I've tried to post my reply to this quite a few times but as best I can tell (and my friends can tell from their CCP4bb subscriptions) I have been foiled by jiscmail every time. Hopefully this will go through. And hopefully I'm not reposting stuff I already sent. Here we go:
Okay, the consensus here seems to be that you don't have a thermal shift assay up and running with your membrane protein or membrane proteins in general. Like a couple of people mentioned the Alexandrov 2008 is the standard reference for membrane proteins (CPM reacts with a thiol). You have options though: Thiol-reaction activated fluorophore (e.g. CPM) High-throughput static light scattering (Harbinger Stargazer, Avacta Optim) High-throughput intrinsic fluorescence (Avacta Optim) High-throughput fluorescence lifetime (NovaFluor PR Fluorescence Lifetime Plate Reader) Western blotting (CETSA, FASTpp) Environmental rigidity sensitive dyes (e.g. DCVJ) Environmental dielecticity / hydrophobicity sensitive dyes (e.g. SYPRO Orange, bis-ANS) Possibly differential scanning calorimetry (DSC) Thiol-reaction activated fluorophore Cysteine side chains are typically buried in the core of a protein. When the protein is denatured they become solvent accessible. A dye like CPM can then react with it and only the thiol-reacted compound is fluorescent. CPM has maximum excitation/emission of ~385/470 nm which is a slight problem. Most qPCR machines have excitation filters that begin at ~450 nm and emission filters that begin at ~500 nm. I have however seen somewhere in the literature someone excite CPM with about ~405 nm and reading fluorescence above 500 nm. I haven't tried it but I have tried bis-ANS where I excited at 455-485 nm but without getting a useful signal (bis-ANS has the same excitation maximum as CPM). Some qPCR machines can be fitted with excitation filters that start at 350 nm (Stratagene MX3000/3005, Qiagen Rotor Gene, possibly more). More fluorescence plate readers have these wavelengths but then often don't have temperature control or the temperature is capped at 42 or 65 C. 65 is probably fine for membrane proteins but I would want to start with a test protein and I can't think of any that would melt at that temperature. But actually, you could just add Gu-HCl to ß-lactoglobulin - ß-LG melts in the 70-80 degrees interval but with enough guanidinium it should be fine. Without temp. control you could incubate outside the plate reader but you should be quick about it. I've been told not to incubate with CPM since it does degrade. CPM is of course incompatible with ß-mercaptoethanol, DTT, and tris (tris because of the primary amine unless you are at pH where it is well protonated and has no buffering capacity). pH is restricted to ~neutral. http://www.ncbi.nlm.nih.gov/pubmed/18334210 I tried what Artem suggested with having a friendly chemist cook me up one of those Korean / Chinese compounds that work like CPM but fluoresces in the visible spectrum. It didn't work for me but perhaps I didn't try hard enough or it wasn't pure enough or something. There are a lot of them but keep in mind that a lot of them are designed to react with both the thiol and the amine of cysteine. In a protein that amine is usually an amide and therefore no go. Update: this is the one Artem used http://www.ncbi.nlm.nih.gov/pubmed/19343759 There's a paper where they used CPM with membrane protein in LCP. They had to centrifuge it after each heating step though because the LCP became cloudy. High-throughput static light scattering Using a specialised machine to read the aggregation state of many wells in parallel using static light scattering while heating. I know of two machines, Harbinger Stargazer and Avacta Optim, but unless you can find one to borrow this might be a bit over budget. High-throughput intrinsic fluorescence The Avacta Optim also reads intrinsic fluorescence at the same time. I don't know if this adds any useful information not already supplied by the light scattering, I haven't tried it. High-throughput intrinsic fluorescence lifetime The lifetime of tryptophan fluorescence differs between folded and unfolded protein and by measuring the lifetime of UV-excited fluorescence at temperature intervals you can get a measurement of the melting temperature of your protein. NovaFluor PR Fluorescence Lifetime Plate Reader is the only machine for this that I know of. Western blotting This is pretty interesting. As far as I can see CETSA is a simplified version of FASTpp but with broader applicability. FASTpp is Fast Parallel Protealysis. Crude lysate + thermolysin. Heat but take out aliquots at intervals. Thermolysin is specific for bulky hydrophobic residues and since most of them are buried the digestion is greatly accelerated when proteins denature (this is true of all proteases though, regardless of their specificity). The aliquots are run on an SDS gel and Western blotting is performed to determine at what temperature the protein of interest becomes degraded. The drawback is that thermolysin is only active to around 80 C. And thermolysin is dependent on calcium for activity (unless you get the calcium independent mutant but then you have to express and purify it yourself). And if you want to probe stability in conditions that inhibit thermolysin you have trouble as well. http://www.ncbi.nlm.nih.gov/pubmed/23056252 CETSA does away with the thermolysin and simply centrifuges away the denatured protein. They also got it to work with whole cells which I am quite excited by. It's the same number of Western blots as FASTpp though. http://www.ncbi.nlm.nih.gov/pubmed/23828940 Environmental rigidity sensitive dyes Using DCVJ, 4-(dicyanovinyl)julolidine, researchers at Ludwig-Maximilians-Universität have investigated antibody stability in detergent. It seems like they had to use a very high concentration and from the paper I can't quite see how they got information with the DCVJ that they didn't get with