Copyright  2001 by the Genetics Society of America

ND9P, a Novel Protein With Armadillo-like Repeats Involved in Exocytosis: Physiological Studies Using Allelic Mutants in Paramecium Marine Froissard, Anne-Marie Keller and Jean Cohen Centre de Ge´ne´tique Mole´culaire, CNRS, 91198 Gif-sur-Yvette Cedex, France Manuscript received August 7, 2000 Accepted for publication October 23, 2000 ABSTRACT In Paramecium, a number of mutants affected in the exocytotic membrane fusion step of the regulated secretory pathway have been obtained. Here, we report the isolation of one of the corresponding genes, ND9, previously suspected to encode a soluble protein interacting with both plasma and trichocyst membranes. Nd9p is a novel polypeptide that contains C-terminal Armadillo-like repeats. Point mutations were found in the first N-terminal quarter of the molecule and in the last putative Armadillo repeat, respectively, for the two thermosensitive mutants, nd9-1 and nd9-2. The different behaviors of these mutants in recovery experiments upon temperature shifts suggest that the N-terminal domain of the molecule may be involved in membrane binding activity, whereas the C-terminal domain is a candidate for protein-protein interactions. The nonsense nd9-3 mutation that produces a short N-terminal peptide has a dominant negative effect on the nd9-1 allele. We show here that, when overexpressed, the dominant negative effect can be produced even on the wild-type allele, suggesting competition for a common target. We suggest that Nd9p could act, like some SNARE proteins, at the membrane-cytosol interface to promote membrane fusion.

I

N the secretory pathway, exocytosis designates the last step of the traffic in which membrane fusion permits delivery of the vesicle contents to the external medium. Many metazoan cell types—and also a few protozoa such as Paramecium—can regulate exocytosis, releasing granules previously accumulated in the cytoplasm in response to extracellular stimuli. The whole mechanism of membrane fusion in exocytosis and its regulation is currently being unraveled by a combination of complementary approaches, genetics in yeast and physiological and biochemical analysis in neurosecretory cells (for review, Su¨dhof 1995; Pfeffer 1999). The discovery of conserved proteins such as N-ethylmaleimide sensitive fusion protein (NSF), soluble NSF attachment proteins (SNAPs), and SNAP receptors (SNAREs) in vesicle and target membranes has allowed the elaboration of an elegant model in which the SNAREs play a major role in vesicle-target recognition and membrane fusion (for review, Ludger and Galli 1998; Rizo and Su¨dhof 1998; Weber et al. 1998). However, some evidence suggests that the machinery for fusion and its regulation involve more partners (for review, TerBush et al. 1996; Gerst 1999; Lustgarten and Gerst 1999). For identification of new molecules participating in regulated exocytosis, Paramecium appears to be an excellent model for three main reasons: (1) It displays a well-developed secretory pathway in which defensive

Corresponding author: Jean Cohen, Centre de Ge´ne´tique Mole´culaire, CNRS, Av. de la Terrasse, 91198 Gif-sur-Yvette Cedex, France. E-mail: [email protected] Genetics 157: 611–620 (February 2001)

secretory granules, called trichocysts, are docked just beneath the plasma membrane in a “prefusion” state, awaiting external stimulation for release. (2) This organism is easily amenable to genetic dissection and, since the secretory process is not essential, numerous viable mutants can be generated. (3) Since the cortical position of trichocysts can be monitored by light microscopy, mutant screening can be specifically focused on exocytosis defects. Exocytosis mutants, called nd (nondischarge) have defects restricted to the postdocking steps of the pathway, namely, signal reception and transduction and exocytotic membrane fusion, without alteration of prior events such as biogenesis or transport (for review, Vayssie´ et al. 2000). To date, 31 alleles defining 17 ND genes have been isolated and the system is far from being saturated, indicating that exocytotic membrane fusion involves numerous proteins. Examination of the exocytotic sites of Paramecium by freeze fracture and transmission electron microscopy reveals specific structures, called rosette and connecting material, normally found, respectively, in the plasma membrane and just beneath. Such structures are absent in the nd mutants (Beisson et al. 1976; Lefort-Tran et al. 1981). The products of the ND genes must then be either integral components of these structures or regulators of their assembly. To understand the role of ND products and bring new clues to the mechanism of membrane fusion, a molecular approach by complementation cloning of these mutants was initiated in our laboratory (Skouri and Cohen 1997) for the ND7 gene, using a newly developed complementation cloning method (Haynes et al. 1996).

612

M. Froissard, A.-M. Keller and J. Cohen

Here, we describe the identification of ND9, the ND gene for which the most extensive information is available to date. By physiological studies of the mutant nd9-1, Beisson et al. (1980) identified Nd9p as a diffusible cytosolic component, capable of interactions with both partners of the fusion event, trichocyst and plasma membranes, in a way influenced by the lipid composition of these membranes. The cytosolic nature of Nd9p has been since confirmed by microinjection experiments of high-speed supernatant of wild-type cell extracts and of 35–40% saturation of ammonium sulfate precipitate of this supernatant to rescue the phenotype of the nd9-1 mutant (J. Cohen, unpublished results). More recently, genetic interactions involving ND9 were discovered: interallelic interactions in the form of a negative dominance of the nd9-3 allele on nd9-1 (although it is recessive against ND9⫹) and intergenic interactions with ND16 and ND18 (Bonnemain et al. 1992) suggesting homopolymeric and heteropolymeric interactions between Nd9p, Nd16p, and Nd18p. In this article, we report the cloning of the ND9 gene and the sequencing of the wild-type and the three mutant alleles. Physiological studies on two mutants, harboring mutations in the N and C terminus, respectively, and a negative dominant effect of a mutation producing a small N-terminal peptide lead us to propose that at least two domains may exist in the molecule, the N terminus for interactions with membranes and the C terminus for interactions with other proteins.

