Identification of a heat-inducible novel nuclear body containing the long noncoding RNA MALAT1 Free

Heat shock causes protein unfolding and nonspecific aggregation that cause proteotoxicity (Richter et al., 2010). The heat-shock response (HSR) is an evolutionarily conserved cellular defense mechanism for eliminating proteotoxicity and maintaining protein homeostasis (proteostasis). The HSR induces transcription of multiple gene products, including molecular chaperones, which assist protein folding. Heat shock factor 1 (HSF1) is a critical transcription factor in the HSR, and induces target genes, such as the chaperone genes (Li et al., 2017). The HSF1–chaperone axis is considered the central HSR in that it increases protein folding capacity and rebalances proteostasis, thereby supporting cell survival. However, a recent genome-wide analysis of transcriptional regulation using HSF1-knockout cells revealed that HSR acts not only through HSF1-dependent transcriptional activation but also through HSF1-independent transcriptional regulation (Mahat et al., 2016).

The cell nucleus contains many nuclear bodies, which are composed of specific proteins and RNAs (Banani et al., 2017; Hyman et al., 2014). Nuclear bodies show liquid-like properties that arise through liquid–liquid phase separation (Banani et al., 2017; Hyman et al., 2014). Nuclear bodies also contribute to cellular stress responses. For example, a nuclear stress body is formed upon heat shock in an HSF1-dependent manner (Cotto et al., 1997; Jolly et al., 1999) and regulates the intron retention of various pre-mRNAs in heat-shocked cells (Ninomiya et al., 2020).

Long noncoding RNAs (lncRNAs) are transcripts that lack obvious protein coding potential and regulate gene expression (Engreitz et al., 2016; Shirahama et al., 2020). A subset of lncRNAs are components of nuclear bodies (Hirose et al., 2019). NEAT1 is an lncRNA essential for forming the nuclear body, paraspeckle (Clemson et al., 2009), and plays an important role in the cellular immune response to pathogenic infection (Imamura et al., 2014, 2018). Metastasis associated lung adenocarcinoma transcript 1 (MALAT1) is a component of other nuclear bodies, termed nuclear speckles, which are composed of pre-mRNA splicing and transcription factors (Hutchinson et al., 2007; Miyagawa et al., 2012). MALAT1 was originally identified as a prognostic marker of metastasis and patient survival in non-small cell lung carcinoma (Ji et al., 2003). We have reported that MALAT1 promotes cell migration through regulation of cell migration-related gene expression at the transcriptional level (Tano et al., 2010) and that it represses transcription of the p53 gene (TP53) (Tano et al., 2018). MALAT1 also regulates alternative splicing by regulating splicing factors in nuclear speckles (Tripathi et al., 2010). These studies suggest that MALAT1 contributes to cancer progression and metastasis through regulation of gene expression (Arun et al., 2020). Despite the importance of MALAT1 in gene regulation in cancer, Malat1-knockout mice show no obvious phenotype under regular breeding conditions (Eißmann et al., 2012; Nakagawa et al., 2012; Zhang et al., 2012). Hence, the physiological function of MALAT1 is an open question to date.

In this study, we showed that MALAT1 translocates in response to heat shock from nuclear speckles to a distinct granule-like structure, named heat-inducible noncoding RNA-containing nuclear body (HiNoCo body), in an HSF1-independent manner.

We examined the nuclear localization of MALAT1 in response to heat shock by RNA fluorescence in situ hybridization (FISH). Under control conditions, MALAT1 was localized in nuclear speckles probed by immunofluorescence staining of the core components, SRSF2 and SRRM2, in human alveolar adenocarcinoma cell line A549 cells (Fig. 1A; Fig. S1A). Heat treatment at 42°C induced the translocation of MALAT1 from nuclear speckles to granule-like structures adjacent to the nuclear speckles (Fig. 1A).

