RESULTS

All three shu alleles cause recessive female sterility with no effects on zygotic viability (Schüpbach and Wieschaus 1991). The two strong loss-of-function alleles, shu^WQ41 and shu^WM40, also result in male sterility while the weaker allele, PB70, does not affect male fertility. Comparison of the phenotypes of shu^WQ41 and shu^WM40 as homozygotes and hemizygotes (over Df(2R)b23) reveals that shu^WQ41 is likely to be a null allele while shu^WM40 is a very strong loss-of-function allele. Due to the presence of a closely linked recessive lethal mutation on the shu^WQ41 and shu^PB70 chromosomes, few homozygous adults could be isolated. The phenotype of shu^WQ41 was therefore studied in hemizygous flies [over Df(2R)b23], but will be referred to in the text as shu^WQ41, while that of shu^PB70 was analyzed in hemizygotes or in the heteroallelic combination with shu^WM40.

shu function is required for the normal function of the germline stem cells: To analyze the shu^WQ41 ovarian phenotype, ovaries were dissected from 0- to 1- and 2- to 4-day-old females and stained with the nuclear stain Hoechst and with antibodies recognizing several different markers of germ cell differentiation. The severity of the phenotypes observed in shu^WQ41 can vary between females, and even between ovarioles, suggesting that there may be some functional redundancy for shu. In contrast to wild-type ovaries (Figure 1B), 40–60% of the ovarioles from newly eclosed shu^WQ41 females do not contain any developing egg chambers (Figure 1, C and D). Usually, the other ovarioles contain only one to three egg chambers and their numbers fail to increase as the females age. Although some egg chambers with 15 nurse cells and an apparently normal oocyte nucleus are observed, subsequent egg chambers have fewer than 16 germ cells, e.g., 2, 4, 8, 10, or 12 (Figure 1, E and F). None of the egg chambers develop into normal eggs but instead degenerate by mid-to late oogenesis. Those egg chambers with the correct number of germ cells may have arisen through the differentiation of prestem cells, i.e., germ cells that developed as cystoblasts without first being established as stem cells (King 1970). The remainder of the shu^WQ41 ovarioles contain no developing egg chambers (Figure 1D), only germaria-containing clusters of germline cells that can sometimes appear pycnotic, indicating that they may be dying.

• Download figure
• Open in new tab
• Download powerpoint

Figure 2.

shu function is required for the normal differentiation of the stem cell progeny. Staining of wild-type and shu^WQ41 germaria, respectively, with anti-Sxl (A and B), anti-Hts/anti-Vasa (C and D), and anti-Bam (E and F). R1, region 1; s, spectrosomes; fu, fusomes.

The strong cytoplasmic staining of these germ cells with antibodies recognizing the Sex Lethal protein (Figure 2, A and B) indicates that shu function is dispensable for the establishment of the female mode of sexual differentiation of the germ cells. Staining of these germ cell clusters with antibodies that recognize spectrosomes and fusomes (Zaccai and Lipshitz 1996) reveals that they do not develop branched fusomes characteristic of older cysts (Figure 2, C and D). To determine the developmental stage of these shu germ cells, they were further analyzed using an antibody specific for the cyto-plasmic form of the Bam protein (McKearin and Ohlstein 1995). In wild-type germaria, the cytoplasmic form of Bam accumulates in mitotic cystoblasts and cystocytes (Figure 2E). However, in newly eclosed shu^WQ41 females no staining was observed in any of the germaria analyzed (>50; Figure 2F). This suggests that, after the first few egg chambers are formed, no further mitotic cystoblasts develop. Instead, these germ cells appear to redivide to form clusters of abnormal germ cells that degenerate or occasionally produce tumorous cysts. In older females (>4 days), ovarioles in which the cells at the tip do not express Vasa, apparently failing to maintain germline cells at the tip (Figure I1), are frequently observed consistent with defects in stem cell renewal or maintenance. In addition, we have also observed some mispackaged egg chambers (Figure 1G). Collectively, these phenotypes suggest that shu's function is essential for the normal activity of the stem cell. The absence of wild-type function could affect the first asymmetric division of the stem cell, resulting in an abnormal cystoblast that is compromised in its ability to undergo the normal four rounds of mitotic division. In addition, this abnormal asymmetric division would affect stem cell renewal, resulting in an abnormal stem cell that divides several times to produce ill-fated germ cell clusters.

