Surface Plasmon Resonance Biomolecular Interaction Analysis

Fiber optic probe-based surface plasmon resonance (SPR) detection has been used in combination with matrix-assisted laser desorption/ionization time-o...
0 downloads 0 Views 152KB Size
Anal. Chem. 1997, 69, 4369-4374

Surface Plasmon Resonance Biomolecular Interaction Analysis Mass Spectrometry. 2. Fiber Optic-Based Analysis Randall W. Nelson*,† and Jennifer R. Krone†

Department of Chemistry and Biochemistry, Arizona State University, Tempe, Arizona 85287-1604 Osten Jansson

Biacore AB, Uppsala Sweden

Fiber optic probe-based surface plasmon resonance (SPR) detection has been used in combination with matrixassisted laser desorption/ionization time-of-flight (MALDITOF) mass spectrometry for the rapid, sensitive, and selective detection of biomolecules. SPR was used to monitor the covalent immobilization of a polyclonal antibody to the surface of a fiber optic probe. The derivatized probe was then used for the selective detection (from solution) of the corresponding antigen and a secondary antibody directed toward the antigen. Species retained during the SPR detection process were next analyzed by direct MALDI-TOF analysis of the probe surface (after exposed to the MALDI matrix and introduction into the mass spectrometer). The combined approach allowed for the two-dimensional detection of biomolecules, with SPR analysis yielding quantitative information pertinent to the binding events and MALDI-TOF providing details on the qualitative nature of the binding partners. As part of continuing investigations into surface plasmon resonance-based biomolecular interaction analysis mass spectrometry (SPR-BIA/MS),1-3 fiber optic probe-based SPR detection4,5 has been coupled to matrix-assisted laser desorption/ ionization time-of-flight (MALDI-TOF) mass spectrometry.6 Fiber optic probe-based SPR analyses, while currently not capable of fully supporting the rigorous kinetic evaluation of binding partners (as is the forte of chip-based SPR-BIA7), do however, offer a number of potential advantages over chip-based SPR analyses. A first advantage is derived from the portability of the fiber optic sensors. Analyses are performed using a derivatized probe † Present address: Intrinsic Bioprobes Inc., 2009 E. 5th St., Ste. 11, Tempe, AZ 85281. (1) Krone, J. R.; Nelson, R. W.; Dogruel, D.; Williams, P.; Granzow, R. Anal. Biochem. 1997, 244, 124-132. (2) Nelson, R. W.; Krone, J. R.; Dogruel, D.; Tubbs, K. A.; Granzow, R.; Jansson, O. Techniques in Protein Chemistry VIII; Academic Press: San Diego, CA, 1997; pp 493-504. (3) Nelson, R. W.; Krone, J. R.; Dogruel, D.; Tubbs, K. A. New Methods for the Study of Molecular Aggregates; Kluwer Academic Press: Norwell, MA, in press. (4) Jorgenson, R. C.; Yee, S. S.; Johnston, K. S.; Compton, B. J. Sens. Actuators B 1993, 12, 213-220. (5) Jorgenson, R. C.; Yee, S. S. Sens. Actuators A 1994, 43, 44-48. (6) Karas, M.; Hillenkamp, F. Anal. Chem. 1988, 60, 2299-2301. (7) Karlsson, R.; Roos, H.; Fa¨gerstam, L.; Persson, B. Methods 1994, 6, 99110.

