MONITTR allows real-time imaging of transcription and endogenous proteins in C. elegans

All organisms experience stress conditions, such as nutrient deprivation, temperature fluctuations, and ER stress. Each type of stress triggers specific cellular responses and modulates signaling pathways for adaptation and survival (de Nadal et al., 2011; Higuchi-Sanabria et al., 2018; Pakos-Zebrucka et al., 2016; Russell and Lightman, 2019). Understanding stress response is crucial for uncovering the mechanisms behind stress adaptation and developing novel strategies to enhance resilience. Gene expression changes are vital to stress responses, involving alterations in metabolism, protein homeostasis, and enzymatic activity. The control of gene expression is tightly regulated, allowing rapid adjustments in transcriptional capacity during stress and subsequent return to basal levels (Almanza et al., 2019; Himanen and Sistonen, 2019; Joutsen and Sistonen, 2019; Mahat et al., 2016; Richter et al., 2010; Vihervaara et al., 2018). Genome-wide analyses of transcription, nucleosome accessibility, and protein–chromatin interactions have provided valuable insights into the regulatory steps of gene expression in response to stress (Dukler et al., 2017; Gasch et al., 2000; Lyu et al., 2018; Mueller et al., 2017; Ni et al., 2009; Niskanen et al., 2015; Vihervaara et al., 2017; Wissink et al., 2019). However, many aspects remain elusive, such as the tissue-specific regulation of stress-responsive genes and the roles of specific transcription factors in this process. Furthermore, it is unclear whether each cell or individual animal in a population responds to a stress signal similarly or distinctly.

Nutrient deprivation in a short period, also known as fasting, can induce a protective state in cells, enhance resistance to environmental stresses and toxicities (Hakvoort et al., 2011; Li et al., 2020; Longo and Mattson, 2014; Mitchell et al., 2010; Raffaghello et al., 2008), and promote tissue regeneration (Kurtovic et al., 2022; Serger et al., 2022; Yousefi et al., 2018) and longevity (Bishop and Guarente, 2007; Fontana and Partridge, 2015; Honjoh et al., 2009; Panowski et al., 2007; Uno et al., 2013). In mammals, most fasting responses are regulated through transcriptional modulation. Increased demand for fat utilization triggers the upregulation of genes involved in fatty acid oxidation while concurrently repressing glycolysis genes to maintain glucose supply. Dysregulation of the fasting response pathway can potentially lead to diseases such as diabetes and obesity (Bideyan et al., 2021; Goldstein and Hager, 2015; Scholtes and Giguère, 2022; Zinke et al., 2002). Although the fasting response is crucial for animals, the molecular mechanisms governing the transcriptional regulation of fasting-induced metabolic genes remain not fully understood. At the tissue level, there is a need for further study on how individual cells respond to fasting signals and dynamically modulate the expression of metabolic genes in time and space. The nematode Caenorhabditis elegans (C. elegans) is an ideal model organism due to its conserved nutrient-sensing mechanisms and metabolic networks (Greer et al., 2007; Jo et al., 2009; Li et al., 2020; Van Gilst et al., 2005).

Transcriptional reprogramming during stress can be a rapid and transient process within minutes (Dukler et al., 2017; Vihervaara et al., 2017). Traditional methods such as quantitative reverse transcription-PCR (qRT-PCR), RNA sequencing, and protein imaging are inadequate for accurately capturing real-time transcriptional dynamics. Therefore, novel approaches are needed for single-cell tracking of transcriptional dynamics in specific tissues and timeframes. Visualizing RNA in living cells is crucial for real-time RNA detection (Buckley and Lis, 2014; Chao and Lionnet, 2018; Das et al., 2021; Le et al., 2022; Rath and Rentmeister, 2015; Sato et al., 2020; Wu and Jaffrey, 2020). The MS2/MCP system, which involves fusing a fluorescent protein with the bacteriophage MS2 coat protein (MCP), is widely used for RNA visualization (Bertrand et al., 1998; Larson et al., 2011; Wu et al., 2015). In the case of live-cell imaging in C. elegans, MS2 has been placed in the untranslated region (UTR) of target genes, which allows dual labeling of nascent and mature RNAs (Hu et al., 2023; Lee et al., 2019; Li et al., 2021). However, it is challenging to specifically monitor nascent RNAs for measuring transcriptional dynamics. Moreover, the presence of MS2 sequences in the UTR may significantly reduce the expression of genes of interest (Lee et al., 2019; Tocchini and Mango, 2024; Xu et al., 2020). Therefore, it is highly desirable to achieve real-time monitoring of mRNA synthesis from endogenous genes without affecting their expression and function in multicellular organisms.

Here, we developed MONITTR (MS2-embedded mCherry-based monitoring of transcription), a new methodology devised with a dual-functional gene tag, DualTag. This system allows for simultaneous visualization of nascent RNA and protein levels. Our findings demonstrate that labeling nascent RNA accurately captures real-time transcriptional changes while minimally affecting the expression of endogenous target genes. Notably, MONITTR represents a valuable tool for precise quantifying endogenous gene expression at single-animal, single-tissue, single-cell, and single-allele levels. This capability enables a deeper understanding of the molecular mechanisms underlying transcriptional regulation of genes in response to various stress conditions. Moreover, our observations reveal the basic principles of how transcription factors regulate transcriptional bursting upon fasting in individual nuclei and specific tissues.

Most previous studies utilizing the MS2/MCP-based RNA tagging method have placed the MS2 sequence in the UTR to label nascent and mature RNA, which may cause defects in RNA stability, transport, and subsequent protein synthesis (Falo-Sanjuan et al., 2019; Garcia et al., 2013; Lee et al., 2019; Li et al., 2022; Lionnet et al., 2011; Lucas et al., 2013; Tutucci et al., 2018). Thus, we sought to develop the DualTag strategy by integrating the 24xMS2 sequence within one of the introns of the mCherry coding sequence to capture dynamic changes in transcription and protein expression levels simultaneously. By incorporating the mCherry^DualTag at either the N- or C-terminus of the target gene of interest, we can monitor the production of nascent RNA using MCP::GFP. To assess any potential impact of MS2 knockin on target gene expression, we employed CRISPR-Cas9-based gene editing technology to insert the coding sequence of mCherry, mCherry^DualTag, or mCherry-MS2 (3′UTR) at the N- or C-terminus of the target gene. These edits result in the fusion of the target gene’s protein with mCherry, enabling direct imaging, and quantitative measurements of protein levels (Fig. 1 A). These genes consisted of two fasting-induced genes, icl-1 and adh-1, and three fasting-repressed genes, acdh-1, hach-1, and acdh-2, which have been previously reported and were validated by our qRT-PCR data (except acdh-2, which was not tested) (Fig. S1, A and B). semo-1 was selected as an additional target gene, which has not been reported to respond to fasting. Through quantitative confocal imaging analysis, we observed that placing MS2 within the intron of mCherry had a significantly smaller impact on target protein expression than placing it in the 3′UTR (Fig. 1, B and C). This finding was further confirmed through qPCR experiments by quantifying mRNA abundance (Fig. 1 D). Therefore, our DualTag design theoretically enables simultaneous monitoring of the synthesis of nascent RNAs and protein levels with minimal impact on target gene expression.

