J. Mol. Biol.
(1991) 218; 387-396
Anion and pH-dependent Conformational Amphiphilic Polypeptide
Transition
of an
Yuji Gotot Department of Biology, Faculty of Science Osaka University, Toyonaka, Osaka 560, Japan
and Saburo Aimoto Institute for Protein Research Osaka University, Suita, Osaka 565, Japan (Received 3 August
1990; accepted 9 October 1990)
While several proteins, including /?-lactamase, cytochrome c and apomyoglobin, are maximally unfolded at pH 2 by HCl in the absence of salt, the addition of anions, either from salt or acid, co-operatively induces the unfolded proteins to refold to a molten globule state, because anions bind preferentially to the compact molten globule state compared to the extended unfolded state. To study the role of the anion-dependent conformational transition at neutral pH, we synthesized a model polypeptide of 51 amino acid residues, consisting of tandem repeats of a Lys-Lys-Leu-Leu sequence and containing a turn sequence, Asn-Pro-Gly, at the center of the molecule. The model polypeptide showed no significant conformation by circular dichroism under conditions of low salt at neutral pH. However, addition of anions, either from salt or acid, induced the folding transition to an a-helical conformational state. The order of effectiveness of various anions in inducing the folding transition was consistent with the series of anions in inducing the molten globule of the acid-denatured protein. This suggests that the helical state of the model polypeptide is equivalent to the molten globule state. At pH values above 9, the model polypeptide also took an a-helical conformation, which was very similar to that induced by anions. On the basis of the chloride and pa-dependent conformational transitions, a phase diagram for the conformational states was constructed. The phase diagram was explained simply by assuming that the conformational transition is linked to the proton and the anion bindings to a limited number of amino groups and that anions bind only to the protonated groups.
1. Introduction
mediate conformation upon addition of anions, either from salt or acid. The intermediate conformation is compact and contains a significant amount of secondary structure, but has a disordered tertiary structure. The intermediate conformation is called a molten globule and is assumed to be a major intermediate of protein folding (Ohgushi & Wada, 1983; Ptitsyn, 1987; Kuwajima, 1989; Ptitsyn et al., 1990). Goto et al. (1990b) examined the salt and aciddependent formation of the molten globule of cytochrome c and apomyoglobin and indicated that the mechanism involves electrostatic interaction of anions with the positively charged groups of the proteins. The conformation of acid-denatured proteins is determined by the balance of the repulsive forces between positive residues, which favor the extended conformation, and mainly hydrophobic forces, which favor the compact molten globule
The conformation of acid-denatured proteins, including /I-lactamase, cytochrome c and apomyoglobin, depends critically on ionic conditions (Goto & Fink, 1989, 1990; Fink et al., 1990; Goto et al., 1990a,b). While these proteins are largely unfolded by HCl at low ionic strength to a conformation similar to that obtained with high concentrations of Gdn-HClf, they refold co-operatively to an intert Author to whom all correspondence should be addressed. 1 Abbreviations used: Gdn-HCl, guanidine hydrochloride; h.p.l.c., high-pressure liquid chromatography; Acm, acetamidomethyl; Boc, t-butyloxycarbonyl; CAPS, 3-(cyclohexylamino)1-propanesulfonic acid; c.d., circular dichroism; CHES, 2-(N-cyclohexylamino)-ethanesulfonic acid; Z, carbobenzoxy.
387 0022~2836/91/060387-10
$03.00/O
0
1991 Academic
Press Limited
Y. Goto and S. Aimoto
388
conformation. The preferential binding of anions to a limited number of basic groups of the compactly folded molten globule state, compared to the extended unfolded state, results in the co-operative refolding transition with an increase in the concentration of anion. Alternatively, the shielding of intramolecular electrostatic repulsive forces in the unfolded state by the anion binding reflects the intrinsic forces that favor the formation of the molten globule. These results indicate the importance of electrostatic interaction in determining the conformational stability of acid-denatured proteins. Such anion-induced conformational transition may not necessarily be limited to acid-denatured proteins, but may have an important role even at neutral pH in determining the conformational stability of proteins and peptides. For example, the conformation of melittin, a bee venom toxin consisting of 26 amino acid residues and containing five basic residues but no acidic residues, depends on salt conditions (Talbot et al., 1979). While melittin is largely unfolded under conditions of low salt’ at pH 75, it assumes a tetramerie helical structure in the presence of a high salt concentration. As another example, Ebert & Kuroyanagi (1982) reported that the conformation of a statistical copolymer of Lys and Leu at a molar ratio of 1 : 1 at pH 7.0 depends on salt conditions (see also De Grado & Lear, 1985). While the polymer is unfolded under conditions of low salt, it assumes a helical conformation in the presence of a high salt concentration. The order of effectiveness of several anions in inducing the coilto-helix transition of the copolymer is consistent with the series of anions in inducing the refolding transition of acid-denatured proteins. These saltinduced conformational transitions at neutral pH may be explained by the same mechanism as that used for the conformational transition of aciddenatured proteins. It is possible that such electrostatic interactions and consequent conformational changes play an important role in the function of some proteins. To understand the anion-dependent formation of the molten globule further, and to clarify the role of this conformational transition at neutral pH, we synthesized a model polypeptide, which consisted of tandem repeats of a Lys-Lys-Leu-Leu sequence (De Grado & Lear, 1985) and contained a turn sequence at the center of the molecule. W’e designed the polypeptide assuming that, while unfolded in the absence of salt at neutral pH due to the electrostatic repulsion between the amino groups, it would form a folded conformation in the presence of salt, in which two amphiphilic helices interact with each other intramolecularly through hydrophobic Leu residues. We report here that the model polypeptide shows an anion-dependent conformational transition very similar to that of acid-denatured proteins, demonstrating tha,t the interactions between anions and basic groups of polypeptides and consequent conformational
changes can occur even at neutral
also show that t.he anion- and pa-dependent
pH. We
forma-
tion of a molten globule-like conformation of the model polypeptide can be explained simply by assuming the linkage of conformational transit’ion with proton and anion bindings to a limited number of amino groups, in which anions bind only to the protonated groups.
