Endocytic Sorting and Recycling Require Membrane Phosphatidylserine Asymmetry Maintained by TAT-1/CHAT-1

Endocytic sorting is achieved through the formation of morphologically and functionally distinct sub-domains within early endosomes. Cargoes destined for recycling are sorted to and transported through newly-formed tubular membranes, but the processes that regulate membrane tubulation are poorly understood. Here, we identified a novel Caenorhabditis elegans Cdc50 family protein, CHAT-1, which acts as the chaperone of the TAT-1 P4-ATPase to regulate membrane phosphatidylserine (PS) asymmetry and endocytic transport. In chat-1 and tat-1 mutants, the endocytic sorting process is disrupted, leading to defects in both cargo recycling and degradation. TAT-1 and CHAT-1 colocalize to the tubular domain of the early endosome, the tubular endocytic recycling compartment (ERC), and the recycling endosome where PS is enriched on the cytosolic surface. Loss of tat-1 and chat-1 function disrupts membrane PS asymmetry and abrogates the tubular membrane structure. Our data suggest that CHAT-1 and TAT-1 maintain membrane phosphatidylserine asymmetry, thus promoting membrane tubulation and regulating endocytic sorting and recycling.


Introduction
In eukaryotic cells, internalized cargoes are transported to early endosomes where they are sorted to be recycled back to the plasma membrane, degraded in lysosomes or delivered to the trans-Golgi network.Early endosomes display a complex and pleiomorphic organization with many tubular processes emanating from central vesicular elements as revealed by three-dimensional reconstruction [1][2][3][4].Internalized receptors dissociate from their ligands in early endosomes which have a slightly acidic internal pH; subsequent segregation of the receptor and ligand is thought to be achieved by a geometry-based mechanism [5].Receptors and other membrane proteins concentrate in the tubular extensions which contain most of the endosomal membrane, whereas soluble contents are enriched in the vesicular components which account for the bulk of the endosomal volume [1,6,7].The recycling vesicles which arise from the tubular extensions may undergo fast recycling, by fusing directly with plasma membranes, or slow recycling, by transporting cargoes through the endocytic recycling compartment (ERC), a collection of tubular membrane structures arranged around the microtubule-organizing center [7,8].Both cargo sorting and subsequent recycling require extensive membrane remodeling to form tubular extensions, which have a high ratio of surface area to luminal volume, thereby effectively concentrating cargoes on recycling membranes.However, it is not clear at present how these tubular processes are formed and maintained.
Both proteins and lipids are required for shaping membranes into various structures including tubular extensions.For example, BAR (Bin/amphiphysin/Rvs) domain proteins, which are central regulators of membrane remodeling, are capable of inducing membrane tubulation [9].Members of the EHDs/RME-1 family of ATPases, which are important regulators of endocytic recycling in mammals (EHD1-4) and C. elegans (RME-1), associate with vesicular and tubular membranes in vivo and tubulate liposomes in vitro [10][11][12][13][14]. On the other hand, phospholipids regulate membrane shaping by either recruiting and activating effector proteins on target membranes or directly affecting membrane curvature.For instance, membrane-shaping proteins like BAR proteins and dynamin are targeted to specific membrane compartments by binding to different phosphoinositides through either a lipid-binding domain (PH or PX) or by electrostatic interaction or both, while phospholipid-binding and membrane deformation by EHDs/RME-1 family proteins appear to be mediated through their helical domains [15,16].In addition to acting through a protein-recruiting mechanism, phospholipids can directly affect membrane curvature.It has been observed that addition of phosphatidylserine (PS) to ATP-containing erythrocyte ghosts stimulates the formation of endocytic vesicles [17].Notably, phosphatidylserine is asymmetrically arranged between the two membrane leaflets, being enriched in the inner leaflet of cell membranes [18].As the most abundant anionic phospholipid of cell membranes, PS regulates surface charge and protein targeting in cultured cells, where it is also observed on the cytosolic surface of endosomes and lysosomes [19].However, it remains to be determined whether PS, or PS asymmetry, is involved in shaping membranes into tubular processes during sorting and recycling.
Previous studies suggest that establishment and maintenance of PS asymmetry require the activity of type IV P-type ATPase family proteins (P4-ATPases), which selectively sequester PS and phosphatidylethanolamine in the cytosolic leaflet of the membrane [17].The P4-ATPases are a large family of putative aminophospolipid translocases with 14 members in human, 5 in yeast and 6 in C. elegans [20,21].The two founding members, mammalian ATP8A1 (ATPase II) and yeast Drs2p, were found to transfer spinor fluorescent-labeled PS analogues from the inner to the outer membrane leaflet when purified and reconstituted into proteoliposomes, indicating that they have intrinsic flippase activity [22,23].Deletion of DRS2 causes defects in phospholipid translocation and protein transport from the TGN to endosomes and vacuoles in yeast [24][25][26][27].Interestingly, loss of function of yeast Cdc50p, a transmembrane protein that does not contain sequence features indicative of a direct involvement in phospholipid translocation, results in similar defects in lipid asymmetry and vesicular transport to those caused by DRS2 deletion [28,29].Cdc50p and the related protein Lem3p were later found to form complexes with P4-ATPases Drs2p and Dnf1p, respectively, and were shown to facilitate transport of the complexes out of the ER [28,29].Moreover, a recent study suggested that Cdc50 proteins are integral components of P4-ATPases and directly participate in the ATPase reaction cycle [30].Nevertheless, the cellular function of Cdc50 family proteins in multicellular organisms remains unknown.
In C. elegans, the P4-ATPase TAT-1 which is most closely related to yeast Drs2p and mammalian ATP8A1, maintains cell surface PS asymmetry, thus preventing appearance of PS in the outer leaflet of the plasma membrane [31].Moreover, an early step of endocytosis and trafficking in the lysosome biogenesis pathway was found to be defective in tat-1(lf) mutants which accumulate large intestinal vacuoles with characteristics of late endosomes or lysosomes [32].However, it is not known how these different endocytic trafficking processes are affected in tat-1(lf) animals, or whether the defects are caused by disruption of PS asymmetry.
In the present study, we identified a novel C. elegans Cdc50 family protein, CHAT-1, which acts as the chaperone of the TAT-1 P4-ATPase.Endocytic sorting is severely affected in chat-1 and tat-1 mutants, causing abnormal cargo recycling and degradation.CHAT-1 and TAT-1 associate with PS-coated tubular membranes of early endosomes, endocytic recycling compartments (ERCs) and recycling endosomes.Loss of tat-1 and chat-1 function disrupts membrane PS asymmetry and abrogates the tubular membrane structure of sorting and recycling compartments.Our data suggest that TAT-1 and CHAT-1 maintain membrane PS asymmetry to regulate membrane tubulation for cargo sorting and recycling.

