Anal. Chem. 1997, 69, 4242-4248
Monitoring the Reactions of Single Enzyme Molecules and Single Metal Ions Weihong Tan† and Edward S. Yeung*
Ames LaboratorysUSDOE and Department of Chemistry, Iowa State University, Ames, Iowa 50011
One approach for studying single molecules is to cycle the molecule through a reaction repeatedly, i.e., to monitor the products from a reaction catalyzed by the molecule of interest. By using a laser-based optical microscope and a CCD detection system, we have determined the chemical activities of individual enzyme molecules. Single molecules are trapped inside femtoliter-size vials (pores in membranes or nanoscopic silica vials manufactured by photolithography) ranging from 5 to 12 µm in diameter and from 4 to 6 µm deep, which act as individual nanoscopic reactors. Single lactate dehydrogenase (LDH1) molecules are isolated by filling these reactors with very low concentrations of LDH-1, excess lactate, and nicotinamide adenine dinucleotide (NAD+). The fluorescent product NADH in more than 100 individual reactors can be followed simultaneously in real time. Different reaction characteristics are observed, both in terms of the instantaneous reaction rates and the average rates over a 30min interval. We found that even when the microenvironments are identical, individual enzyme molecules reveal heterogeneities in their activities. This indicates that distinct molecular conformations may exist in otherwise identical molecules. We have further used this technique for the study of single metal ion-catalyzed reactions. Uniform catalytic activities are observed for individual Os(VIII) ions. Studying single molecules can reveal heterogeneities among them. A variety of methodologies have been developed for singlemolecule detection (SMD) and characterization.1-10 In all these techniques, the amplification of the signal has been an important
issue. The signals used in SMD are either optical or electrochemical. Recent advances in optical detection, such as the avalanche photodiode and the intensified charge-coupled device (CCD), have enabled the efficient detection of single photons. Thus, SMD based on optical detection has become fairly straightforward. Optical signal amplification is mainly performed through two different approaches. The first one is the direct detection of the photons produced by one molecule. It is, however, necessary to overcome photon-counting statistics. Even though single photons can be counted by a photon detector, many photons are still needed to detect single molecules since one can only detect the photons with ∼0.5-5% efficiency.3 The simplest solution is to monitor molecular fluorescence, since the transition can be cycled through 105-106 times3,11 before photochemical destruction occurs. The second approach is the indirect detection of the molecule of interest. In this scheme, the molecule is cycled through a reaction; i.e., the products from a reaction catalyzed by the molecule of interest are monitored. A catalytic species is not consumed during reaction, which then provides amplification of the signal with prolonged reaction time. Substantial chemical amplification of the signal can be obtained. If the reaction product is fluorescent, one can combine both repeated excitation and repeated reaction to gain extreme sensitivity. This has been achieved in the detection of 10-17 molar concentration of enzyme molecules.5 The repeated reaction approach is interesting because it enables us not only to detect single molecules but also to study the activities of single molecules. It is especially significant for biomolecules, such as enzymes, since different conformations of the same molecule may display different biological and biochemical properties.12 It is well-known that enzyme molecules are highly specific and can have very high turnover numbers. It is almost always possible to design an appropriate nonfluorescent substrate for a given enzyme or catalyst, so that a fluorescent product will be formed through a reaction catalyzed by the molecule of interest. One can therefore generate millions of fluorophors from each catalytic molecule, turning detection into a trivial problem. In fact, the very first report of single-molecule detection relied on catalytic amplification to generate easily detectable signals even from a nonlaser excitation source.13 It did, however, take many hours to accumulate enough product molecules for detection. Recent experiments involving individual enzyme molecules were done inside a narrow capillary tube.5 A 7.6 × 10-17 M lactate dehydrogenase (LDH-1) enzyme/substrate solution was filled into a capillary such that only a few molecules existed in the capillary.
† Present address: Department of Chemistry, University of Florida, Gainesville, FL 32611-7200. (1) Keller, R. A.; Ambrose, W. P.; Goodwin, P. M.; Jett, J. H.; Martin, J. C.; Wu, M. Appl. Spectrosc. 1996, 50, A12-A32. (2) Dickson, R. M.; Norris, D. J.; Tzeng, Y.; Moerner, W. E. Science (Washington, D.C.) 1996, 274, 966-969. (3) Barnes, M. D.; Whitten, W. B.; Ramsey, J. M. Anal. Chem. 1995, 67, 418A423A. (4) Funatsu, T.; Harada, Y.; Tokunage, M.; Saito, K.; Yanagida, T. Nature 1995, 374, 555-559. (5) Xue, Q.; Yeung, E. Nature 1995, 373, 681-683. (6) Nie, S.; Chiu, D. T.; Zare, R. N. Science (Washington, D.C.) 1994, 266, 1018-1021. Nie, S.; Emory, S. Science (Washington, D.C.) 1997, 275, 11021106. (7) Betzig, E.; Chichester, R. J. Science (Washington, D.C.) 1993, 262, 14221425. (8) Lee, Y.-H.; Maus, R. G.; Smith, B. W.; Winefordner, J. D. Anal. Chem. 1994, 66, 4142-4149. (9) Fan, F.-R.; Bard, A. J. Science (Washington, D.C.) 1995, 267, 871-874. Fan, F.-R.; Kwak, J.; Bard, A. J. J. Am. Chem. Soc. 1996, 118, 9669-9675. (10) Collinson, M.; Wightman, R. M. Science (Washington, D.C.) 1995, 268, 1883-1885.
(11) Trautman, J. K.; Macklin, J. J.; Brus, L. E.; Betzig, E. Nature 1994, 369, 40-42. (12) Teipel, J.; Koshland, D. E. Biochemistry 1971, 10, 792-805. (13) Rotman, B. Proc. Natl. Acad. Sci. U.S.A. 1961, 47, 1081-1091.
4242 Analytical Chemistry, Vol. 69, No. 20, October 15, 1997
S0003-2700(97)00631-8 CCC: $14.00
© 1997 American Chemical Society
By incubation under appropriate reaction conditions, product zones of NADH were formed inside the capillary where each LDH-1 molecule resided.5 Individual alkaline phosphatase molecules have also been characterized in this manner.14 These experiments clearly demonstrated the feasibility of studying heterogeneities of enzyme molecules. However, only a limited number of enzyme molecules were studied in one capillary. In addition, no detailed reaction kinetics were obtained. Information regarding possible instantaneous changes in enzyme conformation during the reaction was not available. Recently, we have developed a UV laser-based optical microscope and CCD detection system for biochemical imaging. It has been successfully applied in in vivo imaging of neurotransmitter depletion and uptake in glial cells.15,16 This CCD/microscope system can easily be adapted for following biochemical reactions catalyzed by single molecules. In this report, the molecules are confined inside nanoscopic vials together with their substrates. Reaction products are independently monitored by the CCD camera. The activities corresponding to many individual molecules can be followed over time. This enables us to study detailed reaction kinetics and to detect catalytic species at trace concentrations. We have selected LDH-1 molecules and Os(VIII) metal ions for single-molecule reaction studies. LDH is an isoenzyme with five different forms in humans. The distribution in cells is related to metastatic cancers and to cell age.17 The quantification of intracellular LDH activity in individual cells has been done, and unique information about their biological functions has been obtained.17 Similar to enzyme molecules, many metal ions are able to catalyze reactions in which fluorescent products are formed.18,19 We can explore the feasibility of detecting single metal ions. We can also compare single-enzyme-catalyzed reactions with those catalyzed by single metal ions. EXPERIMENTAL SECTION Materials and Reactions. All chemicals were purchased from Sigma Chemical Co. (St. Louis, MO) and used without further purification. All buffers were prepared locally. The two catalytic reactions studied are LDH-1
lactate + NAD+ 98 NADH + pyruvate Os(VIII)
2Ce(IV) + As(III) 98 2Ce(III) + As(V)
(1) (2)
In each case, a fluorescent product is produced, these being NADH and Ce(III) respectively. LDH-1 has the highest activity among the isoenzymes and was purchased as the electrophoretically pure form.5,17 Experimental Setup. The laser-based CCD/microscope system is similar to the one used for intracellular studies of neuron cells.15 A scientific grade CCD camera system (Photometrics, (14) Craig, D. B.; Arriaga, E. A.; Wong, J. C. Y.; Lu, H.; Dovichi, N. J. J. Am. Chem. Soc. 1996, 118, 5245-5253. (15) Tan, W.; Parpura, V.; Haydon, P.; Yeung, E. S. Anal. Chem. 1995, 67, 25752579. (16) Tan, W.; Haydon, P.; Yeung, E. S. Appl. Spectrosc. 1997, 51, 1139-1143. (17) (a) Xue, Q.; Yeung, E. S. Anal. Chem. 1994, 66, 1175-1178. (b) Xue, Q.; Yeung, E. S. J. Chromatogr. B 1996, 677, 233-240. (18) Habig, R. L.; Pardue, H. L.; Worthington, J. B. Anal. Chem. 1967, 39, 600605. (19) Surasiti, C.; Sandell, E. B. Anal. Chim. Acta 1960, 22, 261-269.
