Biochemical systems equation

The biochemical systems equation is a compact equation of nonlinear differential equations for describing a kinetic model for any network of coupled biochemical reactions and transport processes. ^{[1] [2]}

The equation is expressed in the following form:

${\displaystyle {\dfrac {\bf {dx}}{dt}}={\bf {N}}{\bf {v}}({\bf {x}}(p),p)}$

The notation for the dependent variable x varies among authors. For example, some authors use s, indicating species. ^[2] x is used here to match the state space notation used in control theory but either notation is acceptable.

${\displaystyle {\bf {N}}}$ is the stoichiometry matrix which is an ${\displaystyle m}$ by ${\displaystyle n}$ matrix of stoichiometry coefficient. ${\displaystyle m}$ is the number of species and ${\displaystyle n}$ the number of biochemical reactions. The notation for ${\displaystyle {\bf {N}}}$ is also variable. In constraint-based modeling the symbol ${\displaystyle {\bf {N}}}$ tends to be used to indicate 'stoichiometry'. However in biochemical dynamic modeling ^[3] and sensitivity analysis, ${\displaystyle {\bf {N}}}$ tends to be in more common use to indicate 'number'. In the chemistry domain, the symbol used for the stoichiometry matrix is highly variable though the symbols S and N have been used in the past. ^{[4] [5]}

${\displaystyle {\bf {v}}}$ is an n-dimensional column vector of reaction rates, and ${\displaystyle p}$ is a p-dimensional column vector of parameters.

Example

Given the biochemical network:

${\displaystyle X_{o}{\stackrel {v_{1}}{\longrightarrow }}\ x_{1}{\stackrel {v_{2}}{\longrightarrow }}\ x_{2}{\stackrel {v_{3}}{\longrightarrow }}\ x_{3}{\stackrel {v_{4}}{\longrightarrow }}\ X_{1}}$

where ${\displaystyle X_{o}}$ and ${\displaystyle X_{1}}$ are fixed species to ensure the system is open. The system equation can be written as: ^{[1] [6]}

${\displaystyle \mathbf {N} ={\begin{bmatrix}1&-1&{\phantom {+}}0&{\phantom {+}}0\\0&{\phantom {+}}1&-1&{\phantom {+}}0\\0&{\phantom {+}}0&{\phantom {+}}1&-1\\\end{bmatrix}},\ }$${\displaystyle \mathbf {v} ={\begin{bmatrix}v_{1}\\v_{2}\\v_{3}\\v_{4}\\\end{bmatrix}}}$

So that:

${\displaystyle {\begin{bmatrix}{\dfrac {dx_{1}}{dt}}\\[4pt]{\dfrac {dx_{2}}{dt}}\\[4pt]{\dfrac {dx_{3}}{dt}}\\[4pt]{\dfrac {dx_{4}}{dt}}\\[4pt]\end{bmatrix}}={\begin{bmatrix}1&-1&{\phantom {+}}0&{\phantom {+}}0\\0&{\phantom {+}}1&-1&{\phantom {+}}0\\0&{\phantom {+}}0&{\phantom {+}}1&-1\\\end{bmatrix}}}$${\displaystyle {\begin{bmatrix}v_{1}\\v_{2}\\v_{3}\\v_{4}\\\end{bmatrix}}}$

The elements of the rate vector will be rate equations that are functions of one or more species ${\displaystyle x_{i}}$ and parameters, p. In the example, these might be simple mass-action rate laws such as ${\displaystyle v_{2}=k_{2}x_{1}}$ where ${\displaystyle k_{2}}$ is the rate constant parameter. The particular laws chosen will depend on the specific system under study. Assuming mass-action kinetics, the above equation can be written in complete form as:

${\displaystyle {\begin{bmatrix}{\dfrac {dx_{1}}{dt}}\\[4pt]{\dfrac {dx_{2}}{dt}}\\[4pt]{\dfrac {dx_{3}}{dt}}\\[4pt]{\dfrac {dx_{4}}{dt}}\\[4pt]\end{bmatrix}}={\begin{bmatrix}1&-1&{\phantom {+}}0&{\phantom {+}}0\\0&{\phantom {+}}1&-1&{\phantom {+}}0\\0&{\phantom {+}}0&{\phantom {+}}1&-1\\\end{bmatrix}}}$${\displaystyle {\begin{bmatrix}k_{1}X_{o}\\k_{2}x_{1}\\k_{3}x_{2}\\k_{4}x_{3}\\\end{bmatrix}}}$

Analysis

The system equation can be analyzed by looking at the linear response of the equation around the steady-state with respect to the parameter ${\displaystyle {\bf {p}}}$. ^[7] At steady-state, the system equation is set to zero and given by:

${\displaystyle 0={\bf {N}}{\bf {v}}({\bf {x}}({\bf {p}}),{\bf {p}})}$

Differentiating the equation with respect to ${\displaystyle {\bf {p}}}$ and rearranging gives:

${\displaystyle {\dfrac {d{\bf {x}}}{d{\bf {p}}}}=-\left({\bf {N}}{\frac {\partial \mathbf {v} }{\partial \mathbf {x} }}\right)^{-1}{\bf {N}}{\frac {\partial \mathbf {v} }{\partial \mathbf {p} }}}$

This derivation assumes that the stoichiometry matrix has full rank. If this is not the case, then the inverse won't exist.

