P-Type ATPase Apt1 of the Fungal Pathogen Cryptococcus neoformans Is a Lipid Flippase of Broad Substrate Specificity

Lipid flippases of the P4-ATPase family are ATP-driven transporters that translocate lipids from the exoplasmic to the cytosolic leaflet of biological membranes. In the encapsulated fungal pathogen Cryptococcus neoformans, the P4-ATPase Apt1p is an important regulator of polysaccharide secretion and pathogenesis, but its biochemical characterization is lacking. Phylogenetic analysis revealed that Apt1p belongs to the subclade of P4A-ATPases characterized by the common requirement for a β-subunit. Using heterologous expression in S. cerevisiae, we demonstrate that Apt1p forms a heterodimeric complex with the C. neoformans Cdc50 protein. This association is required for both localization and activity of the transporter complex. Lipid flippase activity of the heterodimeric complex was assessed by complementation tests and uptake assays employing fluorescent lipids and revealed a broad substrate specificity, including several phospholipids, the alkylphospholipid miltefosine, and the glycolipids glucosyl- and galactosylceramide. Our results suggest that transbilayer lipid transport in C. neoformans is finely regulated to promote fungal virulence, which reinforces the potential of Apt1p as a target for antifungal drug development.


Introduction
Cryptococcus neoformans is an encapsulated fungal pathogen that causes severe infection of the central nervous system. The pathogenicity of C. neoformans is largely dependent on secretory mechanisms, which results in the transport of important virulence factors to the extracellular space, including fungal melanin, extracellular enzymes (e.g., laccase, urease, and phospholipase B1), and immunomodulatory polysaccharides [1]. Cryptococcal extracellular polysaccharides such as glucuronoxylomannan are considered important regulators of pathogenicity [2] and are required for capsule formation, which protects the fungus against host immune mechanisms (reviewed in reference [3]). Secretion of extracellular molecules is a highly regulated process relying on intense membrane remodeling and involving hundreds of different cellular proteins. Among these are members of the P4 subfamily of P-type ATPases (P4-ATPases), which are lipid flippases that catalyze the translocation of lipids from the exoplasmic to the cytosolic leaflet of cell membranes [4,5].
Plasmid Construction. Primers used are listed in Supplementary Table S1. Standard PCR reactions were performed with Q5 High-Fidelity DNA Polymerase (NEB, Ipswich, MA, USA). The wild-type APT1 coding sequence was originally amplified from H99 C. neoformans cDNA corresponding to GenBank entry CNAG_06469 using primer set #1 and introduced into the BamHI restriction site of the pESC-URA vector (Stratagene, La Jolla, CA, USA) using an InFusion Cloning Kit (Takara Clontech), yielding the plasmid pESC-URA-APT1. The plasmid pESC-URA-APT1-myc was generated with oligonucleotides encoding the myc tag and a GSGSGSG linker (primer set #2) using the pESC-URA-APT1 plasmid as a template and a QuickChange Site-Directed Mutagenesis Kit (New England Biolabs). To generate a vector bearing C-terminal green fluorescent protein (GFP)-tagged APT1 (pESC-URA-APT1-GFP), the PCR-amplified APT1 sequence (primer set #3, pESC-URA-APT1-myc as the plasmid template) and a PCR-amplified GFP fragment (primer set #4, pET-SUMO-LUN-GFP plasmid as the template) containing overlapping sequences were introduced into the BamHI restriction site of pESC-URA vector by InFusion cloning. The CDC50 sequence was amplified from synthetic cDNA (Accession number CNAG_06465) ordered from Integrated DNA Technologies (IDT, Coralville, IA, USA) using primer set #5, and introduced into the NotI site of the corresponding pESC-URA vectors, yielding pESC-URA-APT1-CDC50, pESC-URA-APT1-myc-CDC50, and pESC-URA-APT1-GFP-CDC50. A similar strategy was used to generate pESC-URA-APT1-myc-CDC50-Flag and pESC-URA-APT1-GFP-CDC50-Flag using primer set #6. The point mutant unable to hydrolyze ATP (apt E303Q ) was generated by QuickChange site-directed PCR (primer set #7) following the manufacturer's instructions (NEB, Ipswich, MA, USA). All constructs were verified by sequencing.
ATPase Assay. The purified protein was analyzed for ATPase activity using [γ-32 P]ATP as described by [44]. Briefly, eluates (39 µL, 0.25 µg protein) containing 0.05% (w/v) DDM, in the absence or presence of exogenous lipids (0.5 mg mL −1 ) and inhibitor (1 mM orthovanadate) were incubated for 20 min at room temperature for protein activation. Upon addition of 5 µL of an ATP mixture (10 mM ATP, 50 mM MgCl 2 , 20 µCi [γ-32 P]ATP), ATPase activity was assayed for 60 min at 28 • C. The reaction was stopped by placing the samples on ice and the addition of 1.5 mL of reagent A (10 mM ammonium heptamolybdate in 1 M HCl). The samples were transferred to glass tubes containing 15 µL H 3 PO 4 (20 mM), and 3 mL of reagent B (isobutanol, cyclohexane, acetone, and reagent A; ratio of 5:5:1:0.1 (v/v)) was added. After vortexing for 30 s, the samples were incubated for 10 min on ice to allow complete phase separation. Out of the upper phase, 500 µL was taken and added to 2 mL of scintillation fluid (Rotiszint eco plus for hydrophilic samples, Carl Roth, Karlsruhe, Germany) in a scintillation vial, and β-emission was counted for 1 min per sample (2450 MicroBeta2TM, Perkin Elmer, Groningen, The Netherlands). As a reference, 0.2 µL of the ATPase reaction mix in 2 mL scintillation fluid was measured.
NBD-lipid uptake assays, microscopy, and lipid analysis. Uptake experiments were performed at 30 • C for 30 min, essentially as described previously [43]. Flow cytometry was performed on a Becton Dickinson Flow Cytometer equipped with an argon laser using Cell Quest software. Data were analyzed using Cyflogic (CyFlo, Ltd., Turku, Finland), according to [45]. One microgram of propidium iodide in water was added to cells in 1 mL of PBS just before flow cytometry analysis. Data for at least 20,000 cells were collected prior to gating. NBD fluorescence of living cells was plotted on a histogram, and the geometric mean fluorescence intensity was used for further statistical analysis. Fluorescence microscopy and image acquisition were carried out on living cells immobilized on Concanavalin Acoated glass slides using a spectral confocal laser scanning microscope (Leica TCS SP5, Heidelberg, Germany). All images were acquired using a 63×/1.4 numerical-aperture (NA) oil objective. GFP was excited with a 488 nm argon laser, and emission signals were recorded between 500 and 550 nm, DAPI was excited at 355 nm, and emission signals werecollected between 410 and 470 nm. All fluorescent signals were quantified using ImageJ version 1.53c (National Institutes of Health, Madison, WI, USA; downloaded from https://fiji.sc/ (accessed on 15 July 2021)).
Data analysis. Data were analyzed using Excel (Microsoft) and OriginPro (Origin-Lab, Northampton, MA, USA). Data represent means ± s.d. of at least three experiments.
Phylogenetic analysis. Annotated P4-ATPase protein sequences were retrieved from the Uniprot Database (www.uniprot.org (accessed on 15 July 2021)), except for those corresponding to Tremella mesenterica, which were retrieved from NCBI (www.ncbi.nlm.nih.gov . Sequences were aligned using MUSCLE [46] in MEGA X [47]. The evolutionary history was inferred by using the Maximum Likelihood method and the JTT matrix-based model [48]. Initial tree(s) for the heuristic search were obtained automatically by applying Neighbor-Join and BioNJ algorithms to a matrix of pairwise distances estimated using a JTT model and then selecting the topology with superior log likelihood value.