To achieve nascent RNA labeling, we took advantage of the endogenous lmn-1 promoter to simultaneously express MS2 coat protein (MCP)::Superfolder GFP (sfGFP or GFP hereafter) to label the MS2-tagged nascent transcripts and BFP::LMN-1 to lighten up the nuclear membranes (Fig. 2 A). We first compared fusing a nuclear localization signal (NLS) to MCP (NLS-MCP::GFP) and using MCP::GFP without NLS. The NLS effectively increased the localization of MCP::GFP within the nucleus, primarily enriched in the nucleolus. However, the large size of the nucleolus posed significant challenges for observing RNA labeling in the nucleoplasm. In contrast, MCP::GFP was evenly distributed in the cytosol and nucleus without any dot-like signal (Fig. S1, C and D). Thus, we decided to use the design without an NLS signal, which was also used in two recent studies (Lee et al., 2019; Tocchini and Mango, 2024). We crossed gene of interest (goi)::mCherry^DualTag strains with the lmn-1prom::MCP::GFP::SL2::BFP::LMN-1 strain, which would enable simultaneous imaging of nascent RNA and protein of target genes (Fig. 2, A and B).

We first examined two fasting-induced metabolic genes, icl-1 and adh-1. The mitochondrial protein ICL-1 serves as a pivotal enzyme in the glyoxylate pathway, facilitating the conversion of fatty acid-oxidation products into sugars (Erkut et al., 2016; Van Gilst et al., 2005). adh-1 encodes alcohol dehydrogenase involved in alcohol metabolism and located in the cytoplasm (Ghaddar et al., 2023). We observed that their basal transcription levels were relatively low under well-fed conditions. However, after 8 h of fasting treatment, we observed the presence of 1–4 MCP::GFP-positive spots within numerous epidermal nuclei. These signals likely represent the ongoing production of nascent mRNAs at the transcription sites. Importantly, it should be noted that the nucleus of C. elegans epidermal cells are either diploid or tetraploid, and most of them are syncytia. For example, hyp7 in adult animals, one of the largest epidermal cells, contains 139 nuclei (Chisholm and Hsiao, 2012; Hedgecock and White, 1985). Upon the addition of α-amanitin, a specific RNA Polymerase II inhibitor, the signals for the putative nascent RNAs completely disappeared, supporting that the observed MCP::GFP labeled signals likely represent nascent RNAs (Fig. 2, C–F).

Next, we examined two fasting-repressed metabolic genes, acdh-1 and hach-1. acdh-1 encodes a mitochondrial enzyme that catalyzes the first step of fatty acid beta-oxidation and plays a crucial role in energy production (Van Gilst et al., 2005). HACH-1, another mitochondrial enzyme, is involved in the valine catabolic process (Fox et al., 2022; Sternberg et al., 2024). These genes exhibited higher transcription levels under well-fed conditions, with ∼78% and 27% of nuclei in epidermal cells actively transcribing for acdh-1 and hach-1, respectively. However, upon treatment with the Pol II inhibitor, the MCP::GFP spots significantly diminished, further demonstrating the specificity and efficiency of the DualTag-based labeling for nascent RNA (Fig. 2, G–J). Furthermore, it is worth noting that the subcellular localization of ICL-1::mCherry, ADH-1::mCherry, ACDH-1::mCherry, and HACH-1::mCherry was consistent with predictions shown on Wormbase, indicating the effectiveness of DualTag in simultaneous labeling of nascent RNA and protein.

To test the effectiveness of our DualTag-based nascent transcript detection method in C. elegans embryos, we observed the DualTag labeling of acdh-1 loci. We found that nascent RNAs could be detected as early as the gastrula stage, indicating that the transcription of acdh-1 begins at this stage. However, ACDH-1 protein was only visible from the pre-comma stage and continued to increase as embryonic development progressed. Quantitative analysis revealed that the number of actively transcribing acdh-1 loci per cell at each embryonic stage ranged from 0 to 2, while MCP::GFP showed a diffused pattern when the endogenous acdh-1 was not tagged. The number of the observed foci was consistent with the notion that all embryonic cells are diploid (Hedgecock and White, 1985) (Fig. 3, A and B). To confirm that the MCP::GFP-labeled foci represented nascent transcripts, we performed single-molecule fluorescence in situ hybridization (smFISH) experiments using either control or MS2-tageting probes. Neither probe produced spot-like signals when the endogenous acdh-1 gene was not tagged with the Dualtag (Fig. 3, C and D). For the acdh-1::mCherry^DualTag strain, the majority of the MCP::GFP-positive foci in the embryos were also labeled by the MS2-targeting smFISH probes, while the negative control probes did not label these foci (Fig. 3, E and F).

Previous studies have shown that intestinal cells undergo one round of endoreplication during each molting cycle, resulting in 2n ploidy at the L1, 4n at L2, 8n at L3, and 16n at L4 stages (Hedgecock and White, 1985). We observed active transcription of acdh-1 in the intestinal cells of C. elegans larvae. The number of MCP::sfGFP-positive foci largely correlated with the copy number of the endogenous acdh-1 genes, supporting that these GFP foci specifically represent nascent transcripts produced from the endogenous acdh-1 loci (Fig. 3, G and H).

To assess whether DualTag knock-in impacted the dynamic transcriptional changes of the above-mentioned metabolic genes, we compared pre-mRNA levels between non-tagged and DualTag-tagged strains using qRT-PCR. The results showed that endogenous adh-1, acdh-1, and hach-1 displayed similar increased or decreased levels in response to fasting, regardless of the presence of DualTag. However, pre-mRNA levels of endogenous icl-1 did not change obviously in both strains (Fig. 4, A and B). The discrepancy between the DualTag-based imaging (Fig. 2, C and D) and qRT-PCR analysis (Fig. 4, A and B) likely arises because DualTag-based imaging specifically reports the transcriptional changes in the epidermis, whereas qRT-PCR provides a whole-body analysis. In addition, smFISH assays using either negative control probes or acdh-1 intronic probes were conducted to detect the transcriptional changes of the endogenous acdh-1 with or without DualTag knock-in. Similar numbers of transcriptionally active cells, alleles, and transcriptional output per nucleus were observed in both strains (Fig. 4, C–E). Together, these findings support that the knock-in of DualTag into the endogenous genes does not significantly affect their transcription and that MONITTR can reliably report the transcriptional activity of endogenous genes.

