Evaluating the Interaction Between UCOE and DHFR-Linked Amplification and Stability of Recombinant Protein Expression Zeynep Betts, Alexandra S Croxford, and Alan J Dickson Faculty of Life Sciences, University of Manchester, Michael Smith Building, Oxford Road, Manchester M139PT, UK DOI 10.1002/btpr.2083 Published online April 8, 2015 in Wiley Online Library (wileyonlinelibrary.com)

Chinese hamster ovary (CHO) cells are widely used in the biopharmaceutical industry. In the creation of mammalian cell lines plasmid DNA carrying the gene-of-interest integrates randomly into the host cell genome, which results in variable levels of gene expression between cell lines due to gene silencing mechanisms. In addition, cell lines often show unstable protein production during long-term culture. This means that a large number of clones need to be screened in order to isolate stable, high producing cell lines making mammalian cell line development a long and laborious process. In this study an expression platform incorporating a Ubiquitous Chromatin Opening Element (UCOE; which are proposed to maintain chromatin in an open state) has been utilised for the expression of eGFP in CHO cells. Cell lines containing a UCOE vector, showed a significantly higher and more consistent eGFP expression than the non-UCOE cell lines without DHFR amplification. To further improve recombinant protein production cell lines were amplified with methotrexate (MTX). UCOE cell lines showed improved growth in MTX therefore amplification to 250 nM MTX was achieved following a one-step amplification procedure. However, non-UCOE cell lines showed higher levels of eGFP production following MTX amplification. In addition, UCOE cell lines did not improve stability during long-term culture in the absence of selective pressure. Stable eGFP production was achieved for all cell lines when MTX is present. Finally, UCOE cell lines displayed more consistent response to external stimuli than nonC UCOE cell lines, suggesting that UCOE cell lines are less prone to clonal variability. V 2015 American Institute of Chemical Engineers Biotechnol. Prog., 31:1014–1025, 2015 Keywords: UCOE, Chinese hamster ovary cells, MTX amplification, mammalian cell culture, instability

Introduction Despite recent advances in microbial and yeast expression systems, mammalian cells, in particular CHO cells remain the most used system for the production of therapeutic proteins, especially when the target protein requires a high level of post-translational modifications.1,2 One of the problems associated with heterologous gene expression in mammalian cells is that levels of recombinant protein achieved are highly variable between cell lines, and this variability is not copy number-dependent.2–4 In addition, it is frequently observed that cell lines that initially produce high levels of recombinant protein show a decrease in recombinant protein production during prolonged culture.2,5 These phenomena mean that a large number of cell lines need to be screened

Additional Supporting Information may be found in the online version of this article. Current address of Zeynep Betts: Kocaeli University Umuttepe Yerleskesi Fen Edebiyat Fakultesi B Blok Biyoloji Bolumu, Izmit, Kocaeli 41380, Turkey Current address of Alexandra S, Croxford: Eli Lilly and Company, Liverpool, UK Correspondence concerning this article should be addressed to Z. Betts at [email protected] or [email protected]. 1014

over long-term culture, to select for high, stable recombinant protein production. Decreases in protein expression during prolonged culture in recombinant cell lines can be accounted for by elimination of the transgene from the host genome, which is often the case in CHO cells that have undergone gene amplification.1,6,7 Alternatively, loss of protein production can also occur without any observed loss in gene copy number via mechanisms known as gene silencing.8–10 A number of chromosomal elements, that tend to have positive effects on the levels of transcription including Locus Control Regions (LCRs),11,12 nuclear scaffold/matrix attachment regions (S/ MAR),13,14 and Ubiquitous Chromatin Opening Elements (UCOEs), are utilized in recombinant protein production.15–18 UCOEs are comprised of an extended, methylation-free CpG island, that extends across a region encompassing a transcriptional start site and are endogenously localised in a position adjacent to ubiquitously expressed genes.19 UCOEs have been used as part of retroviral vectors in various cell lines including teratocarcinoma cell line P19 and hematopoietic stem cells and their differentiated progeny to improve and stabilize transgene expression.17,20–23 Antisilencing activity was associated with strongly reduced DNA C 2015 American Institute of Chemical Engineers V

Biotechnol. Prog., 2015, Vol. 31, No. 4

methylation of promoter, which strongly favors its application in PSC-based cell and gene therapy. Studies so far have shown that in CHO and BHK21 cells, UCOE provided a higher level of transgene expression following stable transfection18,24–27 and maintained the stability of protein expression over 100 generations.16 Often recombinant protein production in CHO cells is enhanced by amplifying the gene of interest via methrotrexate (MTX) amplification. In this study, we created DG44-CHO cell lines containing an hCMV-eGFP-IRES-DHFR expression cassette and subsequently amplified the inserted transgenes with MTX to see if the use of UCOEs and MTX amplification could be combined to promote even higher levels of recombinant protein production in CHO cells. We also evaluated whether the stepwise MTX amplification process could be skipped with the use of UCOEs by treating cell lines with higher concentrations of MTX to yield higher-producing cell lines and ultimately shorten the process. In addition, stability of eGFP expression was compared between UCOE and non-UCOE cell lines over long-term culture before and after amplification.

Material and Methods Cell culture and media The parental CHO-DG44 cells were supplied by British Biotech. All transfected cell lines were grown in RPMI medium supplemented with L-glutamine (4 mM final) and 10% (v/v) fetal bovine serum (FBS) (growth medium) either with or without MTX selection. MTX was added to a final concentration of 250 nM when required. 13 Hypoxanthine and Thymidine (HT) solution was added to growth medium for non-transfected parental CHO-DG44 cells. The anchoragedependent CHO cells used in this work were cultured routinely in T-75 flasks at 37  C and 5% CO2. Cells were sub-cultured every 48 to 72 h by rinsing with Dulbecco’s phosphate buffered saline (DPBS), detaching with trypsin, and quenching with the addition of growth medium. An appropriate volume of cells was diluted in fresh growth medium to give a cell density of 1 3 105 viable cells/mL. Long-term culture (LTC) was performed by continuous sub-culture for a total of 80 days in growth medium with or without MTX. Samples for further experimental work were taken on days 10 and 80 of long-term culture. Growth characteristics were assessed by light microscopy using an improved Neubauer haemocytometer. Samples were appropriately diluted and mixed 1:1 with 0.5% [w/v] Trypan Blue in PBS. Expression vectors, stable transfection, amplification, and cloning Two eGFP expression vectors containing hCMV-eGFPIRES-DHFR-pA created either with or without the inclusion R upstream of the hCMV promoter (p1010of 8kb UCOEV eGFP, p901-eGFP, respectively). CHO cells were transfected with the appropriate linearized plasmids and limiting dilution was performed as described in Barnes et al.8 Cell lines were scaled up to six-well plates and eGFP fluorescence was measured by flow cytometry. The ten cell lines that showed the highest amount of eGFP fluorescence, were counted and an equal number of cells from each cell line were pooled and made up to a volume of 50 mL with RPMI 1 10% (v/v) FBS, to give a final concentration of 1.5 3 105 cells/mL. The pooled cells were then diluted in growth medium containing 250 nM MTX to give a range of dilutions (1.5 3 105

