Kinetoplastid‐specific X2‐family kinesins interact with a kinesin‐like pleckstrin homology domain protein that localizes to the trypanosomal microtubule quartet

Kinesins are motor proteins found in all eukaryotic lineages that move along microtubules to mediate cellular processes such as mitosis and intracellular transport. In trypanosomatids, the kinesin superfamily has undergone a prominent expansion, resulting in one of the most diverse kinesin repertoires that includes the two kinetoplastid‐restricted families X1 and X2. Here, we characterize in Trypanosoma brucei TbKifX2A, an orphaned X2 kinesin. TbKifX2A tightly interacts with TbPH1, a kinesin‐like protein with a likely inactive motor domain, a rarely reported occurrence. Both TbKifX2A and TbPH1 localize to the microtubule quartet (MtQ), a characteristic but poorly understood cytoskeletal structure that wraps around the flagellar pocket as it extends to the cell body anterior. The proximal proteome of TbPH1 revealed two other interacting proteins, the flagellar pocket protein FP45 and intriguingly another X2 kinesin, TbKifX2C. Simultaneous ablation of TbKifX2A/TbPH1 results in the depletion of FP45 and TbKifX2C and also an expansion of the flagellar pocket, among other morphological defects. TbKifX2A is the first motor protein to be localized to the MtQ. The observation that TbKifX2C also associates with the MtQ suggests that the X2 kinesin family may have co‐evolved with the MtQ, both kinetoplastid‐specific traits.


| INTRODUC TI ON
The vermiform morphology of the unicellular parasite Trypanosoma brucei that causes African trypanosomiasis in humans and nagana in cattle is defined by a subpellicular array of microtubules (MTs; Robinson et al., 1995;Gull, 1999;Wheeler et al., 2013). These MTs are cross-linked to each other and the plasma membrane via regularly spaced fibrils (Sherwin et al., 1987). A high order is maintained within the corset with the MT plus ends converging at the parasite's posterior end and the minus ends located at the anterior of the cell (Robinson et al., 1995).
The only place where the continuity of the T. brucei MT array is interrupted is at the flagellar pocket, the deep invagination where the flagellum exits the cell body and the only place where endo-and exocytoses occur (Field & Carrington, 2009). The opening of the flagellar pocket is encircled by a collar, which influences the overall shape of this cavity. Above the collar, the so-called hook complex covers the flagellum as it exits the flagellar pocket, with two lateral arms also flanking the flagellum attachment zone (FAZ) filament and the MT quartet (MtQ) inside the cell (Albisetti et al., 2017). The FAZ filament and MtQ are among the interconnected fibers, filaments, and junctional complexes of the multipartite FAZ (Lacomble et al., 2009), which together form a lateral attachment for the flagellum as it follows a helical path along the cell body, connecting the flagellar skeleton to the cytoskeleton through both the flagellar and plasma membranes (Sunter & Gull, 2016). The MtQ originates close to the basal bodies, wraps itself around the flagellar pocket, and interdigitates within the MT corset next to the FAZ filament.
The orientation of MTs in the MtQ is antiparallel to the one found within the MT corset, thus their minus ends are proximal to the basal bodies (Robinson et al., 1995).
The trypanosome MT corset is exceptionally stable and only reorganized during cell division and life cycle transitions (Sinclair & de Graffenried, 2019). Cytoskeletal structures (MtQ, MT corset, flagellum, basal bodies, which anchor the flagella within the cell, and FAZ) need to be duplicated and faithfully segregated during the cell division cycle, just like other single-copy organelles. In addition, this has to be coordinated with replication and division of the nucleus and the single mitochondrial DNA network termed the kinetoplast, which is located on the opposite side of the basal bodies relative to the flagellum (Robinson et al., 1995;Wheeler et al., 2019).
Basal body duplication and segregation are the first visible steps in the cell division cycle and essential for subsequent cytokinesis (Woodward & Gull, 1990). The new MtQ starts growing just prior to basal body maturation and once the new flagellum has grown enough to invade the flagellar pocket, morphogenetic changes result in two separate pockets enveloping both the new and old flagella (Lacomble et al., 2010). Once the new flagellum has been nucleated and elongated, new MTs are integrated into the array between the old and new flagella, thereby increasing the overall width of the cell (Wheeler et al., 2013).
With its antiparallel orientation, the MtQ together with the FAZ filament create an asymmetric seam within the MT corset, which works as a 'cellular ruler' and helps to determine the site of future furrow ingression during cytokinesis (Sunter & Gull, 2016). Any insults to the integrity of these structures result in cytokinesis defects and it is thus tempting to envision the MtQ as a potential specialized 'highway' in charge of transport and delivery of cargo to the site of important cytokinesis events. Furthermore, the endplasmic reticulum associates with parts of the MtQ, although the importance of this association remains mysterious. The first protein localized to this structure, TbSpef1, is found at the proximal end of the MtQ, which wraps around the flagellar pocket between basal bodies and the flagellar pocket collar (Gheiratmand et al., 2013). Its depletion causes problems with the biogenesis of the MtQ and associated structures as well as motility defects. The minus end of the MtQ MTs is anchored to the basal bodies by a protein named TbSAF1, found in close proximity to TbSpef1 but localized between the pro and mature basal bodies (Dong et al., 2020). TbSAF1 was found in a proximal proteomics screen of proteins adjoining to and/or directly interacting with TbSpef1, which found three additional proteins that localize to the MtQ between the basal body and flagellar pocket collar (Dong et al., 2020). However, no motor proteins that localize to the MtQ, and which may potentially bring cargo to the cytokinesis site, have thus far been identified.
Kinesins are well-studied motor proteins that regulate MT dynamics, organize MT networks, and transport cargo along MTs.
Transporting kinesins can form homo-and heterodimers and consist of a head region containing the kinesin motor domain, which is responsible for ATP hydrolysis and MT binding, and a tail region important for dimerization with a kinesin/kinesin-like partner and cargo binding (Hirokawa & Noda, 2008). In trypanosomatid genomes, kinesins and kinesin-like proteins constitute one of the largest protein superfamilies El-Sayed et al., 2005;Ivens et al., 2005). To date, 47 genes encoding highlikelihood kinesins have been identified in the T. brucei genome (Wickstead et al., 2010b;Wickstead & Gull, 2006). Moreover, there is a similar number of kinesin-related proteins, which did not make the threshold due to sequence variance .
Interestingly, numerous kinesins are members of two major clades named X-1 and X-2 that are restricted to the trypanosomatid lineage (Wickstead et al., 2010b;Wickstead & Gull, 2006). Why the kinesin repertoire has expanded so prominently in trypanosomatids remains mysterious.
Hitherto studied kinesins of T. brucei have been shown to be important for various processes occurring in different cellular compartments. KinA and KinB are involved in cell cycle progression and localize to the central spindle during mitosis (Li et al., 2008) while TbKif13-1 localizes either to the nucleus or the mitotic spindle, where it meditates chromosome segregation during mitosis (Wickstead et al., 2010a). KinC and KinD are distributed throughout the cytoskeleton, where they are involved in subpellicular MT organization, basal body segregation, and cytokinesis (Hu et al., 2012;Wei et al., 2013).
Some kinesins are associated with the flagellum and adjacent FAZ. KIN2A and KIN2B are kinesins involved in intraflagellar transport, a process that directly affects flagellar assembly and in turn affects cell motility and cytokinesis (Aslett et al., 2009;Bertiaux et al., 2020). The orphan kinesin KIN-E predominantly localizes to the flagellar tip and is required for elongation of the new FAZ . Another orphan kinesin named KLIF initially localizes to the tip of the new, elongating FAZ and associates with the old FAZ as the cytokinetic furrow ingresses toward the posterior end, thus mediating the final stages of cytokinesis (Hilton et al., 2018).
Other kinesins localizing to specific cytoskeletal structures include FCP2/TbKinX1 and FCP4/TbKin15, which are found at the flagella connector (FC), a trypanosome-specific structure that connects the outgrowing new flagellum to the old one (Varga et al., 2017).
Here we present data on the localization and function of two members of the kinesin superfamily, TbKifX2A and TbPH1, with the former protein belonging to clade X2. Functional characterization and the unique localization of these kinesins further expands our knowledge on how these diverse and dexterous proteins affect the biology of these extremely diverse and successful parasites.
2 | RE SULTS 2.1 | TbPH1 is a kinesin-like protein that fractionates with the T. brucei cytoskeleton TbPH1 (Tb927.3.2490) is a 110 kDa protein named after a C-terminal pleckstrin homology (PH) domain, which is followed by a proximate homeodomain-like (HDL) fold and preceded by a coiled coil (CC) domain ( Figure 1a). The protein undergoes post-translational modifications, with phosphorylation of a serine at the start of the HDL (Urbaniak et al., 2012) and a methylarginine in the middle of the primary structure (Lott et al., 2013). While TbPH1 has an N-terminal kinesin motor domain, two substitutions of highly conserved amino acids within the Walker A motif (G93P and T/S95R) likely ablate the ATP hydrolase activity. Thus, TbPH1 is unlikely to be a bona fide kinesin motor protein.
TbPH1 also belongs to the ~2000 proteins found in the granuleenriched fraction (containing cellular aggregates, nuclei, kinetoplasts, and flagella as well as associated cytoskeletal structures) obtained by 'MT sieving', a method originally conceived to purify stress granules from T. brucei (Fritz et al., 2015). In brief (see Figures S1 for 'MT sieving' pipeline), cells are lysed by non-ionic detergent in low ionic strength buffer, which maintains the cytoskeleton, including the MT corset that in turn cages predominantly detergent insoluble cellular structures with at least one dimension >20 to 30 nm. The MT corset is then disrupted by 300 mM NaCl to release these elements for subsequent analysis.
To verify TbPH1's apparent association with cytoskeletal fractions, we subjected a T. brucei cell line in which the endogenous TbPH1 open reading frame was appended with a C-terminal V5 tag (Dean et al., 2015) to cell fractionation by a procedure identical to 'MT sieving'. The cell equivalents of the ensuing fractions were inspected for the presence of TbPH1-V5 and SCD6, a stress granule marker that served as a control for this method (Figure 2a). SCD6 was mostly found in the soluble fractions as previously reported for unstressed T. brucei (Fritz et al., 2015), with a distribution across the analyzed fractions very distinct from that of TbPH1-V5. TbPH1-V5 was well retained in all cytoskeletal fractions P1 and P2, although a fraction was released in the SN1 with other soluble cytosolic proteins upon initial lysis ( Figure 2a). Importantly, an aliquot of P2 fraction that was fixed on microscopic slides for indirect immunofluorescence (IFA) using an anti-V5 antibody shows that TbPH1-V5 is localized as a line extending from the DAPI-stained kinetoplast (Figure 2b) that decreases in signal intensity distally, suggestive of a cytoskeletal localization. This was confirmed by the observation that the vast majority of TbPH1 was released into the soluble SN3 fraction after high salt disintegration of the cytoskeleton. A small amount of TbPH1-V5 was still found in the final pellet P3, which is consistent with its detection in the cohort of proteins retained by 'MT sieving' (Fritz et al., 2015). Thus, a significant portion of the kinesin-like protein TbPH1 fractionates with the trypanosomal cytoskeleton.

