Wednesday, April 18, 2012

Bulking up....or not: what chaos can tell us about order

A paper in last week's Science, (Using Gene Expression Noise to Understand Gene Regulation, Munsky et al.), as part of the special issue on computational biology, asks whether what looks like noisy gene expression can be informative about gene regulation.  Gene expression within a single cell is a topic of growing interest.  It has seemed to be fairly random, but Munsky et al. suggest if you look closely enough, the randomness can be indicative of some quite regular processes.  Mechanical engineer Brian Munsky and colleagues have used a similar approach to identify gene regulatory networks.

If molecules interact in a probabilistic way--bouncing randomly around the cell until they perchance (literally) bump into each other, and if each cell has countless zillions of molecules buzzing around in this way all the time, and if it's clear that the cells even in the same tissue in the same person (and hence the same genotype) are each a bit different....then how come we're so highly organized into tissues and organs that mostly work mostly correctly.....rather than being just a jiggling blob of formless jelly?

Sad to say, but Prairie Home Companion can't be true!
One obvious possibility is what one could call the law of large numbers, or a principle of central tendency.  All of these random motions have an average behavior, just like everybody's height or glucose levels vary but most of us are somewhere near the middle.  Unlike Minnesotans in Garrison Keillor's Prairie Home Companion, not all the children are above average!

With large numbers of cells in a given organ, there will be variation but only a small percent of cells will behave very differently from the average, and even if they are very naughty indeed, their effect on the organ as a whole--and, say, on the person's health--is trivial: the vast majority of well-behaving cells cover for the wayward ones.  And, indeed, we have bodily systems to detect cells that are far too misbehaving.  When they fail we can get nasty conditions such as cancers.

So how is this stochastic (probabilistic) buzz-fest made manifest at the level of individual genes and their levels of expression (use) by cells?   As the authors of this paper note, because of the vagaries of Brownian motion, two cells, even those produced by the same progenitor cell, will never be identical at the molecular level.  Thus, things like the number of transcription factor molecules per cell, needed to cause specific other genes to be expressed, are unlikely to be identical, and this cell-to-cell variability will affect gene expression levels among cells, and ultimately can lead to phenotypic variability as well.
Consider a single mother cell dividing into two daughter cells of equal volume. During the division process, all the molecules in the mother cell are in Brownian motion according to the laws of statistical mechanics. The probability that each daughter cell inherits the same number of molecules is infinitesimally small. Even in the event that the two daughter cells receive exactly one copy of a particular transcription factor, each transcription factor will perform a Brownian random walk through its cellular volume before finding its target promoter and activating gene expression. Because Brownian motion is uncorrelated in the two daughter cells, it is statistically impossible for both genes to become activated at the exact same time, further amplifying the phenotypic difference between the two daughter cells.
Munsky et al. use cell-to-cell variability in gene expression as a way to understand gene regulation, and they present quantitative models for this. Genes can be expressed all the time, or their expression can be episodic or timed -- this is 'constitutive' vs 'regulated' expression.  Taking into account copy number of transcripts of genes of interest, they determine that when transcript births and deaths are not related, and seem to follow a Poisson distribution (that is, they are independent events that occur at a known average rate, but with different probabilities of any specific rate); this indicates constitutive expression.  Deviation from the Poisson distribution--too many rare or too many common copies, for example--suggests regulated or episodic expression; expression of a gene within a cell can be more or less tightly regulated over time, and can switch states.

Documenting and making sense of transcript levels in single cells, the authors write, can be informative about gene regulation as well as gene networks in ways that looking at gene expression in multiple cells or tissues can't be.  This is because average statistics, such as of the number of transcripts of a particular gene among cells, mask the distribution within each cell, and so regulatory mechanisms can't be inferred. Yet most cellular studies are of test-tubes full of the 'same' kinds of cells, analyzed in aggregate, masking this informative, underlying variation.

Hounds and Hares
If, as the authors predict, sequencing of all the transcribed genes in a single cell becomes routine, understanding gene networks will be easier.  And it will account for our orderliness as organisms.  This can be seen at higher levels, in ordinary experience, too, as this example may help make clear:

Hounds chase hares, but each hound and each hare live individually different lives.  If we want to understand the overall organization of the hound-hare part of the ecosystem, such as how their respective populations vary over space and time, we can look at the aggregate behavior: the chance a hound will sniff a hare, the chance it will catch the hare and so on.  But if we want to understand details, we might have to follow a number of individual hounds and hares, because not all will be equally successful in their hunt, and the circumstances of the hunts will vary.

Bulking up....or battening down?
These ideas apply to situations when there are many 'identical' cells, as in a given organ.  The central tendency would seem to provide safety in numbers.  But if that's the case, how big do the numbers have to be to protect the organism from its fraction of unusually behaving cells?  This will depend on many things, including the 'variance' (variation from cell to cell) of the process, how many cells are in sync at each time, and so on.

There's another important issue. In complex tissues, gene expression is changed via various processes that we can call 'signaling'.  Cells sense their surroundings and respond to them.  They send out signal molecules appropriate for their location.  This reinforces cells in a given tissue to do what's appropriate for the tissue.  And, importantly, signaling can be homeostatic:  some signals are called 'activators' because cells detecting the signal's presence activate that same gene, or some set of response genes, as a result.  Other signals are 'inhibitors' and induce cells to do the opposite.  Waves of expression can generate waves of tissues (like hairs or scales) in an embryo, but activation and inhibition interactions can also generate stability, as signal levels can lead cells too far out of line, so to speak, to fall back into line--to batten down and stay within acceptable limits.  Could this be relevant for the question at hand?

In organs with lots of cells, say in the millions, perhaps that is bulked up enough to bar the door against cellular chaos.  But what about smaller organisms, of which there are many, in whom, like small fleas  biting the backs of larger fleas, the organs can be very small indeed?  Is there any evidence that their organ stability is less, or is more vulnerable to vagrant cells?  Does natural selection work differently in such organisms (e.g., they have to reproduce more or faster, to stay viable as a population)?  We haven't thought about this directly, though we did refer to the issue of embryonic selection as contrasted with competitive Darwinian selection in our book.  Perhaps there's no issue here, or perhaps there is something interesting to follow up.

If we're understanding this paper correctly, it's an interesting application of physics to biology.  This might seem to resemble the ideal gas law in chemistry and physics.  There, for a given container and number of molecules, the pressure or temperature of any gas follows the same law.  But this does not require tracking any individual molecules.  In a way central tendencies of similar cells are like that.  But since each cell is different, each gene is different, and gene expression can affect other gene expression, cells are not like containers of oxygen or hydrogen.  Still, when large numbers of molecules are involved, the distribution of traits among similar cells do seem to follow orderly statistical properties.

No comments: