Exploring the antimicrobial activity of fermented and non‐fermented cocoa bean shell extracts through metabolomics analysis and synergistic studies

Abstract

BACKGROUND

Cocoa bean shell (CBS) extracts have emerged as a promising source of antimicrobial compounds. However, the bioactive compounds responsible for their antimicrobial activity have not been studied sufficiently. This study analyzed the antimicrobial properties of 12 extracts from fermented and non‐fermented CBS, employing solvent and steam distillation extraction techniques to maximize bioactive diversity. The extracts were assessed against selected bacteria and fungi, including the ‘ESKAPE’ pathogens.

RESULTS

The CBS solvent and steam distillation extracts (except for ethyl acetate‐fermented CBS) were found to exhibit antimicrobial activity against Streptococcus mutans with minimum inhibitory concentrations (MICs) as low as 0.0625 mg mL⁻¹, whereas both fermented and non‐fermented steam distillation CBS extracts were found to be active against Candida albicans with MICs at 1 mg mL⁻¹. Overall, steam distillation extracts possessed enhanced antimicrobial activity in comparison with solvent extracts of CBS. Fermented CBS extracts were found to possess better antimicrobial activity than non‐fermented CBS extracts. Metabolomics analysis identified theobromine (TB) and tetramethylpyrazine (TMP) as molecules that contributed to antimicrobial activity against S. mutans. Results showed that caffeine (CAF), TB and TMP were active against S. mutans and Acinetobacter baumannii, whereas CAF and TB were active against C. albicans. Significant synergistic effects of CAF, TB, and TMP with ciprofloxacin (CIP) were observed against Klebsiella aerogenes.

CONCLUSION

These findings highlight the significant potential of bioactive compounds present in CBS for use in the development of sustainable antimicrobial agents. These naturally occurring compounds, including CAF, TB, and TMP, exhibited notable antimicrobial properties, which could also be harnessed to enhance the activity of commonly used antibiotics such as ciprofloxacin. Future studies could focus on determining the mode of action of these bioactive molecules. © 2025 The Author(s). Journal of the Science of Food and Agriculture published by John Wiley & Sons Ltd on behalf of Society of Chemical Industry.

INTRODUCTION

Antimicrobial multidrug resistance (AMR) is a pressing global health issue exacerbated by the overuse and misuse of antibiotics across various sectors, including medicine, agriculture, livestock, and aquaculture. ¹ Although the development of new antibiotics is crucial for combating AMR, it has become increasingly difficult to bring new antibiotics to market, mainly due to the high costs involved and the exit of major pharmaceutical companies from this area.

Driven by the idea to lower the cost of antibiotic drug discovery as well as sustainable management of food and agricultural waste, the waste from natural products has often been explored for its antimicrobial potential. It has been shown repeatedly to offer a promising alternative to conventional drug discovery by harnessing the chemical diversity found in these underutilized resources. ² , ³ , ⁴ , ⁵ , ⁶ , ⁷

Malaysia is one of the largest cocoa traders in Asia and it generates substantial cocoa waste during industrial processing. ⁸ Cocoa waste poses environmental challenges, leading researchers to explore ways to repurpose it for beneficial uses. One of the avenues is to explore its medicinal potential as this waste is known to be rich in bioactive compounds such as polyphenols, methylxanthines, and lipids (Fig. 1). Cocoa bean shell (CBS) is one of the main by‐products. It represents approximately 10% to 17% of the total bean weight and is removed from the cotyledons before or after the roasting process. ⁹ Preliminary reports in recent years have highlighted the antidiabetic, ¹⁰ antioxidant, ¹¹ antimicrobial, ¹² , ¹³ , ¹⁴ , ¹⁵ and anti‐inflammatory ¹⁶ , ¹⁷ , ¹⁸ properties of CBS. The positive findings regarding its antimicrobial activity prompted this investigation into its potential.

Figure 1.

Bioactive compounds that commonly present in the cocoa bean shell (CBS).

Cocoa bean fermentation is a common step in cocoa processing. This process affects the phenolic content, protein, and fat content of cocoa. ¹⁹ , ²⁰ , ²¹ As far as the authors of the current study are aware, studies on the antimicrobial activity comparison of fermented and non‐fermented CBS extracts have yet to be reported.

This study analyzed the bioactive compounds that are present in fermented and non‐fermented CBS using various extraction protocols. The antimicrobial activities of these extracts were evaluated against gram‐positive and gram‐negative bacteria including the ‘ESKAPE’ pathogens, and Candida albicans. Each CBS extract exhibited varying bioactive compound content. Metabolomics analysis was also performed to investigate the relationship between the bioactive profiles of CBS extracts and their antimicrobial ability, as well as to identify the key compounds associated with their inhibitory activity. Selected CBS bioactive compounds were further evaluated for their antimicrobial activity and synergistic effect with ciprofloxacin.

