InteractiveFly: GeneBrief

Ornithine decarboxylase antizyme : Biological Overview | Regulation | Developmental Biology | Effects of Mutation | Evolutionary Homologs | References

BIOLOGICAL OVERVIEW

The sex determination master switch, Sex-lethal, regulates the mitosis of early germ cells in Drosophila. Sex-lethal is an RNA binding protein that regulates splicing and translation of specific targets in the soma, but the germline targets are unknown. In an immunoprecipitation experiment aimed at identifying targets of Sex-lethal in early germ cells, the RNA encoded by gutfeeling, the Drosophila homolog of ornithine decarboxylase (ODC) antizyme (Salzberg, 1996), was isolated (Vied, 2003).

Mammalian Antizyme negatively regulates ODC catalytically as well by directing the inactivated enzyme to the proteasome for degradation. This negative regulation of ODC is part of a feedback loop that controls the levels of polyamines within the cell. Translation of Antizyme is dependent on ribosomal frameshifting, which is promoted by high levels of polyamines. As polyamine levels in the cell rise, more Antizyme is synthesized, leading to the turnover of ODC. Polyamines have been implicated in many processes, including cell growth, transcription, and differentiation (Vied, 2003 and references therein). In mammals Antizyme and ubiquitin are thought to be respectively two types of proteasome targeting devices that mark proteins for both ubiquitin-independent and ubiquitin-dependent degradation by the 26 S proteasome (see Gruendler, 2001).

Drosophila gutfeeling interacts genetically with Sex-lethal. It is not only a target of Sex-lethal, but also appears to regulate the nuclear entry and overall levels of Sex-lethal in early germ cells. This regulation of Sex-lethal by gutfeeling appears to occur downstream of the Hedgehog signal. Gutfeeling appears to regulate the nuclear entry of Cyclin B as well. Hedgehog, Gutfeeling, and Sex-lethal function to regulate Cyclin B, providing a link between Sex-lethal and mitosis (Vied, 2003).

Sxl is necessary for female germ cell development regulating mitosis and the female-specific function of recombination. The targets and mechanism by which Sxl controls these processes are not known. Mitotic germ cells reside in the germarium, which is the anteriormost structure of the ovariole. Each of the 14-16 ovarioles of the ovary contains 2-3 stem cells, which divide to produce a cystoblast and another stem cell. Cystoblasts undergo 4 synchronous mitotic divisions with incomplete cytokinesis to form an interconnected 16-cell cyst. One of these 16 cystocytes becomes the oocyte, and the remaining 15 become nurse cells. This occurs as the 16-cell cyst progresses down the germarium. Toward the posterior of the germarium cells of somatic origin, the follicle cells, surround each cyst. This encapsulation is followed by the cyst pinching off from the germarium to produce an egg chamber. The egg chamber moves down the ovariole as it matures into an egg (Vied, 2003 and references therein).

Given the requirement of Sxl in the germline, attempts were made to identify its RNA targets by immunoprecipitation of the protein. To enrich for early germ cells, where Sxl impacts mitosis, advantage was taken of specific female sterile alleles (the Sxl^fs alleles) which arrest ovarian development at this stage. In these cells, Sxl protein is expressed at high levels and localized primarily in the cytoplasm as seen in the stem cells and cystoblasts. The mutant Sxl protein fails to undergo the dramatic downregulation that normally occurs as the germ cells mature. One of the RNA molecules identified in the immunoprecipitations is encoded by the gutfeeling (guf) gene, the Drosophila ornithine decarboxylase (ODC) antizyme (Vied, 2003).

The connection of Guf with Sxl provides a potential link between Sxl and mitosis. It has been demonstrated in vertebrate cells that both ornithine decarboxylase and polyamine levels undergo changes that coincide with cell cycle transitions and that depletion of polyamines results in a cell cycle arrest. The polyamine synthetic pathway has also been shown to regulate Cyclin B mRNA levels in vertebrate cells. When ODC is inhibited, cyclin B mRNA levels increase. More directly, Guf affects the nuclear localization of Cyclin B, as it does of Sxl. In addition, the data show that the nuclear entry of Cyclin B depends on Sxl. Typically, Cyclin B regulates cell cycle transitions, but this does not appear to be the case in germ cells. Unlike Cyclin A, whose levels change with the cell cycle, Cyclin B levels remain constant in early germ cells. Additionally, Cyclin A associates with the fusome during the G2 phase of the cell cycle, while Cyclin B does not. This suggests that Cyclin A and Cyclin B regulate early germ cell mitosis by different mechanisms. Overexpression of Cyclin B (or Cyclin A) results in only one extra round of cystocyte divisions to produce a 32-cell cyst, and it has been proposed that Cyclin B (and Cyclin A) regulate specialized cystocyte divisions. Cyclin B is downregulated at approximately the 2-cell cyst stage (the same as Sxl) and is not detected again until the 16-cell stage. This timing may reflect a specific requirement for Cyclin B in one of the 4 mitoses that generate the 16-cell cyst. The dependence of Cyclin B nuclear entry on Sxl suggests how Sxl might regulate germ cell mitosis (Vied, 2003).

The genetic data suggest that the effect of Guf on Sxl and Cyclin B is downstream of the Hh signal. In early germ cells, Sxl is in a complex with the cytoplasmic components of the Hh signaling pathway. While the data do not provide evidence to the point of action, it is suggestive that Guf promotes the disassembly of this complex with Sxl. Given that Antizyme (Guf) binds directly to the vertebrate ODC and Smad1 proteins and targets them to the proteasome (reviewed by Coffino, 2001a and b and Hoyt, 2003), it is tempting to speculate that the disassembly of the Hh cytoplasmic components involves targeting of one or more of the components to the proteasome. Additionally, it has been demonstrated that the inhibitory cytoplasmic NF-kappaB complex can be dissociated by changes in polyamine levels. Inhibiting ODC and thus depleting polyamine levels, a condition equivalent to Antizyme overexpression, leads to the proteolysis of Ikappa-Balpha. NF-kappaB dissociates from the cytoplasmic complex and enters the nucleus where it affects transcription. These effects are analogous to the role proposed for Guf in dissociating the components of the Hh cytoplasmic complex (Vied, 2003).

In its effects on Sxl, guf appears to act downstream of Hh and upstream of the cytoplasmic components. If guf functions downstream of Hh, it does not appear to do so in all Hh-regulated ovarian processes. This is exemplified by the lack of an effect on the turnover of the fusome. Also, guf does not affect the known role of Hh in regulating follicle cell proliferation. Finally, a connection between Cyclin B and the Hh signaling pathway has been previously described, but in a different context. Patched 1 (Ptc 1), which is one of two Ptc proteins in mammals, has been shown to directly interact with Cyclin B, inhibiting its nuclear translocation until cells were exposed to the Hh signal. Whether Ptc directly interacts with Cyclin B in Drosophila remains to be determined. These results, nevertheless, provide an intriguing parallel linking the Hh signal and Guf to the nuclear translocation of Sxl and Cyclin B (Vied, 2003).

GENE STRUCTURE

The Drosophila guf locus appears to have four promoters because four different cDNAs are generated. Each cDNA has a different 5' exon with the translation start site (except for the second 5' exon, which does not contain a start codon) which is spliced to two common exons, the first of which contains the translation frameshifting site (Ivanov, 1998). Interestingly, the polypyrimidine tract at the 3' splice site of the first common exon has a run of eight Uridines (U). This is a feature found in all established Sxl targets: msl-2, tra, and Sxl. Additionally, guf has several long polyU tracts in the introns of all its transcripts. For splicing regulation in vivo, Sxl shows a requirement of at least seven Us at its binding site, while runs of six Us gives compromised regulation. In vitro, shortening of target polyU tracts below six severely decreases the affinity of Sxl protein for its RNA (Vied, 2003).

In the 3' UTR short, six nucleotide-long, polyU runs exist in the mature guf message. This raised the question of whether Sxl regulates guf primarily through splicing or translation control. To address this, Sxl was immunoprecipitated from extracts of both wild-type and Sxl^f4 ovaries, and the RNA in the immunoprecipitates was examined by RT-PCR with guf-specific primers. The primers span the two introns of the smallest guf transcript, so the size of the PCR products allows for distinguishing between spliced and unspliced forms of guf RNA (Vied, 2003).

A Drosophila antizyme with high homology to the sequence for mammalian antizyme (ornithine decarboxylase antizyme) has been reported. Homology of this coding sequence to its mammalian antizyme counterpart also extends to a 5' open reading frame (ORF) which encodes the amino-terminal part of antizyme and overlaps the +1 frame (ORF2) that encodes the carboxy-terminal three-quarters of the protein. Ribosomes shift frame from the 5' ORF to ORF2 with an efficiency regulated by polyamines. At least in mammals, this is part of an autoregulatory circuit. The shift site and 23 of 25 of the flanking nucleotides (which are likely important for efficient frameshifting) are identical to their mammalian homologs. In the reverse orientation, within one of the introns of the Drosophila antizyme gene, the gene for snRNP Sm D3 is located. Previously, it was shown that two closely linked P-element transposon insertions caused the gutfeeling phenotype of embryonic lethality and aberrant neuronal and muscle cell differentiation. The present work shows that defects in either snRNP Sm D3 or antizyme, or both, are likely causes of the phenotype (Ivanov, 1998).

Programmed frameshifting is known to occur in single-celled eukaryotes (yeast and protozoa) and in Xenopus but has not previously been discovered in any intermediate organism. The finding of antizyme programmed frameshifting in Drosophila provides the opportunity to study the evolution of this recoding event, which involves a transitory alteration of the rules of readout of the genetic code. The pseudoknot, a 3' mRNA message feature that is an important stimulator for the recoding signal in mammalian systems, is not recognizable in Drosophila, and the identification of its presumed alternative is of major interest. Of the three known RNA elements that stimulate antizyme frameshifting in mammalian cells, the other two are present in guf. One of these is the UGA stop codon of ORF1. It is interesting that even though in vitro translation experiments have shown that the other two stop codons (UAG and UAA) are almost as effective in stimulating frameshifting in decoding mammalian antizyme sequences so far all eukaryotic antizyme genes have UGA as the stop codon of ORF1. Another stimulatory element, a sequence immediately 5' of the UGA stop codon, is also likely to be present in the D. melanogaster antizyme sequence. Of 18 nucleotides 5' of the UGA stop codon, 16 are identical for guf and the rat antizyme sequence (with one of the two mismatches being an antizyme polymorphism). This level of 5' sequence homology is striking and provides a clear indication that this region plays an important role in antizyme frameshifting. The apparent absence of an RNA pseudoknot 3' of the UGA stop codon of guf1 is puzzling. The nucleotides within the stems of this pseudoknot are absolutely conserved among all known vertebrate antizyme sequences. Despite the apparent absence of a 3' pseudoknot, there is nucleotide sequence homology between this region of guf and vertebrate antizyme sequences. Perhaps some other RNA structure is present in this region of guf. Indirect evidence supports such a hypothesis. Comparison of D. melanogaster and D. virilis antizyme nucleotide sequences reveals that the 53 nt between the UGA of the putative frameshift site and intron 2 (a region most likely to contain a stimulatory 3' RNA structure) are completely conserved between the two species, even though the conservation of the rest of the known exonic sequences is only 75%. Computer programs predict several stem loops in this region, but the relevance of any of them to translational frameshifting is unknown. Even though additional frameshift-stimulatory elements in the Drosophila antizyme sequence are not obvious, there is a strong indication that there is more to the guf frameshift site than the 28-nt region homologous to the mammalian antizyme frameshift site. The 28 nt alone cannot stimulate more than 3% frameshifting in vitro. The data for the in vitro translation of guf strongly suggests that the frameshift efficiency in that system is much higher than 3%, thus implying the existence of additional frameshift-stimulatory signals in Drosophila antizyme mRNA (Ivanov, 1998).

