3.1: The Curious Case of Bicoid: Models for Making a Gradient-A Model Emerges

Models for Making a Gradient

Morphogenic gradients are incredibly important in development for patterning tissues. If a tissue is completely homogenous (every cell doing the same task) it cannot be complex and is often not resistant to environmental fluctuations. In the case of Bicoid in Drosophila, we are considering a morphogenic gradient that patterns the anteroposterior axis - as mentioned in the Introduction, too much Bicoid gives us a giant head and too little gives us no head at all. There are somewhere around 60 different genes that respond to various levels of Bicoid protein and one of the earliest to be discovered is Hunchback. Hunchback gives positional information to a fly embryo, it sharpens the boundary between the anterior and posterior of the fly. While the Bicoid protein gradient is very broad and resembles a diffusion gradient, the Hunchback gradient is sharp, with significantly different protein levels within 10 μm, or about 2 nuclei¹.

Figure 1: The Bicoid (Bcd) gradient is shallow, Bicoid concentration (y-axis) degrades over about 1/2 of total embryo length (x-axis). However, this shallow gradient provides positional information for Hunchback (Hb), which is translated as long as there are even low levels of Bicoid protein1,2. This sharp Hunchback gradient neatly splits the embryo into anterior and posterior halves. Later in development the presence of high Hb + high Bcd, high Hb, low Bcd, and low Hb + low Bcd will be read by different transcription factors to produce broad stripes in the embryo (see Regionalization and Organizers for more information).

Obviously the levels of Bicoid must matter tremendously since the Bicoid gradient and its readout (for example Hunchback levels) are incredibly precise. An early model for the observed Bicoid gradient in Drosophila is the SSD model (Synthesis, Solute, Diffusion), where a point source of Bicoid protein (translated from anterior bicoid mRNA) sets up the concentration gradient. The related SDD model (Synthesis, Diffusion, Degradation) incorporates degradation over time of Bicoid protein, keeping the gradient from reaching the posterior end of the embryo. Several other models have been proposed that would allow for fluctuation in embryo size but still lead to a correct Bicoid readout:

1. More mRNA/protein in larger embryos: This would extend the gradient out further in larger embryos.
2. Nuclear shuttling: The same as the SDD model, but Bicoid protein gets taken up by nuclei as it diffuses by them. Under this model, the more densely packed the nuclei are (smaller embryo) the more quickly the Bicoid protein gets pulled out of the syncytial cytoplasm. In larger embryos, the protein can diffuse farther.
3. ARTS model (Active mRNA, Transport, Synthesis): The cytoskeleton actively transports Bicoid mRNA to form an mRNA gradient that gets readout as a protein gradient as it is translated. In larger embryos, the cytoskeleton can transport the mRNA further.
4. mRNA diffusion and degradation: This is similar to the ARTS model, except that in this model Bicoid mRNA degrades over time to maintain an mRNA gradient. Larger flies could have lower degradation rates of Bicoid mRNA or faster transport.
5. Facilitated diffusion of protein. This is similar to the ARTS model, only it involves transport of Bicoid protein by the cytoskeleton. In this case, larger flies could have faster transport of Bicoid protein.

There is some evidence for and against each of these models, suggesting that the Bicoid protein gradient is likely maintained by a combination of these models. Additionally, Bicoid protein readout might be more complex than we think - multiple proteins could be affecting how Bicoid interacts with its targets.

Evidence for and against the models

Data from both Drosophila and other, larger, dipterans have provided evidence supporting each of the models listed above. Additional testing has falsified the predictions of some of the models, and that is the subject of this section. Below I describe each model and a test of that model, I also include a figure that illustrates what the model expects and what was actually observed. I recommend this type of super-reductionist figure to help summarize a large set of data, but we always need to keep in mind that it does not tell the whole story.

1. More protein model. An embryo can get more protein by increasing the amount of mRNA, increasing the rate of translation of that mRNA, or decreasing the rate of protein degradation. Supporting this is the finding that putting in extra copies of Bicoid (more mRNA and more protein) into a fruit fly increases the Bicoid gradient and the size of the head. To see if this is what flies with larger embryos do to make a larger gradient, researchers put Bicoid DNA, complete with cis-regulatory sites and UTRs, from large embryo flies (Calliphora) in Drosophila. They found that the Bicoid gene from large embryos did not make a longer gradient in Drosophila. Therefore, if there are any factors affecting the mRNA stability or protein degradation, they must be specific to the larger flies³.

