Dear Rosetta@home community,

First off, thank you for donating your time and computational resources to help our research! You may have noticed some new jobs predicting the structures of small proteins; I'd like to tell you a bit about why we're working on these.

Allow me to begin with a bit of background information. Most organisms produce small proteins that are rich in disulfide bonds. Disulfide bonds can "staple" different parts of a protein together, greatly increasing a protein's durability when outside of a cell. In higher organisms, like plants and mammals, these proteins are commonly used as hormones or for defense against infection (these are appropriately named "defensins"). Many insects use these proteins as part of their venom. Sometimes, scientists find that one of these small proteins possesses a serendipitous trait that can be exploited for therapeutic benefit. A great example is chlorotoxin, a component of deathstalker scorpion venom. This small protein binds to cancer cells and not to normal cells. Another Seattleite, Dr. Jim Olson, has attached a fluorescent molecule to chlorotoxin and is developing it as a surgical aid. He calls it Tumor Paint, and it allows surgeons to see cancerous tissue while they're performing a surgery. His work has the potential to revolutionize cancer treatment, and it has gotten quite a bit of press coverage.

Tumor Paint is truly amazing, and I find it equally amazing that scientists were able to discover the one specific component of one specific scorpion's venom that can differentiate cancer cells from normal cells. This story inspired us. What if we didn't have to wait for serendipity to deliver the next breakthrough therapy? What if we could build these marvelous little proteins to do whatever we want them to? With your help, this is exactly what we're trying to do.

The first step towards making this dream a reality is to figure out how to make small, disulfide rich proteins from scratch. This is something no one has ever done before. Our preliminary results are very promising, and we have one confirmed success so far. However, to fully realize the therapeutic potential of this family of proteins, we need to expand the diversity of structures we can make; we need to find what the limits are. Then, we can build them with purpose, custom tailored to a given therapeutic task.

If you're interested, the names for these jobs are along the lines of "0_input_0460_ss1_2_ss2_4_ss3_3_ss4_3_0001". Sorry for the convoluted names; I'll try to better about it in the future.

If you have any questions or comments; I'd be delighted to discuss!

Thanks again,
-Chris

Very cool, many thanks for keeping us updated on the meaning behind the work units

we have one confirmed success so far

Does this mean that you have a novel protein which had it's predicted conformation confirmed in lab tests? ...or what is your definition of success in these early stages of developing these defensins?

---

Rosetta Moderator: Mod.Sense

Great question. I performed a previous round of experiments where I tested the feasibility of this project. I attempted to make a protein with the same overall topology as the scorpion toxin I discussed above, but built completely from scratch. The amino acid sequence and placement of the disulfide staples are totally divergent from any naturally existing protein. After months of optimizing methods for the production and purification of these proteins (the disulfide bonds add difficulty), we have finally managed to produce them in the lab. In an exciting breakthrough, we recently received confirmation of our first design success by means of an X-ray crystal structure, working in collaboration with Colin Correnti and Jim Olson.

Here's a picture of the models aligned:


The proteins are shown as cartoon renderings, the design model is colored green, the best-scoring ab initio structure prediction from Rosetta@home is colored cyan, and the experimentally determined X-ray structure is colored gray. The disulfide staples are shown as sticks, with the sulfur atoms colored yellow.

As always, a big thanks to everyone crunching numbers for Rosetta@home. This work is possible because of you!

Also, for those of you who have been following the progress from Rosetta@home for the past few years, this project is a next step after Nobu and Rie Koga's breakthrough where they discovered a general set of rules for de novo protein design. For more info on this topic, see Nobu's explanation in this thread: https://boinc.bakerlab.org/rosetta/forum_thread.php?id=6113#74206.

Thank you for the details. I just wanted to point out that "this project" in the case you are discussing is just one of many that are all underway within BakerLab and all using the Rosetta@home project as a platform to study proteins, disulfide bonds, and etc.

---

Rosetta Moderator: Mod.Sense

just one of many that are all underway within BakerLab and all using the Rosetta@home project as a platform to study proteins, disulfide bonds, and etc.

Yes, thank you for pointing this out. My project is a small cog in the greater research machine that is the Baker Lab, which is why I referred to my work as "a next step" and not "the next step." However, I'll try to be more explicit in subsequent posts.

Hello and thank you to the Rosetta@home community for providing your time and computing power to our research efforts. As a new postdoctoral fellow in the Baker group, I am working with Chris to explore additional disulfide stabilized small proteins. Specifically, we are designing new structural variations with different numbers and orderings of beta sheets and alpha helixes.