MATERIALS AND METHODS

The fermented and non‐fermented dried CBS were obtained from the Malaysian Cocoa Board (Nilai, Negeri Sembilan, Malaysia). The fermented and non‐fermented CBS were prepared following a method described in the literature. ²² , ²³ Briefly, fresh cocoa beans (100 g) were fermented at room temperature for 5 days, followed by a second 5‐day fermentation with the addition of distilled water (60 mL). De‐shelling was performed via infrared micronization at 130 °C for 2 min at 600 rpm. The beans were then broken using a bean breaker, and the shells were separated from the nibs with a winnower. Non‐fermented CBS were de‐shelled using the same protocol, omitting the fermentation steps.

Solvent extraction of fermented CBS and non‐fermented CBS

The extraction of CBS was performed according to previous reports with minor modifications. ²⁴ , ²⁵ The samples were defatted before the extraction procedure. Briefly, ground CBS (50 g) was added to hexane (250 mL) and the mixture was stirred for 5 min. After stirring, the mixture was vacuum filtered and the process was repeated twice. The defatted samples were dried in an oven at 50 °C for 4 h. Five different solvents were used: ethyl acetate, ethanol, methanol, acidified ethanol‐water (pH 3, 80:20 v/v), and acidified methanol–water (pH 3, 80:20 v/v). The pH was adjusted using 0.1 mol L⁻¹ hydrochloric acid. Solvent extraction was carried out using a reflux system. Cocoa bean shell (5 g) was added to each extraction solvent (15 mL) and refluxed for 60 min with stirring. Each solvent was heated to its boiling point or slightly above: ethyl acetate and ethanol at 78 °C, methanol at 65 °C, acidified ethanol‐water at 80 °C, and acidified methanol–water at 70 °C. After it cooled to room temperature the filtrate was collected. The extraction process was repeated twice, and the collected filtrates were then combined and subsequently dried in vacuo to give the final extracts.

Steam distillation

Cocoa bean shell (60 g) and distilled water (600 mL) were mixed in a round‐bottom flask. The mixture was heated to boil on a heating mantle for 1.5 h. Around 100 mL of distillate was collected into the receiving flask. After the distillate was extracted with dichloromethane (DCM) (150 mL × 3), the organic solvent was removed in vacuo to obtain the steam distilled extract fraction.

Liquid chromatography–quadrupole time‐of‐flight mass spectrometry

The liquid chromatography–quadrupole time‐of‐flight mass spectrometry (LC‐QTOF‐MS) data were captured using an Agilent 1290 Infinity LC system coupled with the Agilent 6520 Accurate‐Mass quadrupole time of flight (QTOF) mass spectrometer (Agilent, Santa Clara, CA, USA). The chromatographic separation was achieved using an Agilent Eclipse XDB‐C18 column (150 mm × 2.1 mm, Agilent). The mobile phase consisted of 0.1% formic acid in water: 0.1% formic acid in acetonitrile, applied in a gradient from 95:5 (0–5 min) to 0:100 (5–20 min), followed by isocratic elution at 0:100 (20–25 min), at a flow rate of 0.5 mL min⁻¹. Data were acquired using Agilent MassHunter qualitative analysis B.07.00 software. Both positive and negative ionization modes were used. Secondary metabolites were identified by comparison with the METLIN database; compounds with a matching score below 80% were excluded from further analysis.

Gas chromatography–mass spectrometry

A QP2010 Ultra (Shimadzu, Kyoto, Japan) gas chromatography–mass spectrometry system was fitted with a Zebron ZB‐5MS column (30 m × 0.25 mm × 0.25 μm) of film thickness and maximum temperature of 350 °C. Ultra‐high‐purity helium (99.99%) was used as a carrier gas at a constant flow rate of 1.0 mL/min. The injection, transfer line, and ion source temperatures were all set at 280 °C. The oven temperature was programmed to start at 80 °C (for 2 min) and increased to 280 °C at a rate of 3° C/min. Crude extracts were diluted with a suitable solvent (1/100, v/v), filtered, and 1 μL of the diluted extract was injected with a split ratio of 10:1. Full‐scan mass spectra were collected over a scan range of 40–550 amu. The percentage composition of the crude extract constituents was determined by the percentage of peak area, with chemical compounds identified based on gas chromatography (GC) retention time and matched against the NIST 08 mass spectral library.

