone induction effectively expressed and secreted the hEPO protein in the GST-hEPO knock-in HC11 cell clones. Although this is approximately 1000-fold lower than the 200 IU/mL EPO production in CHO cells, CHO-derived EPO is known to have lower glycosylation efficiency, potentially affecting its activity. In this study, ELISA analysis indicated that hEPO expression from clone No. 67 was approximately twice that of clone No. 68. Both clones were derived from heterozygous knock-in cells generated using the CRISPR/Cas9 system. It is well documented that knock-in approaches employing CRISPR/Cas9 can introduce large deletions, insertions, and complex genomic rearrangements at the target locus²⁸. In addition, previous studies have demonstrated that inherent heterogeneity within the parental cell line including wild-type variability and clonal heterogeneity can substantially influence transgene expression and cellular phenotype²⁹. Factors such as differences in gene copy number, chromosomal context, and epigenetic state among individual clones can result in divergent expression profiles even when identical genome editing procedures are applied. Taken together, the observed two-fold difference in hEPO secretion between clones No. 67 and No. 68 is likely attributable to clonal heterogeneity. It is important to note that, in the present study, detailed analyses of potential large deletions, insertions, complex rearrangements, gene copy number, chromosomal state, or epigenetic modifications around the target gene were not performed for each clone. Future investigations addressing these genomic and epigenetic parameters will be essential to fully elucidate the mechanisms underlying clone-to-clone variability in transgene expression.

In this study, protein expression and purification were conducted 72 h after lactogenic hormone induction. Previous reports have shown that protein expression in HC11 cells is commonly examined at multiple time points (typically 24, 48, and 72 h following lactogenic stimulation) to capture the dynamic progression of hormone-responsive gene activation³⁰. However, in this experiment, we focused on protein purification and concentrated the secreted protein, so protein purification was performed 72 h after lactogenic hormone treatment. However, determining the optimal induction time for transgene expression will require a more comprehensive temporal analysis. In particular, once homozygous knock-in cells become available, detailed time-course studies will be essential to evaluate expression kinetics and identify the most efficient time window for maximal protein production. Typically, lactogenic stimulation triggers a transition from the growth phase to a differentiation state after EGF withdrawal, leading to reduced proliferation and increased expression of milk-associated proteins such as β-casein³⁰. Consistent with these established characteristics, western blot analysis of total protein (Fig. 2A) showed that hEPO expression was detected only in hormone-induced cells. These results indicate that the lactogenic hormone induction protocol used in this experiment functioned as expected.

In addition, although this study focused on validating the initial establishment of the knock-in cell line, the long-term stability of transgene expression remains an important point for further investigation. In principle, fluctuations in the expression of stably integrated genes can emerge after extended passaging; therefore, assessing transcript and protein levels beyond 20 passages using qRT-PCR and ELISA would provide more definitive evidence of stability. Nevertheless, previous studies offer supportive context. It has been reported that HC11-derived stable cell lines maintain long-term transgene expression³¹. Similarly, Jeong et al.³² reported sustained RNA and protein expression when FGF2 was inserted into the β-casein locus of MAC-T bovine mammary epithelial cells³². These findings suggest that mammary epithelial-derived knock-in systems generally support stable foreign gene expression. Therefore, it is reasonable to expect that the knock-in cell line generated in this study will likewise preserve transgene expression over extended passaging. However, direct experimental verification will be essential to conclusively establish long-term stability.

Also, in this study, heterozygous knock-in clones exhibited an approximately 2.6-fold increase in transgene mRNA levels, whereas the corresponding protein levels measured by ELISA increased by nearly 10–20 fold. Such discrepancies between transcript abundance and protein output are well documented, as mRNA levels do not necessarily correlate with protein expression due to multiple layers of post-transcriptional regulation. In particular, enhanced translational efficiency, intracellular protein accumulation, and efficient secretion can collectively amplify protein abundance to a greater extent than changes observed at the RNA level³³.

Furthermore, this study used a tag system to induce stable expression of the hEPO protein. According to a previous study report, mice expressing EPO fused with BLG protein showed higher EPO protein expression levels and significantly reduced erythrocytosis side effects. Importantly, it was previously reported that fusion protein BLG cleavage restores EPO protein activity⁵. Thus, the tag system was reported to play an important role in protein stability. This study performed Western immunoblotting on GST-tagged hEPO protein and GST-cleaved hEPO protein using PreScission protease; the hEPO protein was identified at 35–40 kDa. These results suggest that the PreScission protease protein used in this study is a GST fusion protein to human rhinovirus (HRV) type 14 3C protease, so it was not completely removed from the cleaved hEPO protein, and the GST antibody was used to detect the PreScission protease.

Glycosylation is essential for the in vivo activity of therapeutic EPO³⁴. In this study, we did not analyze the glycosylation status of hEPO purified using the GST SpinTrap column. HC11 cells share a basic mammalian glycosylation framework with human cells. However, because they are mouse-derived, they can produce Neu5Gc as well as Neu5Ac. Their overall glycosylation efficiency is also lower than that of CHO cells³⁵. HC11 cells show additional changes upon lactogenic differentiation. They increase intracellular glycoprotein synthesis. They also show a shift toward more complex-type N-glycans³⁶. The capacity for α(2,6)-linked sialic acid synthesis is limited in mouse cells, but this limitation can be overcome by inducing the expression of ST6Gal sialyltransferases (ST6Gal1 and ST6Gal2)³⁷. This strategy may support the production of recombinant proteins with more human-like glycosylation, and its feasibility should be evaluated in future experiments. These changes suggest that the glycosylation pattern of HC11-derived hEPO may vary depending on the cellular state. Therefore, further research is needed to precisely analyze the glycosylation pattern of the HC11-derived EPO recombinant protein obtained in this study. This analysis will clarify the structural and functional properties of the hEPO produced in this system. It will also be important for evaluating EPO secreted from milk in transgenic livestock generated using the same knock-in strategy.

In addition, the system developed in this study is expected to improve the yield of recombinant proteins through the production of transgenic animals. However, transgenic animal production is a breakthrough method that is superior to using mammalian cells in terms of cost and protein quality^5,7. Therefore, it is considered that solving the ectopic problem of transgenic animals producing hEPO protein in breast milk under the endogenous regulatory region of β-casein will improve not only EPO protein expression. The endogenous EPO protein size in glycated urine reported in a previous study was 35–38 kDa, similar to the hEPO protein size measured in this study. The specific activity of the partially purified hEPO obtained using the GST SpinTrap column was 53.4 mIU/µg, as determined by ELISA. This value is markedly lower than the specific activity reported for CHO-derived hEPO³⁸. This reduction is likely due to the small-scale, partial purification performed in this study, which was intended only to evaluate the feasibility of GST-based purification. For accurate activity assessment, large-scale production using homozygous knock-in cells, followed by high-purity purification methods such as HPLC, will be required. A more precise determination of the specific activity should be obtained using such purified preparations. In addition, functional validation using EPO-responsive cell lines, such as TF-1, and mass spectrometry (MS) will be necessary to confirm the biological activity of the recombinant hEPO. However, the results confirmed in this study indicated that recombinant hEPO protein was expressed and purified at the β-casein gene locus in the mammary epithelial cell line HC11 through genetic knock-in using the CRISPR/Cas9 system.

Although the present study primarily focused on validating locus-specific genomic integration and lactogenic hormone-responsive expression following CRISPR/Cas9-mediated knock-in, additional characterization would further support the broader applicability of this system. In particular, analyses such as detailed glycosylation profiling, expanded functional bioassays, and assessment of production scalability may provide further insights for future therapeutic or industrial applications. These aspects were beyond the scope of the present proof-of-concept study but represent important directions for future research.

Conclusion

In conclusion, we demonstrated that the knock-in transgene within the endogenous regulatory region in β-casein of the mouse mammary gland cell line HC11 was stably expressed. To demonstrate the stable expression system of gene targeting developed in this study, we targeted the hEPO recombinant protein. Finally, we confirmed that the recombinant hEPO protein could be partially purified by the knock-in system used in this study. In addition, although this study obtained heterozygous knock-in cells, the establishment of homozygous knock-in cells in future studies may further enhance transgene expression and improve the productivity of recombinant proteins. If transgenic animals are developed using this system, then this system can significantly contribute to advancing recombinant protein production research. However, additional analyses, including detailed glycosylation characterization and functional validation of the recombinant hEPO, will be necessary to further evaluate the structural and biological properties of proteins produced using this system.

Materials and methods

Knock-in vector design and construction

The knock-in vector used in this study was the same as that described in a previous study and was designed to express the hEPO protein through the GST fusion system at exon 3 in the mouse β-casein gene locus. Initially developed in our lab, the vector contained a 5’ homology arm (1.024 kb), GST-hEPO gene, SV40 polyA, CMV-EGFP, and 3’ homology arm (1.81 kb). To modify the vector, CMV-EGFP was replaced with PGK-Neo as a negative selection marker, and the 3’ homology arm was shortened to 1.015 kb, resulting in the final construct pBSK_1.024 kb_mβC5’arm_GST_hEPO_SV40 polyA_PGK Neo_1.015 kb mβC 3’ arm. Detailed vector sequences are provided in Fig. S1.

Design of sgRNA to target the mouse β-casein gene locus using CRISPR/Cas9

In this study, the sgRNA targeting the β-casein gene locus was the same as that used in a previous study. The sgRNAs were designed based on the nucleotide sequence of the mouse β-casein gene obtained from NCBI (http://www.ncbi.nlm.nih.gov) and designed using the RGEN Tools program (http://www.rgenome.net/cas-designer/). The sgRNA sequence was constructed according to the pGuide-it™ CRISPR/Cas9 system (Takara, Tokyo, Japan) manual, and the oligo was cloned into the pGuide-it-Zs Green1 vector. Moreover, this study used a sgRNA targeting exon 3. The detailed sgRNA sequence and locus are attached as Fig. S2 and Table S1.

Cell culture, transfection, and induction

Mouse mammary epithelial cells (HC11) were cultured in a growth medium consisting of Roswell Park Memorial Institute (RPMI)-1640 medium (Hyclone, Logan, UT, USA) supplemented with 10% fetal bovine serum (FBS, Hyclone), 1% penicillin/streptomycin (Hyclone), 0.1% gentamicin (Sigma, St. Louis, MO, USA), 0.1% epidermal growth factor (EGF) (Gibco, Thermo Fisher, USA), and 0.05% insulin (Sigma). HC11 cells were seeded at a density of 2.75 × 10⁵ cells per well into 6-well plates (SPL) and incubated at 37 °C with 5% CO₂. The cells were co-transfected with PGK-Neo knock-in vector (3 μg), pGuide_it_ZsGreen1_sgRNA (1.5 μg), and overexpression vectors for the DNA DSB repair-related mDmc1 and mRad51 genes, which were expressed under the CMV promoter (1.5 μg), using Xfect transfection reagent (Takara) according to the manufacturer’s instructions. After 24 h of transfection, cells were washed with EDTA–PBS and treated with 0.25% trypsin–EDTA. The cells were resuspended in a culture medium and seeded at a density of approximately 500 cells per well in a 48-well plate (SPL). The cells were selected in the same medium containing 300 μg/mL G418 (Gibco-BRL, Waltham, USA), and the selected clones were plated and cultured at a density of 5.5 × 10⁵ cells per 6 cm dish (SPL) or 10 × 10⁵ cells per 10 cm dish (SPL) with 37 °C and 5% CO₂. The cells were cultured for 72 h for lactogenic hormone induction and washed with PBS to remove FBS completely. Lactogenic hormone induction was performed using the following culture medium. The induction medium consisted of RPMI-1640 (Hyclone), 1% penicillin/streptomycin (Hyclone), 0.01% gentamicin (Sigma), 0.05% insulin (Sigma), 500 μg/mL prolactin (Sigma), and 10 mg/mL dexamethasone (Sigma). Induction was performed at 37 °C with 5% CO₂ for 72 h, and the supernatant was collected after 72 h for analysis and purification.

