A recurring packing contact in crystals of InlB pinpoints functional binding sites in the internalin domain and the B repeat
InlB, a bacterial agonist of the human receptor tyrosine kinase MET, consists of an N-terminal internalin domain, a central B repeat and three C-terminal GW domains. In all previous structures of full-length InlB or an InlB construct lacking the GW domains (InlB392), there was no interpretable electron density for the B repeat. Here, three InlB392 crystal structures in which the B repeat is resolved are described. These are the first structures to reveal the relative orientation of the internalin domain and the B repeat. A wild-type structure and two structures of the T332E variant together contain five crystallographically independent molecules. Surprisingly, the threonine-to-glutamate substitution in the B repeat substantially improved the crystallization propensity and crystal quality of the T332E variant. The internalin domain and B repeat are quite rigid internally, but are flexibly linked to each other. The new structures show that inter-domain flexibility is the most likely cause of the missing electron density for the B repeat in previous InlB structures. A potential binding groove between B-repeat strand β2 and an adjacent loop forms an important crystal contact in all five crystallographically independent chains. This region may represent a hydrophobic `sticky patch' that supports protein–protein interactions. This assumption agrees with the previous finding that all known inactivating point mutations in the B repeat lie within strand β2. The groove formed by strand β2 and the adjacent loop may thus represent a functionally important protein–protein interaction site in the B repeat.
InlB is an invasion protein that is located on the surface of the bacterium Listeria monocytogenes (Dramsi et al., 1995; Lingnau et al., 1995). As a facultative intracellular pathogen, L. monocytogenes can stimulate its own uptake by nonphagocytic cells such as epithelial or endothelial cells (Pizarro-Cerdá et al., 2012). Binding of InlB to the receptor tyrosine kinase MET on host cells is one way to induce internalization of bacteria (Shen et al., 2000). MET acts as a receptor for hepatocyte growth factor/scatter factor (HGF/SF) and is essential during mammalian development (Birchmeier et al., 2003). MET also contributes to tissue regeneration and wound healing, while MET deregulation can promote cancer metastasis (Trusolino et al., 2010). Cellular phenotypes in response to InlB stimulation resemble those observed upon MET activation by HGF/SF. They include cell motility (for example cell scatter of clonally growing MDCK or HT-29 cells) and cell proliferation (Shen et al., 2000; Niemann et al., 2007). At the molecular level, InlB induces MET phosphorylation and the activation of downstream signalling pathways such as phosphorylation of extracellular signal-regulated kinase (ERK) or activation of phosphoinositide 3-kinase (PI3K) and phosphorylation of protein kinase B (PKB)/AKT (Copp et al., 2003; Seveau et al., 2007).
Mature InlB is a 595-amino-acid, multi-domain protein (Fig. 1a). The N-terminal internalin domain (amino acids 36–321; InlB321) is characterized by a central leucine-rich repeat (LRR) region that binds to the first Ig-like IPT domain of MET with high affinity (Niemann et al., 2007). The LRR region is flanked N-terminally by a helical cap and C-terminally by an Ig-like inter-repeat (IR) region (Schubert et al., 2001). The C-terminal part (amino acids 393–630) consists of three GW domains named after a conserved Gly–Trp motif. The GW domains have an SH3-like fold and a high isoelectric point (pI) of about 10 (Marino et al., 2002). The GW domains bind polyanions, including lipoteichoic acid (LTA), on the bacterial surface (Jonquières et al., 1999) and heparan sulfate of host cell surface heparan sulfate proteoglycans (HSPGs; Jonquières et al., 2001). Binding to LTA or HSPGs is mutually exclusive (Jonquières et al., 2001). InlB apparently acts as a soluble protein that induces bacterial uptake into host cells rather than as an adhesin that supports the attachment of L. monocytogenes to host cells (Jonquières et al., 2001; Banerjee et al., 2004). Binding of the GW domains to host HSPGs enhances MET activation through the internalin domain (Banerjee et al., 2004), presumably via an avidity effect or receptor clustering (Jonquières et al., 2001; Niemann et al., 2007). On their own or provided in trans with InlB392, the GW domains showed no effect in cellular assays (Banerjee et al., 2004).
InlB activates MET by dimerization of the MET ectodomain (Niemann, 2013). InlB-induced MET dimerization on cells was shown by fluorescence microscopy (Dietz et al., 2013; Harwardt et al., 2017; Koschut et al., 2016). The crystal structure of a 2:2 complex formed by the InlB internalin domain and a large portion of the MET extracellular domain revealed an arrangement in which two InlB molecules form a twofold-symmetric homodimer via the convex distal surface of their LRR regions (Niemann et al., 2007; Ferraris et al., 2010). Two MET molecules sit on the outside of this InlB dimer, with each being bound to the concave `front side' of the LRR region. Except for a small contact of two IPT2 domains, MET does not contribute to dimerization in this structure. The same 2:2 arrangement of the InlB internalin domain and MET has recently been observed in another structure obtained in the presence of a MET-binding DARPin, lending further support to this activation model (Andres et al., 2019).
