Identification and characterisation of a Theileria annulata proline‐rich microtubule and SH3 domain‐interacting protein (TaMISHIP) that forms a complex with CLASP1, EB1, and CD2AP at the schizont surface

Abstract Theileria annulata is an apicomplexan parasite that modifies the phenotype of its host cell completely, inducing uncontrolled proliferation, resistance to apoptosis, and increased invasiveness. The infected cell thus resembles a cancer cell, and changes to various host cell signalling pathways accompany transformation. Most of the molecular mechanisms leading to Theileria‐induced immortalization of leukocytes remain unknown. The parasite dissolves the surrounding host cell membrane soon after invasion and starts interacting with host proteins, ensuring its propagation by stably associating with the host cell microtubule network. By using BioID technology together with fluorescence microscopy and co‐immunoprecipitation, we identified a CLASP1/CD2AP/EB1‐containing protein complex that surrounds the schizont throughout the host cell cycle and integrates bovine adaptor proteins (CIN85, 14‐3‐3 epsilon, and ASAP1). This complex also includes the schizont membrane protein Ta‐p104 together with a novel secreted T. annulata protein (encoded by TA20980), which we term microtubule and SH3 domain‐interacting protein (TaMISHIP). TaMISHIP localises to the schizont surface and contains a functional EB1‐binding SxIP motif, as well as functional SH3 domain‐binding Px(P/A)xPR motifs that mediate its interaction with CD2AP. Upon overexpression in non‐infected bovine macrophages, TaMISHIP causes binucleation, potentially indicative of a role in cytokinesis.

The transformed phenotype is accompanied by the alteration of multiple host cell signalling pathways, including the constitutive activation of the phosphatidylinositol 3-kinase (PI3-K) pathway that is required for the uncontrolled proliferation of the host cell (Baumgartner et al., 2000), c-Jun NH 2 -terminal kinase (JNK) that renders host cells resistant to apoptosis and contributes to their invasiveness (Lizundia et al., 2006), and JAK2-STAT3 signalling that contributes to the survival of infected cells via c-Myc expression (Dessauge et al., 2005). In most cases, the molecular mechanisms by which Theileria triggers changes in leukocyte signalling remain unknown although two strategies seem likely, namely, secretion of parasite effector molecules into the host cell or sequestration of signalling molecules at the schizont surface. Most apicomplexans develop within a membrane-enclosed parasitophorous vacuole that the parasite remodels to allow access to nutrients and to avoid lysosomal fusion (reviewed in Clough & Frickel, 2017). It is through this physical barrier that parasites such as Toxoplasma and Plasmodium interact with and manipulate their host cells. Theileria, on the other hand, dissolves the surrounding host cell membrane soon after invasion-a prerequisite for survival-and thus, the schizont surface provides a structural platform for direct host-parasite interactions (Shaw, Tilney, & Musoke, 1991).
Theileria associates closely with host cell microtubules (MTs) throughout its entire intracellular life. The transforming schizont remains intracellular and achieves its persistence within the continually dividing host cell by interacting closely with host MTs and the mitotic apparatus, ensuring its distribution to both daughter cells (Hulliger, Wilde, Brown, & Turner, 1964;. Considering the importance of parasite-cytoskeleton interactions in maintaining the transformed phenotype, we recently showed that the host cell MT-associated proteins (MAPs) EB1 (Woods et al., 2013) and CLASP1 (Huber et al., 2017) are recruited to the schizont and speculated on their possible role in recruiting and stabilising MTs on the parasite surface. We discovered that the MT-stabilising protein CLASP1 is sequestered almost completely at the parasite surface, where it remains throughout the entire host cell cycle, and we showed that the kinetochore-binding domain of CLASP1 is necessary and sufficient for parasite interaction.
We decided to use CLASP1 as a bait to search for schizont surface-associated proteins with the aim of identifying novel host-parasite interactions at the parasite-leukocyte interface. To achieve this goal we used proximity-dependent biotinylation (BioID). This proved to be a useful tool for discovering previously undescribed interactions, and here we demonstrate the localisation of the adaptor proteins CD2AP, CIN85, ASAP1 and 14-3-3 epsilon, and the MT binding protein Jakmip1 at the parasite surface. We also show that proteins involved in nuclear transport and constituents of the nuclear pore complex (NPCs) including RanGAP1, RanBP2, Importin B1, Nup214, and Nup160 accumulate close to defined parts of the schizont surface.
We show that the schizont membrane protein Ta-p104, which we have previously described to interact with both EB1 and CLASP1 (Huber et al., 2017;Woods et al., 2013), is a part of the CD2AP/CLASP1 complex, together with a novel secreted T. annulata protein (encoded by TA20980), which we term microtubule and SH3 domain-interacting protein (TaMISHIP). TaMISHIP localises to the schizont surface where it interacts with bovine CD2AP via conserved SH3-binding Px(P/A)xPR motifs. Similar to Ta-p104, TaMISHIP is capable of tracking growing MT plus ends in an SxIP-motif dependent manner when expressed in the cytoplasm. Upon overexpression in non-infected bovine macrophages (BoMac), TaMISHIP induces binucleation, indicative of a role in cytokinesis. Together, these data lead us to propose a role for TaMISHIP in regulating the interaction of the host cytoskeleton with the parasite.
2 | RESULTS 2.1 | Use of BioID to identify proteins at the T. annulata schizont surface Host-parasite interactions in Theileria-transformed leukocytes are poorly characterised, so we performed BioID with the aim of identifying protein-protein interaction networks that occur at the schizont surface. This approach works by fusing a promiscuous Escherichia coli biotin protein ligase (BirA*) to a protein of interest and expressing it in cells. Upon adding biotin to cell culture media, proteins in close proximity to the fusion protein are biotinylated and can be affinity purified using streptavidin coated beads and identified by mass spectrometry (Roux, Kim, Raida, & Burke, 2012). We fused myc-BirA* to the kinetochore binding domain of CLASP1 (amino acids 1256-1538) that was found to be sufficient for parasite binding (Huber et al., 2017). The fusion protein was expressed in T. annulata infected cells (TaC12) by lentiviral transduction, resulting in myc-BirA*-CLASP1 1256-1538 expression in approximately 60% of the population.
Immunofluorescence analysis (IFA) with anti-myc antibodies confirmed that the myc-BirA* tag did not disrupt recruitment of CLASP1  to the parasite surface ( Figure 1a, top panel). Following incubation of transduced cells in culture media containing 50-μM biotin, we used FITC-conjugated streptavidin to confirm that proteins at and near the parasite surface were indeed biotinylated (Figure 1a, bottom panel).
In nontransduced cells, no specific biotinylation of the parasite surface could be observed (not shown). Because the biotinylation reaction results in covalent biotin attachment to proximal proteins, stringent lysis conditions can be employed to solubilise protein complexes prior to affinity purification. This is an advantage when dealing with poorly soluble proteins, such as those incorporated in the parasite membrane.
To test if biotinylated proteins could be detected by Western blotting following cell lysis, TaC12_myc-BirA*-CLASP1 1256-1538 cells were incubated with biotin, lysed exactly as described in (Roux et al., 2012), and probed with HRP-conjugated streptavidin. Lysates from nontransduced cells, or TaC12_myc To facilitate mass spectrometry identification of biotinylated proteins, 2.5 × 10 8 transduced and nontransduced TaC12 cells were incubated with 50-μM biotin, subjected to stringent lysis, and affinity purified by using streptavidin coated beads. Proteins were digested on beads as described in (Roux et al., 2012) and analysed by mass spectrometry. By using this approach, we identified 2 Theileria FIGURE 1 Use of BioID to identify microtubule (MT)-binding proteins and proteins of the nuclear pore complex at the schizont surface. (a) Theileria annulata-transformed cells (TaC12) were transduced with myc-BirA*-CLASP1 1256-1538 lentivirus particles and analysed by immunofluorescence analysis. The localisation of myc-BirA*-CLASP1 1256-1538 was analysed with anti-myc labelling (green); anti-TaSP (red) antibodies were used to label the schizont surface (top panel). TaC12_ myc-BirA*-CLASP1 1256-1538 cells were incubated with 50-μM biotin prior to fixation and analysis with FITC-conjugated streptavidin (green); the parasite surface was labelled with anti-p104 antibodies (red). DNA is labelled with DAPI (blue). (b) The MT-interacting protein Jakmip1 associates with the parasite surface. TaC12 cells were stained with anti-Jakmip1 (green), the parasite was labelled with anti-p104 (red), and host and parasite nuclei were labelled with DAPI (blue). (c) Several proteins involved in nucleocytoplasmic transport were identified with BioID and tested for proximity to the T. annulata schizont surface. TaC12 cells were stained with anti-RanGAP1 (green, top panel), anti-RanBP2 (green, middle panel), or transfected with GFP-Nup214 (bottom panel). The parasite was labelled with anti-p104 (red, top, and middle panels) or anti-TaSP (red, bottom panel), and host and parasite nuclei were labelled with DAPI (blue). Scale bar = 10 μm proteins (encoded by TA08425 and TA03615) with 5 and 10 unique peptides, respectively, and with no peptides identified in the negative control (Table 1 and S1). TA08425 encodes the schizont membrane protein Ta-p104, which we previously showed to interact with host EB1 via an SxIP motif (Woods et al., 2013). TA03615 encodes a hypothetical protein, which is predicted to be secreted and also contains an SxIP-motif and one FAINT ("frequently associated in Theileria") domain.
