The non-glycosylated protein of Toxocara canis MUC-1 interacts with proteins of murine macrophages

Toxocariasis is a neglected parasitic disease caused predominantly by larvae of Toxocara canis. While this zoonotic disease is of major importance in humans and canids, it can also affect a range of other mammalian hosts. It is known that mucins secreted by larvae play key roles in immune recognition and evasion, but very little is understood about the molecular interactions between host cells and T. canis. Here, using an integrative approach (affinity pull-down, mass spectrometry, co-immunoprecipitation and bioinformatics), we identified 219 proteins expressed by a murine macrophage cell line (RAW264.7) that interact with prokaryotically-expressed recombinant protein (rTc-MUC-1) representing the mucin Tc-MUC-1 present in the surface coat of infective larvae of T. canis. Protein-protein interactions between rTc-MUC-1 and an actin binding protein CFL1 as well as the fatty acid binding protein FABP5 of RAW264.7 macrophages were also demonstrated in a human embryonic kidney cell line (HEK 293T). By combing predicted structural information on the protein-protein interaction and functional knowledge of the related protein association networks, we inferred roles for Tc-MUC-1 protein in the regulation of actin cytoskeletal remodelling, and the migration and phagosome formation of macrophage cells. These molecular interactions now require verification in vivo. The experimental approach taken here should be readily applicable to comparative studies of other ascaridoid nematodes (e.g. T. cati, Anisakis simplex, Ascaris suum and Baylisascaris procyonis) whose larvae undergo tissue migration in accidental hosts, including humans.


Introduction
Human toxocariasis is a neglected parasitic disease of global importance [1]. It is caused by the larvae of Toxocara species, usually transmitted from domestic and stray canids or felids to humans and other mammals via the ingestion of infective eggs from the contaminated environment, food or water, or via infective larvae in undercooked meat [2,3]. Although most people infected with Toxocara larvae are asymptomatic, it has been estimated that~1.4 billion people worldwide are exposed to, or infected with, T. canis and T. cati, particularly young children and pet owners [4,5]. The four main forms of human toxocariasis include visceral larva migrans (VLM), ocular larva migrans (OLM), neurotoxocariasis (NT) and covert/common toxocariasis (CT), and there is evidence of associations between anti-Toxocara seropositivity and asthma, idiopathic Parkinson's disease and Alzheimer's disease in humans [6][7][8][9][10][11]. Most anthelmintic drugs are not effective against larvae in tissues of the human host, and no vaccine is available for the prevention of human toxocariasis.
Apart from humans, eggs containing larvae of Toxocara species can infect many mammalian species as paratenic or accidental hosts; in these hosts, larvae that emerge from the eggs in the host gut migrate to various tissues and cause infection or toxocariasis but do not develop to adults in the small intestine [2]. Fundamental investigations of human toxocariasis have relied heavily on the use of rodent (mouse, rat and jird) models [12,13]. The mouse model has been particularly useful to study host-parasite interactions, including the immunobiology of Toxocara/toxocariasis [14]. Clearly, a sound understanding of such interactions in the mouse model at the immunomolecular level is key to underpinning toxocariasis research and to developing new interventions.
Over the past seven decades [15,16], our understanding of the immunobiology of Toxocara has improved significantly [17][18][19]. For example, a series of molecules likely associated with Toxocara development, infection and parasitism have been discovered by informatic analyses of genomic, transcriptomic and proteomic data sets [20][21][22][23][24][25][26] and explored in vitro and in vivo [20]. Importantly, lectins (also known as TES-32 and -70) and mucins (also known as TES-120) excreted/secreted by Toxocara have been inferred to play crucial roles in the escape of larvae from surrounding cells and ensuing tissue inflammation within host animals [20][21][22]. Specifically, mucins of T. canis (i.e. Tc-MUC-2, -3, -4 and -5) have been reported to induce the secretion of cytokines by mouse splenocytes and to modulate the Toll-like receptor signalling in human THP-1 macrophages, indicating an involvement in the host-parasite interplay [23,24]. However, apart from mainly serological investigations [25][26][27][28], little is known about the molecular processes/mechanisms underlying the interactions between Toxocara mucins and host immune cells, particularly at a protein-protein level. In order to gain a better understanding of these aspects, we investigated proteins of murine macrophage cells that interact with a recombinant (non-glycosylated) form of Tc-MUC-1 (designated rTc-MUC-1) using a pull-down assay coupled to liquid chromatography-tandem mass spectrometry (LC-MS/MS), explored key protein-protein interactions by co-immunoprecipitation, and inferred structural and functional roles of the Tc-MUC-1 protein in host-parasite interactions.

