CryoEM structures of membrane pore and prepore complex reveal cytolytic mechanism of Pneumolysin

Katharina van Pee; Alexander Neuhaus; Edoardo D'Imprima; Deryck J Mills; Werner Kühlbrandt; Özkan Yildiz

doi:10.7554/eLife.23644

. 2017 Mar 21;6:e23644. doi: 10.7554/eLife.23644

CryoEM structures of membrane pore and prepore complex reveal cytolytic mechanism of Pneumolysin

Katharina van Pee ¹, Alexander Neuhaus ¹, Edoardo D'Imprima ¹, Deryck J Mills ¹, Werner Kühlbrandt ^1,^*, Özkan Yildiz ^1,^*

Editor: Sjors HW Scheres²

PMCID: PMC5437283 PMID: 28323617

Abstract

Many pathogenic bacteria produce pore-forming toxins to attack and kill human cells. We have determined the 4.5 Å structure of the ~2.2 MDa pore complex of pneumolysin, the main virulence factor of Streptococcus pneumoniae, by cryoEM. The pneumolysin pore is a 400 Å ring of 42 membrane-inserted monomers. Domain 3 of the soluble toxin refolds into two ~85 Å β-hairpins that traverse the lipid bilayer and assemble into a 168-strand β-barrel. The pore complex is stabilized by salt bridges between β-hairpins of adjacent subunits and an internal α-barrel. The apolar outer barrel surface with large sidechains is immersed in the lipid bilayer, while the inner barrel surface is highly charged. Comparison of the cryoEM pore complex to the prepore structure obtained by electron cryo-tomography and the x-ray structure of the soluble form reveals the detailed mechanisms by which the toxin monomers insert into the lipid bilayer to perforate the target membrane.

DOI: http://dx.doi.org/10.7554/eLife.23644.001

Research Organism: Other

Introduction

Gram-positive bacteria, including the common wound-infecting Staphylococcus aureus, the virulent food-borne pathogen Listeria monocytogenes, and Streptococcus pneumoniae that causes pneumonia, utilize cholesterol-dependent cytolysins (CDCs) to attack and kill mammalian and human cells (Gilbert, 2005; Tweten, 2005). The bacteria produce and release CDCs as water-soluble monomers that attach to cholesterol-containing cell membranes, where they assemble into large, 200–500 Å cytolytic pores (Olofsson et al., 1993). X-ray structures of pneumolysin (PLY) from S. pneumoniae (Lawrence et al., 2015; Marshall et al., 2015; van Pee et al., 2016) show that the soluble toxin monomers are roughly rod-shaped and consist of four domains (D1-D4). Sequence comparison suggests that all CDCs have the same domain structure and therefore insert into target membranes in essentially the same way (Tweten et al., 2001). Although numerous biochemical studies have addressed the arrangement of monomers in the CDC prepore or pore complex (Hotze and Tweten, 2012; Tilley et al., 2005) and how they might penetrate the lipid bilayer (Shatursky et al., 1999), the detailed structure of the pore complex, and hence the exact mechanism of membrane insertion is unknown. The size heterogeneity of CDC prepores and pores has so far precluded structure determination at high resolution. We determined the structure of the ring-shaped ~2.2 MDa PLY pore complex by cryoEM. Together with our 2.4 Å x-ray structure of soluble PLY (van Pee et al., 2016) and a map of the prepore complex obtained by electron cryo-tomography, we can now describe the entire process of membrane attachment, prepore and pore formation in near-atomic detail. Given the high degree of sequence conservation amongst CDCs, it is likely that the same mechanism of membrane insertion holds for related bacterial toxins.

Results

CryoEM structure of the PLY pore complex

Ring-shaped pore complexes of wildtype PLY (PLY_WT) forming on unilamellar, cholesterol-containing liposomes were solubilized with detergent. The size and homogeneity of solubilized PLY pores is detergent-dependent, as indicated by negative-stain and cryoEM (Figure 1—figure supplement 1). Pore diameters varied from 310 to 500 Å in DDM or 350 to 400 Å in Cymal-6. Replacing the detergent Cymal-6 by amphipol A8-35 (Althoff et al., 2011; Lu et al., 2014) resulted in a stable population that was sufficiently homogenous for single-particle cryoEM. 2D class averages indicated a roughly even distribution of views (Figure 1—figure supplement 2A). Pore complexes were rings of 42 subunits with an aggregate molecular mass of 2.2 MDa. In total, 6461 ring images were combined to generate a 3D map with 42-fold symmetry at 4.5 Å resolution (Figure 1; Figure 1—figure supplements 2B and 3,). The final map of the pore complex has an outer diameter of 400 Å and a total height of 110 Å, of which ~80 Å protrudes from the membrane surface (Figure 1).

Figure 1. — (A) Single-particle cryoEM map of PLY at 4.5 Å resolution with fitted model. The four PLY domains (D1–D4) are red (D1), yellow (D2), green/cyan (D3) and blue (D4). Inset: refolded β-hairpins (HP1 and HP2) fitted to the map. Cyan β-strands have refolded from helix bundles in D3 of the soluble form. (B) Cross-section with overall dimensions of the pore complex. The grey bar indicates the position of the lipid bilayer. Inset: side view of membrane-inserted monomer with toroid density of disordered amphipol (broken line).

**DOI:** http://dx.doi.org/10.7554/eLife.23644.002

Figure 1—figure supplement 1. — (A) Single-particle cryoEM map of PLY at 4.5 Å resolution with fitted model. The four PLY domains (D1–D4) are red (D1), yellow (D2), green/cyan (D3) and blue (D4). Inset: refolded β-hairpins (HP1 and HP2) fitted to the map. Cyan β-strands have refolded from helix bundles in D3 of the soluble form. (B) Cross-section with overall dimensions of the pore complex. The grey bar indicates the position of the lipid bilayer. Inset: side view of membrane-inserted monomer with toroid density of disordered amphipol (broken line).

**DOI:** http://dx.doi.org/10.7554/eLife.23644.002

Where possible, individual domains of the PLY x-ray structure (van Pee et al., 2016) were moved manually into the 4.5 Å cryoEM map as rigid bodies. Rebuilding of domains that refolded upon membrane insertion and readjustment of secondary structure elements and sidechains within domains yielded an atomic model of the pore complex (Figure 1). Comparison to the soluble form (Figure 2A,B) revealed a complete reorganization of the toxin upon membrane insertion. Of the four protein domains D1-D4 in the x-ray structure, D1 and D4 fitted the map of the pore complex with minimal modifications (Figure 2—figure supplements 1–3, Video 1). In the membrane-inserted form, the loop linking D1 to D2 refolds into a helix (α3a) at the interface between the rearranged domains D1, D2 and D3 (Figures 2A,E, 3 and 4; Figure 2—figure supplement 4). In D4, the highly conserved undecapeptide loop (Figure 3) that renders PLY cholesterol-specific (Soltani et al., 2007b) was shifted by up to ~9 Å into the cryoEM map (Figure 2—figure supplements 2 and 3, Video 1). The undecapeptide loops of adjacent protomers in the pore complex are located in one plane on the outer membrane surface of the target cell, where they interact closely with the lipid head groups (Figures 1B and 2A,C). In the linear rows of soluble PLY monomers that are found in the PLY crystal structures (Lawrence et al., 2015; Marshall et al., 2015; van Pee et al., 2016), the distance between the loops of neighbouring monomers is ~14 Å. In the pore complex this distance decreases to 4–5 Å, which enables an interaction of the loop that connects β-strands 18 and 19 (β18/19) in D4 with the β22/23 loop in D4 of the next monomer along the ring (Figure 2—figure supplements 2 and 3, Video 1). The close proximity of loop β18/19 that contains Asp403, Thr405, and His407 of one monomer to loop β18/19 and the uncedapeptide containing Trp433 of the adjacent monomer suggests a critical role of these D4 loops not only in receptor recognition, but also in oligomer formation.

Video 1. Superposition of domain 4 of the PLY monomer from the crystal structure (blue) on domain 4 of the PLY monomer in the pore complex (green) rotated around an axis perpendicular to the membrane.

Download video file^{(41.2MB, mp4)}

DOI: 10.7554/eLife.23644.014

Open in a new tab

Residues of the undecapeptide and selected residues in other loops are shown as sticks.

DOI: http://dx.doi.org/10.7554/eLife.23644.014

Figure 2. — One subunit of membrane-inserted form of PLY in the cryoEM structure (A) and x-ray structure of soluble PLY (van Pee et al., 2016) (B) seen from the side (left) and from the pore center (right). Both helix bundles (HB1 and HB2, cyan) in the PLY monomer refold to form two long, membrane-spanning β-hairpins (HP1 and HP2). The upper end of the β-hairpins is sandwiched between a helix-turn-helix motif (HTH, green) on the inside and helix α3a (red) of the next monomer on the outside of the pore. One of the helices is a refolded β-strand of D3 (green). Helix α3a is the refolded linker that connects D2 to D1 in the soluble monomer. (C) D4 contains the conserved undecapetide (green sticks) that confers cholesterol specificity to PLY. The peptide includes three resolved Trp sidechains, one of which interacts with Thr405 in the adjacent monomer. (D) Cross section through a segment of the 168-strand β−barrel with resolved bulky sidechains. The two β-hairpins of one monomer are labelled. (E) Helix α3a interacts with HP1 and D1 of the neighbouring monomer, stabilizing the pore complex.

**DOI:** http://dx.doi.org/10.7554/eLife.23644.006

Figure 2—figure supplement 1. — One subunit of membrane-inserted form of PLY in the cryoEM structure (A) and x-ray structure of soluble PLY (van Pee et al., 2016) (B) seen from the side (left) and from the pore center (right). Both helix bundles (HB1 and HB2, cyan) in the PLY monomer refold to form two long, membrane-spanning β-hairpins (HP1 and HP2). The upper end of the β-hairpins is sandwiched between a helix-turn-helix motif (HTH, green) on the inside and helix α3a (red) of the next monomer on the outside of the pore. One of the helices is a refolded β-strand of D3 (green). Helix α3a is the refolded linker that connects D2 to D1 in the soluble monomer. (C) D4 contains the conserved undecapetide (green sticks) that confers cholesterol specificity to PLY. The peptide includes three resolved Trp sidechains, one of which interacts with Thr405 in the adjacent monomer. (D) Cross section through a segment of the 168-strand β−barrel with resolved bulky sidechains. The two β-hairpins of one monomer are labelled. (E) Helix α3a interacts with HP1 and D1 of the neighbouring monomer, stabilizing the pore complex.

**DOI:** http://dx.doi.org/10.7554/eLife.23644.006

Figure 3. — Elements of secondary structure are shown for the crystal structure of soluble PLY (above) and for the cyroEM structure of the pore complex (below). Colors of the secondary structure elements and residue numbers correspond to the PLY crystal structure (Figure 2B; pdb code 5aod). Asterisks indicate conserved residues.

**DOI:** http://dx.doi.org/10.7554/eLife.23644.012

Figure 4. — Slices parallel (A–C) and perpendicular (D–F) to the membrane reveal interactions of secondary structure elements between adjacent monomers in the pore complex. Section planes are indicated on top. Note that the scale of sections D to F is 50% larger to show molecular detail more clearly. The β5-α4-β6 region in domain D1 and the long membrane-parallel helix α5 (red box in A and D) alternate around the top of the pore complex. On the outside, the 168-strand β-barrel is flanked by helix α3a (orange box in B and D) and by the α-barrel of the helix-turn-helix motifs (green box in B and F) on the inside. The 4-strand β-sheets of domain D4 (blue box in C and E) are offset against D4 of the next monomer, forming a ring of 8-strand β-sheets. The 85 Å-long β-strands forming the 168-strand β-barrel are clearly resolved (cyan box in C and E).

**DOI:** http://dx.doi.org/10.7554/eLife.23644.013

Domains 2 and 3 undergo massive rearrangement and refolding (Figure 2A,B) in the membrane-inserted form. The elongated two-stranded β-sheet of D2 rotates from its vertical position in the x-ray structure by 90° around a short glycine linker to an orientation parallel to the membrane plane in the pore complex. D2 connects D1 to D4, and its rotation results in a roughly linear translation of D1 by 35 Å towards the membrane surface and by 30 Å towards the pore centre. Otherwise, D2 required only minor adjustments of secondary structure for an optimal fit to the cryoEM map (Figures 1B and 2A, Figure 2—figure supplement 5).

By contrast, the structure of D3 changes entirely upon membrane insertion. In the soluble toxin, the central, five-stranded β-sheet in D3 is flanked by two bundles of short α-helices (Figure 2B). In the pore complex, both bundles refold into four 85 Å β-strands, which form two parallel β-hairpins that insert into and traverse the lipid bilayer (Figure 2A). The β-hairpins of neighbouring subunits coalesce into one extensive, 168-strand β-barrel with an inner diameter of 260 Å (Figure 1B). The new β-strands can be traced unambiguously, because they are straight, apparently rigid and continuous with four β-strands in the D3 x-ray structure that are preserved in the membrane form. The two new, long β-hairpins are inclined by 20° relative to the membrane normal, in good agreement with predictions for perfringolysin O on the basis of cysteine crosslinking experiments (Sato et al., 2013). The chain trace in the refolded domain is confirmed by the positions of bulky densities for large sidechains in the β-barrel (Figure 2D; Figure 2—figure supplement 4). While five of the six helices in bundles HB1 and HB2 refold into β-strands in the pore complex, one remains intact. Conversely, the loop and fifth β-strand in the central β-sheet of D3 refold into a helix (Figure 2A,B). Together, the two α-helices form a helix-turn-helix (HTH) motif (Figure 2—figure supplement 4). In the ring, the HTH motifs of the 42 subunits line up in one plane on the exoplasmic side of the molecule and form an α-barrel inside the β-barrel (Figures 1B and 4), restricting the pore diameter locally to 250 Å (Figure 1B).

The outside of the membrane-inserted part of the β-barrel is almost entirely hydrophobic (Figure 5A,E) and covered by the toroid density of disordered amphipol that replaces the membrane lipid (Figure 1B), as in the cryoEM structures of amphipol-solubilized membrane proteins (Althoff et al., 2011; Liao et al., 2013). The inside pore surface is highly polar, with three aspartates and two glutamates forming a 15 by 9 Å patch of negative charge, flanked above and below by positive charges (Figure 5B,F,G). Apart from the hydrogen bonds between the 168 β-strands in the barrel and the interactions of helix α3a with D1 and β-hairpin one (Figure 2E, Figure 2—figure supplement 4), three other factors contribute to pore stability: (1) the surface and charge complementarity of the membrane-inserted monomers (Figure 5C); (2) the alternating positive and negative charges of helices α13a and α13b in the α-barrel (Figure 5D); and (3) ionic interactions between charged sidechains in adjacent β-strands of the pore barrel (Figure 5G). In particular, Asp168 and Glu170 in β-strand β7 are in a good position for forming a salt bridge with Lys271 in β-strand β10 of the next-door monomer (Figure 5G).

