Pets & Animals

Assessment of gene expression in many samples using vertical arrays

Assessment of gene expression in many samples using vertical arrays
of 9
All materials on our website are shared by users. If you have any questions about copyright issues, please report us to resolve them. We are always happy to assist you.
Related Documents
  Published online 12 May 2008 Nucleic Acids Research, 2008, Vol. 36, No. 10  e60 doi:10.1093/nar/gkn263  Assessment of gene expression in many samples using vertical arrays Rosa Ana Risques, Gaelle Rondeau, Martin Judex, Michael McClelland and John Welsh* Sidney Kimmel Cancer Center, 10905 Road to the Cure, San Diego, CA 92121, USA  Received January 11, 2008; Revised April 1, 2008; Accepted April 21, 2008  ABSTRACTMicroarrays and high-throughput sequencing meth-ods can be used to measure the expression ofthousands of genes in a biological sample in a fewdays, whereas PCR-based methods can be used tomeasure the expression ofafewgenes in thousandsof samples in about the same amount of time. Thesemethods become more costly as the number ofbiological samples increases or as the number ofgenes of interest increases, respectively, and thesefactors constrain experimental design. To addresstheseissues,weintroduced‘verticalarrays’inwhichRNA from each biological sample is converted intomultiple, overlapping cDNA subsets and spotted onglass slides. These vertical arrays can be queriedwith single gene probes to assess the expressionbehavior in thousands of biological samples in asingle hybridization reaction. The spotted subsetsare less complex than the srcinal RNA from whichthey derive, which improves signal-to-noise ratios.Here,wedemonstratethequantitativecapabilitiesofverticalarrays,includingthesensitivityandaccuracy of the method and the number of subsets needed toachieve this accuracy for most expressed genes.INTRODUCTION Regulated gene expression plays important roles in almostevery aspect of biology, including the differentiation andmigration of cells, maintenance of homeostasis, responsesto stress, damage or infection and aging. Evolutionarychanges in gene expression account for many of thedifferences between species and between individuals withina species. Inappropriate expression can lead to disability,disease, and death, but can also serve as a sensitiveindicator of disease. The immense implications of geneexpression in basic biology and medicine have motivatedthe invention of a wide variety of analytical methods totrack regulated changes, including microarrays (1–3),high-throughput sequencing (HTS) (4–6) and quantitativePCR (7,8). These methods apply to the two extremes of the experimental design spectrum: microarrays and HTScan be used to measure the expression of thousands of genes simultaneously in individual biological samples,whereas quantitative PCR can be used to measure theexpression of individual genes in thousands of biologicalsamples. These methods are very fast and economicalrelative to their predecessors. However, microarrays andHTS methods become less convenient and more costly asthe number of biological samples increases, and quanti-tative PCR becomes more costly as the number of genes of interest increases. Which of these methods to use is astrategic decision based on experimental design factors,such as the number of samples, genes of interest, andreplicates required, and on practical factors, such as theaccessibility of the technology and cost. Experimentaldesigns involving a few thousand biological samples (e.g.cells treated with thousands of different drugs) in whichthe behavior of a few hundred genes is of interest areimpractical using these methods for most laboratories.There are several methods that address this problematicneighborhood of experimental design, including a methodby Kuhn  et al  . (9) that involves the capture of targets to apreassembled array of probe-bearing beads, a methodby Yang  et al  . (10) involving target capture on encodedbeadsanda methodby Geiss  et al  . (11)involving theuseof color-coded probe pairs. Traditional dot blots (12), inwhich total cDNA is spotted on a membrane support andhybridized with