Genetic analysis of sensitivity to temperature fluctuations identifies acclimation stress as a novel type of cellular stress in Caenorhabditis elegans Authors: Noah Lilienfeldt, Yifan Tai, Han Sun, Brandon Waddell, Wen Yang, Robyn Branicky, Siegfried Hekimi, B Conradt Published: 2026 ABSTRACT AbstractAll organisms must quickly acclimate to short-term environmental temperature changes to maintain optimal physiology. Temperature excursions far from the optimum lead to heat- or cold-shock responses, which have been extensively studied. We hypothesized that, even within a physiological range of temperature whose extremes are not stressful by themselves, the need for constant adjustment to frequent temperature fluctuations might be deleterious. We find that frequent fluctuations between physiological temperatures severely impair the short-term survival of C. elegans in a frequency-dependent manner. We show that the fluctuations impact both the soma and the germline, with the mechanisms that lead to short-term death requiring fertility. Using a survival assay with a candidate gene approach, we identified both hypersensitive and resistant mutants. The nature of the corresponding genes suggests that temperature sensing and mitochondrial function are particularly key to survival under temperature acclimation stress. However, the lack of response from reporter genes designed to reveal classic cellular stresses as well as the molecular identities of the hypersensitive, unaffected, and resistant mutants, suggests that acclimation stress represents a novel type of stress with potentially severely deleterious consequences for animals.Keywords: C. elegans, temperature acclimation, survival, germline health, stress responses Introduction Homeostasis, the adjustment of cellular physiology to the demands of environmental and internal change, is the role of the intricate regulatory mechanisms that allow cells to function without damage. Changes in the environment can impact organisms very directly. For example, changes in external temperature can directly alter the temperature at which cells function, and changes in the availability and composition of nutrients and respiratory gases lead to obligatory changes in the pathways of energy generation and biosynthesis (Hochachka and Somero 2002; López-Maury et al. 2008; Somero 2010; Hardie 2011). For example, in yeasts such as Saccharomyces cerevisiae and Schizosaccharomyces pombe, as well as in mammalian cells, changes in nutrient or oxygen availability reprogram energy metabolism and biosynthesis; in yeast, this includes induction of stress-, carbohydrate-metabolism, and energy-generation genes (López-Maury et al. 2008), whereas in mammals glucose deprivation or hypoxia activate AMPK, promoting ATP-generating catabolic pathways and suppressing anabolic biosynthesis (Hardie 2011). Given their need to cope with changes that they regularly encounter, organisms have mechanisms for acclimation. That is, they alter their physiology for optimal function under the new conditions (Seebacher et al. 2015). This is why environmental fluctuations are not acutely stressful as long as they do not exceed the range that is common in the organism's normal environment (Portner and Farrell 2008). However, there are several reasons why acclimation is likely not completely harmless. To maintain optimal physiology following environmental changes, organisms need to change their pattern of gene expression. Changing gene expression takes time and requires many steps and adjustments. For example, the pattern of expression of transcription factors (TFs) needs to change before, and in addition to, the subsequent changes in the expression of effector genes (Schulte 2014). Gene expression involves going through chromatin changes, transcription initiation, elongation, splicing, mRNA export, and translation. Thus, in addition to an energetic cost, during acclimation an organism might suffer from a cost of transient maladaptation, as there is necessarily a delay between the moment an organism senses a need to acclimate and the completion of a change in gene expression. Furthermore, the degree of maladaptation might even worsen when the external conditions change again while the organism is still adapting to the initial change, as this might induce even longer periods of uncoupling between the internal state and the external conditions (Piersma and Drent 2003; Angilletta 2009). Frequent changes in gene expression might by themselves be deleterious. Indeed, epigenetic modifications (DNA methylation, histone acetylation, others), which dynamically regulate gene expression in response to the environment, allow for the possibility of epigenetic drift, where repeated epigenetic reprogramming may lead to the accumulation of stochastic errors that ultimately disrupt gene regulation (Benayoun et al. 2015; Sen et al. 2016). To explore the overall hypothesis that acclimation is deleterious to organisms, we are focusing on temperature change for several reasons. (i) Temperature is easy to control in an experimental setting. (ii) The rate of all biochemical reactions will change with temperature, but not uniformly. Thus, to maintain homeostasis, specific genetic and molecular mechanisms must exist to re-equilibrate the rates of virtually all cellular processes every time cells experience a change in temperature. (iii) Temperature fluctuations are well documented to be experienced by the tissues of poikilotherms, heterotherms, and homeotherms, including humans (Tikuisis 2004; Taylor et al. 2014; Tansey and Johnson 2015; Martin and Falder 2017; Romanovsky 2018; Periard et al. 2021; Cramer et al. 2022). (iv) Temperature fluctuations have been found to have a strong impact on human health (Onozuka and Hagihara 2015; Rehill et al. 2015; Ebi et al. 2021; Wu et al. 2022). (v) Climate change appears to increase the frequency and amplitude of external temperature fluctuations. (vi) Numerous effects of temperature on C. elegans biology have been documented (Byerly et al. 1976; Klass 1977; Branicky et al. 2001; Begasse et al. 2015). We therefore used C. elegans to experimentally explore the hypothesis that although an episode of acclimation might not be acutely stressful, frequently repeated acclimation is detrimental to survival. Materials and methods C. elegans Genetics All C. elegans strains were maintained on NGM agar plates seeded with OP50 E. coli at 20 °C using standard methods (Brenner 1974). N2 (Bristol) was used as the wild-type reference strain. The full genotypes and sources of the strains used in this study are: N2 (CGC), BA784 spe-8(hc50) I (CGC), BX153 fat-7(wa36) V (CGC), CB1370 daf-2(e1370) (CGC), CB4108 fog-2(q71) V (CGC), CE1255 cep-1(ep347) I (CGC), CF1553 muIs84 [(pAD76) sod-3p::GFP + rol-6(su1006)] (CGC), CL2166 dvIs19 [(pAF15) gst-4p::GFP::NLS] III (CGC), CW152 gas-1(fc21) X (CGC), DMS441 acdh-11(n5878) III; nIs590 [fat-7p::fat-7::GFP + lin15(+)] V (CGC), DR26 daf-16(m26) I (CGC), FG125 ocr-2(ak47) osm-9(ky10) IV; ocr-1(ak46) V (CGC), FK134 ttx-3(ks5) X (CGC), FX776 sod-1(tm776) II (CGC), GA184 sod-2(gk257) I (CGC), KQ1366 rict-1(ft7) II (CGC), LD1 ldIs7 [skn-1b/c::GFP + rol-6(su1006)] (CGC), MQ130 clk-1(qm30) III (Wong et al. 1995), MQ1333 nuo-6(qm200) I (Yang and Hekimi 2010), MQ1784 dsc-3(tm1634) I (Liu et al. 2012), MQ887 isp-1(qm150) IV (Feng et al. 2001), MQ920 dsc-4(qm182) IV (Branicky et al. 2001), MT12999 mir-85(n4117) II (CGC), PR671 tax-2(p671) I (CGC), PS3551 hsf-1(sy441) I (CGC), QC115 atfs-1(et15) V (CGC), RB574 alg-2(ok304) II (CGC), RB653 ogt-1(ok430) III (CGC), RB754 aak-2(ok524) X (CGC), SJ4058 zcIs9 [hsp-60::GFP + lin-15(+)] V (CGC), SJ4100 zcIs13 [hsp-6p::GFP + lin-15(+)] V (CGC), TJ356 zcIs356 [daf-16p::DAF-16a/b::GFP; rol-6(su1006)] IV (CGC), TJ375 gpIs1 [hsp-16.2p::GFP] (CGC), TK22 mev-1(kn1) III (CGC), TQ233 trpa-1(ok999) IV (CGC), TU38 deg-1(u38) X (CGC), VC222 raga-1(ok386) II (CGC), VC3201 atfs-1(gk3094) V (CGC), VC446 alg-1(gk214) X (CGC), WS4581 unc-119(ed3) III; opIs263 [rpa-1p::rpa-1::YFP + unc-119(+)] (CGC), YT17 crh-1(tz2) III (CGC), ZG31 hif-1(ia4) V (CGC). T-shifting survival assays Fresh overnight cultures of OP50 E. coli were prepared every 2 wk and were freshly diluted in M9 buffer to a working concentration of 3.35 × 109 CFU/mL before each experiment. For t-shifting assays, a 7.5% (v/v) dilution of this working stock was used to prepare the “OP50 buffer.” The concentration of the bacteria in the OP50 buffer was tightly controlled to ensure reproducibility of survival after t-shifting. Young adult worms were transferred into 0.2 mL PCR strip tubes (Progene, Cat. No. 87-C200-8-TB, Ultident Scientific), 15 worms per tube containing 100 µL of OP50 buffer. The tubes were sealed with