The Tumor Suppressor Rb Critically Regulates Starvation-induced Stress Response in C. elegans

Summary

How animals coordinate gene expression in response to starvation is an outstanding problem closely linked to aging, obesity, and cancer [1–5]. Newly hatched Caenorhabditis elegans respond to food deprivation by halting development and promoting long-term survival (L1 diapause), thereby providing an excellent model to study starvation response [2, 6, 7]. Through a genetic search, we have discovered that the tumor suppressor Rb critically promotes survival during L1 diapause and likely does so by regulating the expression of genes in both insulin-IGF-1 signaling (IIS)-dependent and -independent pathways mainly in neurons and the intestine. Global gene expression analyses suggested that Rb maintains the “starvation-induced transcriptome” and represses the “re-feeding induced transcriptome”, including the repression of many pathogen/toxin/oxidative stress-inducible and metabolic genes, as well as the activation of many other stress-resistant genes, mitochondrial respiratory chain genes, and potential IIS receptor antagonists. Notably, the majority of genes dysregulated in starved L1 Rb(−) animals were not found to be dysregulated in fed conditions. Together, these findings identify Rb as a critical regulator of the starvation response and suggest a link between functions of tumor suppressors and starvation survival. These results may provide mechanistic insights into why cancer cells are often hypersensitive to starvation treatment.

Results and Discussion

Loss of Rb Function Drastically Shortens L1 Starvation Survival

While searching for genetic regulators of starvation survival, we found that the L1 starvation survival (L1SS) rate of a loss-of-function (lf) mutant of lin-35, encoding the worm ortholog of tumor suppressor Rb [8], was dramatically reduced (Figures 1A and 1B). It has previously been shown that lin-35/Rb acts with SynMuv genes to regulate vulval cell induction, cell proliferation, pharyngeal development, RNAi sensitivity, and transgene silencing [8–13]. Our analysis of six Class B and two Class A SynMuv mutants (Figures S1A–S1B) suggests that lin-35/Rb does not act with most of these factors for its role in L1SS.

Figure 1. Tumor Suppressor Rb Functions to Promote L1 Starvation Survival by Affecting Both IIS Pathway-dependent and – independent Functions.

(A) Survival rate of the lin-35/Rb(lf) and daf-18/Pten(lf). Both mutants have a significantly reduced L1 starvation survival rate. The lin-35/Rb(lf) defect can be rescued by expression of a lin-35/Rb(+) transgene. Data for additional synMuv genes are shown in Figure S1A. All data presented as mean ± SEM.

(B) The mean survival calculated with the Kaplan-Meier method and log-rank statistical analyses of indicated mutants with decreased L1 starvation survival. ^ap-values were calculated by log-rank test and show the significance of the difference compared to the wild-type strain. ^bp-values were calculated to show the significance of the difference from the lin-35/Rb(lf) strain.

(C) Survival curves showing that the defect of lin-35/Rb(lf) was partially suppressed by unc-31(lf); age-1(rf) and overexpression (oe) of daf-16/FOXO. All data presented as mean ± SEM.

(D) daf-16(lf); lin-35/Rb(lf) double mutants displayed more severe defects than each single mutant. All data presented as mean ± SEM. Statistical analyses of data shown in (C) and (D) are reported in (B). Raw data for individual starvation survival experiments for Figure 1 are presented in Table S2.

lin-35/Rb is known to have broad functions in many cellular processes, so the mutant animals are not fully healthy. However, the poor L1SS of the lin-35/Rb(lf) animals is not likely due to a general sickness associated with the mutation, as we found that the lin-35/Rb(lf) animals displayed no obvious defects in pharyngeal pumping rate or survival rates upon exposure to pathogenic bacteria and osmotic stress (Figure S1F–S1H). Therefore, Rb likely plays important regulatory roles in cellular processes involved in starvation response.

Rb May Regulate Both IIS-dependent and – independent Functions for Starvation Survival

DAF-2 (IIS receptor), AGE-1 (PI3K), DAF-16 (FOXO protein) and their upstream regulator UNC-31 (calcium-activated regulator) have been shown to regulate L1SS [2, 6, 14] (Figure 1C). Consistent with these findings, a lf mutation in daf-18/Pten, which antagonizes PI3K activity [15], also dramatically reduced L1SS under two different culture conditions (Figure 1A) [16]. We then found that unc-31(lf) and age-1 reduction-of-function (rf) mutations were able to partially suppress the starvation survival defects associated with lin-35/Rb(lf) (Figures 1B and 1C). Furthermore, we consistently observed that overexpression of daf-16/FOXO, which has been shown to extend lifespan [3, 17], caused an increase in L1SS rates in lin-35/Rb(lf) mutants (Figures 1B and 1C). These partial suppression data suggested that lin-35/Rb might function in part by regulating the activity of IIS-independent mechanisms. The observation that the L1SS rates of lin-35/Rb(lf); daf-16(lf) double mutants were significantly lower than that of the single mutants (Figures 1B and 1D) further supports this notion. Therefore, lin-35/Rb likely regulates L1SS by regulating both IIS-dependent and IIS-independent functions.

Rb Affects the Expression of Many IIS-regulated Genes and Potential Insulin-IGF-1 Receptor Antagonists

We performed microarray analysis of genome-wide mRNA levels in lin-35/Rb(lf) to identify potential targets of lin-35/Rb during L1 diapause. Our data indicate that the expression of 1213 genes was significantly different in the mutant compared to wild type (Table S1A). Among them, 43% (517 genes) were down-regulated and 57% (696) were up-regulated. Interestingly, 70% of the 1213 genes dysregulated in starved lin-35/Rb(lf) animals did not overlap with the 1132 genes that were determined to be significantly dysregulated in fed lin-35/Rb(lf) L1 animals [18](Tables S1A and S1H), suggesting starvation-specific regulation of gene expression by lin-35/Rb.

We found that genes known to function in the steps between unc-31 and daf-16 within the IIS pathway (Figure 2A) were not among the genes that were significantly dysregulated in Rb(lf) animals (Table S1A). However, the comparison between Rb(lf) affected genes and daf-2/IIS receptor regulated genes [19] showed a statistically significant overlap of 58 genes, 38 of which had obverse expression changes between lin-35/Rb(lf) and daf-2(rf) mutants (Figure 2B; Table S1B). A total of 22 of the 38 genes, including known daf-16 targets (e.g. dod-19)[19], were repressed by lin-35/Rb but induced by IIS while the remaining 16 genes were induced by lin-35/Rb but repressed by IIS (Figure 2C; Table S1B). Therefore, lin-35/Rb shares common downstream targets with the IIS pathway.

Figure 2. Rb Regulates the Expression of Many IIS-regulated Genes, Potential IIS-regulating Genes, and Genes Regulated by Feeding and Starvation.

(A) A simplified diagram of the insulin receptor pathway’s known effect on L1 starvation survival [2, 6].

(B) Number of genes that changed expression in both daf-2(rf)[19] and lin-35/Rb(lf). The overlaps are significantly greater than the numbers expected by random chance. Changes caused by daf-2(rf) for most of the overlap genes were in the opposite direction of that by lin-35/Rb(lf) (see Table S1B). rf, reduction-of-function.

(C) Bar diagram showing representative genes that were dysregulated in lin-35/Rb(lf) in the opposite direction as that by daf-2(rf) [19].

(D) Expression changes of two potential IIS receptor antagonists in lin-35(Rb(lf). p-values shown in (B and C): * p < 0.05; **p < 0.01; ***p < 0.001.

(E) Number of previously reported FedUP or StarvUP genes that are also regulated by Rb. The expression patterns of these overlap genes were disrupted by Rb(lf) (Tables S1C and S1D). Statistical analysis of overlap expected by random chance and p-values in (B and E) are described in Supplemental Experimental Procedures.

Two genes encoding insulin-like proteins, ins-24 and ins-30, were down-regulated in lin-35/Rb(lf) (Figure 2D). Because both genes were shown to be up-regulated during L1 diapause and in dauer larvae of wild-type animals [20], they may function as the IIS receptor antagonists. Their down-regulation in starved L1 lin-35/Rb(lf) mutants is consistent with Rb’s role in repressing the IIS pathway. Overexpressing either ins-24 or ins-30 did not significantly improve L1SS of the lin-35/Rb(lf) mutant, suggesting only minor contributions by these genes to the Rb-mediated function in starvation survival.

Rb(lf) Severely Disrupts Known Starvation-induced Gene Expression Dynamics

We further compared the transcriptomes of starved lin-35/Rb(lf) mutants with two previously classified gene groups, FedUP and StarvUP [1]. FedUP genes are expressed at higher levels when animals consume food after a period of starvation; StarvUP genes are expressed at higher levels during starvation. Among genes dysregulated in starved lin-35/Rb(lf) animals, 175 (14.4%) overlap with FedUp genes (Figure 2E; Table S1C). More significantly, the vast majority (78%) of these 175 genes were up-regulated, indicating that lin-35/Rb functions to repress these genes during L1 diapause in wild type animals. Among genes dysregulated in starved lin-35/Rb(lf) animals, 140 (11%) were StarvUP genes (Figure 2E; Table S1D). The vast majority (86%) of these 140 StarvUP genes were down-regulated in the lin-35/Rb(lf) mutants, indicating that lin-35/Rb promotes the expression of these genes during L1 diapause in wild-type animals.

Our results, therefore, suggest that lin-35/Rb(lf) alters the “starvation transcriptome” toward a “feeding transcriptome”. In general, FedUP genes are proposed to promote reproductive growth, while StarvUP genes are proposed to support an animal’s survival during starvation [1]. Gene ontology analysis identified that up-regulated genes in lin-35/Rb(lf) mutants were enriched for larval development genes, whereas down-regulated genes were enriched for metabolism genes (Tables S1C and S1D).

Rb Represses the Expression of Many Pathogen-, Toxin-, and Oxidative Stress-responsive Genes

We found significant overrepresentation of pathogen, toxin, and oxidative responsive genes in the lin-35/Rb(lf) mutant microarray dataset (Figures 3A and 3B; Tables S1E–G). The overrepresentation of pathogen-responsive genes was seen in published datasets obtained from several pathogenic conditions [21–23]. The comparison against the 4-hour exposure to Pseudomonas aeruginosa strain PA14 [21] was further analyzed. 57 up-regulated genes in starved lin-35/Rb(lf) animals were PA14_4hr-inducible genes; this number is dramatically greater than expected by chance (Figures 3A and 3B; Table S1E). Similar findings were seen when we compared our microarray datasets with Cry5B toxin-responsive genes and oxidative stress-responsive genes [24, 25] (Figures 3A and 3B; Tables S1F and S1G). 75 of the Cry5B toxin-inducible genes were up-regulated in starved lin-35/Rb(lf) mutants, some of which are also pathogen/oxidative response genes (Tables S1E–G). These results suggest that the LIN-35/Rb protein represses expression of these pathogen-, toxin-, and oxidative stress-inducible genes to benefit starvation survival. In other words, while these genes, many encoding proteins with antimicrobial or detoxification roles, function to protect animals against various environmental threats, they are likely harmful to an animal’s ability to counter starvation. This role of lin-35/Rb appears to be distinct from that of the IIS pathway that plays a prominent role in animal responses to pathogen and toxin stresses [26]. Our recent studies also indicate sharply different miRNA functions in the intestine for starvation survival versus pathogen response [14, 27].

Figure 3. Rb Regulates the Expression of Many Genes That are Induced by Pathogen, Toxin, or Oxidative Stress, and Rb(lf) Mutants are Defective in Oxidative Stress Response.

(A) Number of previously characterized pathogen (Pseudomonas aeruginosa PA14)-inducible genes [21], cry5B toxin-inducible genes [24], and oxidative stress-inducible genes [25] that are up-regulated by Rb(lf) (see Tables S1E–G). These data suggest Rb represses these genes in wild type to promote starvation survival. The calculation of overlap expected by random chance and p-values shown are described in Supplemental Experimental Procedures.

(B) List of representative toxin, pathogen and/or oxidative stress inducible genes that are upregulated in Rb(lf) (See Tables S1E–G for complete lists).

(C) Changes in expression of a group of glutathione transferase genes in lin-35/Rb(lf) mutants. These genes are known to be required for a normal oxidative stress response [25]. P values: * p < 0.05; **p < 0.01; ***p < 0.001.

(D). Rb(lf) worms are hypersensitive to paraquat treatment, indicating that they are defective in oxidative stress response. Day 1 starved L1 animals were treated with or without 1 mM paraquat and examined for viability 60 hours later. p-values were generated by t-test. Error bars, SEM.

Notably, some other oxidative stress-inducible genes, particularly several glutathione transferase genes that are known for critical roles in oxidative stress response [28], were dramatically down-regulated by lin-35/Rb(lf) (Figure 3C). Further tests showed that lin-35/Rb(lf) mutants were supersensitive to paraquat treatment during L1 diapause (Figure 3D), indicating that Rb plays a role in the oxidative stress response and does so partly by promoting the expression of this group of oxidative stress-responsive genes.

Rb Regulates Starvation Survival Partly by Promoting Expression of Mitochondrial Respiratory Chain Proteins

Comparison of our lin-35/Rb(lf) microarray dataset with the previously identified C. elegans mitochondrion (mt) proteomic dataset [29] showed a statistically significant overlap of 120 genes, 67% of which were down-regulated in lin-35/Rb(lf) during starvation (Figure S2A; Table S1I). Many of these genes encode components of various mt respiratory chain (MRC) complexes including those required for normal adult lifespan (e.g., gas-1)[30]. Intriguingly, among these mt genes down-regulated by lin-35/Rb(lf), some repress extended adult lifespan (e.g., clk-1)[31–33]. These genes may function differently during L1 starvation, as clk-1(lf) and isp-1(lf) have been reported to reduce L1SS [2]. Consistent with reduced MRC function in lin-35/Rb(lf), the expression of hsp-6 was significantly up-regulated (3.48 fold) in starved lin-35/Rb(lf) animals. HSP-6, a mt-associated protein chaperone, is an indicator for mt-specific unfolded protein response activities [34]. We also observed that wild-type and lin-35/Rb L1 starved animals are similarly sensitive to potassium cyanide, a potent MRC inhibitor (Figure S2B), confirming that MRC plays critical roles in L1SS. Thus, lin-35/Rb may positively regulate starvation survival partly by promoting MRC activity, but lin-35/Rb(lf) animals with altered MRC gene expression are still sensitive to complete MRC inhibition by cyanide. Our further tests suggested that lin-35/Rb(lf) does not affect mt biogenesis (Figure S2C).

Glucose Supplementation Strongly Rescues L1 Starvation Survival of IIS Pathway Mutants, but Only Weakly Rescues That of lin-35/Rb(lf)

Since mitochondria are crucial organelles for energy production, including glucose metabolism [35], the dysregulation of MRC genes in lin-35/Rb(lf) animals suggests that energy production may be compromised. We thus tested the effect of glucose supplementation on lin-35/Rb(lf) L1SS, and several other mutants that have shortened L1SS but are not known to compromise MRC functions: daf-16/FOXO(lf), daf-18/PTEN(lf)), aak-2/AMPK(lf)), and gpb-2/Gb(lf) [6, 7, 16, 36]. While the L1SS of these mutants was dramatically improved by glucose supplementation (2%), the effect on lin-35/Rb(lf) animals was significantly weaker (Figures 4A and 4B). This result is consistent with a model that lin-35/Rb(lf) significantly reduced the energy production capacity through decreased utilization of glucose, likely by compromising MRC functions. Alternatively, lin-35/Rb(lf) animals may have reduced metabolic rates compared to other mutants and thus demand less energy to survive starvation. This model may be inconsistent with that the transcriptome of lin-35/Rb(lf) resembles that of well-fed worms with higher metabolic rate than starved animals.

Figure 4. Rescue Effect of L1 Starvation Survival by Glucose Supplementation and Tissue-specific Expression of lin-35/RB.

(A) Survival rate of the Rb(lf) and other indicated mutants supplemented with or without 2% glucose. All data presented as mean ± SEM.

