RETRACTED ARTICLE: Impact of different promoters, promoter mutation, and an enhancer on recombinant protein expression in CHO cells

In the present study, six commonly used promoters, including cytomegalovirus major immediate-early (CMV), the CMV enhancer fused to the chicken beta-actin promoter (CAG), human elongation factor-1α (HEF-1α), mouse cytomegalovirus (mouse CMV), Chinese hamster elongation factor-1α (CHEF-1α), and phosphoglycerate kinase (PGK), a CMV promoter mutant and a CAG enhancer, were evaluated to determine their effects on transgene expression and stability in transfected CHO cells. The promoters and enhancer were cloned or synthesized, and mutation at C-404 in the CMV promoter was generated; then all elements were transfected into CHO cells. Stably transfected CHO cells were identified via screening under the selection pressure of G418. Flow cytometry, qPCR, and qRT-PCR were used to explore eGFP expression levels, gene copy number, and mRNA expression levels, respectively. Furthermore, the erythropoietin (EPO) gene was used to test the selected strong promoter. Of the six promoters, the CHEF-1α promoter yielded the highest transgene expression levels, whereas the CMV promoter maintained transgene expression more stably during long-term culture of cells. We conclude that CHEF-1α promoter conferred higher level of EPO expression in CHO cells, but the CMV promoter with its high levels of stability performs best in this vector system.

to CMV and SV40 promoters alone 20,22 . Simian virus 40 (SV40) is also a strong promoter for the production of therapeutic proteins in mammalian cells, providing a lower expression yield, but higher stability than the EF1a and CMV promoters 14 . Mouse phosphoglycerate kinase 1 (PGK) promoter was also used to drive transgene expression in different cell lines 23,24 . Besides natural promoters, hybrid promoters and synthetic promoters were also developed to drive higher and long-term expression of a transgene 24,25 . CAG promoter is a hybrid construct consisting of the cytomegalovirus (CMV) enhancer fused to the chicken beta-actin promoter, which is also used to drive transgene expression in different cell lines 26,27 . Brown et al. designed 140 synthetic promoters to specifically regulate the expression of recombinant genes in CHO cells, which offer precise control of recombinant transcriptional activity in CHO cells spanning over two orders of magnitude 12 . Schlabach et al. took a synthetic biology approach for the generation and screening of transcription factor binding sites for activity in human cells, and yielded compound enhancers that were capable of a twofold greater enhancement activity than the CMV enhancer, with higher levels of activity than the original synthetic enhancer across multiple cell lines 28 .
Although previous reports attempted to identify strong promoters for transgene expression, none of the ideal promoters can significantly increase and maintain stable transgene expression. In this study, we report a systematic comparison of six commonly used promoters (CMV, mouse CMV, CHEF-1α, PGK, CAG, and HEF-1α), a CMV mutant, and a CAG enhancer in transfected CHO cell system. Our findings will benefit those choosing promoters during vector design to generate transfected CHO cell lines with both high expression level and long-term expression stability.

Results
Transfection efficiency and transient transgene expression. Transient gene expression is especially useful in early research when many potential therapeutic candidates are needed for evaluation or when a molecule is needed at short notice 29,30 . At 48 h post-transfection, the fluorescence intensity of CHO cells transfected with different types of promoter was observed under a fluorescence microscope (Fig. 1A). CAG, HEF-1α, mouse CMV, and CHEF-1α showed enhanced transgene expression when compared with CMV. The transfection efficiency of CAG, HEF-1α, mouse CMV, and CHEF-1α was better than that of the CMV mutant and the CAG CMV promoter was regarded as 100, the MFI of other promoter were calculated. CMV, Cytomegalovirus major immediateearly; CAG, the CMV enhancer fused to the chicken beta-actin promoter; CHEF-1α, Chinese hamster elongation factor-1α; mouse CMV, mouse cytomegalovirus; HEF-1α, human elongation factor-1α; PGK, phosphoglycerate kinase; and CMV protein mutant, CAG enhancer. EPO, erythropoietin; SpA, simian virus 40 early polyadenylation signal; eGFP, enhanced green fluorescence protein.

