Cloning and Sequencing of a Drosophila Cyclin A Homolog
We used a degenerate oligonucleotide probe derived from a stretch of amino acids conserved between clam cyclin A and a sea urchin cyclin (Swenson et al., 1986
; Pines and Hunt, 1987
) to screen a λ phage cDNA library made from poly(A)+
RNA from Drosophila ovaries. Several clones were isolated, and the longest insert was sequenced (). The sequence analysis revealed a long open reading frame. Its putative initiation codon is flanked by a sequence that fits the consensus sequence for translational start sites in Drosophila (Cavener, 1987
). The sequences flanking the five ATG codons found upstream of the putative translational start site do not fit the consensus, and all these ATGs are followed by several stop codons. At the 3′ end of the cDNA, a polyadenylation signal followed by a poly(A) tract was found. As shown below, at least two different maternal transcripts were detected on Northern blots (, lane 1). The predominant maternal transcript is about 300 bases longer than our longest insert. At present, we do not know whether this cDNA represents a full-length clone. Some or all of the observed size difference between cDNA and transcript might be accounted for by the presence of a longer poly(A) tail in the mRNA. The sequence of another cDNA clone revealed differences in the 5′ and 3′ untranslated regions (data not shown), arguing for alternative mRNA processing events. According to restriction mapping and partial sequencing, the coding sequence appears to be identical in all 18 cDNA clones.
Nucleotide Sequence and Predicted Amino Acid Sequence of the Drosophila Cyclin A cDNA
Cyclin A mRNA Levels during Embryonic Cell Cycle Progression
Amino acid sequence comparison between the 491 amino acid protein encoded by the Drosophila cDNA and the clam cyclins A and B (Swenson et al., 1986
; Westendorf et al., 1989
), a sea urchin cyclin (Pines and Hunt, 1987
), and the S. pombe cdc13
+ gene product (Booher and Beach, 1988
) revealed extensive conservation. For example, within a 60 amino acid region (indicated by arrows in ) residues at 82% of the positions are similar and 72% are identical in the aligned Drosophila and clam cyclin A sequences. Within this same region, 47% of the positions are identical among all the cyclins, whether A or B type, or whether of yeast, mollusc, echinoderm, or insect origin (see residues marked with an asterisk). Conservation among this group of proteins extends throughout the carboxy-terminal two-thirds of these proteins (). Based on this homology and on the cell cycle–dependent degradation of the encoded protein (see below), we conclude that the cloned sequence represents a Drosophila cyclin. In contrast to the sea urchin and the S. pombe cyclins, the Drosophila sequence is more similar to clam cyclin A than to clam cyclin B (50% identity as opposed to 34% identity in the region shown; see also boxes in ). The presence of amino acid clusters specifically conserved between Drosophila and clam A cyclins, and clusters specifically conserved among yeast, sea urchin, and clam B cyclins (see for instance , residues marked with black dots), allows a tentative classification of these cyclins as A or B types.
Amino Acid Sequence Comparison of the Drosophila Cyclin A with Other Known Cyclin Proteins
Identification of Mutations in the Drosophila Cyclin A Gene
In situ hybridization to polytene chromosomes localized the cyclin A sequence to the genetically well characterized region 68 D/E on the left arm of chromosome three (data not shown). Hybridizations to a series of chromosomal deficiencies that subdivide this region (Akam et al., 1978
) placed the cyclin A sequence between the proximal breakpoints of the deficiencies vin2 and vin3 (). An essentially saturating genetic screen had uncovered five lethal complementation groups (rsg11–rsg15) within this interval (Hoogwerf et al., 1988
Identification of Mutations in the Cyclin A Gene
Independent work in the laboratory of Y.-N. Jan resulted in the cloning of the genomic region flanking the P element insert present in the fly line neo114. This P element insert is located in the chromosomal region 68D/E and the line neo114 does not complement the allele I(3)183 of the complementation group rsg11 (H. Vaessin and Y.-N. Jan, personal communication). On Southern blots, the cyclin A cDNA probe detected restriction fragments in neo114 fly DNA that were not found in control DNAs (). Additionally, the cyclin A cDNA hybridized to the cloned genomic sequences flanking the P element insert (H. Vaessin, Y.-N. Jan, C. F. Lehner, and P. H. O’Farrell, unpublished data). Southern analysis of the I(3)183 allele shows that this mutation is a small deletion that removes cyclin sequences (data not shown). Moreover, in embryos homozygous for either rsg11 allele, neo114, or I(3)183, no zygotic cyclin A was detected (see below). We conclude that neo114 and I(3)183 are mutant alleles of the cyclin A gene that disrupt the cloned sequence and interfere with expression of the encoded protein. The recessive lethality of these mutations suggests that the cyclin A gene encodes an essential function.
