BIP orchestrates bidirectional ER protein trafficking via co-chaperone complexes
Suma Biadsy, Ayelet Gilad, Laila Abu Madegam, Aeid Igbaria
TL;DR
This paper shows how cells redistribute proteins between the ER and cytosol under stress, revealing a new role for BIP in bidirectional protein trafficking.
Contribution
The study identifies a novel, UPR-regulated mechanism of ER-to-cytosol protein export involving BIP, DNAJB12/14, and SGTA.
Findings
ATF6 and IRE1, but not PERK, are essential for initiating ER-to-cytosol protein redistribution.
BIP forms a complex with DNAJB12/14 and SGTA to facilitate ER protein export.
IRE1 simultaneously promotes ER protein export while suppressing ER membrane permeabilization.
Abstract
Interorganellar protein redistribution is an emerging but underexplored aspect of proteostasis and cellular adaptation. Beyond canonical transcriptional and translational regulation, cells dynamically reprogram the spatial distribution of proteins to rapidly respond to environmental stress. This spatial plasticity enables single gene products to acquire novel, context-dependent functions on the basis of subcellular localization. Such relocalization is particularly pronounced in pathological conditions, such as cancer and viral infections, where proteome remodeling enhances cellular survival and adaptability. We previously defined endoplasmic reticulum (ER)-to-cytosol signaling (ERCYS) as a stress-responsive mechanism that alleviates ER burden by redistributing proteins into the cytosol. Despite growing interest, the molecular mechanisms driving ERCYS and related forms of spatial…
Genes, proteins, chemicals, diseases, species, mutations and cell lines named across the full text — each resolved to its canonical identifier and authoritative record.
Click any figure to enlarge with its caption.
Figure 1
Figure 2
Figure 3
Figure 4
Figure 5
Figure 6- —https://doi.org/10.13039/501100003977Israel Science Foundation
- —https://doi.org/10.13039/100001698Israel Cancer Research Fund
Peer Reviews
No public reviews on file for this paper yet. If you reviewed it on a platform where reviews are public (OpenReview, ICLR, NeurIPS, ICML), you can paste yours below so the community can read it here.
Videos
No videos yet. Explain this paper in a talk, walkthrough, or lecture? Add one.
Taxonomy
TopicsEndoplasmic Reticulum Stress and Disease · Fungal and yeast genetics research · Cellular transport and secretion
Introduction
Proteins destined for the secretory pathway are synthesized by either cytosolic ribosomes or ribosomes bound to the endoplasmic reticulum (ER) membrane and are co- or post-translationally translocated into the ER lumen. A key chaperone in this process is binding immunoglobulin protein (BIP, also known as GRP78 or HSPA5), which acts as a molecular ratchet to drive the movement of nascent polypeptides and facilitates efficient translocation of post-translationally imported proteins into the ER [1–3]. Once inside the ER, BIP promotes proper protein folding by binding newly synthesized proteins and preventing their aggregation [4]. With the assistance of other enzymes and chaperones, including ER oxidase 1 (ERO1) and protein disulfide isomerase (PDI) family members, these proteins achieve their native three-dimensional structures and undergo essential post-translational modifications [4, 5].
Perturbations in ER proteostasis result in the accumulation of misfolded proteins, leading to ER stress [6, 7]. To cope with this, cells activate the unfolded protein response (UPR), a signaling pathway that enhances ER folding capacity and restores homeostasis. In mammalian cells, the UPR is regulated by three sensors: inositol-requiring enzyme-1 (IRE1), protein kinase R (PKR)-like endoplasmic reticulum kinase (PERK), and activating transcription factor 6 (ATF6) [8–14]. Under basal conditions, BIP binds the luminal domains of IRE1, ATF6, and PERK, keeping them inactive. During stress, the accumulation of misfolded proteins titrates BIP away, assisting protein folding and leading to dissociation from these sensors, which triggers their activation [15–19]. Activated IRE1 mediates the unconventional splicing of XBP1 mRNA, producing the transcription factor XBP1s, which induces UPR target genes and increases ER folding capacity [9, 20–22]. PERK activation phosphorylates eIF2α, promoting the activation of transcription factors NRF2 and ATF4, which enhances cellular adaptation to ER stress and cellular defense against oxidative stress [11, 23–26]. ATF6, upon stress, translocates to the Golgi where it is cleaved to generate ATF6(N), a transcription factor that induces genes encoding ER chaperones and folding enzymes, including BIP [12–14, 27].
Beyond enhancing protein-folding capacity and cytoprotective functions, unresolved ER stress and prolonged activation of the UPR can trigger apoptotic cell death through multiple mechanisms, including intracellular activation of death receptor 5 [28]. While PERK activation is initially protective and essential for cell survival under mild stress conditions, its sustained activation leads to the induction of CHOP, a key transcription factor that mediates the transition from pro-survival to pro-apoptotic signaling. The PERK–ATF4–CHOP axis is well known for promoting apoptosis and has also been implicated in regulating ferroptosis [29]. Similarly, prolonged activation of IRE1α can surpass its oligomerization threshold, broadening its RNase substrate repertoire to include numerous ER-localized mRNAs, thereby promoting apoptosis. This mechanism has been demonstrated to induce cell death in pancreatic β cells [30]. In addition, persistent IRE1 kinase activity can trigger cell death through the IRE1–TRAF2–JNK signaling pathway [31]. Thus, although the UPR primarily functions to restore ER homeostasis, its prolonged and uncontrolled activation can shift the cellular response from adaptive to apoptotic outcomes.
Furthermore, the UPR reduces ER protein load by limiting substrate entry. Hyperactivation of IRE1’s kinase-regulated RNase triggers the regulated IRE1α-dependent decay (RIDD) pathway, degrading mRNAs and miRNA precursors, many of which encode secretory pathway proteins [32, 33]. In addition, XBP1/IRE1 and ATF6 induce the expression of genes encoding ER chaperones, protein-folding enzymes, and multiple components of the endoplasmic-reticulum-associated protein degradation (ERAD) machinery, including the E3 ubiquitin ligase HRD1, SEL1L, HERPUD1, and other ERAD factors [34, 35]. Together, these pathways enhance ERAD and autophagy, thereby facilitating the clearance of misfolded proteins and alleviating ER proteotoxic stress. PERK-mediated eIF2α phosphorylation also inhibits global protein translation, further limiting the influx of substrates into the ER [33].
The UPR is not the only pathway to decrease ER protein load. Over the past three decades, numerous studies have reported the presence of ER-resident proteins in the cytosolic fraction of various cell types [33, 36–53]. We and others have reported an ER stress-induced and chaperone-mediated mechanism by which proteins in the secretory pathway exit the ER to the cytosol, thereby relieving the ER of its contents [36, 37, 40, 41, 54]. We termed this mechanism “ERCYS” for ER-to-cytosol signaling [33, 36–38, 40]. ERCYS is constitutively activated in cancer cells, causing many proteins to be enriched in the cytosol of stressed cultured cancer cells, murine models of brain tumors, and human patients [37, 38, 55]. In the cytosol, the refluxed proteins acquire new pro-survival functions, thereby increasing the fitness of cancer cells [33, 36–38, 40]. It has a pathophysiological role that was identified as a nongenetic inhibitor of wt-p53 and caspase-3 activities, which was demonstrated by the reflux of a subset of ER proteins that bind and inhibit cytosolic pro-apoptotic proteins [33, 37, 55, 56]. ER and cytosolic chaperones, including DNAJB12, DNAJB14, and the cytosolic HSC70 co-chaperone SGTA, regulate ERCYS. Silencing any of these components abolishes ER protein reflux in both cancer cells and cells undergoing ER stress, underscoring their essential roles [38, 41].
