Analytica Chimica Acta 442 (2001) 15–23 Enzyme-based biosensor for the direct detection of fluorine-containing organophosphates A.L. Simonian a,∗ , J.K. Grimsley a , A.W. Flounders b , J.S. Schoeniger b , Tu-Chen Cheng c , J.J. DeFrank c , J.R. Wild a a Biochemistry and Biophysics Department, Texas A&M University, College Station, TX 77843-2128, USA b Chemical and Radiation Detection Laboratory, Sandia National Laboratories, Livermore, CA, USA c US Army Edgewood Chemical Biological Center, Livermore, CA, USA Received 8 December 2000; received in revised form 3 May 2001; accepted 18 May 2001 Abstract The ability of the enzyme organophosphorus acid anhydrolase (OPAA) to selectively hydrolyze the P–F bond of fluorine containing organophosphates has been used to develop a biosensor for specific detection of these compounds. Hydrolysis rate of diisopropyl fluorophosphate (DFP), paraoxon and demeton-S, by soluble and immobilized OPAA was measured. These compounds were selected as representative substrates of OPAA hydrolysis of P–F, P–O and P–S bonds, respectively. Results indicate that hydrolysis of phosphofluoridates such as DFP is dominant while hydrolysis of phosphotriesters such as paraoxon or of phosphothiolates such as demeton-S, is negligible. Two experimental approaches were used for biosensor development. In the first, OPAA was covalently immobilized on silica gel and used in batch-mode measurements with flat glass pH electrode to detect pH changes due to P–X bond cleavage. In the second approach, the enzyme was covalently immobilized to the porous silica modified gate insulator of a pH-sensitive field effect transistor (pH-FET) and changes in pH relative to a second non-enzyme coated pH-FET were measured in stop-flow mode. Concentrations of DFP down to 25 M with the glass electrode and 20 M with the pH-FET were readily detected. No sensor response was observed with paraoxon or demeton-S indicating that such OPAA-based biosensors could be useful for direct and discriminative detection of fluorine containing organophosphorus neurotoxins (such as the G-type chemical warfare agents sarin GB and soman GD) in samples also containing multiple organophosphate pesticides. © 2001 Elsevier Science B.V. All rights reserved. Keywords: Organophosphorus acid anhydrolase; OPAA; pH-FET; ENFET; Biosensor; Diisopropyl fluorophosphate; Paraoxon; Demeton-S 1. Introduction The detection of phosphofluoridate neurotoxins in field collected water and soil samples is an extremely difficult challenge. Soil and water samples are very likely to contain organophosphate pesticides due to ∗ Corresponding author. Tel.: +1-979-845-6840; fax: +1-979-845-9274. E-mail address: [email protected] (A.L. Simonian). heavy agricultural use of these compounds. It is essential that phosphofluoridate neurotoxins can be readily and unequivocally distinguished from chemically similar agricultural chemicals; ubiquitous organophosphate pesticides must never be interpreted as false positive indication of chemical warfare (CW) agents. Sensitive biosensors based on acetylcholinesterase (AChE) or butyryl cholinesterase (BChE) inhibition have been developed and used for environmental monitoring [1–5]. However, cholinesterase-based sensors 0003-2670/01/$ – see front matter © 2001 Elsevier Science B.V. All rights reserved. PII: S 0 0 0 3 - 2 6 7 0 ( 0 1 ) 0 1 1 3 1 - X 16 A.L. Simonian et al. / Analytica Chimica Acta 442 (2001) 15–23 provide no discrimination between organophosphate pesticides and organophosphate CW agents and produce uncertain and often contradictory signals in response to multiple classes of cholinesterase inhibitors (e.g. organophosphate and carbamate) [6–8]. One approach to improve the specificity of cholinesterase sensors has been to monitor inhibition with both butyryl and acetyl cholinesterase [9] or to analyze inhibition patterns obtained with several acetyl cholinesterase enzymes from diverse sources [10]. We believe the preferred strategy to overcome the disadvantages of inhibited cholinesterase-based sensors is to investigate enzymes capable of selective recognition