- Efficient and reproducible generation of high-expressing, stable human cell lines without need for antibiotic selection

Increased and detailed knowledge about the precise carbohydrate structures has revealed the importance of post-translational modifications for functionality of therapeutic proteins. However, these findings also exposed that the list of modifications is frighteningly long and many of them might affect the immunogenicity, stability, pharmacokinetics and thus efficacy of the protein. The most important capability that distinguishes mammalian cells from other expression systems is N- and O-linked glycosylation and it is assumed that around 2% of the human genome encodes proteins that contribute to glycosylation [28]. Although almost any mammalian cell line possesses the machinery to produce and secrete proteins, only a limited number meet the fermentation requirements and thus can be used for industrial manufacturing: Chinese hamster ovary (CHO), baby hamster kidney (BHK) and mouse myeloma cells NS0 and Sp2/0.

Glycan structures differ significantly among different cell types and species. Only human cells lines promise to produce proteins in a way that species-specific and thus immunogenic differences in glycosylation are absent. For safety and regulatory reasons human cells used for production should not be of tumor origin. In addition, only a limited number of human cell lines are available to date. The development of permanent cell lines using primary human cells and recombinant DNA techniques has been hampered by the fact that human cells are highly resistant to transformation by viral functions. In fact, from human tissue sources only human embryonic kidney (HEK) cells [17], human embryonic lung (HEL) cells [29], human embryonic retinoblasts (HER cells) [30,19] and primary human amniocytes [20] have been successfully transformed with adenoviral functions. For practical and ethical reasons it is very difficult to obtain primary cells from fetal origin. Primary human amniocytes are the only cell type that is readily available without ethical concerns: they can be collected by routine amniocentesis. Human amniocytes can be cultivated as adherent cultures for several passages under standard conditions and consist of three main cell types, fibroblast-like, epithelial-like and amniotic fluid (AF) cells [31]. Recently it has been shown, that about 1% of cells found in amniotic fluid are amniotic-fluid derived stem (AFS) cells [32]. Considering the low transfection efficiency (< 1% in the present analyses) of human amniocytes, the transformation efficiency using an E1-expressing plasmid is surprisingly high. In earlier studies we have calculated to obtain at least one or two transformants in 1 × 105 cells transfected [20]. Even though in the present analyses we have transfected two plasmids, one containing the E1 and pIX genes and a second expressing hAAT with a plasmid ratio of 1:1, the transformation efficiency was comparable to transfection with the E1-plasmid alone (data not shown).

Cotransfection of two plasmids expressing the transforming E1-functions and hAAT, respectively, resulted in stably transformed cell pools, with 6 out of 10 pools expressing hAAT. Moreover, 4 out of 6 cell pools show long-lasting expression for more than 35 passages, with expression levels up to 6 μg/ml or 8 pg/cell/day. Cell cloning by limited dilution from two cell pools resulted in genetically identical cell lines which show stable expression of hAAT for more than 65 passages (the time course of the present study) with expression levels up to 30 pg/cell/day. Since cloning of cells was started in passage 26, this indicates stable expression of hAAT for more than 90 passages. Again it has to be emphasized that this stable protein expression was achieved without any antibiotic selection.

The majority of the mammalian genome is transcriptionally silent. Since integration of transfected plasmids occurs randomly, the position effects generally manifest partial or complete loss of expression. In order to overcome the position effect numerous regulatory elements introduced into the plasmid expressing the gene of interest have been tested including insulators, MARs, strong promoters and enhancers. Even when using such elements a time consuming testing of numerous cell clones over multiple passages cannot be avoided. Other crucial parameters like the choice of promoters for expressing the gene of interest and the marker gene, the ratio of gene of interest to marker gene expressing plasmid, concentration of selection marker in the medium or using linear versus circular plasmids considerably influence cell line development and expression levels.

