Docstoc

Expression profiling at UMC Utrecht

Document Sample
Expression profiling at UMC Utrecht Powered By Docstoc
					DNA microarrays: Applications




      Patrick Kemmeren
       Holstege Lab, UMC Utrecht
               Experiment design


Determined by goal of experiment and degree of accuracy
           Examined thoroughly beforehand

                        Issues
               Replica’s and dye swaps
                 Biological variation
                 Reference sample
Replica’s and dye swaps

           cell culture biological
           harvesting   variation
technical total RNA
variation    cDNA
            labeling
          hybridization

    nr of replica’s? at least 3

      nature of replica’s?
   technical and/or biological
          dye swaps
             Dye swaps
           Counters dye biases

Raw data                         Normalized




        Sources of dye bias
DNA auto fluorescence in cy3 channel
 Cy specific, gene specific labeling
      Cy specific background
   Replica’s and dye swaps

              cell culture biological
              harvesting   variation
   technical total RNA
   variation    cDNA
               labeling
             hybridization

       nr of replica’s? at least 3
          nature of replica’s?
       technical and/or biological
              dye swaps

 replica’s? 4 from biological dye swaps
in combination with statistical error model
                     Biological variation




                Some genes show greater variation

Build an error model from several same vs. same biological replica’s
   Pair wise comparison



Pair-wise comparison (two samples)

     Array #1: A1cy3 vs B1cy5
     Array #2: A2cy5 vs B2cy3
             &: more or
     have genes in duplicate or
   include biological error model


             (screens)
   Reference sample

           Time course
t=0    1       2       3       4       5 (cy5)
t=0    0       0       0       0       0 (cy3)

but t = 0, upregulated gene = 0
  pool samples as reference
               R

  t=0       1      2           3   4     5

 t=0       1       2       3       4    5 (cy5)
 t=R       R       R       R       R    R (cy3)
               Experiment design


Determined by goal of experiment and degree of accuracy
           Examined thoroughly beforehand

                        Issues
               Replica’s and dye swaps
                 Biological variation
                 Reference sample
Applications
mRNA measurement without a microarray



                time (minutes)
           0 15 30 45 60 75




            ….2-3 days work
      mRNA measurement with a microarray
om stationary phase
count                   time
                        Genes
 min              SP   15 60 180   360 min




      ~6000
      genes                                                    Stationary phase exit and entry
   S.cerevisiae                                                               Growth curve
                                                      Text
                                                     10




                                             OD600
                                                      1




                                                     0.1
                                                           0   24   48   72   96    120    144   168   192   216   240
                                                                                   Hours
        Why are mRNA levels important?




    Different cell types have identical set of genes


  Differences in cell phenotype and condition: in part
determined and in part reflected by differences in levels
                  of individual mRNAs
   Applications

    Gene function

    Drug discovery

 Tumor classification

Transcription regulation

        Exotic
       Applications: gene function




The conditions leading to activation of a particular
    gene can lead to insight into its function
Yeast quiescence
Exit and entry from quiescence
Quiescent mutants
           Applications: cellular regulation
om stationary phase
ount       Exit from stationary phase
                     Genes
min              SP   15 60 180   360 min




     ~6000
     genes                                                    Stationary phase exit and entry
  S.cerevisiae                                                               Growth curve
                                                     Text
                                                    10




                                            OD600
                                                     1




                                                    0.1
                                                          0   24   48   72   96    120    144   168   192   216   240
                                                                                  Hours
              Applications: cellular regulation

cellular
environment


                 signal
              transduction
                                               transcription
                                               machinery

         gene specific
          regulators




                             10 000’s genes



                                              Holstege and Young, P.N.A.S. 96, 2-4, 1999
     Mediator complex




   Associated with RNA pol II
Conserved from yeast to mammals
Required for response to activators
      Mechanistic models Srb/Mediator