We analyzed whether other stresses also induced the translocation of MALAT1 into granule-like structures adjacent to the nuclear speckles. Sodium arsenate treatment led to MALAT1 redistributing to two locations; primarily a diffusion into the nucleoplasm and with a minor localization in stress bodies marked by G3BP (also known as G3BP1), a stress body marker protein (Fig. S1D,E). Ultraviolet irradiation caused the diffusion of MALAT1 into the nucleoplasm, with a small granule-like pattern (Fig. S1F). MALAT1 was mainly localized in nuclear speckles but a significant amount was diffused into the nucleoplasm with a small granule-like pattern seen in X-ray-irradiated cells (Fig. S1F). As shown in previous reports (Bernard et al., 2010; Tani et al., 2012a), transcriptional inhibition by actinomycin D caused MALAT1 to diffuse into nucleoplasm. Thus, heat shock induced a distinct localization of MALAT1 in the nucleus.

Next, we investigated whether transcriptional inhibition affects MALAT1 translocation into the granule-like structures under heat shock. As shown in Fig. S1G, transcriptional inhibition by actinomycin D or 5,6-dichloro-β-D-ribofuranosylbenzimidazole (DRB) led to MALAT1 diffusion into the nucleoplasm with small granule-like particles spreading throughout the nucleoplasm under heat shock (Fig. S1G).

To further investigate the MALAT1-containing granule-like structures, we examined the localization of SON and SRSF3, typical nuclear speckle proteins (Fig. S2A) (Sharma et al., 2010; Tripathi et al., 2012). Under heat shock, SON was colocalized with SRSF2. A large amount of SRSF3 was diffuse in the nucleoplasm and a small amount of SRSF3 overlapped with both SRSF2 and MALAT1 in the control and heat-treated A549 cells (Fig. S2A). The number of MALAT1-containing granule-like structures was significantly fewer than the number of nuclear speckles in heat-treated cells (Fig. S2B). A previous study has reported that nuclear speckles have a multi-layered structure (Fei et al., 2017). By using high-resolution imaging with deconvolution, we confirmed that SRSF2 and SON were centered in nuclear speckles and that MALAT1 was present around SRSF2 and SON (Fig. S2C). MALAT1 showed a relatively homogenous distribution in the granule-like structures under heat shock (Fig. S2C).

The heat shock-mediated translocation of MALAT1 was also observed in mouse embryonic fibroblasts (MEFs; Fig. 1B; Fig. S1B) and mouse C17.2 cells (Fig. S1C), indicating that the heat shock-induced change in the MALAT1 nuclear localization is evolutionarily conserved and occurs in normal primary cells as well as in cancer-derived cells.

To investigate whether the heat-shock-induced translocation of MALAT1 is mediated by either pre-existing or newly synthesized MALAT1, we measured MALAT1 RNA turnover upon heat shock through immunoprecipitation of pulse-labeled RNA with bromouridine (BrU), followed by reverse transcription quantitative real-time PCR (qRT-PCR) (Fig. 2A). The overall amount of ACTB mRNA, MYC mRNA and MALAT1 lncRNA did not change upon heat shock, whereas the HSPB1 mRNA level increased by 3.4-fold (P<0.01; Fig. 2B). At 1 h post pulse-labeling, the recovered BrU-labeled ACTB mRNA, HSPB1 mRNA and MALAT1 lncRNA did not significantly reduce in both the control and heat-shock groups, indicating that these RNAs are stable and that most of them pre-existed at 1 h after heat shock (Fig. 2B,C). In contrast, the recovered BrU-labeled MYC mRNA was reduced to 20% (P<0.01) in both the control and heat-shock groups, indicating that MYC mRNA has a high turnover rate and that most of the MYC mRNA was newly synthesized mRNA at 1 h after heat shock (Fig. 2B,C). Northern blot analysis showed that the length and amount of MALAT1 were not altered in heat-shocked cells (Fig. 2D). We investigated whether heat shock affected the processing of MALAT1-associated small cytoplasmic RNA (mascRNA), a tRNA-like transcript generated from the 3′ end of MALAT1 RNA (Wilusz et al., 2008). There was no significant difference in the expression level of unprocessed MALAT1 in non-treated and heat-treated A549 cells, indicating heat shock does not alter the mascRNA processing from MALAT1 (Fig. S2D,E). These results show that heat shock changes the nuclear localization of MALAT1 from nuclear speckles to other nuclear structures without changing its expression level, processing and turnover rate.

