Fred A.M. Asselbergs

Design of a genome-wide siRNA library using an artificial neural network

The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.

Histone deacetylase 1 represses the small GTPase RhoB expression in human nonsmall lung carcinoma cell line

The dynamic balance between histone acetylation and deacetylation plays a significant role in the regulation of gene transcription. Much of our current understanding of this transcriptional control comes from the use of HDAC inhibitors such as trapoxin A (TPX), which leads to hyperacetylated histone, alters local chromatin architecture and transcription and results in tumor cell death. In this study, we treated tumor cells with TPX and HDAC1 antisense oligonucleotides, and analysed the transcriptional consequences of HDAC inhibition. Among other genes, the small GTPase RhoB was found to be significantly upregulated by TPX and repressed by HDAC1. The induction of RhoB by HDAC inhibition was mediated by an inverted CCAAT box in the RhoB promoter. Interestingly, measurement of RhoB transcription in ∼130 tumor-derived cell lines revealed low expression in almost all of these samples, in contrast to RhoA and RhoC. Accumulating evidence indicates that the small GTPase Rho proteins are involved in a variety of important processes in cancer, including cell transformation, survival, invasion, metastasis and angiogenesis. This study for the first time demonstrates a link between HDAC inhibition and RhoB expression and provides an important insight into the mechanisms of HDAC-mediated transcriptional control and the potential therapeutic benefit of HDAC inhibition.

Amplification and expression of recombinant genes in serum-independent Chinese hamster ovary cells

CHO SSF3 cells grow as a suspension culture in unmodified commercial medium with only low–molecular weight ingredients. Continuous serum–free culture unexpectedly induced expression of a low dihydrofolate reductase activity in the originally dhfr– CHO cells. Nevertheless, it was possible with methotrexate to induce amplification of a gene coding for the hybrid plasminogen activator K2tu–PA cotransfected with a dhfr gene. Expression of K2tu–PA expression was proportionally increased to that of dhfr, which was measured with fluorescent methotrexate. Because no serum proteases were present, secreted K2tu–PA was not converted to the enzymatically active form, but was exclusively recovered in proenzyme form.

Scaled-up production of recombinant human renin in CHO cells for enzymatic and X-ray structure analysis.

A process was developed to produce recombinant human renin for X-ray analysis and enzyme inhibition studies. An expression vector containing a human prorenin cDNA and expressing a mouse dihydrofolate reductase selection marker was transfected into dhfr-minus Chinese hamster ovary cells. After selection of cell strains with an increased gene copy number with methotrexate, cultures of the recombinant cells were scaled-up in serum-free media. Major improvements in cellular productivity were achieved by using continuous suspension cultures with cell recycling instead of an adherent culture system or batch-mode suspension cultures. The recombinant zymogen prorenin was purified and preparatively activated with trypsin. Enzymatic properties of the recombinant active renin are described.

Endogenous gene and amplifiable cDNA construct both produce unstable t-PA mRNA in Bowes melanoma cells

Bowes melanoma cells, which naturally produce tissue-type plasminogen activator (t-PA), were transfected with a plasmid containing a human t-PA cDNA under transcriptional control of the promoter/enhancer of the major immediate early gene of human cytomegalovirus (CMV) plus genes expressing geneticin (G418) resistance and dihydrofolate reductase (DHFR). In one of the initial geneticin-resistant transformants, t-PA mRNA transcribed from the chromosomally integrated plasmid had the same short half-life, 20-30 min, as did mRNA transcribed from the endogenous t-PA gene compared to 7-8 h for total poly(A)+ mRNA. After subsequent selection of such cells with methotrexate, a cell line was obtained in which the t-PA cDNA construct was co-amplified with the DHFR gene and which produced 10 times more t-PA protein than the original Bowes melanoma cells.

Use of the Escherichia coli chromosomal DHFR gene as selection marker in mammalian cells

The folA gene, the chromosomal dhfr gene of Escherichia coli, was engineered for expression in mammalian cells. In contrast to plasmid-derived bacterial dhfr genes previously used as selection markers in mammalian cells, the folA gene product is inhibitable by methotrexate (MTX) and trimethoprim (TMP). Therefore, this dhfr may present an alternative to mammalian dhfr species currently used as amplifiable selection markers. Transfected E. coli folA dhfr could complement the lack of endogenous DHFR in Chinese hamster ovary (CHO) cells lacking a functional dhfr gene. Both MTX and TMP inhibited growth of E. coli folA dhfr-transfected CHO cells. Expression of E. coli folA DHFR could be visualized by incubating the transfected cells with a fluorescent methotrexate derivative (F-MTX). Binding of F-MTX to E. coli folA DHFR was inhibitable as by both MTX and TMP, whereas MTX but not TMP blocked binding of F-MTX to recombinant mouse DHFR.

