Flagellar targeting of an arginine kinase requires a conserved lipidated intraflagellar transport (LIFT) pathway in Trypanosoma brucei

Both intraflagellar transport (IFT) and lipidated intraflagellar transport (LIFT) pathways are essential for cilia/flagella biogenesis, motility and sensory functions. In the LIFT pathway, lipidated cargoes are transported into the cilia through the coordinated actions of cargo carrier proteins such as Unc119 or PDE6δ, as well as small GTPases Arl13b and Arl3 in the cilium. Our previous studies revealed a single Arl13b ortholog in the evolutionarily divergent Trypanosoma brucei. TbArl13 catalyses two TbArl3 homologs, TbArl3A and TbArl3C, suggesting the presence of a conserved LIFT pathway in these protozoan parasites. Only a single homolog to the cargo carrier protein Unc119 was identified in T. brucei genome, but its function in lipidated protein transport has not been characterized. In this study, we exploited the proximity-based biotinylation approach to identify binding partners of TbUnc119. We showed that TbUnc119 binds to a flagellar arginine kinase TbAK3 in a myristoylation-dependent manner and is responsible for its targeting and enrichment in the flagellum. Interestingly, only TbArl3A, but not TbArl3C interacts with TbUnc119 in a GTP-dependant manner, suggesting functional specialization of Arl3-GTPases in T. brucei. This study establishes the function of TbUnc119 as a myristoylated cargo carrier and supports the presence of a conserved LIFT pathway in T. brucei.


Introduction
TbUnc119 formed a clad with other Unc119 homologs distinct from PDE6δ proteins. Unc119 is highly conserved among kinetoplastids (Fig. S1B). Notably, PDE6δ homologue could not be found despite extensive searches of the T. brucei genome. Further searches using NCBI BLAST confirmed the absence of PDE6δ in all Kinetoplastid members and most single cellular eukaryotes we have examined, with the possible exception of Paramecium tetraurelia (28). Together these results suggest that Unc119 is likely the only conserved lipidated protein carrier belonging to the Unc119 supergene family (28) in T. brucei and other Kinetoplastid organisms.
The knockdown of TbUnc119 did not produce detectable growth defects in the insect-infectious procyclic form (PCF) cells (Fig. S2A), which corroborates the previous study (25). In the mammal infectious bloodstream form (BSF) T. brucei, a mild growth delay was consistently observed post TbUnc119-RNAi induction (Fig. S2B). Similar to TbUnc119-silencing in the PCF cells, no significant phenotypic changes were observed in the BSF cells. Taken together, these results confirmed that TbUnc119 is not essential for procyclic and BSF cell survival or flagellar biogenesis in culture. These results are also consistent with the non-lethal mutant phenotypes of Unc119 orthologues previously reported in C. elegans (29) or zebrafish (30).

Identification of TbUnc119-interacting proteins by proximity-based biotinylation
In both C. elegans and mammals, Unc119 is characterized as a cargo carrier/chaperone involved in intraflagellar transport of myristoylated ciliary proteins, and Unc119 association with the cargo is regulated by the Arl13b-Arl3 pathway (3,6). C. elegans Unc119 additionally functions in stabilizing the Arl13b-Arl3 interaction, suggesting functional divergence of Unc119 in lower eukaryotes (20). The function of the LIFT pathway and its flagellar cargoes have never been examined in the evolutionarilydivergent T. brucei, we therefore decided to revisit the function of TbUnc119, by investigating its interacting proteins.
LC/MS-MS. A total of 138 and 43 candidates from 3HA-BioID2-TbUnc119 and TbUnc119-BioID2-HA cells, respectively, were identified (Fig. 1A). Among them, 32 high-confidence candidates were found in both 3HA-BioID2-TbUnc119 and TbUnc119-BioID2-HA cells ( Fig. 1B; Table S1). It is not clear why 3HA-BioID2-TbUnc119 had more hits identified than TbUnc119-BioID2-HA cells. One possibility is that the position of the BioID2 tag at the C-terminus of TbUnc119 may interfere with its interaction with other proteins.
To test if TbAK3 interacted with TbUnc119, TbAK3 was fused to a small BB2 tag (39) at the Cterminus and co-expressed with GFP-TbUnc119. Immunoprecipitation was then performed using GFP nAb-conjugated beads. TbAK3-BB2 co-immunoprecipitated with GFP-TbUnc119, but not GFP only ( Fig. 2A). As another control, TbAK1-BB2 did not co-immunoprecipitate with GFP-TbUnc119, suggesting specific interaction between TbAK3 and TbUnc119. The TbAK3-TbUnc119 interaction is myristoylation-dependent, as TbAK3(G2A)-BB2 failed to co-immunoprecipitate with GFP-TbUnc119. The myristoylation mutation also disrupted the flagellar localization of TbAK3 (Fig. 2, B and C). Together these results indicate that the flagellar localization of TbAK3 and its interaction with TbUnc119 are both myristoylation-dependent.
Next we asked whether TbAK3 flagellar targeting required TbUnc119. We tagged one endogenous allele of TbAK3 with fluorescent reporter mNeonGreen in the TbUnc119-RNAi cell line. In control cells without TbUnc119-RNAi induction, TbAK3-mNeonGreen was enriched in the flagella with weak signal in the cytosol (Fig. 3A). Upon TbUnc119-RNAi induction, the flagellum enrichment of TbAK3 diminished and increased cytosolic signal was observed (Fig. 3A, quantitated in Fig. 3B). To further establish the flagellar targeting dynamics of TbAK3, we generated a stable cell line with tetracyclineinducible TbUnc119-RNAi and cumate-inducible expression of TbAK3-BB2. The expression of TbAK3-BB2 was induced for a fixed period of 24 hours, at different times post TbUnc119-RNAi induction. A gradual reduction in flagellar TbAK3 and an increase in the cytoplasmic TbAK3 was observed over the course of TbUnc119-RNAi (Fig. 3C, quantitated in 3D), despite similar expression levels of TbAK3 at different timepoints (Fig. 3E). These results suggested that the loss of TbUnc119 inhibited the entry of TbAK3 into the flagellum.

