A sphingolipid rheostat controls apoptosis versus apical cell extrusion as alternative tumour-suppressive mechanisms

Abstract

Evasion of cell death is a hallmark of cancer, and consequently the induction of cell death is a common strategy in cancer treatment. However, the molecular mechanisms regulating different types of cell death are poorly understood. We have formerly shown that in the epidermis of hypomorphic zebrafish hai1a mutant embryos, pre-neoplastic transformations of keratinocytes caused by unrestrained activity of the type II transmembrane serine protease Matriptase-1 heal spontaneously. This healing is driven by Matriptase-dependent increased sphingosine kinase (SphK) activity and sphingosine-1-phosphate (S1P)-mediated keratinocyte loss via apical cell extrusion. In contrast, amorphic hai1a^fr26 mutants with even higher Matriptase-1 and SphK activity die within a few days. Here we show that this lethality is not due to epidermal carcinogenesis, but to aberrant tp53-independent apoptosis of keratinocytes caused by increased levels of pro-apoptotic C₁₆ ceramides, sphingolipid counterparts to S1P within the sphingolipid rheostat, which severely compromises the epidermal barrier. Mathematical modelling of sphingolipid rheostat homeostasis, combined with in vivo manipulations of components of the rheostat or the ceramide de novo synthesis pathway, indicate that this unexpected overproduction of ceramides is caused by a negative feedback loop sensing ceramide levels and controlling ceramide replenishment via de novo synthesis. Therefore, despite their initial decrease due to increased conversion to S1P, ceramides eventually reach cell death-inducing levels, making transformed pre-neoplastic keratinocytes die even before they are extruded, thereby abrogating the normally barrier-preserving mode of apical live cell extrusion. Our results offer an in vivo perspective of the dynamics of sphingolipid homeostasis and its relevance for epithelial cell survival versus cell death, linking apical cell extrusion and apoptosis. Implications for human carcinomas and their treatments are discussed.

Fig. 1. Comparison of proliferation, apical cell extrusion, and apoptosis phenotypes in the hai1a^hi2217 hypomorphic and the hai1a^fr26 amorphic alleles.

a Comparison of proliferation rates in the hai1a^hi2217 hypomorphic and the hai1a^fr26 amorphic alleles at 2 and 4 dpf, relative to the total number of cells. b Quantification of extruded epidermal cells in the hai1a^hi2217 hypomorphic and the hai1a^fr26 amorphic alleles at 2, 3 and 4 dpf, collected from the embryo growth medium. White bars show the numbers of live cells, blue bars show the number of dead cells, with the proportion of dead cells in each condition indicated by blue text. c, d Representative images of proliferation assay in sibling (c) or the hai1a^fr26 mutant (d) epidermis at 4 dpf in the caudal fin fold with orthogonal views (c’, d’), showing BrdU-positive cells (red) within the peridermal cell layer (green, yellow arrowhead), or the basal cell layer (white, yellow arrow). e, f Proliferation assay in hai1a^fr26 mutant caudal fin fold epidermis without (e) and with (f) knockdown of ∆Np63 by antisense morpholino oligonucleotide (MO) to eliminate basal keratinocytes, with orthogonal views in the lower panels (e’, f’) showing BrdU-positive peridermal cells (yellow arrowheads). BrdU, red; p63, green; nuclei, blue. Scale bar = 50 μm. g–j Representative images of cleaved caspase 3 (aCasp3)-positive cells on the ventral median fin fold at 2 dpf (g, h) and 4 dpf (i, j), sibling (g, i) or hai1a^fr26 (h, j) mutant epidermis. aCasp3, red; p63, green; nuclei, blue. Scale bar = 50 μm. k, l Orthogonal views of the epidermis of the 4 dpf mutant fish shown in (j) displaying apoptotic cells protruding above the epidermis (k), and within the basal layer (l). aCasp3, red; p63, green; nuclei, blue. m Representative image of aCasp3-positive cells in the hai1a^fr26 mutant at 4 dpf in the caudal fin fold with orthogonal view (m’), showing extruding live peridermal cells (green, yellow arrowhead) or extruding aCasp3-positive (red) peridermal cells (green, yellow arrows). Basal cell layer, white. Scale bar = 50 μm. n Quantification of the number of aCasp3-positive cells in the tail fin of sibling control, hai1a^hi2217, and hai1a^fr26 mutant embryos at 1, 2, 3, and 4 dpf, normalised to fin area. o Distributions of apoptotic cells within the epidermis in siblings and hai1a^fr26 mutants at 4 dpf. For all graphs, means within each time point were compared using a one-way ANOVA with post-hoc Tukey’s multiple comparison test. N = number of biological replicates, n = number of fish per condition.

