Introduction
Sophorolipids (SLs) as renewable glycolipid biosurfactants are mainly
produced by microorganisms of the Saccharomyces ssp. (Van Bogaert
et al., 2007). Glucose and fatty acids are the main precursors for the
synthesis of SLs. First, fatty acids are catalyzed by P450 monooxygenase
into hydroxy fatty acids, and then UDP-glucoses are sequentially
connected to form non-acetylated acidic SLs via the action of
glucose transferase I and II (Lodens et al., 2020). Finally, the
functional activities of acetyltransferase and extracellular lactonase
lead to the acetylation and lactonization of SLs. Up to 20 different
structural forms of SLs are known to exist, and this variation is the
result of differences in acetylation, hydroxyl position, the length of
the fatty acid chain, and the unsaturation degree of fatty acids (Hu &
Ju, 2001). Thus, the production of SLs is a complex and multiphase
fermentation process, involving a gas phase (air), a solid phase (cells,
SLs crystals), a hydrophilic liquid phase (acidic SLs, glucose) and a
hydrophobic liquid phase (lactonic SLs, oil), which pose a significant
challenge to the efficient and stable production of SLs (Tian, Li, Chen,
Mohsin & Chu, 2021).
At present, the optimization of
microbial fermentation can be divided into three different aspects: (1)
obtain high-performance producers via mutagenic breeding or
genetic engineering, (2) develop and utilize cheap substrates to reduce
the costs of fermentation , and (3) optimize the fermentation process to
achieve the efficient synthesis of product (Dolman, Kaisermann, Martin
& Winterburn, 2017; Li, Chen, Tian & Chu, 2020;
Tian et al., 2021; Wang et al.,
2020a). Of these processes, the rational and precise regulation of
fermentation remains the most significant factor in achieving
high-efficiency production. The identification of key process parameters
form the basis of process control and optimization. By regulating key
process parameters, it is possible to achieve the regulation of cell
metabolism in a flexible manner, this allowing high titer, productivity,
and yield (Wang et al., 2020c). The continuous development of sensing
detection and information processing technologies, along with the
real-time detection of conventional environmental parameters by on-line
sensors, cellular macro-physiological, and metabolic parameters, has led
to the availability of key parameters on-line, including living cell
amount, oxygen uptake rate (OUR), carbon dioxide evolution rate (CER),
and respiratory quotient (RQ). This information creates a database for
the fermentation process (Chen, Lin, Tian, Li & Chu, 2019; Feng et al.,
2021).
However, the mining of sensitive process parameters still relies upon
correlation analysis and manual experience and has yet to be studied
from the perspective of big data analysis (Davila, Marchal &
Vandecasteele, 1997). On the other hand, it is gratifying that various
data processing and analytical methods are gradually being introduced
into the fermentation process. For example, linear and non-linear
algorithms, neural networks,
support vector machines, and other mathematical models, can quickly
process many on-line and off-line parameters, and can therefore be
correlated with regulatory processes during the fermentation process
(Safarian, Saryazdi, Unnthorsson & Richter, 2021; Zhang et al., 2020).
Overall, the current options for regulating the fermentation process
arise predominantly from the process control perspective and lack
rationality (Kim, Yun & Kim, 2009). Other options include guidance
provided by cellular metabolic characteristics, but this approach lacks
universal applicability.
In terms of SLs fermentation, the supplementation of glucose and oil
substrates is essential for the synthesis of SLs. During the late
fermentation stage, and with the gradual accumulation of SLs, the
rheological properties of the fermentation broth undergoes significant
changes, thus increasing viscosity. These changes exert a key impact on
mass transfer and mixing, thus resulting in a limited oxygen supply,
consequently, the productivity of SLs synthesis decreases notably. The
development of a semi-continuous fermentation process could
significantly alleviate the influence of oxygen limitation on SLs
synthesis (Zhang et al., 2018). It has been found that controlling the
content of oil, and the ratio of oil to SLs, can exert influence on the
morphology of SLs (crystalline or non-crystalline types), thereby
significantly changing the sedimentation characteristics of SLs (Chen et
al., 2021b). In turn, this affects the efficiency of semi-continuous
fermentation. Therefore, it is vital that we are able to precisely
control the process of oil supplementation so that we can achieve the
efficient production of SLs.
In this study,
we established a multi-scale
parameter detection system for the SLs fermentation process by applying
a range of on-line sensors, mainly including a near-infrared
spectrometer and a process mass spectrometer. First, we studied the
differences in macro-physiological and metabolic parameters under
different rates of oil
supplementation. Subsequently, we used a range of process parameters to
construct a data-mechanism fusion model, which was accomplished by
integrating data modeling with cellular metabolic mechanisms, for
feedback feeding of oil and glucose. Finally, this model was applied to
semi-continuous fermentation to achieve a highly efficient production
system for SLs.