Heat shock responses (HSRs) accompany the transcriptional induction of heat shock proteins and their escorting misfolded proteins for proteostasis (Alagar Boopathy et al., 2022). Heat shock causes dynamic changes in nuclear proteomes (Audas et al., 2016; Gallagher et al., 2014; Velichko et al., 2012), and emerging evidence identifies the nucleolus as a key locus for quality control of the misfolded proteins to sustain nuclear proteostasis upon heat shock (Frottin et al., 2019; Theodoridis et al., 2021). Metastable proteins phase-separate into the granular component of the nucleolus upon heat shock, thereby preventing their irreversible aggregations. Heat shock-induced HSP70 protein then translocates to the nucleolus (Pelham et al., 1984; Velazquez and Lindquist, 1984; Welch and Feramisco, 1984), where it disaggregates the misfolded proteins in an ATPase-dependent manner (Frottin et al., 2019; Mediani et al., 2019; Wentink et al., 2020). While damaged proteins are ubiquitinated for their proteasomal degradation, heat shock further induces specific ubiquitination of cellular factors involved in nucleocytoplasmic transport (NCT) and DNA damage repair for recovery of cellular physiology (Maghames et al., 2018; Maxwell et al., 2021). Failure of such compartmentalization or clearance could lead to the formation of irreversible aggregates (e.g., amyloid bodies), impairing DNA damage repair and cell growth (Gallardo et al., 2020; Mediani et al., 2019). Therefore, the nuclear HSR should elaborately operate for genome stability and cell fitness under heat stress conditions.

Intracellular communication between the nucleus and mitochondria contributes to cellular homeostasis, stress responses, and aging (Bennett et al., 2022; Desai et al., 2020; Quiros et al., 2016; Zhu et al., 2022). The nuclear genome encodes most of the mitochondrial genes, thereby controlling mitochondrial biogenesis and function in an anterograde manner. On the other hand, mitochondria cue the nucleus most likely through signaling molecules (e.g., AMP, NAD+, oxygen, calcium, reactive oxygen species [ROS], and metabolites) to activate the expression of stress response genes (Agarwal and Ganesh, 2020; Mottis et al., 2019; Rackham and Filipovska, 2022). The retrograde response is critical for mitochondrial homeostasis, whereas its misregulation causes mitochondrial dysfunction (Andréasson et al., 2019; Quiros et al., 2016).

In fact, modest mitochondrial stress rather increases cell survival by sensitizing stress response pathways, and this phenomenon is called mitohormesis (Yun and Finkel, 2014). Genetic perturbations of mitochondrial activity (e.g., electron transport chain [ETC]) enhance the binding of stress-induced transcription factor HSF1 and RNA polymerase II to HSR gene promoters, thereby amplifying the transcriptional HSR for stress resistance in worms (Labbadia et al., 2017). The partial loss of mitochondrial function also ameliorates the age-dependent decline in cytoplasmic proteostasis and lengthens healthspan in an HSF1-dependent manner (Desai et al., 2020; Labbadia et al., 2017). These observations implicate the mitonuclear interaction in the transcriptional control of early HSR. Nonetheless, it has not yet been addressed whether mitochondrial activity shapes nuclear proteostasis during HSR. Here we provide convincing evidence that suboptimal mitochondrial activity indeed facilitates the nuclear HSR via distinct pathways to sustain nuclear physiology and genome stability under heat stress conditions.

Cell culture, transfection, and heat shock

Mouse embryonic fibroblast NIH3T3 and human embryonic kidney (HEK) 293T cell lines were purchased from American Type Culture Collection (ATCC). NIH3T3 cells were cultured in DMEM (LM001-07; Welgene, Korea) with 10% calf serum (26170043; Gibco, USA). HEK293T cells were cultured in DMEM (SH30243; HyClone, USA) with 10% fetal bovine serum (12483020; Gibco). Plasmid DNA and siRNA were transfected using Lipofectamine 2000 (Invitrogen, USA) and Lipofectamine RNAiMAX (Invitrogen), respectively, according to the manufacturer’s instructions. At 48 h post-transfection, NIH3T3 cells were incubated at 43.5°C in a water bath for 1 h and then recovered at 37°C with 5% CO₂ in a humidified incubator. The chemicals used in cell culture experiments were CCCP (C2759; Sigma, USA), NAC (A7250; Sigma), Chloramphenicol (C0378; Sigma), ISRIB (SML0843; Sigma), VER-155008 (S7751; SelleckChem, USA), Puromycin (P8833; Sigma), and Antimycin A (A8674; Sigma).

