Upgrade to Pro — share decks privately, control downloads, hide ads and more …

The Systems Genetics of Sex Determination

8e4bf6269bc939dfd942996af10e070a?s=47 Steve Munger
November 19, 2014

The Systems Genetics of Sex Determination

Slides from my Center for Public Health Genomics at UVA on 11/19/2014.

8e4bf6269bc939dfd942996af10e070a?s=128

Steve Munger

November 19, 2014
Tweet

Transcript

  1. Finding  Sex  in  the  Circuitry:   Toward  a  predic7ve  network

     model  of   mammalian  sex  determina7on.   Steven  Munger   Postdoctoral  Research  Fellow,  Gary  Churchill  Group   The  Jackson  Laboratory,  Bar  Harbor  Maine  
  2. The  systems  gene7cs  of  sex  determina7on.   •  Antagonis7c  pathways

     drive  sex  determina7on  in  mammals.   •  Gonad  gene  expression  is  highly  dynamic  during  the  cri7cal   window  of  sex  determina7on.   –  Temporal  analysis  iden7fies  gene  ac7va7on  and  repression  in  the  tes7s.   –  Expression  QTL  mapping  iden7fies  gene7c  interac7ons  at  E11.5.   –  In  vitro  valida7on  of  a  novel  regulator  of  sex  determina7on.   •  An  honest  assessment  of  my  limita7ons  circa  2011.   •  A  brief  diversion  into  analysis  of  the  adult  liver  transcriptome.   –  New  mice,  new  technologies,  new  analy7cal  methods,  new  skillset.   •  Present/Future  Plans   –  A  comprehensive  systems  gene7cs  approach  to  elucidate  the  transcrip7on   network  driving  sex  determina7on.        
  3. The  zygote  contains  the  gene7c  informa7on  content     required

     to  make  all  differen7ated  cell  types.  
  4. Bipoten7al  gonad   Ovary   Tes7s   How  does  one

     pluripotent  cell  decide  between  two  fates?     The  example  of  sex  determina7on.   ~36  hour  bipoten7al  window  E10.0-­‐11.5  
  5.           granulosa  cells   suppor-ng  cell

     lineage   steroidogenic  cell  lineage   primordial  germ  cell  lineage   spermatogonia   Leydig  cells   Sertoli  cells   Theca  cells   oogonia   All  cells  in  the  bipoten7al  gonad  are  ini7ally  balanced     between  alternate  sexual  fates.   Anne  McLaren          Paul  Burgoyne          Robin  Lovell-­‐Badge     Ovary Testis
  6. Ovary   Female   Male   Wnt4      

     Rspo1        Sox9   Fgf9   Female   Male   Wnt4        Rspo1        Sox9   Fgf9   E10.5   E11.5   E12.5   Female   Male   Wnt4        Rspo1        Sox9   Fgf9   Bipoten7al  Gonad   Tes-s   Opposing  signals  control  the  sexual  fate  decision.   Yuna Kim, Samantha Jameson, Blanche Capel, and others
  7. 5   1   2   3   4  

    6   7   8   9   10   11   12   13   14   15   16   17   18   19   X   Y   Lhx9 Ptgds SF1 Wt1 Bmp2 Cbln4 Rspo1 Wnt4 Fgfr2 Foxl2 Ctnnb1 Cited2 Amh Sry Sox9 M33 Fst Fgf9 Gata4 Dhh Fog2 Map3k4 Dmrt1 Dkk1 Emx2 Dax1 ~35  genes,  ~15  interac7ons   Few  genes  are  known  to  be  involved  in  sex  determina7on,  and   less  is  known  regarding  their  interac7ons.  
  8. Common  inbred  strains  differ  in  their  sensi7vity  to  sex  

    reversal  in  response  to  gene7c  perturba7on.   129S1/SvImJ  =  “129S1”   DBA/2J  =  “DBA”   C57BL/6J  =  “B6”   Sensitive to XY male-to-female sex reversal Eva Eicher Linda Washburn & others at Jax
  9. 5   1   2   3   4  

