Wild-type Bristol N2 C. elegans were maintained as hermaphrodites at 20°C, grown on modified Nematode Growth Media (NGM) (0.35% instead of 0.25% peptone), and fed Escherichia coli OP50 (Brenner, 1974). A parent stock of wild-type Bristol N2 animals was generated by growing a population on 40 10 cm NGM plates seeded with E. coli OP50 until starvation. The newly starved animals were collected using M9 buffer and transferred to a 50 mL conical tube. The animals were pelleted via centrifugation at 1,000 × g for 3 min and the supernatant removed. The pellet was resuspended in 40 mL of sterile S-buffer +15% (v/v) glycerol, the resuspended animals were separated into 1 mL aliquots and stored at −80°C.

The following bacteria strains were grown using standard conditions (Lagier et al., 2015): Escherichia coli strain OP50 and S. enterica strain SL1344. E. coli OP50::GFP was also cultured using standard conditions with the inclusion of 100 µg/mL of ampicillin.

A new frozen stock of wild-type Bristol N2 was recovered from −80°C for each lineage and allowed to grow at 20°C on NGM seeded with E. coli OP50 for two generations. Well-fed adult wild-type animals were collected using M9 buffer and centrifuge at 1,000 g for 3 min to form a worm pellet. The supernatant was removed, and the animals were lysed using 500 µL a solution consisting of 5 mL of 1N sodium hydroxide and 2 mL of an 8.25% (w/v) sodium hypochlorite solution for a total volume of 7 mL (Final concentrations of the lysis solution are 0.71N sodium hydroxide and 2.36% (w/v) sodium hypochlorite). The animals were gently agitated at room temperature for 4 min or until complete breakage of the adult animals was observed under a dissecting microscope. The lysis solution was diluted using 10 mL of M9 buffer and a pellet was formed by centrifugation at 1,000 × g for 3 min. The supernatant was removed and an additional 10 mL of M9 buffer was added to wash the released eggs, a total of 3 M9 washes were performed. The eggs were synchronized for 22 h in sterile S-buffer + 5 µg/mL of cholesterol at room temperature. Synchronized L1 larval animals were transferred onto modified NGM plates seeded with E. coli OP50 and allowed to grow at 20°C for 48 h. E. coli and S. enterica lawns were prepared by culturing the bacteria for 15∼16 h in Luria Broth (LB) at 37°C in a shaking incubator set to 200RPM. E. coli OP50::GFP was also prepared by culturing the bacteria in LB plus 100 µg/mL of ampicillin for 15∼16 h at 37°C in a shaking incubator set to 200RPM. 500 µL of liquid bacteria cultures were seeded onto 10 cm NGM plates with or without ampicillin and spread aseptically to form a large lawn. Plates were allowed to grow at 37°C for 24 h to form a thick bacterial lawn. After incubation the plates were allowed to cool at room temperature for at least 30 min. L4 larval animals were collected using M9 buffer and split into two separate populations. The two populations of animals were placed onto room temperature plates seeded with either E. coli OP50 or S. enterica SL1344, the two populations were then placed into the same 20°C incubator for 8 h (initial P_o generation). After 8 h, the animals were washed three times in M9 buffer containing 100 µg/mL of kanamycin before being placed onto NGM plates containing 100 µg/mL of ampicillin, seeded with E. coli OP50::GFP to hamper any S. enterica growth. The animals were allowed to grow until they reached 72 h old, at which time the animals were lysed using the previously mentioned method to prepare next-generation. The F₁ through F₁₂ generations in both lineages were grown on NGM plates seeded with E. coli OP50, neither lineage was exposed to S. enterica SL1344 after the initial P_o exposure.

Animals of both lineages were synchronized using the previously mentioned sodium hydroxide and sodium hypochlorite method and allowed to hatch at room temperature for 22 h in S-buffer with 5 µg/mL of cholesterol. The synchronized L1 larval animals were allowed to grow at 20°C for 72 h on fresh NGM seeded with E. coli OP50. The synchronized 72-h old adult animals from both the naïve and trained lineages were placed onto 3.5 cm NGM plates seed with S. enterica SL1344. Three S. enterica SL1344 NGM plates were prepared for each lineage with a 30 µL drop of fresh S. enterica SL1344 liquid bacteria culture which was grown in LB for 15∼16 h at 37°C in a shaking incubator set to 200RPM. The plates were gently swirled to create an approximately 1.5 cm in diameter lawn in the center of the plate. The seeded NGM plates were grown in a 37°C incubator for 15∼16 h. After incubation, plates were allowed to cool to room temperature for at least 30 min before the 72-h old animals were placed on the plates. 20 animals were placed onto each S. enterica SL1344 NGM plates for a total of 60 animals per lineage per assay. The survival assays were incubated at 20°C in a dedicated incubator and live animals were transferred daily to fresh plates until egg laying ceased. Animals were scored once a day and were considered dead when they failed to respond to touch.

