Cell cycle-dependent expression regulation by the proteasome pathway and characterization of the nuclear targeting signal of a Leishmania major Kin-13 kinesin


  • Pascal Dubessay,

    1. UMR5093 CNRS/Université Montpellier I ‘Génome et Biologie Moléculaire des Protozoaires Parasites’, Laboratoire de Parasitologie-Mycologie, Faculté de Médecine, 163 Rue Auguste Broussonet, 34090 Montpellier, France.
    Search for more papers by this author
  • Christine Blaineau,

    1. UMR5093 CNRS/Université Montpellier I ‘Génome et Biologie Moléculaire des Protozoaires Parasites’, Laboratoire de Parasitologie-Mycologie, Faculté de Médecine, 163 Rue Auguste Broussonet, 34090 Montpellier, France.
    Search for more papers by this author
  • Patrick Bastien,

    1. UMR5093 CNRS/Université Montpellier I ‘Génome et Biologie Moléculaire des Protozoaires Parasites’, Laboratoire de Parasitologie-Mycologie, Faculté de Médecine, 163 Rue Auguste Broussonet, 34090 Montpellier, France.
    Search for more papers by this author
  • Lena Tasse,

    1. UMR5093 CNRS/Université Montpellier I ‘Génome et Biologie Moléculaire des Protozoaires Parasites’, Laboratoire de Parasitologie-Mycologie, Faculté de Médecine, 163 Rue Auguste Broussonet, 34090 Montpellier, France.
    Search for more papers by this author
  • Juliette Van Dijk,

    1. CRBM-UPR1086, CNRS, 1919 Route de Mende, 34293 Montpellier, France.
    Search for more papers by this author
  • Lucien Crobu,

    1. UMR5093 CNRS/Université Montpellier I ‘Génome et Biologie Moléculaire des Protozoaires Parasites’, Laboratoire de Parasitologie-Mycologie, Faculté de Médecine, 163 Rue Auguste Broussonet, 34090 Montpellier, France.
    Search for more papers by this author
  • Michel Pagès

    Corresponding author
    1. UMR5093 CNRS/Université Montpellier I ‘Génome et Biologie Moléculaire des Protozoaires Parasites’, Laboratoire de Parasitologie-Mycologie, Faculté de Médecine, 163 Rue Auguste Broussonet, 34090 Montpellier, France.
    Search for more papers by this author

*E-mail gpp@univ-montp1.fr; Tel. (+33) 467 63 55 13; Fax (+33) 467 63 00 49.


The LmjF01.0030 gene of Leishmania major Friedlin, annotated as ‘MCAK-like’, was confirmed as a kinesin with an internally located motor domain and termed LmjKIN13-1. Both the native form of the protein and a green fluorescent protein (GFP)-fused recombinant version were shown to be exclusively intranuclear, and, more specifically, to localize to the spindle and spindle poles. Cell cycle-dependent regulation of the protein levels was demonstrated using synchronized Leishmania cells: LmjKIN13-1 was highly abundant in the G2+M phase and present at very low levels after mitosis. Altogether, these features suggest that this protein participates in mitosis. The construction of systematic deletion mutants allowed the localization of the primary sequence regions responsible for nuclear targeting on the one hand, and for cell cycle-dependent variations on the other hand. A 42-amino-acid region of the carboxy(C)-terminal domain mediates nuclear import and could be defined as an atypical nuclear localization signal. Protein level regulation during the cell cycle was shown to also depend upon the C-terminal domain, where apparently redundant degradation signals are present. Putative degradation signals appear to be present on both sides and inside the nuclear localization signal. Further experiments strongly suggest a role for the ubiquitin/proteasome pathway in this cell cycle-dependent regulation. These data underline the importance of post-translational regulation of protein abundance in this ancestral eukaryote where transcriptional regulation seems to be rare or near absent.


The order Kinetoplastida comprises a large number of parasitic protozoa, among which several trypanosomatids are of high medical importance: Trypanosoma brucei and T. cruzi, the agents of sleeping sickness and Chagas’ disease respectively, and Leishmania, responsible for potentially lethal mucocutaneous and visceral kala-azar diseases. These ancient eukaryotes present highly atypical features in their molecular and cell biology (Gull, 1999; Myler et al., 1999; Clayton, 2002; McKean, 2003). In Leishmania, genes are clustered as large, either convergent or divergent, seemingly polycistronic units that can comprise tens or hundreds of genes located on one DNA strand (Myler et al., 1999; Ivens et al., 2005). Unlike classical operons, however, these clusters comprise genes that have no functional relationship. Furthermore, only very few polymerase II-specific promoters have been identified in Leishmania (Gilinger and Bellofatto, 2001; Martinez-Calvillo et al., 2003; 2004) – thus opening this recent unexpected question: ‘Life without transcriptional control?’ (Clayton, 2002).

The regulation of gene expression in trypanosomatids has actually been seen for several years as depending most probably on post-transcriptional and/or translational processes (Vanhamme and Pays, 1995; Clayton, 2002). In this regard, the role of 3′ UTRs is now well documented in Leishmania (Boucher et al., 2000; 2002). Post-translational modifications of proteins are also known in trypanosomatids. Some of those modifications, like phosphorylation, methylation, acetylation or glycosylation, change the functionality of the protein, while others like ubiquitination target the protein to the proteasome-dependent degradation pathway, thereby acting on the half-life time of the protein (Coux et al., 1996; Muratani and Tansey, 2003).

The proteasome has been found in all eukaryotes studied to date, including trypanosomatids (Robertson, 1999; Yao et al., 1999; de Diego et al., 2001; Wang et al., 2003). In mammals, the cell cycle is one of the processes highly controlled by proteasome-dependent degradation, especially through the degradation of cyclins, cyclin-dependent protein kinases (CDK) and CDK activators or inhibitors (King et al., 1996). In T. brucei, proteasome-dependent cyclin degradation is strongly suspected (Van Hellemond and Mottram, 2000) but the molecular understanding of the cell cycle in trypanosomatids remains very incomplete. As these organisms contain several single-copy organelles of which the accurate replication and correct segregation must be precisely regulated, their cell cycle presents several unique features (Hammarton et al., 2003). For example, distinct cell cycle phases have been described for the nucleus and the single mitochondrial DNA (the kinetoplast) suggesting novel cell cycle checkpoints (Robinson and Gull, 1991; Robinson et al., 1995; Ploubidou et al., 1999). Also, in these organisms, there exists a strong interdependence between cell cycle and cellular morphogenesis (Gull, 1999; McKean, 2003).

