Copyright Q 1992 by the Genetics Society of America
Interactions Between Genes Involvedin Exocytotic Membrane Fusion in Paramecium Hugues Bonnemain, Tadeusz Gulik-Krzywicki, Claude Grandchamp and Jean Cohen Centre de Ginitique Moliculaire, associi a l’llniversiti Pierre et Marie Curie, Centre National de la Recherche Scientiftque, 91 198Gf-sur-Yvette Cedex, France Manuscript received March 13, 1991 Accepted for publication October 16, 1991
ABSTRACT Crosses between members of two independent collections of Paramecium tetraurelia mutants blocked in the final membrane fusion step of trichocystrelease (nd mutants) allowed us to define 13 complementation groups comprising 23 alleles. The mutant nd9“ was then usedas a target in a mutagenesis experiment designed to screen both revertants and new mutants in order to identify interacting genes. This mutant was chosen because it is the best known of its class to date and seems to bealtered in assembly ofthe material connecting the trichocyst membrane to the plasma membrane and inassemblyof the “rosette,” a complex array of intramembranous particlesin the plasma membrane at the trichocyst insertion sites. No revertants were obtained but two new mutants deficient for rosette assembly were identified, nd16’ and nd18, whose gene products appear to interact with that of nd9. Indeed, the double mutants grown at 18”, a permissive temperature for each of the single mutants, are characterized by a deficiency in exocytosis and in rosette assembly, as are also double mutants combining other allelic forms of the same genes. Moreover, aberrant dominance relationships among alleles ofnd9 and of nd16 indicate the existence of interactions between identical subunits, which most likely assemble into multimeric structures. The nd16 gene product was shown by microinjection experiments to be a cytosolic factor, as is the nd9 gene product. It is therefore tempting to propose that the nd16 gene product alsobelongs to the connecting material and is involved in rosette assembly, incooperation with nd9 and nd18.
S
ECRETION relies on apathwayofmembrane traffic common to all eukaryotes (PALADE1975), involving protein synthesis and transit through the endoplasmic reticulum and the Golgi apparatus, vesicular transport andfusion with the plasma membrane. A still poorly understood step in this pathway is the final membrane fusion of the secretory vesicle with the plasma membrane. Geneticdissection of thesecretory pathway has mainly been undertaken in yeast for constitutive secretion [NOVICK,FIELDand SCHEKMAN 1980; SCHEKMAN 1985) and Paramecium and Tetrahymena for regulated secretion (BEISSONet d . 1976, 1980; COHEN and BEISSON1980; LEFORT-TRAN et al. 198 1; POUPHILEet a l . 1986; ORIAS,FLACKS a n d SATIR 1983; TURKEWITZ, MADEDDUand KELLY1991) also see PLATTNER(1987) and ADOUTTE(1988) for reviews]. In particular, the genetic study of secretory processes in Paramecium has shown that proteins a r e necessary for membrane fusion (BEISSONet al. 1976, 1980; LEFORT-TRAN et al. 198 1; VILMARTand PLATTNER 1983; POUPHILE et al. 1986). Thisnotion,in contrast to whatwas thought initially (CHI, LAGUNOFF a n d KOELHER1976; LAWSONet al. 1977; ORCI, PERRELET and FRIEND 1977; PINTODA SILVAand NoGUEIRA 1977; TANAKA, DE CAMILLIand MELDOLESI, 1980) is now documented by data on different sysGenetics 1 3 0 461-470 (March, 1992)
tems. “Connecting materials” have been found at exocytosis sites inchromaffin cells (AUNIS,HESKETH and DEVILLIERS1979;NAKATA, SOBUEand HIROKAWA 1990). Cytosolic proteins have been shown to be involved in exocytosis in mast cells (KOFFERand GOMPERTS 1989) and in BHK cells (DE CURTIS and SIMONS 1989) as are synexin(CREUTZ 1981) and calpactin (BURGOYNE, 1988; DRUSTand CREUTZ 1988) in chromaffin cells. G-proteins were shown to be necessary for in vitro interactionsbetweenzymogengranules and theplasma membrane (NADIN et al. 1989), protein GE for exocytosisinmast and other secretory cells (GOMPERTS1990), “smallG-proteins” for invertase secretion in yeast (GOUDet al. 1988) and neurotransmitter release in nerve terminals (FISCHERVON MOLLARD et al. 1990). Parameciummutantsdeficientintheirsecretory capacity a r e easily obtained. T h e level of the mutational block can be determined firstly by simple cytological observations which permit one to distinguish betweendefectsinsecretory vesicle development, transport or fusion with the plasma membrane (POLLACK 1974; BEISSONet al. 1976; COHEN and BEISSON 1980)and secondly by microinjectionexperiments using the method developedby AUFDERHEIDE (1978) which allow the direct determination of the compart-