C. neoformans Genome Encodes Four Potential P4-ATPases
Previous sequence analysis of the C. neoformans genome identified one ScCDC50 homolog and four P4-ATPase genes, namely, APT1, APT2, APT3, and APT4 [37]. Phylogenetic analysis revealed that Apt4p (CnJ9VM87) is a close homolog to Saccharomyces cerevisiae ScNeo1p and can be classified within the P4B-ATPase subfamily, comprising proteins that are functional in the absence of a Cdc50 protein (Figure 1) [49]. All other C. neoformans Apt proteins belong to the P4A subclade, characterized by the common requirement for a β-subunit but including P4-ATPases with diverse substrate specificities. Apt1p (CnJ9VZ19) and Apt2p (CnJ9VQH2) are closely related to Dnf1/2p from S. cerevisiae, which are broad-specificity transporters capable of recognizing both glycophospholipids and sugar-modified sphingolipids, while Apt3p (CnJ9VGP8) shows homology to PS-transporting ScDrs2p [50]. No homologue of S. cerevisiae Dnf3p, which belongs to a subgroup of P4A-ATPases present exclusively in fungi ( Figure 1) [49], is found in C. neoformans (Suppl. Figure S1). The phylogenetic tree was inferred from maximum likelihood analysis with 1000 bootstrap iterations performed using Mega X. Node values represent maximum likelihood statistical values with a maximum of 100%. All nodes with statistical values less than 80% were collapsed into multifurcations. In the phylogenetic tree, proteins from humans are written in red, those from plants in green, those from worm in blue, and those from yeast in gold letters. Proteins belonging to the same clade are shadowed: red, human-specific clade; green: plant-specific clade; gold: fungi-specific clade; light blue, clades containing proteins from organisms belonging to several life kingdoms. Shadowing and subclade division within the P4A clade (letters a-e) are as described in [49]. For accession numbers, see Materials and Methods.
Our further analysis revealed that Apt1p harbors all functional motifs needed for lipid flipping: a membrane (M) domain with 10 TM helices and a cytosolic region comprising the phosphorylation (P), nucleotide-binding (N), and actuator (A) domains (Figure 2A). Common motifs in the P domain present in Apt1p and found in all P4-ATPases [51] include the DKTGT motif, containing an aspartate that undergoes the phosphorylationdephosphorylation cycle, as well as the TGDx and GDGxND motifs that bind Mg 2+ and link the ATP-binding region to the transmembrane (TM) segments. Likewise, the DGET-like motif in the A domain enabling dephosphorylation of the aspartate residue in the DKTGT motif, thereby facilitating lipid flipping, and the PISL-like motif located in the TM4 helix (M domain) and harboring a proline residue involved in protein folding and lipid flipping events [52,53] are present. Determinants of substrate specificity in TM1, 2, 4, and 6 [53][54][55] are partially conserved among the nearest S. cerevisiae orthologs ScDnf1/2p and ScDrs2p ( Figure 2B). Out of two glutamine (QQ) residues located in the TM1 entry gate position that is needed for the selection of PS in HsATP8A1 [16] and ScDrs2 [54], only the first one (Q111) is preserved in Apt1p (Figures 2B and S2). Apt1p also harbors residues characteristic for PC/PE/ceramide transporters [16], such as a non-polar residue at position 112 (F112) and a polar uncharged residue at position 476 (Q476). Taken together, we can envisage Apt1p as being a functional lipid flippase.

