1: \documentclass[preprint,showpacs]{revtex4}
2: \usepackage{graphicx}
3: \include{epsf}
4:
5: \begin{document}
6:
7: \title{
8: Genetic Toggle Switch Without Cooperative Binding
9: }
10: \author{Azi Lipshtat$^1$, Adiel Loinger$^2$,
11: Nathalie Q. Balaban$^2$ and Ofer Biham$^2$}
12: \affiliation{$^1$
13: Department of Pharmacology and Biological Chemistry,
14: Mount Sinai \\ School of Medicine,
15: New York, NY 10029, USA
16: \\
17: $^2$ Racah Institute of Physics,
18: The Hebrew University,
19: Jerusalem 91904, Israel
20: }
21:
22: \begin{abstract}
23:
24: Genetic switch systems with mutual repression of two
25: transcription factors are studied using deterministic and
26: stochastic methods.
27: Numerous studies have concluded
28: that cooperative binding is
29: a necessary condition for the emergence of bistability
30: in these systems.
31: Here we show that for a range of
32: biologically relevant conditions,
33: a suitable combination of network structure and
34: stochastic effects gives rise to bistability even
35: without cooperative binding.
36:
37: \end{abstract}
38:
39: \pacs{PACS: 87.10.+e, 87.16.-b}
40:
41: \maketitle
42:
43: Recent advances in quantitative measurements
44: of gene expression at the single-cell level
45: \cite{Elowitz2002,Ozbudak2002}
46: have brought new insight on the importance
47: of stochastic fluctuations
48: in genetic circuits
49: \cite{Mcadams1997}.
50: Populations of genetically identical cells show variability
51: due to fluctuations.
52: The role of fluctuations is enhanced due to
53: the discrete nature of the transcription factors
54: and their binding sites, which may appear in low copy numbers
55: \cite{Becskei2000,Kaern2005}.
56: Stochastic behavior
57: may invoke oscillations
58: \cite{Vilar2002,Zhou2005}
59: and spatio-temporal patterns
60: \cite{Shnerb2000,Howard2003,Togashi2004},
61: which are unaccounted for by
62: macroscopic chemical rate equations.
63: Genetic circuits with
64: feedback mechanisms often exhibit bistability,
65: namely, two distinct stable states which
66: can be switched either spontaneously or by an external
67: signal
68: \cite{Atkinson2003,Ozbudak2004,Francois2005}.
69: To qualify as a switch, the spontaneous switching rate
70: must be much lower than the rates of the relevant
71: processes in the cell, namely transcription, translation,
72: binding and unbinding of transcription factors.
73: In particular, genetic switches such as the
74: phage $\lambda$ switch,
75: enable cells to adopt different fates
76: \cite{Ptashne1992}.
77: The toggle switch is
78: a simple genetic circuit that
79: consists of two proteins, $A$ and $B$,
80: with concentrations
81: $[A]$ and $[B]$, respectively,
82: which negatively regulate each other's synthesis
83: (by concentration we mean the average copy number
84: of proteins per cell).
85: The production of protein $A$ is
86: negatively regulated by protein $B$,
87: through binding of $n$ copies of $B$
88: to the $A$ promoter (and vice versa).
89: This process can be
90: modeled by a Hill function, which
91: reduces the production rate of
92: $A$
93: by a factor of
94: $1+k[B]^n$,
95: where
96: $k$ is a parameter and
97: $n$ is the Hill coefficient
98: \cite{Hill}.
99: In case that $n=1$ the binding of a single protein is
100: sufficient in order to perform the negative regulation, while
101: for $n>1$ the {\it cooperative binding} of two or more
102: proteins is required.
103: In numerous studies of the toggle switch system
104: it was concluded that cooperative binding is
105: a necessary condition for the emergence of
106: the two distinct stable states
107: characteristic of a switch
108: \cite{Gardner2000,Cherry2000,Warren2004,Walczak2005}.
109: It was also observed that in presence of cooperative binding,
110: stochastic effects contribute to the broadening of the parameter
111: range in which bistability appears
112: \cite{Kepler2001}.
113:
114: In this letter we show
115: that stochastic effects enable bistability even without cooperative binding
116: of the transcription factors to the operator, namely for
117: Hill coefficient $n=1$.
118: Furthermore, bistability takes place even when
119: the active proteins appear in high copy numbers.
120: These results
121: emphasize the necessity of stochastic methods in the analysis of
122: genetic networks, even under conditions of high concentrations.
123:
124: The mutual repression circuit,
125: referred to as the general switch
126: \cite{Warren2004},
127: is described by
128: the rate equations
129: %
130: \begin{eqnarray}
131: \label{eq:A_dot}
132: [\dot{A}] &=& g_A (1-[r_B]) - d_A [A]-
133: \alpha_0 [A] \left (1-[r_A] \right) + \alpha_1[r_A] \nonumber \\
134: \label{Eq:B_dot}
135: [\dot{B}] &=& g_B(1-[r_A])-d_B[B]-
136: \alpha_0[B]\left(1-[r_B]\right)+\alpha_1[r_B] \nonumber \\
137: \label{eq:rA_dot}
138: [\dot{r_A}] &=& \alpha_0[A]\left(1-[r_A]\right)-\alpha_1[r_A] \nonumber \\
139: \label{eq:rB_dot}
140: [\dot{r_B}] &=& \alpha_0[B]\left(1-[r_B]\right)-\alpha_1[r_B],
141: \label{eq:rate}
142: \end{eqnarray}
143: %
144: \noindent
145: where $g_X$
146: (s$^{-1}$),
147: $X=A,B$, is the maximal production rate
148: of protein $X$
149: and
150: $d_X$ (s$^{-1}$)
151: is its degradation rate.
152: For simplicity, we ignore the
153: mRNA level and take the processes of transcription and translation as a
154: single step of synthesis
155: \cite{mRNA}.
156: The bound repressors are considered
157: as separate species
158: $r_X$
159: and their concentrations
160: are given by
161: $[r_X]$,
162: providing much insight into the
163: repression process
164: \cite{Lipshtat2005}.
165: Here,
166: $r_A$ is a bound $A$ protein
167: that monitors the production of $B$, while
168: $r_B$ is a bound $B$ protein
169: that monitors the production of $A$.
170: Since there is a single promoter of each type,
171: $0\le [r_X] \le 1$.
172: The parameter
173: $\alpha_0$ (s$^{-1}$)
174: is the binding rate of proteins to the promoter
175: and
176: $\alpha_1$ (s$^{-1}$)
177: is the dissociation rate.
178:
179: It is commonly assumed that the
180: binding-unbinding processes
181: are much faster than other
182: processes in the circuit, namely
183: $\alpha_0,\alpha_1 \gg d_X,g_X$.
184: This means that the relaxation times of $[r_X]$
185: are much shorter than other relaxation
186: times in the circuit.
187: Under this assumption,
188: one can take the
189: time derivatives of $[r_X]$
190: to zero, even if the system is away from
191: steady state.
192: This brings the rate equations to the
193: standard Michaelis-Menten form
194: %
195: \begin{eqnarray}
196: \label{eq:dA}
197: [\dot{A}] &=& {g_A / ({1+k[B]})} - d_A [A] \nonumber \\
198: \label{eq:dB}
199: [\dot{B}] &=& {g_B / ({1+k[A]})} - d_B [B],
200: \end{eqnarray}
201: %
202: \noindent
203: where
204: $k=\alpha_0/\alpha_1$
205: is the repression strength.
206: For a given population of free $X$ repressors,
207: the parameter $k$ controls the value of $[r_X]$.
208: The limit of weak repression,
209: $[r_X] \ll 1$,
210: is obtained when
211: $k[X] \ll 1$,
212: while the limit of strong repression,
213: $[r_X] \simeq 1$,
214: is obtained for
215: $k[X] \gg 1$.
216: These equations turn out to have one positive
217: steady-state solution, thus at the level of rate equations
218: this system does not exhibit bistability.
219: For symmetric parameters,
220: where
221: $g_A=g_B=g$
222: and
223: $d_A=d_B=d$,
224: this solution is
225: $[A] = [B] = [ (1+4kg/d)^{1/2} -1 ]/2k$.
226:
227: In order to
228: account for stochastic effects,
229: the master equation approach
230: \cite{Mcadams1997,Kepler2001,Paulsson2000}
231: is applied.
232: In the master equation, the dynamic variables are the
233: probabilities
234: $P(N_A,N_B,r_A,r_B)$
235: for a cell to
236: include
237: $N_X$ copies of free protein $X$
238: and
239: $r_X$ copies of the bound $X$ repressor,
240: where $N_X=0,1,2,\dots$,
241: and
242: $r_X = 0,1$.
