JPH08317797A - Measurement of fermention tank - Google Patents

Abstract

PURPOSE: To provide a method for fuzzy measurement for a fermentation tank changing a culture state with time. CONSTITUTION: In this method for measuring the lapse of a culture in a culture tank by calculating a metabolic balance from measurement of a metabolite measurable in a metabolic composition, a culture state is classified into specific plural culture states and a metabolic balance in each specific culture state is experimentally determined in each specific culture state. A rule for identifying a culture state is decided by using a membership function corresponding to each measurable state variable, the fidelity against the membership function of each state variable obtained from the fermentation tank is determined, the contribution degree of each specific culture state is determined from the obtained fidelity. The ratio of the metabolic composition is determined from the contribution degree. The amount of a substance existing in the fermentation tank is computed from the measured value of the metabolite composition measurable at a point of the determination of the ratio of the metallic composition.

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a fuzzy measuring method for a fermenter.

[0002]

2. Description of the Related Art In a biological reaction process, an unsteady reaction operation in which the reaction state changes with time, such as batch culture or fed-batch culture, is often used. In order to control such a reaction, it is extremely important to measure the culture state over time and online measure various state variables that serve as indices for managing and controlling the process. However, in the actual production process, raw materials with a large amount of impurities such as waste sugar concentration are often used as substrates, and physical state variables such as temperature and pressure and limited chemical state variables such as pH and DO are excluded. , The state variables that can be directly measured online are limited. In addition, sensors for measuring various state variables such as bacterial cell concentration, substrate concentration, and intracellular components are being developed, but there are many problems in terms of sterilization, reproducibility, and accuracy. Therefore, methods such as indirectly estimating state variables such as specific growth rate and metabolite concentration, which cannot be directly measured online, by Kalman filter, or by combining with the material balance are being studied. For example, as a method of indirectly estimating a state variable that cannot be directly measured online from a state variable that can be directly measured online, measure the carbon dioxide concentration in the fermentation flue gas to determine the metabolic balance based on the mass balance. It has been conventional practice to determine and indirectly estimate these non-measurable state variables. This method collects the metabolic balance of how the organic matter was used and metabolized by microorganisms, that is, at what ratio it was used for carbon dioxide, metabolites or bacterial growth based on the mass balance. Among these, by measuring the carbon dioxide generation rate that can be measured online, the amount of organic substances used, the amount of other metabolites, and the bacterial cell concentration are back-calculated and estimated. However, in order to apply these methods, it is necessary to assume that the metabolic balance based on the mass balance of carbon dioxide generation rate, bacterial cell yield, metabolites, etc. is constant during the course of the culture. When the culture state changes with time as in a culture system, the metabolic balance changes with the change, that is, it is difficult to apply the method when the production status of the metabolite or the cell yield changes. Further, in the above method, the cell yield is calculated back from the consumption of ammonia used as a nitrogen source,
When peptone / enzyme extract or the like is used as a nitrogen source, it is impossible to measure the amount of consumption, which makes it difficult to apply.

[0003]

Therefore, an object of the present invention is to perform on-line estimation of state variables such as residual substrate concentration, bacterial cell concentration, metabolite concentration, and by-product concentration, which are difficult to measure directly in culture. Fermentation that can be performed when the culture conditions change over time, such as in batch culture systems or fed-batch culture systems, and the production status of metabolites and byproducts changes, and when peptone / yeast extract is used as a nitrogen source. It is to provide a measuring method for a tank.

[0004]

Means for Solving the Problems The method for measuring a fermenter of the present invention which can solve the above-mentioned problems is to calculate a metabolic balance from the measurement of measurable metabolites in a metabolic composition to determine the inside of the culture tank. In the method of measuring the culture progress in, the culture state is classified into a plurality of specific culture states, the metabolic balance in each specific culture state is experimentally determined, and the membership function corresponding to each measurable state variable is determined. Determine the rules for identifying the culture state using, determine the fitness to the membership function of each state variable obtained from the fermenter, determine the contribution of each specific culture state from the obtained fitness, from the contribution A fermenter in which the culture state changes over time, characterized in that the ratio of the metabolic composition is determined, and the amount of the substance present in the fermenter is calculated from the measurable measured value of the metabolic composition at that time. Measurement of It is the law. The present invention also includes the above method in which the measurable metabolic composition is carbon dioxide and the above method of calculating the amount of the substance present in the fermenter at regular time intervals.

The present invention will be described in detail below. Generally, when an organic substance such as glucose is used as an energy source, a microorganism takes in the organic substance and metabolizes it to acquire energy, proliferate, and discharge carbon dioxide and metabolites. Therefore, if the ratio of the metabolized organic substances to carbon dioxide, metabolites, and bacterial growth was clarified, the metabolism calculated based on the mass balance was calculated by measuring the carbon dioxide excretion rate. It is possible to calculate the utilization amount of organic substances, the concentration of metabolites, and the bacterial cell concentration by back-calculating the balance. However, in this method, it is necessary to assume that the metabolic balance such as cell yield, metabolites, and carbon dioxide generation rate is constant during the course of culture. As described above, it is difficult to apply the method when the state variables change with time and the metabolic balance such as the metabolites and the cell yield changes as in the batch culture system and the fed-batch culture system. Therefore,
When applying this method to batch culture systems or fed-batch culture systems where state variables change over time and metabolic balance such as metabolites and cell yields changes, this culture state It is necessary to somehow estimate the change in erythrocyte and the resulting change in metabolic balance, and if it becomes possible, the scope and effectiveness of this method will be greatly improved.

The basic concept of the measuring method of the present invention is shown in FIG. First, as shown in FIG. 1, the metabolic balance such as cell yield / metabolite and carbon dioxide generation rate corresponding to each culture state is experimentally determined. On the other hand, a rule for identifying each individual culture state is created using fuzzy variables. Then, the fitness of online data in actual culture is calculated to identify which culture state or which culture state contributes at what ratio, and applies the metabolic balance of each culture state described above. Determine the metabolic balance at that time. The concentration of cells, residual substrate, and concentration of metabolites are estimated online by back-calculating the carbon dioxide generation rate obtained online.

The present invention will be described below by taking batch culture of recombinant E. coli using a temperature-sensitive expression system as an example. Here, acetic acid concentration is measured as a metabolite, which is a state variable that is difficult to measure directly, and glucose concentration and microbial cell concentration are measured as residual substrates. The productivity of gene products in a batch culture system of temperature-sensitive recombinant Escherichia coli using a complex medium is largely dependent on the induction pH. Therefore, culture p
Although H is cultivated with various changes, the growth of bacterial cells and the production of acetic acid, which is a metabolite, also differ depending on the induction conditions such as pH, and as a result, there are culture conditions having a plurality of different metabolic balances. Specifically, in the main culture system, induction p
Under the H condition, it was revealed that there are a high microbial cell yield and no metabolite formation, a low microbial cell yield, a state in which acetic acid accumulates as a metabolic by-product, and an intermediate culture state between these. In this way, in the main culture system, the metabolic balance such as the cell yield and the amount of metabolites produced changes according to the culture conditions. Therefore, when the culture conditions are examined while changing the induction conditions, the metabolic balance changes with time.
Therefore, in the method of presuming these in advance and calculating back from the carbon dioxide generation rate, the bacterial cell concentration in the culture system, the glucose concentration,
It was difficult to estimate acetic acid concentration online.
On the other hand, if it is possible to estimate what kind of culturing state and metabolic balance are present from the online data, the above method can be applied to the culturing system targeted here. Therefore, the present invention has developed a method of recognizing the culture state by fuzzy inference. We have constructed a system that indirectly estimates the bacterial cell concentration, substrate concentration, and acetic acid concentration from the carbon dioxide generation rate based on the mass balance, and online even if the metabolic balance changes during the course of culture We have developed a method that can estimate variables.

The control algorithm applied in the present invention is shown in FIG.
Shown in That is, the culture states classified in the previous section are estimated in real time using fuzzy reasoning, and the metabolic balance such as the cell yield at that time is determined according to the fitness of the culture state, and the carbon dioxide generation rate is based on the metabolic balance. Calculate the cell concentration, glucose concentration, acetic acid concentration, etc. online from the above. By repeating this operation at regular intervals, the measurement is continuously performed over the entire culture process. Here, for the sake of convenience, the state in which the microbial cell yield is high and no metabolite is produced is referred to as state A, and the state in which the microbial cell yield is low and acetic acid is accumulated as a metabolite is referred to as state B. Then, the metabolic balance based on the mass balance such as cell yield, acetic acid concentration and carbon dioxide generation rate in each typical culture state is experimentally obtained in advance. As shown in FIG. 3, the culture state was identified by the culture pH and elapsed time (state variables).
A rule for identifying the state A and the state B was created by using the membership function of, and the culture state was identified from the goodness of fit of each rule of the online data. The identification method first calculates, for each state variable, how well the online data matches the membership function of the state variable included in the rules for identifying the state A and the state B. Although two state variables are used here, the average goodness of fit of online data to each membership function is taken as the goodness of fit of the data to that rule. From the fitness, the culture state at that time, that is, the proportion of the state A and the state B contributing, is determined.
The determination of the culture state calculates the suitability of the process data with respect to the rule expressing each culture state included in the fuzzy inference system, that is, the degree to which the current online data fits to which rule. For example, state A
When the goodness of fit with respect to the rule expressing is, the culture state is completely in state A. If the goodness of fit for state A is 0.4 and the goodness of fit for state B is 0.8, it is estimated that state A is moving to state B. This goodness of fit is then used to determine the metabolic balance used in the calculation. For example, when the fitness for state A is Ada, the cell yield is Ya, the fitness for state B is Adb, and the cell yield is Yb, the cell yield Y at that time is calculated by the following formula. To be done. Y = (Ada * Ya + Adb * Yb) / (Ada + Ad
b) The metabolic balance in the culture state was determined by such a calculation, and the bacterial cell concentration, the residual substrate concentration, and the metabolic by-product (acetic acid) concentration were online estimated by back-calculating from the carbon dioxide generation rate. The above method is the procedure of the method for measuring the fermentation tank using the fuzzy rule used in the present invention.

[0009]

EXAMPLES The present invention will be described in more detail below with reference to a batch culture system using recombinant Escherichia coli. Example 1 A culture system comprising a personal computer for fuzzy monitoring and a 3 liter jar fermenter was constructed as an experimental device, and temperature, turbidity, D
Data of O, pH, CO ₂ / O ₂ concentration in exhaust gas, aeration amount, and stirring speed were measured. For online measurement of CO ₂ concentration in exhaust gas, CO ₂ / O ₂ meter (M
OLED EX-1562) was used. The sensor output was transferred to the monitoring computer via RS-232C cable at 1 minute intervals. The fuzzy reasoning program was made by myself using N88BASIC. First, regarding the metabolic balance in the main culture system, the pH suitable for growth
FIG. 4 shows a summary of carbon balance (acetic acid production amount and carbon dioxide production rate) and bacterial cell yields for two different culture conditions of condition A which is the condition of 7.2 and condition B which is suitable for expression of the product. . When calculating the carbon balance, in batch culture under aerobic conditions using a complex medium, the carbon source is mainly used for energy metabolism, and the carbon derived from the metabolized carbon source is a metabolite such as carbon dioxide and organic acid. As a result, it was supposed to be discharged to the outside of the cells. First, as a result of various studies on carbon balance in each of the states A and B, almost 100% of the glucose that was metabolized was used for energy metabolism, almost all of the glucose was excreted as carbon dioxide in the state A, and metabolism in the state B was performed. It was revealed that the product can be represented by acetic acid and is metabolized at a ratio of about 60% to carbon dioxide and about 40% to acetic acid. As a result of examining the microbial cell yield, the microbial cell yield was high in the state A, but the microbial cell growth was hardly observed in the state B, and the microbial cell yield was low. 5 to 7 show the results of online estimation of bacterial cell concentration, substrate concentration, and acetic acid concentration in batch culture of recombinant Escherichia coli under various culture conditions by these metabolic balances and the above-mentioned method. The offline analysis results are also shown. The initial glucose concentration was 20 g / liter, and in all experiments, the culture temperature was adjusted at 5 hours.
Temperature induction was performed to raise the temperature from 23.0 ° C to 42.0 ° C. Figure 5
Shows the time-dependent changes in culture pH, CO ₂ generation rate, cell concentration, substrate concentration, and acetic acid concentration in the experiment in which the present method was applied to the typical condition A of the culture conditions, together with the estimated value according to the present invention and the actually measured value. It is a figure. At this time, the solid line is the result calculated by the method of the present invention, and the large and small white circles and black circles are the measured values. Under this condition, glucose is actively metabolized, acetic acid does not accumulate, and the bacterial cell yield is high. However, as is clear from FIG. 5 (c), the bacterial cell concentration and glucose concentration are the analytical values obtained by actual measurement. It shows a good match. Also, acetic acid is not accumulated under these conditions. Regarding the bacterial cell concentration, a high concentration is shown, reflecting the conditions suitable for growth. FIG. 6 shows the estimation of the culture pH, the CO ₂ generation rate, the bacterial cell concentration, the substrate concentration, and the acetic acid concentration with time according to the present invention in the experiment in which the present method is applied to the typical condition B of the culture conditions, and the measured values are shown. FIG. At this time, the solid line is the result calculated by the method of the present invention, and the large and small white circles and black circles are the measured values. In this experiment, induction production was performed by lowering the culture pH at the same time as temperature induction, but under this condition, glucose metabolism was suppressed,
Although acetic acid accumulates and the cell concentration is low, it is possible to satisfactorily track changes in these state variables over time from the carbon dioxide generation rate for any of glucose concentration, cell concentration, and acetic acid concentration. It indicates that there is.

FIG. 7 shows the culture pH, the CO ₂ generation rate, the bacterial cell concentration, the substrate concentration, and the acetic acid concentration in the experiment in which the present method was applied to the experiment in which the culture conditions were manipulated so that the state A gradually changed to the state B. FIG. 3 is a diagram showing changes with time of the present invention, together with an estimation according to the present invention and an actually measured value. At this time, the solid line is the result calculated by the method of the present invention, and the large and small white circles and black circles are the measured values. Glucose is actively metabolized at first, but the metabolism is suppressed with the decrease of pH after induction, and the bacterial cell concentration grows actively at the initial stage but stops growing after that. It is estimated from. The acetic acid concentration also shows the progress of the culture. Further, FIG. 8 shows a change with time in the degree of conformance with respect to the rule representing each culture state in this experiment. As is clear from the figure, the state in which the culture state transitions from the state A to the state B is well represented. Online estimation from these glucose concentrations, carbon dioxide concentrations such as bacterial cell concentrations and acetic acid concentrations for culture conditions that change over time is extremely difficult to estimate by conventional methods. The method of combining the recognition of the culture state by fuzzy reasoning and the mass balance of the present invention makes it difficult to accurately measure directly even in the culture in which the culture state changes over time and the production conditions such as the cell yield and acetic acid change. It is possible to measure state variables such as glucose concentration, bacterial cell concentration and acetic acid concentration.

[0011]

EFFECTS OF THE INVENTION The culture state in the fermenter is classified into a plurality of specific culture states in advance, the metabolic balance in each culture state is determined, and the culture state at that time is determined by fuzzy inference from online data during the culture. It is possible to determine the contribution of a specific culture state, calculate the metabolic balance at that time from the contribution, and calculate the amount of metabolites in the fermentor that is difficult to measure directly online from the carbon dioxide generation rate. The culture conditions (metabolite concentration, residual air quality concentration, bacterial cell concentration, etc.) can be measured in real time.

[Brief description of drawings]

FIG. 1 is a basic conceptual diagram of a measuring method in a fermenter according to the present invention.

FIG. 2 is a diagram showing a control algorithm applied in the present invention.

FIG. 3 is a diagram showing a rule for identifying a state A and a state B using a membership function of culture pH and elapsed time.

FIG. 4 is a diagram showing metabolic balance of state A and state B.

FIG. 5 is a diagram showing the results of applying the present method to the culture conditions of state A.

FIG. 6 is a diagram showing the results of applying the present method to the culture conditions of state B.

FIG. 7 is a diagram showing the results of applying the present method to culture conditions in which the state A gradually changes to the state B.

FIG. 8 is a diagram showing a time-dependent change in fitness with respect to a rule representing a culture state.

Claims (3)

[Claims] 1. A method for calculating the metabolic balance from the measurement of measurable metabolites in a metabolic composition to measure the progress of culture in a culture tank, classifying the culture state into a plurality of specific culture states, Experimentally determine the metabolic balance in each specific culture state, and define the rules for identifying the culture state using the membership function corresponding to each measurable state variable,
The fitness of each state variable obtained from the fermentor to the membership function is determined, the contribution of each specific culture state is determined from the obtained fitness, and the ratio of the metabolic composition is determined from the contribution at that time. A method for measuring a fermenter in which the culture state changes with time, which comprises calculating the amount of a substance existing in the fermenter from the measurable measured value of the metabolic composition. 2. The method of claim 1, wherein the measurable metabolite is carbon dioxide. 3. The method according to claim 1, wherein the amount of the substance present in the fermenter is calculated at regular intervals.