Copyright 0 1993 by the Genetics Societyof America

The spe-6Gene Is Required for Major Assembly and Shows Second Site Non-Complementation With an Unlinked Deficiency

Jacob P. Varkey,* Patricia L. Jansma,+Alicia N. Minniti* and Samuel Ward* "Department of Molecular and Cellular Biology and TArizona Research Laboratory, Division of Neurobiology, University of Arizona, Tucson, Arizona 85721 Manuscript received July 23, 1992 Accepted for publication September 1 1, 1992

ABSTRACT Caenorhabditis elegans spermatozoa move by crawling. Their motility requires thin cytoskeletal filaments assembled from a unique cytoskeletal protein, the major sperm protein (MSP). During normal sperm development the MSP is segregated to developing sperm by assembly into filaments that form a paracrystalline array in a transient organelle, the fibrous body-membranous organelle. Mutations in the spe-6 gene cause sterility because they lead to defective primary that do not form spermatids. In these mutant spermatocytes the MSP fails to assemble into fibrous body filaments. Instead, the unassembled MSP distributes throughout the cytoplasm and nucleus. Thus, the spe-6 gene product is necessary for normal MSP localization and assembly during sperm develop- ment. In addition to theirMSP assemblydefect, spe-6 mutant spermatocytes arrest at diakinesis although their spindle pole bodies still replicate and separate. This results in spermatocytes with four half-spindles surrounding condensed, but unsegregated, chromosomes. All four spe-6 alleles, as well as a chromosome III deficiency that deletes the spe-6 gene, fail to complement two small overlapping chromosome IV deficiencies, eDfl8 and eDfl9. This non-allele-specific second site non-complementa- tion suggests a concentration-dependent interaction between the spe-6 gene product and products of the gene(s) under eDfl8 and eDfl9, which include acluster of sperm-specific genes. Since MSP filament assembly is highly concentration-dependent in vitro, the non-complementation might be expected if the sperm-specific gene products under eDfl8 and eDfl9 were needed together with the spe-6 gene product to promoteMSP assembly.

ELLULAR differentiation requires ordered syn- eta1 filaments participating in motility (SEPSENWOL, C thesis and organization of new cell components RIS and ROBERTS1989; ROBERTSand KING 1991). which must be integrated into preexisting structures. Although MSP filament assembly in vitro has been Integration requires mechanisms for targeting pro- recently achieved with Ascaris MSP(KING et al. 1992) teins to specific areas of the cell and assembling them and replicated for Caenorhabditis MSP (SMITH1992), into functional structures. The development of Cae- it is not known how MSP assembly is regulated in vivo norhabditis elegans sperm provides an opportunity to in either species, nor is it known how fila- analyze such mechanisms genetically (WARD, ARGON ment assembly in developing spermatocytes compares and NELSON 1981; WARD 1986). to filament reassembly in the spermatozoa. In this In developing spermatocytes, a number of sperm paper we show that the C. elegans - specific gene products are localized into a specialized defective gene, spe-6, encodes a product that is nec- transient organelle known as the fibrous body-mem- essary for the properlocalization and assembly ofMSP branous organelle (FB-MO) complex (ROBERTS,PA- filaments in the FB-MO complex. Wealso present VALKO and WARD 1986). These organelles are then genetic evidence suggesting that the spe-6 gene prod- segregatedinto the developingspermatids. The fi- uct may interact with the product(s) of one or more brous bodies contain paracrystalline arrays of 2-4-nm genes defined by a deficiency mapped to a different filaments that are formed of a small basic protein, the chromosome. major sperm protein, MSP (WOLF,HIRSH and MCIN- MATERIALSAND METHODS TOSH 1978;WARD and KLASS 1982; ROBERTS,PA- VALKO and WARD 1986). TheseMSP filaments disas- Strains and genetics: C. elegans var. Bristol, strain N2, semble in mature spermatids and reassemble in sper- was used as wild type. Most strains were first described by matozoa. Analysis of the structure of MSP filaments BRENNER(1974), except as noted below: spe-6(hc49)dpy- I8(e364) 111, spe-6(hc92)unc-32(eI 89) III, spe-6(hcl43)dpy- and their assembly during crawling motility in Ascaris IS(e364) III, spe-6(hcl46)dpy-l8(e364)III, spe-26(hc138ts) IV, spermatozoa shows that they are the major cytoskel- spe-7(mn242) II, spe-4(hc78) I, spe-32(it129ts) IV, dpy-I8(e364)