| TbPH1 strongly interacts with the trypanosomatid-restricted orphan kinesin TbKifX2A
Because kinesins tend to form dimers (Marx et al., 2009) and kinesin-like TbPH1 contains a CC domain (Figure 1a), which usually mediates protein-protein interactions (Burkhard et al., 2001), we proceeded to see if TbPH1-V5 has an interaction partner by immunoprecipitation (IP) via its C-terminal epitope tag. To enrich for soluble TbPH1-V5 and augment IP stringency, we followed the previously described fractionation technique and used fraction SN3 as the input (Figure 3a). A mock IP control using the parental cell line without an expressed V5 epitope was performed in parallel. Importantly, these bands were not observed in the mock control, indicating their presence is not due to unspecific binding F I G U R E 1 Domain architecture of TbPH1 and TbKifX2A (a) and TbKifX2C (b). Schematic depiction includes phosphorylation and arginine methylation sites. The consensus Walker a motif is shown in between the two schemata for comparison of the respective motifs within the kinesin motor domains of TbPH1 and TbKifX2A. Key residues in ATP hydrolysis shown in red and mutations in purple. The likely inactive state of the TbPH1 "kinesin motor domain" is highlighted by striped pattern. CC, coiled coil domain; HDL, homeodomain-like fold; PH, pleckstrin homology domain.
to the magnetic beads used to immobilize the anti-V5 antibody.
Mass spectrometry analysis of these two excised bands identified TbPH1 itself as expected and a previously uncharacterized kinetoplastid-specific kinesin (Tb927.9.15470) (Dataset S1), which we call TbKifX2A based on a previously proposed kinesin nomenclature (Wickstead et al., 2010b). In contrast to TbPH1, TbKifX2A's kinesin motor domain contains a conventional Walker A motif ( Figure 1a). This motor domain is located on the N-terminus, indicating that TbKifX2A actively moves toward the plus-end of MTs (Hirokawa et al., 2009). It also contains a CC domain and two serine phosphorylation sites (Urbaniak et al., 2012), similarly to TbPH1.
F I G U R E 2 TbPH1-V5 fractionates with the cytoskeleton. (a) Western blot of-trypanosome cell fractions with equal cell equivalents loaded per lane as previously described (Fritz et al., 2015). Fractionation procedure is summarized in Figure S1. Antibodies used were anti-V5 and anti-SCD6 (Kramer et al., 2008). P, pellet; SN, supernatant; T, total lysate. (b) IFA of on an aliquot of fraction P2 placed on microscope slide. Cells were labeled with anti-V5 antibody (green) and DAPI (blue). The scale bar corresponds to 5 μm. F I G U R E 3 TbPH1 interacts with TbKifX2A, a kinetoplastid-specific kinesin. (a) Western blot verifying immunoprecipitation of TbPH1-V5. The blot was probed with anti-V5 and antiα-tubulin antibodies. The parental cell line SmOxP9 was used as a control. E, elution; FT, flow through; SN3, supernatant 3; T, total lysate. (b) SyproRuby-stained gel of elution fractions of TbPH1-V5 IP and mock control (SmOxP9). B1 and B2: Bands excised and by mass spectrometry identified to be TbPH1 and TbKifX2A, respectively. Asterisks: Band corresponding to denatured immunoglobulin light chain that was incidentally eluted. (c) Western blot showing interaction of TbPH1-V5 and TbKifX2A-HA. TbKifX2A-HA was immunoprecipitated in the same way as PH1-V5. Lanes loaded as in Figure 2a. E, elution; FT, flow through; P, pellet; SN, supernatant; T, total lysate. (d) IFA on whole cells detecting TbPH1-V5 (white) and counterstained with DAPI to visualize DNA (magenta). Arrowheads denote increased signal in the area along the proximal part of the flagellum. Asterisks demark an increased signal at the anterior cell body end. Merge of DAPI and phase-contrast images shown on bottom. Scale bar, 10 μm. (e) IFA on whole cells detecting TbKifX2A-HA (white). Images depicted as in D. (f) IFA on whole cells detecting TbPH1-V5 (green) and TbKifX2A (magenta). The merged image is also counterstained with DAPI (blue) to visualize DNA. Scale bar, 5 μm. Close up views of parts with strongest signals (white boxes) also shown. Single channels of the inset images are shown in Figure S2a. Scale bar, 1 μm.
Interaction between these two proteins was further verified by tagging TbKifX2A with a C-terminal HA tag in the TbPH1-V5 cell line.
Indeed, TbKifX2A mirrored the fractionation pattern of TbPH1-V5 ( Figure 3c). As before, the SN3 fraction was used as an input for TbKifX2A-HA IP, which co-immunoprecipitated TbPH1-V5. Thus, we verified that TbPH1 and TbKifX2A exhibit a strong interaction that persists even in 300 mM NaCl.
Whole cells expressing TbPH1-V5 and TbKifX2A-HA were fixed and permeabilized prior to incubation with antibodies recognizing either epitope-tag (Figure 3d,e). We observed that both proteins possess a considerable cell body signal that appears to be nucleusexcluded, with a prominent signal along the proximal region of the flagellum. This pattern is similar to the TrypTag localization data, which visualized mNeonGreen-tagged proteins in immobilized live cells (Dean et al., 2017). The cell body signal likely represents the cytosolic population of both proteins that was also observed in our fractionation data (Figures 2a and 3c). It should be also noted that  (Figures 2a and 3c). The specificity of the antibodies recognizing V5 and HA on extracted cytoskeletons was verified by a lack of signal in the parental cell line lacking either of these epitopes ( Figure S3b).
We proceeded to identify specifically which cytoskeletal element TbPH1 and TbKifX2A associate with. We used antibody rec- This pattern led us to investigate whether the kinesin was bound to the MtQ, as it originates close to the basal body and is considered a part of the FAZ (Lacomble et al., 2009;Sunter & Gull, 2016). To this end, we used the monoclonal antibody 1B41 recognizing the MtQ , a somewhat contentious claim that is belied by its use as a MtQ marker in subsequent studies (Rotureau et al., 2011; see Experimental Procedures for more details about the 1B41 batch used here). Indeed, TbKifX2A-HA shows co-localization with 1B41 To verify that TbKifX2A indeed binds the MtQ, we employed expansion microscopy (ExM), which enlarges cells by embedding F I G U R E 4 Localization of cytoskeleton-associated TbKifX2A and TbPH1. (a) IFA on extracted cytoskeletons probing localization of TbPH1-V5 (green) and TbKifX2A-HA (magenta). (b) IFA on extracted cytoskeletons probing co-localization of TbKifX2A-HA (magenta) with the FAZ (green), the latter labeled with L3B2 antibody. (c) IFA on extracted cytoskeletons probing co-localization of TbKifX2A-HA (magenta) with basal bodies (green), the latter labeled with YL1/2 antibody (d) IFA on extracted cytoskeletons probing co-localization of TbKifX2A-HA (magenta) with the MtQ (green), the latter labeled with 1B41 antibody. In all panels, merge images depict co-localization in white and DIC on right showing observed extracted cytoskeleton. Scale bar, 5 μm. Close up views of parts with strongest signals (white boxes) shown in rightmost images. Single channels of the inset images are shown in Figure S2b. Scale bar, 1 μm. them within a swellable polyacrylamide matrix prior to immunolabeling to allow examination of cellular structures in greater detail (Gambarotto et al., 2019). ExM has recently been implemented in T.
brucei (Amodeo et al., 2021;Gorilak et al., 2021) and accentuates the part of MtQ that wraps around the flagellar pocket upon immunolabeling with antibody recognizing MT-incorporated, acetylated αtubulin ( Figure 5a; Gorilak et al., 2021). Additionally, this procedure allows the proximal MtQ to be distinguished from the nearby hook complex, which lacks MTs (Esson et al., 2012).
Performing ExM allowed us to localize TbPH1-V5 or