Among the many proteins involved in mitotic progression, motor proteins from the kinesin superfamily play essential roles (reviewed in Hunter and Wordeman, 2000). Kinesins are found in all eucaryotes and perform a wide variety of tasks by their ability to modulate interactions with microtubules (MTs) through binding and hydrolysis of ATP by their conserved catalytic core. The latter constitutes the major part of the highly conserved motor domain, which has an N-terminal, C-terminal or internal location. Families of kinesins, termed KinN, KinC and KinI respectively, have been defined from these three motor domain positions, although a more recent and complex nomenclature has now been proposed (Lawrence et al., 2004). KinNs and KinCs have been mostly implicated in MT-based intracellular transport. KinIs, which now essentially comprise the kinesin-13 family (Lawrence et al., 2004), exhibit an ATP-dependent MT-depolymerizing activity (Desai et al., 1999; Ovechkina and Wordeman, 2003). They are mostly involved in mitotic spindle assembly and chromosome segregation, in particular the MCAK (mitotic centromere-associated kinesin) (Walczak et al., 1996; Maney et al., 1998; Ovechkina and Wordeman, 2003).

It is noteworthy that, although differential expression in tissues of mammals has been analysed, little is known about the cell cycle-specific regulation of cellular levels of Kin13s.

Our interest in the function and gene expression regulation of proteins involved in the mitotic cycle of Leishmania brought us to study a kinesin-like protein, at first annotated as MCAK in the Leishmania genome database (GeneDB). In the present work, we have homologously expressed this kinesin gene (hereafter termed LmjKIN13-1) in L. major. The recombinant protein has a strict nuclear localization and appears differentially expressed during the cell cycle. We dissected the protein using a mutagenesis approach in order to characterize the targeting and regulation signals. Our results reveal an atypical nuclear localization signal (NLS) and strongly suggest that the proteasome might be involved in the cell cycle regulation of this MCAK-like kinesin.


Subcellular localization of LmjKIN13-1

The gene LmjF01.0030, first annotated as ‘MCAK-like’ in the sequence of L. major chromosome 1 (GenBank Accession No: NP_047029), encodes a 668-amino-acid protein that contains the highly conserved kinesin motor domain in its centre (residues 130–461). Moreover, the alignment of the motor domain sequence with that of various kinesins revealed the conservation of residues previously identified as specific for the kinesin-13 family (Fig. S1) (Ogawa et al., 2004). Consequently, gene LmjF01.0030 and its encoded protein were termed LmjKIN13-1 and LmjKIN13-1 respectively.

We expressed a green fluorescent protein (GFP)-fused version of LmjKIN13-1 episomally in L. major. Two vectors, pTH6nGFPc and pTH6cGFPn differing only by the position of the GFP gene, were used. These vectors lead to constitutive expression of the GFP and all transfectant cells emit bright green fluorescence in the cytoplasm, nucleus and flagellum (not shown). The entire open reading frame of LmjKIN13-1 was then inserted into the same expression vectors and Leishmania cells were transfected with the corresponding constructs. Both sets of transfections gave identical results, described hereafter.

The observation of asynchronized cells yielded several interesting data (Fig. 1A and B). (i) The recombinant GFP-fused proteins were strictly localized to cell nuclei; (ii) They were found in a low proportion of cells, i.e. 20–30% (26%, mean of 10 countings); (iii) > 80% of decorated cells were about to initiate, or had entered into, mitosis; this could be seen from the presence of a nascent or developing daughter flagellum, an event that precedes initiation of mitosis in trypanosomatids (Ploubidou et al., 1999); (iv) Depending on the stage of mitosis progression, the fluorescence could appear as one or two (more rarely three) dots (Fig. 1C and D) or decorate the putative mitotic spindle (Fig. 1B). The decoration of the mitotic spindle by LmjKin13-1 was further confirmed by using anti-β-tubulin antibodies. The spindle most commonly appeared as a bar-shaped region across the nucleus (Ogbadoyi et al., 2000; Kratzerováet al., 2001) (Fig. 1E). At anaphase, the kinesin-GFP was clearly concentrated at each pole of this spindle. We have further confirmed the subcellular localization of the protein by using a polyclonal anti-LmjKin13-1 C-terminus immune serum. In wild-type cells, the native LmjKin13-1 was thus shown with a similar localization to that the recombinant protein, and also essentially expressed in dividing cells (Fig. 1F). Nevertheless, for practical reasons, most of the following experiments were realized with the GFP-fused LmjKIN13-1.

Figure 1.

GFP-fused (A–E) and native (F) LmjKIN13-1 localize to the nucleus and mitotic spindle of L. major cells at the pre-mitotic and mitotic phases of the cell cycle.
A–D. Colour combinations of GFP (green) and DAPI (blue) fluorescence and phase contrast images. The fusion protein LmjKIN13-1-GFP is expressed in the nucleus of about 25% of the cells. The arrow in (B) shows a second nascent flagellum (see text for comments). LmjKIN13-1-GFP can be seen as one to three dots in the nucleus (A, C and D) or as a spindle-shaped pattern (B).
E. Colocalization of LmjKIN13-1-GFP and the mitotic spindle. The column displays images of one dividing Leishmania cell. Phase: phase contrast images; DNA: the DNA was visualized by DAPI staining; kinetoplast DNA stains more intensely than nuclear DNA; β-tub: the mitotic spindle was visualized using the anti-β-tubulin monoclonal antibody KMX; KIN13-1-GFP: GFP-fused protein; Merged: blue = DNA; red = β-tubulin; green = LmjKIN13-1-GFP.
F. Immuno-localization of LmjKIN13-1. The column displays images of two mitotic Leishmania cells. Phase and DNA: same as above. Anti-KIN13-1: LmjKIN13-1 was visualized using the anti-LmjKin13-1 immune serum. Merged: blue = DNA; green = LmjKIN13-1. Bar in E and F: 10 µm.

It is noteworthy that the morphology of the mitotic spindle did not appear to be affected by the (over)expression of the LmjKIN13-1-GFP (not shown). In actual facts, no phenotype modification was microscopically visible in the cells expressing the GFP-fused kinesin. Also, the growth curve and doubling time in exponential phase of the transfected line were found identical to that of the wild-type one (data not shown). Moreover, FACS analysis indicated that the proportions of cells in the G0/G1, S and G2/M phase in exponential growth were similar in both lines, i.e. 62.8%, 11.9% and 25.3% versus 60.7%, 12.7% and 26.6% for the wild-type and transfected lines respectively (mean of three experiments for each, data not shown).

