The C. elegans sex determination gene laf-1 encodes a putative DEAD-box RNA helicase

Abstract

The C. elegans gene laf-1 is critical for both embryonic development and sex determination. Laf-1 is thought to promote male cell fates by negatively regulating expression of tra-2 in both hermaphrodites and males. We cloned laf-1 and established that it encodes a putative DEAD-box RNA helicase related to S. cerevisiae Ded1p and Drosophila Vasa. Three sequenced laf-1 mutations are missense alleles affecting a small region of the protein in or near helicase motif III. We demonstrate that the phenotypes resulting from laf-1 mutations are due to loss or reduction of laf-1 function, and that both laf-1 and a related helicase vbh-1 function in germline sex determination. Laf-1 mRNA is expressed in both males and hermaphrodites and in both the germ line and soma of hermaphrodites. It is expressed at all developmental stages and is most abundant in embryos. LAF-1 is predominantly, if not exclusively, cytoplasmic and colocalizes with PGL-1 in P granules of germline precursor cells. Previous results suggest that laf-1 functions to negatively regulate expression of the sex determination protein TRA-2, and we find that the abundance of TRA-2 is modestly elevated in laf-1/+ females. We discuss potential functions of LAF-1 as a helicase and its roles in sex determination.

INTRODUCTION

C. elegans sex determination is a highly regulated process during which animals with two sets of autosomes and two X chromosomes (AA XX) develop as self-fertile hermaphrodites, while those with one X chromosome (AA XO) develop as males (reviewed in Zarkower 2006; Ellis, Schedl 2007). The core sex determination pathway in both somatic and germline cells involves a series of negative regulations among proteins that ultimately specify male or female cell fates (Fig. 1A). Stage-specific regulation of tra-2 and fem-3 in the hermaphrodite germ line facilitates production of sperm in the L4 larval stage and oocytes in the adult. Translation of tra-2 mRNA is repressed during the L2 and L3 larval stages, thereby allowing spermatogenesis. Translation of fem-3 mRNA is repressed in adults, thereby causing a switch to oogenesis (reviewed in Puoti, et al 2001; Ellis, Schedl 2007).

Figure 1. C. elegans sex determination.

(A) Simplified sex determination pathway. (B) Models for tra-2 regulation. Export of the tra-2 mRNA is regulated so that it exits the nucleus via a CRM-1-mediated pathway (Segal, et al 2001; Kuersten, et al 2004). GLD-1 binds to sequences in the 3′ UTR of the tra-2 mRNA (Jan, et al 1999) and acts in conjunction with FOG-2 (Clifford, et al 2000) to repress tra-2 translation.

Mutations of laf-1 affect sex determination, and genetic analysis suggests that LAF-1 promotes male cell fates. Laf-1 mutations were isolated as dominant suppressors of fem-3 gain-of-function (gf) alleles (Goodwin, et al 1997). The germ lines of fem-3(gf) XX animals are masculinized, containing excess sperm and no oocytes (Barton, et al 1987). Such animals are sterile, but heterozygosity of laf-1 mutations suppresses the sterility. Laf-1/+; fem-3(gf) animals produce both sperm and oocytes and are self-fertile (Goodwin, et al 1997). In an otherwise wild-type background, 10–30% of laf-1/+ XX heterozygotes are feminized, producing oocytes but no sperm (Goodwin, et al 1997). A similar percentage of XO animals are partially feminized in both the germ line and soma. These data suggest that wild-type LAF-1 promotes male fates in all tissues.

Previous data suggest that the function of LAF-1 involves repression of tra-2 expression (Goodwin, et al 1997). TRA-2 promotes female fates, and regulation of its expression is critical for normal sexual development in all tissues. Translation of tra-2 mRNA is repressed in the hermaphrodite germ line to allow spermatogenesis (Doniach 1986; Goodwin, et al 1993). This repression requires two 28-nucleotide elements termed TGEs (tra-2 and GLI elements) located in the tra-2 3′ untranslated region (3′UTR) (Goodwin, et al 1993; Jan, et al 1997). Disruption or deletion of these elements, such as in tra-2(gf) mutants, causes excess tra-2 activity and feminizes the hermaphrodite germ line (Doniach 1986; Schedl, Kimble 1988). The STAR protein GLD-1 binds to the TGEs and mediates tra-2 repression (Jan, et al 1999) (Fig. 1B). Regulation of tra-2 also requires FOG-2, an F-box protein that physically interacts with GLD-1 (Clifford, et al 2000). Expression of fog-2 and gld-1 is restricted to the germ line (Jones, et al 1996; Clifford, et al 2000), but tra-2 translation is also regulated in the soma (Goodwin, et al 1997). It unknown what factors mediate repression of tra-2 in the soma, but LAF-1 is a good candidate (Goodwin, et al 1997).

Transport of tra-2 mRNA from the nucleus to the cytoplasm is regulated. Most C. elegans mRNAs exit the nucleus via a NXF-1-mediated pathway, but tra-2 mRNA export occurs and is regulated by a leptomycin B-sensitive pathway that likely involves CRM-1 (Segal, et al 2001; Kuersten, et al 2004) (Fig. 1B). Alteration of tra-2 mRNA export such that it exits via the NXF-1-mediated pathway leads to increased levels of TRA-2 protein (Kuersten, et al 2004), suggesting that normal export is needed to establish the translational regulation of tra-2 mRNA described above.

