Comparison of the Antibacterial Activity of Australian Terminalia Spp. Extracts Against Klebsiella Pneumoniae: Novel Treatment for Ankylosing Spondylitis


 Traditional medicines prepared using Terminalia species have been used globally to treat inflammation and pathogenic infections. Recent studies have demonstrated that multiple Asian and African Terminalia spp. inhibit bacterial triggers of some autoimmune inflammatory disease, including ankylosing spondylitis. Despite this, the effects of Australian Terminalia spp. on a bacterial trigger of ankylosing spondylitis (K. pneumoniae) remain unexplored. Fifty-five extracts from five Australian Terminalia spp. were investigated for K. pneumoniae growth inhibitory activity. Methanolic, aqueous and ethyl acetate extracts of most species and plant parts inhibited K. pneumoniae growth, with varying potencies. Methanolic leaf extracts were generally the most potent bacterial growth inhibitors, with MIC values of 66 µg/mL (T. ferdinandiana), 128 µg/mL (T. carpenteriae) and 83 µg/mL (T. petiolares). However, the aqueous leaf extract was the most potent T. grandiflora extract (MIC = 87 µg/mL). All T. catappa extracts displayed low growth inhibitory activity. The Terminalia spp. methanolic leaf extracts were examined by LC-MS and GC-MS. All contained a relative abundance of simple gallotannins (particularly gallic and chebulic acids), the flavonoid luteolin, as well as the monoterpenoids cineole and terpineol. Notably, all Terminalia spp. were non-toxic or of low toxicity in ALA and HDF toxicity assays, highlighting their potential for preventing the onset of ankylosing spondylitis and treating its symptoms once the disease is established.