S0003-2700(97)00537-4 CCC: $14.00

© 1997 American Chemical Society

tethered to the end of a fiber optic lead, a process that essentially lends a high degree of flexibility to an otherwise physically confined method of detection. Along these lines, the physical dimensions of the biosensor unit are relatively small,8 allowing the possibility of transporting the unit into the field. A further advantage is that analyses are quite rapid, generally requiring only a few minutes to perform. Finally, the biospecific component of the biosensor is readily interchangeable, allowing the detection unit to be used serially for multiple analyses directed toward different ligands. A second level of specificity is added to fiber optic SPR screening by the mass spectrometric analysis of compounds retained on the surface of the derivatized probe. As with previously reported chip-based analyses, MALDI-TOF analysis (directly from the surface of the biosensor) adds to the interaction analysis by unambiguously confirming the select retention of the target analyte(s). Of equal importance is the ability of mass spectrometry to detect the presence of nontargeted species (e.g., signals observed at mass-to-charge (m/z) ratios other than those expected for the targeted species) or to evaluate mass shifts due to variations of the targeted species. In addition to being highly specific, MALDI-TOF analyses can be performed with speeds comparable to fiber optic SPR analyses and with equal or greater sensitivities. Combination of the two, fiber optic SPR interaction analysis with MALDI-TOF mass spectrometry, therefore affords a rapid, sensitive, and accurate assay capable of two-dimensional determination of select species. Investigations were performed using the same antibody/ antigen/antibody system reported in an accompanying article.9 All aspects of fiber optic probe derivatization and subsequent analyte detection were monitored in real-time using SPR. Fiber optic sensors, with retentate intact, were then subjected to MALDITOF mass analysis to confirm the presence of retained species. Existing concerns of SPR-BIA/MS (e.g., the ability to detect ligands over a moderate molecular weight range, and the optimization of experimental parameters used in sample preparation and data acquisition) were the topic of study, as well as issues of matrix application to fiber optic probes and the ability to obtain quality mass spectra from the surface of the probe. (8) Biacore probe product description literature. (9) Nelson, R. W.; Krone, J. R.; Jansson, O. Anal. Chem. 1997, 69, 43634368.

Analytical Chemistry, Vol. 69, No. 21, November 1, 1997 4369

Figure 1. General scheme of fiber optic-based SPR-BIA/MS. SPR is used to monitor (in real-time) the derivatization of the fiber optic probe surface (probe dimensions; 10 mm × 0.5 mm (diameter)). The probe is then used in the quantitative SPR analysis of biospecific interactions. MALDI-TOF analysis directly from the surface of the fiber optic probe is used to qualitatively evaluate ligands retained during SPR analysis.

EXPERIMENTAL SECTION Fiber optic SPR analyses were performed using a prototype Biacore probe (Biacore AB, Oppsala, Sweden). A model system of rabbit polyclonal anti-human IgG/human myoglobin/monoclonal anti-human myoglobin was investigated, using the same dervatization/analysis agents as described previously.9 Briefly, stabilizing agents were washed from CM5 (carboxylated dextran) sensor probes by dipping the probes into distilled water several times. The probes were then activated through exposure to N-hydroxysuccinimide [0.1 M prepared in 20 mM HEPES, 0.005% surfactant P20, 150 mM NaCl, 3 mM EDTA, pH 7.4 (HBS) containing 0.1 M N′-ethyl-N′-[(dimethylamino)propyl]carbodiimide], derivatized by exposure to rabbit polyclonal anti-human myoglobin IgG (0.05 mg/mL in 5 mM sodium maleate buffer, pH 6.0) and then blocked by exposure to 1 M ethanolamine hydrochloride (in HBS adjusted to pH 8.5), each step performed by dipping the probes into a vessel filled with the appropriate reagent(s). The probes were next exposed to human myoglobin (400 ng/mL, prepared in HBS with 40 mg/mL human serum albumin, HSA), followed by a short rinse (distilled water) and then exposure to monoclonal anti-human IgG (0.001 mg/mL, prepared in HBS). The probes were rinsed a final time, uncoupled from the biosensor unit, and stored at ambient conditions until preparation for mass spectrometry. Probes were prepared for mass spectrometry by insertion into, and rapid withdrawal from (contact time ∼0.1 s), a vial of sinapinic acid matrix (∼50 mM dissolved in 1:2 acetonitrile/1.5% trifluoroacetic acid). The matrix was allowed to air-dry before the probes were inserted into essentially the same mass spectrometer as previously described,9 with the exception of a source mounting 4370

