Tell me more about x-ray crystallography, Veeky Forums. I hear integral membrane proteins are quite a challenge

>Tell me more about x-ray crystallography
The field is dying. I don't mean that in the science and technology required is on the way out, but rather as a discipline. Even though it has uses in biology and materials science, knowing the skills and math has become kind of pointless. Apparently having a computer that gives the results is cheaper and more accurate for most institutions/companies than having a person dedicated to knowing it. Go figure.

Fusion with GFP for crystallization is not really the best idea. Usually people fuse their protein of choice with GFP for an easy detection in vivo. High throughput screening for variants from different organisms for example is simplified with this method.

Fusion can work wonders though with the right partner and a good linker sequence. T4 lysozyme or thioredoxin are a good choice.

The paper that you linked is weird. SDS is quite a harsh detergent and even if it solubilizes the membrane protein out of the lipid bilayer, the chances are really high that it might denature at least some parts of the protein. Nowadays milder detergents are used routinely, a screening in a small batches makes sense. There are several papers out there on this topic.
Also their use of an IMAC and a hydroxyapatite column is not really the best choice for membrane protein crystallization. Usually a size exclusion column is being used as the last step of purification to yield a homogenous sample with a minimal concentration of buffer/salt and detergent.
Read this paper sci-hub.cc/10.1038/nprot.2009.27

>sci-hub.cc/10.1038/nprot.2009.27
Thanks, will do. Any troubleshooting advice you'd be willing to offer? Never have performed something of this caliber, but I'm willing to give it a whirl.

bump for good thread

Dr. Berry?

There are other aspects of the field however. Growing crystals of organic compounds takes skill. It can be kind of an art form.

Also, though somewhat disheartening, it is not surprising that people don't bother much with the math behind diffraction. A lot of it is pretty esoteric. I mean, in addition to knowing all the symmetry and space groups, you would have to know about the Ewald Sphere and the reciprocal lattice. All kinds of notation need to be learned throughout. Its helpful to be good at visulization.

Who the fuck

?

I get what you're saying, but I think that's still farther from thinking that crystallography is "dying."

At least in biology, x-ray crystallography isn't dying. As a structural technique, there are questions that x-ray crystallography can answer, that NMR and Cryo-EM cannot. I would admit that there is some truth to your point about not needing to know the skills and math -- and, to that point, most grad students and post-docs nowadays really don't want to know that sort of stuff.

But, I think that the field has shifted; to put it another way, which may be more controversial, the bar has been raised. No one is doing technical or methods development any more. And, nowadays, unless the crystal structure has enormous impact, no one is going to publish the structure without functional data.

:^)

There is a lot of truth in your post, especially the last sentence. If you have a new protein structure and publish it, you'll end up in a journal with an impact factor around 1 maybe. With a little bit of biochemical characterization the sky's the limit though. Nobody cares for small molecules structure papers though. Acta E has an impact factor of 0.3 or so.

I don't know whether the method development is true though. At least not in the macromolecular crystallography. There has been a lot of progress in the last five to ten years in the membrane protein crystallization part like the use of lipidic cubic phases or lipid bicelles. The machine development got a good boost last year with the new Eiger detector that produces really awesome data. And at some point maybe us mere mortals will be able to use an XFEL.

>Any troubleshooting advice you'd be willing to offer?
Don't give up when nothing works.
And if you have the possibility to try out different homologs from other organisms then do it as early as possible. I wasted (well not really, still learned a lot of good stuff) one year trying to crystallize an E. coli protein while a homolog from an other bacteria crystallized after the first purification.

> a eukaryotic TMP, they require a motley of posttranslational modifications like glycosylation and palmitoylation for proper transfer and insertion.
In that case I think there's no way around using mammalian/plant/whatever cells.
Insects/yeast are just too far away evolutionary for proper conservation of signal sequences AFAIK.

Most I can tell you is I know some relatives of this guy

nobelprize.org/nobel_prizes/physics/laureates/1914/laue-bio.html

youtube.com/watch?v=9QSndKuLGiA

>Insects/yeast are just too far away evolutionary for proper conservation of signal sequences AFAIK.
Pichia can glycolize some proteins, and its membrane is homologous to that of higher eukaryotes.

Never mind, misread your post.

could spectral analysis be part of this progress?