flat caps (Progene, Cat. No. 87-C8-QPCR, Ultident Scientific). Worms were washed once. This was achieved by gently resuspending the worms by pipetting, followed by pelleting, then the removal of 90 µL of the supernatant, and then a final gentle resuspension in 90 µL of fresh OP50 buffer to a final volume of 100 µL. Sealed tubes were placed in MiniAmp Thermal Cyclers (Thermo Fisher Scientific) for temperature-shifting (t-shifting) assays. T-shifting survival experiments were performed using the general setup described above. Worms were incubated in sealed 0.2 mL PCR tubes containing 100 µL of OP50 buffer and placed in MiniAmp Thermal Cyclers for 48 h or 72 hours, as indicated. At the start of all t-shifting experiments, worms are kept at 20 °C for 3 min. Then, except for the 20 °C controls, the temperature is programmed to shift first to the higher temperature. The maximum rate at which the block of this thermocycler can change temperature is 3 °C per second. At the end of the incubation period, tubes were removed from the thermal cyclers and 75 µL of supernatant was carefully removed from each tube. The remaining 25 µL, which contained the worms, was then pipetted onto fresh 60 mm NGM plates. Once the liquid was fully absorbed into the agar, the number of live, dead, and lost worms was scored. The contents of all 8 tubes of the strip were put on the same plate and counted all together. The proportion of live worms was calculated for each replicate. Each condition was tested in at least 3 biological replicates, with 8 tubes and 15 worms per tube (120 worms) per replicate. Exact sample sizes are provided in the corresponding figure legends. The conditions for successful t-shifting, including the rates of t-shifting, were established experimentally to obtain the high rate of survival at day 2 and the high rate of death at day 3 for the wild type. This revealed that the details of these conditions, such as the treatment of the bacteria, which are washed before using them in M9 buffer to avoid further growth, the exact number of bacteria added, the volume of liquid used, and, more anecdotally, the brand of the PCR tubes used, all affect the outcome quantitatively. We expect that the reproduction of our findings in a different laboratory might require first establishing the exact conditions that yield results similar to what we obtained with the wild type. As has been observed for aging studies, we suspect that differences in the C. elegans wild type strains between laboratories, differences in the OP50 strains, and differences in the conditions of worms and bacterial growth before incubation in liquid could also affect the outcomes of t-shifting quantitatively, albeit not qualitatively. Stress reporters Stress reporter strains were subjected to t-shifting (10-min intervals between 15 °C and 25 °C) for 48 h in OP50 buffer; control worms were maintained at a constant 20 °C and constant 25 °C, following the protocols described for t-shifting survival assays. Three repeats were carried out. Following treatment, worms were allowed to recover for 24 h on OP50-seeded NGM plates at 20 °C before imaging. We observed no difference in reporter expression after 24 or 48 h, with or without 24 h of recovery. For positive controls, young adult worms of each transgenic strain were treated with inducers to confirm GFP induction. Inducers were identified either in the literature or experimentally. A previously constructed strain containing a mutation in the mitochondrial gene isp-1 and the hsp-6p::gfp reporter (Wu et al. 2018) was used as the positive control for hsp-6p::gfp; a strain containing a mutation in the mitochondrial gene gas-1 and the hsp-60p::gfp reporter was constructed (this study) and used as the positive control for hsp-60p::gfp. gst-4p::gfp::NLS worms were heat-shocked at 29 °C for 19 h (this study). A previously constructed strain containing daf-2 and sod-3p::gfp (Bar-Ziv et al. 2020) was used as the positive control for sod-3p::gfp. skn-1b/c::GFP nuclear localization was achieved by inhibiting the negative regulator of SKN-1, GSK-3 (Blackwell et al. 2015) by exposing the transgenic worms to 71.4 mM lithium chloride solution for 48 h (Freland and Beaulieu 2012) (and this study). hsp-16.2p::GFP worms were heat-shocked at 35 °C in M9 solution for 2 h and then left to recover at 20 °C in M9 solution for 16 h (Rea et al. 2005). daf-16p::daf-16a/b::GFP nuclear localization was induced by incubating worms at 35 °C for 2 h, and fluorescent images were immediately taken without recovery period (Singh and Aballay 2009). RPA-1 foci, indicative of DNA damage, in the germ cell nuclei in rpa-1p::rpa-1::YFP worms were induced by incubating transgenic worms in 25 mM hydroxyurea, a genotoxic agent, for 48 h; images were taken following a 24-hour recovery period (Craig et al. 2012b). Images were taken using an Olympus BX63 fluorescence microscope using a 4 × objective. Image insets for nuclear localization were taken using a 20 × objective. Germ cells examined with the rpa-1p::rpa-1::YFP reporter were captured using the 60 × objective. Images were acquired using the CellSens software (Olympus). Exposure times varied for each reporter strain but were kept the same for all conditions tested with a given reporter strain. Germline nomarski imaging Worms were mounted on 5% agarose pads in 10 µL of 20 mM sodium azide for immobilization and immediately imaged using an Olympus BX63 microscope equipped with differential interference contrast (DIC) optics with a 60 × objective. Counting prophase I-arrested oocytes The DAPI-staining protocol was adapted from (Duerr 2006; Liang et al. 2018; Price et al. 2023). Worms were collected immediately following either incubation at 20 °C or 10-min t-shifting using the general setup described above. Samples were centrifuged at 100 g for 1 min and washed once with 500 µL of 0.1% Triton X-100 in M9 buffer. Worms were then fixed in 500 µL of ice-cold 100% methanol for at least 6 min on ice, followed by centrifugation at 100 g for 1 min. Fixed worms were washed twice with 0.1% Triton X-100 and resuspended in 100 µL of the same buffer. Samples were stored at 4 °C in the dark for up to 2 d prior to staining. For staining, most of the supernatant was removed, and worms were transferred onto a microscope slide. After near-complete evaporation of the buffer at room temperature, 6 µL of EverBrite Mounting Medium with DAPI (Cat. No. 23002, Biotium) and 4 µL of ProLong Diamond Antifade Mountant (Cat. No. P36965, Thermo Fisher Scientific) were added. A coverslip was applied and sealed with nail polish. Slides were stored in the dark to prevent photobleaching and could be imaged after 15 min of incubation in the DAPI mounting medium. Oocytes arrested in prophase I were counted using an Olympus BX63 fluorescence microscope with a 60 × objective and an ND1.5 neutral density filter. For each worm, the number of DAPI-stained prophase I-arrested oocytes was counted in both the anterior and posterior gonad arms, and the total number per worm was plotted as the sum of both arms. Acridine orange staining for visualization of germline apoptosis To visualize apoptotic corpses in the germline of worms, we performed acridine orange (AO) staining as described by (Craig et al. 2012a), with minor modifications. Following incubation at 20 °C or after 10-min t-shifting (as described above), worms were transferred to seeded NGM plates. A total of 500 µL of AO staining solution (5 µL/mL of 10 mg/mL AO stock in M9 buffer; Invitrogen, Thermo Fisher Scientific, Cat. No. A3568) was added directly to each plate. Plates were gently rotated to distribute the stain and incubated in the dark at room temperature for 1 hour. Worms were then washed off the plates with M9 buffer, pelleted at 100 g for 2 min, and washed twice more. After a 45 min recovery on fresh NGM plates in the dark at 20 °C, worms were mounted on 5% agarose pads in 10 µL of 20 mM sodium azide for immobilization. Apoptotic corpses were identified by the presence of discrete fluorescent bodies in the germline and scored within 1 h using an Olympus BX63 fluorescence microscope at 20 × objective. Images were acquired using CellSens software (Olympus). Brood size assay To assess brood size, individual L4-stage worms were singled onto 60 mm NGM plates and maintained at 20 °C. Worms were transferred daily to fresh plates until they stopped laying eggs. Progeny on each plate were counted 2 d after the adult was removed to allow for complete hatching and development. The number of dead eggs was also recorded. At least 10 individuals were analyzed per genotype, and brood size (which included all live progeny and dead eggs) was averaged across biological replicates. Defecation assay Young