| Proximity proteomics reveals the interaction network of TbPH1 and TbKifX2A
Since conventional IP of TbPH1-V5 under high salt conditions identified only TbKifX2A as its major interaction partner, we sought a different method to identify potential cargo and/or interactors of this putative kinesin complex. To do this, we tagged TbPH1 with a C-terminal biotin ligase "BioID2" (Kim et al., 2016), appended with an HA tag (Pyrih et al., 2020), to facilitate proximity-dependent biotin identification (BioID) of neighboring proteins. First, we determined by IFA on permeabilized whole cells using anti-HA antibody that TbPH1-BioID2-HA localizes adjacent to the kinetoplast, which was also observed for TbPH1-V5 (cf. Figures 2b and 6a). TbPH1-BioID2-HA also exhibited biotinylation activity, as visualized by streptavidin-conjugated to Alexa Fluor 488 (Figure 6a), in contrast to the parental cell line lacking BioID2 ( Figure 6b). Thus, we confirmed that the chimeric protein is properly localized and appears to biotinylate TbPH1 itself or proximal proteins.
Purification of biotinylated proteins was performed as described elsewhere (Morriswood et al., 2013; Figure S5). Western blotting revealed that the vast majority of TbPH1-BioID2-HA was present in the soluble S1 fraction (Figure 6c), which was used for subsequent analysis. At first glance, this result appears to disagree with previous findings from the 'MT sieving' fractionation, which employs comparable amounts of non-ionic detergent and buffers of low ionic strength; however, we should emphasize that 'MT sieving' fractionation uses a different detergent for a shorter incubation time and at a lower temperature in comparison with the BioID purification (see Methods), which likely explains this difference.
Next, the S1 fraction was incubated with streptavidin-coated magnetic beads to capture biotinylated proteins, which were subsequently trypsinized on beads prior to liquid chromatography-tandem mass spectroscopy (LC-MS/MS). A mock control using the parental cell line lacking BioID2 expression was processed in parallel. A total of 1131 proteins were detected with high confidence in at least one of three biological replicates (Andromeda Protein Score ≥ 20; >1 unique peptide per protein; Dataset S2). Following imputation of missing values, seven proteins were shown to be preferentially biotinylated by TbPH1-BioID2-HA, as visualized by plotting the −10 log ttest p value versus the t test difference in a volcano plot (Figure 6d). Statistically significant hits are found above the cut-off curve in this plot.
TbKifX2A was the most enriched protein based on both parameters, consistent with its strong interaction with TbPH1. As expected, TbPH1 is also among the enriched proteins, indicating that TbPH1-BioID2-HA biotinylated itself and/or a nearby untagged, endogenous form. The sub-cellular location of the five novel proteins was determined using the TrypTag protein localization resource (Dean et al., 2017). This allowed us to already exclude one of them (Tb927.10.12110) as a contaminant, given its nuclear localization. After TbKifX2A, a flagellar pocket protein named FP45 (Tb927.9.9730) was the next most enriched proximal protein.
Interestingly, FP45 is a soluble protein that has been shown to associate with the membrane surrounding the flagellar pocket by immunogold labeling in an independent study (Gheiratmand et al., 2013).
A putative flavin-trafficking protein (Tb927.8.1950) was the next most abundant protein, but does not have a TrypTag localization entry at this time. However, like TbPH1 and TbKifX2A, this protein was found among those that come into proximity with TOEFAZ1 (alias CIF1; Zhou et al., 2018), a protein that associates with the tip of the extending FAZ ( Figure 6e; Hilton et al., 2018). Furthermore, IFA has shown that that N-terminally tagged Tb927.8.1950 exhibited a strong signal in the vicinity of the kinetoplast (Hilton et al., 2018).
The next most enriched protein was a putative kinesin with a Cterminal PH domain (Tb927.6.2880), which is concentrated around the basal body according to TrypTag. Based on phylogenetic analysis described in the next section, we dub this kinesin TbKifX2C.
Cells were lysed in the same buffer as for the BioID procedure to maintain consistent conditions, except for Tb927.8.1950. Because this protein appeared to be cleaved in the BioID buffer, a different buffer that maintained the full-length protein was used for pull-down of this protein ( Figure S6). IP of the three V5 baits was successful, but TbPH1-BioID2-HA co-IP was detected only in the FP45 or TbKifX2C IPs (Figure 6f, Figure S6). FP45 exhibits a robust interaction with TbPH1-BioID2-HA, as the latter is noticeably depleted from the flow through fraction, which contains unbound proteins. In contrast, it appears that a fraction of TbKifX2C interacts with TbPH1, as there is a substantial amount of unbound TbPH1-BioID2-HA in the flow through. TbPH1-BioID2-HA was not detected in any of the mock IPs or the Tb927. 8.1950 IP. Thus, we conclude that FP45 and kinesin TbKifX2C are genuine TbPH1 interactors.
We also compared the TbPH1 proximity proteome to those of TbMORN1, a protein that is part of the hook complex (Morriswood et al., 2013), and TbSpef1, a protein localized to the MtQ (Dong et al., 2020; Figure 6g). Tellingly, there was no overlap among any of these proteomes, suggesting that TbPH1 is not in close proximity to TbMORN1 or TbSpef1 since the biotinylation radius of the BioID biotin ligase is approximately 10 nm (Kim et al., 2014).

