Embryos initiate development with a dowry of maternal transcripts, proteins and other constituents, but these stores are limited and zygotes must transition to self-sustaining production. Whereas the maternal-to-zygotic transition is universal, its timing is thought to vary in different species. For example, transcription in the mouse embryo can be detected before the first cleavage division, and at the two-cell stage, most maternal transcripts have been degraded and activation of the zygotic genome is described as “major” 
. In contrast, the Drosophila embryo has been considered to be transcriptionally silent prior to blastoderm 
, and the provisions that the egg contains have been thought to be sufficient to supply the embryo's first thirteen mitotic cycles. Moreover, studies of fly embryos that lack specific chromosome arms or entire chromosomes concluded that development is controlled entirely from maternal stores up to the mid-blastula stage 
. The present work suggests that the fly embryo may be more like the mammalian embryo than previously thought.
The statement that an embryo is transcriptionally silent is an operational conclusion based on failure to detect transcripts, and is, of course, a negative result. As methods with greater sensitivity become available, this conclusion either becomes more definitive if further experiments verify prior observations, or it must be qualified or contradicted if transcripts are detected. Similarly, assessments of mutant phenotypes depend on the resolution with which the normal can be described, and because the early Drosophila embryo has few distinctive features, our ability to distinguish or evaluate its complexities is limited. But as tools and techniques improve and the embryo's molecular and structural details are better understood, we can recognize perturbations to normal processes that could not previously be resolved.
The results presented in this paper provide evidence for functional pre-blastoderm gene expression in Drosophila. Deep sequencing of RNA isolated from pre-blastoderm embryos identified a cohort of transcriptionally active genes, and Q-PCR analysis confirmed these results for all four genes in the cohort that were tested with this method (h, odd, Kr and eve). Q-PCR was the more sensitive measure, and in contrast to RNA-Seq, it detected transcripts for these genes in embryos prior to cycle 7 (≤32 nuclei). The sensitivity of the RNA-Seq analysis was limited by the number of reads and by the fact that only those sequences that contain a polymorphism that discriminates between the maternal and paternal alleles were informative. Therefore, although the Q-PCR analysis validated the results from RNA-Seq, the cohort must be regarded as a partial list and a more complete one will require the use of more sensitive methods. The RNA-Seq cohort did not, for example, include en, although the evidence we obtained for zygotic production of both en transcripts and En protein is compelling. We suggest therefore that the abundance of en transcripts was below the level of detection by RNA-Seq in our experiments, a possibility that is consistent with the Q-PCR results, which also indicated that en transcripts are present at lower levels than h, odd, Kr or eve transcripts ().
The model of a silent pre-blastoderm genome is based in part on direct measures of transcription 
and on genetic studies (reviewed in 
), and it seemed plausible given the rapidity of the early nuclear cycles. The interphase periods of cycles 2–9 are less than ten minutes, deemed too short for mRNA to be synthesized, processed, and translated. The finding that the zygotic genome is active prior to blastoderm therefore leads us to ask how genes can be expressed under these time constraints. We can, perhaps, look to DNA synthesis in the pre-blastoderm for a conceptual precedent. Because the length of the S phase of the early nuclear cycles is approximately one-fifth that of later cell cycles, replication in the early embryo has distinctive features (an increased number of replication orgins) that reduce the time required to replicate the genome. Although this specific mechanism is not likely to be relevant to transcription or translation, the point is that the early embryo may modify the processes of gene expression for speed and high throughput.
One key attribute of the transcripts that are expressed in the early embryo may be their small size. For example, although the en
gene extends over more than 70 kb 
, its transcript is only 4207 nucleotides. In contrast, the transcript of its homolog invected
, which is expressed after cellular blastoderm, is more than 32 kb. Transcripts of the h, odd, Kr
genes are also relatively small - 3481, 2527, 2920 and 1539 nucleotides, respectively. In addition De Renzis et al (2007) previously noted that 70% of the genes that are expressed during the first blastoderm cycles have no introns, in contrast to the estimated 20% representation of intronless genes in the Drosophila genome. Similar observations have been made in the mosquito Aedes aegypti
. Many of the genes in the cohort of pre-blastoderm genes that we identified are also intronless. Small transcript size is likely to be a pre-requisite for all genes that are expressed in fast-cycling nuclei.