1015

cells/mL, 3 3 104 cells/mL, and 6 3 103 cells/mL). Limiting dilution cloning was then continued as described previously until cell lines were in six-well plates.8 eGFP fluorescence was assessed by flow cytometry and the top ten cell lines, in terms of eGFP fluorescence, were scaled up into T-75 flasks and frozen stocks were made. Flow cytometry Adherent cells were harvested and resuspended in PBS at approximately 1 3 106 cells per mL. The intensity of eGFP fluorescence of cells was assessed using a CYAN ADP flow cytometer, using the 488 nm excitation laser, according to the manufacturer’s instructions. The voltage applied to the photomultiplier (PMT) tube was adjusted; to ensure the histogram plots obtained were within range. Flow cytometry histograms of parental DG44 cells and a DG44-GFP clone by using a photomultiplier tube (PMT) input of 455V were compared with histograms of the same cell lines, this time using a PMT input at 266 V. The mean level of fluorescence (530/40 Log) is reduced by a factor of 6.8 3 1023 when moving from a PMT input of 455 V to 266 V. Therefore, mean fluorescence values measured using an input of 455 V are multiplied by 6.8 3 1023 when compared with cell lines measured at 266 V. Fluorescent data was acquired using a 530/30 nm bandpass filter and analyzed using Summit V.4.3 software. Live cells were gated by eye, using a forward scatter versus side scatter log plot, and single cells were gated using a forward scatter versus pulse width plot. Coefficient of Variance (CV) 5 (SD/m), where m is mean of n observations. Protein extraction and western blotting Cells on day 3 of batch culture were harvested in the beginning and at the end of LTC. The cell pellet was washed with PBS and resuspended in RIPA buffer (0.5% [w/v] sodium deoxycholate, 0.2% [w/v] SDS, 1.0% [v/v] Triton X100, 125 mM sodium chloride, 10 mM sodium floride, 10 mM sodium orthovanadate, 10 mM sodium pyrophosphate, 25 mM HEPES, pH 7.5) using 300 lL for every 1 3 107 viable cells. Protease inhibitors, PMSF, aprotinin and leupeptin were added. The extracts underwent sheering and cell lysates were incubated on ice for 30 min and then centrifuged at 11,000g at 4  C for 30 min. Supernatant were transferred to fresh tubes and stored at 280  C. In order to quantify the protein amount in samples standard BSA solution was prepared (100 lg/mL) and cell lysates were diluted to an appropriate degree for assessment against the standard in 96-well plates. Biorad reagent, diluted in 1:3, was added to all wells and the absorbance was measured at 570 nm in a plate reader after 10 to 15 min. The protein amount in cell lysates was calculated by comparison to the standard curve generated with BSA. Western blotting was performed as described in Barnes et al.8 using GFP, DHFR (Abcam, UK), and ERK antibodies (BD Transduction Laboratories). One cell protein sample from each nonamplified and amplified cell lines was dedicated as “standard” sample. These standard samples were assessed in all related western blot analysis (i.e. nonamplified standard for analyzing samples from nonamplified cell lines, amplified standard for analyzing samples from amplified cell lines). This allowed “internal control” comparison of the protein content of the samples. The density of bands on X-ray film was determined using ImageJ software.

1016

Biotechnol. Prog., 2015, Vol. 31, No. 4

Table 1. Primer Sequences used in PCR Reactions Sequence (50 –30 )

Name b-actin forward (mRNA) b-actin reverse (mRNA) b-actin forward (DNA) b-actin reverse (DNA) GFP forward (mRNA) GFP reverse (mRNA) GFP forward (DNA) GFP reverse (DNA) DHFR forward DHFR reverse

TGTGACGTTGAC ATCCGTAAA GCAATGATCTT GATCTTCATG ACTGCTCTGGCT CCTAGCAC CATCGTACTCC TGCTTGCTG CAGGTACCCAG ACCACATGA TCACCCTCGAA TTTGACCTC CGTGT ACGGTGG GAGGTCT A AACAGCTCTTC CCCTTTCT CAGGAAGCCAT GAATCAACC AGAGAGGACGC CTGGGTA TT

Genomic DNA extraction Genomic DNA was extracted by addition of phenol:chloroform:isoamyl alcohol (25:24:1) and recovered by ethanol precipitation. DNA concentration and purity was measured R UV-Vis Spectrophotometer (Thermo Scienby NanoDrop V tific via ABgene, Epsom, Surrey, UK). RNA extraction RNA was extracted from cells in T-75 plates using TRIzol R (Invitrogen, Paisley, UK) according to the manufacturer’s V

instructions. RNA concentration was quantified using NanoR UV-Vis Spectrophotometer. Extracts were then treated Drop V with DNaseI (Sigma) to remove any trace contamination of genomic DNA. The RNA was subsequently used as a template for cDNA. Reverse transcriptase production of cDNA was performed by using a cDNA Synthesis Kit (Bioline, London, UK), according to the manufacturer’s instructions. qPCR and qRT-PCR analysis The genomic DNA samples, cDNA samples, and negative control samples (parental genomic DNA or samples which had not been reverse transcriptase-treated), were prepared for qPCR by diluting in ddH2O or DEPC-treated water. A series of standards (created from the recombinant plasmid vector or from a chosen cDNA sample) were also used during each essay. The protocol for PCR and primers used are shown in Table 1. Target gene was quantified by normalizing to b-actin. Primers were designed using the Primer 3 output program (Whitehead Institute for Biomedical Research, 1988). Forward and reverse primers for each set were diluted in ddH2O or DEPC-treated water (for DNA and RNA analysis respectively) to give a final concentration of 10 lM and then mixed with R Green I qPCR MasterMix at a ratio of 1:2 (Euro23 SYBRV gentec Ltd); 5 lL sample/standard and 15 lL appropriate primers with reaction mix were set up in 96-well plates in triplicate reactions. qPCR reaction was performed using a Chromo 4 thermal cycler with the following settings: 95  C for 10 min, followed by 35 cycles of denaturation at 95  C for 10 s, annealing at 57  C for 10 s, elongation at 72  C for 20 s and denaturation of primer dimers at 76  C for 1 s. A final elongation step at 72  C for 10 min was performed. Data was quantified using Opticon 4 software (BioRad).

Results Recombinant cell lines were obtained by transfection of CHO-DG44 cells with either UCOE or non-UCOE vector (initial cell lines). Limiting dilution cloning was performed following transfection and over 50 single colonies were observed. These were subsequently scaled up to six-well

plates and eGFP fluorescence was measured by flowcytometry. The mean fluorescence in the UCOE cell lines was significantly higher than the non-UCOE cell lines (Figure 1A). In addition to levels of fluorescence, the degree of population variance (described by the coefficient of variance [CV] values) was compared between cell lines. Although CV values cannot fully depict the population characteristics obtained for each cell line via flow cytometry, it provides simplified information on the degree of variation seen for each cell line, which is useful when assessing many cell lines at once. The mean CV value for UCOE cell lines was lower than the mean CV for non-UCOE cell lines, although this result was not statistically significant (Figure 1B). Analysis of growth and eGFP expression during long-term culture in initial CHO-GFP cell lines Five of the ten highest producing cell lines were randomly chosen and grown continuously for up to 80 days. Viable cell densities and eGFP expression were analyzed in the beginning (day 10) and at the end of long-term cultivation (day 80). There was no difference in viable cell densities between the UCOE and non-UCOE cell lines in early and late generation. Although patterns of growth changed as long-term culture progressed there was no significant change in viable cell densities created at late generation compared with early generation (Figure 2A).eGFP production was analyzed using Western blotting where ERK was used as a loading control (Figures 2B,C). All the UCOE cell lines displayed stable and high amounts of eGFP production over long-term culture. Although some of the non-UCOE cell lines showed an increase in eGFP production throughout the culture, the mean values of eGFP production were higher in the UCOE cell lines, as compared with the non-UCOE group in early and late generations, where the difference was observed as 3- and 1. 6-fold, respectively. Amplification of CHO-GFP cell lines The top ten highest producing initial cell lines from each group were pooled and treated with 250 nM MTX and immediately cloned by limiting dilution cloning. Following MTX treatment and cloning, more than 50 single colonies were observed in the UCOE cells. In comparison fewer colonies were observed in the non-UCOE 96-well plates (Figure 3A). All single colonies observed in the non-UCOE cells and a total of 50 single colonies from the UCOE group were scaled up to six-well plates. In contrast to the results observed in initial cell lines, mean fluorescence was significantly lower for the UCOE cell lines compared with the non-UCOE cell lines (Figure 3C). Furthermore, the mean CV value for UCOE cell lines was significantly lower than the mean CV value for the nonUCOE cell lines (Figures 3B,D). This shows that the variance within individual UCOE cell lines (i.e. intra-cell line variance) was lower than the non-UCOE cell lines. In addition, the variability between cell lines (i.e. intercell line variance) was less pronounced in UCOE cell lines compared with non-UCOE cell lines. According to Levene’s test for equality of variances, there was significantly less variance in levels of eGFP fluorescence between cell lines in the UCOE group, as compared with the non-UCOE group (P < 0.01). In addition to this there was also significantly less variance in the CV values between cell lines in the UCOE group, as compared with the nonUCOE group (according to the Levene’s test for equality of

Biotechnol. Prog., 2015, Vol. 31, No. 4

1017

Figure 1. Level of eGFP fluorescence and variance for initial cell lines. (A) Levels of fluorescence were measured by flow cytometry, with a PMT input at 455V. (B) CVs calculated as described in Methods. Cell lines are sorted according to individual fluorescent intensities, and are aligned between A and B. UCOE non-UCOE. At least 50 cell lines were tested and mean values are quoted 6 SEM. *P < 0.01 using independent samples t-test to compare UCOE & non-UCOE cell lines.

variances, P < 0.05), suggesting that the response of UCOE cells to MTX was more consistent than non-UCOE cells, resulting in a less varied population of cell lines in terms of eGFP fluorescence. Analysis of growth and eGFP production in amplified cell lines To see the effects of the amplification on CHO-GFP cell lines, growth characteristics and eGFP production was analyzed over long-term culture (Figures 4 and 5). Growth of batch cultures was analyzed in early (day 10) and late (day 80) generation cell lines in the presence and absence of MTX selection (Figures 4A,B). When compared with the initial cell

lines both the UCOE and non-UCOE group showed significantly lower cell densities after amplification in early generation (Figures 2A and 4A). Viable cell densities reached a similar level to the initial cell lines when long-term culture progressed with or without MTX selection in UCOE group (Figures 2A and 4A,B). However, viable cell numbers in amplified non-UCOE cell lines at late generation in the presence of MTX were still significantly lower than the initial cell lines (P < 0.05). UCOE cell lines showed better growth characteristics than non-UCOE cell lines in the presence of MTX and achieved higher maximum cell densities (please refer to Supplement 1 for CCT data). In order to address the stability of recombinant protein production in amplified CHO-GFP cell lines, eGFP produc-

Figure 2. Growth and eGFP production for five high producer clones over long-term culture. (A) Viable cell densities were determined by light microscopy and trypan blue exclusion from samples taken during batch culture. Values are preUCOE, non-UCOE, Early generation, 䊊 Late generation. (B) Relative intensity of sented as average of each group 6 SEM (n = 5). Early generation UCOE, Late generation UCOE Early generaeGFP bands was determined by densitometric analysis normalized to ERK. Late generation non-UCOE. Values are quoted 6 range (n = 2), overall mean values are quoted 6 SEM (n = 5). * and * indition non-UCOE, cates P < 0.05 using independent samples t-test and paired samples t-test respectively to compare UCOE & non-UCOE cell lines and early and late generations. (C) eGFP expression was measured by western blot analysis. E 5 early generation, L 5 late generation.

1018

Biotechnol. Prog., 2015, Vol. 31, No. 4

Figure 3. Level of eGFP fluorescence and variance in amplified cell lines. The top 10 highest producing initial cell lines were pooled and treated with 250 nM MTX. (A) Fluorescence levels of amplified CHO-GFP cell lines UCOE, non-UCOE. (B) CVs are calculated as described in Materials and Methods section. Cell were measured with a PMT input of 266V. lines are sorted according to individual fluorescent intensities, and are aligned between A and B. Mean values are quoted 6 SEM. (C) Mean levels of fluorescence (530/40 Log) of initial and 250 nM MTX amplified clones are compared. 0nM MTX cell lines were analyzed using a PMT input of 455V, whereas amplified cell lines were analyzed with a 530/40 input of 266 V. The differing input voltages were compensated as described in Methods (D) Mean CVs are compared before and after MTX amplification. Values are quoted 6 SEM. *P < 0.01 using independent samples t-test to compare cell lines.

