The Common Structural Architecture of Shigella flexneri and Salmonella typhimurium Type Three Secretion Needles

The Type Three Secretion System (T3SS), or injectisome, is a macromolecular infection machinery present in many pathogenic Gram-negative bacteria. It consists of a basal body, anchored in both bacterial membranes, and a hollow needle through which effector proteins are delivered into the target host cell. Two different architectures of the T3SS needle have been previously proposed. First, an atomic model of the Salmonella typhimurium needle was generated from solid-state NMR data. The needle subunit protein, PrgI, comprises a rigid-extended N-terminal segment and a helix-loop-helix motif with the N-terminus located on the outside face of the needle. Second, a model of the Shigella flexneri needle was generated from a high-resolution 7.7-Å cryo-electron microscopy density map. The subunit protein, MxiH, contains an N-terminal α-helix, a loop, another α-helix, a 14-residue-long β-hairpin (Q51–Q64) and a C-terminal α-helix, with the N-terminus facing inward to the lumen of the needle. In the current study, we carried out solid-state NMR measurements of wild-type Shigella flexneri needles polymerized in vitro and identified the following secondary structure elements for MxiH: a rigid-extended N-terminal segment (S2-T11), an α-helix (L12-A38), a loop (E39-P44) and a C-terminal α-helix (Q45-R83). Using immunogold labeling in vitro and in vivo on functional needles, we located the N-terminus of MxiH subunits on the exterior of the assembly, consistent with evolutionary sequence conservation patterns and mutagenesis data. We generated a homology model of Shigella flexneri needles compatible with both experimental data: the MxiH solid-state NMR chemical shifts and the state-of-the-art cryoEM density map. These results corroborate the solid-state NMR structure previously solved for Salmonella typhimurium PrgI needles and establish that Shigella flexneri and Salmonella typhimurium subunit proteins adopt a conserved structure and orientation in their assembled state. Our study reveals a common structural architecture of T3SS needles, essential to understand T3SS-mediated infection and develop treatments.

T3SS are formed by the assembly of ,20-25 different proteins, some of these in multiple copies. They consist of a hollow needle and a basal body which anchors the needle to both bacterial membranes [2,[8][9][10]. The needle extends into the extracellular space and ends in a tip that may make contact with the host cell [11][12][13][14]. Upon contact, translocator proteins form a pore through which effector proteins enter and subsequently alter host cell processes during infection [7][8]. The needle is assembled from multiple copies of a subunit protein, namely MxiH in Shigella, PrgI in Salmonella and BsaL in Burkholderia bacteria. Earlier studies of Shigella flexneri T3SS needles by X-ray fiber diffraction and cryoelectron microscopy (cryoEM) showed that the MxiH subunits are disposed in a helical arrangement, with 5.6 subunits per turn and a 24-Å helical pitch [15]. The atomic structures of C-terminal truncation mutants, resulting in non-functional soluble needle subunits, were determined for BsaL CD5 by solution-state NMR [16], MxiH CD5 by X-ray crystallography [17], and PrgI CD5 by solution-state NMR [18]. All three structures exhibit two a-helical segments separated by a loop at the P-(S/D)-(D/N)-P motif.
In order to gain further insight into the function of Type Three Secretion Systems, it is necessary to obtain an atomic structure of wild-type (WT) needles in their assembled state. The pseudoatomic T3SS needle architecture of Shigella was modeled by Deane and coworkers [17] by docking semi-rigid MxiH CD5 crystal monomers into a 16-Å cryoEM density map [15], with a topology where the N-terminus of the MxiH subunit is facing the inside of the tubular structure. More recently, a refined model of the Shigella MxiH needle was presented by Fujii and coworkers [19] based on a 7.7-Å cryo-electron microscopy (cryoEM) map. In this model (Fig. 1A), the segment ranging from Q51 to Q64, observed as ahelical in the X-ray structure [17], was remodeled as a b-hairpin structure during the molecular docking, while preserving the inwards-facing N-terminus. We have recently presented an atomic model of the Salmonella typhimurium T3SS needle [20] based on solid-state NMR distance restraints collected directly on assembled needles and symmetry from scanning transmission electron microscopy (STEM) data [21]. In this atomic model (Fig. 1B), the N-terminus of the subunit is located at the surface of the needle, as confirmed by immunogold labeling experiments performed both in vitro and in vivo [20].
