The work described in this section is the result of collaboration between Tammer Farid and the author.
HP7 advanced the groundwork set by the uniquely structured holo-HP1. It not only represented an advance in the topological construction of the primary sequence, but, in addition, represented the first four-helix bundle maquette in the Dutton lab to be solely produced by expression given the length of the individual monomers (though more recent maquettes of similar length have been synthesized). However, for all of its benefits, HP7 cannot be used directly for incorporating a single site mutation or covalent modification due
to the homodimeric nature of its structure, and this certainly precludes, without resorting to heterodimeric disulfide bridged helix-loop-helix monomers, incorporation of a single Naq residue. Rather, if possible, it would be much preferable to have a single chain variant of HP7. In work undertaken simultaneously as that presented here, Bender et al. describe the creation of a single chain dual iron(III) porphyrin binding de novo designed four-helix bundle protein with a significantly different cofactor binding architecture, though nearly identical topology, than HP7.11
5.2.1 N-Cap design
In the NMR of HP7, it was clear that the N-termini of the monomers were not structured. The same sequence on the N-terminal end of the second helix in the helix-loop- helix monomer was however structured (Ron Koder, personal communication). To further advance solution structure determination by clarifying the NOE constraints of the N- terminal residues, Ron Koder added a non-ideal Ser-Pro sequence to the beginning of the HP7 sequence producing HP8 (Ron Koder, personal communication), Table 5.1. This had the undesired effect of eliminating one of the heme binding sites (data not shown, collected by T. Farid). To test the more ideal N-capping motif TPEQ,12 BT2 was synthesized as a single helix of HP7 with this N-capping motif in place, Table 5.1, and a C-terminal loop ending at a Cys for covalent cross linking. Presumably forming dimers of cystine-crosslinked dimers BT2 demonstrated full two-heme binding, suggesting the ideal capping sequence TPEQ would not interfere with heme binding in a single chain HP7 variant. Since it was observed that the glycine-serine loops in HP7 were sufficient to cap and stabilize the N-
terminal end of the second helix in the monomer, only one N-terminal TPEQ sequence was needed in the single chain variant.
5.2.2 Cross Bundle Loop Design
The nine residue loops of HP7 are much longer than would be expected to link two adjacent helices.13-15 However, in the development of HP7 it became clear that shorter loops increased the amount of higher order oligomeric states of heme bound maquettes as observed by size-exclusion chromatography (Ron Koder, personal communication), which confirmed an earlier report that suggested shorter loops would be disruptive to the four- helix bundle fold,15 although in that case the variation observed was in loops 3-5 residues long. The need for loops of nine residues belies the fact that the N-terminal and C-terminal ends of the Dutton Laboratory maquettes are still very much identical to the original design and have not been optimized for linking via loops in this manner, especially considering the need to not interfere with cofactor binding. Indeed, the glycine-serine loops in HP7 were designed to be as minimally perturbing as possible so as not to develop a set structure that might interfere with assembly or cofactor binding. Since there was no need to keep the cystine-crosslinks in the single chain HP7 variant, the central cysteine in the found in the HP7 loops was replaced by a homologous serine residue that had been previously found not to globally change the properties of the heme bound HP7C38S.2 Additionally, it was envisioned that the cross-bundle loop (from helices 2 to 3), Figure 5.2C, would need to be longer than the loops found in HP7 so the cross-bundle loop length was increased to eleven residues by the addition of two glycines around the central serine in the HP7C38S loop, Table 5.1. This design process contrasts with that used by Bender et al. in the de novo redesign
of a dual iron(III) porphyrin binding single chain four-helix bundle.11 Using a database of natural interhelical loop structures, the DeGrado group were able to find a loop that best matched (r.m.s.d.) the positioning of at least five Cα backbone atoms at the loop-helix interface of a computationally optimized backbone. This reveals an advantage that de novo
design methods has over the maquette iterative design procedure in that computational approaches do not depend upon the determination of a protein structure when a backbone target is produced during the design process, thus making it easier to systematically design loops and other structures to stabilize a goal structure.
5.2.3 Addition of an N-Terminal Methionine
To allow direct expression of the new single chain maquette in E. coli for use in the intein work detailed in Chapter 6, an N-terminal methionine was appended, giving the final sequence of the single chain HP7 variant, BT3, Table 5.1.
5.2.3 Breaking the Symmetry of HP7
Addition of the cross-bundle loop between helix 2 and helix 3, Figure 5.2C, generates the possibility of two topologically distinct, but nearly structurally identical BT3 conformations. The same conformational rearrangement in HP7 is possible, but it produces a protein that has an indistinguishable topology. This level of symmetry is broken by the addition of the cross-bundle loop. It is anticipated that at least with early BTX designs, this new asymmetry will have limited effects on heme cofactor chemistry since no difference in properties was observed between the non-identical sites in HP7 with histidine knockout
mutants. However, it should be clear from Figure 5.2C that positioning an amino acid like Naq that introduces further asymmetry into BT3 can cause some differentiation between the symmetric pair in a manner that may complicate analysis. To avoid significantly different behavior from the two topological arrangements of helices in such pseudo-symmetric bundles, mutations are effectively limited to pseudo-symmetric positions that are not in different environments with heme bound. Essentially, until one topology can be forced to be favored this requires placement of single site mutations at positions that should remain solvent exposed regardless of topology. The caveat to this design consideration is that mutations that are designed to break the remaining symmetry and effect differentiation of the heme sites will more than likely be found at positions that vary significantly in their environments between the two topologies.