580.

429 Systems Biology III

November 3, 2008

Homework 7 Key

3.3, 4.5 points

Autoregulated cascade: Gene X encodes a repressor that represses gene Y, which also encodes a repressor. Both X
and Y negatively regulate their own promoters. a) At time t=0, X begins to be produced at rate , starting from
an initial concentration of X=0. What are the dynamics of X and Y? What are the response times of X and Y?
Assume logic input functions, with repression thresholds Kxx , Kxy , for the action of X on its own promoter and on
the Y promoter and Kyy for the action of Y on its own promoter. b) At time t=0, production of X stops after a
long period of production, and X concentration decays from an initial steady-state level. What are the dynamics of
X and Y? What are the response times of X and Y?
Diagram of circuit and graph of dynamics:
S

Kxx
X

Kxy

Kyy

a)Assuming logic input function in all parameters:

When X(1 ) = Kxy , Y production stops and the concentration of Y exponentially decays from its initial steady
state value of Kyy to 0. The delay is:
X(t) =

X(1 )

X
(1 et ) = Kxy


1
X
ln

(X Kxy )

With strong auto-repression:

= Kxy
=

Kxy
X

Kyy
) is:
2
ln2
= 1 +

The time for Y (t) to reach half its steady-state value (

Note that (X < Kxy ) = 0 Y = Y Y (t) = Kyy et 12 Kyy et t =

ln 2

580.429 Systems Biology III

Homework 7 Key

b)If X production stops (for example, if its activator becomes inactive) its concentration will exponentially decay
from its steady state level Kxx towards zero. At delay 2 it will cross Kxy
X(2 ) = Kxx e2 = Kxy t2 =

1 Kxx
ln
Kxy

Kyy
2y
Note: In the dynamics graph above, we assume strong auto-regulation (in which Xss is much smaller than what it

X
Y
would be without autorepression Xss = K << ), that Kxx <<
and that Kyy <<

After 2 Y is produced at rate 2 and will reach half of its steady state level Kyy after t 12 = 2 +

3.4, 4 points

Positive feedback. What is the effect of positive auto-regulation on the response time? Use as a model the following
dX
linear equation:
= + 1 X X. Explain each term and solve for the response time. When might such a
dt
design be biologically useful?

represents the basal production rate and 1 X is the positive effect of X on its own production (positive autoregulation). X is the degradation/dilution factor, as usual.
dX
Grouping terms gives us the equation
= ( 1 )X, which helps us to see that the degradation rate is
dt
reduced by positive auto-regulation, to an effective rate of 0 = 1 . Assuming that the auto-regulation is not
too strong, that is, that 1 < ,0 the term multiplying X is negative and we approach a stable steady- state as
described by X(t) = Xss (1 e t ). The response time is defined as the time to reach half of the steady state:
T 1par =
2

ln 2
ln 2
>
= T 1simple
2
0

So positive auto regulation creates a longer response time than simple regulation does, due to the reduced effective
degradation rate. Remember that negative auto-regulation speeds response times.
Strong auto-regulation, in which 1 > , can lead to instability and unchecked growth of X in the model. In real
systems, this instability will be limited by other factors (such as saturation of the input function), locking X in
an ON state of high expression even after its activating input vanishes. Therefore this design creates a bi-stable
system, where X is either low or high and fixed. This is the paradigm seen in commitment-type biological decisions,
such as thouse made during development. Positive feedback characterizes developmental systems that make a switch
that is either OFF or locked ON.

3.5, 1.5 points

Turning off auto-regulation. What is the dynamics of a negatively-auto-regulated gene once its maximal promoter
activity is suddenly reduced from 1 to 2 = 0? What is the response time, and how does it compare to simple
regulation?
A negatively-auto-regulated gene, the promoter activity of which has reduced to 2 = 0, exponentially decays to
ln 2
. The negative auto-regulation has no effect in this case! Assuming an additional
zero with a response time of

activator is present which activates transcription at a constant rate when it is present (the ON step), negative
auto-regulation has an asymmetric accelerating effect, decreasing the response time for an ON signal but not affecting the response time for an OFF signal.

580.429 Systems Biology III

Homework 7 Key

3.6, 2 points

Two-node positive feedback for decision making. During development from an egg to an embryo, cells need to make
irreversible decisions to express the genes appropriate to their designated tissue types and to repress other genes.
One common mechanism is positive transcriptional feedback between two genes. There are two types of positive
feedback made of two transcription factors. The first type is of two positive interactions, X Y and Y X. The
second type has two negative interactions X a Y and Y a X. What are the stable steady states in each type of
feedback? Which type of feedback would be useful in situations where genes regulated by both X and Y belong to the
same tissue? Which would be useful when genes regulated by X belong to different tissues than the genes regulated
by Y?

Positive feedback with two positive interactions has two stable steady-states: X and Y both ON or X and Y both
OFF. This is useful when genes regulated by X and Y belong to the same tissue.

Positive feedback with two negative interactions has two stable steady-states: X ON and Y OFF or X OFF and Y
ON; that is, either X or Y is ON. This is useful when genes regulated by X belong to different tissues or cell fates
than the genes regulated by Y.

580.429 Systems Biology III

Homework 7 Key

4.3, 4.5 points

A decoration on the FFL. The regulator Y in C1-FFLs in transcription networks is often negatively auto-regulated.
How does this affect the dynamics of the circuit, assuming that it has an AND input function at the Z promoter?
How does it affect the delay times? The Y regulator in an OR-gate C1-FFL is often positively auto-regulated. How
does this affect the dynamics of the circuit? How does it affect the delay times?

SX

X
Y

AND

OR

Kyz

Kyy
SX

X
Y

Kyz

Kyy

AND: The negative-auto-regulation on Y speeds its response time, shortening the time needed for Y to cross the
activation threshold for ZKyz . If TON is the delay in Z activation following the introduction of X activating signal
Kyz
SX , and assuming strong auto-repression in the negative-auto-regulation of Y, the delay is TON =
, which is

shorter than the delay in a feed forward loop without negative auto regulation. Following the loss of SX , negativeauto-regulation on Y has no effect and the delay is no different than the simple feed forward loop.
OR: An ON step of SX causes an immediate rise in Z, and the auto-regulation of Y has no effect because one active
input to Z is enough. However, Ys regulation affects the dynamics in the case of loss of SX . A positively-autoregulted gene has the following dynamics in a linear model:
dY
= X + 1 Y Y = X ( 1 )Y
dt
where 1 Y is the term representing the positive auto-regulation and BX is the effect of X. When SX goes to zero,
X = 0 and the solution for Ys dynamics is a decay from an initial value Yss , such that
0

Y = Yss (e(1 )t ) = Yss e t

YTOF F = Kyz = Yss e TOF F = Kyz = TOF F =

TOF F
0
TOF
F

1
Yss
ln(
)
0
Kyz

simplemodel
P.A.R.
> 1 = TOF
> TOF
, longer than the delay
F
F