MATERIALS AND METHODS Strains and culture conditions: The wild-type strain was Paramecium tetraurelia stock d4-2, derived from stock 51 (Sonneborn 1974). The strains bearing the mutations nd9-1, nd9-2, and nd9-3 have been previously described in Cohen and Beisson (1980). Cells were grown at 27⬚ or 35⬚ in grass infusion (wheat grass powder, Pines International, Lawrence, KS), bacterized with Klebsiella pneumoniae the day before use, and supplemented with 0.4 ␮g/ml sitosterol (Sonneborn 1970). For the temperature-shift experiments, cultures were grown at the nonpermissive temperature for a minimum of 48 hr, until the mutant phenotype was fully established, and then starved at the same temperature before shift experiments. For cerulenin (Makor Chemical Co., Jerusalem) and puromycin (Sigma, St. Louis) treatments, drugs were added to the medium just before the shift. Drug experiments: Determination of the cerulenin dose: Different dilutions of cerulenin were tested on the wild type. After starvation of cultures at 35⬚, cerulenin was added before the temperature shift to 15⬚. The dose of 20 ␮g/ml, identical to the one used by Beisson et al. (1980), was tested but appeared to be lethal. Only a 5-␮g/ml dilution was tolerated by the cells. The differences in sensitivity could be due to the difference of culture medium used at the two periods. Determination of the puromycin dose: Puromycin was used at 100 ␮g/ml on wild-type starved cultures. When cells were grown at 35⬚, lethality was observed whether the culture was kept at this temperature or shifted to lower temperature dur-

ing the treatment. At 27⬚, no deleterious effect on the viability of the cells was obtained. Monitoring exocytosis: To visualize individual cells with their own discharged trichocysts, a saturated solution of picric acid is used as a fixing secretagogue. Discharged trichocysts remain clustered around the cell surface and can easily be visualized under dark-field light microscopy with a ⫻10 objective. DNA microinjection: Microinjections were made under an inverted Nikon phase-contrast microscope, using a Narishige micromanipulation device and Eppendorf air-pressure microinjector. Cells clonally derived from microinjected ones were submitted to the picric acid test after 24 hr and 48 hr of growth, respectively, to monitor their exocytotic capacity. For silencing and overexpression experiments, wild-type cells were treated with a solution of aminoethyldextran (Plattner et al. 1984) at 0.01% to stimulate trichocyst exocytosis before microinjection to avoid further discharge that could disturb the microinjection procedure. Gene cloning: The cloning of the ND9 gene was done by functional complementation of the nd9-3 mutant, as described in Skouri and Cohen (1997) for ND7. The ND9 gene was isolated by sib selection using microinjection of DNA prepared from an indexed library of Paramecium macronuclear DNA (Keller and Cohen 2000). Genomic DNA extraction: Total DNA for sequencing was prepared from log-phase culture cells using the DNAzol reagent (GIBCO Life Technologies, Paisley, UK) according to the method recommended by the supplier. Total wild-type DNA for Southern blots was prepared from log-phase cultures according to Duharcourt et al. (1995). PCR on genomic DNA for microinjection and sequencing: Polymerase chain reactions were made with the kit Expand Long Template PCR System (Roche Diagnostics, Mannheim, Germany). Each reaction (50 ␮l), adjusted to a concentration of nucleotides corresponding to the Paramecium A ⫹ T-rich genome composition (740 nm of dATP and dTTP; 260 nm of dCTP and dGTP), contained 150 ng of genomic DNA, 50 pmol of each primer, and 3 units of polymerase mix. The primers used were: Primer 1: 5⬘-ATGATTAGTGTGACAACTAAG-3⬘ Primer 2: 5⬘-ATTGTTTGAGTAGAAAATCGG-3⬘ Primer 3: 5⬘-TTGATTGAAGTAATTCAGCAG-3⬘ Primer 4: 5⬘-GATCAAGCTAAGATACAGATATGATAGATG-3⬘ Amplification was performed with 1 cycle of denaturation (92⬚, 2 min), 10 cycles of denaturation (92⬚, 10 sec), annealing (55⬚, 30 sec), extension (68⬚, 3 min 30 sec), and then 20 cycles of denaturation (92⬚, 10 sec), annealing (55⬚, 30 sec), extension (68⬚, 3 min 30 ⫹ 15 sec/cycle) with a final extension (68⬚, 7 min). Homology-dependent gene silencing: This experiment exploits the fact that microinjection into wild-type cells, at high copy number, of the coding region of a target gene (from the ATG to the TGA without flanking sequences) leads to a specific reduction in expression of the corresponding endogenous gene (Ruiz et al. 1998). Dot blot: To determine the amount of transforming DNA in different clonal cell populations, duplicate aliquots of 50 cells were isolated manually and transferred to 400 ␮l of 0.8 n NaOH. The cell lysates were incubated for 30 min at 65⬚ and loaded on positive membranes Hybond-N⫹ (Amersham, Little Chalfont, UK). After loading, the membranes were left in contact with 0.4 n NaOH for 15 min and washed in 2⫻ SSC. Hybridization was carried out according to Church and Gilbert (1984), at 60⬚. The membranes were then washed at the same temperature with decreasing concentrations of SSC in the presence of 0.1% SDS as follows: 2⫻ SSC for 30 min and