Cell cycle-regulation of LmjKIN13-1

In order to confirm the cell cycle-regulated expression of LmjKIN13-1-GFP, we examined its expression in cell cultures partially blocked in G1 or G2/M phase. In cultures maintained in phosphate buffer saline (PBS) during 40 h (see Experimental procedures), where only 11.7% of cells were in G2+M (Fig. 2A), a similar proportion of cells (12.2%) expressed LmjKIN13-1-GFP, all of them as a single nuclear dot (Fig. 2B). On the contrary, after culture treatment with taxol, a drug that partially stabilizes the mitotic spindle and impedes chromosome segregation, thereby blocking cells in G2/M (Fig. 2C), a high increase in GFP-LmjKIN13-1 expression was observed: the protein was seen in 81% of the cells, mostly as a diffuse intranuclear spindle-shaped pattern (Fig. 2D). These data were further confirmed by immunoblotting using the anti-LmjKin13-1 C-terminus immune serum. In the LmjKIN13-1-GFP-expressing cells, two bands were labelled corresponding to the native and recombinant proteins; both appeared similarly regulated (Fig. 2E). ImageJ analysis of band intensities in Fig. 2E showed an 8.5-fold increase of the native form and a 10.3-fold increase of the GFP-fused protein between G1-blocked and G2-blocked cells (versus 1.07 × for the LACK control).

Figure 2.

LmjKIN13-1-GFP expression is reduced in the G1 and increased in the G2+M phase of the cell cycle.
A. FACS analysis of the DNA content of LmjKIN13-1-GFP – expressing L. major cells synchronized in G1 (see Experimental procedures). The pattern is representative of an accumulation of G1 cells (2N peak).
B. Microscopic examination of the population in (A). About 12% of cells expressed LmjKIN13-1-GFP (green) and all of them exhibited two flagella (arrow). GFP (green) and DAPI (blue) fluorescence were combined with a phase contrast image.
C. Similar FACS analysis of LmjKIN13-1-GFP-expressing mutants after 6 h of taxol treatment, leading to an accumulation of cells in G2+M (4N peak).
D. Microscopic examination of the population in (C). The majority of cells (actually 81% over several experiments) were decorated, exhibiting a diffuse fluorescence essentially restricted to the nucleus. The abnormal fluorescence pattern in these cells is interpreted as due to toxic cumulating effects of taxol on cell MTs after several hours. Most decorated cells (79%) exhibited two flagella, in agreement with the cell cycle arrest in G2/M. GFP: green; DAPI: blue.
E. Immunoblotting. Following synchronization, protein extracts were prepared and assayed for up- and downregulation by Western blotting. Equal amounts of parasites (5 × 106) were fractionated on a 8% SDS-PAGE gel and probed either with anti-LmjKin13-1 immune serum at a 1:50 000 dilution (upper panel) or with anti-LACK immune serum at a 1:20 000 dilution as a loading control (lower panel). a = asynchronous wild-type cells; b = asynchronous LmjKIN13-1-GFP expressing cells; c = same as b blocked in G1; d = same as b blocked in G2+M. Both the native LmjKin13-1 (open arrow) and the recombinant GFP protein (closed arrow) were decreased in G1 cells and increased in G2+M when compared with asynchronous cells. The low-intensity signals of both proteins in cells blocked in G1 may be related to the persistence of a low proportion of G2 cells in this line (see text and Fig. 2A and B). The molecular mass standards (kDa) are shown on the left.

Identification of the sequences responsible for nuclear targeting

In order to identify the amino acid sequences that mediate nuclear targeting, we constructed a series of deletion mutants of the LmjKIN13-1 open reading frame, fused the GFP gene to these partial fragments in C-terminal position, and examined the subcellular location of the resultant chimerical proteins. Among six deletion mutants, only those containing the C-terminal region (mutants b, c and g in Fig. 3A) remained strictly localized to the nucleus (Fig. 3B). In contrast, the mutant lacking this region only (e in Fig. 3A) exhibited a diffuse cytoplasmic localization and obvious reduced nuclear targeting (Fig. 3C). Finally, those containing either the N-terminal region alone (d, Fig. 3A) or the motor domain alone (f, Fig. 3A) localized essentially to the cytoplasm and weakly to the nucleus (not shown). These data strongly suggest that the C-terminal domain of LmjKIN13-1 contains all the information required for strict nuclear targeting. It is noteworthy that mutants b, c and e were also constructed with the GFP in N-terminal position, and in all three instances, the position of the GFP did not influence the phenotype observed (not shown).

Figure 3.

Nuclear localization of LmjKIN13-1-GFP depends on the C-terminal domain.
A. Schematic description and subcellular localization of GFP-fused LmjKIN13-1 deletion mutants. The letter codes of the chimerical proteins are shown on the left: (a) is the full-length recombinant LmjKIN13-1-GFP; (b–g) are mutant proteins bearing deletions of the N-terminal, C-terminal and/or motor domains. Hatched box: motor domain; open and closed boxes: N-terminal (Nter) and C-terminal (Cter) domains respectively. Figures indicate amino acid positions on the full-length native protein. The dashed line in (g) indicates the portion of LmjKIN13-1 that is absent from the protein. N and C: nuclear and cytoplasmic, respectively, subcellular localizations of LmjKIN13-1 mutants. Star: GFP position (for mutants b, c and e, see text).
B. Microscopic examination of L. major cells expressing the GFP-fused C-terminal domain of LmjKIN13-1 (mutant c in A). One or more dots were seen in the nucleus of cells exhibiting two flagella (arrow), hence in the G2+M phase. GFP: green; DAPI: blue.
C. Microscopic examination of a mutant fusion protein where the whole C-terminal domain of LmjKIN13-1 has been deleted (mutant e in A). Left: two mutant e GFP-expressing cells (green). Right: DAPI staining. The kinetoplasts appear larger than normal because the less intense staining of the nuclei (N) had to be artificially increased in the image processing.

To further delimit the nuclear targeting signal, a systematic dissection of the C-terminal region was then undertaken using the same deletion approach as above. Nineteen different deletion mutant proteins, derived from either the whole LmjKIN13-1 or the C-terminal domain alone (Fig. 4A), were fused to the GFP gene and their subcellular localization analysed as above. This allowed the delineation of residues 519–579 as a minimal nuclear targeting domain (see legend of Fig. 4A). A GFP-fusion of these 60 amino acids (protein w in Fig. 4A) was strictly localized to the nucleus (Fig. 4B), allowing to designate this peptide as containing an NLS.

Figure 4.

Characterization of an NLS in the C-terminal domain of LmjKIN13-1.
A. Localization of the putative NLS between residues 519 and 579. Letter codes, box colours and symbols as in Fig. 3A. The C-terminal domain was successively shortened from its C-terminal end using either the full-length LmjKIN13-1 (mutants h-l and e) or the C-terminal region alone (mutants c and m–q). The C-terminal domain was also shortened from its N-terminal end (mutants r–v) and from both ends (mutants w–y). Proteins a, h, i and c, m, n, r and s exhibited a similar nuclear localization (N on the right-hand side). In contrast, deletions in mutants j–l, e, o–q, t–v and x–y induced a loss of nuclear import, indicating peptide 519–579 as the minimal domain containing an NLS. A GFP-fusion of this peptide alone (mutant w) was shown to be strictly nuclear.
B. Nuclear accumulation of the mutant w fusion protein. The fluorescence phenotype is different from that of the full-length recombinant protein (Fig. 1), as here every cell is decorated and the whole nucleus shows homogeneous fluorescence. GFP: green, DAPI: blue (only the kinetoplast is visible).
C. Polyalanine substitutions in the amino acid sequence of peptide 519–579. Basic amino acids arginine (R) and lysine (K) are in bold and underlined. Residues changed to alanine are shown as As, while unchanged residues are indicated as dashed lines. The letters on the right-hand side indicate the subcellular localization of the various mutant proteins. N, nuclear; C, cytoplasmic; C/N, simultaneous cytoplasmic and nuclear localization (in the same cell).
D. Treatment with metabolic inhibitors reduced nuclear accumulation of the GFP-fused peptide 519–579 (mutant w, Fig. 4A and B). L. major cells expressing mutant protein w were treated with 10 mM 2-deoxy-D-glucose and 10 mM sodium azide during 6 h. A loss of motility was observed and the GFP fluorescence equilibrated between the nucleus and the cytoplasm. GFP: green; DAPI: blue.

Further analysis showed that this short sequence contains 12 basic amino acid [arginine (R) and lysine (K)] residues, these being classically involved in monopartite or bipartite NLSs (Jans et al., 2000). In order to confirm the possible role of these residues in nuclear import activity, we carried out a series of polyalanine substitutions across the 60-amino-acid sequence using the full-length GFP-LmjKIN13-1 protein as a substrate (Fig. 4C). The mutation of residues 520–531 did not modify nuclear targeting. In contrast, the three other substitutions did modify the nuclear import of the recombinant protein, either partially (residues 564–574) or completely (residues 532–543 and 544–555) (not shown). This implies (i) that the NLS is delimited by positions 532–574 and (ii) that the residues involved in this NLS are somewhat widely distributed along this peptide, thus indicating an atypical NLS.

The cell cycle-dependent regulation of GFP-LmjKIN13-1 abundance depends upon the C-terminal domain

As we observed marked cell cycle-dependent variations of the intracellular levels of LmjKIN13-1 expressed from a vector normally driving constitutive expression, (i) we could infer that the protein sequence contained all the information necessary for this regulation and (ii) the most likely hypothesis was that the protein was degraded at the end of mitosis. We used the same constructs as above to explore the sequences involved in this regulation of LmjKIN13-1 abundance.

The expression of the GFP-fused C-terminal domain alone (mutant c, Fig. 3A) appeared cell cycle-regulated because it decorated essentially dividing cells, as also observed with other mutants containing the C-terminal domain (mutants b, g, Fig. 3A). On the contrary, the expression of the GFP-fused N-terminal domain alone, motor domain alone or both (mutants d, f and e respectively, in Fig. 3A) yielded a diffuse fluorescence in all cells counted (Fig. 3C). Thus, the regulation appeared to depend exclusively upon the C-terminal domain of the protein.

In order to determine the precise location of the sequences responsible for this regulation, we used the C-terminal deletion mutants (Fig. 4A). With the exception of mutant w (see below), all fusion proteins that had retained a nuclear localization (mutants h, i, c, m, n, r and s) were regulated (not shown). Conversely, all fusion proteins that had lost nuclear targeting (j–l, e, o–q, t–v and x–y) were stabilized, i.e. were expressed in all cells present (not shown). Recombinant protein w was thus the only one being both localized to the nucleus and not cell cycle-regulated (Fig. 4B). Moreover, the addition of the C-terminal domain fragment lying either upstream or downstream of the NLS (fusion proteins n and s respectively, in Fig. 4A) restored the regulation (not shown). However, upon further dissection of the NLS, the whole of the NLS was not found necessary for the expression regulation of LmjKIN13-1, when the rest of the protein was present: in this context, the three polyalanine substitutions described above in the NLS (Fig. 4C), which (completely or partially) abolished nuclear targeting, did not induce any loss of regulation (not shown). These three cytoplasmic mutant proteins were still cell cycle-regulated. On the whole, these results indicate that the NLS is necessary but not sufficient for the regulation of LmjKIN13-1 expression, and that the signals mediating regulation in this gene are dispersed both within the NLS and upstream/downstream of the NLS.

Inhibition of active nuclear transport

We then tested the effect of active nuclear transport inhibition upon nuclear targeting and cell cycle regulation of LmjKIN13-1. For this, we cultivated Leishmania cells expressing the GFP fused to the whole LmjKIN13-1, the C-terminal domain alone or the 60-amino-acid sequence comprising the NLS (mutants a, c and w respectively, in Fig. 4A) in medium containing both beta-2-deoxyglucose and sodium azide (Marchetti et al., 2000). In these conditions, the expression of the first two chimerical proteins yielded a marked reduction in nuclear fluorescence as compared with the controls, but no cytoplasmic fluorescence (not shown). In contrast, the fusion protein w was also reduced in the nucleus but accumulated in the cytoplasm (Fig. 4D). The inhibition of the nuclear import of the latter mutant protein was reversible. Indeed, after drug treatment, cells were washed and cultivated again in complete fresh medium. After 1 h, most cells had recovered their motility and 50% of them exhibited exclusive nuclear fluorescence (not shown). These results tend to suggest that the drugs partially/slowly inhibit nuclear import of the three recombinant proteins.