Previous studies using microarray or RNA sequencing have identified icl-1 and adh-1 as fasting-induced genes, while acdh-1 and hach-1 as fasting-repressed genes (Goh et al., 2018; Harvald et al., 2017; Van Gilst et al., 2005). However, it remains largely unknown how these genes are regulated at the transcriptional and translational levels in live animals during fasting. To explore this, we utilized knock-in alleles with the DualTag and performed imaging on animals exposed to fasting (2, 4, 6, and 8 h) compared with non-fasting animals (0 h). As icl-1, adh-1, acdh-1, and hach-1 were expressed in the epidermis, and we focused on imaging the dynamic changes occurring in epidermal cells. We analyzed various transcriptional parameters by imaging nascent transcripts. We hypothesized that transcriptional alterations could be regulated through distinct mechanisms, which include the proportion of cells (or nuclei for syncytia) displaying active transcription, the number of alleles actively transcribing within each nucleus (indicated by the number of MCP::GFP spots per nucleus), and the transcriptional output of individual alleles (measured by the intensity of individual MCP::GFP spots) (Fig. S1 E). For icl-1, the proportion of epidermal nuclei with icl-1 nascent transcripts showed no significant increase after 2 or 4 h of fasting. However, a notable increase was observed after 6 and 8 h of fasting. The transcriptional output of a single allele showed no difference in animals fasted for 2, 4, 6, or 8 h. Within each nucleus, there were more alleles in the “On” state when animals were fasted for 6 and 8 h. At the translational level, there was no increase in ICL-1::mCherry protein expression after 2 h of fasting, but significant increases were observed for animals fasted for 4, 6, and 8 h (Fig. 5, A and B). Another fasting-induced gene, adh-1, exhibits low levels of both transcription and protein expression under normal conditions. Compared with icl-1, the upregulation of adh-1 transcription occurred at a slower pace, becoming evident after 8 h of fasting, while a significant increase in protein expression was observed only after 12 h of fasting. However, adh-1 displayed similar changes in regulatory modes as icl-1 during the process of transcriptional upregulation (Fig. 5, C and D).

We then investigated the dynamic changes of acdh-1, a gene that is repressed during fasting. Intriguingly, animals that fasted for 2 h showed a higher percentage of nuclei positive for transcription, increased transcriptional output of single alleles, and a greater number of transcriptionally active alleles. These findings suggest that at this time point, the transcription of acdh-1 is induced rather than being repressed. However, when animals were fasted for longer durations (for 4, 6, and 8 h), the portion of nuclei with at least one transcriptionally active allele gradually decreased, and fewer alleles were in the “On” state for transcription. Single allele yield was significantly increased in animals fasted for 2, 4, and 6 h, but not for those who fasted for 8 h. Interestingly, the expression level of the ACDH-1::MCHERRY protein exhibited a gradual and dramatic upregulation (Fig. 5, E and F). For hach-1, another fasting-repressed gene as previously reported, a significantly lower number of nuclei exhibited transcriptional activity in animals fasted for 2 h compared with well-fed animals, and almost none of them were active after fasting for 4 h or longer. Animals that fasted for 2 h also had fewer alleles in the “On” state. However, there was an increased yield of single alleles in animals fasted for 2 h compared with well-fed controls. In contrast to the downregulation at the transcriptional level, the level of HACH-1::MCHERRY protein was dramatically increased during fasting (Fig. 5, G and H). Regarding being fasting-induced or repressed genes, we observed corresponding increases or decreases in total RNA synthesis at the single-nucleus level during fasting, demonstrating the reliability of DualTag quantifications (Fig. S1, E–G). Importantly, we extended our investigation to include the fasting response in the intestinal tissue. Similar transcriptional and translational changes were observed for icl-1 and acdh-1 (Fig. S2, A–H). In conclusion, DualTag-based live-cell imaging demonstrates that changes in nascent RNA and protein levels do not exhibit a positive correlation for fasting-repressed genes.

The previous studies using microarray analysis or qRT-PCR reported that the transcriptional factor NHR-49 (C. elegans homolog of human HNF4a) is partially required for fasting-induced expression of icl-1 and adh-1 (Goh et al., 2018; Van Gilst et al., 2005). Another study using RNA sequencing revealed that loss of hlh-30, which encodes a transcriptional factor homologous to human TFEB, fully or partially abolishes fasting-induced expression of icl-1 and adh-1, respectively (Harvald et al., 2017). Moreover, daf-16 encodes a single FOXO transcription factor, acting as a central regulator to modulate multiple biological processes such as longevity and starvation resistance (Hibshman et al., 2017). Therefore, we conducted quantitative imaging to investigate whether and how these transcriptional factors (TFs) regulated the transcription of icl-1 and adh-1 in the epidermis during fasting response (Fig. 6, A and B).

Through the quantification of the percentage of nuclei with RNA spots, number of active alleles, or total nascent RNAs produced in each nucleus, our results unveiled that the fasting-induced upregulation of both icl-1 was partially perturbed in the absence of nhr-49 and nearly completely abolished in the absence of hlh-30 (Fig. 6, C and D). However, for adh-1, the absence of either nhr-49 or hlh-30 completely abrogated its transcriptional response to fasting (Fig. 6, F and G). Notably, the transcriptional output of each active allele remained largely unchanged during the fasting response, and this characteristic was not altered in the absence of nhr-49 and hlh-30. Moreover, our results suggest that daf-16 does not significantly regulate the fasting response of icl-1 and adh-1 genes in the epidermis (Fig. 6, E and H).

Next, we examined the regulatory role of these TFs in the fasting-mediated reduction in acdh-1 and hach-1 expression. Our results demonstrate that the fasting-induced transcriptional response of acdh-1 in epidermis was not influenced by hlh-30 and daf-16 (Fig. S3, A and B). However, we could not definitively determine the role of nhr-49 in regulating the fasting response of acdh-1 due to their close proximity to the same chromosome. We also found that the transcriptional response of hach-1 in the epidermis remained unaffected by the inactivation of all three TFs (Fig. S3, C–E). Collectively, our comprehensive imaging analysis conducted in the epidermis implies that NHR-49 and HLH-30 are key factors in modulating the transcriptional activation of icl-1 and adh-1 upon fasting. However, it is important to note that our analysis focused specifically on the transcriptional response to fasting in the epidermal tissue. The transcription factors governing transcriptional regulation in other tissues may differ.

Next, we sought to determine how the dynamic transcriptional changes were regulated by transcription factors at single-tissue, single-nucleus, and single-allele levels using time-lapse recording. Through real-time tracking of live animals (at 2-min intervals), where endogenous icl-1 loci were tagged with DualTag, we discovered that the proportion of nuclei displaying active icl-1 transcription in epidermis gradually increased in all animals during the initial 45 min of fasting and reached a plateau in the subsequent 45 min. Notably, this dynamic response was partially repressed by the loss of nhr-49 and completely blocked by the loss of hlh-30 (Fig. 7, A and B).

At the single-nucleus level, we quantified the total fluorescent intensity of all MCP::GFP dots within each nucleus, assessing the total nascent transcripts generated at a specific time point. Our analysis revealed a pronounced initial increase in nascent transcripts within the first 30 min, followed by a subsequent stabilization. Intriguingly, we also observed a simultaneous increase in the number of active icl-1 alleles within each nucleus. Notably, the loss of nhr-49 partially perturbed these dynamic changes, while loss of hlh-30 completely abolished these transcriptional responses (Fig. 7 C).