To determine the biological significance of MALAT1 in the HSR, we examined the cell proliferation of wild-type and MALAT1-knockout A549 cells with or without heat shock treatment. Under control conditions, there was no significant difference in cell proliferation between wild-type and MALAT1-knockout A549 cells (Fig. 2E). However, the cell proliferation of MALAT1-knockout A549 cells was reduced after heat shock, compared with heat-treated wild-type cells. This suggests that MALAT1 is important for the HSR.

Co-staining of MALAT1 and marker proteins of known nuclear bodies in A549 cells after heat shock showed that MALAT1 was not colocalized with any known nuclear bodies, such as nuclear stress body, paraspeckles, Cajal body, nucleoli, gems and promyelocytic leukemia bodies labeled by HSF1, SFPQ, coilin, FBL, SMN1 and PML, respectively (Fig. 3A). These results indicate that heat shock induces the translocation of MALAT1 from nuclear speckles to a distinct nuclear body. We named this MALAT1-containing nuclear body heat-inducible noncoding RNA-containing nuclear body, with the acronym HiNoCo, which means fire spark in Japanese.

HSF1 is the master transcription regulator of the HSR (Li et al., 2017). We therefore examined the formation of HiNoCo-bodies in HSF1-knockdown cells (Fig. 3B,C). HSF1 knockdown was confirmed by qRT-PCR and western blotting (Fig. 3B), in which HSF1 mRNA and protein decreased to less than 10% of that in control cells. Heat shock increased HSPB1 mRNA expression, which is controlled by HSF1 (Trinklein et al., 2004), by 3.5-fold in control cells, but it was unchanged in HSF1-knockdown cells (Fig. 3C), indicating that functional knockdown of HSF1 was successfully established. The MALAT1 RNA level was unchanged in HSF1-knockdown cells before and after heat shock (Fig. 3C). The formation of HiNoCo bodies probed by MALAT1 was similar in control and HSF1-knockdown cells (Fig. 3D), and the size of HiNoCo bodies was not significantly altered in HSF1-depleted cells (Fig. 3E), indicating HiNoCo body is formed through an HSF1-independent mechanism.

To get an insight in the protein factor(s) of the HiNoCo body, we performed knockdown experiments using previously reported protein factors that affected MALAT1 localization in nuclear speckles (Miyagawa et al., 2012; Tripathi et al., 2010; Wang et al., 2019). The knockdown of SRRM1, RNPS1, DHX15, SON, DDX42, HNRNPH1 and HNRNPK, which have been reported to negatively regulate the nuclear speckle localization of MALAT1, promoted the diffusion of MALAT1 in the nucleoplasm of non-treated and heat-treated A549 cells (Fig. S3). Unlike previous studies, the knockdown of PRP6 did not cause the disruption of nuclear speckle localization of MALAT1 in non-treated A549 cells, but promoted the diffusion of MALAT1 into the nucleoplasm in heat-treated A549 cells (Fig. S3). However, PCBP1 knockdown caused the retention of MALAT1 in nuclear speckles even under heat shock (Fig. S3). These results indicate that PCBP1 is required for the relocation of MALAT1 from nuclear speckles in response to heat shock.

Examining the time course of HiNoCo body formation after heat shock at 42°C (Fig. 4A,B) revealed that MALAT1 started to diffuse from nuclear speckles into the nucleoplasm at 5 min and diffused mostly in the nucleus at 10 and 15 min. MALAT1 formed small granules at 30 min and large HiNoCo bodies at 60 min. Even after 3 h of heat shock, HiNoCo bodies remained within the nucleus, although the size of these bodies tended to be smaller than they were 1 h after heat shock (Fig. S4A).

