One unexpected finding of this work is the SycD quaternary structure. We had previously shown that mutations in helix H1B result in monomeric SycD, thereby mapping the homodimerization interface to TPR1 . However, the precise quaternary structure remained ambiguous, because two different crystal forms revealed different arrangements of the protomers in the SycD dimer, an extended and a kinked one . The elongated dimer (PDB ID: 2VGX) had a larger buried surface area, a better shape correlation and a smaller gap volume index, suggesting that it is more likely to represent the solution dimer. Surprisingly, the structure reported here contains a SycD homodimer that is virtually identical to the kinked SycD dimer reported previously (PDB ID: 2VGY). Crystals of the SycD:YopD complex and the kinked apo form of SycD are not isomorphous and in different space groups. This suggests that the kinked dimer may be more favorable than initially thought .
Ambiguity and variability of their quaternary structures appears to be a recurring theme among T3S translocator chaperones. Formation of higher oligomers has been reported for full-length SycD  or Salmonella enterica SicA . It is of note that homodimers of T3S class II chaperones are not constitutive. PcrH shows monomer-dimer equilibrium . Moreover, binding of translocators often disrupts the dimer interaction, resulting in 1:1 complexes [27, 34, 35], although binding of a single translocator to a dimeric chaperone has also been reported . The arrangement of protomers in the chaperone dimers is under debate in several cases as well. For IpgC two different asymmetric dimers  and another 2-fold symmetric dimer involving TPR1  were reported. The crystal structure of PcrH shows two quaternary structures that might be stable in solution, an asymmetric dimer mediated by contacts of the convex side (back-to-back dimer) and a 2-fold symmetric dimer mediated by TPR1 (head-to-head) . For PcrH, no functional data is available to decide which of these dimers is present in solution.
Previous studies had led to the identification of two regions within YopD that are essential for SycD binding , one involving the C-terminal amphipathic domain (aa 278–292) and another one located N-terminally encompassing the residues 53–149, which also contains the six amino acid consensus sequence for binding to class II chaperones identified by Kolbe and co-workers . The concave groove of TPR proteins can accommodate interaction partners in helical [24, 25, 39, 40] or extended conformation [41–43]. Accordingly the SycD concave groove was proposed to bind the amphipathic YopD C-terminus in helical conformation  or a short conserved peptide from the N-terminal region in extended conformation . Our structure verifies that the YopD consensus peptide binds to SycD in the manner predicted by Kolbe and co-workers based on their IpgC:IpaB complex structure . The peptide’s three conserved anchor residues bind into distinct binding pockets within the concave region leading to a peptide conformation that is nearly identical to that of the CBDs of PopD and IpaB in complex with their respective chaperones [26, 27]. This allows understanding how TPR-like chaperones that are well conserved between species (sequence identity SycD/PcrH: 59%, SycD/IpgC: 28%, (see also [21, 26])) bind to their respective cargo that exhibits a lower degree of sequence conservation (sequence identity (1) minor translocator: YopD/PopD:41%, YopD/IpaC: 20%; (2) major translocator YopB/PopB: 43%, YopB/IpaB: 17% (see also [12, 27])) and how they recognize with high specificity two different substrates that share nearly no common sequence features. The presence of the consensus motif for binding of the chaperone’s hydrophobic groove in both the major and the minor translocator from the same species suggests that binding of translocators to the chaperone is mutually exclusive and formation of a ternary complex should not be possible. The data in the literature with regard to this issue is not univocal and the situation may well differ between species [27, 35, 36, 45, 46]. In some cases, e.g. in Edwardsiella tarda, each hydrophobic translocator even requires its own chaperone [47, 48].
The hydrophobic cleft of SycD is lined by aromatic residues, mainly tyrosines, which have been shown to appear with high prevalence in a wide range of protein interaction surfaces . These residues are highly conserved within T3S class II chaperones although they are located at non-canonical positions [21, 34, 44]. Tyr40 and Tyr47 are not only involved in the formation of hydrophobic patches but also keep the peptide in the correct conformation by providing a functional group for hydrogen bond formation with the peptide backbone. The existence of very similar interactions in the chaperone-peptide complexes of IpgC from Shigella and PcrH from Pseudomonas underlines the importance hereof. Thus it is not surprising, that mutations within the conserved aromatic ladder of helix H1A (Tyr40, Phe44, Tyr47 in SycD) have severe effects on the recognition and binding of YopD and YopB . Based on the present structure it is not possible to rationalize why several mutations that map to canonical TPR positions or the binding groove should affect binding and/or secretion only of YopB but not of YopD . Mutations affecting secretion only of YopD but not YopB were mapped to the convex side (L42A, H67A and L76A), which thus was considered to be SycD’s main binding region for YopD, with additional contributions from the loop connecting H2B and H3A [38, 50, 51]. These observations may be explained by the fact that the CBD represents only a small part of a longer N-terminal binding region that together with the C-terminal amphipathic domain is involved in SycD binding. Furthermore, the residual part (aa 150–287) of the minor translocator has been shown to exist in a partially unfolded molten globule state . Thus one might assume that YopD and most likely YopB are recognized by the chaperone’s concave cleft through the conserved binding motif. The partly unfolded region of the protein could wrap around SycD so that YopD might contact the convex side with the remaining binding regions.