adult worms were picked into 100 µL of M9 buffer in 0.2 mL PCR tubes. The tubes were either left at a constant 20 °C or t-shifted between 15 °C and 25 °C in thermocyclers at the frequency indicated for 4 h. After the tubes were removed from the PCR machine, 75 µL of supernatant was carefully removed from each tube. The remaining 25 µL, which contained the worms, was then transferred onto 60 mm NGM plates, which had been kept at 20 °C. Worms were allowed to recover at 20 °C for 15 min and then were scored during the next 30 min. To control the temperature during scoring, 3 tubes and 3 plates were prepared for each condition so that plates were scored for a maximum of 10 min after removal from the 20 °C incubator. The length of time between 2 consecutive posterior body contractions (PBOCs) was considered the defecation cycle length. We observed a significant increase in the mean defecation cycle length of t-shifted animals (at all frequencies) vs the 20 °C control and a significant increase in the variance of the cycle length of animals that had been t-shifted at certain frequencies vs the 20 °C control. An F-test was performed to compare the variances. Heat stress assay Young adult worms were transferred to pre-warmed NGM plates with OP50 and left at 35 °C for either 3.5 or 5 h. Worms were transferred back to 20 °C and allowed to recover overnight. The next day, they were scored for recovery. Worms that were alive and able to crawl spontaneously or in response to being prodded with a platinum wire were scored as recovered; worms that were dead or paralyzed were scored as not recovered. Quantification and statistical analysis All statistical analyses and graph generation were performed using GraphPad Prism (version 10.6.1). Unless otherwise stated, data are presented as mean ± standard error of the mean (SEM). Statistical significance was determined using one-way ANOVA followed by corrections for multiple comparisons using the Bonferroni method, unless otherwise indicated. A P-value < 0.05 was considered statistically significant. Sample sizes are reported in the figure legends and/or shown on the graphs. Results The short-term survival of C. elegans is sensitive to the frequency, the range, the amplitude, and the pattern of temperature fluctuations. In the wild, it is expected that animals will experience temperature fluctuations in an irregular and complex pattern at the same time as fluctuations in other environmental parameters. This cannot be convincingly reproduced in the laboratory. Instead, to explore the effects of environmental fluctuations, we developed an assay relying only on temperature fluctuations. Worms are placed in liquid culture for 3 d in tubes of 100μL of M9 buffer provisioned with a fixed concentration of OP50 E. coli cells. To control temperature, the tubes are placed in a PCR thermocycler (Galbadage and Hartman 2008). The standard C. elegans growth temperatures compatible with high survival and reproduction range from 15 to 25 °C and are not stressful to the wild type under standard conditions (Link et al. 1999; Murray et al. 2007; Prahlad et al. 2008; Ohta et al. 2014; Zevian and Yanowitz 2014).Wild-type (N2) adult hermaphrodites on their first day of adulthood were picked into PCR tubes and subjected to 3 d of thermal fluctuation at different frequencies (each change of temperature is called a t-shift), following which their survival was scored (Fig. 1a). Before being transferred to the PCR tubes, the worms were grown in petri dishes at 20 °C under standard conditions. The controls for these experiments were worms kept under the same liquid conditions at a constant 20 °C. At the start of all t-shifting experiments in liquid, all worms are kept at 20 °C for 3 min. Then, except for the 20 °C controls, the temperature is programmed to shift first to 25 °C. The maximum rate at which the block of the thermocycler can change temperature is 3 °C per second, which means that it takes 3.4 s to change by 10 °C. After the shift to 25 °C, the worms remain at that temperature for the time interval to be tested, following which the temperature is shifted to 15 °C, where again the worms remain for the same time interval before the temperature is shifted back to 25 °C, etc. This procedure continues for 3 d, at which point, the number of surviving worms is scored. It is worth pointing out that the conditions under which we incubate the worms to assess t-shifting sensitivity likely represent a substantially different environment from that of the traditional way of maintaining worms on solid agar plates with a thick lawn of E. coli. This difference might also mean that caution has to be exercised when comparing the phenotypes the mutants display under these quite different conditions. Effect of t-shifting on wild-type survival. a) Survival following 72 h of temperature shifts between 15 °C and 25 °C, with shift frequencies ranging from 10 to 100 min. All statistical comparisons are relative to the survival with constant incubation at 20 °C. Each condition included 10 biological replicates, with approximately 120 animals per replicate. b) Survival following 72 h of 10-min temperature shifts over varying temperature ranges. All statistical comparisons are relative to the survival with constant incubation at 20 °C. Each condition includes 5 biological replicates, with approximately 120 animals per replicate. c) Survival following 72 h of shifts between 15 °C and 25 °C with different dwell times at each temperature per 20-min full cycle (2 shifts). Red bars represent longer dwell times at 25 °C, and blue bars represent longer dwell times at 15 °C per 20-min cycle. For red bars, statistical comparisons are made relative to constant 25 °C incubation; for blue bars, statistical comparisons are made relative to constant 15 °C incubation. Each bar comprises 10 independent biological replicates with approximately 120 animals per replicate. Data are presented as mean ± SEM. ****P < 0.0001; ***P < 0.001; **P < 0.01; *P < 0.05; ns, not significant. All worms survive when kept for 3 d at 20 °C, but when the temperature is shifted every 10 min, ≈80% of the worms are killed (Fig. 1a). The killing is proportional to the shifting frequency, with most worms surviving when shifted only every 40 min or less. To interpret these results, it is important to note that whatever the frequency of shifting, all worms experience the same amount of time at each of the 2 temperatures. Thus, the difference in survival is not attributable to the total time spent at either of the 2 temperatures. The experiment-to-experiment variability is greatest at the intermediate frequencies, likely because any uncontrolled parameter is more likely to affect the experimental outcome when the test conditions are neither harsh (fast t-shifting) nor mild (slow t-shifting). We also established that the rate of temperature change we used (3 °C/s) did not lead to a more severe outcome than the slowest possible rate (0.1 °C/s) (Supplementary Fig. 1 in Supplementary File 1). Next, we explored the effects of varying the range and amplitude of shifting while maintaining a 10-min t-shifting frequency. Figure 1b shows the results of shifting between 20 °C and 15 °C, between 20 °C and 25 °C, and between 17.5 °C and 22.5 °C. The results show that shifting over a narrower range is less deleterious than over a broader range. Shifting between 20 °C and 15 °C has very little effect, but shifting between 20 °C and 25 °C remains clearly deleterious. However, the latter is not as severe as shifting over the full range of 15 °C to 25 °C, even though the animals spend the same total time at 25 °C under both regimens. The fact that shifting between 17.5 °C and 22.5 °C also results in significant death confirms that time spent at 25 °C is not a necessary requisite for death by t-shifting. Thus, amplitude and range of shifting are important, but there is also a temperature asymmetry, with outcomes tending to be more severe when the temperature range is warmer. Following up on these findings, we established that maintaining worms constantly at 25 °C under our conditions led to almost no survival after 3 d (Fig. 1c). We wondered, therefore, what fraction of the time of a 20 min cycle worms need to stay at 25 °C to observe this severe effect. For this, we used a shifting protocol in which worms spent increasing amounts of time at 25 °C and decreasing amounts of time at 15 °C, while keeping the total t-shifting cycle length of 20 min constant, as when t-shifts are sustained every 10 min. Thus, in these experiments, all worms sustained the same number of shifts. We found that even when worms spent as much as 90, 95, or 97.5% of the time at 25 °C and only 10, 5, or 2.5% of the time at 