Into the unknown: expression profiling without genome sequence information in CHO by next generation sequencing

Abstract

The arrival of next-generation sequencing (NGS) technologies has led to novel opportunities for expression profiling and genome analysis by utilizing vast amounts of short read sequence data. Here, we demonstrate that expression profiling in organisms lacking any genome or transcriptome sequence information is feasible by combining Illumina's mRNA-seq technology with a novel bioinformatics pipeline that integrates assembled and annotated Chinese hamster ovary (CHO) sequences with information derived from related organisms. We applied this pipeline to the analysis of CHO cells which were chosen as a model system owing to its relevance in the production of therapeutic proteins. Specifically, we analysed CHO cells undergoing butyrate treatment which is known to affect cell cycle regulation and to increase the specific productivity of recombinant proteins. By this means, we identified sequences for >13,000 CHO genes which added sequence information of approximately 5000 novel genes to the CHO model. More than 6000 transcript sequences are predicted to be complete, as they covered >95% of the corresponding mouse orthologs. Detailed analysis of selected biological functions such as DNA replication and cell cycle control, demonstrated the potential of NGS expression profiling in organisms without extended genome sequence to improve both data quantity and quality.

Figure 1.

The bioinformatics workflow used for CHO data processing. In the first step, reads are assembled by two different strategies: (i) a de novo approach using Velvet starting with all reads and (ii) a knowledge-based approach using all reads mapping to a mouse gene (or the corresponding rat ortholog) again by applying Velvet. After computation and annotation of the final CHO transcriptome assembly, the reads that mapped against all three reference sequence datasets were used for gene expression profiling.

Figure 2.

Coverage of mouse Ensembl transcripts by CHO contig sequences. More than 6000 transcripts can be expected to be nearly complete with a coverage of >95% with respect to mouse. The average mouse transcript coverage is 66.9%.

Figure 3.

Overlap of Ensembl genes detectable on the custom Affymetrix CHO microarray and detectable using the CHO assembly. Please note that not all genes detectable by NGS during expression profiling are contained in the assembly, e.g. because they are expressed at too low levels. In those cases orthologous to mouse or rat can be detected by the mapping strategy and allow for profiling the expression of the corresponding CHO genes.

Figure 4.

Cultivation data of recombinant CHO fed batch process. Normalized viable cell concentration (VCC; normalization to maximal VCC of control process without butyrate addition), cell viability and recombinant IgG product concentration are shown. Butyrate was added to the culture at d = 5.25 as indicated by the arrow. Sampling times for RNA extraction were d = 0, 4, 6 and 8.

Figure 5.

Changes in gene expression mediated by 1.0mM butyrate compared to the control group on Day 6 of the experiment. Results from gene expression analysis of cells treated with 1.0mM butyrate compared to the control on Day 6 of the experiment (18 h after butyrate addition). In the networks shown, genes marked in red are down-regulated in butyrate-treated cells whereas genes shown in green are up-regulated. Pathways and networks are taken from Ingenuity Pathway Analysis. (a) Top three functional categories and number of regulated genes detected by NGS and by Affymetrix, indicating that NGS provides a more detailed look into transcriptional changes mediated by butyrate. (b) and (c) Regulation of cell cycle genes through butyrate. (b) Up-regulation of BTG2 is known to lead to a down-regulation of the Cyclin E/CDK2 complex as well as Cyclin D1 (blue arrows). (c) Pathway of genes involved in mitotic check point control. (d) and (e) Example for the high consistency of differential gene expression detected by CHO NGS shown by the down-regulation of MCM genes at Day 6 in cells treated with 1.0mM butyrate. The MCM complex and associated genes play a role in the initiation and the elongation phases of eukaryotic DNA replication, specifically the formation and the elongation of the replication fork. (d) shows down-regulated genes as detected by the NGS approach while part (e) shows deregulated genes detected by the Affymetrix CHO chip. Several genes (MCM2, MCM4, MCM8 and CDT1) are not measured on the chip but fit well into the complete down-regulation of the complex.