Get 20M+ Full-Text Papers For Less Than $1.50/day. Start a 7-Day Trial for You or Your Team.

Learn More →

Quality control parameters on a large dataset of regionally dissected human control brains for whole genome expression studies

Quality control parameters on a large dataset of regionally dissected human control brains for... JOURNAL OF NEUROCHEMISTRY | 2011 | 119 | 275–282 doi: 10.1111/j.1471-4159.2011.07432.x *Reta Lilla Weston Laboratories and Departments of Molecular Neuroscience, UCL Institute of Neurology, London, UK MRC Sudden Death Brain Bank Project, Department of Neuropathology, University of Edinburgh,Edinburgh, , UK Department of Medical & Molecular Genetics, King’s College London, Guy’s Hospital, London, UK Abstract assessed the accuracy of the array data using QuantiGene, We are building an open-access database of regional human as an independent non-PCR-based method. These quality brain expression designed to allow the genome-wide control parameters will allow database users to assess data assessment of genetic variability on expression. Array and accuracy. We report that within the parameters of this study RNA sequencing technologies make assessment of genome- post-mortem delay, agonal state and age have little impact on wide expression possible. Human brain tissue is a challenging array quality, array data are robust to variable RNA integrity, source for this work because it can only be obtained several and brain pH has only a small effect on array performance. and variable hours post-mortem and after varying agonal QuantiGene gave very similar expression profiles as array states. These variables alter RNA integrity in a complex data. This study is the first step in our initiative to make manner. In this report, we assess the effect of post-mortem human, regional brain expression freely available. delay, agonal state and age on gene expression, and the utility Keywords: brain pH, microarray validation, post-mortem of pH and RNA integrity number as predictors of gene human brain, QTL, RIN, RNA. expression as measured on 1266 Affymetrix Exon Arrays. We J. Neurochem. (2011) 119, 275–282. Microarray analysis and RNA sequencing of the transcip- RNA samples at present is the RNA Integrity Number (RIN) tome in post-mortem human brain tissue is a vital tool in as calculated by the Agilent 2100 Bioanalyzer for electro- investigating the complex genetic mechanisms involved in phoresis (Agilent Technologies UK Ltd, Edinburgh, UK). neurodegenerative and psychiatric disorders (Gilbert et al. 1981; Glanzer et al. 2004; Myers et al. 2007). However, there are many variables which influence the RNA integrity Received May 24, 2011; revised manuscript received July 17, 2011; accepted August 12, 2011. in post-mortem human brain tissues which need to be Address correspondence and reprint requests to John Hardy, Reta Lilla accounted for such data to be highly reliable (Sajdel- Weston Laboratories and Departments of Molecular Neuroscience, UCL Sulkowska et al. 1988; Burke et al. 1991; Glasel 1995; Institute of Neurology, Queen Square, London WC1N 3BG, UK. Imbeaud et al. 2005; Schroeder et al. 2006; Birdsill et al. E-mail: [email protected] 2010; Durrenberger et al. 2010). Abbreviations used: CRBL, cerebellum; eQTL, expression QTL; LRRK2, leucine-rich repeat kinase 2; MAPT, microtubule-associated It is important to have a reliable and stable method to protein tau; MRC, Medical Research Council; OCTX, occipital cortex; assess the quality of RNA samples generated from precious PMI, post-mortem interval; PUTM, putamen; QG, QuantiGene; RIN, heterogeneous tissues, especially from small anatomical RNA Integrity Number; SCN8A, sodium channel, voltage-gated, type regions, such as the substantia nigra and hypothalamus. VIII, alpha subunit; sQTL, splicing QTL; SHRI, Sun Health Research The most widespread measure for estimating the integrity of Institute; WHMT, white matter. 2011 The Authors Journal of Neurochemistry  2011 International Society for Neurochemistry, J. Neurochem. (2011) 119, 275–282 275 276 | D. Trabzuni et al. The RIN ranges from undetectable to ten, with undetectable Materials and methods being completely degraded and 10 being the most intact RNA. The calculation of RIN value is largely based on Human post-mortem brain tissue collection and dissection ribosomal RNA separation although this measure has been Brain tissues originating from 101 control Caucasian individuals shown to be inconsistent (Imbeaud et al. 2005; Schroeder were collected by the Medical Research Council (MRC) Sudden et al. 2006; Sherwood et al. 2011). Death Brain and Tissue Bank (Edinburgh, UK; Millar et al. We are building a publicly accessible database of regional 2007). The bodies were stored refrigerated and were brought up to human brain expression, the UK Human Brain Expression the PM suite just prior to the start of the autopsy. Each post- Consortium, to allow the assessment of the genetic variability mortem brain dissection was carried out in the same way. The whole brain was removed within 15 min of the body as fresh in gene expression (expression quantitative trait loci, eQTLs) tissue. The brain was immediately cut into coronal slices and the and splicing (splicing quantitative trait loci, sQTL) as well as various anatomical regions of interest were immediately sampled. detailed genome-wide expression analysis (Hardy et al. Furthermore, the samples once removed from the coronal slices 2009). To that end, we are collecting a large series of were placed in sealed containers which in turn were placed on control human brain tissues (originating from 130 individ- cool blocks (chilled to )20C) and stored within an insulated box. uals) in which we are dissecting 13 different CNS areas: The samples were dissected into various size pieces approximately prefrontal cortex Brodmann areas 9 and 46, parietal cortex 250–500 mg and were placed in tubes which were immediately Brodmann areas 3,1, and 2, occipital cortex (OCTX) snap frozen in liquid nitrogen. The post-mortem interval (PMI) Brodmann areas 17, temporal cortex Brodmann areas 21,41 was calculated from the time of death to the time of removal of and 42, central white matter (WHMT) below Brodmann the brain from the skull. The total number of samples processed from this source was 1842 as some CNS regions were not areas 39 and 40, hippocampus, thalamus, hypothalamus, available. In all cases, control status was confirmed by histology putamen (PUTM), cerebellum (CRBL), substania nigra, performed on sections prepared from paraffin embedded brain medulla and spinal cord. From each individual brain, we blocks and the diagnosis was determined by a consultant isolated DNA for whole genome genotyping analysis and neuropathologist. Detailed phenotypic information is described in from each region we isolated RNA for whole transcriptome Table 1. exon array analysis. This resulted in a total of 1266 RNA An additional 36 brains originating from neuropathologically samples analysed on Affymetrix Exon arrays and represents confirmed control Caucasian individuals were collected by the Sun by far the largest single CNS expression dataset at present. Health Research Institute (SHRI) an affiliate of Sun Health For this quality control study, we focused on analysing the Corporation, USA (Beach et al. 2008). In this case, whole brains factors that affected the reliability of the RNA samples. were removed as fresh tissue at autopsy and brain coronal slices In this study, we assess: (i) the effects of brain bank, age, were frozen. Anatomical regions of interest were sampled from gender, cause of death, region, post-mortem delay and brain brain coronal slices on dry ice. The time interval from the removal of the brain at the mortuary to the completion of the dissection and pH on RIN-based RNA quality, and, (ii) the effects of RNA placement of samples within the storage freezer ranged from 2.5 to quality on the performance quality of the array experiment, 4 h. The parietal cortex, hypothalamus and spinal cord regions were which was measured by a reliable and widely used param- not available for these samples. The total number of samples eter, present call (%P). %P is the percentage of probe sets processed was 476 as again, some regions were not available from with signal detection above background noise. We examine some brains. Detailed phenotypic information is described in the effects of RNA quality on the cDNA preparation and Table 1. cRNA production as part of the quality control of the array All samples from both sites had fully informed consent experiment, and finally we confirm the reproducibility of for retrieval and were authorised for ethically approved scien- array data using QuantiGene (QG), a novel, PCR-indepen- tific investigation (Research Ethics Committee number 10/H0716/ dent platform (Canales et al. 2006; Arikawa et al. 2008; Hall 3). et al. 2011). Table 1 Demographics of the samples studied. Values (range, mean) for variables in the cohort from both MRC-UK and SHRI-USA data set separately and joined. Of note, however, because of the different practices of the two tissue resources, there is no overlap in the PMIs between the MRC-UK (long PMI) and SHRI-USA (short PMI) Sex Age/year Brain pH PMI (h) RIN no. Brain bank Individuals Male Female Range Mean Range Mean Range Mean Range Mean MRC-UK 101 78 23 16–83 50.4 5.42–6.31 6.3 28–114 52.2 1–8.5 4 SHRI-USA 36 24 12 53–102 80 NA NA 1–5.5 2.6 1–8 3.6 UK + USA 137 102 35 16–102 59 5.42–6.31 6.3 1–114 43.7 1–8.5 3.85 2011 The Authors Journal of Neurochemistry  2011 International Society for Neurochemistry, J. Neurochem. (2011) 119, 275–282 QC parameters for gene expression studies in brain | 277 Post-mortem determination of brain pH AROS Applied Biotechnology AS company laboratories (http:// Brain pH from the MRC-UK samples was recorded from multiple www.arosab.com/). regions using a Hanna HI8424 hand-held pH meter with a glass Total RNA (200 ng) was used as starting material for the cDNA bodied electrode (Fisher Scientific, Loughborough, UK). Consistent preparation. First and second strand cDNA synthesis, the in vitro with previous studies, the pH did not vary between different brain transcription reaction to generate cRNA and the second round of regions (Stan et al. 2006; Monoranu et al. 2009). We therefore used cDNA synthesis was performed using the Ambion WT Expression a single pH value that measured from the lateral ventricle. The Kit according to the manufacturer’s