Small molecule inhibitors of the mitochondrial ClpXP protease possess cytostatic potential and re-sensitize chemo-resistant cancers

The human mitochondrial ClpXP protease complex (HsClpXP) has recently attracted major attention as a target for novel anti-cancer therapies. Despite its important role in disease progression, the cellular role of HsClpXP is poorly characterized and only few small molecule inhibitors have been reported. Herein, we screened previously established S. aureus ClpXP inhibitors against the related human protease complex and identified potent small molecules against human ClpXP. The hit compounds showed anti-cancer activity in a panoply of leukemia, liver and breast cancer cell lines. We found that the bacterial ClpXP inhibitor 334 impairs the electron transport chain (ETC), enhances the production of mitochondrial reactive oxygen species (mtROS) and thereby promotes protein carbonylation, aberrant proteostasis and apoptosis. In addition, 334 induces cell death in re-isolated patient-derived xenograft (PDX) leukemia cells, potentiates the effect of DNA-damaging cytostatics and re-sensitizes resistant cancers to chemotherapy in non-apoptotic doses.


Chapter 2: ClpXP inhibitor 334 impairs the oncogenic potential of leukemia cells. It is known
that dependent on the type of malignancy, ClpP overexpression in human cancer correlates with low patient survival 16 . Hence, the abundance of this protease might also dictate the susceptibility of cancer cells to ClpXPtargeting agents. Therefore, the protein expression of the ClpX chaperone and ClpP were assessed in different leukemia cell lines. Both subunits were most prominently expressed in the chronic myeloid leukemia (CML) cell line K562 and in T-cell acute lymphoblastic leukemia (T-ALL) Jurkat cells. ClpP and ClpX occurred to a lower extent in the T-ALL cell line CEM and the lowest abundance was found in the AML cell line HL-60 (Fig. 2a). Thus, the active HsClpXP inhibitors 319, 334, 339 and the SaClpXP inhibitor 335 were subsequently tested for their proliferation inhibiting potential in K562 and Jurkat cells. In both human leukemic cell lines 334 protruded as the most potent inhibitor of viability, whereas 335 appeared to be least effective (Fig. 2b). A reduction of cell viability by 334 was confirmed in HL-60 and CEM cells (Fig. 2c). Notably, a low ClpXP protein expression in HL-60 cells correlated with a low sensitivity towards 334 (IC 50

Chapter 3: 334 induces cell death in patient-derived xenograft (PDX) cells.
In order to assess the therapeutic relevance, 334 was tested in PDX cells of 5 AML and 2 ALL patients (Supplementary Table S1 online). Therefore, primary tumor cells isolated from the peripheral blood of patients were transduced to establish stable luciferase expression, injected into immune-compromised mice and the development of leukemia was monitored by bioluminescence in vivo imaging 17,18 . AML and ALL PDX cells were re-isolated and harnessed for further ex vivo analysis. Isolated peripheral blood mononuclear cells (PBMCs) were used as healthy control cells.
In fact, applying 334 significantly induced cell death in all tested patient-derived cells (Fig. 3a). The specific cell death of PDX cells and PBMCs was determined by flow cytometry via quantification of forward and side scatter plot signals (FSC-SSC). Spontaneous cell death was subtracted from therapy-induced apoptosis as previously described (Fig. 3b) 19 . However, no correlation between ClpXP protein expression with the 334-induced specific cell death was observed in PDX and PBMC cells (Fig. 3c). This suggests that 334 may have additional targets, the heterogeneous genetic background of PDX cells plays a role in the sensitivity to 334 or that its proapoptotic effect is partially dictated by cellular stress conditions.

