ABNORMAL IONIZABLE
Aug. 5, 1961 [CONTRIBUTION F R O M THE
DEPARTMENT O F CHEMISTRY.
VI.
Structural Studies of Ribonuclease. BP
JAN
HERMANS, JR.,
AND
3293
GROUPS O F RTBOSUCLEASE CORNELL USIVERSITY,
I THACA, NEW
YORK]
Abnormal Ionizable
HAROLD A. SCHERACA
RECEIVED OCTOBER7 , 1960 Evidence has been obtained from ultraviolet difference spectra and optical rotation measurements for the presence of abnormal ionizable groups in ribonuclease. If native ribonuclease is brought from pH 7 t o pH 2 below 15', a small conwith essentially no change in optical density.o Presumably a group with figurational change occurs (small increase in - [a]) P K o b a d . = 2.65 is involved in this transformation. If the p H is further lowered t o 1 (below IS ), another COO- group (or g o u p s ) near a tyrosyl residue acquires a proton (PKobad,= 1.5), giving rise to a decrease in optical density with essentially 110 change in [a]. Although the COO- group is near enough to the tyrcsyl residue to affect its absorption spectrum, t h e two groups are not necessarily hydrogen-bonded t o each other. However, if one is interested in elucidating tlie internal configuration of a protein by determining which groups are near each other, it is not necessary to decide nhether or not a hydrogen bond exists between the two groups. At elevated temperatures. larger changes in optical density and optical rotation are observed and reflect larger changes in configuration of the molecule. The configurational changes are summarized in Fig. 6. T w o possible models, incorporating these abnormally low pK carboxyl groups, were used previously to account for t!ie PH-dependence of the standard free energy of denaturation. These are also used here to account for t h e titration curve of native ribonuclease and for t h e difference titration curve between the native and denatured protein.
Introduction In the preceding paper of this series4ss an investigation of the pH-dependence of the reveisible configurational change in ribonuclease mas reported. By attributing the pH-dependence of the denaturation t o the presence of specific abnormal ionizable groups in the native molecule, it was possible to account for tlie pH-dependent part of the standard free energy of denaturation with the aid of two possible models. These models are characterized by the particular number and nature of abnormal ionizable groups assumed to be present in them. The present paper, being a continuation of the previous ~ n e , provides ~J direct evidence for the presence of these abnormal ionizable groups. This evidence was obtained from pH-dependent spectral changes and optical rotation changes discussed in the previous paper and also from similar types of measurements not previously reported. The models, which were used4 previously to explain the denaturation data, are further used here to account for: (1) the titration curve of native ribonuclease, which is known to show certain abnormalitiesfi a t low PEI, and (2) the difference titration curve between native and denatured molecules. Experimental All of the experiniental niethocls have been described in the preceding paper4 with t ! i r except ion of the following. Titrations.-Direct titrattoils n ere performed with n Radiometer T T T l titrimeter and SBR 2a recorder ds prebiously ~ l e s c r i b e d . ~The standardization R as carried out with the same buffrrs as were used in related nieasurements nith the Beckmail pti-meter Kitrogen was used to keep the so!utions free of COS. The initial pH was about 2, (1) This investieation was supported by research grant E-1473 from the National Institute of Allergy and Infectious Diseases, of the National Institutes o f Health. U. S. Public Health Service, and by research grant G-64iil from the National Science Foundation. (2) Presented before t h e Division of Biological Chemistry a t t h e 131th meeting of the American Chemical Society, Chicago, Il!inois. September, 1958. (3) P a r t of this work was summarized by FI. A . Scheraga, C. Y.Cha, J. Hermans, Jr., and C. L. Schildkraut, "Amino Acids, Proteins and Cancer Biochemistry." Academic Press, New York, N . Y., 1960, p. 31. (4) J. Hermans, Jr.. and H. A. Scheraga. J. A m . Chcm. Sac., 83, 3283 (1961). ( 5 ) T h e previous paper should he consulted for nomenclature and details of the experimental work since t h e previous experiments and their interpretation have a direct bearing on t h e work of this paper ( 6 ) C. T m f o r d and J, D. Hauenstein, J . A m . Chern. Soc.. 18, 5287 (1956). (7) C. Y.C h a and H . A. Scheraga, ibid., 81, 54 (1960).
and the titrations were carried out with 0.8 A: Cog-free KOH in about one-half hour. The proteiii concentration was 25 mg./ml. and 10 nil. was titrated. Titrations were carried out on the solvent and on the protein solution a t 25.0 & 0.1" and 44.5 i 0.2'. The amount of acid bound t o the protein at any pII was computed by subtrarting the amount of base required t o titrate the solvent from that required to titrate the solution. I n order t o calculate the average nurnher of protous dissociated per mole, a molecular weight of 13,6838 mas used. 1)uplicate titration curves agreed with each other t o within 0.1 group a t all pH-values tietween 2.5 and 4.5.