RNA interference was conducted by feeding C. elegans with E. coli strain HT115(DE3) expressing double-stranded RNA (dsRNA) that is homologous to the target gene of interest. Briefly, E. coli with the appropriate vectors was grown in LB broth containing ampicillin (100 μg/mL) at 37 °C overnight, and plated onto NGM plates containing 100 μg/mL ampicillin and 3 mM isopropyl β-D-thiogalactoside (IPTG). RNAi-expressing bacteria were allowed to grow overnight at 37 °C. L2 and L3 larvae animals were placed on the fresh RNAi-expressing bacteria lawns and allowed to develop into gravid adults over 2 days at 20°C. Gravid adults were then transferred to fresh RNAi-expressing bacterial lawns and allowed to lay eggs at 25°C to generate a synchronized RNAi population. After 1 h, the gravid adults were removed from the plate and the eggs were allowed to develop into young adults over 65 h at 20°C. The young adults were then transferred to survival assay plates. Clone identity was confirmed by sequencing at Eton Bioscience Inc. unc-22 RNAi was included as a positive control in all experiments to account for RNAi efficiency.

P_o 56-h old animals were collected after an 8-h exposure to either E. coli OP50 or S. enterica SL1344 at 20°C. The animals were washed three times with 10 mL of M9 buffer and centrifuge at 1,000 × g for 3 min to remove any excess bacteria. The supernatant was removed after the final M9 buffer wash and 400 µL of QIAzol (Qiagen) was added to the worm pellet. The samples were submerged in a bath of ethanol and dry ice immediately after the addition of QIAzol. F₁ through F₁₂ generations were collection followed a similar method, synchronized L1 larval animals were allowed to grow at 20°C on NGM plates seeded with E. coli OP50 until 56 h old before washing and collection in QIAzol. All RNA samples were stored at −80°C. RNA was isolated using the RNeasy Universal Plus Mini kit (Qiagen) following protocol provided by the manufacturer.

Total RNA was obtained as described above. 2 µg of RNA were used to generate cDNA per 100 µL reaction using the Applied Biosystems High-Capacity cDNA Reverse Transcription Kit. qRT-PCR was conducted by following the prescribed protocol for PowerUp SYBR Green (Applied Biosystems) on a Bio-Rad CFX384 Touch real-time PCR machine (Bio-Rad). 10 µL reactions were set up following the manufacturer’s recommendations, and 20 ng of cDNA was used per reaction. Relative fold-changes for transcripts were calculated using the comparative C _T (2^−ΔΔCT) method and were normalized to pan-actin (act-1, -3, -4). Amplification cycle thresholds were determined by the CFX manager software. All samples were run in triplicate. Primer sequences used for this research are the following:

Pan-actin (act-1, -3, -4):

Forward 5’—TCG​GTA​TGG​GAC​AGA​AGG​AC—3’

Reverse 5’—CAT​CCC​AGT​TGG​TGA​CGA​TA—3’

lipl-1:

Forward 5’—GTT​TGT​GAC​GAT​GTG​ATG​TTC​C—3’

Reverse 5’—AAG​TTC​CTG​CGG​GTG​TAT​G—3’

lipl-3:

Forward 5’—CTG​TAC​TGG​AGT​GAT​GCA​GAT​T—3’

Reverse 5’—GAA​GTA​GTT​GTT​CTG​CGC​AAT​TAT -3’

gst-4:

Forward 5’—GAT​ACT​TGG​CAA​GAA​AAT​TTG​GAC—3’

Reverse 5’—TTG​ATC​TAC​AAT​TGA​ATC​AGC​GTA​A—3’

Survival curves were plotted using GraphPad PRISM (version 10) computer software. Survival was considered significantly different from the appropriate control indicated in the main text when p < 0.05. PRISM uses the product limit or Kaplan-Meier method to calculate survival fractions and the log-rank test, which is equivalent to the Mantel-Haenszel test. The TD₅₀ for each survival curve was calculated using the area under the curve function and setting the y-axis baseline to 50 or “50% survival”. The resulting calculated x-axis value or “survival time” was used to plot the survival trend between generations. Statistical details for each figure are listed in its corresponding figure legend.

