Chapter 3. Pinch-valve interface for automated sampling and monitoring of dynamic
3.3.1. Construction of the automated pinch-valve sampling system
When designing the automated device for sampling dynamic chemical processes prior to the analysis by GC-MS, we were inspired by the work conducted by Quintana and co-workers97 who developed an ingenious lab-on-valve system for determination of polychlorinated biphenyls incorporating GC apparatus as a detection tool. In another noteworthy study, Clavijo et al.98 developed a lab-on-valve microextraction system coupled with GC-MS for the determination of polycyclic aromatic hydrocarbons in water. Their automated system showed good precision in quantitative analysis (relative standard deviation (RSD) < 5%). In the current study, it was of paramount importance to take into account the compatibility of the chemical process to be assayed and the analytical requirements. It was crucial that the studied process could be conducted at preset and constant temperature with shaking/stirring. It was also necessary to make the transfer line as short as possible to reduce the delay time between the sample collection and injection. Finally, it was vital that a defined aliquot of the sample could be introduced to the injection port of GC apparatus at pre-defined time points. At the same time, unlike in the work by Quintana and co-workers,97
here it was
not critical to conduct sample preconcentration, which would inevitably slow down the analysis of the consecutive sample aliquots obtained from a dynamic system (e.g. reaction mixture).In the final version of the system, we used polymer (Tygon, silicone, PTFE) and silica tubing to aspirate small aliquots of the studied medium and transfer them hydrodynamically to the injector of the GC-MS instrument. The hydrodynamic flow was exerted by a peristaltic pump. The Reynolds numbers along the flow line are in the order of 10-20 (as calculated for 100% acetonitrile), which points to predominance of laminar flow in the sample conduit.
Directing the sample to the GC injector was facilitated by two pinch valves (NO and NC;
Figure 3.1). Efficient control of the flow was achieved by means of the electronic
40
microsystem triggering five relays (see section 3.2.3., Figure 3.2). All the operations of the system were programmed in a C-language script, and executed according to the optimized time schedule (Table 3.1). After setting up, the operation of the system was triggered by pressing a single button (Figure 3.2). The samples were then injected, and GC-MS signals were automatically recordedand saved in a sequence for post-run processing. The seamless operation of the system significantly minimized the extent of mechanical and manual tasks to be performed during the analysis.
3.3.2. Characterization of the automated pinch-valve sampling system
Figure 3.3 and Table 3.1 present the operation steps of the pinch valves during sample
injection. In a nutshell, the 1st step encompasses withdrawal of an aliquot of the liquid sample from the sample chamber. During this step, the transfer tubing is filled with a fresh portion of sample, and any excess sample is diverted to waste. Considering the applied flow rate (1.25 μL s-1) and the sampling time (79 s; Table 3.1), the volume of sample withdrawn in this step is estimated to be 99 μL. In the 2nd step, the sample is directed to the injection port of the GC apparatus during 28 s, which contributes to the withdrawal of further 35 μL of
sample from the sample vial. Eventually, in the 3rd step, the sample is pushed out toward the liner during 6 s, which adds 8 μL to the volume of sample aspirated by the sampling
capillary (see the following paragraph for further discussion of the injection volume). In the 4th step, 50 μL of the “old” sample are returned to the sample chamber, and a small volume of fresh sample (5th step: 15 s, 19 μL) is immediately introduced to the flow line. Therefore, 111 μL of the sample solution are consumed in every analysis, most of which is discarded to
waste, only a small fraction being dispensed to the GC column inlet. The purpose of withdrawing larger sample volumes than it is actually needed for analysis is to rinse away any residues of the previously analysed sample aliquots from the system. It should be borne inSeveral experiments were conducted to verify injection accuracy and precision of the proposed device. In an offline experiment, an analytical balance was used to determine the
41 injection. Therefore, we also evaluated repeatability of the operation of the automated device after coupling it with the GC-MS instrument (Figure 3.1). The RSD of the peak areas of a test analyte (1-butyl acetate in the mixture: acetonitrile : 1-butanol : isopropenyl acetate = 90 : 9 : 1 (v/v/v)) recorded during 10 consecutive runs was 27%. When considering the peak areas of the test analyte referenced to an internal standard (D-limonene), the RSD decreased to 13%
(Figure 3.4, EICs for analyte, m/z 43 0.5 and internal standard, m/z 93 0.5 u e-1); this was deemed acceptable for further applications. To measure the accuracy of the digital control of sample injection connected to electromechanical actuators (relays, pinch valves), we carried out the following experiment: a video/audio sequence was recorded to capture the action of the relays and the acoustic sounds produced by the relays and the pinch valves. Using a video/audio processing software (Ulead Video Studio, ver. 10.0.0110.0 SE DVD; Ulead Systems, Taipei, Taiwan), we estimated the time gaps between the acoustic signals marking the beginning and the end of the injection. The average interval between the “clicks” of the pinch valves was estimated to be 6.0 0.0 s (n = 3). Considering that the injection time was set to 6 s (in the programme loaded into the Raspberry Pi microcomputer), this result shows that the proposed device provides an accurate and precise control over the injection time.
