The long noncoding RNA NEAT1 and nuclear paraspeckles are up-regulated by the transcription factor HSF1 in the heat shock response

The long noncoding RNA (lncRNA) NEAT1 (nuclear enriched abundant transcript 1) is the architectural component of nuclear paraspeckles, and it has recently gained considerable attention as it is abnormally expressed in pathological conditions such as cancer and neurodegenerative diseases. NEAT1 and paraspeckle formation are increased in cells upon exposure to a variety of environmental stressors and believed to play an important role in cell survival. The present study was undertaken to further investigate the role of NEAT1 in cellular stress response pathways. We show that NEAT1 is a novel target gene of heat shock transcription factor 1 (HSF1) and is up-regulated when the heat shock response pathway is activated by sulforaphane (SFN) or elevated temperature. HSF1 binds specifically to a newly identified conserved heat shock element in the NEAT1 promoter. In line with this, SFN induced the formation of NEAT1-containing paraspeckles via an HSF1-dependent mechanism. HSF1 plays a key role in the cellular response to proteotoxic stress by promoting the expression of a series of genes, including those encoding molecular chaperones. We have found that the expression of HSP70, HSP90, and HSP27 is amplified and sustained during heat shock in NEAT1-depleted cells compared with control cells, indicating that NEAT1 feeds back via an unknown mechanism to regulate HSF1 activity. This interrelationship is potentially significant in human diseases such as cancer and neurodegenerative disorders.

The long noncoding RNA (lncRNA) NEAT1 (nuclear enriched abundant transcript 1) is the architectural component of nuclear paraspeckles, and it has recently gained considerable attention as it is abnormally expressed in pathological conditions such as cancer and neurodegenerative diseases. NEAT1 and paraspeckle formation are increased in cells upon exposure to a variety of environmental stressors and believed to play an important role in cell survival. The present study was undertaken to further investigate the role of NEAT1 in cellular stress response pathways. We show that NEAT1 is a novel target gene of heat shock transcription factor 1 (HSF1) and is up-regulated when the heat shock response pathway is activated by sulforaphane (SFN) or elevated temperature. HSF1 binds specifically to a newly identified conserved heat shock element in the NEAT1 promoter. In line with this, SFN induced the formation of NEAT1-containing paraspeckles via an HSF1-dependent mechanism. HSF1 plays a key role in the cellular response to proteotoxic stress by promoting the expression of a series of genes, including those encoding molecular chaperones. We have found that the expression of HSP70, HSP90, and HSP27 is amplified and sustained during heat shock in NEAT1-depleted cells compared with control cells, indicating that NEAT1 feeds back via an unknown mechanism to regulate HSF1 activity. This interrelationship is potentially significant in human diseases such as cancer and neurodegenerative disorders. NEAT1 (nuclear enriched abundant transcript 1) is a highly abundant long noncoding RNA (lncRNA) 2 that is essential for the formation of specific nuclear bodies called paraspeckles (1)(2)(3). There are two overlapping isoforms of NEAT1 transcribed from the same promoter: NEAT1_1 of 3.7 kb and NEAT1_2 of 22.3 kb (2)(3)(4). NEAT1_2 is indispensable for paraspeckle formation and is generated when the polyadenylation signal, and thus termination of the NEAT1_1 transcript, is suppressed by a heterogeneous nuclear ribonucleoprotein kinasedependent mechanism (4). Unlike NEAT1_1, the 3Ј end of NEAT1_2 is not polyadenylated but is processed by RNase P cleavage and subsequently stabilized through formation of a triple helical structure (3,5,6). Whereas NEAT1_1 is highly expressed in many tissues in mice, the expression patterns of NEAT1_2, and consequently the presence of paraspeckles, are more restricted (7). Recently, NEAT1 was found to be required for mammary gland development and lactation in mice (8). NEAT1 also has a critical role in corpus luteum formation (9). Even though the function of NEAT1 is still not fully understood, several reports have suggested that increased NEAT1 expression regulates the expression of certain genes by sequestering specific mRNAs and proteins into paraspeckles (10 -12). NEAT1 expression is up-regulated in response to different cellular stresses, including viral infections, proteasome inhibition, oncogene-induced replication stress, and hypoxia (11)(12)(13)(14)(15)(16)(17). Emerging evidence suggests that NEAT1 plays a cytoprotective role, as cells deficient of NEAT1 display increased sensitivity toward stress-induced cell death (11,15). In line with this, NEAT1 was found to be transcriptionally activated by HIF2␣ in response to hypoxia in cancer cells and, more recently, reported as a p53 target gene that prevents replication stress and DNA damage induced by mutagenic agents and oncogenes (13,15,18,19). Interestingly, high levels of NEAT1 are associated with tumorigenic characteristics and poor clinical outcome in several human cancers (13,15,20).
Cells are constantly subjected to extrinsic and intrinsic stressors that might have detrimental effects unless neutralized by specific cytoprotective mechanisms. The heat shock response is a universal cellular defense mechanism toward agents causing proteotoxic stress (21,22). Elevated temperatures, as well as wide range of oxidative and electrophilic agents, cause misfolding and damage of cellular proteins that will lead to cellular dysfunction or death unless repaired and/or removed. The heat shock transcription factor 1 (HSF1) plays a key role in this response mechanism (21)(22)(23)(24). Under normal conditions, HSF1 is kept in an inactive form in the cytoplasm by a multichaperone complex consisting of Hsp90, Hsp70, Hsp40, and TriC (23,(25)(26)(27)(28)(29). Upon activation, HSF1 is released from the repressive complex, undergoes a series of post-translational modifications, and forms homotrimers that accumulate in the nucleus (21,23,30). Here, HSF1 stimulates the transcription of genes encoding proteins involved in repair and clearance of damaged proteins (21,23,31). HSF1 specifically binds to heat shock elements (HSE), inverted repeats of nGAAn, where "n" is any nucleotide, in the upstream regulatory regions of its target genes (32,33). Among the best-studied target genes of HSF1 are those encoding protein chaperones, including Hsp70 and Hsp90, that restore proteostasis by regulating folding, activity, and degradation of proteins (34,35). The heat shock response is attenuated when HSF1 is released from the promoters of its target genes and either degraded or re-engaged into the HSF1repressive multichaperone complex by a negative feedback mechanism (21,36).
Here, we report that the isothiocyanate compound sulforaphane (SFN) induces NEAT1 expression and paraspeckle formation in MCF7 cells. This is not dependent on the Keap1-NRF2 pathway but on binding and transcriptional activation of the NEAT1 promoter by HSF1. We have identified a HSE site in the NEAT1 promoter that is highly conserved among vertebrates. Moreover, we show that NEAT1 is up-regulated in response to heat shock demonstrating that up-regulation of NEAT1 is a general event in the heat shock response. Finally, we demonstrate that the expression of HSP70, HSP90, and HSP27 is enhanced and sustained in the heat shock response in NEAT1 knockdown cells, compared with control cells.

