Roles of end‐binding 1 protein and gamma‐tubulin small complex in cytokinesis and flagella formation of Giardia lamblia

Abstract Giardia lamblia is a unicellular organism with two nuclei, a median body, eight flagella, and an adhesive disk. γ‐Tubulin is a microtubule (MT)‐nucleating protein that functions in the γ‐tubulin small complex (γ‐TuSC) in budding yeast. In this study, G. lamblia γ‐tubulin (Glγ‐tubulin) was found to bind to another MT‐binding protein, namely G. lamblia end‐binding protein 1 (GlEB1), via both in vivo and in vitro assays. Hemagglutinin (HA)‐tagged Glγ‐tubulin localized to the basal bodies, axonemes, and median bodies of G. lamblia trophozoites. The knockdown of Glγ‐tubulin expression using an anti‐Glγ‐tubulin morpholino resulted in a decreased growth rate and an increased failed cytokinesis cells of Giardia. The formation of median bodies was affected, and the central pair of MTs in flagella was frequently missing in the Giardia treated with an anti‐Glγ‐tubulin morpholino. G. lamblia γ‐tubulin complex protein 2 (GlGCP2) and GlGCP3, which are putative components of γ‐TuSC, were co‐immunoprecipitated with HA‐tagged Glγ‐tubulin in Giardia extracts. The knockdown of GlGCP2 and GlGCP3 expression also resulted in decreased formation of both the median body and flagella MTs. Knockdown of Glγ‐tubulin, GlGCP2, and GlGCP3 expression affected localization of GlEB1 in G. lamblia. In addition, decreased level of GlEB1 caused reduced formation of median body and the central pair of flagella MTs. These results indicated that Glγ‐tubulin plays a role in MT nucleation for median body formation and flagella biogenesis as a component of Glγ‐TuSC in Giardia and GlEB1 may be involved in this process.

The exact positioning of these organelles should be regulated via the proper function of MTs when this organism divides (Desai & Mitchison, 1997). G. lamblia EB1 (GlEB1) was found at the flagella tips, median bodies, nuclear membranes, and mitotic spindles (Dawson et al., 2007;Kim, Nagami, Lee, & Park, 2014). GlEB1 was also found to complement a BIM1 mutant of Saccharomyces cerevisiae, that is, to induce the proper positioning of the nucleus (Kim et al., 2008). In vitro studies have demonstrated that GlEB1 can be phosphorylated by G. lamblia aurora kinase (GlAK) (Kim, Lee, Lee, & Park, 2017). The ectopic expression of a mutant GlEB1 in which Ser148 was changed to Ala resulted in an increased number of Giardia cells with division defects. The treatment of G. lamblia with an aurora kinase inhibitor triggered cytokinesis defects, and the ectopic expression of a phosphomimetic mutant GlEB1 in which Ser148 was changed to Asp rescued the defects in Giardia cell division caused by that inhibitor, even though it has not yet been determined whether GlEB1 is a direct substrate of GlAK.
In S. cerevisiae, a mutant lacking the C-terminal four residues of γ-tubulin was defective in the proper recruitment of the Kar9p-Bim1p complex (Cuschieri, Miller, & Vogel, 2006). In addition, the overexpression of the EB1 ortholog Bim1p, but not Kar9p, rescued the mutant with defective γ-tubulin. In this study, we examined whether GlEB1 and G. lamblia γ-tubulin (Glγ-tubulin) had any functional relationship in G. lamblia by measuring their physical association.
Little information is available on the γ-tubulin of G. lamblia (Nohynková, Draberb, Reischigc, & Kulda, 2000). While Glγ-tubulin was mainly found in the basal bodies/axonemes of flagella in G. lamblia cells under all the stages, this protein is transiently localized in the centers of mitotic spindles only in the dividing cells. In this study, we examined the roles of γ-tubulin and GCPs in MT modulation in G. lamblia.

| Giardia strain and cultivation
Giardia lamblia WB strain (ATCC 30957; American Type Culture Collection, Manassas, VA) were cultured at 37°C as described in the previous paper (Kim et al., 2014).

| In vitro co-immunoprecipitation assays
The interaction between GlEB1 and Glγ-tubulin was monitored by using the BD Matchmaker Co-IP Kit (Clontech, Mountain View, CA). The pGBKEB1 produced Myc-tagged GlEB1 protein (Kim et al., 2017). A 1,476-bp DNA fragment encoding Glγ-tubulin was cloned into pGADT7 (Clontech) to produce pGADγ-tubulin, in which Glγ-tubulin was expressed in an hemagglutinin (HA)-tagged form. [ 35 S]methionine-labeled Myc-tagged GlEB1 and HA-tagged Glγ-tubulin were synthesized in vitro using the TNT ® Coupled Reticulocyte Lysate Systems (Promega, Madison, WI). These two proteins were mixed in two separate tubes; monoclonal antibodies for the Myc epitope were added into one tube, while polyclonal antibodies specific for the HA epitope were added to the other tube. These antibodies/ labeled protein complexes were precipitated with protein A beads.
The eluted proteins were separated by 12% sodium dodecyl sulfate polyacrylamide gel electrophoresis (SDS-PAGE) and observed using autoradiography.
Glutathione Sepharose ® 4B resin coupled with 5 µg of either purified GST or the GST-GlEB1 protein was incubated with an E. coli lysate expressing His-tagged Glγ-tubulin in a binding buffer (20 mM Tris-HCl, 500 mM NaCl, 0.1% Triton X-100, pH 7.5). After an overnight incubation at 4°C, the resins were washed three times with washing buffer (10 mM Tris-HCl, 150 mM NaCl, 0.1% Triton X-100, pH 7.5) and eluted for western blot analysis.

| Construction of HA epitope-tagged Glγtubulin and HA-tagged GlGCP3′
An 1,676-bp DNA fragment of the glγ-tubulin gene, which is composed of the promoter region near the N-terminus and an HA epitope at the C-terminus of the ORF, was amplified from Giardia genomic DNA by polymerase chain reaction (PCR) using two primers, namely γ-tubulin-NcoI-F and γ-tubulin-HAX3-R (Table 1). The NcoI and NotI sites were used for cloning into the plasmid pGFP.

| Co-immunoprecipitation of GlEB1 or
GlGCP2 or GlGCP3 and Glγ-tubulin with anti-HA antibodies from Giardia carrying pGlγ-tubulinHAX3. pac cells were lysed by sonication with 2-s pulses at 20% amplitude (Pronextech, Seoul, Korea). The supernatants were obtained by centrifugation at 16,000 x g for 20 min and were then precleared with protein A/G beads (Pierce, Waltham, MA) for 1 hr at 4°C.
One milligram of each lysate was reacted with anti-HA agarose beads (Sigma-Aldrich) at 4°C overnight. After two washes with cross-linking wash buffer (50 mM Tris-HCl, pH 7.4, 150 mM NaCl, and 1% Triton X-100), the beads were resuspended with SDSloading buffer and boiled for 5 min. Twenty micrograms of the eluted fraction was analyzed by western blotting using anti-HA (Sigma-Aldrich) to monitor the presence of HA-tagged Glγ-tubulin.
The same sample was treated with antibodies against Glγ-tubulin, GlEB1 (Kim et al., 2017), GlGCP2, or GlGCP3 antibodies. As a control, Giardia caring HA-tagged Glγ-tubulin extracts were incubated with anti-mouse IgG conjugated to Sepharose ® beads (Cell Signaling, Danvers, MA) instead of anti-HA agarose beads.
The primers used to make recombinant Glγ-tubulin (rGlγ-tubulin) were made based on GiardiaDB ORF No: GL50803_114218. A 1,476-bp DNA fragment was amplified by PCR using the two primers rγ-tubulin-F-ERI and rγ-tubulin-R-XhoI (Table 1). The amplified fragment was then cloned into pET32b (Novagen, Darmstadt, Germany) to obtain the plasmid pETγ-tubulin ( and GlGCP3 (amino acids 121-139: KTNKLHGKSKHKSKKSIRSC), were designed. Two peptides (5 mg each) were synthesized, conjugated to keyhole limpet hemocyanin, and then used for immunizing mice or rabbit (Young In Frontier).
These polyclonal antibodies were purified using protein A or protein G resin to obtain specific IgG. The specificity of the purified antibodies was confirmed by a western blot analysis of Giardia extracts.
G. lamblia expressing HA-tagged Glγ-tubulin or G. lamblia expressing HA-tagged GlGCP3 was used to examine co-localization of Glγ-tubulin and GlGCP2 or co-localization of GlGCP2 and GlGCP3, respectively.

| Morpholino knockdown of Glγ-tubulin, GlGCP2, GlGCP3, or GlEB1 expression
Decreased Glγ-tubulin expression was observed by a knockdown experiment using morpholino as described (Carpenter & Cande, 2009). The cells were treated with 25-mer morpholino for Glγ-tubulin, which included 16 nucleotides of the Glγ-tubulin ORF and nine nucleotides upstream of the start codon (Table 1; Gene Tools, Philomath, OR). The non-specific oligomers provided by the company were used as a controls morpholino (Table 1). The lyophilized morpholino was added to 5 × 10 6 cells in 0.3 ml medium at a final concentration of 100 or 200 μM. As another negative control, an equal volume of sterile water was added to the cells. After electroporation, the cells were grown for 24 or 48 hr and then analyzed for the expression of Glγ-tubulin by western blot as described above.
The specific morpholino for GlGCP2 and GlGCP3 was also designed by Gene Tools, and their sequences are listed in Table 1.

| Measuring the growth and cell division of G. lamblia trophozoites
After 24 hr post-treatment with morpholino, the number of parasites per milliliter was determined using a hemocytometer. With G. lamblia trophozoites treated with water, a control morpholino, or the anti-Glγ-tubulin morpholino, the proportions of cells with two or four nuclei were determined to monitor cytokinesis as previously described (Hofstetrova et al., 2010). The cells were attached onto glass slides, fixed with methanol, air-dried, and then mounted in VECTASHIELD anti-fade mounting medium containing DAPI (Vector Laboratories).
The cells with two or four nuclei in each condition were counted among a total of over 300 cells.
lamblia trophozoites (controls, and the cells treated with anti-Glγ-tubulin, anti-GlGCP2, and anti-GlGCP3 morpholino) were grouped by the following phenotypes: disorganized for cytokinesis, defective in furrow formation, disable in cytokinesis, and failed abscission.

| Determination of phenotypes related to the median body and flagella of G. lamblia
For Giardia cells treated with water, control, or anti-Glγ-tubulin morpholino, the percentage of cells having a median body was determined by Giemsa staining. The cells were attached onto glass slides, air-dried, and then fixed with 100% methanol for 10 min. They were then stained with 10% Giemsa solution for 40 min and washed with distilled water. After mounting with dibutyl phthalate xylene mountant (Sigma-Aldrich), the slides were observed with an Axiovert 200 microscope (Carl Zeiss).
To measure the volume of the median bodies, the morpholinotreated cells were stained with 6-11B-1, which is the monoclonal antibodies against α-tubulin (1:600; Sigma-Aldrich), followed by a reaction with AlexaFlour 488-conjugated anti-mouse IgG (1:200; Molecular Probes). The IFA procedure was the same as described above. Samples were observed with an LSM710 laser scanning confocal microscope (Carl Zeiss), and serial sections were acquired at 0.3 μm intervals. For the measurement of median body volume, image analysis was performed using Imaris (Bitplane, South Windsor, CO).
G. lamblia cells stained with 6-11B-1 were also used to observe the effect of morpholino on flagella formation. Specifically, the length of the caudal flagella was measured using Zen 2012 (the blue edition, Carl Zeiss).

| Statistical analysis
Data are presented as the mean ± standard deviation from three independent experiments. To determine the statistical significance of these results, data were performed using Student's t tests by statis-

| Interaction of Glγ-tubulin with GlEB1
In S. cerevisiae, γ-tubulin was found to be involved in the function of Bim1p, which is orthologous to EB1 (Cuschieri et al., 2006). In the subsequent experiment, GST-tagged GlEB1 protein, GST-GlEB1, was examined to determine whether it could interact with histidine-tagged Glγ-tubulin ( Figure 1b). Glγ-tubulin was found to bind to the glutathione resin coupled with GST-GlEB1, whereas incubation with GST alone did not result in the precipitation of rGlγtubulin ( Figure 1b, lanes 2 and 3, respectively).

Using an yeast system, we confirmed the interaction between
Glγ-tubulin and GlEB1 in vivo. The pGADγ-tubulin was constructed to produce Glγ-tubulin with the activation domain of yeast Gal4p.
The AH109 strain transformed with pGBKEB1 and pGADγ-tubulin demonstrated an interaction-positive phenotype, that is, growth on the indicator plates lacking histidine, leucine, and tryptophan ( Figure 1c). Similarly, the AH109 clones with pGBK-p53 and pGAD-T show an interaction-positive phenotype on the indicator plates because of the interaction between the T-antigen and p53. However, the AH109 clones with pGBK-Lam and pGAD-T, which represented a negative control lacking any interaction, showed no growth on the indicator plates.
In this study, pGlγ-tubulinHAX3.pac was made (Figure 1d We performed an additional assay to examine whether GlEB1 could be precipitated with anti-HA antibodies from this transgenic G. lamblia expressing the HA-tagged Glγ-tubulin (Figure 1d(iii)).
Extracts of the HA-tagged Glγ-tubulin expressing cells were incubated with a resin coupled with anti-HA antibodies. As a control, the same extracts were bound to a resin coupled with mouse IgG. The Interestingly, Giardia cells at dividing stages showed the localization of Glγ-tubulin at basal bodies ( Figure 2b). In cells at anaphase, Glγ-tubulin was located at two spots, and the mitotic spindles stained with anti-α-tubulin antibodies were located between them.
On the other hand, Glγ-tubulin was detected outside of two nuclei of the telophase cells. During cytokinesis, it was found in basal bodies and axonemes of the two daughter cells. To confirm the localization of Glγ-tubulin at MTOC (Lauwaet et al., 2011), anti-Glcentrin antibodies were made, and their specificity was confirmed by an immunoreactive protein of 20 kDa in western blot analysis of Giardia F I G U R E 1 Interaction of Glγ-tubulin with GlEB1. (a) Co-immunoprecipitation of Glγ-tubulin with GlEB1. An in vitro-synthesized a Myc-tagged GlEB1 was precipitated with anti-Myc antibodies (lane 1), while Glγ-tubulin with a HA epitope was sedimented with anti-HA antibodies (lane 3). Myc-tagged GlEB1 and HA-tagged Glγ-tubulin mixture were precipitated either anti-Myc or anti-HA antibodies (lanes 2 and 4, respectively). (b) Purified GST-tagged GlEB1 proteins were incubated with E. coli lysates expressing histidine-tagged Glγ-tubulin, and these proteins were precipitated with glutathione Sepharose resin (lane 2). As a control, GST was prepared and incubated with E. coli lysates expressing histidine-tagged Glγ-tubulin (lane 1) and lysates expressing the histidine-tagged Glγ-tubulin was loaded (lane 3). The precipitated proteins were analyzed by western blotting using anti-GST or anti-histidine antibodies. The immunoreactive GST, Glγ-tubulin, and GST-GlEB1 proteins are indicated by arrows. (c) The yeast two-hybrid assay. A serial dilution of yeast cells was spotted on selective indicator plates. Yeast cells bearing pGBK-p53 and pGAD-T were used as a positive control for interaction, while yeast carrying pGBK-Lam and pGAD-T were used as a negative control. L: leucine; T: tryptophan; H: histidine (d) Co-immunoprecipitation of Glγ-tubulin with GlEB1 from HA-tagged Glγ-tubulin expressing G. lamblia lysates. (i) A schematic diagram of the plasmid pGlγ-tubulinHAX3.pac. Glγ-tubulin is expressed from its own promoter, Pglγ-tubulin, as a HA-tagged form (HAX3). Transfected cells are selected by puromycin resistance conferred by the pac gene expressed by the Pggi promoter, a promoter of the γ-giardin protein gene. (ii) The expression of HA-tagged Glγ-tubulin was confirmed by western blot analysis. Extracts were prepared from G. lamblia containing pΔ.pac (lane 1) or pGlγ-tubulinHAX3.pac (lane 2). The membrane was reacted with monoclonal mouse anti-HA (1:1,000). After deprobing in the stripping buffer, the membrane was incubated with polyclonal rat antibodies specific to PDI1 of G. lamblia (1:10,000) as loading control. (iii) Co-immunoprecipitation of Glγ-tubulin with GlEB1 from Giardia carrying pGlγ-tubulinHAX3.pac. Cell extracts containing HA-tagged Glγ-tubulin proteins were pre-cleared with protein A/G beads. As a control, Giardia extracts were incubated with Sepharose bead-conjugated anti-mouse IgG (lane 1). One milligram of lysates was reacted with anti-HA agarose beads at 4°C overnight (lane 2). Twenty micrograms of the eluted fraction was analyzed by western blot using anti-HA or anti-GlEB1 antibodies extracts (data not shown). We then performed an IFA on dividing

| Effect of Glγ-tubulin knockdown on cell division of Giardia
To define the role of Glγ-tubulin in G. lamblia, we designed an anti-Glγ-tubulin morpholino to block the translation of Glγ-tubulin mRNAs (Table 1). A control morpholino (non-specific oligomers) was also made and transfected into cells by electroporation (Table 1)

| Effect of Glγ-tubulin knockdown on the cytoskeletal structure of Giardia
Giardia cells treated with water, the control morpholino, or the anti-Glγ-tubulin morpholino and stained with Giemsa were also monitored to determine the presence of median body, which is a characteristic cytoskeletal structure of G. lamblia (Figure 4a F I G U R E 2 Localization of Glγ-tubulin and Glα-tubulin in G. lamblia expressing HA-tagged Glγ-tubulin. G. lamblia expressing HA-tagged Glγ-tubulin attached to glass slides were reacted overnight with rat anti-HA (1:100) and anti-α-tubulin antibodies (1:800). The cells were then incubated with AlexaFluor 488-conjugated anti-mouse IgG (1:100) and AlexaFluor 564-conjugated anti-mouse IgG (1:100). The cells were mounted with DAPI containing anti-fade mounting medium and observed using an LSM710 laser scanning confocal microscope (Carl Zeiss). A DIC image was acquired to show the cell morphology. Since IFAs using anti-GlGCP3 antibodies did not show any meaningful fluorescent signal under various reaction conditions, we constructed pGlGCP3HAX3part.neo (Supporting information Figure   S1a) and then used to construct transgenic trophozoites expressing the HA-tagged GlGCP3′. The expression of the HA-tagged GlGCP3′ was shown as an immunoreactive protein band of ~81 kDa in a western blot analysis of the resulting G. lamblia extracts (Supporting information Figure S1b). The intracellular level of GlPDI1 was monitored as a loading control.
Additional IFAs were performed to examine the localization of both GCP2 and GCP3 in the Giardia cells at the interphase ( Figure 5c). Giardia trophozoites expressing the HA-tagged GlGCP3 were stained with anti-HA and anti-GlGCP2 antibodies. Both GlGCP2 and GlGCP3 were found in the basal bodies, axonemes, and the median bodies of the Giardia cells.

| Phenotypes of GlGCP2 and GlGCP3 knockdown in G. lamblia
If Glγ-tubulin functions by forming Glγ-TuSC with GlGCP2 and GlGCP3, the knockdown of GlGCP2 and GlGCP3 using anti-GlGCP2 and anti-GlGCP3 morpholino would also be expected to result in a phenotype similar to that of Giardia cells with a decreased expression of Glγ-tubulin. Most of all, extracts of at 24 hr post-transfection were prepared to monitor for their intracellular levels of GlGCP2 or GlGCP3 by western blots using its specific antibodies (Figure 6a(i,ii), respectively). The  Figure   S2b). Still, none of the anterior or the caudal axonemes was found to be devoid of the central pair MTs. In case of Giardia cells treated F I G U R E 4 Effects of morpholino-mediated knockdown of Glγ-tubulin on the formation of median bodies and MT axonemes of Giardia.
(a) Median body formation. (i) For Giardia cells treated with each condition, that is, water, a control morpholino, or an anti-Glγ-tubulin morpholino, the proportions of cells with a median body were determined by Giemsa staining. +MB: cells with the median body; -MB: cells without the median body. (ii) To measure the volume of the median bodies, the cells were stained with anti-α-tubulin antibodies (1:600), followed by a reaction with AlexaFlour 488-conjugated anti-mouse IgG (1:200). The stained cells were observed with a Zeiss LSM710 laser scanning confocal microscope. For the measurement of median body volume, images were analyzed using the Imaris (Bitplane) software. The significance of differences between each condition was evaluated by Student's t tests. Differences with p-values of less than 0.05 were considered significant. (b) Flagella formation. (i) MT axoneme formation. Giardia cells treated with water, control morpholino, or anti-Glγ-tubulin morpholino were treated for TEM. In thin section of G. lamblia cells, the axonemes were scored as a canonical 9 + 2 MT axoneme or an axoneme losing the central pair of MTs. Scale bar: 5 μm. (ii) G. lamblia cells stained with anti-α-tubulin antibodies were also used to observe the effect of the anti-Glγ-tubulin morpholino on flagella formation. The length of external portion of the caudal flagella was measured by Zen 2012 software. Scale bar: 2 μm with anti-GlGCP2, a little increase to 5% was detected only in the axonemes of posterolateral flagella (Supporting information Figure   S2c). On the other hand, the axonemes of both posterolateral and ventral flagella were more affected by the treatment of anti-GlGCP3 morpholino (13% and 5%, respectively).
In addition, the length of the caudal flagella was monitored in G.
lamblia treated with anti-GlGCP morpholino (Figure 6e). The lengths of the caudal flagella of the control cells were 10 ± 1 and 9 ± 1 µm, and this measurement decreased to 6 ± 1 µm upon the knockdown of GlGCP2 and GlGCP3 (p-values = 0.00013 and 0.00007).

| Relationship of Glγ-TuSC with GlEB1 in the median body and flagella biogenesis
In the subsequent experiment, we examined whether Glγ-tubulin affects the function of GlEB1. To do this, the Giardia cells expressing HA-tagged GlEB1 were treated with anti-Glγ-TuSC morpholino, that is, anti-Glγtubulin, anti-GlCGP2, or anti-GlGCP3 morpholino, and then double-stained with anti-HA and anti-α-tubulin antibodies to determine the localization of GlEB1 and MTs (Figure 7a). In control cells, we could observe the localization of GlEB1 at the nuclear membranes and the median bodies. The localization of GlEB1 in the nuclear membrane as well as in the median body was not distinct in Glγ-TuSC knockdown cells. A western blot analysis in these cells showed that the expression level of GlEB1 was not changed in the Glγ-TuSC knockdown cells (Figure 7b). This result suggests the possibility that Glγ-TuSC may be required for the correct positioning of GlEB1 in Giardia.
Therefore, we examined the phenotype of the GlEB1 knockdown cells with respect to the formation of the flagella and median body in the following experiments. A previous study indicated that knockdown of the GlEB1 resulted in a cytokinesis defect (Kim et al., 2017(Kim et al., , 2014. GlEB1 knockdown also resulted in an increased proportion of Giardia cells without a median body (from 24% to 30% p-value = 0.0012; Figure 8a Figure S2d). With respect to the caudal flagella, the length of these flagella was slightly affected by GlEB1 knockdown in G. lamblia and decreased from 7 ± 1 µm in the control group to 6 ± 1 µm in the GlEB1-knockdown group (p-value = 0.04) (Figure 8c).

| D ISCUSS I ON
In S. cerevisiae, a mutant of γ-tubulin was defective in the proper recruitment of the Kar9p-Bim1p complex at the MT tips and only the overexpression of Bim1p restored this mutant phenotype (Cuschieri et al., 2006). Although a direct association between these two proteins has not been shown, their study suggested a functional relationship between Bim1p and γ-tubulin. In the current study, we demonstrated a direct interaction between G. lamblia γ-tubulin and GlEB1, which is the Bim1p ortholog of G. lamblia, through in vivo and in vitro assays ( Figure 1). This interaction seems to occur in G. lamblia as shown by the co-immunoprecipitation of GlEB1 with HA-tagged Glγ-tubulin (Figure 1d(iii)).
γ-Tubulin plays roles in the nucleation and regulation of MT assembly; therefore, it localizes at MTOCs such as centrosomes or spindle pole bodies (Joshi, Palacios, McNamara, & Cleveland, 1992;Wiese & Zheng, 2006). In Trypanosoma brucei, γ-tubulin localizes at basal bodies, which function as an MTOC for the nucleation of flagella/cilia and functions as spindle poles during cell division (Scott, Sherwin, & Gull, 1997;Zhou & Li, 2015). In Giardia, the localization of γ-tubulin at the basal bodies was reported in a study using monoclonal human γ-tubulin antibodies (Nohynková et al., 2000) and a proteomic analysis of the basal bodies (Davids, Shah, Yates, & Gillin, 2011). In this study, we constructed transgenic Giardia expressing HA-tagged Glγ-tubulin and observed the localization of Glγ-tubulin at the basal bodies, axonemes, and median bodies in the interphase trophozoites ( Figure 2a). Interestingly, one-fourth of the interphase cells were found to be devoid of the median body, in which Glγ-tubulin was present at the basal bodies. In the rest of the interphase cells, they had a median body at which Glγ-tubulin was colocalized with MTs ( Figure 2a). This observation is consistent with the study performed by Horlock-Roberts et al. (2017), in which they found that the size of median bodies varies according to the stages of Giardia cell cycle. That is, Giardia cells at G1 phase are present without median bodies, whereas they have bigger median bodies at G2 phase. Thus, our results showing the localization of Glγ-tubulin at the median bodies and its plausible role in the formation of median body via the knockdown of Glγ-tubulin cells should be interpreted with a caution (Figure 4a).
For MT nucleation, γ-tubulin must form a complex with GCPs (Gull, 1999). The most basic MT nucleation machinery is the γ-TuSC complex, which is composed of γ-tubulin and two GCPs (Spc97p and Spc98 in budding yeast; GCP2 and GCP3 in humans; Lin et al., 2015).
A database search and proteomic analysis of basal bodies  indicated that Giardia only has γ-TuSC components, that is, Glγ-tubulin, GlGCP2, and GlGCP3. The presence of γ-TuSC in G.
lamblia was shown in co-immunoprecipitation experiments in which Glγ-tubulin was co-sedimented with putative GlGCP2 and GlGCP3 ( Figure 5a(i,ii), respectively). In addition, the knockdown of GlGCP2 or GlGCP3 resulted in identical phenotype to the knockdown of Glγ-tubulin with respect to reduced volume of the median bodies ( Figure 6c). The median bodies are known as a unique structure in Giardia that may function as a MTOC and a reservoir of polymerized MTs (Piva & Benchimol, 2004). In Giardia, the ectopic expression of mutant kinesin-13, a motor protein depolymerizing MTs at the plus and minus ends, caused significant decreases in the median body volume and resulted in mitotic defects (Dawson et al., 2007).
Our study also suggested that Giardia has a canonical γ-TuSC, protein, which is associated with the central pair MTs of the flagella (Hardin et al., 2017).
Based on the previous investigation (Nohynková et al., 2000), a cartoon was made to predict the position of the axonemes of eight flagella (Supporting information Figure S2a) and used to differentiate the affected flagella axonemes in our TEM figures (Supporting information Figure S2b We then examined whether the knockdown of GlEB1 produced a similar phenotype to that of Glγ-TuSC knockdown. As expected, the decreased expression of GlEB1 resulted in an increased number of cells without the median body, a decreased median body volume, aberrant axonemes losing of the central pair MTs, and shortening of the caudal flagella ( Figure 8). It has been reported in other organisms that mutations or deficiencies in γ-tubulin or GCPs affect the dynamics of plus-end microtubules (Bouissou et al., 2009;Paluh et al., 2000;Vogel et al., 2001;Zimmerman & Chang, 2005). The mutant form of γ-tubulin alters the distribution of the plus-endtracking protein Bim1p, which is homolog to EB1, in S. cerevisiae (Cuschieri et al., 2006) and the depletion of GCPs also affects EB1 in Drosophila (Bouissou et al., 2014). An explanation for γ-tubulin interaction with plus-end protein MT dynamic is that γ-tubulin complexes at MTOCs bind catastrophe or rescue factors or the motor molecules that transport them while there is constant bidirectional transport along microtubules (Oakleya, Paolilloa, & Zheng, 2015).
Further experiments should be performed to observe the changes in the distribution of Glγ-TuSC and GlEB1 during the cell cycle of

Giardia.
Our investigation demonstrated that Giardia has a canonical Glγ-TuSC, which plays a role in MT nucleation for median body formation and flagella biogenesis, and that GlEB1 may be involved in this process. Park, and 2016R1A6A3A11933823 to J. Kim).

CO N FLI C T O F I NTE R E S T
The authors declare that there is no conflict of interest.

AUTH O R S CO NTR I B UTI O N
Juri Kim and Soon-Jung Park conceived and designed the experiments. Juri Kim performed the experiments. Juri Kim and Soon-Jung Park analyzed the data and wrote the manuscript.

DATA ACCE SS I B I LIT Y
All data are included within the manuscript.