Timing Matters in T Cell Differentiation

Size: px

Start display at page:

Download "Timing Matters in T Cell Differentiation"

Eleanore O’Neal’
5 years ago
Views:

1 Virtual PI Meeting April p p3 p Timing Matters in T Cell Differentiation Natasa Miskov-Zivanov University of Pittsburgh 010

Acknowledgements 2 Faeder Lab: Department of Computational and Systems Biology, School of Medicine, University of Pittsburgh John Sekar, James Faeder Morel Lab: Department of Immunology, School of

2 Acknowledgements 2 Faeder Lab: Department of Computational and Systems Biology, School of Medicine, University of Pittsburgh John Sekar, James Faeder Morel Lab: Department of Immunology, School of Medicine, University of Pittsburgh Michael Turner, Penelope Morel Clarke Lab: Computer Science Department, Carnegie Mellon University Paolo Zuliani, Haijun Gong, Edmund Clarke Deepa Sathaye, Alexander Moser Funding: NSF (Expeditions in Computing) NIH

(Interaction map) Set of discrete values for each element Regulatory T (Treg) cells

3 Outline 3 System Antigen presenting cell (APC) Methodology Model design Model elements Naïve T cell Experiments Expert knowledge Literature Influence sets (Interaction map) Set of discrete values for each element Regulatory T (Treg) cells Helper T (Th) cells Model analysis Influence table Model rules Circuit design methods Model simulations

4 T cell differentiation 4 Antigen presenting cell (APC) Tumor cell Tumor secreted cytokines (e.g., TGF ) Naïve T cell T cell subpopulation ratios are critical for numerous immune and auto-immune pathologies Release cytokines that inhibit the immune response () Regulatory T (Treg) cells Helper T (Th) cells Release cytokines that stimulate the immune response () Determinants of the peripheral T cell differentiation are not completely understood

5 Questions to address 5 How can we predict the antigen dose that will induce Treg versus Th? Are there signaling cascades in T cells that are critical in this cell fate decision? Can we use modeling to identify the critical factors? Many clinical trials involving DC vaccines do not take antigen dose into account Could also be important for the ex vivo expansion of Treg for therapeutic purposes

6 Modeling goals 6 Determine whether known mechanisms are sufficient to explain experimental observations Suggest additional experiments to identify missing mechanisms and clarify areas of uncertainty Identify early markers of the response

7 Network model 7 )?A 3)0 (CA!"82 K#? -L )?O $!002 $30- $30N =40 =I@ $%# &#)!"# "0#8% #*+, #*-+ 4(8!& 4(8$9 "./0 "&34 "."/ 1,02 51, &'(! &'(!) "*02!67 )%39 :&;)#2 1$!*/ 4(8!&!"82 1&!&J 4(8!& 1$!*/ 1&!&J %&'() 4(8!&!"82 4(8$9 (CM5/ 1&!&J!"#$*!+ 4(8!&!"82 4(8$9!"#$?D7EF?7EC@ E@GEHE7EC@ 7>?@ABCD?7EC@.<8-K 3P "! #.<8-) =!0/ 1&!&J 4(8!& 4(8$9 1$!*/ 1&!&J (CM5/.<8-)# &1#28&1#- :&;)#-.<8- Receptors: T cell receptor (TCR) Co-stimulation through CD28 receptor (R) TGF receptor (TGF R) Transcription factors: AP-1, NFAT, NF B, SMAD3, STAT5 Genes:, CD25, Other important elements:, PI3K, PIP3, PDK1, Akt, mtorc1, mtorc2, TSC1-TSC2, Rheb, S6K1, ps6

8 Five scenarios 1. High antigen dose 2. Low antigen dose 3. High antigen dose, then removed Ras Raf MEK2 ERK TAK1 MKK7 JNK MHC TCR Ca 2+ APC PKC-θ NF-AT CD86 CD28 NF-κB PI3K mtorc2 PIP3 S6K1 ps6 TGFβ TGFβR PDK1 Akt TSC1-TSC2 RHEB mtorc1 SMAD3 activation inhibition translocation EX γ β α R JAK3 STAT5 4. High antigen dose and TGF 8 5. High antigen dose, then inhibitors added Fos Jun NF-AT NF-κB SMAD3 STAT5 AP-1 NF-AT AP-1 STAT5 NF-AT SMAD3 STAT5 FOXP3 Rα NF-AT AP-1 NF-κB STAT5 Rα NF-AT AP-1 NF-κB

