Aging in Bacteria Redux

Earlier I posted on aging in Bacteria. Today I received an email from Moselio Schaechter, author of the most excellent blog, Small Things Considered. I reprint the email's contents here (with his permission of course).

"About your exciting posting on aging in bacteria, I am reminded of an experiment that I did perhaps 50 years ago. I have no record of it and can only remember the results in general terms.

In an early day preoccupation with the meaning of "balanced growth," it became crucial to have faith in the homogeneity of Salmonella cultures, which speaks to the question bacterial senescence. The experiment could not have been simpler, and had probably been done by others many times. I coated a glass slide with agar, put a thin suspension of cells on it, waited for the liquid to be imbibed, checked under a high dry objective (with phase contrast) to see that the cells looked reasonably single, incubated until ca. 5-6 generations had passed (in rich medium, ca. 2 hours), and scored the number of cell that had not made a microcolony. The answer, best I can remember, is that there were none in the fields examined. I don't remember how many microcolonies I had scored, but it must have been at least 1000. So, under the conditions and by this criterion, there were less than 0.1% non viable cells. Incidentally, my recollection is that the microcolonies were quite uniform in the number of cell each. I walked away thinking that there may not be much senescence in my bugs.

As data go, this is useless at this point, not having anything on paper (and perhaps undergoing senescence myself). However, the experiment can be easily and cheaply repeated using robotic colony counters that can measure the number of cells/microcolony. The results can be obtained in a couple of hours. The level of sensitivity can be quite high, perhaps 1 in 10exp5 or better..

Let's say this is true. It suggests that perhaps the difference between Stewart's result and this one is agar. I have no problem imagining a mechanical obstacle to old cell poles, something that may not matter in liquid medium. Also, cell crowding may be a factor, although it doesn’t show up in his work. As for Caulobacter, even in liquid, I can't really fathom if this is relevant because of the different life styles of the two bugs."

The interesting thing is that the work I have been doing for the past year requires immobilising E. coli on glass slides with poly-lysine and taking movies of them over time. My object was to observe phage lysis times, but conceivably the process could be modified to conduct the work Schaechter suggests. In fact, I might already have the data, albeit encoded in movies waiting to be deciphered.

So what to make of Schaechter's comments. Unfortunately without Schaechter's data, it is difficult to make comparisons between the two studies. However, Stewart et al. write:


"During the growth of the microcolonies, sixteen cells were observed to cease growing; these cells never resumed growth during the course of the experiment. We have defined these cells as potentially dead cells and have analyzed their locations in the lineages. While these apparent deaths may ultimately be due to stochastic events, they show a statistically significant bias (p = 0.01; see Materials and Methods) toward increased divisions spent as an old pole (over the total observation history)."

Given that Stewart et al. observed 35,049 cells, the fraction that "died" is 4.6exp-4.... far below Schaechter's estimated 0.1%. Perhaps we should be asking why did so many cells in Schaechter's experiment die? (Update 9/1/07: From Schaechter: Sorry, but, best I remember, I didn’t see any die. I just guessed that their upper limit had to be less that 1/1000.) Schaechter's impression that microcolony were quite uniform in number of cells suggests that these microcolonies will contain both "old" and "young" cells, and that their growth rates are similar. The question is, given an initial population that supposedly contains both old and young cells, are the differences in growth rate between cells sufficient to produce large differences in microcolony sizes when microcolonies are initiated on agar? This sounds like a great question for Eric Stewart. Presumably he has this data since he tracked microcolonies by film.

Figure above from Stewart et al. 2005. Legend below.
Figure 2. Average Lineage Showing Old Pole Effect on Growth RateThe first division in the microcolonies is not represented, as the identity of the poles is not known until after one division (hence each initial cell gives rise to two lineages that are tracked separately, and subsequently combined from all films to create the single average lineage shown here). The lengths of the lines connecting cells to their progeny are proportional to the average growth rate of that cell; a longer line represents a higher growth rate for that cell. At each division, the cell inheriting the old pole is placed on the right side of the division pair, and shown in red, while new poles are placed on the left side of each pair, and shown in blue (note that this choice of orientation is not the same as that of Figure 1, to compare more easily old and new pole lineages). Because the position of the start of the growth line for each new generation is dependent on the generations that preceded it, the difference in growth rates is cumulative. Green lines indicate the point at which the first cell divides in the last four generations. Nine generations from 94 films encompassing 35,049 cells are included in this tree. The average growth rate of all the cells corresponds to a doubling time of 28.2+/−0.1 min. The data used to generate the average lineage are provided in Dataset S1.

