RETRACTED ARTICLE: Mutagenesis combined with fermentation optimization to enhance gibberellic acid GA3 yield in Fusarium fujikuroi

Abstract

Gibberellic acid (GA3) is a plant growth hormone that plays an important role in the production of crops, fruits, and vegetables with a wide market share. Due to intrinsic advantages, liquid fermentation of Fusarium fujikuroi has become the sole method for industrial GA3 production, but the broader application of GA3 is hindered by low titer. In this study, we combined atmospheric and room-temperature plasma (ARTP) with ketoconazole-based screening to obtain the mutant strain 3-6-1 with high yield of GA3. Subsequently, the medium composition and fermentation parameters were systematically optimized to increase the titer of GA3, resulting in a 2.5-fold increase compared with the titer obtained under the initial conditions. Finally, considering that the strain is prone to substrate inhibition and glucose repression, a new strategy of fed-batch fermentation was adopted to increase the titer of GA3 to 575.13 mg/L, which was 13.86% higher than the control. The strategy of random mutagenesis combined with selection and fermentation optimization developed in this study provides a basis for subsequent research on the industrial production of GA3.

Graphical Abstract

Introduction

Gibberellic acids (GAs) are one of the 5 major groups of plant hormones, along with growth hormone, cytokinin, abscisic acid, and ethylene (Keswani et al. 2022). To date, more than 136 types of gibberellins with different structural compositions have been isolated from plants, fungi, and bacteria, and were numbered according to the order of their discovery (Dilek Tepe 2021). However, only a few gibberellins have known biological activities, such as GA1, GA3, GA4, and GA7 (Hedden and Thomas 2012). Among them, GA3 has attracted the most attention due to its strong activity and wide application prospects for regulating the dormancy of plants as well as promoting the growth of stems and fruits (Mander 2003; Boemke and Tudzynski 2009). Therefore, GA3 was applied in many fields, including agriculture, forestry, and brewing industry with good economic value and social benefits (Kildegaard et al. 2021). Traditional production methods of GA3 mainly include plant extraction, chemical synthesis, and microbial fermentation (Gokdere and Ates 2014). Extraction from plants was quickly abandoned due to its low yield, large amounts of waste, and high costs (Hedden and Sponsel 2015). In addition, the method of chemical synthesis was limited by the large number of steps and resulting pollution (Rodrigues et al. 2012).

With the rapid development of synthetic biology, microbial production of GA3 has attracted increasing attention. At present, GA3 can be produced by many fungi, such as F. fujikuroi (Shukla et al. 2003), Aspergillus niger (Cihangir 2002), and Fusarium oxysporum (Tsavkelova 2016). Among them, F. fujikuroi is the main microorganism used for GA3 industrial production. There exists an entire GA gene cluster in F. fujikuroi, including geranylgeranyl diphosphate synthase (GGS2), the bifunctional ent-copalyl diphosphate/ent-kaurene synthase (CPS/KS), and four cytochrome P450 monooxygenases (P450-1 to P450-4) for GA3 biosynthesis. Also, homologs of the entire F. fujikuroi GA cluster were also present in Fusarium moniliforme (Salazar-Cerezo et al. 2018), which is another strain suitable for industrial production of GA3. However, we found the highest GA3 production of F. moniliforme (7.34 g kg⁻¹, similar to the production of F. fujikuroi) was obtained by solid-state fermentation (SSF), which was not applicable to large-scale industrial production of GA3 (de Oliveira et al. 2017). Therefore, the industrial production of GA3 completely depends on the fermentation of F. fujikuroi due to its natural productivity and controlled fermentation process.

Improving the product titer is one of the main challenges in the development of microbial natural products (Shi et al. 2017a, b), in which both the production strain and external conditions play important roles. However, there are few reports of metabolic engineering in F. fujikuroi due to limited genetic manipulation tools. Gene modification of F. fujikuroi developed slowly, which mainly depended on homologous transformation (Fernandez-Martin et al., 2000). Until 2017, novel technologies, such as CRISPR/Cas9 system, have been developed in F. fujikuroi (Shi et al. 2017a, b). However, CRISPR/Cas9 was operated complexly and less reported at present. In recent years, microbial mutation breeding techniques have been regarded as an efficient tool to improve phenotypes or functions of strains, thus increasing products titer (Kodym and Afza 2003; Zhang et al. 2020). For example, mutants with different morphologies and mycelium colors were obtained by UV mutagenesis. Among them, the mutant morl-189 with shorter hyphae was selected for further optimization of culture conditions and the concentration of GA3 reached 114 mg/L (Lale and Gadre 2010). Similarly combined mutagenesis using both nitrosoguanidine and ⁶⁰Co gamma rays, along with screening for resistance against the fungicide terbinafine resulted in a 11.87% increase of the GA3 titer in the best strain (Wang et al. 2014). ARTP caused DNA damage by chemical active particles with the advantages of lower capital costs, higher mutation rates, and low treatment temperatures compared with traditional UV and chemical mutagenesis methods (Lu et al. 2011; Peng et al. 2020). As reported, ARTP has already applied for obtaining dominant positive mutants from bacteria, fungi, and microalgae (Zhang et al. 2015), which is a potential method for increasing GA3 titer in F. fujikuroi.