9 Scenarios 1 and 2: High and Low antigen dose High dose Low dose 9 CFSE

10 Scenarios 1 and 2: High and Low antigen dose High dose Low dose Simulation trajectories 0 0 Update round Update round 0 Source: N. Miskov-Zivanov et al., in preparation.

11 Scenarios 1 and 2: High and Low antigen dose High dose Low dose Average simulation trajectories Simulation trajectories TCR Ras CD25 PI3K Akt mtorc1 mtorc2 C 0 30 Update round 0 30 Update round Source: N. Miskov-Zivanov et al., in preparation.

12 Scenarios 1 and 2: High and Low antigen dose High dose Low dose 12 Experiments 1000 Simulation trajectories 1023 Simulations Average simulation trajectories 0 TCR Ras CD25 PI3K Akt mtorc1 C mtorc2 0 Update round Update round Source: N. Miskov-Zivanov et al., in preparation.

13 Scenario 1: High antigen dose MHC TCR APC CD86 CD28 TGFβ TGFβR activation inhibition translocation EX γ β α R 13 value = ON (1) value = OFF (0) Ras Raf TAK1 PKC-θ PI3K PIP3 PDK1 Akt SMAD3 JAK3 STAT5 MEK2 MKK7 Ca 2+ mtorc2 S6K1 TSC1-TSC2 RHEB ERK JNK NF-AT NF-κB ps6 mtorc1 Fos Jun NF-AT NF-κB SMAD3 STAT5 AP-1 NF-AT AP-1 STAT5 NF-AT SMAD3 STAT5 FOXP3 NF-AT AP-1 NF-κB STAT5 Rα Rα NF-AT AP-1 NF-κB

14 Scenario 1: High antigen dose MHC TCR APC CD86 CD28 TGFβ TGFβR activation inhibition translocation EX γ β α R 14 value = ON (1) value = OFF (0) Ras Raf TAK1 PKC-θ PI3K PIP3 PDK1 Akt SMAD3 JAK3 STAT5 MEK2 MKK7 Ca 2+ mtorc2 S6K1 TSC1-TSC2 RHEB ERK JNK NF-AT NF-κB ps6 mtorc1 Fos Jun NF-AT NF-κB SMAD3 STAT5 AP-1 NF-AT AP-1 STAT5 NF-AT SMAD3 STAT5 FOXP3 NF-AT AP-1 NF-κB STAT5 Rα Rα NF-AT AP-1 NF-κB

15 Scenario 1: High antigen dose MHC TCR APC CD86 CD28 TGFβ TGFβR activation inhibition translocation EX γ β α R 15 value = ON (1) value = OFF (0) Ras Raf TAK1 PKC-θ PI3K PIP3 PDK1 Akt SMAD3 JAK3 STAT5 MEK2 MKK7 Ca 2+ mtorc2 S6K1 TSC1-TSC2 RHEB ERK JNK NF-AT NF-κB ps6 mtorc1 Fos Jun NF-AT NF-κB SMAD3 STAT5 AP-1 NF-AT AP-1 STAT5 NF-AT SMAD3 STAT5 FOXP3 NF-AT AP-1 NF-κB STAT5 Rα Rα NF-AT AP-1 NF-κB

16 Scenario 1: High antigen dose MHC TCR APC CD86 CD28 TGFβ TGFβR activation inhibition translocation EX γ β α R 16 value = ON (1) value = OFF (0) Ras Raf TAK1 PKC-θ PI3K PIP3 PDK1 Akt SMAD3 JAK3 STAT5 MEK2 MKK7 Ca 2+ mtorc2 S6K1 TSC1-TSC2 RHEB ERK JNK NF-AT NF-κB ps6 mtorc1 Fos Jun NF-AT NF-κB SMAD3 STAT5 AP-1 NF-AT AP-1 STAT5 NF-AT SMAD3 STAT5 FOXP3 NF-AT AP-1 NF-κB STAT5 Rα Rα NF-AT AP-1 NF-κB