REGULATION

Splicing of guf mRNA

Both forms of guf RNA are readily detected although the yield of mature RNA is greater than the precursor transcript in both wild-type and Sxl^f4 ovaries. The overall yield from wild-type ovaries was lower for both products, but since the immunoprecipitations and RT-PCRs are not quantitative, this comparison is only speculative. To determine whether Sxl also binds guf RNA in somatic tissues, the immunoprecipitation and RT-PCR performed using embryonic extracts. Sxl also binds to spliced and unspliced guf RNA in embryos (ratio of spliced to unspliced is approximately equal). These data suggest that Sxl may regulate both the splicing and translation of guf (Vied, 2003).

Since Sxl binds to guf RNA in both the soma and germline, RT-PCR amplifications were performed for the four guf transcripts to determine whether any sex-specific products could be detected. guf RNA was analyzed from males and females (whole animals and carcasses) as well as testes and ovaries. For all cDNA types, no significant difference between the sexes was observed. From these data, it is concluded that Sxl does not regulate the alternative splicing of guf RNA in a global manner. Given the general requirement for guf, splicing regulation of guf RNA by Sxl may be restricted to only a subset of cells, such as the early germ cells, making detection of altered transcripts unlikely. Since there are no apparent alternative exons within guf, it is also possible that Sxl causes a retention of the introns leading to the degradation of the RNA. Alternatively, the regulation by Sxl may occur primarily at the level of translation control (Vied, 2003).

While the RNA binding data suggest that guf is a target of Sxl, a genetic interaction between the two genes would strengthen the idea that the two genes have a related function. Therefore the dose of guf was reduced in homozygous Sxl^f4 females, which normally have small ovaries of tumorous egg chambers. This was done by using a P-element insertion allele (guf^118-3) or a small deletion allele produced by the imprecise excision of the guf^118-3 P-element (guf^lex47; Salzberg, 1996). Reducing guf dose in a homozygous Sxl^f4 background rescues the Sxl^f4 phenotype. The females lay eggs that hatch and produce adults (Vied, 2003).

Since the original description of guf, an additional gene, SmD3, was found in the intron of guf (Ivanov, 1998). SmD3 encodes a snRNP (small nuclear ribonucleoprotein) common to splicing snRNPs and is transcribed in the opposite direction to guf. More relevant to the existing guf alleles is that the P-element insertions that uncover guf are likely to be affecting both genes. The P-elements are inserted in the 5' UTR of SmD3, and guf deletion alleles derived from their imprecise excision affect the coding sequences of both genes. Additionally, it was recently suggested that the reported phenotype of guf (Salzberg, 1996), as caused by transposon insertions, is primarily due to the alteration of SmD3 (Schenkel, 2002). Since Sxl is a splicing regulator, it was unclear whether the effects on Sxl^fs females were a result of reducing guf or SmD3 or both. Additional experments indicate that the rescue is a result of an interaction of Sxl with guf and not with SmD3. Data also indicate that, in the germline, the guf^118-3 allele does affect guf expression. Since reducing the dose of guf allows differentiation of Sxl mutant germ cells, the data suggest that Sxl normally functions as a negative regulator of guf. Presumably, the mutant Sxl proteins are unable to properly regulate the splicing and/or translation of guf RNA (Vied, 2003).

DEVELOPMENTAL BIOLOGY

Embryonic

Northern analysis has revealed that the guf gene is expressed in all stages of embryonic and larval development as well as in pupae and adults. An abundant 2.1-kb transcript is detected in all developmental stages. An additional transcript (-1.6 kb) that labels faintly with guf probes may correspond to a different splicing variant of guf. However, this band could not be detected in other RNA samples prepared from similar developmental stages, suggesting that it may be a degradation product of the 2.1-kb transcript (Salzberg, 1996).

The expression pattern of the guf gene was determined with guf antisense RNA probes using in situ hybridization to whole mount embryos. Low levels of uniformly distributed guf RNA are detected in precellularized embryos, suggesting a maternal contribution of guf transcripts. During gastrulation and germband extension guf transcripts are mostly abundant in the mesoderm and the invaginating posterior midgut. At stage 11, patches of mesodermal cells that, based on their position, probably correspond to fat body precursors begin to express higher levels of guf RNA. A strong signal is also detected at this stage in the Malpighian tubule rudiments and the foregut. As germband retraction takes place, guf expression is maintained in the developing fat body, Malpighian tubules and the posterior and anterior midgut. In addition, at late stage 12 and stages 13-14 guf is expressed at low levels throughout most of the embryo including the subectodermal layer that contains most of the PNS cells. At stage 14, guf transcripts are first detected in the developing body wall muscles. Expression of guf in body wall muscles, fat body and the midgut is maintained throughout the rest of embryonic development (Salzberg, 1996).

The P-element enhancer detectors inserted in the guf gene confer ß-gal expression in a pattern similar to the distribution of guf mRNA. Very high levels of ß-gal activity were detected in the midgut (Kania, 1995). (Due to the strong expression in the gut and the defects in the sensory nervous system caused by these insertions the gene was named gutfeeling). Very weak lacZ expression was detected in the developing somatic muscles starting at stage 14 of embryonic development. Embryos heterozygous for these insertions display weak ß-gal expression in developing muscles, whereas expression in other tissues mentioned above is undetectable (Salzberg, 1996).

Since guf is expressed in Malpighian tubules and fat bodies, the Malpighian tubules and fat bodies of gut minus embryos were examined to determine whether they exhibit morphological defects. Malpighian tubules were visualized by anti-Cut staining and found to be morphologically normal in mutant embryos. Fat bodies were visualized with antibodies against Neurotactin and no obvious morphological defects were observed (Salzberg, 1996).

