Antileishmanial Compounds Isolated from Psidium Guajava L. Using a Metabolomic Approach

With an estimated annual incidence of one million cases, leishmaniasis is one of the top five vector-borne diseases. Currently available medical treatments involve side effects, including toxicity, non-specific targeting, and resistance development. Thus, new antileishmanial chemical entities are of the utmost interest to fight against this disease. The aim of this study was to obtain potential antileishmanial natural products from Psidium guajava leaves using a metabolomic workflow. Several crude extracts from P. guajava leaves harvested from different locations in the Lao People’s Democratic Republic (Lao PDR) were profiled by liquid chromatography coupled to high-resolution mass spectrometry, and subsequently evaluated for their antileishmanial activities. The putative active compounds were highlighted by multivariate correlation analysis between the antileishmanial response and chromatographic profiles of P. guajava mixtures. The results showed that the pooled apolar fractions from P. guajava were the most active (IC50 = 1.96 ± 0.47 µg/mL). Multivariate data analysis of the apolar fractions highlighted a family of triterpenoid compounds, including jacoumaric acid (IC50 = 1.318 ± 0.59 µg/mL) and corosolic acid (IC50 = 1.01 ± 0.06 µg/mL). Our approach allowed the identification of antileishmanial compounds from the crude extracts in only a small number of steps and can be easily adapted for use in the discovery workflows of several other natural products.


Introduction
Leishmaniasis is in the top five of vector-bone diseases identified in the Global Health Estimates 2016 summary table of the World Health Organization (WHO). It is one of 20 neglected tropical diseases that infect one billion people in low socioeconomic populations in 149 countries [1]. Leishmaniasis is caused by the genus Leishmania spp., which includes 29 species. Among them, Leishmania donovani and Leishmania infantum cause visceral leishmaniasis, the lethal form of the disease [2]. Leishmaniasis is transmitted to mammalians by the bites of the female phlebotomine sandfly. The epidemiology of leishmaniasis depends on a variety of factors including the between host, reservoir and vector (human, animal and sandfly), the local ecological characteristics of the transmission sites such as alterations in temperature and water storage, irrigation habits, deforestation, climate changes, exposure

UHPLC-HRMS-based Metabolomics Approach
UHPLC-HRMS (ultra-high-performance liquid chromatography-high resolution mass spectrometry) profiles of nine extracts (three crude extracts from the Champasak province, three crude extracts from the Savannakhet province, and three crude extracts from the Vientiane province ( Figure 2A) provided 448 features (m/z-RT pairs) in negative ionization (NI) mode and 163 features (m/z-RT pairs) in positive ionization (PI) mode. To obtain an unsupervised overview of sample closeness, a principal component analysis (PCA) score plot was performed to project crude extracts and quality control (QC) samples (an aliquot of each sample used in the study) on the latent variable space. This PCA score plot revealed that the analytical workflow was accurate and reproducible due to the central clustering of QC samples. In addition, samples did not cluster according to collection areas; only two samples from the Champasak province did not cluster with the other samples ( Figure 2B(a)).
We then built an orthogonal projection to latent structures (OPLS) regression model to rank the features (m/z-RT pairs) toward the IC 50 values of antileishmanial activity (Y-input) for polar and apolar fractions. The quality of model prediction was adequate (R2Y = 0.985, Q2Y = 0.97) and a permutation test assessed its validity (Supplementary Figure S1). The model revealed that the liquid-liquid extraction method clearly separated polar and apolar fractions in terms of chemical constituents and antileishmanial activity, and the apolar phases showed the most promising antileishmanial potential ( Figure 2B(c)). This supervised technique classified potentially antileishmanial compounds according to their coefficient values ( Figure 2C(a,b)), with positive coefficients relating to a high correlation with antileishmanial potential and negative coefficients to non-active compounds. The first seven hits ( Figure 2C

136
The first seven ranked compounds from the OPLS regression analysis were putatively annotated 137 by matching compounds in the Dictionary of National Products (DNP) database (CRC press, v27.2).

138
Compounds belonging to the Myrtaceae family and the Psidium genus were prioritized to narrow

143
were supported UV spectra of each peak (Table 1; column "UV"). The UV spectra of three

Identification of Putative Antileishmanial Compounds Based on Liquid-liquid Extraction
The first seven ranked compounds from the OPLS regression analysis were putatively annotated by matching compounds in the Dictionary of National Products (DNP) database (CRC press, v27.2). Compounds belonging to the Myrtaceae family and the Psidium genus were prioritized to narrow down the possibilities. For each compound, some candidates were ranked and proposed based on their similarity score according to the comparison between experimental MS/MS fragmentation and the in silico spectra of candidates. This process resulted in the annotation of seven candidates belonging to the Psidium genus, and all were classified as triterpenoid compounds. Annotation results were supported UV spectra of each peak (Table 1; column "UV"). The UV spectra of three triterpenoids (compounds 2, 4, 5) showed an absorption band between 278 and 310 nm highlighting the presence of a conjugated radical ( Figure 3). If available, commercial authentic standards were used to confirm feature annotation (compounds 2, 5, 6) or purified from crude apolar fraction and confirmed by 1D and 2D nuclear magnetic resonance (NMR) (compound 1, Supplementary Table S1).   Table 1 for details).

Putative Mechanism of Action
Antileishmanial assays revealed that pooled apolar fractions had a good activity with an IC50 value of 1.96 ± 0.47 µg/mL and selectivity index (SI) of around 26. The first two ranked compounds, corosolic acid (1) and jacoumaric acid (2) were more active than guajadial B (5) and medicagenic acid (6), with an IC50 of 1.32 ± 0.59 µg/mL for jacoumaric acid and 1.01 ± 0.06 µg/mL for corosolic acid ( Table 2). Since most antileishmanial drugs act by the production of reactive oxygen species (ROS) that lead to the death of parasites [23,24], we measured the production of ROS induced by jacoumaric acid and corosolic acid at the IC50 (1.32 µg/mL and 1.06 µg/mL, respectively) and IC90 (1.90 µg/mL and 1.43 µg/mL, respectively) using H2DCFDA (2′,7′-dichlorodihydrofluorescein diacetate). The corresponding graph presented in Figure 4A clearly demonstrates that jacoumaric acid and corosolic  Table 1 for details).

Putative Mechanism of Action
Antileishmanial assays revealed that pooled apolar fractions had a good activity with an IC 50 value of 1.96 ± 0.47 µg/mL and selectivity index (SI) of around 26. The first two ranked compounds, corosolic acid (1) and jacoumaric acid (2) were more active than guajadial B (5) and medicagenic acid (6), with an IC 50 of 1.32 ± 0.59 µg/mL for jacoumaric acid and 1.01 ± 0.06 µg/mL for corosolic acid ( Table 2). Since most antileishmanial drugs act by the production of reactive oxygen species (ROS) that lead to the death of parasites [23,24], we measured the production of ROS induced by jacoumaric acid and corosolic acid at the IC 50 (1.32 µg/mL and 1.06 µg/mL, respectively) and IC 90 (1.90 µg/mL and 1.43 µg/mL, respectively) using H 2 DCFDA (2 ,7 -dichlorodihydrofluorescein diacetate). The corresponding graph presented in Figure 4A clearly demonstrates that jacoumaric acid and corosolic acid have an antioxidant capacity, since the fluorescence proportional to the production of ROS significantly decreased compared to the control by addition of either compound regardless of the concentration used. We also measured the IC 50 of these compounds when associated with mannitol, a strong antioxidant ( Figure 4B). We found no significant differences in IC 50 values, confirming that the activity of jacoumaric acid and corosolic acid does not involve the production of ROS. acid have an antioxidant capacity, since the fluorescence proportional to the production of ROS significantly decreased compared to the control by addition of either compound regardless of the concentration used. We also measured the IC50 of these compounds when associated with mannitol, a strong antioxidant ( Figure 4B). We found no significant differences in IC50 values, confirming that the activity of jacoumaric acid and corosolic acid does not involve the production of ROS.

Discussion
The aim of this study was to identify antileishmanial compounds from the leaves of P. guajava and their putative mechanism of action. According to the inhibition percentage at 25 µg/mL, the ethanolic leaf extracts of P. guajava killed more than 50% of parasites, which is more than the reported antileishmanial activity of seven Brazilian plant species [25] and plants from the northwest of Morocco [26]. Moreover, the apolar fractions have interesting antileishmanial activity, with IC50 values ranging from 1.62 ± 0.46 µg/mL to 2.89 ± 0.49 µg/mL and an SI measured on L. infantum amastigotes of about 26. This result indicates that this plant is more active than Mangifera indica and Digera muricate, plants that have previously been reported to have antileishmanial activity [27]. To identify compounds responsible for this activity, we compared the in vitro antileishmanial activity and chromatographic profiles of several P. guajava leaf extracts using a metabolomic workflow. The approach depicted here allowed the ranking of detected features based on their correlation to a putative antileishmanial activity in a small number of steps. The annotation results were confirmed by commercial authentic standards (Table 1, compounds 2, 5, and 6) or a purification process (1). Guajadial B exhibited antitumor activity in five human cancer cell lines (HCT116, CCRFCEM, DU145, Huh7, and A549) at an IC50 value of 150 nM [28]. Medicagenic acid has been identified as an antifungal agent against Pyricularia pryzae (minimum inhibitory concentration (MIC) = 0.01 mg/mL and minimum fungicidal concentration (MFC) = 0.03 mg/mL) [29]. In our study, corosolic acid and jacoumaric acid displayed the most interesting antileishmanial activity in the amastigotes stage with respective IC50 values of 1.01 ± 0.06 µg/mL and 1.32 ± 0.59 µg/mL and an SI above five. This makes these two compounds more active than other triterpenoids such as ursolic acid (IC50 value 27 µg/mL) [30], E-caryophyllene (IC50 value 10.7 ± 0.6 µg/mL) [31], and spergulin-A (IC50 value 6.22 µg/mL) [32]. Both compounds have previously been identified in P. guajava and suggested to be good candidates for the development of new treatments for sickle cell anemia [33]. Furthermore, we have

Discussion
The aim of this study was to identify antileishmanial compounds from the leaves of P. guajava and their putative mechanism of action. According to the inhibition percentage at 25 µg/mL, the ethanolic leaf extracts of P. guajava killed more than 50% of parasites, which is more than the reported antileishmanial activity of seven Brazilian plant species [25] and plants from the northwest of Morocco [26]. Moreover, the apolar fractions have interesting antileishmanial activity, with IC 50 values ranging from 1.62 ± 0.46 µg/mL to 2.89 ± 0.49 µg/mL and an SI measured on L. infantum amastigotes of about 26. This result indicates that this plant is more active than Mangifera indica and Digera muricate, plants that have previously been reported to have antileishmanial activity [27]. To identify compounds responsible for this activity, we compared the in vitro antileishmanial activity and chromatographic profiles of several P. guajava leaf extracts using a metabolomic workflow. The approach depicted here allowed the ranking of detected features based on their correlation to a putative antileishmanial activity in a small number of steps. The annotation results were confirmed by commercial authentic standards (Table 1, compounds 2, 5, and 6) or a purification process (1). Guajadial B exhibited antitumor activity in five human cancer cell lines (HCT116, CCRFCEM, DU145, Huh7, and A549) at an IC 50 value of 150 nM [28]. Medicagenic acid has been identified as an antifungal agent against Pyricularia pryzae (minimum inhibitory concentration (MIC) = 0.01 mg/mL and minimum fungicidal concentration (MFC) = 0.03 mg/mL) [29]. In our study, corosolic acid and jacoumaric acid displayed the most interesting antileishmanial activity in the amastigotes stage with respective IC 50 values of 1.01 ± 0.06 µg/mL and 1.32 ± 0.59 µg/mL and an SI above five. This makes these two compounds more active than other triterpenoids such as ursolic acid (IC 50 value 27 µg/mL) [30], E-caryophyllene (IC 50 value 10.7 ± 0.6 µg/mL) [31], and spergulin-A (IC 50 value 6.22 µg/mL) [32]. Both compounds have previously been identified in P. guajava and suggested to be good candidates for the development of new treatments for sickle cell anemia [33]. Furthermore, we have demonstrated their role in ROS scavenging activity, with antioxidant capacity comparable to rosmarinic acid and caffeic acid [34]. Corosolic acid has been shown to prevent inflammation and hypertension in rats [35] and to reduce the plasma glucose level in humans [19]. Corosolic acid is also known for its antibacterial activity against Paenibacillus larvae or Melissococcus plutonius [36], its anticancer activity [37], and its antitumor effect on human cervical adenocarcinoma Hela cells [38]. Jacoumaric acid also displayed antitumor effects in many carcinoma cell lines, such as leukemia [39], colorectal cancer, and human breast cancer cell lines [40]. To the best of our knowledge, this is the first report demonstrating antileishmanial activity of both compounds at the amastigotes stage. However, the close derivatives maslinic and oleanoic acid displayed antileishmanial activity on the promastigote and amastigote forms of L. infantum and L. amazonensis in the same concentration range as recorded in this study (IC 50 on L. infantum amastigote: 0.99 ± 0.09 µg/mL and 2.91 ± 0.066 µg/mL, respectively) [41]. This could highlight a common mechanism of action of these triterpenoid compounds, probably related to ROS scavenging and radical positioning. According to our biological assay results and those obtained by the teams of Torres-Santos et al. and Sifaoui et al. [30,41], the presence of a carboxy group in position C-4 decreases the antileishmanial activity. Overall, the metabolomic workflow depicted in this study allowed us to rapidly target compounds responsible for the antileishmanial activity and limit multiple fractionation steps.

Plant Material
On November 2016, nine samples of P. guajava L. (Pg) leaves were collected from middle and southern parts of Lao PDR (Champasak: Pg1 to Pg3, Savannakhet: Pg4 to Pg6 and Vientiane province: Pg7 to Pg9). Samples were washed and air-dried, protected from the sunlight, before being ground into powder to obtain 1 kg of each. A sample specimen of each sample was collected and deposited at the herbarium of the Institute of Traditional Medicine of Lao, Vientiane, Lao PDR.

Leaf Extraction
Each accession of P. guajava leaves (250 g) sample were extracted two times using a ratio of 1 g of dried leaves for 10 mL of of 80% ethanol under agitation at room temperature for 24 h. The filtrate solutions were evaporated under reduced pressure (Buchi rotavapor R-114, Paris, France) and yield61 ± 2 g for each extract. Crude extracts (1 g) were submitted to a liquid-liquid extraction using 200 mL of a biphasic mixture composed of M1 and M2 (M1: methyl tert-butyl ether: water, 75:25; M2, water: methanol, 75:25), to obtain nine polar and nine apolar fractions. The 18 fractions were dried under reduced pressure and aliquoted for UHPLC-HRMS profiling at 2 mg/mL in a MeOH/Water 80/20 of LC-MS grade (Fisher Scientific, Schwerte, Germany) and for antileishmanial bioassays at 10 mg/mL in DMSO (Sigma-Aldrich, 99.5%, St. Louis, MO, USA).
The structure of compound 1 was determined by 1D and 2D NMR spectroscopy on a Bruker 500 MHz (Avance 500, Billerica, MA, USA) using dimethylsulfoxide-d6. Chemical shifts (relative to tetramethylsilane (TMS)) are in ppm, and coupling constants in Hz. The compound was identified as "corosolic acid" by comparing with data from the literature [40] (Supplementary Table S1).

UHPLC-HRMS Profiling
All extracts (2 mg/mL) were profiled using a UHPLC-DAD-LTQ Orbitrap XL instrument (Ultimate 3000, Thermo Fisher Scientific, Hemel Hempstead, UK). The UV detection from 210 to 400 nm was performed with a diode array detector (DAD) (Hemel Hempstead, UK). Mass detection was performed using an atmospheric pressure chemical ionization (APCI) source in both NI and PI modes at 15,000 resolving power (full width at half maximum at 400 m/z). The mass scanning range was m/z 100-1500 Da. The capillary temperature was 300 • C and the spray voltage was fixed at 3.0 kV. Mass measurements were externally calibrated before starting the experiment. Each full MS scan was followed by data-dependent MS/MS on the four most intense peaks using collision-induced dissociation (35% normalized collision energy, isolation width 2 Da, activation Q 0.250). The LC-MS system was run in binary gradient mode using a BEH C18 Acquity column (100 × 2.1 mm i.d., 1.7 µm, 130 Å, Waters, MA, USA) equipped with a guard column. Mobile phase A (MPA) was 0.1% formic acid (FA) in water and mobile phase B (MPB) was 0.1% FA in acetonitrile. Gradient conditions were: 0 min, 95% MPA; 0.5 min 95% MPA; 12 min, 5% MPA; 15 min, 5% MPA, 15.5 min, 95% MPA, and 19 min, 95% MPA. The flow rate was 0.3 mL/min, column temperature 40 • C, and injection volume was 2 µL.