2. Nuclear shuttling. Early experiments on this suggested that adding a nuclear localization signal to Bicoid protein made a shallower gradient³. Later experiments tested this by making Drosophila with a mutant version of Bicoid that did not accumulate in nuclei. If nuclear import made the Bicoid protein gradient steeper, then they would have seen a broader gradient after inhibiting nuclear import. However, they saw a fairly normal looking Bicoid protein gradient, arguing strongly against this model⁴.

3 and 4. mRNA gradient models. In both models 3 and 4, mRNA is actively transported around the cell by the cytoskeleton. This is in direct opposition to the SDD model, where Bicoid protein is translated from anteriorly-tethered Bicoid mRNA. In these two mRNA gradient models, the protein gradient is simply a "read-out" of the mRNA gradient. Evidence for these two models comes from careful measurement of Bicoid mRNA in developing embryos (Figure 2). These careful measurements show that Bicoid mRNA itself forms a gradient, and is not simply a point source⁵. Its movement is dependent on the cytoskeleton, if the cytoskeleton is disrupted, the mRNA gradient is also disrupted⁶. Evidence against this model comes from the finding that Bicoid protein at the posterior of the embryo is older than Bicoid protein at the anterior. If Bicoid mRNA diffuses towards the posterior to make a gradient, then Bicoid protein is made from that mRNA, we would expect to see younger Bicoid protein at the posterior and mixed age Bicoid protein at the anterior end. However, Durrieu et al found the opposite, suggesting that even though Bicoid mRNA may form a gradient, this is not the main determinant of the protein gradient⁷.

5. Facilitated protein diffusion. Disruption of the cytoskeleton by treating a Drosophila embryo with anti-actin or anti-microtubule drugs leads to disruption of the protein gradient. Since mRNA tethering at the Anterior pole is also dependent on microtubules, just disrupting the cytoskeleton wouldn't tell you whether it is mRNA transport or protein transport that leads to the protein gradient. One study used a funny trick - if you raise Drosophila embryos in low-oxygen (hypoxic) conditions, the Bicoid mRNA will stay at the anterior end even if microtubules are disrupted. This study found that Bicoid protein still forms a gradient even if Bicoid mRNA does not, and that this gradient is dependent on the cytoskeleton⁸.

Figure 2: Evidence for and against Bcd gradient models^3,4,5,6,7,8. Four major models of Bicoid gradient generation are shown. An experiment that supports or falsifies each is circled in yellow. Expected Bicoid concentration gradients after experimental intervention are on the left, observed concentration gradients are on the right.

A Model Emerges

Despite all of the seemingly contradictory data, a model is beginning to emerge that both explains the robustness of the Bicoid protein gradient and explains how the gradient itself can scale to larger or smaller bodies. First, Bicoid mRNA is mostly sequestered at the anterior end of the embryo, but also travels along the cortex (outer edge) of the embryo via microtubules. Bicoid protein is translated off of the Bicoid mRNA, with a higher amount of translation happening at the anterior end, where the concentration of Bicoid mRNA is higher. Some Bicoid protein is likely also translated off of the lower concentration Bicoid mRNA that has moved posteriorly via microtubules. This leads to a gradient of Bicoid protein that is slightly broader than what we would expect if there was a simple point source of Bicoid protein (i.e. translation from anterior pole Bicoid mRNA).

This protein gradient is also modified as Bicoid protein gets transported around the embryo by the cytoskeleton (via actin filaments and microtubules). In this way, there is control over the gradient by the cytoskeleton and by the proteins that mediate interaction between Bicoid (protein and mRNA) and the cytoskeleton. These mediating proteins may be sensitive to local Bicoid levels and may allow Bicoid to move more quickly or slowly depending on the anteroposterior position and the gradient level. For example, a large fly might have less Bicoid protein by diffusion at 15% of its length (EL) than a small fly does at 15% of its length. However, all of the Bicoid protein at 15% EL in the large fly can get taken up by the mediating proteins and moved posteriorly to broaden the gradient. On the other hand in a small fly, the larger amount of Bicoid protein at 15% EL could oversaturate the mediating protein and be transported less efficiently.

Additionally, Bicoid proteins from different species of fly are known to have slightly different properties, for example Bicoid from larger fly species such as Calliphora vicinia and Lucilia sericata are not able to completely rescue Drosophila Bicoid mutants³. These different properties could also play a role in protein-protein interactions involving the cytoskeleton. Finally, the mediating proteins between the species could also differ, some might bind to Bicoid more efficiently and/or move along the cytoskeleton more efficiently.