In addition, the medium composition and culture conditions have a great impact on the overall metabolic network of the strain, thus affecting the quantity and quality of secondary metabolites (Huang et al. 2022). For example, it was found that low concentrations of nitrogen sources (50 kg N.ha(^− 1) nitrate) would improve the cell growth of F. fujikuroi, thus improving productive characteristic of GA3 (Moszczynska et al. 2011). It was also reported that a pH of 5 was more favorable for GA3 accumulation compared to pH of 3 or 7 (Meleigy and Khalaf 2009). In conclusion, current studies had obtained the factors affecting GA3 production, such as nitrogen sources and pH. However, there is still room for improvement. It was reported that the fermentation factors tended to interact with each other (Wang et al. 2022a, b, c). The above-mentioned single-factor test cannot completely reveal the synergistic effect of fermentation factors on GA3 production. Therefore, we carried out multifactorial experiment by central composite design (CCD) and response surface methodology (RSM) in this manuscript to screen the optimum combination for GA3 biosynthesis.

In this study, ARTP mutagenesis was combined with ketoconazole screening to mutagenize the initial strain of F. fujikuroi, resulting in strain 3-6-1 with high GA3 production. Subsequently, other strategies, including the optimization of the medium composition, shake flask fermentation parameters, and the regulation of batch fermentation, were adopted to further improve the GA3 titer of industrial fermentation.

Materials and methods

Strains and culture media

F. fujikuroi NJtech 02 (CCTCC M2015614) was obtained from the China Center for Type Culture Collection. The strain was cultivated at 28 °C with shaking at 200 rpm in seed medium (30 g/L glucose, 5.5 g/L yeast extract, 0.2 g/L MgSO₄-7H₂O, 41.5 g/L KH₂PO₄, 0.05 g/L Na₂MoO₄·2H₂O, 1 ml trace element solution). After 36 h, the seed cultures were used to inoculated the fermentation medium (60 g/L glucose, 5.5 g/L yeast extract, 0.2 g/L MgSO₄. 7H₂O, 1.5 g/L KH₂PO₄, 0.05 g/L Na₂MoO₄.2H₂O, 1 ml trace element solution), followed by further cultivation for 10 days. The aqueous trace element solution consisted of 300 mg/L H₃BO₃, 100 mg/L MnCl₂·4H₂O, 100 mg/L ZnSO₄·7H₂O, 200 mg/L FeCl₃·6H₂O, and 500 mg/L CuCl₂·2H₂O.

ARTP mutagenesis and screening

The ARTP mutagenesis instrument was purchased from Siqingyuan Biologicals Ltd. The operating parameters were 100 W and gas pressure of 0.1 MPa. In each ARTP mutagenesis experiment, 10 μL of the cell suspension was uniformly dispersed on a sterilized metal plate and exposed to the ARTP jet for 0 s, 30 s, 40 s, 50 s, 60 s, 70 s, 80 s, and 90 s, respectively. After mutagenesis, the cells were eluted with sterile saline into a 1 mL centrifuge tube. After washing twice with normal saline, the cells were transferred into a conical bottle with 50 mL sterile water and covered with glass beads at the bottom. The suspension was incubated at 28 °C with shaking at 200 rpm for 20 min. Finally last, 1 mL of the dilution was cultivated on plates containing ketoconazole at a concentration of 0, 5, 10, 20, 30, or 50 mg/L, at 28 °C for 2 days.

The number of colonies was counted to determine the lethality rate, which was calculated using the following formula:

$${\text{Lethality rate }} = \left( {\text{colonies on plate with ketoconazole/colonies on plate without ketoconazole}} \right) \, \times \,100\% .$$

(1)

The preserved strains were subsequently inoculated in shake flasks, and the fermentation experiment was carried out. The content of gibberellin GA3 was detected after fermentation, and the strains with high yield were preserved. In the genetic stability test, the strains obtained by re-screening were passaged on agar medium. The fermentation experiments were carried out with each generation of strains and the yield of GA3 was determined.

Plackett–Burman experimental design

The PB experiment was designed at 2 levels, which is very efficient for screening the most significant factors affecting product yield and has been widely used for media optimization in the process of fermentation. In this study, the test factors were divided into a high level (+ 1) and low level (− 1). A first-order regression equation can be established to describe the relationship between the factors and the response values:

$$Y = \beta 0 + \sum {\beta iXi} \quad i = 1,{\mkern 1mu} 2, \ldots ,K.$$

(2)

The equation parameters included the predicted response value Y, the constant term β₀, and the regression coefficient β_i (Patil et al. 2013).