Role of the proteasome in cell cycle regulation of LmjKIN13-1 expression

We tested whether the proteasome was involved in the cell cycle regulation of LmjKIN13-1 by treating Leishmania cells expressing LmjKIN13-1-GFP independently with two proteasome inhibitors, epoxomicin and MG132. Treatment of a mid-log phase L. major culture for 8 h with epoxomicin (Fig. 5A) or 6 h with MG132 (Fig. 5B) lead to a dramatic increase in cells expressing fluorescence, from about 25% to 91%, with a high proportion (54%) of cells apparently blocked in the G2+M phase. Similar results were obtained when using the GFP-fused C-terminal domain of LmjKIN13-1 (mutant c in Fig. 4A; data not shown). At this stage, this accumulation of LmjKIN13-1-expressing cells could be explained either by a direct effect of proteasome inhibition upon the degradation of LmjKIN13-1 or by an increase in the number of cells in the G2+M phase. In order to decide between both hypotheses, we cultivated L. major in MG132-containing medium, and monitored the number of cells in G1 phase which expressed or not LmjKIN13-1, as well as the number of cells in G2+M phase, over 4 h. The comparison of the curve slopes showed that the proportion of monoflagellated cells expressing LmjKIN13-1 increased much more rapidly and more markedly (4.6 fold) than that of cells blocked in G2+M (1.8-fold) (Fig. 5C and D). These data showed that expression of LmjKIN13-1 preceded the blocking of cells in G2/M and therefore rather support the first hypothesis above, i.e. an involvement of the proteasome in the regulation of LmjKIN13-1.

Figure 5.

Proteasome inhibitors induce an increase in LmjKIN13-1-GFP intracellular levels in L. major. After 8 h of treatment with 10 µM epoxomicin (A) or 6 h with 10 µM MG132 (B), > 90% of cultivated cells expressed the recombinant protein. In (A), nuclei appeared uniformly decorated and cytoplasmic diffusion was seen in about half of the cells. In (B), the intensity of the nuclear fluorescent dots was increased as compared with non-treated controls (not shown). Both in (A) and (B), a microscopic field showing a concentration of cells blocked in the G2+M phase was selected. GFP: green; DAPI: blue.
C. The GFP-LmjKIN13-1 expressing cell line was cultivated in the presence of 10 µM MG132 and culture samples were analysed for fluorescence microscopy at various time intervals from 0 to 4 h. Each measure represents (i) (squares) the percentage of monoflagellated cells expressing LmjKIN13-1-GFP over all monoflagellated cells and (ii) (triangles) the percentage of biflagellated cells over all cultivated cells (see text).
D. Microscopic examination of L. major cells expressing LmjKIN13-1-GFP after 4 h of MG132 treatment. The proportion of monoflagellated cells expressing LmjKIN13-1-GFP (squares in C) is substantially increased as compared with the full-length recombinant protein pattern (Fig. 1).


In the present work, we have studied in the subcellular localization and the expression regulation of the kinesin LmjKIN13-1. The corresponding gene, LmjF01.0030, was clearly identified by computer analysis as a KinI related to the kinesin-13 (formerly MCAK/KIF2) family.

We show here that LmjKIN13-1, both in its native form and as a recombinant protein expressed from an episomal vector, exhibits several interesting and sometimes unusual features as regards KinIs. (i) The protein has an exclusively nuclear localization; and we provide functional evidence for an atypical nuclear targeting signal located in its carboxyl-terminal domain; (ii) LmjKIN13-1 localizes essentially to the poles of the mitotic spindle, a structure that is not perfectly characterized in trypanosomatids; and (iii) Its expression is markedly regulated during the cell cycle; and regulation signals are located in the C-terminal domain sequence.

Characterization of an atypical NLS

LmjKIN13-1 remains strictly localized in the nucleus. This contrasts with the classical MCAK (Wordeman and Mitchison, 1995; Wordeman et al., 1999) and its Xenopus homologue XKCM1 (Walczak et al., 1996), which exhibit a cytoplasmic and/or nuclear localization.

Generally, the import of large proteins into the nucleus is an active process that may depend primarily upon the presence of an NLS (Jans et al., 2000). Using the psort program (http://psort.nibb.ac.jp/form.html), we identified two potential monopartite NLSs, located at position 140 (RKRP) and 548 (KRPR) respectively. Our data show that the first one is not functional. In contrast, polyalanine substitution of a 12-residue peptide encompassing the second NLS (residues 544–555) lead to a complete loss of nuclear targeting. In actual facts, the region mediating nuclear targeting of LmjKIN13-1 is much larger than this simple monopartite NLS and encompasses residues 532–574, i.e. 42 amino acids. This stretch contains 11 dispersed basic amino acids (three Lys and eight Arg), which are classical components of NLSs (reviewed in Jans et al., 2000). Experimental evidence shows that our NLS does not conform to a classical polybasic NLS but rather to an atypical class of NLSs such as that of the adenovirus Ela gene (Standiford and Richter, 1992) or that of the protooncogen c-myc (Makkerh et al., 1996). For example, the first one is made of three distinct groups separated by neutral spacer residues, with deletion of the different groups having variable effects on nuclear import, a fact that recalls our observations. Atypical NLSs may reflect the diversity and complexity of the nuclear import pathway (Jans et al., 2000).

Post-translational regulation of LmjKIN13-1

The expression of LmjKIN13-1 appeared cell cycle-regulated. Our microscopical studies, together with the FACS data, strongly suggest that the protein is expressed before the onset of mitosis, and that it disappears after the end of telophase and cytokinesis, i.e. it is expressed at the G2+M phase.

The regulation of KinIs during the cell cycle remains largely undocumented. The protein levels of mammalian MCAK have been shown to increase as cells progress towards mitosis, but they remain high in the cytoplasm during interphase (Wordeman et al., 1999). In contrast, this report demonstrates drastic variations of a nuclear KinI levels during the cell cycle in Leishmania, with protein degradation at the end of mitosis.

In this study, we have inserted the LmjKIN13-1 gene in the expression vector between the 5′ and 3′ UTRs flanking the Leishmania dihydrofolate reductase-thymidilate synthase (DHFRTS) gene; these regions are widely used and normally lead to constitutive expression in this organism (Lebowitz et al., 1990). Nevertheless, in the same context, the GFP-fused LmjKIN13-1 is only expressed at the G2+M stage. This implies that the protein sequence contains all the information necessary for this regulation: the dissection of the protein permitted us to show the essential role played by the C-terminal domain. More precisely, the study of numerous mutant proteins permitted us to locate the regulation signal(s) at the NLS plus upstream and/or downstream flanking sequences. Indeed, recombinant proteins comprising the latter flanking sequences plus the whole NLS (n and s in Fig. 4A) are still regulated. In contrast, the same proteins but with a ‘leftwards’ or ‘rightwards’ shortened NLS (o and t in Fig. 4A) lose both the regulation and the nuclear localization.

A more precise characterization of these regulation signals may prove complex as they appear, like certain other degradation signals (Ciechanover and Ben-Saadon, 2004), to be functionally redundant at the amino acid level. Indeed, our data suggest a compensation mechanism where the absence of certain amino acids involved in regulation and located on one side of the NLS may be offset by similar residues on the other side of the NLS.

Which regulation/degradation pathway for LmjKIN13-1?

Given that the cell cycle regulation of the expression levels of LmjKIN13-1 depends on its primary sequence, the most likely explanation for this regulation is that there is a degradation of the protein at the end of mitosis and that this degradation is inhibited prior to the next mitosis. The ubiquitin/proteasome degradation pathway appears as an essential mode of regulation of proteins involved in cell cycle transitions (reviewed in King et al., 1996) but is not known for MCAK or its orthologues. In our study, the inhibition of proteasome activity led to a stabilization of LmjKIN13-1-GFP but also to an accumulation of cells in G2+M phase. These results were also observed after utilization of taxol, which blocks the cell cycle in mitosis, and therefore might be interpreted as an indirect effect of the accumulation of cells in G2+M. However, the comparative time-course of increase of LmjKIN13-1-expressing monoflagellated cells versus G2+M cells after MG132 treatment rather suggest a direct effect of the proteasome in the degradation of this protein.

Correct proteolysis by the proteasome involves the interaction of a substrate-specific ubiquitin ligase termed the anaphase-promoting complex (APC) (King et al., 1996) with a sequence signal in the substrate protein. However, no destruction signal sequences could be identified within the C-terminal domain of LmjKIN13-1, whether these were of the destruction-box type (Glotzer et al., 1991; King et al., 1996; D-Box Motif Finder: http://bioinfo.weizmann.ac.il/~danag/d-box/main.html) or of the KEN-box type (Pfleger and Kirschner, 2000). Another pathway for targeting a protein to the proteasome necessitates the presence of a PEST sequence for degradation but independently of its ubiquitination (Gregory and Hann, 2000). PESTfind Analysis (http://www.at.embnet.org/embnet/tools/bio/PESTfind/) identified two potential PEST motifs in the LmjKIN13-1 sequence: one in the N-terminal (residues 76–100) and the other one in the C-terminal (residues 461–480) part. However, neither was found involved in the protein regulation according to our experimental data. It is possible that proteasome targeting may be conditioned by yet unknown signals, particularly in ancestral eukaryotes like Leishmania.

As trypanosomatids apparently lack transcriptional control of gene expression, one can expect numerous and perhaps original examples of post-transcriptional, translational and post-translational regulation in these organisms. As regards gene product levels, if post-transcriptional regulation is well documented in these organisms (Boucher et al., 2000; 2002), we found very few reports (Van Hellemond and Mottram, 2000) of post-translational regulation of protein abundance, of which our study is one more example.

Possible role of LmjKIN13-1?

The nuclear localization, fluorescence pattern and cell cycle-regulated expression of LmjKIN13-1 all strongly suggest that the protein has a role in mitosis, most probably in mitotic spindle formation and perhaps chromosome segregation. Its exact role could not be investigated further in this study. In particular, gene knock-outs (KOs) were not attempted here in view of the extraordinary ability of this organism to create supernumerary genes or chromosomes when subjected to KOs of essential genes (Dubessay et al., 2002; Martinez-Calvillo et al., 2005; our unpublished data). RNAi is not efficient either in Leishmania. In higher eukaryotic systems, overexpression of MCAK/XKCM1 enhanced the catastrophe frequency of the MTs, leading to the formation of aberrant spindles (Maney et al., 1998; Kline-Smith and Walczak, 2002; Newton et al., 2004). In contrast, it is noteworthy that ‘overexpression’ of LmjKIN13-1 did not produce any noticeable phenotype according to morphological, in vitro growth or cell cycle criteria. This might be explained by the fact that this ‘overexpression’ is actually limited here by protein degradation. Alternatively, it is possible that the morphological definition in these microorganisms is too poor to allow for the detection of any but the more severe defects. Finally, another possibility may be that LmjKIN13-1 is a functional homologue of human KIF2A (Ganem and Compton, 2004) that also localizes essentially to centrosomes/spindle pole bodies and to the spindle. Further studies will obviously be necessary to determine the exact role and protein partners of LmjKIN13-1, as well as its evolutionary relationship with its mammalian counterparts. The accumulation of data upon this protozoon will help in clarifying the molecular organization of the mitotic spindle in other ancient eukaryotic models.

Experimental procedures

Parasite, cell culture and transfection

The Leishmania strain used here was the reference strain of the Leishmania genome sequencing project, L. major Friedlin (Ravel et al., 1998). Promastigotes were grown in supplemented RPMI1640 medium as described (Dubessay et al., 2004). For transfection, 8 × 107 promastigotes grown to mid-log phase were electroporated with 100 µg of episomal DNA plasmid, and grown under selection pressure with hygromycin at 30 µg ml−1.

For synchronization assays in G1 phase, mid-log phase cells were pelleted at 1300 g for 10 min and resuspended at 106 ml−1 in PBS pH 7.4 containing 5% RPMI medium, 1% fetal calf serum and 30 µg ml−1 hygromycin B (all reagents Invitrogen). After a 40 h incubation at 26°C, cells were processed for fluorescence imaging as described below, as well as for flow cytometry. For the latter, cells were washed with PBS and fixed with 70% ethanol in PBS for 1 h at 4°C. Cells were pelleted by centrifugation, resuspended in 0.5 ml of PBS containing 20 µg ml−1 RNAse A and incubated for 30 min at 37°C. Cells were again centrifuged, resuspended in PBS containing propidium iodide (Sigma) at 20 µg ml−1 for at least 20 min, and then analysed for DNA content with a FACSCalibur flow cytometer (Becton Dickinson); 104 parasites were counted for each measurement. Cell cycle profiles and subpopulations in G0-G1/S/G2-M phases were analysed using the FlowJo (Tree Star) software.

Drug treatment

In order to block cells in G2 phase, 5 ml of mid-log phase Leishmania cells were treated with 25 µM taxol (USB) during 6 h at 26°C. Four millilitres of the culture were processed for flow cytometry as described above and 1 ml for fluorescence as described below. The inhibitor used against proteasome activity was epoxomicin (Affinity Res. Products) or MG132 (Sigma). Mid-log phase cultures were cultivated with 10 µM epoxomicin for 8 h or 10 µM MG132 for 1–6 h at 26°C before processing for fluorescence analysis. For active nuclear transport inhibition, mid-log phase cells were centrifuged and resuspended in glucose-depleted RPMI1640 medium supplemented as described (Dubessay et al., 2004). 2-deoxy-D-glucose and sodium azide were added to the culture at 10 mM (Marchetti et al., 2000) and cells maintained at 26°C during 6 h before being tested for fluorescence.

Construction of the Leishmania GFP-fused proteins expression vectors pTH6nGFPc and pTH6cGFPn

For all plasmid constructs, the hygromycin-resistance gene (hyg), as well as the 5′- and 3′- regions of the L. major DHFRTS gene were polymerase chain reaction (PCR)-amplified using plasmid pVV-HYG as a template (Dubessay et al., 2004). The coding region of the GFP was a modified version of the mutant GFP S65T (ClonTech), amplified from plasmid pVV917-GFP (P. Dubessay, unpubl. data). The hyg gene, flanked with fragments of the 5′-DHFRTS (600 bp) and 3′-DHFRTS (250 bp) regions, was PCR-amplified using oligonucleotide primers 5′-ggggtaccgcatcgacatcgcgcacg-3′ and 5′-ggggtcgacgtgagacgaggccggtgc-3′, containing KpnI and SalI restriction sites (underlined) respectively. The PCR product (1800 pb) was inserted into the KpnI and XhoI sites of plasmid BlueScript (pBS II KS), thus generating pBS-HYGMtp1.

The GFP gene, flanked with fragments of the 5′- and 3′-DHFRTS regions, and containing a 6X-histidine tag (H6) and the joined MfeI and HpaI cloning sites immediately upstream (for pTH6nGFPc) or downstream (for pTH6cGFPn) of the GFP gene, was introduced into pBS-HYGMtp1 following the different steps described below:

(i) pTH6nGFPc.  First, a 900 bp fragment of the 5′-DHFRTS region was PCR-amplified using Primer ♯1 (5′-gggaagct tggggtgatggagagaatg-3′) and ♯2 (5′-gtgatgatggtgatggtgcat cactagtgctcggacccg-3′) containing the HindIII restriction site (underlined) and a start codon (in bold) flanking a 6X-histidine tag (underlined) respectively. This PCR product was secondarily PCR-amplified using Primer ♯1 and Primer ♯3 (5′-gctctagagttaacgggcaattggtgatgatggtgatggtg-3′) containing XbaI, HpaI and MfeI restriction sites (underlined). The resulting DNA fragment was cloned into the XbaI–HindIII sites of pBS-HYGMtp1, yielding pBS-HYGMtp2. Second, a 1000 bp fragment of the 3′-DHFRTS region was PCR-amplified with Primer ♯4 (5′-gggttaacgtttaaaccagtagatgccgaccgg gat-3′) and ♯5 (5′-gctctagacacaggaatcgtgtgaaacc-3′), containing HpaI–PmeI and XbaI restriction sites (underlined) respectively, and inserted into the HpaI–XbaI sites of pBS-HYGMtp2. Finally, the GFP gene was generated by PCR amplification using Primer ♯6 (5′-ggcaattgggggttaacgtgag caagggcgaggag-3′) containing MfeI–HpaI restriction sites (underlined) and ♯7 (5′-gggtttaaaccttacttgtacagctcgtcc-3′), containing the PmeI site (underlined) and a stop codon (bold). The PCR product was cloned into the PmeI–MfeI sites of pBS-HYGMtp2, thus generating the final plasmid construct pTH6nGFPc which carries a histidine tag and the GFP upstream and downstream, respectively, of the MfeI–HpaI cloning site.

(ii) pTH6cGFPn.  Primer ♯8 (5′-caccatcaccatcatcactagagta gatgccgaccgggat-3′) containing a 6X-histidine tail (under lined) and a stop codon (bold), and Primer ♯5 were used to amplify the 3′-DHFRTS region. This PCR product was in turn PCR-amplified using Primer ♯9 (5′-ggaagcttgggttaaccaccat caccatcatcactag-3′) and Primer ♯5 to add HindIII–HpaI sites (underlined) to the amplified 3′-DHFRTS region, which was inserted into the HindIII–XbaI sites of pBS-HYGMtp1 to generate pBS-HYGMtp3. The 5′-DHFRTS region was PCR-amplified using Primer ♯1 and ♯10 (5′-gggttaacgggcaat tgggggtttaaaccactagtgctcggacccg-3′), containing the HpaI, MfeI and PmeI sites (underlined). The resulting PCR product was inserted into the HpaI–HindIII sites of pBS-HYGMtp3, yielding pBS-HYGMtp4. In a final step, the GFP gene, amplified with Primer ♯11 (5′-gggtttaaacatggtgagcaagggcgag-3′) and ♯12 (5′-ggcaattgcttgtacagctcgtccatg-3′), containing the PmeI site (underlined) plus a start codon (bold) and an MfeI site (underlined) respectively, was inserted into the PmeI and MfeI sites of pBS-HYGMtp4 in order to generate the final construct pTH6cGFPn, in which the fusion protein is tagged in C-terminal with histidine and in N-terminal with the GFP.

Construction of GFP-LmjKIN13-1 fusion proteins and mutant derivatives

The coding region of the LmjKIN13-1 (GenBank Accession ♯ NP_047029), with the start and stop codons removed, was PCR-amplified from L. major‘Friedlin’ genomic DNA with specific oligonucleotides 5′-ggcaattgtcggccgagccgccgtcgtcg cag-3′ and 5′-gggttaaccgttggcggcaggtgctgctgaaaggc, containing the MfeI and HpaI restriction sites (underlined) respectively. The PCR product, purified and digested with MfeI–HpaI, was cloned into the MfeI and HpaI sites of pTH6cGFPn and pTH6nGFPc, generating pTH6cGFPn-LmjKin13-1 and pTH6nGFPc-LmjKin13-1, where LmjKIN13-1 is fused to GFP in C-terminal and N-terminal respectively. For mutant derivative constructs, targeted deletions in the LmjKIN13-1 protein were generated by PCR using specific oligonucleotides, and truncated LmjKIN13-1 proteins were expressed either in pTH6cGFPn or in pTH6nGFPc. Polyalanine substitutions in the C-terminal domain of LmjKIN13-1 were generated using a PCR strategy similar to the one described above. For each substitution mutant, the two parts of the LmjKIN13-1 coding region flanking the targeted element to be substituted were independently PCR-amplified. The upstream part of the gene was amplified using a forward primer containing SacI+MfeI restriction sites and a reverse primer made of one NotI site and a stretch of alanine codons. The downstream part was amplified using a forward primer made as the last one above and a reverse primer containing the HpaI+EcoRV sites. Both PCR products were then successively cloned into pBS (SacI and NotI and NotI and EcoRV sites respectively). The insert was then excised by MfeI–HpaI restriction and cloned into pTH6cGFPn. The conservation of the reading frame between GFP and LmjKIN13-1 (wild type and mutants) was systematically confirmed in all constructions by nucleotide sequence analysis.

Microscopy and fluorescence imaging

For intracellular localization analysis of GFP fusion proteins, transfected cells were grown to mid-log phase, collected by centrifugation, washed twice in PBS and fixed 30 min in 4% paraformaldehyde (PFA). Fixed cells were then washed twice with PBS, and DNA was stained with 1 µg ml−1 4,6-diamino-2-phenylindole (DAPI) in the second wash. Cells were then air-dried on microscope immunofluorescence slides and rinsed briefly with distilled water before mounting.

For immunofluorescence detection of the mitotic spindle, cells were fixed and air-dried on microscope glass slides as above, excluding the DAPI staining. Slides were then incubated sequentially with 0.1% Triton in PBS for 10 min, 5% skimmed milk in PBS for 40 min, 1:100 dilution of mouse anti-β-tubulin monoclonal antibody in 5% skimmed milk-PBS for 45 min, 4 µg ml−1 of rabbit anti-mouse IgG antibody conjugated with the Alexa Fluor 546 dye (Molecular Probes) in 5% skimmed milk-PBS for 45 min and finally 1 µg ml−1 of DAPI in PBS for 5 min; all these steps being interspersed with PBS washes. The anti-β-tubulin antibody (KMX) was kindly provided by Keith Gull, University of Oxford, UK (Ogbadoyi et al., 2000). The protocol for immunofluorescence using the anti-LmjKIN13-1 C-terminus immune serum was identical except that the Ab was used at a 1:2000 dilution and revelation was performed using 4 µg ml−1 of goat anti-rabbit IgG antibody conjugated with the Alexa Fluor 488 dye (Molecular Probes).

All slides were mounted in Mowiol (Calbiochem). Leishmania cells were viewed by phase contrast, and fluorescence was visualized using appropriate filters, on a Zeiss Axioplan 2 microscope with a 100× objective. Digital images were captured using a Photometrics CoolSnap CDD camera (Roper Scientific), and processed with MetaView (Universal Imaging). Certain images were further processed using the Huygens deconvolution software (Scientific Volume Imaging B.V.) at the ‘RIO imaging’ Platform (Montpellier, France).

Estimates of the proportions of cells expressing the GFP-tagged proteins were made by counting 400 cells for each experiment in several experiments.

Antibody production

A 560 bp fragment of the LmjKIN13-1 gene was PCR-amplified using primers 5′-GCCAGGATCCCGGAGGACGA CAACTCGCCTTTC-3′ and 5′-GGCTGCAGGTGCTGCTGA AAGGCGGTGATG-3′. The resulting fragment was cut by BamHI and PstI and cloned in frame into the similarly digested Escherichia coli expression vector pQE30 (Qiagen). This allowed the expression of a truncated version of the LmjKin13-1 protein, corresponding to part of the C-terminal domain of the protein (amino acids 483–668) with an N-terminal 6xHis-tag. The expressed fusion protein was purified from inclusion bodies over a nickel column (HisTrap, Amersham) according to manufacturer's instructions and in the presence of 7.5 M urea. Following dialysis against PBS, the protein was recovered as a precipitate. The precipitated protein was used for production of rabbit polyclonal immune serum.

Western blot analysis

Leishmania promastigotes cellular pellets were lysed with SDS in presence of an anti-protease cocktail (Sigma). Whole cell lysates were boiled for 10 min and 5 × 106 cells per well were electrophoresed on 8% SDS-polyacrylamide gels with standard markers (Fermentas). Proteins were transferred to Hybond PVDF membranes (Amersham) by semidry blotting (Kyhse-Andersen, 1984). Membranes were treated using an ECL Advance Western blotting detection kit (Amersham) as described by the manufacturer. Filters were then incubated for 1 h at room temperature with polyclonal rabbit immune serum at a 1:50 000 dilution and revealed with a goat anti-rabbit IgG peroxydase conjugate (Sigma) at a 1:100 000 dilution. Antibody binding was then visualized by enhanced chemiluminescence (ECL, Amersham) using ECL-Hyperfilm autoradiography films. As an internal control, an equal amount of each cell protein extract was loaded and incubated with a rabbit anti-LACK immune serum at a 1:1000 dilution (Prina et al., 1996). The anti-LACK antibody was generously provided by Eric Prina (Institut Pasteur, Paris). Band intensities were analysed using the ImageJ 1.34S program (Wayne Rasband, http://rsb.info.nih.gov/ij/).


We gratefully acknowledge the expert advice and assistance of Pierre Travo (Platform RIO Imaging, Montpellier) for fluorescence microscopy, and of Olivier Coux (CRBM, CNRS UMR1086, Montpellier) for suggesting pertinent experiments about the proteasome. We also wish to thank Patrick Chaussepied (CRBM, CNRS UMR1086, Montpellier) and Sylvie Gisselbrecht (Institut Cochin, Paris) for stimulating discussions, and the groups of Etienne Schwob and Naomi Taylor at the Institut de Génétique Moléculaire (IGMM, CNRS UMR5535, Montpellier) for access to the FACScalibur equipment and to the FlowJo software respectively. We are also indebted to Eric Prina for the generous gift of anti-LACK antibody, and finally, grateful to Professor Jean-Pierre Dedet for continuous support and encouragement.

Supplementary material

The following supplementary material is available for this article online:

Fig. S1. Structural data for protein LmjKIN13-1.

This material is available as part of the online article from http://www.blackwell-synergy.com