Introduction
Ankylosing spondylitis (AS) is a relatively common autoimmune in ammatory disease that a ict approximately 0.9% of the world's population, with a signi cantly higher incidence in males (Rashid and Ebringer 2007). The onset is most common in middle to older age, although it can also occur in younger individuals (Braun and Sieper 2007;Rashid and Ebringer 2007). AS is a debilitating autoimmune in ammatory disease that affects spinal and pelvic sacroiliac joints of genetically susceptible individuals and is characterized by in ammation of the spinal joints, resulting in chronic back pain and reduced mobility. Chronic AS in ammation may induce ossi cation of the annulus brosis, resulting in the formation of syndesmophytes, which may form bone bridges between the vertebrae, thereby fusing them together. There currently are no effective cures for AS and current treatments instead target the symptoms of the disease (particularly pain and local tissue swelling) using analgesic and anti-in ammatory therapies rather than addressing the etiology or progression of the disease. However, these treatments only address the immediate symptoms of the disease and do not signi cantly affect progression of the disease or the signi cant tissue damage that results from chronic AS (Saag et al. 1994). Furthermore, prolonged usage of anti-in ammatory drugs and analgesics often results in toxicity and unwanted side effects (Saag et al. 1994).
Safe and effective AS chemotherapeutics are urgently required. Ideally, new AS drugs should prevent AS disease progression (rather than just blocking the symptoms), thereby inhibiting the formation of spinal syndesmophytes and subsequent vertebrae fusion. Notably, some causes of AS have been identi ed. It may be triggered by molecular mimicry in genetically susceptible individuals (individuals with the MHC class 1 allele HLA-B27) who contract K. pneumoniae infections (Cock and Cheesman 2018; Rashid and Ebringer 2007). Klebsiella pneumoniae infections induce the production of antibodies that target bacterial surface proteins, which also cross-react with host HLA-B27 QTDRED and DRED epitopes, as well collagens I, III and IV in genetically susceptible people (Rashid and Ebringer 2007). Binding of these bacterial-targeted antibodies to antigens in the host's tissues triggers in ammatory cascades, resulting in increased production of acute phase proteins, complement proteins, vasoactive amines, as well as multiple cytokines and chemokines (Rashid and Ebringer 2007).
Furthermore, binding of the host-reactive antibodies to self-tissue also activates cellular processes, including activating natural killer cells, thereby stimulating further damage to the host tissue (Braun and Sieper 2007).
Multiple antibiotics are already effective inhibitors of K. pneumoniae growth, including the aminoglycosides, chloramphenicol, uoroquinolones and tetracyclines. However, as AS is a chronic condition, preventative therapy would entail ongoing administration of antibiotics. This mode of therapy would be counter-productive as prolonged exposure to antibiotics induces the production of antibiotic resistant bacterial strains, thereby reducing (or even totally negating) the e cacy of the therapy. Indeed, several super-resistant bacterial strains of K. pneumoniae have already been reported, including a strain that is resistant to all common clinical antibiotics (Cheesman et al 2017; Ashayeri-Panah et al 2014). New safe and effective anti-K. pneumoniae therapies are urgently required. A re-examination of traditional medicines is an attractive option for antibiotic drug discovery as the anti-bacterial properties of some medicinal plants have been known for hundreds or (in some cases) thousands of years. Furthermore, due to the established use of traditional medicines, toxicities and side effects are often well documented. Plant-based preparations also often contain multiple compounds with antibacterial activity, as well as potentiating compounds, and are therefore essentially combinational therapies. This is noteworthy because the presence of multiple molecules targeting different aspects of bacterial growth substantially reduces the possibility of the bacteria developing resistance to the therapy. Indeed, we are unaware of any bacteria that have previously developed resistance to crude plant extracts.
Recent studies from our group have identi ed extracts prepared from several Terminalia spp. as being effective against the bacterial triggers of rheumatoid arthritis (RA) ( bark, fruit and leaves were obtained from the Mt Cootha Botanical Gardens in Brisbane and were identi ed by the gardens botanist. Terminalia petiolaris Benth. fruit and leaves were a gift from Jacinta Monck of Kimberley Wild Gubinge, Australia. Voucher specimens of all plant materials were prepared and are stored at the School of Environment and Science, Gri th University, Australia. All plant materials were dried using a Sunbeam food dehydrator and the desiccated material was subsequently ground into a ne powder. Methanol, deionised water, ethyl acetate, chloroform or hexane were added individually to 1 g masses to give a nal volume of 50 mL and extracted for 24 h at 4 o C with gentle mixing on an orbital shaker. All solvents were AR grade and were purchased from Ajax Fine-Chemicals, Australia. The mixtures were ltered through Whatman No. 54 lter paper to remove the solid plant material. The organic solvents were dried at 50 o C in a vacuum oven, whilst the aqueous extracts were lyophilised by at -50 o C. The resultant dry pellets were weighed to determine extraction yield and were subsequently dissolved in 10 mL deionised water (containing 0.5% DMSO), ltered through a 0.22 µm syringe driven lter (Sarstedt) and stored at 4°C until use.

Antibacterial screening
Test bacterial strains All media was supplied by Oxoid Ltd., Australia. The reference Klebsiella pneumoniae strain (ATCC31488) was purchased from American Type Culture Collection (ATCC). The clinical isolate Klebsiella pneumoniae strain was obtained from the science teaching laboratory at Gri th University, Australia. Stock cultures were subcultured and maintained in Mueller-Hinton broth at 4 o C. Culture purity was intermittently monitored by 16S rRNA sequencing to ensure culture purity and homogeneity.

Evaluation of antimicrobial activity
The growth inhibitory properties of the Terminalia spp. extracts against K. pneumoniae were determined using standard disc diffusion assays (Cheesman et al 2019; Sirdaarta et al 2015a) on Mueller-Hinton agar plates. Each antimicrobial assay was performed three times, each with internal triplicates and mean values (± SEM) are reported in this study. Standard discs of ampicillin (10 µg) and chloramphenicol (10 µg) were obtained from Oxoid Ltd., Australia and included on each agar plate as positive controls. Filter paper discs infused with 10 µL of sterile deionised water were also included as a negative control.

Minimum inhibitory concentration (MIC) determination
The minimum inhibitory concentration for each extract was determined using liquid dilution MIC assay and solid phase (agar) MIC assays. Liquid dilution MIC assays are a common method for quantifying antibacterial activity and are considered sensitive compared to other assay systems (Ilanko and Cock 2019; Hübsch et al 2014). Furthermore, antibacterial evaluations using this method allows for comparison of the e cacy with other treatments and between studies.
Solid phase agar disc MIC assays were also used in this study as a con rmatory assay, and as a more realistic approximation of bacterial infections on solid surfaces.