Results Effect o l pH on Difference Spectra.-In the preceding paper4 the transition from native t o reversibly denatured ribonuclease mas follorved by measuting ultraviolet difference spectra at constant p H as a function of temperature. Since the transition occurs a t different temperatures when the pH is varied (see Fig. V39 which is a series of ADz6, us. T curves), it is also possible to determine a set of spectrophotometric titration curve.s (i.e., ADBp? 11s. PH curves) at various temperatures, similar to the one published previously. l o Such a set is shown in Fig. 1, the ionic strength being 0.08 M , that of the solutions of Fig. V2. The reference solution at all temperatures was a t a P€I between 6 and 7 (measured at t.hose temperatures). I t already has been shown in Fig. V2 that the optical density of a pI-1 tj.S3 solution decieaws a t temperatures above 43". Thus, the curves of Fig. 1 a t 45, 50 and 55" level off near p H A a t non-zero values of AD. These limiting values of AD a t high PH were obtained by adjusting tkle low fiH end of the 45, 50 and t55° curves to the values for the 35 and 40" curves a t p€I 1.2, since all the curves of Fig. V2 approach the Saxe asymptote above 43'. The curve of Fig. 1'6 has a horizontal portion a t low P H ; thus (in Fig. 12)if a curve v-ere obtainable at n pH lower than 0.89, it would be expected to superimpose on the PH 0.89 curve. As a result, the largest change in AD as the p H is lowered from 7 to j 0.89 will be constant from about 0 to about (8) C. €1. W.Hirs, S. Moore and W. H . Stein, J . B i d . Chcm., 219, 623 (19BB). (9) Figures and equations of paper VI will be referred to with t h e prefix V, c . P . ,Fig. V1 a n d eq. V2, etc. (10) H. A. Scheraga, Biochim. c l Biophys. A d a , 2 3 , 196 (1957). (1.1) These measurements a t lower ionic strength were made a t an earlier date than those of payer VB and were not repeated a t 0.16 AI ionic strength.
3294
JAN
HERMAKS, JR.,
ASD
HAROLD A.SCHERAGA
Vol. 83
curxres and equation 1 applied to the resulting curves: the data of Table I are obtained. Slopes were taken a t all available temperatures a t values of y = ],'4 and 3' = I, '?. Negative values of Ar indicate that fewer protons are dissociated from the denatured niolecule than from the native molecule. It may be noted that the values of - AT at the same PH depend on y , being loweri3 a t the lower value of 3'. Titration Data.--Iii order to determine whether the AY values of Table I are reasonable, titration experiments were carried out. Since the only groups PH ionizing in the pH-range of interest here ($H 2 Fig. 1.-Spectropliiitoiuetric titration curves a t 0.08 V ionic strength and protein concentration 1.90 mg./ml. Teni- to 5) are carboxyls (which in many cases have peratures in " C . are indicated. Tlic high PH solution was essentially zero heat of ionization) the titration curve of the native molecule should be independent the reference at I'< 43' m t i its o;)tical density was set equal of temperature if no configurational changes occur t o zero at eacli temperature. For 7' > 43' t h e optical denand if the carboxyl groups are not hydrogen bonded sity of t h e !ow pIi solutioii (pH 1.2) is set cqual t o --0.%30 in o d y o mforwi.l " -1ny difference found betweeii the and this was taken as the tie^ reference. At 25 and : 3 0 3 titration curves a t two temperatures must then be lowering of the PH does not lead t o 3 s high a value of due to the presence of denatured molecules with a - A D as at 35'; this is indicated hy (lotted lines since tlie irieasuremeiits could not lie extended to lower PH'F kit fliis different titration curve from that of the native niolecule. The temperatures chosen for the comioiiic strength. parison were 43 and 23". The quantity obtained by l i o ,will increase from about 17 to about 33" aiid be subtracting the number of protons dissociated from constant above 33" (see Fig. I?). Thus, while the ribonuclease a t 2.5" from that dissociated from ri25, 30, 35 and 40' curves approach AD = 0 a t the bonuclease a t 43" may be designated as Ah. I t is high @Hend, they do not all have the same limit on easily seen that Ah then should be the product of the low pH side. Since the ionic strength sets a 1,. and 3'. lower limit to the attainable PH, the low pH end of -It 45" y changes from