Nd9p, a Novel Protein Involved in Exocytosis 0.2⫻ SSC for 30–45 min (Sambrook et al. 1989). Images were obtained by using a Phosphorimager (Molecular Dynamics, Sunnyvale, CA). Hybridization was quantified by ImageQuant software (Molecular Dynamics). Southern blots: Paramecium DNA was digested by restriction enzymes according to the protocols recommended by the supplier (New England Biolabs, Beverly, MA) and then fractionated by electrophoresis on 1% agarose gels and transferred to Hybond-N⫹ filters and treated as described for Dot blot. RNA extractions and Northern blots: Total RNA was prepared essentially according to the method of Chomczynski and Sacchi (1987) using the Trizol reagent (GIBCO Life Technologies) except that the cells were lysed by vortexing in the presence of glass beads. Total RNA was fractionated on formaldehyde/1.25% agarose gels and transferred to positively charged nylon membranes (Ambion, Austin, TX) by capillarity. Hybridization was carried out at 60⬚ in 6⫻ SSC, 2⫻ Denhardt’s solution 0.1% SDS (Sambrook et al. 1989); the filters were then washed and imaged as described for Dot blot. Preparation of radioactive probes: DNA for 32P labeling reactions consisted of 800-bp PCR amplification products of wild-type genomic DNA with ND9 specific primers: 5⬘-AGTAT TGGATAACTCATTCAG-3⬘ (primer 5) and 5⬘-GAGGCTAAG TATCTTGACAC-3⬘ (primer 6). Each PCR reaction (50 ␮l) contained 150 ng of DNA, 50 pmol of each primer, 0.2 mm of each dNTP, and 2.5 units of Taq DNA polymerase (Roche Diagnostics). Reactions were carried out for 1 cycle of denaturation (1 min, 92⬚) and 30 cycles of denaturation (30 sec, 92⬚), annealing (45 sec, 54⬚), and extension (90 sec, 72⬚) with a final extension (10 min, 72⬚). Probes were synthesized by [32P]dATP incorporation using a Random Primers labeling system (GIBCO Life Technologies) according to the supplier’s protocol. Sequencing: The insert of the clone p70k8 containing the wild-type ND9 gene was sequenced by Eurogentec custom sequencing service. The three mutant alleles were sequenced with an ABI 310 sequencer (Perkin-Elmer, Foster City, CA) using the BigDye Primer Cycle Sequencing Ready Reaction kit (Perkin-Elmer). Sequence analysis: Initial characterization of the DNA and protein sequences, as well as hydrophobicity plots using the Kyte-Doolittle algorithm, were performed with the DNA Strider program (Marck 1988). The peptide sequence was deduced from the DNA sequence using the ciliate genetic code where UAA and UAG encode glutamine instead of stop. Homology searches were performed with the BLAST program (Altschul et al. 1997). Secondary structures and motif contained in the protein were detected using different research tools, InterProScan (Apweiler et al. 2000), SMART (Schultz et al. 2000), and Pfam (Bateman et al. 2000).

RESULTS

613

Figure 1.—Organization and expression of the ND9 gene. (a) Restriction map in which the positions of the sites for the following restriction enzymes, BclI (C), EcoRV (E), BglII (G), HindIII (H), and XbaI (X) are represented. (b) The p70k8 insert obtained by complementation cloning displays two putative ORFs, a small one and a large one orientated in opposite directions. Mutations in the three available nd9 mutants, nd9-1, -2, and -3, have been found in the large ORF as indicated. (c) Three kinds of PCR fragments have been generated from this sequence; the first one, using primers 1 and 2, covering the major part of the coding frame (ATG to 20 bases before TGA) and represented in black was used for silencing experiments; the second one, using primers 3 and 4, in which 218 bases upstream from the ATG and 214 bases downstream to the TGA have been included and which is represented in gray, was used in complementation and competition experiments according to the origin of the template DNA, wild-type and nd9-3, respectively; the third one, using primers 5 and 6, was labeled with 32P and used as a probe on Southern and Northern blots. (d) Southern blot of wild-type genomic DNA. Restrictions were made with some of the restriction enzymes used for mapping the insert; see a. (e) Northern blot of total RNAs from wild-type and three mutant strains. The level for each strain was compared with the level of ND7 RNA. The sequence data are available from GenBank/EMBL/DDBJ under accession no. AJ293945.

Identification of the ND9 gene Cloning of the ND9 gene by functional complementation: The plasmid p70k8, able to rescue the Exo⫺ phenotype of the nd9-3 mutant (see materials and methods), was shown by restriction mapping to contain a 4.3-kb insert. After complete sequencing, two open reading frames (ORFs), a large one of 2649 bp and a small one of 753 bp, were identified. Characterization of the ORF responsible for the rescue: The ORF corresponding to the ND9 gene was iden-

tified in three ways (Figure 1): First, the method used to clone the ND7 gene (Skouri and Cohen 1997) was attempted for ND9 prior to availability of the indexed genomic library. The first step consisted of testing the rescuing activity of total digests of wild-type genomic DNA. The results showed that genomic DNA digestions by BclI, BglII, EcoRV, and HindIII did not complement the nd9-3 strain whereas XbaI DNA digestion did. These data indicate that the ND9 gene contains the first series

614

M. Froissard, A.-M. Keller and J. Cohen TABLE 1

TABLE 2

Exocytotic capacity of cell lines with different nd9 genotypes as a function of temperature

Inactivation of ND9 by homology-dependent gene silencing

Exocytosis nd9 genotype

18⬚

27⬚

35⬚

⫹/⫹ 1/1 2/2 3/3 ⫹/1 ⫹/2 ⫹/3 1/2 2/3 1/3

⫹ ⫹ ⫹ ⫺ ⫹ ⫹ ⫹ ⫹ ⫹ ⫺

⫹ ⫺ ⫹ ⫺ ⫹ ⫹ ⫹ ⫹ ⫹ ⫺

⫹ ⫺ ⫺ ⫺ ⫹ ⫹ ⫹ ⫺ ⫺ ⫺

The first four lines represent the phenotypes of homozygous strains, wild type and three allelic nd9 mutants. nd9-1 and nd9-2 are thermosensitive with a different threshold; nd9-3 is altered at all temperatures. Lines 5 to 10 represent the phenotypes of heterozygous lines. All three alleles are recessive to the wild type. Interestingly, nd9-3 displays a specific negative dominance on the nd9-1 allele but not on nd9-2 (Bonnemain et al. 1992).

of sites but no XbaI site. Comparison of this “functional” restriction map with the one established from the sequencing of the insert (Figure 1a) suggests that the ND9 gene corresponds to the large ORF. Second, a PCR product covering the large ORF and flanking sequences was microinjected into the nd9-3 mutant and rescued its exocytotic capacity (Figure 1c, shaded line). Third, and most significantly, single substitutions were found in this ORF in the three allelic mutants known (Figure 1b). The mutations are missense mutations yielding amino acid changes L188S and W846Q for the two thermosensitive mutants, nd9-1 and nd9-2, respectively (Table 1) and an early nonsense mutation (W105stop) for the nd9-3 mutant expressed at all temperatures (Table 1). Southern blot experiments performed on genomic wild-type and mutant DNA indicated that the gene is unique in the genome (Figure 1d). Northern blot experiments using total RNA from wild-type cells and all three mutant alleles, and revealing a 2.8-kb messenger, showed that the gene is expressed at a very low level, but similarly in the four strains (Figure 1e). Inactivation of ND9 by homology-dependent gene silencing: The fact that point mutations in a gene give an nd phenotype does not systematically exclude the possibility that the gene may be involved in other steps of trichocyst secretion. To study the function of the ND9 gene, a silencing approach was undertaken. Some wildtype cells were transformed by microinjection into their macronucleus of a PCR fragment corresponding to the major part of the ORF of the ND9 gene (Figure 1c, solid line). After 48 hr of growth, the clones derived from microinjected cells were tested for exocytosis by picric