    6   7   8   9   10   11   12   13   14   15   16   17   18   19   X   Y   Lhx9 Ptgds SF1 Wt1 Bmp2 Cbln4 Rspo1 Wnt4 Fgfr2 Foxl2 Ctnnb1 Cited2 Amh Sry Sox9 M33 Fst Fgf9 Gata4 Dhh Fog2 Map3k4 Dmrt1 Dkk1 Emx2 Dax1 Dax1-/Y Tda2 Tda1 Tda3 Ods/+ Dax1-/Y Ods/+ Ods/+ 129S1 YPOS Tas Odsm1 Most  modifiers  of  sex  reversal  phenotypes  do  not  harbor   genes  with  known  roles  in  the  process.  
  10. Hypothesis  –  Many  genes  and  pathways  act  to  establish,  

    maintain,  and  ul7mately  disrupt  the  balanced  network.   Female   Male   Wnt4        Rspo1        Sox9   Fgf9   Need  to  understand  the  transcrip7on  network  and   dynamics  underlying  sex  determina7on.  
  11. Many  genes  sensi7ze  the  B6  strain  to     male-­‐to-­‐female

     sex  reversal.   Female     Male   129S1  XY     Female     Male    B6  XY     Hypothesis:  The  transcrip7on  network  underlying  sex   determina7on  in  B6  XY  gonads  is  shided  toward  the  female   fate.  
  12. Fine timecourse analysis 2 strains x 2 sexes x 6

    time points (E11-12) x 3 Biological replicates = 72 samples How  does  gene  expression  differ  between  B6  and  129S1   gonads  in  the  cri7cal  24-­‐hour  window  of  sex  determina7on?  
  13. Thousands  of  genes  differ  by  sex,  strain,  stage,    

    and/or  sex*stage  interac7on.   Munger et al. 2013
  14. Example  genes   Sex  Effect   Stage  Effect   Strain

     Effect   Sex*Stage  Effect  
  15. Can  we  iden7fy  the  exact  7me  of  onset  of  

    sexually  dimorphic  expression?  
  16. Anirudh Natarajan 729 possible Viterbi state paths Only 22 are

    populated. Design HMM to identify onset of sexually dimorphic expression
  17. Most  sexually-­‐dimorphic  gene  expression  between  E11.0-­‐12.0   stems  from  ac7va7on

     or  repression  in  the  tes7s.  
  18. Which  cell  types  are  responsible  for  these   transcrip7onal  cascades?

      Jameson et al. 2012
  19. The  suppor7ng  cell   lineage.  

  20. How  does  B6  differ  from  129S1?  Delayed  ac7va7on  of  the

     tes7s   pathway  and  delayed  repression  of  the  ovarian  pathway.  
  21. Conclusions  Part  I   •  There  are  well-­‐characterized  strain  differences

     in  sex   determina7on  phenotypes.   –  The  common  B6  strain  is  sensi7ve  to  XY  sex  reversal.   •  The  gonad  transcriptome  is  highly  dynamic  during   the  24-­‐hour  cri7cal  period  of  sex  determina7on.     –  Sexually  dimorphic  genes  during  this  window  are   predominately  expressed  in  suppor7ng  cell  precursors.   •  Ac7va7on  of  the  male  tes7s  pathway  and  repression   of  the  female  ovarian  pathway  are  both  delayed  by  5   hours  in  B6  testes.  
  22. Ct RFU Nanoliter-scale qPCR Can  we  iden7fy  the  genes  that

     drive  these  strain  expression   differences  in  the  gonad?  Expression  QTL  (eQTL)  mapping.   B6   129S1   X   F1   Intercross   87  F2  XY  gonads   (from  reciprocal   cross)  
  23. B6  /  B6   B6  /  129S1   129S1  /

     129S1   eQTL  mapping  example   High   Low   Sox9  Expression   LOD  Score  =  0.003   No  correla7on  between  marker   genotype  and  Sox9  expression.   B6  /  B6   129  /  129   B6  /  129   LOD  Score  =  5.94   B6  /  B6   129  /  129   B6  /  129  
  24. Map  modifiers  of  transcript  abundance  for  each  gene.   Munger