17 Scenario 1: High antigen dose MHC TCR APC CD86 CD28 TGFβ TGFβR activation inhibition translocation EX γ β α R 17 value = ON (1) value = OFF (0) Ras Raf TAK1 PKC-θ PI3K PIP3 PDK1 Akt SMAD3 JAK3 STAT5 MEK2 MKK7 Ca 2+ mtorc2 S6K1 TSC1-TSC2 RHEB ERK JNK NF-AT NF-κB ps6 mtorc1 Fos Jun NF-AT NF-κB SMAD3 STAT5 AP-1 NF-AT AP-1 STAT5 NF-AT SMAD3 STAT5 FOXP3 NF-AT AP-1 NF-κB STAT5 Rα Rα NF-AT AP-1 NF-κB

18 Scenario 1: High antigen dose MHC TCR APC CD86 CD28 TGFβ TGFβR activation inhibition translocation EX γ β α R 18 value = ON (1) value = OFF (0) Ras Raf TAK1 PKC-θ PI3K PIP3 PDK1 Akt SMAD3 JAK3 STAT5 MEK2 MKK7 Ca 2+ mtorc2 S6K1 TSC1-TSC2 RHEB ERK JNK NF-AT NF-κB ps6 mtorc1 Fos Jun NF-AT NF-κB SMAD3 STAT5 AP-1 NF-AT AP-1 STAT5 NF-AT SMAD3 STAT5 FOXP3 NF-AT AP-1 NF-κB STAT5 Rα Rα NF-AT AP-1 NF-κB

19 Scenario 1: High antigen dose MHC TCR APC CD86 CD28 TGFβ TGFβR activation inhibition translocation EX γ β α R 19 value = ON (1) value = OFF (0) Ras Raf TAK1 PKC-θ PI3K PIP3 PDK1 Akt SMAD3 JAK3 STAT5 MEK2 MKK7 Ca 2+ mtorc2 S6K1 TSC1-TSC2 RHEB ERK JNK NF-AT NF-κB ps6 mtorc1 Fos Jun NF-AT NF-κB SMAD3 STAT5 AP-1 NF-AT AP-1 STAT5 NF-AT SMAD3 STAT5 FOXP3 NF-AT AP-1 NF-κB STAT5 Rα Rα NF-AT AP-1 NF-κB

20 Scenario 1: High antigen dose trajectory 20 MHC APC CD86 TGFβ activation inhibition translocation EX TCR CD28 TGFβR γ β α R Trajectory example Ras Raf MEK2 ERK Fos AP-1 TAK1 MKK7 JNK Jun Ca 2+ PKC-θ NF-AT NF-κB PI3K mtorc2 PIP3 S6K1 ps6 PDK1 Akt TSC1-TSC2 RHEB mtorc1 SMAD3 NF-AT AP-1 STAT5 NF-AT SMAD3 STAT5 FOXP3 NF-AT AP-1 NF-κB STAT5 Rα NF-AT AP-1 NF-κB JAK3 STAT5 NF-AT NF-κB SMAD3 STAT5 Rα TCR_HIGH PI3K_HIGH PIP3 AKT MTORC1 S6K1 MTORC2 STAT5 CD25 FOXP3 value = ON (1) value = OFF (0)

21 Scenario 1: High antigen dose trajectory 21 MHC APC CD86 TGFβ activation inhibition translocation EX TCR CD28 TGFβR γ β α R Trajectory example Ras Raf MEK2 ERK Fos AP-1 TAK1 MKK7 JNK Jun Ca 2+ PKC-θ NF-AT NF-κB PI3K mtorc2 PIP3 S6K1 ps6 PDK1 Akt TSC1-TSC2 RHEB mtorc1 SMAD3 NF-AT AP-1 STAT5 NF-AT SMAD3 STAT5 FOXP3 NF-AT AP-1 NF-κB STAT5 Rα NF-AT AP-1 NF-κB JAK3 STAT5 NF-AT NF-κB SMAD3 STAT5 Rα TCR_HIGH PI3K_HIGH PIP3 AKT MTORC1 S6K1 MTORC2 STAT5 CD25 FOXP3 value = ON (1) value = OFF (0)

