6. Model Reduction of Reaction Equations 111
7.2. Model Reduction
7.2.4. Examples
In order to demonstrate the application of the graphical simplification rules, we consider two examples. First, we reduce the model of the enzymatic reaction with competitive inhibition from Example 7.2 and Figure 7.2. Then, we consider the more complex PTS model from Figure 7.3.
Example 7.9 (Enzymatic reaction). In the network from Example 7.2 (p. 140) and Figure 7.2, we assume that the rate-limiting step is the actual conversion of the products A and B into the reactant C in reaction 5. Accordingly, as the first step towards a reduced model, we set R1 = R2 = R3 = R4 = R6 = 0 (Figure 7.11 left). After applying the simplification rule from Figure 7.9 to reactions 1, 3 and 6 we get in step 2 the network in Figure 7.11 (right). By applying rules from Figure 7.5 this can be simplified in step 3 to Figure 7.12 (left). In step 4 we apply again the rule from Figure 7.9 and get Figure 7.12 (right). In step 5, a translation of the stoichiometric coefficients (Figure 7.6) finally leads to the reduced model in Figure 7.13. The figure shows that the inhibitor I and the enzyme E do not participate in the overall reaction, but they influence the reaction rate via ‘side effects’ of their thermokinetic potentials on the capacities C˜A, C˜B and C˜C. The reduced concentrations ˜˜cA,c˜˜B, ˜˜cE and c˜I represent the overall concentrations of A, B, E and I, respectively. This result is consistent with the discussion in Section 6.7 (p. 134).
CEB
CEA
CEI
R5
CEAB
CB
CA CI
CE
CC
R5
CEAB
C˜I
C˜A
C˜B
C˜E
CC
˜
cE =cE+cEI+cEA+cEB, C˜E =CE + ˜ξICEI+ ˜ξACEA+ ˜ξBCEB,
˜
cI =cI+cEI, C˜I =CI+ ˜ξECEI,
˜
cA=cA+cEA, C˜A=CA+ ˜ξECEA,
˜
cB=cB+cEB, C˜B=CB+ ˜ξECEB.
Figure 7.11.: Reduction of the enzymatic reaction model – step 1 and 2.
7. Graphical Representation of TK Models
R5
C˜E
CC
C˜I
C˜A
C˜B
CEAB
R5
˜˜ CE
CC
C˜I
˜˜ CA
˜˜ CB
˜˜
cE = ˜cE+ ˜cEAB C˜˜E = ˜CE+ ˜ξAξ˜BCEAB,
=cE+cEI+cEA+cEB+cEAB, =CE+ ˜ξICEI+ ˜ξACEA+ ˜ξBCEB+ ˜ξAξ˜BCEAB,
˜˜
cA= ˜cA+cEAB C˜˜A= ˜CA+ξ˜˜Eξ˜BCEAB,
=cA+cEA+cEAB, =CA+ξ˜˜ECEA+ξ˜˜Eξ˜BCEAB,
˜˜
cB= ˜cB+cEAB C˜˜B= ˜CB+ξ˜˜Eξ˜ACEAB
=cB+cEB+cEAB, =CB+ξ˜˜ECEB+ξ˜˜Eξ˜A.
Figure 7.12.: Reduction of the enzymatic reaction model – step 3 and 4.
˜˜ CE C˜I
˜˜ CB
R˜5
CC
˜˜ CA
R˜5=ξ˜˜E−1R5
Figure 7.13.: Reduced model of the enzymatic reaction.
7. Graphical Representation of TK Models
Example 7.10 (PTS). In Example 5.20 (p. 94) the thermokinetic resistances and capacities for the PTS model from Rohwer et al. [87] were derived. The graphical representation of this model is given in Figure 7.3.
The model contains 10 resistances and 13 capacities. To reduce the model we approximate the 6 lowest resistances by zero: R1 = R2 = R5 = R6 = R9 = R10 = 0. This leads to the approximated but unreduced model in Figure 7.14 with4resistances but still 13capacities. By applying the rules for the reduction of the kinetic submodel (Figure 7.8), we get the reduced model in Figure 7.15. This model contains only 7 capacities but is not suited for a direct simulation because it contains stoichiors that are directly linked without intermediate capacities or resistances. This means that instead of the removed capacitiesCEI, CEI·P, CHP r·P, CIIA·P, CIICB and CIICB·P, the network in Figure 7.15 contains capacities of value zero. We can apply the rules for the reduction of partially serial reactions from Figure 7.7 to resolve this problem partially. In order to avoid crossing lines, the remaining elements need to be rearranged and one gets Figure 7.16.