| TbPH1 interacts with another X2 family kinesin in addition to TbKifX2A
Because we demonstrated that TbPH1 interacts with a kinesin other than TbKifX2A, we decided to further classify this protein, an effort that eventually led us to name it TbKifX2C. In the comprehensive survey of kinesins (Wickstead et al., 2010b), TbKifX2C was deemed too divergent to group into any kinesin family. Nevertheless, it does bear a canonical N-terminal, kinesin motor domain, and a PH domain on its C-terminus ( Figure 1b). Furthermore, there is evidence that TbKifX2C is phosphorylated, like TbPH1 and TbKifX2A (Figure 1a; Urbaniak et al., 2012).
We noticed that the gene encoding TbKifX2C was syntenic to a Leishmania major gene encoding a kinesin placed in the X2 clade (Wickstead et al., 2010b;Wickstead & Gull, 2006). To get initial support that TbKifX2C is part of the X2 family, we performed a BLASTP search of 12 predicted proteomes covering trypanosomatid diversity (Lukeš et al., 2018; Table S1A) using TbKifX2A as a query. A total of 47 protein entries were retrieved below the E-value threshold of 1e-20 (Table S1B). TbKifX2C was in this dataset along with expected hits (e.g. orthologs of TbKifX2A).
Encouraged by this result, we next proceeded with an unbiased molecular phylogenetic approach to test TbKifX2C's placement within the X2 family. We retrieved from the 12 predicted proteomes 549 sequences matching the Pfam "kinesin motor domain" model in a hidden Markov Model search (Table S1C, Supplemental data 1). These sequences were used to produce a trimmed alignment in which extensive gaps were removed. This alignment was then used to construct the phylogenetic tree depicted in Figure 7a.
The majority of X2 family kinesins as defined by Wickstead and colleagues (Wickstead et al., 2010b;Wickstead & Gull, 2006) were grouped into a well-supported clade that was comprised of three distinct and well-supported sub-clades. Each sub-clade contained at least one annotated X2 family kinesin, such as TbKifX2A.
Importantly, TbKifX2C was placed into the X2 clade as well, clustering with the product of the syntenic L. major gene. It should be noted that sequences from the basal trypanosomatid Paratrypanosoma confusum (Skalický et al., 2017) represented early diverging outgroups in all of the subclades except X2C, which contains the apparently divergent sequences from Trypanosoma species. As a testament to this, the Trypansoma congolense member of the X2C clade (TcIL3000_6_2380), which clusters with the divergent TbKifX2C, was the only X2 family sequence not retrieved in the initial BLASTP search (Table S1B). Thus, we conclude from these unbiased in silico approaches that like TbKifX2A, TbKifX2C is a member of the X2 family.
To see if TbKifX2C co-localizes with TbPH1, we performed IFA on fixed whole cells and extracted cytoskeletons (Figure 7b (Morriswood et al., 2013)), and the 11 high confidence proteins in the TbSpef1, a MtQ binding protein (Dong et al., 2020). Proteins exclusive to the TbPH1 proximity proteome are highlighted in green while magenta highlights those that are found to also be in proximity with TOEFAZ. kinetoplast (Figure 7c), which is consistent with a portion of TbPH1 interacting with TbKifX2C (Figure 6f). This partial co-localization is in contrast with the almost complete overlap of TbPH1 with TbKifX2A, which indicates that TbKifX2A is the main partner of TbPH1, perhaps explaining why TbKifX2C was not detected in the initial TbPH1 co-IP under the high salt conditions of 'MT sieving'.
To further explain why TbKifX2C was not observed in our initial IPs identifying the TbPH1-TbKifX2A complex, we decided to look at expression of the T. brucei X2 family kinesins in long slender bloodstream and procyclic stages (Tinti et al., 2022). While TbPH1 is minimally regulated between the two life cycle stages, TbKifX2A is considerably upregulated in the procyclic stage whereas TbKifX2C is expressed at roughly a third of the levels as it is in the long slender stage (Figure 7d). Interestingly, the T. brucei kinesin belonging to the X2B sub-clade (Tb927.8.4840) exhibits more regulated expression in the long slender stage, with only 1 peptide detected in the procyclic stage. Thus, it appears that TbKifX2A is the predominant X2 kinesin in the procyclic stage, which was the experimental model used in this study, which may also explain why this was the only kinesin to be bound to TbPH1 during 'MT sieving'.

