The choanoflagellate pore-forming lectin SaroL-1 punches holes in cancer cells by targeting the tumor-related glycosphingolipid Gb3

Abstract

Choanoflagellates are primitive protozoa used as models for animal evolution. They express a large variety of multi-domain proteins contributing to adhesion and cell communication, thereby providing a rich repertoire of molecules for biotechnology. Adhesion often involves proteins adopting a β-trefoil fold with carbohydrate-binding properties therefore classified as lectins. Sequence database screening with a dedicated method resulted in TrefLec, a database of 44714 β-trefoil candidate lectins across 4497 species. TrefLec was searched for original domain combinations, which led to single out SaroL-1 in the choanoflagellate Salpingoeca rosetta, that contains both β-trefoil and aerolysin-like pore-forming domains. Recombinant SaroL-1 is shown to bind galactose and derivatives, with a stronger affinity for cancer-related α-galactosylated epitopes such as the glycosphingolipid Gb3, when embedded in giant unilamellar vesicles or cell membranes. Crystal structures of complexes with Gb3 trisaccharide and GalNAc provided the basis for building a model of the oligomeric pore. Finally, recognition of the αGal epitope on glycolipids required for hemolysis of rabbit erythrocytes suggests that toxicity on cancer cells is achieved through carbohydrate-dependent pore-formation.

Fig. 1. Classification of β-trefoil lectins according to the TrefLec database.

a Selected examples of β-trefoil lectins from different classes. The binding peptide is represented by a rainbow-colored ribbon, the bound sugar by sticks, and the conserved hydrophobic core-forming amino acids by spheres. b Sunburst statistics for predicted β-trefoil lectins in different classes, in selected domains of life. c Classification of predicted β-trefoil lectins and distribution of sequences in the TrefLec database. d Prediction of β-trefoil lectins with an aerolysin domain based on the corresponding CATH domain (CATH entry 2.170.15.10).

Fig. 2. SaroL-1 sequence information.

a Exerpt of the TrefLec page of the predicted lectin from Salpingoeca rosetta with information about the protein, the domains and the gene. b Peptide sequence of SaroL-1 with separation of the domains and alignment of lobes for the β-trefoil domain. Amino acids corresponding to the signature of Mytilec-like class are highlighted in yellow, amino acids predicted to be involved in carbohydrate binding in green, and amino acids involved in the hydrophobic core in blue.

Fig. 3. SaroL-1 recognizes αGal-containing ligands.

a Representative ITC isotherms of SaroL-1 with αGal1-4Gal (green), GalNAc (red), and lactose (βGal1-4Glc) (cyan), b Comparison of K_A values of various binding partners for SaroL-1, c 200 nM of SaroL-1 (green) binds to GUVs (red; fluorescent lipid Atto 647 N) functionalised with either FSL-Gb3, Gb3 wt, FSL-iGb3, and lactosylceramide (Lac-cer). SaroL-1 induces tubular membrane invaginations in some cases, as visible for FSL-Gb3 GUVs and SaroL-1 clustering on GUVs, as visible for Gb3 wt, FSL-iGb3 and Lac-Cer GUVs. GUVs without functional group (DOPC) serves as negative control and show no binding of SaroL-1. The GUVs were composed of DOPC, cholesterol, glycolipid of choice, and membrane dye to the molar ratio of 64.7:30:5:0.3, respectively. Scale bars are 10 μm.

Fig. 4. SaroL-1 shows dose-dependent binding and intracellular uptake into H1299 cells.

a Histograms of fluorescence intensity of gated living H1299 cells incubated for 30 min at 4 °C with increasing concentrations of SaroL-1-Cy5 (grey, dotted histogram: negative control, blue: 55 nM, green: 135 nM, orange: 271 nm). Shifts in fluorescence intensity indicated that SaroL-1 binds to the H1299 cell surface in a dose-dependent manner. b Histogram of fluorescence intensity of gated living H1299 cells pre-treated for 72 h with the GSL synthesis inhibitor PPMP and incubated for 30 min at 4 °C with increasing concentrations of SaroL-1-Cy5 (grey, dotted histogram: negative control, blue: 55 nM, green: 135 nM, orange: 271 nm). In the absence of glucosylceramide-based GSLs, including Gb3, the binding of SaroL-1 to the plasma membrane was remarkably reduced. c Confocal imaging of H1299 human lung epithelial cells incubated with 271 nM Cy5-conjugated SaroL-1 (red) for indicated time points at 37 °C. The fluorescent signals accumulate partially at the plasma membrane and in the intracellular space of treated cells. Nuclei were counterstained by DAPI. Scale bars represent 10 μm.

