Interactions

Question 1

What is the prey library?

Answer 1

A library of the entire proteome of the specified organism. On the outside our peptide is bound.

Question 2

How are the phage display prey libraries created?

Answer 2

We insert gene fragment into phagemid that contains:
- A phage coat protein gene (often gene VIII) of the M13 phage
- Ori
- M13 ori
- Package protein
It does not contain any proteins necessary for replication or packaging, and if therefore getting help from a helper phage who has all proteins.
The phagemid and helper phage amplifies the library inside of E.coli cells

Question 3

What is the bait proteins?

Answer 3

The proteins within the system you want to investigate interactions for. for example interactions between a mouse and a cat you would introduce the prey library of mice into a cat cell. The proteins present in the cat cell would be bait, and if it binds to anything we will have found interaction.

Question 4

What is biopanning?

Answer 4

Exploiting the phage display system to investigate protein protein interactions

Question 5

What are the steps in biopanning?

Answer 5

1. Make a prey library (add helper phages and move to E.coli)
2. Add bait proteins if in E.coli and let phagemids bind to the proteins
3. Wash away unbound phagemids
4. Elute the bound phagemids (extract them)
5. Repeat but this time with a sublibrary with more stringent conditions to investigate stronger interactions.
6. Iterate until requirements met
7. Sequence the phagemid (that only has the 1 peptide gene) to identify the gene that has bound to the protein.

Question 6

How can we check interaction in biopanning?

Answer 6

We can perform ELISA

Question 7

What is the difference between an expression vector and a plasmid vector?

Answer 7

• E.coli ori
• Res gene
• It will express the phage coat protein fused to a peptide

A phagemid has in addition:
- M13 ori
- Packaging signal

Question 8

What is a bacteriophage?

Answer 8

A virus phage that infects and replicates within bacteria.

Question 9

What are limitations the proteins we can investigate in biopanning?

Answer 9

We can only bind peptides so no whole proteins or protein complexes. For example no antibodies
We are confined to E.coli so we cannot investigate post translational modifications

Question 10

What can we investigate using biopanning?

Answer 10

• Natural proteins
• Synthetic proteins
• Enzymes

Question 11

What is the yeast two hybrid system?

Answer 11

It is a protein-protein interaction identifier. We exploit that the binding domain and activation domain in yeast needs to be bound for transcription.

Question 12

What do we need in the yeast cell for yeast two hybrid system?

Answer 12

• Two different auxotrophic markers to indicate our proteins are being synthesized
• A reporter gene

Question 13

What is an auxotrophic marker?

Answer 13

Something that the cell needs in the growth medium to survive. In yeast it is because we use fx trypsin and leucin genes to report on out proteins. Therefore we have knocked out the genes.

Question 14

Why do we need an artificial nuclear localization signal?

Answer 14

To ensure both proteins being produces enters the nucleus.

Question 15

What are the steps in yeast two hybirid?

Answer 15

1. Create shuttle vectors inside E.coli. The first vector should contain a binding domain, the gene encoding protein 1 and an auxotrophic marker. The next should contain an activation domain, the gene encoding protein 2 and an auxotrophic marker.
2. Knockout following genes in the yeast cell:
   - Binding domain
   - Activation domain
   - Auxotrophic marker of choice 1
   - Auxotrophic marker of choice 2
3. Transfer vectors from E.coli to yeast and add the two auxotrophic markers to the growth medium
4. Incubate the vectors and remove the markers from the growth medium to ensure both vectors are being expressed.
5. If the proteins interact we will see the reporter gene

Question 16

What could be a good reporter gene in yeast two hybrid?

Answer 16

HIS3 that means we are lacking histidine to survive. So if we have interaction, the cell will survive and otherwise not.
3AT is a competitive inhibitor and can be added to make the reporter less sensitive to check for ‘leaking’.

Question 17

What is the tradeoff in yeast two hybrid?

Answer 17

Having false positives or missing some interactions. This is determined for example with the use of 3AT as a competitive inhibitor to the reporter gene.

Question 18

How can we do large scale yeast two hybrid?

Answer 18

1. We can harvest all mRNA from a specific tissue for the prey library.
2. Reverse transcript mRNA to cDNA
3. Fuse every cDNA to a transcription activator domain in the yeast plasmid vector
4. Transfer first prey libary only to yeast, and bait library only to another yeast colony.
5. Extract the prey plasmids from the yeast colony and add to the bait yeast colony
6. Transport to agar plate and look for reporter proteins.

Question 19

What is going from mRNA to cDNA called?

Answer 19

Reverse transcription.