R E T R
Scientific RepoRts | 7:10416 | DOI:10.1038/s41598-017-10966-y enhancer. We also observed that the mouse CMV presented a significantly higher transient efficiency (about 93.5%) than the control vector (about 57.6%), followed by CHEF1-α (90.7%), CAG (78.8%), and CAG enhancer (74.3%) (Fig. 1B). DNA length influences the transfection efficiency, but the PGK promoter was the sixth shortest in length, shorter than CAG, HEF1-α, CHEF-1α, the CMV mutant, and CMV, and showed the lowest activity. This demonstrated that the length of the promoter is not a major factor affecting transfection efficiency.
Compared with the CMV promoter, the enhancement was the highest for CHEF1-α, which improved transgenic eGFP expression by 2.9-fold, followed by HEF1-α (2.4-fold). However, mouse CMV and CAG resulted in a slight increase in transient transgene expression (Fig. 1C).

Recombinant protein expression of stably transfected cells. Transient expression cannot fully
reveal the function of different promoters, and it is important that vectors can be stably expressed in cells. When stably transfected cell lines were screened out, the eGFP protein levels (MFI) were measured by using flow cytometry ( Fig. 2A). The cells transfected with CHEF-1α promoter-containing vectors exhibited the highest expression levels, followed by those containing HEF-1α, CMV mutant, CMV, mouse CMV, CAG, CAG enhancer, and PGK. When the eGFP expression level under the CMV promoter was considered as 100, the expression levels under the CHEF-1α, HEF-1α, CMV mutant, mouse CMV, CAG, CAG enhancer, and PGK promoters were 218. 13, 184.21, 164.33, 49.12, 47.95, 26.31, and 8.77, respectively (Fig. 2B). Therefore, expression under the HEF-1α, CHEF-1α, CMV mutant, and CMV promoters was higher than that under the mouse CMV, CAG, CAG enhancer, and PGK promoter. The highest activity was exhibited by the CHEF-1α promoter, with an MFI 2.18-fold that of the CMV promoter, 24.86-fold that of the PGK promoter, and 8.29-fold that of the CAG enhancer.
Recombinant mRNA expression. The mRNA expression level is closely related to that of recombinant protein 31 . We measured the mRNA expression levels in cells transfected with CMV, CAG, HEF-1α, and CHEF-1α promoters by qRT-PCR using eGFP as the target gene and GAPDH as the internal control. We found differences in recombinant mRNA expression levels among the cells transfected with CMV, CAG, HEF-1α, and CHEF-1α promoters; the highest level was found in CHEF-1α promoter-containing cells, followed by those containing the  CMV, HEF-1α, and CAG promoters (Fig. 3). The mRNA expression level was consistent with the protein expression level, but the increasing fold was not directly proportional, suggesting that a promoter can increase transgene expression not only at the mRNA level, but also by affecting post-transcriptional regulation.
Gene copy number analysis. To further study the mechanism of the promoters, 30 single cell clones transfected with different vectors, HEF-1α, CHEF-1α, CMV, and CAG, were selected and the gene copy number was further analyzed by using qPCR. Some differences in the number of plasmids per genome were detected among the cells, indicating that the ratios of transgene copy number per genome were disparate for all plasmids in this experiment. The copy numbers are presented in Fig. 5A. The relative mean gene copy numbers were 0.73 ± 0.09 (CAG), 6.89 ± 1.43 (CHEF-1α), 10.98 ± 0.94 (HEF-1α), 1 ± 0.17 (CMV) (Fig. 5B). This finding indicated that transgene expression level was not related with gene copy number, suggesting that the promoter enhancing activity did not involve an increase in gene copy numbers.

