Chemical, microbial, and metabolic analysis of Taisui cultured in honey solution

Abstract Taisui, a special substance occasionally found in China, can now be artificially cultured. In order to evaluate the safety of an artificially cultured Taisui (acTS) and develop it into fermented, functional food or oral liquid, the macronutrients, trace elements, microbial community, and extracellular metabolites of Taisui have been investigated in this study. Results showed that the concentrations of total carbohydrates, protein, fat, total ash, and moisture of wet acTS were 2.13 g/100 g, 0.13 g/100 g, 0.07 g/100 g, 0.04 g/100 g, and 88.3%, respectively. The concentrations of top three trace elements of K, Ca, and P, are 1,424.92 mg/kg, 159.96 mg/kg, and 67.89 mg/kg, respectively. Proteobacteria, Euryarchaeota, and Ascomycota were the dominant phyla of bacteria, archaea, and fungi, respectively. Uncultured_bacterium_f_Anaerolineaceae, Alcaligenes, and Ochrobactrum were the three most abundant genera of bacteria; Methanosaeta, Methanosphaera, and Natronomonas, the most abundant genera of archaea; Zygosaccharomyces, Mortierella, and Fusarium, the most abundant genera of fungi. There were 311 metabolites increased in acTS. Most of the metabolites are beneficial to human. These metabolites can be contributed to microbes in acTS. In conclusion, acTS is not a good source of macronutrients and of trace elements, while the safeness of some microorganisms in acTS is also unknown. Nevertheless, it still provides some probiotics and beneficial metabolites for human. It is thus possible to develop acTS into foods when the safety of each microorganism is proved.

various sources differ. For example, Klebsiella oxytoca and Ralstonia eutropha were both found by the plate-culturing and nonculturing methods (Wang, 2007), whereas Pseudomonas flulorescens and Brevundimonas mediterranea were only detected from another Taisui sample (Tong et al., 2018). Of fungi, Candida and Rhodotorula mucilaginosa were identified by 18S rDNA sequencing method (Lin et al., 2013), whereas Acremonium and Trichoderma were the dominant fungi of another type of Taisui (Dai, 2007). In terms of myxomycete, Didymium verrucosporum and Diderma deplanatum have been successfully isolated using either corn meal-agar or oat meal-agar method (Dai, 2007). Moreover, archaea have also been found in Taisui and the dominant archaea were Methanobacterium, Methanobrevibacter, Methanosphaera . In addition, studies have investigated Taisui from the perspective of its chemical composition. Polyvinyl alcohol has been found as the main component of meat-like Taisui, whereas polyacrylic acid or polyvinyl alcohol was the main component of jelly-like Taisui (Li et al., 2020;Zheng & Dong, 2010). Some researchers believe that Taisui may have healthy benefits such as regulating immunity, inhibiting tumors, delaying aging, and eliminating fatigue. Within Taisui, PQQ (pyrroloquinoline quinone), nucleic acids, trace elements etc., have also been found as effective ingredients (Wang, 2018). In order to obtain the functional substances secreted from Taisui, Taisui has now been successfully artificially cultured, some of which are already on the market. For example, commercial Taisui is usually consumed with water or Chinese liquor, or the mixture of brown sugar and herbs (e.g., wolfberry, astragalus, and jujube). Taisui can also be used in food industry. For example, Lactobacillus and Aspergillus, isolated from Taisui (Dai, 2007;Han et al., 2018), can be potentially used in the brewing and soy sauce industry (Fang et al., 2006).
The chemical and microbial compositions of Taisui vary with the source of Taisui (Li et al., 2015;Wang, 2018). Consequently, the chemical compositions, total carbohydrates, protein, fat, total ash, moisture, and trace elements of Taisui cultured in honey solution have been investigated, as well as the microbial structure and extracellular metabolites of its bacteria, archaea, and fungi in this manuscript. Its aim is to fully understand the safeness of acTS products.

| Material
The mixture of Taisui and honey solution (Figure 1) was kindly gifted by Guangzhou KingCell Co., Ltd. Taisui sample is light yellow, cream, and transparent. It is cultured in honey solution at room temperature (about 25°C) with a ratio of honey to water of 6:1 (v/v).

| Chemical components analysis
Wet acTS was washed by distilled water and then was homogenized in a pulverizer. All tests were performed in triplicate.

| Microbial composition of bacteria, archaea and fungi
Due to the difference in appearance, we divided the artificially cultured Taisui solution into three parts: the upper, the lower, and the medium part (denoted as group U, group L, group M, respectively).
Group U and group L were washed by 75% ethanol solutions and distilled water. All tests were performed in triplicate.
DNA from the three groups was extracted using MN NucleoSpin 96 Soil Kit (Macherey-Nagel). Sequencing was analyzed on the Illumina Mi Seq platform (Illumina MiSeq). The sequences were clustered at a similarity level of 97% (USEARCH, version 10.0) (Edgar, 2013), and the OTUs were filtered with 0.005% of the number of all sequences as a threshold (Bokulich et al., 2013). Microorganisms with less than 0.1% of RA were classified into "Others." Phylogenetic Investigation of Communities by Reconstruction of Unobserved States (PICRUSt) software (version1.0.0 6 ) was used to predict functional genes composition. The obtained OTU was F I G U R E 1 Artificially cultured jelly-like Taisui standardized. According to the unique greengene id corresponding to each OTU, the KEGG family information and the abundance of KEGG were obtained. The abundance of each type of function was obtained from KEGG database.

| Metabolomics analysis
Cultured Taisui honey solution was set as culture group (denoted as group Cul), and the uncultured Taisui honey solution was set as control group (denoted as group Con). All tests were performed in sextuplicate.
Samples were mixed (v/v, 1:5) with 80% methanol aqueous solution and were collected for analysis using a Thermo Scientific Vanquish UHPLC system with a Thermo Hyperil Gold (C18) column coupled with a Mass Spectrometer detector Q Exactive HF-X (Thermo) (Dunn et al., 2011). Compared with mzCloud database, the data recognition and quantitative results were obtained. Metabolites were determined as compounds with the Variable Importance in Projection (VIP) value > 1 and p <.05 and Fold Change value (FC) >2 or <0.5 (Svenja et al., 2016).

| Statistical analysis
Statistical analysis was performed with SPSS software (IBM, version 22.0). Results were presented as mean ± standard deviation. Two-tailed T tests were conducted to compare the difference among different groups. Pearson correlation was performed with SPSS software (IBM, version 22.0), and heat maps were performed with Origin (OriginLab, version 2018). Statistical significance was set at a p value < .05.

| Potential macronutrients and trace elements
As shown in Tables 1 and 2, the total content of carbohydrate was 2.13 g/100 g and was the main nutrient of wet acTS compared with other macronutrient. The protein and fat contents were 0.13 g/100 g and 0.07 g/100 g, respectively. Total ash was the lowest component of solid with a number of 0.04 g/100 g. The moisture concentration was the highest with a number of 88.3%. In total, 15 kinds of trace elements were detected. K was the most abundant (1,424.92 mg/ kg), followed by Ca, P, Fe, Mg, and Al within 11-160 mg/kg. The concentrations of macronutrients and trace elements mentioned above were expressed as wet weight.

| OTU, alpha-diversity indices and Venn diagrams
The number of OTU and the alpha diversity indices are shown in Table 3. The OTU numbers and alpha diversity indices indicated that no significant differences were found between group U, L, and M in terms of bacteria and archaea. For fungi, the OTU number, Shannon index, ACE index, and Chao1 index of group L were significantly higher than those of other two groups, and Simpson index was significantly lower than those of other two groups.
The Venn diagram indicated the shared OTUs and unique OTUs of bacteria, archaea, and fungi respectively. In the Venn diagram, there are 1,411 shared OTUs in bacteria, 15 shared OTUs in archaea and 186 shared OTUs in fungi. The shared OTUs accounted for large part of OTUs ( Figure S1).

