The Antimycobacterial Activity, Toxicity and Chemical Composition of South African Propolis

ABSTRACT

Global research has explored the antimicrobial properties of propolis; however, the antimycobacterial activity of South African (SA) propolis has been neglected. This study evaluated 109 propolis samples collected from eight SA provinces where the antimycobacterial activity, chemical composition and toxicity was evaluated. Samples were tested using a broth microdilution assay against various Mycobacterium strains (M. fortuitum ATCC 6841, M. gordonae ATCC 14470 and M. smegmatis ATCC 14468) and the pathogenic M. tuberculosis H37Rv strain. A multi‐step approach combining ultra‐performance liquid chromatography/ mass spectrometry (UPLC–MS) profiling, high‐performance thin‐layer chromatography (HPTLC) bio‐autography, brine shrimp toxicity testing and molecular docking was employed on eight bio‐active and eight non‐active propolis samples. Two samples (GP68 and EC77B) demonstrated promising activity, with MICs ranging from 49–195 µg/mL against non‐tuberculous strains and 125 µg/mL against M. tuberculosis. Chemometric and bio‐autography analyses identified flavonoids and related compounds, including pinocembrin, galangin, chrysin and pinobanksin compounds as major contributors to activity. Both samples were non‐toxic at concentrations (1 mg/mL at 24 h; 0.5 mg/mL at 48 h). Molecular docking indicated favourable binding to key M. tuberculosis targets involved in cell wall biosynthesis (DprE1) and transcription (RNAP). This study demonstrates that selected SA‐derived propolis samples show strong antimycobacterial potential.

This study presents one of the largest geographically diverse propolis collections reported globally, comprising 109 samples sourced from eight provinces across South Africa. The novelty of this work lies in the fact that it presents the first investigation of South African propolis against M. tuberculosis. The study integrates microbiology, advanced chemical profiling, chemometrics, toxicity screening and in silico target validation.

1. Introduction

The genus Mycobacterium comprises over 190 species of aerobic, non‐spore‐forming, acid‐fast bacteria characterised by a complex cell wall that limits permeability and contributes to intrinsic resistance to antimicrobial agents. These bacteria are broadly classified into the Mycobacterium tuberculosis complex (MTBC) and non‐tuberculous mycobacteria (NTM) [1, 2, 3]. Mycobacterium tuberculosis (MTB), the causative agent of tuberculosis (TB), remains one of the leading causes of death from a single infectious agent worldwide [4]. According to the World Health Organisation (WHO), approximately 8.2 million people were diagnosed with TB in 2023, with drug‐resistant forms continuing to pose major treatment challenges [5]. The prolonged treatment duration required for TB and the emergence of drug‐resistant strains significantly complicate disease management and contribute to poor patient compliance [6].

Non‐tuberculous mycobacteria represent a diverse group of environmental organisms commonly found in soil and water. These organisms are increasingly recognised as opportunistic pathogens, particularly in immunocompromised individuals and those with underlying pulmonary disease [7, 8]. These infections are often misdiagnosed as TB due to similarities in clinical presentation and acid‐fast staining, resulting in delayed or inappropriate treatment [8]. In addition, treatment of NTM infections remains challenging because susceptibility profiles vary widely among species and standardised treatment protocols are not available for all NTM species [8, 9, 10]. These challenges highlight the urgent need for novel antimycobacterial agents with improved efficacy and reduced treatment duration.

Natural products remain a valuable source of antimicrobial agents due to their chemical diversity and biological activity [11]. Propolis, a resinous substance produced by honeybees (Apis mellifera L.), has gained considerable interest due to its diverse phytochemical composition and broad‐spectrum antimicrobial activity. Bees collect plant resins and mix them with beeswax and enzymes to produce propolis, which serves as a protective material within the hive by preventing microbial growth and structural decay [12, 13, 14]. More than 800 compounds have been identified in propolis, including flavonoids, phenolic esters, terpenes and other bio‐active constituents, with polyphenols typically representing the major chemical components [15, 16, 17].

Several studies have demonstrated the antimycobacterial potential of propolis from different geographic regions, including Europe, Asia and South America [10, 18, 19, 20, 21]. However, most of these investigations have been limited by small sample sizes and restricted geographic representation, providing only a narrow view of the antimycobacterial potential of propolis. In particular, South African propolis remains underexplored despite its chemical diversity and demonstrated antimicrobial potential observed against other bacterial species [22, 23, 24, 25]. The limited number of studies evaluating antimycobacterial activity highlights a significant knowledge gap.

Although propolis exhibits promising antimicrobial activity, the safety profile must be considered. Previous studies generally indicate that propolis is relatively non‐toxic; however, variability in toxicity has been reported among samples from different geographic regions [26, 27]. Therefore, evaluating both the antimycobacterial activity and toxicity of South African propolis is essential to support the potential as an antimycobacterial agent. This study addresses this gap by evaluating a large and geographically diverse collection of South African propolis samples for antimycobacterial activity, evaluates the chemical composition and toxicity, as well as examines the molecular targets, thereby providing one of the most comprehensive assessments of the antimycobacterial potential of South African propolis to date.

2. Results and Discussion

2.1. Antimycobacterial Activity

The MIC values for the South African propolis ethanolic extracts against the NTM ranged from 49 to 781 µg/mL, while the MIC values for the three Brazilian samples included as comparative references, ranged from 195 to 781 µg/mL. The MIC values of ciprofloxacin ranged from 0.31 to 0.04 µg/mL, while for rifampicin, they ranged from 0.13 to 15.60 µg/mL against the NTM. The MIC for the control drugs, rifampicin, streptomycin and isoniazid against MTB was 0.003, 0.20 and 0.05 µg/mL, respectively, after 14 days of incubation (Table 1).

TABLE 1.

Minimum inhibitory concentration values (µg/mL) of South African propolis against M. smegmatis, M. fortuitum, M. gordonae and M. tuberculosis.

Province	Locations	Sample nos.	M. smegmatis (ATCC 14468)	M. fortuitum (ATCC 6841)	M. gordonae (ATCC 14470)	M. tuberculosis H37Rv
Eastern Cape	Alicedale	EC48	781	781	195 a	>500
	Baviaanskloof	EC27	781	391	391 b	>500
	Brandhoek	EC53	781	391	391	>500
	Gqeberha	EC77A	391	391	195	500
		EC77B	195	195	49	125
		EC77C	391	195	98	250
		EC77D	195	195	49	250
		EC77E	391	195	98	500
	Kariega River Valley	EC50	781	781	195	500
Free State	Assisi Mission Farm Fort Savage	FS71	781	391	195	250
	Bloemfontein	FS24	195	195	98	>500
	Sannaspos	FS2	391	391	195	>500
	Setsoto Local Municipality	FS51A	781	391	391	>500
		FS51B	781	391	391	>500
		FS64A	391	195	195	500
		FS64B	781	391	195	500
		FS64C	195	195	98	500
		FS64D	391	195	98	250
	Smithfield	FS70	391	391	195	500
Gauteng	Craighall Park, Johannesburg	GP17	391	391	195	>500
	Lyttelton Manor, Centurion	GP35	391	391	98	>500
	Bapsfontein	GP11	195	391	98	250
	Beaulieu	GP12	391	391	195	>500
	Benoni	GP26	781	391	195	>500
	Blairgowrie	GP13	781	391	98	500
	Boekenhoutskloof, Tshwane	GP63	781	391	195	500
	Bronkhorstspruit	GP62	781	781	391	>500
	Bryanston	GP4	391	391	98	>500
	Central Pretoria	GP37	391	391	391	>500
	Cradle of Humankind	GP1	781	781	195	>500
	Edenvale	GP25	391	391	98	250
	Equestria	GP78A	391	391	49	250
		GP78B	391	195	98	250
		GP78C	391	391	98	250
	Florida Hills	GP9A	781	391	195	>500
		GP9B	781	781	195	>500
	Fourways	GP52	391	195	49	250
	No specific location provided	GP21	391	391	391	>500
		GP29	195	195	49	>500
		GP82	391	391	195	500
	Hammanskraal	GP38	391	391	195	500
	Sandton	GP68	98	98	98	125
	Midrand	GP23	781	391	391	>500
	Midrand	GP59A	195	195	98	250
		GP59B	391	195	195	250
	Muldersdrift, Driefontein road	GP5	391	195	98	500
	Paulshof	GP69	781	781	195	>500
	Pretoria	GP28A	391	195	98	500
		GP28B	195	195	98	250
		GP31A	391	781	98	500
		GP31B	781	781	391	>500
		GP39	781	391	195	500
	Pretoria East—Stone Hill Conservatory	GP45A	391	391	195	500
		GP45B	391	391	195	500
		GP45C	781	391	195	500
	Randburg	GP22	391	391	391	>500
		GP32	391	391	98	500
	Rietvlei Nature Reserve	GP61	781	781	195	>500
	Sandton, Johannesburg	GP3A	391	391	195	>500
		GP3B	391	391	195	>500
	Springs	GP10	195	195	98	500
	Suikerbosrand Nature Reserve	GP33A	391	391	98	>500
		GP33B	781	391	195	>500
	Sunninghill Gardens	GP6A	391	391	195	>500
		GP6B	391	391	98	>500
	Walkerville	GP16	391	391	195	>500
	Sandton	GP67	391	391	195	125
	Woodmead, Johannesburg	GP60	195	195	195	250
	Vereeniging	GP83	391	391	195	500
KwaZulu‐Natal	Amanzimtoti Scout Hall	KZN54	781	391	391	500
	Glendale Heights	KZN55	391	391	391	500
	Graceland Farm	KZN72A	781	391	195	500
		KZN72B	781	391	195	500
	Ifafa, South Coast	KZN57	781	391	781	>500
	Pietermaritzburg	KZN47	781	391	391	500
	Tala Valley	KZN66	781	781	195	>500
Limpopo	Amandelbult Mine	LP65	391	391	195	>500
	Bushveld	LP36	391	391	195	500
	Mookgophong	LP49	781	781	391	500
	Thabazimbi	LP8	781	391	391	>500
		LP56	781	391	781	>500
	Waterberg	LP58	781	781	781	>500
Northern Cape	Calvinia, Namaqualand	NC34	781	781	391	>500
North West	Beestekraal	NW75	781	781	391	500
	Hartbeespoort	NW79	391	391	195	500
Western Cape	Beaufort West	WC30	391	195	98	500
	Botrivier	WC19	391	391	195	500
	Cape Point	WC44	391	195	195	>500
	Cape Town South	WC80	391	195	98	250
	Constantia	WC42	391	195	98	>500
	Fish Hoek farms	WC41	781	391	98	>500
	Kouebokkeveld	WC73	391	391	195	>500
Western Cape	Lakeside, West Lake	WC15	391	195	98	>500
	Muldersvlei, Stellenbosch	WC46	391	391	195	500
	Newlands Forest	WC43	391	195	49	>500
	Riebeek West	WC74	781	781	195	>500
	Riviersonderend Mountains McGregor	WC14A	391	391	391	>500
		WC14B	781	781	391	>500
	Somerset West	WC18	391	391	195	>500
	Stanford	WC81	195	195	98	500
	Stellenbosch	WC40	391	391	195	>500
	No exact location provided	WC76A	781	781	391	>500
		WC76B	781	781	195	>500
		WC76C	781	391	195	500
		WC76D	781	391	195	>500
		WC76E	781	781	195	>500
	Witsand	WC7A	391	391	49	>500
		WC7B	391	391	391	>500
	Yzerfontein	WC84	781	781	391	250
Brazil	Lagos	BR20A	781	391	195	500
		BR20B	781	781	195	500
		BR20C	781	781	195	>500
Controls
Ciprofloxacin			0.31	0.16	0.04	—
Rifampicin			15.60	15.60	0.13	0.003
Streptomycin			12.50	12.50	— c	0.20
Isoniazid			12.50	6.25	—	0.05
Negative control (acetone)			>6250	>6250	>6250	>6250

^a

Noteworthy anti‐mycobacterial activity annotated in bold.

^b

moderate anti‐mycobacterial activity annotated in italics.

^c

not tested.

Minimum inhibitory concentration values ≤ 195 µg/mL were classified as noteworthy, while MIC values greater than 195 µg/mL but ≤ 500 µg/mL were considered moderate and MIC values > 500 µg/mL were deemed weak. Using this criterion, twelve South African propolis samples showed noteworthy antimicrobial activity (MIC ≤ 195 µg/mL) against M. smegmatis, whereas 54 samples demonstrated moderate activity (195< MIC ≤ 500 µg/mL). Against M. fortuitum, 26 South African samples displayed noteworthy activity, while 62 exhibited moderate activity. Mycobacterium gordonae was the most susceptible NTM, with 84 South African samples showing noteworthy activity and 22 exhibiting moderate activity. When the samples were tested against MTB, 51 samples exhibited moderate activity (16 samples with an MIC of 250 µg/mL and 35 samples with an MIC of 500 µg/mL). Three samples (GP67, GP68 and EC77B) exhibited noteworthy activity (MIC = 125 µg/mL).

The three comparatives Brazilian propolis samples showed weak activity against M. smegmatis, with MIC values of 781 µg/mL. Against M. fortuitum, one sample (BR20A) exhibited moderate activity (MIC = 391 µg/mL), while the remaining two (BR20B and BR20C) were weakly active (MIC = 781 µg/mL). Notably, all three Brazilian samples displayed noteworthy activity against M. gordonae (MIC = 195 µg/mL). Against MTB, Brazilian samples BR20A and BR20B showed moderate activity at an MIC of 500 µg/mL, whilst one sample (BR20C) had an MIC >500 µg/mL.

Propolis samples from Gauteng were more active, followed by those from the Eastern Cape, Free State and Western Cape. From the propolis samples that displayed activity, 11 samples demonstrated noteworthy activity (six from Gauteng, two from Eastern Cape, two from Free State and one from the Western Cape) against all the NTM. Sample GP68, collected from Sandton, Gauteng, South Africa, exhibited the most potent activity, with an MIC of 98 µg/mL against all NTM and 125 µg/mL against MTB. These findings demonstrate that several South African propolis samples exhibit notable antimycobacterial activity, against NTMs with selected samples also showing noteworthy inhibition against MTB. The variability in activity among samples highlights the influence of geographic origin on antimicrobial efficacy.

