What I did last summer (and year)....

Earlier in these pages I gave a brief background of what I've been doing for the past year. I've been interested in how differences in mRNA production and differences in holin structure led to variability in lysis timing. Max Delbruck was the first to consider this question; he looked at variation in the number of babies a single phage produces and found a great deal of variation in this important life history trait. Delbruck's method, diluting a suspension of infected bacteria such each dilution contained on average less than one bacterial cell, and then plating all the (hundreds of) dilutions, is especially onerous, and perhaps explains why his work was not followed up.

(i.e. production rate, and observe the changes in Moreover, his work leaves open the questions of what causes the variation and whether it is evolutionarily significant. My work follows up on Delbruck's, except that I chose to look at a different, but closely related life history trait, lysis time. Earlier work has shown that there is a direct correlation between lysis time and burst size: the longer the phage waits to lyse the cell, the more babies it can produce. So it makes a good proxy for burst size, as well as an interesting life history trait in its own right. Plus, we now have the genetic techniques to manipulate the phage's genome and dissect the causes of variation: to wit, the ability to manipulate the holin protein structure and to alter the strength of the promoter that controls holin production. All we need is a way of observing lysis time for individual cells. Enter the microscope-mounted perfusion chamber! Here is the setup:

The way this works is you place a glass cover slip on the bottom, a cell-binding agent (poly-lysine) and some E. coli infected with lysogenic phage, then place another cover slip on top. The cells form a single layer on the bottom cover slip. The whole unit is placed on a heating platform, under a normal light microscope, and tubes are attached to allow a flow of nutrient broth thru the chamber. The heating platform generates a heat spike, inducing the phage to begin the lysis process. With a microscope mounted camera, I filmed the bound cells following the heat spike until they lysed, then recorded the time of lysis for each cell.

Given are the variation in lysis time for the wild type (JJD3), the most variable altered holin sequence genotype (JJD9) and the most variable altered promoter genotype (SYP028). This histogram (below) clearly shows the greater variability (standard deviation, SD, in lysis time for the latter two genotypes.

In general, SD increases with the mean lysis time (below, P = 0.0087). This result is consistent with the idea that, on average, it takes a longer time for a weak holin-holin interaction to attain the critical holin raft size that is necessary for hole formation. Furthermore, the the timing of attainment also varies widely among individual infected cells. That is, we should expect to observe a positive relationship between the mean lysis times and their standard deviations.

Greater promoter strength leads to a shorter lysis time (P = 0.0065). Interestingly, the lysis time variation seems to show a threshold effect. Within a range of promoter strength (between 150 to 250), different mean lysis times showed a similar stochasticity (P = 0.9593). However, once the promoter strength was dropped to a low level, the stochasticity increased dramatically.

Finally I showed an environmental component to lysis time stochasticity by reducing the nutrients provided to the host cells. Expression of the phage holin protein requires the host synthesis machinery, including the RNA polymerases (RNAPs) for transcription, ribosomes for translation, and many other raw materials for protein synthesis and enzymes for modifications. In general, cells growing at a higher rate will have a higher concentration of the synthesis machinery. Here as growth rate increases, the stochasticity (SD) decreases (P = 0.0346), demonstrating that host physiology can greatly influence the outcome of a viral infection.

The commonly invoked causes for cellular stochasticity are the random events of gene transcription, translation, and degradation of the expressed protein. Besides the usual causes, there is another layer of random event for holin hole formation, namely, the association and disassociation of holin raft on the cell membrane. However, in this preliminary study, I was not able to attribute the relative importance of each cause for the observed lysis time stochasticity. If the main cause is due to host biochemistry and physiology, then it would be difficult for the phage to reduce stochasticity. On the other hand, if the main cause is due to the amount of holin production or holin-holin interaction, then it is possible for mutations to change the promoter strength or holin sequence to increase or reduce the level of stochasticity.

Whether the observed lysis time stochasticity is evolutionarily significant remains to be determined; however it is conceivable that selection can favor genotypes with greater or lower levels of phenotypic stochasticity depending on the circumstances. Future experiments will address (1) whether variation in lysis time stochasticity translates into variation in fitness; (2) whether genotypes expressing greater or lower levels of lysis time stochasticity can be selected for; (3) whether similar patterns of stochasticity exist for different phage species.