Endoplasmic reticulum-unfolded protein response pathway modulates the cellular response to mitochondrial proteotoxic stress

Abstract

Mitochondria and endoplasmic reticulum (ER) remain closely tethered by contact sites to maintain unhindered biosynthetic, metabolic, and signalling functions. Apart from its constituent proteins, contact sites localize ER-unfolded protein response (UPR) sensors like Ire1 and PERK, indicating the importance of ER-mitochondria communication during stress. In the mitochondrial sub-compartment-specific proteotoxic model of yeast, Saccharomyces cerevisiae, we show that an intact ER-UPR pathway is important in stress tolerance of mitochondrial intermembrane space (IMS) proteotoxic stress, while disrupting the pathway is beneficial during matrix stress. Deletion of IRE1 and HAC1 leads to accumulation of misfolding-prone proteins in mitochondrial IMS indicating the importance of intact ER-UPR pathway in enduring mitochondrial IMS proteotoxic stresses. Although localized proteotoxic stress within mitochondrial IMS does not induce ER-UPR, its artificial activation helps cells to better withstand the IMS proteotoxicity. Furthermore, overexpression of individual components of ER-mitochondria contact sites is found to be beneficial for general mitochondrial proteotoxic stress, in an Ire1-Hac1-independent manner.

Supplementary Information

The online version contains supplementary material available at 10.1007/s12192-022-01264-2.

Introduction

Mitochondria, the double membrane–bound cellular powerhouse, is obligatorily dependent on the nucleus and neighboring sub-cellular compartments for its biogenesis and functions. Among its neighboring compartments, mitochondria are almost inseparable from the endoplasmic reticulum (ER) as both the organelles remain connected by specialized inter-organellar tethering structures known as ER-mitochondria contacts sites. Specialized microdomains for the exchange of Ca²⁺ ions between the two organelles within the contacts sites are also known as mitochondria-associated membranes or MAMs (Rowland and Voeltz 2012; Giorgi et al. 2015; Paillusson et al. 2016). ER-mitochondria contact sites have been shown as critical inter-organellar tethering structures essential for several metabolic and signalling functions and organelle dynamics. In yeast, Saccharomyces cerevisiae, the ER-mitochondria contact sites are also known as ER-mitochondria encounter structure or ERMES (Kornmann et al. 2009; Kornmann and Walter 2010). ERMES is constituted of outer mitochondrial membrane proteins like Mdm10, Mdm34 along with integral ER membrane protein Mmm1, and cytosolic protein Mdm12. This proteinaceous inter-organellar structure majorly exchanges Ca²⁺ ions and phospholipids between the two organelles. Apart from constitutive components of ER-mitochondria contact sites, sensors of ER-UPR like Ire1 (Mori et al. 2013; Carreras-Sureda et al. 2019) and PERK (Verfaillie et al. 2012) were shown to localize to MAMs during various stresses indicating the importance of ER-mitochondria communication during cellular stresses. Recently, Ire1α was found to localize to MAMs and directly impact the mitochondrial bioenergetics by controlling Ca²⁺ uptake by the mitochondrial matrix in mammalian cells (Carreras-Sureda et al. 2019). Previous studies have shown that ER stress by small-molecule stressors differentially modulates the expression of many mitochondrial proteins impacting mitochondrial biogenesis and functions (Jonikas et al. 2009; Maity et al. 2016). In yeast, ER integral membrane protein Ire1 acts as the sole sensor for ER-unfolded protein response (UPR) which initiates the UPR signalling upon activation by oligomerization and autophosphorylation. Activated Ire1 causes non-canonical splicing of the HAC1 mRNA by its RNAse domain (Sidrauski and Walter 1997; Walter and Ron 2011), and the protein product of spliced HAC1 mRNA, Hac1p, acts as the transcription factor which upregulates expression of UPR responsive genes (Walter and Ron 2011; Travers et al. 2000). Previously, a study by Maity et al. have shown the importance of IRE1 during ER stress by DTT (dithiothreitol, a reducing agent which imparts ER stress and activates ER-UPR) on mitochondrial morphology, in yeast Saccharomyces cerevisiae (Maity et al. 2014). The authors showed that the presence of IRE1 is critical during ER stress to maintain mitochondrial dynamics, in absence of which mitochondria is severely fragmented during DTT induced ER stress (Maity et al. 2014).

In the current study, after generating mitochondrial sub-compartment (intermembrane space or IMS and mitochondrial matrix or MM)-specific proteotoxic stress model in yeast Saccharomyces cerevisiae, we have explored whether ER-UPR has any modulatory role in the cellular response to mitochondrial proteotoxic stresses in its sub-compartments, especially in IMS. Mitochondrial IMS remains in close connection with ER through the ER-mitochondria contact sites and share a similar protein folding environment like ER (Neupert and Herrmann 2007). Thus, proteotoxic stress in IMS may perturb closely connected ER, and ER-UPR signalling is expected to modulate the stress response in IMS. We show that indeed, the components of the ER-UPR pathway of yeast, the sensory protein Ire1 and its downstream protein Hac1 are important during mitochondrial IMS proteotoxic stress. Notably, deletion of IRE1 leads to more accumulation of misfolded proteins in mitochondrial IMS indicating the importance of IRE1 in clearing the misfolded proteins from mitochondrial IMS. Moreover, artificial activation of ER-UPR by small-molecule ER stressors impart better cellular resilience to IMS proteotoxic stress. Surprisingly, ER-UPR has a contrasting effect during proteotoxic stress originating from the mitochondrial matrix. During proteotoxicity in the mitochondrial matrix, disrupting the ER-UPR pathway becomes beneficial for yeast cells, while inducing ER-UPR by ER stressors become deleterious by a yet unknown mechanism. Interestingly, we show that overexpression of components of ER-mitochondria contact sites (ERMES) is beneficial for withstanding mitochondrial proteotoxic stress irrespective of its location within mitochondria in an IRE1-HAC1 signalling–independent manner. Altogether we report an interesting role of ER-UPR pathway and ER-mitochondria contact sites on mitochondrial proteotoxic stress phenotypes.

Methods

Strains used

S. cerevisiae strain YMJ003 (MATαhis3Δ1 leu2Δ0 met15Δ0 ura3Δ0 LYS + Δcan1::STE2pr-spHIS5 Δlyp1::STE3pr-LEU2 cyh2 Δura3::UPRE-GFP-TEF2pr-RFP-MET15-URA3. BY4741 (MATa, his3Δ1; leu2Δ0; met15Δ0; ura3Δ0) (Jonikas et al. 2009) was used as wild-type yeast strain, and all the strains expressing stressor proteins or control proteins used in this study were made in the background of this strain except for IRE1 and HAC1 deletion strains which are taken from yeast deletion library. Yeast transformation and plasmid preparations were performed by standard methods. Mitochondrial sub-compartment-specific proteotoxic model yeast strains were made by inserting the DNA cassettes using homologous recombination.