Results
chat-1 and tat-1 mutants disrupt plasma membrane PS asymmetry PS is usually confined to the inner leaflet of the plasma membrane and only appears on the cell surface during apoptosis [33].From a forward genetic screen for mutants which disrupt specific labeling of apoptotic cells by a PS-binding protein, we isolated the mutant qx36, and several alleles of tat-1, which encodes a C. elegans P4-ATPase [34] (Materials and Methods).Since P4-ATPases are known to regulate membrane PS asymmetry [17], we investigated whether PS distribution was disrupted in our mutants.Surface-exposed PS can be detected using the secreted fluorescent biosensor GFP::Lact-C2 or Annexin V, both of which bind selectively to PS [19,31].In wild-type animals, surface-exposed PS was only observed around apoptotic cells (Figure 1A, 1D).In qx36 and tat-1 mutants, PS labeling was seen on the surface of virtually all cells, indicating that plasma membrane PS asymmetry was disrupted and PS was exposed on the surface of both living and dying cells (Figure 1B-1F; Figure S1A-S1E) [31].In contrast, no surface labeling was observed in either wild type, qx36 mutants or tat-1(qx30) mutants using a secreted biosensor GFP::Lact-C2(AAA) which does not bind PS (Figure S1H-S1J) [19].
We cloned the gene affected in qx36 mutants and found that it encodes a C. elegans homolog of yeast Cdc50p (Figure S2A, S2L; Materials and Methods).Since Cdc50p is a chaperone and noncatalytic subunit of Drs2p, the P4-ATPase homologous to TAT-1 [28,30], we named this gene chat-1 (chaperone of tat-1).CHAT-1 is most similar to Cdc50p in yeast and CDC50A in human (Figure S2L).All three are predicted to be integral membrane proteins containing two transmembrane domains [35].The chat-1 gene in qx36 mutants contained a C to T transition which resulted in a premature stop codon after Leu 94 (Figure S2A).Similar phenotypes were also observed in ok1681, a deletion mutant of chat-1, or with RNAi inhibition of chat-1 activity (Figure S1F, S1G, and S1K).Although 3 different alternatively spliced chat-1 transcripts are predicted (www.wormbase.org),only chat-1a overexpression rescued both the PS asymmetry and membrane trafficking defects of qx36 mutants (Figure S2A-S2G and see below).

Author Summary
The process by which cells take up nutrients and other large molecules from the extracellular environment is known as endocytosis.At the cell surface, external molecules become enclosed in membrane spheres called endosomes.Early endosomes serve as a sorting station, directing the contents (cargo molecules) to the correct compartment within the cell.This is thought to be achieved by the formation of membrane structures with distinct shape and function.For example, cargoes destined for recycling and degradation are processed through tubular membrane structures and big vesicular compartments, respectively.However, it is poorly understood how early endosome membranes are shaped into different structures.Here we show that two proteins, CHAT-1 and TAT-1, regulate membrane structure and are important for normal endocytic transport in the nematode worm C. elegans.TAT-1 and CHAT-1 are found in tubular membrane structures along the sorting and recycling pathway, where they enrich the outer membrane layer with a lipid called phosphatidylserine (PS) and probably change the membrane curvature.Loss of tat-1 and chat-1 function disrupts the asymmetric distribution of PS, abolishes tubular membrane structures, and abrogates endocytic sorting/ recycling.Our data support a role of TAT-1/CHAT-1regulated membrane PS asymmetry in promoting membrane tubulation for endocytic cargo sorting and recycling.
completely trapped in the ER (Figure S3D, S3E).These data are consistent with previous findings that Cdc50 family proteins and P4-ATPases are mutually required for exiting the ER and that TAT-1 and CHAT-1 are predicted to function as a complex [28,36].
Cargo recycling and degradation are defective in chat-1 and tat-1 mutants Because early and recycling endosomes are affected in tat-1 and chat-1 mutants, we next investigated whether endocytic recycling is defective by examining the trafficking of hTfR, the human transferrin receptor (hTfR::GFP), and hTAC, the a-chain of the human IL-2 receptor TAC (hTAC::GFP), both of which are recycled in a RAB-10-and RME-1-dependent manner in the C. elegans intestine [38].hTfR::GFP accumulated significantly in the cytosol in tat-1 and chat-1 mutants whereas it mainly localized to basolateral membranes in wild type (Figure 3A-3C, 3J).Similarly, increased accumulation of cytosolic hTAC::GFP was also observed in these mutants, albeit to a lesser extent than hTfR (Figure 3D-3F, 3K).These results suggest that recycling of hTfR and hTAC is compromised.Moreover, the glucose transporter 1 (GLUT1), which enters mammalian cells through clathrin-independent endocytosis and constitutively recycles via the Arf6 pathway that require the function of Rab GTPases and RME-1/EHD1 family proteins, primarily localized to apical and basolateral cell membranes when expressed in the C. elegans intestine (Figure 3G) [39].In tat-1(qx30) and chat-1(qx36) mutants, however, cytosolic accumulation of GLUT1::GFP dramatically increased and some of the signal appeared in abnormal vacuoles, indicating that trafficking of GLUT1 to the plasma membrane was disrupted (Figure 3H, 3I, 3L).Intracellularly accumulated hTfR and GLUT1 were found on vesicles positive for RAB-5, RAB-10 or RAB-7 in chat-1(RNAi) animals, especially on vesicles that clustered together, suggesting that they may be trapped in abnormal early endosomes (Figure 3M-3R, 3U-3Z).Consistent with this notion, hTfR and GLUT1 colocalized with RME-1 on basal and lateral membranes in wild type, but partially overlapped with cytoplasmic RME-1 aggregates in chat-1(RNAi) animals (Figure 3S, 3T, 3Z1, and 3Z2).hTfR or GLUT1 did not significantly overlap with either CHC-1, a marker for clathrin-coated pits, or Lysotracker Red in wild type or chat-1(RNAi) animals (data not shown).These data indicate that loss of tat-1 and chat-1 function results in defective recycling of cargoes, which are mainly trapped within abnormal early endosomes.
As abnormal RAB-7 distribution and fewer mature lysosomes were observed in tat-1 and chat-1 intestines, we examined whether the degradative pathway is also affected by using the VIT-2::GFP reporter to monitor yolk trafficking and accumulation [40,41].The initial uptake of yolk in both mature oocytes and fertilized embryos was normal in tat-1 and chat-1 mutants with no obvious accumulation of VIT-2::GFP in the body cavity (Figure 4A-4E).However, the redistribution of yolk to the gut primordium and their degradation appeared to be affected as significantly more VIT-2::GFP was observed in both early and late embryos as well as L1 larvae in tat-1 and chat-1 mutants than in wild type (Figure 4E-4V; Figure S5G-S5I).Moreover, we found that aged tat-1 and chat-1 mutants but not wild-type animals (60 h post L4/ adult molt), accumulated a large number of big yolk granules in the intestine (Figure 4W-4Z; Figure S5J-S5L).As tat-1 and chat-1 adults aged for a shorter period of time (12, 24 or 48 h post L4/ adult molt) contained a similar level of yolk in the intestine as in wild type, our data are consistent with compromised yolk degradation in these mutants (data not shown).In addition, tat-1 and chat-1 intestines accumulated many LGG-1-postive structures, which were disrupted in animals lacking atg-3, atg-5 or atg-7, suggesting that the degradation of autophagic cargo may also be affected (Figure 4Z1-Z4; data not shown) [42].

CHAT-1 and TAT-1 associate with tubular membranes in sorting and recycling compartments
To understand how disruption of CHAT-1 and TAT-1 function results in endocytic defects, we examined their subcellular localization in the intestine by coexpressing CHAT-1::CFP and TAT-1::YFP under control of the intestine-specific promoter vha-6 (P vha-6 chat-1::cfp +P vha-6 tat-1::yfp).These reporters fully rescued the vacuolation phenotype in tat-1 and chat-1 mutants (Figure S2A).TAT-1 and CHAT-1 colocalized to both plasma membranes and intracellular tubular and vesicular structures (Figure 1L; Figure S3A).The intestinal tubular and vesicular localization pattern was also observed when the expression of TAT-1 or CHAT-1 was controlled by the endogenous promoter (Figure S7A, S7B; data not shown).To determine the identities of the cytosolic compartments labeled by CHAT-1 and TAT-1, we coexpressed mCHERRY fusions of different endocytic markers together with CHAT-1::GFP and TAT-1 (P vha-6 chat-1::gfp +P vha-6 tat-1).TAT-1 was included to ensure efficient ER export of CHAT-1::GFP; this combination is subsequently referred to as CHAT-1::GFP for simplicity.CHAT-1::GFP displayed a tubular and vesicular staining pattern, which did not overlap with either the Golgi marker MANS or the lysosomal marker Lysotracker Red, indicating that CHAT-1 is not on the Golgi or mature lysosomes (Figure 5E, 5F).No CHAT-1::GFP was found on the RAB-7-positive ring-like structures, suggesting that it is not enriched on late endosomes or early lysosomes (Figure 5D).However, CHAT-1 partially overlapped with RAB-5 on punctate structures, but not on tubule-like structures that were negative for mCHERRY::RAB-5 (Figure 5A).Thus, a proportion of CHAT-1/ TAT-1 may localize to RAB-5-positive early endosomes.We next examined colocalization of CHAT-1 and RAB-10, which associates with endosomes and Golgi compartments [38].Interestingly, when coexpressed with CHAT-1::GFP, RAB-10 displayed a different staining pattern and mainly localized to CHAT-1-positive tubular structures instead of labeling small cytoplasmic puncta (compare Figure 2G and Figure 5B).Similarly, RAB-11 labeled punctate structures when expressed alone, but colocalized with CHAT-1 on abundant tubules when the two were coexpressed (compare Figure 2J and Figure 5C).Moreover, in animals expressing both mCHERRY::RAB-10 (or RAB-11) and CHAT-1::GFP, the tubular structures became more evident and extensive than in animals carrying only CHAT-1::GFP or animals coexpressing CHAT-1 and RAB-5 or RAB-7 (compare B and C with other panels of Figure 5).These data suggest that RAB-10 and RAB-11 may act together with CHAT-1/TAT-1 to promote extension of the tubular structures.
As CHAT-1 tubules were labeled by RAB-10 and RAB-11, but not RAB-5 or RAB-7, we reasoned that they may be tubular early endosomes and/or endocytic recycling compartments (ERCs).We first examined the tubular structures in rab-10(lf) mutants, in which endocytic transport from early to recycling endosomes is disrupted, and found that CHAT-1 tubules were totally abolished; instead, CHAT-1::GFP overlapped completely with RAB-5 on enlarged early endosomes (Figure 6B, 6E).Inactivation of rab-10 also disrupted tubules labeled by CHAT-1 and RAB-11 (Figure S7E).Next, we examined the CHAT-1 tubules in rme-1(lf) mutants, in which trafficking from recycling endosomes to the plasma membrane is affected, and found that the tubules were not disrupted (Figure 6C).Instead, CHAT-1-and RAB-10-(or RAB-11) positive tubular structures became even more extended when rme-1 function was lost (Figure 6G; Figure S7D).Conversely, loss of rab-5 activity completely abrogated the tubular structures in either wild type or rme-1(b1045) mutants (Figure 6H; data not shown).RAB-5 and RAB-10 are required in early endosomes and for trafficking from early to recycling endosomes, while RME-1 acts downstream of them to promote membrane fission for releasing recycling carriers [14,38].Our data are consistent with the idea that CHAT-1/TAT-1 associates with the tubular membrane of early endosomes and ERCs.Finally, we examined the co-localization of CHAT-1 and RME-1, which is enriched on recycling endosomes [13].We observed that CHAT-1::GFP overlapped with mRFP::RME-1 on basolateral tubulo-vesicular structures, indicating that CHAT-1 also localizes to RME-1positive recycling endosomes (Figure 7A).

CHAT-1 and TAT-1 are required for tubule formation
The formation of tubular extensions in early endosomes, ERCs and recycling endosomes is crucial for sorting and transporting recycling cargoes.We observed that the recycling cargo GLUT1 labeled CHAT-1-positive tubules near basolateral membranes and in the cytoplasm, supporting a role for these tubules in sorting and recycling (Figure 7E; data not shown).GLUT1::GFP was also found on RME-1-positive tubulo-vesicular recycling endosomes (Figure 7F).The tubular membrane structures containing GLUT1::GFP were completely disrupted in tat-1 and chat-1 mutants, in which GFP stained cytoplasmic punctate structures (Figure 7G-7I).Moreover, in tat-1 and chat-1 mutants, the remaining basolateral RME-1-positive puncta completely lost their tubulo-vesicular morphology and became globular (Figure 7B-7D).These data indicate that TAT-1 and CHAT-1 are required for forming and/or maintaining the tubular extensions of sorting and recycling compartments.

