Quality assessment and chemical diversity of Australian propolis from Apis mellifera bees

The propolis industry is well established in European, South American and East Asian countries. Within Australia, this industry is beginning to emerge with a few small-scale producers. To contribute to the development of the Australian propolis industry, the present study aimed to examine the quality and chemical diversity of propolis collected from various regions across Australia. The results of testing 158 samples indicated that Australian propolis had pure resin yielding from 2 to 81% by weight, total phenolic content and total flavonoid content in one gram of dry extract ranging from a few up to 181 mg of gallic acid equivalent and 145 mg of quercetin equivalent, respectively. Some Australian propolis showed more potent antioxidant activity than the well-known Brazilian green, Brazilian red, and Uruguayan and New Zealand poplar-type propolis in an in vitro DPPH assay. In addition, an HPLC–UV analysis resulted in the identification of 16 Australian propolis types which can be considered as high-grade propolis owing to their high total phenolic content. Chemometric analysis of their 1H NMR spectra revealed that propolis originating from the eastern and western coasts of Australia could be significantly discriminated based on their chemical composition.

As a result of geographic isolation and a vast array of geographical and environmental habitats, Australia is classified as one of the most megadiverse countries in the world with 84% of the endemic terrestrial plant species accounting for 6% of global plant species 11 . This native flora is a unique and diverse source for Australian propolis. Therefore, Australia has the capability to produce unique and premium propolis types. However, studies on Australian propolis are limited to the investigation of samples from South Australia, which identified some new and known flavonoids 12 , prenylated cinnamates 13 , stilbenes 13,14 , and diterpenes 15 (Fig. 1). Owing to novel stilbenes coming from the endemic plant species Lepidosperma sp. 14 , propolis from Kangaroo Island in South Australia is considered as a unique propolis type in the world 3 . The Kangaroo Island propolis extract exhibited four times more potent antioxidant activity than the Brazilian green propolis 13 . A stilbene compound, 5,4′-dihydroxy-3,3′-dimethoxy-2-prenyl-(E)-stilbene ( Fig. 1), found in the Kangaroo Island propolis inhibited the growth of a panel of cancer cell lines more potently than the anticancer drug tamoxifen 14 .
Currently, only small-scale propolis production occurs in Australia, mainly in South Australia. There remains a lack of understanding of the chemical composition of Australian propolis and its therapeutic potentials. Considered a nuisance hive product by beekeepers, Australian propolis is regularly discarded 16 . With approximately 530,000 honey bee hives 17 , it is estimated that Australia can secure millions of dollars with the domestic harvesting of propolis. The domestic harvesting will provide extra income for both Australian beekeepers and processors while reducing the reliance on imported propolis 16 . To provide a brief overview of the quality of Australian propolis, this study assessed the resin recovered yield, total phenolic content (TPC), total flavonoid content (TFC), and antioxidant activity of the collected propolis. The assessment utilised the key criteria used to develop the quality standards for propolis in Brazil, Russia, Portugal, Japan, Korea, China, and Taiwan [18][19][20] . The chemical diversity of Australian propolis was also examined from the analysis of their HPLC-UV and 1 H NMR profiling data.

Results
In this study, 158 raw propolis samples were collected from different regions across six Australian states by local beekeepers from 2018 to 2020 ( Fig. 2A). All Australian samples showed variations in appearance with colours including brown, dark brown, black, red, red-brown, orange or yellow, and aroma ranging from strongly resin to sweet honey (Fig. 2B). In addition, Brazilian green and red propolis, as well as Uruguayan and New Zealand poplar-type propolis were used as referenced samples for the quality and chemical diversity assessments.
Assessment of resin recovered yield. In general, the aim of propolis extraction is to dissolve bioactive constituents in resin and eliminate beeswax 21 . Literature indicates most antioxidant propolis compositions solubilise better in ethanol and methanol than other solvents such as water, ethyl acetate, chloroform, and benzene 21 . However, the organic solvents can also dissolve beeswax with the amount dependent on solvent polarity 18 . Therefore, it was suggested to use co-solvents of ethanol-water or methanol-water for propolis extraction 22 . Ethanol is safer to handle over methanol, making it the preferred solvent for propolis extraction in research and industrial settings. The optimal concentration of ethanol in water was found to be between 70-95% ethanol, with 70-80% ethanol often used to extract propolis compounds with minimal wax contamination 21 . This study used 70% ethanol in water to extract raw propolis.