and hydrolysis of organophosphates. For example, the well characterized metalloenzyme organophosphorus hydrolase (OPH; EC 3.1.8.1) originally isolated from Pseudomonas diminuta, is able to cleave P–O, P–F, P–S, and P–CN bonds via an SN 2-type mechanism, resulting in hydrolysis products which change solution pH [11–13]. That capability has been used to develop new biosensors for the direct detection of organophosphates based on pH monitoring of OPH activity with conventional pH electrodes, pH-sensitive field effect transistors (pH-FETs), or pH-sensitive fluorescent dyes [14–17]. This strategy can be extended to further discriminate between different classes of OP neurotoxins by monitoring additional organophosphate hydrolysis products. For example, hydrolysis of phosphofluoridates yields changes in pF as well as pH which can be detected with a fluoride specific ion-selective electrode [18]. Detection selectivity has also been enhanced by genetically modifying OPH enzyme substrate specificity [19]. Finally, OPH has also been combined with the inhibited AChE strategy to produce a multi-enzyme biosensor with the unique ability to monitor pesticide mixtures and discriminate between different pesticide classes [6]. In the present study, another well characterized hydrolytic enzyme-organophosphorus acid anhydrolase, OPAA (EC 3.1.8.2), has been investigated for direct detection of organophosphate compounds. OPAA was originally isolated from a halophilic bacterial isolate designated JD6.5 and demonstrated high levels of DFP hydrolysis [20]. The bacterial isolate was identified as a species of the genus Altermonas and DFP hydrolyzing activity has been observed in several Altermonas species [21]. Subsequently, the gene responsible for the DFP-hydrolyzing capability of the JD6.5 strain was isolated, cloned into Escherichia coli and sequenced [22]. The effects of buffer choice, pH, temperature, storage conditions and reconstitution on enzyme activity have also been reported [23]. As with OPH, the native function of OPAA is not yet known. However, OPAA has been unambiguously identified as a prolidase (EC 3.4.13.9), a type of dipeptidase hydrolyzing dipeptides with a prolyl residue in the carboxyl-terminal position [24]. Significant effort has focused upon development of an OPAA overproducing recombinant cell-line and use of OPAA for enzyme mediated nerve agent decontamination [24,25]. Fig. 1 compares the chemical structure of five organophosphate compounds; all are toxic cholinesterase inhibitors and all are hydrolyzed by OPH [12]. Diisopropyl fluorophosphate (DFP) is a phosphofluoridate containing a P–F bond. The structural similarity between DFP and the two CW nerve agents sarin and soman makes it an appropriate analogue for the development of determination and destruction technologies for these compounds. DFP is frequently used as a surrogate for the fluorine containing chemical warfare agents due to its reduced (but far from negligible) human neurotoxicity. Paraoxon and demeton-S are commercially available organophosphate pesticides. Paraoxon is representative of pesticides with a P–O bond while demeton-S is representative of pesticides with a P–S bond. One of the main parameters which characterizes a biosensor is analyte specificity. Enzymes with broad substrate specificity are suitable for detection of multiple analytes belonging to the same chemical class, while enzymes with high specificity for a single substrate are best for discriminative detection of a single target analyte. Initial investigations of OPAA substrate selectivity [20,21] indicated that the enzyme is capable of cleaving P–F bonds, while P–O or P–S bond hydrolysis is minimal. Since organophosphate pesticides, such as paraoxon and demeton-S, typically possess P–O and P–S bond structures (Fig. 1), such unique substrate preference makes OPAA very attractive for discriminative detection of fluorine containing organophosphates. In this paper, use of OPAA as the recognition component of a highly specific organophosphate biosensor was investigated. Results demonstrated that OPAA immobilized on silica gel enabled detection of DFP down to 25 mol in a batch mode measurement A.L. Simonian et al. / Analytica Chimica Acta 442 (2001) 15–23 17 Fig. 1. Chemical structures of some OP neurotoxins. system with a standard pH electrode and OPAA immobilized to the exposed gate insulator of a pH-FET enabled detection of DFP down to 20 mol. Of greatest interest was the lack of signal when either sensor system was challenged with paraoxon or demeton-S supporting the strategy of OPAA recognition for exclusive detection of fluorine containing organophosphates. 2. Experimental 2.1. Reagents and buffers Paraoxon (diethyl-p-nitrophenyl phosphate), DFP, 3-amino-propyl triethoxysilane glutaraldehyde (grade 1), and all buffer reagents (NH4 Cl, NaCl, MnCl2 ) were obtained from Sigma Chemical Company (St. Louis, MO, USA); demeton-S (O,O-dimethyl-S-2-ethylthiolethyl phosphorothioate) was from Chem Service (West Chester, PA, USA). All solutions were prepared using 18 MW cm ultrapure water (Milli-Q Plus, Millipore, Bedford, MA, USA). 2.2. Enzyme and immobilization 2.2.1. Enzyme OPAA (EC 3.1.8.2) from Altermonas sp. strain JD6.5 was isolated and purified to homogeneity [23] at the US Army Edgewood Chemical Biological Center and obtained as a lyophilized enzyme powder (0.31 mg protein/mg powder). Since OPAA activity is enhanced with ammonia-based buffers with low concentrations of MnCl2 [23], OPAA solutions were prepared in 50 mM NH4 Cl, pH 8.5, with 0.1 mM MnCl2 , 100 mM NaCl (standard buffer). 2.2.2. Immobilization on silica gel OPAA was immobilized on silica gel particles [26] by modification of the method reported previously [18]. All chemical treatments were followed by thorough rinsing with deionized water. Gel particles were cleaned by washing with 1 N HCl, then 20% H2 O2 at room temperature. Particles were silanized in a 10% aqueous solution of 3-aminopropyl triethoxysilane for 6 h at 85◦ C followed by treatment with a 10% aqueous solution of glutaraldehyde for 2 h at room temperature. Finally, 1 g of activated particles was reacted with 1 ml of OPAA (3.12 mg/ml) in standard buffer at 4◦ C for 12 h with gentle stirring. The resulting biocatalyst was stored in standard buffer at 4◦ C. 2.2.3. Immobilization on chip OPAA was immobilized to sensor chips using the same procedure reported previously [17] and almost identical to the silica immobilization described above. Enzyme (3.12 mg/ml in standard buffer) was applied to only one device gate of a dual pH-FET chip (described below) by pipetting into only one epoxy defined well. 18 A.L. Simonian et al. / Analytica Chimica Acta 442 (2001) 15–23 The immobilization was carried out at room temperature for 2 h. After extensive rinsing, chips were stored in standard buffer at 4◦ C. 2.3. Apparatus and procedures A batch mode measurement chamber (5 ml) developed at the Yerevan Physics Institute (Yerevan, Armenia) [18] was fitted with a pH (FTP-2, Lazar Research Laboratories, Los Angeles, CA, USA) and pF (Model 96-09, Orion, Beverly, MA, USA) ion-selective electrode and used to obtain kinetic characteristics of soluble OPAA. Enzyme and substrate were added to the chamber then pH or pF signals were collected by a digital pH-meter (Ion Analyzer 350, Corning, NY, USA), registered by computer (ProComm, Datastorm Technologies, Colombia, MO, USA) and analyzed (KaleidaGraph, Synergy Software, Reading, PA, USA). Sensor chips were prototype pH-FETs (SenDx Medical, Inc., Carlsbad, CA, USA). Each chip contained two discrete depletion mode, n-channel transistors with a non-metallized gate insulator stack of thermal silicon dioxide and chemical vapor deposited (CVD) silicon nitride. A porous ceramic matrix was formed at the gate surface by dip coating the chip prior to packaging into a solution of 200 nm silica microspheres formed via a sol–gel process. This procedure has been described in detail previously [17]. Chips were diced, packaged and installed in a differential control circuit [17]; a simplified schematic of the chip control circuit is shown in Fig. 2. The control circuit was configured as a classical differential pair amplifier and used to monitor the gate to source voltage of the enzyme-modified device relative to the non-enzyme-modified device. The main advantage of the differential pair is that common mode variations such as temperature and bulk pH changes are eliminated while local pH changes at the FET gate with immobilized enzyme are amplified. 