     et  al.  Genes  &  Development,  2009  
  25. 5   1   2   3   4  

    6   7   8   9   10   11   12   13   14   15   16   17   18   19   X   Y   Asns Trim47 Gata4 Lhx9 Ptgds SF1 Wt1 Bmp2 Cbln4 Rspo1 Wnt4 Fgfr2 Foxl2 Ctnnb1 Cited2 Amh Sry M33 Fst Fgf9 Dhh Fog2 Map3k4 Dmrt1 Dkk1 Emx2 Dax1 Sox9 Serpine2 Ren1 Prokr2 Cst9 Defb19 Col9a3 Pld1 Dcamkl1 Col27a1 Smpdl3b Afp Ereg Mmd2 Fbln2 Pglyrp1 Pcsk6 Slco3a1 Ptpre Defb7 Cbln1 Pdgfd Tpd52l1 Actr6 Socs2 Tph2 Axin2 Rtn4rl1 Centb1 Sphk1 Dock4 Dapk1 Rec8 Rpgrip1 Smoc2 Sox8 Gng13 Dtna Taf7l Sox9  eQTL  
  26. 5   1   2   3   4  

    6   7   8   9   10   11   12   13   14   15   16   17   18   19   X   Y   Lhx9 Ptgds SF1 Wt1 Bmp2 Cbln4 Rspo1 Wnt4 Fgfr2 Foxl2 Ctnnb1 Cited2 Amh Sry Sox9 M33 Fst Fgf9 Gata4 Dhh Fog2 Map3k4 Dmrt1 Dkk1 Emx2 Dax1 Sex  Determina7on  2008  
  27. Sox8 Trim47 Axin2 6   5   1   2

      3   4   7   8   9   10   11   12   13   14   15   16   17   18   19   X   Y   Asns Lhx9 Ptgds SF1 Wt1 Bmp2 Cbln4 Rspo1 Wnt4 Fgfr2 Foxl2 Ctnnb1 Cited2 Amh Sry Sox9 M33 Fst Fgf9 Gata4 Dhh Fog2 Map3k4 Dmrt1 Dkk1 Emx2 Dax1 Serpine2 Ren1 Prokr2 Cst9 Defb19 Col9a3 Pld1 Dcamkl1 Col27a1 Smpdl3b Afp Ereg Mmd2 Fbln2 Pglyrp1 Pcsk6 Slco3a1 Ptpre Defb7 Cbln1 Pdgfd Tpd52l1 Actr6 Socs2 Tph2 Rtn4rl1 Centb1 Sphk1 Dock4 Dapk1 Rec8 Rpgrip1 Smoc2 Gng13 Dtna Taf7l Sex  Determina7on  2010   19  eQTLs,  70  total  edges;  44  genes  have  eQTL    
  28. The  temporal  expression  data  is  a  powerful  resource  for  

    predic7ng  candidate  regulatory  genes  within  eQTL.  
  29. RNAi  in  gonad  primary  cells  validates  Lmo4  as  the  Chr

     3   eQTL  regulator.     Temporal Expression
  30. Conclusions  Part  II   •  Gonad  gene  expression  is  highly

     heritable,  and  we   can  map  strain  modifiers  (eQTL).   •  The  fine  7mecourse  expression  data  enabled  us  to   priori7ze  candidates  in  eQTL  intervals.     •  I  developed  a  gonad  primary  cell  assay  and  len7viral   shRNA  knockdown  method  to  validate  Lmo4  as  a   novel  regulator  of  gonad  gene  expression.   •  But…  
  31. B6  and  129S1  lack  gene7c  varia7on  in  most  known  

    sex  determina7on  genes.   SD  Gene   SNPs  –  B6  v  129   Amh   0   Dhh   0   Insrr   0   Cbx2/M33   0   Fgf9   2   Sox9   6   Ctnnb1   0   Wnt4   0   Foxl2   0   Rspo1   2   Lhx9   1   Nr5a1/SF1   2  
  32. In  2010,  increased  gene7c  diversity  =  increased   measurement  error.