InlB structures have been published, in the following order, of the LRR fragment (InlB248; Marino et al., 1999), the internalin domain (InlB321; Schubert et al., 2001) and the full-length protein (Marino et al., 2002). However, the B repeat was not resolved in the structure of full-length InlB. There was some electron density between the internalin and GW domains, but this could not be modelled. Therefore, we determined the structure of a fragment consisting of the internalin domain and the B repeat but lacking the GW domains (InlB392; Ebbes et al., 2011). Again, the B repeat was not resolved and this time there was no electron density at all for the B repeat, suggesting it to be highly flexible. In this crystal form, all packing contacts are formed by the internalin domain. We assume that the B repeat makes no crystal contacts at all and that it is free to move in large solvent channels. Therefore, we crystallized the B repeat alone (Ebbes et al., 2011). These crystals diffracted to high resolution (1.3 Å) and revealed a well defined structure. The B repeat folds into a compact domain resembling a β-grasp fold. The only difference from a canonical β-grasp fold is the absence of the helix connecting strands β2 and β3. Instead, the two parallel edge β-strands are connected by an extended loop structure. While structures of all domains of InlB are now available, the relative orientation of the B repeat and the internalin domain remains elusive.
The B repeat is arguably the least understood domain in InlB. Several internalins contain up to four domains that are homologous to the B repeat, yet none of them has been functionally characterized (Bierne et al., 2007). Initial experiments with recombinant domain-deletion constructs of InlB showed that the B repeat contributes to ERK activation, but found no effect on MET phosphorylation (Copp et al., 2003). This led to the suggestion that the B repeat could bind a co-receptor and activate a second signalling pathway parallel to MET. A more detailed study revealed that InlB392 has a slightly (about twofold) higher potency in inducing MET phosphorylation than InlB321 (Banerjee et al., 2004). Thus, the B repeat apparently does contribute to MET activation. Cell-motility assays with canine MDCK and human cells revealed a profound effect of the B repeat. While InlB392 induced MDCK colony scatter at 10 nM (Ebbes et al., 2011), InlB321 was unable to do so even at 1000 nM (Ferraris et al., 2010). A construct consisting only of the B repeat and GW1–3 had no effect (Ebbes et al., 2011). In human HT-29 cells, InlB392 was about tenfold more active than InlB321 (Bleymüller et al., 2016). In summary, the B repeat on its own shows no effect on cells. When fused to the internalin domain it enhances MET phosphorylation. Its presence increases potency even more in assays of ERK or AKT phosphorylation or cellular phenotypes such as cell motility. So far, however, all of our efforts to identify the postulated co-receptor for the B repeat have failed (Bleymüller et al., 2016).
Analysis of InlB392 variants with single or multiple substitutions on the surface of the B repeat in cellular receptor-activation assays revealed only two functionally important amino acids, both of which are located in the edge strand β2 (Bleymüller et al., 2016). The mutation of Thr332 or Ile334 strongly impaired or completely abolished the activity of InlB392 in cellular assays. The resulting protein variants were at least 100-fold less active than wild-type InlB392 (InlB392_wt). Interestingly, the InlB392 variants T332E and I334D/T336L (variant D in Bleymüller et al., 2016) were inactive in a cell-motility assay with human HT-29 cells even at the highest concentration tested (1000 nM), while the internalin domain alone stimulated cell scatter at 10 nM. Thus, InlB392 variants with a single substitution in the B repeat are substantially less active than a construct that completely lacks the B repeat. One possible mechanistic explanation for this perplexing result is that the wild-type B repeat contributes to MET dimerization, whereas the T332E or I334D mutation hinders dimerization of the internalin domains (Bleymüller et al., 2016). To scrutinize this hypothesis, it would be highly interesting to resolve the relative orientation of the internalin domain and to identify protein–protein interaction sites in the B repeat.
Here, we describe the crystallization and structure determination of wild-type InlB392 (InlB392_wt) and the inactive variant InlB392_T332E. While the wild-type protein was difficult to crystallize and the rare crystals generally showed no or rather low-resolution diffraction, we obtained crystals of InlB392_T332E under several conditions and were able to determine two structures at resolutions of 2.05 and 1.8 Å. Both domains are resolved in all three structures, revealing the arrangement of the internalin domain and the B repeat in InlB for the first time.
All three proteins were expressed and purified as described in Bleymüller et al. (2016). For crystallization, the protein buffer phosphate-buffered saline (PBS), in which the proteins were stored at −80°C, was exchanged for crystallization buffer (10 mM Tris pH 8.0, 20 mM NaCl). All proteins were crystallized in MRC 2-well plates with drops consisting of 100 nl protein (10 mg ml−1) and 100 nl reservoir solution set up using a Crystal Gryphon robot (Art Robbins Instruments). All crystals were harvested directly from commercial or homemade standard screens without further optimization. InlB392_wt crystallized at 20°C in the MORPHEUS screen (Gorrec, 2009) condition E6 [0.1 M (HEPES sodium salt/MOPS acid) pH 7.5, 20%(v/v) ethylene glycol, 10%(w/v) PEG 8000, 30 mM diethylene glycol, 30 mM triethylene glycol, 30 mM tetraethylene glycol, 30 mM pentaethylene glycol]. For harvesting, crystals were cryoprotected in reservoir solution to which an additional 20%(v/v) ethylene glycol was added before flash-cooling the crystal in liquid nitrogen. Crystal form I of InlB392_T332E was obtained at 4°C in the MORPHEUS screen condition G2 [0.1 M (imidazole/MES) pH 6.5, 20%(v/v) ethylene glycol, 10%(w/v) PEG 8000, 20 mM sodium formate, 20 mM ammonium acetate, 20 mM sodium citrate tribasic dehydrate, 20 mM sodium potassium tartrate tetrahydrate, 20 mM sodium oxamate]. These crystals were harvested directly from the crystallization drop and flash-cooled in liquid nitrogen without the addition of a cryoprotectant. Crystal form II of InlB392_T332E was obtained at 4°C in condition D3 of a homemade PEG smear screen with low- and broad-molecular-weight PEGs as described in Chaikuad et al. (2015). The reservoir consisted of 0.1 M succinate pH 7.0, 0.2 M Li2SO4 and 22.5% PEG mixture consisting of equal amounts of eight low-molecular-weight PEGs (PEG 300, PEG 400, PEG 500 MME, PEG 550 MME, PEG 600, PEG 750 MME, PEG 1000 and PEG 1000 MME). These crystals were cryoprotected in reservoir solution additionally containing 15% glycerol before flash-cooling in liquid nitrogen.
