75
acidbindingbeads(10μL)arecombinedwithaconcentratedbuffer(50μL)and60μL applied to the sample. After pipetting up and down three times to mix, the reaction was transferred to the well of a U-bottom plate, 60μL analytical grade isopropanol wasaddedandthemixturepipettedupanddownafurtherthree times. The samples were gently shaken (setting G) on a Grant Bio PMS-1000 microplate shaker (Grant Instruments Ltd., UK) for 5 minutes in order to facilitate the binding of as-cRNA to the nucleic acid binding beads. The U-bottom plate was removed and placed on the MagneticStand-96(LifeTechnologiesLtd.,UK)for10minutesuntilthe solution was completelytransparentandthebeadshadformedatight pellet against the magnets. Thesupernatantwasthencarefullyaspiratedand discarded ensuring not to disturb thepellet.Workingquicklytopreventthebeadsfromdryingout,100μL nucleic acid washsolutionwasadded, the sampleswereshakenata moderate speed (setting M) for 2½ minutes, the beads captured on the magnetic stand for 10 minutes and the supernatant discarded. This step was repeated and the samples shaken vigorously (settingV)for1minutetoevaporateanyresidualethanol.Toeach of the wells, 40μL of pre-heated (550C) elution solution was applied and the magnetic beads pipetted
upand down several times until fully dispersed. Following a 2 minute incubation at RT, the samples were shaken vigorously for 3 minutes, the beads then captured on themagneticstandfor60secondsandpurifiedas-cRNAtransferredtosterile0.2mL Eppendorftubesonice.As-cRNAconcentrationsweremeasuredontheNanoDropTM
1000 Spectrophotometer as previously described (Section 2.3.1.3).
To perform the sense-strand cDNA synthesis, 2μL of random primers were initially combinedwith10μgas-cRNAinatotal reaction volume of 24μL which is equivalent to an as-cRNA concentration of 455ng/μL made up in NF ddH20. The samples were
incubated,withtheheatedlidsetat 750C, to a denaturation temperature of 700C for
5minutes,followedby250Cforanother5minutesandthencooled to 40C for at least
2 minutes. At the end of the programme tubes were centrifuged for 5 seconds and placedonice.Asecond cycle sense-strand cDNA synthesis buffer (8μL) and enzyme (8μL) master mix was prepared and 16μL transferred to the reaction mixture. The tubeswerevortexed,spun-downinamicrocentrifugeandincubated,withtheheated lid set at 750C, to an annealing temperature of 250C for 10 minutes, then 420C for 1
76
being cooled to 40C for a minimum of 2 minutes. At the end of the programme, the
tubes were centrifuged for 5 seconds and placed on ice. The as-cRNA template was then degraded by the addition of 2μL RNase H. Pipetting up and down three times ensuredthecompletetransferoftheenzymefromthepipette tip to the sample. The tubes were vortexed, spun-down in a microcentrifuge and incubated at 370C for 45
minutes,withtheheatedlidsetat750C,followedby950Cfor5minutesbefore being
cooled to 40C for at least 2 minutes. At the end of the programme, tubes containing
the newly synthesised ss-cDNA were centrifuged for 5 seconds and placed on ice. Purification of the ss-cDNA was achieved applying the same basic principles of the magneticbead-basedassaydescribedabove,withafew minor adjustments made to theprotocol.Briefly,introducing18μLofNF ddH20 water brought the final reaction
volumeupto60μL.AtRT nucleic acid binding beads (10μL) were combined with a concentrated buffer (50μL) and 60μL applied to the sample. After pipetting up and down 3 times to mix the reaction was transferred to the well of a U-bottom plate, 120μL of absolute ethanol added and the mixture pipetted up and down a further 3 times.Samples were gently shaken (setting G) for 5 minutes, the beads captured on amagneticstandfor10minutesandthesupernatantdiscarded.Two applications of nucleicacidwashsolution(100μL)wereeachfollowedbyamoderateshake (setting M)for2½minutesanda capture time of 10 minutes before carefully aspirating and discardingthesupernatant.Residualethanolwas evaporated by vigorously shaking theplate (setting V) for 1 minute. Purified ss-cDNA was then eluted by adding 30μL elution solution preheated to 550C and pipetting up and down until the beads were
fullydispersed.Followinga2minuteincubationstepatRTthesampleswere shaken vigorouslyfor a further 3 minutes, the beads captured on the magnetic stand for 60 secondsandthess-cDNAtransferredonice to sterile 0.2mL Eppendorf tubes. Again concentrations were measured using the NanoDropTM 1000 Spectrophotometer as
previously described (Section 2.3.1.3). The products were stored at -200C.
