HPLC-DAD and UHPLC/QTOF-MS Analysis of Polyphenols in Extracts of the African Species Combretum padoides, C. zeyheri and C. psidioides Related to Their Antimycobacterial Activity

Combretum padoides Engl. & Diels, C. psidioides Welv. and C. zeyheri Sond. are used for the treatment of infections and tuberculosis related symptoms in African traditional medicine. In order to verify these uses, extracts were screened for their growth inhibitory effects against M. smegmatis ATCC 14468. Ultra-high pressure liquid chromatography coupled to quadrupole time-of-flight mass spectrometry (UHPLC/QTOF-MS) and GC-MS were used to investigate the polyphenolic composition in the active extracts. The lowest minimum inhibitory concentration (MIC), 625 µg/mL, was shown by a methanol extract of the stem bark of C. psidioides. A butanol extract of C. psidioides gave large inhibition zone diameters (IZD 21 mm) and inhibited 84% of the mycobacterial growth at 312 µg/mL. Combretastatin B-2 and dihydrostilbene derivatives were present in the methanol extract of C. psidioides, whereas the butanol extract of this species contained punicalagin, corilagin, and sanguiin H-4. Methanol and butanol extracts of the stem bark of C. padoides gave large inhibition zone diameters (IZD 26.5 mm) and MIC values of 1250 and 2500 µg/mL, respectively. C. padoides contained an ellagitannin with a mass identical to punicalagin ([M-H]− 1083.0587) and a corilagin like derivative ([M-H]− 633.0750) as well as ellagic acid arabinoside and methyl ellagic acid xyloside. A butanol extract of the roots of C. zeyheri showed mild antimycobacterial activity and contained a gallotannin at m/z [M-H]− 647.0894 as the main compound along with punicalagin and three unknown ellagitannins at m/z [M-H]− 763.0788, 765.0566, and 817.4212. Our results indicate that the studied species of Combretum contain phenolic and polyphenolic compounds with possible potential as leads for antimycobacterial drugs or as adjuvants for conventional anti-TB drugs.


Introduction
Tuberculosis (TB) continues to be of global concern [1]. Worldwide, TB is among the top ten causes of death, and according to World Health Organization [1], 1.2 million people died from TB in 2018. Developing countries have significantly higher incidents of TB due to malnutrition, poverty, and crowded settings [2]. In the African region, the incidence rate of TB is estimated to be 363 per 100,000 compared to 4.4 per 100,000 in the United States [3,4]. Moreover, up to 22.4% of the population Table 1. Summary of applications in traditional medicine and antimicrobial activity as well as identified compounds of the species of Combretum used in this study. Leaves and roots: For treatment of snakebites, wounds, hookworms, bloody diarrhea, malaria and conjunctivitis [30,37,41].
Decoctions of roots, leaf extracts or leaves mixed with maize porridge (Ugali) for treatment of diarrhea and oedema [78].
Broad-spectrum antibacterial profile in our earlier investigation [78] as well as some antifungal effects [79].
Smoke of burnt leaves inhaled for cough (TB?), water extracts of dried leaves for colic, crushed leaves for rheumatism and joint pain [35]; hot water decoctions of roots for diarrhea, dysentery and ankylostomiasis [37]; Pounded roots cooked in porridge for hookworms and dysentery, ground roots cooked and applied to wounds, root decoctions for stomach-ache, cough, pneumonia, vomiting, stomach ulcers and diarrhea, leaf infusions for cough, stem bark infusion for leprosy [80]. Roots, leaves and stem bark made into decoctions or mixed in maize porridge for diarrhea and stomach tumors [78].
Stem bark and leaf extracts inhibitory against several bacteria [81,82]; Fruit, stem bark and root extracts show good antibacterial potential. Triterpenoids from leaves evaluated for anti-Candida effects; terminolic acid was found to be the most active compound. SAR: oleanane and ursane type triterpenoids were the most active ones [54]. We found that extracts of stem bark and roots inhibit the growth of Mycobacterium smegmatis, the BuOH fraction of the roots being especially active (IZ 23 mm).* Triterpenoids and saponins from the leaves [47]; Three unidentified antimicrobial compounds were isolated from the leaves and stem bark [82]; Ursolic acid, maslinic acid, 2α,3β-dihydroxyurs-12-en-28-oic acid, 6β-hydroxymaslinic acid and terminolic acid from leaves [54]. A root butanol extract contained six ellagic acid derivatives including methyl-ellagic acid xyloside, di-methyl-ellagic acid xyloside and 3,3 -Di-O-methyl-4-O-(n -O-galloyl-β-d-xylopyranosyl) ellagic acid, fifteen ellagitannins including punicalagin and nine gallotannins including hexagalloylglucose.* Our results presented in this paper *; AI, activity index in % activity of rifampicin; MIC, minimum inhibitory concentration; IZD, inhibition zone diameter in mm; TB, tuberculosis; TM, traditional medicine. In brackets the voucher numbers of the collected specimen used in this study. Plant uses for symptoms related to TB marked with bold text.

Antimycobacterial Effects of Extracts and Fractions
Altogether, 26 extracts and solvent partition fractions from Combretum psidioides, C. padoides and C. zeyheri, used for the treatment of bacterial infections and even cough (C. zeyheri) in African traditional medicine (Table 1), were screened for their growth inhibitory effects against Mycobacterium smegmatis ATCC 14468. M. smegmatis has been found to be rather resistant to rifampicin [83]. Therefore, inhibitory effects against this bacterial strain could indicate that the extracts and fractions also to a certain extent might inhibit rifampicin resistant strains of Mycobacterium tuberculosis. Moreover, M. smegmatis has some virulence genes in common with M. tuberculosis, and thus serves as a good model bacterium [84]. The results regarding the antimycobacterial activity of crude methanol extracts and their solvent partition fractions of the investigated species of Combretum spp. are shown in Tables 2 and 3.  In brackets the mean percentage of the bacterial growth inhibited by indicated concentration and resulting in no visible growth (IC, inhibitory concentration). * In butanol extract of C. psidioides; ** Not present in the extracts as such but as ellagic acid glycosidic derivatives. The total antimycobacterial activity (mL/g) is and indication on the degree to which 1 g of an extract can be diluted without losing its antimycobacterial activity and is calculated as: extraction yield in mg extracted material/mg plant powder divided to the MIC of the extract [83].
Combretum psidioides gave the best growth inhibitory effects against M. smegmatis of the three species of Combretum used in this investigation. The lowest MIC of 625 µg/mL was shown by a methanol extract of the stem bark of this species. This extract also gave a large inhibition zone diameter of 29.0 mm (Tables 2 and 3) as well as a good total antimycobacterial activity of 313.44 mL/g ( Table 3). The total antimycobacterial activity is dependent on the extraction yield and the MIC of a plant extract, and is calculated as the ratio between the yield (in mg/g) and the MIC (in mg/mL) [84]. Moreover, we also found that chloroform and butanol extracts of the stem bark of C. psidioides were growth inhibitory against M. smegmatis, both giving MIC values of 2500 µg/mL and inhibition zone diameters of 25.5 and 21.5 mm, respectively (Tables 2 and 3). In addition, although the MIC of the butanol fraction of C. psidioides was 2500 µg/mL, this extract was the most effective of the investigated extracts and fractions to inhibit the growth of M. smegmatis at lower concentrations and at 312 µg/mL, this extract resulted in an 84% growth inhibition.
Antimycobacterial effects were observed for methanol and butanol extracts of the stem bark of C. padoides, and both showed inhibition zone diameters of 26.5 mm and an activity index of 0.50 in comparison to rifampicin (IZD 52.5 mm) ( Table 2). The MIC value of the crude methanol extract was 1250 µg/mL and for the butanol extract 2500 µg/mL (Table 3) and the growth inhibition of both extracts was dose-dependent for concentrations ranging down to 312 µg/mL ( Figure 1). For both the crude methanol and butanol extracts, the growth inhibition was stronger at 78 µg/mL when compared to 156 µg/mL (Figure 1). Although the MIC of the butanol extract was higher than the MIC for the methanol extract, the butanol extract gave stronger growth inhibition than the methanol extract at concentrations below 625 µg/mL, and at 78 µg/mL still 51% of the growth was inhibited ( Figure 1).
Compared to C. padoides and C. psidioides, C. zeyheri gave only mild antimycobacterial effects and a root butanol extract gave the best effect (IZD 23.0 mm) ( Table 2).
In comparison to the plant extracts, rifampicin gave an inhibition zone diameter of 52.5 mm and a MIC of 3.9 µg/mL ( Table 2, Table 3, Figure 1

Combretum Psidioides
Based on the growth inhibitory results of both the methanol and the butanol extracts of the stem bark of C. psidioides, a butanol extract was chosen for phytochemical analysis on its polyphenolic composition using HPLC-DAD and UHPLC/QTOF-MS. Although showing a higher MIC than the methanol extract, the butanol extract was chosen for this analysis since the methanol extract contained a high concentration of condensed tannins that could interfere with the analysis ( Figure  S1). However, in terms of the ellagitannins and the ellagic acid derivatives composition, the butanol and methanol extracts contained the same compounds (Supplementary Figure S1). The results of the HPLC-DAD and UHPLC/QTOF-MS analysis of the butanol extract of the stem bark of C. psidioides are shown in Table 4 Figure 3a,b). The thirteen unknown ellagitannins could be identified as ellagitannins based on their UVλ absorption spectra, demonstrating three absorption maxima, characteristic for ellagitannins (Table 4, Figure 3c,d). Epigallocatechin gallate (MW 458.0843) at Rt 9.43 min was tentatively identified according to computer compound libraries available and characteristics such as UVλ absorption maxima data as well as HPLC retention time (Table 4, Figure 3e). To the best of our knowledge, this tea polyphenol has not been reported before in the genus Combretum.
Stilbenoid compounds are known to be present in Combretum spp. [61][62][63][64][65][66][67][68]. Stilbenes in the methanol extract of the stem bark of C. psidioides could contribute to the antimycobacterial effects of this extract (MIC 625 µg/mL). Therefore, a GC-MS analysis was made to study the stilbenoid composition of this extract. This analysis resulted in the characterization of combretastatin B-2 (exact calculated mass for C17H20O5, MW 304.13107) and its dihydrostilbene derivatives ( Figure 4, Table 4). To the best of our knowledge, combretastatin B-2 has not been described before in C. psidioides,

Combretum Psidioides
Based on the growth inhibitory results of both the methanol and the butanol extracts of the stem bark of C. psidioides, a butanol extract was chosen for phytochemical analysis on its polyphenolic composition using HPLC-DAD and UHPLC/QTOF-MS. Although showing a higher MIC than the methanol extract, the butanol extract was chosen for this analysis since the methanol extract contained a high concentration of condensed tannins that could interfere with the analysis ( Figure S1). However, in terms of the ellagitannins and the ellagic acid derivatives composition, the butanol and methanol extracts contained the same compounds (Supplementary Figure S1). The results of the HPLC-DAD and UHPLC/QTOF-MS analysis of the butanol extract of the stem bark of C. psidioides are shown in Table 4 Figure 3a,b). The thirteen unknown ellagitannins could be identified as ellagitannins based on their UVλ absorption spectra, demonstrating three absorption maxima, characteristic for ellagitannins (Table 4, Figure 3c,d). Epigallocatechin gallate (MW 458.0843) at Rt 9.43 min was tentatively identified according to computer compound libraries available and characteristics such as UVλ absorption maxima data as well as HPLC retention time (Table 4, Figure 3e). To the best of our knowledge, this tea polyphenol has not been reported before in the genus Combretum.      Stilbenoid compounds are known to be present in Combretum spp. [61][62][63][64][65][66][67][68]. Stilbenes in the methanol extract of the stem bark of C. psidioides could contribute to the antimycobacterial effects of this extract (MIC 625 µg/mL). Therefore, a GC-MS analysis was made to study the stilbenoid composition of this extract. This analysis resulted in the characterization of combretastatin B-2 (exact calculated mass for C 17 H 20 O 5 , MW 304.13107) and its dihydrostilbene derivatives ( Figure 4, Table 4). To the best of our knowledge, combretastatin B-2 has not been described before in C. psidioides, although a number of other bibenzyls were characterized from the heartwood of this plant by Letcher & Nhamo [63].

Combretum Padoides
Since the HPLC-DAD chromatograms of the methanol and butanol extracts of the stembark of C. padoides were almost identical, but with the butanol part showing a slightly cleaner profile ( Figure  S2), we chose the butanol part for UHPLC/QTOF-MS analysis. In addition, although the MIC was lower for the methanol extract, the IC50 value was lower (<312 µg/mL) for the butanol part, which was also a criterium for choosing this extract for the phytochemical analysis. The results of this analysis are shown in Table 5. Altogether twenty-six ellagitannins were tentatively identified based on their retention times, mass spectrometric data and characteristic UV absorbance spectra ( Figure  5). The ellagitannin present in the highest concentration, at Rt 10.90 min, resembled punicalagin, giving a [M-H] − molecular ion at m/z 1083.0587. The retention time of this ellagitannin (10.90 min) was, however, not in agreement with the retention time that we have found before for β-punicalagin (Rt 14.95-15.64 min). Thus, this ellagitannin could be punicacortein D, with the molecular formula C48H28O30, identical to punicalagin, but with a shorter retention time in HPLC-DAD, compared to punicalagin [86]. Punicacortein D has been found in only one other species of the family Combretaceae, Combretum aculeatum [86]. However, there is also a possibility that this ellagitannin at Rt 10.90 min would be the α-anomer of punicalagin, which is otherwise identical to β-punicalagin to its mass spectrometric data, but has a shorter HPLC-retention time. Moreover, the α-and β-anomers of punicalagin differ by slight shifts in their UVλ absorbance maxima, due to differences in the position of the OH-group and the H-atom linked to the anomeric C-1 atom [86,87]. However, in our analysis, it was impossible to distinguish α-punicalagin from punicacortein D.

Combretum Padoides
Since the HPLC-DAD chromatograms of the methanol and butanol extracts of the stembark of C. padoides were almost identical, but with the butanol part showing a slightly cleaner profile ( Figure S2), we chose the butanol part for UHPLC/QTOF-MS analysis. In addition, although the MIC was lower for the methanol extract, the IC 50 value was lower (<312 µg/mL) for the butanol part, which was also a criterium for choosing this extract for the phytochemical analysis. The results of this analysis are shown in Table 5. Altogether twenty-six ellagitannins were tentatively identified based on their retention times, mass spectrometric data and characteristic UV absorbance spectra ( Figure 5). The ellagitannin present in the highest concentration, at Rt 10.90 min, resembled punicalagin, giving a [M-H] − molecular ion at m/z 1083.0587. The retention time of this ellagitannin (10.90 min) was, however, not in agreement with the retention time that we have found before for β-punicalagin (Rt 14.95-15.64 min). Thus, this ellagitannin could be punicacortein D, with the molecular formula C 48 H 28 O 30 , identical to punicalagin, but with a shorter retention time in HPLC-DAD, compared to punicalagin [86]. Punicacortein D has been found in only one other species of the family Combretaceae, Combretum aculeatum [86]. However, there is also a possibility that this ellagitannin at Rt 10.90 min would be the α-anomer of punicalagin, which is otherwise identical to β-punicalagin to its mass spectrometric data, but has a shorter HPLC-retention time. Moreover, the αand βanomers of punicalagin differ by slight shifts in their UVλ absorbance maxima, due to differences in the position of the OH-group and the H-atom linked to the anomeric C-1 atom [86,87]. However, in our analysis, it was impossible to distinguish α-punicalagin from punicacortein D.   (Table 5), [88]. The retention time of this ellagitannin however was different from that of corilagin, which we have found at Rt 12.52 min in Terminalia laxiflora roots [89] as well as in this investigation in Combretum psidioides stem bark at Rt HPLC-DAD 12.24 min (Table 4). Thus, this ellagitannin is tentatively suggested to be an isomer of corilagin. In addition to the mentioned ellagitannins, we identified the masses of four unknown ellagitannins; an ellagitannin at Rt 7.  (Table 5). When conducting a literature search for possible ellagitannins with molecular masses of 1086 occurring in Combretum spp., we did not find any matches. Instead, we found a reference of an ellagitannin giving a [M-H] − molecular ion at m/z 1085, found in strawberry fruits [90]. The molecular structure of this ellagitannin was not elucidated, however.
We   Table 5). Glucogallin is known from Quercus spp. [91] and from Emblica officinalis [92], but to the best of our knowledge not from Combretum spp.  (Table 5), [88]. The retention time of this ellagitannin however was different from that of corilagin, which we have found at Rt 12.52 min in Terminalia laxiflora roots [89] as well as in this investigation in Combretum psidioides stem bark at Rt HPLC-DAD 12.24 min (Table 4). Thus, this ellagitannin is tentatively suggested to be an isomer of corilagin. In addition to the mentioned ellagitannins, we identified the masses of four unknown ellagitannins; an ellagitannin at Rt 7.  (Table 5). When conducting a literature search for possible ellagitannins with molecular masses of 1086 occurring in Combretum spp., we did not find any matches. Instead, we found a reference of an ellagitannin giving a [M-H] − molecular ion at m/z 1085, found in strawberry fruits [90]. The molecular structure of this ellagitannin was not elucidated, however.
We  Table 5). Glucogallin is known from Quercus spp. [91] and from Emblica officinalis [92], but to the best of our knowledge not from Combretum spp.

Combretum Zeyheri
Since the root part of C. zeyheri gave slightly better growth inhibitory effects than the stem bark, this part was chosen for a phytochemical analysis on its polyphenols. Moreover, C. zeyheri has not been analysed before for its polyphenolic composition. Our UHPLC/QTOF-MS analysis of the butanol fraction of the roots demonstrated that a gallotannin at Rt 15.008 min, showing a [M-H] − molecular ion at m/z 765.0566 was the main compound in this extract, giving a peak area of 20.08% (at UV 280 nm, Table 6, Figure 2g). C. zeyheri contained more gallotannins and condensed tannins than C. psidioides and C. padoides. Altogether eight gallotannins and three procyanidins were found based on the UVλ absorption maxima spectra of these compounds. UVλ absorption maxima peaks with two shoulders, one at 216 nm and another at 258-278 nm, could be seen for the gallotannins (Figure 2g

Antimycobacterial Effects of the Extracts of the Studied Species of Combretum in Relation to Other Studies on the Antimycobacterial Effects of Combretum spp.
A number of studies demonstrate that extracts of Combretum species possess antimycobacterial effects. For example, Combretum comosum was found to give growth inhibitory effects against M. phlei [93]; C. brassii gave antimycobacterial effects against M. tuberculosis (MIC 1250 µg/mL) [94]; extracts from C. platypetalum and C. imberbe gave MIC values of 63-500 µg/mL against M. smegmatis and M. aurum [95]; acetone extracts of the leaves of C. hereroense gave a MIC value of 470 µg/mL against M. smegmatis [96]; acetone extracts of the leaves of C. schumannii gave a MIC value of 313 µg/mL against M. madagascariense and stem bark dichloromethane and root ethanol extracts were effective against M. indicuspranii [28]; C. hartmannianum leaf ethanol extracts showed strong growth inhibition against M. aurum A+ with a MIC value of 190 µg/mL [97] and ethanol extracts of the leaves, bark and root of C. kraussii gave MIC values of 195 µg/mL against M. aurum A+ [98,99].
In our investigation, methanol and butanol extracts of Combretum psidioides, C. padoides, and C. zeyheri were found to give MIC values from 625 to 2500 µg/mL against M. smegmatis. To the best of our knowledge, C. padoides and C. psidioides have not been studied before for their growth inhibitory effects against M. smegmatis. In addition, the roots of C. zeyheri have not been explored before for their antimycobacterial effects.
Contrary to our result, Luo et al. [100] reported that stem bark extracts of C. zeyheri are not antimycobacterial against M. smegmatis and M. tuberculosis. However, Luo et al. [100] defined all extracts as not active with MIC values exceeding 125 µg/mL and the MIC results for C. zeyheri were not shown. Moreover, Magwenzi et al. [95] found that a leaf extract of C. zeyheri was not active against Mycobacterium smegmatis. This could be due to differences in growth inhibitory activities between different organs in C. zeyheri, so that the stem bark and roots might contain more and perhaps different active compounds than the leaves, since we have seen that especially butanol and methanol extracts of the stem bark and roots of C. zeyheri give growth inhibitory effects against M. smegmatis. However, an alkaloid enriched leaf extract of C. zeyheri was found to inhibit the growth of M. smegmatis with a MIC value of 125 µg/mL and the growth inhibition was concentration-and time dependent [101]. Moreover, the same authors also discovered that this alkaloid extract of C. zeyheri was a potent inhibitor of efflux pumps in M. smegmatis. Thus, C. zeyheri (and other African Combretum spp.) might contain a number of alkaloids with antimycobacterial properties yet to be explored.

Ellagitannins in the Species of Combretum and Their Suggested Impact on the Antimycobacterial Effects of These Species
Although some Combretum spp. have been found to be rich in ellagitannins and their derivatives as well as other hydrolysable tannins [13,74,86] (Figure 6), these compound categories and their possible potential role as antimicrobials have not been studied in detail in this genus. This could partly be due to the poor bioavailability of ellagitannins, although they could have potential as topically administered antimicrobials or in the gut as well as in combinations with conventional antibiotics or via their metabolites, the urolithins. Among the few studies available on ellagitannins in Combretum spp. is the study of Jossang et al. [74],  In our study, especially butanol and methanol extracts of the stem bark of C. psidioides and C. padoides and the roots of C. zeyheri, enriched in ellagitannins and gallotannins and their monomers (ellagic acid and gallic acid derivatives), gave good antimycobacterial effects. Thus, the antimycobacterial effects of these extracts are suggested to be partly due to these compound classes, and for the ellagitannins perhaps most likely via their urolithin metabolites [86]. This is supported by Coulidiati et al. [102] who demonstrated that tannins, and especially hydrolysable tannins (ellagitannins and gallotannins), predominate in n-butanol extracts of Combretum sericeum and these compounds were suggested to be responsible for the antimicrobial activity of the extracts.
We found that punicalagin ( Figure 6) is present in the stem bark of Combretum psidioides and in the roots of C. zeyheri. To the best of our knowledge, there are no previous reports on the occurrence of punicalagin in the mentioned Combretum species. Previously, apart from C. glutinosum, punicalagin was found in Combretum molle [13] and both the αand β-punicalagin anomers were characterized in Combretum aculeatum [86]. Punicalagin isolated from the stem bark of C. molle gave growth inhibitory effects against M. tuberculosis typus humanus ATCC 27,294 and against a clinical strain with MIC values of 600 and 1200 µg/mL, respectively [13] and was the first ellagitannin reported to possess antimycobacterial effects. We therefore suggest that part of the antimycobacterial activities of the butanol extracts of C. psidioides stem bark and C. zeyheri roots, reported in this paper, might be due to this ellagitannin, and perhaps in combinations with the other ellagitannins in these extracts. However, again, these activities are in vitro, and further studies should be performed on urolithins resulting from the metabolism of punicalagin, and their effects on mycobacterial strains.
Our study revealed the occurrence of corilagin and sanguiin H-4 in Combretum psidioides and a corilagin like ellagitannin in C. padoides stem bark ( Figure 6). To the best of our knowledge, corilagin and/or its isomers and sanguiin H-4 have not been found in Combretum species earlier. Corilagin has been reported to occur in various species of the closely related genera Terminalia [86,103,104] and Lumnitzera (Combretaceae) [105]. Corilagin was found to give good growth inhibitory effects against S. aureus with a MIC value of 25 µg/mL [106], and to inhibit the growth of methicillin resistant S. aureus [107,108]. Moreover, corilagin increased membrane permeability in E. coli and C. albicans [109]. In addition, corilagin gave bactericidal effects in vitro against Klebsiella pneumoniae [110] and inhibited the growth of Acinetobacter baumanii [111]. Interestingly, corilagin was found to potentiate the activity of β-lactam antibiotics 100-2000-fold against methicillin-resistant S. aureus via inhibition of the activity of penicillin binding protein 2 [108,112]. However, in our screenings on the effects of pure corilagin on M. smegmatis growth, we found that this ellagitannin showed only weak growth inhibition (MIC 1000 µg/mL, Table 3). This could imply that the ellagitannins in the Combretum-extracts work together with each other and with other secondary defense compounds to produce antimycobacterial, and perhaps synergistic or additive effects in combinations. Further investigations on the effects of ellagitannins in the Combretum species studied in this investigation and their urolithin metabolites, both alone and in combinations with isoniazid and rifampicin, on the growth of M. smegmatis and M. tuberculosis are warranted.
Since traditional medicines with Combretum spp. as ingredients are often prepared as decoctions or macerations, this means that polar ellagitannins and other polyphenols, extracted with hot or cold water are important components in them. Macerations of the stem bark of C. psidioides, enriched with polyphenols, are specifically used for the treatment of diarrhea [34] and decoctions of the roots, leaves and stem bark of C. zeyheri are used for cough that could be related to TB [35,37,38], which implies that these species and preparations contain antibacterial (and antimycobacterial) compounds. Interestingly, we have seen that the chromatographic profiles of butanol and methanol extracts of the studied Combretum spp. resembled those of the corresponding water extracts, with ellagitannins as the main compounds also in water extracts ( Figure S3). This could imply that decoctions and macerations of C. psidioides, C. padoides, and C. zeyheri, used in African traditional medicine, would contain ellagitannins with antibacterial (and antimycobacterial) effects (via their urolithins).

Suggested Antimycobacterial Impact of Ellagic Acid Derivatives in the Species of Combretum Used in This Study
According to the results of the present study, all the investigated species of Combretum are rich sources of ellagic acid derivatives. We found that tentatively identified 3 -O-methyl-4-O-(β-d-xylopyranosyl) ellagic acid was present in large quantities in a butanol extract of C. psidioides stem bark and ellagic acid arabinoside was present in the stem bark of C. padoides (Tables 4 and 5). Moreover, dimethyl-ellagic acid xyloside and 3,3 -di-O-methyl-4-O-(n -O-galloyl-β-d-xylopyranosyl) ellagic acid were tentatively characterized in a butanol extract of the stem bark of C. zeyheri (Table 6). To the best of our knowledge, these ellagic acid derivatives have not previously been found in the mentioned species of Combretum. These ellagic acid derivatives are suggested to contribute to the good antimycobacterial effects presented in this study for Combretum psidioides, C. padoides and C. zeyheri since there are a number of reports on good antimycobacterial effects of ellagic acid derivatives. For example, Kuete et al. [25] reported that 3,4 -di-O-methylellagic-acid-3 -O-β-d-xylopyranoside and 4 -O-galloyl-3,3 -di-O-methylellagic acid 4-O-β-d-xylopyranoside, isolated from the stem bark of Terminalia superba gave promising growth inhibitory effects against Mycobacterium smegmatis and M. tuberculosis H37Rv as well as a clinical strain of M. tuberculosis, showing MIC values between 4.88 and 9.76 µg/mL. Moreover, ellagic acid derivatives, such as 3,3 -di-O-methyl-ellagic acid, has been found to inhibit the synthesis of mycolic acid, an important cell wall constituent in Mycobacterium spp. [113] and pteleoellagic acid and some other ellagic acid derivatives revealed good in silico effects on their docking capacity to enzymes important for the biogenesis of the mycobacterial cell wall [114]. Digalloyl-rhamnopyranosyl ellagic acid and diellagic lactone isolated from the leaves of Terminalia brownii did not show growth inhibitory effects against Mycobacterium intracellulare, but showed good growth inhibition against other bacteria such as Pseudomonas aeruginosa (IC50 8.8. and 8.4 µg/mL, respectively) [115]. In summary, the above-mentioned findings warrant further studies on the antimycobacterial effects of ellagic acid derivatives in African species of Combretum.

