fli-1(ky535)affects germ line morphogenesis
The ky535 mutation was isolated in a synthetic lethal screen to identify molecules that act in parallel to the actin-binding protein UNC-115 abLIM [24]. UNC-115 and FLI-1 likely have roles in pharyngeal function underlying the synthetic lethal phenotype. Pharyngeal pumping is severely reduced in unc-115; fli-1 double mutants, and double mutants arrest in the L1 larval stage consistent with a feeding defect (data not shown).
Alone, fli-1(ky535) animals were viable and fertile and displayed a slightly Dumpy (Dpy) body morphology. When observed using differential interference contrast (DIC) microscopy, germ line nuclei were observed in the rachis of the meiotic zone (compare Figures 1A and 1C to Figures 1B and 1D). In most cases, chains of apparently connected nuclei spanned the rachis. Misplaced germ cells in the rachis were observed as soon as the rachis was evident in mid-to-late L4 larval animals (data not shown). This phenotype is referred to here as the germ line morphogenesis (Glm) phenotype. In fli-1(ky535), 94% of gonad arms displayed the Glm phenotype (Figure 2). The Glm phenotype was never observed in wild-type animals.
Transition from mitosis to meiosis is not disrupted in fli-1(ky535)mutant germ nuclei
The misplaced nuclei in the rachis of the meiotic zone in fli-1 might have been due to disruption in the transition of nuclei from mitosis to meiosis. A BrdU incorporation was used to assay nuclei undergoing DNA synthesis in the germ line (e.g. those that have undergone mitosis or S phase of meiosis I) (see Methods) [25]. After 10 minutes of exposure to BrdU, wild-type animals displayed BrdU-positive nuclei in the distal mitotic zone (Figure 1G). fli-1(ky535)-mutant gonads displayed a similar BrdU incorporation profile (Figure 1H), and nuclei in the rachis of the meiotic zone did not incorporate BrdU. A 30-minute exposure to BrdU also resulted in no apparent differences between wild-type and fli-1(ky535) (data not shown). In sum, no differences in BrdU incorporation were detected between wild-type and fli-1(ky535), suggesting that misplaced nuclei in the rachis of the meiotic zone of fli-1(ky535) were not undergoing mitotic divisions, and that normal meiotic progression was not affected (e.g. meiosis I was not delayed). DAPI staining to assay nuclear morphology showed that misplaced nuclei in the rachis of the meiotic zone of fli-1(ky535) animals displayed a meiotic pachytene morphology; the pachytene chromosomes were individually visible with a "bowl of spaghetti" appearance (Figure 1D) [2].
In transmission electron microscopic (TEM) cross-sections, nuclei in the meiotic zone in fli-1(ky535), including misplaced nuclei, were of roughly the same size and shape as those in wild-type (Figure 3; 2.9 ± 0.06 μm diameter for fli-1(ky535) and 3.0 ± 0.6 μm diameter for wild-type). The misplaced nuclei in fli-1(ky535) had a meiotic appearance (Figure 3). Meiotic nuclei appear round and regular as those seen in Figure 3C and 3D, whereas mitotic nuclear membranes have an irregular, "wavy" appearance. These lines of evidence indicate that the transition from mitosis to meiosis is unaffected in fli-1(ky535) mutant germ nuclei.
The previously-published fli-1(bp130) allele caused defects in oocyte production and brood size [21]. Brood size of fli-1(ky535) was comparable to that of wild-type (an average of 272 progeny for fli-1(ky535) compared to 319 for wild type; t-test p = 0.11). Possibly, bp130 is a stronger allele of fli-1 than is ky535 and affects oocyte production more strongly than ky535.
