Protocol for the Thermodynamic Analysis of Some Proteins Using an H

were the protein under study is a two-state folder (i.e., partially folded intermediate states are not populated in the equilibrium unfolding reaction...
0 downloads 0 Views 96KB Size
Download PDF
Anal. Chem. 2005, 77, 693-697

Protocol for the Thermodynamic Analysis of Some Proteins Using an H/D Exchange- and Mass Spectrometry-Based Technique Susie Y. Dai, Myles W. Gardner, and Michael C. Fitzgerald*

Department of Chemistry, Duke University, Durham, North Carolina 27708

SUPREX (stability of unpurified proteins from rates of H/D exchange) is a new H/D exchange- and mass spectrometry-based technique for the measurement of protein folding free energies (i.e., ∆G values) and protein folding m values (i.e., δ∆G/δ[denaturant]). Robust protocols for the acquisition and analysis of SUPREX data have been established and shown to be useful for the analysis of a number of different protein systems. Here we report on the SUPREX behavior of a special class of proteins that are not amenable to conventional SUPREX analyses using previously established protocols. This class of proteins includes protein systems that require an extended time to reach a folding/unfolding equilibrium in chemical denaturant-induced equilibrium unfolding experiments. As part of this work we use ubiquitin as a model system to highlight the complications that can arise in the conventional SUPREX analysis of such protein systems, and we describe a modified SUPREX protocol that can be used to eliminate these complications. Stability of unpurified proteins from rates of H/D exchange (SUPREX) is a new mass spectrometry-based technique for making thermodynamic measurements of a protein’s conformational stability.1-6 The SUPREX technique is an attractive method by which to evaluate protein folding free energies (i.e., ∆G values) and protein folding m values (i.e., δ∆G/δ[denaturant] values) as it offers several unique experimental advantages over conventional methods for making such measurements. These experimental advantages include those of speed, sensitivity, and ability to analyze both highly purified protein analytes as well as protein analytes in complex biological mixtures.7-10 * Corresponding author. Tel: 919-660-1547. Fax: 919-660-1605. E-mail: [email protected]. (1) Ghaemmaghami, S.; Fitzgerald, M. C.; Oas, T. G. Proc. Natl. Acad. Sci. U.S.A. 2000, 97, 8296-8301. (2) Powell, K. D.; Wales, T. E.; Fitzgerald, M. C. Prot. Sci. 2002, 11, 841851. (3) Powell, K. D.; Ghaemmaghami, S.; Wang, M. Z.; Ma, L.; Oas, T. G.; Fitzgerald, M. C. J. Am. Chem. Soc. 2002, 124, 10256-10257. (4) Powell, K. D.; Fitzgerald, M. C. Biochemistry 2003, 42, 4962-4970. (5) Powell, K. D.; Wang, M. Z.; Silinski, P.; Ma, L.; Wales, T. E.; Dai, S. Y.; Warner, A. H.; Yang, X.; Fitzgerald, M. C. Anal. Chem. Acta 2003, 496, 225-232. (6) Ma, L.; Fitzgerald, M. C. Chem., Biol. 2003, 10, 1205-1213. (7) Powell, K. D.; Fitzgerald, M. C. Anal. Chem. 2001, 73, 3300-3304. (8) Ghaemmaghami, S.; Oas, T. G. Nat. Struct. Biol. 2001, 8, 879-882. (9) Powell, K. D. Fitzgerald, M. C. J. Comb. Chem. 2004, 6, 262-269. 10.1021/ac048967z CCC: $30.25 Published on Web 12/10/2004