We recently confirmed that, like Ta-p104, TA03615 is expressed on the schizont surface and that both TA03615 and Ta-p104 interact (directly or indirectly) with CLASP1 (Huber et al., 2017). 2.2 | MAPs and nuclear pore complex proteins localise close to the parasite surface In addition to the two Theileria proteins, we also identified 18 bovine proteins as potential CLASP1 1256-1538 proximal proteins that were Note. Two T. annulata proteins (TA03615 and TA08415) were identified by mass spectrometry that were absent from the control. Seventeen bovine proteins were identified with one or more peptide hit in the myc-BirA*-CLASP1 1256-1538 sample that were absent from the control. One bovine protein (FIMXS1) was identified by 72 peptides, with one peptide identified in the control. Uniprot and GeneDB were used to assign accession numbers (ID), description, and cellular localisation. The %coverage value indicates the percentage of the protein that is covered by the peptides identified by the mass spectrometry analysis. PSM is peptide spectral match, and #PSM reflects the total number of identified peptides for one particular protein. not identified (or in the case of one protein, identified with only one peptide) in the negative control (Table 1 and S1). Because proximitydependent labelling does not discriminate between true interaction partners and proteins that come into transient proximity with the bait, we tested the localisation of identified proteins by IFA. Among the proteins found were several MAPs, including CLASP1 itself, CLASP2, CAP-Gly domain-containing linker protein 1 (CLIP-170), and Janus kinase and MT-interacting protein 1 (Jakmip1). The presence of endogenous CLASP1 and CLASP2 validated our approach, as we have already shown that both proteins associate with the schizont surface (Huber et al., 2017). Jakmip1 interacts with both MTs and members of the Jak family (Jak1 and Tyk2; Steindler et al., 2004). Analysis of endogenous Jakmip1 in TaC12  suggesting that their identification in the BioID pull down is likely due to their abundance in the cytoplasm rather than a specific interaction.
RanGAP1 (Ran GTPase-activating protein 1) and RanBP2 (Ran binding protein 2, also known as Nup358) were also identified in our ASAP1 is known to constitutively bind to CIN85 (Kowanetz et al., 2004) and functions as a GTPase-activating protein (GAP) for Ras-related small GTPases (Brown et al., 1998).
The CIN85 and CD2AP/CMS family of adaptor molecules are ubiquitously expressed and have been implicated in multiple proteinprotein interaction networks and signal transduction pathways (Dustin et al., 1998;Take et al., 2000;Watanabe et al., 2000;reviewed in Dikic, 2002). We investigated the localisation of CD2AP, for which 23 unique peptides were found, in Theileria-infected cells. In paraformaldehyde (PFA) fixed cells, CD2AP is always found at the schizont surface , we think it is likely that the small amount of CIN85 and CD2AP observed in the nucleus of TaC12 cells is an artefact due to non-specific antibody labelling, especially considering that ectopically expressed CD2AP/CIN85 was not found in the nucleus.
