Targeting DNAJB9, a novel ER luminal co-chaperone, to rescue ΔF508-CFTR

The molecular mechanism of Endoplasmic Reticulum-associated degradation (ERAD) of Cystic fibrosis transmembrane-conductance regulator (CFTR) is largely unknown. Particularly, it is unknown what ER luminal factor(s) are involved in ERAD. Herein, we used ProtoArray to identify an ER luminal co-chaperone, DNAJB9, which can directly interact with CFTR. For both WT- and ΔF508 (deletion of phenylalanine at position 508, the most common CF-causing mutant)-CFTR, knockdown of DNAJB9 by siRNA increased their expression levels on the cell surface and, consequently, upregulated their function. Furthermore, genetic ablation of DNAJB9 in WT mice increased CFTR expression and enhanced CFTR-dependent fluid secretion in enteroids. Importantly, DNAJB9 deficiency upregulated enteroids’ fluid secretion in CF mice (homozygous for ΔF508), and silencing one allele of DNAJB9 is sufficient to rescue ΔF508-CFTR in vitro and in vivo, suggesting that DNAJB9 may be a rate-limiting factor in CFTR ERAD pathway. Our studies identified the first ER luminal co-chaperone involved in CFTR ERAD, and DNAJB9 could be a novel therapeutic target for CF.


Results
Discovery of DNAJB9 as CFTR-interacting partner using ProtoArray. Direct protein-protein interactions often have meaningful functions. ProtoArray technology (Invitrogen) has proven to be invaluable in identifying unelucidated protein-protein interactions in a variety of human studies [29][30][31] . The rationale for using ProtoArray for this study was to discover CFTR-interacting ER luminal proteins. To study QC mechanisms, we captured interacting partners on the microarray using purified full length FLAG-CFTR from mammalian HEK293 cells (Fig. S1). The bound CFTR was probed using mouse anti-FLAG monoclonal antibody followed by Alexa-488 conjugated secondary antibody and quantitation by a fluorescence microarray scanner. A stronger interaction results in a higher signal ( Fig. 1A-C).
Using the microarray, over 2,000 interacting partners were identified, and distributed in various subcellular localizations, including cytoplasm, nucleus, extracellular, PM and others. The results were validated by determining the known CFTR-interacting proteins from different compartments; Na + /H + Exchange Regulator Factor 1 (NHERF1, also known as EBP50) and NHERF3 (also known as PDZK1) are the scaffold proteins expressed in epithelial cell apical domains and can directly interact with the C-terminal of CFTR via their PDZ domains 32,33 . Therefore, as expected, NHERF1 and 3 showed strong signal in microarray studies (Fig. 1B). In contrast, proteins typically identified in CFTR co-immunoprecipitation studies, e.g. chaperone protein calnexin [CNX, 25,34 ], co-chaperone HSP70 Binding Protein 1 [HspBP1, 16 ] and ER degradation-enhancing α-mannosidase-like protein [EDEM, 25,34 ] show much lower signal strength with CFTR ( Fig. 1B), suggesting that ProtoArray has distinct ability to identify direct protein-protein interactions.
Multiple ER luminal chaperones bound CFTR with comparable signal intensity to known CFTR interactors; in particular, DNAJB9, a DNAJ (Hsp40) homology subfamily B member 9, bound CFTR with the relatively high signal intensity (Fig. 1C). Although the signal intensity for DNAJB9-CFTR is about 10-fold less than for NHERFs, it is 3-7 fold higher compared to other known proteins in Fig. 1B. DNAJB9 is a soluble ER luminal co-chaperone 35 , identified as inducible Heat Shock Protein upon cell stress 36 , and plays an important role in ERAD of misfolded proteins 36,37 . Other DNAJB family members also bound CFTR (Fig. 1C). Among them, DNAJB1, 6, 7, and 8 are cytosolic proteins [38][39][40][41] , while DNAJB2, DNAJB9, DNAJB11/ERjd3, DNAJB12 and DNAJB14 are ER proteins. Specifically, DNAJB2 is mainly expressed in neuronal cells 42 . DNAJB12 and DNAJB14 was demonstrated to be on the ER membrane with their functional J-domain facing the cytosol 41,43 . Although DNAJB11 is an ER luminal protein, it was suggested to assist BiP-mediated protein folding 44,45 . Given that newly synthesized CFTR is not associated with BiP 26 , we focused on DNAJB9, and wanted to test if the protein plays a role in ERQC for CFTR. DNAJB9 is associated with WT-and ΔF508-CFTR. To validate the interaction between CFTR and DNAJB9 in cells, co-immunoprecipitation (co-IP) was performed using HEK293 