A cross-species whole genome siRNA screen in suspension-cultured Chinese hamster ovary cells identifies novel engineering targets.

High-throughput siRNA screens were only recently applied to cell factories to identify novel engineering targets which are able to boost cells towards desired phenotypes. While siRNA libraries exist for model organisms such as mice, no CHO-specific library is publicly available, hindering the application of this technique to CHO cells. The optimization of these cells is of special interest, as they are the main host for the production of therapeutic proteins. Here, we performed a cross-species approach by applying a mouse whole-genome siRNA library to CHO cells, optimized the protocol for suspension cultured cells, as this is the industrial practice for CHO cells, and developed an in silico method to identify functioning siRNAs, which also revealed the limitations of using cross-species libraries. With this method, we were able to identify several genes that, upon knockdown, enhanced the total productivity in the primary screen. A second screen validated two of these genes, Rad21 and Chd4, whose knockdown was tested in additional CHO cell lines, confirming the induced high productivity phenotype, but also demonstrating the cell line/clone specificity of engineering effects.

CUG binding protein 1 binds to a specific region within the human albumin 3' untranslated region.

3' Untranslated regions (3'UTRs) of messenger RNAs have important roles in post-transcriptional regulation of gene expression and this is partly achieved through binding of specific proteins to sequences or structures within these regions. Previously, replacement of a native luciferase 3'UTR with the human albumin 3'UTR has been found to lead to a 10-fold increase in luciferase reporter activity. In this work we investigated protein binding to the human albumin 3'UTR. Electrophoretic mobility shift and UV cross-linking assays indicate that a ∼50kDa protein from Chinese Hamster Ovary (CHO) cells binds to the albumin 3'UTR, and affinity experiments followed by proteomics identified this protein as CUG binding protein 1 (CUG-BP1, also known as CELF1). Deletion analysis of the albumin 3'UTR showed that nucleotides 1-50 and nucleotides 101-150 are not required for binding but that removal of nucleotides 51-100 caused a loss in binding. The results suggest that CUG-BP1 binds to nucleotides 51-100 of the human albumin 3'UTR. In human cells CUG-BP1 binding may thus play a role in regulation of albumin expression and, additionally, it may have a function in post-transcriptional control in CHO cells.

Albumin 3'untranslated region facilitates increased recombinant protein production from Chinese hamster ovary cells.

Chinese hamster ovary (CHO) cells are used for recombinant protein production in the pharmaceutical industry but there is a need to improve expression levels. In the present work experiments were carried out to test the effectiveness of different 3'untranslated regions (3'UTRs) in promoting production of a naturally secreted luciferase. Seamless cloning was used to produce expression vectors in which Gaussia princeps luciferase coding sequences were linked to the human albumin, immunoglobulin or chymotrypsinogen 3'UTR. Stably transfected CHO cells expressing these constructs were selected. Luciferase activity in the culture medium was increased 2-3-fold by replacing the endogenous 3'UTR with the albumin 3'UTR and decreased by replacement with immunoglobulin or chymotrypsinogen 3'UTR. Replacement of the native 3'UTR with the albumin 3'UTR led to a 10-fold increase in luciferase mRNA levels. Deletion analysis of the albumin 3'UTR showed that loss of nucleotides 1-50, which removed an AU-rich complex stem loop region, caused significant reductions in both luciferase protein expression and luciferase mRNA levels. The results suggest that recombinant protein expression and yield could be improved by the careful selection of appropriate 3'UTR sequences.

CHO-K1 host cells adapted to growth in glutamine-free medium by FACS-assisted evolution.

During the process of recombinant cell line optimisation for production of biopharmaceuticals, multiple cellular properties like robustness against stress, the attainment of high cell concentrations and maintenance of high viability must be considered to maximize protein yield. To improve growth and viability, glutamine is supplemented as an alternative energy source for rapidly dividing cells that oxidize glucose inefficiently. However, the resulting by-product ammonia is toxic at high concentrations and has a negative impact on protein glycosylation, a major quality-determining parameter of biopharmaceuticals. In this work, the CHO-K1 cell line was adapted to a chemically defined medium and suspension growth within 3 weeks. Subsequently, the glutamine concentration was stepwise reduced from 8 to 4 and 2 mM. After each reduction, both the final cell concentration in the batch and the viability decreased. To force a rapid evolution of cells to achieve high final cell concentrations, cells were seeded at high densities (10(7) cells/mL) and surviving cells were sorted by FACS or MACS when viability declined to 10% (typically after 24 h). Sorted cells were grown in batch until viability declined to 10% and viable cells recovered again. The final sorted population was able to reach comparable or even better viable cell concentrations and showed a significantly improved viability compared to their ancestors. The 2 mM glutamine-adapted cell line was directly transferred into glutamine-free medium and was able to grow at comparable rates without requiring further adaptation. Cells compensated the lack of glutamine by increasing their consumption of glutamate and aspartate.

