The Key to the Extraordinary Thermal Stability of P. furiosus Holo-Rubredoxin: Iron Binding-Guided Packing of a Core Aromatic Cluster Responsible for High Kinetic Stability of the Native Structure

Pyrococcus furiosus rubredoxin (PfRd), a small, monomeric, 53 residues-long, iron-containing, electron-transfer protein of known structure is sometimes referred to as being the most structurally-stable protein known to man. Here, using a combination of mutational and spectroscopic (CD, fluorescence, and NMR) studies of differently made holo- and apo-forms of PfRd, we demonstrate that it is not the presence of iron, or even the folding of the PfRd chain into a compact well-folded structure that causes holo-PfRd to display its extraordinary thermal stability, but rather the correct iron binding-guided packing of certain residues (specifically, Trp3, Phe29, Trp36, and also Tyr10) within a tight aromatic cluster of six residues in PfRd's hydrophobic core. Binding of the iron atom appears to play a remarkable role in determining subtle details of residue packing, forcing the chain to form a hyper-thermally stable native structure which is kinetically stable enough to survive (subsequent) removal of iron. On the other hand, failure to bind iron causes the same chain to adopt an equally well-folded native-like structure which, however, has a differently-packed aromatic cluster in its core, causing it to be only as stable as any other ordinary mesophile-derived rubredoxin. Our studies demonstrate, perhaps for the very first time ever that hyperthermal stability in proteins can owe to subtle differences in residue packing vis a vis mesostable proteins, without there being any underlying differences in either amino acid sequence, or bound ligand status.

: Panel a : Control solutions lacking ferrozine. Tube-1 contains a ferric chloride solution. Tube-2 contains ferric chloride and beta-mercaptoethanol. Tube-3 contains ferric chloride, beta-mercaptoethanol and guanidium hydrochloride. Panel b : Sample solutions containing ferrozine. Tube-1 contains ferrozine added to ferric chloride. Tube-2 contains ferrozine added to ferric chloride pre-mixed with betamercaptoethanol. Tube-3 contains ferrozine added to ferric chloride pre-mixed with beta-mercaptoethanol and guanidium hydrochloride. Tube-4 contains ferrozine added to ferrous sulphate. Panel a : Elution of free ferrozine (~17 ml and ~20 ml) on a Superdex Peptide (GE) column in the absence of any iron or protein. Panel b : Fe 2+ -bound ferrozine (two eluting species at 12.5 ml and 14.0 ml) separated from free ferrozine (~17 ml) on the same column. Panel c: Fe 2+ -bound ferrozine (~12.5 ml and ~14.0 ml) separated from free ferrozine (~17 ml and 20 ml) and PfRd protein (~10 ml) on the same column. Panel d: Elution of PfRd protein (~10 ml) on a Superdex Peptide (GE) column in the absence of any ferrozine.

