Labeless Immunosensor Assay for the Stroke Marker Protein

Electrodes containing the antibodies were exposed to solutions of NSE and ... Further work by the same group gave rise to reports of a reversible real...
0 downloads 4 Views 183KB Size
Anal. Chem. 2008, 80, 9411–9416

Labeless Immunosensor Assay for the Stroke Marker Protein Neuron Specific Enolase Based upon an Alternating Current Impedance Protocol Andrew C. Barton, Frank Davis, and Se ´ amus P. J. Higson* Cranfield Health, Cranfield University, Silsoe, Beds, MK45 4DT, U.K. This paper describes the development and characterization of a label-less immunosensor for neuron-specific enolase (NSE) and its interrogation using an ac impedance protocol. Commercial screen-printed carbon electrodes were used as the basis for the sensor. Poly(1,2diaminobenzene) was electrodeposited onto the sensorss and this modified surface was then sonochemically ablated to form an array of micropores. A second electropolymerization step was then used to deposit conductive polyaniline within these pores to give a microarray of polyaniline protrusions with diameters of several µm. This array was then utilized as a substrate to immobilize a biotinylated antibody for NSE using a classical avidin-biotin approach. Electrodes containing the antibodies were exposed to solutions of NSE and interrogated using an ac impedance protocol. The real component of the impedance of the electrodes was found to increase with increasing concentration of antigen. Control samples containing a nonspecific IgG antibody were also studied and calibration curves obtained by subtraction of the responses for specific and nonspecific antibody-based sensors, thereby accounting for and eliminating the effects of nonspecific adsorption of NSE. A linear relationship between the concentration of NSE in buffer solutions from 0 to 50 pg mL-1 and the impedimetric response was observed. The principle of immunoassays was first established by Yalow and Berson1 in 1959. Their work led to the development of the widely used radioimmunoassay to examine the properties of insulin-binding antibodies in human serum, using samples obtained from subjects that had been treated with insulin. Within unconnected work, Clark and Lyons2 in 1962 pioneered the concept of a biosensor. The original methodology involved immobilizing enzymes on the surface of electrochemical sensors so as to exploit the selectivity of enzymes for analytical purposes. The incorporation of antibodies into conducting polymer films was first reported3 in 1991. Antihuman serum albumin (anti-HSA) was incorporated into a polypyrrole film, which was galvanostatically polymerized onto a platinum wire substrate. When the * To whom correspondence should be addressed. Fax (+44) 01525 863433. E-mail: [email protected]. (1) Yalow, R. S.; Berson, S. A. Nature 1959, 184, 1648–1649. (2) Clark, L. C.; Lyons, I. R. Ann. N. Y. Acad. Sci. 1962, 102, 29. (3) John, R.; Spencer, M.; Wallace, G. G.; Smyth, M. R. Anal. Chim. Acta 1999, 249, 381–385. 10.1021/ac801394d CCC: $40.75  2008 American Chemical Society Published on Web 11/14/2008

