« August 2007 | Main | October 2007 »

September 27, 2007

Cooperation and cheating in microbes: quorum sensing and persisters

Two papers on cooperation this week. If you were trying to help someone, but end up causing problems for them, were you being cooperative? I have no idea, so I like to study cooperation in microbes. Microbes don't have brains, so "intent" isn't a factor. And the only definition of "benefit" that makes sense is an increase in Darwinian fitness or reproductive success, which is often easy to measure in microbes; just count them.
I like these definitions:

Cooperation: a behaviour which provides a benefit to another individual (recipient), and which is selected for because of its beneficial effect on the recipient. [Exhaling CO2 isn't cooperation; it evolved as a side-effect of breathing oxygen, not to benefit plants.]
Cheaters: individuals who do not cooperate (or cooperate less than their fair share), but are potentially able to gain the benefit of others cooperating. ["Equal share" might be less ambiguous.]

The evolution of cooperation and cheating in microbes is a hot topic these days, for various reasons. Species without brains simplify the analysis of cooperation and allow interesting comparisons with humans and other animals. (Do humans mostly cooperate with relatives, as bacteria and some animals do, or do our brains lead to cooperation with nonrelatives?) Some kinds of cooperation among microbes may shed light on the origin of multicellular life. Cooperation among microbes (all investing in nitrogen fixation, for example) is critical to the benefits they provide to plants, while other kinds of cooperation may make pathogenic microbes more deadly.

Cooperation can be pointless if the number of cooperators is below some threshold. That's why listener-supported radio stations in small towns are rare. I have suggested a solution to this problem elsewhere, but my solution only works for reasonably intelligent life-forms.

How do bacteria solve the threshold problem? For example, a single bacterial cell can't provide its plant host with enough nitrogen to make much difference; only billions of bacteria, working together, can do that. Similarly, only a large number of harmful bacteria, working together, can produce enough of a "virulence factor" to overcome host defenses.

Bacteria often exchange "quorum-sensing" chemical signals that control activities requiring a large number of cells. Each cell produces a little homoserine lactone or whatever. If there are enough cells in a small enough space, the homoserine lactone concentration gets high enough to trigger cooperative activities.

This week's first paper is "Social cheating in Pseudomonas aeruginosa quorum sensing", by Kelsi Sandoz, Shelby Mitzimberg, and Martin Schuster of Oregon State University, just published online in the Proceedings of the National Academy of Science. (I will also discuss another recent paper on the evolution of "persister" cells.)

Pseudomonas aeruginosa is a pathogen of humans that typically uses quorum sensing to control production of virulence factors. However, mutants that fail to respond to quorum sensing signals are commonly found in patients. The authors of this week's paper suggest that these mutants are "cheaters" (see above) because they benefit from the virulence factors produced by others, without paying the metabolic cost of making these factors themselves. (Just as some humans listen to listener-supported radio without contributing.)

It's hard to measure costs and benefits to individual bacteria inside sick people, so they developed an experimental system to study cooperation and cheating related to quorum sensing in culture flasks. The main energy source they provided to the bacteria was the protein, casein, which has to be broken down outside of the cells, by excreted enzymes, before the bacteria can use it. These excreted enzymes are normally involved in disease and are controlled by quorum sensing. The bacteria benefit, collectively, by producing these enzymes, but there's an individual cost to make them.

Therefore, it is not surprising that the percentage of the bacterial population not making a key extracellular enzyme increased to 20% in about two weeks. Most of these were "defective" in production of several quorum-sensing-controlled factors. When they checked the DNA sequence of a key quorum-sensing gene, all were found to be mutants. The more of these cheaters there were, the slower the overall population growth. When the cheater and cooperator were grown together, the cheater had a generation time of 206 minutes, while the cooperators grew more slowly, with a generation time of 335 minutes.

Given their faster reproduction, what prevents these cheaters from taking over? If there are too many of them, the overall population growth would be slowed by lack of extracellular enzymes. However, I wouldn't expect this to hurt the cheaters as much as it hurts the cooperators, with their higher costs. They say that "compensatory mutations appeared to emerge before the cheater load became detrimental to the entire population, essentially converting cheaters into cooperators." I don't understand this. Mutations are usually random, with respect to their effects on fitness, and I wouldn't usually expect a mutation that helps others to spread. This aspect of the research needs to be explored further.

Thanks to my brother, Glenn, for suggesting this paper.