Chapter 4: Promotion of cellular stress response and mitochondrial alterations by 334. So far,
the ClpXP inhibitor 334 has been shown to possess anticancer potential in a panoply of cancer cell lines derived from the hematopoietic system, liver or breast and in leukemic PDX cells. In order to gain further insights into 334's mechanism of action, we applied quantitative mass spectrometry (MS) using stable isotope labeling with amino acids in cell culture (SILAC). Labeled (light, medium or heavy arginine and lysine) K562 cells were treated with 334 for 24 h or 48 h following quantitative assessment of regulated proteins. Among the significantly altered proteins, mitochondrial stress mediators as well as proteins involved in transcriptional and translational regulation were identified (Fig. 4a,b). In detail, unfolded protein binding (NUDCD3), stress-related proteins (PPP1R12A; GPHN) and enzymes of the mitochondrial lipid and nucleotide metabolism (HMGCS1; DUT) were enriched upon 334 exposure. Notably, the antiapoptotic transcription factor POU4F3, which plays a role in    Tables S2-S5 online). Upon challenges of persistent mtROS production and potential mitochondrial DNA mutations, mitochondria ensure integrity and functionality by an intrinsic adaptive system, the mitochondrial unfolded protein response (mtUPR) 21 . Thereby, in parallel to NF-κB signaling, the mtUPR initiates transcription of the nuclear-encoded mitochondrial protease ClpP and mitochondrial chaperones, such as the heat shock protein 60 (Hsp60) 22,23 . However, when overactivated, the UPR leads to cell dysfunction and apoptotic cell death via the transcriptional mediator C/EBP Homologous Protein (CHOP) 24 . In fact, via quantitative real-time PCR we revealed a significant increase of CHOP mRNA at high 334 concentrations (30 µM) (Fig. 4c). In addition, both Hsp60 and ClpP protein expression were reduced in Jurkat cells (Fig. 4d), suggesting that ClpXP inhibition could promote an uncompensated mtUPR, which mediates cell dysfunction and apoptosis by CHOP expression.
Chapter 5: 334 leads to impairment of the mitochondrial respiratory chain. The induction of mitochondrial stress in response to 334 treatment suggests further investigations of the mitochondrial function and morphology. The electron transport chain (ETC), which is located within the inner mitochondrial membrane is the major production site of cellular ATP 25,26 . We found that short-term treatment with 334 diminished cellular ATP levels in Jurkat cells with higher efficiency then cell viability, indicating a specific effect of 334 on mitochondrial respiration (Fig. 5a). However, the mitochondrial network obtained no major alterations in cellular localization and the overall mitochondrial integrity remained intact upon ClpXP inhibition ( Supplementary  Fig. S5a,b online). At physiological conditions, the ETC produces low concentrations of reactive oxygen species (ROS) as a byproduct of mitochondrial respiration 27 . However, ETC deficiency leads to excessive mtROS production, which affects mitochondrial functionality and cell fate 28,29 . It has recently been shown that in order to avoid oxidative stress, ClpXP specifically removes damaged subunits of the ETC complex I 30 . In fact, mitochondrial ROS production was increased by 334 in a concentration-dependent manner (Fig. 5b). Further, high superoxide levels promoted oxidation reactions and protein carbonylation (Fig. 5c). Protein carbonylations are covalent modifications with carbonyl residues that occur upon oxidative stress and target proteins for selective degradation by the 26S proteasome 31,32 . In general, elevated protein carbonylation is accompanied by diminished ETC activity and a reduction of the mitochondrial membrane potential (MMP) without affecting mitochondrial number, area, or density 33 . In line, a loss of the MMP, which occurs as an initial event of intrinsic cell death, was observed already after 6 h of ClpXP inhibition and led to high levels of apoptosis at later time points (Fig. 5d) 25,34 . In summary, low ATP levels and high mtROS production indicate impairment of the ETC in an early phase of inhibitor treatment. A disturbed MMP, enhanced protein carbonylation, protein aggregation with activation of www.nature.com/scientificreports/ the mtUPR and the induction of intrinsic apoptosis are the consequence (Fig. 5e) 35,36 . In order to pinpoint the metabolic impact of 334 on reduced ETC activity, a mitochondrial stress test was performed in a panoply of leukemia cell lines. In fact, we found that 334 diminished basal respiration and the respiratory capacity of Jurkat, K562 and CEM cells (Fig. 5f, Supplementary Fig. S6 online). Next, we investigated if the ClpXP complex is the target of 334, which confers the inhibitory effect on the mitochondrial metabolism. Therefore, ClpP-deficient HEK293T clones (HEK-KO1 and HEK-KO2), which have been previously characterized, were compared to HEK-WT cells 30 . We found that ClpP-deficient cells are less sensitive towards a reduction of ATP levels by shortterm 334 treatment (3 h). However, a reduction of proliferation (72 h) might be partially caused by off-target effects (Fig. 6a,b). Further, the reduction of the spare respiratory capacity was shown to be partially mediated by targeting the ClpXP complex, as upon low-dose 334 treatment no significant reduction was observed in HEK-KO cells in contrast to HUH7-WT cells (Fig. 6c). In summary, even though, off-target effects cannot be excluded, we found a direct involvement of ClpXP in mediating the metabolic alterations of 334.

Chapter 6: Chemo-sensitivity of leukemia cells is enhanced by 334 treatment. Low-doses of 334
triggered mitochondrial stress in K562, Jurkat and CEM leukemia cells and may thus be harnessed to increase the cancer cells' susceptibility to apoptosis and to sensitize them towards chemotherapeutic treatment 37,38 . In fact, K562 and Jurkat cells were synergistically sensitized towards the tyrosine kinase inhibitor imatinib and the topoisomerase II inhibitor etoposide when applied in combination therapy with non-cytotoxic doses of 334 (10 µM). Further, 334 promoted synergistic cell death in CEM cells treated with vincristine and even resensitized vincristine-resistant (VCR)-CEM cells towards therapy (Fig. 7a) 39 . Of note, chemo-sensitivity towards etoposide or cytarabine was not enhanced in HL-60 cells, which we showed to be refractory towards 334 in viability and cell death assays ( Supplementary Fig. S7a online). To expand the clinical relevance to the chemotherapeutic re-sensitization of other malignancies, 334 was tested in a previously established sorafenib resist- www.nature.com/scientificreports/ ance HCC cell model 40 . The sorafenib-resistant HUH7-R HCC cells were refractory to 10 µM sorafenib and obtained broad chemotherapeutic cross-resistance. Further, HUH7-R cells have been shown to possess ER stress with enhanced proteasomal degradation of mitochondrial proteins and are therefore especially vulnerable to the inhibition of proteostasis which is targeted by the ClpXP inhibitor 334 40 . Interestingly, a combination of sorafenib with 334 achieved cytotoxicity in HUH7-R cells comparable to the parental, non-resistant HUH7-WT cell line and obtained a stronger growth reduction than a combination with doxorubicin, cisplatin, vincristine or gefitinib ( Fig. 7b; Supplementary Fig. S7c online). The chemo-sensitizing effect of 334 is thereby comparable to a combination treatment with the Lon-protease inhibitor bardoxolone methyl (CDDO-ME), which is known to target the mitochondrial Lon protease and affect mitochondrial proteostasis ( Supplementary Fig. S7c online) 41 . Further, in this sorafenib resistance cell model we confirmed the reduction of ATP production in an early phase after treatment with subsequent induction of apoptosis and reduction of cell viability, thus substantiating the anticancer potential of 334 via inhibition of the ETC (Fig. 7c). In order to substantiate the clinical impact of this chemo-sensitizing potential, PDX leukemia cells were treated with 334, a chemotherapeutic compound or a combination of both for 48 h. In fact, low-dose 334 treatment (10 µM) partially sensitized PDX cells to imatinib, etoposide and vincristine treatment (Fig. 7d).

Discussion
Cellular and mitochondrial proteostasis were recently reported as a new target for cancer treatment. The proteasome inhibitor bortezomib (VELCADE) is clinically approved for patients with multiple myeloma as first-line treatment and second-line therapy for relapsed or refractory disease and in cytostatic combination regimens 42,43 .
In addition, mitochondrial chaperones increasingly drew attention as the Hsp90-binding small molecule gamitrinib was shown to impair stability of the ETC and regulate mitochondrial cell death 44,45 . Hsp60, on the other hand, supports folding and assembly of protein precursors in the mitochondrial matrix and restores proteostasis in the course of the mtUPR 21,46 . Interestingly, in a large number of tumors, Hsp60 crosstalks with the mitochondrial apoptosis machinery and inhibits tumor suppressors 47 . Hence, mitochondrial chaperons evoked as promising target structures to counteract the development and progression of malignancies. Herein, we focus on another crucial part of the mitochondrial proteostasis maintaining network, the protease ClpXP, as a novel and druggable target for cancer therapy.
