A potential pathway for flippase-facilitated glucosylceramide catabolism in plants

ABSTRACT

The Aminophospholipid ATPase (ALA) family of plant lipid flippases is involved in the selective transport of lipids across membrane bilayers. Recently, we demonstrated that double mutants lacking both ALA4 and −5 are severely dwarfed. Dwarfism in ala4/5 mutants was accompanied by cellular elongation defects and various lipidomic perturbations, including a 1.4-fold increase in the accumulation of glucosylceramides (GlcCers) relative to total sphingolipid content. Here, we present a potential model for flippase-facilitated GlcCer catabolism in plants, where a combination of ALA flippases transport GlcCers to cytosolic membrane surfaces where they are degraded by Glucosylceramidases (GCDs). GCDs remove the glucose headgroup from GlcCers to produce a ceramide (Cer) backbone, which can be further degraded to sphingoid bases (Sphs, e.g, phytosphingosine) and fatty acids (FAs). In the absence of GlcCer-transporting flippases, GlcCers are proposed to accumulate on extracytoplasmic (i.e., apoplastic) or lumenal membrane surfaces. As GlcCers are potential precursors for Sph production, impaired GlcCer catabolism might also result in the decreased production of the secondary messenger Sph-1-phosphate (Sph-1-P, e.g., phytosphingosine-1-P), a regulator of cell turgor. Importantly, we postulate that either GlcCer accumulation or reduced Sph-1-P signaling might contribute to the growth reductions observed in ala4/5 mutants. Similar catabolic pathways have been proposed for humans and yeast, suggesting flippase-facilitated GlcCer catabolism is conserved across eukaryotes.

Text

P4-type ATPases (i.e., lipid flippases) are enzymes that consume ATP to facilitate the selective transport of lipids across membrane bilayers, specifically the movement from non-cytosolic surfaces to cytosolic surfaces. In plants, flippases are referred to as Aminophospholipid ATPases (ALAs), and require ALA-Interacting Subunits (ALISs) for proper localization and activity.^1,2 Arabidopsis (Arabidopsis thaliana) has 12 ALAs distributed among three of the eight clades comprising eukaryotic flippases, P4A-b, P4A-e, and P4 C,³ which can be further divided into five phylogenetic clusters conserved in angiosperms (Figure 1a).²⁷ Flippases from each of these clusters show both overlapping and distinct lipid transport activities.^5,28 For example, while flippases from all five ALA clusters appear to transport at least some phosphatidylserine (PS), only members from cluster 2, 3, or 4 appear to transport phosphatidylcholine (PC).^2,5,29-31

Figure 1.

Model showing a role for ALA flippases in GlcCer catabolism.

(a) Table listing Arabidopsis Aminophospholipid ATPases (ALAs) classified into phylogenetic clades and clusters, status of GlcCer transport testing, and relevant comments. Verified GlcCer-transporter ALA10⁴ is shaded green. ALA5 is the only other tested ALA⁵ and is shaded orange. P4A-e clade is shaded purple. (b) Structure of the most abundant GlcCer in Arabidopsis leaves, t18:1-Glc-hCer, in comparison to the fluorescent analogue available for testing, NBD-d18:1-Glc-Cer. Both t18 and α–hydroxylation events are highlighted in blue boxes. (c) Legend of enzyme and lipid symbols. (d) Biochemical pathway describing the flow of sphingolipids through a suite of modification enzymes, including Sphingoid Base Hydroxylases [SBHs;⁶], ceramide synthases [Lag One Homologs, i.e., LOHs;⁷], Sphingolipid-FA Hydroxylases [FAHs;⁸], IPC Synthases [IPCSs;⁹], IPC Glucuronosyl Transferase 1 [IPUT1;¹⁰], GIPC Mannosyl Transferase 1 [GMT1;¹¹], Glucosamine IPC Transferase 1 [GINT1;¹²], and GlcCer Synthase [GCS;¹³]. Enzymes involved in sphingolipid catabolism and downstream signaling are also included, such as Glucosylceramidases [GCDs;¹⁴], Alkaline and Neutral Ceramidases [ACERs and NCERs, i.e., A/NCERs;^15–18], and Sphingosine Kinases [SPHKs;^19–21]. Unidentified enzymes are indicated as “???”. Relative abundance of each GlcCer⁸ and GIPC species²² in Arabidopsis leaves is shown. ND is not detected. (e) Model for flippase-facilitated GlcCer catabolism in wild-type Arabidopsis cells, with ALA and ALA-Interacting Subunit (ALIS) complexes transporting GlcCers from lumenal and extracytoplasmic surfaces to cytosolic surfaces. After cytosolic deposition and potentially membrane-to-membrane substrate trafficking, GlcCers are catabolized by Endomembrane (Endo)/PM-localized GCDs and Endo-localized A/NCERs, producing Sphs and FAs. Phosphorylation of Sph to Sph-1-P is performed by SPHKs that localize to the cytosolic surface of the vacuole (Vac). Sph-1-P stimulates Ca²⁺ influx²³ and inhibits K⁺ influx,²⁴ which may be important for growth (contained in blue box). While GlcCers are present at the Vac,^25,26 they are not shown here for simplicity. However, it remains possible that GlcCer flipping and catabolism could occur directly at the Vac. (f) Putative model for GlcCer accumulation and reduced Sph-1-P signaling due to a flippase deficiency (ala4/5 mutant?), which might inhibit growth. Other ALAs remain in this cell but are not shown.