instructions. Biotin labelling electrode was inserted into the lateral ventricle after the brain was was performed using the Terminal Labeling Kit (Affymetrix) coronally cut behind the mamillary bodies. The pH is influenced by according to the manufacturer’s instructions. Following the in vitro that of the tissue comprising the wall of the lateral ventricle and to a transcription reaction, the unincorporated nucleotides were removed lesser extent by the pH of the remaining CSF fluid within the lateral using RNeasy columns (Qiagen). ventricle. The pH was not measured in the SHRI-USA samples. Prior to hybridisation, the fragmented cDNA was heated to 95C for 5 min and subsequently to 45C for 5 min before loading onto RNA extraction the Affymetrix Human Exon 1.0 ST array cartridge. The array Total RNA was isolated from human post-mortem brain tissues cartridge was then incubated for 16 h at 45C at constant rotation based on the single-step method of RNA isolation (Chomczynski (60 rpm). The washing and staining procedure was performed in the and Sacchi 1987) using the miRNeasy 96 kit (Qiagen, Crawley, Affymetrix Fluidics Station 450. The array was exposed to 10 UK). Brain tissues (50–100 mg) were collected and weighed in washes in 6· SSPE-T at 250C followed by four washes in 0.5· RNase-free 96-well plates. A minimum of one extraction was SSPE-T at 50C. The biotinylated cRNA was stained with a performed from each tissue sample. All steps were performed on dry streptavidin-phycoerythrin conjugate, final concentration 2 mg/mL ice prior to the addition of the QIAzol Reagent. All samples were (Affymetrix) in 6· saline sodium phosphate EDTA buffer with homogenised in 4C using the TissueLyser II (Retsch, Castleford, 0.01% Tween-20 (SSPE-T) for 30 min at 25C followed by 10 UK) for 4–5 min at 30 Hz in 800 lL of QIAzol with the addition of washes in 6· SSPE-T at 25C. This was followed by an antibody two 3-mm stainless steel beads. In this step, the solution was amplification step using normal goat IgG as blocking reagent, final homogenised until no large pieces remained. concentration 0.1 mg/mL (Affymetrix) and biotinylated anti-strep- The homogenised tissue samples were incubated at room tavidin antibody (goat), final concentration 3 mg/mL (Affymetrix). temperature (15–25C) for 5 min. After incubation, 160 lLof This was followed by a staining step with a streptavidin– chloroform (CHCl ) (Sigma-Aldrich, Gillingham, UK) was added, phycoerythrin conjugate, final concentration 2 mg/mL (Affymetrix, the plates were mixed vigorously using the TissueLyser II for 30 s at UK) in 6· SSPE-T for 30 min at 25C and 10 washes in 6· SSPE-T 30 Hz. Plates were incubated at room temperature (15–25C) for 7– at 25C. The arrays were scanned at 560 nm using a confocal laser- 10 min. The plates were centrifuged at 6000 g for 45 min at 4C. scanning microscope (GeneChip Scanner 3000 7G). The aqueous phase (upper layer, with approximately 60% of the total volume after the QIAzol was added) was transferred to fresh Array quality control RNase-free 96-well plates. RNA was precipitated by adding 800 lL The Expression Console (EC) software version 1.1 (Affymetrix) of 100% ethanol to the aqueous phase (1.5 volume of the aqueous was used to evaluate the performance quality of the arrays including phase), followed by mixing the samples. The samples were then the labelling, hybridisation, scanning and background signals by transferred into the RNeasy 96-well plates to allow the RNA in the Probe Set summarisation and CHP file generation using Robust solution to bind to the membrane by centrifugation. Multichip Analysis. The quality assessment was performed by The plates were centrifuged at 5600 g for 2 min at 21C, and the generating different parameters for all the probesets analysed by EC; supernatant was discarded. Samples were washed to remove salt traces %P is the main parameter that is used for the array quality in this and impurities by adding 600 lL of washing buffer and centrifuging study. plates for 2 min. Washes were performed 3–4 times. Finally, total In addition, cDNA and cRNA Agilent 2100 Bioanalyzer profiles RNA was eluted in 65 lL of pre-heated RNase-free water (50C). were generated for samples with wide range of RIN numbers (RINs The concentration and purity of each RNA sample was assessed from 2 to7) to assess the cDNA preparation and cRNA production using the NanoDrop ND-1000 Spectrophotometer V3.3.0. The nucleotide lengths from RNA samples with different levels of concentration of each sample was calculated, together with the ratio degradation. of absorbance at 260 nm/280 nm and 260 nm/230 nm. After purification all RNA samples were applied to a RNA 6000 Exon array data analysis Nano-LabChip and analysed using the Agilent 2100 Bioanalyzer All arrays were pre-processed using Robust Multichip Analysis (Agilent Technologies UK Ltd) to obtain the RIN values. RINs and quantile normalisation with GC background correction in Partek’s total RNA electropherograms were calculated by the 2100 Expert Genomics Suite v6.6 (Partek Inc., St. Louis, USA) (Lockstone 2011). Software (Agilent Technologies UK Ltd). Further assessment of the After re-mapping the Affymetrix probe sets onto human Ref Seq RIN was performed by checking each electropherogram visually. build 19 (GRCh37) as documented in the most recent Netaffx annotation file (HuEx-1_0-st-v2 Probeset Annotations, Release 31), Expression profiling using Affymetrix GeneChip Human Exon 1.0 we restricted analysis to 308,717 probe sets that had a gene ST Arrays annotation and contained at least three probes with unique Expression profiling on the Affymetrix GeneChip Human Exon hybridisation. The gene-level expression was calculated for 27 000 1.0 ST Arrays (Affymetrix, High Wycombe, UK) was performed at genes by the median of probe sets corresponding to each gene. 2011 The Authors Journal of Neurochemistry  2011 International Society for Neurochemistry, J. Neurochem. (2011) 119, 275–282 278 | D. Trabzuni et al. Array validation using direct RNA quantification with branched brain regions of interest from each individual, a total of 1302 DNA, QuantiGene 2.0 Assay tissue blocks were available. In the majority of cases (870 of CRBL, OCTX, PUTM and WHMT samples from 12 individuals 1302 tissue blocks) multiple RNA extractions were per- were analysed using the QG platform for validation of exon array formed, resulting in 2318 RNA samples in total. Each of results. We focused on three target genes for validation, leucine-rich these RNA samples was assessed for RNA integrity as repeat kinase 2 (LRRK2), sodium channel, voltage-gated, type VIII, measured by RIN. A single RNA extraction from each tissue alpha subunit (SCN8A), and microtubule-associated protein tau sample was selected on the basis of Agilent 2100 Bioana- (MAPT). We selected ribosomal protein, large, P0 and ubiquitin C lyzer profile for downstream analysis using the Affymetrix as housekeeping genes to normalise the target genes as they showed Exon arrays (Fig. 1). relatively low variability in expression levels (i.e. low coefficient of variation) in all brain regions in our dataset. The approach to the The effect of age, gender, cause of death, region, PMI and selection of reference genes is explained in previous studies (de Jonge et al. 2007; Coulson et al. 2008).In addition, a recent study brain pH on RIN-based RNA quality confirms the efficiency of using this approach in selecting We assessed the dependence of RIN values on the other housekeeping genes to normalise in different tissues (Chervoneva covariates available to us. Our focus was on the explanatory et al. 2010). A summary of the QG probes used for analysis of all power of our available covariates on the RIN for each RNA five genes is provided in Table 2. extraction (2318 RNA samples). In this study, we assessed QuantiGene 2.0 Reagent System was used and the protocol in the 2318 RNA samples. The calculated RIN ranged from 1 to 8.5 QuantiGene 2.0 Reagent System User Manual was followed with with a mean of 3.85 (Table 1). Forty-three per cent of our the exception of the substrate step. Lumigen Lumi-Phos Plus and samples had RIN values of < 3. We found that 33% of the 10% lithium lauryl sulfate was used instead of Lumigen APS-5 variation in RIN was explained by differences among tissue substrate. A Biotek ELx 405 select plate washer was used for all of blocks (adjusted R measure), which set an upper limit for the the wash steps in the assay. The QG 2.0 plates were then read on a Molecular Devices LMax luminometer with the plate incubator set explanatory power of our covariates, which all act at a to 45C to maintain the temperature of the Lumigen Lumi-Phos between-tissue-sample level. Sixteen per cent of the variation Plus substrate. In total, 13 QG 2.0 plates were run to cover all target in RIN was explained by individual-level differences. genes and the house keeping genes. Each house keeping gene Outside of pH and PMI, which were analysed separately, ribosomal protein, large, P0 and ubiquitin C was loaded in the most important covariates were brain region (explained duplicates at 12.5 ng/well. In addition, target genes (LRRK2, )42 9.2% of the variation in RIN number, p = 1.7 · 10 ), age SCN8A and MAPT) were loaded in duplicates at 75 ng/well. )05 (explained an additional 1.1%, p = 4.1 · 10 ) and cause of death (explained an additional 1.9%, p = 0.022). Brain bank Statistical analysis and gender together explained an additional 0.4% of the Linear mixed regression analyses were performed to investigate the variation. relationship between age, gender, cause of death, region, PMI, pH, pH and PMI were investigated separately, because pH was RIN and present call (%P). Statistical analyses were conducted not measured in the SHRI-USA dataset and because the using Partek Genomics Suite version 6.6 and PASW statistic version 18 software. We assessed explanatory power in a forwards range of PMIs from the two brain banks did not overlap. stepwise manner, by examining the increase in variation explained pH explained 2.1% of the variation in the RIN number )4 when a new covariate or set of covariates were added to the existing from the MRC-UK brain bank (p = 1.0 · 10 ) (Fig. 2a).We model, together with a p-value for that increase. note that six individuals had very low pH < 5.90 and when excluded, the correlation was no longer significant (Fig. 2b). Results This study involved the analysis of brain tissue originating from 137 individuals. Since we were not able to obtain all 13 Table 2 QuantiGene probes used to perform array validation Gene Catalogue no. Designed to hybridise to LRRK2 83322 SA-26988 Human LRRK2 SCN8A 83324 SA-17320 Human SCN8A transcript variants 1 and 2 MAPT 81849 SA-15486 Human MAPT, all six variants UBC 80041 SA-10061 Human UBC RPLP0 81152 SA-11148 Human RPLP0 transcript variants 1 and 2 Fig. 1 Description of the data used in the analysis. 