243: The master equation for the mutual repression circuit
244: takes the form
245: %
246: \begin{eqnarray}
247: \label{eq:master}
248: &&\dot{P}(N_A,N_B,r_A,r_B) =
249: g_A \delta_{r_B,0} P(N_A-1,N_B,r_A,r_B) + g_B \delta_{r_A,0} P(N_A,N_B-1,r_A,r_B)
250: \nonumber \\
251: && + d_A (N_A+1) P(N_A+1,N_B,r_A,r_B) + d_B (N_B+1) P(N_A,N_B+1,r_A,r_B)
252: \nonumber \\
253: && - (g_A \delta_{r_B,0} + g_B \delta_{r_A,0}) P(N_A,N_B,r_A,r_B) - (d_A N_A + d_B N_B) P(N_A,N_B,r_A,r_B)
254: \nonumber \\
255: &&+ \alpha_0 [ (N_A+1) \delta_{r_A,1} P(N_A+1,N_B,0,r_B) + (N_B+1)\delta_{r_B,1} P(N_A,N_B+1,r_A,0)]
256: \nonumber \\
257: &&+ \alpha_1 [\delta_{r_A,0} P(N_A-1,N_B,1,r_B)
258: + \delta_{r_B,0} P(N_A,N_B-1,r_A,1)]
259: \nonumber \\
260: &&- \alpha_0 ( N_A \delta_{r_A,0} + N_B \delta_{r_B,0}) P(N_A,N_B,r_A,r_B)
261: -\alpha_1 (\delta_{r_A,1} + \delta_{r_B,1}) P(N_A,N_B,r_A,r_B),
262: \end{eqnarray}
263: %
264: \noindent
265: where
266: $\delta_{i,j}=1$ for $i=j$ and $0$ otherwise.
267: The $g_X$ terms account for the production of proteins.
268: The $d_X$ terms account for the degradation of free proteins,
269: while the $\alpha_0$ ($\alpha_1$) terms describe the
270: binding (unbinding) of proteins to (from) the promoter site.
271: The average copy numbers
272: $\langle X \rangle$,
273: where $X=N_A,N_B,r_A,r_B$,
274: are given by
275: $\langle X \rangle =\sum X P(N_A,N_B,r_A,r_B)$
276: where the sum is over all integer values of $N_A$
277: and $N_B$ up to a suitable cutoffs and over
278: $r_A,r_B=0,1$.
279: Note that for distributions that are skewed or
280: exhibit several peaks, the average does not reflect
281: the actual behavior in a single cell.
282:
283: To analyze the role of fluctuations in
284: this circuit we
285: have calculated the
286: probability distribution
287: $P(N_A,N_B) = \sum_{r_A,r_B} P(N_A,N_B,r_A,r_B)$.
288: We used the symmetric parameters
289: $g=0.05$ (s$^{-1}$),
290: which correspond to average production time of $20$ seconds,
291: and
292: $d=0.005$ (s$^{-1}$)
293: which means degradation
294: time of 200 seconds,
295: in agreement with experimental results
296: \cite{Elowitz2000}.
297: To examine a broad range of relevant values of $k$ we
298: performed two sets of simulations.
299: In the first set we chose
300: $\alpha_1=0.5$ (s$^{-1}$)
301: and varied
302: $\alpha_0$,
303: while in the second set we chose
304: $\alpha_0=0.5$ (s$^{-1}$)
305: and varied
306: $\alpha_1$.
307: We confirmed that the population of free
308: proteins depends only on the ratio, $k$.
309:
310: Under conditions in which the promoter sites are
311: empty most of the time, namely
312: $r_X \ll 1$, the repression is weak and
313: the steady state solution
314: exhibits coexistence of $A$ and $B$ proteins in the cell.
315: In this case the distribution
316: $P(N_A,N_B)$
317: exhibits a single peak
318: [Fig.~\ref{fig:1}(a)].
319: In this case, the values of
320: $\langle N_A \rangle$
321: and
322: $\langle N_B \rangle$
323: obtained from the master equation
324: coincide with $[A]$ and $[B]$, obtained from
325: the rate equations.