Plasmids

Fluc-EGFP expression vectors were gifts from Franz-Ulrich Hartl (Addgene plasmid #90170) (Gupta et al., 2011). The Fluc-EGFP fusion cDNA was subcloned into a pHR lentiviral vector with an N-terminal nuclear localization signal (NLS). cDNA for shuttle-tdTomato reporter was a gift from Jeffrey Rothstein (Addgene plasmid #112579) (Zhang et al., 2015) and subcloned into pcDNA3.1 for transient expression as described previously (Lee et al., 2020). The full-length cDNAs for Npm1 (NM_008722.3) and Pml (NM_178087.4) genes were amplified from NIH3T3 cells and subcloned into a modified pcDNA6/V5-HisA (Invitrogen) vector with C-terminal EGFP fusion.

Recombinant lentiviral infection

Lentiviral packaging plasmids were co-transfected with pHR-NLS-Fluc-EGFP into HEK293T using polyethylenimine as described previously (Lee et al., 2020). Cell culture medium was collected at 48 h and 72 h after transfection and filtered by a 0.45 μm syringe filter after centrifugation at 2,000 rpm for 3 min. Recombinant lentiviruses were harvested by ultracentrifugation at 25,000 rpm at 4°C for 2 h. The lentiviral pellet was resuspended with phosphate-buffered saline (PBS) and used for NIH3T3 transduction with 10 μg/ml polybrene.

Immunofluorescence assay

Cells were washed twice with ice-cold PBS and then fixed using ice-cold 3.7% formaldehyde for 20 min at room temperature (RT). For HSP70 and ubiquitin staining, cells were fixed using ice-cold methanol for 5 min at –20°C. For puromycin labeling, cells were incubated in 2 μM puromycin for 5 min at 37°C before fixation. Fixed cells were washed three times with PBS for 10 min, permeabilized with 0.5% Triton-X in PBS for 15 min at RT, and then incubated with a blocking buffer (1% bovine serum albumin in PBS) for 1 h at RT. Immunostaining was performed by incubating cells with the indicated primary antibodies at 4°C overnight. Cells were washed three times with 0.025% Triton-X in PBS for 10 min and then incubated with Alexa Fluor 488-, 568-, 594-conjugated secondary antibodies (1:2,500 [Invitrogen] or 1:600 [Jackson ImmunoResearch Laboratories, USA]) at 4°C overnight. Hoechst 33342 was incubated for 10 min at RT and washed three times with 0.025% Triton-X in PBS for 10 min. Coverslips were mounted with VECTASHIELD mounting medium (H-1900-10; Vector Laboratories, USA). The primary antibodies used in immunostaining were mouse anti-HSP70 (sc-66048, 1:40; Santa Cruz Biotechnology, USA), rabbit anti-GFP (AE011, 1:200; ABclonal, China), mouse anti-Ubiquitin (sc-8017, 1:250; Santa Cruz Biotechnology), mouse anti-γH2A.X (sc-517348, 1:200; Santa Cruz Biotechnology), mouse anti-Puromycin (PMY-2A4, 1:25; DSHB, USA), mouse anti-RAN (SC-271376, 1:500; Santa Cruz Biotechnology), rabbit anti-FBL (2639, 1:500; Cell Signaling Technology, USA), mouse anti-Exportin T (sc-514591, 1:250; Santa Cruz Biotechnology), mouse anti-RanGAP1 (sc-28322, 1:250; Santa Cruz Biotechnology), mouse anti-Karyopherin β1 (sc-137016, 1:250; Santa Cruz Biotechnology), mouse anti-Karyopherin β2 (sc-365179, 1:250; Santa Cruz Biotechnology), mouse anti-RCC1 (sc-55559, 1:250; Santa Cruz Biotechnology), mouse anti-NXF1 (sc-32319, 1:250; Santa Cruz Biotechnology), mouse anti-ATP5A (ab14748, 1:1,000; Abcam, UK), rabbit anti-Ataxin2 (21776-1-AP, 1:2,000; Proteintech), and mouse anti-G3BP1 (sc-365338, 1:500; Santa Cruz Biotechnology).

Mitochondrial ROS measurement

Cells were incubated in 3 μM MitoSOX (M36008; Invitrogen) for 10 min at 37°C, washed with pre-warmed PBS, and then fixed using pre-warmed 3.7% formaldehyde for 20 min at RT. Hoechst 33342 was incubated for 10 min at RT and washed three times with PBS for 5 min. Coverslips were mounted with VECTASHIELD mounting medium.

