I refer to this so often-- I need to make a name for it.
A name for how Creationists are always parading about saying protein structures are so perfect and pretty colorful Lego blocks, and they all snap together perfectly, but reality is SO not like that? Reality is a buzzing, floppy, invisible mess?
Invisible is the key word here-- Scientists cant look into their microscopes and watch 'Inner Life of a Cell' in real-time.
Figuring out what protein structures 'look like' is really damn hard.
Example #1: VSV, Vesicular Stomatitis Virus? It is like, and old-school virus. Old-school. Like, 1920s.
We just figured out what it really looks like.
Example #2: We have antiretrovirals that work against the retroviral protein integrase. We have no idea how these drugs work, because we have no idea what integrase looks like. Because we dont know how they work, we dont know how to make them better. We dont know how HIV-1 might evolve resistance, to anticipate changes before they happen.
Were just sorta bumbling around in the dark.
Why?
Well, we just cant 'get' the structure of HIV-1 integrase. Tried a whole bunch of approaches, and they just dont work. So, researchers tried to figure out the structure of other retroviral integrases in the hopes these other integrases look something like HIV-1s. They found one that worked, yay! We have got 'a' structure for 'a' integrase! But its not HIV-1 integrase. Popular science journalism has failed completely at noting the importance, the challenges, of using a different virus:
Human foamy viruses (or, prototypic foamy viruses) are a kind of retrovirus, and a kinda retrovirus.
What I mean by that, is that they have the genomic arrangement of a retrovirus (LTRs, gag, pol, env), but they dont behave the same way. Like, HIV-1 is a diploid (+) ssRNA virus. HFV is basically a haploid double-stranded DNA virus-- it reverse transcribes most of itself before it buds out, like Hepatitis B.
Furthermore, pol? Its got some mRNA splicing going on-- most retroviruses transcribe gag and pol together, and they get chopped up into the appropriate proteins (cupcakes vs sheet cake).
And when Ive compared sequences of HIV-1 integrase and HFV integrase... they dont really look alike. I mean, I just played around on Genebank and lined some sequences up, but they dont look alike to me (correct me if Im wrong!). This doesnt mean a 'lot', as viruses are known for having similar structures with apparently no sequence similarity, but Im kinda annoyed at this journalism fail:
99.9% of that article is about HIV-1 integrase. We dont know the 3D structure of HIV-1 integrase. They figured out the 3D structure of HFV integrase, which is only mentioned in passing in that article:
The researchers grew a crystal using a version of integrase borrowed from another retrovirus very similar to its HIV counterpart.
Now, let me be clear, this is really really cool:
Retroviral intasome assembly and inhibition of DNA strand transfer
Theyve got an integrase structure, and we didnt have that before. Also, integrase inhibitors that work against HIV-1 integrase, work against HFV integrase, so their putative active site structure is right.
But this:
The Imperial and Harvard scientists said that having the integrase structure means researchers can begin fully to understand how integrase inhibitor drugs work, how they might be improved, and how to stop HIV developing resistance to them.
is not necessarily true.
Example: We 'know' the structure of HIV-1 envelope, Subtype B. Subtype C envelope looks a lot different. We dont have the structure for Subtype C, we just know because 'variable' regions of Subtype B can be relatively constant in Subtype C. Different parts of the protein are exposed to the immune system, so they evolve differently. Evolution tells us the structures are different.
So its going to be very hard to infer putative evolutionary pathways from a structure that might be, even slightly, different.
Again, 100% cool research, but time will tell if this structure is great, 'good enough', or not really helpful at all for using evolution to improve antiretrovirals.
- Log in to post comments
So we don't know the 3-D design of HIV-1 integrase. Thanks a lot, Designer!
As you said, high sequence divergence in viral genes is nothing new, and it makes alignments tricky. That said if you can get a reasonable alignment and you have a structure for a homolog than threading and modelling at least let you make a reasonable model for the HIV-1 Integrase. Its not perfect but it is a big help. We've been able to get lots of relevant data about how a protein works based on the crystal structure from really distant homologs with very low sequence similarity in our work.
A follow up after looking in the PDB, at least there are several structures all complexed with different metal ions and inhibitors which is fantastic. Unfortunately the crystal resolutions are all pretty low.
Ignoarrogance? Arroignorance?
Brownian spaghetti?
Hmm...How about calling them Knex instead of Legos? There are floppy spaghetti-like ones, and the joins can be setup to rotate.
The press headlines for this story ("Mysteries of HIV solved" and stuff like that) could give the average newspaper reader the idea that the entire problem of HIV/AIDS itself had been solved. I half expected my relatives to call me and ask if I was out of a job yet.
I guess it's hard to strike the right balance of getting people excited about science and dispensing accurate information, and too often science journalists go for the sensationalism without the facts. PR offices at universities don't help.
I like to think of it as a sub-type of Behe blunder.
Where is the difficulty in crystalizing this protein ?
It isn't too big (32kDa), and it shouldn't be membrane-bound, unless I'm mistaken. Is it that it's hard to produce and purify ? Or it just won't produce diffraction-quality crystals for some reason?
As for using the homolog to develop drugs... Long shot indeed. You can develop homology models, but their quality depends on the level of homology. And even with highly homologous protein, you can get nasty surprises.
The wiki article mentions the paper in the same uncritical way, and the person who has written it doesn't seem to understand that several integrases exist, and that the one that was solved isn't HIV integrase.
"Where is the difficulty in crystalizing this protein ?
It isn't too big (32kDa), and it shouldn't be membrane-bound, unless I'm mistaken. Is it that it's hard to produce and purify ? Or it just won't produce diffraction-quality crystals for some reason?"
It likes to clump up in massive aggregates, IIRC. Makes sense if the active form is a dimer/tetramer like they say it might be. However, they do have a structure for the domain that contains the inhibition site that raltegravir and other integrase inhibitors use.
http://www.nature.com/emboj/journal/v20/n24/full/7594220a.html
I think I heard about someone getting a full crystal structure by modifying integrase to prevent aggregation. Of course since they modified it you can't really know how good the structure is.
No doubt, structure of HIV integrase in complex with DNA and inhibitors would be the best thing to have (and remains an important goal!). But while they are trying to get there, a structure of ANY retroviral integrase-DNA complex is very valuable. Because all retroviral integrases are very similar in their essential domain makeup (NTD, CCD, CTD) and serve identical functions, having just one such structure allows us to built a very realistic model for any other retroviral integrase.
It would be worth mentioning that the active sites of PFV and HIV-1 integrases are identical. Of course one had to anticipate them to be similar, all experience in structural biology would suggest that⦠But they did not have to be identical, so there was an element of luck. In fact, all the bits of the PFV IN-DNA complex that are in direct contact with the drugs are present in HIV-1!
Having access to PFV intasome coordinates (freely available from RCSB), anyone even with modest experience in homology modelling can now build a crude model for HIV integrase-DNA-drug complex. With some more experience, effort and additional tools it is possible to improve such models for use in actual drug development. In this sense, the "HIV integrase mystery" has indeed been solved. One can also think of using such models to help identify changes (mutations) to HIV integrase, which would make it more cooperative in crystallisation experiments (!).
P.C.-- I totally agree this is super-cool science! But with the problems weve had with the 'structure' of envelope, Im not overly optimistic, yet, that HFV integrase will transition into understanding/predicting putative evolutionary integrase-inhibitor escape pathways.
One can also think of using such models to help identify changes (mutations) to HIV integrase, which would make it more cooperative in crystallisation experiments (!).
THIS is what Im hoping for :)