The absence of effect of gene copy number and mRNA level on the amount of mAb secretion from mammalian cells.

Recombinant human antibody production represents a major growing class of biopharmaceuticals based on the technological progress within the last decades especially in CHO cells. The HIV neutralizing human monoclonal antibody 2F5 was developed as hybridoma from human lymphocyte preparations. In order to estimate the potential of recombinant 2F5-expressing CHO cells, we generated different recombinant CHO cell lines by varying regulatory sequences, the codon usage, the signal peptides, and the transfection technique. These 2F5-expressing cell lines were developed by selection of the best producer, clone homogeneity, and clone stability. The gene copy number of the clones differed significantly due to methotrexate amplification. In one cell line, we identified only one copy of heavy chain and two copies of light chain. Neither the gene copy number nor the promoter was found to influence the amount of transcript exclusively emphasizing the positioning effect of the transgene. Messenger RNA levels were highest in 2F5/CO and may have resulted from a combination of the promoter and codon-optimized sequences, but unexpectedly, the amount of secreted product was not elevated in this configuration. In our example, translational and post-translational limitations are responsible for decreased antibody secretion.

The level of synthesis and secretion of Gaussia princeps luciferase in transfected CHO cells is heavily dependent on the choice of signal peptide.

There is a great demand for the improvement of mammalian cell production systems such that they can compete economically with their prokaryotic counterparts. Of a number of parameters that need to be explored to accomplish this we have tested the effects of different signal peptides on the synthesis and secretion of Gaussia princeps luciferase in mammalian cells. A series of plasmids were transfected into CHO cells where the coding region for the marine luciferase was fused to the signal peptide coding regions derived from different sources. Both cell extracts and medium samples were analysed for luciferase activity. When the native Gaussia luciferase signal sequence in the vector was substituted by that from human interleukin-2 or albumin then the amount of active recombinant protein produced was substantially reduced, both in transiently and stably transfected cells. Western blotting showed that enzyme activity and protein levels mirrored one another. The major decrease in luciferase activity was shown not to be a result of decreased mRNA levels, indicating the involvement of a post-transcriptional event. When the coding region of human endostatin was fused to that of the Gaussia luciferase signal peptide then an elevated level of secreted endostatin was observed compared to when that of the albumin signal peptide was used. Stable transfection of HepG2 cells with the different signal peptide constructs gave essentially the same results as seen in CHO cells. The overall results indicate that the choice of signal peptide can be imperative to ensure an optimal synthesis and secretion of a recombinant protein in a mammalian cell culture system.

Vectors encoding seven oikosin signal peptides transfected into CHO cells differ greatly in mediating Gaussia luciferase and human endostatin production although mRNA levels are largely unaffected.

The signal peptide of the luciferase secreted by the marine copepod Gaussia princeps has been shown to promote high-level protein synthesis/secretion of recombinant proteins, being far superior to mammalian counterparts. The main aim of the present study was to investigate the effects of seven selected signal peptides derived from oikosins, house proteins of the marine organism Oikopleura dioica, on synthesis/secretion of recombinant proteins. Vector constructs were made in which the coding regions of two naturally secreted proteins, Gaussia luciferase and human endostatin (hEndostatin), were "seamlessly" fused to the signal peptide coding sequences of interest. CHO cells were transfected with the plasmids and populations of stably transfected cells established. The amounts of reporter proteins in cell extract and medium samples were determined and the results compared to those obtained from cells stably transfected with a reference vector construct. In addition, the amounts of luciferase or hEndostatin encoding mRNAs in the cells were determined and related to the protein levels obtained. The levels of reporter protein produced varied greatly among the seven oikosin signal peptides tested. Whereas the oikosin 1 signal peptide resulted in about 40% production of Gaussia luciferase compared to the reference construct, oikosins 2-7 were extremely ineffective (<1%). mRNA levels were not dramatically affected such that inadequate availability of transcript for translation was not the underlying reason for the observations. The oikosin 1 signal peptide was also the most effective regarding synthesis/secretion of hEndostatin. No secreted product was observed using the oikosin 3 signal peptide. Interestingly, the molecular weight of hEndostatin in cell extracts prepared from cells transfected with oikosin 2 and 3 constructs was higher than that using the oikosin 1 signal peptide. The overall findings indicate that the signal peptide affects the efficiency of protein synthesis and secretion through a mechanism operating at the post-transcriptional level. The results described here provide substantial support to our previous observations which suggested that the choice of the signal peptide is imperative when aiming to achieve optimal synthesis and secretion of a recombinant protein using transfected mammalian cells.