tion was analyzed by Western Blotting using samples taken on days 10 and 80 of long-term culture from cells growing in the presence and absence of MTX (Figure 5). Data showed that most cell lines retained eGFP production at a constant level throughout long-term culture in the presence of MTX, as reflected in the overall mean value for both UCOE and non-UCOE group (Figures 5A,B). Furthermore, the mean eGFP expression was significantly higher in the non-UCOE cell lines compared with the UCOE cell lines at the beginning of long-term culture in the presence of MTX (Figure 5A). This was not significant at the end of long-term culture. All cell lines showed a loss of eGFP production over long-term culture, in the absence of MTX, which was reflected in the overall mean of each group (Figures 5C,D). Therefore, eGFP expression was not more stable in UCOE cell lines compared with non-UCOE cell lines. DHFR protein production was also assessed over longterm culture (Figure 6). The mean values of DHFR expression in both UCOE and non-UCOE cell lines showed a decrease over long-term culture in the presence of MTX but it was not statistically significant (Figures 6A,B). Moreover, DHFR expression was similar in all UCOE cell lines. In contrast, a variable DHFR expression was observed in nonUCOE cell lines. A decrease in the intensity of the bands corresponding to DHFR for both UCOE and non-UCOE cell lines maintained in the absence of MTX was clearly visible from Western blot analysis (Figures 6C,D). Although UCOE and non-UCOE cell lines have a similar amount of DHFR

protein, it was shown that UCOE cell lines exhibit higher viable cell numbers in the presence of MTX, and this suggests that UCOE cell lines have a higher tolerance to MTX than non-UCOE cell lines. This is in accordance with the fact that more UCOE cell lines were produced following MTX amplification, although these cell lines showed significantly lower levels of eGFP protein expression. Effect of MTX amplification on plasmid copies per cell and RNA expression To further investigate the reason for the different level of eGFP production between UCOE and non-UCOE cell lines and the instability observed during long-term culture, plasmid copy numbers (Figure 7) and eGFP and DHFR RNA expression (Figure 8) were analyzed by qPCR and qRT PCR, respectively. Plasmid copy numbers in non-UCOE cell lines were significantly higher than the UCOE cell lines in all conditions throughout long-term culture (Figures 7A,B). Despite the differences in individual cell lines, the group mean values of plasmid copy numbers for eGFP in the UCOE cell lines showed no difference over long-term culture in the presence of MTX, whereas a decrease was observed in the non-UCOE cell lines. However, this was not statistically significant (Figure 7A). The gene copy numbers for eGFP were stable in the U5, N4, and N5 cell lines over long-term culture after MTX removal (Figure 7B). The remaining cell lines lost

Biotechnol. Prog., 2015, Vol. 31, No. 4

1019

Figure 4. Growth analysis of amplified CHO-GFP cell lines over long-term culture with (A) or without (B) MTX selection. UCOE,

non-UCOE,

Early generation,

Late generation. Viable cell densities were determined as described in Figure 2.

some of their eGFP genes within 10 days in response to MTX removal when compared with the early generation cell lines in the presence of MTX. A further loss of gene copies was observed in the U1, U3, and N2 cell lines at the end of long-term culture in the absence of MTX. A significant loss in the overall mean value of eGFP gene copy numbers was observed in the UCOE group after 80 days of MTX removal. The difference in the overall mean values of eGFP gene copies was not significant over long-term culture in the absence of MTX in non-UCOE cell lines. All cell lines showed a continuous decrease in recombinant eGFP protein production in the absence of MTX and although some cells showed a decline in gene copy numbers,

the decrease in gene copy number was not proportional to the protein expression. Therefore, it was necessary to confirm the effect of long-term culture and removal of MTX on expression of eGFP at mRNA level. Although the nonUCOE cell lines had significantly higher numbers of eGFP gene copies/cell when compared with the UCOE cell lines, this was not reflected in the eGFP mRNA expression (Figures 8A,B). The amount of eGFP mRNA was found to be significantly higher in the UCOE cell lines than non-UCOE cell lines both in the presence and absence of MTX throughout long-term culture, suggesting that UCOE provides a higher transcriptional activity for the eGFP gene. No significant difference was observed in the overall group mean

Figure 5. eGFP expression in amplified cell lines over long-term culture was assessed by western blot analysis. eGFP expression in the presence (A, B) and absence (C, D) of MTX selection are shown. E 5 early generation, L 5 late generation. (A, C) Relative Early generation UCOE, Late generation UCOE, intensity of eGFP bands was determined by densitometric analysis and normalized to ERK. Early generation non-UCOE, Late generation non-UCOE. Values are quoted 6 range (n = 2). Overall mean values are quoted 6 SEM (n = 5). * and * indicates P < 0.05 using independent samples t-test and paired samples t-test, respectively to compare UCOE & non-UCOE cell lines and early & late generations.

1020

Biotechnol. Prog., 2015, Vol. 31, No. 4

Figure 6. DHFR expression in amplified cell lines over long-term culture was assessed by western blot analysis. DHFR expression in the presence (A, B) and absence (C, D) of MTX selection are shown. E 5 early generation, L 5 late generation. (A, C) Relative Early generation UCOE, Late generation UCOE, Early generband intensities of DHFR protein were quantitated and normalized to ERK. ation non-UCOE, Late generation non-UCOE. Values are quoted 6 range (n = 2). Overall mean values are quoted 6 SEM (n = 5). * and *P < 0.05 using independent samples t-test and paired samples t-test, respectively to compare UCOE & non-UCOE cell lines and early and late generations.

value during long-term culture in the presence of MTX in either the UCOE or non-UCOE group (Figure 8A). In the absence of MTX, all cell lines showed a decrease in eGFP mRNA expression over the course of the culture, which was reflected by the mean values of each group (Figure 8B). Figures 8C,D show DHFR mRNA expression over-long term culture in the presence and absence of MTX selection. Overall, UCOE and non-UCOE cell lines showed similar amounts of DHFR mRNA at the start of long-term culture. Although DHFR mRNA was stable throughout continuous culture in the presence of MTX, a decrease was observed after the removal of MTX in UCOE cell lines. A loss of DHFR mRNA was observed in non-UCOE cell lines both in the presence and absence of MTX selection (Figures 8C,D). Table 2 summarises genetic parameters of amplified cell lines over long-term culture. Data showed a strong correlation between eGFP protein and mRNA expression in UCOE and non-UCOE cell lines (P < 0.001, r 5 0.839 and 0.771, respectively). However, no significant correlation was observed between eGFP gene copy/cell and protein production. To calculate the % decrease for individual cell lines, early and late generations in the absence of MTX selection were compared with the cell lines at early generation with MTX. Despite the eGFP gene copy number remaining relatively constant in the N4 and N5 cell lines and up to a 40% loss in the N1, N2, N3 cell lines, all early generation non-UCOE cell lines lost >30% of their eGFP mRNA expression in the absence of MTX compared with the cell lines with MTX selection. This loss of eGFP mRNA was approximately 90% at the end of long-term culture in the absence of MTX, which was in accordance with the loss of eGFP protein.

According to the results summarised in Table 2, the decrease in eGFP production in U1 cell line is mainly due to loss of eGFP gene copies. Despite having a relatively stable eGFP gene copy number, the U3 cell line showed a loss of protein production accompanied by a decrease in mRNA expression within 10 days after removal of MTX and after 80 days it lost >80% of its eGFP gene copies together with mRNA expression and protein production. This might indicate that the U3 cell line experienced a decrease in transcriptional activity in the first 10 days of MTX-free culture followed by a loss of gene copies within 80 days in the absence of MTX. The U2 cell line displayed 30% decrease in gene copy copies after MTX removal. However the decrease in eGFP protein and mRNA expression was more pronounced. Whereas the U4 cell line lost 50% of its eGFP gene copies within the first 10 days of MTX removal accompanied by a loss of protein production. After 10 days, the gene copy number remained constant for up to 80 days in the absence of MTX. However, cells continued losing their eGFP production and mRNA expression. Although the U5 cell line had the lowest eGFP gene copy per cell, it displayed a relatively high level of eGFP production and mRNA expression at the beginning of long-term culture. In addition, the eGFP gene copies did not change over long-term culture in the absence of MTX. However, this cell line experienced >92% loss of eGFP along with 86% loss of eGFP mRNA within 80 days after they were transferred to MTX-free medium. These results may suggest that some of the remaining eGFP gene copies are no longer transcriptionally active at the end of prolonged culture in U2, U4, and U5 cell lines.

Biotechnol. Prog., 2015, Vol. 31, No. 4

1021

Figure 7. Analysis of plasmid copy numbers in amplified cell lines over long-term culture in the presence (A) and absence of MTX selection (B) by qPCR. Genomic DNA content was normalised using b-actin primers specific for DNA. Error bars represent range of two biological replicates. Early Late generation UCOE, Early generation non-UCOE, Late generation non-UCOE. Overall mean values are quogeneration UCOE, ted 6 SEM (n = 5). *P < 0.05 using independent samples t-test to compare UCOE and non-UCOE cell lines.

Discussion The results presented here demonstrate that the use of UCOE improves the expression of linked transgene without DHFR amplification. In contrast, following MTX-mediated gene amplification UCOEs did not show favorable levels of recombinant protein production, as this was higher in nonUCOE cell lines. Amplification studies showed that the time required for cell line development could be greatly reduced by using UCOEs, which was due to the superior growth of UCOE-containing cell lines in the presence of MTX selection.

It was also shown that UCOE cell lines possess significantly lower CVs compared with non-UCOE cell lines after MTX amplification. Other studies have also demonstrated a more consistent level of reporter gene expression in UCOE cell lines compared with non-UCOE cell lines.16,20,22,23 However, in contrast to initial cell lines, eGFP expression was found to be lower in the UCOE cell lines compared with the non-UCOE cell lines. What appears to be occurring is that non-UCOE cells need to produce higher levels of eGFP to survive a particular concentration of MTX, compared with the UCOE cells. If it was possible for non-UCOE cells to survive 250 nM MTX, with similar low-levels of

Figure 8. Analysis of eGFP (A, B) and DHFR (C, D) mRNA expression over long-term culture in the presence (A, C) and absence of MTX selection (B, D). mRNA expression of the early and late generation cultures were analyzed using qRT-PCR. Relative values of eGFP and DHFR mRNA were calcuEarly generation UCOE, late generation UCOE, early generation non-UCOE, late generation nonlated and normalised to b-actin. UCOE. Error bars represent SD of three biological replicates. Overall mean values are quoted 6 SEM (n = 5). * and *P < 0.05 using independent samples t-test and paired samples t-test, respectively to compare UCOE & non-UCOE cell lines and early & late generations.

1022

Biotechnol. Prog., 2015, Vol. 31, No. 4

Table 2. Comparison of Gene Copy Number Protein and RNA Expression and Their Change During Long-Term Culture eGFP gene % eGFP % eGFP copy/cell Decrease expression Decrease RNA U1

U2

U3

U4

U5

N1

N2

N3

N4

N5

Early 1 MTX Late 1 MTX Early 2 MTX Late 2 MTX Early 1 MTX Late 1 MTX Early 2 MTX Late 2 MTX Early 1 MTX Late 1 MTX Early 2 MTX Late 2 MTX Early 1 MTX Late 1 MTX Early 2 MTX Late 2 MTX Early 1 MTX Late 1 MTX Early 2 MTX Late 2 MTX Early 1 MTX Late 1 MTX Early 2 MTX Late 2 MTX Early 1 MTX Late 1 MTX Early 2 MTX Late 2 MTX Early 1 MTX Late 1 MTX Early 2 MTX Late 2 MTX Early 1 MTX Late 1 MTX Early 2 MTX Late 2 MTX Early 1 MTX Late 1 MTX Early 2 MTX Late 2 MTX

262 6 49 152.7 6 46 110 6 8 70 6 10 141 6 13 280 6 21.7 101 6 15 95 6 11 101 6 5 82.8 6 12 82 6 27 19 6 2.6 165 6 6 126.5 6 9.2 86 6 12 77 6 11 51 6 8 61.8 6 12.6 47 6 15 43 6 8 185 6 6 93.6 6 8.8 123 6 31 65 6 1.5 1377 6 85 1128.2 6 53.6 868 6 54 560 6 28 1213 6 113 875 6 58.3 739 6 36 691 6 67 501 6 85 389.3 6 53.9 483 6 122 704 6 240 569 6 27 550.5 6 34.8 640 6 10 633 6 68

41.8 58.0 73.2 299.6 28.2 32.4 18.1 19.0 81.0 23.5 47.9 53.7 221.3 8.3 14.7 49.4 33.5 64.8 18.1 36.9 59.3 27.9 39.1 43.0 22.4 3.5 240.0 3.3 212.0 211.0

54 6 8 47.8 6 8.6 14 6 3 5 6 1.5 66 6 12 34.2 6 13.5 19 6 9 4 6 1.9 59 6 4 39.6 6 6.7 21 6 0.03 6 6 2.1 85 6 6 70.6 6 7.8 65 6 10 30 6 10 88 6 25 100.4 6 28.6 75 6 5 7 6 1.8 95 6 9 71.6 6 12.5 38 6 4 20 6 3 35 6 9 33.6 6 6.7 44 6 3 15 6 0.1 125 6 55 47.1 6 6.3 57 6 10 17 6 3 177 6 39 198.2 6 23.5 192 6 29 25 6 4 171 6 19 127.3 6 8.2 99 6 22 29 6 3

10.9 74.3 91.1 48.0 70.7 93.9 32.4 63.6 89.5 17.2 24.0 64.7 214.4 14.5 92.4 24.4 60.3 78.4 4.3 225.0 56.6 62.2 54.3 86.2 211.9 28.6 85.8 25.8 42.9 83.1