Statistical analyses

Statistical analyses of the untargeted metabolomic studies were conducted using the one factor module in MetaboAnalyst 6.0. ²⁶ The dataset was sum normalized, log transformed, and Pareto scaled to assess the variance in antimicrobial activity against Streptococcus mutans (active, moderately active, less active, and inactive) in fermented and non‐fermented CBS extracts using principal component analysis (PCA) and partial least‐squares discriminant analysis (PLS‐DA). Principal component analysis, an unsupervised multivariate analysis, used five components to visualize sample clustering based on metabolic profiles. Partial least‐squares discriminant analysis, a supervised method, was used to identify significant differences in metabolites extracted from fermented and non‐fermented CBS extracts. Model validation by permutation testing yielded P < 0.01, and fivefold cross‐validation was used to estimate prediction error and calculate Q ² and R ² values for the five components. Principal component analysis and PLS‐DA were not performed on steam‐distilled CBS extracts (F‐SD and NF‐SD) due to limited data. Variable importance in projection (VIP) scores were calculated to identify metabolites contributing to separation in the PLS‐DA plot; VIP scores greater than 1 indicated metabolites strongly discriminating between active and inactive samples. A heatmap was also generated to visualize metabolite concentrations across CBS extracts.

General protocol for antimicrobial activity test

Broth microdilution assay

The broth microdilution assay was conducted according to the Clinical and Laboratory Standards Institute (CLSI) guidelines. ²⁷ All bacteria strains were cultured on Mueller–Hinton agar (MHA), except for S. mutans, which was cultured on trypticase soy agar (TSA), and C. albicans, which was cultured on potato dextrose agar. After 24 h of incubation, the colonies were transferred into their respective broths and incubated for a further 24 h at 37 °C. The resulting turbid suspensions were then adjusted to a 0.5 McFarland standard containing 1 × 10 ⁶ CFU mL⁻¹. After inoculation, the 96‐well plates were incubated at 37 °C for 18 h. Minimum inhibitory concentrations (MICs) were determined visually as the lowest concentration at which no visible growth was observed. Ciprofloxacin (CIP) and cycloheximide served as positive controls for bacterial strains and C. albicans, respectively. All experiments were performed in triplicate.

Synergistic study of the CIP‐phenolic compound from CBS waste

The synergistic effects between CIP and compounds derived from CBS waste were evaluated against a range of microorganisms, including methicillin‐resistant Staphylococcus aureus (MRSA) ATCC 33591; methicillin‐susceptible S. aureus (MSSA) ATCC 25923 and ATCC 6538; Klebsiella pneumoniae ATCC 10031; Enterococcus faecalis ATCC 29212; E. faecium ATCC 700221; Bacillus cereus ATCC 14579; Pseudomonas aeruginosa ATCC 10145; Acinetobacter baumannii ATCC BAA‐1605 (multidrug resistant); A. baumannii strains C28, C65, and C98 (isolated from Segamat Hospital, Johor, Malaysia); ²⁸ Escherichia coli ATCC 10798; S. mutans ATCC 25175; B. subtilis ATCC 8188; K. aerogenes ATCC 13048; and C. albicans ATCC 10231.

The synergistic effects of CIP combined with bioactive compounds theobromine (TB), caffeine (CAF), and tetramethylpyrazine (TMP) present in the CBS were evaluated using the checkerboard dilution method with minor modifications. ²⁹ All compounds were tested for their synergistic effects at their maximum concentration in different solvents. Theobromine was dissolved in 0.5 mol L⁻¹ of NaOH water, CAF was dissolved in Mueller–Hinton broth (MHB), and TMP was dissolved in methanol. Each solvent was tested against the corresponding bacteria and fungi to ensure that it did not contribute to the killing at its final concentration. Serial dilutions of CBS compounds in combination with CIP were mixed in MHB. The inoculations were prepared from colonies grown overnight on MHA, except for S. mutans, which was grown in trypticase soy agar (TSA) and C. albicans, which was grown in potato dextrose agar (PDA). The synergistic effect was determined after incubation at 37 °C for 18 h. Each experiment was replicated three times. The in vitro interaction between the CBS compounds and CIP was quantified by determining the fractional inhibitory concentration index (FICI) using Eqn (1). The FICI of each compound was interpreted as follows: FICI of 0.5 or less indicates synergy; FICI between 0.5 and 1 indicates additivity; FICI between 1 and 2 indicates no effect, and FICI greater than 2 indicates antagonism. ³⁰

$$\text{FICI value}=\mathit{\text{FICI}}\left(X\right)+\mathit{\text{FICI}}\left(\mathit{CIP}\right)=\frac{\left[X\right]}{X\mathit{MIC}}+\frac{\left[\mathit{CIP}\right]}{\mathit{CIP}\mathit{MIC}}$$ (1)

where X = concentration of TB, CAF, or TMP.