Three stages of PLY pore formation

To investigate the structures of the PLY prepore and pore complex in native membranes prior to detergent solubilisation, we incubated cholesterol-containing liposomes with purified PLY and generated 3D volumes by electron cryo-tomography (cryoET). The liposomes were studded with numerous ring-shaped complexes of 290 to 360 Å outer diameter (Figure 6A). Assemblies were identified as PLY pores or prepores by the absence or presence of a lipid bilayer within the ring (Figure 7A). Pore and prepore complexes were classified independently and processed by subtomogram averaging (Figure 6B). The subtomogram average map of the prepore complex at a resolution of 22 Å (Figure 6C) had an outer diameter of 320 Å and accommodated 34 subunits (Figure 6D). Of the available PLY crystal structures (Lawrence et al., 2015; Marshall et al., 2015; van Pee et al., 2016), the model 5cr6 (Marshall et al., 2015) fitted the map best (Figure 6—figure supplement 1), indicating that in this structure the soluble toxin is in the pre-pore state. The subtomogram average map of the pore complex at 27 Å resolution (Figure 6C) indicated a slightly larger diameter of around 350 Å, but likewise consisted of 34 subunits (Figure 6E). The cross-section profile resembled that of the high-resolution cryoEM map of the pore complex closely (Figure 1), indicating that solubilisation with detergent or amphipol does not change the structure of the membrane form significantly. The model of membrane-inserted PLY fitted the subtomogram average of the pore complex without significant adjustment, including the 20° inclination of the long β-hairpins relative to the membrane normal.

Figure 6. — (A) PLY assembles into prepores and pores upon incubation with cholesterol-containing liposomes. Scale bar, 50 nm. (B) Sections through subtomogram average volumes of the prepore (top) and pore complex (below). Left panel: sections perpendicular to the membrane; right panel: sections parallel to the membrane. Colors indicate section planes. (C) Fourier shell correlation for subtomogram averages indicate 22 Å resolution for the prepore and 27 Å for the pore at FSC_0.5 or 20 Å and 21 Å resolution at at FSC_0.3. Oblique view (D) and cross section (E) of PLY prepore (left) and pore (right). Both maps accommodate 34 PLY monomers. The prepore map was fitted with the crystal structure of the soluble PLY monomer (Marshall et al., 2015), and the pore map with the cryoEM structure of the pore monomer (Figure 2A). PLY domains are red (D1), yellow (D2), green/cyan (D3) and blue (D4). The lipid bilayer is continuous in the prepore complex, but absent in the pore complex.

**DOI:** http://dx.doi.org/10.7554/eLife.23644.016

Figure 6—figure supplement 1. — (A) PLY assembles into prepores and pores upon incubation with cholesterol-containing liposomes. Scale bar, 50 nm. (B) Sections through subtomogram average volumes of the prepore (top) and pore complex (below). Left panel: sections perpendicular to the membrane; right panel: sections parallel to the membrane. Colors indicate section planes. (C) Fourier shell correlation for subtomogram averages indicate 22 Å resolution for the prepore and 27 Å for the pore at FSC_0.5 or 20 Å and 21 Å resolution at at FSC_0.3. Oblique view (D) and cross section (E) of PLY prepore (left) and pore (right). Both maps accommodate 34 PLY monomers. The prepore map was fitted with the crystal structure of the soluble PLY monomer (Marshall et al., 2015), and the pore map with the cryoEM structure of the pore monomer (Figure 2A). PLY domains are red (D1), yellow (D2), green/cyan (D3) and blue (D4). The lipid bilayer is continuous in the prepore complex, but absent in the pore complex.

**DOI:** http://dx.doi.org/10.7554/eLife.23644.016

Figure 7. — (A) Wildtype PLY (PLY_WT) and PLY_D168A form rings on cholesterol-containing liposomes. PLY_WT lyses the majority of liposomes, while PLY_D168Aleaves them mostly intact. Lipid-filled rings with a narrow rim (blue arrows) are prepores, while rings with a wider rim that do not contain lipid (red arrows) are pores. Slits (green arrows) and arcs (yellow arrows) are observed occasionally, but mostly PLY forms complete rings. Mutant PLY_D168A prepores detach easily from the liposomes due to reduced binding affinity, and then break into fragments. Curves indicate the hemolytic activity of PLY_WT, PLY_D168A, and PLY_Δ146/147. PLY_Δ146/147 is inactive, in line with the inability of this mutant to form oligomers on cholesterol-containing liposomes. Scale bar: 50 nm.

**DOI:** http://dx.doi.org/10.7554/eLife.23644.018

Atomic force microscopy (AFM) had revealed two different forms of PLY prepores in planar bilayers, rising 110 or 80 Å above the membrane surface (van Pee et al., 2016). The low form has the same height as the pore complex, but the lipid bilayer is still intact. We propose that in this lower prepore, D1, D2 and D4 have moved to their positions in the pore assembly, while D3 has not yet refolded and the long, membrane-spanning β-hairpins have not yet formed. We assign the tall form to an early prepore state, which then rearranges into the lower, late prepore state. The cross-section profile (Figure 6B,D) shows that the early prepore is essentially a ring-shaped side-by-side arrangement of soluble PLY monomers (Figure 6—figure supplement 1), similar to that in the crystal lattice (Lawrence et al., 2015; Marshall et al., 2015; van Pee et al., 2016). While the late prepore was observed for 13% of the rings on planar bilayers examined by AFM, it was not observed in tomographic volumes of PLY on liposomes, implying that it is an intermediate, transient state which inserts more readily into curved liposomes than into planar bilayers. Cross-sections of pore complexes in liposomes (Figure 6B,E) show that the lipid bilayer around the prepores and pores is curved, whereas in the AFM images it was flat (van Pee et al., 2016). Prepore stability and membrane insertion thus seem to be related to membrane curvature, such that pores form preferentially in lipid bilayers with a convex curvature.

Determinants of lytic activity

To understand the functional role of individual sidechains in oligomerization, prepore and pore formation, we mutated Asp168 that, as the structure suggests, might be involved in forming a salt bridge between adjacent subunits to alanine (PLY_D168A). We also deleted Ala146 and Arg147 (PLY_Δ146/147) in the loop that would clash with the last β-strand of the central β-sheet. The resulting mutants were characterized by negative-stain EM (Figure 7A), hemolytic activity assays (Figure 7B), and x-ray crystallography (Figure 8). X-ray data were collected to 2.45 and 2.5 Å resolution from crystals of PLY_D168A and PLY_Δ146/147 (Table 1). The structures were solved by molecular replacement with the x-ray structure of PLY_WT (van Pee et al., 2016). As in PLY_WT, the mutant monomers crystallized in rows. Superposition of both structures on PLY_WT showed no significant overall differences, although a detailed comparison indicated small changes in domain D2, helix bundle HB2 and in two of the D4 loops (Figure 8). Around the mutated residue in PLY_D168A differences were restricted to sidechain orientations, while the deletion of two residues in PLY_Δ146/147 caused significant conformational changes in the loop connecting helix α5 to strand β7 of the central β-sheet. This deletion also caused a slight shift of helix α5 towards D3 (Figure 8).

Figure 8. — (A) Ala146 and Arg147 in the loop that induces refolding of the last β-strand in the central D3 β-sheet into helix α13a in the late prepore and pore complex. Deletion of both residues renders the toxin inactive. (B) In the pore complex, Asp168 near the end of one long trans-membrane β-hairpin (HP1) forms a salt bridge with Lys271 in the other trans-membrane β-hairpin (HP2) of the adjacent monomer. Replacing Asp168 by alanine inhibits membrane insertion. (C) α-carbon traces in the x-ray structures of PLY_WT (pdb 5aod), PLY_Δ146/147 (pdb 5aof), and PLY_D168A (pdb 5aoe). Minor differences between wildtype and mutant structures are visible in the loop regions of D4 (blue arrows); D2 (yellow arrows) and HB2 (green arrows). In PLY_Δ146/147, one loop connecting D1 to D3 is also slightly different (red arrows).

**DOI:** http://dx.doi.org/10.7554/eLife.23644.019

Table 1.

Data collection and refinement statistics.

DOI: http://dx.doi.org/10.7554/eLife.23644.020

	PLY_D168A (pdb-id 5aoe)	PLY_Δ146/147 (pdb-id 5aof)
Data collection
Beamline	PXII @ Swiss Light Source x10sa
Resolution (Å)	40–2.5 (2.6–2.5)	40–2.45 (2.55–2.45)
Wavelength (Å)	0.9786	0.978
Space group	P2₁	P 2₁ 2₁ 2₁
Cell dimensions
a, b, c (Å)	160.86 24.66 208.35	24.73 163.5 207.8
α, β, γ (°)	90, 90.26, 90	90, 90, 90
Total reflections	316531 (30235)	137706 (13089)
Unique reflections	59222 (5685)	29722 (2937)
Multiplicity	5.3 (5.3)	4.6 (4.5)
Completeness (%)	99 (100)	91 (94)
Mean I / σI	8.1 (1.4)	9.1 (1.2)
Wilson B-factor	46.75	45.05
R_pim	0.08 (0.593)	0.07 (0.607)
CC*	0.998 (0.887)	0.999 (0.792)
Refinement
Reflections used in refinement	59207 (5683)	29718 (2936)
Reflections in test set	2962 (285)	1486 (147)
R_work/R_free (%)	22.25/24.98 (32.9/33.7)	20.87/23.44 (33.23/34.37)
CC(work)/CC(free)	0.946/0.934 (0.769/0.737)	0.955/0.930 (0.626/0.567)
Average B-Factor (Å²)	72.5	64.3
No. atoms in AU	7878	4008
Protein	7724	3856
Water	154	152
r.m.s. deviations:
Bond lengths (Å)	0.003	0.03
Bond angles (°)	0.72	0.69
Ramachandran favored (%)	96	96
Ramachandran allowed (%)	3.6	3.1
Ramachandran outliers (%)	0.5	0.8

Open in a new tab

Note: Values for the highest resolution shell are shown in parentheses.

Even though the structures of PLY_D168A and PLY_Δ146/147 were very similar to PLY_WT, they displayed major differences in their membrane-binding and oligomerization behaviour. Like PLY_WT, PLY_D168A bound to cholesterol-containing liposomes and oligomerized into rings (Figure 7A). However, unlike PLY_WT that lysed the liposomes within a short time, liposomes incubated with PLY_D168A remained intact for hours. Frequently, the rings detached from the liposomes, indicating they were prepores that had not yet inserted into the membrane. This is also reflected by the hemolytic activity, which was reduced by 80% compared to PLY_WT, highlighting the important role of Asp168 in ionic interactions between β-strands that stabilize the pore complex.

PLY_Δ146/147 was not able to bind to the membranes at all and had no detectable hemolytic activity (Figure 7B). Ionic interactions between the β-strands thus contribute significantly to pore stability, and any disruption of these interactions compromises pore formation. Both mutations pinpoint protein regions that are promising targets for drug development. Drugs that interfere with these interactions would render the toxin ineffective, and the bacteria that produce it non-pathogenic.

Discussion

With the ongoing resolution revolution in cryoEM (Kühlbrandt, 2014), structures of macromolecular assemblies can now be determined at high resolution without crystals. This is proving especially useful for membrane proteins and membrane protein complexes (Allegretti et al., 2015; Gu et al., 2016; Hahn et al., 2016; Letts et al., 2016; Liao et al., 2013), which tend to be unstable and flexible, and often do not crystallize. In the case of CDC pore complexes, structure determination represents a particular challenge, because they consist of variable numbers of 30 to 50 monomers (Dang et al., 2005; Shatursky et al., 1999; Shepard et al., 1998), and hence are intrinsically inhomogeneous. This fact has so far precluded structure determination of the membrane-attached or membrane-inserted form at high resolution. By a rigorous screen of the detergents used for solubilisation and purification of the pore complex from cholesterol-containing liposomes we were able to isolate a suitably homogenous population. Exchanging the detergent against amphipols appeared to stabilize rings of uniform size, as an important prerequisite for high-resolution single-particle cryoEM. The number of monomers in the amphipol-stabilized complexes was higher than in the prepores or the pores imaged by cryoET, even though the liposomes were prepared in the same way. This means that either the number of subunits in the ring varies from one liposome preparation to another, or that the rings rearrange into a more stable form upon detergent solubilisation, which we consider more likely. Interestingly, an earlier, low-resolution cryoEM structure of PLY pores and prepores in lipid bilayers deduced that there were 31 ± 3 subunits in the prepore and 38 or 44 in the pore complex (Tilley et al., 2005), in good agreement with our findings. Since the PLY pores in our subtomogram averages were smaller, consisting of 34 rather than 42 subunits, it seems that the rings can grow in the membrane by incorporating further subunits. Apparently, complexes of 42 monomers are more stable than both larger and smaller rings. The inherent variability in ring size restricts the number of particles that can be averaged in any one class. Moreover, the large ring-shaped assemblies are easily distorted and rarely, if ever, perfectly circular, which limits the accuracy to which they can be aligned by image processing. Both factors constrain the attainable resolution of the cryoEM reconstruction.

Earlier cryoEM work, AFM and FRET studies have described the vertical height change of the prepore upon membrane insertion in terms of a collapse or unfolding of D2 (Czajkowsky et al., 2004; Ramachandran et al., 2005; Tilley et al., 2005). Our atomic model of the pore complex now shows that this change is due to a rigid-body rotation of D2, in which the structure of the domain remains intact. Time-lapse AFM of PLY and listeriolysin (LLO) on planar membranes indicated that the vertical height change is not directly linked to pore formation (Mulvihill et al., 2015; van Pee et al., 2016), in line with a mechanism that involves two different prepore states that we refer to as the early and late prepore. Our model of the late prepore monomer (Figure 9) is an intermediate between the x-ray structure of the soluble form and the cryoEM structure of the membrane-inserted form. The 5-stranded central β-sheet in the crystal structure of D3 fits into the cryoET map of the early prepore complex without any modification, but the late prepore model indicates a potential steric clash between the short β-strand in the central β-sheet of D3 and the adjacent monomer. The pivotal role of the loop that contains residues 145–147 connecting helix α5 of D1 to β7 of β-hairpin one is underlined by mutant PLY_Δ146/147 that was completely inactive (Kirkham et al., 2006) and did not even bind to liposomes (Figure 7A). The detachment of PLY_D168A rings indicates that the mutant does not insert into the membrane as easily as PLY_WT to form pores. We have shown by time-lapse AFM that pore formation is irreversible (van Pee et al., 2016). Therefore the rings that become detached from the membrane are prepores, not pores.

