Gut microbes may positively or negatively impact the health of their host via specific or conserved mechanisms. Caenorhabditis elegans is a convenient platform to screen for such microbes. The present protocol describes high-throughput screening of 48 bacterial isolates for impact on nematode stress resistance, used as a proxy for worm health.
With its small size, short lifespan, and easy genetics, Caenorhabditis elegans offers a convenient platform to study the impact of microbial isolates on host physiology. It also fluoresces in blue when dying, providing a convenient means of pinpointing death. This property has been exploited to develop high-throughput label-free C. elegans survival assays (LFASS). These involve time-lapse fluorescence recording of worm populations set in multiwell plates, from which population median time of death can be derived. The present study adopts the LFASS approach to screen multiple microbial isolates at once for the effects on C. elegans susceptibility to severe heat and oxidative stresses. Such microbial screening pipeline, which can notably be used to prescreen probiotics, using severe stress resistance as a proxy for host health is reported here. The protocol describes how to grow both C. elegans gut microbiota isolate collections and synchronous worm populations in multiwell arrays before combining them for the assays. The example provided covers the testing of 47 bacterial isolates and one control strain on two worm strains, in two stress assays in parallel. However, the approach pipeline is readily scalable and applicable to the screening of many other modalities. Thus, it provides a versatile setup to rapidly survey a multiparametric landscape of biological and biochemical conditions that impact C. elegans health.
The human body harbors an estimated 10-100 trillion live microbial cells (bacteria, archaea fungi), which are primarily found in the gut, skin, and mucosal environments1. In a healthy state, these provide benefits to their host, including vitamin production, maturation of the immune system, stimulation of innate and adaptive immune responses to pathogens, regulation of fat metabolism, modulation of stress responses, and more, with an impact on growth and development, disease onset, and ageing2,3,4,5. The gut microbiota also evolves considerably throughout life. The most drastic evolution occurs during infancy and early childhood6, but significant changes also occur with age, including a decrease in Bifidobacterium abundance and an increase in Clostridium, Lactobacillus, Enterobacteriaceae, and Enterococcus species7. Lifestyle can further alter gut microbial composition leading to dysbiosis (loss of beneficial bacteria, overgrowth of opportunistic bacteria), resulting in various pathologies such as inflammatory bowel disease, diabetes, and obesity5, but also contributing to Alzheimer's and Parkinson's diseases8,9,10,11.
This realization has critically contributed to refining the concept of the gut-brain axis (GBA), where interactions between gut physiology (now including the microbes within it) and the nervous system are considered the main regulator of animal metabolism and physiological functions12. However, the precise role of microbiota in gut-brain signaling and the associated mechanisms of action are far from being fully understood13. With gut microbiota being a key determinant of healthy aging, how bacteria modulate the aging process has become a subject of intense research and controversy6,14,15.
With the demonstration that the roundworm Caenorhabditis elegans hosts a bonafide gut microbiota dominated-as in other species-by Bacteroidetes, Firmicutes, and Actinobacteria16,17,18,19,20, its rapid rise as an experimental platform to study host-gut commensal interactions21,22,23,24,25,26 has significantly expanded our investigative arsenal26,27,28,29. In particular, high-throughput experimental approaches available for C. elegans to study gene-diet, gene-drug, gene-pathogen, etc. interactions, can be adapted to rapidly explore how bacterial isolates and cocktails impact C. elegans health and aging.
The present protocol describes an experimental pipeline to screen at once arrays of bacterial isolates or mixtures set in multiwell plates for effects on C. elegans stress resistance as a proxy for health, which can be used to identify probiotics. It details how to grow large worm populations and handle bacterial arrays in 96- and 384-well plate formats before processing worms for automated stress resistance analysis using a fluorescence plate reader (Figure 1). The approach is based on label-free automated survival assays (LFASS)30 that exploit the phenomenon of death fluorescence31, whereby dying worms produce a burst of blue fluorescence that can be used to pinpoint the time of death. Blue fluorescence is emitted by glucosyl esters of anthranilic acid stored in C. elegans gut granules (a type of lysosome-related organelle), which burst when a necrotic cascade is triggered in the worm gut upon death31.
Figure 1: Experimental workflow for high-throughput screening of bacterial isolates with impact on C. elegans resistance to stress. (A) Timeline for worm and bacterial maintenance and assay setup. (B) 96-well bacterial plate array setup and handling. (C) 384-well worm plate setup. Please click here to view a larger version of this figure.
The two C. elegans strains used in parallel for the present study were Bristol N2 wild type and HT1890: daf-16(mgDf50), which grow at similar rates. However, the protocol can be replicated with any combination of two strains that have similar growth rates. Note that, when testing other strains in parallel (for instance, wild type and slow-growing daf-2 mutants), differing growth rates must be considered, and accordingly, the protocol needs to be adjusted. The timescales and quantities of worms and bacteria in the following protocol are optimized for parallel testing of 48 bacterial isolates on two worm strains in two LFASS assays in tetraplicates. Adjustments will be needed if more conditions are to be tested in parallel. Escherichia coli OP50 bacteria strain was obtained from the Caenorhabditis Genetics Center (CGC), University of Minnesota. The 48 bacterial isolates were obtained from the Schulenburg lab and maintained on LB agar.
1. C. elegans culturing on OP50 (Days 1 – 8)
NOTE: The current approach aims to grow C. elegans hermaphrodites on a solid medium at all stages and avoids unnecessary dietary changes (i.e., using alternate faster-growing E. coli strains such as NA22 or richer growth media such as egg plates) to remain as close as possible to the standard growth conditions32,33 that are still widely used. Worm growth temperature (here set at 15 °C) depends on the C. elegans strain(s) used and may need adjusting (for instance, to avoid or trigger the expression of a temperature-sensitive phenotype or biomarker). For information on worm husbandry, please see reference33.
2. Maintenance of gut microbiota isolate collections (Day 9)
3. Growing large C. elegans cultures (Day 10)
4. Preparing gut microbiota isolate collections for re-feeding worms
5. LFASS heat shock and oxidative assay setup (Days 13 – 14)
6. Plate-reader data handling
7. Data inspection
8. LFASS data processing
NOTE: Details are provided at https://github.com/ABA80/LFASS and in the supplementary materials of Reference30.
LFASS assays provide robust, high-throughput, and rapid screening of multiple test conditions at once, such as screening numerous genetic and microbiota parameters that contribute to stress resistance and aging. It only takes 2-3 weeks for the experiment to acquire an extensive dataset of multiple test conditions. L4 + 36 h adult wild-type worm populations were exposed to 42 °C thermal stress and 7% t-BHP-induced oxidative stress after a 36 h culture on 48 gut microbial isolates for 36 h. The assay was performed four times, with each condition replicated four times in each assay. Across all the conditions tested, the median time of death varied between 40-130 min for the thermal stress assay and between 90-240 min for the t-BHP induced oxidative stress assay. Early adulthood thermal stress assays usually display more consistent results and less inter-day and intra-day variability than oxidative stress assays and are better predictors of subsequent longevity30. The difference in the median time of death of worms fed on different microbial diets conclusively exemplifies how gut commensals impact host stress resistance. Figure 2 shows typical results from 16 biological replicates for Bristol N2 wild-type worms on two gut microbiota isolates of interest and the standard laboratory strain E. coli OP50.
Figure 2: Representative results from paired heat/oxidative stress resistance assays from worms grown on two bacterial isolates and a control strain. (A) Heat stress assay. Wild-type worms were fed for 36 h on either MYb11 (Pseudomonas lurida) or MYb115 (Pseudomonassp.) bacterial isolates or OP50 control bacteria. MYb11 led to a significantly earlier median time of death (p < 0.001). (B) Oxidative stress assay. Wild-type worms were fed for 36 h on either MYb11 or MYb115 bacterial isolates or OP50 control bacteria. MYb11 led to a significantly earlier median time of death (p < 0.05). 120-150 worms were analyzed in four sets of samples, and each experiment was performed in quadruplicate. Wells with poor quality/poorly fitted data led to missing values. Box plot representations indicate single well values, minimum, median, and maximum values. *Indicates statistical significance at p < 0.05 and ***p < 0.001, with one-way ANOVA. Please click here to view a larger version of this figure.
