IOX1 activity as sepsis therapy and an antibiotic against multidrug-resistant bacteria

Sepsis is caused by organ dysfunction initiated by an unrestrained host immune response to infection. The emergence of antibiotic-resistant bacteria has rapidly increased in the last decades and has stimulated a firm research platform to combat infections caused by antibiotic-resistant bacteria that cannot be eradicated with conventional antibiotics. Strategies like epigenetic regulators such as lysine demethylase (Kdm) has received attention as a new target. Thus, we sought to investigate the epigenetic mechanisms in sepsis pathophysiology with the aim of discovering new concepts for treatment. A transcriptome analysis of dendritic cells during their inflammatory state identified Kdm as a critical molecule in sepsis regulation. Next, 8-hydroxyquinoline-5-carboxylic acid (IOX1) ability to control endotoxemia induced by Lipopolysaccharide and bacterial sepsis was demonstrated. IOX1 has been shown to regulate endotoxemia and sepsis caused by Escherichia coli and carbapenem-resistant Acinetobacter baumannii and has also contributed to the suppression of multidrug-resistant bacterial growth through the inhibition of DNA Gyrase. These findings show that IOX1 could be a component agent against bacterial sepsis by functioning as a broad-spectrum antibiotic with dual effects.


IOX1 inhibits the immune response via DC maturation induced by LPS. To investigate whether
histone lysine methylation is a targetable therapeutic point, we checked the expression change of enzymes that are responsible for histone methylation and found that Kdm4a and Kdm6b are the most upregulated enzymes upon LPS treatment in dendritic cells (DCs) (Fig. S1). Next, we searched for specific inhibitors of Kdm4a and Kdm6b. Given the advantages of IOX1, we selected IOX1 from among a variety of Kdm inhibitors. The structure of IOX1 (8-hydroxyquinoline-5-carboxylic acid) is shown in Fig. 1A. The cytotoxicity of IOX1 was measured after treatment of dendritic cells with the indicated concentration of IOX1 in the same amount of DMSO and a positive control, H 2 O 2 . In the case of DMSO, up to 0.5% of the volume of DMSO used to dissolve IOX1 was found to be noncytotoxic, and in the case of IOX1, no cytotoxicity was observed below 200 μM (Fig. 1B). Therefore, the concentration of IOX1 for DC function analysis in the future should be less than 200 μM. To analyze the effect of IOX1 on LPS-induced DC activation, we investigated the changes in inflammatory cytokine mRNA expression and secretion ( Fig. S2 and Fig. 1C) and the expression of surface molecules as maturation markers (Fig. 1D). IOX1 significantly reduced the secretion of TNF-α, IL-1β, IL-6, IL-12p70 and IL-10 induced by LPS (Fig. 1C). The expression of surface molecules, such as CD80, CD86, MHC-I and MHC-II, induced by LPS was also significantly reduced (Fig. 1D). These results indicate that IOX1 can effectively regulate the immune activity of bone marrow DCs induced by LPS.

IOX1 suppresses the immune response by inhibiting Kdm4a.
To determine whether the effect of IOX1 is caused by inhibition of Kdm, we performed immunoprecipitation with histone H3 and Kdm4a, followed by Western blot. In correlation with the transcriptome data, Kdm4a was highly upregulated upon LPS addition   Fig. 2A,B). IOX1 significantly suppressed Kdm4a upregulation and restored H3K36me3 and H3K9me3 levels ( Fig. 2A,B). Changes in Kdm4a levels and target histone methylation induced by LPS and IOX1 were mimicked in the lungs of mice (Fig. 2C,D). These results suggest that the inhibitory effect of IOX1 on the immune response occurs via a mechanism of Kdm inhibition.