2.3.2.3.2 Fragmentation
Thess-cDNAtargetswerefragmentedintolengthsofapproximatelyfortytoseventy nucleotides(~40-70nt)applyingtheAffymetrix®GeneChip®WTFragmentationand
77
bringthetotalreactionvolumeupto31.2μL.Afragmentationmaster mix containing a10XcDNAfragmentationbuffer, 1,000 units Apurinic/Apyrimidinic Endonuclease 1 (APE 1), 10 units Uracil-DNA Glycosylase (UDG) and RNase-freeTM water (10μL)
was prepared on ice and 16.8μL transferred to each cDNA sample. The tubes were flicked to mixandthecontentscollectedbycentrifugationimmediately prior to an incubation stepat370Cfor 60 minutesinapre-warmedMJResearchPTC-100®Pel-
tierThermalCyclerwiththeheatedlidsetat 1000C. The enzymes were heated to an
inactivation temperature of 950C for 2 minutes and the tubes snap-cooled on ice. A
1μL aliquot wasretained for size distribution analysis on the Agilent 2100 Bioanal- yser (Section 2.3.1.3).
2.3.2.3.3 Labelling
Toa45μLfragmentationreaction,15μLlabellingmastermixwasappliedcontaining terminal deoxynucleotidyl transferase (TdT) (2μL), 5X TdT buffer (12μL) and 5mM concentrationsofDNAlabellingreagent(1μL).Thetubeswereflickedtomixandthe samples centrifuged for 5 seconds immediately prior to an incubation step at 370C
for60minutesinapre-warmed MJ Research PTC-100® Peltier Thermal Cycler with
the heated lid maintained at 1000C. At the end of the programme the samples were
heatedtoadenaturationtemperatureof700C for 10 minutes, before being cooled
to 40C for a minimum of 2 minutes and placed on ice. The products were stored at -
200C.
2.3.2.3.4 Electrophoretic Mobility Gel-Shift Assay
The biotinylation efficiency of a proportion of the samples selected at random from eachfragmentation and labelling batch was assessed in an Electrophoretic Mobility ShiftAssay(EMSA)usingtheNeutrAvidinTM biotin-bindingprotein(Holden&Tacon
2011, Ludwig et al 1995) in accordance with the recommendations of Affymetrix®
Ltd., UK. Precast 15-well (15μL) 4 to 20% polyacrylamide gels were loaded into the Mini-PROTEAN® TetraCell(Bio-RadLaboratoriesInc.,UK)andthesystemfilledwith a 1X TBE buffer solution containing 89mM Tris-borate and 2mM EDTA, pH 8.3±0.1 at 250C. Two 1μL aliquots for each sample were transferred to separate Eppendorf
78
biotin-binding protein was prepared in a 1X PBS solution, pH 7.2±0.05 at 250C and
5μL added to the first aliquot with the volume of the second aliquot being made up with PBS. The tubes were vortexed, spun-down in a microcentrifuge and incubated atRTfor5minutes.MolecularweightmarkersHyperLadderTMIV(100-1000bp)and
V(25-500bp)wereloadedintothefirstand final lane of each gel. The samples were thencombinedwith5XDNAloadingbuffer(5μL),themixturepipetted up and down 3 times to mix and 10μL loaded onto the gel which was run at 150 volts for 1 hour. After electrophoresis the gel was stained in a 0.001% ethidium bromide solution in 1XTBEbuffer for 20 to 30 minutes and visualised under a UV light source using the Syngene GENi Gel Documentation System.
2.3.2.4 Affymetrix® Human Exon 1.0ST Arrays
2.3.2.4.1 GeneChip® HuEx-1_0-st-v2
TheAffymetrix®high-densityHuman Exon 1.0ST GeneChip® Arrays (HuEx-1_0-st-v
2)containmorethan5.5Mdistinctfeaturesspotteddirectlyontoa5μmsquareglass slidebymeansofphotolithography.Intotalthere are in excess of 1.4M probesets [4 PMprobesassignedtoeachindividualexon]fortheinterrogationofupwardsof one millionexonclusters,withinprobeselectionregionsofempiricallysupportedtrans- cribed sequences, including putative full length mRNA transcripts: from the RefSeq (ncbi.nlm.nih.gov/refseq)(Pruittetal2014),GenBank®(ncbi.nlm.nih.gov/genbank)
(Bensonetal 2014) (n=289,961), Ensembl (ensembl.org) (n=306,583) (Fernandez- Suarez & Schuster 2010, Flicek et al 2014) or dbEST (ncbi.nlm.nih.gov/genbank/d best)(n=665,175)(Boguskietal1993) databases; synteny mapped mouse/rat gen- omes(genome.ucsc.edu)(n=220,262)(Kuhnetal2013), in addition to, those which have been predicted computationally in Ensembl, TWINSCAN (mblab.wustl.edu/ns can) (Brent 2008, Gross & Brent 2006), Exoniphy (cgwb.nci.nih.gov) (Blanchette et al2004,Kentetal2003),MITOMAP(mitomap.org)(Brandonetal2005,Kogelniket al 1996, Ruiz-Pesini et al 2007), VEGA (Vertebrate Genome Annotation) (sanger.ac. uk/resources/databases/vega)(Wilmingetal2008),GENSCAN(genes.mit.edu/gen scan) (Burge & Karlin 1997, Burge & Karlin 1998), or using structural RNA predict- ionsoftware(genome.ucsc.edu) (Karolchik et al 2014), GeneID (genome.crg.es/soft ware.geneid)(Guigo 1998, Parra et al 2000) and the microRNA (miRNA) registry (s
79
anger.ac.uk/software/Rfam/mirna)(Griffith-Jones2004) for the discovery of novel expressionchangesand/oralternativesplicingeventsthathavenotpreviouslybeen reported (n=883,105).
Each25-meroligonucleotideprobeisdesignedtotargetthe sense-strand of the July 2003 NCBI build 34 of the human genome. On average, a typical gene composed of 10 exons will be represented by approximately 30 to 40 probes distributed across theentirelengthof the protein coding transcript and not just the far 3’ end which is associated with the more traditional IVT expression arrays. Another distinguishing featureoftheHumanExon 1.0ST GeneChip® Arrays is the strategy that Affymetrix®
hasadoptedfor subtracting background noise. Rather than possessing a paired MM probefor each PM probe, they have utilised a surrogate approach that incorporates measurementsof background signal intensities from a smaller number of around a thousand, pooled probes with identical GC contents; referred to as an ‘antigenomic’ background probeset.
2.3.2.4.2 Hybridization
Sense-strandcDNA targets, post-fragmentation and labelling, were hybridized onto Human Exon 1.0ST GeneChip® Arrays according to the manufacturer’s instructions
providedintheGeneChip®Hybridization,WashandStain Module (Affymetrix® Ltd.,
UK). GeneChips® were equilibrated to RT immediately prior to use. Reagents were
thawedoniceandthe20XEukaryotichybridizationcontrolstockheatedto650C for
5 minutes or until the cRNA had been completely re-suspended. The hybridization cocktailwaspreparedin1.5mLEppendorftubesby combining the remainder of the sample[58to60μLdependingonwhethera2μLaliquothad been removed in order toperformanEMSA(Section2.3.2.3.4)]with3nMcontrololigonucleotideB2(3.7μL), 20XEukaryotichybridizationcontrols(11μL)[bioB, bioC, bioD and cre at respective concentrationsof1.5(representsthelimitofdetection),5,25and 100pM (American Type Culture Collection)], 2X hybridization mix (110μL), 7% DMSO (15.4μL) and a sufficientquantityofRNase-freeTMwaterin order to bring the total reaction volume
up to 220μL. The tubes were flicked to mix and their contents collected by centrifu- gation. The hybridization cocktail was heated to 990C for 5 minutes, cooled to 450C
80
with 200μL of sample and incubated at 450C in the GeneChip® Hybridization Oven
640 (Affymetrix® Ltd., UK) for 16±0.5 hours overnight with rotation set at 60rpm.
2.3.2.4.3 Washing, Staining and Scanning
Post-hybridization,washingand staining was performed on the GeneChip® Fluidics
Station450(Affymetrix®Ltd.,UK) according to instructions contained in the Whole
Transcript Sense Target Labelling Assay Manual for fluidics protocolFS450_0001 that has been specifically developed for the Human Exon 1.0ST GeneChip® Arrays.