The resin recovered yield of 158 samples was highly variable, ranging from 2 to 81% (Fig. 3A). About 6% of the Australian propolis samples exceeded the resin yield of the Brazilian green propolis (51% www.nature.com/scientificreports/ above the average of 23%. The remaining 92 samples had lower levels of pure resin due to higher levels of wax and other compounds insoluble in ethanol 70%. In addition, drought and bushfire conditions occurring in Australia during 2019 and 2020 might also have a negative effect on the resin recovered yield. The data also revealed old propolis samples often exhibited lower yields compared to fresh samples. Here old propolis samples were defined as being stored at room temperature for a period greater than a year by beekeepers before sending for analysis. Assessment of total phenolic content. Polyphenols or phenolic compounds are the most abundant secondary metabolites in plants 23 . In honey bee propolis, nearly 30 classes of phenolic compounds, have been identified 3 . They consist of both flavonoids (mainly flavanone and flavone) and non-flavonoids (mainly phenylpropanoid) 3 . Chemically, phenolic compounds are good electron donors that can disrupt the oxidation of organic radicals by reacting with the oxygen and nitrogen species present on the radical 24 . Due to their anti-     24,25 . TPC of Australian propolis extracts in this study ranged from 1.3 to 180.5 mg GAE/g extract with a mean value of 68 mg GAE/g extract (Fig. 3B). Australian propolis contained TPC relatively comparable to Brazilian, Croatian, Moroccan and Palestinian propolis and much lower TPC than studies where poplar propolis was the dominant type in the country (Table S1, Supplementary information). However, it is not easy to compare the TPC values in this study with others due to different assay protocols and reagent ratios used in different labs. By including the Brazilian green, Brazilian red, and Uruguayan poplar propolis as references, an overview of the phenolic level in Australian propolis can be assessed undoubtedly (Fig. 3B). The referenced propolis had TPC ranging from 82.7 to 134.4 mg GAE/g extract (Table 1), and 13% of the Australian samples had TPC over the upper limits of this range. Assessment of total flavonoid content. Among over 8000 phenolic compounds found from plants, half of them are flavonoids presenting as aglycone, glycosides and methylated derivatives 23 . Most flavonoids identified from propolis are aglycone and methylated analogues, while only three flavonoid glycosides have been isolated so far 3 . China suggests the TFC as a reference standard for their propolis which is mainly the poplar propolis type 26 . While Chinese legislation requires a minimum flavonoid content of 8% (w/w) in raw propolis 27 , Brazilian legislation accepts 0.5% (w/w) as a minimum 6 . Studies indicated that the poplar propolis typically contained higher total flavonoid content than the Brazilian green propolis 28 .
Owing to a specific chemical scaffold of a carbonyl C-4 in the B-ring and a hydroxy group at C-5 in the A-ring, flavonoids can bind strongly to aluminum ions to form yellow stable complexes [Al III (flavonoid-H) 2 ] + showing an electronic band at a wavelength of 415 nm 29 . The flavanone-aluminum complex has been known to display less absorbance than the flavone-aluminum complex 30 . Often, the measurement of the yellow [Al III (flavonoid-H) 2 ] + complex provides quantification of TFC in natural sources, expressed in an equivalent term of milligrams of quercetin 20,31-43 .