Confocal microscopy and quantitative image analysis

Confocal images were acquired using LSM 780 confocal laser scanning microscope (Zeiss, Germany) with 405/488/561/633 nm lasers, processed with ZEN software (Zeiss), and analyzed by ImageJ/Fiji software (NIH). Live imaging was performed using confocal dish (200350; SPL Life Sciences, Korea) in a live cell chamber maintaining 5% CO₂ at 37°C. Mitochondrial membrane potential was visualized by Mitoview633 (70055; Biotium, USA) fluorescent dye staining. The nuclear to cytoplasmic ratio of S-tdT, puromycylation, and RAN expression was calculated in individual cells by measuring their intensities of nuclear and cytoplasmic fluorescence as described previously (Lee et al., 2020). Mitochondrial morphology was analyzed using the Mitochondrial Analyzer plugin in ImageJ/Fiji (https://github.com/AhsenChaudhry/Mitochondria-Analyzer) (Chaudhry et al., 2020). Nucleolar HSP70 was detected using Stardist plugin in Fiji (https://github.com/stardist/stardist/) (Schmidt et al., 2018).

Quantitative RNA analysis

Total RNA was isolated from NIH3T3 cells using RNA extraction kit (9767A; Takara, Japan) and used for cDNA synthesis by M-MLV reverse transcriptase (M1705; Promega, USA) according to the manufacturer’s instructions. Real-time PCR was performed using SYBR Green-based Prime Q-Mastermix (Q-9202; GeNet Bio, Korea) on QauntStudio 1 Real-Time PCR Instrument (Applied Biosystems, USA).

qPCR primer sequences (5'-3')

Hsf1 F: GCCCCTCTTCCTTTCTGCAT, R: TCATGTCGGGCATGGTCAC

Gapdh F: GCCATCAACGACCCCTTCATT, R: GCTCCTGGAAGATGGTGATGG

Mrpl2 F: TCACATAGGCAGGATGGCAG, R: CATGTCCCTGCAGCTCGGAT

Hsp70 F: ATGGACAAGGCGCAGATCC, R: CTCCGACTTGTCCCCCAT

Bag3 F: TTCGAGCCGCTTCTCCATTC, R: TTCGGGTTGGGTAACAGGTG

Fos F: GGGGACAGCCTTTCCTACTA, R: CTGTCACCGTGGGGATAAAG

Jun F: ACGACCTTCTACGACGATGC, R: CCAGGTTCAAGGTCATGCTC

Hspd1 F: GCAGAGTTCCTCAGAAGTTGG, R: GCATCCAGTAAGGCAGTTCTC

Immunoblotting

Cells were briefly washed twice, harvested in ice-cold PBS, and then collected by centrifugation at 2,500 × g at 4°C for 5 min. Cells were lysed in RIPA buffer (50 mM Tris pH 8.0, 150 mM NaCl, 5 mM EDTA, 10% NP-40, 0.5% sodium deoxycholate, and 1 mM phenylmethylsulfonyl fluoride) and incubated for 10 min on ice. After adding SDS loading dye, cell lysates were resuspended by a 26 gauge needle with a syringe. Protein samples were resolved by SDS-PAGE and transferred to the NC membrane (10600002; GE Healthcare, USA). The primary antibodies used in immunoblotting were rabbit anti-MRPL2 (HPA064814, 1:1,000; Atlas Antibodies, Sweden), mouse anti-HSP70 (sc-66048, 1:500; Santa Cruz Biotechnology), mouse anti-β tubulin (66240-1-Ig, 1:5,000; Proteintech, China), mouse anti-eIF2a (sc-133132, 1:500; Santa Cruz Biotechnology), and rabbit anti-phospho eIF2a (ab32157, 1:10,000; Abcam). Horseradish peroxidase-conjugated secondary antibodies (Jackson Immunoresearch Laboratories) were used for detecting primary antibodies and reacted with ECL solution (1705061; Bio-Rad, USA). Luminescence signals were detected by iBright FL1500 (Invitrogen).

Cellular ATP measurement

The relative amount of ATP in whole-cell lysates was quantified using luciferase-based ADP/ATP Ratio assay kit (MAK135; Sigma) according to the manufacturer’s instruction. Luminescence signals were detected using a GloMax luminometer (Promega).