At the single-allele resolution, real-time recording unveiled that icl-1 displayed transcriptional bursting kinetics, characterized by alternating periods of transcriptional on and off states. The average duration of transcriptional bursts significantly increased during fasting. Moreover, our analysis indicated that nhr-49 and hlh-30 also played regulatory roles in modulating the bursting kinetics during fasting response. Specifically, burst durations were significantly decreased, and pause durations were increased in mutants lacking nhr-49 or hlh-30 (Fig. 7, D–G). Furthermore, we quantified the burst amplitude and found that these two TFs did not regulate the burst amplitude of the icl-1 gene (Fig. 7 H). Utilizing the same quantitative methods, we further examined how nhr-49 and hlh-30 regulate the transcriptional response of adh-1 and discovered similar regulatory features (Fig. S4, A–H). One notable difference is that the loss of hlh-30 caused an increase in the burst duration of adh-1 during the initial 60 min of image recording. However, between 60 and 120 min, the transcription was mostly shut down (Fig. S4, C and D). As a result, there were minimal changes in average burst duration and pause duration compared with the wild type overall.

Collectively, our findings demonstrate that NHR-49 and HLH-30 contribute to the upregulation of icl-1 and adh-1 expression during fasting through various mechanisms. These include increasing the number of nuclei and alleles involved in nascent transcript production, extending the duration of transcriptional bursts, and reducing pause durations. Remarkably, these changes are accompanied by a relatively stable burst amplitude (Fig. S5, A and B).

Next, we sought to determine whether MONITTR works to report the dynamic transcriptional changes induced by other types of stress, such as heat shock stress and ER stress. Upon heat shock, heat shock proteins (HSPs) are induced and serve as conserved molecular chaperones during evolution. They play crucial roles in the transport, folding, and assembly of unfolded or misfolded proteins (Burdon, 1988; Janowska et al., 2019; Joutsen and Sistonen, 2019; Tedesco et al., 2022). In C. elegans, HSPs are induced in the presence of thermal or ER stress (Baird et al., 2014; Kumsta and Hansen, 2017; Prahlad and Morimoto, 2009; Taylor et al., 2021). C. elegans is an established metazoan model for studying the effect of heat stress in vivo, which can be easily applied by exposing the animal to temperatures between 30°C and 37°C (Kumsta et al., 2017; Prahlad et al., 2008; Prahlad and Morimoto, 2009; Rodriguez et al., 2013; Xu et al., 2023). Specifically, hsp-16.41 has been identified as a heat-shock response gene essential for acquired heat stress tolerance (Kourtis et al., 2012).

We utilized mCherry^DualTag to tag the endogenous hsp-16.41 gene. After heat shock treatment, each cell displayed 0–2 MCP::GFP dots for various embryonic stages, which is consistent with the fact that embryonic cells are diploid (Fig. 8, A and B). Notably, nearly all the dots were also labeled by the MS2 targeting FISH probes, but not the negative control probes (Fig. 8, C and D). Additionally, we observed hsp-16.41 expression in the cell bodies of the diploid PVD neurons (Hedgecock and White, 1985), where most cells had two alleles actively transcribing (Fig. 8, E and F). hsp-16.41 was also robustly induced in the intestinal cells of C. elegans larvae, and the numbers of MCP::GFP dots were consistent with the copies of the endogenous hsp-16.41 gene (Fig. 8, G and H) (Hedgecock and White, 1985). These results further support that MONITTR is specific for monitoring nascent RNA transcription.

Promoter–GFP fusion served as a widely used method to report transcriptional dynamics. We first examined whether the response of transgenic strains carrying hsp-16.41prom::gfp responded to acute heat stress by monitoring GFP expression. When cultured at 20°C, the GFP signal was barely detectable. However, upon transferring the worms to 30°C, noticeable induction of GFP expression was observed after 2 h, with peak expression likely occurring between 2.5 and 5 h (Fig. 9, A and B). For the DualTag system, neither hsp-16.41 RNA nor HSP-16.41::mCherry protein expression was detected at 20°C. However, upon just 5 min of heat shock, DualTag imaging revealed that nearly 47.4% of epidermal nuclei exhibited active transcription of nascent RNA. This proportion was increased to 63.3% after 15 min and remained stable after 30 min. However, prolonged heat stress for 1 h resulted in the transcriptional shutdown of hsp-16.41, with similar outcomes observed for longer heat stress durations of 2–5 h (Fig. 9, C and D). During the initial 5–30 min of transcriptional response to heat stress, total RNA synthesis per nucleus increased due to an increase in the number of active hsp-16.41 alleles in each cell, while the transcriptional output of individual alleles remained relatively unaltered (Fig. 9, E–G). Notably, the expression of HSP-16.41::mCherry protein, as detected by DualTag, exhibited a significant lag behind transcription. Notable expression was observed only after 2 h of heat shock, with more pronounced expression occurring between 3 and 5 h (Fig. 9 H). These findings emphasize the precise monitoring of heat stress-induced transcriptional changes using DualTag while indicating a significant time lag in the detection of protein expression.

Next, we utilized the hsp-16.41::mCherry^DualTag strain to examine whether absolute temperature or temperature differences influence transcriptional changes following heat shock. Worms were initially cultured at 15°C or 20°C and then subjected to heat shock at 30°C, 32°C, or 34°C. We observed that all temperature differences rapidly induced hsp-16.41 transcription within 5 min, reached peak expression at ∼15 min, and maintained it for 30 min. At the milder heat shock temperature of 30°C, transcription gradually paused after 30 min. However, at 34°C, the signal of nascent RNA decreased to a certain level after 2 h of heat shock and plateaued thereafter, possibly due to temperature-induced disruptions in transcriptional shutdown regulation or RNA splicing (Fig. 9 I). Notably, irrespective of initial culturing temperature (15°C or 20°C), the optimal temperature range for inducing high HSP-16.41 expression in worms was between 30°C and 32°C, suggesting their sensitivity to absolute temperature (Fig. 9 J). HSF-1, a highly conserved transcription factor, induces the expression of HSPs, which provide cellular protection against diverse cytotoxic conditions (Anckar and Sistonen, 2007; Baird et al., 2014; Howard et al., 2016; Kumsta et al., 2017; Kumsta and Hansen, 2017; McMillan et al., 1998). Quantitative DualTag imaging confirmed the indispensable role of hsf-1 in mediating the transcriptional regulation of hsp-16.41 during heat shock (Fig. 9 J).

In response to ER stress, which results from the accumulation of unfolded or misfolded proteins, the unfolded protein response (UPR) triggers transcriptional and translational regulations to restore ER homeostasis (Hetz, 2012; Hetz et al., 2020; Oakes and Papa, 2015; Schröder and Kaufman, 2005; Yoshida, 2007). Therefore, we asked whether DualTag imaging could detect the transcriptional dynamics of ER stress response in time and space. The transcription of hsp-4, C. elegans ortholog of mammalian GRP78/BiP chaperone, is upregulated during ER stress induced by DTT or tunicamycin (Howard et al., 2016; Jo et al., 2009; Urano et al., 2002). We first examined a transgenic strain carrying hsp-4prom::gfp and found that GFP protein expression was only detectable after 2 h of tunicamycin treatment, with a further increase after 3 h (Fig. 10 A). Using DualTag to label the endogenous hsp-4, RNA imaging revealed that under control conditions, only 1.8–14.8% of epidermal nuclei exhibited active transcription of hsp-4. Following 30 min of tunicamycin treatment, around 39.2% of epidermal nuclei showed active transcription of hsp-4, reaching its peak after 1 h with up to 87.6% of nuclei actively transcribing hsp-4. However, the transcriptional levels remained relatively unchanged between 1 and 3 h of treatment (Fig. 10, B and C). During the time frame of 30 min to 3 h of tunicamycin treatment, there was an increase in the number of nuclei involved in transcription, the number of active alleles per nucleus, and the total RNA yield per nucleus. Notably, the transcriptional output per allele remained relatively constant throughout this period (Fig. 10 D). Overall, our findings highlight the sensitivity of DualTag in detecting and characterizing the dynamic regulatory features of transcriptional modulation under ER stress.

