The AHABh-motif is structurally and functionally significant

The AHABh-motif (strands 1-2, helices C-D) encompasses both of the structurally efficient fold-motifs seen in PglC: a hydrogen bonding network, and structurally and functionally critical proline kinks. The tertiary fold of the AHABh- motif is established through a network of hydrogen bonds stabilizing the close interaction between β-strand 2 with the full length of helix D (Figure 3.14). The C- terminal end of helix D is held in close proximity to the β-hairpin through interactions between both the backbone amide and tyrosyl hydroxyl of Tyr58 to the side chain carbonyl of Asn100 and Lys103 ammonium group, respectively. Similarly, N-terminus of helix D is stabilized within the AHABh-motif through a hydrogen bond network formed between Asp93, Phe60, and Thr62.

Figure 3.14. The AHABh(alpha-helix-associated beta-hairpin)-motif that defines the superfamily fold is formed by a β-hairpin comprising β-strands 1 and 2

packing against helix D.

Quantitative analysis of helical geometry confirms a kink formed by Pro96 in concert with residues with low helix propensity (89) (Ser91, Asp93, Glu94) distorts the !-helical geometry of the N-terminal end (residues 92-96) of helix D into 310-helix geometry (figure 3.15; Table 3.2). The 310-helix geometry is defined by the amide N-H group of each amino acid forming a hydrogen bond with the amide carbonyl group of the amino acid residues three residues prior in sequence

(90, 91). Thus, each turn of the helix formed by the hydrogen bond is comprised of ten atoms, making it a more tightly wound helix than a canonical α-helix. As a result, the 310 geometry is considered relatively unstable in comparison; however, it has been postulated that inherent instability imparts a unique dynamic character to these helices (92). Unsurprisingly, 310-helices are often found in functionally- dynamic regions of protein structure such as the active-site, voltage switches in ionchannels, and heme-iron coordination sites (92, 93). Despite lacking the some of the stabilizing hydrogen bonds canonical to !-helix geometry, the 310-helix extension of helix D in PglC is stabilized through the presence of the Asp93 at the helical N-cap position (94) to satisfy the positive electrostatic environment of the helix dipole. Additionally, the hydrophobic packing of the two leucine residues (L92 and L95) at the N′ and C0 positions of the helix add further stability to the secondary structure geometry.

Figure 3.15. Active-site helix D 310-geometry allows for co-facial positioning of Asp-Glu catalytic dyad. A, Helix D (residues 92- 104) shown in color ramp representation from N-terminus (blue) to C-terminus (red). B, Looking down helical axis from N-terminus. N-terminus of helix D adopts 310–helix geometry. C, Looking down helical axis from C-terminus. C-terminus of helix D adopts canonical α-helix geometry.

Table 3.2. Results of helical geometry analysis for helix D. Helix Partition

Helix Length

(residues) Twist (°) Residues/ Turn Unit Height (Å)

Full-length 13 107.2 ± 6.35 3.36 ± 0.20 1.75 ± 0.28

N-terminal 5 112.8 ± 8.08 3.19 ± 0.23 2.06 ± 0.40

C-terminal 9 104.3 ± 5.14 3.45 ± 0.17 1.58 ± 0.14

310 (ideal*) -- 120 3.0 2.0

α (ideal*) -- 105 3.6 1.5

* Ideal helix geometries defined as reported by Barlow and Thornton, 1988 (91).

In PglC, this stablized 310-helical geometry is both structurally and functionally relevant as it architects proximal positioning of the catalytic-dyad by favoring the observed Asp93 rotomer via interactions with the 310-helix dipole (Figure 3.14, dashed gray bonds). Participation of one of the Asp93 sidechain oxygen atoms in these helix-capping hydrogen bonds in addition to coordination of the catalytically required Mg2+ _{ion contributes to the nucleophilic reactivity of} the non-coordinating Asp oxygen via a disruption of resonance stabilization. This increased nucleophilic reactivity for one oxygen of the Asp sidechain is distinctive of covalent-intermediate enzymatic mechanisms (95). As such, the AHABh-motif is not solely a structural motif in PglC, but also plays a critical role in the catalytic mechanism.