By applying the rules for the equivalence of stoichior elements (Figure 7.5), we can completely resolve this problem in the two steps shown in Figure 7.17 and Figure 7.18. This network could be simulated as a TK model. However, the negative stoichiometric coefficients are uncommon and we can get rid of them by a translation of the stoichiometric coefficients (Figure 7.6) and get the reduced model in Figure 7.19. The model contains a conservation relation for each of the four proteinsEI, HP r,IIA and IICB. Only the conservation relations for EI and IICB in Figure 7.19 have the form required for using the simplification rule in Figure 7.10. After reduction of these two conservation relations we get the reduced model in Figure 7.20.
Figure 7.21 shows simulation results of the original and the reduced model. The respective curves are qualitatively similar but partly deviate quantitatively. The fluxes in the reduced model are overestimated in steady state because some of the resistances that add to the overall resistance of the PTS are set to zero. Thus, by increasing the remaining resistances we can get a much better fit of original and reduced model. In a real modeling work, the parameters of the detailed model are not known, but the reduced parameters are estimated using the measurement data. Then, the adapted resistance values can be directly identified from the measurement values.
7. Graphical Representation of TK Models
CEI·P CHP r CIIA·P CIICB
ξP EP
ξP yr
R3R4 R7R8
ξGlc
CEI CIIA
ξGlc·P CEI·P·HP r
CP yr·P·EI CHP r·P·IIA CIIA·P·IICB CIICB·P·Glc
CHP r·P CIICB·P
Figure 7.14.: Approximated but unreduced PTS model.
ξP EP
ξP yr
R3R4 R7R8
ξGlc
ξGlc·P C˜IICB·P·Glc
C˜HP r·P·IIA
C˜P yr·P·EI CEI·P·HP r CIIA·P·IICB
C˜HP r
C˜IIA
C˜P yr·P·EI =CP yr·P·EI+ξP EP−1 CEI+ξP yr−1 CEI·P, ˜cP yr·P·EI =cP yr·P·EI+cEI+cEI·P, C˜HP r·P·IIA=CHP r·P·IIA+ ˜ξIIA−1 CHP r·P + ˜ξHP r−1 CIIA·P, c˜HP r·P·IIA =cHP r·P·IIA+cHP r·P +cIIA·P, C˜IICB·P·Glc =CIICB·P·Glc+ξ−1Glc·P CIICB+ξ−1GlcCIICB·P, ˜cIICB·P·Glc =cIICB·P·Glc+cIICB+cIICB·P,
C˜IIA=CIIA−ξHP r·P·IIAξ˜IIA−2 CHP r·P, ˜cIIA =cIIA−cHP r·P, C˜HP r =CHP r−ξHP r·P·IIAξ˜HP r−2 CIIA·P, c˜HP r =cHpr−cIIA·P.
Figure 7.15.: Reduction of the PTS model – step 1.
7. Graphical Representation of TK Models
C˜HP r·P·IIA
CEI·P·HP r C˜P yr·P·EI C˜IICB·P·Glc CIIA·P·IICB
R˜3
R˜4
ξP yr R˜7
R˜8
ξGlc·P ξGlc
ξP EP C˜IIA
C˜HP r
R˜3=ξP yrR3, R˜4=ξP EPR4, R˜7=ξGlc·PR7, R˜8=ξGlcR8.
Figure 7.16.: Reduction of the PTS model – step 2.
C˜HP r·P·IIA
CEI·P·HP r C˜P yr·P·EI C˜IICB·P·Glc CIIA·P·IICB
R˜3
R˜4
ξP yr R˜7
R˜8
ξGlc·P ξGlc
ξP EP −1
−1
C˜IIA
C˜HP r
Figure 7.17.: Reduction of the PTS model – step 3.
7. Graphical Representation of TK Models
C˜HP r·P·IIA
CEI·P·HP r C˜P yr·P·EI C˜IICB·P·Glc CIIA·P·IICB R˜3
R˜4
ξP yr R˜7
R˜8
ξGlc·P
C˜IIA ξGlc
ξP EP −1
−1 C˜HP r
Figure 7.18.: Reduction of the PTS model – step 4.
R˜3
ξP yr
ξGlc
ξGlc·P
CIIA·P·IICB C˜HP r·P·IIA C˜IICB·P·Glc
C˜P yr·P·EI CEI·P·HP r
R˜8
ˆ˜ R4
ξP EP
ˆ˜ R7
C˜IIA
C˜HP r
ˆ˜
R4= ˜ξIIAR˜4= ˜ξIIAξP EPR4, Rˆ˜7= ˜ξHP rR˜7= ˜ξHP rξGlc·P R7.