15 °C, survival increased compared to constant 25 °C (Fig. 1c). To explore this further, we carried out the converse experiment, in which the worms spend most of their time at 15 °C and only short moments at 25 °C (Fig. 1c). We found that t-shifts to 25 °C as short as 30 s were sufficient to lower survival, and 1 or 2 min at 25 °C produced even more severe effects. Thus, the time spent at 25 °C, even for a short time, has deleterious effects when most of the time is spent at 15 °C. In sum, even a short period of time spent at a particular temperature changes the physiology of the worms in a way that changes the reaction to another temperature. Stress reporter-gene responses to t-shifting We used well-characterized transcriptional and translational GFP reporter constructs to test whether previously characterized cellular stresses are induced by t-shifting (Fig. 2 and Supplementary Fig. 2 in Supplementary File 1). We compared the expression of the reporters in worms kept at constant 20 °C and constant 25 °C to worms that have been t-shifted for 2 d, at which time almost all worms are still alive. We tested the heat-shock response (with an hsp-16.2 reporter) (Link et al. 1999), the UPRmt response (with hsp-6 and hsp-60 reporters) (Yoneda et al. 2004), the response to oxidative stress and other stresses (with sod-3, gst-4, daf-16 and skn-1 reporters) (Henderson and Johnson 2001; Link and Johnson 2002; An and Blackwell 2003; Libina et al. 2003). The gst-4 reporter includes a nuclear localization sequence. skn-1 and daf-16 are transcription factors, and their reporters are translational and thus express all or part of the activity of the corresponding proteins. None of these reporters revealed any measurable increase in expression or nuclear localization in response to 2 d of 10-min t-shifting compared to constant incubation at 20 °C (Fig. 2 and Supplementary Fig. 2 in Supplementary File 1). With constant 25 °C incubation reporter expression appears to be decreased in some cases, especially in the case of the skn-1 reporter. Effect of t-shifting on stress pathway reporters. Representative fluorescent micrographs show reporter expression for: the mitochondrial unfolded protein response (UPRmt; hsp-6p::GFP and hsp-60::GFP), oxidative stress (gst-4p::GFP, sod-3p::GFP), oxidative stress and other stresses (skn-1b/c::GFP), the heat shock response (HSR; hsp-16.2p::GFP), and insulin/IGF-1 signaling (IIS)/general stress response (daf-16p::daf-16a/b::GFP). Control worms were maintained at a constant 20 °C for 48 h and recovered for 24 h on plates at 20 °C. 25 °C worms were maintained at constant 25 °C for 48 h and recovered for 24 h on plates at 20 °C. T-shifted worms were subjected to temperature shifting (every 10 min between 15 °C and 25 °C) for 48 h and recovered for 24 h on plates at 20 °C. See Methods for a description of the conditions used for the positive controls for each reporter. Insets for gst-4, skn-1, and daf-16 reporters demonstrate nuclear localization under positive control conditions. Twelve animals were examined for each condition, and the experiment was repeated 3 times. See Supplementary Fig. 2 in Supplementary File 1 for enlarged images and additional information. Scale bars: 200 µm for all main images; 100 µm for insets. Identifying hypersensitive and resistant mutants We observed that while most wild-type worms are killed after 3 d of 10-min t-shifting, most are not killed after 2 d. This provided the basis for a mutant screen (Fig. 3, a and b and Supplementary Fig. 3 in Supplementary File 1). We compared the survival of mutants that had been t-shifted between 15 °C and 25 °C every 10 min for 48 and 72 h to the survival of the mutants kept at constant 20 °C for the same amount of time. We considered mutants that have significantly reduced survival compared to the controls after 2 d to be hypersensitive and mutants with significantly increased survival compared to the controls after 3 d to be resistant. The culture conditions for our tests are labor-intensive (Materials and Methods), only allowing for a candidate-gene screen of 31 strains. All hypersensitive mutants were also tested at the much lower shifting frequency of 60-min shifts (Fig. 3b). Under these conditions, the wild type shows ≥89% survival compared to controls (Figs. 1 and 3). However, some hypersensitive mutants were strongly affected; we call them ultra-hypersensitive (Fig. 3, a and b). Survival of the wild type and mutants following 48- and 72-h high-frequency temperature shifts. a) Scatter plot schematic showing survival at 72 h (y-axis) vs survival at 48 h (x-axis) for wild-type and mutant strains exposed to 10-min t-shifting between 15 °C and 25 °C. Each point represents the average survival for the indicated mutants, which are color-coded by functional category: blue for stress resistance, red for temperature sensation, purple for homeostasis, and green for mitochondrial function. Mutants that are ultra-hypersensitive to acclimation stress (explained below) are marked with an asterisk (*). The wild-type strain is shown in black. b) Summary table of the survival of wild-type and mutant strains after 10-min t-shifting for 48 or 72 h. Mutants are grouped by functional categories. Aging phenotypes summarize findings from the literature, with the following acronyms: U (unknown), D (disputed), L (lengthened), S (shortened), and WT (wildtype). See Supplementary Table 1 in Supplementary File 1 for further explanation and references. Survival represents mean survival fraction ± SEM from ≥3 independent replicates with approximately 120 animals per replicate. For statistical analysis, for each genotype, the survival after t-shifting is compared to the survival of the wild type after t-shifting at that timepoint. ****P < 0.0001; ***P < 0.001; **P < 0.01; *P < 0.05; ns, not significant. Mutants that had significantly reduced survival compared to the wild type after 10-min temperature shifts for 48 h were considered hypersensitive (HS); mutants that had significantly reduced survival compared to the wild type after 60-min temperature shifts for 72 h and compared to themselves at constant 20 °C for 72 h were considered ultra-hypersensitive (UHS). Only HS mutants were tested for ultra-hypersensitivity. Mutants that had significantly increased survival compared to the wild type following 10-min temperature shifts for 72 h were considered resistant (R). Mutants with significantly reduced survival at 20 °C compared to the wild type at 20 °C for that timepoint are also indicated. A bar graph representation of the data is presented in Supplementary Fig. 3 in Supplementary File 1. For screening, we chose mutants in categories for which we expected an altered sensitivity to t-shifting. The screen appears to be highly discriminating, capable of finding that even moderate differences in survival are statistically significant. In this respect, surviving t-shifting might be similar to aging, for which many mutants live moderately but significantly shorter or longer than the wild type (Hsu et al. 2003; Pan et al. 2007; Blackwell et al. 2015; Zhang et al. 2019). However, we observed no correlation between the aging and t-shifting of the tested mutants (Fig. 3b). Hypersensitivity to t-shifting spans a broad range from only 26% survival (dsc-4) to 68% survival (rict-1) after 2 d. Resistance after 3 d ranges from 98% (nuo-6) to 41% survival (fat-7) (Fig. 3, a and b and Supplementary Fig. 3). Temperature sensing and reaction to temperature changes We tested mutations in crh-1, hsf-1, fat-7, trpa-1, mir-85, ttx-3, deg-1, dsc-3, dsc-4, tax-2, the triple mutant ocr-2 osm-9;ocr-1, and acdh-11 which all affect some aspect of reaction to temperature (Hobert et al. 1997; Branicky et al. 2001; Hajdu-Cronin et al. 2004; Murray et al. 2007; Prahlad et al. 2008; Chatzigeorgiou et al. 2010; Nishida et al. 2011; Ma et al. 2015; Chen et al. 2016; Ohnishi et al. 2020; Takagaki et al. 2020; Pagliuso et al. 2021) (Fig. 3, a and b and Supplementary Fig. 3). fat-7 is moderately resistant, and acdh-11 is strongly resistant. All others are clearly hypersensitive, with mir-85 and dsc-3 being ultra-hypersensitive. Thus, surviving 10-min t-shifting for 2 d requires many aspects of temperature sensation. In many organisms, temperature adaptation involves changes in the composition of lipids because of its effect on the fluidity of membranes (Sinensky 1971; Ernst et al. 2016). Several of the genes we tested were related to lipid metabolism (dsc-4, dsc-3, fat-7, and acdh-11). However, given the variety of phenotypes they display (hypersensitive, ultra-hypersensitive, moderately resistant, and strongly resistant), it is unclear how lipid metabolism is involved in the reaction to t-shifting. hsf-1 