Distinct plant phenotypes have now been reported for lines deficient in each specific ALA cluster. For example, RNAi lines that target ALA1 (P4A-b; cluster 1) display vegetative growth reductions under chilling stress,²⁹ ala2 mutants (P4C; cluster 5) have increased sensitivity to viral pathogens,^32,33 ala3 mutants (P4A-e; cluster 4) have disrupted vesicular trafficking and reductions in vegetative growth and pollen fertility,^34–36 ala6/7 double mutants (P4A-e; cluster 3) have reduced pollen fertility,³⁷ and ala10 mutants (P4A-e; cluster 2) have perturbed stomatal conductance and membrane desaturation levels.^31,38 There is also evidence that an ala3/4/5/9/10/11 mutant (P4A-e; cluster 2, 3, and 4) is dwarfed and non-viable, suggesting members from the P4A-e clade together are essential to Arabidopsis.³⁹

Recently, we demonstrated that the loss of both ALA4 and −5 (ala4/5 mutant; P4A-e; cluster 3) is sufficient to cause severe plant dwarfism and reduced cell growth.⁵ These growth defects correlated with changes in the concentrations of various membrane lipids, including perturbations in glycosphingolipids that are thought to be critical for vegetative growth and development.^13,40 Specifically, ala4/5 mutants displayed 1.4-fold increases in glucosylceramides (GlcCers; Figure 1b) and corresponding 1.3-fold decreases in glycosylinositolphosphoceramides (GIPCs), relative to total sphingolipid content. While it is not yet clear why GlcCer levels are increased in ala4/5 mutants, or whether this specific sphingolipid accumulation (i.e., sphingolipidosis) is the underlying cause of ala4/5 mutant dwarfism, we speculated that ALA4 and −5 might flip GlcCers to facilitate their catabolism on cytosolic membrane surfaces. Importantly, this speculation does not rule out the potential for ALAs to transport GIPCs, or for ala4/5 mutants to be dwarfed due to GIPC reductions.

