The DEAD-box helicase Hlc regulates basal transcription and chromatin opening of stress-responsive genes.

Stress-responsive genes are lowly transcribed under normal conditions and robustly induced in response to stress. The significant difference between basal and induced transcription indicates that the general transcriptional machinery requires a mechanism to distinguish each transcription state. However, what factors specifically function in basal transcription remains poorly understood. Using a classic model stress-responsive gene (Drosophila MtnA), we found that knockdown of the DEAD-box helicase Hlc resulted in a significant transcription attenuation of MtnA under normal, but not stressed, conditions. Mechanistically, Hlc directly binds to the MtnA locus to maintain the accessibility of chromatin near the transcriptional start site, which allows the recruitment of RNA polymerase II and subsequent MtnA transcription. Using RNA-seq, we then identified plenty of additional stress-responsive genes whose basal transcription was reduced upon knockdown of Hlc. Taken together, these data suggest that Hlc-mediated basal transcription regulation is an essential and widespread mechanism for precise control of stress-responsive genes.

INTRODUCTION

External and internal stimuli provoke a significant transcriptional reprogramming, during which RNA polymerase II (Pol II) rapidly accumulates at the promoter and coding region of stress-responsive genes to enhance their expression (1,2). The adaptive transcription enables cells to respond within minutes upon exposure to various stresses, thereby improving cell survival and adaptation (1,2). For example, copper stress can quickly activate the transcription activity of Drosophila Metallothionein A (MtnA), whose encoded protein is able to sequester overloaded copper ions inside cells (3,4).

Note that a great many stress-responsive genes are not completely silent under normal conditions. Instead, they are continuously transcribed at the basal state, despite being at relatively low levels (1,2). Classic cases in point are MtnA and multiple heat shock protein genes (Hsps), whose basal expression can be readily detectable by different techniques (5–8). In particular, the basal activities of Hsps were found to have fundamental roles in normal cell growth and differentiation (9,10). These studies support that the basal transcription program of stress-responsive genes is also an essential aspect of biological processes in living cells.

To date, most transcription factors identified function in both basal and stress-induced transcription. For example, the metal-responsive transcription factor-1 (MTF-1) is indispensable for the transcription activity of MtnA under both normal and heavy metal stressed conditions (11–14). Given the significant difference between basal and stress-induced transcription, we speculate that cells may require sophisticated mechanisms to assist the general transcriptional machinery and/or transcription factors to distinguish between normal and stressed conditions. In support, our previous study has demonstrated that knockdown of Drosophila gawky significantly reduced the level of metal-induced transcription but had no effect on basal transcription (15). Mechanistically, gawky translocates into the nucleus and directs the metal-activated version of MTF-1 to the promoter regions of metal-responsive genes to promote their transcription in response to heavy metal stress (15). This suggests that gawky is specific for dictating metal-induced transcription. However, what factors exclusively regulate basal transcription and the detailed mechanism remain an outstanding question.

Drosophila Hlc and its mammalian homolog DDX56 belong to DEAD-box helicases, a highly-conserved RNA-binding protein family whose members are involved throughout cellular RNA metabolism, such as splicing, nuclear export, translation and degradation (16–18). DDX56 was first identified as a ribosome biogenesis factor modulating the assembly process of the late-stage 60S ribosomal subunit in human and yeast cell lines (19,20). Consistently, loss of planarian DDX56 induces ribosome biogenesis disruption and leads to dysregulation of ribosomal proteins, which ultimately impairs cell proliferation and triggers caspase-dependent apoptosis (21). DDX56 was also found to exert an inhibitory or promoting effect on infectivity of various viruses (22–25). In the case of the West Nile virus (WNV), DDX56 directly binds to the capsid protein to form a stable complex which facilitates packaging of the infectious WNV into particles (22). By contrast, DDX56 attenuates infection of the chikungunya virus (CHIKV) by destabilizing incoming CHIKV genomic RNAs as well as reducing CHIKV protein production on early steps of viral replication (23). Moreover, recent studies have revealed the oncogenic roles of DDX56 in different cancer types (26–28). For example, DDX56 can promote tumor growth by affecting the alternative splicing of a subset of cell cycle‐related genes in colorectal cancer (26). Although emerging roles of DDX56 have been revealed, studies regarding the molecular functions of this helicase are still very limited.

Using a classic model stress-responsive gene (MtnA), we here identified Drosophila Hlc as an essential regulator specific for dictating basal transcription. Mechanistically, Hlc directly interacts with the genomic locus of MtnA to maintain its open chromatin status. Knockdown of Hlc resulted in a condensed chromatin architecture near the transcriptional start site (TSS), thereby disassociating Pol II from the genomic MtnA. In addition, we found a negative cooperativity between gawky and Hlc at the genomic locus of MtnA during the transition from the basal to copper-induced transcription, confirming the specificity of Hlc-meditated regulation. Using RNA-seq, we found plenty of additional stress-responsive genes whose basal transcription was significantly down-regulated upon knockdown of Hlc. Focused studies on heat stress genes further verified the general role of Hlc in basal transcription. In summary, we proposed that Hlc-mediated regulation acts as an essential and widespread mechanism to coordinate the basal transcription of stress-responsive genes.

MATERIALS AND METHODS

Drosophila cell culture and RNAi

Drosophila Schneider 2 (S2) cells were maintained in Schneider's Drosophila Medium (Sigma, S9895) containing 10% fetal bovine serum (v/v; Excell, FSP500) and 1% penicillin-streptomycin (v/v; Thermo Fisher Scientific, 15140122) at 25°C. DNA templates for double-strand RNA (dsRNA) preparation were obtained by PCR reactions with primer pairs containing the T7 promoter sequence (Supplementary Table S1). DsRNAs were generated by in vitro transcription using ScriptMAX® Thermo T7 Transcription Kit (TOYOBO, TSK-101). For Hlc RNAi in S2 cells, 1.5 × 106 cells were incubated with 8 μg of the indicated Hlc dsRNA in 600 μl of serum-free medium for 30 min at room temperature. After addition of 400 μl of medium containing 20% fetal bovine serum (v/v), dsRNA-treated cells were cultured at 25°C for 3 days. β-gal dsRNA served as a negative RNAi control in this study. To activate metal-induced transcription, where indicated, a final concentration of 500 μM copper sulfate (CuSO4, Macklin, C805782) was added into the medium for the final 12 hr before collection. To activate heat shock-induced transcription, where indicated, cells were cultured at 33 or 37°C for the final 1 hr before collection.

