counter:10587


Development of an automatic system for sampling through a gas-diffusion membrane and detection of hydrogen peroxide in the nanomolar concentration range.





John Olsson
20 poäng

Handledare: Einar Pontén




Avdelningen för analytisk kemi
Kemiska instutitionen, Umeå universitet
1999






ABSTRACT
INTRODUCTION
EXPERIMENTAL
Reagents and solutions.
Flow system
Flow cell
Sample system A B and C
Sample probe
RESULTS AND DISCUSSION
Addendum
REFERENCES



Abstract

In this work the sampling and determination of hydrogen peroxide (H2O2) by a gas diffusion sampling probe coupled to a chemiluminescence detection system has been investigated. The automatic flow system developed was intended for the monitoring of H2O2 in a cell cultivation media. Selectivity was obtained by diffusion of hydrogen peroxide through a hydrophobic Gorte-Tex gas diffusion membrane. Detection was accomplished by peroxyoxalate chemiluminescence (PO-CL) with 1,1-oxalyldiimidazole (ODI) as reagent. A solid phase reactor with immobilised 3-aminoflouranthene (3-AFA) in front of photomultiplier tube was used for detection. On-line and semi automatic detection of hydrogen peroxide in a final single-line system has successfully been applied in this work. However it has been found difficult to develop a gas diffusion probe for sampling of hydrogen peroxide in nanomolar concentration levels. At higher concentrations (>mM) this approach still may have it’s applicability, but low concentration sampling require further membrane development.


Introduction

Hydrogen peroxide can be generated in bacterological media exposed to light and/or oxygen [1].

Hydrogen peroxide is directly cytotoxic and it may not be a coincident that high concentrations of hydrogen peroxide kill cells. However, it is suggested that hydrogen peroxide can induce apoptosis (programmed cell death), therefore any enzyme system capable of generating hydrogen peroxide in a regulated fashion could be a mediator of programmed cell death. Experiments have shown that the presence of catalase in cell cultivation, significant decreases the percentage of dead cells [2]. The inhibition of growth or bactericidal effects of hydrogen peroxide is mainly depending on the concentration, therefore there is a great interest in methods for detection and quantification of hydrogen peroxide formed or present in biological systems.

Quantification of hydrogen peroxide in cell cultivation is rather difficult, due to the complex matrix of the cultivation media and the fairly low concentrations of hydrogen peroxide that induce cell death. We expect that the concentrations of hydrogen peroxide causing cell growth inhibition or apoptosis is in the sub micro molar range. Monitoring the release of hydrogen peroxide in a cell growth media at this low concentration demands a sensitive and selective method for quantification and sampling. However, the sampling procedure should not interact with the bioprocess.

Several techniques and methods for the measurement of the cellular release of hydrogen peroxide are suggested in ref [1].
(i) the change of absorbance upon oxidation of phenol red [3].
(ii) formation of enzyme-substrate complexes with peroxidases [4].
(iii) catalase-dependent oxidation of formate to release CO2 [5].
(iv) generation of fluorescent products due to a hydrogen peroxide mediated oxidative reaction [6].
(v) the loss of fluorescence upon oxidation of scopoletin [7]

Imidazole mediated peroxyoxalate-chemiluminescence (PO-CL) appears to be among the most sensitive and selective techniques for quantification of hydrogen peroxide, with limit of detection as low as 10 nM in water samples [8]. Unfortunately PO-CL is not directly suitable to cell cultivation media, which consists of a large amount of different species, that can precipitate and cause self-absorbtion of the chemiluminescent light generated. Fairly of selective sampling of hydrogen peroxide from cell media may be accomplished with the use of a microdialysis sampling technique [9]. This also provides a method of sampling that can be used continuously without affecting the biological system. Gaseous hydrogen peroxide has previous been sampled through membranes and introduced to flow systems for quantification with a detection limit 5*10-12 atm [10,11].

Therefore the use of a diffusion membrane in combination with a PO-CL detection system for monitoring the release of hydrogen peroxide in a cell growth media has been investigated.



Experimental

Reagents and solutions.

Bis(2,4,6-trichlorophenyl)oxalate was either purchased from Aldrich, 97% or synthesised by co-worker Bjarnestad using the method described in ref [12]. Oxalyldiimidazole was synthesised as described by Murata [13] using toluene instead of benzene by co-worker Jonsson. Acetonitrile (MeCN) was of Baker analysed HPLC-grade (Baker, Deventer, Holland) and was additionally dried using a 3-Å molecular sieve (KeboLab, Stockholm, Sweden). Water was purified using Milli-Q (Millipore, Bedford, MA) equipment. This water was thoroughly degassed with He, protected from light, and continuously recirculated through a column (110 mm long by 10 mm i.d.) containing manganese dioxide granulate (Aldrich; 99+%) at room temperature to reduce and maintain a low background level of hydrogen peroxide. 12.5 mM and 471 mM hydrogen peroxide stock solutions were prepared from PerhydrolÒ 30% hydrogen peroxide (Merck, Darmstadt, Germany) and stored at 5-10°C in the refrigerator. Dilute solutions were prepared in brown high-density polyethene (HDPE) bottles prior to use (Nalgene, Rochester, NY) from stock solutions. Reagent carrier solution were prepared daily, by dissolving ODI and TCPO in acetonitrile to final concentration of approximately 1 and 0.4 mM, respectively.