This Week's Citation Classic

W. Fiers, R. Contreras, F. Duerinck, G. Haegeman, D. Iserentant, J. Merregaert, W. Min Jou, F. Molemans, A. Raeymaekers, A. Van den Berghe, G. Volckaert & M. Ysebaert. 1976. Complete nucleotide sequence of bacteriophage MS2 RNA: primary and secondary structure of the replicase gene. Nature 260:500-7.

Today full genome sequences are popping up almost daily, so it's almost hard to believe that over 30 years ago, the first genome was sequenced. I've often heard that the first genome sequenced was phage Phi-X174 by two-time Nobelist Fred Sanger and colleagues. Not true! Phi-X174 was the first DNA genome sequenced. Walt Fiers and colleagues sequenced the first full genome, the ~3.5kb RNA genome of the bacteriophage MS2.

Photo: Electron micrograph of Leviviridae courtesy of Hans-Wolfgang Ackermann, Laval University, Quebec.

One Species' Entire Genome Discovered Inside Another's

From Science Daily

I confess I was a little mystified when I saw this headline. From my perspective as a phage biologist, this is not really news. Temperate phages, like lambda, do this all the time. But then wolbachia is such a neat beastie. It deserves a little press.

However, the article got me thinking about another host/parasite association: caterpillars and wasps (including Darwin's famous ichneumonidae). The female wasp, as is famously known, paralyses the caterpillar and lays her eggs inside the unfortunate creature. The eggs hatch and the resulting larvae devour the caterpillar from the inside out, while taking care to avoid the major organs so as to keep the caterpillar alive and the meat fresh.

Turns out this weird story gets even weirder. These wasps are infected with polydnaviruses. The full genome of the virus is integrated into the wasp's genome and the virus replicates only in specific cells in the female wasp's reproductive system. Here things get really interesting. The polydnaviruses are injected into the caterpillar along with the wasp eggs. The resulting infection does not lead to viral reproduction, but rather the viruses affect the caterpillar's immune system. Specifically they suppress the caterpillar's phagocytic hemocytes which, in the absence of viral infection, would encapsulate and kill the wasp eggs.

There are two known genera of polydnaviruses: the ichnoviruses and bracoviruses, infecting ichneumonid and braconid wasps respectively. There is little sequence homology between the two genera, suggesting that this virus "lifestyle" evolved independently in two wasp lineages. Rather than being parasitic to the wasp, the virus has evolved a symbiotic relationship and gang up to parasitise the caterpillars. How cool is that?

Photo by: David Wahl, American Entomological Institute

This Week's Citation Classic

John J. Dennehy, Stephen T. Abedon, Paul E. Turner. HOST DENSITY IMPACTS RELATIVE FITNESS OF BACTERIOPHAGE Φ6 GENOTYPES IN STRUCTURED HABITATS. Evolution (OnlineEarly Articles). doi:10.1111/j.1558-5646.2007.00205.x

Citation classic? Ah...not quite. Just kidding. No, this is just my latest published paper, now appearing online early in Evolution. The work stems from a side project I conducted while I was a postdoc in Paul Turner's laboratory at Yale. What originally began as a simple experiment in the spring of 2005 to determine whether the density of hosts in a habitat affected parasite competition quickly snowballed as additional experiments were conducted, data analyses and reanalyses performed, revisions made and resubmitted (3x! The first submission was on 1/31/06), authors added (the estimable Stephen Abedon) and much sweat, angst and time sacrificed. This paper is the fruit of the most difficult effort I've undertaken in my fledgling career. (However, as Homer Simpson might say, "The most difficult effort of your career... SO FAR!").

Here we competed two Phi6 strains over a range of host densities in two separate habitats: in liquid culture and on agar plates. In liquid culture, the results were not surprising. The more fit phage out-competed the less fit phage over all host densities. This result was expected because, in a well-mixed liquid culture, the phages are not spatially limited in their access to hosts, and their net reproduction should be a product of the number of hosts and the reproduction per host. Since the more fit phage produced more babies per host (in the paper, greater burst size), its advantage over the less fit phage should be consistent over all host densities, and the total number of babies produced by both strains should increase with increasing host density (See Fig. 5 below. Note: the slopes are not significantly different despite appearing to converge.). By contrast, in agar, phage dispersal is limited by its ability to diffuse through the viscous agar. This causes a shift from direct competition to indirect competition between the phage strains. That is, in the liquid culture, phage compete globally for hosts, whereas in the agar culture, competition is limited locally. Here the results were surprising; the relative fitness of the less productive phage strain increased with increasing host density (up to a point where it leveled off). (See Fig. 4 below with fitness comparisons to unstructured habitat).
This result caused us considerable consternation and its cause is still under debate. Since what is actually occurring in a growing plaque (i.e., the region on a lawn of bacteria where hosts are infected and lysed) is somewhat of a black box, we can only speculate as to why we observe increasing relative fitness with increasing host density for the less productive strain. One possible explanation, as described in our paper, is:

...bacteria may differ physiologically over space depending on their initial densities. This phenomenon may be attributed to the fact that bacteria form microcolonies on a lawn, and microcolony size depends on initial density (Kaplan et al. 1981). Lower initial inocula lead to larger microcolony sizes. This outcome makes intuitive sense if we assume that microcolonies are spheres that are packed within a constant volume (the top-agar layer). Each microcolony is initiated by a single bacterial cell seeded in the top agar. Thus, if fewer bacteria are seeded, then microcolonies must grow to a larger size in order to attain the same cumulative volume.

Large microcolonies contain relatively fewer outer-surface bacteria with access to oxygen and nutrients, and with relatively unobstructed diffusion of wastes. For these reasons, large microcolonies may contain lower numbers of bacteria that are competent for phage infection. Thus, the final 20% of infections at low initial bacteria densities likely result in reduced burst sizes per cell (due to the larger microcolony size) and, therefore, less particles per plaque. This effect could be substantial with a Pseudomonas host given that it is an obligate aerobe, and that bacteria in the center of Pseudomonas microcolony may be particularly physiologically inappropriate for phage infection. To summarize, greater input of bacteria into a habitat may lead to smaller microcolonies that contain greater numbers of bacteria competent for phage infection, and this may lead to better phage growth.

Why does the less fit strain display even greater increase in phage density as plaques form under greater bacterial densities? We speculate that the presumptive poorer host physiology with larger microcolony size has a greater impact on Φ6M relative to wild type Φ6. Alternatively, Φ6M may be less able to efficiently penetrate into larger microcolonies, resulting in fractionally fewer bacteria infected within the confines of the plaque, rather than fewer phage produced per bacterium infected.

Whatever the cause of the changes in relative fitness, one thing is clear: the strength of selection against the less productive phage strain was reduced at greater host densities. Consequently these genotypes may gain time to adapt to the habitat conditions, and may eventually out-compete the more productive phage strain in evolutionary fitness. This result is relevant in a practical sense with regard to phage therapy, and is consistent with the general finding that biofilms are more resistant to phage or antimicrobial attachment than are planktonic bacterial populations. It may also be relevant to the emergence of infectious diseases where habitat structure and increased host densities permit the persistence of less fit genotypes.

Photo credit: Dennis Bamford. Phi6 adsorbing to Pseudomonas phaseolicola pili.

This Week's Citation Classic

This week's citation classics are a pair of papers published by John Hubby and Richard Lewontin in 1966.

J. L. Hubby and R. C. Lewontin, "A Molecular Approach to the Study of Genic Heterozygosity in Natural Populations. I. The Number of Alleles at Different Loci in Drosophila pseudoobscura," Genetics 54 (1966): 546-595.

