Finding method in biological madness
February 8, 2011 § Leave a comment
Our tiny human minds are often boggled by the complexity of biological systems. If that system evolved to do the job we think it does, we say to ourselves, why couldn’t it have been simpler? Surely all those extra components and those extra feedback loops aren’t really necessary. Or are they?
A multinational, multidisciplinary team led by Hans Westerhoff recently provided possible rationales for the complexity of a key type of signaling pathway, the nuclear hormone receptor pathway (Kolodkin et al. 2010. Design principles of nuclear receptor signaling: how complex networking improves signal transduction. Mol. Syst. Biol. 6 446). This is the class of signaling pathway responsible for responding to important hormones such as thyroid hormone, testosterone and estrogen, and many others. The receptors are transcription factors, with a DNA-binding domain and a ligand-binding domain, and the textbook view of how they work used to be that they sat in the nucleus bound to DNA and waited for their ligand to come along and activate them. Because their ligands are small, hydrophobic (or, better, lipophilic) molecules that can readily pass through cell membranes, nothing special was thought to be necessary to get the signal to the nucleus.
But things are more complicated now. (Ah, for the dear dead days of innocence.) Kolodkin et al. show us just how much more complicated with a figure describing the 7 distinguishable modules of the core nuclear hormone signaling network. This figure depicts over 100 moving parts, and it only covers the pieces that are essentially the same in all nuclear hormone signaling pathways. The most prominent difference between the old model and the new mega-model is that there is an enormous amount of work going on to move the receptors in and out of the nucleus. I’ve drawn you the old “sit and wait” model; I’m not even going to think about trying to draw the new model, but you can find it here.
Why all these moving parts? Kolodkin et al. identify eight different puzzles in their core network, of which one is central: why are nuclear hormone receptors ever allowed to move out of the nucleus into the cytoplasm? This has at least two apparently negative consequences. First, it takes the receptors away from the neighborhood of the DNA; if they’re not near the DNA, they can’t signal, so you would think this would reduce the efficiency of signaling. Second, as the receptors can leave the nucleus, there is now a need to move the receptors back in again, which happens through interactions with importins; this process costs energy which you would not have to spend in the simple model.
To find out whether these extra features affect the behavior of the network, Kolodkin et al. first modeled the function of the simplest possible “sit and wait” model, using parameter values that are as realistic as they can make them. To their surprise, this simple model doesn’t provide effective signaling at all: at low but realistic hormone concentrations (0.005 nM) only 1/200 of the receptors in the nucleus end up with bound ligand, and the transcriptional response is negligible. This isn’t just a trick of the parameters they picked; in their supplementary figures, they show that varying the parameters used by 5 or 10-fold makes little difference.
If the simple model simply doesn’t work, that would seem to be a pretty good clue as to why the real biological situation is more complicated. But why is allowing the receptor into the cytoplasm helpful? The authors separate this into two subquestions: if you have receptor in the cytoplasm, but the receptor doesn’t move between cytoplasm and nucleus, could that be a good thing? And if you allow movement between compartments, is that better?
In both cases, Kolodkin et al. argue that the answer is yes. A lipophilic ligand, they point out, can easily cross cell membranes but may not find it quite so easy to cross the aqueous cytoplasmic compartment. If you have receptor in the cytoplasm, this might act as a chaperone, or “ferry boat”, helping the ligand to move across the relatively large distance from one membrane to the next. Although the receptor is larger than the ligand, and therefore diffuses more slowly, there’s more ligand-receptor complex than free ligand in the cytoplasm (again, using parameters that are as realistic as the authors can make them) and so the mere presence of receptor in the cytoplasm increases the rate of ligand accumulation in the nucleus, enhancing transcriptional activity. (Of course, the ligand can still diffuse across directly.) This effect is particularly important at early time points, because the rate at which ligand makes its way across the cytoplasm is limiting for the rate of the transcriptional response. [There is also the possibility that the nuclear receptor does something unrelated to its transcriptional activity while in the cytoplasm; let’s not go there, this story is complicated enough.]
Now what if you allow (energy-consuming) shuttling between nucleus and cytoplasm? Kolodkin et al. try various implementations of this, all of which enhance transcription at least a little. The one that really looks as if it would make a difference is a scheme in which ligand-bound receptor is selectively imported into the nucleus. There’s a potential molecular mechanism for this: at least some nuclear receptors (such as the glucocorticoid receptor) have two nuclear localization sequences, one of which only becomes available when the receptor binds to ligand. And there are experiments in the literature suggesting that selective import happens (one of which Kolodkin et al. reproduce in their Supplementary Material). If you allow selective import of liganded receptor, you effectively use receptor shuttling as a way of pumping ligand into the nucleus. This results in a response that is faster and larger than any of the other models.
There is much more in this paper for those who are interested in the specifics of nuclear receptor function, and several interesting ideas that could (and probably will) be experimentally tested. For me what’s most interesting is the way this analysis shows that an informal model that looks perfectly reasonable at first glance (the sit-and-wait model) becomes much harder to defend when you articulate your assumptions more formally — and how features that might initially seem counterintuitive turn out to be quite sensible after all.
If you want to play with the models yourself, you can: the simple model is here, and the selective transport model is here. Each of the models described in Kolodkin et al.’s Figure 2 can be found at http://jjj.bio.vu.nl/webMathematica/Examples/run.jsp?modelName=kolodkinX, where X is the number of the model.
Kolodkin AN, Bruggeman FJ, Plant N, Moné MJ, Bakker BM, Campbell MJ, van Leeuwen JP, Carlberg C, Snoep JL, & Westerhoff HV (2010). Design principles of nuclear receptor signaling: how complex networking improves signal transduction. Molecular systems biology, 6 PMID: 21179018