© 2005 American Chemical Society

Robust protocols for the acquisition and analysis of SUPREX data have been previously established and shown to be useful for the analysis of a number of different protein systems.1-10 The basic SUPREX protocol involves diluting aliquots of the protein sample into a series of deuterated H/D exchange buffers that contain different concentrations of a chemical denaturant, either guanidinium chloride (GdmCl) or urea. Labile protons in the protein are allowed to exchange with solvent deuterons for a specified length of time. After this specified time, the deuterium content of the protein sample in each H/D exchange buffer is determined using matrix-assisted laser desorption/ionization (MALDI) mass spectrometry. The deuterium content of the protein in each buffer (i.e., the change in mass relative to the fully protonated sample, or ∆mass) is plotted as a function of [denaturant] in the buffers, and the data are fit to a sigmoidal equation to extract a C1/2SUPREX value (i.e., the [denaturant] at the transition midpoint). In cases were the protein under study is a two-state folder (i.e., partially folded intermediate states are not populated in the equilibrium unfolding reaction) and exhibits EX2 exchange behavior (i.e., the protein’s folding rate is much faster than the intrinsic exchange rate of an unprotected amide proton, kint), the extracted C1/2SUPREX value can be used to calculate the protein’s folding free energy (i.e., ∆G value).2-6 Here we report on a class of proteins that is not amenable to SUPREX analysis using the basic SUPREX protocol outlined above. This class of proteins includes protein systems that require an extended time to reach a folding/unfolding equilibrium in chemical denaturant-induced equilibrium unfolding experiments. As part of this work we use ubiquitin as a model system to highlight the complications that can arise in the conventional SUPREX analysis of such protein systems, and we describe a modified SUPREX protocol that can be used to eliminate these complications. MATERIALS AND METHODS Reagents. Deuterium oxide (99.9 atom % D), sodium deuterioxide, and deuterium chloride were purchased from Aldrich (Milwaukee, WI). Deuterated phosphoric acid was obtained from Cambridge Isotope Laboratories (Andover, MA), and GdmCl (OmniPur) was from EM Science (Gibbstown, NJ). Sinapinic acid (SA) was either from Acros Organics (Pittsburgh, PA) or from Aldrich. Trifluoroacetic acid was from Halocarbon (River Edge, (10) Wang, M. Z.; Shetty, J. T.; Howard, B. A.; Campa, M. J.; Patz, E. F.; Fitzgerald, M. C. Anal. Chem. 2004, 76, 4343-4348.

Analytical Chemistry, Vol. 77, No. 2, January 15, 2005 693

The protonated and deuterated SUPREX buffers employed in our analysis of ubiquitin contained 20 mM sodium phosphate buffer (pH 7.4 or pD 7.4) and increasing amounts of GdmCl. A series of ubiquitin SUPREX curves were recorded using H/D exchange times that varied between 8 s and 2 h. These ubiquitin SUPREX curves were collected using both the modified SUPREX protocol outlined in Figure 1 and the basic protocol that has been generally used in all SUPREX analyses reported to date. The C1/2SUPREX values extracted from each SUPREX curve were plotted as a function of the H/D exchange time according to eq 1 as described by Powell and Fitzgerald in ref 4.

RT ln(〈kint〉t/0.693 - 1) ) -mC1/2SUPREX - ∆G

Figure 1. Schematic representation of the modified SUPREX protocol described in this work.

NJ), and acetonitrile was from Fisher (Pittsburgh, PA). Trizma base and the protein samples, bovine ubiquitin and horse heart cytochrome c, were purchased from Sigma and used without further purification (St. Louis, MO). General Methods and Instrumentation. MALDI mass spectra were collected on a Voyager DE Biospectrometry Workstation (Perspective Biosystems, Inc., Framingham, MA) in the linear mode using a nitrogen laser (337 nm). SA was used as the matrix in all experiments. All spectra were obtained in positiveion mode using an acceleration voltage of 25 kV, a grid voltage of 23 kV, a guide wire voltage of 75 V, and a delay time of 225 ns. GdmCl concentrations were determined using a Bausch & Lomb refractometer as described elsewhere.11 A Jenco 6072 pH meter equipped with a Futura calomel pH electrode from Beckman Instruments was used for pH measurements. The measured pH values of the deuterated buffer solutions were corrected for isotope effects to obtain pD values. These pD values were obtained by adding 0.4 to the pH value. Modified SUPREX Protocol. A schematic representation of the modified SUPREX protocol developed in this work is shown in Figure 1. The modified SUPREX protocol includes a step (i.e., step 2 in Figure 1) that is not included in the basic protocol utilized in all SUPREX experiments reported to date. This additional step in the modified SUPREX protocol involves a dilution of the protein sample into protonated SUPREX buffers that contain increasing concentrations of the chemical denaturant. The protein samples are allowed to reach a folding/unfolding equilibrium in these protonated SUPREX buffers before they are transferred to a series of deuterated SUPREX buffers. This equilibration time can vary from protein to protein. In the case of ubiquitin, a 30-min equilibration time was sufficient for the protein to reach a folding equilibrium in each denaturant-containing buffer. It is important that the composition (e.g., pH and [denaturant]) of the protonated and deuterated SUPREX buffers be identical. This ensures that the protein is at equilibrium prior to the initiation of H/D exchange. (11) Pace, C. N. Methods Enzymol. 1986, 131, 266-280.