Figure 9. — Stepwise conformational changes of PLY toxin during pore formation shown for one monomer (above, side view) and for three neighbouring monomers (below, view from the pore centre). In the first step, soluble PLY monomers attach to the surface of cholesterol-containing cell membranes via the conserved D4 undecapeptide (Figure 2C) to form circular oligomers of the early prepore. A 90° rotation of D2 moves D1 and D3 towards the membrane in the late prepore. Helix bundles HB1 and HB2 are poised above the membrane surface to refold into 85 Å β-hairpins HP1 and HP2. In the final step of pore formation, both hairpins traverse the hydrophobic membrane core and assemble into a 168-strand, 260 Å β-barrel. Reorganization of the PLY monomer exposes numerous charges on the inside of the β-barrel (Figure 4) that would destabilize the lipid bilayer and repel membrane lipids, resulting in pore opening and cell lysis.

**DOI:** http://dx.doi.org/10.7554/eLife.23644.021

In earlier models of pore formation, the conserved domain D4 (Figure 3) was thought to be important for toxin targeting, receptor recognition and binding to cholesterol-containing membranes (Farrand et al., 2010; Ramachandran et al., 2002; Soltani et al., 2007a, 2007b), whereas oligomer formation was mainly attributed to intermolecular interactions via D1 and D3 (Lawrence et al., 2015; Ramachandran et al., 2002). It has however been shown that D4 of streptolysin, pyolysin and LLO can oligomerise by itself on cholesterol crystals (Harris et al., 2011; Weis and Palmer, 2001) or erythrocyte ghosts (Köster et al., 2014). Our cryoEM structure confirms that D4 can indeed play a role in oligomer formation through intra- and intermolecular interactions of its loops (Figure 2—figure supplements 2 and 3). Loop β22/23 and the conserved undecapeptide loop with Trp433 at its tip appear to interact with loop β18/19 of the neighbouring monomer, thereby stabilizing the pore complex by inter- and intramolecular interactions (Figure 2—figure supplements 2 and 3).

The surface charge distribution on the toxin monomers is critical for CDC activity. Modification of surface charges affects ring formation, as we demonstrated by mutagenesis both for LLO (Köster et al., 2014) and PLY (van Pee et al., 2016). The alternating positive and negative charges of the HTH motif (Figure 5D) and the positive charges at the tips of the β-hairpins (Figure 5E,F) demonstrate the importance of charge complementarity for PLY ring formation. A similar HTH motif was found in the 8 Å cryoEM map of poly-C9 component of the human membrane attack complex (Dudkina et al., 2016). This HTH motif was already present in the soluble monomer of the C6 component (Aleshin et al., 2012) As in PLY, the HTH in the poly-C9 pore forms an α-barrel that determines the effective pore diameter.

Together with the x-ray structure of the soluble form, our cryoEM structures of the pore and prepore explain the step-by-step process of membrane insertion and pore formation in near-atomic detail. Movies showing the transition from the soluble monomer to the pore complex via the intermediate stages of the early and late prepore illustrate the complete mechanism of pore formation (Videos 2–4). In the early prepore, PLY monomers assemble side by side into rings on the membrane surface. In the late prepore, the D3 helices are poised for re-folding immediately above membrane surface. Helix unfolding may be a stochastic yet cooperative event, such that the spontaneous transition of one monomer triggers the refolding of its neighbours in the ring, comparable to a zipper. As the 168-strand β-barrel assembles in the membrane, the charged patches on its inside surface would repel any trapped hydrophobic lipid. Most likely the lipid is pushed out of the nascent pore by successively inserted toxin monomers, rather than being ejected in the form of micelles or small vesicles, as has been proposed for suilysin (Leung et al., 2014). Upon pore formation, the membrane potential collapses, the cytoplasm leaks out and the cell dies.

Video 2. Mechanism of pneumolysin pore formation.

Download video file^{(6.9MB, mp4)}

DOI: 10.7554/eLife.23644.022

Open in a new tab

(1) Soluble PLY monomers attach to cholesterol-rich membranes by the cholesterol-binding undecapeptide (light green) of domain D4 (blue) and oligomerize into rings. For simplicity, only three ring subunits are shown. (2) Domain D2 (yellow) rotates by 90°, bringing domain D3 (green) with its two helix bundles (cyan) close to the membrane surface. (3) The helix bundles insert into the membrane and unfold into two trans-membrane 85 Å β-hairpins. β-hairpins of the 42 subunits in the ring merge into one large 168-strand β-barrel, which perforates the membrane.

DOI: http://dx.doi.org/10.7554/eLife.23644.022

Video 3. Model of conformational changes in an entire ring of 42 PLY monomers from early prepore to late prepore to pore, seen from the side and from above.

Download video file^{(4.7MB, mp4)}

DOI: 10.7554/eLife.23644.023

Open in a new tab

DOI: http://dx.doi.org/10.7554/eLife.23644.023

Video 4. Oblique view of PLY ring inserting into the membrane.

Download video file^{(3.2MB, mp4)}

DOI: 10.7554/eLife.23644.024

Open in a new tab

DOI: http://dx.doi.org/10.7554/eLife.23644.024

The mechanism of membrane insertion by rearrangement of the conserved domains 1–4 is likely to be the same for all CDCs. Therefore, compounds that interfere with refolding and membrane insertion of soluble monomers would prevent infection by Streptococcus pneumonia and other CDC-producing Gram-positive bacteria that attack human cells with similar pore-forming toxins. The 4.5 Å cryoEM structure of the PLY pore complex thus paves the way towards the design of new drugs that inhibit pore formation as a promising approach towards combatting infections by dangerous and wide-spread Gram-positive pathogens.

Materials and methods

Protein expression and purification

The gene coding for N-terminal His₆-tagged PLY was inserted into the pET15b vector. E. coli BL21 (DE3) cells transformed with the expression plasmid were grown in TB medium containing 50 µg · ml⁻¹ ampicillin. Protein expression was induced with 1 mM isopropyl-β-D-1-thiogalactopyranoside upon reaching an optical density of one at 600 nm. After 4 hr at 37°C the cells were pelleted, resuspended in Buffer A (50 mM Tris pH 7.0, 150 mM NaCl, 30 mM Imidazole, 5 mM β-mercaptoethanol) and disrupted with a Microfluidizer (M-110L, Microfluidics Corp., Newton, MA). PLY was purified on a HisTrap FF column, equilibrated with Buffer A. The protein was eluted in Buffer A containing 300 mM imidazole. Protein fractions were pooled and diluted in 50 mM Tris pH 7.0, 5 mM β−mercaptoethanol to a final NaCl concentration of 50 mM. The His₆-tag was removed by overnight cleavage with thrombin at 4°C. The protein was further purified on a HiTrap Q FF ion-exchange column equilibrated in Buffer B (50 mM Tris-HCl pH 7.0, 50 mM NaCl, 5 mM β-mercaptoethanol). PLY was eluted in Buffer B containing 170 mM NaCl. Protein fractions were pooled, concentrated to 10 mg · ml⁻¹ and stored at −80°C.

Site-directed mutagenesis and hemolysis assays

Site-directed mutagenesis was performed with the QuikChange site-directed mutagenesis kit (Stratagene) according to the manufacturer’s instructions with the wildtype construct as a template. All constructs were verified by nucleotide sequencing. The hemolytic activity of PLY was determined by lysis of sheep red blood cells (SRBC) (Darji et al., 1995). Purified protein was serially diluted in hemolysis buffer (50 mM sodium phosphate pH 6.6, 150 mM NaCl, 5 mM DTT, 0.1% (v/v) BSA) in final volumes of 50 µl and incubated for 30 min at 37°C with 50 µl of a suspension of SRBC (I0⁸ ml⁻¹). Release of hemoglobin was monitored by recording the absorbance at 405 nm. The amount of toxin necessary to lyse 50% of erythrocytes was determined and expressed as percentage of the value for PLY_WT. The absorbance upon incubation with 1% Triton-X100 was used as reference value for 100% lysis of erythrocytes. Three independent measurements were performed for each PLY mutant.

Liposome preparation and amphipol stabilization

A lipid mixture containing 70 mol-% di-oleyl phosphatidyl choline (DOPC) and 30 mol-% cholesterol in chloroform was dried under a constant nitrogen stream. The lipid film was taken up in 50 mM Tris-HCl pH 7.0, 150 mM NaCl, 5 mM β-mercaptoethanol at a final concentration of 10 mg · ml⁻¹ and stirred overnight at room temperature. After three freeze-thaw cycles (liquid nitrogen, 37°C), the suspension was passed through an extruder to obtain unilamellar ~200 nm liposomes. The extruded liposomes were flash-frozen in liquid nitrogen and stored at −20°C. The liposome suspension was incubated with PLY at a final lipid-to-protein ratio of 1:2 (wt/wt) at 37°C for 30 min. For the preparation of PLY pore complexes, the proteoliposomes were solubilized at a final concentration of 0.56% Cymal-6 at room temperature overnight. Amphipol A8-35 was added in fivefold molar excess and the suspension was incubated for 30 min at room temperature. Detergent was removed by dialysis against 50 mM Tris-HCl pH 7.0, 150 mM NaCl, 5 mM β-mercaptoethanol at room temperature for 72 hr.

Specimen preparation for negative-stain EM

PLY mutant pores were compared to wildtype protein formed on planar lipid layers or cholesterol-containing liposomes prepared as above. The liposome mixture was diluted to a final concentration of 0.5 mg · ml⁻¹ and 25 µg · ml⁻¹ PLY was added. Planar lipid bilayers and PLY were incubated in Teflon well plates. 25 µg · ml⁻¹ PLY in reaction buffer (150 mM NaCl, 50 mM Tris-HCl pH 7.0, 5 mM β-mercaptoethanol) was transferred into a well and overlaid with a droplet of a 0.5 mg · ml⁻¹ DOPC/cholesterol solution (1:1 molar ratio) in chloroform. Planar lipid layers formed upon chloroform evaporation. Wells were incubated for 30 min at 37°C, after which PLY-containing membranes were transferred to EM grids and stained with 1% (wt/vol) uranyl acetate. Negatively stained specimens were examined in an FEI Tecnai Spirit electron microscope at an acceleration voltage of 120 kV. Images were recorded on a 2 K Gatan CCD camera at a magnification of 25,000x–45,000x and ∼1.0–1.5 µm underfocus.

Single-particle cryoEM and image processing

Amphipol-solubilized PLY pores were diluted to a final concentration of 1 mg · ml⁻¹. A 3 µl aliquot was applied to freshly glow-discharged R2/2 holey carbon grids (Quantifoil Micro Tools, Jena, Germany), blotted for 3 to 4 s at 10°C and 100% humidity in a Vitrobot Mark IV (FEI). Preferential orientation of toxin rings on the air-water interface was overcome by carbon backing. Dose-fractionated 6.0 s movies of 30 frames with an electron dose of 1.02 e^- · (Å² · frame)⁻¹ were recorded after coma-free alignment (Allegretti et al., 2014) with an FEI Polara electron microscope operating at 300 kV with 0.5–3.6 µm underfocus and a specimen pixel size of 1.4 Å on a K2 Summit direct electron detection camera (Gatan, Pleasanton, USA) operating in counting mode with an energy filter slit of 20 eV. Movie frames were corrected for beam-induced motion with Motioncorr (Li et al., 2013) and again with UNBLUR (Grant and Grigorieff, 2015), which also applied a dose-dependent filter. The contrast transfer function of each image was determined using CTFFIND3 (Mindell and Grigorieff, 2003). A total of 12308 particle images were hand-picked from 983 micrographs in RELION 1.3 or 1.4 (Scheres, 2012, 2015) and extracted into 360-pixel boxes. Initial 2D classification indicated that rings with 42-fold symmetry were most common, although rings with different symmetries were also present. For 3D classification in RELION, the pore structure of suilysin (EMD-2983) low-pass filtered to 50 Å was used as a reference. For the next processing step, the unsymmetrized low-resolution PLY map was used as the reference. Subsequent 3D classification yielded three classes, of which two had clear 42-fold symmetry. Rings that were distorted, damaged or of different size were sorted out at the 3D classification stage. A total of 6461 particles from 3D classes of 42-fold symmetry were auto-refined in RELION. A B-factor of −175 Å² for map sharpening was determined using the modulation transfer function of the K2 Summit detector. The resolution of the processed data after B-factor sharpening was 4.5 Å (FSC_0.143), with an estimated orientation accuracy of 0.75°. Local resolution was assessed with RELION 2.0 (Kimanius et al., 2016).

CryoET of PLY prepores and pores

Preformed liposomes were incubated with PLY (1 mg · ml⁻¹) at room temperature for 30 min to obtain prepores, or at 37°C for 30–180 min to obtain pores. For cryoET the liposomes were diluted 1:3 with buffer (150 mM NaCl, 50 mM Tris-HCl pH 7.0, 5 mM β-mercaptoethanol) and then 1:1 with 6 nm gold particles conjugated with protein A (Aurion) as fiducial markers. Samples of 3 µl were applied to glow-discharged Quantifoil EM grids (R2/2, Cu 300 mesh), excess liquid was blotted off with filter paper (Whatman #4) and samples were vitrified by plunge-freezing into liquid ethane (Dubochet et al., 1988). Single-axis tilt series were typically collected from −62° to +62° at 2° increments and 3–4 µm underfocus with a total electron dose of 60–80 e^- · Å^-2, on a Tecnai Polara electron microscope equipped with a field emisson gun operating at 300 kV (FEI, Hillsboro, OR), a post-column energy filter (GIF Quantum, Gatan) and a K2 summit direct electron detector (Gatan). Images were recorded in counting mode with a pixel size of 0.35 nm. Tilt series were CTF-corrected, binned 2 × 2 and aligned. Tomographic volumes were generated by weighted back projection in IMOD (Kremer et al., 1996).

Sub-tomogram averaging

For particle picking and initial rounds of sub-tomogram averaging (Frangakis and Hegerl, 2001), tomograms were filtered by nonlinear anisotropic diffusion to enhance contrast. Ring-shaped complexes were picked manually in 3dmod. A total of 752 prepore complexes and 2400 pore complexes were picked for averaging. Ring volumes were pre-aligned in one plane in 2 × 2 binned unfiltered tomograms. All prepores had roughly the same size and shape and were averaged in PEET (Nicastro et al., 2006). The average volume was used as an initial reference for symmetry determination by sub-tomogram averaging with IMOD (Kremer et al., 1996). Symmetries of 42-fold or lower were applied to each prepore volume on a trial-and-error basis. Volumes were aligned and averaged in an iterative process, which converged within about 15 iterations on an average volume with clear 34-fold symmetry. The best 70% of the prepore volumes contributed to the final average, in which 34 PLY monomers were resolved in the ring.

