NSAIDs-dependent adaption of the mitochondria-proteasome system in immortalized human cardiomyocytes

The progressive consumption growth of non-steroidal anti-inflammatory drugs (NSAIDs) has progressively raised the attention toward the gastrointestinal, renal, and cardiovascular toxicity. Increased risk of cardiovascular diseases was strictly associated with the usage of COX-2 selective NSAIDs. Other studies allowed to clarify that the cardiovascular risk is not limited to COX-2 selective but also extended to non-selective NSAIDs, such as Diclofenac and Ketoprofen. To date, although a less favorable cardiovascular risk profile for Diclofenac as compared to Ketoprofen is reported, the mechanisms through which NSAIDs cause adverse cardiovascular events are not entirely understood. The present study aimed to evaluate the effects of Ketoprofen in comparison with Diclofenac in immortalized human cardiomyocytes. The results obtained highlight the dose-dependent cardiotoxicity of Diclofenac compared to Ketoprofen. Despite both drugs induce the increase in ROS production, decrease of mitochondrial membrane potential, and proteasome activity modulation, only Diclofenac exposure shows a marked alteration of these intracellular parameters, leading to cell death. Noteworthy, Diclofenac decreases the proteasome 26S DC and this scenario may be dependent on the intracellular overload of oxidized proteins. The data support the hypothesis that immortalized human cardiomyocytes exposed to Ketoprofen are subjected to tolerable stress events, conversely Diclofenac exposition triggers cell death.

via the intrinsic/mitochondrial pathway 34 . Furthermore, the imbalance between physiological and pathological ROS levels could be associated with proteasome dysfunctions leading to decreased degradation of several proteasome substrates, including IκB, p53, Bax, and p27, and induced apoptosis 35,36 . ROS induction is not the only mechanism by which NSAIDs may induce proteasome dysfunction. Even though the mechanism is not fully understood, proteasome inhibition contributes to impair the proteasome protective function resulting in a higher risk of cardiac proteinopathy 37 .
A recent study demonstrated that, unlike aspirin, Diclofenac treatment in cardiomyocytes induced ROS generation, alterations of mitochondrial functions, and decreased proteasome activity 38 .
Hence, the present study aimed to evaluate the effects of Ketoprofen (K) in comparison with Diclofenac (Dic) in immortalized human cardiomyocytes.

Results
In vitro model characterization. As in vitro model, human immortalized cardiomyocytes were cultured as described in Materials and Methods. The characterization of the in vitro model was performed by assessing the localization of proteins specifically expressed in mature cardiomyocytes by immunofluorescence assay 39 .
Confocal microscopy analyses were carried out by comparing the cells maintained for 24 h (48 h from the seeding, indicated as 2 days in vitro, 2DIV) in not-proliferating conditions with the cells maintained for six days in the same medium (seven days from the seeding, indicated as 7 days in vitro, 7DIV). The representative pictures of immunolocalization analysis of myosin heavy chain 7 (MHC7), a crucial late differentiation marker and connexin 43 (Cx43), the predominant gap junction in heart tissue, essential to ensuring cardiac electric activity 40 , are reported in Fig. 1A,B. Although we did not observe typical myosin striation, an increase in the red fluorescence intensity in 7DIV condition compared to 2DIV cells was observed. Cx43 immunostaining showed the increase of gap junction network in 7DIV cells compared to 2DIV cells. Myoblast determination protein 1 (MyoD), a skeletal muscle-specific bHLH transcription factor, which is early activated during myogenesis, and normally down-regulated in differentiated human cardiomyocytes 41,42 , was significantly reduced in 7DIV cardiomyocytes compared to 2DIV condition, as shown by western blotting analysis (Fig. 1C). These results provide evidence about the validity of this cellular model, as a useful tool to analyze the effects of NSAIDs on cardiac cells.  Ketoprofen and Diclofenac affect MMP in a different way. Mitochondrial metabolism is the primary ROS source; therefore the loss of normal mitochondrial homeostasis may result in the imbalance of ROS production. The electrochemical gradient between the inner and outer mitochondrial membranes drives the ATP synthesis and generates the mitochondrial membrane potential (MMP, ΔΨ). MMP should be maintained in a homeostatic range to ensure the correct mitochondrial functions. This electrochemical parameter can be assayed to evaluate the mitochondrial state, which can be modulated by several xenobiotic compounds. In this regard, JC-1 cationic dye is an useful tool to detect MMP in adherent cells [43][44][45] . The reduction of aggregate/  www.nature.com/scientificreports/ monomer ratio related to JC-1 dye indicates a decrease of the mitochondrial membrane potential resembling the effect induced by uncoupling agents, such as FCCP. This condition could be associated with cell death, but it is not mandatory, due to the possibility that some compounds induce a negative modulation of MMP without triggering cellular death pathways, as previously demonstrated by some investigations on salicylates [46][47][48] . Therefore, to evaluate the potential cytotoxicity of a specific compound, MMP detection assays should be supported by cell viability and cytotoxicity detection assays and by the assessment of mitochondrial number. We evaluated the MMP modulation induced by the NSAIDs of interest through the use of fluorescent cationic JC-1 dye and by TRME live imaging. A decrease of the JC-1 ratio at 100 (  5B) for 24 h promoted a decrease of JC-1 ratio compared to the untreated cells, however, the effect of K was significantly less severe, especially at 200 μM, compared to Dic. This result is also supported by the TRME live analysis, showing a substantial reduction of red fluorescence intensity upon Dic only (Fig. 5F), while no decrease of TRME is observed with K. In line with previous results obtained by other research groups on cardiac tissue, cardiac cells 33,38 , hepatic tissue, and hepatocyte culture 49 , Dic exposure induces a detrimental reduction of MMP. On the other hand, it is possible to suppose that K effects on MMP result in a mechanism resembling the effect of salicylate derivatives on MMP, thus without significant cytotoxic events 46-48 . All the observed events are, however, accompanied by a significant reduction of the mitochondrial marker Mitotraker upon Dic (Fig. 5C), thus suggesting a decrease of mitochondria in Dic-treated cells, not observed with K. Also, in Dic-treated cells a significant increase of the fluorescence intensity for MitoSox (a specific marker of mitochondrial superoxide) is observed (Fig. 5C), while this parameter is only slightly affected by K, thus suggesting a specific increase of mitochondrial superoxide in Dic-treated cells. Since the effect of ROS in triggering the opening of mitochondrial permeability pore (mPTP), inducing changes of mitochondrial membrane potential, likely activating intrinsic apoptotic pathway 34 ; these parameters were assessed with live-imaging assay by Incucyte device. mPTP opening was evaluated by Mitochondrial PT Pore Assay. Briefly, this assays uses a calcein/cobalt quenching technique, where calcein stains the entire cell, while cobalt is able to quench the calcein fluorescence signal outside the mitochondrial matrix. If the inner mitochondrial membrane (IMM) is in physiological condition cobalt can not cross the IMM and the cells exhibit green fluorescence. Conversely, if the IMM is damaged the green fluorescence is quenched by cobalt, and the cells exhibit a decrease of the green fluorescence intensity.
Notably, upon Dic exposure (at both concentrations tested) the mPTP is significantly compromised compared to untreated and K-treated cells, as shown by the decrease of the green fluorescence intensity (Fig. 5D,E). These data were further supported by TMRE assay performed on the same cellular sample (Fig. 5D,F). Upon Dic  www.nature.com/scientificreports/ exposure (at both concentrations tested) the red fluorescence intensity related to TMRE is decreased compared to untreated and K-treated cells. Moreover, the induction of the apoptotic pathway was evaluated by Incucyte caspase-3/7 green assay, where the apoptosis activation appears proportional to the green fluorescence intensity.
In agreement with the previous data, after Dic exposure (at both concentrations tested), we observed a dramatic increase of caspase 3/7 activity, mainly after 10 h of treatment, compared to untreated and K-treated cardiomyocytes, as reported in Fig. 5G,H.
Proteasome activity is modulated upon Ketoprofen and Diclofenac exposition. Altered intracellular levels of ROS may trigger protein damage resulting in proteotoxic stress and eventually in cell death 50 . In this scenario, the ubiquitin proteasome system (UPS) is deputed to recognition, degradation, and recovery (if possible), of damaged or unfolded proteins. The proteasome activities ensure the degradation of damaged proteins in an ATP-dependent process 51 . Proteasome or 26S proteasome is composed of three principal components, the core 20S or CP (core particle) containing proteolytic activity, and two ATP-dependent 19S or regulatory particles (RP). Three different protease activities within the proteasome core have been identified: caspase-like post acidic (β1), trypsin-like post basic (β2), and chymotrypsin-like post hydrophobic (β5). These core protease activities can be detected employing specific fluorescently tagged substrates. In intact substrate, the fluorescence is quenched, while after substrate cleavage the fluorescence is released and its intensity is proportional to proteasome activity therefore it can be detected. Despite the binding specificity of fluorescent compound used to detect proteasome activity, in the cellular crude extract, these fluorescent substrates can also be digested by other non-proteasome proteases. For this reason, the analysis should be performed in the presence and absence of a specific proteasome inhibitor, such as MG-132, bortezomib, or epoxomicin. To obtain a more reliable result from the assay, proteasome activity with the inhibitor is subtracted from proteasome activity without inhibitor [52][53][54] .
We, therefore, analyzed the effects of tested NSAIDs, at 100 μM and 200 μM for 24 h, on chymotrypsin-like activity, the most recurring proteasome activity, in the presence and absence of MG-132 inhibitor (Fig. 6A,D). Both assessed compounds, at each concentration tested, show a significant decrease of chymotrypsin-like activity compared to untreated cardiomyocytes. Interestingly, a significant difference in this activity was observed only at high concentration in cells exposed to Dic compared to K (Fig. 6A,D). Although no significant differences between the compounds at 100 μM concentration are observed in trypsin-like activity (Fig. 6B), upon 200 μM concentration, cells treated with Dic show a significant increase of this proteasome activity compared to K-treated and untreated cells (Fig. 6E). Caspase-like activity is significantly reduced in cells upon Dic exposure compared to K-treated and untreated cells at both concentrations assayed (Fig. 6C,F).