In common strategies to select for highly expressing cell lines, the marker gene is linked to the gene of interest and employing a selection strategy can circumvent the problem of silencing. However, even under selective pressure only a very minor number of cell clones yield in high and stable expression of the gene of interest. In addition, the selective pressure has to be maintained throughout cell line development. Currently, aminoglycoside antibiotics such as hygyromycin B and Geneticin are frequently used. These substances interfere with protein translation and exhibit highly toxic effects in mammalian cells not containing the corresponding bacterial genes. However, these antibiotics are also described to have undesired side effects like increasing frequency of sister chromatid exchange [33] and altering expression of glucose-regulated genes [34].

A different very popular strategy for high-level protein production is based on the use of dhfr-deficient CHO cells in combination with expression vectors carrying a functional dhfr gene and the gene of interest. Cultivation of these transfected cells in methotrexate (MTX) containing medium results in amplification of the dhfr gene and gene of interest sequences and thus high-level expression of the gene of interest. The disadvantage of this production system however is, that cells grown in the presence of MTX often show substantial heterogenicity in chromosomal location and copy number of amplified sequences, they exhibit rearrangement and highly variable amplification of transfected sequences, and they contain chromosomes with highly extended regions, chromosomes joined at amplified regions or even circular chromosomes consisting entirely of exogenous DNA. Thus the use of the dhfr-amplification system very often results in a high degree of genetic instability in the production cell line [35].

As described earlier, the permanent expression of E1-functions is crucial for maintaining the transformed character of stable cell lines [36]. Thus, the E1-functions are replacing the selection marker and prevent the repressive effect of the surrounding heterochromatin. Earlier analyses have shown, that transfection of two plasmids into CHO cells results in co-integration at a common site in all clones examined [37]. Since in the present analyses 4 out of 6 cell pools show long-lasting expression of hAAT, we assume that the hAAT-expressing plasmid has co-integrated in a transcriptionally active site. Single cell cloning of the cell pools resulted in stable cell lines exhibiting very high protein expression of up to 30 pg/cell/day. The E1-functions have been shown to increase expression from several promoters including the CMV promoter [38-40] used in the present analyses, which might contribute to the high expression levels.

Primary human amniocytes are efficiently transformed by adenoviral E1-functions. Based on this observation, we exhibit an improved method for developing high protein expressing stable human cell lines. These cell lines can be easily adapted to serum-free suspension culture by gradually exchanging the medium to a serum-free, chemically defined medium for suspension cells (data not shown). Moreover, hAAT expressed as reference protein in human amniocyte cell lines is fully glycosylated and sialylated. In future experiments we plan to perform a more detailed analyses of the glycan structure of hAAT expressed in human amniocyte cell lines in comparison to the protein expressed in CHO cells.

Industrial protein expression demand short time lines for cell line development, use of chemically derived serum free medium, growth in suspension and the possibility to scale up production process. Only during the very early passages the primary amniocytes depend on fetal serum but are soon transferred to chemically derived serum-free medium. For technical reasons transformation and isolation of permanent cell clones occurs in adherent culture, and thus an additional step for adaptation to growth in suspension has to be admitted. We have started additional experiments in order to simplify and speed up this adaptation step and thus shorten the time frame for cell line development.

The potential of the present novel method of cell line development however is not restricted to production of biopharmaceuticals. For example, cotransfection of primary amniocytes with the E1-expressing plasmid and a second plasmid expressing SV40 T-antigen or Epstein-Barr virus EBNA-1 protein would result in cell lines that would be optimized for transient protein expression. Moreover, overexpression of glycosylation enzymes like the α2–6 sialyltransferase would result in cells optimized for production of glycosylated proteins with high sialic acid content. Additional examples would be the expression of certain viral proteins in cells lines for improved production of viruses for vaccination or gene therapy. The current method would also allow the insertion of certain DNA-sequences like FLP recombinase targets (FRT) sites by simply introducing FRT sites in the E1-expressing plasmid. By cotransfecting such new cells with a plasmid expressing the gene of interest flanked by FRT sites and a FLP expressing plasmid, a recombinase mediated cassette exchange would occur. The insertion of the gene of interest would then occur at a predicted site and thus would drastically simplify cell line development.

rating: 3.00 from 2 votes | updated on: 29 Nov 2008 | views: 17367 |

Rate article:

excellent! bad…