2011 The Authors Journal of Neurochemistry  2011 International Society for Neurochemistry, J. Neurochem. (2011) 119, 275–282 QC parameters for gene expression studies in brain | 279 (a) arrays was performed using Expression Console software. This software produces a number of array quality measures. The most reliable and widely used parameter is the present call (%P) (Tomita et al. 2004; Weis et al. 2007). The percent present call is the percentage of probe sets with signal detection above background probe level p -value of £ 0.01. The range of %P in this study was 1.7–77.3% (mean of 61.4%). Thrity-six per cent of the variation in %P was explained by individual-level differences (adjusted R measure). Outside of pH and PMI, which for reasons described previously were analysed separately, the most important covariates were brain region (explained 12.4% of the )53 variation in %P, p = 3.6 · 10 ), followed by brain bank )6 (b) (explained an additional 4.7%, p = 9.6 · 10 ), and then )05 RIN (explained an additional 2.7%, p = 9.1 · 10 ). Age, gender, and cause of death together explained an additional 2.2% of the variation. The effect of brain region on %P was most obvious when comparing CRBL and WHMT. CRBL showed highest %P (mean = 68%) whereas WHMT showed the lowest %P (mean = 57.5%) (Fig. 3). pH explained 12.0% of the variation in %P from the )9 MRC-UK brain bank (p = 2.3 · 10 ). (Fig. 4). However, as with RIN, this correlation was highly dependent on the six individuals with low pH < 5.90. When these samples were excluded from the analysis no significant correlation was obtained. Fig. 2 (a) Scatter plot for total RNA samples with linear regression line Finally, no significant correlation was found between PMI of RIN numbers for pH. Plot shows the effect of pH on RIN number for and %P, either for the MRC-UK dataset or the SHRI-USA RNA samples isolated from 13 region of control brain tissue. Test dataset. )4 p-values is test p-value = 1.0 · 10 , r-value = 0.145. Including the Moreover, no difference was observed in the expected low pH values of < 5.9. (b) Scatter plot for total RNA samples with nucleotide lengths following cDNA preparation (200–400 linear regression line of RIN numbers for pH. Plot shows the effect of nt) and cRNA production (ranges from 200 to 2000 nt) pH on RIN number for RNA samples isolated from 13 region of between different samples with different RIN values (as control brain tissue. No significant correlation was obtained with shown by Agilent Bioanalyser profiles). r-value = 0.0016 and p-value = 0.82. Samples with low pH value of < 5.9 were excluded. The effect of PMI on RIN differed between the MRC-UK and SHRI-USA datasets. There was no significant correlation between PMI and RIN in the MRC-UK dataset, which had PMIs ranging from 28 to 114 h. In contrast, PMI explained 8.4% of the variation in RIN in the SHRI-USA dataset (p = 0.0053), which had PMIs ranging from 1 to 5.5 h. Longer PMI was associated with a higher RIN in this data set, a counterintuitive result which might represent con- founding with some unmeasured variables in the study. The effects of age, gender, cause of death, region, PMI, brain pH and RIN-based RNA quality on array performance Fig. 3 Bar chart to show variation in %P by brain region (CRBL, (%P, cDNA and cRNA profile) WHMT).This graph shows the different performance of samples from We assessed the dependence of RNA array performance specific brain regions on the array. These results are highly significant )24 quality on RIN-based RNA quality and the other covariates p-values (1.3 · 10 ). The heights of the bars represent the mean. available to us. A systematic quality control check of the The error bars represent the SEM. 2011 The Authors Journal of Neurochemistry  2011 International Society for Neurochemistry, J. Neurochem. (2011) 119, 275–282 280 | D. Trabzuni et al. Fig. 6 QuantiGene validation of microarray expression data for Fig. 4 Scatter plot for total RNA samples with linear regression line of LRRK2 expression level in different regions. The graph shows higher Present call (%P) for pH. Scatter plot shows that pH significantly )9 expression in OCTX compare with other regions. Wilcoxon-signed explains 12.0% of the variation in %P (p = 2.3 · 10 , r = 0.353), rank test was performed and these results confirm the difference including samples with low pH values. Low pH values are driving the between regions in array data is significant. The expression level is regression analysis. presenting the median and the stars indicating the significant differ- ence in expression with p-values of < 0.01. In this case, as LRRK2 The reproducibility of array data using QuantiGene, a expression level was very low, it was unreliable to use SEM and PCR-independent platform to confirm expression results Wilcoxon’s test was performed using the median values. QuantiGene is a technique for mRNA expression quantifi- cation. Its strength is that it is a non-PCR based technique two platforms for both high (SCN8A and MAPT) (Fig. 5) that is therefore not subject to the systematic biases that PCR and low expression transcripts in brain (LRRK2) (Fig. 6). could create when applied to degraded RNA. This technique Furthermore, there was an excellent correlation in fold has the potential to be used in place of qRT-PCR (Canales change (r = 0.855) at both the gene and exon expression et al. 2006; Hall et al. 2011). levels and signal intensity values (r = 0.91) in the four brain We validated the exon array results in a subset of 12 regions we studied. individuals by comparing the normalised mRNA expression level for three genes (LRRK2, SCN8A and MAPT), in four brain regions (CRBL, OCTX, PUTM and WHMT). We Discussion observed a similar regional pattern of expression across the To our knowledge, the UK Human Brain Expression Consortium data set is the largest control brain microarray data set generated to date. It is based on the analysis of tissue samples from 13 different CNS regions originating from 137 individuals and containing 2318 processed samples. The main goal of this project is to build a large reference database for eQTL and sQTL analysis. This study showed considerable variation in RIN values among RNA samples. Sixty-seven per cent of the variation resides in differences among extractions from the same tissue blocks and most of the remaining variation is unexplained by the available covariate information. We found that pH is the most important post-mortem factor influencing RIN-based RNA integrity, a result consistent with previous studies (Hardy et al. 1985; Mexal et al. 2006; Fig. 5 QuantiGene validation of microarray expression data. (a) Chevyreva et al. 2008; Monoranu et al. 2009; Durrenberger SCN8A expression level between different regions. The graph shows et al. 2010). Samples with very low pH values (ranging high expression in OCTX compare with other regions. It is clear that from 5.42 to 5.90) were responsible for the positive this gene in mostly not expressed in WHMT region (expression level correlation seen between pH and %P (and also RIN). close to zero) also it presenting large error bar. (b) MAPT, showing However, when these low pH samples were removed from higher expression in OCTX compare with other brain regions. These the analysis we no longer observed any significant correla- results confirm the array data with significant p-values of < 0.01. The expression level is presenting the mean and the error bars is SEM. tion. This may in part explain contradictory observations 2011 The Authors Journal of Neurochemistry  2011 International Society for Neurochemistry, J. Neurochem. (2011) 119, 275–282 QC parameters for gene expression studies in brain | 281 regarding the effect of pH on RNA integrity and sample Furthermore, there is a possibility that samples with the performance on arrays (Hardy et al. 1985; Monoranu et al. shortest PMIs (1–2.5 h) within the SHRI-USA sample set 2009; Birdsill et al. 2010) and confirms the findings of a may originate from those individuals who suffered longer recent, but smaller study on control brain tissue (Tomita agonal states prior to death and agonal stress has been shown et al. 2004; Sherwood et al. 2011). to affect gene expression differently in different brain regions We found our array-based expression data to be reliably (Li et al. 2007). Other factors may contribute to this such as validated by the QuantiGene PCR-independent method, intermittent edge effects, especially on small samples during when tested on two high expression genes (MAPT and dissecting and tissue handling procedures. SCN8A) and one low expression gene (LRRK2). However, Finally, we note that our brain samples have been derived the performance quality of the array, as defined by %P, was from only two sources: one was a rapid death brain bank with not profoundly affected by age, gender, region, PMI, RIN long post-mortem intervals and the other specialises in and cause of death. This confirms findings from previous obtaining very short post-mortem intervals. Both brain banks studies on much smaller sample sizes (Tomita et al. 2004; had separately optimised their protocols to facilitate gene Birdsill et al. 2010; Durrenberger et al. 2010). We found that expression studies and this may limit the generalisability of only 2.7% of the variation in %P was explained by RIN. these conclusions to tissue collected in other ways. Indeed, 80 RNA samples with undetectable RINs performed These results are important for several reasons. Firstly, well on the arrays with %P values ranging from 45 to 76%. they confirm the practical feasibility of using post-mortem Thus, we found RIN to be a poor predictor of array quality control brain tissue to study the transcriptome of the human performance even at the low end of the RIN scale. brain by array technology. Secondly, they show that micro- Furthermore, the latter was confirmed since the cDNA and arrays can give reliable results over a wide range of RIN cRNA length synthesis was not affected by the wide range of