Interestingly, this mechanism is not unique to proteins in the ER, but rather occurs on a more global scale, including mitochondria, peroxisomes, lysosomes, and the nucleus, under stress conditions [37, 48, 57]. Those changes are not captured by abundance and lead to a global remodeling of subcellular organization [48]. Despite that, the mechanism and the signaling pathways that regulate this phenomenon are yet to be studied.
ERCYS and the UPR are activated during ER stress and are constitutively active in patients with cancer and cancer murine models [22, 36–38, 58, 59]. However, the relationship between these two important mechanisms is still poorly understood. In this study, we uncover a previously unrecognized function of the UPR in orchestrating the spatial redistribution of ER-resident proteins, with a central role for ATF6 and IRE1 signaling pathways. IRE1 exerts dual control: facilitating ERCYS while simultaneously restricting BAX/BAK-mediated ER membrane permeabilization. Moreover, we identify BIP as a key effector of protein reflux by assembling into a stress-inducible complex with ER membrane-anchored and cytosolic co-chaperones. These findings challenge the canonical view of BIP as an import chaperone functioning as a molecular ratchet for post-translationary translocated proteins and instead establish it as a bidirectional regulator of ER protein trafficking, critical for spatial proteome remodeling during stress adaptation.
Materials and methods
Cell culture and reagents
Adenocarcinoma human alveolar basal epithelial cells (A549), human breast cancer cells (MCF-7), human embryonic kidney cells (HEK293), and mouse embryonic fibroblasts (MEFs) were cultured in Dulbecco’s modified Eagle medium (DMEM) supplemented with 10% fetal bovine serum (FBS; Gibco, no. 10270106) and 1% penicillin–streptomycin (Gibco, no. 15140122). Tunicamycin (Tm; no. 654380) was obtained from Calbiochem. Thapsigargin (Tg; no. T9033), cisplatin (Cis; no. PHR1624), etoposide (Eto; no. 341205), MKC-3946 (no. S8286), Ceapin-A7 (no. SML2330), GSK2606414 (no. 516535), and GSK2656157 (no. 504651) were purchased from Sigma-Aldrich, while AMG-18 hydrochloride (no. HY-114368A) was obtained from MedChemExpress (MCE). All UPR inhibitors were prepared as 10 mM stock solutions in DMSO; working concentrations are indicated in the figure legends.
XTT assay
The TACS XTT cell proliferation assay kit (Cat. no. 4891-025-k) was purchased from RnDsystems. In brief, 3000 cells were seeded in 96-well plates and treated with Tm and the indicated UPR inhibitors, as described in the figure legends. The XTT working solution was prepared by mixing XTT reagent with XTT activator. In total, 50 μL of the working solution was added to each well and incubated for 6 h at 37 °C in a 5% CO_2_ incubator. Absorbance was then measured at 490 nm with reference wavelengths of 630–690 nm.
RNA extraction and real-time PCR (qPCR)
Total RNA was extracted from A549, HEK293T, and MCF-7 cells using the NucleoSpin RNA Mini kit (Macherey–Nagel, no. 740955.50). Then, 500–1000 ng of the total RNA was used for cDNA synthesis using Maxima™ Reverse Transcriptase (ThermoFisher, no. EP0742).
Quantitative real-time PCR (qPCR)
qPCR was performed using the QuantStudio™ 1 Real-Time PCR System (Applied Biosystems, Thermo Fisher Scientific) in 96-well plates. Each reaction (10 µL SYBR Green [Thermofisher, no. 4344463], 1.2 µL forward primer, 1.2 µL reverse primer, 4 µL cDNA template, and 3.6 µL nuclease-free water) was performed in technical triplicates using gene-specific primers, as described previously in [60]. Thermal cycling conditions followed the standard SYBR Green protocol provided by the instrument software (QuantStudio1-Thermo Fisher Scientific). ROX dye was used as the passive reference. Gene expression levels were calculated using the comparative Ct (ΔΔCt) method and normalized to the housekeeping gene GAPDH.
Subcellular protein fractionation
Cytosolic and membranous fractions were separated as previously described [37, 61]. In brief, cells were trypsinized for 2 min and pelleted at 100 × g for 5 min at 4 °C. The resulting cell pellets were washed with ice-cold phosphate-buffered saline (PBS) and centrifuged again under the same conditions. After the second centrifugation, the pellets were resuspended in digitonin buffer (50 mM HEPES, pH 7.4; 150 mM NaCl; 10 μg/mL digitonin [0.001%]) and incubated for 10 min at 4 °C. Cells were then pelleted at 2000 × g for 5 min at 4 °C, and the supernatant was collected as the “cytosolic (digitonin)” fraction, representing the soluble, non-membrane-bound portion of the cell lysate. The remaining pellet was resuspended in NP-40 buffer (50 mM HEPES, pH 7.4; 150 mM NaCl; 1% NP-40), incubated for 30 min on ice, and centrifuged at 7000 × g for 5 min at 4 °C. The resulting supernatant was collected as the “membranous (NP-40)” fraction, representing proteins contained within or associated with membranes, including those from the plasma membrane and intracellular organelles such as the ER, mitochondria, Golgi, and endosomes.
Immunoprecipitation
Cells were harvested in co-immunoprecipitation lysis buffer (50 mM Tris/HCl pH 8, 150 mM NaCl, 0.5% TritonX100, and 1 mM EDTA). Cells were incubated on ice for 30 min, and the lysate was cleared by 10 min of centrifugation at 11,000 × g at 4 °C. Total proteins were quantified using a Pierce reagent gold BCA protein assay kit (Thermo Scientific). Equal protein amounts were taken and incubated with different primary antibodies at (1 μg Ab/1000 μg protein) at 4 °C overnight. Dynabeads protein A (Life Technologies) were added to the protein/antibodies mixture for 3 h at 4 °C with rotation. After the incubation, the beads were washed and eluted with 50 μL of Laemmli sample buffer, heated for 5 min at 70 °C, and loaded onto SDS–PAGE.
Endoglycosidase H (Endo H) assay
Endo H was purchased from New England Biolabs (NEB; no. P0702L). Following immunoprecipitation, total protein lysates or co-immunoprecipitation eluates were denatured in glycoprotein denaturing buffer at 100 °C for 10 min. Samples were then cooled to room temperature, after which Endo H GlycoBuffer and Endo H enzyme were added. Reactions were incubated at 37 °C for 1 h and subsequently analyzed by SDS–PAGE and western blotting.
Western blot
Cells were washed with ice-cold PBS and lysed in RIPA buffer (25 mM Tris–HCl, pH 7.5; 150 mM NaCl; 1% sodium deoxycholate; 0.1% SDS; and 1% NP-40). After a 30-min incubation on ice, lysates were vortexed for 10 min using a thermomixer at 4 °C and then centrifuged at maximum speed. Equal amounts of protein were separated by SDS–PAGE. Primary antibodies (1:1000 dilution) were incubated overnight at 4 °C. Following three washes, membranes were incubated with fluorescent secondary antibodies (1:10,000 dilution) for 1 h and visualized using an iBright Imaging System (Thermo Fisher). All antibodies used in this study are listed in Supplementary Table S1.
Statistical analysis
All experiments were performed in biological triplicate. Band quantification was carried out using ImageJ/Fiji software and validated against the quantification obtained from the iBright Imaging System (Thermo Fisher). Background signals were removed from all gels, and quantifications were performed consistently across experiments. The percentage of cytosolic protein was determined by dividing the intensity of the cytosolic signal by the total signal intensity of both the digitonin and NP-40 fractions. Data were analyzed using GraphPad Prism software. Unless otherwise stated, datasets were normally distributed and are presented as means ± SDs. Comparisons between groups were performed using unpaired, two-tailed Student’s t-tests. Statistical significance was defined as p < 0.05 (*p < 0.05, **p < 0.01, ***p < 0.001).