      Most  expression  plamorms  were  hybridiza7on  microarrays   designed  against  the  reference  (~B6)  strain.   ≠
  33. I  lacked  the  computa7onal  skills  and  sta7s7cal  training  to  be

      successful  in  the  “Big  Data”  integra7ve  genomics  arena.      
  34. None
  35. The  Jackson  Laboratory,  Bar  Harbor,  Maine  

  36. 18M 18M 4M 4M 4M 4M 7M More  gene7c  diversity

     =     More  phenotypic  diversity   129S1/SvImJ C57BL/6J Brynn Voy
  37. The  Collabora7ve  Cross:  A  large  panel  of  recombinant  inbred  

    lines  derived  from  eight  inbred  founder  strains.   CC001– 98% Homozygous
  38. The  complementary  Diversity  Outbred  heterogeneous  stock.   Collaborative Cross Funnel

    Diversity Outbred … G2:F4-F12 mice from 144 different funnels Random Outbreeding
  39. Diversity  Outbred  mice  exhibit  phenotypes  far  exceeding  the   range

     observed  in  the  founder  strains.  
  40. Rodents  of  Unusual  Size  

  41. 2011:  The  emergence  of  High  Throughput  Sequencing  (HTS)   Technology

     for  transcriptome  profiling.  
  42. High  Fat   Chow   Diversity  Outbred  Mice    

    Genera7ons  G4-­‐G12   Phenotyping   Livers   Euthanize  @  26  weeks  
  43. Resources  from  Svenson  850  DO  Project   Animal Information Diet

    Genotype Sex Generation (4-12) Litter No. Born Date Coat Color Other Coat Color Feature Chow or HF Urinalysis Urinary Glu 1 ACR1 Urinary Glu2 ACR2 Change in ACR Electrocardiogram HR HRV PQ PR QRS QTc RR ST QTc dispers ion mean SR amplitude mean R amplitude pNN50 (>6ms) rMSSD Body Composition at two ages: (1) 12 weeks; (2) 21 weeks Length 1 Weight 1 % Fat 1 LTM1 BMD1 BMC1 B Area1 T Area1 RST1 TTM1 Length 2 Weight 2 % Fat 2 LTM2 BMD2 BMC2 B Area2 T Area2 RST2 TTM2 Change in Weight Change in % Fat Organ Weights Heart Spleen Kidney L Kidney R Weekly Body Weights; Growth Curve BW 4 BW 5 BW 6 BW 7 BW 8 BW 9 BW 10 BW 11 BW 12 BW 13 BW 14 BW 15 BW 16 BW 17 BW 18 BW 19 BW 20 BW 21 BW 22 BW 23 BW 24 BW 25 BW 26 Liver Harvest RNAseq Metabol- omics Protein Abundance Other Tissue Harvest Kidney/R RNAseq Muscle (gastrocnemius) RNAseq Pellets: Microbiome Gen 8 Ltr 2 (n=100) 18w Gen 9 Ltr 1 (n=100) 18w, 26w N  =  146*  traits   Genome Dynamics Center for *Other  calculated  traits  can  be  derived   Karen  Svenson  and  Lisa  Somes   15  August  2012   650  Diversity  Outbred  animals  have  been  phenotyped  for  clinically  relevant  traits.  QTL  analysis  is  underway.   Plasma chemistries at two ages: (1) 8 weeks; (2) 19 weeks CHOL1 HDLD1 TG1 Glu1 NEFA1 Ca1 Phos1 GLDH1 BUN1 Change in CHOL2 HDLD2 TG2 Glu2 NEFA2 Ca2 Phos2 GLDH2 BUN2 Chol TG NEFA Glu Hematological parameters at two ages: (1) 10 weeks; (2) 22 weeks WBC1 RBC1 mHGB1 HCT1 MCV1 MCH1 MCHC 1 CHCM1 RDW1 HDW1 PLT1 MPV1 NEUT1 LYM1 MONO1 EOS1 LUC1 BASO1 Retic1 cHGB1 WBC2 RBC2 mHGB2 HCT2 MCV2 MCH2 MCHC 2 CHCM2 RDW2 HDW2 PLT2 MPV2 NEUT2 LYM2 MONO2 EOS2 LUC2 BASO2 Retic2 cHGB2 Insulin (17w) Leptin (17w)
  44. ~  30  million  SE  100bp  reads   Yfg   1.