Although the two proposed models show fundamental differences in their topology, MxiH and PrgI present a high sequence identity (.60%, Fig. 2). From these two models, strong experimental predictions regarding the orientation of termini and the secondary structure present in S. flexneri T3SS needles can be made and experimentally verified. In the current study, we thus present solid-state NMR data and immunogold labeling results obtained on WT needles of S. flexneri which allow us to address the discrepancy between the two models and establish the common architecture of T3SS needles.
The solid-state NMR assignment of 13 C and 15 N chemical shifts (supplementary Table S2 in Text S1) was carried out using standard experiments, 2D 13 C-13 C, 15 N-13 C and 15 N-( 13 C)-13 C experiments. Due to the high spectral resolution of the sample, most of the assignment could be carried out on uniformly-labeled ([U-13 C 6 ]Glc) needles using 2D spectroscopy. An example of the spectral resolution obtained in [U-13 C 6 ]Glc needles is presented in a 13 C-13 C proton-driven spin diffusion (PDSD) spectrum (supplementary Fig. S1 in Text S1). A near-complete resonance assignment (96%) was obtained using an extensive dataset of 2D spectra (supplementary Table S1 in Text S1) recorded on [U-13 C 6 ]Glc and 13 C spin diluted (produced with [1-13 C]Glcand [2-13 C]Glc) T3SS needles. In sparsely-labeled samples, the dilution of 13 C spins results in an improved 13 C resolution due to the removal of one-bond 13 C-13 C dipolar and J CC couplings, and in the simplification of spectra [37] (Fig. 3C). For instance, the complementarity of the [1-13 C]Glc-and [2-13 C]Glc labeling patterns [37][38][39][40] enabled the facile assignment of the repetitive LSSLS motif (L12-S16) of MxiH.
Residues extending from S2 to R83 in the MxiH sequence are all observed in cross-polarization based solid-state NMR spectra, indicating a high structural rigidity. Consequently, with the exception of Met1 detected in an INEPT 1 H-13 C spectrum, all residues in the MxiH subunit protein exhibit a rigid conformation.
No peak doubling could be observed in solid-state NMR spectra (Figs. 3C and S1 in Text S1), denoting the absence of polymorphism in the needle samples. This is in contrast to the X-ray structure of monomeric MxiH CD5 (PDB ID: 2CA5) where two distinct conformations of MxiH are present in the unit cell [17], molecules A and B. Conformation B, which was not consistent with the low-resolution EM density, had been attributed either to crystal packing or to a different activation state of the needle [17].
Shigella flexneri and Salmonella typhimurium needle proteins adopt a similar conformation in the assembled state As backbone chemical shifts are strongly correlated to secondary structure [41][42][43], the secondary structure elements present in the MxiH subunit can be identified from our assignment

Author Summary
Gram-negative bacteria use a molecular machinery called the Type Three Secretion System (T3SS) to deliver toxic proteins to the host cell. Our research group has recently solved the structure of the extracellular T3SS needle of Salmonella typhimurium. Employing solid-state NMR, we could determine local structure parameters such as dihedral angles and inter-nuclear proximities for this supramolecular assembly. Concurrently, a high-resolution cryo-electron microscopy density map of the T3SS needle of Shigella flexneri was obtained by Fujii et al. Modeling of the Shigella needle subunit protein to fit the EM density produced a model incompatible with the atomic model of the Salmonella needle in terms of secondary structure and subunit orientation. Here, we determined directly the secondary structure of the Shigella needle subunit using solid-state NMR, and its orientation using in vitro and in vivo immunogold labeling in functional needles. We found that Shigella subunits adopt the same secondary structure and orientation as in the atomic model of Salmonella, and we generated a homology model of the Shigella needle consistent with the EM density. Knowing the common T3SS needle architecture is essential for understanding the secretion mechanism and interactions of the needle with other components of the T3SS, and to develop therapeutics.
b-sheet propensity [43]. The analysis of Dd thus reveals two stretches of predominantly positive values corresponding to two long a-helices. The first a-helix is 27-residues long (L12 to A38), the second a-helix comprises 39 residues and extends up to the Cterminus (Q45 to R83). The intervening region of negative values (E39 to P44) corresponds to a rigid loop and contains the conserved P-(S/D)-(D/N)-P turn structure previously identified in structures of monomeric [16][17][18]44] and assembled [19][20] T3SS needle subunits. Kinks or bends in the a-helices can be inferred from the isolated negative values of Dd at T23 and Q64. The Nterminal segment from S2 to T11 is not a-helical but adopts instead a more extended conformation, as identified from the presence of negative Dd values.
