An intracellular membrane junction consisting of flagellum adhesion glycoproteins links flagellum biogenesis to cell morphogenesis in Trypanosoma brucei

Kinetoplastid parasites including Trypanosoma brucei, Trypanosoma cruzi and Leishmania spp. are important human and veterinary pathogens that still remain a threat to global health and productivity (Mathers et al., 2007). These single-cellular organisms are characterized by the presence of a single flagellum, which is important for normal cell motility, cell division, immune evasion and host-parasite interactions (Broadhead et al., 2006; Engstler et al., 2007; Oberholzer et al., 2011; Ralston et al., 2006). Like many eukaryotic organelles, size regulation has been observed for the flagellum in trypanosomatid parasites and the control of the flagellum length is important for their parasitic life style (Gluenz et al., 2010). In Leishmania, the flagellum undergoes tight regulation so that a long flagellum is present in the extracellular, motile parasite and only a small residual flagellum is present in the intracellular, immotile amastigote cells. Flagellum length also varies during T. brucei life cycle development and its length is tightly coupled to the FAZ length and cell body length (Rotureau et al., 2011; Sharma et al., 2008).

In procyclic form T. brucei, which proliferates in the midgut of tsetse flies, each parasite cell contains a single flagellum seeded by the basal bodies that are physically attached to the kinetoplast (mitochondrial DNA) (Ogbadoyi et al., 2003). During the cell cycle, a new flagellum is assembled posterior to the existing flagellum, with its distal end attached to the old flagellum through the flagella connector (FC). As cell cycle progresses, little growth occurs at the old flagellum, whereas the new flagellum, guided by the FC, elongates along the old flagellum, duplicating the exact same length and helical pattern as the old structure (Moreira-Leite et al., 2001). Elongation of the new flagellum is accompanied with the assembly of a new flagellum attachment zone (FAZ) (Kohl et al., 1999), a mostly intracellular structure containing protein filament, microtubules and associated endoplasmic reticulum (Sherwin and Gull, 1989). The coordinated flagellum/FAZ assembly is required for the highly ordered segregation of single-copy organelles including the kinetoplast and the associated basal bodies (Absalon et al., 2007; Vaughan, 2010). It also plays an important role in cell morphogenesis, by regulating the organization of the subpellicular microtubules that subtends the plasma membrane (Kohl et al., 2003; Sherwin and Gull, 1989; Zhou et al., 2011). However, the molecular mechanism of the coordinated flagellum/FAZ assembly remains unknown.

In trypanosomes, the membrane-bound flagellum adheres to the cell body via membrane-associated glycoproteins. The first such molecule characterized was GP72, a stage-specific, glycosylated membrane protein in T. cruzi (Cooper et al., 1991). GP72 was readily detected along the flagellum adjacent to the adhesion region using indirect immunofluorescence on live, non-permeabilized parasites (Haynes et al., 1996). While GP72 null mutants are viable, T. cruzi cells without GP72 contain detached flagella and drastically changed cell morphology (Cooper et al., 1993). FLA1, the T. brucei homolog to GP72, is an essential protein, functioning in both flagellum attachment and cell division, likely through regulation of FAZ assembly (LaCount et al., 2002; Nozaki et al., 1996). As a transmembrane protein with a large extracellular domain and a 16-amino-acid cytoplasmic tail (LaCount et al., 2002), how FLA1 mediates flagellum membrane adhesion is not yet clear.

In this study, we performed domain analysis of the FLA1 protein in procyclic T. brucei where the function of FLA1 is best characterized. While the 16-amino-acid tail is required for correct targeting of FLA1 to the FAZ, the extracellular domain is important for flagellum attachment through its interaction with a novel FLA1-interacting protein, FLA1BP. The specific interaction between FLA1BP on the flagellum and FLA1 on the cell body, provides the molecular basis for the membrane–membrane adhesion between flagellum and cell. This unique intracellular membrane junction facilitates the coordinated assembly of flagellum and FAZ, and mediates flagellum regulation of cell morphogenesis.

FLA1 is a surface glycoprotein dually required for flagellum attachment and cytokinesis (LaCount et al., 2002). It contains an N-terminal signal peptide, followed by a glycosylated, extracellular domain, a transmembrane domain and a C-terminal 16-amino-acid tail (Fig. 1A). The extracellular domain was predicted to contain NHL repeats, which have been found in various proteins including growth regulators and membrane receptors and have been shown to be required for protein–protein interactions (Edwards et al., 2003; Slack and Ruvkun, 1998). Previous immunofluorescence analyses showed that FLA1, like GP72, was present along the flagellum adhesion region (Haynes et al., 1996; Nozaki et al., 1996). But it was not clear whether FLA1 (or GP72) was associated with the cell membrane or the flagellum membrane or both. Being the only flagellum adhesion protein characterized in T. brucei thus far, it was also not clear how FLA1 mediated the flagellum/cell membrane adhesion.