Heterologous Expression and Localization of Apt1 and Cdc50 Proteins in S. cerevisiae
To study the biochemical function of Apt1p, we generated a series of expression plasmids encoding untagged and tagged versions of Apt1p alone or in combination with its potential subunit Cdc50p under the control of a galactose-inducible bidirectional promoter ( Figure 3A). In addition, a plasmid encoding a non-functional APT1 mutant was generated. The corresponding mutant protein lacks the glutamate in a conserved DGET motif (residues 301-304), which is the key amino acid required for dephosphorylation of the P4-ATPase during the catalytic cycle and is consequently blocked in the turnover of the phosphorylated protein. To study the functionality of untagged and tagged proteins, the resulting constructs or the empty vector was used to transform a S. cerevisiae mutant (ZHY709). This mutant lacks the endogenous plasma membrane P4-ATPases ScDnf1p and ScDnf2p, as well as the Golgi-localized P4-ATPase ScDrs2p, resulting in defective lipid transport and an aberrant exposure of endogenous aminophospholipids at the plasma membrane [14,21,56,57]. Consequently, this mutant is hypersensitive to the membrane impermeable cytotoxic peptide duramycin that binds the aminophospho-lipid PE. In the presence of this compound, galactose-induced co-expression of APT1 and CDC50 (either untagged or tagged) complemented the ZHY709 mutant, whereas cells expressing Apt1p alone or catalytically inactive Apt1p E303Q were not viable ( Figure 3C). Miltefosine, edelfosine, and perifosine are alkyl phosphocholine analogues that are toxic only after uptake across the plasma membrane ( Figure 3B). Wild-type yeast is sensitive to all three alkyl phosphocholine analogues, whereas ZHY709 is not, due to the lack of the phosphocholine-transporting P4-ATPases ScDnf1/2p at the plasma membrane ( Figure 3B). Upon galactose-induced co-expression of APT1 and CDC50, mutant cells became sensitized to miltefosine, whereas cells bearing Apt1p alone or catalytically inactive Apt1p E303Q remained tolerant ( Figure 3C). By contrast, edelfosine and perifosine impaired the growth of mutant cells upon expression of APT1 variants, and co-expression of CDC50 did not exert a notable effect. Taken together, these results demonstrate that co-expression of both tagged and untagged versions of Apt1p and Cdc50p partially complemented the phenotypes of the P4-ATPase-deficient S. cerevisiae strain ZHY709. Yeast strain BY4741 (wild type) and ZHY709 transformed with empty plasmids (E.V.) were used as a positive and a negative control, respectively. Yeast strains were grown at 30 • C on plates containing a concentration gradient of the indicated toxins (the direction of the gradient is indicated by a triangle) or on control plates with glucose (SD, control) or galactose (SG, induced gene expression) with no further additions. The experiment was repeated at least three times with identical results.
Taking advantage of the engineered GFP tags, we studied the subcellular localization of Apt1p and catalytically inactive Apt1p E303Q by confocal fluorescence microscopy ( Figure 4). Apt1p-GFP alone displayed a staining pattern characteristic of the yeast ER, with intense labeling of the perinuclear ER ( Figure 4A). Upon co-expression with Cdc50p, however, Apt1p-GFP moved from the ER into a more peripheral localization. Apt1p E303Q -GFP co-expressed with Cdc50 displayed a localization similar to that observed for wild-type Apt1p-GFP with Cdc50p. Western blot analysis confirmed the expression of both full-length proteins ( Figure 4B).

