FYCO1 Contains a C-terminally Extended, LC3A/B-preferring LC3-interacting Region (LIR) Motif Required for Efficient Maturation of Autophagosomes during Basal Autophagy*

Background: FYCO1 binds to LC3 and is involved in transport of autophagosomes. Results: FYCO1 uses a C-terminally extended LIR motif for specific interaction with LC3A/B stimulating autophagosome maturation during basal autophagy. Conclusion: The transport adaptor FYCO1 engages in a specific interaction with LC3A/B to stimulate autophagosome maturation. Significance: This work will lead to an increased understanding of LIR interactions and the role of FYCO1 in autophagy. FYCO1 (FYVE and coiled-coil protein 1) is a transport adaptor that binds to phosphatidylinositol 3-phosphate, to Rab7, and to LC3 (microtubule-associated protein 1 light chain 3) to mediate transport of late endosomes and autophagosomes along microtubules in the plus end direction. We have previously shown that FYCO1 binds to LC3B via a 19-amino acid sequence containing a putative core LC3-interacting region (LIR) motif. Here, we show that FYCO1 preferentially binds to LC3A and -B. By peptide array-based two-dimensional mutational scans of the binding to LC3B, we found FYCO1 to contain a C-terminally extended LIR domain. We determined the crystal structure of a complex between a 13-amino acid LIR peptide from FYCO1 and LC3B at 1.53 Å resolution. By combining the structural information with mutational analyses, both the basis for the C-terminally extended LIR and the specificity for LC3A/B binding were revealed. FYCO1 contains a 9-amino acid-long F-type LIR motif. In addition to the canonical aromatic residue at position 1 and the hydrophobic residue at position 3, an acidic residue and a hydrophobic residue at positions 8 and 9, respectively, are important for efficient binding to LC3B explaining the C-terminal extension. The specificity for binding to LC3A/B is due to the interaction between Asp1285 in FYCO1 and His57 in LC3B. To address the functional significance of the LIR motif of FYCO1, we generated FYCO1 knock-out cells that subsequently were reconstituted with GFP-FYCO1 WT and LIR mutant constructs. Our data show that FYCO1 requires a functional LIR motif to facilitate efficient maturation of autophagosomes under basal conditions, whereas starvation-induced autophagy was unaffected.


FYCO1 (FYVE and coiled-coil protein 1) is a transport adaptor that binds to phosphatidylinositol 3-phosphate, to
Rab7, and to LC3 (microtubule-associated protein 1 light chain 3) to mediate transport of late endosomes and autophagosomes along microtubules in the plus end direction. We have previously shown that FYCO1 binds to LC3B via a 19-amino acid sequence containing a putative core LC3-interacting region (LIR) motif. Here, we show that FYCO1 preferentially binds to LC3A and -B. By peptide array-based twodimensional mutational scans of the binding to LC3B, we found FYCO1 to contain a C-terminally extended LIR domain. We determined the crystal structure of a complex between a 13-amino acid LIR peptide from FYCO1 and LC3B at 1.53 Å resolution. By combining the structural information with mutational analyses, both the basis for the C-terminally extended LIR and the specificity for LC3A/B binding were revealed. FYCO1 contains a 9-amino acid-long F-type LIR motif. In addition to the canonical aromatic residue at position 1 and the hydrophobic residue at position 3, an acidic residue and a hydrophobic residue at positions 8 and 9, respectively, are important for efficient binding to LC3B explaining the C-terminal extension. The specificity for binding to LC3A/B is due to the interaction between Asp 1285 in FYCO1 and His 57 in LC3B. To address the functional significance of the LIR motif of FYCO1, we generated FYCO1 knock-out cells that subsequently were reconstituted with GFP-FYCO1 WT and LIR mutant constructs. Our data show that FYCO1 requires a functional LIR motif to facilitate effi-cient maturation of autophagosomes under basal conditions, whereas starvation-induced autophagy was unaffected.
Macroautophagy (hereafter autophagy) is an evolutionary conserved degradation pathway that directs surplus or damaged cytosolic components to lysosomal degradation to maintain cellular homeostasis (1). A double membrane structure, the phagophore, grows around part of the cytoplasm, or targeted components, and closes upon itself to form the autophagosome (1). The autophagosome matures by fusion, either with a late endosome, or directly with a lysosome.