TbUnc119 binds to TbSMP1-1, but is not required for TbSMP1-1 intracellular distribution
To address whether TbUnc119 may have non-ciliary functions as observed with animal Unc119 orthologs, we examined TbUnc119-BioID candidates for non-ciliary myristoylated proteins. The small myristoylated protein TbSMP1-1 (encoded by Tb927.1.2230) contains an N-terminal myristoylation site (G2) and is enriched at the cell membrane of T. brucei (37). The interaction between TbSMP1-1 and TbUnc119 was confirmed by co-immunoprecipitation (Fig. 4A). While TbSMP1-1-GFP was enriched in cell periphery consistent with plasma membrane association, TbSMP1-1(G2A)-GFP mutant lost cell membrane enrichment and was found throughout the cytosol (Fig. 4, B and C). This observation was further confirmed by profiling the fluorescent intensity across randomly selected TbSMP1-1-GFP and TbSMP1-1(G2A)-GFP expressing cells (Fig. 4, B and C). Silencing of TbUnc119 however, had no detectable effects on TbSMP1-1-GFP distribution in the cell (Fig. 4D).

TbUnc119 interacts specifically with Arl3A in a GTP-dependent manner
The best characterized function of Unc119 is in the context of the LIFT pathway as a carrier for myristoylated cargoes. Once the cargo-Unc119 complex is inside of the ciliary lumen, Arl3-GTP acts as a displacement factor, binds to Unc119 and releases the cargo (4,6). Unlike vertebrates that contain only a single Arl3 protein, T. brucei has two Arl3 homologues, TbArl3A and TbArl3C, that are both associated with the flagellum and exhibit flagellar phenotypes when overexpressed as GTP-locked forms (24). Interestingly, both TbArl3A and TbArl3C were found in the TbUnc119 BioID screen, albeit only in the 3HA-BioID2-TbUnc119 cells.
Our previous studies have shown that TbArl13 acts as a GEF on TbArl3A, Our previous studies have shown that TbArl13 acts as a GEF on TbArl3A,as has been reported in mammals (3,24). We hypothesized that in cells depleted of TbArl13, the level of TbArl3A-GTP in T. brucei should decrease, which in turn will affect the interaction between TbUnc119 and TbArl3A. To test this, stable cells expressing tetracycline-inducible TbArl13-RNAi and cumate-inducible TbUnc119-YFP and TbArl3A-BB2 were generated. While TbArl3A-BB2 co-immunoprecipitated with TbUnc119-YFP in control, their interaction was abolished in TbArl13-RNAi (Fig. 6C). Together, these results demonstrate specific interaction of TbUnc119 with only one of the TbArl3 homologs, TbArl3A, in a GTP-dependent manner, suggesting that TbUnc119 is an effector of TbArl3A that is regulated by TbArl13.