Notable antimycobacterial activity has been reported in propolis samples from various geographical regions, demonstrating that antimicrobial efficacy is strongly influenced by botanical origin and chemical composition. For example, a previous study [28], the reported antimycobacterial activity of four propolis samples collected in Turkey, demonstrated noteworthy activity against M. smegmatis with MIC values ranging from 31 to 125 µg/mL. These values are comparable to the lower MIC values observed for selected South African samples in the current study, particularly GP68 and EC77B, suggesting that flavonoid‐rich propolis types may share similar inhibitory properties across different geographical regions. Similarly, chestnut‐origin and polyfloral‐origin from Turkey derived propolis demonstrated susceptibility of M. smegmatis, with MIC values of 19.50 and 39.06 µg/mL, respectively [29]. The stronger activity reported for these Turkish samples compared to some South African samples may reflect differences in plant foraging sources, which influence flavonoid diversity and concentration [17].

Furthermore [10], evaluated Brazilian propolis against various NTM species (M. kansasii, M. avium, M. fortuitum, M. abscessus and M. chelonae) and reported MIC values > 200 µg/mL against M. fortuitum. This finding closely aligns with the moderate to weak activity observed for the comparator Brazilian samples included in the current study, supporting the notion that Brazilian propolis may exhibit reduced activity against certain NTMs compared to South African propolis. Collectively, these findings suggest that although propolis from different regions demonstrates variable MIC values, the inhibitory trends observed across NTMs remain broadly consistent.

The antimycobacterial activity of propolis (albeit sourced from other countries and thus representing distinct chemotypes) against MTB has been documented in only a limited number of studies [18, 19, 20, 21, 30], highlighting the continued need for more geographically diverse investigations. These studies collectively demonstrate that propolis from different origins exhibits variable but measurable inhibitory activity against MTB. For instance, Brazilian propolis has been shown to produce moderate inhibition against the MTB H37Rv strain [18], while Polish propolis extracts demonstrated partial to complete inhibition of several MTB strains, with complete susceptibility observed in six clinical isolates at concentrations ≤1.8 mg/mL [19]. Similarly, Chilean propolis inhibited 92% of MTB H37Rv growth at 300 µg/mL [30] and Nepalese propolis exhibited bacteriostatic activity (MIC 8–32 µg/mL) against the H37Ra strain [20].

More recently [21], a study reported notable antimycobacterial activity of Brazilian propolis, with MIC values of 62.5 µg/mL against MTB clinical and laboratory strains and 31.25 µg/mL against rifampicin‐resistant MTB strains. In contrast with these findings, the Brazilian propolis samples evaluated in the current study exhibited weak activity against MTB.

This contrast may be attributed to differences in chemical composition, geographic origin or extraction methodology, all of which can significantly influence propolis bio‐activity. Additionally, variations in resin sources available to bees in different regions may account for differences in the abundance of key bio‐active flavonoids [17], which play a critical role in antimycobacterial activity.

South African propolis displayed stronger activity against NTMs than against MTB, with inhibition of MTB generally requiring higher concentrations (≥500 µg/mL). This observation may be attributed to the unique lipid‐rich structure of the MTB cell wall, which restricts drug penetration and contributes to resistance to antimycobacterial agents. Such structural characteristics have been widely recognised as key determinants of reduced antimicrobial susceptibility in mycobacteria [31, 32]. Notably, two samples (EC77B and GP68) that were particularly active against NTMs also demonstrated relatively lower MIC values against MTB. This suggests a partial correlation in susceptibility patterns across NTMs and MTB. These findings suggest that screening against NTMs may provide useful preliminary insight into antimycobacterial potential. However, NTMs may overestimate the potency of propolis relative to MTB due to differences in cell wall complexity and metabolic behaviour. Therefore, although M. smegmatis and M. fortuitum serve as valuable preliminary screening models, testing against MTB remains essential. Overall, these results highlight the variability in antimycobacterial activity among propolis samples and reinforce the potential of selected South African propolis as candidates for further antimycobacterial investigation.

2.2. Chemical Analysis

Compounds were identified by matching retention times (Rt), mass‐to‐charge values (m/z) and fragmentation patterns with data available in the literature and databases (Table 2). Eight South African propolis samples displaying the best antimycobacterial activity against all NTMs and MTB were selected, along with eight samples exhibiting weak activity against all the tested Mycobacterium species. A total of 25 compounds were identified in the selected South African propolis samples (Figures S1 and S2), and 14 compounds have been identified in the Brazilian sample (Figure S3). All the active samples contained pinobanksin (7), pinostrobin (10), 5‐methoxygalangin (13), 3‐O‐acetylpinobanksin/pinobanksin‐3‐O‐acetate (21), pinobanksin‐3‐O‐methylbutyrate (34) and pinobanksin‐3‐O‐hexanoate (37), whereas the Brazilian sample was found to contain artepillin C (33). Pinocembrin (18) and galangin (19) were found in all the active samples except for sample GP10, and chrysin (16) was found in all the active samples except for EC77D.

TABLE 2.

UPLC–MS identification of compounds in eight active South African propolis and one Brazilian sample.