Microplate liquid dilution MIC assay
The MICs of the extracts were evaluated by previously described standard methods (Ilanko and Cock 2019; Hübsch et al, 2014). A 0.2 mg/mL volume of piodonitrotetrazolium violet (INT; Sigma, Australia) solution was used to indicate bacterial growth as a colour change (from clear to red). The MIC was visually determined as the lowest dose at which colour development was inhibited.

Disc diffusion MIC assay
The MICs of all extracts was also evaluated by disc diffusion assay across a range of extract concentrations as previously described (Cock et

Toxicity screening
The toxicity of all extracts was assessed using the Artemia nauplii lethality assay (ALA) for rapid toxicity screening, as well as an MTS cellular proliferation assay for an evaluation of cytotoxicity.
Artemia franciscana Kellogg nauplii toxicity screening Standard Artemia franciscana nauplii lethality assays were used as an initial evaluation of the toxicity of the Terminalia spp. extracts (Ruebhart et al 2009). Potassium dichromate (1 mg/mL) (AR grade, Chem-Supply, Australia) was included with each assay as a reference toxin. The LC 50 with 95% con dence limits for each treatment was calculated using Probit analysis.

Cellular viability assay
All Terminalia spp. extracts were also screened for cytotoxicity against normal human primary dermal broblasts (HDF ATCC PCS-201-012) purchased from American Type Culture Collection. All extracts were screened at 200 µg/mL following 24 h exposure using standard protocols and absorbances were monitored at 540 nm and a blank wavelength of 690 nm using a Molecular Devices, Spectra Max M3 plate reader (Shalom and Cock, 2018). All tests were performed three times in triplicate (n=9) and triplicate controls were included on each plate and the % cellular viability (compared to the untreated control) was recorded. Cellular viability ≤50% of the untreated control was classi ed as toxic, whereas extracts that yielded cell viability >50% untreated control were deemed to be nontoxic.

HPLC-MS QTOF ngerprint analysis
Separation, identi cation of extract components and determination of the relative abundance was achieved using an Agilent 1290 HPLC system equipped with a Zorbax Eclipse C18 column (2.1 x 100 mm, 1.8 µm particle size) linked to an Agilent 6530 quadrupole time-of-ight (QTOF) mass spectrometer. The chromatographic separation utilised parameters previously optimised for the identi cation of phytochemical components in related Terminalia spp. extracts (Sirdaarta et al 2015a). Mass spectra were obtained in the electron ionisation mode and analytes were recorded for 45 mins in total ion count (TIC) mode.

GC-MS head space analysis
The Terminalia spp. extract components were separated, identi ed and their relative abundance was analysed using a Shimadzu GC-2010 plus (USA) chromatography system linked to a Shimadzu MS TQ8040 (USA) mass selective detector system using previously developed parameters (Sirdaarta et al 2015b). For analysis, the mass spectrometer's ionisation mode was set to 70eV and the mass signals were recorded as total ion count (TIC) mode for 45 mins across a mass range of 45-450 m/z.

Statistical analysis
Data is expressed as the mean ± SEM of three independent experiments, each with three internal replicates (n=9). Statistical signi cance was calculated using one-way analysis of variance (ANOVA) followed by Tukey's post hoc analysis. P values <0.01 were considered signi cant.

Liquid extraction yields
Extraction of the Terminalia spp. plant materials with solvents of varying polarity yielded dried plant extracts ranging from 32 mg (T. grandi oria leaf ethyl acetate extract) to 426 mg (aqueous T. ferdinandiana leaf extract) from 1 g of dried, powdered plant material. These dried extracts were resuspended in 10 mL deionised water (containing 1 % DMSO) for further testing. The aqueous and methanolic extracts generally had substantially higher yields of extracted material compared to the corresponding chloroform, ethyl acetate and hexane extracts (Table 1). Additionally, the higher polarity methanol and water solvent extractions yielded the greatest relative abundance and widest diversity of phytochemical classes for all Terminalia spp. across all plant materials examined. In general, all of the methanolic and aqueous Terminalia spp. extracts contained high levels of phenolics, avonoids and tannins, as well as moderate levels of saponins. The ethyl acetate extracts also contained similar phytochemical classes, although in substantially lower relative abundances compared to the methanolic and aqueous extracts.

