The barks were collected in July 2022 in Chiautla de Tapia, Puebla, located at 18°15’30.30’’N, 98°35’0.9’’E at 1020 AMSL (Figure 1). Non-probabilistic sampling was carried out, and 10 juvenile trees with an average diameter of 44.4 ± 6 cm and different bark colours were selected. One branch of each tree was stripped of bark, separating the primary and secondary branches and the bark from the stem. Following the instructions of the folk method, the biological material was dried in the sun for 2 days and ground in a mechanical mill (Figure 2 and Figure 3). Jorge Alberto Gutiérrez Gallegos, PhD, correctly identified the biological material, and some type specimens (voucher number 19745) were stored in the FEZA-UNAM Herbarium.

Ten grams of ground bark were meticulously extracted with 50 mL of ethyl acetate (J.T. Baker) in Soxhlet equipment at 50 °C, and three cycles of approximately 2 h each were carried out. The solvent was then carefully removed under reduced pressure in a rotary evaporator (BUCHI-R114).

The extracts (10 mg/mL) were dissolved in methanol (J.T. Baker), ultrasonicated for 3 min, and mixed with a vortex. The sample was separated on silica gel HPTLC plates (20 × 10 cm, _F254) (Merck). The analysis was performed in a CAMAG-HPTLC system (Muttenz, Switzerland) equipped with a sample applicator (ATS4), an automatic chamber developer (ADC2), a visualiser (version 2), an automatic immersion derivatiser (version 1.0 AT) and a hot plate (version 3).

All extracts were applied as 6 mm bands with 10 mm between each band. A quality control sample (QC), consisting of a mixture of all samples, was applied at the right end of each chromatographic plate.

The samples were separated with an elution system of hexane: EtOAc (7:3, v/v). The chamber’s saturation time was 20 min, and the humidity was 47% using a saturated KSCN solution. After development, the plates were automatically dried and derivatised with anisaldehyde-sulfuric acid at 100 °C for 3 min. Images of the plates were acquired under white light or UV at 366 nm.

For visual representation and multivariate data analysis, the plate images were processed with rTLC (Fichou, Ristivojevic & Morlock 2016). The data extraction dimensions were the same as those used for the sample application. For data extraction, 28 pixels were used as the integration dimension, and data from the grey channel were normalised to the QC sample of each corresponding plate and used for further analysis.

Principal component analysis (PCA) was performed using information between the 0.14 and 0.37 R_f range, corresponding to the anacardic acid, masticadienoic acid and 3α-hydroxymasticadienoic acid bands. The model was scaled by unit variance (UV), and the score plot was coloured according to branch type.

Ten grams of bark were boiled in 200 mL of water for 10 min and allowed to cool for 30 min at room temperature. The aqueous extract was filtered with paper (Whatman No. 1), frozen and lyophilised. One gram of lyophilised material was extracted by maceration with 45 mL of methanol under stirring overnight. The methanol extract was filtered with paper (Whatman No. 1), and the solvent was removed under reduced pressure with a rotary evaporator (BUCHI-R114).

The sample analysis was performed as reported by Gómez-Salgado et al. (2024). They were dissolved in DMSO (2 mg/mL) and analysed using an Agilent G6530AB HPLC-MS-SQ-TOF system. Separation was performed using a Gemini-NX C18 reversed-phase column (3 µm, 2.0 × 75 mm, Phenomenex) in a binary elution system of CH₃CN-H₂O and 0.1% formic acid (Gómez-Salgado et al. 2024). A gradual elution gradient (constant flow rate at 0.4 mL/min) was used, starting with a 15:85 ratio until reaching 100:0 in 8 min, followed by an isocratic retention of 1.5 min (Gómez-Salgado et al. 2024). The mass spectrometric analysis was performed in positive ionisation mode (range 100 m/z–2500 m/z), and the two most abundant ions per cycle were selected for fragmentation using the automatic mode (Auto MS²) (Gómez-Salgado et al. 2024).