Response surface methodology

Central Combinatorial Design (CCD) is a response surface design with the advantages of simple structure and accurate prediction. Through the experimental design, a quadratic polynomial equation model can be obtained to explain the relationship between various factors and response values:

$${\text{Y}} =\, \beta 0 + \sum \beta i {\text{X}}i + \sum \sum \beta {\text{iiXi}}^{2} + \sum \beta {\text{ij}} {\text{XiXj}} \quad {\text{i, j}} = 1,2, \ldots , {\text{k}}{.}$$

(3)

The equation model included parameters for the predicted response value Y, the constant term β0, independent factors Xi, Xj, the quadratic coefficient βii, and the correlation coefficient βij (Ji et al. 2009).

Dry cell weight measurement

The dry cell weight was measured every 24 h. Samples comprising 50 mL of the fermentation broth are vacuum filtered through a glass fiber filter and then washed twice with distilled water. The wet mycelia were dried at 60 °C to a constant weight. The content of glucose was detected using a SBA-40C biosensor (Institute of Biology, Shandong Academy of Sciences). The fermentation broth was centrifuged at 12,000 g for 5 min. The supernatant was filtered through a 0.22 μm pore-size aqueous membrane, followed by liquid phase analysis in the liquid phase injection bottle.

Analysis of gibberellic acid GA3

GA3 was analyzed by high-performance liquid chromatography (Dionex U3000) equipped with a Venusil MPC18 column (5 μm, Agela Technologies). The pretreated samples were separated using a mobile phase composed of methanol/water/phosphoric acid (68:32:0.05) at a flow rate of 0.7 mL/min. The detection wavelength was 210 nm, and the injection volume was 10 μL. The retention time of GA3 was 24.12 min. The standard curve was prepared by diluting the 0.1 g GA3 standard with 10 ml methanol to yield solutions of 50, 100, 200, 400, 600, and 800 mg/L. Then, the standard solutions were filtered through a 0.22 μm pore-size organic membrane for liquid phase analysis. The standard curve was made, and the formula was as follows:

Y = 0.1256*X-0.2476 (R² = 0.9991),

where Y is the peak area and X represents the concentration of GA3.

Fermentation in a 5 L bioreactor

The bioreactor fermentation was conducted in medium containing 90 g/L glucose, 12.95 g/L defatted soybean meal, 0.1 g/L MgSO₄·7H₂O, 1.5 g/L KH₂PO₄, and 0.05 g/L Na₂MoO₄·2H₂O, and 1 ml trace element solution consists of 300 mg/L H₃BO₃, 100 mg/L MnCl₂·4H₂O, 100 mg/L ZnSO₄·7H₂O, 200 mg/L FeCl₃·6H₂O, and 500 mg/L CuCl₂·2H₂O. After cultivation at 26 °C with shaking at 220 rpm for 36 h, an inoculum comprising 5% of the pre-culture in seed culture medium was used to inoculate a 7.5 L bioreactor and cultivated at pH 4 for 10 days. The GA3 titer, glucose content, and OD₆₀₀ were measured every 24 h.

Results and discussion

ARTP mutation combined with ketoconazole screening to obtain mutant strains with high GA3 titer

Optimization of the ARTP treatment time

ARTP is the use of a variety of chemically active particle components in the normal pressure and room-temperature plasma source, such as OH- and nitrogen molecule two-positive system, to achieve the purpose of mutagenesis (Songnaka et al. 2022; Zhang et al. 2014). We firstly utilized ARTP technology for the mutagenesis of F. fujikuroi protoplasts, and the number of clones was counted to calculate the lethality rate. In order to obtain positive mutant strains with improved GA3 production, the ARTP treatment time was optimized to control the lethality rate between 70 and 80% (Gao et al. 2020). As shown in Fig. 1a, the lethality rate increased with prolonged ARTP treatment time. When protoplasts were exposed for 60 s, the lethality rate reached 58.5%. When the treatment time was increased to 90 s, the lethality rate reached 100%. Consequently, 70 s was chosen as the optimal treatment time, resulting in a lethality rate of 75.1%.