RESULTS AND DISCUSSION

This study investigated the phytochemical composition and antimicrobial activity of fermented and non‐fermented cocoa bean shell (CBS) extracts. Twelve fermented (F) and non‐fermented (NF) CBS extracts were yielded from solvent extraction using ethyl acetate (F‐EtOAc, NF‐EtOAc), acidified ethanol (F‐A‐EtOH, NF‐A‐EtOH), ethanol (F‐EtOH, NF‐EtOH), acidified methanol (F‐A‐MeOH, NF‐A‐MeOH), methanol (F‐MeOH, NF‐MeOH), and steam distillation (F‐SD, NF‐SD). The non‐volatile phytochemical content of solvent extracts was determined by LC‐QTOF‐MS and the volatile content from steam distillation was analyzed using gas chromatography–mass spectrometry (GC–MS).

Analysis of the Non‐volatile and Volatile Phytochemicals from Fermented and Non‐fermented CBS Extracts

A total of 50 compounds were proposed from the LC‐QTOF‐MS analysis of both fermented and non‐fermented CBS solvent extracts. The non‐volatile compounds and their volume peak intensity of fermented extracts (F‐EtOAc, F‐A‐EtOH, F‐EtOH, F‐A‐MeOH and F‐MeOH) and non‐fermented extracts (NF‐EtOAc, NF‐A‐EtOH, NF‐EtOH, NF‐A‐MeOH and NF‐MeOH) are summarized in Supporting Information, Table S1, and depicted in Fig. S1, and the metabolites in each extract are illustrated in the heatmap in Fig. 2. In general, the major compounds that were found in fermented and non‐fermented solvent extracts are methylxanthines (TB, CAF) and polyphenol (proanthocyanidine, catechin derivatives, and caffeoyl derivatives) and a large variety of lipids (sphinganine, oleamide, and N‐pentadecylcyclohexanecarboxamide), with non‐fermented extracts exhibiting more diversity than fermented CBS. Higher concentrations of TB and CAF were found in fermented CBS extracts. This finding is in line with the results, where the non‐fermented CBS exhibited less TB and CAF content due to exudation from the bean to the shell after fermentation. ¹⁹ , ³¹ , ³² , ³³ , ³⁴

Figure 2.

Heatmap generated by MetaboAnalyst software, represents the abundance of metabolites from both fermented (F‐EtOAc, F‐A‐EtOH, F‐EtOH, F‐A‐MeOH and F‐MeOH) and non‐fermented CBS extracts (NF‐EtOAc, NF‐A‐EtOH, NF‐EtOH, NF‐A‐MeOH and NF‐MeOH). The color scale ranged from red (higher abundance) to blue (lower abundance).

Forty‐six volatile compounds were detected from the GC–MS analysis of fermented (F‐SD) and non‐fermented (NF‐SD) CBS steam distillation extracts. Most detected compounds are lipids, with the non‐fermented extracts containing a greater variety of lipids than fermented CBS. Supporting Information, Table S2 contains a list of volatile compounds found in F‐SD and NF‐SD. They are depicted in Supporting Information, Fig. S2. It is notable that TMP was found only in fermented CBS (F‐SD).

This study provided valuable insights into the phytochemical composition of fermented and non‐fermented CBS identified with LC‐QTOF‐MS and GC–MS analyses. However, the absence of structural confirmation from nuclear magnetic resonance (NMR) spectroscopy or further analytical data limited a more conclusive analysis of the compounds.

Antimicrobial Activity of Fermented and Non‐fermented CBS Extracts

The antimicrobial activity of the extracts was assessed against a panel of bacteria and fungus as outlined in Supporting Information, Table S3. The results revealed that none of the extracts exhibited bacterial inhibition at concentrations of 2 mg mL⁻¹ or lower, except for S. mutans and C. albicans (Table 1). Apart from F‐EtOAc, all CBS extracts demonstrated inhibition against S. mutans, with MICs ranging from 0.0625 to 2 mg mL⁻¹. Both fermented and non‐fermented steam distillation extracts displayed significant inhibition against C. albicans, with MICs of 1 mg mL⁻¹. Several studies have investigated the antimicrobial activity of extracts derived from cocoa bean, cocoa pod husk and CBS, with most of the cases reporting higher MIC values. It was reported that the cocoa powder extract exhibited MIC of 5 mg/mL against C. albicans, E. coli, S. aureus, and P. aeruginosa, while C. albicans was the most sensitive microorganism with the lowest MIC of 5 mg mL⁻¹. ³⁵ They also found that the cocoa powder extract had slightly higher potential against gram‐positive bacteria than gram‐negative bacteria. One study found that crude cocoa husk water–methanol extract inhibited S. choleraesuis and S. epidermidis with minimum inhibitory concentrations (MICs) of 1 mg mL⁻¹ and 2.5 mg mL⁻¹, respectively. ³⁶ Another study similarly demonstrated antimicrobial activity of a CBS ethanol–water extract against S. mutans, with an MIC as low as 0.0625 μg·mL⁻¹. However, no activity was observed against MSSA, MRSA, E. coli, P. aeruginosa, C. albicans, or S. cerevisiae. ¹²

Table 1.