Here we offer a working model for GlcCer catabolism in plant cells that integrates a potential role for ALA flippases (Figure 1c–f). A key feature of this model is a need to flip GlcCers across a membrane to facilitate their degradation into glucose and a ceramide backbone (Cers). Once cytosolically exposed, glucose removal could be catalyzed by glucosylceramidases. In Arabidopsis, there are four genes encoding Glucosylceramidases (GCDs) that show homology to a well-characterized mammalian glucosylceramidase, named GBA2. For GCD3, the GlcCer-degrading activity has been confirmed and a GFP-fusion was shown to localize to the ER and PM,¹⁴ likely on cytosolic membrane surfaces similar to GBA2.⁴¹ After glucose removal, the Cer portion is then further degraded to a sphingoid base (Sph, e.g., phytosphingosine) and a fatty acid (FA) by Golgi/ER-localized Neutral Ceramidases (NCERs)^16,17 and two phylogenetically distinct Alkaline Ceramidases (ACERs), named ACER1¹⁸ and Turgor Regulation Defect 1 [TOD1;¹⁵]. Sphs can also be phosphorylated by tonoplast-localized Sphingosine Kinases (SPHKs) to produce Sph-1-phosphate (Sph-1-P, e.g., phytosphingosine-1-P),^19–21 a turgor-regulating secondary messenger that stimulates Ca²⁺ influx²³ and inhibits K⁺ influx.²⁴ We propose that a GlcCer-flippase deficient mutant might result in a GlcCer-catabolism defect that 1) increases GlcCer concentrations on lumenal and/or extracytoplasmic (i.e., apoplastic) membrane surfaces, and 2) reduces the production of GlcCer-catabolism byproducts, including signaling molecules such as Sphs and Sph-1-Ps. We speculate that either GlcCer accumulation or reduced Sph signaling could reduce plant growth and contribute to the underlying mechanism of dwarfism in ala4/5 mutants (Figure 1f).

An important feature of this model is the predicted spatial separation between GlcCers and the GlcCer-degrading GCDs, which require an ALA lipid flippase for convergence. Evidence suggests GlcCers accumulate on the apoplastic surface of the plant PM under normal growth conditions,^42,43 where they are protected from any cytosol-localized degrading enzymes. Other plant membranes known to accumulate GlcCers include the vacuole and other parts of the endomembrane system, ^25,44,45 but assays have not been performed to determine the leaflet-specific distribution of GlcCers in these membranes. As for GCDs, while GCD3-GFP fusions have been shown to localize to PM and endomembrane systems,¹⁴ these experiments did not address leaflet-specific localization. GCD1, −2, and −4 also remain uncharacterized for activity or localization. However, studies on the mammalian homolog GBA2 suggest that this class of enzymes localizes to cytosolic membrane surfaces,⁴¹ which provides an expectation that plant GCDs also have cytosolic orientations.

In the proposed model, it is not yet clear which ALAs are directly transporting GlcCers (Figure 1a). ALA10 from cluster 2 is the only plant flippase with demonstrated GlcCer transport activity, shown via the uptake of a fluorescent NBD-d18:1-GlcCer substrate in yeast.⁴ ALA1 from cluster 1 might also transport GlcCers as it has the highest sequence homology to ATP10D (human) and Dnf1/2 (yeast),³ which have been shown to transport GlcCers.⁴⁶ ALA3 might also flip GlcCers as they belong to the same hierarchal P4A flippase clade as Dnf1/2, ATP10D, and ALA10.³ Currently, most ALAs have not been tested for GlcCer transport activity (Figure 1a).

As to whether ALA4 or −5 from cluster 3 can transport GlcCers, ALA5 failed to show detectable uptake of an NBD-d18:1-GlcCer substrate in a yeast assay, despite showing uptake for at least one other NBD-labeled sphingolipid (sphingomyelin – a non-plant lipid).⁵ However, evidence suggests that some lipid flippase activities are not reliably detected by fluorescent lipid uptake assays. For example, ALA5 failed to show uptake of an externally supplied NBD-PS substrate when expressed in a yeast flippase mutant, but in an independent assay still showed the capacity to flip and remove endogenous PS from the extracytoplasmic surface of these same lines and confer resistance to the toxin papuamide A.⁵

The sphingolipid test substrate NBD-d18:1-GlcCer is also an imperfect proxy for plant sphingolipids, which have several important variations that could alter substrate recognition by ALA flippases (Figure 1b). For example, plant GlcCers are modified by hydroxylation events near the head group. One modification, α-hydroxylation of the FA chain by Sphingolipid-FA Hydroxylase 1 and −2 (FAHs) converts Cers to hydroxyCers (hCers), a modification that occurs in >98% of GlcCers in Arabidopsis leaves.⁸ Additionally, C4-hydroxylation of the Sph chain by Sphingoid Base Hydroxylase 1 and −2 (SBHs) converts d18 sphingolipids to t18, a modification that occurs in ~81% of GlcCers in Arabidopsis leaves.⁸ Interestingly, both fah1/2 mutants⁸ and sbh1/2 mutants⁶ display growth defects, suggesting that these modifications are important for plant growth. Thus, it remains possible that ALA4 and −5 could still facilitate the transport of t18-GlcCer or d/t18-Glc-hCer plant sphingolipids, despite the inability to recognize a related NBD-labeled test substrate.