With the advancement of gene editing technologies, we can efficiently modify endogenous genes and incorporate various gene tags, particularly fluorescent protein tags (Chen et al., 2013; Dickinson et al., 2013; Schwartz et al., 2021; Schwartz and Jorgensen, 2016). In the MONITTR system, we have introduced a dual-functional tag named DualTag, which functions as a fluorescent protein with embedded MS2 stem loops within its artificial introns. By employing DualTag labeling of endogenous genes, we have monitored both pre-mRNA and protein levels of target genes in live animals. Unlike traditional approaches that typically insert MS2 in the 5′ or 3′UTRs, which often severely interfere with target gene expression, our strategy overcomes this limitation, as demonstrated by our side-by-side comparison. While we previously employed a similar approach in cultured human cells (Xu et al., 2020), this study, for the first time, introduces a novel gene tag specifically designed for simultaneous imaging of transcription and endogenous protein levels in multicellular animals. In the future, the versatility of DualTag can be further expanded by incorporating alternative RNA aptamers such as PP7, N22, and pepper into fluorescent proteins with distinct colors (Chen et al., 2019; Daigle and Ellenberg, 2007; Larson et al., 2011; Sun and Zou, 2022), thus addressing the demand for simultaneous monitoring of multiple genes.

MONITTR provides a significant advantage in specifically visualizing nascent RNA production. Our results clearly illustrate that MONITTR offers a more precise and real-time depiction of transcriptional dynamics than the widely used promoter-GFP fusion-based transcriptional reporters. For example, DualTag rapidly detected transcriptional activation at hsp-16.41 loci within 5 min of heat shock, whereas changes in protein level were only observed after 2 h. Similarly, during adh-1 fasting response, DualTag revealed transcriptional reprogramming at 8 h, while protein level increases were observed at 12 h. When monitoring fasting-repressed genes, the stability and lifespan of mature RNA and proteins introduce intrinsic delays, limiting their ability to capture real-time transcriptional alterations accurately. Therefore, the MONITTR system emerges as the ideal choice when investigating the dynamic changes of transcriptional repression. Moreover, DualTag facilitates transcriptional analysis at multiple levels, including individual animals, specific tissues, individual nuclei, and single gene loci, providing valuable insights into tissue-wide transcriptional regulation across organisms. While our study primarily focused on epidermal and intestinal tissues, future investigations can expand to diverse tissues by utilizing tissue-specific promoters to express MCP fused with a fluorescent protein.

Organisms face constant and diverse stresses throughout their lifespan, leading to the evolution of various mechanistic stress responses and quality control mechanisms. These include the cytosolic heat-shock response (HSR), endoplasmic reticulum unfolded protein response (UPR-ER), short-term food deprivation response (fasting), and the ubiquitin-proteasome system (Balch et al., 2008; Hetz, 2012; Higuchi-Sanabria et al., 2018; Morimoto, 1998, 2008; Vihervaara et al., 2018). These molecular pathways serve as both quality control mechanisms to maintain cellular homeostasis and as responses to mitigate the damaging effects of stresses such as heat exposure. Currently, there is limited understanding of the spatiotemporal regulation of stress-response genes in vivo under stress. To address this gap, we applied MONITTR to image and analyze transcriptional dynamics in C. elegans under three types of stress stimuli, including fasting, heat shock, and ER stress.

Our quantitative imaging illustrated distinct response kinetics of stress-responsive genes. Upon heat shock, hsp-16.41 showed the fastest transcription response, with changes detected within ∼5 min. While hsp-4 exhibited transcriptional alterations at around 30 min during ER stress, fasting response induced a slower transcriptional activation, with icl-1 showing changes at 6 h and adh-1 at 8 h. Intriguingly, during long-term live imaging of fasting response, where images were captured every 2 min, we noticed a significant upregulation of icl-1 transcription ∼30 min after the initiation of imaging, while adh-1 started exhibiting transcriptional upregulation at around 60 min. This observation suggests that additional factors such as laser exposure during high-frequency imaging might contribute to accelerated transcriptional remodeling in addition to fasting stress. In conclusion, our findings indicate that transcriptional remodeling in response to different stress stimuli occurs with varying response times, ranging from minutes to hours. These studies highlight the complexity of stress-induced transcriptional activation, suggesting the involvement of stress-specific mechanisms (de Nadal et al., 2011; Himanen and Sistonen, 2019; Vihervaara et al., 2018). Notably, distinct genes display unique transcriptional response kinetics, even when exposed to the same stressor. Furthermore, significant heterogeneity is observed among different animals in their response to a specific stress within the same gene. This observation supports the previous conclusion that stress-responsive genes favor expression plasticity, leading to a greater degree of variability. This increased variability is vital for better adaptation to adverse environmental changes (Charlebois, 2015; Charlebois et al., 2011; Gasch et al., 2017; Pascual-Ahuir et al., 2020). These findings emphasize the complex and dynamic nature of stress-induced transcriptional regulation.

Transcription is a complex process involving multiple steps, such as TF binding, coactivator recruitment, Preinitiation Complex formation, RNA Pol II recruitment, nucleosome eviction, and transcription termination (Brown et al., 2013; Jonkers and Lis, 2015; Kim and O’Shea, 2008; Malik and Roeder, 2023; Rodríguez-Molina et al., 2023; Soutourina, 2018). Each of these steps is highly stochastic in nature (Raj et al., 2006; Raj and van Oudenaarden, 2008; Raser and O’Shea, 2004). Different stressors may influence and regulate specific transcriptional steps through distinct signaling pathways, resulting in diverse regulatory patterns (Fujimoto et al., 2018; Lyu et al., 2018; Mueller et al., 2017; Niskanen et al., 2015; Vihervaara et al., 2017). Investigating stress-specific mechanisms presents an intriguing avenue for future research. MONITTR holds immense potential as a crucial approach for dissecting gene regulation at a highly detailed level, offering exquisite spatial and temporal sensitivity.

To access the regulatory networks involved in sensing and adapting to changes in nutrient availability, previous studies have employed quantitative RT-PCR or RNA-Seq to monitor gene expression changes in a population of C. elegans during fasting (Goh et al., 2018; Harvald et al., 2017; Van Gilst et al., 2005). However, our MONITTR system allows us to study how different tissues or cell types achieve transcriptional reprogramming during fasting response in their natural context. We observed significant fasting-induced expression of icl-1 and adh-1 in the epidermis, which was found to depend on both HLH-30 and NHR-49 through genetic analysis. Moreover, our real-time imaging approach provides single-nucleus, single-gene, and single-allele resolution, revealing that transcriptional upregulation primarily occurs by increasing the number of nuclei and alleles involved in transcription. Importantly, icl-1 and adh-1 transcription occur in discontinuous bursts during fasting response, which is conserved and offers a wide tunable regulatory range for individual genes (Chubb et al., 2006; Golding et al., 2005; Raj and van Oudenaarden, 2008).

