ABSTRACT
Polyclonal (PAT) and monoclonal (AXO 49) antibodies against Paramecium axonemal tubulin were used as probes to reveal tubulin heterogeneity. The location, the nature and the subcellular distribution of the epitopes recognized by these antibodies were, respectively, determined by means of: (i) immunoblotting on peptide maps of Paramecium, sea urchin and quail axonemal tubulins; (ii) immunoblotting on ciliate tubulin fusion peptides generated in E. coli to discriminate antibodies directed against sequential epitopes (reactive) from post-translational ones (non reactive); and (iii) immunofluorescence on Paramecium cells, using throughout an array of antibodies directed against tubulin sequences and post-translational modifications as references.
AXO 49 monoclonal antibody and PAT serum were both shown to recognize epitopes located near the carboxylterminal end of both subunits of Paramecium axonemal tubulin, whereas the latter recognized additional epitopes in α-tubulin; AXO 49 and a fraction of the PAT serum proved to be unreactive over fusion proteins; both PAT and AXO 49 labelled a restricted population of very stable microtubules in Paramecium, consisting of axonemal and cortical ones, and their reactivity was sequentially detected following microtubule assembly; finally, both antibodies stained two upward spread bands in Paramecium axonemal tubulin separated by SDS-PAGE, indicating the recognition of various α- and β-tubulin isoforms displaying different apparent molecular masses.
These data, taken as a whole, definitely establish that PAT and AXO 49 recognize a post-translational modification occurring in axonemal microtubules of protozoa as of metazoa. This modification appears to be distinct from the previously known ones, and all the presently available evidence indicates that it corresponds to the very recently discovered polyglycylation of Paramecium axonemal α- and β-tubulin.
INTRODUCTION
Structural and functional diversity is commonly observed among the microtubular networks in eukaryotic cells. This diversity can be generated from the structure and localization of the microtubule-organizing centers (MTOCs), through interactions between microtubules (MTs) and microtubule-associated proteins (MAPs), and from the heterogeneity of MTs at a molecular level. At the cellular level, different isotypes of α- and β-tubulins are found which are encoded by multigenic families and the number of tubulin genes increases from unicellular to multicellular eukaryotes (Raff, 1984; Cleveland and Sullivan, 1985; Little and Seehaus, 1988; Sullivan, 1988; Silflow, 1991; Gaertig et al., 1993; Luduena, 1993). However, the great variety of isoforms is generated by combinations of numerous post-translational modifications (PTMs) occurring on relatively few primary gene products (McKeithan et al., 1983; Suprenant et al., 1985; Denoulet et al., 1988; Eddé et al., 1991; Fouquet et al., 1994). As concerns α-tubulin, acetylation of Lys 40 (L’Hernault and Rosenbaum, 1985; LeDizet and Piperno, 1987; Eddé et al., 1991), polyglutamylation of Glu 445 (Eddé et al., 1990), and at the C-terminal end, detyrosination (Barra et al., 1974; Thompson, 1982) and removal of a glutamyl-tyrosine dipeptide (Paturle-Lafanechère et al., 1991) have been reported. As for β-tubulin, phosphorylation of Tyr 437 and Ser 444 in the βIII isotype and of Ser 441 in the βVI isotype (Eipper, 1974; Luduena et al., 1988; Diaz-Nido et al., 1990; Alexander et al., 1991; Rüdiger and Weber, 1993), and glutamylation of glutamate residues (Glu 434, 435, 438 or 441) differing according to the isotype (Alexander et al., 1991; Redeker et al., 1992; Rüdiger et al., 1992; Mary et al., 1994) have been described. Among these modifications, all demonstrated in metazoa, only acetylation has been thoroughly characterized in a protist.
The study of the relation between MT diversity and tubulin heterogeneity has been approached here through the use of immunological tools in a ciliate protist, Paramecium. Such a model combines a higher MT diversity than a single metazoan cell with a more reduced genetic diversity. Thus, the 13 microtubular arrays displaying different stability and immunological properties identified in this unicellular organism (Cohen and Beisson, 1988; Torres and Delgado, 1989; Adoutte et al., 1991; Fleury et al., 1995) appear to be built from few types of tubulin molecules: all the gene sequence data available indicate that the resulting proteins are extremely similar (Dupuis, 1992a,b). Therefore, in this protist, tubulin heterogeneity is expected to be mostly, if not exclusively, generated through PTMs, and the present study illustrates this statement.
Previous work had shown that a serum raised against Paramecium axonemal tubulin (1ater named anti-PA tubulin antibody then PAT) reacts specifically with axonemal and cortical MTs of Paramecium (Cohen et al., 1982) and other protists (Brugerolle and Adoutte, 1988), and also with axonemal tubulins from a broad range of species, extending from protists to mammals, whereas it is unreactive with cytoplasmic tubulins of metazoa (Adoutte et al., 1985). Taking into account the fact that most stable, long-lived MTs are posttranslationally modified (Sullivan, 1988; Greer and Rosenbaum, 1989, for reviews), the PAT immunoreactivity restricted to a subset of stable MTs in Paramecium and its late occurrence in the course of cell division provided the first arguments permetting to hypothesize that the serum would recognize a post-translational modification taking place on previously assembled MTs (Adoutte et al., 1991).
In order to identify the major epitope recognized by the PAT serum, we recently produced monoclonal antibodies (mAbs), using Paramecium axonemes as antigen, and one of them, AXO 49, appeared to exhibit immunocytochemical properties similar to those of the serum (Callen et al., 1994).