SFN induces NEAT1 expression and paraspeckle formation
Several lines of evidence clearly point toward NEAT1 being a stress-induced lncRNA that is involved in cytoprotection (11,13,15). NEAT1 expression has recently been shown to be induced by hypoxia and confers protection to hypoxia-induced cell death in breast cancer cells (15). To further determine the role of NEAT1 in oxidative stress, MCF7 cells were treated with the isothiocyanate sulforaphane (SFN), which triggers an antioxidative response in cells by modifying thiol groups in several proteins, including Keap1 in the Keap1-NRF2 pathway (37,38). NEAT1 expression was assessed by RT-qPCR using two different primer sets: one recognizing both isoforms, and one solely recognizing the long NEAT1_2 isoform (Fig. 1A). SFN potently and rapidly induced the expression of NEAT1 in MCF7 cells (Fig. 1A). Pretreatment of cells with N-acetylcysteine, a strong antioxidant and precursor of cellular GSH, counteracted the effect of SFN on NEAT1 expression (Fig. 1B).
Paraspeckles are dynamic ribonucleoprotein complexes that form around the NEAT1_2 isoform in the nucleus (4). To determine whether SFN-induced NEAT1 expression is associated with increased paraspeckle formation, we performed RNAfluorescence in situ hybridization (RNA-FISH) on untreated and SFN-treated MCF7 cells using probes recognizing the long NEAT1_2 isoform. Whereas NEAT1_2-containing punctas appeared small and scarcely distributed in the nucleus of untreated MCF7 cells, SFN treatment potently increased the numbers and the overall signal intensity of the paraspeckles (Fig. 1, C and D).

