Copyright 0 1988 by the Genetics Society of America

Developmental Genetics of ChromosomeZ Spermatogenesis-Defective Mutants in the Nematode Caenorhabditis elegans

Steven W. L’Hernault,’ Diane C. Shakes2and Samuel Ward’ Department of Embryology, Carnegie Institution of Washington, Baltimore, Maryland 21210 Manuscript received December 30, 1987 Revised copy accepted July 1, 1988

ABSTRACT Mutations affecting Caenorhabditis elegans spermatogenesis can be used to dissect the processes of meiosis and spermatozoan morphological maturation. We have obtained 23 new 1 mutations that affect spermatogenesis (spe mutations). These mutations, together with six previously described mutations, identify 11 complementation groups, of which six are defined by multiple alleles. These spe mutations are allrecessive and causenormally self-fertile hermaphrodites to produce unfertilized oocytes that can be fertilized by wild-type male sperm. Five chromosome I mutation/ deficiency heterozygotes have similar phenotypes to the homozygote showing that the probable null phenotype of these is defective sperm. Spermatogenesis is disrupted at different steps by mutations in these genes. The maturation of 1 spermatocytes is disrupted by mutations in spe-4 and spe-5. Spermatids from spe-8 and spe-12 mutants develop into normal spermatozoa in males, but not in hermaphrodites. fer-6 spermatids are abnormal, and fer-1 spermatids look normal but subsequently become abnormal spermatozoa. Mutations in five genes (fer-7, spe-9, spe-11, spe-13 and spe-15) allow formation of normal looking motile spermatozoa thatappear to be defective in either sperm- spermatheca1 or sperm-oocyte interactions.

PERMATOGENESIS is a complexprocess during duringthe L4 (last) larval stage and continues S which undifferentiated germ cells develop into throughout the life of the animal (KLASS,WOLF and mstile spermatozoa. Dramatic cytological changes oc- HIRSH 1976). The hermaphroditegonad produces cur as spermatocytes undergo the two meiotic divi- sperm only during the L4 stage, and these sperm are sions that precede formation of the haploid gamete. stored until they fertilizeone of the oocytes produced Mutations that affect spermatogenesis exist in a num- by the adult gonad (HIRSH, OPPENHEIMand KLASS ber of organisms, includingman (e.g., AFZELIUS1985), 1976). Sperm introduced into a hermaphrodite dur- but genetic analysis of this process is limited to a few ing copulation with a male will outcompeteher- organisms such as mice (reviewed by HANDEL1987), maphrodite-producedsperm and outcross progeny Drosophila (reviewed by HACKSTEIN1987 and LIF- will result (WARDand CARREL1979). Fertilization SCHYTZ 1987) and the nematode Caenorhabditis ele- with either male or hermaphrodite sperm occurs in- guns (HIRSH and VANDERSLICE1976; WARD and side the hermaphrodite, and embryonated eggs are MIWA1978; ARGONand WARD 1980; EDGAR1982). laid 8-10 hr after fertilization (HIRSH,OPPENHEIM The hermaphroditic mode of reproduction in C. ele- and KLASS1976). guns makes it especially well-suited for geneticanalysis Many mutations that eliminate spermatogenesis or of spermatogenesis. Screening for defective sperm in cause production of defective sperm will cause adult normally self-fertile hermaphrodites avoids selection hermaphrodites to lay oocytes rather than embryo- of mutants in male mating behavior,but permits iden- nated eggs. Hermaphrodites can be scored for this tification of animals that will produce progeny when phenotype several hours after they have been picked mated to wild-type males. simply by inspecting plates for the presenceof oocytes Sperm are produced by both the C. elegans male, under the dissecting microscope. Such self-sterile mu- which has one sex chromosome (XO), and the her- tant hermaphrodites are spe (spermatogenesis-defec- maphrodite, which has two sex (XX) tive) if their oocytes can be fertilized by mutant sperm (NIGON 1949).Spermatogenesis begins in the male inseminated by a heterozygous wild-type male (SIG- URDSON, SPANIERand HERMAN1984). To date, more ’ To whom correspondence should be addressed. Present address: De- than 60spe mutations have been recovered by various partment of Biology, Emory University, 1555 Pierce Drive, Atlanta, Georgia 303222. laboratories (e.g., HIRSH and VANDERSLICE1976; Present address: Department of Molecular and Cellular Biology, Biosci- WARD and MIWA 1978; ARGONand WARD 1980; ences West Building, University of Arizona, Tucson, Arizona 85721. ’ Present address: Section of Genetics and Development, 220 Bradfield EDGAR1982; BURKE1983; SIGURDSON, SPANIERand Hall, Cornell University, Ithaca, New York 14853. HERMAN1984; this paper and unpublished observa-

