CEXS-UPF

Grau dEnginyeria Biom`

edica

BIOCOMPUTING
PROBLEMS

1 Transcriptional repression. A promoter is repressed by a transcription factor with concentration R, with Kd being the dissociation constant of the binding reaction between the repressor
and the promoter, and m being the maximal rate of transcription from the promoter.
(a) Calculate the activity of the promoter (input function), assuming that the repressor binds
to the promoter only as a homomultimer with n subunits.
(b) Consider now that an inductor with concentration S binds the repressor molecule reversible
with dissociation constant Ks . Assuming that the total concentration of repressor is RT , and
that the cooperativity of its binding to the promoter is n = 1, calculate the corresponding
input function.
2 Transcriptional cascade. Consider a cascade of three proteins, in which protein X, upon
activation (by phosphorylation, for instance), leads to transcription of protein Y , which in turn
activates transcription of protein Z. Assume that the input functions of Y and Z are Hill
functions with coefficients of activation Ky and Kz , respectively, maximal transcription rates y
and z , and Hill coefficients much larger than 1 in the two cases (so that we can assume step
functions). The decay rates of Y and Z are y and z , respectively. At t = 0 protein X becomes
phosphorylated very quickly, at a level larger than Ky (so that protein Y starts to be transcribed
immediately). Calculate the response time of protein Z, defined as the time at which the level
of Z reaches half its steady state level. Compare with the result with the response time for the
case of constitutive expression.
3 Positive feedback. Calculate the characteristic response time of a protein (i.e. the time needed
for the protein to reach half its steady-state level starting from 0 concentration), in the case in
which the protein activates its own expression (with a certain leak):
pn
dp
=+ n
p
dt
Kd + pn
To that end, assume that the protein regulates its own transcription in a non-cooperative manner
(n = 1), that the system operates far from saturation (in the linear regime, p  Kd ), and that
> /Kd . Describe briefly what would happen if < /Kd (dont do the calculation in this
case).
4 Hysteretic activation. Consider a protein that activates its own transcription in a positive
feedback loop, while its promoter has a certain level of basal expression:
dp
pn
=+ n
p
dt
K + pn

where is the basal transcription rate, is the maximal transcription rate, K is the activation
coefficient, and is the decay rate of the protein. Assuming that n is large:
(a) Decomposing appropriately the right-hand-side of the equation above, plot the phase line
of the system and indicate the number and type (stable/unstable) of equilibrium points for
three different values of , such that (i) K > , (ii) K = /2, and (iii) K < 0.
(b) Starting from the value of used in case (i) above, represent all the equilibrium points of
the system as a function of increasing (this plot is known as a bifurcation diagram).
(c) Discuss what happens as grows and decreases, and show that such a pulsed stimulation
leads to hysteresis (i.e. the system remembers its recent history).
5 Irreversible expression. Consider a gene that activates its own expression in a cooperative
manner:
pn
dp
p
=S+ n
dt
K + pn
S is represents an external factor that controls the basal expression level, K is the expression
threshold for the positive feedback to kick in, is the decay rate, and we assume a very high
cooperativity. We will consider the case K = /2.
(a) Assuming S = 0, determine the two stable equilibrium points of the system, and determine
in which of the two states the cell will end if the initial concentration of the protein is (i)
K/2 and (ii) 3K/2. Link the answer with the phenomenon of hysteresis.
(b) Assume now that the cell starts with a zero initial concentration of the protein and with
no signal, S = 0. If S starts to increase, calculate for what value of S a first bifurcation
occurs. What type of bifurcation is it?
(c) If S reaches a value twice the one obtained in the previous question, and next goes back
down to zero, describe the response of the system (in terms of the value of p at the end of
the process). Is this process irreversible? If so, what could we change to make it reversible
(or vice versa)?
6 Irreversible stimulation. Consider the phosphorylation/dephosphorylation cycle shown in the
figure below, in which phosphorylation is induced in two different ways: by an stimulus signal
S and by the phosphorylated protein itself, in the form of a positive feedback loop. Assuming
that the latter process is cooperative, the model of this system can be written as:
Anp
dAp
= kp SA + n
kd Ap
dt
K + Anp
where A is the concentration of the unphosphorylated protein and Ap is
the concentration of the phosphorylated protein. We will assume that
the total protein concentration, AT = A + Ap , is constant.
In the model, kp is the activation (phosphorylation) rate and kd is the inactivation (dephosphorylation) rate. We will consider n large, K = AT /2, and = kd AT .