Genetics 133: 79-86 Uanuary, 1993) 80 J. P. Varkey et al. Ill, unc-?2(e189) Ill. The deficiencies used in this analysis The tissue was stained in a moist chamber for 30 min in the wereeDfl8(IV)(H0m~1~ 1986),eDfl9(1V)(H0~~~1~1986) dark with 100 Hg/ml DAPI in PBS. A coverslip was placed and eDf2 (Ill) (ANDERSONand BRENNER1984). The strains on the tissue and pressed gently prior to microscopic ex- were maintained at appropriate temperatureson Escherichia amination. coli-seededplates, and genetic manipulations were per- Electron microscopy: Adult virgin males were hand dis- formed as described by BRENNER (1974). Anew allele of sected in SM salts, and the testes were transferred to 1% spe-6 was isolated by crossing ethyl methanesulfonate (EMS)- formaldehyde 1% glutaraldehyde inSM in a depression mutagenized N2 males with spe-6(hc49)dpyls(e364) homo- slide. The testes were fixed in a moist chamber for 1 hr at zygotes and screening the Fl outcross progeny for sterile room temperature and then embedded in 1% agar, and hermaphrodites. Similar Fl non-complementation screens small agar blockswith the tissue were incubated in the to chromosome IV deficiencies were done by crossing EMS- fixative overnight at 4". The blocks were rinsed in 10 mg/ mutagenized wild-type males to appropriate deficiency het- ml lysine for 20 min and then rinsed inSM. The blocks erozygous hermaphrodites and screening the F1 outcross were post fixed in 1% oso4in SM. Excess Os04was rinsed progeny for sterile hermaphrodites at 25 O. Instead of using out with water, and theblocks were stained in 0.5% aqueous morphological markers to distinguish the outcross progeny, uranyl acetate. The blocks were then dehydrated in a graded mated hermaphrodites were transferred daily to new plates series of ethanol and embedded in EPON/Araldite resin for and progeny hermaphrodites were selected onlyfrom plates sectioning. Thin sections were examined with a Philips 420 two days after mating that had approximately 50% males, electron microscope operating at 80 kV. ensuring that the hermaphrodites were also outcross. For Immuno-electronmicroscopy: Hand-dissected testes the analysis of the complementation of the chromosome IV were fixed as described above for conventional electron deficiencies eDfl8 and eDfl9 to different spe genes, herma- microscopy except that they were post fixed in 0.1 % OsOl phrodites heterozygous for the deficiency were crossed to in SM for 30 min. The testes were embedded in LR white homozygous males of temperature-sensitive alleles at per- resin following the manufacturer's directions (Polysciences, missive temperature or heterozygous malesof noncondi- Warrington, Pennsylvania) without en bloc staining in uranyl tional alleles. acetate. The sections were picked on nickel grids, etched Light microscopy:Spermatogenesis was analyzed in both with sodium metaperiodate, washed thoroughly and then males and hermaphrodites by hand dissecting individual incubated in a 1:200 dilution of anti-MSP serum (BURKE worms in sperm medium (SM) and examining with Nomar- and WARD1983) for 30-60 min. Excess antibody was re- ski differential interference microscopy. SM is 5 mM HEPES moved by washing four times for a total of 20 min in PBS (pH '7.8) with 50 mM NaCI, 25 mM KC], 5 mM CaC12, 1 mM with 1% bovine serum albumin (BSA). These grids were MgS04. then incubated for 30 min in a freshly prepared dilution of Nuclearstaining: Adult virginmales or young adult goat anti-rabbit immunoglobulin conjugated to gold beads hermaphrodites were hand dissected in SM, and the gonads (AuroprobeEM, Janssen Pharmaceuticals). The grids were were transferred to 100 pg/ml 4,6-diamidino-2-phenylin- washed thoroughly in PBS with 1% BSA followed by washes dole (DAPI) (Sigma, St. Louis) in phosphate-buffered saline in 0.05% Tween-20 and distilled water. The grids were (PBS, pH 7.0) and incubated in a moist chamber in the dark stained with uranyl acetate and lead citrate and carbon- for 30 min. The stained gonads were mounted on glass stabilized before examination. slides and examined under the microscope. Cytolocalization of MSP and : Adult virgin males were hand dissected in SM containing 10% polyvinyl-pyr- RESULTS rolidone (w/v), and the testes were transferred to a poly-^- lysine spotted glass slide. A small drop of 4% paraformal- spe-6 mutant phenotype:All four of the sped alleles dehyde in SM was added to cover the testes, and they were have essentially identical phenotypes, which are un- then incubated in a moist chamber for 30 min at room changed over the deficiency eDf2(III): they produce temperature. A cover glass was placed on top of the testes, no progeny as hermaphrodites, and spermatogenesis and pressure was gently applied. The slide was placed on a arrestswithout formation of spermatids(SHAKES block of dry ice for 5 min, and the coverslip was removed while the tissue was still frozen. The tissue was rinsed in 1988). PBS (pH 7.0) three times for a total of 15 min and then Figure la shows the testis of a heterozygous (spe- rinsed in PBS + 10 mg/ml glycine five times for a total