Germ line plasma membrane partially surrounded misplaced germ nuclei in fli-1(ky535)
TEM analysis revealed that wild-type meiotic zone nuclei were near the cortex. The germ line plasma membrane protruded between and partially enveloped each nucleus, forming a characteristic "T" shaped membrane described above and elsewhere (Figure 3A and 3C) [4, 5]. In TEM cross-sections of meiotic regions of fli-1(ky535), germ line plasma membrane was clearly associated with each misplaced nucleus in the rachis, suggesting that germ line plasma membrane invaginated to partially enclose misplaced germ nuclei (Figure 3B and 3D and Figure 4). A similar phenotype was observed in cross-sections of animals heterozygous for a deletion of the fli-1 locus called tm362 (data not shown). No defects in the organization of the distal mitotic zone were observed in cross sections of fli-1(ky535) (e.g. the germ cell arrangement resembled wild-type and distal tip cell filopodia between germ cells was observed). While the shape and diameter of wild-type distal meiotic gonads was relatively uniform (a diameter range of 16–23 μm, n = 10), fli-1 gonads were often of irregular diameter (a range of 12–33 μm, n = 10) and irregular shape (compare Figures 3A with Figure 3B and 4A).
Sheath cell extensions were associated with misplaced nuclei in the rachis in fli-1(ky535)
The plasma membrane surrounding interior nuclei in fli-1(ky535) formed gaps between nuclei similar to the gaps formed by plasma membrane invagination around cortical nuclei (Figure 3B and 3D and Figure 4). In fli-1(ky535) mutants, additional membranes were frequently observed occupying these interior gaps formed by invaginated germ line plasma membrane (Figure 4B and 4C). Less frequently, electron-dense laminar structures were present in the interior gaps (Figure 4C). Cross sections of heterozygous fli-1(tm362)/+ deletion animals showed a similar phenotype (data not shown).
The nature of these membrane-like structures between misplaced germ cells observed by TEM was unclear. The germ line is partially surrounded by the somatic sheath cells, which extend filopodia across the bare regions of the germline not covered by the cell body [4]. Sheath cell protrusions occupy gaps between nuclei formed by germ line plasma membrane invagination. In wild-type, sheath cell protrusions do not extend deeply between nuclei but rather stay near the cortex [4]. Possibly, the membrane-like structures between misplaced germ nuclei in fli-1 mutants were somatic sheath cell extensions.
To assay sheath cell morphology, a transgene consisting of the lim-7 promoter driving gfp expression was analyzed. lim-7::gfp is expressed in the sheath cells but not the germ line [4]. In wild-type harboring lim-7::gfp, no GFP fluorescence was detected in the rachis of the meiotic region (Figure 5A, B, and 5G), although GFP was detected at the surface of the germ line in a "honeycomb" pattern as previously described [4], due to the thin cytoplasm of regions of the sheath cells covering the germ nuclei.
In fli-1(ky535) mutants, the cortical "honeycomb" pattern was observed, although it was often irregular and disorganized, suggesting cortical nucleus arrangement was disorganized. Fingers of GFP expression were observed protruding into the rachis and associating with misplaced germ nuclei (Figure 5C, D, and 5H). These protrusions were from the somatic sheath cells and not the distal tip cell, as a lag-2::gfp transgene, expressed only in the distal tip cell [9], did not show these patches in the rachis in fli-1(ky535) animals (data not shown). The TEM and lim-7::gfp results combined indicate that misplaced nuclei in the rachis were bounded by germ line plasma membrane, and extensions of the sheath cells protruded between the misplaced germ cells in the rachis (Figure 6 is a depiction of these results).
FLI-1 encodes a molecule similar to Flightless 1/Fliih
The ky535 mutation was mapped genetically to linkage group III by standard linkage analysis with visible markers using synthetic lethality with unc-115 (data not shown). Three-factor mapping with dpy-17 and unc-32 using the Glm phenotype indicated that ky535 was close to and to the left of unc-32 (approximately 0.22 cM) (Figure 7A). The fli-1 gene (B0523.5 on Wormbase), which encodes an actin-binding protein of the Flightless 1/Fliih family, resides in this region of the genome (Figure 7B and 7C). The FLI-1 polypeptide is composed of N-terminal leucine-rich repeats (LRRs) and 5 C-terminal gelsolin-like domains (Figure 7D). A fli-1 cDNA was isolated in a previous study [26] (U01183 in Genbank). This transcript was used as the basis for Figure 7C and 7D. The cDNA is likely to contain the entire fli-1 coding region, as an in-frame stop codon is present 5 codons upstream of the presumed initiation methionine (data not shown). Furthermore, two independent fli-1 cDNAs were sequenced (yk48g9 and yk294b7, courtesy of Y. Kohara). While incomplete at the 5' ends, these cDNAs were identical in structure to the U01183 cDNA.
To test if the fli-1 gene is involved in germ line morphogenesis, RNA-mediated gene interference (RNAi) of fli-1 was performed. fli-1(RNAi) phenocopied the germ line morphogenesis defect of ky535 (Figure 1F and Figure 2). Furthermore, the cosmid B0523, which contains the fli-1 gene, rescued the synthetic lethality of unc-115(mn481); fli-1(ky535) animals harboring a transgene containing the cosmid (Figure 7B). The B0523 cosmid contains two other genes, B0523.1 and B0523.3. RNAi of these genes did not cause a Glm phenocopy (data not shown). fli-1 RNAi in both wild-type and rrf-1(pk1417 and ok289) backgrounds caused the Glm phenocopy (Figure 1F and Figure 2). rrf-1 mutations attenuate RNAi in somatic cells but do not apparently affect RNAi in the germ line [27].
A PCR-generated fragment of B0523 containing only the fli-1 gene and a tryptophan tRNA (Figure 7C, see Methods) rescued the synthetic lethality of unc-115(mn481); fli-1(ky535) mutants (Figure 7C). Furthermore, a fli-1::gfp full-length fusion transgene (see Methods) partially rescued the Glm phenotype of fli-1(ky535) animals (Figure 2) as well as the lethality of unc-115(mn481); fli-1(ky535) double mutants. The nucleotide sequence of the entire region included in the rescuing fli-1(+) transgene was determined from ky535 mutants. No nucleotide changes were detected in this region in three independent PCR amplifications of the fli-1 gene from ky535 genomic DNA. Possibly, ky535 is regulatory mutation outside of the region necessary for rescue, and transgenic fli-1(+) expression, which can often lead to overexpression, can overcome the ky535 mutation. A fli-1 transcript was detected by RT-PCR in fli-1(ky535) mutants (data not shown). As described below, the fli-1 locus is haploinsufficient for the Glm phenotype, indicating that lowering fli-1 gene dosage by as little as one-half can cause the Glm phenotype.
The fli-1(tm362)deletion causes a germ line morphogenesis defect
To confirm that fli-1 controls germ line morphogenesis, a deletion in the fli-1 locus was analyzed (isolated and kindly provided by The National Bioresource Project for the Experimental Animal C. elegans, S. Mitani). The deletion, tm362, removed bases 10973 to 11931 relative to the cosmid B0523 (Genbank Accession number L07143) with breakpoints in coding exons 9 and 11 of fli-1 (Figure 7C). The out-of-frame tm362 deletion removed coding region encompassing parts of gelsolin domains 3 and 4 (Figure 7D).
fli-1(tm362) homozygotes from a heterozygous mother arrested during embryogenesis and failed to hatch. Of arrested embryos, 70% displayed the Pat phenotype (paralyzed and arrested at the two-fold stage of embryonic elongation) (Figure 7F). The Pat phenotype is characteristic of defects in body wall muscle function [28]. Indeed, fli-1(ky535) mutants displayed slightly disorganized myofilament lattice structure in body wall muscles (data not shown), suggesting that body wall muscle development was also affected by fli-1(ky535). The remaining 30% of tm362 homozygous embryos arrested earlier in embryogenesis with severe defects in embryonic organization (Figure 7E). Defects in muscle organization and embryonic development in a fli-1 mutation have been described [21].
