The AAA + ATPase valosin-containing protein (VCP)/p97/Cdc48 interaction network in Leishmania

Valosin‐containing protein (VCP)/p97/Cdc48 is an AAA + ATPase associated with many ubiquitin-dependent cellular pathways that are central to protein quality control. VCP binds various cofactors, which determine pathway selectivity and substrate processing. Here, we used co-immunoprecipitation and mass spectrometry studies coupled to in silico analyses to identify the Leishmania infantum VCP (LiVCP) interactome and to predict molecular interactions between LiVCP and its major cofactors. Our data support a largely conserved VCP protein network in Leishmania including known but also novel interaction partners. Network proteomics analysis confirmed LiVCP-cofactor interactions and provided novel insights into cofactor-specific partners and the diversity of LiVCP complexes, including the well-characterized VCP-UFD1-NPL4 complex. Gene Ontology analysis coupled with digitonin fractionation and immunofluorescence studies support cofactor subcellular compartmentalization with either cytoplasmic or organellar or vacuolar localization. Furthermore, in silico models based on 3D homology modeling and protein–protein docking indicated that the conserved binding modules of LiVCP cofactors, except for NPL4, interact with specific binding sites in the hexameric LiVCP protein, similarly to their eukaryotic orthologs. Altogether, these results allowed us to build the first VCP protein interaction network in parasitic protozoa through the identification of known and novel interacting partners potentially associated with distinct VCP complexes.

www.nature.com/scientificreports/ their mammalian orthologs. Altogether, these results allowed us to build the Leishmania VCP protein network, the first characterized in parasitic protozoa.

Results
Identification of the Leishmania VCP/p97 cofactors and core proteome. VCP/p97 is a hexameric protein that can interact with a large number of protein cofactors through its N-terminal and C-terminal domains 27 . These interactions are key for its activity and functional diversity. This study aimed to identify the Leishmania VCP interactome based on the human VCP/p97 network from the BioGrid database (Supplementary Table S1) as well as on protein-protein interaction experiments. According to the e-value similarity accessed by protein BLAST (blastP), most proteins from the human VCP network have homologs in Leishmania ( Supplementary Fig. S1).
Seeking for the identification of Leishmania VCP cofactors, we carried out a series of immunoprecipitation (IP) experiments with either C-or N-terminally HA-tagged Leishmania infantum VCP (LiVCP; LinJ.36.1420) 10 followed by LC-MS/MS, database searching, and peptide identification. To allow VCP expression from its endogenous locus, we replaced one LiVCP genomic copy by a HA-epitope tagged LiVCP (LiVCP-HA) (Fig. 1A). Both episomal and integrated LiVCP-HA versions were expressed at levels comparable to the endogenous VCP protein (Fig. 1B) and yielded similar IP results. From seven independent IP-MS/MS experiments (see Methods), 218 proteins were initially identified as potential LiVCP interacting partners (Supplementary Table S2). To decrease the number of false positives in the LiVCP network building, we applied a stringent filter. First, we calculated the average number of peptides for each protein in all seven IPs and then we excluded proteins also found in five independent IP-MS/MS studies conducted with HA-tagged proteins unrelated to the VCP complex ( Fig. 1C and Supplementary Table S2). From this analysis, 24 proteins were selected to specifically associate with LiVCP. Among those, 15 proteins with an average of exclusive unique peptide count (EUPC) ≥ 2.0 were classified as the LiVCP core partners ( Fig. 1D and Supplementary Table S2). These share conserved domains with known VCP cofactors in other eukaryotes, such as the p47 (Shp1) UBX/UBA-and SEP-containing protein (LinJ. 22.0200), the UBX-containing protein FAF (Fas-associated factor) shown a higher homology to the human FAF2 protein (UBXD8 or ETEA; LinJ. 35.1960), the ubiquitin fusion degradation protein 1 (UFD1; LinJ. 36.6780), the nuclear protein localization 4 (NLP4; LinJ. 25.1320) known to form a heterodimer with UFD1 34 , an ubiquitin associated UBA/TS-N domain protein (LinJ. 24.1650) with an identified ER membrane signal peptide, and two PUB (PNGase/UBA or UBX) domain containing proteins (LinJ.11.0920 and LinJ.09.1060). In addition, the LiVCP proteome core includes the serine/threonine phosphatase type 1 (PP1) (LinJ. 34.0900), the PP1 regulator-like protein (LinJ.05.1200) whose human homolog SDS22 is known to interact with VCP 36 , three mitochondrial proteins-a stomatin-like protein (LinJ.05.1040), an ATP-dependent zinc metallopeptidase (LinJ. 36.2850) and the AAA + FtsH protease (LinJ. 36.2850), a tetratricopeptide repeat protein part of the ER membrane protein complex subunit 2 (LinJ. 17.0230) and three hypothetical proteins (LinJ. 36.5080, LinJ.03.0250, LinJ.15.1570) ( Fig. 1D) with some similarity to the yeast nuclear envelope protein Nsp1-like C-terminal region, syntaxin-like domain, and putative nuclear jumonji-like domain, respectively.