Three lines of evidence suggest that laf-1 regulates tra-2. First, the phenotypes of laf-1/+ animals are similar to those of strong tra-2(gf) mutants. Both of these genotypes cause XX animals to develop as females and partially feminize the XO soma and germ line (Doniach 1986; Schedl, Kimble 1988; Goodwin, et al 1997). Additionally, both tra-2(gf) mutations and laf-1/+ heterozygosity suppress the germline masculinization of fem-3(gf) mutants (Barton, et al 1987; Goodwin, et al 1997). Second, tests of epistasis placed laf-1 upstream of tra-2. Tra-2(null); laf-1/+ double mutants are pseudomales indistinguishable from tra-2(null) single mutants (Goodwin, et al 1997). Third, reporter transgenes carrying the tra-2 3′UTR are misregulated in laf-1/+ mutants (Goodwin, et al 1997). Transgenes carrying the wild-type tra-2 3′UTR are repressed in wild type but are derepressed in laf-1/+ mutants. Transgenes that lack the TGEs fail to be repressed in wild type and exhibit no additional derepression in laf-1/+ heterozygotes, indicating that laf-1 acts through the TGEs (Goodwin, et al 1997). These data suggest that laf-1 acts upstream of tra-2 and inhibits its expression. Feminization of laf-1/+ heterozygotes may thus result from overexpression of TRA-2.

In addition to its role in sex determination, laf-1 is essential for embryogenesis. Laf-1 heterozygotes are feminized, but laf-1 homozygotes die as embryos or early larvae (Goodwin, et al 1997). The embryonic lethality of laf-1 homozygotes is probably not due to misregulation of tra-2, as such lethality is not associated with either increased or decreased tra-2 activity. Therefore, laf-1 likely regulates other mRNAs in the embryo or serves another function required for normal development.

This paper describes the molecular analysis of laf-1 and further characterization of its role in sex determination. We demonstrate that laf-1 encodes a putative DEAD-box RNA helicase related to S. cerevisiae Ded1p and Drosophila Vasa. We investigated laf-1 mRNA expression and the sub-cellular localization of LAF-1 protein. Finally, we tested whether TRA-2 expression is increased in laf-1/+ mutants.

MATERIALS AND METHODS

Worm strains

The following strains were used in this work: N2 Bristol (wild type), CB4856 (Hawaiian isolate), glp-4(bn2ts) I, tra-2(e2020gf) II, tra-2(q122gf) II, tra-2(e1095) II; unc-24(e138) fem-3(e1996)/++ IV, unc-119(ed3) III, glp-1(q231ts) III, laf-1(q267) unc-32/qC1 III, dpy-1(e1) laf-1(q267)/qC1 III, laf-1(q267)/+ III, dpy-1(e1) laf-1(q217)/qC1 III, laf-1(q217)/+ III, laf-1(q80)/qC1 III, fem-2(e2105)/unc-45(r450) dpy-1(e1) III, fem-1(e1965)/unc-5(e53) mor-2(e1125) IV, fem-3(e1996)/unc-24(e138) dpy-20(e1282) IV.

Fine mapping laf-1

dpy-1 laf-1(q267)/qC1 or laf-1(q267) unc-32/qC1 females were mated with males of the Hawaiian isolate CB4856, and F1 hermaphrodites were allowed to self-fertilize. Dpy non-Laf or Unc non-Laf F2 animals were isolated, and single nucleotide polymorphisms (snpA-snpF, see Fig. 2A) in the laf-1 region were assayed using single worm PCR followed by restriction enzyme digest or sequencing. From mothers of genotype dpy-1 laf-1(q267)/++(Hawaiian), 113/461 Dpy non-Laf recombinants were homozygous for the N2 Bristol allele of snpA. In 9/65 of these animals, the crossover occurred to the right of snpC and in 1/39 it occurred to the right of snpD, placing laf-1 to the right of snpD. From mothers of genotype laf-1(q267) unc-32/++(Hawaiian), 96/1048 Unc non-Laf recombinants were homozygous for the N2 Bristol allele of snpB. In 4/61 of these animals, the crossover occurred to the left of snpF and in 2/96 it occurred to the left of snpE, placing laf-1 to the left of snpE. SnpA-snpE are WormBase alleles snp_Y71H2B[2], pkP3093, snp_y71H2AM[4], hw41389, and hw41420, respectively. SnpF was identified by sequencing and is an A at nucleotide III: 2,793,914 in the Hawaiian isolate compared to a G in N2 Bristol.

Figure 2. laf-1 cloning and phylogeny.

(A) The laf-1 region of chromosome III. Above, the genetic map of the left arm of chromosome III between dpy-1 and unc-32. Below, a closer view of the laf-1 region of the physical map. SNP markers used for mapping are indicated, and the light gray region indicates the smallest interval into which laf-1 was mapped. The black arrow is laf-1, and gray arrows are other predicted open reading frames near laf-1. (B) laf-1 mRNA and protein. The laf-1 transcript is shown with boxes and lines representing exons and introns, respectively; the coding region is shaded gray. Laf-1 mRNA is trans-spliced to SL1. LAF-1 protein contains a DEAD-box RNA helicase domain. The nine conserved motifs, Q-VI, within the helicase domain are indicated by gray boxes. (C) Alignment of motifs II, III and surrounding regions of LAF-1 and related proteins. Shading indicates amino acids that are identical (black background) or similar (gray background) in >80% of the proteins compared. Conserved motifs II (DEAD) and III (SAT) are indicated below the alignment, and the amino acids affected by laf-1 mutations are identified by allele number above the alignment. (D) Phylogenetic tree showing LAF-1 and related proteins. The value at each node is the number of bootstrap replicates (out of 1000) that included that node. Letters in parentheses indicate species as follows:, Ce - Caenorhabditis elegans, Dm -Drosophila melanogaster, Hs - Homo sapiens, Mm - Mus musculus, Sc - Saccharomyces cerevisiae, Xl - Xenopus laevis.