Diclofenac-dependent alteration of the proteasome configuration. To support previous results,
we assessed the effects of tested NSAIDs (100 μM and 200 μM for 24 h) on the proteasome structure by native gel electrophoresis/Western blotting, as described in previous investigations 33,38 . The representative images of the membranes are reported in Fig. 7A,B. The membranes were incubated with primary antibody anti-PSMA6 (proteasome sub-unit α6), which recognizes all proteasome structures. At both concentrations tested, in cells exposed to Dic, it is possible to observe a clear absence of the band corresponding to proteasome 26S double capped (26S DC); conversely, this band is present in untreated and K-treated cells (Fig. 7A,B). In Fig. 7C,D, the histograms of the relative proportion (in percentage) of each proteasome form (26DC and 20S) are reported. This is a qualitative analysis of proteasome configuration in a specific moment of cell life. The results show a proportional increase of proteasome 20S (about 80% of the total) and a clear proportional decrease of proteasome 26S DC (about 15% of the total) in immortalized human cardiomyocytes exposed to Dic. This effect is not found in cells exposed to K, which appear similar to untreated cells (Fig. 7C,D). Hence, we can postulate that Dic induces alterations of 26S structure, which establishes a qualitative loss of proteasome configuration.

Increased levels of intracellular oxidized proteins after Diclofenac exposition. High levels
of oxidized proteins, due to non-physiological oxidative stress levels associated with mitochondrial damage, induce the proteasome 26S dismantling [55][56][57][58] . In this situation the UPS is ineffective, and only proteasome 20S can remove oxidized proteins. Oxyblot assay is used to evaluate the intracellular levels of oxidized proteins. In Fig. 8A,B representative images of OxyBlot with respective blue Coomassie staining of the same membrane are reported. In Fig. 8C,D, the histograms of the respective analyses are reported. In our experimental conditions, the immortalized human cardiomyocytes exposed to Dic show a marked increase of oxidized protein compared to untreated cells and K-treated cells, suggesting an accumulation of oxidized proteins that may be involved in the replacement of 26S with 20S proteasome in cardiomyocytes, as previously demonstrated [55][56][57][58] .

Discussion
Reports on cardiovascular adverse reactions began to emerge in early 2003 59 . Later, several placebo-controlled trials focused on COX-2 inhibitors showing an increased risk of atherothrombotic vascular events associated with the use of these drugs 60,61 . More recent data from meta-analyses of randomized trials and observational studies have contributed to clarifying that cardiovascular side effects are not a peculiar characteristic of COX-2 inhibitors, but are also associated with the use of some NSAIDs [60][61][62][63][64][65][66][67] .
Several findings support the concept that both COX-2 selective inhibitors and tNSAIDs may increase cardiovascular risk, although this effect greatly varies among individual drugs and strictly depends on the dose 62 . The difference among individual NSAIDs-associated side effects requires specific investigation since it may depend in part on the specific pharmacodynamic properties, but it is very likely to rely also on unique COX-independent activities of the molecule.
Increased levels of 20-Hydroxyeicosatetraenoic acid (20-HETE) were observed in mice treated with COX-2 inhibitors and associated with decreased tail bleeding and increased platelet aggregability 69 . Although increased 20-HETE levels are likely related to COX-2 inhibition and probably contribute to the adverse cardiovascular outcome, this effect was observed across the NSAIDs class thus excluding a specific COX-2 dependent effect.
Prostaglandins activate a class of receptors called E-type prostanoid (EP) receptors, which play a key role in the development of pain and inflammation and are also involved in the control of apoptosis and cell survival 70,71 .
Thus, the block of prostaglandin signaling by NSAIDs may be responsible for NSAID-induced apoptosis. However, other mechanisms are emerging that better explain the individual behavior of different NSAIDs. Specifically, NSAID-induced ROS generation in cardiomyocytes was found to be a critical step in the induction of apoptosis, associated with mPTPs opening and proteasome dysfunction [32][33][34][35][36][37][38] .
NSAID-induced ROS can damage proteins causing multiple effects on the proteasome: the oxidation of proteasomal subunits may indirectly result in proteasome inhibition but the increased levels of oxidized proteins may also overload the proteasome, inducing dysfunction.
To better understand the COX-2 independent effects accounting for cardiovascular side effects, we compared the effects of Ketoprofen and Diclofenac in differentiated immortalized human cardiomyocytes.  . Data were expressed as R.U., and each sample was normalized on its respective blue coomassie staining. Neg. Ctr (negative control) is a sample without protein derivatization. Ctr versus treated, *p < 0.05, ****p < 0.0001; Keto versus Diclofenac, + p < 0.05, ++++ p < 0.0001. Fulllength Oxyblot and membranes stained with blue coomassie can be found in Supplementary Fig. S3.
Scientific Reports | (2020) 10:18337 | https://doi.org/10.1038/s41598-020-75394-x www.nature.com/scientificreports/ Interestingly, the results obtained in the in vitro model revealed a markedly deleterious effect of Dic in comparison to K. We show that Dic exposure causes cardiotoxicity and a strong decrease of cell viability, in line with a recent investigation 38 . Considering the cardiomyocytes attitude to produce ROS during the metabolic activity to ensure the lifespan of cardiac tissue, it is conceivable to assume that a correct balance between physiological and pathological ROS levels is indispensable for cellular homeostasis. Some NSAIDs, such as naproxen sodium, are ROS inductors without cytotoxic effect, while others, such as Diclofenac and meclofenamate sodium, are strong ROS inductors associated with cytotoxicity 33,38 .
Mitochondria play a crucial role in cellular ROS production, therefore the loss of normal mitochondrial homeostasis results in an imbalance of ROS generation. Some NSAIDs can uncouple oxidative phosphorylation and dissipate MMP as extensively demonstrated 72 . In our experimental conditions, K treatments did not affect cell viability, while increases total ROS production, but not mitochondrial ROS and slightly decreases MMP in a dose-independent way, without affecting the mitochondrial number. On the contrary, Dic treatments strongly altered these parameters in a dose-dependent manner, reducing mitochondrial number and increasing mitochondrial ROS, leading to cell death. In summary, a potential tolerable effect of K on ROS production and MMP was observed, in agreement with that described for salicylate derivatives and naproxen 33,47,48 .
Conversely, in our experimental conditions, after Dic administration, higher ROS production (both total and mitochondrial) associated with a strong mitochondrial membrane depolarization and a decrease of mitochondria was found. Although K exposition triggers a total ROS imbalance, it appears not detrimental and may induce, as a consequence, the expression of cytoprotective genes, as already demonstrated 73 , resembling the cellular response to physiological levels of ROS. Conversely, Dic triggers a marked ROS imbalance (both total and mitochondrial), which is proportionally linked to the exposure dose. Due to high ROS-induced stress, Dic-treated cells are unable to trigger cytoprotective response, thus favoring cell deathlikely by apoptosis as demonstrated by activation of caspases 3/7.
The underlying mechanism of NSAIDs-dependent depolarization effect may be due to their ability toactivate the mPTPs opening, which ensures the free passage of low molecular weight compounds between the inner mitochondrial matrix and cytosol. The opening of mPTPs is promoted by Ca 2+ accumulation in mitochondria, pro-oxidants, and low MMP. In this regard, NSAIDs such as aspirin and derivatives, directly affect mPTPs resulting in depolarization of the mitochondrial membrane, as previously described [46][47][48] . This action is linked to impaired mitochondrial Ca 2+ uptake, as already observed in colon cancer cell lines 48 and vascular smooth muscle cells (VSMCs) 47 . The proton conductance of the inner mitochondrial membrane is increased by salicylate, thus this net proton influx leads to the uncoupling of mitochondria 46 . Our hypothesis on the observed K-dependent mitochondrial membrane depolarization is that the drug has the ability to affect proton influx in the mitochondrial respiratory chain, similarly to salicylate derivatives. Also, it was previously shown that the uncoupling effect of Dic is about 50-fold greater than salicylate 74 . In agreement, our results show an alteration of mPTP in cells exposed to Dic; conversely cells exposed to K show the same behaviour of untreated cells. These data were further confirmed by TMRE staining, where cells exposed to Dic show a significant decrease in the fluorescence related to TMRE. Therefore, Dic enhances the permeability of transition pore on the IMM, which, in turn, may induce the depolarization of mitochondrial membrane leading to activation of the apoptotic pathway.
As mentioned above, mitochondria metabolism plays a central role in ROS production. Non-physiological levels of ROS, together with inadequate antioxidant defences trigger protein damage that results in proteotoxic stress and eventually in cell death 50 . The management of damaged or unfolded proteins is mediated by UPS, which delivers ubiquitin tagged proteins to the proteasome and ensures their degradation. UPS is an ATPdependent process and each event is realized with energy consumption to counteract intra-and extra-cellular proteotoxicity 51 . The proper activity of UPS is responsible for the correct turnover of proteins that are required for cardiac homeostasis. Impaired UPS function has been implicated in heart diseases 75,76 .
