Matrine-producing endophytic fungus Galactomyces candidum TRP-7: screening, identification, and fermentation conditions optimization for Matrine production

Matrine (MA) is an alkaloid extracted from the root of genus Sophora with various pharmacological activities. Production of MA by endophytic fungi offers an alternative challenge to reduce the massive consumption to meet the increasing demand of MA. In the current study, the positive strains with MA producing ability were screened from endophytic fungal isolated from the root of Sophora tonkinensis Gagnep. Chromatographic analyses verified the identity of the produced MA. Among these fungi, Galactomyces candidum strain TRP-7 was the most valuable strain for MA production with the initial yield 8.26 mg L−1. The MA production was efficiently maximized up to 17.57 mg L−1 of fermentation broth, after optimization of eight process parameters using Plackett–Burman and Box-Behnken designs. The statistical optimization resulted in a 1.127 times increase in MA production as compared to the initial yield of TRP-7. This is the first report to isolate endophytic fungi with MA-producing activity from S. tonkinensis Gagnep., and to identify an endophytic fungus G. candidum TRP-7 as a new promising start strain for a higher MA yield.


Introduction
Matrine (MA), a naturally occurring tetracyclic quinolizine alkaloid, has received intensive attention due to its extensive pharmacological activity, including anticancer (Chang et al. 2013;Lin et al. 2018), antiviral (Sun et al. , 2020, anti-inflammatory (Jiang and Wang 2020;Li et al. 2021;Pecoraro et al. 2021) and antioxidant effects (Zhao et al. 2015;Liu et al. 2017;Zhang et al. 2018). It is also used to control numerous pests of various crops, in addition to being widely utilized in pharmacy (Liu et al. 2007). Thus, it is required broadly in clinical medication and biological pesticides.
Currently, the main source of MA is isolation from the roots of Sophora specie (Chen et al. 2004). According to previous report, these plant species grow slowly and provide a low MA yield. Moreover, the removal of roots for extraction practices lead to the death of the tree and diminish the natural resource. Previous trials for MA production by tissue culturing (Gao et al. 2016) or chemical synthesis processes (Mandell et al. 1965;Chen et al. 1986) showed some disadvantages and limitations.
Endophytic fungi are microorganisms colonizing internal plant tissues at a certain period but do not cause obvious infection of host plant tissues (Schulz and Boyle 2005;Tešanović et al. 2017). Studies have shown that endophytic fungi of medicinal plants can synthesize the same or similar secondary metabolites as the host plants (Chen et al. 2010;Morales-Sánchez et al. 2021). Stierle et al. (1993) isolated the first endophytic fungus that can produce taxol from Taxus brevifolia, opening up a new biotechnological investment area. Recently, some endophytic fungi isolated from plants of Sophora species were found to produce MA, representing an easily manipulated and alternative source to eliminate the excessive harvesting of Sophora species. The previously reported MAproducing endophytic fungal genera were Simplicillium (Yu et al. 2013), Acremonium (Mao et al. 2015) and Aspergillus . Regrettably, the MA yield produced by these microorganisms is too low for mass production. Thus, screening MA-producing positive strains from different Sophora plants is necessary to develop a new technique for MA manufacturing.
In a previous study, a total of 655 fungal strains representing 47 taxa were isolated (Yao et al. 2017).
This study aimed to screen MA-producing endophytes from these fungal strains isolated from the Chinese medicinal plant S. tonkinensis and enhance the MA yield from the endophytic strain through optimization of the fermentation conditions by the combination of the Plackett-Burman design, the steepest ascent experiment and the response surface design.

Chemicals and strains
Standard MA (99%, purity) was purchased from Solarbio Co. (Beijing, China). The solvent (acetonitrile) used in the high-performance liquid chromatographic (HPLC) analysis was of HPLC grade. The tested strains were saved by the Research Group of Medicinal Plants, College of Agriculture, Guangxi University.

