Life and Death of a Single Enzyme Molecule - American Chemical

---

Report

Life and Death of a Single Enzyme Molecule nzymes are proteins that catalyze biochemical reactions. As with other catalysts, enzymes do not change the equilibrium between substrate and product; they simply speed the reaction rate. By modifying enzymes, nature can control the rate at which spontaneous, but otherwise sluggish, reactions proceed. Biochemists are concerned about the details of enzyme action because of the ubiquitous role of enzymes in biological transformations. Analytical chemists have long used enzymes as reagents in ultrasensitive immunoassays in which the amplifies the signal associated with antibodyantigen binding (2) Given the importance of enzymes to biology and chemistry it is surprising that enzvmes have been the of much research Until recently, almost all enzyme experiments required a huge ensemble of molecules; the results of which measure the average properties of these molecules. An early exception was a paper in 1961 by Rotman, who studied the reaction of a single )3-galactosidase molecule held in small droplets over a 10- to 15-h incubation (2). Recently, Yeung's group and our group have developed sensitive analytical techniques to follow the reaction kinetics and thermodynamics of an individual enzyme molecule (3-6). Surprisingly, enzyme molecules are not identical and instead demonstrate a range of behaviors Details of enzyme behavior which have been hidden in the ensemble average of classic methods are revealed hv studying individual molecules These details provide instant into the

life and death of a molecule Douglas B. Craig Edgar Arriaga Jerome C. Y. Wong

Details of enzyme behavior, previously hidden in the ensemble average of classic methods, are revealed by studying individual molecules. Enzymology review In a simple view of a typical enzymatic reaction, substrate S and enzyme E are in equilibrium with an enzyme-substrate complex ES, which undergoes irreversible reaction to form product P and free enzyme E + S E + P

Hui Lu Norman J . Dovichi University of Alberta (Canada) 0003-2700/98/0370-39A/$15.00/0 © 1997 American Chemical Society

in which kx is the rate of formation of the enzyme-substrate complex, k_x Yi the eate

of dissociation of the complex, and &cat is the rate of production of productfromthe complex. The reaction rate has units of s"1 and measures the number of substrate molecules converted to product by one enzyme molecule; it is sometimes called the "turnover number". According to the MichaelisMenten analysis, the overall reaction rate, or velocity V, is given nb V=d[P]/dt = -d[S]/dt = *cat[ES] = I^max[S]/([S] + Ku)

Analytical Chemistry News & Features, January 1, 1998 39 A

Report in which Vmax is the saturated reaction rate (the reaction rate in the presence of high substrate concentration) that is equal to jfecat[E]0 in whiih [E]0 is the total lcncentration of enzyme. KM is the Michaelis constant that is equal to the concentration of substrate required to generate a rate that is halfof V u u u

VA r

max

All enzymes are not alike

These analyses neglect an important point—enzymes are not pure compounds. Rather, they often generate multiple bands in isoelectric focusing and other electrophoretic purification methods (7,8). These bands can be caused by variations in primary structure, because of the presence of enzymes coded by different genes, or by variations in post-translational modifications such as glycosylation (9). Heterogeneous enzymatic populations also may be attributable to variations in secondary structure as the same primary structure folds into different conformations (3) Several forms of can have different k k and k The reaction rate must be replaced with a weighted average over all the enzyme's forms These averages

phosphate) generates a weak fluorescence signal when excited by an argon-ion laser at 458 nm. The substrate is converted to the highly fluorescent [2-benzthiazole]-6'hydroxy-benzthiazole, which is efficiently excited by the laser. Individual alkaline phosphatase molecules were studied by filling an electrophoresis capillary with a slug of dilute enzyme mixed with a high concentration of substrate. The substrate is present at a concentration well above KM, and the enzyme reaction proceeds at a rate close to the saturated value V . During incubation, each enzyme molecule converts the substrate to pools of highly fluorescent product; one enzyme molecule generates several thousand product molecules. After incubation, the contents of the Ccipillciry cire flushed to a high-sensitivity fluorescence detector. Each pool offluorescentproduct generates a peak with an intensity that is proportional to the amount of product present This technology is a variation of Regnier's enzyme mediated (EMMA) albeit taken