326: For strong repression,
327: the distribution $P(N_A,N_B)$
328: exhibits a peak
329: in which the $A$ population is
330: suppressed and a peak in which the $B$ population
331: is suppressed, as expected for a bistable system.
332: However, a third peak appears near the origin,
333: in which both populations of free proteins diminish
334: [Fig.~\ref{fig:1}(b)].
335: This peak represents a
336: dead-lock situation, caused by the
337: fact that both $A$ and $B$ repressors can be bound
338: simultaneously, each
339: bringing to a halt the production of the other specie.
340: This result is in contrast to the rate
341: equations which exhibit a single solution,
342: $[A]=[B]$, for the entire range of parameters.
343: Below, we present three biologically sensible variants
344: of the circuit in which the third peak is suppressed,
345: giving rise to a bistable switch.
346:
347: Consider the exclusive switch, where
348: there is an overlap between the promoters
349: of $A$ and $B$ and thus
350: no room for both to be occupied simultaneously.
351: Such a situation is encountered in nature,
352: for example, in the
353: lysis-lysogeny switch of phage $\lambda$
354: \cite{Ptashne1992}.
355: It was shown that
356: in presence of cooperative binding,
357: the exclusive switch is more stable than
358: the general switch
359: \cite{Warren2004}.
360: This is because in the exclusive switch the access of the
361: minority specie to the promoter site is blocked by the
362: dominant specie.
363: Here we show that in the exclusive switch,
364: stochastic effects give rise to bistability even
365: without cooperativity between the transcription factors.
366: To model this system recall that
367: $[r_A]$
368: ($[r_B]$)
369: can be defined as the fraction of
370: time in which the promoter
371: is occupied by a bound $A$ ($B$) protein.
372: The fraction of time in which the promoter is vacant is
373: $1 - [r_A] - [r_B]$.
374: Incorporating this into Eq.~(\ref{eq:rate})
375: gives rise to the following modification:
376: in the $\alpha_0$ terms,
377: each appearance of
378: $[r_A]$
379: or
380: $[r_B]$
381: should be replaced by
382: $[r_A]+[r_B]$.
383: For symmetric parameters,
384: the resulting equations still exhibit
385: a single solution,
386: in which
387: $[A]=[B]$
388: and
389: $[r_A]=[r_B]$.
390: The Michaelis-Menten
391: equations for the exclusive switch are given by
392: Eqs.~(\ref{eq:dB}),
393: where in the first equation
394: $k$ is replaced by
395: $k/(1+k[A])$
396: and in the second equation it is replaced by
397: $k/(1+k[B])$.
398: To account for the discreteness
399: of the transcription factors and their fluctuations,
400: the master equation
401: should be applied,
402: with the constraint that
403: $P(N_A,N_B,1,1)=0$.
404: It takes the form of
405: Eq.~(\ref{eq:master}),
406: except that
407: in the $\alpha_0$ and $\alpha_1$ terms,
408: each time
409: $\delta_{r_A,j}$
410: ($\delta_{r_B,j}$)
411: appears it should be multiplied by
412: $\delta_{r_B,0}$,
413: ($\delta_{r_A,0}$).
414: In the exclusive switch, under conditions of
415: weak repression,
416: $P(N_A,N_B)$
417: exhibits a single peak
418: [Fig.~\ref{fig:2}(a)],
419: for which
420: $\langle N_A \rangle$
421: and
422: $\langle N_B \rangle$
423: coincide with
424: $[A]$ and $[B]$, respectively.
425: For strong repression,
426: the distribution $P(N_A,N_B)$
427: exhibits two peaks.
428: In one peak the $A$ population is
429: suppressed, while in the
430: other peak the $B$ population
431: is suppressed,
432: as expected for a bistable system
433: [Fig.~\ref{fig:2}(b)].
434: The dead-lock situation is impossible in this system.
435:
436: To examine the time dependence of the populations of free
437: proteins in a single cell, we have performed Monte Carlo
438: simulations, based on the master equation for the exclusive
439: switch.
440: In Fig.~\ref{fig:3}
441: we present the copy numbers of free and bound
442: $A$ and $B$
443: proteins
444: vs. time. The population size of the dominant specie is in the
445: range of 20-60, while the minority specie is almost completely
446: suppressed. The typical switching time is around $10^{5}$ seconds.
447:
448: Consider a different variant of the genetic switch, which
449: exhibits bound-repressor degradation (BRD).