2.4 | Use of CD2AP and CIN85 to identify further protein interaction networks at the schizont surface by using BioID Because CD2AP and CIN85 have been implicated in multiple proteinprotein interaction networks and signal transduction pathways (reviewed in Dikic, 2002), we asked the question if they could contribute to Theileria-induced transformation by recruiting signalling molecules to the schizont surface. CD2AP has been shown to bind to the tumour suppressor protein p53 in the cytoplasm of human lung carcinoma cells (Panni et al., 2015), and considering that p53 has been FIGURE 2 Analysis of bovine adaptor protein (ASAP1, CD2AP, and CIN85) localisation, and function of myc/BirA*-CD2AP. (a) TaC12 cells were stained with anti-ASAP1 (green, top panel), anti-CD2AP (green, middle panel) or anti-CIN85 (green, bottom panel). The parasite was labelled with anti-p104 (red), and host and parasite nuclei were labelled with DAPI (blue). (b) TaC12 cells were transduced with the myc/BirA*-CD2AP construct and analysed by immunofluorescence microscopy. Cells were stained with anti-myc (green), and the parasite was labelled with anti-TaSP (red); DNA was labelled with DAPI (blue, top panel). TaC12_myc-BirA*-CD2AP cells were incubated with 50-μM biotin and subsequently stained with FITC conjugated streptavidin (green). The schizont was labelled with anti-TaSP (red), and DNA was labelled with DAPI (blue, bottom panel). Scale bar = 10 μm reported to be sequestered at the surface of the Theileria schizont (Haller et al., 2010), we hypothesised that schizont-associated CD2AP could play a role in recruiting p53 to the parasite. To test this, we first expressed short hairpin RNAs (shRNAs) in Theileria-infected macrophages to deplete bovine CD2AP and obtained a mixed population in which the majority of cells expressed no detectable CD2AP ( Figure   S2A). Following selection with puromycin, CD2AP was barely detectable by Western blot ( Figure S2B). Selected cells displayed a 2-fold reduction in relative proliferation rate following 7 days in culture when compared to wild-type (WT) TaC12 cells or those transduced with GFP-CD2AP, or a non-CD2AP-targeting shRNA control plasmid (Figure S2C). To address our hypothesis that CD2AP functions to recruit signalling molecules to the parasite surface, we analysed the localisation of p53 and IKK in TaC12 cells following CD2AP depletion.
Although the pattern of both p53 and IKK was rather heterogeneous in TaC12 cells (this has been previously described for IKK expression in  et al., 2015). Importins bind to cargo proteins in the cytoplasm, and the complex is transported through the nuclear pore into the nucleus (reviewed in Kubitscheck & Siebrasse, 2017). Transient transfection of TaC12 cells revealed that both GFP-Nup160 and GFP-Importin B1 localise close to the schizont surface, as well as to the nuclear membrane or nucleus as expected ( Figure S1B). The pattern of these two proteins at (or near) the schizont surface is comparable to the staining pattern of RanGAP1, RanBP2, and Nup214 (compare Theileria lestoquardi (DQ00498) and one with a sequence identity of 42% in Theileria parva (TP01_0380; Figure S3). Interestingly, no homologue could be found in the nontransforming T. orientalis genome, making TaMISHIP a candidate for being a transforming factor (Hayashida et al., 2012). The T. lestoquardi homologue was isolated as an immunogenic protein ("Clone 5") that localises to the schizont surface and is expressed in two splice variants (Bakheit, Scholzen, Ahmed, & Seitzer, 2006). The two splice variants were also found in T. annulata and T. parva, although a function or life cycle specific expression of one or the other variant has not yet been described (Bakheit, Ahmed, & Seitzer, 2008). Microarray data shows that TA20980 is expressed in all life cycle stages, exhibiting a slight increase in gene expression in the schizont stage, with a gradual decrease through merozoites to piroplasms (Shiels and Weir, unpublished). Indirect IFA in TaC12 cells revealed that endogenous TaMISHIP is localised at the schizont surface (Figure 4a). Although TaMISHIP contains several nuclear localisation signals, and no domains that would suggest integration into the schizont membrane, we did not detect TaMISHIP in the host nucleus. We conclude that TaMISHIP is most likely secreted and then binds to other schizont surface proteins, although we cannot fully exclude the possibility that the endogenous protein is exported to the nucleus in amounts that are below the detection threshold. Analysis of TaMISHIP localisation in freshly invaded peripheral blood mononuclear cells indicates that it co-localises with Ta-p104 in sporozoites, likely in the microneme/rhoptry complex (Iams et al., 1990), and translocates to the surface of the parasite as it differentiates into a schizont ( Figure S4A). Recently, it was reported that CD2AP is involved in Toxoplasma gondii invasion (Guérin et al., 2017), and so we also tested the localisation of bovine CD2AP following infection with Theileria sporozoites ( Figure S4B). We could detect no convincing accumulation of CD2AP in the early stages of infection, suggesting that CD2AP is unlikely to play a role in invasion. We could first detect CD2AP close to the parasite surface after 24 hr when differentiation into the schizont stage had started, suggesting that CD2AP recruitment occurs only after the parasite establishes its intracellular niche.