Analysis of knock-in integration by PCR

To confirm the single-cell knock-in integration of the PGK-Neo knock-in vector in HC11 cells, a polymerase chain reaction (PCR) was performed as described below. Genomic DNA was extracted from the transfected cells using the G-DEX IIc Genomic DNA Extraction kit (for cell/tissue) (Intron, Seongnam-si, Korea) according to the manufacturer’s instructions. The PCR screening was performed using primers targeting the 5’ arm, 3’ arm, and targeted and non-targeted regions to verify the knock-in integration. The PCR primer sequences used in this experiment are presented in Table S4. The 5’ arm PCR was performed with an mβC5 CV S5 primer and mβC3 GST AS9 primer using Solg™ 2X Multiplex (Solgent) as follows: 33 cycles of amplification at 95 °C for 20 s, 62 °C for 40 s, and 72 °C for 2 min. The 3’ arm PCR was performed with a Neo 3–4 sense primer and mβC3’ arm AS1 primer using Solg™ 2X Multiplex (Solgent) and the same conditions that were used for the 5’ arm screening PCR. To determine the homozygous and heterozygous knock-in statuses, targeted and non-targeted PCR was performed with a mβC5 CV S5 primer and mβC3’ arm AS2 primer using KOD Fx Neo (Toyobo) as follows: 5 cycles of amplification at 94 °C for 15 s, 74 °C for 5 min, 5 cycles of amplification at 94 °C for 15 s, 72 °C for 5 min, 5 cycles of amplification at 94 °C for 15 s, 70 °C for 5 min, 20 cycles of amplification at 94 °C for 15 s, 68 °C for 5 min, and 68 °C for 7 min. The PCR fragments were analyzed by electrophoresis on a 0.8% agarose gel.

RT-qPCR analysis

Total RNA was isolated from HC11 cells using the RNeasy Mini kit (QIAGEN, Hilden, Germany), according to the manufacturer’s instructions. For cDNA synthesis, 3 μg of total RNA was used as a template. RT-PCR was performed using random primers and Superscript II reverse transcriptase (Invitrogen, Carlsbad, CA, USA) according to the manufacturer’s protocol. To confirm the GST-hEPO transgene expression in knock-in single-cell clones, RT-qPCR was performed using primer pairs (Table S4) with Solg™ 2X Multiplex (Solgent) under the following conditions: 33 cycles of amplification at 95 °C for 20 s, 64 °C for 40 s, 72 °C for 1 min. The PCR fragments were confirmed by electrophoresis on a 2% agarose gel. Quantitative real-time PCR was performed to measure the GST-hEPO transgene mRNA expression in a total volume of 20 μL, containing 20 ng cDNA, 10 pmol of primer pairs (Table S4), and 10 μL of TOPreal™ qPCR 2X PreMix (Enzynomics, Daejeon, Korea). All reactions were repeated three times and were conducted on Mx3000P (Agilent Technologies, Santa Clara, CA, USA) under the following conditions: denaturation at 95 °C for 10 min, followed by 40 cycles of amplification at 95 °C for 10 s, annealing at 60 °C for 15 s, and strand extension at 72 °C for 15 s. For comparative quantification, mRNA expression levels of each gene were normalized to the mouse Rplp0 primer pairs (Table S4).

Western blotting

To confirm the protein expression of hEPO, the protein was extracted from single-cell clones harboring the knock-in vector using PRO-PREP™ protein extraction solution (Intron, Korea). For protein extraction from the culture medium of lactogenic hormone-induced single-cell clones, the medium was centrifuged at 1000 rpm for 5 min. Then, 250 μL of 100% trichloroacetic acid (TCA) was added to 1 mL of the collected supernatant, and the mixture was incubated on ice for 10 min. Following incubation, the sample was centrifuged at 13,000 rpm and 4 °C for 20 min, and the supernatant was carefully removed. The remaining protein pellet was washed with 200 μL of 100% acetone to remove residual TCA. The sample was incubated at 95 °C for 5 min to evaporate any remaining acetone. Finally, the pellet was resuspended in 30 μL of alkaline SDS-PAGE sample buffer to extract the protein. Immunoblotting was performed to analyze the GST tag cleavage in GST-fused hEPO, and the full-volume samples were loaded and separated on a 10% SDS-PAGE gel for 1.5 h. The loaded proteins were transferred from the gel to a polyvinylidene difluoride membrane (Bio-Rad) at 80 V, 2 h, in Tris–glycine buffer. The membrane was then blocked in 5% non-fat milk or BSA in TBST (Tris-buffered saline containing 0.1% Tween 20) at room temperature for 2 h and then blotted with mouse monoclonal anti-EPO antibody (sc-5290, Santa Cruz Biotechnology; dilution 1:1000), anti-GST antibody (91G1, Cell Signaling Technology; dilution 1:2000) at 4 °C overnight. After overnight incubation, the membranes were washed with TBST and incubated with horseradish peroxidase (HRP)-conjugated secondary antibodies, mouse immunoglobulin G (HRP) kappa binding protein (m-IgGk BP)-conjugated (sc516102, Santa Cruz Biotechnology; dilution 1:1000), and goat anti-rabbit IgG H&L (ab205718, Abcam; dilution: 1000) at room temperature for 2 h. The membranes were washed with TBST. The hEPO bands were detected using the EZ-Western Lumi Pico kit and EZ-Western Lumi Femto kit (Dogen, Seoul, Korea).