Transcriptional bursting is widely recognized as a key factor contributing to the heterogeneity in transcriptional activity observed among cells (Chang et al., 2008; Falo-Sanjuan et al., 2019; Fritzsch et al., 2018; Wernet et al., 2006). Currently, burst frequency, burst duration, burst amplitude and pause duration appear to be the main parameters that can be modulated in different tissues and contexts (Lammers et al., 2020; Liang et al., 2022; Senecal et al., 2014; Xu et al., 2015, 2020). However, it remains to be determined whether transcription factors alter some of these parameters to confer appropriate transcriptional outputs in a live animal under fasting stress. By analyzing bursting kinetics, we found that burst and pause durations are regulated, while burst amplitude remains relatively stable during fasting. Disruption of crucial transcription factors alters these regulatory features, causing defects in transcriptional activation. Further research should combine single-molecular tracking (Morisaki et al., 2014; West, 2023) to monitor individual TF molecules along with DualTag readout of downstream transcriptional bursting, enabling analysis at high spatial and temporal resolution. This approach may provide direct evidence of how transcription factors perceive stress signals to initiate transcription remodeling (Wagh et al., 2023).

Unless otherwise specified, C elegans were cultured on nematode growth medium (NGM) plates seeded with OP50 E. coli at 20°C (Brenner, 1974), except for the worms in Fig. S4 which were fed with HT115 E. coli. N2 Bristol was used as the wild-type strain. hsf-1(sy441) (Hajdu-Cronin et al., 2004), hlh-30(tm1978) (Settembre et al., 2013), and daf-16(mu86) (Murphy et al., 2003) are loss-of-function alleles as reported previously. The strains used in this study are listed in Table S1.

The pSM delta vector (a derivative of pPD49.26) was used as the backbone for most plasmids constructed in this study. The sgRNAs were constructed from pU6 (GB)-sgRNA as backbone (Dickinson et al., 2013).

To generate donor plasmids for CRISPR/Cas9-mediated genome editing, about 600 bp homologous arms (5′ HA or 3′ HA) were amplified from a home-made C. elegans genomic DNA library. Recutting was prevented by introducing one or multiple silent mutation(s) in the repair template. The mCherry coding sequence with Cbr-unc-119 (+) sequence was amplified from pMLS291 (# 73724; Addgene Plasmid) (Schwartz and Jorgensen, 2016). The 24xMS2V5 sequence was amplified from Addgene Plasmid # 84561 (Wu et al., 2016). A seamless cloning kit (C113-02; Vazyme, 502 ClonExpress MultiS One Step Cloning Kit) was used for multifragment ligation.

Plasmids to express sgRNAs were generated by PCR-based Quick-Change cloning method. Briefly, a 5′ phosphorylated reverse primer, a forward prime with target sequence, and part of the sgRNA scaffold were used to amplify the entire sgRNA template plasmid. The template DNA concentration used in the PCR reaction was as low as 0.25 ng/μl. T4 DNA ligase was used to ligate PCR products. The ligation product was transformed into competent cells and correct plasmids were identified by Sanger sequencing. The plasmids used in this study are listed in Table S2. The primers used to generate sgRNA-expressing plasmids are listed in Table S3.

To generate icl-1(zac416[icl-1::mCherry]), Peft-3::cas9 (50 ng/μl, kindly provided by Dr. Suhong Xu), a donor plasmid (5′ homology arm+ mCherry+ 3′ homology arm) (50 ng/μl), two plasmids to express the single-guide RNAs (25 ng/μl each), and negative selection markers Podr-1::rfp (30 ng/μl), Pmyo-2::mCherry (#19327; 2 ng/μl; Addgene), and Pmyo-3::mCherry (#19328; 3 ng/μl; Addgene) were mixed and injected into unc-119(ed4). Notably, in the donor plasmid Cbr-unc-119(+) was inserted in the third intron of mCherry (Schwartz and Jorgensen, 2016), and several silent mutations were included to avoid recutting. About 8 days later, the progenies that moved normally and showed no negative selection marker expression were selected for PCR-based genotyping and Sanger sequencing. adh-1(zac534[adh-1::mCherry]), acdh-1(zac354[acdh-1::mCherry]), hach-1(zac537[hach-1::mCherry]), acdh-2(zac453[acdh-2::mCherry]), and semo-1(zac429[semo-1::mCherry]) were generated using a similar unc-119 selection-based knockin protocol as described above.

To generate icl-1(zac356[icl-1::mCherry(24xMS2 in the second intron), or mCherry^DualTag hereafter]), the donor plasmid containing 5′ homology arm, mCherry, and 3′ homology arm was constructed using a seamless cloning protocol. Noted that 24 repeated MS2V5 sequences were inserted in the second intron of mCherry and Cbr-unc-119(+) was inserted in the third intron. Several silent mutations were included to avoid recutting. The donor plasmid (50 ng/μl), Peft-3::cas9 (50 ng/μl), sgRNAs (20 ng/μl each), and negative selection markers Podr-1::rfp (30 ng/μl), Pmyo-2::mCherry (2 ng/μl), and Pmyo-3::mCherry (3 ng/μl) were mixed and injected into unc-119(ed4). A similar unc-119 selection-based knockin protocol as described above was used to generate adh-1(zac510[adh-1:: mCherry^DualTag]), acdh-1(zac344[acdh-1:: mCherry^DualTag]), hach-1(zac511[hach-1:: mCherry^DualTag]), acdh-2(zac355[acdh-2:: mCherry^DualTag]), semo-1(zac381[semo-1:: mCherry^DualTag]), hsp-4(zac460[mCherry^DualTag::hsp-4), and hsp-16.41(zac462[hsp-16.41:: mCherry^DualTag]).

To generated icl-1(zac315[icl-1::mCherry::24xMS2::3′UTR)]), the mCherry fragment and 24xMS2 sequence were inserted between the 5′ homology arm and the 3′ homology arm in the donor. Noted that the 24xMS2V5 sequence was inserted upstream of the 3′ UTR of icl-1 while Cbr-unc-119(+) was inserted in the third intron of mCherry. Several silent mutations were included to avoid recutting. Other steps were similar as described above. adh-1(zac535[adh-1::mCherry::24xMS2::3′UTR)]), acdh-1(zac430[acdh-1::mCherry (24xMS2::3′UTR)]), hach-1(zac539[hach-1::mCherry::24xMS2::3′UTR)]), acdh-2(zac313[acdh-2::mCherry::24xMS2::3′UTR)]), and semo-1(zac435[semo-1::mCherry::24xMS2::3′ UTR)]) were generated via a similar protocol.