Recently, involvement of ClpXP was demonstrated in tumorigenesis and progression of AML, prostate, and breast cancer 4,48 . However, thus far predominantly electrophilic inhibitors such as beta-lactone and phenylester have been reported 4,11 . In general, these ClpP subunit inhibitors suffer from limited stability and therefore weaker effects on tumor proliferation and clonogenic growth was observed in comparison with previous gene silencing experiments 11,48 . We hypothesized that ClpXP inhibition by compounds with enhanced plasma stability could be a superior strategy. Repurposing of bacterial ClpXP inhibitors revealed 334 to block the reconstituted mitochondrial protease with anti-proliferative effects on cancer cell-lines. Although the engagement with other cellular targets cannot be excluded, we observed a correlation between low ClpXP protein expression in HL-60 and corresponding poor sensitivity towards the ClpXP inhibitor 334, suggesting a selective effect on cells expressing the target protein. However, PDX and HUH7-R cells revealed no obvious correlation between ClpXP levels and 334 sensitivity, which may additionally be attributed to further off-targets as well as cellular stress conditions paired with an insufficient antioxidant defense 40 . Moreover, ClpP /-KO cells showed impaired proliferation and were less sensitive to the reduction of ATP production by 334 treatment compared to the parental HEK-WT cell line, emphasizing that 334 directly targets the ClpXP complex and thereby impairs the mitochondrial metabolism. Nonetheless, the drug-mediated reduction of the basal respiration was comparable for HEK-WT and HEK-KO cells, indicating off-target cytotoxicity of 334.
Notably, PBMCs possess comparable ClpXP expression and 334 response as their leukemic PDX counterparts. We conclude that the use of 334 in cell death-inducing concentrations involves a high risk of adverse events and requires local or tumor-targeted drug application. However, in non-apoptotic doses, 334 treatment efficiently sensitizes leukemic cells towards imatinib, etoposide, or vincristine and reestablishes the sorafenib responsiveness of HUH7-R cells 49 . In summary, this promising chemo-sensitizing potential of 334 may be harnessed in combination therapies, in order to lower the drug dose and to minimize related side-effects.
In general, the transcription of chaperones and proteases, such as Hsp60 and ClpP, is induced by the mtUPR, which promotes cell survival by preventing aberrant proteostasis and organelle damage 21,50 . However, upon 334 treatment, leukemic cells exhibit low Hsp60 levels and ClpP expression is not increased, indicating insufficient prosurvival mtUPR signaling 16 . In addition, proteomics indicates that 334 promotes a cellular stress response, while attenuating DNA repair and RNA turnover 51 . Interestingly, the transcriptional regulator POU4F3, which potentially counteracts chemotherapy-induced apoptosis, was found to be upregulated upon both 24 h and 48 h of 334 treatment and may partially compensate the mtUPR 20 . Hence, a combined use of ClpXP inhibitors with DNA-damaging agents may optimize the obtained anti-tumor effects. However, optimal drug partners for 334 remain to be identified.
To date, the small molecule 334 faces major obstacles for clinical translation. First, the identified ClpXP inhibitors possess low cancer cell selectivity and potential off-target effects, which limit the application as monotherapy in cytotoxic doses and requires further structural optimization and target validation. Second, there is no clinical experience on the toxicity profile and potential side-effects of ClpXP inhibition in vivo. We hypothesize that a high mitochondrial mass in healthy tissues, such as liver, heart, and muscles, may correlate with a high susceptibility to ClpXP inhibitors. On the other hand, identifying malignancies which, according to their metabolic profile, likely respond to ClpXP inhibition, could be of particular interest for personalized cancer therapy. In fact, we found that HCC cells were more sensitive towards ClpXP inhibitors compared to leukemia cells and therefore suggest that especially tumors with a strong dependency on oxidative phosphorylation are druggable with high www.nature.com/scientificreports/ efficiency. This may further be harnessed for the eradication of stem-cell like and quiescent cancer cells, which often account for tumor recurrence, aggressiveness and therapy resistance [52][53][54] .
In summary, we identified novel human ClpXP inhibitors as potential drug scaffolds and present 334 as an efficient combinational agent to common cytostatics, in order to enhance their anti-cancer potential and to regain the sensitivity of chemo-resistant cancers. Culture conditions. For the cultivation of HUH7-WT, HUH7-R, HepG2, MCF7 and MDA cells, DMEM medium (PAN Biotech GmbH, Aidenbach, Germany) was used. HUH7-R cells were additionally cultured in the presence of 10 µM sorafenib to maintain resistance. HL-60 cells were cultured in IMEM medium (PAN Biotech GmbH, Aidenbach, Germany). K562, CEM and CEM/VCR-R cells were cultured in RPMI 1640 (PAN Biotech GmbH, Aidenbach, Germany) and the medium of Jurkat cells was additionally supplemented with pyruvate (100 mM). All media were supplemented with 10% FCS (PAA Laboratories GmbH, Pasching, Austria) and cells were cultured at 37 °C with 5% CO 2 in constant humidity. Before cell seeding, culture flasks, multiwell plates and dishes of adherent cell lines were coated with collagen G (0.001% in PBS, Biochrom AG, Berlin, Germany).

