Design and Synthesis of Novel Xyloketal Derivatives and Their Protective Activities against H2O2-Induced HUVEC Injury

In this work, we designed and synthesized a series of amide derivatives (1–13), benzoxazine derivatives (16–28) and amino derivatives (29–30) from xyloketal B. All 28 new derivatives and seven known compounds (14, 15, 31–35) were evaluated for their protection against H2O2-induced HUVEC injury. 23 and 24 exhibited more potential protective activities than other derivatives; and the EC50 values of them and the leading compound 31 (xyloketal B) were 5.10, 3.59 and 15.97 μM, respectively. Meanwhile, a comparative molecular similarity indices analysis (CoMSIA) was constructed to explain the structural activity relationship of these xyloketal derivatives. This 3D QSAR model from CoMSIA suggested that the derived model exhibited good predictive ability in the external test-set validation. Derivative 24 fit well with the COMSIA map, therefore it possessed the highest activity of all compounds. Compounds 23, 24 and 31 (xyloketal B) were further to examine in the JC-1 mitochondrial membrane potential (MMP) assay of HUVECs using flow cytometry (FCM). The result indicated that 23 and 24 significantly inhibited H2O2-induced decrease of the cell mitochondrial membrane potential (ΔΨm) at 25 μM. Collectively, the protective effects of xyloketals on H2O2-induced endothelial cells may be generated from oxidation action by restraining ROS and reducing the MMP.


Introduction
Cardiovascular disease (CVD) has drawn significant attention in recent years because it has become the leading cause of mortality worldwide, affecting people from every income level. Reactive oxygen species (ROS), including H2O2, OH − , NO, and ONOO − , play key roles in the pathogenesis of many CVDs, such as hypertension and atherosclerosis (As). The ROS-induced oxidative stress in cardiac and vascular is closely connected with the endothelial dysfunction in disease initiation and progression. Reactive oxygen species (ROS) are generated under pathological conditions, such as ischemia-reperfusion and inflammation, and activate pro-apoptotic and anti-apoptotic signaling programs in endothelial cells [1]. As one of the most common ROS, hydrogen peroxide (H2O2) can easily cross the plasma membrane, produce a highly reactive radical OH· , and lead to cell and tissue damage [2,3]. The generation of H2O2 plays a key role in the atherosclerotic progression. H2O2 mediates various cellular responses. Direct or indirect stimulation by H2O2 due to its intracellular production could activate various cellular pathways, including calcium release, protein tyrosine kinase, mitogen-activated protein kinases (MAPKs), transcription factor NF-κB, and the induction of cell apoptosis [4][5][6][7]. Thus, H2O2 has been extensively used as an oxidative stimulus to induce oxidative stress in in vitro models. As the major type of endothelial cells, human umbilical vein endothelial cells (HUVECs) are commonly accepted as a model cell to explore the mechanisms involved in the pathogenesis of CVDs [8].
Mitochondrion serve as a pivotal decision center in many types of apoptotic response: they release a variety of death-promoting factors from their inter-membrane spaces into the cytosol, triggering an increase in mitochondria permeability and leading to consequences of mitochondrial dysfunction (e.g., disruption of the mitochondrial membrane potential ΔΨm) [9,10]. Mitochondria are considered the main source of ROS in the cell. Unless adequately detoxified, superoxide causes mitochondrial oxidative stress and may contribute to a decline in mitochondrial function.
Xyloketals are a type of novel compounds that possess unique molecular structures. They are isolated from the marine mangrove fungus Xylaria sp. (#2508) (Chart 1) [11,12]. We previously demonstrated that xyloketal B has protective action against a variety of pathophysiological stimuli, such as oxLDL, oxygen-glucose deprivation (OGD) and 1-methyl-4-phenylpyridinium (MPP+), in different disease models [13][14][15][16][17][18]. Thus, xyloketal B might be a good candidate for further development as an antioxidant medicine in cardiovascular diseases. However, its clinical development may be difficult due to water insolubility. Structure-activity relationship analyses in previous reports have demonstrated that the characteristic substituted groups at the C-12 or C-13 position of xyloketal B are key functional groups for its antioxidative effect. To improve the solubility and biological activity of xyloketal B, some amino groups can be introduced at the C-12 or C-13 position of this type of structure, and the corresponding acid salts could be prepared in the future. Because of the complexity of the stereoselective synthesis of xyloketals, it is difficult to provide a significant amount of optically pure samples for biological activity evaluation. We decided to begin the studies using racemic xyloketal B. In this paper, we designed and synthesized a new series of derivatives (Chart 2) from xyloketal B, including a series of C-13 xyloketal amide derivatives (1)(2)(3)(4)(5)(6)(7)(8)(9)(10)(11)(12)(13); xyloketal benzoxazine derivatives (16-28) using a one-pot reaction of xyloketal B, formaldehyde and different primary amines; and xyloketal amino derivatives (29-30) that C-13 substituted using different secondary amines. All 28 new derivatives and 7 known compounds (14,15,(31)(32)(33)(34)(35) were evaluated for their protection against H2O2-induced HUVEC injury. Then, a comparative molecular similarity indices analysis (CoMSIA) was constructed using the SYBYL programming package (version 7.3.5) to explain the structural activity relationship of these xyloketal derivatives [19,20]. The training set and test set were randomly divided out of a total of 35 molecules. A training set of 30 molecules was used to construct the QSAR model, and a training set of five molecules was used to validate it. Mitochondria are considered the main source of reactive oxygen species (ROS) in cells [21,22]. Therefore, we investigated whether xyloketals could protect mitochondria through inhibition of ROS. Any compound with high antioxidative action was further investigated in the JC-1 mitochondrial membrane potential (MMP) assay of HUVECs using flow cytometry (FCM).