Culturing and alkaloid extraction
The disk of the strain (5 mm size) was inoculated in 100 mL potato dextrose broth (PDB) liquid medium and cultured at 28 °C for 7 days. After incubation, the fermentation broth and mycelia were separated by vacuum filtration. Alkaloids were extracted by the acid-water method described by Jian et al. (2013) with some modifications. The samples were extracted overnight with twice the volume of fermentation broth in absolute ethanol and sonicated for an hour in a 45 °C water bath. The pH of the extract was adjusted to 3-4 with 1 M HCl after filtration and sonication for an hour in a 45 °C water bath. Then, the pH was raised to 9 with 1 M NaOH. One hour later, the water phase was extracted three times with equal volumes of chloroform. The mycelium was dried at 48 °C, ground into powder, added to twice the volume of chloroform, and then sonicated for 30 min in a 45 °C water bath for 3 cycles. Finally, the mycelium extract was filtered, and the residue was discarded. Thereafter, the chloroform extracts were evaporated under vacuum at 100 rpm in a 48 °C water bath using a rotary evaporator and dissolved in 1.0 mL methanol.

Thin-layer chromatography
Chloroform extracts that showed MA-producing activity were used for MA production on GF254 silica gel plates using a standard sample of MA (100 μg mL −1 ) as described by Bai et al. (2006). A run solvent consisting of chloroform/methanol/ ammonia (9:1:0.2, v/v/v) was applied. MA spots were detected under an ultraviolet lamp (254 nm) and as orange-red spots after spraying with diluted iodized bismuth potassium solution spray reagent.

HPLC analysis of MA
The identity and concentration of the produced MA from the positive endophytic fungal strain TRP-7 was also confirmed using HPLC (Shimadzu LC-40 with a multichannel full wave PDA (photodiode array) detector using an X-Bridge C18 Waters column (4.60 mm × 250 mm, 5 μm). The chromatography conditions were described by Yin et al. (2013) with slight modifications. The mobile phase consisted of acetonitrile and phosphate solution (triethylamine 2 mL L −1 ) by gradient elution: 0-6 min, acetonitrile 6%; 6-35 min, acetonitrile 6-12%; 35-45 min, acetonitrile 12-6% at a 1.0 mL min −1 flow rate. The injected volume was 10 μL, and the wavelength of the detector was adjusted to 210 nm. The MA concentration was estimated using a standard curve of MA with a concentration range of 20-100 μg mL −1 .
Morphological and molecular identification of the MA-producing strain The colony morphology and cultural characteristics of the MA-producing endophytic fungal TRP-7 were examined on potato dextrose agar (PDA). The MA-producing endophytic fungal isolate TRP-7 was morphologically examined for cultural and microscopic characteristics according to Moubasher (1993). Molecular identification was also carried out depending on the 18S rRNA sequence analysis. Fungal genomic DNA was extracted by a DNA Kit (EZgene™ fungal gDNA miniprep kit). The 18S ribosomal RNA genomic sequence was intensified through PCR (polymerase chain reaction) using ITS1 and ITS4 as universal specific fungal primers, ITS1 (5′-TCC GTA GGT GAA CCT GCG G-3′) and ITS4 (5′-TCC TCC GCT TAT TGA TAT GC-3′) (Zaki et al. 2019). The PCR product (the amplified ITS region) was run in 1% agarose gel electrophoresis, and the gel bands underwent two-directional ITS sequencing by sending to Sangon Biotech Co. (Shanghai, China). The TRP-7 isolate was identified to the genus and species levels by comparing its ITS sequence with those in the NCBI and GenBank databases. A phylogenetic tree was designed using the MEGA 7 program. The neighbour-joining (NJ) method was used to minimize the total distance of the system tree by determining the nearest (or adjacent) paired classification units (Zhang and Sun 2008).