Weak loss-of-function alleles of shu reveal later requirements in oogenesis: Females heteroallelic for the weak loss-of-function allele PB70 and the strong loss-of-function allele WM40 lay only ~30% as many eggs as wild-type females, none of which developed. Typically, these eggs are abnormally shaped and dorsal appendages, if present, are reduced or fused (see Figure 4B). Inspection of the ovaries of these females revealed at least five developing egg chambers in the majority of ovarioles. About 80% of these egg chambers contain the correct 15:1 ratio of nurse cells to oocyte (Figure 3B) but never develop into wild-type eggs. Although an oocyte is established, as seen by the localization of Bic-D and Orb proteins to the posterior-most cell in the cyst (Figure 3F), its identity is not maintained and the oocyte appears to fail in its further development. Subsequent to stage 2 or 3, the distribution of the two proteins becomes diffuse and accumulation of the proteins is observed in nurse cells (Figure 3F). This pattern is particularly striking for the Bic-D protein (Figure 3G) and suggests that the transport system from the nurse cells to the oocyte has collapsed in these egg chambers. Subsequently, oocyte development fails and the egg chambers degenerate. These observations corroborate the results of Tirronen et al. (1993). We have also observed at low frequency (5–10%) egg chambers in which the oocyte was not correctly positioned (Figure 3D) and egg chambers in which the oocyte nucleus appears polyploid rather than diploid, although not to the same extent as the nurse cells. This is similar to phenotypes observed in Bic-D partial loss-of-function mutations (J. Y. Oh and R. Steward, unpublished results). These observations suggest that there may be a later requirement for shu function during egg chamber development.

• Download figure
• Open in new tab
• Download powerpoint

Figure 3.

shu function may be required at later stages of egg chamber development. In ovaries from 0- to 2-day-old shu^PB70/shu^WM40 females, oocyte identity is established. A and B show Hoechst and phalloidin-stained egg chambers from wild-type and shu^PB70/shu^WM40 females, respectively. The oocyte is the cell with the diploid oocyte nucleus and four ring canals positioned at the posterior of the egg chambers (arrows). In C, the oocyte appears to have undergone endoreduplication and no longer looks diploid. The egg chamber shown in D has a mispositioned oocyte (arrow). In wild-type egg chambers the Bic-D and Orb proteins both localize to the presumptive oocyte and show overlapping patterns of localization. (E) Wild-type egg chambers stained with anti-Orb. (F) Corresponding stages of shu^PB70/shu^WM40 ovaries stained with anti-Orb. Both Bic-D and Orb proteins localize to the oocyte but their localization is not maintained and the focus of staining within the oocyte becomes diffuse (arrow). (G) shu^PB70/shu^WM40 egg chamber stained with anti-Bic-D showing diffuse staining in the oocyte and patches of accumulation among the nurse cells.

shu functions in the germline: The phenotype of shu indicates that its function is required for the normal activity of the germline stem cell. To address whether shu functions autonomously in the germline, we made use of the FRT-mediated, dominant female sterile (DFS) technique and generated mutant clones in the germline by mitotic recombination (Chou and Perrimon 1992). Females heterozygous for autosomal insertions of the DFS mutation ovo^D1 fail to lay eggs and their ovaries contain no egg chambers that have developed beyond stage 4–5 of oogenesis (Figure 4C). In contrast, females of the genotype y w P[FLP]/w; FRT ovo^D1/FRT shu^PB70, which were subjected to heat shock, lay abnormal eggs (Figure 4B). These eggs are similar to those laid by females hemizygous or heteroallelic for the PB70 allele. Inspection of ovaries from females in which germline clones of the strong loss-of-function alleles WQ41 and WM40 were induced revealed the presence of abnormal egg chambers that appear, by Hoechst staining, to be degenerating by mid-to late oogenesis (Figure 4D). These results indicate that shu functions in the germline. They do not, however, exclude the possibility of an additional somatic function.

• Download figure
• Open in new tab
• Download powerpoint

Figure 4.

shu function is required in the germline. Germline clones were generated by mitotic recombination in the ovo^D1 background. (A and B) Dorsal views of eggshells produced by germline clones induced in mothers of the genotype y w P[FLP]/w; FRT ovo^D1/FRT (A) and y w P[FLP]/w; FRT ovo^D1/FRT shu^PB70 (B). (C and D) Hoechst-stained ovarioles from mothers of the genotype y w P[FLP]/w; FRT ovo^D1/FRT shu^WM40 without (C) and with (D) heat-shock treatment. The arrow in D indicates an ovariole in which a clone has produced an egg chamber containing only two polyploid cells. In the germarium an individual germ cell that has undergone premature endoreduplication to produce a polyploid cell is present. Also shown to the right of this is an older degenerating egg fused to two younger egg chambers.