EPO expression levels.
To evaluate the effect of promoter on the secreted protein expression, we selected a therapeutic protein, erythropoietin (EPO), and analyzed its expression in CHO cells under serum-free medium culture conditions. We transfected the EPO-expressing vectors driven by the CHEF1-α or CMV promoter into CHO cells and analyzed the expression levels of EPO by using ELISA and Western blot. The mean EPO expression level of cells transfected with the CHEF1-α driven vector yielded 49.07 ± 2.4 mg/L (Fig. 6A), and those transfected with the CMV driven vector yielded 23.56 ± 0.9 mg/L (Fig. 6B). The results showed that EPO expression driven by the CHEF1-α promoter was about 2.08-fold higher than that driven by the CMV promoter. The highest production of EPO in the single cell clone driven by CHEF1-α and CMV was 75.3 mg/L, 54.9 mg/L, respectively. In addition, Western blot results were further demonstrated that EPO expression driven by the CHEF1-α promoter was higher than that of CMV promoter (Fig. 6C).

Discussion
An ideal vector should ensure persistent transgene expression without epigenetic effects 32 . The landscape of epigenetic silencing is complex, and many factors could influence transgene silencing 33,34 . The promoter is a major element in the expression cassette of gene therapy vectors, and optimal promoter selection can increase target specificity and gene expression 14 . Several reports investigated the effect of promoters on transgene expression 35,36 . At present, CMV is the most commonly used promoter for the production of recombinant protein 37 . However, the CMV promoter cannot maintain production stability over time and it has many potential methylation sites; mutations and methylation will lead to lower productivity of recombinant proteins 38 . Nonetheless, we demonstrated that the CMV promoter is more stable than the mouse CMV, CHEF-1α, PGK, CAG, HEF-1α, and CMV mutant promoters, and the CAG enhancer. Previous studies revealed that EF-1α can enhance transgene expression in a lentiviral vector-mediated system and produce high-level, stable transgene expression 39 . We also found that CHEF-1α can significantly improve exogenous gene expression, but with reduced stability in CHO cells. One report demonstrated that the CAG promoter could enhance exogenous gene expression when it was cloned into the pCAGGS-GFP vector and transfected into Balb/c mice 40 . However, we found that the CAG promoter cannot enhance transgene expression in long-term culture. This phenomenon may result from methylation of the CAG promoter 41 . Mammalian DNA is frequently methylated at cytosine bases that are part of CpG dinucleotides 42 . In this study, the CMV mutant was unable to improve exogenous gene expression when the cytosine at position 404 was point-mutated to guanine; hence, we speculate that C-404-G point mutations may inhibit transgene expression.
In the present study, the relative activities of six commonly used promoters (CMV, CAG, HEF-1α, mouse CMV, CHEF-1α, and PGK), the CAG enhancer, and a CMV mutant were compared to explore the effects on transgene expression in CHO cells. During the long-term passage of cells, non-viral-mediated vectors that contain large DNA fragments can cause deficient transgene expression. Among the six promoters, CMV mutant, and