single gene probes, have been used tomeasure the expression of single genes in multiplebiological samples, but dot blots have poor performancecharacteristics and the membrane format is inconvenient.To remedy this, Rogler and colleagues (13) devised RNAexpression microarrays (REM) that are essentially dotblots implemented in a glass slide microarray format, Present addresses:Rosa Ana Risques, Department of Pathology, University of Washington, Seattle, WA 98195, USAMartin Judex, Landrat-Wagner-Str. 43a, 84085 Langquaid, GermanyThe authors wish it to be known that, in their opinion, the first two authors should be regarded as joint First Authors*To whom correspondence should be addressed. Tel: +1 619 450 5990, ext. 282; Fax: +1 619 550 3998; Email:   2008 The Author(s)This is an Open Access article distributed under the terms of the Creative Commons Attribution Non-Commercial License ( which permits unrestricted non-commercial use, distribution, and reproduction in any medium, provided the srcinal work is properly cited.   b  y g u e  s  t   on S  e  p t   e m b  e r 2  ,2  0 1 4 h  t   t   p :  /   /  n a r  . oxf   or  d  j   o ur n a l   s  . or  g /  D o wnl   o a  d  e  d f  r  om   greatly improving both performance and convenience.A potential drawback of this direct approach is that raretranscripts remain rare in cDNA made using methods thatseek to preserve representation. We previously introducedtheideaofprintinglowcomplexityrepresentations(LCRs)of mRNA population glass slide microarrays to monitorthe expression of individual genes in many biologicalsamples (14). The strategy is outlined in Figure 1. TheseLCRs comprise overlapping subsets of the RNA popula-tion. The representation of rare transcripts is enhanced inthese subsets, leading to the surmise that their usemight permit expression profiling of rare transcripts.However, the quantitative behavior of the method hasnot been established, either with regard to sensitivity torare transcripts, or with regard to the number of LCRsneededtoachievehighsensitivityformostexpressedgenes,and the method cannot be used effectively without thiscritical information. Here, we describe these quantitativeaspects of vertical arrays in comparison to microarrays,real-time RT–PCR and ‘spike-in’ experiments. MATERIALS AND METHODS RNA samplepreparation Normal human diploid fibroblasts (cell line ATCC CRL2091) were deprived of serum for 48h and serum wasreintroduced as described by Iyer  et al  . (15). Total RNAwas isolated at 0, 20 and 240min after reintroduction of serum using RNeasy Mini Kits (Qiagen, Valencia, CA,USA). RNA was treated with DNase I, and purified againusing the RNeasy Mini Kit cleanup protocol. RNAconcentration was determined by UV absorbance at260nm and adjusted to 25ng/ m l. LCR preparation Reverse transcription and RNA arbitrarily primed poly-merase chain reaction (RAP-PCR) were performed in asingle reaction mixture containing 1   M-MLV buffer(Promega, Madison, WI, USA), 0.2mM each dNTP (ICN,Aurora, OH, USA), 1 m Ci [ a -P 32 ] dCTP (ICN, Irvine, CA,USA), 5 m M arbitrary primer (Proligo, Boulder, CO,USA), 50U M-MLV (Promega), 50U AmpliTaq DNApolymerase Stoffel fragment (Applied Biosystems, FosterCity, CA, USA) and 100ng of RNA. The reaction wasincubated at 37 8 C for 60min, heated at 94 8 C for 3min andtemperature cycled through 94 8 C for 15s, 35 8 C for 2minand 72 8 C for 2min for 35 cycles. Eleven different 10-merarbitrary primers were used for RAP-PCR: c8(TCACCAGCCA), d8(ACGGGCCAGT), g8(CAAGGGCAGT),h8(GGCAGGCTGT), c9(GGGCACCAGG), d9(GGGGCACCAC), f9(CACCAGGGGC), g9(CTGACTGCCT), a10(ACCTGGGGAG), c10(ACAGCCCCCA) andOPN28(GCACCAGGGG). The reactions were assembledusing a Biomek FX liquid handling workstation. Productswere purified using the PSI    Clone PCR 96 purificationkit (Princeton Separations, Adelphia, NJ, USA), elutedwith 80 m l of distilled water (pH 9.5), and DNA con-centration was measured for one replicate of each RAP-PCR reaction type. RAP-PCR repeatability was assessedqualitatively by electrophoresis through 4% polyacrylam-ide, 8M urea gels and autoradiography. Standard microarray analysis of RAP-PCR products To identify differentially regulated genes with which tocharacterize vertical arrays, standard microarray analysisusing LCRs was done as shown previously (16). LCRswere labeled for hybridization to standard arrays asfollows:   500ng of LCR was mixed with 8 m g of random Color-coded gene probes...LCR1LCR2LCR3LCR41 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 LCR1a LCR2a LCR3a ...LCRs from multiple biological sources (a,b,c…)LCR1b LCR2b LCR3b . . . LCR1c LCR2c LCR3c . . . . . . LCRs from a single biological source LCRs spotted on glass slidesAB Figure 1.  Production of a vertical array. ( A ) Multiple LCRs are produced using different arbitrary primers, indicated by different colors. The 16horizontal bars represent a small number of the many different RNAs in a sample. Arbitrary primers match opposing sequences in the RNApopulation by chance, generating partially overlapping arbitrary sample sequences. ( B ) Several LCRs are prepared from each biological sample andthese are spotted on a glass slide to form a vertical array. Fluorescently tagged single gene probes hybridize to LCRs that have sampled sequencefrom the mRNA corresponding to the gene. e60  Nucleic Acids Research, 2008, Vol. 36, No. 10  P AGE 2 OF 9   b  y g u e  s  t   on S  e  p t   e m b  e r 2  ,2  0 1 4 h  t   t   p :  /   /  n a r  . oxf   or  d  j   o ur n a l   s  . or  g /  D o wnl   o a  d  e  d f  r  om   hexamer (final vol. 36 m l), boiled for 5min and cooled onice. Five mocroliters of a 10  reaction mixture was added,and the volume was adjusted to 50 m l, such that the finalreaction contained 10mM Tris–HCl (pH 7.5), 5mMMgCl 2 , 7.5mM DTT, 0.025mM dGTP, dATP anddCTP, 0.009mM dTTP, 0.04mM Cy3- or Cy5-dUTP(Amersham Pharmacia Biotech, Buckinghamshire,England) and 10U of DNA Polymerase I Klenowfragment (New England Biolabs, Beverly, MA, USA).The reaction was incubated at 37 8 C overnight and thenheated at 70 8 C for 10min. The samples were purified usingQIAquickPCRpurificationkit(Qiagen)andelutedin25 m lof distilled water. Each labeling reaction was done induplicate, the purified products from duplicate reactionswere pooled (50 m l) and incorporation was measuredby spectrophotometery (Cy3: 550nm, Cy5: 650nm).Approximately 80pmols of dye per microgram of DNAwere incorporated. The  t ¼ 0 sample was mixed with t ¼ 240min sample and hybridized to arrays assembled onUltraGAPS coated slides (Corning, Corning, NY, USA)containing three replicates of PCR products from 3840human cDNA clones (I.M.A.G.E) (Invitrogen, Carlsbad,CA, USA). Each slide contained three replicates, for sixdata points per gene. Prehybridization, hybridization andwashes were performed following the manufacturer’sinstructions for UltraGAPS slides, except that the iso-propanol wash step was omitted and probes were notallowedtocooltoroomtemperature.Formamidewasusedat 25% final concentration and 0.1mg/ml denaturedsalmon sperm DNA was used as blocking agent.Standard microarrays were scanned using a ScanArray5000 Laser scanner using ScanArray version 2.1 softwareand were quantified using Quantarray version 2.0 software(PerkinElmer, Waltham, Massachusetts, USA). Reciprocaldye-swapping was performed for every experiment. Vertical microarrayprinting RAP-PCR reaction products were dried and resuspendedin 22 m l of distilled water to achieve average DNAconcentrations of 100ng/ul for printing. A total of 4 m lof DNA was mixed with 4 m l of DMSO. Each of the eightreplicate RAP-PCR reactions for every time-point wereprinted 12 times on the slide. The printing was done withan Omnigrid microarrayer (GeneMachines, San Carlos,CA, USA) on Ultra GAPS coated slides (Corning). Afterprinting, the DNA was cross-linked to the slides by UVirradiation (300mJ) using a UV StrataLinker (Stratagene,La Jolla, CA, USA) and baked for 2h at 80 8 C. Slides werethen washed in water and spin-dried for storage. Salmonella  LT2 DNA digested with EcoRV and ClaI at50ng/ m l was printed as negative controls. This digested Salmonella  LT2 DNA was also used to produce thepositive controls, wherein 12 of the sequences selected tobe probed in the vertical arrays were PCR amplified andspiked in as serial