Your points on methods development (LCP, detector) are well-taken, I had not considered them. Thank you for pointing them out.

OK anons, ignorant chemfag here
Why don't we just always use NMR for membrane proteins? I mean I know it's difficult, but you would gain the following benefits:
>No longer need to crystallize your protein - you'll get to see what it looks like in the bilayer, as opposed to as a crystal
>You'll get structural, kinetic, and thermodynamic data about conformations

It sounds like there's a lot of trouble getting these things to crystallize. Is NMR already used often here?

>Why don't we just always use NMR for membrane proteins?

Few reasons. There is a size limit to proteins that can be studied by solution-NMR, which is around 50 kDa. Most membrane proteins are quite larger, and any membrane protein-signaling complex would be even bigger.

Also, one of the enormous challenges of membrane proteins is expressing and purifying them. (So, whether you're using NMR or crystallography, you still need to get a significant amount of homogenous, pure protein.) Then, for protein-observed NMR in solution, you'd have to stabilize membrane proteins -- with transmembrane segments that are otherwise embedded in the cell membrane -- in aqueous solution.

Aw shit I see. Are those the only two techniques that exist for macromolecule structure elucidation? X-ray and NMR? I know you can do some fuckery with the various flavors of anisotropy spec but it seems to me like they're pretty useless in terms of elucidation.

Is it possible to elucidate on the basis of the sequence alone? I'm assuming that's well beyond our computational capabilities, I'm just wondering if it's at all possible or if the structure depends on more variables beyond the amino sequence.

You can get a somewhat basic idea about how a protein folds using structure-prediction via i-TASSER, or other similar tools.

>Is it possible to elucidate on the basis of the sequence alone?
we can come up with predictions for helices and sheets versus unordered regions, but when you get up to tertiary/quaternary structures, the complexity is just way too high to get anything beyond a shitty estimate

you might appreciate this
jcrystal.com/steffenweber/
jcrystal.com/

I developed a small interest in crystallography because it seems to be the only application of "classical" geometry.

There's also cryo-EM, SAXS for low resolution structures and there are probably also a few autists using neutron diffraction and shit.

The main techniques are EM (negative-stain and Cryo), x-ray crystallography, and NMR.

X-ray has been the gold standard. There are absolutely limitations, but, IMO, if you want a high-resolution structure to answer a question, a crystal structure is the way to go.

NMR has limitations in the size of proteins, and the process has not really been streamlined. Nowadays, an undergrad can go through the process of "solving" a structure by following the programs that automate everything.

The same sort of automation really hasn't been doable in NMR. (This is due to how you go about solving structures from NMR. It is non-trivial, and I highly doubt it will ever be as, say, "automated" as x-ray crystallography.) IMO, that is one of the things that has heped keep the guys and girls who work on solution structures to a small amount. Another factor here is how x-ray crystallography has been able to be streamlined.

Negative-stain is mid- to low-resolution, you can resolve domains but not secondary structure elements. Cryo-EM is hot, now being pushed to sub-3 A on large, (mostly) highly-symmetrical structures. That can be difficult, and many big targets -- for example, multi-protein complexes --- are still quite difficult. Often, cryo-EM is used to elucidate "architecture" (read: domains and motifs, maybe secondary structures and backbones if you're really lucky) of complexes, with crystal structures docked in.

SAXS is very low resolution, it presents "envelopes". It will give you an estimate of size and rough shapes, not much else.

Correction: Nowadays, an undergrad can go through the process of "solving" a crystal structure by following the programs that automate everything. This is the case in x-ray crystallography, not NMR.

>Is it possible to elucidate on the basis of the sequence alone?
No, but there's work being done using sequence comparisons to help predict the structure.

Love these old websites.

that's sad. 3rd year chem major and taking mat sci classes. crystallography is probably the coolest thing i've learned so far in undergrad.

>Is it possible to elucidate on the basis of the sequence alone?
It is possible. But ab-initio structure prediction only works well for small soluble proteins. There are also websites that provide structure prediction using homologous structures but the results will be more imprecise the more dissimilar the target is from the template. Recently I solved the structure of a plant protein and compared it to the prediction results. The best template had an amino acid sequence similarity of around 30% and while the overall fold was similar the part of the protein that catalyzes the reaction was predicted totally wrong. I guess at some point we will have filled up the PDB with enough structures to have a broad enough range for prediction but for now that is not the case. Also most of these websites like I-Tasser or Swiss-model are made for biologists who want a quick result in the easiest way possible. Waiting several days for a more precise structure is already too much. And don't get me started on installing (usually Linux-based) software yourself, most biologists are too dumb for that.