SFN-induced NEAT1 expression is not dependent on NRF2
SFN stimulates several stress signaling pathways in cells, of which the Keap1-NRF2 pathway is the most prominent. To determine whether NRF2 is involved in SFN-induced NEAT1 expression, MCF7 cells were transfected with an siRNA toward NRF2 and stimulated with SFN for 6 h. The NRF2 protein accumulated after 6 h of SFN treatment, but its depletion did not interfere with the induction of NEAT1 ( Fig. 2A). We also assessed the NEAT1 expression in control and NRF2-depleted cells after a prolonged treatment with SFN for 24 h. Elevated levels of NEAT1 were observed in both control and siNRF2transfected cells (Fig. 2B). In contrast, SFN-mediated induction of NQO1 mRNA, a well-established target of NRF2, was severely reduced in NRF2-depleted cells (Fig. 2C). We conclude that SFN-induced NEAT1 expression is not dependent on the Keap1-NRF2 pathway.

SFN-induced NEAT1 expression and paraspeckle formation are dependent on HSF1
SFNs, as well as other oxidants, have recently been shown to stimulate HSF1, the key transcription factor conferring cellular protection to agents causing protein misfolding (39,40). We therefore sought to determine whether SFN-induced NEAT1 expression depends on a mechanism involving HSF1. SFN treatment indeed induced a mobility shift of HSF1, which is associated with its activation, and nuclear accumulation of the protein (Fig. 3, A and B). Consistent with the observed shift and nuclear translocation of HSF1, SFN potently induced the expression of the Hsp70 mRNA, a prominent target gene of HSF1 (Fig. 3C). We next transfected MCF7 cells with two different siRNAs specifically silencing HSF1 expression, and we determined the effect on SFN-induced NEAT1 expression. Both siRNAs significantly reduced the increase in NEAT1 levels observed after SFN treatment (Fig. 3D). The same was observed when HSF1 expression was silenced in SFN-treated HeLa cells (Fig. 3E). To determine whether SFN-induced paraspeckle formation depends on HSF1, we performed co-immuno-FISH analyses on control and HSF1-depleted cells using an HSF1 antibody and probes specifically binding to NEAT1_2. In line with the observations described above, SFN enhanced the nuclear staining of HSF1 (Fig. 4, A and B) and the formation of NEAT1_2 containing paraspeckles (Fig. 4, A and C). Importantly, SFN-induced paraspeckle formation was severely compromised in HSF1-depleted cells (Fig. 4, A and C). Taken together, our data clearly demonstrate that HSF1 is essential for increased NEAT1 expression and paraspeckle formation as a response to SFN-treatment in MCF7 cells.

NEAT1 is transcriptionally regulated by HSF1
Having established that SFN induces NEAT1 expression by an HSF1-dependent mechanism, we next asked whether SFN treatment leads to transcriptional activation of the NEAT1 gene. A luciferase reporter vector containing nucleotides  Fig. 1. Depletion of NRF2 expression in whole-cell extracts was verified by Western blotting analyses using an NRF2 antibody. The membrane was re-probed with an anti-actin antibody to ensure equal loading. B and C, MCF7 cells were transfected as described in A and subjected to a long-term treatment with SFN (10 M) for 24 h. The expression of NEAT1 and NEAT1_2 (B) and NQO1 (C) was determined by RT-qPCR. Experiments were performed in triplicate, and the graph is representative of three independent experiments. (**, p Յ 0.01; *, p, Յ 0.05; ns, not significant.)

NEAT1 is a novel HSF1 target gene in the heat shock response
Ϫ4040 to ϩ144 of the NEAT1 upstream regulatory region was generated and transfected into MCF7 cells. Reporter gene assays were performed in extracts from untreated and SFN-treated cells. SFN significantly stimulated the NEAT1 promoter-driven luciferase activity (Fig. 5A). This stimulation was severely compromised upon co-transfection with an HSF1-di-