numbers (1–8.5) and pH measurements with drop off in array RIN values (from 2 to 7) in our array experiments. The validity only being observed below brain pH 5.9. Thirdly, robust performance of the Affymetrix Exon arrays in the face they show that the results from Affymetrix exon arrays are of degraded RNA may be due to recent changes to the RNA reproducible by other technologies, making it possible for amplification process. In keeping with the manufacturer’s database users to use the data generated with confidence. instructions, this was performed using the Ambion WT Furthermore, this study is the first step of an ongoing Expression kit, which uses both non-polyA and polyA-based multi-regional human brain expression project that has been mRNA priming for first strand cDNA synthesis. This meant established to build an open-access database of identified that RNA amplification did not require an intact polyA tail. genome-wide genetic variability in relation with gene eQTLs In addition, increasing the quantity of the starting material of and sQTL as well as for detailed expression analysis (Hardy RNA from 500 to 750 ng improved the array performance. et al. 2009). We hope this will move the field forward in our Through the analysis of this observational study, we understanding of the underlying molecular mechanisms of experienced different limitations. For example, we had complex neurological and psychiatric diseases, and will expected that cause of death would greatly influence both support the neuroscience community with a resource which RIN-based RNA quality and %P-based array quality, but will bring functional insights. cause of death only explained 1.9% of variation in RIN and we did not find any significant relationship with %P. It may Acknowledgements be that cause of death is an imperfect reflection of the true This work was supported by the MRC through the MRC Sudden medical and drug treatment history of the individual, and that Death Brain Bank (CS) and through MRC Project Grants (John access to that history, were it available, would reveal other Hardy and Michael Weale) and Training Fellowship (Mina Ryten) factors of greater relevance. Likewise, in the range of 28– schemes. Daniah Trabzuni was supported by King Faisal Specialist 114 h PMI did not affect on either RIN or %P, nor could we Hospital and Research Centre, Saudi Arabia. We would like to thank see a loss of RIN-based RNA quality over the 1–5 h range. It AROS Applied Biotechnology AS company laboratories. We would remains possible that there may be selective loss of RNA also like to specially thank Jane Ramsey, Geoff Scopes and Wilson within an hour because the half-life of some mRNA species Lew (Affymetrix) for their valuable input. The authors have no has been reported to be as short as 15 min, while others may conflicting financial interests. be as long as 22 days depending on the tissue type and storage conditions (Ross 1995; Barrachina et al. 2006; Bahar References et al. 2007; Beach et al. 2008; Vennemann and Koppel- kamm 2010). This issue has been studied in detail for a range Arikawa E., Sun Y., Wang J., Zhou Q., Ning B., Dial S. L., Guo L. and Yang J. (2008) Cross-platform comparison of SYBR Green real- of PMIs by Harrison et al. 1995 and confirmed by a more time PCR with TaqMan PCR, microarrays and other gene recent study by Tomita et al. (2004). The same conclusion expression measurement technologies evaluated in the MicroArray was made that PMI had a limited effect on mRNA (Harrison Quality Control (MAQC) study. BMC Genomics 9, 328. et al. 1995; Tomita et al. 2004). 2011 The Authors Journal of Neurochemistry  2011 International Society for Neurochemistry, J. Neurochem. (2011) 119, 275–282 282 | D. Trabzuni et al. Bahar B., Monahan F. J., Moloney A. P., Schmidt O., MacHugh D. E. acidosis and postmortem interval for human brain gene expression and Sweeney T. (2007) Long-term stability of RNA in post-mortem studies: selective mRNA vulnerability and comparison with their bovine skeletal muscle, liver and subcutaneous adipose tissues. encoded proteins. Neurosci. Lett. 200, 151–154. BMC Mol. Biol. 8,108. Imbeaud S., Graudens E., Boulanger V., Barlet X., Zaborski P., Eveno Barrachina M., Castano E. and Ferrer I. (2006) TaqMan PCR assay in E., Mueller O., Schroeder A. and Auffray C. (2005) Towards the control of RNA normalization in human post-mortem brain standardization of RNA quality assessment using user-independent tissue. Neurochem. Int. 49, 276–284. classifiers of microcapillary electrophoresis traces. Nucleic Acids Beach T. G., Sue L. I., Walker D. G., Roher A. E., Lue L., Vedders L., Res. 33, e56. Connor D. J., Sabbagh M. N. and Rogers J. (2008) The Sun Health de Jonge H. J., Fehrmann R. S. and de Bont E. S. et al. (2007) Evidence Research Institute Brain Donation Program: description and based selection of housekeeping genes. PLoS ONE 2, e898. experience, 1987–2007. Cell Tissue Bank 9, 229–245. Li J. Z., Meng F. and Tsavaler L. et al. (2007) Sample matching by Birdsill A. C., Walker D. G., Lue L., Sue L. I. and Beach T. G. (2010) inferred agonal stress in gene expression analyses of the brain. Postmortem interval effect on RNA and gene expression in human BMC Genomics 8, 336. brain tissue. Cell Tissue Bank [Epub ahead of print]. Lockstone H. E. (2011) Exon array data analysis using Affymetrix Burke W. J., O’Malley K. L., Chung H. D., Harmon S. K., Miller J. P. and power tools and R statistical software. Brief Bioinform, [Epub Berg L. (1991) Effect of pre- and postmortem variables on specific ahead of print]. mRNA levels in human brain. Brain Res. Mol. Brain Res. 11, 37–41. Mexal S., Berger R., Adams C. E., Ross R. G., Freedman R. and Canales R. D., Luo Y. and Willey J. C. et al. (2006) Evaluation of DNA Leonard S. (2006) Brain pH has a significant impact on human microarray results with quantitative gene expression platforms. postmortem hippocampal gene expression profiles. Brain Res. Nat. Biotechnol. 24, 1115–1122. 1106, 1–11. Chervoneva I., Li Y., Schulz S., Croker S., Wilson C., Waldman S. A. Millar T., Walker R., Arango J. C., Ironside J. W., Harrison D. J., Mac- and Hyslop T. (2010) Selection of optimal reference genes for Intyre D. J., Blackwood D., Smith C. and Bell J. E. (2007) Tissue and normalization in quantitative RT-PCR. BMC Bioinform. 11, 253. organ donation for research in forensic pathology: the MRC Sudden Chevyreva I., Faull R. L., Green C. R. and Nicholson L. F. (2008) Death Brain and Tissue Bank. J. Pathol. 213, 369–375. Assessing RNA quality in postmortem human brain tissue. Exp. Monoranu C. M., Apfelbacher M. and Grunblatt E. et al. (2009) pH Mol. Pathol. 84, 71–77. measurement as quality control on human post mortem brain tis- Chomczynski P. and Sacchi N. (1987) Single-step method of RNA sue: a study of the BrainNet Europe consortium. Neuropathol. isolation by acid guanidinium thiocyanate-phenol-chloroform Appl. Neurobiol. 35, 329–337. extraction. Anal. Biochem. 162, 156–159. Myers A. J., Gibbs J. R. and Webster J. A. et al. (2007) A survey of genetic Coulson D. T., Brockbank S., Quinn J. G., Murphy S., Ravid R., Irvine human cortical gene expression. Nat. Genet. 39, 1494–1499. G. B. and Johnston J. A. (2008) Identification of valid reference Ross J. (1995) mRNA stability in mammalian cells. Microbiol. Rev. 59, genes for the normalization of RT qPCR gene expression data in 423–450. human brain tissue. BMC Mol. Biol. 9, 46. Sajdel-Sulkowska E. M., Majocha R. E., Salim M., Zain S. B. and Durrenberger P. F., Fernando S. and Kashefi S. N. et al. (2010) Effects Marotta C. A. (1988) The postmortem Alzheimer brain is a source of antemortem and postmortem variables on human brain mRNA of structurally and functionally intact astrocytic messenger RNA. quality: a BrainNet Europe study. J. Neuropathol. Exp. Neurol. 69, J. Neurosci. Methods 23, 173–179. 70–81. Schroeder A., Mueller O. and Stocker S. et al. (2006) The RIN: an RNA Gilbert J. M., Brown B. A., Strocchi P., Bird E. D. and Marotta C. A. integrity number for assigning integrity values to RNA measure- (1981) The preparation of biologically active messenger RNA from ments. BMC Mol. Biol. 7,3. human postmortem brain tissue. J. Neurochem. 36, 976–984. Sherwood K. R., Head M. W., Walker R., Smith C., Ironside J. W. and Glanzer J. G., Haydon P. G. and Eberwine J. H. (2004) Expression Fazakerley J. K. (2011) RNA integrity in post-mortem human profile analysis of neurodegenerative disease: advances in speci- vCJD and control brain tissue. Neuropathol. Appl. Neurobiol., 37, ficity and resolution. Neurochem. Res. 29, 1161–1168. 633–642. Glasel J. A. (1995) Validity of nucleic acid purities monitored by Stan A. D., Ghose S., Gao X. M., Roberts R. C., Lewis-Amezcua K., 260nm/280nm absorbance ratios. BioTechniques 18, 62–63. Hatanpaa K. J., Tamminga C. A. (2006) Human postmortem tissue: Hall J. S., Leong H. S. and Armenoult L. S. et al. (2011) Exon-array what quality markers matter? Brain Res. 1123, 1–11. profiling unlocks clinically and biologically relevant gene signa- Tomita H., Vawter M. P. and Walsh D. M. et al. (2004) Effect of agonal tures from formalin-fixed paraffin-embedded tumour samples. Br. and postmortem factors on gene expression profile: quality control J. Cancer 104, 971–981. in microarray analyses of postmortem human brain. Biol. Psychi- Hardy J. A., Wester P., Winblad B., Gezelius C., Bring G. and Eriksson atry 55, 346–352. A. (1985) The patients dying after long terminal phase have aci- Vennemann M. and Koppelkamm A. (2010) mRNA profiling in forensic dotic brains; implications for biochemical measurements on au- genetics I: possibilities and limitations. Forensic Sci. Int. 203, topsy tissue. J. Neural Transm. 61, 253–264. 71–75. Hardy J., Trabzuni D. and Ryten M. (2009) Whole genome expression as Weis S., Llenos I. C., Dulay J. R., Elashoff M., Martinez-Murillo F. and a quantitative trait. Biochem. Soc. Trans. 37, 1276–1277. Miller C. L. (2007) Quality control for microarray analysis of human Harrison P. J., Heath P. R., Eastwood S. L., Burnet P. W., McDonald B. brain samples: The impact of postmortem factors, RNA character- and Pearson R. C. (1995) The relative importance of premortem istics, and histopathology. J. Neurosci. Methods 165, 198–209. 2011 The Authors Journal of Neurochemistry  2011 International Society for Neurochemistry, J. Neurochem. (2011) 119, 275–282 http://www.deepdyve.com/assets/images/DeepDyve-Logo-lg.png Journal of Neurochemistry Pubmed Central