The interesting proteins are however the huge complexes and membrane proteins. Prediction works quite poorly for them. There are much less structures of membrane proteins deposited in the PDB which of course limits the template based prediction. But also the ab-initio approach is limited since less restrictions are available for the prediction. At some point a reliable prediction will be possible but confirmation of the results will still need to to happen via experiments.

>s it possible to elucidate on the basis of the sequence alone?

There's also some collaborative macrotasks works like fold.it, but it has a lot of limitations.

>There's also some collaborative macrotasks works like fold.it, but it has a lot of limitations.
Fold.it is just a game. Do you mean folding@home?

what plant protein were you working on?

nah, fold.it in the end was supposed to be able to help researchers that could upload protein structures that they wanted resolved and the community would spawn out a prediction, or something like that (don't take my word for it, based on something I read long ago)

I do have one last question, if you don't mind. Would it be best to mutate an amino-acid such that it wouldn't needs PT modifications, or would it be best to find (or, God help me, engineer) a host that would perform these duties? Wouldn't the mutation affect the fold? How useful would a crystal of a mutated protein be for subsequent analysis (especially, say, if glycosylation were needed for oligomer formation, which is still a big problem concerning my protein)?

Ahhhh, piss. So much to consider.

>Would it be best to mutate an amino-acid such that it wouldn't needs PT modifications, or would it be best to find (or, God help me, engineer) a host that would perform these duties?
I have no experience with post-translational modifications, sorry. I guess you will end up trying out both ways anyway. Maybe you're lucky and the path of least resistance will work for you.

>Wouldn't the mutation affect the fold?
It might. It might not. You won't know of you don't do any experiments. I had a project once where the mutation of two adjacent amino acids led to the movement of two large loops.

>How useful would a crystal of a mutated protein be for subsequent analysis (especially, say, if glycosylation were needed for oligomer formation, which is still a big problem concerning my protein)?
That depends on your protein. Does it oligomerize? Is the oligomerization a part of the function or just a by-product of the purification? Maybe it is even better if the oligomerization does not occur.

>Ahhhh, piss. So much to consider.
If your group does not have enough experience with structural biology try to find somebody who does and is willing to help you. Check the question section of Researchgate (if you are willing to sift through the basic questions of all those poo in loo's). And consider going to a workshop with a focus on membrane protein structure elucidation. There is one in march for example: biochem.utah.edu/hill/wcpcw.html That one is about structural biology in general I think.

>I do have one last question, if you don't mind. Would it be best to mutate an amino-acid such that it wouldn't needs PT modifications, or would it be best to find (or, God help me, engineer) a host that would perform these duties?

Well, you wouldn't have to worry about engineering your own strain of host cells for expression. Baculovirus expression systems in insect cell lines (High Five and Sf9) can handle most proteins from eukaryotes.

I would go to one of these expression systems. If you want to express in a prokaryotic system (BL21s or perhaps codon+), then you would have to look into playing around with constructs (chopping it up, etc.).

>Wouldn't the mutation affect the fold?

Certainly, that is a possibility. There are number of ways you can look into this, from computational (secondary structure prediction) to expression (mutant construct solubility) to functional (activity assay) to disorder (NMR, limited proteolysis).

>How useful would a crystal of a mutated protein be for subsequent analysis (especially, say, if glycosylation were needed for oligomer formation, which is still a big problem concerning my protein)?

It comes down to the question you are trying to answer with the crystal structure. Going off your assumption that glycosylation is needed for oligomerization, and further extrapolating that, say, oligomerization is needed for biological activity, then yes, you should present a crystal structure that is glycosylated, and functional assay that connects glycosylated oligomer to biological activity.

If the glycosylation has not effect on your particular system (say, the glycosylation is different from the oligomerization domain), again, you'll need biochemical and cell biological data to show that your construct still has activity.

Thank you both for your considerate answers.

You're welcome, glad to help.

Seconding this. Tell us the protein.

thirding