When the incubation temperature was changed from 42°C to 37°C, the MALAT1 was not localized in nuclear speckles at 10 min after the temperature shift, but it partially returned to the nuclear speckles 30 min after recovery and completely after 60 min of recovery (Fig. 4A,B). Quantitative analysis using LAS-AF software clearly showed that MALAT1 exited from nuclear speckles in response to heat shock and it returned to nuclear speckles after reducing the temperature back to 37°C (Fig. 4B). When a second heat shock at 42°C was applied after 60 min of recovery, MALAT1 again formed HiNoCo bodies (Fig. S4B). MALAT1 localization was not altered at 10 min after the change in incubation temperature from 42°C to 37°C after the second heat shock treatment, but returned to nuclear speckles 30 min after recovery similar to the first recovery step (Fig. S4B).

Poly(A)+ RNA was localized in nuclear speckles at every time point of the heat shock and the recovery step (Fig. S4C). We also found that U1snRNA was colocalized with nuclear speckles in the non-treated and heat-treated cells (Fig. S4D), indicating that the formation of the HiNoCo body is specific for MALAT1, but not for poly(A)+ RNA and U1snRNA.

HiNoCo bodies showed a granule-like shape that is a typical shape of phase-separated structures (Hyman et al., 2014). HiNoCo bodies gradually grow over time upon stress (Fig. 4A), similar to what is seen for stress granules, which are cytoplasmic phase-separated structures (Wheeler et al., 2016). We therefore hypothesized that the HiNoCo body is formed by liquid–liquid phase separation.

The HiNoCo body granules containing MALAT1 in heat-shocked cells disintegrated upon treatment with 1,6-hexanediol (Fig. 5A), an inhibitor of phase-separated structures but not solid protein assemblies (Kroschwald et al., 2017). HiNoCo body formation was observed in more than 80% of nuclei and 1,6-hexanediol treatment disrupted HiNoCo body formation in 80% of nuclei (Fig. 5B). However, this treatment did not affect MALAT1 localization in nuclear speckles in non-heat-treated cells nor the formation of nuclear speckles themselves in both non-treated and heat-treated cells (Fig. 5A; Fig. S5). In contrast, 2,5-hexanediol, which does not disrupt phase-separated structures (Lin et al., 2016), did not affect the localization of MALAT1 in both non-treated and heat-treated cells (Fig. 5A; Fig. S5). Finally, the expression level of MALAT1 was not altered by 1,6-hexanediol and 2,5-hexanediol (Fig. 5C). These results suggest that HiNoCo body is a phase-separated structure.

In this study, we showed that MALAT1 translocates thermo-responsively and reversibly to a distinct nuclear granular structure in an HSF1-independent manner in response to heat shock. We named this distinct nuclear granular structure the HiNoCo body. The HiNoCo body is a novel nuclear body that is evolutionarily conserved in human and mouse cells, and is distinct from nuclear stress body, which is formed only in primate cells upon heat shock at the primate-specific pericentromeric satellite III arrays near their own loci in HSF1-dependent manner (Jolly et al., 2004; Rizzi et al., 2004). Of note, we found that MALAT1 knockout reduced cell proliferation after heat-shock treatment, suggesting that HiNoCo body is important for the HSR.

The number of granules containing either Malat1 or Srrm2 in MEFs (Fig. 1B) and in C17.2 cells (Fig. S1C) tended to be more numerous under heat shock, and their size was smaller than nuclear speckles in non-treated cells. These differences may be due to the difference in the HSR system between humans and mice, such as the presence or absence of nuclear stress bodies. Upon heat shock stress, the number of HiNoCo bodies was reduced compared with nuclear speckles (Fig. S2B). This may reflect the difference in components present in nuclear speckles and HiNoCo bodies.