Analytical Chemistry, Vol. 69, No. 21, November 1, 1997

bracket capable of holding the fiber optic probe. MALDI-TOF mass spectrometry was performed using front-side desorption/ ionization by targeting the surface of the probe with a laser while viewing the operation using a video camera. Mass spectra were obtained in the positive ion mode, and were the signal average of multiple (∼50) laser desorption/ionization events. Spectra were calibrated using an equation generated by equating the flight times and m/z values of the singly and doubly charged signals of horse heart cytochrome c (MW ) 12 360.7; the calibration sample was prepared on a discrete area of the probe). RESULTS AND DISCUSSION Compatibility/Sensitivity. Figure 1 shows an overview of fiber optic-based SPR-BIA/MS. In general, biospecific probes are constructed on an SPR-active medium tethered at the end of a fiber optic lead. Derivatization and interactions steps are monitored, in real-time, using SPR and proceed by placing the probe in vials of either derivatization agents or analyte. After use in SPR detection of analytes, the probes are uncoupled from the biosensor unit and prepared for MALDI-TOF mass spectrometry by dipping in (and quickly withdrawing from) a vial of matrix. The probes are next inserted into an appropriately configured mass spectrometer and targeted with a laser for MALDI-TOF mass spectrometry. Species retained during SPR analysis are identified on the basis of molecular weight. Figure 2 shows an SPR response plot taken during the derivatization of a fiber optic probe with polyclonal anti-human myoglobin IgG and subsequent use of the probe in the analysis for human myoglobin and secondary (monoclonal) anti-human myoglobin IgG. The derivatization procedure was comprised of three steps: an activation step, followed by the covalent coupling

Figure 2. Response plot of the derivatization of an SPR active fiber optic probe with polyclonal anti-human myoglobin IgG and use of the probe in the analysis of human myoglobin and secondary IgG. Response changes due to each stage of derivatization (activation, immobilization, blocking) and analysis (myoglobin, secondary antibody) are as marked (R ) rinse).

of IgG to the sensor surface, and then blocking of the remaining activated sites. Each step proceeded by placing the fiber optic probe in a vial containing the necessary reagents. All steps were monitored real-time using SPR, with an integration time of ∼100 s. SPR response is report in kilo resonance units (kRU), with 1 kRU equaling a surface density of 1 ng of analyte/mm2. Upon activation, immobilization, and blocking, an overall change in response of ∼25 kRU is observed. During the myoglobin analysis (placing the probe in a vial containing human myoglobin, in the presence of HSA, for ∼15 min), a change in response of ∼2 kRU was observed. The probe was then exposed to monoclonal antihuman myoglobin IgG (by dipping in a vial of antibody for ∼7 min), resulting in a further response change of ∼7.5 kRU. Considering the dimensions of the fiber optic probe (cylindrical; 10 mm long × 0.5 mm diameter; surface area ∼15 mm2), the following estimates are made for the amount of material retained on the surface of the probe during the derivatization and analysis steps. The response change observed during the immobilization of the polyclonal antibody (∼25 kRU) translates to ∼375 ng (2.5 pmol) of antibody covering the active surface of the probe. Considering an antibody valence of 2, the probe is capable of retaining a maximum of 5 pmol of antigen. During incubation with myoglobin, the SPR signal is observed to approach a plateau, indicating impending saturation. Upon removal of the probe from the myoglobin solution, a net response change of ∼1.8 kRU (∼27 ng/probe surface) is observed. The SPR response translates to ∼1.6 pmol of myoglobin retained during the incubation. It is apparent that not all of the primary antibody was appropriately oriented for binding, as the 1.6 pmole retained myoglobin represents only ∼32% (1.6 pmol/5 pmol) of the maximum possible antigen valence sites. Similar hindrance of antigen binding sites has been observed during previous studies,1-3,9 and is presumably due to random orientation of the antibody during the amine