dilutions at 3ng/ul, 0.6ng/ul, 0.12ng/ul,24pg/ul and 4.8pg/ul, with 50ng/ul  Salmonella  LT2. Vertical microarrayprobe synthesis andhybridization Twenty-four genes that were differentially regulated wereinitially chosen for study on the vertical arrays, and sixgenes exhibiting no change in expression were selected asnegative controls. Three of these were eventually excludedfrom the analysis due to the presence of repetitiveelements, and one is an independent cDNA clone fromthe same Unigene. The corresponding I.M.A.G.E. cloneswere grown, and the inserts were amplified by PCR usingthe primers M13F (GTTTTCCCAGTCACG- ACGTTG)and M13R (TGAGCGGATAACAATTTCACACAG).The PCR products were purified using the QIAquickPCR purification kit (Qiagen), and concentrations weremeasured spectrophotometrically. Insert sizes were con-firmed by electrophoresis. 25–50ng of insert DNA waslabeled by  in vitro  transcription (IVT) using a Megascriptkit (Ambion, Austin, TX, USA), in a reaction containing7.5mM GTP, ATP and CTP, 2.5mM UTP, 1.75mMCy5-UTP; (Amersham Pharmacia Biotech) and 1 m l of enzyme mix, with a final volume of 10 m l. The reaction wasincubated at 37 8 C for 3h. A total of 7.5U of T7 RNAPolymerase (Promega) were then added, and after 3h at37 8 C, 1U of RNase-free DNase 1 was added and tubeswere incubated at 37 8 C for 15min. RNA was purifiedusing an RNeasy Mini Kit (Qiagen). IVT produced8–10 m g of RNA labeled with 400–800pmols of dye.The vertical arrays were prehybridized, hybridized andwashed following the manufacturer’s protocol for UltraGAPS slides with a few modifications. 0.1mg/ml of PolydT (Amersham Pharmacia Biotech), 0.1mg/ml of humanCot-1 DNA (Invitrogen) and 25% formamide were used.A probe consisted of 1–1.2 m g of Cy5-labeled RNA(  50–80pmols of dye) corresponding to a gene wasmixed with 10–12ng of a Cy3-labeled (0.7–1pmols of dye) pool of all LCRs, blocking agents, formamide andbuffer. Cy3-dUTP labeling of this pool followed theprotocol described above for labeling LCRs. Real-time RT-PCR Transcript abundances for 20 genes studied using verticalarrays were also quantified by real-time RT–PCR(Table 1S, a and b). The primers were designed withPrimer Express software version 2.0.0 (Applied Bio-systems), and chosen to span splice junctions to avoidamplification of possible contaminating genomic DNA orunspliced transcript. Primers used can be found inTable 1Sb. First-strand cDNA synthesis was performedusing oligo (dT) 15  and real-time RT–PCR was carried outin the presence of SYBR Green using the ABI Prism7900HT sequence detector (Applied Biosystems). A melt-ingcurvewasusedtoidentifyatemperaturewhereonlytheamplicon, and not primer dimers, accounted for SYBRGreen-bound fluorescence. Standard curves for candidatecDNAs were prepared from a four-point 1/10 serialdilution and were run in duplicate, as were all the samplesand the nontemplate control. cDNA quantities werenormalized to an internal glyceraldehyde-3-phosphatedehydrogenase mRNA control. Spike-inexperiments Sequences from 10  Arabidopsis thaliana  cDNA clones wereamplified by PCR using specific primer pairs, with one of each pair having a 5 0 T7 promoter sequence extension P AGE 3 OF 9  Nucleic Acids Research, 2008, Vol. 36, No. 10  e60   b  y g u e  s  t   on S  e  p t   e m b  e r 2  ,2  0 1 4 h  t   t   p :  /   /  n a r  . oxf   or  d  j   o ur n a l   s  . or  g /  D o wnl   o a  d  e  d f  r  om   (AR34, AF325016.2, 545bp; AR37, AF325019.2, 817bp;AR43, AF325025.2, 874; AR44, AF325026.2, 647bp;AR45, AF325027.2, 600bp; AR46, AF325028.2, 767bp;AR50, AF325032.2, 1056bp; AR51, AF325033.2, 1224bp;AR53, AF325035.2, 1170bp; AR60, AF325042.2, 634bp).T7RNA polymerase wasthen usedto synthesize thespike-in transcripts. The transcripts were purified and added tohuman fibroblast total RNA at the proportions discussedin the text. LCRs were prepared as described above forarbitrary primers c8, c9, c10, d9, g8 and h8, and these wereeach spotted five times on glass slide arrays, as describedpreviously. Fluorescent probes for each of these spike-intranscripts were prepared from the same PCR productsusingCy3orCy5andhybridizedtothearrayedLCRs.Themean