Ethics statement
Adult worms of T. canis were collected from dogs with naturally acquired infection in the Animal Hospital affiliated to the College of Animal Sciences. Collection and experimentation were approved (permit no. 20170177) by the Ethics Committee of Zhejiang University, Hangzhou, China.

Worms and cell lines
Eggs were collected from the uteri of adult females of T. canis, incubated in H 2 O for 8 weeks at 25˚C and manually hatched with glass beads. The third-stage larvae (L3s) of T. canis released from the eggs were purified and enriched using the Baermann method [29], snap frozen in liquid nitrogen and stored in -80˚C until use. Murine macrophage cells (RAW264.7) and human embryonic kidney cells (HEK 293T) were purchased from Beyotime Biotechnology, Shanghai, and maintained according to the manufacturer's instructions.

RNA extraction and cDNA synthesis
Total RNA was extracted from pooled T. canis L3s (n = 3000) using Trizol reagent (Ambion, Thermo Fisher Scientific). RNA was reverse-transcribed into the first-strand cDNA using a PrimeScript RT Reagent Kit (Takara Bio, Dalian), according to the manufacturer's protocol. The synthesised cDNA was stored at -20˚C until use.

Prokaryotic expression of rTc-MUC-1
A recombinant form of the T. canis mucin 1 protein (rTc-MUC-1) fused to a 6× His tag was produced in a prokaryotic expression system. In brief, the Tc-muc-1 protein-coding sequence was amplified with a primer set containing Kpn I and EcoR I restriction sites (5'-GGGGTACC ATGCACGTCCTTA-3'; 5'-GGAATTCTTAACAGAAGCCGCACGT-3'), and inserted into the pCold TF plasmid (Takara Bio, Dalian) employing the Kpn I and EcoR I restriction sites. Recombinant plasmids were purified, verified by specific PCR-amplification and competent BL21 (DE3) E. coli (TransGen Biotech, Beijing) transformed. Following induction with 0.6 mM isopropyl β-d-1-thiogalactopyranoside (IPTG), rTc-MUC-1 was expressed in a culture volume of 250 ml at 37˚C for 24 h. Proteins were released from cells by sonication (1 s-on, 3 s-off for 15 min on ice) at 400 W using a Scientz-IID ultrasonic homogenizer (Scientz, China). The sonicates (25 ml each) were centrifuged at 14,000 × g and 4˚C for 20 min, and supernatants filtered (0.22 μm aperture) and run through a Ni-NTA column (Sangon Biotech, Shanghai); rTc-MUC-1 was eluted from the column with 250 mM imidazole and subjected to SDS-PAGE analysis using an established protocol [32]. E. coli transformed with the pCold TF empty plasmid was used as a negative control throughout experimentation.
A poly-His affinity pull-down method [33] was used to isolate proteins in RAW264.7 macrophage cells that bind rTc-MUC-1. E. coli expressing rTc-MUC-1 was used to produce the bait protein, which was immobilised on a Ni-NTA column (Sangon Biotech, Shanghai) and washed three times with 2 mL of 10 mM imidazole. The immobilised bait protein was incubated with 1 mL of protein extract from RAW264.7, incubated at 4˚C for 4 h and washed three times with 2 mL of 10 mM imidazole. Bait and target proteins were eluted (three times) from the column with 250 mM imidazole. The His-tagged protein expressed by E. coli containing the empty pCold TF plasmid vector was used as a negative control. Eluted proteins were resolved by SDS-PAGE and stained with silver (Silver Stain for Mass Spectrometry kit, Pierce, Thermo Fisher Scientific).
Raw mass data were processed using ProteinPilot software (SCIEX) using five search parameters (enzyme of trypsin, Cys alkylation of iodoacetamide, bias correction of true, background correction of true, protein mass of unrestricted) against the Mus musculus proteome in the UniProt database. Peptide identification was performed with a confidence of � 95%. Common contaminants (including keratin, antibodies and serum albumin or globulins) were filtered-out prior to the identification of proteins that bind rTc-MUC-1. Identified proteins were annotated based on the UniProt database and Gene Ontology (GO) resources using the OmicsBean data integration analysis platform, and based on the Kyoto Encyclopedia of Genes and Genomes database using the KEGG Orthology-Based Annotation System (KOBAS 3.0) [35]. Protein network analysis was performed using STRING 10.0 software and displayed using Cytoscape 3.4.0 platform [36,37].

