An animal toxin-antidote system kills cells by creating a novel cation channel

Abstract

Toxin-antidote systems are selfish genetic elements composed of a linked toxin and antidote. The peel-1 zeel-1 toxin-antidote system in C. elegans consists of a transmembrane toxin protein PEEL-1 which acts cell autonomously to kill cells. Here we investigate the molecular mechanism of PEEL-1 toxicity. We find that PEEL-1 requires a small membrane protein, PMPL-1, for toxicity. Together, PEEL-1 and PMPL-1 are sufficient for toxicity in a heterologous system, HEK293T cells, and cause cell swelling and increased cell permeability to monovalent cations. Using purified proteins, we show that PEEL-1 and PMPL-1 allow ion flux through lipid bilayers and generate currents which resemble ion channel gating. Our work suggests that PEEL-1 kills cells by co-opting PMPL-1 and creating a cation channel.

Fig 1. PMPL-1 is necessary for PEEL-1 toxicity in C. elegans.

(A) Selfish activity of the peel-1 zeel-1 toxin-antidote system in C. elegans shown in a genetic cross of strains from Bristol (which has the genetic element) and Hawaii (which lacks the genetic element). Hermaphrodite worms heterozygous for the presence of peel-1 zeel-1 (p(+) z(+)/p(−) z(−)) have 25% inviable progeny. This is due to sperm-delivered PEEL-1 toxicity causing developmental arrest of zeel-1(−) progeny. (B) Proportion of worms dying from ectopic, heat shock-PEEL-1 expression. Two pmpl-1 mutant alleles (yak103 and yak52) provide resistance to toxicity. (C) Proportion of dead, arrested embryos from self-fertilizing hermaphrodites heterozygous for peel-1 zeel-1. n = total progeny scored. (D) Predicted domain structure of the PMPL-1 protein with mutant alleles shown. The hydrophobic helices are predicted to be monotopic, passing through one leaflet of a lipid bilayer. (E) Body wall and vulval muscle (magenta) of pmpl-1(yak103) worms with vulval muscle-specific expression of PEEL-1 alone (top) or PEEL-1 and PMPL-1 (bottom). Vulval muscles appears normal with PEEL-1 alone but are missing or atrophied when PEEL-1 and PMPL-1 are co-expressed in these cells. All worms expressing PEEL-1 and PMPL-1 in vulval muscle cells had missing or severely deformed vulval muscles (n = 22). peel-1 and pmpl-1 are both GFP tagged. All channels are shown in S2D Fig. Scale bar = 20 µm. Underlying data are available in S1 Data.

Fig 2. PEEL-1 and PMPL-1 are sufficient for toxicity in HEK293T cells.

(A) Cytotoxicity (measured by LDH release) of combinations of three constructs transfected in HEK293T cells. Each data point is a biological replicate, normalized to LDH release from transfection with peel-1::eGFP in the same experiment. All plots show means with SD. Transfections combine constructs coding for mCherry or eGFP (−) or a fluorescent-tagged protein (+): PEEL-1::eGFP (top), PMPL-1::mCherry (middle), or mCherry::ZEEL-1 (bottom). (B) Cytotoxicity of PMP3-like proteins alone or with PEEL-1::eGFP. The PMPL-1 yak52 (A47T) mutant protein, C. elegans PMPL-2, and the yeast homolog yPMP3 are shown. (C) Live-cell imaging of a single cell transfected with an ER-marker (mCherry::KDEL) and peel-1::eGFP or (D) pmpl-1::eGFP. The cell nucleus is indicated (N). PMPL-1 is also seen on the plasma membrane (PM) and on lipid droplets (inset). Scale bar = 10 µm. (E) Cytotoxicity is suppressed by addition of the GBR1 ER-retention tag on the C-terminus of PEEL-1::eGFP or PMPL-1::mCherry. P-values in (A) and (E) calculated using one-way ANOVA with Dunnett’s multiple comparisons test, comparing all samples to PEEL-1 alone in (A) and all samples to PEEL-1 with PMPL-1 in (E). In (B), multiple unpaired t-tests were used with Holm-Šídák test, comparing each PMP3-like protein alone to PMP3-like with PEEL-1 (***p < 0.001; ****p < 0.0001). Underlying data are available in S1 Data.

Fig 3. PEEL-1 toxicity results in HEK293T cell swelling.