Our working model that correlates ala4/5 mutant growth defects to reductions in GlcCer catabolism predicts several other mutations that might cause similar sphingolipidosis-related dwarfisms. For example, the model predicts that a mutant lacking all four GCDs would be completely deficient in GlcCer catabolism, and thus should accumulate GlcCers and be dwarfed like ala4/5 mutants. At present, only knockdown mutants in GCD3 have been characterized, which failed to show any obvious growth defects.¹⁴ The normal growth associated with gcd3 mutants might indicate that the four GCDs in Arabidopsis provide some degree of functional redundancy. Thus, a mutant combination that is completely lacking all four GCDs is needed to determine whether a disruption in this first stage of GlcCer catabolism can partially pheno-mimic an ala4/5-like growth deficiency.

The model also predicts that mutants lacking signaling components downstream of GCD activity might show dwarfism. However, functional redundancy might also be an issue for finding phenotypes associated with the loss of three NCERs and two ACERS. At present, no ala4/5-like growth phenotypes have been reported for single mutants of ncer1,¹⁶ ncer2, ⁴⁷ acer1,¹⁸ or tod1 (an ACER),¹⁵ or for an ncer1/2 double mutant.⁴⁷ Similarly, double mutants lacking both SPHK1 and −2 might show growth defects, due to a complete deficiency in direct Sph-1-P synthesis. However, at present only sphk1 and sphk2 single mutants have been characterized, both of which have normal vegetative growth.^48,49 Thus, additional genetic studies are needed to test the relative importance of each subsequent stage in GlcCer catabolism.

It is also worth considering the changes in sphingolipid metabolite concentrations predicted by this model. Besides GlcCer increases, a GlcCer-catabolism deficiency is expected to decrease the concentrations of hCer and Sph byproducts. While neither of these changes were observed in ala4/5 mutants, evidence suggests plants compensate for reduced sphingolipid production by reducing growth, as seen in RNAi lines deficient in the Sph biosynthesis-gene Long Chain Base 1 (LCB1), which are dwarfed but still have normal Sph/sphingolipid contents.²²

While the model illustrated here is for a plant cell, the potential importance of flippase-facilitated GlcCer catabolism is not restricted to plants. In humans, mutations in the flippase ATP10D have been correlated with elevated plasma GlcCer levels.⁵⁰ Recently, ATP10D was confirmed to transport GlcCers,⁴⁶ and a similar model for GlcCer catabolism in humans has been proposed where ATP10D flips GlcCers across membranes to facilitate their cytosolic degradation by GBA2. Additionally, Saccharomyces cerevisiae has two flippases, Dnf1p and −2p, that can uptake GlcCers,⁴⁶ and at least one cytosolic enzyme capable of glucosylceramidase activity, Egh1.⁵¹ As S. cerevisiae is one of several yeast species that lack GlcCers,^52,53 the continued presence of an enzyme for GlcCer uptake might indicate a functional role in the context of nutrition or perception of other organisms. However, other yeast species such as Cryptococcus neoformans still produce endogenous GlcCers, and their homologs of Dnf1p/2p⁵⁴ and Egh1⁵⁵ might still function in the context of a flippase-facilitated GlcCer homeostasis pathway. Thus, similar flippase-facilitated GlcCer-catabolism pathways appear to be conserved in plants, animals, and at least some fungi.

Funding Statement

Funding for this work was provided by the United States Department of Agriculture (HATCH grant no. NEV00384; to J.F.H.), the National Science Foundation (IOS grant no. 1656774; to J.F.H.), the Innovation Fund Denmark (LESSISMORE; to M.P.), the Carlsberg Foundation (RaisingQuinoa; project number CF18-1113; to M.P.), and the Novo Nordisk Foundation (NovoCrops; project number 2019OC53580; to M.P. and R.L.-M.).