close to zero a t pH 4 to the 23 and 30" curves of Fig. 1 mas not coinpletely essentially unity a t pH 2.4, while essentially no realizable and is therefore shown as a dashed line. denatured molecules are present a t 23" in this pHIf the PK's of the ionizable groups of the tleria- range (see Fig. 1). In Fig. 2 open circles represent tured protein are not the same as those of the the values of A h . The curve drawn in Fig. 2b is native protein," the value5 of y (cq. I v l l , the that representing the spectral data a t 45", i.e., y as a fraction denatured, will depend on P H . l y e may rep- function of pH a t 45".and was taken from Fig. 1. resent the difference iii the average number of 111Fig. 2a solid circles represent the analog uf the groups ionized, between denatured a ~ i r l natire spectral curve of Fig. 1 a t 45" but a t an ionic molecules, by the symbol A? (AY = r ~ i- YX). strength of 0.16 -11.The ordinates of the curves of I t is shown in Appendix I1 (eq. X13) that this yuan- Figs. 2a and 2b are so adjusted that AD = ADmax tity is related to the slope of tlie curves representing (or y = I) corresponds to Ah = - 2.0 groups, a kind 3' a s a function of pH, by oi rnaxiniuin value of AY from Table I. IT-ith this adjustment the difference in the direct titration tly;dpH = 2 . ; w j ~ '1( - ?)A, (1) If the cur\-es of Fig. 1 are con\-erted to ?' s's. pH curves !rcpresentecl by open circles) corresponds well with the sI'ectrophotonictric titration curve a t ,.1 ri I_P I $H's ah0x-v i 3 2.3 t i . ( ,,, slnpes and absolute values are then matched) 51I-IFW- of I ~ i hC I X \ E C (IF F r c ; 1 , 1 trE L ' . ~ L L Y S III .L, The results nbtained by applying equation I CCIMPT~I.EI> ?"E.REPR~IV [ 7 s i s ~ ~ ?;I J I - A . T I O ~1 and those obtained by direct titration are thus seeri 'l'emp., tdy ' C . dbH),-i/? A!' ($H)v-ij>d$H)y-i/i AI (DH)v-i/+ to be in good agreement with each other. The fact '
, \ \ ~ j
25 ' :30 35 40 45
,
.io
I
_- I .,.,
(-0.75)"
(-1.3)
(-0.88)'' -1.25 -1.?Y -1.31 - 0 89 0 . 67
(-15) - 2.1 -2 2 - -2 2 - I .j - 1 1
-
11.5) (1.8) 2.2 2.2 2.9 : 2
R i
(-0
-~
04)'' .titi
-
.78
-
5s
(-1 5 ) -1.5 -1.8
-
(1 8) 2 1 2 4
1 :j
.-
--,IS
-0.4
1 . i
The computed valuci if this table iit p H 2 arid l u w t ~ i i i i error for the follo\ving reason. Figure Y2 shows t l i < l i ; i t low PH A D does not become equal t o zero a t low temperatures, even though 3' does (as \vas fouiid from a coinpiiriwii betveen spectra1 and optical rotation data in Pig. \ - 4 ) . Thus, is not proportional t o A D be!ow p H 2 , t i i n tl dpH to dyldpH, as we have assumed in computing t h r , d p H values in this table. 'I
(12) Dissociation constants m a y differ for m a n y reasons, e.,?., if loniziui: groups act as donors or acceptors in hydrogen bonds or if electrostatic inrcractioni are different lvhen the configuration of t h e molecule has chauged or if hydrophobic regions surround the ionizable group i n m e oi its ronfigurations but not in the other
( 1 3 ) 11 I S interesting t o note t h a t a t y = d , 4 the value of -A;' IS still higher. .4t this value of 3 t h e curves in F i g . 1 are n o t well defined b y experimental points. W r do, however. possess a series of measurements of AD as t i function of P H a t 4,i' [ a n d ionic strength 0 l ( i W , which do deiine the curve n e w y = 1 :These d n t t i nrr reivesented h y the tilled circles of F i g 2a ) From the +lop? :it = a!a o n e ohtairih - A Y 2 i : i t fit1 '2 7 . \vhich is higher th:in nnq ol t h e - AI T - ~ I I I I C c)f ~ 'I'al,l? I 8\14) Actiiallp, ,i c~,tiIigurational chanjic. has iireii shown to occur( a n d , further, i t will be shown below t h a t a carboxyl groupts) is prolj 7
COOH and as COO -) ably hydrogen bonded i n both of its forms See also .4ppendix I. (15) Below pE1 2.5 the direct titration d a t a aparoach a - A h value of 3 and do not w i n c i d e h t all with the spectral d a t a . However, this is in agreenient with the increase in - A T with y a t constant p H , which was noted's in Table I . T h e inflection point of Fig. 2 is in reality the inflection point of t h e y vs. pH, riot the P ws. $11 curl-e. Therefore, p a n t of Pig. 2 is iint t u he Interpreted :i, the groiip(4
ABNORMAL IONIZABLE GROCPSOF RIBONUCLEASE
Aug. 5 , 1961
3295
L
-3.08
\o
$
IC
-4
4
IC
Ionic strength
0.08M
8.