pyrrole anti-HSA electrode was exposed to 50 µg mL-1 HSA for 10 min, a new reduction peak was observed at a potential of approximately +600 mV versus Ag/AgCl. Further work by the same group gave rise to reports of a reversible real-time immunosensor.3 Other early work utilized a pulsed amperometric detection technique for other analytes, including p-cresol,4 thaumatin,5 and polychlorinated biphenyls.6 Since this early work there has been burgeoning interest in the development of electrochemical immunosensors sas detailed in several recent reviews.7-9 Antibody-antigen interactions are by their very nature complex, and the reproducible response characteristics of immunosensors require that the affinity reaction is minimally perturbed by the fabrication procedure. We have previously shown that up to 2-3 µg of antibodies for bovine serum albumin (BSA) and digoxin may be successfully incorporated into conducting polymer films by entrapment in a growing polypyrrole film with no detrimental effect to antibody activity.10 Electrochemical interrogation of these films demonstrated selective interactions with the target antigens. Further work utilized an ac impedance protocol11 as the method of interrogation for these filmssand led to the development of immunosensors for digoxin and bovine serum albumin. Later work by our group studied approaches for immobilization of antibodies onto polyaniline-coated screen-printed carbon electrodes utilizing a classical avidin-biotin chemistry. This enabled the construction of immunosensors for the fluoroquinolone antibody ciprofloxacin12 and myelin basic protein.13 Our group has also pioneered the development of sonochemically fabricated microarrays of conductive polymers,14 the sche(4) Barnett, D.; Laing, D. G.; Skopec, S.; Sadik, O. A.; Wallace, G. G. Anal. Lett. 1994, 27, 2417. (5) Sadik, O. A.; John, M. J.; Wallace, G. G.; Barnett, D.; Clarke, C.; Laing, D. G. Analyst 1994, 119, 1997–2000. (6) Bender, S.; Sadik, O. A. Environ. Sci. Technol. 1998, 32, 788–797. (7) Rodriguez-Mozaz, S.; de Alda, M. J. L.; Barcelo, D. Anal. Bioanal. Chem. 2006, 386, 1025–1041. (8) Diaz-Gonzalez, M.; Gonzalez-Garcia, M. B.; Costa-Garcia, A. Electroanalysis 2005, 17, 1901–1918. (9) Cosnier, S. Electroanalysis 2005, 17, 1701–1715. (10) Grant, S.; Davis, F.; Pritchard, J. A.; Law, K. A.; Higson, S. P. J.; Gibson, T. D. Anal. Chim. Acta 2003, 495, 21–32. (11) Grant, S.; Davis, F.; Law, K. A.; Barton, A. C.; Collyer, S. D.; Higson, S. P. J.; Gibson, T. D. Anal. Chim. Acta 2005, 537, 163–168. (12) Garifallou, G. Z.; Tsekenis, G.; Davis, F.; Millner, P. A.; Pinacho, D. G.; Sanchez-Baeza, F.; Marco, M.-P.; Gibson, T. D.; Higson, S. P. J. Anal. Lett. 2007, 40, 1412–1442. (13) Tsekenis, G.; Garifallou, G. Z.; Davis, F.; Millner, P. A.; Gibson, T. D.; Higson, S. P. J. Anal. Chem. 2008, 20, 2058–2062. (14) Higson, S. P. J. Sensor. International Patent PCT/GB96/0092, 1996.

Analytical Chemistry, Vol. 80, No. 24, December 15, 2008

9411

Figure 1. Formation of polyaniline microarrays: (a) deposition of insulating layer, (b) sonochemical formation of pores, (c) polymerization of aniline, (d) schematic of antibody modified electrodes, and (e) commercial screen-printed electrodes used within this work.