To assess the extent of maternal contribution of guf mRNA and to determine whether zygotic guf expression is eliminated in a deletion allele, mutant embryos were examined by in situ hybridization with guf antisense RNA probe. This probe did not contain sequences from the 5' terminus that were not removed by this deletion (the probe corresponds to nucleotides 1310-2164 of the cDNA). These experiments revealed that guf mRNA is present in guf homozygous mutant embryos until cellular blastoderm at levels similar to those found in wild-type embryos. The store of maternal guf mRNA is depleted rapidly during gastrulation, and by the end of stage 7 only low levels of residual guf transcripts are detected in the invaginating posterior midgut. No guf expression was detected in guf embryos at later stages of embryogenesis. These data suggest that zygotic guf expression is eliminated in guf mutant embryos (Salzberg, 1996).

If Sxl is a negative regulator of guf, it was reasoned that a correlation might be detectable between the expression of guf RNA and Sxl protein. RNA in situ hybridizations using guf sequences as a probe on wild-type ovaries has shown that the expression of guf is dynamic and changes considerably as the germ cells mature. Early germ cells have very low levels of guf RNA, but further down the germarium, the levels increase. guf RNA is also detected in the follicle cells of the germarium. In stage 2 egg chambers (the first chamber past the germarium), expression decreases to very low levels in both the follicle and germ cells. After stage 2, guf RNA is expressed in the germ cells but is essentially undetectable in the follicle cells. As vitellogenesis begins, the levels in germ cells drop again and guf RNA is detected primarily in the oocyte. After this stage, high levels are seen in the nurse cells and oocyte, while follicle cell expression remains very low (Vied, 2003).

The cells in the germarium that show very low levels of guf RNA are the early germ cells, which also express high levels of cytoplasmic Sxl. guf RNA levels begin to increase in the region of the germarium where Sxl is downregulated, suggesting that Sxl might act negatively on guf. By contrast, in Sxl^f4 ovaries, guf RNA levels are found to be high throughout, including the germ cells immediately below the terminal filament. An increase in guf RNA is also seen in Sxl^f5 ovaries. These observations support the hypothesis that, normally, Sxl downregulates guf expression and suggest that the mutant Sxl^fs proteins are inefficient at this function. Consistent with the observation that decreasing the dose of guf partially rescues the Sxl^fs female sterile phenotype, wild-type levels of guf RNA are seen in the germ cells of Sxl^f5/Sxl^f1;guf^118-3/+ ovaries (Vied, 2003).

EFFECTS OF MUTATION

The gutfeeling (guf) gene was uncovered in a genetic screen for genes that are required for proper development of the embryonic peripheral nervous system. Mutations in guf cause defects in growth cone guidance and fasciculation and loss of expression of several neuronal markers in the embryonic peripheral and central nervous systems. guf is required for terminal differentiation of neuronal cells. Mutations in guf also affect the development of muscles in the embryo. In the absence of guf activity, myoblasts are formed properly, but myoblast fusion and further differentiation of muscle fibers is severely impaired. The guf gene was cloned and found to encode a 21-kD protein with a significant sequence similarity to the mammalian ornithine decarboxylase antizyme (OAZ). In mammals, OAZ plays a key regulatory role in the polyamine biosynthetic pathway through its binding to, and inhibition of, ornithine decarboxylase (ODC), the first enzyme in the pathway. The elaborate regulation of ODC activity in mammals still lacks a defined developmental role and little is known about the involvement of polyamines in cellular differentiation. Guf is the first antizyme-like protein identified in invertebrates (Salzberg, 1996).

Mutations in the gutfeeling, gene were uncovered in a genetic screen designed to identify genes that play a role in the development of the embryonic PNS (Kania, 1995). Embryos from 2000 strains carrying homozygous lethal P-element insertions on the second chromosome were stained with monoclonal antibody (MAb) 22C10 and examined for defects in the PNS. The guf gene was identified by three insertional mutations that cause similar phenotypes and were mapped to cytological position 48E5-12. Precise or near precise excision of a P-element insertion in guf reverts both the lethality and the PNS phenotype, demonstrating that the insertion causes the phenotype. Additional guf alleles were generated by imprecise excision of a P-element insertion. One allele corresponds to a small deficiency that removes most of the guf transcription unit and the entire open reading frame (Salzberg, 1996).

This deletion was found to abolish zygotic expression of guf and hereafter is referred to as a null allele. The insertional alleles cause phenotypes that are very similar to those observed in guf null embryos, suggesting that they are severe loss of function or null alleles (Salzberg, 1996).

Mutations in guf are embryonic lethal and cause multiple defects in the embryonic PNS. A smaller number of PNS neurons express 22C10 in homozygous guf embryos than in wild-type embryos and the overall intensity of 22C10 immunoreactivity is much decreased when compared to wild-type embryos (Kania, 1995). The lateral and ventral PNS clusters are more severely affected than the dorsal and ventral clusters. On average, six to seven neurons are labeled with MAb 22C10 in lateral clusters of guf embryos as compared to 10-11 in wild-type, and only three to four neurons label in the ventral cluster of mutant embryos as compared to seven in wild-type. The lateral chordotonal neurons seem to be most severely affected by mutations in guf and only one to three chordotonal neurons are evident in each lateral cluster as compared to five in wild-type embryos. However, the loss of 22C10 immunoreactivity is not restricted to chordotonal neurons and is evident in all types of sensory neurons (Salzberg, 1996).

In addition to the lack of 22C10 expression, PNS axons of guf embryos display severe defects in growth cone guidance and fasciculation. One of the most noticeable defects that is observed in all guf embryos is splitting of the intersegmental nerve (ISN) into two fascicles. The ISN 'loop' phenotype was observed in 57% of abdominal segments in guf homozygous mutant embryos. Crossing of the ISN into neighboring segments is much less frequent and was observed in single segments of -40% of the examined guf null embryos. Defects in growth cone guidance and fasciculation are not restricted to ISNs but are also evident in segmental nerves (Salzberg, 1996).