Data Processing
The UHPLC-HRMS raw data were converted to abf files (Reifycs, Japan) and processed with MS-DIAL version 3.90 [42] for mass signal extraction between 100 and 1500 Da from 0.5 to 16.5 min. Respective tolerances for MS1 and MS2 were set to 0.01 and 0.2 Da in centroid mode. The optimized detection threshold was set to 10 5 (negative) and 2 × 10 5 (positive) for MS1. The peaks were aligned on a quality control sample (an aliquot of each fraction) reference file with a retention time tolerance of 0.15 min and a mass tolerance of 0.025 Da. Adducts and complexes were identified to exclude them from the final peak list along with features from blanks sample (injection of dilution solvent). Additionally, features with a relative standard deviation above 30% in QC sample were also deleted. The resulting peak list was then exported to comma-separated value (CSV) format prior to multivariate data analysis using SIMCA-P+ (version 15.0.2, Umerics, Umea, Sweden).

Statistical Analysis
CSV files were directly imported into SIMCA-P+ (version 15.0.2, Umerics, Umea, Sweden). For multivariate data analysis, all data were pareto scaled. The OPLS regression analysis was carried out with IC 50 of antileishmanial activity as the Y input. Coefficient scores were used to rank variables according to their antileishmanial potential.

Identification of Significant Features
Molecular formulae of significant features were calculated with MS-FINDER 3.24 [18]. Various parameters were used in order to reduce the number of potential candidates, such as exclusive selection of the elements C, H, and O; mass tolerance fixed to MS1:0.01 Da and MS2:0.2 Da; and the isotopic ratio tolerance set to 20%. Only natural product databases focused on plants were selected from the DNP (CRC press, v27.2). Compounds from the Psidium genus or Myrtaceae family were prioritized. The results were presented as a list of compounds sorted according to the score value of the match. This value encompassed uncertainty on accurate mass, the isotopic pattern score, and the experimental MS/MS fragmentation mirrored to in silico matches. Only chemical identities with a final score above five were retained.

Antileishmanial Activity on Promastigotes
We use luciferase assays to evaluate the effect of the tested compounds on the growth of L. infantum promastigotes. Briefly, logarithmic phase promastigotes suspended in RPMI 1640 medium supplemented with 10% fetal calf serum, 2 mM l-glutamine, and antibiotics (100 U/mL penicillin, 100 mg/mL streptomycin, and 50 µg/mL geneticin) were incubated at a density of 10 6 parasites/mL in sterile 96-well plates with crude extract and fractions (diluted at 50 and 25 µg/mL) in duplicate. Amphotericin B (purchased from Sigma-Aldrich) was used as a positive control. After a 72-h incubation period at 24 • C, we examined each well plate under a microscope to detect possible precipitate formation. To estimate the luciferase activity of promastigotes, 80 µL of each well was transferred into white 96-well plates after mild resuspension. Steady Glow ® reagent (Promega, Madison, WI, USA) was added according to the manufacturer's instructions, and plates were incubated for 2 min at room temperature. The luminescence was measured by a MicroBeta luminescence counter (PerkinElmer, Waltham, MA, USA). For the IC 50 evaluation, the most active fractions or compounds were measured by eight dosage dilutions (0.39-50 µg/mL). IC 50 was calculated by non-linear regression analysis processed on dose-response curves, using the GraphPad Prism 6.0. (San Diego, CA, USA). IC 50 values represent the mean value calculated from three independent experiments.