Fig. 1

ARTP mutagenesis was combined with ketoconazole selection to obtain mutant strains with increased GA3 production. a Lethality rate curve of F. fujikuroi under ARTP exposure. b Lethality rates of the strain at different ketoconazole concentrations. c Genetic stability of the improved mutants 3-6-1 and 1-6-4 during serial passaging

Optimization of the ketoconazole concentration

After optimizing the ARTP treatment time, it was necessary to establish an efficient screening method. Ergosterol is an important component of the fungal cell wall, and its absence leads to cell death (Koselny et al. 2018). Ketoconazole is one of the members of the imidazole series, which has a broad-spectrum antifungal profile by inhibiting ergosterol biosynthesis and interfering with other membrane lipids (Borgers et al. 1983). As reported, the products which competed for the precursor GGPP (geranylgeranyl pyrophosphate) with ergosterol were overaccumulated in strain under pressure of ketoconazole. For instance, combined with overexpression of the 3-hydroxy-3-methylglutaryl-coenzyme A (HMG-CoA) reductase gene (rate-limiting enzyme for terpenoid synthesis), the addition of ketoconazole has the ability of increasing the β-carotene concentration by 206.8% in Saccharomyces cerevisiae with a 22.9% decrease in the ergosterol content (Yan et al. 2012). The results indicated that ketoconazole has the potential of increasing GA3 titer in F. fujikuroi by downregulating ergosterol biosynthesis, which increased flux of GGPP applied for GA3 production. We therefore selected ketoconazole as the screening agent, based on its ability to inhibit the production of ergosterol. Only strains with high production of the general precursor (Farnesyl Pyrophosphate: FPP) could maintain the synthesis of ergosterol under the screening pressure, and the increased precursor pool could potentially also increase the production of GA3. Firstly, the cell suspension of the control group was spread on the screening plate supplied with different ketoconazole concentrations, and the lethality was determined by counting the number of colonies. As shown in Fig. 1b, when the concentration of ketoconazole was set to 30 mg/L, the lethality rate already reached 100%. Therefore, 30 mg/L of ketoconazole was chosen for the subsequent test.

Next, the protoplasts of F. fujikuroi were exposed to ARTP for 70 s and firstly screened on the screening plate supplied with 30 mg/L ketoconazole. After that, 240 well-grown colonies were selected and transferred to seed medium and fermentation medium for re-screening, which yielded 12 positive mutants (Table 1). The two mutants 1-6-4 and 3-6-1 exhibited 57 and 64% higher GA3 production than the control strain, reaching titers of 186.86 and 194.64 mg/L, respectively. Notably, their biomass also increased by 7.2% and 19.5% compared to the control strain, reaching 19.75 and 22.02 g/L, respectively. This result confirmed that ketoconazole could be used as a favorable selection pressure to improve the production of GA3.

Table 1 The productivity of the high-yield mutants by ARTP mutagenesis

Analysis of the genetic stability

Considering that mutant strains are often unstable and prone to reversion mutations (Yamayoshi et al. 2018), the two positive mutant strains were grown on agar slants for serial sub-cultivation. Subsequently, their GA3 contents were measured. It was found that the GA3 titer of the mutant strain 3-6-1 was reduced by 7% compared with the starting strain after sub-culturing for 10 generations, and finally stabilized at approximately 180 mg/L. By contrast, the GA3 titer of the mutant strain 1-6-4 was only 85% that of the starting strain (Fig. 1c). Therefore, although the yield of the 3-6-1 mutant strain was slightly reduced, it was more genetically stable.

Thus, ARTP mutation combined with ketoconazole screening yielded the mutant strain 3-6-1 with high GA3 titer and good genetic stability.

Systematic optimization of fermentation medium for GA3 production

Single factor optimization of nitrogen source

In previous studies, the transcription factor was obviously affected by culture conditions, and the nitrogen regulatory proteins participated in GA3 biosynthesis (Wang et al. 2022a, b, c). For example, by multi-omics analysis, it was found that the transcription factor AreA would bind to promoter of the target gene, such as P450-1/P450-4, under the condition of nitrogen starvation, thus upregulating GA3 biosynthesis (Michielse et al. 2014). Carbon and nitrogen sources are usually considered as the most influential factors affecting gene expression (Basiacik Karakoc and Aksoz, 2004). Therefore, this study focused on carbon and nitrogen source optimization to further improve the yield of GA3.

It was reported that the production of GA3 begins when the nitrogen source is depleted (Rios-Iribe et al. 2011), so the type and concentration of the nitrogen source are crucial for the fermentation of GA3. In this study, ten representative organic and inorganic nitrogen sources were tested, including ammonium chloride, ammonium nitrate, potassium nitrate, yeast extract, corn steep liquor, soybean peptone, defatted soybean meal, soybean powder, peanut powder, and wheat protein.

When corn steep liquor, defatted soybean meal, peanut powder, soybean peptone, or wheat protein were used as nitrogen sources, the yield of GA3 was significantly higher than in the control (Fig. 2a). Among them, the optimal nitrogen source was defatted soybean meal, with which the yield of GA3 reached 381.86 mg/L, and nearly doubled compared with the control (180 mg/L). By contrast, when ammonium chloride was used as nitrogen source, the pH decreased to 0.38, which seriously affected the cell growth with the final biomass of 6.81 g/L. In addition, the cell growth was normal when ammonium nitrate and potassium nitrate were used as nitrogen sources, but the yield of GA3 was extremely low. These results showed that organic nitrogen sources were more conducive to cell growth and GA3 synthesis than inorganic nitrogen sources.