To generate lmn-1(zac434[Plmn-1::MCP::sfGFP::SL2::mCherry::lmn-1]), lmn-1(zac455[Plmn-1::NLS::MCP::sfGFP::SL2::mCherry::lmn-1]), and lmn-1(zac472[Plmn-1::MCP::sfGFP::SL2::BFP::lmn-1]), MCP::sfGFP with or without the NLS sequence was ligated with SL2::mCherry or SL2::BFP coding sequence and inserted into the N-terminus of the endogenous lmn-1 locus using CRISPR/Cas9-mediated homologous recombination. Notably, Cbr-unc-119(+) was inserted in the third intron of the mCherry coding sequence and several silent mutations were included to avoid recutting. For lmn-1(zac434[Plmn-1::MCP::sfGFP::SL2::mCherry::lmn-1]) and lmn-1(zac455[Plmn-1::NLS::MCP::sfGFP::SL2::mCherry::lmn-1]), a donor plasmid (50 ng/μl), Peft-3::cas9 (50 ng/μl), sgRNAs (20 ng/μl each) and negative selection markers Podr-1::rfp (30 ng/μl), Pmyo-2::mCherry (2 ng/μl), Pmyo-3::mCherry (3 ng/μl) were mixed and injected into unc-119 (ed4). To generate lmn-1(zac472[Plmn-1::MCP::sfGFP::SL2::BFP::lmn-1]), a donor plasmid (50 ng/μl), Peft-3::cas9::NLS::pU6::dpy-10 sgRNA(50 ng/μl, kindly provided by Dr. Suhong Xu) and sgRNAs (20 ng/μl each) were mixed and injected into N2. F1 animals with dumpy or roller phenotype were selected as candidates and the F2 animals were subjected to PCR-based genotyping.

nhr-49(zac598) was generated by a modified dpy-10-based co-conversion protocol (Arribere et al., 2014). In brief, Peft-3::cas9::NLS::pU6::dpy-10 sgRNA (50 ng/μl) was mixed with sgRNA #1(target sequence: 5′-AGT​TTA​ATT​GTT​TCA​GTC-3′) and sgRNA #2 (target sequence: 5′-TTC​TGC​TCA​CTG​TTC​AAA​A-3′) (20 ng/μl each) and injected into lmn-1(zac472). Dumpy or roller F1 worms were selected for PCR-based genotyping and Sanger sequencing. F2 animals with homozygous nhr-49 deletion alleles were used for function analysis. A 2,243 bp genomic DNA region flanked by 5′-CTG​GTT​TTT​GAA​AGT​TAT​GA-3′ and 5′-CAG​TGA​GCA​GAA​TAA​TCA​TA-3' was deleted in nhr-49(zac598).

zacEx2292(Phsp-16.41::gfp) and zacEx2573(ser2prom3::myr-bfp) were generated using standard gonad transformation by injection. Podr-1::rfp (50 ng/μl) and Pmyo-2::mCherry (2 ng/μl) were used as co-injection markers.

Worms were cultured in 6-cm plates, washed three times with M9 buffer, and collected into a 2 ml centrifuge tube, and then 200 μl of TRizol was added (15596018; Ambion). For fasting response and ER stress response groups, worms were synchronized at the L4 stage. For heat shock response groups, 1-day adult worms were collected. The above tubes were repeatedly frozen and thawed twice in liquid nitrogen and room temperature and then stored in a −80°C refrigerator until the next step. A cell/tissue total RNA extraction kit (19221ES50; YEASEN) was used to extract RNA. RNA was reverse transcribed with a reverse transcription kit (HiScript III first Strand cDNA Synthesis Kit; Vazyme). Real-time qPCR was then performed on a CFX-96 thermal cycler (Bio-Rad) using 2xUniversal SYBR Green Fast qPCR Mix (ABclonal) master mix. The qPCR program was set as follows: 95°C for 10 min and 40 cycles at 95°C for 15 s/60°C for 30 s. ama-1 was used as the reference gene. At least three biological and technical replicates were performed and shown. The primers for total mRNA were designed in the exon region, and the primers for pre-mRNA were mainly in the intron region. qRT-PCR primers were listed in Table S3.

The α-amanitin (23109-05-9; MCE) stock solution was dissolved in M9 to a working concentration of 20 µg/ml. For icl-1 and adh-1, worms at the L4 stage were soaked in 10 μl M9 buffer or α-amanitin in 200 μl PCR tubes for 3 h at 20°C and then transferred to fresh NGM plates seeded with or without OP50 E. coli for drying. For fasting 0-h groups, images were taken immediately after the worms were transferred to NGM plates with OP50 E. coli. For fasting 8 h groups, the animals were imaged after cultured on NGM plates without OP50 E. coli for 8 h. For acdh-1, worms at the L4 stage were imaged after being soaked in 10 μl M9 buffer or α-amanitin in 200 μl PCR tubes for 4 h at 20°C. For hach-1, L4-stage worms were imaged after being treated with M9 or α-amanitin for 2 h.

Tunicamycin (11089-65-9; Aladdin) was dissolved in DMSO to a stock concentration of 5 mg/ml. The final concentration of tunicamycin was 50 µg/ml in NGM plates. Synchronized L4 stage worms were transferred to chemical-containing plates seeded with OP50 E. coli. Worms were cultured at 20°C for different durations as shown in Fig. 10.

For results shown in Fig. 4, A and B; and Fig. 9, A–H, animals were initially grown at 20°C until 1-day-old adult. Images were taken immediately after the worms were incubated in a 30°C water bath at different times. For results shown in Fig. 8, A–D, embryos were imaged or collected after being incubated at 33°C for 30 min. For the results shown in Fig. 8, E and F, 1-day-old adult worms were imaged after being incubated at 33°C for 30 min. For results shown in Fig. 8, G and H, worms at different larval stages were imaged after being incubated at 33°C for 30 min. For results shown in Fig. 9 I, worms were initially cultured to 1-day adult stage at 15°C or 20°C and then imaged after incubated at a 30°C, 32°C, or 34°C water bath for different durations. For results shown in Fig. 9 J, worms grown to 1-day adult at 15°C or 20°C were imaged after being incubated at different temperatures for 30 min.

Worms were fed to L4 stage and then washed twice with M9, then transferred to NGM plates without E. coli and cultured at 20°C for 0–12 h before subjecting to confocal imaging or being collected for qRT-PCR.

A series of short probes were designed and synthesized by GD Pinpoease Biotech Co., Ltd. smFISH probes complementary to MS2V5 sequence were designed to cover the region of MS2V5^24x (Cat # 2903461-D1). Probes (Cat # 1723581-D1, Cat # 1723582-D1) are complementary to the intron sequence of acdh-1, covering the entire intronic region of acdh-1. All probe sets used in this study were tagged with PIN670 (Cat # PIF0017; GD Pinpoease Biotech Co. Ltd.). smFISH was performed using PinpoRNATM RNA in situ hybridization kit (Cat # PIT1000; GD Pinpoease Biotech Co. Ltd.). In detail, C. elegans embryos or larvae were harvested into an Eppendorf tube and fixed in 1 ml of 4% paraformaldehyde for 24 h at room temperature. Then the samples were frozen in liquid nitrogen for 1 min. After the sample was thawed in water at room temperature, the embryos or larvae were placed on ice for 20 min. Then the sample was washed twice with 1 ml of 1× PBS. To permeabilize, the samples were resuspended in 1 ml 70% EtOH and stored overnight at 4°C. Ethanol was removed and embryos or larvae were washed twice by 1 ml of 1× PBS. Target RNA molecules were exposed to protease treatment and hybridized with probes for at least 5 h at 40°C. Then the signal was amplified sequentially by three reactions according to the manufacturer’s instructions. The target RNAs were then fluorescently labeled by Tyramide Signal Amplification (TSA) assay. To stain the nuclei, the samples were incubated in 5 μg/ml DAPI for 20 min at room temperature. Excessive DAPI was removed before imaging.