Proteasome or 26S proteasome is composed of three principal components, the core 20S or CP with proteolytic activity, and two ATP-dependent 19S or RP. Four stacked rings with a central cavity form the core particle. These rings are arranged in a particular manner: two outer rings composed of seven α-subunits and two inner rings composed of seven β-subunits. Only three β-subunit present the active protease site, each one diverging from the others due to the different amino acidic cutting site recognized. To date, three different protease activities within the proteasome core have been identified: caspase-like post acidic (β1), trypsin-like post basic (β2), and chymotrypsin-like post hydrophobic (β5). The regulatory particle is composed of ATPase subunits accountable for protein cargo translocations in the 20S core. Subunits composing 19S particles can bind ubiquitin, thus participating in the recognition and de-ubiquitination of substrates 51,77 . Each of the three 20S core protease activities can be detected utilizing specific fluorescently tagged substrates. Our results show the direct role of Dic in triggering 26S structure alterations, which establish the qualitative loss of proteasome activity. Although 26S proteasome is involved in the recognition and degradation of ubiquitinated proteins, the loss of 19S triggers an alternative mechanism of degrading damaged proteins, which is dependent only on 20S proteasome activity. Therefore, sustained stress conditions into the cell, such as excessive production of ROS, could reduce UPS activity leading to disassembling of proteasome 26S. When two 19S regulatory particles are separated from the 20S core, the degradation of damaged proteins is carried out only by 20S core, without ubiquitin labeling and ATP consumption. This condition is especially induced by high oxidative stress [55][56][57][58] .
Although proteasome chymotrypsin-like activity is negatively affected by K exposure, it is possible to hypothesize that cells treated with K maintain a functional UPS, like untreated cells, since they shows an intact 26S proteasome, also called 26S DC. As a matter of fact, in our experimental condition, upon K exposure the immortalized human cardiomyocytes preserve proteasome caspase-like activity. On the contrary, cells treated with Dic show only 20S proteasome structure and a significant loss of chymotrypsin-like and caspase-like proteasome activity. Noteworthy, upon Dic exposure cardiomyocytes show a significant increase of trypsin-like proteasome Scientific Reports | (2020) 10:18337 | https://doi.org/10.1038/s41598-020-75394-x www.nature.com/scientificreports/ activity. Cytotoxic events due to Dic exposure are probably associated with excessive ROS production, leading to a marked increase of oxidized proteins. In this context, ATP production could be limited, and oxidized protein levels may increase. The concurrence of these events could trigger the disassembly of proteasome 26S and the loss of ubiquitin-dependent proteasome activity. The high amount of damaged protein could accumulate into the 20S core and then overload proteases activity leading to a reduction in the overall proteasome activity, as observed after Dic exposure.
Probably, immortalized human cardiomyocytes exposed to K maintain responsiveness to counteract tolerable levels of stress, conversely Dic treatment triggers a cardiotoxic response, ultimately leading to cell death as summarized in Fig. 9, where a possible sequence of events is depicted for both compounds. .

Materials and methods
Immortalized human cardiomyocytes culture. Immortalized human cardiomyocytes cells were used as a cardiac model and purchased from Applied Biological Materials Inc. (abm). They derived from the ventricular tissue of 62 old years male. The culture media was composed of Dulbecco's modified Eagle's medium/Ham's F12 50/50 mix containing 10% Foetal bovine serum (FBS), 100 U/ml penicillin, 100 U/ml streptomycin, 2 mM glutamine (Corning, Manassas, VA, USA) and it was replaced day by day. The cells (used at passage 4-8) were subcultured by 0.25% trypsin-EDTA (Corning, Manassas, VA, USA) enzymatic digestion. For cardiomyocytes characterization, cells were seeded in 10% FBS supplemented DMEM F12 at a seeding density of 2.5 × 10 4 cells/ cm 2 . 24 h after seeding, 10% FBS supplemented DMEM F12 was replaced with the culture medium supplemented with 1% of FBS, which was changed every day until the sixth day. Then, human cardiomyocytes were exposed to NSAIDs of interest.
Cell treatments. Ketoprofen powder was dissolved in sterile water (containing 20 μl of NaOH 5 N) to obtain a 50 mM stock solution, and used in a final concentration range of 100-600 μM. Diclofenac powder was dissolved in sterile DMSO (Sigma, St. Louis, Mo, USA) at an initial concentration of 50 mM and used in a final concentration range of 100-600 μM. Each solution was freshly prepared for every experiment. As mentioned above, seven days after seeding, (seeding density of 2.5 × 10 4 cells/cm 2 , one day in 10% FBS-supplemented medium and 6 days in 1% FBS-supplemented medium) cardiomyocytes were exposed to each treatment for 24-72 h. www.nature.com/scientificreports/ Contrast phase images. Immortalized human cardiomyocytes were seeded into a collagen I-coated flask T75 cm 2 (2.5 × 10 4 cells/cm 2 seeding density) in 10% FBS supplemented DMEM F12. 24 h later the 10% FBS supplemented media was replaced with 1% FBS supplemented media, and the cardiomyocytes were maintained in this culture conditions for 6 days. On the seventh day of culture, the cells were exposed, for 24 h, to 200 μM of NSAIDs in 1% FBS-supplemented medium. Then the images were acquired at 20 × magnification by Leica DMi1-CH-9435 optical microscope.