Data for InlB392_wt were collected on beamline P13 operated by EMBL Hamburg at the PETRA III storage ring (Cianci et al., 2017) using a PILATUS 6M detector (Dectris). For InlB392_T332E, measurements were carried out on the BL14.2 beamline at the BESSY II electron-storage ring operated by the Helmholtz-Zentrum Berlin für Materialien und Energie (Mueller et al., 2015). Data for both crystal forms of InlB392_T332E were collected on a PILATUS 3S 2M detector (Dectris). Between 240 and 360° of fine-sliced data (0.1° per frame) were collected. The data were indexed and integrated with XDS (Kabsch, 2010) and scaled with XSCALE using zero-dose extrapolation (Diederichs et al., 2003). Data-collection statistics are shown in Table 1.
2.3. Structure determination and refinement
The structures were solved by molecular replacement with Phaser (McCoy et al., 2007) from the CCP4 package (Winn et al., 2011). The internalin domain (PDB entry 1h6t; Schubert et al., 2001) was placed first, followed by placement of the B repeat (PBD entry 2y5p; Ebbes et al., 2011). The structures were completed by iterative cycles of model building in Coot (Casañal et al., 2020) and refinement in REFMAC5 (Kovalevskiy et al., 2018) during the early stages and phenix.refine (Liebschner et al., 2019) during the final stages of rebuilding. TLS refinement was used for all three structures. For InlB392_wt with three chains in the asymmetric unit, local noncrystallographic symmetry (NCS) restraints and restraints to InlB392_T332E crystal form II (PDB entry 7nms) as a reference model were applied. Refinement statistics are shown in Table 1.
We extensively screened for new crystallization conditions of wild-type InlB392. Crystals only grew in condition E6 of a MORPHEUS screen (Gorrec, 2009) stored beyond its `use by' date. We were unable to reproduce or optimize these crystals with homemade solutions. At 20°C we obtained single crystals shaped as hexagonal plates typically of about 60 × 60 × 10 µm in size and reaching up to 100 × 100 × 15 µm. These crystals showed no diffraction even on beamlines BL14.2 of BESSY II and P13 of PETRA III. At 4°C we obtained thin rod-shaped crystals with a cross-section of about 10 × 10 µm and a length of 80–100 µm. The best crystal diffracted to 3.3 Å resolution (Table 1). For flash-cooling, we had to add ethylene glycol (see Section 2) as the reservoir solution alone did not freeze clearly, although all conditions of the MORPHEUS screen should be inherently cryoprotected.
InlB392_T332E, an InlB392 variant with a single Thr-to-Glu substitution in the B repeat, readily yielded single crystals of varying morphologies in several conditions from the commercial MORPHEUS screen and a homemade PEG smear screen (Chaikuad et al., 2015). These crystals reached a size and quality sufficient for data collection without further optimization. The best crystal from condition G2 of the MORPHEUS screen (crystal form I) was a rhombic plate with a cross-section of about 125 × 50 µm and diffracted to 2.05 Å resolution (Table 1). The best crystal from condition D3 of the PEG smear (low- and broad-molecular-weight) screen (crystal form II) was a rod with a cross section of about 35 × 35 µm and a length of about 180 µm. This crystal diffracted to 1.8 Å resolution (Table 1). We also collected data from InlB392_T332E crystals grown under different conditions from the MORPHEUS or PEG smear screens. Some were isomorphous to crystal form I and diffracted to lower resolution, so we did not pursue structure determination. Others had larger unit cells and showed signs of translational noncrystallographic symmetry in the native Patterson map. Our attempts to solve these structures by molecular replacement in Phaser both with and without the tNCS option have so far failed.
All three structures described in this paper were easily solved by molecular replacement. Phaser was able to place both the internalin domain and the B repeat for all five crystallographically independent molecules. The refinement statistics are listed in Table 1. The presence of many weak reflections due to translational pseudosymmetry (see below) may explain the relatively high Rwork and Rfree values for InlB392_wt.
As expected, there are no major differences between the internalin domains in the InlB392 structures and the previously reported structures of the InlB internalin domain. An overlay of the LRR region revealed some flexibility in the cap and especially in the IR region (Fig. 1b). Likewise, there are no major differences between the B repeats in the InlB392 structures and the published high-resolution structure of the isolated B repeat, with the exception of chain C in the InlB392_wt structure, which shows no clear density for residues 354–372 corresponding to strand β3 and the following loop in the other B-repeat structures (Fig. 1c). These residues may be flexible or adopt multiple conformations which cannot be resolved at this rather low resolution of 3.3 Å.
