Counting Caenorhabditis elegans: Protocol Optimization and Applications for Population Growth and Toxicity Studies in Liquid Medium

The nematode Caenorhabditis elegans is used extensively in molecular, toxicological and genetics research. However, standardized methods for counting nematodes in liquid culture do not exist despite the wide use of nematodes and need for accurate measurements. Herein, we provide a simple and affordable counting protocol developed to maximize count accuracy and minimize variability in liquid nematode culture. Sources of variability in the counting process were identified and tested in 14 separate experiments. Three variables resulted in significant effects on nematode count: shaking of the culture, priming of pipette tips, and sampling location within a microcentrifuge tube. Between-operator variability did not have a statistically significant effect on counts, even among differently-skilled operators. The protocol was used to assess population growth rates of nematodes in two different but common liquid growth media: axenic modified Caenorhabditis elegans Habitation and Reproduction medium (mCeHR) and S-basal complete. In mCeHR, nematode populations doubled daily for 10 d. S-basal complete populations initially doubled every 12 h, but slowed within 7 d. We also detected a statistically significant difference between embryo-to-hatchling incubation period of 5 d in mCeHR compared to 4 d in S-basal complete. The developed counting method for Caenorhabditis elegans reduces variability and allows for rigorous and reliable experimentation.


Nematode and E. coli OP50 Culture Details
E. coli OP50 was cultured in Lennox L. Broth Base (Invitrogen) for 17 h at 37 °C (with shaking), washed twice (centrifugation at a fixed angle at 2000 x g for 10 min) and re-suspended in C. elegans M9 growth buffer (medium composition described below). The OP50 is stored at 4 °C for up to one week (microbes start to die and aggregate after one week). A volume of 100 μL OP50 suspension was spread onto agar Nematode Growth Medium (NGM, recipe in SI) plates and incubated overnight at room temperature (approximately 24 °C). "Chunks" of C. elegans on agar were transferred from the stock plate to NGM OP50 plates and incubated in a dark incubator at 21 °C for a period of 5 d to 7 d. Once nematodes consumed the E. coli and cleared the plates, chunks of the NGM nematode culture were again transferred to fresh OP50 plates. The "chunking" process was repeated 4 to 5 times to establish healthy, gravid nematode populations. Nematodes were monitored with a light microscope at 40X magnification.
After they had consumed most of the OP50 lawn, 5 mL M9 was added to the place and a sterile bacteria spreader was used to gently dislodge the eggs; egg and medium slurry was pipetted into a centrifuge tube and bleached 1 to obtain a sterile solution of nematode eggs. Eggs were washed twice with sterile water and re-suspended in the two different nematode growth media. mCeHR 1 (recipe in SI) was supplemented with 10 % or 20 % volume fraction fat-free milk (ultra-pasteurized; opened in sterile hood only) + 100 μg/mL tetracycline hydrochloride (Sigma). S-basal complete (SI) was supplemented with 10 % volume fraction OP50 suspension with or without 1 % Penicillin-Streptomycin-Amphotericin B and 0.5 % Amphotericin B (MP Biomedicals). Buffer and medium protocols are described in detail in the SI. It is important to note, culture flasks must be vented for adequate air supply; when swirling/shaking flasks, take care to not wet the vent with culture medium so as to not contaminate cultures.

Variability studies
Data from the counting protocol experiments were analyzed using a generalized mixed effects The simplicity of this expression facilitates easy interpretation and presentation, and for this reason modeling results are discussed in terms of median counts.
In the first implementation of the model described above (henceforth referred to as "protocol model 1"), random effects attributed to the slide were modeled as independently following a normal distribution with mean 0 and standard deviation [i.e., ( )~n ormal(0, )]. Random error terms for counts occurring on day were modeled as independently following a normal distribution with mean 0 and standard deviation [i.e., ~normal(0, )], allowing the modeled degree of overdispersion to vary day to day. The median nematode concentration as observed under the standard counting protocol for day ( ) was assigned a uniform "prior" from 0 to 40 nematodes per dot. For clarity, a prior is a Bayesian probability distribution for a quantity (i.e. number of nematodes per dot) that is not supported by defined experimental observations (data). The dot location effects were assigned independent normal prior distributions with mean 0 and standard deviation 0.5, and were normalized such that ∑ 10 =1 = 0. The prior assigned to the effect of the counting protocol ( ) was a normal distribution with mean 0 and standard deviation 4. The standard deviations of random slide effects and the random error terms were assigned independent exponential prior distributions with mean 0.5 [i.e., ~ exponential(0.5) and ~ exponential(0.5)].
In the second implementation of the model (henceforth referred to as "protocol model 2"), random effects for slide were modeled as independently following a double exponential distribution with mean 0 and standard deviation [i.e., ( )~d ouble exponential(0, )]. Random error terms for counts occurring on day were modeled as independently following a double exponential distribution with mean 0 and standard deviation [i.e., ~ double exponential (0, )], allowing the modeled degree of overdispersion to vary day to day. The median nematode concentration as observed under the standard counting protocol for day ( ) was assigned a uniform prior from 0 to 40 nematodes per dot. The prior assigned to the effect of counting protocol ( ) was a uniform distribution from -5 to 5. The dot location effects were assigned independent uniform prior distributions from -0.5 to 0.5, and were normalized such that ∑ 10 =1 = 0. The standard deviations of random slide effects and the random error terms were assigned independent uniform prior distributions from 0 to 1 [i.e., and ~ uniform(0,1)].
A separate analysis was conducted on data obtained from two inter-operator variability studies. Let represents the log ratio of the median concentration found in dot location to the average of the median concentrations across all ten locations (parameterized such that ∑ 10 =1 = 0; these effects were only included when modeling data from the first inter-operator variability study); ( ) denotes a random effect influencing the concentration of all dots on slide (nested within each operator) by a multiplicative factor of exp( ( ) ) (these effects were only included when modeling data from the first inter-operator variability study); and 1 and 2 are random error terms unique to a single dot allowing for overdispersion among observed counts.
In the first implementation of the model for inter-operator variability data (henceforth referred to as "operator model 1"), random effects attributed to operator and slide were modeled as following a normal distributions with means 0 and standard deviations and , respectively [i.e., ~normal(0, ) and