intensity values at each of the five dilutions werecalculated for each LCR, and averaged between dye-swapchips, log-transformed and plotted. Figure 5a shows anexample, and Figure 3S, shows results for each spike-inand the corresponding probe. To select the best LCR foreach gene, the correlation between log 4  measured inten-sitiesandlog 4 spike-inconcentrations,excludingthelowestconcentration (i.e. zero spike-in) was determined, and r p > 0.95 was used as the first selection criterion. Thesecond criterion for the best LCR was to choose the bestStudent’s pairwise  t -test  P -value indicating a measurabledifference between the lowest nonzero spike-in concentra-tion and the highest. These criteria resulted in theselections in Figure 5b. The zero spike-in concentrationswereexcludedbecausethesesampleswereoftenindifferentphysical locations on the microarrays, resulting in greatervariance due to background issues. Student’s  t -tests werethen calculated for every pairwise difference in spike-inconcentrations (Table 2S). RESULTS In these experiments, LCRs were prepared using RAP-PCR (17), in which arbitrarily chosen oligonucleotideprimers are used in low stringency reverse transcriptionand PCR (Figure 1a). LCRs made in this way are similarto multiplex PCR products, except that single shortoligonucleotide primers are used, and these primers findfrequent matches, or approximate matches, in the RNAdue to their short length. Regions of the RNA that areflanked by sequences with partial matches to the arbitraryprimers succeed in reverse transcription and PCRamplification. The sequence complexity of LCRs is lowerthan that of the RNA from which they are derived becauseonly a subset of the sequences in the template moleculesamplifies. Successful sequences amplify reproducibly butwith different efficiencies, such that rare mRNAs can beabundantly represented in an LCR, while abundantmRNAs can be represented at low levels. Thus, anyindividual sequence in an LCR can have higher repre-sentation than in the mRNA population from which theLCR was derived. While the relative abundances of dif-ferent sequences within a sample can be highly distorted,relative abundances any particular sequence  between samples are maintained, as in multiplex PCR, andconsequently, LCRs can be used to infer relative transcriptabundances between samples.When printed on microarrays (Figure 1b), the higherrepresentation of rare transcript sequences in LCRs leadsto better signal-to-noise behavior in hybridization experi-ments (16–18), and multiple LCRs can be prepared suchthat most RNAs have enhanced representation in at leastone LCR. Arrays prepared in this manner can be queriedwith gene-specific probes to explore differential expressionin potentially thousands of biological samples with veryhigh sensitivity. In the discussion that follows, singlesequences will be referred to as ‘probes’, and complexmixtures will be referred to as ‘targets’. In standardmicroarrays, the probes are affixed to the array surfaceand the target is used in solution. In vertical arrays, it isthe other way around: the complex targets are spotted andthe simple probe is in solution. The LCRs prepared usingRAP-PCR were used in these two different capacities, firstas solution-phase targets for standard microarrays andthen as spotted targets on vertical arrays. Detection ofLCR-specific differentially regulated genes To initiate these experiments, we used LCRs as hybridiza-tion targets for standard cDNA expression arrays. Thisprocedure revealed genes that were differentially regulatedin response to serum-starvation and refeeding of fibro-blasts,andalsoidentifiedtheLCRsinwhichthesegenesarerepresented.TotalRNAfromaserumstarvation-refeedingtreatment in fibroblasts performed according to Iyer  et al  .