The orientation of the B repeat relative to the internalin domain varies substantially (Figs. 2a–2c). There are no polar contacts between the internalin domain and the B repeat to stabilize their arrangement. Many internalins have at least one domain C-terminal to the LRR or IR region (Bierne et al., 2007). Besides InlB, InlK is the only internalin for which a structure extending beyond the internalin domain is known (Neves et al., 2013). The LRR-adjacent D2 domain and the following D3 domain of InlK are structurally distinct from the InlB IR region and B repeat, respectively. Nevertheless, as in our InlB392 structures, few contacts and large flexibility between D2 and D3 were found in InlK (Neves et al., 2013). In InlB, the short linker between the internalin domain and the B repeat acts as a pivot point rather than a hinge (Figs. 2a–2c).
Glu321 was previously regarded as part of the internalin domain (Schubert et al., 2001), but may rather belong to the B repeat, as it forms two polar contacts with other B-repeat residues but none with the internalin domain. In all five crystallographically independent InlB392 chains reported here there are hydrogen bonds between the backbone carbonyl O atom of Glu321 and the backbone NH of Ala340 and between the side-chain carboxylate of Glu321 and the phenolic OH of Tyr323 (Fig. 2d). The only interaction of Glu321 and the internalin domain is a CH–π interaction of the aliphatic part of the glutamate side chain with the aromatic ring of Phe290.
None of the structures contained a twofold-symmetric arrangement of InlB392 indicative of a dimer, and the PISA server (Krissinel & Henrick, 2007) predicted InlB392 to be monomeric in solution for all three structures. The packing of InlB392_T332E crystal form I is closely related to that of InlB392_wt and the surroundings of all three molecules in the wild-type structure are similar. In a way, the packing of InlB392_wt can be viewed as a slightly distorted version of InlB392_T332E crystal form I with a tripled b axis (Fig. 3). Accordingly, the native Patterson map of InlB392_wt has an off-origin peak with a peak height of 41.7% of the origin peak at coordinates u, v, w = 0.0000, 0.3234, 0.0000 according to the tNCS detection of Phaser (38.0% at 0.000, 0.322, 0.000 according to phenix.xtriage). The packing of InlB392_T332E crystal form II is different.
A common contact of all four molecules of InlB392_wt and InlB392_T332E crystal form I is formed between the concave LRR side of one molecule and the cap region and the convex LRR side of another molecule (Fig. 3). The concave side of the LRR is the primary binding site for the MET receptor (Niemann et al., 2007). The crystal contact involves exposed aromatic side chains of InlB that are essential for MET binding (Machner et al., 2003). The contact area is between 397 and 647 Å2, with a mean of 549 Å2, according to the PISA server (Krissinel & Henrick, 2007). This crystal contact is not present in crystal form II of InlB392_T332E.
However, all five crystallographically independent InlB392 molecules in the three crystal forms have one recurring crystal contact in common. The contact area lies between 594 and 687 Å2, with a mean of 621 Å2, according to the PISA server (Krissinel & Henrick, 2007). This contact is formed between the B repeat of one molecule and the IR region of a neighbouring molecule (Fig. 4). On the B repeat it involves residues from strand β2 (mainly Val329, Thr/Glu332, Val333 and Ile334) and the long loop (residues 347–353) connecting strands β2 and β3 that is a helix in canonical β-grasp fold proteins. The PISA server calculates a favourable interaction for this contact (negative solvation-energy effect ΔiG ranging from −10.0 to −23.4 kJ mol−1; mean −15.1 kJ mol−1), while the previously described LRR–cap contact has a positive ΔiG ranging from −0.4 to 16.7 kJ mol−1 (mean 5.4 kJ mol−1). The overall arrangement of the contact between the B repeat and the IR region is similar for all five instances of InlB392 described in this paper. The precise geometry varies somewhat and two groups can be distinguished. One group has three members, namely both crystal forms of InlB392_T332E and the B repeat of chain C packing against the IR region of chain A in InlB392_wt (Fig. 4a). The second group comprises two crystal contacts of InlB392_wt, namely the B repeat of chain B packing against the IR region of a symmetry-related chain B and the B repeat of chain A packing against the IR region of chain C (Fig. 4b). The different geometry of the two contact groups is illustrated in Fig. 4(c). In the T332E variant, the mutated residue Glu332 substantially contributes to formation of this crystal contact and is well defined in the electron density (Fig. 4d).
Taking into account the packing of all three crystal forms described in this paper, the binding of the IR region to the groove between B-repeat strand β2 and the loop connecting β2 and β3 is clearly the dominant packing interaction. This crystal contact has the largest mean interface area, the most negative ΔiG and is formed by all molecules. Potential functional implications will be considered in Section 4.
The T332E substitution is one of two point mutations that we had previously found to have a negative effect on the biological function of the B repeat (Bleymüller et al., 2016). Thr332 is surface-exposed. Therefore, we had expected the mutation to glutamate not to impair the B-repeat structure. The circular-dichroism spectrum and the elution behaviour on a gel-filtration column confirmed this assumption, as they were basically identical for the wild-type B repeat and the T332E variant (Bleymüller et al., 2016). The crystal structures presented in this work provide additional and conclusive proof that the T332E mutation does not change the structure of the B repeat. Therefore, the negative effect of this substitution in cellular assays is most likely due to the destruction of a binding site, preventing the interaction with a functionally important binding partner.