(B) Mean survival as calculated with the Kaplan-Meier method, and p-values as calculated with the log-rank statistical analyses, of indicated strains supplemented with or without 2% glucose. p-values are relative to the same strain without supplementation of glucose. Glucose supplementation has a significantly weaker rescuing effect on the L1 survival rate of lin-35/Rb(lf) than that of other mutants tested.

(C) Survival rate of lin-35/Rb L1 mutant animals carrying extrachromosomal arrays expressing wild-type lin-35/Rb driven by three tissue specific promoters (intestine, hypodermis, and pan-neurons) and the lin-35 promoter. Transgenic animals were scored based upon the expression of a co-injection marker using a Leica fluorescence microscope. The average from multiple independent transgenic lines for each genotype is reported with the standard error of the mean for each time point (±SEM).

(D) Mean survival rates calculated with Kaplan-Meier method. ^a,b,c p-values were calculated by log rank test showing the significance of the difference from wild type, lin-35/Rb(lf) and lin-35/Rb(lf) rgef-1P::lin-35, respectively. Raw data for individual starvation survival experiments for Figure 4 are presented in Table S2.

Rb plays prominent roles in Neurons and the Intestine to Promote L1 Starvation Survival

lin-35/Rb is broadly expressed in nearly all cells in newly hatched L1 larvae [8]. As shown in Figure 1, full length genomic lin-35/Rb DNA rescued L1SS of lin-35/Rb(lf) mutants. We used tissue-specific promoters (elt-2p, rgef-1p and dpy-7p) to express the functional LIN-35/Rb protein in three major tissues/cells: intestine, neurons and hypodermis. Expression of lin-35/Rb in the neurons or intestine partially restored the L1SS of lin-35/Rb(lf) animals (Figures 4C and 4D), identifying neurons and the intestine as important sites for this specific lin-35/Rb function. Furthermore, neuronal expression consistently showed a stronger rescue effect than intestinal expression, and transgenic animals carrying both elt-2P::lin-35 and rgef-1P::lin-35 constructs showed similar rescue effect as animals carrying rgef-1P::lin-35 alone (Figures 4C and 4D), suggesting lin-35/Rb activity in the neurons is more prominent for L1SS. Global gene expression analysis identified that 13% (93) of the total dysregulated genes in lin-35/Rb mutants are expressed in neurons [37], and 85 of these genes are down-regulated by lin-35/Rb(lf) (Table S1J). The partial rescue effects of neuronal and intestinal expression suggest that lin-35/Rb expression in additional tissues likely contributes to its function in L1SS. lin-35/Rb expression in the hypodermis is critical for its role in regulating EGF/Ras signaling during vulval development [8, 38, 39], but failed to rescue the L1SS phenotype (Figures 4C and 4D), supporting that Rb function in L1SS is distinct from previously identified functions.

Concluding remarks

Our study suggests that lin-35/Rb likely executes its role in starvation response in neurons and other tissues by regulating the expression of a large number of genes involved in multiple cellular pathways and mechanisms, including both IIS pathway-dependent and – independent mechanisms. Additionally, Rb’s function in L1 starvation response is largely independent of its well-known role in regulating the expression of cell cycle regulators (Figures S1A–E).

Our results may provide important insights into mammalian Rb function in nutrient/starvation responses and the validity of fasting-coordinated cancer therapy [5]. The fact that mutations in tumor suppressor genes Rb, Pten and FOXO have profound impacts on starvation responses may lead to two non-mutually exclusive implications regarding the relationship between starvation response and cancer. On the one hand, cancer cells have likely lost the ability to maintain a normal starvation-induced transcriptome, analogous to what is shown in the C. elegans lin-35/Rb(lf) mutant in this study, rendering cancer cells supersensitive to starvation. On the other hand, maintaining a “fit” transcriptome capable of dynamic changes in response to stress may be pivotal for suppressing tumorigenesis. Regardless, the tumor suppressor roles of these genes are likely to be far beyond just inhibiting undesired cell proliferation and growth.

Highlights.

• Rb plays critical roles in promoting L1 starvation survival.

• Rb may function in both IIS-dependent and -independent pathways.

• Rb represses many genes during starvation that are induced by other stresses.

• Rb promotes mitochondrial respiratory chain functions during starvation.