Chemistry
The general synthetic routes of compounds 1-35 are outlined in Schemes 1-3. All the new compounds were prepared from xyloketal B and xyloketal B acid that were gained from synthetic way in the ordinary state without any asymmetric factors [16]. New xyloketal amides 1-13 were obtained via a condensation reaction between xyloketal B acid and the corresponding amines in the presence of (benzotriazol-1yloxy)tris(dimethylamino)phosphonium hexafluorophosphate (BOP) and N,N-diisopropylethylamine (DIEA) (Scheme 1). Interestingly, the one-pot Mannich reaction of xyloketal B, formaldehyde and different primary amines afforded a series of novel xyloketal derivatives 16-28 bearing an 1,3-oxazine moiety. Instead of primary amines, the Mannich reaction of secondary amines, xyloketal B and formaldehyde generated C-13 substituted amine derivatives 29-30 (Scheme 3). All the new compounds were fully characterized using MS and NMR. Moreover, all of the examined compounds were synthesized as racemic mixtures from synthetic xyloketal B and xyloketal B acid and no asymmetric synthesis was applied. Their stereo features are the same as xyloketal B and xyloketal B acid. The stereochemistry of these xyloketal derivatives was complicated. In principle, the two oxygen-containing pyran and furan rings B and C can be connected in a cis or trans fashion. The methyl group at C-5 or C-5′ could be cis or trans with respect to the stereogenic centers at the junction at C-2 or C-2′ and C-6 or C-6′. However, previous studies indicated that rings B and C or B′ and C′ were cis for all condensations leading to xyloketal derivatives in the natural and synthetic compounds [18,[23][24][25][26][27][28][29], thus only two sets of stereoisomers of xyloketals can be formed: syn, anti and syn, syn types. Moreover, C-2/C-5 methyl in cis orientation occupied dominant position both in experimental and theoretical results [29]. We previously also reported that synthetic xyloketal B and xyloketal B acid were characterized as mixtures of stereoisomers, including the enantiomers and diastereoisomers [16,18,30], and the ratio of two sets of diastereoisomers syn, anti and syn, syn was about 1:1 via NMR analysis. Similar to our previous studies, at this time, racemic mixtures of all new derivatives were consisted of two sets of diastereoisomers (Chart 3, syn, anti a and syn, syn b). Every diastereoisomer had four enantiomer pairs depending on C-5/C-5′ methyl in cis or trans, and the isomer with C-2/C-5 and C-2′/C-5′ methyl all in cis may take greatest proportion in these four enantiomer pairs in according to previous studies. The very close relationship of the diastereoisomers syn, anti and syn, syn was evident in NMR spectra. Though with overlapping of nearly identical sets of signals, two sets of signal peaks could still be detected in 1 H and 13 C NMR spectra assigned to isomers syn, anti and syn, syn with approximate ratio of ~1:1. Taking 24 as an example, both the 1 H and 13 C NMR spectra of 24 showed evidence of the diastereoisomers 24a and 24b ( Figure 1). Obviously, in 1 H NMR, methyl (10 and 10′) at C-2 and C-2′ showed as two peaks respectively (δ = 1.49, 1.48, and 1.52, 1.50 ppm) relative to two single peak (δ = 1.50 and 1.52 ppm) of 10 and 10′ in the natural xyloketal B [11] (Figure 1A), in addition, the integrals of the hydrogen atoms of two peaks indicated an approximately 1: 1 ratio of diastereoisomers 24a and 24b. The 13 C NMR spectrum was more instructive ( Figure 1B). The methyl (C-10, C-10′) and (C-11, C-11′) both presented as four closely packed peaks (δ = 23.0, 22.9, 22.8, 22.6 and 16.0, 15.9, 15.9, 15.8 ppm). Moreover, the aromatic carbon atom (C-13) also appeared as two peaks (δ = 98.9 and 99.0 ppm). These peaks all proved that compound 24 consisted of two sets of diastereoisomers. However, the enantiomers could not be found by NMR analysis because of their identical NMR spectra. Separating these stereoisomers via chromatography was very difficult. Therefore, all xyloketal derivatives were used directly in the biological screening without separating the stereoisomers this time. These compounds possessed the same structural framework; the only differences were different substituents at the C-12 or C-13 position of the aromatic ring. Although the test compounds are enantiomeric and diastereomeric mixtures, their activities and SAR analysis could be obtained.

Xyloketal Derivatives Protected Endothelial Cells against H2O2-Induced Injury Assay
The apoptosis of HUVECs caused by ROS has been implicated in numerous pathophysiological processes of CVDs. An important source of endogenous ROS is generated from H2O2, and it has been proven that ROS are involved in the apoptosis of ECs [31,32]. Using a similar culture system, we have shown that xyloketals had no significant effects on cell viability up to 100 μM in the MTT assay. Accordingly, the protection of xyloketals 1-35 at concentrations of 1 and 10 μM was applied for the following H2O2 (600 μM)-induced injury of HUVECs, and apocynin (1 and 10 μM) was used as a positive control. The results (Table 1, Figure 2) showed that some compounds exhibited strong antioxidative activities, in both morphological changes and inhibition of cell apoptosis. Among them, benzo-1,3-oxazine xyloketal derivatives 23 and 24 displayed greater potential protective activities than other derivatives with cell viabilities 83.07% and 86.08%. Furthermore, to evaluate the activities of these two most significant compounds clearly, the EC50 values in HUVECs ranged from 1-50 μM of 23, 24 and the leading compound xyloketal B (31) were determined with 5.10, 3.59 and 15.97 μM, respectively (Table 2). Thus, new candidates with amino groups, which could be prepared the corresponding acid salts in the future to improve their water-insolubility, will be the promising compounds for further evaluation in the treatment of cardiovascular diseases.   HUVECs were pre-incubated with 10 μM xyloketal derivatives (blue bars) or 1 μM xyloketal derivatives (red bars) for 30 min, and 600 μM H2O2 was added to the medium. After incubation for 20 h, cell viability was determined using MTT reduction assay. Apocynin was used as positive control. Values are the mean ± SD (n = 6).