159 6 16.1 208.8 6 3.2 40.3 6 1.6 34.3 6 0.4 111.9 6 3.4 72.6 6 9 55.6 6 6.1 24.4 6 0.3 107.2 6 6.6 160.6 6 14.6 47.7 6 6.1 28 6 3.1 121 6 4.8 186.2 6 24.1 133.5 6 19.5 86.8 6 9.3 195.8 6 5.8 186.5 6 8.7 149.9 6 13.6 26.9 6 0.7 39.4 6 6.9 58.1 6 4.1 27.8 6 0.7 5 6 0.4 41.3 6 6.3 45.9 6 5.3 22.3 6 4.7 2.7 6 0.2 55.1 6 3.8 23.5 6 0.8 25.7 6 4.3 5.7 6 0.1 46 6 1.2 59.7 6 5.7 57.1 6 3.8 4.3 6 0.6 53.8 6 6.9 39.7 6 3.5 26.6 6 1.6 5 6 0.1

% Decrease 231.3 74.6 78.5 35.1 50.3 78.2 249.8 55.5 73.9 253.9 210.3 28.3 4.8 23.5 86.3 247.7 29.4 87.4 211.1 46.0 93.4 57.4 53.3 89.7 229.7 224.1 90.7 26.2 50.5 90.8

Summary of eGFP gene copy number, eGFP expression (relative densitometer values) and mRNA for individual cell lines over long-term culture. Values for “% decrease” was calculated by comparing the results obtained from cell lines 10 and 80 after MTX removal to the results of early generation (day 10) cell lines with MTX selection.

eGFP expression, then more cell lines would have been obtained after amplification. It seems that UCOE cells have a higher tolerance to MTX than non-UCOE cells, containing the same amount of DHFR protein per cell. This is consistent with the fact that more UCOE cell lines were produced following MTX amplification, although these cell lines showed significantly lower levels of fluorescence, compared with the non-UCOE cell lines. eGFP production remained constant when MTX selection was present in both the UCOE and the non-UCOE cell lines. MTX removal resulted in a decrease in eGFP expression over long-term culture in all cell lines (whether constructed with or without UCOE inclusion). These results suggested that UCOEs appear to offer no preferential advantage in terms of stability when using MTX amplification. However, it was shown that the UCOE cell lines displayed better growth characteristics in the presence of MTX than the nonUCOE cell lines. For the purposes of recombinant protein production in an industrial setting, the main priority is to obtain the maximum amount of recombinant protein per liter of medium i.e. volumetric productivity. High volumetric production is achieved through a combination of high specific productivity and the ability of cells to grow to a high maximum cell density. Therefore, UCOEs may still confer

an advantage in producing recombinant proteins following MTX amplification, if improved growth characteristics offset the lower levels of specific productivity. In addition, as eGFP fluorescence was less variable between UCOE cell lines, fewer cell lines would need to be screened to select the best candidates for scale-up. Plasmid copies and mRNA expression were analyzed in order to assess the reason for instability of recombinant protein production and it was observed that the higher gene copy number did not necessarily result in higher protein expression either in the UCOE or in the non-UCOE cell lines which was reported in other studies.28–30 Williams et al. have also used UCOE element linked to the hCMVeGFP expression cassette and shown no significant correlation between level of expression and plasmid copies per cell.16 Another study that used an 8 kb UCOE to drive expression of a linked transgene in transgenic mice stated that copy number dependent expression of the transgene was not observed, suggesting that the transgenes were prone to position effects in the presence of UCOE.31 Moreover, despite having lower plasmid copies/cell, eGFP mRNA was found to be significantly higher in UCOE cell lines compared with non-UCOE cell lines, suggesting a higher transcriptional activity for the eGFP gene in UCOE cell lines.

Biotechnol. Prog., 2015, Vol. 31, No. 4

The lower amount of eGFP protein in MTX-treated UCOE cells of early generation was not consistent with the high mRNA amount. Whilst we have no direct explanation for this finding, we hypothesize that this may reflect processing events that occur under conditions of high translation of intracellular eGFP, leading to processes such as aggregation. In partial support of this we observed the appearance of fluorescent “crystals” in cell lines during development (Supplement 2) but this observation was not subject of systematic investigation. It should also be mentioned here that gene expression does not always occur continuously within the cell but cycles between active and inactive states. Genes are prone to “transcriptional bursts” of expression, which occurs both in engineered reported genes and natural genes.32–34 Essential genes display lower protein noise than nonessential genes and for most proteins the level of expression was inversely proportional to the level of protein noise.35,36 The transcriptional bursting in CHO cells was also verified and it was shown that a lower rate of protein degradation results in a decrease in the correlation between mRNA and protein levels.33,37 Chromatin remodeling is believed to be a logical candidate to mediate such regulation and it was found that the inclusion of a MAR element when creating recombinant CHO cells reduced the expression noise.13 UCOEs are derived from genomic loci associated with essential genes and proposed to maintain an open chromatin structure. Therefore if UCOEs decrease protein noise of associated transgenes, this may explain the difference between mRNA and protein levels and why more UCOE cell lines were obtained following MTX treatment (43 vs. 12 clones from UCOE and non-UCOE constructs, respectively), although producing lower levels of eGFP (please see Supplement 3 for more explanation). Direct comparison of cells of UCOE and non-UCOE derivation at equivalent stages of MTX amplification raises the interesting observation that, whilst each exhibit comparable amounts of mRNA encoding DHFR (i.e. responding to the MTX challenge), the UCOE cell lines express greater amounts of mRNA encoding eGFP. The manner in which a single IRES-based bicistronic transcriptional unit can result in detection of imbalanced amounts of the two linked coding sequences has no obvious explanation. The PCR-based detection process used in our studies did not account for the integrity of the entire DHFR encoding regions and there may be selfregulatory systems involving miRNA or exonucleases that limit DHFR coding sequences (whether monocistronic or part of the 3’-regions of multicistronic transcripts).38–40 Such hypotheses offer intriguing possibilities in relation to fine control of MTX-mediated recombinant gene amplification but were beyond the immediate aims of the current project. DHFR and eGFP protein expression was moderately correlated; however, the pattern of expression was not parallel to each other. IRESs are complex RNA structures, mainly found within the mRNAs of proteins involved in cell growth and apoptosis.41 It has been shown that IRES frequently becomes activated under stress conditions, when capdependent translation is compromised.42–44 A study conducted by Yang et al. showed that IRES mediated translation was activated and increased the expression of an endogenous protein involved in cell death in specific diseases when the lymphoid cancer cells were treated with the chemotherapeutic drug vincistrine.45 Therefore, it may be suggested that the IRES usage is limited and an expression of a gene of interest from an IRES-containing vector might be unpredictable.

1023

In most of the UCOE cell lines the instability of protein production was partly due to a loss of transgene copies, which is commonly observed in amplified cell lines.1,46 However, in some cases (i.e. U2 and U5), the cell lines experienced instability of eGFP expression with little or no change in plasmid copies per cell during long-term culture in the absence of MTX. Therefore, it may be suggested that UCOEs may not be totally resistant to the effect of the surrounding chromatin environment. Heller-Harrison et al. have reported that loss of transgene expression may occur by genetic rearrangement of transgenes resulting in uncoupling of bi-cistronic transcript and subsequent loss of monoclonal antibody production.47 These findings suggest that the use of UCOEs may still require the screening of clones over prolonged culture for desirable properties when used in combination with MTX amplification. The instability of protein production observed in the nonUCOE cell lines was also partly due to a loss of transgene copies over long-term culture in the absence of MTX. However, it should be noted that the non-UCOE cell lines still contained over 100 transgene copies per cell (>500 in most cases) at the end of prolonged culture in the absence of MTX. Therefore, the very low level of protein expression in non-UCOE cell lines in the absence of MTX selection cannot solely be attributed to the loss of plasmid copies. Some of the non-UCOE cell lines did not show any loss in transgene copies, although they displayed a profound loss of protein production and mRNA expression during prolonged culture in the absence of MTX. It has been reported that CHO-DG44 derived recombinant cell lines lost their protein expression during long-term culture in the absence of selective pressure through gene silencing by DNA methylation, with no change in plasmid copies.30,47,48 Therefore, it may be suggested that in the current study, cell lines which showed an instability of protein production despite having stable numbers of transgene copies over long-term culture may be experiencing gene silencing in the absence of MTX.

Conclusion The results in this study showed that UCOE cell lines are capable of surviving higher levels of MTX, following a onestep amplification procedure, compared with non-UCOE cell lines. However, UCOE cell lines did not show improved levels of stability during long-term culture in the absence of selective pressure. Stable recombinant protein production was maintained for all cell lines in the presence of MTX, although non-UCOE cell lines showed higher levels of recombinant protein production than UCOE cell lines. Nevertheless, UCOE cell lines may still confer an advantage in recombinant protein production, following MTXamplification, as they displayed better growth characteristics. Furthermore, transcriptional activity per gene copy was greater when UCOE was present in the expression construct both in the presence and absence of MTX. MTX amplification has provided further insights into how UCOEs may affect recombinant protein expression. UCOE cells showed a more consistent response following MTX addition, resulting in less variation in cell lines obtained. Similarly, removal of MTX resulted in a consistent response, in terms of eGFP expression, in UCOE cell lines. This suggests that UCOE cell lines are less prone to clonal variability, and so will react in a more predictable manner following exposure to external stimuli.

1024

Biotechnol. Prog., 2015, Vol. 31, No. 4

Acknowledgment The authors gratefully acknowledge the financial support of the Republic of Turkey Ministry of National Education for providing the scholarship. We would also like to thank Dr Steven Williams (Cobra Biologics) for providing the UCOE construct.

Literature Cited 1. Almo SC, Love JD. Better and faster: improvements and optimization for mammalian recombinant protein production. Curr Opin Struct Biol. 2014;26:39–43. 2. Bandaranayake AD, Almo SC. Recent advances in mammalian protein production. FEBS Lett. 2014;588:253–260. 3. Jiang Z, Huang Y, Sharfstein ST. Regulation of recombinant monoclonal antibody production in chinese hamster ovary cells: a comparative study of gene copy number, mRNA level, and protein expression. Biotechnol Prog. 2006;22:313–318. 4. Vishwanathan N, Le H, Jacob NM, Tsao Y-S, Ng S-W, Loo B, Liu Z, Kantardjieff A, Hu W-S. Transcriptome dynamics of transgene amplification in Chinese hamster ovary cells. Biotechnol Bioeng. 2014;111:518–528. 5. Barnes LM, Bentley CM, Dickson AJ. Molecular definition of predictive indicators of stable protein expression in recombinant ns0 myeloma cells. Biotechnol Bioeng. 2004; 85:115–121. 6. Fann CH, Guirgis F, Chen G, Lao MS, Piret JM. Limitations to the amplification and stability of human tissue-type plasminogen activator expression by Chinese hamster ovary cells. Biotechnol Bioeng. 2000;69:204–212. 7. Jun SC, Kim MS, Hong HJ, Lee GM. Limitations to the development of humanized antibody producing chinese hamster ovary cells using glutamine synthetase-mediated gene amplification. Biotechnol Prog. 2006; 22:770–780. 8. Barnes LM, Bentley CM, Dickson AJ. Characterization of the stability of recombinant protein production in the GS-ns0 expression system. Biotechnol Bioeng. 2001;73:261–270. 9. Kim M, O’Callaghan PM, Droms KA, James DC. A mechanistic understanding of production instability in CHO cell lines expressing recombinant monoclonal antibodies. Biotechnol Bioeng. 2011;108:2434–2446. 10. Osterlehner A, Simmeth S, G€opfert U. Promoter methylation and transgene copy numbers predict unstable protein production in recombinant chinese hamster ovary cell lines. Biotechnol Bioeng. 2011;108:2670–2681. 11. Dean A. On a chromosome far, far away: LCRs and gene expression. Trends Genet. 2006;22:38–45. 12. Lisowski L, Sadelain M. Current status of globin gene therapy for the treatment of beta-thalassaemia. Br J Haematol. 2008; 141:335–345. 13. Galbete L, Buceta J, Mermod MN. MAR elements regulate the probability of epigenetic switching between active and inactive gene expression. Mol BioSyst. 2009;5:143 14. Harraghy N, Regamey A, Girod P-A, Mermod N. Identification of a potent MAR element from the mouse genome and assessment of its activity in stable and transient transfections. J Biotechnol. 2011;154:11–20. 15. Antoniou M, Harland L, Mustoe T, Williams S, Holdstock J, Yague E, Mulcahy T, Griffiths M, Edwards S, Ioannou PA, Mountain A, Crombie R. Transgenes encompassing dualpromoter CpG islands from the human TBP and hnrpa2b1 loci are resistant to heterochromatin-mediated silencing. Genomics. 2003;82:269–279. 16. Williams S, Mustoe T, Mulcahy T, Griffiths M, Simpson D, Antoniou M, Irvine A, Mountain A, Crombie R. CpG-island fragments from the hnrpa2b1/cbx3 genomic locus reduce silencing and enhance transgene expression from the hCMV promoter/enhancer in mammalian cells. BMC Biotechnol. 2005;5:17 17. Zhang F, Frost AR, Blundell MP, Bales O, Antoniou MN, Thrasher AJ. A ubiquitous chromatin opening element (UCOE) confers resistance to DNA methylation-mediated silencing of lentiviral vectors. Mol Ther. 2010;18:1640–1649.