coupling procedure. The SPR response change due to the secondary antibody was ∼7.5 kRU, indicating the retention of ∼750 fmol of the antibody. Again, the less than unity (∼47%) retention of solution-phase component indicates hindered access to the epitope (due to shielding of the epitope by the primary antibody). The amount of sample present on the fiber optic probe (750 fmol-1.6 pmol) is, under the usual conditions, ample for MALDITOF analysis. Indeed, MALDI-TOF analyses are quite sensitive and have been performed at the subfemtomole level for analytes in the 10-100 kDa mass range10 and with as little as ∼70 fmol of analytes with molecular weights in the 1 MDa mass range.11 Even during the previous studies on chip-based SPR-BIA/MS, analyte quantities in the low-femtomole range were sufficient for mass spectrometric analysis of retained proteins.1-3,9 Analysis of the SPR fiber optic probes, however, was not performed in a fashion typical of MALDI-TOF mass spectrometry (or chip-based SPRBIA/MS) in which all the sample present on the two-dimensional surface was available for analysis. Considering that the probe was cylindrical, at least half of the sample retained during SPR was not available for mass spectrometric analysis (either shadowed from the laser or on the side of the probe opposite the entrance of the mass spectrometer). Furthermore, the fiber optic probes presented a 10 mm × 0.5 mm cross section to the entrance of the mass spectrometer. This area (5 mm2) is somewhat larger than the sample spots used in most MALDI-TOF instruments (most commercial instruments present a target spot of e1 mm2). As a result, most of the sample was thinly dispersed on the fiber optic probe and could not be addressed for mass analysis (using (10) Li. L.; Golding, R. E.; Whittal, R. M. J. Am. Chem. Soc. 1996, 118, 1166211663. (11) Nelson, R. W.; Dogruel, D.; Williams, P. Rapid Commun. Mass Spectrom. 1994, 8, 627-631.

Analytical Chemistry, Vol. 69, No. 21, November 1, 1997

4371

Figure 3. MALDI mass spectrum obtained from a fiber optic SPR probe derivatized with polyclonal anti-human myoglobin and exposed first to human myoglobin and then to monoclonal anti-human myoglobin IgG. Ion signals are readily observed for myoglobin (m/z ) 17 200 Da) and the secondary antibody [m/z ∼144 500 Da (see inset)].

the existing mass spectrometer configuration). Regardless, mass spectra of reasonably high quality were obtained from the surface of the fiber optic probes. Figure 3 shows a MALDI-TOF mass spectrum obtained from targeting the surface of the fiber optic probe used in generating Figure 2. Observed in the mass spectrum are ion signals indicating the selective retention of myolobin and anti-myoglobin IgG. Overall, the mass spectrum is relatively clean, indicating little or no retention of unspecified ligands. Specificity/Accuracy. SPR detection is, on a whole, nonspecific, measuring the mass amount of material retained on the surface of the fiber optic probe largely independent of chemical or physical properties. The qualitative accuracy of the SPR analysis therefore relies on the specificity of the immobilized binding partner. Although this specificity can approach absolute (the retention of only a single analyte), there exists the chance to bind nontargeted compounds to the surface of the probe in an unspecified manner. Alternatively, different versions (variants) of the target species stand a chance of being retained by the immobilized binding partner. These situations can result in the false detection of the targeted analyte, which, in turn, is of some consequence during both qualitative and quantitative evaluations of SPR data. However, SPR detection is universal and, more importantly, nondestructive, allowing secondary analysis of the retained species. MALDI-TOF mass spectrometry, although being destructive, is extremely specific, detecting species at defined molecular masses. Thus, MALDI-TOF analysis is capable of discerning the nature of species (targeted analytes, variants, or nonspecific) retained during SPR analysis. As shown in Figure 3, such mass spectrometric analysis of retained species is of great value in validating the presence of the targeted compounds. The validation of retained (targeted) species is critical in establishing the degree of analytical specificity of the immobilized ligands. For instance, the polyclonal antibodies immobilized to the surface of the fiber optic responsible for Figures 2 and 3 appear to be of sufficient specificity to isolate the myoglobin with little cross-reactivity toward any components present in the carrier solution (40 mg/mL HSA). However, fiber optic probe SPR mass spectrometry can be used as easily to detect deficiencies in compounds or processes. Figure 4 shows a mass spectrum 4372 Analytical Chemistry, Vol. 69, No. 21, November 1, 1997