MAb 22C10 is a relatively late marker for PNS development, since neuronal cells express the 22C10 antigen only after they exit the cell cycle and start differentiating as neurons. Thus, lack of 22C10 expression may indicate a true loss of neurons stemming from a failure of precursor formation, failure of precursor division, transformation of neurons into lineage-related support cells, or cell death. Alternatively, this phenotype may reflect a failure of neuronal cells to fully differentiate and express late neuronal markers. To distinguish among these possibilities, a variety of cell-type-specific markers were used to assess number and identity of PNS cells in guf embryos of various developmental stages (Salzberg, 1996).

Since chordotonal neurons are among the most severely affected neurons the analysis focused on the formation of the conspicuous cluster of lateral chordotonal organs. To determine whether the precursors of chordotonal organs were affected in guf embryos, they were immunocytochemically stained with antibodies against the Atonal protein. The pattern of chordotonal SOPs in the mutant was found to be indistinguishable from that of wild-type embryos. To determine whether SOP division is impaired, the PNS of mutant embryos was examined with anti-Couch potato (Cpo) antibodies. The Cpo protein is present in nuclei of all the cells that constitute each sensory organ. Thus, the number of CPO positive cells reflects the number of cell divisions each precursor went through. The level of CPO protein is reduced in guf mutant embryos when compared to wild-type, but a wild-type number of CPO positive nuclei is evident in each chordotonal organ (Salzberg, 1996).

These observations indicate that zygotic guf is not required for the determination of SOPs or their division but is required during later stages of PNS development. Loss of 22C10 immunoreactivity may reflect a transformation of chordotonal neurons into another type of neurons that normally express lower levels of this antigen. Hence, embryos were stained with anti-Cut antibody that labels only external sensory organs and a subset of multiple dendritic neurons appraisal of whether the identity of chordotonal neurons is altered in guf mutants. The pattern of Cut expression appears normal in guf embryos, indicating no change in the identity of sensory organs. To determine whether the identity of cells within each chordotonal organ is altered, embryos were stained with anti-Prospero to reveal sheath cells, and the anti-Repo/RK2 antibody to reveal ligament cells. Normal numbers of sheath and ligament cells are evident in most lateral chordotonal organs of guf embryos. The results of these experiments suggest that loss of guf activity does not alter the identity of sensory organs, nor does it alter the identity of cells within each organ. It is concluded that guf is not required for cell fate determination but rather for terminal differentiation of the developing PNS cells (Salzberg, 1996).

Whereas cell type-specific antigens are expressed in a normal spatial pattern in the PNS of guf mutants, the level of many markers is greatly reduced. Reduced expression levels were observed for the 22C10 antigen and CPO protein in mature embryos. In addition, the level of the glial homeo protein Repo in ligament cells of the lateral chordotonal organs, but not in exit and peripheral glia, is severely decreased when compared to wild-type embryos. However, the levels of Prospero and Cut proteins in the PNS seem to remain unaffected by mutations in guf. These observations indicate that mutations in guf may exert their effect on late PNS differentiation by altering protein levels (Salzberg, 1996).

Since mutations in guf seem to affect most or all types of sensory neurons, the CNS of mutant embryos was examined to determine whether CNS neurons are affected in a similar fashion. guf embryos exhibited a severe reduction in the number of 22C10-expressing cells in the ventral nerve cord when compared to wild- type embryos. However, the number of neuroblasts in the CNS of stage 10-11 guf embryos appeared normal when visualized with antibodies against Prospero. The overall width of the ventral nerve cord appeared normal when examined with Nomarski optics. This suggests that as in the PNS, the lack of 22C10 expression in the CNS does not reflect a true loss of neuronal cells but a failure of these cells to fully differentiate as neurons (Salzberg, 1996).

To further assess late neuronal differentiation in the CNS, embryos were stained with antibodies against the synaptic vesicle protein Synaptotagmin. This protein serves as a terminal differentiation marker, as expression and synaptic localization of Synaptotagmin just precedes synaptic activity. Synaptotagmin is barely detectable in guf embryos when compared to wild type. This observation again supports the view that mutations in guf impair late neuronal differentiation by altering protein levels (Salzberg, 1996).

To examine the axonal pathways in the CNS of guf mutants, embryos were immunohistochemically stained with MAb 1D4 that recognizes the Fasciclin-II (FasII) protein. In wild-type embryos, anti-FASII staining reveals three distinct longitudinal tracts on either side of the midline. The organization of these tracts in guf embryos is aberrant. Although three distinct fascicles can be identified in some segments, most often only two fascicles are evident and these fascicles appear less tight than normal. The most medial fascicle is the least severely affected, whereas the most lateral fascicle is the most severely affected. In addition, the posterior commissure appears thicker than normal (Salzberg, 1996).

In summary, the defects observed in the CNS of guf embryos are very similar to those observed in the PNS and consist of alterations in the protein levels of neuronal-specific markers coupled to misrouting of axons and defects in fasciculation. These data suggest that guf plays a similar role in PNS and CNS development. However, further analysis of cell identities in the CNS of guf mutants is required to exclude the possible involvement of guf in cell fate determination in the CNS (Salzberg, 1996).

Examination of guf embryos with Nomarski optics indicates that mutant embryos also display defects in muscle development. To study the fully developed muscle pattern, embryos were generated with antibody against Drosophila muscle Myosin. This antibody labels all somatic muscle fibers, the cardial cells and the visceral muscles. Somatic muscles of the body wall are organized in a highly stereotyped pattern in wild-type embryos. This pattern is considerably disrupted in guf homozygous mutant embryos where numerous muscles are missing, displaced or exhibit aberrant shapes (Salzberg, 1996).

In addition, these embryos contain many mononucleate myoblasts that failed to fuse into muscle syncytia. Similar defects were observed in guf mutant embryos stained with antibodies against the spectrin-like MSP-300 protein that is expressed in muscles and chordotonal neurons. The level of MSP-300 protein is greatly reduced in muscle fibers and chordotonal neurons of mutant embryos when compared to wild type (Salzberg, 1996).