Cloning of the shu gene: Previous studies have localized shu to genetic map position 2-105 in the cytological interval 59D8–60A2 (Schüpbach and Wieschaus 1991). To refine its position, we performed complementation tests between the shu alleles and deficiencies in the region. Our results placed shu in the interval defined by the proximal breakpoint of Df(2R)b23 and the distal breakpoint of Df(2R)OV1 (Figure 5A). Simultaneously, we mapped shu genetically relative to the female-sterile mutant bgcn that is also uncovered by Df(2R)b23. We screened females for recombination events between the cn bw shu^WQ41 sp and cn bw bgcn^QW34 chromosomes that would result in females that were fertile in trans to Df(2R)b23. Of 6500 females screened, 2 fertile females were obtained. Both of these females were mutant for sp, placing shu 0.063 cM proximal to bgcn, i.e., within ~30 kb. We mapped the position of bgcn (probe provided by D. McKearin) and the distal breakpoint of Df(2R)OV1 (Figure 5B), which lies between shu and bgcn, using a cosmid contig obtained from the European Drosophila Genome Project (data not shown).

On the basis of the shu phenotype we would expect the gene to be expressed in early stages of oogenesis and during spermatogenesis. We therefore looked at the expression pattern of candidate ovarian cDNAs that were isolated using genomic fragments in the region proximal to the Df(2R)OV1 breakpoint. A 1.65-kb transcript in this region (Figure 5B) is expressed in the germarium of the ovary and in the apical tip of the testes (see below and Figure 6). This transcript is of identical size on Northern blots of ovary and testes mRNA. The 2.6-kb HindIII fragment of genomic DNA containing this transcript, plus 700 bp of upstream sequence, can rescue the sterility of shu males and females (see materials and methods). This fragment encodes no other complete transcript. In addition, we have identified mutations in the open reading frame of this transcript in genomic DNA from all three shu alleles (see below and Figure 7). These data confirm that this transcript encodes shu.

• Download figure
• Open in new tab
• Download powerpoint

Figure 5.

Genetic and molecular map of the region containing the shu locus. (A) Genetic mapping of shu. DNA deleted by the deficiency chromosomes is represented by black boxes, with the uncertainties in the breakpoints indicated. The region defining the shu interval is shown by a stippled box. (B) Restriction map of an ~16-kb region around the shu locus showing the position of the distal breakpoint of Df(2R)OVI. Over-lapping bacteriophage λ clones covering the region are shown underneath. Ovarian transcripts identified by Northern analysis are indicated. The 2.6-kb HindIII fragment, which encodes shu but no other transcripts, was used for transformation rescue of the shu phenotype. Both the 1.65-kb transcript and the adjacent 1-kb transcript are transcribed in the distal to proximal direction. B, H, X, and Xb represent BamHI, HindIII, XhoI, and XbaI sites, respectively. Polymorphisms identified in genomic DNA from Ore-R, and the cn bw, cn bw sp, and Df(2R)tid chromosomes are indicated (*).

shu is expressed in a dynamic pattern in the germline: We examined the expression pattern of shu during oogenesis, spermatogenesis, and embryogenesis using whole mount in situ hybridization with RNA probes made from the cDNA clone corresponding to its transcript. As shown in Figure 6, A and C, the mRNA can first be detected at the apical tip of the germarium in the germline stem cells and cystoblasts. The level of expression decreases in the remainder of region 1, where the cystocytes are dividing to produce 2-, 4-, 8-, and 16-cell cysts. However, in region 2b, where 16-cell cysts are present, strong levels of expression are again seen. In region 3, equivalent levels of expression can be seen in all 16 germ cells of the stage 1 egg chamber. This pattern of expression corresponds well with the earliest phenotypes of shu and confirms the results of our clonal analysis that indicated that shu function is required in the germline. Subsequent to stage 4 of oogenesis, there appears to be an abrupt downregulation of shu expression until stage 10, where the transcript can be detected at high levels in the nurse cells (Figure 6B). In early cleavage-stage embryos uniform levels of shu mRNA are detected (Figure 6E), suggesting a possible maternal function for shu. Interestingly, by the cellular blastoderm-stage shu mRNA is exclusively found in the pole cells (Figure 6F). The mRNA can be detected in the germ cells throughout their migration through the midgut and into the gonadal mesoderm (data not shown) and is present in the embryonic gonad of stage 15 embryos (Figure 6G). No staining above background level can be detected in any tissues other than the gonads. The shu mRNA can also be detected at the apical tip of the testes where the stem cells are dividing to produce cysts of primary spermatocytes (Figure 6D).