450: Even a low degradation rate, $d_r$,
451: of the bound repressors tends to remove the
452: mutual suppression of both species, and gives
453: rise to a binary switch.
454: The rate equations that describe this circuit are
455: identical to
456: Eq.~(\ref{eq:rate}),
457: except that
458: a degradation term of the form
459: $-d_r[r_A]$
460: ($-d_r[r_B]$)
461: is added to the equation for
462: $[\dot{r_A}]$
463: ($[\dot{r_B}]$).
464: For symmetric parameters,
465: the Michaelis-Menten form of these equations,
466: applicable in the limit of fast switching,
467: is given by
468: Eq.~(\ref{eq:dB})
469: where
470: $k=\alpha_0/(\alpha_1+d_r)$
471: and $d$ is replaced by an effective degradation rate
472: $d_{\rm eff}=d+ d_r k/(1+k[A])$
473: in the first equation
474: and by the analogous term in the second equation.
475: This equation exhibits a bifurcation
476: at $k_c=(d/d_r)(\sqrt{g}+\sqrt{d_r})/(\sqrt{g}-\sqrt{d_r})$,
477: in which the
478: symmetric solution
479: $[A]=[B]$
480: becomes unstable, giving rise to two stable solutions
481: in which one specie is dominant and the other is suppressed
482: (Fig.~\ref{fig:4}, inset).
483: We thus find that in case that bound repressors exhibit
484: degradation, bistability appears even at the level of
485: rate equations.
486: The emergence of bistability can be attributed to the
487: fact that the effective degradation rate for the minority
488: specie is larger than for the dominant specie,
489: enhancing the difference between the population sizes.
490: The master equation for this circuit is
491: obtained by adding the term
492: $d_r [ (\delta_{r_A,0}-\delta_{r_A,1}) P(N_A,N_B,1,r_B)
493: + (\delta_{r_B,0}-\delta_{r_B,1}) P(N_A,N_B,r_A,1)]$
494: to
495: Eq.~(\ref{eq:master}).
496: This term represents transitions of the cell from
497: $r_X=1$ ($X=A,B$) to $r_X=0$,
498: {\em without}
499: changing the number of free proteins.
500: The degradation of bound repressors gives rise to suppression
501: of the peak near the origin,
502: leading to the emergence of bistability.
503:
504: A third variant of the genetic switch exhibits
505: protein-protein interactions (PPI) such that
506: an A protein and a B protein may form an AB complex,
507: which is not active as a transcription factor.
508: This circuit
509: exhibits bistability within a range of parameters,
510: both for the rate equations and for the master equation.
511:
512: We have calculated the switching time using the master equation,
513: for an initial state that includes only free A proteins.
514: The distribution
515: $P(N_A,N_B)$ vs. time was calculated and
516: the function
517: $f(t)=P(N_A>N_B)-P(N_A<N_B)$
518: was found to decay exponentially according to
519: $f(t)=\exp(-t/\tau)$,
520: where $\tau$
521: is defined as the switching time.
522: In Fig.~\ref{fig:4} we present the switching time
523: $\tau$,
524: obtained from
525: the master equation vs. $k$ for the exclusive switch
526: ($\circ$)
527: and for the BRD switch
528: ($\times$).
529: We also examined the dependence of $\tau$ on the copy
530: number, $N$, of the dominant specie.
531: For the exclusive switch, we found that
532: when $d$ is varied, $\tau \sim N^2$, while in case
533: that $g$ is varied, $\tau \sim N$. This dependence
534: is weaker than found for the cooperative switch
535: \cite{Warren2004}.
536:
537: The results presented in this paper
538: (except for Fig.~\ref{fig:3})
539: were obtained by direct
540: integration of the master equation rather than by Monte
541: Carlo methods
542: \cite{Gillespie1977}.
543: Direct integration is much more efficient and provides more accurate
544: results, without the need to accumulate statistics.
545: Recent improvements in the methodology enable to use
546: direct integration for complex networks that involve large
547: numbers of active proteins
548: \cite{Lipshtat2004},
549: which will enable to go beyond elementary circuits into simulations
550: of complete networks.
551:
552: In contrast to previous knowledge that bistability requires cooperative
553: binding of transcription factors,
554: we have shown that
555: bistability is possible
556: without cooperative binding.
557: We have analyzed three
558: variants of the genetic toggle switch, that exhibit bistability
559: without cooperative binding.
