You are on page 1of 26

Annual Reviews www.annualreviews.

org/aronline
Ann. Rev. Phys. Chem. 1987. 38:463~88 Copyright 1987 by Annual Reviews Inc. All rights reserved

BIOCHEMICAL APPLICATIONS OF DIFFERENTIAL SCANNING CALORIMETRY


Julian M. Sturtevant Departments of Chemistry and of Molecular Biophysics and Biochemistry, Yale University, NewHaven, Connecticut 06511-8118 INTRODUCTION Differential scanning calorimetry (DSC)and the closely related differential thermal analysis (DTA)have been widely employedduring the last several decades in the thermodynamicstudy of processes that are initiated by either an increase or a decrease in temperature. This review focuses on biochemical applications of DSC. Macromolecular and polymolecular structures stabilized by the cooperation of numerous weak forces are important to most biochemical processes. Since such highly cooperative structures undergo conformational or phase transitions upon being heated, significant information concerning these structures can be obtained by DSC. Small molecules cannot be studied by DSC unless they form aggregates showing intermolecular cooperation, as in crystals. This is illustrated in Table 1. Since the enthalpies of chemical processes rarely are as large as 20 cal g-l, it is evident that molecules having molecular weights, or molecular aggregates having aggregate weights, in the thousands of daltons are required to give transitions sufficiently sharp for useful DSC observation. In a scanning calorimeter, one measuresthe specific heat of a system as a function of the temperature. For a solution, the apparent specific heat of the solute, c2, is given by the expression 1 cz = c~+--(c-cl)
W2

1.

wherec is the specific heat of the solution, c~ is that of the solvent, and wz 463 0066-426X/87/1101-0463 $02.00

Annual Reviews www.annualreviews.org/aronline 464


STURTEVANT Table 1 Transition widths for a two-state transition as observed by DSC for various values of the transition enthalpy. Temperature of half completion = 50C Transition enthalpy -t kcal mol 20 40 60 extent of conversion. Temperature/C ~ = 0.1 ~ = 0.9 29 39 43 75 62 58

is the weight fraction of the solute. Since the quantity c--ct is usually relatively small, for example, approximately -0.7% of c t for a 1%aqueous solution of a protein, it is essential to employa differential schemeof measurementin which c- c ~ is directly measured. This is accomplishedin a differential scanning calorimeter by using two closely matchedcells filled with equal weights or, more usually, with equal volumes of solution and solvent. When one considers that in general a significant, or even major, fraction of the total change in apparent specific enthalpy is due to the simple heating or cooling of the solvent, it becomes evident that the highest possible sensitivity and accuracy should be realized. The so-called excess apparent specific heat, cox, is the amountby which the apparent specific heat during a transition involving the solute exceeds the baseline specific heat. A recurring problem in DSC,as in manyother measurementtechniques, is the determination of the appropriate baseline since, as is evident, no direct observation of it is possible during the transition. Figure 1 showsa typical DSC curve observed for the reversible thermal denaturation of a globular protein. The apparent specific heat of the native form of the protein increases with increasing temperature while that of the denatured form is independent of temperature. The dashed curve, obtained by procedures outlined below, is the baseline specific heat, and Cex is as indicated. The integral of Cexover temperature then gives the specific calorimetric enthalpy, Ah=l, for the transition. In this review, we discuss very briefly the instrumentation for DSC and then consider the theoretical aspects of the various applications of DSC; these applications are illustrated with examplestaken from the literature. We do not attempt a comprehensive coverage of the rapidly growing literature in this field. INSTRUMENTATION In the opinion of the author, the instruments best suited for work with biochemical systems are the DASM-1M described by Privalov et al (1), its

Annual Reviews www.annualreviews.org/aronline


DIFFERENTIAL SCANNING CALORIMETRY

465

Observed ted ~mt-"~

~/~

0.1 col

-tg-t

15

25 55 TEMPERATURE /C

45

Fi#ure I A tracing of the DSCcurve observed with a solution of the Arg 96 ~ His mutant of the lysozyme of T4 phage (kindly supplied by Dr. John Schellman of the University of Oregon). Protein concentration 8,29 mgml-~, pH2.34, 0.02 Mpotassium phosphate buffer containing 0.025 MKC1.Nonoise was visible in the original recording. The calculated curve -t g-~ and cB ~ was obtained as outlined in the text with % = 0.0358+0.00192 t col K -~ 0.235 col K g-~. The calculated curve differs from the observed curve with a standard deviation of 1%of the maximalapparent specific heat.

successor the DASM-4 (2), and the Microcal 2. The DASM-4 is available from V/O Mashpriborintorg, MoscowG-200, USSR, and the Microcal 2 from Microcal, Inc., Amherst, Massachusetts, USA. The DASM-4 is shown schematically in Figure 2. Two cells composed of platinum or gold capillary tubing are suspended within two adiabatic shields with a 200-junction thermopile between them. The cells have an effective volumeof 0.5 ml and are filled through vertical extensions of the capillaries. They are heated by electric heaters in good thermal contact with them; the power to the heaters is adjusted by means of a control circuit, activated by the output of the thermopile betweenthe cells, which maintains them at closely equal temperatures. Thermopiles between the cells and the adiabatic shields activate control circuits, which hold the shield temperatures close to that of the cells. The instrument supplies signals that showthe cell temperature and the differential heating power to the cells; the signals maybe registered on an X-Yrecorder or maybe fed to a computer for subsequent analysis. Each cell is equipped with a

Annual Reviews www.annualreviews.org/aronline 466 STURTEVANT

Figure 2 Schematic representation of the cells and adiabatic shields of the DASM-4 differential scanning microcalorimeter. 1. Capillary inlets; 2. capillary cell; 3. inner shield; 4. thermopile between the cells; 5. outer shield. (By permission of Pergamon Press, Ltd.)

second electric heater that is used for calibration purposes. Scan rates of 0.125 to 2.0 K min- ~ are available; lower scan rates, whichare occasionally advisable, require modification of the heater circuits. The subject of DSCinstrumentation has been reviewed by Privalov (2). THEORETICAL CONSIDERATIONS

Interpretations of DSC data are usually based on the equilibrium thermodynamic expression

9 In K~ AH, H

where K is the equilibrium constant for the process under study, T is the absolute temperature, and AHvH is the apparent or vant Hoff enthalpy. (Although AHvH is a standard state quantity, the generally small variation of enthalpy with concentration makes it permissible in most cases to compareAHvr~ with the true or calorimetric enthalpy, AHcal. ) It is immediately evident that any equilibrium process observed during increase of the temperature is necessarily endothermic and that any exothermic process so observed must involve rate limitation. For a discussion of the appli-

Annual Reviews www.annualreviews.org/aronline


DIFFERENTIAL SCANNING CALORIMETRY

467

cation of DSC and DTA to rate-limited processes, the reader is referred to Borchardt &Daniels (3). In the following discussion, it is assumedthat no rate limitation of either chemical or instrumental character is present. That this assumption is valid in any particular case can be demonstrated by showing that the DSCobservations are independent of the scan rate employed. Reversibility will also be assumed, with consideration of the problemof irreversibility postponedto a later section. Two-State Processes

The vant Hoff equation, Eq. 2, is directly applicable only to two-state processes in which states intermediate between the initial and final states are not significantly populated at equilibrium. For a process of the simplest possible form A ~ B ; AHvn= MAho,~= AHoa~ 3.

the indicated equality must hold, with AH~n being given by the expression
Anvr~ = ART~/2 cexl/2 Ahca " I

Here Mis the molecular weight, Ahca~ is the calorimetric specific enthalpy, T~/2 = t~/2+273.15; t~/2 is the temperature (C) at which the process half completed, C~x,m is the excess specific heat at t~/2, R is the gas constant, and the factor A has the value of 4.00. (The expression AH 2R~/~T~/~Cox,~/~, where Cex,~/2 is the excess molar heat capacity at t~/2, has been widely employed in the calculation of AHvn.However, it is evident from Eq. 4 that in general, AHin this expression is not a single enthalpy but is actually the geometrical mean of AH~and AHva.) The finding that the equality in Eq. 3 holds within close limits for the thermal denaturation of many globular proteins is the best indication available that protein thermal denaturation can be of all-or-none or twostate character. The value of ll/2, or of tin, the temperature at which Cex reaches its maximalvalue c ...... for the two-state process in Eq. 3, should be independent of concentration. On the other hand, an increase of tm with increase in concentration is a clear indication of a decrease in the degree of oligomerization during the reaction, and conversely a decrease of with increase of concentration shows that the degree of oligomerization increases. If the reacting species is knownto be oligomeric at ordinary temperature, and tm is found to be independent of concentration, it may be concluded either that the reacting species has becomemonomericby

Annual Reviews www.annualreviews.org/aronline 468 STURTEVANT

the time the reaction temperature is reached or that no dissociation .or association accompaniesthe reaction. Weshall consider two cases involving change in degree of oligomerization. In the first of these, there is either complete dissociation or complete association involving an oligomer in one state and a monomer in the other: (a) Ao~nB or (b) nA~Bn. 5.

If we substitute the appropriate expression for K in terms of ~, the extent of conversion, in Eq. 2 and perform the differentiation, and then use the expression cex -- Ahcal(d~/dT), we find, since ~ = 0.5 at T,/2, that AHvH = 2(n + 1)RT~/2Cex. 1/2 Ahc~t 6.

so that the factor A= 2(n + 1). If in place of T,/2 we use Tin, the temperature at whichCexhas its maximal value, c ....... in Eq. 4, and c ...... in place of Cex,1/z, A has the values 4.00, 5.83, 7.47, and 9.01 for n --- l, 2, 3, and 4, respectively. The second case involves incomplete dissociation or association to a dimer in either the initial or the final state or both: A~B KN ,IT T~, KD 7.
K I

In curve fitting to this model, it seems reasonable to assume that KNand KDdo not vary significantly over the temperature range of the reaction controlled by K,. The equilibrium constant for the A to B step is given by 1 -- (1 + 8aK~D(A)0)/2K~ K, = 1 - (l + 80 - e)K~(A)0) ~/2" K~ 8.

where (A0) is the total concentration expressed in monomer units. Differentiation of In K, gives for the factor A whene = 1/2 the value 4K~(A) 4K~(A) /2 + 1 +4K~(A)0-(I 1/~" A = 1 +4KD2(A)0--(1 +4KD2(A)0) +4K~(A)o) 9. Obviously, no simple statement concerning the values of A can be made in this case. It was pointed out above that if association or dissociation accompanies

Annual Reviews www.annualreviews.org/aronline DIFFERENTIAL SCANNING CALORIMETRY 469 a process observed by DSC, the value of tm should vary with reactant concentration. For the case represented by Eq. 5a, whenct -- 1/2, In K~/2 = constant + (n- 1) In (A)0 = - RTI/~2 +constant. Thusa vant Hoff plot of In (A)0 versus 1/T~/2 will have a slope SA/-/vH

10.

A/-/vH (n- 1)R

11.

evaluated in this way for a two-state process should agree with the value given by Eq. 4. For the case represented by Eq. 7, we see from Eq. 8 that In KI, ~/z = constant + In [1 -- (1 + 4K~(A)o) ~/z] _ In [1 - (1 + 4K~(A)o)1/2] AHvn + constant. RTI,I/2 12.

Simulations in which T that L 1/2 is calculated as a function of In (A)0 show -1/2 and for this case, at least with Ks and KD in the range 0.01 to l0 #M (A)0 in the range 50 to 1000 #M, and AHvH independent of temperature, a plot of In (A)0 versus 1/Tl,~/2 is linear. Such simulations also showthe factor by whichthe slope of the vant Hoff plot must be multiplied to give AHvH. This factor can vary over a wide range if KNand KDare not very different. The evaluation of AHvH by the means outlined above gives a very important general application of DSC. For a strictly two-state process, carried out under essentially equilibrium conditions, AHvH = AHcal ----MAhcav If AHvH < AH~,I it can be concluded that one or more intermediate states are of significance in the overall process, whereasif AHvH > AHc,~intermolecular cooperation is clearly indicated. DSC is the only experimental technique that gives such positive indications concerning these characteristics of a process. Figure 3 showsplots of the excess apparent specific heat, taken arbitrarily as zero at 0C, versus temperature for two hypothetical reactions following Eq. 5a, one with n = 1 (curve A) and one with n -- 4 (curve The parametric values for each reaction are ti/z = 60C and Ahca I = 8 cal g-l, with a monomermolecular weight of 12,500. Curve A is nearly but not quite symmetric about tl/~ whereas curve B is markedly asymmetric. Curve C is the baseline for curve B and is calculated by changing the initial baseline (CA = 0+0.003 tcal -t g-~) t o t he f inal one (ca = 0.180 + 0.001 tcal K-~ g-~) in proportion to the integrated area under the curve as the temperature increases from t = 20C to t = t.

Annual Reviews www.annualreviews.org/aronline 470 STURTEVANT

1.00

0 25 Figure 3 The calculated hypothetical reactions weight baseline = 12,500 for curve daltons. B with variation in which Curve

45 TEMPERATURE,

65 C

85

of excess apparent specific heat with temperature for 4/2 = 60C, Ahc~ = 8 cal g-, and monomer molecular A, A ~ B; curve t, B, A4 ~ 4B. tcal CB = 0.180+0.001 Curve -I K C is g-~. the calculated

c a = 0-t-0.003

A process involving association instead of dissociation would have asymmetry opposite to that of curve B. Eq. 5a maybe extended to include the frequently interesting situation of ligand binding: AnLm ~ nB + rnL. 13.

Here we are assumingcomplete dissociation of ligand from the species B. If the total ligand concentration (L)0, is muchlarger than (A)0, logarithm of the equilibrium constant in Eq. 13 at half completionmaybe written In K1/2 = constant+ (n-- 1) In (A)0 + rn In (L)0

ZXHv.
RT1/2

+ constant. 14.

Annual Reviews www.annualreviews.org/aronline DIFFERENTIAL SCANNING CALORIMETRY 471 Thus a plot at constant (A)0 of In (L)0 versus 1/T1/2 should be a straight line with slope S given by S-

Any.
mR

15.

Eqs. 6 (or 9), 10 (or 12), and 15 provide three meansfor estimating AHvH. If the value obtained by Eq. 15 differs from the other two, this maybe due to incomplete dissociation of the ligand, or even to greater binding of the ligand to species B than to A. It may be noted that the dissociation of a tightly boundligand in the absence of any excess ligand in solution will lead to asymmetry of the DSC curve in just the same way as does a dissociation involving the ligand binding species. Two-State Processes with Permanent Changes in Specific Heat The preceding treatment must be modified if the process under study is accompaniedby a large permanent change in the apparent specific heat. Anexampleof this situation is shownin Figure 1, which presents a tracing of the DSCcurve observed for the thermal denaturation of T4 lysozyme at pH 2.34 (S. Kitamura and J. M. Sturtevant, unpublished observations). -~ at the start of the transition at 15C In this case, AHca1 = 31 kcal mol and 108 kcal mol-~at the end of the transition at 45C, so that significant errors might arise if the data are analyzed on the basis of a temperatureindependent enthalpy. In such cases, the best way to evaluate tl/2, Ahcal, and AHvH is by fitting the experimental data to the appropriate theoretical curve according to the least squares criterion. Weoutline here the algebra involved for a case such as that shown in Figure 3 where the specific heats observed before and after the transition are themselves temperature dependent. Consider again the process represented by Equation 13 with (L)0 assumed to be much larger than (A)0, and with the initial and final specific heats given by the expressions c A = A+Bt, ca -- C+Dt, t -- C. 16.

The enthalpy of denaturation as a function of temperature is then 2 Ahcal = Aho+(C--A)t+ 1/2(D--B)t where,with Ahl/2
=

17.

Ahcalat

tl/2,

Annual Reviews www.annualreviews.org/aronline 472 STURTEVANT 18.

Ah0= Ah1/2- (C- A) t ~/2- 1/2(D- B) t ~12. If we set -A = Aho-273.15(C-A)+ B = (C-A)-273.15(D-B); 1/2(273.15)2(D-B), C 1/2 (D-B),

19.

then the integrated form of Eq. 2 can be written RK" ~1n~/2~/2 = A (1T J ~+Bln 1]/2] T T~/2 -I-C(T-- T,/2) 20.

where K= K/(L)0, K~/2 is the value of K at 4/2, and the ratio fl = AHvH/Ahcaj is assumed to be independent of temperature. The extent of conversion is given by a n= ~-~ y(1-~); Y=2 1 K 21.

and the excess specific heat, as before, by cCx = Ahcat (d~/dT). The calculated baseline for a two-state reaction is Car = (1 - ~) (A + Bt) + ~(C + so that the total excess specific heat is
Ctot = Cex "[- Cav. 23.

22.

The adjustable parameters used in minimizing the standard deviation, [E(ctot- Cob~)2/(--1/2 whe re Cob ~ arethe ~ observed values, may be t aken in the formt 1/2, Ah1/2, and ft. For a simple two-state reaction, fl is expected to be equal to the molecular weight. K/K~/2, 7, and ~ are calculated for temperatures corresponding to the observed data points (~ is obtained by successive approximationsfor n > 3) and CCx,C~v, and tot are evaluated. It can be shown by simulation that in cases of modest permanent increases in specific heat, adequate accuracy can be obtained by drawing a baseline calculated according to Eq. 22, ~ is determined by integration on the assumption of temperature-independent enthalpy and then by using data points relative to this baseline in the equations given earlier, which assume no permanent change in specific heat.

Non-Two-State Processes (AHo~ < An~at)


At first sight, it would seem obvious that a large protein molecule with a very complex three-dimensional structure would unfold gradually when heated, with manystates populated betweenthe initial and final states. It was therefore surprising whenit becameevident, largely as a result of the

Annual Reviews www.annualreviews.org/aronline


DIFFERENTIAL SCANNING CALORIMETRY 473

DSC demonstration, in some cases, of equality of AHca~ and A//vH, that at least some proteins unfold in a very nearly all-or-none manner. It is, however, not surprising that in manycases the unfolding is more complex than two-state. In such cases, as noted above AHvH < AH~a~with AHvH calculated according to Eq. 4 with .4 -- 4. Twodifferent models for processes involving multi-state unfolding are considered. The first assumes the existence of several domains in the molecule, each of which unfolds totally independently in a two-state manner. It is, of course, conceivable, and perhaps probable, that there may be interactions between domains, but there is no general method by which such interactions can be included in our treatment. ~ If we include the possibility of dissociation at any or all steps, the ith step is of the form
Ai,mi ~- miB i

24.

and the parametersto be adjusted for this step are mi, t~/2,i, AhCaLi, and It is expected that in most cases, fl,- will be equal to the molecularweight of the entire molecule. A serious problem arises in cases of multi-state transitions for which there is a large over-all permanent change in specific heat, namelythat the experimental observations give direct information concerning the specific heats of the initial and final states, but none concerningthe specific heats of the intermediate states. Wehave adopted the purely arbitrary procedure of partitioning the over-all changein specific heat betweenthe steps in the process at each temperature in proportion to their individual enthalpies at that temperature. It is then simply a matter of applying the algebra developed above to each componenttransition, and of summing the individual ctot.t to evaluate the standard deviation of the data from the theoretical equation for the assumedmodel. The second non-two-state model to be considered is that of strictly sequential two-state steps. For five sequential steps, the over-all process maybe represented as P~ ~
P2 ~

P3 ~ P4 ~ P~ ~ P6

25.

where for each step we have Ah~, K~, t~/2,t, and AHvr~,; = fl~Ah~. For a pure substance fl~ = f12 ..... molecular weight unless the process is
~ In the ease of a protein containing two domains, any interaction between the domains would presumably be manifested by changes in t~/2 and/or Ah~for a fraction of either or both domains whenthat fraction of the other domain is denatured. Wehave found that the simulated DSC curves for such a case with changes in either or both t~/:s of + 10C (original t~/~s = 50 and 53) and either or both Ahabs of _+0.5 cal g 1 (both original A~s 2.5 cal g-l) can still be resolved into two two-state components on the basis of the independent model with practically as good accuracy as for the unperturbed case, but of course with different derived parameters.

Annual Reviews www.annualreviews.org/aronline 474 STURTEVANT

complicated by dissociation or association at one or more steps. The following treatment does not include that situation. Using the equations (P,+1) K,. = (p,~-; (P)0 = (P,)+"" 26.

it follows that, with D = I+K~+K1K2+ "" (P1)=_I. (P2)_K~. +K1...K~ (P3) K1K~. 27. 28.

(P)0 O (e)0 O (P)0


Integration of the vant Hoff equation for each step gives In Ki - R

since Ki = 1 at T1/~.i. Eqs. 2%29 are employedto calculate (P,)/(P)o T+3Tand T-6T, where fit is a small temperature interval. The change in (P3/(P)o in the temperature interval 26Tis then A[(P3/(P)o] = (p)~ (at

(P~)

T+~T) -(Pi) (at (~0

T--3T).

30.

This change will lead to a heat absorption due to the ith step of Aq, = A [(P,+ t)/(P)o]" [Ahl + ah2 +.-" + 31. since to arrive at species (i+ 1), Eq. 25 showsthat heats Ahl, Ah2.... , Ahg have to be absorbed. The excess specific heat at T is then
Cex = (X Aq3/23T.

32.

As before, if Ac o is large, it seems that the best that can be done is to assume that the fraction of the total specific heat change at each temperature due to step i is proportional to Ah]XAhi, and to use the resulting Ac~to correct the Ahi.

Numerical Treatment of DSCData


The simplest DSC curves readily yield enthalpies by the evaluation of the areas under the curves, for example by means of a planimeter, and vant Hoff enthalpies by means of Eq. 4. It was pointed out above that in cases showing large permanent changes in heat capacity, it is best to obtain calorimetric and vant Hoff enthalpies by fitting the experimental data to a theoretical curve. In more complex cases, some form of curve fitting seems to be unavoidable. Privalov and his colleagues (4, 5) showed that the complex DSCcurves

Annual Reviews www.annualreviews.org/aronline


DIFFERENTIAL SCANNING CALORIMETRY 475

observed for the melting of transfer RNAs could be accurately represented as the sums of several two-state curves, which were attributed to the independent melting of various substructures within the molecules. In an important series of papers published in 1978, Freire & Biltonen (6-8) pointed out that the partition function for a macromolecularsystem can be calculated by a double integration of the apparent heat capacity as measured by DSC. A recursive method was developed to evaluate the thermodynamicparameters of each step in a multistate transition, and such quantities as the fractional occupancyof each state were calculated as functions of the temperature. This analysis was applied in general form to cooperative phenomena (7) including the helix-coil transitions of nucleic acids (8) and phase transitions of phospholipid bilayers (9). The Freire-Biltonen procedure is in principle independent of any assumedmodel, although their treatment for the helix-coil transition of nucleic acids assumedthat the transition occurs betweena double-stranded and a single-stranded form. Filimonov et al (10) suggested ways to improve the procedure developed by Freire & Biltonen (6). In particular, they demonstrated that incorporation of feedback into the iterative calculation reduced cumulative inaccuracies arising from such sources as noise in the DSC data. Gill et al (11) have recently developeda generalized binding formulation based on the postulation of multiple allosteric forms of macromolecules. In applying this theory to the melting of a transfer RNA,they showed that with the assumption of six allosteric forms they could obtain a representation of the experimental data that was as accurate as one based on six independently melting domains. The latter model implies 25= 32 energy levels for the molecule while the former model has only 6 energy levels corresponding to the 6 allosteric states. For analyzing a wide range of DSCdata we have employed a procedure that is based on either of the modelsoutlined earlier involving independent or sequential two-state events, with or without association or dissociation. The parameters defining each step are adjusted to minimize the standard deviation of points on the calculated sum curve for Ctot from observed points. In general, as mentionedearlier, three parameters are required for the definition of a two-state curve, which maybe taken, as tl/2, Ahead, and AHvH. If Acd 5~ 0, Ahcal, and AHvH are the values at tl/2. In simple cases, where there is no intermolecular cooperation, no dissociation, and no association, the ratio AHvH/Ahcal should be taken equal to the molecular weight of the substance, thus reducing the numberof adjustable parameters to two per step. Even so, the numberof parameters to be determined by fitting to experimental data of limited accuracy is frequently excessive, so that unique solutions maynot be obtainable. Furthermore, it is frequently

Annual Reviews www.annualreviews.org/aronline 476 STURTEVANT

difficult or impossible to distinguish betweendifferent modelson the basis of the goodnessof the fit finally obtained. This is true, for example,with the independent and sequential models with transitions close together in temperature. The minimization of the standard deviation can be accomplished either by a "brute force" adjustment of each parameter in succession [Edge et al (12)] or by a nonlinear least squares approach such as supplied by the Marquehardt or.Simplex algorithms (B. G. Barisas, private communication). In the brute force procedure, the numberof successive adjustments of each parameter should be limited to four or five in each cycle of the calculation, and the increments applied should at first be rather large, in order to minimizethe danger of arriving at a local rather than the global minimum. If shortening the computer time required for the calculation is not of much importance, the brute force method has some advantage in being more readily followed as the computation proceeds. Changet al (13, 14) have developed a similar approach to the analysis of DSCcurves into the sum of simple two-state curves. Their procedure as described does not accommodate a significant permanent change in heat capacity nor the possibility of self association or dissociation. Non-Two-State Processes (AHvH > AHc,t) As noted earlier, when AHvn as calculated by Eq. 4 with a suitable value of A exceeds AHca it may be concluded that the process under study 1 involves intermolecular cooperation. Phase transitions offer the most commonexampleof this situation; since the melting of a perfectly pure crystalline substance is an isothermal process, AHvH approaches infinity. The so-called gel-to-liquid-crystal phase transition of a multilamellar bilayer suspension of carefully purified dipalmitoylphosphatidylcholine in water was reported by Albon & Sturtevant (15) to have a ratio of AHvn/AHcal equal to 1400 whenscanned at 0.023 K min-1. It is probable that the ratio wouldhave been still higher at a lower scan rate because of further relief of instrumental lags. In systems of this sort, the ratio AHvH/AHcaI can be taken as a lower limit for the size in monomer molecules of the average "cooperative unit" for the process. Similarly, in the helix-coil transition of a polynucleotide of high degree of polymerization, although the reaction is far from all-or-nothing in character so far as the entire molecule is concerned, the ratio AHvH/AHcal, where AH.I is the enthalpy per mole of base pairs, gives a measure of the number of base pairs in the average cooperative unit. Irreversible Processes The foregoing treatment is strictly valid only for reversible processes that are subject solely to thermodynamiclimitation during their observation

Annual Reviews www.annualreviews.org/aronline DIFFERENTIAL SCANNING CALORIMETRY 477

by DSC.Unfortunately, many, if not most, processes of interest in biochemistry and otherwise suitable for DSC study are, operationally, at least partially irreversible as judged by the DSC criterion of the repeatability of the DSCscan on rescanning the sample. The thermal denaturations of many proteins are found on the basis of this criterion to be irreversible, even in cases where denaturation at lower temperatures by treatment with guanidinium chloride or urea is found to be reversed on dialyzing out the denaturing agent. The thermal helix-coil transitions of naturally occurring double helical polynucleotides are usually incompletely reversible in DSC experiments, primarily because of the entropic difficulty of achieving exact realignment of the polynucleotide chains after chain separation. Possible justification for the application of reversible thermodynamics to apparently irreversible processes has been briefly discussed by Privalov [Ref. (16), pp. 28-29]. Wehave found that reasonable vant Hoff plots based, for example, on Eq. 10 are observed even in cases of apparently irreversible protein denaturations. Thus, t mfor the denaturation of the regulatory subunits of aspartyl transcarbamoylase in the presence of ATP increases with increasing ATPconcentration in the manner expected on the basis of Eq. 14 (12). In other words, the protein during denaturation responds to the concentration of free ATPin the solution, even at concentrations of ATPwhere the protein is effectively saturated with the ligand. Similarly, tm for the denaturation of the tetrameric core protein of the lac repressor of E. coli increases with increasing protein concentration in accordance with Eq. 14 (17). Furthermore, the highly asymmetric DSC curve observed with this protein can be accurately fitted to Eq. 5a with n --- 4 and values for tl/2, Ahead,and AHvH agreeing well with the observed values. These and other empirical results provide somemeasureof validity to the application of equilibrium thermodynamics to apparently irreversible processes. Simulations show that results similar to those outlined above can be obtained for a modelwhere an equilibrium dissociative process is followed by a rate-limited irreversible step, with selection of a rate constant, k 1, and an activation energy, Ea, which lead to as muchas 75%conversion to the irreversible species whenthe A to B conversion is 95%complete. APPLICATIONS The Thermal Denaturation of Proteins Perhaps the most important applications of DSCin biochemistry are concerned with the thermal unfolding of proteins, since the calorimetric analysis can give information concerning the fundamental nature of this process and the forces involved in the stabilization of the native structures

