Electron microscopy of the VcGspD-B₅ complex.

Recombinant full-length VcGspD and CTB₅ were each expressed in E. coli and purified to homogeneity for electron microscopy as described in reference 21. VcGspD (50 µgml⁻¹) was either applied directly to a negatively charged carbon coated copper EM grid or first incubated with CTB₅ (7.5 µgml⁻¹) for 30 minutes at 25°C. Proteins were negatively stained with uranyl formate and visualized on a 100 kV transmission electron microscope (FEI Morgagni) as previously described in reference 23. Images were recorded on a 2k × 4k CCD camera (Gatan) at a nominal magnification of 44,000x at the specimen level (Fig. 1B and C). For image processing, micrographs were digitally binned twice to yield a final pixel size of 4.1 Åpixel⁻¹. Particles were picked manually in Ximdisp24 and processed in SPIDER25 to generate two-dimensional (2D) reference free class averages (Fig. 1B and C, insets).

The class averages of VcGspD particles oriented perpendicular to the channel axis (side view) reveals a protein density that is consistent with the features observed in our cryoEM reconstruction (compare Fig. 1A and B, inset). The periplasmic domain is 125 Å long and forms a large unobstructed periplasmic vestibule with a constriction located just below the periplasmic gate. The periplasmic gate appears in a closed conformation, effectively separating the periplasmic vestibule from the extracellular chamber. The extracellular chamber is capped at the top by a weak protein density corresponding to the extracellular gate that was observed in the VcGspD cryoEM reconstruction.21 The identification of these features suggests the channel architecture is well preserved under the conditions used for EM of negatively stained specimens.

When VcGspD was mixed with CTB₅ we observed two distinct groups of particles in the class averages. The first group (corresponding to ∼35% of all side views) appeared similar to class averages of free VcGspD, in that they contained an unobstructed periplasmic vestibule (Fig. 1B and inset). However, a second group of class averages (corresponding to ∼45% of all side views) clearly contained an additional punctate density at the entrance of the periplasmic vestibule (Fig. 1C and inset arrow head). This density has approximate dimensions 60 × 40 Å that correspond well to the size of the B-pentamer. Therefore, our EM analysis suggests that CTB₅ binds within the VcGspD periplasmic vestibule. The distribution of free and CTB₅-bound VcGspD particles is consistent with the relatively weak binding affinity observed between CTB₅ and the periplasmic domain of GspD in solution (K_D 15.5 ± 1.7 µM),21 and the low protein concentrations required for EM sample preparation.

Molecular model of the secretin-toxin complex.

The class averages obtained from the CTB₅-bound VcGspD particles were used to guide the placement of the complete cholera toxin AB₅ heterohexamer (PDB ID 1S5E),26 within the cryoEM reconstruction of VcGspD to obtain further structural insight into the secretion mechanism. First, the atomic structure of CTB₅ was placed into the EM density obtained from the class averages of the VcGspD-CTB₅ complex (Fig. 2A). The CTB₅ structure overlays well with the punctate density located within the periplasmic vestibule of VcGspD. This placement positions the CTB₅ complex at the entrance of the VcGspD periplasmic vestibule, roughly 85 Å below the periplasmic gate. Based on this placement, we modeled the complete AB₅ cholera toxin molecule into the periplasmic vestibule of the VcGspD cryoEM reconstruction (Fig. 2B, left). The enormous 55.4 × 10⁴ ų periplasmic vestibule is capable of housing the 9.3 × 10⁴ ų AB₅ toxin. The resulting model places the cholera toxin just below the VcGspD periplasmic constriction. This placement is almost identical to what we proposed previously based on the steric consideration of the constriction diameter, which narrows the channel to 55 Å.21 Indeed, the B-pentamer is 65 Å wide, suggesting that the toxin could not pass through the constriction without major conformational changes occurring in the channel.

We propose that the VcGspD-toxin complex that we observed in this study corresponds to the priming stage in the piston-driven secretion mechanism (Fig. 2B).7,21 In this model, the toxin is bound within the periplasmic vestibule of the secretin with the channel gate in a closed conformation, forming a substrate loaded or primed state (Fig. 2B, left). Then, following this priming stage, a signal is somehow transferred via the inner membrane platform to the secretion ATPase GspE in the cytoplasm, leading to ATP hydrolysis and polymerization of pseudopilins and resulting in formation or extension of the pseudopilus.27,28 Eventually, the pseudopilus tip complex GspK-GspI-GspJ reaches the secretin and possibly interacts with the periplasmic domain of the channel, sealing the channel's entrance. We have previously shown by SPR that the trimeric pseudopilus tip complex interacts with the periplasmic domain of secretin.21 As the pseudopilus extends further—probably through continuing ATP-driven polymerization—it would act as a piston forcing the toxin to interact with the VcGspD constriction (Fig. 2B, middle). This interaction with the secretin constriction could provide the trigger that causes a conformation change in the channel to open the periplasmic gate. This conformational change may resemble the conformational change that was observed in the T3SS secretin following needle attachment.10,11,21 With the periplasmic gate open, the toxin may pass into the extracellular vestibule for secretion (Fig. 2B, right). Once the toxin is secreted, the pseudopilus may transiently remain extended, acting as a plug to prevent channel leakage. By an as yet unknown mechanism, the GspD periplasmic gate closes, and the pseudopilus retracts and/or disassembles partially or fully to reset the system. Whereas this model is in line with our current structural knowledge of the T2SS, a complete understanding of this mechanism will require future high-resolution structural studies of T2SS assembly and secretion.