The mechanism of toxicity in HET-S/HET-s prion incompatibility

Abstract

The HET-s protein from the filamentous fungus Podospora anserina is a prion involved in a cell death reaction termed heterokaryon incompatibility. This reaction is observed at the point of contact between two genetically distinct strains when one harbors a HET-s prion (in the form of amyloid aggregates) and the other expresses a soluble HET-S protein (96% identical to HET-s). How the HET-s prion interaction with HET-S brings about cell death remains unknown; however, it was recently shown that this interaction leads to a relocalization of HET-S from the cytoplasm to the cell periphery and that this change is associated with cell death. Here, we present detailed insights into this mechanism in which a non-toxic HET-s prion converts a soluble HET-S protein into an integral membrane protein that destabilizes membranes. We observed liposomal membrane defects of approximately 10 up to 60 nm in size in transmission electron microscopy images of freeze-fractured proteoliposomes that were formed in mixtures of HET-S and HET-s amyloids. In liposome leakage assays, HET-S has an innate ability to associate with and disrupt lipid membranes and that this activity is greatly enhanced when HET-S is exposed to HET-s amyloids. Solid-state nuclear magnetic resonance (NMR) analyses revealed that HET-s induces the prion-forming domain of HET-S to adopt the β-solenoid fold (previously observed in HET-s) and this change disrupts the globular HeLo domain. These data indicate that upon interaction with a HET-s prion, the HET-S HeLo domain partially unfolds, thereby exposing a previously buried ∼34-residue N-terminal transmembrane segment. The liberation of this segment targets HET-S to the membrane where it further oligomerizes, leading to a loss of membrane integrity. HET-S thus appears to display features that are reminiscent of pore-forming toxins.

Figure 1. HET-S in the presence of HET-s(218–289) amyloid seeds makes holes in liposomes observed by freeze-fracture electron microscopy and liposome leakage assays.

(A) TEM images of the replica of 100-nm diameter extruded liposomes incubated at 4°C in the absence of protein, with a higher magnification on the right. (B) Liposomes incubated in the presence of a mixture of HET-S and HET-s(218–289) fibril seeds display membrane damage: Hole-like structures ranging from a ∼10–∼60-nm width indicated by white arrows are present. In addition, species interpreted as protein aggregates are labeled by a black arrow. (C) The time-dependent release of calcein from E. coli polar liposomes induced by HET-S and HET-s(218–289) 10°C was measured at HET-S concentrations of 0.12–4 µM in the presence (solid lines) or absence (dashed lines) of 0.8 µM HET-s(218–289) amyloid seeds (monomer-equivalent concentration). (D) The calcein leakage measured in the presence (solid lines) or absence (dashed lines) of 8, 0.8, or 0.08 µM HET-S with either 0.008, 0.08, or 0.8 µM HET-s(218–289) amyloid seeds.

Figure 2. HET-S but not HET-s expression is toxic in E. coli.

The growth of E. coli cultures expressing various constructs of (A) HET-S (full-length, 1–227, or N-terminal Histag) or (B) HET-s (full-length, 1–227) was monitored by their OD₆₀₀. After induction of protein expression, the cell growth of cultures with HET-S or HET-S(1–227) but not HET-s or HET-s(1–227) was suppressed, indicating that the HET-S HeLo domain, but not HET-s or its HeLo domain, is toxic to E. coli. Further analysis of the cell pellet showed that HET-S is associated with the inner membrane of E. coli. (C) SDS-PAGE analysis of soluble and insoluble fractions after expression in E. coli at 37°C for 4 h. Lanes 1 and 2 represent the supernatant and pellet of the cell lysate, respectively. The pellet and supernatant fractions after treatment with 1 M urea are in lanes 5 and 6, respectively. HET-S remains in the insoluble fraction after incubation in 1 M Urea (lane 5).The pellet and supernatant fractions after extraction with the detergent N-lauroylsarcosine are shown in lanes 3 and 4, respectively, showing that HET-S is solubilized by this detergent. The identity of the HET-S and OmpF bands were confirmed by trypsin digest followed by MS analysis and are marked with an asterisk. (D) Coomassie stain (top) and Western blot (bottom) visualization of SDS-PAGE fractions from a sucrose density gradient separation of E. coli cell lysates after HET-S expression at 37°C for 4 h (lanes 1–13). Lanes L and M are the cell lysate before density gradient separation and the protein marker (SeeBlue Plus2, Invitrogen), respectively. The relative NADH oxidase activity of each fraction (indicating the location of the inner membrane) is plotted over the Western blot. HET-S is found in fractions that correspond to the inner membrane, whereas the outer membrane protein OmpF moves to a higher density. The majority of HET-S is in the soluble protein fractions while a small amount is also found on the very bottom of the tube (55% sucrose, fraction 1) whose density (>1.25 g/ml) indicates that it likely to be lipid-free (or low-lipid-content) protein aggregates.

