Characterisation of polyglutamylases in trypanosomatids

Microtubules are subject to post-translational modiﬁcations, which are thought to have crucial roles in the function of complex microtubule-based organelles. Among these, polyglutamylation was relatively recently discovered, and was related to centrosome stability, axonemal maintenance and mobility, and neurite outgrowth. In trypanosomatids, parasitic protozoa where microtubules constitute the essential component of the cytoskeleton, the function of polyglutamylated microtubules is unknown. Here, in order to better understand the role of this conserved but highly divergent post-translational modiﬁcation, we characterised glutamylation and putative polyglutamylases in these parasites. We showed that microtubules are intensely glutamylated in all stages of the cell cycle, including interphase. Moreover, a cell cycle-dependent gradient of glutamylation was observed along the cell anteroposterior axis, which might be related to active growth of the microtubule ‘corset’ during the cell cycle. We also identiﬁed two putative polyglutamylase proteins (among seven analysed here) which appeared to be clearly and directly involved in microtubule polyglutamylation in in vitro activity assays. Paradoxically, in view of the importance of tubulins and of their extensive glutamylation in these organisms, RNA interference- based knockdown of all these proteins had no effect on cell growth, suggesting either functional redundancy or, more likely, subtle roles such as function modulation or interaction with protein partners.


Introduction
Microtubules (MTs) are highly conserved dynamic structures made of aand b-tubulin heterodimers that are encoded by a family of polymorphic genes and are involved in a variety of essential cell processes such as intracellular transport, mitosis and mobility. In addition to their length and structure variations, MTs can be subject to a series of reversible post-translational modifications (PTMs) (acetylation, phosphorylation, polyglutamylation, polyglycylation, palmitoylation and detyrosination), some of which are very unusual (reviewed in Wloga and Gaertig (2010) and Janke and Bulinski (2011)). Polyglutamylation is a PTM that generates glutamate side-chains of variable length on the gamma-car-boxyl group of glutamate residues within the primary sequence of the target protein (reviewed in Janke et al. (2008)). It occurs in a variety of proteins, essentially aand b-tubulins (Eddé et al., 1990;Rüdiger et al., 1992), but also the nucleosome assembly proteins NAP1 and NAP2 (Regnard et al., 2000) and other nucleo-cytoplasmic shuttling proteins  that, in higher eukaryotes, are proposed to function in a broad array of physiological activities, through molecular mechanisms as diverse as phosphatase inhibition, chromatin regulation, caspase activation and intracellular transport (Okada et al., 2011;Reilly et al., 2011;Kuryan et al., 2012). The modification is generated by relatively recently discovered polyglutamylating enzymes, members of the Tubulin Tyrosine Ligase-Like (TTLL) family (Janke et al., 2005). TTLLs constitute a large family of proteins that share a homology domain with TTL, another tubulin-modifying enzyme, and whose members can catalyse the ligation of glutamate amino acids to tubulins or other substrates . In mammals, 13 TTLL enzymes exist, among which nine are glutamylases (van Dijk et al., 2007) and three are catalysing a related modification, glycylation Wloga et al., 2009). Each enzyme displays defined reaction preferences for modifying the aor b-tubulin and for generating short or long sidechains http://dx.doi.org/10.1016/j.ijpara.2014.09.005 0020-7519/Ó 2014 Australian Society for Parasitology Inc. Published by Elsevier Ltd. All rights reserved. (van Dijk et al., 2007;Rogowski et al., 2009). Both the polyglutamylation and the polyglycylation reactions consist of two biochemically distinct steps: initiation and elongation, which are often mediated by distinct TTLLs. Initiation is the formation of an isopeptide bond with the gamma-carboxyl group of the glutamate acceptor site, whereas side chain elongation consists of the formation of regular peptide bonds (Redeker et al., 1991;Wolff et al., 1994). It has been proposed that the range of signals generated by the variable length of the added sidechain may allow a fine tuning of the function of MTs and of their interactions with protein partners Janke, 2011) (reviewed in Janke et al. (2008) and Janke and Kneussel (2010)). Thus, tubulin PTMs, and more specifically polymodifications, are thought to have crucial roles in the assembly, maintenance and function of complex MT-based organelles (reviewed in Janke and Bulinski (2011)).