The Structural Activity Relationship of Xyloketals on a COMSIA Model
To explore the SAR of these xyloketals, a COMSIA model was constructed to explain the structural activity relationship of xyloketal B and its analogs. These compounds had the same structural framework, to unify the evaluation standard; therefore, a dominating stereoisomer of a previously reported xyloketal structure was selected for use in this SAR analysis (Chart 4) [16,18,30]. The statistical parameters of the 3D-QSAR models are shown in Table 3. For an acceptable standard of a 3D-QSAR model, the q 2 training (cross-validated regression coefficient) of the training set should be greater than 0.5, and the r 2 training (conventional regression coefficient) should be greater than 0.9. The LOO PLS analysis of the model gives a q 2 value of 0.577 at six components, together with the conventional regression coefficient r 2 of 0.988 and a standard error of estimate of 0.041. Furthermore, the significance and predictability of QSAR models should be further proven by an external test set using the following criteria: q 2 > 0.5, r 2 > 0.6 and 0.85 < k < 1.15 (k refers to the slope of the regression line between the experimental and the predicted biological activities). A graphical representation of the predicted and actual values is displayed in Figure 3. An excellent correlation between the experimental and predicted biological activities is shown in Figure 3 for the test set (q 2 = 0.648, r 2 = 0.858, and k = 0.987). In summary, all statistical data satisfied the recommended criteria, suggesting that the derived model exhibits good predictive ability in the external test-set validation. The predicted and actual values are shown for comparison in Table 4.   To determine how to modify the structure of xyloketal B, we built a model using a series of derivatives to explain the structure-activity relationship. Figure 4 shows a contour map of each field in the presence of xyloketal B. These maps indicated the favorable and unfavorable modification of the compounds in the colored regions. They are (a) an electrostatic map highlighting the regions where electropositive components were favorable (shown in blue) and unfavorable (shown in red) for the activity; (b) a hydrophobic map highlighting the regions where hydrophobic components were favorable (shown in yellow) and unfavorable (shown in white) for the activity; (c) a hydrogen donor map highlighting the regions where hydrogen donor components were favorable (shown in cyan) and unfavorable (shown in purple) for the activity; and (d) a hydrogen acceptor map highlighting the regions where hydrogen donor components were favorable (shown in magenta) and unfavorable (shown in green) for the activity.   The compounds in Group A yielded relatively lower activities. This result could be explained by the mismatch of the CoMSIA force field with the substitution groups. As shown in Figure 5a, the overlay of the compounds in group A had a common amide group, and the oxygen fell into the blue area, EP1. As mentioned previously, the blue contour map indicates areas where positively charged components would increase the activity, but negatively charged components would decrease the activity. Take compound 5 as a more specific example; it not only had the aforementioned common amide group in EP1, but also an -OH group in the purple area DN1. Purple areas indicate that a hydrogen bond donor in the area would have a negative effect on the compound activity (Figure 5b). The above analysis justified why compound 5 had the lowest activity. Some of the compounds in Group B had stronger activities. An overlay of the compounds in Group B in Figure 5c revealed that the substitution groups of compounds in group B took a different orientation to fit in the yellow region HF1. Compound 24 extended its nitro group into the magenta area AF1 and the red area EN1, where hydrogen bond acceptor and electro-negative groups would elevate the activity of the compound, respectively. Compound 24 matched the CoMSIA map well ( Figure 5d); therefore, it possessed the highest activity of all compounds. In contrast, compounds 25-28 had reduced activities due to replacing the nitro group with halogens, which were not hydrogen bond acceptors. Compounds in Group C produced similar activities to that of xyloketal B because no drastic structural changes were made to these compounds. The substitution groups did not specially fall into regions that increased or decreased the activities.

Xyloketal Derivatives Restored the H2O2-Induced Reduction of the Mitochondrial Membrane Potential (ΔΨm)
Mitochondria are considered the main source of reactive oxygen species (ROS) in cells. Therefore, we investigated whether xyloketals could protect mitochondria via inhibition of ROS. Compounds 23 and 24 were examined in the JC-1 mitochondrial membrane potential (MMP) assay of HUVECs using flow cytometry (FCM). As shown in Figures 6 and 7, 23 and 24 significantly inhibited the H2O2-induced the decrease in the cell mitochondrial membrane potential (ΔΨm) at 25 μM. Collectively, 23 and 24 effectively protected HUVECs against oxidative damage and further mitochondrial membrane integrity impairment and prevented H2O2-induced apoptosis of HUVECs by regulating the ROS-mediated mitochondrial dysfunction pathway.

Chemistry
All reagents and solvents were of commercial quality and used without further purification. 1 H and 13 C NMR data were recorded on a Bruker AVANCE 400 MB NMR spectrometer (Bruker, Fallanden, Switzerland) operating at 400 and 101 MHz for 1 H and 13 C respectively. All chemical shifts are in ppm (δ) with respect to tetramethylsilane (TMS) as internal standard, and coupling constants (J) are in Hz. Mass spectra were obtained on DSQ (low resolution mass spectrometer) (Thermo, San Jose, CA, USA) and MAT95XP (high resolution mass spectrometer) instruments (Thermo, San Jose, CA, USA).    The training set and test set were randomly divided out of a total of 35 molecules. A training set of 30 molecules was used to construct the QSAR model. In addition, a training set of 5 molecules was used to validate it. All of the molecules were aligned using the most active compound, 22, as the template. Each compound was mapped onto a 3D lattice with grid points 2.0 Å apart. The mapped region was created automatically by the program with an attenuation factor of 0.3. The electrostatic, hydrophobic, donor and acceptor columns were used to construct the model. The model was constructed using the partial-least-squares (PLS) analysis without any column filtering.
The robustness of the model was addressed based on the internal cross-validation using the leave-one-out (LOO) procedure and the external validation of the test set. All of the statistical parameters are listed in Table 3.

Mitochondrial Membrane Potentials Assay
The JC-1 probe was used to measure the mitochondrial depolarization in HUVECs. Briefly, HUVECs were cultured in six-well plates at a density of 2.5 × 10 5 /mL, incubated with 23 and 24 at different concentrations (25 μM and10 μM) for 30 min, then exposed to 400 μM H2O2 for 20 h. According to the instructions for the test kits, all cells were collected into 1.5-mL tubes, incubated with JC-1 for 20 min at 37 °C and rinsed twice with PBS. The mitochondrial membrane potentials (ΔΨm) were monitored by determining the relative amounts of dual emissions from JC-1 monomers or aggregates using a BD FACS Aria flow cytometry. The red/green fluorescence was calculated using the Flowjo (v7.6.5, Ashland, OR, USA).

Statistics
Results are presented as mean ± S.E.M. Comparisons between multiple groups were performed using the Excel by t-test. Differences were considered to be significant at p ≤ 0.05.

Conclusions
The water insolubility of the xyloketal compounds from marine fungus may be challenging for further clinical development. Therefore, a new series of derivatives with the introduction of amino groups at the C-12 and C-13 positions of xyloketal B were designed and synthesized to improve the solubility and biological activity. All 28 new derivatives and seven known compounds (14, 15, 31-35) were evaluated for their protection against H2O2-induced HUVEC injury. The results indicated that some compounds exhibited strong anti-oxidative activities, especially compounds 23 and 24, which displayed the best excellent protective activities out of all of the derivatives. Then, a CoMSIA was constructed using the SYBYL programming package (version 7.3.5) to explain the structural activity relationship of the xyloketal derivatives. A 3D QSAR model generated using the CoMSIA was analyzed and provided good advice to modify the molecules for better activity in the future. Compounds 23 and 24, which had the most remarkable anti-oxidative activities, were further examined in the JC-1 mitochondrial membrane potential (MMP) assay of HUVECs. The results showed that compound 23 and 24 would significantly inhibit H2O2-induced the decrease in the cell mitochondrial membrane potential (ΔΨm) at 25 μM. In conclusion, we designed and synthesized a new series of xyloketal derivatives to improve solubility and biological activity. Among them, compounds 23 and 24 effectively protected HUVECs against oxidative damage and further mitochondrial membrane integrity impairment. These derivatives will be new candidates for the treatment of CVD.
OxLDL oxidized low density lipoprotein PLS partial-least-squares ROS reactive oxygen species SAR structure-activity relationship THF tetrahydrofuran

Conflicts of Interest
The authors declare no conflict of interest.