Question 20

How much cDNA do we need for the prey library? And why this number

Answer 20

We need a factor 12 more than predicted number of clones in Clarke & Carbon. Factor 12 because there is only a 1/3 chance that it will be in frame with the activation domain and reverse so 1/6. This equals a factor 6 which we double as a safety.

Question 21

What are the bias with large scale two yeast hybrid?

Answer 21

If we do a cDNA library we will not have equal representation of the genome because the cDNA stems from mRNA. Highly expressed genes produce more mRNA, leading to higher representation in the cDNA library.

Question 22

What if we want to screen for two whole organisms at the same time?

Answer 22

Library-library screening complexity.
IF we want to find all interactions between two organisms we need to do N*M interaction analysis.

Question 23

How can we avoid manually building two libraries in library-library screening?

Answer 23

We exploit that yeast cells can be haploid and diploid. Diploid cells reproduce via mitosis creating daughter cells identical to the parent cells. Haploid reproduce via meiosis producing offsprings or cells different from other parent but containing a little bit of each parent and each cell different from the other.
1. Create a bait library in a A type haploid yeast cell
2. Create a prey library in an alpha type haploid yeast cell
3. Add both libraries to a matrix (with a robot) and allow to replicate. We now get a haploid yeast cell that contains a gene from both library which we can now start screening for.

Question 24

What does A and alpha type haploid cell mean?

Answer 24

Haploid yeast cells come in two mating types, a and ‘α’, each producing specific pheromones to identify and interact with the opposite type, thus displaying simple sexual differentiation.

Question 25

What is yeast one hybrid?

Answer 25

Directly study the interactions between a protein (prey) and a specific DNA sequence (bait). If we have a binding domain for which we want to identify the interaction with the DNA binding motif. So basically we have again a reporter gene downstream from the bait sequence and using transcription factor. Gal4p recognition sequence is replaced by the DNA binding site to be studied. The protein (potentially) recognizing the DNA binding site is fused to the activation domain.

Question 25

What is the advantage of library-library yeast two hybrid?

Answer 25

- Its automated since a robot places the libraries onto the matrix - We can screen for false positives (if a prey cell with the same genome is positive for all bait cells, there might be a problem)

Question 25

What is yeast three hybrid?

Answer 25

Interactions that are non-binary (we can examine interactions that has more than 2 proteins). Adding a third protein, other under inducible expression control. The third protein will act as a bridge formation or stabilizing conformation.

Question 25

What is reverse two hybrid?

Answer 25

If we look at an interaction we know is taking place between two proteins, we can add a third protein to see if it will interrupt the interaction. We can combine it with a lethal reporter gene, so that if we do not interrupt the interaction, the cell will die. This method will not work well if we have membrane bound proteins. If we want to examine these, we can use the SOS recruitment system.

Question 25

What is the SOS recruitment system?

Answer 25

For membrane bound proteins! Here we use the Ras pathway in yeast. The bait is membrane bound (x) and is fused to an SOS which is truncated (C-terminally). Now if we have an interaction between the bait (x) and the prey (y) we will activate the Ras pathway which is needed for survival. Before we start the screening we need the cells to survive which they would not do on their own. Therefore we are using a temperature sensitive allele, so we can grow it at 25 degrees and the natural SOS will be active. Then when we want to start screening we change the temp to 36 degrees where the natural SOS will not be active. Now the interaction will keep the cell alive if it takes place, and otherwise it will die.

Question 26

What is bacterial two hybrid?

Answer 26

In E.coli. We are using the lambda C1 repressor that recognises the specific DNA sequence called the lambda operator. Its fused to a bait protein and that should interact with a prey protein that is fused to the C terminal end of the alpha subunit of the RNA polymerase in E.coli. If there is an interaction the alpha subunit bound to the DNA is going to recruit the RNA polymerase which will induce transcription. If no interaction we will have no transcription of the reporter gene. We need three specific antibiotic resistance genes to accommodate the plasmids. The ori also needs to be compatible to the others used. We need a multiple cloning site that makes an in frame fusion for the bait with the alpha c1 protein and here the pray and in frame fusion with the RNA alpha domain in frame fused to our prey/prey library. Its mostly used for small numbers and not big library screenings. Miller assay

Question 27

What is mammalian two hybrid assay?

Answer 27

The mammalian two-hybrid system is used to study protein-protein interactions in mammalian cells. This system is particularly useful for studying interactions between mammalian proteins that may not fold correctly in yeast or require post-translational modifications or external stimuli not present in yeast. We build it in E.coli and transport into mammalian cells. We need three vectors - Target construct - Reporter plasmid - Bait construct

Question 28

What are the key principles behind protein folding?

