Author Correction: Conjugates of urolithin A with NSAIDs, their stability, cytotoxicity, and anti-inflammatory potential

Urolithin A (UA, 1), a gut microbiota postbiotic metabolite is attributed to express interesting biological activities indicated by in vitro, in vivo and clinical studies. Due to its strong anti-inflammatory properties it is considered as a promising lead molecule for further drug development, however, its strong phase II metabolism, severely limits its oral application. Therefore, monoesterified UA derivatives with selected NSAIDs: ibuprofen (Mix 3a/3b), mefenamic acid (Mix 4a/4b), diclofenac (Mix 5a/5b) and aspirin (Mix 6a/6b) were designed. Performed array of stability assays indicated Mix 4a/4b as a most suitable candidate for further studies due to its exceptional stability in human plasma. Thus, we evaluated effects of Mix 4a/4b on cell viability as well as the impact on cytokines secretion in THP-1 derived macrophages and compared it to UA. At high concentration (50 µM) Mix 4a/4b expressed a cytotoxic effect, however at concentration of 5 µM it significantly suppressed TNF-α secretion, and significantly increased ani-inflammatory IL-10 secretion at 10 µM without affecting cell viability. This work has led to selection of a novel UA derivatives, which are stable in solutions and in human plasma as well as posess anti-inflammatory activity towards THP-1 macrophages at non-cytotoxic concentrations.

www.nature.com/scientificreports/ the purified metabolites should not be considered as postbiotics per se. Nonetheless, these products of bacterial metabolism are continuously studied, given that their biological activity often surpasses maternal compounds. As a perfect example of purified metabolites marked with high biological activity may serve vitamin B 12 , equol or 8-prenylnaringenin 5,6 . In recent years the gut microbiome-derived metabolites of ellagitannins (ETs) gained a significant attention as candidates for new bioactive molecules 7 . Plants containing high amounts of ellagitannins have been used in traditional medicine for centuries [8][9][10] . Ellagitannins exhibit several beneficial biological effects in vitro; however, their potential therapeutical use is limited by their low bioavailability. In vivo, after consuming products rich in ellagitannins, they are hydrolysed into hexahydroxydiphenoic acid, which undergoes lactonisation, resulting in the formation of ellagic acid (EA). EA is furtherly metabolised by gut microbiota to urolithins, characterised, contrary to ETs and EA, with good bioavailability 11,12 . Nonetheless, it should be noted that based on the ability to produce urolithins, 3 separate metabotypes can be distinguished in population: metabotype A producing as a final product only urolithin A (UA, compound 1, Fig. 1), metabotype B producing in addition isourolithin A and urolithin B, and metabotype 0 deprived of appropriate bacteria capable to produce urolithins 13 . Knowing that the administration of ETs does not result in the formation of UA in around 10% of the population (metabotype 0 non-urolithin producers) and UA producing bacterial species are not known yet, the supplementation of ET rich products does not benefit the large group of people 14,15 . The application of UA has been widely studied, for instance, in the context of age-related conditions, muscle function, cancer or inflammation and its healthpromoting and anti-inflammatory mechanism of action include induction of mitophagy, suppression of NF-κB signaling pathway and activation of aryl hydrocarbon receptor (AhR) signalling 16 .
Despite unambiguously confirmed pharmocological properties of UA, the pharmakokinetic studies revealed significant limitations in its utilisation as an active dietary supplement or active pharmaceutical ingredient for oral use 7 . The phase 2 metabolism occurring in the intestinal wall and the formation of glucuronide and sulfate conjugates significantly narrows the biological activity of maternal compound 17 . In consequence following the ingestion of either ellagitannins or pure UA they are present in bloodstream and tissues in a conjugated form. These limitations highlight the need for alternative strategies to obtain a pharmacologically sufficient concentrations of UA or its biologically active derivatives in human plasma.
Taking the key role of inflammation in the pathogenesis of various chronic diseases and dubious safety profile of currently used medications, there is a special need for novel, safe anti-inflammatory drugs 18,19 . The modification of the chemical structure of natural products is a well-known strategy in novel drugs development. It was presented that 28% of new drugs approved by the US Food and Drug Administration between 1990 and 2008 originated from naturally occurring substances 20 .
Researchers investigating the anti-inflammatory properties of naturally occurring compounds can use various in vitro models of immune modulation. THP-1 human leukemia monocytic cells that can be differentiated into THP-1 derived macrophages are commonly used as an in vitro inflammatory response model. THP-1 cells present several benefits comparing to human peripheral blood mononuclear cells (PBMC), with high speed of growth or homogeneous genetic background. On the other hand, small differences with PBMC within the expression of receptors or degree of inflammatory response were observed. Nonetheless, THP-1 monocytes and macrophages are considered as suitable for preclinical studies or screening purposes 21 .