To examine the precise localization of FLA1 and the targeting requirements of this protein, various YFP fusions were constructed and stably overexpressed in procyclic T. brucei (Fig. 1). The localization of YFP-tagged FLA1 mutants was then monitored in non-permeabilized cells by direct YFP fluorescence or indirect immunofluorescence using anti-GFP (Fig. 1B–G). Whereas direct YFP fluorescence reflected the distribution of YFP reporters all over the cell, anti-GFP labeling of intact cells allowed specific detection of YFP reporters at cell surface. All FLA1 mutants, with YFP inserted immediately after the signal peptide, were found in reticulated structures throughout the parasite cell that overlapped with the endoplasmic reticulum (data not shown). This observation was consistent with the presence of the N-terminal signal peptide in FLA1 that allowed its entry into the secretory pathway. The strong YFP signal observed in the endomembrane system may also be due to YFP overexpression (Fig. 1H) or poor folding in these cells. Anti-GFP probing of non-permeabilized YFP::FLA1 cells specifically labeled along the flagellum attachment region (Fig. 1B), confirming the exclusive presence of the YFP reporter extracellularly at the flagellum attachment region. This was consistent with previous reports on FLA1 localization and membrane topology in T. brucei (LaCount et al., 2002; Nozaki et al., 1996). Similarly, HA-tagged GP72 was also found along the flagellum when live, non-permeabilized cells were probed with anti-HA (Haynes et al., 1996). Interestingly, FLA1 truncation mutants lacking the extracellular domain (YFP::FLA1ΔECD) or the C-terminal 16-amino-acid tail (YFP::FLA1ΔC16 and YFP::FLA1ΔC16RR, in which two arginine residues were inserted immediately after the transmembrane domain as stop transfer signal), could all be incorporated into the flagellum attachment region (Fig. 1C–E). YFP fusion to the transmembrane domain only (YFP::FLA1–TM) and YFP reporter alone, however, were both readily detected intracellularly but showed no specific targeting to the flagellum adhesion region (Fig. 1F,G). The extracellular accessibility of the YFP reporters was further evaluated by surface biotinylation assays (supplementary material Fig. S1). Whereas YFP::FLA1, YFP::FLA1ΔECD, YFP::FLA1ΔC16 and YFP::FLA1ΔC16RR could all be biotinylated and affinity purified with streptavidin beads, YFP::FLA1-TM and YFP alone could not, verifying the results of the immunofluorescence assays. These results suggested that either the extracellular domain or the C-terminal 16-amino-acid tail of FLA1, together with the transmembrane domain, could mediate protein targeting to the flagellum adhesion region.

Due to the flagellum attachment along the cell body, we reasoned that the precise localization of the FLA1 mutants could only be distinguished if the FLA1 mutants were expressed in cells with flagellum detached from the cell body. To achieve this, FLA1 cDNA with scrambled coding sequence (supplementary material Fig. S3) was synthesized and used as template for the construction of RNAi-resistant FLA1 mutants, which were then expressed in FLA1-RNAi cells (Fig. 2). The tetracycline-inducible depletion of endogenous FLA1 and tetracycline-inducible expression of RNAi-resistant, YFP-tagged FLA1 truncation mutants were monitored by immunoblots using a monoclonal antibody against FLA1 (Nozaki et al., 1996) and anti-GFP, respectively (Fig. 2B). Consistent with previously observed (LaCount et al., 2002), upon the addition of tetracycline to induce FLA1-RNAi, the endogenous FLA1 was depleted (Fig. 2B), leading to flagellum detachment from the cell body (Fig. 2A). In these cells, the RNAi-resistant YFP::FLA1ΔECD^R localized exclusively to the cell bodies in a distinct, continuous line overlapping with the FAZ filament marked with L3B2 antibody (Kohl et al., 1999). YFP::FLA1ΔC16^R, on the other hand, localized mainly in punctate structures along the detached flagellum. Some scattered labeling of YFP::FLA1ΔC16^R was also found on the cell body side, aligning with the L3B2-labeled FAZ (Fig. 2A). There may be other explanations for these observations, but the simplest interpretation of these results is that the C-terminal 16 amino acids may be required for anchoring to the cell body along the FAZ, and the extracellular domain may interact specifically with a flagellum membrane protein, thus mislocalizing YFP::FLA1ΔC16^R to the flagellum in punctate dots.