Stilbenes in Combretum Psidioides and Their Possible Antimycobacterial Effects
The present study demonstrated that the bibenzyl, combretastatin B-2, and some related dihydrostilbene derivatives occurred in a methanolic stem bark extract of Combretum psidioides (Figure 4). Earlier, fourteen phenanthrenes and some bibenzyl derivatives (not including combretastatin B-2) have been characterized from the stem bark of C. psidioides [63], but these compounds were not investigated for their antibacterial effects. Combretastatins and phenanthrenes are known to occur in some other species of Combretum, the most wellknown source being the stem bark of the South African species C. caffrum from which a series of antineoplastic combretastatins, among them combretastatin as well as combretastatins A-1, A-2, A-3, B-1, B-2, B-3, and B-4 have been isolated and characterized [65,116]. Some work has been performed on the antimicrobial effects of stilbenoids and phenanthrenes from Combretum species: Combretastatin B-5 from the leaves of C. woodii gave good antibacterial effects against S. aureus with a MIC of 16 µg/mL [67] and phenanthrenes from C. collinum, C. hereroense and C. apiculatum gave a MIC value of 25 µg/mL against Mycobacterium fortuitum [117]. These results warrant further studies on the antibacterial and antimycobacterial effects of stilbenoids and phenanthrenes from the Combretum species used in this study as well as from African Combretum spp. in general.

Epigallocatechin Gallate
Our results indicated that the green tea polyphenol, epigallocatechin gallate, was present in a crude methanol extract of the stem bark of C. psidioides (Table 4). This compoud has not been found before in C. psidioides. Interestingly, epigallocatechin gallate has been shown to affect cell wall integrity of M. smegmatis mc2155, and is suggested to be a good prophylactic agent for TB [118]. In addition, it was found that epigallocatechin gallate inhibits the survival of M. tuberculosis within human macrophages [119]. Moreover, (−)-epigallocatechin gallate has been found to repress the expression of transcription factor lasB, involved in the quorum sensing system of Pseudomonas aeruginosa [120]. Our study suggests that more species of Combretum, used for TB in African traditional medicine, should be explored for their contents of epigallocatechin gallate. Epigallocatechin gallate could be one of the effective antimycobacterial compounds in the stem bark of C. psidioides.

Extraction Yield and Its Impact on the Total Antimycobacterial Activity of the Extracts of the Species of Combretum Used in This Study
When a traditional medicinal plant is evaluated for its usefulness, it is important to measure the total activity of its extract or isolated fractions/compounds [84]. The total activity is a measure of the extraction yield of the extracts/fractions divided to the antimicrobial activity of the extracts/fractions (mg extraction yield/MIC in mg/mL) and indicates the volume in ml to which the active compounds in 1 g of the plant material can be diluted and still inhibits bacterial growth [84].
We found that methanol extraction of the stem bark of C. psidioides results in a high extraction yield of 19.59% (195.9 mg extracted from 1 g plant material) compared to 10.64% and 15.33%, for C. padoides and C. zeyheri, respectively ( Figure 7). Thus, the methanol stem bark extract of C. psidioides gives a good total activity of 313.44 mL/g (resulting from dividing the extraction yield to the MIC; 195.9/0.625 = 313.44), and indicates that standardized C. psidioides stem bark extracts could have applications as antimycobacterial phytomedicines. The extraction yields for C. padoides and C. zeyheri obtained in the present study are in agreement with the yields mentioned by Masoko et al. [121]. Moreover, Masoko et al. [121], found that methanol and acetone are suitable solvents for extracting antimicrobials from Combretum spp.
When the extraction yields of different fractions resulting from solvent partition were compared, it was observed that the butanol fraction gave the highest yield of 47.52% for Combretum padoides, and thus resulting in a total antimycobacterial activity of 190.12 mL/g (Table 2, Figure 8). This means that butanol is an optimal solvent for extracting antimycobacterial compounds, such as ellagic acid derivatives and ellagitannins from C. padoides. The total extraction yield when using Soxhlet extraction with methanol was, however, observed to be quite low, 10.61% (Figure 7), resulting in a total antimycobacterial activity of 85.12 mL/g of this extract ( Table 2). thus resulting in a total antimycobacterial activity of 190.12 mL/g (Table 2, Figure 8). This means that butanol is an optimal solvent for extracting antimycobacterial compounds, such as ellagic acid derivatives and ellagitannins from C. padoides. The total extraction yield when using Soxhlet extraction with methanol was, however, observed to be quite low, 10.61% (Figure 7), resulting in a total antimycobacterial activity of 85.12 mL/g of this extract ( Table 2).

Plant Material and Ethnopharmacological Background Data
The Combretum species were collected from Miombo woodland and riverine forest habitats in Mbeya and Iringa districts in Tanzania in February-March 1999 [32]. Species identification was

Plant Material and Ethnopharmacological Background Data
The Combretum species were collected from Miombo woodland and riverine forest habitats in Mbeya and Iringa districts in Tanzania in February-March 1999 [32]. Species identification was performed by the botanist Mr. Leonard Mwasumbi, the former superintendent of the Herbarium of the University of Dar-es-Salaam, Tanzania. Voucher specimens are deposited in the Botanical Museum (H) of the Finnish Museum of Natural History in Helsinki, Finland and at the Herbarium of the University of Dar-es-Salaam, Tanzania. The Combretum species collected for this study are presented in Table 1 together with data on their ethnopharmacological uses (with special emphasis on cough as a symptom of TB), antimicrobial activity, and their phytochemical composition.

Soxhlet Extraction
For each plant sample, 20 g of dried and milled plant material (roots, stem bark, leaves) was extracted with 500 mL MeOH in a Soxhlet apparatus for 4 h. The extracts were reduced to dryness under vacuum using a rotary evaporator, the temperature of the water bath not exceeding +40 • C. For complete drying, the extracts were then further freeze-dried in a lyophilizer for 1-2 days.
The resulting residues were redissolved in MeOH (crude extract) to a final concentration of 50 mg/mL for antimycobacterial screening.

Solvent Fractionation
Lyophilized crude Soxhlet methanol extracts (1500 mg) of different plant organs of the Combretum species were dissolved in 50 mL distilled water for 30 min using an ultrasonic bath. The extracts were centrifuged in order to separate material, which did not dissolve in water (aqueous insoluble fraction). The aqueous part was used for further fractionation starting with 3 × 25 mL CHCl 3 :EtOH (2:1), whereafter 3 × 25 mL n-BuOH was used. The fractionations thus resulted in CHCl 3 , n-BuOH, aqueous and aqueous-insoluble fractions. The fractions were evaporated to dryness using a rotary evaporator and re-dissolved in MeOH to stock solutions of 50 mg/mL for antimycobacterial testing.

HPLC-UV/DAD Method I
HPLC analyses for qualitative analysis of extracts of Combretum species were performed as described in Fyhrquist et al. [122]. The liquid chromatographic system (Waters 600 E) consisted of a pump and a controller coupled to a 991 PDA detector and a 717 plus automatic sampler under control of Waters Millennium 2000 software. Samples of 20 µL (20 mg/mL inMeOH) were injected. A reversed phase Hypersil BDS-C-18 analytical column (250 mm × 4.6 mm ID 5 µm) was used for the separations. Elution was performed as gradient elution using A) 0.1% formic acid in water and B) 100% acetonitrile. The gradient started with 85% A and ended with 100% B and the length of the runs were 50-60 min. The flow rate was 1 mL/min. UV chromatograms were constructed at 254, 280, and 340 nm. UVλ absorption maxima spectra of the compounds of interest were recorded between 200 and 400 nm using Millennium 2000 software and compared to reference compounds in the computer library.

HPLC-UV/DAD Method II
A second HPLC-method described in Julkunen-Tiitto et al. [123] and Fyhrquist et al., [122], developed especially for the detection of polyphenols, was used. The HPLC-system consisted of a Waters 600 E pump and a controller coupled to a 991 PDA detector and an autosampler under control of Agilent Chemstation software (Waters Corp., Milford, USA). Samples of 10 µL (2 mg/mL in 50% MeOH) were injected. Separations were performed on a reversed phase Hypersil Rp C-18 analytical column (length: 10 mm; ID: 2 mm; particle size 5 µm). Gradient elution was performed by using solvent systems as follows: A) Aqueous 1.5% tetrahydrofuran + 0.25% orthophosphoric acid and B) 100% MeOH. The flow rate was 2 mL/min. UV chromatograms were constructed at 220, 270, 280, 320, and 360 nm. UVλ absorption maxima spectra of compounds, with special emphasis on ellagitannins and ellagic acid derivatives were compared to the Agilent Chemstation library and to literature [88].

UHPLC/Q-TOF MS Method
Ultra-high pressure liquid chromatography coupled to quadrupole time-of-flight mass spectrometry was used as a sensitive method to detect molecular masses with four decimal precision. UHPLC-DAD (Model 1200 Agilent Technologies)-JETSTREAM/QTOFMS (Model 6340 Agilent Technologies) equipped with a 2.1 × 60 mm, 1.7 µm C 18 column (Agilent technologies) was used for the identification of phenolic compounds based on the method described in Taulavuori et al. [124]. Solvent A was 1.5% tetrahydrofuran and 0.25% acetic acid in HPLC quality water and solvent B was 100% methanol. Gradient runs were as follows: from 0 to 1.5 min, B 0%, from 1.5 to 3 min, 0 to 15% B, from 3 to 6 min, 10 to 30% B, from 6 to 12 min, 30 to 50% B, from 12 to 20 min, 50 to 100% B, and from 20 to 22 min, 100 to 0% B. Negative ion mode were used to acquire the mass spectra depending on the chemical class of the compounds and a mass range from 100 to 2000 m/z was used. The negative mode was found to be favorable for ellagitannins and ellagic acid derivatives. Pfundstein et al., [88] was used as the main reference for ellagitannins, ellagic acid derivatives and gallotannins.
Mass measurement error (mass accuracy) was calculated according to Brenton and Godfrey [125]: Difference between an individual measurement and the true value ∆Mi (in ppm, parts per million) = (M measured − M calculated ) × 10 6 /M calculated , where M measured is the measured mass in QTOF-MS and M calculated is the exact calculated mass according to the molecular formula of the compound. Negative mode of qtof was used, and thus the mass of the hydrogen atom (1.0078) was subtracted from all the calculated masses.