Fig. 2. Cell death in hai1a^fr26 mutants is caspase-dependent and tp53-independent.

a–h Apoptotic cells in the caudal fin fold of 4 dpf embryos, immunostained for aCasp3 (red), p63 (green) and DAPI (blue), with tp53 mutation or upon treatment with ZDEVD-FMK to inhibit apoptosis. Scale bar = 50 μm. i Quantification of the numbers of aCasp3-positive cells in the tail fins of embryos, normalised to fin area. Means were compared using one-way ANOVA with post-hoc Tukey’s multiple comparison test. j Survival curves showing the effects of tp53 mutation or ZDEVD-FMK treatment on embryo viability.

Fig. 3. Sphingosine kinase activity modulates cell death in hai1a^fr26 mutant epidermis.

a–d S1P levels are elevated in hai1a^fr26 mutants at 4 dpf. a–c Whole mount immunostaining in the caudal fin fold, anti-S1P (white), nuclei are labelled with DAPI (blue). c shows an individual extruding cell, with the orthogonal view shown in c’. d Quantification of S1P fluorescence in the caudal fins of embryos, normalised to fin area. For full dataset including SphK inhibition, see Fig. S3g. Scale bars: (a, b) 200 µm, (c) 20 µm. e Quantification of the numbers of extruded cells collected per fish at 4 dpf upon inhibition of sphingosine kinase by MPA08 treatment. White bars show the numbers of live cells, blue bars show the numbers of dead cells, with the proportion of dead cells indicated with blue text. f Survival curves showing the effect of sphingosine kinase inhibition on embryo viability. g–j Apoptotic cells in the caudal fin fold of 4 dpf hai1a^hi2217 mutant embryos, with and without MPA08 treatment. Apoptotic cells are labelled with aCasp3 (red), basal keratinocytes with p63 (green), and nuclei using DAPI (blue). Scale bar = 50 μm. l–o Apoptotic cells in the caudal fin fold of 4 dpf hai1a^fr26 mutant embryos, with and without MPA08 treatment. Apoptotic cells are labelled with aCasp3 (red), basal keratinocytes with p63 (green), and nuclei using DAPI (blue). Scale bar = 50 μm. q, r Apoptotic cells in the caudal fin fold of 4 dpf hai1a^fr26 mutant embryos upon S1P treatment. Apoptotic cells are labelled with aCasp3 (red), basal keratinocytes with p63 (green), and nuclei using DAPI (blue). Scale bar = 50 μm. k, p, s Quantification of numbers of aCasp3-positive cells in the tail fins of embryos, normalised to fin area. t–v Apoptotic cells in the caudal fin fold of 4 dpf wild-type embryos, treated with ceramide to induce apoptosis (u), or concomitant treatment with ceramide and S1P (v). Apoptotic cells are labelled with aCasp3 (red), basal keratinocytes with p63 (green), and nuclei using DAPI (blue). Scale bar = 50 μm. w Quantification of numbers of aCasp3-positive cells in the tail fins of embryos, normalised to fin area. For (d), means of controls and mutants were compared using an unpaired, two-tailed Student’s t-test. For all other graphs, means were compared using one-way ANOVA with post-hoc Tukey’s multiple comparison test. N = number of biological replicates, n = number of fish per condition.