Figure 1. SEM micrographs of nanoscopic vials: (a, top) membrane vials; (b, bottom) vials nanofabricated on a quartz slide by photolithography.
Tucson, AZ) was connected to a fluorescence microscope (Axioskop, Carl Zeiss). The CCD camera was mounted on the top entry port of the microscope, facing downward. About 5 mW of the 305-nm laser beam from an Ar+ laser (Model 2045, SpectraPhysics, Mountain View, CA) was isolated from the other lines with an external prism. It was focused to a 1-2-mm spot and directed by mirrors and lenses to the microscope sample stage for excitation of the fluorescent product molecules inside the nanovials. Fluorescence was collected through quartz microscope objectives (Zeiss Ultrafluor, 10× and 40× magnifications, 0.25 and 0.6 numerical aperture). These objectives transmit in the UV and have very limited autofluorescence. In front of the CCD camera, a combination of band-pass filters was used to selectively collect the desired signal. The exposure time and data acquisition time were all controlled by the commercial CCD software. During data acquisition, all images were stored in cache memory, which were later transferred to hard disk for further data analysis. Nanoscale Sampling. We used nanoscopic vials to form independent reactors of femtoliter volumes. Two different kinds of vials were used. The first ones were made from Nuclepore polycarbonate membranes (Costar Scientific), which are used in filtration. The sizes of the membrane pores are available over a broad range, from 0.015 µm to over 20 µm in diameter. We used membranes with pore diameters from 5 to 10 µm, as shown in Figure 1a. These pore sizes are uniform, and the material is confirmed in bulk studies to be chemically inactive in the enzymatic reaction in this work. The thickness is ∼6 µm, producing volumes between 120 and 470 fL for each vial. The Analytical Chemistry, Vol. 69, No. 20, October 15, 1997
4243
second kind of nanoscopic reactors were manufactured on fusedsilica plates by photolithography. They are shown in Figure 1b. Standard etching techniques were used to produce semispherical holes of ∼8 µm in diameter. The depth of these vials is ∼4 µm. The volume for the fused-silica vials is ∼135 fL each. Single enzyme molecules can be studied in either type of vials, while single metal ions have only been successfully studied in the fusedsilica vials. Liquid filling of nanoscopic volumes has been a difficult technological problem in many fields, including biological studies where micromanipulators coupled with microinjection have been employed. Since we are studying more than 100 vials simultaneously, it is impossible to fill the vials one by one. We have developed a variety of methodologies to fill the vials with the selected reagents. One of the most successful techniques was to combine ultrasonic vibration with vacuum degassing. The fusedsilica plate containing the nanovials was first immersed in the substrate solution for ∼20 min, during which an aspirator was used to apply a vacuum to get rid of the air in solution and inside the vials. At the same time, ultrasonic vibration (produced by a laboratory ultrasonic cleaner) was used to agitate the vials and the solution, which facilitated the expulsion of air from the vials. Then, the enzyme solution of predetermined concentration was added to the substrate solution. Quick manual mixing was done by inverting the closed vial, followed by a 2-min liquid filling using the same procedures described above. We used rhodamine 6G dye solution to test this procedure. From the fluorescence image of the nanovials, we conclude that this procedure ensured that all holes were filled with the desired solution. The fused-silica slide was then taken out of the solution, and a small drop of the enzyme/substrate solution was added onto the top of the vials. A very thin quartz coverslip (0.08 mm) was put on the top to cover the liquid-filled vials. A tight seal is expected to form between the fused-silica plate and the quartz coverslip. The quartz coverslip prevented any evaporation from these small vials and mixing among the vials during the monitoring of the enzyme reactions. Similar procedures were used to fill the vials in the Nuclepore polycarbonate membranes, which were sandwiched between two quartz plates to create the vials. Experimental Procedure and Data Analysis. The singlemolecule experiments will not be successful without careful design of the experiment. One of the key elements is to reduce the background signal and to avoid contamination to the enzyme/ substrate or metal ion/substrate solutions. The background level must be low