Worms were anesthetized using 10 mM levamisole in M9 buffer and placed on 4% agarose pads for static imaging. All confocal images were acquired on an Olympus IX83 fluorescence microscope equipped with a spinning-disk confocal scanner (CSU-W1; Yokogawa), an sCMOS camera (Prime 95B), 10×/20×/40× magnification, a 60× NA 1.49 oil Apochromat objective, 488/561/405 nm lasers (OBIS), and a PIEZO stage (ASI). For results shown in Fig. 2, C, E, G, and I; Fig. 3, A, C, E, and G; Fig. 4 D; Fig. 5, A, C, E, and G; Fig. 8, A, C, E, and G; Fig. 9 C; Fig. 10 B; Fig. S1 C; and Fig. S2, A and E, z-stacks with 0.8 µm step size were captured and proper focal-plane images were shown. For results shown in Fig. 1 B, Fig. 9 A, and Fig. 10 A, z-stacks with 0.8, 1, or 2 µm step size were obtained, and the maximum intensity of projections was shown.

For time-lapse imaging, worms at L4 stage were anesthetized with 10 mM levamisole in M9 buffer and subjected to confocal imaging using a protocol developed by Chai et al. (2012). Z-stacks with 0.5 µm step size were taken every 2 min (exposure time: 100 ms) for 1.5–2 h. The percentage of nuclei with RNA spots, number of nuclei, RNA spot intensity, burst/pause time, and burst/pause duration were analyzed manually using Image J. At least seven animals were quantified for each strain. For results shown in Fig. 7 A and Fig. S4 A, the maximum intensity of projections was shown.

For results shown in Fig. 2, D, F, H, and J; Fig. 5, B, D, F, and H; Fig. 6, C–H; Fig. 10, C and D; Fig. S1 D; Fig. S2, B, C, F, and G; and Fig. S3, A–E, worms were synchronized to L4 stage. For results shown in Fig. 8 F; and Fig. 9, D–G, I, and J, worms were synchronized to 1-day-old adult stage. For results shown in Fig. 3, B, D, and F; and Fig. 8, B and D, the samples were at embryonic stages. For results shown in Fig. 3 H and Fig. 8 H, worms were synchronized to different larval stages. To quantify % of the transcriptionally active nucleus, the percentage of nuclei with nascent RNA spots among all nuclei in the defined area was calculated. To quantify the transcriptional output of single alleles, the mean intensity and area of each RNA signal were manually measured and multiplied. The average intensity of the background was also quantified. The subtraction of the fluorescence intensity between the signal point and the background was defined as the intensity of a single allele. To quantify the number of transcriptional active alleles, the number of alleles being transcribed in each nucleus was counted. For the transcriptional output of each nucleus, the RNA fluorescence intensity of all transcribed alleles in each nucleus was summed, which reflected the overall transcription yield of each nucleus.

The statistical methods of transcribing the nucleus, number of active alleles, and transcription of the nucleus in dynamic imaging were similar to that in static imaging. Nuclei with no more than two alleles transcribing at the same time were selected when quantifying burst and pause-related results. To quantify burst (pause) time, all burst (pause) time of a single allele within 1.5–2 h of dynamic imaging was added together. To quantify burst (pause) duration, the total duration of each burst (pause) time of a single allele during the whole imaging process was measured. To quantify the amplitude of transcriptional bursts, the maximum value of the fluorescence intensity of RNA signals during a complete burst of a single allele throughout the imaging process was measured.

For results shown in Fig. 1 C, Fig. 9 B, and Fig. 10 C, the protein fluorescence intensity was measured by the average fluorescence intensity of the whole animal through Image J, and then the average intensity of the background outside of the animals was subtracted. For results shown in Fig. 5, B, D, F, and H; Fig. 9 H; and Fig. S2, D and H, the average intensity was measured in the area of interest, and the average background fluorescence intensity was similarly subtracted.

GraphPad Prism was used to analyze data. For most bar chart results, values are shown as mean ± SEM. For violin plot results, quartiles and medians were shown. Data distribution was assumed to be normal but this was not formally tested. For all data comparing multiple groups, One-way analysis of variance (ANOVA) was performed with the Tukey’s post-test. Two-tailed unpaired Student’s t test was used when comparing two sets of data. Statistical analysis was reflected by the P value. (*P < 0.05, **P < 0.01, ***P < 0.001, ****P < 0.0001, and ns = not significant if P > 0.05).

Fig. S1 shows the mRNA levels of icl-1, adh-1, acdh-1, and hach-1 are dramatically adjusted upon fasting. Fig. S2 shows that DualTag-based imaging allows monitoring the dynamic transcriptional changes of icl-1 and acdh-1 in the intestinal cells. Fig. S3 shows NHR-49, HLH-30, and DAF-16 are not required for the transcriptional down-regulation of acdh-1 and hach-1 in the epidermis. Fig. S4 shows MONITTR allows real-time tracking of the dynamic changes of adh-1 transcription in wild-type, nhr-49, and hlh-30 mutant animals. Fig. S5 shows a schematic diagram of the transcriptional patterns of fasting-induced genes in response to fasting. Table S1 shows the C. elegans strains used in this study. Table S2 shows the plasmids used in this study. Table S3 shows the primers used in this study.

Data are available in the article itself and its supplementary materials.

We thank Drs. Suhong Xu (Zhejiang University), Lijun Kang (Zhejiang University), Daniel J Dickinson (UT Austin), Christopher M. Hammell (Cold Spring Harbor Laboratory), and Erik M. Jorgensen (University of Utah) for sharing reagents.

This work was supported by the National Natural Science Foundation of China grants 32370822 and 31970919 (to W. Zou), 32371507 and 32171444 (to B. Chen), and National Key R&D Program of China 2021YFC2700904 (to B. Chen). Some strains were provided by the CGC, which is funded by the NIH Office of Research Infrastructure Programs (P40 OD010440), and the MITANI Lab through the National Bio-Resource Project of the Ministry of Education, Culture, Sports, Science and Technology, Japan.

Author contributions: X. Liu: Data curation, Formal analysis, Investigation, Methodology, Project administration, Resources, Supervision, Validation, Visualization, Writing - review & editing, Z. Chang: Data curation, Investigation, P. Sun: Investigation, Methodology, Project administration, B. Cao: Investigation, Methodology, Y. Wang: Validation, J. Fang: Investigation, Resources, Validation, Visualization, Y. Pei: Conceptualization, Data curation, Formal analysis, Investigation, Methodology, Project administration, Resources, Supervision, Validation, Writing - original draft, Writing - review & editing, B. Chen: Conceptualization, Funding acquisition, Investigation, Methodology, Project administration, Supervision, Writing - original draft, Writing - review & editing, W. Zou: Conceptualization, Data curation, Formal analysis, Funding acquisition, Project administration, Resources, Supervision, Visualization, Writing - original draft, Writing - review & editing.