The TaMISHIP protein sequence includes two EB1-binding SxIP motifs, three putative SH3-binding Px(P/A)xPR motifs, two regions consisting of several nuclear localisation signals, and one nuclear export signal (data from elm.eu.org; Figure 4b). Although the sequence homology between the T. annulata protein and the corresponding T. parva and T. lestoquardi proteins is rather low, these domains are conserved in all three species ( Figure S3). Px(P/A)xPR motifs can mediate interactions with SH3 domains (Kowanetz et al., 2003), and this motif is found 3 times in TaMISHIP (Figure 4b, Figure S3). Because CD2AP and CIN85 possess three SH3 domains each, an interaction of TaMISHIP with CD2AP and/or CIN85 is possible. This was tested by immunoprecipitating (IP) endogenous CD2AP, CIN85, and TaMISHIP from Theileria-infected cells. Co-precipitated proteins were identified by mass spectrometry, and for selected proteins, the interaction was confirmed by Western blotting. CD2AP was found in TaMISHIP-IPs, and vice versa, indicating that TaMISHIP and CD2AP interact with one another either directly or indirectly. We also detected four EB1 peptides by mass spectrometry and a faint band by Western blot following IP of TaMISHIP, indicating a weak interaction between TaMISHIP and EB1. Considering that MT plus ends come into transient contact with the parasite surface, the partial or weak interaction between these two molecules is not surprising and is similar to what we have previously described for Ta-p104-EB1 interaction (Woods et al., 2013). Ta-p104 and bovine CLASP1 were found repeatedly in both IPs. Together, these data suggest that Ta-p104, CLASP1, and EB1 are part of the TaMISHIP-CD2AP complex (Figures 4c,d, S5, and S6). CIN85 was also identified by mass spectrometry in both IPs, although we could not confirm this by Western blotting. The IP of FIGURE 3 Venn diagram summarising the proteins identified in three BioID experiments. The proteins identified by mass spectrometry in three independent BioID experiments (myc-BirA*-CLASP1 1256-1538 , myc-BirA*-CD2AP, and myc-BirA*-CIN85) are summarised in a Venn diagram. The proteins are grouped into Theileria annulata proteins and bovine proteins that localise or do not localise to the schizont, or were not tested for schizont localisation. Immunofluorescence analysis (IFA) of proteins that were found to associate with the schizont are shown in Figures 1, 2, and S1 (Jakmip1, RanGAP1, RanBP2, GFP-Nup214, GFP-Nup160, GFP-Importin B1, ASAP1, CD2AP, and CIN85). The following proteins were tested in IFA, and no association with the parasite was found: GFP-Ezrin, GFP-Cortactin, mCherry-Clathrin, GFP-Talin1, CrkL-GFP, and Coronin1B-pmCherry. The Venn diagram was made by using the web page https://creately.com/app/# (c) TaC12 cells were lysed and subjected to co-immunoprecipitation (co-IP) with rabbit polyclonal anti-CD2AP or rabbit IgG (control) antibodies. For each sample the nonsoluble pellet (P), lysate supernatant (SN), IP flow through (FT; for each 1.5% of total amount), and 3.3% of total bound fraction after IP (IP) was analysed by SDS-PAGE. Analysis with anti-CD2AP antibodies confirmed the successful immunoprecipitation of CD2AP. The coprecipitation of the parasite proteins TaMISHIP and Ta-p104, and the host cell protein CLASP1 could be confirmed with the corresponding antibody probes. (d) TaC12 cells were lysed and subjected to co-immunoprecipitation (co-IP) with rat polyclonal anti-TaMISHIP or rat IgG (control) antibodies. As a control, the anti-TaMISHIP immunoprecipitation was also performed with non-infected BoMac that do not express the T. annulata protein TaMISHIP. For each sample, the nonsoluble pellet (P), lysate supernatant (SN), IP flow through (FT; for each 1.5% of total amount), and 5% of total bound fraction after IP (IP) was analysed by SDS-PAGE. Analysis with anti-TaMISHIP antibodies confirmed the successful immunoprecipitation of TaMISHIP. The co-precipitation of host cell CD2AP, Ta-p104, CLASP1, and EB1 could be confirmed with the corresponding antibody probes. (e) HEK 293T cells transfected with WT GFP-TaMISHIP or GFP-TaMISHIP in which all three Px(P/A)xPR motifs were mutated to Px(P/A)xPA (GFP-TaMISHIP 3XKPR➔KPA ) were lysed 16 hr after transfection and subjected to co-IP using GFP-TRAP magnetic beads. For each sample, the nonsoluble pellet (P), lysate supernatant (SN), IP flow through (FT; for each 1% of total amount), and 10% of total bound fraction after IP (IP) was analysed by Western blotting. Analysis with anti-eGFP (enhanced GFP) antibodies confirmed the expression of the fusion proteins and their precipitation with GFP-TRAP beads. GFP-TaMISHIP is predicted to run at around 136 kDa, and the additional band detected with the anti-GFP antibody might be degradation products of the fusion protein. Although CD2AP is co-precipitated with WT GFP-TaMISHIP, no interaction can be seen between mutated GFP-TaMISHIP and CD2AP.