The second paper is a theoretical analysis of bacterial "persisters." These cells are alive, but so dormant that they are resistant to many antibiotics that would kill them if they were more active.

Andy Gardner, Stuart West, and Ashleigh Griffin, of the University of Edinburgh, point out that becoming a persister can benefit bacteria in two different ways. The direct benefit is resistance to catastrophes, such as antibiotics.

But there is also an indirect benefit. Persisters don't consume resources, so they free up resources for other bacteria nearby. If those other bacteria are close relatives (especially clone-mates), then a gene that increases the chance of a bacterium becoming a persister may increase the survival and reproduction of other copies of itself, consistent with the selfish-gene and kin-selection hypotheses. On the negative side, as long as a cell remains in the persister state, it is not reproducing.

So, what percent of a given bacterial clone should become persisters? It depends on:
1) the frequency of catastrophes (make more persisters if catastrophes are frequent)
2) the availability of resources (if resources are scarce, a cell is giving up less potential reproduction in becoming a persister, and relatives will benefit more from the reduction in demand for resources), and
3) how related the bacteria are (freeing up resources for use by nonrelatives is a losing strategy).

Testing these predictions should be fun.

September 21, 2007

Thanks, Google Scholar!

I'm sure I could make more money doing something else, but this made my day. That and getting another rhizobium paper accepted by Proceedings of the Royal Society. Would it be greedy to go for the top three?

September 20, 2007

Menopause trade-offs

Why do women, in contrast to our closest relatives, stop giving birth while they are still relatively young and healthy? This week's paper. "Testing Evolutionary Theories of Menopause", by Daryl Shanley and coauthors, published in Proceedings of the Royal Society, uses data from people living in The Gambia to test two different hypotheses.

Both hypotheses assume that human physiology has been shaped by natural selection to maximize the number of surviving descendants. (This seems to be the philosophy of fundamentalists of various religions as well, despite alternate interpretations of ancient religious texts.) As Darwin wrote:

I use the term Struggle for Existence in a large and metaphorical sense, including dependence of one being on another, and including (which is the more important) not only the life of the individual, but success in leaving progeny.

The two evolutionary hypotheses tested were:

1) if older women are more likely to die in childbirth, leaving their existing children at risk, then caring for existing children may be a better (DNA-programmed) strategy than bearing additional children.

2) grandmothers may enhance the survival of their grandchildren enough to more than make up for having more children of their own.

In this paper, they used data from The Gambia, collected from 1950 to 1975. (After that, improvements in medical care decreased infant mortality to levels no longer representative of most of our evolutionary history. )

Having a living mother increased survival of children under 2 more than tenfold. Having a living maternal grandmother increased survival twofold. Other relatives, including fathers, had no effect.

They used these data in a mathematical model simulating mothers of different ages, children, and grandmothers. If they assumed death in childbirth increased sharply with age, there was little net benefit (more children minus risk of existing children dying) in reproduction late in life, but no net cost. Combining the effect of mothers and grandmothers on survival to age 2 gave at most a slight benefit to menopause. However, if the effects of grandmothers on survival to age 15 were included, menopause at 55 was optimum (open circles in figure). Similar extensions of the maternal benefit had less effect, apparently because there are more children without grandmothers than without mothers.
There was some uncertainty in the estimates of the effects of having a mother or grandmother on survival, and even more uncertainty as to how dangerous childbirth at 60 would be (no data!). Data on chimps probably would not be helpful, as baby chimps have smaller heads. But maybe data for other human populations could be used to fit a curve that could then be extrapolated beyond the current range of human reproduction. I also wonder how relatively recent cultural advances, including agriculture and midwifery, affect lifespan, child-birth associated mortality, etc., even in hunter-gatherer societies.

This entry is dedicated to my mother in honor of her 80th birthday.

September 15, 2007

The pirate code

R_RRR R_!R_!R!!_R!R !R!!_!!_R!R_! !R !RR!_!!_!R!_!R_R_! !!_!!! !!R!_!!_R!_! R!!!_!!R_R R_!!!!_!!_!!! R!R!_RRR_R!!_! RRR_!RR!_! R!_!!! R_!!!!_! !RR!_RRR_!R!_R_!!!!_RRR_!R!!_! R_RRR !R_!!!R_!R_!!!_R !RR_RRR_!R!_!R!!_R!! RRR_!!R! !RR!_!!_!R!_!R_R_! !R!_!R_R!!_!!_RRR !RR_!!_R_!!!! !R_R!!_R!!_!!_R_!!_RRR_R!_!R_!R!! !_R!_R!R!_!R!_R!RR_!RR!_R_!!_RRR_R! !!_ !!R! R_!_!_R!!_!_R!!

September 14, 2007

Money for monkeys

This week's paper is "Do capuchin monkeys (Cebus apella) use tokens as symbols?" by E. Addessi and coauthors, published in Proceedings of the Royal Society.

Humans use symbols in various ways, from drawings that somewhat resemble the object represented to national flags and religious symbols that represent complex ideologies (or at least group identity). Is symbolic reasoning a uniquely human trait, at least on this planet?

Chimps have been shown to use symbols, correctly choosing between previously learned symbols for "food" and "tool" when presented with a new tool or food item, while another chimp learned numbers well enough to pick the higher one when that resulted in more food. But what about monkeys?

The authors trained capuchin monkeys to exchange tokens (poker chips, metal nuts, etc.) for food. "Denominations" were assigned to each of two tokens so that one could be exchanged for 3 times as much food as the other. After the monkeys had some opportunity to learn the exchange-value of each token, through repeated exchange sessions, they were given opportunities to choose between one 3-treat token B versus 1 to 5 1-treat tokens A. A smart monkey would pick one B token over 2 A tokens but not over 4 A tokens.

Actual performance varied among monkeys. Four of ten followed the optimum strategy. Four monkeys learned to prefer the high-value token B, but apparently didn't count how many low-value token A's were available as an alternate choice.

Only one monkey consistently chose correctly between several B versus several A tokens. The question is, could that monkey find a girlfriend?

In other words, to what selective forces, operating on our ancestors, can we attribute the ability of some of us to compose music or prove mathematical theorems?

Also this week:

Placing late Neanderthals in a climatic context

Crystal Structure of an Ancient Protein: Evolution by Conformational Epistasis

September 11, 2007

Evolving enzymes in the lab

This week's paper is another example of how nonrandom selection from among random variants can solve problems so difficult that we are unable to "design" a solution. As in an earlier post, the selection process was automated, not requiring the human judgement used in breeding crops or dogs.

"Selection and evolution of enzymes from a partially randomized non-catalytic scaffold" was written by Burckhard Seelig and Jack Szostak, both of Boston, and published in Nature (448:828). Their goal was to evolve an enzyme to link two RNA bases together in a particular way, a reaction not found in nature.

Enzymes are biological catalysts, which speed the rates of chemical reactions. Most natural enzymes (that we know of) are made from protein, but a few are made from RNA, possible relics from a hypothesized "RNA world" where RNA acted both as enzymes and as genetic material. Enzymes made of DNA have never been seen in nature, but laboratory conditions have been designed that allow them to evolve from random DNA sequences (Science 286:2441).

Designing an environment where protein-based enzymes could evolve, without using living cells, was actually trickier than earlier evolutionary systems for RNA- and DNA-based enzymes. The authors used a huge library of 1000 billion random DNA variants. All the DNA sequences had two loops, but the contents of the loops varied randomly.

Transcribing all of the DNA variants into messenger RNA and translating the messenger RNAs into proteins is a routine operation; the trick was to leave each protein variant linked to its particular mRNA. Then they stuck one of the substrates of the desired reaction to the RNA-protein combination and turned them loose over a surface with lots of the other substrate bound to it. If the protein could catalyze a reaction to link the two substrates together, the whole complex got tied to the surface and used in the next generation. All the nonfunctional variants got washed away.

The DNA sequences that survived 8 generations of this selective "sieve" were randomly mutated and subject to additional cycles of selection. The whole process took a few days. Evolution can be fast, if the conditions are right. The final product accelerated the reaction over a million times. The authors suggest that this approach could be used to evolve other enzymes that link substrates together. A modified method, saving the enzyme complexes that wash away, could be used to select
enzymes that break chemical bonds rather than make them.