compared to the signal level to allow discrimination. Careful rejection of stray light with optical filters is critical in background suppression. To avoid contamination, all sample solutions were carefully filtered (0.22 µm). Here we use the LDH-1-catalyzed enzymatic reaction as an example. Before the samples were put on the microscope stage, the laser beam was directed toward the center of the microscope field of view and focused onto a similar fused-silica plate. The standard illumination system of the microscope was used to align those nanovials with the objective. After filling with sample solution, the sample vials were put on the microscope stage and were focused in the standard transmission mode. This was followed by acquiring a fluorescence image with laser excitation. Once good focus was confirmed, a group of consecutive images was taken as the reaction proceeded. We used a fixed time interval between two images, ranging from 5 s (enzyme experi4244
Analytical Chemistry, Vol. 69, No. 20, October 15, 1997
ments) to a few minutes (metal ion experiments). To avoid heating the solution or bleaching the fluorescent products, the laser beam was blocked except during data collection. More than 100 consecutive images could be taken, depending on the image size and the computer RAM memory. For a given image, there were spatial variations due to nonuniformities in illumination by the laser beam and uneven sensitivities of the CCD pixel elements. In addition, because of dark count, stray light, and fluorescence from the buffer solutions and the coverslips, even the nonvial areas produced a background signal in the CCD images. This background level must be subtracted from the intensities recorded in each pixel to determine the fluorescence signal originating from the molecules inside the vials; i.e., flat-fielding was applied. To account for temporal variations in laser intensity and in light collection for each sample studied, the background-subtracted intensities of each vial should be further normalized by the adjacent background intensity (nonvial area) in the corresponding image. As shown in Figure 2A, there are three intensities associated with each vial at position (i, j), I0 (dark counts with no laser excitation), Iat (fluorescence intensity of vial at time t with laser excitation), and Ibt (background intensity of the adjacent nonvial area for the vial at time t with laser excitation). The relative intensity is defined as
R ) (Iat - I0)/(Ibt - I0)
(3)
For consecutive images, the same adjacent nonvial area was used for normalization. Thus, variations in laser power and signal collection efficiency will cancel out in the data analysis. For the membrane vial experiments, we have further normalized the relative intensities R with respect to that in the first image, which was recorded as soon as possible (100 s) after mixing the sample solutions. For the fused-silica vials, the low background in the nonvial areas prevents us from using it for proper normalization. RESULTS Bulk Solution Experiments. Before using the nanoscopic vials for single-molecule reaction studies, we performed bulk solution experiments to determine the catalytic efficiencies of LDH-1 molecules and of Os(VIII) ions under optimal experimental conditions. Experiments with LDH-1 involve monitoring the formation of the fluorescent product, NADH, during continuous reaction in a spectrometer (SPEX Fluoromax). The cell was filled with 3 mL of the substrate solution containing 1 mM NAD+, 3 mM lactate, and 20 mM Tris buffer at pH 8.5. At this pH, LDH-1 shows the highest activity for catalyzing lactate to pyruvate conversion. NADH also shows higher fluorescence efficiency at pH 8.5 compared to that at pH 7. NADH was monitored at its fluorescence maximum of 465 nm. As shown in Figure 3a, the lowest concentration we were able to detect with confidence was ∼5 × 10-13 M LDH-1. The small bumps in the plots for the two lowest concentrations were due to mechanical disturbance to the spectrometer. It is clear that there was no autoreaction when only the substrate was present. These curves showed extremely good linear relationships over time. When still higher concentrations of LDH-1 were used, the linear relationship only held on for a few minutes before the slope decreases. In the Os(VIII) experiments, 2 M sulfuric acid was used as the buffer for the reaction. Solutions of As(III), Ce(IV), and Os(VIII) were prepared either from their corresponding salts or
Figure 3. (a) NADH production rate as a function of [LDH-1] (molecular ratio of NAD+/LDH-1 fixed at 107); (b) Ce(III) production rate as a function of [Os(VIII)].