Construction of mitochondrial sub-compartment-specific proteotoxic models in yeast

For the construction of the DNA cassettes to generate mitochondrial sub-compartment-specific proteotoxic models in yeast, respective gene sequences of model stressor proteins and control protein were amplified and cloned in ApaI and AvrII site of pYMN23 plasmid. Mitochondrial targeting signal sequence of yeast Cytochromeb2 (cyb2) was amplified using Cyb2-DHFR plasmid as template, and its 19 amino acid–truncated version lacking the membrane sorting signal (Cyb2Δ19) was cloned upstream of the ORF to target the protein to mitochondrial IMS and matrix, respectively. Galactose inducible Gal1 promoter was cloned to express the proteins with galactose as the inducer. Cyc1 terminator sequence was used for transcription termination. The whole DNA cassette including antibiotic selection marker (cloNAT) was amplified using Kapa HiFi DNA polymerase and transformed in the yeast YMJ003 strain. Yeast strains were selected on a nourseothricin antibiotic (cloNAT) containing selection in YEPD (1% yeast extract, 2% peptone, and 2% dextrose) medium. The integration of the DNA cassette at the correct locus was confirmed by genotyping (plating on URA ± plates) and PCR-based methods (using locus-specific upstream and gene-specific internal primers). For ER targeting, ER-specific targeting signals of Kar2 and ER-retention signals were added to the gene sequences of the stressor proteins and control proteins.

Growth conditions and drop dilution assay

All the strains are grown in poorly fermentable media YPR (1% yeast extract, 2% peptone, and 2% raffinose) overnight. The overnight-grown culture was re-inoculated in the same medium at an initial OD₆₀₀ of 0.1. For drop dilution assay, different yeast strains were grown in YPR medium until mid-log phase (OD₆₀₀ 0.4–0.6) and serially diluted and spotted on different plates containing YPR (1% yeast extract, 2% peptone, and 2% raffinose) or on YPR-gal (YPR with 1% galactose as inducer) plates. For specific treatment with ER stressors, 5 mM DTT or 0.5 µg/ml of tunicamycin was added to the YPR-gal plates. For plasmid selection (Ura), yeast strains were grown in synthetic media with 2% raffinose without the uracil (SR-Ura) and SR-Ura with 1% galactose (SR-Ura + gal).

Crossing of stressor protein-expressing yeast strains with deletion strains from YKO library

Yeast strains expressing stressor proteins with targeting signals for mitochondrial IMS and matrix with genotype (MATα his3Δ1 leu2Δ0 met15Δ0 ura3Δ0 LYS + Δcan1::STE2pr-spHIS5 Δlyp1::STE3pr-LEU2 cyh2 Δura3::UPRE-GFP-TEF2pr-RFP-MET15-URA3::signal sequence stressor protein-NAT derived from S288C) is referred to as query strain here onwards. The query strains were crossed with knockout strain in BY4741 background (MATa, his3Δ1; leu2Δ0; met15Δ0; ura3Δ0;gene::KANMX) from Yeast KO Library (YKO, Mat A complete set, ThermoFisher Scientific cat no. 95401.H2) as described previously (Baryshnikova et al. 2010). In brief, query strain and single KO strain (ire1Δ, hac1Δ) were cultured in YPD media until saturation. Five microliters of the saturated culture of each strain were inoculated together into 400 µl of YPD in 2.2-ml 96-well deep-well plates and were co-cultured overnight for mating at 30 °C, at 200 rpm continuous shaking in a shaker incubator. For Mat a/ Mat α diploid selection, 10 µl of mated culture was re-inoculated in 2.2-ml 96-well deep-well plates containing 400 µl of diploid selection medium (SD media with glutamic acid and without lysine, arginine, and leucine) with antibiotics nourseothricin 100 µg/ml and geneticin 200 µg/ml. The culture plate was incubated for 48 h at a shaker incubator at 200 rpm at 30 °C. Subsequently, for sporulation, 10 µl of diploid cells was re-inoculated in 400 µl of sporulation medium (1% potassium acetate, 0.1% yeast extract, 0.05% glucose, 0.1% amino acid supplement powder from a mixture containing 2 g histidine, 10 g leucine, 2 g lysine, and 2 g uracil) in 2.2-ml 96-well deep-well plates, and the plates were incubated at 25°C in static condition for 5 days. From sporulation media, 10 µl of culture was seeded in haploid selection media (SD without leu/Arg/Lys) along with canavanine and thialysine to select meiotic cells in Matα background, and plates were incubated at 30°C for the next 2 days. The second round of haploid selection was done similarly in the presence of antibiotics nourseothricin and geneticin. After the second round of haploid selection, strains were inoculated for genomic DNA isolation and were confirmed for KANMX cassette integration, gene deletion, and mating type PCR.

Western blot

Wild-type yeast strains along with IMS-PMD and MM-PMD strains were inoculated in YP-raffinose (YPR) media and grown overnight at 30°C. The cultures were again re-inoculated in YPR media and grown until it reaches OD₆₀₀ 0.5–0.6. Then, the cultures were induced with 1% galactose and incubated at 30°C in the shaker incubator. At the same time, one set of cultures were kept separately as uninduced control. Small amounts of cultures were collected at 2 different time points (4h and 12h) after galactose induction. After the induction, cells were collected (equivalent to a total of 40 OD₆₀₀ cells for each strain and condition) and were lysed using the glass bead rupture method. The total protein content of each cell lysates was estimated by BCA assay, and a total of 30 µg of total proteins were loaded in each lane for SDS-PAGE. For Western blot with anti-Kar2 antibody, after transferring the proteins from SDS-PAGE in PVDF membrane, the membrane was blocked with 5% BSA in TBST for 2 h followed by overnight incubation with Kar2 primary antibody (y-115, cat no. sc-33630, Santa Cruz Biotechnology) in 1:5000 dilutions at 4°C. After incubation with HRP-conjugated anti-rabbit secondary antibody, the Western blot was developed using electrochemiluminescence (Thermo Fisher). For Western blot of DMMBP protein in IMS-DMMBP and IMS-DMMBP-ire1Δ strains or IMS-DMMBP and IMS-DMMBP-hac1Δ strains, whole cell lysates after 4 and 8 h of galactose induction were taken along with uninduced controls. Overnight incubation with polyclonal anti-MBP antibody (1:5000) was done followed by secondary antibody incubation and development by ECL as described for Kar2 Western blot.

Protease protection assay

To determine the localization of targeted model proteins in mitochondrial sub-compartments, samples were subjected to Proteinase K digestion as reported earlier (Gabriel et al. 2007; Zulkifli et al. 2020). Fifty micrograms of isolated mitochondria was spun down and resuspended in respective isotonic and hypotonic buffers, 400 µl of SH and H buffers (20 mM HEPES pH 7.4), respectively, for 15 min on ice. Intermitted tapping was done in between. Then, samples were treated with Proteinase K (total 8 µg) and incubated for 15 min on ice and then protease digestion was stopped with 2 mM PMSF (phenyl methyl sulfonyl fluoride) on ice for 15 min. Tubes were spun down at 17,500 × g for 30 min at 4 °C. The reisolated mitochondrial pellet was resuspended in SH buffer and 5X SDS loading dye. All the samples were boiled at 95 °C for 10 min and loaded on 10% SDS-PAGE.

RNA isolation and RT-PCR

Cultures were inoculated in YP-raffinose media and grown overnight. Secondary cultures were reinoculated with primary culture with the starting OD₆₀₀ of 0.1 in YP-raffinose media. After it reaches mid-log phase (OD₆₀₀ of 0.6), strains were induced with 1% galactose. As uninduced control, strains were grown in separate flask without the inducer. After 4 h of induction, cells were harvested and washed with 1X PBS and transferred to 1.5-ml tubes. RNA isolation was performed following cell lysis by glass bead method using Trizol reagent (Ambion, Catno. 15596018). One microgram of RNA was used for cDNA synthesis (Verso cDNA synthesis kit, Thermo AB1453/A) in 20-µl reaction. Master mix was made for qRT-PCR using Takara SYBR green mix and 1µl of cDNA (from 1:5dilution) was taken. DMMBP_F (5′ GGCGAAACAGCGATGACCA 3′) and DMMBP_R (5′ ACGAACGGTTTGGATGGTTG 3′) primers were used. Amplification was done using Biorad C1000Touch Thermal Cycler CFX384 Real time system with the condition of initial 95 °C for 3 min followed by 95 °C for 10s, 60°C for 30s, and 72°C for 30s for a total of 39 cycles. All RT-PCR were done in triplicates, and Ct value was automatically determined by the system.

Measurement of UPR induction by flow cytometry

YMJ003 strain (wild type) or the mitochondrial IMS and matrix specific proteotoxic stress models made in the YMJ003 background contain the UPRE-GFP reporter for reporting ER-UPR induction. As an internal control of transcription and translation, the second reporter protein mCherry induction is checked under constitutive promoter Tef2. By plotting the GFP/mCherry ratio for any strain harboring these reporters, the level of ER-UPR induction can be assessed as reported earlier (Jonikas et al. 2009; Maity et al. 2016). Cultures of different stressor protein expressing strains (exogenous control protein or stressor proteins specifically targeted to mitochondrial IMS and matrix as well as ER) along with control wild-type strains were inoculated in YP raffinose (YPR) media and grown overnight at 30 °C. The cultures were again re-inoculated in YPR media and grown until it reaches OD₆₀₀ 0.5–0.6. Then, the cultures were induced with 1% galactose and incubated at 30 °C in the shaker incubator. At the same time, one set of cultures was kept separately as uninduced control. Small amounts of cultures were collected at 2 different time points (4h and 8h) after galactose induction for measurement of GFP and mCherry fluorescence using LSRII (BD) and Beckman Coulter Cytoflex S flow cytometer. The ratio of mean GFP values and mean mCherry values of each strain were plotted and compared with wild-type yeast strain without stressor proteins (negative control) for assessing the ER-UPR induction with the expression of misfolded proteins.

Propidium iodide (PI) staining of yeast cells

To measure any cell death due to mitochondrial proteotoxic stress, yeast strains were grown in YPR and YPR + 1%-gal media. Cells were harvested after 24h and 48h of post-galactose induction. Harvested cells were washed with 1X PBS to remove the media and were resuspended in 1X PBS. As positive control, yeast cells were heated at 100°C for 15 min, and unstained cells were taken as negative control. Yeast cells were stained with 5 µg/ml propidium iodide for 10 min and samples were examined in CytoFlex flow cytometer (Beckman Coulter) using PE channel. A total of 50,000 events were recorded in each experiment; live and dead cell populations were selected from negative and positive controls, respectively, and applied to all test samples.

Imaging of yeast cells by confocal microscopy

To check the mitochondrial morphology, yeast strains were transformed with mitochondria targeted yeGFP plasmid (pVT-100U-mtGFP). Primary cultures were grown in synthetic raffinose media without uracil (SR-Ura) to select the plasmid and reinoculated in the same media and grown until 0.5 OD₆₀₀. Cultures were induced with 1% galactose for stressor protein expression and grown for 12h at 30°C. Aliquot of cells were taken and washed with 1X PBS and fixed with 3.5% formaldehyde. Furthermore, cells were washed with 1X PBS and coated on concanavalin A slides for adherence. After 15 min of incubation, images of slides were taken with Leica TCS SP8 confocal microscope. Uninduced strains (without galactose induction) were taken as a control.

Results

We have generated sub-mitochondrial (IMS and matrix-specific) proteotoxic stress model in yeast, Saccharomyces cerevisiae, by using well-described bipartite signal sequence of yeast cytochrome b2 (also known as Cyb2 SS) for IMS targeting (Fig. 1Ai) and a truncated version of it lacking 19 amino acid stop-transfer signal (also known as Cyb2Δ19 SS) for matrix targeting (Geissler et al. 2000; Popov-Celeketic et al. 2008) (Fig. 1Aii) as described before (Rao et. al 2020). To generate proteotoxic stress, two exogenous model misfolding and aggregation-prone proteins are used, the first one is double mutant version of E. coli maltose-binding protein (MBP), also known as DMMBP possessing two substitution mutations V8G and Y283D which render the protein slow folding and misfolding-prone (Wang et al. 1998; Mapa et al. 2012). The second stressor protein PMD (protein with misfolded domains) exhibits exposed hydrophobicity and amyloid-forming property as described previously (Rao et. al 2020) (Fig. 1Aiii). In parallel, we expressed wild-type well-folded MBP protein (Fig. 1Aiii) as control of exogenous protein overexpression within mitochondrial sub-compartments. The model stressor or control proteins are expressed from galactose-inducible Gal1 promoter. The whole DNA cassette containing Gal1 promoter, signal sequences, gene sequence encoding stressor (or control) proteins, and Cyc1-transcription termination sequence (Fig. 1Ai and Aii) is genome integrated in the URA3 locus of wild-type yeast strain YMJ003. The targeting of proteins by the signal sequence to mitochondrial sub-compartments has been described in detail recently by our group using control protein MBP by isolation of mitochondria (Rao et. al 2020) after induction of the protein with galactose. IMS targeted MBP protein is found in three forms within mitochondria (p or precursor form, I or intermediate form, and m or mature form) almost half of which are leaked by mitoplasting from the mitochondria and are not retrieved in pellet fraction equally as in case of matrix targeted one. In the matrix targeted form, the protein is found in two forms only (the p and the m forms) (Figure S1A and B) and the matrix targeted form remains protected after mitoplasting as expected. Notably, the expression of either of these stressor proteins, DMMBP and PMD, imparts sufficient stress to mitochondrial IMS which is reflected as a visible growth phenotype (Fig. 1B, growth phenotypes denoted as I). In contrast, only PMD protein imparts stress and results in visible growth phenotype in the mitochondrial matrix, while DMMBP do not show a visible growth defect upon its expression in the matrix (Fig. 1B, growth phenotypes denoted as M). The similar growth phenotypes are nicely recapitulated by measuring the growth curves of the strains in liquid media (Fig. 1C and D). Absence of visible toxicity of DMMBP upon expression in matrix can be explained by its faster folding assisted by Hsp60/10 chaperonin system present in the mitochondrial matrix. DMMBP protein being a known GroEL/ES (E.coli Hsp60/10 chaperonin) substrate (Chakraborty et al. 2010; Sharma et al. 2008) is expected to fold faster to its native form in the presence of the mitochondrial Hsp60/10 chaperonin which are close homologs of the E.coli GroEL/ES system. In contrast, IMS being devoid of Hsp60/10 chaperonin system, DMMBP would fold slower being a slow-folding mutant protein and the chance of being in non-native state would be longer leading to more misfolding of the protein in IMS. It is important to note that expressing well-folded control protein MBP with the same inducer does not impart any growth defects in either of the mitochondrial sub-compartments (Fig. 1B, C, and D). Without inducer (in YPR media), all strains grow similarly indicating that there is no discernible leaky expression from the Gal1 promoter and there are no growth defects due to integration of the DNA cassette for expression of the control or stressor proteins in the yeast strains (Fig. 1B, C, and D). Expression of PMD protein in either of the mitochondrial sub-compartment led to severe mitochondrial fragmentation and disruption of its tubular network as appreciated by comparing the images of the yeast cells expressing mitochondria targeted yeGFP with and without the galactose-induction (Fig. 1E and F and Figure S1C and D). The amyloid-forming PMD protein forms large aggregates as observed by microscopy (Figure S1D) and as shown before also biochemically (Rao et. al 2020)(large aggregates of the protein are found in stacking gel of SDS-PAGE during Western blot). Interestingly, despite drastic growth defects and severe mitochondrial fragmentation due to mitochondrial proteotoxic stress, the cell death was negligible for both IMS (Figure S2A) or matrix (Figure S2B) stress. Altogether, we show that proteotoxic stress within mitochondrial IMS or matrix leads to mitochondrial fragmentation and growth arrest of yeast cells. The phenotype depends on the nature of stressor proteins expressed and the sub-mitochondrial compartment in which the protein is targeted.