Our time course analysis of HiNoCo body formation revealed that MALAT1 diffused from nuclear speckles to the nucleoplasm and formed small granules that increased in size (Fig. 4). PCBP1 depletion prevented MALAT1 disassociation with nuclear speckles during heat shock (Fig. S3), suggesting that PCBP1 is required for the release of MALAT1 from nuclear speckles (Fig. 5D). Moreover, we found that MALAT1 could not form HiNoCo bodies in the presence of a transcriptional inhibitor (Fig. S1G), suggesting that active transcription is important for the formation of HiNoCo body. Based on these results, we propose a model in which MALAT1 is released from nuclear speckles by a PCBP-mediated pathway soon after heat shock, and MALAT1 localizes into the HiNoCo body, in a manner that is dependent on active transcription (Fig. 5D). Of note, there is a possibility that the diffusion of MALAT1 into the nucleoplasm reflects transcriptional inhibition by heat shock stress for the following reasons: (1) heat shock causes global transcriptional repression and certain genes are downregulated rapidly after heat shock treatment (Mahat et al., 2016); and (2) transcriptional inhibition prevents the nuclear speckle localization of MALAT1 (Bernard et al., 2010; Tani et al., 2012b).

Heat shock also enhances the formation of stress granules, which are phase-separated membraneless compartments in the cytoplasm (Protter and Parker, 2016). Stress granules in the cytoplasm serve as a sensing system for heat shock or pH stress using the physicochemical properties of liquid–liquid phase separation in yeast (Franzmann et al., 2018; Kroschwald et al., 2018; Riback et al., 2017). The sensitivity of HiNoCo body to 1,6-hexanediol supports the idea that HiNoCo body is a phase-separated membraneless granule, like stress granules in cytoplasm (Fig. 5A), although 1,6-hexanediol is widely used but not necessarily definitive for liquid-liquid phase separation (Alberti et al., 2019). In addition, we speculate that HiNoCo body formation may act as a biosensor in the nucleus to monitor temperature changes through the phase separation phenomenon. Further analysis of HiNoCo body formation may contribute to elucidating the cellular stress response system.

MEFs (kindly supplied by Dr Yumiko Kawamura, University of Tokyo, Japan), A549 cells, A549 MALAT1 knockout cells (Gutschner et al., 2013) and C17.2 cells (kindly provided by Dr Yasunobu Uchijima, University of Tokyo, Japan) were cultured in Dulbecco's modified Eagle's medium (DMEM) supplemented with 10% heat-inactivated fetal bovine serum and antibiotics at 37°C in a humidified incubator with 5% CO₂. To induce heat shock stress, MEFs, A549 and C17.2 cells were placed in a humidified incubator at 42°C with 5% CO₂ for the indicated periods. For sodium arsenate treatment, 0.5 mM of sodium arsenate (Sigma-Aldrich, St Louis, MO, USA, #S7400) was added to A549 cells and incubated in a humidified incubator with 5% CO₂ at 37°C for 1 h. To induce ultraviolet (UV) stress, A549 cells were irradiated by 10 J/m² UV light and placed in a humidified incubator with 5% CO₂ at 37°C for 8 h. For X-ray irradiation, A549 cells were exposed to 5 Gy of X-rays and incubated in a humidified incubator with 5% CO₂ at 37°C for 2 h after irradiation. To inhibit transcription, 2 μg/ml of actinomycin D (Sigma-Aldrich, #A9415) was added to A549 cells that were incubated in a humidified incubator with 5% CO₂ at 37°C for 27 h. For inhibition of liquid–liquid phase separation, 1,6-hexanediol (Wako, Osaka, Japan, #087-00432) or 2,5-hexanediol (Sigma-Aldrich, #11904-10G) (final concentration 50 mM) was added to non-treated (incubated at 37°C) and heat-treated (incubated at 42°C for 50 min) A549 cells for 10 min.

Total RNA was extracted from cells with RNAiso Plus (#9109, TaKaRa Bio Inc., Shiga, Japan) according to the manufacturer's instructions. Isolated RNA was reverse transcribed into cDNA using PrimeScript RT Master Mix (Perfect Real Time; #RR036B, TaKaRa Bio Inc.). The cDNA was amplified using the primer set listed in Table S1. SYBR Premix Ex Taq2 (Perfect Real Time; #RR820W, TaKaRa Bio Inc.) was used according to the manufacturer's instructions. qRT-PCR analysis was performed using a Thermal Cycler Dice Real Time system (TaKaRa Bio Inc.), according to the manufacturer's instructions.