560: The first circuit
561: is the exclusive switch, in which
562: the two promoter sites
563: cannot be occupied simultaneously.
564: The second circuit exhibits
565: degradation of bound repressors, while in the third circuit
566: free $A$ and $B$ proteins may form a complex which is not
567: active as a transcription factor.
568: Rate equations predict a single stable state
569: in the first circuit and bistability in the second and third
570: circuits.
571: However, the master equation predicts
572: bistability in all the three circuits.
573: These findings are not limited to cases in which proteins exist in
574: low numbers, but are due to the low copy number
575: of the promoter itself.
576: The results presented here are expected to
577: have significant implications on the understanding of
578: non-genetic
579: variability in cell populations,
580: and may shed new light on the way cells
581: differentiate despite uniform environmental conditions.
582:
583: N.Q.B. was supported by the Center of Complexity of the Horowitz
584: Foundation and the Bikura Program of the Israel Science Foundation.
585:
586: \begin{thebibliography}{10}
587:
588: \bibitem{Elowitz2002}
589: {M.B. Elowitz, A.J. Levine, E.D. Siggia and P.S. Swain},
590: Science {\bf 297}, 1183 (2002).
591:
592: \bibitem{Ozbudak2002}
593: {E.M. Ozbudak, M. Thattai, I. Kurtser, A.D. Grossman and A. van Oudenaarden},
594: Nature Genetics {\bf 31}, 69 (2002).
595:
596: \bibitem{Mcadams1997}
597: {H.H. McAdams and A. Arkin},
598: Proc. Natl. Acad. Sci. US {\bf 94}, 814 (1997);
599: A. Arkin, J. Ross and H.H. McAdams,
600: Genetics {\bf 149}, 1633 (1998).
601:
602: \bibitem{Becskei2000}
603: A. Becskei and L. Serrano,
604: Nature {\bf 405}, 590 (2000).
605:
606: \bibitem{Kaern2005}
607: M. Kaern, T.C. Elston, W.J. Blake and J.J. Collins,
608: Nature Reviews Genetics {\bf 6}, 451 (2005).
609:
610: \bibitem{Vilar2002}
611: {J.M. Vilar, H.Y. Kueh, N. Barkai and S. Leibler}, Proc. Natl. Acad. Sci. US
612: {\bf 99}, 5988 (2002).
613:
614: \bibitem{Zhou2005}
615: T. Zhou, L. Chen and K. Aihara,
616: Phys. Rev. Lett. {\bf 95}, 178103 (2005).
617:
618: \bibitem{Shnerb2000}
619: {N.M. Shnerb, Y. Louzoun, E. Bettelheim and S. Solomon}, Proc. Natl. Acad. Sci.
620: US {\bf 97}, 10322 (2000).
621:
622: \bibitem{Howard2003}
623: {M. Howard and A.D. Rutenberg}, Phys. Rev. Lett. {\bf 90}, 128102 (2003).
624:
625: \bibitem{Togashi2004}
626: {Y. Togashi and K. Kaneko}, Phys. Rev. E {\bf 70}, 020901 (2004).
627:
628: \bibitem{Atkinson2003}
629: M.R. Atkinson, M.A. Savageau, J.T. Myers and A.J. Ninfa,
630: Cell {\bf 113}, 597 (2003).
631:
632: \bibitem{Ozbudak2004}
633: E.M. Ozbudak, M. Thattai, H.N. Lim, B.I. Shraiman, A. van Oudenaarden,
634: Nature {\bf 427}, 737 (2004).
635:
636: \bibitem{Francois2005}
637: P. Francois and V. Hakim,
638: Phys. Rev. E {\bf 72}, 031908 (2005).
639:
640: \bibitem{Ptashne1992}
641: {M. Ptashne}, {\em {A Genetic Switch: Phage $\lambda$ and Higher Organisms, 2nd
642: edition.}} ({Cell Press and Blackwell Scientific Publications}, {Cambridge,
643: MA}, 1992).
644:
645: \bibitem{Hill}
646: Hill-function models are simplifications of rate-law equations.
647: When derived directly from rate laws, $n$ is expected to take only
648: integer values. However, when these models are used for fitting
649: empirical data, $n$ is a fitting parameter which may take
650: non-integer values.