CIN85 did not, however, co-precipitate CD2AP or TaMISHIP ( Figure   S7). These data might indicate that CIN85 associates rather weakly with the CD2AP/CLASP1/TaMISHIP/Ta-p104 complex. We also found several isoforms of the adaptor protein 14-3-3 in both CD2AP and TaMISHIP IPs. Like CD2AP and CIN85, 14-3-3 family proteins can interact with multiple proteins-170 associated proteins that play roles in signal transduction or cellular communication have been identified for 14-3-3 gamma (Jin et al., 2004), and like CD2AP and CIN85, 14-3-3 epsilon is clearly recruited to the vicinity of the schizont surface ( Figure S1C). The proteins identified by mass spectrometry following CD2AP, TaMISHIP, and CIN85 IPs are summarised in a Venn diagram ( Figure S8), and the raw data can be found in Tables S3 and S4 TaMISHIP forms a complex with CD2AP that leads to reduced solubility. Importantly, while WT GFP-TaMISHIP co-precipitated CD2AP, no interaction of the mutated protein (GFP-TaMISHIP R173A/R189A/R205A ) with CD2AP could be detected by Western blot (Figure 4e, Figure   S7D). These results confirm that the interaction between TaMISHIP and CD2AP is specific and is mediated by the Px(P/A)xPA motifs contained within TaMISHIP.

| TaMISHIP can track MT plus ends and causes binucleation upon overexpression
Because TaMISHIP possesses two SxIP motifs, we considered that a localisation to growing MT plus ends is possible. We had previously tested whether the two SxIP motifs found within TaMISHIP were sufficient to mediate MT plus-end tracking by separately expressing fragments containing one or the other of the SxIP motifs in COS7 cells, and analysing their localisation following fixation. This approach failed to reveal any interaction with MT plus ends (Woods et al., 2013). Now, to investigate the potential interaction of TaMISHIP with MT plus ends more fully, we tested the ability of full length GFP- TaMISHIP

| DISCUSSION
In this work, we have discovered a core network of interacting proteins, including CLASP1, EB1, CD2AP, Ta-p104, and TaMISHIP (TA20980), that coats the Theileria surface throughout the cell cycle, and which likely plays a role in mediating interactions between the parasite and the host cytoskeleton. We also demonstrated the presence of nuclear pore complex proteins close to the surface of the schizont.
