Lecture 12: Intro to Bacterial Evolution: Luria-Delbrück
Today�
Bacteria evolve
Et c . . .
There are errors in this process, resulting in mutants.
If a selective pressure arrives, some mutants will have an advantage and take over the population.
Can we design an experiment to understand this process?
First we will need a controlled selective pressure to perform experiments.
Phages (viruses) can infect bacteria
DNA
Phage
2.
1.
3.
4.
5.
Phages (viruses) can infect bacteria
DNA
Intercalating dye
Observing phage-cell interactions:
Dyes bind to DNA and become fluorescent
Phages (viruses) can infect bacteria
Direct observation of phage injecting DNA
Phages (viruses) can infect bacteria
Direct observation of phage injecting DNA
Phages (viruses) can infect bacteria
Direct observation of phage injecting DNA
Bacteria can evolve resistance to phages
Inoculate cells onto standard media plate
Colonies will grow
Inoculate cells onto plate with phage
Small number of colonies will grow
Bacteria can evolve resistance to phages
Inoculate cells onto plate with phage
Small number of colonies will grow
Inoculate these cells onto phage plate
Many more colonies!
Once a cell acquires resistance to the phage, it can hand that resistance down to its offspring.
The Luria-Delbrück Question
When we add the bacterial cells to the phage plate, does that
T1 phage
E. coli
Note: this experiment was in the early 40s, before molecular biology and genetics was understood. This was still a very open question!
In Nashville
Salvador Luria
Max Delbrück
Vanderbilt University
The Luria-Delbrück Question
Scenario 1: induced mutation
non-mutant
phage-resistant mutant
Inoculate cells
Mutation induced in some cells
Mutant survive and grow; others die
The Luria-Delbrück Question
Scenario 2: pre-existing mutation
non-mutant
phage-resistant mutant
Inoculate cells
Non-mutants die
Mutant grows
What experiment will tell us the difference between the two scenarios???
Is it Darwinian evolution (pre-existing) or Lamarckian (induced)?
(in 1943 it was not at all clear that natural selection applied to bacteria!)
The Luria-Delbrück Question
What do we expect to happen in the induced mutation scenario?
Grow cells in liquid media
Spread particular volume of liquid culture on the plate
Count resistant colonies
If we assume the same rate of induced mutation, i.e. each time we put cells on the plate, the same fraction become resistant, how many colonies do we expect to see on the plate?
What do we expect to happen in the induced mutation scenario?
Let’s say 1 out of every 1 million cells becomes resistant when it comes into contact with the phage.
If each time we spread culture onto the plate, we inoculate ~10 million cells, on average how many resistant colonies do we expect to see?
10 million cells
1/million becomes resistant
10 resistant colonies
If we perform this experiment many times, will we get 10 every time? What do we expect to see?
Sound familiar???
With induced mutations, we expect a Poisson distribution in the number of mutant colonies
If we plate the culture 100 times, and there are induced mutations with average 10 mutants per plate
These are the number of plates we expect to see
With these number of colonies
~4 plates with 5 colonies
~12 plates with 10 colonies
~12 plates with 16 colonies
Et c.
For example, in this case:
For the Poisson distribution, the mean is equal to the variance
This means there is not much spread in the data!
There is ~0 chance to observe more than 25 colonies on a plate.
Even if you repeated the experiment many, many times, you’d never observe this number if the mutations are induced.
What about the pre-existing mutation scenario?
The pre-existing mutations case
As cells divide in the tube, some of them may randomly acquire the resistance mutation.
The daughter cells of this mutant will inherit the mutation lead to resistance colonies of the phage plate.
If the mutation rate is very low, there will almost always be very few resistant colonies on the plate.
However, what about the very small chance of having a mutation arise extremely early during liquid culture growth?
The pre-existing mutations case
What about the very small chance of having a mutation arise extremely early during liquid culture growth?
Because the mutation rate is very low, you’ll rarely see this situation, but if you repeat the experiment many, many times, you’ll occasionally see a large number of colonies—not at all what’s expected with induced mutations!
Then after growth a large number of cells will be resistant and, upon plating, you’ll get many colonies!
The pre-existing mutations case
The pre-existing mutation scenario will result in a distribution of colony numbers with long tail!
With induced mutations, we expect a Poisson distribution.
Variance = mean → no points far above the mean
With pre-existing mutations, because the mutations are inherited, in the rare cases that a mutant arises early during growth, you’ll get a large number of colonies!
Non-zero!
This blue curve is a hypothetical sketch, not the actual distribution!
A schematic depiction of the two cases
(wikipedia)
Growth in liquid culture
Phage selection plating
Repeats
Summing up the prediction
If the number of resistant colonies for repeated experiments has a low variance, then that is evidence of induced mutations
Small fluctuations in the number of colonies!
If the number of resistant colonies for repeated experiments has a high variance, then that is evidence of spontaneous, pre-existing mutations
Large fluctuations in the number of colonies!
This experiment is sometimes called the “fluctuation test”.
Luria and Delbrück’s experiment
2. Inoculate cells onto plate coated with T1 phage
1. Grow E. coli in liquid culture
3. Wait 24-48 hours and count resistant colonies
4. Repeat many times to build up distribution
What did they observe?
First they checked that repeated platings from the same culture were Poisson-distributed. This will check for sampling error or other experimental error.
Even though pre-existing mutations should have a long-tailed distribution across many cultures, within the same culture, the colony counts should be Poisson distributed!
What did they observe?
mean = 16.7
mean = 51.4
mean = 3.3
variance = 15
variance = 27
variance = 3.8
Within a single culture, the number of resistant colonies does not exhibit large fluctuations ✓
What did they observe between different cultures?
Et c . . .
What did they observe between different cultures?
mean = 26.8
variance = 1217!
(std = 35)
mean = 11.4
variance = 694!
(std = 26)
mean = 30
variance = 6620!
(std = 81)
In all cases, the variance is significantly larger than the mean. There is a small but non-zero chance of observing many resistant colonies!
How do intra-culture vs. inter-culture fluctuations compare?
…
Small variance compared to mean.
Large variance compared to mean.
The distribution of resistant colonies has a long tail
Experimental data
Expected number assuming pre-existing, spontaneous mutations (a lot of math)
Long tail!
The huge variance is incompatible with the induced mutation scenario!
Likely scenario
Now with molecular biology/DNA understood, pretty much everyone believes this based on molecular mechanisms.
Molecular note
What is the most common mechanism for resistance to the T1 phage?
Outer membrane transporter FhuA
What have we learned?