A point-of–no-return leading to death during heat-shock in C. elegans

Longevity is a complex trait determined by genes, the environment, and their interactions. There is considerable insight into the genes associated with longevity and the interplay with environmental conditions. Most genes involved in the stress response play a major role in determining longevity. Yet, there is limited understanding of the mechanisms that determine how long stress can be tolerated before death becomes inevitable. Here, we leveraged the detection of an irreversible switch to death by studying global gene expression profiles in combination with survivorship following heat stress in the nematode C. elegans. By analysing the transcriptional response in a high-resolution time series of increasing stress exposures, we found a distinct shift in gene expression patterns between 3-4 hours into the stress response, separating an initially highly dynamic phase from a later mostly stagnant phase. Remarkably, this turning point in expression dynamics coincided with a phenotypic point of no return, as shown by a strong decrease in movement, survival and, progeny count in the days after; ultimately leading to death.


28
Longevity differs strongly within and between species. Over the past few decades detailed insight has 29 been obtained about the genetic and environmental factors that determine longevity. Many genes have 30 been identified that play a key role in lifespan determination, many of which are involved in 31 environmental stress response pathways. Environmental stress is a potentially harmful condition 32 beyond the optimum range of the organism, for example through shifts in temperature and exposure to 33 oxidants, or toxic compounds [1,2]. On a cellular level, these environmental stressors can interfere 34 with protein homeostasis, leading to an accumulation of misfolded proteins and protein aggregates 35 [3]. To avoid the detrimental effects of misfolded proteins and protein aggregates, multiple stress 36 response systems have evolved as a first line of defence to maintain proteostasis, of which the highly-37 conserved heat-shock response (HSR) pathway is prominent [3,4]. The accumulation of misfolded 38 proteins is a hallmark of aging and age-related diseases such as Alzheimer's and Parkinson's disease 39 [5][6][7]. The connection between the processes involved in stress and aging is further substantiated by 40 the fact that several components of the stress response pathways were found to function as regulators 41 of lifespan [8,9]. For example, the evolutionary highly conserved transcription factor HSF-1 is a key 42 component in the initiation of the HSR, as well as a regulator of lifespan [10]. Therefore, 43 understanding how an organism perceives and handles stress is fundamental for understanding the 44 molecular mechanisms that underlie aging [6]. 45 The nematode Caenorhabditis elegans is an established metazoan model for studying the 46 effect of -and response to -stress and aging in vivo [6,8-10]. One of the most widely studied stress 47 responses in C. elegans is acute heat stress, which can be easily applied by exposing the animal to 48 temperatures between 33-37 o C [10][11][12]. Often, the effects of the stress are quantified on a phenotypic 49 level by recording complex traits such as survival rate, mobility, and reproduction [12][13][14]. Generally, 50 the inflicted damage accumulates with increasing temperature and exposure time. For example, brood 51 size decreases with moderate increases in temperature beyond the optimum [14,15], whereas a strong 52 decrease in survival rates is only observed after prolonged exposures to heat stress [12,16,17]. It was 53 shown that C. elegans detects and responds to heat stress via transient receptor potential channels and 54 a neuropeptide signaling pathway [18]. At the level of the transcriptome, a heat shock induces a 55 strong response. Genome wide gene expression analysis in C. elegans shows that a two hour exposure 56 to 35 o C affects genes associated with development, reproduction and metabolism [19]. Furthermore, 57 an exposure of 30 minutes to 33°C already induced a massive global gene expression shift highly 58 dependent on HSF-1, affecting genes associated with a wide range of functions such as cuticle 59 structure, development, stress response, and metabolism [20]. 60 Yet, there is limited understanding of the mechanisms that determine how long stress can be 61 tolerated before death becomes inevitable. Given the range of phenotypic effects, it is to be expected 62 that the transcriptional response during heat stress is highly dynamic. For example, the initial 63 transcriptional response to heat shock probably does not resemble the transcriptome after a lethal 64 exposure to heat stress. To gain more insight into the underlying dynamics of the stress response, we 65 have generated a high-resolution time-series of transcriptomic and phenotypic data of C. elegans 66 exposed to heat stress conditions at 35 o C for 0-12h. Transcriptomic analysis revealed a global shift in 67 expression dynamics occurring between 3 and 4 hours into the heat exposure. The shift marks the end 68 of an initially highly dynamic transcriptional response to heat stress that plateaus at longer exposures. 69 On a phenotypic level, longer exposures (> 4h) were associated with low chances of recovery, thus 70 indicating that the critical shift observed in the global gene expression marks a point of no return 71 ultimately leading to death. 72

73
Transcriptional variation during prolonged heat stress 74 We first assessed the impact of heat stress durations on genome-wide expression levels. Wild type 75 Bristol N2 populations were exposed to heat stress conditions at 35 o C for increased exposure 76 durations between 0.5-12 hours ( Figure 1A). To find the main sources of variation during the 77 transcriptional response to heat-shock, we used principal component analysis (PCA). The first two 78 principal components (PCs) captured 77% (1 st 57%, 2 nd 20%) of the total variation ( Figure 1B). The 79 first PC sorted the time points in chronological order, showing that variation in gene expression 80 between samples was largely due to the increasing length of heat exposure. Furthermore, the distance 81 along the 1 st PC was larger for early time points in comparison to later exposure times, indicating that 82 a large part of the changes in gene expression occurred early in the stress response. Together, the 1 st 83 and 2 nd PCs indicated 3-4 hours of heat exposure as a turning point in transcriptional patterns during 84 the prolonged stress response. 85 Bristol N2 populations were grown at 20°C for 46 hours before the start of the heat-shock at 35°C. Clock-symbols indicate the time of sampling for subsequent transcriptome analysis of the dynamic stress response. Each time point (0, 0.5, 1, 2, 3, 4, 6, 8, and 12 hours) was sampled 3-5 times. (B) Principal component analysis of gene expression data averaged per time point. The first two components retain 77% of the variation in the data set, and placed the exposure duration (as indicated by the clock symbol) in chronological order.

Changes in gene expression reach a plateau 106
We further investigated the temporal dynamics of global transcriptome changes. About ~6200 (~30%) 107 genes contributed significantly (q-value < 0.01) to the variation explained by the first two principal 108 components. This sub-set was used as input for k-means clustering to extract common patterns in 109 gene expression changes, identifying six distinct stress-response groups (Figure 2A-B). Cluster 1, 4, 110 and 6 (representing ~3790 (60%) of the genes) contained genes downregulated during exposure to 111 heat (Figure 2A), and cluster 2, 3, and 5 contained upregulated genes (~2450 (40%) of genes; Figure  112 2B). The largest changes were found in cluster 1 and 3 with an average 5-fold down-and 32-fold up-113 regulation, respectively. 114 Within these clusters, the initial in-or decrease in transcript levels started rapidly, between 115 0.5-1 hour after initiation of the heat stress exposure, and reached a plateau before 3 hours into the 116 stress response. One exception is cluster 1, consisting of 420 genes that were down-regulated after 4 117 hours. Interestingly, this was the only pattern clearly distinguishing the later (>4h) time points. As 118 previously indicated by the PCA, the transcriptional patterns reveal a global change in expression 119 dynamics after approximately 3-4 hours into the stress response, starting with a highly dynamic 120 adaptive phase and ending with a plateau phase of minimal overall changes. 121 To explore the biological functions associated with the gene sets within the individual 122 response clusters, an GO-enrichment analysis was performed (Figure 2C-D; Supplementary table 123 S1). Overall, the down-regulated clusters were enriched with structural constituents of the cuticle, 124 particularly collagens (col, dpy, rol, sqt), as well as genes associated with transcription (nhr), 125 metabolic processes, and locomotion ( Figure 2C). In the upregulated clusters, genes involved in 126 nucleosome assembly (his) were found to be overrepresented, as well as those regulating embryo and 127 larval development ( Figure 2D). Cluster 3, the smallest group (54 genes), had an immediate and 128 strong reaction to the stress. This cluster could not be associated with an enrichment term. Half of the 129 genes within this cluster have not previously been classified with any GO term yet are very likely 130 involved in the response to heat stress. 131 Through transcriptome analysis, we identified a critical time point around 3-4 hours into the stress 134 response, separating an initially highly dynamic phase from a later mostly stagnant phase. Next, we 135 tested how these observed transcriptional patterns correlate with the effects of increasing heat stress 136 durations on the phenotypic recovery of the animals. To measure the effects, we observed survival, 137 progeny count, and movement in populations that were allowed to recover at 20°C after different heat 138 stress durations ( Figure 3). Since it has previously been shown that it can take three days after the 139 exposure to a transient lethal heat-shock to observe the fatal effects in the survival scores of C. 140 elegans [12], we recorded daily phenotypic observations over a four day recovery period following 141 the stress. 142 The heat exposure durations resulted in three phenotypically distinct groups. First, for 143 survival, the animals exposed to heat for up to two hours show high survival chances equal to the 144 control ( Figure 3A). An intermediary group was formed by animals exposed to heat for 3-4 hours 145 with about 80% surviving the first day, which steadily declined to ~60% survival by day 4. It is of 146 note that the exposure duration of this group coincides with the critical time point (3-4h) identified in 147 the transcriptomic data. In the third group, with heat-exposures over 6 hours, survival chances were 148 already drastically reduced after the first day (<20%). 149 Analogous to the 3 distinct survivorship groups found for short-, intermediate-and long-term 150 stress exposures, this division was also present in the fraction of nematodes regaining a healthy 151 movement during the recovery period, as well as regaining a normal number of progeny ( Figure 3B  152 and 3C). The movement in populations exposed to a short heat stress (< 3 hours) did not differ from 153 that of control populations. While the heat stress initially causes slightly lower numbers of progeny, 154 the reproduction peeked 2-3 days after the heat stress together with the control population. For   development. Next, we selected gene clusters with a strong decrease in expression levels ( Figure 4B). 174 While most of the transcriptional patterns differed between development and heat stress conditions, 175 about 20% of genes were present in both groups. An enrichment analysis of these genes found a 176 strong overrepresentation of genes associated with the cuticle structure and locomotion. 177 Together, these results showed that heat stress disrupted the major transcriptional changes delay. Furthermore, it shows that the animal almost fully shifts its transcriptional program to deal with 180 the acute heat stress conditions. 181 with distinct trends in the ability to recover from the stress, presenting a 'point of no return' as seen 203 by a drastic decline in survival rates, the ability to recover normal movement, and to produce viable 204 offspring in the four days of recovery following the stress. 205 To our knowledge, this is the first study that links the dynamics of heat stress response at the 206 transcriptome level to the ability to recover. Gene expression regulation under stress conditions is 207 strictly controlled, its kinetics are rapid and very often it is reversible. This allows for extremely rapid 208 adaptation of cells and tissues in response to general stress, in particular heat stress, and for returning 209 to a baseline level [21]. We analysed the phenotypic recovery from these rapid adaptive changes 210 occurring during stress and found that already a relatively early response to heat stress abruptly 211 changes development. 212

213
Early dynamic response to heat-stress disrupts development 214 During the early phase of heat stress, the transcriptional response is highly dynamic. About 400 genes 215 (cluster 2 and 5) are highly upregulated. Comparison with transcriptional patterns normally occurring 216 during development has shown that this gene-set uniquely reacts in the response to stress. 217 Furthermore, genes highly active during development show low transcriptional changes during stress 218 conditions. These results indicate that the animal almost fully switches its transcriptional focus on proteins are the first set of genes to show a strong and rapid increase in transcript levels. Shortly after, 221 histones and genes associated with the nucleosome assembly are highly enriched in upregulated gene 222 clusters. Nucleosome remodelling has previously been shown to be an important part of the stress 223 response, e.g. by allowing access to transcription sites of stress responsive genes [21, 23,24]. In C. 224 elegans, depletion of a nucleosome remodelling complex leads to a higher thermal sensitivity [25]. 225 Packaging of DNA into nucleosomes could be an additional protective mechanism during the stress 226 response. 227 Phenotypic analysis of the recovery process indicates that the disruption of normal 228 transcriptional processes is fully reversible up to the 'point of no return'. After short exposures to 229 stress, the animals recovered a healthy movement phenotype and started reproducing, indicating that 230 the protective mechanisms put in place by the early transcriptional heat shock response are sufficient 231 in this time frame. However, the disruption of normal transcriptional development could be one of the 232 causes for the observed delay in reproduction. A study in which a two hour 35 o C heat-shock was 233 compared to two hour recovery of that heat-shock showed that the transcriptional patterns in the 234 recovery population had still not returned to normal [19]. Also, a delay in reproduction has previously 235 been shown in heat-shocked pre-gravid adult C. elegans exposed to temperatures between 30-32 o C 236 [15]. Arresting reproduction ensured limited damage to reproductive compartments during stress 237 conditions. In the early heat stress response, we found this delay on a transcriptional level as 238 development and reproduction related genes did not show their normal up regulation. 239 240

Attenuation of dynamic response 241
At medium-to-long exposures, the transcriptional stress response attenuates corresponding 242 phenotypes, i.e. a ~40% decrease in survival and an increased occurrence of animals with an abnormal 243 movement phenotype. The attenuation of the heat shock response has mostly been studied in several 244 cell lines [26,27]. An integral part is the activation and subsequent suppression of the HSF-1 245 transcription factor activity through a negative feedback loop, which is partially mediated by those 246 chaperones that are transcriptionally induced by HSF-1, such as HSP-70 [28]. The attenuation of the 247 heat shock response is believed to serve a protective function, as cell lines with defects in the process display lower growth rates and reduced fitness [21,27]. In C. elegans, it was shown that a gain-of-249 function mutation in a negative regulator of the heat-shock response (HSB-1) results in severe effects 250 on survival after heat stress [26]. In our data set, transcripts of chaperones induced by HSF-1 increase 251 immensely within the first 30 minutes of the stress response (Supplementary Figure S1). The drastic 252 increase slows down until peak levels are reached at 4 hours into the stress response, followed by a 253 small decrease and complete attenuation. It is unclear if the observed global transcriptional slowing 254 down is due to an actively regulated process, such as the HSF-1 feed-back loop, or due to a passive 255 process, such as the accumulation of damage to key cellular processes. Another explanation might be 256 a developmental cue. During normal development without stress, the C. elegans transcriptome is 257 Overall, our study links a strong shift in transcriptional dynamics upon exposure to heat stress 274 with an inability to recover from the stress response. The inability to recover was reflected in a 275 decrease in worm activity, progeny count, and survival in the days after. Therefore, we think this 276 critical shift in the dynamics of gene expression marks a point of no return ultimately leading to death. 277 The selected traits (movement, survival, and progeny count) were observed using a stereomicroscope 300 at approximately 24, 48, 72, and 96 hours post heat-shock. To allow for accurate scoring of all 301 individual animals, the population size per dish was kept at a maximum of 25 animals at the start of 302 the experiment. In total, 3 dishes per heat-shock duration were scored, which amounts to a total of 303 approximately 60 animals per treatment. Animals were transferred to fresh NGM dishes every day 304 during the reproductive phase using a platinum wire. Bagging and suicidal animals were censored.

Movement and Survival 307
Movement was scored based on classification systems that have previously been described in 308 association with aging studies, where it acts as a measure of the biological age [33,34]. These systems 309 were combined and adapted to score the impact of the heat-shock. Healthy nematodes are actively 310 moving in a sinusoidal pattern (Hosono: type I; Herndon: Class A). As a result of the heat shock, a 311 proportion of the animals deviated from the healthy phenotype in varying degrees such as visibly 312 lower levels of activity, low responsiveness to touch with a platinum wire and/or an irregular shape of 313 movement (for example due to a partially paralysed tail). This is corresponding to Class B and C of 314 Herndon or Type II and III of Hosono). Worms were scored as dead, when no head movement was 315 observed after 3 touches with a platinum wire. 316 317

Progeny count 318
It has previously been shown that C. elegans can lay non-viable eggs after heat shock [12].    Table S1: Gene lists used for GO enrichment analysis, and detailed output of the enrichment analysis 423 performed with the functional annotation tool provided by DAVID 6.8. 424