Sciencemadness Discussion Board - Some notes on chromatography (in the interest of protein purification)

Some notes on chromatography (in the interest of protein purification)

eznurak - 13-7-2009 at 05:26

Hey all,

I typed up some notes on chromatography, and protein purification in particular.

http://heybryan.org/books/papers/protein-purification/notes....

I especially like the section on the logical combinations of chromatographic operations.

Logical combinations of chromatographic steps

crude sample -> GF (desalt mode) -> AC -> ___ -> GF
crude sample -> GF (desalt mode) -> IEX -> ___ -> GF
crude sample -> Hlc (dilution may be needed) -> IEX -> GF
crude sample -> GF (desalt mode) -> IEX -> HlC -> GF

clear or very dilute samples -> AC -> GF or RPC
clear or very dilute samples -> IEX -> GF or RPC
clear or very dilute samples -> IEX -> HlC -> GF
clear or very dilute samples -> precipitation (e.g., in high ionic strength) -> resolubilise -> treat as for sample in high salt concentration

chemoleo - 13-7-2009 at 14:30

Except it doesn't work like that in practise

For example, many proteins aren't stable (they irreversibly aggregate) at low salt ('desalt mode'). So you lose your precious sample just by removing the salt.

Many proteins also don't bind hydrophobic interaction columns (HIC = hydrophobic interaction chromatography). They'll just end up in the flowthrough.

Many proteins *irreversibly* bind to ion exchange columns, particularly the Q-type (quaternary ammonium - meaning negatively charged proteins bind). Elution can only occur with SDS (detergent, which wrecks a Q column) or guanidinium hydrochloride, which unfolds, precipitates and usually inactivates the protein.

Protein purification is not for no reason an art unto its own!

eznurak - 13-7-2009 at 14:38

Quote: Originally posted by chemoleo

Except it doesn't work like that in practise

For example, many proteins aren't stable (they irreversibly aggregate) at low salt ('desalt mode'). So you lose your precious sample just by removing the salt. Many proteins also don't bind hydrophobic interaction columns (HIC = hydrophobic interaction chromatography). They'll just end up in the flowthrough. Many proteins *irreversibly* bind to ion exchange columns, particularly the Q-type (quaternary ammonium - meaning negatively charged proteins bind). Elution can only occur with SDS (detergent, which wrecks a Q column) or guanidinium hydrochloride, which unfolds, precipitates and usually inactivates the protein. Protein purification is not for no reason an art unto its own!

At the beginning of the document there was this other section that basically listed out the different types of considerations to take into account for the "window of stability" of your target compound.

Quote:

* temperature stability - need to work rapidly at lowered temperature
* pH stability - selection of buffers for extraction and purification; selection of conditions for ion exchange, affinity or reversed phase chromatography
* organic solvents stability - selection of conditions for reverse phase chromatography
* detergent requirement - consider effects on chromatographic steps and the need for detergent removal. consider choice of detergent.
* salt (ionic strength) - selection of conditions for precipitation techniques and hydrophobic interaction chromatography
* cofactors for stability and activity - selelection of additives, pH, salts, buffers
* protease sensitivity - need for fast removal of proteases or addition of inhibitors
* sensitivity to metal ions - need to add EDTA or EGTA in buffers
* redox sensitivity - need to add reducing agents
* molecular weight - selection of gel filtration media
* charge properties - selection of ion exchange conditions
* biospecific affinity - selection of ligand for affinity medium
* post translational modifications - selection of group specific affinity medium
* hydrophobicity - selection of medium for hydrophobic interaction chromatography

I am a programmer and amateur science tech enthusiast. So, I want to encode this into some sort of software to help me automatically generate steps of different purification procedures, based off of initial properties of a sample encoded into the system. For many different analytes that I might need, there are going to be this usual range of nominal characteristics that have to be taken into account. In some cases, there are extremely unusual characteristics that require special consideration, however for the common "every day" mad science, that shouldn't be the case (otherwise you wouldn't be able to get anything done).

Remember rougelike games? Nethack, maybe. The way that the dungeon level generators work is by a simple algorithm: starting at one point on the map, walk in a certain direction and make this path the "walkable" part of the map. Instead of a 2D map though, we're just doing a one dimensional walk when it comes to "correct next steps given what we know about our poor protein that we want to purify".

- Bryan

chemoleo - 13-7-2009 at 16:25

I commend your enthusiasm - but again - practical experience on protein purification shows - it's pretty empirical.

An algorithm might work on small proteins - such as globular domains and single domain proteins.

But many, if not most proteins, have multiple domains, each with different properties.

This is why a large, multi-domain protein with an overall pI of 7 (neutral) can bind to i.e. both anion and cation exchange columns. Or even one with an overall pI of 9. Different parts within the same molecule have different properties, and they are enough to produce 'unpredictable' behavior.

