Engineered Cytochrome P450-Catalyzed Oxidative Biaryl Coupling Reaction Provides a Scalable Entry into Arylomycin Antibiotics.

We report herein the first example of a cytochrome P450-catalyzed oxidative carbon-carbon coupling process for a scalable entry into arylomycin antibiotic cores. Starting from wild-type hydroxylating cytochrome P450 enzymes and engineered Escherichia coli, a combination of enzyme engineering, random mutagenesis, and optimization of reaction conditions generated a P450 variant that affords the desired arylomycin core 2d in 84% assay yield. Furthermore, this process was demonstrated as a viable route for the production of the arylomycin antibiotic core on the gram scale. Finally, this new entry affords a viable, scalable, and practical route for the synthesis of novel Gram-negative antibiotics.

FX knockout CHO hosts can express desired ratios of fucosylated or afucosylated antibodies with high titers and comparable product quality.

During antibody dependent cell cytotoxicity (ADCC) the target cells are killed by monocytes and natural killer cells. ADCC is enhanced when the antibody heavy chain's core N-linked glycan lacks the fucose molecule(s). Several strategies have been utilized to generate fully afucosylated antibodies. A commonly used and efficient approach has been knocking out the FUT8 gene of the Chinese hamster ovary (CHO) host cells, which results in expression of antibody molecules with fully afucosylated glycans. However, a major drawback of the FUT8-KO host is the requirement for undertaking two separate cell line development (CLD) efforts in order to obtain both primarily fucosylated and fully afucosylated antibody species for comparative studies in vitro and in vivo. Even more challenging is obtaining primarily fucosylated and FUT8-KO clones with similar enough product quality attributes to ensure that any observed ADCC advantage(s) can be strictly attributed to afucosylation. Here, we report generation and use of a FX knockout (FXKO) CHO host cell line that is capable of expressing antibody molecules with either primarily fucosylated or fully afucosylated glycan profiles with otherwise similar product quality attributes, depending on addition of fucose to the cell culture media. Hence, the FXKO host not only obviates the requirement for undertaking two separate CLD efforts, but it also averts the need for screening many colonies to identify clones with comparable product qualities. Finally, FXKO clones can express antibodies with the desired ratio of primarily fucosylated to afucosylated glycans when fucose is titrated into the production media, to allow achieving intended levels of FcγRIII-binding and ADCC for an antibody. Biotechnol. Bioeng. 2017;114: 632-644. © 2016 Wiley Periodicals, Inc.

Combining regulated and constitutive protein expression significantly boosts protein expression by increasing productivity without affecting CHO cell growth.

Chinese hamster ovary (CHO) cells are commonly used for the expression of therapeutic proteins. To increase the titer output of CHO production cultures either specific productivity (Qp), growth, or both need to be increased. Generally, Qp and growth are inversely correlated and cell lines with high Qp have slower growth and vice versa. During the cell line development (CLD) process, the faster-growing cells tend to take over the culture and represent the majority of the isolated clones post single cell cloning. In this study, combinations of regulated and constitutive expression systems were used to supertransfect targeted integration (TI) cell lines expressing the same antibody either constitutively or under-regulated expression. Clone screening with a hybrid expression system (inducible + constitutive) allowed identification and selection of higher titer clones under uninduced conditions, without a negative impact on cell growth during clone selection and expansion. Induction of the regulated promoter(s) during the production phase increased the Qp without negatively affecting growth, resulting in approximately twofold higher titers (from 3.5 to 6-7 g/L). This was also confirmed using a 2-site TI host where the gene of interest was expressed inducibly from Site 1 and constitutively from Site 2. Our findings suggest that such a hybrid expression CLD system can be used to increase production titers, providing a novel approach for expression of therapeutic proteins with high titer market demands.

Activation of the PERK branch of the unfolded protein response during production reduces specific productivity in CHO cells via downregulation of PDGFRa and IRE1a signaling.