Fig. ?.-The difference in groups titrated a t 45' and 25' (open circles) compared with the spectrophotometric titration curve a t 45'. T h e spectrophotometric data were so represented t h a t they coincide with t h e titration-curve points a t high PH and t h a t they level off a t low pH a t two groups difference (ie., AD = AD^^^ corresponds t o Ah = -2). The curve in b was taken from Fig. 1, t h e spectrophotometric d a t a in a (filled circles) were determined in a separate experiment but are entirely analogous. In Fig. 2a there is a point (not shown) on t h e horizontal portion a t pH 0.9.
t h a t at one pH different values of Ar are found, depending on the value of y (see Table I ) , can best be explained by assuming t h a t more than one denatured form of ribonuclease exists (each with a different titration curve) and t h a t the distribution of the denatured material over these forms is a function of PH and temperature and therefore varies with y . As a simplification we shall assume that a t values of y 6 l / Z only one denatured form exists,16 even though small variations occur in Ar as a function of y a t constant PH at low y-values (see Table I). The good agreement between direct titration data and the results of Table I, based on ey. 1. shown in Fig. 2 , provides confidence in the reasonableness of the Ahr values of Table I. These Ar values will be compared later with those computed theoretically for different models. Difference Spectrum at Low Temperatures.When a solution of ribonuclease a t pH 0.9 and ionic st.rength 0.16 M is cooled below l5', essentially all of the molecules are in the native form (see Fig. V4). Nevertheless, as may De seen in Fig. V2, this solution has a negative difference spectrum with respect to a solution a t pH 6.8 a t the same temperature (i.e., AD - 0.06 a t 10') and also (from Fig. V3) a higher negative specific rotation than does a solution of pH 6.8 a t the same temperature. The form of the difference spectrum under these conditions is slightly different from that accompanying the configurational change (Fig. Vl) and is shown iti Fig. 3, the peaks occurring a t 288.5 and 281 m p instead of a t 287 and 280 mp as in Fig. VI. The value of this optical density difference a t 288 inp has been determined a t various temperatures, PH's and ionic strengths. Data a t 10' are shown in Fig. 4. Changing the temperature to 0" or to 15' did not affect the data noticeably indicating that the heat of ionization i s essentially zero. The observed spectrophotometric titration curves again correN
0 6 ) The hypothesis of the occurrence of only one form of reversibly denatured ribonuclease below, but more than one above, the transition temperature. was made in the preceding paper4 to explain the asymmetry (skewness) of the curves representing y us. T a t constant p H .
AD.
Wavelength (mp).
Fig. 3.-Difference spectrum a t ionic strength 0.38 -11,pH 0.60 against a reference solution of zero optical density a t pH 6.5, both a t 0' (protein concentration 1.90 mg./ml.). -
0
,
0
8
8
lonlc strength 0
-O,O4I
0.32
0 0.16
, '++;\o
-0.02
0
I
2
3
4
PH
Fig. &-Optical densities of solutions a t various ionic strengths as a function of pH at 10" compared with solutions of the same ionic strength and concentration a t pH = 6 to 7 . The two curves are of the form pH
- log --;I' (1
- x)
- pK, where
2.Ll/ADrnax,with AD,,, = 0.074 and with pK's chosen so t h a t they fit the datd a t ionic strength 0.32 and 0.16, respectively. The verticel lines indicate the value of p K (where x = 1/2). N =
spond to the ionization of groups on the protein molecule. Two curves have, therefore, been drawn in Fig. 4 corresponding to the equation
adjusting the value of pKobd. to obtain the best agreement with the experimental data. Since the value of cannot be obtained a t these ionic strengths, the arbitrary choice of -0.074 is supported by a single measurement a t the higher ionic strength of 1.3. It may be noted that the P K o b s d increases from 1.53 to 1.67 when the ionic strength increases from 0.16 to 0.32 M. This is theincrement to be expected on the basis of genecal electrostatic interactions'' for a sphere of 17.1 A. radiusL8with no change in configuration upon ionization in a molecule of net positive charge of 18.6e19Using the appropriate electrostatic interaction factor a value of 2.48 is obtained for the (pKH)5 when the charge on the molecule is zero. Thus, the low-temperature (17) See, e x . , I. M. Klotz, in "The Proteins" (Ed. by H. Neurath and K. Bailey), I B , Academic Press, Inc.. New York, N . Y., 1953, p. 9135. (18) C. Tanford, J, D. Hauenstein and D. G . Rands, J . A m . Chcm, Soc., 77, 6409 (1956). (19) See the next subsection for evidence that the configuration does uot change i u this range of gR a n d leinpcralrtrc.