(A) Brightfield images of cells transfected with indicated constructs. Cells were imaged 48 hrs after transfection. Scale bar = 25 µm. Arrows point to swollen cells. (B) Percent of transfected cells that appear normal or swollen under brightfield. Mean with SD of three biological replicates is shown. One hundred cells were scored for each condition in each biological replicate. (C) Selected frames of a live-cell imaging time-course experiment (see S1 Movie). A single cell is shown, co-expressing PEEL-1::eGFP and PMPL-1::mCherry. The nucleus is labeled (“N”). t = 0 is 16 hrs after transfection. Acute swelling can be seen (t = 2 hr 30 min), followed by a transient blob or protrusion (arrow) jetted out by the cell (t = 4 hr 15 min) and later reabsorbed (dotted arrow) (t = 4 hr 20 min). Scale bar = 10 µm. (D) A typical phenotype from a cell transfected with PEEL-1::eGFP and PMPL-1::mCherry at 48 hrs post-transfection. PEEL-1 and PMPL-1 can be seen on the plasma membrane (arrow) and in fragmented ER (arrowhead). Scale bar = 10 µm. Underlying data are available in S1 Data.

Fig 4. PEEL-1’s amphipathic helix is critical for toxicity.

(A) The AlphaFold2 predicted structure of PEEL-1 and (B) a helical wheel representation of the putative PEEL-1 amphipathic helix. Amphipathic helix residues are colored (pink = hydrophobic, blue = hydrophilic). (C) Cytotoxicity of a series of PEEL-1 C-terminal truncations expressed in HEK293T cells. The number of amino acids removed are indicated (ex. “−28” means the last 28 residues were removed). Each truncation removes an additional alpha helix. The “−65” truncation removes the amphipathic helix. (D) Percent dead worms after heat-shock PEEL-1 expression of the indicated truncation mutant. Fifty worms were assayed for each data point, and two independent transgenic lines were tested for each construct. (E) Cytotoxicity of PEEL-1 amphipathic helix missense mutants in HEK293T cells. Mutants are ordered by descending hydrophobic moment (µH). Six single mutants (left of dotted line) and three double mutants are shown (right of dotted line). All bar graphs show mean with SD. Statistics were performed using multiple unpaired t-tests with Holm-Šídák test, comparing each PEEL-1 alone to PEEL-1 and PMPL-1 (**p < 0.01; ***p < 0.001; ****p < 0.0001). (F) The average toxicity of missense mutants from (E) plotted against their hydrophobic moment. Underlying data are available in S1 Data.

Fig 5. PEEL-1 and PMPL-1 create a monovalent cation channel.

Current–voltage plots of whole-cell patch-clamp electrophysiology on (A) transfected cells, (B) Control cell line, and (C) Experimental cell line. High intracellular potassium (140 mM K⁺/8.6 mM Na⁺) and high extracellular sodium (145 mM Na⁺/4 mM K⁺) solutions are used. Currents elicited by a family of 0.5 s voltage steps from a −30 mV holding potential, from −100 mV to 60 mV, in 10 mV increments. Currents normalized to cell capacitance (pF). Negative and positive currents indicate cation flow into or out of the cell, respectively. (A) Plots from HEK293 cells acutely transfected with peel-1::eGFP or pmpl-1::mCherry. (B) Plots from Control cell line expressing constitutive eGFP and tetracycline-inducible pmpl-1::mCherry. (C) Plots from Experimental cell line expressing constitutive peel-1::eGFP and tetracycline-inducible pmpl-1::mCherry. Control and Experimental cell lines are shown without tetracycline or with tetracycline at the indicated time after addition of tetracycline. All plots shown as mean with SEM. (D) Western blot for PMPL-1::mCherry in Control and Experimental cell lines without tetracycline (−tet, left) and 18 hrs after addition of tetracycline (+tet, right). Leaky expression of PMPL-1 is seen in both cell lines in the absence of tetracycline. Less background PMPL-1 expression is seen in Experimental cells than in Control cells, likely because of selection against higher background PMPL-1 expression when in combination with PEEL-1 but not eGFP. GAPDH loading control shown. Original blots available in S1 Raw images. (E) Permeability of indicated ions was assayed in Experimental cells without tetracycline. Test ionic solutions substituted previous bath solution (145 mM Na⁺/4 mM K⁺) with 140 mM pure cations (external) or 20 mM anions (internal), except Ca²⁺ (20 mM external, 1 mM internal Ca²⁺ with 2.5 mM EGTA). All plots show mean with SEM. Statistical tests compare all results to NMDG (treated as control) in one-way ANOVA with Dunnett’s multiple comparisons test (*p < 0.05; ***p < 0.001). (F) Schematic of planar lipid bilayer experiment. Liposomes containing purified PEEL-1 or PMPL-1 are added to the cis well to deliver proteins to the lipid bilayer. (G) Conductance traces of one experiment (left) and a histogram of the trace (right, 2 pS bin width, normalized based on probability density) at an applied voltage of +180 mV. SDS–PAGE gels of purified proteins are shown in S15A Fig and more example conductance traces and controls are shown in S16 Fig. Underlying data are available in S1 Data.