The development of the molecular network was based on a classical analysis using the Global Natural Products Social Molecular Networking (GNPS) website (http://gnps.ucsd.edu) (Wang et al. 2016). The obtained raw (.d) files were converted to mzML format using the MSConvert programme and analysed on the GNPS server (Chambers et al. 2012). Some parameters were followed for the molecular network analysis, such as filtering MS² fragment ions within ±0.5 Da of the precursor m/z ratio. Also, the precursor and fragment ion tolerances were set to 0.5 Da (Gómez-Salgado et al. 2024). Network node connections were formed by assigning edge values based on a cosine score greater than 0.7 and at least six aligned peaks (Gómez-Salgado et al. 2024). Subsequently, the molecular network spectra were compared with the GNPS spectral libraries. Finally, a MolNetEnhancer analysis was performed to determine the structural diversity of the molecules present in the samples (Ernst et al. 2019). Cytoscape 3.9.1 was used to visualise the molecular networks graphically (Shannon et al. 2003), and clusters were assigned colours based on their chemical subclass (Gómez-Salgado et al. 2024).

Thirty female mice of the CD1 strain, 10–12 weeks old, were used, whose backs were depilated with an electric shaver and depilatory cream (Loquay^®). To induce the third-degree burn, they were previously anesthetised with 24 mg/kg of sodium pentobarbital (pisabental, PiSA). The shaved area was exposed in a 1 × 2 cm opening of a metal plate and immersed in 95 °C water for 12 s. Subsequently, the burn was inoculated subcutaneously with 10⁷ CFU/60 µL of P. aeruginosa PA14.

Mice were kept in a heat source until they fully recovered from the procedure. Sterilised sawdust bedding was also used, and the acrylic cages, drinkers and metal grates were washed with 10% sodium hypochlorite.

The bark of the primary and secondary branches was applied following the instructions of the popular method, which consists of washing the wound with 5% aqueous extract and applying the powdered bark until the lesion is covered (Gómez-Salgado et al. 2024). This procedure was performed every 24 h for 12 days.

Mortality was determined every 24 h, and on the 12th day, surviving animals were anesthetised with 24 mg/kg sodium pentobarbital (pisabental, PiSA) and sacrificed by cervical dislocation. Burn tissue and liver were processed to quantify the presence of bacteria using the viable count method, with results expressed as CFU/g of tissue.

The ethyl acetate extracts were diluted in DMSO (Sigma-Aldrich) to obtain final concentrations of 250, 500 and 1000 µg/mL in the in vitro assays.

An overnight culture of P. aeruginosa PA14 was performed in LB broth at 37 °C, 120 rpm, and adjusted to O.D. 660 nm = 0.1. In a 96-well plate (Corning^®), 180 µL of culture was mixed with 20 µL of the ethyl acetate extract, incubated at 37 °C and 160 rpm for 24 h.

Pyocyanin production: A total of 800 µL of culture supernatants were mixed with 420 µL of chloroform (J.T. Baker^®) and shaken vigorously for 2 min. The samples were centrifuged at 14 000 rpm for 8 min, and the organic phase was recovered and mixed with 800 µL of 0.2 N HCl. For pyocyanin quantification, the aqueous phase (650 µL) was diluted 1:1 in distilled water and measured at 520 nm. The values obtained were normalised based on the growth of the cultures measured at 660 nm.

The caseinolytic activity was carried out according to that reported by Gómez-Salgado et al. (2024): A total of 5 µL of centrifuged culture supernatant (14 000 rpm for 2 min) was mixed with 100 µL of 1.2% azocasein (20 mM Tris-base, 1 mM CaCl₂, pH 8) and incubated for 35 min at 37 °C (Gómez-Salgado et al. 2024). Subsequently, 50 µL of the reaction mixture was transferred to glass tubes with 200 µL of 1% HNO₃, shaken, and centrifuged at 14 000 rpm for 2 min. Finally, 50 µL of the supernatants was transferred to a 96-well plate (Corning^®) with 150 µL of 0.5% NaOH and measured at 440 nm.

Biofilm production: In a 96-well plate (Corning^®), bacteria were incubated at 28 °C for 48 h in the dark. After, the cultures were discarded, and the wells were washed with distilled water three times. They were placed in an oven at 40 °C for 20 min to eliminate humidity.

Subsequently, 200 µL of 0.1% crystal violet was added to each well, allowed to stand for 5 min, and washed three times with double-distilled water. Finally, the dye adhered to the walls of the wells was solubilised with 200 µL of 80% ethanol, and the absorbance was measured at 570 nm. The values obtained were normalised based on the growth of the cultures measured at 660 nm.

Experimental data were analysed with SigmaPlot version 14.0. The figure legends indicate the different parametric and non-parametric statistics used.

The burn mouse model was carried out following the procedure described by Gómez-Salgado et al. (2024). Ethical clearance to conduct this study was obtained from the Institution of Teaching and Research in Agricultural Sciences, Postgraduate College, Animal Welfare Committee (Project research number: COBIAN/016/23 and ethical clearance No. 016). The animals were fed a standard diet (Rodent Diet 5001, LabDiet) and housed in an experimental room with 12 h light and dark cycles.