Peak #	Rt (mins)	Elemental formula	m/z calculated	m/z theoretical	Mass error (Da)	Fragments	Compound ID	Active								Control
								GP 10 a	GP 11 b	GP 28B c	GP 52 d	GP 60 e	GP 68 f	EC 77B g	EC 77D h	BR 20A i
1	3.05	C9H8O3	163.04	163.04	3.4	‐ j	p—coumaric acid	−	−	−	−	−	+ k	−	−	+
2	3.49	C25H23O3	515.12	515.12	0.4	353, 191	di‐caffeoylquinic acid	−	−	−	−	−	−	−	−	+
3	3.72	C25H23O3	515.12	515.12	0.4	353, 191	di‐caffeoylquinic acid	−	−	−	−	−	−	−	−	+
4	5.30	C16H14O5	285.08	285.07	4.5	267, 252, 239, 139	Pinobanksin‐methyl ether	+	−	+	−	+	+	+	−	−
5	5.38	C16H14O5	543.15	543.15	1.7	—	Unidentified	−	−	−	−	−	−	−	−	+
6	5.64	C15H10O6	285.04	285.04	−1.2	255	Kaempferol	−	+	−	+	−	+	+	+	−
7	5.84	C15H12O5	271.06	271.06	−4.9	253	Pinobanksin	+	+	+	+	+	+	+	+	−
8	5.86	C16H14O6	301.08	301.07	−5.5	283,152	Methoxy pinobanksin	−	−	−	−	−	−	−	−	+
9	6.04	C16H11O6	299.06	299.05	−5.7	284	Kaempferol‐3‐methyl ether	+	+	+	−	−	+	+	+	−
10	6.11	C16H14O4	269.08	269.08	−12.9	254, 191, 165	Pinostrobin	+	+	+	+	+	+	+	+	−
11	6.27	C17H14O7	329.06	329.08	−6.0	314, 299, 271	Dimethoxyquercetin isomer	+	+	+	−	+	+	+	+	−
12	6.46	C17H14O7	247.10	247.10	−1.3	—	Caffeic acid prenyl ester	−	−	−	−	−	−	−	−	+
13	6.72	C16H12O5	283.06	283.06	−5.2	268, 239, 211	5‐Methoxygalangin	+	+	+	+	+	+	+	+	−
14	6.99	C14H15O3	231.10	231.20	−3.6	—	p‐Coumaric acid isoprenyl ester	−	−	−	−	−	−	−	−	+
15	7.30	C19H24O4	315.16	315.16	−4.0	297, 241	Capillartemisin A	−	−	−	−	−	−	−	−	+
16	7.52	C15H10O4	253.04	253.05	4.7	145	Chrysin	+	+	+	+	+	+	+	−	−
17	7.59	C16H14O5	285.08	285.07	−3.0	271, 164, 151	Vestitone	−	−	−	−	−	−	−	−	−
18	7.67	C15H12O4	255.07	255.06	−6.4	213, 151	Pinocembrin	−	+	+	+	+	+	+	+	+
19	7.76	C15H10O5	269.05	269.04	−4.1	227	Galangin	−	+	+	+	+	+	+	+	+
20	7.76	C16H12O6	299.06	299.05	−1.8	284, 255, 227, 151	Kaempferide	−	−	−	−	−	−	−	−	−
21	7.88	C17H14O6	313.07	313.07	−3.8	253	3‐O‐Acetylpinobanksin / Pinobanksin‐3‐O‐acetate	+	+	+	+	+	+	+	+	−
22	8.00	C17H14O7	329.06	329.06	1.8	299, 271	Quercetin‐dimethyl‐ether	+	+	+	−	−	−	−	−	+
23	8.09	C23H22O6	329.13	393.13	0.1	349, 163, 145	5‐isoprenyl caffeic acid‐p‐coumaric acid ester	+	−	−	−	−	−	−	−	−
24	8.18	C16H12O5	283.06	283.06	−5.2	268, 239, 211	5‐Methoxygalangin	+	+	+	−	+	+	+	+	−
25	8.24	C21H29O6	377.20	377.19	−1.6	273, 201	Unidentified	−	−	−	−	−	−	−	−	+
26	8.40	C16H20O9	355.10	355.11	8.7	253	Pinobanksin‐3‐O‐pentanoate	+	+	−	−	−	+	+	−	−
27	8.52	C15H10O4	253.04	253.05	4.7	145	Chrysin	+	−	−	+	−	+	−	−	−
28	8.61	C18H16O4	295.10	295.09	−9.7	178, 134	Caffeic acid cinnamyl ester	+	+	−	−	−	+	−	−	+
29	8.79	C18H16O6	327.09	327.08	−5.2	271, 253	Pinobanksin‐3‐O‐propionate	+	+	+	−	+	+	+	+	−
30	9.23	C19H18O6 (C20H20O8)	341.10 (387.12)	341.10 (387.12)	−1.1	319, 271, 253	Pinobanksin‐3‐O‐butyrate/isobutyrate (Formic acid adduct)	−	−	+	−	+	+	+	+	−
31	9.46	C18H16O3	279.10	279.10	−6.0	__	p‐Cinnamyl coumarate	+	+	+	−	+	+	+	+	−
32	9.62	C19H18O6	341.10	341.10	−5.2	271, 253	Pinobanksin‐3‐O‐butyrate/isobutyrate	+	+	−	+	+	+	+	+	−
33	9.86	C19H24O3	299.16	299.16	−0.7	255, 200	Artepillin C	−	−	−	−	−	−	−	−	+
34	10.40	C20H20O6	355.10	355.11	9.8	253	Pinobanksin‐3‐O‐methylbutyrate	+	+	+	+	+	+	+	+	−
35	10.62	C24H20O6	403.12	403.12	−3.0	299, 271, 253	Pinobanksin‐3‐O‐phenylpropionate/hydroxycinnamate	−	+	+	−	+	−	+	−	−
36	10.89	C28H32O5	447.21	447.22	−0.9	297, 253, 198, 149	(E)‐3‐[2,3‐dihydro‐2‐(1‐methylethenyl)‐7‐prenyl‐5‐benzofuranyl] ‐2‐propenoic acid	−	−	−	−	−	−	−	−	+
37	11.11	C21H22O6	369.13	369.13	−6.1	271, 253	Pinobanksin‐3‐O‐hexanoate	+	+	+	+	+	+	+	+	−
38	11.17	C24H37O5	405.26	405.26	−3.0	−	Unidentified	−	−	−	−	−	−	−	−	+
39	11.31	C24H37O5	405.26	405.27	−1.2	−	Unidentified	−	−	−	−	−	−	−	−	+

^a South African active sample 10 (Gauteng).

^b sample 11 (Gauteng).

^c sample 28B (Gauteng).

^d sample 52 (Gauteng).

^e sample 60 (Gauteng).

^f sample 68 (Gauteng).

^g sample 77B (Eastern Cape).

^h sample 77D (Eastern Cape).

ⁱ Brazilian sample A.

^j Undetected.

^k compound present.

Pinobanksin derivatives (pinobanksin‐3‐O‐butyrate or isobutyrate (32), pinobanksin‐3‐O‐propionate (29), pinobanksin‐3‐O‐pentanoate or 2‐methylbutyrate (26), pinobanksin‐3‐O‐hexanoate (37), 3‐O‐acetylpinobanksin (21)), flavonoids (kaempferol‐3‐methyl ether (9), dimethoxyquercetin (11), 5‐methoxygalangin (13)) and phenolic esters such as p‐cinnamyl coumarate (31) have previously been identified in South African propolis [22, 23, 25, 33]. Additionally, Brazilian propolis was previously reported to contain high concentrations of only one flavonoid, artepillin C [25]. The findings from previous studies are consistent with findings from the current study, demonstrating that South African propolis is chemically distinct from Brazilian propolis, which is typically characterised by high concentrations of artepillin C, whereas South African propolis generally contains diverse flavonoids. The predominance of flavonoids such as pinocembrin, galangin, chrysin and pinobanksin compounds indicates that the South African propolis samples investigated in this study are chemically consistent with poplar‐type propolis, a classification widely reported for South African propolis [22, 25, 33].

Active and inactive samples differed in the chemical composition. Inactive samples (GP1, WC14B, GP31B, LP56) did not contain compounds that are present in active samples, namely pinobanksin (7), pinostrobin (10), 5‐methoxygalangin (13), pinocembrin (18), galangin (19), 3‐O‐acetylpinobanksin / pinobanksin‐3‐O‐acetate (21), pinobanksin‐3‐O‐methylbutyrate (34) and pinobanksin‐3‐O‐hexanoate (37) (Table S1). Although certain inactive samples (LP8, NC34, LP58 and WC76E) contained some of the same compounds identified in active samples, the heatmap analysis (Figure 1) shows that these compounds were present at low concentrations, potentially accounting for the lack of biological activity.

FIGURE 1.

Heatmap (A) and VIP plot (B) obtained from the PLS–DA analysis, indicating potential markers and distinguishing between eight active and eight inactive samples based on relative peak intensities.

The heatmap analysis and a VIP plot were obtained from the PLS‐DA model, which highlighted potential chemical markers and distinguished active from inactive samples based on relative peak intensities (Figure 1). Pinocembrin (18) was identified as the compound found in high concentration in the South African active propolis samples. The predominance of pinocembrin in South African propolis has previously been reported [34], where this compound may contribute to the antimicrobial activity of the South African propolis samples was proposed.