extreme with single en-

Not only can the reaction of a single molecule be studied, but the measure-

ments can be replicated. A 4 x 10"16 M enzyme solution is used to ensure that only one or two enzyme molecules are in the reaction capillary. After an 8-min incubation, a brief electrophoresis pulse is applied to the capillary. From conventional EMMA experiments, we know that the enzyme has higher mobility than the reaction product. The electrophoretic pulse drives the enzyme a short distance into fresh substrate, leaving a pool of fluorescent product behind. A second incubation is performed for 4 min, followed by another electrophoretic pulse to move the into fresh substrate This process is repeated for 2- and 1-min incubations creating four pools of fluorescent product in the capillary After thefinalincubation, the products are electrokinetically swept through the detector. Four peaks are observed with monotonically increasing amplitude (Figure 1). The dashed curve is the least squaresfitof four Gaussian peaks to the data. Peak spacing is uniform, which is expected for a single molecule undergoing electrophoretic separation from the fluorescent product. The peak spacing at 11.1 + 0.3 s is smaller than the 15-s pulse used to move the enzyme into fresh substrate, re-

result

in deviations from Michaelis-Menten behavior Recent advances in high-sensitivity laser-induced fluorescence technology allow the study of reactions that are catalyzed by a single enzyme molecule. Our systems are based on a postcolumn sheathflow cuvette, which dramatically reduces light scatter and allows facile detection of small numbers of molecules, including single molecule detection after separation by capillary electrophoresis (CE) (10-12) Reaction rate of one enzyme molecule

We have worked with alkaline phosphatase, found widely in nature, which cleaves phosphate groups from numerous substrates. The bacterial and mammalian versions of this enzyme are heterogeneous, generating numerous bands in isoelectric focusing (7,8). Several fluorogenic substrates are available to study this enzyme. AttoPhos (2'[2-benzthiazole ] -6' -hydroxy-benzthiazole 40 A

Figure 1 . Peaks generated by multiple incubation. The solid line represents the data; the dashed line represents a least squares fit of four Gaussian peaks to the data.

Analytical Chemistry News & Features, January 1, 1998

fleeting the eelative eelocity oo the enzyme molecule and its pools of fluorescent product during the electrophoretic pulse. The first peak to migrate from the capillary corresponds to the last incubation time, which shows that the enzyme molecule migrates faster than the product molecules. Peak area increases linearly with incubation time. The slope of the plot is an unambiguous and direct measurement of the reaction rate for this enzyme molecule. In contrast, reaction rate measurements performed on bulk enzyme solutions are fraught with systematic errors, because it is difficult to accurately determine the amount of enzyme used in the experiment. Extraneous protein and denatured enzyme confound conventional assays. Of course, the bulk assays are averaged over all forms of the enzyme in solution. The data in Figure 1 verified that the reaction proceeds with pseudo-zerothorder kinetics. As a result, the rate can be estimated after a single incubation, which greatly simplifies the experimental procedure. The activity of 83 different enzyme molecules was measured and found to be heterogeneous (Figure 2). The drop-off in activity below 25 s"1 appears to be real; our instrument should detect activity of ~6 s"1 in a 20-min. incubation. The activity tends to cluster around 60 160 and 290 s_1; howit is not clear if this clustering is real or if it is caused by the relatively small number of points used to define the distribution The general shape of the activity distribution is similar to that for lactate de-

Figure 2. Histogram of reaction rates for 83 alkaline phosphatase molecules. Activity was determined from peak area, which was determined by a nonlinear least-squares fit of either one or two Gaussian functions.

