Abstract
The Drosophila gene for snRNP SmD3 (SmD3) is contained in reverse orientation within the first intron of the Ornithine Decarboxylase Antizyme (AZ) gene. Previous studies show that two closely linked P elements cause the gutfeeling phenotype characterized by embryonic lethality and aberrant neuronal and muscle cell differentiation. However, the exact nature of the gene(s) affected in the gutfeeling phenotype remained unknown. This study shows that a series of P inserts located within the 5′-untranslated region (5′-UTR) of SmD3 or its promoter affects only the expression of SmD3. Our analysis reveals that the gutfeeling phenotype associated with P elements inserted in the 5′-UTR of SmD3 results from amorphic or strongly hypomorphic mutations. In contrast, P inserts in the SmD3 promoter region reduce the expression of SmD3 without abolishing it and produce larval lethality with overgrown imaginal discs, brain hemispheres, and hematopoietic organs. The lethality of these mutations could be rescued by an SmD3+ transgene. Finally, inactivation of AZ was obtained by complementing with SmD3+ the deficiency Df(2R)guf lex47 that uncovers both SmD3 and AZ. Interestingly, AZ inactivation causes a new phenotype characterized by late larval lethality and atrophy of the brain, imaginal discs, hematopoietic organs, and salivary glands.
ESTIMATION based on the sequencing of the Drosophila genome indicates that ∼7% of the genes may be nested within other genes (Ashburneret al. 1999). The first example of a nested gene was that of the pupal cuticle protein (Pcp) that is fully encoded within an intron of ade3 (Henikoffet al. 1986). Although numerous molecular data were collected on the organization of nested genes, the gathering of genetic data is impeded by the difficulty of assigning mutations to these genes and identifying their phenotypes.
In this article we describe the mapping and phenotypic analysis of mutations in two Drosophila genes encoding the ornithine decarboxylase antizyme (AZ) and the spliceosome core protein SmD3, respectively. Recent studies revealed that the SmD3 (CG8427) gene is included in reverse orientation within the first intron of the AZ (CG16747) gene (Ivanovet al. 1998) in band 48E on chromosome 2. Moreover, a series of independently isolated P elements are inserted in this intron. They cause embryonic (Kaniaet al. 1995; Salzberget al. 1996) or larval lethality with specific tissue overgrowth (Töröket al. 1993), although the exact nature of the affected genes remained unknown (Ivanovet al. 1998).
AZ controls and limits polyamine accumulation by blocking biosynthesis and transport of polyamines (Lindenet al. 1985). The best-understood function of AZ is its action on ornithine decarboxylase (ODC) catalyzing the first step in the synthesis of polyamines. Regulation of ODC half-life is mediated by AZ (Hayashi 1989; Hayashiet al. 1996) that binds to ODC, inhibits its activity, and targets it to degradation by the proteasome (Murakami et al. 1992, 1999; Li and Coffino 1994; Matsufujiet al. 1995). The AZ activity is induced by a striking unusual translation mechanism involving a +1 translational frameshift to decode AZ mRNA (Matsufujiet al. 1995). The AZ proteins are the products of a short first open reading frame 1 (ORF1) and a longer ORF2. At the codon preceding the stop codon of ORF1, translation shifts into the +1 frame to align translation with ORF2. Since the mammalian genome contains five nonallelic AZ homologs, inactivation of a specific AZ gene may remain phenotypically undetectable due to the supply of redundant AZ functions (Coffino 2000). However, overexpression of ODC, the reverse of AZ inactivation, causes dramatic changes in oocyte or sperm production of transgenic mice (Halmekytöet al. 1991) and results in neoplastic transformation in cultured cells (Auvinenet al. 1992; Cliffordet al. 1995) and transgenic mice (Megoshet al. 1995; O'Brienet al. 1997). Reciprocally, targeted AZ expression in the skin of transgenic mice reduces tumor promoter induction of ODC (Feithet al. 2001), indicating that polyamine regulation may play an important role in the control of cell proliferation.
The SmD3 protein belongs to a family of polypeptides present in eukaryotes that share conserved motifs forming the Sm domain, followed by variable C-terminal extensions. The Sm proteins are constituents of the small nuclear ribonucleoprotein particles (snRNPs) that participate in many different RNA-processing reactions and are essential for viability (Rymond 1993; Royet al. 1995; Bordonné and Tarassov 1996). In particular, the U1, U2, U4, and U5 snRNPs are involved in the splicing of U2-dependent introns (Yuet al. 1998), whereas U7 snRNP is essential for the 3′-end formation of histone mRNA (Grimmet al. 1993). Moreover, the telomerase of the budding yeast was shown to be an Sm-containing snRNP (Setoet al. 1999). Thus, the current model for the function of the Sm proteins envisages their involvement in snRNP assembly and nuclear transport (Hamm et al. 1987, 1990; Fischeret al. 1993; Nelissenet al. 1994; Grimmet al. 1997). A common feature of snRNPs is the occurrence of RNA polymerase II transcripts whose 5′-end is usually capped and modified by hypermethylation (Reddy and Busch 1983). Structural analysis reveals that the Sm domains mediate protein-protein interactions and form a ring/doughnut structure in which the positively charged interior is thought to bind with the Sm site of the snRNAs (Kambachet al. 1999). In particular, three of the Sm proteins, SmB, SmD1, and SmD3, make direct contact with the 5′ exon region close to the 5′ splice site (Zhang and Rosbash 1999). Interestingly, among the seven Sm proteins, only those three contain long and positively charged C-terminal tails that make direct contact with the pre-mRNA in a largely sequence-independent manner (Zhanget al. 2001). The truncation of the tail of the Sm proteins is unessential for cell viability, although the absence of the SmD3 tail and combinations of truncated SmD3 with truncated forms of SmD1 and SmB result in poor growth and temperature sensitivity (Zhanget al. 2001).
Thus, genetic characterization of both AZ and SmD3 genes may provide new insights into their respective functions in a metazoan organism. The data presented here show that AZ inactivation in Drosophila leads to a drastic reduction of cell proliferation in numerous larval and imaginal tissues. By contrast, downregulation of SmD3 results in tissue overgrowth, whereas complete SmD3 inactivation causes embryonic lethality.
MATERIALS AND METHODS
Drosophila stocks and mutant phenotype determination: The P-LacW mutant lines (Töröket al. 1993) were provided by Istvan Kiss and the EP lines by the Bloomington Drosophila Stock Center. To select homozygous and hemizygous mutant larvae we made use of a CyO balancer chromosome expressing a green fluorescence protein, CyO, P[GFP+], which was kindly provided by Gunter Reuter. All the stocks were maintained on standard Drosophila medium at room temperature; all crosses were done at 25°. The phenotype of the different organs was examined by dissecting 5-day-old wild-type larvae and 12- to 15-day-old mutant larvae in Ringer's (Becker 1959). For examination of cell nuclei, dissected tissues were fixed for 20 min in 4% paraformaldehyde in PBS at room temperature. The tissues were treated for 1 hr with 400 μg/ml RNAse and then stained for 1 hr with 0.2 ng/ml 4′,6-diamidino-2-phenylindole (DAPI), mounted in Vectashield embedding medium (Vector Laboratories, Burlingame, CA), and examined under a Zeiss fluorescence microscope.
Isolation of the ΔLR51 deletion by imprecise excision of the P element: The isolation of the ΔLR51 deletion by imprecise excision of the P-LacW element present at position 48E in l(2)k090-37 was carried out as previously described (Töröket al. 1995).
SmD3+ transgene: The 2.3-kb SalI genomic fragment containing the SmD3 gene sequence was cloned in the pCaSpeR4 transformation vector, and this construct was transformed into w1118 homozygous embryos as described previously (Töröket al. 1999). For these studies, a line with an insertion on the third chromosome was used.