Experiment Control 1 2 3

No. of injected cells

Phenotypic classes Lethality

I

II

III

IV

0 17 51 27

0 0 6 0

10 5 22 8

7 14 9

1 3 3

4 6 7

Three rounds of microinjection with the PCR fragment corresponding to the ND9 ORF (in black in Figure 1c) have been done into the macronucleus of wild-type cells. Exocytotic capacity of ⵑ10 cells per clone, derived from each injected cell after 2 days of culture at 27⬚, was tested by picric acid. The clones were classified according to the average phenotypes of individual cells into four classes: I, wild-type phenotype, ⵑ1000 trichocysts (tric.)/cell; II, 150–500 tric./cell; III, 1–100 tric./ cell; IV, no trichocyst. The control represents noninjected cells otherwise processed as injected ones.

acid treatment. Some of them showed partial or total abolishment of exocytosis (Table 2). Careful examination of exocytosis-defective cells by phase-contrast light microscopy revealed trichocysts with normal morphology anchored to the cortex. The defect generated in silenced cells is indistinguishable from an nd phenotype. It therefore seems that in the secretory pathway ND9 is involved only in the late membrane fusion step of exocytosis, as was also deduced for ND7 in silencing experiments (Ruiz et al. 1998). Presence of Armadillo-like repeats in the Nd9p primary structure: The 882-amino-acid predicted polypeptide (Figure 2) has no clear homologue in other species and therefore appears to be a novel protein. The only BLAST hits, with low scores on the C terminus, are yeast Vac8p (Fleckenstein et al. 1998; Pan and Goldfarb 1998) and human smgGDS proteins (Yamamoto et al. 1990; Kaibuchi et al. 1991). Both proteins contain Armadillo repeats, first described in the armadillo protein of Drosophila melanogaster (Riggleman et al. 1989), and which consist of ⵑ42-amino-acid-long stretches placed in tandem, with low conservation of residues between repeats but characterized by the presence of a hydrophobic core. Crystallographic analysis of ␤-catenin Armadillo repeats (Huber et al. 1997) revealed a tridimensional structure of the protein in a superhelix of helices with a positively charged groove, each repeat forming three ␣-helices. Neighbor Armadillo repeats share strong interactions to each other for coherence of the superhelix and at least six Armadillo domains are necessary to form this tridimensional structure. The BLAST alignment of Nd9p involved two Armadillo repeats for Vac8p and five of them for smgGDS although only one was recognized as a bona fide Armadillo repeat in Nd9p using the Pfam tool (Figure 2a). The regular alternation of the hydrophobic cores of the Armadillo

Nd9p, a Novel Protein Involved in Exocytosis

615

Figure 2.—Identification of Armadillo-like repeats in ND9p. (a) Alignments of ND9p with Vac8p (P39968) (middle) and with the Armadillo consensus (top and bottom), as performed, respectively, using the BLAST and Pfam research tools. (b) Hydrophobicity plots of Vac8p and ND9p. The repeated hydrophobic cores are highlighted by segments on top of the figures. These cores correspond to the Armadillo repeats in Vac8p and have the same length (ⵑ40 amino acids) in Nd9p. Below each plot is represented the region of similarity between the two proteins. Box gray scales are as in c. (c) Overall organization of the 882-amino-acid long polypeptide Nd9p. The three mutations are indicated on the protein. The C-terminal Armadillo-like repeats (amino acids 572 and 882) are represented as boxes and are aligned with the Armadillo regions of two proteins with which they share weak similarity, Vac8p and smgGDS (AAF32290), as determined by the BLAST program. Black boxes indicate bona fide Armadillo repeats, gray boxes are putative ones, according to alignments to Armadillo-containing proteins, and open boxes represent hydrophobic cores harboring the same organization as Armadillo repeats.

repeats produces a regular pattern in hydrophobicity plots, as shown for Vac8p in Figure 2b. When the same analysis is performed on the Nd9p sequence, a similar regular pattern of seven 40-amino-acid cores is observed in the C terminus. The second of these cores is recognized as an Armadillo repeat and alignments could be made between repeats 2 and 3 of Nd9p and Vac8p and between repeats 2 to 6 of Nd9p and smgGDS (Figure 2c). The possible existence of seven Armadillo repeats

in the C terminus of Nd9p argues that a superhelix of helices could be formed and the presence of a mutation in the last of these repeats suggests an essential role of this region for the protein. Physiological studies of the mutant strains The determination of the sequence of the ND9 gene and of its mutated forms leads to a primary observation