Trypanosoma and Leishmania are flagellated parasitic protozoa of the Trypanosomatidae family. They have a simple but precisely ordered cytoskeleton, primarily made of stable MTs  and, with MTs constituting the most abundant part of their cytoskeleton, they show a reduced dependence on the acto-myosin network (Kohl and Gull, 1998;Berriman et al., 2005). MTs constitute four sub-structures in trypanosomatids: the mitotic spindle, the flagellar axoneme, the basal body of the flagellum and, most importantly, the subpellicular 'corset'. This corset is made exclusively of a dense network of MTs cross-linked to each other and to the plasma membrane, forming a helical pattern along the long axis of the cell (reviewed in Robinson et al. (1995)). It is responsible for the cell shape and plays a major role in events such as positioning of organelles, mitosis and cytokinesis (Sasse and Gull, 1988).
Similar to other organisms, MTs in these parasites are subject to a series of PTMs (detyrosination, acetylation, polyglutamylation) (Wolff et al., 1992;Gull, 1999). Using mass spectrometry analysis, trypanosomatid tubulin has been shown to be extensively glutamylated and non-glutamylated tubulins were reported to be almost absent . In contrast, the closely related modification glycylation was not detected in Trypanosoma brucei . The discovery that TTLL proteins catalyse tubulin glutamylation (Janke et al., 2005;van Dijk et al., 2007) led us to identify eight TTLL genes in silico in the genome of T. brucei. These enzymes belong to five of the diverse subtypes of this protein family (TTLL1, TTLL4, TTLL6, TTLL9 and TTLL12). To understand the role of this evolutionarily conserved but highly divergent PTM (Wloga and Gaertig, 2010) and to better characterise tubulin glutamylation in trypanosomatids, the distribution of polyglutamylated MTs during the cell cycle in Leishmania major and T. brucei was examined using specific antibodies (Abs). The putative TTLL genes of T. brucei and L. major were then cloned and the subcellular localisation of the gene products determined using GFP-fused recombinant proteins; their enzymatic activities were analysed in vitro and an insight into their putative biological function obtained using RNA interference (RNAi).

Bioinformatics
The sequences of genes encoding putative TTLL proteins in L. major and T. brucei were searched in the MapView database (http://www.genedb.org/genedb/leish/). The corresponding trypanosomatid protein sequences were aligned using Clustal X and corrected manually as described previously (Wloga et al., 2008).

Construction of L. major cell lines expressing GFP fused proteins
The genes encoding putative TTLLs were PCR-amplified from genomic DNA. The PCR products were cloned into a pGEM-T-Easy vector (Promega Ò , WI, USA) and then inserted into the expression vectors pTH6cGFPn and pTH6nGFPc (Dubessay et al., 2006) which place the GFP fusion at the N-terminus and C-terminus, respectively. The presence of the reading frame of the recombinant proteins was confirmed by sequencing. Transfection in L. major was performed as described previously (Casanova et al., 2009). Briefly, 5 Â 10 7 cells with 80 lg of plasmid DNA were electroporated using a Bio-Rad GENE Pulser II and a Pulse Controller (Bio-Rad, Hercules, USA) at 25 lF and 1500 V with two pulses of 0.5 ms each and 10 ms between each pulse. Hygromycin B (Sigma Ò ) was added at 30 lg/ml for selective pressure.

Microscopy and immunofluorescence imaging
For the intracellular localisation of GFP-fused proteins or for immunofluorescence of whole cells, transfected cells grown to mid-log phase were fixed in 4% paraformaldehyde (PFA) and airdried on microscope fluorescence slides (Dubessay et al., 2006). For immunofluorescence analysis of cytoskeletons, cells were washed in PBS and deposited on 8-well slides. Cytoskeletons were extracted in 0.25% Nonidet P40 (NP40), 100 mM piperazine-N,N 0bis(2-ethanesulfonic acid) (PIPES), 1 mM MgCl 2 , pH 6.9, for 5 min at room temperature (Pradel et al., 2006) and then fixed in 4% PFA for 20 min for L. major or for 5 min for T. brucei.
To study the mitochondrion, cultivated cells were primarily incubated for 10 min at 27°C with Mitotracker Red CMXRos (500 nM, Molecular Probes Ò ) and washed in culture medium without FCS. After a second wash in culture medium with FCS, the cells were fixed as described above.
DNA was visualised by DAPI staining (0.01 lg/ml). Slides were mounted with Mowiol mounting medium (Calbiochem Ò , CA, USA) and viewed by phase contrast and immunofluorescence microscopy using appropriate filters on a Zeiss Ò Axioplan 2 microscope with a 100Â objective. Digital images were captured using a Photometrics CoolSnap CDD camera (Roper Scientific Ò , FL, USA) and processed with Metaview Software (Universal Imaging Corporation Ò , PA, USA).
2.5. Northern blot and western blot 2.5.1. Northern blot Total RNA from promastigote forms of L. major and procyclic forms of T. brucei was extracted using a RNeasy extraction kit (Qiagen Ò , Netherlands) and denaturated in a solution of 2.5 Â 3-(N-morpholino)propanesulfonic acid (MOPS), 9.25% formaldehyde and 50% formamide. RNAs were then incubated for 10 min at 65°C and 5-10 min on ice, before being separated on an agarose gel (1.4% agarose, 6% formaldehyde and 1Â MOPS). RNAs were transferred to a nylon membrane (Amersham Hybond-N+, GE Healthcare Ò , UK) and hybridized with a specific DNA probe radio-labelled with a 32 P-dCTP by random priming. The GPI-anchor transamidase subunit 8 gene (GPI8; LmjF.18.0360 and Tb10.6 1.3060) which is constitutively expressed, was used as a control.
Quantifications were done by scintillation counting of the aand b-tubulin bands after loading of the samples on a 10% SDS-PAGE and electro-transfer onto nitrocellulose, as previously described (Regnard et al., 1998). Polyglutamylase activity was also determined using ANP32B and NAP1 as substrates, at 0.2 mg/ml.

RNAi in T. brucei
Amplification of parts of T. brucei TTLL genes was performed using genomic DNA with the primers listed in Supplementary  Table S1. The PCR products including the HindIII (AAGCTT) and SacII (CCGCGG) restriction sites were cloned into a pGEMT-Easy vector (Promega Ò ) and into a p27T7tiB/GFP vector (trypanoFAN: http://trypanofan.path.cam.ac.uk/trypanofan/main). Linearised plasmid DNA (10 lg) was transfected into 3 Â 10 7 T. brucei procyclic forms using an exponential protocol (1500 V, 25 lF) on a Bio-Rad Ò Gene Pulser Xcell™ electroporation system. Transfectants were grown under drug selective pressure (phleomycin, 5 lg/ml) for 15-20 days prior to induction by tetracycline (1 lg/ml) as previously described (Blaineau et al., 2007). Cell type labelling (GT335-positive or -negative) was estimated in the presence and absence of tetracycline as described in Section 2.4. Approximately 800 cells were counted.

Distribution of glutamylated and polyglutamylated tubulin in L. major and T. brucei
While MT polyglutamylation has been reported as extensive in T. brucei from mass spectrometry data , no data are available for Leishmania. To visualise glutamylation in trypanosomatid cells, we performed immunofluorescence in L. major and T. brucei using two Abs directed against polyglutamylated tubulin: GT335 is a mAb that recognises all forms of polyglutamylated tubulin independently of the length of the polyglutamate side chain (Wolff et al., 1992), whereas PolyE is a polyclonal Ab that is specific to tubulin modified by the presence of side chains at least three glutamates long ). Contrary to the low level of polyglutamylation in interphase MTs, which is observed in most proliferating mammalian cells (Regnard et al., 1999), the trypanosomatid interphasic cells appeared intensely stained with both antibodies (Fig. 1). Moreover, in L. major cells, GT335 labelling was clearly more intense in G2/M cells (as defined from the appearance of the nascent flagellum) than in non-mitotic cells (Fig. 1, Supplementary Fig. S1). This shows (and confirms in T. brucei) that MTs are highly polyglutamylated at all stages of the cell cycle. Interestingly, the posterior third of the cell body lacked GT335 labelling in interphasic cells of both species, which was clearly confirmed by the merging of the fluorescence of both antibodies (Fig. 1). However, during the G2/M phase, this relative reduction was no longer observed in L. major promastigotes (Fig. 1A, Supplementary Fig. S1), although it persisted in T. brucei procyclic forms (Fig. 1B). This specific immunofluorescence pattern was confirmed using detergent-extracted cytoskeletons in both species (Supplementary Figs. S1 and S2). This experiment also confirmed that polyglutamylation is located in the cortex of MTs and not inside the cell body. Finally, the flagellum was always decorated with the antibodies, although less intensely than the rest of the cell (Supplementary Figs. S1-S3). In contrast with mammalian cells (Bobinnec et al., 1998), the centrioles of the basal body, as well as the mitotic spindle, were not distinctly labelled by the antibodies.
The differential distribution of GT335 labelling being reminiscent of the distribution of (de)tyrosinated a-tubulin (Sasse and Gull, 1988), we wanted to see if there was a correlation between the distribution of glutamylated and tyrosinated MTs. As expected (Sherwin et al., 1987), the tyrosination-specific antibody YL1/2 clearly labelled the basal body in both species (Figs. 2 and 3). It also intensely decorated the cleavage furrow in L. major mitotic cells (Fig. 3). Moreover, also as previously described (Sasse and Gull, 1988), YL1/2 labelling was distributed over the whole cell body, with an increase in the posterior third of the cell in T. brucei (Fig. 2), i.e. the cell region where GT335 labelling was partially deficient. This observation was also made in non-mitotic cells of L. major (Fig. 3).
3.2. Identification of the genes encoding putative TTLL proteins in L. major and T. brucei Using bioinformatics, we identified nine and 10 genes encoding proteins with a TTL domain in the L. major and T. brucei genomes, respectively. Two genes in L. major and one in T. brucei code for proteins with a high sequence homology to TTL (Table 1, Supplementary Table S1). Following phylogenetic studies described in Janke et al. (2005), the trypanosomatid TTLL proteins are classified into five families: TTLL1, TTLL4, TTLL6, TTLL9 and TTLL12. The analysis of the predicted domains of most of these putative TTLLs identified a core TTL domain with an ATP binding site, a domain which is the common hallmark of TTLL proteins, and an extended domain which is characteristic of the polyglutamylase activity (Fig. 4). Two proteins, classified as TTLL12 and present only in T. brucei, did not exhibit an extended domain and their homologue in mammals has previously been shown to not act as a tubulin glutamylating enzyme (van Dijk et al., 2007). Thus the seven remaining putative polyglutamylase proteins were examined in the two species (Table 1). mRNA levels of the putative polyglutamylase genes were evaluated by northern blot analysis of the promastigote forms of L. major and procyclic forms of T. brucei. All of the genes were clearly transcribed compared with GPI8, a constitutively expressed gene ( Supplementary Fig. S4). Using this crude measurement, ttll4a seems to show high expression levels in both species, and ttll6b seems to be expressed at very high levels in L. major, while ttll6a and ttll9 appeared to be expressed at low levels in T. brucei. These results are actually corroborated by the comprehensive RNA-seq study in L. major by Rastrojo et al. (2013), which showed RNA levels of ttl4a and ttl6b threefold higher than those of most the other genes.

Subcellular distribution of GFP-fused putative polyglutamylases
We constructed L. major recombinant strains expressing GFPfused TTLL proteins. The expression of the GFP-fused proteins in recombinant cells was verified by western blot using an anti-GFP antibody ( Supplementary Fig. S5) and their subcellular localisation was analysed using fluorescence microscopy. In order to rule out any bias due to the tag insertion, the GFP tags were introduced Trypanosoma brucei procyclic forms. For each organism, non-mitotic and G2/M cells are shown (here termed 'interphasic' and 'mitotic', respectively, for convenience). Each representative cell is viewed in phase contrast microscopy and in fluorescence microscopy using DAPI, the GT335 monoclonal antibody which labels all glutamate side chains (red), the PolyE monoclonal antibody (which labels long glutamate side chains) (green) and GT335-PolyE merging. The nucleus (N) and the single dense, intensely stained, mitochondrial DNA (K for kinetoplast) are stained with DAPI. Bar = 5 lm.
at the N-and at the C-terminal ends in independent experiments: both constructs (LmTTLL-GFPn/LmTTLL-GFPc) gave identical results for all genes except for TTLL4B (see below), indicating that the GFP tag did not modify the localisation of these proteins.
Four of the recombinant TTLL proteins (LmTTLL1, LmTTLL6A and 6B, and LmTTLL9) showed a diffuse distribution in the cell body but were excluded from the nucleus (Fig. 5). In addition, LmTTLL6B-GFP exhibited an intensely fluorescent dot at the very posterior end of the cell. LmTTLL4A-GFP showed a diffuse localisation and was particularly concentrated in the nucleus (Fig. 6). Surprisingly, LmTTLL4B-GFPc was located exclusively in the mitochondrion, as confirmed by co-localisation with the mitochondrial marker Mitotracker™ (Fig. 6). Accordingly, an in silico analysis of the sequence of LmTTLL4B revealed a mitochondrial addressing signal at the N-terminus of the protein, which also explained why the N-terminally tagged protein did not localise to the mitochondrion. Finally, LmTTLL4C-GFP was exclusively seen in the nucleus, with a pronounced accumulation at the mitotic spindle in G2/M cells, and at the midbody in anaphase (Fig. 6).

In vitro polyglutamylation activity and substrate specificity of the L. major recombinant TTLL proteins
In vitro assays with taxol-stabilised MTs (Regnard et al., 1998;van Dijk et al., 2007) were performed in order to test the tubulin polyglutamylase activity of recombinant TTLL proteins from GFP-TTLL expressing L. major cells. To determine whether these were involved in the initiation or the elongation of glutamate side  chains, we used two kinds of substrates: highly polyglutamylated MTs from brain tissue and almost unmodified MTs from HeLa cells. While both initiating and elongating enzymes can act on the brain MTs by either generating new side chains or increasing the length of the pre-existing side chains, HeLa MTs are solely modified by enzymes able to initiate the formation of side chains (van Dijk et al., 2007).
LmTTLL4A and LmTTLL6B were found to be active on both MT substrates. Their ability to modify HeLa tubulin demonstrates that these enzymes possess a chain-initiating polyglutamylase activity (Fig. 7A). LmTTLL6B preferentially modifies b-tubulin subunits, and can therefore be classified as a b-tubulin chain-initiating polyglutamylase.
In order to study the glutamylase activity towards substrates other than the tubulins, we performed the same in vitro assays for LmTTLL4A, LmTTLL4B, LmTTLL4C, LmTTLL6A and LmTTLL6B using recombinant NAP1 and ANP32B, two non-tubulin substrates strongly modified by some of the mammalian TTLLs (Regnard et al., 2000;van Dijk et al., 2008). Again, LmTTLL4A and LmTTLL6B proved to be the only clearly active enzymes: LmTTLL6B presented a high glutamylase activity on NAP1 and ANP32B, whereas LmTTLL4A presented activity on ANP32B only ( Fig. 7B and C). Very slight activity was seen on NAP1 only for LmTTLL4C and LmTTLL6A, while LmTTLL4B exhibited no significant glutamylase activity on either of the substrates (Fig. 7B and C).

Functional analysis of trypanosomatid TTLL proteins by RNAi in T. brucei
Studying loss-of-function phenotypes by gene knockout is usually cumbersome in Leishmania due to a 'mosaic aneuploidy' that leads to frequent generation of extra chromosomal copies in this organism (Dubessay et al., 2002;Sterkers et al., 2011). Moreover, RNAi is not functional in L. major due to the absence of proteins of the interference pathway (Robinson and Beverley, 2003), but can be used in T. brucei (Shi et al., 2000). We therefore independently transfected RNAi vectors targeting all seven TTLL genes into T. brucei procyclic forms, and subsequently studied the RNAi-   Supplementary Table S1. aa, amino acids. induced phenotypes such as cell growth, cytokinesis and glutamylated tubulin distribution. RNAi against TbTTLL4B, the TTLL protein identified as mitochondrial in L. major and with a similar mitochondria-addressing signal in T. brucei, induced a significant cell growth defect 5 days after induction, which was accompanied by a blockage of cytokinesis. The cytokinesis phenotype was revealed by a significant increase in the numbers of multinucleated (P3 nuclei) and anucleated cells termed 'zoids' (Robinson et al., 1995), contrasting with a low number of kinetoplasts ( Fig. 8A and B).
In contrast, RNAi targeting the remaining TbTTLL genes had no effect on growth, although all of those efficiently depleted their respective targets as shown by northern blots ( Supplementary  Fig. S6). With respect to cell morphology, a moderate but statistically significant increase in multinucleated cells and in zoids was noted in TbTTLL4A-and TbTTLL6B-RNAi cell lines, respectively, as well as a reduction in mononucleated cells (Fig. 8C). Since both proteins act as initiating polyglutamylases in vitro, we hypothesised that they could have redundant functions in vivo. A double RNAi for TbTTLL4A and TbTTLL6B was performed; induced cells exhibited no growth defect ( Supplementary Fig. S7), but the cell cycle disruption appeared even more significant (Fig. 8C, Supplementary Table S2), suggesting a possible synergy between both proteins.
We then studied the effect of RNAi on glutamylation of MTs in TbTTLL4A-RNAi, TbTTLL6B-RNAi and the corresponding double RNAi cell lines, using the GT335 and PolyE Abs (Fig. 9). Cells labelled by GT335 were counted and the mean percentage of GT335-stained cells in the population determined in all lines at day 5 after tetracycline induction. After TbTTLL4A RNAi induction, the fluorescence of PolyE observed by microscopy appeared not to be affected by RNAi, suggesting a compensating elongation of the side chains by other enzymes. By contrast, the proportion of GT335-labelled cells was reduced from 93.7% to 63.5% (noninduced and tetracycline-induced populations, respectively). No difference was observed after the TbTTLL6B-RNAi. However, following the double TbTTLL4A + TbTTLL6B RNAi (Fig. 9), the reduction in the proportion of GT335-labelled cells was much greater, from 89.2% to 47.6% in non-induced and tetracycline-induced cells, respectively. The reduction in GT335-tubulin staining following double RNAi could also be seen using western blots (Fig. 9).

Discussion
In trypanosomatids, MTs constitute a particularly essential component of the cytoskeleton, as other classical cytoskeletal components such as actin have a minimal function. As major determinants of flagellar length dynamics (Blaineau et al., 2007) and , and (C) ANP32B substrates were subjected to an in vitro glutamylation assay using extracts of L. major cells expressing the recombinant Tubulin Tyrosine Ligase-Like proteins. The controls were wild-type L. major cells, not transfected with Tubulin Tyrosine Ligase-Like proteins, from which the reference ''100%'' activity (dotted line) was defined. Incorporation of 3 H glutamate was determined by scintillation counting. The relative activity of each protein is expressed as a percentage over the total activity in the control reaction.  t-test). Cell growth decreased from day 6 after tetracycline induction. Distribution of morphological phenotypes induced by TbTTLL4B-RNAi induction shows an accumulation of multinucleated cells from day 7 (grey: NI; white: I). Significant differences compared with NI cell lines are indicated as follows (t-test, ⁄⁄ P < 0.01, ⁄ P < 0.05). (C) Phenotype analysis after induction of TbTTLL4A-, TbTTLL6B-and TbTTLL4A + TbTTLL6B-RNAi. Five hundred to 600 cells were counted in each cell line. The number of zoids (anucleated cells) and multinucleated cells (P3 N) increased, especially after performing the double RNAi. Significant differences compared with NI are indicated as follows (t-test, ⁄⁄ P < 0.01; ⁄ P < 0.05). Fig. 9. Effect of RNA interference (RNAi) targeting putative polyglutamylases on the distribution of glutamylated microtubules (MTs) in Trypanosoma brucei. The distribution of glutamylated MTs was studied in TbTTLL4A + TbTTLL6B-RNAi cell lines (ÀTet: RNAi-non-induced; +Tet: RNAi-induced) using the monoclonal antibodies GT335 (red) and PolyE (green). Only merged fluorescence images are shown. Insert: the reduction of glutamylation in the TbTTLL4A + 6B double RNAi cells was also visible using western blot with GT335 and 12G10 (anti-tubulin control antibody) monoclonal antibodies (NI: non-induced cells; I: induced cells; Tet, tetracycline). (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.) morphogenesis (Gull, 1999), they are essential for parasites that display transformations in cell shape and organelle positioning during their life cycle (Robinson et al., 1995). Finally, and classically, they participate in intracellular transport, mitosis and cell division. Due to the particular structure of MTs, assembled from heterodimers of aand b-tubulin, PTMs along these structures are likely to encode novel information for the cell, both linked to the nature, length and spacing patterns of these modifications. Indeed, the modified tubulin subunits are non-uniformly distributed along MTs and, as stated previously, ''analogous to the model of the 'histone code', diverse PTMs are proposed to form a biochemical 'tubulin code' that can be 'read' by factors that interact with microtubules'' (Verhey and Gaertig, 2007). Polyglutamylation indeed affects processes such as the interaction of MTs with kinesins, MT-associated proteins (MAPs) or MT-severing factors through a modulation of affinity depending on the polyglutamate chain length (Boucher et al., 1994;Wolff et al., 1994;Larcher et al., 1996;Bonnet et al., 2001;Lacroix et al., 2010). Here, we sought to explore the extension and relevance of this process in trypanosomatids. This report presents an updated insight into MT glutamylation and is, to our knowledge, the first published study of putative glutamylases in these parasites.
Contrasting with what is known in mammalian proliferative cells, the study of the distribution of polyglutamylation confirmed that interphasic cells are highly glutamylated in trypanosomatids. As glutamylation often accumulates in stable MTs, this is consistent with the extreme stability of the cytoskeleton of trypanosomatids (Sasse and Gull, 1988), in contrast to the highly dynamic MTs in interphasic mammalian cells. Furthermore, we observed a distinctive labelling of the posterior third of the cells, which was decorated by the Ab PolyE but not by GT335. The simplest explanation for this finding would be that fewer glutamylated tubulins are present at the posterior end of the cell but they carry, on average, longer side chains. An alternative hypothesis might be that GT335 is not as insensitive to the side chain organisation as claimed. In any case, this differential distribution of polyglutamate side chains is strikingly reminiscent of the distribution of detyrosinated atubulin. Indeed, stable MTs also often accumulate detyrosinated tubulins, whereas tyrosinated a-tubulin is a marker for newly assembled MTs; these were shown, in T. brucei, to be more present and to elongate at this same posterior end, termed 'dynamic', during the G1/S phase (Robinson et al., 1995;Sherwin et al., 1987;Sherwin and Gull, 1989; our data). The differential staining between GT335 and YL1/2, which recognises tyrosinated tubulins, is in agreement with the idea that MTs are more dynamic in the posterior third of the cell. Thus, our data show a differential and dynamic pattern of MT glutamylation in the posterior end of the cell, which we hypothesise is related to active growth of the MT corset during the cell cycle.
The differential pattern of glutamylation is modified at the G2/ M stage in L. major since, at this stage, we constantly observed intensification, as well as extension, of both GT335 and YL1/2 labelling over the whole cell body (Figs. 1 and 3). Both were particularly intense on the cytokinesis furrow during cytokinesis. Surprisingly, this was not observed in T. brucei, suggesting differences in the use of PTMs of MTs between both trypanosomatids, which might be related to the major differences (e.g. the symmetric/asymmetric nature) in cell division between both organisms (Wheeler et al., 2011).
Among the seven TTLL proteins studied here two, LmTLLL4A and LmTTLL6B, appear clearly and directly involved in MT glutamylation in vitro. This function is further supported by the marked reduction in GT335 labelling in vivo when the levels of both proteins were jointly reduced. It can be stressed that both enzymes are expressed at high levels in vivo in L. major and T. brucei ( Supplementary Fig. S4). In vitro glutamylation assays further allowed classification of LmTTLL6B as a b-tubulin initiating glutamylase. It is noteworthy that the inhibition of TbTLLL4A and/or TbTTLL6B had little effect on growth curves; nevertheless, zoids and multinucleated cells increased, particularly after double RNAi, thereby suggesting that polyglutamylation might (directly or indirectly) play a role in mitosis/cytokinesis. Finally, when we tested the in vitro polyglutamylation of two other well-known substrates of polyglutamylases, NAP1 and ANP32B, only LmTTLL4A and LmTTLL6B demonstrated clear activity on these nuclear substrates, of which two orthologues (Lmjf.31.1750/Tb927.9.5730 and Lmjf.19.0440/Tb927.10.15180) are annotated in the genome sequence.
Different hypotheses could explain the failure to observe a clear in vitro glutamylation activity for the remaining recombinant LmTTLL proteins: expression at low levels relative to the endogenous proteins, inhibition of the protein expression in the particular life cycle stage (promastigote) used for this study, inadequate substrates used to test for these proteins, or the lack of an activation step or cofactors for enzyme activity (Janke et al., 2005;van Dijk et al., 2007). In particular, TTLL1 in higher eukaryotes is known to be active only as part of a multiprotein complex (Janke et al., 2005). Therefore, the in vitro polyglutamylation assay used here does not allow us to rule out enzymatic activity of the remaining TTLL proteins.
Two putative polyglutamylases, LmTTLL4A and LmTTLL4C, were partially or exclusively (respectively) located at the nucleus. As seen above, LmTTLL4A was strongly active on ANP32B and MTs. LmTTLL4C, which is exclusively nuclear, showed weak activity on NAP1, whereas mammalian TTLL4 has very strong activity on NAP1 (van Dijk et al., 2007). As a consequence, the correct substrate for LmTTLL4C might be a nuclear protein that has not been tested here. Since inhibition of the expression of TTLL4C did not cause any notable growth reduction, we hypothesise that this action might take place at more subtle levels of regulation or differentiation, and/or that redundancy between TTLLs might occur.
Finally, LmTTLL4B exhibited a highly peculiar localisation, being exclusively present in the single mitochondrion of L. major. This putative polyglutamylase showed no activity on either MTs or NAP1/ANP32B. It is likely that the substrates of this enzyme, which were not searched for in this study, are mitochondrial proteins. Furthermore, the inhibition of this enzyme induces growth arrest and, specifically, a blockage of cytokinesis, which is induced by its expression inhibition. In trypanosomatids, cytokinesis is directly dependent upon smooth progress of the part of the cell cycle associated with the kinetoplast, the single and complex mitochondrial DNA (Robinson et al., 1995). The identification of the substrate(s) of this enzyme should assist in clarifying the complex mechanisms regulating this original step in the cell cycle (Hammarton, 2007).
In conclusion, PTMs are essential regulators of protein function and they could be particularly important in organisms such as trypanosomatids where RNA pol II promoters are nearly absent. This study about putative polyglutamylases in trypanosomatids provides insights into an original PTM, polyglutamylation, which is extensively present in these parasites. Our data strongly suggest that LmTTLL4A and LmTTLL6B are bona fide MT polyglutamylases. The putative role of the other TTLLs is not understood. It is paradoxical, in view of the importance of tubulins in these organisms and their extensive glutamylation, that the inhibition of most TTLLs had no effect on cell growth or the cell cycle. This may suggest redundancy between polyglutamylases, as indicated here by the increased effect of the double TTLL4A + TTLL6B RNAi. Also, the role of polyglutamylation may be more subtle than can be shown here and/or limited in time: for example, as shown in several organisms, it may help to stabilise or, conversely, destabilise MTs; or it may produce varying signals for MAPs, MT-severing fac-tors or molecular motors (reviewed in Janke and Bulinski (2011)). Here, as in most cases, PTMs appear not to induce a loss/gain of function but rather a subtle modulation of this function, thereby likely playing a much larger and more complex role than presently suspected, yet difficult to grasp.