Proteins interacting with Atg8 family proteins via a LIR motif are either selective autophagy receptors or proteins interacting with Atg8 proteins on the outside of autophagic structures and other vesicles. Only two LIR-containing proteins, MAPK8IP1/ JIP1 and FYCO1 (FYVE and coiled-coil protein 1), are reported to be involved in transport of autophagosomes (21,22). FYCO1 is a 1478-amino acid-long phosphatidylinositol 3-phosphatebinding protein and Rab7 effector that interacts with LC3, and is involved in the transport of autophagosomes along microtubules in the plus end direction (21). Mutations in FYCO1 cause autosomal-recessive congenital cataracts, suggesting that FYCO1 is required for human lens development, transparency, or both (23). FYCO1 is recruited by LC3 to Dectin-1 phagosomes during LC3-associated phagocytosis to facilitate maturation of early p40phox-containing phagosomes into late LAMP1-positive phagosomes (24). In macrophage cell lines exposed to LPS, tubular lysosomes form. Rab7 and its effectors RILP (Rab7-interacting lysosomal protein) and FYCO1 are required for this formation through modulation of dynein-and kinesin-driven transport along microtubules, respectively (25). Intriguingly, FYCO1 is involved, together with the endoplasmic reticulum protein protrudin, in mediating microtubule-dependent transport of late endosomes via endoplasmic reticulumendosome contact sites to produce cell protrusions and neurite outgrowth (26).
Here, we describe the details of the interaction between FYCO1 and LC3B based on x-ray crystallography and mutational analyses. FYCO1 binds LC3 via a canonical F-type core LIR motif. However, a C-terminal extension of the LIR is required because a vital interaction for the binding is between Glu 1287 (position 8) and Arg 70 of LC3. FYCO1 preferably interacts with LC3A and -B. This is, at least in part, due to a specific interaction between His 57 in LC3B and Asp 1285 in the FYCO1 LIR. His 57 is found in LC3A and -B, but not in other ATG8 family members. By reconstituting FYCO1 KO cells with FYCO1 WT and FYCO1 with the core LIR mutation F1280A/ I1283A, we found that a functional LIR is required for efficient maturation of autophagosomes under basal conditions.
Cell Culture and Transfections-HeLa cells were grown in Eagle's minimum essential medium supplemented with 10% fetal bovine serum (Biochrom AG, S0615), non-essential amino acids, 2 mM L-glutamine, and 1% streptomycin-penicillin (Sigma, P4333). HEK293 cells were maintained in Dulbecco's modified Eagle's medium with the same supplements as described above. Subconfluent cells were transfected with plas-FYCO1 Contains an LC3-preferring, C-terminally Extended LIR mids using TransIT-LT1 (Mirus, MIR2300) (for microscopy analysis) or Metafectene PRO (Biontex) (for immunoprecipitation and zinc finger nucleases) following the supplier's instructions. Twenty-four hours after transfection, cells were fixed and permeabilized in pre-chilled (Ϫ20°C) methanol for 10 min and washed two times in PBS. Fixed cells were blocked with 3% pre-immune goat serum in PBS for 30 min at room temperature before incubation for 1 h at room temperature with primary antibodies diluted in PBS with 1% goat serum. Cells were washed five times in PBS before incubation with Alexa Fluor secondary antibodies (Life Technologies) diluted 1:500 supplemented with 1% goat serum. Before imaging, cells were washed five times in PBS. Stable Flp-In GFP-FYCO1 and GFP-FYCO1 LIRmut cell lines were generated using the Flp-In recombination system following the manufacturer's protocol (Invitrogen). Cells were treated as indicated with 0.2 M bafilomycin A1 (Sigma, B1793).
Immunoprecipitation-Transfected cells were rinsed twice with PBS prior to lysis in a modified radioimmunoprecipitation assay buffer (50 mM Tris, pH 7.4, 150 mM NaCl, 2 mM EDTA, 0.25% sodium deoxycholate, 1% Nonidet P-40) supplemented with cOmplete Mini EDTA-free protease inhibitor mixture tablets (1 tablet/10 ml) (11836170001, Roche Applied Science). Lysates were centrifuged for 5 min at 13,000 rpm 4°C followed by incubation with anti-Myc affinity gel (B23401 (Biotool)) for 18 h at 4°C. Beads were washed five times with lysis buffer and eluted with Myc peptide (B23411 (Biotool)) while shaking at 4°C for 30 min. Supernatant was added to SDS-PAGE loading buffer with 1 mM DTT and boiled for 5 min. Samples were resolved by SDS-PAGE and transferred to Nitrocellulose membrane (LI-COR). Membranes were blocked with Odyssey chemical blocking buffer (LI-COR) (diluted 50% in PBS) for 30 min.