A | |||
Parameter | Setting | Comment | |
Temperature | 25 °C | ||
Fluorescence excitation | 365 nm | 10 nm bandwidth | |
Fluorescence emission | 430 nm | 20 nm bandwidth | |
Number of flashes | 8 flashes, 1ms settle time | ||
Gain | 100 to 130 | To be adjusted depending on sample. Aim for 2-5 k read value at start of assay | |
Lag time | 1 μs | ||
Integration time | 30 μs | ||
Read time interval | Every 2 min | Depending on stress severity, duration and time interval may need adjusting (8 h is enough at 7% t-BHP) | |
Duration | 8-12h | ||
Read direction | From below | ||
B | |||
Parameter | Setting | Comment | |
Temperature | 42 °C | ||
Fluorescence excitation | 365 nm | 10 nm bandwidth | |
Fluorescence emission | 430 nm | 20 nm bandwidth | |
Number of flashes | 8 flashes, 1ms settle time | ||
Gain | 100 to 130 | To be adjusted depending on sample. Aim for 2-5 k read value at start of assay | |
Lag time | 1 μs | ||
Integration time | 30 μs | ||
Read time interval | Every 2 min | Depending on stress severity, duration and time interval may need adjusting (6 h is enough at 42 °C) | |
Duration | 6-12h | ||
Read direction | From below |
Table 1: Oxidative and heat stress assay settings. Examples of settings used for oxidative (A) and heat (B) stress assays on the fluorescence plate readers used in this study.
Supplementary File 1: Media, buffer, and culture recipes. Composition of different media, buffers, and cultures used in the present study. Please click here to download this File.
Supplementary File 2: Raw fluorescence data from the plate reader after being exported and converted to excel .xls format. (A) The excel file shows the raw fluorescence data for a heat shock assay. Raw fluorescence data is shown at 2 min time interval. (B) The column with "well position" on the 384-well plate is highlighted, which can be used to label the worm and bacterial strains. (C) The file is labeled with worm and bacterial strains. Please click here to download this File.
Supplementary File 3: LFASS analysis using Matlab. (A) After opening the LFASS folder in Matlab, a command window pops up. (B) "fitfolder" is typed in the command window to start the program, and the folder's name with the file to be analyzed is indicated. (C) Once the program finds the data excel file for analysis, the user must follow on-screen instructions, providing the various parameters requested. Please click here to download this File.
Supplemental File 4: Analyzed LFASS data using Matlab. (A) An example of a curve fitted through Matlab analysis, where the black line indicates the time point corresponding to the earlier time boundary and the blue line the later time boundary for the curve fitting. The fitted sigmoid curve is displayed in red. (B) The results file generated by Matlab is opened in .xls format. The values are highlighted and displayed in number format. The second column provides the time when the value on the raw data curve first exceeds 50% of the maximum value. The third column provides the time when the fitted curve's value equals 50% of the maximum value, as determined during the batch-fitting analysis. The fourth column provides the time at which the value on the fitted curve equals 50% of the maximum value after batch analysis and final refitting where necessary. The fourth column should, thus, provide the most reliable result. Please click here to download this File.
Supplementary Table 1: Example experimental plate layouts. (A) 96-well plate layout for testing 47 different worm gut microbiota bacterial isolates on two separate worm strains, with live OP50 being used as a control. Four copies of this plate are required for each assay being run. (B) 384-well layout for heat or oxidative stress assays, corresponding to the previous 96-well plate assay in (A). This setup allows tetraplicates to be performed for each condition, with each worm strain having its own OP50 control. Please click here to download this Table.
C. elegans offers many advantages for rapidly screening multiple experimental parameters at once, owing to its small size, transparency, fast development, short lifespan, inexpensiveness, and ease of handling. Its considerably simpler genome, body plan, nervous system, gut, and microbiome, yet complex and similar enough to humans, make it a powerful preclinical model, where mechanistic insight can be gained while testing for bioactive efficacy or toxicity. As interest is growing in developing microbial interventions for diseases8,9,34,35, taking advantage of such preclinical models to rapidly investigate probiotics is an attractive option with the largely untapped and near-unlimited therapeutic potential of microbes34. The protocol described here is suitable to study the impact of environmental bacteria on C. elegans health but can be readily adapted for other purposes.