Agar Disk Diffusion
An agar disk diffusion method, explained in detail in Fyhrquist et al. [122], was used. Mycobacterium smegmatis ATCC 14,468 was grown for five days at +37 • C on Löwenstein-Jensen agar slants (Becton-Dickinson & Company, USA). 200 µL of bacterial culture containing 1.0 × 10 8 CFU/mL was inoculated on Petri dishes (∅ = 14 cm, Bibby Sterilin, UK) containing 30 mL Middlebrook 7H10 agar (Difco) supplemented with OADC supplement (Difco) as a top layer and 30 mL base agar (Antibiotic agar No 2, Difco) as a bottom layer. Filter paper disks (∅ = 12.7 mm, Schleicher & Schuell) loaded with 200 µL extracts or fractions (50 and 20 mg/mL, respectively), 200 µL rifampicin (10 mg/mL, Sigma-Aldrich) and 200 µL MeOH (as negative control) were allowed to dry completely before placing them equidistantly from each other on the petri dishes. Prior to incubation, the petri dishes were kept in +4 • C for 1 h. The petri dishes were incubated at +37 • C for five days. Each extract/fraction was tested in triplicate and three separate experiments were performed. The diameter of zones of inhibition (IZD) were measured with a caliper and the mean of 3 diameters ± SEM was calculated. Activity index (AI) was measured as the percentage of activity compared to rifampicin, as described by Fyhrquist et al. [122].

Microplate Dilution Method
A modified microplate assay based on Collins & Franzblau [126] and described in Fyhrquist et al. [122] was used for measuring minimum inhibitory concentrations of extracts of Combretum spp. Mycobacterium smegmatis ATCC 14468 was incubated for five days at +37 • C on Löwenstein-Jensen agar slants and transferred into Dubos broth after completed incubation. The absorbance of the samples was adjusted to 0.1 at 625 nm (approx. 1.0 × 10 8 CFU/mL) using Dubos broth supplemented with Dubos broth albumin (Difco) and diluted further to 5.0 × 10 5 CFU/mL. 100 µL of this inoculum was added to the wells of the microplates (Nunc, Nunclone, Denmark). 100 µL of two-fold dilutions of plant extracts (9.76-5000 µg/mL) and of rifampicin (0.9-1000 µg/mL) in Dubos broth or 100 µL Dubos broth (growth control) were added to the wells of the microplate, so that the final volume in the wells was 200 µL, thus containing 2.5 × 10 5 CFU/mL. In order to measure eventual absorption at 620 nm by the plant extracts or the antibiotic, solvent partition fractions, extracts and rifampicin, alone without bacteria were used as controls and the absorption produced by these samples was subtracted from the corresponding samples containing bacterial cells. The solvent control, MeOH at 5% or less, did not affect the growth of M. smegmatis. The microplates were incubated for four days at +37 • C, whereafter turbidity of the wells at 620 nm was measured using a Victor (Wallac, Finland) spectrophotometer. The results were calculated as the mean percentage inhibition of the growth control of three replicate samples ± SEM. The smallest concentration of the extracts and of rifampicin inhibiting 90% or more of the growth of M. smegmatis and thus resulting in no visible growth was considered as the MIC.

Calculation of Total Antimycobacterial Activity
The total activity is a measure of the relation of the exraction yield of a plant extract divided to the MIC of that extract. The total activity indicates the volume into which 1 g of plant extract can be diluted without losing its antibacterial activity [84]. The total antimycobacterial activity for plant extracts in this study was calculated as follows: Total activity (mL/g) = extraction yield of extract A (mg/1000 mg) divided to the MIC of extract A (in mg/mL).

Conclusions
The results of the present study confirm the antimycobacterial potential of butanol, methanol, and watersoluble extracts of the stem bark of Combretum padoides and C. psidioides and the stem bark and roots of C. zeyheri. The traditional use of decoctions and macerations of C. zeyheri for the treatment of cough [81] is justified by these results. The results from this study also suggest that standardized ethanol and water-based extracts of C. padoides and C. psidioides could be used for the treatment of tuberculosis, although these species are not mentioned to be used for this purpose in African traditional medicine.
The growth inhibitory effects of the butanol, methanol and watersoluble extracts of the Combretum species in this study are suggested to be partly due to ellagitannins and ellagic acid derivatives, which are abundantly present in these extracts. However, the therapeutic use of ellagitannins for TB is likely to be limited to their metabolic products, the urolithins, since there is so far only one report on the occurrence of intact ellagitannins in the plasma of rats after a prolonged intake of ET rich foods [127].
The ellagic acid derivatives, and especially the glycosides that we have found in the Combretum species studied, could have an interesting potential as new anti-TB drug scaffolds, since some ellagic acid derivatives, such as dimethyl ellagic acid xyloside, that was found in the roots of C. zeyheri in this investigation, have been found to possess promising antimycobacterial effects, amongst others as inhibitors of the synthesis of mycolic acid [113].
The results from this study indicate that extracts of Combretum psidioides, C. padoides, and C. zeyheri should be tested against other mycobacteria, and especially Mycobacterium tuberculosis, since moderate MIC values against M. smegmatis could imply good MIC values against M. tuberculosis. The activity guided isolation of active compounds from the butanol extracts, with special emphasis on ellagic acid derivatives, remains to be performed. Moreover, urolithins resulting from the metabolic conversion of the ellagitannins punicalagin, corilagin, and saguiin H-4, should be tested for their antimycobacterial effects.
Lastly, the potential of ellagitannins and ellagic acid derivatives, and other phenolic compounds from the investigated species of Combretum spp. should be tested as antibiotic adjuvants to enhance the effects of rifampicin, isoniazid, and other conventional anti-TB drugs.