694

Analytical Chemistry, Vol. 77, No. 2, January 15, 2005

(1)

In eq 1, R is the gas constant, T is the temperature (in kelvin), 〈kint〉 is the estimated average intrinsic exchange rate of an unprotected amide proton, t is the H/D exchange time used in SUPREX, m is defined as δ∆G/δ[denaturant], and ∆G is the folding free energy of the protein in the absence of denaturant. A linear least-squares analysis of the data was used to evaluate the slope and y-intercept that correspond to the protein folding m value and ∆G value, respectively. The error reported with the SUPREXderived ∆G and m values in this work represents the fitting error in our linear least-squares analysis. Either one of two different 〈kint〉 values were used in our calculations of ∆G values using eq 1. One value, 18.1 s-1, was determined using the SPHERE program and the entire primary amino acid sequence of ubiquitin.12,13 The other value, 6.1 s-1, was determined using the SPHERE program and the portion of ubiquitin’s primary amino acid sequence that is buried in the hydrophobic core of the folded protein. These buried regions were identified by visual inspection of ubiquitin’s three-dimensional structure.14 The buried amide protons included in this second calculation of 〈kint〉 were the backbone amide protons of the following residues in ubiquitin’s primary amino acid sequence: 1, 3-5, 7-9, 13, 15, 17, 19, 23, 25-28, 30, 34, 36-38, 40, 41, 4345, 47, 50, 53, 56, 61, 67, and 69-73. We have previously shown that eq 1 can be used to evaluate reasonably accurate and precise ∆G values and m values in a wide variety of protein systems provided that the protein folding reaction under study is well modeled by a two-state folding process (i.e., significant populations of partially folded intermediate states do not exist) and that the protein under study is in the EX2 exchange regime (i.e., the protein’s folding rate is greater than kint).2-6 Under the conditions of the SUPREX experiments performed here, ubiquitin appeared to exhibit EX2 exchange behavior as judged by the fact that only one population of deuterated protein molecules was detected in the mass spectrometry readout of our SUPREX experiment. There was no mass spectral evidence for ubiquitin being in the so-called EX1 exchange regime (i.e., its protein folding rate was slower than the intrinsic exchange rate of an unprotected amide proton) as two distinct populations of (12) Zhang, Y. Z. Ph.D. Thesis, Structural Biology and Molecular Biophysics, University of Pennsylvania, 1995. (13) (a) Bai, Y.; Milne, J. S.; Mayne, L.; Englander, W. S. Proteins: Struct. Func. Gen. 1993, 17, 75-86. (b) Loftus, D.; Gbenle, G. O.; Kim, P. S.; Baldwin, R. L. Biochemistry 1986, 25, 1428-1436. (14) Vijay-Kumar, S.; Bugg, C. E.; Cook, W. J. J. Mol. Biol. 1987, 194, 531544.

Table 1. Transition Midpoints (i.e., C1/2SUPREX Values) of Ubiquitin SUPREX Curves Obtained in This Work C1/2SUPREX values (M)a

Figure 2. Typical SUPREX curves recorded for ubiquitin. The closed circles represent data collected using the basic SUPREX protocol, and the open circles represent data collected using the modified SUPREX protocol described in this work. The same H/D exchange time, 30 s, was used in both cases. The lines represents the best fit of each data set to a standard sigmoidal equation. The arrows denote the C1/2SUPREX value of each curve. The error bars indicate (1 standard deviation from at least five replicate determinations.

exchange time (s)

conventional protocol

8 15 30 60 120 180 300 600 900 1800 3600 5400 7200

3.98 3.83 3.41 2.88 2.44 2.38 2.02 1.86 1.83 1.72 1.54 1.29 1.28

modified protocol 3.02 2.87 2.70 2.30 1.60 1.17

a Values were taken from the best fit of each SUPREX curve. Errors were typically between (0.1 and (0.2.

deuterated protein molecules were not detected (i.e., one population in which all the globally protected Hs were exchanged and one population in which none of the globally protected Hs were exchanged). We also note that the results of our conventional biophysical studies on ubiquitin under solution conditions similar to our SUPREX analyses (see below) as well as the results of other studies on ubiquitin under lower pH conditions indicate that the denaturant-induced equilibrium unfolding properties of ubiquitin are well modeled by a two-state process.15,16 RESULTS AND DISCUSSION Ubiquitin was subject to SUPREX analyses using both the modified SUPREX protocol described here and the basic SUPREX protocol that has been generally used in all SUPREX experiments reported to date. Both protocols were used to generate a series of ubiquitin SUPREX curves using different H/D exchange times. Shown in Figure 2 are typical SUPREX curves that were obtained in these experiments. Summarized in Table 1 are the C1/2SUPREX values extracted from the SUPREX curves generated in this work. In Figure 3, these C1/2SUPREX values are plotted as a function of H/D exchange time and fit to eq 1. Equation 1 predicts that all the points in the plots shown in Figure 3 should be collinear. A visual inspection of the plots in Figure 3 reveals that all the data points lie very close to a single line with the exception of the three data points collected using the conventional protocol at short exchange times (i.e., e30 s). These data points are marked with an arrow in Figure 3. In the case of these three data points, it appears that the observed C1/2SUPREX values are shifted to aberrantly high denaturant concentrations. Interestingly, the magnitude of this aberrant shift is increased as the exchange time is decreased. For example, when the exchange time was 60 s, the observed shift was small (i.e., 0.18 M GdmCl) and within the error of our C1/2SUPREX value determinations, typically (0.15 M. However, when shorter exchange times were used (i.e., e30 s), the observed shift was significant (i.e., g0.5 M GdmCl) and clearly measurable. We attribute the aberrantly large C1/2SUPREX values observed in our SUPREX analyses of ubiquitin using the conventional (15) Ibarra-Molero, B.; Loladze, V. V.; Makhatadze, G. I.; Sanchez-Ruiz, J. M. Biochemistry 1999, 38, 8138-8149. (16) Pan, Y.; Briggs, M. S. Biochemistry 1992, 31, 11405-11412.

protocol at short exchange times to the relatively long time it takes the protein to reach a folding/unfolding equilibrium in the H/D exchange buffers used in our experiments. Using CD spectroscopy, we determined that as long as 20-30 s was required for the folded ubiquitin sample to reach a folding/unfolding equilibrium in some of the deuterated H/D exchange buffers employed in our conventional SUPREX analyses (data not shown). This means that the H/D exchange reaction in some of our SUPREX buffers proceeded well before the ubiquitin sample reached a folding/unfolding equilibrium in the denaturant-containing SUPREX buffer. Consequently, the ∆mass measurements recorded in our SUPREX experiments did not accurately reflect the “equilibrium” unfolding/folding properties of the protein as it was not in a true equilibrium state. The modified SUPREX protocol described here provides a specific time for the protein to reach its appropriate folding/ unfolding equilibrium in each denaturant containing SUPREX buffer before the H/D exchange reaction is allowed to proceed. This separation of the equilibration and H/D exchange times was critically important for the accurate determination of ubiquitin’s protein folding m value by SUPREX. A linear least-squares analysis of the ubiquitin SUPREX data collected using only the conventional protocol yielded an m value of 1.40 ( 0.06 kcal mol-1 M-1 (see Figure 3A). This m value is significantly lower than the value obtained from our linear least-squares analysis of the ubiquitin SUPREX data collected using the conventional protocol at long exchange times yielded, 2.08 ( 0.15 kcal mol-1 M-1(see Figure 3B). It is also lower than the value obtained from our linear leastsquares analysis of the ubiquitin SUPREX data collected using the modified SUPREX protocol, 1.90 ( 0.08 kcal mol-1 M-1(see Figure 3C). Moreover, it is noteworthy that the latter two SUPREXderived m values for ubiquitin (i.e., the values obtained from the data fitting in Figure 3B and C) are both very similar to the ubiquitin m value of 2.03 kcal mol-1 M-1, which can be estimated from the graphical m value data presented in ref 15. We note that the data in ref 15 was obtained under buffer conditions (10 mM sodium acetate, pH 4.2) different from our SUPREX buffer conditions (20 mM sodium phosphate, pD 7.4). However, we note that the results of our own CD denaturation studies on ubiquitin Analytical Chemistry, Vol. 77, No. 2, January 15, 2005