Antileishmanial Activity on Axenic Amastigotes
Logarithmic phase L. infantum promastigotes were centrifuged at 900 g for 10 min. The supernatant was removed carefully and was replaced by the same volume of RPMI 1640 complete medium at pH 5.4 and incubated for 24 h at 24 • C. The acidified promastigotes were incubated for 24 h at 37 • C in a ventilated flask and were transformed into amastigotes. The effect of the tested compounds on the growth of L. infantum axenic amastigotes was assessed as follows: L. infantum amastigotes were incubated at a density of 2 × 10 6 parasites/mL in sterile 96-well plates with crude extract and fractions (diluted at 50 and 25 µg/mL) in duplicate. Amphotericin B was used as a positive control. After a 48-h incubation period at 37 • C, each well plate was then examined under a microscope to detect any precipitate formation. Subsequently, to determine the IC 50 , we used the same method described in Section 4.8.1.

Cytotoxicity Evaluation
An MTT (3-(4,5-Dimethylthiazol-2-yl)-2,5-Diphenyltetrazolium Bromide) assay with the J774A.1 cell line (mouse macrophage cell line, Sigma-Aldrich) was performed to evaluate the cytotoxicity of the tested compounds. Briefly, cells (5 × 10 4 cells/mL) in 100 µL of complete medium (DMEM high glucose supplemented with 10% fetal calf serum (FSC), 2 mM l-glutamine and antibiotics (100 U/mL penicillin and 100 µg/mL streptomycin)) were seeded into each well of 96-well plates and incubated at 37 • C and 5% CO 2 . After incubation for 24 h, 100 µL of medium with various concentrations of jacoumaric acid, carasolic acid, and appropriate controls (doxorubicin and amphotericin B, purchased from Sigma-Aldrich) were added and the plates were incubated for 72 h at 37 • C and 5% CO 2 . Each 96-well plate was then examined under a microscope to detect possible precipitate formation before the medium was aspirated from the well. 100 µL of MTT solution (0.5mg/mL in DMEM) was then added to each well. Cells were incubated for 2 h at 37 • C and 5% CO 2 . Then, the MTT solution was removed and DMSO (100 µL/well) was added to dissolve the resulting formazan crystals. Plates were shaken vigorously (300 rpm) for 5 min. The absorbance was measured at 570 nm with a microplate spectrophotometer (Eon Bio Tek, Winooski, VT, USA). DMSO was used as blank. CC 50 values were calculated by non-linear regression analysis processed on dose-response curves, using GraphPad Prism 6.0. (San Diego, CA, USA). CC 50 values represent the mean value calculated from three independent experiments.

Determination of Intracellular ROS Generation
ROS levels in treated and untreated L. infantum amastigotes were monitored using the cell permeable fluorogenic dye H 2 DCFDA [43]. 1 × 10 6 cells/mL amastigotes were treated with IC 50 and IC 90 doses of jacoumaric acid and corosolic acid for 24 h. The cells were then centrifuged, washed with PBS, resuspended in PBS, and incubated for 30 min in the dark with 10 µM H 2 DCFDA at 37 • C. ROS were measured as an increase in fluorescence caused by the conversion of non-fluorescent dye to highly fluorescence H 2 DCFDA (excitation wavelength 490 nm, emission wavelength 525 nm) in a fluorescence microplate reader (SAFAS Xenius XM, Monaco, France).

Conclusions
We set up a metabolomic workflow to rapidly decipher antileishmanial compounds from P. guajava L. leaves extract. Our dereplication approach lead to the identification of corosolic acid and jacoumaric acid as the most active compounds against amastigotes L. infantum, along with the ROS scavenging in the antileishmanial mechanism. This is the first report in antileishmanial activity of corosolic acid and jacoumaric acid. Further studies would be required to understand the mechanism of action of these triterpenoids' compounds on leishmaniasis parasites.

Supplementary Materials:
The following are available online, Figure S1: Permutation plot of the OPLS correlation model between antileishmanial activity and the liquid chromatography-mass spectrometry dataset; Figure S2: MS/MS fragmentation pattern of standard compound 1, 2, 5, 6 versus identified peaks in crude extract chromatograms.; Table S1: 1H-NMR and 13C-NMR data (DMSO-d6) of compound 1. LC-MS raw files and treated data spreadsheet has been archived to zenodo with the linked DOI: 10.5281/zenodo.3555529.