IOX1 anti-inflammatory effect on an LPS-induced endotoxicity mouse model. We initiated an in vivo study by validating the anti-inflammatory effect of IOX1. A 30-min pretreatment with IOX1 (20 mg/ kg, i.p.) prior to LPS (20 mg/kg, i.p.) injection ameliorated the survival of endotoxemia-induced mice. The survival rate increased within 18 h from 0% in mice injected solely with LPS to 40% in mice pretreated with IOX1 ( Fig. 3A). In mouse serum, the increased levels of TNF-α, IL-1β and IL-6 induced by LPS were significantly reduced by IOX1 administration (Fig. 3B). In lung tissues, the expression of TNF-α, IL-1β and IL-6 was increased by LPS injection, and infiltration of PMNs (polymorphonuclear leukocytes) was also significantly reduced by IOX1 administration (Fig. 3C,D). Furthermore, the levels of AST, ALT, BUN and creatinine, biomarkers of organ failure, that were enhanced by LPS injection were reduced by IOX1 administration (Fig. 3E). These results confirmed that IOX1 has an anti-inflammatory effect on the LPS-induced endotoxicity mouse model.  (Fig. 4A). The levels of the proinflammatory cytokines TNF-α, IL-1β, IL-6, IL-12p70, and IL-10 in the serum (Fig. 4B) and in the lungs (Fig. 4C) of the A. baumannii-derived sepsis mouse model treated with IOX1 were significantly reduced. H&E staining demonstrated that IOX1 could repress PMN (polymorphonuclear leukocyte) infiltration in the tissue of the corresponding sepsis mouse model (Fig. 4D) and could also mitigate AST, ALT, BUN and creatinine levels in the lungs of the same mice (Fig. 4E). Interestingly, IOX1 showed an effective inhibition of colony growth (CFU) of A. baumannii in the major organs. A total repression of colony numbers was observed in the lung, kidney and spleen, and a significant reduction was reported in the liver (Fig. 4F). Additionally, the endotoxin levels in mouse serum were also significantly reduced (Fig. 4G). Notably, IOX1 effects were reproducible in an E. coli K1 sepsis mouse model, with a survival rate of 60% (Fig. S3). These results show that IOX1 can act as an endotoxin removal agent and an effective inhibitory molecule against bacterial colony growth, resulting in effective treatment of A. baumannii-induced sepsis. Lungs from BALB/c mice were injected with 10 mg/kg LPS in the presence or absence of 20 mg/kg IOX1 overnight. Total protein extracts from DCs and lungs treated with IOX1 were used to detect Kdm4a (A,C), H3K9me3 and H3K36me3 (B,D) by Western blotting, Histone H3 was used as a loading control (full-length blots are presented in Figs. S10-S13); Immunoprecipitation (IP) assays were performed on DCs treated in the presence or absence of IOX1, using specific antibodies to Kdm4a and histone H3 as controls (fulllength blots are presented in Figs. S10 and S12). www.nature.com/scientificreports/ IOX1 antibiotic effect on gram-negative bacteria. Next, we sought to elaborate on the data obtained above (Fig. 4F) by investigating the potential antibacterial activity exerted by IOX1 against bacteria. First, we analyzed the time-kill kinetics profile of IOX1 against E. coli (DH5α) and A. baumannii following different addition of concentrations of 12.5, 25, 50 and 100 μg/ml at various checkpoint times of 0, 4, 6, 8, 10 and 12 h. IOX1 completely inhibited the growth of E. coli (DH5α) at up to 12 h at a concentration of 25 μg/ml, the same concentration for which antibiotic activity was shown for the positive control imipenem (Fig. 5A), and inhibited A. baumannii at a concentration of 50 μg/ml (Fig. 5B). IOX1 exhibited inhibitory growth effects on different types of gram-negative bacterium, S.enteritidis (Fig. S4A), S.typhimurium (Fig. S4B), K.pneumoniae (Fig. S4C), and P.aeruginosa (Fig. S4D), with complete bacterial growth inhibition at a concentration of 50 μg/ml at up to 12 h. The antibacterial effect of IOX1 was also confirmed for the gram-positive bacterium S. aureus (Fig. S5A) and colistin-resistant AB (Fig. S5B) in a concentration-dependent manner, with complete bacterial growth inhibition at a concentration of 50 μg/ml at up to 12 h and 8 h, respectively. The antibacterial activity against not only gram-negative bacteria but also gram-positive bacteria was shown. This antibacterial effect against different types of bacteria was further proven in a cecal content injection (CCI) mouse model (Fig. S6). These results indicate that IOX1 exhibits a bactericidal effect that led to a significant reduction in the bacterial colony count and endotoxin levels.

IOX1 antiseptic effect on an
IOX1 inhibits DNA gyrase activity similarly to quinolone antibiotics. To further understand the mode of action (MOA) of IOX1, we tested whether JIB-04, a well-known cell-permeable pan-Jumonji histone demethylase inhibitor, could also exhibit the same antimicrobial activity against E. coli (DH5α). However, unlike IOX1, JIB-04 showed a negligible effect (Fig. S7). These data suggest that the antimicrobial activity of IOX1 was not due to the cell permeability of histone demethylase inhibitors but probably to the ability of IOX1 to inhibit DNA gyrase, a mechanism feature of quinolone antibiotics (2), as IOX1 is structurally similar to this family of antibiotics (Fig. 1A). Therefore, a molecular docking study was performed to investigate the binding modes between IOX1 and E. coli DNA gyrase. The 3D structure of E. coli DNA gyrase was obtained from the Protein Data Bank (http://www.rcsb.org, PDB ID: 6RKW). IOX1 was docked into the active site of E. coli DNA gyrase using GOLD software 20,21 . The information for the initial IOX1-binding site in E. coli DNA gyrase for www.nature.com/scientificreports/ the docking simulation was provided from the recently published 3D structure for the complex system of the ligand and DNA gyrase 22 . The best docked conformation was selected based on cluster analysis and the GOLD fitness score (Fig. 6A,B). The best cluster number from 100 trials for IOX1 was 80, and the GOLD fitness score was 43.2 (Table S1). Our molecular docking results showed that IOX1 formed hydrogen bonds with the D1-T2 and D2-T2 bases of both DNA strands (Fig. 6C,D). IOX1 also bound with the D1-A3 and D2-A3 bases of the   www.nature.com/scientificreports/ DNA strands via hydrophobic interactions and electrostatic interactions. Interestingly, the models showed that IOX1 interacted mainly with the DNA and that the interaction with the DNA gyrase protein seemed to be very weak or minor. In conclusion, the molecular docking study clearly confirmed that IOX1 interacted mainly with the cleft of double-stranded DNA, which is bound to DNA gyrase, rather than with DNA gyrase itself. The overall binding pattern of IOX1 is in good agreement with the experimental data 22 . Furthermore, we performed a gyrase inhibition assay, and the relaxed circular PBR322 plasmid, which usually forms a supercoil structure in the presence of gyrase activity, was used as a substrate. Supercoils and relaxed plasmids were determined by agarose gel electrophoresis (Fig. 6E). Inhibition of gyrase activity increased the relaxed plasmid morphology by regulating the activity of subunit A in DNA gyrase 16 . In addition, it has been suggested that antibacterial activity is potentially demonstrated through inhibition of both subunit A and subunit B of DNA gyrase 12 ; moreover, our binding test confirmed the binding of IOX1 to the whole DNA gyrase (Fig. S8A) and precisely its binding to the DNA Gyrase Subunit B (Fig. S8B). Therefore; we tried to determine whether IOX1 directly regulates the activity of DNA Gyrase Subunit B by performing a DNA gyrase ATPase binding assay. This method is known to measure the conversion of NADH to NAD + by ATP hydrolysis and assess NAD + production via a 340 nm spectrogram 19 . DNA gyrase ATPase activity was reduced in the presence of IOX1 compared to that in the negative control (Fig. 6F). These results demonstrate that IOX1 is involved in both the A and B subunits of DNA gyrase, inhibiting overall gyrase activity.

IOX1 increases the bactericidal effect against carbapenem-resistant A. baumannii.