Total RNA was extracted from non-treated or heat-treated (42°C for 1 h) cells with RNAiso Plus according to the manufacturer's instructions. Total RNA (10 µg) was separated on a 1.1% (w/v) agarose gel containing 0.08% formaldehyde and transferred to a positively charged nylon membrane (Millipore, Bedford, MA, USA). After UV cross-linking, blots were hybridized to previously described ³²P-labeled riboprobes (Tano et al., 2010) at 52°C overnight in Ultrasensitive Hybridization Buffer (Applied Biosystems, Foster City, CA, USA). An autoradiograph was captured and quantified using an FLA9000 biomolecular imager (FUJIFILM, Tokyo, Japan).

BrU pulse-chase analysis (BRIC-qPCR analysis) was performed as described previously (Tani et al., 2012a). Briefly, A549 cells were incubated at 37°C in the presence of 150 µM BrU (Wako, Osaka, Japan) for 24 h in a humidified incubator with 5% CO₂. After replacing the BrU-containing medium with BrU-free medium, cells were incubated at 37°C or 42°C for 1 h. Cells were harvested and total RNA was isolated using RNAiso Plus (TaKaRa), followed by isolation of BrU-labeled RNA by anti-BrdU mouse antibody (clone 2B1, MBL, Nagoya, Japan). The isolated RNA was quantified by qRT-PCR.

Cell viability was assessed using a Cell Counting Kit-8 (Dojindo, Kumamoto, Japan) at 0, 24, 48, and 72 h after heat shock treatment (42°C, 1 h). The procedures were performed according to the manufacturer's instructions.

Cells cultured on cover glasses in 12-well plates were washed with PBS(-) (PBS without Ca²⁺ and Mg²⁺) and fixed with 4% PFA in PBS(-) for 20 min at room temperature. Then, the cover glasses were washed twice with PBS(-) and permeabilized with 0.5% Triton X-100 in PBS(-) for 5 min at 4°C. The cover glasses were washed twice with PBS(-) and incubated in hybridization buffer [50% formamide, 1× Denhardt's solution, 1× SSC, 10 mM EDTA, 0.01% Tween 20, 0.1 mg/ml yeast tRNA (Sigma-Aldrich, #R8508)] for 2 h at 55°C. Digoxigenin (DIG)-labeled antisense RNA probe mixed in hybridization buffer was added to the cover glasses, and incubated for 16 h at 55°C. The cover glasses were washed twice with wash buffer (50% formamide, 2× SSC and 0.01% Tween 20) for 15 min at 55°C, and then incubated in 5 µg/ml RNaseA (Sigma-Aldrich, #R4642) in TNET (50 mM Tris-HCl, pH 7.4, 140 mM NaCl, 5 mM EDTA, 0.01% Triton X-100) for 30 min at 37°C. Subsequently, the cover glasses were incubated in 2× SSC plus 0.01% Tween 20 for 15 min at 55°C and washed twice with 0.1× SSC plus 0.01% Tween 20 for 15 min at 55°C. After blocking with 1% BSA in PBS(-) with 0.01% Tween 20 for 1 h at room temperature, the cover glasses were incubated in 1% BSA in PBS(-) with 0.01% Tween 20 containing anti-DIG sheep antibody (Roche, Mannheim, Germany) and antibodies against the proteins of interest (Table S2). After washing with PBS(-) plus 0.2% Tween 20, a solution of 1% BSA in PBS(-) with 0.01% Tween 20 with Alexa Fluor 488-, Alexa Fluor 555-, or Alexa Fluor 647-conjugated secondary antibodies was added to the cover glasses for 1 h at room temperature. After washing with PBS(-) plus 0.2% Tween 20, the cover glasses were incubated in DAPI in PBS(-) for 10 min at room temperature. The cover glasses were mounted with Prolong Gold (#P36934, Invitrogen, Carlsbad, CA, USA).