Antibacterial activity
To evaluate the growth inhibitory activity of the Australian Terminalia spp. extracts against K. pneumoniae, all extracts were initially screened by disc diffusion assays and the zones of inhibition (ZOI) were measured. The methanolic, aqueous and ethyl acetate T. ferdinandiania fruit (Figure 1a), leaf (Figure 1b) and seed ( Figure 1c) extracts all inhibited K. pneumoniae growth. Notably, the T. ferdinandiania leaf extracts were substantially stronger than the extracts produced from either the fruit or seeds of that species (as evaluated by size of the ZOI). Indeed, ZOIs of approximately 18 mm were measured for the methanolic leaf extract against both the reference and clinical isolates strains of K. pneumonia ( Figure 1b). Furthermore, this extract was a substantially better inhibitor of K. pneumoniae growth than both control antibiotics. Both strains of K. pneumonia were completely resistant to the ampicillin control, although both bacterial strains were highly susceptible to chloramphenicol (ZOIs ~14 mm). Interestingly, the ZOI measured for the methanolic T. ferdinandiania leaf extract was substantially larger than the chloramphenicol control disc, indicating that the extract is a substantially more potent anti-K. pneumoniae treatment than the pure antibiotic. Additionally, a relatively high dose (10 µg) of pure chloramphenicol was infused into the control antibiotic disc. In contrast, the extracts tested herein contain complex mixtures of compounds, with the antibacterial components expected to account for a relatively low percentage of the total extracted material. Therefore, the inhibitory activity of the individual methanolic T. ferdinandiania leaf extract components are particularly noteworthy. In contrast, the methanolic T. ferdinandiana fruit and seed extracts produced ZOIs of approximately 10 and 7 mm respectively.
The methanolic T. carpenteriae leaf extract was also a strong inhibitor of both K. pneumoniae strains (ZOIs ~12 mm) ( Figure 2). Indeed, the K. pneumoniae growth inhibition by this extract compares favourably to the inhibition by the antibiotic controls (as judged by the ZOI size). Similarly, the methanolic T. grandi ora fruit ( Figure 3a) and leaf (Figure 3b) extracts were also good inhibitors of the growth of both K. pneumoniae strains, with ZOIs of approximately 11 and 13 mm respectively. However, a notable difference were evident for the T. grandi ora leaf extracts compared to fruit extract (and to the other Terminalia spp. screened in our study). Speci cally, the aqueous T. grandi ora leaf extract was a substantially stronger inhibitor of both K. pneumoniae strains than the methanolic extract was, with ZOIs of 16 and 16.7 mm against the reference and clinical isolate strains respectively.
The methanolic T. petiolares fruit (Figure 4 a) and leaf extracts (Figure 4b) were also strong inhibitors of both K. pneumoniae strains (ZOIs of approximately 13.5 and 17 mm respectively). The aqueous and ethyl acetate T. petiolares fruit and leaf extracts were also good inhibitors of bacterial growth, albeit with substantially smaller ZOIs than measured for the corresponding methanolic extracts. In contrast, relatively weak anti-K. pneumoniae activity was noted for the T. catappa bark (Figure 5a), fruit ( Figure 5b) and leaf extracts (Figure 5c). Indeed, weak growth inhibition was only seen for the methanolic T. catappa extracts (ZOIs ≤7.5 mm against both K. pneumoniae strains), whilst all other T. catappa extracts were completely ineffective inhibitors of K. pneumoniae growth.
Quanti cation of minimum inhibitory concentration (MIC) The relative level of K. pneumoniae growth inhibitory activity was also quanti ed and further evaluated by determination of MIC values of each extract ( Table  2). MIC values were only determined for those extracts that displayed activity in the disc diffusion screening assays. With some noteworthy exceptions, the methanolic and ethyl acetate leaf extracts were generally substantially more potent inhibitors of K. pneumoniae growth than extracts produced from other plant parts, or using other solvents. The T. ferdinandiana methanolic (66 and 53 μg/mL against the reference and clinical strains respectively) and ethyl acetate leaf extracts (78 and 75 μg/mL against the reference and clinical strains respectively) were particularly good inhibitors of both K. pnuemoniae strains. Similarly, the T. petiolares leaf methanolic (83 and 69 μg/mL against the reference and clinical strains respectively) and ethyl acetate extracts (105 and 92 μg/mL against the reference and clinical strains respectively) were also noteworthy. As K. pneumoniae can trigger ankylosing spondylitis in genetically susceptible people (Cock and Cheesman 2018), these extracts may be particularly promising for the prevention and control of that disease, as well as other diseases caused by K. pneumoniae infections.
The growth inhibitory activity of the methanolic and ethyl acetate extracts of most parts of most of the other Terminalia spp. was also noteworthy, with MIC values generally substantially below 1000 μg/mL. The T. catappa extracts (prepared using all plant parts) were a notable exception to this trend, with high MIC values (1500->5000 μg/mL) determined for all T. catappa extracts against both K. pnuemoniae strains. Similarly, MIC values substantially >1000 μg/mL were also determined for all of the T. ferdinandiana seed extracts against these bacteria. Thus, the T. catappa extracts and the T. ferdinandiana seed extracts are likely to be of limited use for the prevention and treatment of ankylosing spondylitis. A further trend was also evident: The leaf extracts were generally substantially more potent inhibitors of K. pneumoniae growth than the corresponding fruit, bark or seed extracts. Additionally, the anti-K. pneumoniae activity of the aqueous extracts of most Terminalia spp. and plant parts studied were also noteworthy, albeit with substantially higher MIC values than determined for the methanolic and ethyl acetate extracts.