NEAT1 is a novel HSF1 target gene in the heat shock response
rected siRNA, demonstrating that SFN-induced activation of the NEAT1 promoter depends on HSF1 (Fig. 5B). HSF1 binds to heat shock elements (HSE) within its target genes that are composed of alternating inverted repeats of 5 bp, nGAAn where "n" is any nucleotide (32,33). We carefully inspected the NEAT1 promoter and identified three putative HSEs. One of these, located between nucleotides Ϫ445 and Ϫ431, specifically caught our attention as it is highly conserved between species (Fig. 5C). To determine whether this region is involved in SFNactivated NEAT1 transcription, a truncated construct of the The result is representative of three independent experiments. B, MCF7 cells were co-transfected with pNEAT1-luc and siHSF1_#2 as described under "Experimental procedures." Cells were left untreated or treated with SFN for 8 h, and luciferase assays were performed. C, sequence conservation within NEAT1 upstream regions is illustrated by PhyloP Basewise Conservation scores from 100 vertebrates (USCS Genome Browser). An alignment of conserved HSE core sequences from human, rhesus, mouse, dog, and elephant is shown. D, truncated mutant of the NEAT1 promoter luciferase reporter construct encompassing the putative HSE site was generated and transfected into MCF7 cells along with a version harboring four point mutations within the HSE consensus sequence. SFN-induced luciferase activity was measured 24 h post-transfection. E, MCF7 cells were left untreated or treated with SFN (20 M) for 6 h, and ChIP assays were performed using an anti-HSF1 antibody. qPCR was performed with primers flanking the HSE site. Primers flanking a region further upstream in the NEAT1 promoter ("upstr"), as well as primers amplifying a region of the GAPDH promoter, were used as negative controls. The relative amount of immunoprecipitated DNA compared with input DNA for each primer set is shown for the HSF1 ChIP. The values obtained by the IgG ChIP was less than 0.003% for the HSF1 and control primers. The result is representative of three independent ChIP experiments, where qPCR reactions were done as triplicates. (***, p Յ 0.001; **, p Յ 0.01; ns, not significant.) NEAT1 is a novel HSF1 target gene in the heat shock response NEAT1 promoter reporter vector was made containing nucleotides Ϫ470 to ϩ144. We also made a mutated version where we introduced four point mutations in the predicted HSE core, and both constructs were transfected into MCF7 cells. SFN potently stimulated transcription from the truncated NEAT1 promoter (Fig. 5D). This stimulation was absolutely dependent on an intact HSE core, as point mutations in this region totally abolished the SFN-induced increase in NEAT1 promoterdriven luciferase activation. To analyze whether HSF1 can bind to the NEAT1 promoter in vivo, ChIP experiments were conducted on untreated and SFN-treated MCF7 cells using an antibody against HSF1 and RT-qPCR primers amplifying a 100-bp fragment of the NEAT1 promoter encompassing the HSE site. HSE-containing NEAT1 promoter fragments co-precipitated with the HSF1 antibody (Fig. 5E). Importantly, SFN robustly increased HSF1 binding to NEAT1 HSE fragments. Primers amplifying a GAPDH fragment and a region of the NEAT1 promoter upstream of the HSE site ("upstr") were used as controls. Control ChIPs with IgG gave very high Ct values compared with that of the HSF1 antibody and, importantly, showed no differences upon SFN stimulation.

NEAT1 is induced by heat shock
Having established that NEAT1 levels are enhanced by an HSF1-dependent mechanism upon SFN treatment, we next sought to determine whether NEAT1 is induced as response to heat shock (HS). MCF7 cells were incubated at 43°C for 30 min and either harvested directly or after recovery at 37°C for the indicated periods. HSF1 was rapidly activated during HS as assessed by a mobility shift in Western blotting (Fig. 6A). This was accompanied by increased expression of the Hsp70 mRNA (Fig. 6B). Importantly, HS rapidly and transiently stimulated the expression of NEAT1 (Fig. 6C). This indicates that elevated NEAT1 expression is a general mechanism in the heat shock response pathway.

Proliferation is compromised and expression of HSF1 target genes is amplified in NEAT1-depleted cells
Elevated NEAT1 levels and paraspeckle formation in response to cellular stress are widely observed and believed to play a prosurvival role by regulating the expression of specific genes. To start unraveling the function of NEAT1 in the heat shock response, we measured the sensitivity of control and NEAT1-depleted cells to heat shock by cell confluence proliferation assays. MCF7 cells were transfected with NEAT1-specific GapmeR antisense oligonucleotides (ASOs), which generally reduced the NEAT1 expression by 70 -80% for up to 120 h or a control GapmeR. Cell confluence was then monitored for 96 h using the IncuCyte live cell analysis system. After the first 48 h, half of the cells were subjected to heat shock for 30 min and then returned to IncuCyte system for another 48 h. Strikingly, NEAT1 depletion severely decreased the confluency of MCF7 cells, indicating that NEAT1 is necessary for their proliferation or survival (Fig. 7A). The proliferation rate was not further decreased after heat shock compared with cells kept at 37°C over the whole-monitoring period (Fig. 7, A and B). Taken together, this suggests that NEAT1 is generally required for the proliferation or survival of MCF7 cells and that an additional stress, such as heat shock, does not further affect the already growth-inhibited cells. Control-transfected cells generally recovered well after heat shock with only a slight reduction in confluency (Fig. 7, A and B).
To further analyze the role of NEAT1 in the heat shock response, we assayed the expression of the HSF1 target genes encoding Hsp70, Hsp90, and Hsp27 in control and NEAT1depleted cells. MCF7 cells were transfected with two different GapmeR ASOs, which either targeted both isoforms of NEAT1 or solely the long NEAT1_2 isoform. Transfected cells were exposed to heat shock, and HSP70, HSP90, and HSP27 expression was assessed by RT-qPCR. Interestingly, the expression of