Apart from the above core partners, LiVCP co-immunoprecipitated heat shock proteins, chaperonins, T-complex proteins, mitochondrial integrity and stress response proteins, translation factors and ribosomal proteins, components of the trypanothione and peroxidase systems, RNA-binding proteins, and proteasome subunits (Supplementary Table S2) in line with its function in a broad array of ubiquitin-dependent protein quality control pathways 13-16 . 'Network proteomics' analysis to identify interacting partners of the core LiVcp cofactors and associations between cofactor-bound proteins. The interaction of p47, UFD1, NPL4, FAF2 and PUB1 with the Leishmania VCP ortholog was further confirmed by inverse co-immunoprecipitation studies. C-or N-terminally HA-tagged Lip47, LiNPL4, LiUFD1, LiFAF2, or LiPUB1 proteins cloned into vector pSP α ZEO α were transfected and stably expressed in L. infantum promastigotes ( Supplementary Fig. S2A). Immunoprecipitation of these HA-tagged cofactor versions using an anti-HA antibody followed by western blotting with a specific anti-TbVCP antibody 10,37 confirmed their association with LiVCP ( Fig. 2A). The co-immunoprecipitation of LiUFD1, LiNPL4, Lip47, LiFAF2 and LiPUB1 with LiVCP ( Fig. 1D and Table 1) and the detection of LiVCP in LiUFD1, LiNPL4, Lip47, LiFAF2 and LiPUB1 immunoprecipitates by western blotting ( Fig. 2A) and mass spectrometry (Fig. 2B, Table 1, Supplementary Table S2) confirmed their close partnership.
Lip47 co-immunoprecipitated LiPUB1, the serine/threonine protein phosphatase 1 (PP1) catalytic subunit beta (LinJ.28.0730), a PP1 regulator-like protein that is homologous to the yeast SDS22 protein (LinJ.05.1200), and the PP1 phosphatase inhibitor that is homologous to the yeast Ypi1 protein (LinJ.07.0840) ( Table 1). In Saccharomyces cerevisiae, these three proteins form a ternary complex that is important for the nuclear localization of PP1 and whose assembly and quality control requires Cdc48 and its adaptor p47 38 .
As reported in other eukaryotes 34 , the Leishmania LiUFD1 protein complexes with NPL4 to form the heterodimer cofactor UFD1-NPL4 but also associates with LiFAF2 (Table 1). Complex formation between VCP-NPL4-UFD1, FAF1 and polyubiquitinated proteins was shown previously to promote ER-associated degradation 39 . Other proteins only associated with the LiUFD1 proteome include a mitochondrial serine carboxypeptidase (CBP1; LinJ.18.0450), a phenylalanine-4-hydroxylase (LinJ. 28 1 and Supplementary Table S2). Interestingly, the hypothetical protein LinJ.03.0250 was also found in LiVCP immunoprecipitates (Fig. 1C), and some of the other LiUFD1 interacting proteins have homologs known to be part of the VCP network in other systems (see Discussion). Proteins solely detected in LiFAF2 pull-down include a tetratricopeptide repeat containing protein of the ER membrane protein complex (LinJ.17.0230) also found in LiVCP immunoprecipitates (Fig. 1C), several   (Table 1). The LiPUB1 cofactor co-immunoprecipitated a serine palmitoyltransferase 1-like protein (LinJ.34.3530; SPT1), also found in the LiUFD1 proteome (Table 1). Serine palmitoyltransferase catalyzes the first and ratelimiting step in sphingolipid (ceramide) biosynthesis and has been associated with the resistance to heat stress and apoptosis 40,41 .
General features of the LiVCP core cofactors. Proteins specifically co-immunoprecipitated with LiUFD1, Lip47, LiFAF2, or LiPUB1 were submitted to Gene Ontology -cellular component (GO-CC) enrichment analysis using the tool provided by TriTrypDB ( Fig. 3B and Supplementary Table S3). An enrichment of a 19-fold for vacuolar proteins was seen in the LiUFD1 proteome. Nuclear proteins were exclusively enriched within the Lip47 proteome and endoplasmic reticulum-related proteins were enriched only within the LiFAF2 and LiPUB1 proteomes.
Molecular function (MF) and biological process (BP) analyses were also done aside comparison with the LiVCP proteome (Supplementary Table S3). Globally, the molecular function analysis predicted proteins with translation factor activity and structural constituents of the ribosome, RNA-binding activity, structural molecule activity and peptidase activity. ATPase and ligase activities were exclusively enriched within the LiFAF2 proteome while transmembrane transporter activity was only associated with the LiUFD1 proteome.  www.nature.com/scientificreports/ The biological process analysis showed that cofactor-associated proteins were mostly involved in translation, protein targeting and folding, and several metabolic processes. Protein folding and tRNA metabolic processes were found only associated with the LiFAF2 proteome while protein targeting and mitotic cell cycle were enriched Table 1. Proteins specifically associated with one or more Leishmania LiVCP cofactors as determined by immunoprecipitation and LC-MS/MS studies. Proteins in bold represent the core LiVCP cofactors. Proteins indicated with an asterisk (*) are also found in LiVCP immunoprecipitates. TM: transmembrane helix. Average of exclusive unique peptide count (EUPC) is from seven independent IP experiments. www.nature.com/scientificreports/ in the LiPUB1 proteome (Supplementary Table S3). Overall, Gene Ontology analysis of each proteome indicated specific compartment-association for each cofactor.