matic for the formation of which is shown within Figure 1a. Poly(1,2-diaminobenzene) can be electrodeposited on a variety of conductive surfaces to form an insulating layer.15 This process is self-limitingsand thus reproducible. Sonochemical ablation is then used to ablate or “drill” holes in this insulating material with sizes of 0.1 to several micrometers and a density of up to 120 000 pores cm-2. These micropore arrays have been used in the detection of aqueous chlorine.16 Further electropolymerization reactions can then be used to generate arrays of conducting polyaniline protrusions, of either just the polymer or also containing entrapped biological species.17 Previous work within our group has utilized these microarrays containing entrapped enzymes for the amperometric detection of glucose17,18 and alcohol19 and have also been utilized for the determination of organophosphate pesticides20,21 with extreme sensitivity (10-17 M). We have also applied these arrays to the formation of immunosensors for prostate specific antigen (PSA)22 via either the entrapment of antiPSA during the electrochemical process or alternatively via the immobilization of the anti-PSA in a controlled manner at the surface of a polyaniline microarray. Use of this procedure led to a thousandfold increase of sensitivity to the antigen compared to the entrapment procedure.22 Immunosensors for the stroke marker protein S-100 [β] have also been developed via this process.23 Stroke or “brain attack” is clinically defined as a rapidly developing syndrome of vascular origin that manifests itself in (15) Myler, S.; Eaton, S.; Higson, S. P. J. Anal. Chim. Acta 1997, 357, 55–61. (16) Davis, F.; Collyer, S. D.; Gornall, D. D.; Law, K. A.; Mills, D. W.; Higson, S. P. J. Chim. Oggi/Chem. Today 2007, 25, 28-31. (17) Barton, A. C.; Collyer, S. D.; Davis, F.; Gornall, D. D.; Law, K. A.; Lawrence, E. C. D.; Mills, D. W.; Myler, S.; Pritchard, J. A.; Thompson, M.; Higson, S. P. J. Biosens. Bioelectron. 2004, 20, 328–337. (18) Myler, S.; Davis, F.; Collyer, S. D.; Higson, S. P. J. Biosens. Bioelectron. 2004, 20, 408–412. (19) Myler, S.; Davis, F.; Collyer, S. D.; Gornall, D. D.; Higson, S. P. J. Biosens. Bioelectron. 2005, 21, 666–671. (20) Pritchard, J. A.; Law, K. A.; Vakurov, A.; Millner, P. A.; Higson, S. P. J. Biosens. Bioelectron. 2004, 20, 765–772. (21) Law, K. A.; Higson, S. P. J. Biosens. Bioelectron. 2005, 21, 1914–1924. (22) Barton, A. C.; Davis, F.; Higson, S. P. J. Anal. Chem. 2008, 20, 6198– 6205. (23) Barton, A. C. Ph.D. Thesis, Cranfield University, 2008.

9412

Analytical Chemistry, Vol. 80, No. 24, December 15, 2008

focal loss of cerebral function. Stroke is the most common devastating neurological disease, the second most common cause of neurological disability,24 and the third leading cause of death in the United States and the UK after heart disease and all cancers.25,26 The importance of promptly diagnosing stroke after the first symptoms are oberved cannot be overstated. Delays in diagnosis and medical intervention beyond 3 h after stroke onset may contribute to clinical deterioration and disability. An early diagnosis enables doctors to more effectively choose appropriate emergency intervention such as antiplatelet or neuroprotective therapy as well as also facilitating better prognoses of disease outcome. The delay in achieving an accurate and certain diagnosis wastes the limited amount of time available in which the brain can respond to reperfusion and this significantly increases the risk of hemorrhage even after most of the permanent injury has occurred.27-30 It therefore follows that the successful treatment of stroke requires rapid diagnosis. Accordingly, there is a need for a rapid, sensitive, and specific diagnostic assay for stroke events to help identify those individuals at risk for delayed neurological deficits. The main focus of recent research relates to the identification and use of diagnostic markers for stroke and cerebral injury. Enolase is a 78-kDa homo- or heterodimeric cytosolic protein produced from [R], [β], and [γ] subunits. The [γ][γ] enolase isoform is most specific for neurons and is referred to as neuronspecific enolase (NSE). Elevations of NSE in serum can be attributed to cerebral injury due to physical damage or ischemia caused by infarction or cerebral hemorrhage, coupled with increased permeability of the blood brain barrier. The serum (24) Adamson, J.; Beswick, A.; Ebrahim, S. J. Stroke Cerebrovasc. Dis. 2004, 13, 171–177. (25) U.S. Department Of Health And Human Services, Center for Disease Control and Prevention, National Vital Statistics Report, 2007, 55, 1-120. (26) Office of National Statistics Health Statistics Quarterly (12) Winter. Stroke incidence and risk factors in a population based cohort study; London, UK, 2001. (27) National Institute of Neurological Disorders and Stroke rt-PA Stroke Study Group., N. Engl. J. Med. 1995, 33, 1581-1587. (28) Kwiatkowski, T. G.; et al. N. Engl. J. Med. 1995, 340, 1781–1787. (29) Marler, J. R. J. Ann. Emerg. Med. 1999, 33, 450–451. (30) Marler, J. R.; et al. Neurology 2000, 55, 1649–1655.