of 6(hc49)/+) male with many spermatids (arrow) which 25 min to block unreacted aldehydes. After permeablizing is indistinguishable fromwild type. Figure lb shows a the tissue with 0.5% Triton X-100 (Sigma) in PBS for 4-5 homozygous see-6 mutant testis that contains defective min, it was rinsed in PBS followed by incubation for 15 min in PBS + 5% goat serum. The tissue was then incubated in spermatocytes (arrow). The defective spermatocytes 1:200 dilution of rabbit anti-MSP serum (BURKEand WARD show no indication of forming spermatids (Figure 2a) 1983) for 30 min and rinsed in PBS + 5% goat serum before unlike the wild type (Figure 2b). Chromosomal stain- counterstaining with appropriate fluorescent-tagged anti- ing with DAPI suggests that the nuclei of these sper- bodies. The same protocol was followed for tubulin staining matocytes are arrested neardiakinesis, since fully con- except that anti-tubulin antibodies to yeast a-tubulin [kindly provided by A. ADAMS(KILMARTIN, WRIGHT andMILSTEIN densed chromosomes are revealed by chromosomal 1982)] were used. A Bio-Rad confocal microscopewas used spreading (Figure2). for generating some immunofluorescence images. When mutant spermatocytes are stained with anti- Chromosome cytology: Adult virginmales were hand tubulin antibodies (Figure3, a and c) to reveal spindle dissected in SM, and the testes were transferred to a well and DAPI to reveal chromosomes (Fig- slide containing fresh fixative (methano1:glacial acetic acid ure 3, b andd), the firstmeiotic prophase looks 3:l ratio) at room temperature for 10-15 min. The tissue was then gently transferred into a drop of PBS on a clean normalup to diakinesis. Then it appearsthat the slide and rinsed three times by blotting off the excess PBS. chromosomes fail tomove to the metaphase plate, C. elegans spe-6 Gene Analysis 81

FIGURE‘L.--Nomarski inlage of DAPI-stained spe-6(hc49) mutant spermatocytes showing undivided nuclei (a). In contrast, wild-type spermatocytes (b) show four haploid spermatids and a central resid- ual cytoplast. Chromosome spreads of these defective spermatocytes show individual chromosomes condensed as at diakinesis (c). The chromosomes remain loosely clustered near the center of the cell.

was performed on conventionally stained and anti- FIGURE1 .-Light micrographs of testes from ‘L-day-old males. (a) spe-6(hc49)/+ heterozygotes showing large number of spermatids MSP immunogold labeled thin sections in order to (arrow). (b) spe-fi(hc49) homozygotes showing defective spermato- examine the FB-MO complexes. The wild type ap- cytes (arrow). pearance of the FB-MO complex is shown in Figure 5a where the MSP fibers are clearly seen surrounded although they become more loosely associatedthan at by the membranous organelle membrane. In contrast, diakinesis (Figure 3, e and Surprisingly, although Q. no fibers are seen in spe-6 mutant spermatocytes (Fig- the chromosomes do not segregate, the spindle pole ure 5b). The MOs appear similar to wild type in young bodies replicate and separate so that the final arrested spermatocytes often have four half-spindles located at defective Spermatocytes with their characteristic the cell periphery oriented perpendicular to each spherical head, electron dense collar and invaginated other (Figure 3e). The condensed chromosomes re- membrane body (Figure 5) (WOLF, HIRSHand Mc- mainloosely clustered near the middle of the cell INTOSH 1978; WARD,ARGON and NELSON 198 1 : ROB- (Figure 39. ERTS, PAVALKO and WARD 1986). In older spermato- To determine if these defective spermatocytes ac- cytes, the MO’s invaginated membranes, which would cumulated the major sperm protein (MSP), they were normally fold around the MSP fibers, balloon out into stained for immunofluorescence with anti-MSP anti- a vesicle (Figure 5c). Immuno-electron microscopy bodies. In wild-type spermatocytes, MSP staining reveals that MSP is found throughout both the cyto- shows oblong patches representing the fibrous body plasm and nuclei of mutant spermatocytes(Figure 6b), (Figure 421 and WARDand KLASS 1982). In contrast, but is confined to only the FBs of wild-type sperma- no such patches were observed in spe-6 mutant sper- tocytes (Figure 6a and ROBERTS, PAVALKOand WARD matocytes; instead, diffuse bright staining was ob- 1986). served throughout the cytoplasm (Figure 4b). All aspects of the homozygous spe-6(hc49) pheno- This diffuse immunofluorescencestaining with anti- type were indistinguishable fromthe spe-G(hc49)leDfl MSP antibodies suggest that the FBs might be either phenotype. The deficiency eDfl deletes the right arm missing or defective in spe-6 mutants. Consequently, of chromosome III, which includes the spe-6 locus. As electron microscopyof spe-6 mutant spermatocytes discussedbelow, spe-6 showsanomalous failure in 82 J. P. Varkev et al.