However, other sources of inaccuracy and imprecision exist; they are believed to be related to the mechanic actuation of the peristaltic pump, tube plasticity, and other causes. In another experiment, we estimated the actual injection volume obtained when the automated injection system was coupled with the GC-MS apparatus (Figure 3.5): 1.7 0.4 μL (n = 3; based on comparison of peak area obtained using the automated system with the peak areas obtained by performing manual injections of different volumes of the standard sample). We think that the lower injection volume in this on-line experiment is due to the influence of gas pressure (10 kPa) inside the liner of the GC-MS instrument. It can be concluded that although the RSDs observed in this study were slightly higher than those obtained with lab-on-valve systems, the injection repeatability can be regarded as satisfactory. In order to bring the RSD
42
values down, and obtain reliable analytical data, it is certainly required to use internal standards (spiked to the samples).
To further characterise the performance of the automated sampling/analysis system (Figures 3.1 and 3.2), we verified its quantitative capabilities. Figure 3.6 presents calibration plots for 1-butyl acetate, D-limonene,
-pinene, and -terpinene. All these compounds are
relevant considering the foreseen applications of the system (see sections 3.3.3. and 3.3.4.).The coefficients of determination in these plots are within the range: 0.944 – 0.989 (Table 3.2); which is believed to be acceptable for the anticipated applications. The increased standard deviation – observed at higher concentration values (Figure 3.6) is believed to be caused by inaccuracy of curve fitting while measuring areas of highly deformed peaks. The range of the assay was narrowed down to maintain the coefficients of determination at a high level (> 0.9; Table 3.2).
Figure 3.3 Schematic diagram of the injection sequence executed using the automated device (cf. Figures 3.1 and 3.2) incorporating two pinch valves (NO – normally open; NC – normally closed). Labels: S – sample, W – waste, C – column.
43
Figure 3.4 Injection repeatability test. EICs (analyte m/z 43 0.5 u e-1, internal standard m/z, 93 0.5 u e-1) from 10 consecutive analyses carried out using the automated sampling system (Figures 3.1 and 3.2) combined with GC-MS instrument. Sample: 86.1 μM 1-butyl acetate, 36.7 μM D-limonene (in acetonitrile).
44
Table 3.1 Operation sequence of the custom device for the introduction of liquid samples to the injector of gas chromatography apparatus using pinch valves.
Figure 3.5 Determination of the injection volume using the automated sampling/injection system. The black solid markers correspond to the manual injections (without the automated system). The black solid line is a linear function fitted to those data points. The red line extrapolates the peak area (EIC analyte m/z 43 0.5 u e-1, internal standard m/z, 93 0.5 u e-1) obtained with the automated injection system (n = 3) to the volume axis. The concentration of the 1-butyl acetate standard was 0.5 μM in both cases. Each data point in this graph corresponds to arithmetic average of two replicate results.
45
Figure 3.6 Quantitative capabilities of the proposed automated sampling system (Figures 3.1 and 3.2) coupled with GC-MS. (A) Calibration plot for 1-butyl acetate (D-limonene as internal standard). (B) Calibration plot for D-limonene, (C) -pinene and (D) -terpinene (thymol as internal standard). Error bars correspond to standard deviations (n = 3). For calibration equations, see Table 3.2
46
Table 3.2 Calibration equations (cf. Figure 3.6) and limits of detection (LODs) for the four analytes discussed in this report. The LODs were calculated using the S/N = 3 criterion, where S is the peak amplitude and N is the RMS noise of baseline. The injection volume resulting from the estimation described in section 3.3.2.was used to calculate LODtot.. Note well, the conditions used to analyse the first (1-butyl acetate) and the other three (limonene, pinene, terpinene) compounds are different (cf. section 3.2.4.).