Sequencing laf-1 mutations

Genomic DNA from laf-1/qC1 heterozygotes was PCR amplified in ~2kb sections using Elongase enzyme (Invitrogen) and cloned into a vector for sequencing. Base changes (relative to the sequence of wild-type C. elegans) seen in multiple clones were confirmed by sequencing PCR products from individual laf-1/qC1 heterozygotes or homozygous laf-1 dead eggs. Wild-type animals were also tested to verify the WormBase sequence.

Phylogenetic analysis of LAF-1 and related proteins

Proteins related to LAF-1 were identified by BLAST search through the UniProt website (http://www.pir.uniprot.org/), using the full LAF-1 protein sequence as the query. An alignment of the most closely related proteins from selected species was made with the ClustalX program (http://bips.u-strasbg.fr/fr/Documentation/ClustalX/) using a Gonnet 250 protein weight matrix. A phylogenetic tree based on the alignment was constructed by the Neighbor-Joining method using the Phylip software package (http://evolution.genetics.washington.edu/phylip.html). S. cerevisiae eIF4A, a more distantly related DEAD-box protein, was included in the analysis and assigned as the outgroup. The reliability of the tree was tested by bootstrap resampling with 1000 replicates. UniProt accession numbers for all included proteins are: S. cerevisiae: eIF4A (P10081), Ded1p (P06634), Dbp1p (P24784). C. elegans: GLH-1 (P34689), GLH-2 (Q966L9), VBH-1 (Q65XX1). D. melanogaster: Vasa (P09052), Belle (Q9VHP0). M. musculus: DDX4 (Q61496), DDX3X (Q62167), PL10 (P16381), DDX3Y (Q62095). H. sapiens: DDX4 (Q9NQI0), DDX3 (O00571), DBY (O15523). X. laevis: VLG1 (Q91372), An3 (P24346), PL10 (Q7ZXJ0).

laf-1 and vbh-1 RNAi

Wild-type L4 hermaphrodites were grown at 20°C on RNAi plates (NGM plates with 1mM IPTG and 25μg/mL carbenicillin) seeded with bacteria expressing double stranded RNA against nucleotides 1885–2127 of the laf-1 open reading frame, nucleotides 1557–1935 of the vbh-1 open reading frame, a mixture of both cultures, or the empty vector (L4440) as a negative control. Animals were moved to a new plate each day for three days, and their offspring subsequently scored. Embryos that had not hatched within 24 hours were scored as dead. Adult phenotypes were scored on the third day after being collected as eggs. Animals with embryos in their uterus were scored as fertile, and those with stacked oocytes were scored as female. Any animal that was adult size but that had neither embryos nor stacked oocytes was scored as sterile. Sterile animals were included in the “other defects” category, along with a few animals that died as adults, burst through the vulva, or grew slowly and failed to reach adult size.

Northern blots

All worms were grown at 20°C unless otherwise noted, using standard methods. For laf-1(RNAi) and vbh-1(RNAi) blots, wild-type embryos were isolated by hypochlorite treatment (6.25mL sodium hypochlorite, 3mL 4M KOH, 15.75mL H₂O) and grown to adulthood on RNAi plates (NGM plates with 1mM IPTG and 25μg/mL carbenicillin) seeded with the bacterial strains describe above in “laf-1 and vbh-1 RNAi”. For the male/hermaphrodite blot, animals of each type were hand-picked from a mating population. For the glp blot, worms were grown in liquid culture at 15°C, and then embryos were isolated by hypochlorite treatment and grown to young adult stage at 25°C. For the developmental blot, worms were synchronized by hypochlorite treatment and grown in liquid culture to the indicated stages. Each developmental stage was verified at the time of collection by examining vulva and germline development of a sample of worms using DIC microscopy.

RNA was extracted using Trizol reagent (Invitrogen), and Northern blots were performed by standard methods. 5μg total RNA/lane was separated on a 1.5% agarose gel with formaldehyde and blotted to ZetaProbe membrane. RNA probes were synthesized from digested plasmid templates using the Maxiscript kit (Ambion) and α-P³²-UTP. Laf-1 mRNA was detected with a probe antisense to nucleotides 1–575 of the open reading frame for the male/hermaphrodite blot and with a probe antisense to the entire laf-1 open reading frame for all others. The rpl-6, vit-2, gld-1, and act-1 probes were antisense to nucleotides 20–508, 86–400, 975–1378, and 1–503 of their open reading frames, respectively. The probes were cleaned over a G-50 microspin column (GE Healthcare) to remove unincorporated free nucleotides and hybridized to the blots overnight at 68°C in hybridization buffer (recipe in Ambion Maxiscript kit instructions). The blots were washed at 65°C as specified in the Northern blot protocol in the Maxiscript instruction booklet, exposed overnight with a storage phospor screen, scanned with a Storm scanner, and quantified with ImageQuant software. The abundance of laf-1 mRNA was normalized to that of the rpl-6 or act-1 loading control. The mean laf-1 abundance from 2–4 independent blots using separate worm samples is reported.