Figure 4. MALDI-TOF mass spectrum taken from a second SPR probe showing the presence of nontargeted compounds. The nontargeted compounds stem from polypeptides present myoglobin carrier (HSA breakdown products) and are most likely due to inadequate rinsing of the probe after exposure to myoglobin (see Figure 5).

obtained from the surface of a fiber optic probe used in a screening similar to that demonstrated above, with the exception of inadequate washing between incubation with myoglobin and the secondary antibody. Obvious in the mass spectrum are signals due to the presence of species other than those targeted. The origin of the contamination was later determined to be the carrier solution (data not shown); the low molecular weight fragments presumably due to some form of breakdown product of the albumin. At present, it is difficult to make rigorous quantitative estimates using MALDI-TOF without the use of standards;12,13 however, upon review of the SPR response plot (Figure 5), a response of ∼2.3 kRU (30% higher than above) was observed during the myoglobin incubation, while treatment with the secondary antibody yielded a response change of ∼7.5 kRU (the same as in Figure 2). The additional signal observed during myoglobin incubation can most likely be attributed to the nonspecific species. Information such as this is of great benefit not only when protocols necessary for reducing the retention of nontargeted species are being developed but when the relative binding efficiencies (quantity bound vs immobilization chemistry) of the immobilized and secondary ligands are determined. An additional question regarding specificity arises from the degree of mass accuracy possible during MALDI-TOF analysis from the surface of a fiber optic probe. As stated, spectra were calibrated using an equation generated from a standard applied to a discrete region of the fiber optic probe. Figure 6 shows a mass spectrum taken from the region where the calibration sample adjoined the analytical sample. Ion signals for both the calibrant (cytochrome c) and analyte species are observed, thereby allowing the spectrum to be internally calibrated. Such procedure should make possible a mass accuracy of ∼0.01%;14 however, the myoglobin signal clearly lacks the resolution necessary to establish an accurate centroid (required for highest mass accuracy)sthe broadness of the peak is due to heterogeneity, (presumably) a product of the analyte stabilization procedure (the myoglobin is (12) Nelson, R. W.; McLean, M. A.; Hutchens, T. W. Anal. Chem. 1994, 66, 1408-1415. (13) Muddiman, D. C.; Gusev, A. I.; Hercules, D. M. Mass Spectrom. Rev. 1996, 14, 383-429. (14) Beavis, R. C.; Chait, B. T. Anal. Chem. 1990, 62, 1836-1840.

Figure 5. Response plot of from sensor probe used in the generation of Figure 4. Responses from all steps of the derivatization and analysis processes are virtually the same as in Figure 2, with the exception of an ∼30% increase in response during exposure to myoglobin. The use of these data in combination with the mass spectral data suggests the retention of nontargeted compounds during incubation with myoglobin (present in the HSA carrier) (R ) rinse).