Subcellular localization of the LiVCP core cofactors.
To determine the subcellular localization of the core LiVCP cofactors, we carried out digitonin fractionation and indirect immunofluorescence studies. As a central player to the endoplasmic reticulum (ER)-associated protein degradation 19,48 and the translocation of damaged mitochondrial proteins from the outer mitochondrial membrane (OMM) into the cytosol 20,49 , VCP has been found associated with both ER and OMM fractions, similarly to what we have described previously for the Leishmania LiVCP 10 . First, we treated Leishmania parasites with increasing concentrations of digitonin (20 μM-10 mM) and carried out western blotting on the different cellular sub-fractions using an anti-HA antibody (all cofactors were HA-epitope tagged). Our results indicate an association of LiFAF2 and partly of LiPUB1 proteins with the organellar fraction and an enrichment of LiUFD1, Lip47 and LiNPL4 proteins with the cytosolic fractions (Fig. 4A). www.nature.com/scientificreports/ Immunofluorescence studies using an antibody directed against the ER BiP protein confirmed that LiFAF2 and partly LiPUB1 co-localize to endoplasmic reticulum (Fig. 4B, 4C). No co-localization was found with the mitochondrion (data not shown). Immunofluorescence studies to detect Lip47 demonstrated a partial co-localization with the histone H3 nuclear marker (Fig. 4D), as also suggested by GO analysis (Fig. 3B).

Predicted molecular interactions between the conserved domains of LiVCP cofactors and the
Leishmania VCP protein based on 3D homology modeling and protein-protein docking. It has been shown previously that the VCP/p97/Cdc48 hexamer can bind to its various cofactors through the UBX or PUB domains or a ubiquitin-like fold or through one of the linear binding motifs such as SHP 32 . The L. infantum LiVCP encodes a protein of 784 amino acids that is highly conserved among its eukaryotic orthologs (75% aa identity with the human VCP/p97 protein) 10 and harbors all the characteristic domains of VCP orthologs (Fig. 5A, top). Based on sequence similarity and conservation to mammalian VCP cofactor orthologs (Supplementary Fig. S1) and on the EUPC for each cofactor in LiVCP immunoprecipitates (Fig. 1), we selected Lip47, LiUFD1, LiNPL4, LiFAF2, and LiPUB1 as the core LiVCP cofactors for further investigation. Multiple sequence alignments of these LiVCP cofactors along with mammalian, yeast, trypanosomatid and non-trypanosomatid species helped us to generate phylogenetic trees revealing that Lip47, LiFAF2, LiUFD1, LiNPL4 and LiPUB1 are evolutionary distant from other eukaryotes but they harbor binding modules/motifs that are conserved and could represent specific domains for interaction with LiVCP (Supplementary Figs. S3-S7).
To investigate whether LiVCP-cofactor interactions identified by immunoprecipitation studies take place within the previously predicted binding sites in the hexameric VCP protein 30,32 and the conserved binding modules in Lip47, LiFAF2, LiUFD1, LiNPL4 and LiPUB1, we constructed 3D models by homology modeling and a protein-protein docking approach. 3D homology models are helpful in refining protein-protein interaction predictions that have been based on a sequence match alone as interface binding is more directly determined by the structure of the binding site rather than its sequence. The LiVCP homohexamer 3D molecular structure was obtained through homology modeling by superposition with the H. sapiens p97 homohexamer (PDBid: 5C18) template (Fig. 5A, bottom).
The Lip47 protein presents the same domain composition (UBA, SEP and UBX) than its human ortholog (Fig. 5B, upper). Similarly to the human p47, Lip47 harbors the sequence GMPP in the UBX domain (Supplementary Fig. S8) that docks in silico into the Nn-Nc pocket of LiVCP ( Fig. 5B and Supplementary Fig. S9) 50 . Although Lip47 shares only 25% amino acid identity with its human ortholog, it has 41% sequence similarity and 73% sequence coverage and contains the two SHP motifs (SHP1 and SHP2) ( Supplementary Fig. S8) that together with the UBX domain are known to be key for the interaction with VCP 31 . The SHP site is typically characterized by the consensus sequence FxGxGx 2 h that was recently summarized as h(x) 1−2 F/W(x) 0−1 GxGx 2 L (h, hydrophobic residue; x, any amino acid) 32 and in Lip47, SHP1 and SHP2 sites, FYGRGQRL and FQGHGHRL, respectively are conserved ( Supplementary Fig. S8). In addition, the Ramachandran plot showed that the amino acid residues of the Lip47 3D homology model are found in most favoured (87%) and additional allowed regions (8.7%) (Supplementary Table S4), which makes structure predictions reliable.
The LiFAF2 protein interacts with LiVCP through the GFPP motif in its UBX C-terminal domain as predicted by 3D homology modeling (Fig. 5C, Supplementary Fig. S10) and protein-protein docking upon superposition of the available structure of the human 40 GYPP 43 motif in the ovarian tumor domain-containing protein 1 (OTU1) and the homologous part within the Leishmania FAF2 protein (Supplementary Fig. S11). Surprisingly, LiFAF2 lacks the ubiquitin-associated (UBA) domain that is present in the human FAF1 and FAF2 proteins and is fundamental for interacting with ubiquitinated substrates 51 .
UFD1 is known to assemble with NPL4 to form the heterodimer UFD1/NPL4 (U/N), one of the most important p97/VCP cofactors, that recognizes polyubiquitinated substrates and together with VCP extracts them from organellar membranes or macromolecular complexes 52,53 . U/N binds to the N-terminal domain of VCP via the ubiquitin-like domain (UBL) of NPL4 and the SHP box of UFD1 32 . The Leishmania LiUFD1 protein also contains the UBA domain and an UT6 region that accommodates the sequences QPTFAGA GRT L and RLKALGGGGRS corresponding to the SHP1 and SHP2 sites, respectively (Fig. 5D, top and Supplementary Fig. S12). 3D homology modeling predicts binding of LiUFD1 to LiVCP via its SHP1 site (Fig. 5D, bottom). Furthermore, docking essays propose that the SHP1 motif of LiUFD1 (reversed comparatively to the human SHP1) interacts with the same surface as in its human ortholog using amino acid residues Phe273, Arg278 and Leu280 (Phe228, Arg234 and Leu235 in human Ufd1) (Fig. 5D bottom and Supplementary Fig. S13).
In contrast to its eukaryotic orthologs, the Leishmania NPL4 lacks the N-terminal ubiquitin regulatory X (UBX)-like domain bound to the N-domain of Cdc48 hexamer, as well as the two Zn 2+ -finger domains important for anchoring the adjacent Mpr1/Pad1 N-terminal (MPN) domain to the top of the D1 ATPase ring 34 . Interestingly, protein-protein docking simulations showed that despite the absence of the UBX-like domain, LiNPL4 seems to interact with the N-terminus of LiVCP through its C-terminal GMPP motif (aa 350-353) (Supplementary Figs. S14, S15). Similarly to other eukaryotes 34 , the U/N heterodimer in Leishmania also recognizes polyubiquitinated substrates as shown by LiNPL4 and LiUFD1 IPs followed by western blotting (Supplementary Fig. S16).
A representative member of the PUB (PNGase/UBA or UBX-containing) domain proteins in Leishmania, LiPUB1, which has no homolog in humans, contains a conserved region (Supplementary Fig. S17) that is predicted by 3D homology modeling (Fig. 5E) and protein docking (Supplementary Fig. S18) to interact with the last five amino acids 780 DDLYS 784 of the LiVCP C-terminus, similarly to other PUB domain proteins 54 .
Altogether, despite the evolutionary distance of LiVCP cofactors from their mammalian orthologs, in silico models based on structural data from the human orthologs using 3D homology modeling and protein-protein docking indicated that the conserved binding modules of Leishmania VCP core cofactors (except LiNPL4) interact with specific binding sites in the hexameric LiVCP, similarly to their mammalian orthologs. www.nature.com/scientificreports/ Nuclear and kinetoplast DNA was stained with DAPI (blue). Endoplasmic reticulum (ER) putative co-localization was assessed using an anti-BiP antibody (red in B and C) as an ER marker. An Alexa Fluor 555 anti-rabbit was used as secondary antibody for BiP. Putative nuclear co-localization for Lip47 was assessed using an anti-histone H3 (H3) antibody (red in D).  6A) and allowed us to build the first LiVCP interaction network (Fig. 6B). To decrease the number of false positives, we excluded proteins that were co-immunoprecipitated with unrelated HA-tagged proteins (average of EUPC = 0.0). From the total number of proteins co-immunoprecipitated with Lip47, LiUFD1, LiFAF2, and LiPUB1 (Fig. 3A) only 10, 17, 23 and 9, respectively together with 24 proteins specifically associated to LiVCP (Fig. 1 , Table 1 and Supplementary Table S2) were considered to build the topology of LiVCP network (83 proteins in total, 47 shown in Fig. 6B). The list of proteins and their intersections are detailed in Supplementary  Table S5 whereas the complete list of interacting proteins is presented in Supplementary Table S2.
Our analysis revealed that the LiUFD1 and LiFAF2 proteomes share five proteins in common with LiVCP. These include the mitochondrial stomatin-like protein 2 (SPL-2) (EUPC > 20), the membrane-integrated mitochondrial AAA/FtsH protease 57 Fig. 6B). LiVCP and Lip47 mutually associate with the serine/threonine protein PP1 phosphatase catalytic subunit, the PP1 phosphatase inhibitor Ypi1, and the SDS22 (EUPC = 20) PP1 regulator-like protein (  Fig. 6B and Supplementary Table S5). Specific interactions between cofactor-associated partners that do not involve an association with LiVCP are described in detail in the 'network proteomics' section (see also Table 1 and Fig. 6B).
Overall, using combined datasets from multiple IP/MS-MS studies for LiVCP and its main cofactors, we unveiled the first Leishmania LiVCP interactome. Many of the LiVCP interactors have homologs that are known to interact with VCP in other eukaryotic systems but others do not and may be relevant to study further as part of distinct Leishmania VCP complexes.

Discussion
In this study, we have characterized the valosin-containing protein (VCP) interaction network, the first in parasitic protozoa. We have employed a series of co-immunoprecipitation and mass spectrometry analyses coupled with in silico models that allowed us to identify the Leishmania VCP proteome with its major cofactors and their interacting partners, as well as, to predict molecular interactions between conserved domains within these cofactors and specific binding sites in the hexameric LiVCP. Our data support several similarities but also important differences between the Leishmania VCP protein network and VCP complexes characterized in other eukaryotes.
Most of the known classes of VCP cofactors in other eukaryotes have orthologs in Leishmania. However, cofactors harboring the VIM (VCP-interacting motif)/VBM (VCP-binding motif) motifs 58 were not found in any of our immunoprecipitation experiments, and with the exception of the ERAD-associated E3 ubiquitin-protein ligase HRD1 (LinJ.15.1460) no homologs of this class of proteins were depicted in the Leishmania genome. While www.nature.com/scientificreports/ LiVCP is highly conserved among different eukaryotic organisms and harbors all the characteristic domains of the human p97/VCP ortholog 10 , most of the major LiVCP cofactors are phylogenetically distant from their eukaryotic orthologs. However, despite their evolutionary distance from yeast and mammalian orthologs, in silico models based on 3D homology modeling and protein-protein docking indicated that the conserved binding modules of Leishmania VCP cofactors interact with specific binding sites in the hexameric LiVCP, similarly to their eukaryotic orthologs. This is the case for Lip47 and LiFAF2 that bind the Nn-Nc pocket of LiVCP through the sequence GMPP/GFPP in their UBX domain. Surprisingly, LiFAF2 lacks the UBA domain and theoretically should have lost the ability of binding ubiquitinated substrates 59 . Similarly to its yeast ortrholog 34 , LiUFD1 www.nature.com/scientificreports/ interacts with the N-terminus of LiVCP through its conserved SHP motif. LiPUB1 while has no human homolog yet interacts with the C-terminal region of LiVCP, as described for the human VCP-PUB complex. One of the important differences unveiled by our study concerns LiNPL4, which lacks the UBX-like domain and the two Zn 2+ -finger motifs shown to bind VCP. Indeed, recent studies on the structure of the Cdc48-Ufd1-Npl4 complex from the thermophilic fungus Chaetomium thermophilum demonstrated that Npl4 interacts with the Cdc48 N-terminal region through its UBX-like domain and uses the Zn 2+ -finger motifs to anchor its MPN domain to the top of the D1 ATPase ring 34 . On the other hand, Ufd1 interacts with Npl4 to form the Ufd1-Npl4 heterodimer through a short segment of its UT6 domain 34 which is also conserved in the Leishmania Ufd1 ortholog. Our 3D homology modeling and docking experiments predicted binding of LiNPL4 to LiVCP through its C-terminus GMPP motif. All NPL4 orthologs of the genus Leishmania, Leptomonas and Phytomonas harbor the GMPP motif and a variant of this motif, GM-[EDS]-PP, is present in Trypanosoma species. This unique feature in Trypanosomatidae paves the way for the design of inhibitors specifically targeting NPL4 binding to VCP. Ufd1-Npl4 and p47 are substrate-recruiting cofactors shown to bind VCP in a mutually exclusive manner as they compete for the same binding modules 30,31 . This seems also to be the case in Leishmania. Lip47 was neither found in the LiUFD1 proteome nor LiUFD1 in the Lip47 pull-down. Moreover, they do not share common interacting partners, except for the V-type proton ATPase subunit D.
Our analysis led us to identify the first PUB domain proteins in Leishmania and to unveil LiPUB1 as one of the major LiVCP cofactors interacting with the C-terminal region of LiVCP. Furthermore, new prospects were opened with the yet non-studied PUB domain protein LiPUB2 identified here as a putative LiVCP cofactor. In addition, a third UBX-and PUB domain-containing protein (LinJ.36.0140; LiPUB3) was retrieved by searching the L. infantum genome database. PUB domain proteins known to bind to the C-terminus of the VCP hexamer have been associated with autophagy of damaged lysosomes 23 but the physiological significance of these interactions remains to be determined. Here, we found that a serine palmitoyltransferase-like protein was associated solely with the LiPUB1 proteome. Interestingly, a homolog of this protein in yeast is required for resistance to heat shock and plays an essential role in the removal or refolding of denatured or aggregated cytoplasmic proteins 60 . We have reported recently that LiVCP is essential for the parasite survival under heat stress 10 , and PUB proteins with the help of a serine palmitoyltransferase may contribute to this process.
Interestingly, our study revealed new LiVCP interacting partners exclusively associated with the LiVCP proteome that might function as LiVCP cofactors. These include a PUB domain protein (LiPUB2), an outer mitochondrial membrane protein, a P-type H + -ATPase with homology to fungal and plant proton pumps 61 , a FYVE zinc-finger protein known to bind phosphatidylinositol 3-phosphate in membranes of endocytic vesicles that regulate membrane trafficking and receptor signaling 56 , a microtubule-associated protein, and several hypothetical conserved proteins that may be part of distinct VCP-complexes. LiVCP, and also Lip47, associates with the type 1 protein serine/threonine phosphatase (PP1) complex consisting of the PP1 catalytic subunit Glc7/PP1-B, the SDS22 regulator-like protein and the protein phosphatase PP1 inhibitor YPI1. In S. cerevisiae, nuclear localization of Glc7 requires Sds22 and Ypi1 62 . It has been shown recently that nuclear PP1 activity is positively regulated by the AAA-ATPase Cdc48 and its cofactor Shp1 (p47) 63 to promote the assembly of the Glc7-Sds22-Ypi1 PP1 complex and to ensure its quality control 38 . Interestingly, here we found that Lip47 interacts with a large number of nuclear proteins and that is partly localized to the nucleus. From an evolutionary point of view, it is interesting that the interaction of LiVCP with Lip47 may regulate the assembly of the Glc7-Sds22-Ypi1 complex like is the case in yeast. LiVCP was found also associated, although not in all immunoprecipitates, with substrate-processing cofactors like the thioesterase ovarian tumor domain-containing protein 1 (OTU1), a deubiquitinating enzyme involved in ERAD through the interaction of its UBX-like (UBXL) domain with the N-terminus of Cdc48 64 . Although it is yet unclear how the polyubiquitinated substrates are released from the p97/VCP/Cdc48 complex and passed on to the proteasome 65 , OTU1 can be the deubiquitinase involved in this process as its UBXL domain interacts with VCP 55 . Although the Leishmania OTU1 ortholog does not harbor a well-defined UBXL domain, interestingly the 39 GYPP 42 loop of UBXL in S. cerevisiae shown not only to be critical for the interaction with VCP but also for its role in the ERAD pathway 55 it is conserved in Leishmania ( 49 GFPP 52 ).
Another novel finding here is the association of the Leishmania VCP and also of its cofactors LiFAF2 and LiUFD1 with a stomatin-like protein 2 (SLP-2). Stomatins are members of the SPFH (stomatin, prohibitin, flotillin, HflC/K) superfamily that localize to the mitochondrial inner membrane scaffolding for the spatial organization of inner membrane proteases regulating mitochondrial dynamics, quality control, and cell survival 66 . Accordingly, our data support the association of LiUFD1 with transmembrane transport and peptidase activity proteins. It is possible that LiVCP associates with SLP-2 through UFD1, which also interacts with FAF2. Also present in the LiVCP proteome and associated with both LiFAF2 and LiUFD1 cofactors is the mitochondrial ATP-dependent AAA + protease FtsH whose proteolytic and chaperone-like activity is crucial to the protein degradation protein quality control of mitochondrial and chloroplast membranes 67 . This suggests a new model for membrane protein degradation mediated by ATP-dependent proteolytic systems 57 .
Our data support a subcellular compartmentalization for the key LiVCP cofactors. LiFAF2 and LiPUB1 are associated with organelles, LiFAF2 most likely with endoplasmic reticulum (ER), Lip47 is partially localized to the nucleus, LiUFD1 is possibly associated with cytosolic vacuoles, and NPL4 is enriched in the cytoplasm. Interestingly, LiFAF2 is associated with the peroxisomal biogenesis factors PEX2, PEX10 and PEX11. Similarly to other eukaryotes 68 , recent data in trypanosomatids support de novo biogenesis of peroxisomes (glycosomes) from the ER 69 . In higher eukaryotes PEX2, PEX10 and PEX12 are part of the peroxisomal E3 ubiquitin ligase complex required for pexophagy, a pathway to degrade ubiquitinated peroxisomes that involves an AAA ATPase complex with a striking similarity to p97 70 . Glycosome turnover in Leishmania is mediated by autophagy 71 and the VCP/p97 quality control system, possibly through its FAF2 cofactor, may contribute to the control of glycosome homeostasis and turnover in these parasites. LiFAF2 also associates with two fatty acid elongases of the ELO GNS1/SUR4 family involved in the membrane-bound fatty acid chain elongation 72  www.nature.com/scientificreports/ ELO2 fatty acid elongase was found in the proteome of all four LiVCP cofactors (p47, UFD1 FAF2, PUB1). In trypanosomatids, the fatty acid elongation pathway occurs in the membrane of the ER 73 and this could explain its association with components of the ERAD pathway. In summary, this study allowed us to build the first VCP protein interaction network in trypanosomatids through the identification of known and novel interacting partners potentially associated with distinct VCP complexes. Our proteomics datasets identified biologically relevant functions for the Leishmania VCP cofactors and provided an important resource for further investigation of VCP function in several cellular processes related to protein quality control in these parasites.  25.1320/ LINF_250019000), and LiPUB1 (LinJ.09.1060/LINF_090016300) genes were amplified and cloned into pSP72αZEOα. An HA epitope tag was added either at their N-or C-terminus. Primers used in this study are shown in Supplementary Table S6. The generated plasmids were independently transfected into L. infantum promastigotes. Purified plasmid DNA (10-20 μg, Qiagen Plasmid Mini Prep Kit, Toronto, Ontario, Canada) or linearized fragments (10-20 μg for the targeting cassette 5′flank-αZEOα-VCP HA -3′flank from 10 were transfected into Leishmania by electroporation as described previously 74 . Stable transfectants were selected and cultured with 0.6 mg/ml zeocin (Sigma).