Seven of the randomly selected trees were male and three were female. Each presented a unique colour gradient in the SB, ranging from red to white (Figure 2 and Figure 3a). Instead, the BBs were white, regardless of the colour of the SB. These colour differences become imperceptible after drying, and all barks become reddish brown.

Additionally, with its thicker vascular cambium, the SB accumulates more water, requiring a longer drying time. In contrast, BB dries faster, resulting in thin strips that are easier to grind (Figure 2 and Figure 3b).

The organic extract with ethyl acetate from the SB and BB exhibited similar physical characteristics: a viscous texture, a dark green colour, and similar yield (Figure 3c). Meanwhile, the 5% aqueous extracts were reddish and generated foam when shaken vigorously (Figure 3d). Likewise, the physical appearance and yield of the lyophilised extract was 10%, a similar value in all samples (Figure 3e).

The ethyl acetate extract was analysed by HPTLC to identify the main bioactive metabolites described for cuachalalate. Visual inspection of the chromatographic plates indicated slight quantitative differences between biological replicates and branch type (Figure 4a and Figure 4b). However, the area showing the three compounds did not show relevant differences between samples.

The chromatographic data were subjected to PCA for more information on possible differences. The model resulted in four principal components (PCs) that explained 87% of the variability of the model (RX2_cum = 0.87). PC1 captured 50% of the model variability, while PC2 captured only 19%. However, no differentiation of groups was observed in the score, indicating that there are no qualitative or quantitative differences in the content of anacardic acid mixture (R_f 0.14), masticadienoic acid (R_f 0.23) and 3α-hydroxymasticadienoic acid (R_f 0.37) between the samples (Figure 4c).

The molecular networks were built considering 740 MS spectra condensed in 110 nodes clustered in 62 groups forming 16 molecular families (groups of fragmentation spectra with a high degree of similarity, represented as a cluster of connected nodes in the network) and 46 isolated nodes (Figure 5a). Analysis of the MNs using the MolNetEnhancer displayed the presence of terpenoids, flavonoids and catechins in the samples. Furthermore, dereplication analysis allowed the putative identification of some molecules. Among the terpenoids, cholic acid, saponins, β-sitosterol and panaxatriol (Figure 5b) were detected, whereas the polyphenolic compounds epicatechin, epigallocatechin gallate, catechin gallate, kaempferol-3-O-rhamnoside, kaempferol-3-O-glucoside and acacetin-7-O-rutinoside were putatively identified (Figure 5b).

In a previous study using a folk method and a burn model in mice, the antipathogenic capacity of the cuachalalate SB was demonstrated (Gómez-Salgado et al. 2024). The folk procedure consisted of washing the infected burn with the aqueous extract of the SB and applying a poultice with the ground bark (Gómez-Salgado et al. 2024).

By applying the folk method with the bark of the branches, we also recorded the antipathogenic ability. In the burns model, the untreated animals succumbed to infection within the first 24 h, and this trend continued until 12 days when a final survival rate of 40% was recorded. Compared to the group treated with the folk method, the animals’ deaths were avoided, and a survival of 90% was recorded (Figure 6a).

Although the folk method with the bark branches did not reduce the establishment of the bacteria in the burned tissue of the surviving animals (Figure 6b), the treatment prevented systemic dispersion because no bacteria were recorded in the livers (Figure 6c).

Ethyl acetate extract of BB markedly reduced the production of virulence factors regulated by quorum sensing in P. aeruginosa.

The SB extract showed remarkable efficacy, inhibiting biofilm formation by 78% with 250 µg/mL and reaching 90% inhibition with 1000 µg/µL. Although the BB extract did not show a dose-response effect, all concentrations tested inhibited the biofilm by approximately 70%, comparable to the QS mutant strain (ΔlasR/ΔrhIR) used as a positive control (Figure 7a).

For pyocyanin production, the SB extract reduced it by about 75%, while the BB extracts completely inhibited it at 1000 µg/mL (Figure 7b). In particular, the PBB extract registered the most significant activity because it completely inhibited its production with 500 µg/mL (Figure 7b).

Finally, the SB extract inhibited caseinolytic activity by 20% and 40%, with 500 µg/mL and 1000 µg/mL, respectively, while the BB extract did so at around 80% with 500 µg/mL (Figure 7c).