Quality control parameters on a large dataset of regionally dissected human control brains for whole genome expression studies

Download PDF
8 pages

Loading next page...
 
/lp/pubmed-central/quality-control-parameters-on-a-large-dataset-of-regionally-dissected-eoftK4Qk0J

References (41)

Publisher
Pubmed Central
Copyright
© 2011 The Authors. Journal of Neurochemistry © 2011 International Society for Neurochemistry
ISSN
0022-3042
eISSN
1471-4159
DOI
10.1111/j.1471-4159.2011.07432.x
Publisher site
See Article on Publisher Site

Abstract

JOURNAL OF NEUROCHEMISTRY | 2011 | 119 | 275–282 doi: 10.1111/j.1471-4159.2011.07432.x *Reta Lilla Weston Laboratories and Departments of Molecular Neuroscience, UCL Institute of Neurology, London, UK MRC Sudden Death Brain Bank Project, Department of Neuropathology, University of Edinburgh,Edinburgh, , UK Department of Medical & Molecular Genetics, King’s College London, Guy’s Hospital, London, UK Abstract assessed the accuracy of the array data using QuantiGene, We are building an open-access database of regional human as an independent non-PCR-based method. These quality brain expression designed to allow the genome-wide control parameters will allow database users to assess data assessment of genetic variability on expression. Array and accuracy. We report that within the parameters of this study RNA sequencing technologies make assessment of genome- post-mortem delay, agonal state and age have little impact on wide expression possible. Human brain tissue is a challenging array quality, array data are robust to variable RNA integrity, source for this work because it can only be obtained several and brain pH has only a small effect on array performance. and variable hours post-mortem and after varying agonal QuantiGene gave very similar expression profiles as array states. These variables alter RNA integrity in a complex data. This study is the first step in our initiative to make manner. In this report, we assess the effect of post-mortem human, regional brain expression freely available. delay, agonal state and age on gene expression, and the utility Keywords: brain pH, microarray validation, post-mortem of pH and RNA integrity number as predictors of gene human brain, QTL, RIN, RNA. expression as measured on 1266 Affymetrix Exon Arrays. We J. Neurochem. (2011) 119, 275–282. Microarray analysis and RNA sequencing of the transcip- RNA samples at present is the RNA Integrity Number (RIN) tome in post-mortem human brain tissue is a vital tool in as calculated by the Agilent 2100 Bioanalyzer for electro- investigating the complex genetic mechanisms involved in phoresis (Agilent Technologies UK Ltd, Edinburgh, UK). neurodegenerative and psychiatric disorders (Gilbert et al. 1981; Glanzer et al. 2004; Myers et al. 2007). However, there are many variables which influence the RNA integrity Received May 24, 2011; revised manuscript received July 17, 2011; accepted August 12, 2011. in post-mortem human brain tissues which need to be Address correspondence and reprint requests to John Hardy, Reta Lilla accounted for such data to be highly reliable (Sajdel- Weston Laboratories and Departments of Molecular Neuroscience, UCL Sulkowska et al. 1988; Burke et al. 1991; Glasel 1995; Institute of Neurology, Queen Square, London WC1N 3BG, UK. Imbeaud et al. 2005; Schroeder et al. 2006; Birdsill et al. E-mail: [email protected] 2010; Durrenberger et al. 2010). Abbreviations used: CRBL, cerebellum; eQTL, expression QTL; LRRK2, leucine-rich repeat kinase 2; MAPT, microtubule-associated It is important to have a reliable and stable method to protein tau; MRC, Medical Research Council; OCTX, occipital cortex; assess the quality of RNA samples generated from precious PMI, post-mortem interval; PUTM, putamen; QG, QuantiGene; RIN, heterogeneous tissues, especially from small anatomical RNA Integrity Number; SCN8A, sodium channel, voltage-gated, type regions, such as the substantia nigra and hypothalamus. VIII, alpha subunit; sQTL, splicing QTL; SHRI, Sun Health Research The most widespread measure for estimating the integrity of Institute; WHMT, white matter. 2011 The Authors Journal of Neurochemistry  2011 International Society for Neurochemistry, J. Neurochem. (2011) 119, 275–282 275 276 | D. Trabzuni et al. The RIN ranges from undetectable to ten, with undetectable Materials and methods being completely degraded and 10 being the most intact RNA. The calculation of RIN value is largely based on Human post-mortem brain tissue collection and dissection ribosomal RNA separation although this measure has been Brain tissues originating from 101 control Caucasian individuals shown to be inconsistent (Imbeaud et al. 2005; Schroeder were collected by the Medical Research Council (MRC) Sudden et al. 2006; Sherwood et al. 2011). Death Brain and Tissue Bank (Edinburgh, UK; Millar et al. We are building a publicly accessible database of regional 2007). The bodies were stored refrigerated and were brought up to human brain expression, the UK Human Brain Expression the PM suite just prior to the start of the autopsy. Each post- Consortium, to allow the assessment of the genetic variability mortem brain dissection was carried out in the same way. The whole brain was removed within 15 min of the body as fresh in gene expression (expression quantitative trait loci, eQTLs) tissue. The brain was immediately cut into coronal slices and the and splicing (splicing quantitative trait loci, sQTL) as well as various anatomical regions of interest were immediately sampled. detailed genome-wide expression analysis (Hardy et al. Furthermore, the samples once removed from the coronal slices 2009). To that end, we are collecting a large series of were placed in sealed containers which in turn were placed on control human brain tissues (originating from 130 individ- cool blocks (chilled to )20C) and stored within an insulated box. uals) in which we are dissecting 13 different CNS areas: The samples were dissected into various size pieces approximately prefrontal cortex Brodmann areas 9 and 46, parietal cortex 250–500 mg and were placed in tubes which were immediately Brodmann areas 3,1, and 2, occipital cortex (OCTX) snap frozen in liquid nitrogen. The post-mortem interval (PMI) Brodmann areas 17, temporal cortex Brodmann areas 21,41 was calculated from the time of death to the time of removal of and 42, central white matter (WHMT) below Brodmann the brain from the skull. The total number of samples processed from this source was 1842 as some CNS regions were not areas 39 and 40, hippocampus, thalamus, hypothalamus, available. In all cases, control status was confirmed by histology putamen (PUTM), cerebellum (CRBL), substania nigra, performed on sections prepared from paraffin embedded brain medulla and spinal cord. From each individual brain, we blocks and the diagnosis was determined by a consultant isolated DNA for whole genome genotyping analysis and neuropathologist. Detailed phenotypic information is described in from each region we isolated RNA for whole transcriptome Table 1. exon array analysis. This resulted in a total of 1266 RNA An additional 36 brains originating from neuropathologically samples analysed on Affymetrix Exon arrays and represents confirmed control Caucasian individuals were collected by the Sun by far the largest single CNS expression dataset at present. Health Research Institute (SHRI) an affiliate of Sun Health For this quality control study, we focused on analysing the Corporation, USA (Beach et al. 2008). In this case, whole brains factors that affected the reliability of the RNA samples. were removed as fresh tissue at autopsy and brain coronal slices In this study, we assess: (i) the effects of brain bank, age, were frozen. Anatomical regions of interest were sampled from gender, cause of death, region, post-mortem delay and brain brain coronal slices on dry ice. The time interval from the removal of the brain at the mortuary to the completion of the dissection and pH on RIN-based RNA quality, and, (ii) the effects of RNA placement of samples within the storage freezer ranged from 2.5 to quality on the performance quality of the array experiment, 4 h. The parietal cortex, hypothalamus and spinal cord regions were which was measured by a reliable and widely used param- not available for these samples. The total number of samples eter, present call (%P). %P is the percentage of probe sets processed was 476 as again, some regions were not available from with signal detection above background noise. We examine some brains. Detailed phenotypic information is described in the effects of RNA quality on the cDNA preparation and Table 1. cRNA production as part of the quality control of the array All samples from both sites had fully informed consent experiment, and finally we confirm the reproducibility of for retrieval and were authorised for ethically approved scien- array data using QuantiGene (QG), a novel, PCR-indepen- tific investigation (Research Ethics Committee number 10/H0716/ dent platform (Canales et al. 2006; Arikawa et al. 2008; Hall 3). et al. 2011). Table 1 Demographics of the samples studied. Values (range, mean) for variables in the cohort from both MRC-UK and SHRI-USA data set separately and joined. Of note, however, because of the different practices of the two tissue resources, there is no overlap in the PMIs between the MRC-UK (long PMI) and SHRI-USA (short PMI) Sex Age/year Brain pH PMI (h) RIN no. Brain bank Individuals Male Female Range Mean Range Mean Range Mean Range Mean MRC-UK 101 78 23 16–83 50.4 5.42–6.31 6.3 28–114 52.2 1–8.5 4 SHRI-USA 36 24 12 53–102 80 NA NA 1–5.5 2.6 1–8 3.6 UK + USA 137 102 35 16–102 59 5.42–6.31 6.3 1–114 43.7 1–8.5 3.85 2011 The Authors Journal of Neurochemistry  2011 International Society for Neurochemistry, J. Neurochem. (2011) 119, 275–282 QC parameters for gene expression studies in brain | 277 Post-mortem determination of brain pH AROS Applied Biotechnology AS company laboratories (http:// Brain pH from the MRC-UK samples was recorded from multiple www.arosab.com/). regions using a Hanna HI8424 hand-held pH meter with a glass Total RNA (200 ng) was used as starting material for the cDNA bodied electrode (Fisher Scientific, Loughborough, UK). Consistent preparation. First and second strand cDNA synthesis, the in vitro with previous studies, the pH did not vary between different brain transcription reaction to generate cRNA and the second round of regions (Stan et al. 2006; Monoranu et al. 2009). We therefore used