Methods
Protein lysate preparations and western blots. Western blots were performed following standard procedures. The anti-mouse HA tag monoclonal antibody (1:3,000; ABM), the anti-mouse HSP70 (cytosolic) antibody (1:400, Acris), the anti-rabbit HSP70 (mitochondrial) antibody (1:2000, kindly provided by Dr. Osvaldo de Melo Neto, Recife, Brazil), and the anti-mouse T. brucei (Tb)VCP antibody (1:5,000; 37 kindly provided by Dr. James D. Bangs (Department of Microbiology & Immunology, Jacobs School of Medicine & Biomedical Sciences, University at Buffalo, USA) were used in this study. As a secondary antibody, we used the anti-mouse Horseradish peroxidase (HRP) (Cell Signalling) or the anti-rabbit-HRP antibody (GE Healthcare). Blots were visualized by chemoluminescence with Pierce ECL2 western blotting kit (Thermo Scientific). Digitonin fractionation was carried out in the presence of increasing concentrations of digitonin (20 μM to 10 mM) as described previously 75 and each fraction was analysed by western blot.
Immunofluorescence studies. Immunofluorescence studies were done as described previously 9,10 . The anti-mouse HA tag monoclonal antibody (1:1,000; ABM), the anti-mouse H3 Histone 3 antibody (1:500; ABM) and the rabbit polyclonal anti-T. brucei BiP (kindly provided by Dr. JD Bangs) were used as a primary antibody followed by Alexa Fluor 488 anti-mouse and Alexa Fluor 555 anti-rabbit as secondary antibodies. Nucleus and kinetoplast DNA was stained with DAPI (blue). The cells were observed under a Nikon epifluorescence microscope with a 100X objective and oil immersion. Images acquisition was performed with ImagePro Plus software and ImageJ.
Immunoprecipitation studies. Immunoprecipitation (IP) studies and mass spectrometry analysis were carried out as described previously 9 . Briefly, L. infantum promastigotes expressing HA-tagged protein were lysed, mixed with Pierce anti-HA magnetic beads (Thermo Scientific, Canada) and kept at -20 °C and subjected to mass spectrometry analysis or alternatively resolved in SDS-PAGE, excised, trypsin-digested, and analyzed by mass spectrometry.

Sample preparation and LC-MS/MS analysis. Protein digestion and mass spectrometry analyses
were performed by the Proteomics Platform of the CHU de Québec Research Center (Quebec, Qc, Canada) as described previously 76 . Briefly, bands of interest were extracted from gels, placed in 96-well plates and then washed with water. Tryptic digestion was performed on a liquid handling robot (MultiProbe, Perkin Elmer) according to manufacturer's instructions using the protocol detailed in 77 with some modifications as suggested in 78 . Proteins were reduced with 10 mM DTT and alkylated with 55 mM iodoacetamide. Iodoacetamide derivative of cysteine was specified as a fixed modification and oxidation of methionine and deamidation of Asparagine and Glutamine were specified as a variable modification. Two missed cleavages were allowed. Trypsin digestion was performed using 126 nM of modified porcine trypsin (Sequencing grade, Promega, Madison, WI) at 37 °C for 18 h. Digestion products were extracted using 1% formic acid, 2% acetonitrile followed by 1% formic acid and 50% acetonitrile. The recovered peptides were pooled, vacuum-centrifuged, dried and then resuspended into 12 µl of 0.1% formic acid, and 5 µl were analyzed by mass spectrometry. Peptide samples were separated by online reversed-phase (RP) nanoscale capillary liquid chromatography (nanoLC) and analyzed by electrospray mass spectrometry (ES MS/MS). The experiments were performed with an Ekspert NanoLC425 (Eksigent) coupled to a 5,600 + mass spectrometer (Sciex, Framingham, MA, USA) equipped with a nanoelectrospray ion source. Peptide separation took place on a self-packed picofrit column (New Objective) with reprosil 3u, 120A C18, 15 cm × 0.075 mm internal diameter, (Dr Maisch). Peptides were eluted with a linear gradient from 5-35% solvent B (acetonitrile, 0.1% formic acid) for 35 min at a flow rate of 300 nL/min. Mass spectra were acquired by a data dependent acquisition mode using Analyst software version 1.7. Each full scan mass spectrum (400 to 1,250 m/z) was followed by collision-induced dissociation of the twenty most intense ions. Dynamic exclusion was set for a period of 12 s and a mass tolerance of 100 ppm. Peptide identifications were accepted if they could be established at greater than 5.0% probability to achieve an FDR less than 1.0% by the Scaffold Local FDR algorithm. Protein identifications were accepted if they could be established at greater than 99.0% probability to achieve a FDR less than 1.0% and contained at least 2 identified peptides. Protein probabilities were assigned by the Protein Prophet algorithm 79 . Proteins that contained similar peptides and could not be differentiated based on MS/MS analysis alone were grouped to satisfy the principles of parsimony.