is hypersensitive although no heat shock temperatures are involved in the test, at least not as heat shock conditions were previously studied and defined (Lithgow et al. 1994; Pagliuso et al. 2021; Miles et al. 2023); and mir-85, a microRNA that participates in turning off expression of the heat shock protein HSP-70 after a heat shock (Pagliuso et al. 2021), is ultra-hypersensitive. Several of the hypersensitive genes encode ion channels involved in sensory neuron temperature sensing (trpa-1, ttx-3, deg-1, tax-2, ocr-1, ocr-2, and osm-9), suggesting that, as with heat shock (Prahlad et al. 2008; Lee and Kenyon 2009), a neuronal signal affects how non-neuronal cells react to moderate changes in temperature. This also suggests that having insufficient information about temperature, or the wrong information, leads to implementing the wrong program of acclimation, via changes in gene expression or other mechanisms. Stress responses We tested 8 mutants known to have altered responses to particular stresses: daf-2 (insulin/insulin-like growth factor receptor) and daf-16 (FOXO) (Lithgow et al. 1995; Murakami and Johnson 1996; Honda and Honda 1999; Scott et al. 2002; Lamitina and Strange 2005), cep-1 (p53 homologue) (Derry et al. 2001; Chang et al. 2017), sod-1 (main cytoplasmic superoxide dismutase) (Yang et al. 2007; Horspool and Chang 2017), sod-2 (main mitochondrial superoxide dismutase; also considered a mitochondrial mutant due to its localization) (Yang et al. 2007; Van Raamsdonk and Hekimi 2009), atfs-1 (master transcription factor regulating the mitochondrial UPR; both loss- and a gain-of-function alleles were tested) (Nargund et al. 2015; Soo et al. 2021; Uma Naresh et al. 2022), and hif-1 (hypoxia-inducible factor) (Jiang et al. 2001) (Fig. 3, a and b and Supplementary Fig. 3). cep-1 mutants showed no statistically significant resistance, sod-1, sod-2, atfs-1(lf), and daf-16 mutants were hypersensitive, and sod-1 and sod-2 were ultra-hypersensitive. daf-2, atfs-1(gf) and hif-1 mutants were resistant. The pattern we observed suggests that mitochondrial stress, oxidative stress, and thermal stress are among the stresses produced by t-shifting. See the Discussion about the discrepancy between stress-reporter expression and mutant findings. Homeostasis We tested mutants in genes involved in general homeostasis, including alg-1 and alg-2 (RISC), ogt-1 (O-GlcNAc transferase), rict-1 (mTORC2), raga-1 (mTORC1), and aak-2 (AMPK) (Fig. 3, a and b and Supplementary Fig. 3). alg-2 was wild-type, alg-1, ogt-1 and rict-1 were hypersensitive, and raga-1 and aak-2 were resistant. The fact that losses of crucial activities, such as that of AMP kinase and RAGA-1, lead to greater resistance to t-shifting is surprising and suggests that some of the effects of t-shifting are due to a homeostatic response that is partially maladaptive. That is, a weakened homeostatic response to t-shifting might be less deleterious than the full-strength response. Note that the term maladaptive here refers to the response to the acute stress in the screening assay, not to the less frequent and less regular temperature changes expected in nature. Mitochondria When looking at genes linked to stress resistance, we observed that loss- and gain-of-function alleles of atfs-1, the master regulator of the UPRmt, had opposite effects on resistance. Thus, mitochondrial function might be particularly relevant in determining hypersensitivity or resistance. We therefore tested the 5 following mutants that impair the respiratory chain function, the main mitochondrial function: clk-1, isp-1, nuo-6, gas-1 and mev-1 (Fig. 3, a and b and Supplementary Fig. 3). All mutants were strongly resistant, despite having poor energy generation and appearing fragile and sick. Many of the mitochondrial mutants we chose to study have altered lifespans, strongly extended (nuo-6, isp-1 and clk-1) (Wong et al. 1995; Feng et al. 2001; Yang and Hekimi 2010) or severely shortened (mev-1 and gas-1) (Kayser et al. 2004; Van Raamsdonk and Hekimi 2011). However, there is again no correlation between the aging and t-shifting phenotypes (Fig. 3b). Thus, the physiological changes that lead to altered acute sensitivity to a lethal temperature acclimation stress are distinct from those that promote or impair long-term survival under protective laboratory conditions. The germline is affected by t-shifting The function and organization of the germline appear compromised by t-shifting. T-shifting decreases oocyte production and maturation and leads to the presence of ectopic sperm (Supplementary Fig. 4 in Supplementary File 1). These effects already appear on day 1 of t-shifting even though there is no death at that time. However, at later times, the declining health of the animals and the greater disorganization of the germline prevented quantifiable observations. Mature oocytes arrested in prophase I are characterized by their size, shape, and internal morphology when using DIC microscopy (McCarter et al. 1999). DIC identification of oocytes was confirmed by DAPI staining, revealing the condensed bivalent chromosomes of prophase I. Compared to controls, there are fewer mature oocytes in the wild-type germline after 1 d of t-shifting (Fig. 4a and Supplementary Fig. 4). Furthermore, sperm is frequently localized outside of the spermatheca, including between the developing oocytes in the proximal gonad (Supplementary Figs. 4 and 5 in Supplementary File 1). Because t-shifting is carried out with adult animals, it means that upon t-shifting sperm frequently leaves the spermatheca and migrates to ectopic locations. Effect of t-shifting on the germline. a) Total number of prophase I-arrested oocytes (sum of anterior and posterior gonad arms) in wild-type (WT) and mutant strains under control conditions (20 °C, gray bars) or following 24 h of 10-min t-shifting between 15 °C and 25 °C (red bars). Genotypes are grouped by t-shifting survival phenotype: hypersensitive mutants are shown on the left, and resistant mutants on the right. Ultra-hypersensitive mutants are grouped with the hypersensitive mutants and indicated with an asterisk besides their gene name. Data represent mean ± SEM from ≥20 animals per genotype and condition. For each strain, the t-shifting condition was compared with the 20 °C control using t-tests; P values were adjusted for multiple comparisons using the false discovery rate (FDR) procedure. All * represent P < 0.001 except for daf-16 where * represents P < 0.01. b) Bar graph comparing the mean number of oocytes in mutants exhibiting a wild-type, hypersensitive or resistant phenotype with respect to t-shifting. Each point represents the mean number of oocytes for one mutant. The resistant mutants are significantly different from both the WT and hypersensitive mutants although only the difference to the WT group is shown. The WT mutants are not significantly different from the hypersensitive mutants. See Materials and Methods for full genotypes of mutants. c) Bar graph showing the mean brood size of mutants grouped by their t-shifting phenotype as in B. Each bar represents the mean brood size from ≥10 individual animals; error bars indicate the standard error of the mean (SEM). d) Bar graph comparing the mean brood size in mutants exhibiting a wild-type, hypersensitive or resistant phenotype with respect to t-shifting. Each point represents the mean brood size for one mutant. For B, D **P < 0.01 and ***P < 0.001. e) Representative fluorescent images show RPA-1 foci in germ cells in the distal ends of the germline using the reporter strain rpa-1p::rpa-1::YFP. For the positive control, worms were either treated with the toxic agent 25 mM hydroxyurea (positive control) for 48 h, kept at constant 20 °C or subjected to t-shifting (10-min intervals between 15 °C and 25 °C) for 48 h. Fifteen animals were examined for each condition, both immediately after removal from the PCR machine and also after 24 h recovery. Scale bar: 20 µm. f) Representative brightfield and fluorescent micrographs display germline apoptosis in the posterior gonad arm, visualized by acridine orange staining. Worms were either maintained at a constant 20 °C or subjected to temperature shifting at 10-min intervals between 15 °C and 25 °C for 24 h. Fifteen animals were examined for each condition, both immediately after removal from the PCR machine and also after 24 h recovery. Scale bar: 100 µm. We quantified the extent of ectopic sperm in all mutants with and without t-shifting (Supplementary Fig. 5 in Supplementary File 1). T-shifting increased the ectopic sperm phenotype in most mutants, with a greater effect, on average, in resistant mutants. However, these mutants also frequently showed an ectopic sperm