concentration of NSE has also been reported to correlate with the extent of damage (infarct volume) and neurological outcome.31 Additionally, a secondary elevation of serum NSE concentration may be an indicator of delayed neuronal injury resulting from cerebral vasospasm.32 NSE, which has a biological half-life of 48 h and is normally detected in serum at an upper limit of 12.5 ng mL-1 (160 pM), is typically elevated after stroke and cerebral injury. Serum NSE is elevated after 4 h from onset, with concentrations reaching a maximum between 1 and 3 days after onset.33 After the serum concentration reaches its maximum (maybe >300 ng mL-1, 3.9 nM), it gradually decreases to normal concentrations over approximately one week. We have within this work developed a label-less immunosensor for NSE. The sensor utilizes screen-printed carbon electrodes and is modified first by deposition of a insulating polymer, poly(1,2diaminobenzene). The insulated electrode is then sonochemically ablated to form a micropore array. Conductive polyaniline protrusions are then grown and chemically modified with a commercial biotinylating reagent. Complexion of the immobilized biotin with neutravidin allows the further binding of biotinylated specific antibodies (or controls containing nonspecific IgG to allow subtraction of nonspecific binding) via standard avidin-biotin interactions as described earlier12,13 and as shown schematically in Figure 1b. EXPERIMENTAL SECTION Materials and Equipment. Sodium dihydrogen orthophosphate, disodium hydrogen orthophosphate, sodium chloride, and hydrochloric acid were obtained from BDH (Poole, Dorset, UK). Aniline, polyclonal human anti-IgG (AIgG), the biotinylation kit (part no. BK101), biotin 3-sulfo-N-hydroxysuccinimide, neutravidin, HSA, BSA, tris(hydroxymethyl)methylamine, magnesium sulfate, potassium ferrocyanide, and potassium ferricyanide were obtained from Sigma-Aldrich (Gillingham, Dorset, UK). NSE and monoclonal antibody against NSE (ANSE), protein S-100 [β], PSA, and CA125 (tumor marker protein), all with sodium azide preservative, were supplied by Canag Diagnostics, Ltd. (Gothenburg, Sweden). All water used was obtained from a Purelab UHQ Deioniser (Elga, High Wycombe, UK). Commercial screen-printed carbon electrodes (Figure 2) containing carbon working and counter electrodes as well as an Ag/AgCl reference electrode were obtained from Microarray Ltd. (Manchester, UK). The surface area of the working electrode was 0.2178 cm2. Aniline buffer (pH 1-2) was prepared containing 0.5 mol L-1 KCl, 0.3 mol L-1 HCl, and 0.2 mol L-1 aniline. Phosphate buffer (PBS, pH 7.4) was prepared comprising 52.8 mmol L-1 disodium hydrogen orthophosphate 12-hydrate, 13 mmol L-1 sodium dihydrogen orthophosphate 1-hydrate, and 5.1 mmol L-1 sodium chloride. Tris buffer (pH 6.8) comprising 50 mmol L-1 tris (hydroxymethyl)methylamine and 6.5 mg mL-1 magnesium sulfate was prepared; the pH was adjusted to pH 7.4 with 10 mmol L-1 sodium hydroxide. For antibody biotinylation, the procedure outlined in the BK101 kit was followed (see manufacturer’s instructions for details). (31) Martens, P.; Raabe, A.; Johnsson, P. Stroke 1998, 29, 2363–2366. (32) Laskowitz, D. T.; Grocott, H.; Hsia, A.; Copeland, K. R. J. Stroke Cerebrovasc. Dis. 1998, 7, 234–241. (33) Missler, U.; Wiesmann, M.; Friedrich, C.; Kaps, M. Stroke 1997, 28, 1956– 1960.

Figure 2. Specific binding of NSE: (a) Nyquist plot, concentrations of NSE (A) 0, (B) 1, (C) 2, (D) 3, (E) 4, (F) 5, and (G) 10 pg mL-1 and (b) calibration plot showing changes in electron-transfer resistance vs NSE concentration.