Immunofluorescence. Immortalized human cardiomyocytes were seeded on collagen I-coated ( Cell index (CI). Immortalized human cardiomyocytes were seeded into the wells of 16-well E-plate at a seeding density of 2.5 × 10 4 cells/cm 2 and cultured for seven days. Each treatment was performed after the seventh day of culture; a suitable time to obtain differentiated human cardiomyocytes.
The xCELLigence system (Roche Applied Science) provides a quantitative parameter called cell index (CI), which reflect the cell status. Briefly this system measures cell-electrode impedance, thus the CI represents a quantitative measure of the cell number, cell viability, adhesion degree, and morphology. The results are reported as delta cell index (DCI). For each well, DCI represents the CI at a given time point (CI ti ) plus delta value. The difference between a reference DCI value and the CI at the delta time point provides the well DCI: IncuCyte Cytotox Green and Caspase-3/7 Green. Immortalized human cardiomyocytes were plated (seeding density 2.5 × 10 4 cells/cm 2 ) in 10% FBS-supplemented medium w/o phenol red into a 96 black well plate. The next day the culture medium was replaced with 1% FBS-supplemented medium w/o phenol red, this culture medium was daily replaced until the seventh day of culture. Then, the cells were exposed to 100 and 200 μM of NSAIDs in 1% FBS-supplemented medium w/o phenol red, and 250 nM of IncuCyte Cytotox Green Reagent (Essen BioScience) was added in the experimental culture medium for counting dead cells. For the apoptosis detection, 5 μM IncuCyte Caspase-3/7 Green (Essen BioScience) was added in the experimental culture medium for counting caspases activation The plates were placed in IncuCyte device (20 × objective), the cytotoxicity and caspases activation were recorded (three images for well, six replicates) every 3 h by both phase contrast and fluorescence scanning for 24 h at 37 °C and 5% CO2. Images were analysed using the Incucyte ZOOM software and the data were reported as green object count per mm 2 .

Measurement of cellular ROS.
2′-7′-dichlorofluorescein diacetate (DCFDA) cellular ROS detection assay kit (Abcam, Cambridge, UK) to analyze ROS production in our in vitro model was used according to manufacturer's instructions. Briefly, immortalized human cardiomyocytes were plated (seeding density 2.5 × 10 4 cells/cm 2 ) in 10% FBS-supplemented medium w/o phenol red into a 96 black well plate. The next day the culture medium was replaced with 1% FBS-supplemented medium w/o phenol red, this culture medium was daily replaced until the seventh day of culture. Then, the cell monolayer was washed one time with 1X buffer, and was incubated with DCFDA 10 μM for 30 min at 37 °C protected from the light. Later the cell monolayer was washed with PBS and the cells were exposed to 100 and 200 μM of NSAIDs in 1% FBS-supplemented medium w/o phenol red. H 2 O 2 800 μM was used as a positive control. Every single experiment was performed in quadruplicate. ROS production was immediately determined by measuring the formation of fluorescent dichlorofluorescein (DCF), using a PerkinElmer VICTOR 3 , at an Ex-485 and Em-535 nm. Measurements were done every 30 min for six hours, this being the optimal condition to measure ROS production, as suggested by the manufacturer. The value of fluorescence intensity at each time point is reported. The value reported was obtained by the ratio of fluorescence at a specific time point on fluorescence at time 0, which was measured immediately after DCFDA incubation.

Determination of mitochondrial membrane potential (MMP).