phenotype even without t-shifting (Supplementary Fig. 5 in Supplementary File 1). Ultimately, the ectopic sperm phenotype could not be mapped exactly onto t-shifting sensitivity as measured by survival. We also quantified oocyte numbers with and without t-shifting in all mutants (Fig. 4, a and b). Many hypersensitive and resistant mutants had slightly fewer oocytes than the wild type without t-shifting. However, on average, the resistant mutants had significantly fewer oocytes than the wild type (Fig. 4b). However, the resistant mutants with the fewest number of oocytes, daf-2, acdh-11, clk-1 and nuo-6, were not more resistant to t-shifting than resistant mutants with more oocytes (Fig. 3). Following t-shifting the average number of oocytes in the wild type was reduced by almost half. This effect was not observed in hypersensitive mutants (Fig. 4b), except, but to a lesser extent, in deg-1 and tax-2 (Fig. 4a). However, this effect was clearly present in most resistant mutants and, on average, for the resistant mutants as a group (Fig. 4, a and b). Given that these mutants already tend to have fewer oocytes even without t-shifting, the effect resulted in resistant mutants having only a few oocytes following t-shifting (Fig. 4b and Supplementary Fig. 6). Interestingly, the number of oocytes after shifting did not correlate with t-shifting survival among either the hypersensitive or resistant mutants (Supplementary Fig. 6 in Supplementary File 1). To further explore differences in the function of the germline between hypersensitive and resistant mutants, we scored the brood size of the mutants grown on plates (Fig. 4, c and d). Although the brood size of any strain is not necessarily the same under different conditions, here liquid culture and on-plate culture, one would expect a high correspondence for most strains. We observed that hypersensitive mutants tended to have wild-type brood sizes on average while resistant mutants had reduced brood sizes. However, there was a large overlap between the 2 groups and, as for oocytes numbers after shifting, within each phenotypic group individual brood sizes showed no correlation with t-shifting survival (Supplementary Fig. 6 in Supplementary File 1). Further effects of t-shifting on the germline To assess whether germline DNA was damaged by t-shifting, we used the rpa-1 translational reporter for DNA damage (Stergiou et al. 2011), but no increase in expression was observed after t-shifting (Fig. 4e). To ensure healthy gametes, germline apoptosis is believed to help remove potentially damaged nuclei (Andux and Ellis 2008; Fausett et al. 2021). Indeed, apoptosis is a normal homeostatic function in the germline (Gumienny et al. 1999), and DNA damage increases germline apoptosis (Gartner et al. 2000). We also used acridine orange to stain apoptotic nuclei in the germline (Gumienny et al. 1999; Craig et al. 2012b), and found that it was almost completely inhibited by t-shifting (Fig. 4d), pointing to an effect of t-shifting on germline physiology. Fertility is required for t-shifting lethality Given the numerous effects of t-shifting on the germline, we asked whether fertility per se affected the outcome of t-shifting. We tested 2 sterile mutants: fog-2 mutants, which have no sperm, and spe-8 mutants, which have defective sperm (L'Hernault et al. 1988; Schedl and Kimble 1988). Both were fully refractory to death by t-shifting (Supplementary Fig. 7 in Supplementary File 1). This suggests that the killing mechanism involves the production of progeny or its physiological consequences. This could involve mechanical damage from the misplacement of the eggshell-enclosed embryos, or internal hatching of progeny, or germline to soma signaling. However, the fertile animals' germline anatomy was already too disturbed after 2 d of t-shifting to provide visual evidence for any particular mechanism. t-shifting also acts on the soma The survival of the sterile mutants suggests that germline dysfunction is necessary for the t-shifting killing mechanism. However, this does not necessarily mean that the process directly impacted by t-shifting is located in the germline, especially given the importance of the somatic gonad for germline function (Killian and Hubbard 2005). To investigate this, we further examined the skn-1::gfp and daf-16::gfp reporter strains, which express translational fusions that overexpress wild-type forms of the proteins exclusively in somatic cells (Henderson and Johnson 2001; An and Blackwell 2003). In fact, the corresponding strains were almost completely resistant to t-shifting (Fig. 5a), suggesting that although a functional germline is required for lethality, the germline dysfunction induced by t-shifting might result from an action of the soma on the germline. Effects of t-shifting on the soma. a) Bar graph showing survival of the wild type (N2), ldIs7 [skn-1b/c::GFP + rol-6(su1006)], and zIs356 [daf-16p::daf-16a/b::GFP + rol-6(su1006)] transgenic strains following 72 h of t-shifting between 15 °C and 25 °C every 10 min, compared to constant incubation at 20 °C. Transgenic lines express translational fusion proteins for SKN-1 and DAF-16; no expression is observed in the germline. Comparisons are made between the temperature-shifted and 20 °C control conditions. Each bar represents the mean of 3 independent biological replicates, with ≈120 animals per replicate; error bars represent SEM. ****P < 0.0001, ns = not significant. b) Bar graph showing defecation cycle length for animals kept at constant 20 °C or t-shifted between 15 °C and 25 °C for 4 h at the indicated frequency. Each point represents one defecation cycle; bars represent the mean and error bars the SEM. Statistical analysis of the variance of each condition vs the wild type is shown. ****P < 0.0001, ns = not significant. Coefficients of variation of bars from left to right are: 0.0571, 0.0586, 0.0958, 0.0965. c) Bar graph showing percent recovery at 20 °C after 3.5 h at 35 °C. d) Bar graph showing percent recovery at 20 °C after 5 h at 35 °C. For C and D mutants, which are grouped by their temperature-shift phenotype and ordered by their brood size as in Fig. 4. Each bar represents the mean of multiple independent biological replicates, with approximately 50 animals per replicate; error bars represent SEM. All bars are compared to the wild type. Only significant differences are indicated. *P < 0.01 to P < 0.0001. Mutants that have significantly decreased survival after 3.5 h at 35 °C compared to the wild-type are considered more sensitive to heat stress than WT; mutants that have significantly increased survival after 5 h at 35 °C compared to the wild-type are considered resistant to heat stress. See Materials and Methods for full genotypes of mutants. We also scored germline-independent phenotypic effects of t-shifting. The defecation cycle is a behavior of fixed duration that involves the function of the intestine and the nervous system (Branicky and Hekimi 2006; Mahoney et al. 2008; Pfeiffer et al. 2008; Wang et al. 2013). We tested the effect of short-term t-shifting (4 h of t-shifting between 15 °C and 25 °C) on the defecation cycle and observed a significant increase in the variability of cycle lengths in a frequency-dependent manner (Fig. 5b). As for t-shifting survival, very slow t-shifting (60 min) had no effect. To test whether t-shifting sensitivity correlated with sensitivity to heat stress, another somatic phenotype, we developed an assay that allowed us to screen for sensitivity and resistance to impaired survival following heat stress (Fig. 5, c and d). When adult worms are transferred to 35 °C for 3.5 h, most wild-type worms (>80%) will recover when transferred back to 20 °C. In contrast, when wild-type worms are left at 35 °C for 5 h most (>80%) will not recover. The worms did not appear to die from egg retention, indicating that any effect on the germline is unlikely to be responsible for heat sensitivity. These conditions represent a milder heat stress than conditions used in other studies as both hsf-1 and mir-85 mutants, which have been shown to be sensitive to heat stress under other assay conditions (Pagliuso et al. 2021; Miles et al. 2023), recover as well as the wild type in this assay. We found no correlation between sensitivity to heat stress on plates and sensitivity to t-shifting (Fig. 5, c and d). Among the hypersensitive mutants only alg-1 and daf-16 are more sensitive to heat than the wild type and 2 (sod-2 and tax-2) are even resistant to heat. Among the resistant mutants, aak-2 is more sensitive to heat than the wild type and several are resistant to heat. This includes mutants whose heat resistance was previously known, such as daf-2, and several of the mitochondrial