(15) was purified at 0, 20 and 240min after the reintroduc-tion of serum. This RNA was converted to LCRs usingRAP-PCR and 11 different arbitrary primers. There aretwo different RAP-PCR procedures, one of which gen-erates LCRs from an initial oligo(dT)-primed first strandcDNA template (19), and the other of which generatesLCRs directly from RNA using arbitrary priming of reverse transcription to make first strand cDNA(17,20,21).Thelatterwasusedintheseexperimentsbecauseit can be done in a single well, with fewer pipetting steps,and without cDNA purification, facilitating preparationusing a pipetting robot. The LCRs were radioactivelylabeled and reproducibility was assessed qualitatively bygel electrophoresis and autoradiography. The 0 and240min LCRs were fluorescently labeled and used astargets against standard cDNA microarrays containingabout 4000 human cDNA probes (16). Reciprocal dyeswap experiments were also performed. This procedureidentified differentially regulated genes and the LCR inwhich each differentially regulated gene was represented.Analysis involved print-tip loess normalization and scalingbetween arrays using the limma package in BioConductorand the R programming environment (22–24). A modified t -statistic was used to estimate the probability that a genewasdifferentially regulated (22),with  P -values adjustedformultiple testing to predict the false discovery rate (25). Tentranscripts having a modified  t -statistic with  P  0.05 fromat least one LCR target, four with  P -values in the range0.05  P  0.33, and four having larger  P -values ( P  0.5)weretestedusingreal-timeRT–PCRtoconfirmdifferentialgene expression. RT–PCR spanned splice junctions to e60  Nucleic Acids Research, 2008, Vol. 36, No. 10  P AGE 4 OF 9   b  y g u e  s  t   on S  e  p t   e m b  e r 2  ,2  0 1 4 h  t   t   p :  /   /  n a r  . oxf   or  d  j   o ur n a l   s  . or  g /  D o wnl   o a  d  e  d f  r  om   avoid possible interference from unspliced transcripts orresidual genomic DNA. Table 1S, contains real-time PCRresults, gene names, accession numbers and the associatedLCR. All of these genes met the additional criterion thattheir average signal intensities exceeded the mean intensityof96 control probesequences derived fromthe ratby threestandard deviations. The real-time RT–PCR measure-ments correlated well ( r p ¼ 0.92) with the correspondingmicroarray measurements for those transcripts that hadmodified  t -statistics with  P  0.20 in the standard micro-arrays (Figure 1S). Detection of differential gene expression usingvertical arrays Vertical arrays were then assembled by spotting LCRsprepared from the same RNA preparations and arbitraryprimers on glass slides using a microarray printer. Elevendifferent LCRs for each of the three time-points ( t ¼ 0, 20,240min) after reintroduction of serum were preparedin eight replicates. Eight replicate oligo(dT)-primedcDNAs from the same RNAs were prepared in parallel.Each LCR and oligo(dT)-primed cDNA was spotted fourtimes in each of three subarrays, for a total 3168 spotsfrom the LCRs and 288 spots from oligo(dT)-primedcDNA. In addition, a complete replicate of the LCRsprepared using arbitrary primer OPN28 was included, foran additional 288 spots. The final array contained 3456LCR spots and 288 oligo(dT)-primed cDNA spots. Spotswere also included comprising serial dilutions of theanticipated probe sequences diluted in restriction digested Salmonella  genomic DNA as positive controls forhybridization. Additional spots containing  Salmonella sequences were included as controls for cross-hybridiza-tion and other ill-defined foreground nuisance problems.Fluorescently labeled probes corresponding to 28 genesselected from standard microarrays were made by reversetranscription with incorporation of Cy5-labeled nucleo-tides. A control comprising equal masses of all 11 LCRsfrom each time point was labeled by random primedsynthesis with Cy3-labeled nucleotides and these werehybridized to the vertical arrays simultaneously with thegene-specific probes to allow normalization for spottedDNA mass and probe availability to hybridization. Thiscontrol mixture is sufficiently complex that individual geneexpression differences do not contribute significantly tovariance in the hybridization signal.LCRs from all three time points were spotted adjacentlyin small groups throughout the chip. Ratios of measure-ments from the time points  t ¼ 20 and  t ¼ 240 weregenerated by dividing by a measurement from adjacent t ¼ 0 spots on the chip after