Pathogen exposure during L4 larval pre-adult development in C. elegans has been shown to cause transgenerational effects on subsequent generations (Rechavi et al., 2011; Sterken et al., 2014; Ma et al., 2019; Gabaldón et al., 2020). Offspring of animals exposed to members of the Pseudomonas family exhibited an enhanced avoidance behavior, increased dauer formation, or an increased expression of stress response genes (Palominos et al., 2017; Moore et al., 2019; Burton et al., 2020). However, these studies either did not measure survival over multiple generations or exposed multiple generations of offspring to the training pathogen before measuring survival. We sought to determine if a single prolonged exposure during the L4 larval stage would alter offspring survival against a slow killing pathogen over multiple generations. First, we exposed half a population of synchronized wild-type L4 larval animals to Escherichia coli OP50 and the other half to S. enterica SL1344 for 8 h to generate both naïve (E. coli OP50) and trained (S. enterica SL1344) lineages. S. enterica was selected as the training pathogen due to its low ability to invoke an avoidance response in C. elegans over 12 h, its lack of toxin production, and its predictability to slowly kill C. elegans via intestinal colonization over the course of 11 days as compared to 4 days using Pseudomonas aeruginosa (Sun et al., 2011; Sun et al., 2012; Sellegounder et al., 2018; Burton et al., 2020; Wibi et al., 2022). After exposure, the parental animals were washed to remove all bacteria and allowed to develop into gravid adults on E. coli OP50::GFP in the presence of 100 µg/mL of ampicillin to prevent S. enterica contamination. The animals were then lysed to obtain the F₁ generation offspring. Subsequent generations (F₁—F₁₂) of both trained and naïve lineages were maintained on E. coli and were not exposed to S. enterica during development. F₁ to F₁₂ animals from both the trained and naïve lineages were transferred after 72 h to plates seeded with a partial lawn of S. enterica and the survival time was measured (Figure 1).

Seven different replicates were measured using this method, each replicate was established by thawing a new tube of frozen wild-type animals from the −80°C and allowing the animals to grow for two generations on E. coli OP50. All the replicates descend from a single large population of animals to account for any previously unknown genetic or epigenetic changes in the population. All replicates were maintained in the same dedicated incubator to account for fluctuations in temperature and humidity. Animals were synchronized by removing all bacteria and live animals using a combination of sodium hydroxide and sodium hypochlorite, and allowing the newly collected eggs to hatch in sterile S buffer with 5 µg/mL cholesterol. Larval L1 animals in the absence of food, halt development and allow for easier synchronization of the two lineages. All eggs which did not hatch in the allotted 22 h were removed to prevent poor synchronization. The naïve and trained lineages of a replicate were maintained on the same fresh batch NGM plates and seeded with E. coli OP50 from the same single colony. Both lineages of a replicate were washed with buffer from the same batch and all antibiotics were premeasured and aliquoted to account for any differences in drug exposure. All survival assays between the naïve and trained lineages shared the same S. enterica broth grown from a single colony. Both lineages were removed from the incubator during survival assay measurement to account for difference in temperature at the bench versus the incubator. The initial replicate was performed alone, while the other six replicates were performed in pairs. It is possible that differences between the naïve and trained lineages may have been caused by an unknown variable in the environment, the methods used attempted to minimize these outside influences.

Taking the average of all the survival assays, the trained animals did not show a significant difference in survival when exposed to S. enterica as compared to the naïve animals (Figure 2A). We then plotted the time that it took for 50% of the assay animals to die (TD₅₀) for each generation against S. enterica to see if there were any differences between the two lineages on a generation-to-generation basis. When the TD₅₀ of the naïve lineage was plotted over generations, the animals’ survival displayed a tendency to oscillate between generations forming a repeating pattern of peaks and valleys. This repeating pattern throughout the 12 generations was unexpected and has yet to be reported. A one-way ANOVA examining the TD₅₀ values of the naïve lineage found a significant difference between generations. When the TD₅₀ of the trained lineage was plotted and examined in the same manner, the oscillation was present but noticeably suppressed as compared to the naïve lineage and an ANOVA test found no significant difference between generations of the trained lineage (Figure 2B). The TD₅₀ of all the survival assays for the naïve and trained lineages were also plotted alongside the means to better illustrate the data range (Figures 2C, D). Next, we examined the frequency distribution of all the survival assays. Though there was no significant difference between the two distributions, we do note that the naïve lineage had less defined peaks in frequency as compared to the trained lineage which centered around a single larger peak (Figures 2E, F). Taken together, these results demonstrate that TGIP does occur after an 8-h exposure to S. enterica, but the TGIP does not increase nor decrease survival on average as compared to the naïve lineage. Instead, the TGIP from the P_o generation stabilizes the later generations survival, reducing the generation-to-generation variability observed in the naïve lineage.