Anti-Myosin staining also revealed defects in the visceral and cardiac musculature. In wild-type embryos, a thin layer of evenly spread visceral muscles is detected around the gut. In guf embryos the visceral muscles seem to spread properly around the developing gut, however the thickness of this muscle layer appears less uniform than that of wild-type embryos leading to its rippled appearance. The constrictions of the gut form properly in guf mutants. The contractile part of the heart normally consists of a double row of Myosin expressing cardial cells that form a linear tube underneath the dorsal midline. Although the cardial cells form properly and the assembly of the heart into a linear tube appears relatively normal in guf mutants, Myosin expression in the cardial cells is essentially absent (Salzberg, 1996).

To determine whether the defects observed in the somatic musculature of guf embryos stem from aberrant specification of these muscles or from later differentiation defects, guf embryos were stained for nautilus RNA and for the Evenskipped (Eve) protein. nautilus is normally expressed in a subset of developing muscles in stage 12/13 embryos. This expression pattern does not appear to be altered in guf embryos. The Eve protein is also present in the nuclei of the dorsal-most muscle and pairs of pericardial cells in each abdominal segment. In guf mutants, the number of Eve positive cells appears similar to wild type, but the distribution of nuclei in the dorsal somatic muscle is abnormal. Instead of segregating into two crescent rows of nuclei on either side of the segment, these nuclei remain clumped together. Similarly, these dorsal muscles display abnormal morphology when visualized with anti-Myosin antibody and do not appear as broad and spread out as in wild type. The aberrant distribution of nuclei within muscle fibers is not restricted to the dorsal-most muscles and is evident in all muscle fibers when stained with antibodies against Mef2. These results indicate that the initial specification of individual muscles occurs properly in the absence of guf activity, but myoblast fusion and distribution of the syncytial nuclei within the developing muscle fibers is aberrant. Hence, it has been concluded that in the myogenic lineages, as in the PNS, guf is not required for cell type specification but for terminal differentiation of muscle cells. It is proposed that mutations in guf affect muscle development by reducing the level of various muscle-specific proteins (Salzberg, 1996).

These data suggest that guf plays a role in terminal differentiation of PNS cells following their specification as neurons or support cells but is not required for the specification of cell identities. Based on the observation that the levels of several neuronal-specific proteins are abnormally low in mutant embryos, it is proposed that guf is required for the regulation or maintenance of gene expression in the differentiating PNS. The failure of PNS cells to express late neuronal markers may explain the abnormal morphology of neuronal cells and the defects in growth cone guidance and fasciculation (Salzberg, 1996).

The pattern of body wall muscles is also disrupted in guf mutant embryos since many muscle fibers are missing or misplaced. In addition, the layer of visceral muscles surrounding the gut appears rippled and less regular than normal. Early subdivision of the mesoderm to somatic, visceral and cardiac mesoderm was found to occur normally in the absence of guf, and the initial specification of individual muscles does not seem to be affected. However, myoblast fusion and segregation of the syncytial nuclei within the developing muscles is aberrant leading to abnormal morphology or complete loss of specific muscle fibers. Based on these data it has been proposed that guf is required for differentiation but not cell fate determination of all types of embryonic muscles (Salzberg, 1996).

guf function in female germ cell development

To determine whether guf has a function in female germ cell development, germ cell clones were generated that were homozygous mutant for guf. Clones were induced by using the FLP/FRT mitotic recombination system in a background that had the SmD3+ genomic transgene, to preclude effects of this gene. When induced during the larval stage, clones of germ cells mutant for guf were not recovered in adult ovaries, suggesting a requirement for guf in the early stages of germ cell development. When induced in adults, germ cell clones did not develop beyond stage 2 egg chambers. Within 7 days of induction, many guf mutant cysts were observed in the germarium, and these showed no obvious defects, including their expression of Sxl. However, 14 days after induction, no guf mutant cysts were detected in germaria, but multiple degenerating egg chambers that had been pinched off were observed. These data suggest that, in the adult ovary, guf is not required in the early stages of germ cell development but in the cysts past the germarium, and that guf is needed either for the survival or the prevention of differentiation of germline stem cells. Since guf mutant germ cells show near normal Sxl expression, the data indicate that loss of guf does not have a major effect on Sxl expression in the early dividing germ cells. The proposal that Sxl maintains guf at low levels in region 1 of the germarium is consistent with this view (Vied, 2003).

Germ cells mutant for guf appear to develop through the germarium normally. Therefore whether ectopic expression of guf would affect early germ cell development was examined. Overexpression of Guf causes a decrease in the levels of cytoplasmic Sxl, resulting in fewer germ cells with high levels of the protein under the terminal filament. This effect requires efficient expression of Guf since it is only observed when the frameshift-independent variant of guf, is used. This latter observation also confirms that Guf translation requires frameshifting in vivo (Vied, 2003).

As a different means of increasing Guf levels, wild-type ovaries were treated with polyamines (spermidine). Polyamines increase the translation of endogenous guf RNA by increasing the rate of translational frameshifting. Spermidine causes a decrease in the number of early germ cells with high levels of cytoplasmic Sxl. This effect was observed within 3 h, implying that the change in Sxl is not caused by differentiation of the germ cells. The possibility that spermidine affects Sxl by a Guf-independent mechanism cannot be excluded. However, it is likely that an increase in Guf is involved given the overexpression data above (Vied, 2003).

The effect of Guf overexpression on Sxl in early germ cells strongly resembles the effect of overexpressing the patterning gene, hh (Vied, 2001). In the ovary, Hh is expressed in the terminal filament and cap cells, the somatic cells at the tip of the germarium. Hh regulates the proliferation of the follicle cells of the ovary and has minor effects on germ cell development (Vied, 2003).