cDNA synthesis was performed using the Ambion WT Expression a single pH value that measured from the lateral ventricle. The Kit according to the manufacturer’s instructions. Biotin labelling electrode was inserted into the lateral ventricle after the brain was was performed using the Terminal Labeling Kit (Affymetrix) coronally cut behind the mamillary bodies. The pH is influenced by according to the manufacturer’s instructions. Following the in vitro that of the tissue comprising the wall of the lateral ventricle and to a transcription reaction, the unincorporated nucleotides were removed lesser extent by the pH of the remaining CSF fluid within the lateral using RNeasy columns (Qiagen). ventricle. The pH was not measured in the SHRI-USA samples. Prior to hybridisation, the fragmented cDNA was heated to 95C for 5 min and subsequently to 45C for 5 min before loading onto RNA extraction the Affymetrix Human Exon 1.0 ST array cartridge. The array Total RNA was isolated from human post-mortem brain tissues cartridge was then incubated for 16 h at 45C at constant rotation based on the single-step method of RNA isolation (Chomczynski (60 rpm). The washing and staining procedure was performed in the and Sacchi 1987) using the miRNeasy 96 kit (Qiagen, Crawley, Affymetrix Fluidics Station 450. The array was exposed to 10 UK). Brain tissues (50–100 mg) were collected and weighed in washes in 6· SSPE-T at 250C followed by four washes in 0.5· RNase-free 96-well plates. A minimum of one extraction was SSPE-T at 50C. The biotinylated cRNA was stained with a performed from each tissue sample. All steps were performed on dry streptavidin-phycoerythrin conjugate, final concentration 2 mg/mL ice prior to the addition of the QIAzol Reagent. All samples were (Affymetrix) in 6· saline sodium phosphate EDTA buffer with homogenised in 4C using the TissueLyser II (Retsch, Castleford, 0.01% Tween-20 (SSPE-T) for 30 min at 25C followed by 10 UK) for 4–5 min at 30 Hz in 800 lL of QIAzol with the addition of washes in 6· SSPE-T at 25C. This was followed by an antibody two 3-mm stainless steel beads. In this step, the solution was amplification step using normal goat IgG as blocking reagent, final homogenised until no large pieces remained. concentration 0.1 mg/mL (Affymetrix) and biotinylated anti-strep- The homogenised tissue samples were incubated at room tavidin antibody (goat), final concentration 3 mg/mL (Affymetrix). temperature (15–25C) for 5 min. After incubation, 160 lLof This was followed by a staining step with a streptavidin– chloroform (CHCl ) (Sigma-Aldrich, Gillingham, UK) was added, phycoerythrin conjugate, final concentration 2 mg/mL (Affymetrix, the plates were mixed vigorously using the TissueLyser II for 30 s at UK) in 6· SSPE-T for 30 min at 25C and 10 washes in 6· SSPE-T 30 Hz. Plates were incubated at room temperature (15–25C) for 7– at 25C. The arrays were scanned at 560 nm using a confocal laser- 10 min. The plates were centrifuged at 6000 g for 45 min at 4C. scanning microscope (GeneChip Scanner 3000 7G). The aqueous phase (upper layer, with approximately 60% of the total volume after the QIAzol was added) was transferred to fresh Array quality control RNase-free 96-well plates. RNA was precipitated by adding 800 lL The Expression Console (EC) software version 1.1 (Affymetrix) of 100% ethanol to the aqueous phase (1.5 volume of the aqueous was used to evaluate the performance quality of the arrays including phase), followed by mixing the samples. The samples were then the labelling, hybridisation, scanning and background signals by transferred into the RNeasy 96-well plates to allow the RNA in the Probe Set summarisation and CHP file generation using Robust solution to bind to the membrane by centrifugation. Multichip Analysis. The quality assessment was performed by The plates were centrifuged at 5600 g for 2 min at 21C, and the generating different parameters for all the probesets analysed by EC; supernatant was discarded. Samples were washed to remove salt traces %P is the main parameter that is used for the array quality in this and impurities by adding 600 lL of washing buffer and centrifuging study. plates for 2 min. Washes were performed 3–4 times. Finally, total In addition, cDNA and cRNA Agilent 2100 Bioanalyzer profiles RNA was eluted in 65 lL of pre-heated RNase-free water (50C). were generated for samples with wide range of RIN numbers (RINs The concentration and purity of each RNA sample was assessed from 2 to7) to assess the cDNA preparation and cRNA production using the NanoDrop ND-1000 Spectrophotometer V3.3.0. The nucleotide lengths from RNA samples with different levels of concentration of each sample was calculated, together with the ratio degradation. of absorbance at 260 nm/280 nm and 260 nm/230 nm. After purification all RNA samples were applied to a RNA 6000 Exon array data analysis Nano-LabChip and analysed using the Agilent 2100 Bioanalyzer All arrays were pre-processed using Robust Multichip Analysis (Agilent Technologies UK Ltd) to obtain the RIN values. RINs and quantile normalisation with GC background correction in Partek’s total RNA electropherograms were calculated by the 2100 Expert Genomics Suite v6.6 (Partek Inc., St. Louis, USA) (Lockstone 2011). Software (Agilent Technologies UK Ltd). Further assessment of the After re-mapping the Affymetrix probe sets onto human Ref Seq RIN was performed by checking each electropherogram visually. build 19 (GRCh37) as documented in the most recent Netaffx annotation file (HuEx-1_0-st-v2 Probeset Annotations, Release 31), Expression profiling using Affymetrix GeneChip Human Exon 1.0 we restricted analysis to 308,717 probe sets that had a gene ST Arrays annotation and contained at least three probes with unique Expression profiling on the Affymetrix GeneChip Human Exon hybridisation. The gene-level expression was calculated for 27 000 1.0 ST Arrays (Affymetrix, High Wycombe, UK) was performed at genes by the median of probe sets corresponding to each gene. 2011 The Authors Journal of Neurochemistry  2011 International Society for Neurochemistry, J. Neurochem. (2011) 119, 275–282 278 | D. Trabzuni et al. Array validation using direct RNA quantification with branched brain regions of interest from each individual, a total of 1302 DNA, QuantiGene 2.0 Assay tissue blocks were available. In the majority of cases (870 of CRBL, OCTX, PUTM and WHMT samples from 12 individuals 1302 tissue blocks) multiple RNA extractions were per- were analysed using the QG platform for validation of exon array formed, resulting in 2318 RNA samples in total. Each of results. We focused on three target genes for validation, leucine-rich these RNA samples was assessed for RNA integrity as repeat kinase 2 (LRRK2), sodium channel, voltage-gated, type VIII, measured by RIN. A single RNA extraction from each tissue alpha subunit (SCN8A), and microtubule-associated protein tau sample was selected on the basis of Agilent 2100 Bioana- (MAPT). We selected ribosomal protein, large, P0 and ubiquitin C lyzer profile for downstream analysis using the Affymetrix as housekeeping genes to normalise the target genes as they showed Exon arrays (Fig. 1). relatively low variability in expression levels (i.e. low coefficient of variation) in all brain regions in our dataset. The approach to the The effect of age, gender, cause of death, region, PMI and selection of reference genes is explained in previous studies (de Jonge et al. 2007; Coulson et al. 2008).In addition, a recent study brain pH on RIN-based RNA quality confirms the efficiency of using this approach in selecting We assessed the dependence of RIN values on the other housekeeping genes to normalise in different tissues (Chervoneva covariates available to us. Our focus was on the explanatory et al. 2010). A summary of the QG probes used for analysis of all power of our available covariates on the RIN for each RNA five genes is provided in Table 2. extraction (2318 RNA samples). In this study, we assessed QuantiGene 2.0 Reagent System was used and the protocol in the 2318 RNA samples. The calculated RIN ranged from 1 to 8.5 QuantiGene 2.0 Reagent System User Manual was followed with with a mean of 3.85 (Table 1). Forty-three per cent of our the exception of the substrate step. Lumigen Lumi-Phos Plus and samples had RIN values of < 3. We found that 33% of the 10% lithium lauryl sulfate was used instead of Lumigen APS-5 variation in RIN was explained by differences among tissue substrate. A Biotek ELx 405 select plate washer was used for all of blocks (adjusted R measure), which set an upper limit for the the wash steps in the assay. The QG 2.0 plates were then read on a Molecular Devices LMax luminometer with the plate incubator set explanatory power of our covariates, which all act at a to 45C to maintain the temperature of the Lumigen Lumi-Phos between-tissue-sample level. Sixteen per cent of the variation Plus substrate. In total, 13 QG 2.0 plates were run to cover all target in RIN was explained by individual-level differences. genes and the house keeping genes. Each house keeping gene Outside of pH and PMI, which were analysed separately, ribosomal protein, large, P0 and ubiquitin C was loaded in the most important covariates were brain region (explained duplicates at 12.5 ng/well. In addition, target genes (LRRK2, )42 9.2% of the variation in RIN number, p = 1.7 · 10 ), age SCN8A and MAPT) were loaded in duplicates at 75 ng/well. )05 (explained an additional 1.1%, p = 4.1 · 10 ) and cause of death (explained an additional 1.9%, p = 0.022). Brain bank Statistical analysis and gender together explained an additional 0.4% of the Linear mixed regression analyses were performed to investigate the variation. relationship between age, gender, cause of death, region, PMI, pH, pH and PMI were investigated separately, because pH was RIN and present call (%P). Statistical analyses were conducted not measured in the SHRI-USA dataset and because the using Partek Genomics Suite version 6.6 and PASW statistic version 18 software. We assessed explanatory power in a forwards range of PMIs from the two brain banks did not overlap. stepwise manner, by examining the increase in variation explained pH explained 2.1% of the variation in the RIN number )4 when a new covariate or set of covariates were added to the existing from the MRC-UK brain bank (p = 1.0 · 10 ) (Fig. 2a).We model, together with a p-value for that increase. note that six individuals had very low pH < 5.90 and when excluded, the correlation was no longer significant (Fig. 2b). Results This study involved the analysis of brain tissue originating from 137 individuals. Since we were not able to obtain all 13 Table 2 QuantiGene probes used to perform array validation Gene Catalogue no. Designed to hybridise to LRRK2 83322 SA-26988 Human LRRK2 SCN8A 83324 SA-17320 Human SCN8A transcript variants 1 and 2 MAPT 81849 SA-15486 Human MAPT, all six variants UBC 80041 SA-10061 Human UBC RPLP0 81152 SA-11148 Human RPLP0 transcript variants 1 and 2 Fig. 1 Description of the data used in the analysis. 