Signal transduction?              Negative components

                                  Enzymatic activities




  Opposite effects upon deleting different components?
Mediator structure




             In collaboration with Benjamin Guglielmi &
             Michel Werner - CEA/Saclay, France
Non-essential components
    Expression profiling mutants:
   detailed molecular phenotypes


               Requires:
       good arrays and protocols
  rejection of low quality experiments
proper experimental design and statistics
      Expression profiling deletion mutants



                  WT       mt   mt   wt   wt




srb2 Δ1   REF      med2 Δ1 REF                   WT1        REF
  REF   srb2 Δ2     REF med2 Δ2                  REF        WT2
Array#1 Array#2    Array#3 Array#4             Array#31   Array#42

                  15 mutants                    12 WT controls
                        Data analysis
             4 ratio measurements per gene per mutant
    24 control ratio measurements per gene (WT vs REF WT)
Filtered out variation due to culture, array, dye etc., including stress

                               Anova
Gal11.1
Gal11.2
Med3.1
Med3.2
Med2.1
Med2.2
 Srb2.1
 Srb2.2
 Srb5.2
 Srb5.1
 Sin4.1
 Sin4.2
Med1.1
Med9.2
Med1.2
Med9.1
 Nut1.1
 Nut1.2
 Srb8.1
          Duplicates cluster




 Srb8.2
Srb11.1
Srb11.2
 Srb9.1
Srb10.1
 Srb9.2
Srb10.2
Subunit architecture
Subunit architecture
Subunit architecture
Antagonistic effects
Antagonistic effects
Mediator antagonism




      +    -     +
      target genes
Does Cdk/cyclin regulate tail?




           +
           target genes
Epistasis: tail domain is downstream of Srb10/11
Epistasis: tail domain is downstream of Srb10/11
Epistasis: tail domain is downstream of Srb10/11
Med2 is phosphorylated in vivo




                     Med2
Med2 is phosphorylated in a SRB10/11
         dependent manner




                      Med2
Med2 is phosphorylated at S208




      *
      *
Tail phosphorylation by Srb10/11 functional?
Med2 ser 208 is required for response to
          specific activators
         Med2 ser 208 is required for response to
                   specific activators
         S208A
                       50            9 genes with
                                        PDRE
                  18

             77
Dsrb10                 p = 4x10-19




                        p
           Srb10/11 is not only Med2 kinase?


         S208A
                       50            other Med2 kinase?
                  18

             77
Dsrb10                 p = 4x10-19
                                                   Med2


other Srb10/11 kinase targets
Pho85 deletion


                 S208A
                               Dpho85
                     22

                12        28
                     6
           70                  191
                     7
  Dsrb10
Expression profiling Mediator subunits
 Profiles can be viewed as detailed phenotypes

 Agrees with subunit architecture
 Reveals:
 Soh1 is a positive factor
 functional/physical link between Rox3 and Srb10/11
 functional/physical link between Srb2/5 and Tail
 antagonistic submodules within same complex

 Epistasis:
 Tail module downstream of Srb10/11 CDK/cyclin
 Tail subunits phosphorylated in SRB10/11 dependent manner
 Srb10/11 represses through Med2 phosphorylation

 Med2 phosphorylation prevents recruitment through PDR
 activator?
Genome-wide location analysis
       ChIP on chip
                           Grow yeast cultures
                           Crosslink protein to DNA

                           Break open cells and shear
                           DNA


                           Immunoprecipitate


                           Reverse-crosslinks



                           Amplify and label DNA


                           Hybridize to microarray

In collaboration with Michel Werner- CEA/Saclay, France
Applications: diagnostics




                            ?
Diagnosis and treatment of HNSCC

Diagnosis
                                              HNSCC
1. HNSCC
2. neck lymph node metastasis
                                                  clinical neck examination



                          clinical negative            clinical positive
                                 (N0)                        (N+)



                      • primary tumor resection    • primary tumor resection
                                                   • radical neck dissection