651:
652: \bibitem{Gardner2000}
653: {T.S. Gardner, C.R. Cantor and J.J. Collins}, Nature {\bf 403}, 339 (2000).
654:
655: \bibitem{Cherry2000}
656: {J.L. Cherry and F.R. Adler}, J. Theor. Biol. {\bf 203}, 117 (2000).
657:
658: \bibitem{Warren2004}
659: {P.B. Warren and P.R. ten Wolde}, Phys. Rev. Lett. {\bf 92}, 128101 (2004);
660: J. Phys. Chem. B {\bf 109}, 6812 (2005).
661:
662: \bibitem{Walczak2005}
663: A.M. Walczak, M. Sasai and P. Wlynes,
664: Biophysical J. {88}, 828 (2005).
665:
666: \bibitem{Kepler2001}
667: {T.B. Kepler and T.C. Elston}, Biophysical Journal {\bf 81}, 3116 (2001).
668:
669: \bibitem{mRNA}
670: Extended circuits that include the mRNA level have also been
671: studied. It was found that
672: there is
673: a perfect agreement between the rate equation
674: results for the simplified and
675: the extended circuits,
676: and excellent agreement with only slight difference in peak
677: shapes in the master equation.
678:
679: \bibitem{Lipshtat2005}
680: {A. Lipshtat, H.B. Perets, N. Q. Balaban and O. Biham}, Gene {\bf 347}, 265
681: (2005).
682:
683: \bibitem{Paulsson2000}
684: {J. Paulsson and M. Ehrenberg}, Phys. Rev. Lett. {\bf 84}, 5447 (2000);
685: {J. Paulsson}, Nature {\bf 427}, 415 (2004).
686:
687: \bibitem{Elowitz2000}
688: {M.B. Elowitz and S. Leibler}, Nature {\bf 403}, 335 (2000).
689:
690: \bibitem{Gillespie1977}
691: {D.T. Gillespie}, J. Chem. Phys. {\bf 81}, 2340 (1977).
692:
693: \bibitem{Lipshtat2004}
694: {A. Lipshtat and O. Biham}, Phys. Rev. Lett. {\bf 93}, 170601 (2004).
695:
696: \end{thebibliography}
697:
698: \clearpage
699:
700: \begin{figure}
701: \caption{
702: The probabilities
703: $P(N_A,N_B)$
704: for the general switch,
705: under conditions of
706: (a) weak repression ($k=0.005$)
707: where there is one symmetric peak
708: and
709: (b) strong repression ($k=50$)
710: where three peaks appear, one dominated by $A$, the
711: second dominated by $B$ and the third in which
712: both species are mutually suppressed.
713: The weights of the three peaks are about the same.
714: }
715: \label{fig:1}
716: \end{figure}
717:
718: \begin{figure}
719: \caption{
720: The probabilities
721: $P(N_A,N_B)$
722: for the exclusive switch,
723: under conditions of
724: (a) weak repression ($k=0.005$)
725: where there is one symmetric peak
726: and
727: (b) strong repression ($k=50$)
728: where bistability is observed.
729: }
730: \label{fig:2}
731: \end{figure}
732:
733: \begin{figure}
734: \caption{
735: The populations of free and bound
736: $A$ and $B$ proteins vs. time, obtained
737: from Monte Carlo simulations of the exclusive switch
738: with the parameters
739: $g=0.2$, $d=0.005$,
740: $\alpha_0=0.2$
741: and
742: $\alpha_1=0.01$.
743: The bistable behavior is clearly observed, where the population size
744: of the dominant specie is between 20-60 and the other specie is nearly
745: diminished. Failed switching attempts are clearly seen.
746: }
747: \label{fig:3}
748: \end{figure}
749:
750: \begin{figure}
751: \caption{
752: The switching time vs. the repression strength, $k$, for the
753: exclusive switch ($\circ$) and for the case in which bound
754: repressors exhibit degradation ($\times$). For the bistable range
755: (roughly $k>1$)
756: the switching time increases as $k$ is increased.
757: The inset shows the steady state solution
758: for $[A]$ and $[B]$ vs. $k$,
759: obtained from the rate equations for the BRD switch.
760: Note that for the BRD switch, the parameter $\alpha_0$
761: varies, while $d_r=d$ and $\alpha_1=0.01$ are held fixed.
762: }
763: \label{fig:4}
764: \end{figure}
765:
766: \end{document}
767:
768:
769:
770: