yeast two hybrid

[Microbio]

Hello,

 

I am wondering what I could expect when doing a yeast two hybrid experiment with 96 spots.

 

When I run my experiment (1 bait vs 96 preys) I often end up with lots of growth on my final selection plate (the one you use to check for interaction).

I often have 15 spots that show growth.

Now I plate them out twice (I am doing it manually, so I just make 2 selection plates for every test): so on 2 plates to select for interaction. The weird thing is that the growth often does not match. I see growth on different spots.

Often I have, eg. 15 spots that grow on plate 1 and 17 spots on plate 2 and 10 of them will match for example.

(and I already quit incubating them after 5-7 days because if I incubate even longer I get growth on even more spots , up to 70 per plate, out of 96)

 

Now is this normal?

(I already dilute the cells in water before plating them out, since going straight from ager to agar seems to give too many growth)

 

Also: when I repeat the y2h experiment and plate out again , I often see growth on other spots (where there was no growth before) , if I add everything up, I end up having 1 or 2 spots where I see growth 3 out of 4 times or 4 out of 4 times.

 

Is this something I can expect or?

 

It seems, to me, that this system is pretty "random".

I know there are lots of false positives, but this seems a bit too much?

 

The way how they grow it seems very hard to figure out the true ones ! At the moment I am very harsh in my selection and I just look at the ones that grew 4 times (so 2 seperate , independent tests, 2 times 2 plates).

 

I also saw papers where they use a robot, often spotting 4 times on 1 plate, however I find this a bit weird since its not very independent since its just 4 times from the same spot/colony, its not really indepedent in my opinion.

 

 

[CharonY]

Well, there is a lot that can go wrong and without looking at the raw data or precisely revising protocols it is tricky to see whether there is something off with the protocol.

That being said, Y2H has been shown to create massive amounts of false positives, if conduct large meta- and/or validation studies. However, typically under controlled conditions it is not as bad as what you described.

[Microbio]

Well, there is a lot that can go wrong and without looking at the raw data or precisely revising protocols it is tricky to see whether there is something off with the protocol.

That being said, Y2H has been shown to create massive amounts of false positives, if conduct large meta- and/or validation studies. However, typically under controlled conditions it is not as bad as what you described.

 

I use a standard protocol and use a manual 96 pinner.

 

1) grow cells on the appropriate medium (5-6 days)

2) plate cells from the appropriate medium to the mating medium (5-6 days)

3) Replate cells on the medium to check for mating (I work with an auxotrophy, so if they mated, they should survive). Till this step: everything goes smooth.

So there is not really a huge problem here.

4) replate the surviving cells on the selection medium by using the 96 pinner, dissolving the cells in 60microliter of H2O and then plate them on the medium for the interaction and grow for 5 days.

 

Its step 4 where I find that things get a bit more random.

 

I can not use 3AT because I use a different gene for the interaction check.

[CPG]

Not to derail the hard work you've already put in, but for your consideration if this continues to be an issue... If you're worried about false positives and troubleshooting isn't going well, do you have the option available to you to take a different approach to the same question? I don't know what would suit your purposes well, but assuming you're just looking for protein-protein interactions there is FRET, co-IP, BiFC / DERB, that sort of thing?

[Microbio]

No,

it is not really an option to use FRET or other techniques for a high-throughput selection.

I guess I'll have to go through lots of potential interactions to finally find true ones.

 

Not to derail the hard work you've already put in, but for your consideration if this continues to be an issue... If you're worried about false positives and troubleshooting isn't going well, do you have the option available to you to take a different approach to the same question? I don't know what would suit your purposes well, but assuming you're just looking for protein-protein interactions there is FRET, co-IP, BiFC / DERB, that sort of thing?