Pore complexes were separated into seven classes according to their shape and diameter by principal component analysis and clusterPCA in IMOD (Kremer et al., 1996). The best class with 844 pore complexes was used for further processing as described above. In the final pore complex average, individual subunits were clearly distinguished in one third of the ring, indicating some heterogeneity of ring size and symmetry (Figure 5B). The number of PLY monomers in the pore was determined as 34 from the angular distance between distinct subunits in the average volume and the ring diameter. As for the prepores, all averaging steps were performed in PEET (Nicastro et al., 2006). The resolution at FSC_0.5 was 27 Å for the pore and 22 Å for the prepore (Figure 6c). At FSC_0.3, the resolution of the pore and prepore complex would be 21 Å and 20 Å, respectively.

Model building and analysis

Model building was performed in COOT (Emsley and Cowtan, 2004) based on the x-ray structure of the water-soluble PLY monomer (pdb 5a0d; (van Pee et al., 2016). Initially, individual domains in the x-ray structure of soluble PLY were fitted manually into the cryoEM map as rigid bodies using COOT. Refolded or flexible protein regions were re-fitted manually, followed by geometry regularization in COOT. The complete backbone of PLY was traced in the pore complex map. Densities of bulky side chains were observed in well-ordered regions. The transmembrane β-hairpins, which form upon unfolding of HB1 and HB2 in domain 3 were readily fitted to the map density. The pore model with 42 subunits was generated in UCSF Chimera (Pettersen et al., 2004). Sub-tomogram average maps of the PLY prepore were fitted in UCSF Chimera with the x-ray structures of the PLY monomer (pdb 5cr6; [Marshall et al., 2015]). Sub-tomogram averages of the pore complex were fitted with the cryoEM structure of the membrane-inserted form. Figures were drawn with PyMol (Schrödinger, 2015). FSC curves of model versus map were calculated using the EMAN package (Tang et al., 2007).

Crystallization, data collection, structure determination, and refinement

Intitial crystallization trials were carried out with a protein concentration of 8 mg · ml⁻¹ (PLY_D168A) or 10 mg · ml⁻¹ (PLY_Δ146/147) after addition of glycerol to a final concentration of 10% in 96-well plates by vapour diffusion. 300 nl of protein solution were mixed with 300 nl of a commercially available crystallization solution (PGA Screen from Molecular Dimensions, JB Screen Classics I from Jena Bioscience, and Classics I Suite from Qiagen) in a pipetting robot (Mosquito, TTP Labtech). Hanging drops were incubated over 100 µl of reservoir solution at 18°C. Initial crystals hits were refined by varying the protein-to-reservoir ratio with a drop volume of 3 µl, incubated over 500 µl reservoirs at 18°C in 24-well plates. PLY_D168A crystals grew after one day at 18°C in a 3 µl drop of 1.5 µl protein solution (8 mg · ml^-1 in 10% glycerol) and 1.5 µl of reservoir solution (0.1 M sodium cacodylate pH 6.5, 1% PGA-LM, 0.2 M potassium bromide and 0.2 M potassium thiocyanate). PLY_Δ146/147 crystals grew in a few days at 18°C in a 3 µl drop of 2.0 µl protein solution (10 mg · ml^-1 in 10% glycerol) and 1.0 µl of reservoir solution (0.1 M imidazole pH 6.5 and 1.2 M sodium acetate). Crystals were transferred to reservoir solution as a cryo-protectant and flash-frozen in liquid nitrogen. Data were collected at beamline PXII (Paul Scherrer Institute, Villigen, Switzerland) under a constant stream of cold nitrogen gas (100 K). Data processing, integration and scaling was performed with the XDS package (Kabsch, 1993). Structures of the PLY mutants were solved by molecular replacement with PLY_WT (pdb-id 4AOD) as a search model using PHASER (McCoy, 2007) in the CCP4 software package (Collaborative Computational Project, Number 4, 1994). The initial electron density map was improved by cycles of density modification, automatic model building in RESOLVE (Terwilliger, 2004) and refinement by REFMAC (Murshudov et al., 1997). The model was subjected to iterative rounds of rebuilding into 2F_o-F_c and F_o-F_c maps using COOT (Emsley and Cowtan, 2004) and refined with the phenix.refine subroutine in the PHENIX program suite (Zwart et al., 2008). Data collection, refinement, and model statistics are summarized in Table 1. Figures were generated with PyMOL (Schrödinger, 2015).

Accession numbers

The cryoEM density map has been deposited in the Electron Microscopy Data Bank under the accession number EMD-4118. The structure coordinates have been deposited in the protein data bank under accession number 5LY6 (PLY pore complex), 5AOE (PLY_D168A) and 5AOF (PLY_Δ146/147)

Acknowledgements

We thank Janet Vonck for discussion and advice on image processing, Sabine Häder and Heidi Betz for technical assistance, Sebastian Schaupp for help with screening solubilisation conditions and negative-stain EM, and Juan Castillo for computer support. This work was funded by the Max Planck Society.

Funding Statement

The funders had no role in study design, data collection and interpretation, or the decision to submit the work for publication.

Funding Information

This paper was supported by the following grant:

Max-Planck-Gesellschaft DepartmentSB to Katharina van Pee, Alexander Neuhaus, Edoardo D'Imprima, Deryck J Mills, Werner Kühlbrandt, Özkan Yildiz.

Additional information

Competing interests

WK: Reviewing editor, eLife.

The other authors declare that no competing interests exist.

Author contributions

KvP, Data curation, Investigation, Methodology, Writing—original draft, Performed biochemical experiments, Carried out single-particle cryoEM and processing, Interpreted the data and wrote the paper.

AN, Investigation, Methodology, Performed cryo-tomography and sub-tomogram averaging and analysis of data.

ED, Investigation, Methodology, Interpreted the data and wrote the paper.

DJM, Investigation, Methodology, Devised the cryoEM data collection strategy.

WK, Resources, Formal analysis, Funding acquisition, Writing—review and editing, Interpreted the data and wrote the paper.

ÖY, Conceptualization, Data curation, Formal analysis, Supervision, Validation, Investigation, Visualization, Methodology, Writing—original draft, Project administration, Writing—review and editing, Initiated and directed the project, Interpreted the data and wrote the paper.

Additional files

Major datasets

The following datasets were generated:

van Pee K,Neuhaus A,D'Imprima E,Mills DJ,Kuehlbrandt W,Yildiz O,2016,CryoEM structure of the membrane pore complex of Pneumolysin at 4.5A,https://www.ebi.ac.uk/pdbe/entry/pdb/5ly6,Publicly available at the EMBL-EBI Protein Data Bank (accession no: 5LY6)

van Pee K,Neuhaus A,D'Imprima E,Mills DJ,Kuehlbrandt W,Yildiz O,2016,CryoEM structure of the membrane pore complex of Pneumolysin at 4.5A,http://www.ebi.ac.uk/pdbe/entry/emdb/EMD-4118,Publicly available at the EMBL-EBI Protein Data Bank (accession no: EMD-4118)

van Pee K,Yildiz O,2016,Crystal structure of pneumolysin deletion mutant Delta146_147,http://www.rcsb.org/pdb/explore/explore.do?structureId=5AOF,Publicly available at the RCSB Protein Data Bank (accession no: 5AOF)

van Pee K,Yildiz O,2016,Crystal structure of pneumolysin D168A mutant,http://www.rcsb.org/pdb/explore/explore.do?structureId=5AOE,Publicly available at the RCSB Protein Data Bank (accession no: 5AOE)

The following previously published dataset was used:

van Pee K,Yildiz O,2016,Crystal structure of wild type pneumolysin,http://www.rcsb.org/pdb/explore/explore.do?structureId=5aod,Publicly available at the RCSB Protein Data Bank (accession no: 5AOD)