Herein we proposed a synthesis of urolithin a derivatives (UADs), with an esterified phenolic hydroxyl group in 3-or 8-position with non-steroidal anti-inflammatory drugs (NSAIDs). Additionally, we performed stability assays of monoesterified mixes of UADs, and for the most stable one tested cytotoxicity and evaluated impact on TNF-α secretion in THP-1 derived macrophages.

Results and discussion
Synthesis of UADs. UA undergoes strong phase II metabolism, being conjugated to glucuronic acid by uridine 5′-diphospho-glucuronosyltransferases (UGTs) enzymes. Knowing that some of the NSAIDs present inhibitory activity towards UGTs, we decided to perform synthesis of esterified UA with ibuprofen, mefenamic acid, diclofenac or acetylsalicylic acid (aspirin) 22,23 . Compounds were chosen as representatives of diverse classes of NSAIDs (propionic acid derivatives, fenamates, acetic acid derivatives and salicylates) and marked with potential different isozyme-specific inhibitory activity toward UGTs 24 . So far there were no studies conducted on synthesis of monoesterified derivatives of UA. Due to the presence of two phenolic groups in UA molecule, initially attempts to conduct esterification resulting in selective obtaining 3-substituted isomer were undertaken. Three-stage strategy was applied to synthesise monoesterified UADs. Firstly, 2 was obtained, and in the next step of synthesis underwent esterification with a particular NSAID. The esterification process was followed by the demethylation of UAD deprived of unblocked -OH groups using AlCl 3 . Unfortunately, during demethylation, the cleavage of ester bound occurred, resulting in the formation of UA. Regardless of lowering the temperature of a chemical reaction or changing the demethylating agent on BBr 3 , the hydrolysis of ester bond (combined with demethylation of UADs in 8-position) repeated. Hence it was decided to synthesise mixtures of monoesterified UADs, with esterified -OH group in 3-or 8-position (Fig. 1). Mixtures of esters with ibuprofen, mefenamic acid, diclofenac and aspirin are referred to as Mix 3a/3b, Mix 4a/4b, Mix 5a/5b and Mix 6a/6b, respectively. In the case of synthesis of ibuprofen and aspirin esters, appropriate acyl chlorides were used, prepared previously according to the literature 25,26 . Mefenamic acid and diclofenac esters were synthesised using DCC/DMAP as a coupling system (Fig. 2).
Hitherto, only the synthesis of etheric derivatives of urolithins was performed 27 . Additionally, obtained compounds were deprived of free hydroxyl groups. In contrast to that study, the goal of our research was to synthesise monosubstituted, esterified UADs. The regioselective synthesis and investigation of monoesterified UADs will www.nature.com/scientificreports/ provide more detailed information regarding the structure-activity relationship and may benefit further searching of postbiotic metabolites-derived drugs.  www.nature.com/scientificreports/ Characterisation of UADs. After successfully synthesising UADs' mixtures, the identity of obtained compounds was confirmed using the HPLC-MS-DAD method and NMR spectroscopy. All pairs of UADs eluted either as single peaks (Mix 3a/3b, Mix 5a/5b, Mix 6a/6b) or with minimal differences in retention times (Mix 4a/4b) preventing the separation of isomers using C 18 stationary phases (Fig. S1). The chromatographic strategies to separate the isomeric 3-and 8-substituted monoconjugates were introduced using Phenomenex Kinetex Biphenyl column (150 mm × 2.1 mm, particle size 1.7 μm) and methanol as a mobile phase. No improvement of isomers separation was achieved, what excluded application of preparative HPLC separation using these conditions. However, since the obtained UADs mixtures are similar to the in vivo produced mixture of urolithin A 3-and 8-glucuronides it was decided to conduct further studies on their mixtures. Analogously to our study, the mixture of these phase II metabolites of UA was not separated using the HPLC method 28 . The analysis