Generation of FYCO1 Knock-out Cell Lines-Flp-In T-REx HEK293 cells were seeded in 60-mm plates. Subconfluent cells were transfected with plasmids encoding zinc finger nucleases targeting FYCO1 (Sigma-Aldrich). After 5 days, cells were sorted singularly in 96-well plates using a FACSAria cell sorter (BD Biosciences). Genomic DNA was isolated from each subsequent cell line and probed for nuclease cleavage with the Surveyor mutation detection kit (Transgenomic). Cleavage was visualized by polyacrylamide-Tris borate-EDTA gel electrophoresis. Cell lines positive for nuclease cleavage were subjected to Western blotting and probed with antibodies against FYCO1.
GST Pulldown Experiments-All GST-tagged proteins were expressed in Escherichia coli SoluBL21 (Genlantis). GST fusion proteins were purified on glutathione-Sepharose 4 Fast Flow beads (GE Healthcare 17-5132-01). 35 S-labeled Myc-tagged proteins were synthesized in vitro using the TNT T7-coupled reticulocyte lysate system (Promega). Translation reaction products from 0.25 g of plasmid DNA were incubated with GST-labeled proteins on glutathione-Sepharose beads in NETN-E buffer (50 mM Tris, pH 8.0, 100 mM NaCl, 1 mM EDTA, 0.5% Nonidet P-40) supplemented with cOmplete Mini EDTA-free protease inhibitor mixture tablets (1 tablet/10 ml) (11836170001, Roche Applied Science) for 1 h at 4°C. The beads were washed five times with 400 l of NETN-E buffer, boiled with 2ϫ SDS-PAGE gel loading buffer with 1 mM DTT, and subjected to SDS-PAGE. Gels were stained with Coomassie Brilliant Blue and vacuum-dried. 35 S-labeled proteins were detected using a Fujifilm bioimaging analyzer BAS-5000 (Fuji), and quantifications were performed using the Image Gauge software (Fuji).
Protein Expression and Purification for X-ray Crystallography-LC3B was expressed from pGEX4T plasmid in BL21 (DE3). LC3 was expressed as a GST-tagged protein. The protein was purified using glutathione-Sepharose 4B, cation exchange, and gel filtration chromatography. The GST moiety was proteolytically removed by thrombin protease. The protein solution was concentrated to 21.2 mg/ml by ultrafiltration in 25 mM Tris-HCl (pH 7.5) and 1 mM dithiothreitol. The concentrated LC3 was mixed with a peptide representing residues 1276 -1288 from FYCO1 with 1:2 molar ratio and incubated for 24 h at 4°C.
Crystallization and Data Collection, Structure Determination, and Refinement-Crystals of LC3-FYCO1 peptide complex were obtained by the sitting-drop vapor diffusion method at 293 K in drops containing a mixture of 1 l of protein and the 1 l of reservoir solution, which consisted of 0.1 M potassium thiocyanate and 30% w/v polyethylene glycol monomethyl ether 2000. Crystals were flash-cooled in a nitrogen gas stream. X-ray diffraction data sets for LC3-FYCO1 peptide complex were collected at 100 K on beamline BL44XU (SPring-8, Hyogo, Japan). Data processing and reduction were carried out with HKL2000 (29). The crystals belonged to space group P2 1 with one molecule in the asymmetric unit. Data collection, phasing, and refinement statistics are summarized in Table 1. The structure of the LC3-FYCO1 peptide complex was determined by molecular replacement using MOLREP (30) with LC3 (Protein Data Bank (PDB) ID code 1UGM) (31) as a search model. Models were subsequently improved through an alternate cycle of manual rebuilding using COOT (32), and refinement with the program REFMAC5 (33). The final refined model consists of residues 5-123 of LC3 in the asymmetric unit. For the FYCO1 peptide, the density allowed building on residues 1277-1288 complexed to the molecule. Refinement statistics are summarized in Table 1. There are no residues in disallowed regions of the Ramachandran plot. Structure figures were generated using PyMOL.
SPOT Synthesis and GST Overlay Assays-Peptides were synthesized on cellulose membranes using a MultiPrep automated peptide synthesizer (INTAVIS Bioanalytical Instruments AG, Cologne, Germany), as described previously (34). Membranes were blocked in Tris-buffered saline-Tween 20 with 5% nonfat dry milk and probed by overlaying with GST fusion of LC3B at 1 g/ml for 2 h at room temperature. Membranes were washed in Tris-buffered saline-Tween 20, and bound proteins were detected with HRP-conjugated anti-GST antibody (1:5000; clone RPN1236; GE Healthcare).