Transcription of the Cyclin A Gene during Wild-Type Development
The relative cyclin A mRNA levels during different stages of embryogenesis were estimated by Northern blot experiments (). During all stages, two bands (around 2.7 kb and 3.0 kb, respectively) were detected by the cyclin A cDNA probe. The lower band was especially prevalent in RNA from 0–1 hr embryos (, lane 1). These embryos are transcriptionally inactive and therefore contain only maternal RNA.
In general, cyclin A mRNA levels appeared to be roughly correlated with the mitotic activity in the embryo. Consistent with the high frequency of clones found in the ovary cDNA library (0.1%), the maternal mRNAs appeared to be quite abundant in 0–1 hr embryos that are engaged in extremely rapid cleavage divisions (, lane 1). Considerably lower levels of cyclin A mRNA were found in total RNA from early cycle 14 embryos (, lane 2), suggesting that much of the maternal mRNA was degraded before gastrulation. The increase in cyclin A mRNAs during cycles 15 and 16 must have represented reaccumulation by zygotic transcription (, lane 3). Only very low levels of mRNA were detected in older embryos, when only cells in the nervous system still divide (, lane 4). The cyclin A gene might therefore be transcriptionally inactive in cells that have completed their limited number of mitotic divisions.
Characterization of Anti–Cyclin A Antibodies
In order to follow the accumulation of the cyclin A protein during development, an affinity purified rabbit antibody against bacterially produced cyclin A was used for immunoblotting and immunofluorescence experiments. A number of observations indicated that the affinity purified antibodies are monospecific for Drosophila cyclin A. Immunofluorescent labeling was reduced to background levels in embryos homozygous for either of the cyclin A mutant alleles, I(3)183 or neo114, at stages after the exhaustion of the maternal cyclin A supplies (see below). In addition, our anti–cyclin A antibodies did not react with either the sea urchin cyclin or the clam cyclin A made in a reticulocyte lysate (A. Murray and C. F. Lehner, unpublished data).
On immunoblots, two bands around 61 kd were detected in total embryo extracts from different developmental stages (). These bands were not detected with a control antibody (data not shown). The lowest band corresponded in size to the product of in vitro translation reactions in which T7 RNA polymerase transcripts of the cyclin A cDNA were used to program a reticulocyte lysate (data not shown). Therefore, it is unlikely that the multiple bands are caused by proteolysis. More likely, they reflect modification of cyclin A. Modification of cyclin proteins has been reported in other organisms (Pines and Hunt, 1987
; Standart et al., 1987
). Interestingly, the ratios of the signal intensities in the different bands varied at different developmental stages (, compare lanes 1 and 2 with lanes 3 and 4), suggesting that cyclin A modification might be developmentally regulated.
Cyclin A Abundance during Embryonic Development
Consistent with the results of Northern blot experiments, considerably lower signals were observed in older embryos, where only cells in the nervous system divide (, lanes 6–8).
Subcellular Location and Degradation of Cyclin A during Mitotic Divisions
After egg activation, a fertilized Drosophila embryo undergoes 13 rapid and synchronous cleavage divisions during which only the nuclei divide (Foe and Alberts, 1983
). When early embryos were stained with affinity purified anti–cyclin A antibodies in indirect immunofluorescence experiments, an almost uniformly distributed, bright signal was observed throughout the syncytial embryo. Surprisingly, only subtle differences in staining were observed between embryos in interphase and those in different mitotic stages (data not shown).
After the 13 cleavage divisions, nuclei become cellularized. In striking contrast to the early cleavage divisions, cyclin A was completely degraded during the subsequent cell divisions. shows cells from an embryo during mitosis 14 after double labeling with anti–cyclin A antibodies and a DNA stain. Cyclin A staining accumulated during interphase and was found exclusively in the cytoplasm (, Interphase, see also ). In contrast, during prophase the most intense cyclin A staining was localized in the region of the condensing chromatin (, Prophase). Metaphase cells were either brightly stained (, Metaphase A), intermediately stained (data not shown), or unstained (, Metaphase B), indicating that cyclin A was completely and rapidly degraded within metaphase. Consistently, no cyclin A was detected in anaphase (, Anaphase), telophase, and early in the following interphase (data not shown, but see ).
Intracellular Distribution of Cyclin A during Cell Division
Accumulation and Degradation of Cyclin A during Embryonic Cell Cycle Progression
Cyclin A and the Temporal and Spatial Program of Embryonic Cell Divisions
The divisions following cellularization (mitoses 14–16) are no longer synchronous: the cells enter mitosis in a spatially and temporally defined, bilaterally symmetrical pattern (Foe and Alberts, 1983
; Hartenstein and Campos-Ortega, 1985
; Foe, 1988). Anti–cyclin A labeling permitted dramatic visualization of this spatio-temporal program of embyronic cell divisions. The first cells to enter the 14th mitosis were found in a cluster in the head region of the embryo. Shortly thereafter, cells in another cluster in the head region underwent mitosis 14. In the embryo shown in , most of the cells in these first two clusters had already passed metaphase and therefore cyclin A was no longer detectable. In addition, divisions in another domain were anticipated by cyclin A staining in nuclei, a distribution characteristic of early mitotic stages (, arrows). In a slightly older embryo (), the cells in this domain, just anterior to the cephalic furrow, had completed mitosis and were therefore no longer stained. With time, existing domains expanded as more cells divided and additional domains appeared in the head and lateral region (, and ).