The minimum inhibitory concentrations (MICs) of fermented and non‐fermented cocoa bean shell (CBS) extracts

Sample ID	Minimum inhibitory concentrations (MIC) (mg mL−1)
	S. mutans (ATCC 25175)	C. albicans (ATCC 10231)
F‐EtOAc	>2	>2
F‐A‐EtOH	2	>2
F‐EtOH	0.5	>2
F‐A‐MeOH	0.5	>2
F‐MeOH	0.5	>2
F‐SD	0.0625	1
NF‐EtOAc	1	>2
NF‐A‐EtOH	1	>2
NF‐EtOH	2	>2
NF‐A‐MeOH	1	>2
NF‐MeOH	1	>2
NF‐SD	0.125	1
Ciprofloxacin (CIP)	1 μg mL−1	0.156 μg mL−1
Cycloheximide (CHX)	‐	0.0625 μg mL−1

Comparative analysis between fermented and non‐fermented solvent extracts revealed that the fermented CBS extract exhibited marginally better antimicrobial activity than non‐fermented CBS extracts against S. mutans. The solvent extraction of fermented CBS extracts, F‐A‐EtOH, F‐EtOH, F‐A‐MeOH, and F‐MeOH showed a lower MIC of 0.5 mg mL⁻¹ against S. mutans than non‐fermented extracts, NF‐EtOAc, NF‐A‐EtOH, NF‐EtOH, NF‐A‐MeOH and NF‐MeOH, which had a MIC of 1 mg mL⁻¹. Earlier literature suggested that non‐fermented cocoa extracts often exhibit superior antimicrobial activity due to higher diversity of phytochemicals compared with their fermented cocoa extracts. These studies are focused on comparing the antimicrobial activity of fermented and non‐fermented cocoa bean extract instead of CBS extracts. Smullen et al. reported that the non‐fermented cocoa bean extract with a MIC of 4 mg mL⁻¹ demonstrated marginally better antibacterial activity against S. mutans than fermented cocoa, which had a MIC of 8 mg mL⁻¹. ³⁷ Moreover, the non‐fermented cocoa bean extract was reported to show better antibacterial activity against Porphyromonas gingivalis than fermented cocoa. ³⁸ In contrast, the findings of the current study indicated that fermented CBS extracts exhibited greater inhibitory activity against S. mutans than non‐fermented CBS extracts. The enhanced activity might be attributed to the higher levels of TB and CAF present in the fermented extracts in comparison with the non‐fermented CBS extracts.

Previous studies on CBS have focused solely on using organic solvents as extractants. The aim of the current study was to investigate the antimicrobial potential of steam distillates of CBS, a method that has not been reported. In this study, steam distillation extracts, which contained volatile phytochemicals, exhibited antimicrobial activity three times greater than that of solvent‐extracted samples. The fermented steam distillation CBS extract also demonstrated superior antibacterial activity against S. mutans in comparison with its non‐fermented counterpart (MIC of F‐SD: 0.0625 μg mL⁻¹; MIC of NF‐SD: 0.125 μg mL⁻¹).

Identification of Key Compounds for S. mutans Inhibition

It is essential to understand the specific phytochemicals responsible for the enhanced antimicrobial activity of fermented CBS extracts in order to evaluate their therapeutic potential. To identify the active compounds inhibiting S. mutans, the phytochemical compositions of CBS extracts were analyzed using LC‐QTOF‐MS and GC–MS. Inactive F‐EtOAc extracts were compared with active fermented and non‐fermented extracts using MetaboAnalyst. Unsupervised PCA and supervised PLS‐DA were performed (Fig. 3(a),(b), respectively) to classify non‐volatile bioactive compounds based on their metabolite profiles. Principal component analysis, which showed no clear group separation, explained 76% of the total variance across five components: PC1 (25.4%), PC2 (14.9%), PC3 (13.4%), PC4 (11.3%), and PC5 (10.7%). In contrast, PLS‐DA revealed separation of non‐volatile compounds among all extracts, with PC1 (11.5%), PC2 (17.2%), PC3 (16.7%), PC4 (8.8%), and PC5 (7.8%) accounting for a total variance of 62%. These findings indicate that the model effectively discriminated metabolites associated with varying levels of activity (active, moderately active, less active, and inactive) in fermented and non‐fermented solvent extracts. Principal component analysis and PLS‐DA analyses of volatile compounds from CBS steam distillation extracts (F‐SD and NF‐SD) were not performed due to the limited dataset, which comprised only two categories: highly active and less active.