Compound
(0.05500.5481) 0.989 0.0005-0.0400 1.97×10-4
±2.48×10-5
0.03730.0198) 0.944 0.0001-0.0600 2.96×10-5
±6.89×10-6
(0.008300.00510) 0.965 0.0001-0.0600 1.50×10-4
±5.93×10-5
(0.006600.00490) 0.949 0.0001-0.0600 9.66×10-5
±1.92×10-5
1.64×10-10
±3.26×10-11
47
3.3.3. Application in the monitoring of single-microbead biocatalysis
While many assays for lipases (e.g. spectrophotometry-based)99 focus on the hydrolytic activity of these enzymes,100 for practical applications it is desirable to probe their transesterification activity.72, 101
Therefore, to exemplify capabilities of the proposed
automated system (Figures 3.1 and 3.2) in bioanalysis, first we applied it in the monitoring of a transesterification reaction catalyzed by lipase immobilized on macroporous resin microbeads:(eq. 3.1)
It should be noted that transesterification is a crucial step in the production of biodiesel fuel;
therefore, it is of considerable relevance when it comes to harnessing green energy.16 From the model transesterification reaction depicted in eq. 3.1, it is evident that transferring acyl group from isopropenyl acetate onto the molecule of 1-butanol leads to formation of a new ester (1-butyl acetate) and a by-product (acetone). Using the GC method described in section 3.2.4., it is possible to separate 1-butyl acetate from all other components of the reaction mixture (Figure 3.7)
We further applied the proposed system in the monitoring of transesterification catalyzed by varied numbers of lipase microbeads (n = 1, 2, 3, 4, 5, and 10). As expected, the enzyme progress curves obtained for different numbers of microbeads have different slopes (Figure
3.9). Following 126-min incubation, the relative reaction yield – as expressed by the ratio of
the product 1-butyl acetate area vs. the internal standard (D-limonene) peak area – recorded for 10 enzyme microbeads – is highest (Figure 3.9, dark blue diamonds). Notably, the relative reaction yield – recorded for 1 enzyme microbead (Figure 3.9, red circles) – is only slightly higher than the blank (reaction mixture incubated without biocatalyst microbeads; Figure 3.9, black triangles). Nonetheless, it is pleasing to note that the amount of product, obtained in the presence of individual microbeads, is detectable. Motivated by this success, we carried out a comparative study of biocatalytic performance of single enzyme microbeads. As shown inFigures 3.8 and 3.10, the activities of individual microbeads (n = 8) are not exactly the same.
The reaction velocities (calculated by fitting linear functions to the data sets in Figure 3.10
48
and feeding the slopes into the calibration equation from Table 3.2) range from 308 to 432 nmol L-1 min-1 (Table 3.3).
Note well that the 126-min period corresponds to the initial stage of reaction since the concentration of isopropenyl acetate is in the order of 0.1 M, and only few percent of it are depleted (assuming 1:1 reaction stoichiometry, eq. 1). Interestingly, there is no evident correlation between the relative reaction yields (for the 126 min time point) and the specific dimensions of the studied microbeads (diameter, outer surface area, volume; Figure 3.11). In fact, one outlier point in Figure 3.11 indicates the incidence of particularly large microbeads with disproportionately low enzymatic activity within the microbead population. This unexpected result can be explained with the anticipated heterogeneous immobilization of lipase molecules on the surface of the acrylic resin. This observation was only possible when conducting single-microbead assay. It can be of use whenever it is required to probe transesterification activities of very small samples of heterogeneous biocatalysts. The heterogeneity of lipase microbeads indicates that multiple microbeads need to be tested one-by-one before a conclusion on the overall quality of the “microbead population” can be drawn.
In conclusion, the proposed method enables observation of catalytic polydispersity of immobilized enzymes – adding to the conventional bioanalytical toolkit used in biotechnology.
Unlike a previously published method,6 the current one enables temporal monitoring of transesterifications catalyzed by single microbeads, and there is no need that the product absorbs ultraviolet (UV) light. Thus, it can be noticed that the analytical approach presented in this work is different than many conventional assays. The current method targets transesterification activity of relatively small samples of heterogeneous biocatalysts (microbeads < 1 mm). Standard methods often target larger samples, focus on homogeneous catalysis, or probe hydrolytic activity (as opposed to transesterification activity of lipase).
While GC-MS is an established analytical platform, the current method enables convenient coupling of the GC-MS instrument with a reaction system. Nonetheless, it should also be pointed out that the use of GC-MS apparatus – as opposed to the use of UV-Vis spectrophotometer – makes the overall cost of the current method higher. However, the automated system proposed here does not add much to the cost of the existing GC-MS instruments because the sampling device (Figures 3.1 and 3.2) has been constructed using inexpensive parts. It is estimated that the capital cost of the proposed sampling technology is
10 lower than the cost of a commercial autosampler. Near-infrared spectroscopy is yet
another technique for real-time monitoring of transesterifications.102, 103 It is envisaged that49
these optical methods – along with the “automated-sampling GC-MS approach” – will form a part of modern analytical toolkit for biocatalyst screening.