Local protein conformations can be directly compared on the basis of the secondary chemical shifts, as they are not influenced by amino acid type. The similarity in secondary chemical shifts between the MxiH and PrgI needles is remarkable (Figs. 4C and 4D), suggesting that the local structure of these needle subunits is highly similar. The agreement is especially apparent in the two ahelical regions, and the slight bend in the C-terminal helix occurs at the same position in both subunits, Q64 for MxiH corresponding to Q61 for PrgI. Two minor differences are observed: the kink appears at T23 for MxiH rather than at T25 (which would correspond to N22 in PrgI); more importantly, the rigid-extended region spans a larger region in MxiH, which has three additional residues, and the deviations of Dd from random coil values are more pronounced. On the other hand, the secondary structure elements proposed in the model of Fujii et al. are not compatible with the experimental chemical shifts of MxiH (Fig. 4A). In order to quantitatively compare the cryoEM model to our chemical shift values, we back-predicted secondary chemical shifts from the cryoEM model ( Fig. S2B in Text S1) using SPARTA+ [45]. It was reported that the back-prediction of solid-state NMR 13 C chemical shifts in SPARTA produces outliers, i.e. predictions deviating from experimental shifts by more than 3 ppm, in approximately 4% of cases [46]. The number of 13 C chemical shift outliers for the cryoEM-based model of Fujii et al., 13.1%, is significantly higher, showing that its secondary structure cannot be reconciled with our NMR data.
Despite the aforementioned discrepancies in secondary structure between the MxiH [19] and PrgI [20] models, there is a striking similarity in the shape of the 3D envelope of their constituting subunits (Fig. 1). We thus wanted to test whether the architecture identified in PrgI needles could explain all experimental data for MxiH needles, both solid-state NMR chemical shifts and the 7.7-Å cryoEM density map. We generated homology models of the MxiH needle by Rosetta modeling calculations [47] using the structure of the PrgI needle as a template. A representative model from this ensemble (Fig. 5) shows good overlap with the cryoEM density map (EM correlation of 0.66), The N-terminus is located at the surface of the needle. Top views of the needle assemblies are shown in the top corners. The structures in the bottom corners present the location of highly conserved residues that were identified from multiple sequence alignment (Fig. 2). The atoms of conserved amino acids are represented as spheres and are colored respectively blue or red depending if their amino acid belongs to the N-or the C-terminal region. Conserved residues are lining the lumen of the needle in the Salmonella model but are exposed to the extracellular milieu in the Shigella model. doi:10.1371/journal.ppat.1003245.g001 demonstrating that the architecture found in the PrgI needle ( Fig. 1B) is compatible with all experimental data. Using the nomenclature of Fujii et al., the long C-terminal a-helix identified in PrgI occupies the density region H1 facing towards the needle pore in the cryoEM map of MxiH (Fig. 5B). The density regions H2 and H3 are filled by the N-terminal a-helix. The kink present in density region H3 corresponds to the kink identified at T23 from the analysis of secondary chemical shifts.