Going through it by your list:

* Temp stability: most single domain proteins are happy at RT. Multiple domain proteins usually less so. Then again, these multi-domain proteins may very well be happy in complex wiht their natural binding partner - which you sometimes, or often simply do not know. A protein may prove not purifiable (it just aggregates whenever produced in isolation) simplly because it lacks its binding partner(s). But normally 4 deg C is a safer temp - but who wants to work in the coldroom for many days at a time?
* pH stability - again - it may be impossible to predict from the proteins isoelectric point (pI). Normally pH 7.4 is ok. But for ion exchange chromatography, pH will have to be adjusted to usually betw 6 and 8. Some proteins irreversibly aggregate at pH 6.5, but not pH 6.7. Again there's no way to predict this.
* organic solvents stability: Only small proteins or domains can be usually purified by HPLC and *refolded*! This is not predictable whatsoever.
*detergent: normally used with larger proteins, small ones have no need for it. Again it is unpredictable at which point the protein becomes just more soluble and stable, or just unfolds and remains soluble. I'm not happy with the use of detergents at all!
* ionic strength - a key factor. Some proteins (usually small domains and small proteins) are happy at low IS. larger proteins usually aren't. Some proteins are i.e. only happy between 200 and 500 mM NaCl, above and below they aggregate. Others aggregate at any concentration of NaCl. Others again are only happy at low NaCl, alhtough this is less common. Usually higher IS makes them more stable in solution.
* cofactors- a key factor indeed! This may be the all-decider between something that is isolatable, and not.
* metal ions - NOOO!!! EGTA/EDTA are there to remove metal ions from contaminant PROTEASES! Adding this to an unknown protein, which potentially may coordinate Zn or other ions, will undoubtely lead to its unfolding, inactivation and aggregation! Strongly advised against! The literature is FULL of EDTA-induced artefacts!
*redox sensitivity: the common reagent there is dithiothreitol (DTT) . Rule of thumb - unless it is antibodies, or some extracellular proteins, DO use DTT! Intracellular environments are reducing. But again it may be the opposite.
* molecular weight - size exclusion chromatography is good. But still lacks in real separation power, despite all the column media available nowadays.
*affinity - the method of choice. But again it is marred by specificity. People try to select between i.e. di and tri-methylated amino groups (lysines). It is questionable to the extreme whether a single methyl group will decide between a binding and not-binding event!

Also - mad science usually doesn't include in depth biochemistry, because its reagents aren't easily accesible, and equipment is usually horrendously expensive!

If you can still see a one-dimensional walk to purification there, let us know!!

Best of luck!

PS Making things more complicated: Proteins from Halophiles and extremophiles (organisms adapted to either hot or cold temperatures) have even more divergent properties. Thermophilic (hot) proteins may only be able to fold, and be active, at high temperatures. It's the opposite for proteins of organisms living in the cold.
Bacteria don't have disulphides, so DTT is always a must. Not so with eukaryotes (from yeast to humans). Bacteria don't have multi-domain proteins, so purification of these proteins is inherently easier.
Also, another one in the list is the issue of protein expression: Mammalian proteins, when produced in bacteria - can work very well, but sometimes it just does not express. Then eukaryotic expression systems have to be tested. A painful, laborious procedure, which may not lead to success. Here the issue is more of appreciable levels of protein production, rather than its isolation.
Oh, and this is why people usually use affinity tags (GST, His(6) etc) - even very badly expressed protein can be purified that way...

[Edited on 14-7-2009 by chemoleo]

eznurak - 13-7-2009 at 16:32

Quote: Originally posted by chemoleo

I commend your enthusiasm - but again - practical experience on protein purification shows - it's pretty empirical.

Thank you for the reply. I will read over the biochemistry more thoroughly when I have my neurons properly aligned. At the moment however, I want to focus on something in particular you mentioned--

Quote: Originally posted by chemoleo

Also - mad science usually doesn't include in depth biochemistry, because its reagents aren't easily accesible, and equipment is usually horrendously expensive!

I have long known about the hive, synthetikal, sciencemadness, rougesci, wetdreams.ws, and so on (although I always appreciate learning about other sites), most recently the newbie on the block is diybio. Yes, the equipment and reagents are horrendously expensive, but it doesn't have to be that way. Thermostable Taq polymerase, for instance, might make for a good affinity chromatography target. One of my projects involves making an open source hardware package distribution service, much like debian and other linux distributions (but if you don't know what that means, it's ok- just think of it as a way to "download" instructions to build hardware given what you already have). Hopefully this will make it easier to make all of this equipment on our own.

- Bryan