shu encodes a FK506BP-like protein with a TPR motif: Sequence analysis of the longest ovarian cDNA clones corresponding to the shu transcript revealed an open reading frame of 1236 bp. This is predicted to encode a 392-amino-acid protein (Figure 7A) with an expected molecular weight of 45 kD. The first conceptual translation initiation site, at position 58, has 2/4 matches (GATA) to the Drosophila consensus site (Cavener 1987). The 57 bp upstream of this ATG does not contain an in-frame stop codon. The longest cDNA is apparently full length because its size, 1682 bp, corresponds well with the size predicted by Northern analysis (data not shown), and an expressed sequence tag available from the Berkeley Drosophila genome project, which was isolated from a different cDNA library (0- to 3-hr embryo), has an identical 5′ end. Also, this cDNA can rescue the sterility of shu females and males when expressed under the control of the ovarian tumor (otu) promoter (data not shown; Robinson and Cooley 1997).

• Download figure
• Open in new tab
• Download powerpoint

Figure 6.

In situ hybridization to wild-type tissue using shu antisense riboprobes. (A) Wild-type ovariole showing strong germline expression of shu from germarial stages through stage 4. The level of expression is reduced subsequent to this and increases again in the nurse cells of the stage 10 egg chamber (B). In the germarium (C), low levels of shu expression can be detected in the germline stem cells and cystoblasts. The level of expression appears to be reduced in dividing cystocytes in the posterior of region 1. Similarly, in the testes (D), shu mRNA can be detected in the dividing stem cells at the tip of the testes. Uniform levels of the RNA are present in cleavage-stage embryos (E). During the cellular blastoderm stage, the RNA is incorporated into the pole cells and can be detected in the embryonic gonad at stage 15 (G).

A search for recognized protein motifs, using the Prosite database, revealed a peptidyl-prolyl cis-trans isomerase (PPIASE) domain at amino acids 103–193 and one tetratricopeptide repeat (TPR) at amino acids 303–334. PPIASE domains are protein-protein interaction motifs that have been shown, in vitro, to catalyze the cis-trans isomerization of the peptide bond of proline residues, resulting in changes in protein folding (Fischer and Schmid 1990). Their activity is inhibited by binding of the drugs FK506 and rapamycin. Hence, proteins containing these domains are commonly referred to as FK506-binding proteins (FK506BP). The TPR motif is a degenerate 34-amino-acid sequence, which is also thought to mediate protein-protein interactions (Lambet al. 1995). In addition, the Shu sequence contains a putative nuclear localization signal (NLS), and 18 predicted phosphorylation sites, although the functional significance of these, if any, remains unknown. A schematic representation of the protein is shown in Figure 7B.

Comparison of the predicted amino acid sequence of Shu with other proteins in the database, using the Gapped BLAST search program (Altschulet al. 1997), revealed that it is a novel Drosophila protein that shows similarity to proteins of the high-molecular-weight immunophilin family. This class of protein typically contains both PPIASE and TPR motifs. They are evolutionarily conserved, multifunctional proteins, examples of which have been identified in mammals and plants. Human FKBP6 (p36) (Menget al. 1998) shows 28% identity and 46% similarity (probability 10⁻²⁴ with 3% gaps) to Shu along the length of the entire protein sequence (Figure 7C). A similar level of homology is also observed with human FKBP4 (Sanchezet al. 1990), FKBP52 (the mammalian p59 heat-shock protein-binding immunophilin; Peattieet al. 1992), and several plant immunophilins. Functional studies of human FKBP12, the best-characterized FKBP, have identified 14 residues important for enzymatic activity and drug binding (Kay 1996). As shown in Figure 7C, Shu retains 8 of these residues and a further amino acid is also conserved, suggesting that this domain is functionally important. Similarly, comparison of the Shu TPR motif with the consensus TPR sequence reveals that 5 of 6 consensus residues are conserved (Figure 7C).