Cell proliferation assay

Transfected cells were washed once with PBS and then incubated with a staining solution (0.25% w/v crystal violet [C3886; Sigma] in 20% methanol) for 10 min at RT. The stained cells were washed three times with PBS and then lysed in a solution (0.1 M sodium citrate with pH 4.2 in 25% ethanol) for 30 min with gentle shaking. The extracts were diluted to 20% using the lysis solution and the absorbance was measured at 590 nm using plate reader infinite M200 (Tecan, Switzerland).

Statistics

All statistics were analyzed using Prism 6 (GraphPad Software, USA) or R (ver. 3.6.1) with ARTool library (Kay and Wobbrock, 2016). Appropriate statistical analyses were chosen based on Shapiro–Wilk test for normality (P < 0.05) or F-test and Brown-Forsythe test for equal variance (P < 0.05). The statistical significance of the dataset, which passed both normality and equal variance, was determined by Student’s t-test, ordinary 1-way ANOVA with Dunnett post hoc test, or ordinary 2-way ANOVA with Tukey post hoc test. If the dataset didn’t pass normality and equal variance, Mann–Whitney U test or aligned ranks transformation ANOVA with Wilcoxon rank-sum test were conducted. The dataset which passed only normality or equal variance was statistically analyzed by Welch’s t-test or Kruskal–Wallis 1-way ANOVA with Dunn’s post hoc test as described previously (Kim et al., 2020).

siRNA sequences

siMrpl2: ATGGTTAACGTTGGATACTCG

siMrpl10: GCGGAAACACAAGATCTTCAT

siMrpl37 (sc-106246; Santa Cruz Biotechnology)

siAfg3l2: CCTGCCTCCGTACGCTCTATCAATA

siOpa1: CCGACACAAAGGAAACTATTT

siHsf1: CCCAAGTACTTCAAGCACA

RNA interference-mediated depletion of mitochondrial ribosomal proteins (MRPs) enhances nucleolar HSP70 accumulation during HSR

The expression of overall MRPs decreases with age in mice, whereas MRP depletion confers the resistance to heat stress and extends lifespan in worms (Houtkooper et al., 2013; Labbadia et al., 2017). We thus examined whether mammalian MRP depletion would also benefit cellular physiology under heat stress conditions. More specifically, we assessed the nuclear HSR in control and MRP-depleted cells. The nucleolar accumulation of HSP70 protein was readily detectable during the recovery period after heat shock (Figs. 1A-1C). The siRNA-mediated depletion of large subunit MRPs (i.e., MRPL2, MRPL10, and MRPL37) elevated the nucleolar HSP70 levels, resulting in high percentages of the nucleolar HSP70 granule-positive cells at 2 h post-heat shock (HS) (Figs. 1D and 1E).

Among large subunit MRPs, MRPL2 constitutes the intersubunit bridges with the small ribosomal subunit protein MRPS6 and the two small ribosomal subunit rRNAs (h23 and h24) (Amunts et al., 2015), playing an important role in mitochondrial ribosome dynamics during translation (Yusupova and Yusupov, 2017). MRPL2 depletion did not significantly affect cellular ATP levels under our heat shock conditions (Supplementary Fig. S1A), possibly due to a metabolic compensation by cytoplasmic glycolysis (Wang et al., 2012). Nonetheless, we found that MRPL2 depletion substantially slowed down cell proliferation in long-term cell cultures (i.e., 72 h after siRNA transfection) (Supplementary Fig. S1B). These results support that MRPL2 depletion indeed compromises mitochondrial activity for growth. Accordingly, we selected MRPL2 as a representative target for manipulating mitochondrial translation and activity in our subsequent analyses.

Downregulation of ETC activity has been shown to increase HSF1 binding to its downstream gene promoters, including hsp-70 (Labbadia et al., 2017). HSP70 then interacts with HSF1 to inhibit its DNA-binding activity as a negative feedback mechanism for HSF1-dependent transcription (Masser et al., 2019). Nonetheless, MRPL2 depletion had no remarkable effects on Hsp70 expression levels in whole-cell lysates, modestly elevating Hsp70 mRNA levels only immediately after heat shock (Figs. 1F and 1G, Supplementary Figs. S2A and S2B). MRPL2-depleted cells also displayed the transcriptional induction of other HSF1 target genes comparably to control cells (Supplementary Figs. S2C-S2F). MRPL2 depletion significantly elevated the nucleolar HSP70 levels even in HSF1-depleted cells (Supplementary Figs. S3A-S3C) although HSF1 depletion dampened HSP70 protein levels in both control and MRPL2-depleted cells at 2 h post-HS (Supplementary Fig. S3D) (Kovács et al., 2019). Finally, the HSP70 ATPase inhibitor VER-155008 has been shown to block the disassembly of the nucleolar HSP70 granules during HSR (Frottin et al., 2019; Mediani et al., 2019), while the MRPL2 depletion phenotype was still evident in the presence of the HSP70 inhibitor (Fig. 1H). These data together suggest that MRP depletion may facilitate the HSP70 translocation into the nucleolus, and the nuclear HSR phenotype unlikely implicates the transcriptional activity of HSF1 under heat shock conditions.