NEAT1 is a novel HSF1 target gene in the heat shock response
all target genes was repeatedly amplified and sustained in cells where NEAT1 was silenced, compared with cells transfected with a control GapmeR (Fig. 8). Moreover, the background expression in unstressed cells was slightly enhanced. Of note, a stronger effect on the HSF1 target genes was observed for cells transfected with the GapmeR targeting both isoforms of NEAT1, compared with those transfected with the GapmeR only silencing the NEAT1_2 isoform. Taken together, our data suggest that NEAT1 depletion, by some mechanism, potentiates the HSF1 activity by either creating additional proteotoxic stress in the cells or by regulating the turnover or the activity of the HSF1 protein.

Discussion
High-throughput RNA-Seq has demonstrated that most cells express a plethora of long noncoding transcripts (41,42). During the last few years, huge efforts have been made to reveal their biological function, and many of them now appear as important contributors to gene regulation at different levels. NEAT1 is the architectural component of nuclear ribonucleoprotein complexes called paraspeckles, and it has recently gained considerable attention as several reports have shown that the transcript is abnormally expressed in human diseases, including cancer (13,15,20). The function of NEAT1 remains elusive, but emerging evidence suggests that NEAT1 and paraspeckles have a role in cytoprotection. Here, we show that NEAT1 is induced at the transcriptional level by the isothiocyanate compound SFN. This is accompanied by increased paraspeckle formation. SFN mimics oxidative stress in cells by modifying thiol groups in cellular proteins and induces antioxidative response pathways of which Keap1-NRF2 is the most prominent (37,38). We demonstrate that SFN-induced NEAT1 expression is not dependent on NRF2. In contrast, depletion of HSF1 severely abrogates SFN-induced NEAT1 expression and paraspeckle formation. Several reports have shown that SFN and other sulfhydryl-reactive compounds can stimulate the heat shock response pathway in cells by activating HSF1 (39,40,43,44). The mechanism for how SFN activates HSF1 is somewhat obscure, but previous studies have shown that oxidative compounds might promote the DNA-binding activity of HSF1 by modifying cysteine residues in the DNA-binding domain (45,46). SFN has also been shown to modify Hsp90 and thereby disrupt complex formation between Hsp90 and its protein partners (47,48). Recently, Naidu et al. (44) reported that phenethyl isothiocyanate indeed modified cysteine residues within Hsp90 leading to dissociation and activation of HSF1.
Our results show that HSF1 accumulates in the nucleus upon SFN treatment and binds to the NEAT1 promoter in vivo. We have identified a conserved HSE in the NEAT1 promoter that is critical for SFN-induced transcriptional activation of the NEAT1 gene. Intriguingly, this site overlaps with a recently reported NF-Bbinding site, which is necessary for lipopolysaccharide-induced NEAT1 expression in lung cancer cells (49). An overlapping NF-B-and HSF1-binding site has been identified previously in the promotor of the gene encoding MHC class I chain-related protein A (MICA) (50). Here, HSF1 and NF-B bind mutually exclusive to the site, and overexpression of a truncated version of HSF1 containing only the DNAbinding domain outcompetes NF-B binding and abolishes TNF␣-induced MICA expression. The question as to whether the overlapping HSF1/NF-B site in the NEAT1 promoter represents a regulatory hub, coordinating outputs from different signaling pathways, remains to be resolved.
In this study, we show that NEAT1, as well as being induced by SFN, is also induced upon heat shock. This clearly suggests that NEAT1 up-regulation is a general phenomenon in the heat shock response. This is supported by a study by Hirose et al. (11), demonstrating that NEAT1 expression and paraspeckle formation are induced by inhibition of the 26S proteasome by MG132 or bortezomib. Proteasome inhibition causes a proteotoxic stress in the cells as proteins that are destined for degradation form aggregates in both the cytoplasm and the nucleus (11,51). Activation of HSF1 to induce expression of molecular chaperones is a general cellular response mechanism to proteasome inhibition (52-54). Thus, we envision that NEAT1 induction upon proteasome inhibition might be mediated by HSF1mediated transcriptional activation of the NEAT1 promoter.
Several reports have shown that NEAT1 depletion sensitizes cells to a variety of stressors. Thus, we hypothesized that knockdown of NEAT1 expression would make cells more susceptible to heat shock. However, we repeatedly observed that transient transfection with NEAT1 antisense oligos by itself dramatically reduced the proliferation of MCF7 cells and that this tendency was not reinforced by heat shock. This shows that MCF7 cells cultivated in vitro are highly dependent on NEAT1. To further dissect the function of NEAT1 in the heat shock response, we knocked down NEAT1 expression by antisense oligos and

