Caenorhabditis elegans Life Span Studies: The Challenge of Maintaining Synchronous Cohorts

Introduction

T he nematode Caenorhabditis elegans with its short life span of only around 20 days has been increasingly adopted as a convenient way to explore the life span effects of drugs, antioxidants, and other compounds in vivo. ^{1 –5} This increasing popularity is illustrated by the accelerating increase in the publication number of C. elegans life span studies. For instance, a keyword search of the PubMed database (National Center for Biotechnology Information, National Library of Medicine) determining the number of references containing the terms “Caenorhabditis elegans” and “life span” for each year between 1980 and 2009 reveals more than twice as many such publications in the 4 years between 2005 and 2009 than for the entire 25-year period before 2005. This increasing popularity means that many researchers with little or no prior experience in this model organism embark on life span studies in this organism. Although the C. elegans system has many advantages apart from its short life span, including its morphological simplicity and relative ease of maintenance, ^{6 –8} the apparent conceptual simplicity of life span studies can be misleading. For more detailed discussions regarding some challenges faced in this context, see refs. 2, 9, and 10. Here we discuss a specific problem associated with synchronization methods often used in life span studies.

C. elegans has a rapid life cycle and produces a very large number of offspring during a short fertile period in early adulthood. Maturity is typically reached 3–4 days after hatching, and up to 300 progeny distributed over the next 4–6 days are typical for wild-type animals under standard conditions. ⁶ This poses a significant challenge in the context of life span studies because the aging cohort can be rapidly overcrowded by offspring if fecundity is not curtailed. A common approach for this purpose is the use of fertility mutants. These mutants produce progeny normally at a permissive temperature but are sterile at higher temperatures. ¹¹

Adult nematodes are postmitotic apart from their germ line, and an alternative approach for the control of offspring is the use of the drug 5-fluoro-2′-deoxyuridine (FUdR). ^12,13 FUdR inhibits DNA synthesis and therefore selectively affects dividing cells of the germ line and in the developing larvae, making it useful as a form of contraceptive. ^8,12,13

Using either fertility mutants or FUdR, large batches of synchronized worms can be grown relatively easily. However, there have been suggestions that some temperature-sensitive mutants are not entirely sterile but can produce up to 0.1 progeny per adult. ^11,14 Similarly, contraception by FUdR is not always perfectly effective, especially in the context of drug screening studies where interactions of FUdR with test compounds are possible. ¹⁰

Whereas 0.1 progeny per adult represents only <0.05% of baseline fecundity, it should be noted that at this level a cohort of 100 animals would eventually contain up to 10 second-generation animals (∼10%). This problem is particularly worrying considering that maximum life span is often determined as the mean of the longest living 10% of the cohort. Hence, this effect could lead to significant artefacts when biomarker studies are extended to higher ages.

Materials and Methods

To quantitatively evaluate the level of background fecundity that might be acceptable, we have performed a simple in silico experiment using Monte Carlo analyses. ¹⁵ Two tests were performed; in the first test we determined the mean and maximum life span distribution of 10,000 independent in silico populations of 500 worms each. Survival time was based on an idealized normal life span distribution derived from actual experimental data. For each individual worm in each population, two uniform pseudo-random numbers were generated. The first random number determined if progeny were produced by the adult worm under consideration. The second number determined the day at which this offspring, if any, hatched. The latter was determined on the basis of the inverse transformation method based on the cumulative density function of the number of progeny per adult wild-type worm constructed, again, from experimental data. ¹⁶ Three different progeny frequencies (f ) were considered (f = 0.1, 0.05, and 0.01 progeny per adult worm). On the basis of its hatching day, the day of death was calculated for each progeny (using the same procedure as for the adult animals in the cohort) and this additional death event was appended to the adult cohort population data. The mean and maximum life span of each cohort with and without progeny was then calculated. Maximum life span was determined as the average life span of the longest-lived 10% of a worm population. Two types of hypothesis testing procedures were used to evaluate the significance of life span differences: (1) two sample t-test for both mean and maximum life span and (2) Kaplan–Meyer (logrank) test comparing life span curves (Fig. 1G). All simulations were performed in MATLAB (R2008b).

OPEN IN VIEWER

FIG. 1.

Statistics related to the progeny artefact. (A–C) Comparison of the mean life span of 10,000 in silico worm population comprising 500 worms each with or without progeny at frequencies of 0.1, 0.05, and 0.01 per adult. (D–F) Comparison of the maximum life span in the same set. (G) Relevant test statistics of the two hypotheses tests performed. The p value listed is the average of all 10,000 comparisons performed. The frequency indicates the number of individual tests of the 10,000 total that showed significant differences (p < 0.05).

Results

Figure 1, A–F, illustrates the comparison of the histograms of mean and maximum life span values obtained from the 10,000 independent populations with and without progeny at the three progeny frequencies. As expected, our data indicate that as the progeny frequency increases, the mean value of the distribution shifts from the corresponding control mean. The statistical significance of progeny on mean and maximum life span was evaluated for each of the 10,000 sets of populations (Fig. 1G). For the lowest value of f = 0.01 offspring per adult, both the hypotheses tests failed to indicate a significant population difference for either mean or maximum life span, meaning that the effect of offspring is undetectable in a typical population of 500 animals. However, 34% of maximum life span comparisons were significantly affected at f = 0.05 and nearly all (93%) of maximum life span comparisons were confounded at f = 0.1. At this level of fecundity, 18% of life span curves were significantly different (as judged by logrank test) and even 8.5% of mean life span results were significantly confounded by progeny effects. It should be noted that this level of f = 0.1 is close to the level of offspring that can be expected for one of the more commonly utilized fertility mutants. ^{11, 14}

Discussion

This analysis clearly illustrates that even relatively low levels of proliferation and carryover of offspring have the potential to affect both baseline aging studies (e.g., biomarkers) and related intervention studies. Some simple measures can be taken to ameliorate the risk of this artefact. Whatever method for control of offspring is used, it is important to establish its effectiveness under realistic conditions by carefully monitoring of the number of offspring generated under the actual assay conditions used. Although this might not always be possible during experiments, e.g., for very large cohorts, it should be possible to determine baseline levels in preliminary tests or to carefully monitor a subset of the main study. Fertility mutant strains differ in the amount of offspring they produce and double fertility mutants have been generated that have very low levels of residual offspring. These might be preferable if the choice of mutant strain is not otherwise constrained. ^11,14 Dose and timing of FUdR treatment can be adjusted to prevent low-level proliferation, and procedures for manual transfer can be optimized to improve confidence in their effectiveness (e.g., perform daily transfers rather than every other day, check for carry over of offspring between transfers). Even if FUdR or fertility mutants are used, additional manual transfers can sometimes be implemented to further reduce the risk of carrying forward any offspring.