Materials and methods
Human ClpP and ClpXP assays. Human ClpX and ClpP were expressed and purified as described previously 64 . Human ClpXP activity measurements were carried out in a total reaction volume of 100 µL in black, flatbottom 96-well plates. All data were recorded in triplicates.   www.nature.com/scientificreports/ The residual hClpP peptidase activity was measured upon treatment with inhibitors by monitoring the cleavage of the fluorogenic substrate Ac-Ala-hArg-(S)-2-aminooctanoic acid-7-amino-4-carbamoylmethylcoumarin (Ac-Ala-hArg-2-Aoc-ACC, custom-synthesis by Bachem) as described previously 54 . In a black 96-well plate . Basal respiration, spare respiratory capacity and proton leak were assessed as described in (Supplementary Fig. S6a online). The FCCP titration for HEK-WT cells is shown in Supplementary Fig. S6b online (last panel) (*P < 0.05, **P < 0.01, ***P < 0.001, Oneway ANOVA, Tukey's Multiple Comparison Test, n = 3). In vivo amplification of patient-derived xenograft (PDX) cells. PDX acute lymphoid (ALL) and acute myeloid leukemia (AML) were generated and amplified as previously described 17,55 . In short, patient-derived cells were serially transplanted into the tail vein of 6-12 week old male or female NOD scid gamma mice (NSG; The Jackson Laboratory, Bar Harbour, ME, USA). At late stage leukemia, PDX cells were re-isolated from bone marrow and spleen, washed in PBS, and used for ex vivo studies using respective growth media 56.  www.nature.com/scientificreports/ FA in acetonitrile (ACN, buffer B). Samples were separated using a gradient raising buffer B from 5 to 22% in 112 min, followed by a buffer B increase to 32% within 10 min. Buffer B content was further raised to 90% within the next 10 min and held another 10 min at 90%. Subsequently buffer B was decreased to 5% and held until end of the run (total: 152 min). During sample separation ion transfer tube temperature was et to 275 °C and MS full scans were performed at 120,000 resolution in the orbitrap with quadrupole isolation. The MS instrument was operated in a 3 s top speed data dependent mode. The scan range was set from 300 to 1500 m/z with 60% RF lens amplitude. The automatic gain control (AGC) target was set to 200,000, the maximum ion injection time was 50 ms and internal calibration was performed using the lock mass option. Peptides with intensity higher than 5000 and charge state 2-7 were fragmented with higher-energy collisional dissociation (HCD) (30%). Dynamic exclusion time was set to 10 ppm low and high mass tolerance. MS2 scans were recorded in the ion trap operating in rapid mode. The isolation window was set to 1.6 m/z and the AGC target to 1.0e4 with maximum injection time of 35 ms. Ions were injected for all available parallelizable time. Data were analyzed using MaxQuant software using standard settings for SILAC quantification experiments 56 . Searches were performed against a UniProt database of Homo sapiens proteome (taxon identifier: 9606). Data were further processed using Perseus software 57 . Proteins only identified by site, in reverse database search or marked as potential contaminants were excluded from further analysis. Protein ratios were log 2 -transformed, the replicates were grouped and only proteins with valid values in at least 2 out of the 3 replicates were included in further analysis. Data were analyzed by two-sided one sample Student's t-test with Benjamini-Hochberg correction (false-discovery rate = 0.05). t-test significance was plotted over log 2 -fold protein ratio using GraphPad Prism.
Immunoblotting. Proteins were separated via SDS-PAGE 58  . Chemiluminescence was detected with the Chemidoc Touch Imaging system (Bio-Rad, Munich, Germany) and the protein expression was quantified using Stain-free technology and the Image Lab Software. This technique enables a quantification of the whole lane protein, and therefore can be used for the normalization of protein bands 59 . The protein expression was normalized to the protein load or the respective β-actin protein expression.