Figure 3.

Data analysis using MetaboAnalyst software: (a) principal component analysis (PCA) and (b) partial least‐squares discriminant analysis (PLS‐DA) scores plot of metabolites from fermented (F‐EtOAc, F‐A‐EtOH, F‐EtOH, F‐A‐MeOH, and F‐MeOH) and non‐fermented extracts (NF‐EtOAc, NF‐A‐EtOH, NF‐EtOH, NF‐A‐MeOH, and NF‐MeOH). The plots distinguish between extracts based on their activity against S. mutans, categorized as active (minimum inhibitory concentration (MIC): 0.5 mg mL⁻¹), moderately active (MIC: 1 mg mL⁻¹), less active (MIC: 2 mg mL⁻¹), and inactive (MIC: >2 mg mL⁻¹).

The variable importance in projection (VIP) scores, based on the PLS‐DA model of non‐volatile and volatile compounds, are reported in Fig. 4(a),(b). This estimates the importance of each metabolite in the PLS model. In this study, 16‐hydroxy hexadecanoic acid, observed only in the F‐EtOH, F‐A‐MeOH and F‐MeOH CBS extracts, was the most significant metabolite and might be associated with the inhibition of S. mutans. However, no prior studies have reported on 16‐hydroxy hexadecanoic acid (juniperic acid) possessing antimicrobial activity against S. mutans or any other pathogens. Despite that, structurally related compounds such as palmitic acid, stearic acid, lauric acid, caprylic acid, and 3‐hydroxyundecanoic acid have all demonstrated antimicrobial properties. ³⁹ , ⁴⁰ , ⁴¹ , ⁴² Further experimental validation is thus required to assess its significance accurately.

Figure 4.

Data analysis using MetaboAnalyst software. The variable importance in projection (VIP) schematic scores of partial least‐squares discriminant analysis (PLS‐DA) analyses for (a) solvent cocoa bean shell (CBS) extracts (F‐EtOAc, F‐A‐EtOH, F‐EtOH, F‐A‐MeOH and F‐MeOH, NF‐EtOAc, NF‐A‐EtOH, NF‐EtOH, NF‐A‐MeOH and NF‐MeOH) and (b) steam distillation CBS extracts (F‐SD and NF‐SD). A high VIP score indicates greater discrimination between active and inactive samples against S. mutans. The color scale reflects metabolite abundance, with red representing higher abundance and blue representing lower abundance.

Based on the VIP score for volatile compounds, TMP, present only in F‐SD could potentially be one of the important metabolites that inhibit S. mutans. Tetramethylpyrazine has been reported for its anti‐inflammatory and anti‐cancer effects. ⁴³ , ⁴⁴ , ⁴⁵ , ⁴⁶ While numerous studies have reported on TMP's diverse biological activities, only limited research has investigated the analogs of TMP against S. aureus, B. subtilis, S. agalactiae, and E. coli. ⁴⁷ , ⁴⁸ , ⁴⁹ Moreover, the antibacterial activity of TMP against S. mutans has not been reported previously. In this study, TMP alone showed activity against S. mutans with MIC of 1.5 mg mL⁻¹ which is 24‐fold higher than the MIC of steam distillation extract (F‐SD, MIC: 0.0625 mg mL⁻¹), suggesting the synergistic effect of TMP with other compounds in the F‐SD extracts.

The weaker antibacterial activity of the NF‐SD extract against S. mutans might be due to the absence of TMP. NF‐SD contains more fatty acid esters and less free fatty acids, which are known to have stronger antibacterial properties, and this might also be one of the reasons for its weaker inhibition toward S. mutans. Studies indicated that the esterification of fatty acids tends to reduce antibacterial activity, except for esterified forms of sucrose. ⁵⁰ This study found that the NF‐SD extract mainly contains methylated and ethylated fatty acid esters with less free fatty acid than F‐SD.

The comparison of the key secondary metabolites (TB and CAF) in CBS waste showed that only TB significantly enhanced activity against S. mutans. Notably, TB exhibits substantial levels among the various extracts, with a false discovery rate (FDR) lower than 0.05, indicating its significance for antimicrobial activity. This is consistent with the results where TB (MIC: 1.25 mg mL⁻¹) exhibited better antimicrobial activity against S. mutans than CAF (MIC: 4 mg mL⁻¹), as shown in Table 2. Studies have also shown the antimicrobial effect of TB against caries‐associated bacteria, such as S. mutans and Actinomyces naeslundii, and non‐caries‐associated bacteria like Lactobacillus acidophilus and E. faecalis. ⁵¹ , ⁵²

Table 2.

The minimum inhibitory concentrations (MICs) of TB, CAF, and TMP against S. mutans, C. albicans and A. baumannii strains

Sample ID	Minimum inhibitory concentrations (mg mL−1)
	S. mutans (ATCC 25175)	C. albicans (ATCC 10231)	A. baumannii (ATCC BAA 1605)	A. baumannii (C65)
TB	1.25	0.625	>1.25	1.25
CAF	4	4	2	2
TMP	1.5	>1.5	1.5	>1.5
CIP	1 μg mL−1	0.156 μg mL−1	>1 μg mL−1	0.256 μg mL−1
CHX.	‐	0.0625 μg mL‐1	‐	‐

Abbreviations: TB, theobromine; TMP, tetramethylpyrazine; CAF, caffeine; CIP, ciprofloxacin, CHX, cycloheximide.

Caffeine, TB and TMP showed activity against A. baumannii, with MICs of 2, 1.25 and 1.5 mg mL⁻¹, respectively (Table 2). Although CAF has been reported to be active against A. baumannii at 4 mg mL⁻¹, activity against A. baumannii by TB and TMP have not been reported previously. ⁵³ Thus, results from this study show the antimicrobial potential of CAF, TB and TMP. Their synergistic potential with CIP against various pathogens is discussed in the next section.

Synergistic effects

The aim of the synergistic activity study was to assess the antimicrobial activity of common bioactive compounds in CBS like TB, CAF, and TMP against several microbes, alone or combined with CIP. Ciprofloxacin, a commonly prescribed antibiotic, was selected for the synergistic study due to its broad‐spectrum antimicrobial activity. However, resistance to CIP is a growing concern, which makes it an ideal candidate for synergy studies, as combining it with other bioactive compounds such as, CAF, TB and TMP could potentially help to restore its effectiveness, reduce the required dose, and potentially delay the development of resistance. The synergistic effect of TB, CAF, and TMP combined with CIP was determined by the FICI derived from the checkerboard synergism method. These synergistic effects were determined against a range of bacteria and fungi, including the ‘ESKAPE’ pathogens, B. cereus, E. faecalis, B. subtilis, S. mutans, E. coli, and C. albicans as outlined in Table 3.

Table 3.

The combined effects of TB, TMP and CAF with CIP against various microbial strains, as determined by the checkerboard method

Bacteria	Minimum inhibitory concentrations (MIC)				Fractional inhibitory concentration index (FICI)	Combination effect
	CBS bioactive mg mL−1		CIP μg mL−1	Mixed μg mL−1
MRSA	TB	>1.25	0.5	0.5	1.00	Indifference
ATCC 33591	CAF	16		>0.5	1.00	Indifference
	TMP	>1.5		0.5	1.00	Indifference
MSSA	TB	>1.25	0.004	0.004	1.00	Indifference
ATCC 25923	CAF	16		0.004	1.00	Indifference
	TMP	>1.5		0.004	1.00	Indifference
MSSA	TB	>1.25	0.002	0.002	1.00	Indifference
6538	CAF	16		0.002	1.00	Indifference
	TMP	>1.5		0.002	1.00	Indifference
K. pneumoniae	TB	>1.25	0.256	0.256	1.00	Indifference
ATCC 10031	CAF	8		0.256	1.00	Indifference
	TMP	>1.5		0.512	2.00	Indifference
E. faecalis	TB	>1.25	1	1	1.00	Indifference
ATCC 29212	CAF	8		1	1.00	Indifference
	TMP	>1.5		1	1.00	Indifference
E. faecium	TB	>1.25	0.032	0.032	1.00	Indifference
ATCC 700221	CAF	4		0.032	1.00	Indifference
	TMP	1.5		0.032	1.00	Indifference
C. albicans	TB	0.625	CHX	0.625	1.00	Indifference
10 231	CAF	4	0.625	0.625	1.00	Indifference
	TMP	>1.5		0.313	0.50	Synergy
B. cereus	TB	>1.25	0.128	0.256	2.00	Indifference
ATCC 14579	CAF	8		0.256	2.00	Indifference
	TMP	>1.5		0.256	2.00	Indifference
P. aeruginosa	TB	>1.25	0.256	0.256	1.00	Indifference
ATCC 10145	CAF	>16		0.128	0.50	Synergy
	TMP	>1.5		0.256	1.00	Indifference
A. baumannii	TB	>1.25	>0.512	>0.512	1.00	Indifference
BAA 1605	CAF	2		>0.512	1.00	Indifference
	TMP	1.5		>0.512	1.00	Indifference
A. baumannii	TB	1.25	0.256	>0.512	2.00	Indifference
C65	CAF	2		0.256	1.00	Indifference
	TMP	>1.5		0.256	1.00	Indifference
A. baumannii	TB	>1.25	0.128	0.128	1.00	Indifference
C28	CAF	2		0.128	1.00	Indifference
	TMP	>1.5		0.128	1.00	Indifference
A. baumannii	TB	>1.25	0.256	0.256	1.00	Indifference
C98	CAF	4		0.256	1.00	Indifference
	TMP	>1.5		0.256	1.00	Indifference
E. coli	TB	>1.25	0.016	0.016	1.00	Indifference
ATCC 10798	CAF	4		0.016	1.00	Indifference
	TMP	>1.5		0.016	1.00	Indifference
S. mutans	TB	1.25	1	1	1.00	Indifference
ATCC 25175	CAF	4		1	1.00	Indifference
	TMP	1.5		1	1.00	Indifference
B. subtilis	TB	1.25	0.256	0.125	0.48	Synergy
ATCC 8188	CAF	16		0.256	1.00	Indifference
	TMP	>1.5		0.256	1.00	Indifference
K. aerogenes	TB	1.25	0.5	0.0625	0.31	Synergy
ATCC 13048	CAF	16		0.0625	0.25	Synergy
	TMP	>1.5		0.125	0.38	Synergy

Abbreviations: TB, theobromine; TMP, tetramethylpyrazine; CAF, caffeine; CIP, ciprofloxacin; CBS, cocoa bean shell; MSSA, Methicillin‐susceptible S. aureus; MRSA, Methicillin‐resistant S. aurues.

Multiple synergistic relationships were noted between CIP/CHX and common compounds found in CBS waste. It is interesting to note that a synergistic effect was observed when CIP was combined with all three tested metabolites found in CBS (TB, CAF, and TMP) against K. aerogenes. This finding is significant as K. aerogenes is one of the most virulent pathogens and has displayed resistance to multiple antibiotics.

Similarly, synergy was observed between TMP and CHX against C. albicans, and between CAF and CIP against P. aeruginosa. Theobromine and CIP also demonstrated a positive synergistic effect against B. subtilis. The mechanisms underlying these interactions are not yet understood and this warrants further investigation.

CONCLUSION

Both fermented and non‐fermented CBS extracts were found to exhibit antimicrobial activity against S. mutans and C. albicans with fermented extracts showing better activity. Steam distilled extracts also showed better antibacterial activity than solvent extracts. Using a metabolomic approach, TMP was identified as a potential antibacterial agent against S. mutans. The results indicated that the composition of fatty acid esters and free fatty acids impacts the antimicrobial efficacy significantly. The stronger antibacterial activity of fermented steam distillation extracts could be attributed to the higher content of free fatty acids, with less methylated and ethylated fatty acid. A positive synergistic effect was also observed when CAF, TB, and TMP were combined with CIP against K. aerogenes but remained indifferent against most of the other bacterial and fungal strains tested. Synergy was also observed when TMP was combined with CHX against C. albicans. Caffeine and CIP also worked synergistically against P. aeruginosa while TB and CIP also showed positive synergy against B. subtilis.

In conclusion, this study shows the complex interplay of various metabolites found in fermented and non‐fermented CBS extracts and their contributions to antimicrobial activity. Although some of the key bioactive compounds have been identified, the lack of structural confirmation by NMR spectroscopy or further analytical data limits a more conclusive analysis of the proposed compounds. The mechanism of action also remains unclear. Future studies could thus focus on targeted metabolic or enzymatic analyses to elucidate the mode of action of these bioactive molecules.

AUTHOR CONTRIBUTIONS

Yie Kie Chong: investigation; formal analysis; writing – original draft. Wei Khang Gan: investigation; formal analysis; writing – original draft. Joash Ban Lee Tan: writing – review and editing. Ahmad Kamil Hj Mohd Jaaffar: writing – review and editing. Zainal Baharum: writing – review and editing. Keng Yoon Yeong: conceptualization, supervision, writing – review and editing. All authors have read and agreed to the published version of the manuscript.

CONFLICT OF INTEREST

The authors declare no conflict of interest.

Supporting information

Data S1. Supporting Information.

DATA AVAILABILITY STATEMENT

The data that support the findings of this study are available from the corresponding author upon reasonable request.