616

M. Froissard, A.-M. Keller and J. Cohen

that the alleles nd9-1 and nd9-2, which give different thermosensitivity thresholds, have mutations at opposite ends of the molecule. It has been shown for nd9-1 cells that the thermosensitivity was not due to a conformation change of the mutated product but rather stemmed from the natural physiological variations in the lipid composition of membranes during adaptation to temperature changes (Beisson et al. 1980). Indeed, transfer of cytoplasm from nd9-1 cells grown at the permissive temperature (exo⫹) to nd9-1 cells grown at the nonpermissive temperature (exo⫺), be it by microinjection or by conjugation at the permissive temperature, did not rescue the exo⫺ partner as would be the case with the wild-type donor cells. In addition, the analysis of the phenotype reversion during shifts from nonpermissive to permissive temperatures in the presence or absence of a protein synthesis inhibitor (puromycin) and a lipid metabolism inhibitor (cerulenin) leads to the conclusion that Nd9-1p was unable to function in a membrane context with a high proportion of saturated fatty acids (as in membranes of a cell grown at 27⬚). In this context, we wondered whether the nd9-2 mutation conferred the same properties. We addressed this question by extending the recovery kinetics and inhibitor experiments to nd9-2 cells. Kinetics of recovery of nd9-1 and nd9-2 mutant phenotypes upon temperature shifts: We first compared the kinetics of reversion of both mutants in control conditions. Due to the different thresholds of thermosensitivity, the shifts were made from nonpermissive (35⬚) to permissive (15⬚) temperatures common to both of them. Therefore, the mutant nd9-1 not only serves as a comparison with results obtained 20 years earlier, but also as an internal control since the shifts were made between different temperature ranges. Exocytosis capacity of the two mutants after shifting was determined by picric acid treatment at regular time intervals. The nd9-2 strain began to revert as early as 4 hr after transfer, whereas nd9-1 did so only after 18 hr of culture at 15⬚ (Table 3). These two mutants, in addition to displaying distinct thermosensitivity thresholds, thus have distinct recovery kinetics. Effects of cerulenin and puromycin on phenotypic recovery: Cerulenin and puromycin effects on wild-type cells: The use of inhibitors that are toxic during growth necessitates the use of stationary phase cells. When applied to stationary cells at 35⬚, puromycin was lethal, possibly by preventing the synthesis of heat-shock proteins necessary for life at sublethal temperatures, and even in shifts to 15⬚ if puromycin was added soon after the shift. This inhibitor could not be used in experiments of shifts with nd9-2, since the nonpermissive temperature is strictly 35⬚ for this mutant. In contrast, cerulenin is not toxic for stationary cells at any temperature. However, cerulenin—an inhibitor of fatty acid synthase (Omura 1976; Inokoshi et al. 1994; Moche et al. 1999)—applied during a shift of stationary wild-type cells from 35⬚ to

TABLE 3 Kinetics of phenotype recovery of nd9-1 and nd9-2 mutants Time after temperature shift 0 2 4 6 12 18 30 40 60 72

hr hr hr hr hr hr hr hr hr hr

Phenotype nd9-1

nd9-2

⫺ ⫺ ⫺ ⫺ ⫺ ε/⫺ ε⫹ / ε / ⫺ ⫹ / ε⫹ / ε / ⫺ ⫹ / ε⫹ / ε ⫹

⫺ ⫺ ε/⫺ ε⫹ / ε / ⫺ ⫹ / ε⫹ / ε / ⫺ ⫹ / ε⫹ / ε ⫹ ⫹ ⫹ ⫹

The recovery kinetics of nd9-1 and nd9-2 were compared during the same temperature shift from 35⬚ to 15⬚, respectively, nonpermissive and permissive temperatures for both mutants. The recovery of exocytotic capacity was tested by picric acid treatment on 10–15 cells of each culture at regular times after the shift. The two mutants showed different behavior: nd9-2 started to recover exocytosis 4 hr after the shift whereas for nd9-1, the lag was about 18 hr. ⫹, wild-type, ⵑ1000 tric./cell; ε⫹, 100–500 tric./cell; ε, ⬍20 tric./cell; ⫺, 0 tric.

15⬚ inhibited exocytosis (Table 4). This result may be related to exocytosis inhibition by this drug in other systems (Martinez et al. 1982; Metz et al. 1993). Cerulenin applied during transfers of smaller temperature intervals, 35⬚ to 27⬚ and 27⬚ to 15⬚, had no effect on exocytosis (Table 4). Therefore, inhibition experiments on the mutants were carried out with temperature shifts of smaller amplitude, encompassing their thermosensitivity threshold. Cerulenin and puromycin effects on the mutants: Stationary cultures of the nd9-1 and nd9-2 mutants, equilibrated at their respective nonpermissive temperatures, 27⬚ and 35⬚, were shifted to the permissive temperatures (15⬚ and 27⬚, respectively) in the absence or presence of cerulenin, and puromycin for nd9-1. As shown in Table 4, a clear inhibition of the recovery was observed only with nd9-1, in the presence of cerulenin but not of puromycin. This is in complete agreement with the previous results of Beisson et al. (1980). What is new is the complete absence of effect of this inhibitor on nd9-2, as if the defect caused by the mutation were insensitive to the lipid composition of the membranes. This observation, added to the previous ones on the two alleles, indicates that only the N-terminal mutation gives sensitivity to the lipid composition of the membranes, suggesting the presence of an N-terminal domain in Nd9p interacting with membranes. Dominant negative effects of nd9-3 Bonnemain et al. (1992) have shown a dominant negative effect of the nd9-3 allele on nd9-1, although it is

Nd9p, a Novel Protein Involved in Exocytosis

recessive to wild-type and nd9-2 alleles. Considering that the nd9-3 nonsense mutation is, like nd9-1, in the N terminus of the molecule, it was interesting to reexamine this selective negative dominance. The conclusions of Bonnemain et al. (1992) favored an interaction between several Nd9p products to generate a polymer, an interaction poisoned by Nd9-3p when Nd9-1p is the partner. Knowing now that Nd9-3p corresponds to a truncated version of the protein, we can extend the model and imagine that the negative dominance could correspond to competition toward any target, not solely itself. We wondered whether it would be possible to increase the negative effect of nd9-3 by overexpression and observe an effect in the wild type. PCR amplification products corresponding to the full ND9 gene both in the wild type (a sequence sufficient for rescue) and in the nd9-3 mutant were injected in the wild-type strain. The clones were tested 48 hr after injection both for the presence of plasmid DNA by dotblot experiments and for exocytotic capacity by picric acid treatment. Among the clones scored positive by dot blot, all of those that were microinjected with the nd9-3 version of the gene lost exocytotic capacity, whereas the clones microinjected with ND9⫹ DNA were unaffected. The actual overexpression of messenger RNA was checked on Northern blots and indeed observed in both microinjection experiments (Figure 3). We conclude that a dominant negative effect of the nd9-3 allele in wild-type cells can be produced.

617

Figure 3.—Effect of overexpression of wild-type and nd9-3 versions of the ND9 gene in wild-type cells. PCR amplification products corresponding to the full ND9 open reading frame with flanking sequences obtained with wild-type (WT) and nd9-3 (3) DNA as templates (see Figure 1c) were respectively injected into the macronucleus of wild-type cells. All clones that were microinjected with the nd9-3 version of the gene lost exocytotic capacity, whereas the clones microinjected with ND9⫹ DNA were unaffected. The actual overexpression of messenger RNA was checked on Northern blots for one cell line in each experiment (B) and is indeed observed in both microinjection experiments as compared with the level of ND7 expression (A).

DISCUSSION

ND9, a novel gene: The aim of the present study was to isolate the ND9 gene, a member of the group of genes involved in the final membrane fusion event of regulated exocytosis in Paramecium, and to gain inTABLE 4 Effects of cerulenin and puromycin on exocytosis recovery upon temperature shifts Exocytosis Strain Wild type

nd9-1 nd9-2

Cerulenin ⫺ ⫹ ⫹ ⫹ ⫺ ⫺ ⫺ ⫺ ⫹ ⫺ ⫺ ⫹

Puromycin ⫺ ⫺ ⫺ ⫺ ⫹ ⫹ ⫹ ⫺ ⫺ ⫹ ⫺ ⫺

35⬚

27⬚

⫹ ⫹ ⫹

→ → ⫹ ⫹

→

→ Lethal ⫹ → Lethal ⫹ → ⫺ → ⫺ → ⫺ → ⫺ → ⫹ ⫺ → ⫹

15⬚ ⫹ ε/⫺ ⫹ ⫹ ⫹ ε/⫺ ⫹

618

M. Froissard, A.-M. Keller and J. Cohen

sights into its function. We cloned the ND9 gene by functional complementation. Sequencing of the three existing mutant alleles revealed substitutions in an ORF of 2649 bp, confirming that this region corresponds to ND9. Homology-dependent gene silencing of this sequence provokes a loss of exocytotic capacity of wildtype cells and no other detectable phenotype. The ND9 gene thus seems to be necessary only for trichocyst exocytotic membrane fusion. Physiological analysis of two mutant alleles reveals two functional domains: Studies of the two allelic mutants, nd9-1 and nd9-2, by kinetics of recovery from the Exo⫺ to the Exo⫹ phenotype and by sensitivity of this recovery to cerulenin (inhibitor of fatty acid metabolism) permitted us to say that the two mutants were affected in different ways. Earlier work (Beisson et al. 1980) showed that the thermosensitivity of nd9-1 reflected a physiological variability of fatty acid composition of the membranes according to growth temperature and that Nd9-1p itself was not thermosensitive. In contrast, the lack of effect of cerulenin on the nd9-2 mutant suggests that Nd9-2p is thermolabile by itself, independent of the lipidic environment. These differences, together with the positions of the two mutations in nd9-1 and nd9-2 at the opposite ends of the primary sequence, argue in favor of two independent functional domains in ND9p, with the N-terminal domain in direct or indirect contact with membrane lipids. Negative dominance of nd9-3 confirms the binding of the N terminus of Nd9p to a target: The results obtained by Beisson et al. (1980) demonstrating the interaction of Nd9p with both trichocyst and plasma membranes, together with those of Bonnemain et al. (1992) showing a dominant effect of nd9-3 on nd9-1, lead to the conclusion that the protein could act as a homopolymer connecting both membranes. In this case, Nd9-3p would poison Nd9-1p-containing polymers but not wild-typecontaining ones. Our present observation that the mutation nd9-3 is a nonsense mutation in the N terminus

leads us to revisit these conclusions in the context of two different domains. The poisoning by a small Nd93p peptide can be understood if we consider that there is competition with a common target. In this case, we have to propose that Nd9-1p has less affinity for the target at permissive temperature than the wild-type protein, so that Nd9-3p competes more easily with it. By overexpression of Nd9-3p, we obtained a negative dominance also on the wild-type Nd9p product, in favor of our hypothesis of competition with a target (Figure 4). The N terminus therefore seems to be the site of competition for a target and a region in which a “lipidsensitive” mutation has been obtained. We propose that the N terminus is the domain of the protein interacting with membranes. Whether the interaction is directly with lipids or occurs via another protein whose expression or nature depends on the lipidic composition of membranes cannot be deduced from these experiments. An Armadillo-like protein-protein interaction domain: Although no strong similarity between Nd9p and known proteins has been found, Nd9p displays a putative protein-protein interaction domain in the form of a region resembling Armadillo repeats. Although the similarity is weak, this region could actually be an Armadillo domain since alignments could be made up to five contiguous Armadillo domains in the smgGDS protein and that the order of the repeats is preserved (indeed, within a domain, there is more conservation of the repeats according to their position; Peifer et al. 1994). In a study on double mutants, negative genetic interactions have been demonstrated between ND9, ND16, and ND18 alleles (Bonnemain et al. 1992), suggesting physical interactions between the encoded products. It is tempting to propose that such interactions occur through the C-terminal Armadillo domain. Promotion of membrane fusion by Nd9p at the membrane-cytosol interface? Membrane fusion is a universal process in eukaryotic cells whose underlying mechanism

Figure 4.—Model of the interallelic interactions between ND9 gene products visible at 18⬚. Nd9p is represented as a hook in which the short arm is the N-terminal domain. The substitution in Nd9-1p is represented by a black dot and the truncated Nd9-3p peptide by a small open box. In the wild type, Nd9p is supposed to interact with a target, the interaction in turn provoking normal exocytotic capacity (⫹/⫹, Exo⫹). This interaction is weakened with Nd9-1p but still functional (1/1, Exo⫹), but completely abolished if Nd9-3p is present (1/3, Exo⫺). Nd9-3p is not able to compete with wild-type Nd9p in the stoichiometric ratio found in heterozygotes (⫹/3, Exo⫹) but can do so when it is overexpressed by high copy number transformation (⫹/⫹ inj. nd9-3, Exo⫺) compared to the control injection (⫹/⫹ inj. ND9, Exo⫹).