Nucleic acid procedures: Basic DNA manipulation, including plasmid rescue, genomic and cDNA library screening, DNA sequencing, RNA purification, and Northern blotting, were done as previously described (Török et al. 1995, 1999). Mapping of the breakpoints of the ΔLR51 and Df(2R)guflex47 deletions was performed as previously described (Hartensteinet al. 1997) by using the following primers. For ΔLR51 the primers were: Sal-271, 5′-TAACGAAGCCACTACTAAACACT-3′; Sal-1981, 5′-TAGAGAATAGAAATGCGTTGT-3′. For Df(2R)guflex47 the primers were: 1-1089, 5′-CTCTTAACGTGGCTGGCCATAA TTG-3′; Sal-2226, 5′-TAGAGAATAGAAATGCGTTGT-3′. Following PCR amplification, the products were cloned in the pCR2.1 vector (Invitrogen, Carlsbad, CA) and sequenced using T3 and T7 primers. Quantification of mRNA levels was performed on total RNA extracted from wild-type, l(2)k090-37, and l(2)k131-07 third instar larvae with the following primers for the A class of AZ transcripts: 5′-TATACTAACTTTTG ACGACATCA-3′ in combination with 1-1089 and, for the class D transcripts, 5′-TTAGCGTCATAAAAGCG-3′ in combination with 1-1089. PCR products were resolved on 1.0% agarose gel and stained with ethidium bromide, and the DNA bands were quantified using a Bio-Print UV transilluminator and a Bio-Capt program (Vilber Lourmat, Marne-La-Vallée, France).
Antibody preparation and Western blot analysis: Polyclonal antibodies against a synthetic peptide SIGVPIKVLHEAEGHI ITC corresponding to residues 2–20 of the deduced SmD3 protein sequence conjugated to maleimide-activated keyhole limpet hemocyanin were raised in rabbits and purified as previously described (Strandet al. 1994). Western blots and immunodetection were performed as previously described (Töröket al. 1995).
In vitro translation: In vitro translation was performed by using the TNT transcription-translation-coupled reticulocyte lysate system (Promega Biotec, Madison, WI) with AZ cDNAs subcloned in pBlueScript SK+ vector as templates (Stratagene, La Jolla, CA).
RESULTS
The spliceosome core protein SmD3 gene is nested within the first intron of the Ornithine Decarboxylase Anti-zyme gene: As shown in Figure 1A, the SmD3 gene is located within the first intron of the gene encoding the AZ gene, in reverse orientation by comparison to that of AZ (Ivanovet al. 1998). The SmD3 gene consists of two exons and the 5′-end of the largest SmD3 cDNA that we isolated is located 522 nucleotides upstream from the acceptor site of the second AZ exon. SmD3 encodes a single type of transcript made of two exons. Sequence alignment of >70 isolated cDNAs and available expressed sequence tags (ESTs) in databases revealed four different classes of AZ transcripts, designated as A, B, C, and D, that share common second and third exons but differ by their first exon. In all classes of AZ transcripts, with the exception of class B, the first exon contains an AUG initiator codon in the ORF. Since AZ protein synthesis needs a frameshifting from ORF1 to ORF2 that is triggered by the presence of polyamines (Rom and Kahana 1993; Matsufujiet al. 1995), we tested whether the four classes of transcripts could produce full-size AZ polypeptides. We found that only the transcripts A and C could be translated in vitro into polypeptides with the expected masses of 28 and 30 kD, corresponding to proteins with a length of 254 and 272 amino acid residues, respectively (Figure 2B). As expected, the synthesis of these two proteins was enhanced by the presence of 0.8 mm spermidine. Although the D class of transcripts contains an AUG initiator codon in ORF1 in exon 1 that could give rise to a product of 249 residues, it remained untranslatable in the reticulocyte in vitro translation system. Moreover, no translation product could be detected with transcript B that contains no AUG codon in ORF1 of exon 1. Interestingly, transcripts B and D correspond to infrequent cDNAs (Figure 2A) and may thus represent rare splicing products. In vitro translation of an SmD3 cDNA produces a polypeptide with the expected mass of 18 kD (Figure 2C), whose synthesis is inhibited in the presence of 0.8 mm spermidine (data not shown).