Figure 6. MALDI-TOF mass spectrum taken from an SPR-active fiber optic probe showing analyte (human myoglobinsretained during SPR analysis) and internal calibrant (equine cytochrome csapplied to probe in the presence of MALDI matrix; MW ) 12 360.7). Although the heterogeneity of the analyte prohibits the accurate determination of a (single) molecular weight for the human myoglobin, the reasonable mass resolution observed for the calibrant indicates only a minor reduction of instrumental performance (during acquisition of mass spectra from the surface of the fiber optic).

supplied cyano-stabilized, and has been observed as heterogeneous during previous investigations2). Therefore no real claims can be made as to the experimentally determined degree of mass accuracy, although judging from the adequate resolution of the calibrant signal (m/dm ∼350 vs ∼500 typical), instrumental performance appears to not be severely degraded, and a reasonable mass accuracy (e.g., on the order of 0.01-0.05%) can be expected. A few issues regarding the fiber optic SPR-BIA/MS approach are worth comment. The first of these issues regards sample

utilization. During the analyses, signals adequate for determining the presence, and origin, of retained species were observed. While the detection limits of neither SPR-BIA nor MALDI-TOF were apparently reached, it is worth noting that the vast majority of sample retained on the fiber optic probe was never subject to mass spectral analysis (for the reasons stated above). In fact, by SPR-BIA estimate, enough analyte (∼1.6 pmol of myoglobin) was retained on the fiber optic probe for perhaps as many as 20, traditionally prepared MALDI-TOF analyses. It is for this single reason that introduction of the fiber optic probe into the mass spectrometer may not be the most universally applied method for mass spectrometric analyses of species retained on the surface of the probe. Given the opportunity, elution of retained species from the fiber optic probe onto a MALDI-TOF mass spectrometer target (or into an electrospray ionization source) appears a much better alternative, especially when considering the re-use of the fiber optic sensors. (Preliminary investigations into such an approach have been undertaken, and have indeed shown such benefits of eluting from the probe onto a MALDI-TOF mass spectrometer target.) Another issue requiring address is that of the purpose of the analyses. As demonstrated above, the fiber optic SPR-BIA/MS approach is well suited for the rapid, two-dimensional qualitative screening of fluids for the presence of specific compounds. Additional benefits of the analysis are the ability to determine the presence of untargeted species (either nonspecifically retained or variants of the target analyte) and use in evaluating immobilization/binding/rinse protocols for optimal interaction conditions. A major use of the fiber optic SPR-BIA/MS method might therefore lie in the evaluation of biological systems (e.g., fermentation broths or expression media) for the high-efficiency production (as estimated with quantitative SPR-BIA) of the correct species Analytical Chemistry, Vol. 69, No. 21, November 1, 1997

4373

(as verified by MALDI-TOF). Like analyses are possible for chipbased SPR-BIA/MS; however, the true forte of the chip-based sensors is the accurate determination of the affinity constants involved in biomolecular interactions. Mass spectrometry aids in the chip-based analyses by confirming that only the targeted species has been involved in the SPR-monitored interaction and, therefore, that the affinity constants are in fact for the described complex. Alternatively, MALDI-TOF analyses are quick to detect nontargeted species retained during chip-based SPR-BIA, and as such, data generated during the BIA analysis can either be rejected or possibly corrected for by using quantitative MALDITOF methods. CONCLUSION The use of fiber optic-based SPR detection with MALDI-TOF mass spectrometry for the two-dimensional detection of biospecifically retained analytes has been reported. The two methods

4374

Analytical Chemistry, Vol. 69, No. 21, November 1, 1997

of detection have demonstrated considerable complementarity, in terms of quantitative determination (SPR) vs absolute specificity (MALDI-TOF), and high compatibility in terms of speed and sensitivity. The combined approach promises to be a means of accurately assessing the specificity of ligands involved in biomolecular recognition events, as well as a rapid method for system optimization during the development of biospecific assays. ACKNOWLEDGMENT The financial and technical support of Biacore AB has been greatly appreciated. Data presented in part at the 6th Biasymposium; October 8-10, 1996, Washington, D.C. Received for review May 27, 1997. Accepted August 25, 1997.X AC9705374 X

Abstract published in Advance ACS Abstracts, October 1, 1997.