Human blood genomics: distinct profiles for gender, age and neurofibromatosis type 1

Please download to get full document.

View again

of 13
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.
Document Description
Human blood genomics: distinct profiles for gender, age and neurofibromatosis type 1
Document Share
Document Tags
Document Transcript
  Research report  Human blood genomics: distinct profiles for gender, age andneurofibromatosis type 1 $ Yang Tang a,b , Aigang Lu a,b , Ruiqiong Ran a,b , Bruce J. Aronow c , Elizabeth K. Schorry d ,Robert J. Hopkin d , Donald L. Gilbert  e , Tracy A. Glauser  e , Andrew D. Hershey e , Neil W. Richtand  b,f,g,h , Michael Privitera a  , Arif Dalvi a  , Alok Sahay a  , Jerzy P. Szaflarski a  ,David M. Ficker  a  , Nancy Ratner   b,g , Frank R. Sharp a,b,e, * a   Department of Neurology, University of Cincinnati, Vontz Center, Room 2327, 3125 Eden Avenue, Cincinnati, OH 45267-0536, USA  b The Neuroscience Program, University of Cincinnati, Cincinnati, OH 45267, USA c  Division of Molecular Developmental Biology and Informatics, Cincinnati Children’s Hospital Medical Center, Cincinnati, OH 45229, USA d  Division of Human Genetics, Cincinnati Children’s Hospital Medical Center, Cincinnati, OH 45229, USA e  Division of Pediatric Neurology, Cincinnati Children’s Hospital Medical Center, Cincinnati, OH 45229, USA f   Department of Psychiatry, Cincinnati, OH 45267, USA g  Department of Cell Biology, Neurobiology and Anatomy, Cincinnati, OH 45267, USA h  Psychiatry Service, Cincinnati Veterans Affairs Medical Center, Cincinnati, OH 45267, USA Accepted 14 October 2003Available online 1 January 2004 Abstract Application of gene expression profiling to human diseases will be limited by availability of tissue samples. It was postulated that germline genetic defects affect blood cells to produce unique expression patterns. This hypothesis was addressed by using a test neurologicaldisease-neurofibromatosis type 1 (NF1), an autosomal dominant genetic disease caused by mutations of the  NF1  gene at chromosome17q11.2. Oligonucleotide arrays were used to survey the blood gene expression pattern of 12 NF1 patients compared to 96 controls. A groupof genes related to tissue remodeling, bone development and tumor suppression were down-regulated in NF1 blood samples. In addition,there were blood genomic patterns for gender and age: Y chromosome genes showing higher expression in males, indicating a gene-dosageeffect; and genes related to lymphocyte functions showing higher expression in children. The results suggest that genetic mutations can bemanifested at the transcriptional level in peripheral blood cells and blood gene expression profiling may be useful for studying phenotypicdifferences of human genetic diseases and possibly providing diagnostic and prognostic markers. D  2003 Elsevier B.V. All rights reserved. Theme:  Disorders of the nervous system Topic:  Genetic models  Keywords:  Neurofibromatosis; Age; Gender; Genomics; Microarray; Blood; Gene expression profiling; Marker  1. Introduction Gene expression profiling using microarray technologyhas the potential to be a powerful tool in diagnosis andclassification of cancers [1,8,11,35]. However, its applica- tion to the study of other human diseases may be limited bydifficulties in obtaining routine tissue samples from patients.This stimulated us to determine if blood—the most readilyavailable tissue in the human body—can be used for expression profiling of some human diseases. This notionwas supported by our previous findings that acute injurymodels in rats including stroke, cerebral hemorrhage, seiz-ures, hypoglycemia and hypoxia produced a unique geneexpression profile in lymphocytes [27]. However, it was unclear whether this could be applied to human diseases. 0169-328X/$ - see front matter   D  2003 Elsevier B.V. All rights reserved.doi:10.1016/j.molbrainres.2003.10.014 $ Supplementary data associated with this article can be found, in theonline version, at doi:10.1016/j.molbrainres.2003.10.014.* Corresponding author. Department of Neurology, University of Cincinnati, Vontz Center, Room 2327, 3125 Eden Avenue, Cincinnati,OH 45267-0536, USA. Tel.: +1-513-558-7082; fax: +1-513-558-7009.  E-mail address: (F.R. Sharp) Brain Research 132 (2004) 155–167  Leukemias associated with a specific chromosome trans-location or a trisomy produced unique expression patternsthat were distinct from other types of leukemia [2,31].Hereditary breast cancers due to mutations of two separategenes (BRCA1 and BRCA2) exhibit distinct gene expres-sion profiles that differ from the profiles of sporadic breast cancers [9]. These findings raised the possibility that chro-mosome abnormalities or individual gene mutations might result in characteristic gene expression profiles in a varietyof diseases in addition to cancer.We postulated that mutations of genes, that were passedthrough the germline and that were expressed in blood cells,would produce downstream transcriptional changes in the blood cells. If so, gene expression profiling of peripheral blood might provide diagnostic and prognostic markers andcould provide insights into how the genetic mutations produce end-organ phenotypes.To address this hypothesis, Affymetrix human U95Aarrays were used to survey the gene expression patternsfrom 108 human blood samples. First, male and femalesamples were compared to determine if the difference insex chromosomal composition affected the transcriptional pattern in peripheral blood cells. Then samples fromdifferent age groups were compared to determine if devel-opment or aging affected blood genomic expression pat-terns. Finally, neurofibromatosis type 1 (NF1), anautosomal dominant disease caused by the mutation of   NF1  gene on chromosome 17q11.2, was used as a test case to determine whether a single gene defect can cause acharacteristic gene expression pattern in blood. The resultsare consistent with the suggestion that genetic defectsincluding chromosomal abnormalities and single genemutations can be manifested at the transcriptional level in peripheral blood cells. This suggests that blood genomicexpression profiling can be used to supplement end-organtissue genomic studies. 2. Materials and methods 2.1. Blood sample collection and processing  After an informed consent was obtained, a peripheralvenous blood sample was drawn during 2001–2002 either at the University of Cincinnati Medical Center, CincinnatiChildren’s Hospital Medical Center or the Cincinnati Vet-eran Affairs Medical Center. The 108 subjects in this studyincluded healthy people and patients with miscellaneousdiagnoses including NF1, epilepsy, bipolar affective disor-ders, schizophrenia, idiopathic Parkinson’s disease, progres-sive supranuclear palsy, acute migraine headache, chronicdaily headache, or Tourette’s syndrome. The diagnoses of  NF1 were based on clinical criteria established by the National Institute of Health (NIH) Consensus Development Conference on Neurofibromatosis [17,18]. The detailed demographic informations on samples that passed qualitycontrol criteria and were used for the microarray studies are provided in Supplemental Table 1.A 10–15 ml blood sample was drawn from the cubitalvein into lavender top-EDTA containing Vacutainer tubes.The blood was mixed with Trizol LS reagent (Invitrogen,Carlsbad, CA) within 15 min. Total RNA was isolatedaccording to the protocol provided by the manufacturer and was further purified using RNeasy mini kit (Qiagen,Chatsworth, CA). This whole blood RNA isolation protocolwas chosen over several other currently available protocolsthat require either gradient separation of white cells or hypotonic red cell lysis that might add to the technicalvariability. The quality of total RNA was assessed using anAgilent 2001 Bioanalyzer (Agilent, Palo Alto, CA) andquantified by spectrophotometry. 2.2. Microarray processing  Sample labeling, hybridization to arrays and imagescanning were carried out as described in the AffymetrixExpression Analysis Technical Manual. Briefly, 10  A g totalRNA was used to synthesize cDNA that was subsequentlyused as a template to generate biotinylated cRNA. cRNAwas fragmented and hybridized to Affymetrix human U95Agenechips, which contain probe sets for more than 12,000genes and ESTs. After hybridization, microarrays werewashed and scanned with a laser scanner (Agilent, PaloAlto, CA). The Affymetrix GENECHIP software (MAS 4.0)was used to calculate the raw expression value of each genefrom the scanned image. RNA quality was further assessed by examination of the 3  V  –5  V  ratios for actin and glyceral-dehyde-3-phosphate dehydrogenase (GAPDH). Sampleswere excluded if the ratio