Therefore we have undertaken an exhaustive comparison of the location, the nature and the subcellular distribution of the epitopes recognized by both polyclonal and monoclonal antibodies, by means of: (i) immunoblotting on peptide maps of Paramecium axonemal tubulin; (ii) immunoblotting over fusion proteins expressed in E. coli, a procedure which can discriminate antibodies that are directed against sequential epitopes (reactive ones), from those directed against post-translational epitopes (non reactive); or (iii) single and double labelling immunofluorescence of Paramecium cells, the latter method permitting to compare kinetics of appearance of the respective epitopes in newly assembled MTs. AXO 49 and PAT turned out to be both directed to a post-translational modification of tubulin, located in the extreme C-terminal part of both subunits and distinct from previously known ones detected by reference mAbs. These data have led us to search for a new modification of Paramecium axonemal tubulin. In a separate work, C-terminal peptides have been subjected to physico-chemical analysis and have been shown to carry a new post-translational modification, consisting of the addition of glycyl units to the γ carboxyl groups of Glu 445 and Glu 437 of the α and β subunits, respectively, and termed polyglycylation (Redeker et al., 1994). The characteristics of the PTM recognized by both antibodies described here fit very well with those of the polyglycylation. In addition, they provide the first clues suggesting the occurrence of different forms of polyglycylation in a cell. Finally, we have shown that this previously undescribed PTM occurs in axonemal tubulins from very distant species in the evolutionary scale, ranging from Paramecium to sea urchin and quail.
MATERIALS AND METHODS
Cells
Paramecium tetraurelia cells (strain d4-2) were grown as reported by Callen et al. (1994).
Protein preparations
Paramecium axonemes and axonemal tubulin were prepared as previously described (Geuens et al., 1989).
Quail oviduct ciliary axonemes were prepared by M. C. Marty (CNRS, Ivry-sur-Seine, France), as described by Adoutte et al. (1991). Sea urchin (Paracentrotus lividus) spermatozoan flagellar axonemes (provided by M. P. Cosson, CNRS, Villefranche-sur-Mer,
France) were prepared according to the method of Gibbons and Fronk (1972).
Fusion proteins were prepared as detailed by Callen et al. (1994). Briefly, the plasmids encoding β-galactosidase/α-tubulin fusion proteins were constructed as follows. DNA of one α-tubulin gene from Stylonychia lemnae (Helftenbein and Müller, 1988) cloned in pBR322 (a kind gift from F. Caron, Ecole Normale Supérieure, Paris) was digested with various restriction endonucleases. The gene fragments were then inserted in the pEX vectors (Boehringer, Mannheim, Germany) downstream from the lacZ gene under control of a λ promoter. The expressed inserts correspond to the 5 following domains of α-tubulin lying at positions 22-176, 214-445, 274-338,
339-445 and 412-445 (Fig. 4). The protein domain 177-213 cannot be expressed because of the presence of an universal STOP codon UAA encoding glutamine (position 177) in the α-tubulin gene of Stylonychia, and of the absence of a suitable restriction endonuclease site upstream of position 214.
Anti-tubulin antibodies
Reference antibodies Polyclonals
Affinity-purified C85, C105 and C140 monospecific antibodies (Arévalo et al., 1990) were generously provided by Dr J. M. Andreu (Centro de Investigaciones Biologicas, Madrid, Spain).
Monoclonals
DM1A and DM1B mAbs (Blose et al., 1984) were purchased from Amersham (Les Ulis, France); 6-11B-1 cell culture supernatant (Piperno and Fuller, 1985) was a generous gift from Dr G. Piperno (Mount Sinai Medical School, New York, USA) whereas ascites fluid was purchased from Sigma (Saint-Quentin-Favallier, France); TU-01 (Viklicky et al., 1982) was kindly provided by Dr V. Viklicky (Czechoslovak Academy of Sciences, Prague, Czech Republic) and GT335 (Wolff et al., 1992) was generously given by Dr A. Wolff (Collège de France, Paris).
Preparation of anti-axonemal tubulin antibodies
The serum (4th sample taken in the course of rabbit immunization) raised against Paramecium axonemal tubulin in acrylamide (Cohen et al., 1982) was employed crude (PAT) or affinity-purified using Paramecium axonemal tubulin, as previously described (Geuens et al., 1989). Two peaks, PAT 1 and PAT 2, were, respectively, eluted at pH 2.8 and 2.2.
AXO 49 mAb was prepared as described by Callen et al. (1994), and used as cell culture supernatant or (when mentioned) affinitypurified by means of Affi-Prep protein A MAPS II kit from Bio-Rad (Ivry-sur-Seine, France).
Limited proteolysis
Proteolysis of soluble Paramecium axonemal tubulin, 0.5-4 mg/ml in MES buffer at pH 6.8 (50 mM MES, 1 mM EGTA, 1 mM MgCl2, 1 mM GTP) was performed by incubation for 30 minutes at 30°C with the following proteases at the indicated weight ratios: trypsin, 1.2%; pronase, 0.2%; subtilisin B (Novo, Copenhagen, Denmark), 2%. The reaction was stopped by addition of 2 mM PMSF.
Gel electrophoresis, immunoblotting and dot-blotting
SDS-PAGE was conducted according to the method of Laemmli (1970), using 0.75 mm thick, 12 cm long, 8% polyacrylamide gels or (when mentioned) 10% polyacrylamide mini-gels at pH 8.3, with 0.1% (w/v) SDS (99% pure) from BDH Laboratory Supplies (Poole, England). Gels were stained with Coomassie Blue or electro-transferred according to Kyhse-Andersen (1984), and membranes were processed as described previously (Callen et al., 1994). Dot-blot analysis of peptides was performed after covalent binding to Immobilon-AV affinity membrane (Millipore, Saint-Quentin en Yvelines, France), according to Canas et al. (1993). Peroxidaselabelled secondary antibodies (sheep anti-mouse or goat anti-rabbit IgGs from Diagnostics Pasteur, Marnes-la-Coquette, France) were used at 1:2,000 and peroxidase was detected using enhanced chemiluminescence (ECL, Amersham).
Peptide mapping
For in-gel digestion, α- and β-tubulin bands were separated then cut out from 0.75 mm thick, 12 cm long, 7.5-15% or 8% SDS-polyacrylamide gels performed either at pH 8.25 to get an optimal separation of Paramecium tubulin subunits, or at pH 9.25 to separate sea urchin and quail tubulin subunits. Under these respective conditions, Paramecium α-tubulin migrates faster than β-tubulin, whereas the relative migrations of both subunits of metazoan tubulins are reversed (Suprenant et al., 1985). Then the Paramecium and metazoan tubulin bands were loaded onto 1 mm thick, 11 cm long SDS-polyacrylamide gels made of 15-20% acrylamide gradients at pH 8.25 and 9.25, respectively.
The bands were submitted to proteolysis either by formic acid, according to Sonderegger et al. (1982), or by Staphylococcus aureus V8 protease (type XVII-B, Sigma), after Cleveland et al. (1977). The latter procedure is detailed by Callen et al. (1994). Briefly, gels were either silver-stained according to the method of Merril et al. (1981, 1983), or electro-transferred according to Towbin et al. (1979). After overnight incubation with primary antibodies, the filters were successively incubated with biotinylated secondary antibodies (goat antimouse or donkey anti-rabbit IgG at 1:1,600), then with streptavidinbiotinylated horseradish peroxidase complex at 1:1,600 (Amersham), and detection was carried out using ECL. In one case, detection was performed using peroxidase or alkaline phosphatase-labelled secondary antibodies and DAB or NBT/BCIP, respectively, as previously described (Adoutte et al., 1991).
Immunofluorescence microscopy
The immunocytochemical procedure is reported by Callen et al. (1994). Fluorescein-labelled sheep anti-mouse and goat anti-rabbit IgGs at 1:200 (Diagnostics Pasteur) and Texas Red-conjugated goat anti-rabbit IgG at 1:50 (Jackson Immunoresearch Laboratories, West Grove, PA, USA) were used as secondary antibodies. Cells were examined on a Bio-Rad Lasersharp MRC 600 confocal microscope with an argon laser for single staining and a two-laser equipment (argon and helium-neon) for double staining.
RESULTS
AXO 49 mAb and PAT serum recognize epitopes located in a broad C-terminal domain of Paramecium axonemal α- and β-tubulin
Following in-gel digestion of Paramecium axonemal α- and β-tubulins by V8 protease (see below, Fig. 7, lanes a-c, and Fig. 5 in Callen et al., 1994), immunoblotting was achieved with AXO 49 mAb and PAT serum (Fig. 1). Both antibodies react more strongly at the level of the β subunit than of the α one (1anes e,f,m,n). After digestion, AXO 49 reacts only with one β peptide, of 21 kDa (1anes c,d) and one α peptide, of 19 kDa (1anes g,h). PAT stains the same peptides as AXO 49 (1anes k,l,o,p), but relatively more strongly the α ones (1anes g,h,o,p), suggesting the recognition of several epitopes in this domain. These α and β peptides are also, respectively, labelled by DM1A and DM1B (Fig. 1, lanes r,s, and Fig. 7, lanes e,f and h,i) which react with C-terminal (Ct) epitopes of α- and β-tubulin, respectively (Breitling and Little, 1986; and Fig. 4). Therefore the α 19 kDa and β 21 kDa peptides are situated in the C-terminal domain of each subunit. PAT additionally stains 34 and 30 kDa peptides arising from digestion of α-tubulin (Fig. 1, lanes o-r) also labelled by 6-11B-1 (1ane t), directed to acetylated Lys 40 of α-tubulin (LeDizet and Piperno, 1987; and Fig. 4); these peptides are therefore comprised in the N-terminal (Nt) domain of α-tubulin.
Paramecium axonemal tubulin was also cleaved in solution by pronase and formic acid which generated α and β 36 and
18 kDa peptides, whereas α-tubulin only was digested by trypsin (not shown). PAT mostly labelled the smaller α and β Ct domains, stained by DM1A and DM1B, respectively, and less strongly the larger α Nt ones labelled by 6-11B-1 (not shown).
The proteolysis patterns of Paramecium tubulin are in agreement with those obtained with brain tubulin by Brown and Erickson (1983), Mandelkow et al. (1985), Serrano et al. (1986) and de la Vina et al. (1988). The peptide locations deduced here from antibody immunoreactivity fit with those determined in brain tubulin by sequencing (Mandelkow et al., 1985). The formic acid main cleavage sites were located at peptide bonds Asp 306-Pro 307 in both subunits of porcine brain tubulin (Serrano et al., 1984b); from the conservation of these residues in tubulin from brain (Krauhs et al., 1981) with respect to Paramecium (Dupuis, 1992a,b), identical formic acid cleavage sites can be inferred in Paramecium tubulin, yielding large Nt and smaller Ct fragments, as actually found.
In conclusion, AXO 49 and PAT recognize epitopes located in a broad C-terminal domain of each subunit (covering 1/3 of the molecules) of Paramecium axonemal tubulin, whereas PAT recognizes (an) additional epitope(s) lying in a larger N-terminal domain (2/3 of the molecule) of the α subunit (see Fig. 4).
AXO 49 and PAT react with extreme C-terminal epitopes of Paramecium axonemal α- and β-tubulin
AXO 49 and PAT stain two upward spread bands in Paramecium axonemal tubulin at the level of both subunits, but the staining of α-tubulin by AXO 49 is slightly shifted upwards relative to that provided by DM1A (Fig. 2, lanes a-d).
Limited proteolysis of axonemal tubulin with subtilisin generates two polypeptides, αs and βs, displaying electrophoretic mobilities greater than those of the α and β subunits (CB, lanes e,f), indicating the removal of small fragments, as for brain tubulin (Serrano et al., 1984a). The spread staining of the whole subunits with Coomassie Blue (1ane e) as well as with DM1A and DM1B mAbs (1anes a,c) contrasts with the distinct staining of the αs and βs polypeptides (1anes f,g,i); the lack of labelling at the level of the uncleaved subunits shows that the proteolysis was complete. AXO 49 is completely unreactive with the αs and βs polypeptides (1ane h) while still reactive with the whole set of digestion products (see inset). Consequently, the epitopes recognized by AXO 49 were not merely destroyed by cleavage and are born by the small peptides separated after electrophoresis. PAT fairly reacts with αs but hardly with βs (1ane j), indicating the presence of (an) additional reactive epitope(s) born by the αs (and βs) polypeptide(s).
In order to locate the sites of subtilisin cleavage in Paramecium tubulin and the epitopes recognized by AXO 49 mAb and PAT serum, V8 protease peptide maps of the whole and subtilisin-digested subunits, αs and βs, were compared. C85 and C105 antibodies, which are directed to epitopes located near the C-terminal end of α- and β-tubulin, respectively (Arévalo et al., 1990; and Fig. 4), label αs 18 and βs 19 kDa peptides, downward shifted relative to the α 19 and β 21 kDa ones (Fig. 3, lanes b,e,f,i). As expected, AXO 49 is unreactive towards the former peptides (1anes d,g) while strongly reactive with the latter ones (1anes c,h). Unlike AXO 49, PAT still reacts moderately with the αs 18 kDa peptide (1anes c,d,k,l) but hardly with the βs 19 kDa one (1anes g,h,n,o). In contrast, neither the migration rates nor the reactivities of the α 34 and 30 kDa peptides with 6-11B-1 and PAT are affected by subtilisin digestion (1anes j-m). Remarkably, one purified fraction from the serum, PAT 1, displays a very weak reactivity with the αs 18 and βs 19 kDa peptides (1anes p-s), suggesting that it is enriched in an antibody similar to the AXO 49 mAb. In contrast, the reactivity pattern of the second fraction, PAT 2 (1anes t-w), is very similar to that of the whole serum.
Thus, double proteolysis, yielding unmodified Nt peptides and smaller Ct ones than with V8 protease only, demonstrates that subtilisin cleaves, in Paramecium, as in brain tubulin (Serrano et al., 1984a,b), both subunits near their C-terminal ends. The loss of AXO 49 mAb reactivity towards subtilisincleaved Paramecium axonemal tubulin, and the conservation of the epitopes recognized by DM1A and DM1B (Fig. 2) and C85 and C105 antibodies (Fig. 3) permitted us to localize: (i) the subtilisin cleavage sites approximately downstream of the most rightward of these epitopes, i.e. probably on the Ct side of residues 443 and 431 of the α and β subunits, respectively, namely 6 to 11 residues from their Ct extremities; (ii) the epitopes recognized by AXO 49 downstream of the subtilisin cleavage sites, therefore in the extreme C-terminal part of both subunits, probably in the respective sequence segments 444-449 and 432-442 of α- and β-tubulin. In addition, the important decrease of PAT serum reactivity in single (Fig. 2) and double (Fig. 3) proteolysis experiments shows that the latter recognizes several epitopes: (i) mainly the same ones as those reacting with AXO 49 mAb, located near the C-terminal end of both subunits; (ii) epitopes lying upstream of the previous ones, i.e. in broad C-terminal domains of both subunits delimited at their Nt sides, by the position of the αs 18 and βs 19 kDa peptides, starting at residues 267 and 257, respectively, and at their Ct ends, by the subtilisin cleavage sites; (iii) an epitope located in a broad N-terminal region of the α subunit (1-213), delimited by the smallest reactive Nt peptide, of 21 kDa (Fig. 1, lane q). The positions of the epitopes recognized by AXO 49 and PAT are represented in Fig. 4.
The locations of the Ct epitopes rely upon determinations of the subtilisin cleavage sites, inferred from reactivities of sitedirected antibodies. For α-tubulin, the epitope was restricted to a small segment, thanks to the availability of an antibody directed to a very Ct sequential epitope (C85). However, the accuracy of such determinations is limited by the lack of information on the minimal sequences compatible with immunoreactivity. In fact, two conflicting views have been reported for the locations of the subtilisin cleavage sites in mammalian brain tubulin, probably partly due to different conditions used. On the one hand, from SDS-PAGE (Serrano et al., 1984a) and amino acid analysis (Maccioni et al., 1986), these sites were previously located at positions lying from 30 to 40 residues from the Ct end. On the other hand, native gel electrophoresis (Sackett et al., 1985) and reactivity of site-directed antibodies (de la Vina et al., 1988; Paschal et al., 1989; Lobert et al., 1993) indicated the occurrence of one or several cleavage sites between residue 430 and the end of each subunit. Our results on Paramecium tubulin are consistent with the second set of data as well as with the direct determination of the major cleavage sites in pig brain α- and β-tubulin, by sequencing, at positions Asp 438-Ser 439 and Gln 433-Gly 434, respectively (Redeker et al., 1992).
AXO 49 mAb and a fraction of PAT serum are unreactive with bacterially expressed tubulin
Fusion peptides comprising various segments of ciliate (Stylonychia) α-tubulin were expressed in E. coli and permitted to cover the α-tubulin sequence of ciliates except for the 21 Nt amino acids and the central segment 177-213 (Fig. 4). Their relative migrations after SDS-PAGE are shown in Fig. 5, beside Paramecium axonemes (CB, lanes a-f).
After immunoblotting with TU-01 mAb, directed towards an N-terminal sequential epitope of α-tubulin (Grimm et al., 1987), only the Nt fusion peptide, N, is reactive, as expected, similarly as whole α-tubulin from Paramecium axonemes (TU-01, lanes a,b,d). With DM1A, all of the Ct fusion peptides embracing the sequence 426-430 are reactive, similarly as α-tubulin (DM1A, lanes a-f).
In contrast to the anti-sequence antibodies, AXO 49 is fully unreactive with all fusion peptides (AXO 49, lanes a,b,d). The lack of reactivity of the fusion peptide C, covering the sequence 214-445, potentially bearing the epitope, as inferred from peptide mapping of axonemal tubulin, was ascertained by probing it at an amount 30-fold greater than that shown here.
As concerns the whole serum, it is clearly reactive with all of the α Ct fusion peptides comprising the sequence 412-445, but neither reacts with the Nt peptide, N, nor with an internal peptide, C3, covering the sequence 274-338 (Figs 4 and 5, PAT, lanes a-f). PAT 2 fraction from the serum, which exhibits the same reactivity as the latter on subtilisin-V8 protease peptide maps, is similarly reactive with the α Ct fusion peptides (not shown). In contrast, PAT 1, enriched in AXO 49-like antibody (see Fig. 3), stains very weakly the α Ct fusion peptides (Fig. 5, PAT 1, lanes a-f) in comparison with Paramecium axonemes.
Therefore, the lack of reactivity of AXO 49 mAb and of a fraction of PAT serum with ciliate α-tubulin fusion peptides reveals that they very likely are both directed to post-translational epitopes present in Paramecium axonemal tubulin (near the C-terminal end of both subunits). In contrast, the reactivity of the whole serum with fusion peptides indicates that it additionally contains (at least) one antibody directed to the α-tubulin sequence 412-445, which is thus further narrowed with respect to the broad localization (267-443) inferred from V8 protease and subtilisin cleavage sites (Fig. 4). In order to account for the weak residual labelling of the Ct βs peptide, the same antibody could be involved in the recognition of an homologous β-tubulin sequence; alternatively, this may be due to a β sequence-specific antibody. Finally, another antibody directed to an N-terminal region (1-213) of the α subunit has to be considered, but the non reactivity of the serum with the Nt fusion peptide, N, limited to the sequence portion 22-176 (Fig. 4), does not permit us to infer the nature of the epitope involved.
The target of AXO 49 is different from the polyglutamylation recognized by GT 335 mAb
As polyglutamylation was the most important PTM of tubulin yet found in both subunits, it was crucial to determine whether it corresponded to the epitopes investigated here. Therefore the reactivity of GT335 mAb, directed against tubulin glutamylation (Wolff et al., 1992), was compared to that of AXO 49 on immunoblots of whole or proteolyzed axonemal tubulin.
When large amounts of Paramecium axonemes are separated after SDS-PAGE, an upward spread of staining is visible above the major α- and β-tubulin species, and a third minor diffuse band, βh, can be resolved above the level of the β band (Fig. 6CB, lane a).
Immunoblotting of Paramecium axonemes with GT335 yields a remarkable staining beginning slightly above the level of β-tubulin located by means of C140 antibody, directed to the β-tubulin sequence 153-165 (Arévalo et al., 1990); this staining is either spreading upward or appearing as a doublet (as observed by Bré et al., 1994) the slower component of which migrates at the level of βh (1anes a-c). The latter species likely represents (an) isoform(s) of β-tubulin, given that C140 can yield a spread staining extending up to its level, provided that the antigen and antibody concentrations are adequately adjusted. At the level of the α subunit, located with DM1A, no staining is detected with GT335 (1anes c,e). In contrast, with AXO 49, a labelling of both subunits is conspicuous, and again slightly upwards shifted with respect to that yielded by the antisequence antibodies (1anes b,d,e). The staining of the α and β subunits is also spread upward and extends to the level of the β and βh bands, respectively.
V8 protease peptide mapping shows that the proteolysis pattern of the βh species is very different from that of α-tubulin (Fig. 7, lanes b,d), but resembles that of β-tubulin, except for a βh 22 kDa peptide which exhibits a higher apparent molecular mass than the corresponding β 21 kDa peptide (1anes c,d). This confirms that the βh species represents (a) modified isoform(s) of β-tubulin.
Comparative immunoblotting of peptide maps of Paramecium axonemal α-, β- and βh-tubulins with AXO 49 and GT335 shows that: (i) in contrast to GT335, which is unreactive, AXO 49 strongly labels the α Ct 19 kDa peptide; moreover, this staining is shifted upward with respect to that provided by DM1A (1anes e,f,g). (ii) The β Ct 21 kDa peptide is reactive with GT335 as well as with AXO 49, and the staining yielded by both mAbs is slightly upward shifted with respect to that provided by DM1B (1anes h-j). (iii) Remarkably, the βh 22 kDa peptide, which is neither stained by DM1A (not shown) nor by DM1B, is nevertheless fairly reactive with AXO 49 and especially with GT335 (1anes k-m), indicating that it is C-terminal. Taking into account the relative amounts of the α, β and βh Ct peptides (1anes b-d), it can be inferred that AXO 49 reactivity is actually stronger at the level of the β and βh species than that of the α one, whereas GT335 reactivity is the highest with the slower migrating isoform(s) of tubulin, βh.
AXO 49 and PAT label a subpopulation of stable microtubular arrays in Paramecium
The labelling of Paramecium cells by AXO 49 and PAT displays very characteristic features which are completely different from those provided by other antibodies, either directed against sequential or post-translational epitopes of tubulin.
These antibodies are the only ones which strongly label the ciliary axonemes along their length, except their tip (Fig. 8A,B,D,E). In contrast, the anti-sequence antibodies, such as DM1A, strongly label only the tip of the cilia (along ∼2 μm length), similar to 6-11B-1 (Fig. 8C,F; Adoutte et al., 1991) and other anti-sequence and anti-acetylated α-tubulin mAbs (Callen et al., 1994) while GT335 labels both the axoneme proximal part and the tip (Bré et al., 1994). Moreover, all cilia are not similarly labelled with AXO 49 and PAT. Those from the anterior left area of the cell are more brightly stained with AXO 49, and less intensely with PAT, whereas the reverse is true for the cilia located in the other cortical areas (Fig. 8A,D). Strikingly, the largest part of the anterior left cell area, where no basal body duplication occurs during division, has specific invariant morphogenetic properties (Iftode et al., 1989).
Contrary to DM1A which decorates all microtubular networks in Paramecium, and in particular the intracytoplasmic labile one, AXO 49 and PAT do not label cytoplasmic MTs (Fig. 8B,C,E), in agreement with blotting data (M. H. Bré, personal communication). Similarly, 6-11B-1 does not label the cytoplasmic MTs (Fig. 8F; Adoutte et al., 1991) and very weakly labels cytoplasmic tubulin on immunoblots (Callen et al., 1994). In contrast, a fraction of cytoplasmic MTs is recognized by GT335 (Bré et al., 1994).
The labelling of Paramecium cells with AXO 49 and PAT is limited to very stable MTs, associated with the cortex and probably the membranes, and usually found as bundles. In addition to axonemes, it comprises particularly the contractile vacuole system and the postoral fiber; this population of stable MTs is more restricted than that recognized by 6-11B-1 (Fig. 8B,E,F; Adoutte et al., 1991) or by GT335 (Bré et al., 1994). In the course of cell division, the reactivity of the new microtubular structures with PAT is detected only after a certain lag time following their assembly revealed by their reactivity with DM1A or DM1B (Adoutte et al., 1991). Two differences between AXO 49 and PAT are clearly visualized by double labelling experiments (Fig. 8G,H). On the one hand, the cytospindle, consisting of a transient set of cortical MTs assembling beneath the cortex during division, is unreactive with AXO 49 while labelled with PAT. On the other hand, the new microtubular structures, such as the contractile vacuole system, become reactive with AXO 49 still later than with PAT.
The epitopes detected in Paramecium with AXO 49 and PAT are present in axonemal tubulin of metazoa
As PAT serum reacts with axonemal tubulins of metazoa (Adoutte et al., 1985), we tested AXO 49 mAb in comparison with the serum on peptide maps of axonemal tubulins from an invertebrate and from a vertebrate, i.e. those from sea urchin spermatozoan flagella and from quail oviduct cilia.
The V8 protease cleavage patterns of axonemal β-tubulins from sea urchin (Fig. 9A, lanes a-e), quail (1ane l; Adoutte et al., 1991) and Paramecium (Fig. 7, lane c; Callen et al., 1994) display a distribution of numerous peptides of various mobilities, whereas the corresponding α-tubulins yield, after proteolysis, a few groups of major peptides (Fig. 9A, lanes f-j,o; Adoutte et al., 1991; Callen et al., 1994). Such differences had been previously observed between cleavage patterns of β- and α-tubulins from protists to metazoa (Little et al., 1981; Adoutte et al., 1985). Both subunits of sea urchin flagellar tubulin are reactive with AXO 49, whereas only α-tubulin of quail cilia is clearly reactive (Fig. 9B, lanes e,f,m,n).
After proteolysis, AXO 49 strongly reacts with α-tubulin 21 kDa peptides from both species which are the major ones present in the corresponding peptide maps (Fig. 9A,B, lanes g,o), as in the case of Paramecium (Fig. 9A, lane j and Fig. 1, lane g). AXO 49 also weakly reacts with minor α-tubulin 35 kDa peptides from both metazoan species, but not from Paramecium (Fig. 9B, lanes g,o and Fig. 10, lanes d-f). Both 21 and 35 kDa peptides, labelled by DM1A (Fig. 10, lanes d,j), are C-terminal. PAT serum, like AXO 49 mAb, labels strongly the major Ct 21 kDa peptides from metazoan and Paramecium axonemal tubulins, and weakly the minor Ct 35 kDa peptides from metazoan tubulin (Fig. 10, lanes d-i). In addition, PAT labels supplementary peptides. These are 6-11 B-1 reactive metazoan tubulin 37, 35 and 18 kDa Nt peptides, and Paramecium tubulin 32-35 kDa peptides (1anes g-i, k-m) as is also a 21 kDa one, observed at high protease concentration (Fig. 1, lane q).
On peptide maps of axonemal β-tubulins from sea urchin and quail, 36 and 22 kDa C-terminal peptides are very weakly stained by AXO 49, PAT and DM1B (Fig. 9B, lanes d,l and Fig. 10, lanes a-c). The very weak staining of the metazoan β-tubulin 22 kDa peptides contrasts with the strong labelling of the corresponding α 21 kDa peptides, as well as of the Paramecium β-tubulin 21 kDa peptide (Fig. 1, lanes d,l). This cannot be accounted for solely on the basis of differences in amounts of the corresponding peptides (see below).
In conclusion: (1) in spite of an overall resemblance in the metazoan and protozoan axonemal α-tubulins, mainly yielding Nt 35 kDa and Ct 21 kDa complementary peptides, additional Ct 35 kDa and Nt 18 kDa complementary fragments were detected in the metazoan tubulins only (Fig. 9A, lanes g-j,o, and Fig. 10, lanes j-m). These observations are in agreement with those of Little et al. (1984a) showing differences between the proteolysis patterns of axonemal α-tubulins from metazoa and protozoa, contrasting with the resemblance of the β-tubulin patterns. (2) The similar reactivity of AXO 49 mAb and PAT serum with axonemal α-tubulin Ct 21 kDa peptides from metazoa and Paramecium permits us to: (i) extend to metazoa the presence of the post-translational epitopes detected in Paramecium, and (ii) narrow their localization near the α-tubulin C-terminal end in these metazoa. As for the β subunit, the very weak reactivity of the metazoan Ct peptides with AXO 49 and PAT contrasts with the strong reactivity of the Paramecium homologue. Therefore, in these metazoan species, either the β-tubulin post-translational epitopes are less abundant than in Paramecium, or they differ in such a way that their affinity for both antibodies is weaker. (3) The secondary reactivity of the PAT serum with metazoan α-tubulin Nt 18 kDa peptides whose sequence position, 1-203 (determined as by Callen et al., 1994), overlaps with the N-terminal epitope location (1-213) found in Paramecium α-tubulin, suggests a conservation of the latter epitope in metazoan α-tubulins.
DISCUSSION
AXO 49 mAb and PAT serum reveal a posttranslational modification of tubulin in axonemes and other stable microtubular arrays
In ciliated and flagellated cells of both protists and metazoa, a range of data involving cleavage pattern similarities (Little et al., 1982; Russell et al., 1984), reflagellation studies (McKeithan et al., 1983; Russell and Gull, 1984), tubulin gene homogeneity and tubulin isotype multifunctionality (Little et al., 1984b; Raff, 1984; Joshi and Cleveland, 1990; Silflow, 1991; Gaertig et al., 1993; Luduena, 1993; Renthal et al., 1993) have provided evidences that axonemal tubulin is drawn from the same pool as the cytoplasmic one and is then post-translationally modified.
In this context, the reactivity of axonemal MTs of Paramecium and metazoa with AXO 49 mAb and PAT serum, combined with the lack of reactivity of their respective cytoplasmic counterparts with AXO 49 (this report) and with PAT (Adoutte et al., 1985, 1991), strongly suggest that both antibodies recognize a post-translational modification of axonemal MTs occurring on the same tubulin species as those involved in the cytoplasmic pool. The alternative hypothesis, which could account for the specificity of our antibodies, namely the recognition of axonemal-specific isotypes, does not hold by reason of the lack of homologous sequences in the C-terminal part of α- and β-tubulins from Paramecium (Dupuis, 1992a,b) and from metazoan testis-specific isotypes (Little and Seehaus, 1988) which could distinguish them from other isotypes (Fig. 11).
The non reactivity of AXO 49 mAb and of a fraction of PAT serum over ciliate α-tubulin expressed in E. coli as fusion protein, conjugated with their reactivity with Paramecium axonemal α- and β-tubulin, provide decisive evidence permitting us to affirm that both antibodies are directed against a posttranslational modification present in both subunits of axonemal tubulin, not occurring in bacteria. Effectively, we have previously shown that, given the great conservation of ciliate tubulin, antibodies directed against tubulin sequential epitopes (including conformational determinants) were indeed reactive with bacterially expressed ciliate tubulin whereas, conversely, antibodies directed against PTMs of tubulin proved to be totally unreactive (Callen et al., 1994), even with large amounts of fusion protein like those probed with AXO 49.
The apparent molecular mass heterogeneity of Paramecium axonemal α- and β-tubulin revealed by SDS-PAGE and immunoblotting and its decrease after removal of the Ct end of both subunits concomitant with loss of AXO 49 reactivity (Fig. 2) suggest that different α- and β-tubulin isoforms are generated by successive PTMs located in their C-terminal end which would each cause slight upward shifts in electrophoretic mobility. This assertion is supported by the following observations: (i) an increase in the apparent molecular masses of numerous isoforms of axonemal tubulins as a function of the glutamylation degree (Bré et al., 1994; Fouquet et al., 1994); (ii) a decrease in brain tubulin isoform amount after subtilisin cleavage (Redeker et al., 1992). Thus the upward shifts of AXO 49 and GT335 reactivities and the lack of shift of PAT reactivity with respect to the staining yielded by anti-sequence antibodies indicate that AXO 49 and GT335 detect only modified isoforms, whereas PAT also reacts with unmodified axonemal tubulin, certainly through its anti-α-tubulin sequence antibodies.
Although acetylation slows down tubulin migration (Callen et al., 1994), this upward shifting of immunoreactive isoforms with respect to the bulk of tubulin cannot be caused by acetylation since: (i) it is also observed at the level of Ct α 19 kDa and β 21 kDa peptides whereas acetylation is located in the Nt domain of α-tubulin (Figs 3, 7); (ii) axonemal tubulin is mostly acetylated so that its labelling with 6-11B-1 and DM1A is observed at the same level (Callen et al., 1994). The apparent molecular mass heterogeneity cannot be accounted for by acetylation either, since there is probably a single acetylation site in Paramecium α-tubulin Nt domain, as inferred from brain tubulin (Eddé et al., 1991; Callen et al., 1994).
Remarkably, the high apparent molecular mass isoforms are conspicuously stained by antibodies directed against Nt sequential and post-translational epitopes of α-(Callen et al., 1994) and β-tubulin (such as C140), but are less well or even not labelled by DM1A and DM1B (Fig. 7, lane l); this may reflect either a differential affinity of the antibodies for these minor isoforms or a reduction of accessibility of antibodies towards sequential Ct epitopes by PTMs located in their vicinity (see also Bré et al., 1994).
Therefore, in Paramecium axonemal tubulin, in addition to polyglutamylation detected next to and above the level of the β subunit, another C-terminal polymodification, recognized by AXO 49 and PAT, is responsible for the considerable apparent molecular mass heterogeneity and presumably for the reduction of accessibility of Ct sequential epitopes of the α (and β) subunit(s).
The similarly restricted reactivities of AXO 49 mAb and PAT serum with Paramecium microtubular networks and their late occurrence during cell division confirm that these antibodies are directed against the same post-translational modification taking place after MT assembly.
The presence of anti-sequence antibodies in the serum can neither account for the opposite relative reactivities of cilia from different parts of the cell with AXO 49 and PAT, nor for the lag time remaining between MT assembly and appearance of PAT reactivity. These observations as well as the delayed reactivity of AXO 49 with respect to PAT, and the non reactivity of transient structures with AXO 49 would be rationalized if AXO 49 recognized a later stage of a multi-step PTM than PAT.
AXO 49 and PAT detect a new post-translational modification of tubulin
Both AXO 49 and PAT decorate, in Paramecium, a narrower subpopulation of stable MTs than other antibodies directed to PTMs, namely acetylation and polyglutamylation. Both antibodies yield a unique labelling pattern of Paramecium ciliary axonemes, consisting of a strong decoration of the axoneme body except the tip, roughly complementary to those yielded by other anti-tubulin antibodies. This suggests that they are targeted against a bulky modification leading to a decreased accessibility of other antibodies to their epitopes along the axoneme. Although all tubulin PTMs occur preferentially on the MTs (Greer and Rosenbaum, 1989; Audebert et al., 1993; Paturle-Lafanechère et al., 1994), the modification revealed by AXO 49 and PAT is the only one which appears to be delayed with respect to MT assembly (Fleury et al., 1995). These unique immunocytochemical features point to a new type of tubulin modification, distinct from previously known ones.
(1) The detyrosination-tyrosination of tubulin, associated with the presence of a Gly-Glu-Glu-(Tyr) sequence, occurs at the C-terminal end of the α subunit in metazoa (Gu et al., 1988; Little and Seehaus, 1988; Rüdiger et al., 1994), and has only been detected in few cases in protists (Stieger et al., 1984; Birkett et al., 1985). In ciliates, it has not yet been found, in agreement with the available sequence data (Fig. 12). Moreover, in Paramecium, the anti-tyrosinated tubulin antibody YL 1/2 (Kilmartin et al., 1982; Wehland et al., 1984) is unreactive. In conclusion, the tubulin subunit specificity and the distribution of the detyrosination-tyrosination of tubulin (Barra et al., 1974; Gundersen and Bulinski, 1986), are completely different from those of the modification investigated here.
(2) The same comments can be made for the PTM involving the removal of a glutamyl-tyrosine dipeptide at the C-terminal end of the α-tubulin subunit (Paturle-Lafanechère et al., 1994).
(3) The subunit and/or species distribution yet found for phosphorylation (see Introduction; and Hirano-Ohnishi and Watanabe, 1989; Matten et al., 1990) do not correspond to those of the modification we are looking for.
(4) We have particularly paid attention to acetylation of tubulin, on account of several analogies of this PTM with that revealed in Paramecium, namely its detection in cilia, flagella and other stable MTs of metazoa and protists (L’Hernault and Rosenbaum, 1983; LeDizet and Piperno, 1991), including Paramecium (Cohen and Beisson, 1988; Torres and Delgado, 1989).
The immunocytochemical and peptide mapping data we obtained here and previously (Adoutte et al., 1991; Callen et al., 1994) have shown that the modification revealed by PAT and AXO 49 was different from the acetylation, and that the post-translational epitopes recognized by both antibodies are located near the C-terminal end of both subunits from Paramecium and metazoan axonemal tubulins, contrary to the acetylation site which is located in the N-terminal part of the α subunit only.
(5) We have further focussed our attention on polyglutamylation which shares some common features with the modification investigated here. It has been detected near the C-terminal end of both tubulin subunits (see Introduction), and in Paramecium, the reactivity of GT335 mAb was abolished after subtilisin cleavage (Bré et al., 1994), like that of AXO 49.
However, we have shown that various isoforms of Paramecium axonemal tubulin are differentially reactive with GT335 and AXO 49 mAbs. Furthermore, although widely spread among axonemal tubulins (Fouquet et al., 1994), glutamylation also occurs on labile cytoplasmic MTs in Paramecium (Bré et al., 1994) and metazoa (Wolff et al., 1992), and is a major PTM of mammalian brain tubulin (Eddé et al., 1991), contrary to the modification recognized by AXO 49 (Callen et al., 1994).
(6) A remaining candidate for the PTM analysed here is the very recently discovered polyglycylation (Redeker et al., 1994). The latter has been localized at residues Glu 445 and Glu 437 of Paramecium axonemal α- and β-tubulin, respectively, which fits with the location determined here for the epitopes recognized by AXO 49, namely in the sequence segments α(444-449) and β(432-442). The final characterization confirms that the target of AXO 49 indeed corresponds to the polyglycylation (M. H. Bré et al., unpublished). Thus, the potential high degree of polymodification, consisting of the addition of up to 34 glycyl units, and the great diversity of polyglycyl chain lengths displayed (Redeker et al., 1994) can well account for the reduced accessibilities of various antibodies to their epitopes in Paramecium axonemal tubulin and axonemes, respectively observed by immunoblotting and immunocytochemistry, and for the considerable apparent molecular mass heterogeneity of axonemal tubulin exhibited on gels and immunoblots. Moreover, the multi-step nature of polyglycylation could account for the sequential appearance of PAT and AXO 49 immunoreactivity in the course of assembly of microtubular structures in Paramecium, and also for the non reactivity of some transient structures (such as the cytospindle) with AXO 49; PAT and AXO 49 could recognize differently glycylated forms of tubulin. In addition, the opposite relative reactivities of the cilia with the two antibodies would imply a specific distribution of glycylated forms of tubulin through the cell. It is striking that this differential PTM distribution correlates with specific morphogenetic stability properties of the corresponding cell territories.
Moreover, the extension of AXO 49 reactivity to axonemal tubulins of metazoa, such as sea urchin and quail, indicates that this new PTM revealed by our antibodies is widely spread through evolution.
Our data, by increasing the number of PTMs occurring in the C-terminal end of both tubulin subunits, known to play a regulatory role in MT assembly and dynamics, and also in interactions with MAPs and motor proteins, raises the question of the role of this new modification. The uniqueness of the polyglycylation resides in its occurrence in microtubular assemblies including the most stables ones known, namely the axonemes of cilia and flagella, therefore suggesting, in contrast to others, a possible role of this modification in MT stability. This could be achieved at three levels (Gelfand and Bershadsky, 1991), either by acting directly on polymer stability, or/and indirectly through interactions with associated proteins or with the membrane.
This work was supported by the CNRS and the Université Paris-Sud. We are very grateful to A. Adoutte for his constant encouragements and for his important contribution to this work. We thank particularly J. C. Clérot and R. Jeanmaire-Wolf for the production of AXO 49 mAb without which this work could not have been carried out; A. Baroin-Tourancheau and P. Delgado for the construction of fusion proteins; A. M. Callen for performing experiments with fusion proteins; M. H. Bré for helpful discussions and critical reading of the manuscript; G. Coffe for fruitful discussions and F. Iftode for constant help in Paramecium culture. We also would like to thank the colleagues who provided us with protein samples and antibodies. We are grateful to H. Philippe for sharing aligned sequences; M. Laurent for help in confocal microscopy (Service d’Imagerie Cellulaire, Orsay);
ACKNOWLEDGEMENTS
G. Géraud for his assistance in the use of a two-laser equipped confocal microscope (Service d’Imagerie, Institut Jacques Monod, Paris); G. Johannin for his help in picture scanning (Service d’Imagerie Cellulaire, Orsay); C. Couanon for help in picture and manuscript editing; and N. Narradon for extensive photographic work.