of mass spectra confirmed the identity of UADs, which gave specific pseudomolecular ions [M-H]at m/z 415, 450, 505 and 389 for Mix 3a/3b, Mix 4a/4b, Mix 5a/5b and Mix 6a/6b, respectively. Analogously to previously presented urolithin A glucuronides' NMR spectra, the downshifts for H2 and H4 of UADs esterified in 3-position and for H7 and H9 of UADs esterified in 8-position were observed (Fig. 3, Tables 1, 2) 28 . Furthermore, a comparison of the proton signals integration ratio revealed the proportion of isomers in mixtures of UADs. The ratio of 3-esterified UADs to 8-esterified UADs were 1:1.27, 1:1.58, 1:1.89 and 1:1.12 for Mix 3a/3b, Mix 4a/4b, Mix 5a/5b and Mix 6a/6b, respectively. Stability assays. The high susceptibility to degradation of novel lead compounds may negatively affect the potential usability of promising APIs (Active Pharmaceutical Ingredients). Furthermore, the evaluation of compounds' stability is often not included during in vitro assays. Therefore, the stability assays for Mix 3a/3b, Mix 4a/4b, Mix 5a/5b and Mix 6a/6b, were performed during our study. Results of stability assays are presented in Table 3. Conditions were selected on the basis of previously described stability-examination protocol and possible exposure of UADs on stress conditions during drug formulation, storage, oral or intravenous administration 29 .    were characterised with the highest tolerance to incubation in − 70 °C (7.8 ± 7.9% and 11,9 ± 3.0% respectively), while the Mix 3a/3b presented the highest degradation during the cycle of freezing and thawing − 87.5 ± 1.7%. UADs were tested for their stability in 1 M HCl, and based on received outcomes, they can be divided into two separate groups. Mix 3a/3b and Mix 6a/6b almost completely vanished after incubation for 24 h in 37 °C or 12 h in 80 °C in an acidic environment. Contrarily, Mix 4a/4b and Mix 5a/5b moderately tolerate 24 h incubation in 37 °C with 1 M HCl (49.4 ± 0.6% and 41.4 ± 4.9% degradation, respectively). Nonetheless, Mix 4a/4b and Mix 5a/5b were also undetected after incubation at a higher temperature for 12 h.
In terms of UADs' stability in basic conditions, all native compounds diminished completely or almost completely, no matter of time and temperature of incubation in 1 M NaOH.
UADs' susceptibility to oxidative stress varied widely, and Mix 3a/3b turned out to be the least affected by the incubation in 30% H 2 O 2 at 37 °C for 24 h with only 43.1 ± 8.9% degradation. At the other end of the scale, the level of Mix 6a/6b was decreased by almost 99% compared to the control. Unsurprisingly, none of UADs' peaks was detected after incubation at 80 °C for 12 h.
All UADs except Mix 4a/4b underwent strong degradation in human plasma incubated at 37 °C, resulting in the total absence of UADs' peaks after 4 h. Surprisingly, Mix 4a/4b was marked with good toleration to the human plasma environment with a ratio of degradation of 18.0 ± 6.1% after 4 h.

Cytotoxicity.
Taking into account the exceptional stability of Mix 4a/4b in human plasma, this derivatives were chosen for further in vitro assays. The cytotoxicity of Mix 4a/4b on THP-1 derived macrophages was tested using a standard MTT assay performed after administration of UADs, one hour resting and stimulation of cells with LPS for 3 or 24 h. Mix 4a/4b did not exhibit cytotoxic effect up to 10 µM (Fig. 4a,b, Tables S1, S2). However, when 50 µM concentration was used, cell viability significantly declined. Additionally, the impact of UA on THP-1 macrophages viability was assessed, and no cytotoxic effect was observed in the tested concentration range (2-50 µM), similarly to previously obtained results 8,17 . The paradoxical increased viability after 24 h incubation with 5 and 10 µM Mix 4a/4b may result from the dependance of MTT assay on metabolic rate 30 . Furthermore, the cytotoxic effect of 50 µM Mix 4a/4b was independently proved using Neutral Red Uptake (NRU) assay performed 3 and 24 h after LPS stimulation (Fig. 4c,d, Tables S3, S4).