—The snRNP SmD3 gene is nested in reverse orientation within the AZ gene. (A) Molecular organization of the AZ-SmD3 gene locus and transcripts. The AZ-SmD3 locus encompasses ∼8 kb of unique genomic DNA from nucleotide 7,167,000 in the 48E region on chromosome 2 (Celera Genomics contig: AE003823). The SmD3 gene encodes a single size transcript made of two exons whereas the AZ gene produces four classes of transcripts characterized by four divergent exons 1 and two shared exons 2 and 3. The positions of the six P elements affecting the expression of the SmD3 gene are indicated by red triangles, whereas the location of the P insert in EP(2)2104 that exerts no effect on the expression of both SmD3 and AZ is indicated by a green triangle. Translatable coding domains of SmD3 and AZ transcripts, as tested in the reticulocyte cell-free system, are represented by solid blue boxes. The orientation of the SmD3 and AZ transcription is indicated by an arrow. The SmD3+ genomic fragment used to generate transgenic flies is displayed above the map by a blue bar. The extensions of the interstitial deletions ΔLR51 and Df(2R)guf lex47 are shown below the map by yellow bars. (B) Nucleotide sequence of the upstream control domain (lowercase letters) and transcribed region (uppercase letters) of SmD3 where the P elements (triangles) analyzed in this study are located.
—In vitro translation of Drosophila Antizyme isoforms in the reticulocyte cell-free system. (A) Antizyme isoforms. cDNA and EST sequences can be grouped into four classes: A, B, C, and D, according to the ORF1 sequence present in the first exon (see Figure 1A). The relative abundance of the identified sequences is indicated on the right. The AUG initiator codon of ORF1 present in the first exon of the A, C, and D transcripts, as well as the UAG termination codon in ORF2, is indicated above each transcript. No initiator codon could be detected in ORF1 of the first exon of the B transcript. Sequences translated in the reticulocyte cell-free system are represented by boxes with oblique dots. The position of the +1 translational shift is indicated by a box with vertical dots. (B) In vitro translation of the four classes of AZ transcripts. The four classes of AZ cDNAs were translated in a coupled transcription-translation reticulocyte lysate system with (+) or without (–) 0.8 mm spermidine in the presence of [35S]methionine, and the products were separated on a 12% SDS-polyacrylamide gel. Products were identified by autoradiography. The products translated from the A and C transcripts display the expected masses of 28 and 30 kD and are indicated by arrow and arrowhead, respectively. Their translation is increased in the presence of 0.8 mm spermidine. (C) In vitro translation of the SmD3 transcript. In vitro translation of SmD3 sense RNA strand produced a single polypeptide with a molecular mass of 18 kD (arrow) whereas antisense RNA gave rise to no specific products.
Disruption of SmD3 expression by P inserts exerts no effect on AZ function: Since the SmD3 gene is contained within the first intron of AZ (Figure 1A), it was important to determine whether the various P elements found to be inserted in this intron affect the expression of both SmD3 and AZ genes or only one of them. Previous genetic analyses of recessive mutations causing tissue overgrowth during larval development (Töröket al. 1993) or embryonic lethality (Kaniaet al. 1995; Salzberget al. 1996) revealed that five independently isolated P-element-induced mutations, including l(2)k090-37, l(2)k131-07, l(2)k154-01, l(2)k095-40, and l(2)k118-03 were mapped in the AZ-SmD3 locus. In addition, a search in the Berkeley Drosophila Genome Project (http://www.fruitfly.org) and FlyBase (http://www.flybase.bio.indiana.edu) showed the occurrence of three additional P elements, namely l(2)02833, EP(2)2104, and EP(2)2176, inserted in the vicinity of the other five transposons. With the exception of the line l(2)02833 that was discarded from available stock collections, we genetically analyzed these mutations. By complementation tests we found that all mutations were allelic with the exclusion of EP(2)2104 (Table 1). The six allelic mutations were also lethal in combination either with Df(2R)guflex-47, which uncovers part of the proximal coding sequences of both AZ and SmD3 genes, or with ΔLR51, a small interstitial deletion (Figure 1A) generated by imprecise excision of the P insert in l(2)k090-37. This deletion removed a segment of 337 nucleotides upstream from the insertion site of l(2)k090-37, uncovering the promoter region of SmD3. As shown in Figure 1B, the six allelic P inserts were located within a segment of 79 nucleotides spanning the promoter and 5′-untranslated region (5′-UTR) of the SmD3 gene, whereas the site of the EP(2)2104 insertion was positioned farther away from the 5′-end of SmD3. Investigations of the lethal phase of homoallelic and heteroallelic mutant combinations allowed us to classify these mutations in two categories (Table 1). One category of P insert, including l(2)k095-40 and l(2)k118-03, alone or in combination with Df(2R)guf lex47, produced embryonic lethality, whereas the other category, including l(2)k090-37, l(2)k131-07, l(2)k154-01, and EP(2)2176, led to late larval lethality. These findings suggest that embryonic lethality results from a complete or nearly complete loss of gene function, whereas the late larval lethality is caused by hypomorphic mutations.
Lethal phase of SmD3 and AZ mutant combinations and rescue with an SmD3+ transgene
To determine whether the P inserts affect only the expression of SmD3 or both SmD3 and AZ, we examined the level of transcripts for both genes in wild-type and homozygous mutant larvae. Northern blot analysis (Figure 3A) showed that the level of SmD3 transcript was strongly reduced in homozygous l(2)k090-37 and l(2)k131-07 larvae whereas AZ expression remained essentially unaffected. Similar effects were noticed in homozygous l(2)k154-01 larvae (data not shown). Western blot analysis confirmed that the level of SmD3 protein expression (Figure 3B) was lowered in l(2)k131-07/+ larvae and strongly reduced in homozygous l(2)k131-07 larvae. These data indicate that, with the exception of EP(2)2104, all the other P inserts located in the first AZ intron between the 5′-UTR of SmD3 and the second AZ exon alter only the expression of SmD3 without affecting that of AZ.
To further confirm that the lethality associated with the P inserts was due only to a disruption of SmD3 expression, we tested whether a cloned DNA segment containing the sequence of the wild-type SmD3 gene could restore viability of homozygous mutant animals. By using P-element transformation and genetic crosses we found that a 2.3-kb SalI-DNA fragment containing the SmD3 gene (Figure 1A) was fully sufficient to restore the viability of homozygous mutant animals displaying either embryonic or larval lethality (Table 1). Altogether our results showed that the six lethal P inserts affect only the SmD3 function, but not that of AZ, although they are located in the first intron of this gene.
Inactivation of AZ and SmD3 leads to distinct phenotypes: Since the SmD3+ transgene could fully restore the development of SmD3-deficient animals, albeit not that of animals deficient for both SmD3 and AZ, we investigated the phenotype of animals bearing the deficiency Df(2)guf lex47 complemented with an SmD3+ transgene. In these animals only the AZ gene is inactivated. We found that AZ inactivation produced larvae whose growth was considerably delayed. These larvae were unable to pupariate. We found that organs, such as the brain, the hematopoietic organs, the imaginal discs, and the salivary glands of 12-day-old mutant larvae (Figure 4, C, F, I, and L), were considerably reduced in size by comparison to wild-type tissues (Figure 4, A, D, G, and J). In particular, we noticed that the number of cells in the imaginal rings were less numerous in AZ than in wild-type salivary glands (inserts of Figure 4, J and L). In contrast, the gut (Figure 4O), the Malpighian tubules, and the fat bodies (not shown) were nearly normal in size and shape.
Determination of the lethal phases in the SmD3 mutant lines revealed two distinct phases of lethality. With respect to the location of the P inserts, those located in the 5′-UTR of SmD3 were found to produce embryonic lethality whereas the inserts in the promoter region cause late larval lethality. We investigated the phenotype in both categories of mutants. Dissection of 12-day-old l(2)k090-37 mutant larvae revealed overgrowth of the brain (Figure 4B), hematopoietic organs (Figure 4E), and imaginal discs (Figure 4H) and a reduction of the size of the salivary glands (Figure 4K). Similar abnormalities were noted in l(2)k131-07, l(2)k154-01, and EP(2)2176 or in combinations of the three mutant lines with Df(2R)guf lex47 or Df(2R)ΔLR51 (data not shown).
—Expression of SmD3 and AZ in mutant and wild-type third instar larvae. (A) Northern blot containing poly(A)+ RNA extracted from 12-day-old mutant l(2)k131-07 and l(2)k090-37 larvae and 5-day-old wild-type larvae was successively probed with 32P-labeled SmD3, AZ, and β-tubulin cDNAs. (B) Expression of SmD3 in proteins extracted from wild-type, heterozygous, and homozygous l(2)k090-37 third instar larvae shows that the level of SmD3 is greatly reduced in mutant larvae. The amount of proteins in the different lanes was first equalized by comparing aliquots in a Coomassie blue-stained gel.