Quanti cation of toxicity
All of the Australian Terminalia spp. extracts were initially screened across a range of concentrations in the Artemia nauplii toxicity assay and LC 50 values were quanti ed ( Table 2). As LC 50 values >1000 μg/mL have been de ned as nontoxic in this assay (Ruebhart et al 2009), the majority of the Australian Terminalia spp. extracts were classi ed as nontoxic. Indeed, only the methanolic T. ferdinandiana fruit (LC 50 = 925 μg/mL), T. grandi ora fruit (LC 50 = 934 μg/mL) and leaf (LC 50 = 884 μg/mL), as well as the T. petiolares fruit extracts (LC 50 = 925 μg/mL) displayed toxicity. Furthermore, the relatively high LC 50 values calculated for those extracts indicate that they are only of low toxicity. Additionally, it is noteworthy that whilst Artemia nauplii are generally robust, they are sensitive to acidic pH (Kwast, 1996) and extracts containing high levels of ascorbic acid may provide falacious results (Wright et al., 2016b;Kwast, 1996). Notably, high ascorbic acid content is a characteristic observed in all Terminalia spp. (Cock, 2015) and previous studies have reported that the apparent toxicity of Indian (Mandeville and Cock Cellular cytotoxicity assays are generally considered to provide a more relevant evaluation of the toxicity of plant extracts. Notably, acidic phytochemicals (including ascorbic acid) generally do not signi cantly in uence the viability of eukaryotic cell lines and thus we also evaluated the toxicity of the Australian Terminalia spp. extracts using normal human dermal broblasts HDF viability assays. Notably, exposure of HDFs to all of the Terminalia spp. extracts (including those classi ed as low toxicity in the Atremia nauplii assay) resulted in substantially >50% for all tests (compared to untreated control viability). Thus, all extracts were determined to be nontoxic to HDFs.

Phytochemical ngerprint analysis
The T. ferdinandiana, T. carpentariae, T. grandi ora and T. petiolares methanolic leaf extracts generally displayed substantially greater bacterial inhibitory activity than the other solvent extractions (or the T. catappa extracts). Thus, those extracts were analysed by metabolomic phytochemical ngerprint analysis. In recent studies, LC-MS analyses have been used to pro le the phytochemical composition of multiple Terminalia spp. extracts for the presence of speci c compounds, and to quantify their relative abundances in those extracts (Nel et  Notably, a number of compounds were detected in high relative abundance in the leaf extracts of multiple Terminalia spp. leaf extracts. In particular, the low molecular mass tannins ellagic acid and ellagic acid dehydrates, gallic acid, chebulic acid, and exifone were present in relative abundance. Additionally, the higher molecular mass tannins castalagin, chebulagic acid, chebulinic acid and corilagin were also previously reported in relative abundance in the leaf extracts of other Terminalia spp. leaf extracts. The avonoid luteolin was also previously reported to be in relative abundance in multiple Terminalia spp. leaf extracts (Nel et  Notably, all of these compounds were also identi ed in relative abundance in the Australian Terminalia spp. leaf extracts examined in our study (Table 3). Gallotannins were particularly prevalent across all of the Australian Terminalia spp. methanolic leaf extracts, with the levels of gallic acid (5.8-6.9 % relative abundance), chebulic acid (4-6.9 % relative abundance), corilagin (5-6.9 % relative abundance), chebulinic acid (3.1-4.5 % relative abundance) and exifone (5.7-6.6 % relative abundance) particularly noteworthy. The percentages cited in Table 3 correspond to the chromatographic percent peak area compared to the total area for all peaks in the chromatogram. The relative abundances of ellagitannins was also notable, albeit generally at lower levels than noted for the gallotannins. In particular, ellagic acid (1.3-1.7 % relative abundance), ellagic acid dehydrates (2.1-2.8 % relative abundance) and castilagin (2.1-2.8 % relative abundance) were present across the extracts of all of the Australian Terminalia spp. studied herein. The avonoid luteolin was also present in particularly high relative abundance (5.8-7.4 %) across all of the extracts tested. Notably, the relative abundances of the tannin and avonoid compounds in the methanolic T. ferdinandiana, T. carpentariae and T. grandi ora leaf extracts in our study correlate closely with the previously reported values (Cock et al 2019). To the best of our knowledge, the presence and abundance of these compounds has not previously been reported in T. petiolares. Interestingly, the LC-MS ngerprint of the T. petiolares leaf extract reported herein is similar in both identity and relative percentage abundance to the other Australian Terminalia spp. leaf extracts.
Notably, these relative abundances are also similar to the levels previously reported for other Terminalia spp. leaf extracts using similar analytical methods (Wright et al 2019).