NEAT1 is a novel HSF1 target gene in the heat shock response
assessed the effect on the expression of three HSF1 target genes encoding Hsp70, Hsp90, and Hsp27. Interestingly, knockdown of NEAT1 amplified and prolonged the expression of these target genes. The mechanism for this is still obscure. NEAT1 depletion abrogates the formation of paraspeckles (4). This might lead to mislocalization of paraspeckle-associated proteins that disturbs proteostasis in the cells and thereby contribute to the activation of HSF1. Alternatively, NEAT1 might regulate the turnover of the HSF1 protein or activity by a negative feedback mechanism. Interestingly, the effect of NEAT1 depletion on HSF1 target genes was significantly stronger when cells were transfected with a GapmeR targeting both isoforms compared with one only reducing NEAT1_2 expression. This indicates that the short NEAT1_1 isoform has an important function in the regulation of the heat shock response.
HSF1 plays a critical role in the cellular defense to proteotoxic stress. Many neurodegenerative diseases, including amyotrophic lateral sclerosis (ALS), Huntington's disease, and Alzheimer's disease are associated with the formation of protein aggregates (31,55). Loss of HSF1 expression or activity is frequently observed in these diseases (55)(56)(57)(58). Our results demonstrate that HSF1 activates the expression of NEAT1 during the heat shock response. Interestingly, several reports have shown that NEAT1 is abnormally expressed in ALS and Huntington's disease (59 -61). Moreover, mislocalization of two paraspeckle proteins, FUS (Fused in sarcoma) and TDP-43 (TAR DNAbinding protein-43) is well-known to be associated with ALS (62). It has been speculated that NEAT1 expression and paraspeckle formation might have a protective role in neuronal cells in early stages of ALS and Huntington's disease (60,61,63). In line with this, Hirose et al. (11) showed that mouse embryonic fibroblasts from NEAT1 knockout cells displayed an increased sensitivity to proteasome inhibitors causing formation of protein aggregates, compared with WT cells. The cross-talk between NEAT1, paraspeckle formation, subcellular localiza-tion of FUS and TDP-43, and HSF1 in these devastating diseases should be a focus of future research.
Constitutive activation of HSF1 and abnormal expression of NEAT1 are both frequently observed in human cancers (13, 15, 20, 64 -66). There is clear evidence that both HSF1 and NEAT1 have cytoprotective roles in tumors and are associated with poor prognosis. In this study, we demonstrate that NEAT1 is a novel target gene of HSF1. It remains to be determined whether there is any correlation between HSF1 activation and NEAT1 expression in cancer.

Cell culture and treatments
MCF7 (ATCC HTB-22 TM ) and HeLa (ATCC CCL-2 TM ) cells were purchased from American Type Culture Collection and maintained in minimal essential medium (Sigma) supplemented with 10% fetal bovine serum (Biochrom) and 1% penicillin/streptomycin (Sigma). MCF7 cells were cultured in the presence of 0.01 mg/ml insulin (Sigma). All cells were grown at 37°C in humidified condition containing 5% CO 2 . SFN (catalog no. S4441) and N-acetylcysteine (NAC, catalog no. A9165) were purchased from Sigma. SFN was added to the cells at a final concentration of 20 M for short-term treatments up to 8 h and at a final concentration of 10 M for long-term treatment (24 h). When included, NAC (5 mM) was added to the media 1 h before SFN treatment. To induce a cellular heat shock response, cells were incubated at 43°C for 30 min and then either harvested directly or returned to 37°C for recovery.