(a) Decomposing appropriately the right-hand-side of the equation above, plot the phase line
of the system and indicate the number and type (stable/unstable) of equilibrium points in
the absence of stimulus (S = 0).
(b) Repeat the graph above for increasing values of S, and taking into account the observed
behavior of the equilibrium points, plot the bifurcation diagram of the system for increasing
stimulus S. Show that the system is irreversible, in the sense that if the cell starts with
no phosphorylated protein and no stimulus, as the stimulus increases the protein activates,
but decreasing then S back to zero does not lead to deactivation of the protein.
(c) Repeat the analysis made in the two sections above, but this time for a value of kd much
larger than /AT , and discuss what happens with the irreversibility in this case.
7 Mutual inhibition and decision making. In the infection process of E. coli by the bacteriophage (bacterial virus) , the virus is faced with the decision of remaining dormant inside the
cell (lysogeny) or multiplying and spreading to other bacteria after inducing lysis of its cell host.
This decision is mediated by a switch based on two genes, the repressor gene cI and the lytic
gene cro. The transcription of cro is repressed by cI and, according to the standard view, the
transcription of cI is repressed by Cro. Assuming that repression is cooperative in the two cases,
we can write down the following model:
x
dx
=
x x
dt
1 + (y/Ky )n
y
dy
=
y y ,
dt
1 + (x/Kx )m
where x represents the concentration of cI, and y the concentration of Cro.
(a) Assuming that x /x > Kx , y /y > Ky , and that the decay rates are similar for the two
proteins, plot the nulclines and vector field of the system. Identify the fixed points and
their stability (on the basis of the vector fields around them) and analyze the behavior of
the system depending on the initial values of the cI and Cro concentrations.
(b) Discuss, on the basis of the results above, how the behavior of the system changes when
the maximal expression rates of the two proteins vary, in two cases: x /x < Kx and
y /y < Ky .
8 Mutual activation. Two proteins A and B mutually activate their own expression, in such a
way that their dynamics is described by the following coupled differential equations:
dA
B n
= n
A,
dt
k + Bn

dB
An
= n
B ,
dt
k + An

where is the maximum expression, k the activation threshold, and the decay rate. All three
parameters are assumed to be the same for the two proteins.
(a) Assuming that n  1, represent graphically the nulclines of the system in the plane (A,B),
for / > k. Determine, both graphically and algebraically, the equilibrium points.