Fig. 4. Sphingolipid levels in the hai1a^fr26 mutant epidermis agree with mathematical modelling of the rheostat including a negative feedback loop controlling de novo ceramide synthesis.

a Schematic of the sphingolipid rheostat showing the three principal lipids, ceramide, sphingosine, and S1P, and the enzymes responsible for their inter-conversion and de novo synthesis. b–d Lipidomic analysis of sphingolipids of the rheostat in hai1a^fr26 amorph mutants and sibling controls. b Selected ceramide species (Cer_{16_0} and Cer_{16_1}) at 2 dpf, 4 dpf, and 4 dpf upon MO-mediated knockdown of cers5 and cers6. c Sphinganine levels at 2 dpf and 4 dpf. d Sphingosine levels at 2 dpf and 4 dpf. Means of controls and mutants within each condition were compared using an unpaired, two-tailed Student’s t-test. e–g Lipidomic analysis of sphingolipids of the rheostat in hai1a^hi2217 hypomorph mutants and sibling controls, at 2 dpf and 4 dpf. Means of controls and mutants within each condition were compared using an unpaired, two-tailed Student’s t-test. h Simplified scheme of the sphingolipid rheostat model for MCA. Ceramide (${\mathrm{x}}_1$), sphingosine (${\mathrm{x}}_2$) and S1P (${\mathrm{x}}_3$) are considered as variables, reactions ${\mathrm{v}}_1-{\mathrm{v}}_3$ are reversible, S1P degradation (${\mathrm{v}}_4$) is considered irreversible. The system has an effective downstream flux (from ${\mathrm{x}}_1$ to ${\mathrm{x}}_3$) in the steady state. Ceramide synthesis (${\mathrm{k}}_0$) includes de novo synthesis and synthesis via sphingomyelin conversion and is modelled as a boundary condition, i.e. a constant synthesis rate of order zero is assumed. i Schematic of the ODE modelling, including sphinganine. A Boolean operator ${\mathrm{S}}_{\mathrm{on}}$ is active ${(\mathrm{S}}_{\mathrm{on}}=1)$ in wild type and inactive ${(\mathrm{S}}_{\mathrm{on}}=0)$ in the mutant, and S1P synthesis by k₃ is described by ${\mathrm{k}}_{3,\mathrm{app}}={\mathrm{k}}_3\cdot \left(1-{{\mathrm{S}}_{\mathrm{on}}\cdot \mathrm{\alpha }}_{\mathrm{hai}1\mathrm{a}}\right),0.5\le {\mathrm{\alpha }}_{\mathrm{hai}1\mathrm{a}}<1$ so that it is at least 50% lower in wild type than in the mutant. A possible negative feedback loop is shown, which is turned off when the ceramide concentration decreases below a certain threshold, and implemented via a scaling factor such that k₀₀ can be described by ${\mathrm{k}}_{00,\mathrm{app}}={\mathrm{k}}_{00}\cdot {(1+{\mathrm{S}}_{\mathrm{on}}\cdot \mathrm{\alpha }}_{\mathrm{cer}}),-1<{\mathrm{\alpha }}_{\mathrm{cer}}<0$. j ODE model-predicted sphinganine and ceramide levels (horizontal lines) at 4 dpf compared to experimental lipidomics data (vertical bars) without the implementation of a negative feedback loop (${\mathrm{\alpha }}_{\mathrm{cer}}=0$). The 2 dpf data were set as the initial condition and are therefore perfectly met by the simulation results. k ODE model-predicted sphinganine and ceramide levels (horizontal lines) at 4 dpf compared to experimental lipidomics data (vertical bars) with the implementation of a negative feedback loop at the level of k₀₀ (${\mathrm{\alpha }}_{\mathrm{cer}}=-1$).

Fig. 5. Manipulation of the sphingolipid rheostat enzymes or ceramide de novo synthesis alters cell death in the hai1a^fr26 mutant.

a Schematic of sphingolipid rheostat showing inhibition of ceramidases with ceranib-2, inhibition of sphingosine kinases with MPA08 or SKI-II, inhibition of S1P phosphatase with a MO targeting sgpp1, inhibition of serine palmitoyltransferase with myriocin or SPT-IN-1 and sphingomyelinases with desipramine, and MOs targeting cers5 and cers6 at the level of de novo synthesis and salvage from sphingosine. b Quantification of aCasp3-positive cells in the tail fins of 4 dpf embryos upon DMSO or ceranib-2 treatment, normalised to fin area. c Quantification of aCasp3-positive cells in the tail fins of control or sgpp1 MO-injected 4 dpf embryos, normalised to fin area. d Quantification of aCasp3-positive cells in the tail fins of 4 dpf embryos upon DMSO, myriocin, or desipramine treatment, normalised to fin area. e Quantification of aCasp3-positive cells in the tail fins of control or cers5 alone, cers6 alone, or both cers5 and cers6 MO-injected 4 dpf embryos, normalised to fin area. For all graphs, means were compared using one-way ANOVA with post-hoc Tukey’s multiple comparison test. N = number of biological replicates, n = number of fish per condition.