Figure 2. Schematic of single enzyme molecule experiments in nanoscopic vials. The amount of products produced by each enzyme molecule is different.
from their oxides by using 2 M sulfuric acid.18 The Os(VIII)catalyzed reaction exhibited a slower rate compared with that of LDH-1. Also, there was a detectable autoreaction (blank reaction) between As(III) and Ce(IV). The blank reaction can be a result of a very slow uncatalyzed reaction between As(III) and Ce(IV) or due to the catalysis by unknown impurities.18,19 The blank reaction sets a limit to the lowest concentration of Os(VIII) that can be determined. The lowest detectable concentration was ∼5 × 10-12 M, as shown in Figure 3b. The linearity over time for the Os(VIII)-catalyzed reaction was not as good as that for LDH1. Actual reasons for this are unknown, but it is conceivable that the reaction becomes autocatalytic as the product ions accumulate. Single-Molecule Reaction in Membrane Vials. Membrane vials with 8-µm diameter have a volume of ∼300 fL. If each contains only one LDH-1 molecule, the LDH-1 concentration will be 5.5 × 10-12 M. Therefore, if 2.75 × 10-12 M LDH-1 is employed, statistically, only 40% of the vials will contain one or more enzyme molecules. Using this rationale, we have tested different concen-
trations of LDH-1 and related the concentration to the number of vials where the fluorescence intensity increased during incubation. We found that there was general agreement, with (50% accuracy. Subsequent experiments were performed by controlling the concentration of the catalyst so that between 50 and 25% of the vials contained a single enzyme molecule. There is the possibility that more than one enzyme molecule resides in a vial. However, according to Poisson statistics, such are less probable events. We need to consider the minimum exposure time for adequate S/N and the frame rate (interval between exposures) that will produce a detectable change in the NADH concentration. At 1-s exposure with this CCD system, we were able to detect a NADH standard solution in the range of 2 × 10-7 M without difficulty. If one can confine one enzyme molecule in a nanoscopic reactor of 300 fL, the NADH formation rate is ∼5.5 × 10-9 M/s, which was calculated on the basis of the specifications of the LDH-1 sample purchased. This means that the CCD system can detect an obvious increase (S/N > 3) in NADH after 40 s of reaction. On the other hand, even after 2000 s of reaction, the NADH concentration is only ∼2 × 10-5 M. Since the reagent concentrations are ∼1 mM, no depletion of NAD+ or lactate will occur over the course of the experiment. In separate experiments where standard solutions of NADH were imaged, we determined that photobleaching was not a factor over the cumulative exposure time at these excitation powers. Single-molecule reaction was monitored by imaging the membrane pores. For every 10 s, a CCD image was taken to record the formation of the product, NADH. One image would Analytical Chemistry, Vol. 69, No. 20, October 15, 1997
4245
Figure 4. Single enzyme molecule reaction kinetics inside polycarbonate membrane vials. Individual LDH-1 molecules are trapped inside membrane holes with the substrate solution. The product NADH is monitored over time. The background-corrected intensity in each vial has been normalized with respect to that in the first fluorescence image.
usually cover between 50 and 80 vials in most of the experiments. From each of the vials, a reaction rate curve can be constructed. When the membrane vials were filled with only the substrate solution, we did not observe any significant fluorescence intensity change during the same time period of incubation. Similarly, when the vials were rinsed and then filled with substrate solution immediately after an enzyme assay, incubation for the same period did not produce any recognizable NADH signal. This shows that memory effects (adsorption) and other artifacts are negligible. Some of the vials do not show significant increases in fluorescence intensity, consistent with the absence of any enzyme molecules in those vials. One such rate curve is shown at the bottom of Figure 4. Even this rate curve is not completely smooth or horizontal, reflecting the noise level and the drift of the detection system. Data from 12 representative (selected on the basis of the number of total occurrences of each type of feature) vials which exhibit increases in fluorescence are also plotted in Figure 4. The signal levels at 1500 s correspond to the cumulated amounts of products formed after the experiment. These show the same large variations in activities for individual molecules as in experiments done in capillary tubes.5 The majority of the rate curves are similar to the middle group of seven (solid lines). The fluorescence intensities changed fairly linearly between 300 and 1500 s, although the slopes of each are different. The different activities can be attributed to heterogeneities in conformation among the LDH-1 enzyme molecules.5 The average rate of change in intensity roughly parallels the rate curves in Figure 3a for this concentration, indicating that Kcat is still applicable to an ensemble of single molecules. KM, however, cannot be determined here because by design there is an excess of substrates. Quite interesting is the observation of a different initial slope for each of the solid curves (between 100 and 300 s). Since the sample vials were filled off-line, there can be an initial equilibration period after the fused-silica slide was put on the microscope stage, particularly the solution temperature. Closer examination of the curves reveals that this slope change occurs at different times (varying by tens of seconds) for each vial. Any temperature effects should affect all vials equally since they are very closely spaced. These sudden and random rate changes are even more obvious for some other vials (five dashed curves in Figure 4) and can even 4246
Analytical Chemistry, Vol. 69, No. 20, October 15, 1997
occur several times (hundreds of seconds apart) for each vial during the 1500-s observation period. Since all plots in Figure 4 are derived from the same experiment, and since many data points define each rate change, it is unlikely that such sudden jumps are experimental artifacts due to, for example, laser power fluctuations, temperature changes, contamination, or mechanical instabilities. This emphasizes the advantage of monitoring 100 vials at a time. There are two possible explanations for such rate changes. The molecules can be going through conformational changes in this time scale, thereby altering the enzyme’s catalytic efficiency. Or, there can be some influence from the microenvironment in which the molecule resides. In particular, adsorptive or electrostatic interactions with the surface (fused silica or polycarbonate) can affect the activity of the enzyme, although permanent adsorption on the surface can be ruled out because we did not observe any memory effects in control experiments. Figure 4 shows the unique features of our experimental protocol. While the activities of individual enzyme molecules have been measured after successive incubation periods,5,14 continuous monitoring of product formation is required to produce detailed instantaneous rate information. Such information was not available when only the total amount of products produced was measured at the end of each incubation period.5,14 The ability to follow many vials simultaneously allows molecular heterogeneities to be distinguished from differences in conditions from one experiment to the next. Single-Molecule Reaction in Silica Vials. The same solution and procedures similar to those used in the membrane vials were employed in the fused-silica vial experiments. There are more than 400 vials on each fused-silica plate. Depending on the magnification used in imaging (10× or 40× objectives), ∼100 vials were monitored through the CCD camera, providing 100 sets of reaction rates at a time. In one experiment, in ∼63% of the vials, the relative increase in signal is between 0.925 and 1.1; while the other 37% of the vials exhibit relative increases between 1.125 and 1.35. By definition, when the relative increase in signal is 1, there is no product formed. A value less than 1 can result from the normalization process if