Methacrylacid-(3-(trimethoxysilyl)propylester) (MTP). Trimethylolpropane trimethacrylate(TRIM) and 2,3-epoxypropyl methacrylate(GMA) and Benzoin methyl ether (BME) were purchased from Aldrich. TRIM and GMA were treated with Al2O3(s) to be free from phenolic polymerisation inhibitors. Isooctane and toluene were of pa-grade and dried with 3-Å molecular sieve. 3-aminofluoranthene (3-AFA), (Aldrich, 97%).

Initial experiments on the cellgrowth media were performed by adding ODI to a mixture of cell media and 3-AFA, and the chemiluminescence was detected by a laboratory constructed detection cell.

Flow system

The chemiluminescence flow system consisted of a regent pump (Merck Hitachi 655-A reaction pump) operating at a flow rate of 0.5 or 1 ml /min. The reagent solutions were kept in a polypropylene bottle (Nalgene) fitted with a drying tube. Samples were injected using a pneumatically operated poly (ether ether ketone) (PEEK) six-port injection valve with Tefzel rotor seal (Rheodyne Model 9010, Cotati, CA) fitted with a 20 mL poly tetrafluoroethylene (PTFE) loop. The injection interval was controlled either manually or from the computerised sampling system.

Flow cell

The detector flow cell was constructed by washing, a quarts glass with NH4F. The washed glass surface was treated with MTP for 16h. The MTP treated glass was put into a PTFE mould which was filled with a He degassed mixture of BME, GMA, TRIM, isooctane and toluene, about 2, 19, 19, 30, 30 (w/w) % respectively. The mould was put into UV photo polymerisation oven (10.300 mW/cm2) for 1 h, keeping ventilation at minimum to get a porous polymer on the glass surface. The polymer was washed and dried prior to immobilisation of fluorophore. The porous polymer was suspended into a solution of dried acetonitrile and 3-AFA (5mM) over the weekend. The flow cell was washed, and mounted into a laboratory built holder. The chemiluminescence detector consisted of a photomultiplier tube (PMT), H5784,(Hamamatsu) fitted to the flow cell, operating at 600 V, controlled from a laboratory constructed control unit. The PMT output was recorded by a HP3396A integrator (Hewlett-Packard, Palo Alto, CA).

Sample system A B and C

Three sample systems were constructed, two using continuous flow through the sampling probe (A,B) and one using stopped flow in the sampling probe (C) according to Figure 1. System A and B used a Altex -XV (Alitea, Tyresö, Sweden) peristaltic pump, system A upstream the injector and B downstream the injector, a valve (Rheodyne Model 9010, Cotati, CA or an Upchurch right angle valve) was inserted between the sampling probe and the injector to allow external standards to be analysed.

The stopped flow system consisted of a Kloehn piston pump (Kloehn, Las Vegas, Nevada) with a eight port distribution port upstream the injector. An automatic 10-port valve (VICI model E10) inserted between the sample probe and injector was used to allow external standards. The piston pump and the automatic valve were both controlled by the software Winpump®.

Sample probe

The sampling probe was constructed from a piece of microporous PTFE tubing (ST-1084), and a polypropylene T-cross which were sealed with silicon tubing according to figure 2. The system setup for detection and sampling of hydrogen peroxide was tested and evaluated using water standards.

Figure 1.
Illustration of the sample system A, B and C with the PO-CL flowsystem.



Figure 2.
Gas diffusion probe, arrows indicate flow direction. membrane: 250 mm long, 1.7mm i.d.,
0.4 mm wall thickness, 30-40% porosity. PTFE tube: 1/16"



Results and discussion

Initial batch experiments showed that blank signals from the cell media were high. Never the less adding acetonitrile, used as a solvent for the PO-CL reaction, to the cell media lead to an opaque solution that had an substantial self-absorbtion. Thus direct measurements on the cell media, showed that it was necessary to use a sampling method to discriminate between the unwanted species from the cell media before detection of hydrogen peroxide by PO-CL. In addition, for long time monitoring of the hydrogen peroxide content in the cell media, an in situ sampling method is preferable. In order to allow continuous monitoring of the content of hydrogen peroxide in the cell media, the technique of gas diffusion through a membrane was adopted for further experiments.

For obvious reasons, the area to volume ratio of the sampling probe, should be kept as high as possible, in order to shift the equilibrium over the membrane. A larger volume probe is more sensitive but have lower sample through-put. The probe should be adapted to the measurement. Two probes were used, one with high bulk volume and the other according to figure 2.