3296
JAN
-1521I
I I
I
HERMANS, JR., .cmHAROLD A . SCHERAGA
I
I
3
4
5
PH.
Fig. 5.-Optical rutation as a function of pH a t Iso,ionic strength 0.16 111. The curve was drawn through the experimental points. T h e line is the tangent of maximum slope to the curve representing the variation of [0rl438 betwzen the observed limits, following the equation describing the ionization of ii monovalent weak acid with pK = 2.66.
spectrophotometric titration curvc indicates that there is at least one carboxyl group in the native niolecule with a low p K H of 2.5. It should be pointed out that the data a t still lower ionic strength (0.08 M ) do not fit into this simplified scheme. If a curve is drawn to fit the three experimental points (according to eq. 2), agreement is obtained only with a value of (AD288)max which is much higher than 0.074. Unfortunately, no data could be obtained below pH 1.2 a t this ionic strength, and the origin of this discrepancy is unexplained. It is important to note the absence of a similar difference spectrum at temperatures (I'> 43'; see Fig. V2) where all the molecules are in the unfolded form. l y e may conclude from this that the group whose ionization produces the difference spectrum a t 10" is in all likelihood one (or two) which contributes to the pH-dependence of the equilibrium between native and unfolded ribonuclease, and thus to the value of - A r , which has a niaximum of about two groups. Further support for this conclusion was provided in the preceding paper, where the presence of this group (or these two groups) with pKobad.= 1.5 (at 0.1G dlionic strength and pH 1.5) and zero heat of ionization was used to account for the pH-dependence of the standard free energy of denaturation. Effect of p H on Optical Rotation at Low Ternperahre.-We shall now show that the low temperature optical density changes reported in the previous subsection are not accompanied by changes in the configuration of the molecule. Since the specific rotation at ];io is different at pH 0.9 from that at pH G.S (see Fig. V3), it was necessary to determine whether the ionization (with ,@Kobad. = 1.5) is accompanied by a change in [ c Y ~ G ~ At . an ionic strength of 0.32 Ai? and temperature of Iso, the value of is 159 (i.e., 8 units greater than that at pH 7, see Fig. V3) and independent of p H in the range 0.7 to 1.8, even though AD288 is changing (see Fig. 4). This shows t h a t the pH-dependence of [a1 does not follow that of AD shown in Fig. 4. Thus, apparently no configurational change ac-
Vol. s3
companies the ionization of the abnormally strong group with P K o b s d . = 1.5. T h e fact that the ionization of this group i s accompanied by a change in the tyrosyl ultravgolet absorption spectrum implies that t h r e exists some direct interaction between a tyrosyl group and the carboxyl group with abnormally low PK. The alternative explanation for the occurrence of the change in spectrum, namely, t h a t the ionization of the abnormally strong group can trigger a configurational change in which a tyrosyl group is also involved, has thus been ruled "ut,",?l,2~ The optical rotation measurements, just cited, provided evidence that one or more carboxyl groups (with p K o b s d . = 1.5) are responsible for the low temperature difference spectrum in the p H range of 0.S to 1.Swithout any accompanying configurational change. At a somewhat higher pH (1.8 to 4) there is a small change in optical rotation at low temperature, as can be seen from Fig. 5 . These nieasurements indicate the possible presence of a carboxyl groups(s) with p k - o b s d . = 2.6,5, this group being involved in a configurational change. This small configurational change is not accompanied by any change in AD (Fig. 4). It is remarkable that at l l i g h temperatures, where all the molecules are completely unfolded, there is also a change in optical rotation with p H , as can be seen from Fig. V3, and t h a t this change is of about the same magnitude as the one a t low temperatures, just discussed. It would, therefore, appear that the two configurational changes, namely. the one under discussion and the one forming the subject of the preceding paper,4 occur to a great deal independently of one another. This point needs further investigation, however, and our conclusion should be regarded as a tentative onc. Discussion Schematic Representation of the Configurational Changes.-To avoid confusion b y the rather large number of different causes (i.e., configurational changes and ionizations) which lead to various effects (changes in optical density and optical rotation) when the pH or temperature is changed, we shall make use of Fig. 6, which schematically represents the various forms in which the ribonuclease molecule can occur. Helical and random portions of the molecule are indicated by the usual symbols. (20) See also S. J. Leach and H. A. Scheraga, J. R i d . Chern., 236, 2827 (19601, (21) We have demonstrated here t h e effect of t h e ionization of a COOH group on t h e absorption of a nearby tyrosyl residue. From the point of virw of its effect on t h e spectrum, this neighboring charge effect is indistinguishable from D tyrosyl-carboxyl hydrogen bond. Ilowever, if one wishes t o elucidate t h e internal con6guration of a protein by determining which side-chain groups are near each other. it is not necessary t o decide whether or not a hydrogen bond exists hetween the two groups, This kind of charge effect, in which t h e thrredimensional folding brings rcmolc groups near each ulher, is different from t h e one discussed previously,Zo based on experiments of Donovan, c f al.,22 on low molecular weight model compounds. In the latter cases, attention was focused20 on charge effects due t o groups which were nearby i n fhr polypcplidc chain. (22) J. W. Donovan, M. Laskowski, Jr., and H . A. Scheraga, Biochim. et B i o p h y s A d a , 29, 455 (1958). (23) In the Discussion section and in Appendix I it will be shown how t h e experimentally observed zero value for the heat of ionization of t h e abnormal carboxyl group with pXH = 2.5 supports t h e conclusion t h a t no configurational change accompanies t h e ionization of this group below 1.5.'
Aug. 5, 19Gl
ABNORXAL IONIZABLE GROUPSOF RIBONUCLEASE
3207
V3 is observed. A small helical region is retained in I V because the high-temperature limiting value CL-e.rgi5. pH of - [aju6 a t $H 6.5 is less than t h a t a t PH 0.9 0 '3(p~6: PH-2 0 pHvI (see Fig. V3) by an amount corresponding to the - - Tg4O' T415' total change observed in Fig. 5 . This conclusion is 1 II tentative, however (see above). m , ' ; - . . (V) At low pH the small helical portion of IV $p can unfold givin rise to an additional increase in -- [ a ] 4(Fig. ~ ~ IJ37; also the COO- groups take up protons in the transformation of I V to V without a >L"" change in optical density since the tyrosyl and carT>-o* T>7C' boxyl groups in IV are already far apart Form I - , , 1,450 ,can be converted to form V by lowering the PH a t 1 -,-A temperatures above 15'. 7 ZI Titration Curves.-Tanford and Hauensteid Fig. B.-,%hematic representation of the various forms of have published measurements of the direct titraribonuclease. A particular combination of temperature and tion of ribonuclease iii 0.15 AI KCI. The data in p H , where any form predominates, is indicated, as are the the range where the carboxyl groups titrate (PH 1 changes in p H and/or temperature which will reversibly to 6) involve certain abnormalities, which indicate transform one form into others. Helical and random portions that an undetermined number of groups have a pK are indicated by t h e usual symbols; t h e circle in forms which is lower than that which is normally found I, I1 and I11 indicates an unspecified interaction, which low- ( p K o = 4.6). Since we have found evidence, ers the pK of a t least one carboxyl group and increases t h e which in part determines the extent of these abnorpK of a t least one tyrosyl group. The pictures are not meant malities, we have computed titration curves (with to resemble the actual structure of ribonuclease on any but eq. V, A4) in the region p H 1 to 7 , using different these points Also, the whole molecule is not represented, sets of ionizing groups, according to models A, B, but only those portions pertinent to the present discussion. C1 and C2described in Table V-IT. A group with The circled region implies the existence of inter- pKobsd.= 2.65 ( p K H = 3.65), whose possible existactions, which presumably stabilize the molecule ence was inferred above, was included in models B, and which modify the pK of the ionizing groups C1 and Cf since i t improves the agreement with the participating in these interactions. The following experimental titration data; it was omitted from model A since i t makes the agreement worse. five forms may exist: (I) Native ribonuclease, with two helical regions, While this group does affect the titration curve, it one consisting (in the diagram only) of two large does not affect AF$ of the preceding paper very helices, the other of one small helix. Somewhere in much. I n computing the titration curve of ribonuclease the large helical region there is a zone of interaction, with (at least) one carboxyl and one tyrosyl group. a t 25", 0.15 Af ionic strength, we have to take into The carboxyl and tyrosyl groups in the interaction account the configurational change. From the data zone are close but not necessarily hydrogen bonded i!i Fig. V2 we can see t h a t the fraction denatured a t to each other. No other features of the ribonu- p H 2.0 and 25" is only 0.09, so that the correction clease molecule are meant to be represented in the for the presence of the denatured form is only a minor one since the measured titration curve does diagram (11) At low temperature and QH the small helical not descend below pH l.S. Since not very much is gained by inspection of region becomes random (with an ionization of agroup with PKo,,sd = 2 G), and we observe the small the graphs containing calculated and measured chanve in [a1416 of Fig. 5. No spectral change titration data, lhe data are presented in Table I1 as acconipanies the transformation of I to IT since the TABLE I1 tyrosyl-carboxyl interaction zone is unaffected. V A L C EOF ~ A MINUSCOIZPUTED VALVEOF h (111) .At still lower p H and low temperature, MEASURED (Absolute largest values) the carboxylate ion in the circled region takes up a Model In pH-range 1.8 t o 2 5 In 913-range .? t o 4 5 proton (bKobsd = 1.5); no configurational change 4 -1.2 $0 3 occurs, however. The proximity of the tyrosyl I3 - .5 +0.4 group to the newly introduced proton gives rise to C' .2 +1.0 the change in optical density of Fig. 4. Ca - .5 +0.4 (IV) If the temperature is raised a t high p H , the large hclical portion of I unfolds, and the change in D of Fiq. V2 is observed because the tyrosyl residue the absolute largest value which the differences is no longer in the circled region, ;.e., no longer ab- attain in two pH-ranges. Above pH 4.5 all curves normalz4;also, the change in optical rotation of Fig. are in good agreement with the experimental d a t a 0.2 group). Below pH 4.5 models B (within (24) In the transformation I 111. the tyrosyl OH group remains and COgive the best agreement. near the COOH group and a small value of A D is observed In the Difference in Titration Curves.-The difference IV, the interacting tyrosyl and carboxyl groups transformation I in numbers of groups titrated, Ar, was obtained by are removed f a r apart from each other a n d a large A D results As discussed below, this larger value of A D may he the result of both the subtracting the computed titration curve for fully rupture of a tyrosyl-carboxylate ion hydrogen bond and t h e removal native molecules from that of fully denatured moleof the tyrosyl group from a hydrophobic region (which surrounds t h e cules. The theoretical curves for models A, B, C1 tyrosyl and carboxyl interaction zone io form I) t o a n aqueous medium in form IV. and CZare shown in Fig. 7. The open circles are 7 i
c*
,C X 0 dI
0
E=
q =+pb
-vc % ??: p-"
0
c%:ower
3
L -
q
COO"
-.
-
01
DH
i
+
-.
-
*
Alternatively, it might be possible to determine the number of tyrosyl groups which become normal upon reversible denaturation a t low bH and high temperature from the value of the :I change in molar extinction E a t 287 mp, which is about - lS00 for complete reversible denaturation (Fig. 1). Recent data of Bigelow, ~t a / . , indicate a decrease of about 900 in €287 for every tyrosyl group becoming normal, following various changes in the ribonuclease molecule which lead to complete or partial normalization of these groups.3"131From this it would follow t h a t of the three groups which are abnormal in native ribonuPH. clease, one is still abnormal in the reversibly deFig. ;.--Number of groups ionized on the denatured ribo- natured molecule. Recent experiments in this nuclease molecule niinus t h a t on the native molecule for four Laboratory32with low pH irreversibly heat denamodels. Open circles are experimental points, taken from tured ribonuclease (which is indistinguishable from Table I, i.e., from the slopes of the curyes giving fraction de- reversibly denatured ribonuclease by spectral measnatured ( y ) as a function of pH. at y = l / + urementsj32i.e., has LIE^^, = - 1800 with respect to inative ribonuclease), have failed to establish unthe data2j of Table I, with y = I t should be equivocally the number of abnormal tyrosyl groups pointed out that the experimental A7 values are iii this derivative. In light of the findings of subject to an error of a t least 15yo. Thus, curve Bigelow and O t t e ~ e n it , ~would ~ still appear most A can be said to be in fairly good agreement with likely, however, t h a t one abnormal tyrosyl group the experimental data and curve B to be in reason- per molecule is present in both reversibly and irable agreement with them; the curves for models C1 reversibly denatured ribonuclease. and CZapparently give poor agreement. The fact Nature of the Abnormality of the Tyrosyl and that the experimental data were obtained a t ionic Carboxyl Groups.