Closed since about 1 year ago
Novice Overall Prediction Electron DensityWe are very excited about this puzzle, because while our collaborators have been able to fit a cryoEM map well enough to build an atomic model of this channel (chain A in this puzzle) they have been unable to fit the toxin variant (chain B). We can't wait for you to fit both chains into this cryo-EM map, but especially residues 119-148 (chain B). Please read this blogpost for all the juicy scientific details!
Read all the juicy scientific details in the blogpost: https://fold.it/forum/blog/new-unsolved-vgic-cryo-em-density-puzzle
100 pts.
18,608
This puzzle is open for 2 weeks to give you enough time to solve this Unsolved VGIC Cryo-EM Density Puzzle!
Note that there is no Refine Density tool in this puzzle, because there is no way to refine this density… the electron density cloud comes from cryo-EM, not X-ray crystallography, so we only have this one "image" of the density to work with. Good luck!
Comparing the sequence from Puzzle 2541 with that from https://www.rcsb.org/structure/6N4R is confusing.
I think Puzzle 2541's segments 119-148 are supposed to be segments 1-30 of protoxin-2 (PTx2).
At least the cysteines for both seem to agree (Puzzle 2541's segments 120 127 133 134 139 143
and PTx2's segments 2 9 15 16 21 25).
I think Puzzle 2541's segments 1-118 are supposed to be segments 717-834 in the pdf file for 6N4R.pdb
(https://doi.org/10.1016/j.cell.2018.12.018 or https://www.cell.com/action/showPdf?pii=S0092-8674%2818%2931632-5).
I think this because the sequences for Nav1.7 VSD2 on this pdf's p.22 supp fig E and p.28 supp figs A & B seem to match
many of segments 1-118 in Puzzle 2541.
It seems like Puzzle 2541's starting structure's segments 119-148 form the 3 desired disulfide bonds,
but for segment range 1-118, there are 2 more cysteines (segments 9 and 26) that start with no disulfide bond.
Do these 2 cysteines form disulfide bonds to cysteines in other parts of the protein not included in Puzzle 2541?
https://fold.it/forum/blog/new-unsolved-vgic-cryo-em-density-puzzle says Voltage-gated ion channels (VGICs) are membrane proteins. Does this mean that the protein in Puzzle 2541 could have 3 different zones: (1) intracellular, (2) trans-membrane, and (3) extracellular. Should parts of the protein in zones (1) & (3) have their hydrophobic (orange) residues in the protein interior while their hydrophilic (blue) residues lie on the outermost surface? Should parts of the protein in zone (2) have their hydrophilic (blue) residues in the protein interior while their hydrophobic (orange) residues lie on the outermost surface? Do the force fields in Puzzle 2541 account for these 3 different zones?
https://en.wikipedia.org/wiki/Disulfide#Occurrence_in_proteins says that "since most cellular compartments are reducing environments, in general, disulfide bonds are unstable in the cytosol", and https://en.wikipedia.org/wiki/Disulfide#In_eukaryotes says that "disulfide bonds are mostly found in secretory proteins, lysosomal proteins, and the exoplasmic domains of membrane proteins." Do these statements mean that we should only expect disulfide bonds to form in zone (3), the extracellular zone, in Puzzle 2541?
Can you tell us which segments in Puzzle 2541 you expect to be in each of zones (1)-(3)?
Thanks!
Hi Jeff,
Thank you for your insightful and thoughtful questions regarding Puzzle 2541. It’s great to see your interest in unraveling the complexities of this VGIC cryo-EM density puzzle. I’ll address each of your points below in detail.
Sequence Discrepancy Between Puzzle 2541 and PDB Structure 6N4R:
The mismatch you’ve identified between the sequences in Puzzle 2541 and the 6N4R structure is indeed expected. The experimental structure you refer to, which we used as a basis to redesign and optimize protoxin-2, is not the sequence of the full human Nav1.7 channel. Instead, it is a chimeric construct: the top region of the voltage-sensing domain (VSD2) comes from the human Nav1.7 channel, while the rest is derived from a bacterial sodium channel (see this paper https://pubmed.ncbi.nlm.nih.gov/30661758/, figure S1, panel D). This approach of creating chimeric constructs with simpler bacterial Nav channels has been used because obtaining structures of full human Nav channels has traditionally been very challenging due to their large size.
In contrast, the cryo-EM density in Puzzle 2541 corresponds to the full human Nav1.7 sequence. Additionally, the protoxin-2 sequence in the puzzle has been optimized for selectivity, which is why it does not match the native protoxin-2 sequence. While the chimeric construct was instrumental in designing the toxin variant, the puzzle now reflects the complete human Nav1.7 sequence, explaining the differences you observed.