cell lines that stably express WT-and ΔF508-CFTR. Parental HEK293 cells (Par) were also recruited as specificity control. DNAJB9-HA 36 was transiently overexpressed in these cells. Immunoprecipitation of FLAG-CFTR recovered a fraction of DNAJB9 from cells stably expressing WT-CFTR, while in DNAJB9 immuneprecipitates using anti-HA antibody, WT-CFTR was present as well (Fig. 1D). Although ΔF508 alters CFTR global conformation 46 , the structure within ER lumen has not been elucidated. We tested whether DNAJB9 was associated with ΔF508-CFTR. As expected, a fraction of DNAJB9 was present in the immuneprecipitates of ΔF508-CFTR (Fig. 1E). Conversely, in DNAJB9-HA immunoprecipitates, a small fraction of ΔF508-CFTR was able to be detected as well (Fig. 1E). In contrast, Par control shown no CFTR or little DNAJB9 in co-IP, although it was noticed that DNAJB9 expression in Par control is much lower which is probably due to low transfection efficiency. To exclude the possibility that interaction between CFTR and DNAJB9 seen in the co-IP experiments was random association between these two proteins during lysis and incubation, chemical crosslinking was performed using DSP crosslinker before co-IP and the immunoprecipitates were subjected to immunoblotting using WES system (ProteinSimple Inc.). As shown in Fig. S2A,B, in both WT-and ΔF508-CFTR immuneprecipitates, a small fraction of DNAJB9-HA was detected. The interaction between CFTR and DNAJB9 was further validated by looking at their endogenous www.nature.com/scientificreports www.nature.com/scientificreports/ association using proximity ligation assay (PLA) in T84 cells. As shown in Fig. 1F, PLA signal was detected only when antibodies against both CFTR and DNAJB9 were used. Together with ProtoArray data, these results suggest that DNAJB9 is very likely a direct, interacting partner of both WT-and ΔF508-CFTR. DNAJB9 promotes turnover of CFTR. DNAJB9 is a soluble ER-localized Type II DnaJ homologue, containing N-terminal J domain and the Gly/Phe-rich domain, and C-terminal substrate binding domain 47 . Initially DNAJB9 was identified as an ER stress inducible co-chaperone with a role in protecting cells from ER stress 48 . Consistently, subsequent studies suggested that the general function for DNAJB9 was to mediate ERAD of misfolded proteins, e.g. surfactant protein C 36 and ENaC 49 ; knockdown of DNAJB9 by siRNA increased the target proteins' stability 36 . Several lines of evidence have suggested that DNAJB9 is able to couple the substrate binding and association with ERAD machinery in a BiP-independent manner [35][36][37] . Given the critical roles of DNAJB9 in maintaining protein homeostasis, DNAJB9-deficient mice, although viable, showed constitutive ER stress 50 . Based on these findings and the potential direct interaction between CFTR and DNAJB9, we hypothesized that DNAJB9 acts as a functional component in CFTR ERAD pathway.
In previous studies, knockdown of functional ERAD components e.g. co-chaperone DNAJB12 14 or E3 ligase RNF5 51 , increased total CFTR levels. Therefore, we asked if downregulation of DNAJB9 would lead to increased CFTR levels. Using WES system, compared to empty vector treated negative control, downregulation of DNAJB9 by siRNA increased total protein levels of WT-and ΔF508-CFTR ( Fig. 2A).
Subsequently, we asked whether decreasing DNAJB9 expression would increase or rescue WT-and ΔF508-CFTR function on the surface, as several lines of evidence suggested that knockdown of a functional component of CFTR ERAD pathways would increase CFTR surface expression as well as its function 14,51 . In-Cell Western (ICW) was used to measure the surface levels of CFTR, and in parallel SPQ assay was performed to determine CFTR function. As expected, based on previous studies 51 , knockdown RNF5 by siRNA upregulated CFTR surface levels (Fig. 2B). Interestingly, knockdown of DNAJB9 by siRNA also increased surface expression of both forms of CFTR (Fig. 2B). Consistently, downregulation of DNAJB9 by siRNA enhanced CFTR function www.nature.com/scientificreports www.nature.com/scientificreports/ ( Fig. 2C). It is noteworthy that knockdown of DNAJB9 using specific siRNA in our experimental settings could only lead to around 20-30% knockdown of DNAJB9 mRNA expression (Fig. S3), arguing that partially inhibiting DNAJB9 function can substantially change the folding kinetics of newly synthesized CFTR.
To further test the hypothesis that DNAJB9 plays a role in ERAD of CFTR, the ubiquitination levels of CFTR was examined upon DNAJB9 overexpression. As expected, overexpression DNAJB9 upregulated CFTR ubiquitination (Fig. S4); taken together, these results indicated that DNAJB9 is a functional component of the ERAD of CFTR and suggested that the DNAJB9-dependent ERAD pathway is a novel target for CF therapy.
DNAJB9 deficiency rescued both WT-and ΔF508-CFTR in mice. Given the robust phenotype implicating DNAJB9 in the ERAD of CFTR in cultured cells, we examined the role of DNAJB9 in vivo. CF mice (homozygous for ΔF508) are a useful model to understand the pathophysiology of disease as well as to evaluate therapies, such as an in vivo testing of the strategies of altering the interaction between ΔF508-CFTR and WT-or ΔF508-CFTR were transiently transfected with empty vector (Ø) or siRNA against DNAJB9 (siRNA) for 36-48 hours before assays. Parental cells that do not express CFTR were recruited as a negative control. All data were representative of three independent experiments. (A) WES analysis of CFTR and Vinculin (loading control). Total cell lysate was analyzed by WES system using anti-CFTR and anti-Vinculin antibodies. Each sample was analysed in triplicate. Student's t-test was performed to determine the statistical significance. (**P < 0.01; ***P < 0.005). (B) In-Cell western to determine surface CFTR levels. Cells were reverse transfected by siRNA against DNAJB9 and RNF5 in 96-well plate. Forty-eight hours after transfection, cells were then fixed and surface CFTR was probed by anti-FLAG antibody followed by IRDye 800CW conjugated Goat anti-Rabbit 2 nd antibody. CellTag 700 was used as cell quantity control. Samples without anti-FLAG antibody incubation were used as background control. Representative images were shown on the top and quantification was shown at the bottom. Each condition was done in triplicates. Student's t-test was performed to determine the statistical significance. (*P < 0.05; **P < 0.01). (C) SPQ assay to assess CFTR function. Cells were seeded into 96-well plate 24 hours after transfection and then subjected to SPQ assay described in "Materials and Methods" section. A representative graph is shown for WT-CFTR (top) and ΔF508-CFTR (bottom). Each condition was performed in triplicate. (D) Knockdown DNAJB9 increased CFTR expression. Immunoblotting of mouse Jejunum membrane fraction from different genotypes, DNAJB9 +/+ (WT), DNAJB9 +/− (Het), and DNAJB9 −/− (KO). CFTR and Na/K ATPase was probed using anti-CFTR and anti-Na/K ATPase antibodies, and Na/K ATPase was served as a membrane marker and loading control. Protein levels was quantified and ratio of KO/ WT was determined. Full-length blots are presented in Fig. S11. www.nature.com/scientificreports www.nature.com/scientificreports/ chaperone proteins 52 . Recently, primary intestinal organoids derived from mice have been successfully used by us and other groups as an important tool to test CFTR function and its interacting partners [53][54][55] . We hypothesized that DNAJB9 hypomorphic mice (KO) generated by gene trap technology 50 should have increased functional CFTR on the apical membrane in the enteroids. Interestingly, DNAJB9 was highly expressed in the intestine in both WT and CF mice and expression in CF mouse intestine was comparable to WT mouse (Fig. S5). RT-PCR analysis and our previous studies 50 show that KO mouse have minimal DNAJB9 mRNA expression and unaltered CFTR mRNA levels (Fig. S6). We first compared WT to KO mice (Fig. 3A,B). Although the extent of organoid fluid secretion was similar upon maximum forskolin (FSK) stimulation between WT and KO mice, the basal activity of CFTR was over 50% higher in KO mice. Consistently, immunoblotting of the membrane fraction from KO mouse intestine detected higher CFTR levels compared to WT (Fig. 2D), supporting the hypothesis that loss of DNAJB9 improve CFTR surface function.
DNAJB9 partial deficiency rescues ΔF508-CFTR. It has been observed that DNAJB9 −/− mice have constitutive ER stress and developmental problems 50 , and that DNAJB9 −/− CFTR ΔF508/ΔF508 mice are generally smaller than their littermates DNAJB9 +/+ CFTR ΔF508/ΔF508 . However, DNAJB9 heterozygous mice are asymptomatic 50 and in HEK293 cells 30% reduction of DNAJB9 could rescue ΔF508-CFTR (Fig. 2); therefore, we rationalized that partial knockdown of DNAJB9 in DNAJB9 heterozygous (DNAJB9 +/− ) is likely to rescue ΔF508-CFTR. As shown in Fig. 4A, compared to DNAJB9 +/+ CFTR ΔF508/ΔF508 mice, DNAJB9 +/− CFTR ΔF508/ΔF508 mice had increased organoid fluid secretion by ~31%. To further assess this "rescue" effect in vivo, intestinal closed-loop experiments were performed, which assess CFTR function of the intact gut as previously described 54,56 . In contrast to organoid www.nature.com/scientificreports www.nature.com/scientificreports/ fluid secretion assay where FSK was used to activate CFTR, cholera toxin (CTX) was used to stimulate CFTR in the closed-loop assay. DNAJB9 heterozygosity upregulated CFTR-dependent fluid secretion by ~60% and ~107% following 2 µg/mL and 10 µg/mL CTX treatment, respectively (Fig. 4B), consistent with an increase in functional ΔF508-CFTR when DNAJB9 gene dose is decreased. Lastly, we examined whether heterozygous DNAJB9 could improve the CF mice development by measuring their body weight since CF mice usually have reduced body weight because of lack of CFTR function 51 . It is found that in a sibling pair of same sex, CF mice with DNAJB9 heterozygosity had increased body weight compared to CF mice (Fig. 5), which is not dependent on sex and age.  www.nature.com/scientificreports www.nature.com/scientificreports/

Discussion
Endoplasmic Reticulum-Associated Degradation (ERAD) pathways have been widely appreciated to play a major role in the removal of "misfolded" CFTR from ER. Using different approaches, many important players have been identified, including cytosolic chaperones and ER membrane proteins. Increasing evidence suggests that ER luminal factors are also needed to mediate ERAD of misfolded proteins, including transmembrane protein opsin, mutation of which is associated with autosomal dominant retinitis pigmentosa 57,58 . However, the luminal factor(s) that participate in CFTR ERAD are still completely unknown. It may be possible that the involvement of ER luminal factors in CFTR ERAD has been simply neglected because (1) majority of the CFTR mutations occur in the cytosolic or membrane domains (www.cftr2.org), (2) the luminal loops or extracellular loops (ECLs) of CFTR are small, accounting for only 7% CFTR mass with the biggest loop, ECL4, harboring only 39 amino acids, and (3) the cytosolic or membrane factors appear to have enough "power" to mediate CFTR ERAD. It might be also due to lack of tools, such as antibodies, to identify these factors. Using purified full-length active CFTR coupled with ProtoArray, we have shown here that protoArray had distinct ability to identify CFTR-interacting partners in CFTR structure of various cellular compartments, including the ECLs of CFTR. It is found that DNAJB9 interacted with CFTR with relatively high signal intensity (Fig. 1C). Overexpression of DNAJB9 increased CFTR ubiquitination (Fig. S4). Knockdown DNAJB9 increased CFTR surface and functional expression in vitro and in vivo (Figs 2-4). These evidences suggest that DNAJB9 regulates the CFTR ERAD pathway and that interrupting DNAJB9 function may be a novel strategy to rescue mutant CFTR because heterozygosity of DNAJB9 improve overall CF mice development (Fig. 5).
Interestingly, in an earlier microarray gene analysis 59 , gene expression profile of ΔF508 homozygous CF patients was compared between patients with mild lung disease and patients with severe lung disease. It is found that DNAJB9 was one of the highly expressed genes in human nasal epithelial cells, and compared to patients with mild lung disease, patients with severe lung disease were associated with higher DNAJB9 expression (Fig. S7). This evidence further supports that targeting DNAJB9 could be an effective strategy for CF treatment.
DNAJB9 is a unique ER luminal co-chaperone 35 . Our data indicates that DNAJB9 interacts directly with CFTR. However, it is unclear exactly how CFTR interacts with this co-chaperone. DNAJB9 belongs to type II DNAJ (Hsp40) homology subfamily B, containing an N-terminal J-domain which was suggested to regulate nucleotide binding cycles of Hsp70, a G/F domain in the middle, and C-terminal domain (CTD) which was shown to interact with its substrate 47 . Therefore, it is possible that it is through its CTD that DNAJB9 interacts with CFTR. However, we do not rule out the possibility that their interaction is indirect (i.e., via an intermediary partner), which would result in the same outcome. More experiments are needed to determine this and is beyond the scope of this study.