R. C. Lewontin and J. L. Hubby, "A Molecular Approach to the Study of Genic Heterozygosity in Natural Populations. II. Amount of Variation and Degree of Heterozygosity in Natural Populations of Drosophila pseudoobscura," Genetics 54 (1966): 595-609.

A central tenet of Darwin's theory of natural selection is that genetic variation exists in natural populations. Selection on genetic variation is the fundamental basis for evolutionary change. However, at the time of Hubby and Lewontin's work, whether or not this genetic variation, in fact, existed was largely unknown.

Lewontin, a population geneticist, paired up with Hubby, a biochemist, to use a revolutionary new technique to examine genetic variation at dozens of loci for the fruit fly, Drosophila pseudoobscura. This new technique was acrylamide gel electrophoresis.

When a mutation occurs in a gene, it can change the amino acid sequence of that gene following translation. Different amino acid sequences can have different net electrical charges. Hubby and Lewontin isolated proteins from D. pseudoobscura, placed them in wells of a slab of acrylamide gel and ran a slight current across the gel. The proteins then migrated thru the gel at a speed proportional to their net electrical charge. Hubby and Lewontin observed that the same protein isolated from different members of the population frequently migrated across the gel at different speeds (represented by the "bands" in the photo above), a result they correctly attributed to genetic variation. This supposition was supported by the fact that the variation segregated in a Mendelian fashion.

Hubby and Lewontin's results were stunning. While some genetic variation was expected, no one was quite prepared for the enormous amounts of variation their experiments revealed. Hubby and Lewontin concluded that there was genetic variation at 39% of loci in the D. pseudoobscura genome. In fact, this method leads to an underestimate of the true amount of genetic variation because it does not account for mutations that do not lead to amino acid substitutions (i.e., silent mutations) or changes in the net electrical charge of the protein.

"This was a breakthrough, a revolutionary finding," recalled Brian Charlesworth, "It led to an explosion of work. Everyone in the field rushed out to duplicate these studies in other organisms, including humans, and they found more and more examples of genetic diversity."

The data was distinctly at odds with the Classic and Balance theories of molecular evolution (a series of posts on this from hpb are available here: controversy, redux and final) championed by R. A. Fisher and Sewall Wright respectively. In fact, data such as that obtained by Hubby and Lewontin led Motoo Kimura to propose a neutral theory of molecular evolution.

Curiously, paper II (Lewontin and Hubby) is cited far more often than paper I (Hubby and Lewontin), despite Dick Lewontin's assertions that the papers form an indivisible pair. Lewontin attributes this to the Matthew Effect.

Update: Larry Moran provides some background.

There is No Spoon

I enjoyed The Matrix (but not the sequels) because it was thought provoking, original, and causes you to question conventional descriptions of reality. John Tierney, writing in yesterday's Science Times, points out that some academics have taken the question of reality much further. Nick Bostrom, a philosopher at Oxford University, suggests there is a one in five chance that we’re living in a computer simulation.

Dr. Bostrom assumes that technological advances could produce a computer with more processing power than all the brains in the world, and that advanced humans, or “posthumans,” could run “ancestor simulations” of their evolutionary history by creating virtual worlds inhabited by virtual people with fully developed virtual nervous systems.

Interesting hypothesis, but how would you disprove it? Just like the question of the existence of God, the idea is unfalsifiable. "Are we just binary digits in a vast computer universe?" is a metaphysical question, and beyond the scope of science. Whether we are made up of atoms or computer circuits is irrelevant. We still have bodies, and there are still chairs and tables: it's just that their fundamental nature is a bit different from what we may have thought. Nonetheless, it is a bit fun to speculate. What happens at the end of history? Does it say, “Insufficient Memory to Continue Simulation.”?

This Week's Citation Classic

One of the first science books I ever read I bought at a yard sale in my early teens. Mixed among various Sidney Sheldons and Jackie Collins was a dog-eared copy of David Lack's Darwin's Finches. Up to that point, evolutionary biology was not covered in any of my science classes in junior high, nor was it discussed later when I was in high school. Focus, instead, was largely directed to anatomy and physiology, particularly of humans. Lack's book made me aware of a vast world of biology outside the strict confines of what my instructors considered "relevant." Thus, in his honor, I submit as this week's citation classic:

Lack, D. (1947) The significance of clutch size I-II. Ibis, 89, 302-352. (Not available online).