FYCO1 Contains a C-terminally Extended LIR Motif-We
have previously reported that FYCO1 contains a LIR motif that we functionally mapped to the 19-amino acid-long region 1276 -1294, located between the FYVE and GOLD domains (21). This region contains the sequence FDII confined to the core LIR motif consensus sequence (W/F/Y)XX(L/I/V) (5). To test whether this is the core LIR motif of an F-type LIR, we mutated the Phe 1280 and Ile 1283 residues to alanine and performed pulldown assays with GST-LC3B bound to beads and in vitro translated Myc-tagged, full-length WT and the F1280A/ I1283A double mutant. As seen in Fig. 1A, the LIR mutant completely lost binding to GST-LC3B. To identify the minimal peptide sequence required for efficient binding to LC3B, we employed a peptide array strategy with 18-mer peptides spanning amino acids 1265-1298 of FYCO1. The peptide walk was done with steps of one amino acid to allow high resolution mapping (10). Interestingly, the minimal binding motif was determined to encompass the octapeptide FDIITDEE starting with the aromatic residue of the core motif, but extending 4 amino acids C-terminal to the core (Fig. 1B). Weak binding was also seen for peptides shortened from the C-terminal end down to the conserved hydrophobic position of the core motif (here represented by Ile 1283 ). However, strong, binding required the four-amino acid C-terminal extension. In comparison, for ULK1 and ATG13, we have previously defined the minimal motifs to the pentameric sequences DFVMV and DFVMI, respectively (10). To study the extent of the functional LIR motif, and to assess the relative importance of specific residues involved in the interaction, we performed a peptide array with two-dimensional amino acid substitutions of peptides covering the region between 1276 and 1293 (Fig. 1C). This approach confirmed the C-terminally extended LIR motif and even suggested that Leu 1288 is part of a nonameric LIR sequence. Strikingly, Glu 1287 can only be substituted with another acidic residue, Asp, to restore binding. The most C-terminal Leu 1288 is only successfully replaced by the hydrophobic, Ile, Val, or Phe residues. As expected, the conserved aromatic Phe 1280 in the core LIR can only be substituted with the two other aromatic residues, Trp and Tyr. Similarly, in the hydrophobic position of the core LIR only Leu or Val, and to some extent Phe, can replace Ile 1283 . There is also a tendency for preference of acidic charge in the residue immediately C-terminal to Phe 1280 . As noted previously for the ATG13 and ULK1 LIRs and apparent from alignments of known LIR motifs (10), Gly or Pro residues affecting secondary structure were not tolerated at the core positions. Basic Lys or Arg residues were not tolerated, either (Fig. 1C). In conclusion, whereas the previously identified canonical LIR motifs are 4 amino acids long, FYCO1 contains a C-terminally extended 9-amino acid-long F-type LIR motif with an acidic residue and a hydrophobic residue at positions 8 and 9 important for efficient binding to LC3B.
FYCO1 Interacts Preferentially with LC3A and -B-When a peptide spanning FYCO1 amino acids 1276 -1294 was fused to the C terminus of GFP, in vitro translated and assayed against all human ATG8 homologues in a GST pulldown assay, only LC3A and -B bound strongly to the FYCO1 LIR ( Fig. 2A, upper panel) (10). In vitro translated full-length FYCO1 also shows preference for LC3A and -B in GST pulldown assays ( Fig. 2A, lower panel), although we see some binding to LC3C and GABARAP in vitro. However, we have earlier observed a stronger tendency for unspecific binding by GABARAP in such assays (10). The preferential binding to LC3A and -B was confirmed in vivo by co-immunoprecipitation experiments of Myctagged FYCO1 and GFP fusions of human ATG8 family proteins expressed in HEK293 cells (Fig. 2B). GFP-LC3A and GFP-LC3B were efficiently co-precipitated with Myc-FYCO1, whereas GFP-LC3C, GFP-GABARAP, GFP-GABARAPL1 and -L2 were not. We conclude that FYCO1 preferentially binds to LC3A and -B via its C-terminally extended F-type LIR motif. Consistently, we could also show that endogenous FYCO1 colocalized with endogenous LC3A/B in cells (Fig. 2C).  Structure of the LC3B-FYCO1 LIR Complex-To further explore the molecular mechanism responsible for the preferred binding of FYCO1 to LC3A and -B, we determined the struc-ture of the LC3B-FYCO1 LIR complex by x-ray crystallography. The complex consists of full-length LC3B-(1-125) bound to a 13-amino acid LIR peptide of FYCO1 (residues 1276 -1288). The crystal structure of the LC3B-FYCO1 LIR complex was determined by molecular replacement using the crystal structure of the LC3B monomer (residues 5-120, PDB code 1UGM), and refined to 1.53 Å resolution. The structure of LIR-bound LC3B (Fig. 3, A and B), which consists of a five-stranded ␤-sheet and five ␣-helices, is essentially identical to the previously reported structures of peptide-free LC3B and p62 LIR-bound LC3B (11,31,35); these structures have an average 0.7 and 1.0 Å root mean square deviation for the C-␣ positions, respectively. FYCO1 LIR binds within the LC3 groove in an extended conformation and forms a short 3 10 helix at the C-terminal containing residues Asp 1285 to Glu 1287 (Fig. 3, A and B). The FYCO1-LIR-binding surface of LC3B consists of two loops (␤1-␤2 and ␤2-␣3), two ␣-helices (␣2 and ␣3), and two ␤-strands (␤1 and ␤2). The side chains of the core FYCO1-LIR residues (Phe 1280 and Ile 1283 ) are bound deeply into the two hydrophobic pockets (HP1 and HP2) of LC3B (Fig. 3C), similar to that observed previously for other canonical LIR interactions (5,36). Moreover, the LC3B-FYCO1 LIR complex structure displays three remarkable electrostatic interactions: 1) Arg 10 of LC3B with Asp 1277 of FYCO1 (Fig. 3D); 2) Arg 70 of LC3B with Asp 1281 and Glu 1287 of FYCO1 (Fig. 3, E and F); and 3) His 57 of LC3B with Asp 1285 of FYCO1 (Fig. 3G). These would potentially form a series of salt bridges between the two molecules. In particular, the structure suggests that Arg 70 plays an important role for the interaction between LC3B and FYCO1. However, Arg 70 is conserved in GABARAP, -L1, -L2, LC3A, -B, and -C and cannot therefore explain the preferential binding of FYCO1 to LC3A and -B. It should be noted that the conformation of Arg 70 in the LC3B-FYCO1 LIR complex is different from that of the LC3B-p62 LIR complex (Fig. 3F). Arg 70 does not form strong interactions with the acidic residues of the p62 LIR. Contrary to Arg 70 , His 57 is only conserved in LC3A and -B. Hence, the His 57 residue is likely also contributing to the specific interaction of FYCO1 with LC3A and -B. Finally, the most C-terminally located important residue of the FYCO1 LIR determined from the two-dimensional peptide array mutational analysis, Leu 1288 , makes hydrophobic interactions with Pro 55 , Val 58 , and Ile 66 of LC3B (Fig. 3H). A sequence alignment of human ATG8 proteins shows that His 57 is found only in LC3A and -B and that Arg 10 is not present in GABARAP proteins (Fig. 3I).
The C-terminally Extended FYCO1 LIR Is Required for Both Binding Strength and Selective Binding to LC3A/B-Analyses of more than 40 verified LIR motifs show that the LIR motifs (W/F/Y)XX(L/I/V) frequently contain acidic residues (Glu or Asp) or phosphorylatable residues (Ser or Thr) N-or C-terminally (or both) to the core aromatic residue (5). For the FYCO1 LIR, there is no acidic residue directly N-terminal to the core aromatic Phe 1280 residue. However, mutation of Asp 1277 engaging in an electrostatic interaction with Arg 10 in the N-terminal arm of LC3B (Fig. 4C), and also its neighbor Asp 1276 , had a negative effect on binding. This was particularly evident for the D1276A/D1277A double mutant, as assayed by GST pulldown assays (Fig. 4, A and B). The importance of the electrostatic

FYCO1 Contains an LC3-preferring, C-terminally Extended LIR
interaction between LC3B Arg 10 and Asp 1277 revealed from the structure was confirmed by GST pulldown analyses showing that mutation of either Arg 10 or Asp 1277 to Ala had a similar negative effect (50% of WT), whereas mutation of both residues to Ala reduced the binding proportionally to about 25% of WT (Fig. 4, C and D). The binding of D1276A and D1277A mutants varied in several experiments. A possible explanation may be that they compensate for each other. This may explain why the double mutant had a more dramatic effect than the single mutants.