Figure 3. Prion-independent (thermodynamic) activation of HET-S reveals that membrane binding and liposome leakage are distinct activities and that liposome binding is mediated through the N-terminus of HET-S.

(A) TEM image of freeze-fractured liposomes treated with HET-S at 30°C show that many liposomes contain a single large hole, indicated by white arrows similar to those with prion-activated HET-S (Figure 1B). Consistent with the TEM images, is (B) the prion-independent leakage caused by HET-S 30°C. The HET-s-like variant HET-S[E86K] and the HeLo domains of HET-S and HET-s do not cause leakage even at this elevated temperature. (C) SDS-PAGE analysis of 150,000 g supernatant (S) and pellet (P) fractions from a 20 µM HET-S incubation with liposomes reveals that HET-S associates with liposomes at 30°C but not at 4°C. (D) HET-s and HET-s(1–227) do not interact with liposomes and are found in the supernatant. In contrast, HET-S(1–227) and HET-S(E86K)—a HET-S mutant that cannot dimerize in solution and that has a [Het-s]-like phenotype in vivo—bind to liposomes after an extended period of incubation at 30°C. Despite being able to bind liposomes, neither of these two proteins causes leakage from liposomes (B).

Figure 4. Solid-state NMR reveals that HET-S adopts the HET-s(218–289) β-solenoid fold upon aggregation induced by HET-s.

NCA solid-state NMR spectra of various protein samples are shown. The NCA spectrum of the HET-s(218–289) fibrils alone is depicted in black contours in A, B, and C. The blue contours represent the NCA spectrum of [¹³C, ¹⁵N]-HET-S (A) co-fibrillized with unlabeled HET-s(218–289) monomer, (B) aggregated by amyloid seeds of unlabeled HET-s(218–289), (C) co-fibrillized with unlabeled monomeric full-length HET-s. For the amino acid residue G271, two cross-peaks are observed that are attributed to the presence of two interfaces in the mixed fibril sample, i.e., HET-S/HET-S and HET-S/HET-s. (D) Overlay of the PAIN spectrum (in red) of aggregates formed by [¹³C]-HET-S and [¹⁵N]-HET-s(218–289) and the PAIN of HET-s(218–289). The PAIN spectrum shows cross peaks only if there is an intermolecular polarization transfer between the two differently labeled molecules. The peak assignments are based on the chemical-shift assignment obtained by comparison with HET-s(218–289). (E) Chemical-shift differences between the β-solenoid fold of HET-s(218–289) fibrils and the PFD of HET-S upon aggregation with HET-s(218–289) fibril seeds. The absolute values of the differences in the N, C^α, and C^β chemical shifts between HET-s and HET-S for the residues assigned in HET-S (within the HET-s(218–289)/HET-S-aggregate) are displayed. Most chemical shift differences are within the errors of the measurements, which are about 0.2 ppm for ¹⁵N and 0.1 ppm for ¹³C. Larger shift differences are observed close to the amino acid difference between HET-S and HET-s, K235E. (F) A model of the PFD in the mixed aggregates depicting the interfaces that give rise to the two resonances for residue G271. The spectra show the relevant region from (A). The ratio of intensities G271∶G271′ indicates that there are an equal number of HET-S/HET-s and HET-S/HET-S interfaces in the aggregates from the 1∶1 mixture. (G) The N-terminus of HET-S remains associated to liposomes upon proteolytic degradation. Proteinase K-treatment (0°C, 30 min) of HET-S that had been pre-incubated in presence or absence of liposomes at 37°C was analyzed by SDS-PAGE. The liposome sample was centrifuged at 180,000 g for 30 min to separate the liposome-bound protein from the soluble protein. This analysis reveals that residues ∼1–35 of HET-S are associated with the liposome fraction and are protected from proteolysis by the non-specific protease PK. The bands indicated by an asterisk have been analyzed by Edman degradation. The N-terminal sequences of three major degradation products of the pelleted liposome-embedded material (masses of approximately 20 kDa and two at 7 kDa) are ⁻¹⁷[MRGSHHH] and ¹[MSEPFEI] and ¹¹[ALGVAG] as indicated. More bands could not be analyzed attributed to the high degree of overlap and low of concentration of the remaining bands. The two strongest bands in the supernatant of the protease-treated HET-S/liposome sample with approximate masses 28 kDa and 5 kDa and have N-terminal residues starting with ³⁵[GRD] and ²¹⁸[KID]. These two bands are consistent with fragments spanning residues 35–289 and 218–289 (the PFD) respectively. The first lane is the SeeBlue Plus2 protein standard and the last lane is PK alone. These data were reproduced three times.

Figure 5. Correlations between P. anserina heterokaryon incompatibility phenotype and in vitro liposome calcein leakage data.