Parallelly, the cytotoxicity of UA and Mix 4a/4b on THP-1 monocytes was investigated using flow cytometry analysis of PI-stained cells with 25 µM parthenolide as a positive control (Fig. 4e, Table S5). As previously described, 25 µM parthenolide decreased THP-1 viability 31 . Similarly to results obtained for THP-1 derived macrophages using MTT and NRU assays, Mix 4a/4b did not influence THP-1 monocytes viability up 10 µM. Impact of Mix 4a/4b on cytokines secretion. The TNF-α secretion of LPS-stimulated THP-1 derived macrophages was assessed using ELISA kits. UA was chosen as a control compound based on the well-established inhibitory activity of UA on TNF-α secretion and thoroughly examined the underlying intracellular mechanism of action 16 . As expected, UA exhibited an inhibitory effect on TNF-α secretion in a dose-dependent manner (Fig. 5a, Table S6). Mix 4a/4b affected TNF-α secretion 3 h after stimulation. While the significant inhibitory effect on TNF-α level was observed already at 5 µM Mix 4a/4b, this impact was not detected using concentration of 10 µM (Fig. 5a, Table S6). The lack of concentration-dependency of determined activity between 5 and 10 µM concentrations (not observed in the case of UA), can arise from the hybrydic properties of the Mix 4a/4b. As described above, UA exhibits an inhibitory effect on TNF-α secretion, affecting the NF-κB signalling pathway in cells. On the other hand, mefenamic acid non-selectively inhibits COX-2, and it is known, that inhibition of this enzyme can increase TNF-α expression in THP-1 macrophages 32,33 . The additional set of experiments using THP-1 derived macrophages confirmed the stimulating effect of mefenamic acid on TNF-α secretion (Fig. S7). Thus, the contrary effect of two Mix 4a/4b subunits might explain the dose-independent manner of action and will be closely evaluated in future research. Moreover, the cytotoxic effect of Mix 4a/4b at 50 µM concentration was presented. After administration of 10 µM Mix 4a/4b, we did not observe the decrease of the viability of THP-1 macrophages. Nevertheless, it was previously demonstrated that despite the not observed decline in THP-1 cells www.nature.com/scientificreports/ viability, the induction of apoptotic intracellular signalling pathways, including caspases activity and parallel upregulated production of proinflammatory TNF-α, were detected after administration of several stressors 34,35 . Additionally, the impact of UA and Mix 4a/4b on anti-inflammatory interleukin-10 (IL-10) and interleukin-6 (IL-6) secretion 24 h after stimulation with LPS was tested (Fig. 5b,c, Tables S7, S8). In accordance with previously published results, UA did not significantly affect IL-10 secretion 17 . Interestingly, Mix 4a/4b increased levels of IL-10 comparing to stimulated control in a dose-dependent manner, reaching statistical significance at a concentration of 10 µM. Consistent with hitherto presented results, 50 µM UA inhibited secretion of IL-6 36 . Contrary to this specific UA activity, Mix 4a/4b did not affect IL-6 secretion in THP-1 derived macrophages. While the mefenamic acid alone did not influence IL-10 secretion, it increased IL-6 production in a dosedependent manner (Fig. S7). These findings suggest distinctive mechanism of action of Mix 4a/4b from that previously determined for UA 37 .

Conclusion
In summary, we performed the successful synthesis of mixtures of UADs monoesterified to ibuprofen (Mix 3a/3b), mefenamic acid (Mix 4a/4b), diclofenac (Mix 5a/5b) and aspirin (Mix 6a/6b). Despite similarities in conjugation mode, obtained compounds fundamentally differ in terms of their stability, and contrary to other UADs, Mix 4a/4b well tolerated the human plasma environment. The determined impact of Mix 4a/4b on TNF-α secretion may result either from their cytotoxic effect at high concentration or the opposing activity of Mix 4a/4b subunits at lower concentrations. The influence of Mix 4a/4b on IL-10 secretion and lack of activity towards IL-6 secretion imply mechanism of action, which is different from that previously determined for UA 37 . Presented results create a novel approach in anti-inflammatory drug development, focusing on chemical modification of the structure of postbiotic metabolites. Additional studies should be oriented on the synthesis of monoesterified UADs and close investigation of their intracellular mechanism of action. Further work evaluating UADs bioavailability in vitro, with the special interest in the role of UGTs, are currently in progress. (2). The synthesis of the first intermediate, 2, was performed according to the previously described protocol 38 . Briefly, after heating under reflux mixture of resorcinol (1 g; 9.08 mmol, 2.1 eqv), 2-Bromo-5-methoxybenzoic acid (1 g; 4.33 mmol, 1.0 eqv) and NaOH (0.375 g; 9.38 mmol, 2.2 eqv) in water (4.5 mL) for 30 min, a 5% aqueous solution of CuSO4 (1.8 ml; 0.56 mmol, 0.1 eqv) was added dropwise. Afterwards, the solution was heated for additional 10 min. The obtained precipitate was filtered, dried and the identity of 2 was confirmed using the UHPLC-DAD-MS method; 610 mg (58%). (UA, 1). The goal of this procedure was to obtain mixtures of UADs with esterified 3-or 8-position. A solution of 2 (0.51 g; 2.11 mmol, 1 eqv) and AlCl 3 (1.46 g; 10.95 mmol, 5.2 eqv) in chlorobenzene was refluxed, and the demethylation process was monitored using the UHPLC-DAD-MS method. After complete demethylation of 2, resulting in the formation of UA, a cooled mixture was added to ice. In the next step, the mixture was extracted with diethyl ether (3 × 125 ml). Subsequently, the organic solvent was evaporated on a rotary evaporator equipped with a vacuum system to reduce the pressure. The identity of obtained UA was confirmed using the UHPLC-DAD-MS method; 196 mg (41%).

