Screening in situ by pigment ring in PBF plate
The spores of the original strain were diluted and plated for culturing. Colonies were selected for shake flask fermentation. We found that the fermentation ability of offspring was highly variable.
As shown in Figure 2a, b, strains grown on PBF medium plates produced black pigments around the colony, which was absent when cultured on ABK medium plates. Therefore, colonies grown on ABK medium were selected based on the colony size and spore quantity. In addition to these characteristics, the colonies on the PBF medium plates could be selected based on the size and depth of the pigmentation ring. In the avermectin production facility, strain engineers focus on studying and summarizing the relationship between fermentation ability and colony morphology, including growth state, color and spore quantity. Generally, there are three types of colonies of Streptomyces avermitilis, namely, the gray powder type, white type and bare type. Among them, the strain with gray powder type colonies usually has the highest production capacity. In contrast, the strains with bare colonies are usually poor in fermentation. On PBF medium plates, 44 strains were selected according to the spore quantity and the black pigmentation ring. Among them, 16 strains fermented 3000-4000 μg/mL avermectin B1a, and only two strains fermented more than 4000 μg/mL B1a (Figure 3), which confirms the genetic instability of Streptomyces avermitilis described in the literature (Xi et al. 2020). Without an efficient screening method, the trait of avermectin production will degenerate quickly with passaging.
The correlation between the colony titer on the plate and the shake flask fermentation unit was studied, and the results are shown in Figure 4a. The regression equation is Y = 3.705X + 122.63 (Y stands for the shake flask fermentation titer of B1a, X stands for the colony fermentation titer of B1a on the plate), and the Pearson r value is 0. 990, indicating a positive correlation. The correlation between the titer of total avermectin and the titer of avermectin B1a was analyzed (Figure 4b), which shows that AVMs = 11.9 + 2.51B1a (Pearson r = 0.994). This high correlation means that high-yield strains of avermectin B1a can be identified simply by measuring the total titer of avermectin on the plate.
It has been proven that there is a negative correlation between the production ability of secondary pigments and avermectin. The aveI gene positively regulates the production of pigments, and deletion of the aveI gene led to increased biosynthesis of avermectin B1a by approximately 16-fold (Chen et al. 2008). Another study indicated that aveI functioned as a global regulator in S. avermitilis and controlled not only secondary metabolism and morphological differentiation but also primary metabolism. Consequently, on PBF medium plates, colonies with abundant spores, plump morphology and smaller pigmentation rings tend to be high-yielding colonies. However, in the application, the pigment circle is closely related to the growth of the colony (Figure 2). The larger the colony grows, the larger the pigment circle, and the smaller the colony, the smaller the pigment circle. Therefore, it was necessary for us to further develop a relatively objective in situ screening method.
Avermectin B1a is usually detected by HPLC, whose pretreatment is tedious, and the process is rather complex, labor-intensive, time-consuming and not suitable for high-throughput screening. Fortunately, the titer of the colony on the plate shows a very close correlation with the titer of the shake flask culture, and so does the B1a titer and the total titer of AVMs. Therefore, we can develop an in situ screening method by analyzing the titer of the colony on the plate.
Rapid screening of colonies
Avermectins are intracellular products that mainly exist within the mycelium of Streptomyces avermitilis and very little avermectins are present in the supernatant of the fermentation. However, during fermentation, many brown pigments accumulate in the culture, which have strong UV absorption at 200-300 nm wavelengths. We were concerned that water-soluble substances such as these pigments might be co-extracted by methanol and potentially interfere with the rapid analysis of colonies. Therefore, we analyzed the UV absorption of these water-soluble impurity substances. Fifty milliliters of avermectin fermentation broth was centrifuged at 3000× g for 5 minutes, and the supernatant was analyzed by UV scanning. We found that the maximum absorption wavelength of the impurity was 254 nm (Figure 5). The maximum absorption wavelength of avermectin B1a is approximately 245 nm. The punctured colony with agar was extracted with methanol, and the supernatant of the extraction was also scanned. The data in Figure 6 show that most of the chromatograms had a maximum absorption peak at approximately 245 nm. Two samples lost the ability to produce both pigment and avermectin (Figure 6). The ideal strain should produce only avermectin B1a without producing any pigment at all because these pigments make industrial wastewater hard to treat. In this study, we did not find such an ideal strain in our mutant library. This maximum absorption peak at 245 nm can be used to determine the avermectin content, which can be used to screen for high-yielding strains in situ.
The data of in situ colony screening are shown in Figure 7, and the colony from the original strain was chosen as a control. The absorbance of the control group was 0.33. Those above the control level were considered the desirable and positive mutants. The majority of mutants were negative and undesirable, and only approximately 6.7% of strains carried the positive mutation. Sample No. 18 showed the highest absorption, reaching more than 0.5, a 51.5% increase compared to the control. This strain was selected and cultured for further study. DNA is a double-stranded structure. Mutations often cause base changes in only one strand, leaving the other strand intact. As a result, there will be at least two genotypes in the same colony. We therefore further diluted the spores of the high-yield strains obtained from the preliminary screening and discarded the heterozygotes to obtain pure high-yield strains. All of the screened strains were studied by flask fermentation, and the results are shown in Figure 8. The flask titer of avermectin B1a of the start strain was only 4582 μg/mL, and that of AVMs was 10587 μg/mL. By screening the library of ARTP-induced mutants, we selected a strain with a flask titer of avermectin B1a of 6137 μg/mL and AVMs of 16919 μg/mL. By secondary screening, the average flask titer of avermectin B1a increased to 7017 μg/mL, and AVMs increased to 17027 μg/mL. The flask titers of avermectin B1a and AVMs were maintained during the next two generations of fermentation, suggesting stable inheritance of this excellent trait in the mutant.