| Depletion of TbPH1 and TbKifX2A causes prominent morphological defects
To assay the function of the novel kinesin TbKifX2A and its asso- suggesting that the protein might depend on its partner for stability.

F I G U R E 7
TbKifX2C is part of the X2 clade of kinesins and partially co-localizes with TbPH1. (a) Phylogenetic tree of 549 trypanosomatid sequences matching the "kinesin motor domain" model with high confidence (see materials and methods). Clades containing sequences that lie outside of the X2 family were collapsed and named based on the presence of L. major and/or T. brucei representatives belonging to kinesin families as defined by Wickstead and colleagues (Wickstead et al., 2010b;Wickstead & Gull, 2006). The numbers of sequences belonging to each clade is shown within each triangle that is also shaded based on the number of contained sequences. Branch support values >50 are depicted; those values >95 and 100 are marked by closed and open circles, respectively. Four pointed star marks X2 family sequences as annotated by Wickstead and co-workers. Scale bar, number of substitutions per site. (b) IFA on whole cells probing co-localization of TbKifX2C-V5 (green) with TbPH1-BioID2-HA (magenta). Scale bar, 5 μm. Close up views of parts of strongest signal (white boxes) also shown. Scale bar, 1 μm. (c) IFA on extracted cytoskeletons probing co-localization of TbKifX2C-V5 (green) with TbPH1-BioID2-HA (magenta). Scale bar, 5 μm. Close up views of parts of strongest signal (white boxes) also shown. Scale bar, 1 μm. Single channels of insets shown in Figure S2c (d) Tukey plot showing differential expression of T. brucei X2 family kinesins in long slender bloodstream (B) and procylic (P) stages based on LC-MS/MS iBAQ values from Tinti and Ferguson (Tinti et al., 2022) depicted on logarithmic scale on y-axis. Numbers above x-axis report number of iBAQ spectra per sample. Colored numbers below each boxplot denote the median iBAQ value. ****p < 0.0001; *p < 0.05.
To get a more robust RNAi phenotype for phenotypic analysis, we generated a cell line in which both proteins were downregulated, which showed more pronounced growth inhibition than the single knock-downs (Figure 8b, Figure S7a). Growth slowed after 3 days of RNAi induction and did not reach levels of the uninduced counterparts at any point during the 7 day time course (Figure 8b). Depletion of both proteins was confirmed by western blotting (Figure 8b, inset), although neither protein was decreased below the detection limits of this assay, perhaps explaining the lack of a more severe growth phenotype.
Because both TbPH1 and TbKifX2A are modified in a way suggestive of post-translational regulation (Figure 1a), we wondered whether the growth inhibition of the double RNAi knockdown is due to a cell cycle defect. Thus, we assayed the proportion of cell cycle stages present over a course of 7 days of TbPH1/TbKifX2A depletion, sorting individual DAPI-stained cells into different cell cycle stages by scoring their number of nuclei (N) and kinetoplasts (K) (Robinson et al., 1995). While we observed a moderate decrease in 1N1K cells and increase in 2N2K cells (Figure 8c), these were not to the extent that would correspond to an unequivocal cell cycle defect. The number of zoids (cells with 1 kinetoplast and lacking a nucleus) reached a peak at day 5, making up about 7% of the population, which may suggest a cytokinesis defect. However, a conclusive cytokinesis defect phenotype should also include a concurrent emergence of 2N1K or multinucleated cells, which was not observed here (Robinson et al., 1995).
The cell cycle was also assayed by flow cytometry on propidium iodide-stained cells, which quantifies the portion of cells in the G1, S, and G2/M phases, as well as cells with aberrant DNA content (Darzynkiewicz et al., 1999). While the aforementioned cell cycles phases were not markedly affected upon TbPH1/TbKifX2A depletion, this assay also revealed the statistically significant emergence of zoids from 3 days post-induction (Figure 8d, Figure S7c). Thus, while we cannot conclude that TbKifX2A and TbPH1 play a direct role in cell cycle progression or cytokinesis, the two assays we employed here indicate that their depletion leads to a slight accumulation of anucleate zoids.
To gain unbiased insight into potential effects of simultaneous depletion of TbPH1 and TbKifX2A, we prepared whole cell lysates Andromeda Protein Score ≥ 10 and >1 unique peptide per protein) were detected for the 3 and 5 day time points, respectively. post-translationally modify the C-terminal tails of α-and β-tubulin by addition of polyglutamate side chains (Jentzsch et al., 2020).