stray light in the nonvial areas fluctuates during the experiment. This actually allows us to estimate the maximum uncertainty in defining an “empty” vial. So, 1.1 was used as the cutoff value. The number of vials without enzyme molecules thus roughly corresponds to that expected based on the bulk concentration. We have plotted in Figure 5 signals from the 11 vials that contain presumably one enzyme molecule each. Again the observed reaction rates (slopes) were different. There is ∼3× difference (slopes ranging from 0.64 to 0.23 with a mean of 0.40 and a relative standard deviation of 25%) in the activities of the individual enzyme molecules. This is a significant difference in activities and indicates that there may be structural variations among these molecules.12 Single Metal Ion Reaction. One of the difficulties associated with single metal ion detection is contamination control. There are many different sources of contamination, such as the container for solutions, the reactor, and the solvents. We first attempted single metal ion experiments by using the membrane vials. Contamination (background reaction) was found to be a major problem. We then used the nanoscopic fused-silica vials. If proper precautions were taken, contamination could be controlled. A portion of the bulk solution, similar to the one described above,
is 13.3-8.6, with a mean of 10.9 and a relative standard deviation of 15%. Since metal ions have invariant conformation, we expect each one to have the same catalytic activity if they are put in the same environment where the catalytic reaction takes place. Note that in our experiments the metal ions are likely adsorbed onto the negatively charged fused-silica surface and not diffusing freely in solution. Figure 6 implies that the adsorption sites are quite homogeneous. A corollary of these results is that it is possible to detect single metal ions by designing an appropriate catalytic reaction.
Figure 5. Single enzyme molecule reaction kinetics in vials manufactured on a fused-silica plate. Individual LDH-1 molecules are trapped inside the vials with the substrate solution. The product NADH is monitored over time. Background-corrected fluorescence intensities are plotted.
Figure 6. Single metal ion reaction kinetics in vials manufactured on a fused-silica plate. Background-corrected fluorescence intensities are plotted.
was used to fill the vials. The remainder of the bulk solution was monitored visually with respect to its color change while the single metal ion reactions were followed by the CCD/microscope system. If the bulk solution exhibits a color change over time, it constitutes evidence for contamination. Since the background reaction has a reaction rate comparable to that catalyzed by 1.4 × 10-12 M Os(VIII) (shown in Figure 3b), we selected the Os(VIII) concentration to be ∼5 × 10-12 M. This resulted in 33% of the vials (135 fL each in volume) being filled with one or more Os(VIII) ion. The experimental procedure was very similar to that described above. However, the reaction rates for single metal ions were much slower. Thus it took longer (∼1 h) to accumulate enough product ions for establishing a rate curve. For each experiment, more than 100 different sets of data were obtained. We have analyzed the first 20 vials to determine representative reaction kinetics for single metal ions. Of these 20, ∼11 have slight increases in the fluorescence intensity. These correspond to background reaction in vials where there were no Os(VIII) ions present. One vial shows a relatively large increase in signal, as indicated by the heavy line in Figure 6. The slope is 75% larger than the other slopes. This is consistent with the occasional occurrence of a vial with two metal ions. The other eight show moderate increases in signal, indicating that there was one metal ion in each of the vials. These are also plotted in Figure 6. The slopes for these curves are similar, in contrast to those obtained for single enzyme molecule studies. The range in slopes
DISCUSSION There are a few issues of great interest. The first one is the comparison of the two different types of vials for the enzyme reaction. When the membrane vials were used, we observed many short-term changes in the reaction rate plot (Figure 4). In contrast, these slope changes were not observed in the fusedsilica vials (Figure 5). One explanation is that the membrane wall has stronger interactions with the enzyme molecules than the fused-silica wall. If the interaction happens to be near the active sites for catalytic activity of the enzyme molecule, the reaction efficiency will be altered. Apparently, interactions with the membrane wall are reversible. If only two data points on each plot (the beginning and the end) were recorded,5,14 these intermediate changes would not have been recognized. The overall (long-term) activities of individual enzyme molecules are not influenced by interactions between the enzyme