Authors

Eva Adriaensen (OA), Griet Laenen (OA), Vikki Keersmaekers (JNJ) (?), Marjolein Crabbe (JNJ), Hans De Wolf (JNJ), Bie Verbist (JNJ)
Others? (Acknowledgement: Dea Putri?, Maaren Derks?)

Abstract

Introduction

Gene \cite{D_Ambrosio_2017}expression\cite{Ritz_2015} , through the quantification of mRNA is commonly used in biomedical research for patient diagnostics and/or therapeutics \cite{Pabinger_2014} \cite{Oh_2017} \cite{El_Hadi_2017}. The quantification of mRNA is routinely performed using real-time quantitative PCR (qPCR), measuring gene expression levels in a highly sensitive and specific manner \cite{Bustin_2000}. However, there are limitations to this technique, which relate to the need for RNA extraction and the enzymatic based reverse transcription and target mRNA amplification steps which are prone to errors \cite{Yang_2006} \cite{Morten_2016}.  Branched chain DNA (bDNA) technology, in which the signal and not the mRNA target sequence is amplified, provides a non enzymatic alternative to qPCR \cite{Zhang_2005} \cite{Flagella_2006} \cite{Tsongalis_2006}. The QuantiGene Plex 2.0 platform (Affymetrix) combines bDNA with the luminex/xMAP magnetic bead capturing technology. This platform does not require an RNA extraction step, as it measures mRNA levels directly from cultured cells \cite{Severyn_2016}, cell lysates \cite{Ferrer_2014}, tissue homogenates \cite{Metzger_2013}, formalin-fixed tissues \cite{Knudsen_2008}, to name only a few starting points. The amplification of the signal depends on the cooperative hybridization between the target mRNA and three oligonucleotide probes. These probes are capture extenders (CE), label extenders (LE) and blocking probes (BL), whose sequences depend on the mRNA target sequence \cite{Ferrer_2014}.  The hybridized mRNA target sequence is immobilized on the bead via a capture probe that links the bead with parts of the CE sequence, which provides the specificity of the signal \cite{Tsongalis_2006}. The signal is subsequently amplified by adding a pre-amplifier sequence, which partly overlaps with the LE sequence, and by adding several biotinylated amplifiers, which generate the branched DNA structure \cite{Ferrer_2014}. The bDNA binds to streptavidin conjugated R-phycoerythrin (SAPE) and  a luminex reader detects the individual beads by flow cytometry \cite{Ferrer_2014} \cite{Hall_2015}. The resulting fluorescent signal is proportional to the hybridized mRNA quantity \cite{Tsongalis_2006} \cite{Hall_2015}.  The luminex/xMAP magnetic bead capturing technology allows for multiplexing in a single well of a 94 or 384 multi-well plate and is thus able to quantify the expression of a series of genes in a high throughput mode \cite{Ferrer_2014}. As such, the QuantiGene Plex 2.0 platform (Affymetrix) does not only offer the possibility to quantify mRNA levels in the context of patient diagnostics and/or therapeutics, but can also be used in a high throughput drug discovery setting. Indeed, entire compound libraries could be tested, in single dose or dose response, against disease specific gene signatures in search for new disease relevant chemical starting points \cite{Severyn_2016}. However, in order to analyse mRNA levels in high throughput mode, a proper data analysis frame work should be put in place. The data analysis flow proposed by Affymetrix is relatively straightforward and aims to translate and normalize gene expression, in the linear range of the assay, to fold change values. The latter is achieved by averaging all signals, subtracting the average background signal for each gene, normalizing against housekeeping genes and dividing the normalized values for the treated samples by the normalized value of the untreated sample. Hence, QuantiGene data is currently often processed in a local spreadsheet environment and normalized against two to four commonly used house keeping genes \cite{Yim_2010} \cite{Hall_2011} \cite{Sun_2011}\cite{Hall_2015} \cite{Morten_2016}.  The expression of generic housekeeping genes such as ACTB, GADPH, HPRT1 or B2M can, however, vary considerably across tissue types and/or under different experimental conditions, which make them less suited for normalisation  \cite{de_Jonge_2007}.  In addition to a proper housekeeping gene assessment, the optimal cell density, limit of quantification and number of beads will have to be analysed as well, prior to the start of a high throughput QuantiGene Plex 2.0 drug discovery campaign.  
Against this background we present experimental data and introduce a newly developed open source available R based shiny application: QGprofiler, that allows for proper QuantiGene Plex 2.0 assay optimisation, choice of housekeeping genes and data pre-processing from raw gene expression to normalized fold change values. In addition, we propose a way to assess cytotoxicity and introduce a step-wise dose response fold change analysis. QGprofiler is available at URL and will accept both 96 and 384 multi well plate format in single and dose response .