Reassuringly, TbPH1 and
Upon depletion of either TTLL, protrusions were observed emanating from the posterior end, which prompted the authors to name this morphological change the "glove" phenotype. To investigate whether TbPH1/TbKifX2A downregulation truly phenocopies that of each TTLL, we immunolabeled cells 5 days post RNAi-induction with antibodies recognizing glutamyl side chains (GT335 antibody).
Upon depletion of either TTLL, polyglutamylation was decreased and α-tubulin tyrosination increased in the posterior end, from which the protrusions emerged (Jentzsch et al., 2020). We did not observe any changes in polyglutamylation upon TbPH1/TbKifX2A silencing by imaging individual cells by IFA or on the population level by western blot analysis ( Figure S8b), suggesting the mechanism underlying the emergence of posterior-end protrusions differs from that of the TTLLs.

| DISCUSS ION
Kinesins and related kinesin-like proteins constitute one of the largest protein families encoded by the T. brucei genome  and are represented by diverse gene repertoires in other trypanosomatids as well (El-Sayed et al., 2005;Ivens et al., 2005;Wickstead et al., 2010b;Wickstead & Gull, 2006). Among the eukaryotes, trypanosomatids have experienced one of the largest expansions of the kinesin motor superfamily, suggesting these proteins endow trypanosomes with unique biological properties, many of which still await elucidation Wickstead et al., 2010b;Wickstead & Gull, 2006). Indeed, this expansion has resulted in the emergence of two trypanosomatid-specific kinesin families named X1 and X2. A kinesin belonging to the former, named FCP2/TbKinX1, has been shown to maintain a connection of a new flagellum's distal tip along the side of the old flagellum, thus facilitating the unique mechanism of templated flagellar replication in T. brucei (Varga et al., 2017).
In this work, we have shed light on the thus far neglected X2 family of kinesins. First, by a widescale phylogenetic analysis using kinesin sequences from 12 trypanosomatid species, we have shown that the X2 family is comprised of three sub-clades. Sub-clades X2A and X2B contain kinesins from 11 of the 12 searched predicted proteomes, whereas X2C has representative from all 12. Thus, it appears that the three sub-clades are well-conserved in trypanosomatids, with gene losses being the exception, suggesting that this X2 kinesin repertoire is fundamental for these flagellates. Indeed, the model used in this study, T. brucei, possesses kinesins from all three sub-clades.
The well-supported X2 clade of our study essentially agrees with the composition of the X2 clade that was initially revealed in a phylogenetic tree of kinesins throughout eukaryotes, with one notable exception (Wickstead et al., 2010b;Wickstead & Gull, 2006). Two L. major and T. brucei kinesins that were originally annotated as X2 family kinesins were placed outside of the X2 clade in our analysis (marked 'X2?' in Figure 7a). This can be explained by the fundamental different datasets and goals of the two studies. Wickstead and colleagues aimed to understand general kinesin evolution, and thus sampled a wider range of eukaryotes, necessitating the inclusion of only two trypanosomatid species. Our study aimed to address whether a divergent kinesin should be included in the kinetoplastid-specific X2 family, and thus we restricted our dataset to a wider range of 12 trypanosomatid species. In the end, the trees are not as contradictory as they ostensibly seem at first. The outlying kinesins in our 'X2?' clade were placed sister to the otherwise well-supported X2 clade with less than 50% support in Wickstead's tree (Wickstead & Gull, 2006), which is consistent with the placement of these sequences in relation to the X2 clade in our study. Thus, we propose that our tree reveals the structure of the X2 family sensu stricto (s.s.), given the level of support of this clade and contained sub-clades, implying the stability of its internal branching. The evolution of the kinesin superfamily in kinetoplastids is out of the scope of this study, but we hope this result provides impetus for future investigation of this topic.
Another property of X2 s.s. kinesins we have serendipitously discovered in this study is that they directly interact with TbPH1, a likely catalytically-dead, kinesin-like protein with a C-terminal PH domain that localizes to the MtQ. While we do not address here whether TbPH1 has truly lost ATP hydrolase activity, it contains two point mutations to conserved residues of the Walker A motif that are likely to ablate this activity. A glycine to proline mutation exchanges the amino acid that confers the most flexibility to the peptide backbone for one that rigidifies it. The polar threonine/serine amino acid responsible for the requisite coordination of Mg 2+ (Deltoro et al., 2016) is replaced with a positively charged arginine, which would repel this cation. Divergent Walker A motifs that still retain ATP hydrolase activity do not have two such mutations at once (Deltoro et al., 2016;Hu et al., 2012). Thus, the notion that TbPH1 has an inactive kinesin motor domain is likely but empirically unproven. Unfortunately, we are not able to corroborate this result as regulatable RNAi-cell lines targeting FP45 were reported to be unstable, precluding functional analysis (Gheiratmand et al., 2013). Indeed, we have also encountered problems in obtaining robust phenotypes downregulating TbPH1 and TbKifX2A individually or simultaneously upon extended growth and/or cryopreservation. Thus, from our and other's experience, complete ablation of this cohort of MtQassociated proteins is difficult, which complicates their functional analysis.
We have also observed some morphological defects affecting a minor proportion of cells upon TbPH1/TbKifX2A downregulation that are at this time difficult to explain. One is the emergence of anucleate zoids, intensifying at the same RNAi time point when we observe the flagellar pocket expansion. At this time, we rule out that this is due to cytokinesis defects, as this would also give rise to a more conspicuous enrichment of 2N1K or multinucleate cells, which we did not observe. In the final RNAi time point examined here, we also observed cells exhibiting posterior-end protrusions, resembling the 'glove phenotype' observed upon depletion of two TTLL enzymes, which are responsible for post-translational modifications appending the C-terminal tails of α-and β-tubulin (Jentzsch et al., 2020). As neither TTLL12A (TTLL6A was not among the detected proteins, Dataset S3) nor the modifications both enzymes catalyzes are affected by TbPH1/TbKifX2Asilencing, we conclude that these protrusions arise from a different mechanism.
Altogether, we conclude that TbPH1 and TbKifX2A play a still undefined role in maintaining compact flagellar pocket morphology. Whether one or both spuriously occurring defects are a direct consequence of the primary flagellar pocket expansion phenotype or due to incomplete ablation of TbPH1 and TbKifX2A remains unknown. Another possibility is that the cytosolic pool of TbPH1 and/ or TbKifX2A may play a role in posterior end morphogenesis. What we can say is that it is unlikely that they are arising from secondary depletion of proteins other than the targeted ones or the downreg- TbKifX2A is the first motor protein that has been demonstrated to localize to the MtQ. This and our finding that TbKifX2C partially interacts with the MtQ-binding TbPH1 suggests the exciting possibility that the X2 family s.s. kinesins coincides with the occurrence of the MtQ, a cytoskeletal component that is considered to be part of the FAZ and restricted to trypanosomatids (Sunter & Gull, 2016).
It has been suggested that the MtQ contains β-tubulin with a unique but still unknown modification that is recognized by the 1B41 monoclonal antibody Gallo & Precigout, 1988), which may at least partially explain why these kinesins and TbPH1 perferentially bind to the MtQ over the corset MTs. This specificity may be mediated by interaction with TbPH1, perhaps explaining this instance of a kinesin interacting with a catalytically inactive kinesin-like protein. An example of this phenomenon is the budding yeast Kar3 kinesin interacting with kinesin-like Vik1, which confers to the Kar3/Vik1 complex a strong interaction with a microtubule (Allingham et al., 2007).
Another striking feature of the three X2 family s.s. kinesin paralogs in T. brucei is their apparent regulation in the long slender bloodstream and procyclic stages, with TbKifX2A being solely upregulated in the latter. In contrast, TbPH1 exhibits more consistent expression levels between these two stages. This suggests that the X2 s.s. kinesin repertoire is remodeled to adapt each stage to their specific milieus. Perhaps this regulation is due to the higher endocytic flux in long slender bloodstream forms in comparison with the procyclic forms, a process that occurs via the flagellar pocket . This is certainly a worthy line of future investigation.
Given that the X2 family s.s. kinesin may associate with the MtQ in other trypanosomatids, it will be also interesting to see whether the X2 kinesins have adapted to the presence of the cytostomecytopharynx complex, an organelle that is closely associated with the MtQ in Trypansoma cruzi (Alcantara et al., 2017) and the basal trypanosomatid P. confusum (Skalický et al., 2017). While our results indicate that TbKifX2A or TbPH1 do not contribute to the biogenesis of the MtQ or FAZ, the expansion of the flagellar pocket we observe in cells depleted for these proteins has led us hypothesize that they may contribute to how the MtQ operates in trypanosomatids.
Thus, our study represents a starting point for investigation of this kinetoplastid-specific family of kinesins.