22 Scenario 2: Low antigen dose trajectory 22 MHC APC CD86 TGFβ activation inhibition translocation EX TCR CD28 TGFβR γ β α R Trajectory example Ras Raf MEK2 ERK Fos AP-1 TAK1 MKK7 JNK Jun Ca 2+ PKC-θ NF-AT NF-κB PI3K mtorc2 PIP3 S6K1 ps6 PDK1 Akt TSC1-TSC2 RHEB mtorc1 SMAD3 NF-AT AP-1 STAT5 NF-AT SMAD3 STAT5 FOXP3 NF-AT AP-1 NF-κB STAT5 Rα NF-AT AP-1 NF-κB JAK3 STAT5 NF-AT NF-κB SMAD3 STAT5 Rα TCR_LOW PI3K_LOW PIP3 AKT MTORC1 S6K1 MTORC2 STAT5 CD25 FOXP3 value = ON (1) value = OFF (0)

23 Scenario 2: Low antigen dose trajectory 23 MHC APC CD86 TGFβ activation inhibition translocation EX TCR CD28 TGFβR γ β α R Trajectory example Ras Raf MEK2 ERK Fos AP-1 TAK1 MKK7 JNK Jun Ca 2+ PKC-θ NF-AT NF-κB PI3K mtorc2 PIP3 S6K1 ps6 PDK1 Akt TSC1-TSC2 RHEB mtorc1 SMAD3 NF-AT AP-1 STAT5 NF-AT SMAD3 STAT5 FOXP3 NF-AT AP-1 NF-κB STAT5 Rα NF-AT AP-1 NF-κB JAK3 STAT5 NF-AT NF-κB SMAD3 STAT5 Rα TCR_LOW PI3K_LOW PIP3 AKT MTORC1 S6K1 MTORC2 STAT5 CD25 FOXP3 value = ON (1) value = OFF (0)

24 Scenarios 1 and 2: High and Low antigen dose High dose Low dose 24 Simulation trajectories Experiments * * * * * Average simulation trajectories 0 TCR Ras CD25 PI3K Akt mtorc1 mtorc2 0 Update round 30 Source: N. Miskov-Zivanov et al,. in preparation. C 0 30 Update round )"*+*,*-./"01.2"34+5*"'" 0.0 '!!" &!" %!" $!" #!"!" ;<=#" >1?2"4-@?*-"8:/*" <:0"4-@?*-"8:/*"!" (" '!" '(" #!" 6784.*"9:5-8" Simulations

25 Model does not capture low dose scenario 25 Model simulations of low antigen dose result in 100% + cells, in experiments no more than 50% + cells

26 Model does not capture low dose scenario 26 Model simulations of low antigen dose result in 100% + cells, in experiments no more than 50% + cells Low dose antigen is modeled as a change in rules: PKCTHETA* = TCR_HIGH or (TCR_LOW and CD28 and MTORC2) PI3K_LOW* = (TCR_LOW and CD28) or (PI3K_LOW and IL2_EX and IL2R) PI3K_HIGH* = (TCR_HIGH and CD28) or (PI3K_HIGH and IL2_EX and IL2R) * = (not TCR_HIGH and ) or (not TCR_HIGH and FOXP3)

Model does not capture low dose scenario 27 Model simulations of low antigen dose result in 100% + cells, in experiments no more than 50% + cells Low dose antigen is modeled as a change in rules:

27 Model does not capture low dose scenario 27 Model simulations of low antigen dose result in 100% + cells, in experiments no more than 50% + cells Low dose antigen is modeled as a change in rules: PKCTHETA* = TCR_HIGH or (TCR_LOW and CD28 and MTORC2) PI3K_LOW* = (TCR_LOW and CD28) or (PI3K_LOW and IL2_EX and IL2R) PI3K_HIGH* = (TCR_HIGH and CD28) or (PI3K_HIGH and IL2_EX and IL2R) * = (not TCR_HIGH and ) or (not TCR_HIGH and FOXP3) A more dynamic analysis of TCR signal strength necessary: duration of stimulation

28 Analysis of duration of stimulation 28 Remove TCR after 18 hrs Experiments Source: Sauer et al., PNAS 105:7797, CFSE

29 Scenario 3: Antigen removal at rounds 1-12 (T1-T12) T6 29 No removal Attractors A1 A2 A3 A4 A5 A6 A7 A8 A9 A10 A11 HD LD TCR Ras CD PI3K Akt mtorc mtorc Attractor size :;"<346"=".7>:67"06?3@.A" '!!" -"./ " &!" %!" $!" #!" B3CD)E"FGE#E"" FGE#=" B3CD)="!" (!" ('" (#" ()" ($" (*" (%" (+" (&" (," ('!" (''" ('#" "

trajectories -"./0.12304" '!!" &!" %!" $!

A" B3CD)E"FGE#E"" FGE#=" B3CD)=" TCR Ras

30 Scenario 3: Antigen removal at rounds 1-12 (T1-T12) simulation trajectories -"./ " '!!" &!" %!" $!" #!" B3CD)E"FGE#E"" FGE#=" B3CD)=" TCR Ras CD25 PI3K Akt mtorc1 mtorc2 Average trajectories for attractor !" (!" ('" (#" ()" ($" (*" (%" (+" (&" (," ('!" (''" ('#" "

31 Scenario 3: Antigen removal at round 6 (T6) Experiments Source: Sauer et al., PNAS 105:7797, Remove TCR after 18 hrs 31 CFSE Model simulations HD A5 A6 A7 A4 Akt! mtorc1 LD Akt! mtorc1 A2 A1 A3 A10 A11 A9 A8

Ras CD25 PI3K mtorc2 A6 A7 A5 A4 Akt! mtorc1 LD Akt!

32 Scenario 3: Antigen removal at round 6 (T6) Experiments Source: Sauer et al., PNAS 105:7797, Remove TCR after 18 hrs 32 CFSE Model simulations HD Ras CD25 PI3K mtorc2 A6 A7 A5 A4 Akt! mtorc1 LD Akt! mtorc1 A2 + population + CD25+ population - - population Ras CD25 PI3K mtorc2 A1 Ras A3 A10 A11 A9 A8 PI3K mtorc2

33 Time to reach steady state (T6, HD, LD) -./012"/034"256718"25756" '&" '%" '$" '#" '!" &" %" $" #"!" ('" (#" ()" ($" (*" (%" (+" (&" (," ('!" (''" (9:7;5.:" 84+9/5"+9:*"5;.</="5;<;." '&" '%" '$" '#" '!" &" %" $" #"!" ()" *+,-./" ()0%" 1231" /45." (''" *+,-./" (60''" *47" /45." Treg cells take longer to differentiate than Th cells Simulation trajectories T Update round

34 Model checking 34 SPIN provides yes/no answers SPIN High dose Low dose Specification LTL Formula Verified Specification LTL Formula Verified Does become false forever?!(<>[]!foxp3) Yes Does become true forever?!(<>[]foxp3) Yes Does ps6 become true forever?!(<>[]ps6) Yes Does ps6 stay false forever?!([]!ps6) Yes Does PIP3 become true forever?!(<>[]pip3) Yes Does become true but eventually!(<>il2 && <>[] Yes become false forever?!il2) Does Akt become true forever?!(<>[]akt) Yes Is always true?!([]pten) Yes Does mttorc1 become true forever?!(<>[]mtorc1) Yes Does mtorc become true forever?!(<>[]mtorc) Yes Does S6K1 become true forever?!(<>[]s6k1) Yes Does CD25 become true forever?!(<>[]cd25) Yes Does STAT5 become true forever?!(<>[]stat5) Yes Does STAT5 become true forever?!(<>[]stat5) Yes Does CD25 become true forever?!(<>[]cd25) Yes High dose + TGF! Does become true forever?!(<>[]il2) Yes Does become true forever?!(<>[]il2) Yes Does become true forever?!(<>[]pten) Yes Does become false forever?!(<>[]!pten) Yes Deepa Sathaye, Alexander Moser

35 Probabilistic model checking 35 PRISM allows for analyzing transient behavior Deepa Sathaye, Alexander Moser

36 Probabilistic model checking 36 PRISM allows for analyzing transient behavior simulations )"*+*,*-./"01.2"34+5*"'" '!!" &!" %!" $!" #!"!" ABC#" ;1<2"4-=<*-"8:/*" )"*+*,*-./"01.2"34+5*"'"!" (" '!" '(" #!" 6784.*"9:5-8" '!!" &!" %!" $!" #!"!" A;B#" ;:0"4-<=*-"8:/*" )"*+*,*-./"0."10+2*"'"!" (" '!" '(" #!" 6784.*"9:5-8" '!!" &!" %!" $!" #!"!" )"*+*,*-./"01.2"34+5*"'"!" (" '!" '(" #!" 3450.*"672-5" '!!" &!" %!" $!" #!"!" ;<=#" (" '!" '(" #!" 6784.*"9:5-8" PRISM

37 Statistical model checking 37 Statistical model checker Model Simulations Trace statistics More traces? New model New experiments

38 Statistical model checking 38 Low antigen dose scenario: 1. Does always go to 1? Property: F[20] (IL2 == 1) Test: BEST Result: estimated probability close to 1 2. Probability that stays at 0 before becomes 1? Property: (IL2 == 0) U[15] (FOXP3 == 1) Test: BEST )"*+*,*-./"01.2"34+5*"'" '!!" &!" %!" $!" #!"!" >:?7@" A;B#" ;:0"4-<=*-"8:/*"!" (" '!" '(" #!" 6784.*"9:5-8" Result: estimated probability = rare event Paolo Zuliani

39 Statistical model checking 39 High antigen dose + antigen removal scenario: Studies of relative timing on mtor vs. CD25/ STAT5 pathway Property 1 G 7 ~(MTORC1 = 1 & MTORC2 = 1) 2 F 7 (MTORC1 = 1 & MTORC2 = 1) 3 F 10 (MTORC1 = 1 & MTORC2 = 1 & CD25 = 0 & (F 18 (CD25 = 1))) 4 F28 (MTORC1 == 1 & MTORC2 == 1 & CD25 == 0 & (F 1 (CD25 == 1)) ) 5 F10 (MTORC1 = 1 & MTORC2 = 1 & CD25 = 0 & (F 1 (G 17 (CD25 = 1)))) Probability estimate and sample size estimate = samples = 200,160 estimate = samples = 2,352 estimate = samples = 25,968 estimate = samples = 26,160 estimate = samples = 25,920 Elapsed time [s] 1, )"*+*,*-./"01.2"34+5*"'" '!!" &!" %!" $!" #!"!" ;-<=*-"9*,:34+" >?#@",ABC>'",ABC>#"!" '!" #!" (!" 6784.*"9:5-8" Paolo Zuliani

40 Conclusion 40 Logical modeling approach allows development of comprehensive models of cell fate Model of peripheral T cell differentiation recapitulates a wide range of experimental observations Circuit analysis reveals key elements of the mechanism for expression Timing is critical for Treg differentiation: Treg cells take longer to differentiate than Th cells Race between activating and inhibiting pathways Feedback between and

41 Conclusion 41 Model checking Allows for more efficient studies of the model Probabilistic model checking: Provides transient results that match simulations Statistical model checking: Further analysis of of transient behavior Provides insights into timing relationships between elements

42 Next steps 42 Analyze different removal scenarios using model checking Expansion of the model (keep up with fast pace of developments in the field) Develop a model for several cell types Develop population models that embed intracellular circuitry

43 Thank you! 43

Pancreatic Cancer Research and HMGB1 Signaling Pathway

Pancreatic Cancer Research and HMGB1 Signaling Pathway Haijun Gong*, Paolo Zuliani*, Anvesh Komuravelli*, James R. Faeder #, Edmund M. Clarke* * # The Hallmarks of Cancer D. Hanahan and R. A. Weinberg