Figure 7.19.: Reduction of the PTS model – step 6.
7. Graphical Representation of TK Models
C˜HP r·P·IIA
ξP yr
ξP EP
ξGlc
ξGlc·P
˜˜ R8
˜˜
CIICB·P·Glc
˜˜
CP yr·P·EI
˜˜ R3
˜ˆ R˜4
˜ˆ R˜7
C˜HP r
C˜IIA
ξ˜P yr·P·EI = cEI,tot CEI,tot
˜˜
ξP yr·P·EI, ξ˜IICB·P·Glc= cIICB,tot CIICB,tot
˜˜
ξIICB·P·Glc,
˜
cP yr·P·EI = ˜˜cP yr·P·EI =cP yr·P·EI+cEI+cEI·P,
˜
cIICB·P·Glc= ˜˜cIICB·P·Glc =cIICB·P·Glc+cIICB+cIICB·P,
˜˜
CP yr·P·EI = cEI,tot CEI,tot
C˜P yr·P·EI = cEI,tot
CEI,tot(CP yr·P·EI+ξP EP−1 CEI+ξ−1P yrCEI·P),
˜˜
CIICB·P·Glc= cIICB,tot CIICB,tot
C˜IICB·P·Glc = cIICB,tot
CIICB,tot(CIICB·P·Glc+ξGlc−1·PCIICB+ξGlc−1CIICB·P),
˜˜
R3= CEI,tot cEI,tot
R˜3= CEI,tot
cEI,tot ξP yrR3, R˜ˆ˜4= CEI,tot cEI,tot
ˆ˜
R4= CEI,tot cEI,tot
ξ˜IIAξP EPR4,
˜ˆ
R˜7= CIICB,tot cIICB,tot
ˆ˜
R7= CIICB,tot cIICB,tot
ξ˜HP rξGlc·P R7, R˜˜8= CIICB,tot cIICB,tot
R˜8= CIICB,tot
cIICB,tot ξGlcR8, with
cEI,tot= ˜cP yr·P·EI+cEI·P·HP r =cEI+cEI·P +cP yr·P·EI+cEI·P·HP r,
cIICB,tot= ˜cIICB·P·Glc+cIIA·P·IICB=cIICB+cIICB·P +cIICB·P·Glc+cIIA·P·IICB, CEI,tot= ˜CP yr·P·EIξ˜˜P yr·P·EI+CEI·P·HP r
= (CP yr·P·EI+ξ−1P EPCEI+ξP yr−1 CEI·P)ξ˜˜P yr·P·EI+CEI·P·HP r, CIICB,tot= ˜CIICB·P·Glcξ˜˜IICB·P·Glc+CIIA·P·IICB
= (CIICB·P·Glc+ξ−1Glc·P CIICB+ξ−1GlcCIICB·P)ξ˜˜IICB·P·Glc+CIIA·P·IICB.
Figure 7.20.: Reduced PTS model.
7. Graphical Representation of TK Models J3 [mM min−1]
t [10−1min]
0.5 200
100
00 1.0
300
J4 [mM min−1]
t[10−1min]
0.5 200
100
00 1.0
300
J7 [mM min−1]
t [10−1min]
0 0.5
0 1.0
40 20
J8 [mM min−1]
t[10−1min]
0 0.5
0 1.0
20 40
cIIA [µM]
t [10−1min]
0 0.5
0 1.0
10 5
cHP r [µM]
t[10−1min]
0 0.5
0 1.0
10 20
cHP r·P·IIA [µM]
t [10−1min]
0 0.5
0 1.0
30 20 10
cP yr·P·EI [µM]
t[10−1min]
0 0.5
0 1.0
4 2
Figure 7.21.: Simulation results with the original (solid) and the reduced (dashed) PTS model.
The dotted line shows the reduced model with corrected resistances: R03 = 1.3R3, R04 = 1.3R4, R70 = 1.3R7 and R08 = 1.3R8. The clamped concentrations are cP EP = 2800µM, cP yr = 900µM, cGlc = 500µM, cGlc·P = 50µM. The initial conditions of the detailed model are cEI,0 = 5µM, cHP r,0 = 50µM, cIIA,0 = 40µM, cIICB,0 = 15µM, cEI·P·HP r,0 = cEI·P,0 = cHP r·P·IIA,0 = cHP r·P,0 = cIIA·P·IICB,0 = cIIA·P,0 = cIICB·P·Glc,0 = cIICB·P,0 = cP yr·P·EI,0 = 0.01µM. The initial conditions of the reduced model were computed numerically from the initial conditions of the original model.
7. Graphical Representation of TK Models