The mitochondria dye JC-1 (Abcam, USA) was utilized to evaluate the NSAIDs effect on MMP in immortalized human cardiomyocytes, as indicated by the manufacturer. Briefly, cells were plated (seeding density 2.5 × 10 4 cells/cm 2 ) in 10% FBS-supplemented medium w/o phenol red into a 96 black well plate. The next day the culture medium was replaced with 1% FBSsupplemented medium w/o phenol red, this culture medium was daily replaced until the seventh day of culture. Then, the cells were exposed, for 24 h, to 100 and 200 μM of NSAIDs in 1% FBS-supplemented medium w/o phenol red. As depolarization control, FCCP 100 μM for 4 h was used. FCCP acts as an uncoupling agent, thus preventing ATP synthesis. After exposure to treatments, the cell monolayer was washed with PBS and then incubated with JC-1 dye 10 μM for 20 min at 37 °C protected from light. Later, the cell monolayer was rinsed with 1X dilution buffer and the proportionally fluorescence to MMP was immediately measured by using a PerkinElmer VICTOR 3 . Every single experiment was performed in quadruplicate. The fluorescence of the JC-1 aggregate form was measured by setting the Ex-531 and Em-595 nm wavelengths, while the fluorescence of JC-1 monomer form was measured by setting the Ex-485 and Em-535 nm wavelengths. The fluorescence intensity values were expressed as the ratio JC-1 aggregate form/JC-1 monomer form. www.nature.com/scientificreports/ substrate, which releases free highly fluorescent AMC (7-amido-4-methyl coumarin) in the presence of proteasome proteolytic activity. The assay was performed in the presence and absence of MG132 proteasome inhibitor. Briefly, cells were plated (seeding density 2.5 × 10 4 cells/cm 2 ) in 10% FBS-supplemented medium w/o phenol red into a flask T75 cm 2 . The next day, the culture medium was replaced with 1% FBS-supplemented medium w/o phenol red, this culture medium was daily replaced until the seventh day of culture. On the seventh day of culture, the cells were exposed, for 24 h, to 100 and 200 μM of NSAIDs in 1% FBS-supplemented medium w/o phenol red. After exposure to treatments, the cell monolayer was detached by trypsin and centrifuged 6 min at 250 g. The cell pellets were washed with cold PBS and transferred into 1.5 ml tubes, then centrifuged 6 min at 250 g. 0.5% NP-40 (tergitol) in PBS was used to suspend cell pellets to obtain protein extract. About 500 μl, for cell pellet, of 0.5% NP-40 extraction buffer was used. After homogenization by pipetting up and down ten times, the extracts were centrifuged 15 min at 16,000 g by refrigerated mini centrifuge. The supernatants were collected and maintained at 4 °C, ready for the assay. Extract samples and AMC standards (1-10 μM) were placed in 96 black well plate in a total volume of 100 μl. In all sample wells, the fluorescent substrate AMC (final concentration 50 μM) was placed with or without MG132 proteasome inhibitor (final concentration 100 μM). After mixing all the components in the wells, the plate was incubated at 37 °C for 20 min protected from light (T1 measure). Chymotrypsin-like activity at T1 was determined by measuring the fluorescence released from the AMC substrates, using a PerkinElmer VICTOR 3 , at an Ex-355 and Em-460 nm. After the first measurement, the plate was incubated at 37 °C for 30 min protected from light (T2 measure). Chymotrypsin-like activity at T2 was determined by measuring the fluorescence released from the AMC substrates, using a PerkinElmer VICTOR 3 , at an Ex-355 and Em-460 nm. Jurkat cell lysate, with significant proteasome activity, was used as a positive control, and each experiment was performed in triplicate. To quantify proteasome activity, described as "one unit of proteasome activity is defined as the amount of proteasome which generates 1 nmol of AMC per minute at 37 °C", the manufacturer's instructions were followed. First, at each T (T1 or T2), the fluorescence values from the wells without inhibitor were subtracted to the fluorescence values from the wells with inhibitor, to obtain tRFU (total relative fluorescence unit). Measurement of the well without the proteasome inhibitor showed total proteolytic activity, and the wells containing proteasome inhibitor showed non-proteasome activity. Then, delta RFU = tRFU2-tRFU1 was calculated. Delta RFU values were applied to the AMC standard curve to obtain B, which is the amount of AMC in the sample well expressed as pmol/well. Proteasome activity was obtained by: where B is the amount of AMC (pmol) in the sample, calculated by the AMC standard curve. V is the total volume reaction (μl) in the well; T1 and T2 are the time (min) of the first and second readings, respectively. D is the sample dilution factor.
Proteasome (trypsin-like and caspase-like) activity assay. Proteasome activity assay to evaluate trypsin-like and caspase-like activity of the proteasome in immortalized human cardiomyocytes was used. Trypsin-like and caspase-like activities were determined utilizing an AMC-tagged peptide substrate, which releases free highly fluorescent AMC (7-amido-4-methyl coumarin) in the presence of proteasome proteolytic activity. The assay was performed in the presence and absence of bortezomib (Santa Cruz Biotechnology, Dallas, TX, USA) proteasome inhibitor. Briefly, cells were plated (seeding density 2.5 × 10 4 cells/cm 2 ) in 10% FBSsupplemented medium w/o phenol red into a flask T75 cm 2 . The next day, the culture medium was replaced with 1% FBS-supplemented medium w/o phenol red, this culture medium was daily replaced until the seventh day of culture. On the seventh day of culture, the cells were exposed, for 24 h, to 100 and 200 μM of NSAIDs in 1% FBS-supplemented medium w/o phenol red. After exposure to treatments, the cell monolayer was detached by trypsin and centrifuged 6 min at 250 g. The cell pellets were washed with cold PBS and transferred into 1.5 ml tubes, then centrifuged 6 min at 250 g. Proteasome lysis buffer (50 mM Tris-Hcl pH 7.5, 250 mM sucrose, 5 mM MgCl 2 , 0.5 mM EDTA free acid, 1 mM DTT, 2 mM ATP, 0.025% digitonin, 10% glycerol, all chemicals were purchased from all chemicals purchased from Sigma, St. Louis, Mo, USA) in Milli-Q-water was used to suspend cell pellets, to obtain crude protein extract. About 120 μl, for cell pellets, of proteasome lysis buffer were used. After homogenization by pipetting up and down ten times, the extracts were incubated 10 min at 4 °C. After centrifugation 30 min at 20,000 g (by refrigerated mini centrifuge), the supernatants (protein crude extract) were collected and maintained at 4 °C, ready for the assay. The total protein content was determined by extrapolation from a BSA standard curve (0.025-2 mg/ml To quantify specific proteasome activity at each T (T1 or T2), the fluorescence values from the wells without inhibitor were subtracted to the fluorescence values from the wells with inhibitor, to obtain total relative fluorescence unit (tRFU). Measurement of the well without proteasome inhibitor showed total proteolytic activity and the wells containing proteasome inhibitor showed non-proteasome activity. Then www.nature.com/scientificreports/ delta RFU = tRFU2-tRFU1 was calculated. Delta RFU values were applied to the AMC standard curve to obtain B, which is the amount of AMC in the sample well, expressed as pmol/well. Proteasome activity was obtained by: where B is the amount of AMC (pmol) in the sample, calculated by the AMC standard curve. V is the total volume reaction (μl) in the well, T1 and T2 are the time (min) of the first and second reading, respectively.

