1 of 36

Quick grading update

  • HW1 is graded. I will post an answer key and figure out how to distribute the graded assignments via Blackboard
  • Remarks from Friday presentations will be back this week also
  • HW2 is in the works; should be back this week
  • HW4 will be posted tomorrow evening

2 of 36

Lecture 16: Gene regulation and networks

Today:�

  • Review gene expression model�
  • Examine the gene regulation network of E. coli
  • Look at the key features of the network or “motifs”�
  • Analyze a major motif: self-regulation

3 of 36

A simple model of gene regulation

DNA

mRNA

protein

transcription

translation

 

 

Rate of change of mRNA conc.

transcription from DNA

Active degradation by RNase

Rate of change of protein conc.

Dilution from cell growth

translation from mRNA

4 of 36

Constitutive expression:

Due to fast degradation, mRNA reaches a maximum level very quickly!

mRNA production perfectly canceled by mRNA degradation

Due to much slower degradation, protein takes a long time to reach a maximum level!

Concept of “separation of time scales

(Started from an unrealistic condition of no mRNA or protein)

Because mRNA reaches a constant level so quickly, we’ll make a simplification:

 

“steady state” assumption

5 of 36

Constitutive expression

 

 

 

 

 

 

 

What is the steady-state protein concentration?

6 of 36

Constitutive expression

 

 

 

Stable steady state!

7 of 36

Constitutive expression

 

What is the steady-state protein concentration?

 

 

 

 

Directly proportional to translation rate

Directly proportional to transcription rate

To maintain an unregulated protein at a low concentration, you have to synthesize it slowly!

(Will become important when we talk about regulation)

8 of 36

Gene regulation

Gene X

Gene Y

X promoter

X coding sequence

Y promoter

Y coding sequence

Y

repression

This interaction depends critically on the promoter DNA sequence!

9 of 36

Network Motifs

If there are innumerable possibilities for gene regulatory networks, and mutation can constantly change, them, how do you figure which networks are important to understand?

Network motifs:

  1. compare the measured regulatory network of E. coli to a random network that has many of the same properties
  2. Identify the structures that are much more common in the E. coli network then the random network. These are called “motifs” and provide us a guide for what to study

How does this work?

10 of 36

Network Motifs

We will look at networks as graphs.

These circles represent genes. Called “nodes” in network language.

These lines represent regulatory relationships. Could be repression or activation. Will call the lines “arrows”

Gene X

Gene Y

X represses Y

Gene X

Gene Y

X activates Y

Both of these scenarios would be represented by the graph above.

11 of 36

Network Motifs

We will quantify the structure of the graph with a few metrics

Network 1

Network 2

 

 

 

 

1

2

1

2

3

3

4

4

5

5

6

6

7

7

8

8

9

9

10

10

1

2

3

4

5

6

7

8

9

10

11

12

13

14

1

2

3

4

5

6

7

8

9

10

11

12

13

14

 

 

(“natural” E. coli network)

(random network)

12 of 36

Network Motifs

How do you generate a set of random graphs to compare to the E. coli network?

Node/gene

Arrow/regulatory interaction

Erdös-Rényi algorithm:

  1. Pick two nodes at random (can pick the same twice in a row), draw an arrow between them
  2. Repeat until you have the appropriate number of arrows (with no double arrows)

5 nodes

6 arrows

Identify structures or “motifs” and count how many there are of each!

“feed-forward loop”

Self-regulation

Et cetera . . .

13 of 36

Paul Erdös

Eccentric, traveling mathematician from the 20th century

Authored papers with over 500 different collaborators!

Prolific collaboration led to the creation of “Erdös number”, how many degrees one is away from Erdös in terms of co-authoring papers.

14 of 36

Erdös number

15 of 36

Kevin Bacon number

16 of 36

What are the properties of the E. coli network?

This subset of the network contains:

  • 424 operons
  • 116 transcription factors
  • 577 interactions (inhibition or activation)

 

 

Alon, An Introduction to Systems Biology

17 of 36

What are the properties of the E. coli network?