MRP depletion facilitates misfolded protein clearance and ubiquitin recycling from the nucleolus during heat shock recovery

Heat stress disrupts nuclear homeostasis, leading to protein misfolding and aggregation, elevated DNA damage, and impaired NCT (Frottin et al., 2019; Mediani et al., 2019; Ogawa and Imamoto, 2018). Considering the role of HSP70 in nucleolar protein quality control under cellular stress (Frottin et al., 2019), we next asked whether MRP-depleted cells would better sustain nuclear proteostasis during heat shock recovery. Accordingly, we performed a series of imaging experiments in control and MRPL2-depleted cells to compare the kinetics of their nuclear HSRs.

Upon heat shock, a structurally unstable luciferase with N-terminal NLS and C-terminal GFP fusion (NLS-Fluc-EGFP) localizes to the nucleolus, likely due to misfolding (Frottin et al., 2019; Gupta et al., 2011; Nollen et al., 2001). We found that the nucleolar translocation of NLS-Fluc-EGFP was comparable between control and MRPL2-depleted cells, but it disappeared more rapidly in MRPL2-depleted cells during the recovery period (Figs. 2A and 2B, Supplementary Fig. S4A). Heat shock-induced protein ubiquitination has been shown to play a crucial role in targeting misfolded proteins for degradation or recovering from the heat-induced shutdown of cellular activities (e.g., general translation, NCT) (Maxwell et al., 2021). We indeed observed that ubiquitin signals gradually increased in the nucleolus during the heat shock recovery (Figs. 2C and 2D, Supplementary Fig. S4B). MRPL2-depleted cells, however, quickly normalized the nucleolar accumulation of free ubiquitin or ubiquitinated proteins. These phenotypes were consistent with the high levels of nucleolar HSP70 proteins in MRPL2-depleted cells that would facilitate the clearance of the misfolded proteins from the nucleolus during the heat shock recovery.

MRP depletion sustains genome stability and NCT during heat shock recovery

Heat shock disrupts DNA damage repair of topoisomerase-induced DNA single/double-strand breaks during the cell cycle (Kantidze et al., 2016), and the evidence for the implication of ubiquitin recycling in DNA damage repair is abundant (Maxwell et al., 2021; Mediani et al., 2019). However, free ubiquitin may become rate-limiting for DNA damage repair when cellular stress such as heat shock depletes it by ubiquitination of misfolded proteins and stress-relevant factors (Maxwell et al., 2021; Mediani et al., 2019). We thus hypothesized that efficient ubiquitin recycling in MRPL2-depleted cells would buffer heat shock-induced DNA damage. To assess DNA damage repair during the heat shock recovery, we quantified γH2A.X levels, a molecular marker for DNA double-strand breaks. MRPL2 depletion indeed reduced the number of heat shock-induced γH2A.X foci and facilitated their disappearance during the recovery period (Figs. 2E and 2F).

Since heat shock also induces the ubiquitination of NCT factors (Maxwell et al., 2021), we further examined whether MRPL2 depletion modulates NCT during heat shock recovery. We employed a transgenic tdTomato reporter harboring both nuclear localization and nuclear export signals (Shuttle-tdTomato/S-tdT) (D'Angelo et al., 2012; Lee et al., 2020). Under baseline conditions, the S-tdT reporter protein localizes dominantly to the nucleus, whereas NCT imbalance induces its cytoplasmic translocation. We found that heat shock reduced the ratio of nuclear to cytoplasmic S-tdT expression levels in control cells, and NCT activity was gradually restored during the recovery period (Figs. 2G and 2H). In contrast, MRPL2 depletion substantially suppressed NCT disruption upon heat shock, facilitating the nuclear import of newly synthesized proteins (i.e., puromycin-labeled proteins) during the heat shock recovery (Figs. 2I and 2J).