Example

For example, consider the same problem from the previous section of a linear chain. The matrix ${\displaystyle {\frac {\partial \mathbf {v} }{\partial \mathbf {x} }}}$ is the unscaled elasticity matrix:

${\displaystyle {\mathcal {E}}={\begin{bmatrix}{\dfrac {\partial v_{1}}{\partial x_{1}}}&\cdots &{\dfrac {\partial v_{1}}{\partial x_{m}}}\\\vdots &\ddots &\vdots \\{\dfrac {\partial v_{n}}{\partial x_{1}}}&\cdots &{\dfrac {\partial v_{n}}{\partial x_{m}}}\end{bmatrix}}.}$

In this specific problem there are 3 species (${\displaystyle m=3}$) and 4 reaction steps (${\displaystyle n=4}$), the elasticity matrix is therefore a ${\displaystyle m\times n=3\ {\mbox{by}}\ 4}$ matrix. However, a number of entries in the matrix will be zero. For example ${\displaystyle \partial v_{1}/\partial x_{3}}$ will be zero since ${\displaystyle x_{3}}$ has no effect on ${\displaystyle v_{1}}$. The matrix, therefore, will contain the following entries:

${\displaystyle {\mathcal {E}}={\begin{bmatrix}{\dfrac {\partial v_{1}}{\partial x_{1}}}&0&0\\{\dfrac {\partial v_{2}}{\partial x_{1}}}&{\dfrac {\partial v_{2}}{\partial x_{2}}}&0\\0&{\dfrac {\partial v_{3}}{\partial x_{2}}}&{\dfrac {\partial v_{3}}{\partial x_{3}}}\\0&0&{\dfrac {\partial v_{4}}{\partial x_{3}}}\\\end{bmatrix}}.}$

The parameter matrix depends on which parameters are considered. In Metabolic control analysis, a common set of parameters are the enzyme activities. For the sake of argument, we can equate the rate constants with the enzyme activity parameters. We also assume that each enzyme, ${\displaystyle k_{i}}$, only can affect its own step and no other. The matrix ${\displaystyle {\frac {\partial \mathbf {v} }{\partial \mathbf {p} }}}$ is the unscaled elasticity matrix with respect to the parameters. Since there are 4 reaction steps and 4 corresponding parameters, the matrix will be a 4 by 4 matrix. Since each parameter only affects one reaction, the matrix will be a diagonal matrix:

${\displaystyle {\mathcal {E}}={\begin{bmatrix}{\dfrac {\partial v_{1}}{\partial k_{1}}}&0&0&0\\0&{\dfrac {\partial v_{2}}{\partial k_{2}}}&0&0\\0&0&{\dfrac {\partial v_{3}}{\partial k_{3}}}&0\\0&0&&{\dfrac {\partial v_{4}}{\partial k_{4}}}\\\end{bmatrix}}.}$

Since there are 3 species and 4 reactions, the resulting matrix ${\displaystyle {\frac {d{\bf {x}}}{d{\bf {p}}}}}$ will be a 3 by 4 matrix

${\displaystyle D={\mathcal {E}}_{1}^{1}{\mathcal {E}}_{2}^{2}({\mathcal {E}}_{3}^{3}-{\mathcal {E}}_{3}^{4})+{\mathcal {E}}_{1}^{1}{\mathcal {E}}_{2}^{3}{\mathcal {E}}_{3}^{4}-{\mathcal {E}}_{2}^{1}{\mathcal {E}}_{2}^{3}{\mathcal {E}}_{3}^{4}}$

${\displaystyle {\vphantom {}}}$

${\displaystyle {\frac {d{\bf {x}}}{d{\bf {p}}}}={\frac {1}{D}}\left[{\begin{array}{ll}{\mathcal {E}}_{k_{1}}^{1}({\mathcal {E}}_{2}^{2}({\mathcal {E}}_{3}^{3}-{\mathcal {E}}_{3}^{4})+{\mathcal {E}}_{2}^{3}{\mathcal {E}}_{3}^{4})&-{\mathcal {E}}_{2}^{3}{\mathcal {E}}_{3}^{4}{\mathcal {E}}_{k_{2}}^{2}\\{\mathcal {E}}_{2}^{1}{\mathcal {E}}_{k_{1}}^{1}({\mathcal {E}}_{3}^{3}-{\mathcal {E}}_{3}^{4})&{\mathcal {E}}_{1}^{1}{\mathcal {E}}_{k_{2}}^{2}({\mathcal {E}}_{3}^{3}-{\mathcal {E}}_{3}^{4})\\{\mathcal {E}}_{2}^{1}{\mathcal {E}}_{2}^{3}{\mathcal {E}}_{k_{1}}^{1}&{\mathcal {E}}_{1}^{1}{\mathcal {E}}_{2}^{3}{\mathcal {E}}_{k_{2}}^{2}\\\end{array}}\right.}$