This subset of the network contains:

  • 424 operons
  • 116 transcription factors
  • 577 interactions (inhibition or activation)

 

 

This network is actually very “sparse”: it has a low proportion of the possible connections. To see this, again imagine you have a network of 5 nodes/genes. How many possible connections are there?

5 possible connections for the first gene.

Because regulation can go in either direction, every other gene can also have 5 connections. So the maximum possible number of connections is:

 

# genes

# connections each can make

 

 

18 of 36

What motifs dominate the E. coli network?

Alon, An Introduction to Systems Biology

 

Two network motifs very common compared to random:

  1. Autoregulation (self-repression or self-activation)���
  2. Feed-forward loops

19 of 36

How many autoregulatory interactions do we expect to see in a random network?

 

 

 

 

 

If we know the mean number of self-regulations, and self-regulation is a rare, independent event, how is the number of self-interactions distributed in the population of all random networks?

 

For a random network like E. coli:

 

 

20 of 36

How many autoregulatory interactions do we expect to see in a random network?

 

For a random network like E. coli:

 

 

How many autoregulatory interactions are actually in E. coli?

40

 

Self-regulation is a very important component of gene regulation in E. coli. We should understand its dynamics and what its function might be!

Next time we’ll look at the other major motif in E. coli, the feed-forward loop.

21 of 36

Auto-regulation

 

 

How do we incorporate auto-regulation into our gene expression model?

DNA

mRNA

protein

Auto-repression, for example:

 

22 of 36

Auto-regulation

 

Transcription rate

 

How do we account for this mathematically?

 

Let’s use this to see what one very important function of self-repression is.

 

23 of 36

Self-repression enables faster response

 

Remember that for unregulated expression, the steady-state protein concentration depended on transcription and translation rate:

Synthesis rates depend on sequence of promoter and ribosome-binding site of mRNA

To keep the concentration low…

…the cell has to synthesize it slowly

How can a cell maintain a low concentration, but be able to synthesize the protein rapidly? Self-repression!

24 of 36

Self-repression enables faster response

Let’s look at self-repression with the simplified step-function regulation:

 

 

 

Time

 

 

 

 

 

…synthesis can be very fast and reach a low concentration quickly!

For an unregulated gene, this would need to be very slow!

25 of 36

Self-repression enables faster response

An experimental test of this prediction:

Two strains of E. coli:

GFP

Induced constitutive GFP expression

Induced self-repressed GFP expression

GFP

26 of 36

What do they see?

Self-repression enables fast gene expression response!

What’s a less simplified way to model-self regulation mathematically?

27 of 36

The Hill Function

 

 

maximum transcription rate; depends on RNA polymerase binding to promoter.

transition concentration; depends on transcription factor affinity for promoter sequence.

sets how steep the transition is

(n = 5 on the left)

(= 15 here)

28 of 36

The Hill Function

 

29 of 36

Hill function for activation

 

 

One problem here is that with this formula, there is no transcription at all without the activator.

That is not necessarily the case. How do we take that into account?

30 of 36

Hill function for activation

 

First, some math:

 

 

 

Now introduce one more term to the formula.

31 of 36

Hill function for activation

 

 

 

 

 

 

What about transcriptional repression?

32 of 36

Hill function for repression

 

 

 

 

 

 

33 of 36

What kind of gene expression dynamics does autoregulation predict?

We saw with our simplified, step-function self-repression that there is a stable, steady-state protein concentration. We can see that from the more detailed Hill function too:

 

 

 

 

Stable steady-state!

The steady-state concentration will depend on synthesis rates (RNAp affinity for promoter), K (repressor affinity for promoter), and cell growth rate!

34 of 36

What about self-activation? Let’s do the same graphical analysis.

 

 

 

 

Stable steady-state

Self-activation can have bi-stability!!!!

 

 

Stable steady-state

Unstable steady-state!!

35 of 36

Stable states of self-activated genes

 

protein concentration rate of change

 

Depending on relative synthesis rates, binding affinities, and cell growth rate, self-activation can lead to bistability or one stable steady state.

Steady state over there

bistability

Stable steady state

36 of 36

Bistability not to be common because parameters need to be fine-tuned

Can be realized experimentally though!

 

Next time: feed-forward loop!