NMR.
The NMR spectra were recorded on a Bruker AVANCE (Bruker, Karlsruhe, Germany) spectrometer operating at 300 or 500 MHz for 1 H NMR and at 125 MHz or 75 MHz for 13 C NMR. The spectra were measured in CDCl 3 and are given as δ values (in ppm) relative to TMS, and coupling constants (J) are reported in Hz.
Stability assays. The stability assays under several stress conditions were performed according to modified, previously described protocol 29 . Detailed information regarding stability assays' conditions are described in Supporting Information. Samples were dissolved in DMSO to obtain 20 mg/ml concentration and added to aliquots to achieve 80 mg/ml concentration at the start of incubation. All experiments were performed in triplicates (n = 3). The degree of degradation was calculated as follows: THP-1 derived macrophages. THP-1 human monocytic cells (DSMZ, Braunschweig, Germany) were studied between passages 4 and 19. Monocytes were differentiated into THP-1 derived macrophages according to modified, previously described protocol 17 . THP-1 monocytes were seeded in 24-well plates at a density of 400,000 cells per well. Cells were incubated at 37 °C under humidified 5% CO 2 in the culture medium containing RPMI 1640 without phenol red supplemented with 10% FBS and 2 mM glutamine. In order to differenti- MTT assay. THP-1 monocytes were seeded in 24-well plates at a density of 400,000 cells per well and differentiated into THP-1 derived macrophages according to the method described above. After treatment with the test compounds for 1 h and subsequent stimulation with LPS for 3 or 24 h, cells were washed with warm PBS. Afterwards, a 1 mL of solution of 0.5 mg/mL MTT in culture medium was added, and macrophages were incubated at 37 °C under humidified 5% CO 2 for 30 min. After medium removal, crystals were dissolved in 400 µL DMSO, 200 µL of solution was transferred to 96-well plate and the absorbance of the solution was measured at λ = 570 nm with the correction to 630 nm using a microplate reader. The relative cell viability was shown as the experimental group's ratio to stimulated cells.
Neutral red uptake (NRU) assay. NRU assay was performed according to modified, previously described protocol 39 . 40 mg of neutral red dye in was dissolved in 10 mL distilled water to obtain 4 mg/ mL neutral red stock solution. Subsequently, 140 µL of neutral red stock solution was added to 14 ml of culture medium and sterile filtered in order to obtain 40 µg/mL neutral red medium. THP-1 monocytes were seeded in 24-well plates at a density of 400,000 cells per well and differentiated into THP-1 derived macrophages according to the method described above. After treatment with the test compounds for 1 h and subsequent stimulation with LPS for 3 or 24 h, cells were washed with warm PBS. Afterwards, a 500 µL of warm 40 µg/mL neutral red medium was added and macrophages were incubated at 37 °C under humidified 5% CO 2 for 2 h. After this time cells were washed with PBS and 500 µL of neutral red destain solution was added and the plate was placed in plate shaker for at least 15 min. The absorbance of the solution was measured at λ = 530 nm with the correction to 645 nm using a microplate reader. The relative cell viability was shown as the experimental group's ratio to stimulated cells. The experiment was performed in triplicate (n = 3).

Enzyme-linked immunosorbent assay (ELISA).
After treatment with the test compounds for 1 h and subsequent stimulation with LPS for 3 h (for TNF-α secretion measurements) or 24 h (for IL-10 and IL-6 secretion measurements) supernatants were collected. The secretion of cytokines levels was measured by ELISA kits according to the manufacturer's instruction (Human TNF-α, Human IL-10 and Human IL-6 ELISA kits, BD Biosciences, Franklin Lakes, NJ, USA). The relative cytokines levels secretion was shown as the experimental group's ratio to LPS stimulated cells.
Statistical analysis. All in vitro experiments were performed in triplicates (n = 3), at least from 3 independent experiments, if not otherwise stated. The significance of obtained differences was determined using one-way ANOVA, and post hoc Dunnett's test was used to compare results with control groups. Tukey's post hoc test was performed to compare differences in inhibitory activity between groups. The p value < 0.05 was considered statistically significant. Statistica 13 software was used to perform all analyses.

Data availability
All data generated or analysed during this study are included in this published article (and its supplementary information files).