Intriguingly, wild-type InlB392 was a much more problematic crystallization target than the InlB392_T332E variant. This is unexpected because glutamate side chains are statistically underrepresented in interfaces of oligomeric proteins, presumably due to their high conformational entropy (Derewenda & Vekilov, 2006). Moreover, crystal contacts are systematically depleted of residues with high side-chain entropy, and Glu, along with Lys, has the lowest propensity to form crystal contacts (Cieślik & Derewenda, 2009). In the semi-rational surface-entropy reduction (SER) approach, surface-exposed glutamates are mutated to alanine, threonine or tyrosine in order to increase the likelihood of crystallization (Mateja et al., 2002; Cooper et al., 2007). Analysis of the actual crystal contacts with the PISA server indicates an unfavourable contribution of Thr332 (positive solvation-energy effect ΔiG ≃ 4.2 kJ mol−1) but an almost neutral effect of Glu332 (ΔiG ≃ 0 kJ mol−1). One explanation for this unexpected effect of the T332E substitution would be that the B repeat evolved to prevent fortuitous binding to the IR domain and that the T332E mutation counteracts this anti-aggregation property.
In the structure of full-length InlB there were two possibilities to connect the GW domains to the internalin domain, with similar distances between the C-terminal residue of the internalin domain and the N-terminal residue of the GW domains (Marino et al., 2002). The authors deposited an asymmetric unit (shown in grey in Fig. 5) containing the combination with the slightly shorter distance of 47 Å between the Cα atoms of residues 319 and 393 in the Protein Data Bank. The distance to residue 393 of a symmetry-related copy of the GW domains (shown in pink in Fig. 5) is 52 Å.
Upon overlaying all five copies of InlB392 described in this work with one copy of the InlB full-length dimer, the C-terminal residues of the B repeat are about halfway between the two possible GW domains (Fig. 5). For chain A of InlB392_wt, the distances to residue 393 of the GW domains of the deposited asymmetric unit and the symmetry-related GW domains are 22.1 and 20.8 Å, respectively. The largest difference between these distances is found for InlB392_T332E crystal form II. The distances to residue 393 of the GW domains of the deposited asymmetric unit and the symmetry-related GW domains of PDB entry 1m9s (Marino et al., 2002) are 34.4 and 9.0 Å, respectively. For all five instances of InlB392 reported here, the shortest distance between the C-terminal residue of the B repeat and the N-terminal residue of the GW domains is found for the symmetry-related GW domains. However, none of the orientations of the B repeat observed in InlB392 can represent the position present in the full-length protein in PDB entry 1m9s, as no residues are left that could span the distance of at least 9.0 Å. Moreover, if one overlays two copies of InlB392 onto the two copies of the functional full-length dimer, there would be severe clashes of the B repeat (Supplementary Fig. S1). Given the highly flexible linkage between the internalin domain and the B repeat, this suggests that crystal-packing forces govern the observed orientations of the B repeat.
Strikingly, the contact between B-repeat strand β2 and the IR region of a neighbouring molecule is similar in all five chains. This is unusual as monomeric proteins rarely have common interfaces in more than one-third of their crystal forms and generally much less (Xu et al., 2008). Recurring crystal contacts indicate energetically favourable interactions and sometimes they are even biologically relevant (Xu et al., 2008). The contact between strand β2 and the IR region most likely does not represent a physiological contact. Firstly, in the complex with the MET receptor (Niemann et al., 2007) the IR region contacts the Sema domain of the MET receptor with residues that contact strand β2 of the B repeat in the InlB392 structures (Fig. 6). Upon binding to MET, the IR region will therefore not be able to form this contact with the B repeat. Secondly, the contact between the B repeat and the IR region is formed regardless of the T332E mutation. As the T332E mutation strongly impairs the biological function of the B repeat in cellular assays, it appears highly unlikely that this represents a physiologically relevant contact.
Instead, we assume that this is a fortuitous interaction between two binding sites lacking their native binding partner. The MET binding site in the IR region and strand β2 of the B repeat may thus represent `sticky patches', i.e. specific surface patches with properties that are thermodynamically favourable for cohesive interactions (Derewenda & Godzik, 2017). This hypothesis is supported by the ΔiG P-value reported by the PISA server. The ΔiG P-value is a measure of interface specificity. A P-value larger than 0.5 means that the interface is less hydrophobic than it could be and therefore the interface is likely to be an artefact of crystal packing. A P-value smaller than 0.5 indicates an interface with a hydrophobicity that is higher than the average for the given structures would be, implying that the interface surface can be interaction-specific (Krissinel & Henrick, 2007). The cap–LRR interface that is present in crystal form I of the T332E variant and three times in the wild-type structure has a mean ΔiG P-value of 0.62, indicating a pure crystal-packing contact. In contrast, the interface between the B repeat and the IR region that is present in all structures has a ΔiG P-value of 0.28, indicating higher hydrophobicity than would be expected on average (Fig. 7).
We previously observed a similar phenomenon for the Y. enterocolitica type III secretion protein SycD. Recombinantly produced SycD showed weak, concentration-dependent homodimerization in solution. Two crystal forms showed putative homodimers; however, the arrangement of protomers in these dimers differed (Büttner et al., 2008). Both homodimers involved similar surface patches in the first tetratricopeptide repeat (TPR), but the geometry of the two contacts differed. Subsequently, a structure of the complex of the Aeromonas hydrophila homolog AscH with AopB suggested that SycD probably employs this region of TPR1 for interaction with its binding partner YopB, when present (Nguyen et al., 2015). In the absence of YopB, the YopB binding site in TPR1 of SycD could be a `sticky patch' that promotes fortuitous SycD homodimerization and supports different geometries of the dimer interface.