Recently, it has been elegantly demonstrated that DNAJB9 prefers binding to aggregation-prone polypeptide in a BiP-independent manner and mediates the substrate to ubiquitination-proteasomal degradation pathway 37 . Interestingly, when CFTR ECL4 sequence was subjected to TANGO algorithm analysis which has been used by Behnke et al. to identify aggregation-prone region in their designed peptide library 37 , it was found that CFTR ECL4 has one aggregation-prone region at each end of the loop (Fig. S8). Particularly, a region at the C-terminal of ECL4 has great potential to form aggregates. Therefore, we propose that DNAJB9 via its CTD domain interacts with CFTR via ECL4. Interestingly, it seemed that DNAJB9 has no preference to either WT-CFTR or ΔF508-CFTR (Fig. 1). Importantly, deficiency of DNAJB9 could rescue both WT-and ΔF508-CFTR. These data suggest that the interacting region in both WT-CFTR and ΔF508-CFTR have equal accessibility to DNAJB9.
It has been widely accepted that DNAJB9 could mediate ERAD of its clients 36,37 , however, the molecular mechanism is unclear. Previous studies have suggested that interrupting ERAD-related (co-)chaperone function by siRNA could rescue ΔF508-CFTR 14,51 . This effect was also observed in our studies of DNAJB9, using both siRNA studies (Fig. 2) and a genetic ablation animal model (Figs 3-5), suggesting a critical role of DNAJB9 in CFTR ERAD pathway. Importantly, DNAJB9 is a soluble ER luminal protein 35 , distinct from previous studies focusing on cytosolic regulators, suggesting that ER luminal factors indeed participate in ERAD of misfolded integral membrane proteins, including CFTR. But, how does DNAJB9 execute its role? It is noteworthy that BiP protein is not required for DNAJB9-client interaction 36,37 and that BiP is not associated with CFTR complex [26][27][28] , suggesting that DNAJB9-mediated ERAD of CFTR is likely independent of BiP. It has been shown that multiple E3 ligases on the ER-membrane are involved in CFTR ubiquitination for degradation 14,51,60,61 . It is possible that DNAJB9 plays a role in directing misfolded CFTR to one of these E3 ligases and thus promoting CFTR ubiquitination because overexpressing DNAJB9 increased ubiquitination of CFTR at steady state (Fig. S4). Therefore, it would be interesting to determine which E3 ligase is involved in this pathway. Given the fact that DNAJB9 is not a membrane anchored protein, one could postulate that there must be an adaptor protein linking DNAJB9 to the membrane E3 ligase. Therefore, it is equally important to identify any potential adaptor protein which delivers DNAJB9-substrate complex to E3 ligase. Lastly, our siRNA study (Figs 2 and S3) and animal heterozygosity (Figs 4 and S6), where DNAJB9 mRNA were only partially downregulated, suggesting that DNAJB9 might be a limiting factor in CFTR ERAD pathway and that targeting DNAJB9 using specific inhibitor may turn out to be an important strategy in correcting Cystic Fibrosis effectively.

Materials and Methods
Chemical and antibodies. Chemical used in this study included Forskolin, CTX were from MilliporeSigma. www.nature.com/scientificreports www.nature.com/scientificreports/ Cell transfection. HEK293 cells were obtained from ATCC and were maintained in DMEM/F-12 medium containing 10% FBS and 1% Penicillin-Streptomycin. To do vector transfection, cells were transfected with human DNAJB9-HA [3HA was expressed at the very C-terminal of the protein 36 ] using Lipofectamine 3000 (ThermoFisher Scientific) according to the manufacturer's protocol. To do siRNA transfection, cells were transfected with DNAJB9 or RNF5 [HSS155077, ThermoFisher] specific siRNA by using RNAiMax (ThermoFisher) according to the manufacturer's protocol. Transfected cells were studied 36-48 hours post-transfection.

Mouse jejunum membrane fraction preparation. Mouse Jejunum in sucrose buffer (sucrose 250 mM,
Tris-HCl 10 mM, EDTA 1 mM, pH 7.2) were homogenized in a tissue grinder on ice in the presence of protease inhibitors (PMSF, Leupeptin, APR). After brief centrifuge at 800 × g for 10 min at 4 °C, the supernatant was centrifuged at 200,000 × g for 60 min at 4 °C. The pelleted membrane fraction was used for immunoblotting.