695

Figure 3. Best fits of our SUPREX data to eq 1 using a linear leastsquares analysis. Plots of RT ln(〈kint〉t/0.693 - 1) vs C1/2SUPREX value are shown, and the same data points are plotted in each panel. The closed circles represent data collected using the conventional SUPREX protocol that has been previously described, and the open circles represent data collected using the modified SUPREX protocol described in this work. The data points marked with an arrow were acquired using the conventional SUPREX protocol and short (i.e., e60 s) exchange times. The lines represent the best fit of the data. (A) Only the data collected using the conventional SUPREX protocol are fit to eq 1. (B) Only the data collected using the conventional SUPREX protocol and long exchange times are fit to eq 1. (C) Only the data collected using the modified SUPREX protocol described here are fit to eq 1.

performed at pH 7.4 using a 20 mM tris buffer produced an m value of 2.13 ( 0.16 kcal mol-1 M-1. The similarity between the m values we determined for ubiquitin in our CD denaturation studies under the conditions of our SURPEX experiments and the m values reported in ref 15 strongly suggest that the equilibrium unfolding properties are very similar (i.e., both two-state) under the two different pH conditions. The three different linear least-squares analyses of our data in Figure 3A-C yielded ubiquitin ∆G values of -8.61 ( 0.16, -9.78 ( 0.28, and -9.35 ( 0.19 kcal mol-1, respectively. The correlation coefficients for these linear least-squares analyses were 0.9763, 0.9822, and 0.9912, respectively. It is noteworthy that the highest correlation coefficient was observed in the linear least-squares analysis of the data points collected using the modified SUPREX protocol. It is also interesting to note that the ∆G value extracted from the linear least-squares analysis in Figure 3A (i.e., using all the data points collected using the conventional protocol) is within ∼10% of the values obtained from the other two linear least-squares analyses in which the data points collected using the conventional protocol and short exchange times (i.e., the data points in Figure 696 Analytical Chemistry, Vol. 77, No. 2, January 15, 2005

3 that are highlighted with an arrow) were excluded. These three data points had a relatively small impact on the ∆G value determined for ubiquitin using the SUPREX technique. This happened to be the case in this situation because a large majority of data points in the plots shown in Figure 3 were acquired using relatively long exchange times. If the data points collected using the conventional protocol and the six shortest exchange times were used to generate a ∆G value, for example, a substantially different value of 6.58 kcal mol-1 would be calculated and a high correlation coefficient would be observed (i.e., 0.9947). A ∆G value of -7.77 kcal mol-1 can be estimated from the graphical data presented in ref 15 for the conventional chemical denaturant-induced equilibrium unfolding of ubiquitin. We note that the solution conditions (i.e., 10 mM sodium acetate, pH 4.2) in this earlier study that employed circular dichroism as the structural probe were not exactly the same as the solution conditions (i.e., 20 mM sodium phosphate buffer, pD 7.4) in the SUPREX study described here. However, the results of a CD denaturation experiment performed on ubiquitin in our laboratory using a 20 mM tris buffer, pH 7.4 (i.e., essentially the same solution conditions as those used in our SUPREX experiments) yielded a ∆G value of -8.47 ( 0.25 kcal mol-1 (data not shown). This value and the corresponding m value (i.e., 2.13 ( 0.16 kcal mol-1 M-1) are within 10% of the values experimentally determined from the best fit of our SUPREX data, -9.35 ( 0.19 kcal mol-1 and 1.90 ( 0.08 kcal mol-1 M-1, respectively (see Figure 3C). The accuracy of ∆G values determined by SUPREX is critically dependent on the 〈kint〉 value that is used in eq 1. The 〈kint〉 value used in eq 1 represents the average intrinsic exchange rate of all the globally protected amide protons in the protein under study. The temperature, pH, and primary amino acid sequence dependence of the intrinsic exchange rates of unprotected amide protons in model dipeptide systems have been well studied.12,13a The GdmCl concentration dependence of intrinsic exchange rates of amide protons has also been studied on poly(DL-alanine) at pHs 1-5, and the rates do not change by more than a factor of 2 in GdmCl concentrations ranging from 0 to 6 M.13b The largest changes in kint values are observed with changes in amino sequence.12,13a Depending on the amino acid side chains that are adjacent to the amide bond, kint values can vary by as much as 10-fold. However, it is important to note that the impact of such variabilities in kint values on our SUPREX-derived ∆G values is minimal as the 〈kint〉 term in eq 1 is in the natural log term. A reasonably good estimate of 〈kint〉 values for SUPREX analyses can be obtained by averaging the estimated kint values of each dipeptide segment in regions of the protein where the amide protons are globally protected in the protein’s native threedimensional structure. In theory, only those amide protons that are globally protected should be included in the calculation of 〈kint〉. Amide protons that are unprotected or only locally protected from exchange should not be included in the calculation as their H/D exchange behavior is not probed in the SUPREX experiment. This is because the H/D exchange rates (i.e., kex values) of these amide groups do not display a strong denaturant dependence. In our earlier SUPREX studies, we have typically calculated 〈kint〉 values by averaging the appropriate kint values of each dipeptide segment in the protein’s entire amino acid sequence (i.e., whether the region was globally protected or not). This

method of estimating 〈kint〉 values is especially convenient as it does not require prior knowledge about which regions of the sequence are globally protected and which regions are not. In the case of ubiquitin, it is possible to identify the globally protected amides (i.e., those that are buried in the hydrophobic core of the protein) by examination of its three-dimensional structure. A 〈kint〉 value of 6.1 s-1 can be calculated for ubiquitin using only the amino acid sequences of the regions of the protein’s polypeptide chain that are buried in the protein’s three-dimensional structure (see Materials and Methods). This 〈kint〉 value is ∼3-fold lower than the 〈kint〉 value of 18.1 s-1 derived from the protein’s entire amino acid sequence. Interestingly, if such a 〈kint〉 value of 6.1 s-1 (rather than 18.1 s-1) is used in eq 1 to analyze the data in Figure 3C, the m value remains unchanged (i.e., 1.90 ( 0.08 kcal mol-1 M-1) and essentially identical to that determined by us (data not shown) and by others15 in a conventional CD denaturation experiment on ubiquitin (i.e., 2.13 ( 0.16 and 2.03 kcal mol-1 M-1, respectively). However, the ∆G value changes from -9.35 ( 0.19 kcal mol-1 to a new value of -8.71 ( 0.17 kcal mol-1. This new value is within experimental error of that measured in our conventional CD denaturation experiment on ubiquitin (i.e., -8.47 ( 0.25 kcal mol-1) (data not shown). For two-state folding proteins, SUPREXderived ∆G and m values are expected to be the same as those values derived in conventional CD denaturation experiments. The number of proteins that require relatively long times to reach a folding/unfolding equilibrium in the SUPREX experiment is difficult to estimate. Certainly, the acquisition of preliminary CD spectroscopy data can be useful for the identification of such behavior in unknown protein systems. However, in the absence of such data, our results reveal that this behavior can be detected in the conventional SUPREX experiment. For example, our conventional SUPREX data were not well fit to eq 1 when SUPREX

curves with a wide range of C1/2SUPREX values are considered in the analysis. More specifically, there were two different linear regions to the data (one at low [denaturant] and one at high [denaturant] values). It is also noteworthy that the slope (i.e., m value) of the linear region at high [denaturant] was significantly smaller than that of the linear region at low [denaturant]. Such an observed change in slope appears to be a useful sign in conventional SUPREX analyses that the protein under study is slow to reach a folding/unfolding equilibrium. In such cases, the modified protocol outlined here should be employed for SUPREX analyses. CONCLUSION The modified SUPREX protocol described here provides a specific time for the protein to reach the appropriate folding/ unfolding equilibrium in each denaturant-containing SUPREX buffer before the H/D exchange reaction is allowed to proceed. Our results on ubiquitin illustrate that such a separation of the equilibration and H/D exchange times can be important for the accurate determination of ∆G and m values by SUPREX. It is critically important to use the modified SUPREX protocol presented here in cases where a protein’s equilibration time is not short compared to the H/D exchange time. ACKNOWLEDGMENT This work was supported with funds from a PECASE Award (NSF-CHE-00-94224) to M.C.F.

Received for review July 14, 2004. Accepted October 18, 2004. AC048967Z

Analytical Chemistry, Vol. 77, No. 2, January 15, 2005

697