Accurate development of metazoan embryos requires precise production, reception, and interpretation of patterning cues along appropriate spatial axes on realistic timescales. Many embryonic patterning events utilize graded spatial distributions of patterning molecules, or morphogens, whose activities rely fundamentally on molecular interactions susceptible to environmental fluctuations and stochasticity in gene expression 
. Despite the widespread occurrence of morphogen-mediated tissue patterning, in most cases little or no quantitative data exist regarding the dynamics of gradient establishment or the spatio-temporal regulation of morphogen production.
The wealth of molecular and genetic tools available in Drosophila melanogaster
offers an optimal context in which to study the basis of developmental accuracy in embryonic patterning by morphogens. In the Drosophila
embryo, anterior-posterior (AP) axial patterning originates with maternal cues deposited into the developing egg 
. Among these cues is the transcription factor Bicoid (Bcd), the mRNA of which localizes at the anterior cortex of the oocyte 
. Translation of bcd
mRNA is believed to commence upon fertilization, after which the embryo undergoes 13 rapid nuclear mitotic cycles (n.c.) without cytokinesis. By the start of interphase 14 about 2 h after egg deposition (AED), the embryo consists of a syncytial blastoderm layer of about 6,000 nuclei at the cortical surface, surrounding the interior core of yolk and vitellogenic nuclei. During the blastoderm stage Bcd protein distributes along the AP axis as an exponentially decaying gradient 
, and nuclear Bcd activates target genes in a dosage-dependent manner (reviewed in 
Recently, quantitative analysis of living embryos expressing Bcd-GFP revealed that the protein gradient remains nearly unchanged subsequent to the arrival of nuclei to the cortex at n.c. 10 (about 85 min AED), and therefore that the nuclear gradient achieves stability within only about 80 min at 25°C 
. Moreover, nuclear gradients at n.c. 14 exhibit remarkable reproducibility between embryos: along the AP axis, similarly positioned nuclei in different embryos contain Bcd-GFP concentrations that differ by only about 10% 
. In principle, this level of precision would be sufficient for nuclei to distinguish their AP positions with an error of only a single nuclear diameter 
. These observations raise fundamental questions regarding the underlying cell-biologic mechanisms responsible for the rapid, precise establishment of nearly equivalent gradients in essentially every embryo.
One such question concerns the dependence of protein gradient formation on the strength, localization, and dynamics of the underlying bcd
mRNA. Females carrying altered genetic dosages of bcd
produce gradients of altered amplitude, resulting in mispatterned embryos 
. Moreover, the packaging of bcd
mRNAs into discrete ribonucleoprotein (RNP) complexes, and their subsequent localization to the anterior oocyte cortex, requires a suite of maternal factors (reviewed in 
). Ooctyes lacking these factors produce embryos with distorted protein gradients resulting from bcd
mRNA translation at inappropriately posterior locations 
. The reliable formation of the Bcd protein gradient must depend, at least in part, on the spatial distribution and the number of bcd
mRNA molecules. While prior work has utilized in situ hybridization to document the spatial distribution of bcd
mRNA during embryogenesis 
, no work has yet determined bcd
mRNA particle numbers or examined their global localization in the developing embryo.
A second question regards the biophysical processes affecting the Bcd protein, such as its transport and degradation (reviewed in 
). The observed exponentially decaying nuclear steady state Bcd protein distribution is consistent with analytical models of gradient formation via constant anterior protein synthesis, coupled with diffusion and uniform degradation throughout the embryo (the SDD model). Within this model, the effective diffusion constant of the Bcd protein D and the protein lifetime τ determine the dynamics of gradient formation. A relatively short τ would allow the gradient to reach an equilibrium distribution (i.e. when protein degradation is matched by new synthesis) within the available developmental time of ~2 h. Conversely, a relatively long protein lifetime will prohibit the achievement of such an equilibrated state within that time. Likewise, larger or smaller values of D will increase or decrease, respectively, the predicted distance a protein can move away from its source mRNA. A direct measurement of D for cytoplasmic Bcd-GFP at n.c. 13 
yielded a substantially slower diffusion constant than for comparably sized, biologically inert molecules 
, and about an order of magnitude too low to account for the rapid achievement of a steady state gradient with the appropriate length constant. More recent work has suggested the presence of multiple populations of Bcd-GFP, where a fraction of Bcd-GFP may diffuse rapidly enough to establish stability in the time available 
Slow diffusivity, however, would be consistent with an alternate model in which the protein gradient arises from graded bcd
mRNA distribution 
. Low values of D coupled with rapid protein degradation would minimize protein movement away from source molecules, so that the local mRNA amount would predominately determine protein concentration along the AP axis. Under these conditions, sufficiently mobile mRNA would greatly impact the dynamics of gradient formation, and protein diffusion contributes only minimally. However, at present it is difficult to evaluate the legitimacy of these models, or to determine what values of D and τ, if any, might accurately describe the dynamics of Bcd gradient formation, because no quantitative data exist which span the first 80–90 min of development.
To address these questions, we developed two novel quantitative measurement approaches. First, we adapted a method of fluorescent in situ hybridization to the Drosophila
embryo employing fluorescently labeled DNA oligonucleotides 
. This method allowed us to identify individual bcd
mRNA particles and determine their positions and intensities. We found that bcd
mRNA particles undergo movement into the core of the embryo, away from their initial site of localization at the ooctye cortex, by the end of the third cleavage division (within 30 min AED). Subsequently, bcd
mRNA is relocalized to the egg cortex during cortical nuclear migration (mitoses 6 to 9, 55 to 75 min AED), during which bcd
particles tend to dissociate without bcd
mRNA degradation. Despite the dynamic behavior of bcd
particles, at all times >90% of mRNA is localized within the anterior 20% of the embryo, a distribution which is insufficiently extended to produce the observed protein gradient in the absence of large-scale protein redistribution along the anterior-posterior axis.
Second, we measured nuclear Bcd-GFP levels during presyncytial stages using GFP fluorescence in fixed embryos. We discovered that Bcd-GFP does not begin accumulating in nuclei until interphase 6, about 45 min AED, reaching peak levels 50 min later in n.c. 11 and 12. Unexpectedly, in fixed samples we found that the nuclear gradient declines between n.c. 13 and n.c. 14, a result which differs from prior analysis of living embryos. We present evidence that attributes this difference in part to delayed fluorescence maturation of GFP in living embryos.
Finally, to examine whether the bcd mRNA distribution impacts protein gradient formation, we incorporated its observed spatio-temporal dynamics into numerical simulations of protein production and movement. We found that models utilizing the actual mRNA distribution result in improved predictions of protein gradient dynamics, compared to models employing a single point source at the anterior pole. Therefore, we conclude that although the spatially extended mRNA localization contributes to the protein gradient, mRNA localization alone cannot account for protein gradient dynamics. Our results demonstrate that protein movement, whether active or passive, is a necessary component of Bcd protein gradient formation.