Screening of process variables by Plackett-Burman design (PBD)
The Plackett-Burman design (PBD) was used to screen factors that significantly affect biomass production (Gu et al. 2018). According to the fermentation characteristics of TRP-7, temperature, pH, inoculum amount, incubation period, liquid loading, KH 2 PO 4 concentration, MgSO 4 concentration and NaCl concentration were selected as variables (Table 1). In PBD, each factor is set with high and low levels (1, − 1), and the difference between the high and low levels of a factor cannot be too large. The independent variables temperature, pH, inoculum, incubation period, media amount, KH 2 PO 4 concentration, MgSO 4 concentration and NaCl concentration were coded as A, B, C, D, E, F, G and H, respectively. All experiments were repeated three times. The predicted response (Ŷ) was coded as follows: where Ŷ is the predicted response (biomass concentration) and b 0 , b 1 , b 2 , b 3 , b 4 , b 5 , b 6 , b 7 and b 8 are (Uncu and Cekmecelioglu 2011).
The experiment was designed by using Design-Expert 12.0 software. The statistical software MINITAB 20.0 was used to analyse the experimental design.
The steepest ascent experiment The steepest ascent experimental method was used to find and further analyse the factors having the greatest impact on the fermentation of products (Zhang et al. 2012). Based on the first-order equation obtained from PBD, the step with the factor ratio of the corresponding factors is taken as the benchmark until the step moves to the highest point; then, the range near the highest point is taken as the corresponding range of response surface optimization using the steepest rise method test. The coefficient of each variable in the fitting equation determines the climbing direction and change step. The test was repeated 3 times to obtain the average value and find the central point of the response surface test.

Box-Behnken design
According to the factors obtained from the Plackett-Burman design test, the centre point was obtained from the steepest ascent test, and the best level of the response surface test was designed according to the central combination design principle. A second-order polynomial equation was related to biomass production as follows: where Y is the response (biomass concentration), b 0 is the intercept, b i is the coefficient of the linear effect of the key factor, b ii is the coefficient of the secondary effect of the key factor, b ij is the coefficient of interaction between key factors, e is the correction factor, and x i and x j are independent variables used in this study. The optimum conditions for maximizing biomass production were determined by using the Response Optimizer Tool in MINITAB20.0.

Validation tests for the optimization results of culture conditions
The MA-producing strain TRP-7 was grown under optimized fermentation conditions, and the production of MA was measured at the end of the fermentation.

Statistical analysis
All experiments were done in triplicate. The Plackett-Burman design and the BBD response surface optimization test were carried out by using Design Expert ver. 12.0 software. One-way ANOVA was used to test all individual data groups to determine the significant differences using the IBM SPSS ver. 22 statistics software. A Fisher's least significant difference (LSD) multiple comparison test, at P < 0.05, was applied to determine differences among data groups. To construct the neighbour-joining tree and to calculate the bootstrap values, MEGA 7 software was applied.