To test for an epistatic interaction between hh and guf, hs-hh; guf^118-3/+ flies were generated. Removing one copy of guf counteracts the overexpression of Hh, and there is an increase in the number of early germ cells with high levels of cytoplasmic Sxl. Previous data (Vied, 2001) suggested that Hh regulates not only the levels but also the nuclear localization of Sxl in early germ cells. The coding sequence of Sxl suggests a leucine-rich Nuclear Export Signal (NES). Whether the Sxl NES is functional was tested using leptomycin B (LMB). LMB prevents the nuclear export of proteins by binding to exportin 1/CRM1, which binds directly to leucine-rich NES sequences. Wild-type ovaries treated with LMB show an increase in detectable nuclear Sxl. By contrast, early germ cells that have not been treated with LMB show no distinct nuclear foci of Sxl. Consistent with Hh affecting Sxl nuclear entry, overexpression of Hh in the presence of LMB significantly increases the amount of nuclear Sxl. Overexpression of Guf in the presence of LMB has a similar effect, and the effect of Guf is usually stronger than that of Hh (Vied, 2003).

To demonstrate that guf acts downstream of Hh in Sxl nuclear entry, Hh was overexpressed in a background with a reduced dose of guf (hs-hh; guf^118-3/+) and the ovaries were treated with LMB. As was seen for the degradation of Sxl, removal of one copy of guf counteracts Hh-regulated nuclear entry of Sxl, and Sxl remained predominantly in the cytoplasm. This observation suggests that Guf functions downstream of Hh in both nuclear entry and degradation, promoting the release of cytoplasmic Sxl in early germ cells (Vied, 2003).

Besides Sxl, Hh affects other germ cell-specific processes. Hh overexpression decreases the ß-Spectrin staining of the fusome. ß-Spectrin is one of the proteins associated with the fusome, a germline specific organelle that plays a role in establishing the interconnected 16-cell cyst. Overexpression of Guf does not appear to affect the fusome. Additionally, reducing the dose of guf does not reverse the effect of Hh overexpression on the fusome. These data implicate guf as a downstream effector of Hh, but only in regulating Sxl (Vied, 2003).

In Drosophila, Cyclin B is not essential for viability but is necessary for female fertility. Since Sxl has been shown to be important for the mitosis of early germ cells, a correlation between Sxl and Cyclin B was sought. In early germ cells, Cyclin B colocalizes with Sxl and is downregulated concomitantly with Sxl. Since Sxl and Cyclin B colocalize in early germ cells, the effect of Hh on Cyclin B was examined. Overexpression of Hh results in fewer Cyclin B-expressing early germ cells, as seen for Sxl. Furthermore, the germ cells that express Cyclin B show reduced levels of the protein. This similarity in response prompted a test to see whether Hh also regulates the nuclear entry of Cyclin B. Overexpression of Hh and treatment with LMB results in higher levels of nuclear Cyclin B than in wild-type early germ cells treated with LMB only. This observation suggests that Hh promotes the nuclear entry of Cyclin B (Vied, 2003).

Since Guf appears to act downstream of Hh to regulate Sxl, whether Cyclin B would also respond to Guf was also examined. Ectopic expression of Guf has the same effect on Cyclin B as on Sxl. Fewer early germ cells with cytoplasmic Cyclin B were observed, and Sxl and Cyclin B continued to colocalize. LMB treatment with Guf overexpression increases the nuclear levels of Cyclin B. As for Sxl, Cyclin B responds more strongly to Guf than Hh (Vied, 2003).

To determine whether Guf also acts downstream of Hh in affecting Cyclin B nuclear entry, hs-hh; guf^118-3/+ ovaries treated with LMB were examined. Under these conditions, Cyclin B was predominantly cytoplasmic, suggesting that Guf also functions downstream of Hh in the regulation of Cyclin B (Vied, 2003).

In vertebrates, the intracellular localization of Cyclin B is critical to its function. An NES within the cytoplasmic retention signal (CRS) allows Cyclin B to rapidly shuttle out of the nucleus. Phosphorylation of the CRS results in nuclear accumulation of Cyclin B and its associated Cyclin-dependent kinase with which Cyclin B initiates mitosis. Since Sxl and Cyclin B appear to undergo similar changes in response to Hh and Guf, tests were made to determine whether Cyclin B is affected by changes in Sxl.

Cyclin B was examined in ovaries with mutant Sxl protein. Like Sxl, Cyclin B is localized to the cytoplasm of Sxl^f4 germ cells. Treatment of Sxl^f4 ovaries with LMB caused the Sxl, but little Cyclin B, protein to accumulate in the germ cell nuclei. As overexpression of Hh with LMB treatment increases the nuclear levels of both Sxl and Cyclin B in wild-type ovaries, the intracellular localization of both proteins was examined in Sxl^f4 ovaries after similar treatment. The mutant Sxl protein accumulates in the nuclei, but surprisingly, Cyclin B remains in the cytoplasm. Consistent with this observation, ovaries doubly homozygous for Sxl^f4 and Su(fu) treated with LMB show nuclear accumulation of Sxl^f4 protein, but not Cyclin B. These results suggest that, normally, Sxl affects the rate of nuclear entry of Cyclin B. Interestingly, the mutations in the Sxl^fs alleles alter a region of the protein that is proline-rich. Proline-rich domains have been described as being involved in protein-protein interactions (Vied, 2003).

Sxl can enter the nucleus without the concomitant entry of Cyclin B. Cyclin B mutant ovaries were examined to test the assumption that Sxl does not require Cyclin B for nuclear exiting. A null mutation for cyclin B derived from the imprecise excision of a P-element in the 5' UTR (cycB3) is not lethal except when combined with other cell cycle regulators. The cycB3 mutation results in female sterility with many agametic ovarioles. When germ cells were present, the localization of Sxl was primarily cytoplasmic. Treatment with LMB shows that Sxl is able to enter the nucleus at levels comparable to those observed in wild-type ovaries\. These results suggest that Sxl does not require Cyclin B for nuclear exiting. Otherwise, Sxl would be primarily nuclear in the absence of Cyclin B (Vied, 2003).