Versatility
The different "modules" that form this pipeline can be adapted to match experiment- or lab-specific requirements. For instance, bacterial isolates can be substituted for bacterial mixtures, bacterial mutants, or RNAi-producing bacteria, and the setup can be used to screen for bacterial genes involved in host-microbe interactions, screen for host genes involved in stress resistance, or study the impact of individual microbes as part of defined bacterial communities. This method describes the screening of several bacterial isolates in a single worm's genetic background. However, the assay is scalable, and multiple host and bacterial genotypes can be studied at once. This can be further combined with differential drug treatments and/or dietary regimens to gain insight into host gene-microbe gene-diet interactions.
The protocol is yet to be attempted on other nematode species, but it should also be directly applicable to C. remanei and C. briggsae, and it may apply to L3 stage parasitic nematode species that share morphological, size, and death fluorescence features with C. elegans30. Finally, other stressors (juglone, paraquat, arsenate, other temperatures) can be used within the same pipeline to provide complementary or confirmatory information.
Potential challenges
The time investment up to day 12 to grow worms and bacteria before the assays is significant, but due to the throughput of the approach, it is far outweighed by the breadth of the collected data. Nevertheless, care in planning such assays is essential to avoid wasting time by repeating the preparation steps because materials are suboptimal by the time the assays can start. Main early issues include sample contamination (worms or bacteria), poorly synchronized worm populations, and running out of materials (worm populations eat through their food faster than expected and re-feed or transfer animals more often than planned).
Avoiding contaminations
Due to the scale and duration of the protocol, one major potential issue is contamination by fungi and bacteria, whose occurrence would certainly confound results. The ideal setup is, thus, an air-conditioned and air-filtered dedicated lab space without direct airflow above the experimental bench, using sterile consumables. However, observing regular cleaning procedures (decontaminating benches before and after work, cleaning plate boxes, and using sterile consumables) and working by a flame in a contained space (avoid hallways, spaces near open doors, and windows) are usually sufficient to maintain a clean environment. If mild contaminations occur prior to egg preparation in Step 2.2.1., the bleaching procedure will take care of them, and the protocol can progress as planned. Conversely, contaminations that affect the growing worm population prior to transferring onto test bacterial plates will signify the end of that attempt, and the protocol will need to be restarted from Step 1. Later contaminations will tend to affect single bacterial strains or test conditions. Depending on their extent, it may be worth going on or restarting the protocol.
Worm synchronization
As resistance to thermal and oxidative challenges changes when worms develop and grow, test conditions that lead to different developmental stages by the time LFASS assays are carried out cannot be readily compared. It is essential to yield well-synchronized populations for endpoint assays. As the assays are performed on worms collected in bulk at L4 or L4 + 48 h stages, in the absence of a worm sorter or other sorting procedure to collect perfectly staged worms, the worm populations must be very well synchronized at L1. It should be ensured that hatchlings are kept off food long enough for all live eggs to hatch before growing the test cohort (>15 h at 20 °C and >18 h at 15 °C). Genotype may affect the spread in developmental timing, which is also affected by temperature. Growing worm populations at 15 °C give better control over this while limiting the impact of temperature-sensitive mutations during the worm population expansion phase (notably when working with insulin signaling mutants). The main reason to study the effects of microbes from L4 or L4 + 48 h is to bypass the developmental impact of a differing diet and study the impact of live gut bacteria rather than their impact on diet. Worms grown in an earlier larval stage on different bacterial isolates or mixtures will inevitably lead to uneven developmental rates and a loss of synchrony.