R E T R A C T E D A R T I C L E
Scientific RepoRts | 7:10416 | DOI:10.1038/s41598-017-10966-y CAG enhancer tested here, the CAG enhancer was the shortest in length, followed by the mouse CMV and PGK promoters. The highest positive eGFP gene expression rate was achieved with the HEF1-α promoter-containing vector, indicating that transgene expression is not only affected by the length of the fragment, but also by multiple factors, including cell growth conditions, cell type, and cell state. Our results show that, among the six promoters, CAG enhancer and CMV mutant, CMV and CHEF-1α, are better promoters than the CMV mutant, CAG, CAG enhancer, mouse CMV, HEF-1α, and PGK in mammalian cells for long-term cell culture. Results obtained by flow cytometry and qPCR experiments revealed that the vector containing HEF-1α was integrated into the host chromosome with the highest copy number, followed by CHEF-1α, CMV, and CAG. These results demonstrated that there was no direct relationship between exogenous gene expression levels and gene copy number. qRT-PCR was used to analyze the expression levels of mRNA and showed that the cells transfected with CHEF-1α had the highest expression levels, followed by HEF-1α, CMV, and CAG. These results are consistent with the results of eGFP expression.
To further investigate the effect of strong promoter on the expression of a gene of interest, we evaluated the expression of a therapeutic secreted protein, erythropoietin (EPO), driven by the CHEF-1α and CMV promoters under serum-free medium culture conditions. The results showed that EPO expression level in CHO cells transfected with the vector containing CHEF-1α was significantly higher than of that of cells transfected with the vector containing the CMV promoter, which was consistent with eGFP gene expression.
Many studies investigated synthetic promoters 43 . Human CMV and HEF-1α constructs increased transgene expression by up to ten-fold 44 . However, human CMV usually peaks 1-2 days after transfection and the activity is rapidly lost 43 . Considering the effects of CMV on the relative maintenance of transgenes and the activity of CHEF-1α on transgene expression, we can use a CMV core and CHEF-1α to construct a synthetic promoter. The synthetic promoter would shorten the vector length and potentially contribute to the improvement of transfection efficiency.
In conclusion, we found that CHEF-1α showed high transgene expression activity and CMV derived from an IRES-mediated vector presented the highest retention rate. However, this experiment was performed only in CHO cells and the results cannot be extrapolated to other cell lines. The development of such expression systems is a major strategic task continually required for the expression of target proteins for research and is necessary for any future long-term clinical application. Our study makes a significant contribution to such research and applications.  45 (Fig. 7A, supplement 1). The CMV mutant, CAG enhancer, and CAG, CHEF-1α, mouse CMV, HEF1-α, PGK, and CMV promoters were 589 bp, 287 bp, 1662 bp, 1335 bp, 523 bp, 1659 bp, 555 bp, and 589 bp in length, respectively (Fig. 7B); eGFP was inserted into the vectors as a reporter gene. The sequences of elements were synthesized by General Biosystems (Chuzhou, China). Meanwhile, to detect the effect of promoter on secreted protein, we selected two promoters, CMV and CHEF1-α, to construct vectors containing the EPO cDNA (Fig. 7A). The EPO gene was synthesized according to the codon optimization sequence and further cloned into the vector using standard methods 46 .

Stable expression of transfected cells and flow cytometry analysis. Long-term stable expression
of the target gene is required for industrial production of recombinant proteins. Thus, cells transfected with plasmid vectors were passaged in pools and further cultured under G418 (800 μg/mL) selection for 16 days. When the non-transfected cells were dead, the stably transfected cell colonies were screened out and the concentration of G418 was reduced to 500 μg/mL to maintain the selection pressure. When cell colonies reached 90% confluence, the cells were harvested by using 0.25% trypsin/EDTA. Analyses were performed with a Guava EasyCyte ™ 8HT flow cytometer (MilliporeSigma, Darmstadt, Germany) using FlowJo software 7.6 (Tree Star, Ashland, OR, USA). This software provides values for the media fluorescence intensity (MFI) based on the fluorochromes in each cell captured by flow cytometry. Therefore, the eGFP MFI of each sample can be used as a reporter of transgene expression level. For EPO gene expression, the cells transfected with EPO gene were screened under G418 (800 μg/mL) selection for about 15 days and the stably transfected cells appeared. When cell colonies reached 90% confluence, the cells were harvested and then were seeded into 96-well plate with a limited dilution method to produce the one cell/well. The cell colonies took up about 15% of the surface of the well when cells grew about one week, and further transferred into 24-well, 6-well, gradualy. When the total cell number achieved 1 × 10 7 , the cells were cultured in protein-free, serum-free, chemically-defined CD CHO medium (Life Technologies # Analysis of long-term transgene expression stability. CHO cells stably transfected with the vectors containing human CMV, CMV mutant, CAG, CHEF-1α, CAG enhancer, mouse CMV, and HEF-1α were passaged in pools and further cultured. The MFI for each vector type was measured by using the Guava EasyCyte ™ 8HT flow cytometer, and the relative retention of eGFP expression for each vector was calculated as the ratio of the MFI after generation 60 to that at the start of stability testing. Each ten passages, the expression level of eGFP was determined by fluorescence intensity. qPCR analysis of gene copy number. Genomic DNA was isolated from 4 × 10 6 transfected CHO-S cells using a genomic DNA Mini Preparation kit with spin column (Beyotime). The gene copy numbers were determined through qPCR technique using the eGFP and the GAPDH reference gene. Oligonucleotide primer sequences are provided in Table 1. Copy numbers were determined in triplicate and are presented as ratios of individual copy numbers relative to the control. Double-stranded DNA fragments were prepared from the oligonucleotides by polymerization: a 10 μL reaction mix consisting of 4 μL template DNA (0.05 μg/μL), 5 μL SYBR Green (TAKARA, Dalian, China), 0.2 μL of each forward and reverse oligonucleotide (10 μM each), and 0.6 μL deionized water was subjected to heating at 95 °C for 3 min, 30 cycles of 94 °C for 30 s, 50 °C for 30 s, and elongation for 30 s at 72 °C, then 60 °C for 30 min for data acquisition.
Recombinant mRNA expression analysis. After approximately 10 passages, the expression levels of recombinant mRNA were analyzed. We isolated total RNA from cells transfected with each plasmid, and reverse transcription reaction was performed with a HiFiScript first strand cDNA synthesis kit (CWBIO, Beijing, China) according to the manufacturer's protocol. qRT-PCR was undertaken using a PikoReal ™ Software 2.2 Real-Time PCR System (Thermo Scientific, Waltham, MA, USA). cDNA template (40 ng/reaction) was quantified by using a Multiscan Spectrum spectrophotometer (SpectraMax i3x, Silicon Valley, CA, USA). PCR reactions were performed according to standard procedures using primer sets designed for the eGFP and GAPDH sequences ( Table 1). The mRNA expression level of the target gene was calculated by comparison with that of the internal reference gene.

ELISA analysis
Cells transfected vectors were screened using G418 (800 μg/mL) after 48 h post-tranfection. Colonies arose after 10-14 days. Briefly, like the method demonstrated in stable expression of transfected cells and flow cytometry analysis. 30 stable clones were picked out and transferred to 24-well plates. Volumetric EPO production (mg/L) for initial CHO-EPO cell lines cloned were scaled up to 125 mL Corning shake flasks. Cells were grown in protein-free, serum-free, chemically-defined CD CHO medium supplemented with 8 mM L-glutamine in 125 mL Corning shake flasks with 30 mL medium; at 60% density compared with 1.5 × 10 7 . Cells supernatant were collected for analysis by ELISA to determine volumetric EPO production as previously described 47 .
Western blot analysis. The cells transfected with EPO-containing vector were suspended. When the cell number reached 8 × 10 6 /mL, the supernatant was collected and EPO was detected by immunoblotting. The supernatant containing EPO mixed with 5× SDS sample buffer was boiled. Ten microliters of sample were subjected to electrophoresis on a 15% SDS-polyacrylamide gel and transferred to a nitrocellulose membrane by electro-blotting. A 1:1500 dilution of an anti-EPO (EPO resistance protein) rabbit antiserum (Baoankang Biotechnology Co., Ltd., Shenzhen, China) was incubated with the membrane followed by a secondary incubation with a 1:2000 dilution of goat anti-rabbit antibody conjugated to alkaline phosphatase (Jackson Immuno Research Lab, West Grove, PA, USA). Densitometric analysis was performed by using ImageJ v2.1.4.7 software (National Institutes of Health, Bethesda, MD, USA).

Statistical analysis.
All experimental data were analyzed by using SPSS 18.0 software (SPSS Inc., Chicago, IL, USA). Data are reported as mean ± standard deviation. All experiments were performed three times and t-tests were used for comparisons. Differences with P values <0.05 were considered statistically significant.