I;IGVRE :~.-,~~lti~t~~l~~~li~l-(it, c, e) ;tnd1)AI'l-st;ti~tcd (b. d. 1) i1n;tges of wild-type (a. 1)) and spe-6 mutant testis (c-Q. The spe- 6(hr4Y) mutant spermatocytes initially form spindles (c, arrow) that look identical to the wild type (a, arrow).The mutant spermatocytes then fail to undergo nuclear division (d. r) but their spindle pole bodies still replicate, resulting in cells with four half spindles (e, arrow). The chromosomes in the defective cells remain condensed in a loose group near the centerof the cell (f, arrow). (Bar represents 1 0 pm.) complementation to the chromosome IV deficiencies pDfl8 and eDflY. The sterile phenotype of spe-6/+ 111; eDfl8/+ IV or eDflY/+ IV was similar to spe-6 homozygotes that were wild type for chromosome IV. "I he testes of trans-heterozygotes were also examined FIGURE4.-~hlf<~ll microscopic images of'anti-\lSI' antibody by light and electron microscopy. Although some of staining of wild-type (a) and spP-h(hcY2)rnutant sperrnatocytes (b). In mutant cells the anti-MSP antibodies show diffuse but strong the testes examined had a few spermatids, immunoflu- staining in contrast to the distinct oblong staining of the FBs in the orescence microscopy with anti-MSP antibodies and wild-typecells. Both preparations were stained in parallel and ultrastructural examination includingimmunogold photographed at identical exposure. (Bar represents 10 pm.) labeling of MSP showed that, as in sped homozygotes, spe-6(hc4Y)/+; eDflY/+ mutant spermatocytes fail to (spe-6(hc4Y) and spe-6(hcY2)) that blocked spermato- form the FBs, and the MSP was distributed through- genesis prior to the formation of haploid spermatids out the cytoplasm (data not shown). (SHAKFS 1988). Another allele, spe-6(hc143), was ob- Interactions between spe-b(hc143)/+ andeither tained serendipitously in a search for mutations on eDflN/+ IV or eDflY/+ IV appear to be weaker than chromosome IV because it failed to complement a the other spe-6 alleles and, on this basis, this allele chromosome IV deficiency, eDflY(IV), as a trans-het- might be hypomorphic (see below and Table 1). How- erozygote. This small deficiency on the right arm of ever, the otherspe-6 alleles are likely to be null because chromosome IV, as well as an overlapping small defi- en)/+111 has the same phenotype as spe-6/+ III when ciency, eDfl8(IV), deletes a cluster of eight sperm- either construct is trans-heterozygous with either specific genes which had been identified molecularly, eDfl8/+ IV or eDflY/+ IV. as well as other genes (WARDet al. 1988; S. WARD, Genetic interactions between spe-6 and a region unpublished). of chromosome N: The spe-6 gene on chromosome To test if this non-complementation between un- III was initially identified by two sterile mutant alleles linked trans-heterozygotes was specific to either the C. elegans spe-6 Gene Analysis 83

FIGURE6.-1mnluno-electron micrograph of wild tvpe (a) and 5pt-6 mutant (b) spermatocytes showing the localization of MSP using anti-MSP antibodies. In wild type cells the gold particles are localized to the FBs (arrow) while in mutant spermatocvtes the gold particles itre scattered throughout the cvtoplasm and the nucleus (N). FIGURE5.-Convention;d tr;lnsmission electron microscopic im- . Surprisingly, the spe-6(hc143) allele comple- ages of wild type (a) and spe-6(hr4Y) mutant (b, c) spermatocytes. mented better than the others, yielding only about The FB-MO complex which is clearly distinguishable in wild-type cells (a. arrow) is missing in the mutant cells. However, normal MOs 10% spermatogenesis-defective sterile progeny. Since are seen in young mutant spermatocytes (b. arrow). In older defec- a11 four spe-6 alleles failed to complement the defi- tive spermatocytes, the membranes of the MOs balloon out (c, ciency, we tested whether a deficiency of the spe-6 arrow). The individual condensed chromosomes can be seen in region, eDj2(1II). which deletes the entire spe-6 gene, many of these cells (n). wouldalso fail to complement thechromosome IV deficiencies. As shown in Table 1, the trans-hetero- deficiency or to the spe-6(hc143) allele, all spe-6 alleles zygotes of these two deficiencies were also sterile. This were tested against both deficiencies (Table 1). All of suggests thatthe apparent interaction between the the other alleles fail to complement either deficiency spe-6 gene and the unlinked deficiencies is due to the 21s trans-heterozygotes, as indicated by the approxi- reduced dosage of hemizygous genes and not due to mately 25% sterile progeny among the F1 progeny of allele-specific interactions. the crosses described in the legend to Table 1. These As a control for the specificity of this interaction, sterilehermaphrodites had theexpected spermato- we performed complementation tests with two other genesis-defective phenotype, laying many unfertilized Spemutants that arrest development as spermato- 84 J. P. Varkey et al.