We were particularly intrigued at the presence of several adaptor proteins at the host-parasite interface. Adaptor proteins are known to recruit signalling proteins into complexes, which led us to speculate that the schizont recruits these proteins to initiate signalling cascades on its surface. CD2AP and CIN85 share significant sequence and structural similarity and are comprised of three N-terminal SH3 domains, a  (Cormont et al., 2003), cytokinesis (Monzo et al., 2005), and downregulating receptor tyrosine kinases (Soubeyran, Kowanetz, Szymkiewicz, Langdon, & Dikic, 2002;Szymkiewicz et al., 2002). CD2AP interacts with p85, the regulatory subunit of PI3-K (Huber et al., 2003), and in podocytes knocked out for CD2AP, protein kinase B (PKB/Akt) is less phosphorylated and therefore less active (Schiffer, Mundel, Shaw, & Böttinger, 2004;Tossidou et al., 2007). CIN85 links B-cell receptor signals to the canonical NF-κB pathway. Following knock out of CIN85 in B cells, the level of both IκB and JNK1 phosphorylation was decreased, and in vitro kinase activity of the IKK complex impaired (Kometani et al., 2011). Both CD2AP and CIN85 can directly bind the tumour suppressor p53, although the functional significance of this is unknown (Panni et al., 2015). All of these signalling pathways are known to have important functions in Theileria-infected cells. First, the PI3-K pathway is constitutively active (Baumgartner et al., 2000), which leads to PKB/ Akt activation and contributes to the proliferation of transformed cells (Heussler et al., 2001). Second, IKK is recruited to the schizont surface where it is constitutively phosphorylated, leading to the degradation of NF-κB inhibitors and subsequent NF-κB pathway activation, which is essential for the survival of Theileria-infected cells (Heussler et al., 2002). Finally, p53 was found to associate with the schizont surface to prevent its translocation to the host cell nucleus, therefore inhibiting p53-mediated apoptosis (Haller et al., 2010). Considering the contribution of these signalling pathways in Theileria-induced transformation, we hypothesised that CD2AP and CIN85 could recruit signalling molecules to the schizont surface. To test this, we performed BioID experiments using CD2AP and CIN85 to target the biotin ligase to the schizont surface. Surprisingly, we did not pull down any proteins known to be involved in pro-survival or proliferative signalling, and localisation of p53 and IKK remained unchanged in Theileria-infected cells following depletion of CD2AP. This suggests that those signalling molecules present on the parasite surface (p53 and proteins of the IKK complex) are located more than 40 nm from CD2AP/CIN85 and that these adaptor proteins do not play a direct role in their recruitment.
In addition to its roles in signalling, CD2AP is involved in regulating cytokinesis. In mitosis CD2AP concentrates on midzone MTs and the midbody, and downregulation in human cells causes defects in cell separation (Monzo et al., 2005). Although depletion or overexpression of CD2AP in Theileria-infected cells did not induce any detectable defects in cell cycle progression, an important finding of our work was the discovery that TaMISHIP interacts specifically with CD2AP via its Px(P/A)xPR motifs and induces binucleation when overexpressed in non-infected cells. Interestingly, IP of both CD2AP and TaMISHIP co-precipitated the GPI-anchored parasite surface protein Ta-gp34 ( Figure S8, Tables S3 to S5). Ta-gp34 has been previously characterised and, like CLASP1, CD2AP, and TaMISHIP, localises in a punctate pattern at the schizont surface (see Figure S9 for CD2AP and TaMISHIP structured illumination microscopy and Huber et al., 2017, for CLASP1). Although neither the function nor the binding partner of gp34 was elucidated, a potential involvement in cytokinesis was proposed . Our data suggest that Ta-gp34 is part One specific aim of our study was to identify novel parasite molecules with the potential to interact with the host. We identified 12 T. annulata proteins in our co-IP experiments, after excluding the highly abundant and intracellular ribosomal proteins. These proteins are summarised and described in Table S5. In order for a Theileria protein to qualify as a potential "host cell manipulator," several criteria must be fulfilled. The candidate proteins should be uniquely expressed in transforming Theileria species, and absent from the nontransforming species T. orientalis or other nontransforming apicomplexans such as Toxoplasma or Plasmodium. The protein should be expressed in the schizont stage, as only this stage is associated with host cell transformation, and finally, the protein should be expressed at the parasite surface or secreted into the leukocyte cytosol (Shiels et al., 2006). Of the 12 proteins we identified in our experiments, only TaMISHIP fulfils these strict criteria and was chosen for further study, and indeed an attempt to use bioinformatics screening to identify potential manipulators of the host phenotype revealed TaMISHIP as a potential candidate (Hayashida et al., 2012;Shiels et al., 2006). We showed that TaMISHIP, like Ta-p104, interacts with CLASP1 (Huber et al., 2017) and EB1 (Woods et al., 2013)

FIGURE 6
Graphical representation of host-parasite interactions on the Theileria annulata schizont surface. Parasite surface-associated proteins identified in our BioID and co-IP experiments (Ta-p104, gp34, TaMISHIP, and TA03615) are shown in grey. Microtubule-associating proteins found with BioID (CLASP1, Jakmip1), or previously published as binding to the parasite (EB1), are represented in green. Adaptor proteins (CD2AP, CIN85, ASAP1, and 14-3-3 epsilon) are represented in yellow, and proteins involved in nuclear transport (RanGAP1, RanBP2, Nup214, Nup160, and Importin B1) are represented in blue. The cell cycle-dependent association of Plk1 with the schizont and the recruitment of the IKK signalosome complex and p53 have been described previously, and P in a circle represents phosphorylation 4.3 | Primers, antibodies, and expression constructs Primers, antibodies, and expression constructs are listed in Table S6.  Figure S9 were analysed on the DeltaVision OMX Blaze system (GE Healthcare) with sCMOS cameras at the Biozentrum in Basel, Switzerland. Fiji (ImageJ) software (Schindelin et al., 2012) and Photoshop (Adobe) were used to process the images.

