In addition to the alpha-, beta- and gamma-tubulins, essential for microtubule assembly in all eukaryotes, several new tubulin subfamilies have been recently identified in a cascade of discoveries [1]. Complementation cloning of the UNI3 mutation in Chlamydomonas led to the characterization of delta-tubulin, an unexpected fourth member of the tubulin subfamily, involved in assembly of basal bodies [2]. Genome search for delta-tubulin led to identify not only deltas in mammals and protozoa, but also to disclose further new divergent tubulins, epsilon and zeta [3,4,5]. A sixth subfamily, eta, was characterized in Paramecium [6] by complementation cloning of the sm19 mutation affecting basal body duplication [7]. In contrast to alpha-, beta- and gamma-tubulins, the new tubulins, of which a few sequences only are available, do not seem to be present in all eukaryotes and their function might concern such elaborate microtubule arrays as centrioles and basal bodies.

For each of the new subfamilies, sequence conservation is weak and, as only a few sequences are presently available, their phylogenetic relationships are unclear [1,6]. It is therefore important to ascertain if the members of a given presumed subfamily have the same function. For the delta-tubulin subfamily, the function is known only for Chlamydomonas: deletion of this gene results in the loss of the C-tubule, in each of the nine microtubule triplets which are the hallmark of centrioles and basal bodies. In the absence of a functional assay, a similar role of the delta-tubulin homologs identified in mammals [3,4] and protozoa [5,6] had not yet been demonstrated. We report here a functionnal analysis of delta-tubulin in Paramecium, a favourable organism because of the high number of its basal bodies and of their differentiated duplication pattern.

In the course of random sequencing (see http://caroll.vjf.inserm.fr/pt/) of clones from the recently described indexed genomic library of Paramecium [8], a likely δ-tubulin homolog was characterized [6]. Southern blots (not shown) demonstrate that this sequence corresponds to a unique gene, designated as δPT1. The deduced polypeptide, 397 amino acids long, is shorter than tubulins; alignment with the known δ-tubulin sequences shows (Figure 1) that 43% of its residues are identical to the corresponding residue in at least one of the other sequences. After those of Chlamydomonas [2], man [4], mouse [3] and Trypanosoma [5], δPT1 is the fifth δ or putative δ-tubulin identified to date.

To ascertain whether δPT1 had a similar function, we took advantage of the phenomenon of homology-dependent gene silencing operating in Paramecium, previously discribed [9] and used to demonstrate the role of δ-tubulin in basal body duplication [10]. Microinjection of the coding sequence of a gene, at high copy number, into the macronucleus results in the inactivation of the corresponding endogenous gene.

The effect of δPT1 inactivation was studied on the individual progenies of 132 microinjected cells from two independent series of 33 and 99 cells respectively. In both sets of clones, growth rate, cell morphology and cytoskeletal organization remained normal over the four first divisions after injection. Then, over the next 2-3 divisions, the same syndrome progressively developed in a large majority of the clones (25/33 and 86/99 respectively for the two experiments): disorganisation of the basal body complex forming the oral apparatus, disorders in the cortical pattern with abnormal positioning of basal bodies and orientation of striated and microtubule rootlets, presence of intra-cytoplasmic basal bodies and erratic microtubule bundles suggesting occurence of microtubule nucleation at abnormal sites, possibly in relation to the intracytoplasmic basal bodies (Figure 2). All such transformed cells reached a terminal phenotype: rounded shape, smaller cell size with reduced number of basal bodies in the oral apparatus and on the cortex and after 7-8 divisions died of starvation as their oral apparatus became non-functional.

Basal body ultrastructure was examined on cells just beginning to display cortical disorders and shape changes, around the 5th division. At this early stage of expression of the transformant phenotype, a striking defect, lack of the C-tubule, was found in practically all basal bodies (156 out of 171 unambiguous basal body cross-sections), corresponding to at least 20 different cells from several transformed clones (Figure 3). It must be pointed out that this effect is highly specific: loss of the C-tubule was never observed in other experimental or genetical conditions interfering with basal body properties or duplication, such as γ-tubulin or centrin inactivation [10, 11; Garreau de Loubresse, unpublished] or in basal body defective mutants such as sm19 cells, mutated in η-tubulin [6], or sm2 cells [7].

In some basal bodies, both triplets and doublets coexisted (Figure 3c', e'), suggesting, as demonstrated in the case of other genes [9], that inactivation of the δPT1 gene was not total. In several instances, a mother "all doublets"basal body had produced a daughter basal body with more pronounced tubule deficiencies (Figure 3c'; Figure 4). Absence of the C-tubule clearly does not prevent duplication but may lead over successive duplication cycles to more severe ultrastructural defects accounting for abortive structures. Positioning and nucleation properties are also affected: intracytoplasmic basal bodies were found (not shown), in agreement with immunocytological observations (Figure 2b,c). Another recurrent anomaly concerned the ciliary rootlet. This thick striated fiber is normally nucleated and anchored on the three anterior right triplets (Figure 3c); on basal bodies composed of doublets (Figure 3c'), the rootlet arises from a narrower region covering only two doublets and it appears fuzzier and less tapered than in control cells. The meshes of the infraciliary lattice, a contractile meshwork whose pattern depends on the basal body pattern [11] were also distorted. Cross-sections of cilia were essentially normal, in agreement with the fact that axonemal outer doublets originate from the A- and B-tubules of the basal body.

Inactivation of the δPT1 gene results in loss of the C-tubule, specific of the centriolar structure, as previously observed in the UNI3 mutant of Chlamydomonas, carrying a total deletion of the δ-tubulin gene [2]. In Paramecium, further loss, namely of B- and even A-tubules, is also occasionally observed. The latter observation may imply either that the function of δ-tubulin is not restricted to C-tubule assembly, or that absence of the C-tubule destabilizes B- and, in turn, A-tubules. While δ-tubulin function remains to be characterized, the available observations in the two organisms at least delineate likely functions of the C-tubule. First, in both organisms, the C-tubule plays a role in anchoring, development and positioning of basal body appendages which form the cortical cytoskeleton; in Paramecium, these defects also correlate with mislocalization and internalization of basal bodies, yielding cortical disorders and reduction of basal body number and cell size. Second, the C-tubule does not play a direct role in ciliogenesis: axoneme structure and cilary activity seem normal in Paramecium ; in Chlamydomonas, doublet basal bodies can form flagella, although after some maturation delay [2]. Third, the C-tubule is not directly required for basal body duplication: doublet basal bodies duplicate normally in the algae and in the ciliate; however, in Paramecium, further tubule loss may appear (Figure 3c', Figure 4). This difference may be due to the fact that in Chlamydomonas, basal bodies duplicate once per cell cycle, while in Paramecium, more than half of the ca 3500 cortical basal bodies, and all those destined to the neo-formed oral apparatus, undergo two or three fast cycles of duplication at each division [12,13]: the C-tubule might then contribute to stability of the microtubule scaffold (see Figure 4) and in turn to the efficiency of the "duplication complex" assumed to relay the structural information from mother to daughter basal body [14,15]. Conversely, the striking fact that absence of one or more triplets does not alter the underlying ninefold symmetry of the organelle, as also observed in Chlamydomonas mutants [16], provides an experimental argument in favour of the existence of a complex providing pattern information.

Finally, considering that addition of the C-tubule (a hemi-tubule, like the B-tubule) is the last step in basal body/centriole assembly [17,18], its absence upon depletion of δ-tubulin reveals not only the existence of subtle specificities at microtubule nucleation sites, but also a role of tubulin isotype diversity in generating microtubule arrays, well documented in Drosophila [19,20].