An co-immunoprecipitation approach was employed to identify the hypothetic, FLA1-binding protein on the flagellum membrane. As outlined in Fig. 3A, FLA1-RNAi cells expressing YFP::FLA1-ΔC16^R were biotinylated to label all surface proteins. After cell lysis with sonication and detergent extraction, centrifugation-cleared cell lysates were used for co-immunoprecipitation with anti-GFP followed by affinity purification with streptavidin-coated beads. Proteins eluted from the streptavidin beads were then analyzed by silver staining and immunoblots (Fig. 3B–D). Two bands were specifically detected in YFP::FLA1ΔC16^R cells relative to cells expressing YFP only, which was used as a negative control (Fig. 3B). One was a ∼110 kDa protein that was labeled with both anti-GFP and anti-biotin (Fig. 3C,D), representing the YFP::FLA1ΔC16^R protein and this was confirmed by mass spectrometry analysis. The other was an ∼80 kDa band which could be detected only by anti-biotin, but not by anti-GFP, possibly representing a FLA1-binding protein. Mass spectrometry identified a single, previously uncharacterized protein (encoded by two identical genes Tb927.8.4050 and Tb927.8.4100; see Table 1), which we named FLA1BP. Bioinformatic analyses of FLA1BP revealed a domain organization strikingly similar to that of FLA1, despite the lack of sequence similarity between these two proteins. Following a signal peptide near the N-terminus, FLA1BP possessed an NHL-repeat-containing domain, followed by a transmembrane domain and a C-terminal 40-amino-acid tail (Fig. 4A). The interaction between FLA1 and FLA1BP was also verified by reverse co-immunoprecipitation, where FLA1 was shown to co-precipitate with YFP::FLA1BP (supplementary material Fig. S4).

YFP::FLA1BP was constructed by inserting the YFP reporter immediately after the N-terminal signal peptide and before the NHL repeats (Fig. 4A). Localization of YFP::FLA1BP was then evaluated by YFP fluorescence or anti-GFP labeling of non-permeabilized cells induced for FLA1-RNAi or not. In un-induced control cells, YFP::FLA1BP was found all over the cell by direct YFP fluorescence, similar to YFP::FLA1 (Fig. 4B). By surface labeling with anti-GFP, YFP::FLA1BP localized exclusively along the flagellum attachment region (Fig. 4B). This result also indicated that the YFP moiety in YFP::FLA1BP was accessible extracellularly, supporting that the region between the signal peptide and the transmembrane domain, including the NHL repeats, was also extracellular, just like FLA1.

Interestingly, in FLA1-RNAi cells, YFP::FLA1BP was detected only along the detached flagellum in a dotted pattern by anti-GFP (Fig. 4B). This localization pattern did not require the extracellular domain of FLA1BP, as YFP::FLA1BPΔECD targeted properly to the detached flagellum, in a dotted pattern similar to that observed for YFP::FLA1BP (Fig. 4D).

The dotted labeling pattern of FLA1BP along the flagellum suggested a possible anchoring of FLA1BP to flagellum skeleton, which contains a 9+2 microtubular axoneme and a paraflagellar rod (PFR) lattice, both with regular structural repeats. Neither the localization of YFP::FLA1BP to the flagellum, nor its dotted pattern was affected in cells lacking PFR that were additionally depleted of PFR2 (Fig. 4E) (Bastin et al., 1998), suggesting that FLA1BP may interact directly with flagellum axoneme.

The presence of FLA1BP on the flagellum and its interaction with FLA1 suggested that their interaction could mediate the flagellum–cell adhesion observed in T. brucei. To test this, FLA1BP was depleted using inducible RNAi. Efficient RNAi was verified by reverse transcriptase (RT)-PCR and immunoblots (Fig. 5B,C). As specific antibody against FLA1BP was not available, YFP::FLA1BP was stably expressed in FLA1BP-RNAi cells and its depletion upon tetracycline induction was monitored using anti-GFP. Despite efficient FLA1BP depletion, the FLA1BP-RNAi cells did not show obvious defects in proliferation rate. Cells lacking FLA1BP continued to duplicate at a rate similar to the un-induced control even 7 days post- induction (Fig. 5A). However, FLA1BP-RNAi cells exhibited drastic changes in morphology. Flagellum detached from cell body in >90% of the cells observed at 2 days post-induction (cf. Fig. 6), supporting the role of FLA1BP in flagellum adhesion.

In addition to flagellum detachment, FLA1BP-depleted cells also showed drastic morphological changes in organelle positioning and cell length, which were likely due to disruption of the coordinated flagellum/FAZ assembly (Fig. 6A,B). To test this, we analyzed FLA1BP-depleted cells that contained a single nucleus, a single kinetoplast and a single flagellum, which represented ∼60% of the asynchronous population at 7 days post-induction. Detailed measurements of PFR length (an indicator of flagellum length), FAZ length and nucleus/kinetoplast positioning relative to the posterior and anterior tips of the parasite cells were summarized in Fig. 6C. The flagellum, despite being detached from the cell body, measured at 14.3±2.3 µm in FLA1BP-RNAi cells, similar to that of the control cells (13.7±1.9 µm). Whereas in control cells, the flagellum length and FAZ length are strongly correlated (R² = 0.75), this correlation was completely abolished in FLA1BP-RNAi cells (R² = 0.21; Fig. 6D). The FAZ length reduced drastically from 12.1±2.1 µm in control cells to 5.8±1.6 µm in FLA1BP-ablated cells.

Short FAZ led to short cell length. The distance between the nucleus and the posterior end of the cell was 6.3±1.8 µm, decreased only slightly than that in control cells (7.1±1.4 µm; Fig. 6C). The distance between the nucleus and the anterior tip of the parasite (anterior region marked by the presence of the FAZ), however, is shortened from 9.2±1.7 µm in control cells to 6.6±1.5 µm in FLA1BP-RNAi cells (Fig. 6C), suggesting that shortening of the cell body was mainly due to shortening in the anterior region of the parasites. On the other hand, the distance from the kinetoplast and associated basal bodies to the posterior tip, significantly increased from 4.5±1.0 µm to 6.8±1.6 µm (Fig. 6C), thus resulting in the kinetoplast located slightly anterior to the nucleus in FLA1BP-RNAi cells (see Fig. 6A).

Despite the short length, the FAZ appeared normally assembled in FLA1BP-RNAi cells. In addition to the FAZ markers CC2D and FAZ1 (Fig. 6A; supplementary material Fig. S5), YFP::FLA1 was also properly localized to the FAZ region on the cell body side (Fig. 4C). Restoring FLA1BP expression in FLA1BP-RNAi cells by tetracycline removal restored flagellum adhesion, correlation between flagellum length and FAZ length, as well as normal cell morphology (Fig. 6A–D).

To further understand how FLA1BP-RNAi cells divided with a short FAZ, major cell cycle events including the duplication and segregation of nucleus, basal bodies and associated kinetoplast, as well as the assembly of new FAZ and new flagellum were examined in FLA1BP-RNAi cells throughout the cell cycle (Fig. 7). In both control and FLA1BP-RNAi cells, the order and timing of each cell cycle event remained largely unaltered. This was consistent with the observation that FLA1BP-RNAi cell proliferated at the same rate as the control cells and no significant change in cell cycle duration was observed (cf. Fig. 5A). Whereas the new flagellum was assembled along the old flagellum, with its distal tip attached to the old flagellum via the FC, the new FAZ also extended along the old FAZ, with it distal tip attached to the old structure, in both control and FLA1BP-RNAi cells (Fig. 7A,B). However, unlike in control cells where new FAZ and new flagellum assembled in a highly coordinated fashion (R² = 0.92), elongation of the new FAZ was slowed and uncoupled from new flagellum elongation in FLA1BP-RNAi cells (R² = 0.35; Fig. 7C). The short FAZ could also account for the inhibited basal body/kinetoplast segregation observed in these parasites (Fig. 7D), hence the change in kinetoplast positioning in daughter cells (cf. Fig. 6C). Nuclear division, however, was only moderately affected by the short FAZ (Fig. 7E).

Lithium chloride treatment has been shown to lengthen flagellum/cilium in Chlamydomonas reinhardtii and some mammalian cells (Miyoshi et al., 2009; Wilson and Lefebvre, 2004), by inhibition of the glycogen synthase kinase 3 (GSK3). To confirm that flagellum adhesion proteins play an active role in linking FAZ assembly to flagellum elongation, T. brucei cells induced with FLA1BP-RNAi or not were cultured in the presence or absence of 5 mM LiCl for 24 hours, which showed little inhibitory effects on parasite proliferation (Fig. 8A). Flagellum and FAZ lengths were then measured in cells containing one nucleus, one kinetoplast and one flagellum, as described in Fig. 6. The LiCl treatment resulted in a significant increase of flagellum length in both control and FLA1BP-RNAi cells (Fig. 8B,C, P<0.0001). Significant FAZ length increase, however, was only observed in control cells where flagellum adhesion was intact (Fig. 8B, P<0.0001). Cells lacking FLA1BP did not show difference in FAZ length (Fig. 8C, P = 0.35), supporting an essential role of flagellum membrane adhesion in transmitting flagellum length information to the intracellular FAZ.

Using a FLA1-truncation mutant that lacked the C-terminal 16 amino acids and mislocalized to the flagellum membrane, we isolated a novel FLA1BP, which is the first FLA1-binding protein identified to date. Despite the lack of similarity in amino acid sequences, FLA1BP and FLA1 have similar domain organization: both are transmembrane proteins with extracellular, NHL-repeat-containing domains and short cytoplasmic tails. Further analyses of YFP-tagged FLA1 and FLA1BP in cells with detached flagella, revealed a distinct localization of FLA1BP to the flagellum membrane and FLA1 to the cell membrane along the FAZ. Co-immunoprecipitation confirmed the interaction between FLA1 and FLA1BP, probably through their extracellular domains. A model of their interaction and how this interaction mediates flagellum–cell adhesion is proposed in Fig. 9. In this model, FLA1 spans the cell membrane along the FAZ with its C-terminal 16 amino acids in the cell lumen, possibly playing a role in intracellular FAZ assembly (LaCount et al., 2002). FLA1BP, on the other hand, spans the flagellar membrane, with the C-terminal 40 amino acids facing flagellum lumen and possibly anchored to flagellar axoneme. FLA1 and FLA1BP interact through their extracellular domains, hence linking the flagellum membrane to cell membrane.

In T. brucei, the flagellum is a multifunctional organelle. In addition to motility and sensory functions, it also regulates cell morphogenesis (Vaughan, 2010). This latter function is via the coordinated assembly of flagellum with the FAZ, which is an intracellular complex important for cell division, organelle segregation and subpellicular microtubule organization (LaCount et al., 2002; Robinson et al., 1995; Vaughan et al., 2008; Zhou et al., 2011). Further dissection of the exact mechanisms of the coordinated assembly required molecular characterization of the flagellum membrane attachment structures.

The identification of FLA1BP, an interacting partner of FLA1, provided the molecular basis of flagellum membrane adhesion to the cell membrane. In cells lacking FLA1BP and flagellum adhesion, the coordinated flagellum/FAZ assembly was disrupted, during both cell cycle development and LiCl treatment. Although the new flagellum formed normally, the FAZ was assembled at a slower rate, to a short but relatively fixed length (5.8±1.6 µm), as indicated by different FAZ markers. This short FAZ was ∼50% of the average FAZ length observed in control cells (12.1±2.1 µm). It is not clear if this partial assembly was due to incomplete depletion of FLA1BP, or the presence of other FLA1-related proteins (see below), or an intrinsic length regulation of core FAZ components. However, it appeared that a ∼6 µm FAZ was sufficient to support organelle segregation, cell division and cell survival, at least in culture.

The short FAZ is responsible for the shortened cell length, particularly in the anterior region, suggesting that the subpellicular microtubules in the posterior region may be regulated differently to the subpellicular microtubules in the anterior region. The short FAZ was also responsible for the limited segregation of the kinetoplasts and associated basal bodies, but had only moderate effect on the proper segregation of the nuclei.

Reversing FLA1BP-RNAi by removing tetracycline restored flagellum–cell adhesion and the coordinated flagellum–FAZ assembly, thus allowing the length information contained in the flagellum to be efficiently transmitted to the cell body and thus restoring normal cell morphology. Previous work on GP72 in T. cruzi produced similar results in cell morphology (Cooper et al., 1993). Together, these results supported a conserved function of the FLA1-related membrane adhesion proteins in linking flagellum biogenesis to FAZ elongation, therefore regulating cell morphogenesis in trypanosomes.

To understand how the molecular interaction between FLA1 and FLA1BP coordinates flagellum/FAZ assembly, we investigated the targeting of FLA1 and FLA1BP to the flagellum adhesion region. Our present studies indicated that the extracellular domains of FLA1 and FLA1BP were dispensable for their respective localization to the FAZ and the flagellum membrane. The correct targeting of FLA1 and FLA1BP was likely mediated by their short, C-terminal cytoplasmic tails, through specific interaction with other proteins in the cell body or in the flagellum. The C-terminal 16 amino acids of FLA1 has been implicated in directing FAZ assembly (LaCount et al., 2002) and depletion of FLA1 led to inhibited FAZ formation (Vaughan et al., 2008). The C-terminal 40 amino acids of FLA1BP resided in the flagellum lumen and may be associated with flagellum proteins that led to the dotted pattern observed with YFP::FLA1BPΔECD. Whereas FLA1BP-depleted cells were able to proliferate in culture, FLA1 is an essential protein possibly as an integral component of FAZ (Rotureau et al., 2011). Interestingly, the PFR was not essential for YFP::FLA1BP localization to the flagellum in a dotted line, suggesting that FLA1BP may be associated with other structural component, possibly the flagellum axoneme. This is also consistent with the normal flagellum attachment and lack of morphological changes observed in snl-1 cells lacking PFR (Bastin et al., 1998). The FLA1–FLA1BP interaction, and their respective interactions with flagellum axoneme and FAZ proteins, may direct the formation of the membrane–cytoskeleton nexus at flagellum attachment region and thus mediate the highly coordinated flagellum/FAZ assembly during the cell cycle.

The adhesion between the flagellum membrane and the cell membrane (Fig. 9) is reminiscent of cell–cell membrane adhesions observed in multicellular animals, such as the adherens junctions (Yonemura, 2011). The interaction between FLA1 and FLA1BP and their possible association with cytoskeletal components (see Fig. 9), is also similar to cadherins that are transmembrane glycoproteins forming dimers with the extracellular domains and interacting with actin cytoskeleton on the cytoplasmic side. Adherens junctions are important for tissue morphogenesis and signal transduction between linked cells. Given that sensory functions have become increasingly evident in trypanosome flagella, whether FLA1 and FLA1BP also function in signal transduction between the flagellum and cell body will be of great interest.

In T. brucei database, four additional genes were predicted to encode proteins containing transmembrane domains and NHL repeats arranged in similar fashion as FLA1 and FLA1BP (Table 1). Among them, the amino acid sequences of FLA2 (encoded by Tb927.8.4060) and FLA3 (encoded by Tb927.8.4110) are >97% identical, and each exhibits ∼60% identity and 70% similarity, respectively, to FLA1. While FLA1BP (encoded by Tb927.8.4050 and Tb927.8.4100) shows little similarity to FLA1, a FLA1BP-like protein (encoded by Tb927.5.4570 and Tb927.5.4580) was found in the database, showing 40.5% identity and 54.9% similarity to FLA1BP. A sixth FLA1-like protein is encoded by Tb927.10.6180, which shows no homology to either FLA1 or FLA1BP. Interestingly, all FLA1 related proteins with the exception of Tb927.10.6180 appear to be developmentally regulated. While FLA1 and FLA1BP appeared more abundant in the procyclic cells, FLA2/3 and FLA1BP-like protein were more abundant in blood stream form parasites by transcriptome and mRNA expression analyses (Kabani et al., 2009; Queiroz et al., 2009). Owing to the sequence similarity between FLA1 and FLA2/3, and FLA1BP and FLA1BP-like proteins, it is tempting to speculate that whereas FLA1–FLA1BP interaction mediates flagellum–cell membrane adhesion in procyclic cells, FLA2/3 may interact with FLA1BP-like protein in BSF parasites. Previous studies suggested significant FAZ remodeling during parasite development in tsetse flies (Rotureau et al., 2011). The presence of six FLA1-related proteins, with similar domain organization but distinct sequences and developmental specificity, substantiates the remarkable complexity of the flagellum–cell membrane adhesion during development.

Orthologs to most of the FLA1-related proteins are also found in other kinetoplastid parasites including T. cruzi and Leishmania spp. (Table 1). Each T. cruzi cell contains an attached flagellum similar to T. brucei. Flagella in Leishmania parasites, however, are only partially attached with a short FAZ (Weise et al., 2000). This unique intracellular membrane adherence junction in a unicellular parasite represents a fascinating opportunity to study signal transduction from flagellum to cell body. The possible specific interaction between two non-variant surface proteins that is not essential for normal cell functions also opens up new ways to combat these lethal human pathogens.

Procyclic form YTat1.1 cell line (Ruben et al., 1983) was cultured in Cunningham’s medium supplemented with 15% heat-inactivated fetal bovine serum (Hyclone) at 28°C. Procyclic form 29.13 cell line engineered for tetracycline-inducible expression (Wirtz et al., 1999) was maintained in Cunningham’s medium supplemented with 15% heat-inactivated, tetracycline-free fetal bovine serum (Clontech) in the presence of 15 µg/ml G418 and 50 µg/ml hygromycin at 28°C.

For FLA1-RNAi, a 302-bp fragment (nucleotides 715–1016) or a 390-bp fragment of FLA1 (nucleotides 1364–1641 of the coding sequence and 112 nucleotides in 3′-untranslated region) was PCR amplified and cloned into a pZJM construct (Wang et al., 2000) with blasticidin resistance, which allows tetracycline-inducible RNAi. Both RNAi constructs produced inhibitory effects on cell growth and flagellum attachment, similar to the RNAi phenotypes previously observed with a 1006-bp fragment of FLA1 (nucleotides 64–1070) (LaCount et al., 2002) (supplementary material Fig. S2; Fig. 2A; Fig. 4B). RNAi cell line created with the 302-bp fragment was used for the expression of FLA1-RNAi resistant mutant YFP::FLA1ΔECD; RNAi cell line created with the 389-bp fragment was used for the expression of RNAi-resistant YFP::FLA1ΔC16^R (see below). For FLA1BP-RNAi, a 518-bp fragment within FLA1BP coding sequence (nucleotides1385–1902) was PCR-amplified and cloned into pZJM with blasticidin-resistance marker.

A pXS2/YFP vector (Bangs et al., 1996; He et al., 2004) was used for stable and transient expression of YFP-tagged proteins. The N-terminal 99-bp of FLA1, which codes for the signal peptide was PCR-amplified from T. brucei genomic DNA with primers 5′-CCCAAGCTTATGGGTGGCAGGACTGAATCACGGG-3′ (HindIII site underlined) and 5′-CTAGCTAGCGGTTTGGTCTCCTATAACGGGACTC-3′ (NheI site underlined), digested with HindIII and NheI, and fused to the N-terminus of YFP in pXS2/YFP; the remaining 1542-bp coding region of FLA1 (nucleotides 100–1641) was then PCR-amplified with primers 5′-GAAGATCTGTGGGCACCAAGGTGATTGTGAACC-3′ and 5′-GAAGATCTTTATCACTCGAACGCCGGCAACCGTACA-3′ (BglII sites underlined), digested with BglII and fused to the C-terminus of YFP, thus producing the pXS2/YFP::FLA1 construct. The FLA1ΔECD truncation mutant (containing nucleotides 1516–1641 of FLA1 coding sequence) was PCR-amplified with primers 5′-CGGGATCCGTGTGTCTCGGTGGTATCGTTTCCTC-3′ (BamHI site underlined) and 5′-CGGAATTCTCACTCGAACGCCGGCAA-3′ (EcoRI site underlined), digested with BamHI and EcoRI, and fused into the C-terminus of FLA1 signal peptide and YFP, producing the YFP::FLA1ΔECD. Using YFP::FLA1 or YFP:FLA1ΔECD as a template, a stop codon TGA was introduced after the transmembrane domain (nucleotides 1519–1590) with primers 5′-TTTCCTGATGGTGGTATGAAGCCCGTACAACGTAA-3′ and 5′-TTACGTTGTACGGGCTTCATACCACCATCAGGA AA-3′ (stop codons underlined), thus producing YFP::FLA1ΔC16 and YFP::FLA1TM, respectively. YFP::FLA1ΔC16RR was constructed by placing two genetic codons for arginine (R) and a stop codon immediately after the transmembrane domain (nucleotides 1519–1590) through PCR amplification with primers 5′-GAAGATCTTCACCGACGTACCACCATCAG-3′ (stop codon underlined and genetic codon of arginine under dotted line). To construct RNAi-resistant FLA1 truncation mutants, nucleotides 1360–1590 of FLA1 with scramble codons (supplementary material Fig. S3A), was synthesized and used to replace all corresponding regions in the FLA1 mutants. For tetracycline inducible expression, the YFP-tagged FLA1 mutants were cloned into the plew100 vector with phleomycin resistance (Wirtz et al., 1999). Proper expression of the RNAi-resistant YFP::FLA1ΔC16^R was verified by immunoblots with anti-FLA1 and anti-GFP, respectively (supplementary material Fig. S3B).

For expression of FLA1BP as YFP fusions, the N-terminal 135 nucleotides that contained the N-terminal transmembrane domain was PCR amplified with primers 5′-CCCAAGCTTATGCCGTTGTGGAAACAAACC-3′ (HindIII site underlined) and 5′-CTAGCTAGCACTCTCACTACAATCGTTGCG-3′ (NheI site underlined), digested with HindIII and NheI, and fused to the N-terminus of YFP in a pHD1034 vector with puromycin resistance (Quijada et al., 2002); the rest of FLA1BP coding sequence (nucleotides 136–2253) was then PCR-amplified with primers 5′-CGGGATCCGGAGCCGCTCCCATTGAA-3′ (BamHI site underlined) and 5′-GGAATTCTCACTGCTCGACCCTTTCTTT-3′ (EcoRI site underlined), and fused to the C-terminus of YFP, thus generating YFP::FLA1BP. Alternatively, FLA1BP coding sequence containing only the C-terminal transmembrane domain and 40-amino-acid tail (nucleotides 2059–2250) was PCR amplified with primers 5′-CGGGATCCCGCGTGTGCCTAATCATAATC-3′ (BamHI site underlined) and 5′-GGAATTCTCACTGCTCGACCCTTTCTTT-3′ (EcoRI site underlined), and engineered to the C-terminus of YFP to generate YFP::FLA1BPΔECD.

For transient transfections, 50 µg of plasmid DNA was introduced into parasites by electroporation. Transient expression of YFP fusions was monitored typically at 16 hours post-transfection. For stable transfections, 15 µg linearized plasmid DNA was used for electroporation, and stable clones selected by appropriate antibiotics and serial dilution.

Total RNA was isolated from 5×10⁷ cells using Trizol® reagent (Invitrogen, USA). Isolated RNA was treated with RNase-free DNase I (Roche, Germany) and the first-strand cDNA was synthesized using M-MLV reverse transcriptase (Invitrogen, USA) with oligo(dT)12–18 primer (Fermentas, USA). The cDNA level of FLA1BP was then monitored by PCR amplification of a 202 base pairs fragment with primers 5′-CGGGATCCCGCGTGTGCCTAATCATAATC-3′ and 5′-GGAATTCTCACTGCTCGACCCTTTCTTT-3′ using varying PCR cycles. As a control, the cDNA level of α-tubulin was monitored by PCR amplification of a 164-bp fragment with primers 5′-AAGCGCGCCTTCGTGCACTG-3′ and 5′-CGGGATCCCTAGTACTCCTCCACATCCTCC-3′.

Log-phase cells were diluted to 2×10⁵ cells/ml to initiate a growth assay. Cell densities were measured every 24 hours by a hemocytometer. If cell density exceeded 10⁶ cells/ml, the culture was diluted with fresh medium to 10⁶ cells/ml to avoid overgrowth in the next 24-hour period. Duplication index was calculated as Log₂ (N_t×Df/N₀), where Df is the dilution factor, N_t is cell density at each time point and N₀ is the cell density at t = 0.

Cells with stably transfected pLew100/YFP:FLA1ΔC16^R and pZJM/FLA1 were cultured in the presence of 10 µg/ml tetracycline for 40 hours to induce FLA1-RNAi and expression of YFP::FLA1ΔC16^R. Approximately 4×10⁹ cells were harvested, washed with pre-chilled gPBS (phosphate-buffered saline containing 1 g/l glucose), and resuspended with cold gPBS containing 1.5 mg/ml biotin (Pierce 21335). Surface biotinylation was carried out on ice for 30 minutes, and 2 M Tris buffer (pH 6.8) was added to 100 mM final concentration to terminate biotinylation. Parasite cells were washed twice, resuspended in 1 ml gPBS containing 100 mM Tris and protein inhibitors (5 µg/ml pepstain, 5 µg/ml leupeptin, 1 µM PMSFs and 5 µg/ml aprotinin; Sigma), and lysed by brief sonication. Triton X-100 was then added the cell lysate to 1% and the cells were extracted at 4°C for 1 hour with gentle mixing. After centrifugation at 20,000 g for 30 minutes at 4°C, the cleared cell lysate was incubated with 50 µl anti-GFP bound Dynabeads (Invitrogen) at 4°C overnight. The Dynabeads were washed twice with PBS containing 1% Triton X-100, 100 mM Tris and protein inhibitors, and twice with PBS containing 100 mM Tris. For elution of proteins bound to anti-GFP, the Dynabeads were incubated with 200 µl 200 mM glycine (pH 1.8) for 3 minutes and the eluate rapidly neutralized with Tris buffer (pH 10). 40 µl streptavidin beads (Invitrogen) were then incubated with the neutralized eluate. After the incubation, the beads were washed three times in buffer containing 8 M urea, 1% SDS and 100 mM Tris, pH 8. Proteins bound to the streptavidin beads were eluted by boiling in the presence of gel loading buffer containing 1% SDS and 11.9 mM β-mecaptoethanol and fractionated on 10% SDS-PAGE. The gels were then further processed for silver staining, mass spectrometry (MS) and immunoblotting.

Each dried peptide was reconstituted in 3% acetonitrile and 0.1% formic acid and analyzed by an LTQ-FT ultra mass spectrometer (ThermoFisher Scientific) and a Prominence™ HPLC unit (Shimadzu). The peptide samples were injected, concentrated and subsequently resolved in a capillary column (200 µm ID×10 cm). Mobile phase buffer A (0.1% formic acid in H₂O) and buffer B (0.1% formic acid in acetonitrile) were used to establish the 60-minute gradient before re-equilibration of the column. HPLC was performed at a constant flow rate of 20 µl/min, and ∼300 nl/min at the electrospray emitter (Michrom BioResources). Under an electrospray potential of 1.5 kV, samples were ionized in an ADVANCE™ CaptiveSpray™ Source (Michrom BioResources). Data acquired by LTQ-FT ultra in the positive ion mode was used to perform a full MS scan at a resolution of 100,000 and a maximum ion accumulation time of 1000 ms. To collect peptides and measure peptide fragments by collision-induced dissociation, a linear ion trap was used with the AGC setting. All MS/MS spectra were searched against Trypanosoma brucei GeneDB (Berriman et al., 2005) for protein identification using Mascot (version 2.2.07, Matrix Science, Boston, MA, USA) search engines.

Unless otherwise stated, cells were attached to coverslips, fixed and permeabilized in methanol at −20°C for 10 minutes and blocked with 3% BSA in PBS for 1 hour before immunolabeling. For surface labeling with the anti-GFP antibody, live cells were incubated with antibodies on ice (Haynes et al., 1996), fixed in situ with 2 or 4% EM grade paraformadehyde (PFA) and then observed under microscope. Alternatively, cells were fixed in situ with 2 or 4% EM grade PFA for 20 minutes. Fixed but non-permeabilized cells were then probed with antibodies and processed for microscopy. Membrane permeabilization under these conditions was not detected using cells expressing YFP only (cf. Fig. 1G). The antibodies anti-PAR (1∶2000), L3B2 (1∶25) or anti-CC2D (1∶500) were used to label the paraflagellar rod, FAZ 1 in the FAZ filament, and CC2D at the FAZ, respectively. Images (single slices or z-stacks) were acquired by a Zeiss Axio Observer Z1 fluorescence microscope equipped with a 63× NA 1.4 objective and a CoolSNAP HQ2 CC2D camera (Photometrics). Z-stack images were collected at 0.3–0.5 µm steps, deconvolved and projected as 2D images. Image processing was performed with AxioVision Rel. 4.8 (for deconvolution), ImageJ (for length/distance measurements) and Photoshop CS extended version 12.0 (for figure preparation).

For scanning electron microscopy, cells were fixed with 2.5% glutaraldehyde in Cunningham’s medium situ for 2 hours and attached to coverslips by centrifugation at 4000 rpm for 10 minutes. Cells were then washed twice with 0.1 M cacodylate buffer (pH 7.4, filtered), one time with distilled water, and post-fixed with 1% O_SO₄ in H₂O for 1 hour at room temperature. After fixation, cells were washed with distilled water once and dehydrated in increasing concentrations of ethanol: 30% ethanol, 50% ethanol, 70% ethanol, 80% ethanol, 90% ethanol, 95% ethanol each for 10 minutes and 100% ethanol twice for 20 minutes. After dehydration, cells were critical point dried, coated with carbon by a vacuum evaporator (JEOL), and viewed using a Helios NanoLab DualBeam electron microscope (FEI).

We thank Ms Meng Mei at Nanyang Technological University for assistance with mass spectrometry and Dr Zakayi Pius Kabututu for technical assistance. We also thank Professor George Cross for the Flap2 antibody (anti-FLA1), Professor Keith Gull and Philippe Bastin for the L3B2 antibody, and Professor Diana McMahon-Pratt for anti-PAR.