Fig. 2

Optimization of carbon and nitrogen sources. a The influence of different nitrogen sources on DCW and GA3 concentration. b The influence of different carbon sources on DCW and GA3 concentration

Single factor optimization of carbon source

Similar to the nitrogen source, the carbon source is also an important factor affecting GA3 synthesis (Camara et al. 2018). The carbon source not only provides energy for cell growth but is also converted into the GA3 precursor acetyl-CoA (Rodriguez-Ortiz et al. 2010). In this study, 6 representative carbon sources were tested, including glucose, fructose, sucrose, starch, glycerol, and mannitol. The yield of GA3 reached the maximum of 202.00 mg/L when glucose was used as the sole carbon source, followed by mannitol with a yield of 180.28 mg/L. Conversely, fructose resulted in the lowest GA3 production (Fig. 2b). In addition, when glycerol was used as the sole carbon source, the biomass reached the highest value of 28.02 g/L, while the other carbon sources resulted in similar biomass accumulation, indicating that glycerol was the most favorable carbon source for cell growth. In summary, carbon sources had different effects on the cell growth and GA3 synthesis, but glucose was the most favorable carbon source for GA3 production.

Plackett–Burman design for determining the main factors

It was reported that different nutritional compounds are interactional factors rather than isolated from each other (Nkhata et al. 2018). Therefore, response surface methodology (RSM) was used to search for the optimal conditions in the multi-factor system in order to further improve the GA3 titer. The Plackett–Burman (PB) design is generally regarded as an indispensable support for RSM (Dil et al. 2021). PB is extremely efficient in ranking the factors that affect the yield of products, and has been widely used for the optimization of fermentation media (Zhou et al. 2019).

In the PB experiment, six different factors were selected, including defatted soybean meal, glucose, KH₂PO₄, MgSO₄·7H₂O, Na₂MoO₄·2H₂O, and trace elements, while the GA3 yield was selected as the response value. Based on the software Design Expert (version8.0, Stat-Ease, Inc., USA), each factor was classified into two levels, and the actual values are shown in Table 2. In this study, a total of 12 groups of experiments were conducted in shake flasks, followed by measurement of the GA3 titers. The experimental design and results are shown in Table 3. According to the results shown in Table 4, variance analysis was carried out using the software Design Expert. The correlation coefficient of the regression model was R² = 0.9969, and the adjustment correlation coefficient was R² (Adj) = 0.9827, indicating that the regression model had a high fitting degree. The P values of glucose, defatted soybean meal and magnesium sulfate were less than 0.05, indicating that these three factors were the main factors affecting the GA3 content. Therefore, the RSM strategy was adopted for these three factors in the subsequent experiments.

Table 2 Variables and their levels employed in the Plackett–Burman design

Table 3 Matrix of Plackett–Burman design and results of the response (GA3 yield)

Table 4 Effects and statistical analysis of variables

Response surface methodology

As shown in Table 5, the concentrations of glucose, defatted soybean meal and magnesium sulfate were optimized. Using the software Design Expert, a 3-factor and five-level Central Composite Design (CCD) experiment was performed, and the GA3 yield was used as the response value. The actual values of the experimental factors are shown in Table 6 and the variance analysis is presented in Table 7.

Table 5 Variables and their levels employed in CCD

Table 6 Matrix of CCD and results of the response (GA3 yield)

Table 7 Effects and statistical analysis of variables

Based on Design Expert, a total of 20 groups of experiments were designed to analyze the effects of defatted soybean meal, glucose, and magnesium sulfate on the yield of GA3. The response diagram is shown in Fig. 3a–c. According to Table 7, the model was significant and lack of fit was not significant, which proved that the model had the ability to accurately predict the best combination of three factors. At this moment, the prediction reached 392.914 mg/L with the combination of 13 g/L defatted soybean meal, 90 g/L glucose, and 0.1 g/L MgSO₄.7H₂O (4.412% lower than the real value). Based on these predicted values, we carried out fermentation experiments and obtained a high GA3 yield of 410.25 mg/L, which was twofold higher than the titer obtained under the initial conditions. The yield was only slightly lower than the predicted results with deviation reaching 10%, indicating that the model could predict the actual results.