Cell Lysate preparation and immunoprecipitation. HEK293 cells stably express FLAG-ΔF508-CFTR and FLAG-WT-CFTR were lysed in IP buffer (87787, ThermoFisher) containing EDTA-free protease-inhibitor cocktail (MilliporeSigma). Cell debris were removed by centrifugation at max speed for 10 min at 4 °C. FLAG-CFTR and DNAJB9-HA were immuneprecipitated from whole-cell lysates using anti-FLAG-conjugated resin (MilliporeSigma) and anti-HA-conjugated resin (MilliporeSigma), respectively. Proteins immobilized on beads were eluted by SDS-PAGE Sample Buffer. Samples were incubated for 20 min at 37 °C before subjected to SDS-PAGE and western blotting following standard protocols. Images were acquired using Bio-Rad ChemiDoc TM Touch imaging System, processed using Image Lab (version 6.0), and collected in Photoshop (version 19

Full length FLAG-CFTR purification and activity measurement. HEK293 cells expressing full-length
FLAG-CFTR were lysed using lysis buffer (0.2% Triton X-100 in PBS) in the presence of protease inhibitors and cells debris was removed by centrifugation 63 . Whole cell lysates (or purified microsomes) were subjected to anti-FLAG-conjugated resin column. After extensive washing, bound proteins were eluted with elution buffer (100 mM Glycine, pH 2.2) containing 0.2% Triton X-100 and then neutralized with 1 M Tris. Purified protein mixture were dialyzed with dialysis buffer. The purity were determined by SDS-PAGE followed by Coomassie Blue staining and quantitated by densitometry analysis. The ATPase activity of purified CFTR was measured using radiolabeled [γ-32 P]ATP by incubation for the indicated time followed by Thin-layer Chromatography (TLC).
ProtoArray. ProtoArray (ThermoFisher) was performed using the above purified full-length FLAG-CFTR according to manufacturer's protocol.
Tango analysis. Tango analysis was done using the online tool at http://tango.switchlab.org/ for prediction of aggregating regions in unfolded polypeptide chains.
WES assay. Total cell lysate or IP samples were prepared in lysis buffer (RIPA from Cell Signaling or IP lysis buffer from ThermoFisher) containing protease inhibitor cocktail, and then subjected to WES system according to manufacturer's protocol using the indicated antibodies. Data were analyzed using Compass for SW version 3.1.7. SPQ assay. HEK293 cells stably expressing WT-CFTR and ΔF508-CFTR were transfected with DNAJB9-specific siRNA [HSS106410, ThermoFisher, 36 ]. Cell were plated in a 96 well plate with clear bottom and black well plate 24 hours after transfection. After overnight incubation, cells were acutely loaded with 10 mM SPQ prepared in 1:1 Opti-MEM:water for 5 min at 37 °C. After aspirating SPQ, cells were washed once with NaI buffer [130 mM NaI, 20 mM HEPES, 10 mM glucose, 4 mM KNO 3 , 1 mM Ca(NO 3 ) 2 •H 2 O, and 1 mM Mg(NO 3 ) 2 ], and then continue incubation at room temperature (RT) in NaI buffer for 1 hour with one time replacement with fresh NaI buffer. Cells were then washed once with NaNO 3 buffer [130 mM NaNO 3 , 20 mM HEPES, 10 mM glucose, 4 mM KNO 3 , 1 mM Ca(NO 3 ) 2 •H 2 O, and 1 mM Mg(NO 3 ) 2 ] and then 100 µL NaNO 3 buffer was added to each well and SPQ signal were measured at 360 EX/460 EM for 5 minutes using Flex-station3 (Molecular Devices) to record the baseline. To measure SPQ signal at FSK stimulated phase, additional 100 µL NaNO 3 buffer containing 10 µM FSK was added to each well. Lastly, additional 100 µL NaI buffer was added to quench the SPQ signal. Signal from each well was normalized to signal at 5 th min.
In-cell western (ICW). HEK293 cells 24 hours after transfection were plated on 96 well plate and fixed with 4% formaldehyde for 15 min at RT. After washing with PBS, cells were incubated with blocking buffer (10% BSA in PBS) for 1 hour at RT and then with anti-FLAG antibody in 3% BSA in PBS for overnight at 4 °C. After thorough washing, cells were incubated with IRDye ® 800CW-conjugated anti-Rabbit antibody (ProteinSimple) containing CellTag 700 (ProteinSimple) for 1 hour. After washing, buffer was aspirated from each well and the plate was scanned and quantitated using LI-COR Odyssey ® CLx imaging system.