Here we used assemblies of lysenin prepores and pores on supported lipid bilayers (SLB) composed of SM/Chol 1:1, to mimicking lipid raft domains composition22. First, we screened different monomer incubation concentrations. Remarkably, these incubations revealed that the incubation concentration directly affects the percentage of pores present in the ensemble. We focused on two incubation concentrations, a monomer concentration at 1.6 µM, where the pore state was predominant, being 78±2% of the total number of oligomers (Fig. 1a), and a monomer concentration at 5.6µM, where the pore state was in minority with 13±4% of pores (Fig. 1b). As the pore state is more abundant at lower levels of crowding and less at higher levels of crowding, where the prepore state is more abundant, we concluded that the prepore-to-pore transition got blocked at higher levels of crowding. This finding suggested that the prepore-to-pore transition required the presence of available space around the toxin prepores, or otherwise the prepore-to-pore transition got sterically inhibited. We also found that the incubation concentration affected the supramolecular organization on the membrane. Incubations at a concentration of 1.6µM resulted in clusters of lysenin pores surrounded by protein-free membrane areas (Fig. 1a), whereas incubations at 5.6µM resulted in hexagonal close packed (hcp) assemblies composed mainly of lysenin prepores (Fig. 1b). The negative hydrophobic mismatch of the β-barrels of the lysenin pores is the probable cause of the observed attraction between pores (details on Supplementary section).
To exclude uncontrolled effects derived from the presence of the mica substrate below the membrane1, we repeated the high concentration incubation on non-supported lipid bilayers (nSLB)23 of SM/Chol 1:1. The imposition of a flat surface to the membrane or the anchoring of individual lysenin oligomers10 could affect previous observations. The use of nSLB mimics the curvature and tension of the native environment of membranes in cells, and anchoring of oligomers is excluded14. The 5.6µM incubation on nSLB, with a 78% of lysenin oligomers in the prepore state, confirmed that the prepore state is dominant regardless of the substrate, the membrane tension or the curvature (Suppl. Fig. 1). The results on nSLB suggest that the steric inhibition is a general phenomenon of the crowded assemblies of lysenin.
To get further insight on the phenomenon of prepore-to-pore transition steric inhibition in lysenin toxin, we filmed at 0.45 seconds per frame the evolution of a SM:Chol 1:1 membrane exposed to lysenin monomers solution (Fig. 1d, Suppl. movie 1). We time-tracked changes in the occupancy of the membrane by monomers, prepores and pores. A pixel height count methodology was used in order to localize each population (details on Methods section). We evidenced a sequence of events that lead to steric inhibition, with five stages in the filling of the membrane (Fig. 1c). First, the arrival of lysenin monomers (Fig. 1d, t<13 s, in blue). Second, the beginning of oligomerization (Fig. 1d, t∼25 s, in green). Third stage, where most of the oligomerizations took place, in which we observed an abrupt decrease on the lysenin-free membrane area lead by monomer membrane occupancy. This stage, called oligomerization end, finish with an almost full coverage of membrane with lysenin oligomers (Fig. 1d, t<47 s, in pale green). Fourth, the steric blockage of prepore insertions, where the number of prepores and pores remained constant (Fig. 1d, t<70 s, orange). At this stage with a vast majority of Lysenin oligomers in the prepore stage where we think that the prepore-to-pore transition got sterically blocked. Moreover, at that time the oligomer assembly exhibited certain degree of dynamism. For tens of seconds the oligomers searched for a stable state until every oligomer position got fixed in the last stage, the 2D crystal assembly (Fig. 1d, t>70 s, red). We note that this time series represents one of the few data sets showing the process of 2D protein crystal growth24.
How did lysenin evolutionary overcome the steric blockage on the prepore-to-pore transition owing its ability to form pores?25. We first want to check if as suggested by Podobnik et al.18 data, prepore-to-pore transition liberate space. Following previous results in which a pH-decrease induced prepore-to-pore transitions in lysenin 2D prepore assemblies17, we induce new insertions and measured the occupy area by each population. We decided to use a membrane patch sufficiently small to be fully visualized by the HS-AFM. Previous HS-AFM works on lysenin were performed on large membrane patches where local effects could be hidden out of visualized area10 14 15 17. A small SM/Chol 1:1 patch (around 600nm in diameter) we performed 5.6µM lysenin incubation inducing the full filling of the patch. We counted ∼315 lysenin oligomers, out of which 95% were in prepore state. We induced the prepore-to-pore transition by a continuous pH decrease from 7.5 to 5.2 inducing ∼135 insertions, 43% of all prepores (Fig. 2a,b and suppl. movie 2). Along with the induced insertions we observed a decrease of the membrane area occupied by lysenin oligomers of ∼16%. Thanks to this results we check directly a membrane occupancy reduction by Lysenin oligomers induced by the prepore-to-pore transition. The prepore and prepore-to-pore transition models should, as previously suggested19, be modify to adequate our findings. Following our results the superimposition of lysenin monomeric structure on C-terminal domains of the pore structure that impose the shame pore and prepore diameter is not correct. Interestingly, the total area of the patch remained invariant during the insertions, this was combined with the appearance of seemly lipid protrusions on top of the pores (Suppl. Fig. 2). The lysenin seemly punches the membrane during the pore formation. This type of mechanism was recently identified in the suilysin toxin26.