Here Lack applied evolutionary theory to the question of "Why don't birds lay more eggs?" At the time, many thinkers (e.g. Wynne-Edwards) suggested that animals limited the number of offspring for the good of the species. As Lack later wrote in "The Significance of Litter Size,"

"The popular view, repeated in explicit and implicit forms in many text-books, is that the reproductive rate of each species has been adjusted by natural selection so as to equal the total mortality, thus keeping the population balanced. It is claimed that such an arrangement has survival value through preventing over-population."

Lack's key insight was that clutch size is ultimately determined by the number of offspring parents can provide with food, not the number of eggs the mother can lay, or by any consideration of the "good of the species." This applies not just in the course of a single year, but over the parent's lifetime. As a considerable number of studies have shown, birds can feed more young than eggs they lay in a single season. Lack realized that natural selection will act to maximise lifetime reproductive success, not reproduction per brood. If birds maximised reproduction per brood, the energy expenditures required would limit their ability to survive harsh winters and provision future clutches. This hypothesis can be generalised to include, with some exceptions, all organisms that provide parental care to offspring.

Interesting evidence supporting Lack's hypothesis comes from the observation that clutch size and latitude are positively correlated. Summer day length increases with latitude providing birds with more time to find food to provision clutches thus increasing the number of offspring they can rear (however, this is of course dependent upon resource availability in the habitat).

The paper is important for several reasons. It was one of the first to suggest that life history traits of organisms are evolutionary adaptations. Second, Lack was one of the first to link evolutionary theory with population biology. Third it sharpened the focus of the study of adaptation to dispute the notion that adaptations can provide for "the good of the species", an argument later thoroughly disabused by George Williams.

Lack is also renown for studying Darwin's Finches, the iconic Galapagos birds first observed scientifically by Darwin and later studied to great effect by Peter and Rosemary Grant. Lack's influence is seen widely in the study of life history theory, sex allocation, bird behavior, population biology, speciation, and evolutionary ecology. He was a Fellow of the Royal Society and received their Darwin medal 1972 before passing away March 12, 1973.

Great Egret and chicks in Morro Bay, CA Heron Rookery, photo by Mike Baird, bairdphotos.com

Even bacteria get old....

Did you know even bacteria get old? Scientists traditionally assumed that bacteria were immortal, since these single-celled organisms split into two apparently identical daughter cells, which in turn divide, and so on. We now believe that this is not true. Eric Stewart of INSERM, the French institute for health and medical research in Paris, and his colleagues took fluorescent images of individual E. coli cells over ten generations. Each generation the E. coli cells divide down the middle, giving each daughter cell one new tip and an old tip from its mother, or grandmother, or some older ancestor. Using computer software, Stewart et al. identified and tracked the tips of each bacterial cell. The results indicated that the cell that inherits the old tip suffer a diminished growth rate, decreased offspring production, and an increased incidence of death.

More recently, Martin Ackermann, whom I met at the recent GRC Microbial Population Biology conference, and colleagues have published two papers on aging in bacteria in BMC Evolutionary Biology and Aging Cell (both open access). In the BMC Evolutionary Biology paper, Ackermann et al. evolved Caulobacter crescentus for 2000 generations under conditions where selection was strong early in life, but weak late in life. This selection had the effect of increasing the age of first reproduction and faster growth rates, but led to the unexpected evolution of slower aging. However, late acting deleterious mutations did invade and spread in populations.

In the Aging Cell paper, Ackermann et al. construct simple models to show why organisms might evolve aging, and test these models using age-specific performance data of C. crescentus to test the assumptions of the models. C. crescentus cell division is assymetric, and results in a sessile stalked cell and a motile swarmer cell. Ackermann et al.'s results showed that rate of cell division (hence fitness) declined with age for stalked cells, presumably because of the accumulation of damage in the stalked cell. Naturally it would have been nice to see what happened to the motile cells, but unfortunately this data is quite difficult to obtain. Nonetheless, the implication is there that the assymetric cell division results in the partitioning of damage between 'parent" and "offspring" cells.