Fig. 3

Response Surface Methodology. Response surface plots of effect of interaction of a Defatted soybean meal and glucose addition. b Defatted soybean meal and MgSO₄•7H₂O addition. c glucose and MgSO₄•7H₂O addition on GA3 yield

Optimization of fermentation parameters

Inoculation amount

The inoculation amount affects the lag phase of the strain at the beginning of the fermentation. The lag period was shortened when the inoculation amount was large, which resulted in the rapid propagation of cells and fast synthesis of products (He et al. 2013). However, the increase of inoculation amount affected the normal metabolism of cells due to large amounts of metabolic byproducts. Therefore, we investigated the effect of inoculation amount as a single factor on the GA3 titer in this study, with inoculation of 2, 5, 10, and 20%, respectively. The highest yield of GA3 was obtained when the inoculation amount was set to 5%, reaching 406.95 mg/L (Fig. 4a), while too high or too low inoculation amounts reduced the yield of GA3.

Fig. 4

Optimization of fermentation parameters. The influence of a inoculum size, b inoculation period, c agitation speed, d temperature, and e initial pH on the GA3 concentration. f Comparison of the initial and optimized conditions

Inoculation time

The quality of the seed culture has a great influence on the fermentation result (Wang et al. 2015). Generally, inoculation with a seed culture in the logarithmic growth phase has strong advantages for cellular metabolism and growth, which effectively shortens the lag phase, increasing the product yield (Ji et al. 2009). In this study, we took the inoculation time as a single factor, and the inoculation time was varied at 24, 36, and 48 h, respectively. It was found that the yield of GA3 reached the maximum of 413.09 mg/L when the inoculation time was set to 36 h (Fig. 4b). However, the results showed that the inoculation time generally had little effect on GA3 production.

Shaking speed

Oxygen was necessary for the cell growth and GA3 synthesis of F. fujikuroi (Kai et al. 2016). As a result, it was essential to keep the medium well aerated in the fermentation process. In this study, we took the rotation speed as a single factor, which was varied at 150, 200, 250, and 300 rpm, respectively. The results showed that when the rotation speed was 200 rpm, the yield of GA3 reached the maximum of 414.97 mg/L (Fig. 4c). By contrast, the yield of GA3 decreased at low rotation speed, which may be due to the low dissolved oxygen, which was insufficient for the synthesis of GA3. In addition, GA3 production significantly decreased when the rotation speed was increased above 250 rpm, most likely due to shear stress, leading to cell damage, which decreased the total biomass.

Temperature

Microbial growth and metabolism are dependent on various enzymes, which are directly affected by temperature (Bai et al. 2020). Accordingly, temperature was also identified as one of the essential factors affecting the cell growth and synthesis of GA3. In this study, we refered to the gradient settings for temperature by Inacio da Silva et al., who designed the temperature as 28 °C, 31.5 °C, and 35 °C separately, and the highest GA3 titer was obtained at 28 °C (da Silva et al. 2021). Based on that, “28 °C -centered” temperature design (24 °C, 26 °C, 28 °C, 30 °C, 32 °C) was used for exploration of effects of high and low temperature on GA3 biosynthesis. The results showed that as the temperature increased, the yield of GA3 first increased and then decreased. When the temperature was 26 °C, the yield of GA3 reached a maximum of 450.09 mg/L (Fig. 4d). Additionally, when the temperature was higher than 32 °C, the yield of GA3 decreased significantly, indicating that the synthesis of GA3 was limited by high temperature. We speculated that the reason is the combination of enzyme activity and characteristic of F. fujikuroi. It was reported that the enzyme activity of CPS/KS was downregulated at 31 °C compared to 22 °C (Piombo et al. 2020). Also, the cell wall on the surface of filamentous fungi is constantly being regenerated, and high temperature, such as 32 °C, can exacerbate environmental stress to have negative effects on the hyphae formation of Fusarium sp. (Yoshimi et al. 2016).

Initial pH

It was reported that the initial pH of the medium has a great influence on the growth and metabolism of microorganisms. For example, GA3 was the main product of F. fujikuroi when the pH value was between 3.5 and 5.8, while it tended to produce GA4 and GA7 when the pH value was above 6.0 (MacMillan 2001). Therefore, the GA3 production could be guided by adjusting the pH value in the fermentation process (Wang et al. 2020). In order to explore the optimal initial pH, we took the initial pH as a single factor, which was varied at 3.0, 4.0, 5.0, 6.0, and 7.0, respectively. The results showed that with the increase of initial pH, the yield of GA3 rose at first and then decreased. When the initial pH was 4.0, the yield of GA3 reached the maximum of 452.19 mg/L (Fig. 4e), indicating that the optimal initial pH was 4.0.

Fermentation verification experiment

In order to comprehensively evaluate the effects of the optimized medium components and fermentation parameters on GA3 synthesis, fermentation experiments were carried out in shake flasks under combined conditions. The strains were cultivated at optimal fermentation parameters, including an inoculation volume of 5%, inoculation time of 36 h, shaking speed of 200 rpm, temperature of 26 °C, and initial pH of 4.0.