Antibodies against NSE were reconstituted in pH 7.4 phosphate buffer. Biotinylated antibodies were kept frozen in aliquots of 200 µL at a concentration of 1 mg mL-1 until required. Formation of Polyaniline Microarrays. Polyaniline microarrays were constructed as described earlier.17 To deposit the insulating layer, a 5 mmol L-1 solution of 1,2-diaminobenzene in pH 7.4 phosphate buffer was utilized. Prior to the immersion of the carbon electrode, the monomer solution used was thoroughly purged with N2 for 20 min in a sealed cell to provide an oxygenfree atmosphere. An initial 1-s blast of ultrasound was also applied to a submerged electrode to displace air bubbles trapped at the surface of the electrodes. Homogenous insulation of a planar carbon electrode was achieved by sequentially scanning the working electrode potential from 0 to +1000 mV (vs Ag/AgCl) and back to the starting potential at a scan rate of 50 mV s-1 for 20 sweeps. Sonication experiments were performed using a custom-built 2-kW, 25-kHz ultrasound tank with internal dimensions of 750 × 750 × 600 mm (working volume 750 × 750 × 500 mm) (Ultrawave Ltd., Cardiff, UK). Ultrasound was applied at a frequency of 25 kHz for 10-s duration. The potentiodynamic electrodeposition of polyaniline into the microelectrode array was achieved electrochemically by sequentially cycling the working electrode potential from -200 to +800 mV (vs Ag/AgCl) and back to the starting potential at a scan rate Analytical Chemistry, Vol. 80, No. 24, December 15, 2008

9413

of 50 mV s-1. A linear sweep from -200 to +800 mV (50 mV s-1) was performed at the end of cyclic voltammetry to leave the polyaniline in its protonated emeraldine salt form. Attachment of Antibody to the Microarray. The 30-µL sample of biotin-sulfo-NHS (10 mg mL-1 in water) was placed on the microarray electrode surface for 24 h. The sensors were rinsed with copious water and 30 µL of avidin (10 µg mL-1 in water) placed on the working electrode for 1 h sfollowed by further rinsing in water. Then 30 µL of biotinylated antibody (1 mg mL-1 in water, 1 h) was added followed by rinsing. Finally, nonspecific interactions were blocked by HSA (10-6 mol L-1 in PBS, 1 h). Alternating current impedance measurements were performed using an ACM Auto AC DSP frequency response analyzer. Following immobilization of antibody, impedance analyses were performed from 1 to 10 000 Hz (±5 mV amplitude perturbation) in pH 7.4 tris buffer, i.e., containing no antigen, as a baseline trace. This buffer solution did, however, contain a 50:50 mixture of [Fe(CN)6]3-/4-, at a concentration of 10 mmol L-1 as redox mediator so as to perform faradic impedance spectroscopy. The potential of the electrochemical cell is offset to the formal potential of the redox probe (+0.12 V vs Ag/AgCl via amperometric cyclic voltammetry). Subsequently, ANSE-doped sensors are exposed to a range of concentrations from 1 to 300 pg mL-1 NSE, suspended in pH 6.8 tris buffers containing [Fe(CN)6]3-/4-. RESULTS AND DISCUSSION Impedance Profiles of the Electrodes. Figure 2a shows the Nyquist curves obtained for specific response to NSE for a novel ANSE-immobilized, PANI-modified, carbon microelectrode array sensor assembly. The concentration of NSE analyte represented is 0-10 pg mL-1. The Z′ (real) component of the impedance increases steadily with decreasing frequency whereas the Z′′ (imaginary) component increases to a maximum value before falling to a minimum as the frequency approaches 1 Hz followed by a further increase in Z′′ rather than forming a semicircular Nyquist plot. This type of impedance spectrum is indicative of a surface-modified electrode system where the electron transfer is slow and the impedance is controlled by the interfacial electron transfer34 at high frequency. As the frequency falls further, impedance is mainly controlled by the rate of diffusion of the redox probe.34 What is interesting is that in this system at the peak of the semicircle the capacitive impedance is greater than the faradic component; we obtained similar Nyquist behavior for our previous work on microelectrode immunosensors for PSA.22 A computer fitting of the experimental data to a theoretical model, represented by a simple equivalent circuit, is performed by the software accompanying the frequency response analyzer. For a more thorough evaluation of the data obtained, the changes in the Nyquist curves may be translated into electron-transfer impedance changes to provide a clear and consistent format. We compared experimentally determined and simulated values of electron-transfer impedance and found a 100% correlation to two decimal places. The relative electron-transfer resistance changes from the baseline response at each concentration are plotted in Figure 2b. The insulation of the modified electrode upon formation of stable antibody-antigen immunocomplexes hinders the electron-transfer (34) Katz, E.; Willner, I. Electroanalysis 2003, 15, 913–947.