Results
UPR activation remodels the spatial redistribution of ER proteins
We initially sought to investigate a potential link between activation of the UPR signaling branches and the reflux of ER-resident proteins to the cytosol. To this end, A549 cells were treated with varying concentrations of tunicamycin (Tm), an inhibitor of N-linked glycosylation, and thapsigargin (Tg), an inhibitor of the sarco/endoplasmic reticulum Ca^2+^-ATPase (SERCA). We monitored the activation of the three principal UPR branches: IRE1, PERK, and ATF6. Subsequently, 16 h post-treatment, IRE1 activity was assessed by measuring XBP1 mRNA splicing as well as the expression levels of BLOC1S and GALNT10—two known RIDD targets [32, 62, 63]. Tm treatment induced robust activation of the IRE1/XBP1 axis starting at concentrations as low as 50 ng/mL, with maximal activation observed at concentrations exceeding 50 ng/mL (Supplementary Fig. S1A). Correspondingly, RIDD activity was confirmed under these conditions, as evidenced by a dose-dependent decrease in BLOC1S and GALNT10 mRNA levels beginning at 50 ng/mL Tm and plateauing beyond 50 ng/mL (Supplementary Fig. S1B, C).
To evaluate the PERK pathway, we measured the expression of its downstream targets HMOX1, GADD34, and CHOP [60, 64, 65]. PERK activity followed a similar activation profile, with all three genes upregulated at 50 ng/mL Tm, and expression levels plateauing at 50-100 ng/mL (Supplementary Fig. S1D–F). Likewise, activation of the ATF6 pathway was confirmed by increased expression of its target genes, BIP and HERPUD1, which were induced at 50 ng/mL Tm and reached maximal levels between 100 and 150 ng/mL (Supplementary Fig. S1G, H). Consistent activation of all three UPR branches—IRE1, PERK, and ATF6—was also observed following treatment with Tg. Concentrations as low as 15 nM Tg were sufficient to induce degradation of BLOC1S, and upregulation of CHOP and BIP, indicating coordinated UPR activation (Supplementary Fig. S1I–K). Notably, BIP, but not HERPUD1, was also upregulated at 7.5 nM Tg, suggesting differential sensitivity among ATF6 target genes (Supplementary Fig. S1I–K). These findings were also confirmed in MCF-7 cells. IRE1 activity was induced by a minimal concentration of 50 ng/mL Tm and reached a plateau at 50–100 ng/mL, as evidenced by XBP1s induction and BLOC1S mRNA degradation (Supplementary Fig. S1L, M). PERK and ATF6 pathways in MCF-7 cells exhibited a similar activation profile, with GADD34 and BIP expression induced at 50 ng/mL Tm (Supplementary Fig. S1N, O). Upon treatment with Tg, MCF-7 cells displayed comparable activation kinetics, with all three UPR branches activated at a minimal concentration of 15 nM (Supplementary Fig. S1P–S). Together, these data establish the minimal concentrations of Tm and Tg required to induce ER stress and activate the three arms of the UPR, providing a defined experimental framework for dissecting downstream stress responses.
To examine the role of UPR activation in regulating the spatial remodeling of ER-resident proteins in cancer cells, we treated cells with the above-defined concentrations of Tm (50 ng/mL) and Tg (15 nM) to activate the UPR. We compared them with sub-threshold concentrations that induced only minimal UPR activation (25 ng/mL Tm and 7.5 nM Tg) (Supplementary Fig. S1). We then quantified the cytosolic localization of a subset of ER proteins, including protein disulfide isomerase 1 (PDIA1/PDI), protein disulfide isomerase 4 (PDIA4), and DnaJ heat shock protein family (Hsp40) member B11 (DNAJB11), under the same conditions described in Supplementary Fig. S1. The digitonin-soluble fraction, containing cytosolic proteins, was extracted and compared with the NP-40-soluble fraction, representing membranous proteins. In A549 cells treated with 25 ng/mL tunicamycin, the levels of PDIA1, PDIA4, and DNAJB11 in the cytosolic fraction were undetectable (Fig. 1A–D). Treatment with 50 ng/mL Tm or higher significantly increased the reflux of PDIA1, PDIA4, and DNAJB11 into the cytosol (Fig. 1A–D). Under these conditions, the levels of cytosol-localized proteins plateaued at approximately 40% of the total signal. Treatment with 7.5 nM Tg induced a modest increase in cytosolic PDIA1 and DNAJB11 (Fig. 1E–H), while 15 nM Tg or higher further elevated their cytosolic levels to values comparable to those observed with Tm (Fig. 1E–H).Fig. 1UPR activation remodels the spatial redistribution of ER proteins. A Subcellular protein fractionation (digitonin fraction) of DNAJB11, PDI, and PDIA4 in A549 cells treated with different concentrations of Tunicamycin (Tm) (ng/mL). B–D Quantification of the refluxed ER proteins DNAJB11, PDI, and PDIA4, as in A, respectively. E Subcellular protein fractionation of DNAJB11, PDI, and PDIA4 in A549 cells treated with different concentrations of Thapsigargin (Tg) (nM). F–H Quantification of the refluxed ER proteins DNAJB11, PDI, and PDIA4, as in E, respectively. I, J Percentage of annexin-V-positive A549 and MCF7 cells (respectively) treated with tunicamycin. K, L Percentage of annexin-V-positive A549 and MCF7 cells (respectively) treated with Tg. All experiments were done in biological triplicates, with means ± SDs calculated using Prism 10 (GraphPad) (***p < 0.001, **p < 0.01, *p < 0.05)
We next assessed cell viability under these conditions. Treatment with 0–250 ng/mL Tm did not result in a significant increase in cell death after 16 h. However, higher concentrations of Tm (1000 ng/mL) caused a modest but statistically significant increase in cell death in both A549 and MCF-7 cell lines (Fig. 1I–J). Similarly, A549 cells treated with Tg showed no detectable increase in cell death up to 15 nM Tg and exhibited only a slight but significant increase at 25 nM. In contrast, MCF-7 cells showed no measurable toxicity across the 0–25 nM Tg concentration range (Fig. 1K, L). These findings indicate that subtoxic levels of ER stress are sufficient to activate UPR and promote the reflux of ER proteins to the cytosol. Moreover, low concentrations of Tm and Tg that failed to activate the UPR did not induce ER protein reflux. Together, these results demonstrate a strong correlation between UPR activation and ER protein redistribution within the cell.