2011 The Authors Journal of Neurochemistry  2011 International Society for Neurochemistry, J. Neurochem. (2011) 119, 275–282 QC parameters for gene expression studies in brain | 279 (a) arrays was performed using Expression Console software. This software produces a number of array quality measures. The most reliable and widely used parameter is the present call (%P) (Tomita et al. 2004; Weis et al. 2007). The percent present call is the percentage of probe sets with signal detection above background probe level p -value of £ 0.01. The range of %P in this study was 1.7–77.3% (mean of 61.4%). Thrity-six per cent of the variation in %P was explained by individual-level differences (adjusted R measure). Outside of pH and PMI, which for reasons described previously were analysed separately, the most important covariates were brain region (explained 12.4% of the )53 variation in %P, p = 3.6 · 10 ), followed by brain bank )6 (b) (explained an additional 4.7%, p = 9.6 · 10 ), and then )05 RIN (explained an additional 2.7%, p = 9.1 · 10 ). Age, gender, and cause of death together explained an additional 2.2% of the variation. The effect of brain region on %P was most obvious when comparing CRBL and WHMT. CRBL showed highest %P (mean = 68%) whereas WHMT showed the lowest %P (mean = 57.5%) (Fig. 3). pH explained 12.0% of the variation in %P from the )9 MRC-UK brain bank (p = 2.3 · 10 ). (Fig. 4). However, as with RIN, this correlation was highly dependent on the six individuals with low pH < 5.90. When these samples were excluded from the analysis no significant correlation was obtained. Fig. 2 (a) Scatter plot for total RNA samples with linear regression line Finally, no significant correlation was found between PMI of RIN numbers for pH. Plot shows the effect of pH on RIN number for and %P, either for the MRC-UK dataset or the SHRI-USA RNA samples isolated from 13 region of control brain tissue. Test dataset. )4 p-values is test p-value = 1.0 · 10 , r-value = 0.145. Including the Moreover, no difference was observed in the expected low pH values of < 5.9. (b) Scatter plot for total RNA samples with nucleotide lengths following cDNA preparation (200–400 linear regression line of RIN numbers for pH. Plot shows the effect of nt) and cRNA production (ranges from 200 to 2000 nt) pH on RIN number for RNA samples isolated from 13 region of between different samples with different RIN values (as control brain tissue. No significant correlation was obtained with shown by Agilent Bioanalyser profiles). r-value = 0.0016 and p-value = 0.82. Samples with low pH value of < 5.9 were excluded. The effect of PMI on RIN differed between the MRC-UK and SHRI-USA datasets. There was no significant correlation between PMI and RIN in the MRC-UK dataset, which had PMIs ranging from 28 to 114 h. In contrast, PMI explained 8.4% of the variation in RIN in the SHRI-USA dataset (p = 0.0053), which had PMIs ranging from 1 to 5.5 h. Longer PMI was associated with a higher RIN in this data set, a counterintuitive result which might represent con- founding with some unmeasured variables in the study. The effects of age, gender, cause of death, region, PMI, brain pH and RIN-based RNA quality on array performance Fig. 3 Bar chart to show variation in %P by brain region (CRBL, (%P, cDNA and cRNA profile) WHMT).This graph shows the different performance of samples from We assessed the dependence of RNA array performance specific brain regions on the array. These results are highly significant )24 quality on RIN-based RNA quality and the other covariates p-values (1.3 · 10 ). The heights of the bars represent the mean. available to us. A systematic quality control check of the The error bars represent the SEM. 2011 The Authors Journal of Neurochemistry  2011 International Society for Neurochemistry, J. Neurochem. (2011) 119, 275–282 280 | D. Trabzuni et al. Fig. 6 QuantiGene validation of microarray expression data for Fig. 4 Scatter plot for total RNA samples with linear regression line of LRRK2 expression level in different regions. The graph shows higher Present call (%P) for pH. Scatter plot shows that pH significantly )9 expression in OCTX compare with other regions. Wilcoxon-signed explains 12.0% of the variation in %P (p = 2.3 · 10 , r = 0.353), rank test was performed and these results confirm the difference including samples with low pH values. Low pH values are driving the between regions in array data is significant. The expression level is regression analysis. presenting the median and the stars indicating the significant differ- ence in expression with p-values of < 0.01. In this case, as LRRK2 The reproducibility of array data using QuantiGene, a expression level was very low, it was unreliable to use SEM and PCR-independent platform to confirm expression results Wilcoxon’s test was performed using the median values. QuantiGene is a technique for mRNA expression quantifi- cation. Its strength is that it is a non-PCR based technique two platforms for both high (SCN8A and MAPT) (Fig. 5) that is therefore not subject to the systematic biases that PCR and low expression transcripts in brain (LRRK2) (Fig. 6). could create when applied to degraded RNA. This technique Furthermore, there was an excellent correlation in fold has the potential to be used in place of qRT-PCR (Canales change (r = 0.855) at both the gene and exon expression et al. 2006; Hall et al. 2011). levels and signal intensity values (r = 0.91) in the four brain We validated the exon array results in a subset of 12 regions we studied. individuals by comparing the normalised mRNA expression level for three genes (LRRK2, SCN8A and MAPT), in four brain regions (CRBL, OCTX, PUTM and WHMT). We Discussion observed a similar regional pattern of expression across the To our knowledge, the UK Human Brain Expression Consortium data set is the largest control brain microarray data set generated to date. It is based on the analysis of tissue samples from 13 different CNS regions originating from 137 individuals and containing 2318 processed samples. The main goal of this project is to build a large reference database for eQTL and sQTL analysis. This study showed considerable variation in RIN values among RNA samples. Sixty-seven per cent of the variation resides in differences among extractions from the same tissue blocks and most of the remaining variation is unexplained by the available covariate information. We found that pH is the most important post-mortem factor influencing RIN-based RNA integrity, a result consistent with previous studies (Hardy et al. 1985; Mexal et al. 2006; Fig. 5 QuantiGene validation of microarray expression data. (a) Chevyreva et al. 2008; Monoranu et al. 2009; Durrenberger SCN8A expression level between different regions. The graph shows et al. 2010). Samples with very low pH values (ranging high expression in OCTX compare with other regions. It is clear that from 5.42 to 5.90) were responsible for the positive this gene in mostly not expressed in WHMT region (expression level correlation seen between pH and %P (and also RIN). close to zero) also it presenting large error bar. (b) MAPT, showing However, when these low pH samples were removed from higher expression in OCTX compare with other brain regions. These the analysis we no longer observed any significant correla- results confirm the array data with significant p-values of < 0.01. The expression level is presenting the mean and the error bars is SEM. tion. This may in part explain contradictory observations 2011 The Authors Journal of Neurochemistry  2011 International Society for Neurochemistry, J. Neurochem. (2011) 119, 275–282 QC parameters for gene expression studies in brain | 281 regarding the effect of pH on RNA integrity and sample Furthermore, there is a possibility that samples with the performance on arrays (Hardy et al. 1985; Monoranu et al. shortest PMIs (1–2.5 h) within the SHRI-USA sample set 2009; Birdsill et al. 2010) and confirms the findings of a may originate from those individuals who suffered longer recent, but smaller study on control brain tissue (Tomita agonal states prior to death and agonal stress has been shown et al. 2004; Sherwood et al. 2011). to affect gene expression differently in different brain regions We found our array-based expression data to be reliably (Li et al. 2007). Other factors may contribute to this such as validated by the QuantiGene PCR-independent method, intermittent edge effects, especially on small samples during when tested on two high expression genes (MAPT and dissecting and tissue handling procedures. SCN8A) and one low expression gene (LRRK2). However, Finally, we note that our brain samples have been derived the performance quality of the array, as defined by %P, was from only two sources: one was a rapid death brain bank with not profoundly affected by age, gender, region, PMI, RIN long post-mortem intervals and the other specialises in and cause of death. This confirms findings from previous obtaining very short post-mortem intervals. Both brain banks studies on much smaller sample sizes (Tomita et al. 2004; had separately optimised their protocols to facilitate gene Birdsill et al. 2010; Durrenberger et al. 2010). We found that expression studies and this may limit the generalisability of only 2.7% of the variation in %P was explained by RIN. these conclusions to tissue collected in other ways. Indeed, 80 RNA samples with undetectable RINs performed These results are important for several reasons. Firstly, well on the arrays with %P values ranging from 45 to 76%. they confirm the practical feasibility of using post-mortem Thus, we found RIN to be a poor predictor of array quality control brain tissue to study the transcriptome of the human performance even at the low end of the RIN scale. brain by array technology. Secondly, they show that micro- Furthermore, the latter was confirmed since the cDNA and arrays can give reliable results over a wide range of RIN cRNA length synthesis