9414

Analytical Chemistry, Vol. 80, No. 24, December 15, 2008

Figure 3. Nonspecific binding of NSE (a) Nyquist plot, concentrations of NSE (A) 0, (B) 5, and (C), 10 pg mL-1 and (b) calibration plot showing changes in electron-transfer resistance vs NSE concentration.

kinetics of the redox probe resulting in the increase of electrontransfer resistance. The electron-transfer resistance increases with increasing antigen concentration and gives rise to a linear-type response from 1 to ∼50 pg mL-1. Between 50 and 300 pg mL-1, the curve begins to plateau, suggesting that the immunosensor approaches saturation. Attempts were made to extend the useful range to 100 pg mL-1 by fitting a curve to the data. Several models were tried but the best fit for the data was using the standard curve fitting software supplied with Microsoft Excel. Up to 100 pg mL-1, the experimental data could be fitted (with R2 ) 0.998) by the curve y ) -1.044 × 10-9[NSE]4 + 1.840 × 10-6[NSE]3 8.443 × 10-4[NSE]2 + 0.140[NSE] - 0.116. Nonspecific interactions have the potential to interfere with immunosensor performance, leading to erroneously elevated results. For this reason, an identical set of immunosensors was fabricated utilizing a nonspecific IgG antibody in place of the specific NSE antibody. Results for these electrodes were obtained in exactly the same manner as for the specific electrodes and the Nyquist plots are shown in Figure 3a. As can be seen, although there is binding of NSE to the nonspecific immunosensor, the responses are much smaller. A calibration profile for these measurements can also be plotted (Figure 3b) and shows there is a much lower response for the nonspecific antibody. A lineartype nonspecific response is observed from 1 to 100 pg mL-1 NSE. Upon comparison of Figure 3b with Figure 2b, it is apparent that ∼10% of the anti NSE-immobilized sensor response encountered is, in fact, nonspecific. The calibration curve tends toward a plateau

Figure 4. Corrected calibration plot showing changes in electrontransfer resistance vs NSE concentration: (a) 0-300 and (b) 0-10 pg mL-1 NSE.

at ∼100 pg mL-1 NSE, suggesting that an optimum electrontransfer resistance change from the baseline response has been reached. To obtain a calibration profile, in terms of “corrected” electrontransfer resistance change for an ANSE immunosensor (so accounting for any nonspecific responses), the response obtained in Figure 3b (nonspecific) was subtracted from those obtained in Figure 2b (specific) over the entire analytical concentration range. This matched sensor approach enables a true understanding of the ANSE-NSE immunosensor system. In total, 10 ANSEimmobilized sensors and 10 corresponding AIgG-immobilized sensors were interrogated over the concentration range, to allow an assessment of the reproducibility of the responses for this system. Results are presented in Figure 4. The error bars indicate the standard deviation obtained for the 10 matched sensor pairs from the mean corrected values. At these concentrations, the responses in terms of electrontransfer resistance change from the baseline trace (0 pg mL-1 NSE) showed reasonable linearity. From Figure 4a, a linear-type response was also observed from 1 to 50 pg mL-1 NSE. At concentrations above 50 pg mL-1, the response begins to tend toward a plateau as the sensor approaches saturation. Figure 4b shows responses in the range 1-10 pg mL- and shows linear behavior with R2 > 0.99. The reproducibility of the corrected response obtained from the standard deviation for 10 matched replicate sensor pairings enables discrimination to be possible at 1 pg mL-1 NSE antigen and a limit of detection for the immunosensor (three times the standard deviation of the baseline value) of 0.5 pg mL-1 NSE. Regeneration Studies. Attempts were made to see whether it was possible to regenerate the sensors by acid washing. Five ANSE-immobilized sensors and five matched nonspecific AIgGimmobilized sensors were exposed to the full interrogation procedure prior to reversibility investigations. Sensors were