During the course of biopharmaceutical production, heterologous protein expression in Chinese hamster ovary (CHO) cells imposes a high proteostatic burden that requires cellular adaptation. To mitigate such burden, cells utilize the unfolded protein response (UPR), which increases endoplasmic reticulum (ER) capacity to accommodate elevated rates of protein synthesis and folding. In this study, we show that during production the UPR regulates growth factor signaling to modulate growth and protein synthesis. Specifically, the protein kinase R-like ER kinase (PERK) branch of the UPR is responsible for transcriptional down-regulation of platelet-derived growth factor receptor alpha (PDGFRa) and attenuation of the IRE1-alpha (IRE1a) branch of the UPR. PERK knockout (KO) cell lines displayed reduced growth and viability due to higher rates of apoptosis despite having stabilized PDGFRa levels. Knocking out PERK in an apoptosis impaired (Bax/Bak double KO) antibody-expressing cell line prevented apoptotic cell death and revealed that apoptosis was likely triggered by increased ER stress and reactive oxygen species levels in the PERK KO hosts. Our findings suggest that attenuation of IRE1a and PDGFRa signaling by the PERK branch of the UPR reduces ER protein folding capacity and hence specific productivity of CHO cells in order to mitigate UPR and prevent apoptotic cell death. Last, Bax/Bak/PERK triple KO CHO cell lines displayed 2-3 folds higher specific productivity and titer (up to 8 g/L), suggesting that modulation of PERK signaling during production processes can greatly improve specific productivity in CHO cells.

Bax and Bak knockout apoptosis-resistant Chinese hamster ovary cell lines significantly improve culture viability and titer in intensified fed-batch culture process.

In the field of therapeutic protein production, process intensification strategies entailing higher starting cell seeding densities, can potentially increase culture productivity, lower cost of goods and improve facility utilization. However, increased cell densities often trigger apoptotic cell death at the end of the cell culture process and thus reduce total viable cell count. Apoptosis-resistant Chinese hamster ovary cell lines may offer the possibility to diminish this undesired outcome of the intensified production process. In this study, we have generated and tested Bax/Bak double-knock-out (DKO) apoptosis resistant hosts to express standard and bispecific antibodies, as well as complex molecules in intensified production processes both as pools and single cell clones, and at different scales. In all cases, therapeutic proteins expressed from clones or pools generated from the Bax/Bak DKO hosts showed not only better viability but also enabled extended productivity in the later stages of the 14-day intensified production process. The product qualities of the produced molecules were comparable between Bax/Bak DKO and wild type cells. Overall, we showed that Bax/Bak DKO apoptosis-resistant host cell lines significantly improve viability and volumetric productivity of the intensified production cultures without altering product qualities.

Expressing antigen binding fragments with high titers in a targeted integration CHO host by optimizing expression vector gene copy numbers and position: A case study.

Antigen binding fragments (Fab) are a promising class of therapeutics as they maintain high potency while having significantly smaller size relative to full-length antibodies. Because Fab molecules are aglycosylated, many expression platforms, including prokaryotic, yeast, and mammalian cells, have been developed for their expression, with Escherichia coli being the most commonly used Fab expression system. In this study, we have examined production of a difficult to express Fab molecule in a targeted integration (TI) Chinese Hamster Ovary (CHO) host. Without a need for extensive host or process optimization, as is usually required for E. coli, by simply using different vector configurations, clones with very high Fab expression titers were obtained. In this case, by increasing heavy chain (HC) gene copy numbers, clones with titers of up to 7.4 g/L in the standard fed-batch production culture were obtained. Our findings suggest that having a predetermined transgene integration site, as well as the option to optimize gene copy number/dosage, makes CHO TI hosts an effective system for expression of Fab molecules, allowing Fab expression using platform process and without significant process development efforts.

Preventing pyruvate kinase muscle expression in Chinese hamster ovary cells curbs lactogenic behavior by altering glycolysis, gating pyruvate generation, and increasing pyruvate flux into the TCA cycle.

Deletion of the pyruvate kinase muscle (PKM) gene, which is involved in conversion of phosphoenolpyruvate to pyruvate, has been shown to curb lactogenic behavior in Chinese hamster ovary (CHO) cells. This study describes the generation of pyruvate kinase muscle isoforms 1 and 2 knockout (PKM-KO) and pyruvate kinase muscle isoform-1 knockout (PKM1-KO) CHO host cells to understand metabolic shifts that reduce lactate secretion in these cells. Glucose and amino acids uptake levels in wild-type (WT), PKM-KO, and PKM1-KO stable cell lines, expressing two different antibodies, were analyzed in 14-day fed-batch production assays using different vessels. PKM-KO and PKM1-KO cells consumed more glucose per cell, altered amino acids metabolism, had higher flux of pyruvate into the tricarboxylic acid (TCA) cycle, and as previously shown reduced lactate secretion levels compared with the WT cells. Additionally, both PKM-KO and PKM1-KO cells had higher specific productivity and lower cell growth rates compared with the WT cells. Our findings suggest that rewiring the flux of pyruvate to the TCA cycle by deletion of PKM or PKM1 reduced cell growth and increased specific productivity in CHO cells. Overall, PKM1-KO cells had similar product quality and comparable or better titers relative to the WT cells, hence, targeted deletion of this isoform for curbing lactogenic behavior in CHO cells is suggested.