Fig. 1.

Proteotoxic stress in mitochondrial sub-compartments leads to growth defects and mitochondrial fragmentation of yeast cells. Ai and Aii A schematic representation of DNA cassette used for integration in yeast genome for expression and targeting of stressor proteins (and folded protein control) to mitochondrial intermembrane space (IMS) and matrix (MM). Cyb2SS indicates a signal sequence of yeast cytochrome b2 (Cyb2) (Ai) which is a bipartite targeting signal of total 167 amino acids containing N-terminal positively charged presequence (shown in gray) followed by hydrophobic membrane sorting stop transfer sequence of 19 amino acids shown in orange. Cyb2SS is used for specific targeting to mitochondrial IMS. The truncated version of Cyb2SS (Cyb2Δ19) which is deleted of stop transfer sequence (Aii) is used for specific targeting to mitochondrial matrix (MM). The proteins are expressed from the inducible Gal1 promoter and the Cyc1 terminator has been used for transcription termination (Gal1p and Cyc term in the schematic, respectively). Aiii Schematic representation of stressor proteins (and control) used for generating mitochondria IMS or MM specific stress model in yeast. Maltose-binding protein or wild-type MBP (as control of folded exogenous protein) from E. coli, its slow folding mutant version, DMMBP containing two substitutions V8G and Y283D and a third protein PMD which forms an amyloid type of aggregates are shown schematically. B Drop-dilution assay of yeast strains expressing the misfolded and aggregation-prone stressor proteins DMMBP and PMD along with control protein MBP targeted to mitochondrial IMS and matrix. Yeast strains grown until mid-log phase (OD₆₀₀ 0.5) were serially diluted (1:10) and spotted on solid agar media without inducer (YPR) and with inducer [YPR + 1% gal (yeast-extract-peptone-raffinose + 1% galactose as inducer)]. Slow growth phenotype of IMS-PMD, IMS-DMMBP, and MM-PMD strains are visible exclusively after induction of expression of stressor proteins in media with the inducer (YPR-gal plates) in comparison to the control (YPR plates). “I” and “M” indicate spots with visible growth phenotype in the case of IMS-PMD/DMMBP and MM-PMD strains, respectively. No growth phenotype was observed in YPR-gal plates in case of wild type yeast strain or strains expressing control folded protein MBP. Growth assay was performed at least three times and one representative picture has been shown. Plate pictures were taken after 36h of incubation at 30°C. C. Growth curves of Wt, IMS-MBP, IMS-DMMBP, and IMS-PMD strains were plotted after growing all the strains in liquid media without inducer (YPR) (upper panel) and with inducer (YPR + 1%gal) (lower panel). As shown in panel B in solid agar media, in media without inducer (YPR), all strains grow similarly (upper panel). In contrast, in presence of the proteotoxic stress due to the expression of stressor proteins DMMBP and PMD by galactose induction (YPR + gal media), prominent growth defect is visible in the case of IMS-DMMBP and IMS-PMD strains compared to Wt strain (lower panel). D Similar to panel C, growth curves of MM-MBP, MM-DMMBP, and MM-PMD strains are shown along with Wt yeast cells as control.E–F. Confocal fluorescence microscopy images of mitochondria of yeast strains (D panel-IMS-PMD and E panel-MM-PMD) by mitochondria-targeted yeGFP (yeast enhanced GFP) in the absence (upper panels) and presence (lower panels) of galactose (inducer of mitochondria-targeted stressor proteins). The upper panels show (without inducer) a tubular network of healthy mitochondria and lower panels (with inducer) display fragmentation and disruption of the tubular network of mitochondria due to proteotoxic stress. The scalebar represents 5 µm

Genetic disruption of the ER-UPR pathway aggravates the IMS-proteotoxic stress phenotype due to ineffective clearance of misfolded proteins from IMS

As the ER-UPR sensors are found to be localized in the ER-mitochondria contact sites, especially during stress, it was interesting to check whether altering the ER-UPR pathway would change the mitochondrial proteotoxic stress phenotypes due to the close connection between the organelles. Thus, to understand any modulatory function of the ER-UPR pathway on mitochondrial proteotoxic stress phenotypes, we deleted IRE1 or HAC1 (role of IRE1 and its downstream HAC1 in ER-UPR in yeast is summarized in Fig. 2A) in the background of strains expressing stressor proteins (PMD and DMMBP) within mitochondrial sub-compartments, IMS and matrix, or in the control strain expressing the wild-type protein MBP in the sub-mitochondrial compartments. Although the Ire1-Hac1-mediated signalling of ER-UPR is the exclusive UPR signalling pathway in yeast, Saccharomyces cerevisiae, there are evidences of Ire1-independent Hac1 activation in literature (Leber et al. 2004) which prompted us to check the phenotypes upon deletion of both of these components of ER-UPR (ire1Δ or hac1Δ) independently. It is important to mention here that ire1Δ or hac1Δ strains do not exhibit any growth phenotype (ire1Δ or hac1Δ strains grows like the wild-type yeast strain) in absence (Fig. 2B YPR plates and Fig. 2C upper panel) or in presence of the inducer (galactose) (Fig. 2B YPR-gal and Fig. 2C lower panel). Interestingly, deletion of IRE1 aggravated the growth phenotype of the strain expressing stressor protein PMD in mitochondrial IMS (Fig. 2B—upper panel—and Figure S3A) indicating an important role of IRE1 during proteotoxic stress response of mitochondrial IMS. In contrast, IRE1 deletion imparted growth benefits to cells possessing the same proteotoxic stress in the mitochondrial matrix (Fig. 2B—upper panel—and Figure S3B). A similar observation was made when the downstream gene HAC1 was deleted. HAC1 deletion aggravated the growth phenotype of the IMS-PMD strain while alleviating the phenotype of the strain harboring the same stressor protein in the mitochondrial matrix (Fig. 2B, lower panel). The growth phenotype due to IRE1 or HAC1 deletion combined with PMD protein–induced IMS proteotoxic stress was also evident from growth curves obtained from liquid cultures (Fig. 2C lower panel). To rule out any modulatory role of Ire1-independent Hac1 activation on the mitochondrial proteotoxicity, we made the double deletions of both IRE1 and HAC1 in IMS-PMD and MM-PMD background. In case of IMS proteotoxicity, the aggravation of growth phenotype remained similar in double deletion in comparison to single deletions (Figure S3C) indicating exclusive role of Ire1-Hac1 signalling mediated modulatory role of ER-UPR pathway on IMS proteotoxicity. Interestingly, the stress-alleviating role of single deletion of IRE1 on matrix proteotoxicity became more prominent in double deletion of IRE1 and HAC1 indicating a possible modulatory role of an Ire1-Hac1-independent pathway on the matrix proteotoxicity (Figure S3C). This result is interesting but is puzzling simultaneously and needs further exploration for its molecular explanation. Although the growth phenotype aggravation due to IRE1 deletion was not so evident in the case of IMS-DMMBP strain, when we compared the level of misfolding-prone DMMBP protein in the absence and presence of IRE1 (IMS-DMMBP-ire1Δ and IMS-DMMBP strains, respectively), we found a significant accumulation of the DMMBP protein in strains deleted of IRE1 (Fig. 2D). A similar level of DMMBP protein accumulation was also observed upon deleting HAC1 in IMS-DMMBP strain (IMS-DMMBP-hac1Δ) (Figure S4A). Although there is only nominal increase (1.2–1.4-fold) in the gene expression level of DMMBP in IMS-DMMBP-ire1Δ or IMS-DMMBP-hac1Δ strains compared to IMS-DMMBP (Figure S4B), the accumulation of proteins is much larger (~ 3–4 fold) upon deletion of IRE1 or HAC1. This data attribute to important role of IRE1-HAC1 mediated ER-UPR signalling in clearing misfolded proteins from mitochondrial IMS. Altogether, we show that during mitochondrial proteotoxic stress, the ER-UPR pathway plays an important role in the stress tolerance of IMS proteotoxic stress. By contrast, disruption of the ER-UPR pathway is beneficial during mitochondrial matrix proteotoxic stress.