A feature of Drosophila embryogenesis that may be important to early gene expression is the duration of cycle 1. Fertilization takes place in the reproductive tract of the Drosophila female, and because the time to egg-laying varies, we do not have an accurate measure of the length of the first cycle. Nevertheless, based on the proportion of cycle 1 embryos in populations of eggs that are collected at short intervals, cycle 1 can be roughly estimated to be approximately 20–30 minutes – 2–3 times longer than the ensuing cycles. The fact that the male and female pronuclei remain separate during cycle 1 and do not join to form diploid zygotic nuclei prior to the cycle 1–2 division is a further complication, but our data showing that En:GFP encoded by the paternal genome is present in cycle 2 nuclei and that en mutant embryos are abnormal at the cycle 2–3 division shows that embryos are competent for gene expression at these early stages.
The antibody stainings that detected En protein in cycle 2 nuclei do not allow us to calculate the amount of En protein that the embryo has made. We can nevertheless estimate minimum times needed to make an en
transcript at the elongation rate that has been measured for Drosophila Hsp70
(at 1.5 kb/min; 
) or to make an En protein (at 540 amino acids/min): 2.8 min and 1 min, respectively. Although productive expression is conceivable given the length of cycle 1–2 and the interphases of the later cycles, it is possible (perhaps likely) that expression in the pre-blastoderm is more efficient than these numbers suggest. For example, RNA polymerase II elongation rates similar to those in human tissue culture cells (4.3 kb/min; 
) would reduce the time to make an en
transcript to less than one minute, and early embryos may tailor the protein synthesis machinery for high production levels (perhaps by increasing the load rate and density of ribosomes on mRNAs). Also, because histones and mitotic cyclins are provided maternally and are not among the genes that are expressed in the pre-blastoderm, it is possible that mRNAs from the cohort of early genes are translated throughout the cell cycle without interruption.
The phenotypes we identified in en and eve mutant embryos are perplexing given the presumptive roles of En and Eve as transcription factors. en mutants do not distribute nuclei normally, they do not maintain mitotic synchrony and their pole cells are not symmetrical at the embryo midline. These are curious phenotypes for several reasons. First, a number of genes have been identified that are required to break left/right symmetry, but en may be the only gene known that is required to establish symmetry. Second, although both the pole cell asymmetry and the abnormal behavior of the mutant nuclei are suggestive of defects to the cytoskeleton, En appears to have a strictly nuclear localization in the pre-blastoderm. Our working assumption therefore is that En functions as a transcription regulator in the pre-cellular embryo and that the cytoskeletal defects in the mutants are an indirect consequence of its loss-of-function. It is possible that En functions as a repressor to silence transcription generally and that the mitotic asynchrony is a consequence of aberrant transcripts that interfere with some aspect that is required for the rapid cell cycle; alternatively, En may regulate expression of specific targets whose mis-regulation leads to the mitotic abnormalities. Future experiments that monitor transcripts in en mutant embryos may reveal how En functions at these early stages.
The stereotypic and precisely synchronized nuclear divisions during the first hour of Drosophila development are the first recognized occurrence of pattern formation in the embryo. Although this period is also a time of complex cytoplasmic re-organization and metabolic activity, the opacity of the embryo has effectively obscured the interior of live embryos from view, and the few markers that have been available have limited studies to characterizations of nuclear behaviors and the cytoskeletal structures that orchestrate them. It is likely that these technical impediments have contributed to the failures of genetic screens to identify mutants that affect the patterned activities of the early embryo, but our finding that en expression is required in the second division indicates that the period of pre-cellular development can be analyzed with genetic tools. Our work shows that the fly embryo is not entirely pre-programmed, and although the phenotypes we found appear to be subtle, our functional studies have sampled only a small fraction of the zygotically active genes and therefore do not yet reveal the full extent to which control of early development is under zygotic control.