The antimicrobial activity of IOX1 was also tested against resistant strains of bacteria. Carbapenem-resistant A. baumannii (CRAB) was found to proliferate in the presence of high concentrations of imipenem; in contrast, IOX1 succeeded in inhibiting bacterial growth in a concentration-dependent manner at up to 8 h (Fig. 7A). Additionally, analysis of time-kill kinetics profiles of imipenem (carbapenem antibiotic), ciprofloxacin (quinolone antibiotic) and IOX1 showed that, unlike IOX1, imipenem and ciprofloxacin exhibited no inhibitory growth effects on multidrug (carbapenem and quinolone)-resistant (MDR) A. baumannii (Fig. S9A-D). Interestingly, IOX1 treatment with imipenem (12.5 μg/ml) suggested a synergetic effect on CRAB growth (Fig. 7B). This observation was further demonstrated when enhanced antimicrobial effects were observed by combining imipenem with IOX1 (12.5, 25, 50 and 100 μg/ml) (Fig. 7C). This result shows that IOX1 exerts a promising antibacterial effect on MDR bacteria.  www.nature.com/scientificreports/ mouse). Within 18 h, the survival rates of mice injected solely with CRAB and mice treated with imipenem were 0%. However, IOX1 was able to increase the survival rate to 80% and to maintain it at 60% for 36 h (Fig. 8A). Additionally, unlike imipenem, which showed no inhibitory effect, the levels of inflammatory cytokines, including IL-10, TNF-α, IL-1β, IL-6 and IL-12p70, in serum (Fig. 8B) and in the lung (Fig. 8C) were reduced by IOX1 administration. IOX1 was also found to be effective in reducing lung PMN infiltration (Fig. 8D) and in attenuating organ damage, as reflected by a reduction in the levels of AST, ALT, BUN and creatinine (Fig. 8E). IOX1-mediated bacterial clearance was also investigated, and the data showed an abrogated number of remaining bacteria in the liver, lungs, kidneys and spleen (Fig. 8F). Furthermore, IOX1 reduced the endotoxin levels in mouse serum that were originally induced by CRAB inoculation (Fig. 8G). These results show the promising therapeutic effects of IOX1 on sepsis triggered by carbapenem-resistant bacteria.

Discussion
Unlike other histone modifications, histone methylation was discovered relatively late, and study of the reversible reaction is still underway 23 . However, recent histone demethylase studies have been actively conducted in several fields; for example, Kdm6b is involved in the expression of comprehensive inflammatory genes in macrophages 24 and Kdm4a is markedly expressed in neuroectodermal stem cells to induce tumor development 25 . The most widely used 2OG analogs N-oxalyl glycine (NOG) and pancreatic ductal adenocarcinoma (PDAC), general inhibitors of 2OG oxygenase, have poor cell permeability 26 , requiring the use of prodrug diester derivatives, whereas IOX1 can be transferred to the cytoplasm and nucleus without ester modification 27 . Herein, we confirmed that IOX1 inhibited the induction of proinflammatory cytokines in LPS-stimulated DCs and, in parallel, inhibited the activity and expression of Kdm4a and Kdm6b induced by LPS (Figs. 1 and 2). In addition, IOX1 showed significant therapeutic effects on bacterial-induced mouse sepsis models, including those with antibiotic-resistant bacteria. In particular, the bacterial count in each organ caused by bacterial infection was demonstrated to be reduced (Figs. 4F, 8F, and S3F). Similar to the other derivatives, IOX1, one of the small molecules with improved physicochemical properties among 8-hydroxyquinolines, displayed direct antimicrobial activity. Remarkably, it showed broad antimicrobial effects on E. coli, Acinetobacter baumannii (AB), double antibiotic-resistant (MDR) www.nature.com/scientificreports/ bacterium (quinolone and carbapenem-resistant AB), colistin-resistant AB and gram-positive bacteria, thus demonstrating excellent antimicrobial performance. We elucidated the mechanism of action that modulates the broad-spectrum antimicrobial activity of IOX1 (Fig. 6). In addition to immunosuppressive activation of IOX1 through Kdms, we found that nonquinolone antibiotics rarely inhibit bacterial DNA gyrase, resulting in bacterial antimicrobial activity and anti-inflammatory responses. As a demonstration of dual function, we presented molecular modeling based on structural analysis that can be directly linked as direct and indirect evidence ( Fig. 6 and Table S1) and showed that IOX1 simultaneously binds to bacterial DNA gyrase subunit B through isothermal titration calorimetry (ITC) analysis (Fig. S8B). Its mode of action is unique in that it directly interacts with DNA gyrase to prevent its binding to DNA. This study provides the first structural explanation of the Escherichia coli DNA gyrase-targeting mechanism of IOX1. Developing sepsis drugs has been limited by the availability of a mouse model for drug screening, which is currently regarded as a challenge, comprehending that inflammatory conditions, such as trauma, burns, and infections, in mice are significantly different from those of the human immune system. Moreover, most studies are based on the host's immune system; therefore, it is very likely that the compounds that show efficiency in mice will not be as effective in humans because of the variation in the immune system response and the whole physiological system between the two species. Extensive studies on the introduction of Kdm or topoisomerase inhibitors have recently been initiated, but none have been shown to simultaneously inhibit Kdm and DNA gyrase, as IOX1 does. These results demonstrate that IOX1 can be used as a broad-spectrum antibiotic, with the possibility of expanding the opportunity to develop and improve new drugs.