was not affected by the wide range of numbers (1–8.5) and pH measurements with drop off in array RIN values (from 2 to 7) in our array experiments. The validity only being observed below brain pH 5.9. Thirdly, robust performance of the Affymetrix Exon arrays in the face they show that the results from Affymetrix exon arrays are of degraded RNA may be due to recent changes to the RNA reproducible by other technologies, making it possible for amplification process. In keeping with the manufacturer’s database users to use the data generated with confidence. instructions, this was performed using the Ambion WT Furthermore, this study is the first step of an ongoing Expression kit, which uses both non-polyA and polyA-based multi-regional human brain expression project that has been mRNA priming for first strand cDNA synthesis. This meant established to build an open-access database of identified that RNA amplification did not require an intact polyA tail. genome-wide genetic variability in relation with gene eQTLs In addition, increasing the quantity of the starting material of and sQTL as well as for detailed expression analysis (Hardy RNA from 500 to 750 ng improved the array performance. et al. 2009). We hope this will move the field forward in our Through the analysis of this observational study, we understanding of the underlying molecular mechanisms of experienced different limitations. For example, we had complex neurological and psychiatric diseases, and will expected that cause of death would greatly influence both support the neuroscience community with a resource which RIN-based RNA quality and %P-based array quality, but will bring functional insights. cause of death only explained 1.9% of variation in RIN and we did not find any significant relationship with %P. It may Acknowledgements be that cause of death is an imperfect reflection of the true This work was supported by the MRC through the MRC Sudden medical and drug treatment history of the individual, and that Death Brain Bank (CS) and through MRC Project Grants (John access to that history, were it available, would reveal other Hardy and Michael Weale) and Training Fellowship (Mina Ryten) factors of greater relevance. Likewise, in the range of 28– schemes. Daniah Trabzuni was supported by King Faisal Specialist 114 h PMI did not affect on either RIN or %P, nor could we Hospital and Research Centre, Saudi Arabia. We would like to thank see a loss of RIN-based RNA quality over the 1–5 h range. It AROS Applied Biotechnology AS company laboratories. We would remains possible that there may be selective loss of RNA also like to specially thank Jane Ramsey, Geoff Scopes and Wilson within an hour because the half-life of some mRNA species Lew (Affymetrix) for their valuable input. The authors have no has been reported to be as short as 15 min, while others may conflicting financial interests. be as long as 22 days depending on the tissue type and storage conditions (Ross 1995; Barrachina et al. 2006; Bahar References et al. 2007; Beach et al. 2008; Vennemann and Koppel- kamm 2010). This issue has been studied in detail for a range Arikawa E., Sun Y., Wang J., Zhou Q., Ning B., Dial S. L., Guo L. and Yang J. (2008) Cross-platform comparison of SYBR Green real- of PMIs by Harrison et al. 1995 and confirmed by a more time PCR with TaqMan PCR, microarrays and other gene recent study by Tomita et al. (2004). The same conclusion expression measurement technologies evaluated in the MicroArray was made that PMI had a limited effect on mRNA (Harrison Quality Control (MAQC) study. BMC Genomics 9, 328. et al. 1995; Tomita et al. 2004). 2011 The Authors Journal of Neurochemistry  2011 International Society for Neurochemistry, J. Neurochem. (2011) 119, 275–282 282 | D. Trabzuni et al. Bahar B., Monahan F. J., Moloney A. P., Schmidt O., MacHugh D. E. acidosis and postmortem interval for human brain gene expression and Sweeney T. (2007) Long-term stability of RNA in post-mortem studies: selective mRNA vulnerability and comparison with their bovine skeletal muscle, liver and subcutaneous adipose tissues. encoded proteins. Neurosci. Lett. 200, 151–154. BMC Mol. Biol. 8,108. Imbeaud S., Graudens E., Boulanger V., Barlet X., Zaborski P., Eveno Barrachina M., Castano E. and Ferrer I. (2006) TaqMan PCR assay in E., Mueller O., Schroeder A. and Auffray C. (2005) Towards the control of RNA normalization in human post-mortem brain standardization of RNA quality assessment using user-independent tissue. Neurochem. Int. 49, 276–284. classifiers of microcapillary electrophoresis traces. Nucleic Acids Beach T. G., Sue L. I., Walker D. G., Roher A. E., Lue L., Vedders L., Res. 33, e56. Connor D. J., Sabbagh M. N. and Rogers J. (2008) The Sun Health de Jonge H. J., Fehrmann R. S. and de Bont E. S. et al. (2007) Evidence Research Institute Brain Donation Program: description and based selection of housekeeping genes. PLoS ONE 2, e898. experience, 1987–2007. Cell Tissue Bank 9, 229–245. Li J. Z., Meng F. and Tsavaler L. et al. (2007) Sample matching by Birdsill A. C., Walker D. G., Lue L., Sue L. I. and Beach T. G. (2010) inferred agonal stress in gene expression analyses of the brain. Postmortem interval effect on RNA and gene expression in human BMC Genomics 8, 336. brain tissue. Cell Tissue Bank [Epub ahead of print]. Lockstone H. E. (2011) Exon array data analysis using Affymetrix Burke W. J., O’Malley K. L., Chung H. D., Harmon S. K., Miller J. P. and power tools and R statistical software. Brief Bioinform, [Epub Berg L. (1991) Effect of pre- and postmortem variables on specific ahead of print]. mRNA levels in human brain. Brain Res. Mol. Brain Res. 11, 37–41. Mexal S., Berger R., Adams C. E., Ross R. G., Freedman R. and Canales R. D., Luo Y. and Willey J. C. et al. (2006) Evaluation of DNA Leonard S. (2006) Brain pH has a significant impact on human microarray results with quantitative gene expression platforms. postmortem hippocampal gene expression profiles. Brain Res. Nat. Biotechnol. 24, 1115–1122. 1106, 1–11. Chervoneva I., Li Y., Schulz S., Croker S., Wilson C., Waldman S. A. Millar T., Walker R., Arango J. C., Ironside J. W., Harrison D. J., Mac- and Hyslop T. (2010) Selection of optimal reference genes for Intyre D. J., Blackwood D., Smith C. and Bell J. E. (2007) Tissue and normalization in quantitative RT-PCR. BMC Bioinform. 11, 253. organ donation for research in forensic pathology: the MRC Sudden Chevyreva I., Faull R. L., Green C. R. and Nicholson L. F. (2008) Death Brain and Tissue Bank. J. Pathol. 213, 369–375. Assessing RNA quality in postmortem human brain tissue. Exp. Monoranu C. M., Apfelbacher M. and Grunblatt E. et al. (2009) pH Mol. Pathol. 84, 71–77. measurement as quality control on human post mortem brain tis- Chomczynski P. and Sacchi N. (1987) Single-step method of RNA sue: a study of the BrainNet Europe consortium. Neuropathol. isolation by acid guanidinium thiocyanate-phenol-chloroform Appl. Neurobiol. 35, 329–337. extraction. Anal. Biochem. 162, 156–159. Myers A. J., Gibbs J. R. and Webster J. A. et al. (2007) A survey of genetic Coulson D. T., Brockbank S., Quinn J. G., Murphy S., Ravid R., Irvine human cortical gene expression. Nat. Genet. 39, 1494–1499. G. B. and Johnston J. A. (2008) Identification of valid reference Ross J. (1995) mRNA stability in mammalian cells. Microbiol. Rev. 59, genes for the normalization of RT qPCR gene expression data in 423–450. human brain tissue. BMC Mol. Biol. 9, 46. Sajdel-Sulkowska E. M., Majocha R. E., Salim M., Zain S. B. and Durrenberger P. F., Fernando S. and Kashefi S. N. et al. (2010) Effects Marotta C. A. (1988) The postmortem Alzheimer brain is a source of antemortem and postmortem variables on human brain mRNA of structurally and functionally intact astrocytic messenger RNA. quality: a BrainNet Europe study. J. Neuropathol. Exp. Neurol. 69, J. Neurosci. Methods 23, 173–179. 70–81. Schroeder A., Mueller O. and Stocker S. et al. (2006) The RIN: an RNA Gilbert J. M., Brown B. A., Strocchi P., Bird E. D. and Marotta C. A. integrity number for assigning integrity values to RNA measure- (1981) The preparation of biologically active messenger RNA from ments. BMC Mol. Biol. 7,3. human postmortem brain tissue. J. Neurochem. 36, 976–984. Sherwood K. R., Head M. W., Walker R., Smith C., Ironside J. W. and Glanzer J. G., Haydon P. G. and Eberwine J. H. (2004) Expression Fazakerley J. K. (2011) RNA integrity in post-mortem human profile analysis of neurodegenerative disease: advances in speci- vCJD and control brain tissue. Neuropathol. Appl. Neurobiol., 37, ficity and resolution. Neurochem. Res. 29, 1161–1168. 633–642. Glasel J. A. (1995) Validity of nucleic acid purities monitored by Stan A. D., Ghose S., Gao X. M., Roberts R. C., Lewis-Amezcua K., 260nm/280nm absorbance ratios. BioTechniques 18, 62–63. Hatanpaa K. J., Tamminga C. A. (2006) Human postmortem tissue: Hall J. S., Leong H. S. and Armenoult L. S. et al. (2011) Exon-array what quality markers matter? Brain Res. 1123, 1–11. profiling unlocks clinically and biologically relevant gene signa- Tomita H., Vawter M. P. and Walsh D. M. et al. (2004) Effect of agonal tures from formalin-fixed paraffin-embedded tumour samples. Br. and postmortem factors on gene expression profile: quality control J. Cancer 104, 971–981. in microarray analyses of postmortem human brain. Biol. Psychi- Hardy J. A., Wester P., Winblad B., Gezelius C., Bring G. and Eriksson atry 55, 346–352. A. (1985) The patients dying after long terminal phase have aci- Vennemann M. and Koppelkamm A. (2010) mRNA profiling in forensic dotic brains; implications for biochemical measurements on au- genetics I: possibilities and limitations. Forensic Sci. Int. 203, topsy tissue. J. Neural Transm. 61, 253–264. 71–75. Hardy J., Trabzuni D. and Ryten M. (2009) Whole genome expression as Weis S., Llenos I. C., Dulay J. R., Elashoff M., Martinez-Murillo F. and a quantitative trait. Biochem. Soc. Trans. 37, 1276–1277. Miller C. L. (2007) Quality control for microarray analysis of human Harrison P. J., Heath P. R., Eastwood S. L., Burnet P. W., McDonald B. brain samples: The impact of postmortem factors, RNA character- and Pearson R. C. (1995) The relative importance of premortem istics, and histopathology. J. Neurosci. Methods 165, 198–209. 2011 The Authors Journal of Neurochemistry  2011 International Society for Neurochemistry, J. Neurochem. (2011) 119, 275–282

Journal

Journal of Neurochemistry – Pubmed Central

Published: Oct 1, 2011

There are no references for this article.