Temporal and Spatial Expression of the Cell-Cycle Regulator Cul-1 in Drosophila and Its Stimulation by Radiation-Induced Apoptosis

In eukaryotic cells, passage through different stages of the cell cycle is controlled by a variety of processes including the synthesis or activation of regulatory proteins as well as the destruction of others. Moreover, cell-cycle transitions are triggered by degradation of specific proteins (Cohen-Fix and Koshland, 1997; Lyapina et al., 1998), one of which is associated with ubiquitin-dependent proteolysis of cell-cycle regulators by the proteasome complex (Deshaies, 1995). Cell regulators targeted for destruction, including phosphorylated cyclins, are labeled by specific ubiquitination with the help of evolutionarily conserved ubiquitin ligase complexes known as SCFs (Lyapina et al., 1998). Studies of these complexes in budding yeast and human cell cultures revealed that they contain Cdc53 and Cul-1 proteins, respectively (Feldman et al., 1997; Skowyra et al., 1997). Both proteins belong to a family, the cullins, of which there are at least five members in nematodes and six in humans (Kipreos et al., 1996). Though cullins share a significant homology, probably reflecting their common biochemical properties, the functions of most, except for Cul-1 and Cdc53, have not been identified. Mutations of both cdc53 in budding yeast and cul-1 in nematode affect cell-cycle control and are lethal, indicating that other members of the cullin family are then unable to perform their function.

The high level of similarity between the yeast and nematode proteins suggests that their genes are orthologues, but mutations in their genes in these two organisms are manifested differently. While the cdc53 gene mutation causes developmental arrest at the G₁/S-phase transition in yeast, the nematode cul-1 mutant displays hyperplasia in all postembryonic tissues because affected cells are unable to exit the mitotic cycle and thus divide excessively. This latter observation led Kipreos et al. (1996) to propose that Cul-1 is a negative repressor of proliferation in nematodes. The observed phenotypic difference between yeast and nematode mutants necessitates further investigation into the role of cul-1 orthologues in other organisms.

Here we describe the isolation and characterization of cul-1 genes in two popular model organisms: Drosophila melanogaster and mouse. We also show that the Drosophila cul-1 gene is able to restore the ability of the cdc53 defective yeast mutant to pass through the G₁/S-phase transition, showing these two genes are true orthologues. We also characterize the temporal and spatial patterns of cul-1 gene expression at both the transcript and protein levels during Drosophila development. Moreover, we show increased cul-1 transcription when cell proliferation is inhibited and apoptosis is induced by γ-irradiation. These data suggest Drosophila cul-1 is probably a negative repressor of cell division, as previously observed in nematodes. We conclude that in higher multicellular eukaryotes Cul-1 participates in the regulation of several different phases of the cell cycle.

A Drosophila central nervous system-specific cDNA library in λgt10 was prepared as previously described (Li et al., 1995). A mouse cDNA λgt10 library, prepared from day-8.5 mouse embryos, was kindly provided by Dr S. Agulnik (Princeton University) (Agulnik et al., 1996). The libraries were plated on E. coli strain NM514. The Drosophila library was screened with a 560 bp EcoRI/SalI fragment of cDNA encoding Galleria mellonella neuropeptide sericotropin (Kodrik et al., 1995) to screen for neurospecific genes. This screen identified the Drosophila cul-1 cDNA. The 1.8 kb EcoRI/HindIII fragment from this cDNA was then used to screen the mouse library. The cDNA probes fragments were ³²P-labelled with Random Primed DNA Labeling Kit (USB) and hybridized to 2×10⁵–3×10⁵ plaques transferred to Hybond-N membranes (Amersham). Hybridization was performed in 5× SSPE (0.9 M NaCl, 0.05 M NaH₂PO₄, 5 mM EDTA, pH 8.0), 5× Denhardt’s solution, 0.1 % SDS and 100 µg ml⁻¹ calf thymus DNA at 50 °C for the Drosophila library, and at 55 °C for the mouse library. After a 1 h prehybridization, the membranes were incubated with heat-denatured probes for 12–16 h and then washed twice at the hybridization temperature in 1× SSPE, 0.1 % SDS, and twice in 0.1×SSPE, 0.1 % SDS. Positive clones isolated were sequenced using Sequenase Version 2.0 kit (USB).

Nucleotide sequences of the isolated clones and their open reading frames (ORFs) were searched against the GenBank database using the BLAST program (Altschul et al., 1997). Protein alignments and phylogenetic tree construction were performed with the aid of the DNASTAR software package.

The S. cerevisiae KJB-1 (MATaade2 trp1 ura3-52 mip1::URA3 cdc53-2) strain carrying the temperature-sensitive mutation of cdc53 gene was kindly provided by Dr M. Goebl (Indiana University). To prepare constructs containing the Drosophila cul-1 gene we generated a polymerase chain reaction (PCR) fragment using the cDNA clone as a template and two primers. The upstream primer 5’-CCCGAATTCGCCGCCAGAATGAACCGCTC contained an EcoRI site and the ATG start site (underlined), while the downstream primer 5’-CGCGGGCCCATTTAACAAAAA-CTTAGGC contained the stop codon (underlined) and a SmaI site. The PCR fragment containing the entire Drosophila cul-1 ORF (2.6 kb) was cloned into an EcoRI/SmaI-linearized vector pUC57. To ensure that no PCR-induced mutations were in the final construct a portion (2220 bp) of the DNA obtained from the PCR, between the unique EcoRV and XbaI sites, was exchanged with the corresponding wild-type fragment. The flanking regions were then sequenced to ensure that no mutations were present in these regions. The 2.35 kb EcoRI/SmaI fragment containing the full-length cul-1 gene ORF was then cloned into the corresponding sites of pBDGal4 (Stratagene), producing a hybrid protein, which consisted of the GAL4 binding domain and the Cul-1 protein. Similarly, C-terminal truncated protein derivatives of Cul-1 were prepared using the 2.00 kb EcoRI/SalI and 1.82 kb EcoRI/PstI fragments. All three constructs were used for transformation of the KJB-1 yeast strain by the lithium acetate procedure (Gietz and Woods, 1994).

RNA was extracted from whole insects or cultured cells with TRIzol Reagent (Gibco BRL), electrophoresed on formaldehyde 1 % agarose gels, and transferred onto Hybond-N membranes (Amersham). The 1.8 kb EcoRI/HindIII fragment of Drosophila cul-1 cDNA was radiolabeled and used as a probe. Hybridization was conducted essentially as described above for the cDNA library screenings except that it was performed at 65 °C. Hybridization membranes were analyzed using a Personal Molecular Imager (Biorad) with Quantity One software package.

For chromosomal localization of the cul-1 gene, the digoxygenin (DIG)-labeled 1.8 kb EcoRI/HindIII cDNA fragment was hybridized to salivary gland polytene chromosomes according to the method of Hankeln et al. (1994). The cul-1 gene was mapped using polytene chromosomes of the Oregon R strain and the deletion mutants Df(2R)ca53/+, Df(2R)ca58/+ and Df(2R)cn89e1/+ (Alexandrov and Alexandrova, 1991). Detection of the hybridization signal was performed using anti-DIG-alkaline phosphatase conjugate (Boehringer Mannheim) according to the manufacturer’s protocol.

Localization of cul-1 transcripts during embryogenesis and oogenesis was performed using antisense and sense cul-1 RNAs, which were synthesized by SP6 and T7 RNA polymerase, respectively. DIG-labeled probes were hybridized as described (Mason et al., 1994). Dissected ovaries and dechorinated eggs were fixed in 4 % paraformaldehyde. Hybridization of labeled probes was performed at 65 °C in 50 % formamide for 24–36 h. Washes were carried out at the same temperature for 2 days. Oogenesis and embryo development were staged according to King (1970) and Bownes (1982), respectively.

To prepare antibodies against the Cul-1 Drosophila protein, the 675 bp SacI/SalI fragment of the Drosophila cul-1 cDNA clone was inserted into the pQE30 vector. Isopropyl β-D-thiogalactoside (IPTG) induction of protein expression and Ni²⁺-agarose purification of the fused poly(His)-containing protein were done according to the manufacturer’s protocol (Qiagen). The purified protein was used for antiserum production as described (Filippova et al., 1998). Western blotting was performed as described (Filippova et al., 1998). Immunohistochemistry was conducted using whole mounts of embryos or selected organs from postembryonic developmental stages (Filippova et al., 1998).

Drosophila Kc 167 cells, kindly provided by Dr L. Cherbas (Indiana University), were cultured in M3 medium with 3 % fetal bovine serum at 25 °C. Cells were seeded at 1×10⁶cells ml⁻¹and grown for 24 h. This semiconfluent culture was irradiated with a dose of 28 krad. 2 ml cell culture samples were collected at the indicated time, washed twice with PBS and used in a TUNEL assay (Boehringer Mannheim) or for total RNA extraction using TRIzol Reagent (Gibco BRL). Serial dilutions of total RNA samples were used for RT–PCR analysis of reaper gene expression utilizing an Access PCR kit (Promega). Primers CTACCAGTTGTGTA-ATTCCGAACGAG and TTGGACTGGACTATTGGTTTT-CCCCG produced a reaper-specific fragment 293 bp long. The amount of RNA was also tested using actin-specific primers GTCGACAATGGCTCCGGCATGTGCA and TCAGCGCG-CGTTAGAAGCACTTGCGGT.