Characterization of proteasomes. Characterization of different proteasome structures was done by
native gel electrophoresis/western blotting (5%), as already described 33,38 . Briefly, immortalized human cardiomyocytes were plated (seeding density 2.5 × 10 4 cells/cm 2 ) in 10% FBS-supplemented medium w/o phenol red into a flask T75 cm 2 . The next day, the culture medium was replaced with 1% FBS-supplemented medium w/o phenol red, this culture medium was daily replaced until the seventh day of culture. The cells were exposed, for 24 h, to 100 and 200 μM of NSAIDs in 1% FBS-supplemented medium w/o phenol red. Later, the cell pellets were collected and homogenized in lysis buffer containing: 50 mM Tris-HCL pH 7.5, 5 mM MgCl 2 , 0.5 mM EDTA, 2 mM ATP and 0.5% NP-40 (tergitol) in Milli-Q-water. For cell homogenization, pellets were vortexed for 5 min at 4 °C and then incubated for 30 min at 4 °C. After centrifugation at 15,000 g and 4 °C for 30 min the supernatants were collected and stored at − 20 °C. The total protein content was determined by extrapolation from a BSA standard curve (0.025-2 mg/ml). Later, the samples were diluted 1:1 in native sample buffer (Bio-Rad) and then were loaded in native gel condition. The run gel was composed of stacking upper gel (3.5%) and resolving gel (5%) with freshly added 1 mM ATP. Electrophoresis was carried out in TBE buffer (90 mM Tris, 90 mM Borate, 0.5 mM EDTA and freshly added MgCl 2  (1:5000) (Abcam). Later, incubation with secondary HRP-conjugated anti-rabbit IgG antibody diluted 1:10,000 (Cell Signaling Technology, Danvers, MA, USA) was performed and the immunoreactive bands were visualized by ECL, according to the manufacturer's instructions. Bands from whole cell lysate obtained using Alliance 4.7 UVITEC (Cambridge, UK) were analyzed by ImageJ software, and values were given as relative proportion %.
OxyBlot assay. OxyBlot protein oxidation detection kit (Merck Millipore, Burlington, MA, USA) to evaluate oxidized protein levels in immortalized human cardiomyocytes were used according to the manufacturer's instructions. Briefly, cells were plated (seeding density 2.5 × 10 4 cells/cm 2 ) in 10% FBS-supplemented medium w/o phenol red into a flask T75 cm 2 . Later, the culture medium was replaced with 1% FBS-supplemented medium w/o phenol red, this culture medium was daily replaced until the seventh day of culture. On the seventh day of culture, the cells were exposed, for 24 h, to 100 and 200 μM of NSAIDs in 1% FBS-supplemented medium w/o phenol red. Subsequently, the cell pellets were collected and homogenized in lysis buffer (provided in the kit) containing DTT 50 mM. After protein extraction, 5 μg/μl of protein extracts were used to perform derivatization, as suggested by the manufacturer. Negative control (Neg. Ctr) is a sample without derivatization. Lysates from control and treated cells (20 µg total proteins per sample) were run on 10% polyacrylamide SDS denaturing gels, as previously performed by us 78 . The following primary antibody was used: rabbit anti-DNP diluted 1:150. After incubation with secondary HRP-conjugated anti-rabbit IgG antibody diluted 1:300 the immunoreactive bands were visualized by ECL, according to the manufacturer's instructions. Bands from whole cell lysate obtained using Alliance 4.7 UVITEC (Cambridge, UK) were analyzed by ImageJ software and normalized to blue Coomassie staining and values were given as relative units (R.U.).

Statistical analyses.
Data are expressed as mean ± standard error mean (SEM). Samples were processed by Graph Pad Prism 6 software (RRID: SCR_002798). Two-tailed unpaired student's t-test Welch-corrected was used to determine statistical differences among groups. A p value of < 0.05 was considered statistically significant.