IRE1’s dual role: promoting ERCYS and restricting BAX/BAK-dependent permeabilization
To investigate the role of IRE1 activation in the ER proteins spatial redistribution, we employed low concentrations of Tm and Tg that are sufficient to activate the IRE1 signaling arm of the UPR. Under these conditions, a significant cytosolic accumulation of PDIA1, PDIA4, and DNAJB11 was observed in A549 cell line (Fig. 1). Treatment with low-dose Tm and Tg induced splicing of XBP1 and a corresponding decrease in BLOC1S expression (Supplementary Figs. S1 and S2A; Fig. 2A–D). To determine whether these effects were dependent on IRE1 activity, cells were pretreated for 1 h with either AMG-18—a mono-selective IRE1α inhibitor—or MKC-3946, which specifically targets the endoribonuclease domain of IRE1 [66, 67] prior to Tm exposure for an additional 16 h. Both inhibitors efficiently suppressed XBP1 mRNA splicing and prevented BLOC1S degradation in response to Tm and Tg (Fig. 2A–D; Supplementary Fig. S2A), confirming that these compounds effectively attenuate the catalytic activity of IRE1 in A549 and MCF-7 cell lines.Fig. 2IRE1 inhibition induces an alternative route for ER-to-cytosol trafficking. A, B qPCR of relative mRNA levels of XBP1s and BLOC1S, respectively, in A549 cells treated with IRE1 inhibitors AMG-18 and MKC at a final concentration of 0.5 and 5 μM, respectively, and tunicamycin (Tm) ng/mL, N = 3. C, D qPCR of relative mRNA levels of XBP1s and BLOC1S, respectively, in A549 cells treated with IRE1 inhibitors AMG-18 and MKC at a final concentration of 0.5 and 5 μM, respectively, and different concentrations of thapsigargin (Tg) nM, N = 3. E, F qPCR of relative mRNA levels of CHOP and GADD34 in A549 cells treated with IRE1 inhibitors AMG-18 and MKC at a final concentration of 0.5 and 5 μM, respectively, and Tm, N = 3. G, H qPCR of relative mRNA levels of HERPUD1 and BIP in A549 cells treated with IRE1 inhibitors AMG-18 and MKC at a final concentration of 0.5 and 5 μM, respectively, and Tm, N = 3. I Subcellular protein fractionation of DNAJB11, PDI, and PDIA4 in A549 cells treated with different concentrations of Tm (ng/mL), Tg (nM), and the IRE1 inhibitors AMG-18 and MKC at a final concentration of 0.5 and 5 μM, respectively. J–L Quantification of the refluxed ER proteins DNAJB11, PDI, and PDIA4, as in I, respectively. M Representative immunoblot showing the interaction between SGTA and PDIA1 (PDI) in A549 cells treated with IRE1 inhibitors AMG-18 and MKC at a final concentration of 0.5 and 5 μM, respectively, and Tm 50 ng/mL. N Subcellular protein fractionation of DNAJB11 and PDIA4 in DNAJB12/14-silenced A549 cells treated with Tm and the IRE1 inhibitor AMG-18 at a final concentration of 0.5 μM. O, P Quantification of the refluxed ER proteins DNAJB11 and PDIA4 as in N, respectively. All experiments were performed in biological triplicates, with means ± SDs calculated using Prism 10 (GraphPad) (***p < 0.001, **p < 0.01, *p < 0.05)
Despite their strong inhibitory activity toward IRE1, we examined whether the IRE1 inhibitors might also affect other branches of the UPR. To this end, we performed qPCR analyses to directly assess UPR activation during IRE1 inhibition by AMG-18 or MKC-3946. MKC-3946 modestly induced the PERK but not the ATF6 pathway under basal conditions, as evidenced by increased CHOP and GADD34 expression; however, no significant changes were observed following tunicamycin treatment, indicating that MKC-3946 does not inhibit PERK or ATF6 activities (Fig. 2E–H). Under the same experimental conditions, AMG-18 had no detectable effect on either the ATF6 or PERK pathways (Fig. 2E–H), consistent with previous reports that AMG-18 (KIRA-8) selectively targets IRE1 without influencing the other UPR branches [68, 69].
We next assessed the levels of cytosolically localized ER-resident proteins under the conditions described above, in the presence of IRE1 inhibitors. Treatment of A549 and MCF-7 cells with MKC-3946 or AMG-18 alone, in the absence of ER stress inducers, resulted in increased cytosolic levels of DNAJB11, PDIA1, and PDIA4 (Fig. 2I–L; Supplementary Fig. S2B–E). Subsequent addition of Tm and Tg further enhanced the cytosolic accumulation of these proteins. Notably, IRE1 inhibition did not extensively affect the extent of ER protein reflux in cells treated with Tm or Tg (Fig. 2I–L; Supplementary Fig. S2B–E). These results suggest that IRE1 may negatively regulate ER protein reflux under basal conditions, and that its inhibition promotes increased localization of ER proteins to the cytosol.
To further validate these findings and strengthen our conclusions, we examined protein reflux in IRE1 knockout (IRE1^−/−^) mouse embryonic fibroblasts (MEFs) compared with wild-type (IRE1^+/+^) controls. IRE1^−/−^ MEFs exhibited elevated basal cytosolic levels of DNAJB11, PDIA1, and PDIA4 even in the absence of ER stress inducers (Supplementary Fig. S2F–I). Treatment with Tm and Tg led to a further increase in cytosolic localization of these proteins. Although inhibition of IRE1 alone altered ER protein distribution, it did not affect the extent of reflux triggered by ER stress (Supplementary Fig. S2F–I). Since both MKC-3946 and AMG-18 induced spatial redistribution of ER proteins, and only MKC-3946 activated the PERK pathway (Fig. 2E–H), we conclude that this phenomenon is independent of the PERK activation observed under these conditions. Collectively, these results suggest that IRE1 activity plays a role in maintaining ER protein compartmentalization under basal conditions, potentially serving as a protective mechanism to prevent uncontrolled leakage of ER proteins into the cytosol.
ERCYS is facilitated by chaperones from both the ER and the cytosol, and relies critically on the interaction between the cytosolic HSC70 co-chaperone SGTA and refluxed ER-resident proteins [37, 38, 41]. Furthermore, silencing SGTA or the ER-associated co-chaperones DNAJB12 and DNAJB14 significantly impairs ER protein reflux and reduces the cytosolic accumulation of multiple ER proteins [37, 38, 41]. To determine whether IRE1 inhibition promotes ER protein reflux through an ERCYS-dependent mechanism in conditions without Tm and Tg, we examined the role of SGTA in this process. A549 cells were transfected with SGTA-targeting siRNA and treated with Tm, AMG-18, or a combination of both. In control cells, Tm treatment led to increased cytosolic levels of DNAJB11 and PDIA4, whereas SGTA silencing markedly reduced cytosolic DNAJB11 and PDIA4 levels during ER stress (Supplementary Fig. S2J–M), consistent with SGTA-dependent ERCYS. AMG-18 treatment alone elevated DNAJB11 and PDIA4 levels in the cytosol, with further increases observed upon co-treatment with Tm. SGTA silencing did not reduce the cytosolic levels of these proteins in the presence of AMG-18 (Supplementary Fig. S2J–M), suggesting that IRE1 protects against spatial remodeling of ER proteins independently of SGTA and thus via an ERCYS-independent mechanism.
To further validate this conclusion, we performed SGTA immunoprecipitation and assessed its interaction with the refluxed protein PDIA1. Although PDIA1 was detectable in the cytosolic fraction of AMG-18- or MKC-3946-treated cells, no interaction with SGTA was observed under these conditions (Fig. 2M). In contrast, Tm-treated cells showed robust PDIA1–SGTA interaction, confirming that under ER stress, PDIA1 reflux is mediated through SGTA-dependent ERCYS (Fig. 2M). In order to validate that the SGTA–PDIA1 interaction is due to the reflux of PDIA1 from the ER and not as a result of problems in targeting proteins to the ER, we used a glycosylated version of PDIA1 harboring a glycosylation site close to the C-terminus of the protein to ensure full entry to the ER. Under those conditions we observed that SGTA binds glycosylated PDIA1, suggesting that the ER proteins were ER-localized prior to their reflux and interaction with SGTA (Supplementary Fig. S2N). Furthermore, silencing both DNAJB12 and DNAJB14 led to similar results as the ones found with SGTA silencing (Fig. 2N–P; Supplementary Fig. S2O). These findings suggest that IRE1 inhibition induces an alternative pathway for ER protein redistribution that operates independently of SGTA and DNAJB12/14, and is therefore likely distinct from the canonical ERCYS mechanism. This supports the existence of an IRE1-regulated, DNAJB12/14/SGTA-independent route by which ER proteins relocalize to the cytosol under non-stress conditions.