Development of a targeted integration Chinese hamster ovary host directly targeting either one or two vectors simultaneously to a single locus using the Cre/Lox recombinase-mediated cassette exchange system.

Cell line development (CLD) by random integration (RI) can be labor intensive, inconsistent, and unpredictable due to uncontrolled gene integration after transfection. Unlike RI, targeted integration (TI) based CLD introduces the antibody-expressing cassette to a predetermined site by recombinase-mediated cassette exchange (RMCE). The key to success for the development of a TI host for therapeutic antibody production is to identify a transcriptionally active hotspot that enables highly efficient RMCE and antibody expression with good stability. In this study, a genome wide search for hotspots in the Chinese hamster ovary (CHO)-K1-M genome by either RI or PiggyBac (PB) transposase-based integration has been described. Two CHO-K1-M derived TI host cells were established with the Cre/Lox RMCE system and are described here. Both TI hosts contain a GFP-expressing landing pad flanked by two incompatible LoxP recombination sites (L3 and 2L). In addition, a third incompatible LoxP site (LoxFAS) is inserted in the GFP landing pad to enable an innovative two-plasmid based RMCE strategy, in which two separate vectors can be targeted to a single locus simultaneously. Cell lines generated by the TI system exhibit comparable or higher productivity, better stability and fewer sequence variant (SV) occurrences than the RI cell lines.

Concurrent transfection of randomized transgene configurations into targeted integration CHO host is an advantageous and cost-effective method for expression of complex molecules.

Complex recombinant proteins are increasingly desired as potential therapeutic options for many disease indications and are commonly expressed in the mammalian Chinese hamster ovary (CHO) cells. Generally, stoichiometric expression and proper folding of all subunits of a complex recombinant protein are required to achieve the desired titers and product qualities for a complex molecule. Targeted integration (TI) cell line development (CLD), which entails the insertion of the desired transgene(s) into a predefined landing-pad in the CHO genome, enables the generation of a homogeneous pool of cells from which clonally stable and high titer clones can be isolated with minimal screening efforts. Despite these advantages, using a single transgene(s) configuration with predetermined gene dosage might not be adequate for the expression of complex molecules. The goal of this study is to develop a method for seamless screening of many vector configurations in a single TI CLD attempt. As testing vector configurations in transient expression systems is not predictive of protein expression in the stable cell lines and parallel TI CLDs with different transgene configurations is resource-intensive, we tested the concept of randomized configuration targeted integration (RCTI) CLD approach for expression of complex molecules. RCTI allows simultaneous transfection of multiple vector configurations, encoding a complex molecule, to generate diverse TI clones each with a single transgene configuration but clone specific productivity and product qualities. Our findings further revealed a direct correlation between transgenes' configuration/copy-number and titer/product quality of the expressed proteins. RCTI CLD enabled, with significantly fewer resources, seamless isolation of clones with comparable titers and product quality attributes to that of several parallel standard TI CLDs. Therefore, RCTI introduces randomness to the TI CLD platform while maintaining all the advantages, such as clone stability and reduced sequence variant levels, that the TI system has to offer.

Application of dynamic flux balance analysis to an industrial Escherichia coli fermentation.

We have developed a reactor-scale model of Escherichia coli metabolism and growth in a 1000 L process for the production of a recombinant therapeutic protein. The model consists of two distinct parts: (1) a dynamic, process specific portion that describes the time evolution of 37 process variables of relevance and (2) a flux balance based, 123-reaction metabolic model of E. coli metabolism. This model combines several previously reported modeling approaches including a growth rate-dependent biomass composition, maximum growth rate objective function, and dynamic flux balancing. In addition, we introduce concentration-dependent boundary conditions of transport fluxes, dynamic maintenance demands, and a state-dependent cellular objective. This formulation was able to describe specific runs with high-fidelity over process conditions including rich media, simultaneous acetate and glucose consumption, glucose minimal media, and phosphate depleted media. Furthermore, the model accurately describes the effect of process perturbations--such as glucose overbatching and insufficient aeration--on growth, metabolism, and titer.