Discussion
The current repertoire of therapies to treat AS are not ideal and only target the symptoms of the disease, whilst allowing irreversible tissue degradation to continue in the vertebral and sacroiliac joints. Therapies targeting the etiological events would prevent the onset of the disease, which would not only alleviate the symptoms of AS, but would also prevent tissue damage. The role of K. pneumoniae infections in the etiology of AS is well established and is supported by a wealth of evidence (Cock and Cheesman 2018;Rashid and Ebringer 2007). Despite this, there are currently no clinical therapeutics for AS that target K. pneumoniae. Inhibiting the growth of K. pneumoniae using prophylactic antibiotic therapies may be effective in the prevention and treatment of AS, and other autoimmune in ammatory diseases. However, prophylactic usage of conventional antibiotics would also induce the development of further antibiotic resistance, rendering the target bacterium (and other microbial species) unsusceptible to the effects of the antibiotic. The use of combinations of antibiotics (or antibiotics and potentiator compounds) may be more effective than antibiotic monotherapies and may also decrease the possibility of the bacteria developing further resistance.
Traditional medicines and herbal therapies are promising targets for antibiotic drug discovery as they have often been used for long periods (in some cases, thousands of years) and are perceived (often erroneously) to be safer than synthetic drugs ( they have yet to be tested for anti-K. pneumoniae activity. Methanolic extracts prepared from all of the Terminalia spp. screened in our study displayed noteworthy K. pneumoniae growth inhibitory activity. Additionally, the aqueous and ethyl acetate extracts of all Terminalia spp. tested (except T. catappa) also inhibited K. pneumoniae growth. Notably, for all of the Australian Terminalia spp., the leaf extracts had substantially greater potency than the corresponding fruit, bark or seed extracts. Therefore, the methanolic T. ferdinandiana, T. carpentariae, T. grandi ora and T. petiolares leaf extracts were further analysed to determine their phytochemical composition using both LC-MS and GC-MS headspace analysis.
Metabolomic ngerprint analyses of the methanolic T. ferdinandiana, T. carpentariae, T. grandi ora and T. petiolares leaf extracts con rmed the presence of some noteworthy compounds with therapeutically relevant properties. In particular, our study targeted the tannins, avonoids and volatile terpenoids. All of the methanolic leaf extracts evaluated contained a diversity of tannins in relative abundance. Notably, tannins inhibit the growth of multiple bacterial pathogen Another feature of the Australian Terminalia spp. phytochemistry reported herein was the diversity and relative abundance of volatile terpenoids identi ed in the methanolic leaf extracts. In particular, the monoterpenoids cineole, terpineol and linalool were present in relative abundance. Interestingly, each of these compounds has potent broad-spectrum bacterial growth inhibitory activity against both gram positive and gram negative pathogens (Zenegin and Baysal, 2014;Paduch et al, 2007). Their antibacterial activities are linked to their small sizes and lipophilic characteristics (Paduch et al, 2007). Cineole, terpineol and linalool (as well as other monoterpenoids) can insert into cytosolic membranes, thereby altering bacterial membrane uidity and permeability, as well as changing the conformation and function of bacterial membrane proteins. These membrane changes dramatically alter bacterial respiration, as well as several other crucial cellular processes (Cristani et  The diversity and abundance of small volatile monoterpenoids in the Australian Terminalia spp. extracts is also particularly noteworthy as their non-speci c antibacterial mechanisms are inherently di cult for bacteria to adapt to and to develop resistance against. Indeed, we were unable to nd any reports of the development of resistance to monoterpenoids by any bacterial species. Furthermore, multiple terpenoid compounds that may function via several antibacterial mechanisms were identi ed in these extracts. These combinations may enhance the growth inhibitory potency of the combination relative to the individual components, and may also substantially reduce the development of antibiotic resistance in the target bacteria. It is unlikely any of these compounds is solely responsible for the bacterial growth inhibitory activity reported in our study. Instead, it is likely that multiple components with several antibacterial properties may collectively interact to produce the noteworthy activity reported herein. Furthermore, synergistic interactions may potentiate the anti-K. pneumoniae activity, thereby increasing their e cacy of the individual components. Our study was limited to examining the inhibitory activity of the Australian Terminalia spp. extracts towards a bacterial trigger of AS. However, these are crude extracts, which contain complex phytochemical mixtures. It is therefore possible that these extracts may also modulate other in ammatory processes such as cytokine release and therefore effect several aspects of AS disease progression. Indeed, several of the compounds identi ed in our study have been reported to have pleuripotent anti-in ammatory effects by acting on both the initiator and downstream in ammatory stages of the disease. For example, several of the terpenoids that we have identi ed herein can suppress NF-κB signalling, and may therefore also directly inhibit in ammation (Salminen et al, 2008). Further studies are required to evaluate Australian Terminalia spp. extracts for direct anti-in ammatory properties, and for additional combinational effects that may enhance the therapeutic properties of individual components of these extracts.

Conclusions
The results of this study demonstrate the potential of T. ferdinandiana, T. carpentariae, T. grandi ora and T. petiolares as inhibitors of K. pneumoniae, a bacterial trigger of AS. The methanolic and ethyl acetate extracts prepared from T. ferdinandiana and T. petiolaris leaves were particularly good inhibitors of K. pneumoniae growth. Additionally T. ferdinandiana, T. grandi ora and T. petiolaris fruit extracts, as well several aqueous extracts, also had noteworthy activity. Whilst the extracts examined in this report are promising as anti-AS agents, caution is needed before these compounds can be applied to medicinal purposes.
In particular, further toxicity studies using human cell lines are needed to verify the suitability of these extracts for these purposes.

Supplementary Files
This is a list of supplementary les associated with this preprint. Click to download. Table1.docx