| Plasmids and generation of cell lines
For epitope tagging at the endogenous locus with HA and V5, long PCR products were generated with primers detailed in Table 1 using pPOT-V5-HygR and pPOT-HA-NeoR vectors as templates (based on [Kaurov et al., 2018] and modified from pPOTv4 [Dean et al., 2015]).
For tagging with the BioID2 ligase (Kim et al., 2016), a modified version of pPOTv7-Blast-mNG (Pyrih et al., 2020)  into BamHI-HindIII sites of p2T7-177 (Wickstead et al., 2002) in a three way ligation using XbaI as the restriction site in between the two RNAi fragments. The plasmid was linearized with NotI before electroporation.

| RNAi and growth curves
Procyclic form cell lines for depletion of either TbPH1 or TbKifX2A were induced at a density of 2 × 10 6 cells/ml with 1 μg/ml doxycycline and growth monitored daily with a Z2 Coulter Counter Analyzer (Beckman Coulter). Cells were diluted back to the initial cell density every day. Cell lines carrying the vector for simultaneous depletion of TbPH1 and TbKifX2A were induced at a density of 5 × 10 5 cells/ ml with 1 μg/ml doxycycline and growth monitored daily with a Neubauer counting chamber. Cells were diluted back to the initial cell density every 2 days.

| Flow cytometry
For flow cytometry, 1-2 × 10 6 cells were fixed in 30% methanol-PBS at least overnight and up to seven days at 4°C. Samples were washed once in PBS-T (0.5% Triton) and subsequently incubated at 37°C for 1 h in propidium iodide staining solution (10 μg/ml propidium iodide and 9.6 μg/ml RNaseA in PBS-T, both from Sigma).

| Cell fractionation via 'microtubule sieving'
To entrap non-soluble and cytoskeletal proteins within the MT corset, a 'MT sieving' technique as described in Fritz et al. (2015) was employed. The experiment was conducted exactly as described before in (Fritz et al., 2015; see also Figure S1).
Aliquots were taken of the different steps for western blot analysis.
For IP of TbKifX2A-HA the same protocol was followed except that the antibody used was anti-HA (ThermoScientific) instead of anti-V5.
For IP of potential TbPH1 interactors, 2 × 10 8 cells expressing FP45-V5 and TbKifX2C-V5 in addition to TbPH1-BioID2-HA were lysed and processed exactly as described in the BioID section below except that the beads used were paramagnetic V5-Trap® beads (Chromotek). Samples corresponding to 2% of the total (start and flow through fractions) and 20% of the total (beads fraction) were loaded on SDS-PAGE gels, blotted onto PVDF membranes and probed with anti-HA and anti-V5 antibodies as described below.