Directly C-terminal to Phe 1280 lies also the acidic Asp 1281 residue. The charge of this residue is not absolutely crucial for the LC3B binding because the D1281A mutant still retained binding in the pulldown assay (Fig. 4, E and F). However, the two-dimensional peptide array clearly suggested that Asp or Glu is preferred at this position (Fig. 1C). The requirement for the C-terminally extended LIR and crucial role of Glu 1287 is clearly seen because the E1287A mutant was almost as detrimental as the double mutant of the core LIR F1280A/I1283A (Fig. 4, A and B). The L1288A mutant also had a clear negative  effect on the binding, confirming its importance as the most C-terminal residue in the 9-amino acid-long FYCO1 LIR (Fig. 4,  A and B). It is also likely that the Ala mutant is the least severe mutation in this position as judged from the peptide array and structure of the LC3B-FYCO1 LIR complex (Figs. 1C and 3H). The crystal structure shows that both Asp 1281 and Glu 1287 engage in electrostatic interactions and form hydrogen bonds with Arg 70 in the LIR docking site of LC3B (Fig. 3, E and F). The R70A mutant of LC3B completely abolished binding to FYCO1 in the GST pulldown assays. The D1281A mutant was less affected, whereas E1287A only retained minimal binding and no binding at all was measured with LC3B R70A versus FYCO1 E1287A (Fig. 4, E and F). Because FYCO1 Asp 1285 and LC3B His 57 engage in an electrostatic interaction, we tested the importance of this interaction by pulldown assays. His 57 was mutated to Asp, which is present in the corresponding position of GABARAP family proteins. An ϳ50% reduction in binding was observed for both FYCO1 D1285A versus WT LC3B and -B H57D versus FYCO1 WT (Fig. 4, G and H). Taken together, the mutational analyses show the Glu 1287 residue of the extended LIR motif to be essential for binding to LC3B via its interaction with Arg 70 , and His 57 is essential for the preferred binding to LC3A/B.

FYCO1 Contains an LC3-preferring, C-terminally Extended LIR
Generation of Flp-In T-REx HEK293 Cells Knocked Out for FYCO1-We employed the zinc finger nuclease strategy targeting the large exon 8 of the human FYCO1 gene on chromosome 3p21.3 to generate a HEK293 Flp-In T-REx cell line lacking expression of FYCO1 (Fig. 5A). The nuclease cut is at amino acid position 295 of the coding sequence. The presence of a deletion in the FYCO1 gene was verified using the surveyor kit from Transgenomic, Inc. The expected bands with sizes of 310, 182, and 128 bp were present, confirming cleavage by the CEL-1 enzyme (Fig. 5B). Loss of FYCO1 protein expression was confirmed by Western blotting with an antibody against FYCO1 (Fig. 5C). Three cell lines lacking FYCO1 expression were generated. We chose clone 12 for our further studies and denoted it FYCO1 KO.
A Functional LIR Motif of FYCO1 Is Required for Efficient Maturation of Autophagosomes-The generation of the FYCO1 KO cells allowed us to stably express inducible mutants of FYCO1 in a FYCO1 null background. FYCO1 KO cells were reconstituted with tetracycline-inducible Flp-In GFP-FYCO1 WT and the F1280A/I1283A LIR mutant (LIRmut) constructs (Fig. 5D). Without tetracycline induction, the expression levels of GFP-FYCO1 WT and GFP-FYCO1 LIRmut constructs were comparable with that of endogenous FYCO1 in the parental Flp-In T-REx HEK293 cells (Fig. 5D).
The interaction of FYCO1 with LC3 is essential for its role in the maturation of LC3-positive phagosomes into LAMP1-positive structures (24). Because the LIR motif is also used for binding of FYCO1 to autophagosomes (21), we next examined the subcellular localization pattern of GFP-FYCO1 WT and GFP-FYCO1 LIRmut in the reconstituted KO cell lines without the interference of endogenous FYCO1. To be able to clearly see the GFP signal in these cells, we induced the expression with tetracycline. WT and LIRmut GFP-FYCO1 both localized to clusters of relatively large vesicular structures (Fig. 5E), likely resulting from homotypic fusion events. FYCO1 is localized on the ring perimeters, and the vesicles vary in diameter from 2 m in HeLa cells to 15 m in HEK293 cells. Also, the co-localization with LAMP1 was unchanged for WT and LIR mutant GFP-FYCO1 (data not shown). However, both the co-localization of FYCO1 with LC3 and that of LC3 with LAMP1 were dramati-  DECEMBER 4, 2015 • VOLUME 290 • NUMBER 49 cally reduced in the LIRmut reconstituted cells relative to the WT reconstituted cells (Fig. 6, A and B). This was the case both in full medium and upon serum and amino acid starvation for 2 h. When the cells were treated with bafilomycin A1, to inhibit fusions between autophagosomes and lysosomes and the lysosomal acidification, the differences between FYCO1 WT and LIRmut were less prominent. This is likely due to accumulation of LC3-positive autophagosomes that co-cluster with FYCO1positive vesicles in a LIR-independent manner.