We also examined the embryonic phenotype of l(2)k095-40 and l(2)k118-03 and found that it was consistent with the previously described gutfeeling phenotype (Salzberget al. 1996). In particular, immunostaining of l(2)k095-40 or l(2)k118-03 embryos with the monoclonal antibody Mab 22C10 (Fujitaet al. 1982; Goodmanet al. 1984) showed that the lateral chordotonal neurons and the ventral nerve cord were severely affected when compared to wild-type embryos (data not shown), confirming that the gutfeeling phenotype results from an inactivation of the SmD3 gene.
DISCUSSION
Narrow-range effects of P elements inserted in the AZ-SmD3 locus: To elucidate the respective function of both SmD3 and AZ genes, we addressed the question of whether P elements inserted within the SmD3 gene alter the expression of only the SmD3 gene or of both genes. In this article, we demonstrate genetically and molecularly that six P elements inserted in SmD3 exert narrowrange effects on SmD3 but have no effect on AZ expression. These elements inserted over a region of 77 nucleotides covering part of the promoter and 5′-UTR of SmD3 (from nucleotides –29 to +48 with regard to the putative transcription start point) repress only the expression of SmD3. Furthermore, genetic investigation of the EP(2)2104 P transposon located at position –70 showed that this insert fully complements the mutation generated by the other six inserts. This complementation indicates that, first, the P insert in EP(2)2104 exerts no detrimental effect on the expression of both SmD3 and AZ and, second, that the SmD3 promoter region is composed of a relatively short sequence of <70 nucleotides.
The sequence of the SmD3 promoter is relatively GC rich (45 G + C/70 nucleotides) and contains no obvious TATA or CAAT motifs. Moreover, the beginning of the SmD3 transcription unit is devoid of an initiator T-C-A+1-G/T-T-T/C consensus sequence (Arkhipova 1995) and a downstream promoter element with a conserved sequence A/G-G-A/T-C-G-T-G (Kutach and Kadonaga 2000). These features indicate that the basal transcription of SmD3 requires other core promoter elements that remain yet to be discovered. The relatively small size of the promoter region suggests that the expression of SmD3 should be constitutive and indeed we found that both SmD3 transcripts and proteins are ubiquitously expressed during embryogenesis and late larval development (data not shown).
Among the 7 P-element inserts scattered over <120 nucleotides, only 2 are inserted at the same nucleotide position. This is in contrast to P elements located in the singed promoter region of which 7 out of 10 are inserted in the same position (Roihaet al. 1988; Paterson and O'Hare 1991). The scattered distribution of the P inserts may reflect the transcriptional activity of SmD3 in premeiotic germline cells, where P elements usually transpose. By contrast, the absence of recorded P inserts in the AZ gene (Spradlinget al. 1999) suggests that this gene either is expressed at a low level in premeiotic germline cells or contains no P-element target sequence.
—Morphology of wild-type and mutant SmD3 and AZ tissues. Whole-mount organs were stained with DAPI. (A, D, G, J, and M) Fiveday-old wild-type, (B, E, H, K, and N) 12-day-old SmD3 l(2)k090-37, and (C, F, I, L, and O) 12-day-old AZ Df(2R)guflex47; P{w+ SmD3+} mutant larvae. (A–C) Brain and associated imaginal discs with the position of optic lobes (o) and the ventral ganglion (v). Mutant SmD3 optic lobes are hypertrophied whereas mutant AZ brains are considerably reduced in size. (D and E) Brain and associated hematopoietic organs (h). These organs are considerably enlarged in mutant SmD3 larvae but strongly reduced in mutant AZ larvae. Inset in F displays a twofold enlargement of the hematopoietic organs. (G–I) Wing (w), haltere (ha), and leg (l) imaginal discs. By comparison to wild-type discs, the SmD3 discs are enlarged whereas the AZ discs are reduced in size. (J–L) Salivary glands. In both SmD3 and AZ larvae, the size of these organs is reduced. Insets in J display a fourfold enlargement of the imaginal rings at the neck of the salivary glands. (M–O) Midgut and proventriculus (p). In both SmD3 and AZ mutant larvae, the size of the digestive tract is nearly normal.
