TE~S 22 - MAY 1997 3 Brack, A. arid Orge!, L E. (1975) Nature 256, 383-387 4 Wachteshauser, G. (1988) Microbiol, Rev. 52, 452-484 5 Joyce, G. F. (1989) Nature 338, 217-224 6 Piccirilli, J. A. et al. (1990) Nature 343, 33-37 7 Finkelstein, A. (1994) Curr. Opin. Struct. Biol. 4, 422-428 8 Dill, K. A. et al. (1995) Protein Sci. 4,561-602 9 Dinner, A. R., Sali, A. and Karplus, M. (1996) Proc. Natl. Acad. Sci. U. S. A. 93, 8356--8361 10 Cech, T. R. (1993) in The RNA World (Gesteland, R. F. and Atkins, J. F., eds), pp. 239-269, Co!d Spring Harbor Laboratory Press 11 Herschlag, D. (1995) J. Biol. Chem. 270, 20871-20874 12 Draper, D. E. (1996) Trends Biochem. Sci. 21, 145-149 13 Narlikar, G. J. and Herschtag, D. (1997) Annu. Rev. Biochem. 66, 19-59 14 Koshland, D. E., Jr (1976) Fed. Proc. 35, 2104-2111 15 Monod, J., Wyman, J. and Changeux, j-P. (1965) J. Mol. Biol. 12, 88-118 LETTERS 15 Kacser, H. and Beeoy, R. (1984) ; L~,: E~o; 20, 38-51 17 Bryngelson, J. D. et al. (1995) Prote, ~c~ Funct Genet. 21,167-195 18 Todd, M. J. et aL (1996) Proc. Natl. A¢o:~ :~,.~ U S. A. 93, 4030-4035 19 Orengo, C. A. et al. (1994) Nature 372, 631-634 20 Pande. V. S. et al. (1994) Proc. Natl. Acad. Sc/ U. S. A. 91, 12976-12979 21 Hartl, F. U. (1996) Nature 381, 571-580 22 Allain, F. H. et aL (1996) Nature 380, 646-650 Kinetic data remiabitity In a recent Talking Point article ~, Schuck and Minton propose two simple consistency tests to ascertain the reliability of kinetic measurements performed with surface plasmon resonance (SPR) biosensors. We agree totally with their suggestions and the necessity to perform these tests. However, we believe that the inclusion of some of our published experiments ~as examples of data that fail these tests is inappropriate. The interaction we studied was characterised by substantial re-binding of dissociating material to the surface. This is a common problem in binding studies, especially for interactions with fast as.sociation rates, and can be overcome either by 'infinite' dilution (which is not practical) or by the addition of an excess of competing ligand during the dissociation phase to prevent re, .nding :~ (see Fig. 2a in Ref. 2). Schuck and Minton wrongly compare the k,,,~ values obtained in this way (and shown in our Table !) with the k,,~,~ calculated from Fig. ld. The k,,,~ should be and is consistent with the 'apparent' low kd~,,value obtained in buffer flow (Fig. 2a), not with the high k,~,, value obtained in the presence of the competing peptide. Concerning consistency test 1, Schuck and Minton somehow calculate a value of approximately 1.3 nM for the equilibrium constant K~ from our data shown in Fig. la. These are six points with the first and last differing by tenfold. We are sure that Schuck and Minton would agree that, in order to obtain an accurate value, a range of concentrations differing by at least several hundred-fold has to be used, including points close to saturation 3. It is also obvious from the same figure that true equilibrium was not reached during the time course of these experiments, again making these data unsuitable for a Eo calculation. Moreover, Schuck and Minton compare the value of 1.3 nM with the K n obtained using the 'true' off rate from Table 1, not the 'apparent' one, as discussed above. We have recently repeated these experiments with recombinant proteins that are not fused to GST. This diminishes the re-binding problem and equilibrium is reached much faster than with the GST-fusion proteins. Using this approach, both the association and dissociation phases can be fitted accurately with simple kinetic models, confirming our published finding of fast association and very fast dissociation rates for SH2 domain-phosphopeptide interactions. The calculated K~ values also match those defined by equilibrium binding assays. We conclude that our published data are not 'self-inconsistent' .....I ,~.., pro.~ded experimental conditions E,UaU LI IQL, and materials are chosen carefully, SPR biosensors using continuous flow (Bib.core) can be reliably used for calculation of kinetic and equilibrium constants. References 1 Schuck, P. and Minton, A. P. (1996) Trends Biochem. Sci. 21, 458-460 2 Panayotou, G. et al. (1993) Mol. Cell. Biol. 13, 3567-3576 3 LJmbJrd,L E. (1986) Jn Cell Surface Receptors: a short course on theory and methods, Martinus Nijhoff Publishing GEORGE PANMGTGLIAND MIKE WATERRELD Ludwig Institute for Cancer Research, 91 Riding House Street, London, UK WlP 8BT. Copyr|ght © 1997, El:~evler 5dence Ltd. All rights reserved. 0968-0004/97/$17.00 Reply to Paaayotou and Wateffie[d Panayotou et al. ~ have evaluated the dissociation rate constant k via analysis of a dissociation experiment conducted i ' the presence of a large excess of competing peptide. We agree that this value of k is much more likely to reflect the true chemical rate constant for dissociation of peptide from SH2 domain than the value obtained from analysis of the association experiment in the context of the elementary 1:1 association model. However, the use of a value of k obtained from the same analysis of the same association experiment with the same oversimplified model only compounds the internal inconsistency of the calculation of the association equilibrium constants K,, (mistakenly labeled dissociation constants) in Table i o~ Panayotou et al.~. We have shown Published by ElsevierScience Ltd. 0968-0004/97 (. PII: S0968-0004(97)t) 1037-2 elsewhere 2 that mass transport effects can result in quasi-linear plots of dR/dr vs R, analysis of which via the elementary association model (neglecting mass transport) yields apparent rate constants k and k~ s' that both are far below the actual intrinsic chemical rate constants for binding. If one calculates an apparept equilibrium association constant by dividing an artifactually low estimate of k by a realistic estimate of k_, then the resulting estimate of K a will be depressed by the same factor as k. The internal inconsistency inherent in the analysts employed by Panayot'Ju and colleagues ~is evident in the limiting long-time behavior of the association experiments plotted in Fig. la, c. One may estimate the equilibrium response R= of the system from Fig. la, d, as described in our note (Table I caption) or, equivalently, by extrapolating the straight lines plotted in Fig. lc to the x-intercept (dR/dr = 0). The resulting dependence of R= on free ligand concentration can be modeled via PII: S0968-0004(97)01038-4 our Eqn 4 to obtain a reasonable estimate of K~ (or K,). We emphasize that this calculation is not subject to the influence of mass transport. When done, one obtains a value of K. approximately equal to 1 riM-= (Ref. 1) for the particular set of data plotted in Fig. 1. It is quP" impossible to describe the long-time limit of these data using the value of 0.0237 nM- (Ref. 1) for the association equilibrium constant as reported by the authors. References I Panayotou, G. et al. (1993) Mol. Cell. Biol. 13, 3567-.3576 2 Schuck, P. and Minton, A. P. (1996) Anal. B~octlem. 240, 262-272 P~ER $CHUCKAND ALLEN P. M|NTON Laboratory of Biochemical Pharmacology, National Institute of Diabetes, Digestive and Kidney Diseases, National Institutes of Health, Bethesda, MD 20892, USA. 149