     Align  reads  to  transcriptome.   Yfg   Yfg   Yfg   Mouse  1   Mouse  2   Mouse  3   x  453  mice   2.  Es7mate  gene  and  isoform  expression.   3.  Map  expression  QTL   RNA-­‐seq  -­‐>  Gene  Expression  -­‐>  expression  QTL  
  45. DO  mice  are  unique  diploid  mosaics  derived  from  the  8

      founder  strains.  300+  recombina7on  events  per  mouse,   ~50M  SNPS,  ~5M  indels.   A/J   A   BL6   B   129   C   NOD   D   NZO   E   CAST   F   PWK   G   WSB   H   1   2   3   4   5   6   7   8   9   10  11  12  13  14  15  16  17  18  19  20  21  22  23  24  25  26  27  28  29  30  31  32  33  34  35  36   A   A   A   A   A   A   A   A   B   B   B   B   B   B   B   C   C   C   C   C   C   D   D   D   D   D   E   E   E   E   F   F   F   G   G   H   A   B   C   D   E   F   G   H   B   C   D   E   F   G   H   C   D   E   F   G   H   D   E   F   G   H   E   F   G   H   F   G   H   G   H   H   Founder strains – 8 possible genotypes Diversity Outcross – 36 Possible Genotype states
  46. Building  an  individualized  diploid  genome.   F F D D

    D D D C C A A A A A A A G G E E E E E E H H H H G G G G B B B B B A A H H H H F F F F F F E E C C C D D Psuedo-phased Chromosomes Two Haploid Chromosomes Genotype DNA @ 8K informative SNPs FG FG DG DG DB DB DB CB CB AA AA AH AH AH AH AF GF GF EF EF EF EE EE EC HC HC HD HD Infer 36-state genotypes Gatti et al. 2014 Munger et al. 2014
  47. Seqnature   Munger  et  al.  2014   Gat  et  al.

     2014  
  48. One  alignment  error  can  cause  two  spurious  eQTL.   Alignment

     to  individualized  transcriptomes  results  in  fewer   spurious  eQTL.     Rps12-ps2 Aligned to NCBIm37 Aligned to DO IRGs
  49. Hebp1 Aligned to NCBIM37 Aligned to DO IRGs Alignment  to

     individualized  transcriptomes   reveals  significant  local  eQTLs  for  2,000+  genes.  
  50. Are  these  unmasked  local  eQTLs  real?   CC/DO Founder Strain

    samples
  51. The  founder  origin  of  each  allele  provides  direct  es7mates  of

      allele  specific  expression.   N=277 DO samples N=554 allele-specific estimates Only alleles derived from 129S1 express Gm12976 in the DO population.
  52. Liver  Transcriptome  Map   21,316  genes  expressed   above  threshold

      11,932  (56%)  have  eQTL     10,243  (86%)  are  local   1,689  are  distant  
  53. Predic7ng  candidate  causal  variants   from  DO  founder  sequences.  

  54. The  power  of  8  (founder  strains).     Of  50,000

     SNPs  in  the  region,  only  22  match  the  inferred   strain  effect  pavern.  
  55. Integra7ng  ENCODE  data   Slc22a2 Get Liver ENCODE DNase Hypersensitive

    regions Get 129/NOD/PWK SNPs Intersect and search for TFBSs Dan Gatti, DOQTL
  56. Nkx3-­‐2   T T G A G T G A

    A 129S1/SvImJ NZO/HlLtJ PWK/PhJ Reference T A G A G T G A A JASPAR database
  57. Trans-­‐eQTL  are  the  key  to  building  a  network.    

    Lrtm1 Hypothesis:  Gene  on  Chr  7   Lrtm1  exp  on  Chr  14  
  58. Predic7ng  regulators  underlying  trans-­‐eQTL.   Lrtm1   Lrtm1  -­‐  leucine-­‐rich

     repeats  and  transmembrane  domains  1   cis-­‐eQTL   trans-­‐eQTL   331  genes  within  +/-­‐  5Mb  of  the  trans-­‐eQTL  peak  SNP  
  59. Condi7on  Lrtm1  eQTL  scan  on  local  genotype.   Lrtm1  

  60. Regress  out  the  expression  of  every  candidate  gene  in  

    the  region.   Lrtm1   Condi7oning  the  Lrtm1  eQTL  scan  on  the  expression  of  Igsf23     knocks  down  the  Chromosome  7  peak  from  15  -­‐>  6  LOD.  
  61. Founder  paverns  at  Chr  7  suggest  that  Igsf23   represses