Nonparametric analysis of dot location effects
We considered the set of nematode counts across the collection of 63 slides analyzed during the variability source examination phase where the first dot was placed in spot 1 (but spanning various settings of the other controlled factors). To examine whether the observed variability between spots was statistically significant, the following leave-one-out comparison was conducted. For each slide, we determined whether the count from spot 1 was greater than, less than or equal to the collective average count from the other 9 spots on the same slide. The results across all 63 slides were evaluated using a sign test, which only requires the assumption that the observed counts are independent from one slide to another. The same process was repeated with interest focusing on each of the 10 spots in turn. The results of this test, summarized in Table   S8, suggest that the average counts obtained from spots 1 and 10 had a statistically significant tendency to be lower than the average count of the other spots. The sign test did not facilitate the estimation of the effect size in terms of influence on median number of nematodes.

Growth rate experiments
Nematode count data from the two population growth experiments were analyzed from the perspective of exponential population growth rate and absolute count. so that normalized counts from slides with greater dilution fraction were less influential.
The average counts described above were used to estimate growth rates in the following manner.
The population growth rate exhibited in experiment for passage of treatment condition between consecutive counting sessions occurring on days and ′ was characterized by the following equation: The resulting growth rate was translated into the corresponding time required for a population to double in size according to: where ∆ ′ provides an estimate of the number of days required for population number to double in experiment for passage of treatment condition , according to the average growth rate observed between consecutive counting sessions occurring on days and ′ .
Analysis of the growth experiment data is predominantly exploratory due to the longitudinal nature of population growth observations and the relatively small number of experimentally independent replicates. In such instances, the results of a model-focused analysis are likely to be heavily influenced by modeling choices and assumptions, rather than information present in the data itself.
S-basal growth medium amount reagent 5.9 g NaCL 1 g K2HPO4 6 g K2HPO4

S-basal complete
To make 0.

Modified C. elegans Habitation and Reproduction Medium (mCeHR)
To make 1 L mCeHR base: -Add the choline, vitamin and growth factor mix, i-inositol, hemin chloride and 250 mL sterile water to a 1 L sterile filter bottle in a sterile biological hood. -Apply suction to filter -Add nucleic acid mix, mineral mix, lactalbumin hydrosylate, 50 X MEM Amino Acids Solution (Gibco), 100 X MEM Non-Essential AA Solution (Gibco), KH2PO4, d-Glucose and HEPES sodium salt with 256 mL autoclaved water; filter. -Add cholesterol in ethanol without filtration.
Final volume ratio of mCeHR is 80% base to 20% non-fat milk. Query: Do differences in the sizes of the pipette tips used for transferring nematodes between containers result in a bias in the nematode counts (e.g. small pore pipette tips might potentially exclude large nematodes/adults from being counted)? Test: Step 1: From the primary container (culture flask), sample aliquots were transferred (using a standard 10 µL primed pipette tip) onto glass microscope slides and the culture was counted using the counting protocol.
Step 2: From the primary container (culture flask), a 1000 μL primed pipette tip was used to transfer a 1000 µL aliquot of nematode culture into a secondary container (culture flask) that was then diluted 150X with Millipore water. The primed 1000 μL pipette tip was used to transfer 1000 µL of culture into 6 wells of a 24-well microtitre plate and the plate was heated at 80 °C to heat-kill the nematodes. Each well was independently imaged using a CoolSNAPHQ2 CCD camera (Photometrics, Tucson, AZ) coupled to an automated Zeiss microscope (Axio Vert.A1, Carl Zeiss Microscopy, Oberkochen, Germany) with Zen software (Carl Zeiss Microscopy, 2012 Blue Edition). The microscope was calibrated using a stage micrometer (Electron Microscopy Services) at 5X. Images were exported as .tiff files and viewed in ImageJ. Nematode counts obtained using the counting protocol were compared to nematode counts obtained using the 1000 µL pipette tip.