FYCO1 Contains an LC3-preferring, C-terminally Extended LIR
We then compared the levels of the lipidated LC3-II form in the four different cell lines by Western blotting. For these experiments, the reconstituted cell lines were not induced with tetracycline to keep the expression levels as close to endogenous levels as possible. Both in the FYCO1 KO cell line and in the LIRmut cell line, LC3-II levels were increased in the full medium situation (Fig. 7, A and B). This may suggest either an increase in formation of lipidated LC3 or an inhibition of clearance. To distinguish between these possibilities, we treated the cells with bafilomycin A1 for 6 h to block lysosomal acidification and halt degradation. This resulted in a lower increase in LC3-II levels in the KO and LIRmut cell lines as compared with parent HEK293 and KO cells reconstituted with GFP-FYCO1 WT (Fig. 7, A and B). Therefore, the increased LC3-II level in full medium was not caused by increased formation of LC3-II, but rather inhibition of clearance. To visualize this effect on autophagic flux more clearly, we calculated the amount of LC3-II accumulated upon bafilomycin A1 treatment by subtracting the LC3B-II level without bafilomycin A1 from the LC3-II level in the bafilomycin A1-treated sample. Strikingly, in cells lacking FYCO1, the flux was greatly inhibited, seen as

FYCO1 Contains an LC3-preferring, C-terminally Extended LIR
strongly reduced accumulation of LC3-II. The same effect was also seen in FYCO1 KO cells reconstituted with GFP-FYCO1, either WT or LIRmut (Fig. 7C). Additionally, we quantified the amount of endogenous LC3 puncta in the uninduced cells. Here, we found that for the LIRmut cells, the number of puncta in the full medium situation was higher than for the WT cells. Because the number of LC3 puncta of FYCO1 WT and LIRmut cells was equal upon bafilomycin A1 treatment, we conclude that there is a maturation defect in the LIRmut cells (Fig. 8). Taken together, these data indicate that, in the full medium situation, the LIRmut cells have a reduced ability to clear autophagosomes.

Discussion
We showed previously that FYCO1 induces microtubule plus end transport of LC3-positive autophagic vesicles (21). In that study, we also noted that FYCO1 has a canonical F-type core LIR motif with a phenylalanine at the aromatic position and an isoleucine at the hydrophobic position. The LIR motif is generally defined as having the core consensus sequence (W/F/ Y)XX(L/I/V) (5). Here, we show that FYCO1 preferentially interacts with LC3A and -B, and that the functional LIR extends with five residues C-terminal to the hydrophobic position of the core LIR. Detailed analyses, including data obtained from twodimensional peptide array-based mutation scanning, x-ray crystallography, and mutation analyses by GST pulldown assays, allow us to firmly conclude that the C-terminal extension is involved in determining both binding strength and specificity toward LC3A and -B. In particular, Asp 1285 of FYCO1 provide specificity toward binding to LC3A and -B by interacting with His 57 in LC3B. The corresponding residue Asp 54 of GABARAP, GABARAPL1 and -L2, and Glu 63 of LC3C would lead to charge repulsion and act unfavorably for interaction with Asp 1285 of FYCO1. ALFY (autophagy-linked FYVE protein, also called WDFY3) is a large phosphatidyl-inositol 3-phosphate-binding protein interacting with the autophagy receptors p62/SQSTM1 and NBR1 and helps to clear ubiquitinated protein aggregates by autophagy (37,38). ALFY binds preferentially to the GABARAP subfamily of ATG8 proteins (18). This specificity is in part mediated by a tyrosine residue (Tyr 3351 ) C-terminal to the core LIR at the corresponding positon of Asp 1285 in FYCO1. Tyr 3351 interacts with Asp 54 in GABARAPs, but is sterically hindered by His 57 of LC3B (18). Hence, in both these LIRs, a residue located two residues C-terminal to the conserved hydrophobic position of the core LIR is involved in distinguishing between the LC3A and -B and GABARAP subfamilies.
Many LIR motifs contain acidic residues N-terminally to the core aromatic residue that engage in electrostatic interactions with basic residues in the N-terminal arm of ATG8s (5). Asp 3344 of the ALFY-LIR is able to form ionic interactions with Lys 24 and Tyr 25 of GABARAP, but not with the corresponding Gln 26 and His 27 of LC3B. This way these residues also contribute to the preferential binding to GABARAPs (18). For FYCO1, Asp 1277 binds to Arg 10 in the N-terminal arm of LC3B, and a double mutation of both Asp 1277 and Asp 1276 reduced binding to LC3B by 80%. Clearly, these two N-terminal residues are important for the binding affinity and may also impact on the specificity as seen for the ALFY-LIR.