molecules and the walls (Figure 4 vs Figure 5). It is known that fused silica does not favor adsorption of LDH-1 5,17 at these pHs, so the lack of surface interactions in Figure 5 is expected. The second comparison is between the reactions catalyzed by enzyme (Figure 5) molecules and by metal ions (Figure 6). It is well-known that the conformations of enzyme molecules play an important role in their catalytic activities.12 For electrophoretically pure enzyme molecules obtained from the same batch from the same company, all the molecules are supposed to be identical. This is in contrast to enzyme preparations where different glycoforms are expected to coexist.14 When identical molecules are put into the same environment, we expect them to possess the same activities. However, for enzyme molecules, we observed different activities for individual molecules, while for metal ions, the observed activities are more similar. Furthermore, if the environments in the different vials are affecting the activities, similar variations will show up in both types of reactions. The variations in enzyme activities can be attributed only to the enzyme molecules themselves. It is the different stable conformations that cause heterogeneities in activities among individual enzyme molecules. This implies that molecular modeling and drug design calculations need to become more sophisticated to account for such conformers at secondary energy minima in biomolecules that coexist over long times. CONCLUSION We have utilized a unique nanosampling technique, together with a UV laser-based optical microscope and CCD detection system for the study of reaction kinetics of enzyme molecules and for the detection of metal ions. We demonstrated that single enzyme molecules or single metal ions can be detected at low concentrations in an ultrasmall volume defined by a nanoscopic vial. More than 100 individual molecules can be followed Analytical Chemistry, Vol. 69, No. 20, October 15, 1997
4247
simultaneously. In principle, this scheme can be extended to any molecule that can catalyze the production of a suitable fluorophor. We not only show that the enzymes can be counted but also that their reaction dynamics can be studied. We note that the large numbers of product molecules contributing to each data point eliminate counting statistics in determining the rate of reaction, while single-event studies4 cannot readily provide rate information. Each enzyme molecule possesses a unique activity. The activities show a 3× difference for the collection of pure LDH-1 molecules. This indicates that different conformations in enzyme molecules may have contributed to the differences in their catalytic activities. Using the same methodology, we have detected single metal ions. These ions display more similar reaction kinetics and catalytic activities among themselves, which is consistent with the lack of conformational variations. One of the major motivations in SMD research is to apply novel SMD schemes to characterize physical and chemical properties that may lead to the discovery of new fundamental principles and to the development of new technology. We believe that SMD based on repeated reaction is well suited for this goal. In this approach, the emphasis is not the detection of single molecules but the study of the properties of individual molecules under different conditions. This can have important implication in many scientific fields, such as drug design and drug interaction with (20) Xu, X.; Yeung, E. S. Science (Washington, D.C.) 1997, 275, 1106-1109.
4248
Analytical Chemistry, Vol. 69, No. 20, October 15, 1997
biomolecules. As far as the detection of one molecule is concerned, the catalytic reaction approach enables us to detect extremely low concentrations of enzymes (7.6 × 10-17 M5) and of metal ions (5 × 10-12 M, this work). If a biomolecule can be tagged with an enzyme, it should then be detectable at the same low level, potentially leading to early disease diagnosis. In addition, detection by reaction amplification allows one to monitor an actual chemical reaction in progress. While not all chemical reactions are expected to be heterogeneous, the ability to follow individual steps in a reaction20 should lead to new kinetic insights. ACKNOWLEDGMENT We thank Dr. Qifeng Xue for technical help in the beginning phase of the metal ion research. W.T. is a U.S. Department of Energy Distinguished Postdoctoral Fellow sponsored by the Office of Science Education and Technical Information and administered by Oak Ridge Institute for Science and Education. The Ames Laboratory is operated for the U.S. Department of Energy by Iowa State University under Contract No. W-7405-Eng-82. This work was supported by the Director of Energy Research, Office of Basic Energy Sciences, Division of Chemical Sciences. Received for review June 17, 1997. Accepted July 23, 1997.X AC970631K X
Abstract published in Advance ACS Abstracts, September 15, 1997.