Reverse transcription and quantitative PCR
Cells were lysed in TRI Reagent, and total RNA was isolated with Direct-zol RNA MiniPrep (Zymo Research) according to the manufacturer. RNA concentration was measured by Nano-Drop 2000 (ThermoFisher Scientific), and cDNA synthesis of total RNA was performed with SuperScript TM IV reverse transcriptase (ThermoFisher Scientific). 2.5 M of random hexamer primer (ThermoFisher Scientific) and ϳ250 ng of template were used for the reaction. Total RNA was denatured at 65°C for 5 min, and cDNA was synthesized at 50°C for 10 min. Quantitative PCR was run on a LightCycler 96 (Roche Applied Science) with the SYBR Green Reaction Mix FastStart Essential DNA Green Master (Roche Applied Science) and 0.25 M forward and reverse primer. (Thermal cycle conditions used are as follows: 95°C for 10 min and 40 cycles of 95°C for 10 s, 60°C 10 s, and 72°C for 10 s.) All primers sequences are provided in Table 1. Experiments were done in triplicates, and the ⌬⌬Cq method was used for fold change calculations. GAPDH was used as reference gene.

Immunoblotting
Whole-cell extracts (WCE) were made by lysing cells directly in 2ϫ NuPAGE LDS Sample Buffer (ThermoFisher Scientific). Nuclear extracts (NE) were isolated using the NE-PER TM nuclear and cytoplasmic extraction kit (ThermoFisher Scientific) according to the manufacturer's instruction. In brief, cells were resuspended in Cytoplasmic Extraction Reagent I and II, and nuclei were pelleted by centrifugation at 16,000 ϫ g. The pellet was resuspended in ice-cold Nuclear Extraction Reagent, vortexed for 1 min, and incubated on ice for 10 min. This step was repeated three more times before centrifugation at 16,000 ϫ g for 10 min. Proteins were resolved on SDS-polyacrylamide gels and transferred to nitrocellulose membranes. Equal loading of proteins was verified by probing the membranes with an antibody recognizing actin (WCE) or lamin B (NE). The following primary antibodies were used, all at 1:000 dilution: rabbit anti-NRF2 (Abcam, catalog no. ab62359); rabbit anti-HSF1 (Cell Signaling, catalog no. 4356); rabbit anti-lamin B (Proteintech, catalog no. 12987-1-AP); and mouse anti-actin (Millipore, MAB1501). The blots were detected with IRDyeconjugated secondary antibodies (LI-COR Biosciences) at a 1:10,000 dilution (800CW goat anti-rabbit, catalog no. 926- Table 1 Primer and siRNA/ASO sequences F is forward, R is reverse. NEAT1 is a novel HSF1 target gene in the heat shock response 32211; 680LT goat anti-mouse, catalog no. 926-68020), and the Odyssey CLx IR Imaging System.

RNA-fluorescence in situ hybridization and immunofluorescence staining
Stellaris NEAT1 RNA FISH probes recognizing the NEAT1_2 isoform (VSMF-2251-5, Quasar 670-conjugated) were purchased from LGC Biosearch Technologies. Preparation of cells, hybridization, and mounting were performed according to the Stellaris RNA-FISH Probes manual. In brief, cells were seeded onto circular coverslips in 12-well dishes and allowed to attach for 2-3 days. They were fixed with 4% freshly made formaldehyde at room temperature, and permeabilized with 70% ethanol. Hybridization was done at 37°C in a humidifying chamber overnight. For co-immuno-FISH experiments, the hybridization was performed as described above, and cells were subsequently incubated in 1% RNase-free BSA for 30 min and then stained with anti-HSF1 antibody for 1 h (1:50, Cell Signaling, catalog no. 4356A). Cells were incubated with goat anti-rabbit Alexa 488-conjugated secondary antibody (1:500, ThermoFisher Scientific, catalog no. A11070) and mounted using Vectashield antifade mounting medium containing DAPI (Vector Laboratories, H-1200). Images were generated using a Zeiss LSM780 confocal microscope (Carl Zeiss Microscopy GmbH, Jena, Germany). In all samples, Z-stack (five slices, 2.5 m total height) images were taken at ϫ40 magnification. For all images, the middle Z-slice was positioned at DAPI's best focus. The same treatment and setting were applied to all replicates, and for each slide at least 10 pictures were taken for Volocity analysis. The Volocity software (PerkinElmer Life Sciences, version 6.3) was used to measure signal intensities for both NEAT1_2 and HSF1 signals. At least 250 cells in each group of treatments were analyzed by Volocity software. The mean intensity of NEAT1_2 or nuclear HSF1 signals in the SFN-treated group was normalized against control.