Figure 7.

(A) Sequence of shu cDNA and predicted protein. The DNA sequence of the longest cDNA clone is shown with the amino acid sequence indicated below the open reading frame. The ATG start codon and the TGA stop codon are shown in bold, and the polyadenylation signal is underlined. Introns were identified by comparison of the cDNA and genomic sequences and are indicated above the sequence (▿). The introns at positions 116 and 870 are 59 and 68 bp in size, respectively, and both have the eukaryotic consensus splice sites. Underlined amino acid sequences represent the PPIASE domain (103–193) and TPR motif (304–337). The putative NLS at amino acid 283 is boxed. The codons mutated in each of the three shu alleles are underlined and the altered nucleotide is shown in bold. (B) Schematic representation of the structure of the protein with the two domains and positions of the mutations indicated. The PPIASE and TPR domains are indicated by black and hatched boxes, respectively. (C) Alignment of the Shu sequence with that of hFKBP6 (accession no. AF038847). Identical residues are indicated as white on black, and conserved residues are shown as black on gray. Dashes represent gaps introduced to maximize the alignment. The positions of the PPIASE domains and TPR motifs in each protein are indicated by lines above (Shu) and below (FKBP6) the sequence. Functionally important residues of hFKBP12 and the consensus TPR residues are indicated above the sequence (the consensus TPR residues of hFKBP6 are indicated below its additional two repeats), with residues conserved in Shu shown in boldface type.

To investigate the nature of the shu mutations, we sequenced the genomic DNA from flies hemizygous for each allele (shown in Figure 7, A and B). The WQ41 allele has a C to T transversion at position 88 that creates a premature stop codon (Q11 to STOP). This confirms our prediction, based on genetic analysis, that WQ41 is a complete loss-of-function allele. The WM40 allele has a G to A transversion at position 1142 that creates a premature termination of translation 31 amino acids before the end of the protein (W342 to STOP). The weakest loss-of-function allele of shu, PB70, is due to a G to A transversion at position 1051 that results in an alanine (332) to threonine substitution within the TPR motif. This result suggests that the TPR motif is likely to be important for shu function. Interestingly, this mutation does not affect male fertility.

Shu protein is expressed in stem cells and 16-cell cysts: Rat antisera were raised against a 6× His-tagged protein containing amino acids 31–392 of Shu. Western analysis revealed that the antisera specifically recognize a 52-kD band in wild-type ovary extracts (Figure 8A, lane 1). This protein is not detected in ovary extracts from hemizygous shu^WQ41 females (Figure 8A, lane 2), confirming the specificity of the antisera. To determine the subcellular localization of the Shu protein, affinity-purified antibodies were used for immunofluorescent staining of ovaries. The Shu protein is expressed in the germline stem cells and cystoblasts (Figure 8B). Low levels of staining were observed in dividing cystocytes in the posterior of region 1. In region 2b, staining was evident in all cells of the newly formed 16-cell cysts. Like the shu mRNA, this persisted until stage 4–5 egg chambers. The staining was present in the cytoplasm of the germ cells. Only background levels were detected in the follicle cells. Germline staining was eliminated in shu^WQ41 ovaries. Cytoplasmic staining was also seen later in the nurse cells of stage 10 and older egg chambers (data not shown). However, as the strong loss-of-function alleles of shu do not develop normal eggs at this stage it is difficult to ascertain if this later staining is specific. We have been unable to detect staining in the germ cells of embryos with this antiserum. This could be due to technical reasons, such as the antibody sensitivity or specificity, as we have only been able to detect very low levels of the protein on immunoblots of embryonic extracts.

Figure 8.

(A) Western analysis of wild-type (lane 1) and shu^WQ41 (lane 2) ovarian extracts using a rat antiserum raised against the Shu protein. A band of ~52 kD is recognized by the antiserum that is absent from the shu^WQ41 extract. Below, the same blot is shown after detection of α-tubulin to control for loading. (B) Immunofluorescence staining of wild-type germarium using anti-Shu antiserum. The protein is expressed in the cytoplasm of the germline stem cells and cystoblasts. The level of expression is greatly reduced in dividing cystocytes but increases again to high levels in the stage 1 egg chamber.