Briefly, the hybridization cocktail was removed and stored at 40C. Cartridges were
loadedwith200μL non-stringent Wash Buffer A [6X SSPE (3M NaCl, 0.2M NaH2PO4,
0.02M EDTA) and 0.01% Tween-20] and equilibrated to RT. Fresh solutions of the stainingreagentswereprepared.SAPEmixforthefirstand third stain contained 2X stain buffer (300μL) [100mM MES, 1M [Na+] and 0.05% Tween-20], 50mg/mL BSA
(24μL),1mg/mLSAPE(6μL)and270μLofmolecular biology grade water for a final volumeof600μLpersample.TheAbmixforthesecondstaincontained the same 2X stainbuffer(300μL),10mg/mLGoatIgGstock(6μL),0.5mg/mLbiotinylatedAb(3.6 μL), 50mg/mL BSA (24μL) and 266.4μL of molecular grade water, again to achieve a final volume of 600μL per sample.
Two post-hybridization washes, one with non-stringent Wash Buffer A at 300C for
10cycleswith2 mixes per cycle and another with stringent Wash Buffer B [100mM MES, 0.1M [Na+] and 0.01% Tween-20] at 500C for 6 cycles with 15 mixes per cycle,
was followed by the first stain in SAPE solution at 350C for 5 minutes. A post-stain
washwith non-stringent Wash Buffer A at 300C for 10 cycles with 4 mixes per cycle
wasfollowedbyasecondandthirdstainat350Cfor5minutesin the Ab solution and
5minutesintheSAPEsolution.Afinalwashstepwasperformed with non-stringent Wash Buffer A at 350C for 15 cycles with 4 mixes per cycle and the cartridges filled
with 800μL Array Holding Buffer prior to being scanned on the GeneChip® Scanner
3000 (Affymetrix® Ltd., UK).
Itisimportantto note that in the design of this study several steps were taken so as toavoidanypotentialconfoundingeffectsofmergingdatasetsfromdifferentbatches ofmicroarrays:1)Kitsandreagentswereorderedinbulkfromthesamesupplier so
81
consistentLOTnumberscouldbemaintained;2)Thetimethathadelapsedbetween processingeachofthebatcheswaskepttoaminimum;3)Thesamestoppingpoints, storageconditions&numberoffreezethaw cycles were strictly adhered to for each run; 4) Groupings were assigned randomly and re-assigned at each level from RNA extractionsrightthewaythroughtothe point at which the cartridges were scanned and 5) Each of the groups contained roughly equal proportions of female and male samples with a mixture of controls, FALS and SALS cases of variable age, onset and disease duration.
2.3.2.5 Bioinformatics Analysis
2.3.2.5.1 Affymetrix® GeneChip® Operating Software
ImagesprocessedbytheAffymetrix®GeneChip®Scanner3000wereextractedusing
GCOSversion1.2(Affymetrix®Ltd.,UK).Reportsweregenerated at gene (transcript
clusters)andexon(probesets)levelwhichplottedsignaldistributions and other QC
metrics used to evaluate the efficiency of individual hybridizations.
2.3.2.5.2 Partek® Genomics SuiteTM
CELfileswhichhadpassedtheQC metrics assessed by Expression ConsoleTM softw-
are version 1.3.0.187 (Affymetrix® Ltd., UK) (Section 2.3.2.5.2) were imported into
Partek® Genomics SuiteTM software version 6.60 copyright © 2013 (Partek® Inc. St
Louis, MO, USA) (Downey 2006, Fan et al 2011). Raw expression values were log2 transformed,apre-backgroundadjustmentforGCcontentwas applied and a quant- ilenormalisationprocedureperformedusingtheRobustMulti-arrayAverage(RMA) medianpolishprobesetsummarizationalgorithm(Irizarryetal2003,Quetal 2010) whichexcludesoutlyingvaluesfallingbelow1orgreaterthan3standarddeviations awayfromthemean(Wuetal 2004). Only core meta-probesets were considered (n =287,329).Attranscriptleveltheseinclude~22,000uniqueclusterswhichrelateto full-lengthprotein coding mRNA sequences in the RefSeq and GenBank® databases.