Answer 28

- Hydrophobic collapse theory: Hydrophobic residues will fold inwards to avoid water and the hydrophilic will be outwards toward the water - Physics and knowledge based potentials: Van der waals interactions (A weak force of attraction between electrically neutral molecules that collide with or pass very close to each other), sidechain interactions, charge interactions etc - Contacts in proteins are evolutionary conserved (cheatcode): When two proteins are close to each other on the surface then if one protein mutates then the other one will also mutate (!!) to keep the interaction. Therefore proteins in close proximity to each other are often similar in structure and sequence.

Question 29

How can we utilize evolutionary conserved contacts in proteins?

Answer 29

If we then want to examine this we can do multiple sequencing to conduct a distance map. We find proteins that have similar mutations and those will often be close to each other and/or interact. The map is basically a matrix with distances for all residues.

Question 30

What is AlphaFold2?

Answer 30

Predicts protein structures and solves the problem of going from sequence to structure. It does NOT predict interactions. Works by using AI models that predicts a protein structure based on amino acid sequences.

Question 31

How does AlphaFold2 work? Part 1: creating representations and embedding

Answer 31

1. Make multiple sequence alignment to utilize the evolutionary conservation (see above). 2. Search genetic databases 3. Embed the multiple sequence to create a multiple sequence alignment (MSA) representation. Again we use the evolutionary conservation to create the representation. 4. Bottom parts: AlphaFold2 will start looking for similar proteins that have been experimentally solved, to use as a template for the final structure. 5. It will use the distance map (see above under evolutionary conservation) which shows how close the different residues are to create another. 6. This gives us a contact map which is a pair representation giving a first hypothesis of the proximity among the residues. 7. Everything we create is embedded using a neural network where they are transformed from discrete to continuous variables (this is embedding)

Question 32

How does AlphaFold2 work? Part 2: the evoformer: the intuition

Answer 32

1. We have our MSA representation and we are making a transformer. The transformer is looking at the MSA representation and looking for correlations between the residues. For example X and Y always seem to correlate so therefore they probably co-mutate (both mutate at same time same place). 2. It passes the information onto making a pair representation 3. Based on the correlations the pair transformer suggests a correlation that is passed on to the MSA in the very first step. These steps are iterated 48 times to keep learning about which residues should be close to each other based on the MSA.

Question 33

How does AlphaFold2 work? Part 3: the structure module

Answer 33

1. The model treats all residues as if they were gas particles, so in the beginning they are not connected. In the beginning all residues are places at the centre of a core unit system. 2. Each residue is moved (either left/right or rotated up or down) one at at time, meaning we iterate over how many residues we have

Question 34

How do we quantify the models preduction?

Answer 34

- plDDT score (predicted local distance difference test score) which is a per-residue confidence - Another quality score PAE (expected positional error) indicating the error expected at residue x if the predicted and actual structures are aligned on residue y. It tells us in more detail if the residues are pointing the right directions in regards to each other (important for interactions). We get a plot where we can see the aligned residuals and the score of the residue.

Question 35

What can alphafold2 not be used for?

Answer 35

Mutants and protein interactions (only protein structures)

Question 36

What is FoldSeek?

Answer 36

A good first step if we dont know what our protein of interest does A structure based similarity search software FoldSeek is to proteinstructures what BLAST is for sequences. It is quick to search in the database because they did a structure alphabet, where each alphabet signifies a structural element. This means that we can transform sequences to structures by using the alphabet system.

Question 37

What is alphafold multimer?

Answer 37

Predicting protein-protein interactions. 1. MSA pairing: find sequences from both proteins (needs to be from the same species) and append them to each other. If we homologs we need to stack them. This ensures we can find co-evolutionary patterns between proteins 2. Penalize clashes: biologically it does not make sense for them to overlock so the added penalties to teach the model not to do it. 3. Multi-chain cropping during training: the model does not see the full sequence all at once, only 384 residues at a time. When comparing proteins tho, its important to see both chains at the same time. Thats why we do the multi chain cropping.

Question 38

What is AlphaPulldown?

Answer 38

Automates the process of AlphaFold multimer. It generates a nice output thats easy to view.

Question 39

What is MolPC2?

Answer 39

Estimating stoichiometry and building large complexes from subcomplexes. AlphaFold could not predict stoichiometry, so we can use this instead

Question 40

What is AlphaFill?

Answer 40

A software that adds ligands and cofactors to AlphaFold models by homology and superimposition.

Question 41

What is RosaTTAfold-AllAtom?

Answer 41

An AI model that the improves on AlphaFold2 so we can look at all atoms, not just proteins