Fig. 6. Early inhibition of ceramide synthesis in wild-type embryos recapitulates ACE and apoptosis phenotype.

a Loss of Matriptase inhibition leads to parallel pro-oncogenic and tumour-suppressive pathways. Sphingolipid rheostat-mediated tumour-suppressive methods include S1P-dependent apical cell extrusion, which preserves epithelial integrity, and C₁₆ ceramide-dependent apoptotic cell death, which may lead to loss of epithelial barrier function. When levels of C₁₆ ceramide are above a certain threshold, a negative feedback loop prevents further de novo ceramide synthesis. In hai1a^fr26 mutants by 2 dpf, ceramide levels drop due to sustained SphK activity catalysing S1P production, repression of de novo synthesis is lost, and resultant C₁₆ ceramide levels trigger apoptosis. b-c Epidermal aggregates and apoptotic peridermal cells in the caudal fin fold of 4 dpf WT fish, transiently treated with inhibitor of de novo ceramide synthesis myriocin at 2 dpf, with orthogonal views (b’, c’) showing extruding live (yellow arrowhead) or apoptotic (yellow arrow) peridermal cells. Apoptotic cells are labelled with aCasp3 (red), peridermal cells with krt4:GFP (green), basal keratinocytes with p63 (white), and nuclei using DAPI (blue). Scale bar = 50 μm. d Quantification of the numbers of extruded cells collected per fish at 4 dpf upon inhibition of de novo sphingolipid synthesis by myriocin treatment at 2 dpf. White bars show the numbers of live cells, blue bars show the numbers of dead cells, with the proportion of dead cells indicated with blue text. e Quantification of aCasp3-positive cells in the tail fins of 4 dpf embryos upon myriocin treatment, normalised to fin area. f Quantification of proliferating cells in the tail fins of 4 dpf embryos upon myriocin treatment, normalised to fin area. For all graphs, means were compared via two-tailed, unpaired Student’s t-test. N = number of biological replicates, n = number of fish per condition.

Fig. 7. Embryo death is caused by apoptosis-induced loss of epidermal barrier.

a–d Scanning electron microscopy images of the epidermal surface of sibling control (a), hai1a^hi2217 hypomorphs (b), and hai1a^fr26 amorph mutants (c, d). e–g Cartoons of the effects of hai1a mutation on epidermal morphology (basal keratinocytes in red, peridermal cells in green, underlying basement membrane in grey). h, i Biotin epidermal barrier assay on transverse cryosections in 4 dpf sibling control (h) and hai1a^hi2217 (i) mutants. Basal keratinocytes are labelled with p63 (green), nuclei using DAPI (blue), and biotin using streptavidin-Cy3 (SA-Cy3, red). Scale bar = 50 μm. j, k Biotin epidermal barrier assay on transverse cryosections in 4 dpf sibling control (j) and hai1a^fr26 (k) mutants. Basal keratinocytes are labelled with p63 (green), nuclei using DAPI (blue), and biotin using streptavidin-Cy3 (SA-Cy3, red). Scale bar = 50 μm. l–o Biotin epidermal barrier assay on transverse cryosections in 4 dpf sibling control (l, n) and hai1a^fr26 (m, o) mutants, treated with caspase inhibitor ZDEVD-FMK (l, m) or injected with cers5 and cers6 MOs (n, o). Basal keratinocytes are labelled with p63 (green), nuclei using DAPI (blue), and biotin using streptavidin-Cy3 (SA-Cy3, red). Scale bar = 50 μm. p–r Biotin epidermal barrier assay upon peridermal cell ablation, without (p) or with (q) the prodrug metronidazole (MTZ), and using TUNEL to label dying cells (r). Nuclei are labelled using DAPI (blue), biotin using streptavidin-Cy3 (SA-Cy3, red), and dying cells using TUNEL-fluorescein (green). Scale bar = 50 μm. s Survival curves of control (DMSO) or MTZ-treated fish at different concentrations, with treatments starting from 4 dpf.