3D homology modeling and protein-protein docking. To 25.1320) and LiVCP proteins, we searched for their orthologs in PDB database using BlastP and Delta-Blast. Then, 3D structure models were built using the modelling software MODELLER 80 based on their homologous structure, PDBid : 5C1B_V, 1S3S_H, 2MX2_A, 2HPL_B, 6CDD_ A and 5C18_D, respectively. The QPTFAGA GRT L SHP1 peptide of LiUFD1 and the DDLYS peptide of LiVCP were modeled using as templates PDBid: 5C1B_V and 2HPL, respectively. The LiVCP hexamer was obtained by superposition with the H. sapiens p97 hexamer (PDBid: 5C18). The quality of the models was assessed by Ramachandran plot analysis through PROCHECK 81 . Proteins and peptides of L. infantum were docked into LiVCP using HDOCK 82 and HPEPDOCK 83 , respectively. The electrostatic potential surfaces of 3D models and images were generated with PyMOL software (https ://pymol .org/).

Multiple alignments and phylogeny.
We used the amino acid sequence of each L. infantum JPCM5 protein to search by BlastP and Delta-Blast for close homologs in bacteria, archaea, fungi, plants and animals. The sequences extracted from databases were aligned with Clustal Omega 84 . To establish the phylogenetic relationships between L. infantum proteins and those of prokaryotic and other eukaryotic organisms, the sequences were aligned with Clustal Omega 84 and a phylogenetic tree was constructed using PhyML 85 . The gene ontology (GO) term enrichment analysis was performed using the tool provided by TriTrypDB (https ://tritr ypdb.org) with Fisher exact test filtering for false discovery rate (FDR) lower than 0.05. Cellular Component, Molecular Function and Biological Process were carried out separately.

Experimental design and statistical rationale. Filter for LiVCP IP-MS/MS: Seven IP-MS/MS studies
were considered for LiVCP (Supplementary Table S2). To select proteins that specifically associate with LiVCP, we calculated the average number of peptides for each protein in all seven experiments, and then excluded all proteins also found in five independent IP-MS/MS studies conducted with HA-tagged proteins unrelated to the VCP complex. From the selected 24 LiVCP specific interacting proteins, 15 with an average of exclusive unique peptide count (EUPC) ≥ 2.0 were classified as the LiVCP core and are presented in Fig. 1C and Supplementary  Table S2.
Filter for LiVCP cofactors IP-MS/MS: Two independent experiments were performed for each LiVCP cofactor (Lip47, LiFAF2, LiUFD1 and LiPUB1) and only proteins with an average of exclusive unique peptide count EUPC ≥ 2.0 in both experiments for each cofactor were considered (Supplementary Table S2). To exclude nonspecific partners, IP-MS/MS studies with proteins unrelated to LiVCP complexes were carried out and proteins with an average of EUPC > 3.6 (LiVCP threshold) were used to filter out non-specific LiVCP cofactor associated proteins. The remaining proteins co-immunoprecipitated with Lip47, or LiFAF2, or LiUFD1 and or LiPUB1 were considered as putative LiVCP cofactors. Cytoscape v 3.5.1 86 was used to plot network analyses of all new interactions for LiVCP, Lip47, LiFAF2, LiUFD1 and LiPUB1.
Filter for network plotting: To decrease the number of false positive results in the LiVCP network building, we only show proteins that were not co-immunoprecipitated with all unrelated HA-tagged proteins (average of EUPC = 0.0). From the total number of proteins co-immunoprecipitated with Lip47, or LiUFD1 or LiFAF2 and or LiPUB1, only 10, 17, 23 and 9, respectively were considered together with the 24 LiVCP specific interactors (Fig. 1, Table 1, and Supplementary Table S2) to build the topology of LiVCP network. No statistical analysis was needed.