Additional compounds found in high concentrations included galangin (19), 3‐O‐acetylpinobanksin (21), pinobanksin‐3‐O‐methylbutyrate (34) and pinobanksin‐3‐O‐hexanoate (37). Pinocembrin and galangin, identified as abundant compounds in active samples, have previously demonstrated strong antimycobacterial activity, with MIC values ranging from 3.5 to 4.0 µg/mL against M. tuberculosis H37Rv [35, 36]. The presence of these compounds in high concentrations in the most active samples, particularly GP68, strongly supports their contribution to the antimycobacterial activity observed in the current study. Similar associations between high flavonoid content and antimicrobial activity have also been reported in Turkish propolis, where elevated levels of pinocembrin were linked to enhanced biological activity [29].

Agar overlay high‐performance thin‐layer chromatography (HPTLC) coupled with bio‐autography was employed to target and isolate bio‐active compounds from the South African propolis sample (GP68), which exhibited the best antimycobacterial activity (Table 1). Figure 2 presents the developed HPTLC plate of GP68 under UV radiation at 366 nm and white light reflectance, alongside the corresponding bio‐autogram against M. smegmatis. This bio‐autographic approach provided valuable insights into the contribution of individual chemical constituents to the overall antimycobacterial efficacy of propolis.

FIGURE 2.

The developed HPTLC plate of South African propolis is shown under UV 366 nm (A) radiation and white reflectance (B), along with a bio‐autogram against M. smegmatis (C).

The compounds that demonstrated inhibitory activity against M. smegmatis included pinobanksin (17; Rf = 0.40), chrysin (16; Rf = 0.42), galangin (19; Rf = 0.48), 3‐O‐acetylpinobanksin (21; Rf = 0.50), pinobanksin‐3‐O‐propionate (29; Rf = 0.54), pinocembrin (18; Rf = 0.55), pinobanksin‐3‐O‐butyrate (32; Rf = 0.56), pinobanksin‐3‐O‐pentanoate/2‐methylbutyrate (34; Rf = 0.58) and pinobanksin‐3‐O‐hexanoate (37; Rf = 0.60) (Table S2).

The bio‐autography results further confirmed that five compounds previously identified through the heatmap and VIP score analysis as abundant were also identified as being responsible for the antimycobacterial activity of the South African propolis sample. This supports the hypothesis that propolis efficacy arises from synergistic interactions among chemical constituents rather than the activity of a single compound [37]. This observation is consistent with a previous study demonstrating that combinations of compounds such as pinocembrin, chrysin and galangin exhibit enhanced antimicrobial activity compared to individual compounds alone [34]. Therefore, compounds 16 (chrysin), 17 (pinobanksin), 29 (pinobanksin‐3‐O‐propionate) and 32 (pinobanksin‐3‐O‐butyrate) present in the South African sample (GP68) may contribute to the overall antimycobacterial effect.

2.3. Molecular Docking Analysis

Based on the potent antimycobacterial effects observed for the South African propolis samples GP68 and EC77B against the various Mycobacterium strains, particularly MTB, a molecular docking analysis was carried out. This aimed to elucidate the potential molecular interactions and mechanisms responsible for the observed bio‐activity. Following a thorough literature evaluation of essential MTB proteins, three pivotal targets were chosen for investigation: DprE1, F₀F₁ ATP synthase and RpoB (RNA polymerase). These proteins play crucial and distinct roles in bacterial viability, governing cell wall biosynthesis, energy production and genetic transcription, respectively. The docking results for the nine flavonoid compounds identified from South African propolis against the three validated MTB targets suggest possible mechanisms of action based on their predicted binding affinities. In molecular docking studies, the docking score serves as an estimate of the binding energy between a ligand and its target protein [38, 39].

These scores are calculated based on scoring functions that approximate the thermodynamic stability of the ligand–protein complex by considering key interactions such as hydrogen bonding, hydrophobic contacts, electrostatic forces and van der Waals interactions. A more negative docking score generally indicates a stronger and more favourable predicted interaction, suggesting that the ligand may form a more stable complex with the target protein under physiological conditions [40, 41].

As illustrated in Table 3, pinocembrin exhibited the strongest binding affinity for DprE1 with a score of −9.2 kcal/mol, surpassing the reference DprE1 inhibitor from the BTZ series (−8.5 kcal/mol). Compounds such as chrysin (−8.2 kcal/mol) and pinobanksin‐3‐O‐butyrate (−8.1 kcal/mol) also displayed moderately favourable docking scores. These results indicate that the flavonoid scaffolds present in South African propolis can bind to the DprE1 binding pocket with predicted affinities that are comparable to established lead compounds. The biological relevance of targeting DprE1 is well supported by prior chemical validation through benzothiazinone inhibitors and structural studies, which have characterised it as a druggable, FAD‐containing enzyme essential for cell wall biosynthesis [42]. Further analysis of the binding mode for the top‐performing ligand, pinocembrin, reveals key stabilising interactions with several residues within the catalytic pocket, including Tyr60, His132, Gly117, Ala141 and Pro116 (Figure 3A). These interactions, which comprise conventional hydrogen bonds and π–alkyl contacts, are positioned near the FAD‐binding region critical for the enzyme's redox activity in arabinan synthesis. The favourable docking score is likely a consequence of this combination of specific hydrogen bonding and broader hydrophobic contacts. This finding is consistent with previous structural work demonstrating that potent DprE1 inhibitors frequently stabilise their binding through multiple hydrogen bonds and van der Waals interactions with key catalytic residues [42].

TABLE 3.

Molecular docking results of selected targets against active extracts from African propolis.

IDs	Compounds	DprE1	F0F1 ATP‐Synthase	RpoB‐RNAP
1	3‐O‐Acetylpinobanksin	−7.8	−6.2	−7.3
2	Chrysin	− 8.2	−6.4	−9.1
3	Galangin	−7.9	−6.1	−8.6
4	Pinobanksin‐3‐O‐Butyrate	−8.1	−6.0	−7.0
5	Pinobanksin‐3‐O‐Propionate	−7.5	−6.1	−7.1
6	Pinobanksin	−7.8	−6.2	−8.9
7	Pinocembrin	−9.2	−6.3	−9.0
8	Pinobanksin‐3‐O‐Hexanoate	−7.3	−5.2	−8.1
9	Pinobanksin‐3‐O‐Methylbutyrate	−7.5	−6.0	−9.0
10	Positive control	−8.5 (G1T)	−6.6 (BDQ)	−8.8 (RFP)

FIGURE 3.

Molecular docking interaction plots of selected African propolis flavonoids with Mtb targets. The 2D and 3D representations show key amino acid interactions of pinocembrin with DprE1, chrysin with RNA polymerase (RNAP) and chrysin with ATP synthase (AtpE).

Furthermore, the docking results for RNAP showed a similar pattern of strong binding for several of the flavonoids. The majority of tested compounds, including chrysin, pinocembrin, pinobanksin, galangin and pinobanksin‐3‐O‐methylbutyrate, displayed binding energies clustered between −8.6 and −9.1 kcal/mol. Primarily, chrysin (−9.1 kcal/mol), pinocembrin (−9.0 kcal/mol) and pinobanksin‐3‐O‐methylbutyrate (−9.0 kcal/mol) all had better scores than the rifampicin control (−8.8 kcal/mol). Since rifampicin is known to bind to the RpoB subunit of RNAP and block the RNA exit channel, compounds with similar or stronger binding scores may interact with the same or nearby regions of the enzyme. Also, the notably strong binding of the compounds against RpoB‐RNA polymerase is a significant finding, as RNAP is the target for first‐line TB treatment [43]. Additionally, the interaction plot of the top binding ligand, chrysin, to RNAP reveals hydrogen bonding with residues such as Arg469, Gln438 and Pro489, and hydrophobic contacts with Ile490 and Leu439, located within the β‐subunit binding cleft (Figure 3B). Chrysin also exhibited π–π stacking interactions with aromatic residues, suggesting that it fits well within the rifampicin‐binding region or nearby sub‐pockets. Since RNAP is essential for transcription, these binding interactions may disrupt RNA synthesis, similar to the mechanism of rifampicin [43]. The strong docking affinities observed here are consistent with previous reports of flavonoids interacting with nucleic acid‐processing enzymes through hydrogen bonding and aromatic stacking [44].

In contrast, none of the screened flavonoids matched or exceeded the binding energy of bedaquiline for the F₁F₀ ATP synthase target (bedaquiline control −6.6 kcal/mol); all compounds gave weaker predicted affinities (range −5.2 to −6.4), indicating that these propolis samples are less likely to inhibit AtpE at concentrations comparable to bedaquiline. The bedaquiline‐AtpE interaction is a well‐studied clinical example of targeting mycobacterial energy metabolism, and the comparatively low docking score against AtpE suggests that the propolis flavonoids’ primary modes of action are more likely to involve cell‐wall or transcriptional targets than the ATP synthase. The interaction maps show that the flavonoids formed only a few hydrophobic and π–π contacts with residues such as Ala66, Gly62 and Phe65, located along the transmembrane helices (Figure 3C). The limited number of hydrogen bonds and the mostly surface‐level contacts explain the less favourable docking energies (ranging from –5.2 to –6.4 kcal/mol). This weak interaction pattern supports the findings that these compounds are less likely to inhibit the energy‐generating F₁F₀ ATP synthase, wherein the hydrophobic binding pocket is known to favour bulky diarylquinoline structures like bedaquiline [45].

The docking results corroborate the experimental findings of the current study, which identified high antimycobacterial activity in propolis samples enriched with these flavonoid compounds. These findings are also consistent with previous reports showing that flavonoids, including chrysin, galangin and pinocembrin, possess broad antimicrobial activity through inhibition of key bacterial enzymes and membrane‐associated targets [34, 44, 46].

The favourable binding scores and predicted interaction patterns support the potential of these natural compounds to act as potential enzyme inhibitors in critical MTB pathways, particularly those involved in cell wall biosynthesis (DprE1) and transcription (RNAP). Also, the relatively low toxicity of the active samples further strengthens their potential as lead candidates for antimycobacterial drug development.

2.4. Toxicity

The brine shrimp lethality results for the eight active South African propolis samples are presented in Table 4, and the toxicity dose‐response curves are shown in Figure 4. After 24 h, none of the samples were toxic at 1 mg/mL, with average percentage mortality ranging from 0.62% to 30.63%. At 48 h, all samples except EC77B and EC77D remained non‐toxic at 1 mg/mL, with mortality ranging from 1.09% to 43.68%. Toxicity dose‐response curves indicated that both EC77B and EC77D were non‐toxic at 500 µg/mL (Figure 4). Similarly, the comparator Brazilian propolis extract (BR20A) was non‐toxic at 1 mg/mL, showing average mortalities of 13.17% at 24 h and 20.68% at 48 h.

TABLE 4.

The percentage mortality of active South African propolis at 24 and 48 h.

Samples	Percentage mortality
	24 h	48 h
GP10	1.59 a	43.68
GP11	8.75	34.11
GP28B	13.99	19.79
GP52	0.62	1.09
GP60	4.00	14.37
GP68	2.76	11.27
EC77B	30.63	89.75 b
EC77D	30.16	82.64
BR20A	13.17	20.68
Controls
Saltwater	0	0
DMSO (2% v/v)	0	0
Potassium dichromate	100	100

^a Percentage mortality less than 50% is considered non‐toxic

^b Percentage mortality above 50% is considered toxic (bold).

FIGURE 4.

The toxicity dose response of propolis samples EC77B and EC77D at 48 h.

These results are corroborated [22], who evaluated the toxicity of 51 South African propolis samples. All the samples showed no signs of toxicity (ranging from 0 to 33.09 at 24 h and 0 to 40.57% at 48 h). The low percentage mortality recorded in the brine shrimp lethality assay suggests that these propolis samples display antimycobacterial activity without inducing significant toxicity in eukaryotic cells.

3. Conclusion

This study represents the first comprehensive evaluation of the antimycobacterial activity of South African propolis against both NTM and a virulent strain of MTB. A total of 109 South African propolis samples were assessed, constituting the largest dataset of propolis tested for antimicrobial activity to date. The South African propolis extracts demonstrated notable antimycobacterial and anti‐tuberculosis effects, with some samples displaying better activity (GP68 displaying MIC ranging from 98–125 µg/mL against all Mycobacterium strains) compared to the Brazilian propolis, often considered the ‘gold standard’ in propolis research. The antimycobacterial activity of the South African propolis samples was associated with the presence of flavonoids such as pinocembrin, galangin, chrysin and various pinobanksin derivatives. The observed antimycobacterial activity was associated with high concentrations of flavonoids such as pinocembrin, galangin, chrysin and various pinobanksin derivatives, which were abundant in active samples compared to inactive ones. All active samples exhibited low toxicity in the brine shrimp lethality assay, with percentage mortality below 50% at 24 h and no toxicity observed at 500 µg/mL after 48 h. The molecular docking analysis revealed a consistent binding pattern by multiple compounds against key MTB targets, including DprE1 enzyme and RNAP, where some of the compounds rivalled or even surpassed the docking scores of the reference drugs. Collectively, these findings provide robust evidence supporting the therapeutic potential of selected South African propolis samples as antimycobacterial agents. Moreover, the study highlights a clear correlation between the antimycobacterial activity, chemical composition and safety profile of South African propolis. Future research requires investigation of the synergistic interactions between propolis and conventional anti‐tuberculosis drugs to enhance therapeutic efficacy. Additionally, the antibiofilm and anti‐quorum‐sensing properties of South African propolis should be investigated to determine the capacity to inhibit Mycobacterium biofilm formation, disrupt established biofilms and interfere with mycobacterial communication pathways.

4. Materials and Methods

4.1. Sample Collection

In co‐operation with the South African Bee Industry Organisation (SABIO) and South African apiarists, 109 propolis samples (Figure 5) were collected from eight provinces in South Africa. To the best of our knowledge, this is currently the largest propolis sample size ever studied globally for antimicrobial activity. Brazilian propolis has gained global commercial importance due to its favourable biological attributes and health benefits. Hence, for comparative purposes, three Brazilian propolis samples were obtained from Lagos, Brazil (collected by apiarist Mr Kai Hichert) and added to the collection.

FIGURE 5.

The geographical distribution of 109 South African propolis samples used in this study.

4.2. Sample Preparation

Ground propolis samples were submerged in ethanol at a ratio of 1:10 (w/v) and incubated in the orbital shaker incubator at 37°C for 7 days. Thereafter, the excess ethanol was evaporated, and the extract was air‐dried under a fume hood [22, 24].

The dried ethanolic extracts were resubmerged in acetone to a starting concentration of 25 mg/mL for antimycobacterial screening. For chemical profiling, 1 mg of the extract was submerged in 1 mL of ethanol and the solution was incubated for 24 h on an orbital shaker and sonicated for 10 min. The final suspension was filtered using a 0.22 µL Clarinert syringe filter.

4.3. Culture Preparation

Mycobacterium fortuitum (ATCC 6841), M. smegmatis (ATCC 14468) and M. gordonae (ATCC 14470) were cultured in 7H9 Middlebrook broth containing glycerol and oleic albumin dextrose catalase (OADC) growth supplement. Mycobacterium fortuitum and M. smegmatis were incubated for approximately four days, while M. gordonae was incubated in an orbital shaker for approximately 10 to 15 days at 37°C. Post‐incubation, bacterial cultures were standardised by diluting to 1:100 in OADC‐supplemented Middlebrook broth to obtain the 0.5 McFarland turbidity standard. An ethics waiver to use these micro‐organisms was approved by the University of the Witwatersrand Human Research Ethics Committee (reference no. W‐PR‐240118‐01).

Mycobacterium tuberculosis H37Rv was grown in 10 mL of OADC‐supplemented Middlebrook 7H9 at 37°C for 3‐4 days to an OD_600nm of between 0.5 and 0.8. The MTB culture was standardised by diluting to 1:500 in OADC‐supplemented Middlebrook 7H9 broth. All procedures with MTB were conducted in a BioSafety Level III laboratory, registered with the South African Department of Agriculture, Forestry and Fisheries (registration number: 39.2/NHLS‐20/010), and were approved by the Institutional Biosafety Committee of the University of the Witwatersrand (approval number: 20200502).

4.4. Microdilution Assay

4.4.1. Non‐Tuberculous Mycobacterium (NTM)

The minimum inhibitory concentration (MIC) of the propolis samples against the three NTM strains was determined using the modified broth microdilution assay [24]. Middlebrook broth (100 µL) was aseptically added to each well of the 96‐microtiter plates. Thereafter, 100 µL of each propolis ethanolic extract was added in duplicate in the first row, followed by a two‐fold serial dilution. Ciprofloxacin (0.01 mg/mL) and rifampicin (1.0 mg/mL) were used as positive controls since they form part of the antimicrobials used to treat NTM [8]. Acetone was used as the diluent control since it was used to resuspend the propolis extracts. A standardised Mycobacterium culture (100 µL of 0.5 McFarland turbidity standard) was added to the wells. The microtiter plates were hermetically sealed with a sterile adhesive film and incubated at 37°C for 4 days (M. smegmatis and M. fortuitum) or 10 days for M. gordonae. Post‐incubation, 40 µL of p‐iodonitrotetrazolium violet (INT) at a concentration of 0.4 mg/mL was added to each well as a growth indicator. The change in colour to purple‐pink indicated microbial growth. The MIC value was confirmed by the absence of colour change and was defined as the lowest propolis concentration that prevented growth of the bacterial cells. The plates were read when the positive control wells started to change colour. All assays were performed in triplicate on consecutive days to ensure reproducibility.

4.4.2. Mycobacterium tuberculosis H37Rv

The extracts were diluted to 1 mg/mL in media for use in the broth microdilution assay. An amount of 50 µL of OADC‐supplemented 7H9 Middlebrook broth was added to each well in a 96‐well microtitre plate. In the first row, 100 µL of propolis extract was added to the first well in duplicate, and a two‐fold serial dilution was performed. The standardised MTB culture (50 µL) was added to all the wells. As a positive control, rifampicin, isoniazid and streptomycin were included at starting concentrations of 3.2, 6.4 and 100 µg/mL, respectively. Acetone at a concentration of 1 mg/mL was used as the negative control. The inoculated plates were incubated at 37°C in a CO₂ incubator for 14 days. Growth was scored visually using an inverted mirror on day 7 and 14. Post‐incubation, 10 µL of Alamar Blue was added to each well. Bacterial growth was indicated by a colour change from blue (no growth) to pink (growth). The MIC value was confirmed as the propolis concentration that showed no blue‐to‐pink colour change.

4.5. Ultra Performance Liquid Chromatography—Mass Spectrometry (UPLC–MS) and Multivariate Statistical Analysis

The South African propolis samples that demonstrated the best antimicrobial activity against the Mycobacterium species, together with eight samples that showed weak activity (for comparative purposes), were selected for chemical analysis using UPLC–MS. The analysis was performed on a Waters Acquity I Class Ultra Performance Liquid Chromatographic system with a PDA detector (Waters, Milford, MA, USA). The separation was achieved on an Acquity UPLC BEH C18 column (150 mm × 2.1 mm i.d., 1.7 µm particle size, Waters) maintained at 40°C. The mobile phase consisted of 0.1% formic acid in water (solvent A) and 0.1% formic acid in acetonitrile (solvent B) at a flow rate of 0.35 mL/min. The gradient elution was as follows: 90% A: 10% B, changing to 60% A: 40% B in 2.5 min, changing to 55% A: 45% B in 8.5 min, changing to 20% A: 80% B in 1.5 min, keeping for 0.5 min, back to initial ratio in 0.5 min and equilibrating the system for 1.5 min. The total run time was 15 min. The samples were injected into the mobile phase with an injection volume of 1.0 µL (full‐loop injection). The data was collected by chromatographic software, Masslynx 4.2. The mass spectrometry (G3QToF, Waters) was operated in negative ion electrospray mode. Nitrogen was used as the desolvation gas. The desolvation temperature was set to 350°C at a flow rate of 550 L/hr, and the source temperature was 100°C. The capillary and cone voltages were set to 2500 and 40 V, respectively. Data was collected between 100 and 1200 m/z. During acquisition, the LockSpray interface was used to ensure mass accuracy and reproducibility. A heatmap and a variable importance projection (VIP) plot were generated using MetaboAnalyst (https://www.metaboanalyst.ca/), where the peak intensities were obtained from the UPLC‐MS‐generated data, normalised and subjected to Pareto scaling. The parameters were as follows: autoscale sample features for standardisation, Euclidean for distance measure, ward for clustering method and partial least squares discriminant analysis (PLS‐DA) as the statistical analysis.

4.6. Agar Overlay High‐Performance Thin‐Layer (HPTLC) Chromatography Bio‐Autography

The HPTLC‐agar overlay bio‐autography was used to perform target‐directed isolation of active compounds from the South African propolis samples that showed the best antimycobacterial activity against all the Mycobacterium species. The analysis was carried out using a semi‐automated HPTLC system (CAMAG, Muttenz, Switzerland) consisting of an automatic sampler 4, a chromatogram immersion device, an automatic developing chamber (ADC 2), a TLC plate heater and a documentation device (visionCATS) used for operating parameters and data analysis. A volume of 2 µL (50 mg/mL) of the propolis extract was applied on pre‐coated 60 F254 glass plates with 20 × 10 cm dimensions as a 10 mm band, and it was applied 35 mm from the bottom of the edge and 25 mm from the side edge with a delivery speed of 150 nL/s using an autosampler fitted with a 25 µL Hamilton syringe. The plate was exposed to air for 3 min before development. The development was calibrated by conditioning the layer to 33% relative humidity (RH) through the use of a saturated aqueous magnesium chloride solution for 10 min. The plates were developed to a migration distance of 80 mm in a chamber pre‐saturated for 20 min using toluene: ethyl acetate: methanol (78:22:5 v/v/v). The plates were dried for 5 min and visualised under white reflectance at 254 and 366 nm radiation.