Transformation of Bowes melanoma cells with SV40 T antigen

A serum-dependent and two serum-independent variants of the Bowes melanoma cell line, RPMI7272, were transfected with plasmids containing a geneticin-resistance (neo) gene transcribed by the HSV thymidine kinase promoter and an SV40 T antigen gene under control of the mouse metallothionein I promoter. T-antigen increased the cloning efficiency of the serum-dependent cell line in soft-agar more than 50-fold, but cloning efficiency of serum-independent lines was not increased. Trypsinization of serum-independent lines required 100 times lower concentrations of trypsin than serum-dependent cells. Human metal-inducible T-antigen-producing (HMT) melanoma cells supported replication of transfected plasmids containing an SV40 origin of replication. Transient expression of interferon or plasminogen activator from such plasmids was 40-fold higher than in untransformed melanoma cells and could be enhanced 30-fold more by stimulation of transcription of the T antigen gene with cadmium chloride. HMT cells can be grown in suspension and thus may represent an attractive alternative to monkey kidney COS cells.

Production of the chimerical plasminogen activator K2tu-PA in CHO cells

Development of a CHO cell-based production system for the hybrid plasminogen activator K2tu-PA is described. Using the major immediate-early promoter of mouse cytomegalovirus (MCMV) transient and stable expression levels were 3-10-fold higher than those obtained with several other strong promoters. Splicing and polyadenylation signals from the rabbit beta-globin gene were used downstream of the DNA segment coding for K2tu-PA. The strong enhancer moiety of the MCMV promoter also stimulated strongly the promoter of the dihydrofolate reductase (DHFR) gene, placed adjacently for selection/gene amplification purposes. One construct with opposing K2tu-PA and DHFR RNA transcripts yielded the highest expression level with a single copy of the plasmid, but K2tu-PA expression was consistently lost after amplification of such genes, possibly as a result of the formation of antisense RNA. With other constructs, K2tu-PA production leveled off at 6.5 micrograms per million cells per day despite a high gene copy number. This was due to a combination of inefficient mRNA translation and mRNA instability, caused by elements from the untranslated portions of tissue-type and urokinase-type plasminogen activator cDNA which were included in the expression vector. After elimination of these inhibitory DNA segments, 4-5-times higher expression levels were reached.

Corrigendum: Design of a genome-wide siRNA library using an artificial neural network

Nat. Biotechnol. 23, 995–1001 (2005), published online 17 July 2005; corrected after print 25 July 2006 In the Methods section, p. 1,000, col. 2, paragraph 2, the text beginning: “The whole plate was discarded if for any time point...” and ending, “The final data set contained 2,431 sequences.” inaccurately described the procedure.

Conditionally adherent growth of serum-independent CHO cells for automated drug screening and biopharmaceutical production.

SSF3 is a CHO cell line adapted for growth in protein-free medium. It grows in suspension unless serum-derived attachment factors such as vitronectin are added to the medium. Serum-independent cell lines, which adhere to the substrate after induction with dexamethasone or constitutively, were created by transfection with a human vitronectin gene under control of the mouse mammary tumor-virus promoter. Substrate attachment and SSF3VN-cell spreading could be prevented with an RGD peptide (arginine-glycine-aspartic acid) confirming that attachment is mediated by an intregrin receptor. Hormone-inducible attachment could be blocked by glucocorticoid antagonist promegestone. All steps in the isolation of stable transfected SSF3VN cell lines could be done in a chemically defined medium avoiding the risk of introduction of serum-derived infectious agents. SSF3VN cells could be grown in protein-free medium in solid-phase large-scale bioreactors. Application in microplates as used in high-throughput screening was demonstrated in an assay of Ca(2+) release from internal stores induced by agonist-binding to recombinant human metabotropic glutamate receptor hmGluR1b.