Under these conditions, the final biomass was 17.89 g/L, and the yield of GA3 reached 462.16 mg/L (Fig. 4f) when F. fujikuroi was cultivated in shake flask cultures comprising 90 g/L glucose, 5.5 g/L yeast extract, 0.2 g/L MgSO4·7H2O, 1.5 g/L KH2PO4, 0.05 g/L Na2MoO4·2H2O, and 1 ml trace element solution. Notably, there was little difference in terms of biomass compared with the control, indicating that the optimized conditions had little effect on cell growth, while the yield of GA3 improved from 180 to 462.16 mg/L, representing a 2.5-fold increase over the initial conditions before optimization.

Effects of surfactants and vitamins on the GA3 titer

Surfactants

It was reported that the addition of small amounts of surfactants could increase the dissolved oxygen level in the fermentation broth, promoting the cell growth and improving product synthesis (Song et al. 2016). In order to promote the accumulation of GA3, we tested six different surfactants at a concentration of 2%, including sodium dodecyl sulfate (SDS), betaine, Tween-80, polyethylene glycol, span-80, and stearic acid, respectively.

The results showed that there was no significant change in biomass (Fig. 5a), indicating that the surfactants had no side effects on the growth of F. fujikuroi, but they had an impact on the yield of GA3. For instance, the addition of Tween-80 increased the yield of GA3 from 462.16 to 496.78 mg/L, and its concentration was further optimized. When 1% Tween-80 was added, the titer of GA3 reached the highest value of 498.53 mg/L (Fig. 5b). Additionally, the results showed that an excessive concentration of Tween-80 had obvious side effect on cell growth. Therefore, 1% of Tween-80 was finally selected as the optimal concentration.

Fig. 5

The effects of different surfactants and vitamins on DCW and GA3 concentration. The effects of a different surfactants, b concentrations of tween-80, c different vitamins, d concentrations of VB6, e concentrations of VB12, f concentrations of biotin on DCW and GA3 concentration

Vitamins

Vitamins are organic compounds that are necessary in small amounts for the growth of many heterotrophic organisms and are added as important nutrients in many fermentation processes (Han et al. 2019). Here, we tested five vitamins related to cell metabolism, including vitamin B1 (VB1), vitamin B6 (VB6), vitamin B12 (VB12), biotin, and vitamin C (VC), which were added to a final concentration of 0.2 mg/L. The biomass of the experimental groups supplied with the five different vitamins increased to different extents, indicating that vitamins could promote the growth of F. fujikuroi (Fig. 5c). Moreover, the yield of GA3 was promoted by the addition of VB6, VB12, and biotin, while it was decreased by VB1 and VC. According to Fig. 5c, we concluded that 0.2 mg/L of vitamins could affect the GA3 production in F. fujikuroi. From the view of production cost, we next tried to reduce the dose of vitamins. According to Fig. 5d–f, there is an obvious increase in GA3 titer (> 15%) when 0.05 mg/L vitamins were supplemented to the medium. Hence, we designed the concentration vitamin at the range of 0.05–0.5 mg/L for subsequent experiments to further optimize the added amounts of VB6, VB12, and biotin.

When 0.2 mg/L vitamin B6 was added to the fermentation medium, the biomass was increased to 21.55 g/L, which was 21.0% higher than in the control (Fig. 5d). It stands to reason that VB6 promoted the accumulation of the GA3 precursor acetyl-CoA and thus increased the GA3 titer (Yasuda et al. 2022). By contrast, the biomass did not change significantly, but the GA3 titer was improved when VB12 was added. When the concentration of VB12 was set to 0.1 mg/L, the titer of GA3 reached 572.17 mg/L, which was 24.34% higher than that in the control (Fig. 5e). In addition, when the concentration of biotin was set to 0.1 mg/L, the biomass reached the maximum of 24.65 g/L, which was 38.25% higher than that in the control (Fig. 5f), while the titer of GA3 reached 519.37 mg/L.

Feedback-controlled fed-batch fermentation based on the residual sugar concentration

It was reported that feedback-controlled fed-batch fermentation based on the residual sugar concentration could eliminate the substrate inhibition and catabolite repression induced by high concentrations of glucose (Wang et al. 2022a, b, c). It is reported that GA3 can only accumulate when the nitrogen source is depleted, so it is necessary to limit the nitrogen source in the medium. To maintain a high C/N ratio, we first added 90 g/L of glucose in the medium (Shi et al. 2017a, b). As shown in Fig. 6a, batch fermentation supplied with 90 g/L initial glucose resulted in a stagnant GA3 yield of 520 mg/L, at which point the glucose was exhausted. In order to further improve the fermentation yield of GA3, fed-batch fermentation was carried out in a 7.5 L stirred-tank fermenter.