The PO-CL flow system constructed had a limit of detection (LOD) of 0.03 mM hydrogen peroxide when using 20 mL injection volume of external standards. The precision was better than 5% rsd. A good precision is needed since the blank level of hydrogen peroxide normally present in lab water otherwise limits the attainable detection limit. The ODI/TCPO reagent mixture used in this work did not have the long-term stability needed for high reproducibility. However, this reagent combination has been found suitable previously and, moreover, on-line generation of ODI can easily be achieved as an alternative [14].


Figure 3.
The reagent long-term stability tested with multiple injections of 590 nM external standard.


Three different set-up’s of sampling system were investigated (see Figure 1). Systems using the pump upstream the sample probe (system A and C), may have fouled the sample probe due to the high pressure inside the membrane and hereby reduce probe efficiency. On the other hand placing the pump downstream the sampling probe (system B) may cause air leakage into the system thus fouling the detection of hydrogen peroxide. The best system constructed was the totally automatic system using stopped flow (C) in the sampling probe, due to endurance, reproducibility and indolent operation.

There was no significant differences between the three sampling systems in terms of precision. Recovery, signal from the sample probe compared with signals from an external standard, varied very much with all three systems, but system recovery were generally increased with lower flow rate and longer equilibrium time through the sampling probe (see Figure 4).

Table 1.

Recovery, LOD and %RSD for the different sampling systems.
Recovery Conc. LODprobe1a LODext1b. %Rsd2.
System A 104 %3 590 nM 150 nM 150 nM 5%
System B 10 %4 2.47 mM n.a. n.a. 10%
System C 61 % 312 nM 10nM 30 nM 5%

1: a, Estimated LOD for the prob. b, LOD for external standards
2: Relative standard deviation, external standard.
3: High recovery due to low flow in high volume probe and long equilibrium time.
4: This was one of the initial experiments using the probe, no replicate were
higher than this.
n.a.: not available


Figure 4.
Recovery, signal from the sampling probe compared with
signal from external standard, using sample system A.


There were problems getting a high recovery from the sample systems. Therefore the systems were checked for degradation of hydrogen peroxide. Degradation did not explain the low recovery. This was confirmed by comparing the response from a low concentration standard that was pumped through the inner-channel of the probe or by-passed. Instead the low recovery was due to insufficient equilibrium time for the sample probe. By using stopped flow sampling (system C) the recovery was improved, at best to 61 % for a 312 nM hydrogen peroxide standard solution. However using longer equilibrium time reduces the sample through-put and increase the consumption and potential risk for degradation of the PO-CL reagent.

Further investigations on the membrane sampling probe system, should preferable be based on an automatic stopped flow sampling system, online generation of ODI and automatic external standard injection, so an immediate calibration of the system is possible. The use of other diffusion membranes may also be a way to increase sampling efficiency but more investigations are required.

Addendum

An important experimental problem was discovered after this work had been finalised. In front of the photomultiplier tube window a piece of black protection tape was still present. Thus, that the data presented in this report should be viewed as preliminary results, to avoid exam-work ad infinitum. It is likely that sensitivity and detection limit values can be improved.



Acknowledgement

Jag skulle vilja tacka mina handledare Einar och Knut, för att de visat mej vägen, och mina medarbetare, speciellt Tobias, Malin, Wen, Anna och Camilla, för alla trevliga pratstunder och råd. Jag vill också tacka alla andra på Analytisk kemi som har hjälpt mej med det här arbetet vare sig ni vet det eller inte. Tack Solveig, utan dej är det inte värt att gå hem.


References

[1] Juven, B. J., Pierson, M. D.; Journal of food protection, 1996, 11, 1233-1241
[2] Parchment, R. E.; In Vivo, 1991, 5, 493-500
[3] Pick, E. ;Keisari, Y.; J. Immunol. Methods 1980, 38, 161-170
[4] Boveris, A.; Oshino, N.; Chance, B.; Biochem. J.; 1972, 128, 617-630
[5] Iyer, G. Y. N.; Islam, M. F.; Quastel, J. H.; Nature (London); 1961, 192, 535-541
[6] Black, M. J.; Brandt, R. B.; Anal. Biochem. 1974, 58, 246-254
[7] Boveris, A.; Martino, A.; Stoppani, A. O. M.; Anal. Biochem. 1977, 80, 148-158
[8] Stigbrand, M.;Ponten, E. ;Irgum, K.; Analytical Chemistry, 1994, 66, 1766-1770
[9] Torto, N., Thomas, L., Gorton, L., Marko-Varga, G.; Analytica Chimica Acta; 1998, 374, 111-135
[10] Zhang, G.; Dasgupta, P.,K.; Analytica Chimica Acta; 1992, 260, 57-64
[11] Stigbrand, M.; Karlsson, A.; Irgum, K.; Anal Chem 1996, 68, 3945-3950
[12] Mohan, A. G.; J. Chem. Educ. 1974, 51, 528-529
[13] Murata, S.; Chem. Lett. 1983, 1819-1820
[14] Pontén, E., Stigbrand, M., Irgum, K., Anal. Chem. 1995, 67 , 4302-4308