Fig. 2.

Disrupting ER-UPR signalling is harmful during mitochondrial IMS proteotoxic stress while it imparts cellular fitness during matrix stress. A A schematic picture depicting ER-UPR signalling events of yeast S. cerevisiae. Step 1, ER-transmembrane protein Ire1 is the sole sensor of ER-UPR in yeast. Ire1 has a sensory domain within ER lumen (luminal domain or LD) and cytosol exposed domains harbouring kinase (KD) and RNAse activities (RnaseD). Step 2, upon sensing the stress within ER lumen, Ire1 binds to the misfolded proteins leading to its activation and oligomerization. Step 3, activated Ire1 binds and splices the HAC1 mRNA by its RNAse activity. Step 4, Spliced HAC1 mRNA translates to Hac1p protein and translocates to nucleus where it acts as a transcription factor to upregulate UPR-responsive genes. B Upper panel: IMS-PMD, MM-PMD, IMS-PMD-ire1Δ, and MM-PMD-ire1Δ strains were spotted along with wild-type yeast strain and ire1Δ strains as controls on YPR and YPR + gal plates to assess the any changes in growth phenotypes of IMS-PMD and MM-PMD strains in combination with ire1Δ. Plate pictures were captured after 52 h of incubation at 30 °C to capture the prominent growth defect of IMS-PMD-ire1Δ (denoted as I⁺) compared to IMS-PMD (I) and growth phenotype alleviation of MM-PMD-ire1Δ (denoted as M⁻) compared to MM-PMD (M). Lower panel: IMS-PMD, MM-PMD, IMS-PMD-hac1Δ, and MM-PMD-hac1Δ strains were spotted and pictures were taken after 52 h of incubation at 30 °C to display the prominent growth defect of IMS-PMD- hac1Δ (I⁺) compared to IMS-PMD (I) and growth rescue of MM-PMD-hac1Δ (M⁻) compared to MM-PMD (M). “*” indicates a contaminating spot due to spillage during spotting. C Growth curves of Wt, ire1Δ, hac1Δ, IMS-PMD, IMS-PMD-ire1Δ, IMS-PMD-hac1Δ strains were plotted after growing all the strains in liquid media without inducer (YPR) (upper panel) and with inducer (YPR + 1%gal) (lower panel). As shown in panel B in solid agar media, in media without inducer (YPR), all strains grow similarly (upper panel). In contrast, in presence of the proteotoxic stress due to the expression of stressor protein PMD by galactose induction (YPR + gal media), prominent growth defect is visible in the case of IMS-PMD strain compared to Wt strain (lower panel). Like Wt strain, ire1Δ, hac1Δ strains do not exhibit any difference in growth rate in YPR + gal media but when combined with IMS-PMD, IMS-PMD-ire1Δ, IMS-PMD-hac1Δ strains grow slower than IMS-PMD strain in YPR + gal media indicating the importance of ER-UPR signalling by IRE1-HAC1 pathway in stress tolerance during mitochondrial IMS proteotoxic stress. D Western blot by anti-MBP antibody to show the steady-state level of stressor protein DMMBP was performed using yeast whole-cell lysates made from IMS-DMMBP and IMS-DMMBP-ire1Δ strains, after 4 h and 8 h of galactose induction along with uninduced control. Middle and lower panel: GAPDH level and Amido-back stained PVDF membrane are shown as loading controls

Activation of ER-UPR is beneficial during mitochondrial IMS proteotoxic stress but is detrimental during matrix proteotoxic stress

As genetic perturbation of the ER-UPR pathway affected the fate of the mitochondrial IMS or matrix proteotoxic stress phenotypes differently, it was interesting to check whether artificial activation of ER-UPR may differentially affect proteotoxic stress phenotypes of mitochondrial IMS and matrix. Before proceeding to experiments, we reanalyzed microarray data available in the public domain from previous studies (Maity et al. 2016) and found that DTT (reductive) stress in the wild-type yeast strain (same strain used in this study) causes substantial changes in the gene expression of components of protein quality control machineries of mitochondrial IMS like Mia40, Erv1 system, and small Tim proteins indicating perturbation of mitochondrial protein homeostasis during DTT-induced ER stress (Fig. 3A) (Maity et al. 2016). Additionally, DTT stress was shown to significantly alter mitochondrial dynamics (severe fragmentation) in the absence of the ER-UPR sensor IRE1, in yeast (Maity et al. 2014). Altogether, it is evident from previous literature that perturbation of ER protein homeostasis leads to changes in mitochondrial protein expressions and the morphology of the organelle. To check the effect of ER-UPR induction with ER stressors, we treated the yeast strains expressing stressor proteins in mitochondrial IMS and matrix with small-molecule ER stressors, DTT (induces ER stress by imparting reductive stress) and tunicamycin (induces ER stress by blocking N-linked glycosylation). Interestingly, treatment with 5 mM DTT or 0.5 µg/ml of tunicamycin alleviated the growth phenotype of IMS-PMD strain indicating a beneficial role of activation of ER-UPR pathway during mitochondrial IMS proteotoxic stress (Fig. 3B). We did not use higher concentrations (2.5 µg/ml or more) of tunicamycin that is usually used to impart ER stress in yeast, as a combination of multiple proteotoxic stresses (tunicamycin and misfolded protein) severely slowed down cell growth at higher tunicamycin concentrations. The lower concentrations of tunicamycin used (0.5 µg/ml) was sufficient to impart ER stress as evident by negligible growth of ire1Δ or hac1Δ cells (Fig. 3D). In contrast to IMS-PMD, the phenotype of matrix proteotoxic stress in MM-PMD strain was significantly aggravated when ER stress was imparted with these ER stressors (Fig. 3C). The phenotype of wild-type yeast strain or control strains expressing folded proteins in both mitochondrial sub-compartments remained unaltered (spots of IMS-MBP and MM-MBP). This data indicated a beneficial role of ER-UPR activation during specific proteotoxic stresses in mitochondrial IMS and a detrimental role during mitochondrial matrix. Importantly, the alleviated phenotype of IMS-PMD during ER stress (induced by DTT or tunicamycin) could not support the growth of IMS-PMD-ire1Δ or IMS-PMD-hac1Δ. Like the single deletions strains ire1Δ and hac1Δ strain, the IMS-PMD-ire1Δ (Fig. 3D) and IMS-PMD-hac1Δ (Figure S4C) strains too were unable to grow in media with ER stressors. This data indicated that intact ER-UPR signalling through IRE1-HAC1 is key for better proteotoxic stress tolerance of IMS-PMD strain during ER stress. Deletion of ER-UPR components combined with matrix proteotoxicity (MM-PMD-ire1Δ and MM-PMD-hac1Δ) also rendered these strains unable to grow in plates with ER stressors probably due to the combined effect of IRE1 or HAC1 deletion and aggravation of stress phenotypes of MM-PMD strain by ER stressors (Fig. 3D and Figure S4C).