Generation of murine bone marrow DCs (BMDCs).
Bone marrow from the tibias and femurs was used to isolate DCs after euthanasia of C57BL/6 mice following IACUC guidelines and in compliance with ARRIVE guidelines. All methods were carried out in accordance with relevant guidelines and regulations. Red blood cells were removed from the bone marrow using red blood cell lysis buffer, leaving only progenitor cells. These progenitor cells were placed into cell culture plates with RPMI 1640 (containing 10% FBS, 1% penicillin/ streptomycin, and 10 ng/ml of GM-CSF) and incubated for 6 d at 37 °C under a 5% CO2 atmosphere. After six days, precursor cells had differentiated into immature DCs. To produce mature DCs, immature DC were treated with 50 ng/ml LPS and incubated overnight at 37 °C under 5% CO 2 30,31 .
Detection of AST, ALT, BUN and creatinine in mouse serum. Serum levels of AST (aspartate aminotransferase), ALT (alanine aminotransferase), BUN (blood urea nitrogen) and Creatinine were analyzed using total laboratory automation (Hitachi, Japan) and TBA-200FR NEO (Toshiba, Japan) systems at the Konkuk University Medical Center 30,31 . Hematoxylin and Eosin (H&E) staining. Lungs were extracted from mice, fixed in 4% paraformaldehyde (PFA) solution, and dehydrated to prepare paraffin blocks. The paraffin blocks were then cut to a thickness of 5 μm to prepare tissue slides. Any remaining paraffin was removed with xylene. Tissue slides were then hydrated, stained with H&E, and dehydrated. After mounting the tissue slides, images were captured under a microscope 30,31 . www.nature.com/scientificreports/ Western blotting and immunoprecipitation (IP). Proteins (a total of 25 μg per sample) were separated by 10% sodium dodecyl sulfate polyacrylamide gel electrophoresis (SDS-PAGE) and transferred to polyvinylidene fluoride membranes. After blocking the membranes with 5% skim milk in Tris-buffered saline containing 0.05% Tween-20 (TBS-T), they were incubated with the indicated antibodies. After washing with TBS-T, the membranes were incubated with a secondary antibody conjugated to horseradish peroxidase and visualized using an enhanced chemiluminescence kit (Merck, Darmstadt, Germany). For immunoprecipitations (IP), the corresponding Ab was added to 1 ml each of the cellular extracts and incubated at 4 °C for 1 h on a rotator. Fifty microliters of a 50% slurry of prewashed protein A-agarose beads was then added to each sample, followed by incubation for an additional 12 h at 4 °C. The samples were washed four times in lysis buffer and subjected to Western blot analyses as described above 30,31 .
Determination of bacterial counts in organ tissues. At the time of sacrifice, the lungs, liver, and kidneys were removed aseptically and placed separately in 1 ml of sterile PBS. The tissues were then homogenized on ice using a tissue homogenizer under a vented hood. The lung, liver, and kidney homogenates were diluted with PBS to 1:1000. After plating 10 μL of each diluted sample onto LB agar, the plates were then incubated at 37 °C for 24 h. The numbers of colonies were then counted and used to assess the relative abundances of bacteria 30,31 . Detection of endotoxin in mouse serum. The levels of endotoxin in mouse sera were measured using a LAL chromogenic end point assay (Lonza Group Ltd., Allendale, NJ, USA), according to the manufacturer's recommendations. Mouse sera were diluted tenfold in endotoxin-free PBS before the assay. After subtracting background levels, the results were calculated relative to an E. coli endotoxin standard provided with the assay kit. The relative amounts of endotoxin in each sample are expressed as EU/ml 30,31 .