The developmental rescue of SmD3 mutants with an SmD3+ transgene demonstrates that the AZ transcription is not affected by the insertion of P-LacW elements within its first intron that nearly triples the size of this intron from ∼5.9 to ∼16.7 kb in length. Conversely, the ΔLR51 interstitial deletion that removes a segment of ∼0.3 kb yields no effect on AZ function. These findings sustain the idea that this intron could accept additional sequences or suffer deletions without impairing AZ transcription and may thus provide an explanation for a fortuitous insertion of the SmD3 gene during evolution. However, the finding that transcriptional initiation of the D transcript occurs between the SmD3 gene and the second AZ exon provoked us first to test whether expression of this type of transcript would not be enhanced when a P element is inserted in the first AZ intron. Therefore we first investigated the translational ability of this category of transcripts and, second, checked whether the amount of D transcripts would be increased in mutant third instar larvae carrying a P insert in the SmD3 gene. Among the four classes of AZ transcripts we found that only the two most abundant transcripts displayed translational ability in the reticulocyte cell-free system. Among the other two classes of AZ transcripts, the B transcript was devoid of initiator codon in ORF1 and, as expected, was unable to give rise to a translation product. In contrast the D transcript contains an initiator codon present in the ORF1 that is preceded by a sequence in conformity with the Drosophila translation start consensus sequence ANN(C/A) A (A/C) A/C) ATGN (Cavener 1987) but for unknown reasons remained untranslatable. By using specific primers for the A and D transcripts, we were unable to notice any enhanced level of D transcripts in homozygous l(2)k090-37 and l(2)k131-04 mutant larvae by comparison to wild type (data not shown). In both mutant and wild-type larvae the most abundant AZ transcripts are represented by class A. Thus, we conclude that the high level of AZ expression detected in homozygous l(2)k090-37 and l(2)k131-07 larvae results from the transcription of AZ mRNAs initiated distally from the SmD3 gene, indicating that the insertion of relatively large sequences has no effect on the transcription and processing of AZ mRNAs.
Inactivation of the SmD3 gene causes the gutfeeling phenotype: Previously, the gutfeeling phenotype was attributed to the inactivation of the AZ gene (Salzberget al. 1996). Here, we show that insertion of P elements in the 5′-UTR of the SmD3 gene is responsible for this phenotype that is characterized by defects in the organization of the embryonic peripheral and central nervous systems and in the differentiation of embryonic muscles (Salzberget al. 1996). Thus, the absence of a spliceosome core protein produces developmental defects that become apparent at midembryogenesis at the time of nerve and muscle cell differentiation. Until that stage, development of the mutant embryos appears to be normal due to the maternal contribution of SmD3. However, when zygotic SmD3 expression is required, differentiation of specific cells becomes impaired and development is arrested. No SmD3 mutation has yet been found in mammals.
However, the gutfeeling features identified in Drosophila are highly reminiscent of those found in human spinal muscular atrophy (SMA) in which the anterior horn cells of the spinal cord degenerate (Lefebvreet al. 1998). This disorder leads to progressive paralysis of the trunk and limb associated with muscular atrophy. Interestingly, SMA is caused by mutations in the survival of motor neurons (SMN) gene (Lefebvreet al. 1995). The SMN protein localizes both to the cytoplasm and the nucleus (Liu and Dreyfuss 1996), where it is essentially involved in splicing (Pellizzoniet al. 1998). This protein is part of a large complex containing Sm proteins of spliceosomal uridine-rich small ribonucleoproteins, including SmD3 (Liuet al. 1997; Friesen and Dreyfuss 2000). Since defects in the interactions between the SMN and Sm proteins inhibit assembly of snRNP complexes (Bühleret al. 1999), it is conceivable that inactivation of the human SmD3 gene function would similarly affect the assembly of these complexes and result in a pathogenesis resembling that of SMA.
Reduced expression of the SmD3 gene leads to tissue overgrowth: By contrast, we found that reduction in SmD3 expression by P elements inserted in its presumed promoter region produces a distinct phenotype, characterized by prolonged larval life and moderate overgrowth of numerous imaginal tissues. Due to an estimated slower, albeit persistent, rate of cell proliferation, the overgrowth of the brain, imaginal discs, and hematopoeitic organs becomes visible only during the prolonged larval development. This finding indicates that a strongly reduced, but detectable, production of the SmD3 protein may alter processes involved in the regulation of cell growth and proliferation. The tissue overgrowth resulting from a reduced expression of SmD3 is in apparent contrast with the arrest of embryonic development noticed in the absence of SmD3 expression. This discrepancy can be explained by a residual activity of SmD3 that is sufficient to allow the splicing of transcripts required for cell growth but not those required for cell differentiation.
A similar pattern of tissue overgrowth was observed in the case of mutations reducing the expression of the ribosomal-associated RpS21 and P40 factors (Töröket al. 1999) and the ribosomal protein RpS6 (Watsonet al. 1992; Stewart and Denell 1993). These proteins play a critical role in the initiation of protein synthesis and may contribute to the ability of the 40S ribosomal subunit to recognize specific mRNAs (Palen and Traugh 1987). A similar role could be assumed when the expression of SmD3 is reduced. This reduction could specifically affect the splicing of transcripts involved in the regulation of cell growth and proliferation and thus lead to a slower rate of cell proliferation. In these respects it would be important to determine whether defects in splicing or translation control characterize human tumors.
Inactivation of the AZ gene results in a tissue-specific block of cell growth and proliferation: By contrast to the mammalian genome that comprises at least four different AZ genes (Ivanovet al. 2000), the Drosophila genome contains a single AZ gene. Our analysis reveals that zygotic inactivation of the Drosophila AZ gene blocks cell growth and proliferation in numerous larval and imaginal tissues and demonstrates that AZ is an essential gene. Our data show that the multifunctional activities of AZ in the regulation of polyamine metabolism that inhibits ODC activity, stimulates ODC degradation, and suppresses polyamine uptake could not be replaced in Drosophila by another enzyme. Moreover, we were unable to detect any embryonic phenotype, presumably due to a substantial maternal contribution in AZ proteins that should become depleted at the beginning of the larval life. In AZ-deficient larvae the majority of the tissues are atrophic and made of cells smaller than normal, indicating that defects in polyamine metabolism exert detrimental consequences on the rate of cell growth and proliferation. However, cells in a few tissues, such as the digestive tract, the Malphigian tubules, and the fat bodies, display a nearly normal size, suggesting that these cells either are insensitive to high levels of polyamines or possess different mechanisms to reduce excessive amounts of polyamines. Interestingly, no expression of AZ could be detected by in situ hybridization in these tissues with the exception of the imaginal cells in the gut (data not shown). In no case were we able to detect any tissue overgrowth resulting from an AZ deficiency, as would have been expected from studies on transgenic mice (Megoshet al. 1995; O'Brienet al. 1997) or cell cultures (Auvinenet al. 1992; Cliffordet al. 1995). However, as in the case of SmD3, it is possible that a reduction in the production of AZ may give rise to a distinct phenotype. Availability of point mutations or P inserts in the AZ promoter, which would decrease but not abolish AZ expression, will provide answers to this question. High levels of polyamines are detected in numerous tumor cell lines, and overexpression of ODC in transgenic mice favors tumor development whereas overexpression of AZ reduces tumor promoter induction by ODC and sensitivity to chemical carcinogenesis (Feithet al. 2001). In these respects, similar experiments could be performed in a range of Drosophila tumors to determine whether spatiotemporal controlled expression of AZ could suppress tumor development.
In conclusion, we have shown that P-element inserts can be used to assign distinct functions to the SmD3 gene and the AZ gene in which SmD3 is included in the first intron. Furthermore, we have demonstrated that the gutfeeling phenotype is the result of amorphic or strongly hypomorphic mutations in the SmD3 gene. We therefore propose that the two genes studied in this article should be designated by the name of their respective product, namely SmD3 and AZ.
Acknowledgments
We thank Sylvia Oksas, Corinna Klein, and Christine Schmitt-Mbamunyo for excellent technical assistance and Istvan Kiss for providing us with the mutant fly stocks. This work was supported by grants of the Deutsche Forschungsgemeinschaft (436UNG113/81/0-2) within the framework of a German-Hungarian program of scientific cooperation and the European Commission (QLRI-CT-2000-00915).
Footnotes
-
Communicating editor: T. Schüpbach
- Received November 19, 2001.
- Accepted March 20, 2002.
- Copyright © 2002 by the Genetics Society of America