References

Aleshin AE, Schraufstatter IU, Stec B, Bankston LA, Liddington RC, DiScipio RG. Structure of complement C6 suggests a mechanism for initiation and unidirectional, sequential assembly of membrane attack complex (MAC) Journal of Biological Chemistry. 2012;287:10210–10222. doi: 10.1074/jbc.M111.327809. [DOI] [PMC free article] [PubMed] [Google Scholar]
Allegretti M, Klusch N, Mills DJ, Vonck J, Kühlbrandt W, Davies KM. Horizontal membrane-intrinsic α-helices in the stator a-subunit of an F-type ATP synthase. Nature. 2015;521:237–240. doi: 10.1038/nature14185. [DOI] [PubMed] [Google Scholar]
Allegretti M, Mills DJ, McMullan G, Kühlbrandt W, Vonck J. Atomic model of the F420-reducing [NiFe] hydrogenase by electron cryo-microscopy using a direct electron detector. eLife. 2014;3:e01963. doi: 10.7554/eLife.01963. [DOI] [PMC free article] [PubMed] [Google Scholar]
Althoff T, Mills DJ, Popot JL, Kühlbrandt W. Arrangement of electron transport chain components in bovine mitochondrial supercomplex I1III2IV1. The EMBO Journal. 2011;30:4652–4664. doi: 10.1038/emboj.2011.324. [DOI] [PMC free article] [PubMed] [Google Scholar]
Collaborative Computational Project, Number 4 The CCP4 suite: programs for protein crystallography. Acta Crystallographica. Section D, Biological Crystallography. 1994;50:760–763. doi: 10.1107/S0907444994003112. [DOI] [PubMed] [Google Scholar]
Czajkowsky DM, Hotze EM, Shao Z, Tweten RK. Vertical collapse of a cytolysin prepore moves its transmembrane beta-hairpins to the membrane. The EMBO Journal. 2004;23:3206–3215. doi: 10.1038/sj.emboj.7600350. [DOI] [PMC free article] [PubMed] [Google Scholar]
Dang TX, Hotze EM, Rouiller I, Tweten RK, Wilson-Kubalek EM. Prepore to pore transition of a cholesterol-dependent cytolysin visualized by electron microscopy. Journal of Structural Biology. 2005;150:100–108. doi: 10.1016/j.jsb.2005.02.003. [DOI] [PubMed] [Google Scholar]
Darji A, Chakraborty T, Niebuhr K, Tsonis N, Wehland J, Weiss S. Hyperexpression of listeriolysin in the nonpathogenic species Listeria innocua and high yield purification. Journal of Biotechnology. 1995;43:205–212. doi: 10.1016/0168-1656(95)00138-7. [DOI] [PubMed] [Google Scholar]
Dubochet J, Adrian M, Chang JJ, Homo JC, Lepault J, McDowall AW, Schultz P. Cryo-electron microscopy of vitrified specimens. Quarterly Reviews of Biophysics. 1988;21:129–228. doi: 10.1017/S0033583500004297. [DOI] [PubMed] [Google Scholar]
Dudkina NV, Spicer BA, Reboul CF, Conroy PJ, Lukoyanova N, Elmlund H, Law RH, Ekkel SM, Kondos SC, Goode RJ, Ramm G, Whisstock JC, Saibil HR, Dunstone MA. Structure of the poly-C9 component of the complement membrane attack complex. Nature Communications. 2016;7:10588. doi: 10.1038/ncomms10588. [DOI] [PMC free article] [PubMed] [Google Scholar]
Emsley P, Cowtan K. Coot: model-building tools for molecular graphics. Acta Crystallographica Section D Biological Crystallography. 2004;60:2126–2132. doi: 10.1107/S0907444904019158. [DOI] [PubMed] [Google Scholar]
Farrand AJ, LaChapelle S, Hotze EM, Johnson AE, Tweten RK. Only two amino acids are essential for cytolytic toxin recognition of cholesterol at the membrane surface. PNAS. 2010;107:4341–4346. doi: 10.1073/pnas.0911581107. [DOI] [PMC free article] [PubMed] [Google Scholar]
Frangakis AS, Hegerl R. Noise reduction in electron tomographic reconstructions using nonlinear anisotropic diffusion. Journal of Structural Biology. 2001;135:239–250. doi: 10.1006/jsbi.2001.4406. [DOI] [PubMed] [Google Scholar]
Gilbert RJ. Inactivation and activity of cholesterol-dependent cytolysins: what structural studies tell us. Structure. 2005;13:1097–1106. doi: 10.1016/j.str.2005.04.019. [DOI] [PubMed] [Google Scholar]
Grant T, Grigorieff N. Measuring the optimal exposure for single particle cryo-EM using a 2.6 å reconstruction of rotavirus VP6. eLife. 2015;4:e06980. doi: 10.7554/eLife.06980. [DOI] [PMC free article] [PubMed] [Google Scholar]
Gu J, Wu M, Guo R, Yan K, Lei J, Gao N, Yang M. The architecture of the mammalian respirasome. Nature. 2016;537:639–643. doi: 10.1038/nature19359. [DOI] [PubMed] [Google Scholar]
Hahn A, Parey K, Bublitz M, Mills DJ, Zickermann V, Vonck J, Kühlbrandt W, Meier T. Structure of a complete ATP synthase dimer reveals the molecular basis of inner mitochondrial membrane morphology. Molecular Cell. 2016;63:445–456. doi: 10.1016/j.molcel.2016.05.037. [DOI] [PMC free article] [PubMed] [Google Scholar]
Harris JR, Lewis RJ, Baik C, Pokrajac L, Billington SJ, Palmer M. Cholesterol microcrystals and cochleate cylinders: attachment of pyolysin oligomers and domain 4. Journal of Structural Biology. 2011;173:38–45. doi: 10.1016/j.jsb.2010.07.010. [DOI] [PubMed] [Google Scholar]
Hotze EM, Tweten RK. Membrane assembly of the cholesterol-dependent cytolysin pore complex. Biochimica Et Biophysica Acta (BBA) - Biomembranes. 2012;1818:1028–1038. doi: 10.1016/j.bbamem.2011.07.036. [DOI] [PMC free article] [PubMed] [Google Scholar]
Kabsch W. Automatic processing of rotation diffraction data from crystals of initially unknown symmetry and cell constants. Journal of Applied Crystallography. 1993;26:795–800. doi: 10.1107/S0021889893005588. [DOI] [Google Scholar]
Kimanius D, Forsberg BO, Scheres SH, Lindahl E. Accelerated cryo-EM structure determination with parallelisation using GPUs in RELION-2. eLife. 2016;5:e18722. doi: 10.7554/eLife.18722. [DOI] [PMC free article] [PubMed] [Google Scholar]
Kirkham LA, Kerr AR, Douce GR, Paterson GK, Dilts DA, Liu DF, Mitchell TJ. Construction and immunological characterization of a novel nontoxic protective pneumolysin mutant for use in future pneumococcal vaccines. Infection and Immunity. 2006;74:586–593. doi: 10.1128/IAI.74.1.586-593.2006. [DOI] [PMC free article] [PubMed] [Google Scholar]
Kremer JR, Mastronarde DN, McIntosh JR. Computer visualization of three-dimensional image data using IMOD. Journal of Structural Biology. 1996;116:71–76. doi: 10.1006/jsbi.1996.0013. [DOI] [PubMed] [Google Scholar]
Köster S, van Pee K, Hudel M, Leustik M, Rhinow D, Kühlbrandt W, Chakraborty T, Yildiz Ö. Crystal structure of listeriolysin O reveals molecular details of oligomerization and pore formation. Nature Communications. 2014;5:3690. doi: 10.1038/ncomms4690. [DOI] [PubMed] [Google Scholar]
Kühlbrandt W. Biochemistry. The resolution revolution. Science. 2014;343:1443–1444. doi: 10.1126/science.1251652. [DOI] [PubMed] [Google Scholar]
Lawrence SL, Feil SC, Morton CJ, Farrand AJ, Mulhern TD, Gorman MA, Wade KR, Tweten RK, Parker MW. Crystal structure of Streptococcus pneumoniae pneumolysin provides key insights into early steps of pore formation. Scientific Reports. 2015;5:14352. doi: 10.1038/srep14352. [DOI] [PMC free article] [PubMed] [Google Scholar]
Letts JA, Fiedorczuk K, Sazanov LA. The architecture of respiratory supercomplexes. Nature. 2016;537:644–648. doi: 10.1038/nature19774. [DOI] [PubMed] [Google Scholar]
Leung C, Dudkina NV, Lukoyanova N, Hodel AW, Farabella I, Pandurangan AP, Jahan N, Pires Damaso M, Osmanović D, Reboul CF, Dunstone MA, Andrew PW, Lonnen R, Topf M, Saibil HR, Hoogenboom BW. Stepwise visualization of membrane pore formation by suilysin, a bacterial cholesterol-dependent cytolysin. eLife. 2014;3:e04247. doi: 10.7554/eLife.04247. [DOI] [PMC free article] [PubMed] [Google Scholar]
Li X, Mooney P, Zheng S, Booth CR, Braunfeld MB, Gubbens S, Agard DA, Cheng Y. Electron counting and beam-induced motion correction enable near-atomic-resolution single-particle cryo-EM. Nature Methods. 2013;10:584–590. doi: 10.1038/nmeth.2472. [DOI] [PMC free article] [PubMed] [Google Scholar]
Liao M, Cao E, Julius D, Cheng Y. Structure of the TRPV1 ion channel determined by electron cryo-microscopy. Nature. 2013;504:107–112. doi: 10.1038/nature12822. [DOI] [PMC free article] [PubMed] [Google Scholar]
Lu P, Bai XC, Ma D, Xie T, Yan C, Sun L, Yang G, Zhao Y, Zhou R, Scheres SH, Shi Y. Three-dimensional structure of human γ-secretase. Nature. 2014;512:166–170. doi: 10.1038/nature13567. [DOI] [PMC free article] [PubMed] [Google Scholar]
Marshall JE, Faraj BH, Gingras AR, Lonnen R, Sheikh MA, El-Mezgueldi M, Moody PC, Andrew PW, Wallis R. The crystal structure of pneumolysin at 2.0 å resolution reveals the molecular packing of the pre-pore complex. Scientific Reports. 2015;5:13293. doi: 10.1038/srep13293. [DOI] [PMC free article] [PubMed] [Google Scholar]
McCoy AJ. Solving structures of protein complexes by molecular replacement with phaser. Acta Crystallographica Section D Biological Crystallography. 2007;63:32–41. doi: 10.1107/S0907444906045975. [DOI] [PMC free article] [PubMed] [Google Scholar]
Mindell JA, Grigorieff N. Accurate determination of local defocus and specimen tilt in electron microscopy. Journal of Structural Biology. 2003;142:334–347. doi: 10.1016/S1047-8477(03)00069-8. [DOI] [PubMed] [Google Scholar]
Mulvihill E, van Pee K, Mari SA, Müller DJ, Yildiz Ö. Directly observing the lipid-dependent self-assembly and pore-forming mechanism of the cytolytic toxin listeriolysin O. Nano Letters. 2015;15:6965–6973. doi: 10.1021/acs.nanolett.5b02963. [DOI] [PubMed] [Google Scholar]
Murshudov GN, Vagin AA, Dodson EJ. Refinement of macromolecular structures by the maximum-likelihood method. Acta Crystallographica Section D Biological Crystallography. 1997;53:240–255. doi: 10.1107/S0907444996012255. [DOI] [PubMed] [Google Scholar]
Nicastro D, Schwartz C, Pierson J, Gaudette R, Porter ME, McIntosh JR. The molecular architecture of axonemes revealed by cryoelectron tomography. Science. 2006;313:944–948. doi: 10.1126/science.1128618. [DOI] [PubMed] [Google Scholar]
Olofsson A, Hebert H, Thelestam M. The projection structure of perfringolysin O (Clostridium perfringens theta-toxin) FEBS Letters. 1993;319:125–127. doi: 10.1016/0014-5793(93)80050-5. [DOI] [PubMed] [Google Scholar]
Pettersen EF, Goddard TD, Huang CC, Couch GS, Greenblatt DM, Meng EC, Ferrin TE. UCSF chimera--a visualization system for exploratory research and analysis. Journal of Computational Chemistry. 2004;25:1605–1612. doi: 10.1002/jcc.20084. [DOI] [PubMed] [Google Scholar]
Ramachandran R, Heuck AP, Tweten RK, Johnson AE. Structural insights into the membrane-anchoring mechanism of a cholesterol-dependent cytolysin. Nature Structural Biology. 2002;9:823–827. doi: 10.1038/nsb855. [DOI] [PubMed] [Google Scholar]
Ramachandran R, Tweten RK, Johnson AE. The domains of a cholesterol-dependent cytolysin undergo a major FRET-detected rearrangement during pore formation. PNAS. 2005;102:7139–7144. doi: 10.1073/pnas.0500556102. [DOI] [PMC free article] [PubMed] [Google Scholar]
Sato TK, Tweten RK, Johnson AE. Disulfide-bond scanning reveals assembly state and β-strand tilt angle of the PFO β-barrel. Nature Chemical Biology. 2013;9:383–389. doi: 10.1038/nchembio.1228. [DOI] [PMC free article] [PubMed] [Google Scholar]
Scheres SH. RELION: implementation of a bayesian approach to cryo-EM structure determination. Journal of Structural Biology. 2012;180:519–530. doi: 10.1016/j.jsb.2012.09.006. [DOI] [PMC free article] [PubMed] [Google Scholar]
Scheres SH. Semi-automated selection of cryo-EM particles in RELION-1.3. Journal of Structural Biology. 2015;189:114–122. doi: 10.1016/j.jsb.2014.11.010. [DOI] [PMC free article] [PubMed] [Google Scholar]
Schrödinger LLC. The PyMOL Molecular Graphics System. Version 1.8PyMol. 2015
Shatursky O, Heuck AP, Shepard LA, Rossjohn J, Parker MW, Johnson AE, Tweten RK. The mechanism of membrane insertion for a cholesterol-dependent cytolysin: a novel paradigm for pore-forming toxins. Cell. 1999;99:293–299. doi: 10.1016/s0092-8674(00)81660-8. [DOI] [PubMed] [Google Scholar]
Shepard LA, Heuck AP, Hamman BD, Rossjohn J, Parker MW, Ryan KR, Johnson AE, Tweten RK. Identification of a membrane-spanning domain of the thiol-activated pore-forming toxin clostridium perfringens perfringolysin O: an alpha-helical to beta-sheet transition identified by fluorescence spectroscopy. Biochemistry. 1998;37:14563–14574. doi: 10.1021/bi981452f. [DOI] [PubMed] [Google Scholar]
Soltani CE, Hotze EM, Johnson AE, Tweten RK. Specific protein-membrane contacts are required for prepore and pore assembly by a cholesterol-dependent cytolysin. Journal of Biological Chemistry. 2007a;282:15709–15716. doi: 10.1074/jbc.M701173200. [DOI] [PMC free article] [PubMed] [Google Scholar]
Soltani CE, Hotze EM, Johnson AE, Tweten RK. Structural elements of the cholesterol-dependent cytolysins that are responsible for their cholesterol-sensitive membrane interactions. PNAS. 2007b;104:20226–20231. doi: 10.1073/pnas.0708104105. [DOI] [PMC free article] [PubMed] [Google Scholar]
Tang G, Peng L, Baldwin PR, Mann DS, Jiang W, Rees I, Ludtke SJ. EMAN2: an extensible image processing suite for electron microscopy. Journal of Structural Biology. 2007;157:38–46. doi: 10.1016/j.jsb.2006.05.009. [DOI] [PubMed] [Google Scholar]
Terwilliger T. SOLVE and RESOLVE: automated structure solution, density modification and model building. Journal of Synchrotron Radiation. 2004;11:49–52. doi: 10.1107/S0909049503023938. [DOI] [PubMed] [Google Scholar]
Tilley SJ, Orlova EV, Gilbert RJ, Andrew PW, Saibil HR. Structural basis of pore formation by the bacterial toxin pneumolysin. Cell. 2005;121:247–256. doi: 10.1016/j.cell.2005.02.033. [DOI] [PubMed] [Google Scholar]
Tweten RK, Parker MW, Johnson AE. The cholesterol-dependent cytolysins. Current Topics in Microbiology and Immunology. 2001;257:15–33. doi: 10.1007/978-3-642-56508-3_2. [DOI] [PubMed] [Google Scholar]
Tweten RK. Cholesterol-dependent cytolysins, a family of versatile pore-forming toxins. Infection and Immunity. 2005;73:6199–6209. doi: 10.1128/IAI.73.10.6199-6209.2005. [DOI] [PMC free article] [PubMed] [Google Scholar]
van Pee K, Mulvihill E, Müller DJ, Yildiz Ö. Unraveling the pore-Forming steps of pneumolysin from Streptococcus pneumoniae. Nano Letters. 2016;16:7915–7924. doi: 10.1021/acs.nanolett.6b04219. [DOI] [PubMed] [Google Scholar]
Weis S, Palmer M. Streptolysin O: the C-terminal, tryptophan-rich domain carries functional sites for both membrane binding and self-interaction but not for stable oligomerization. Biochimica Et Biophysica Acta (BBA) - Biomembranes. 2001;1510:292–299. doi: 10.1016/S0005-2736(00)00360-6. [DOI] [PubMed] [Google Scholar]
Zwart PH, Afonine PV, Grosse-Kunstleve RW, Hung LW, Ioerger TR, McCoy AJ, McKee E, Moriarty NW, Read RJ, Sacchettini JC, Sauter NK, Storoni LC, Terwilliger TC, Adams PD. Automated structure solution with the PHENIX suite. Methods in Molecular Biology. 2008;426:419–435. doi: 10.1007/978-1-60327-058-8_28. [DOI] [PubMed] [Google Scholar]

eLife. 2017 Mar 21;6:e23644. doi: 10.7554/eLife.23644.035

Decision letter

Editor: Sjors HW Scheres¹

In the interests of transparency, eLife includes the editorial decision letter and accompanying author responses. A lightly edited version of the letter sent to the authors after peer review is shown, indicating the most substantive concerns; minor comments are not usually included.

Thank you for submitting your article "CryoEM structures of membrane pore and prepore complex reveal cytolytic mechanism of Pneumolysin" for consideration by eLife. Your article has been reviewed by three peer reviewers, one of whom, Sjors HW Scheres, is a member of our Board of Reviewing Editors, and the evaluation has been overseen by Richard Aldrich as the Senior Editor. The following individuals involved in review of your submission have agreed to reveal their identity: Christos G Savva (Reviewer #2).

The reviewers have discussed the reviews with one another and the Reviewing Editor has drafted this decision to help you prepare a revised submission.

Summary:

The study by van Pee and co-workers describes the cryo-EM structures of the cholesterol dependent cytolysin (CDC) pneumolysin (PLY) in the membrane-inserted and pre-pore states by single particle and tomography respectively. Although the first crystal structure of a CDC (perfringolysin O) in the water soluble state was solved almost 20 years ago, a high-resolution structure of the membrane-inserted form has been elusive due to the inherent structural heterogeneity of these complexes with the best results so far limited to the resolution of ~10-15Å where domains can be approximately identified but the exact mechanics of the transition from soluble to membrane-inserted cannot be seen. Despite the recent developments of direct electron detectors and better software that have pushed the resolution limits of Cryo-EM, this class of pore forming toxins still poses difficulties due to the irregular size and shape of the pores. In this study authors have succeeded in preparing more homogeneous samples by systematically screening various detergents. The end result is the first near-atomic resolution structure of a CDC in the membrane-inserted state, which provides valuable information for the mechanics of pore formation including the large movements of domains 2 and 3, which have not been resolved previously by other studies. In general the manuscript is well written and is an important step for the field of pore forming proteins of the CDC/MACPF family. Therefore, the three reviewers were overall positive about publishing this work, although several important issues were raised that should be addressed in a revised version.

Essential revisions:

1) The overall resolution of the single-particle reconstruction is limited to 4.5A, whereas the entire paper discusses the resulting atomic model as if this were a atomic-resolution structure. Locally the resolution is probably even worse than 4.5A, as for example the β strands are not well separated in Figure 1A. Overall, too little information is given on the quality of the map and the corresponding atomic model. In fact, at 4.5A, the map alone will probably still leave considerable scope for registry or main chain tracing errors. Therefore, the authors should explicitly discuss these caveats where relevant in the main text. In addition, they should provide the following (supplementary?) information in their revision:

– A map that displays variations in local resolution (e.g. as calculated in resmap).

– A view of a segmented monomer from the sharpened map (possibly also colored by local resolution) would be helpful to further evaluate the quality of the map

– Zoomed-in views of density and model, or movie(s) of such, for various regions in the structure

– FSC curve between the refined model and the map. And because overfitting can be important at this resolution, preferably also FSC curves for refinement in one half-map, combined with FSC curves between the refined model and the other half-map.

– A more detailed description of the model building and refinement procedures, which is transparent about which parts of the model are less well defined than others.

2) At the reported resolution the salt bridge between Asp168 and Lys271 should be interpreted carefully. A close up of the EM densities around these residues should be provided. Although the authors have mutated Asp168 and seen a 20% reduction in hemolytic activity this is not conclusive of a salt-bridge disruption. Did mutation of Lys271 also result in the same phenotype? In addition the mutation seems to affect membrane binding as the authors mention (Figure 7 legend) and the ring attachment from the membranes (subsection “Determinants of lytic activity”). Does this imply that pores are formed but are then released? Previous AFM studies by the authors (van Pee et. Al, Nano Letters, 2016) suggest the process is irreversible. Some clarity and caution interpreting this salt bridge would improve the manuscript.