Apt1 and Cdc50 Proteins Form a Stable Complex Showing ATPase Activity
To investigate the ability of Apt1p and Cdc50p to form a stable complex, detergentsolubilized membrane preparations obtained from ZHY709 cells expressing APT1-GFP and CDC50-FLAG were subjected to anti-FLAG affinity chromatography, and bound proteins were subsequently eluted by treatment with a buffer containing the FLAG peptide ( Figure 5A). This procedure efficiently co-purified Apt1p and Cdc50p ( Figures 5B,C and S3). Analysis of the Apt1p-Cdc50p complex by SDS-PAGE followed by Coomassie blue stain-ing revealed a band of the expected molecular mass for Apt1p-GFP (~203 kDa), a band at the expected molecular mass for glycosylated Cdc50p-FLAG (~60 kDa), and minor contaminating proteins. Deglycosylation experiments using PNGase F resulted in a single small-molecular-size band (~50 kDa), confirming the glycosylation of Cdc50p-FLAG ( Figure 5D). The purified, detergent-solubilized complex displayed maximal ATP hydrolysis in a PC/PS mixture (molar ratio, 90:10), with an apparent ATPase activity of 177 ± 31 nmol ATP min −1 mg protein −1 (n = 3) ( Figure 5E). Sodium orthovanadate inhibited 65% of this activity, indicating that Apt1p-Cdc50p was the dominant active ATPase in the eluate. The protein complex was also active in the presence of PC alone (63% relative to PS/PC), while the absence of added lipids resulted in a basal activity (22% relative to PS/PC). We conclude that Apt1p and Cdc50p form a stable membrane complex showing lipid-stimulated ATPase activity. and 1 mM orthovanadate (OV). Results are the means ± S.D. from duplicate determinations (protein with PC/PS/OV, n = 3; protein with PC, n = 2; protein without lipid, n = 1). Abbreviations: S1-3, supernatants; P1-3, pellets; FT, flow-through; W1-3, washes; M, Flag Matrix; E1-3, pooled eluates; C, concentrated eluate. PC, phosphatidylcholine; PS, phosphatidylserine.

Apt1p in Complex with Cdc50p Is a Broad-Specificity Lipid Flippase
We next determined whether co-expression of Apt1p with Cdc50p complemented the lipid internalization defect of the P4-ATPase-deficient S. cerevisiae strain ZHY709 by employing fluorescent nitrobenzoxadiazole (NBD) acyl-labelled lipid reporters ( Figure 6). To avoid overlapping fluorescent emission signals from NBD and GFP, cells expressing untagged Apt1p were employed. Cells were incubated with different NBD-labelled lipids and, following the removal of the surface-exposed probe, the remaining fluorescence was detected by flow cytometry ( Figure 6A). For the quantitative assessment of NBDlipid internalization, fluorescence intensities were expressed as percentages relative to control cells harboring empty vectors. Co-expression of Apt1p with Cdc50p resulted in a population of cells with increased internalization of a broad range of lipids ( Figure 6A). Analysis of the data indicated that the P4-ATPase complex preferentially transported PC, GlcCer > PE > PS, GalCer > SM > PG, but not the diosylceramide lactosylceramide (LacCer) ( Figure 6B). Furthermore, a non-functional version of the transporter (Apt1p E303Q ), in combination with Cdc50p, was unable to support the uptake of fluorescent lipids, confirming that P4-ATPase activity is a requirement for lipid internalization.

Discussion
Previous studies revealed overlapping functions of both the P4-ATPase Apt1p and the Cdc50 protein in the maintenance of membrane integrity, implying that Cdc50p serves as a β-subunit for Apt1p to form a functional lipid flippase [36,37,39,58]. By using heterologous expression in S. cerevisiae, the present study provides direct evidence that both proteins form a stable heterodimeric complex resulting in a functional lipid flippase. These findings are consistent with Cdc50 proteins being an integral part of the P4A-ATPase lipid flippases. In S. cerevisiae, the Cdc50 family members ScCdc50p, ScLem3p, and ScCrf1p form stable complexes with ScDrs2p, ScDnf1p/Dnf2p, and ScDnf3p, respectively [59,60]. Formation of these complexes is required for proper expression and ER export of either partner [59][60][61].
In agreement with this notion, we observed ER retention of Apt1p in the absence of Cdc50p expression. In all eukaryotic genomes examined so far, P4-ATPase genes greatly outnumber CDC50 genes, suggesting that a given Cdc50 protein can interact with multiple P4-ATPases. This notion is supported by studies on fungal, human, and plant P4-ATPases. In S. cerevisiae, ScLem3p does associate with both P4A-ATPases ScDn1p and ScDnf2p. Likewise, the human P4A-ATPases HsATP8B1, HsATP8B2, HsATP8B3, HsATP8A1, and HsATP8A2 each interact with either HsCdc50A or HsCdc50B [13,62], and various Arabidopsis Cdc50p homologs form stable complexes with several plant P4A-ATPases [14,63,64]. Conceivably, C. neoformans Cdc50p may also serve as a binding partner for the P4A-ATPases Apt2p and Apt3p. Apt4p, as a P4B-ATPase, might not require a β-subunit to function, as known for the P4B-ATPase ScNeo1p [35,61] and HsATP9B [65]. The heterologous expression system utilizing S. cerevisiae mutants presented here may help clarify this issue.