mutants (Lithgow et al. 1995; Labbadia et al. 2017). There was also no correlation between heat resistance on plates and resistance to t-shifting. For example, several of the mutants that are particularly resistant to t-shifting, such as mev-1, raga-1 and acdh-11 (Fig. 3) were no more resistant to heat than the wild type. Thus, resistance to heat stress is neither sufficient nor necessary for resistance to t-shifting. Although, like the brood size assay, for technical reasons the heat stress assay is carried out on plates rather than in liquid culture, a strong correlation between the effect of the 2 conditions on mutants would be expected. Sensitivity to continuous 25 °C incubation To allow for controlled repeated t-shifting our experiments had to be carried out in liquid culture in closed PCR tubes with a finite amount of food. Under these conditions we observed that, like high frequency t-shifting, continuous incubation at 25 °C did not lead to reporter activation (Fig. 2 and Supplementary Fig. 2 in Supplementary File 1) but led to a high death rate (Fig. 1c). This suggests that our culture conditions create a sensitizing environment in which the effects of temperature shifting can be observed. This raised the possibility that altered sensitivity to deleterious aspects of liquid-culture incubation, particularly time spent at 25 °C, could underlie the apparent increased or decreased sensitivity to t-shifting. However, when t-shifting is carried out, the animals experience 25 °C only half of the total incubation time, which is not acutely deleterious per se as it does not lead to any death in the wild type when the frequency of t-shifting is low (Fig. 1a). Since survival of the wild type is uncoupled from time spent at 25 °C, it strongly suggests that resistance to 25 °C plays no role in resistant mutants. Even so, hypersensitive mutants could in principle be hypersensitive to t-shifting because they are hypersensitive to 25 °C. All hypersensitive mutants identified by 10-min t-shifting were therefore also tested at 60-min t-shifting frequency (Fig. 3). Most hypersensitive mutants were not hypersensitive at 60 min despite still being incubated half of the time at 25 °C as with 10-min t-shifting. This excludes the possibility that time spent at 25 °C is the cause of their hypersensitivity. However, the ultra-hypersensitive mutants sod-1, sod-2, hsf-1, mir-85, and dsc-3 were still sensitive at 60 min. For these mutants, we therefore cannot yet exclude the possibility that their hypersensitivity to t-shifting includes, or is dependent on, a hypersensitivity to 25 °C. Nonetheless, some of the mutants that we found to be resistant to t-shifting are known to be resistant to a number of stresses, such as daf-2, isp-1 and nuo-6 (Lithgow et al. 1995; Murakami and Johnson 1996; Honda and Honda 1999; Scott et al. 2002; Dues et al. 2017; Wu et al. 2018). We therefore tested all mutant strains for resistance to continuous 25 °C for 3 d (Supplementary Fig. 8a in Supplementary File 1). Unsurprisingly, all mutants with a wild-type and hypersensitive phenotype were as sensitive to continuous incubation at 25 °C as the wild type. In addition, 3 of the t-shifting resistant mutants, hif-1, atfs-1(gf) and fat-7 were also as sensitive to 25 °C as the wild type. The daf-16 and skn-1 overexpressing strains, which show high resistance to t-shifting, were only weakly resistant to 25 °C. However, several of the t-shifting mutants were also significantly resistant to continuous incubation at 25 °C to various degrees. Despite this, the overall correlation with t-shifting resistance was poor and non-significant (Supplementary Fig. 8b in Supplementary File 1). Interestingly, the fully sterile mutants, fog-2 and spe-8, which are fully resistant to t-shifting (Supplementary Fig. 7 in Supplementary File 1), were also fully resistant to 25 °C, indicating that sensitivity to 25 °C also requires a functional germline, possibly because egg retention is involved. Discussion All organisms are expected to react physiologically to changes in external temperature in order to benefit from an optimal adaptation to their environment. However, it is possible that when the demands on the organism to adapt are too extreme, the normal physiological response can itself become deleterious. We are familiar with this principle when considering processes like fibrosis, inflammation, the cytokine storm, apoptosis or stress responses, which are protective processes that become deleterious when too acutely or chronically stimulated. Using extreme experimental conditions, we can reveal processes that would otherwise be hidden in an all-encompassing general homeostatic response. This is what we have sought to do here with the t-shifting assay, which leads to profoundly deleterious consequences on the reproductive system, to the point of killing the animal. The dysfunction of the reproductive system from t-shifting suggests that the main function of the reaction to temperature change, which only becomes deleterious when too frequently triggered, is to preserve germline function. What could be the mechanism by which an appropriate response becomes deleterious when repeated too frequently? Temperature has the potential to affect every biochemical process in the cell. The Arrhenius equation tells us that the degree to which each biochemical process will be affected is roughly similar. However, there will remain differences between processes and, independently of biochemical reaction rates, the properties of macromolecules will change with temperature in ways that can weaken or destroy their function. It is therefore not surprising that organisms alter their patterns of gene expression when moved to different temperatures (Becerra et al. 2003; Gadgil et al. 2005; Brunquell et al. 2016; MacMillan et al. 2016; Lee et al. 2019), as they cannot rely on a uniform change in biochemical rates. We thus propose 2 potential hypotheses for why t-shifting is deleterious. Firstly, in small organisms like C. elegans the temperature experienced by all its cells will mirror the temperature of the liquid (which in our case takes only 3.4 s for a 10 °C change). Implementing changes in gene expression, however, takes substantially more time. For example, some of the fastest transcriptional changes that can be observed occur in the order of min, and more time is then required to produce proteins and localize them appropriately (Zobeck et al. 2010; Mahat et al. 2016; Muhlhofer et al. 2019; Liu et al. 2025). Thus, for a while, the worm's gene expression is not adapted to the temperature at which it finds itself, which should be deleterious. Secondly, the mechanisms of changes in gene expression cannot be error-free and could lead to partial maladaptation whenever a major change is required. In fact, some mechanisms of gene expression are intrinsically imprecise, such as epigenetic changes to chromatin. Indeed, repeated epigenetic reprogramming may lead to the accumulation of stochastic errors that ultimately disrupt gene regulation (Benayoun et al. 2015; Sen et al. 2016). We observed that the killing by t-shifting is frequency-dependent (Fig. 1a). However, over a fixed time (3 d), t-shifting at different frequencies also means differences in the total number of shifts. A number of hypersensitive mutants were also ultra-hypersensitive, showing significantly increased death at a 60-min t-shifting frequency that does not affect the wild type nor the majority of hypersensitives (Fig. 3b). This suggests the possibility that there is unrepaired or unrepairable damage at each t-shift. In the ultra-hypersensitive mutants, this damage could be particularly severe or particularly difficult to repair. The hypothesis that the total number of t-shifts is what matters suggests the possibility of future experiments to test the effect of t-shifting on the aging process using very low frequency but long-term t-shifting. Temperature fluctuations have been studied in the context of aging in C. elegans using very different conditions (Galbadage and Hartman 2008). These authors examined the effect of temperature fluctuation under conditions optimized for aging studies. They used large tubes (600 µL) with holes for aeration, only 5 worms per tube, in S-medium. They added S-medium daily to compensate for evaporation as well as more E. coli. The worms used for the experiments were obtained by bleaching gravid adults and incubating the synchronized larvae for 3 d at 16 °C before transferring to microfuge tubes. Under these conditions, the worms are viable at all temperatures and temperature fluctuation protocols tested and can be used for lifespan experiments. These conditions are very different to those we used to obtain