Fig. 3.

Inducing ER stress imparts cellular fitness against IMS proteotoxic stress but aggravates matrix proteotoxic stress in an IRE1-dependent manner. A Transcript levels of mitochondrial IMS oxidative folding machinery and small Tim proteins which act as chaperones and work in concert with the oxidative folding machinery were plotted. The Y-axis represents the Log2fold change of transcript levels compared to untreated control cells. Microarray data of yeast after DTT-induced ER stress compared to untreated control, as reported previously (Maity et al. 2016) was analyzed to plot the bars. B Drop-dilution assay of yeast strains expressing the stressor proteins DMMBP and PMD along with control protein MBP targeted to mitochondrial IMS (IMS-MBP, IMS-DMMBP and IMS-PMD) along with wild-type yeast strain as control, by spotting on the control plate (YPR), with inducer plate (YPR + gal) and on inducer with ER stressors (0.5 µg/ml) and DTT (5 mM) was performed as described in Fig. 1C. “I” indicates spots with visible growth phenotype in the case of IMS-DMMBP and IMS-PMD strains, while “I-” indicates alleviation in the phenotype of IMS-PMD strain while combined with treatment with ER stressors tunicamycin or DTT. The arrowheads indicate alleviation of phenotypes in plates with ER stressors combined with IMS proteotoxic stress in comparison to only IMS proteotoxic stress. Plate pictures were taken after 48 h of incubation at 30 °C. Experiments were performed at least three times and one representative picture has been displayed. C. Similar to panel B, MM-MBP, MM-DMMBP, and MM-PMD strains along with wild-type yeast strains were spotted on YPR, YPR + gal, and YPR + gal with ER stressors (0.5 µg/ml) and DTT (5 mM) plates as described in Fig. 1C. The arrowheads indicate an aggravation of phenotypes in plates with ER stressors combined with matrix proteotoxic stress in comparison to only matrix proteotoxic stress. Plate pictures were taken after 48 h of incubation at 30 °C. “M” indicates spots with visible growth phenotype in case of MM-PMD, while “M + ” indicates aggravation of phenotype of MM-PMD strain while combined with treatment with ER stressors tunicamycin or DTT. D IMS-PMD, MM-PMD, IMS-PMD- ire1Δ, and MM-PMD- ire1Δ strains were spotted along with wild-type yeast strain and ire1Δ strain as controls on YPR, (YPR + gal) and (YPR + gal) with ER stressors (0.5 µg/ml) and DTT (5 mM) plates as described in panel A and B. “I” and “M” indicate spots with visible growth phenotype in case of IMS-PMD and MM-PMD strains respectively on YPR + gal plates, while “I-” indicates alleviation of the phenotype of IMS-PMD strain when spotted on YPR + gal plates with ER stressors, DTT or tunicamycin and “M + ” indicates aggravation in the phenotype of MM-PMD strain while combined with ER stressors

Altogether, we show that an intact Ire1-Hac1 signalling of ER-UPR is important to cope up with IMS proteotoxic stress and artificially activating ER-UPR helps to alleviate IMS proteotoxicity. In contrast, disrupting basal ER-UPR signalling by Ire1-Hac1 is beneficial for matrix stress which becomes evident as alleviation of growth phenotypes of matrix proteotoxicity when combined with ire1∆ or hac1∆ deletions. In corroboration, activation of ER-UPR by ER stressors aggravates the proteotoxic phenotypes of matrix by an yet elusive mechanism.

Proteotoxic stress in mitochondrial IMS does not lead to activation of ER-UPR despite its close association with ER through contact sites

As we observed the beneficial effect of ER-UPR induction on mitochondrial IMS proteotoxicity, it was interesting to check if ER-UPR is elicited as an adaptive cellular response to cope with mitochondrial IMS proteotoxic stress. For this purpose, induction of ER-UPR was measured by monitoring the expression of reporter protein GFP (Green Fluorescent protein) from UPR element (UPRE-GFP) which is genome integrated in the wildtype strain YMJ003 used in this study (Fig. 4A) (Jonikas et al. 2009; Maity et al. 2016). Simultaneously, second fluorescent protein mCherry is expressed from constitutive Tef2 promoter which acts as internal control of cellular transcription and translation (Fig. 4A). An increased GFP/RFP ratio thus indicates induction of ER-UPR. Notably, there was no significant increase in GFP/mCherry ratio with IMS proteotoxic stress (with any of the stressor protein, PMD or DMMBP) indicating no induction of ER-UPR due to insufficient ER stress during proteotoxic stresses in mitochondrial IMS despite its close connection with ER through inter-organellar contact sites (Fig. 4B). Proteotoxic stress in mitochondrial matrix stress also did not induce ER-UPR. As a positive control of ER-UPR induction, same stressor proteins PMD and DMMBP were expressed within ER lumen which showed a significant increase in GFP/mCherry ratio after expression of the stressor proteins by galactose induction. Importantly, expression of folded control protein MBP did not induce ER-UPR (Fig. 4B and C). Likewise, ER-UPR marker protein Kar2 (ER resident Hsp70 molecular chaperone) which gets upregulated with induction of ER-UPR, remained unaltered during mitochondrial IMS or matrix stress (Fig. 4D) further confirming that ER-UPR is not induced during proteotoxic stress in closely apposed mitochondrial sub-compartments.

Fig. 4.