Cyclin A staining persisted in cells of the amnioserosa (, region A) and in the neurogenic region (, region N). The cells in the neurogenic region eventually enter mitosis 14, some of them synchronously in a segmentally repeated pattern (). Cell cycle 14 in the neurogenic region can be up to three times longer than in the head where cells enter mitosis 14 first. In contrast, amnioserosa cells never go into mitosis 14. Rather than disappear abruptly, the cyclin A in these cells appeared to be slowly degraded. Except for the amnioserosa, the patterned loss of cyclin A labeling was strictly correlated with the pattern of mitotic divisions revealed by double labeling using a DNA stain (data not shown).
In contrast, to degradation, the accumulation of cyclin A did not reflect the pattern of mitosis 14. Cyclin A accumulated very uniformly in all cells throughout the embryo (). The rate of accumulation appeared to be identical in cells that complete mitosis 14 very early (head region), very late (neurogenic region), or never (amnioserosa).
Cyclin A cycled during all embryonic cell divisions after cellularization. Following degradation during mitosis 14, cyclin A reaccumulated during interphase 15. This can be seen in , where the cells in the head and the lateral regions of the embryo were again labeled. In cycle 15, the cells never stained as intensely as those cells that enter mitosis 14 very late (). In fact, the first cells to divide in the thoracic region of the embryo shown in (arrow) had already passed metaphase 15 and were therefore no longer labeled. Cyclin A staining in a slightly older embryo demonstrated that not only mitosis 14 but also mitosis 15 is highly patterned (). Note that the segmentally repeated pattern of the 15th division in the lateral region of the embryo () was not predictable from the pattern of reaccumulation (). Thus, like mitosis 14, the timing of mitosis 15 is not correlated with the rate of cyclin accumulation.
Cyclin A Distribution and Nuclear Density in Mutant Embryos
A similar accumulation and degradation accompanied cycle 16 (data not shown). Subsequently, mitotic divisions in the embryo stop except for cells in the central and peripheral nervous system and cells in the germ cell lineage. At these stages, cyclin A staining was detected exclusively in these cells (). Cyclin A expression was not restricted to embryonic cells; it was detected in imaginal disc cells and by immunoprecipitation in Schneider G2 tissue culture cells (data not shown). Thus, cyclin A appears to be expressed in all dividing cells, but not in nondividing cells (except for the nondividing aminoserosa cells where cyclin A is initially present but later disappears).
Cyclin A Accumulation and Cell Divisions in Mutant Embryos
It is clear that early embryos in Drosophila and other species have large stores of maternal cyclin mRNA. The amount of cyclin A produced from this maternal mRNA can be estimated by immunofluorescent labeling of embryos that are unable to produce cyclin A zygotically. The consequence of a zygotic defect can be examined after exhaustion of the maternal contribution. We analyzed mutant embyros homozygous for either neo114 or I(3)183 or carrying these alleles over the deficiency vin3. No clear difference between wild-type and mutant embryos was detectable during the early cleavage stages (data not shown). In older mutant embryos, no anti–cyclin A staining was detected (data not shown, but see below), suggesting that the maternal cyclin A mRNA pool was exhausted.
We analyzed cyclin A expression and the cell division program by double labeling timed embryo collections with anti–cyclin A antibodies and a DNA stain. The mutant embryos proceeded normally through the first 13 divisions. In the cell cycle preceding mitosis 14, they never accumulated cyclin A to the normal level (compare ). Nevertheless, the pattern of mitosis 14 was normal, as visualized by DNA staining (data not shown) and by the spatial pattern of cyclin A (compare ). Comparing normal and mutant embyros of the same age demonstrated that not only the pattern but also the timing of mitosis 14 was unaffected in mutant embryos despite the lower levels of cyclin A ().
After mitosis 14, cyclin A accumulated only to barely detectable levels in mutant embryos (compare ). Nevertheless, at least some cells went through division 15 (data not shown). During the stage when embyros would normally be engaged in mitosis 16, no cyclin A was detected in mutant embryos. Interestingly, no mitotic figures were detected in these mutant embryos, clearly indicating that cell divisions were blocked in the absence of cyclin A. Consistently, at stages when the ectodermal division program was completed, the nuclear density revealed by labeling of nuclear envelopes was clearly lower in mutant embryos (). The number of nuclei in the ectoderm of mutant embryos was slightly less than half of the number of normal embryos.