The N-terminus of MxiH is exposed on the exterior of the assembly In order to determine the placement of the N-and the Cterminus in the MxiH needle, immunogold labeling experiments were performed both in vitro and in vivo. For in vitro experiments, the MxiH protein was expressed as an N-terminal His tag fusion construct, mxiH-N(His), and needles were polymerized. The tag is recognized by an Anti-His tag antibody which is imaged using protein A Gold (Fig. 6B). Immunogold labeling was not observed for WT needles lacking the His tag (Fig. 6A) and for mxiH-N(His) needles incubated without the Anti-His tag antibody (Fig. 6C), excluding the possibility that the labeling resulted from nonspecific binding of the primary antibody or protein A Gold. mxiH expressed as a C-terminal His tag fusion construct prevented needle polymerization as previously reported [44]. Similar results were obtained for in vivo immunogold labeling experiments, where DmxiH Shigella cells were complemented with WT mxiH (Fig. 6E), an N-terminal Strep-tag fusion construct, mxiH-N(Strep), (Fig. 6F) or a C-terminal Strep-tag fusion construct, mxiH-C(Strep), (Fig. 6G). The strains expressing WT mxiH and mxiH-N(Strep) could assemble secretion-competent needles, as shown in a protein secretion assay (Fig. 6D), but the strain expressing mxiH-C(Strep) did not assemble any needle (Fig. 6G) and was not competent for secretion, consistent with previous mutagenesis data showing that C-terminal insertions compromise needle assembly and secretion function [48]. The strain expressing mxiH-N(Strep) formed needles decorated with gold particles. These results show that the MxiH subunit N-terminus is exposed at the surface of the needle and corroborate that MxiH needles adopt the topology found in the PrgI needle solid-state NMR structure [20] but not the topology of previous MxiH needle models [17,19].
The common structural architecture of Shigella flexneri and Salmonella typhimurium needles is reflected in sequence conservation patterns As T3SS are employed by a variety of bacterial pathogens [3], we generated a multiple sequence alignment of T3SS needle proteins to evaluate the generality of the needle architecture. The sequences group in two clusters based on a pattern of highly conserved residues in the loop and C-terminal region (Fig. 2). Three distinct regions of the protein are identified according to the conservation level: the first ,10 residues forming the N-terminal extension are poorly conserved (average conservation level of 2.1), the subsequent N-terminal region is moderately conserved (6.5), and the loop (7.2) and C-terminal region (8.7) are highly conserved. Both MxiH and PrgI belong to the same cluster, in contrast to the needle protein of Yersinia pestis, YscF, which shows lower sequence identity to MxiH (25%). Sequences for EscF proteins are more evolutionary distant and were not included in any of the clusters. They also lack the canonical P-(S/D)-(D/N)-P loop motif and the variable extension.
The loop and C-terminal region contain a large number of residues, 11, that are highly conserved in both clusters, compared to only two residues in the N-terminal region. Those highly conserved residues, indicated with stars in Fig. 2, are all located lining the lumen of the needle in the PrgI needle structure (Fig. 1B,  bottom right corner); however, no conserved residues are present in the interior of the needle in the cryoEM MxiH model (Fig. 1A,  bottom left corner). Conserved residues are colored according to amino acid type following the Clustal X color scheme [75]. Highly conserved residues ($12 identical amino acids in both clusters) are highlighted with a star (N-terminal region: blue; C-terminal region: red). The numbering of residues corresponds to the sequence of MxiH. Database accession identifiers for the primary sequences are given in supplementary Table S3 in Text S1. doi:10.1371/journal.ppat.1003245.g002

Discussion
In the current study, WT S. flexneri needle assemblies were prepared by in vitro polymerization and studied by solid-state NMR. The MxiH needle subunit is predominantly rigid (S2-R83). Our chemical shift assignment (supplementary Table S2 in Text S1) provides for the first time empirical information on the conformation of the MxiH termini in the context of the needle assembly, as the first 19 (molecule A) or 14 (molecule B) residues are disordered in the crystal structure of MxiH CD5 , as well as the last 2 (molecule A) or 8 (molecule B) C-terminal residues [17].
The following secondary structure elements were identified for MxiH: a rigid-extended N-terminal conformation (S2-T11), followed by the N-terminal a-helix (L12-A38) with a kink at T23, a loop (E39-P44) and the C-terminal a-helix (Q45-R83). The conformation of Shigella MxiH and Salmonella PrgI subunits in their needle assembled state is highly similar, as revealed by the comparison of secondary chemical shifts (Figs. 4C and 4D). Equally, the structure of the PrgI subunit solved by solid-state NMR comprises a rigid-extended N-terminal conformation (T3-Y8), an a-helix (L9-A35) with a kink at N22, a loop (A36-P41), and a Cterminal a-helix (A42-R80).