3) Also related to the limited resolution: interactions of the Trp-rich loop are discussed (Results section) but the electron density map in that region looks poorly resolved. This is a very important part of a CDC and the motif defines the family so another panel is required to show the quality of the electron density map in that region. (A close up stereo showing side-chains would help). There is great interest in seeing what conformation this loop adopts in the prepore and pore states since it is thought to control prepore to pore conversion. In Results paragraph two, the authors state that the undecapeptide is interacting with Thr405. This sentence should be re-worded to provide more detail. What is the nature of this interaction and again is the resolution of this area sufficient to justify this comment? From Figure 2C the loop itself does not seem to be resolved even at the Cα backbone level.

4) A major finding of this work is that domain 2 collapses but does not lose its structure (Results section). Earlier EM studies suggested it does lose its structure but modeling and FRET studies suggested otherwise. More discussion of this controversy is required. Importantly, the electron density of this key region should be shown in a figure to satisfy the reader that the structure has been maintained. (Figure 2E is not convincing enough. A stereo with side-chains focused on D2 only would be easier to evaluate). Some new detailed insights are discussed; namely, the remodeled helix-turn-helix motif and helix alpha3a. Can stereo figures of the electron density with side-chains displayed of these regions be shown?

5) The electron cryo-tomography of the prepore reveals a structure of 34 protomers in the prepore, a number expected based on numerous other reports. However, the CryoEM structure of the pore reveals a larger 42 protomer pore. This is unexpected but interestingly the authors note a range of ring structures were present with this being the most stable. So the question arises whether the pore structure is an artifact. Can the authors model a 34 protomer pore based on their 42 protomer structure and convince that the details they discuss in the 42 protomer pore are likely relevant to the smaller pore? Is the 20º incline in the β-hairpin maintained in modeling the smaller pore? Like the authors, other workers have found that PLY form linear oligomers in their crystals. The authors should discuss how linear oligomers fit into their detailed mechanism of pore formation.

6) The authors have skirted around another controversial issue of whether arcs and incomplete rings observed by other workers are physiological relevant as they are not observed in this work. Some discussion of why PLY does not form such structures in their hands but do in others would be insightful.

7) Subsection “Three stages of PLY pore formation”: In the previous AFM study (van Pee et. Al, Nano Letters, 2016) only 13% of the lower height rings were pre-pores on supported lipid membranes. This could be mentioned to indicate that the occurrence of a "late prepore" is rare event even on membranes without curvature.

8) The literature review, particularly in the Discussion, has been narrowly focused to a large degree on the PLY literature and should be broadened out to other CDCs, particularly to PFO where there is an extensive literature on how it forms pores. Does the current study support the major findings in the PFO literature and does it disagree with any aspects?

9) Parts of the Discussion should be moderated. The limited resolution and a few static snapshots do not reveal all the details of membrane insertion and pore formation. Such details will come gradually from a host of biochemical and biophysical approaches. The suggestion that the CryoEM structure forms a good basis for drug discovery to treat numerous Gram-positive infections is not supported by any results in the manuscript so should be restricted to a speculative claim about S. pneumoniae.

10) Please include crystallization and crystallography methods.

[Editors' note: further revisions were requested prior to acceptance, as described below.]

Thank you for resubmitting your work entitled "CryoEM structures of membrane pore and prepore complex reveal cytolytic mechanism of Pneumolysin" for further consideration at eLife. Your revised article has been favorably evaluated by Richard Aldrich (Senior editor) and three reviewers, one of whom is a member of our Board of Reviewing Editors.

The reviewers agreed that this manuscript has improved considerably and is almost ready for publication. However, there was also consensus that a few of the main points raised in the first round have not been dealt with satisfactorily. These are:

1) Over-interpretation of lower-resolution parts of the map in terms of the atomic model. Although the new supplementary figures now reveal the limitations of the map more clearly, over-interpretation in the text still remains a problem. Specifically, W433 is mentioned to move wrt the crystal structure, yet there is no density visible for W433 in Video 1, Figure 2—figure supplement 2 or 3 to support this statement. The map in this region is ~6Å. The same can be said for H407 that does also not have clear density. The authors should stay away from over-interpretation in this region and simply state that this domain is likely to contribute to the inter-molecular interactions that stabilize the pore, as they have done in the Discussion section, paragraph three. Likewise, there is no obvious density that connects D168 and K271 (Figure 5 instead shows connecting density between β-strands that is expected at resolutions lower than 4.7A). The resolution in this area is again ~6-7Å according to the local resolution estimation. The authors should be more cautious and can suggest that a salt bridge may exist as evident by the point mutant D168A, which has an 80% reduction in hemolytic activity.

2) The comment in the Discussion section: "The 4.5 Å cryoEM structure of the PLY pore complex thus paves the way towards the design of new drugs that inhibit toxin oligomerization and pore formation as a promising approach towards combating infections by wide-spread and dangerous Gram-positive pathogens" jars with the reply "the different CDCs seem to follow different pathways of pore formation…. Any more discussion of PFO papers would distract from the main focus of our work, which is on PLY." Therefore, the authors should focus the therapeutic claim focus to diseases that PLY has shown to contribute.

eLife. 2017 Mar 21;6:e23644. doi: 10.7554/eLife.23644.036

Author response

Essential revisions:

1) The overall resolution of the single-particle reconstruction is limited to 4.5A, whereas the entire paper discusses the resulting atomic model as if this were a atomic-resolution structure.

We are aware, and do not approve of the general tendency to overstate the resolution and quality of cryoEM maps. In this case, however, we have a high-resolution x-ray structure of the soluble form of PLY, and several of its domains move as rigid bodies. We would therefore be justified to describe at least these parts as an atomic resolution structure. Nevertheless we have refrained from such a description, and instead describe the structure in terms of an atomic model, which it no doubt is.

Locally the resolution is probably even worse than 4.5A, as for example the β strands are not well separated in Figure 1A. Overall, too little information is given on the quality of the map and the corresponding atomic model. In fact, at 4.5A, the map alone will probably still leave considerable scope for registry or main chain tracing errors.

Register and chain-tracing errors in the β-strands of the barrel are actually not a problem, because the first and last stretches of each strand are fixed in the x-ray structure of domain 3. Together with the hydrophobic residues facing the lipid bilayer on the outside of the barrel and the polar and charged residues facing the inside, this leaves virtually no room for chain trace errors even at 4.5 Å. Moreover, we know that the β-strands in the barrel are well-resolved because of the sharp peak at 4.8 Å in the FSC curve (Figure 1—figure supplement 2B), which is the characteristic distance between β-strands. We can therefore be sure that our atomic model of the β-barrel, which is the major new part of the structure, is correct and accurate within the constraints of the map.

Therefore, the authors should explicitly discuss these caveats where relevant in the main text. In addition, they should provide the following (supplementary?) information in their revision:

– A map that displays variations in local resolution (e.g. as calculated in resmap).

– A view of a segmented monomer from the sharpened map (possibly also colored by local resolution) would be helpful to further evaluate the quality of the map

It is true that the local resolution of some parts of the map, especially the loops connecting the long β-strands and one loop in the periphery of domain 4, is worse than 4.5 Å, but the resolution of domains 1, 2 and parts of domain 4 is significantly better. The local resolution estimate (new Figure 1—figure supplement 3) shows this clearly for the whole pore complex and for an individual protomer seen from different directions. The new supplementary figures demonstrate the excellent quality of the map, which resolves many of the tryptophan and tyrosine sidechains (Figure 2—figure supplements 1–3). This would be expected at 4 Å rather than at 4.5 Å resolution.

– Zoomed-in views of density and model, or movie(s) of such, for various regions in the structure

We included five new zoomed-in figures as supplements to Figure 2:

Figure 2—figure supplement 1: stereo diagram of the interface between domain 1 (red) and domain 3 (green) drawn as a ribbon-and-stick model. The local resolution estimate indicates 4 – 4.5 Å in this map region.

Figure 2—figure supplement 2: superposition of domain 4 of the membrane-inserted form (cryoEM structure, green) and the soluble form (x-ray structure, blue, pdb-id 5aod) in the cryoEM map drawn at two contour levels.

Figure 2—figure supplement 3: stereo diagrams of the undecapeptide (bright green) and domain 4 loops (blue) of the neighboring monomer on the membrane surface, seen from two directions.

Figure 2—figure supplement 4: stereo diagram of domain 3 (green) with the upper part of the β-barrel and the new helix-turn-helix motif (HTH) that forms the α-barrel inside the pore as a stick model in the cryoEM map. The local resolution estimate indicates 4 to 4.5 Å for the β-barrel outside the lipid bilayer and around 5 Å for the α-barrel. Also shown is the new helix α3a (red and yellow) of the neighboring monomer, indicating a potential ionic inter-monomer interaction of Lys196 with Asp59 that would stabilize the membrane-inserted pore complex.

Figure 2—figure supplement 5: stereo diagram of β-strands in domain 2 as stick model in the cryoEM map at two contour levels. The local resolution is around 4.5 Å.

– FSC curve between the refined model and the map.

The FSC curve between the model and the map is now included in Figure 1—figure supplement 2B (blue curve). Note that the model was used as built into the map, but not refined, for the reasons given below. Like the masked FSC, this curve shows a sharp peak at 4.8 Å, due to the strong contribution of the resolved long β hairpins in the barrel. The correlation factor at the steep drop beyond this peak at 4.5 Å is not 0.5 (which is an arbitrary value), but approximately 0.3. We draw the reviewers’ and editor’s attention to two papers, which will be familiar to the BRE of our study, that report similar thresholds for the map against model FSC of the 3.4 Å structure of the ribosome/Sec61 complex (Voorhees et al., Cell 2014) and the tri-snRNP spliceosome at 5.9 Å (Nguyen et al., Nature 2015).

And because overfitting can be important at this resolution, preferably also FSC curves for refinement in one half-map, combined with FSC curves between the refined model and the other half-map.

We can rule out overfitting for three reasons: (1) Substantial portions of the model were not built from scratch, but they were rigid-body fitted domains from our 2.4 Å x-ray structure of soluble PLY. These map regions cannot be overfitted. (2) The major new part of the membrane-inserted form of the PLY protomer is a β-barrel with 84 long β-hairpins of rigid, pre-defined geometry, so the risk of overfitting in this region is minimal. (3) The comparison of the phase-randomized vs masked FSC (Figure 1—figure supplement 2B) indicates that there is no overfitted noise.

– A more detailed description of the model building and refinement procedures, which is transparent about which parts of the model are less well defined than others.

The individual domains of the PLY x-ray structure were fitted manually into the cryoEM map as rigid bodies with COOT. Where necessary, flexible or refolded protein regions were then refitted manually, followed by geometry regularization in COOT. We had described this procedure clearly in the methods of the original manuscript, but now describe it also in the main text, where it is less easily overlooked.

Fitting and refinement with Phenix or Refmac was tried but did not work, because these automatic procedures attempted to e.g. fit the membrane-inserted β-hairpins into the amphipol density around the hydrophobic part of the pore. The densities of amphipol and protein are comparable at 4-4.5 Å resolution, and only manual fitting can differentiate between them. Therefore, it was unfortunately not possible to employ automatic model fitting and refinement. It is well known that refinement of x-ray structures at similar resolutions also does not work, for the same reasons.

2) At the reported resolution the salt bridge between Asp168 and Lys271 should be interpreted carefully. A close up of the EM densities around these residues should be provided.

Even though this salt bridge is in the least well-defined map region near the tip of the long β-hairpins, there was significant density for it, as shown in the new panel G of Figure 5. We would like to point out that we did not look for this salt bridge, but deduced that there had to be one in this position from our chain trace, and then found the corresponding map density, which again shows there were no chain tracing errors in the β-barrel. Moreover, the interaction of adjacent β-hairpins that includes this salt bridge is in excellent agreement with the S-S crosslinking studies of Tweten and colleagues in the related PFO (Sato et al. 2013, NatChemBiol). This is now explained more fully in subsection “CryoEM structure of the PLY pore complex” paragraph four of the revised manuscript.

Although the authors have mutated Asp168 and seen a 20% reduction in hemolytic activity this is not conclusive of a salt-bridge disruption. Did mutation of Lys271 also result in the same phenotype?

The reviewer did not read our manuscript carefully. The mutation of Asp168 to alanine reduces the hemolytic activity by 80%, not to 80%! This is exactly the strong effect that one would expect from the disruption of a salt bridge stabilizing the membrane-inserted form. We do not see the point of mutating Lys271 as well.

In addition the mutation seems to affect membrane binding as the authors mention (Figure 7 legend) and the ring attachment from the membranes (subsection “Determinants of lytic activity”). Does this imply that pores are formed but are then released? Previous AFM studies by the authors (van Pee et. Al, Nano Letters, 2016) suggest the process is irreversible. Some clarity and caution interpreting this salt bridge would improve the manuscript.

We apologize if this was not clear in the original manuscript. In the original Figure 7 legend we wrote “Many PLY_D168A rings become detached from the liposomes, probably due to reduced binding affinity, and tend to break into fragments.” At this stage, the rings are of course prepores, and it is these, not the pores, that become detached and break into fragments. In the revised manuscript (subsection “Determinants of lytic activity” paragraph two) and Figure legend, we have clarified this point.

Note that the AFM study (van Pee et. al, Nano Letters, 2016, Figure 5) describes the transformation of membrane-attached rings into pores as irreversible, not the membrane attachment of the rings. Rings of this PLY mutant in the early prepore state detach from the membrane easily, but only before they (irreversibly) insert into the membrane to form the pore. This is now discussed in on paragraph two of the Discussion section.

The new Figure 2—figure supplement 2 shows domain 4 of PLY in the cryoEM (green) and x-ray structure from the inside of the ring. Differences between the undecapeptide structure in the soluble and pore forms of PLY are surprisingly small. The closest distance between Trp433 and Asp403 or Thr405 is 4-5 Å.

The stereo image of Figure 2—figure supplement 3 (see response to comment 1 above) shows the interactions of the D4 loops between two monomers. These interactions and their consequences are now discussed in more detail in the revised manuscript in paragraph three of the Discussion section.

In the earlier cryoEM studies no intramolecular details were resolved and the rotation of domain 2 was interpreted as a “collapse”. To propose a rotation of domain 2 would have been more plausible but was not suggested. The FRET study (Ramachandran et al. 2005, PNAS) was interpreted in terms of a bending of domain 2 to account for the height change observed by AFM (Czajkowsky et al. 2004, PNAS). Our present cryoEM structure now shows that both interpretations were incorrect. We rewrote the corresponding text passage (Discussion section paragraph three) that now includes a reference to the FRET study. The AFM study had already been cited.