Mitochondrial proteotoxic stress in its IMS or matrix does not induce ER-UPR. A Schematic representation of the UPRE-GFP reporter that is integrated into the genome of wild type yeast strain (YMJ003 strain) (Jonikas et al. 2009; Maity et al. 2016) used in this study. GFP is expressed under the UPR element (UPRE) and reports for induction of ER-UPR. As the reporter of cellular transcription and translation, a second fluorescent reporter protein, mCherry is expressed under constitutive Tef2 promoter. B Mitochondrial sub-compartment-specific proteotoxic stressor model yeast strains (IMS-PMD/DMMBP and MM-PMD/DMMBP) along with control strains expressing folded protein (IMS-MBP, MM-MBP), wild-type yeast strain and the yeast strain expressing the control protein MBP or stressor protein DMMBP specifically targeted to the endoplasmic reticulum (ER) were grown until mid-log phase and stressor (or control) proteins were induced with 1% galactose. After 4h and 8 h of galactose induction, GFP and mCherry fluorescence were measured by flow cytometry. The ratio of mean GFP and mCherry fluorescence was plotted as bar plots for all strains. Error bars represent standard deviation between repeats (n = 3). P value was calculated by unpaired Student’s T-test (2-tailed). NS indicates “not significant” with a P value of more than 0.05. C As described in panel B, yeast strains expressing the control protein MBP or stressor proteins DMMBP and PMD specifically targeted to the endoplasmic reticulum (ER) were taken and GFP and mCherry fluorescence were measured by flow cytometry after 4 h and 8 h of galactose induction (4h Ind and 8h Ind, respectively). At the same time points, uninduced cells (UI) were also taken for GFP and mCherry fluorescence measurement by flow cytometry. The ratio of mean GFP and mCherry fluorescence was plotted as bar plots for all strains. Error bars represent standard deviation between repeats (n = 3). P value was calculated by unpaired Student’s T-test (2-tailed). D Western blot by Kar2 polyclonal antibody was performed using yeast whole-cell lysates made from wild-type yeast strain, IMS-PMD and MM-PMD strain after 4h and 12h of galactose induction along with uninduced control (cells taken out at “0” time point of galactose induction). The Amido black-stained PVDF membrane after Western transfer of proteins has been shown as the loading control

Altogether, we show that proteotoxic stress in mitochondria remains limited to the organelle and does not impart stress to neighboring organelle ER sufficiently to activate ER-UPR despite its close connection with mitochondrial sub-compartments especially IMS through inter-organelle contact sites. Although, we show that artificial induction of ER-UPR is helpful to ameliorate the IMS proteotoxicity, it is not naturally elicited as an adaptive response to IMS proteotoxic stress.

Overexpressing ERMES components alleviate the phenotypes of proteotoxic stress within mitochondrial sub-compartments independent of the ER-UPR pathway

Previous studies have shown that yeast ERMES components like MDM12 and MDM34 are substantially upregulated and components like MMM1 is marginally upregulated while MDM10 is downregulated during DTT-induced ER stress (Fig. 5A) (Maity et al. 2016). As ER stress leads to changes in expression of ER-mitochondria tethering proteins, it is expected that mitochondrial communication with ER and alteration in protein homeostasis within its sub-compartments, especially IMS, would change during ER stress. To check whether changing the expression level of ERMES components would modulate the proteotoxic stress phenotypes of mitochondrial sub-compartments, we overexpressed the components of yeast ERMES, MDM12, MDM34, MDM10, and MMM1 from the yeast ORF overexpression plasmid library. Interestingly, overexpression of MDM34 and MDM12 showed prominent growth phenotype alleviation in IMS-PMD strain (Fig. 5B). Overexpression of MDM10 and MMM1 was also helpful during IMS proteotoxic stress although the extent of phenotype rescue by overexpression of these two ERMES components was less prominent than the overexpression of MDM34 and MDM12. This alleviation of growth phenotypes remained mostly similar during IMS proteotoxic stress when combined with ER stress with 5 mM DTT (Fig. 5B). During ER stress with 0.5 µg/ml of tunicamycin, IMS-PMD growth defect was rescued by overexpression of MDM12 as in the case of without tunicamycin stress (Fig. 5B). Notably, overexpressing other ERMES components, MDM34, MDM10, and MMM1 were not helpful; rather, overexpression of these three components aggravated the growth phenotype of IMS-PMD strain when combined with tunicamycin stress (Fig. 5B). The alleviation of phenotype with overexpression of ERMES components specially MDM12 and MDM34 was also observed in matrix proteotoxic stress (Fig. 5C). In the presence of ER stressors, DTT, or tunicamycin, the growth phenotype alleviation by overexpressed ERMES components remained similar in the case of MM-PMD strain. As Ire1 has been recently shown to be localized at ER-mitochondria contacts sites in mammalian cells where it helps in Ca²⁺ transfer to mitochondria matrix from ER (Carreras-Sureda et al. 2019), and directly impact the mitochondrial respiration, it was interesting to check whether the modulatory effect of overexpressed ERMES components is related to ER-UPR pathway. To check whether the beneficial effect of overexpression of ERMES components is independent of the ER-UPR pathway, we overexpressed the ERMES components in the background of HAC1 deleted IMS-PMD and MM-PMD strains. As shown before, the IMS-PMD-hac1Δ strain is significantly more sensitive to IMS proteotoxic stress (slow growth phenotype of IMS-PMD-hac1Δ compared to IMS-PMD is more prominent in synthetic media) than IMS-PMD strain as described before (Fig. 2B, lower panel). When the individual ERMES components were overexpressed in the IMS-PMD-hac1Δ, all individual ERMES components except MMM1 could prominently rescue the phenotype of IMS-PMD-hac1Δ indicating the HAC1-independent role of ER-mitochondria tethering structure in the IMS proteotoxic stress tolerance (Fig. 6A). Due to deletion of HAC1, MM-PMD strain showed significant fitness during matrix proteotoxicity (Fig. 6B, and Fig. 2B lower panel) and any further effect of overexpression of MDM12, MDM34, or MMM1 was not discernible (Fig. 6B). Only overexpression of MDM10 in MM-PMD-hac1Δ aggravated the phenotype, the reason of which needs further exploration.

Fig. 5.

Overexpression of components of ER-mitochondria encounter structure (ERMES) leads to alleviation of proteotoxic stress phenotype of mitochondrial IMS and matrix. A Transcript levels of yeast ERMES components after DTT-induced ER stress measured by microarray as reported previously (Maity et al. 2016) was plotted. The Y-axis represents the Log2fold change of transcript levels compared to untreated control cells. B–C Drop-dilution assay of yeast strains IMS-PMD (panel B) and MM-PMD (panel C) transformed with plasmids overexpressing (OE) individual components of yeast ERMES (MDM10, MDM12, MDM34, and MMM1) was performed by spotting the yeast strains along with wild-type yeast strain and same strains with empty vector (EV) as controls on the (SR-Ura) plates (synthetic media without uracil with 2% raffinose), with inducer plate (SR-Ura + gal) and (SR-Ura + gal) with ER stressors, tunicamycin (0.5 µg/ml), or DTT (5 mM) as described in Fig. 1C. Plate pictures were taken after 48 h of incubation at 30 °C

Fig. 6.