| SYPRO ruby staining
Following electrophoresis, polyacrylamide gels were placed into fixing solution (7% glacial acetic acid, 50% methanol) for 30 min at room temperature. Following another, identical fixation step, the gel was incubated with SYPRO Ruby solution overnight at room temperature. A washing step in washing solution (7% glacial acetic acid, 10% methanol) for 30 min at room temperature was followed by three 5 min washes in milliQ water before imaging on a ChemiDoc MP (Bio-Rad).

| Immunofluorescence and cytoskeleton extraction
Approximately 2 × 10 6 cells were used per slide. Cells were washed in PBS and fixed in 2.3% PFA in PBS for approximately 30 min before being transferred to microscopy slides (ThermoScientific). Following neutralization in 0.1 M glycine in PBS, slides were incubated in methanol at −20°C overnight for permeabilization. Slides were then rehydrated in PBS and blocked in 1% BSA in PBS for 1 h followed by incubation with primary antibody for 1 h in a humid chamber (see Table 2 for antibodies used in this study). The slides were then washed three times with PBS before incubation in AlexaFluorconjugated secondary antibody (goat anti-rabbit/mouse, used at 1:1000; Invitrogen). Following three further washes in PBS, a drop of ProLong Gold Antifade reagent with DAPI (ThermoScientific) was added, a coverslip applied, and the slide sealed with nail polish.
Slides were imaged with a Zeiss Axioscope or an Olympus FluoView Fv1000 confocal microscope.
For immunofluorescence analysis of cells extracted using the 'MT sieving' fractionation method described above, the pellet fraction P2 was fixed in 2.3% PFA as described. All subsequent steps were the same as for whole cell immunofluorescence.

TA B L E 2 Antibodies used in immunofluorescence
To extract cytoskeletons conventionally, approximately 1 × 10 7 PCF cells were collected and centrifuged at room temperature for 5 min at 800× g and the supernatant discarded. Thereafter, the cell pellet was washed once in PBS and after resuspension in PBS applied in a drop-wise fashion to Superfrost plus® slides (ThermoScientific). The cells were left to settle before the addition of PEME buffer (100 mM Pipes, pH 6.9, 1 mM MgSO 4 , 2 mM EGTA, 0.1 mM EDTA) containing 0.5% (v/v) NP-40 (Igepal) for 10 s. This cytoskeleton extraction was followed by a 10 min fixation step in 4% PFA. All subsequent steps were as described above for whole cell immunofluorescence.

| Scanning and transmission electron microscopy
For TEM, cell pellets were high-pressure frozen, freeze substituted in the presence of 2% OsO4 in acetone, and embedded into Spi-Pon812 resin (SPI) as described previously (Kaurov et al., 2018). For SEM of the TbPH1/TbKifX2A RNAi cell line, cells were fixed in 2.5% glutaraldehyde in 100 mM PBS overnight at 4°C and then spotted onto poly-l-lysine-coated glass cover slips. The cells were post-fixed in 2% osmium tetroxide in 100 mM PBS for 1 h at room temperature and finally washed in the same buffer. After dehydration, cells were critical-point dried, coated with gold palladium and imaged.
Prior to imaging pieces of gels containing stained expanded cells were transferred to a glass-bottom dish coated with poly-L-lysine, and imaged using a Leica TCS SP8 confocal microscope with an HC PL apochromatic 63x/NA 1.40 oil immersion objective. Z-stacks were acquired with the step size of 100 nm. Confocal z-stacks were processed in Fiji (Schindelin et al., 2012). 3D reconstructions were performed in the 3D viewer plugin.

| Flagellar pocket diameter measurements
For FP diameter measurements, random DIC images were taken of

| Mass spectroscopy analysis of captured biotinylated proteins
Trypsin-digestion of captured biotinylated proteins was performed on bead prior to liquid chromatography-tandem mass spectroscopy (LC-MS/MS) as previously described (Pyrih et al., 2020).
Carbamidomethylation of cysteine was set as a fixed modification and oxidation of methionine and N-terminal protein acetylation were allowed as variable modifications. The experimental design included matching between runs for biological replicates. Peptides were required to be at least seven amino acids in length, with false discovery rates (FDRs) of 0.01 calculated at the levels of peptides, proteins and modification sites based on the number of hits against the reversed sequence database. The obtained data were subsequently processed in Perseus version 1.6.14 as described in Zoltner et al. (2020).

| Mass spectroscopy analysis of wholecell lysates
TbPH1/TbKifX2A RNAi cells were induced for 3 and 5 days with doxycycline, 5 × 10 7 cells per replicate collected by centrifugation, and washed once in 1x PBS before being snap-frozen in liquid nitrogen. Non-induced and parental control samples were also processed similarly as a control. MS LFQ analysis was performed on biological triplicates. For sample prep, SP4 (Solvent Precipitation SP3) protocol without beads was used (Johnston et al., 2021). Briefly, cell pellets (with a minimum of 2 in at least one group). The obtained data (comparing TbPH1/TbKifX2A cells at T3 and T5 with T0) was additionally processed in Perseus version 1.6.14 as described previously (Zoltner et al., 2020).

| Bioinformatic analyses
BLASTP search was performed in TriTrypDB using KIFX2A amino acid sequence as a search query, returning hits with E = values <1e-20. The 12 trypanosomatid predicted proteomes listed in Table S1A was searched. These 12 proteomes were chosen based on their distribution within the consensus trypanosomatid phylogenetic tree presented in Lukeš et al. (2018).
For constructing the kinesin phylogenetic tree depicted in Figure 7a, protein sequences that match the Pfam "kinesin motor domain" model (PF00225) from the same 12 predicted proteomes were mined using HMMER v3.3 (Eddy, 2009). All sequences with an E-value <1e-40 (listed in Table S1C) were taken for MAFFT alignment (Katoh et al., 2002). Regions of the alignment that comprised of >50% were removed. The phylogenetic tree based on the trimmed alignment was inferred with IQTree v1.3.12 with JTT + F + G4 chosen as the best fit model according to BIC (Nguyen et al., 2015).
Branch support was estimated using ultrafast bootstrap method.

ACK N OWLED G M ENTS
We

CO N FLI C T O F I NTE R E S T
The authors declare that there's no conflict of interest.

DATA AVA I L A B I L I T Y S TAT E M E N T
The mass spectrometry proteomics data have been deposited to the ProteomeXchange Consortium via the PRIDE partner repository with the dataset identifiers PXD025802 (BioID) and PXD033717 (RNAi proteomes).