LAF-1 transgenes

We constructed gfp::laf-1 and laf-1::gfp transgenes expressed in embryos and the adult germ line by cloning the full length LAF-1 open reading frame into vectors pJK3 and pJK7 (modified versions of pFJ1.1 (Batchelder, et al 2007)), using the Spe I and Mlu I restriction sites. The vector contains the pie-1 promoter, pie-1 5′ and 3′ UTRs, and unc-119(+) as a marker for transformation. We introduced each transgene into unc-119(ed3) by biolistic transformation (Pratis, et al 2001) and isolated worms with wild-type movement. Transformants homozygous for one or more integrated copies of the transgene were identified, and the transgenic lines were maintained at 25°C, which is suggested to aid germline expression of GFP transgenes (Evans 2006). Expression of the intended GFP::LAF-1 or LAF-1::GFP fusion protein was confirmed by Western blots using anti-GFP antibodies.

Antibody staining

Embryos were dissected onto poly-L-lysine treated slides and prepared by freeze cracking and 5 minute fixation in cold methanol. The slides were blocked for 30 minutes with PBSBT (1xPBS + 0.05% BSA + 0.25% Tween) and then incubated overnight at 4°C with primary antibody (rabbit anti-GFP (Invitrogen #A11122) diluted 1:200 and mouse anti-PGL-1 (O1C1D4, courtesy of Susan Strome) diluted 1:1 in PBSBT). The slides were washed 3x with PBST (1xPBS + 0.25% Tween) and incubated for one hour at room temperature with secondary antibody (DyLight488 donkey anti-rabbit and Cy3 donkey anti-mouse (Jackson ImmunoResearch) diluted 1:500 in PBST). The slides were then washed 4x with PBST, with DAPI diluted 1:1000 in the third wash, and mounted with Vectashield (Vector labs).

TRA-2 Western blots

The TRA-2 antibody was raised against a GST-tagged C-terminal fragment of TRA-2A (amino acids 1134–1475 of TRA-2A) fusion protein in rabbits and affinity purified against His-tagged TRA-2B (Yoo 2003). For all strains except wild type, 300–350 adult XX females were picked into 25μL M9 and frozen at −80°C. For wild type, 850 young adult N2 Bristol hermaphrodites were picked and frozen in the same way. 25μL of sample buffer was added to each sample, and the worms were lysed by three cycles of freezing in liquid nitrogen and boiling for 5 minutes. SDS-PAGE and Western blotting were performed by standard methods using 10% gels and blotting to Immobilon-P PVDF membrane (Millipore). Blots were incubated with TRA-2 antibody at 1:2500 overnight at 4°C and with secondary antibody (HRP-linked donkey anti-rabbit Ig, Amersham) at 1:5000 for 2 hours at room temperature. The blots were incubated with SuperSignal West Pico Chemiluminescent Substrate (Pierce) for 5 minutes and imaged using X-ray film. Membranes were blocked for 1 hour at room temperature before reprobing with tubulin antibody following the same procedure (1°- mouse monoclonal anti α-tubulin, Sigma, 1:5000, 2° - HRP labeled anti mouse, BD Biosciences 1:5000). Estimates of the amount of TRA-2 and tubulin in each lane were obtained by scanning the films with a document scanner and quantitating the bands using ImageQuant software. Values for multiple exposures of the same blot were averaged, and the TRA-2 values were divided by the tubulin values to correct for differences in loading.

RESULTS

laf-1 encodes a putative DEAD-box helicase

Previous work mapped laf-1 to the left arm of chromosome III between dpy-1 and daf-2, and three-factor mapping placed it ~1.5–1.8 map units left of daf-2 (Goodwin, et al 1997). Using single nucleotide polymorphisms as markers, we further narrowed the location of laf-1 to an approximately 11kb region spanning nucleotides 2,766,977 – 2,777,892 of chromosome III (Fig. 2A, see Materials and Methods for mapping data). Two predicted open reading frames (Y71H2AM.18 and Y71H2AM.19) overlap this region. We sequenced three independent laf-1 mutations (q80, q217, and q267), and all are single base changes in Y71H2AM.19 (see below).

We characterized the 3.7kb laf-1 transcript using Northern blots and a series of overlapping RT-PCR and RACE experiments. Laf-1 mRNA consists of six exons, is trans-spliced to SL1, and has a 1.5kb 3′ untranslated region (Fig. 2B, sequence provided in Supplemental Fig. 1, GenBank accession FJ348231). The predicted LAF-1 protein contains a DEAD-box RNA helicase domain with the nine motifs characteristic of and highly conserved in all DEAD-box proteins (Fig. 2B). The three sequenced laf-1 mutations are all located in or near conserved motif III: laf-1(q267) changes Thr434 to Ile within motif III, laf-1(q80) changes nearby Arg426 to Cys, and laf-1(q217) changes Met430 to Ile (Fig. 2C and Supplemental Fig. 1). The identity of all three laf-1 mutations as missense alleles and the fact that they cluster within a small portion of the protein suggests that they are not null alleles. Results presented below, however, indicate that they are reduction- or loss-of-function alleles.

LAF-1-related helicases function in the germ line

We identified proteins related to LAF-1 by BLAST analysis and assembled a phylogenetic tree of LAF-1-related DEAD-box helicases. The closest relative of LAF-1 is C. elegans VBH-1 (Fig. 2D). LAF-1 and VBH-1 are 47% identical across their full length. Expression of VBH-1 is germline specific, and vbh-1(RNAi) causes a premature switch from spermatogenesis to oogenesis, resulting in reduced numbers or absence of sperm (Salinas, et al 2007). VBH-1 may also play a role in oocyte production or function, because fertility of vbh-1(RNAi) females is not fully restored by mating to wild-type males (Salinas, et al 2007). LAF-1 and VBH-1 are most closely related to the group of helicases that includes S. cerevisiae Ded1p, human DDX3, and Drosophila Belle. LAF-1 is more distantly related to Drosophila Vasa and its close homologs.

Many LAF-1-related helicases function in the animal germ line. Drosophila Belle is required for both spermatogenesis and oogenesis (Johnstone, et al 2005). Mutations in human DBY are a frequent cause of male infertility (Foresta, et al 2000). Mouse PL10 is testis-specific and may function in translational regulation during spermatogenesis (Leroy, et al 1989). Drosophila Vasa and it homologs in virtually all animals, including its closest C. elegans homologs, the GLH proteins, are found in granules that segregate with primordial germ cells during early development (Hay, et al 1988; Gruidl, et al 1996, reviewed in Raz 2000). These germ granules, called polar granules in Drosophila and P granules in C. elegans, contain maternal RNA and proteins required for specification of the germ line (reviewed in Wylie 1999). Drosophila Vasa is required for establishment of the germ line and for oogenesis (Schupbach, Wieschaus 1986; Lasko, Ashburner 1988). Deletions of C. elegans glh-1 cause temperature-sensitive sterility with a strong maternal effect (Spike, et al 2008), and both GLH-2 and GLH-4 function in germ cell proliferation and gamete formation (Gruidl, et al 1996; Kuznicki, et al 2000). LAF-1 joins this list of DEAD-box RNA helicases required for germline development.

Proteins related to LAF-1 affect various post-transcriptional aspects of gene expression, most notably translation (Chuang, et al 1997; Carrera, et al 2000; Berthelot, et al 2004; Johnstone, Lasko 2004; Shih, et al 2008) and export of RNA from the nucleus (Askjaer, et al 2000; Yedavalli, et al 2004). We discuss these LAF-1-related proteins in more detail below and suggest models for LAF-1 function based on the known roles of its homologs (see Discussion).

laf-1 and vbh-1 function in germline sex determination

We compared the phenotypes of laf-1(RNAi) and vbh-1(RNAi) individually and in combination to investigate the degree to which their functions are related. We also tested whether the phenotypes of laf-1 mutants are similar to those of laf-1(RNAi), which we assume to be reduction- or loss-of-function. We fed wild-type L4 hermaphrodites bacteria expressing either laf-1 dsRNA, vbh-1 dsRNA, a mixture of the two, or no dsRNA and scored the phenotypes of their progeny. To lessen potential off-target RNAi effects, we chose regions of laf-1 and vbh-1 outside of the conserved helicase domains. Figure 3 demonstrates that our RNAi procedures specifically reduced abundance of laf-1 or vbh-1 mRNA. Offspring of both laf-1(RNAi) and vbh-1(RNAi) were feminized (Table 1). Feminization of laf-1(RNAi) was strongest among day 2 offspring (12% females), but the feminization of most day 3 offspring could not be ascertained because they died as embryos (88% Emb). The phenotypes of laf-1 mutants are similar to those of laf-1(RNAi), and we therefore conclude that the mutations are likely reduction- or loss-of-function alleles. Feminization of vbh-1(RNAi) was evident on all three days (10–16% females), with very little embryonic lethality. We observed a higher percentage of female offspring in laf-1(RNAi); vbh-1(RNAi) double RNAi than with either laf-1(RNAi) or vbh-1(RNAi) alone. The similarity of laf-1(RNAi) and vbh-1(RNAi) phenotypes, together with the enhanced phenotypes when both are silenced, indicate that both are involved in germline sex determination and likely act partially redundantly. Laf-1 mutants and laf-1(RNAi) have additional embryonic lethal phenotypes not evident in vbh-1(RNAi), suggesting that laf-1 has additional roles in embryonic development.

Figure 3. Specificity of laf-1(RNAi) and vbh-1(RNAi).

Northern blots of wild-type worms treated with either empty vector, laf-1(RNAi), or vbh-1(RNAi) were probed with either laf-1, vbh-1, or rpl-6 as a loading control. Numbers below each panel show the relative abundance of laf-1 and vbh-1 mRNAs +/− one standard deviation from two independent experiments, with empty vector being arbitrarily defined as 1.0.

laf-1 is widely expressed and developmentally regulated

We assayed laf-1 expression using Northern blots (Fig. 4) to further understand where and when it might act. Both males and hermaphrodites express laf-1 (Fig. 4A), which is consistent with both sexes being affected by laf-1 mutations. Laf-1 mRNA is expressed in both the germ line and soma of young adult hermaphrodites (Fig. 4B). We used temperature sensitive alleles of glp-1 and glp-4 to deplete adults of germ cells (Austin, Kimble 1987; Beanan, Strome 1992). The absence of gld-1 mRNA, which is expressed only in the germ line (Jones, et al 1996), demonstrates that this was successful (Fig. 4B). Young adults depleted of germ cells express reduced but substantial amounts of laf-1 mRNA. We interpret the reduced expression to indicate that laf-1 mRNA is expressed in the germ line, and we interpret the continued expression of laf-1 in animals without a germ line to indicate that laf-1 is also expressed in somatic cells. Expression of laf-1 in both the germ line and soma is consistent with the germline feminization of laf-1/+ heterozygotes and previous data showing that transgenes carrying the tra-2 3′UTR are misregulated in the soma of laf-1/+ heterozygotes (Goodwin, et al 1997). We conclude that laf-1 is expressed in and functions in both the germ line and soma.