Fig. 5. Crystal structure of SaroL-1.

a Cartoon representation of monomeric SaroL-1 in complex with GalNAc. β-trefoil-domain colored in blue and aerolysin domain in green. The GalNAc ligands are displayed in their electron density map as sticks. b Superimposition of β-trefoil lectin domains, (7QE4, light magenta), (7R55, blue) in complex with 3 molecules of GalNAc (violet) and 2 molecules of Gb3 (cyan). c Zoom on α, β and γ binding sites with GalNAc (violet) polar contacts are represented as dashed lines and bridging water molecules as red spheres. d Zoom on the interactions with Gb3 (cyan) in binding β and γ sites, polar contacts are represented as dashed lines and bridging water molecules as red spheres. e Overlay of β-trefoil domains of SaroL-1 (blue) in complex with Gb3 (cyan) and of monomeric CGL (5F90) (yellow) in complex with Gb3 and αGal1-4Gal (yellow). f Comparison of the structures of monomeric SaroL-1, pore-forming lectin LSL (1W3A) and ε-toxin (1UYJ) from left to right. The pro-aerolysin domain is colored in green and the membrane-binding domain in blue (SaroL-1), red (LSL) and orange (ε-toxin).

Fig. 6. Pore-forming and hemolytic activity of SaroL-1.

a SaroL-1 (unlabeled, 200 nM) triggers the influx of 3 kDa dextran-AF488 (green) into wt Gb3-containing GUVs (red) via its pore-forming activity. In the control group without SaroL-1, there was no visible influx of dextran-AF488 detected. Yellow arrows indicate events of dextran-AF488 influx to wt Gb3-containing GUVs. The GUVs were composed of DOPC, cholesterol, wt Gb3, and membrane dye to the molar ratio of 64.7:30:5:0.3, respectively. The scale bars represent 10 µm. b Kinetics of SaroL-1 driven dextran-AF488 influx to wt Gb3-containing GUVs. Mean values ± SD are shown. Data represent three independent experiments, n = 3. The molecular weight of fluorescently labelled dextran is 3 kDa. The total amount of control GUVs was at 0 min—225 GUVs, 30 min—344 GUVs, 60 min—349 GUVs and 120 min—393 GUVs. For SaroL-1 experiment with wt Gb3-containing GUVs, the total amount of GUVs was 0 min—230 GUVs, 30 min—185 GUVs, 60 min—183 GUVs and 120 min—178 GUVs. In the PNPG-treated group there was a total amount of GUVs at 0 min—322 GUVs, at 30 min—435 GUVs, 60 min—447 GUVs and at 120 min 446 GUVs. c Relative hemolytic activity of SaroL-1 with estimation of IC₅₀ as 6.3 μg/mL (170 nM). Mean values ± SD are shown. Data represent two independent experiments, n = 2. d Relative inhibition of the hemolytic activity of SaroL-1 by PNPG, GalNAc, melibiose and lactose. Mean values ± SD are shown. Data represent two independent experiments, n = 2. All error bars correspond to mean value ± SD. Data for the graphs are available in Supplementary Data 1.

Fig. 7. Cytotoxic activity of SaroL-1 against H1299 cells.

a Dose-dependent increase of cytotoxicity following the addition of purified SaroL-1 in a standard cell proliferation assay (MTT) demonstrating increment of cytotoxicity after 24 h of incubation compared to treatment with PBS. Cell viability is reduced by ~87% after stimulation with 1.36 µM SaroL-1. Data represent three independent experiments, n = 3. b Cell proliferation assay (MTT) of H1299 cells pre-treated with PPMP for 72 h before addition of increasing concentrations of purified SaroL-1. Cytotoxicity was remarkably reduced after 24 h of incubation with SaroL-1 in absence of the glycosphingolipid Gb3 at the plasma membrane of treated cells. Data represent three independent experiments, n = 3. c The soluble sugar PNPG inhibited SaroL-1 cytotoxicity. H1299 were incubated with increasing concentrations of SaroL-1 pre-treated with 10 mM PNPG. Cell proliferation assay (MTT) was used to assess SaroL-1´s cytotoxic activity after 24 h in comparison to the treatment with PBS. The results indicate that cell viability is preserved when SaroL-1 glycan-binding sites are saturated with soluble 10 mM PNPG. Data represent three independent experiments, n = 3. d H1299 cells suffer of acute cytotoxicity and membrane damage in presence of SaroL-1. A lactate dehydrogenase (LDH) assay revealed impairment of cell membrane integrity upon incubation with 0.27 µM and 1.36 µM SaroL-1 after 2 and 4 h. Data represent two independent experiments, n = 2. Differences to the control were analyzed for significance by using two-tailed unpaired t-test. *p < 0.05, **p < 0.01, ***p < 0.001, ****p < 0.0001.

Fig. 8. Prediction and visualization of Sarol-1 pore oligomeric assembly.

a Preliminary model of membrane-bound heptameric SaroL-1 built by homology modeling using the structure similarity displayed in Fig. 3. b Crystal structure of heptameric ε-toxin of C. perfringens (PDB 6RB9). c Crystal structure of heptameric lactose-binding lectin CEL-III from Cucumaria echinata complexed with βGal derivative (PDB 3WT9). d–g Cryo TEM images of SaroL-1 clusters bound to Gb3-decorated LUVs. Red arrows indicate mushroom-like oligomers of SaroL-1 with estimated sizes (indicated by blue and red scale bars, respectively) corresponding to the heptameric predicted model (grey surface miniature). Scale bars are 50 nm in d and 20 nm in e–g.