Abstract
We present the theory of a microfluidic bioreactor with a two-compartment growth chamber and periodic serial dilution. In the model, coexisting planktonic and biofilm populations exchange by adsorption and detachment. The criteria for coexistence and global extinction are determined by stability analysis of the global extinction state. Stability analysis yields the operating diagram in terms of the dilution and removal ratios, constrained by the plumbing action of the bioreactor. The special case of equal uptake function and logistic growth is analytically solved and explicit growth curves are plotted. The presented theory is applicable to generic microfluidic bioreactors with discrete growth chambers and periodic dilution at discrete time points. Therefore, the theory is expected to assist the design of microfluidic devices for investigating microbial competition and microbial biofilm growth under serial dilution conditions.
Similar content being viewed by others
References
Allen LJS (2007) Introduction to mathematical biology. Prentice Hall, Upper Saddle River
Balagadde FK, You LC, Hansen CL, Arnold FH, Quake SR (2005) Long-term monitoring of bacteria undergoing programmed population control in a microchemostat. Science 309:137–140
Balagadde FK, Song H, Ozaki J, Collins CH, Barnet M, Arnold FH, Quake SR, You L (2008) A synthetic Escherichia coli predator–prey ecosystem. Mol Syst Biol 4:187
Costerton JW, Stewart PS, Greenberg EP (1999) Bacterial biofilms: a common cause of persistent infections. Science 284:1318
Freedman HI, So JWH (1989) Persistence in discrete semi-dynamical systems. SIAM J Math Anal 20:930–938
Ghannoum MA, O’Toole GA (eds) (2004) Microbial biofilm. ASM Press, Washington Toole (2004)
Gu GY, Lee YW, Chiang CC, Yang YT A nanoliter microfluidic serial dilution bioreactor (preprint)
Hale J (1969) Ordinary differential equations. Wiley-Interscience, Hoboken
Hegab HM, ElMekawy A, Stakeborg T (2013) Review of microfluidic microbioreactor technology for high-throughput submerged microbiological cultivation. Biomicrofluidics 7:021502
Hsu SB, Smith HL, Waltman P (1996) Competitive exclusion and coexistence for competitive system on ordered Banach space. AMS Trans 348:4083–4094
Hsu SB (2013) Ordinary differential equations with applications, 2nd edn. World Scientific, Singapore
Klapper I, Dockery J (2010) Mathematical description of microbial biofilms. SIAM Rev 52:221
Melin J, Quake SR (2007) Microfluidic large-scale integration: the evolution of design rules for biological automation. Annu Rev Biophys Biomol Struct 36:213
Novick A, Szilard L (1950) Description of the chemostat. Science 112:715
Pylyugin SS, Waltman P (1999) The simple chemostat with wall growth. SIAM J Appl Math 5:1552–1572
Smith HL (1995) Monotone dynamical systems. An introduction to the theory of competitive and cooperative systems. American Mathematical Society, Washington
Smith HL (2011) Bacteria competition in serial transfer culture. Math Biosci 229:149
Smith HL, Waltman P (1995) The theory of the chemostat. Cambridge University Press, Cambridge
Stewart FM, Levin BR (1973) Partitioning of resources and outcome of inter-specific competition: a model and some general considerations. Am Nat 107:171
Toprak E, Veres A, Mitchel JB, Chait R, Hartl DL, Kishony R (2012) Evolutionary paths to antibiotic resistance under dynamically sustained drug selection. Nat Genet 44:101–106
Zhao XQ (2003) Dynamical systems in population biology. Canadian Mathematical Society, Springer Verlag, Berlin
Acknowledgments
The authors would like to thank W. Jager for useful discussions and suggestions. Y. T. Y. would like to acknowledge the funding support from the Ministry of Science and Technology under the grant number MOST 103-2220-E-007-026 and 104-2220-E-007-011 and from the National Tsing Hua University under the Grant Numbers 102N204E1 and 103N204E1.
Author information
Authors and Affiliations
Corresponding author
Appendices
Appendix A: Stability analysis
We can rewrite the initial value problem in Eq. (2.1) in vector form
where the vector X and F(X) are defined as
Then we have
Differentiating Eq. (5.2) with respect to \(x_{0}\in {\mathbf {R}}^{3}\) yields
Setting \(x_{0}=(\tilde{S},0,0)\) in Eq. (5.3), then we obtain
where \(V(t)=D_{x_{0}}\Phi _{t}(x_{0})\bigg |_{x_{0}=(\tilde{S},0,0)}\) and
Substituting \((S,u,w)=\Phi _{t}(\tilde{S},0,0)\equiv (\tilde{S},0,0)\) we can obtain
To prove the local stability of the extinction fixed point \((\tilde{S},0,0)\), we need to show that the spectral radius of \(D_{x_{0}}P(\tilde{S},0,0)\) is less than 1. From Eq. (3.4),we have
From (5.6), the eigenvalues of the matrix A are \(0, \lambda _1\), and \(\lambda _2\) . Here \(\lambda _1,\lambda _2\) are the eigenvalues of \(2\times 2\) matrix \(A_1\) given by
Then \(\lambda _1,\lambda _2\) are the roots of characteristic polynomial of \(A_1\)
The discriminant \(\vartheta \) of \(g({\lambda })\) is
Thus \(\lambda _1,\lambda _2\) are real value and given by
To evaluate the spectral radius \(r(D_{x_0}P(\tilde{S},0,0))\) from (5.7), we need to compute the matrix \(V(t)=[V_1(t)\,\,\,V_2(t)\,\,\,V_3(t)]\).Then the dynamics of the matrix is described by
where \(e_i\) are the basis vectors given by
It is easy to show that V(t) take the following form
Then
Note that from Eqs. (5.11) and (5.12), the matrix \(\hat{V}(t)\) satisfies
Let
The matrix B have eigenvalues \(\mu _1,\mu _2\).
Then the eigenvalues of \(D_{x_{0}}P(\tilde{S},0,0)\) are \(\eta \), \(\mu _1\) and \(\mu _2\). Since \(0<\eta <1\), the spectral radius \(r(D_{x_{0}}P(\tilde{S},0,0))<1\) if and only if \(|\mu _i|<1,\,\,i=1,2\).
From (Allen 2007) \(|\mu _i|<1,\,\,i=1,2\) if and only if
From (5.10) (5.14) and (5.15) and Liouvilles’ formula (Hsu 2013; Hale 1969), we can calculate the determinant det B as
To evaluate trace(B), we need compute \(V_{11}(T)\) and \(V_{22}(T)\). Let \(\tilde{V}_i\) be an eigenvector of the eigenvalue \(\lambda _i\) of matrix \(A_1\), \(i=1,2\).
It is easy to show that
are eigenvectors of \({\lambda }_1\) and \({\lambda }_2\) respectively.
Since \({\lambda }_1\ne {\lambda }_2\) then from theory of linear system (Hsu 2013; Hale 1969),
we obtain
From (5.10), we note that
or
From (5.18),(5.19), (5.20), and (5.21), we can obtain
where
Using (5.16), the stability condition for the extinction fixed point \(E_0=(\tilde{S},0,0)\) is given by
Appendix B: Proof of Lemma A
Lemma 1
Define \((S_n,u_n,w_n)=P^n(S_0,u_0,w_0)\). For any \((S_0,u_0,w_0)\in {\mathbb {R}}^3\), the sequence \({(S_n,u_n,w_n)}^{n=\infty }_{n=0}\) is bounded.
Proof
From the first equation in (3.1), we have \(S'(t)\le 0, 0<t<T\), and \(S(T)\le S_0\) since \(S_1=\eta S(T)+FS^{(0)}\le \eta S_0+FS^{(0)}\). Inductively we have \(S_{n+1}=\eta S_n+FS^{(0)}\), n = 0,1,2......, then
Given \(\epsilon >0\) small for n large
Let \(U(t)=\frac{1}{\gamma }(u(t)+\delta w(t))\) and \(U_n=\frac{1}{\gamma }(u_n+\delta w_n)\), \(n\ge 0\).
Then \(S(t)+U(t)=S_0+U_0\) for \(0\le t\le T\) and
Hence \(U(t)\le S_0+U_0\), \(0\le t\le T\).
Let \(\theta ^*=max\{\eta ,\theta _w\}\), then
Inductively for \(n\ge 1\), we have
For large n, we have
Hence \({(S_n,u_n,w_n)}^{n=\infty }_{n=0}\) are bounded.
Proof of Theorem A, part(i)
Consider the following auxiliary system of (3.1)
\(S(0)=S_0,\,\,u(0)=u_0,\,\,w(0)=w_0\) with with the same resetting of initial condition (3.1) at \(t=T^{+}\).
Define map \(\hat{P}(\hat{S}_0,\hat{u}_0,\hat{w}_0)=L \circ \psi _{T}(\hat{S}_0,\hat{u}_0,\hat{w}_0)\), with \((\hat{S}_0,\hat{u}_0,\hat{w}_0)=(S_{0},u_{0},w_{0})\), L defined in (4.3) and \(\psi _t(\hat{x}_0)\) is the semi-flow defined by the system Eq. (5.25), \(0\le t\le T\), \(\hat{x}_0=(\hat{S}_0,\hat{u}_0,\hat{w}_0)\). We write initial value problem (5.25) in vector form
Then \(F(X)\le \hat{F}(X)\). We note that (5.3) is a cooperationsystem for \(0\le t\le T\).