Complementation tests and uptake assays employing fluorescent lipids served as an approach to analyze lipid transport activity of the Apt1p-Cdc50p complex. We found that the Apt1p-Cdc50p complex facilitates the uptake of a broad lipid spectrum, including several phospholipids and monoglycosylceramides. The physiological relevance of this broad substrate specificity remains to be established but is reminiscent of the broad substrate specificity of the two plasma membrane P4A-ATPases Dnf1/2p in S. cerevisiae [29,30]. In the heterologous expression model used here, co-expression of Apt1p and Cdc50p reversed an aberrant exposure of endogenous aminophospholipids, as evidenced by reduced hypersensitivity of the cells to the PE-binding peptide duramycin. These results are in line with a direct role of the Apt1p-Cdc50p complex in pumping natural PE to the cytosolic leaflet to generate an asymmetric membrane, as recently suggested for P4-ATPases from other organisms [5,66]. The ability of the Apt1p-Cdc50p complex to transport NBD-PC and NBD-PS suggests that natural PC and PS are maintained in the cytoplasmic leaflet and that the outer leaflet is primarily composed of higher sphingolipids. In line with this notion, C. neoformans mutant cells lacking CDC50 expose PS on the cell surface [40]. The dynamic regulation of transbilayer lipid asymmetry by P4-ATPases has been found to be crucial for actin organization and membrane transporter function [25]. Thus, several phenotypes observed for C. neoformans mutant cells lacking CDC50 or APT1 might be caused by a disturbed lipid membrane asymmetry. Further studies are required to identify the subcellular localization of Apt1-4p in C. neoformans to dissect their individual cellular roles.
Notably, Apt1p in complex with Cdc50p was found to transport the monoglycosylceramides GlcCer and GalCer. GlcCer is an important regulator of cryptococcal virulence [67]. This neutral glycosphingolipid is found at the fungal cell wall [68] and in extracellular vesicles [69]. An alteration in the membrane lipid structure may result in an altered raft formation, thereby affecting fungal membrane fluidity and rigidity [70]. Thus, Apt1p might be needed for a tight regulation of the GlcCer contents of the cellular membranes and cell wall. In line with this concept, deletion of APT1 affects intracellular membrane organization and results in altered lipid metabolism, with reduced levels of GlcCer and inositol phosphoryl ceramides in C. neoformans [38].
Apt1p might also serve to scavenge GlcCer from the host. In the yeast S. pombe, the SpDnf2p ortholog transports GlcCer and GalCer but not PC and PE, suggesting that glycosphingolipid transport is a consolidated function of SpDnf2p [30]. S. pombe is unable to synthesize GlcCer, and it has been hypothesized that P4-ATPase-mediated transport of GlcCer in this organism serves a role in scavenging this lipid from the plant material the yeast grows upon [30,71]. The interconnections between Apt1p and GlcCer flipping are intriguing, since both molecules are required for both pathogenic and secretory mechanisms. This implies that lipid transport in C. neoformans is finely regulated to promote fungal virulence, which reinforces the potential of both Apt1p and GlcCer as targets for antifungal drug development [68,72].
In conclusion, our work identifies Apt1p in complex with Cdc50p as a lipid flippase with broad substrate specificity. This information represents a first step towards the understanding of the essential role of this transporter during infection and towards its validation as a drug target. The availability of a heterologous expression system for Apt1p-Cdc50p may prove useful for the expression and purification of the complex for further detailed biochemical analysis aimed at identifying novel antagonists of this transporter class that might help in the treatment of cryptococcal meningitis and related infections in humans.