In contrast, the chemical shifts observed in this work are incompatible with the secondary structure identified in the previous MxiH needle model [19] (Figs. 4A and S2 in Text S1). Two major discrepancies exist between this model (Fig. 1A) and the experimental chemical shifts: the first 11 N-terminal residues were previously modeled as an a-helix and the segment extending from residues Q51 to Q64 was previously modeled as a b-hairpin. One explanation for the discrepancy in secondary structure elements in MxiH subunits between the current study and the model of Fujii et al. is that the measurements were carried out on S. flexneri serotype 6 (this study) and serotype 2. Compared to serotype 6, the sequence of serotype 2 has the following mutations: D7N, K8D, Q29E, S32L, N35D, A36K and E39K. Since no amino acid substitutions between the two serotypes are located in the conjectured b-hairpin region of MxiH (Q51-Q64), rearrangement of the secondary structure of the C-terminal helix by these mutations would likely involve even greater alterations locally at the site of the mutation, which are not observed. The finding that PrgI sustains a significantly greater number of N-terminal substitutions while preserving a C-terminal conformation highly similar to MxiH further discredits this hypothesis. We conclude that the incompatibility between this model and the experimental data results from remodeling of the monomeric subunit structure to fit the cryoEM density map without the guidance of site-specific local structural information such as experimentally-determined dihedral angles and distance restraints.
As shown here and recently for the T3SS needle of S. typhimurium [20], solid-state NMR data can be used in complementation to cryoEM data for the generation or the validation of structural models of large assemblies. Owing to the high similarity of secondary chemical shifts between PrgI and MxiH, we generated a homology model of S. flexneri needles which accounts for all available experimental data, including the 7.7-Å cryoEM density map. Although the correlation to the cryoEM density is higher for the previous model by Fujii et al. (0.72) compared to our PrgI-based homology model (0.66), this is expected as no EM bias was used in the building of the current homology model. The previous model also has significantly higher full-atom energy after Rosetta refinement (Table 1). It is notable that the symmetry and helical parameters of the PrgI-based homology model are highly similar to the parameters derived from the cryoEM density map (Table 1), consistent with the view that Shigella and Salmonella needles are built upon a common architecture. From a methodological point of view, it is also interesting that both independent procedures, solid-state NMR in combination with results from STEM and Rosetta modeling -and high-resolution cryoEM, result in the same obtained symmetry. Furthermore, the fact that the homology model of PrgI fits well into the completely independently and simultaneously determined high-resolution cryoEM map (Fig. 5) supports the validity of the PrgI structure (Fig. 1B).
Using immunogold labeling, we confirmed that the N-terminus is oriented on the outside face of the needle assembly (Fig. 6). This subunit orientation is consistent with sequence conservation patterns observed by multiple sequence alignment (Fig. 2), where residues in the C-terminal helix are more frequently conserved, and with mutagenesis data, where sequence alterations are more easily tolerated at the N-terminus. Indeed, deletion of 2-5 residues or addition of 3-6 residues at the C-terminus results in dysfunctional needles as detected in effector secretion assays and Congo Red (CR) induction assays [19,48] for MxiH, or in epithelial cell invasion assays [44] for PrgI. In contrast, deletion of 2-3 residues or addition of 3-7 residues at the N-terminus of MxiH subunits still produces functional needles (Ref. [19] and Fig. 6D). Moreover, PrgI subunits still assemble into functional T3SS needles when expressed with a deletion or addition of six residues at their N-terminus [44] as well as with a 8-residue Nterminal Strep-tag [20], as demonstrated by epithelial cell invasion assays. A deletion of eight N-terminal residues resulted in a S. typhimurium strain incapable of invasion [44], indicating that the two residues G7 and Y8 must be preserved to form the axial interface connecting subunits (i) and (i611) and are necessary for the polymerization of PrgI subunits [20].
As the biological role of the T3SS needle is to ensure the secretion of effector proteins, it is consequent that the residues lining the lumen, directly involved in the transport of effectors, are highly conserved (Fig. 1B). On the other hand, the diversity of residues exposed at the surface of the needle can confer a selective advantage to pathogenic bacteria, which must evade recognition by host cells [7,[49][50]. Accordingly, the mxiH gene was shown to be the most variable gene in the virulence plasmid of S. flexneri [51].