SYPRO Orange at the same (high) concentration. Maybe I misread something. I saw 40 mg/ml and moved on. But it might work better with a different dye. http://www.ncbi.nlm.nih.gov/pubmed/23212746 Environmental dielecticity / hydrophobicity sensitive dyes According to researchers at University of Colorado they got something useful with 3 out the 4 membrane proteins they tested with SYPRO Orange by using careful background subtraction (but protein at 1-2 mg/ml). They don't use this technique any more though, they've gone over to the lifetime fluorescence thing. http://www.ncbi.nlm.nih.gov/pubmed/16552147 If you look in the original Pantoliano they also looked at bacteriorhodopsin with a non-ionic detergent at ~CMC and got a beautiful signal. So you might just want to try that. Disclaimer: They didn't use SYPRO Orange, they used dyes that you excite with UV light. http://www.ncbi.nlm.nih.gov/pubmed/11788061 Differential scanning calorimetry (not sure if it’s okay for membrane proteins) There’s at least one machine with a reasonable sample use and okay throughput (MicroCal VP-Capillary DSC from GE Healthcare) and there are papers about using DSC for accessing protein stability (like this one http://www.ncbi.nlm.nih.gov/pubmed/23022410) but I suspect that you might also see a phase transition for the detergent micelles and lipids (if you have them) and I have not come across any stories about whether membrane proteins are fine in this assay - may possibly just require a simple background subtraction correction. Now, once you think you have an assay that works with your protein you have to validate it. I think the best way is to destabilise your protein slightly. Collect data with guanidinium hydrochloride at the following concentrations: 0 mM, 50 mM, 100 mM, 200 mM, 500 mM, 1 M. You should see a lower melting temperature as Gu-HCl concentration goes up. Does-response curves are one of the best experimental validations of any result. Now you should examine if there are divalent cations bound to your protein. Do an experiment with 5 mM EDTA pH ~7, 5 mM EGTA pH ~7, and one with just water. If your protein co-purified with anything then you should see an effect on your stability. Usually that is a destabilising effect. Then you screen against a panel of divalent cations and you should see a stabilising effect because after purification you only have partial occupancy of whatever binding site there may be. If it were fully occupied then that would indicate either a covalet, irreversible binding or that the ion is present in your buffer at a higher concentration than the k_d. I have seen a stabilising effect of chelating agents (once) but then there was a stabilising effect of one divalent cation and a destabilising of another. So the destabilising ion bound in the metal site but induced an unfavourable conformation. Now you should investigate whether there are non-specific ion stabilising effects. Often divalent cations like sulfate and malonate have a stabilising effect and things like sodium and ammonium could be positive, neutral, or negative. There are crowding effects and ion strength effects that influence the stability of your protein and these are individual for each protein. Applying the Hoffmeister series to protein stability is bollocks. At best you see a higher probability of stabilising effects from ions high in the Hoffmeister series. It is supposed to be about the ability of salts to precipitate proteins anyway. You could do all of these above in one go. There’s an example of the current salt screen we use if you look in the example data at http://github.com/grofte/NAMI - NAMI is an open source, Python program from Durham for analysis of thermofluor data. Just a quick plug for my stuff. The manual is an okay read as well. Of course in your case… It’s a membrane protein so you might see a difference in salt effects based on which detergent you are using. I would expect that ionic detergent would act highly different depending on the salt composition of your buffer (salt in the scientific sense, not just NaCl). So if you change salts you should go back and do the analytical SEC detergent selection screens again. I hope this helps, Morten On 12 April 2014 16:00, Artem Evdokimov <[email protected]> wrote: > There is an alternative method that does not rely on hydrphobic > interaction of dye with the protein interior: it relies instead on reaction > between fluorogenic dye and interior cysteine residues of the protein. When > protein melts these Cys residues become exposed, react with the dye and > generate fluorescence. It works very well, with two caveats: 1) the really > good yellow dye is not commercial, last time I checked (there is a masked > blue dye, but it's not as good and it requires excitation in UV) and 2) you > need 'unusual' excitation and emission wavelengths. > > I have not checked recently, maybe someone developed a nice green or red > emitter, and then this is an ideal method for membrane proteins and > anything else that requires detergents... > > I believe the method was first reported by one of the GPCR-structure > teams, who used the blue version -- I tried it a long time ago with the > yellow version made for me by a friendly chemist. There was a chinese paper > describing the dye synthesis (it was a Michael-reactive double bond that > masked the fluorophore). > > Cheers, > > Artem > > - Cosmic Cats approve of this message > > > On Sat, Apr 12, 2014 at 3:38 AM, Theresa Hsu <[email protected]> wrote: > >> Dear all >> >> Does anyone has experience with Thermofluor assay to find the substrate >> transported/binding by a membrane protein? My protein does not have any >> similar structures and the substrate suggested by sequence analysis is not >> being transported in proteoliposome. I know ITC is good but I am looking >> for a more high-throughput way. >> >> Thank you. >> > > -- Morten K Grøftehauge, PhD Pohl Group Durham University