     Lrtm1  expression  in  the  liver.   Expression  of  Lrtm1   129,  NOD,  PWK  Low   is  controlled  by  Igsf23     129,  NOD,  PWK  High   Testable  Hypothesis:      IGSF23   Lrtm1   One  example  of  100’s  in  the  DO  liver  transcriptome.  
  62. Conclusions  Part  III   •  Diversity  Outbred  mice  are  ideal

     for  high-­‐resolu7on   mapping  of  expression  QTL.   •  We  have  developed  methodology  for  dealing  with   exploi7ng  the  high  gene7c  diversity  in  this   popula7on.   •  Combining  mapping  studies  in  gene7cally  diverse   popula7ons  with  mul7ple  levels  of  “seq”  data  (RNA-­‐ seq,  Dnase  I/ATAC-­‐seq,  ChIP-­‐seq)  provides  the   informa7on  required  to  predict  gene  regulatory   networks.  
  63. Can  we  apply  this  approach  to  sex  determina7on?   Lrtm1

  64. Diversity  Outbred  mice  have  10x  more  gene7c   varia7on  in

     known  sex  determina7on  genes.   SD  Gene   SNPs  –  B6  v  129   SNPS  –  CC/DO   Founders   Amh   0   20   Dhh   0   19   Insrr   0   250   Cbx2/M33   0   99   Fgf9   2   279   Sox9   6   65   Ctnnb1   0   241   Wnt4   0   255   Foxl2   0   16   Rspo1   2   389   Lhx9   1   218   Nr5a1/SF1   2   235  
  65. Problem  1:  Hybrid  vigor.   Sex  reversal  has  not  been

     observed  in  the   outbred  Diversity  Outbred  popula7on  L.   Nondisjunc7on  is  observed  (XO,XXY)  at  an   appreciable  rate.  
  66. Solu7on:  Sensi7ze  the  background  to  sex  reversal.  Loss  of  

    Dax1  causes  XY  sex  reversal  only  on  the  B6  background.     Bouma  et  al  2005   Sex  reversed  XY  Ovary  
  67. XY * XX * XX * XY XY * *

    XX XY …x200   X DO  Dax1+/Y   B6  Dax1-­‐/+   (B6xDO)F1  Dax1-­‐/+  and  Dax1-­‐/Y  embryos   A  sensi7zed  screen  of  gonad  eQTL  in  the  DO.   *Only showing X and Y Chromosomes from (B6xDO)F1’s
  68. Can  we  apply  this  approach  to  sex  determina7on?  Yes.  

    Lrtm1 DATA  COMING  SOON!  
  69. Collaborative Cross Strains Validate  predic7ons  by  “breaking”  the  sex  

    determina7on  network  with  crosses  of  CC  strains.   CC  Strain   Sox9   Fgf9   Mapk3   Wnt4   Foxl2   Predic-on   CC001   +++   +   +   -­‐-­‐   -­‐   Tes7s++   CC002   -­‐   +   +   +   -­‐-­‐   Balanced   CC003   -­‐   -­‐   -­‐   +   +   Ovary+   CC004   -­‐   -­‐   -­‐-­‐-­‐   -­‐   +++   Ovary++   CC005   +   -­‐-­‐   ++   -­‐   +   Tes7s+   CC006   -­‐   +   -­‐   +   -­‐   Balanced   CC007   -­‐   -­‐   +   +++   -­‐   Ovary++   CC004  x  CC007   =  XY  Sex  Reversal?  
  70. Acknowledgements   Anirudh  Natarajan   Collaborators   David  Threadgill  -­‐

     NCSU   David  Aylor  -­‐  UNC   Paul  Magwene  –  Duke   Loren  Looger  –  HHMI   Eva  Eicher  -­‐  JAX   Funding   Dept  of  Cell  Biology  Innova7on  Award   DUMC  Bridge  Funding   Center  for  Systems  Biology  Seed  Grant   NICHD  ARRA  Supplement     Equipment/Training   Fernando  Pardo-­‐Manuel  de  Villena   Tim  Bell,  Haider  Ali   Blanche Capel
  71. Acknowledgements  

  72. “Inferences  about  biological  phenomena  are  rarely  separable  from   the

     geneQc  context  in  which  they  are  embedded.  If  you  wish  to   generalize  your  results  across  geneQc  backgrounds,  you  need  to  do   experiments  across  geneQc  backgrounds.”                                                                                                                                -­‐  William  Valdar    
  73. Thank  you!