Discussion
In this study, we revisited the functions of TbUnc119 in light of recent understanding of the LIFT pathway in ciliary biogenesis. Overall our results supported a function of TbUnc119 in flagellar targeting of myristoylated TbAK3. This is consistent with the cargo carrier function of Unc119 observed in mammals and C. elegans (6,19). There are, however, some differences between TbUnc119 and its higher eukaryotic counterparts. In C. elegans, Unc119 forms mutual interactions with both Arl3 and Arl13, facilitating GTP loading to Arl3. As the interaction between Unc119 and Arl3 is GTPindependent, Unc119 is unlikely an Arl3 effector in C. elegans (20). In T. brucei, no detectable interaction was observed between TbUnc119 and TbArl13. TbUnc119 directly interacts with TbArl3A in a GTP-dependent fashion, similar to mammalian Unc119 (5,6). Thus C. elegans Unc119 may represent a case of functional divergence, though it appeared more conserved with mammalian Unc119 in the phylogenetic analyses (Fig. S1A). One important difference between T. brucei and mammalian Unc119 is the lack of Unc119-Arl2 interaction in T. brucei. In mammals, Arl2 is shown to interact with Unc119 and displace low-affinity cargoes in the cytosol (5). TbArl2 is essential for cytokinesis in T. brucei (40) but this effect is unlikely mediated by TbUnc119.
In T. brucei, we showed that the binding between TbUnc119 and TbAK3 depended on the myristoylation state of the cargo. TbUnc119 is able to bind to other myristoylated proteins such as TbSMP1-1, though the function of this binding remained unclear. Depletion of TbUnc119 had no obvious effects on TbSMP1-1 distribution. Considering the lack of growth phenotypes in cultured TbUnc119-RNAi PCF and BSF cells, depletion of TbUnc119 is unlikely to cause gross perturbation in myristoylated protein distribution or functions, unlike those observed in cells with the myristoylation pathway inhibited (41-43). Depletion of TbAK3, a flagellar protein identified as TbUnc119 cargo in this study, is also shown to be dispensable for cell growth in culture (34). However, TbAK3-depletion impairs cell motility and parasite infectivity in the tsetse flies (34), suggesting that TbAK3 is crucial for flagellar function and parasite development in tsetse fly. TbUnc119 is thus expected to be important for parasite survival in hosts, which remained to be tested. Furthermore, TbUnc119 has many cytoplasmic BioID candidates without predicted myristoylation modification. During the bioinformatic analyses, we could not identify a canonical homologue of PDE6δ, a prenylated cargo carrier, in T. brucei and most other single-cellular organisms. Yet protein prenylation and the molecular machinery is clearly present in T. brucei (44-46) and several other protists (47). This raised an interesting possibility that TbUnc119 and other protist Unc119 orthologs, may be able to carry other lipidated cargoes, particularly prenylated proteins. This possibility should be examined in the future as little is currently known about prenylated targets in T. brucei.
A single Arl3 homolog is present in mammals and C. elegans, and it is known to interact with and regulate IFT components in addition to its function in displacing cargoes associated with Unc119

Bioinformatic analyses
The amino acid sequences of Unc119 and PDE6δ from various model organisms were obtained from UniProt, Tritrypdb and NCBI protein database. Multi-sequence alignments were performed using MUSCLE (50). The output of multi-sequence alignments was formatted using Multiple Align Show of the Sequence Manipulation Suite (JavaScript application) (51). For phylogenetic analyses, the Unc119 and PDE6δ sequences were aligned using MAFFT (LINSI). ProtTest (v. 3.4.2) (52) was used for model selection. Maximum likelihood tree was generated using RAxML (v. 8.2.10) and Multiparametric bootstrapping was done using automatic bootstrapping option (autoMRE).

Expression constructs, Cell culture and Transfection
All T. brucei sequences used in this study were retrieved from the Tritryp database (http://tritrypdb.org/tritrypdb/). RNAi target sequences were selected using RNAit (53). The details of plasmid constructs used in this study are summarized in Table S2. For immunofluorescence assays, T. brucei cells expressing fluorescent or small tags were washed and resuspended in 1 × PBS and attached to coverslips. Cells were fixed with 4% PFA and permeabilized with 0.25% Triton X-100 unless otherwise stated. DNA was stained with DAPI (2.5 µg/ml). Images were captured by Zeiss Axio Observer Z1 fluorescence microscope with a 63× objective (NA=1.4) and a CoolSNAP HQ2 CCD camera (Photometrics).

Image quantification and statistical analyses
For quantifications, images acquired using fixed exposure conditions were processed using ImageJ. Fluorescence intensity of the flagellum was performed using the plot profile function, by were calculated using two-tailed t-test with 95% confidence interval.
After 2 washes with 1× PBS, the cells were resuspended in 1× PBS supplemented with protease inhibitor cocktail (Sigma) and homogenized by sonication. The cell lysates were centrifuged at 17000 g for 15 mins at 4°C and the cleared supernatants were incubated with magnetic GFP-nAb TM beads (Allele Biotechnology) to co-precipitate GFP-fusion proteins together with their binding partners. Cells co-expressing GFP and BB2 tagged TbArl-GTPase were used as negative controls. Proteins bound to the magnetic beads were eluted by boiling in 1×Laemmli buffer and analysed by SDS-PAGE followed by immunoblotting.
GST-Arl GTPases and His-TbUnc119 fusion proteins were expressed in E. coli (0.1mM IPTG). Purification of these tagged proteins was performed using Ni-NTA beads (Qiagen) or Glutathione Sepharose TM 4B beads (GE healthcare) according to manufacturer's instructions. His-TbUnc119 bound to Ni-NTA beads were incubated with cell lysates containing GST-Arl3A, GST-Arl3C and GST only. Alternatively, Glutathione Sepharose TM 4B beads bound to GST or GST-fusions including GST-Arl2 were incubated with His-TbUnc119. Interaction between TbUnc119 and Arl-GTPases was then examined by SDS-PAGE followed by immunoblotting.

Nucleotide exchange assay
GST-TbArl3A bound to glutathione beads was treated with alkaline phosphatase (ALP,10 units/ml) to enzymatically dephosphorylate purified GST-TbArl3A (containing a mixture of GTP-or GDP-bound forms) to nascent guanosine state in 1 ml exchange buffer (20 mM HEPES pH 7.4, 1mM MgCl2, 1 mM DTT) supplemented with 50 mM EDTA for one hour at room temperature. The beads were washed thrice with 1ml exchange buffer and then incubated with GTP (100 µM), GDP (100 µM) or no nucleotide, in the presence of 100 mM MgCl2 for 1 hour at room temperature. Beads were washed once each with exchange buffer and 1 × PBS and incubated with purified His-TbUnc119 for 4 hours at 4°C.
Beads were then washed thrice each with 1% Triton X-100 in 1× PBS followed by 1× PBS and bound proteins eluted by boiling in 1×Laemmli buffer.

Antibodies for immunostaining and immunoblots
Antibodies used for immunostaining were anti-HA (1:500; Santa Cruz, #sc-7392) and streptavidin-Alexa Fluor 568 (1:2000). For immunoblots, the samples were boiled in 1×Laemmli buffer and separated on SDS-PAGE. Antibodies used for probing were anti-YFP (1:1000, rabbit), anti-mCherry             TbUnc119 and TbAK1-BB2, TbAK3-BB2 or  TbAK3(G2A)-BB2, respectively, were incubated with GFP-nAb beads. Proteins bound to the beads were fractionated on SDS-PAGE followed by immune-blotting with anti-GFP and anti-BB2 antibodies. Cells co-expressing GFP and TbAK3-BB2 were used as a negative control. Asterisks indicate possible degradation products of GFP-TbUnc119. Input: 3% of cell lysates. (B, C) Cells expressing TbAK3-YFP or TbAK3(G2A)-YFP were viewed after fixation with 4% PFA. Nuclear and kinetoplast DNA were stained with DAPI (blue). Scale bar: 5 μm.   4. The cell membrane association of TbSMP1-1 is myristoylation-dependent but TbUnc119independent. (A) TbSMP1-1-BB2 co-immunoprecipitated with GFP-TbUnc119, but not GFP only. (B, C, D) Cells expressing TbSMP1-1-GFP (B) or TbSMP1-1(G2A)-GFP (C) were immobilized on agarose gel and imaged live, to best visualize the plasma membrane association of TbSMP1-1-GFP. The intracellular distribution of TbSMP1-1-GFP was also monitored in live cells induced for TbUnc119-RNAi (D). All GFP images were collected at constant exposure time. The distribution of TbSMP1-1-GFP in control and TbUnc119-RNAi cells was measured using plot profiling. A line of 5µm length was drawn across the entire cell body, encompassing cell membranes at both ends, and fluorescence intensity along the length of this line was plotted. 10 representative cells were shown for control and TbUnc119-RNAi cells. Insets show enlarged images of a representative cell from each sample that were selected for intensity measurements. Scale bar: 5 µm.  GST-TbArl3A was treated with or without alkaline phosphatase (ALP) and then loaded with GTP or GDP was used to pull down His-TbUnc119. (C) Control and TbArl13-RNAi cells co-expressing TbUnc119-YFP and TbArl3A-BB2 were immunoprecipitated with GFP-nAb beads. TbArl13-RNAi did not affect TbUnc119-YFP and TbArl3A-BB2 expression levels, but their binding was abolished. Input: 5% of cell lysates; UB: 5% of unbound fraction; B: proteins eluted from GFP-nAb beads.