Figure 3.7 Full range GC-MS chromatograms (total ion currents, TICs, m/z range: 42-250 u e-1) of the reaction (cf. eq. 3.1) mixture (5 mL; acetonitrile : 1-butanol : isopropenyl acetate = 90 : 9 : 1 (v/v/v)). (A) Blank containing the internal standard (no microbeads added). (B) Sample after 126-min incubation using 10 microbeads of lipase (see section 3.2.1.). Temperature: 30 C. Shaking speed: 20 rpm. The peak of the reaction product (1-butyl acetate) can be seen in (B). Internal standard: 10-5 M D-limonene.
50
Figure 3.8 Raw EICs (analyte m/z 43 0.5 u e-1, internal standard m/z, 93 0.5 u e-1) for the data in Figure 3.10 (analyses at 126 min). Conditions are the same as in Figure 3.7. Asterisk (*) indicates a contaminant peak which has a longer retention time when the amount of the reaction product is higher than usual. Two asterisks (**) indicate a fronting feature which is most probably related to an injection artifact.
51
Figure 3.9 Synthesis of 1-butyl acetate catalyzed by small number of macroporous resin microbeads with immobilized lipase (from Candida antarctica; expressed in Aspergillus oryzae) monitored by the setup shown in Figures 3.1 and 3.2. Size range of microbeads: 400-600 μm. Markers: (, black triangle) 0 microbeads; (, red circle) 1 microbead; (, blue square) 2 microbeads; (, green tilted triangle) 3 microbeads; (, green reversed triangle) 4 microbeads; (, violet tilted triangle) 5 microbeads; (, dark blue diamond) 10 microbeads. Conditions are the same as in Figure 3.7. All the data point had been subtracted with the peak area ratio value at time “zero”.
Figure 3.10 Single-microbead biocatalysis. Each curve corresponds to one lipase microbead. For EICs (analyte m/z 43 0.5 u e-1, internal standard m/z 93 0.5 u e-1) of the analyses at 126 min, see Figure 3.8. For the calculated reaction rates, see Table 3.2. Conditions are the same as in Figure 3.7. All the data points had been subtracted with the peak area ratio value at time “zero”. The data marked with red circles (, red circle) correspond to the same experiment as the one depicted in Figure 3.9. The inset shows the sample vial with a reaction mixture and one lipase microbead (indicated with yellow arrow).
52
Figure 3.11 Dependence of the relative yield of the enzymatic reaction (after 126 min) on physical dimensions (diameter, outer surface area, volume) of the analyzed lipase microbeads according to the results of single-microbead assay (Figures 3.8 and 3.10). Peak areas were measured based on the EICs (analyte m/z 43 0.5 u e-1, internal standard m/z, 93 0.5 u e-1).
Table 3.3 Calculation of the lipase-catalyzed transesterification velocities obtained during the single-microbead transesterification experiment (Figure 3.10) facilitated by the proposed automated system (Figures 3.1 and 3.2).
Microbead symbol (cf. Figure 3.10)
Fitted line A – area ratio t – time / min
Reaction velocity / nmol L-1 min-1
1, A = (0.01900.0022)t 308
2, A = (0.02080.0014)t 315
3, A = (0.02410.0012)t 329
4, A = (0.02930.0030)t 351
5, A = (0.03920.0007)t 392
6, A = (0.03430.0020)t 372
7, A = (0.03780.0023)t 386
8, A = (0.04870.0029)t 432
53
3.3.4. Application in time-resolved extraction of plant tissue samples
To further demonstrate the capabilities of the proposed automated sampling system, we applied it in time-resolved monitoring of the extraction of a real sample (fresh peel of citrus fruit). In this experiment, the sample vial was filled with 5 mL of pure acetonitrile, incubated at 25
C, and shaken. At time “zero”, a small fragment of freshly obtained fruit peel was
inserted to the vial, and the monitoring started. As the chromatograms were recorded, the increasing signals of certain plant metabolites could readily be observed. For example, the on-line extraction of lemon peel led to the emergence of several signals, including those identified as limonene, pinene and terpinene (Figures 3.12 and 3.13), which are abundant ingredients of citrus fruit.104, 105 In the case of lemon, the signals of the three identified metabolites became strong after 30 min extraction, and they were constant till the end of the experiment (Figure 3.12A). In the case of the kumquat sample, a substantial amount of limonene was extracted immediately on contact with extractant (Figure 3.12B). The time-resolved extraction profiles provide information on the availability of target analytes for extraction. In fact, a previous study – conducted in this laboratory (N.B. also utilizing universal electronic modules) – clearly demonstrated the usefulness of time-resolved extraction monitoring to reveal extractability of sample/specimen constituents.106 Although no generalizations can be made based on the limited number of real samples, the current result proves that the method can reveal differences in extractability of selected plant tissue specimens. The extraction monitoring in real time – enabled by the proposed analytical system – can further facilitate optimization of extraction conditions by reducing the number of manual operations to minimum.Figure 3.12 Monitoring extraction of real samples in real time. Time evolution of the extraction profiles.
Extraction solvent: 5 mL acetonitrile. Temperature: 25 C. Shaking speed: 20 rpm. Internal standard:
10-5 M thymol. (A) Lemon peel (m = 3.81 mg): (●) limonene, (▓) pinene, (▲) terpinene. (B) Kumquat peel (m = 4.00 mg): (●) limonene. The analyses were conducted in duplicate and representative results are displayed.
54
Figure 3.13 Monitoring extraction of the lemon peel sample in real time. Extraction solvent: 5 mL acetonitrile.
Temperature: 298 K. Shaking speed: 20 rpm. Internal standard: 10-5 M thymol. (A) EIC at m/z 93 0.5 u e-1 obtained at time “zero” (right after inserting the sample to the extraction solvent), and (B) EIC at m/z 93 0.5 u e-1 obtained at a later stage of extraction (140 min). These mass spectra are from the same experiment as the one illustrated in Figure 3.12A.
3.4. Summary
We have demonstrated a simple device for automated sampling of dynamic chemical processes, and instantaneous analysis by GC-MS. The device coupled with a GC-MS apparatus is easy to use. It also offers the advantage of reducing the extent of mechanical operations during sampling and analysis. As exemplified in this report, it enables analyses of dynamic processes such as enzymatic reaction and extraction. The optimum conditions for these processes (e.g. temperature, shaking) can readily be achieved and maintained. One disadvantage of using this system is that the septum of the injector has to be replaced often in order to prevent leakage of the carrier gas. This problem can readily be rectified by substituting septum with a metal plate to which the needle of the injection system could be soldered permanently. Another consideration is that the current setup is mainly applicable to
55
short term monitoring since a few hundred microliters are withdrawn from the sample chamber – depleting the medium of reaction or extraction. This issue can be mitigated by using larger volumes of reaction/extraction solvents, in which case the depletion of solvent during sampling would only have negligible effect on the analysis result. To further improve the quantitative aspects of the presented method, it is suggested that isotopically labeled internal standards are used as additives to the reaction or extraction mixtures. Nonetheless, when doing so, attention must be paid so as to avoid experimental artifacts (by possible shifting of the reaction/extraction equilibria). Overall, we envisage that the proposed system can further be used for monitoring, optimization, and mechanistic studies on other chemical and biochemical processes. Following additional alterations to the valve unit assembly and the program, one might also consider coupling this system with other types of analytical instruments (for example, mass spectrometers, without separation systems), which would increase the range of potential applications. It is also interesting to note that – in this study – we have demonstrated the usefulness of a popular microcomputer (Raspberry Pi) in the construction of automated analytical systems, thus expanding the application realm of universal electronic modules in chemistry laboratories. Down this path, we believe that such miniature electronic systems will soon become an indispensable component of the instrumental tacklebox in chemistry, enticing industrious analysts into development of automated systems for a wide range of analytical tasks.
56
Chapter 4.
Summary and conclusions
In the two studies described above, mass spectrometry-based methods have been developed and used in the enzymatic reaction monitoring.
In the first study (Chapter 2), an atmospheric pressure ionization mass spectrometric method was implemented in the analysis of a bienzymatic reaction in a high aspect ratio cell.
Using this simple analysis system, we demonstrated chemical signal transduction due to the progress of the enzymatic reaction in real time. The non-labelled and isotopically labeled ATP was used to trigger the reaction. This allowed us to disentangle the contributions of passive and active transport. The use of isotopic label helped to overcome the repeatability problems associated with experimental instabilities. The results shown in this thesis may give the aspect
Using this simple analysis system, we demonstrated chemical signal transduction due to the progress of the enzymatic reaction in real time. The non-labelled and isotopically labeled ATP was used to trigger the reaction. This allowed us to disentangle the contributions of passive and active transport. The use of isotopic label helped to overcome the repeatability problems associated with experimental instabilities. The results shown in this thesis may give the aspect