(A) The TMHMM per-residue predictions (see also Figure S3) for HET-s (left) and HET-S (right) are mapped as a color gradient onto a ribbon diagram of the crystal structure of HET-S(1–227). A star indicates the locations of the two critical amino acid residues, 23 and 33, whose identities can interconvert the functionality of HET-s and HET-S. (B) The time course of prion-induced calcein leakage is shown for HET-s, HET-S, and several phenotype-interconverting variants at 4 µM protein concentration with 0.8 µM HET-s(218–289) amyloid seeds. Only HET-S and HET-s[D23A,P33H] show calcein leakage consistent with the HET-S toxic phenotype. The leakage is expressed as fraction based on the positive control experiment using 1% SDS. A comparison of the TM prediction, incompatibility phenotype, and liposome leakage activity of these variants is presented in Table S1. A detailed per-residue TMHMM prediction of these and several other variants of known phenotype is displayed in Figure S3).

Figure 6. HET-S oligomerizes in the membrane-like environment of the detergent FC-12.

(A) The time dependence of the oligomerization state of HET-S in 0.4% FC-12 was followed by size-exclusion chromatography with multiple-angle light scattering, UV, and refraction index detection. With time the monomeric species is depleted with a concomitant increase in the oligomeric one (violet = 0 h; blue = 1 h; green = 2.5 h; yellow = 3 h; orange = 3.5 h; red = 16 h). The grey trace is the size exclusion profile of HET-S that was extracted by 0.4% FC-12 from liposomes with which it had been incubated at 37°C for 2 h. The extracted sample is similar in monomer and oligomer content to the 16-h non-liposome measurement. The masses of the HET-S entities (not including the bound FC-12 molecules) as calculated by the Astra V software conjugate analysis are listed on above the chromatograms. The inlay shows that under the same conditions, HET-S 1–227 remains monomeric for 16 h with only a minor accumulation of a high molecular weight aggregate (no intermediate oligomers detected). (B) CD spectra of HET-S in 0.4% FC-12 were independently acquired for the same time course showing a time-dependent loss of alpha-helical content for the first 3 h upon the addition of detergent, after which the sample reaches a stable alpha helical content. Colors are as in (A).

Figure 7. Proposed mechanism for the generation of toxicity by the HET-s prion/HET-S system.

(1) In the fusion cell, HET-S (in blue with a red TM segment) encounters the β-solenoid structure of the HET-s prion (in brown). (2) HET-S binds to the β-solenoid structure through its own PFD segment, itself adopting the β-solenoid structure. The structural overlap of the HeLo domain and the PFD causes a partial unfolding of the HeLo domain of HET-S, represented here by the transition to a random coil conformation of its three C-terminal helices. (3) The destabilized HeLo domain of HET-S then expels its N-terminal TM segment (residues 1–34, in red). (4) The exposed TM segment targets the activated HET-s/HET-S complex to the membrane where it is able to penetrate the membrane through the formation of a TM helix. The catalytic nature of the activation of HET-S by the HET-s amyloid suggests that activated HET-S may also be released from the co-aggregate before it enters the membrane. The HET-S membrane-disrupting oligomer may form as a direct result of its complex with the HET-s amyloid or only first as it inserts into the membrane. However, the membrane integrity is disrupted by hole-like structures, thus triggering cell death. The model for the HET-s fibril was created from the PFD fibril structure and the HeLo domain structure with an unwinding the last three helices of the HeLo domain (residues 177–222) to make space for the HeLo domains around the fibril. The HET-s HeLo domains are depicted as dimers between adjacent monomers in the fibril, but these are speculative and it should be emphasized that the structures of the HeLo domains of HET-s and HET-S, in the context of a fibril, are not known except that they lose tertiary structure (more molten globule-like), with a local loss of secondary structure around residues 190–220 , indicated, in the figure, by the spheres around the HeLo domains.

Figure 8. HeLo domain alignment of the TM segment and HeLo/HET domain architecture comparison.

(A) The sequences in this alignment are the non-redundant output of a PSI-BLAST search with residues 4–33 of HET-S (carried to convergence at an E-value threshold of 0.005). Therefore, these 35 sequences are a subset of HeLo domains that are more similar to HET-S in their TM region. The output score of the TMHMM algorithm (sum of per-residue probability) is plotted to the right of the sequences with a double asterisk to indicate those that are predicted to have a TM helix. When only residues 1–38 are input into the algorithm many more (21 of 35) sequences are predicted to have a TM helix (indicated by a single asterisk). The GI accession numbers in red are for HET-S and HET-s and those in green are for the sequences which have the HET-S domain architecture (HeLo-PFD). (B) Two examples of HeLo domain-containing STAND proteins that have similar architectures to HET domain-containing STAND proteins. The known role of the HET domain as an inducer of cell death suggests that the HeLo domain may have a similar role.