Genetics 120 435-452 (October, 1988) 436 S. W. L'Hernault, D. C. Shakes and S. Ward tions). A subset of previously studied spe mutants that unpublished results) and nDf 23, nDf 24 and nDf 25 (FERGU- made haploid sperm in normal numbers were called SON and HORVITZ1985) and the free duplication sDp2 (ROSE,BAILLIE, and CURRAN 1984; HOWELLet al. 1987) fer (forfertilization-defective)to distinguish them from were also employed.Culturing, handling and genetic manip- premeiotic mutants in several previous studies (e.g., ulations of C. elegans were performed as described (BRENNER WARDand MIWA 1978; ARGONand WARD 1980). 1974), and standard nomenclature is used (HORVITZet al. This distinction is nolonger being made, and the 1979). The entire broods of cis (++/ab) heterozygotes, designation spe is now used for all new mutations that reared at 20 " , were analyzed during two factor recombina- tion mapping (ROSE and BAILLIE1979), and symmetric 95% interfere with spermatogenesis whether they act pre- confidence limits were tabulated (MAINLAND,HERRERA and or postmeiotically (mutations previously namedfer will SUTCLIFFE1956). Nonconditional chromosome I spe muta- retain their names). tions were balanced either by the free duplication sDp2 or Feminization (fern; NELSON,LEW and WARD1978; to complementing chromosome I deficiencies. Experiments DONIACHand HODGKIN1984; KIMBLE,EDGAR and were performed at 16", 20" or 25", as noted. Screen for new chromosome Z sperm-defective steriles: HIRSH1984; HODGKIN1986) or feminization of the Ethyl methanesulfonate (EMS)was used asa mutagen (BREN- germline (fog; SCHEDLand KIMBLE1988; M. K. BAR- NER 1974) for generating all the spe mutations used in this TON and J. KIMBLE,unpublished observations) muta- study. We have used four methods to identify picked mutant tions can also cause hermaphrodites to stop laying hermaphrodites that lay oocytes but produce progeny if embryonated eggs and start laying oocytes but, unlike mated to wild-type males. The first two methods are essen- tially equivalent to the M and S sets of BRENNER(1974). spe mutants, fern and fog mutant hermaphrodites do Mutant hermaphrodites that laidoocytes were mated to not produce sperm. The germline of fern and fog X0 wild-typemales and, if outcross progeny resulted, were animals (which in wild type makes only sperm) makes considered spe candidates. These candidates were out- either sperm and oocytes or only oocytes, and the crossed once more and assigned to achromosome by crosses soma of X0fern animals (which in wild type is always to two triply marked strains that have a marker on each linkage group (TRENT,TSUNG and HORVITZ1983). Our male) can behermaphrodite. Consequently, these triply marked strain for chromosomes I-111 uses the original non-spe hermaphrodites that are recovered asoocyte- markers for chromosome I (dpy-5) and 111 (unc-32) but laying animals during spe mutant hunts can be easily replaces bli-2 11 with rol-I II; bli-2 is very sick at 25" and identified. cannot be used to score ts spe mutants. Only spe mutations We would like to identify all genes that confer a spe located on chromosome I were examined further during the course of this study. phenotype, and approximately 36 genes on all six C. The other two methods used to screen for spe mutants elegans chromosomes that conferthis phenotype have, involved mutagenesis of morphologically marked strains to thus far, been discovered (HIRSHand VANDERSLICE permit recovery of mutations linked to easily scored mark- 1976; WARD and MIWA 1978; ARGONand WARD ers. Young hermaphrodites that carried dpy-5; rol-I; unc-32 1980; EDGAR1982; BURKE1983; SIGURDSON, SPAN- were mutagenized in the first of these methods, outcrossed to wild-type males and 2-12 F1 progeny were picked from IER and HERMAN1984; our unpublished observa- each Po to individual plates. These F1 progeny were allowed tions). This paper describes our genetic analysis of to lay eggs forone day at 20" after which they were spermatogenesis genes on chromosome I;we confined removed, and the plate of eggs was shifted to 25" so poten- our efforts to a single chromosome in order to obtain tial ts spe mutants would not be missed. When an F1 is spe-/ multiple alleles of spe genes and to allow the use of +, about % of its F2 progeny are spelspe; young non-Spe hermaphrodites lay only embryonated eggs while Spe her- genetic strategies for recovery and maintenance of maphrodites lay large numbers of oocytes. Platescontaining nonconditional spe mutations. We have found 11 large numbers of oocytes were identified 2 days after shift- chromosome Z spe genes, six withmore than oneallele, ing to 25", and six hermaphrodites homozygous for each andboth temperature sensitive (ts) and noncondi- one of the three morphological mutations were picked off tional mutations have been recovered. each of these candidate plates. The plates containing these picked hermaphrodites were inspected one day later and, if oocytes were present, the spe mutation was recovered by mating these Spe hermaphrodites to wild-type males. IfSpe MATERIALS AND METHODS hermaphrodites were predominantly of one morphological type, this suggested linkage of the spe mutation tothat Strains, culture conditions and genetic nomenclature: marked chromosome. We also recovered spe mutations out C. elegans var. Bristol (strain N2) was the wild-type strain of mutagenized dpy-5 hermaphrodites by essentially the usedin all experiments (BRENNER1974). The following same methods. Only mutations derived from different mu- genes and mutations were used in this study: dpy-5(e61) I, agenized Po hermaphrodites were kept in order to ensure unc-29(e1072)I, unc-II(e47) I, unc-I3(e51, e450, e1091)I, that mutations were of independent origin. All mutations unc-I5(e73) I, dpy-I4(e188) I, bli-4(e937) I, rol-l(e91) 11, were outcrossed at least twice to wild type, and only those unc-32(e189) III (BRENNER1974); hirn-5(e1490) V (HODG- that exhibited linkage to dpy-5 were examined in this study. KIN,HORVITZ and BRENNER 1979); lin-II(n566) I, lin- Complementation: Initially, all mutations were tested for 17(n671) I (FERGUSONand HORVITZ1985); tra-I(e1099) IZI complementation to one another until allelic series and map (HODGKIN and BRENNER 1977);eDp6 III (HODGKIN 1980); positionswere established. Construction of spedpy-5 cis sup-7(st5) X (WATERSTON1980). The chromosome I defi- double mutants made complementation tests of noncondi- ciencies sDf4 (ROSE 1980), sDf5,sDf6 (ROSEand BAILLIE tional spe mutations easier to perform because dpy trans 1980), hDf6 (K. S. MCKIM,A. M. HOWELLand A. M. ROSE, heterozygotes (spe-a dpy-5/spe-b dpy-5) could be easily iden- Spermatogenesisin C. elegans 437 tified. Later in the analysis,two factor mapping, duplication suggests that at least 42 of the 48 known map units mapping and phenotypic data were obtained for new spe on chromosome I have beensampled for spe mutations mutations, and these suggested appropriate complementa- tion tests. Mutations that did not fall into previously identi- in our study. Our collection includes both tempera- fied genes were three factor mapped and complementation ture-sensitive and nonconditionalmutations. Many tested to all spe genes within the same map interval. nonconditional spe mutations were found to lie under Suppression studies: All nonconditional chromosome I the free duplication sDp2 (ROSE, BAILLIEand CUR- spe mutations were tested for sup-7(st5) X suppressibility. REN 1984; HOWELLet al. 1987), which could be used Males that were unc-Ijr(e450) /+ I; sup-7(st5) X were mated to spe dpy-5(e61) hermaphrodites. The F2 broodswere to balance the sterile mutant allele, usually on a dpy-5 scored for the presence of self-fertile dpy-5 hermaphrodites markedchromosome. These duplication-bearing (with correction for crossing over betweendpy-5 and thespe strains are stable, and they permit both strain main- mutation). If necessary, tentativespe dpy-5 I; sup-7 X candi- tenanceand geneticmanipulation. Nonconditional dates were picked and outcrossed towild type to check for steriles that do not lie under sDp2 are maintained as segregation of the oocyte-laying phenotype in the F2. No chromosome I spe mutations were suppressible. heterozygotes to a complementing chromosomeI de- Phenotypic analysis: Brood and/or oocyte counts were ficiency. Genes are positioned on the chromosome I performed on individual hermaphrodites that were picked map (Figure 1) by two and three factor, duplication and transferred either every day (if at 25") or every second and deficiency mappingand the results appear in day (if at 16") until they stopped laying or died. Tempera- ture sensitive periods were determined as previously de- Tables 1, 2 and 3. scribed (HIRSHand VANDERSLICE1976). Male sperm devel- Description of wild-type spermatogenesis: Wild- opment and morphology were examined in vitro after dis- type spermatogenesis in C. elegans has been described section in sperm medium (SM): 50 mM HEPES titrated to in detail previously (reviewed by WARD1986). These pH 7.0 or 7.8 with NaOH, 50 mM NaCI, 25 mM KC], 5 mM results are summarized in diagrammatic form in Fig- CaC12, 1 mM MgS04, plus 10 mg/ml polyvinylpyrrolidone, PVP 40 (NELSON1979; NELSONand WARD1980; NELSON, ure 2 and explained here in order to make our de- ROBERTSand WARD 1982). SMallows primary spermato- scriptionsof mutant phenotypesunderstandable. cytes to develop into haploid spermatids. Some spermatids Spermatogenesisoccurs similarly inboth the her- were treated in vitro with the sperm activator pronase so maphrodite andmale. Primary spermatocytes develop that both their differentiation and the motilityof resulting from spermatogoniaas a syncytium with a central core spermatozoa couldbe examined (NELSONand WARD1980; WARD,HOGAN and NELSON1983). Sperm development and of cytoplasm, the rachis, connecting peripheral nuclei. motility in hermaphrodites were studied by examining the Primary spermatocytes (which are 4N) separate from spermathecae of live worms dissected in SM medium. Nu- the rachis afterentering meiosis, gothrough one clear events werestudied by either fixing worms in Carnoy's division to give rise to secondary spermatocytes(which solution and staining the DNA with DAPI (4,6-diamidino- 2-phenylindol) or Hoechst 33258, or dissecting worms in are 2N) and a second division to give rise to four SM containing lipid-soluble Hoechst 33342. Light and elec- haploid (1N) spermatids.Cellular components not tron microscopic techniques were essentially as previously needed by the mature sperm, such as ribosomes, are described (WARD,ARGON and NELSON 1981). partitioned into theresidual body before the sperma- Many of our phenotypic observations were performed tids separate. The hermaphroditegonad produces before our spe mutationswere properly balanced (see above). This was possible because linkage of a spe mutation sperm only during the L4 (last) larval stage and then to dpy-5 permits identificationof probable spe dpy-5mutants switches to produce only oocytes in the adult. Sper- in mixed populations. Phenotypes that were initiallydeter- matids made during the L4 stage are activated, be- mined by examining spe dpy-5 animals have all been con- come crawling spermatozoa (Figure2b) and are stored firmed by subsequent analysis ofspe non-dpy-5 animals. in a chamber called the spermatheca where they fer- tilize incoming oocytes. RESULTS Male spermatogenesis, asin the hermaphrodite,also Isolation of mutations and genetic analysis: We begins during the L4 stage but, unlike the herma- have identified 23 new, independently derived chro- phrodite, continues throughout the adult life of the mosome I mutations that result in abnormal sperma- worm.Conversion of male-produced spermatids to togenesis, in addition to the six chromosome I muta- amoeboid spermatozoa (termed either sperm activa- tions previously described (ARGONand WARD1980). tion or spermiogenesis,as it is in hermaphrodites) We used four different strategies to recover these spe occurs during copulation with hermaphrodites. These mutations (see MATERIALS AND METHODS),and all mu- male-derived spermatozoa crawl throughthe her- tationsproved tobe recessive. The most efficient maphrodite uterus to reach the spermatheca, where strategy was isolation of mutations incis to dpy-5 since they fertilize many oocytes because they displace her- this established linkage to chromosomeI and allowed maphrodite-produced spermatozoa (WARDand CAR- convenientidentification of sterile hermaphrodites REL 1979). Male spermatids canalso be converted into during outcrossing and complementation testing. spe spermatozoa in vitro by treatment with a protease, genes thatare distant from dpy-5, suchas spe-13, weak bases such as triethanolamine or the ionophore (about 2 1 map units; Table 1) still exhibit linkage; this monensin (NELSONand WARD 1980; WARD,HOGAN 438 S. W. L’Hernault, D. C. andShakes S. Ward

TABLE 1 Two-factor mapping in cis”

Genotype of het- Percent recombination erozygote ParentalRecombinanttypes types (100 P)b

spe-4 spe-4 dpy-5/++ spe-4 spe-4 222 WtC Spe 4 0.63-3.8 78 Dpy Spe 1 DPY spe-8 spe-8 spe-8 lin-l7/++ 223 wt 13 Spe 5.7- 13 65 Lin Spe 13 Lin ~pe-9~ dpy-5/++ spe-9 234 Dpy Spe 36 DPY 5.4-8.7 spe- 1 1 spe-I 1 dpy-5/++ 1 spe-I 1 1 spe- 1352 wt 1 Spe 0.06 460 Dpy Spe 0 DPY

12 spe-12 dpy-5/++spe-12 spe- 12 493 Wt 14 Spe 2.8-6.1 138 Dpy Spe 13 DPY 13 spe-13 dpy-5/++spe-13 spe- 13 639 wt 92 Spe 18-24 144 Dpy Spe 90 DPY spe- 15 spe-15 dpy-5/++spe-15 15 spe- 512 wt 60 Spe 14-2 1 11 1 Dpy56 Spe DPY fer-6 fer-6 dpy-5/++ fer-6 fer-6 665 wt 32 Spe 6.8-1 1 183 Dpy Spe 44 DPY

a Two-factor crosses were performed as described by BRENNER(1 974) for markers in cis. The 95%confidence values limits on two-factor data are listed, except for spe-12. ‘ wt = wild type. Only dpy animals were scored. and NELSON1983). Resulting spermatozoa, whether to the number observed in a wild-type male (data not produced in vivo or in vitro, use their single pseudopod shown). to crawl across the substrate. Sperm motility is neces- The spe-4 adult hermaphrodites, unlike males, do sary for in vivo fertility because spermatozoa can be not retain thedevelopmentally arrested spermatocytes swept out of the spermatheca and into the uterus by that they form. Some apparently haploid sperm nuclei passing oocytes. These spermatozoa must crawl out of appear in L4 spe-4 larvae during the normal time of the uterus and back into the spermatheca or they will spermatogenesis (data not shown), but they must be eventually be pushedout of the hermaphrodite repro- lost or resorbed because no sperm or haploid nuclei ductive tract (WARDand CARREL19’79). Unlike most are observed after DAPI staining of the adult repro- animal sperm, C. eleguns sperm do not have a flagel- ductivetract of hermaphrodites(Figure 4a). spe-4 lum or an acrosome. oocytes pass through the spermatheca, resumemeiosis Phenotypes of mutations affecting meiosis: spe- and eventually become polyploid. These polyploid 4(hc78, hc81)”Spermatogenesis arrests in spe-4 males oocytes resemble the oocytes in old wild-type her- during the formationof haploid spermatids from ma- maphrodites that have exhausted their sperm supply ture primary spermatocytes. The dissected testis of (WARDand CARREL1979) or the oocytes of later- wild-type virgin adult males reveals spermatocytes de- acting spe mutants such as fer-1 (WARDand MIWA veloping in syncytium along the rachis, a few free 1978). spe-4 bearing worms lay fewer of these oocytes spermatocytes (not joined to therachis) that are com- than wild type or many of the later acting mutants pleting the nuclear and cytokinesis events of meiosis (Table 4). and hundreds of accumulated spermatids (like Figure spe-5(hc93, hc1lO)”Spermatogenesis in spe-5 males, 3a). In contrast, spe-4 males produce a few cells that as in spe-4 males, arrests during developmentof sper- appear to be in the process of completing meiosis but matids from spermatocytes. spe-5 males never accu- fail to produce any spermatids (Figure 3b);spe-4/sDf6 mulatelarge numbers of spermatids,but they do males exhibit the same phenotype (data not shown). produce some free sperm cells that consist of a mix- Some of these freecells will form thetypical cloverleaf ture of primary and secondary spermatocytes as well structures of syncytial spermatid/residual body clus- as a few spermatids (Figure 5, a and b). Many of these ters (Figures 2a and 3c), but most look like primary free sperm cells contain haploid nuclei; the number spermatocytes (Figure 3c). There “spermatocytes”dif- of haploid nuclei generally correlates with the cyto- fer from wild type because they have completed the logical stage of arrest (compare Figure 5, b and c). nuclear events of meiosisand have four haploid nuclei Although spe-5males producea large number of in syncytium (Figure 3d).The total numberof haploid haploid nuclei, diploid nuclei and disorganized nu- nuclei visualized ina DAPI stained spe-4 male is similar clear divisions are also observed in DAPI stained Spermatogenesis in C. elegans 439

H IH I I SDf4 fnDf23 eDf3 sDf6 I I hDf6 eDf6

b .5 map unit'

L

I I nDf23 'sDf5 ' I-. I-. .. sDf6 - FIGURE1 .-a, A partial genetic map of C. elegans chromosome I. Genes that cause spermatogenesis defects (spe) are drawn above the line that represents chromosome I, and morphological and other markers are drawn below the line (the positions of these markers are from EDGLEYand RIDDLE(1 988). Ninety-five percent confidence intervals of spe gene two factor mapping data are indicated by closed circles (see MATERIALS AND METHODS and Table 2). The extent of deficiencies (Dfs)and duplications (Dp's) are indicated below the lines. b, Expanded genetic map of C. elegans showing the region of chromosome I extending from unc-15 to unc-29. whole mounts of spe-5 males (data not shown). This spe-4 hermaphrodites; Figure 4a). spe-5 hermaphrod- variable arrest of spermatogenesis has been observed ites lay fewer of these oocytesthan wild type or many for spe-5(hc93), spe-5 (hcl IO), spe-5(hc93/hclIO) and of the later acting spermatogenesis defective mutants spe-5(hc93/sDf4). (Table 4). spe-5 adult hermaphrodites, unlikemales, do not Postmeiotically acting spe mutations: spe-8(hc40, retain the developmentally arrested spermatocytes hc50, hc53,hc79, hc85, hc108, hcl34ts) and spe- that they form. Some apparently haploid sperm nuclei 12(hc76)-Mutations in spe-8 and spe-12 cause indis- appear in L4 larvae during the normal time of sper- tinguishablephenotypes so our data on thesetwo matogenesis, but they must be resorbed or lost be- genes will be discussed together. The spe-8 gene is cause few,if any, sperm or haploid nucleiare observed presently defined by seven nonconditional alleles. spe- after DAPI staining of the reproductive tract of young 8 mutant hermaphrodites are completely self-sterile, adult hermaphrodites (data not shown, but similar to with the exception of hc134ts hermaphrodites, which spe-4 in Figure 4a). spe-5(hc93)hermaphrodites some- produce a few progeny at 16" (Table 4 and data not times produce one or two progeny indicating that shown). There is one allele of spe-12,and mutant spe- fertile spermatozoa must develop occasionally (Table 12 hermaphrodites rarely produce any young (Table 4). Most oocytes produced by spe-5 hermaphrodites 4); this is also true for spe-I2/nDf24 hermaphrodites do not get fertilized and become polyploid (similar to (data not shown). 440 S. W. L'Hernault, D. C. andShakes S. Ward TABLE 2 produced. Thus, 96-8 or spe-12 males are normally Three-factor crossesa fertile while unmated spe-8 or spe-12 hermaphrodites are self-sterile. Combining spe-8 males with spe-8 her- Genotype of se- maphrodites or spe-12 males with spe-12 hermaphro- Phenotypelected hermaphro- Genotype of of selected dite (withrespect to ditescreates a stable male/"female" gonochoristic Gene heterozygote hermaphrodite unselected marker) strain. spe-5spe-5/dpy-5 unc-11 Dpy non-Unc3/20 spe-5/+ C. elegansmales differfrom hermaphrodites be- Uncnon-Dpy 5/2 1 spe-5/+ cause they have only one X chromosome instead of spe-8spe-8/lin-l7 dpy-5 Dpy non-Lin12/12 spe-8/+ two (NIGON1949). In order to test whether the dif- Linnon-Dpy 011 2 spe-8/+ fering fertility of spe-8 and spe-12 males and her- spe-9spe-9/unc-75 dpy-5 Dpy non-Unc 57/57 spe-9/+ maphrodites was a result of X chromosome dosage, Unc non-Dpy 0/44 spe-9/+ we have used the transformer gene tra-1 to construct spe-11 spe-ll/unc-13 dpy-5 Dpy non-Unc0/22 spe-Ill+ strains that have fertile (albeit, at reduced levels rela- Unc non-Dpy24/24 spe-I I/+ tive to wild type) phenotypic males with a XX genotype spe-12spe-12/unc-29 dpy-5 Dpy non-Unc 10111 spe-12/+ (HODGKINand BRENNER 1977).The fertility of these Unc non-Dpy 1/ 1 1 spe-I 2/+ males was assessed by the mating scheme outlined in spe-13spe-13/dpy-5 unc-11 Dpy non-Unc18/18 spe-l3/+ Figure 6, and the results indicate that spe-8 and spe- Uncnon-Dpy 0/12 spe-l3/+ 12 phenotypic males produce functional spermatozoa, spe-15spe-l5/dpy-5 lin-17 Dpy non-lin21/21 spe-l5/+ even if they have the XX genotype of a hermaphro- Linnon-Dpy 0/16 spe-15/+ dite. fer-6fer-6/dpy-5 unc-75 Dpy non-Unc 18/18fer-6/+ spe-9 (hc52ts, hc88ts)-Unmated spe-9 males, grown Unc non-Dpy 011 8 fer-6/+ at therestrictive temperature (25O), accumulate large a Three factors crosses were performedas described by BRENNER numbers of normal-appearing spermatids that activate (1974). FI heterozygotes of genotype spelab will segregate A non- in vitro following pronase treatment to give cytologi- B and B non-A FP recombinants, which were picked and scored for cally normal, crawling spermatozoa (data not shown). expression of a Spe phenotypein 114 of the Fa self-progeny. spe-9(hc88ts) males are apparently infertile at restric- TABLE 3 tive temperatures; eight matings of four spe-9(hc88ts) Deficiency and duplication mappingof spe genes" males and onedpy-5 hermaphrodite failed to yield any outcross progeny. spe-9 hermaphrodites (either spe-9 Gene sDp2 hDf6 sDf4 SOP sDf6 nDf25nDf24 nDf23 eDf3 eDf6 allele), raised under restrictive conditions,contain spe-4 + +o + + spermatozoa that are motile in the spermatheca and spe-5 0 + 0 crawl in vitro (data not shown), but are slowly swept spe-8 0 out of the spermatheca by passing oocytes (Table 5). spe-9 + ++ spe-11 0 + 0 spe-9(hc88ts) hermaphroditesare completely self- spe-12 + ++o 0 + sterile whilespe-9(hc52ts) hermaphrodites can still spe-15 0 producea few young under restrictive conditions fer-I + +o 0 0 (Table 4). fer4 + + + The timeperiod during development when spe- * + indicates complementation,0 indicates noncomplementation; 9(hc88ts) is temperature sensitive for fertility (the no symbol indicates the testwas not performed. TSP) has been determined by growing mutant her- spe-8 or spe-12 adult hermaphrodites contain sper- maphrodites at 16" (the permissive temperature) or matids that fail to differentiate into spermatozoa. 25" (the restrictive temperature) and shifting worms These nonmotile spermatids are rapidly swept out of of various ages to the other temperature. The total the spermatheca (Table 5) into the uterus by passing number of progeny produced by animals that were oocytes (Figure4b); a similar phenotype has been shifted from one temperature to the other was then described previously for fer mutants, which have sper- determined and is summarized in Figure 7. The TSP matids that form abnormal spermatozoa (ARGON and for the sterile phenotype is between 30 and 45 hr of WARD 1980). Unmated males bearing spe-12 (either postembryonic development; this is duringthe L4 spe-12/spe-I2 or spe-I2/nDf 24; data not shown) or larval period, when spermatogenesis occurs in the any of the fourspe-8 alleles, hc40, hc50, hc79or hcl08, wild-type hermaphrodite(HIRSH, OPPENHEIM and accumulatenormal looking spermatids in normal KLASS 1976). numbers. Surprisingly, mating induces maturation of spe-11(hc77ts, hc90)-Hermaphrodites bearing the spe-8 and spe-12 male spermatidsinto cytologically temperature sensitive allele of spe-1I, (hc77), produce normal spermatozoa. These male-derived spermato- a few progeny under restrictive growthconditions zoa are fertile in hermaphrodites of the same or while mutant hermaphrodites bearing the otherallele different genotype and normal outcross progeny are (hc90) are nonconditional and completely self-sterile Spermatogenesis in C. eleguns 44 1

FIGURE 2,"Summary of C. ele- gam spermatogenesis. a, Stages of meiosis. Primary spermatocytes de- rachis velop while in syncytium with the rachis. Matureprimary spermato- Prim ary SecondaryPrimary c Spermatocytesspermatocytes Spermatid cytes bud off the rachis, undergo a first meiotic division and resulting ""..". secondary spermatocytes can sepa- A- + 0 *:::::iijii::j. rate or stayin syncytium; only the ""."_""."".. latter is illustrated. Four haploid spermatids subsequently develop 2N nucleus from these syncytial secondary sper- 4N nucleus Residual matocytes during the second meiotic IN nucleus Body division. Much of the cytoplasm pre- sent in the secondary spermatocytes SpermatidSpermatozoa ends up in the residual body during the cytoplasmic volume reduction surface projection that accompanies spermatid differ- B- @ entiation. b, Postmeiotic stages. The -v sessile spermatid, lacking any obvious cellular polarity, is converted into an Pseudopod actively crawling spermatozoan with a polarity that is indicated by its sin- gle projection-studded pseudopod. (Table 4). The trans heterozygote hc77/hc90 is also produce a number of round (rather than the normal leaky, producing small broods of approximately the oblong) thinly shelled eggs, in addition to many un- same size as hc77 homozygotes (Table 4). Brood sizes shelled oocytesand a few apparently normal embryos of hc77/sDf4, however, are lower than hc77 homozy- (Table 4). These thinly shelled eggs, unlikethe normal gotes (Table 4). The hc90 hermaphrodites never pro- embryos, do not hatch. duce any progeny, and the phenotype of hc90/sDf4 Unmated spe-I 1 males accumulate spermatids that animals is the same as hc90 homozygotes. are cytologicallywild type and canbe induced by The few progeny produced by spe-ll(hc77ts) her- pronase treatment to differentiate in vitro into sper- maphrodites at 25" are produced early in the egg- matozoa that propel themselves across the substrate laying period. Preliminary temperature shift experi- normally (data not shown). The fertility ofspe- ments indicate that spe-I I(hc77ts) hermaphrodites can 1I(hc77ts) him-5 males was assessed by mating them produce more progeny if they are shifted to permis- in excess to dpy-5 hermaphrodites. A few of the total sive conditions as young adults after spermatogenesis young present after three days of mating were out- has been completed. This demonstrates that mature cross phenotypic wildtypes (13/472 or about 3%) spe-I I spermatozoa can reverse their sterile phenotype indicating that spe-I I(hc77ts)males, like spe-II(hc77ts) after shift-down to the permissive temperature, and hermaphrodites, are only slightlyfertile under restric- suggests that the spe-I I gene product is itself temper- tive conditions. ature reversible, since mature C. elegans sperm do not spe-13(hcl37ts)-spe-l3males grown at 25" pro- synthesize (they lack ribosomes: reviewed by duce normal numbers of spermatids that activate in WARD1986). vitro after pronase treatment to form normal-appear- Sperm are swept out of the spermatheca very slowly ing spermatozoa. Electron microscopic analysis ofun- by passingoocytes in spe-1 I(hc90) or restrictively matedspe-13 males reveals no obviouscytological raised spe-I I(hc77ts) hermaphrodites (Figure 4c and defects in spermatocytes or spermatids (data not Table 5), and motile spermatozoa are observed in the shown). spe-I3 malesgrown at 25" are apparently spermatheca of live hermaphrodites (data not shown). infertile, even though their spermatozoa appear cy- spe-I I oocytes sometimesundergo a few cell divisions tologically normal, since they fail to sire progeny in (either allele; Figure 4c). This oocyte phenotype ap- mating experiments (80 spe-I3 him-5 males tested; 20 pears to be dueto the presence ofspe-11 sperm attempted crossesof four males to one dpy-5 her- because we haveoccasionally observed her- maphrodite at 25 "). maphrodites that lack sperm in one of the gonadal spe-I3 hermaphrodites are completely self-sterileat arms (this phenomenon alsooccasionally occurs in 25" and weakly self-fertile at 16" (Table 4). spe-I3 wild-type hermaphrodites; unpublished observations), hermaphrodites grown at 25" produce spermatozoa and the oocytes formed by the spermless arm remain that cancrawl in the spermatheca but are rapidly as single cells, but become polyploid like most other swept out into the uterus by passing oocytes and are spe mutants (data not shown). Restrictively grown spe- eventuallylost (Table 4). These oocytes resume I1(hc77ts), but not spe-I I (hc90), hermaphrodites also meiosis after passing through the spermatheca and FIGLIRE3.-a and b. Low magnific;ltion microgr;~phsof dissected male gonads of spe-4(hr78)/+ (a) and spe-q(hr78) (b) X310. spe-l/+ is indistinguishable from wild type. accumulating hundreds of spermatids (representative spermatid at arrow). spe-4 males do not accumulate spermatids but arrest spermatogenesis at what appears to be primary spermatocytes (representative spermatocyte at thearrow), most of which remain attached to the syncytial gonad (g). c and d, Nomarski and fluorescence micrographs of DAPl stained spe-4(hc81) spermatocytes, X750. arrow, a cell that looks like a Spermatocyte by Nomarski optics (c) actually has four haploid nuclei when viewed by fluorescence microscopy (d). arrowhead, a cell that has attempted to complete meiosis but the haploid nuclei do not separate into spermatids, and the residual body is much smaller than wild type. 442 " I FIGURE4.-Fluorescent photographs of DAPl stained hermaphrodites. The spermatheca appears in the boxed regions, sperm nuclei are indicated by small arrows and oocyte nuclei by large arrowheads. Analyses of nonconditional spe mutants were carried out in a dpy-5 background, which appeared to have no effect on the Spe phenotype (see MATERIALS AND METHODS). a, spe-4(hc81) dpy-5(e61)hermaphrodites do not accumulate mature sperm in their spermatheca, yet oocytes are produced that become polyploid (arrowhead), X500. b, spe-8(hd0) dpy-5(e61) hermaphrodites accumulate many sperm in their spermatheca but these are swept into the uterus (arrows to the right of the box) by passage of the oocytes that then become polyploid in the uterus (arrowhead), X800. c, spe-1 J(hc90)dpy-5(e61) hermaphrodites accumulate many sperm in their spermatheca (arrow in box), and these sperm are not swept into the uterus by passing oocytes. A few oocytes undergo an orderly series of nuclear divisions (arrowhead) but they do not form normal embryos, X800. 443 444 S. W. L'Hernault, D. C. Shakes and S. Ward

TABLE 4 Quantitation of sterile phenotypen

Growth tem- Shelled Gene Allele perature Progeny (SE) eggs6 Oocytes (SE) n Wild type (Bristol) 16" 350 +. 54 >1 179 f 25 15 25' 183 f 40 >1 122 f 22 15 spe-4 hc78 16" 0 0 36 f 4 15 hc81 16" 0 0 77 f 11 17 spe-5 hc93 16"

hc77/sDf4 25" >1 ND ND 11 hc90/sDf4 20" 0 ND ND 15 hc90 16" 0 0 378 f 26 17 ' Oocyte and/or progeny countswere performed on 15-30 individual hermaphrodites until they either died or had stopped laying. These shelled eggs do not subsequently hatch. ' ND = not determined. eventually become polyploid (data notshown but sim- dites are almost completely self-sterile and lay many ilar to spe-8 in Figure 4b). Surprisingly, while the self- oocytes at 25" (Table 4). spe-15 males produce large sterility phenotype of spe-I3 is slightly temperature numbers of spermatidsthat activate in vitro after sensitive (Table 4), thesweep of hermaphrodite sper- pronase treatment to form spermatozoa that are cy- matozoa is dramatically temperature sensitive (Table tologically indistinguishable from wild type (data not 6). spe-13 hermaphrodites raised at 16" retain most shown). of their sperm, even after 196oocytes have been laid. fer-l(hclts, hclhs,hc24ts, hc82ts,hc9lts, hc47, hc80, In contrast, spe-I3 hermaphroditesraised at 25" have hc136, b232ts)fer-1 has been studied in detail previ- swept out all of their sperm after 197 oocytes have ously (WARD and MIWA 1978; ARGONand WARD been laid (Table 6). Interestingly, while 25" grown 1980; WARD,ARGON and NELSON198 l), and only spe-I3 worms lay about as many oocytes as the total new information will be discussed. The nine alleles all number of oocytes produced by unmated wild-type show the same sperm phenotype: infertile spermato- hermaphrodites, 16 " grown spe-I3 hermaphrodites, zoawith stubby motile pseudopods. The alleles do which containsperm that are infertile but are not differ from one another enough toindicate they must swept out by passing oocytes, produce 1.4 times as represent at least five distinctive mutations. The al- many oocytes as unmated wild-type hermaphrodites leles hc47, hc80 and hc136 are nonconditional, exhib- (note that most oocytes get fertilized and give rise to itingcomplete sterility at both16" and 25". The progeny in the case of wild type; Table 4). remaining alleles are temperature sensitive and are spe-I 5(hc75ts)-spe-I 5 hermaphrodites are slightly fertile at 16" in all cases. The restrictive temperature self-fertile at16" (Table 4), andtheir young are is 20" for hc82, hc91 and b232 while it is 25" for hcl, produced throughout thelife cycle,unlike most of the hc13 and hc24. As previously noted (WARDand MIWA other spe mutants whose progeny (if any) are produced 1978; ARGONand WARD1980), hc24 produces signif- during the first day of adulthood. spe-15 hermaphro- icantly more progeny at 25" than either hcl or hc13. Spermatogenesis in C. eleguns 445

FIGURE5.-a, Low magnification Nomarski micrograph of a spe-5(hc93)dissected gonad (g). The arrow indicates a field of spermatocytes and spermatids, X290. b and c,Nomarski (b) and fluorescence (c) micrographs of Hoechst 33342 stained spe-5(hc93)sperm. Cells that appear by Nomarski optics (b) to be primary spermatocytes (boxed) actually have four haploid nuclei (c). The cells that look like spermatids (arrow) also have normal looking haploid nuclei, X740.

TABLE 5 Hermaphrodite sperm"

Sperm/ Initial Later Hermaphrodite sweep/oocyte Hermaphrodite sweep/oocyte

G ene Allele Gene N x SE N x SE N x SE f=O e0 e12 or 18

dPY-5 e6 dPY-5 I 225 8 f 14 9 0.1 f 0.03 spe- I I hc90 e61 11 116+9 15 0.1 + 0.02 spe-9 hc52ts spe-9 9 121 f 16 9 1.0 f 0.1 hc88ts 5 167 +. 28 5 0.8 +. 0.3 spe-8 hc50 e6 hc50spe-8 I 9 183 f 16 4 3.7 * 1 3.6 9 f 0.4 spe- I2 hc76 8 246 f 22 7 11.0 f 2.3 8 6.2 2 0.5 spe- 13 13hc137ts 282 f 16 13 9.1 f 1.6 Synchronized groups of worms were grown at 20" or 25" for temperature sensitive (ts) mutations. Worms were fixed and stained with Hoechst 33258 or DAPl at the onset of egg laying (t = 0) and, in two cases, after a numberof hours of egg laying (t = 12 or 18). The counts at t = 0 include some primary Spermatocytes that each contribute four to the sperm count. The sweep/oocyte shows the number of sperm lost from the spermatheca (excluding those that fertilized an egg) for each egg or oocyte that has passed. At t = 0, all oocytes and/or eggs are still in the uterus while a number of oocytes and/or eggs have been laid by t = 12 or 18 hr. and these expel sperm from the uterus.

The trans heterozygote hcl3lhc24 is more penetrant one young each and the other 14 were completely than hc24/hc24 (virtually all hc24/hc24 produce in sterile). Electron microscopic examination of fer- excess of eight young/hermaphrodite/generation, see l(hcl)/nDf23spermatozoa isolated from males raised ARGONand WARD 1980; 2/16 hcl3lhc24 produced under restrictive growth conditions (data not shown) 446 S. W. L’Hernault, D. C. Shakes and S. Ward

PO

+ F1 tra- 1 eDp6

F2 X unc-73 dpy-5

(only if spe tra I males are fertile) t spe t + nonUnc unc-13 dpy-5 tre- 1 t spe t + nonUnc F3 rc unc-13 dpy-5 tra- 1 unc- 73 + Uric unc-73 dpy-5 tra- 7 Y

FIGURE6.-Scheme by which spe-8 and spe-12 XX males were created and tested for fertility. Phenotypic unc-13 hermaphrodites bearing a recessive allele (e1099)of the transformer mutation tra-Z (balanced by the free duplication eDp6) were crossed to spe/spe (either spe-8(hc50) or spe-l2(hc76) males. Resulting F1 non-Unc hermaphrodites were picked and allowed to segregate FZ broods. All FZ males will be tra-1 homozygotes, except for rare cases of X chromosome nondisjunction, and about one-thirdof the FZ non-Unc males will be spe homozygotes. One hundred twenty FZ non-Unc males of unknown genotype were individually mated to several unc-13 dpy-5 tester hermaphrodites and scored 4 days later for the presence of outcross progeny. Overall mating efficiency of tra-1 XX males was 30% for the spe-8 crosses and 22% for spe-12 crosses. The genotype of most of the FZnon-Unc males (about two thirds) is spe +/-t unc-13, so most outcross Fs broods contain Unc hermaphrodites (arrow at “a,”between FZand Fs); mating efficiency of tra-1 males with this genotype was 24% for either spe gene. Some of the outcross Fs broods should be composed exclusively of phenotypic wild-type hermaphrodites if spe tra FP males were fertile (arrow at “b,” between F2 and Fs). A fraction of the successful outcrosses (17/36 for spe-8 and 7/26 for spe-12) lacked both males and unc-13 hermaphrodites in their FS broods, which indicated that these tru-1 males were most likely homozygous spe. However, FZspe unc-l3(+)/++; tru-1 males can arise because of F, recombination, and these would be false positives, if they successfully mated. The frequency of these false positives is a function of both the map distance between the spe gene and unc-13 (see Figure 1) and the mating efficiency of these males. If the mating efficiency of spe unc-13(+)/++; tra-1 males is assumed to be about 24%, then, of the above-mentioned crosses attributed to spe-8 or spe-12 tra-1 males, no more than 6/17 spe-8 or 1/7 spe-I2 crosses are likely to be false positives due to F, recombination.

reveals aspermatozoan phenotype that is indistin- found that the hc6 mutation fails to complementfer- guishable from the fer-1 (hcl)/fer-1 (hcl) homozygote 5(hc23ts), agene previously described as beingon described previously (WARD,ARGON and NELSON chromosome ZZZ (ARGONand WARD 1980). Linkage 1981). of hc23 was reinvestigated and it was found to be fer-3(hc3ts)-fer-3(hc3ts) had previously been as- linked to dpy-5. Furthermore, hc23 fails to comple- signed to chromosome Z (ARGONand WARD 1980). ment hc6 in all aspects of the phenotype described Linkage of this gene was reinvestigated and it was above. Creation of afer-6(hc6ts) dpy-5 cis double re- found to be unlinked to dpy-5 and was linked to rol-1 sulted in loss of the temperature sensitive effects on and unc-4. This gene has not been pursued further in worm movement and growth rate. Two factor map- this study since it is clearly on chromosome ZZ. ping of fer-6(hc6ts) dpy-5 produced non-Dpy isolates fer-6(hc6ts)-One temperature sensitive allele of that were non-Unc and grew at wild-type rates, sug- fer-6 has been previously described as a male-sterile gesting that these original aspects of the fer-6 pheno- mutant with abnormal spermatids (ARGONand WARD type were conferred by mutation(s) that were separa- 1980; WARD, ARGON and NELSON198 1). We have ble fromthe mutation causing the spe phenotype. foundthat, in additionto male-sterility, the fer-6 These data suggest thatfer-5(hc23ts) is a reisolate of strains characterized previously (ARGONand WARD fer-6(hc6ts) since it is unlikely that two ts mutations of 1980; WARD, ARGON andNELSON 1981) are temper- independent origin would also have the same second- ature sensitive foruncoordinated movement, and ary mutations; thefer-5 gene no longer exists. have a growth rate that is retarded by - 1 day at 25 O fer-7(hc34ts)-We have confirmed the previously with respect to wild type (data not shown). We have published data onfer-7 (ARGONand WARD 1980) and Spermatogenesis in C. elegans 447

TEMPERATURE SENSITIVE PERIOD OF SPE-S(hc 88) and fer genes. hc94 has been determined to be an allele of fog-I (feminization of the germline; M. K. ""Y BARTONand J. KIMBLE,unpublished data), and het- 450 -. erozygous hc94/+ males contain oocytes. The other 400- 8 four mutants all contain spermatids that become ap 350- parently normal spermatozoa after pronase treatment. 300- n LL 250- DISCUSSION O 200- 16' 150-, p 25' We have identified and described 23 new C. elegans 13 loo-, mutations on chromosome I that affect spermatogen- Z 50- /'I esis. These mutations could be identified because they 0-d bsw 0 cause normally self-fertile hermaphrodites to lay oo- 20 30 40 30 20 50 60 70 80 cytes. We have chosen to confine our efforts to chro- mosome Z because several well-studied spermatogen- TIME OF SHIFT (25' hr) esis-defective (spe) genes were already mapped to this FIGURE7.-Results of shifting spe-9(hc88ts) wormsbetween 16" and 25". Each point is the mean number of progeny produced per chromosome (ARGONand WARD 1980), balancers worm after shifting 10 hermaphrodites down (open circles) or up wereavailable for muchof this chromosome (e.g., (solid circles). The error bars are the standard deviations of the HOWELLet al. 1987), and we wanted to recover mul- mean values. Times are normalized to the 25" growth rate which tiple alleles. We have recovered 16/23 of our new is about twice the 16" growth rate. chromosome I spe mutations in cis to the morpholog- TABLE 6 ical marker dpy-5 to facilitatebalancing and other genetic manipulationsof these sterile strains. The +e-13 hermaphrodite sperm atdifferent temperatures" remaining seven chromosome Z spe mutations were ~ ~ ~~~ not recovered on a morphologically marked chromo- Raised at 16" Raised at 25" some I, but were subsequently shown to exhibit link- Oocytesand Sperm Oocytesand Sperm eggs laid present n eggs laid present n age to dpy-5. A few chromosome Z spe genes might 0 320f 8 9 0 310+ 16 8 not be identified because they show only weak linkage 51 359f 24 5 54 74f 13 3 to dpy-5 due to the effects of recombination. Such spe 132 206f 17 5 110 77* 7 10 genes, if they exist, will most likely be on the distal 196 211 k 11 5 197 0 5 right arm of chromosome I; it is about 27 map units Temperature dependence of sperm sweep rate in spe-13 her- from dpy-5 to the right end of the chromosome Z maphrodites. Hermaphrodites were raised at 16" or 25", and genetic map. Two spe-8 alleles (18 map units to the synchronized L4 larvae were placed as groups of 15 worms/plate. left of dpy-5) and one spe-9 allele (eight map units to The worms were transferred daily to fresh plates, a few were fixed the right of dpy-5) were actually recovered as linked and stained with DAPI and the average number of oocytes and eggs laid per worm, up to that point, was calculated. Sperm are dpy-5 doubles. The spe-13 gene is unambiguously quickly swept out of the hermaphrodite gonad by passing oocytes linked to dpy-5 and, since this gene is 21 map units to at 25", but most sperm are retained in the spermatheca of worms the left of dpy-5, it seems probable that our screening raised at 16". The sperm count (* the standard error of the mean) procedures have sampled at least 85% (42 map units) for which no oocytes or eggs have been laid was determined in of chromosomeI. young hermaphrodites that had oocytes in their uterus. At 25", the worms laid about 200 oocytes during the next 48 hr, whereas at The new chromosome I spe mutations, together 16". the worms took 120 hr to lay about 200 oocytes. with six previouslyisolated mutations (ARGONand WARD1980), define 1 1 complementation groups; six have shown thatfer-7 complements at least one allele of these complementation groups are defined by mul- of all the chromosome I spe genes that map in the tiple alleles. Twenty-two of the 23 new chromosome same region. Z spe mutations were recovered from 3838 picked Other mutations: We have also recovered five mu- mutant Fl's for an overall EMS induced forward mu- tations that appeared to be spe because unmated her- tation rate of -6 X lo-' chromosome I spe mutations/ maphrodites lay oocytes but produce progeny follow- gamete. BRENNER(1 974)calculated the average EMS ingsuccessful mating to wild-typemales. However, induced forward mutation rate forC. elegans to be -5 these mutations were not studied in detail because X 10-4 mutations/gene/gamete. This figure suggests they either hadgonadal and somatic abnormalities there should be about 12 spe genes in the 42 map unit (hc45), erratic expressionof the sterile phenotype interval of chromosome I we have sampled. Two of (hc45, hc84, hc109, hclll), or result in a switchin these genes, spe-8 and fer-I, have mutation rates that germline specific sex determination (hc94) and were, are somewhat higher than the other spe genes (1.6 X therefore, not sperm specific. These mutations com- 1O-' and 1.3 X 10-' mutations/gene/gamete, respec- plement each other and all other chromosome I spe tively) suggesting that they are favored targets for 448 S. W. L’Hernault, D. C. Shakes and S. Ward

Primary Secondary Spermatozoa spermstocyte Spormstld apermstocytes 10 fer-7spe- 10 fer- 14 spe- 1 1 spe-23 spe-20 \ \ fer- 1 spe- 15 t \ ler-3 spe- 19 fer-6 fer-4 \ fer-6 spe-21 spe-4 \ VARIABLE‘

\ Prlrnary fer-2 spermatocytes spe- 1 . onrachis t w I I her. lra fern,fog genes FIGURE8.-Stages of normal spermatogenesis are shown diagranlmatic;rlly as an ordered pathway of morphogenesis. Genes are placedon thc pathway at the stage that is altered by mutations in a gene. Sperm phenotypesof spe-8 and spe-12 are different in the hermaphrodite (*) and the male.which has spermato7oathat are cytologically indistinguishable from wild type. Relow the pathway areshown abnormal intermediates that accumulate when the gene is mutated. The last step on the pathway represents mutant spernmatozoa that are cytologically norm;d and can enter oocytes.

EMS mutagenesis. Poisson analysis of mutant allele duce mostly primary spermatocytes, but mutations in frequencies by the method ofMENEELY and HERMAN either gene occasionally allow partial progression of (19759, which reduces overemphasis on the more mu- spermatocytesthrough the second meiotic division. table genes such as spe-8 and fer-1, indicates there are spe-4 spermatocytes never proceed further than the about 14 chromosome I spe genes. We conclude that formation of incomplete spermatids connectedto the there could be less than three unidentified chromo- residual body, whereas spe-5 animals produce a few some I spe genes in the 42 mapunit interval of normal looking spermatids that mustoccasionally ma- chromosome I we have sampled. tureinto functionalspermatozoa since spe-5(hc93) Many, although notall, spe mutants have sperm that hermaphrodites sometimes produce a few progeny. appear obviously different from wild type. The phe- Most spe-4 and spe-5 spermatocytes,no matter at notypic effects of all of these spe mutations arereces- whichcytological stage they arrest, contain haploid sive; since none are semidominant, they probably do nuclei. This indicates that nuclear divisions of meiosis not act in cis on sperm. A previously studied series of are not dependent oncytokinesis; similar results have temperature sensitice spe mutations were all found to beenfound in otherexperimental systems,such as act late in spermatogenesis (ARGON and WARD 1980; yeast (reviewed by ESPOSITOand KLAPHOLZ 1981). WARD,ARGON and NELSON1981). We have demon- Mutations in the other nine chromosomeI spe genes stratedthat our screening procedures (described affectspermatogenesis postmeiotically. The pheno- above) allow the recovery of mutations that permit types of three of these chromosome I spe genes (fer- analysis of most, if not all, of spermatogenesis. All 1, fer-6 and fer-7) have been discussed in detail else- cytologicallyidentifiable stages in spermatogenesis where(WARD and MIWA 1978; ARGON and WARD nowhave at least one mutation (chromosome I or 1980; WARD, ARGON and NELSON1981). Mutations elsewhere, unpublished data) that either prevents or in any of thesix recently identified chromosomeI spe alters its transition to thenext cellularstage. The genes (spe-8, spe-9, spe-11, spe-12, spe-13 and spe-15) phenotypes of the new chromosome I and previously permit formation of cytologically normal spermatids. described mutants are summarizedin Figure 8. Morphogenesis of spermatids into spermatozoa and/ Mutations in two chromosome I genes, spe-5 and or spermatozoan function are altered by mutation in spe-4, actduring meiosis to preventformation of any of thesesix genes. normal spermatids. Bothspe-5 and spe-4 mutants pro- spe-8 and spe-12 hermaphrodites accumulate sper- Spermatogenesis in C. elegans 449 matids that do not mature into spermatozoa. The mutation probably is caused by defective spermatozoa- spermatids that spe-8 and spe-12 males deposit in her- spermatheca interactions such that spermatozoa either maphrodites during copulation become spermatozoa get forced out of the spermatheca by passing oocytes that crawl to the spermatheca and, like sperm from or do not correctly recognize and/or position them- wild-type males (WARDand CARREL1979), success- selves within the spermatheca in the first place. The fully fertilize the hermaphrodite's oocytesbecause analysis of spe-13 hermaphrodites proves to be more they outcompete wild-type hermaphrodite spermato- complicated; the sperm are rapidly swept in animals zoa. Thus, single mutations in either the spe-8 or spe- grown at 25" (-100 times wild type; Table 5)but are 12 gene can convert the mode of C. elegans reproduc- not swept at all in animals grown at 16" (Table 6), tion from hermaphroditic to male/"female," and spe- despite the fact thatthe animals are onlyslightly 8 and spe-12 mutations have revealed a new class of temperature sensitive for self-sterility. Perhaps a sin- spe mutants that are male-fertile. These differences gle encoded by the spe-13 gene is required for between the hermaphrodite and male phenotype are sperm to interact with the spermatheca and oocytes. not due to X-chromosome dosage because genotypic These two interactions might be differentially sensi- XX spe-8 or spe-12 wormscan sire progeny when tive to the levelsof this protein present in spe- converted into males by tra-1. Therefore, the pheno- 13(hc137ts) mutant animals at 16" and 25". typic sex and not the genotype of the worm deter- Spermatozoa made by spe-11 hermaphrodites, un- mines if fertile spermatozoa can form. Recently, the like all other chromosome Z spe mutations, are not response of hermaphrodite and male spermatids to swept out by passing oocytes,but they do interact with various in vitro sperm activators has been examined. oocytes and often stimulate several nuclear divisions. These results indicate that spe-8 and spe-12 directly Additionally, hermaphrodites carrying the hypo- affect spermatids rather than in vivosperm activator(s) morphic allele hc77ts (see below), produce a number (D. C. SHAKES,S. W. L'HERNAULT andS. WARD,in of round thinly shelled eggs (as opposedto wild-type preparation). oblong eggs). These thinlyshelled "eggs" do not Four other new chromosome Z spe mutants (spe-9, hatch, and it recently has been discoveredthat spe-11 spe-11, spe-13 and spe-15) all produce spermatids that (hc90) hermaphrodite sperm enter and activate oo- activate in vitroto form crawling spermatozoa. Mating cytes during self-fertilization but subsequently form experiments, however, indicate that males carrying a an abnormal zygote (D. HILLand S. STROME,personal mutation in three of these spe genes are sterile (spe-15 communication).Sperm from spe-11 males cause these has not been tested). Crawling spermatozoa are also same abnormalities when they cross-fertilize wild-type observed in the spermatheca of unmated her- oocytes, suggesting that defective spe-11 sperm cause maphrodites carrying these mutations (spe-15 has not abnormal zygote formation (D. HILLand S. STROME, been tested). Thus, in these mutants, as in most of personal communication). These analysesof spe-11 our spe mutants, the hermaphrodite and male phe- mutations have revealed a possible role for sperm- notype appears to be the same. specific components in early embryogenesis.Our pre- Mutants that produce cytologically normal sper- liminary analysis of the spe-1 l(hc77ts) temperature matozoa are likely to have defects in spermatozoa- sensitive period indicates thatthe defect affecting spermatheca interactions and/or spermatozoa-oocyte hermaphrodite self-sterility is temperature reversible interactions. These two phenomena canbe distin- during adulthood. These data suggest that mutant guished by examining sperm behavior in the her- spe-11 gene product can recover (at least) partial wild- maphrodite reproductive tract. The passage of oo- type function if sperm are shifted from restrictive to cytes through the reproductive tract usually, but not permissive conditions because mature sperm lack ri- always,sweeps nonfunctional spe sperm out of the bosomes and, therefore, do not synthesize proteins spermatheca (Tables 5 and6) ashas been shown (reviewed by WARD1986). previously (WARDand MIWA1978; ARGONand WARD Our genetic analysis of the spe mutations discussed 1980). For example, spe-8 and spe-12 hermaphrodite in this paper indicates that their phenotypes probably spermatids are swept very rapidly (respectively, -30 result from single gene recessive mutations. Alleles of or - 100 times wild type; Table 5) by passing oocytes, two genes,fer-1 and spe-8 arise at or slightly abovethe presumably because theylack pseudopods and, there- average forward (knockout) mutation frequency (see fore, can neither move nor interact with the sperma- above), suggesting that some alleles should be null. theca; both of these processes are known to be impor- All seven spe-8 alleles cause hermaphrodites, but not tant in preventing the sweeping of wild-type sperma- males, to produce nonfunctional sperm, and only tozoa (WARDandCARREL 1979). Although hcl34ts shows any significant hermaphrodite self-fer- hermaphrodites with mutations in spe-9 form normal tility at 16" . The uniformity of these phenotypic traits looking spermatozoa, these sperm are alsorapidly in animalshomozygous for anyof the seven spe-8 swept (-10 times wild type; Table 5). Therefore, this alleles strongly suggeststhat this is the null phenotype. 450 S. W. L'Hernault, D. C. andShakes S. Ward

Likewise, all nine alleles of fer-1 cause identical phe- spermatogenesis stimulates oogenesis. The number of notypes in 25" grownmutant sperm at the light oocytes produced by a spe hermaphrodite, however, microscope level, despite the fact that this phenotype seems to depend on when during spermatogenesis a is only conditionally expressed in the sixts alleles; spe mutation acts. Two chromosome I spe genes (spe- three ts alleles also cause the same ultrastructural 4 and spe-5) block spermatogenesis during meiosis, defects (WARD, ARGON and NELSON 198 1). fer- and hermaphrodites mutant in either of these genes l(hc1ts)appears to be conditionally null since the produce - 10-20% of the wild-type number of oocytes ultrastructural phenotype of sperm from either hcl/ (Table 4). Mutants that arrest spermatogenesis post- hcl or hcllnDf23 males grown under restrictive con- meiotically generally produce about thesame number ditions have the same Spe phenotype (fer-1 fails to of oocytes as wild type (see spe-8 and spe-12, Table 4) complement nDf23). In contrast,fer-1 (hc24ts) appears as has been noted beforeby others (ARGONand WARD to be conditionally hypomorphic since hermaphro- 1980). One striking exception to this rule is the post- dites become more penetrant for self-sterility when meiotically acting temperature sensitive mutation spe- hc24ts is in trans to fer-l(hcl3ts). 13(hc137ts) that has an unusual correlation between The spe-4, spe-5 and spe-11 genes are each defined oocytes produced and sperm sweep rate. spe-13 her- by two alleles, and noncomplementing deficiencies maphrodites rapidly sweep their sperm when raised exist for all three genes. Both alleles of spe-4 and spe- at 25 " and produce about thesame number of oocytes 5 have the same phenotype, and spe heterozygotes in as unmated wild-type hermaphrodites. In contrast, trans to noncomplementing deficiencies have the same spe-13 hermaphrodites raised at 16" do not sweep cytological phenotype as the spe homozygote, suggest- their sperm and produce 1.4 times more oocytes than ing that these nonconditional alleles eliminate gene unmated wild-type hermaphrodites. The longer per- activity. One allele (hc77ts) of spe-11 is temperature sistence of sperm in the spermatheca at 16" appears sensitive and incompletely penetrant forsterility while to be responsible for stimulatingexcessive oocyte pro- theother allele (hc90) is nonconditional and com- duction in these mutant hermaphrodites, and is con- sistent with production of a direct signal by sperma- pletely sterile. When heterozygous in trans to a non- tozoa that stimulates oogenesis. complementing deficiency, the sterility of spe- Spermatogenesis-specific mutations can be readily ll(hc77ts) is enhanced while spe-ll (hc90) remains isolated in C. elegans because mutantscreening is completely sterile. This indicates that hc77ts is hypo- carried out in hermaphrodites. Delivery of sperm to morphic for gene function, but hc90 might eliminate the in vivo location where oocytes are fertilized is far gene activity. There are six other chromosome I spe less complicated in self-fertilizing hermaphrodites genes (fer-6, fer-7, spe-9, spe-12, spe-13, and spe-15) than in many dioecious organisms where male mating for which the effects of existing mutations on gene competence and successful copulation are required. function are less certain. The presence of defective sperm causes hermaphro- The spe mutantsstudied in this paper were re- dites to lay oocytes, ratherthan shelled eggs, and covered because they cause hermaphroditesto lay picked hermaphrodites that lay oocytes can be easily oocytes. C. elegans is unusual among animals in that identified because self-fertilization in youngadult wild type hermaphrodites produce more eggs than wild-type hermaphrodites is quite efficient. Screens the available sperm (reviewed by NIGON1965). The for oocyte-laying mutant hermaphrodites have,so far, gonad produces about 300 sperm during the herma- permitted the identification of 36 C. elegans spe genes phrodite L4 larval stage, and they are stored in the on all six chromosomes (this paper and our unpub- spermatheca. Sperm production then ceases, and the lished results). All of these spe mutations affect only gonad produces oocytes throughout adulthood. Wild- sperm and do nothave any apparent effects on worm type oocytes are fertilized in the spermatheca with growth, behavior or viability. Complementation close to 100% efficiency (WARD and CARREL1979), screens have not been used to recover new alleles of and each hermaphrodite lays about 350 fertilized eggs any spe gene, and so, for many spe genes, we cannot followed by 180 unfertilized oocytes at 16" (Table 4). be sure that available mutations reflect the null phe- However, wild type hermaphrodites can produce notype. However, 13 spe genes have been mapped to 1000-1 500 progeny if they are repeatedly mated to regions for which noncomplementing deficiencies ex- males (HODGKIN1983). Unmated fern-1 XX animals, ist (the five chromosome I genesdescribed in this which are true females because they never make any paper and our unpublished results). Twelve speldefi- sperm (NELSON, LEWand WARD1978; DONIACHand ciency trans heterozygotes do not have different phe- HODGKIN1984), produce a few (- 10) oocytes but can notypes from the spe homozygote; the one exception produce hundreds of progeny when mated (NELSON, is the fer-$(hclts)/deficiency heterozygote, which has LEWand WARD1978). These dataindicate that either gonadal abnormalities not observed in the fer-4 ho- the presence of sperm or a signal associated with mozygote (our unpublished observations). These re- Spermatogenesis in C. elegans 45 1

sults suggest that many spe genes probably exert their Service grant GM09684 to S.W.L. and a Carnegie Corporation phenotypic effects only in sperm and do not affect Fellowship to D.C.S. somatic tissue. Screening for spermatogenesis mutations in other LITERATURECITED experimental systems, such as Drosophila (reviewed by AFZELIUS,B. A,, 1985 The immotile cilia syndrome: a microtu- HACKSTEIN1987 and LIFSCHYTZ1987) or mice (re- bule-associated defect. CRC Crit. Rev. Biochem. 19 63-87. viewed by HANDEL1987) throughmale-sterile screens ARGON,Y., and S. WARD,1980 Caenorhabditiselegans fertiliza- tion-defective mutants with abnormal sperm.Genetics 96 413- leads to recovery of many mutations that affect cells 433. in addition to or other than sperm. In fact, no genes BRENNER,S., 1974 The genetics of Caenorhabditiselegans. Ge- that affect spermatogenesis exclusively have been un- netics 77: 71-94. covered in mice (reviewed by HANDEL1987), andonly BURKE,D. J., 1983 A molecular genetic analysis of the major sperm proteins of the nematode Caenorhabditis elegans. Ph.D. a few such genes have been discovered in Drosophila thesis, Johns Hopkins University. (reviewed by HACKSTEIN1987 and LIFSCHYTZ1987). DONIACH,T., and J.HODGKIN, 1984 A sex-determining gene,fem- Commonly, either male mating behavioror secondary 1 required for both male and hermaphrodite development in sexual characteristics are alteredso that sperm cannot Caenorhabditis elegans. Dev. Biol. 106: 223-235. be deposited in the female. This phenotypic class, as EDGAR,L. G., 1982 Control of spermatogenesis in the nematode Caenorhabditis elegans Ph.D. thesis, University of Colorado. discussed above, is eliminated when studying C. ele- EDGLEY,M. L., and D.L. RIDDLE, 1988 Caenorhabditiselegans guns because the hermaphroditic mode of reproduc- genetic map. Caenorhabditiselegans Stock Center, University tion does not require mating. Many male-sterile Dro- of Missouri, Columbia. sophila mutations cause cytologically defective sperm, BPOSITO,R. E., and S. KLAPHOLZ, 1981 Meiosis and ascospore but these mutations subsequently proved to be alleles development. pp. 21 1-287. In: TheMolecular Biology of the Yeast Saccharomyces, Lfe Cycle and Inheritance, Edited by J. N. of genes that have a null-lethal phenotype. The C. STRATHERN,E. W. JON= and J. R. BROACH.Cold Spring elegans spe mutations that have been analyzed (a total Harbor Laboratory, Cold Spring Harbor, N.Y. of more than 30 genes on chromosome I and else- FERGUSON,E., and H. R. HORVITZ,1985 Identification and char- where) (WARDand MIWA 1978; ARGONand WARD acterization of 22 genes that affect the vulval cell lineages of the nematode Caenorhabditis elegans. Genetics 110 17-72. 1980; WARD,ARGON and NELSON 1981 ; ROBERTS HACKSTEIN,J. H. P., 1987 Spermatogenesis in Drosophila. pp. 63- and WARD 1982a, b; our unpublished observations) 1 16. In: Spermatogenesis: Genetic Aspects. Results and Problems are always associated with sperm defects, and the null inCell Dzfferentiation 15, Edited byW. HENNIG.Springer- phenotype of many of these genes probably does not Verlag, Berlin. HANDEL,M. A., 1987 Genetic control of spermatogenesis in mice. affect tissues otherthan sperm (see previous para- pp. 1-62. In: Spermatogenesis: Genetic Aspects. Results and Prob- graph). Thisdoes notmean that somatically expressed lems in Cell Dzfferentiation 15, Edited by W. Hennig. Springer- C. elegans genes are not also expressed in sperm but, Verlag, Berlin. rather, that our screeningmethod has identifieda HIRSH,D., and R. VANDERSLICE,1976 Temperature-sensitive de- class of genes whose activity seems limited to sperm. velopmental mutants of Caenorhabditis elegans. Dev. Biol. 49: 220-235. Mutations that act either pre or postmeiotically can HIRSH,D., D. OPPENHEIMand M. KLAS, 1976 Development of be obtained, and in fact, mutant hermaphrodites that the reproductive system of Caenorhabditis elegans. Dev. Biol. do not make any sperm but do make a few oocytes, 49: 200-219. such as fern (NELSON,LEW and WARD1978; DONIACH HODGKIN,J., 1986 Sex determination in the nematode C. elegans: analysis of tra-3 suppressors and characterization offem genes. and HODGKIN1984) andfog (this paper) mutants can Genetics 114: 15-32. be identified by this screening method. This means HODGKIN,J., 1980 More sex-determination mutants of Caenor- that our screening method probably permitsrecovery habditis elegans. Genetics 96 649-664. of any spe mutation, no matter when it acts during HODGKIN,J., 1983 Male phenotypes and mating efficiencyin spermatogenesis, and that many aspects of C. elegans Caenorhabditis elegans. Genetics 103: 43-64. HODGKIN,J., and S. BRENNER,1977 Mutations causing transfor- spermatogenesis are accessible by genetic analysis. mation of sexual phenotype in the nematode Caenorhabditis elegans. Genetics 86: 275-287. We thank DANIELBURKE, TOM ROBERTS andJAcoB VARKEY for HODGKIN,J., H. R. HORVITZand S. BRENNER,1979 Non-disjunc- the original isolation of some of the alleles described here, and tion mutants of the nematode Caenorhabditis elegans. Genetic's thank EILEEN HOGANand MICHAELSEPANSKI for excellent technical 91: 67-94. assistance. We thankJuDITH KIMBLE,ANN ROSE and SUSANSTROME HORVITZ,H. R., S. BRENNER,J. HODGKINand R. K. HERMAN, and their laboratories for communicating unpublished results and 1979 A uniform genetic nomenclature forthe nematode thank the Rose laboratory for nematode strains. We thank ANDREW Caenorhabditis elegans. Mol. Gen. Genet. 175: 129-133. FIREfor suggesting the spe Ira experiments and many helpful HOWELL,A. M., S. G. GILMOUR,R. A. MANCEBOand A. M. ROSE, discussions. Some nematode strains used in this study were provided 1987 Genetic analysis of a large autosomal region in Caenor- by the Caenorhabditis Genetics Center, which is supported by con- habditis elegans by use of a free duplication. Genet. Res. 49 tract number NO1-AG-9-2113 between the National Institutes of 207-213. Health and the Curator of the University of Missouri. This work KIMBLE,J., L. EDGARand D. HIRSH,1984 Specification of male has been supported by U.S. Public Health Service grant GM2524.3 development in Caenorhabditis elegans:thefm genes. Dev. Biol. to S.W., the Carnegie Institution of Washington, Public Health 105 234-239. 452 S. W. L’Hernault, D. C. Shakes and S. Ward

KLASS, M., N. WOLFand D. HIRSH, 1976 Development of the ROSE, A. M., and D. L. BAILLIE,1979 The effect of temperature male reproductive system and sexual transformation in the and parental age onrecombination and nondisjunction in Cae- nematode Caenorhabditis elegans. Dev. Biol. 52: 1-18. norhabditis elegans. Genetics 92: 409-4 18. LIFSCHYTZ,E., 1987 The developmental program of spermio- ROSE, A. M., and D. L. BAILLIE,1980 Genetic organization of the genesis in Drosophila: a genetic analysis. Int. Rev. Cytol. 109 region around unc-15 (I), a gene affecting paramyosin in Cae- 21 1-258. norhabditis elegans. Geneti0 96 639-648. MAINLAND,D. L., L. HERRERAand M. I. SUTCLIFFE,1956 ROSE, A. M., D. L. BAILLIEand J. CURRAN,1984 Meiotic pairing Statistical Tablesfor use with Binomial Samples, Contingency Tests, behavior of two free duplications of linkage group I in Caenor- ConfidenceLimits and Sample Size Estimates. Department of habditis elegans. Mol. Gen. Genet. 195 52-56. Medical Statistics, New York University School of Medicine. SCHEDL,T., and J. KIMBLE,1988 fog-2, a germ-line specific sex determination gene required for hermaphroditespermatogen- MENEELY,P. M., AND R. K. HERMAN,1979 Lethals, steriles and deficiencies in a region of the X chromosome of Caenorhabditis esis in Caenorhabditis elegans. Genetics 119: 43-6 1. SIGURDSON,D. C., G. J. SPANIERand R. K. HERMAN, elegans. Genetics 92: 99-1 15. 1984 Caenorhabditiselegans deficiency mapping. Genetics NELSON, G. A,, 1979 Nematode sperm motility. Carnegie Inst. 108: 331-345. Wash. Year Book 78: 62-66. TRENT,C., N. TSUNGand H. R. HORVITZ,1983 Egg laying NELSON, G. A., and S. WARD,1980 Vesicle fusion, pseudopod defective mutants of the nematode Caenorhabditis elegans. Ge- extension and amoeboid motility are induced in nematode netics 104: 619-647. spermatids by the ionophore monensin. Cell 19 457-464. WARD,S., 1986 Asymmetric localization of gene products during NELSON,G. A., K. K. LEWand S. WARD,1978 Intersex, a tem- the development of Caenorhabditiselegans spermatozoa. pp. perature sensitive mutant of the nematode Caenorhabditis ele- 55-75. In: Society for Developmental Biology 44th Annual Sppo- guns. Dev. Biol. 66 386-409. sium, Edited by J. G. GALL.Alan R. Liss, New York. NELSON,G. A., T. M. RoBERTSand S. WARD,1982 Caenorhabditis WARD,S., and J. S. CARREL,1979 Fertilization and sperm com- elegans spermatozoan locomotion: amoeboid movement with petition in the nematode Caenorhabditis elegans. Dev. Biol. 73: almost no actin. J. Cell Biol. 92: 12 1-1 3 1. 304-321. NIGON,V., 1949 Les modalitks de la reproduction et le dktermi- WARD,S., and J. MIWA, 1978 Characterization of a temperature- nisme du sexe chez quelques nkmatodes libres. Ann. Sci. Nat. sensitive fertilization-defective mutant of the nematode Cae- ZOO^.) 11: 1-132. norhabditis elegans. Genetics 88 285-303. ROBERTS,T. M., and S. WARD,1982a Centripetal flowof pseu- WARD,S., Y. ARGONand G. A. NELSON,1981 Sperm morpho- dopodial components could propel the amoeboid movement of genesis in wild type and fertilizationdefective mutants of Cae- Caenorhabditis elegans spermatozoa. J. Cell Biol. 92: 132-1 38. norhabditis elegans. J. Cell Biol. 91: 26-44. ROBERTS, T. M., and S. WARD,198213 Membrane flow during WARD,S., E. HOGANand G. A. NELSON,1983 The initiation of nematode spermatogenesis. J. Cell Biol. 92: 1 13-120. spermatogenesis in the nematode Caenorhabditis elegans. Dev. ROSE, A. M., 1980 Genetic studies of the gene coding for para- Biol. 98: 70-79. myosin in Caenorhabditis elegans: unc-15 and the adjacent re- WATERSTON,R. H., 1980 A second informational suppressor, sup- gion. Ph.D. thesis, Simon Fraser University, Burnaby, B. C. 7 X,in Caenorhabditis elegans. Genetics 97: 307-325. Canada. Communicating editor: R. K. HERMAN