DISCUSSION OF FIGURE S3
Confirming that Apo-2 PfRd does not support the formation of any disulfide bonds amongst any of the four cysteines involved in holding the iron atom, once the iron atom is no longer present (and has been removed).
Any formation of disulfide bonds amongst PfRd's cysteine residues could call into question the detailed analytical reasoning and scenario that appears to be emerging from our examinations of the structural-biochemical behavior of Apo-1 PfRd, Apo-2 PfRd and holo-PfRd. Therefore, we decided to examine, in particular, Apo-2 PfRd, the second of the two forms of PfRd which was generated and examined for the very first time in this paper (since Apo-1 PfRd has already been characterized by other authors, albeit using naturally-sourced protein). Our basic approach was to first examine the molecular weight of intact Apo-2 PfRd. This is shown in Supplementary Figure  S3a.
Next, as a control, we assumed that there could be some disulfides formed upon departure of the iron atom, in Apo-2 PfRd, and so we first treated it with beta-mercaptoethanol (to reduce disulfides) and then react the cysteine residues with an alkylating agent, iodoacetic acid (IAA). IAA is expected to form adducts with the sulphydryl groups of cysteine residues by displacing any cross-disulfide species involving beta-mercaptoethanol. The mass spectrometric data involving this treated protein is shown in Supplementary Figure S3b.
Finally, we alkylated an Apo-2 PfRd population that had not been pre-treated with betamercaptoethanol, so that only free cysteine residues would be seen to react with IAA. The mass spectrometric data for this population is shown in Supplementary Figure S3c. Briefly, the idea was that each alkyl group would increase the mass of PfRd by 57 Da, and that a comparison of the beta-mercaptoethanol unreacted, and reacted, populations would reveal whether any cysteines were non-reactive, either on account of being disulfide-bonded, or being buried within the structure and (therefore) inaccessible to the alkylating agent. To facilitate exposure of all parts of the protein to the alkylating agent, IAA, the alkylation was done at 90 degrees Centrigrade in the presence on 6 M Gdn.HCl, such that the combination of denaturant and temperature would achieve some unfolding of the protein where individually either denaturant or temperature were incapable of doing so.
The result was that with the Apo-2 PfRd population that was never exposed to betamercaptoethanol [Supplementary Figure S3c], we were able to see that the population is dominated by species in which all four cysteine residues were accessible to IAA (and thus not engaged in disulfide bonds). In addition, minority populations with three, two and one cysteine alkylated were also seen as would be expected in any such reaction with a protein with hyperthermostable structure. In the mercaptoethanol-treated population too [Supplementary Figure S3b   PfRd alkylated by iodoacetic acid (IAA) without any treatment with beta mercaptoethanol. The masses of ~7523.95, ~7462.27, ~7406 and ~7345 Da represent species carrying four, three, two, and one IAA aductions, respectively, with the mass peak with the highest intensity representing the population with all four of PfRs's cysteine residues modified. The molecule's cysteine residues are thus free and available to be alkylated (and not disulfide bonded).

Structural examination of the stacking, orthogonal and other ring-ring interactions amongst the six aromatic residues of holo-PfRd
We conducted a detailed and thorough examination of the aromatic interactions in the aromatic cluster in the core of holo-PfRd. Fig. S4 displays all of these interactions in a pair-wise fashion, and also in groups of two or three residues.

Y10 and Y12 contact each other (Panel f in
F48 interacts with both Y10 and W36 (Panel g in Fig. S4).
Therefore, F29 is directly in contact with W3, W36 and F48.
No other aromatic residue forms such a huge number of interactions as F29 and W36.
It is clear that F29 and W36 act as key players for this aromatic cluster, almost like the central glue which keeps the cluster together. Organization of the aromatic cluster in holo-PfRd, showing different aromatic interactions amongst the molecules six aromatic residues, namely W3 (green), Y10 (orange), Y12 (magenta), F29 (blue), W36 (red) and F48 (black).

Examination of changes in the structural characteristics of holo-PfRd and different aliphatic and aromatic residue substitution mutants.
The near-UV and far-UV CD spectra of holo-PfRd and all the variants produced by alanine (or serine) substitute of selected aliphatic and aromatic residues were examined, presented and discussed in Figure 10 in the main paper.
Results from the examination of the corresponding protein species in respect of changes in structural characteristics are presented in Fig. S5, both in respect of proteins heated in the presence and in the absence of denaturant.
Panels a, c and e, respectively, in Figure S5, Collectively, the data reveal the following points : • Neither holo-PfRd nor any of the aliphatic substitution mutants undergoes any significant unfolding upon heating in the absence of denaturant (panel a). Only mutant I23A loses its additional negative MRE upon heating, such that at 95 °C all variants and holo-PfRd have comparable MRE values. The aromatic substitution mutants have different MRE values to begin with (as discussed in the main manuscript); however, these two do not undergo any changes with heating (panel b). • Holo-PfRd and all of the aliphatic substitution mutants undergo some thermo-chemical unfolding in the presence of 6 M Gdm.HCl (panel c). With the aromatic substitution mutants, the data is different for different mutants. W3A and Y10A show thermo-chemical unfolding like holo-PfRd. However, W36A and F29A, which have much lower ellipticity to start with, start developing higher negative values of MRE upon heating, while with Y12A and F48A, there is an initial drop in the negative MRE signal, followed by an increase in the negative MRE value (panel d).