The assessment of Australian propolis indicated the TFC of Australian propolis was in the range of 0.2-144.8 mg QE/g extract with a mean value of 24.0 mg QE/g extract. Argentina 37 , Croatia 42 , Japan 39 , Morocco 19 , Palestine 19 , and South Korea 40 also have TFC reported within the range recorded for Australian propolis (Table S1, Supplementary information). Within the control samples, Brazilian red propolis had the highest TFC with 122.3 mg QE/g extract while the Uruguayan poplar and Brazilian green propolis contained 75 and 57 mg QE/g extract on average, respectively (Fig. 3C). There were 2, 13 and 20 Australian propolis samples having higher TFC than the Brazilian red, Uruguayan poplar and Brazilian green propolis, respectively (Fig. 3C). From the assessment of the total phenolic and flavonoid contents, it can be seen that Australian propolis samples have the wide variation of both phenolic and flavonoid levels which reflect the relatively diverse range of Australian plant resins the honey bees forage. www.nature.com/scientificreports/ Assessment of free radical scavenging activity. The formation of reactive oxygen and nitrogen species has been known to associate with the oxidative deterioration of food products and human diseases caused by oxidative stress processes such as atherosclerosis, diabetes mellitus, chronic inflammation, neurodegenerative disorders, and cancer [44][45][46] . In the present work, the scavenging capacity and reducing power of 2,2-diphenyl-1-picrylhydrazyl (DPPH) free radical was chosen for the antioxidant evaluation of propolis. The DPPH radical scavenging activity of the Australian propolis extracts varied between 4 to 100% inhibition at 100 µg/mL (Fig. 3D). Samples with TPC over 100 mg GAE/g extract showed 80-100% inhibition at 100 µg/mL (Fig. 3E). A positive linear correlation was observed between the DPPH radical scavenging property and TPC when TPC values were below 100 mg GAE/g extract (Fig. 3F). The results of this study agree with the data from a previous study reported by Kasote et al. 47 , confirming that propolis with high levels of phenolic compounds shows a strong antioxidant activity. However, there was no clear positive correlation between TFC and the DPPH radical scavenging activity (Fig. 3G).
To compare the antioxidant potency, IC 50 values of those propolis samples showing 100% inhibition at 100 µg/ mL were determined. The results indicated that the antioxidant activity of some Australian propolis was comparable or better than the Brazilian green, Brazilian red, and Uruguayan poplar propolis (Table 1). Notably, these Australian propolis samples displayed significantly higher TPC. While they all showed less TFC than the Brazilian red propolis, over 50% of them had more TFC than the Brazilian green and Uruguayan poplar propolis (Table 1).
HPLC profiling and classification of high-grade propolis. For pharmaceutical and food applications, knowledge of the chemical composition of raw materials is necessary 48 . Chromatography and spectroscopy are often utilised for the qualification, traceability and authentication of raw materials 49 . Numerous studies on the chemical composition of propolis by various techniques including thin layer chromatography (TLC) 50 , high performance TLC 51,52 , high-performance liquid chromatography (HPLC) [53][54][55] , liquid chromatography coupled with mass spectrometry (LC-MS) [56][57][58] , gas chromatography coupled with mass spectrometry 59 , near infrared spectroscopy 60 , and nuclear magnetic resonance (NMR) 26,43,47,49,[61][62][63][64][65] have been reported. Among them, HPLC has proved to be an effective and reliable technique in the separation of complex natural product mixtures and routine analysis 48,55,66 . Based on retention times and the UV adsorption spectra of the peaks in a chromatogram, HPLC-UV is often selected as a profiling method to provide a quick classification of propolis samples without detailed identification of individual components [53][54][55] . This study classified propolis samples into two groups. High-grade propolis (57 samples) had TPC greater or equal to 75 mg GAE/g extract (10% over the TPC mean value). Low-grade propolis (101 samples) consisted of the remaining samples. Defining Australian high-grade propolis allows the prioritisation of samples for future developments.
Most flavonoids and phenolics show two absorption maximums at around 240-285 and 300-400 nm, which correspond to the benzoyl and the cinnamoyl systems, respectively 67,68 . Due to the high molar absorptivity of the different phenolic classes, 280 nm is the most generic wavelength for phenolic analysis using HPLC-UV 18,48,68,69 . By profiling the chemical composition of 57 high-grade propolis extracts, 16 high-grade propolis types were identified (Fig. 4). These Australian propolis types showed TPC, TFC and antioxidant activity relatively competitive to the three referenced propolis (Table S2, Supplementary information). The HPLC profiles revealed that propolis types 4-12 and 16 were unique and characteristic for each state while types 1-3 and 13-15 shared some similarities. In particular, types 1-2 contained similar compounds eluting from 6.8 to 8.6 min. Compared to type 1, type 2 displayed more polar compounds in a region of 5.1-6.1 min while type 3 had other extra compounds between 2.9 and 3.8 min. A similar case occurred for propolis types 13-15, which shared five compounds eluting at 4.2, 4.6, 5.5, 5.6 and 5.7 min. The results indicated these propolis types were likely produced from multiple resin sources. Surprisingly, the chromatographic profile of the Australian high-grade propolis type 15 was almost identical to that of Uruguayan and New Zealand poplar propolis (Fig. 4). This finding provides evidence that Australia also produces poplar propolis similar to other countries in its temperate regions (NSW, TAS and VIC). Whether Populus spp. or other plant species is the botanical source of this Australian propolis type warrants further investigation. The Brazilian green and red propolis are considered premium propolis types, and they represent two of 12 propolis types defined in Brazil 70 . No Australian propolis tested in this study showed similar HPLC profiles to that of Brazilian green and red propolis. However, the identification of 16 Australian phenolicrich propolis types indicates that Australia is also home to a diverse range of world-class propolis.

H NMR profiling and chemometric analysis of Australian propolis. Recently H NMR has been
receiving considerable attention for metabolic fingerprinting analysis to test food quality, origin, manufacture, or authenticity due to its high reproducibility, unbiased structural information production and possibility of detecting fraudulent compounds in a sample 71 . This non-destructive technique has been utilised to evaluate and identify possible adulterations in coffee 72,73 , honey 74,75 , milk and dairy products [76][77][78] , liquor [79][80][81][82] , oil 83,84 and juice 85,86 . For propolis, the advantage of the 1 H NMR method is that it can detect waxes and terpenoids, which are generally not observed by HPLC-UV due to their lack of UV chromophores 62 . In general, the 1 H NMR spectra of propolis extracts display all phytochemical compound classes, which can be determined based on their NMR fingerprints present in specific regions of the spectra. The peaks that occur in the region 0.5-3.0 ppm are mainly terpenoids, steroids or linear fatty acids from wax residues, whereas peaks found in regions 3.5-5.5 and 6.5-8.0 ppm are sugars and phenolics, respectively 47 . The singlet around δ H 11.0-12.0 ppm could be attributed to an intramolecular hydrogen bond formed by the -OH group at C-5 and the ketone at C-4 of the flavonoid molecules 43 . However, this signal is only observable in the 1 H NMR spectrum when the sample is recorded in aprotic deuterium solvents 87 . Nine studies utilised the 1 H NMR profiling method to examine the chemical diversity of propolis associated with seasonal variations or different regions in Brazil, China, India, Mexico and several European countries ( www.nature.com/scientificreports/ used in these studies. Among them, DMSO-d 6 was the most selected solvent due to its advantages in solubilising a wide range of compounds with different polarities and facilitating the detection of exchangeable proton signals, which is beneficial for the assignment of flavonoids 88 . Therefore, DMSO-d 6 was also chosen in this study to relatively support the level of TPC and TFC determined. With the profiling of 158 samples, this study has generated the largest 1 H NMR propolis database so far ( Table 2). The 1 H NMR analysis confirmed that all highgrade propolis contained characteristic phenolic signals, particularly from 6.50-7.80 ppm, with relatively higher intensities than other non-phenolic signals. In addition, the typical proton signal at 11.0-12.0 ppm was mostly   www.nature.com/scientificreports/ observed in the 1 H spectra of propolis having high total flavonoid content. Propolis in the low-grade group were mainly terpenoid-rich or sugar-rich as assigned by high terpenoid or sugar fingerprint signals (Fig. 5). The 1 H NMR spectral data of both high-grade and low-grade propolis samples were further processed and analysed by chemometric methodologies using untargeted multivariate statistical principal component analysis (PCA) and partial least squares-discriminant analysis (PLS-DA) to understand chemical similarity or differences according to the geographical regions the samples were collected. The PCA explains in an unsupervised way the variance of each dataset when increasing the number of principal components without referring to any class label while the PLS-DA extracts the information that can predict all possible class memberships from linear combinations of original NMR bins with the use of multivariate regression techniques and assess all class discriminations 71 . The PCA score plots of high-grade propolis did not show any separate grouping among six datasets with a 95% confidence level (Fig. 6A). In contrast to the PCA, the supervised PLS-DA (Fig. 6B) provided some discriminations enabling the differentiation of QLD, SA and WA from other clusters. In particular, the separated Hotelling's T2 ellipses of QLD and SA clusters indicated the uniqueness in chemical compositions of their high-grade propolis (Fig. 6B). Since the SA and WA states are adjacent ( Fig. 2A), they shared some similar propolis types but also had some distinct propolis as shown in Fig. 6B. The results also revealed that chemical constituents of propolis from NSW and VIC were relatively similar. Interestingly, despite having a small number of samples in the dataset, the TAS cluster spread across other states, indicating this state possesses highly diverse propolis samples, some of which are similar to NSW and VIC propolis. The conclusion for the chemical diversity of TAS propolis becomes more certain if additional samples are discovered. The PCA of the low-grade propolis did not show any significant discriminations (Fig. 6C). Further PLS-DA analysis only exhibited a partial separation of some propolis in QLD and WA (Fig. 6D). This discrimination reflected the greater geographic and floral differences between QLD and WA, located on Australia's eastern and western coasts, respectively ( Fig. 2A).
