Saturday, June 29, 2019

The Subtle Mechanics of Regulatory Enhancers

Synergy is an overused buzzword elsewhere. But in biology, it has real meaning.

Our genes are mostly off, yet each cell needs some select portion of the genome turned on to do its job. Gene expression is an enormous field of study, encompassing the full life cycles of both mRNA and proteins, and much else. But turning the transcription of a gene on is by far the most powerful and typical way to control its expression. This is done by regulatory proteins that bind the DNA near the gene, either nearby at segments called promoters, (which is where the RNA polymerase complex is assembled), or far away at modular segments called enhancers. Proteins bound at enhancers are thought to loop around by way of the flexibility of DNA to touch the proteins bound at the promoters, forming a somewhat disordered protein mush that is more active (firing off more RNA polymerases) the bigger it is, and the more of its components are activating vs repressing.

A very general flow of early drosophila embryonic development, from egg to gastrula and from rough-out to finer position specification.

The fly model system has been a wonderful place to study enhancers, since it has a lot of them, they are clearly modular for different developmental tasks, and they frequently have critical and very visible influences on the fate of tissues in the adult. A recent paper used one developmental gene, hunchback (hb) to study in detail how its expression pattern is developed from the proteins bound to its enhancers.

Experimentally labeled micrographs of hunchback expression (in mRNA form, bottom) as driven by bicoid protein (middle), expressed from bicoid mRNA messages (top). How does the sharp midline cutoff of hunchback expression develop from the hazy gradient of bicoid protein, its primary transcriptional activator? We can ignore the odd posterior hunchback band, which is driven by some non-bicoid inputs. At this early stage, the drosophila embryo is a bag with cells mostly on the surface, and has not yet begun the body segmentation process, though glimmers are beginning on the molecular level, as shown here.

Hunchback was named for the shape of the larva of the fly when this gene is mutated. They are missing most of their thoracic segments. Hunchback is a very early gene in the lengthy and complex chain of events that specify the developmental pattern of the fly. It encodes a DNA-binding regulator that acts on the next set of developmental genes, the gap genes. It is expressed in a simple pattern of - on in the front/anterior, and off in the posterior, of the early embryo. A good deal of it is supplied by the mother in the egg, so it has been difficult to tease apart the effects of that inherited pool of mRNA, vs those of new gene expression within the embryo. But this paper studies only the embryo-based expression pattern, specifically using broken up parts of the enhancers to figure out how they generate a very sharp half-embryo-on / half-embryo-off pattern.

Some upstream portions of the hunchback gene from drosophila. Modular enhancer cassettes lie upstream, and the one studied in the current paper is the P2 enhancer, right next to the coding area of the gene, which for these experiments has been excised and put upstream of an easily assayed reporter gene, LacZ. 

The researchers study one of the three hunchback enhancers- the one that drives this early anterior-only expression (P2, above). The regulator that binds to this enhancer (at 6 distinct sites) is bicoid, a fully maternal factor supplied in the anterior part of the egg. (Mutants of bicoid have no head- the problems of mutants get more dramatic the earlier you go into the developmental cascade.) Bicoid has a very gradually tapering / diffusing distribution, from high in the anterior to low in the posterior. The six sites that bind bicoid in this main hunchback enhancer are known to have cooperative effects, and thus could account for the (non-linear) sharpness of the hunchback expression pattern- high in the anterior, then dropping off sharply at the midline. In this way a much more gradual gradient of bicoid is recomputed into a finer dividing line between front and back. This is a dynamic that is employed over and over again as finer divisions are made throughout development. Yet the authors maintain that this DNA site binding cooperativity is, on its own, not quantitatively sufficient to account for the pattern, and go in search of other explanations for how the bicoid and other factors drive this high-definition pattern.

As bicoid binding sites are removed from the hunchback P2 enhancer (left), the expression of the test gene becomes shallower in its anterior-posterior gradient and migrates towards the anterior.

Later on they cop to the fact that their system is more complex than they portrayed it at first. The enhancer they are working with (isolated from the rest of the enhancers and driving a fluorescent reporter gene) may have a few binding sites for some other factors, including krüppel, tailless, zelda, and indeed hunchback itself, forming a small positive feedback loop. When they scrubbed out those extra sites, expression was quite a bit farther from the wild-type condition, pushed towards the anterior (below). As extra demonstrations, they individually knock out hunchback protein expression, which clearly accounts for some of this effect, pushing expression of the wild-type enhancer farther anterior. And they do a similar demonstration for zelda, which has a similar, though much smaller, effect.

Some further experiments with a purified enhancer, where all non-bicoid sites have been removed (top, red). It is clear that the other sites (present in the wild-type P2 enhancer, black) have a key role driving expression to a more posterior position, but have little roll in the steepness of the cutoff of hunchback expression. Specific regulators are knocked out of the embryo and assayed on the wild-type P2 enhancer in the lower panels, hunchback and zelda, respectively, to show their individual effects.

Next, in search of the additional cooperativity factors, they make deficiencies in several of the common transcriptional components that occupy, not the enhancer, but the promoter where the RNA polymerase is going to be assembled. This is a bit tricky, since these will have quite general deleterious effects, complicating interpretation. But one example is shown below- CBP (Creb Binding Protein). This is an enzyme that modifies histones and is a common part of the core transcription complex, and knocking down its activity shifts the hunchback expression curve not only anteriorly, but also to a more shallow profile, indicating that this protein helps the cooperative effects of multiple bicoid activator proteins bound at the enhancer to take effect. This leads to a clear model of the system whereby the bicoid proteins are only partially cooperative among themselves as they bind to DNA. But their cooperativity is enhanced by all of them binding in concert to their targets in the core transcription complex, one of which may be this CBP protein, or others attached to it (below).

A sample experiment with the full enhancer, in a fly where the expression of one of the core transcription components, in this case CBP, has been knocked down. Not only is expression reduced in the sense of moving anteriorly, but the slope of the expression vs the gradient of bicoid is also reduced, indicating that this defect reduces the effective cooperativity of the six bicoid proteins bound at this enhancer. 

Surprisingly, this work mostly recapitulates work done twenty years ago in the same system. That earlier work had demonstrated that bicoid binds to its DNA sites in a cooperative fashion, specifically in directionally-specific pairs clearly suggesting a single side-by-side cooperative interface on the protein. It also showed that interactions with the core transcriptional apparatus, in that case TAFii60 and TAFii110 in yeast, could account for remaining amounts of cooperativity among the six or more bicoid binding sites on the hunchback enhancer. Thus all this is hardly news, whatever the new mathematical machinery brought to bear by Park et al. in the current paper (and this from Harvard, no less). By this point, we would be expecting to see a full structural reconsitution and recapitulation of the transcriptional activation system, with titrations to demonstrate its accuracy with respect to the concentrations of bicoid found in vivo.

Whatever the pace of progress, however, it is of small bricks like these that knowledge of biological mechanisms is built. Even this system, which is so well defined and long-studied, has endless complexities that arise when one looks closer than the schematic models that were originally advanced to explain it. For example, a recent paper discussed how bicoid activates some genes in the very posterior of the embryo, despite occurring at vanishing (nano-molar) concentrations. They showed that aggregations arise in the posterior that, despite the low average concentration, provide high local concentrations, and thus, plausible transcription activation activity by bicoid. Then there is the complexity of the core transcriptional apparatus, which has clearly impeded efforts to fully delineate the cooperative structures that form between it and activating / repressing regulators bound at peripheral sites.

  • RNAs enter the mix at enhancers and promoters as well.
  • Loving the Dead.
  • Jared Diamond on the current situation.

No comments: