Application and automation of flow injection analysis (FIA) using fast responding enzyme glass electrodes to detect penicillin in fermentation broth and urea in human serum

An enzyme immobilization technique has been developed to determine the concentration of biological compounds. This technique has been applied to penicillinase and urease, which are crosslinked as very fine films directly onto the sensitive ends of pH glass electrodes, thereby dispensing with the need of an on-line enzyme reactor. The biosensor is incorporated in an FIA system within a magnetically stirred detection cell. Penicillin-V in fermentation broth and urea in human serum samples were detected and the results were compared with HPLC and spectrophotometric methods. On-line measurement is achieved through the automation of this FIA system.


Introduction
Samples containing penicillin or urea are frequently measured in pharmaceutical and clinical laboratories.
Several types of enzyme electrodes for the determination of penicillin in batches have been developed and are described elsewhere [10]. Penicillin detection based on the enzymic hydrolysis by coupling an immobilized enzyme reactor with an electrochemical detector have been reported [11][12][13]. In all cases the electrochemical detector was a pH electrode.
A penicillin sensor has been incorporated in an FIA system, without using an enzyme reactor, which ensured a high sample throughput [2]. This detection system has possibilities for on-line monitoring of penicillin fermentation processes and also for the routine determination of urea in body fluids.
The present paper describes the automation and application of the FIA system in which a urea electrode and a penicillin electrode are directly incorporated in a homemade stirred-flow detection cell. The sample is diluted in inlcllllnastJ ureae elctrod r|cordlr Irlstalbc pump ataor-Inrfa regul=Uon of d lztor volume Figure 2. Experimental set-up for the FIA system using fast responding enzyme glass electrodes to detect penicillin and urea.
the cell by mixing with a buffer solution (figure 2). Urease and penicillinase were immobilized onto a combined pH glass electrode. This sytem was used to detect penicillin in fermentation samples manually and urea in human serum measured by the automated system. Various parameters influencing the electrode response were studied and the performances of penicillin sensor and the urea sensor were compared.

Experimental
Hardware Detection cell and agitator The configuration is shown in figure 2. Penicillinase and urease were immobilized over a combined pH-glass electrode: type LoT 403-M8-$7 (Ingold, Paris). A homemade stirred-flow cell is used as detection chamber. PVC and PTFE were used to construct the cell; small stainlesssteel tubes were used as fluid inlets (in order to connect the tubes to the cell). The potentiometric measurements were obtained using a Radiometer priM-64 research pH meter connected to a SEFRAM-Recorder (Sefram-Servotrace, Paris) and to the computer. The detection cell was agitated by a stirrer (MIKROLAB, Aarhus, Denmark) at a moderate speed.

Injection valve and actuator
Automatic sample injection is carried out by use of a timing control pneumatic actuator connected to a fourway teflon injection valve (both Rheodyne, Cotati, USA).
The actuator permits automatic operation of the valve.

Automation elements
The analogue minivolt signal arising from the pH-meter is converted to digital equivalent by a 6Bll module mounted on a 6BP04-2 backplane. The digital I/O backplane is a 6B50-1 module which provides 24 digital I/O channels-these backplanes are from Analog Devices (Norwood, USA). The I/O channels are connected to solid-state relays plugged in a EGS08000 backplane from Celduc (Sorbiers, France). These relays control the injection valve and the pumps.
The microcomputer (Macintosh Plus) is connected to the 6BPO4 backplane via the serial port (RS232).

Software
The software is written in Quick Basic, and it is structured with subroutines for control of the injection valve and the pumps, data acquisition, calibration, and calculations. Subroutines ensure recording of the baseline, injection of sample, recording of response signal and preparation of the next sample injection. The record signals are saved, and, by means of calculation and calibration subroutine, the mV-peak height is converted to the appropriate concentration of the substrate. Four calibration samples with different concentrations are used to trace the calibration curve.
Reagents 0"01 t sodium-phosphate buffer (NaH2PO4 and Na2-HPOe.2H20, pH 6.5 from Riedel-de Hain, Germany) and 0.005 M sodium-phosphate buffer (NaHPO4 and Na2HPO4.2HO, pH 7"75, also from Riedel-de HaEn, This was similar to that for the enzyme electrode described previously [15]. However, to immobilize penicillinase, the combined pH electrode was immersed in a penicillinase solution containing 4 mg m1-1 of the enzyme. The concentration of the urease solution for electrode immersion solution was 16 mg m1-1. The diaphragm of the reference electrodes was covered by plastic clips to prevent a contact with the enzyme and glutaraldehyde solution during the immobilization period.

Flow-injection system
The manifold is shown in figure 2.

Results and discussion
Measuring pencillin in fermentation broth and urea in human serum samples is important; penicillin fermentations typically run for 10 days and a high frequency of analysis is therefore unnecessary. However, this determination is done within a large concentation range and in a complex media. During the fermentation process there is a significant change in the medium composition, whereas the pH value of the broth remains approximately constant. The concentration range of urea, however, is smaller in human serum samples and a higher sample throughput is needed for routine clinical analysis. To fulfil these requirements, injection volume, flow rate and buffer strengths were adjusted to have good reproducibility, as well as acceptable sensitivity and response time.

Determination of the penicillin-V in fermentation broth
Linearity, sensitivity and reproducibility The electrode response to penicillin-V at four different injection volumes is shown in figure 3(a). Sensitivity is improved using a large injection volume. However, higher injection volumes produce a decrease in the linear concentration range and a lower injection rate. A good compromise between adequate sensitivity and linear range of response was obtained using a sample volume of 0"1 ml, which ensures a linear range between 0 and 100 mM. Consequently, the measurements during the whole fermentation time were done without changing any operating parameter.
Theoretically, the response to pencillin should be proportional to the logarithm ofits concentration because the hydrogen ions (H+) produced from the enzymic reaction are detected by a pH electrode, the response of which is related to the Nermst equation. In the absence of any enzymic catalysis, the recorded response for a batch system reflects the pit-change between the initial signal E0 and the stationary response Estat: AN Estat E0 k log A [H+] (1) where [H+] is the proton concentration change and kl the batch pH-operation constant factor for a given temperature respectively.
Logarithmic responses have been reported for batch measurements of penicillin by different authors [10,16,17]. In all cases the penicillinase enzyme was immobilized. If the enzymatic production of H + is proportionate to the substrate concentration (first order kinetics), where k2 is a constant factor for a given temperature, the response of the electrode is proportional to the logarithm of the substrate (penicillin) concentration: where ka is the enzymic batch-operation constant factor.
However, using flow systems for potentiometric detection of penicillin, a linearity of the response-to penicillin concentration has been observed 11,13]. This can be due to the fact that the batch steady state response is not reached with the FIA-system and so the Nernst equation is no longer valid. Figures 3(b)    Results of the analysis of penicillin-V fermentation broth samples were compared with results from an HPLC method [14]. The operating conditions chosen for FIA allow measurements without any change in parameters during the whole fermentation process. The pH of the broth is regulated to about 6"5 + 0"2. Figure 5 shows the penicillin analysis for the fermentation FBO-13. The electrode response to penicillin-V is shown for two different injection volumes. The results demonstrate good agreement in biosensor responses over a wide range of concentrations with the HPLC measurements. The change of the composition of the fermentation medium with time affects the sensitivity of the biosensor. At the end of the fermentation time, the agreement between both detection methods is better for the smaller volume  injected (50 lal) in the FIA system. This could be due to reduced interference with the broth elements by the biosensor using a higher dilution rate in the detection cell.
Automatic determination of urea Linearity, sensitivity and reproducibility The electrode response to urea in the stirred-flow detection cell using the automated FIA system is shown in figure 6. Typical response peaks were recorded using a t , flow rate of 4 ml min -1 (figure 6(a)). From this measurement, the potential change AE is plotted against the urea concentration (figure 6(b)). Sensitivity is sufficient using a buffer at lower concentration, in comparison with the penicillin measurements. However, the linear range from 0 to about 10 mM is not large, but the normal urea concentrations in blood range from 2 to 8 mM [9]. Reproducibility of the peaks obtained for urea samples are shown in figure 7. The r.s.d, of the measurements was 1"94%.

Measurement of urea in human serum samples
The determination of the concentration of urea in serum samples in a clinical laboratory by a spectrophotometric method 18] and direct injection into the carrier stream of an FIA system are compared in figure 8. A good agreement is obtained between these two methods in the normal amounts of urea in blood.

Stability of the biosensors
The long-term stability of the two electrodes was examined at room temperature as shown in figure 9. After construction, two pencillinase and two urease electrodes were used in the FIA system for about 8 hours per day. Between measurements, the sensors were stored in the detection cells at room temperature. The penicillinase electrode was stable over the test period of 33 days.
However, the stability of the urease electrode due to the urease enzyme [15], is not as good. Long-term stability of the penicillinase electrode () when stored in phosphate buffer (0"01 M/0"I M NaCl, pH 6.5) and of the urease electrode A when stored in phosphate buffer (0"005 M/0"I M NaCl, pH 7"75); substrate concentration: 5 raM; temperature: 25 C. detection cell. Various parameters influencing the electrode response were studied, and the performance of the penicillin sensor and the urea sensor were compared. The various conditions required for penicillin determination during the fermentation process and urea in human serum samples are reduced or completely eliminated by the sample dilution, which takes place in the detection cell. The optimal operating conditions of the FIA system have also been established.
The FIA system described here is simple, reliable and inexpensive and well-suited to the determination of urea in human serum samples and penicillin in fermentation broth samples.