The names of my most recent Boinc jobs are a bit long but designed to give a general overview of the structure at a glance. Each protein is composed of several secondary structure elements (loops, beta sheets/strands, and alpha helixes) listed from the start (N-terminal) to end (C-terminal) of each protein. The ordering of these structures are accompanied by a digit indicating the number of amino acids in that structure. For brevity, secondary structures are coded L=loop region, E=beta strand (which align to form beta sheets as described below), and H=alpha helix.

For example, 1L-5E-2L-10H-2L-5E-1L_1-2.P.0 is a design with two beta strands, each five amino acids in length, and one alpha helix, ten amino acids long. These three structural elements are connected by short loop regions and both ends of the protein are allowed to be flexible as length one loops. Finally, the designation "_1-2.P.0" means that the two beta strand elements, strand 1 and strand 2, will form a sheet with Parallel alignment as opposed to Anti-parallel. I mention this because these are the first designs in which we've included parallel sheets in disulfide stabilized small proteins.

At present, we are validating some of our smallest designs, in the laboratory, which were selected based on Rosetta@home results. These include EHE, EEH, and HEE designs. The jobs you're most likely to see right now are going to be of the EHEH variety, such as 1L-5E-2L-10H-2L-5E-2L-10H-1L_1-2.P.0. Over time, our designs will continue to increase in size as we determine the best way to make stable structures of desired configurations. Eventually, we aim to establish stable scaffolds of various structures which can serve as a starting point for more targeted protein design against specific pathogenic targets.

Once more, thank you for providing the vast computing resources needed to evaluate our many designs and feel free to share any comments/questions.

-Jason

Thanks for the detailed update. The naming convention you've employed makes perfect sense and now we can all understand what types of structures are underlying some of these jobs. I am curious as to the timeframes involved in your research cycles? How long does in-lab verification work, and how long does the BOINC side of things take (in general)? I have friends in research and they tell me tales of watching paint dry sometimes. Just curious. Again, thanks for the update!

Great question Timo! The in-lab verification of these proteins is very labor intensive. Recently, Jason and I worked together to produce and test 48 different designed proteins, all of which he first screened with Rosetta@home. Our preliminary biochemical experiments are very encouraging, and it looks like many of our designed proteins have folded up into the intended conformation. We're currently working with our collaborators at the Fred Hutchinson Cancer Research Center to determine structures for these; hopefully we'll be able to report back with good news soon!

Best regards,
-Chris

This is another "hello and thank you" post! I'm a new graduate student rotating in the David Baker lab, and I'm helping Chris and Jason to explore different types of these small, disulfide-bonded proteins. I just came to UW from Fred Hutchinson Cancer Research Center, where I worked in Jim Olson's lab -- the folks Chris mentioned above who are developing drugs based on naturally-occurring proteins like these. It's fun to step a bit away from what nature has built and see what different types of molecules we can create.

I've started things off by putting out some jobs to try to fold very simple proteins, just a single beta sheet. There are some models for these in nature, but we're trying to see if we can make some that look a bit different -- in particular, the ones I'm designing won't generally have a bunch of tryptophan amino acids, which is common in the versions from nature, and will instead be held together by their disulfide bonds. Here's what a nice-looking example of one of these looks like:

(OK, there will actually be a picture here when I can get it uploaded to the static server)

My jobs are named in the same way as Jason's -- the name indicates the sequence and length of secondary structure elements, in this case strands of the beta sheet.

Soon, I'll put out some more jobs testing out a very different topology -- a beta sheet sandwiched between two alpha helices. I'm looking forward to telling the community about what we discover as we design and express these proteins!

All right! I'm submitting a bunch of designs for the second topology I mentioned, which I'm calling "HEHE" for "helix-strand-helix-strand". The names of the jobs all indicate the lengths of the alpha helix and beta sheet strand elements in the design -- for example:

1L-10H-3L-5E-2L-10H-2L-5E-1L_1-2.P.0_0001

means 1 amino acid (AA) at the N terminus ("start"), then a 10-AA helix, then a loop of 3 AAs, then a 5-AA beta sheet strand, then a loop of 2 AAs, then a 10-AA helix, then a 2-AA loop, then a 5-AA sheet strand, and then 1 AA at the C terminus.

OK, that's pretty dry stuff, I know, but the lengths of those elements will have a big impact on whether they're able to fold, and whether the whole molecule is short enough that we can order them in batches of thousands of DNA sequences.

Before I kicked these off, I made sure we had some great-looking designs in here. The ones we're particularly interested in look like a sandwich, with helices as the bread and a two-strand beta sheet as the meat. I promise cool-looking pictures when I can get some images uploaded!

OK, here's a pretty picture of one of the single-beta-sheet designs I mentioned above. The ab initio folding results from the best of these look really good, and I'm getting ready to express 12 of those proteins right now!