                        60% N0         40% N+
   Diagnosis and treatment of HNSCC

   Diagnosis
                                                    HNSCC
   1. HNSCC
   2. neck lymph node metastasis
                                                           clinical neck examination



                                clinical negative               clinical positive
                                       (N0)                           (N+)


2 policies for N0 patient
therapy                     • primary tumor resection       • primary tumor resection
                                                            • radical neck dissection
• “watch and wait”
                                              “wait” for metastasis

                                                                           radical neck
                             60% N0          40% N+                        dissection
                                             risk for distant metastasis
   Diagnosis and treatment of HNSCC

   Diagnosis
                                                         HNSCC
   1. HNSCC
   2. neck lymph node metastasis
                                                              clinical neck examination



                                     clinical negative             clinical positive
                                            (N0)                         (N+)


2 policies for N0 patient
therapy                         • primary tumor resection      • primary tumor resection
                                • neck dissection (smaller)    • radical neck dissection
• “watch and wait”
• neck surgery of all
clinical N0 patients

                                  60% N0             40% N+
              over-treatment of “real” N0 patients
Goal

 Design new diagnostic tool for lymph node metastasis based
 on expression profiling of primary tumors


 • diagnosis is based on primary tumor
 • reliable identification of N-status
Hybridization on 25K Microarray

25K microarray
• 21,521 70-mer oligos
• 3871    control oligos
Selection of tumors
• primary tumors                                                       reference sample
• at least 3 year follow-up
• no treatment before surgery
•  50% tumor cells




                                                      ratio (log)
          82 tumors




                                                                       tumor sample




                                                         ratio (log)




                         red genes: different in tumor compared to reference
 Identify set of genes that is able to distinguish
 between tumors from N0 and N+ patients
                                                up   ratio   down
          102 genes




optimal accuracy

8 FN (25%)
13 FP (26%)

accuracy (74%)
Does prediction accuracy correlate with tumor/patient characteristic ?

            year of surgery




                                                                     optimal accuracy
                                                                     4 FN (24%)
                                                                     2 FP (9%)

                                                                     accuracy 86%
Validate predictor on an independent set of tumors


28 primary tumor (2000 & 2001)




                                                     0 FN (0%)
                                                     3 FP (18%)

                                                     accuracy 89%
Accuracy of predictor and clinical diagnosis




  N0 PV = TN/(TN+FN)
  N+ PV = TP/(TP+FP)
  accuracy = (TN+TP)/(TN+TP+FN+FP)
Neck treatment based on diagnosis


   clinical diagnosis                 microarray predictor
   neg  SOHND                       neg  no neck treatment
   pos  radical                     pos  radical




 accurate treatment – 29%           accurate treatment – 89%
                Applications: drug discovery




Functional Discovery via a Compendium of Expression Profiles
Timothy R. Hughes, ..(30x)..Stephen H. Friend 1, Cell, Vol. 102, 109–126, July, 2000
Holstege Lab
Group leader               Bioinformatics
Frank Holstege             Philip Lijnzaad
                                                                               _
                           Sander van Hooff
Microarray technology      Linda Bakker
Dik van Leenen             Hanneke van Deutekom
Marian Groot-koerkamp      Rodrigo Aldecoa Garcia
Diane Bouwmeester          Patrick Kemmeren

Transcription regulation   Former members
Joris Benschop             Erik Sluiters
Sake van Wageningen        Nynke van Berkum               Genes

Tony Miles                 Harm van Bakel           SP   15 60 180   360 min



Mariel Brok                Jeroen van de Peppel
Nathalie Brabers           Marijana Radonjic
Loes van de Pasch          Paul Roepman
Eva Apweiler               Jean-Christophe
Tineke Lenstra             Andrau
Ines de Castro             Nienke Kettelarij
Thanasis Margaritis        Theo Bijma
                           Joop van Helvoort

				
DOCUMENT INFO
Shared By:
Categories:
Tags:
Stats:
views:0
posted:3/13/2013
language:English
pages:62