While homozygous fli-1(tm362) animals arrested in embryogenesis, heterozygous tm362/+ animals displayed the Glm phenotype similar to ky535 animals (49%; Figure 1E and Figure 2). TEM cross sections of tm362/+ heterozygotes were analyzed and found to have a similar ultrastructural defect as described for fli-1(ky535) (data not shown), including germ line plasma membrane and sheath cell invaginations around misplaced nuclei. These results suggest that fli-1 is haploinsufficient for the Glm phenotype. Indeed, heterozygous ky535/+ animals also displayed the Glm phenotype (60% compared to 94% for ky535 homozygotes; Figure 2). Trans-heterozygous ky535/tm362 animals were viable and had a severe Glm phenotype (91%; Figure 2), suggesting that ky535 and tm362 failed to complement for this phenotype. However, the additive effect of each heterozygote alone could explain this effect.
The lethality of fli-1(tm362) was rescued by the fli-1(+) transgene that also rescued the unc-115(mn481); fli-1(ky535) lethality (Figure 7C), and the Glm phenotype of rescued homozygous fli-1(tm362) animals was significantly less severe than fli-1(ky535) homozygotes (Figure 2). The Glm phenotype was likely due to fli-1 loss of function, as fli-1 RNAi caused the Glm phenocopy and the Glm phenotype of both fli-1(ky535) and fli-1(tm362) was rescued by transgenic fli-1(+). Thus, the viable fli-1(ky535) allele might be hypomorphic and the lethal fli-1(tm362) allele might be null. If this is the case, fli-1 might be haploinsufficient for the Glm phenotype as heterozygotes displayed the Glm phenotype. It is also possible that either or both of the two fli-1 alleles are not simple loss-of-function alleles and thus cause a dominant Glm phenotype. Indeed, fli-1(tm362) was rescued more efficiently than ky535 by transgenic fli-1(+) (Figure 2), suggesting that ky535 might have some dominant character that is more difficult to rescue. In either case, the Glm defect is likely a loss-of-function phenotype of fli-1 as RNAi of fli-1 caused the Glm defect.
Germ line actin organization in fli-1(ky535)mutants
FLI-1 can bind to and sever actin filaments [20], suggesting that it might modulate cytoskeletal organization. The effect of fli-1(ky535) on the actin cytoskeleton of the germ line was analyzed by staining with rhodamine-labeled phalloidin. Hermaphrodite somatic sheath cells contain much actin, which was difficult to distinguish from germ line actin. To circumvent this problem, male gonads, which lack sheath cells, were analyzed, although hermaphrodites showed a pattern consistent with that seen in males (data not shown).
fli-1(ky535) males displayed a Glm phenotype similar to hermaphrodites, as displaced nuclei were observed in the rachis of the single male gonad arm (Figure 8A and 8B). In the wild type male germ line, a cortical layer of phalloidin staining was associated with the germ line plasma membrane that surrounded each germ nucleus (Figure 8C). In fli-1(ky535), nuclei at the cortex displayed an apparently normal actin organization. However, a cortical layer of actin was observed surrounding the misplaced nuclei in the rachis, apparently associated with the invaginated plasma membrane (Figure 8D). While actin was associated with misplaced nuclei in fli-1(ky535), no obvious defects in the organization of the actin cytoskeleton per se were observed. The fli-1(bp130) allele caused defects in gonad actin organization [21] not seen in fli-1(ky535). fli-1(ky535) might be a hypomorphic allele, and actin organization might not be affected to the extent observed in bp130.
fli-1is expressed in the gonad and in muscle
A transcriptional fli-1promoter::gfp reporter transgene was constructed that contained the fli-1 5' upstream region driving gfp (see Methods). Expression was observed in body wall muscle, pharyngeal muscle, and vulval muscle of embryonic, larval, and adult animals (Figure 9A). This is consistent with the Pat phenotype of fli-1(tm362) animals and the pharyngeal pumping defects of unc-115; fli-1(ky535) animals. In complex arrays (see Methods), expression was occasionally observed along the entire length the adult gonad in a "honeycomb" pattern characteristic of gonad expression (Figure 9B). This expression was faint and variable (not observed in all animals) and tended to dissipate as the complex array lines were maintained for more than three generations. This pattern could reflect expression in the germ line, the somatic sheath cells, or both. Male gonads, which are devoid of sheath cells, also showed faint and variable expression along their lengths (Figure 9B inset), suggesting that expression might be in the germ line. However, sheath cell expression cannot be excluded from these experiments.