Let \((\hat{S}_n,\hat{u}_n,\hat{w}_n)=\hat{P}^n (\hat{S}_0,\hat{u}_0,\hat{w}_0)\).
By Kamke theorem (Smith 1995) and the resetting mechanism (3.1), it follows that
We want to show that if (3.6) holds, i.e., the extinction state \((S^{(0)},0,0)\) is locally stable for the map P, then \((\hat{u}_n,\hat{w}_n)\rightarrow (0,0)\) as \(n\rightarrow \infty \).
From the proof of Lemma A1, it follows that
For \(\epsilon >0\) small,
Let
Then
By Kamke’s Theorem and resetting mechanism, it follows that
Since \(r(D_{x_0}\hat{P}(S^{(0)},0,0))=r(D_{x_0}P(S^{(0)},0,0))\), then \(r(D_{x_0}\hat{P}(S^{(0)},0,0))<1\) implies \(r(D_{x_0} \hat{P}^{\pm }(S^{(0)} \pm \epsilon ,0,0) <1\) for \(\epsilon >0\), \((u^{\pm }_n,w^{\pm }_n)\rightarrow (0,0)\) as \(n\rightarrow \infty \). Hence if the stability condition (3.6) holds then \((u_n,w_n)\rightarrow (0,0)\) as \(n\rightarrow \infty \).
Proof of Theorem A, part(ii)
We shall prove uniformly persistence of the map P if \(r(D_{x_0}P(S^{(0)},0,0))>1\) i.e., Eq. (3.6) does not hold with strict inequality. Since \(E_0=(S^{(0)},0,0)\) is the only fixed point on the boundary of \(Int({\mathbb {R}}^3_{+})\), from Theorem 1.3.1 (Zhao 2003; Freedman and So 1989) and Lemma A1, it suffices to show that
where \(W^s(E_0)=\{x_0\in {\mathbb {R}}^3_{+}:P^n(x_0)\rightarrow E_0\,\,\,as\,\,\,n\rightarrow \infty \}\) is the stable set of fixed point \(E_0\).
We prove by contradiction. If this is not the case, there exists \(x_0=(S_0,u_0,w_0)\in Int({\mathbb {R}}^3_{+})\) such that \(P^n(x_0)\rightarrow E_0\) as \(n\rightarrow \infty \).
Then we have \(S_n\rightarrow \tilde{S},\,\,u_n\rightarrow 0\,\,w_n\rightarrow 0\) as \(n\rightarrow \infty \). From Eqs. (5.1) and (5.2) and stability condition \(r(D_{x_0}P(S^{(0)},0,0))>1\) , it follows that \((u^{-}_n,w^{-}_n)\rightarrow \infty \) as \(n\rightarrow \infty \), a contradiction. From Theorem 1.3.7 (Zhao 2003), there exists a positive fixed point \(E=(S^*,u^*,w^*)\) of the map P.
Proof of Theorem B
First we claim that under the assumptions (3.15) and (3.16) we have
From (2.1) it follows that
By the assumption (3.15) we have
From above the set \(W=\{(S, u, w):\gamma S+u+\delta w=\gamma S^{(0)}\}\) is positively invariant under the map P. Furthermore, from (5.28) inductively we have
Let \(U=u+\delta w\). Under the assumption (3.16) we have
From (5.27) we consider the following limiting system
from [4] let \(R_1=\frac{f(S^{(0)})T}{ln(\frac{1}{\eta })}\), it follows that if \(R_1<1\), then \(U(T_n^+)\rightarrow 0\); if \(R_1>1\), then there exists a unique fixed point \(\hat{U}\) of the system (5.29) such that
For the case \(R_1>1\) equivalently \(\eta >\eta ^*=e^{-f(S^{(0)})T}\), we consider the limiting equation of the second equation of (2.1)
The above equation can be written as
It is easy to show that
Since \(\eta >e^{-f(S^{(0)})T}\), it is easy to check that \(\eta e^{AT}<1\) holds.
Hence
We note that from chapter 1 (Zhao 2003) it is easy to lift the limiting systems (5.29) and (A.30) to the original system (2.1). Thus we complete the proof.
Rights and permissions
About this article
Cite this article
Hsu, SB., Yang, YT. Theory of a microfluidic serial dilution bioreactor for growth of planktonic and biofilm populations. J. Math. Biol. 72, 1401–1427 (2016). https://doi.org/10.1007/s00285-015-0913-0
Received:
Revised:
Published:
Issue Date:
DOI: https://doi.org/10.1007/s00285-015-0913-0