Fig. 6

Fed-batch fermentation. a Batch fermentation of F. fujikuroi. b Fed-batch fermentation with an initial glucose concentration of 40 g /L and residual glucose controlled at 30–40 g/L. c Fed-batch fermentation with an initial glucose concentration of 30 g /L and residual glucose controlled at 10–20 g/L

It has been reported that fermentation at low glucose concentrations in the range of 10–60 g/L is beneficial for the accumulation of target products (Li et al. 2021). For F. fujikuroi, low glucose has been proved to be more favorable for GA3 accumulation than high glucose during fermentation process. For example, with an initial glucose concentration of 100 g/L, the titer of GA3 reached 1.2 g/L (Uthandi et al. 2010), only half of low-glucose experiment (50 g/L glucose) (Wang et al. 2017). In addition, 300 g/L of glucose was continuously fed at the flow rate of 0.005 L/h with the GA productivity of 0.0168 gL⁻¹ h⁻¹, 2.9-fold higher than that obtained by adding all glucose at once, which was considered as a typical example of low-glucose fermentation of F. fujikuroi (Shukla et al. 2005). Because the specific range of glucose concentration remained unknown, the feeding mode of glucose was optimized in this study.

Two feedback feeding methods based on residual sugar were designed as follows: a. 40 g/L initial glucose, followed by control of the residual sugar in the range of 30–40 g/L; b. 30 g/L initial glucose with the controlled residual sugar in the range of 10–20 g/L. The fermentation profiles are shown in Fig. 6b–c. It was found that the biomass accumulation was almost the same as that of batch fermentation, indicating that the glucose concentration had little effect on the cell growth. However, when the residual sugar concentration was maintained at 30–40 g/L, the yield of GA3 was obviously superior to the other two fermentation modes. The gap of GA3 titer appeared to widen in the later stage of fermentation, especially after 192 h. Eventually, the GA3 titer reached the maximum of 575.13 mg/L at 240 h, representing a 13.86% increase over the conventional batch fermentation (Fig. 6b).

By contrast, when the residual sugar concentration was maintained at 10–20 g/L, the GA3 accumulation reached saturation at 24–48 h (Fig. 6c). We speculated that the early stage of fermentation was a critical period for rapid accumulation of biomass (Anand and Srivastava 2022), during which the glucose consumption rate exceeded the glucose supply rate, resulting in reduced accumulation of GA3 at an early stage. In conclusion, we added glucose when the concentration was under the desired range, and we totally used glucose of 450 g (Fig. 6A), 280 g (Fig. 6B), and 260 g (Fig. 6C) separately. The glucose consumption of fed-batch fermentation was only half of batch fermentation (90 g/L), which indicated that fed-batch fermentation greatly reduced glucose consumption. Therefore, the fed-batch fermentation strategy of maintaining the residual sugar concentration within the range of 30–40 g/L was adopted for GA3 production. To our best knowledge, this is the first study to determine the lower bound of glucose concentration suitable for GA3 synthesis.

In summary, ARTP was used for strain mutagenesis for the first time. Response surface analysis was firstly used to analyze the carbon–nitrogen ratio (glucose/defatted soybean meal) most suitable for GA3 synthesis, and the low-glucose feeding mode was proposed for increasing GA3 titer. In this study, we achieved a fivefold increase in GA3 production merely by fermentation engineering, which was expected to provide effective suggestions for industrial production of GA3 in F. fujikuroi.

Conclusion

In this study, ARTP mutagenesis was applied for the breeding of F. fujikuroi. According to the metabolic network of F. fujikuroi, ketoconazole was firstly used as screening agent of the mutant strain, and a high-yield strain 3-6-1 was obtained with GA3 yield of 194.64 mg/L. Afterward, the optimal fermentation medium and fermentation parameters were achieved comprising defatted soybean meal and glucose as the nitrogen and carbon source, respectively, as well as an inoculation time of 36 h, shaking at 200 rpm, temperature of 26 °C, and initial pH of 4.0. In this study, we obtained the optimal concentrations of defatted soybean meal and glucose by PB and CCD experiments, which were 13.0 and 90.0 (g/L), respectively. The GA3 titer under optimal condition was 410.25 mg/L, 2.1-fold higher than that under the original conditions. Furthermore, we discussed a new feedback regulation mode superior to one-step feeding method of glucose (90 g/L), in which the residual sugar concentration was dynamically regulated for cell growth and GA3 synthesis. Finally, the GA3 titer reached the maximum of 575.13 mg/L at 240 h when the initial glucose was 40 g/L, and then the glucose was maintained in the range of 30 ~ 40 g/L. In conclusion, we obtained a potential GA3-producing strain by ARTP mutation and fermentation optimization, which revealed the effects of fermentative factors on GA3 production in F. fujikuroi, thus laying a foundation for its industrial production.