-Having attributed the abstrength 0.08 -411, but the curves calculated a t 0.16 normality of one or more tyrosyl residues to the M introduces little discrepancy because of this presence of neighboring carboxyl groups with p K o b s d . same large experimental error. = 1.5, it is now appropriate to attempt to account In paper V4 we conclude from denaturation data for the abnormal pK's of the tyrosyl and carboxyl that models C1 and C?can be excluded and that no groups. choice could be made between models X and €3. I n order to make clear the difficulties involved, it The comparison between the models on the basis of is instructive to consider the theoretical parameters the titration curve and the difference titration describing the hydrogen bond between a tyrosyl curve presented in this paper leads to the same con- group and a carboxylate ion.Y3 Since the hydrogen clusion. bonding constant33Kij is approximately 1, the Pk' Tyrosyl Groups.-It previously has been of the carboxyl group is lowered 0.3 unit, that of shown18s26from spectrophotometric titration data the tyrosyl group is raised by the same amount that 3 of the 6 tyrosyl groups of ribonuclease have and the heat of ionization of the carboxyl group is an abnormally high p K . Since the unfolding of the changed from zero to -33000 cal.,/mole. I n conmolecule (transformation I to I\.' or I-) is accompanied by a change in the ultraviolet absorption trast, it is observed for ribonuclease that the $KH spectrum, it is simplest to assume that one or more of the carboxyl group is 2 units lower than pKOand of the 3 abnormal tyrosyl groups (circled region of I ever, a t high pH. the reversibly denatured form then undergoes modiin Fig. 6) have become normal. The exact number fication t o irreversibly denatured material. I n support of t h e postuexistence of a reversibly denatured form a t high OH a t room could be ascertained in principle by determining the lated temperature, t h e following may be pointed o u t : Reversibly denatured number of groups (by direct spectrophotometric material occurs in measurable amounts a t room temperature a t p H 1 titration) which remain abnormal in the denatured but is absent a t p H 7 . Theoretically, it is t o be expected t h a t t h e protein, ;.e., a t temperatures above 75". Unfor- equilibrium cnnstant for unfolding should also increase. a t high OH and room temperature, t o values where reversible denaturation is tunately, even though such solutions are stable be- appreciable (cq. V, A23, in which t h e p K of t h e tyrosyl groups in the low pH 7, there is a rapid" irreversible denaturation native form is ahnormally high b u t in the denatured form normal). accompanied by precipitation, above pH T a t i < i o .While we cannot make these measurements a t high g H in solufion, Since the tyrosyl groups ionize above PH 8, i L i s such a decrease of t h e transition temperature, both a t l o w and high pH, has been observed with ribonuclease films.29 therefore impossible to titrate them'a a t '7.5". (29) .1 Nakajima and H. .4. Scheraga, J . A m . C k e m . Sor., 83, 157,;
!
( 2 5 ) If our models were strictly applicahle (i.e.. i f , as airearly pointed o u t , only one denatured form of ribonuclease existed), Ar should be independent of y. However, as shown i n Table I , AT depends on y. Therefore, t h e choice of y = 1/1 is somewhat arbitrarp. (26) D. Shugar, Biochem. J . , 62, 142 (1952). (27) J. Hermans, Jr , and H . A. Scheraga. unpublished results. (28) T h e presence of 3 abnormal tyrosines was deduced from t h e interpretation of t h e irreversible titration o f these groups a t room temperature a t high pH.1b2i I t may thus appear t h a t a n irreversible change of configuration is required in order t o normalize a n y of these groups. Yet we observe a renersible normalization of abnormal tyrosyl group(s) a t low pH. This apparent discrepancy may be resolved a s follows. T h e transformation I to IV is reversible. How-
(1961).
(30) C. Rigelow and 31. Ottesen, Biochim. d f Biophys. A f l a , 32, 574 (1959). (31) C Bigelow, Compf. i ~ e n d .trap. L a b . Ca?!shc,i,g, S P Y .> 1 and AH,,^ = A H , k D must hold. The equality of the two heats of hydroge.1 bond formation might seem surprising, in view of the large Jifference between the equilibrium constants for formation of the hydrogen bond but is in agreement with data on salicylic acid. The analo y between the structural effect in salicylic acid and in rigonuclease is elaborated in the last subsectiori of the Discussion section. (35) M. Laskawski, Jr.. and H. A. Scheraga, ;bid., 78, 5783 (1956).
3300
LEOXIDAS ZERVAS, I I I L T O N
TvISITZ A X D
Appendix 11
J. P.GREENSTEIN
1'01. s3
Since
Denaturation and Change in Hydrogen-ion Binding.LVhen, in a certain PN-range, the ratio of native to denatured molecules is pH-dependent, denaturation must be accornpanied by a change in the average number of hydrogen ions dissociated from the protein molecules, Using equation Y, A11 of the precedinypaper,4 we hdve
dy/dPH = -2.303 dy/d In H = -2.303 (dyld In Kut,a,i.) ( d In I