Given the survival of the naïve lineage tended to fluctuate between generations as opposed to the trained lineage, we questioned if the expression of genes related to S. enterica infection were also fluctuating in the absences of pathogen infection. To this end, we collected uninfected samples of the two lineages at 56-h old, the same age as the P_o generation collection, and examined the relative gene expression of antimicrobial genes which are induced by S. enterica infection, lipl-1, and lipl-3 (Xiao et al., 2017). Lipase-like 1 and 3 (lipl-1, -3) are predicted lipases located in lysosomes and expressed in the intestine of C. elegans (O'Rourke and Ruvkun, 2013). The lipl family of genes are regulated by MXL-3 and HLH-30, and control the utilization of internal energy reserves through autophagy and lipophagy (O'Rourke and Ruvkun, 2013). Since both lipl-1 and lipl-3 are upregulated during both S. enterica infection and food deprivation (Xiao et al., 2017; O'Rourke and Ruvkun, 2013), the stress response gene gst-4 was also measured as a control. gst-4 codes for the antioxidant enzyme Glutathione S-Transferase 4 and is induced by both starvation and pathogen infection (Hoeven et al., 2011; Tao et al., 2017), making it a well-suited indicator of false positives due to starvation or pathogen contamination.

Quantifying the expression of lipl-1, -3 and gst-4 in the F₁—F₁₂ generations and comparing the RNA levels to the level in the P_o generation showed that the expression of lipl-1 varied from generation to generation in the naïve lineage. By contrast, there is almost no variation in the expression of lipl-1 in the trained lineage between generations F₂ to F₈ (Figure 3A). The expression of lipl-3 however did fluctuate, regardless of lineage (Figure 3B). The F₉ generation in the naïve population has an interesting spike in expression of lipl-1 and lipl-3, reaching a mean relative quantity of 4.05 and 4.33 respectively. The cause of this spike during the F₉ and the sudden drop in expression during the F₁₀ is not clear. In neither lineage across all generations tested did the expression of gst-4 significantly increase as compared to the P_o generation (Figure 3C), indicating that the increase of lipl-1 expression in the naïve lineage was not due to a lack of food availability or pathogen contamination, but rather a possible innate fluctuation of gene expression. To confirm if lipl-1 plays a role in survival against S. enterica, the naïve population was treated with lipl-1 RNAi prior to pathogen challenge. Animals treated with lipl-1 RNAi displayed an enhanced survival against S. enterica as compared to the empty vector control (Figure 3D). This enhanced survival phenotype of the lipl-1 knockdown animals supports our previous observation that an overexpression of lipl-1 is detrimental to the survival of C. elegans during S. enterica infection.

To better examine the expression of lipl-1 and lipl-3, we plotted the mean relative quantity of each gene by generation to visualize any trends in expression (Figures 4A, B). Interestingly, when the mean relative quantity of lipl-1 was plotted, the naïve generations with higher lipl-1 expression tended to have a lower TD₅₀ as compared to other generations (Figure 4C). In contrast, the expression of lipl-1 in the trained lineages remained consistent between the F₂—F₇ generations, which correlates to the flattened survival tendency (Figures 4D, E). This near constant expression of lipl-1 in the trained lineage falters starting on generation F₈. Surprisingly, the two lineages share a similar expression trend with lipl-3 which does not correlate to change in survival in either lineage (Figures 4F–H). Since lipl-1 and lipl-3 expression are regulated by the translocation of the transcription factor HLH-30 into the nucleus, it is assumed that translocation of HLH-30 would increase expression of both gene proportionally (O'Rourke and Ruvkun, 2013). It is unlikely that modifications to HLH-30 signaling are the cause of the lipl-1 variation suppression. Comparing the expression of gst-4 between the two lineages again showed no difference in expression over the course of 12 generations (Figure 4I). These results suggest a possible plasticity in the expression of genes involved in the innate immune response of C. elegans. Lineages without TGIP do not have the information to attenuate the expression of lipl-1 and adjust its expression to their benefit. By contrast, the lineages with TGIP suppress lipl-1 variation which correlates to a more consistent survival across multiple generations. However, TGIP does not suppress the whole HLH-30 pathway as lipl-3 expression is unaffected after S. enterica exposure. These results point to a selective suppression of variation lipl-1 that does not affect related genes such as lipl-3.