Mechanisms of unintended amino acid sequence changes in recombinant monoclonal antibodies expressed in Chinese Hamster Ovary (CHO) cells.

An amino acid sequence variant is defined as an unintended amino acid sequence change and contributes to product heterogeneity. Recombinant monoclonal antibodies (MAbs) are primarily expressed from Chinese Hamster Ovary (CHO) cells using stably transfected production cell lines. Selections and amplifications with reagents such as methotrexate (MTX) are often required to achieve high producing stable cell lines. Since MTX is often used to generate high producing cell lines, we investigated the genomic mutation rates of the hypoxanthine-guanine phosphoribosyltransferase (HGPRT or HPRT) gene using a 6-thioguanine (6-TG) assay under various concentrations of MTX selection in CHO cells. Our results show that the 6-TG resistance increased as the MTX concentration increased during stable cell line development. We also investigated low levels of sequence variants observed in two stable cell lines expressing different MAbs. Our data show that the replacement of serine at position 167 by arginine (S167R) in the light chain of antibody A (MAb-A) was due to a genomic nucleotide sequence change whereas the replacement of serine at position 63 by asparagine (S63N) in the heavy chain of antibody B (MAb-B) was likely due to translational misincorporation. This mistranslation is codon specific since S63N mistranslation is not detectable when the S63 AGC codon is changed to a TCC or TCT codon. Our results demonstrate that both a genomic nucleotide change and translational misincorporation can lead to low levels of sequence variants and mistranslation of serine to asparagine can be eliminated by substituting the TCC or TCT codon for the S63 AGC codon without impacting antibody productivity.

Insulin degrading enzyme (IDE) expressed by Chinese hamster ovary (CHO) cells is responsible for degradation of insulin in culture media.

Chinese hamster ovary (CHO) cells cultured in serum-free chemically-defined media (CDM) are used for manufacturing of therapeutic proteins. Growth factors, such as insulin are commonly utilized in manufacturing platforms to enhance CHO cell viability and growth. Here we report that insulin is degraded in the culture media over time mainly due to the activity of the insulin degrading enzyme (IDE). Insulin degradation was faster in cell lines that released more IDE, which negatively impacted cell growth and in turn, production titers. Deletion of the IDE gene in a representative CHO cell line nearly abolished insulin degradation in seed train and end-of-production media. In summary, our data suggests that selecting cell lines that have lower IDE expression or targeted-deletion of the IDE gene can improve culture viability and growth for insulin-dependent CHO production platforms.

Maximizing antibody production in a targeted integration host by optimization of subunit gene dosage and position.

Historically, therapeutic protein production in Chinese hamster ovary (CHO) cells has been accomplished by random integration (RI) of expression plasmids into the host cell genome. More recently, the development of targeted integration (TI) host cells has allowed for recombination of plasmid DNA into a predetermined genomic locus, eliminating one contributor to clone-to-clone variability. In this study, a TI host capable of simultaneously integrating two plasmids at the same genomic site was used to assess the effect of antibody heavy chain and light chain gene dosage on antibody productivity. Our results showed that increasing antibody gene copy number can increase specific productivity, but with diminishing returns as more antibody genes are added to the same TI locus. Random integration of additional antibody DNA copies in to a targeted integration cell line showed a further increase in specific productivity, suggesting that targeting additional genomic sites for gene integration may be beneficial. Additionally, the position of antibody genes in the two plasmids was observed to have a strong effect on antibody expression level. These findings shed light on vector design to maximize production of conventional antibodies or tune expression for proper assembly of complex or bispecific antibodies in a TI system.

UBR E3 ligases and the PDIA3 protease control degradation of unfolded antibody heavy chain by ERAD.