Our assumption that the hydrophobic groove formed by strand β2 and the loop connecting strands β2 and β3 forms the primary ligand-binding site in the B repeat (Fig. 7) is supported by our previous observations. In an attempt to identify potential binding sites in the B repeat by mutagenesis, the only two substitutions of surface residues that resulted in a loss of function were located in strand β2. We had suggested that the B repeat potentiates MET activation by forming a weak homodimer contact through which it could promote the dimerization of MET bound to the internalin domain. The T332E mutation would prevent homodimerization of the B repeat and thereby suppress MET dimerization through the internalin domain (Bleymüller et al., 2016). The InlB392 structures presented here are compatible with the role of the B repeat in MET activation that we proposed previously. However, additional experiments will be required to corroborate or disprove this model, as we did not observe homodimerization of the B repeat even in crystals of wild-type InlB392.
While crystal-packing contacts are often suspected to lead to structural artefacts, they can also help to reveal functional information when comparing different packing environments. Here, five crystallographically independent molecules reveal protein dynamics, highlighting the high inter-domain flexibility between the internalin domain and the B repeat of InlB. However, the five structures presented here apparently do not completely map out the conformational space of InlB392, as the B repeat needs to adopt yet another position to fit into the structure of full-length InlB. Therefore, additional crystal forms and further structures are likely to show even larger inter-domain movements.
The largest packing contact that is similarly formed by all five crystallographically independent instances of InlB392 involves the known binding site for the MET Sema domain in the IR region and a surface patch in the B repeat that has previously been shown to be functionally important through mutagenesis and cellular assays. The structures analysed in this paper thus represent a good example showing that recurrent crystal contacts can highlight physiologically relevant binding sites. Due to the extremely high protein concentration in crystals, crystallography allows the visualization of protein–protein interactions that are too weak to be studied in solution. The biological relevance of such contacts needs to be addressed by complementary functional assays (Kobe et al., 2008). If an actual binding partner is missing, the interface area can still show up as a crystal contact interacting with heterologous protein surfaces in a nonphysiological way (Forwood et al., 2007). In a time where protein crystallography is increasingly sandwiched between accurate protein structure prediction (Baek et al., 2021; Jumper et al., 2021) and high-resolution cryo-EM of large proteins and complexes (Kühlbrandt, 2014), the careful analysis of crystal contacts might turn out to be a valuable asset for crystallography.
Synchrotron MX data were collected on beamline P13 operated by EMBL Hamburg at the PETRA III storage ring, DESY, Hamburg, Germany. We would like to thank Dr Isabel Bento and Saravanan Panneerselvam for assistance in using the beamline. Measurements were carried out on the BL14.2 beamline at the BESSY II electron-storage ring operated by the Helmholtz-Zentrum Berlin für Materialien und Energie. We would like to thank Dr Christian Feiler for assistance during the experiment. Open access funding enabled and organized by Projekt DEAL.
We gratefully acknowledge the financial support from HZB.
Andres, F., Iamele, L., Meyer, T., Stüber, J. C., Kast, F., Gherardi, E., Niemann, H. H. & Plückthun, A. (2019). J. Mol. Biol. 431, 2020–2039. CrossRef CAS PubMed Google Scholar
Baek, M., DiMaio, F., Anishchenko, I., Dauparas, J., Ovchinnikov, S., Lee, G. R., Wang, J., Cong, Q., Kinch, L. N., Schaeffer, R. D., Millán, C., Park, H., Adams, C., Glassman, C. R., DeGiovanni, A., Pereira, J. H., Rodrigues, A. V., van Dijk, A. A., Ebrecht, A. C., Opperman, D. J., Sagmeister, T., Buhlheller, C., Pavkov-Keller, T., Rathinaswamy, M. K., Dalwadi, U., Yip, C. K., Burke, J. E., Garcia, K. C., Grishin, N. V., Adams, P. D., Read, R. J. & Baker, D. (2021). Science, 373, 871–876. Web of Science CrossRef CAS PubMed Google Scholar