In their assembled state, the only characteristic structural distinction observed between the homologous MxiH and PrgI proteins on the basis of the secondary structure analysis is the length of the N-terminal rigid-extended segment, which is composed of 10 residues for MxiH and six for PrgI. An important observation for the understanding of needle elongation is that T3SS needle proteins from different species are not interchangeable. For instance, Yop effector secretion could not be restored in a yscF null mutant strain of Yersinia enterocolitica by the expression of MxiH [52], which can be attributed to the low sequence identity of the two sequences (25%). However, despite the high sequence identity between MxiH and PrgI (.60%), a Shigella flexneri mxiH null mutant strain could not be complemented by PrgI and was not invasive due to the lack of needle formation [18]. Similarly, purified T3SS needles could be elongated in vitro from soluble monomer subunits of their own species, but not from subunits of other species, Salmonella typhimurium with MxiH* and Shigella flexneri with PrgI* [44]. Differences in the electrostatic surfaces from soluble monomeric subunit structures were suggested to explain the incompatibilities between needle subunits [18]. Our findings hint that the N-terminal rigid-extended segment could also play a crucial role in determining the species specificity. The excellent NMR spectral quality of needle samples produced by the in vitro preparation protocol [20] paves the way for the determination of high-resolution structures of MxiH filaments. Ongoing investigations in our group will thus address the question of the atomic structure of the rigid-extended N-terminal segment in MxiH needles.
In this study, it was found that Shigella flexneri and Salmonella typhimurium needles present a common architecture, providing a structural basis essential for the interpretation of previous functional data on type-three secretion in Shigella. Knowledge of the correct needle topology is essential to develop a key understanding of the operating mechanism of T3SS. The topology has an impact on the study of needle-tip interactions [53][54][55][56][57], needle-basal body interactions [58][59][60][61][62] and regulation of secretion [48,[63][64][65][66], which can contribute to the development of treatments against T3SS-mediated pathologies.

Sample preparation
MxiH was expressed and purified following the protocol established for S. typhimurium PrgI needles [20]. Similarly, the Nterminal hepta-Histidine (His) tag was cleaved using tobacco etch virus protease, releasing MxiH proteins containing the non-native N-terminal residues glycine and histidine. The protein concentration was raised to 0.2 mM during polymerization, which took place at 37uC for sixteen days. Approximately 20 mg of needle proteins were produced for each of three labeling schemes. All three samples were labeled with 15

Solid-state NMR
MxiH needles were ultra-centrifuged and transferred into 4.0mm magic-angle spinning rotors. Solid-state NMR experiments were conducted at 600 MHz, 800 MHz and 850 MHz 1 H frequencies on Avance I and Avance III spectrometers (Bruker Biospin, Germany) at a spinning rate of 11 kHz and a temperature of 5.5uC. Additional experimental details can be found in supplementary Table S1 in Text S1. For the calculation of secondary chemical shifts (Dd), average random coil chemical shifts were taken from Ref. [43].

Immunogold labeling
Recombinantly-produced MxiH needles were labeled using an Anti-penta-Histidine tag monoclonal antibody (Invitrogen). The labeling was performed using a 1:200 diluted antibody and 10-nm protein A Gold (Posthuma, Utrecht). Non-polar Shigella mxiHknockout cells (DmxiH) were complemented with either wild-type mxiH, mxiH with N-terminal Strep-tag or mxiH with a C-terminal Strep-tag in pASK-IBA vectors (IBA lifesciences) and induced with 0.2 mg/ml anhydrotetracycline for 1 hr. The cells were fixed in 2% paraformaldehyde and immunogold labeled using 1:40 diluted StrepMAB-Classic (IBA lifesciences) for 1 hr followed by 1:40 diluted 12-nm gold-conjugated secondary antibody for 1 hr. Protein secretion assays were performed by addition of 20 mg/ ml CR followed by western blot analysis against IpaB, IpaC and DnaK.

Multiple sequence alignment
Sequences of commonly studied T3SS needle proteins were aligned using Clustal W version 2 [67] with default settings and the GONNET substitution matrix [68]. A first round of alignment revealed two clusters of highly conserved patterns in the loop and C-terminal region with the N-terminal regions presenting a higher variability. In a second round, the two sequence clusters were aligned as two separate groups. The conservation level and consensus sequence for each cluster were calculated using JalView [69]. Uniprot accession identifiers for the primary sequences are specified in supplementary Table S3 in Text S1. Pairwise sequence alignments used in the Rosetta modeling were generated using BLAST [70].