Given that inhibition of IRE1 led to increased cytosolic accumulation of ER proteins in DMSO-treated cells, we next investigated whether IRE1 overexpression could protect against ER protein reflux. Elevated levels of IRE1α within the ER promote spontaneous oligomerization, bringing the cytosolic kinase domains into close proximity and enabling autophosphorylation, which subsequently activates the RNase domain [30, 70]. Overexpression of wild-type IRE1 (IRE-WT) in A549 and HEK293 cells enhanced XBP1 splicing and reduced BLOC1S and GLANT1 expression, consistent with activation of the IRE1 branch of the UPR (Supplementary Fig. S3A–F). In contrast, overexpression of a dominant-negative IRE1 construct, comprising the luminal and transmembrane domains of IRE1 fused to NCK1 (IRE1-DN) [68], failed to induce XBP1 splicing or activate the IRE1 arm of the UPR (Supplementary Fig. S3A–F). IRE1-DN also did not activate this pathway under ER stress conditions induced by tunicamycin (Supplementary Fig. S3D–F). Under these conditions, HEK293T cells exhibited ER stress response kinetics comparable to those observed in A549 cells following tunicamycin treatment (Supplementary Figs. S1 and S3I–L).
IRE1 overexpression selectively activated the IRE1 signaling branch without affecting activation of the ATF6 or PERK pathways, nor did it inhibit their induction (Supplementary Fig. S3A–L). Moreover, when combined with ER stress inducers, IRE1 overexpression resulted in higher BIP mRNA levels than either IRE1 overexpression or tunicamycin treatment alone, suggesting a potential regulatory role for IRE1 in modulating BIP expression (Supplementary Fig. S3F–H).
Overexpression of IRE1-WT alone was sufficient to promote cytosolic accumulation of DNAJB11, PDI, and PDIA4 (Fig. 3A–D; Supplementary Fig. S3L–P). In contrast, overexpression of IRE1-DN failed to induce protein relocalization to the cytosol and exerted a dominant-negative effect by suppressing ER protein reflux even under Tm-induced stress conditions (Fig. 3A–D; Supplementary Fig. S3L–P).Fig. 3IRE1 overexpression is sufficient to activate ERCYS. A Representative immunoblot of DNAJB11, PDI, and PDIA4 in the cytosolic-digitonin and membranous-NP40 fractions in A549 cells overexpressing IRE1-WT versus IRE1-DN mutant. B–D Quantification of DNAJB11, PDI, and PDIA4 in the cytosolic fraction as shown in A. E Representative immunoblot of DNAJB11 and PDIA4 in the cytosolic-digitonin and membranous-NP40 fractions in cells overexpressing IRE1-WT versus IRE1-DN mutant after silencing SGTA and DNAJB12/14. F, G Quantification of DNAJB11 and PDIA4 in the cytosolic-digitonin fraction as shown in E. H Representative immunoblot showing the interaction between SGTA, DNAJB11, and PDIA1 in A549 cells overexpressing IRE1-WT and IRE1-DN. I Subcellular protein fractionation of DNAJB11 and PDIA4 in BAX/BAK^+/+^ and BAX/BAK^−/−^ MEFs. J, K Quantification of the refluxed ER proteins DNAJB11 and PDIA4 as shown in I, respectively. All experiments were performed in biological triplicates, with means ± SDs calculated using Prism 10 (GraphPad) (***p < 0.001, **p < 0.01, *p < 0.05)
Given that both IRE1 inhibition and activation promote ER-to-cytosol protein reflux, we hypothesized that IRE1 regulates ER protein redistribution through two distinct mechanisms. To test this, we silenced SGTA, DNAJB12, and DNAJB14, and subsequently overexpressed either IRE1-WT or IRE1-DN. In control cells, overexpression of IRE1-WT significantly increased the cytosolic levels of DNAJB11 and PDIA4, whereas IRE1-DN had no effect despite the same expression levels of the WT and DN in those cells (Fig. 3E–G; Supplementary Fig. S3L). In contrast, IRE1-WT overexpression failed to induce cytosolic accumulation of these proteins in SGTA-silenced cells, indicating that SGTA is required for IRE1-driven ERCYS (Fig. 3E–G). A similar effect was observed upon silencing DNAJB12 and DNAJB14, further supporting the dependency of this process on these ER-resident co-chaperones (Fig. 3E–G). Together, these findings demonstrate that IRE1 activation is sufficient to promote ER protein reflux through a mechanism that requires SGTA as well as DNAJB12 and DNAJB14, highlighting a chaperone-mediated, ERCYS-dependent pathway regulated by IRE1 activity.
To further investigate the role of SGTA in this process, we examined its association with refluxed ER proteins in cells overexpressing IRE1. SGTA co-immunoprecipitated with PDIA1 and PDIA4 in cells expressing wild-type IRE1 (IRE1-WT), but not in those expressing the dominant-negative mutant (IRE1-DN) (Fig. 3H). This pattern mirrors the interaction observed in cells subjected to Tm-induced ER stress (Fig. 2M; Supplementary Fig. S2N). These findings indicate that SGTA-dependent ERCYS requires active IRE1 signaling and that IRE1-mediated ER protein reflux involves a functional interaction between SGTA and refluxed ER-resident proteins, similar to that observed under ER stress conditions.
Collectively, our results suggest that both IRE1 inhibition and hyperactivation enhance ER protein reflux, albeit through distinct mechanisms. When IRE1 activity is suppressed, the chaperone-mediated ERCYS pathway becomes nonfunctional, implying that IRE1 signaling is essential for this process. In its absence, alternative mechanisms—such as ER membrane permeabilization mediated by the pro-apoptotic proteins BAX and BAK [45]—may promote the redistribution of ER proteins to the cytosol.
To test this hypothesis, we analyzed the effect of IRE1 inhibition in MEF cells lacking BAX and BAK. Wild-type and BAX/BAK double-knockout MEFs were treated with Tm and Tg in the presence or absence of AMG-18, and ERCYS was subsequently assessed. In wild-type MEFs, AMG-18 treatment led to pronounced ERCYS and cytosolic accumulation of ER-resident proteins. In contrast, BAX/BAK-deficient cells failed to accumulate ER proteins in the cytosol upon AMG-18 treatment, and cytosolic ER protein levels remained reduced even under Tm-induced stress (Fig. 3I–K).
These findings suggest that IRE1 exerts a dual regulatory role—facilitating ERCYS while simultaneously preventing ER membrane permeabilization and uncontrolled protein leakage through BAX/BAK. Accordingly, ERCYS and ER membrane permeabilization may represent two opposing mechanisms of ER protein redistribution, both modulated by IRE1 activity. In this model, IRE1 inhibition suppresses ERCYS but promotes BAX/BAK-dependent ER membrane permeabilization.
ERCYS activity is alienated from PERK activity.
Next, we investigated whether inhibition of PERK influences ER protein reflux. Cells treated with Tm or Tg were pretreated with either GSK2656157 or GSK2606414, both selective inhibitors targeting the kinase domain of PERK [71, 72]. Pretreatment with these inhibitors reduced the expression of PERK downstream targets—CHOP, GADD34, and HMOX1—during ER stress, confirming effective suppression of PERK signaling (Supplementary Fig. S4A–F). Treatment with GSK2656157 or GSK2606414 alone did not alter the spatial distribution of PDIA1, PDIA4, or DNAJB11, in contrast to IRE1 inhibition (Supplementary Fig. S4G–J). As expected, addition of Tm or Tg to control cells increased the cytosolic levels of these ER-resident proteins. Pretreatment with GSK2656157 or GSK2606414 did not affect the extent of cytosolic accumulation, as the levels of PDIA1, PDIA4, and DNAJB11 were comparable to those in untreated control cells (Supplementary Fig. S4G–J). This lack of effect is not due to inhibition or activation of the other UPR branches, as PERK inhibitors selectively target PERK without impacting IRE1 or ATF6 signaling (Supplementary Fig. S4K–L). These results suggest that, in contrast to IRE1 inhibition, suppression of PERK signaling does not impact ER protein reflux, indicating that PERK may not play a major role in regulating ER protein redistribution.