acute temperature-fluctuation stress (see Materials and Methods). We do not know which aspects of the different conditions lead to different outcomes. Determining this will require future dedicated studies that also include aging experiments. Temperature acclimation stress such as we generate by t-shifting has not been studied and appears to be a novel type of stress. Indeed, our analysis with reporter genes does not indicate that well-studied stresses, such as heat stress, oxidative stress, mitochondrial stress, or DNA damage are visibly induced by t-shifting. In addition, the pattern of hypersensitive and resistant mutants cannot be readily predicted by a priori considerations based on the effect on stress reporter genes. For example, t-shifting does not stimulate the expression of mitochondrial stress reporters (based on hsp-6 and hsp-60 promoters) nor that of oxidative stress reporters (based on gst-4, skn-1, sod-3 and daf-16 promoters), yet atfs-1(lf), sod-1 and sod-2 mutants are hypersensitive (Fig. 2). How can we reconcile these observations? One possibility is that the threshold for increased sensitivity or resistance to t-shifting is much lower than the sensitivity of the reporter constructs. This suggests that, directly or indirectly, t-shifting generates a moderate increase in several distinct stresses which all, in aggregate, are involved in the physiological dysfunction that leads to death. Sensitivity is increased when any protection against a particular stress is lowered by mutation. In contrast, the threshold for the level of t-shifting sufficient for killing is elevated when protection from one or more of the stresses generated by t-shifting is increased, for example by overexpression of SKN-1 (Fig. 5a). The observation that some mutants involved in general homeostasis, such as aak-2 and raga-1, have increased resistance is particularly intriguing, as one would expect that the loss of homeostatic mechanisms would worsen cellular processes when constant readjustments to homeostasis are required. One interpretation is that under abnormally stringent conditions that require homeostatic readjustment, such as t-shifting, it is less deleterious to fail to adjust than to adjust with an exaggerated response. Again, we have examples of such phenomena in immunology or stress responses, where therapeutic interventions seek to dampen natural responses that have become deleterious. Indeed, the germline phenotypes support this notion. It is the resistant mutants who suffer more from t-shifting (in terms of oocyte production and maturation and sperm displacement) than the hypersensitive mutants, yet this appears to facilitate their survival. One particularly clear signal from the list of mutants and their sensitivity is the involvement of mitochondrial function, with loss- and gain-of-function atfs-1 alleles having opposite effects and with every respiratory chain function mutant tested being resistant or highly resistant (Fig. 3). Some of the respiratory chain mutants have been found to be resistant to multiple stresses (Dues et al. 2017; Wu et al. 2018), which in some cases might be lowered by the loss of atfs-1 (Wu et al. 2018; Campos et al. 2021). We observed that under the conditions of the t-shifting test continuous growth at 25 °C is deleterious per se (Fig. 1c). And killing by t-shifting is more acute when the temperature range of the shifting involves higher temperatures (Fig. 1b). This indicates that under our assay conditions, higher temperatures lead to accelerated deterioration of the worms for unknown reasons. However, spending time at 25 °C is not required for killing (Fig. 1b) and, when identical time is spent at 25 °C and 15 °C, killing depends on the frequency of t-shifting (Fig. 1a). Furthermore, mutant heat sensitivity does not correlate with hypersensitivity or resistance to t-shifting (Fig. 5, c and d). A revealing observation is that when t-shifting between 15 °C and 25 °C is asymmetric, very little time spent at one of the temperatures is sufficient to lead to disproportionately large changes in survival (Fig. 1c). Given the very short time (3.4 s) it takes to move from one temperature to the next, it indicates that spending any time at 15 °C or 25 °C has long-lasting effects. What could these effects be? Some changes to gene expression due to temperature could be very rapid. For example, the rate at which RNA polymerases proceed, or stall, or even fall off DNA (Abbondanzieri et al. 2005; Mejia et al. 2008; Miguel et al. 2013; Kindgren et al. 2020; Plaskon et al. 2021). Temperature also affects the stability of RNAs (Becskei and Rahaman 2022). Similarly, the rate of ribosome translation affects the fate of the mRNA being translated (Kulkarni et al. 2019; Jaquet et al. 2022). Finally, protein aggregation or unfolding triggered by temperature could be the source of a signal that impacts gene expression via the UPR, UPRmt, and the assembly and disassembly of protein aggregates and stress granules (Walter and Ron 2011; Shpilka and Haynes 2018; Glauninger et al. 2022; Marijan et al. 2024). We believe our findings and the future studies of the underlying molecular mechanisms might apply to homeotherms such as humans as much as to poikilotherms. Humans use temperature-control mechanisms to attempt to keep a stable internal temperature. However, this applies mainly to the temperature of the most vital organs of the core, while peripheral tissues, e.g. skin, muscle, and entire limbs with all their tissue types, can experience and tolerate quite large fluctuations due to changes in environmental temperatures without producing acute injury (Romanovsky 2018). For example, exposed skin sustains freezing cold injuries only when skin temperature falls below −5 °C, and burn injuries occur only when the basal layer of the epidermis reaches at least 44 °C (Tikuisis 2004; Martin and Falder 2017). Moreover, peripheral vasoconstriction in response to cold to protect core temperature will decrease peripheral temperature even further (Tikuisis 2004; Tansey and Johnson 2015). Although keeping the core temperature stable is prioritized by temperature-control mechanisms, even the core temperature can vary (Taylor et al. 2014). Furthermore, conditions such as fever or intense physical exertion can elevate core temperature by several degrees, whereas injury, anesthesia, and starvation can decrease core temperature (Romanovsky 2018; Periard et al. 2021; Cramer et al. 2022). In other words, throughout life, most human cells will experience frequent excursions from the ideal temperature, which might be deleterious in ways that are neither appreciated nor sufficiently studied. Supplementary Material Acknowledgments Some strains were provided by the CGC, which is funded by NIH Office of Research Infrastructure Programs (P40 OD010440). Contributor Information Noah Lilienfeldt, Department of Biology, McGill University, Montreal, Canada H3A 1B1. Yifan Tai, Department of Biology, McGill University, Montreal, Canada H3A 1B1. Han Sun, Department of Biology, McGill University, Montreal, Canada H3A 1B1; Department of Biochemistry and Molecular Cell Biology, Shanghai Key Laboratory for Tumor Microenvironment and Inflammation, Shanghai Jiao Tong University School of Medicine, Shanghai 200025, China. Brandon Waddell, Department of Biology, McGill University, Montreal, Canada H3A 1B1. Wen Yang, Department of Biochemistry and Molecular Cell Biology, Shanghai Key Laboratory for Tumor Microenvironment and Inflammation, Shanghai Jiao Tong University School of Medicine, Shanghai 200025, China. Robyn Branicky, Department of Biology, McGill University, Montreal, Canada H3A 1B1. Siegfried Hekimi, Department of Biology, McGill University, Montreal, Canada H3A 1B1. Data availability The authors affirm that all data necessary for confirming the conclusions of this article are represented fully within the article and its tables and figures, including the Supplemental material available at GENETICS online. Supplemental material available at GENETICS online. Funding This work was supported by a grant to S.H. from the Canadian Institutes of Health Research, FDN-159916. H.S. was supported by a Fellowship Grant from the Visiting Programs for Graduate Students and Research Assistants of Shanghai Jiao Tong University School of Medicine. S.H. holds the Campbell Chair of Developmental Biology . Literature cited Associated Data Supplementary Materials Data Availability Statement The authors affirm that all data necessary for confirming the conclusions of this article are represented fully within the article and its tables and figures, including the Supplemental material available at GENETICS online. Supplemental material available at GENETICS online.