Sequence analysis of Drosophila and mouse cul-1 homologues While screening a brain Drosophila cDNA library for neurospecific clones, we isolated a clone with a 2.8 kb insert that hybridized to sericotropin cDNA probe from the moth Galleria mellonella (Kodrik et al., 1995) at low stringency. Sequence analysis revealed the presence of an open reading frame (ORF) 2419 bp long capable of encoding a protein of 773 amino acid residues with a calculated molecular mass of 89.4 kDa. The GenBank accession number of the sequence is L41642. The protein had no signal peptide or transmembrane regions, but contained a potential bipartite nuclear targeting sequence at amino acid residues 704–721 (Dingwall and Laskey, 1991). Both of these features suggested that the cDNA clone encoded a nuclear cytosolic protein. GenBank BLAST analysis using this deduced protein sequence identified several related proteins belonging to the cullin family, a novel family of cell-cycle regulation proteins recently identified in higher eukaryotes (Kipreos et al., 1996). A BLAST search using G. mellonella sericotropin DNA and protein sequences showed only low levels of similarity to cullins, and this homology was restricted to a small region. Therefore, we believe that sericotropin, which belongs to the lipocalin superfamily, is functionally unrelated to the cullin family of proteins.

Screening of an embryonic mouse cDNA library with the Drosophila cul-1 cDNA clone as a probe under mild hybridization conditions led to the isolation of a 3.0 kb cDNA clone, encoding a 776-residue protein, which is highly similar to product of the human cul-1 gene. Its GenBank accession number is AF176910. Phylogenetic analysis confirmed that the Drosophila and mouse gene products are homologous to cullins, with the highest similarity to cul-1 orthologue products of this gene family (Fig. 1A). Fig. 1B displays the alignment of Drosophila and mouse protein sequences designated as dmcul-1 and mmcul-1, respectively, with the human cullin-1 protein (hscul-1) and its yeast orthologue Cdc53. The human and mouse proteins are 98.9 % identical, the major difference being the presence of a 24-residue stretch in the mouse protein, amino acids 59–83, which is absent in the human gene. The Drosophila protein has 59 % similarity to both human and mouse Cul-1 proteins and 24.7 % to the yeast Cdc53. These data suggest that we isolated the Drosophila and mouse homologues of the cul-1 gene.

To confirm the phylogenetic analysis suggesting that Drosophila cul-1 is an orthologue of the yeast cdc53 gene, we complemented the cdc53 yeast conditional mutant with the Drosophila cul-1 gene. The yeast KJB-1 strain has a temperature-sensitive cdc53-2 mutation that causes developmental arrest at 36 °C (Mathias et al., 1996). We transformed the KJB-1 strain using constructs having either the full-length Drosophila cul-1 gene or its truncated forms. Drosophila cul-1 genes that gave C-terminal truncated proteins (Dmcul-1ΔPst, truncated at amino acid 604, or Dmcul-1ΔSal, truncated at amino acid 664) did not change the mutant phenotype at 36 °C. In contrast, full-length Drosophila cul-1 protein complemented the mutation, allowing the transformant to propagate at the nonpermissive temperature (Fig. 2).

The temporal pattern of Drosophila cul-1 gene transcription was determined using northern and western analyses (Fig. 3). Hybridization of total RNA isolated from whole animals at different developmental stages with the Drosophila cul-1 cDNA probe revealed the presence of a single approx. 3 kb mRNA band (Fig. 3A). The expression of cul-1 mRNA was high in 0–4 h embryos, moderate in the older (4–16 h) embryos and very low in the larvae. It rises at the pupal stage and high expression levels were observed in adult females. Western blotting with Cul-1 antibodies also showed the highest concentration of the Cul-1 protein in early embryos. Both protein and transcript levels decrease by the third instar larvae; however, transcript levels show a greater decrease than Cul-1 protein levels (Fig. 3B).