Figure 5. Effect of (a) regeneration and (b) storage on the mean change in electron-transfer resistance at 50 pg mL-1 NSE.

exposed to the full range of NSE antigen concentrations for 30 min at each concentrationsand in between concentrations a washing/flushing of weakly bound or nonspecifically adsorbed matter from the sensor surface was performed before the impedance trace was recorded in redox-containing buffer. Once the trace had been recorded for both ANSE and AIgG-immobilized sensors at a particular concentration, the experimental protocol was continued up to 300 pg mL-1. A mean corrected electrontransfer change for each concentration was obtained with a standard deviation for the five pairs. After the regeneration treatment, the same experimental procedure was followed again to assess the recovery of the initial response in terms of corrected electron-transfer change over the concentration range. If the faradic response can be restored, the sensor has potential for reusability. Five regenerations were attempted and the saturation concentration of 50 pg mL-1 NSE was selected for detailed analysis. In order to split the antibody-antigen complex and to regenerate the sensors, 0.1 M HCl acidic buffer (pH 2.3) was applied for three separate 1-min time periods. In between each acidic buffer exposure, pH 7.4 PBS was used to rinse the sensor surface. Antigens are released from the sensor surface and antibodies are still bound when the NSE antibody/NSE antigen immunocomplexes are disrupted. Finally, the sensors were rinsed with 50 mL of PBS to produce NSE-free sensors, which could then be used for fresh analyses. The results in Figure 5a showing the change in electrontransfer resistance following exposure to 50 pg mL-1 NSE indicate that negligible changes in corrected electron-transfer response at the saturation concentration of NSE are encountered after the first regeneration for the five sensor pairings. Both specific and nonspecific sensors exhibited reproducible behavior. Following the second regeneration, it is apparent that the corrected electrontransfer change has been significantly lowered with a response Analytical Chemistry, Vol. 80, No. 24, December 15, 2008

9415

∼60% that of the initial response. All subsequent regeneration attempts for the five sensor pairings show a similar response of 50-60% of the original. The reversible phenomenon observed following one sensor regeneration attempt can be attributed to the restoration of the availability of the binding sites of the antibody after a short-time acidic washing. Unfortunately, repeated washing of the sensing layer lowered the activity of the antibody, which could account for the decrease of the sensitivity of the sensors following the initial regeneration. Subsequent attempts to regenerate the surface show that further exposure to the acidic conditions does not in this case remove all analyte recognition properties and that the antibody still acts as the receptor component of the immunosensor, although to a lesser degree. The results are significant in the fact that they demonstrate the potential for the development of the regeneration process. Further work should yield an optimum protocol for further, or perhaps continual, true reversibility. Greater initial response recovery over a larger number of cycles would clearly be highly beneficial. The phenomenon of reversible regeneration at immunosensor surfaces is a behavior that is rarely encountered and is a highly desirable property of these ANSE immunosensors. Stability of the Immunosensors. Stability of the ANSE-NSE immunosensor system was assessed from storage in dry state, at 4 °C, for a period of 12 weeks. An identical experimental protocol was employed for electrochemical impedance interrogation, and the corrected electron-transfer change was chosen as the analysis approach for response recovery. Initially, 18 matched sensor pairings of ANSE (specific) and AIgG (nonspecific) immunosensors were refrigerated at 4 °C in sealed, sterilized vessels. Every 2 weeks, a set of three matched sensor pairings was removed from storage and interrogated over the full NSE antigen analyte concentration range from 1 to 300 pg mL-1. Figure 5b shows the mean change in corrected electrontransfer resistance obtained at each time interval and a NSE concentration of 50 pg mL-1. As can be seen, only minimal changes occur over a storage period as long as 12 weeks, indicating that the activity of the antibody receptor biocomponent is unaffected by storage over this time. The results lend themselves well to possible packaging and suggest a satisfactory shelf life. Effect of Interferents. The experimental protocol for the assembly and testing of NSE was repeated in an identical manner except that the antigen solutions contained one of either 10 ng mL-1 BSA, HSA, CA125 (tumor marker), PSA, or S-100 [β] along with the NSE. In these cases, the calibration results did not differ significantly from the original calibration figures (Figure 4a and b). The mean corrected electron-transfer changes showed a variation of less than 5% for all concentrations and standard deviations that were comparable to the original. Thus, adsorption of BSA, HSA, CA125, PSA, and S-100 [β] and any nonspecific interactions at the sensor surface do not occur or do not affect significantly the biorecognition events (stable immunocomplex formation) and their impedance response. The matched AIgG sensor (nonspecific) response does not alter, and so it is postulated that the HSA blocking during fabrication minimizes nonspecific effects almost totally. The AIgG sensor response is thus not due

9416

Analytical Chemistry, Vol. 80, No. 24, December 15, 2008

to the presence of the target analyte or the presence of foreign proteins. This control sensor must be responding to any other effects at the sensor surface that may be occurring such as changes in the polymersor effects from the solution itself on the sensor. In any case, the effects are always minor compared to the signal obtained when a sensor immobilized with a specific receptor antibody is exposed to its target analyte. Again, these results are promising for prototype development since they suggest that sensing of a target antigen in “real” samples is a plausible proposal. CONCLUSIONS The results obtained verify the fabrication techniques and experimental interrogation protocol as a viable approach toward the label-less sensing of markers for stroke and demonstrate the generic nature of the technology. The ANSE-NSE immunosensor system has a low nonspecific response associated with the faradic sensing of NSE. The ANSE immunosensor gave a corrected lineartype response from 1 to 50 pg mL-1 with saturation occurring above these concentrations. The limit of detection of the ANSE-NSE immunosensor system was found to be 0.5 pg mL-1, and there was an ability to discriminate at a concentration of 1 pg mL-1 NSE analyte. The immunosensor was reversible for one regeneration attempt before the acid treatment has a detrimental effect on the performance of the immobilized antibody as a receptor biocomponent. The stability of the sensor assemblies was encouraging with negligible loss of sensitivity after storage for 12 weeks at 4 °C in dry state. No interference was observed upon exposure to high levels of other proteins. The low quantification levels and limits of detection for NSE have illustrated the potential use of these immunosensor assemblies for the label-less detection of markers for stroke within the range of clinical significance. Currently, ELISA testing kits permit the quantification of elevated levels of NSE in human serum following stroke from approximately 1 to 150 ng mL-1 NSE (Canag Diagnostics, Diagnostic Automation, Inc.). Although this initial study shows great promise, there still remains further work to be performed. Measurements need to be performed on clinical samples. However, it should be noted the extremely low limits of detection of the affinity-based sensors indicate that clinical samples can be highly diluted in PBS before use, thereby simplifying sample handling and also diluting and minimizing the effects of potential interferents. The minimal effect of high levels of potentially interfering proteins also bodes well for potential applications. ACKNOWLEDGMENT This work, including funding for A.C.B., has been supported by the European Community QLRT-2001-02583 (SMILE) and NMP2-CT-2003-505485 (ELISHA) Framework VI contracts. The authors thank Dr Josephine M. Higson for help with mathematical modeling and curve fitting. Received for review July 7, 2008. Accepted October 27, 2008. AC801394D