2.3.1. Batch-mode assay with conventional pH electrode The temperature-controlled (25◦ C) measurement cell was filled with modified standard buffer (1 mM NH4 Cl), and the pH electrode-stopper assembly was plugged carefully to avoid air bubble formation. For analysis of soluble enzyme kinetics, substrate was injected into the measurement cell and enzymatic reactions were initiated by 10 l of OPAA (0.098 mg/ ml solution). In the immobilized enzyme experiments, 500 mg of carrier with immobilized OPAA was injected into the measurement cell and the reaction was initiated by injection of different concentrations of substrate. The velocity of pH or pF changes (in mV/s) was calculated for each concentration of substrate and the results were used for determination of kinetic parameters using the Michaelis–Menten equation V = Vmax S KM + S where V is the reaction rate, S the substrate concentration, Vmax the maximum reaction rate, and KM the apparent Michaelis constant. To remove adsorbed fluoride, the electrode, stir bar, and chamber were rinsed between calibration readings and between reactions in standard buffer with 20% methanol (v/v) followed by a thorough rinsing with methanol-free standard buffer. 2.3.2. Stop-flow assay with pH-FET The flow system consisted of a peristaltic pump and a flow cell mounted over packaged chips which provided a head space of approximately 75 l (Fig. 2). The pH response of the device was tested by Fig. 2. Schematic of the constant current differential sensor circuit. See details in Section 2.3. A.L. Simonian et al. / Analytica Chimica Acta 442 (2001) 15–23 monitoring changes in the common source voltage (Vcs , Fig. 2) with pH 4, 7, and 10 standard buffers before and after enzyme immobilization. Enzyme specific responses were measured as changes in differential signal (Vdiff , Fig. 2). A silver/silver chloride flow-through reference electrode (Microelectrodes 16-702, New London, CT, USA) was incorporated in the inlet stream manifold. Measurements were taken with flow stopped to avoid streaming potential interference. Each experimental point was the average of 3–5 measurements. 19 3.2. Batch-mode assay with conventional pH electrode Fig. 3 compares the change in pH generated by OPAA immobilized to silica gel in response to varying concentrations of each organophosphate compound tested. Response to DFP was very good; DFP concentrations as low as 20 mM were detectable and a linear response was observed from 20 to 500 mM. The response to paraoxon was extremely low, and no activity was detected for demeton-S (Fig. 3, curves 3 and 4). 3.3. Stop-flow assay with pH-FET chip 3. Results and discussion 3.1. Soluble OPAA kinetic characteristics Table 1 presents measured kinetic characteristics of OPAA, as well as previously reported characteristics of OPH [19]. The comparison of OPAA and OPH kinetic parameters emphasizes the unique recognition capability of OPAA relative to OPH. OPAA displays a strong preference for P–F bond hydrolysis making it suitable for discriminative detection of fluorine containing organophosphates, while OPH displays significant activity for both P–O and P–F bond hydrolysis making it suitable for detection of multiple organophosphates. Apparent kinetic parameters for soluble OPAA with DFP as a substrate, obtained by pH elecapp app trode, are Vmax = 2.70 ± 0.28 mV/s and Km = 0.54 ± 0.08 mM. The same parameters in the presapp ence of 1 mM paraoxon are Vmax = 2.54 ± 0.30 mV/s app and Km = 0.55 ± 0.23 mM and in the presence of app 1 mM demeton-S are 2.63 ± 0.30 mV/s and Km = 0.57 ± 0.23 mM. A typical FET-chip sensor response to different concentration of DFP is presented in Fig. 4. Detection time required for 90% of the total signal change was between 100 and 250 s, depends on concentration of DFP. Fig. 5 presents response of the OPAA-modified pH-sensitive chips as a function of substrate concentration for different substrates. As expected, based on OPAA kinetic characteristics, differences between responses to DFP and other substrates are quite large. Paraoxon generated barely measurable responses, even at high concentrations; there was no detectable signal for demeton-S. Good linearity was observed for DFP concentrations from 12.5 to 500 mM (Fig. 5b). In addition to excellent discrimination, the presence of 0.5 mM paraoxon plus 0.5 mM demeton-S had little effect on sensor response to DFP (Fig. 5). 3.4. pH dependence The previously reported pH optimum for OPAA acapp tivity (kcat ) with DFP as a substrate was 8.5, while Table 1 Comparative kinetic constants for OPH and OPAA enzymes Substrate/hydrolyzed bond Enzyme DFP [P–F]a OPAA OPH Paraoxon [P–O]b OPAA OPH Demeton-S [P–S]b OPAA OPH a b Detected by fluoride-specific electrode. Detected by pH-sensitive electrode. kcat (s−1 ) KM (mM) kcat /KM (M−1 s−1 ) 770∗ 3500 2.81 1.42 2.7 × 105 2.4 × 106 4.64 11500 0.6 0.12 7.75 × 103 9.6 × 107 3.5 4.0 8.0 1.2 × 103 0.028 4.8 20 A.L. Simonian et al. / Analytica Chimica Acta 442 (2001) 15–23 Fig. 3. Calibration curves for batch-mode biosensor with a flat pH electrode: (1) fresh immobilized OPAA with DFP as a substrate; (2) the same, after 5 months; (3) fresh immobilized OPAA with paraoxon as a substrate; (4) with demeton-S. Fig. 4. Biosensor response for 1, 0.5 and 0.25 mM of DFP. Stop-flow mode with pH-sensitive field effect transistor. Measurement condition: 1 mM NH4 Cl buffer, containing 0.1 mM MnCl2 and 100 mM NaCl. A.L. Simonian et al. / Analytica Chimica Acta 442 (2001) 15–23 21 Fig. 5. Signal versus concentration for DFP (black squares) paraoxon (inverted triangles), and mixture of 0.5 mM of paraoxon, 0.5 mM of demeton-S and DFP (triangles). (a) Stop-flow assay with pH-sensitive field effect transistor (full curves). (b) The linear ranges of the calibration curves; f (x)DFP = 43.5 + 1469x, R = 0.99976; f (x)mix = 42.5 + 1255x, R = 0.98891; f (x)PX = 13.9 + 77.1x, R = 0.85367. 22 A.L. Simonian et al. / Analytica Chimica Acta 442 (2001) 15–23 Fig. 6. pH profile for DFP biosensor response. Stop-flow mode with pH-sensitive field effect transistor. Measurement condition: 1 mM NH4 Cl buffer, containing 0.1 mM MnCl2 and 100 mM NaCl. DFP concentration 1 mM. app app the highest catalytic efficiency (kcat /Km ) was observed at pH 6.8 [20]. However, enzyme immobilization can influence enzyme operational capability and change the pH optimum, depending on which sites on the enzyme were involved in binding to a surface. To determine the pH optimum for the OPAA-modified pH-FET sensor, response to a fixed concentration of DFP (1 mM) at pH 6.2–9.0 was measured (Fig. 6). Maximum sensor response was obtained at pH 8.5, however, the sensor response was not as sensitive to pH as the soluble enzyme [20] and satisfactory signal was obtained from pH 7.5 to 9.0. 3.5. Long-term stability The enzyme stability was significantly different for the two sensor formats. OPAA immobilized on silica gel retained about 60% of the starting activity after 5 months (Fig. 3, curve 2). However, OPAA on the pH-sensitive chip lost 60% of its activity in 22 days, and there was no response after 30 days. This was especially surprising since previous results demonstrated that the silica microspheres modification of the pH-sensitive chips enhanced enzyme stability and enabled stable sensor response for more than 2 months [17]. Such dramatic differences in long-term stability may be due to the significant differences in enzyme capacity of the two sensor formats; the silica gel has a much greater surface area than the pH-sensitive chip surface. 