18. Nair AR, Jinger X, Hermiston TW. Effect of different UCOEpromoter combinations in creation of engineered cell lines for the production of factor VIII. BMC Res Notes 2011;4:178 19. Crombie RL, Williams SG. Vectors Comprising CpG Islands Without Position Effect Variegation and Having Increased Expression. U.S. Patent 7,812,148, 2010. 20. Zhang F, Thornhill SI, Howe SJ, Ulaganathan M, Schambach A, Sinclair J, Kinnon C, Gaspar HB, Antoniou M, Thrasher AJ. Lentiviral vectors containing an enhancer-less ubiquitously acting chromatin opening element (UCOE) provide highly reproducible and stable transgene expression in hematopoietic cells. Blood 2007;110:1448–1457. 21. Antoniou MN, Skipper KA, Anakok O. Optimizing retroviral gene expression for effective therapies. Hum Gene Ther. 2013; 24:363–374. 22. Pfaff N, Lachmann N, Ackermann M, Kohlscheen S, Brendel C, Maetzig T, Niemann H, Antoniou MN, Grez M, Schambach A, Cantz T, Moritz T. A ubiquitous chromatin opening element prevents transgene silencing in pluripotent stem cells and their differentiated progeny. Stem Cells 2013;31:488–499. 23. Ackermann M, Lachmann N, Hartung S, Eggenschwiler R, Pfaff N, Happle C, Mucci A, G€ohring G, Niemann H, Hansen G, Schambach A, Cantz T, Zweigerdt R, Moritz T. Biomaterials 2014;35:1531–1542. 24. Benton T, Chen T, McEntee M, Fox B, King D, Crombie R, Thomas TC, Bebbington C. The use of UCOE vectors in combination with a preadapted serum free, suspension cell line allows for rapid production of large quantities of protein. Cytotechnology 2002;38:43–46. 25. Boscolo S, Mion F, Licciulli M, Macor P, De Maso L, Brce M, Antoniou MN, Marzari R, Santoro C, Sblattero D. Simple scaleup of recombinant antibody production using an UCOE containing vector. New Biotechnol. 2012;29:477–484. 26. Dharshanan S, Chong H, Cheah SH, Zamrod Z. Stable expression of h1c2 monoclonal antibody in ns0 and CHO cells using pFUSE and UCOE expression system. Cytotechnology 2014;66: 625–633. 27. Hou JJC, Hughes BS, Smede M, Leung KM, Levine K, Rigby S, Gray PP, Munro TP. High-throughput ClonePix FL analysis of mAb-expressing clones using the UCOE expression system. New Biotechnol. 2014;31:214–220. 28. Lattenmayer C, Loeschel M, Schriebl K. Analysis of genetic parameters in order to get more information on high producing recombinant CHO cell lines. Cell Technol Cell 2007;3: 95–98. 29. Lattenmayer C, Trummer E, Schriebl K, Vorauer Uhl K, Mueller D, Katinger H, Kunert R. Characterisation of recombinant CHO cell lines by investigation of protein productivities and genetic parameters. J Biotechnol. 2007;128:716–725. 30. Chusainow J, Yang YS, Yeo JHM, Toh PC, Asvadi P, Wong NSC, Yap MGS. A study of monoclonal antibody-producing CHO cell lines: what makes a stable high producer? Biotechnol Bioeng. 2009;102:1182–1196. 31. Katsantoni EZ, Anghelescu NE, Rottier R, Moerland M, Antoniou M, de Crom R, Grosveld F, Strouboulis J. Ubiquitous expression of the rtTA2S-m2 inducible system in transgenic mice driven by the human hnRNPA2B1/cbx3 CpG island. BMC Dev Biol 2007;7:108. 32. Sigal A, Milo R, Cohen A, Geva-Zatorsky N, Klein Y, Liron Y, Rosenfeld N, Danon T, Perzov N, Alon U. Variability and memory of protein levels in human cells. Nature 2006;444: 643–646. 33. Raj A, Peskin CS, Tranchina D, Vargas DY, Tyagi S. Stochastic mRNA synthesis in mammalian cells. Plos Biol. 2006;4:e309. 34. Kaufmann BB, van Oudenaarden A. Stochastic gene expression: from single molecules to the proteome. Curr Opin Genet Dev. 2007;17:107–112. 35. Newman JRS, Ghaemmaghami S, Ihmels J, Breslow DK, Noble M, DeRisi JL, Weissman JS. Single-cell proteomic analysis of S. Cerevisiae reveals the architecture of biological noise. Nature 2006;441:840–846. 36. Bar-Even A, Paulsson J, Maheshri N, Carmi M, O’Shea E, Pilpel Y, Barkai N. Noise in protein expression scales with natural protein abundance. Nat Genet. 2006;38:636–643.

Biotechnol. Prog., 2015, Vol. 31, No. 4 37. Pilbrough W, Munro TP, Gray P. Intraclonal protein expression heterogeneity in recombinant CHO cells. Plos One 2009;4:111 38. Herranz H, Cohen SM. MicroRNAs and gene regulatory networks: managing the impact of noise in biological systems. Genes Dev. 2010;24:1339–1344. 39. Jacquier A. The complex eukaryotic transcriptome: unexpected pervasive transcription and novel small RNAs. Nat Rev Genet. 2009;10:833–844. 40. Shevelev IV, H€ubscher U. The 3 “5” exonucleases. Nat Rev Mol Cell Biol. 2002;3:364–376. 41. Bonnal S. IRESdb: the internal ribosome entry site database. Nucleic Acids Res. 2003;31:427–428. 42. Holcik M, Sonenberg N, Korneluk RG. Internal ribosome initiation of translation and the control of cell death. Trends Genet. 2000;16:469–473. 43. Spriggs KA, Bushell M, Mitchell SA, Willis AE. Internal ribosome entry segment-mediated translation during apoptosis: the role of IRES-trans-acting factors. Cell Death Differ. 2005;12:585–591.

1025 44. Spriggs KA, Bushell M, Willis AE. Translational regulation of gene expression during conditions of cell stress. Mol Cell 2010; 40:228–237. 45. Yang L, Gu L, Li Z, Zhou M. Translation of traf1 is regulated by IRES-dependent mechanism and stimulated by vincristine. Nucleic Acids Res. 2010;38:4503–4513. 46. Kim NS, Kim SJ, Lee GM. Clonal variability within dihydrofolate reductase-mediated gene amplified chinese hamster ovary cells: stability in the absence of selective pressure. Biotechnol Bioeng. 1998;60:679–688. 47. Heller-Harrison RA, Crowe K, Cooley C, Hone M, Maccarthy K, Leonard M. Managing cell line instability and its impact during cell line development. Biopharm Int. 2009;22:1–12. 48. Yang Y, Mariati, Chusainow J, Yap MGS. DNA methylation contributes to loss in productivity of monoclonal antibodyproducing CHO cell lines. J Biotechnol. 2010;147:180–185. Manuscript received Jan. 29, 2015, and revision received Mar. 17, 2015.