Various cellular stresses, including heat shock, can induce the formation of cytoplasmic stress granules (SGs) (Maxwell et al., 2021; Zhang et al., 2018). SG has been proposed to sequester NCT factors, thereby temporally compromising NCT (Zhang et al., 2018). However, MRPL2 depletion did not remarkably affect the dynamics of SG assembly and disassembly during HSR (Supplementary Figs. S5A-S5H). We further found that MRPL2-depletion effects on the nucleolar HSP70 granules were still detectable when p-eIF2alpha-dependent SG assembly was inhibited by ISRIB treatment (Zhang et al., 2018) (Supplementary Figs. S5I-S5L). These results exclude the implication of SG and possibly p-eIF2alpha-triggered integrated stress response (Koncha et al., 2021) in MRPL2-dependent assembly of the HSP70 granules upon heat shock.

While there was no strong correlation between MRPL2 depletion and RAN protein levels during HSR, MRPL2-depleted cells displayed a relatively high ratio of nuclear to cytoplasmic RAN protein (Figs. 2K-2M), explaining robust NCT in heat shock stress (Stewart, 2007). We further observed that a sub-population of nuclear RAN was localized in the nucleolus upon heat shock (Supplementary Fig. S6A), and it was relatively specific among other NCT-relevant factors (Supplementary Fig. S6B). We reason that nucleolar RAN may be one of the HSP70 clients under heat shock conditions (Ryu et al., 2020), possibly explaining MRPL2-depletion effects on NCT. These findings together suggest that suboptimal mitochondrial activity by MRPL depletion may better sustain nuclear proteostasis and genome stability upon heat shock.

Pharmacological and transgenic perturbations of mitochondrial function tune nuclear HSR

How does MRP depletion modulate the nuclear HSR? Treatment of a mitochondrial translation inhibitor (i.e., chloramphenicol) negligibly affected the nucleolar assembly of HSP70 granules in control and MRPL2-depleted cells during HSR (Supplementary Fig. S7). A simple loss of mitochondrial translation may thus not be responsible for the HSP70 phenotype in MRPL2-depleted cells. We further manipulated mitochondrial activity using the mitochondrial proton gradient uncoupler carbonyl cyanide 3-chlorophenylhydrazone (CCCP) and the OXPHOS inhibitor antimycin A and compared their effects on the nucleolar HSP70 translocation during HSR. CCCP treatment suppressed nucleolar HSP70 assembly and silenced MRPL2-depletion effects (Figs. 3A and 3B, Supplementary Fig. S8). By contrast, antimycin A treatment alone was sufficient to increase nucleolar HSP70 levels and masked the MRPL2 phenotype (Supplementary Figs. S9A and S9B). Moreover, we detected additive effects of CCCP and antimycin A on the nucleolar HSP70 levels under our heat shock conditions (Supplementary Figs. S9C and S9D). Although both CCCP and antimycin A block mitochondrial ATP synthesis, they display opposing effects on mitochondrial oxygen consumption (Divakaruni and Jastroch, 2022). Given negligible effects of MRPL2 depletion on cellular ATP levels (Supplementary Fig. S1A), we speculate that low oxygen consumption by antimycin A may mimic MRPL2 depletion to promote nucleolar HSP70 accumulation during HSR, whereas maximal oxygen consumption by CCCP likely silences the MRPL2 effects. Indeed, CCCP suppressed MRPL2 phenotypes in both nucleolar ubiquitin granules and γH2A.X foci while generally reducing nucleocytoplasmic RAN gradient during the heat shock recovery (Figs. 3C-3I). These results indicate that MRPL2-depletion phenotypes on the nuclear HSR require the functionality of mitochondrial oxidative phosphorylation, and the transgenic effects could be modulated by transient manipulations of the mitochondrial activity (i.e., oxygen consumption).