was greater than 2, or if therewere visible defects on the arrays, or if the hybridizationwas much weaker or stronger than most other arrays. A totalof 108 samples passed these quality controls.A linear scaling procedure was performed so that signalintensities for all genes on an array are multiplied by ascaling factor that makes the average intensity value for each array equal to a preset value of 1500. This procedurescaled the average intensity of all the arrays to the samelevel and made the comparison among different samples possible. Based on the knowledge that microarray measure-ments are more reliable for transcripts of high abundance, atotal of 4528 genes with relatively high expression values(average difference more than 1000 in at least 54 out of 108arrays) were subjected to the following statistical analyses.The other genes were discarded. Analyses were performedusing log 2  expression values to reduce skew for low andhigh expressing genes and improve estimates of variance. 2.3. Microarray data analysis with BRB-Array Tools To determine if there was an expression profile in bloodcorresponding to a particular phenotype (gender, age, or  NF1), a class comparison tool (BRB-Array Tools 2.0, Y. Tang et al. / Molecular Brain Research 132 (2004) 155–167  156  developed by Dr. Richard Simon and Amy Peng at the National Cancer Institute and Emmes Corporation) was usedfor comparing pre-defined classes. This tool performed a parametric  t  -test (two groups) or   F  -test (more than twogroups) for each gene and then computed the number of genes that were differentially expressed between the pre-defined classes at a  P  <0.001 significance level. We chose  P  <0.001 in this study to reduce the number of false- positive genes (4.5 genes were expected by chance from4528 genes studied). The class comparison tool then per-formed 2000 random permutations of the class labels (i.e.which arrays correspond to which classes) and computed the proportion of the random permutations that gave as manygenes significant at   P  <0.001 level. This estimated proba- bility of randomness provided a global test of whether theexpression profiles in the pre-defined classes were signifi-cantly different. In this global testing situation, obtaining a  P  value of less than 0.05 is sufficient to establish that expression profiles between two classes are different (seeTechnical report 001 at  f  brb/ TechReport.htm). 2.4. Microarray data analysis with significance analysis of  microarray (SAM) software Significance analysis of microarray (SAM) was also usedto assess the significance of observed expression differences between pre-defined classes (such as different age groups, NF1 vs. controls) and provide visually intuitive results.When comparing two pre-defined classes, SAM assigns ascore to each gene on the basis of expression change relativeto the standard deviation of replicates. Instead of using afixed statistical cutoff such as  P  <0.001, SAM choosesgenes with scores greater than an adjustable threshold  D ,and uses permutation of the repeated measurements toestimate the percentage of genes identified by chance [29]. 2.5. Hierarchical cluster analysis Genes differentially expressed between two classes weresubjected to hierarchical cluster analysis [6] using GENE-SPRING 4.2 (Silicon Genetics, Redwood City, CA). GENE-SPRING normalizes each gene to median of allmeasurements (per gene normalization) and calculates thecorrelation for each gene with every other gene in the set and takes the highest correlation and pairs those two genes,averaging their expression values. It then compares this newcomposite gene with all of the other unpaired genes. This isrepeated until all of the genes have been paired. Genes withsimilar expression profiles were grouped in rows whereassamples with similar impacts on the overall expression wereclustered in columns, with a standard correlation coefficient of 0.95 used as the measure for significant statisticalsimilarity. The branching behavior of the tree was controlledusing a separation ratio setting of 0.5 and a minimumdistance setting of 0.001. 2.6. Quantitative RT-PCR Real-time Taqman RT-PCR was performed on selectedgenes using the 5700 Sequence Detection System (PEBiosystems, Foster City, CA, USA). All primers and probeswere designed using Primer Express 2.0 (PE Biosystems). Aone-step reverse transcription PCR was performed accord-ing to the Taqman One-Step RT-PCR Master Mix ReagentsKit protocol (PE Biosystems). Input RNA amounts werecalculated with relative standard curves for all mRNAs of interest and GAPDH. Normalization to GAPDH was per-formed to account for variability in the initial concentrationand quality of total RNA, and in the conversion efficiencyof the reverse transcription reaction. 3. Results 3.1. Blood genomic expression pattern for gender  To test whether sex chromosome differences produceddifferences in blood genomic expression profiles, 26 female blood samples were compared to 26 male blood samples.These samples were randomly selected from 51 female blood samples and 57 male blood samples. There was no bias toward a particular disease when selecting samples andthere were no significant age differences between the femaleand male groups. Using the parametric  t  -test, 24 genes weredifferentially expressed in whole blood in males comparedto females (  P  <0.001, Table 1). A global permutation-based test (using BRB-Array Tools, see Section 2) showed that among the 2000 random permutations, only 2.7% of the permutations gave as many as 24 genes at the  P  <0.001level, suggesting that the observed differences in the tran-scription of these 24 genes in the blood related to gender were statistically significant. Further supporting this con-clusion, hierarchical clustering demonstrated a clear separa-tion of male from female whole blood samples (Fig. 1a) andthe gender of the remaining human samples can be accu-rately determined by using the expression profile of these 24genes.The most regulated genes related to gender were locatedon the Y chromosome (Table 1). Ribosomal protein S4 Y- linked (RPS4Y) and DEAD/H box polypeptide Y (DBY)showed the greatest difference between genders, and thesegenes alone could potentially serve as gender markers in blood (Fig. 1b, c). In contrast, Smcy homologue Y (SMCY) and protein kinase Y-linked (PRKY) only showed a slightlyhigher expression in blood from males (Fig. 1d, e). Two other genes on the Y chromosome—USP9Y (NM _ 004654)and UTY (AF000994)—though not highly expressed in blood (not included in the list of 4528 genes of highexpression, see Section 2), showed higher expression inmale samples. It should be noted that the detection of the Ychromosome genes in female blood samples is likely to bespurious, and can relate to other transcripts in the female Y. Tang et al. / Molecular Brain Research 132 (2004) 155–167   157   blood samples that cross-hybridize with the probes for themale genes on the Y chromosome.However, besides genes on the Y chromosome, genes onother chromosomes also showed differential expression between genders (Table 1). Although only one gene on the X chromosome—ribosomal protein S4 X-linked(RPS4X) (Fig. 1f) was more highly expressed in femalescompared to males using the criteria in this study, at least 10other genes on the X chromosome showed higher expres-sion in females if less stringent criteria were used (  P  <0.05,data not shown). Furthermore, genes located on other chromosomes such as eukaryotic translation initiation factor 1A (Table 1, Fig. 1g), also showed significant differential expression between female and male blood samples. 3.2. Blood genomic expression pattern related to age Since subjects in our study had a wide age range, wedetermined whether age affected blood genomic expression profiles. The 108 samples were arbitrarily separated intothree groups: children (<15 years old,  n =50); younger adults (15–50 years old,  n =36); and older adults (>50 yearsold,  n =22). A parametric  F  -test indicated that 144 geneswere differentially expressed in blood samples from differ-ent age groups (  P  <0.001). A global permutation-based test showed that among 2000 random permutations, only <0.1% permutations gave as many as 144 genes at   P  <0.001 level,thus establishing age as an important determinant of bloodgene expression. Hierarchical cluster analysis using these144 genes can roughly segregate children from adults but cannot separate younger and older adults (Fig. 2a). Howev- er, it is apparent that even the children form a heterogeneousgroup that can be separated into at least three subgroups.The basis of the three subgroups of children is unclear, but does not relate to the further refinement of age group. It neither relates to their concomitant neurological diseases,which are evenly distributed in the subgroups. The genesthat correlate best with age are the immunoglobulins that show higher expression in most of the children compared toadults.Three separate parametric  t  -tests were then performed tocompare children to older adults, children to younger adultsand younger adults to older adults. For each of thesecomparisons, a global permutation-based test was per-formed to test if the observed differential expression wassignificant using 2000 random permutations. For three Table 1Selected genes that are differentially expressed in whole blood of males compared to females using Affymetrix oligonucleotide arraysSystemic # Genbank # Description Chromosome locationHigh in male 41214 _ at NM _ 001008 Ribosomal protein S4, Y-linked Yp11.338355 _ at NM _ 004660 DEAD/H (Asp-Glu-Ala-Asp/His) box polypeptide,Y chromosomeYq1137583 _ at NM _ 004653 Smcy homologue, Y chromosome (mouse) Yq1140030 _ at NM _ 002760 Protein kinase, Y-linked Yp11.2198 _ g _ at NM _ 002513 Non-metastatic cells 3, protein expressed in 16q1338451 _ at NM _ 006830 Ubiquinol-cytochrome  c  reductase (6.4 kDa) subunit 19p13.339347 _ at NM _ 004069 Adaptor-related protein complex 2, sigma 1 subunit 19q13.2–q13.338483 _ at NM _ 015343 Hypothetical protein 17p1339317 _ at AK000716 Cytidine monophosphate-  N  -acetylneuraminic acidhydroxylase (CMP-  N  -acetylneuraminatemonooxygenase)6p22–p23High in female 34643 _ at AK026741 Ribosomal protein S4, X-linked Xq13.138110 _ at NM _ 005625 Syndecan binding protein (syntenin) 8q1234498 _ at NM _ 004665 Vanin 2 6q23–q2434951 _ at NM _ 006018 Putative chemokine receptor; GTP-binding protein 12q24.3134278 _ at NM _ 001412 Eukaryotic translation initiation factor 1A 1q42.337685 _ at NM _ 007166 Phosphatidylinositol binding clathrin assembly protein11q14663 _ at NM _ 001412 Eukaryotic translation initiation factor 1A 1q42.336322 _ at NM _ 004479 Fucosyltransferase 7 (alpha (1,3) fucosyltransferase) 9q34.332877 _ i _ at AA524802 EST36979 _ at NM _ 006931 Solute carrier family 2 (facilitated glucosetransporter), member 312p13.335830 _ at AB002368 KIAA0370 protein 16p11.21031 _ at NM _ 003137 SFRS protein kinase 1 6p21.3–p21.231499 _ s _ at J04162 Fc fragment of IgG, low affinity IIIb, receptor for (CD16)1q2338411 _ at AK025583 Homo sapiens cDNA: FLJ21930 fis, cloneHEP04301, highly similar to HSU90916 Humanclone 23815 mRNA sequence1137493 _ at NM _ 000395 Colony stimulating factor 2 receptor, beta,low-affinity (granulocyte-macrophage)22q13.1 Y. Tang et al. / Molecular Brain Research 132 (2004) 155–167  158  comparisons, the probability of generating the same number of genes by chance was 0.1%, 0.35%, and 14.4% for eachcomparison, respectively. The results of the above compar-isons were further demonstrated by using SAM software[29] (see Section 2). Consistent with the results obtainedfrom BRB-Array Tools, there were significant numbers of  Fig. 1. Genes differentially expressed in the blood of males compared to females. (a) Hierarchical cluster analysis of 24 genes demonstrates differentialexpression between 26 male and 26 female blood samples. A parametric  t  -test (BRB-Array Tools 2.0) was performed on 4528 genes that were highly expressedin blood to derive a group of 24 genes that were significantly regulated in males vs. females (  P  <0.001). These genes were subjected to a hierarchical cluster analysis using Genespring software. Each gene was normalized to the median of 52 measurements so that its relative expression in each sample was indicated by the fold change relative to the median as represented by the color of the squares. (b–g) The expression of six representative genes obtained from themicroarray data with differential expression in male and female blood samples. Each gene was normalized to the median of 108 measurements and thenormalized values were plotted for a total of 57 male and 51 female samples in our data set. For RPS4Y (b) and DBY (c), the expression values in somefemales were negative using the Affymetrix software, which indicates they were too low to be accurately detected. Y. Tang et al. / Molecular Brain Research 132 (2004) 155–167   159
Similar documents
View more...
Search Related
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