Figure 4. laf-1 mRNA expression.

(A) Northern blot comparing laf-1 expression in wild-type adult males, L4 hermaphrodites, and young adult hermaphrodites. The same blot was probed for rpl-6 as a loading control and for vit-2, which is expressed only in hermaphrodites (Blumenthal, et al 1984). (B) Northern blot comparing laf-1 expression in wild-type (N2 Bristol), glp-1(q231ts), and glp-4(bn2ts) animals. All strains were grown at 25°C and harvested as young adults, before the wild-type animals produced embryos. The same blot was probed for rpl-6 as a loading control, and a separate blot of the same RNA was probed for gld-1, which is germline specific (Jones, et al 1996). Numbers below the blot show the mean laf-1 abundance +/− one standard deviation from three independent experiments, with wild type being arbitrarily defined as 1.0. (C) Northern blot showing laf-1 expression in wild-type embryos, L1-L4 larvae and adults. The same blot was probed for act-1 as a loading control. Numbers below the blot show the mean laf-1 abundance +/− one standard deviation from four independent experiments, with adults being arbitrarily defined as 1.0.

Expression of laf-1 is developmentally regulated. Laf-1 mRNA is present at all stages (embryo, L1-L4, and adult), with the highest levels occurring in embryos (Fig. 4C). The amount of laf-1 mRNA decreases during larval development and then increases as the worms reach adulthood. A high level of laf-1 expression in embryos is consistent with a role for laf-1 in embryonic development as suggested by the lethal phenotype of laf-1 homozygotes.

LAF-1 is predominantly cytoplasmic and concentrated in P granules of the germ line

To investigate where within the cell LAF-1 is located, we expressed N-terminal and C-terminal GFP fusion proteins (GFP::LAF-1 and LAF-1::GFP, respectively) in the embryo using a pie-1 promoter and pie-1 5′ and 3′ UTRs. Such constructs are known to be expressed in both the adult germ line and embryos (Strome, et al 2001). Both GFP::LAF-1 and LAF-1::GFP are predominantly, if not exclusively, cytoplasmic (Fig. 5, panels A and E), with a substantial portion of the signal concentrated in discrete cytoplasmic foci. The largest and most conspicuous foci are confined to the germline precursor cells (“P1” of the two-celled embryos in Fig. 5; P2, P3, etc. cells of later embryos). GFP::LAF-1 and LAF-1::GFP colocalize in these foci with PGL-1 (Fig. 5), identifying them as P granules (Kawasaki, et al 1998).

Figure 5. Sub-cellular localization of LAF-1.

Confocal images of two-celled embryos expressing either GFP::LAF-1 (left column) or LAF-1::GFP (right column). (A) An embryo expressing GFP::LAF-1. The arrow points to one of the two nuclei, and the triangle indicates a P granule in the P1 cell. (B) The same embryo as in panel A stained with mouse anti-PGL-1, a component of germline-specific P granules (Kawasaki, et al 1998). The triangle indicates the same P granule as in panel A. (C) DAPI stain showing DNA in the nucleus. (D) Merged image showing co-localization of GFP::LAF-1 and PGL-1 in P granules. The triangle indicates the same P granule as in panels A and B. (E–H) An embryo expressing LAF-1::GFP.

TRA-2 expression is elevated in laf-1/+ mutants

Previous data suggest that LAF-1 negatively regulates expression of TRA-2 (Goodwin, et al 1997). In principle, feminization of laf-1/+ heterozygotes could be due entirely to elevated amounts of TRA-2. To test this model, we examined TRA-2 protein levels in laf-1/+ heterozygotes. Two distinct TRA-2 proteins are predicted to be derived from two different mRNAs (Fig. 6A). TRA-2A is a ~170kDa transmembrane protein, provides the main feminizing activity of tra-2, and results from translation of the ubiquitously expressed 4.7kb tra-2 mRNA (Okkema, Kimble 1991; Kuwabara, et al 1992; Kuwabara, Kimble 1995). A smaller protein corresponding to the intracellular domain of TRA-2A is thought to arise from translation of the germline-specific 1.8kb tra-2 transcript (Okkema, Kimble 1991; Kuwabara, et al 1998) or from TRA-3-mediated cleavage of TRA-2A (Sokol, Kuwabara 2000). This protein, termed TRA-2B or TRA-2ic, (“ic” denotes “intracellular”) also promotes female development (Kuwabara, et al 1998; Lum, et al 2000). The 4.7kb and 1.8kb transcripts share the same TGE-containing 3′ UTR (Kuwabara, et al 1992; Kuwabara, et al 1998) and thus may be subject to the same translational regulation.

Figure 6. laf-1/+ heterozygotes express elevated quantities of TRA-2.