2
Absence or presence of culture shaking and type of shaking Settling bias Query: Does settling of the nematodes in the culture flask result in a bias in the nematode counts? Test: A nematode culture was gently removed from a 20 °C incubator and placed on a laboratory bench top for 5 min to settle. The culture was sampled and counted with no shaking and with minimal movement of the flask. The flask was then re-counted using Shake Style A (vigorous swirling of the flask three times clockwise and then three times counter clockwise, with the flask oriented upright and the "clock" positioned on the floor). The nematode culture was allowed to settle for 5 min, and was re-counted with Shake Style B (gentle rocking of the flask back and forth, three times in each direction, with the flask oriented upright and the rocking occurring from left to right). Flasks were gently rocked between sampling aliquots for both shaking styles. Query: Does transferring sample aliquots using unprimed, new or primed ("old") pipette tips or transferring from a primary container to a secondary or tertiary container induce a bias in the nematode counts? Test 1: From the primary nematode container (culture flask), a 300 μL sample aliquot was transferred into a 1.7 mL microcentrifuge tube (secondary container). A 200 μL sample aliquot was then removed from the secondary container and transferred into a second microcentrifuge tube (tertiary container). Nematode counts obtained in secondary and tertiary containers were compared to nematode counts obtained in the primary container. Test 2: From the primary container (culture flask), sample aliquots were transferred onto glass microscope slides with (1) new, unprimed pipette tips, (2) new, primed pipette tips and with (3) old (re-used), primed pipette tips. New, unprimed pipette tips were used straight out of the box. New, primed pipette tips were primed by pipetting up and down in growth culture four times.

4
Counting nematodes in different locations within a single container Sampling location bias Query: Does a specific sampling location within a container result in a bias in the nematode counts? Test: From the primary container (culture flask), sample aliquots from the middle of the flask were transferred onto glass microscope slides and the nematodes were counted using the counting protocol. Then, a 1000 µL sample aliquot was transferred to a 1.7 mL microcentrifuge tube (tube was mixed by vortex). Sample aliquots were taken from the top, middle and bottom portions of the total sample volume in the microcentrifuge tube. Nematode counts obtained in the flask were compared to the nematode counts obtained in the 3 different locations within the microcentrifuge tube.
5 Inherent counting differences between people Person-to-person bias Query: Is there a significant difference between individual technicians using the same counting protocol that results in a bias in the nematode counts? Test: From a single primary container (flask), three different individuals determined the nematode counts in the flask using the counting protocol; the flask contained a nematode culture prepared by a non-counting operator who hand-selected and transferred a specified number of nematodes into a known volume of culture medium. In addition, four different individuals counted a nematode culture with an unknown nematode concentration. Test: From the primary nematode container (culture flask), sample aliquots composed of 1 x 300 μL, 3 x 100 µL (into one tube) and 3 x 50 µL (into one tube) were transferred into separate secondary containers (1.7 mL microcentrifuge tubes) and the nematodes in each secondary container were independently counted. Nematode counts obtained in secondary containers were compared to nematode counts obtained in the primary container after correcting for dilution.

7
Order of sample aliquots (dots) on microscope slide Location bias Query: Does the placement order or location of the 2 µL sample aliquot (dot) on the microscope slide induce a bias in the nematode counts? Test: Nematode cultures were sampled from flasks and counted with alternating dot placement ( Figure 1). First, the culture was counted with the counting protocol (dot placement from spot 1 to spot 10). Next, the culture was re-counted with dots placed on slides from spot 6 to spot 10 and from spot 1 to spot 5. Three slides were prepared for each ordering sequence. The set of nematode counts from all 42 slides evaluated during the variability source examination phase of the study (dot placement from spot 1 to spot 10) were included in the final statistical analysis; these 42 slides spanned various settings of other controlled factors.
Supplementary Tables S2 through S7 and Table S9  0.007 *The proportion listed in row corresponds to fraction of slides for which the count observed at dot position was greater than the average count across the other nine positions on the same slide. If dot location had no effect on nematode count, one would expect this value to vary randomly around 0.5. Confidence intervals from dot positions 1 and 10 fall entirely below 0.5, which indicates a potential negative bias. a/ Out of a total of 63 measurements. Figure 1. Schematic of for Sampling Nematodes from Liquid Culture. The sampling protocol for counting nematodes in liquid culture is briefly described. The three most important aspects of sampling are italicized in the text and emphasized in red in the figure. These include swirling the sample, priming the pipette tip in the culture medium and removing aliquots from the middle of the culture (as opposed to the top). We also recommend only submersing the tip about halfway into the medium, and not allowing the pipette shaft to come into contact with the medium to avoid contamination.

FIGURES Supplementary
Dilutions are recommended for nematode cultures that are too dense to count easily (cultures that contain ≥ 15 to 20 nematodes per μL or ≥ 30 to 40 nematodes per dot). It is strongly recommended that cultures are kept at < 20 nematodes per μL in order to ensure optimal health of the culture.