Ninety five strains of bacteria were obtained from the Department of Veterinary and Animal Science, University of Copenhagen, Denmark, Department of Microbiology, Sokoine University of Agriculture, Tanzania, and Muhimbili University of Health and Allied Sciences, Tanzania ( Supplementary Table S1 ). In addition, a set of reference strains were included: Staphylococcus aureus ATCC25923, Escherichia coli ATCC25922, Klebsiella pneumonia ATCC700603, Pseudomonas aeruginosa ATCC27853, Enterococcus faecium ATCC51559, Streptococcus uberis ATCC854 and Streptococcus agalactiae ATCC12403. Bacteria were propagated on Mueller Hinton agar (MHA) and Mueller Hinton broth (MHB) (Sigma Aldrich, St Louis, USA) at 37°C for 16-24 hours.

Fresh plant samples of S. glaucescens were collected in Njombe district in the Southern Highlands of Tanzania (S08°47’08.5” E 34°53’22.2”). The voucher specimen number HOS/FM 3672 is deposited at the herbarium of Botany Department, College of Natural and Applied Sciences,University of Dar es Salaam, Tanzania. Plants names have been verified at http://www.theplantlist.org. The leaves, roots and stem of the target plants were collected. The roots were manually separated into root bark and rootwood and the stems into stem bark and stem wood. All plants parts were chopped into small pieces and dried at 20 °C at the Tanzania Tree Seed Agency Laboratory, Morogoro. The dried samples were milled to particle size of 2 mm using milling machine (Christy Hunt Engineering Ltd, England). Solvents of different polarities including hexane, dichloromethane and methanol (Sigma Aldrich, Saint Louis, USA) were used to obtain crude extracts from the plant parts using cold (total and sequential) and hot methods of extraction with some modification (11, 12). For cold-total extraction, powdered materials were soaked in a solvent for 48 hours then filtered and for cold-sequential method, residue from dichhloromethane were resoaked with methanol for 48 hours before filtered. The obtained liquid extracts were filtered using Whatman™qualitative filter paper1, 24.0 cm (GE Healthcare Life Sciences, China) in syntax glass. In hot method, powdered materials were boiled for 4 hours in a soxhlet extractor with solvents of different polarities and temperatures to obtain liquid extracts. For the decoction method, powdered samples were soaked in water for about 30 seconds; boiled in a charcoal burner for 30 minutes and then allowed to cool then filtered. All filtrates from organic solvents were concentrated in a rotary evaporator at 40-55 °C to get extracts. The filtrates from water were freezed at -20 °C and taken to the freeze drier (ATS company, Pennsylvania, USA) and other extracts were completely frozen and placed in a freeze-dryer to remove traces of solvents. The dried extracts were stored at -20°C before running bioassays. The matrix of extracts from different parts of the plant (leaves, roots and stems) and with different extraction solvents are summarized in Supplementary Table S2 . Further, the methanol root bark from cold-total extraction was fractionated sequentially with hexane, ethyl acetate and methanol solvents by vaccum liquid chromatography through Buchner funnel. The resulting fractions were subjected to column chromatograph to obtain sub-fractions, which were then purified to yield two compounds. The compounds were epifriedelanol (1) ( Figure 1 ) and euphol (2) isolated from ethyl acetate fraction. Other compounds used were isolated from leaves extract tetracosane (3), hexadecyl palmitate (4), octacosane (5), tetracosanol (6) and β-sitosterol (7); stem wood extract were 1-hexacosene (8), campesterol (9) and tetracosanoic acid (10). These compounds were kindly donated by Frank Rewgoshora and David Credo and were published (13, 14).

Growth inhibition assay was performed in honey comb multi well plate using the Bioscreen C machine (Thermo Fischer Scientific, Finland). Bacteria were grown in MHB overnight, and the broth was diluted by adding MHB to the desired Optical Density (OD₆₀₀) equal to 0.10 (10⁸ CFU/mL). The extracts were dissolved in Dimethyl-Sulphoxide (DMSO) (Sigma Aldrich, Germany) to make concentrationsof 1500 µg/mL and 150µg/mL in MHB, which was mixed with the bacteria in MHB. The plates were incubated for 24 hours with continuous shaking at 37°C and Optical Density (OD₆₀₀) readings at intervals of 20 minutes. The experiment was performed in triplicates, and negative controls in the form of wells with bacteria and solvent used to dissolve the extract and wells with broth and bacteria alone were included in the plates. Bacteria were also added to wells with tetracycline (4 µg/mL) and gentamicin (4 µg/mL) (Sigma Aldrich, Germany). Growth curves of Log₁₀ optical density against time were visualized using GraphpadPrism8 software (San Diego, USA).

Minimum Inhibitory Concentration (MIC) was determined in duplicates by two folds micro-dilution tests performed in a 96-well microtitre plates (15). In brief, test organisms in concentrations equivalent to 0.5 McFarland (1.5 x 10⁸ CFU/mL) were diluted to 100 folds to reach a concentration of 10⁶ CFU/mL. The inoculum was added to well containing 2 fold dilutions of extracts (range 1500 µg/mL to 0.00075 µg/mL) or the antimicrobial gentamicin (range 0.016 µg/mL to 0.000004 µg/mL). Two additional rows were used, one as sterility control, in which, neither drug nor bacteria were added, and one as growth control in which only bacteria and broth were added. The microtitre plates were incubated at 37°C for 16-18 hours. The MIC was taken visually from the well of the lowest concentration showing no bacteria turbidity. The antimicrobial activity of extracts was classified as strong if the MIC range was <500 µg/mL, moderate if MIC range was >500 to 1500 µg/mL and weak if the MIC range was >1500 µg/mL (16).

Time-kill assay of Staphylococcus aureus (ATCC25923) and Escherichia coli (ATCC25922) were carried out according to the procedures described by (17) with slight modifications. One mL of MHB was added to test tubes 1-9. Then 1mL of plant extracts was added to test tube 1 followed by two-fold serial dilutions starting with the 2^nd tube and continued to the 5^th tube leaving a volume of 1 mL in each test tube. A total of 1mL of gentamicin (4 µg/mL) was added in tube 6. Tube 7, 8 and 9 were reserved as controls for the broth with no bacteria, bacteria alone and bacteria with solvent (DMSO) respectively. Then, 1 mL of MHB was added to each tube to make a volume of 2 mL. The turbidity of bacteria was adjusted and visually compared to 0.5 McFarland corresponding to approximately 1.5 x10⁸ CFU/mL and then diluted 100 folds. One hundred µl of the bacteria suspension was added to all tubes except tube 7 and mixed by vortex at low speed. This resulted into an initial inoculum size of approximately 3.5 x10⁵ CFU/mL. The number of colony-forming units (CFU) was determined by plating serial dilutions in Phosphate Buffer Saline on MHA at times 0, 2, 4, 6, 8, 10, 12 and 24 hours after inoculation. Plates were incubated over night at 37 °C. Graphs of the Log₁₀ CFU/mL were plotted against time using GraphpadPrism8 software (San Diego, USA). Time kill curves were determined in triplicates.

Haemolysis activity was measured according to (18). Briefly, sheep blood was collected in vacuutainer tubes. The blood was centrifuged at 3500 rpm followed by three times wash for 5 minutes to obtain red blood cells (RBCs). The obtained RBCs suspension were prepared at 2% in phosphate buffer pH= 7.4. Extracts were prepared at concentrations of 100, 200, 500, 1000 µg/mL. Two mL of RBCs suspension, PBS and plant extract were added to reach the final volume of 4 mL. Hydrogen peroxide (H₂O₂₎ at a concentration of 3% was used as control to induce total haemolysis. The solvent (DMSO) was used as negative control. After 30 minutes of incubation at room temperature, the mixtures were centrifuged at 3500 rpm for 10 minutes and the resulting supernatant was removed and used to evaluate haemolytic activity using a spectrophotometer at the absorbance wavelength of 540 nm. All the experiments were done in three independent experiments. Determination of a percentage haemolysis was calculated by the substraction of the negative control absorbance from the extracts absorbance relative to the positive control absorbance. The final result was multiplied by 100. The interpretation of percentage haemolysis was compared to positive control and considered as low haemolytic activity between 0-40% (18).

The MIC data were expressed as medians and interquartile ranges (IQR). In cases where there were growths in all dilutions, the MIC value was set to the highest tested concentration for statistical purposes. Shapiro-Wilk test was used to test the normality of the data. Comparison of the mean between groups was performed using a Mann-Whitney U test. Comparison of more than two groups was performed by the Kruskal-Wallis test. The IBM Statistical Package for Social Sciences (SPSS) programme, version 20 was used. A two-tailed p-value of < 0.05 level was considered statistically significant.

All plant extracts were first tested for their ability to inhibit the growth of two reference bacteria; the Gram-positive S. aureus ATCC25923 and the Gram-negative E.coli ATCC25922. For the first 5 hours, growth in all samples resembled the growth of the controls. After this time point, the growth of S. aureus was inhibited by cold–total methanol extract of root bark and leaves, and cold-sequential methanol extract of the whole root of S. glaucescens at concentrations of 150 µg/mL. No growth was observed in controls treated with tetracycline or gentamicin ( Figure 2A ). None of the extracts inhibited the growth of E. coli at a concentration of 150 µg/mL and 1500 µg/mL ( Figure 2B ). Full data with inactivation curves for all extracts in different concentrations appear in Supplementary Figure S1 .

Active extracts selected from the growth inhibition assay were subjected to MIC determination against a wider collection of Gram-positive (Staphyllococcus aureus, Streptococci species and Enterococci species) and Gram-negative bacteria (Escherichia coli, Pseudomonas aeruginosa, Klebsiella pneumoniae, Salmonella species, Shigella species, Yersinia enterocolitica). The median MIC of extract was significantly lower for Gram-positive bacteria [90 µg/mL (IQR 30-750 µg/mL), n = 114] than for Gram-negative bacteria [3000 µg/mL (IQR 1500->3000 µg/mL), n = 171], P<0.0001. The extracts from the whole root [MIC = 40 µg/mL (IQR 30-210 µg/mL)] and root bark [MIC =20 µg/mL (IQR 10-190 µg/mL)] exhibited higher activity against Gram-positive bacteria than extracts from leaves [MIC =380 µg/mL (IQR 380-1500 µg/mL)] (P<0.001). The whole root and root bark extracts exhibited strong activity against the Gram-positive bacteria, S. aureus, Enterococci spp. and Streptococci spp., while extracts from leaves demonstrated moderate activity against these bacteria.

When tested against Gram-negative bacteria, extracts of the whole root and root bark exhibited moderate activity against Shigella species and Pseudomonas aeruginosa while they displayed weak activity against Escherichia coli and Klebsiella pneumoniae. Similarly, extracts from leaves displayed weak activity except against P. aeruginosa, where it demonstated moderate activity. In addition, the extracts from whole root and root bark showed strong activity against Yersinia enterocolitica, while the extract from leaves exhibited moderate activity. However, the number of strains tested in this species was low (n = 2). The MIC values of all extracts were significantly higher than that of gentamicin for the same bacteria (p<0.0001) ( Table 1 ).

An extended study was done on determination of MIC of the most active extract (root bark with MIC 20 µg/mL (IQR 10-190 µg/mL)), its fractions, sub-fractions and the isolated compounds against three strains S. aureus and an E. Ccoli. The fractions, sub-fractions and the compound epifriedelanol (1) demonstrated the same activity against the tested S. aureus strains as the crude extract, suggested that the activity was related to these fractions and the specific compound. Other compounds displayed weak activity against S. aureus. All fractions, sub-fractions and compounds (1-10) exhibited weak activity against E. coli, similar to the crude extract ( Table 2 ).

The kinetic effect (bacteriostatic versus bactericidal) of the three active extracts (whole root, root bark, and leaves), fractions, sub-fractions and epifriedelanol (1) were determined against Staphylococcus aureus (ATCC25923) and Escherichia coli (ATCC25922). The time kill curves revealed that the effect was static, such as, in the situation where antibactial activity was observed, it caused stop of growth, but it did not kill the bacteria. There was a significant growth inhibition in CFU/mL after 24 hours for the whole root and root bark extracts at a concentration of 3000-1500 µg/mL, root bark fractions, sub-fractions and a compound epifriedelanol at the concentration of 3000-750 µg/mL. Extracts from leaves also exhibited significant inhibition at the concentrations of 6000-3000 µg/mL when compared to the untreated groups. In contrast at lower concentration, extracts, fractions, sub-fractions and active compound allowed growth of the bacteria, slightly higher than in the untreated groups. However, the difference was not statistically significant (p> 0.05). Treatment with gentamicin resulted in fast reduction of CFU, and no colonies were formed after 4 hours of treatment with this drug. Full account of the comparison within and between treated group against control groups are shown in Figures 3A–H and Suplementary Tables S3 and S4 . For E. coli, only the root bark of S. glaucescens was tested at the concentrations of 24000, 12000, 6000 and 3000 µg/mL, and, no significant difference to the growth curve of the untreated control was observed (data not shown).

The three active extracts (whole root, root bark, and leaves), fractions, sub-fractions and epifriedelanol tested showed haemolytic activity which was significantly above the negative control ( Figures 4A-H ). No haemolytic activity was observed for the solvent control. Eventhough, the haemolytic activities were above the negative control, the haemolytic activities were below 40% when compared to the positive control (H₂0₂), except for extracts from the leaves at the concentration of 500 and 1000 µg/mL ( Figure 5 ). The percentage haemolysis indicated low haemolytic activities of these extracts, fractions, sub-fractions and the compound. Furthermore, the minimum inhibitory concentration observed in the antimicrobial activity for the extracts (whole root, root bark, and leaves), fractions, sub-fractions against S. aureus was < 100 µg/mL, such that the percentage haemolysis at this concentration for the however was less than 14.04%. In contrast, extracts of leaves showed high haemolysis activity (75.01%) at the highest concention of 1000 µg/mL. Generally, the percentage haemolysis showed concentration dependent.