ATF6 promotes ERCYS and protects against DNA damage during ER stress
Next, we tested whether Ceapin-A7 (an ATF6 inhibitor [73]) can affect the reflux of protein to the cytosol. Treating cells with Ceapin-A7 inhibits the upregulation of ATF6-regulated genes BIP and HERPUD1 in cells treated with Tm and Tg (Supplementary Fig. S5A–D). Treatment with Ceapin-A7 selectively inhibits the ATF6 branch of the UPR without affecting PERK or IRE1 signaling under both stressed and non-stressed conditions (Supplementary Fig. S5E, F). Unlike the IRE1 inhibitors, pretreatment with Ceapin-A7 alone was insufficient to reflux ER proteins to the cytosol without stress (Fig. 4A–D; Supplementary Fig. S5G-J). However, during ER stress induced by Tm or Tg, Ceapin-A7 effectively decreased the cytosolic accumulation of DNAJB11, PDIA1, and PDIA4, maintaining their levels comparable to those observed in DMSO-treated cells (Fig. 4A–D; Supplementary Fig. S5G–J).Fig. 4. Disruption of ATF6 signaling abrogates ER protein reflux to the cytosol. A Representative immunoblot of DNAJB11, PDI, and PDIA4 in the in the cytosolic-digitonin and membranous-NP40 fractions in A549 cells treated with DMSO, tunicamycin (ng/mL), and thapsigargin (nM) in the presence and absence of Ceapin-A7 at a final concentration of 3.5 μM. B–D Quantification of DNAJB11, PDI, and PDIA4 in the cytosolic fraction as shown in A. E Representative immunoblot of DNAJB11, PDI, and PDIA4 in the cytosolic and membrane fractions of ATF6^+/+^ and ATF6^−/−^ MEF cells treated with tunicamycin and thapsigargin. F–H Quantification of DNAJB11, PDI, and PDIA4 in the cytosolic fraction as shown in E. I wt-p53 transcriptional activity in A549 cells treated with DMSO, etoposide, and tunicamycin in the presence of the ATF6 inhibitor Ceapin-A7 at a final concentration of 3.5 μM. J XTT assay of A549 cells treated with etoposide and tunicamycin in the presence of the ATF6 inhibitor Ceapin-A7 at a final concentration of 3.5 μM. K Representative immunoblot showing the interaction between AGR2 and wtp53 in A549 cells in the presence and absence of Ceapin-A7 and ER stress. All experiments were conducted in biological triplicates, with means ± SDs calculated using Prism 10 (GraphPad) (***p < 0.001, **p < 0.01, *p < 0.05)
To further validate these findings, we examined the cytosolic levels of PDIA4, PDIA1, and DNAJB11 in ATF6-knockout (ATF6^−/−^) and wild-type (ATF6^+/+^) MEF cells. While Tm and Tg treatment promoted ER protein reflux in wild-type cells, this response was completely abrogated in ATF6-deficient cells (Fig. 4E–H), further supporting a critical role for ATF6 in regulating ER protein redistribution during stress.
Previously, we found that ERCYS is associated with the acquisition of pro-survival functions by cytosolic AGR2, which exits the ER and binds to wild-type p53 in the cytosol, leading to its inhibition [37]. In addition to its interaction with p53, cytosolic AGR2 has been reported to perform several other functions, including binding to TRAF3 and associating with cytoplasmic microtubules during viral infection [52, 53]. To test whether Ceapin-A7 inhibits this gain of function, we treated cells with cisplatin or etoposide to increase wt-p53 activity under different conditions. In the absence of ER stress, cisplatin and etoposide increased the activity of wt-p53 (Fig. 4I). Adding Tm or Tg before adding cisplatin and etoposide decreased wt-p53 activity (Fig. 4I). The inhibition of ATF6 with Ceapin-A7 significantly restored wt-p53 transcriptional activity during ER stress, compared with cells treated with ER stress alone, in the presence of DNA damage agents. Thus, ATF6 signaling is important for AGR2 spatial distribution and the inhibition of proapoptotic pathways in the cytosol.
We next asked whether ATF6 inhibition affects cell survival in cells treated with etoposide. To investigate this, we treated A549 cells with etoposide in the presence of Tm and Ceapin-A7. Treatment with etoposide decreased the viability of cancer cells in the control condition (Fig. 4J). Pretreatment cells with subtoxic Tm concentrations partially rescued cells and increased their viability (Fig. 4J). Pretreatment with Ceapin-A7 before Tm decreased cell viability compared with cells pretreated with Tm alone (Fig. 4J). Mechanistically, AGR2 binds the cytosolically localized wt-p53 during ER stress and inhibits its activity in an ATF6-dependent manner, as this interaction is abrogated when cells are treated with Ceapin-A7 (Fig. 4K).
Collectively, these findings reveal a previously unrecognized role for the UPR in regulating protein redistribution in eukaryotic cells. Specifically, they demonstrate that ATF6 signaling is essential for the translocation of ER-resident proteins to the cytosol, likely through the modulation of ERCYS activity. These results underscore the critical contribution of the ATF6 pathway to ER protein reflux and highlight the broader importance of the UPR in orchestrating spatial proteome remodeling under stress conditions.
BIP facilitates ER protein reflux via a DNAJB12/14–SGTA complex
Given that BIP is a shared transcriptional target of both IRE1 and ATF6, and is capable of functioning in multiple cellular compartments [50, 51], we investigated its potential role in the spatial redistribution of ER proteins. To this end, BIP was silenced in A549 cells, and the cytosolic accumulation of AGR2, DNAJB11, PDIA1, and PDIA4 was assessed under ER stress. In control cells treated with scrambled siRNA, ER stress induced a marked redistribution of these proteins into the cytosol (Fig. 5A–E). In contrast, BIP knockdown significantly reduced their cytosolic presence, restoring levels comparable to those in DMSO-treated cells (Fig. 5A–E). These results identify BIP as an essential effector of ERCYS and implicate it as a regulator of stress-induced spatial proteome remodeling.Fig. 5ER-luminal BIP promotes ERCYS through a DNAJB12/14–SGTA-dependent reflux complex. A Representative immunoblot of AGR2, DNAJB11, PDI, and PDIA4 in the cytosolic and membrane fractions of cells treated with tunicamycin (Tm; ng/mL) and thapsigargin (Tg; nM) in A549 BIP-silenced cells. B–E Quantification of AGR2, DNAJB11, PDI, and PDIA4 in the cytosolic fraction as shown in A. F Representative immunoblot of DNAJB11, PDI, and PDIA4 in the cytosolic and membrane fractions of BIP-silenced A549 cells expressing FLAG–DNAJB12-WT or FLAG–DNAJB12-H139Q mutant. G–I Quantification of DNAJB11, PDI, and PDIA4 in the cytosolic fraction as shown in F. J Representative immunoblot showing the interaction between BIP, DNAJB12, DNAJB14, and SGTA in A549 cells treated with Tm and Tg. K Representative immunoblot showing the interaction between BIP and FLAG–DNAJB12-WT or DNAJB12-H139Q in A549 cells. L Representative immunoblot showing the interaction between AGR2 and wild-type p53 in A549 cells overexpressing IRE1-WT or the IRE1 dominant-negative (DN) variant following BIP silencing, compared with control cells. All experiments were done in biological triplicates, with means ± SDs calculated using Prism 10 (GraphPad) (***p < 0.001, **p < 0.01, *p < 0.05)
To determine whether BIP facilitates ER protein reflux through its interaction with DNAJB12 and DNAJB14, we silenced BIP in A549 cells and overexpressed wild-type (WT) DNAJB12, which is known to promote ERCYS independently of UPR activation (Supplementary Fig. S6A–C). DNAJB12 contains a cytosol-facing J-domain that enables interaction with HSP70 family chaperones at the cytosolic interface of the ER membrane. The highly conserved HPD motif within this J-domain (comprising His-Pro-Asp) is preserved across all DNAJB12 homologs and is essential for the interaction between DNAJ co-chaperones and HSP70/HSC70 [38]. To assess the functional significance of this motif, we generated a point mutation substituting HPD with QPD at position 139 (H139Q), thereby disrupting its co-chaperone activity and capacity to recruit HSC70. The H139Q-DNAJB12 mutant, expressed at levels comparable to the WT protein, was unable to induce ER protein reflux [38].