Rosetta modeling
We used the Rosetta symmetric modeling framework [71] to generate a full 29-subunit model of MxiH based on the solid-state NMR structure of the PrgI needle (PDB ID: 2LPZ). This framework makes conformational sampling in symmetric systems The first 11 N-terminal residues are not depicted due to the poor sequence homology of the rigid-extended segment (Fig. 2); however the protrusion in the EM map at subunit (i), labeled prot. and highlighted by a purple dashed contour trace, may be well explained by the N-terminus of subunit (i+5). The central subunit (i), colored as in Fig. 1B, forms a lateral interface with subunits (i65), in red, and subunits (i66), in dark blue. The axial interface is formed with subunits (i611) shown in pink. doi:10.1371/journal.ppat.1003245.g005 more efficient by 1) only considering conformations that are consistent with the symmetry of the system and 2) performing a minimal number of energy and derivative evaluations by explicitly simulating only a subset of the interacting monomers and propagating conformational changes to symmetry-related subunits. The PrgI template was used to define the initial rigid-body placement and backbone conformation of the MxiH subunits as well as their relative orientation with respect to each other, Figure 6. Localization of the MxiH subunit N-terminus in vitro and in vivo. A-C, Immunogold labeling of recombinant Shigella flexneri MxiH needles using a monoclonal Anti-His tag antibody. A, WT MxiH needles B, needles polymerized from N-terminal His tag fusion construct, mxiH-N(His), C, mxiH-N(His) needles labeled without primary Anti-His tag antibody. In B, some gold particles are indicated by arrows. Proteins expressed from a Cterminal His tag fusion construct did not polymerize. D, Effector protein secretion assay of Shigella strains used in Fig. E-G by addition of Congo Red (CR) followed by western blot analysis against IpaB, IpaC and DnaK. E-G, Non-polar Shigella mxiH-knockout cells (DmxiH) expressing E, wild-type mxiH, F, mxiH with N-terminal Strep-tag, mxiH-N(Strep) or G, MxiH with C-terminal Strep-tag, mxiH-C(Strep). In E and F, needles are indicated by arrows and two needles complexes are highlighted. doi:10.1371/journal.ppat.1003245.g006 therefore dictating the helical symmetry parameters. Unaligned residues at the N-termini of the subunits were remodeled using symmetric, Monte Carlo based 3-and 9-residue backbone fragment insertions in a coarse-grained representation of a helically-arrayed 29-subunit system. The entire structure, including all backbone, side chain and rigid body degrees of freedom was subsequently refined using a physically realistic all-atom energy function [72]. 1,000 homology-based models were generated of which the 10 lowest in energy showed strong convergence with 0.6 Å backbone RMSD for the entire complex, excluding the 11 residues comprised in the N-terminal variable extension. The final models did not diverge significantly from the starting PrgI template structure (0.8 Å backbone RMSD) and present almost identical helical parameters to the initial template (5.6 subunits/ turn, 23.5 Å radius, 4.2 Å axial rise/monomer). Evaluation of the full-atom energy of the previous cryoEM model by Fujii et al. was done in the following manner: the full 29-subunit assembly was generated from the coordinates of the monomeric subunit using the symmetry-related operations described in the header of the PDB file (PDB ID: 3J0W), followed by the aforementioned symmetric refinement in Rosetta. The refinement led only to minor changes relative to the starting coordinates (,1 Å backbone RMSD for the rigid helical parts of the structure, ,0.5 Å change in axial rise/monomer and radius, ,10u in rotation/monomer). 100 models were generated in this manner and the median energy of the 10 lowest-energy models was reported ( Table 1). Fitting of the homology model to the EM density map and calculation of the EM correlations were done using the program UCSF Chimera [73][74].

BMRB
The 15 N-and 13 C-chemical shift assignments for MxiH, the T3SS needle protein of Shigella flexneri, were deposited at the Biological Magnetic Resonance Data Bank (BMRB) under accession number 18651.

Supporting Information
Text S1 Supporting information. Figure S1, solid-state NMR 13 C-13 C spectrum of MxiH needles. Figure S2, backprediction of secondary chemical shifts.