LPS-induced mouse endotoxemia model. The  After the 2 h, the serum was harvested for pro-inflammatory cytokine ELISA. Next day, these mice were sacrificed depending on animal ethics. The half of the lung was fixed by paraformaldehyde. And then, these lungs were sliced for making tissue slide and that slides were stained by hematoxylin and eosin. The other side of lung was homogenized using a Bullet Blender homogenizer (Next Advance, NY, USA). The pro-inflammatory cytokines in lung were measured by ELISA. The AST, ALT, BUN and creatinine levels in the serum were measured using total laboratory automation (Hitachi, Japan) and TBA-200FR NEO (Toshiba, Japan) systems. The study was carried out in compliance with the ARRIVE guidelines.
A. baumannii (KUMC ATCC 19606)-induced mouse sepsis model. week-old female BALB/c mice were intraperitoneally injected 20 mg/kg of IOX1 (Selleckchem). For survival rate analysis, after the 30 min, IOX1-injected mice were i.p. injected 1.9 × 10 4 CFU/mouse of A. baumannii (5 mice/group). The survival rate was observed for 36 h. For other analysis, after 30 min, IOX1-injected mice were i.p. injected 9.5 × 10 3 CFU/ mouse of A. baumannii (5 mice/group). After the 2 h, the serum was harvested for pro-inflammatory cytokine ELISA. Next day, these mice were sacrificed depending on animal ethics. The half of the lung was fixed by paraformaldehyde. And then, these lungs were sliced for making tissue slide and that slides were stained by hematoxylin and eosin. The other side of lung, liver, kidneys and spleens were homogenized using a Bullet Blender homogenizer (Next Advance, NY, USA). The pro-inflammatory cytokines in lung lysates were measured by ELISA. The remained bacteria in tissue lysates were diluted by PBS and incubated at LB agar plate during overnight. The AST, ALT, BUN and creatinine levels in the serum were measured using total laboratory automation (Hitachi, Japan) and TBA-200FR NEO (Toshiba, Japan) systems. The levels of serum endotoxin were conducted by limulus amebocytes lysate (LAL) assay method and the color reaction were measured at 405 nm. The study was carried out in compliance with the ARRIVE guidelines.

Carbapenem-resistant A. baumannii (CRAB)-induced mouse sepsis model. The 6-week-old
female BALB/c mice were intraperitoneally injected 20 mg/kg of IOX1 (Selleckchem). For survival rate analysis, after the 30 min, IOX1-injected mice were i.p. injected 1.4 × 10 4 CFU/mouse of CRAB (5 mice/group). The survival rate was observed for 48 h. For other analysis, after the 30 min, IOX1-injected mice were i.p. injected 7 × 10 3 CFU/mouse of CRAB (5 mice/group). After the 2 h, the serum was harvested for pro-inflammatory cytokine ELISA. Next day, these mice were sacrificed depending on animal ethics. The half of the lung was fixed by paraformaldehyde. And then, these lungs were sliced for making tissue slide and that slides were stained by hematoxylin and eosin. The other side of lung, liver, kidneys and spleens were homogenized using a Bullet Blender homogenizer (Next Advance, NY, USA). The pro-inflammatory cytokines in lung lysates were measured by ELISA. The remained bacteria in tissue lysates were diluted by PBS and incubated at LB agar plate during overnight. The AST, ALT, BUN and creatinine levels in the serum were measured using total laboratory automation (Hitachi, Japan) and TBA-200FR NEO (Toshiba, Japan) systems. The levels of serum endotoxin were conducted by limulus amebocytes lysate (LAL) assay method and the color reaction were measured at 405 nm.
The study was carried out in compliance with the ARRIVE guidelines.