(b) Determine graphically the stability of all the equilibria, drawing the slope field of the system
in the different regions in which the nulclines divide the phase space. What is the biological
meaning of this behavior?
(c) Show (also graphically) that if k decreases a bifurcation occurs. What is the type of this
bifurcation?
9 Ultrasensitivity amplification in a transcriptional cascade. Consider a cascade of three
proteins, in which protein X activates transcription of protein Y , which in turn activates transcription of protein Z. Assume that the input functions of Y and Z are Hill functions with
coefficients of activation Ky and Kz , respectively, maximal transcription rates y and z , and
Hill coefficients n and m. The decay rates of Y and Z are y and z , respectively.
(a) Assuming that the activation of Y by X is ultrasensitive (n > 1) while the second is not
(m = 1), determine whether the total cascade is ultrasensitive and, if it is, calculate the
corresponding Hill coefficient.
(b) Consider that the activation threshold Kz of the second activation is small, so that a small
amount of Y is enough to activate Z. In that situation, demonstrate that the effective
ultrasensitivity of the full cascade (measured by the Hill coefficient of the corresponding
effective input function) is equal to the product of the ultrasensitivities of the individual
reactions.
10 Ultrasensitivity by molecular titration. Two proteins A and B form a complex with
binding rate kon . In turn, the complex dissociates into its two constituent proteins with rate
koff . We assume that the proteins are stable and that the cell does not divide (so that the total
concentrations of the two proteins, which we will call AT and BT , can be considered constant in
time), and that the dissociation constant kd = koff /kon is much smaller that BT (which occurs
when the binding affinity between A and B is very high).
(a) Calculate the equilibrium concentration of free A as a function of AT , BT , and kd .
(b) Calculate the sensitivity coefficient of the response of free A to AT for AT = BT , assuming
again that kd  BT . Apply the result to the case kd = 1 nM, BT = 104 nM. To interpret this
result, consider a Hill-function response with Hill coefficient n, and calculate its sensitivity
at threshold (i.e. when the input I is equal to the half-maximal activation K of the Hill
function). Determine the value of n for which the sensitivity of the Hill function would be
similar to that of the binding reaction studied above.
11 Bandpass regulation. A protein acts on its own promoter, activating its expression above a
certain threshold ka , and repressing it above a second threshold kr . The dynamics of the protein
concentration thus follows the equation:
dA
An
1
 m A
= n
n
dt
ka + A 1 + A
kr

Consider that ka < kr , and that n, m  1.

 In these conditions, and for > ka , plot the phase line of the system and determine
graphically the number and types of equilibrium points.
Repeat the study for decreasing values of , and show that as decreases a bifurcation
occurs. What type of bifurcation is it, and for what approximate value of does it occur?
12 Turn-off delay in a coherent feedforward loop. The
flagella motor in E. coli is built from the expression of a fli
promoter that is activated directly by the environmental sensor FlhDC, which is turned on by signals such as glucose
limitation and osmotic pressure, and by FliA, which is triggered when the first motors are completed. FlhDC, in turn,
also activates FliA, thus forming a coherent feedforward loop
(FFL, see figure).

FlhDC
FliA

Sx
Sy

fli

Notably, activation of the fli operon requires only one of its two inputs to be present (i.e. the
integration is similar to an OR gate, instead of an AND gate). Let x , y and z be the degradation rates of FlhDC, FliA, and a relevant gene of the fli operon (such as FliL), respectively, and
x , y and z their expression rates. We also consider that FliA has no effect on the fli operon
when its concentration is below Kyz , and that FliL has no effect on its target gene when it is
below Kzt .
(a) Assuming that the input to FliA (Sy ) is continuously present, let us consider the situation
in which the input to FlhDC (Sx ) starts at a high level (for which FlhDC is active), and
after a while (long enough to allow the circuit to reach a steady state) fluctuates between
this high level and a low level (for which FlhDC is inactive), as shown in the figure above.
In that situation, analyze the expression of fli in response to the sudden drops in Sx for
both short and long durations of the drop.
(b) Calculate the minimum duration of a drop necessary for the perturbation to have an effect
in the activity of FliL.
13 Perfect adaptation. Consider a protein P whose expression is activated by a transcription
factor S. This transcription factor also activates the expression of a protease X, which binds to
P and degrades it:
P +X X
Consider that the expression rate of both P and X is linear in the concentration of transcription
factor and that the protease X is stable, and ignore the dilution of P (in front of the much more
important enzymatic degradation).
(a) Obtain the differential equations for P and X that describe the dynamics of this system.
(b) Plot the nullclines and vector field of the system in the phase plane formed by X and P .
Use the information to establish the number and type of fixed points of the system.