be measured. As in the repllcate incubation study, a dilute enzyme solution is introduced into the capillary and thefirstincubation is performed at 16 °C for 15 min. After the first incubation, the enzyme is moved into fresh substrate with an electrophoretic pulse. The temperature is simultaneously raised to 24 °C by blowing hot air ovee the eapillary with a thermal controller. After a second 15-min incubation, the enzyme molecule is hvHrotrpnase bacterial alkaline phosmoved with another electrophoretic pulse The thermodynamic life of a phatase and B-ealactosidase {2-6) and the temperature is raised to 30 °C. After molecule a final incubation the contents of the capilEnzymes are catalysts that lower the activaSeveral possible factors can cause this lary are flushed to the detector tion energy requiied to convert the eubstrate broad distribution in enzyme reaction rate. to product. As wiih other thermodynamic Enzyme aggregates would cause heterogeFigure 3 shows the results from one exneous activity. However, zone electrophore- measurements, activation energy has always periment. A set of three equally spaced been measured from a large ensemble of sis shows that the enzyme migrates as sevpeaks is shown, corresponding to the incumolecules. Also, as wiih other thermodyeral closely spaced bands with no evidence bation at the three temperatures. Eight molfor aggregates. The distribution could reflect namic properties, activation energy is govecules were studied by this method. Peak poor experimental precision. However, repli- erned by the ergotic hypothesis, which esarea, estimated from the regression paramecate incubation studies measure activity with sentially states that experiments on numerters, increases with temperature. Arrhenius a precision of 10% or better; poor precision is ous molecules can be replaced by many plots (In peak area vs. 1/T) are linear with experiments on a single molecule (14,15). not a major factor in the distribution. Enslope = -EJR, in which E ai the ectivation The development of high-sensitivity anazymes could stick to the capillary walls, with energy for the reaction catalyzed by the enlytical tools now allows us to probe individthe active site partially hidden hindering zyme molecule and R is the gas constantt ual molecules and to test the ergotic hypoth- The activation energies varied from 39 to access to the substrate Conventional esis In particular the activation energy of EMMA studies and our multiple incubation 91 kj/mol with a mean of 53 kj/mol lnd the the reaction catalyzed by one molecule can study confirm that the enzyme molecules standard deviation of the distribution migrate in solution. Enzymes could denature during the assay; those molecules with low apparent activity would have denatured early in the incubation period. However, the multiple incubation data show no evidence for loss of activity during subsequent incubations. For numeeous srasons, we eelieve that these differences in activity reflect variances in molecular structure.

Analytical Chemistry News & Features, January 1, 1998 41 A

Report

Figure 3. Electropherogram of alkaline phosphatase thermodynamic analysis. A solution of 4.6 x 10~16 M alkaline phosphatase in 100-mM borate (pH 9.5) containing 1 mM AttoPhos was injected onto the capillary. Sample was incubated for three 15-min periods at 16, 24, and 30 °C, interrupted by brief periods of high voltage. Following the third incubation, the capillary contents were electrokinetically swept through the detector.

16 kj/mole. This range of activation energy is much larger than the experimental uncertainty, and it reflects the heterogeneous nature of this enzyme. Activation energy and the pre-exponetial term of the Arrhenius equation were highly correlated with an intercept near zero. We also measured the activation energy of a bulk solution of alkaline phosphatase. The activation energy was 50 kj/mol, which is quite close to the average of our single molecule results. As predicted by the ergotic hypothesis, the activation energy obtained from the time average of the eight molecules is equal at the 90% confidence limit to the ensemble average obtained from ~3 x 107 molecules. No correlation is found between activation energy and activity. The dephosphorylation of substrates by alkaline phosphatase has been proposed to occur via

42 A

Analytical Chemistry News & Features,

in which EH is the free enzyme, R-OP is the substrate, EHR-OP is the enzyme substrate complex, EH*R-OP is the activated enzyme substrate complex, EP is the phosphorylated enzyme, and P is inorganic phosphate (16). The first step in the mechanism represents ligand binding. The second step represents the catalytic reaction, and it should dominate the activation energy. The third step represents reaction product loss. The equilibrium associated with ligand binding may dominate the activity of the enzyme. Death of an enzyme by thermal denaturation

How do enzymes die? An enzyme can be permanently denatured by heating it or treating it with reagents that disrupt hydrogen bonds. Thermal denaturation is particularly interesting because it mimics, in reverse, the steps necessary for a protein to fold into an active conformation. During heating, hydrogen bonds are broken in succession, unfolding the molecule. For protein molecules, birth and

death are similar processes carried out in reverse order. Until now, all studies of denaturation have been performed on large ensembles of molecules. During denaturation, activity is gradually lost. For example, if alkaline phosphatase is heated to 64 °C for 5 min, the activity is half the starting activity. Longer heating ultimately leads to a complete loss of activity. We postulated two different models for denaturation. In the first, the "Cheshire Cat" model, hydrogen bonds break sequentially leading to a gradual unfolding of the molecule (4). The breaking of bonds far from the active site results in subtle changes in the conformation of the active site. As more hydrogen bonds break, activity is gradually lost. This model explains the heterogeneity that we observe in enzyme activity; the range of activity is a result of a range of secondary structures of the molecule. Essentially different enzyme molecules have been caught in numerous conformations along the denaturation pathwav If the Cheshire Cat model accurately describes thermal denaturation, then heating alkaline phosphatase to a temperature of 64 °C for 5 min should result in the shift of the distribution of enzyme activity to an average that is half the activity of a control solution. It is expected that the number of active molecules does not change dramatically upon heating. Instead, all molecules become less active. In the second model, "catastrophic denaturation", the breaking of hydrogen bonds is viewed as a cooperative phenomenon. The breaking of one hydrogen bond increases the probability that a neighboring bond will break, leading to a catastrophic disruption of a large domain in the molecule (4). This model is similar to that used to describe thermal denaturation of double-stranded DNA, which is a highly cooperative breaking of the hydrogen bonds that hold the molecule together. If the catastrophic denaturation model is accurate then heating alkaline phosphatase to 64 °C for 5 min should result in a population of molecules whose distribution of activity does not change aooreciably imon heating Instead we expect that the loss in overall activity arises becaime only half the molprnlpQ QiirvivpH heatincr

Experiments were performed to study thermal denaturation of alkaline phosphatase. A control solution of enzyme was left untreated. Another solution of alkaline phosphatase was heated to 64 °C for 5 min and then immediately diluted by 6 orders of magnitude to quench the denaturation process. The number and activity of the enzyme molecules in both solutions were measured. Within experimental error, half the enzyme molecules survived the treatment and their activity distribution was indistinguishable from the control solution. Our single molecule study allows detailed characterization of the death of an enzyme. The molecule denatures catastrophically, with no evidence of a gradual shift in conformation. This result may also hold in reverse: Protein folding may be a gradual process until a critical set of hydrogen bonds is formed whereupon the enzyme molecule cooperatively folds into an active conformation Postmortem The intimate details of the life and death of individual enzyme molecules can now be studied. Unlike the textbook model, which illustrates all molecules of a given chemical species as being identical, enzyme molecules act as if they have different personalities. The kinetic and thermodynamic properties of mammalian enzyme molecules are broadly distributed. What accounts for these distributions in the behavior of different alkaline phosphatase molecules? The central paradigm of chemistry states that structure affects function. Differences in function must arise from some differences in structure Two possible causes of structural differences exist. In the simplest case, all enzyme molecules consist of the same atoms arranged in the same order. Differences in chemical behavior arise from variances in the secondary structure into which the molecule has folded. This argument is a particularly attractive explanation for the behavior of enzymes, such as lactate dehydrogenase that generates a single band in gel electrophoresis. However, individual molecules generate a broad distribution of activities (3) On the other hand, our thermal dena-