| IFA and time-lapse imaging
4.5 | Use of BioID to identify proteins proximal to and interacting with CLASP1 1256-1538 , CD2AP, and CIN85 The BioID approach was applied as described before (Roux et al., 2012). Briefly, TaC12 WT cells (control) and TaC12 cells expressing a myc-BirA* fusion protein were incubated in media containing 50-μM biotin (Serva) and subjected to stringent lysis. Biotinylated proteins were affinity purified with 600-μl MyOne Streptavidin C1 Dynabeads (Invitrogen). Proteins were eluted from 10% of the beads for Western blot analysis, and the remaining 90% of the beads were subjected to on beads tryptic digest and analysis by mass spectrometry on an ESI-ion trap or QTOF mass spectrometer at the Proteomics and Mass BoMac (control) cells, the anti-CIN85 and anti-TaMISHIP anti sera (300 μl of each polyclonal affinity purified antisera) and IgG from rat serum (12 μg, control) antibodies were added to the lysates (8 mg total protein each) overnight at 4°C. The next morning 1.5 mg of protein L magnetic Dynabeads were added for 6 hr at 4°C. After incubation, the beads were washed 3 times with lysis buffer and 3 times with PBS.
Proteins were eluted from 30% of the beads in Laemmli SDS-sample buffer and subjected to analysis by Western blot. The remaining 70% of the beads were subjected to on beads tryptic digest and analysis by mass spectrometry on an ESI-QTOF mass spectrometer at the Proteomics and Mass Spectrometry Core Facility, Department of Clinical Research, University of Bern, Switzerland. HEK 293T cells transiently expressing GFP-TaMISHIP fusion proteins were lysed in lysis buffer containing 0.5% NP-40 and subjected to GFP-TRAP co-IP (Chromotek) following the manufacturer's instructions.

| Sporozoite infection
A Holstein calf was infected by subcutaneous inoculation with 1 × 10 6 cells infected with T. annulata Ankara 279. The calf was monitored for the progress of infection by taking rectal temperature and examining lymph node biopsy and blood smears for the presence of schizonts and piroplasms, respectively. Unfed nymphal Hyalomma excavatum ticks were applied to the ears of the calf when piroplasms were first detected in the blood smears. After the ticks detached and moulted, the sporozoite stabilate Ankara 279 was prepared as described by Brown (1987).
Peripheral blood mononuclear cells were isolated from venous bovine blood taken from a jugular vein by flotation on Ficoll-Paque as described previously (Goddeeris & Morrison, 1988). Cells were resuspended in RPMI 1640 medium (containing 2-mM glutamine, 5 × 10 −5 -M 2-mercaptoethanol, 100-IU/ml penicillin, 100-μg/ml streptomycin, and 10% FCS) to obtain 10 7 cells/ml; 500 μl of cell suspension was mixed with an equal volume of T. annulata Ankara 279 stabilate diluted in culture medium to obtain an equivalent of one tick/ml. The suspension was incubated for 5 min, 30 min, 1 day, or 3 days at 37°C with occasional agitation, then centrifuged at

FUNDING INFORMATION
This work was supported by the Swiss National Science Foundation (SNF), project number PZ00P3_154689 awarded to Kerry Woods.
The funders had no role in study design, data collection and interpretation, or the decision to submit the work for publication.

AUTHOR CONTRIBUTIONS
S. H. and K. W. performed the conception and design of the study.
S. H., K. W., T. C., and D. R. carried out the acquisition, analysis, or interpretation of the data. S. H. and K. W. did the writing of the manuscript.