normalization to the mixedLCR control signal, and these ratios were plotted for eachgene and each LCR. The  t ¼ 0 adjacent spots were used tohelp correct for local variation in background. Figure 2shows one such graph for gene AA428473, which maps toNuclear receptor subfamily 1, group D, member 2(NR1D2). Four-fold down-regulation is implied at t ¼ 240 by two different LCRs, d9 and h8. QuantitativeRT–PCR (real-time RT–PCR) indicated 5.3-fold down-regulation of this gene. The other nine LCRs do not reporta change, nor is the change reflected in vertical array dataacquired from the oligo(dT)-primed first strand cDNAtargets. The signals from the oligo(dT)-primed targetswere typically 20-fold or more larger than the LCRsignals, and differential regulation detected in LCRs werenot detected in the oligo(dT) targets. However, we did notattempt to optimize for detection in the oligo(dT) targets.Consistent with this observation, in a standard microarrayexperiment using an oligo(dT)-primed probe and fourreplicate arrays, only two among these 28 genes(AA251800 and H77766) had changes with  P -values of  P  0.05. For all LCRs where standard arrays implied achange in transcript abundance with  P  0.20, thecorresponding LCR on the vertical array implied a similarchange.More than one LCR might report a change for anygene, and several examples of this are shown in Figure 2S.To decide which LCR was the best reporter of differentialexpression for that gene, we employed a prescreen of datafalling outside two standard deviations on the log 2  scale toexclude outliers, which can usually be attributed to defectsin the microarray such as high-local background, andfollowed this with  t -tests. Boxplots of vertical array resultsfor 27 of the 28 transcripts are shown in Figure 3. Onegene failed quality control and was omitted. Plotted foreach gene are the data from the LCR having the largest t -statistic and  P  10  5 . The log 2 -transformed data isapproximately normally distributed for each gene.Figure 4a shows strong Pearson’s correlation ( r p ¼ 0.94)between measurements made using standard arrays andvertical arrays, and Figure 4b shows the correspondingcomparison between vertical arrays and those that weretested using real-time RT–PCR ( r p ¼ 0.92). The standarderror of the estimate calculated from the data in Figure 4bwas  s est ¼ 0.62 on the log 2  scale, and assumes that the real-time RT–PCR measurements were error-free. Thesestudies indicate that vertical arrays measure changes intranscription abundances quite accurately. Correlationbetween real-time PCR and the oligo(dT)-primed targetswas  r p ¼ 0.16, indicating that the direct oligo(dT)-primingapproach did not provide useful information for mosttranscripts. Vertical arrays did not show changes after20minofre-exposure toserum, withthepossible exceptionof H79778 (Histone deacetylase 3) (Figure 2S).Thegenesusedinthisstudywerechosenwithoutspecificreference to their biological functions. Their expressionprofiles were largely in accord with results of Chang  et al  .(26) deposited in GEO (, with the exceptions of CTSB (AA598950), whichis down-regulated after 4h in our data but only slightlyupregulated in (26) NR1D2 (AA428473), which is verystrongly down-regulated in our data, but not so in (26)LRRFIP1 (AA085597), which is strongly up-regulated inour data, but very modestly up-regulated in (26) andHDAC3 (H79778), which is not regulated in our data,but is moderately upregulated in (26). Detectionsensitivity We performed ‘spike-in’ experiments to determine thenumber of LCRs needed to detect changes in most P AGE 5 OF 9  Nucleic Acids Research, 2008, Vol. 36, No. 10  e60   b  y g u e  s  t   on S  e  p t   e m b  e r 2  ,2  0 1 4 h  t   t   p :  /   /  n a r  . oxf   or  d  j   o ur n a l   s  . or  g /  D o wnl   o a  d  e  d f  r  om 
Similar documents
View more...
Related Search
We Need Your Support
Thank you for visiting our website and your interest in our free products and services. We are nonprofit website to share and download documents. To the running of this website, we need your help to support us.

Thanks to everyone for your continued support.

No, Thanks