Purification of GST-tagged hEPO protein

The GST-tagged hEPO protein was purified using a GST Spin Trap column (GE Healthcare Life Sciences) with a capacity of 500 μg for GST proteins according to the manufacturer’s instructions. The column was equilibrated by adding pH 7.4 and phosphate-buffered saline (PBS) at room temperature. To bind GST-hEPO protein to the column, 20 mL of the supernatant was concentrated 12-fold with an Amicon® Ultra Centrifugal Filter (Millipore) on ice. Then, 500 μL of the concentrated sample was applied to a GST Spin Trap column (GE Healthcare Life Sciences) and gently inverted on ice for 10 min, performed approximately 4 times. After three washes with PBS buffer, 100 μL of an elution buffer containing 10 mM reduced glutathione was applied to the column, incubated for 10 min, and centrifuged twice at 100 × g for 30 s. To isolate the recombinant hEPO protein, GST cleavage was performed with PreScission protease (GenScript Biotechnology, Nanjing, China) according to the manufacturer’s on-column manual of the GST Spin Trap column (GE Healthcare Life Sciences). The spin column was loaded with 600 μL PreScission protease buffer and washed twice by centrifugation at 100 × g for 1 min. Then, 200 μL of 50 mM Tris (pH 8.0), 150 mM NaCl, 1 mM EDTA, 1 mM DTT, and 2 units of PreScission protease were added to the column and incubated at 4 °C for 16 h. Afterward, pure hEPO protein was obtained by centrifugation at 100 × g for 1 min.

ELISA analysis

To analyze the concentration of human EPO protein, ELISA (enzyme-linked immunosorbent assay) analysis was performed using the Human EPO ELISA kit (Invitrogen) according to the manufacturer’s protocol. The sample was diluted by 1/4 or 10 and used for the analysis. The activities of purified GST-hEPO and cleavage EPO proteins were measured at 450 nm using a Synergy™ HTX Multi-mode microplate reader (Bio Tek, Winooski, USA). The analytical method for this result expressed EPO concentrations in mIU/mL, which were converted to pg/mL using the conversion factor provided by the manufacturer (1 mIU/mL = 7.86 pg/mL) (Thermo Fisher Scientific, BMS2035).

Statistical analysis

Densitometric quantification of DNA and protein bands was performed using UN-SCAN-IT gel Analysis Software (Silk Scientific Inc., Orem, UT, USA). Statistical analysis of mRNA levels and densitometric quantification of DNA and protein bands were conducted using GraphPad Prism 5 (GraphPad Software Inc., San Diego, CA, USA). Data analysis was performed using two statistical approaches. First, one-way ANOVA, followed by Dunnett’s test, was used to compare all experimental groups against the control group. Second, a t-test was applied to compare differences between two specific groups. A 95% confidence interval was used for all statistical analyses.

Supplementary Information

Below is the link to the electronic supplementary material.

Author contributions

Hyeong-Ju You and Ga-Yeon Kim contributed equally to this work. M.K. conceived the project. H.Y. performed all the bioinformatic analyses and drafted the manuscript. H.Y. and G.K. prepared samples and performed the experiments. H.Y., G.K. and M.K. revised the manuscript. All authors read and approved the final manuscript.

Funding

This work was supported by a National Research Foundation of Korea (NRF) grant awarded by the Korean government (MSIT) (RS-2024-00343478).

Data availability

The datasets generated and/or analyzed during the current study are available in the GenBank repository, [Accession numbers: PV992698]

Declarations

Competing interests

The authors declare no competing interests