PS asymmetry across endomembranes is disrupted in tat-1 and chat-1 mutants
To investigate whether TAT-1 and CHAT-1 regulate endocytic sorting and recycling by maintaining membrane PS asymmetry, we examined PS asymmetry across endomembranes.We first determined PS distribution by expressing the biosensor GFP::Lact-C2 specifically in intestine cells (P ges-1 GFP::Lact-C2) [43].The cell membranes and surfaces of virtually all internal vesicles were labeled, indicating that PS was exposed on the cytosolic surface of both plasma membranes and various intracellular compartments (Figure S8A).We next coexpressed GFP::Lact-C2 with different endolysosomal markers and found that it labeled intracellular structures that were positive for RAB-5, RAB-7, RAB-10, RME-1 or Lysotracker Red (Figure 8A-8E).Thus, PS appeared on the cytosolic surfaces of recycling, early and late endosomes as well as lysosomes.mCHERRY::Lact-C2 coincided well with CHAT-1::GFP on tubular membranes, indicating that the tubules are coated by PS (Figure 8F).
In coelomocytes, which are scavenger cells that actively endocytose and degrade soluble material, GFP::Lact-C2 (P unc-122 GFP::Lact-C2) stained plasma membranes and the surfaces of endosomes and lysosomes, as observed in intestine cells (Figure S8D, S8E).To determine whether PS is absent from the luminal leaflet of endomembranes in wild type and whether PS asymmetry is affected in tat-1 and chat-1 mutants, we focused on coelomocytes, which contain abundant endocytic vesicles that are larger than those in intestine cells.To detect luminal PS in endocytic vesicles, we examined expression of ssGFP::Lact-C2 driven by the myo-3 promoter (P myo-3 ssGFP::Lact-C2), which after secretion from body wall muscle cells is taken up by coelomocytes through endocytosis and transferred to lysosomes via endocytic transport [44].In wildtype coelomocytes, the endocytosed GFP::Lact-C2 mainly accumulated in lysosomes as indicated by an endocytic cargo ssCHERRY, whereas no clear GFP signal was detected in endosomes, suggesting that PS is likely absent from the luminal side of endomembranes (Figure 8G; Figure S8F, S8G).Therefore, like the plasma membrane, PS is preferentially distributed on the cytosolic side of endomembranes.Remarkably, we found that in both tat-1(qx30) and chat-1(qx36) mutants, the internalized GFP::Lact-C2 labeled endosome membranes with a ring-like staining pattern, indicating that PS appeared on the luminal side of endomembranes (Figure 8H, 8I).By contrast, expression of GFP::Lact-C2(AAA), which is deficient in PS binding, gave no or very faint and diffuse GFP signal in endosomes (Figure S8H-S8J).These data suggest that in tat-1 and chat-1 mutants, PS asymmetry across endomembranes is disrupted, causing PS to appear on both cytosolic and luminal leaflets of the membrane.

CHAT-1 acts together with TAT-1 to regulate PS asymmetry and endocytic transport
From a genetic screen for altered distribution of a PS-binding protein, we recovered mutant alleles of chat-1 and tat-1, which displayed identical PS asymmetry phenotypes on both plasma membranes and endomembranes.Mutants of chat-1 and tat-1 also showed identical endocytic defects.CHAT-1 and TAT-1 are expressed in the same tissues and colocalize to both plasma membranes and various intracellular compartments in the intestine.Furthermore, CHAT-1 and TAT-1 are co-dependent for exiting the ER, similar to P4-ATPase and Cdc50p proteins in yeast and mammalian cells [28,36].Therefore, like Drs2p and Cdc50p, TAT-1 and CHAT-1 may act as a complex to regulate membrane PS asymmetry and endocytic traffic.

TAT-1/CHAT-1 regulates endocytic sorting
We observed pleiotropic phenotypes associated with the presence of multiple endocytic vesicles in tat-1 and chat-1 mutants, many of which can be attributed to defective endocytic sorting.For example, in tat-1 and chat-1 mutants, enlarged early endosomes accumulated while recycling and late endosomes were disrupted.The abnormal vacuoles appear to be heterogeneous, since they were labeled by markers of early, late and recycling endosomes as well as early lysosomes [32].Cargo recycling and degradation are also defective in these mutants.Therefore, loss of tat-1 and chat-1 function likely disrupts endocytic sorting through early endosomes, thereby affecting subsequent trafficking through both recycling and degradative pathways.
In a recent study, TAT-1 was found to be required at an early step of endocytosis [32].Consistent with this, we observed a defect in endocytosis of fluid cargo from both basolateral and apical intestinal cell membranes in tat-1 and chat-1 mutants (data not shown).However, defective endocytosis was not seen in oocytes when yolk was taken up.Instead, we observed defects in yolk redistribution and digestion in tat-1 and chat-1 embryos (Figure 4).

TAT-1/CHAT-1 may regulate membrane tubulation by maintaining PS asymmetry
How is endocytic sorting through early endosomes regulated by the P4-ATPase TAT-1 and its chaperone CHAT-1?The tubular elements of early endosomes serve as sorting platforms to enrich and transport transmembrane cargoes.Several lines of evidence indicate that TAT-1 and CHAT-1 are required for generating and/or maintaining the tubular membrane structure.Firstly, tat-1 and chat-1 mutants disrupt endocytic transport through early endosomes, leading to the accumulation of enlarged early endosomes positive for RAB-5, which labels the vesicular but not the tubular element.Secondly, TAT-1 and CHAT-1 associate with tubular membranes at sorting and recycling compartments, which contain the recycling cargo GLUT1.Thirdly, loss of tat-1 and chat-1 function abolishes the tubular membrane structure containing GLUT1 and disrupts the tubulo-vesicular morphology of RME-1-positive recycling endosomes.
Phospholipids can have significant effects on membrane curvature when their distributions between the two membrane leaflets are altered.It was observed that addition of exogenous phospholipids including PS to the outer leaflet of discoid platelets caused expansion of the outer surface in the form of numerous extensions [45].In another case, incorporation of PS, phosphatidylethanolamine and phosphatidylcholine into the outer leaflet of discoid erythrocytes increased the outer membrane surface area and induced a crenated shape with a higher ratio of external to internal surface area, whereas transverse diffusion of exogenous phospholipids from the outer to the inner leaflet reversed the shape change [17,45].Our findings that CHAT-1-assoicated tubules are coated by PS and that loss of tat-1 and chat-1 function disrupts PS asymmetry of endomembranes and abrogates tubular extensions strongly suggest a role of PS and/or PS asymmetry in membrane tubulation.As an aminophospholipid transporter, TAT-1/CHAT- 1 may catalyze the active translocation of PS from the luminal to the cytosolic leaflet of endomembranes, which results in a high ratio of PS in the cytosolic leaflet versus the inner leaflet and an increased outer monolayer area, leading to the deformation of membranes into tubular extensions [46] (Figure 9).Consistent with this, we observed extensive tubular membrane structures labeled by the PS biosensor mCHERRY::Lact-C2 in animals overexpressing CHAT-1 and TAT-1.
Our data suggest that RAB-10 and RAB-11 may also contribute to tubule formation or extension as both of them, when overexpressed, become enriched on CHAT-1::GFP-positive tubules and enhance tubule extension, whereas loss of rab-10 function completely disrupts tubular structures and traps CHAT-1 on enlarged RAB-5-positive early endosomes (Figure 9).However, we do not know whether RAB-10 plays a direct or indirect role in this process, or how its function is achieved.One interesting possibility is that RAB-10 and RAB-11 may contribute to tubule formation by regulating membrane-bending proteins, as BAR proteins are found to interact with a variety of effectors of small GTPases including Rabs [9].
Our data support a role of TAT-1/CHAT-1 in membrane tubulation for cargo sorting and recycling.However, we do not know how the sorting of cargo for degradation may be affected by loss of tat-1 and chat-1 function.Interestingly, it was reported that tat-1(lf) mutants accumulate giant multi-vesicular bodies (MVBs) in hypodermal cells [32].As translocation of PS and PE across the membrane bilayer is thought to provide the driving force for membrane bending [17,47,48], we suspect that loss of TAT-1 and CHAT-1 function may lead to defective budding from multivesicular compartments of early endosomes, thereby affecting the sorting of cargo into the degradative pathway.Further experiments need to be performed to test this hypothesis, especially the examination of MVBs in intestine cells.
Isolation, mapping, and cloning of chat-1 and tat-1 TTR-52 is a secreted protein that specifically recognizes apoptotic cells through its binding to surface exposed phosphatidylserine (PS) [34].In wild-type embryos carrying P hsp TTR-52::mCHERRY, apoptotic cells are surrounded by mCHERRY, which is absent from the surface of living cells.In order to understand how the PS engulfment signal is regulated, we performed a forward genetic screen to look for mutants which disrupt or alter the staining of apoptotic cells by TTR-52::mCHERRY.From this screen, we isolated the qx36 mutant and several alleles of tat-1, which resulted in TTR-52::mCHERRY staining of virtually all cells, both dying and living [34].
qx36 was mapped to linkage group IV.Two rounds of threepoint mapping were performed using unc-17 (23.11) dpy-13 (0.00) and dpy-13 (0.00) unc-8 (+3.29), which mapped qx36 to a small genetic interval between 0.00-1.02.Transformation rescue experiments were performed and one fosmid clone in this region, WRM0637aA04, rescued the qx36 defect.Long PCR fragments covering different open reading frames within this fosmid were tested and only the fragment covering R08C7.2 possessed rescue activity.R08C7.2 encodes a Cdc50p-like protein of 348 amino acids, which we named chat-1 (chaperone of tat-1).Sequencing of the locus in the qx36 mutant identified a C to T transition, which  (B, E, H) and chat-1(qx36) (C, F, I) that express hTfR::GFP (A-C), hTAC::GFP (D-F) or GLUT1::GFP (G-I).GFP signal was mainly seen on plasma membranes in wild type (arrows) but accumulated intracellularly in tat-1 and chat-1 mutants (arrowheads).(J-L) Quantification of intracellular accumulation of hTfR::GFP (J), hTAC::GFP (K) and GLUT1::GFP as shown in (A-I).Data are shown as mean numbers of labeled structures 6 SEM.*P,9.0610 229.(M-Z2) hTfR::GFP (M-T) and GLUT1::GFP (U-Z2) are trapped in abnormal early endosomes in chat-1(RNAi) animals.Merged images of hTfR::GFP or GLUT1::GFP with mCHERRY RAB-5 (M, N, U, V), mCHERRY::RAB-10 (O, P, W, X), mCHERRY::RAB-7 (Q, R, Y, Z) or mRFP::RME-1 (S, T, Z1, Z2) in wild-type and chat-1(RNAi) intestines are shown.Overlap of hTfR or GLUT1 with different endocytic markers is indicated by arrows in wild type and arrowheads in chat-1(RNAi) animals.Scale bars: 5 mm.doi:10.1371/journal.pgen.1001235.g003resulted in a premature stop codon after Leu 94.Given that similar phenotypes were observed in ok1681, a chat-1 deletion mutant containing an 1152 bp deletion that removes the region from exon 2 to intron 5 of the chat-1 gene, or when chat-1 is inactivated by RNAi, qx36 is probably a null or strong loss-offunction mutation of chat-1.9)).Since the tat-1 gene locates at 17.59 of LGIII, a transformation rescue experiment was performed and a DNA fragment containing the tat-1 gene fully rescued the qx30 defect.The sequence of the tat-1 gene was determined in qx30 and three other mutants, qx22, qx23, qx24, which are allelic to qx30.The qx22 and qx23 mutants carry the same mutation that results in substitution of the Ala at codon 627 with Val, whereas the qx24 mutant has a splicing mutation (G to A transition) at the junction of intron 12 and exon 13.The qx30 mutant has a G to A transition that results in a premature stop codon after Leu 1028.

RNAi
To inactivate tat-1 and chat-1 by RNAi, dsRNA synthesized in vitro (550 ng/ml) was injected into the gonad of young adult hermaphrodites (see Table S1 for primer sequences).Embryos laid between 16 to 24 h post-injection were either used for analyzing PS asymmetry or cultured until the L4 larval or young adult stage for examining intestinal phenotypes.We found that tat-1 RNAi significantly diminished the expression of tat-1.For example, 65% of embryos (n = 40) transgenic for P tat-1 tat-1::gfp had bright GFP fluorescence before injection, but only 4% of them showed similar GFP intensity after injection.RNAi treatment of tat-1 also resulted in a 56% reduction of TAT-1::GFP expression in larvae (n = 40).Similarly, 100% of animals carrying an integrated P chat-1 chat-1::gfp reporter had strong GFP fluorescence before injection, but after chat-1 RNAi treatment, 99% had only weak GFP signal (n = 30).For rab-5 and rab-10 RNAi, the bacterial feeding protocol was used as described before [54].Briefly, L3 larvae were treated with either rab-5 RNAi (I-4J01) or control RNAi (pPD129.36) and adult animals of the same generation were scored as most F1 progeny die due to inactivation of rab-5.For rab-10 RNAi, L3 or L4 larvae were treated with either rab-10 RNAi (I-3G23) or control RNAi (pPD129.36) and F1 progeny were examined at the adult stage.

Examination of PS asymmetry and yolk accumulation
The ex vivo staining of dissected gonads by Annexin V was performed as described before [55].To determine membrane PS asymmetry in embryos, mixed stage animals carrying P hsp ss GFP::Lact-C2 or P hsp ssGFP::Lact-C2(AAA) were incubated at 33uC for 1 h followed by recovery at 20uC for 2.5 h before examination.To examine PS asymmetry across endomembranes in coelomocytes, mixed stage P0 animals carrying both P myo-3 ss GFP::Lact-C2 and P hsp ssCHERRY were incubated at 33uC for 0.5 h to induce the expression and secretion of ssCHERRY.Culture was continued at 20uC for one more generation before F1 adults were examined.This long incubation allows the complete uptake and transport of ssCHERRY to lysosomes, which otherwise is seen in the body cavity and many tissues other than coelomocytes.To examine yolk uptake in oocytes, L4 hermaphrodites from the strains indicated were aged for 12 and 24 h and confocal images were taken with the same exposure time.The uptake of yolk was observed at the same two time points in wild type and tat-1 and chat-1 mutants.Images of animals that were aged to 24 h post L4/adult molt are shown in Figure 4. To examine yolk accumulation in the intestine, L4 larvae were aged for 12, 24, 48, and 60 h and images were taken with the same exposure time.No obvious difference in yolk accumulation was observed in wild-type, tat-1 or chat-1 intestines when animals were aged less than 60 h.

Quantification of cargo recycling
Intracellular accumulation of hTfR::GFP, hTAC::GFP and GLUT1::GFP was quantified by determining the number of labeled structures within a 250 mm 2 area in the intestine.Lysotracker Red-positive granules and mRFP::RME-1-positive vesicles were counted within a unit area of 150 mm 2 .LGG-1 puncta and GFP::RAB-10-or GFP::RAB-11-labeled aggregated structure (clustering of .2labeled puncta) were scored within a unit area of 500 mm 2 .Five different areas were chosen and quantified in each animal at the L4 larval or young adult stage and 8 animals were scored for each strain.The average total intensity per unit area of GFP::RAB-5 and GFP::RAB-7 in the intestine and VIT-2::GFP in fertilized early embryos was measured using Image J 1.42q software.For GFP::RAB-5 and GFP::RAB-7, 5 different areas (40 mm 2 each) in the intestine of L4 larvae were chosen for each animal and 8 animals were quantified.For VIT-2::GFP in fertilized early embryos, 3 different regions (20 mm 2 each) were chosen for each embryo and 32 embryos were scored.Axiovision Rel.4.7 software (Carl Zeiss, Inc.) was used to quantify average total intensity per unit area of VIT-2::GFP in 4-fold stage embryos or in the intestine of aged adult (60 h post L4/adult molt).3 different regions (12.6 mm 2 each) were chosen for each animal and 32 embryos or 28 intestines were quantified.Student's two tailed unpaired t-test was performed and the P value was indicated in the figure legend.