In order to discard any possible electrostatic effect of pH change on the appearance of protein-free membrane area, we induced prepore insertions using single molecule compressions27. We performed force mapping experiments at constant pH of 7.5 in hcp lysenin assemblies presenting 94% of prepores. We imaged a 50x50 nm area where we increased the applied force of the AFM imaging from 300pN to 600pN. As a result, prepore-to-pore transitions were induced (Fig. 2c). Visualizing the area after a zoom out at 300pN forces, more than half of the formed pores presented lateral displacements >1nm with respect their previous positions (Fig. 2c). All the lysenin that remained in prepore state stay in the same position. We concluded that the insertion of lysenin is related to the liberation of certain membrane space. The motion of the pores formed by AFM compression from their initial positions confirms the findings by pH-decrease. Biologically, this liberation of space could be the answer found by lysenin to confront the necessity of crowded environments to enhance oligomerization with the necessity of scape steric inhibitions. Remarkably, all new pores appear to form clusters with no pore surrounded just by prepores.
If prepore-to-pore transition is blocked in crowded environments and pore diameter is smaller than prepore one, we hypothesize that prepore-to-pore transition should liberate certain space that could lead to unblock other insertions. To test for the effect of liberation of space after prepore insertion, we induced prepore-to-pore transitions by a sudden decrease of pH from 7.5 to 5.2. We used a lysenin assembly incubated at 5.6µM monomer concentration and filmed the process of prepore insertions in the assembly (Fig. 3a), selecting five representative HS-AFM frames. For each of the five selected frames we performed cross-correlation searches to identify the position of each oligomer. We calculated the first neighbours by Delaunay triangulation. Finally, each oligomer was assigned a prepore or pore status based on the measured height. We used the list of first neighbours to identify isolated pores and lines of three consecutive pores. We named isolated pores those pores with all neighbours in the prepore state (Fig. 3, dashed circles). To find the lines of pores, we started with two neighbour pores and localized adjacent pores within the small arc form by connecting the centres of the oligomers (Fig. 3, grey ellipsoids). To cope with the complexity of crowding, we moved beyond the simple analysis of experimental images; previous analysis only based on experimental data could not detect spatially chained unblocking of the prepore-to-pore transitions 17. We used computer simulations to generate bias-free prepore/pore neighbour statistics. Based on the list of positions of the five experimental frames we simulated prepore/pore distributions; equal probability of being a pore for all the oligomers was imposed (see five simulated frames as example in Fig. 3b). The percentage of pores was the same as in the experimental distributions. Next, we counted the number of isolated pores and of lines of three pores (Fig. 3b). We plotted the ratio of isolated pores (simulated/experimental) as a function of the percentage of pores, where we consistently observed fewer isolated pores on the experimental data than in the random simulations (Fig. 3c). The comparison between the experimental and simulated data show that the number of isolated pores is lower in experiments than in the simulations. Moreover, we find that it is more probable to find new pores in the vicinity of pores in the experimental data than in the simulations based on the number of isolated pores, which is higher for the simulations starting with the same number in mean. We concluded that the liberation of space after the prepore insertion is enough to unblock the prepore-to-pore transition from its steric inhibition. Experimental and simulated three pore lines ratio shows, at low pore percentages, more lines in the experimental lysenin assembly than in the simulations (Fig. 3c). At higher percentages we find no difference. New form pores suffer a membrane-mediated attraction toward other pores and liberate space behind, increasing insertion probability in the direction opposite to attraction; the repetition of this chain insertion mechanism create the lines of pores. The appearance of lines and not concentric circles is the result of the small diameter difference between pore and prepore, 5.5 nm2, which does not allow the insertion of all prepore neighbours at the same time.
From our study we can conclude that the steric inhibition of the prepore-to-pore transition emerge from the overall enlarge of the prepore, probably cause by the negative residues in the cytoplasmatic domain, during the invagination that results in the pore structure18. The actuals prepore and prepore-to-pore transition models18, should be refine in order to accommodate our observations. Furthermore, the electrophysiology9 and vesicle permeabilization experiments 20 can be now understood as a manifestation of crowding steric inhibition of pore formation. Nevertheless this work is the first direct proof of prepore enlargement structure during prepore-to-pore transition. Moreover, the prevalent appearance of pores in the borders of Lo raft-like phases show by Yilmaz et al.15, can now be explain in terms of the reduction of lateral pressure on SM-rich borders which facilitates prepore insertions. Since the pores can travel in the membrane without restrictions, as shown by Yilmaz et al.15. We hypothesize that the role of the steric blockage ensures that pores are delocalized from oligomerization environment in which just prepores stay, avoiding therefore plasma membrane repair mechanisms 28 29.
Our findings highlight the importance of crowding in membranes, especially during infectious processes, and the power combining different force microscopy techniques with dedicated randomization analytical approaches to assess relevant role of crowding. We envision the steric inhibition as a new antibiotic strategy30, using PFTs binding domains ‘crowders’ that would block prepore-to-pore transition, the activity of the toxin and in last term infection.