The higher level of cul-1 mRNA observed in adult females when compared to males (Fig. 3A) may be due to its high expression in ovaries. We therefore also monitored the spatial and temporal expression of the cul-1 gene during oogenesis. In situ hybridization using a cul-1 antisense probe showed the presence of the transcript in nurse cells during most of oocyte development. Transcript levels increased progressively as the egg chambers developed and reached maximum levels at ovarian stage 10 (Fig. 4A). At stages 11 and 12, during nurse cell dumping, cul-1 transcripts passed from the nurse cells into the oocytes. A strong reaction was observed in the ooplasm during nurse cell degeneration, but all detectable cul-1transcripts disappeared abruptly just before the end of nurse cell breakdown. No cul-1 mRNA was found in any region of stage-14 oocytes or in mature eggs (Fig. 4A).

cul-1 gene transcription during oogenesis was coupled with its translation. Cul-1 antibodies showed that Cul-1 protein is predominantly localized in nurse cells until stage 10 (Fig. 4B). During nurse cell breakdown the Cul-1 protein, just as the cul-1 mRNA, was transported into oocytes (Fig. 4C). In contrast, however, to the disappearance of the cul-1 gene transcripts at the last stage of oogenesis, mature unfertilized eggs contained high levels of the Cul-1 protein (Fig. 4D).

No cul-1 transcripts were found in embryos at the beginning of zygotic development (Fig. 5A), but a wave of cul-1 gene transcription was observed shortly afterwards. The first cul-1 mRNA was detected at the embryonic stage 3 before syncytial blastoderm formation (Fig. 5B), and transcripts were detected throughout the subsequent cleavage divisions with highest levels observed at stage 4 (Fig. 5C). After cellularization, cul-1 expression declined progressively during gastrulation and reached a constant low level of expression at stage 8 (Fig. 5D–F). Distribution of mRNA was uniform over the entire embryo during all stages of embryogenesis. The Cul-1 protein level was high at all stages of embryo development starting from the very first divisions (Fig. 6A–D). The protein, as with the mRNA, was distributed evenly throughout the embryo during the early stages of development. At the late stages, Cul-1 protein levels gradually decreased, especially in tissues that have low proliferation rates (Fig. 6D).

Various larval and pupal tissues were analysed for the presence of cul-1 transcripts, which were present in the nuclei of all cell types, whether they were dividing or quiescent (Fig. 7A). The cul-1 protein, however, is preferentially expressed in actively proliferating tissue. For example, high levels of cul-1 protein were found in the gonads (Fig. 7B) and the growing imaginal discs (Fig. 7C).

To investigate the influence of cessation of cell proliferation on cul-1 expression in Drosophila we irradiated Kc167 cells with a lethal dose of γ-irradiation. Irradiation of a vigorously growing semiconfluent culture with a 28 krad dose caused, within 24 h, an increase in the number of apoptotic cells from approximately 5 % in the initial culture to more than 50 %. Induction of apoptosis was monitored by the TUNEL assay (data not shown) and expression of the apoptotic gene reaper (McCall and Steller, 1997) (Fig. 8A). The irradiation dose that we applied was higher than the lethal 25 krad dose that is commonly used to prepare feeder layer cells (Echalier, 1997). This dose sharply blocks cell division, stops proliferation and induces apoptosis; however, expression of cul-1 gene, which was monitored by northern analysis of RNA isolated from the irradiated cells, increased twofold at 72 h after irradiation (Fig. 8B).

In situ hybridization was performed using the wild-type Oregon R and the hybrid strains Df(2R)ca53/+; Df(2R)ca58/+ and Df(2R)cn85e1/+, which contain deficiencies in the regions 43E6–44B5-9, 43A3–43F6 and 42E4-F1–43E16-F1, respectively (Alexandrov and Alexandrova, 1991). Results of in situ hybridization revealed that the deficiencies Df(2R)ca53 and Df(2R)ca58 cover the corresponding cul-1 gene, while Df(2R)cn85e1 does not (Fig. 9). Thus, using the cytological boundaries of these deletions, the cul-1 gene maps to position 43F1-6 of the 2R chromosomal arm.

The cul-1 gene in nematodes and its orthologue in yeast, cdc53, belong to the cullin gene family (Kipreos et al., 1996). Since mutations of cul-1 and cdc53 are lethal, the products of these genes must play a vital role and cannot be substituted with other cullins, despite their considerable structural homology. Mutations were shown to affect cell-cycle regulation and cul-1 and cdc53 were postulated to share a biochemical mechanism of specific destruction of the cell-cycle regulatory molecules via ubiquitination (Kipreos et al., 1996; Mathias et al., 1996; Willems et al., 1996). The phenotypic effects of mutations in the cul-1 and cdc53 genes are, however, surprisingly different. While the yeast cdc53 mutant is unable to perform the G₁-to-S-phase transition and cease dividing (Mathias et al., 1996), the nematode cul-1mutant displays excessive divisions in postembryonic tissues resulting in hyperplasia, suggesting that the Cul-1 protein is required for exit from the cell cycle (Kipreos et al., 1996). The diverse cell-cycle effects of the orthologous regulators in yeast and nematodes raise the question of how cul-1 orthologues function in other organisms.