Disulfide Bonds in Puzzle 2541:
You’re correct that the three cysteines in the toxin sequence are expected to form disulfide bonds, as reflected in the starting structure. The two cysteines in the channel sequence (positions 9 and 26) do not form disulfide bonds with any other parts of the protein, so there is no need for concern here.
Membrane Environment Considerations:
Great catch! You’re right that VGICs are membrane proteins, and the protoxin-2 toxin partially embeds into the membrane. Ideally, the energetic scoring in Foldit would account for the hydrophobic residues being more stable in the membrane and hydrophilic residues favoring the aqueous extracellular or intracellular environments. However, the current Foldit energy framework does not include a force field or scoring function that considers the presence of a membrane. While software like Rosetta includes membrane energy functions (see https://pubmed.ncbi.nlm.nih.gov/32224301/), this has not been implemented in Foldit. Nevertheless, your attention to these factors as a player is appreciated and aligns well with real-world membrane protein behavior.
Zones and Disulfide Bond Formation:
Your observation about disulfide bonds primarily forming in extracellular zones is spot on, as these bonds are generally unstable in reducing environments like the cytosol. In Puzzle 2541, the extracellular zone is indeed where the toxin variant binds, and only the three disulfide bonds within the variant are relevant here. The two cysteines in the channel sequence are not involved in disulfide bond formation.
Segment Zoning in Puzzle 2541:
To clarify the zones:
• Extracellular Zone: This is where the toxin binds. The extracellular zone can be approximated as the region above a plane formed by residues 51, 37, 97, and 102.
• Transmembrane Zone: This region spans the membrane interface and includes residues embedded in the lipid bilayer. The toxin partially embeds into this region.
• Intracellular Zone: This is the bottom part of the VSD, representing residues closer to the cytoplasmic environment.
While these distinctions are critical for structural biology and protein design, the Foldit puzzle does not currently account for these specific environmental factors.
Thank you again for your questions and for diving deep into the complexities of this puzzle. Best of luck, and feel free to reach out if you have more questions or need further clarifications!
Warm regards,
Diego
Wow Diego. Thanks for all of the information. I think it will help.
The Trim Tool has now been added to this puzzle.
Here is how to use it: https://foldit.fandom.com/wiki/Trim
You will need to restart your Foldit client for it to show up.
As I play Puzzle 2541, I wonder if some parts of the ED cloud come from atoms that aren't part of the protein. For an ion channel, you might expect nearby or internal water molecules plus ions like Na+, K+, Ca++, and Cl- to contribute to the ED cloud. For a membrane protein, you might also expect nearby molecules from the membrane to contribute to the ED cloud. Either of these cases would give parts of the ED cloud that won't be filled by Puzzle 2541's protein segments 1-148 alone.
As an example, the somewhat-related pdb file 6N4I.pdb starts with:
HEADER METAL TRANSPORT 19-NOV-18 6N4I
TITLE STRUCTURAL BASIS OF NAV1.7 INHIBITION BY A GATING-MODIFIER SPIDER
TITLE 2 TOXIN
COMPND MOL_ID: 1;
COMPND 2 MOLECULE: NAV1.7 VSD2-NAVAB CHANNEL CHIMERA PROTEIN;
COMPND 3 CHAIN: A, B, C, D;
COMPND 4 ENGINEERED: YES;
COMPND 5 MOL_ID: 2;
COMPND 6 MOLECULE: BETA/OMEGA-THERAPHOTOXIN-TP2A;
COMPND 7 CHAIN: E, F, G, H;
COMPND 8 SYNONYM: BETA/OMEGA-TRTX-TP2A,PROTX-II,PT-II,PROTOXIN-2,PROTX2;
COMPND 9 ENGINEERED: YES
SOURCE MOL_ID: 1;
SOURCE 2 ORGANISM_SCIENTIFIC: HOMO SAPIENS;
SOURCE 3 ORGANISM_TAXID: 9606;
SOURCE 4 EXPRESSION_SYSTEM: TRICHOPLUSIA NI;
SOURCE 5 EXPRESSION_SYSTEM_TAXID: 7111;
SOURCE 6 MOL_ID: 2;
SOURCE 7 ORGANISM_SCIENTIFIC: THRIXOPELMA PRURIENS;
SOURCE 8 ORGANISM_COMMON: PERUVIAN GREEN VELVET TARANTULA;
SOURCE 9 ORGANISM_TAXID: 213387;
SOURCE 10 EXPRESSION_SYSTEM: SYNTHETIC CONSTRUCT;
SOURCE 11 EXPRESSION_SYSTEM_TAXID: 32630
KEYWDS SODIUM CHANNEL, TOXIN, GATING-MODIFIER, VOLTAGE-GATED, METAL
KEYWDS 2 TRANSPORT
EXPDTA X-RAY DIFFRACTION
AUTHOR H.XU,C.M.KOTH,J.PAYANDEH
and later lists many lines containing "HET" or "OU", like:
HETNAM 6OU [(2~{R})-1-[2-AZANYLETHOXY(OXIDANYL)PHOSPHORYL]OXY-3-