Accumulation of unfolded antibody chains in the ER triggers ER stress that may lead to reduced productivity in therapeutic antibody manufacturing processes. We identified UBR4 and UBR5 as ubiquitin E3 ligases involved in HC ER-associated degradation. Knockdown of UBR4 and UBR5 resulted in intracellular accumulation, enhanced secretion, and reduced ubiquitination of HC. In concert with these E3 ligases, PDIA3 was shown to cleave ubiquitinated HC molecules to accelerate HC dislocation. Interestingly, UBR5, and to a lesser degree UBR4, were down-regulated as cellular demand for antibody expression increased in CHO cells during the production phase, or in plasma B cells. Reducing UBR4/UBR5 expression before the production phase increased antibody productivity in CHO cells, possibly by redirecting antibody molecules from degradation to secretion. Altogether we have characterized a novel proteolysis/proteasome-dependent pathway involved in degradation of unfolded antibody HC. Proteins characterized in this pathway may be novel targets for CHO cell engineering.

Endothelial intercellular cell adhesion molecule 1 contributes to cell aggregate formation in CHO cells cultured in serum-free media.

Chinese hamster ovary (CHO) cells have been adapted to grow in serum-free media and in suspension culture to facilitate manufacturing needs. Some CHO cell lines, however, tend to form cell aggregates while being cultured in suspension. This can result in reduced viability and capacity for single cell cloning (SCC) via limiting dilution, and process steps to mitigate cell aggregate formation, for example, addition of anti-cell-aggregation agents. In this study, we have identified endothelial intercellular cell adhesion molecule 1 (ICAM-1) as a key protein promoting cell aggregate formation in a production competent CHO cell line, which is prone to cell aggregate formation. Knocking out (KO) the ICAM-1 gene significantly decreased cell aggregate formation in the culture media without anti-cell-aggregation reagent. This trait can simplify the process of transfection, selection, automated clone isolation, and so on. Evaluation in standard cell line development of ICAM-1 KO and wild-type CHO hosts did not reveal any noticeable impacts on titer or product quality. Furthermore, analysis of a derived nonaggregating cell line showed significant reductions in expression of cell adhesion proteins. Overall, our data suggest that deletion of ICAM-1 and perhaps other cell adhesion proteins can reduce cell aggregate formation and improve clonality assurance during SCC.

Utilizing a regulated targeted integration cell line development approach to systematically investigate what makes an antibody difficult to express.

Chinese hamster ovary (CHO) cells are conventionally used to generate therapeutic cell lines via random integration (RI), where desired transgenes are stably integrated into the genome. Targeted integration (TI) approaches, which involve integration of a transgene into a specific locus in the genome, are increasingly utilized for CHO cell line development (CLD) in recent years. None of these CLD approaches, however, are suitable for expression of toxic or difficult-to-express molecules, or for determining the underlying causes for poor expression of some molecules. Here we introduce a regulated target integration (RTI) system, where the desired transgene is integrated into a specific locus and transcribed under a regulated promoter. This system was used to determine the underlying causes of low protein expression for a difficult-to-express antibody (mAb-A). Interestingly, we observed that both antibody heavy chain (HC) and light chain (LC) subunits of mAb-A independently contributed to its low expression. Analysis of RTI cell lines also revealed that while mAb-A LC triggered accumulation of intracellular BiP, its HC displayed impaired degradation and clearance. RTI pools, generated by swapping the WT or point-mutant versions of difficult-to-express antibody HC and LC with that of an average antibody, were instrumental in understanding the contribution of HC and LC subunits to the overall antibody expression. The ability to selectively turn off the expression of a target transgene in an RTI system could help to directly link expression of a transgene to an observed adverse effect. © 2018 American Institute of Chemical Engineers Biotechnol. Prog., 35: e2772, 2019.

Pyruvate Kinase Muscle-1 Expression Appears to Drive Lactogenic Behavior in CHO Cell Lines, Triggering Lower Viability and Productivity: A Case Study.

Chinese hamster ovary (CHO) cell lines are used to express a variety of therapeutic proteins. However, lactogenic behavior displayed by some CHO cell lines during manufacturing processes may result in a decline in viability, productivity, and possible alterations in product quality. In cultured cells, lactate is produced during glycolysis through irreversible conversion of phosphoenolpyruvate to pyruvate and then lactate via sequential function of pyruvate kinase and lactate dehydrogenase (LDH) enzymes. In the process of cell line development (CLD), two lactogenic cell lines expressing different antibody molecules are identified. The lactogenic behaviors of these cell lines can be differentially mitigated through optimization of either nutrient feeds or culture pH, depending on the cell line. Analysis of various proteins involved in the glycolysis pathway reveal a direct correlation between the pyruvate kinase muscle-1 (PKM-1) isoform levels and lactogenic behavior. CRISPR mediated knockout of the PKM-1 isoform abolishes lactate accumulation even under lactogenic conditions. Furthermore, a cell line lacking expression of both PKM-1 and PKM-2 enzymes capable of maintaining productivity, viability, and growth without displaying lactogenic behavior is identified. Targeted deletion of PKM in CHO cells may be tolerated due to expression of PKL (liver) and PKR (red blood cell) isoforms of pyruvate kinase. All together, these findings suggest that PKM-1 up-regulation during antibody production could trigger lactogenic behavior and that this enzyme is dispensable for CHO cell survival.