To prepare the template cDNA for the synthesis of the RNA probe, the MALAT1 fragment (1930–5644 nt in NR_002819.4) was subcloned into pSC-B vector (Stratagene, La Jolla, CA, USA). A DIG-labeled MALAT1 RNA probe was synthesized using DIG RNA-labeling mix (#11277073910, Roche) and T3 RNA polymerase (#11031163001, Roche) according to the manufacturer's instructions. For in situ hybridization of nuclear poly(A)+ RNA, an Alexa Fluor 594-conjugated oligonucleotide dT34 probe (previously described in Miyagawa et al., 2012)] was used. The antibodies used are listed in Table S2.

The images were acquired with a Leica TCS SP5 confocal microscope (Leica microsystems, Wetzlar, Germany). More than 100 cells were observed in each experiment. To examine the colocalization of MALAT1 and SRSF2, the colocalization rate was determined using the colocalization tool in the LAS-AF software (Leica microsystems) under the following conditions and calculations: threshold 30%, background 35%; colocalization rate (%)=area colocalization/area foreground, and area foreground=area image – area background. To analyze the number or size of HiNoCo bodies, images were taken as Z-stacks and projected in one plane with maximum intensity using Fiji/ImageJ (Image/Stack/Z Project). Z-projected images were binarized using Fiji/ImageJ (Image/Adjust/threshold, threshold 30) and the size of HiNoCo bodies were measured by the Analyze particle tool in Fiji/ImageJ.

High-resolution images were acquired with a Leica STELLARIS5 confocal microscope (Leica Microsystems, Wetzlar, Germany) with the following conditions: objective lens 63×oil lens, pinhole 0.5 AU; and pixel size 29 nm×29 nm. The images were submitted to automated internal Leica Lightning deconvolution software with the default setting.

The siRNA sequences are listed in Table S3. To improve knockdown efficiency, siRNAs were transfected into cells sequentially using Lipofectamine RNAi MAX (Invitrogen), according to the manufacturer's instructions. Briefly, siRNA duplexes (final concentration 10 nM) were transfected into 4×10⁴ cells (first transfection), followed by incubation for 6 h at 37°C and 5% CO₂. After changing the medium, the cells were incubated overnight at 37°C and 5% CO₂. The following day, siRNA duplexes (final concentration 10 nM) were transfected into the cells again (second transfection), followed by incubation for 6 h at 37°C and 5% CO₂. After changing the medium, the cells were incubated for 24 h at 37°C and 5% CO₂.

Total protein extracts were prepared in RIPA buffer [50 mM Tris-HCl pH 8, 150 mM NaCl, 1 mM EDTA, 0.1% SDS, 1% Triton X-100, 0.1% DOC, and protease inhibitor cocktail (#P8340, Sigma-Aldrich)] and separated on 10% SDS-PAGE gels. The proteins were transferred onto a PVDF membrane (Millipore). The membrane was blocked with 3% BSA in Tris-buffered saline containing 0.1% Triton X-100 (TBST) and incubated with anti-HSF1 or anti-GAPDH antibody (Table S2) for 1 h at room temperature. After washing the membrane with TBST, the membrane was incubated with horseradish peroxidase (HRP)-linked secondary antibodies for 1 h at room temperature. The proteins were visualized using Immobilon western chemiluminescent HRP substrate (#WBLUF0500, Millipore) and LAS-4000UVmini Luminescent Image Analyzer (FUJIFILM), according to the manufacturer's instructions.

The qRT-PCR analysis was performed on biological triplicates. The statistical significance was assessed by Welch's t-test with Bonferroni correction.

We thank Dr Yoshihiro Urade and Dr Kenzui Taniue for fruitful suggestions. MEFs were kindly supplied by Dr Yumiko Kawamura (University of Tokyo, Japan). C17.2 cells were kindly provided by Dr Yasunobu Uchijima (University of Tokyo, Japan). We thank Michal Bell, PhD, and J. Ludovic Croxford, PhD, from Edanz Group (https://en-author-services.edanz.com/ac) for editing a draft of the manuscript.

The peer review history is available online at https://journals.biologists.com/jcs/article-lookup/doi/10.1242/jcs.253559