HETNAM 2 6OU HEXADECANOYLOXY-PROPAN-2-YL] (~{Z})-OCTADEC-9-ENOATE
FORMUL 9 6OU 16(C39 H76 N O8 P)
and:
ANISOU 8316 CE3 TRP H 30 22631 21474 28179 32 633 84 C
ATOM 8317 CZ2 TRP H 30 97.265 201.143 212.000 1.00168.74 C
ANISOU 8317 CZ2 TRP H 30 19971 18768 25372 184 698 -38 C
ATOM 8318 CZ3 TRP H 30 95.296 201.746 213.259 1.00174.22 C
ANISOU 8318 CZ3 TRP H 30 20585 19508 26101 129 686 181 C
ATOM 8319 CH2 TRP H 30 96.684 201.612 213.145 1.00165.14 C
ANISOU 8319 CH2 TRP H 30 19468 18370 24906 204 714 119 C
TER 8320 TRP H 30
HETATM 8321 C08 6OU A1001 97.253 256.111 231.579 1.00 92.47 C
ANISOU 8321 C08 6OU A1001 11866 12323 10943 773 -764 29 C
HETATM 8322 C09 6OU A1001 96.010 256.741 232.204 1.00 83.88 C
ANISOU 8322 C09 6OU A1001 10838 11209 9826 845 -728 62 C
HETATM 8323 C10 6OU A1001 94.891 256.969 231.191 1.00 68.53 C
ANISOU 8323 C10 6OU A1001 8909 9221 7907 809 -706 99 C
HETATM 8324 C11 6OU A1001 93.659 257.602 231.820 1.00 68.23 C
ANISOU 8324 C11 6OU A1001 8921 9162 7842 882 -669 127 C
containing atomic coordinates for extra non-protein molecules (here membrane phospholipids?) in the X-ray Diffraction (ED) structure.
Do you think the cryo-EM ED cloud data we are using in Puzzle 2541 contains any contributions from atoms besides the protein segments 1-148 that we can model with Foldit? If there are immobile water molecules hydrogen-bonded to parts of our protein, what would they look like in a cryo-EM ED cloud? Also, what would immobile ions like Na+, K+, Ca++, or Cl- look like in a cryo-EM ED cloud? Did you filter the raw cryo-EM ED cloud data to remove all contributions from nearby non-protein atoms before including this data in Puzzle 2541?
Thanks for reading.
jeff101, I'd guess the protein just wasn't properly crystallized or wasn't stable enough during the imaging.
Hi Jeff,
Thank you again for your insightful question! Let me clarify the possibilities of non-protein contributions to the electron density (ED) cloud in Puzzle 2541 and how cryo-EM works in this context.
First, you’re correct that this is a membrane protein and an ion channel, so it’s reasonable to wonder if components like phospholipids, ions, or water molecules might be part of the ED cloud. Here’s a breakdown:
Ions: Although this is a sodium channel, Puzzle 2541 is a trimmed version of the full channel, focusing exclusively on the Voltage Sensing Domain 2 (VSD2), which is a regulatory peripherical domain. Since this puzzle does not include the central conduction pathway (named Pore Domain) where sodium ions would typically bind or pass through, sodium ions are not represented in the puzzle’s ED cloud.
Phospholipids: Since this is a membrane protein, nearby lipids might interact with the protein. However, in cryo-EM, the electron density is generated by averaging thousands of micrographs of the protein. Lipids are typically too dynamic and and fluid (establishing short-lived interactions with the protein ) to appear consistently across all these images, so they rarely contribute recognizable density. Exceptions occur when specific lipids interact stably with the protein (e.g., regulatory lipids).
Water Molecules: Similarly, most water molecules move too freely to appear in the ED cloud. However, water molecules that form stable and long-lived hydrogen bonds with the protein could create consistent density and might be visible in the final map.
The cryo-EM map used in Puzzle 2541 has been refined and filtered based on proximity to the atomic model, but it’s not completely stripped of potential non-protein contributions. However, in Foldit, the focus is exclusively on fitting the protein structure, so these putative extra densities aren’t explicitly modeled or considered, although might affect the scoring. However, if you notice any density that seems unaccounted for—such as regions where no side chain fits well and that might correspond to water molecules or other non-protein atoms—please feel free to share it in this forum, this type of input is highly valuable to us as researchers!
It’s really great that you’re considering the broader biological context of this puzzle! it reflects the real-world complexity of membrane proteins and highlights how such components (lipids, water, etc.) can be critical in actual structural biology research.
Thanks again for engaging so thoughtfully with the puzzle, and happy folding!
Best regards,
Diego