(c) Show in the phase plane the trajectory of the system that results from a sudden increase in
the transcription factor S (which can be considered as the stimulus here). Plot the resulting
behavior of P and X in time. Show that the system responsive in an adaptive manner, since
there will be an initial response to the stimulus increase, but the protein P will return to
its initial state eventually, even though the stimulus is maintained at a high level.
14 Fold-change detection. Consider the incoherent feedforward circuit of the figure, for which
the state of X i Y combine in Z by means of an AND logic with input function:
z X
Y
Assume that the activation of Y by X is linear in X (fy (X) = y X), and let y i
z be the decay rates of Y i Z. In these conditions:
fz (X, Y ) =

(a) Write down the differential equations that describe the dynamical behavior of Y i Z.
(b) Analyze the behavior of the system in the phase plane and calculate the fixed points.
(c) Represent graphically the response of Y i Z in time to a sudden increase of X at t = T0 :
X = X0 for t < T0 , and X = Xf for t T0 , with Xf > X0 (assume that for t < T0 the
system is in steady state).
(d) Consider next the case in which both X0 and Xf increase by the same multiplicative factor
, in such a way that they keep the same ratio: for t < T0 , X = X0 , and for t T0 ,
X = Xf . Show that in this case the response of Z is independent of : the circuit detects
multiplicative increases (fold changes) of X, but not its absolute value.
15 Expression dynamics of a SIM motif. Considerem un m`odul dentrada u
nica en el que
una protena X activa lexpressi
o de dos protenes diana Y1 i Y2 . Siguin k1 < k2 els llindars
dactivaci
o per X de les dues protenes, per sobre dels quals els ritmes dexpressio son 1 i 2 ,
respectivament. Siguin , 1 i 2 els coeficients de decament de X, Y1 , i Y2 , respectivament.
Suposem que inicialment el nivell de totes tres protenes es zero, i que en un instant t = T1 es
comena a expressar X a un ritme constant . Posteriorment, en t = T2 , es deixa de produr X.
Amb aquestes dades, calculeu levolucio temporal dels nivells de Y1 i Y2 per 0 < t < . Quin
nom rep aquest programa temporal dactivacio?
16 Equispaced sequential activation from a SIM motif. Consider a SIM motif controlled by
a regulator X that activates a collection of targets Zi (with cooperativities ni much larger than
1). Assuming that at time t = 0 X begins to be produced at a constant rate :
(a) design the activation thresholds Ki in such a way that the targets are activated one after
the other at equal intervals t.
(b) Let us consider now that the master regulator X has only one DNA binding site, which is the
one used in the transcriptional activation of all targets Zi . In that situation, the differences

between the activation thresholds Ki will come from the different binding sequences in the
Zi promoters. Assume that, under those conditions, the minimum attainable difference
between the activation thresholds is 10%. Calculate, in the limit t  1, the maximum
number of targets that can be activated sequentially in this SIM motif.
17 Dynamics of the bi-fan motif. The minimal dense-overlapping regulon is the bi-fan motif,
in which two proteins X1 and X2 activate the transcription of two genes Z1 and Z2 (all to all).
The input signal of X1 , Sx1 , appears at time t = 0 and disappears at t = D, and the input signal
of X2 , Sx2 , appears at time t = D/2 and disappears at t = 2D. Assuming large cooperativity in
the interactions, represent graphically the dynamics of Z1 and Z2 , if their input functions follow
the AND and OR logic, respectively.
18 A biochemical adder. Design a molecular circuit that adds the number of two proteins V and
W, and stores the result into the molecule number of a third protein, U. To that end,
(a) design a Bare Bones code that implements the addition described above, and
(b) translate the Bare Bones code designed above into a molecular circuit.
19 A non-destructive biochemical multiplier. Design, using the Bare Bones programming
language, a molecular circuit that can multiply the molecule numbers of two proteins V and W,
without destroying them. The result should be stored in the molecule number of a third protein,
U. To that end, use as many auxiliary variables T as you need.