The full-length fli-1::gfp transgene, predicted to encode a full-length FLI-1 polypeptide with GFP at the C-terminus, rescued fli-1 lethality and partially rescued the Glm phenotype of fli-1(ky535) and fli-1(tm362), suggesting the FLI-1::GFP molecule was functional. No FLI-1::GFP fluorescence was detected in the gonads of these transgenic animals, and muscle expression was very faint and inconsistent. Possibly, FLI-1::GFP was expressed at very low levels, below detection in the gonad.
To detect low levels of FLI-1::GFP expression, gonads from animals expressing full-length fli-1::gfp were excised and stained with an antibody against GFP. Specific GFP immunoreactivity was predominantly associated with germ nuclei (Figure 10A–F). Gonads from animals without the fli-1::gfp transgene showed no such reactivity (Figure 10G–I). FLI-1::GFP was associated with nuclei along the length of the entire gonad, and no obvious differences in FLI-1::GFP accumulation or nuclear association were detected along the length of the distal gonad from the mitotic zone through the meiotic zone.
let-60 mutations display a germ line morphogenesis phenotype similar to fli-1
Previous studies described defects in germ line organization in mutants of Ras signaling pathway components: mpk-1 and ksr-2 mutations caused germ line clumping [17, 29]; and mek-2 and let-60 Ras mutants displayed misplaced nuclei in the meiotic zone [16]. C. elegans LET-60 is similar to human k-Ras [30, 31], and has been shown to control transition of germ nuclei from meiotic pachytene to diakinesis and germ line organization [16].
To begin to characterize Ras signaling in the Glm phenotype, alleles of let-60 Ras that cause loss of function, constitutive activation, and dominant negative effects were analyzed for the germ line morphogenesis defect by DIC optics and DAPI staining (Figure 11A) [30, 32, 33]. The hypomorphic loss-of-function allele n2021 caused a ky535-like germ line defect in 44% of gonad arms, and the stronger let-60 loss-of-function alleles s1124, s1155, and s59, which are homozygous lethal, caused the Glm phenotype when heterozygous (52%, 23%, and 47%, respectively). These data suggest that let-60 is haploinsufficient for the Glm phenotype as was fli-1. Three different dominant-negative alleles of let-60 also displayed the germ line phenotype as homozygotes or as heterozygotes (e.g. 94% for homozygous sy93) (Figure 11A). TEM sections of let-60(n2021) showed a similar ultrastructural defect as fli-1(ky535) (data not shown), including germ line plasma membrane and sheath cell protrusions between misplaced nuclei. Furthermore, let-60 loss-of-function and dominant negative mutants displayed sheath cell lim-7::gfp expression associated with misplaced nuclei (Figure 5E, F, and 5I). While loss-of-function and dominant negative let-60 alleles caused the Glm phenotype, constitutively-active let-60 alleles n1700 and n1046 caused little or no Glm phenotype (Figure 11A).
LET-60 activity can compensate for loss of FLI-1 in germ line morphogenesis
fli-1 and let-60 Ras mutations cause the Glm defect, and constitutively-active let-60 alleles, which presumably cause let-60 overactivation, had no apparent effect on germ line morphogenesis. The constitutively-active let-60(n1700) mutation partially suppressed the Glm defect of fli-1(ky535) heterozygotes and homozygotes and fli-1(tm362)/+ heterozygotes (Figure 11B). For example, fli-1(tm362) heterozygotes displayed 49% defective gonad arms, reduced to 15% by heterozygous let-60(n1700). let-60(n1046), another constitutively-active let-60 mutation, suppressed the Glm phenotype of fli-1(ky535)/+ heterozygotes (60% versus 18%). These data indicate LET-60 Ras overactivation can partially compensate for loss of fli-1 function and suggest that fli-1 and let-60 Ras act together to control germ line morphogenesis. Possibly, FLI-1 and LET-60 act in the same pathway or in parallel pathways to control germ line morphogenesis. It is also possible that FLI-1 and LET-60 control each others' expression.