DEtranscriptsweredeterminedatthe5%significancethreshold(Partek®unadjusted
p<0.05)bymeansofa2-wayANalysisofCOVAriance(ANCOVA)statisticaltestwhich controlsfortheeffectsofageandgender.Listsweresubsequentlyfiltered to remove
82
unannotated transcripts (n=4,377/22,011) and a FC threshold of ≥ ±1.20 applied. Aberrant splicing events were predicted at the same level of statistical significance using the Partek® alternative splicing index (ALT-SPLICE ANOVA) in which exon to
exoncomparisonsweredrawnbetweendiseaseandcontrolgroupingsforeachgene in relation to the transcript’s overall DE [Partek® unadjusted p-value (fold-change)
≥0.05andp-value(alt.splicing)<0.05]. To reduce the probability of predicting false outcomes attributed to a type I error: a) Probesets with a maximum value for log2 transformed signal intensities below 3 across all the GeneChips® in the experiment
were removed except for instances where a significant difference in the expression ofasingleexonexistsbetweenthetwogroupings;b) Exons residing in genes whose expression at the level of the whole transcript was found to be significantly altered were excluded and c) Filtered transcript clusters with no observable differences in the expression levels of individual exons [Partek® unadjusted p<0.05, FC ≥ ±1.20]
(Gillettetal2009,Rabinetal2010,Tianetal 2011). Additional transcripts (n=2,077 /17,273)withfewerthanfourormorethanfortyprobesetswerealsoremovedsince the presence of a such small/large number of markers makes it prohibitively more difficult to interpret the alternative splicing events predicted to occur within these genes (Whistler et al 2010). Graphical outputs were obtained using easyExon 1.0.4 Java-basedplatformcopyright©2008whichisfreelyavailableonlineatmicroarray.
ym.edu.tw/easyExon (National Yang-Ming University, Taiwan) (Chang et al 2008).
2.3.2.5.3 Qlucore Omics Explorer
Partek®GC-RMAnormalisedpivotdatacouldalsobeinterpretedvisuallyinQlucore
Omics Explorer software version 2.3 (Lund, Sweden). F-test statistical outputs for the ANOVA were filtered at the 5% significance level (p<0.05) in order to generate lists of variables or genes (transcripts) which are best able to separate patient and controlgroupingsonaPCAplotorhierarchicallyclusteredHeatMap.Anypotentially confoundingeffectsdueto‘Age’(continuous)or‘Gender’(nominal)were eliminated asfactorsandthelistsfilteredtoremoveunannotatedtranscripts(n=3,356/22,011) using theNetAffxTM AnalysisCentre, which is freely available online at affymetrix.co m/analysis/index/affx(Affymetrix®Ltd.,UK) (Liu et al 2003), before applying a FC
83
2.3.2.5.4 Gene Ontology Enrichment Analysis
Affymetrix®transcript cluster ID’s (gene level) or probeset identification numbers
(exonlevel)wereuploadedintoDAVIDbioinformaticsresource version 6.7 (Section 2.3.1.8.3). With a Homo sapiens background applied, function annotation clustering analysis was performed on GOTERM_BP_FATandGOTERM_MF_FATgeneontology (GO)terms using ‘Medium’ classification stringency [Kappa:similaritytermoverlap 3,similaritythreshold0.50; Classification:Initial group membership 3, final group membership 3 and multiple linkage threshold 0.50]. Those categories identified as being significantly enriched withinthepatientpopulationsweredefinedashavinga minimumEASEscoreabove 1.30andBenjamini-HochbergFDR corrected p<0.05. An overview of the workflow and analysis pipeline for the Human Exon 1.0ST Gene Chip® ArraysusedforthetranscriptionalprofilingofperipheralEBV-transformedB- lymphocytes (LCL’s) in C9ORF72-related ALS [Chapter 5: ‘C9ORF72 GEP Study’ and Chapter 6: ‘C9ORF72 Survival Study’] is provided in Figure 2.4.
2.3.2.6 Quantitative Real-Time PCR
2.3.2.6.1 Primer Design
GenelevelexpressionchangesobservedontheHumanExon1.0STGeneChip®Arrays
werevalidatedbyqRT-PCR.Standardprimersordered from Eurofins MWG Operon, GermanyweredesignedusingtheNCBIPrimer-BLASTtool which is freely available online at ncbi.nlm.nih.gov/tools/primer-blast. The reference sequence (RefSeq) of the full protein-coding transcript was obtained in FASTA format from ensembl.org andprimers were designed to the region exhibiting the largest and most significant FC between the patient and control groupings. This was determined by plotting the meanprobesetsignalintensitiesfromPartek® GC-RMA normalised pivot data along
the y-axis against the Affymetrix® probeset identification numbers along the x-axis
usingMicrosoftExcel2010(MicrosoftCorporation,UK).Standarderrorof the mean (SEM) was calculated by dividing the standard deviation (stdev) by the square root ofnwhichrepresentsthenumberof samples in each group. Oligonucleotide probes