X-ray structures of protein-protein complexes involving toxins

The bacteria causing diphtheria, whooping cough, and cholera and other diseases secrete mono-ADP-ribosylating toxins that modify intracellular proteins. Recently, we reported four structures of a catalytically active complexes between a fragment of Pseudomonas aeruginosa exotoxin A (ETA) in complex with its protein substrate, translation elongation factor 2 (eEF2). The target residue in eEF2, diphthamide (a modified histidine), spans across a cleft and faces the two phosphates and a ribose of the NAD⁺ analogue, βTAD. This suggests that the diphthamide is involved in triggering NAD⁺ cleavage and interacting with the proposed oxacarbenium intermediate during the nucleophilic substitution reaction explaining the requirement of diphthamide for ADP-ribosylation. Diphtheria toxin may recognise eEF2 in a manner similar to ETA. Remarkably, the toxin-bound βTAD phosphates mimic the phosphate backbone of two nucleotides in a conformational switch of 18S rRNA thereby achieving universal recognition of eEF2 by ETA.

We have determined four crystal structures of complexes between yeast eEF2 and a catalytical fragment of ETA (ETA_c):

1. The apo complex eEF2-ETA_c;
2. the putative Michaelis complex eEF2-ETA_c-βTAD (b-methylene-thiazole-4-carboxamide adenine dinucleotide);
3. The post-reaction complex between ADP-ribosylated eEF2 (ADPR-eEF2) and ETA_c;
4. the eEF2-ETA_c complex with a water-soluble ETA inhibitor, PJ34 bound within the active site.

Diffraction data extend to a maximum resolution of 2.8-3.1 Å for the four complexes.

All structures were determined by molecular replacement with the structures of eEF2 and ETA_c. They are highly isomorphous and have three complexes in the asymmetric units (complex 1: chains A and B; complex 2: chains C and D; complex 3: chains E and F). In general, the molecules are well ordered, especially domain IV of eEF2 and ETA_c, which form most of the intermolecular contacts. Only a single b-hairpin in eEF2 domain III contributes to the interaction. In complex 3 eEF2 domains I, G’, III and V are somewhat mobile. The conformation of eEF2 in the toxin complex is similar to that of apo-eEF2, but due to its interaction with ETA_c, eEF2 domain III has rotated 13-18° towards domain IV.

There are no major conformational changes such as domain reorientations or large shifts in loops in either eEF2 or ETA_c when comparing the four different structures. In contrast, there are roughly two types of eEF2-ETA_c complexes in all four structures. The NAD⁺ binding sites of complexes 2 and 3 are rotated 5-9 degrees towards the diphthamide located in domain IV of eEF2 compared to complex 1. The axis of rotation passes through ETA residues 493-494 and 578-579 within the eEF2-toxin interface. This causes the distance between the NC1 of the N-ribose, and the nucleophilic NE2 from the diphthamide residue to be 11.0 Å in βTAD complexes 2 and 3 but 12.1 Å in βTAD complex 1. In these two complexes, the diphthamide substituent reaches towards the βTAD b-phosphate. The bulky and positively charged quaternary ammonium substituent on the diphthamide is oriented towards the βTAD b-phosphate whereas the amide group faces towards the N-ribose.

The large distance between the N-ribose NC1 and the nucleophilic diphthamide NE2 atom observed in the Michaelis complex is surprising. To rigorously test the relevance of the structures we demonstrated that the ADPRT reaction readily occurs in crystals of eEF2-ETA_c soaked with NAD⁺. These crystals are physically intact after reaction, and still diffract to at least 3.5 Å. The eEF2-ETA_c-βTAD complex is a close analogue to the E-S complex (Michaelis complex) for the ADPRT reaction, but it clearly does not represent the transition state complex, nor does it indicate what structural changes are required for that complex to form. There must be at least one intermediate complex in the reaction pathway, which then will lead to the formation of the transition state species for the reaction. Although we do not have any structural data on the nature of these additional complexes involved in the reaction mechanism, a governing feature of these complexes is that the distance between the ribose NC1 and the nucleophilic NE2 atom must be considerably smaller than what is observed in our Michaelis complex structure. Kinetic isotope effect studies suggest a separation between the NAD⁺ NC1 and diphthamide NE2 atoms of 2.6 Å at the transition state of the eEF2-DTA complex.

There are potentially several ways of reducing the nucleophile-electrophile distance to a similar value in the eEF2-ETA_c complex. One option is that the oxarbenium ion migrates from the NAD⁺ pocket in ETA towards the diphthamide NE2 atom, but that would inevitably require that the highly reactive oxacarbenium ion was protected from reacting with water or other nucleophiles during transfer across the intermolecular cleft which probably contains 1-2 layers of bound water. Presently, there is no experimental data supporting such a mechanism. The distance could also be reduced if eEF2 rotated substantially relative to ETA_c, but that seems to be prevented in the enzymatically active crystalline state by the tight crystal packing around ETA_c and eEF2 domain IV. A third possibility is that the loop region containing the diphthamide may undergo a transient conformational change during reaction, thereby bringing the diphthamide NE2 atom closer to the oxacarbenium ion. Such a conformational change has not yet been observed, but the diffraction data for all known structures of eEF2 were collected at 100 K. The diphthamide loop region may be more mobile or capable of adapting an alternative conformation at room temperature, where the complex is enzymatically active in the crystal. The interaction between the diphthamide ammonium group and the NAD⁺ phosphates may promote a conformational change in the loop. Furthermore, the relaxation process associated with cleavage of the NC1-NN1 bond of NAD⁺ and alleviation of the strained conformation may also contribute to partial reduction of the distance between the oxacarbenium ion and the diphthamide without exposing the positive charge. However, no matter how the nucleophile-electrophile distance is reduced it seems likely that the quaternary ammonium of the diphthamide modification remains directed towards the phosphates of NAD⁺ thus anchoring the oxacarbenium ion to the diphthamide.

Presently, we are working to determine the crystal structures of various members of the ADPRT toxin family in complex with their target proteins.

top