As expected, overexpression of WT DNAJB12 was sufficient to drive cytosolic accumulation of DNAJB11, PDI, and PDIA4 compared with cells transfected with an empty vector or the HPD mutant (Fig. 5F–I). The HPD mutation completely abolished ER protein reflux, demonstrating that this process strictly depends on the integrity of the J-domain HPD motif. Furthermore, BIP silencing suppressed the cytosolic accumulation of ER-resident proteins, even in cells overexpressing DNAJB12 (Fig. 5F–I). Although UPR activation was evident in BIP-silenced cells, this did not enhance cytosolic accumulation in DNAJB12-overexpressing cells; instead, the amount of refluxed protein was markedly reduced upon BIP knockdown in DNAJB12-WT cells (Fig. 5F–I; Supplementary Fig. S6D–G). Collectively, these results indicate that BIP is essential for mediating ER-to-cytosol reflux under conditions that do not activate the UPR or cause ER stress, supporting a noncanonical, signaling-independent role for BIP in this chaperone-driven process of spatial protein redistribution.
We next examined whether BIP promotes ER protein redistribution by directly engaging the DNAJB12/DNAJB14/SGTA complex at the ER membrane. To this end, we performed BIP co-immunoprecipitation under basal and ER stress conditions. In DMSO-treated cells, BIP showed no detectable association with DNAJB12, DNAJB14, or SGTA. In contrast, upon ER stress induction, BIP robustly associated with all three components, indicating stress-dependent assembly of a functional reflux complex (Fig. 5J).
BIP interaction with the cytosolic co-chaperone SGTA suggests that this complex is assembled on the cytosolic side of the ER membrane (Fig. 5J). These findings were further supported using cells overexpressing wild-type (WT) or H139Q mutant DNAJB12. Overexpression of WT-DNAJB12 increased ER protein reflux and enhanced the interaction between DNAJB12 and BIP even under non-stress conditions (Fig. 5K). In contrast, this interaction was abolished in the H139Q mutant, indicating that an intact HPD motif is required for DNAJB12–BIP interaction (Fig. 5K).
Consistent with these observations, BIP associated with DNAJB12 under ER stress conditions (Fig. 5J) and co-immunoprecipitated with cytosolic SGTA, further supporting a cytosolic localization of the BIP–DNAJB12/DNAJB14/SGTA complex. Together, these data demonstrate that BIP engages the DNAJB12/DNAJB14/SGTA complex in an HPD motif-dependent manner to promote ER protein reflux during ER stress.
Given that the HPD motif of DNAJB12 is oriented toward the cytosol, we hypothesized that BIP must interact with DNAJB12 from the cytosolic face of the ER membrane. To test whether a cytosolic form of BIP can bind the cytosol-facing J-domain of DNAJB12 and substitute for ER-localized BIP in promoting ER protein reflux, A549 cells were co-transfected with either wild-type BIP containing its ER signal peptide or a signal-peptide-truncated BIP variant that localizes exclusively to the cytosol. Expression of cytosolic BIP alone did not alter the cytosolic levels of PDI or PDIA4. In contrast, overexpression of ER-localized BIP modestly increased the cytosolic levels of DNAJB11, PDIA1, and PDIA4 (Supplementary Fig. S6H–N).
These data indicate that BIP must originate from the ER in order to engage the cytosol-facing J-domain of DNAJB12. We propose that BIP undergoes ER-to-cytosol relocalization, where it directly binds the DNAJB12 HPD motif and associates with the cytosolic co-chaperone SGTA. Together, these findings suggest that simple cytosolic expression of ER proteins is insufficient, and that an organized, regulated exit from the ER is required for their functional engagement in the cytosol.
Moreover, when IRE1-WT was induced in control-scrambled cells, a significant amount of PDI and PDIA4 underwent reflux. This reflux was abolished in BIP-silenced A549 cells, indicating that BIP is both necessary and sufficient for driving ER protein reflux during ER stress, whether triggered by, Tm/Tg, IRE1 overexpression, or DNAJB12 overexpression (Fig. 5F–I; Supplementary Fig. S6H–N).
Together, our results reveal a noncanonical role for BIP in mediating ER protein export through a chaperone-guided, directional mechanism that requires both its ER origin and interaction with the DNAJB12/14–SGTA complex. This expands BIP’s functional repertoire from its classical role in protein import and UPR regulation to a bidirectional regulator of ER protein trafficking.
Finally, we examined the interaction between AGR2 and wild-type p53 by inducing IRE1-WT, which promotes ER protein reflux to the cytosol (Fig. 3; Supplementary Fig. S3). A549 cells overexpressing IRE1-WT, but not the dominant-negative mutant, exhibited an increased interaction between AGR2 and wild-type p53. This interaction was abolished upon BIP silencing (Fig. 5L), indicating that IRE1-driven ER protein reflux and the resulting AGR2–wt-p53 interaction are BIP-dependent. BIP silencing significantly restored wild-type p53 activity under tunicamycin-induced stress and DNA damage conditions, consistent with its inhibitory effect on AGR2 reflux and on the AGR2–wt-p53 interaction in the cytosol (Fig. 5A; Supplementary Fig. S6O). Together, these findings establish BIP as a critical effector of ERCYS and suggest that its inhibition—alone or in combination with ATF6 blockade—may impair the gain-of-function activities of refluxed proteins such as AGR2, thereby sensitizing cancer cells to apoptosis.
Discussion
Spatial protein remodeling is a key adaptive response of eukaryotic cells to various stressors. Recent studies have shown that interorganellar protein redistribution is a prominent feature during viral infection, with the ER exhibiting the most extensive remodeling [48]. Over the past decades, numerous ER-resident proteins have been detected at non-ER locations, where they modulate diverse signaling pathways, including those involving caspase-3/7, HIF1α, STAT3, wild-type p53, and TRAF3 [33, 36–52]. Such non-genetic relocalization events contribute to disease pathogenesis by conferring new, often context-dependent functions on the mislocalized proteins, as observed in viral infections [52, 74, 75], cancer [38, 55, 76, 77], and NGLY1 deficiency [78, 79]. Moreover, DNAJB12, DNAJB14, and SGTA have been implicated in neurodegenerative diseases [80]. In this study, we identify a mechanistic link between spatial protein remodeling and the relatively underexplored ATF6 branch of the UPR in cancer cells. We further reveal that BIP acts as a critical factor assembling a protein complex at the ER membrane to mediate the relocation of ER proteins to the cytosol through a mechanism independent of its canonical ratcheting function.
Under conditions where low concentrations of tunicamycin failed to activate the UPR, ER protein localization remained unchanged. In contrast, low concentrations of thapsigargin caused a modest induction of BIP expression, accompanied by a slight cytosolic accumulation of ER-resident proteins (Fig. 1E–H; Supplementary Fig. S1K). Treatment with mild, subtoxic concentrations that partially activated the UPR resulted in detectable reflux of ER proteins into the cytosol. At 7.5 nM Tg—a concentration insufficient to trigger full UPR activation—BIP was the only UPR marker upregulated. Despite the lack of broad UPR induction, partial cytosolic relocalization of DNAJB11 and PDIA1 was observed (Fig. 1E–G). These findings suggest that ATF6/BIP signaling contributes to ER protein redistribution even in the absence of full UPR engagement.