turation study of alkaline phosphatase suggests that these conformations do not easily interconvert upon heating; we saw no change in the distribution of molecular activities after heating. The linear kinetic curves further demonstrate that the activity of a molecule is constant over a 15-min period. In the second case, which appears to be quite common for most enzymes and particularly for eukaryotic enzymes, a vast menagerie of post-translational modifications are found in nature. In this case, molecules vary in activity because of differences in primary structure. Proteins, particularly those found in eukaryotes, are usually decorated with sugars, lipids, phosphates, and other functional groups. A protease often forms the protein from a larger apoprotein which is the original translation product from mRNA; different cutting sites lead to different product molecules. These post-translational modifications are generated in a somewhat haphazard fashion leading to a population of product moleniles that contain a range of modifications FurtVier r n m p l i c a t i n n n r n i r s for enTVtnp'! cn^Ti QC alV^linf* ntiocntiataQP wVlirti zire

more monomer pen combine bled products. We have recently investigated bacterial alkaline phosphatase, which also demonstrates a range of activities (5). This molecule is not glycosylated in nature. However, the molecule generates at least three bands in gel electrophoresis, which arise from removal of an ./V--erminal arginine from one or both strands of the dimeric enzyme. We measured the activity of 30 bacterial alkaline phosphatase molecules; the activity ranged fivefold Unlike the mammalian enzyme the activity of the bacterial enzyme clustered into three groups which is consistent with the observation of three forms of the dimeric en7vme Future work will measure the activity of individual molecules isolated from the isnplprtric fftciisinp" (rel These studies on single enzyme molecules are exciting and interesting because we can study several basic properties that are otherwise lost in the ensemble average of conventional measurements. We need to understand the role that these different

classes of alkaline phosphatase (and other enzymes) play in nature. We can also confirm some basic tenets of statistical thermodynamics. Finally, it's a very powerful analytical tool. The ability to detect the products generated by a single enzyme molecule will be valuable in enzyme assays with the ultimate in sensitivity. The work was supported by an operating grant from the Natural Sciences and Engineering Research Council. DBC acknowledges a postdoctoral fellowship and JCYW acknowledges a predoctoral summer fellowship from the Alberta Heritage Foundation for Medical Research. References (1) Engvall, E.; Perlmann, P.J. Immunology 1972,109,129-35. (2) Rotman, B. Proc. Natl. Acad. Sci. USA 1961, 47,1-6. (3) Xue, Q.; Yeung, E. Nature e195,373,68182. (4) Craig, D. B.; Arriaga, E. A.; Wong, J.C.Y.; Lu, H.; Dovichi, N. J./. Amer. Chem. Soc. 1996,118, 5245-53. (5) Craig D. B.; Dovichi, N. J. Unpubllshed results. (6) Tan, W.; Yeung, E. S. Anall Chem. 1997, 69,4242-48. (7) Engstrom, L. Biochim. Biophys. Acta 1961,52,36-48. (8) Saini, P. K.; Done, J. Biochim. Biophys. Acta 1972,258,147-53. (9) Rudd, P. W. et all Biochemistry 1994,33, 17'-22. (10) Cheng, Y. F.; Dovichi, N. J. Science 1988, 242, 562-64. (11) Wu,S.; Dovichi, N.J./. Chromatogr. 19898 480,141-55. (12) Chen, D.Y.; Dovichi, N. J. Anall Chem. 1996, 68, 690-96. (13) Bao, J.; Regnier, F. E.J. Chromatorgr. 1992, 608, 217-24. (14) Hill, T. L. An Introduction to Statistical Thermodynamics; Addison-Wesley Publishing Company: Don Mills, Ontario, 1960; p.l. (15) Gladstone, S.; Laidler, K. J.; Eyring, H. The Theory of Rate Processes; McGrawHill: New York, 1941; p. 108. (16) Price, N. C; Stevens, L. Fundamentals of Enzymology; Oxford University Press: New York, 1984; pp. 147-48. Douglas B. Craig is an assistant professor at the University of Winnipeg (Canada). Edgar Arriaga is a research associate, and Jerome C. Y. Wong is an undergraduate, at the University ofAlberta. Hui Lu is an employee of Sperry-Sun (Canada). Norman J. Dovichi is a professor, and correspondence about this article can be addressed to him at the Dept. of Chemistry, University ofAlberta, Edmonton Alberta T6G 2G2 Canada.

Analytical Chemistry News & Features, January 1, 1998 43 A