Hi Shirley

I have not worked with the machine you are using, myself. However, I have done thermal shift assays on several proteins before (GM-CSF and antibodies a.o.). However, I ran it on a Roche Lightcycler 480, in white 384-well plates.

This is the protocol:

In each well:

- 1-5 µg of protein (final concentration ~ 50 - 300 ng/µl)

- A final concentration of '20x' Sypro orange dye (supplied as '5000x' concentrate in DMSO from Life Technologies).

- Dilute with the buffer your protein is dissolved in, to get a final volume of 15 µl. Make sure you be consistent with the buffer, because the melting temperature of your protein will change with the buffer!

That's it, no other things added. As long as you're not excessive with salts or fluorescent things, you should not run into problems. Beware of detergents though, as they might interact with the hydrophobic SYPRO.

You should also prepare a blanco without protein, as you might get some change in fluorescence solely due to the temperature.

When you're done, seal the plate or wells.

Then I recommend you write your own method (if this is possible though) for the machine you are working with! This assay requires very SLOW heating, because SYPRO orange will only work in a hydrophobic environment of the molten globule stage. If you heat too fast, this stage is too short-lived.

I used the following settings on the Roche Lightcycler.

A filter combination of 498 nm (excitation) and 610 nm (emission).

Start off with 25°C (you can ramp up quickly if you wish, mine was 4.4°C/sec).

Then, ramp SLOWLY between 25°C and 95°C. The ramp rate is 0.02°C per second!! Really, the heating rate is key here.

Then cool down (this is just to avoid burns...) quickly to 25°C again (ramping at 4.4°C/sec again).

The run takes about 1 hour.

You should get something like the file added. Note that the samples were run in triplicate and the thickness of the curve is due to all the error flags on the datapoints.

After that, You use only the start and rising part of the curve, until it reaches the maximum (omit the later data points), to fit a Boltzmann sigmoidal. The V50 gives you the midpoint of the slope and corresponds to the melting point.

It's indeed possible that the start of the curve is quite high in signal. However, with a correct heating rate, you should get a drop in signal until right before the melting point. Use the lowest value as your starting value. I think this phenomenon is due to partial association of the SYPRO dye with the protein, maybe at some hydrophobic patches, or maybe when the protein is slightly precipitated. As you melt outr the precipitates, the signal drops, until right before the melting point.

Best of luck with this...

Leander

what is RFU? did you take the first derivative of RFU?

RFU is relative fluorescence units, it's basically just fluorescence intensity, dimensionless.

No, I did not take the first derivative of RFU, the plot you see is raw measurement data, as they come from the machine. You do a Boltzmann sigmoidal fit to this data and the V50 gives you the Tm (as I explain above). However, you only use the first part of the data (until the maximum RFU value) to fit your boltzmann sigmoidal, trying to fit it to the dropping part of the curve doesn't make sense.

To give you a rough idea about the V50; it's the midpoint of the rising slope. In this case, the melting points are slightly below or above 60°C.

Hi,

if you just want to establish the assay I would suggest to use another protein instead of BSA. There is a reason why you can buy BSA as "free from fatty acids". It has some hydrophobic properties and can bind fatty acids (that's one of its functions!) that might be a problem because the fluorescent dye will bind at those sites too. So you will have a high fluorescence signal from the beginning and not observe any increase due to melting. That doesn't explain the lysozyme result though. If you get no signal at all, you should just run a few samples in all available channels, one after the other, you have to get something with one of those at the end.

I'm beginning to wonder if the problem arises from my PCR tubes/plates.

For tubes, I'm using DIATEC420-1375 from Diamed and the lids are from Froggabio AVST-01 (Optically Clear Flat Caps). For plates, I'm using Sarstedt's 96 PCR Plate without Skirt (ref: 72.1978.202) and the seals are Bio-Rad's Microseal B Film (MSB1001).

Hopefully these are it. If they're not good, any recommendations?

Thanks.

Hi again Shirley

In my case, the plates were definitely white. If I look up the Bio-RAD machine, you can get white plates as well. Maybe order those (e.g.: http://www.bio-rad.com/en-be/sku/mll9651-multiplate-96-well-pcr-plates-low-profile-unskirted-white).

With the white plates, the fluorescence signal is more intense; some 2.5 to 3x. This should help in any case...

And of course you have to use optically clear caps/covers, but that seems to be OK with the covers/seals you mention.

Although the signal with the clear plates is lower, you would not really expect to see nothing at all.

Best of luck

Hi Leander,

Thanks for your reply.

I agree with you that it would be weird for me to not see anything.

I read somewhere that white plates will help enhance fluorescent signal by as much as 10X. Funny thing is that I have already increased the concentration of SYPRO accordingly but to no avail :(

I'm starting to think that it may not be a materials problem but the instrument itself. Even though others have used Bio-Rad's CFX96 to do this. Mine is probably not working properly. I can't find any other way in doing this.

The CFX96 wouldn't allow us to customize the filters. If it did, I think it would have been better.

Dear Shirley, how did it end? Did you manage to make it work? I would alo like to try setting up the assay for BSA, because it's cheap and we have heaps, but I am afraid it will bind immediately to Sypro as Huseyin suggests. Your experience would be very helpful for us!

Shirley,

Have you figured out what's happening?

Klara

Hi,

that curve of BSA is correct, it is happening due BSA have a lot of hydrophobics exposed residues, in the initial temperature the sypro binding to this residues, while you heat the protein the sypro unbinding to this residues

Did same experiment use BSA on Biorad machine, see exactly same thing, but for other proteins I can see peaks. Problem is the RCU always start with something high instead of close to zero, any idea?

Hi Qianqian Ma

I am planning to do a thermal melt assay using BSA, I was wondering if you were able to solve your problem with high RCU at the beginning for your melting curves.

Thank you.