To validate whether general mitochondrial function tunes the nuclear HSR, we further manipulated mitochondrial activity and assessed nucleolar HSP70 assembly during the heat shock recovery. Optic atrophy 1 (OPA1) and ATPase family gene 3-like 2 (AFG3L2) are mitochondrial stress-sensitive proteins implicated in mitochondrial fusion, morphology, and function (Richter et al., 2015). Impairment of mitochondrial protein quality control leads to the cleavage of the inner mitochondrial membrane protein OPA1 by mitochondrial proteases, including AFG3L2 (Richter et al., 2019; Tulli et al., 2019). We found that OPA1 depletion enhanced nucleolar HSP70 granule assembly during the heat shock recovery while masking the MRPL2-depletion effects (Figs. 4A and 4B). Consistent with previous findings (Cipolat et al., 2004), OPA1 depletion provoked mitochondrial fragmentation (Figs. 4C-4E). MRPL2 depletion alone did not cause any significant change in mitochondrial morphology, but it suppressed the OPA1-depletion phenotype during the heat shock recovery (Figs. 4C-4E). A simple explanation could be that the equilibrium between mitochondrial fusion and fission somehow becomes more static in MRPL2-depleted cells, thereby blunting OPA1 effects on mitochondrial fusion. The lack of the phenotypic correlation between nucleolar HSP70 and mitochondrial morphology suggests that OPA1-dependent mitochondrial fusion is unlikely implicated in MRPL2-dependent nuclear HSR. In contrast, MRPL2 and OPA1 depletion additively suppressed the heat shock-induced generation of mitochondrial ROS (Figs. 4F-4H), possibly suggesting mitochondrial ROS as a key molecule mediating MRPL2-depletion effects on the nuclear HSR.

Heat shock-induced mitochondrial ROS is specifically implicated in DNA damage response

To address the hypothesis mentioned above, we examined whether treatment of the antioxidant ROS scavenger N-acetylcysteine (NAC) could suppress MRPL2 phenotypes in the nuclear HSR. As expected, NAC reduced mitochondrial ROS levels in control and MRPL2-depleted cells, substantially masking MRPL2-depletion effects (Figs. 5A and 5B). Nonetheless, MRPL2-depletion effects on nucleolar HSP70 assembly were still detectable in NAC-treated cells (Figs. 5C and 5D). NAC treatment and MRPL2 depletion also showed no interactive effects on nucleolar ubiquitin granule formation and nucleocytoplasmic RAN gradient (Figs. 5E-5H). However, NAC treatment reduced the number of γH2A.X foci in control cells during HSR (Figs. 5I-5K), and non-additive effects of NAC treatment and MRPL2 depletion were evident in the DNA damage response. These results suggest a model that nuclear DNA damage response is triggered by heat shock-induced mitochondrial ROS likely via DNA oxidation (Agarwal and Ganesh, 2020). In contrast, other nuclear HSRs (i.e., nucleolar HSP70 assembly, nucleolar ubiquitin granule, nucleocytoplasmic RAN gradient) may depend on retrograde signals other than mitochondrial ROS (Fig. 6A).

Mitonuclear interaction is an intracellular communication essential for cell physiology and stress response (Eckl et al., 2021; Melber and Haynes, 2018; Suhm et al., 2018). The odd ratio of mitochondrial-encoded to nuclear-encoded gene expression (i.e., mitonuclear protein imbalance) has been proposed as a mechanism for mitochondrial perturbations to amplify nuclear transcriptional response and sustain cytoplasmic proteostasis (Houtkooper et al., 2013). Our study discovers an independent mechanism of the mitochondria-to-nuclear communication that acts on nuclear homeostasis under heat shock conditions. We define distinct retrograde pathways (i.e., ROS-independent vs ROS-dependent) from suboptimal mitochondrial activity that support nuclear proteostasis and genome stability. These findings hint at how the endosymbiotic correlation between the mitochondria and nucleus has been optimized over evolutionary challenges by multiple pathways of the mitonuclear interaction (Yun and Finkel, 2014).

Time-point assessments of cellular HSR revealed their unique kinetics upon heat shock (Fig. 6B). Cytoplasmic SG assembly was immediately induced upon heat shock while rapidly reversed during the recovery period. It contrasted with protein synthesis or NCT that was silenced by heat shock and barely recovered at 2 h post-HS. In the nucleus, heat shock promoted the nucleolar accumulation of misfolded proteins. The nucleolar condensates began to disappear as the nucleolar levels of ubiquitin and HSP70 gradually increased during the post-HS period. These HSR dynamics would minimize any unnecessary synthesis of new proteins and their nuclear entry under a proteotoxic environment while activating the nucleolar quality control pathway for compartmentalization and clearance of damaged proteins during HSR. Interestingly, we noticed that the disassembly of nucleolar ubiquitin preceded that of nucleolar HSP70, particularly in MRPL2-depleted cells. The phase difference may indicate that a ubiquitin-dependent mechanism primarily degrades damaged proteins in the nucleolus, and the HSP70 chaperones subsequently refold the remaining proteins.