(A) Diagram of tra-2 transcripts and proteins. The 4.7kb and 1.8kb tra-2 mRNAs are shown, with their predicted protein products, TRA-2A and TRA-2B. The extracellular - EC, transmembrane - TMDs, and intracellular - IC portions of TRA-2A are marked. TRA-3 cleavage of TRA-2A produces an intracellular fragment, termed TRA-2ic (Sokol, Kuwabara 2000). The portion of the protein used to raise and purify the antibody is indicated by a black bar. TGEs (represented by black half-arrows) are present in the 3′ UTR of both the 4.7kb and 1.8kb transcripts (Kuwabara, et al 1992; Kuwabara, et al 1998). The portions of the tra-2 3′UTR deleted in tra-2(e2020gf) and tra-2(q122gf) are indicated (Goodwin, et al 1993). (B) Western blots probed with an antibody that detects TRA-2B/TRA-2ic. The same blots were probed with an antibody against α–tubulin as a loading control. The estimated abundance of TRA-2 is shown below the blots, with wild type (left) or fem-2 (right) being arbitrarily defined as 1.0.

We used Western blots and an affinity purified, polyclonal antibody directed against the C-terminal region of TRA-2 (Yoo 2003) to measure TRA-2 protein abundance in wild type, laf-1/+ heterozygotes, and control strains (Fig. 6B). The antibody is directed against a region shared by both TRA-2A and TRA-2B/ic, but we detect only TRA-2B/ic on Western blots. This ~50kDa protein is present in extracts from wild-type strain N2 (Fig. 6B, lane 3) and absent in tra-2(null);fem-3 mutants (Fig. 6B, lane 1). Thus, the antibody is specific for TRA-2B/ic. We used feminized tra-2(null);fem-3 animals for this comparison so that we could compare females to females rather than to tra-2(null) pseudomales.

The abundance of TRA-2 protein is increased several fold in tra-2(gf) mutants. Tra-2(q122) deletes one of the two TGEs in the tra-2 3′UTR, and tra-2(e2020) deletes both TGEs plus some additional sequence (Goodwin, et al 1993) (Fig. 6A). We observe a 4- to 5-fold increase in TRA-2 expression in tra-2(e2020gf) mutants (Fig. 6B, compare lane 2 to 3 and lane 9 to 13) and a 3-fold increase in tra-2(q122gf) mutants (compare lane 10 to lane 13). Thus, deletion of one or both TGEs leads to increased accumulation of TRA-2. The magnitude of the increase correlates with the extent to which the TGE sequences are deleted.

We compared the abundance of TRA-2 in laf-1(q217)/+ and laf-1(q267)/+ adult females to that of fem-1, fem-2 or fem-3 females. We analyzed adults, because laf-1/+ mutants are incompletely penetrant. Approximately 20% of laf-1/+ heterozygotes develop as females, and we manually picked these animals for analysis by Western blots. If elevated expression of TRA-2 is responsible for their development as females, then such animals should have the highest levels of TRA-2. FEM-1, FEM-2, and FEM-3 act downstream of TRA-2 in the regulatory cascade (Doniach, Hodgkin 1984; Hodgkin 1986) (Fig. 1A), and we used them to distinguish effects on TRA-2 expression caused by feminization per se, as opposed to effects caused by laf-1 mutations in particular. We observed a 1.5- to 1.9-fold increase in TRA-2 expression in laf-1/+ heterozygotes compared to the fem controls (Fig. 6B, compare lanes 4 and 5 to lanes 6–8 and lanes 11 and 12 to lane 13). Although the increase in TRA-2 expression in laf-1/+ females is modest, it is consistent and reproducible (see Supplemental Fig. 2). We conclude that laf-1 negatively regulates TRA-2 expression either directly or indirectly, but that the magnitude of the regulation is modest. We discuss below whether this increase is sufficient to cause feminization of laf-1/+ heterozygotes.

DISCUSSION

LAF-1 is a putative DEAD-box helicase

DEAD-box helicases influence many aspects of RNA metabolism, including transcription, splicing, mRNA export, translation, and mRNA degradation (reviewed in Rocak, Linder 2004). They bind ATP and RNA cooperatively (Lorsch, Herschlag 1998; Polach, Uhlenbeck 2002) and unwind short stretches of double stranded RNA (Rozen, et al 1990; Rogers, et al 1999) or remodel ribonucleoprotein complexes (Jankowsky, et al 2001; Fairman, et al 2004; Bowers, et al 2006; reviewed in Linder, et al 2001) in an ATP hydrolysis-dependent manner.

The proteins most closely related to LAF-1 affect translation or export of RNA from the nucleus. Ded1p and Dbp1p of S. cerevisiae function during translation initiation and are thought to unwind double-stranded portions of 5′ UTRs to facilitate 40S ribosome scanning (Chuang, et al 1997; Berthelot, et al 2004). Human DDX3 and mouse PL10 both rescue the lethality of ded1 deletions when expressed in S. cerevisiae, suggesting that they also function in translation (Chuang, et al 1997; Mamiya, Worman 1999). DDX3 additionally inhibits cap-dependent translation by binding to eIF4E and disrupting eIF4E-eIF4G interactions (Shih, et al 2008). Drosophila Vasa promotes translation of several maternal mRNAs including oskar, nanos and gurken (Markussen, et al 1995; Gavis, et al 1996; Styhler, et al 1998; Tomancak, et al 1998) through interactions with translation initiation factor eIF5B (Carrera, et al 2000; Johnstone, Lasko 2004). Other LAF-1-related helicases affect RNA export. Human DDX3 and Xenopus An3 both shuttle between the nucleus and cytoplasm through interactions with the CRM1 export receptor (Askjaer, et al 1999; Yedavalli, et al 2004). DDX3 mediates export of unspliced or partially spliced HIV-1 viral transcripts through the Rev/CRM1 pathway (Yedavalli, et al 2004). Nuclear export of An3 is linked to its helicase activity, suggesting that it too may function in RNA export. Being a member of the larger family of these RNA regulatory proteins, LAF-1 may function similarly and affect translation or nuclear export of its targets.

The nature of the laf-1 mutations

Silencing of laf-1 by RNAi yields offspring whose lethal or feminized phenotypes are the same as those of laf-1 mutations. Thus, our laf-1 mutations are likely loss- or reduction-of-function alleles. Two lines of evidence, however, indicate that laf-1 alleles have special properties. First, laf-1/+ heterozygotes are more strongly feminized than saDf1/+ heterozygotes, which deletes laf-1. Approximately 10–30% of laf-1/+ animals develop as females, while only 6% of saDf1/+ do so (Goodwin, et al 1997). Laf-1 alleles were identified because of their dominant feminizing effect, and the screen may have demanded alleles more strongly dominant than typical null alleles. Second, all three laf-1 mutations that we sequenced are amino acid substitutions in or near conserved motif III of the helicase domain. Mutations in this domain of other DEAD-box proteins eliminate helicase activity without affecting ATP binding/hydrolysis or RNA binding (Pause, Sonenberg 1992; Schwer, Meszaros 2000). Motif III is thought to participate in interdomain interactions necessary for conformational changes that couple ATP hydrolysis to RNA unwinding (Pause, Sonenberg 1992; Schwer, Meszaros 2000; Sengoku, et al 2006). Mutations that disrupt interdomain interactions in other DEAD-box proteins are known to have dominant-negative effects (Plumpton, et al 1994; Cheng, et al 2005; Sengoku, et al 2006).

LAF-1 regulates tra-2 expression

We observe a 1.5- to 1.9-fold increase in the abundance of TRA-2 in laf-1/+ females compared to fem-1(−), fem-2(−), or fem-3(−) control females. Based on this, we conclude that LAF-1 indeed negatively regulates tra-2 expression, either directly or indirectly. Is this modest but reproducible increase in TRA-2 expression sufficient to feminize laf-1/+ heterozygotes, or must laf-1 have other targets of regulation in the germ line in addition to tra-2? We observe a 3-to 5-fold increase in TRA-2 expression in tra-2(gf) alleles (see Fig. 6), 100% of which are feminized (Doniach 1986; Schedl, Kimble 1988). Tra-2(gf)/+ heterozygotes, which likely express less TRA-2 than tra-2(gf) homozygotes, are also 100% feminized (Doniach 1986; Schedl, Kimble 1988). Thus, it appears that relatively small increases in expression of TRA-2 are sufficient to feminize a high proportion of animals. Only about 20% of laf-1/+ animals are feminized, and it is therefore reasonable that the 1.5- to 1.9-fold increase in TRA-2 expression that we observe is sufficient to cause their feminization. We expect that laf-1 homozygotes would exhibit an even higher proportion of feminized animals and higher levels of TRA-2 expression compared to laf-1/+ heterozygotes, but we were unable to test these predictions due to lethality of laf-1 homozygotes.

Models for LAF-1 activity

The identity of LAF-1 as an RNA helicase suggests ways that it might function in regulation of tra-2. Four models are diagramed in Figure 7:

Figure 7. Models for LAF-1 regulation of tra-2.

Diagram of possible roles for LAF-1 in tra-2 regulation. The number beside each gray line refers to the corresponding model in the Discussion.

Model 1

LAF-1 represses global translation. LAF-1 may not specifically regulate tra-2, but instead have a global effect on translation initiation like that of human DDX3 (Shih, et al 2008). Widespread misregulation of translation would explain the lethality of laf-1 homozygotes, and feminization of laf-1/+ heterozygotes could result from sensitivity of the system to small changes in the amount of TRA-2 or other proteins required for sex determination.

Model 2

LAF-1 remodels a tra-2 mRNP complex. Some DEAD-box proteins, including Ded1p, affect mRNA:protein interactions (Jankowsky, et al 2001; Fairman, et al 2004; Bowers, et al 2006; reviewed in Linder, et al 2001). LAF-1 may remodel a protein complex on the tra-2 3′UTR to facilitate repression of tra-2 translation by other regulators, such as GLD-1 and FOG-2. It could unwind stretches of dsRNA to promote GLD-1 binding to the TGEs or selectively remove proteins that interfere with tra-2 repression.

Model 3

LAF-1 promotes translation of a tra-2 repressor. LAF-1 might indirectly affect TGE-mediated repression by activating translation of GLF-1, FOG-2, or other undiscovered negative regulators of tra-2. This mechanism would be similar to translational activation of specific Drosophila mRNAs by Vasa (Markussen, et al 1995; Gavis, et al 1996; Styhler, et al 1998; Tomancak, et al 1998).

Model 4

LAF-1 affects export of tra-2 mRNA from the nucleus. Export of tra-2 mRNA is regulated (Segal, et al 2001; Kuersten, et al 2004), and LAF-1 may affect this process. For example, LAF-1 might interact with C. elegans CRM-1 much as LAF-1 homologs An3 and DDX3 interact with CRM1 in Xenopus and humans (Askjaer, et al 1999; Yedavalli, et al 2004). Regulated export of tra-2 mRNA from the nucleus relies on the CRM-1-mediated pathway and may influence translational control of tra-2 mRNA (Kuersten, et al 2004). If LAF-1 mediates tra-2 mRNA export, disruption of this process in laf-1 mutants could lead to increased TRA-2.