| Composition of microbiota at the phylum level
The microbial community structure of bacteria, archaea, and fungi at the phylum level was presented in Figure 2, and the corresponding relative abundance (RA) was listed in Table S1.
The RA of Proteobacteria was the highest in L, U, and M groups, accounting for 43.10%, 38.73%, and 41.70%. Followed by Chloroflexi, Firmicutes, Acidobacteria, Actinobacteria, Bacteroidetes, and Nitrospirae, these phyla of bacteria were the micro part of bacteria and their RAs were 15%-4%.
The dominant phylum of archaea among 3 groups was Euryarchaeota (>88%), while Crenarchaeota and Diapherotrites only accounted for a low ratio of the community < 10%), respectively. However, only Ascomycota was predominant among fungi with higher than 46% RA among 3 groups. The RA of Rozellomycota was 19.09% in group U. The RA of Basidiomycota and Mortierellomycota was 7.47% and 3.45% in group U, respectively, and was 6.05% and 2.09% in group M. The RA of Basidiomycota only accounted for 1.77% in group L.

| Composition of microbiota at the genus level
The microbial community structure of bacteria, archaea, and fungi at the genus level was presented in Figure 3 while the top 10 relative abundance (RA) was listed in Table S2

| Functional genes prediction
Based on KEGG database, PICRUSt analysis is used to predict functional genes composition by comparing species composition obtained from the sequencing data. As shown in Figure 4, it was found that the functional genes of Metabolism was the most dominant category. The KEGG level 2 result showed that group U, L, and M had high abundance of carbohydrate metabolism, amino acid metabolism, energy metabolism, metabolism of cofactors and vitamins, and nucleotide metabolism among bacteria and archaea. The abundance of those functional genes categories respectively accounted for more than 27% and 30% of the bacteria and archaea. However, group U, L, and M had similar abundance in the same category and showed no significant difference among three groups.

| Extracellular metabolites
The metabolites in group Cul were analyzed. The KEGG pathway and related metabolites were presented in Figure S2 and Table 3. A total number of 720 metabolites were detected, among which 311 metabolites were upregulated and 178 metabolites were downregulated. The FC of some metabolites was showed in Table S3.
Most metabolites were related to environmental information processing, metabolism, and/or organismal systems, whereas only a small part were related to drug development and/or genetic infor- Aldehydes, such as cuminaldehyde, 3,4-dihydroxybenzaldehyde, and phenylacetaldehyde.

| Pearson correlation between the microbes and metabolites
Pearson correlation heat map showed the correlation between 10 microbes with the highest abundance and the 10 metabolites with highest abundance and five metabolites with lowest abundance ( Figure 5).   Our results showed that the acTS contains 2.13% carbohydrate, lower than that in ordinary food. The protein content of Taisui was significantly lower than the normal wild type of wet Taisui and also the standard of "high protein food" (12%) (Zhang Tao et al., 2018). In the meantime, the fat content was also significantly lower than that of the normal wet wild-type Taisui and the standard of "low fat food" (3%) (Chen et al., 2010). The concentration of total ash and trace elements was far lower than the normal concentration range of wet wild-type Taisui (Zhu et al., 2011). In short, the acTS used in current study is a substance with low carbohydrate, low protein, low fat, low total ash, low trace elements, and high moisture, while wild Taisui is a substance with high protein. The differences between these two types of Taisui may due to their culture environment. Since all of carbohydrate, protein, fat, total ash, and trace elements were low, which means acTS is not a good source of macronutrients and trace elements.
In order to investigate whether or not it is safe to consume acTS as food ingredient, we have thus investigated the composition of microorganism in the acTS. In terms of bacteria, Proteobacteria contains Alcaligenes, Ochrobactrum, Methylotenera, and Ralstonia, and Lactobacillus is a well-known probiotic parasitizing in intestine and vagina (Setiarto et al., 2017). This acTS contains potential pathogenic bacteria and probiotics. However, it cannot infer that there is a safety risk in this acTS, because the safeness is not consistent for different bacterial strains. (Arellano et al., 2020).
In terms of archaea, Euryarchaeota contains many species includ- and Aspergillus glaucus are widely used in fermentation (Deng et al., 2009;Fang et al., 2006;Ishchuk et al., 2016;Zhang et al., 2016;Zhu et al., 2003). Dai (2007) reported that A. glaucus was the dominant fungi in wild Taisui. Lactarius, belonging to Chordata, can produce chemicals such as sesquiterpenes and has the activity of anti-tumor (Barros et al., 2007). Fortunately, these fungi are relatively safe because they are widely present in traditional foods. In addition, there are some commonly used industrial species in acTS, such as Mortierella for n-6 polyunsaturated fatty acids production, Pichia for protein, Chaetomium for cellulase, and Penicillium for penicillin (Ahmad et al., 2014;García-Estrada et al., 2020;Ho et al., 2007;Sun et al., 2019). The safeness of some genera of fungi in acTS is still undefined. Several types of Alternaria could cause diseases, but it could produce anticancer drugs such as vinblastine (Duan et al., 2008). Some genera of fungi are relatively unsafe. For example, Fusarium could produce toxin (Geiser et al., 2004); some species of Cladosporium can cause allergy (Bensch et al., 2012); Candida could cause inflammation (Tarang et al., 2020) and Malassezia could cause dandruff (Sommer et al., 2015). So, this acTS contains potential pathogenic and probiotics. However, it also cannot infer that the   (Lim et al., 2008). Salicylic acid is pharmaceutical intermediates, which is used for anti-inflammatory medicine like aspirin (Su et al., 2017). In conclusion, most of the metabolites are beneficial to human.
The KEGG pathways of functional genes prediction and extracellular metabolites were consistent. Pearson correlation analysis showed that some microbes are significantly associated with the above metabolites. According to previous report, Aspergillus is one of the most important strains commercially produced citric acid by starch or sucrose-based medium fermentation (Aboyeji et al., 2020).
It is known that honey contains a high amount of sucrose and thus is a good source for Aspergillus to produce citric acid. As reported, Candida could also be potentially used to produce citric acid (Uzah et al., 2020). Aspergillus produces D-biotin with a structure similar that of desthiobiotin, suggesting that Aspergillus may potentially to produce desthiobiotin (Zheng, 2007). Candida can utilize its own lipase to catalyze phenylethylamine synthesis from lipid substrate (Wen, 2012). Except for Candida, the main organisms commonly used to produce stereoselective lipases are Pseudomonas, Fusarium, and Aspergillus (Qin, 2006). Therefore, microbes may also be responsible for the production of metabolites. This study indicates that the acTS used in current study could be potentially developed into fermented food, functional food, or oral liquid.

| CON CLUS ION
In current study, the chemical compositions, the microbial structure of bacteria, archaea, and fungi, and the extracellular metabolites of acTS were analyzed. The concentrations of macronutrient and trace elements were low, indicating that acTS will not be a good source of macronutrients and trace elements. Microbial composition of bacteria, archaea, and fungi, and metabolomics analysis showed that acTS can provide some probiotics and some beneficial metabolites for human. It is therefore possible to develop acTS it into fermented food, functional food or oral liquid, only if when safeness of each microorganism is proved.

ACK N OWLED G M ENTS
The program was supported by the funds of Key-Area Research and Development Program of Guangdong Province (Nos. 2020B020226005 and 2020B020226008). The authors thank Ruixia Qiu, Bing Yu and Shulin Deng from the Department of Food Science and Engineering, Jinan University, for their contributions to this study. We would also like to thank Biomarker Technologies Co., Ltd., (Beijing, China) for the analysis of Microbial Community Diversity, and Novogene Co., Ltd., (Beijing, China) for the analysis of Non-targeted Metabolomics.

CO N FLI C T O F I NTE R E S T
All authors declare that they have no conflicts of interest.