At some point in the history of life, aging must have evolved (I've covered some of this theory in an earlier post, but a wonderful resource is senescence.info). Presumably the segregation of older material into one individual (termed 'parent') had to begin somewhere, and it would be nice to know the factors that led to this phenomenon. Bacteria may be the ideal model to explore this event. I'm looking forward to more studies in the near future.

Photo From: Aging and Death in E. coli PLoS Biology Vol. 3, No. 2, e58 doi:10.1371/journal.pbio.0030058

This Week's Citation Classic

After a short hiatus for vacation and a conference, This Week's Citation Classic returns with this gem from Marshall Nirenberg and Heinrich Matthaei:

Nirenberg, Marshall W., and J. Heinrich Matthaei. "The Dependence of Cell-Free Protein Synthesis in E. Coli Upon Naturally Occuring or Synthetic Polyribonucleotides." Proceedings of the National Academy of Sciences 47, 10 (October 1961): 1588-1602.

Following Oswald Avery's and Hershey and Chase's demonstration that DNA was the genetic material and Watson and Crick's solution of its structure, the race was on to determine how DNA was translated into proteins. Upstart Biochemist PI Marshall Nirenberg, fresh from a NIH postdoc in Dr. DeWitt Stetten, Jr.'s laboratory, and his first postdoc Heinrich Matthaei, conducted what was to be known as the poly-U experiment. (Many of Nirenberg's colleagues felt that it was naive for a biochemist untrained in molecular genetics to commence a brand-new area of research; at least one felt that Nirenberg was committing "professional suicide.") However, as he wrote in his laboratory notebook, Nirenberg realized that he “would not have to get polynucloetide synthesis very far to break the coding problem. Could crack life's code!”

As are many of the most important experiments in scientific history, the experiment was simple and elegant: Nirenberg and Matthaei used a mortar and pestle to grind up E. coli cells to obtain a cell-free extract. To the extract, they added a synthetic RNA, poly-uracil (UUU), and some DNase to dissolve any DNA present. The experiment used 20 test tubes, each filled with a different amino acid. For each individual experiment, 19 test tubes contained an unlabeled “cold” amino acid and one contained a "hot" radioactively tagged amino acid so they could watch the reaction. The “hot” amino acid was changed each time they did the experiment. In the experiment where the "hot" amino acid was phenylalanine, the results were "spectacular"! After an hour, the control tubes showed a background level of 70 counts per milligram of protein, whereas the "hot" phenylalanine tube showed 38,000 counts per milligram, thus demonstrating that UUU coded for phenylalanine.

In August 1961, Nirenberg presented the results to a small group of about thirty scientists at the International Congress of Biochemistry in Moscow. Francis Crick, who was among the original thirty, arranged to have Nirenberg deliver his paper again, this time to an audience of ~1000 people. It is not an exaggeration to suggest that Nirenberg and Matthaei's results were a worldwide sensation and made Nirenberg an instant scientific celebrity. For example , in January 1962, the Chicago Sun-Times announced that "No stronger proof of the universality of all life has been developed since Charles Darwin's 'The Origin of Species' demonstrated that all life is descended from one beginning. In the far future, the hope is that the hereditary lineup will be so well known that science may deal with the aberrations of DNA arrangements that produce cancer, aging, and other weaknesses of the flesh." In January 1962, Nirenberg joked to Crick, "[T]he American press has been saying that [my] work may result in (1) the cure of cancer and allied diseases (2) the cause of cancer and the end of mankind, and (3) a better knowledge of the molecular structure of God. Well, it's all in a day's work."

Nirenberg later shared the 1968 Nobel Prize with Robert Holley and Gobind Khorana.

NIH has a nice exhibit on Deciphering the Genetic Code. The photo of the vial of poly-U used in the experiment was taken from that exhibit.

Larry Moran at Sandwalk also posted on this experiment. Hopefully he will remove it now :P