4. Conclusions The results obtained in this study demonstrate the ability of the enzyme OPAA to serve as a highly discriminative biorecognition element for organophosphate biosensors. OPAA can uniquely hydrolyze fluorine containing organophosphates, such as DFP (k cat = 1650 s−1 ) or the chemical warfare agents sarin (k cat = 611 s−1 ) and soman (k cat = 3145 s−1 ) [24], without responding to non-fluorine containing organophosphate pesticides (this study). OPAA was tested in a batch mode system by immobilizing to silica gel and monitoring changes in solution pH as a function of DFP concentration with a standard pH electrode. In addition, OPAA was covalently immobilized to the silica microspheres modified gate A.L. Simonian et al. / Analytica Chimica Acta 442 (2001) 15–23 insulator of a pH-FET and changes in pH relative to a second non-enzyme coated pH-FET were measured in stop-flow mode. Concentrations of DFP down to 25 M with the standard pH electrode and 20 M with the pH-FET were readily detected. Little or no sensor signal was generated in response to paraoxon or demeton-S. Such excellent discrimination between compounds with such dramatic differences in application yet within the same chemical class is highly desirable. OPAA should also prove valuable in combination with other organophosphate hydrolyzing enzymes, such as OPH, so as to enable construction of an array of biosensors able to discriminate between several classes of neurotoxins. Acknowledgements The authors gratefully acknowledge the support of the US Department of Energy, Office of Nonproliferation Research and Engineering, NN-20, and the US Army Medical Research and Materiel Command (JRW). We thank Joanne Volponi of Sandia National Laboratories for chip preparation and pH testing. Sandia is a multiprogram laboratory operated by Sandia Corporation, a Lockheed Martin Company, for the United States Department of Energy under contract DE-AC04-94AL85000. References [1] G.A. Evtugyn, H.C. Budnikov, E.B. Nikolskaya, Analyst 121 (1996) 1911. [2] N. Mionetto, J.-L. Marty, I. Karube, Biosens. Bioelectron. 9 (1994) 463. [3] G. Palleschi, M. Bernabei, C. Cremisini, M. Mascini, Sens. Actuators B 7 (1992) 513. [4] P.SkladalM. Mascini, Biosens. Bioelectron. 7 (1992) 335. [5] J.J. Kulys, E.J. D’Costa, Biosens. Bioelectron. 6 (1991) 109. 23 [6] A.L. Simonian, E.I. Rainina, J. R Wild, Anal. Lett. 30 (14) (1997) 2453. [7] A. Herbert, L. Guilhermino, H.C. Da Silva De Assis, P.-D. Hasen, Z. Angew. Zool. 3 (1995–1996) 1–15. [8] M. Brufani, M. Marta, M. Pomponi, Eur. J. Biochem. 157 (1986) 115–120. [9] A.M. Nyamsi Hendji, N. Jaffrezic-Renault, C. Martelet, P. Clechet, A.A. Shul’ga, V.I. Srtkha, L.I. Netchiporuk, A.P. Soldatkin, W.B. Wolodarski, Anal. Chim. Acta 281 (1993) 3. [10] T.T. Bachmann, R.D. Schmid, Anal. Chim. Acta 401 (1999) 95. [11] D.P. Dumas, J.R. Wild, F.M. Raushel, Biotech. Appl. Biochem. 11 (1989) 235. [12] D.P. Dumas, H.D. Durst, W.G. Landis, F.M. Raushel, J.R. Wild, Arch. Biochem. Biophys. 277 (1990) 155. [13] K. Lai, N.J. Stolowich, J.R. Wild, Arch. Biochem. Biophys. 318 (1995) 59. [14] E. Rainina, A. Simonian, E. Efremenko, S. Varfolomeyev, J. Wild, Biosens. Bioelectron. 11 (1996) 991–1000. [15] A. Mulchandani, I. Kaneva, W. Chen, Anal. Chim. Acta 70 (1998) 4140–4145. [16] R. Russell, M. Pishko, A. Simonian, J. Wild, Anal. Chem. 71 (1999) 4909–4912. [17] A.W. Flounders, A.K. Singh, J.V. Volponi, S.C. Carichner, K. Wally, A.L. Simonian, J.R. Wild, J.S. Schoeniger, Biosens. Bioelectron. 14 (1999) 715–722. [18] A.L. Simonian, B.D. diSioudi, J.R. Wild, Anal. Chim. Acta 389 (1999) 189–196. [19] B.D. diSioudi, C.E. Miller, K. Lai, J.K. Grimsley, J.R. Wild, Chem.-Biol. Interact. 119/120 (1999) 211–223. [20] J.J. DeFrank, T.-C. Cheng, J. Bacteriol. 173 (1991) 1938– 1943. [21] J.J. DeFrank, W.T. Beaudry, T.-C. Cheng, S.P. Harvey, A.N. Stroup, L.L. Szafraniec, Chem.-Biol. Interact. 87 (1993) 141– 148. [22] T.-C. Cheng, S.P. Harvey, G.L. Chen, Appl. Environ. Microbiol. 62 (1996) 1636–1641. [23] T.-C. Cheng, J.J. Calomiris, Enzyme Microbial Technol. 18 (1996) 597–601. [24] T.-C. Cheng, J.J. DeFrank, V.K. Rastogi, Chem.-Biol. Interact. 119/120 (1999) 455–462. [25] T.-C. Cheng, V.K. Rastogi, J.J. DeFrank, G.P. Sawiris, Ann. New York Acad. Sci. 864 (1998) 253–258. [26] A.L. Simonian, G.E. Khachatrian, S.Sh. Tatikian, Ts.M. Avakian, I.E. Badalian, Biosens. Bioelectron. 6 (1991) 93–99.
© Copyright 2025 Paperzz