REFERENCES

Search PubMed for articles about Drosophila Ornithine decarboxylase antizyme

Almrud, J. J., et al. (2000). Crystal structure of human ornithine decarboxylase at 2.1 A resolution: structural insights to antizyme binding. J. Mol. Biol. 295(1): 7-16. 10623504

Chattopadhyay, M. K., Murakami, Y. and Matsufuji, S. (2001). Antizyme regulates the degradation of ornithine decarboxylase in fission yeast Schizosaccharomyces pombe. Study in the spe2 knockout strains. J. Biol. Chem. 276(24): 21235-41. 11283013

Chen, H., MacDonald, A. and Coffino, P. (2003). Structural elements of antizymes 1 and 2 are required for proteasomal degradation of ornithine decarboxylase. J. Biol. Chem. 277(48): 45957-61. 12359729

Coffino, P. (2001a). Regulation of cellular polyamines by antizyme. Nat. Rev. Mol. Cell Biol. 2(3): 188-94. 11265248

Coffino, P. (2001b). Antizyme, a mediator of ubiquitin-independent proteasomal degradation. Biochimie 83: 319-323. 11295492

Hoyt, M. A., Zhang, M. and Coffino, P. (2003). Ubiquitin-independent Mechanisms of Mouse Ornithine Decarboxylase Degradation Are Conserved between Mammalian and Fungal Cells. J. Biol. Chem. 278(14): 12135-43. 12562772

Gruendler, C., Lin, Y., Farley, J. and Wang, T. (2001). Proteasomal degradation of Smad1 induced by bone morphogenetic proteins. J. Biol. Chem. 276: 46533-46543. 11571290

Gupta. R., et al. (2001). Effect of spermidine on the in vivo degradation of ornithine decarboxylase in Saccharomyces cerevisiae. Proc. Natl. Acad. Sci. 98(19): 10620-3. 11535806

Ichiba, T., et al. (1994). Functional regions of ornithine decarboxylase antizyme. Biochem. Biophys. Res. Com. 200: 1721-1727. 8185631

Ivanov, I. P., Simin, K., Letsou, A., Atkins, J. F. and Gesteland, R. F. (1998). The Drosophila gene for antizyme requires ribosomal frameshifting for expression and contains an intronic gene for snRNP SmD3 on the opposite strand. Mol. Cell. Biol. 18: 1553-1561. 9488472

Ivanov, I. P., et al. (2000a). Conservation of polyamine regulation by translational frameshifting from yeast to mammals. EMBO J. 19(8): 1907-17. 10775274

Ivanov, I. P., Rohrwasser, A., Terreros, D., Gesteland, R. F. and Atkins, J. F. (2000b). Discovery of a spermatogenesis stage-specific ornithine decarboxylase antizyme: Antizyme 3. Proc. Natl. Acad. Sci. 97: 4808-4813. 10781085

Kania, A., et al. (1995). P-element mutations affecting embryonic peripheral nervous system development in Drosophila melanogaster. Genetics 139: 1663-1678. 7789767

Li, X., Stebbins, B., Hoffman, L., Pratt, G., Rechsteiner, M. and Coffino P. (1996). The N terminus of antizyme promotes degradation of heterologous proteins. J. Biol. Chem. 271(8): 4441-6. 8626796

Matsufuji, S., et al. (1995). Autoregulatory frameshifting in decoding mammalian ornithine decarboxylase antizyme. Cell 80: 51-60. 7813017

Murakami, Y., Matsufuji, S., Kameji, T., Hayashi, S., Igarashi, K., Tamura, T., Tanaka, K., and Ichihara, A. (1992). Ornithine decarboxylase is degraded by the 26S proteasome without ubiquitination. Nature 360: 597-599. 1334232

Murakami, Y., Matsufuji, S., Hayashi, S. I., Tanahashi, N., and Tanaka, K. (1999). ATP-Dependent inactivation and sequestration of ornithine decarboxylase by the 26S proteasome are prerequisites for degradation. Mol. Cell. Biol. 19: 7216-7227. 10490656

Rom, E., and Kahana, C. (1994). Polyamines regulate the expression of ornithine decarboxylase antizyme in vitro by inducing ribosomal frame-shifting. Proc. Natl. Acad. Sci. 91: 3959-3963. 8171019

Salzberg, A., Golden, K., Bodmer, R. and Bellen, H. J. (1996). gutfeeling, a Drosophila gene encoding an Antizyme-like protein, is required for late differentiation of neurons and muscles. Genetics 144: 183-196

Satriano, J., et al. (1998). Agmatine suppresses proliferation by frameshift induction of antizyme and attenuation of cellular polyamine levels. J. Biol. Chem. 273(25): 15313-6. 9624108

Schenkel, H., Hanke, S., De Lorenzo, C., Schmitt, R. and Mechler, B. M. (2002). P elements inserted in the vicinity of or within the Drosophila snRNP SmD3 gene nested in the first intron of the Ornithine Decarboxylase Antizyme gene affect only the expression of SmD3. Genetics 161: 763-772. PubMed ID: 12072471

Scorcioni, F., Corti, A., Davalli, P., Astancolle, S. and Bettuzzi, S. (2001). Manipulation of the expression of regulatory genes of polyamine metabolism results in specific alterations of the cell-cycle progression. Biochem. J. 354(Pt 1): 217-23. 11171097

Vied, C. and Horabin, J. I. (2001). The sex determination master switch, Sex lethal, responds to Hedgehog signaling in the Drosophila germline. Development 128: 2649-2660. 11526072

Vied, C., Halachmi, N., Salzberg, A. and Horabin, J. I. (2003). Antizyme is a target of Sex-lethal in the Drosophila germline and appears to act downstream of Hedgehog to regulate Sex-lethal and Cyclin B. Dev. Bio. 253: 214-229. 12645926

Zhang, M., Pickart, C. M. and Coffino, P. (2003). Determinants of proteasome recognition of ornithine decarboxylase, a ubiquitin-independent substrate. EMBO J. 22(7): 1488-96. 12660156

Zhu, C., Lang, D. W. and Coffino, P. (1999). Antizyme2 is a negative regulator of ornithine decarboxylase and polyamine transport. J. Biol. Chem. 274(37): 26425-30. 10473601

date revised: 12 May 2003

Home page: The Interactive Fly © 2017 Thomas Brody, Ph.D.