Antibody generation and immunostaining
CHAT-1(74-306) or full-length RAB-7 protein tagged with six Histidine residues (CHAT-1(74-306)-His 6 or RAB-7-His 6 ) was expressed in and purified from E. coli and used to raise rat polyclonal antibody against CHAT-1 or RAB-7.The antibodies were further purified by incubating 3 ml rat serum with nitrocellulose membrane strips containing 5 mg CHAT-1(74-306)-His 6 or RAB-7-His 6 protein.Bound antibodies were eluted with 100 mM glycine-HCl (pH 2.5).Purified anti-RAB-7 antibody recognized a single band of expected size (24 KD) in a western blot analysis using lysate prepared from mixed staged wild-type worms, while purified anti-CHAT-1 antibody failed to detect endogenous CHAT-1 expression in wild-type animals, but recognized CHAT-1::GFP (64 KD) or CHAT-1-His 6 (39 KD) when the western blot was performed using lysate prepared from transgenic animals expressing CHAT-1::GFP or CHAT-1-His.In a whole-mount immunostaining experiment, anti-CHAT-1 antibody stained plasma membranes in wild-type but not chat-1(qx36) embryos or oocytes.RAB-7 antibody staining revealed a specific pattern reminiscent of GFP::RAB-7 in the intestine of wild-type but not rab-7 RNAi-treated animals except for the staining of apical membranes which appears to be non-specific.Monoclonal anti-GFP antibody was purchased from Roche (USA) and anti-RME-1 antibody was obtained from Developmental Studies Hybridoma Bank (University of Iowa, USA).For immunostaining, mixed stage embryos or dissected intestines were fixed with methanol/acetone followed by blocking in phosphate buffered saline (PBS) containing 1% BSA and 10% fetal calf serum for 2 h at 4uC.The samples were then incubated with primary antibodies in blocking buffer at 4uC overnight with 1:50 dilution for RAB-7, RME-1, CHAT-1 antibodies, 1:200 for GFP antibody and 1:500 for LGG-1 antibody.After washing three times with PBST (PBS + 0.2% Tween 20), the samples were incubated with secondary antibody conjugated to Alexa-488 or Alexa-546 (Molecular Probes) at a 1:50 dilution for 2 h at room temperature.The stained samples were washed three times as before and mounted in 15% VECTASHIELD mounting medium with DAPI (VECTOR) and visualized using a Zeiss LSM 510 Meta inverted confocal microscope.

Microscopy and imaging analysis
DIC and fluorescent images were captured with a Zeiss Axioimager A1 equipped with epifluorescence and an AxioCam monochrome digital camera and were processed and viewed using Axiovision Rel.4.7 software (Carl Zeiss, Inc.).A 100x Plan-Neofluar objective (NA1.30) was used with Immersol 518F oil (Carl Zeiss, Inc.).For confocal images, a Zeiss LSM 5 Pascal inverted confocal microscope with 488, 543, 514, 458 and 405 lasers was used and images were processed and viewed using LSM Image Browser software.The top bar indicates the genetic map of the chat-1 genomic region and the lower panels show the rescue of chat-1(qx36).At least 15 animals (non-transgenic and transgenic) from each independent transgenic line were scored for all lines obtained as indicated in parentheses.Rescue activity was determined by examining the intestine vacuolation phenotype when chat-1 expression was driven by the vha-6 promoter (indicated by asterisks); whereas both PS asymmetry and intestine vacuolation phenotypes were scored when chat-1 expression was controlled by the endogenous promoter.The chat-1 gene structure is shown with filled boxes representing the exons and thin lines indicating the introns.The arrows pointing away from the 39 exons delineate the direction of transcription.Three different transcripts of chat-1 gene are predicted due to alternative splicing.The positions of the intragenic mutation identified in the qx36 mutant and the genomic deletion in the chat-1 deletion mutant ok1681 are also indicated.(B-G) Fluorescent images of chat-1(qx36) mutants expressing ssGFP::Lact-C2 driven by heat-shock promoters (P hsp ssGFP::Lact-C2) and/or P chat-1 chat-1a (B, C) or P chat-1 chat-1b (D, E) or P hsp chat-1c (F, G).Expression of chat-1a but not chat-1b or chat-1c rescued the PS asymmetry and intestinal vacuolation phenotypes of chat-1(qx36) mutants.(H-K) DIC images of the intestine in tat-1(qx30) (H) and chat-1(qx36) (J) mutants with or without overexpression of tat-1 (I) or chat-1 (K) driven by the intestine-specific promoter vha-6.The intestinal vacuolation phenotype was rescued by expressing tat-1 or chat-1 specifically in the intestine.Scale bars: 5 mm.(L) Phylogenetic tree of yeast, C. elegans, and human CDC50 family proteins.Sequences of CDC50 family proteins from yeast, C. elegans, and human were compared using the CLUSTALW program.The phylogenetic tree was generated based on the multiple sequence alignment and is shown as a Neighbour-Joining Tree with branch length.Found at: doi:10.1371/journal.pgen.1001235.s002(0.49 MB PDF)

Figure 9 .
Figure 9. Proposed functions of TAT-1/CHAT-1 in endocytic sorting and recycling.TAT-1 and CHAT-1 localize to plasma membranes and to tubular early endosomes, ERCs and recycling endosomes where they regulate sorting and recycling events by restricting PS to the cytosolic membrane leaflet.TAT-1/CHAT-1 probably promotes tubule formation by enriching PS on the cytosolic leaflet, thereby affecting membrane curvature.RAB-10 and RAB-11 may also contribute to membrane tubulation in early endosomes and ERCs, whereas RME-1/AMPH-1 is involved in membrane fission and tubulation of recycling endosomes.PS, phosphatidylserine; PE, phosphatidylethanolamine; PC, phosphatidylcholine; SM, sphingomyelin.doi:10.1371/journal.pgen.1001235.g009