Development and characterization of an automated imaging workflow to generate clonally-derived cell lines for therapeutic proteins.

In the development of biopharmaceutical products, the expectation of regulatory agencies is that the recombinant proteins are produced from a cell line derived from a single progenitor cell. A single limiting dilution step followed by direct imaging, as supplemental information, provides direct evidence that a cell line originated from a single progenitor cell. To obtain this evidence, a high-throughput automated imaging system was developed and characterized to consistently ensure that cell lines used for therapeutic protein production are clonally-derived. Fluorescent cell mixing studies determined that the automated imaging workflow and analysis provide ∼95% confidence in accurately and precisely identifying one cell in a well. Manual inspection of the images increases the confidence that the cell line was derived from a single-cell to >99.9%. © 2017 American Institute of Chemical Engineers Biotechnol. Prog., 34:584-592, 2018.

Supplementation of Nucleosides During Selection can Reduce Sequence Variant Levels in CHO Cells Using GS/MSX Selection System.

In the process of generating stable monoclonal antibody (mAb) producing cell lines, reagents such as methotrexate (MTX) or methionine sulfoximine (MSX) are often used. However, using such selection reagent(s) increases the possibility of having higher occurrence of sequence variants in the expressed antibody molecules due to the effects of MTX or MSX on de novo nucleotide synthesis. Since MSX inhibits glutamine synthase (GS) and results in both amino acid and nucleoside starvation, it is questioned whether supplementing nucleosides into the media could lower sequence variant levels without affecting titer. The results show that the supplementation of nucleosides to the media during MSX selection decreased genomic DNA mutagenesis rates in the selected cells, probably by reducing nucleotide mis-incorporation into the DNA. Furthermore, addition of nucleosides enhance clone recovery post selection and does not affect antibody expression. It is further observed that nucleoside supplements lowered DNA mutagenesis rates only at the initial stage of the clone selection and do not have any effect on DNA mutagenesis rates after stable cell lines are established. Therefore, the data suggests that addition of nucleosides during early stages of MSX selection can lower sequence variant levels without affecting titer or clone stability in antibody expression.

Beating the odds: The poisson distribution of all input cells during limiting dilution grossly underestimates whether a cell line is clonally-derived or not.

Establishing that a cell line was derived from a single cell progenitor and defined as clonally-derived for the production of clinical and commercial therapeutic protein drugs has been the subject of increased emphasis in cell line development (CLD). Several regulatory agencies have expressed that the prospective probability of clonality for CHO cell lines is assumed to follow the Poisson distribution based on the input cell count. The probability of obtaining monoclonal progenitors based on the Poisson distribution of all cells suggests that one round of limiting dilution may not be sufficient to assure the resulting cell lines are clonally-derived. We experimentally analyzed clonal derivatives originating from single cell cloning (SCC) via one round of limiting dilution, following our standard legacy cell line development practice. Two cell populations with stably integrated DNA spacers were mixed and subjected to SCC via limiting dilution. Cells were cultured in the presence of selection agent, screened, and ranked based on product titer. Post-SCC, the growing cell lines were screened by PCR analysis for the presence of identifying spacers. We observed that the percentage of nonclonal populations was below 9%, which is considerably lower than the determined probability based on the Poisson distribution of all cells. These results were further confirmed using fluorescence imaging of clonal derivatives originating from SCC via limiting dilution of mixed cell populations expressing GFP or RFP. Our results demonstrate that in the presence of selection agent, the Poisson distribution of all cells clearly underestimates the probability of obtaining clonally-derived cell lines. © 2017 American Institute of Chemical Engineers Biotechnol. Prog., 34:559-569, 2018.