${\displaystyle \qquad \qquad \qquad \quad \left.{\begin{array}{ll}{\mathcal {E}}_{2}^{2}{\mathcal {E}}_{3}^{4}{\mathcal {E}}_{k_{3}}^{3}&{\mathcal {E}}_{2}^{2}{\mathcal {E}}_{3}^{3}{\mathcal {E}}_{k_{4}}^{4}\\{\mathcal {E}}_{3}^{4}{\mathcal {E}}_{k_{3}}^{3}({\mathcal {E}}_{1}^{1}-{\mathcal {E}}_{2}^{1})&{\mathcal {E}}_{3}^{3}{\mathcal {E}}_{k_{4}}^{4}({\mathcal {E}}_{2}^{1}-{\mathcal {E}}_{1}^{1})\\{\mathcal {E}}_{1}^{1}{\mathcal {E}}_{2}^{2}{\mathcal {E}}_{k_{3}}^{3}&-{\mathcal {E}}_{k_{4}}^{4}({\mathcal {E}}_{1}^{1}({\mathcal {E}}_{2}^{2}-{\mathcal {E}}_{2}^{3})+{\mathcal {E}}_{2}^{1}{\mathcal {E}}_{2}^{3})\\\end{array}}\right]}$

Each expression in the matrix describes how a given parameter influences the steady-state concentration of a given species. Note that this is the unscaled derivative. It is often the case that the derivative is scaled by the parameter and concentration to eliminate units as well as turn the measure into a relative change.

Assumptions

The biochemical systems equation makes two key assumptions:

1. Species exist in a well-stirred reactor, so there are no spatial gradients. ^{[8] [9] [10]}
2. Species concentrations are high enough so that stochastic effects are negligible ^{[11] [12] [13]}

See also

• Stoichiometry matrix
• Chemical reaction network theory
• List of systems biology modeling software

References

1. Reder, Christine (November 1988). "Metabolic control theory: A structural approach". Journal of Theoretical Biology. 135 (2): 175–201. doi:10.1016/S0022-5193(88)80073-0.
2. Hofmeyr, Jan-hendrik S. (2001). "Metabolic control analysis, in a nutshell,". In Proceedings of the 2 Nd International Conference on Systems Biology: 291–300. CiteSeerX   10.1.1.324.922 .
3. Stucki, Jörg W. (1979). "Stability analysis of biochemical systems— A practical guide". Progress in Biophysics and Molecular Biology. 33: 99–187. doi:10.1016/0079-6107(79)90027-0.
4. Fjeld, M.; Asbjørnsen, O.A.; Åström, K.J. (September 1974). "Reaction invariants and their importance in the analysis of eigenvectors, state observability and controllability of the continuous stirred tank reactor". Chemical Engineering Science. 29 (9): 1917–1926. doi:10.1016/0009-2509(74)85009-8.
5. Park, David J. M. (1 September 1975). "SMISS, stoichiometric matrix inversion for steady state metabolic networks". Computer Programs in Biomedicine. 5 (1): 46–60. doi:10.1016/0010-468X(75)90026-4.
6. Cornish-Bowden, Athel; Hofmeyr, Jan-Hendrik S. (May 2002). "The Role of Stoichiometric Analysis in Studies of Metabolism: An Example". Journal of Theoretical Biology. 216 (2): 179–191. doi:10.1006/jtbi.2002.2547.
7. Heinrich, Reinhart; Schuster, Stefan (1996). The Regulation of Cellular Systems. New York, NY: Springer. ISBN   0412032619.
8. Cowan, Ann E.; Moraru, Ion I.; Schaff, James C.; Slepchenko, Boris M.; Loew, Leslie M. (2012). "Spatial Modeling of Cell Signaling Networks". Methods in Cell Biology. 110: 195–221. doi:10.1016/B978-0-12-388403-9.00008-4. PMC   3519356 .
9. Fell, David A. (May 1980). "Theoretical analyses of the functioning of the high- and low-Km cyclic nucleotide phosphodiesterases in the regulation of the concentration of adenosine 3′,5′-cyclic monophosphate in animal cells". Journal of Theoretical Biology. 84 (2): 361–385. doi:10.1016/S0022-5193(80)80011-7.
10. Kholodenko, Boris N. (March 2006). "Cell-signalling dynamics in time and space". Nature Reviews Molecular Cell Biology. 7 (3): 165–176. doi:10.1038/nrm1838. PMC   1679905 .
11. Gillespie, Daniel T. (December 1977). "Exact stochastic simulation of coupled chemical reactions". The Journal of Physical Chemistry. 81 (25): 2340–2361. doi:10.1021/j100540a008.
12. Gillespie, Daniel T. (1 May 2007). "Stochastic Simulation of Chemical Kinetics". Annual Review of Physical Chemistry. 58 (1): 35–55. doi:10.1146/annurev.physchem.58.032806.104637.
13. Andrews, Steven S; Bray, Dennis (September 2004). "Stochastic simulation of chemical reactions with spatial resolution and single molecule detail". Physical Biology. 1 (3): 137–151. doi:10.1088/1478-3967/1/3/001.