In conclusion, by utilising chemical assays with HPLC and 1 H NMR profiling, this study introduced an overview of the Australian propolis's quality and chemical diversity. Although the production cost of Australian propolis might be greater compared to international sourced propolis, targeting premium propolis market is a strategy for Australia to establish a propolis industry. The data from this study indicates it is possible to find premium Australian propolis types which have higher total phenolic content, total flavonoid content and antioxidant property than some well-known international propolis. The study has identified and reported 16 types of Australian high-grade propolis for the first time. These findings will contribute to the premiumisation of Australian honey bee propolis. In addition, the 1 H NMR database generated in this study will help address the issues of adulterated, unnaturally enhanced and faux propolis in the future for the Australian propolis industry. Although more work is necessary to build a comprehensive picture for Australian propolis regarding their chemical composition and therapeutic values, the results of this study provide a foundation for the commercial production of the new premium propolis in the domestic and global food, and cosmeceutical industries.

Materials and methods
Solvents and reagents. All solvents including ethanol (EtOH), methanol (MeOH) and acetonitrile    Evaluation of antioxidant activity using DPPH free radical scavenging assay. The DPPH free radical scavenging activity of the propolis extracts at different concentrations was evaluated using the DPPH assay as described previously with some modifications 31 . Briefly, the DPPH solution was prepared on the day of measuring at a concentration of 100 µM in MeOH. The propolis extracts (200 μL) at different concentrations were added to 600 μL of DPPH solution in Eppendorf tubes. The mixtures were kept in the dark at room temperature for 20 min before being plated to a 96-well plate (200 µL/well) and measured at 518 nm using a Perkin Elmer Enspire microplate reader. All evaluations were performed in triplicates. Gallic acid and MeOH were used as positive and negative controls. The % inhibition of the DPPH radical for each sample was normalised and calculated using the following formula: where A S is the absorbance of the sample, A P is the absorbance of the positive control and A B is the absorbance of the blank sample (negative control). The IC 50 values of the extracts were determined as the concentration required to inhibit 50% of DPPH free radical.
HPLC analysis. The HPLC analysis was performed on the HPLC system (Agilent 1290) equipped with autosampler, quaternary pump, and DAD detector. Each sample (10 mg/mL in MeOH) was injected (2 µL) onto a C 18 uHPLC Zorbax column (2.1 × 50 mm, 1.8 µm) and eluted by H 2 O (0.1% formic acid) and MeCN (0.1% formic acid) as mobile phases A and B, respectively. Detection was achieved at 280 nm. The mobile phases were used at a flow rate of 0.4 mL/min with a 15-min gradient program which was started at 2% B for 0.5 min, increased to 100% B for 9.0 min, kept at this level for the next 3.0 min, then reduced to 2% B for 1 min and finally reequilibrated for 1.5 min.
NMR analysis and processing. The 1 H NMR spectra were acquired at 300 K on a Bruker Ascend 400 MHz spectrometer equipped with a 5 mm room temperature probe and processed by Bruker TopSpin 3.6 software. The spectrum was recorded using the standard pulse sequence, with a 90° pulse length of 9.61 µs, 64 scans, a spectral width of 16 ppm, a relaxation delay of 5 s, and an acquisition time of 3.75 s. The spectra were referenced to the DMSO residual solvent signal δ H 2.50 ppm.
Chemometric analysis. The 1 H NMR spectral data between 0.00-8.00 ppm in an Excel format were exported to MetaboAnalyst 5.0 software 90 (https:// www. metab oanal yst. ca/) for principal component analysis (PCA) and partial least-squares discriminant analysis (PLS-DA). In all cases, Hotelling's T2 regions depicted by ellipses in score plots of each model defined a 95% confidence interval.

Data availability
All data generated or analysed during this study are included in this published article and its Supplementary Information file.