Sequence and phylogenetic analyses of the two genes that we isolated from Drosophila and mouse clearly show they are orthologues of cul-1 and cdc53. The mouse protein sequence is practically identical to human Cul-1, except for an additional block of 24 amino acids present at the N terminus. It is noteworthy that this block is also present in the Drosophila Cul-1 (Fig. 1B). The level of identity between Drosophila Cul-1 and yeast Cdc53 proteins does not exceed 25 %, which is comparable to the level of identity between cullin family members from the same species (Kipreos et al., 1996).

The Drosophila cul-1 gene complements the yeast temperature-sensitive cdc53 mutant phenotype, as was previously demonstrated for the human cul-1 gene (Lyapina et al., 1998). Complementation is possible only with the full-length Drosophila cul-1, and not with its truncated forms. These results prove that the isolated Drosophila gene is a true orthologue of cdc53. Since it can control the transition from late G₁ to S phase in yeast, it might also participate in the regulation of this phase transition in Drosophila; however, we observed high levels of the Cul-1 protein during the first division cycles after fertilization. Since these divisions proceed without the G₁ phase (Edgar and Datar, 1996), Drosophila cul-1 gene might be required not only for the G₁-to-S transition, but also for other cell-cycle phase transitions. The Cul-1 protein is known to participate in the SCF complexes associated with ubiquitin-dependent proteolysis. The maternal zygotic transition in Drosophila embryos also requires degradation of the maternally derived regulators and thereby inactivation of the maternal cell-cycle control. Thus it is possible that the maternally derived Cul-1 protein is needed for destruction of maternally derived cell-cycle control regulators to change cell-cycle kinetics.

Analysis of Cul-1 protein expression in human cell cultures showed that its level is not periodic throughout cell cycle. It was also shown that Cul-1 protein levels remained high even in serum-deprived cells (Lisztwan et al., 1998). It is not clear, however, whether these unchanged Cul-1 protein levels in G₀–G₁ quiescent cells result from continued gene transcription or stability of the Cul-1 protein. To monitor cul-1 gene transcription in cells that exit the cell cycle we analyzed cul-1 mRNA levels in cells that enter apoptosis. During the first 6 h after irradiation we observed a drop in mRNA levels, probably because overall gene expression is sharply inhibited following treatment (Nordstrom et al., 1996). After 24 h of irradiation we found a twofold increase of cul-1 gene transcription in cells undergoing apoptosis. This response was surprising, and contrary to what would be expected for cell-cycle regulatory proteins. The lethal dose of radiation causes the sharp decline in expression of proteins engaged in cell proliferation. The elevation of cul-1 expression in irradiated cells suggests that it is required for exit from cell cycle to apoptosis. That Cul-1 is indispensable for the exit to quiescence was shown in C. elegans (Kipreos et al., 1996).

Studies of the cul-1 gene in nematodes failed to detect transcription in early embryos at the four-cell stage but clearly demonstrated an early zygotic gene expression prior to the 24-cell stage (Seydoux and Fire, 1994). Since a maternal effect was observed in the cul-1 mutants in nematodes, Kipreos et al. (1996) suggested that either the presence of maternally derived transcripts in early embryos was below the threshold level of in situ detection, or the Cul-1 protein was stored in oocytes (Kipreos et al., 1996). We therefore specifically analyzed Drosophila cul-1 gene expression at the mRNA and protein levels during oogenesis and early embryo development. We found high levels of cul-1 mRNA and the Cul-1 protein in developing oocytes. By the end of egg development, however, cul-1 mRNA disappears. It is clear that the maternal effect described for the nematode also occurs in Drosophila and it is executed via the protein product and not mRNA deposition. The maternal Cul-1 is utilized at early embryonic stages but transcription of the zygotic cul-1 gene begins before formation of the syncytial blastoderm (Fig. 5B).

The Cdc53 protein is a part of the multiprotein complex that specifically labels and destroys cell regulators through ubiquitination. The same is true for its Cul-1 homologues in higher eukaryotes. We show here, however, that the Drosophila cul-1 gene product is required for at least two different stages of the cell cycle. This increased complexity of the Cul-1 in multicellular organisms, where stages of quiescence and apoptotic death are parts of programmed development, indicates that the cells of such organisms contain an array of cell-cycle regulators whose breakdown is controlled by Cul-1.

This work was supported in part by a grant from NIH AI32572, the UC Agricultural Experimental Station, and by grant 305/94/0975 from the Grant Agency of the Czech Republic.