Consequences of relying on blue fluorescence
Being able to pinpoint death in LFASS assays without the need for individual worm tracking or additional reagents such as Sytox green36, but instead exploiting an endogenous signal in a fluorescence range that is unlikely to overlap with traditional worm biomarkers, is a clear benefit. However, it remains critical to ensure that worm death fluorescence can be consistently detected across many test conditions. While the assay does not rely on fluorescence quantification but on its dynamics, sufficient fluorescence still needs to be emitted by the worms to estimate the median time of death (TOD) accurately. As the assay does not allow for resolution of single worm death, the blue fluorescence bursts emitted by dying individuals need to have sufficient temporal overlap for fluorescence accumulation to give out a single population fluorescence peak. Only then can median TOD be accurately inferred. This means worms unable to produce sufficient levels of blue fluorescence cannot be used in this protocol, which notably includes some mutants of the kynurenine pathway such as tdo-2 and kynu-130,31, as well as starved animals. Hence, it is important to ensure worms are kept in replete conditions throughout. This can notably be an issue when performing HT115-driven RNAi screens as HT115 lawns on NGM medium supplemented with IPTG and antibiotics tend to be thinner than OP50 lawns. Monitoring worm food supply during the days leading to the assays is, thus, critical.
Another consequence of the dynamics of death fluorescence is that the quicker the assay, the better the sensitivity, as worms from the same population will die more synchronously and produce a more defined population death fluorescence peak. Thus, the longer the assay, the larger the population size needed to generate good death fluorescence peaks. Mutants with low metabolisms, such as eat-2 and daf-230, will also give out less fluorescence and require higher numbers per well. While 10-15 wild-type worms per well are sufficient for a 42 °C thermal stress assay that kills all of them within 4 h, nearly double that amount is required for daf-2(e1370) worms in the same assay, and >100 wild-type worms are required for an assay that would kill worms over a 24h period.
In the context of environmental bacterial isolates, differing microbial metabolism can also differentially impact the kynurenine pathway that produces the blue fluorescing compounds. Hence, some bacterial isolates lead to lower peak death fluorescence, while others lead to higher peaks than OP50. As a result, one needs to aim for enough worms per well to accommodate all eventualities.
Lastly, correctly optimizing plate-reading parameters is essential. This protocol covers the setting of fluorescence excitation and emission parameters and the depth at which readings should ideally be performed. With a possibly wide range of death fluorescence yields across multiple conditions in a single assay, it is important to ensure that weak signals are detected but that detectors are never saturated. Instruments with detectors that boast high dynamic ranges will, thus, perform better.
Limitations of the method
The method described here is robust and flexible and can be readily adapted to various conditions. For convenience, key restrictions and limitations are mentioned along the protocol where they come into play. Briefly, the reliance on death fluorescence as an endogenous marker of death is both a strength and a limitation of these assays.
Under a limited number of specific dietary conditions (low tryptophan diets or starvation) and in mutant conditions affecting tryptophan metabolism (i.e., tdo-2, kynu-1, some daf-2 alleles), the death fluorescence (DF) may be strongly attenuated, to the extent that the peak fluorescence is difficult to detect and TOD estimates are less reliable.
Conversely, some microbes may produce blue fluorescing compounds whose excitation and emission spectra overlap with DF signals (i.e., Enterococcus faecalis). While their dynamics differ from the DF and can often be uncoupled, they may confuse the analysis.
Next, the assays rely on the cumulation of signals from dying individual worms, but DF is quenched over time. Hence, if worms die too far apart, individual fluorescence peaks will fail to produce a population peak, preventing the measurement of a population median TOD. The slower the assay kills, the higher the worm number required in each well30. Hence, 10-15 adult worms per well are sufficient for a 42 °C heat stress assay that kills half the worms within 2 h, but over 70 worms per well are necessary for a bacterial fast-killing assay that kills half the worms in 12 h.
Lastly, users need access to a plate reader to perform such assays that can perform time-lapse fluorescence measurements at the required wavelengths. Excitation and emission wavelengths may also vary slightly between systems and vary between nematode species, as mentioned30.