Other recent papers on evolution in major journals:

Genome-wide expression dynamics of a marine virus and host reveal features of co-evolution

Recombination Speeds Adaptation by Reducing Competition between Beneficial Mutations in Populations of Escherichia coli

Cryptic Population Dynamics: Rapid Evolution Masks Trophic Interactions

High-Resolution Genome-Wide Dissection of the Two Rules of Speciation in Drosophila

Heritable Stochastic Switching Revealed by Single-Cell Genealogy

Evolution in the Social Brain

Social Components of Fitness in Primate Groups

A Basal Dromaeosaurid and Size Evolution Preceding Avian Flight

A new theory for the evolution of polyandry as a means of inbreeding avoidance

Adaptive evolution of genes underlying schizophrenia

Reed bunting females increase fitness through extra-pair mating with genetically dissimilar males

Evolution of a single niche specialist in variable environments

September 6, 2007

Evolution avoidance syndrome

That's the title of an essay by my colleague Scott Lanyon. He notes that "development" refers to changes within an individual, whereas changes in the genetic composition of a population are known as "evolution." Apparently some public officials were afraid to say that a fish population could "evolve" resistance to a newly arrived pathogen, so they say they hope resistance will "develop." This is confusing, because individual susceptibility to pathogens can develop, increasing or decreasing with age, but that's not what they were talking about.

I used to run into a similar problem when I worked in an agronomy department. Some of the people I interacted with would say that an herbicide had "broken down", when actually the weed species it once killed had evolved resistance to it. The change was in the weeds, not in the pesticide. This misuse of the English language is particularly harmful because herbicides do break down (chemically degrade), which is usually a good thing; we don't want them polluting lakes, for example.

Populations evolve, but don't worry, fish and weeds didn't evolve from apes.

September 5, 2007

If it's junk, can we get rid of it?

This week's paper is "Deletion of ultraconserved elements yields viable mice" by Nadav Ahituv and collaborators, published online in PLoS Biology.

The instructions for "life as we know it" are coded in DNA, but it appears that only a fraction of our DNA is ever used. (This is probably not true of our brains, myths notwithstanding.) At least, only a fraction of it is ever translated into proteins such as enzymes. Some of the untranslated (noncoding) DNA has known functions, such as coding for the RNA part of the ribosomes that translate messenger RNA into protein, but much appears to be junk. Much of the junk is multiple copies of transposons, bits of unusually selfish DNA that reproduce like rabbits and burrow into the chromosomes, sometimes presumably disrupting functional DNA.

But if the noncoding DNA is mostly useless junk, why has some of it apparently been preserved by natural selection?

That was the question raised in an earlier post. About 5% of noncoding DNA is highly conserved. That is, the DNA sequence in related species is very similar, as if it hadn't changed in the millions of years since they diverged from a common ancestor. In that length of time, you would expect changes due to mutation.

Conserved DNA sequences are easy to understand when they code for protein. Mutant versions of the protein didn't work as well, so only those individuals with the original version survived and reproduced. But why would noncoding DNA sequences be conserved? Maybe they have some function, after all?

The authors of this week's paper conducted a direct test of this hypothesis, by deleting highly conserved (identical in mouse and man) but noncoding regions from the DNA of mice. They only deleted four sequences, but those were chosen to maximize the chance of seeing an effect. For example, although none of the deleted DNA coded for proteins, the sequences were near proteins previously shown to have a large effect. So if the deleted DNA regulated expression of nearby proteins, the mice should have been noticeably different. Dead, for example.

The DNA-deleted mice appeared normal, however, in growth and biochemistry. Matings among themselves or with normal mice gave the usual number of offspring per litter. 2% of the mice had only one kidney, versus an estimated 0.1% in normal mice, but that was the only abnormality found, and it wasn't necessarily caused by their deletions.

What can we conclude from these results? There seem to be several possibilities:

1) The deleted sequences (and by extrapolation, much noncoding but highly conserved DNA) really is junk. If so, why are they conserved? Maybe those parts of the chromosome are somehow more protected from mutation than other regions.

2) The deleted sequences are important, but only under conditions not tested in the laboratory. Although they don't appear to control nearby protein-coding genes, maybe they control ones farther away. Or maybe they code for RNA that (without translation into protein) interferes with a virus not found in the laboratory. The kidney defect needs more research.

3) The deleted sequences serve some important function, but there are backups with similar function (though perhaps with different sequence) elsewhere in the genome. The authors seem to like this hypothesis.

This paper reminded me of an earlier paper (Nature 395:905) in which genes for myoglobin (thought to be critical to oxygen supply in muscles) were knocked out in mice. Although their muscle tissue was pale rather than red, the mice were able to exercise normally. If we have backup systems that let us (us mice, that is) survive without myoglobin, perhaps it's not surprising that deletion of DNA sequences with no known function had little apparent effect.

T. Ryan Gregory has also posted an interesting discussion of this week's paper.