Annual Reviews www.annualreviews.org/aronline 478 STURTEVANT

of proteins. A very recent and complete listing of changes in free energy, enthalpy, and heat capacity accompanyingthe unfolding of proteins has been published by Pfeil (18). This listing includes the numerousvalues obtained by methods other than DSC. Wefirst consider relatively simple two-state denaturations and then more complex multistate denaturations. TWO-STATE DENATURATIONS A comprehensive review of the literature (up to 1979) on the thermal denaturations of small globular proteins, most of which are of very nearly all-or-none or two-state character, has been published by Privalov (19). In this section, we consider a few more recent examplesof two-state denaturations that illustrate important possibilities of the DSCmethod. Two-state denaturations with self-dissociation or association The first protein shownby DSCto undergo dissociative unfolding was Streptomyces subtilisin inhibitor (SSI). This protein was known to be dimeric at ordinary temperatures. Its denaturational DSCcurves were found to be slightly asymmetric, and the values of t mincreased with increasing protein concentration. Quantitative analysis of the DSC curves showed clearly that the denaturation follows the schemeA2 ~- 2B (20). Thus the native protein remains dimeric up to temperatures in the vicinity of 80C. The so-called core protein obtained by limited proteolysis from the lac repressor of E. eoli is tetrameric at ordinary temperatures. It is interesting that the DSC data for the thermal denaturation of this protein, which is an operationally irreversible process, yield three different estimates for AHvH that are in reasonable agreement: (a) Curve fitting of the data to the model A4 ~-4B as outlined in an earlier section gave AHvr~= 525+ 30 kcal mol-~; (b) calculation according to Eq. 4 with A = 10 gave AHvH = 585 + 31 kcal mol 1; (c) the slope of a vant Hoff plot of In (L)0 (L0 = total protein concentration) versus I/T~/2, with n = 4 in Eq. 11, gave AHvH = 498 _ 24. These can be compared with AH~al = 594 + -~. This agreement, modest though it is, gives further support 32 kcal mol to the unorthodox procedure of applying equilibrium thermodynamics to irreversible processes. Two-state denaturations with ligand dissociation The first application of DSCto protein/ligand association reactions involved the binding of Larabinose and D-galactose to the arabinose binding protein (ABP) of coli (21). The thermal denaturation of A BPis reversible both in the absence and presence of the ligands. The ratio AH~H/AHca ~ is 1.26 in the absence of ligands, indicating somedegree of association, and 0.91 in the presence of ligands, indicating approximately two-state behavior. It was found that

Annual Reviews www.annualreviews.org/aronline DIFFERENTIAL SCANNING CALORIMETRY 479 the tm for the denaturation increased with increasing ligand concentration, and that a vant Hoff plot of In (L0) versus 1/Tm was a slightly curved line (ACd> 0) the slope of which gave (taking rn = 1 in Eq. 15) AHvn = kcal mo1-1 at 58C and 161 at 65C kcal mol-l for both ligands. These values are considerably smaller than the values (183 kcal tool-~ for arabinose and 174 kcal mol-l for galactose) calculated from the DSCcurves by means of Eq. 4 (A = 4), perhaps because of some nonspecific binding by the native protein or more residual binding by the unfolded protein. It is interesting that the data involvedinthis vant Hoff plot, with tm increasing from 58.4C to 65.1C, are for protein ranging from 97.7 to 99.98% saturated by ligands, and that still higher concentrations of ligands would lead to further increases in t m. It is thus quite clear that this continuing increase in tm is not due to ligand-induced structural changes in the proteins. The value for AHcal in the presenceof glucose at 59Cis 200.7___1.8 kcal tool- ~ while that in the absenceof glucose at 59Cis 169.2 _ 1.1 kcal mol-1. The difference betweenthese figures, 31.5 kcal mol-1, agrees very well with the value 30.1 kcal mol-1, calculated from the equation AHdisso~= 15.26_+0.47+(0.436+_0.047) (t-25 ) kcal m01-1 33.

which expresses the results of isothermal calorimetric measurementson this system (21). Wethus see that the excess enthalpy of denaturation the presence of the ligands is simply the enthalpy of the dissociation of the ligand that accompanies the denaturation. The binding ofisopropyl fl-D-thiogalactoside (IPTG) to the core protein of lac repressor has effects on the apparently irreversible denaturation of the protein similar to those of arabinose on the denaturation of ABP as indicated by the fact that a plot of In (L)0 versus 1/Tmis a straight line the slope of which gives a value for AHvH in reasonable agreement with the AHo,Ifor the denaturation plus ligand dissociation. It thus appears that the protein during denaturation is influenced by the concentration of unboundligand free in solution, which could hardly be the case if the denaturation were an absolutely irreversible process. Two-state denaturations with large permanent specific heat changes The denaturation of a mutant of T4 lysozyme shown in Figure 2 is a good sample of a denaturation with a large change in specific heat (S. Kitamura and J. M. Sturtevant, unpublished observations). The thermodynamic properties of this transition were evaluated by curve fitting to a two-state model including a temperature-dependent ACd. The denaturation of the iso- 1 form of yeast cytochromec in which the single thiol group has been blocked by treatment with methyl methane

Annual Reviews www.annualreviews.org/aronline


480 STURTEVANT

thiol sulfonate has been found to have both a large change in specific heat and an increase in dimerization accompanying denaturation as shown by a decrease in tl/2 upon increase of protein concentration (B. T. Nail and J. M. Sturtevant, unpublished results). Data for this denaturation over concentration range of 63 to 630/~M were well fitted by the model of Eq. 7, with no significant dimerization of the native form of the enzyme, but with the extent of dimerization of the denatured protein ranging from 0.79 to 0.93 over the concentration range covered. MULTI-STATE DENATURATIONS Privalov and his colleagues have applied their curve resolution methodto the DSC data obtained for a large number of complex protein denaturations, including those of pepsin and pepsinogen, various calcium-binding proteins, plasminogen, immunoglobulins, histones H1 and H5, fibrinogen, tropomyosin, paramyosin, and others. This work, together with reports by others in this field up to 1982, has been reviewed in detail by Privalov (16). Someof these proteins, for exampleplasminogen(22), have been the subjects of detailed reports since the publication of Privalovs review. The thermal denaturation of taka-amylase A gives a DSCcurve with a single asymmetric peak, and with AHvH/AHcal about 0.17, indicating a multi-state transition (H. Fukada, K. Takahashi, J. M. Sturtevant, submitted for publication). This protein contains a single tightly bound2+ Ca , and in the absence of any added Ca2+ its denaturation curves can be accurately resolved into the sumof three, independent two-state transitions 2+ during the last step. As expected, including the dissociation of the Ca the dissociation of a tightly bound ligand in the absence of added excess ligand has the same effect on the DSC curve as wouldthe self-dissociation 2+, the temperature of denaturation of a dimer. In the presence of added Ca 2+ increases with increasing Ca concentration, further confirming that the 2 dissociates during the unfolding. This indication of the tightly boundCa existence of three domainsin the molecule is consistent with the structure of the molecule as revealed by X-ray crystallography, which shows a cleft dividing the molecule into a smaller and a larger part. The suggestion based on the DSC result is that the larger part is composed of two domains and the smaller part of one. Aspartyl transcarbamoylase is a complex protein composed of six socalled catalytic polypeptide chains and six regulatory chains per molecule of 310,000 daltons. The c6r6 molecule can be separated into two catalytic subunits, c3, and three regulatory subunits, r2. DSCstudy (12) shows that the denaturational curve of c3 can be resolved into three two-state components, that of r2 into two components, and that of c6r6, composed of two separate peaks, into five two-state components. The values of tm

Annual Reviews www.annualreviews.org/aronline


DIFFERENTIAL SCANNING CALORIMETRY

481

for the denaturations of c3 and r2 are independent of concentration, indicating that the polypeptide chains do not separate on denaturation, while those for the denaturation of c6r6 increase with increasing concentration, showingthat dissociation, presumablyinto ca and r2 subunits, accompanies the denaturation. The effects of the ligands ATP,CTP,and the bisubstrate analog N-(phosphonoacetyl)-L-aspartate (PALA)on the DSCcurves of and r2 are consistent with the view that the former two bind to r2 and are dissociated on denaturation, while PALA binds more strongly to denatured r2 than to the native protein; PALA binds to c3 and is dissociated on denaturation, whereas the two nucleotides are more strongly bound to denatured than to native c3, As might be expected, the effects of these three ligands on the DSC curve for err6 are quite complicated, but qualitatively consistent with their various effects on the curves for c3 and r2. With the development of techniques for accomplishing single amino acid replacements in proteins, and for achieving high yield biosynthesis of these altered proteins, much interest attaches to determination of the thermodynamiceffects of such "synthetic mutations." Perhaps at present the protein most carefully studied from this point of view is the Arg 96 --, His mutant of the lysozymeof T4 bacteriophage (23). DSC measurementsin the pH range 2.2 to 2.84 (S. Kitamura and J. M. Sturtevant, unpublished observations) showed the unfolding of this protein to be reversible and gave the results summarized as follows:
DENATURATIONOF MUTANTFORMS OF PROTEINS

Wild type: AHca~= 5o97+2.33t kcal mo1-1 (standard deviation __+4,20) tl/2 = 2.11 + 17.29t C Arg 96--, His: (standard deviation ___ 0.58) -~- -t -8.58+2.66t kcal mol (standard deviation _+ 4.48) tl/2 = - 19.84+21.31t C (standard deviation _ 0.51)

34.

maca 1

35.

Fromthese data, since the free energy of unfolding, AGo,equals zero at t 1/2, one can calculate by meansof the Gibbs-Helmholtz equation the AGotemperature profiles at various pH values within the experimental range. Taking the free energy of stabilization as the value of AGu for the mutant form less that for the wild type at fixed pHand temperature, values of the -l, i.e. an apparent destabilization, are obtained, order of --3.5 kcal mol

Annual Reviews www.annualreviews.org/aronline 482 STURTEVANT

a quantity surprisingly large for a replacement of one charged group by another. Of course, as mentionedearlier, it cannot be decided on the basis of the DSC data alone whether the mutation caused destabilization of the active form or stabilization of the unfolded form, although the former certainly seems more likely. In any case for which ACd#- 0 for either for both wild type or mutant forms, the free energy of stabilization may undergo a change of sign of dGu/dT within the temperature range of interest. Unless stated otherwise, free energies of stabilization quoted belowwill be the values at the tl/2 of the wild type protein. An early study of the effects of single aminoacid replacements involved the replacement of Gly 211 of the a-subunit of tryptophan synthase by either arginine or glutamic acid (24). The arginine replacement had barely perceptible effect on tn, but caused an increase in enthalpy of 17 kcal mol- ~, while the glutamic acid substitution increased t~ by 1.8C and enthalpy by 10 kcal mol-l (enthalpies calculated to the tm for the wild type using the observed values for ACd). Thus very small effects on the free energy were accompaniedby relatively large effects on the enthalpy. A more elaborate DSCstudy of glutamine and serine replacements of Glu 49 in this protein (25) showed decreases in t,, of 3.0 and 1.9C respectively at pH7.0, and increases of 8.1 and 4.3C respectively at pH 9.3. In terms of free energy changes at the tm of the wild type, 54.3, these changes correspond to apparent stabilizations of 2.8 and 1.4 kcal mol-1 respectively. Within experimental uncertainty, all three proteins have the same enthalpy of denaturation, varying from 81 kcal tool -I at 45C to -~ . 124 kcal mol at 60 The denaturational DSCcurve for the 2 repressor of E. coli shows two clearly separated peaks: The one at lower temperature is due to denaturation of the N-terminal portion of the molecule and the other at higher temperature to the denaturation of the C-terminal portion (26). Fourteen single amino acid replacements and one double replacement, all in the N-terminal portion of the molecule, have been subjected to DSC study (27 29). Within experimental uncertainty, none of these replacements affected the peak that results from the denaturation of the Cterminal portion, which indicated that the two domains do indeed unfold approximately independently. A maximalapparent stabilization of 1.3 kcal mol- l (Gin 33 -~ Tyr) and destabilization of 2.8 kcal mol- ~ (Ala 66 -~ Thr) were observed. In the five cases of replacements of glycine (a poor helix former) in an a-helix by another residue, stabilization was observed. In most cases, the enthalpy change (maximal decrease 22 kcal tool -~) was larger in magnitude than the free energy change, indicating enthalpyentropy compensation. Seven mutant forms of the tail spike protein of phage P22 have been

Annual Reviews www.annualreviews.org/aronline


DIFFERENTIAL SCANNING CALORIMETRY

483

studied by DSC (M.-H. Yu, J. King, J. M. Sturtevant, unpublished observations). In this case, a maximaldestabilization of + 17 kcal mol- ~ (Arg 283 -~ Ser) was determined, the largest destabilization so far observed for a single amino acid replacement. An interesting point is that all these proteins were elaborated at 30C by temperature-sensitive mutants that are unable to form mature protein at 40C. All the proteins have values for tm between 83 and 88C. Thus the temperature sensitivity is not a simple matter of thermodynamicsand must involve some kinetic effects. Preliminary DSCwork with eleven mutant forms of staphylococcal nuclease has shownsmall destabilizations (maximalvalue 0.4 kcal tool-1) -~) (J. Gerlt, D. Shortle, and stabilizations (maximal value 1.4 kcal mol and J. Sturtevant, unpublished observations). The Asp 27-o Asn and Asp 27-~ Ser mutant forms of dihydrofolate reductase have been prepared and studied (E. Howell, J. Kraut, and J. Sturtevant, unpublished observations). In each case, tm is raised (3.8 and 5.2C respectively) and there is thus apparent stabilization (1.3 and 1.0 kcal mol-~respectively). It is interesting that in the former case the denaturational enthalpy is increased by 15 kcal tool-~ and in the latter -~. case it is decreased by 7 kcal mol Conformational Transitions of Nucleotides The helix-coil transitions of oligo- and polynucleotides have been extensively studied by means of DSC.As a rough generalization, the enthalpy increase accompanying this transition maybe taken to be 8-10 kcal (mole of base pairs) ~. If allowance is madefor "fraying" at each end of a double helix of nucleotides, it appears that a transition of useful amplitude and sharpness requires an oligonucleotide containing at least 6 base pairs. The early literature in this field was summarizedby Privalov in 1974 (30). A recent systematic DSC study of 19 deoxy oligonucleotides and deoxy polynucleotides by Breslauer and his colleagues has enabled them to compile a consistent set of nearest neighbor base-stacking enthalpies, which has been shownto be successful in predicting helix-coil enthalpies in DNAsof known sequence. This work has recently been summarized (31). The first DSCstudy of the unfolding of a tRNAwas made by Bode et al (32). Privalov et al have employedDSC to study the melting of several tRNAs. It was in the course of this work that the first resolutions of complex DSCcurves into two-state component curves were introduced. These and other research are well summarized by Privalov & Filimonov (5). DSChas been applied to the study of DNA-ligandinteractions in much the same way as to protein-ligand interactions. Particularly important

Annual Reviews www.annualreviews.org/aronline 484 STURTEVANT

has been its application to the study of binding of antitumor drugs to DNA(33). In recent years there has been muchinterest in the left-handed double helical form of DNAknown as Z DNA. The thermodynamics of the transition from the more usual right-handed helical B form to the Z form has recently been determined by DSCboth for poly (dG-mSdC) (34) and for poly (dGdC)-poly (dGdC) (J. B. Chaires and J. M. Sturtevant, unpublished observations). The B-helix-to-coil and Z-helix-to-coil transitions of these polynucleotides, which occur at temperatures as high as 125C depending on experimental conditions, were also observed. This work illustrates an important feature of the DASM-4 calorimeter, namely that an excess pressure of N2of approximately 2 atm is applied above the liquids in the cells, thus permitting scanning of aqueous solutions up to 130C. Phase Transitions of Phospholipids and Phospholipid Mixtures Phospholipid bilayers have been extensively studied during the last two decades both because they serve as simple models for complex biological membranes (35) and because they are intrinsically interesting as quasi-two dimensional systems. Muchhas been learned about model and biological membranes from studies of their thermotropic properties, and DSCis in most cases the method of choice for such studies, especially when thermodynamicdata are of importance. McElhaneyin 1982 (36) published an extensive review of this application of DSC. Here we shall briefly consider certain general points and some recent developments. EXPERIMENTAL CONSIDERATIONS Phospholipids in bilayer suspensions in aqueous media undergo several different phase transitions. Of these, the so-called main, or gel-to-liquid crystal, transition of phosphatidylcholines (PCs) has been shownto be a first order phase transition (15). In principle this meansthat for a highly purified PC the transition should be close to isothermal, and this in turn meansthat very low scan rates, of 0.1 K min-1 or less, should be employed to minimize instrumental broadening of the transition. Since the size of a DSC signal decreases with a decrease in the time rate of heat flow into the sample, low scan rates require high instrumental sensitivity. Phospholipid phase transitions have in some eases been shown to produce two or more closely spaced DSC peaks (37, 38), sometimes for unknown causes. The existence of such transitions is another reason for using low scan rates, at least with single component lipid systems. It has been found that there are kinetic limitations in the formation of

Annual Reviews www.annualreviews.org/aronline


DIFFERENTIAL SCANNING CALORIMETRY 485

certain phospholipid phases (39, 40). These phenomenaseem to occur more often in the formation of phases during lowering rather than raising of the temperature, but in any case they emphasize the need for more careful experimentation than was formerly thought to be necessary. EFFECTS OF ADDED SOLUTES It has frequently been observed that the addition of any one of a wide variety of substances to a phospholipid bilayer lowers the phase transition temperature and broadens the transition. The cooperativity of a transition of a pure lipid is manifested in its sharpness and maybe expressed in terms of the size in lipid molecules of the cooperative unit given by the ratio AHvH/AHca I where AH~is the enthalpy of the transition in cal per mole of lipid and AHvn is estimated from the transition curve as outlined in an earlier section. Althoughit has usually been assumedthat solute broadeningof a transition curve indicates a decrease in cooperativity, it has been shownrecently, on the basis of ideal solution theory, that such broadeningdoes not necessarily reflect loss of cooperativity and, conversely, that there maybe a loss of cooperativity in cases where the transition breadth is not decreased or is even increased by addition of a foreign substance to the lipid (41, 42). Suchconsiderations are considerably complicated if an added solute causes a change from the gel phase to the recently discovered interdigitated phase (43). Recent work has demonstrated the value of fitting DSCdata obtained at modest added solute concentrations to theoretical curves based on ideal solution theory, including independently determined aqueous phase-lipid phasedistribution coefficients (44) in the case of solutes showing significant water solubility (45). an illustration of the application of DSC to the study of the thermotropic behavior of lipids, which are somewhat more complicated than PCs, we may cite the recent work of Maggioet al (46-48). Twenty chemically related glycosphingolipids were studied, and also mixtures of certain of these with dipalmitoylphosphatidylcholine and/or myelin basic protein.
PHASE TRANSITION PROPERTIES OF COMPLEXLIPIDS As

Conformational Transitions of Polysaccharides Although DSC has not as yet been extensively applied to the thermallyinduced transitions that occur in certain polysaccharides, it has recently been shownto be potentially as useful in this field as in those involving other biopolymers such as proteins and nucleic acids. One of the most extensively investigated processes involving polysaccharides is the double helix-coil transition of iota- and kappa-carrageenan (49, 50). The observed enthalpy for these systems can be interpreted in terms of the Manning theory (5 l, 52), since these polysaccharides,

Annual Reviews www.annualreviews.org/aronline 486 STURTEVANT

which carry charged sulfonate groups, can be regarded as linear polyelectrolytes. Current examples of this are the studies by Paoletti et al (53) and Kitamura et al (54) of the order-disorder transition of xanthan polyelectrolyte, which is composed of a cellulosic backbonewith trisaccharide side chains carrying carboxyl groups, some of which are on pyruvate residues condensed at the carbonyl group. Significant differences in the transition properties of native and partially depyruvated xanthans were observed, which according to the Manningtheory suggest that the native form has a mixed structure with single and double helical regions while the depyruvated form is entirely double helical. Itou et al (55) reported that the triple helical polysaccharide schizophillan undergoes a sharp transition at about 6C. Recently this polysaccharide has been studied in water-dimethyl sulfoxide mixtures (S. Kitamura, T. Kuge, J. M. Sturtevant, unpublished results), and it has been found that two transitions are observed over most of the range of solvent compositions. The transition at higher temperature may be due to a triple helix-single coil change in conformation. In DSC studies ofpolysaccl~arides, as in those involving polynucleotides, polydispersity must be taken into account. The ratio ofvant Hoffenthalpy to calorimetric enthalpy maybe employed, as outlined earlier, to obtain an estimate of the size of the cooperative unit for the process under study.

APPLICATIONS

TO COMPLEX

SYSTEMS

The first attempt to apply DSCto a complex system was probably the work of Steim et al (56) on the plasma membrane of Aeholeplasmalaidlawii B. Two endotherms were observed: The one at lower temperature was fully reversible and appeared to be due to the phase transition of gel phase lipids while the irreversible transition at higher temperature was pr.obably due to denaturation of membrane proteins. Brandts and his co-workers have made very effective use of DSCin studying the humanerythrocyte membrane (57). Four well-defined transitions were observed, of which two have been demonstrated to be due to relatively simple unfolding transitions of proteins (58). Loike et al (59) employed DSCto study the heat produced by suspensions of murine macrophages. The total heat that was evolved in the interval 10-37C, during scanning at 1 K rain -~, ranged from 300 to 2500x 10-12 cal per cell, dependingon cell density, glucose concentration, and the presence or absence of various drugs. Approximately 24%of the heat liberated was due to conversion of glucose to lactic acid, and an

Annual Reviews www.annualreviews.org/aronline


DIFFERENTIAL SCANNING CALORIMETRY 487

additional 14-~1%was due to hydrolysis of ATP, depending on whether the final product was ADP or AMP. Thompson et al (60) found that the complex DSCcurve observed heating a suspension of photosystemII in the temperaturerange 30-70C could be resolved into 5 two-state curves. Anearlier study (61) with a less sensitive instrument had given similar results. A partial assignmentof these peaks to the componentsof the system has been made, including assignment of a peak at 48.2C to the functional denaturation of the oxygen-evolving complex. This peak, which accounts for only 2%of the total enthalpy absorbed, is extremely sharp With AHvn/AhcaI = 3 106. The cause of this unusualeffect is unknown. DSCgives a convenient method for following the thermally induced reversible polymerization of the coat protein of tobaccomosaicvirus (62).
ACKNOWLEDGMENTS

Thedifferential scanningcalorimetric work in the authors laboratory has been supported by grants from the National Institute of Health (GM04725) and the National Science Foundation (PCM-8117341and DMB8421173). Comments, suggestions, and corrections by Drs. S. Kitamura and A. Tanakawere very helpful.
Literature Cited 1. Privalov, P. L., Plotnikov, V. V., Filimonov, V. V. 1975. J. Chem. Thermoo dyn, 7:41-47 2. Privalov, P. L. 1980. Pure Appl. Chem. 52:479-97 3. Borchardt, H. J., Daniels, F. 1951. J. Am. Chem. Soc. 79:41-46 4. Filimonov, V. V., Privalov, P. L., Hinz, H.-J., vonderHaar, F., Cramer, F. 1976. Eur. J. Biochem. 70:25-31 5. Privalov, P. L., Filimonov, V. V. 1978. J. Mol. Biol. 122:447-64 6. Freire, E., Biltonen, R. L. 1978. Biopolymers 17:463-79 7. Freire, E., Biltonen, R. L. 1978. Biopolymers 17:481-96 8. Freire, E., Biltonen, R. L. 1978. Biopolymers 17:492510 9. Freire, E., Biltonen, R. L. 1978. Biochim. Biophys. Acta 514:54~8 I0. Filimonov, V. V., Potekhin, S. A., Matveev, S. V., Privalov, P. L. 1982. Mol. Biol. 16:551~52 11. Gill, S. J., Richey, B,, Bishop, G., Wyman,J. 1985. Biophys. Chem. 21: 114 12. Edge, V., Allewell, N. M., Sturtevant, J. M. 1985. Biochemistry 24:5899-5906 13. Chang, L.-H., Li, S.-J., Ricca, T. L., Marshall, A. G. 1984. Anal Chem. 56: 1502-7 14. Chang, L.-H., Marshall, A. G. 1986. Biopolymers 25:1299-1313 15. Albon, N., Sturtevant, J. M. 1978. Proc. Natl. Acad. Sci. USA 75:225840 16. Privalov, P. L. 1982. Adv. Protein Chem. 35:1-104 17. Manly, S. P., Matthews, K. S., Sturtevant, J. M. 1985. Biochemistry 24: 3842-46 18. Pfeil, W. 1986. In Thermodynamic Data for Biochemistry and Biotechnology, ed. H.-J. Hinz, pp. 349 76. Berlin/Heidelberg: Springer-Verlag 19. Privalov, P. L. 1979. Adv. Protein Chem. 33:167-241 20. Takahashi, K., Sturtevant, J. M. 1981. Biochemistry 21: 6185-90 21. Fukada, H., Sturtevant, J. M,, Quiocho, F. A. 1983. J. Biol. Chem. 258: 1319398 22. Novokhatny, V. V., Stanizlav, A. K., Privalov, P. L. 1984. J. MoL Biol. 179: 215-32 23. Grutter, M., Matthews, B. 1982. J. Mol. Biol. 154:525-35

Annual Reviews www.annualreviews.org/aronline 488 STURTEVANT


43. Simon, S. A., Mclntosh, T. J. 1984. Biochim. Biophys. Acta 773:169-72 44. Katz, Y., Diamond, J. H. 1984. J. Membr. Biol. 17:101-20 45. Constantinides, P. P., Ohosaini, L., Inouchi, N., Kitamura, S., Seshadri, R., Israel, M., Sartorelli, A. C., Sturtevant, J. M. 1987. Biophys. J. 5l: 239a 46. Maggio,B., Ariga, T., Sturtevant, J. M., Yu, R. K. 1985. Biochemistry 24: 108492 47. Maggio,B., Ariga, T., Sturtevant, J. M., Yu, R. K. 1985. Biochim. Biophys. Acta 818:1-12 48. Maggio,B., Sturtevant, J. M., Yu, R. K. 1986. J. Biol. Chem. 262:2652-59 49. Rochas, C., Rinaudo, M. 1982. Carbohydr. Res. 105:227-36 50. Norton, I. J., Goodall, D. M., Morris, E. R., Rees, D. A. 1983. J. Chem. Soc., Faraday Trans. 1 79:2475-88 51. Manning,G. S. 1969. J. Chem. Phys. 51 924-34, 934-49 52. Paoletti, S., Smidsrod, O., Grasdalen, H. 1984. Biopolymers 23:1771-94 53. Paoletti, S., Cesaro, A., Delben, F. 1983. Carbohydr. Res. 123:173-78 54. Kitamura, S., Kuge, T., Sturtevant, J. M. 1986. Abstr. US Calorimetry Conf., p. 171 55. Itou, T., Teramoto, A., Matsuo, T., Suga, H. 1986. Macromolecules 18: 1234-40 56. Steim, J. M., Tourtellotte, M. E., Reinerr, J. C., McElhaney,R. N., Rader, R. L. 1969. Proc. Natl. Acad. Sci. USA63: 104-9 57. Jackson, W., Brandts, J. F. 1970. Biochemistry 9:2294-2301 58. Brandts, J. F., Erickson, L., Lysko, K., Schwartz, A. T., Taverna, R. D. 1977. Biochemistry 16:3450-54 59. Loike, J. D., Silverstein, S. C., SturUSA 78:5958-62 60. Thompson, L. K., Sturtevant, J. M., Brudvig, G. W. 1986. Biochemistry 25: 6161~9 61. Cramer, W. A., Whitmarsh, J., Low, P. S. 1981. Biochemistry 20:157~2 62. Sturtevant, J. M., Velicelebi, G., Jaenicke, R., Lauffer, M. A. 1981. Biochemistry 20:3792-3800

24.Matthews, C. R., Crisanti, M. M., Gcpncr, G. L.,Vclicclebi, G.,Sturtcvant, J. M. 1980. Biochemistry 19: 129093 25. Yutani, K., Khechinashvili, N. N., Lapshina, E. A., Privalov, P. L., Sugino, Y. 1982. Int. J. Pept. Protein Res. 20: 33136 26. Pabo, C. O., Sauer, R. T., Sturtevant, J. M., Ptashne, M. 1979. Proc. Natl. Acad. Sci. USA 76:1608-12 27. Hecht, M. H., Sturtevant, J. M., Sauer, R. T. 1984. Proc. Natl..4cad. Sci. USA 81:5685-89 28. Hecht, M. H., Hehir, K. M., Nelson, H. C. M., Sturtevant, J. M., Sauer, R. T. 1985. J. Cell. Biochem. 29:217-24 29. Hecht, M. H., Sturtevant, J. M., Sauer, R. T. 1986. Proteins: Struct. Funct. Genet. I: 43-46 30. Privalov, P. L. 1974. FEBS Lett. 40: S140-53 31. Breslauer, K. J., Frank, R., Blocker, H., Marky, L. A. 1986. Proc. Natl. Acad. Sci. USA 83:3746-50 32. Bode, D., Schernau, U., Ackermann, T. 1973. Biophys. Chem. 1:214-21 33. Marky, L. A., Snyder, J. G., Remeta, D. P., Breslauer, K. J. 1983. J. Biomol. Struct. Dyn. 1:487-507 34. Chaires, J. B., Sturtevant, J. M. 1986. Proc. Natl. Acad. Sci. USA83:5479-83 35. Singer, S. J., Nicolson, G. L. 1972. Science 175:720-31 36. McElhaney, R. N. 1982. Chem. Phys. Lipids 30:229-59 37. Chowdhry,B. Z., Lipka, G., Dalziel, A. W., Sturtevant, J. M. 1984. Biophys. J. 45:901-4 38. Chowdhry,B. Z., Lipka, G., Hajdu, J., Sturtevant, J. M. 1984. Biochemistry 23: 2044-49 39. Chen, S. C., Sturtevant, J. M., Gaffney, B. J. 1980. Proc. Natl. Acad. Sci. USA 77:5060-63 40. Kodama, M., Hashigami, H., Seki, S. 1985. Biochim. Biophys. Acta 814: 3006 41. Sturtevant, J. M. 1982. Proc. Natl. Acad. Sci. USA 79:3963~57 42. Sturtevant, J. M. 1984. Proc. NatL Acad. Sci. USA 81:1398-1400

tevant,J. M.1981. Proc.Natl.Acad. Sci.

You might also like