Overexpression of individual ERMES subunits alleviates mitochondrial proteotoxic stress phenotype independent of ER-UPR signalling. A–B Drop-dilution assay of yeast strains IMS-PMD-hac1Δ (panel A) and MM-PMD hac1Δ (panel B) transformed with plasmids overexpressing (OE) individual components of yeast ER-mitochondria contact sites (MDM10, MDM12, MDM34, and MMM1) was performed by spotting the strains along with wild-type yeast strain and same strains with empty plasmid vector (EV) as controls (SR-Ura), (SR-Ura + gal), (SR-Ura + gal) with ER stressors, tunicamycin (0.5 µg/ml), or DTT (5 mM) as described in Fig. 1C. Plate pictures were taken after 48 h of incubation at 30 °C. C A comprehensive model of the current work is summarized in panel C. The upper box of panel C shows the proteotoxic stress in mitochondrial IMS. The basal level of UPR is beneficial for stress tolerance for mitochondrial IMS proteotoxicity. Furthermore, activated ER-UPR by environmental stressors helps to ameliorate the IMS proteotoxic stress. Apart from role of ER-UPR, enhanced ER-mitochondria contact sites by ERMES complex is also helpful in stress tolerance in IMS in an ER-UPR-independent way. The lower box shows proteotoxic stress in mitochondrial matrix. In case of matrix stress, even basal UPR is not beneficial and mitochondrial matrix proteotoxicity is better managed if basal ER-UPR is disrupted. In corroboration, activated UPR by ER stressors is detrimental and aggravates the matrix proteotoxicity. Similar to IMS proteotoxic stress, increased ER-mitochondria contacts are also helpful in matrix proteotoxicity

In summary, we show that overexpression of most of the individual components of ER-mitochondria encounter structure (ERMES) of yeast, Saccharomyces cerevisiae is helpful in stress tolerance to mitochondrial IMS or matrix proteotoxic stress. The beneficial role of overexpression of ER-mitochondria contact sites during mitochondrial proteotoxicity is independent of the IRE1-HAC1 pathway of ER-UPR indicating multiple routes of ER-mitochondria communication during mitochondrial proteotoxic stress.

Discussion

Mitochondria, apart from its role in essential biosynthetic, metabolic, and energy production processes and as the master regulator of cell death pathway, have gained a lot of attention as a cellular hub of proteotoxic stresses. Recently we have reviewed various causes of mitochondrial proteotoxicity and its implication on ER-mitochondria communication and Ca²⁺ homeostasis (Ali et al. 2021). Evidence of the presence of disease-associated misfolded and aggregated mutant proteins within the mitochondrial sub-compartments in different pathologies are plenty in literature (Mattiazzi et al. 2002; Caspersen et al. 2005; Devi et al. 2006; Hansson Petersen et al. 2008; Manczak and Reddy 2012). Such accumulation of aggregated or misfolded proteins within mitochondrial sub-compartments would result in proteotoxic stress of the organelle. Mitochondrial stress thus generated may perturb the homeostasis of neighboring compartments like ER due to its close connection with the later through contacts sites. It is known that ER-UPR signalling is important for pathways like ERAD (ER-associated degradation) (Travers et al. 2000) which is not only an important quality control process for ER protein homeostasis but also equally important for mitochondria as an unimpaired ERAD is critical for preventing mitochondrial dysfunctions (Liu et al. 2020). Furthermore, some of the components of ERAD like Ubx2 is a major player of Translocation associated degradation (MitoTAD), a surveillance pathway that prevents the blocking of TOM complex pore from jamming by precursor proteins (Martensson et al. 2019) which play important role in mitochondrial protein homeostasis. Thus, ER-UPR signalling is expected to have a profound effect on mitochondrial protein homeostasis. In the current work, by generating a proteotoxic stress model in yeast, Saccharomyces cerevisiae, by specifically targeting stressor proteins to mitochondrial IMS and matrix, we show that the cellular response to mitochondrial proteotoxic stress is uniquely modulated by the ER-UPR pathway. We show that upon disrupting the ER-UPR signalling by deletion of ER-UPR sensor IRE1 or its downstream effector HAC1, the proteotoxic stress phenotype of IMS is much aggravated indicating the importance of intact basal ER-UPR signalling (Jonikas et al. 2009) during IMS proteotoxic stress. We speculate that basal ER-UPR signalling maintains the surveillance mechanisms like ERAD which plausibly impart a protective role on closely apposed mitochondria by removing the mitochondrial outer membrane-associated misfolded or aggregated proteins that can reduce the impact of proteotoxic stresses in IMS (Fig. 6C, upper panel). Surprisingly, deletion of IRE1 or HAC1 alleviated the phenotype of matrix proteotoxicity indicating a beneficial role of disrupted basal UPR during matrix proteotoxic stress. The opposite effect of the ER-UPR pathway on mitochondrial matrix stress is interesting but baffling (Fig. 6C, lower panel). The possible explanation for the beneficial effect of the absence of IRE1 can be due to Ire1’s role in Ca²⁺ transfer from ER to mitochondria and maintenance of mitochondrial respiration as shown in mammalian cells (Carreras-Sureda et al. 2019). We have reported that during mitochondrial matrix proteotoxic stress, mitochondrial respiratory apparatus is significantly downregulated, and activation of mitochondrial respiration is detrimental for cells (Rao et. al 2020). We speculate that Ire1 maintains Ca²⁺ transfer from ER to mitochondria which activates the mitochondrial dehydrogenases and mitochondrial respiration and in absence of IRE1 the respiratory load on mitochondria is less and it helps in better stress tolerance during matrix proteotoxicity (Carreras-Sureda et al. 2019). Interestingly, we show that despite the beneficial role of ER-UPR induction on mitochondrial IMS proteotoxic stress (Fig. 6C, upper panel), IMS stress is not able to induce ER-UPR as an adaptive response. This is in corroboration to our earlier finding that perturbations of cellular proteostasis network mostly elicits non-optimal adaptive responses, at least in a unicellular eukaryote like yeast Saccharomyces cerevisiae (Ghosh et al. 2019). The current findings again indicate that although ER-UPR induction is beneficial during IMS stress, yet yeast cells are unable to mount the response sufficiently as an adaptive mechanism. We show that artificial activation of ER-UPR by ER stressors can be beneficial during IMS proteotoxicity, and this effect can be exploited therapeutically in case of pathologies associated with mitochondrial dysfunctions due to IMS proteotoxicity.

We speculate that not only ER-UPR, but ER-mitochondria communication would be important during mitochondrial proteotoxic stress. Many mitochondrial proteins especially inner membrane proteins are known to get synthesized on the ER surface (Williams et al. 2014) and ER surface play a crucial role in salvaging and redirecting many mitochondrial proteins by a unique surveillance mechanism known as ER-SURF (Hansen et al. 2018). All these evidences indicates that ER-mitochondria contacts are not only important for maintaining signalling and metabolic functions by mitochondria but also crucial for the biogenesis of mitochondria (Fig. 6C). Indeed, we show that overexpressing ERMES components helps in coping with mitochondrial proteotoxic stress in an ER-UPR independent way. This finding indicates the beneficial role of the apposition of two organelles during mitochondrial proteotoxic stress. The exact mechanism of stress tolerance by overexpressed ERMES components needs to be explored in detail, in future.

Supplementary Information

Below is the link to the electronic supplementary material.

Author contribution

The work was conceived by KM. All yeast strains were generated by KBN and RS. Yeast experiments were done by RS, KBN. FACS experiments were done by RS, KBN, and MJ. Western blot experiments were performed by KBN and MJ. KM analyzed the data and wrote the manuscript.

Declarations

Conflict of interest

The authors declare no competing interests.