PolyQ Prions: 4FE8 analysis in R Studio Bio3d

Code is being updated. 

Here we're looking at one of the best resolved polyQ regions in the Huntingtin protein. The resolution of small beta structures was made possible by histidine inserts. Beta turns contribute to nucleation in Huntington's disease, enabling the formation of larger beta structures, and downstream aggregation and propagation. Further analysis of the paper methodology would be updated in this same post.

This is a test-drive of bio3d 2.3-0 in R studio. Since 4FE8 contains multiple copies of HTT in various conformations, I first got familiar with the 3D structure in Samson, and then deleted non-relevant duplicate chains. You can also use the pdbsplit function if you're familiar with the structure. 

Although 4FE8 consists of 410 amino acids, the first 370 amino acids are actually from the maltose binding protein (MBP), which aids in solubility and crystallisation of the structure. The MBP is fused to the N-terminal of HTT. Therefore, the HTT protein itself begins at residue 371 (Methionine), and ends at residue 410 (Glutamine). The entire polyQ region spans 388Q to 410Q, inclusive of 3 H inserts at 395H, 397H and 399H. The proline chain right after the polyQ region is not visible in this structure, but can be seen in full HTT structures such as 7DXJ. The full HTT protein consists of about 3100 amino acids.

To analyse the N-termal + polyQ region, I first extracted the HTT-EX1, removing the MBP, creating a dataframe of 40 amino acids from 371M to 410Q. 

Let's analyse the composition. Glutamine (Q) itself is polar, hydrophilic and soluble, capable of H-bonding. However, long tracts of Q form complex H-bonds with each other, condensing the overall structure in the nucleation process. This increases propensity towards beta conformations that sequester hydrophilic regions, resisting degradation and enhancing aggregation via templating. In bio3d, there is a dm function to calculate distances between atoms in PDB structures. This can be applied to compare normal vs pathogenic HTT structures, provided that the polyQ regions are reasonably well resolved, which is unfortunately not the case. 

Then we look at the secondary structure propensities of each amino acid, according to established experimental scores per amino acid,. Here I analysed it by individual residues. For larger proteins, using tetrapeptide (or other multipeptide) sliding windows would make more sense for a comprehensive picture. You can also use aa.index function in bio3d for more parameters, but today we will analyse it manually. 

This analysis actually reminds me of the PLAAC algorithm, although we're not specifically investigating prion-like propensity here. Since the PLAAC uses a Hidden Markov Model, assigning prion-like scores to each amino acid, you can rebuild the function in R and apply it, adjusting parameters according to latest research. 

Update - We are using experimental propensities here. The previous version of this post confused thermodynamic propensities with structural forming propensities, and mixed in some outdated theoretical values. 

In Pace et al. 1998, the lower the thermodynamic barrier value ΔG, the higher the structural propensity towards helix formation. This ranges from 0 (alanine, reference) to 3.16 for proline (helix breaker).

However in the Minor 1994 paper on beta propensity, the ΔG measured thermodynamic stability of the resultant beta formation. Therefore, the higher the thermodynamic stability, the higher the propensity towards beta formation. This ranges from -3 for proline (beta sheet breaker) to 1.1 for threonine (favours beta sheets). As you can see proline is very disobedient due to its bulky structure, and has a high propensity towards disorder, although it can also form a distinctive polyproline helix. 

The coil propensity is a direct statistical scale. A lower coil propensity indicates favouring structural formation (helix or beta), while a higher coil propensity favours disorder. So the negative values indicate a propensity towards order. There are very small values, though not zero, as amino acids have at least a slight tendency towards either structure or disorder. As expected, proline scores the highest here.

We then normalise all 3 parameters in the same direction for easier comparison. The original helix values are inverted as well to match the direct scale of the other two values. As expected, higher propensities towards both helix and beta formations tend to yield lower coil propensities. 

So here we see the original raw value plots. When reading the helix (blue) plot, lower value = higher helix propensity. 

Normalization makes for easier comparison across structural propensities. This is not a strict quantitative comparison, as statistical scale ≠ thermodynamic scale, so distortions are inevitable. But for a qualitative comparison (ranking), it would work. In general, helix and beta propensities tend to move in opposite directions, though not always, some amino acids have approximately similar propensities towards either. Although helix propensity remains highest for the polyQ region, beta propensity follows closely, which makes sense since polyQ structures are relatively flexible until they reach a pathogenic length and tend towards beta structure. In particular, the relatively high coil index (green) indicates the structural flexibility within this region, facilitating shifting between helix, beta, or coil. This structural flexibility hinders experimental resolution of polyQ structures.

If we analyse the entire protein of MBP + HTT-EX1, 410 a.a., this is what we get. It would be more meaningful to run this composition analysis on full-length HTT proteins, and compare various polymorphisms and the resultant composition changes. 

 · · ─ ·✶· ─ · · 

References 

Kim M. (2013). Beta conformation of polyglutamine track revealed by a crystal structure of Huntingtin N-terminal region with insertion of three histidine residues. Prion, 7(3), 221–228. https://doi.org/10.4161/pri.23807

Campen, A., Williams, R. M., Brown, C. J., Meng, J., Uversky, V. N., & Dunker, A. K. (2008). TOP-IDP-scale: a new amino acid scale measuring propensity for intrinsic disorder. Protein and peptide letters, 15(9), 956–963. https://doi.org/10.2174/092986608785849164

Pace, C. N., & Scholtz, J. M. (1998). A helix propensity scale based on experimental studies of peptides and proteins. Biophysical journal, 75(1), 422–427. https://doi.org/10.1016/s0006-3495(98)77529-0

Minor, D. L., Jr., & Kim, P. S. (1994). Measurement of the β-sheet-forming propensities of amino acids. Nature, 367(6464), 660–663. https://doi.org/10.1038/367660a0