TABLE 1 lead to aberrant spermatocytesthat do not divide Phenotypes of F1progeny from crosses between chromosomeN further to form haploid spermatids. The eartiest and deficiencies eBfI8 or eDfl9 with thedifferent Spe mutant most strikingdefect in thesespermatocytes is the strains as wellas with chromosomeIII deficiency eDf2 absence of MSP filaments that normally form the fibrous bodies. MSP is present in the mutant sperma- Cross F, tocytes as shown by both light microscopic immuno- spe-6(hc49)III X eDfl8(IV) 318s fluorescencestaining and electron microscopic im- spe-6(hc49)III X eDfl9(IV) 611 4s munogold staining. However, the MSP does not lo- spe-fj(hc92) III X eDfl8(IV) 119s spe-6(hc92)III X eDfl9(IV) 3112s calize or assemble into filaments associated with the spe-6(hc 146) III X eDfl8(IV) 5138s membranous organelles, where they normally form, spe-6(hc146)III X eDfl9(IV) 3112s or into filament bundles elsewhere in the cytoplasm. spe-6(hc 143) III X eDfl8(IV) 3142s Because neither cytological staining procedurefor spe-6(hc 143) III X eDfl9(IV) 41383 MSPis quantitative, we cannotbe sure that the eo!0 X eDfl8(IV) 31 16s eo! (114 X eDfl9(IV) 31 16s amount of MSP is normal, but theintensity of staining spe-26(hc138ts)IV X eDfl8(IV) 011 2s suggests it cannot be substantially less than wild type. spe-26(hc138ts)IV X eDfl9(IV) 011 2s A possible explanation of the failure of MSP fila- spe-4(hc78)I X eDfl9(fV) 0136s spe-32(it129ts)IV X eDfl8(IV) 0152s ment assembly in spe-6 mutant spermatocytes is that spe-7(mn242)I1 X eDfl8(IV) 011 1s the function of the spe-6 gene product is to nucleate spe-7(mn242)I1 X eDfl9(IV) 011 2s the assembly of the MSP filaments. Since this assembly efl91+ (w 011 00s is normally associated with the membranesof the MO, @fl8/+ (IV) 011 00s and most of the filaments appear to end close to the eDf2/+ 0150 S membrane (WARD,ARGON and NELSON198 1 ; ROB- The column “cross”summarizes the relevant genes in each cross, and the column “F,”shows the fraction of sterile progeny (S) in the ERTS, PAVALKO and WARD 1986), thenormal spe-6 FI. All of these sterile progeny laid large numbers of oocytes but gene product could be a MO that no embryos or progeny, like spe-6 homozygotes. The actual geno- promotes MSP assemblyeither by nucleating assembly types of the parents used for the crosses are the following. The eDfl8 and eDfl9 hermaphrodites used in all crosses were +eDfl8 and remaining associated with the end of each fila- +/unc-24(e138)+dpy-20(el282)IV and +eDfl9 +/unc-24(e138) ment or perhaps by binding MSP and increasing its +dpy-20(e1282)1V. The heterozygous males of the Spe genes used local concentration. If the spe-6 gene product were a were spe-6(hc49) dpy-l8(e364)/++ Ill, spe-6(hc92)unc-32(e189)/++ lII,spe-6(hc143)dpy-18(e364)/++III,spe-6(hc146)dpy-l8(e364)/++ MO membranecomponent then its absence there Ill, spe-4(hc78)unc-5(e6 I)/++ I, spe-32(it129ts) dpy-20 (el 282)/++ could also explain why, following an initially normal and spe-7(mn242) unc-4(e120)/++ II. Homozygous males of the temperature-sensitive allele of spe-26(hc138ts)were used at permis- appearance, the MO membrane eventually becomes sive temperature for crosses with the deficiency strains. Chromo- distended and opens out into a vesicle in the terminal some III deficiency heterozygous males eo!/+ (generated by cross- defective spermatocytes. ing eDfl/eDfl/eDp6 hermaphrodites with N2 males) were used for generating the double deficiency strains. The fraction of sterile The geneticinteraction between spe-6 mutations progeny obtained from crosses with all spe-6 alleles with chromo- and the deficiencies on chromosome ZV is a striking some IV deficiencies was not significantly different from the ex- example of second site non-complementation. Many pected 25% for such a non-complementation test (WILKS1935; WILLIAMS1976). The only exception was the allele hc143 which other examples have been reported, most commonly, showed a weaker interaction with both deficiencies. allele-specific interactions (ATKINSON1985; KLEIN and DEPPE1985; KUSCH and EDGAR1986; FULLERet cytes, spe-26(hc138ts) IV and spe-T(rnn242) They II. al.1987; RINE and HERSKOWITZ 1987; DUTCHER, show no apparent interaction as trans-heterozygotes GIBBONSand INWOOD 1988; HOMYK and EMERSON with either eDfl8 or eDfl9 (Table 1) or with spe-6 1988; REGAN and FULLER 1988; STEARNS andBOT- alleles. In addition,the heterozygous deficiencies themselves yield no sterile worms (Table 1). STEIN 1988;FULLER et al. 1989; HAYS et al.1989). In an attemptto identify individual genes that failed These are usually interpreted to imply specific physi- to complement spe-6/+ heterozygotes, EMS-mutagen- cal interaction between the protein products of the ized males were crossed to spe-6 hermaphrodites and mutated genes. For example, certain &-tubulin alleles sterile F1 progeny were selected. From approximately fail to complementcertain a-tubulin mutations in 3000 mutagenized chromosomes, one new allele of Drosophila (FULLERet al. 1989;HAYS et al. 1989). spe-6, hc146, was found, but no unlinked mutations Since it is known thatthe a- and /3- form were identified. heterodimers, the non-complementation could result fromthe formation of nonfunctionalheterodimers that reduce the concentration for assembly of micro- DISCUSSION tubules (STEARNSand BOTSTEIN1988; FULLERet al. In spe-6 mutants,spermatogenesis appears to be 1989). Alternatively, the presence of mutant hetero- normal up to diakinesis, but all of the mutant alleles dimers could interfere withassembly or poison the C. elegans spe-6 Gene Analysis 85

function of assembled microtubules(STEARNS and NETT and WARD1986), one would expect Caenorhab- BOTSTEIN1988). ditis MSP assembly tobe similar. Indeed, in vitro Second site non-complementation can also be non- filament assembly of Caenorhabditis MSP has recently allele-specific. Microtubules again provide a clear ex- been achieved following theprocedure for Ascaris ample. Mutations in the TUB-1 gene of yeast fail to MSP, and it is also highly concentration dependent complement null mutations in TUB-3 gene as trans- (SMITH1992). If the sperm-specific genes under the heterozygotes (STEARNSand BOTSTEIN1988). Both eDfl9 deficiency, as well as the spe-6 gene product, TUB-1 and TUB-3 genes code for a-tubulinswhich are both participate in MSP assembly in vivo, then a gene present in microtubules. It is thought that the non- dosage dependent interaction among them is not sur- complementation between alleles of these two genes prising, since filament assembly is so concentration is due to the decreased dosage of their functionally dependent. There could be a single critical gene un- interchangeableproducts (STEARNS and BOTSTEIN der thedeficiency, or perhaps it is the reducedexpres- 1988). sion of a combination of several genes which causes Since all four spe-6 alleles fail to complement both the defect. We have attempted to distinguish these deletions eDfl8 and eDfl9, and most importantly, so alternatives by selecting for new mutations that fail to does the deficiency eo$?, the second site non-comple- complement spe-6 mutations, but as described in the mentation described here is not allele-specific. It must results, this experiment identified only a new spe-6 be due to the decreased gene dosage of the hemizy- allele and did not identify unlinked mutations. There- gous genes and could occur in several ways. Like the fore, a more extensive screen will be needed to see if yeast a-tubulin case, there could be genes similar to there are mutations mapping under eDfl9 that fail to spe-6 under the chromosome ZV deficiencies whose complement spe-6 mutations. Several new spermato- products participate in a particularly concentration- genesis-defective mutationsthat map under eDfl9 sensitive reaction. Alternatively, there could be dis- have been isolated recently, but they all complement tinct genes underthe deficiencies whose products spe-6 u. VARKEY,unpublished observations). interact with the spe-6 gene product during a devel- It is not clear how the failure to assemble MSP into opmental pathway that is sensitive to concentration. filaments in spe-6 mutant spermatocytes is related to If the spe-6 gene product functions to nucleateMSP the failure of the nucleus to divide in these cells. The filament assembly, why should spe-6/+ heterozygotes meiotic nuclei arrest near diakinesis without chromo- fail to complement eDfl8/+ or eDfl9/+ heterozy- some segregation, but the meiotic spindle pole bodies gotes? We do not know all of the genes under these duplicate and appear to orient at rightangles. This is deficiencies, but we do know that there is a cluster of similar to mutations in the NDCl gene of yeast, which sperm-specific genes which include four expressed disruptchromosome segregation without affecting MSP genes and four additional sperm-specific genes progression of the cell cycle (THOMASand BOTSTEIN identified molecularly because they encodesperm- 1986). One possible explanation for the lack of seg- specific mRNAs (WARDet al. 1988; s. WARD,unpub- regation of chromosomes in spe-6 mutants is that this lished). In the heterozygous deficiencies the products is theconsequence of unpolymerized MSPin the of all these genes would be expected to be abouthalf cytoplasm and nucleus that interferes with chromo- the normal amount. It is unlikely that the interaction some attachment to the spindle or with chromosome with spe-6 would be caused by the absence of four movement. Alternatively, the spe-6 gene productitself copies of MSP genes alone since there are about 30 could play arole in chromosomeattachment and movement as well as in MSP assembly. The observa- nearly identical functional MSP genes in the genome tion that, in spite of the failure of chromosome seg- (BURKE and WARD1983; WARDet al. 1 988), so halving regation, the spindle pole bodies still mature, dupli- the expression of only four of them would only reduce cate andorient at right angles shows that spindle the expected amount of MSP by about 7%. It is more maturation is not dependent on chromosome segre- likely that the interaction is due to one or moreof the gation. Thus, these spe-6 mutations, like mutations in other identified sperm-specific genes or other as yet several previously described spermatogenesis-defec- unrecognized genes under these deficiencies. tive genes, reveal that precisely ordered steps in the MSP filament assembly in vitro has been recently complex assembly of the sperm can be separated from achieved for Ascaris MSP utilizing alcohols and other each other yetstill proceedindependently (WARD, agents that reduce wateractivity to stimulate filament ARGONand NELSON198 1; L’HERNAULT, SHAKES and formation (KING et al. 1992). Like the assembly of WARD1988; SHAKESand WARD 1989; L’HERNAULT other filaments, such as microtubules and microfila- and ARDUENGO199 1). ments, Ascaris MSP filament assembly is concentra- tion dependent and requires a critical threshold con- This work was supported by U.S. Public Health Service grant centration. Since the Ascaris protein is 86% identical GM25243 to S.W. and Institutional Research Support grant IN1 10 to the consensus Caenorhabditis MSP protein (BEN- from American Cancer Society to J.P.V. Strains were obtained from 86 J. P. Varkey et al. the Caenorhabditis elegansGenetic Stock Center supported by NIH. L'HERNAULT,S. W., D. C. SHAKES and S. WARD, We thank A. ADAMSfor the anti-tubulin antibodies and the Neu- 1988 Developmental genetics of chromosome I spermatogen- robiology Program for use of the Confocal Microscope. We thank esis-defective mutants in the nematode Caenorhabditis elegans. members of the laboratory for helpful suggestions. Genetics 120 435-452. REGAN,C. L., and M. T. FULLER,1988 Interacting genes that LITERATURECITED affect function: the nc2 allele of the hapire locus fails to complement mutations in the testis-specific @-tubulin ANDERSON,P., and S. BRENNER, 1984A selection of heavy- gene of Drosophila. Genes Dev. 2: 82-92. chain mutant in the nematode C. elegans. Proc. Natl. Acad. Sci. RINE, J., andI. HERSKOWITZ, 1987Four genes responsible for a USA 81: 4470-4474. position effect on expression from HML and HMR in Saccha- ATKINSON,K. D., 1985 Two recessive suppressors of Saccharomy- romyces cerevisiae. Genetics 116 9-22. ces cereuisiae CHOl that are unlinked but fallin the same ROBERTS,T. M., and K. L. KING, 1991 Centripetal flow and complementation group. Genetics 111: 1-6. directed reassembly of the major sperm protein (MSP) cyto- BENNETT,K. L., and S. WARD, 1986 Neither a germ line-specific skeleton in the amoeboid sperm of the nematode, Ascaris suum. nor several somatically expressed genes are lost or rearranged Cell Motil. 20 228-241. during embryonic chromatin diminution in the nematode As- ROBERTS,T. M., F. M. PAVALKOand S. WARD,1986 Membrane caris lumbricoides var. suum. Dev. Biol. 118: 141-147. and cytoplasmic are transported in the same organelle BRENNER,S., 1974 The genetics of Caenorhabditiselegans. Ge- complex during nematode spermatogenesis. J. Cell Biol. i02: netics 77: 7 1-94. 1787-1 796. BURKE,D. J., and S. WARD,1983 Identification of a large multi- SHAKES,D. C.,1988 A genetic and pharmacological analysisof gene family encoding the major sperm protein of Caenorhabditis spermatogenesis in the nematode,Caenorhabditis elegans. Ph.D. elegans. J. Mol. Biol. 171: 1-29. Thesis, The Johns Hopkins University. DUTCHER,S. K., W. GIBBONSand W. B. INWOOD,1988 A genetic SHAKES,D. C., and S. WARD,1989 Mutations that disrupt the analysis of suppressors of the PF10 mutation in Chlamydomonas morphogenesis and localization ofa sperm-specificorganelle in reinhardtii. Genetics 120 965-976. Caenorhabditis elegans. Dev. Biol. 134 307-3 16. FULLER,M. T., J. H. CAULTON,J. A. HUTCHENS,T. C. KAUFMAN SEPSENWOL,S., H. RIS and T. M. ROBERTS,1989 A unique and E. C. RAFF,1987 Genetic analysis of microtubule struc- cytoskeleton associated with crawling in the amoeboid sperm ture: a @-tubulinmutation causes the formation of aberrant of the nematode, Ascaris mum. J. Cell Biol. 108: 55-66. microtubule in vivo and in vitro. J. Cell Biol. 104 385-394. SMITH,M. J., 1992 InVitro filament formation by major sperm FULLER,M. T., C. L. REGAN, T. S. HAYSand L. L. GREEN, protein (MSP) of Caenorhabditiselegans. M.S. Thesis, University 1989 Interacting genes identify interacting proteins involved of Arizona. in microtubule function in Drosophila. Cell Motil. Cytoskeleton STEARNS,T., and D. BOTSTEIN,1988 Unlinked noncomplemen- 14: 1-8. tation: isolation of new conditional-lethal mutations in each of HAYS,T. S., R. DEURING,B. ROBERTSON,M. PROUTand M. T. the tubulin genes of Saccharomyces cerevisiae. Genetics 119 FULLER,1989 Interacting proteins identified by genetic in- 249-260. teractions: a missense mutation in a-tubulin fails to complement THOMAS,J. H., and D. BOTSTEIN,1986 A gene required for the alleles of the testis-specific @-tubulingene of Drosophila mela- separation of chromosomes on the spindle apparatus in yeast. nogaster. Mol. Cell. Biol. 9 875-884. Cell 44 65-76. HODGKIN, J.,1986 Sex determination in the nematode C. elegans: WARD,S., 1986 The asymmetric localizationof gene products analysis of tra-3 suppressors and characterization offem genes. during the development of Caenorhabditis elegansspermatozoa, Genetics 114 15-52. pp. 55-75 in Gametogenesis and the Early Embryo, edited by J. HOMYK, T., JR., andC. P. EMERSON,JR., 1988 Functional inter- G. GALL.Alan R. Lis, New York. actions between unlinked muscle genes within haploinsufficient WARD,S., Y. ARGONand G. A. NELSON, 1981 Sperm morpho- regions of the Drosophila genome. Genetics 119 105- 12 1. genesis in wild-type and fertilization-defective mutants of Cae- KILMARTIN,J. V., B. WRIGHTand C. MILSTEIN,1982 Rat mono- norhabditis elegans. J. Cell Biol. 91: 26-44. clonal anti-tubulin antibodies derived by using a new nonse- WARD,S., and M. KLASS, 1982 The location of the major protein in Caenorhabditis elegans sperm and spermatocytes. Dev. Biol. creting rat cell line. J. Cell Biol. 93: 576-582. KING, K. L., M. STEWART,T. M. ROBERTSand M. SEAVY, 92: 203-208. WARD,S., D. J. BURKE,J. E. SULSTON,A. R. COULSON,D. G. 1992 Structure and macromolecular assembly of twoiso- ALBERTSON,D. AMMONS, M. KLASS and E. HOGAN, forms of themajor sperm protein (MSP) from the sperm of the 1988 Genomic organization of major sperm protein genes nematode, Ascaris mum. J. Cell Sci. 101: 847-857. and pseudogenes in the nematode Caenorhabditiselegans. J. KLEIN,K. K., and C. S. DEPPE,1985 Complementation and non- Mol. Biol. 1-13. complementation among nonallelic mutations altering devel- 199: WILKS,S. S., 1935 The likelihood test of independence in contin- opment in Schizophyllum commune. Genetics 109 333-339. gency tables. Ann. Math. Statist. 6: 190-196. KUSCH,M., and R. S. EDGAR,1986 Genetic studies of unusual loci WILLIAMS,K., 1976 The failure of Pearson's goodness of fit that affect body shape of the nematode Caenorhabditis elegans statistic. Statistician 25: 49. and may code for cuticle structural proteins. Genetics 113: WOLF,N., D. HIRSH andJ. R. MCINTOSH,1978 Spermatogenesis 621-639. in males of the free-living nematode, Caenorhabditis elegans. J. ~.'HERNAULT,S. W., and M. ARDUENGO,1991 Genetic and mo- Ultrastruc. Res. 63: 155-169. lecular analysis of Spe-4. 1991 C. elegansMeeting (Abstr.), p. 366. Madison, Wisc. Communicating editor: R. K. HERMAN