Cell viability assays. Cell viability was measured using CellTiter-Blue and CellTiter-Glo Luminescent Cell
Viability Assay (Promega, Madison, WI, USA) according to the manufacturer's protocol. Therefore, cells were seeded in 96-well plates and treated with DMSO or respective compounds in the indicated concentrations for the indicated time. Cellular ATP levels were measured via luminescence signal after the addition of CellTiter-Glo using an Orion II microplate luminometer (Titertek Berthold, Pforzheim, Germany). For the determination  www.nature.com/scientificreports/ of cell viability CellTiter-Blue reagent was added to the cells and after four hours the fluorescence intensity was measured using SpectraFluor PlusTM (Tecan, Crailsheim, Germany).
Apoptotic cell death. For the evaluation of apoptotic cell death, cells were treated as indicated for the given time points. Cell death analysis was done using two different approaches.
Apoptotic cell death. First, apoptosis rates were evaluated as described by Nicoletti et al. 60 . After treatment cells were harvested, washed with ice cold PBS and pelleted (600×g, 5 min, 4 °C), before being incubated with hypotonic fluorochrome solution containing propidium iodide (HFS-PI solution). Cells were then analyzed via flow cytometry. Apoptotic cells are represented as a broad sub-G1 peak in the histogram plot and were gated accordingly.
Specific cell death. In the second approach, forward and side scatter plot signals (FSC-SSC) were used to determine the cell death of PDX cells and PBMCs. The FSC signal is a measure for cell size, while the SSC signal can be used to evaluate the granularity of the cells. Cells were harvested and analyzed via flow cytometry. Apoptotic cells are characterized by shrinkage (reduced FSC signal) and higher particle content (increased SSC signal). Viable and dead cells were gated accordingly in a FSC-SSC dot plot.  www.nature.com/scientificreports/ cyanide 3-chlorophenylhydrazone (CCCP) (Sigma-Aldrich, Taufkirchen, Germany), which causes a disruption of the MMP, served as a positive control and was added during the staining process (50 µM). Red and green fluorescence was analyzed using flow cytometry and cells with intact MMP were gated in dot plots. For signal compensation BD CompBeads Anti-Mouse Ig, κ particles and AlexaFluor488/PE mouse IgG2b, κ isotype control antibodies (BD Biosciences, Heidelberg, Germany) were used according to the manufacturer's protocol.
Migration-modified Boyden chamber assay. Migrational ability of cells was assessed using the boyden chamber assay. HUH7 cells were seeded in 6-well plates and either left untreated or pretreated with the indicated compound for 24 h. After pre-incubation, 1 × 10 5 cells per condition were resuspended in growth medium containing the respective compounds and were transferred into collagen G coated Transwell Permeable Supports (8 µm pore polycarbonate inserts, Corning Inc., New York, NY). The transwell inserts were then placed into a 24-well plate containing 700 µL of DMEM (containing 10% FCS) and incubated for 16 h (37 °C, 5% CO 2 ). Migrated cells were stained with crystal violet and counted using the particle counter plugin of the ImageJ software.
Mitochondrial morphology. Fluorescence staining. Mitochondrial morphology was analyzed using Mi-toTracker Red CMXRos (Thermo Fisher Scientific, Waltham, MA, USA) according to the manufacturer's instructions. For this purpose, MDA-MB-231 cells were seeded in 8-well ibiTreat μ-slides (ibidi GmbH, Munich, Germany) and treated with respective compounds as indicated. After the treatment cells were incubated with growth medium containing staining solution (200 nM) for 30 min (37 °C, 5% CO 2 ), before mitochrondial morphology was evaluated via live-cell imaging with a Leica-SP8 confocal microscope (Leica Microsystems, Wetzlar, Germany) equipped with an incubation chamber (37 °C, 5% CO 2 , 80% humidity; okolab S.r.l., Pozzuoli, Italy). Cells were lysed and carbonyl groups in the protein side chains were derivatized to 2,4-dinitrophenylhydrazone (DNP-hydrazone) by incubation with 2,4-dinitrophenylhydrazine (DNPH). SDS-PAGE and immunoblotting against DNP were performed as described above. The DNP abundance was quantifies by normalization to the respective protein load.