Nuclear HSP70 protein binds the HSF1 complex and interferes with its DNA-binding activity (Masser et al., 2019). Translocation of the nuclear HSP70 to an insoluble fraction (e.g., sequestration by misfolded proteins) is thought to relieve the inhibitory interaction between HSP70 and HSF1 proteins, thereby derepressing HSF1-dependent transcription. Although MRPL2 depletion facilitated nucleolar HSP70 assembly during the recovery period after heat shock, we did not observe any significant effects of MRPL2 depletion on HSF1-dependent induction of HSR genes, including a mitochondrial unfolded protein response (UPR^mt) marker gene Hspd1 (Hsp60) (Pellegrino et al., 2013). In fact, genetic studies in C. elegans showed that downregulation of mitochondrial activity (e.g., depletion of ETC-relevant factors or MRP) triggers the UPR^mt owing to the mitonuclear protein imbalance. The UPR^mt then promotes HSF1-dependent transcription of HSR genes that sustain cytoplasmic proteostasis and increase fitness in the context of physiological stress (e.g., heat shock, aging) (Houtkooper et al., 2013; Labbadia et al., 2017) although lack of the correlation between UPR^mt and longevity has also been documented (Bennett et al., 2014). The mitochondrial translation inhibitor chloramphenicol has been shown to induce UPR^mt (Houtkooper et al., 2013). However, it did not alter nucleolar HSP70 assembly in either control or MRPL2-depleted cells under our heat shock conditions. These lines of evidence exclude the implication of UPR^mt and HSF1-dependent transcription in MRPL2-sensitive nuclear homeostasis, suggesting a more direct role for MRPL2-dependent mitochondrial activity in the nucleolar protein quality control. Considering the relevance of low mitochondrial oxygen consumption to MRPL2 effects on the nuclear HSR, we speculate that MRPL2-deficient mitochondrial ribosomes may generate aberrant translation products on the mitochondrial membrane. The partial loss of mitochondrial ribosome function may not be sufficient to trigger UPR^mt but down-regulate the overall OXPHOS flux and heat shock-induced ROS generation.

Diverse mitochondrial retrograde signals have evolved to tune distinct aspects of host cell physiology (Andréasson et al., 2019; Bohovych and Khalimonchuk, 2016; Chakrabarty and Chandel, 2022; Mottis et al., 2019), likely mediating the adaptive advantage of suboptimal mitochondrial activities (Lee et al., 2022). These signaling molecules include ATP/ADP/AMP, NADH/NAD+, oxygen, ROS, calcium, and metabolic intermediates (e.g., acetyl-CoA, alpha-ketoglutarate) that could modulate enzyme activities (e.g., AMP-activated protein kinase, sirtuins, dioxygenases), epigenetic/chromatin remodeling, and posttranslational modifications (e.g., thiol oxidation) in non-mitochondrial compartments (Andréasson et al., 2019; Martínez-Reyes and Chandel, 2020; Mottis et al., 2019). We propose oxygen consumption as a key event for the retrograde signaling during HSR, yet our current study is limited to specifically define a ROS-independent signaling molecule for MRPL2-relevant nuclear proteostasis under heat shock conditions. Future studies should better elucidate the molecular nature of these retrograde signals and their mechanism of action. Considering that mitochondrial dysfunction is frequently observed in neurodegenerative diseases with nuclear proteotoxic aggregates (e.g., Huntington’s disease, amyotrophic lateral sclerosis) (Panchal and Tiwari, 2019; Riguet et al., 2021), it will also be interesting to see if titration of the mitochondrial activity can increase the capacity of nuclear proteostasis to ameliorate relevant nuclear pathologies.

We thank Franz-Ulrich Hartl, Jeffrey Rothstein, Addgene, and Developmental Studies Hybridoma Bank for reagents. This work was supported by grants from the Suh Kyungbae Foundation (SUHF-17020101); from the National Research Foundation funded by the Ministry of Science and Information & Communication Technology (MSIT), Republic of Korea (NRF-2021R1A2C3011706; NRF-2021M3A9G8022960; NRF- 2018R1A5A1024261).

Conceptualization, D.P., C.L., and K.T.M.; methodology, D.P., Y.Y., J.K., J.L., J.P., K.H., J.K.S., and C.L.; validation, D.P.; formal analysis, D.P., J.K., and C.L.; investigation, D.P.; writing – original draft, D.P.; writing – review & editing, D.P. and C.L.; visualization, D.P. and C.L.; supervision, C.L.; funding acquisition, C.L. All authors read and approved the final manuscript.

The authors have no potential conflicts of interest to disclose.