Both IRE1 knockout and chemical inhibition of IRE1 led to a pronounced accumulation of ER proteins in the cytosol, even under non-stress conditions, indicating that IRE1 may play a critical role in preventing BAX/BAK-dependent ER membrane permeabilization and subsequent protein leakage. Although PERK was substantially activated in MKC-3946-treated cells, we concluded that PERK does not contribute to ERCYS induction. Because both IRE1 inhibitors, MKC-3946 (which leads to PERK activation) and AMG-18 (which does not), produced similar phenotypes, the effects observed with MKC-3946 are unlikely to be mediated by PERK activation.
Conversely, IRE1 overexpression also increased cytosolic levels of ER proteins, revealing a dual role for IRE1. In its absence, ER membrane permeabilization appears unrestrained, promoting protein leakage and potentially competing with regulated export mechanisms such as ERCYS. When overexpressed, however, IRE1 enhances ER-to-cytosol protein reflux through an SGTA-dependent pathway that operates independently of ATF6 or PERK activation. This regulated export (ERCYS) is associated with cytoprotective outcomes by limiting pro-apoptotic signaling in the cytosol. Two distinct mechanisms mediate ER protein reflux: one driven by BAX/BAK in the absence of IRE1 activity, and another through the IRE1-dependent ERCYS pathway. In BAX/BAK knockout cells, inhibition of IRE1 further reduces reflux, as both pathways are disrupted, BAX/BAK-mediated release is absent, and ERCYS is partially blocked by IRE1 inhibition. Collectively, these findings suggest that IRE1 orchestrates cell fate by simultaneously suppressing pro-death ER membrane permeabilization and promoting adaptive ER protein export via ERCYS.
Under ER stress, ERCYS is active. In BAX/BAK-deficient cells, inhibition of IRE1 further reduces protein reflux, as both pathways are compromised: BAX/BAK-dependent reflux is absent, and ERCYS is partially suppressed by IRE1 inhibition. Consistently, dominant-negative IRE1 inhibited ERCYS but did not induce BAX/BAK-dependent ER membrane permeabilization, despite effective blockade of IRE1 signaling. Thus, the relative contributions of canonical and noncanonical IRE1 functions to BAX/BAK-dependent processes remain unresolved. How BAX and BAK are activated or recruited in the absence of functional IRE1 is an important question for future studies. In addition, whether proteins exiting the ER via SGTA-independent, non-ERCYS pathways acquire distinct cytosolic functions linked to cell death remains to be determined.
Suppressing PERK does not affect ER protein reflux to the cytosol. However, ATF6 inhibition with Ceapin-A7 does prevent ER protein reflux. This led to the attenuation of ERCYS and subsequently increased activity of wt-p53 during ER stress. This is evident by the decreased proliferation index under these conditions. Hence, ATF6 is important for the activation of ERCYS, leading to the inhibition of wt-p53 and thus preventing cell death. Targeting ATF6 could be a promising strategy for enhancing DNA-damage-related cell death by preventing the inhibition of p53 during ERCYS, thereby sensitizing cancer cells to DNA-damaging drugs.
Although ER protein reflux has been described in both yeast and mammalian systems, important mechanistic differences exist between these organisms. Yeast lack ATF6, and IRE1 constitutes the sole arm of the UPR. Consistent with this simplified UPR architecture, genetic ablation of either IRE1 or its downstream transcription factor HAC1 (the functional homolog of XBP1) results in enhanced ER protein reflux compared with wild-type cells. Interestingly, loss of HAC1 produces a more pronounced reflux phenotype than loss of IRE1, suggesting that IRE1-dependent but HAC1-independent mechanisms may also contribute to restraining ER protein reflux in yeast. These findings indicate that, in yeast, the UPR regulates ER-to-cytosol reflux through pathways that differ from those operating in mammalian cells. At the same time, the requirement for IRE1 in limiting ER protein reflux highlights a conserved role for this sensor across species, underscoring both evolutionary continuity and divergence in the regulation of ER proteostasis.
These observations suggest that ATF6 activation, as well as forced IRE1 activation by overexpression, may regulate distinct downstream effectors or parallel signaling pathways that converge on ERCYS. BIP, a shared downstream target of both the IRE1/XBP1 and ATF6 arms of the UPR [81], was shown previously to act as a molecular ratchet facilitating protein translocation across the ER membrane. Proteins that undergo post-translational translocation are first synthesized in the cytosol and subsequently delivered to the ER membrane, where their entry into the ER lumen is facilitated by the Sec62/63 complex. Once inside, BIP binds to the translocating polypeptides, acting as a molecular ratchet that prevents their backsliding through the translocon and ensures their movement into the ER [3, 82]. This mechanism, which is evolutionarily conserved from yeast to humans, is essential for the efficient import and folding of small, secretory proteins such as hormones and growth factors [1, 3, 82]. In contrast to the majority of proteins that enter the ER through cotranslational translocation, post-translational translocation is limited to a small subset of proteins encoding small secreted peptides and hormones (Fig. 6). In this study, we show that under ER stress conditions, BIP interacts with the ER membrane-associated proteins DNAJB12/14 and the cytosolic co-chaperone SGTA, forming a complex required for protein reflux from the ER to the cytosol [38]. We further demonstrate direct physical interactions between BIP and the co-chaperones DNAJB12, DNAJB14, and SGTA in cells overexpressing wild-type DNAJB12, but not the J-domain–deficient H139Q mutant. These findings support a previously unrecognized, signaling-independent function of BIP as a core molecular component that directly facilitates the reflux mechanism through its interactions with these co-factors. Although BIP depletion activates the UPR (Supplementary Fig. S6D–F), protein reflux is markedly impaired in the absence of BIP, indicating that UPR activation alone is insufficient to drive reflux without BIP. This underscores BIP as a critical effector required for the reflux process itself. Recent studies have shown that BIP stabilizes DNAJB12 in cells treated with the reducing agent dithiothreitol [83], suggesting that BIP may promote ERCYS under stress by maintaining DNAJB12 stability. Moreover, BIP has previously been identified as part of the DJANGOS complex, which facilitates the retrotranslocation of simian virus 40 (SV40) from the ER to the cytosol [74]. This mechanistic parallel suggests that ERCYS may reflect a broader, endogenous trafficking pathway co-opted by viruses for cytosolic entry and is regulated by BIP and other cytosolic and ER chaperones.Fig. 6. Working model. Schematic illustrating the dual role of BIP as both a molecular chaperone and a molecular ratchet. A Normal conditions: BIP binds and assists in the folding of unfolded or misfolded proteins within the ER and facilitates their proper translocation into the ER lumen. B ER stress: BIP may translocate to the cytosol via an as-yet-undefined mechanism, where it interacts with the cytosol-facing HPD motif of DNAJB12 and the cytosolic HSC70-co-chaperone SGTA to drive ER protein reflux through an IRE1- and ATF6-dependent pathway. In parallel, IRE1 promotes ERCYS-mediated reflux while protecting against BAX/BAK-dependent ER membrane permeabilization. AGR2, an ER-derived protein, can gain cytosolic functions following reflux by forming inhibitory interaction with p53. Green arrows indicate activation, and red arrows indicate inhibition. The precise cytosolic localization of BIP remains to be definitively established, and alternative models, such as membrane-spanning interactions, cannot be excluded
In conclusion, we define a noncanonical, signaling-independent function of the ER-resident chaperone BIP as a central mediator of ER protein reflux. While BIP is classically viewed as essential for protein import into the ER and regulation of UPR sensors, our findings demonstrate that under stress, BIP forms a functional complex with ER membrane-anchored DNAJB12 and DNAJB14, along with the cytosolic co-chaperone SGTA, to actively facilitate protein export to the cytosol. This reflux mechanism requires BIP to operate from within the ER lumen and depends strictly on an intact DNAJB12 J-domain, establishing a directional, chaperone-guided pathway (Fig. 6A, B). Beyond expanding the functional landscape of BIP, this work identifies ERCYS as an integral component of UPR-driven spatial proteome remodeling and highlights it as a potential therapeutic target in cancer.
Supplementary Information
Additional file 1.