Nd9p, a Novel Protein Involved in Exocytosis

is still largely unknown. The NSF/SNAP/SNARE machinery has been demonstrated to be essential to complete the membrane fusion/membrane retrieval cycle in other systems (Weber et al. 1998), but the precise step at which the machinery acts is still under discussion (Rizo and Su¨dhof 1998). A particular feature of this machinery is that it contains two-domain components, the SNAREs, which provide an interface between the membrane and other proteins (from the cytosol or the partner membrane). Indeed, SNAREs are either proteins with a transmembrane domain (syntaxin, synaptobrevin) or with covalent lipid modifications (SNAP25), which make them transmembrane- or membrane-associated proteins, and coiled-coil regions involved in the protein-protein interactions that bridge the membranes. Nd9p has no homology with any of the known SNAREs but, from our present analysis, seems to share overall organization with them: a membrane-binding domain (N terminus) and a protein-protein interaction domain (Armadillo-like domain in the C terminus). It is interesting to note that a yeast protein, Vac8p, involved in vacuole morphogenesis and protein targeting to the vacuole through membrane fusion of small vesicles, is built essentially of 11 Armadillo domains (Fleckenstein et al. 1998; Pan and Goldfarb 1998). Vac8p also has a potential to bind to membranes via myristoylation, palmitoylation, and acylation in the N terminus (Wang et al. 1998; Schneiter et al. 2000), necessary for its localization to the vacuolar membrane at sites of membrane-membrane contacts, but not for its function in protein targeting. This analogy between Nd9p and Vac8p may reflect the existence of novel proteins and new types of interactions, mediated by Armadillo domains, involved in membrane fusion. In particular, Armadillo-containing proteins have been demonstrated to interact with SmgGDS, which acts as a GDP exchange factor (Yamamoto et al. 1990; Kaibuchi et al. 1991; Vithalani et al. 1998). The peculiarity of trichocyst exocytosis in Paramecium is that the membranes to fuse are arrested in a prefusion state and remain bound by a so-called connecting material. Further studies, using in particular the nd9-2 mutant altered in the last Armadillo-like repeat, could help determine whether ND9p actually mediates the assembly of the connecting material through the recruitment of key factors for exocytosis such as G proteins and their regulators, since we suspect that they are also involved in trichocyst exocytosis (Lumpert et al. 1990). We thank Carl Creutz (Charlottesville, VA) and France Koll, Linda Sperling, and Janine Beisson (Gif-sur-Yvette, France) for critical reading of the manuscript. Grant 96024 from Centre National de la Recherche Scientifique “Biologie cellulaire: du normal au Pathologique” and a grant from the Ministe`re de l’Education National, la Recherche et la Technologie “Programme de Recherche Fondamentale en Microbiologie, Maladies Infectieuses et Parisitaires” are gratefully acknowledged.

619

LITERATURE CITED Altschul, S. F., T. L. Madden, A. A. Schaffer, J. Zhang, Z. Zhang et al., 1997 Gapped BLAST and PSI-BLAST: a new generation of protein database search programs. Nucleic Acids Res. 25: 3389– 3402. Apweiler, R., 2000 Protein sequence databases. Adv. Protein Chem. 54: 31–71. Bateman, A., E. Birney, R. Durbin, S. R. Eddy, K. L. Howe et al., 2000 The Pfam protein families database. Nucleic Acids Res. 28: 263–266. Beisson, J., M. Lefort-Tran, M. Pouphile, M. Rossignol and B. Satir, 1976 Genetic analysis of membrane differentiation in Paramecium. Freeze-fracture study of the trichocyst cycle in wildtype and mutant strains. J. Cell Biol. 69: 126–143. Beisson, J., J. Cohen, M. Lefort-Tran, M. Pouphile and M. Rossignol, 1980 Control of membrane fusion in exocytosis. Physiological studies on a Paramecium mutant blocked in the final step of the trichocyst extrusion process. J. Cell Biol. 85: 213–227. Bonnemain, H., T. Gulik-Krzywicki, C. Grandchamp and J. Cohen, 1992 Interactions between genes involved in exocytotic membrane fusion in Paramecium. Genetics 130: 461–470. Chomczynski, P., and N. Sacchi, 1987 Single-step method of RNA isolation by acid guanidinium thiocyanate-phenol-chloroform extraction. Anal. Biochem. 162: 156–159. Church, G. M., and W. Gilbert, 1984 Genomic sequencing. Proc. Natl. Acad. Sci. USA 81: 1991–1999. Cohen, J., and J. Beisson, 1980 Genetic analysis of the relationships between the cell surface and the nuclei in Paramecium tetraurelia. Genetics 95: 797–818. Duharcourt, S., A. Butler and E. Meyer, 1995 Epigenetic selfregulation of developmental excision of an internal eliminated sequence on Paramecium tetraurelia. Genes Dev. 9: 2065–2077. Fleckenstein, D., M. Rohde, D. J. Klionsky and M. Rudiger, 1998 Yel013p (Vac8p), an armadillo repeat protein related to plakoglobin and importin alpha is associated with the yeast vacuole membrane. J. Cell Sci. 111: 3109–3118. Gerst, J. E., 1999 SNAREs and SNARE regulators in membrane fusion and exocytosis. Cell Mol. Life Sci. 55: 707–734. Haynes, W. J., K.-Y. Ling, Y. Saimi and C. Kung, 1996 Toward cloning genes by complementation in Paramecium. Neurogenetics 11: 81–98. Huber, A. H., W. J. Nelson and W. I. Weis, 1997 Three-dimensional structure of the armadillo repeat region of beta-catenin. Cell 90: 871–882. Inokoshi, J., H. Tomoda, H. Hashimoto, A. Watanabe, H. Takeshima et al., 1994 Cerulenin-resistant mutants of Saccharomyces cerevisiae with an altered fatty acid synthase gene. Mol. Gen. Genet. 244: 90–96. Kaibuchi, K., T. Mizuno, H. Fujioka, T. Yamamoto, K. Kishi et al., 1991 Molecular cloning of the cDNA for stimulatory GDP/GTP exchange protein for smg p21s (ras p21-like small GTP-binding proteins) and characterization of stimulatory GDP/GTP exchange protein. Mol. Cell. Biol. 11: 2873–2880. Keller, A. M., and J. Cohen, 2000 An indexed genomic library for Paramecium complementation cloning. J. Eukaryot. Microbiol. 47: 1–6. Lefort-Tran, M., K. Aufderheide, M. Pouphile, M. Rossignol and J. Beisson, 1981 Control of exocytotic processes: cytological and physiological studies of trichocyst mutants in Paramecium tetraurelia. J. Cell Biol. 88: 301–311. Ludger, J., and T. Galli, 1998 Exocytosis: SNAREs drum up! Eur. J. Neurosci. 10: 415–422 (Erratum: Eur. J. Neurosci. 10: 1936). Lumpert, C. J., H. Kersken and H. Plattner, 1990 Cell surface complexes (‘cortices’) isolated from Paramecium tetraurelia cells as a model system for analysing exocytosis in vitro in conjunction with microinjection studies. Biochem J. 269: 639–645. Lustgarten, V., and J. E. Gerst, 1999 Yeast VSM1 encodes a v-SNARE binding protein that may act as a negative regulator of constitutive exocytosis. Mol. Cell. Biol. 19: 4480–4494. Marck, C., 1988 ‘DNA Strider’: a ‘C’ program for the fast analysis of DNA and protein sequences on the Apple Macintosh family of computers. Nucleic Acids Res. 16: 1829–1836. Martinez, J. P., M. V. Elorza, D. Gozalbo and R. Sentandreu, 1982 Regulation of alpha-galactosidase synthesis in Saccharo-

620

M. Froissard, A.-M. Keller and J. Cohen

myces cerevisiae and effect of cerulenin on the secretion of this enzyme. Biochim. Biophys. Acta 716: 158–168. Metz, S. A., M. E. Rabaglia, J. B. Stock and A. Kowluru, 1993 Modulation of insulin secretion from normal rat islets by inhibitors of the post-translational modifications of GTP-binding proteins. Biochem. J. 295: 31–40. Moche, M., G. Schneider, P. Edwards, K. Dehesh and Y. Lindqvist, 1999 Structure of the complex between the antibiotic cerulenin and its target, b-ketoacyl-acyl carrier protein synthase. J. Biol. Chem. 274: 6031–6034. Omura, S., 1976 The antibiotic cerulenin, a novel tool for biochemistry as an inhibitor of fatty acid synthesis. Bacteriol. Rev. 40: 681–697. Pan, X., and D. S. Goldfarb, 1998 YEB3/VAC8 encodes a myristylated armadillo protein of the Saccharomyces cerevisiae vacuolar membrane that functions in vacuole fusion and inheritance. J. Cell Sci. 111: 2137–2147. Peifer, M., S. Berg and A. B. Reynolds, 1994 A repeating amino acid motif shared by proteins with diverse cellular roles [letter]. Cell. 76: 789–791. Pfeffer, S. R., 1999 Transport-vesicle targeting: tethers before SNAREs. Nat. Cell Biol. 1: E17–E22. Plattner, H., H. Matt, H. Kersken, B. Haacke and R. Sturzl, 1984 Synchronous exocytosis in Paramecium cells. I. A novel approach. Exp. Cell Res. 151: 6–13. Riggleman, B., E. Wieschaus and P. Schedl, 1989 Molecular analysis of the armadillo locus: uniformly distributed transcripts and a protein with novel internal repeats are associated with a Drosophila segment polarity gene. Genes Dev. 3: 96–113. Rizo, J., and T. C. Su¨dhof, 1998 Mechanics of membrane fusion [news]. Nat. Struct. Biol. 5: 839–842. Ruiz, F., L. Vayssie, C. Klotz, L. Sperling and L. Madeddu, 1998 Homology-dependent gene silencing in Paramecium. Mol. Biol. Cell 9: 931–943. Sambrook, J., E. F. Fritsch and T. Maniatis, 1989 Molecular Cloning: A Laboratory Manual. Cold Spring Harbor Laboratory Press, Cold Spring Harbor, NY. Schneiter, R., C. E. Guerra, M. Lampl, V. Tatzer, G. Zellnig et al., 2000 A novel cold-sensitive allele of the rate-limiting enzyme of fatty acid synthesis, acetyl coenzyme A carboxylase, affects the

morphology of the yeast vacuole through acylation of Vac8p. Mol. Cell. Biol. 20: 2984–2995. Schultz, J., R. R. Copley, T. Doerks, C. P. Ponting and P. Bork, 2000 SMART: a web-based tool for the study of genetically mobile domains. Nucleic Acids Res. 28: 231–234. Skouri, F., and J. Cohen, 1997 Genetic approach to regulated exocytosis using functional complementation in Paramecium: identification of the ND7 gene required for membrane fusion. Mol. Biol. Cell 8: 1063–1071. Sonneborn, T. M., 1970 Methods in Paramecium research. Methods Cell Physiol. 4: 242–335. Sonneborn, T. M., 1974 Paramecium aurelia, pp. 469–594 in Handbook of Genetics, Vol. 2, edited by R. C. King. Plenum Press, New York. Su¨dhof, T., 1995 The synaptic vesicle cycle: a cascade of proteinprotein interactions. Nature 375: 645–653. TerBush, D. R., T. Maurice, D. Roth and P. Novick, 1996 The Exocyst is a multiprotein complex required for exocytosis in Saccharomyces cerevisiae. EMBO J. 15: 6483–6494. Vayssie´, L., F. Skouri, L. Sperling and J. Cohen, 2000 Molecular genetics of regulated secretion in Paramecium. Biochimie 82: 269–288. Vithalani, K. K., C. A. Parent, E. M. Thorn, M. Penn, D. A. Larochelle et al., 1998 Identification of darlin, a Dictyostelium protein with Armadillo-like repeats that binds to small GTPases and is important for the proper aggregation of developing cells. Mol. Biol. Cell 9: 3095–3106. Wang, Y. X., N. L. Catlett and L. S. Weisman, 1998 Vac8p, a vacuolar protein with armadillo repeats, functions in both vacuole inheritance and protein targeting from the cytoplasm to vacuole. J. Cell Biol. 140: 1063–1074. Weber, T., B. V. Zemelman, J. A. McNew, B. Westermann, M. Gmachl et al., 1998 SNAREpins: minimal machinery for membrane fusion. Cell 92: 759–772. Yamamoto, T., K. Kaibuchi, T. Mizuno, M. Hiroyoshi, H. Shirataki et al., 1990 Purification and characterization from bovine brain cytosol of proteins that regulate the GDP/GTP exchange reaction of smg p21s, ras p21-like GTP-binding proteins. J. Biol. Chem. 265: 16626–16634. Communicating editor: S. L. Allen