In situ screening technology is widely used in microbiology. It is well known that Penicillium was discovered by observing the plate in situ. Bacteriostatic circles and transparent circles can be classified as in situ screening techniques. In fact, in situ screening is often the first choice for screening a special strain because it is fast, intuitive and scalable. However, strains of avermectin are notoriously hard to screen in situ, and there was no report of successes prior to this study. We considered two effective methods: one is to examine the pigmentation ring on the PBF medium plate, and the other is to analyze the colony extract. QPix™ Microbial Colony Picker can be applied directly on PBF medium plates. High-throughput analysis of colony extracts can be performed in 96-well plates, combined with background subtraction of 254 nm absorption, to achieve fast and accurate assessment of strain performance.
For industrial manufacturing, an excellent avermectin strain should not only produce more avermectin B1a but also have a high B1a/B1b ratio. In the separation process, avermectin B1a was purified by a recrystallization process. Because of the similarity between B1a and B1b in chemical structure and solubility, if the concentration of B1b is too high, the content of B1b in the product is also high, and B1b is an undesirable impurity in the product. Strains in the paper were separated, and 30 of them were selected for flask fermentation analysis (Figure 9). Although we did not deliberately screen for high B1a/B1b strains, the majority of strains isolated naturally had a higher B1a/B1b ratio than the control group. This may be due to the high B1a/B1b ratio of the starting strain. As a result, despite B1b also having a peak absorption at 245 nm, high B1a strains can still be screened out by this method.
Finally, the inconsistency of colony performance in shake flask fermentation versus in industrial large-scale fermentation should be discussed. In laboratory settings, strains were selected by colony potency analysis and shake flask fermentation tests. However, their high yield and other favorable characteristics are often not reproduced when cultured in large fermenters in industry. We have experienced this phenomenon in our previous work with an avermectin manufacturer (company name). We sent several excellent strains to the plant for trial fermentation. These strains in general are better than the strains they were using, but it is interesting to note that the best performer in the fermenter was often not the strain with the highest performance in the lab; instead, the second highest-producing lab strain was the best in the fermenter (unpublished data). Shaking flasks and fermenters are very different in many aspects. In addition to temperature and nutrient supply, oxygen availability is vital to the growth of microbes and cells in culture. In shaking flasks, the transfer of oxygen takes place via two liquid surfaces, and the efficiency is determined by the size and shape of the flask, agitation speed, filling volume, and ambient conditions. Cultivation processes in shaking flasks are never under absolute control, as they are not monitored in real time for critical parameters such as temperature, dissolved oxygen or optical density. Temperature and dissolved oxygen, for example, can only be regulated by controlling ambient conditions in a sealed shaker system or a cultivation room. Automated processes, such as feeding the culture according to defined profiles or the integration of control loops, cannot be accomplished. The fermenter is equipped with multiple sensors that allow for real-time monitoring. Temperature, dissolved oxygen, pH and biomass are measured constantly and displayed numerically and graphically. Exhaust, metabolites, and redox potential can be calculated based on these measured parameters in the fermenter. Various types of impellers operating at specific agitation speeds can meet the individual requirements of different cell types regarding shearing force and mixing efficiency. Strains sensitive to shearing force may produce poorly in fermenters but better in flasks. Antifoaming agents are necessary in fermenters to reduce the surface viscosity of the film, increase the velocity of drainage, and enhance the diffusion of gas. However, sensitive strains tend to age easily in the presence of defoamers and strong shearing force, which are likely the causes of loss of genetic stability, persistent poor fermentation and suboptimal performance in fermenters. Therefore, it is necessary to add a certain amount of defoamers, such as soybean oil and silicon polyether, to the media during screening to help select strains with good tolerance to shearing force and defoamers. This could be tested in future studies to improve our screening method.
In support of the significance of shearing force, our lead authors have made the following observation while working for an avermectin manufacturer (unpublished work, Hebei Veyong Biochemical Co., LTD). In 2000, the titer of avermectin was only approximately 1800 µg/mL, and the fermentation process could not continue for a long time because of hyphal aging. When the shaft of a fermenter broke by accident, stirring had to be stopped. The whole fermentation process was carried out only by blowing sterile air. While this fermentation unit was expected to be particularly low due to this accident, we were pleasantly surprised that the fermentation titer of avermectin was drastically improved in subsequent analysis. The hyphae in this particular fermentation broth appeared globular under the microscope because high-speed stirring is not conducive to mycelium ball formation. This suggests that Streptomyces avermitilis is sensitive to shearing force, causing the discrepancy between the titer in a flask culture and in a fermenter. This also implies that the airlift fermenter may be better for Streptomyces avermitilis fermentation. There have been some previous studies on the effects of stirring on fermentation (Lopez-Ramirez et al. 2018; Zhang et al. 2010). The correlation between the colony titer and the shake flask titer was studied in detail by Xi Cheng (2020, Hebei Xingbai Agricultural Technology Co., Ltd). Eleven single colonies were fermented in shaking flasks, and their potency was analyzed by HPLC. The correlation coefficient was 0.924 between the titer of a single colony and the titer of the shaking flask. At the significance level of 0.05, there was a positive correlation between the single colony titer and the corresponding shake flask titer. Despite the difference in medium compositions, moisture contents, and modes of nutrient transfer, plates and flasks are nonetheless highly correlated in terms of fermentation level.