Comparison and complementarity with other C. elegans assays
Over the past 10-15 years, many assays have been developed to assess C. elegans health in multiple conditions at once, employing a diversity of strategies and offering different levels of data depth and breadth30,36,37,38,39,40,41,42,43,44,45,46. Several powerful assays have notably been developed to automatically monitor many worms at once on solid media across their lifespan, using camera arrays and scanners with multi-worm tracking algorithms and/or microfabricated individual worm arrays37,38,39,40,41,43. In the context of host-microbe interactions, similar approaches to the protocol described above but not relying on DF have also enabled genome-wide RNAi screening of C. elegans genes involved in bacterial infections46,47,48. However, these high-throughput assays rely on handling worms in liquid throughout, which is not desirable when dealing with multiple bacterial types at once due to the increased risks of cross-contamination and because liquid culture affects host and microbial metabolisms. Hence, in traditional infection, stress, healthspan, and longevity assays, C. elegans are maintained and aged on bacterial lawns grown on solid media. As is also the case in this protocol, results from screens performed with this pipeline can be more readily related to and predict the outcome of downstream confirmatory experiments with screen hits.
The advantages of the LFASS approach itself have been well-covered in Reference30. Compared to more elaborate continuous imaging setups that could be repurposed for similar studies, LFASS is cheaper (provided median TOD is the only readout needed), and its simplicity means it does not require much technical ability or specialized equipment. LFASS is mainly about breadth over depth. LFASS data do not provide information other than the median time of death of a worm population and an indirect readout of a kynurenine pathway output. Conversely, with lower but still decent throughput, complex traits can be measured from other assays37,38,39,40,41,43 over the whole lifespan of worms, making them excellent downstream approaches to study hits from an LFASS screen.
The authors have nothing to disclose.
We thank the CGC Minnesota (Madison, USA, NIH – P40 OD010440) for providing worm strains and OP50 and Pr. Hinrich Schulenburg (CAU, Kiel, Germany) for providing all the environmental microbial isolates depicted here. This work was funded by a UKRI-BBSRC grant to AB (BB/S017127/1). JM is funded by a Lancaster University FHM PhD scholarship.
10 cm diameter plates (Non-vented) | Fisher Scientific | 10720052 | Venting is not necessary for bacterial cultures |
15 cm diameter plates (Vented) | Fisher Scientific | 168381 | |
384-well black, transparent flat bottom plates | Corning | 3712 or 3762 | Not essential to be sterile for fast stress assays |
6 cm diameter plates (Vented) | Fisher Scientific | 150288 | Venting is necessary for worm cultures to avoid hypoxia |
96-well transparent plates (Biolite) | Thermo | 130188 | |
Agar (<4% ash) | Sigma-Aldrich | 102218041 | Good quality agar is important for the structural integrity of the culture media, to avoid worm burrowing |
Agarose | Fisher Scientific | BP1356 | |
Avanti Centrifuge J-26 XP | Beckman coulter | ||
Bleach | Honeywell | 425044 | |
Calcium chloride | Sigma-Aldrich | C5080 | |
Centrifuge 5415 R | Eppendorf | ||
Centrifuge 5810 R | Eppendorf | ||
Cholesterol | Sigma-Aldrich | C8667 | |
LB agar | Difco | 240110 | |
LB broth | Invitrogen | 12795084 | |
LoBind tips | VWR | 732-1488 | Lo-bind reduce worm loss during transfers |
LoBind tubes | Eppendorf | 22431081 | |
Magnesium sulfate | Fisher Scientific | M/1100/53 | |
Plate reader- infinite M nano+ | Tecan | Monochromator setup enables fluorescence tuning but adequate filter-based setups may be used | |
Plate reader- Spark | Tecan | ||
Potassium phosphate monobasic | Honeywell | P0662 | |
Sodium chloride | Sigma-Aldrich | S/3160/63 | |
Stereomicroscope setup with transillumination base | Leica | MZ6, or M80 | Magnification from 0.6-0.8x up to 40-60x is necessary, as is a good quality transillumination base with a deformable, titable or slidable mirror to adjust contrast |
t-BHP (tert-Butyl hydroperoxide) | Sigma-Aldrich | 458139 | |
Transparent adhesive seals Nunc | Fisher Scientific | 101706871 | It is important that it is transparent and that it can tolerate the temperatures involved in the assays. |
Tryptophan | Sigma-Aldrich | 1278-7099 | |
Yeast extract | Fisher Scientific | BP1422 |