Screening fungal endophytes for Matrine production
Endophytic fungal strains were screened for Matrine production using PDA medium as a cultivation medium. By screening the MA-producing activity of the chloroform extracts from the endophytic cultures isolated from the root of S. tonkinensis, the obtained results in Table 2 show that a total of 2 endophytic fungal isolates (TRP-7 and TRP-95) belonging to Galactomyces sp. and Phoma sp. evidenced MAproducing activity. Among them, the Ma yield of the endophytic fungus isolate TRP-7 was the highest, at 8.26 mg L −1 . Isolation of endophytic Galactomyces sp. from S. tonkinensis with MA-producing activity is first mentioned in this study.
The presence of MA in the extracts of the TRP-7 strain was confirmed by TLC and HPLC analysis. Both TLC and HPLC analysis, as illustrated in Fig. 1a-c, confirmed that the endophytic fungal isolate TRP-7 was an MA producer. Moreover, the PDA absorption spectrum of MA in chloroform extracts matched that of authentic MA with a maximum absorption at 201 nm. As demonstrated, the produced MA showed a retention factor (R F ) of 0.28 on the plate and a retention time (R T ) of 18.929 min by HPLC, similar to the standard MA sample under the same conditions (Fig. 2).
Morphological and molecular identification of the TRP-7 strain Colony morphology of the TRP-7 strain was observed on PDA medium for 7 days at 28 °C. The colony diameter was approximately 6.3 cm, and the strain was identified as Galactomyces candidum. It is characterized by the mycelium growing creeping, the conidia stalk being erect and short, and there being strings of conidia on the top; the conidia are colourless or light and cylindrical. Conidia dimensions: 5-10 × 10 μm.
Additionally, identification of the TRP-7 strain was confirmed by molecular studies as PCR amplification of the 18S-rRNA coding gene sequence (1316 bp), and compared with sequences recovered from the GenBank database using BLAST programs, the TRP-7 strain was identified as a G. candidum strain. Based on the ITS1-5.8S-I TS2 region alignment of the TRP-7 strain, a phylogenetic tree (Fig. 3d) was built. The identified sequence was deposited under accession number MZ148443 into GenBank.

Plackett-Burman design (PBD)
The MA yield (mg L −1 ) obtained from the PB model trials is demonstrated in the PBD matrix (Table 3). The obtained data were statistically analysed as in Table 4 and represented graphically as in Fig. 4a and b. As shown, there was a variation in the MA concentrations from 6.02 mg L −1 (run 1) to 12.97 mg L −1 (run 4), supporting the importance of the studied variables. The maximum MA yield (12.97 mg L −1 ) was achieved when the production process was carried out using PDA medium containing potato extract 300 g L −1 , glucose 20 g L −1 , KH 2 PO 4 2 g L −1 , MgSO 4 0.5 g L −1 , NaCl 0.7 g L −1 , an incubation period of 13 days, inoculum amount 5% (v/v), liquid loading 80/250 mL, pH 5, and temperature 30 °C. The influence of variable factors on MA production was as follows: E (liquid loading) > D (incubation period) > A (temperature) > B (pH) > C (inoculum amount) > F (KH 2 PO 4 concentration) > G (MgSO 4 concentration) > H (NaCl concentration).
Only 3 variables, liquid loading, fermentation time and temperature, significantly affected MA production (P < 0.05). The effect of liquid loading was negative, while the effects of fermentation time and temperature were positive. The determination coefficient (R 2 ) was 0.9730, indicating goodness of fit, as the regression equation could explain 97.30% of the total variations. The PB model provided a linear polynomial correlation equation describing the correlation between the 8 studied factors and the response (MA yield) as follows:

Results of the steepest ascent experiment
According to the linear polynomial correlation equation obtained from the Plackett-Burman test, the MA yield mg L −1 = 9.682 + 0.790A − 0.508B was negative, so we reduced its actual value in the steepest ascent test. The MA yield (mg L −1 ) obtained from the steepest ascent trials is presented in Table 5. As shown, the Fig. 2 The chromatographic analyses of a standard MA and the extracted sample from the endophytic fungal isolate G. candidum TRP-7. a Is the TLC plate image after spraying bismuth potassium iodide reagent (I: Sample; II: Mixed standard of MA and OxyMa). b HPLC chromatogram of the standard and PDA spectral analysis of the standard MA. c Is the HPLC of MA from G. candidum TRP-7 and the PDA spectral analysis of G. candidum TRP-7. The samples were dissolved in 1 mL of HPLC-grade methanol, and a 10-μL portion was injected into the HPLC under the conditions described in the "Materials and methods" section. A peak with a retention time of 18.9 min was detected in the fungal sample and standard Fig. 3 Morphological and molecular identification of the endophytic G. candidum TRP-7. a and b Are the front and reverse colony views, respectively, on PDA media at 28 °C for 7 days. c Shows the microscopic features at × 40. d Shows the neighbour-joining phylogenetic tree based on the sequence of PCR-amplified ITS1-5.8S-ITS2 rRNA gene analysis. All sequence data were retrieved from the NCBI and GenBank databases maximum MA yield (17.79 mg L −1 ) was achieved when the production process was carried out using PDA medium containing potato extract 300 g L −1 , glucose 20 g L −1 , KH 2 PO 4 1 g L −1 , MgSO 4 0.5 g L −1 , NaCl 0.7 g L −1 , an incubation period of 11 days, inoculum amount 5% (v/v), liquid loading 100/250 mL, pH 5, and temperature 27 °C. Therefore, the central point of the Box-Behnken design (BBD) matrix of the three selected significant variables was temperature (27 °C), fermentation time (11 days), and liquid loading (100/250 mL). Table 6 shows the Box-Behnken design (BBD) matrix of the three selected significant variables, temperature (A), fermentation time (B), and liquid loading (C), with their coded values and experimental data of MA yield (mg L −1 ) drawn from the design trials. The regression coefficients for biomass production are presented in Table 7. The model correlation coefficient (R 2 ) value equals 0.9973, the adjusted R 2 equals 0.9939, and the predicted R 2 equals 0.9706, supporting the high correlation and goodness of fit. The insignificant lack of fit (P = 0.1937 > 0.05) also demonstrated that the model fit well to the experimental data. By studying the interaction effects between the selected variables, it was found that the variable temperature showed significant effects (P < 0.05), and the variables fermentation time and liquid loading showed insignificant effects (P > 0.05). The interaction between temperature and liquid loading was significant (P < 0.05), whereas the interactions between temperature and fermentation time and fermentation  where A, B, and C were uncoded variables for temperature, fermentation time and liquid loading, respectively. The response surface plots described by the regression model were plotted for better understanding (Fig. 5). All surface plots demonstrated that as the level of variables increased, biomass production increased up to a certain level. In Fig. 5a, the biomass concentration demonstrated a nonlinear effect with increasing liquid loading from 90 to 110 mL/250 mL and with increasing temperature from 25 to 29 °C MA yield mg L −1 = 17.51 + 1.43A + 0.0363B

Box-Behnken design
under constant fermentation time. An increase in temperature with a simultaneous increase in fermentation time (liquid loading constant) led to an increase in biomass production until they reached their optimal values (Fig. 5b). In Fig. 5c, the fermentation time showed a nonlinear effect on biomass production. As the fermentation time was increased to 11 days, the biomass concentration increased to approximately 17 mg L −1 . The maximum response value was predicted by Design-Expert 12 software. The optimum conditions for biomass production predicted by the model were found as a fermentation time of 11.04 days and liquid loading of 101.008 mL/250 mL at 28 °C, and the maximum predicted yield of MA was 17.614 mg L −1 .  To verify the reliability of the response surface design results, the medium was prepared with the optimal combination of the main factors determined by the response surface experiment for shake flask fermentation and repeated 3 times, and the results of the original medium were used as the control. The results showed that under optimal conditions, the yield of MA was 17.57 ± 0.36 mg L −1 , which was basically consistent with the predicted value (17.614 mg L −1 ).

Discussion
The study of plant endophytic fungi began in 1898, when Vogl isolated the first endophytic fungus from ryegrass seeds (Zhao et al. 2020). The first paclitaxelproducing endophytic fungus was obtained from the phloem of Taxus brevifolia (Stierle et al. 1993), which initiated extensive subsequent research on the endophytic fungi of medicinal plants. Studies have shown that most endophytic fungi of medicinal plants can produce the same or similar medicinal active compounds with the host plants (Zaki et al. 2019;El-Sayed et al. 2020;Li et al. 2022;Lin and Huang 2022;Tang and Huang 2022a, b). In a previous study, some endophytic fungi were isolated from S. tonkinensis Gagnep. that demonstrated inhibitory activity against fungal phytopathogens of Panax notoginseng (Yao et al. 2017), although there was no report on the production of medicinal active compounds. Despite the enormous therapeutic potential of MA, its high production costs have prevented its largescale production. This study signifies the potential of G. candidum TRP-7 as a viable replacement for MA production. The endophyte employed in this study, G. candidum TRP-7, is a filamentous yeast-like fungus (Thornton et al. 2010). Studies have shown that G. candidum is usually used as a plant pathogen, which can affect tomatoes, potatoes and other fruits and vegetables (Su-qin et al. 2020;Paes et al. 2021). The G. candidum strain screened in this study showed MA production using TLC and HPLC. This is also the first report that the endophytic fungus G. candidum has the ability to produce MA. Although optimization of MA production from endophytes is an important strategy that is recommended for enhancing MA yield from endophytes, there are few studies on optimizing endophytic MA production Zhou et al. 2022). Increasing the yield of metabolites of strains can be achieved by improving the fermentation and culture conditions of microorganisms. The Plackett-Burman design is an economical and effective experimental design method that can quickly and effectively screen out the factors that have a significant impact on MA production from many factors (Huang et al. 2019;Wyantuti et al. 2021;Tang and Huang 2022a, b). The Box-Behnken design is one method in response methodology that can identify the variable condition located at the midpoint edges of the variable space as well as at the centre (Anwar et al. 2015). A Box-Behnken design (BBD) model was applied to optimize the composition for the endophytic Aspergillus terreus medium to achieve a maximum yield of 20.63 μg L −1 MA by Zhang et al. (2017). Only three variables, temperature, fermentation time and liquid loading, were selected from the currently investigated PB model as the key significant variables for further investigation of their interaction effects on MA yield and determination of their optimum levels using response surface methodology represented in a BBD. The predicted MA yield from BBD was higher than that obtained from the PB model, with a 19.2% increase, reflecting the significant interaction effects of the three selected key variables.
It is widely known that different microorganisms need appropriate fermentation conditions to produce metabolites to their own strain characteristics (Al-Ghazali and Omran 2017). Therefore, research on the optimization of fermentation processes at home and abroad mainly focuses on the selection of different components of culture media, such as carbon sources, nitrogen sources and trace elements, as well as the optimization of fermentation conditions, including temperature, pH and dissolved oxygen (Zhou et al. 2014;Posada-Uribe et al. 2015;Thi Nguyen et al. 2018; Bandara 2019). Zhang et al. (2017) optimized the carbon source and nitrogen source of the MAproducing strain A. terreus, and the results showed that potato starch and glucose were better suited for A. terreus to produce MA than others. Therefore, PDB medium was chosen as the basic medium and enhanced MA production by optimizing the fermentation conditions of the endophytic fungi G. candidum TRP-7. The optimum fermentation levels of MA production by the fungus G. candidum TRP-7 were determined by using response surface methodology represented in a BBD. The fermentation conditions were a liquid loading of 101.008/250 mL and culturing at 28 °C for 11.04 days. Temperature is one of the important conditions for ensuring the growth of microorganisms (Burman and Bengtsson-Palme 2021). The main reason is that high temperature will denature or coagulate proteins inside microbial cells, while it also destroys the enzymatic activity inside microbial cells, which kills the microorganisms, and low temperature will inhibit microbial growth. Obtaining the maximum MA production at higher temperatures is in agreement with the data reported by Dubey and Behera (2011). The optimum incubation temperature reported for 3-DMC alkaloid production by endophytic Bacillus megaterium ACBT03 was 28 °C, and production decreased by elevating the temperature to 30 °C (Dubey and Behera 2011).
In this study, the optimum fermentation time for maximum MA production was determined to be 11 days, and the yield of MA decreased gradually when the fermentation time exceeded 11 days. This is due to the growth of MA-producing microorganisms caused by nutrient limitation and the change in pH in the culture media or inhibitory end products (Papagianni 2004;Asgher et al. 2006). On the other hand, it may be due to biotransformation during the growth of strain G. candidum TRP-7, and MA was transformed into other intermediate products, which may have affected the growth of strain G. candidum TRP-7 and the accumulation of MA. The effect of fermentation time on DNJ alkaloid yield from Streptomyces lavendulae TB-412 was previously studied by Wei et al. (2011), who reported that the optimal fermentation time for DNJ alkaloid production was 11 days. This result indicates that fermentation time was also an important factor affecting microbial fermentation.
Dissolved oxygen is another important factor in fermentation; only by providing enough oxygen can the normal growth and metabolism of the strain be maintained (Huang and Tang 2007). Dissolved oxygen not only has a significant impact on the synthesis pathway of secondary metabolites but also has a certain impact on the synthesis rate of metabolism (Pfefferle et al. 2000). Due to high cell density and large oxygen uptake, changing the liquid loading and shaking speed to is generally used to increase the dissolved oxygen in shaking flask fermentation (Subrahmanyam et al. 2010;Al-Askar et al. 2014). According to the research results of Long et al. (2020), 50/250 mL is the best liquid loading in the production of epothilone B. When the liquid loading is less than 50 mL, the yield is not high due to the lack of nutrients in the fermentation system; when the liquid loading is greater than 50 mL, the yield decreases due to the limitation of oxygen supply in the fermentation system. The yield of MA reached its peak when the liquid loading was 100/250 mL in this study, and the result is in accordance with that of Jia et al. (2017), who reported 100 mL liquid loading in a 250 mL triangular flask as optimal for TMP alkaloid production by the Monascus strain.
The target G. candidum strain TRP-7 was identified as G. candidum by morphological and molecular identification. The liquid state fermentation conditions of G. candidum strain TRP-7 were further determined by the PB design and BBD of RSM: PDA medium containing potato extract 300 g L −1 , glucose 20 g L −1 , KH 2 PO 4 1 g L −1 , MgSO 4 0.5 g L −1 , NaCl 0.7 g L −1 , culture temperature of 27 °C, fermentation time of 11 days, inoculum amount 5% (v/v), liquid loading of 100/250 mL, and pH 5. Under the optimum liquid state fermentation conditions, MA production could be as high as 17.614 mg L −1 . The final yield of MA is far from reaching the standard of industrial production, and it is necessary to improve the yield of MA by physical or chemical mutagenesis. In addition, the specific metabolic pathway by which G. candidum produces MA needs to be further studied.

Conclusions
In summary, G. candidumis strain TRP-7 with the MA production ability was screened from S. tonkinensis from different areas of Guangxi province in China. The target G. candidumis strain TRP-7 was identified as G. candidumis by morphological and molecular identification. The liquid state fermentation conditions of G. candidumis strain TRP-7 were further determined by the PB design and BBD of RSM: PDA medium containing potato extract 300 g L −1 , glucose 20 g L −1 , KH 2 PO 4 1 g L −1 , MgSO 4 0.5 g L −1 , NaCl 0.7 g L −1 , culture temperature of 27 °C, besides fermentation time of 11 days, inoculum amount 5% (v/v), liquid loading of 100/250 mL, and at pH 5. Under the optimum liquid state fermentation conditions, the MA production could be as high as 17.614 mg L −1 .
This is the first report on the optimization of fermentation conditions of MA production, and the study is expected to provide a new approach of MA production through biosynthesis by G. candidumis strain. The final yield of MA is far from reaching the standard of industrial production, it is necessary to improve the yield of MA by physical or chemical mutagenesis. In addition, the specific metabolic pathway of G. candidumis producing MA needs to be further studied.

Data availability
The datasets generated during and/or analysed during the current study are available from the corresponding author on reasonable request.

Conflict of interest
The authors have no relevant financial or non-financial interests to disclose.