H. Bonnemain et al.
462
ment that is affected by the mutation, such as the plasma membrane, secretoryvesicle (called trichocyst) o r cytosol. Inaddition,thepredetermined sites of exocytosis are marked by a particle arrayin the plasma membrane known as the rosette(JANISH 1972; PLATTNER, MILLERand BACHMAN 1973) which can be visualized by freeze fracture electron microscopy. Using mutants, BEISSON et al. (1976, 1980), LEFORT-TRAN et al. (1 98 1) and POUPHILE et al. (1 986) correlated the presence of the rosette and underlying electron-dense "connecting material" with exocytotic capacity. In wild-type Paramecium cells, the trichocysts are docked at predetermined sites of the cell cortex in a "pre-fusion" state, awaiting an external stimulus that triggers fusion of the trichocyst membrane with the plasma membrane. Thus, a major advantage of the Paramecium system is that final membrane fusion can be dissociated from previous steps and studied per se. The nd mutants have normal trichocysts docked at the plasma membrane that are not excretable. They display a very specific defect in the last step of exocytosis, the membrane fusion and/or its regulation. The objective of the present work was to dissect genetically the machinery involved in membrane fusion, firstly by enumerating the known nd genes by allelism tests since mutants have beenobtained in differentlaboratories and secondly by looking for direct interactions betweennd genes. Mutagenesis was carried out on the most studied nd mutant to date, nd9", whose expression is thermosensitive. We looked bothforrevertantsatthe restrictive temperature (27")and for new mutants at the permissive temperature (18 "). New mutants were obtained and two of them carry mutationsin the genes nd16 and nd18 that respectively appear to interact with nd9 and to be involved in rosette assembly. The gene product of at least one of these mutants ( n d 1 6 ) is a cytosolic factor which is therefore likely to be a structural component of the connecting material, as already proposed for the nd9 gene product. MATERIALS AND METHODS
Strains: The wild-type strain used in this study was the stock d4-2,a derivative of stock 5 1of Paramecium tetraurelia (SONNEBORN 1975). The strains bearing the mutations nd2, nd7, nd6, nd7, ndY, nd9b, ndY, nd12, nd16 and nd17 have been previously described (COHEN and BEISSON1980). The mutations nd126 (and itsallele nd242),nd139,nd146, nd163, ndl69 and nd203 were found in wild stocks and transferred to the stock 51 (NYBERG1978). The mutations ndT, nd16b and nd18 were obtained inthis study. Cells bearing the mutations tam6,tam8 (normally shaped but unattached trichocysts; BEISSONand ROSSIGNOL 1975) and tam38 (abnormal and unattached trichocysts; RUIZ et aE. 1976) were used as donors and recipients in microinjection experiments (see below). Culture conditions: Cells were grown in grass infusion bacterized the day before use withKlebsiella pneumoniae and supplemented with 0.4 pg/ml @-sitosterolthree in depression
slides according to SONNEBORN (1970) or 96 well plates as previously described (COHEN and BEISSON1980). Mutagenesis: Mutagenesis with 400 J/m' of ultraviolet light at a wavelength of 254 nm was performed according to COHENand BEISSON (1980) onpreautogamous nd9" cells. After mutagenesis, cells were rendered 100% homozygous by induction of autogamy, in order to allow expression of recessive mutations. Autogamous cells were isolated in fresh medium in 96-depression plates. Tests on lethality, growth rate and exocytotic performance were done on post autogamous clones grown for at least four to seven fissions at 27 O . Cells observed at 18" weregrown for at leastsix additional fissions at this temperature. Determination of exocytotic capacity: T o monitor exocytotic capacity during the screening procedure, the plates containing the post-autogamous mutagenized clones were replicated into picric acid-containing plates and observed under a stereo microscope with pseudo dark field illumination (COHENand BEISSON1980). In other routine tests, samples of 20-50 picric acid-treated cells were observed at 10 X magnification in dark field and thenumber of secreted trichocysts estimated. In the tables, indicates that exocytosis is comparable to the one of the wild type and represents discharge of70-100% of the cell'strichocyst(eachcell contains ca. 1000 trichocysts), f refers to partial and variable exocytosis in the range of 5-30% of the trichocysts being discharged, e represents a discharge of fewer than 10 trichocysts and - means that notrichocysts weredischarged. Genetic analysis: Crosses were performed as described by SONNEBORN (1970) and the F, and FP progeny were cultured in 96-depression plates and their phenotype observed as previously described (COHEN and BEISSON1980). Functional complementation tests by microinjection: Microinjection experiments were carried out at 18" as described by AUFDERHEIDE (1978), COHEN and BEISSON (1980) et al. (1981) with minor modification of and LEFORT-TRAN the microinjection device. To avoid the frequent leaks that occur in oil circuits, pressure variations for cytoplasm aspiration and injection were transmitted through an air circuit governed by a peristaltic pump. tam6 cells were used as a donors of wild type trichocysts and tam38 cells asrecipients. For microinjections using cells grown at 35"C, tam8 was used instead of tam38 because of the thermolethality of this mutant. Like tam38, the mutant tam8 possesses non functional trichocysts (AUFDERHEIDE 1978), although of normal shape, with a competent cytoplasm and plasma membrane. Appropriate transfers of cytoplasm from cell to cell permit deduction of the precise localization of the physiologicallesion generated by the mutation to the trichocyst compartment, the plasma membrane or the cytosol, asexplained in detail by LEFORT-TRAN et at. (1981). Briefly, when trichocysts of the mutant tested cannot be secreted after transfer into tam38 or tam8 cells, and when tam6 trichocysts are perfectly functional upon transfer into the mutant, it is concluded that the site of action of the mutation is in the trichocyst compartment itself. When the reverse situation is observed, the mutation is said to affect the non-trichocyst compartment; forthe nd mutants in particular, the non-trichocyst compartment can be equated to the plasma membrane or to structures firmly linked to the plasma membrane since all the events preceeding direct membrane interactions are unaffected in the nd classof mutants. When microinjections in both directions allow trichocyst secretion, as does microinjection ofcytoplasm devoid of functional trichocysts intothe mutant being tested, it is concluded that the site of action of the mutation is the cytosol.