Banerjee, M., Copp, J., Vuga, D., Marino, M., Chapman, T., Van Der Geer, P. & Ghosh, P. (2004). Mol. Microbiol. 52, 257–271. CrossRef PubMed CAS Google Scholar
Bierne, H., Sabet, C., Personnic, N. & Cossart, P. (2007). Microbes Infect. 9, 1156–1166. Web of Science CrossRef PubMed CAS Google Scholar
Birchmeier, C., Birchmeier, W., Gherardi, E. & Vande Woude, G. F. (2003). Nat. Rev. Mol. Cell Biol. 4, 915–925. CrossRef PubMed CAS Google Scholar
Bleymüller, W. M., Lämmermann, N., Ebbes, M., Maynard, D., Geerds, C. & Niemann, H. H. (2016). J. Biol. Chem. 291, 25567–25577. PubMed Google Scholar
Büttner, C. R., Sorg, I., Cornelis, G. R., Heinz, D. W. & Niemann, H. H. (2008). J. Mol. Biol. 375, 997–1012. Web of Science PubMed Google Scholar
Casañal, A., Lohkamp, B. & Emsley, P. (2020). Protein Sci. 29, 1069–1078. Web of Science PubMed Google Scholar
Chaikuad, A., Knapp, S. & von Delft, F. (2015). Acta Cryst. D71, 1627–1639. Web of Science CrossRef IUCr Journals Google Scholar
Cianci, M., Bourenkov, G., Pompidor, G., Karpics, I., Kallio, J., Bento, I., Roessle, M., Cipriani, F., Fiedler, S. & Schneider, T. R. (2017). J. Synchrotron Rad. 24, 323–332. Web of Science CrossRef CAS IUCr Journals Google Scholar
Cieślik, M. & Derewenda, Z. S. (2009). Acta Cryst. D65, 500–509. Web of Science CrossRef IUCr Journals Google Scholar
Cooper, D. R., Boczek, T., Grelewska, K., Pinkowska, M., Sikorska, M., Zawadzki, M. & Derewenda, Z. (2007). Acta Cryst. D63, 636–645. Web of Science CrossRef CAS IUCr Journals Google Scholar
Copp, J., Marino, M., Banerjee, M., Ghosh, P. & van der Geer, P. (2003). J. Biol. Chem. 278, 7783–7789. CrossRef PubMed CAS Google Scholar
Derewenda, Z. S. & Godzik, A. (2017). Methods Mol. Biol. 1607, 77–115. CrossRef CAS PubMed Google Scholar
Derewenda, Z. S. & Vekilov, P. G. (2006). Acta Cryst. D62, 116–124. Web of Science CrossRef CAS IUCr Journals Google Scholar
Diederichs, K., McSweeney, S. & Ravelli, R. B. G. (2003). Acta Cryst. D59, 903–909. Web of Science CrossRef CAS IUCr Journals Google Scholar
Dietz, M. S., Hasse, D., Ferraris, D. M., Göhler, A., Niemann, H. H. & Heilemann, M. (2013). BMC Biophys. 6, 6. Google Scholar
Dramsi, S., Biswas, I., Maguin, E., Braun, L., Mastroeni, P. & Cossart, P. (1995). Mol. Microbiol. 16, 251–261. CrossRef CAS PubMed Web of Science Google Scholar
Ebbes, M., Bleymüller, W. M., Cernescu, M., Nölker, R., Brutschy, B. & Niemann, H. H. (2011). J. Biol. Chem. 286, 15496–15506. CrossRef CAS PubMed Google Scholar
Ferraris, D. M., Gherardi, E., Di, Y., Heinz, D. W. & Niemann, H. H. (2010). J. Mol. Biol. 395, 522–532. CrossRef PubMed CAS Google Scholar
Forwood, J. K., Thakur, A. S., Guncar, G., Marfori, M., Mouradov, D., Meng, W., Robinson, J., Huber, T., Kellie, S., Martin, J. L., Hume, D. A. & Kobe, B. (2007). Proc. Natl Acad. Sci. USA, 104, 10382–10387. Web of Science CrossRef PubMed CAS Google Scholar
Gorrec, F. (2009). J. Appl. Cryst. 42, 1035–1042. Web of Science CrossRef CAS IUCr Journals Google Scholar
Harwardt, M.-L. I. E., Young, P., Bleymüller, W. M., Meyer, T., Karathanasis, C., Niemann, H. H., Heilemann, M. & Dietz, M. S. (2017). FEBS Open Bio, 7, 1422–1440. CrossRef CAS PubMed Google Scholar
Jonquières, R., Bierne, H., Fiedler, F., Gounon, P. & Cossart, P. (1999). Mol. Microbiol. 34, 902–914. PubMed Google Scholar
Jonquières, R., Pizarro-Cerdá, J. & Cossart, P. (2001). Mol. Microbiol. 42, 955–965. PubMed Google Scholar
Jumper, J., Evans, R., Pritzel, A., Green, T., Figurnov, M., Ronneberger, O., Tunyasuvunakool, K., Bates, R., Žídek, A., Potapenko, A., Bridgland, A., Meyer, C., Kohl, S. A. A., Ballard, A. J., Cowie, A., Romera-Paredes, B., Nikolov, S., Jain, R., Adler, J., Back, T., Petersen, S., Reiman, D., Clancy, E., Zielinski, M., Steinegger, M., Pacholska, M., Berghammer, T., Bodenstein, S., Silver, D., Vinyals, O., Senior, A. W., Kavukcuoglu, K., Kohli, P. & Hassabis, D. (2021). Nature, 596, 583–589. Web of Science CrossRef CAS PubMed Google Scholar
Kabsch, W. (2010). Acta Cryst. D66, 125–132. Web of Science CrossRef CAS IUCr Journals Google Scholar
Kobe, B., Guncar, G., Buchholz, R., Huber, T., Maco, B., Cowieson, N., Martin, J. L., Marfori, M. & Forwood, J. K. (2008). Biochem. Soc. Trans. 36, 1438–1441. Web of Science CrossRef PubMed CAS Google Scholar