The new Figure 2—figure supplement 4 shows the remodeled helix-turn-helix (HTH) motif as stick models and Figure 2—figure supplement 5 the β-strands of domain 2.

We do not think that the numbers of subunits we found in the PLY prepore and pore are at all unexpected. Tilley et al. (Cell, 2005) reported 31 ± 3 monomers in the PLY prepore, and two classes with 38 or 44 monomers in the PLY pore, in excellent overall agreement with our findings. Both the prepore and pore of Tilley et al. were in liposomes, so there can be no question that the number of subunits in our detergent-solubilized PLY pores is artefactual.

Can the authors model a 34 protomer pore based on their 42 protomer structure and convince that the details they discuss in the 42 protomer pore are likely relevant to the smaller pore? Is the 20º incline in the β-hairpin maintained in modeling the smaller pore?

It must have escaped the reviewer’s notice that we already built a 34-protomer model into the smaller pore of the subtomogram average, and that this was shown in Figure 6E of the original manuscript. No significant modification of the pore subunit, in particular not of the 20° incline of the β-hairpins relative to the membrane normal, was necessary to fit 34 copies of the pore-forming subunit into the 27 Å-resolution cryoET map. This is now pointed out more clearly in subsection “Three stages of PLY pore formation” of the revised manuscript

Like the authors, other workers have found that PLY form linear oligomers in their crystals. The authors should discuss how linear oligomers fit into their detailed mechanism of pore formation.

A description of how the linear arrays of the various x-ray structures relate to the prepore and pore form of PLY, and how this fits our detailed mechanism of pore formation can be found in our recent AFM paper (van Pee et al., Nano Lett 2016; supporting figure S2 and text relating to it).

We certainly have no intention of skirting around this issue. Arcs and slits of PLY_wt are actually visible in Figure 7A. This figure was there in the original manuscript, but we now point them out specially with coloured arrows. However, in contrast to LLO, PLY forms mostly complete rings.

We thank the reviewer for this suggestion, which we have taken up in paragraph two of subsection “Three stages of PLY pore formation2” of the revised manuscript.

Notwithstanding their high degree of sequence identity and similar monomer and oligomer structures, the different CDCs seem to follow different pathways of pore formation. Of the 21 papers on pore-forming toxins that we cite in our manuscript, more than half deal with PFO. Any more discussion of PFO papers would distract from the main focus of our work, which is on PLY.

The cryoEM structure of the PLY pore complex shows clearly the interactions of the secondary structure elements of neighboring protomers. The structure further shows how elements of secondary structure present in the soluble form refold and take on a completely different structure in the membrane. While the formation of the long membrane-inserted β-hairpins was predicted as such, no one could know or predict the exact structure of the β-strands and how they interact to form the membrane-perforating barrel. Other structural elements, including the inner α-barrel and the newly formed helix in domain 1, are entirely new. These elements are involved in inter- and intramolecular interactions in the pore complex. Even at 4.5 to 5 Å resolution there can be no reasonable doubt of their existence and interaction. Targeting these regions with newly developed drugs would be an elegant way to inhibit pore formation and infection, and thus an effective means of putting these pathogen out of action. We think that our discussion of this aspect of the PLY structure is reasonable and adequate.

10) Please include crystallization and crystallography methods.

Done

[Editors' note: further revisions were requested prior to acceptance, as described below.]

As explained in our first response letter, the location of most residues in domain 4, which hardly changes its conformation upon membrane insertion, is predefined by the crystal structure of the soluble PLY monomer. Although the resolution of the cryoEM map in the loop regions is lower, there is little room for misinterpreting the loop conformations. Nevertheless, we changed the manuscript text as requested by the reviewers.

The statement:

“In D4, Trp433 at the tip of the highly conserved undecapeptide loop (Figure 3) that renders PLY cholesterol-specific (Soltani et al., 2007b) was shifted by ~9 Å into the map density (Figure 2—figure supplement 2, 3, Video 1). “

was changed to:

“In D4, the highly conserved undecapeptide loop (Figure 3) that renders PLY cholesterol-specific (Soltani et al., 2007b) was shifted by up to ~9 Å into the cryoEM map (Figure 2—figure supplement 2, 3, Video 1).”

The statement:

“The short distances of Asp403 and Thr405 to Trp433 and of His407 to Try461 in adjacent monomers suggest a critical role of the D4 loop not only in receptor recognition, but also in oligomer formation.“

was changed to:

“The close proximity of loop β18/19 that contains Asp403, Thr405, and His407 of one monomer to loop β22/23 and the uncedapeptide containing Trp433 of the adjacent monomer suggests a critical role of these loops not only in receptor recognition, but also in oligomer formation.”

The statement:

“The conserved Trp433 in the undecapeptide loop seems to form a polar interaction with Asp403 and Thr405 in the loop that connects β18/19 in the neighbouring monomer, and Try461 interacts with His407 of the next monomer, most likely by π-stacking (Figure 2—figure supplements 2, 3).”

was changed to:

“Loop β22/23 and the conserved undecapeptide loop with Trp433 at its tip appear to interact with loop β18/19 of the neighbouring monomer, thereby stabilizing the pore complex by inter- and intramolecular interactions (Figure 2—figure supplements 2, 3).”

Likewise, there is no obvious density that connects D168 and K271 (Figure 5 instead shows connecting density between β-strands that is expected at resolutions lower than 4.7A). The resolution in this area is again ~6-7Å according to the local resolution estimation. The authors should be more cautious and can suggest that a salt bridge may exist as evident by the point mutant D168A, which has an 80% reduction in hemolytic activity.

The statement:

“[…]and (3) an ion bridge between Asp168 and Lys271 in adjacent b-strands of the pore barrel (Figure 5G). Even though the hairpins at the ends of the trans-membrane b-strands are the least well-ordered part of the ring (Figure 1—figure supplement 3), there is significant map density for the salt bridge connecting them (Figure 5G).”

was changed:

“[…]and (3) ionic interactions between charged sidechains in adjacent β-strands of the pore barrel (Figure 5G). In particular, Asp168 and Glu170 in β–strand β7 are in a good position for forming a salt bridge with Lys271 in β–strand β10 of the next-door monomer (Figure 5G).”

The statement:

“[…]we mutated Asp168 that is involved in ionic inter-subunit interactions[…]”

was changed to:

“[…]we mutated Asp168 that, as the structure suggests, might be involved in forming a salt bridge between adjacent subunits[…]”

We apologize for not making this clear. As far as anyone knows, the mechanism of membrane insertion by rearrangement of the conserved domains 1 – 4 is the same for all CDCs. What we meant was that the exact pathways by which the membrane-inserted toxin monomers assemble into pores may differ between species. While most CDCs form circular pores, a small number of bacteria, for example Listeria monocytogenes or Streptococcus pyogenes, give rise to slit-like or arch-shaped pores. However, the structure of the membrane-inserted monomers in these slit-like or arch-shaped pores would be essentially the same as the membrane-inserted form of PLY. Therefore, compounds that interfere with refolding and membrane insertion of soluble monomers would prevent infection by Streptococcus pneumoniae and other CDC-producing Gram-positive bacteria. We have modified the last paragraph of the revised manuscript as follows:

“The mechanism of membrane insertion by rearrangement of the conserved domains 1 – 4 is likely to be the same for all CDCs. Therefore, compounds that interfere with refolding and membrane insertion of soluble monomers would prevent infection by Streptococcus pneumoniae and other CDC-producing Gram-positive bacteria that attack human cells with similar pore-forming toxins.

The 4.5 Å cryoEM structure of the PLY pore complex thus paves the way towards the design of new drugs that inhibit pore formation as a promising approach towards combatting infections by dangerous and wide-spread Gram-positive pathogens.”

[bib1] Aleshin AE, Schraufstatter IU, Stec B, Bankston LA, Liddington RC, DiScipio RG. Structure of complement C6 suggests a mechanism for initiation and unidirectional, sequential assembly of membrane attack complex (MAC) Journal of Biological Chemistry. 2012;287:10210–10222. doi: 10.1074/jbc.M111.327809. [DOI] [PMC free article] [PubMed] [Google Scholar]

[bib2] Allegretti M, Klusch N, Mills DJ, Vonck J, Kühlbrandt W, Davies KM. Horizontal membrane-intrinsic α-helices in the stator a-subunit of an F-type ATP synthase. Nature. 2015;521:237–240. doi: 10.1038/nature14185. [DOI] [PubMed] [Google Scholar]

[bib3] Allegretti M, Mills DJ, McMullan G, Kühlbrandt W, Vonck J. Atomic model of the F420-reducing [NiFe] hydrogenase by electron cryo-microscopy using a direct electron detector. eLife. 2014;3:e01963. doi: 10.7554/eLife.01963. [DOI] [PMC free article] [PubMed] [Google Scholar]

[bib4] Althoff T, Mills DJ, Popot JL, Kühlbrandt W. Arrangement of electron transport chain components in bovine mitochondrial supercomplex I1III2IV1. The EMBO Journal. 2011;30:4652–4664. doi: 10.1038/emboj.2011.324. [DOI] [PMC free article] [PubMed] [Google Scholar]

[bib5] Collaborative Computational Project, Number 4 The CCP4 suite: programs for protein crystallography. Acta Crystallographica. Section D, Biological Crystallography. 1994;50:760–763. doi: 10.1107/S0907444994003112. [DOI] [PubMed] [Google Scholar]

[bib6] Czajkowsky DM, Hotze EM, Shao Z, Tweten RK. Vertical collapse of a cytolysin prepore moves its transmembrane beta-hairpins to the membrane. The EMBO Journal. 2004;23:3206–3215. doi: 10.1038/sj.emboj.7600350. [DOI] [PMC free article] [PubMed] [Google Scholar]

[bib7] Dang TX, Hotze EM, Rouiller I, Tweten RK, Wilson-Kubalek EM. Prepore to pore transition of a cholesterol-dependent cytolysin visualized by electron microscopy. Journal of Structural Biology. 2005;150:100–108. doi: 10.1016/j.jsb.2005.02.003. [DOI] [PubMed] [Google Scholar]

[bib8] Darji A, Chakraborty T, Niebuhr K, Tsonis N, Wehland J, Weiss S. Hyperexpression of listeriolysin in the nonpathogenic species Listeria innocua and high yield purification. Journal of Biotechnology. 1995;43:205–212. doi: 10.1016/0168-1656(95)00138-7. [DOI] [PubMed] [Google Scholar]

[bib9] Dubochet J, Adrian M, Chang JJ, Homo JC, Lepault J, McDowall AW, Schultz P. Cryo-electron microscopy of vitrified specimens. Quarterly Reviews of Biophysics. 1988;21:129–228. doi: 10.1017/S0033583500004297. [DOI] [PubMed] [Google Scholar]

[bib10] Dudkina NV, Spicer BA, Reboul CF, Conroy PJ, Lukoyanova N, Elmlund H, Law RH, Ekkel SM, Kondos SC, Goode RJ, Ramm G, Whisstock JC, Saibil HR, Dunstone MA. Structure of the poly-C9 component of the complement membrane attack complex. Nature Communications. 2016;7:10588. doi: 10.1038/ncomms10588. [DOI] [PMC free article] [PubMed] [Google Scholar]

[bib11] Emsley P, Cowtan K. Coot: model-building tools for molecular graphics. Acta Crystallographica Section D Biological Crystallography. 2004;60:2126–2132. doi: 10.1107/S0907444904019158. [DOI] [PubMed] [Google Scholar]

[bib12] Farrand AJ, LaChapelle S, Hotze EM, Johnson AE, Tweten RK. Only two amino acids are essential for cytolytic toxin recognition of cholesterol at the membrane surface. PNAS. 2010;107:4341–4346. doi: 10.1073/pnas.0911581107. [DOI] [PMC free article] [PubMed] [Google Scholar]

[bib13] Frangakis AS, Hegerl R. Noise reduction in electron tomographic reconstructions using nonlinear anisotropic diffusion. Journal of Structural Biology. 2001;135:239–250. doi: 10.1006/jsbi.2001.4406. [DOI] [PubMed] [Google Scholar]

[bib14] Gilbert RJ. Inactivation and activity of cholesterol-dependent cytolysins: what structural studies tell us. Structure. 2005;13:1097–1106. doi: 10.1016/j.str.2005.04.019. [DOI] [PubMed] [Google Scholar]

[bib15] Grant T, Grigorieff N. Measuring the optimal exposure for single particle cryo-EM using a 2.6 å reconstruction of rotavirus VP6. eLife. 2015;4:e06980. doi: 10.7554/eLife.06980. [DOI] [PMC free article] [PubMed] [Google Scholar]

[bib16] Gu J, Wu M, Guo R, Yan K, Lei J, Gao N, Yang M. The architecture of the mammalian respirasome. Nature. 2016;537:639–643. doi: 10.1038/nature19359. [DOI] [PubMed] [Google Scholar]

[bib17] Hahn A, Parey K, Bublitz M, Mills DJ, Zickermann V, Vonck J, Kühlbrandt W, Meier T. Structure of a complete ATP synthase dimer reveals the molecular basis of inner mitochondrial membrane morphology. Molecular Cell. 2016;63:445–456. doi: 10.1016/j.molcel.2016.05.037. [DOI] [PMC free article] [PubMed] [Google Scholar]

[bib18] Harris JR, Lewis RJ, Baik C, Pokrajac L, Billington SJ, Palmer M. Cholesterol microcrystals and cochleate cylinders: attachment of pyolysin oligomers and domain 4. Journal of Structural Biology. 2011;173:38–45. doi: 10.1016/j.jsb.2010.07.010. [DOI] [PubMed] [Google Scholar]

[bib19] Hotze EM, Tweten RK. Membrane assembly of the cholesterol-dependent cytolysin pore complex. Biochimica Et Biophysica Acta (BBA) - Biomembranes. 2012;1818:1028–1038. doi: 10.1016/j.bbamem.2011.07.036. [DOI] [PMC free article] [PubMed] [Google Scholar]

[bib20] Kabsch W. Automatic processing of rotation diffraction data from crystals of initially unknown symmetry and cell constants. Journal of Applied Crystallography. 1993;26:795–800. doi: 10.1107/S0021889893005588. [DOI] [Google Scholar]

[bib21] Kimanius D, Forsberg BO, Scheres SH, Lindahl E. Accelerated cryo-EM structure determination with parallelisation using GPUs in RELION-2. eLife. 2016;5:e18722. doi: 10.7554/eLife.18722. [DOI] [PMC free article] [PubMed] [Google Scholar]