Eukaryotic expression
The protein-coding sequence of Tc-muc-1 was amplified with primers 5'-CGGGATTCATGC ACGTCCTTA-3' (BamH I restriction site underlined) and 5'-GGAATTCTTAACAGAAGC CGCACGT-3' (EcoR I restriction site underlined), and then inserted into the pcDNA3.1-Flag vector, in order to express a FLAG-tagged Tc-MUC-1 fusion protein. HEK 293T cells containing plasmids expressing selected His pull-down proteins of RAW264.7 macrophage cells were prepared. In brief, coding sequences were PCR-amplified using the EcoR I-and Bgl II-site containing primers (S1 Table), and then inserted into the pCMV-HA vector to produce haemagglutinin (HA)-tagged fusion proteins. Recombinant plasmids were purified using the Qiagen Ultrapure 100 system. HEK 293T cells were transfected with recombinant endotoxin-free plasmids (5 μg) using the ExFect 2000 Transfection Reagent (Vazyme), and cultured in DMEM at 37˚C and 5% CO 2 . The culture medium was exchanged with DMEM supplemented with 10% v/v foetal bovine serum 12 h after transfection, and incubated at 37˚C and 5% CO 2 for 36 h.

Co-IP
The Co-IP assay was employed to confirm the interaction between rTc-MUC-1 and individual partner proteins from RAW264.7 macrophage cells. To extract total proteins, cells transfected with recombinant plasmids were washed with PBS (pH 7.5), suspended in 20 μL NP40 lysis buffer (Beyotime Biotechnology, Shanghai) with 1 mM phenylmethylsulfonyl fluoride (PMSF) on ice for 30 min, and then centrifuged at 14,000 × g (4˚C) for 15 min. Protein A superparamagnetic beads (Dynabeads, Thermo Fisher Scientific) were washed twice in PBS (pH 7.5), and then incubated with 3 μL mouse DYKDDDK tag monoclonal antibody (binds to FLAG tag epitope; α-Flag) (Proteintec, Wuhan). Pre-immune serum IgG was used as the negative control. Processed beads were incubated with proteins (50 μL) extracted from HEK 293T cells at 4˚C for 4 h, washed three times with PBS (pH 7.5), and mixed with 20 μL PMSF-containing NP40 lysis buffer and 5 μL loading buffer (1.5% Tris, 9.4% glycine and 0.5% SDS, pH 8.3). Protein samples were then boiled for 10 min, separated from the beads and transferred to new tubes.

Structural modelling and docking
The three-dimensional structures of non-glycosylated Tc-MUC-1 (without a signal peptide) and its interacting proteins were modelled using the programs I-TASSER [38-40] and Alpha-Fold2 [41], and GO annotations predicted. In silico protein-protein docking was performed using the ClusPro server [42,43] based on resolved/modelled structures [44]. Protein complexes were analysed and displayed using the PyMOL molecular graphics system v2.5 (Schrödinger, LLC), and predicted binding sites in the Tc-MUC-1 and protein networks in murine macrophages.

Tc-MUC-1 is predicted to have protein binding activity
A complete cDNA sequence, called Tc-muc-1 (531 nt in length), encoding the surface coat glycoprotein TES-120 precursor (nmuc-1) of infective larvae of T. canis (Fig 1A) was obtained. From this sequence, an amino acid sequence of 176 aa in length was deduced (called Tc-MUC-1) that contained a signal peptide, a low complexity region with STSSSS(P)A repeats, and two ShKT (also known as nematode six-cysteine domain or ion channel regulator) domains ( Fig  1B). A U-shaped structure of Tc-MUC-1 was modelled, with two ShKT domains folded in the centre (Fig 1C). The fold of each ShKT domain contained two nearly perpendicular stretches of helices, stabilised by three disulfide bridges (Fig 1C). With reference to the functional information in the PDB database, ShKT domain-containing Tc-MUC-1 was predicted to have a binding activity (GO:0005488), particularly protein binding (GO:0005515) and metal ion binding (GO:0046872). No specific binding site was predicted with confidence, due to limited structural and functional information on mucins of T. canis and related parasitic worms.

The rTc-MUC-1 protein interacts with at least 219 proteins of murine macrophages
To understand aspects of the host-parasite interplay at the protein-protein level, molecules in murine RAW264.7 macrophage cells that directly or indirectly bound rTc-MUC-1 of T. canis were analysed using His pull-down and subsequent mass spectrometry (Fig 2A-2D). A total of 618 peptides representing 219 proteins were identified by LC-MS/MS, 109 proteins for which at least two unique peptides (S2 and S3 Tables) were detected. Apart from 31 presently uncharacterised molecules, the other proteins identified (n = 188) were comprehensively annotated with Gene Ontology (GO) terms and/or linked to Kyoto Encyclopedia of Genes and Genomes (KEGG) pathways. The annotated proteins, predominantly associated with extracellular region and exosome, were functionally enriched in terms of RNA, protein, lipid and/or substrate binding, and are suggested to play roles in organonitrogen compound metabolism and biosynthetic processes (Figs 2D and S1 and S4 Table). These proteins appear to be involved particularly in multiple pathways, including metabolism, genetic information processing (e.g. protein processing in endoplasmic reticulum and RNA degradation), cellular processes (transport and catabolism, cellular community) and organismal systems (e.g. peroxisome, antigen processing and presentation pathway and nucleotide binding oligomerization domain (NOD)-like  Table). Based on information available for functional protein associations in the mouse, a complicated protein-protein interaction network was predicted for 147 of the mouse proteins identified (S3 Fig and S6 Table).

Interactions suggest that the Tc-MUC-1 protein is linked to cytoskeletal dynamics and signalling in macrophages
Focusing on the rTc-MUC-1-CFL1 and rTc-MUC-1-FABP5 interactions, functional information on other proteins from murine macrophages initially identified in the pull-down assay were manually curated. CFL1 has a relatively complicated network with 17 of the identified proteins of RAW264.7 cells (S6 Table), including ARHGDIA and PFN1, which were not, or only weakly, identified using the co-immunoprecipitation (co-IP) assay. These molecules appear to dominate functionally in cell activation (e.g. cellular response to stimulus) and actin filament-based processes (e.g. extracellular matrix organization, cell migration and adhesion), and in actin-based cytoskeletal dynamics regulation-associated signalling (S7 and S8 Tables). Although no protein association network was predicted for FABP5 (S6 Table), this fatty acidbinding protein as well as malic enzyme 1 (ME1) and ubiquitin C (UBC) were predicted to be involved in the peroxisome proliferator-activated receptor (PPAR) signalling pathway (S8 Table). An integration of the biological information for CFL1, FABP5 and associated proteins indicates roles for Tc-MUC-1 in regulating the cytoskeletal dynamics and adipocytokine signalling of murine macrophages, particularly in the positive regulation of extracellular matrix organization (Fig 5).

Discussion
Using an integrated pull-down, mass spectrometry, co-IP and bioinformatic approach, we investigated the interactions between the non-glycosylated form of the Tc-MUC-1 protein and macrophages of the murine host; rTc-MUC-1 was shown to be physically interactive with CFL1 and FABP5 from murine RAW264.7 macrophage cells. The integration of information from the protein association networks/pathways for the two proteins suggested that the Tc-MUC-1 protein plays a role in interfering cytoskeletal organisation and signal transduction in host macrophages.
The surface coat glycoprotein MUC-1 of T. canis has a protein binding activity. In the 1990s, it was shown that the mucin-and glycan-rich assembly form the "fuzzy coat" of T. canis, which appears to play a key role in shedding the surface coat with bound immune cells [20,45,46]. To date, a family of secreted mucins, namely MUC-1, -2, -3, -4 and -5, has been identified in the infective larvae of T. canis [47,48]. The proteins of each of these molecules are "hybrids", being unique among mucins in having acquired two ShKT domains [47,48]. One possibility is that this acquisition has conferred new properties, such as the interaction with proteins of host immune cells that promote infection. Although the ShKT domain is found in the potassium channel inhibitor ShK (a toxin) in sea anemone [49], it has been proposed to bind molecules without exhibiting a toxic effect [47]. Indeed, by performing His pull-down and co-IP assays, we confirmed that rTc-MUC-1 did bind to the actin binding protein CFL1 of macrophages and a fatty acid binding protein FABP5 of the RAW264.7 cell line. The Tc-MUC-1-CFL1 and Tc-MUC-1-FABP5 complexes were modelled in silico, and strong hydrogen bonds were predicted, supporting previous speculation that the formation of intermolecular complexes with the ShK/SXC domains might be through non-covalent means [47]. Crystal structures of these complexes should confirm the binding sites of the Tc-MUC-1 protein, although protein structures were modelled with the machine learning-based program Alpha-Fold2 [41]. N-glycosylation of Tc-MUC-1-FLAG in HEK 293T cells might explain why no interactions with AKR1B3, ARHGDIA, PFN1 or PRDX1 that were detected by pull-down experiment. Future mutation experiments should define which key residues in rTc-MUC-1 are directly involved in binding CFL1 and/or FABP5 of murine RAW264.7 macrophages.
By binding to the actin binding protein cofilin 1, the Tc-MUC-1 protein is likely to interfere with the migration and phagocytosis of macrophage cells. Like other pathogen molecules detected by host animals, the mucins of parasites can be recognised by innate immune cells [20,50]. It is well known that the specialised monocytes (e.g. macrophages) play essential roles in the detection and elimination of pathogens, including parasitic worms [51][52][53]. In this process, cell migration and phagocytosis are involved, which rely on a dynamic remodelling of actin cytoskeleton. Cofilin is one of the actin-binding proteins demonstrated to play roles in the actin cytoskeletal remodelling, migration directionality and phagosome formation of macrophages [54,55]. Interestingly, in the current study, we showed that the rTc-MUC-1 interacts with the actin binding protein CFL1 of the macrophage-like Abelson leukaemia virus-transformed cell line derived from BALB/c mice (RAW264.7) that has been commonly used to explore parasite-host interactions [56][57][58]. CFL1 was also predicted to associate with other proteins that were identified in the pull-down assay, including actin, cyclase associated actin cytoskeleton regulatory protein, coronin actin binding protein, lymphocyte cytosolic protein and Rho GDP dissociation inhibitor alpha, which are known to play roles in the myosin contraction and cofilin-mediated disassembly during macrophage migration [54,55]. These results strongly suggest that the Tc-MUC-1 protein, by interacting with CFL1, affects the migration and phagosome formation of host macrophages in innate immune responses.
The Tc-MUC-1 protein might also play a role in regulating the signal transduction in host immune cells. This statement could be supported by an association network of CFL1 with the identified signalling molecules involved in cell shape, attachment and motility, including Rho GTPases and 14-3-3 protein zeta [59][60][61]. Therefore, by interacting with CFL1, the Tc-MUC-1 protein is likely to influence the cell locomotion of stimulated macrophages; this protein might also play a role in modulating inflammation-associated signalling pathways, as interacting molecules (i.e. fatty acid binding protein 5 and the associated NADP-dependent enzyme ME1 and polyubiquitin precursor UBC) are known to transfer specific fatty acids from the cytosol to the nucleus, wherein they activate nuclear receptors linked to the modulation of NF-kappa-B signalling during inflammation [62,63]. Indeed, mucins of T. canis have been reported to play roles in the stimulation of immune cells of host animals, particularly via IL-10 and proinflammation cytokine production and Toll-like receptor signalling in vitro [23,24]. Further studies should explore the regulatory role(s) of Tc-MUC-1 on immune recognition and/or evasion (e.g., via macrophage proliferation/activation, expression of major histocompatibility complex (MHC) class II molecules, cytokine secretion and/or calcium flux).
Questions surrounding Toxocara mucins that also warrant investigation in the future include: Do sugar moieties of Tc-MUC-1 have an impact on its recognition by host macrophage cells?-as it has been reported that such elements are important in antibody-mediated immune responses [27]. Do the ShKT domains of the Tc-MUC-1 protein play roles in altering innate immune responses of host animals?-as ShKT domain-containing peroxidase and associated cuticle development of C. elegans has been reported to be involved in pathogen resistance [64][65][66]. Are mucins involved in the tuft cell activation?-which has been emerging a novel aspect for understanding intestinal immune responses to parasitic infections [67]. Addressing these questions should improve our understanding of the interplay between T. canis larvae and host cells.
In conclusion, we demonstrated that a non-glycosylated, recombinant form of the protein of Tc-MUC-1 interacts with the actin binding protein CFL1 and the fatty acid binding protein FABP5 of murine RAW264.7 macrophages in vitro. These findings suggest that the Tc-MUC-1 protein might interfere with the migration and phagocytosis of host macrophages, providing new insight into the host-parasite interplay at the molecular and cellular levels. Further investigations of glycosylated Tc-MUC-1 in the interaction with host (murine and human) macrophages, and other species of Toxocara and ascaridoids should establish whether immune recognition and modulation are genus-or species-specific.
Supporting information S1 Table. Oligonucleotide primer sets used for the PCR-based amplification of coding sequences for subsequent protein expression in HEK 293T cells.