+
463
Interacting Exocytotic Mutants Freezefracture electron microscopy: In order to prevent unwanted stimulation during the centrifugation and fixation steps, 100-ml cultures of the desired genotypes, grown at the desired temperature, were made 20 mM MgC12with a 0.5 M stock solution for15 min before harvesting(GILLICAN and SATIR 1983).The pellets were fixed in 10 ml of 0.5% glutaraldehyde, 10 mM Na-phosphate, pH 7.0, for 20 min at room temperature and washed twice in the same buffer without glutaraldehyde. Fixation was used in these experiments since it allows rosette particles to partition into the plasmaface,like the other particles (double ring) of the exocytotic site(LEFORT-TRAN et al. 1978). The final pellets were placed between two highly hydrophilic 10-pm thick round copper plates fixed on Balzers thin copper sample holders. The high hydrophilicity of the copper plates was obtained by washing in nitric acid and distilled water.The samples so "sandwiched" between copper plates were then rapidly plunged into liquid propane (AGCERBECK and GuLIK-KRZYWICKI,1986). The openingofthesandwichat -125 under a vacuum ofabout 1.33X 10-5 Pa in a freezefracture unit (Balzers BAF301) produces two complementary fracturesof the frozen sample. Replication of the fractured surfaces was performed using platinum-carbon. The replicas were cleanedin chromic acid, washed with distilled water, and observedwith a Philips 301 electron microscope.
TABLE 1 Phenotypes andallelism relationshipsof nd mutants exo phenotype at Strains
Previous names 18" 27" 35"
-
" -
nd3" nd3' nd3' nd139 nd3d nd3e ndtib
-
separate mutageneses, composedof 12 mutatedalleles in8genes (COHENand BEISSON1980), theother obtained by examination of wild type stocks, consisting of eightmutated alleles in six genes (NYBERG 1978). As these collections could containnd mutations in identical genes, systematic complementation tests were performed between the two collections by crosses and F1 analysis. T w o cases of allelism were found: ndP is allelic to nd139 (now nd3d)and nd6 to nd163 (now nd6'). For the crosses where allelism was found, a linkage test was performed by analyzing the F2 progeny. N o wild typerecombinantsappeared, indicating thattheconfrontedmutationsare separated by less than one centimorgan. Altogether, as summarized in Table 1,if we include the new nd gene (nd18) and thenew alleles (nd?, nd16') of previously identified genes obtained in this work and to be described below, we possess 23 nd mutations belonging to 13 genes, one of the mutations (nd246) being very leaky. Despite this large number of genes involved in the simple process of membrane fusion,many remain to bediscovered. Indeed, only one gene has been independentlymutated five times, onegenethree times and fourgenes twice; the seven remaining genes have only one mutated allele (Table 1). Many other genes should exist that could be revealed in future mutageneses. Search for new mutants and nd9" revertants: Mutagenesis: A way to identify interactions between gene products is to look for revertants of aknown mutant. In the absence of a positive screen for revertants of
-
"
nd6 nd163
nd 7 nd9" nd9 nd9' nd9c nd12
ndl7 nd18 nd126" nd126 nd126' nd242 nd146 nd146" nd146' nd148
-
-
c
-
-
-
-
d,e, g, k f. k
tric.
-
d, e, g
cyt.
-
a, b, c, e, g, k d d
+
d, g, j
-
-
CVt.
+ + + + -
k
-
CVt.
+ + + + + + -
h
tric.
-
d.
k
-
f f. k f f
f
nd I69
- "
nd203
A
+
tric.
-
d k
-
+ + + f
d, h f, k
pl. mb.
- pl. mb.
- "
-
a, e, i 1
-
-
+ + + -
-
References
1
+ + -
nd16 nd16" nd16'
Number of nd genes: Two independent collections of nd mutantsare available, one obtained after several
tric.
" -
O ,
RESULTS
-
Rosette
+ -
+ + +
WT*
ndB nd2
nd6"
Site of action
-
f. k
-
f, k
Column "previous name": when new alleles were discovered for a known gene, the name of the previously described mutations was changed by adding the letter a to its name. When two mutants already named were found to contain allelic mutations, the name with the lowest number was kept to designate the gene. For homogeneity, we propose to rename ndB nd2. Column =exo" (for exocytotic performance): - indicates no exocytosis; c means that only a few trichocysts (less than 10 per cell) can be excreted; f is for partial exocytosis of 50-300 trichocysts per cell; + indicates complete exocytosis of a "cloud" of >500 to 1000 trichocysts per cell. Column"site of action": tric. = trichocyst, pl. mb. = plasma et al. (1981) and membrane, cyt. = cytosol [see LEFORT-TRAN MATERIALS AND METHODS].
Column "References": a, AUFDERHEIDE (1978); b, BEISSONet al. (1976); c, BEISSONet al. (1980); d, COHENand BEISSON(1980); e, LEFORT-TRAN et al. (198 1); f, NYBERC(1978); g, POUPHILEet al. (1986); h, M. RossIcNoLandJ. BEISSON. unpublished; i, SONNEBORN (1 974); j, KERBOEUFand COHEN(1990); k , this paper. * Wild type.
nd mutants, we used a routine protocol (previously used for mutant isolation, COHENand BEISSON1980) where a few thousand individual clones were monitored for exocytosis. Previous mutageneses on wildtype cells have generally yielded one to three trichocyst mutation@) per 1000 clones tested (COHEN1980). However, the frequency of reversion-suppression of a particular nd gene could be expected to be far lower. A UV mutagenesis on nd9" cells and screeningon
464
H. Bonnemain et al. TABLE 2 Isolation of the new nd genes from the nd9" genetic background crosses FI segregations in
exo phenotype at Phenotypic class
I [nd9+-m+] 18 I1 [new] 111 [new] IV [nd9"-m+] V [ndY-m] 47 Total number of FZclones analyzed
18"
27"
+
+ +
+ + + -
35"
nd9"-w X W T *
nd9"-x x W T
+
28
-
13
0
0 21 20 82
20
- 25 - 20 -
*
29
-
94
86
nd9"-y x W T
nd9"-r x W T
0 0 17
55 0 0 0 31
82
Each of the new nd9"-m double mutants (where m designates the second mutation) obtained in the mutagenesis was crossed with the wildtype and the FZsegregation analyzed. Class I represents the parental wild-type phenotype, class V the parental double mutant phenotype and class IV the recombined nd9"-m+phenotype. FZsegregation for nd9"-w X WT yielded a new phenotype (class 11) which revealed the presence of the new gene w in the nd9+ context. For the cross nd9"-x X WT, the new Fz phenotype obtained (class 111) was barely distinguishable from class 1V [nd9"] so that the genotypes were ascertained by backcrosses with nd9", and clones carrying the mutation x in the nd9+ context identified. The cross nd9"-y X WT yielded no new FZ phenotype, butclass V was majoritary. Phenotypic segregation fits with a 2:l:l segregation ( x 2 = 1.78, P = 0.38)and in fact, by backcrossing clones of class V with nd9", clones carrying the new gene y in the nd9+ context gene could be segregated from nd9" indicating a strong linkage between the two were found in this class. In the cross nd9"-z X WT, no new " mutations (distance 5 1.2 cM). * Wild type.
5952 clones were nevertheless performed. AS shown in Table 1, the nd9" strain is thermosensitive and displays an exo' phenotype at 18" and an exo- phenotype at 27" so that revertants could be detected at the nonpermissive temperature (27") as exo+ clones while new trichocyst mutations would be detected by an exo- phenotype at thepermissive temperature (18 "). Out of 5952 post autogamous clones isolated after mutagenesis, 30% containeda lethal gene, 5% were slow growers, five clones displayed a phenotype exo- at 18 " , and not one was a revertant (exo+ at 27"). From observations by phase contrast microscopy, one of the new double mutants appeared to possess "football-shaped" trichocysts, unattached to the cell cortex (ft mutant), which was irrelevant for our study. The four other double mutants contained normal trichocysts, attachedtothe cell cortexand therefore had an nd phenotype. These four double mutants were genetically analyzed. Geneticanalysis of thenewdouble mutants: As the four new mutations (which we shall call w, x, y and z) appeared in a nd9" genetic background, the first experiment was to isolate these new mutations by crossing the double mutants by the wild type, as shown in Table 2. (1) The mutation w was in a new n d gene, called nd18, unlinked to nd9" and thermosensitive in that it gives an exo- phenotype only at 35". (2) The mutation x was also in an nd gene, allelic to nd16, and therefore called nd16'. Thismutation,unlinkedto nd9", is also thermosensitive with a threshold around 27" so that it was difficult to differentiate this mutation from nd9" in the F2 analysis. Backcrosses of F2 clones with nd9" were performed and theresulting Fs analyzed to ascertain their genotype, nd9" or nd16'. (3) The mutation y was an nd mutation, allelic to nd3, and therefore called ndT. The nd3c mutation, un-
linked to nd9", confers an exo- phenotype at all temperatures. (4) The mutation z could not be separated from nd9" in the cross performed, being at less than 1.2 cM from nd9", and was not studied further. The most important of these results is that two out of thethree new nd mutations, nd16' and nd18, increase the defect due to the nd9" mutation in conditions where they do not present an nd phenotype themselves. Indeed, at 18", both double mutants are exo- although each of the corresponding single mutants is exo+. This denotes the presence of negative interactions, at least at the phenotype level, between each of these genes and nd9". In order to be sure that the observed phenotypes do correspond to the newly identified mutations in combination with nd9" and not to other mutations induced by the mutagenesis, the genes nd16' and nd18 were reassociated with nd9" by crosses. Inthe FS progenies of these two crosses, 25% of the clones, the double mutant class, were unable to discharge their trichocysts when grown at 18" and therefore display the phenotype originally observed. nd16' and nd18 actually modify the phenotype of nd9". Analysis of intergenicinteractions: Systematic construction of double mutants combining all nd9, nd16 and nd18 alleles was carried out. As shown in Table 3, depending upon which alleles were associated, it was possible to observe either epistasis or negative interactions at the phenotypic level. Indeed, only one double mutant,nd9"-nd16", failed to present any kind of interaction, since the phenotypic transitionwas that of the mutant nd16", without interference due to the presence of the mutation nd9". In four other double mutants, three bearing nd9", and one nd16', the mutation with the most severe phenotype (with the lower temperature threshold) was found to be epistatic to
Mutants Interacting Exocytotic
465 TABLE 4
TABLE 3
Number of particles perrosette for nd mutants in nonpermissive conditions
Exocytotic phenotype of double mutants exo phenotype at Strains
nd9"-nd16" nd9"-ndltjb nd9"-nd18 nd9'-nd16"
27"
35"
Relationships between the genes
f (-,*)
-
No interaction
(+,+)
e,-)
(+,+)
(-,*)
18"
+ -
(+.+I (-*+I
+ +
(+,+) nd9b-nd16b
(+,+)
nd9'-nd18 nd9'-nd 16" nd9'-nd16' nd9'-nd18
+
*
Interaction
(-,-I -
Interaction ~~
(-,-)
(-,-I
Epistasis Interaction
(-.-)
(-.+)
(-,+)
(-,-)
(-,+I -
+
f
(+,+)
-
-
-
(-,f) (-,-)
-
-
(-,+) e>-) f (+>+) e , - ) (*,+) (-,-)
Epistasis Epistasis Epistasis Interaction Interaction
All the double mutant combinations between the three genes nd9 (three alleles a, b and c), nd16 (two alleles a and b) and nd18 (one allele) were constructed by crosses. The exocytotic capacity was monitored for each double mutant at three temperatures and compared to the performanceof its single mutant parent.The signs in parenthesis under the phenotype of the double mutants at each temperature indicate the exocytotic capacity of the parental single mutants and appear in the order in which their names are given in the double mutant name (data fromTable 1 ) . When the phenotype of the double mutant was similar to the one of the most affected parent, i.e., with the lowest temperature threshold exo+to exo-, we concluded that an epistatic relationship exists between the genes. When the phenotype of the double mutantwas worse than the one of each of its parents, that is with less exocytotic capacity than the parentsata given temperature, we concluded that a negative interaction between the t w o confronted genes was taking place.
the mutation giving the less severe phenotype. In the six remainingcombinations, thephenotype of the double mutant displayed a lower temperature threshold for shifting from exo+to exo- than either parent, indicating the presence of negativeinteractions. Moreover, the genes nd16 and nd18, which present interactions with nd9, also interact with each other when they are associated in double mutants. Possible interpretations of the molecular nature of these interactions will be presented in the Discussion. Analysis of interallelic interactions: In order to look for interallelic interactions within the complementation groups studied, we systematically analyzed the phenotypes of all the possible heterozygotes. The allelism of nd9"~b* (COHENand BEISSON1980) and of nd16"9b(this paper) has been deduced from the absence of F1 complementation at a temperature restrictive for the parents. However, the phenotypes of the heterozygotes can be observed at other temperatures,
0
1
Wildtype . . nd9"* 7320 nd16'(35")4673 nd18(35")20 9 nd126b 36 10 nd169 34 11 nd203(35")85 8
Interaction
-
(+*+)
(+,+) nd16'-nd18
f (+,+) f (+&)
(-,-I
(+,+I -
(-,+)
nd16"-nd18
-
Strains
2
3 4 5 6 7
Total No. Mean No. of 8 9 10 of sites particle/rosette
. . . . 6 12 1 1 2 . 10 6 3 1 . . 1 . . . . 4 . . . . . * . 2 2 . . . . . 10 1 0 2 . . . . .
..
. . ... . .
.
. .
.
. . . .
31 113 57 33 50 0.8 84 09.32
7.3 0.6 0.3 0.5 0.4
~
Fields of plasmic faces of freeze-fracture replicas presenting a sufficient number of visible exocytotic sites were photographed and the numberof particles per rosette was determined in sites occupied by a trichocyst, recognizable by the presence of a "ring" of particles and not "parentheses" (BEISSON et al. 1976). The table presents the number of rosettes against the number of particles they contain. Growth temperature was 27 a unless otherwise stated. * Data from BEISSONet al. (1976).
permissive for at least one of the parents. In the case of nd9, an aberrant dominance appeared in one combination, between nd9" and nd9": at 18 O , a temperature permissive for nd9" but not for nd9c (Table l), the heterozygote a/c presents aslightly leaky mutated phenotype (exocytosis of less than 10 trichocysts per cell), although the allele c is recessive when heterozygous with the wild type or the b allele. This indicates that nd9 gene productsmust interact with each other, probably for assembly into multimeric structures. In this case, the a mutated form is not compatible with the c one. A similar conclusion can be drawn fornd16 whose more affected b allele is dominant over a (a/b being exo+" at 27"), although recessive to the wild type allele. Freeze-fracture analysis of the new nd mutants and of interacting double mutants: All the nd mutants obtained to date but one(nd12; POUPHILE et al. 1986) lack the rosette of intramembranous particles as judged by freeze-fracture electron microscopy in the conditions wherethe exo- phenotype is expressed. It was therefore of interest to investigate the new nd mutants by this technique. Table 4 clearly shows that all the new mutants lack a rosette at trichocyst insertion sites at the non permissive temperature, as illustrated in Figure 1 for nd16', nd18 and nd126b. We also wondered whether the lack of exocytotic functions in interacting double mutants, at the temperature permissive for each of thecorresponding single mutants, was due to the loss of rosette activity or to the absence of rosette particles. Counts of rosette particles frominteractingdoublemutants are presented in Table 5 and indicate thatthey do not possess rosette particles at18" (Figure 2). Therefore,the interactions interfere with rosette assembly. Compartments affectedby some nd mutations: To localize the site of action of those nd mutations that have never been analyzed by this test, microinjections
H.Bonnemain et al.
466
FIGURE1.-Freeze-fracture images of exocytotic sites of wild-type and mutant cells. a and b, wild type grown at 27". c and d, nd126' grown at 27". e and f, nd16' grown at 35". g and h, nd18 grown at 35". In c to h, the growth temperatures indicated were nonpermissive ones. Thecells presented an exo-phenotype at the timeof fixation. Bar = 0.25 pm. TABLE 5 Number of particles per rosettein interacting double mutants
0.9
Strains(l8")
0
Wild type nd9"* nd16' n d 1l 8 nd9"-nd16' nd9"-nd18 53
7 1 1 30 23
1
1
1 8 15
2
3
7 1 2 4 12
8 9 3 4 3
4
5
6
7
1 8 14 6
5 11 18 8
10 8 10 16 21
14 6 5
*
20
8
21 5 2 8 0.6
9
1 1 .
IO 1 1 4.9 *
Total No. of sites
Mean No. of particle/rosetre
60 65 68 70 46
7.3 5.8 5.8
See legend to Table 4. Growth temperature was 18".
were performed for the strains nd7, nd169, nd126' and nd16'. T h e strains nd12, nd17, nd18 and nd203 were not analyzed because of the difficulty of maintaining their thermosensitive mutated phenotype (expressed only above 35") under our microinjection conditions. Table 6 shows that one nd mutant ( n d 7 ) is affected in the plasma membrane (according to the extrapolation explainedin MATERIALS AND METHODS), two in the trichocyst compartment (nd169, nd126') and onein the cytosol (nd16),although the repairwas weaker for this latteronethanforthe previously 1978). known reparable mutant, nd9" (AUFDERHEIDE This indicates that the nd16+ geneproduct is less or less exchangeable with its stableupontransfer mutated partner than thend9 gene product. DISCUSSION
In Paramecium, exocytotic membrane fusion, which can be prevented by mutations in numerous genes, involves the presence of a complex structural array detectable by electron microscopy as a rosette of intramembranous particles and its underlying con-
necting material, a cytoskeletal element thought to induce the rosette assembly (POUPHILEet al. 1986). Our objective is to know thenumber,natureand function of the genes involved in the process, as well as the nature of their interactions, in order to approach the underlying mechanisms. In this work, we have enumerated the known nd genes, which control membrane fusion in trichocyst 23 mutated alleles exocytosis, andfoundthatthe already obtained belong to 13 genes. In addition, we have observedinteractions between various allelic forms of three nd genes involved in rosette assembly. T h e products of t w o of them are shown to exist as diffusible cytosolic factors; this information is lacking for the thirdone. (1) T h e possible roles of the at least 13 gene products whose absence gives an nd phenotype and (2) the fact that the products of the three interacting genes could participate in assembly of the connecting material and of the rosette will successively be discussed. Number of nd genes and functions of the exocytotic site: The number of nd genes involved in the
Interacting Exocytotic Mutants
467
FIGURE2.-Freeze-fracture images of interacting double mutants. Cells were grown at 18", a permissive temperature for the single mutants and a non permissive one for the double mutants. a, nd16'; b, nd18; c, nd9"-nd16'; d, nd9"-ndl8. Exocytotic sites are indicated by arrows. Bar = 0.25 pm.
TABLE 6
Site of action of four mutants No. of cells with
Total No. of
Genes tested
Donor
Recipient
0 tric.
1-30 tric.
8
injected cells
Siteof actlon
tam6 nd3'
nd3' tam38
9 1
nd169
tam6 nd 169
nd169 tam38
3 11
7
10 11
tric.
nd126'
tam6 nd126'
nd126' tam38
2
6
8 11
tric.
ndl6'
tam6 nd16' tam8
nd16' 7 tam8 5 nd16' 23 15
6 16
13 21 38
cyt.
nd3'
11
9 9
mb. pl.
Four mutants were submitted to the microinjection test. Only cells for whichcytoplasmic transfer was certain and which were healthy at the test time, 2 hr 30 min-3 hr 30 min after microinjection, are reported. For nd16', both donor and recipient cells were cultured at 35" and microinjected cells were transferred to 35" immediately after microinjection. The number of trichocysts that could be secreted by each cell was determined by the picric acid test. The site of action was deduced as explained by LEFORT-TRAN et al. (1 981) and summarized in MATERIALS AND METHODS. PI. mb. =plasma membrane (or nontrichocyst compartment, see MATERIALS AND METHODS), tric. = trichocyst, cyt. = cytosol.
sole membrane fusion step of secretion might be much more than the 13 genes identifiedto date as the system is far from saturated. This number may seem elevated for a function restricted in space and time. However, as secretion is regulated in Paramecium, the complete exocytotic site is expected to ensure several different functions. First, as membrane fusion does not occur
without induction, the exocytotic site must be subject to permanent inhibitory control to prevent unwanted exocytosis. Second, when induction occurs, the site has to "sense" either thepresence of the inductor itself or the presence of a second messenger emitted elsewhere in the cell in response to the inductor. Third, the site has to respond to the induction either by actively fusing the membranes or by removing inhibition so as to permit "spontaneous" fusion of the closely apposed membranes. The mechanisms underlying these processes are not as simple as it was initially imagined. In Paramecium, a growing numberof physiological activities, enzymes and other proteins have been demonstrated or suggested to be associated with exocytosis by cytological, physiological and biochemical methods. A Ca2+ATPase activity located in the region of the exocytotic site is correlated with exocytotic capacity (MATT,BILINSKI and PLATTNER1978; PLATTNER et al. 1980); a 6365-kDa phosphoprotein is dephosphorylated upon triggering of exocytosis in wild-type cells but not in nd mutants (GILLIGAN and SATIR1982; ZIESENISSand PLATTNER 1985) and a calcineurin-like protein phosphatase and calmodulin have been suspected to be involved in exocytosis (MOMAYESIet al. 1987). A Ca2+ influx governed by a putative channelis induced when exocytosis is triggered and is inactivated by the mutation nd22 (KERROEUFand COHEN1990) and aGTPdependent mechanism seems to be involved in membrane fusion (LUMPERT,KERSKEN and PLATTNER 1990). A cytoskeletal function can be proposed for the
468
H. Bonnemain et al.
connecting material, a structure resistant to mild detergent attackthatstandsbetween trichocyst and plasma membranes. The connecting material represents a preformed link between the two areas to be fused upon induction. More interestingly, it seems to be an anchor for rosetteparticles. Indeed, as demonet al. (1978), rosetteparticles strated by LEFORT-TRAN are partitioned into the external face of the plasma membrane by freeze-fracture when cells are frozen without treatmentand into thecytoplasmic face when cells are frozen after prefixation, suggesting interactions between rosette particles and subjacent connecting material which can be stabilized by fixation. Ifwe consider, in addition,thatsome activities found in other secretory cells resulting fromthe action of proteases, protein kinases, Ca2+-binding proteins, phospholipases A2 and C (for a review see PLATTNER 1989), may exist in Paramecium, the number of genes controllingmembrane fusion alreadydetectedno longer appears so high. Interaction between nd genes: Our study was designed to look for genes that interact with nd9 since it is the most extensively studied nd gene to date and because of the availability of a thermosensitive allele, nd9", which allowed us to look both for revertants at the restrictive temperature and new mutants at the permissive temperature. BEISSONet al. (1976) demonstrated a correlation between the presence of the rosette and exocytotic capacity in the mutant nd9". AUFDERHEIDE (1978) reported that this mutant could be repaired by microinjection of cytoplasm from a nonallelic strain and this was confirmed for the allele nd9' (COHENand BEISSON1980). BEISSON et al. (1980) showed that the diffusible cytosolic product of the gene nd9" interacts with both plasma and trichocyst membranes in a way dependent on the lipidic composition of these membranes, and suggested that it is a component of the connecting material. In addition, fromthepresent work, we can notethataberrant dominance occurs between some alleles of this gene. This reflects direct monomer to monomer interactions between nd9 gene products andis in agreement with the hypothesis that the gene nd9 encodes a component of the connecting material. Our mutagenesis yielded no revertants, which a posteriori does not seem so surprising given the absence of a system for positive selection. However, new nd mutations were obtained. Two mutated genes isolated from the mutagenesis, nd16" and nd18, increase the exocytotic defect given by nd9": they confer an exo- phenotype at 18 to cells bearing nd9" although none of these alleles alone gives an nd phenotype at that temperature. It is remarkable that thesenegative effects provided by the genetic associations of nd9" with nd16' and nd18 can also be found using the alleles nd9' and nd16", and even in nd16-nd18 double
mutants, in which no mutated form of the gene nd9 is present (Table 3). The aggravated exo phenotype found in double mutants can also be observed ultrastructurally at thelevel of rosette assembly. At restrictive temperatures, all three single mutants nd9", nd16' and nd18 are unable to assemble rosettes, as judged by freeze fracture electron microscopy (Table 4, Figure 1) and it is concluded that all three gene products are necessary for rosette assembly. At 18', a permissive temperature forthese single mutants, they display sub-normal rosettes (Table 5 , Figure 2). However, no rosettes could be detected in the double mutants at 18". We conclude that the defect produced by the interaction of these genes bears on the mechanism of rosette assembly. Our arguments for concluding that the two newly identified genes alongwith nd9 form a groupof three interacting genes are as follows. Two alternative hypotheses can be put forward to explain that a double mutant has a worse phenotype than each of the single mutants. Hypothesis 1: To have a wild-type phenotype, a critical threshold of some substance(s) is required.Thermosensitivemutantsproduce this substance at a level below this threshold at the restrictive temperature and above it at the permissive one, without this level necessarily being as high as in wild-type cells. Double mutants carrying two mutations of this kind could present a phenotype more affected than each of the single mutants, if the level of substance synthesized becomes lower than the threshold. However, this hypothesis implies that the two confronted mutations affect independent pathways for producing this substance, giving a common phenotype. Hypothesis 2:The aggravation of the phenotype given by the association of two mutations in a single cell reflects the direct interactions between the products of the genes. In this case, the genes concerned would affect a common stepin the pathway yielding the phenotype. In ourcase, some genetic combinations show epistasis of one gene over the other, which indicates that the genes associated encode productsinvolved in the same pathway. Moreover, different levels of negative interactions are observed (including epistasis and even an absence of interaction in one case), dependingon which alleles of the combined genes are confronted. Inthecontext of the first hypothesis, this would suppose different activity thresholds for thesubstance according tothe genic combination in the double mutants which does not seem probable. We favor the simpler and more likely hypothesis that the products of the three genes nd9, nd16 and nd18 physically interact andare involved in the same process of rosette assembly and exocytotic capacity. In addition, aberrant dominance relationships between some alleles were found for both genes nd9 and nd16. This indicates that the gene products belongto
Mutants Interacting Exocytotic
multimeric structures built up through interactions betweenidentical monomers. In microinjection experiments (Table 6), the nd16 gene product was shown to be present in the cytosol, as previously shown forthe nd9 gene product. Altogether, these data strongly suggest that the products of the three genes nd9, nd16 and nd18 act in the same cytosolic multimeric structure in which they undergo two types of interactions: interactions between subunits encoded by the same gene and interactions between subunits encoded by different genes. It has been previously supposed that the nd9 gene product was a component of the connecting material (BEISSON et al. 1980). It is therefore tempting to extend this hypothesis by suggesting that the products
469
470
H.Bonnemain et al.
NAKATA, T., K. SOBUEand N. HIROKAWA, 1990 Conformational change and localizationof calpactin I complex involved in exocytosis as revealed by quick-freeze, deepetch electron microscopy and immunocytochemistry.J. Cell Biol. 110 13-25. NOVICK,P., C.FIELD and R. SCHEKMAN, 1980 Identification of 23 complementation groupsrequiredfor post-translational events in the yeast secretory pathway. Cell 21: 205-215. NYBERG, D., 1978 Genetic analysis of trichocyst discharge of the wild stocks of Paramecium tetraureliu. J. Protozool. 25: 107112. ORCI,L., A. PERRELET and D. S. FRIEND,1977 Freeze-fracture of membrane fusions during exocytosis in pancreatic B-cells. J. Cell Biol. 7 5 23-30. ORIAS,E., M . FLACKS and B. H. SATIR,1983 Isolation and ultrastructural characterization of secretory mutants of Tetrahymena thermophila. J. Cell Sci. 64: 49-67. PALADE, H., 1975 Intracellular aspects of the process of protein synthesis. Science 1 8 9 347-358. PINTODA SILVA,P., and M. L. NOGUEIRA, 1977 Membrane fusion during secretion. A hypothesis based on electron microscope observation of Phytophthora palmivora zoospores during encystment. J. Cell Biol. 73: 161-181. PLATTNER, H., 1987 Synchronous exocytosis in Paramecium cells, pp. 69-98 in Cell .Fusion, edited by A. E. SOWERS.Plenum, New York. PLATTNER,H., 1989 Regulation of membrane fusion during exocytosis. Int. Rev. Cytol. 119: 197-286. PLATTNER, H., F. MILLER and L. BACHMANN, 1973 Membrane specializationsin the form of regular membrane-to-membrane attachment sites in Paramecium. A correlated freeze-etching and ultrathin-sectioning analysis.J. Cell Sci. 13: 687-719. PLATTNER, H., K. REICHEL,H. MATT,J. BEISSON,M. POUPHILE and M . LEFORT-TRAN, 1980 Genetic dissection of the final exocytosis steps in Paramecium cells: cytochemical localization
of Ca++ATPase activity over preformedexocytosissites. J. Cell Sci. 4 6 17-40. POLLACK, S., 1974 Mutations affecting the trichocysts in Paramecium aureliu. I. Morphology and description of the mutants. J. Protozool. 21: 352-362. POUPHILE, M., M. LEFORT-TRAN, H. PLATTNER, M. ROSSIGNOL and J. BEISSON, 1986 Genetic dissection of the morphogenesis of exocytosis sites inParamecium. Biol. Cell 5 6 151-162. RUIZ,F.,A. ADOWTE,M. ROSSIGNOL and J. BEISSON, 1976 Genetic analysis of morphogenetic processes in Paramecium. I. A mutation affecting trichocyst formation and nuclear division. Genet. Res. 27: 109-122. SCHEKMAN, R., 1985 Protein localization and membrane traffic in yeast. Annu. Rev. Cell Biol. 1: 115-143. SONNEBORN, T. M., 1970 Methods in Paramecium research. Methods Cell Physiol. 4 24 1-339. SONNEBORN, T. M.,1974 Paramecium aurelia, pp. 469-594 in Handbook of Genetics, edited by R. KING. Plenum, New York. SONNEBORN, T. M., 1975 The Paramecium aurelia complex of 14 sibling species. Trans. Am. Microsc. SOC.94: 155-178. TANAKA, Y.,P. DE CAMILLI and J. MELDOLESI, 1980 Membrane interactions between secretion granules and plasmalemma in three exocrine glands. J. Cell Biol. 84: 438-453. TURKEWITZ, A. P., L. MADEDDU and R. B. KELLY, 1991 Maturation of dense core granules inwild type and mutant Tetrahymena termophila. EMBO J. 10: 1979-1987. VILMART, J., and H. PLAITNER,1983 Membrane-integrated proteins at preformed exocytosis sites. J. Histochem. Cytochem. 31: 626-632. ZIFSENISS, E., and H. PLATTNER,1985 Synchronous exocytosis in Paramecium cells involves very rapid ( ~ l s )reversible , dephosphorylation of a 65-kD phosphoprotein in exocytosis-competent strains. J. Cell Biol. 101: 2028-2035. Communicating editor: S. L. ALLEN