The HPTLC plates and samples were sterilised for 30 min under UV light in the laminar flow unit. Thereafter, 100 mL of OADC‐supplemented 7H10 Middlebrook agar was poured into a 25 × 25 cm sterile tray and allowed to solidify. Thereafter, the HPTLC plates were covered with 100 mL of agar inoculated with M. smegmatis while it was still liquid (∼40°C). The agar was allowed to solidify, and the Petri dishes were stored at −4°C for 30 min. Thereafter, they were incubated for 4 days at 37°C. After incubation, the bio‐autographs were sprayed with 0.4 mg/ml of INT to visualise inhibition zones. The plates were further incubated for 4–6 h at room temperature until a red colour developed to indicate microbial growth. The zones were measured and correlated with the spots observed on the reference HPTLC plate under UV/visible light. Images of all the bio‐autography plates were captured under 366 nm UV radiation and white reflectance.

To identify the compounds in the zone of inhibition, HPTLC analysis was carried out as previously described. Silica gel from bands sharing the same Rf values was carefully scraped off the plates using a spatula, and the collected material was dissolved in methanol. These solutions were filtered through 0.2 µm syringe filters and analysed by UPLC–MS to determine the chemical composition of each band. Compound identification was based on comparisons of elemental composition, accurate mass and mass spectral fingerprints with information available in scientific literature and databases such as the Mass Bank of North America (https://mona.fiehnlab.ucdavis.edu) and PubChem (https://pubchem.ncbi.nlm.nih.gov).

4.7. In‐Silico Molecular Docking

The three‐dimensional (3D) x‐ray crystal structures of selected M. tuberculosis targets; decaprenylphosphoryl‐β‐D‐ribofuranose 2′‐oxidase (DprE1), F₀F₁ ATP synthase (AtpE subunit) and the RNA polymerase β‐subunit (RpoB)—were obtained from the Protein Data Bank (PDB) with the accession IDs 6HFW [47], 4V1F [48] and 5UHG [49], respectively. Each protein structure was prepared for molecular docking using UCSF Chimaera version 1.12 [50]. Protein preparation followed a standard protocol to ensure accurate docking and minimise computational complexity. All non‐essential components, such as water molecules, co‐crystallised ligands and non‐standard amino acid residues, were removed to simplify the system. Missing hydrogen atoms were added to ensure accurate valency and partial atomic charges were assigned using the Gasteiger charge parameter. The AutoDock graphical user interface provided by MGL tools was utilised to define the AutoDock atom types. The resulting optimised protein structures were then saved in PDBQT format for docking simulations. This preparation ensured that each receptor was properly orientated and energetically minimised for reliable molecular docking analysis.

Also, the 3D structures of the active compounds; pinocembrin, 3‐O‐acetylpinobanksin, galangin, pinobanksin‐3‐O‐methylbutyrate, pinobanksin‐3‐O‐hexanoate, chrysin, pinobanksin, pinobanksin‐3‐O‐propionate, pinobanksin‐3‐O‐butyrate and positive controls G1T, bedaquiline and rifampicin, were retrieved from the PubChem database. Each structure was then imported into Avogadro 2.0 software [51] for geometry optimisation and energy minimisation. The minimisation process was performed using the MMFF94 force field to achieve the most stable conformations with minimised potential energy, ensuring that the ligand structures were in their optimal forms prior to molecular docking.

Subsequently, molecular docking was performed using AutoDock Vina software [41] to predict the binding affinities and interaction modes between the prepared ligands and target proteins. The docking simulations were carried out within the defined active sites of each target, and the results were evaluated based on the binding energy scores and the nature of ligand–protein interactions.

4.8. Toxicity Study

The brine shrimp lethality assay [52], with modifications, was used. Brine shrimp eggs (Artemia franciscana) (0.5 g) were hatched in 16 g/mL salt water, exposed to light and incubated at ambient temperature for 24–48 h. All the propolis samples were prepared to a concentration of 2 mg/mL using 2% (v/v) dimethyl sulfoxide (DMSO). Subsequently, 400 µL of the saltwater containing brine shrimp was added to each well of a 48‐well microtiter plate, followed by 400 µL of the propolis samples, which were added in triplicate. Saltwater with a concentration of 16 g/l and 2% DMSO was used as a negative control, and 1.6 mg/mL potassium dichromate was used as the positive control. The shrimp were observed under a light microscope to record the initial number of dead shrimp. The percentage mortality was calculated after 24 and 48 h using Equation 1, and a mortality rate above 50% was considered toxic [53].

$$\begin{array}{ccc}\hfill \%\mathrm{Mortality}& =& \frac{\begin{array}{c}\mathrm{Dead}\mathrm{shrimp}\mathrm{at}48\mathrm{h}\left(\mathrm{before}\mathrm{acetic}\mathrm{acid}\right)\hfill \\ -\mathrm{Dead}\mathrm{shrimp}\left(\mathrm{time}0\right)\hfill \end{array}}{\mathrm{Total}\mathrm{number}\mathrm{of}\mathrm{dead}\mathrm{shrimp}\left(\mathrm{after}\mathrm{acetic}\mathrm{acid}\right)}\hfill \\ & & \times 100\hfill \end{array}$$ (1)

Samples that displayed toxicity were tested at variable concentrations of 1, 0.5, 0.25, 0.125 and 63 mg/mL to determine the concentrations at which the extracts are non‐toxic. Each dilution was tested in triplicate in two separate experiments. Positive and negative controls were also included. The results were presented as a dose‐response curve.

Author Contributions

Nyiko F. Mzimba: investigation, writing – original draft, methodology, validation, visualisation, writing – review and editing, formal analysis and data curation. Efficient N. Ncube: investigation, writing – original draft, methodology, validation, visualisation, writing – review and editing and formal analysis. Elliasu Y. Salifu: investigation, writing – original draft, methodology, validation, visualisation, writing – review and editing and formal analysis. Alvaro Viljoen: conceptualisation, investigation, funding acquisition, writing – review and editing, project administration, supervision and resources. Bhavna Gordhan: investigation, writing – original draft, methodology, validation, visualisation, writing – review and editing, formal analysis, data curation and supervision. Bavesh Kana: funding acquisition, writing – review and editing, project administration, data curation, supervision and resources. Sandy van Vuuren: conceptualisation, funding acquisition, writing – review and editing, project administration, supervision and resources.

Funding

This study was supported by the National Research Foundation, the South African Medical Research Council, the Faculty of Research Committee Individual Research Grant, the Oppenheimer Memorial Trust and the University of the Witwatersrand Postgraduate Scholarship.

Conflicts of Interest

The authors declare no conflicts of interest.

Supporting information

Supporting File 1: cbdv71577‐sup‐0001‐SupMat.docx.

Data Availability Statement

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