Reporter gene assays
Sub-confluent MCF7 cells in 12-well plates were transfected with 150 ng of luciferase reporter plasmids using Lipofectamine2000 reagent (ThermoFisher Scientific) according to the manual provided by the manufacturer. After 24 h, cells were either left untreated or treated with SFN (20 M) for 8 h. Cells were harvested, and luciferase assays were performed using the Dual-Light luciferase and ␤-galactosidase reporter gene assay system (ThermoFisher Scientific). Of note, cells were initially co-transfected with luciferase reporter plasmids and an expression vector for ␤-gal, but as SFN repeatedly interfered with the ␤-gal activity in the cells, the expression vector was omitted from the transfections, and only the luciferase activity was included in the analyses. Co-transfections with siRNA and plasmid DNA were performed in two steps using Lipofectamine2000. First, siRNAs were introduced into the cells by reverse transfection. After 48 h, plated cells were retransfected with plasmid DNA and left for another 24 h.

Chromatin immunoprecipitation (ChIP) assays
MCF7 cells were seeded at a density of 6 million cells per 10-cm dish the day before use. The cells were left untreated or treated with SFN (20 M) for 6 h before harvesting. Two 10-cm dishes were used per condition. The "iDeal ChIP-seq kit for Transcription Factors" (Diagenode, C01010055) was used for harvest and ChIP according to the manufacturer's instruction. The two dishes for each treatment were combined, and the approximate cell number was estimated to be 15 million cells. Volumes of buffers used in the kit were adjusted to this. Cells were fixed for 15 min. Sonication was performed in ice-cold water on a Bioruptor UCD-200 (Diagenode) for 3 ϫ 10 minutes using 30-s on/off pulses. Samples run on an agarose gel showed a majority of DNA with size from 100 to 400 bp after shearing. For immunoprecipitation, 10 l of anti-HSF1 antibody (Cell Signaling, 4356) or 1 l of IgG (provided with the kit) was used with 200 l of sheared chromatin. Two l (1%) of input chromatin was set aside. The eluate had a volume of 25 l, which was diluted 1:10 before 5 l was used in a qPCR. qPCR was performed in triplicate on a LightCycler 96 (Roche Applied Science). The relative amount of immunoprecipitated (IP) DNA compared with input DNA was calculated using the "percent input method" as follows: Because the input chromatin was 1%, a dilution factor of 100 (6644 cycles, log2 of 100) was subtracted to adjust input Ct value to 100%. To calculate the percentage of specific chromatin co-immunoprecipitated with the HSF1 antibody or the IgG control, the triplicate average Ct values, Ct(IP), for the specific qPCR primers (HSE, "upstream," and GAPDH) were used in the equation: 100⅐2∧(adjusted input Ϫ Ct(IP)). Primer sequences are given in Table 1.

Cell confluence proliferation assay
MCF7 cells were transfected in solution with indicated LNA-GapmeR antisense oligos and seeded in 96-well plates at an initial confluency of ϳ30% (20,000 cells per well) and immediately placed in an IncuCyte S3 live-cell analysis system, which is a fully equipped automated microscope for cell confluence monitoring. Three phase-contrast images were acquired from each well at 120-min intervals over a period of 96 h, using a ϫ20 objective. For each condition, five wells were monitored. Data were analyzed using the IncuCyte S3 software.

Statistics
GraphPad software (Prism version 7, Mac OS X) was used to analyze quantitative data. Statistical significance was evaluated with unpaired Student's t test or one-way ANOVA followed by the Dunnett's multiple comparison test. The data were considered statistically significant when p Յ 0.05. For all experiments, significance is expressed as ***, p Յ 0.001, **, p Յ 0.01, and *, p Յ 0.05. The error bars indicate Ϯ S.D. in all figures. All the experiments were performed at least three times.