The most dramatic mutation affecting the binding of FYCO1 to LC3B, apart from the core LIR F1280A/I1283A double mutation, was the E1287A mutant. This residue is located four residues C-terminal to the conserved core hydrophobic residue and interacts electrostatically with Arg 70 in LC3B. The E1287A mutation resulted in an 85% reduction in binding. The crucial importance of this ionic interaction is underscored by the peptide array scan showing that the only allowed substitution was to the other acidic residue, i.e. E1287D. Because the Arg 70 residue is conserved in GABARAPs too, this is not an interaction contributing to the specificity, but it is crucial for binding strength for the FYCO1-LC3B interaction. Our two-dimensional mutation analysis by peptide array scan also showed that Leu 1288 , the most C-terminal residue interacting with LC3B, can only be substituted with the other hydrophobic residues, Ile, Val, and Phe, and this residue interacts with other hydrophobic residues in LC3B. However, based on sequence conservation between ATG8 family members at these positions, it is not likely that these interactions contribute to the binding preference.
The presence of at least six different human ATG8 homologues is intriguing, as yeast has only one Atg8 and Caenorhabditis elegans and Drosophila have two. The degree of redundancy relative to specific individual roles for the mammalian  DECEMBER 4, 2015 • VOLUME 290 • NUMBER 49 ATG8 family members is only beginning to be studied. Based on knockdown and overexpression studies in HeLa cells, a division of labor between LC3B for initiation and GABARAPL2 for completion of autophagosomes has been proposed (8). On the other hand, studies in C. elegans indicate that the GABARAPlike ATG8 homologue LGG-1 is required for autophagosome formation and the LC3 homologue LGG-2 is required for fusion with the lysosome (39). In knockdown studies, the LC3 subfamily has been found to be dispensable for starvation-induced autophagy in hepatocytes and prostate cancer cells, whereas the GABARAP subfamily is required (40). This, as well as the fact that most autophagy studies in mammals have had a focus on LC3B, has motivated us to ask whether LIR-containing proteins have any specificity toward any of the six human ATG8 homologues. A handful of studies have identified LIR-containing proteins with a preference for the GABARAP subfamily, such as the scaffold protein ALFY involved in selective autophagy (18), and proteins of the Unc-51 like kinase (ULK) complex (10). These proteins interact with ATG8s on the outer surface of the phagophore/autophagosome. Also, Rab effectors such as FYCO1 and Rab GTPase-activating proteins interact with LC3/ GABARAP on the outer surface (21,22,41,42). The preference of FYCO1 for the LC3 subfamily is therefore interesting, and FYCO1 is as far as we know the only LIR-containing protein with a preference for LC3A and -B. Another group of LIR-containing proteins are the autophagy receptors that interact with LC3/GABARAP on the inside of the forming phagophore, and end up being degraded by autophagy (reviewed in Refs. 4 and 5). Among the autophagy receptors, there are different binding preferences, arguing against the idea that a specific subset of ATG8 homologues is responsible for the attachment of cargos to the inner membrane. The autophagy receptor p62/SQSTM1 binds very well to all ATG8 family proteins in in vitro binding studies. However, for selective autophagy of p62/SQSTM1, it has been found that LC3B is required and not the GABARAP subfamily (43,44).

FYCO1 Contains an LC3-preferring, C-terminally Extended LIR
We previously showed that FYCO1 uses its LIR motif to bind to autophagosomes (21). Knockdown of FYCO1 in HeLa cells stably expressing GFP-LC3B leads to a perinuclear clustering of GFP-LC3B under basal conditions (21). This indicated a role for FYCO1 in plus end-directed transport of phagophores and/or autophagosomes. To look more specifically at the importance of the LIR motif, in this study, we made FYCO1 knock-out cell lines stably expressing wild type or LIR mutated FYCO1. Our data confirmed that the LIR motif is required for co-localization of FYCO1 with autophagosomes, and expression of a LIR

FYCO1 Contains an LC3-preferring, C-terminally Extended LIR
mutant construct negatively affected maturation of autophagosomes under basal conditions. The effects we see on late steps of basal autophagy on mutation of the FYCO1 LIR motif correlate with a role of FYCO1 in regulating kinesin-mediated transport of LC3-positive autophagic structures. The effects we observed by mutating the FYCO1 LIR motif were seen only for cells grown in full medium. We have recently shown that FYCO1 interacts with kinesin 1 (26), and it has been demonstrated that anterograde movement of autophagosomes is dependent on kinesin 1 only when cells are grown in full medium (45). This supports our finding that we see the effect on maturation in basal autophagy when cells are fed, and not during starvation.