Koschut, D., Richert, L., Pace, G., Niemann, H. H., Mély, Y. & Orian-Rousseau, V. (2016). Biochim. Biophys. Acta, 1863, 1552–1558. CrossRef CAS PubMed Google Scholar
Kovalevskiy, O., Nicholls, R. A., Long, F., Carlon, A. & Murshudov, G. N. (2018). Acta Cryst. D74, 215–227. Web of Science CrossRef IUCr Journals Google Scholar
Krissinel, E. & Henrick, K. (2007). J. Mol. Biol. 372, 774–797. Web of Science CrossRef PubMed CAS Google Scholar
Kühlbrandt, W. (2014). Science, 343, 1443–1444. Web of Science PubMed Google Scholar
Liebschner, D., Afonine, P. V., Baker, M. L., Bunkóczi, G., Chen, V. B., Croll, T. I., Hintze, B., Hung, L.-W., Jain, S., McCoy, A. J., Moriarty, N. W., Oeffner, R. D., Poon, B. K., Prisant, M. G., Read, R. J., Richardson, J. S., Richardson, D. C., Sammito, M. D., Sobolev, O. V., Stockwell, D. H., Terwilliger, T. C., Urzhumtsev, A. G., Videau, L. L., Williams, C. J. & Adams, P. D. (2019). Acta Cryst. D75, 861–877. Web of Science CrossRef IUCr Journals Google Scholar
Lingnau, A., Domann, E., Hudel, M., Bock, M., Nichterlein, T., Wehland, J. & Chakraborty, T. (1995). Infect. Immun. 63, 3896–3903. CrossRef CAS PubMed Google Scholar
Machner, M. P., Frese, S., Schubert, W. D., Orian-Rousseau, V., Gherardi, E., Wehland, J., Niemann, H. H. & Heinz, D. W. (2003). Mol. Microbiol. 48, 1525–1536. CrossRef PubMed CAS Google Scholar
Marino, M., Banerjee, M., Jonquières, R., Cossart, P. & Ghosh, P. (2002). EMBO J. 21, 5623–5634. CrossRef PubMed CAS Google Scholar
Marino, M., Braun, L., Cossart, P. & Ghosh, P. (1999). Mol. Cell, 4, 1063–1072. Web of Science CrossRef PubMed CAS Google Scholar
Mateja, A., Devedjiev, Y., Krowarsch, D., Longenecker, K., Dauter, Z., Otlewski, J. & Derewenda, Z. S. (2002). Acta Cryst. D58, 1983–1991. Web of Science CrossRef CAS IUCr Journals Google Scholar
McCoy, A. J., Grosse-Kunstleve, R. W., Adams, P. D., Winn, M. D., Storoni, L. C. & Read, R. J. (2007). J. Appl. Cryst. 40, 658–674. Web of Science CrossRef CAS IUCr Journals Google Scholar
Mueller, U., Förster, R., Hellmig, M., Huschmann, F. U., Kastner, A., Malecki, P., Pühringer, S., Röwer, M., Sparta, K., Steffien, M., Ühlein, M., Wilk, P. & Weiss, M. S. (2015). Eur. Phys. J. Plus, 130, 141. Web of Science CrossRef Google Scholar
Neves, D., Job, V., Dortet, L., Cossart, P. & Dessen, A. (2013). J. Mol. Biol. 425, 4520–4529. Web of Science CrossRef CAS PubMed Google Scholar
Nguyen, V. S., Jobichen, C., Tan, K. W., Tan, Y. W., Chan, S. L., Ramesh, K., Yuan, Y., Hong, Y., Seetharaman, J., Leung, K. Y., Sivaraman, J. & Mok, Y. K. (2015). Structure, 23, 2022–2031. CrossRef CAS PubMed Google Scholar
Niemann, H. H. (2013). Biochim. Biophys. Acta, 1834, 2195–2204. CrossRef CAS PubMed Google Scholar
Niemann, H. H., Jäger, V., Butler, P. J. G., van den Heuvel, J., Schmidt, S., Ferraris, D., Gherardi, E. & Heinz, D. W. (2007). Cell, 130, 235–246. Web of Science CrossRef PubMed CAS Google Scholar
Pizarro-Cerdá, J., Kühbacher, A. & Cossart, P. (2012). Cold Spring Harb. Perspect. Med. 2, a010009. PubMed Google Scholar
Schubert, W. D., Göbel, G., Diepholz, M., Darji, A., Kloer, D., Hain, T., Chakraborty, T., Wehland, J., Domann, E. & Heinz, D. W. (2001). J. Mol. Biol. 312, 783–794. Web of Science CrossRef PubMed CAS Google Scholar
Seveau, S., Tham, T. N., Payrastre, B., Hoppe, A. D., Swanson, J. A. & Cossart, P. (2007). Cell. Microbiol. 9, 790–803. CrossRef PubMed CAS Google Scholar
Shen, Y., Naujokas, M., Park, M. & Ireton, K. (2000). Cell, 103, 501–510. Web of Science CrossRef PubMed CAS Google Scholar
Trusolino, L., Bertotti, A. & Comoglio, P. M. (2010). Nat. Rev. Mol. Cell Biol. 11, 834–848. CrossRef CAS PubMed Google Scholar
Winn, M. D., Ballard, C. C., Cowtan, K. D., Dodson, E. J., Emsley, P., Evans, P. R., Keegan, R. M., Krissinel, E. B., Leslie, A. G. W., McCoy, A., McNicholas, S. J., Murshudov, G. N., Pannu, N. S., Potterton, E. A., Powell, H. R., Read, R. J., Vagin, A. & Wilson, K. S. (2011). Acta Cryst. D67, 235–242. Web of Science CrossRef CAS IUCr Journals Google Scholar
Xu, Q., Canutescu, A. A., Wang, G., Shapovalov, M., Obradovic, Z. & Dunbrack, R. L. (2008). J. Mol. Biol. 381, 487–507. CrossRef PubMed CAS Google Scholar
This is an open-access article distributed under the terms of the Creative Commons Attribution (CC-BY) Licence, which permits unrestricted use, distribution, and reproduction in any medium, provided the original authors and source are cited.