Vector transfection, rescue assay and stable cell line construction

To generate Hlc expression vectors for transfection into Drosophila S2 cells, the indicated sequences were inserted into our previously described Hy_pAct5C FLAG MCS vector (15,29,30), a reporter which can express a FLAG-tagged protein under control of the Act5C promoter. These Hlc expression vectors were used for stable cell line construction and rescue assays. The detailed information of all vectors used in this study is provided in Supplementary Vector Information.

For transient transfection, 1μg of the indicated vector was introduced into 1.0 × 106 S2 cells using Lipo6000™ Transfection Reagent (Beyotime, C0526), and transfected cells were cultured at 25°C for 2 days before collection. For Hlc rescue assay, 24 hr after treatment with the Hlc UTR dsRNA, cells were transfected with the indicated Hlc vector and cultured at 25°C for another 4.5 days in the presence of 150 μg/ml hygromycin B (Biofroxx, 1366ML010). Note that these Hlc expression vectors only harbor the coding region and are insensitive to UTR dsRNA-mediated Hlc knockdown. For stable cell line construction, transfected cells were selected in the medium containing 150 μg/ml hygromycin B for 4 weeks.

RT-qPCR

RNA was isolated using RNAiso Plus (Takara, 9109), and complementary DNA (cDNA) was produced through reverse transcription (RT) using PrimeScript RT Master Mix Kit (Takara, RR036A) according to the manufacturer's instructions. Quantitative PCR (qPCR) was carried out with the indicated primer pairs and Hieff® qPCR SYBR Green Master Mix (YEASEN, 11201ES03) using the CFX connect real-time PCR system (Bio-Rad). Ct values were normalized to rp49 mRNA expression. The sequences of all qPCR primers are provided in Supplementary Table S2.

Fluorescence in situ hybridization (FISH)

Fluorescent RNA probes were generated by in vitro transcription (ScriptMAX® Thermo T7 Transcription Kit, TOYOBO, TSK-101) and labeled with the Alexa Fluor 488 dye (ULYSIS Nucleic Acid Labeling Kit, Thermo Fisher Scientific, U21650). The sequence of the MtnA fluorescent probe is provided in Supplementary Table S3. FISH was conducted as previously described with minor modifications (15). In brief, cells were fixed by fixative solution (75% methanol (v/v) and 25% glacial acetic acid (v/v)) at room temperature for 10 min, denatured at 80°C for 10 min, incubated with 100 ng of denatured probes in the presence of 20 ng/μl yeast RNA (Beyotime, R0038) at 42°C overnight and stained with 1 μg/ml DAPI (Beyotime, C1002) for 5 min before confocal microscopy analyses. Fluorescence signals were detected using the Leica TCS SP8 system and quantified using ImageJ.

Nuclear run-on (NRO)

NRO assay was used to measure the production of the nascent MtnA transcript in Drosophila nuclei. NRO was performed as previously described (15,30). In brief, a total of 1.5 × 108 S2 cells were washed twice with 1 ml of ice-cold 1× PBS buffer (phosphate buffer saline pH 7.4: 137 mM NaCl, 2.7 mM KCl, 10 mM Na2HPO4 and 1.8 mM KH2PO4) and resuspended into 5 mL of ice-cold Hypotonic Buffer (150 mM KCl, 4 mM MgOAc and 10 mM Tris–HCl pH 7.4) containing 80 units/ml RNase inhibitor (Beyotime, R0102). After 10 min incubation on ice, cells were collected by centrifugation at 1000g and resuspended into 1 ml of ice-cold Lysis Buffer (Hypotonic Buffer supplemented with 0.5% NP40 (v/v) and 80 units/ml RNase inhibitor) followed by ∼20 rounds of slow pipetting. Nuclei were then isolated by density gradient centrifugation using 4.5 ml of 0.6 M sucrose, resuspended into 100 μl of NRO Buffer (50 mM Tris–HCl pH 7.5, 5 mM MgCl2, 150 mM KCl, 0.1% srkosyl (w/v), 10 mM dithiothreitol and 80 units/ml RNase Inhibitor) containing each of 1 mM NTPs (ATP, GTP, CTP and BrUTP (Sigma, B7166)) and incubated at 28°C for 5 min to allow incorporation of BrUTP into nascent transcripts. The NRO reaction was shut off by addition of 500 μl of RNAiso Plus (Takara, 9109). After RNA isolation, nuclear RNA was resuspended into 500 μl of RNA Immunoprecipitation Buffer (20 mM Tris–HCl, pH 7.5, 200 mM NaCl, 2.5 mM MgCl2, 0.5% NP40 (v/v) and 10% glycerol (v/v)). BrUTP-labeled transcripts were then purified using Dynabeads® Protein G (Thermo Fisher Scientific, 10003D) and anti-BrUTP (Abcam, ab1893). Efficient nascent RNA purification was verified by measuring the level of the MtnA precursor mRNA by RT-qPCR analysis.

Chromatin immunoprecipitation (ChIP)

ChIP was used to evaluate Pol II, Hlc and gawky density at the genomic locus of MtnA. A total of 7.5 × 107 S2 cells were washed twice with 1× PBS (phosphate buffer saline pH 7.4: 137 mM NaCl, 2.7 mM KCl, 10 mM Na2HPO4 and 1.8 mM KH2PO4) containing 1 mM phenylmethanesulfonyl fluoride (PMSF) and cross-linked with 1% formaldehyde for 10 min at 37°C. The cross-linking reaction was stopped by addition of a final concentration of 125 mM glycine. Fixed cells were collected by centrifugation at 300g for 5 min, resuspended into ChIP Lysis Buffer (1% sodium dodecyl sulfate (SDS; w/v), 10 mM EDTA, 50 mM Tris–HCl pH 8.1 and 1 mM PMSF) and sonicated using a Scientz-IID Sonicator to yield 200–500 bp DNA fragments. The mixture was then centrifuged at 12 000g for 10 min at 4°C to remove cell debris, and the supernatant was saved as chromatin solution. 5% of the chromatin solution was used as the input sample, and the remaining was immunoprecipitated with the indicated antibody using ChIP Assay Kit (Beyotime, P2078) according to the manufacturer's instructions. These antibodies were used in ChIP: anti-FLAG (Beyotime, AF519), anti-Hlc (purified from rabbit antisera, raised against amino acids 545–560 of the endogenous Drosophila Hlc), anti-gawky (previously described in (15)) and anti-RNA polymerase II (Abcam, ab5095; this antibody recognizes Pol II phosphorylated at the serine 2 residue of the C-terminal domain repeat YSPTSPS). Note that a negative IgG (Beyotime, A7028) was also used in each ChIP experiment to exclude the potential of non-specific binding. DNA from the input and ChIP samples was purified using Cycle Pure Kit (OMEGA, D6492) and subjected to qPCR analysis. For Hlc and gawky ChIP, the DNA binding sites of Hlc and gawky (as well as the negative IgG) were defines as genomic regions enriched over the input DNA. For Pol II ChIP, signals (relative to input) were normalized to signals obtained from the negative IgG, and the normalized ChIP signal in the RNAi control sample was set to ‘1’.

Micrococcal nuclease (MNase) protection assay

MNase protection assay was used to examine the role of Hlc in chromatin accessibility. After RNAi and/or rescue experiments, a total of 1.5 × 108 S2 cells were washed twice with 1× PBS buffer (phosphate buffer saline pH 7.4: 137 mM NaCl, 2.7 mM KCl, 10 mM Na2HPO4 and 1.8 mM KH2PO4), cross-linked with 1% formaldehyde in 1× PBS buffer for 10 min and quenched with 125 mM glycine for 10 min. Fixed cells were then resuspended into 2 ml of ice-cold solution I (10 mM HEPES pH 7.6, 10 mM KCl, 1.5 mM MgCl2, 0.34 M sucrose, 10% glycerol (v/v), 1 mM dithiothreitol, 0.1% Triton X-100 (v/v) and 1× protease inhibitor cocktail (Beyotime, P1045)) followed by ∼50 rounds of slow pipetting. Nuclei were collected by centrifugation at 1000g for 5 min and resuspended into 1 ml of Solution II (3 mM EDTA, 0.2 mM EGTA, 1 mM dithiothreitol and 1× protease inhibitor cocktail). After 30 min incubation on ice, released chromatin was washed twice with 1mL of Solution II, resuspended into 1 mL of MNase Buffer (50 mM Tris–HCl pH 7.4, 25 mM KCl, 12.5% glycerol (v/v), 10 mM CaCl2, 4 mM MgCl2 and 1 mM PMSF) and then divided into two equal aliquots (0.5 ml). One was incubated with 5 units MNase (Aladdin, 9013-53-0) for 5 min at 37°C, and the other, as the negative control sample (undigested), was lightly sonicated at 27% amplitude for 6 rounds of 24 s using a Scientz-IID Sonicator. Both aliquots were incubated at 65°C for 6 h and treated with 0.5 mg/ml proteinase K (Beyotime, ST535) for 30 min at 55°C to release the cross-linked DNA. After DNA purification, qPCR experiments were performed to quantify the extent of MNase digestion with 7 overlapping amplicons tiling through the genomic region of MtnA (–250 to +257 bp relative to the TSS). The relative protection level at each examined site was defined as the ratio of MNase-digested to undigested genomic DNA.

Immunofluorescence staining (IF)

IF was used to visualize the nuclear-cytoplasmic distribution of the wild-type and mutant Hlc. IF was performed as previously described (15,30,31). In brief, after treatment with fixative solution (75% methanol (v/v) and 25% glacial acetic acid (v/v)), fixed cells were permeabilized with 0.1% Triton X-100 (v/v) in TBST buffer (Tris-Buffered Saline Tween-20 pH 7.6: 0.242% Tris–HCl (w/v), 0.8% NaCl (w/v) and 0.05% Tween-20 (v/v)) at room temperature for 10 min and incubated with the indicated primary antibody in the presence of 5% bovine serum albumin (w/v; Beyotime, ST025) at 4°C overnight. After incubation with the fluorophore-conjugated secondary antibody, 1 μg/ml DAPI (Beyotime, C1002) was used to stain nuclei for 5 min before confocal microscopy analyses. Fluorescence signals were detected using the Leica TCS SP8 system and quantified using Image J. Line profiles were analyzed using Image-Pro Plus 6.0. These antibodies were used in IF: anti-Hlc (raised against amino acids 545–560 of the endogenous Drosophila Hlc), anti-FLAG (Beyotime, AF519), Alexa Fluor 555-labeled anti-mouse IgG (Beyotime, A0460) and Alexa Fluor 555-labeled anti-rabbit IgG (Beyotime, A0453).

Western blotting

Protein extracts from β-gal (the RNAi control) and Hlc RNAi cells were prepared using RIPA buffer (50 mM/l Tris–HCl pH 7.4, 150 mM/l NaCl, 0.1% SDS (w/v), 1% sodium deoxycholate (w/v) and 1% Triton X-100 (v/v)). Western blotting experiments were performed according to the standard ECL protocol (Thermo Fisher Scientific, EI9051). Antibodies used were anti-Hlc (raised against amino acids 545–560 of the endogenous Drosophila Hlc), anti-α-Tubulin (Beyotime, AT819), HRP-labeled goat anti-rabbit IgG (H + L) (Beyotime, A0208) and HRP-labeled goat anti-mouse IgG (H + L) (Beyotime, A0216).

RNA sequencing (RNA-seq) and bioinformatic analysis

Whole-cell RNA was isolated from two biological replicates using RNAiso Plus (Takara, 9109), and RNA quality was verified using the Agilent Bioanalyzer 2100 system. RNA-seq libraries were generated with 1 μg of RNA using the Illumina TruSeq™ RNA Sample Preparation kit and were subject to 150 bp paired-end sequencing using the Illumina NovaSeq 6000 system. Adapter sequences and low-quality reads were removed with Trimmomatic (version 0.36) with parameters LEADING:3 TRAILING:3 SLIDINGWINDOW:4:15 MINLEN:36 (32). The remaining high-quality reads were then aligned to the reference genome (the Drosophila melanogaster genome (BDGP6)) from Ensembl (https://www.ensembl.org) using TopHat (version 2.1.0) (33,34). StringTie (version 2.1.7) was used to estimate global gene expression of each sample based on the Ensembl gene annotation (BDGP6.32, version 104) (33,35). Differential gene expression (DGE) analysis was performed using the DEseq2 R package (version 1.32) (36), and gene ontology (GO) analysis was performed using DAVID bioinformatics resources (2021 update) (37).

For ATAC-seq data analysis, the raw data, generated from Drosophila S2 cells, were downloaded from the Gene Expression Omnibus (GEO) repository (GSE197224, SRX10915169) (38), quality-trimmed using Trimmomatic (version 0.36) (32) and mapped to the Drosophila melanogaster genome (BDGP6) using TopHat (version 2.1.0) (33,34). Sambamba (version 0.8.1) was used to remove duplicate reads from the BAM file (39). The bigWig file with RPKM normalization was generated from the trimmed BAM file using deepTools (version 3.3.0) with the option ‘computeMatrix’ (40). The density plot reflecting ATAC-seq signals was generated using deepTools (version 3.3.0) with the option ‘computeMatrix reference-point’ (40).

Statistical analysis

Unless otherwise indicated, statistical significances were evaluated by Student's t-test. Statistical details are provided in each figure legend.

RESULTS

Identification of the role of Hlc in the mRNA expression of MtnA

To identify factors that specifically regulate the basal transcription of stress-responsive genes, we utilized the previously published screening dataset (15). Among all 127 factors screened, the DEAD-box helicase Hlc was the only candidate that modulated the expression level of MtnA (a well-characterized metal-responsive gene that belongs to the metallothionein family (41)) under normal, but not copper-stressed, conditions in Drosophila S2 cells (15). To ascertain the phenotype from the screening and exclude the possibility of off-target effects of RNAi-mediated transcript knockdown, we took advantage of 3 independent dsRNAs to individually reduce Hlc expression in unstressed and copper-treated S2 cells (Figure 1A, B; Supplementary Figure S1A). Consistently, the level of MtnA mRNA was significantly down-regulated under normal conditions, but had almost no change in response to copper treatment (Figure 1B). RNA-FISH of MtnA mRNA further verified the function of Hlc in MtnA expression at the single-cell level (Figure 1C, D).

Figure 1.

Hlc-mediated regulation is required for basal transcription but is bypassed for stress-induced transcription. (A) A schematic representation of the endogenous Drosophila Hlc mRNA and the approximate targeted positions of three independent Hlc dsRNAs. (B) RT-qPCR quantification of MtnA and Hlc expression using whole-cell RNA extracts purified from the control (β-gal) or Hlc RNAi cells (n = 3 biological replicates; each replicate present as a black dot). ∗∗P< 0.01; ∗P < 0.05, calculated by Student's t-test. (C) Confocal microscopy analysis of the MtnA FISH signals after Hlc RNAi. Representative images from 4 biological replicates are shown. Scale bar = 5 μm. (D) Violin plots comparing the MtnA FISH signals from samples described in Figure 1C (n = 100 cells for fluorescence signal quantification per sample). Lines inside each violin plot represent the first quartile, median and third quartile of the data. ∗∗P< 0.01; ∗P < 0.05, calculated by Student's t-test. (E) A schematic overview of the experimental setup (NRO, Pol II ChIP and rescue assay) for the data shown in (G)–(I). (F) The approximate locations of primer pairs detecting the 5′ end, middle or 3′ end of the MtnA gene body. (G) NRO analysis of the nascent MtnA transcript abundance (i.e. the transcription activity of MtnA) after Hlc RNAi (n = 5 biological replicates; each replicate present as a black dot). The reference gene U6 served as a negative control. ∗∗P< 0.01; ∗P < 0.05, calculated by Student's t-test. (H) Pol II ChIP assay measuring Pol II density at the locus of MtnA after Hlc RNAi (n = 3 biological replicates; each replicate present as a black dot). The reference gene Act42A served as a negative control. ∗∗P< 0.01; ∗P < 0.05, calculated by Student's t-test. (I) RT-qPCR quantification of MtnA expression using whole-cell RNA extracts purified from the control RNAi, Hlc RNAi, empty vector control rescued or Hlc vector rescued cells (n = 3 biological replicates; each replicate present as a black dot). The reference gene U6 served as a negative control. ∗∗P< 0.01; ∗P < 0.05, calculated by Student's t-test. Data in (B), (G), (H) and (I) are shown as means ± SEM. For samples under copper stress, 500 μM copper sulfate was added into the medium for the final 12 h before collection. NT, normal condition; Cu, copper stress.

Hlc regulates basal but not stress-induced transcription

Based on the above observations, we speculated that Hlc may somehow regulate MtnA at the transcriptional level. To test the hypothesis, we monitored the transcription activity of MtnA by nuclear-run on (NRO) assays (Figure 1E, left panel), a classic method for examining the nascent transcript abundance of a gene of interest (42,43). The newly-generated MtnA transcript was purified with anti-BrUTP and analyzed by RT-qPCR with primer pairs detecting the 5′ end, middle or 3′ end of the precursor (Figure 1F). Compared to the treatment with the negative control (β-gal dsRNA), Hlc RNAi led to a ∼80% reduction in the abundance of the nascent MtnA transcript under normal conditions regardless of which primer pair was used (Figure 1G, higher panel), but had no effect under copper stress (Figure 1G, lower panel). The reference gene U6 was also measured to minimize technological artifacts of each biological replicate under both conditions (Figure 1G). The MtnA precursor mRNA was highly enriched in NRO samples, excluding steady-state RNA contamination (Supplementary Figure S1B). Moreover, transcription inhibition assays, in which a 12 h transcription pulse of MtnA was triggered by addition of copper sulfate and turned off by Actinomycin D, demonstrated that neither knockdown nor overexpression of Hlc affected the half-life of MtnA mRNA (Supplementary Figure S2). This excludes the possibility that Hlc modulates the level of MtnA mRNA by enhancing its stability.

Given that MtnA is transcribed by RNA Pol II (8), we next investigated whether Hlc affects the density of Pol II at the locus of MtnA by ChIP experiments (Figure 1E, middle panel). Like the NRO assay, same primer pairs were used to measure the 5′ end, middle and 3′ end of the MtnA gene (Figure 1F). Following transfection with dsRNAs that target Hlc, the overall binding of Pol II to MtnA was significantly decreased under normal conditions (Figure 1H, higher panel), but had almost no change upon copper treatment (Figure 1H, lower panel). As a control, Pol II association of the house-keeping gene Act42A was not affected by Hlc knockdown under both conditions (Figure 1H).

In addition, we developed a rescue vector that only contains the coding region of Hlc and is insensitive to UTR dsRNA-mediated Hlc knockdown. As expected, rescue assays (Figure 1E, right panel) demonstrated that reexpression of Hlc significantly restored the level of MtnA mRNA under normal conditions (Figure 1I).

It is now known that circular RNAs are common outputs of many protein-coding genes via backsplicing (44–46). Therefore, we also investigated whether different splicing patterns influence Hlc-mediated transcription regulation using our previously described Hy_pMtnA dati Exons 1–3 vector (15,47), which is under control of the MtnA promoter and can produce a linear dati RNA as well as a circular version (Supplementary Figure S3A). Compared to the β-gal dsRNA control, both linear and circular dati RNA were significantly down-regulated upon Hlc knockdown under normal, but not stressed, conditions (Supplementary Figure S3B, C). These data suggest that (i) Hlc regulates basal transcription in a canonical splicing or backsplicing independent manner, and that (ii) Hlc-mediated transcription regulation seems to rely on the context of the promoter region.

Collectively, these results strongly demonstrate that Hlc is specifically required for the basal transcription of the metal-responsive gene MtnA and indicate the essential role of Hlc in discrimination of basal and stress-induced transcription.

Hlc directly targets the promoter and gene body of MtnA

Given that transcription is a tightly-controlled process that mainly occurs in the nucleus (48), we thus investigated whether the endogenous Drosophila Hlc is a nuclear protein by immunofluorescence staining assays and confocal microscopy analyses. As observed, Hlc was detected in both the nucleus and cytoplasm (Figure 2A, B). Relatively, ∼75% was present in the nuclear fraction and ∼25% was present in the cytoplasmic fraction (Figure 2C). Same experiments examining the subcellular localization of the FLAG-tagged Hlc confirmed the above phenotype (Supplementary Figure S4). Therefore, we concluded that Hlc translocates between the nucleus and cytoplasm and is primarily enriched in the nucleus.

Figure 2.

Hlc is a nuclear-localized protein and interacts with the genomic MtnA. (A) Confocal microscopy analysis of the endogenous Hlc immunofluorescence signals in S2 cells. Representative images from three biological replicates are shown. Scale bar = 5 μm. (B) Line profiles of both fluorescence intensities (Hlc and DAPI) along the dashed line, present in (A), confirming the nuclear localization of Hlc. (C) Violin plots reflecting the relative proportion of nuclear and cytoplasmic Hlc from the sample described in (A) (n = 50 cells for fluorescence signal quantification). Lines inside each violin plot represent the first quartile, median and third quartile of the data. (D) Schematic representations of the genomic MtnA and Act5C with the approximate locations of ChIP amplicons. (E) Hlc ChIP assay measuring the binding of Hlc to the genomic MtnA in cells treated with or without copper sulfate (n = 3 biological replicates). The reference gene Act5C served as a negative control. The negative IgG was used in each ChIP experiment to exclude artifacts of non-specific binding. The DNA binding sites of Hlc were defines as genomic regions enriched over the input DNA. Data are shown as means ± SEM. (F) Confocal microscopy analysis of the endogenous Hlc immunofluorescence signals in cells treated with or without copper sulfate. Representative images from three biological replicates are shown. Scale bar = 5 μm. (G) Violin plots reflecting the relative proportion of nuclear and cytoplasmic Hlc from samples described in (F) (n = 50 cells for fluorescence signal quantification per sample). Lines inside each violin plot represent the first quartile, median and third quartile of the data. (H) Western blots of the endogenous Hlc with protein extracts from cells treated with or without copper sulfate. α-Tubulin served as a loading control. (I) Bar chart showing the quantified protein level of Hlc from three independent blots (each replicate present as a black dot). Data were normalized to the unstressed sample and are shown as means ± SEM. For samples under copper stress, 500 μM copper sulfate was added into the medium for the final 12 h before collection. NT, normal condition; Cu, copper stress.

To ask whether Hlc directly regulates the transcription of MtnA, we examined the density of Hlc at the locus of MtnA by ChIP experiments. Primer pairs spanning the promoter region and gene body of MtnA were used (Figure 2D). Compared to the negative IgG control, Hlc was found to interact with both the promoter region (0–400 bp upstream of the TSS) and gene body of MtnA under normal conditions (Figure 2E, left panel). Intriguingly, Hlc only exhibited a limited binding capacity towards MtnA in response to copper treatment (Figure 2E, left panel). No obvious interaction between Hlc and the house-keeping gene Act5C was observed, which further confirmed the specificity of Hlc-mediated regulation (Figure 2E, right panel). Moreover, the subcellular localization (Figure 2F, G) and protein level (Figure 2H, I) of Hlc were unaffected in copper-stressed cells. These results rule out the possibility that copper ions induce the disassociation of Hlc from MtnA by triggering Hlc nuclear export or affecting Hlc protein level.

The binding of Hlc to MtnA is required for the basal transcription of MtnA

To search the potential DNA binding domain of Hlc, we took advantage of DP-Bind (a web server for sequence-based prediction of DNA-binding sites) and found that amino acids 517–557 of Hlc had a potential DNA-binding capacity (49). In support, ChIP-qPCR experiments demonstrated that the interaction between Hlc and MtnA was drastically reduced upon deletion of the predicted DNA-binding domain (Supplementary Figure S5). This mutant was referred to ΔDBD herein.

Since thousands of studies have proved that protein–DNA interactions play a significant role in the transcription regulation of many eukaryotic genes (48,50,51), we next asked whether the direct interaction between Hlc and MtnA is essential to the accurate function of Hlc in the process of basal transcription. To this end, the vector expressing the ΔDBD mutant (Figure 3A) was introduced into the Hlc UTR dsRNA treated cells. The expression level of MtnA and the occupancy of Pol II were measured with primer pairs detecting the 5′ end, middle or 3′ end of the MtnA gene body (Figure 3B–D). Different from the wild-type Hlc, the ΔDBD mutant failed to restore MtnA expression (Figure 3C) or Pol II density at the MtnA locus (Figure 3D), confirming the contribution of the DNA binding domain of Hlc in promoting MtnA transcription under normal conditions. Recall that Hlc was released from the MtnA locus in response to copper treatment (Figure 2E, left panel), which can explain the specificity of Hlc-mediated regulation in basal transcription.

Figure 3.

Hlc functions as a DNA binding protein. (A) Schematic representations of the wild-type Hlc (WT) and its ΔDBD mutant. The position of the DNA binding domain is marked in pink. (B) The approximate locations of primer pairs detecting the 5′ end, middle or 3′ end of the MtnA gene body. (C) RT-qPCR quantification of MtnA expression using whole-cell RNA extracts purified from the control RNAi, Hlc RNAi, empty vector control rescued, WT Hlc vector rescued or ΔDBD Hlc vector rescued cells (n = 3 biological replicates; each replicate present as a black dot). The reference gene U6 served as a negative control. ∗∗P< 0.01; ∗P < 0.05, calculated by Student's t-test. (D) Pol II ChIP assay measuring Pol II density at the locus of MtnA after the same knockdown-rescue experiment as described in Figure 3C (n = 3 biological replicates; each replicate present as a black dot). The reference gene Act42A served as a negative control. ∗∗P< 0.01; ∗P < 0.05, calculated by Student's t-test. Data in (C) and (D) are shown as means ± SEM.

Hlc functions in a gawky-independent manner

In contrast to Hlc, it was recently demonstrated that Drosophila gawky is only required for metal stress-induced transcription activation but is bypassed from basal transcription (15). In addition, the binding capacity of gawky towards the promoter of MtnA was robustly elevated under metal stress (15). Therefore, we asked whether gawky affects the interaction between Hlc and MtnA during the transition from the basal to copper-induced transcription. To this end, we performed Hlc ChIP experiments using gawky RNAi cells treated with or without copper. As observed, the binding of Hlc to MtnA was not affected by gawky knockdown under normal conditions (Figure 4A), supporting that gawky is not involved in Hlc-mediated basal transcription regulation. Surprisingly, gawky knockdown resulted in an elevated interaction between Hlc and MtnA under copper stress (Figure 4B). This can be explained by the assumption that Hlc rebound to MtnA to sustain its basal transcription after the metal-induced transcription had been impaired.

Figure 4.

Hlc functions in a gawky-independent manner. (A, B) Hlc ChIP assay measuring the binding of Hlc to the genomic MtnA after cells had been individually treated with two independent gawky dsRNAs (n = 3 biological replicates). The reference gene Act5C served as a negative control. The negative IgG was used to exclude artifacts of non-specific binding. The DNA binding sites of Hlc were defines as genomic regions enriched over the input DNA. Schematic representations of the genomic MtnA and Act5C with the approximate locations of ChIP amplicons are shown on top. Data are shown as means ± SEM. For samples under copper stress, 500 μM copper sulfate was added into the medium for the final 12 h before collection. NT, normal condition; Cu, copper stress.

Next, we performed gawky ChIP experiments and found that knockdown of Hlc had almost no effect on the binding of gawky to MtnA, indicating that Hlc is dispensable for gawky-mediated transcription activation (Supplementary Figure S6). As a control, the house-keeping gene Act5C exhibited no changes in Hlc or gawky ChIP signals in above experiments (Figure 4; Supplementary Figure S6). Taken together, these results support that Hlc specifically controls basal transcription while gawky specifically regulates metal-induced transcription.

Hlc enhances the genomic accessibility around the TSS of MtnA

The compaction of genomic DNA into chromatin has a fundamental role in the transcription regulation of many genes (52–54). Emerging studies have demonstrated that a group of DEAD-box helicases (e.g. DDX5 and DDX17) are able to modulate nucleosome compaction, chromatin organization and the accessibility of DNA into chromatin in an active or repressive way (55–59). Informed by these findings as well as our data present in Figure 1, we reasoned that Hlc may maintain the transcription activity of MtnA through regulating chromatin organization. To test this hypothesis, we measured the accessibility of the genomic region surrounding the MtnA TSS by MNase protection assays (Figure 5A), a classic method used to profile the active chromatin region of a gene of interest (60,61). Limited digestion of chromatin with the endo- and exo-nuclease MNase cleaves the exposed DNA between adjacent nucleosomes, but significantly enriches the genomic DNA fragments that are wrapped around histone octamers (60,61). As expected, agarose gel electrophoresis showed that mononucleosome-protected DNA fragments (∼150 bp) were released by MNase cleavage from chromatin isolated from β-gal (the RNAi control) and Hlc RNAi cells (Figure 5B). Subsequent qPCR analysis was performed with seven overlapping amplicons to tile through the genomic region surrounding the MtnA TSS (–250 to +257 bp) and quantify the digestion extent of MNase treatment (Figure 5B). The relative protection level was defined as the ratio of MNase-digested to undigested genomic DNA. Compared to the β-gal dsRNA control (set to ‘1’), knockdown of Hlc resulted in a significant increase in the MNase protection level at the genomic region spanning from –84 to +257 bp relative to the TSS (Figure 5C), suggesting the formation of a closed/inactive chromatin architecture near the TSS. Taken together, the result supports a model in which Hlc maintains an open chromatin architecture near the TSS of MtnA to allow loading of the general transcriptional machinery (e.g. Pol II) for the basal transcription of MtnA.

Figure 5.

Hlc is required for chromatin opening near the TSS of MtnA. (A) A schematic overview of the experimental setup for the MNase protection assay. (B) Left, the approximate locations of seven overlapping qPCR amplicons spanning the MtnA TSS. These amplicons were used to measure the MNase protection level at each position. Right, agarose gel electrophoresis verifying that mononucleosome-protected DNA fragments were released by MNase digestion. (C) MNase protection analysis in the control (β-gal) or Hlc RNAi cells (n = 3 biological replicates). The relative protection level was defined as the ratio of MNase-digested to undigested genomic DNA. To better reflect the effect of Hlc knockdown on chromatin opening, the β-gal dsRNA control was set to ‘1’ at each examined site. Data are shown as means ± SEM. Statistically significant differences between the control and Hlc RNAi samples (P< 0.05, calculated by Student's t-test), if any, are marked with the grey box.

The helicase ATP binding domain of Hlc is required for open chromatin maintenance

Considering that helicases are a class of ubiquitous and highly-conserved regulators that function in the structure modulation of double-stranded nucleic acids in an energy-dependent manner (16–18), we thus asked whether the helicase ATP binding domain of Hlc contributes to open chromatin maintenance. To fill the gap, we constructed a vector which can express an RNAi-resistant Hlc mutant deleted of the helicase ATP binding domain (referred to ΔHD herein) (Figure 6A). The control, wild-type and ΔHD vector were individually introduced into the Hlc UTR dsRNA treated cells, and MNase protection experiments were performed to measure the chromatin architecture at the MtnA TSS. Distinct from what was observed with the wild-type Hlc, reexpression of the ΔHD mutant failed to eliminate the closed chromatin architecture, but instead slightly increased the MNase protection level at the MtnA TSS relative to the Hlc knockdown sample (Figure 6B). This phenotype suggests that the ΔHD mutant can function like a dominant-negative protein. Indeed, loss of the helicase ATP binding domain did not affect the interaction between Hlc and MtnA (Supplementary Figure S7A) or the nuclear-cytoplasmic distribution of Hlc (Supplementary Figure S7B, C).

Figure 6.

The helicase ATP binding domain of Hlc is required for chromatin opening. (A) Schematic representations of the wild-type Hlc (WT) and its ΔHD mutant. The position of the helicase ATP binding domain is marked in orange. (B) MNase protection analysis in the control RNAi, Hlc RNAi, empty vector control rescued, WT Hlc vector rescued or ΔHD Hlc vector rescued cells (n = 3 biological replicates). The relative protection level was defined as the ratio of MNase-digested to undigested genomic DNA. The β-gal dsRNA control was set to ‘1’ at each examined site. Statistically significant differences between the WT and ΔHD Hlc vector rescued samples (P< 0.05, calculated by Student's t-test), if any, are marked with the grey box. The approximate locations of seven overlapping qPCR amplicons spanning the MtnA TSS are shown on top. (C) RT-qPCR quantification of MtnA expression after the same knockdown-rescue experiment as described in (B) (n = 3 biological replicates; each replicate present as a black dot). The approximate locations of primer pairs detecting the 5′ end, middle or 3′ end of the MtnA gene body are shown on top. The reference gene U6 served as a negative control. ∗∗P< 0.01; ∗P < 0.05, calculated by Student's t-test. (D) Pol II ChIP assay measuring Pol II density at the locus of MtnA with primer pairs shown in (C) after the same knockdown-rescue experiment as described in (B) and (C) (n = 3 biological replicates; each replicate present as a black dot). The reference gene Act42A served as a negative control. ∗∗P< 0.01; ∗P < 0.05, calculated by Student's t-test. Data in Figure 6B–D are shown as means ± SEM.

To further examine whether the helicase ATP binding domain of Hlc regulates the basal transcription of MtnA, we quantified the level of MtnA mRNA and the density of Pol II after rescue experiments same to what was described in Figure 6B. Consistently, the ΔHD mutant failed to restore MtnA expression (Figure 6C) or Pol II occupancy at the MtnA locus (Figure 6D). Taken together, we concluded that the proper activity of Hlc in open chromatin maintenance and MtnA transcription depends on its helicase ATP binding domain.

Hlc plays a widespread and potent role in the basal transcription of a subset of Drosophila stress-responsive genes

In order to substantiate the importance of Hlc-mediated transcription regulation, we thus asked whether additional stress-responsive genes are similarly regulated in Drosophila. To this end, two independent dsRNAs against Hlc (#2 and 3) were used to knockdown Hlc expression in unstressed S2 cells and RNA-seq was subsequently performed to measure global gene expression. As observed, out of 1983 differentially expressed protein-coding genes, 1101 were significantly down-regulated upon treatment with Hlc dsRNA#2 (fold change (log2) >1 and P < 0.001) (Figure 7A). Gene ontology analysis indicated that the set of down-regulated genes was enriched for the category related to ‘response to stress’ (Figure 7B). Notably, stress-responsive genes were more sensitive to Hlc knockdown among these down-regulated genes, since loss of Hlc reduced the expression level of stress-responsive genes to a greater extent than that of non-stress genes (Figure 7C). In particular, a number of Hsp mRNAs were identified among the most down-regulated genes (Figure 7D). For example, the basal transcription of Hsp23 and Hsp26 was almost completely silent upon Hlc knockdown (Figure 7E, F). Moreover, we utilized the previously published ATAC-seq (a high-throughput method for mapping global chromatin accessibility) dataset (38) and found that down-regulated stress-responsive genes had a more accessible chromatin structure around their TSSs relative to unchanged stress-responsive genes (Figure 7G). Consistently, similar phenotypes were observed with the Hlc dsRNA#3 sample (Supplementary Figure S8). These results support the function of Hlc in the basal transcription and open chromatin maintenance of a subset of stress-responsive genes.

Figure 7.

Hlc plays a general role in the basal transcription of Drosophila stress-responsive genes. (A) Volcano plot showing differentially expressed protein-coding genes in Drosophila S2 cells treated with Hlc dsRNAs (#2). MtnA is present as a purple dot and Hlc is present as a light blue dot. Threshold used to define Hlc-regulated genes is fold change (log2) > 1 and P < 0.001. P value is calculated by Wald chi-squared test. For each gene selected in following analyses (B, C and G), combined reads from the control and Hlc RNAi (#2) samples are ≥1000. (B) Gene ontology (GO) analysis identifying categories of genes that are enriched among down-regulated genes. Note that subcategories with red arrows belong to the main category ‘response to stress’ (GO:0006950). Subcategories are ranked by their statistical significance, and P value is calculated by Fisher's exact test. (C) Rank-ordered plot reflecting the fold change of down-regulated genes. Genes are ranked by their fold change, and stress-responsive genes are highlighted as red dots. A boxplot comparing the reduction extent between stress-responsive and non-stress genes is also shown in the middle. P value is calculated by Student's t-test. (D) Heat map highlighting the fold change of Hsp genes upon Hlc knockdown. MtnA served as a positive control and Gapdh1 served as a negative control. (E, F) Integrative Genomics Viewer (IGV) snapshot of Hsp23 and Hsp26 expression in the control and Hlc RNAi cells. (G) Density plot reflecting ATAC-seq normalized signals around TSSs of down-regulated (blue line) or unchanged (green line) stress-responsive genes in the RNA-seq dataset. (H) RT-qPCR quantification of Hsp expression using whole-cell RNA extracts purified from the control and Hlc RNAi cells. To activate heat shock-induced transcription, cells were cultured at 33 or 37°C for the final 1 h before collection. ∗∗P< 0.01; ∗P < 0.05, calculated by Student's t-test. Data are shown as means ± SEM. NT, normal condition; 33 and 37°C, heat stress.

Based on the above observation (Figure 7D–F), we next focused on Hsps to validate our RNA-seq result. As expected, RT-qPCR analysis demonstrated that knockdown of Hlc led to a significant and robust reduction in the basal, but not heat shock-induced, transcription of all Hsp mRNAs examined (Figure 7H), analogous to our prior results with MtnA. Taken together, we concluded that Hlc-mediated transcription regulation represents a general mechanism for cells to specifically regulate the basal transcription of a subset of stress-responsive genes in Drosophila.

DISCUSSION

Significant transcription reprogramming occurs to help cells establish multiple adaptive responses in the presence of stressed conditions (1,2). Typically, the transcription activity returns to its basal level after the stress has been removed (1,2). Despite employing the same transcription machinery (e.g. Pol II), cells may use different pathways to dictate basal and stress-induced transcription states. Based on the hypothesis, we aimed to search regulators that function exclusively in each transcription state. As expected, our previous study has identified gawky as a metal stress ‘checkpoint’, which specifically activates metal-induced transcription by promoting MTF-1 nuclear import and recruiting MTF-1 to the promoter of metal-responsive genes (15). Here, this study proposes that the DEAD-box helicase Hlc specifically regulates the basal transcription of a subgroup of stress-responsive genes but is dispensable for their induced transcription. The focused investigation regarding MtnA highlights the molecular mechanism of Hlc-mediated transcription regulation: Hlc binds to the genomic locus of MtnA to increase its DNA accessibility by maintaining an open chromatin architecture, which ultimately allows assembly of the general transcriptional machinery (Figure 8A). Taken together, these studies propose that the changes in mechanisms of gene expression and transcription regulation are accompanied by changes in the chromatin environment which is modified by the binding of different trans-activators. At least in the case of metal stress, Hlc and gawky serve as essential factors to dictate basal and stress-induced transcription, respectively (Figure 8B).

Figure 8.

A working model for Hlc-mediated transcription regulation. (A) Hlc ensures the basal transcription of stress-responsive genes by maintaining an open and accessible chromatin structure. (B) In the case of transition between normal and metal-stressed conditions, Hlc specifically regulates basal transcription while gawky specifically regulates induced transcription.

Although it was long thought that Hlc functions as an RNA binding protein (19,20,23), our study suggests that Hlc also possesses a DNA binding capacity (Figure 2E). In addition, the binding of Hlc to DNA loci is indispensable for its proper function in basal transcription, as supported by the phenotype that (i) Hlc was almost completely disassociated from the MtnA locus in the presence of overloaded copper (Figure 2E) and that (ii) the ΔDBD mutant did not restore MtnA expression or Pol II recruitment in the Hlc UTR dsRNA treated cells (Figure 3C, D). We also noticed that Hlc rebound to the MtnA locus when copper-induced transcription was impaired by gawky knockdown (Figure 4B), recall that the binding of gawky to the MtnA locus is in a metal-dependent manner (15). These results further confirm that Hlc and gawky function anticooperatively on chromatin (Figure 8B).

The DEAD-box family modulates gene expression by controlling various aspects of RNA metabolism, such as splicing, nuclear export and degradation (16–18). DDX5 is required for appropriate splicing of genes necessary for spermatogenesis (62). UAP56/DDX39B, as a component of the exon-junction complex, binds to nascent transcripts during splicing and recruits the export factor ALY to facilitate the nuclear export of bulk mRNAs (63–65). DDX6 destabilizes transcripts of differentiation-relevant genes through association with mRNA degradation factors (66). However, we found that knockdown of Hlc did not affect the splicing pattern (Supplementary Figure S3), subcellular distribution (Figure 1C) or stability (Supplementary Figure S2) of examined mRNAs. These results somehow exclude the possibility that Hlc may also function post-transcriptionally, at least for the regulation of MtnA expression. In addition, it is worth noting that nuclear quality control (i.e. RNA nuclear surveillance, a process to degrade incorrectly processed nascent transcripts in the nucleus (67–69)) can be bypassed for Hsp mRNAs only in response to heat hock (70), reminiscent of the phenotype of Hlc knockdown under stressed conditions. In this regard, we noticed that Hlc was not involved in premature transcription termination (Figure 1G, H). Therefore, it is not likely that Hlc co-transcriptionally affects steps of mRNA nuclear quality control. However, further investigation about this hypothesis is required.

Chromatin opening is a prerequisite for transcription. In this study, we found that Hlc is indispensable for open chromatin maintenance in the basal transcription state (Figure 5) and its helicase ATP binding domain contributes to this process (Figure 6). These data raise a hypothesis that Hlc acts as a direct functional enzyme to remodel chromatin and increase the accessibility of DNA. However, we cannot rule out the possibility that Hlc indirectly affects 3D chromatin organization through regulating the functional activity of chromatin remodeling factors and/or affecting genome stability. In fact, DDX5 and DDX17 have been demonstrated to regulate histone modifications in association with different histone acetyltransferases and deacetylases in a context-specific manner (55–57). Moreover, UAP56/DDX39B has been found to interact with active chromatin and unwinds harmful RNA-DNA hybrids (71), a structure which is very likely formed by nuclear-retained linear and circular RNAs (47,64). Therefore, more effort is needed to clarify the detailed mechanism of Hlc-mediated chromatin opening, which is under way in our lab.

DATA AVAILABILITY

RNA-seq data were deposited in the Gene Expression Omnibus (GEO) repository (GSE205804) of NCBI.

Click here for additional data file.