3.4 Conclusions

Independent validations of the topology of PglC with respect to the membrane establish the unique membrane interaction modality inclusive of the reentrant membrane helix (helices A, B) and three co-planar amphipathic helices (C, D, and I). This modality acts to stabilize the minimal functional unit of PglC in the membrane. As previously described, the overall structure and membrane- association mode of PglC is largely divergent from other monotopic proteins of known structure in both domain structure, membrane-association interface, and free energy of transfer to the membrane. Additionally, as there are no known structural homologs, either soluble or membrane-associated, it is plausible that the evolution of this the novel PglC fold was intimately associated with the membrane.

As a result, the extensive membrane-resident volume of PglC together with the membrane in which it is positioned effectively forms a folding core and membrane interaction modality distinct from other structurally characterized monotopic proteins. These monotopic membrane proteins of known structure can be classified by the canonical modes of monotopic protein membrane interaction (Figure 3.2): (A) interactions mediated through an amphipathic helix, (B) hydrophobic loops extending into the membrane, (C) electrostatic interactions

between ions and lipid headgroups, and (D) covalent lipid anchors. From the comparison of membrane interaction interface of PglC versus the other monotopics of known structure, it is clear that the structure of PglC represents a first-in-class example of a new modality in which the membrane interaction interface of PglC is comprised of an extensive hydrophobic membrane-embedded domain formed by the reentrant membrane helix and three amphipathic helices (Figure 3.16).

Figure 3.16. Schematic representation of interactions between monotopic membrane proteins and the membrane. A, Interaction via an amphipathic α-helix parallel to the plane of the membrane (prostaglandin H2 synthase – 1CQE); B, Interaction via hydrophobic loops (carboxylesterase – 3CN9); C, Electrostatic or ionic interactions with lipid head groups (arachidonate 15-lipoxygenase – 2P0M); D, membrane association via a covalent fatty acyl, prenyl or

glycosylphosphatidyl-inositol post-translational modification

(acetylcholinesterase – 1N5M) ; E, Reentrant membrane helix (RMH) with C- and N-termini on the same face of the membrane and amphipathic α-helices parallel to the membrane plane (PglC – 5W7L).

Though it is minimal by comparison to other monotopics of known structure, the soluble domain of PglC demonstrates an efficiency with its fold that

is both structurally and functionally significant. Small parsimonious structural motifs not only underpin the critical structural feature of membrane-association, the RMH, they also directly influence active-site geometry through specific positioning of catalytically-important residues (Arg 112 and Asp93). Overall, the unique structure of C. concisus PglC represents a novel architecture for membrane interaction and the first structurally characterized member of the monotopic PGT superfamily.

CHAPTER FOUR

Structural underpinnings of PglC-Substrate Interactions

The interpretation of PEG and PO42- electron density and phosphate release kinetics presented in this chapter are included in the following accepted article: Ray, L.C., Das, D., Entova, S., Lukose, V., Lynch, A.J., Imperiali, B., and Allen, K.N. Membrane association of monotopic phosphoglycosyl transferase underpins function. Nature Chemical Biology (2018). (in press).

4.1 Introduction

Although the monotopic and polytopic PGT superfamilies perform the same general PGT reaction, a transfer of a C1-phosphosugar from a UDP-sugar to a polyprenol phosphate embedded in the membrane, they rely on vastly different scaffolds and molecular logic. The structure of PglC, an exemplar of the monotopic PGT superfamily reveals a novel architecture for membrane association and function that is largely divergent from that of the polytopic PGT class. As these two families are so structurally and mechanistically divergent, it follows that the active-site and binding-site geometry should be dissimilar as well.