[bib22] Kirkham LA, Kerr AR, Douce GR, Paterson GK, Dilts DA, Liu DF, Mitchell TJ. Construction and immunological characterization of a novel nontoxic protective pneumolysin mutant for use in future pneumococcal vaccines. Infection and Immunity. 2006;74:586–593. doi: 10.1128/IAI.74.1.586-593.2006. [DOI] [PMC free article] [PubMed] [Google Scholar]

[bib23] Kremer JR, Mastronarde DN, McIntosh JR. Computer visualization of three-dimensional image data using IMOD. Journal of Structural Biology. 1996;116:71–76. doi: 10.1006/jsbi.1996.0013. [DOI] [PubMed] [Google Scholar]

[bib24] Köster S, van Pee K, Hudel M, Leustik M, Rhinow D, Kühlbrandt W, Chakraborty T, Yildiz Ö. Crystal structure of listeriolysin O reveals molecular details of oligomerization and pore formation. Nature Communications. 2014;5:3690. doi: 10.1038/ncomms4690. [DOI] [PubMed] [Google Scholar]

[bib25] Kühlbrandt W. Biochemistry. The resolution revolution. Science. 2014;343:1443–1444. doi: 10.1126/science.1251652. [DOI] [PubMed] [Google Scholar]

[bib26] Lawrence SL, Feil SC, Morton CJ, Farrand AJ, Mulhern TD, Gorman MA, Wade KR, Tweten RK, Parker MW. Crystal structure of Streptococcus pneumoniae pneumolysin provides key insights into early steps of pore formation. Scientific Reports. 2015;5:14352. doi: 10.1038/srep14352. [DOI] [PMC free article] [PubMed] [Google Scholar]

[bib27] Letts JA, Fiedorczuk K, Sazanov LA. The architecture of respiratory supercomplexes. Nature. 2016;537:644–648. doi: 10.1038/nature19774. [DOI] [PubMed] [Google Scholar]

[bib28] Leung C, Dudkina NV, Lukoyanova N, Hodel AW, Farabella I, Pandurangan AP, Jahan N, Pires Damaso M, Osmanović D, Reboul CF, Dunstone MA, Andrew PW, Lonnen R, Topf M, Saibil HR, Hoogenboom BW. Stepwise visualization of membrane pore formation by suilysin, a bacterial cholesterol-dependent cytolysin. eLife. 2014;3:e04247. doi: 10.7554/eLife.04247. [DOI] [PMC free article] [PubMed] [Google Scholar]

[bib29] Li X, Mooney P, Zheng S, Booth CR, Braunfeld MB, Gubbens S, Agard DA, Cheng Y. Electron counting and beam-induced motion correction enable near-atomic-resolution single-particle cryo-EM. Nature Methods. 2013;10:584–590. doi: 10.1038/nmeth.2472. [DOI] [PMC free article] [PubMed] [Google Scholar]

[bib30] Liao M, Cao E, Julius D, Cheng Y. Structure of the TRPV1 ion channel determined by electron cryo-microscopy. Nature. 2013;504:107–112. doi: 10.1038/nature12822. [DOI] [PMC free article] [PubMed] [Google Scholar]

[bib31] Lu P, Bai XC, Ma D, Xie T, Yan C, Sun L, Yang G, Zhao Y, Zhou R, Scheres SH, Shi Y. Three-dimensional structure of human γ-secretase. Nature. 2014;512:166–170. doi: 10.1038/nature13567. [DOI] [PMC free article] [PubMed] [Google Scholar]

[bib32] Marshall JE, Faraj BH, Gingras AR, Lonnen R, Sheikh MA, El-Mezgueldi M, Moody PC, Andrew PW, Wallis R. The crystal structure of pneumolysin at 2.0 å resolution reveals the molecular packing of the pre-pore complex. Scientific Reports. 2015;5:13293. doi: 10.1038/srep13293. [DOI] [PMC free article] [PubMed] [Google Scholar]

[bib33] McCoy AJ. Solving structures of protein complexes by molecular replacement with phaser. Acta Crystallographica Section D Biological Crystallography. 2007;63:32–41. doi: 10.1107/S0907444906045975. [DOI] [PMC free article] [PubMed] [Google Scholar]

[bib34] Mindell JA, Grigorieff N. Accurate determination of local defocus and specimen tilt in electron microscopy. Journal of Structural Biology. 2003;142:334–347. doi: 10.1016/S1047-8477(03)00069-8. [DOI] [PubMed] [Google Scholar]

[bib35] Mulvihill E, van Pee K, Mari SA, Müller DJ, Yildiz Ö. Directly observing the lipid-dependent self-assembly and pore-forming mechanism of the cytolytic toxin listeriolysin O. Nano Letters. 2015;15:6965–6973. doi: 10.1021/acs.nanolett.5b02963. [DOI] [PubMed] [Google Scholar]

[bib36] Murshudov GN, Vagin AA, Dodson EJ. Refinement of macromolecular structures by the maximum-likelihood method. Acta Crystallographica Section D Biological Crystallography. 1997;53:240–255. doi: 10.1107/S0907444996012255. [DOI] [PubMed] [Google Scholar]

[bib37] Nicastro D, Schwartz C, Pierson J, Gaudette R, Porter ME, McIntosh JR. The molecular architecture of axonemes revealed by cryoelectron tomography. Science. 2006;313:944–948. doi: 10.1126/science.1128618. [DOI] [PubMed] [Google Scholar]

[bib38] Olofsson A, Hebert H, Thelestam M. The projection structure of perfringolysin O (Clostridium perfringens theta-toxin) FEBS Letters. 1993;319:125–127. doi: 10.1016/0014-5793(93)80050-5. [DOI] [PubMed] [Google Scholar]

[bib39] Pettersen EF, Goddard TD, Huang CC, Couch GS, Greenblatt DM, Meng EC, Ferrin TE. UCSF chimera--a visualization system for exploratory research and analysis. Journal of Computational Chemistry. 2004;25:1605–1612. doi: 10.1002/jcc.20084. [DOI] [PubMed] [Google Scholar]

[bib40] Ramachandran R, Heuck AP, Tweten RK, Johnson AE. Structural insights into the membrane-anchoring mechanism of a cholesterol-dependent cytolysin. Nature Structural Biology. 2002;9:823–827. doi: 10.1038/nsb855. [DOI] [PubMed] [Google Scholar]

[bib41] Ramachandran R, Tweten RK, Johnson AE. The domains of a cholesterol-dependent cytolysin undergo a major FRET-detected rearrangement during pore formation. PNAS. 2005;102:7139–7144. doi: 10.1073/pnas.0500556102. [DOI] [PMC free article] [PubMed] [Google Scholar]

[bib42] Sato TK, Tweten RK, Johnson AE. Disulfide-bond scanning reveals assembly state and β-strand tilt angle of the PFO β-barrel. Nature Chemical Biology. 2013;9:383–389. doi: 10.1038/nchembio.1228. [DOI] [PMC free article] [PubMed] [Google Scholar]

[bib43] Scheres SH. RELION: implementation of a bayesian approach to cryo-EM structure determination. Journal of Structural Biology. 2012;180:519–530. doi: 10.1016/j.jsb.2012.09.006. [DOI] [PMC free article] [PubMed] [Google Scholar]

[bib44] Scheres SH. Semi-automated selection of cryo-EM particles in RELION-1.3. Journal of Structural Biology. 2015;189:114–122. doi: 10.1016/j.jsb.2014.11.010. [DOI] [PMC free article] [PubMed] [Google Scholar]

[bib45] Schrödinger LLC. The PyMOL Molecular Graphics System. Version 1.8PyMol. 2015

[bib46] Shatursky O, Heuck AP, Shepard LA, Rossjohn J, Parker MW, Johnson AE, Tweten RK. The mechanism of membrane insertion for a cholesterol-dependent cytolysin: a novel paradigm for pore-forming toxins. Cell. 1999;99:293–299. doi: 10.1016/s0092-8674(00)81660-8. [DOI] [PubMed] [Google Scholar]

[bib47] Shepard LA, Heuck AP, Hamman BD, Rossjohn J, Parker MW, Ryan KR, Johnson AE, Tweten RK. Identification of a membrane-spanning domain of the thiol-activated pore-forming toxin clostridium perfringens perfringolysin O: an alpha-helical to beta-sheet transition identified by fluorescence spectroscopy. Biochemistry. 1998;37:14563–14574. doi: 10.1021/bi981452f. [DOI] [PubMed] [Google Scholar]

[bib48] Soltani CE, Hotze EM, Johnson AE, Tweten RK. Specific protein-membrane contacts are required for prepore and pore assembly by a cholesterol-dependent cytolysin. Journal of Biological Chemistry. 2007a;282:15709–15716. doi: 10.1074/jbc.M701173200. [DOI] [PMC free article] [PubMed] [Google Scholar]

[bib49] Soltani CE, Hotze EM, Johnson AE, Tweten RK. Structural elements of the cholesterol-dependent cytolysins that are responsible for their cholesterol-sensitive membrane interactions. PNAS. 2007b;104:20226–20231. doi: 10.1073/pnas.0708104105. [DOI] [PMC free article] [PubMed] [Google Scholar]

[bib50] Tang G, Peng L, Baldwin PR, Mann DS, Jiang W, Rees I, Ludtke SJ. EMAN2: an extensible image processing suite for electron microscopy. Journal of Structural Biology. 2007;157:38–46. doi: 10.1016/j.jsb.2006.05.009. [DOI] [PubMed] [Google Scholar]

[bib51] Terwilliger T. SOLVE and RESOLVE: automated structure solution, density modification and model building. Journal of Synchrotron Radiation. 2004;11:49–52. doi: 10.1107/S0909049503023938. [DOI] [PubMed] [Google Scholar]

[bib52] Tilley SJ, Orlova EV, Gilbert RJ, Andrew PW, Saibil HR. Structural basis of pore formation by the bacterial toxin pneumolysin. Cell. 2005;121:247–256. doi: 10.1016/j.cell.2005.02.033. [DOI] [PubMed] [Google Scholar]

[bib53] Tweten RK, Parker MW, Johnson AE. The cholesterol-dependent cytolysins. Current Topics in Microbiology and Immunology. 2001;257:15–33. doi: 10.1007/978-3-642-56508-3_2. [DOI] [PubMed] [Google Scholar]

[bib54] Tweten RK. Cholesterol-dependent cytolysins, a family of versatile pore-forming toxins. Infection and Immunity. 2005;73:6199–6209. doi: 10.1128/IAI.73.10.6199-6209.2005. [DOI] [PMC free article] [PubMed] [Google Scholar]

[bib55] van Pee K, Mulvihill E, Müller DJ, Yildiz Ö. Unraveling the pore-Forming steps of pneumolysin from Streptococcus pneumoniae. Nano Letters. 2016;16:7915–7924. doi: 10.1021/acs.nanolett.6b04219. [DOI] [PubMed] [Google Scholar]

[bib56] Weis S, Palmer M. Streptolysin O: the C-terminal, tryptophan-rich domain carries functional sites for both membrane binding and self-interaction but not for stable oligomerization. Biochimica Et Biophysica Acta (BBA) - Biomembranes. 2001;1510:292–299. doi: 10.1016/S0005-2736(00)00360-6. [DOI] [PubMed] [Google Scholar]

[bib57] Zwart PH, Afonine PV, Grosse-Kunstleve RW, Hung LW, Ioerger TR, McCoy AJ, McKee E, Moriarty NW, Read RJ, Sacchettini JC, Sauter NK, Storoni LC, Terwilliger TC, Adams PD. Automated structure solution with the PHENIX suite. Methods in Molecular Biology. 2008;426:419–435. doi: 10.1007/978-1-60327-058-8_28. [DOI] [PubMed] [Google Scholar]

PERMALINK

CryoEM structures of membrane pore and prepore complex reveal cytolytic mechanism of Pneumolysin

Katharina van Pee

Alexander Neuhaus

Edoardo D'Imprima

Deryck J Mills

Werner Kühlbrandt

Özkan Yildiz

Roles

Abstract

Introduction

Results

CryoEM structure of the PLY pore complex

Figure 1. Overall structure of the PLY pore complex.

Figure 1—figure supplement 1. Negative stain and cryoEM of PLY solubilized in DDM, Cymal-6 and Amphipol.

Figure 1—figure supplement 2. Image processing of PLY rings.

Figure 1—figure supplement 3. Local resolution estimate of the PLY monomers and of the complete pore complex.

Video 1. Superposition of domain 4 of the PLY monomer from the crystal structure (blue) on domain 4 of the PLY monomer in the pore complex (green) rotated around an axis perpendicular to the membrane.

Figure 2. Soluble and membrane-inserted PLY monomer.

Figure 2—figure supplement 1. Stereo view of domain 1 presented as ribbon-and-stick model.

Figure 2—figure supplement 2. Superposition of domain 4 of the PLY monomer in the crystal structure (blue) and in the pore complex (green), viewed from the pore center.

Figure 2—figure supplement 3. Stereo view of domain 4 in the pore complex with potential inter- and intramolecular interactions of loops viewed from two sides.

Figure 2—figure supplement 4. Stereo view of domain 3 with the upper part of the β-barrel and the new helix-turn-helix motif (HTH) that forms the α-barrel inside the pore.

Figure 2—figure supplement 5. Stereo view of domain 2 seen along the ring axis.

Figure 3. Sequence alignment of selected cholesterol-dependent cytolysins.

Figure 4. Intermolecular interactions in the 4.5 Å cryoEM map of the PLY pore complex.

Figure 5. Charge distribution in the PLY pore complex.

Three stages of PLY pore formation

Figure 6. CryoET of PLY prepores and pores.

Figure 6—figure supplement 1. CryoET map of the PLY prepore with rigid-body fitted x-ray structures of the water-soluble toxin that forms rows in the 3D crystals.

Figure 7. Membrane binding and hemolytic activity of PLY .

Determinants of lytic activity

Figure 8. Location of functionally important PLY residues.

Table 1.

Discussion

Figure 9. Mechanism of membrane insertion and pore formation.

Video 2. Mechanism of pneumolysin pore formation.

Video 3. Model of conformational changes in an entire ring of 42 PLY monomers from early prepore to late prepore to pore, seen from the side and from above.

Video 4. Oblique view of PLY ring inserting into the membrane.

Materials and methods

Protein expression and purification

Site-directed mutagenesis and hemolysis assays

Liposome preparation and amphipol stabilization

Specimen preparation for negative-stain EM

Single-particle cryoEM and image processing

CryoET of PLY prepores and pores

Sub-tomogram averaging

Model building and analysis

Crystallization, data collection, structure determination, and refinement

Accession numbers

Acknowledgements

Funding Statement

Funding Information

Additional information

Competing interests

Author contributions

Additional files

Major datasets

References

Decision letter

Roles

Author response

ACTIONS

PERMALINK

RESOURCES

Similar articles

Cited by other articles

Links to NCBI Databases