2. Materials
and Methods (a) Materials
wit.h the A sequence: model polypeptide H-(Acm-Cys)-(Lys-Lys-Leu-Leu)~-Lys-Zys-~4sn-Pro-G1~-(Leu-Leu-Lys-Lys)G-Tyr-NH, was synthesized on an ABI 430A peptide synthesizer employing the Boc-amino acid anhydride method. The side-chain protected Boc-amino acid derivatives used were Cys(Acm), Lys(ZCl-Z) and Tyr(2Br-Z). 4-Methylbenzhyrylamin resin was used as a starting material. The pept,ide resin obtained was treated with anhydrous hydrogen fluoride containing 10 “/0 anisole at 0°C for 75 min. The peptide was extracted with 65 M-acetic a,cid and freeze-dried. The crude peptide was applied to a preparative C,, h.p.1.c. column and the main fractions. eluted with a gradient of acetonitrile in 61 Y0 trifluoroacetic acid. were collected. The amino acid composition of the polypeptide was determined with an Trica amino acid analyzer, model A-5500 (Kyoto, Japan). The sample was hydrolyzed in sn evacuat,ed, sealed tube wit,h 6 X-HCl at 110 “C for 72 h. The Asp : Gly : Leu : Tyr : Lys : Pro composition was 1.1 : I.1 : 21.6 : I,3 : 23.8 : 1.1, being consistent with the ratio expected from the designed sequence: 1 : 1 : 22 : 1 : 24 : 1. However, the analgtma! h.p.1.c. pattern of the prepared polypeptide indicated that it consisted of several polypeptides with a sequence close to the designed one. We did not try to obtain the pure material in view of the purpose of the present research. (b) Methods (i) Circular dichrokm measwements All measurements in this work were carried out at 20 “(> with thermostatically control!ed cell holders. c.d. measurements were carried out with a Jasco spectropolarimeter. model J-50OA, equipped with an interface and a personal computer. The instruments were ammonium d-lo-camphorsulfonate calibrated wit.h (Takakuwa et al., 1985). The results are expressed as mean residue ellipticity [Ol, which is defined as [flj = in. 100 x e,,,,/(le), where dobsdis t,he observed ellipticity degrees, c is the concentration in residue mole/l, and 2 is the length of the light patb in cm. The c.d. spectra were measured, except where otherwise specified, at a protein concentration of 10 pM with a l-mm cell from 250 to 195 nm. The polypeptide concentration was determined from amino acid analysis. (ii) Conformational transitions Usually; 61 ml of the polypeptide soiution, dissolved in deionized water, was mixed with 0.9 ml of acid or salt solution. The acid and salt solutions were prepared before the measurements and used as soon as possible. The buffers used were sodium acetate (pH 3 to 65) Tris HCJ (pH 6.5 to 85), CHES (pH 8.5 to 9.5) and CAPS (pH 9.5 to 11). The solutions below pH 3 were prepared by using t’he appropriate concentration of HCl. pH was measured soon after spectroscopic measurement with a Radiometer PHM83 at 20°C. In the measurements with K,Fe(CX),, K,Fe(CX),, sodium triphosphate and sodium diphos-
Conformation of an Amphiphilic
Polypeptide
389
-5
* b x -10 5 ii 0 -15 z
,-
I
I
I
200
I
I
220 Wavelength
Figure 1. c.d. spectra function of sodium 10 mw-Tris’HCl buffer rhlorat,e concentrations: 4.9 mM: 5. 18 rnM. The
I
-2c
240 ( nm )
of the model polypeptide concentration perchlorate (pH 7.5) at 20°C. Sodium 1; 0 mM; 2,27 mM; 3, 63 polypept,ide concentration
as a in permM; was
10 /lM.
phate, the polypeptide showed a tendency to form aggregates at the later stage of the transition. The formation of aggregates resulted in disruption of the isosbestic point seen in Fig. 1. At higher concentrations of these salts, the solutions became turbid and a sudden decrease in c.d. intensity at 222 nm was observed. In cases where the maximal ellipticity could not be estimated because of aggregation, it was assumed to be the same as that for the transition induced by sodium perchlorate, in which we did not observe aggregation.
3. Results (a) Salt-induced conforma,tional transition Figure 1 shows the effect on the c.d. spectrum of adding sodium perchlorate to the model polypeptide in 10 mM-‘l’ris. HCI buffer at pH 7.5. In the absence of salt, the c.d. spectrum had a minimum at 200 nm,
showing that the polypeptide takes no significant secondary structure. Addition of sodium perchlorate changed the spectrum to one with minima at 208 and 222 nm, representing the formation of an a-helical conformation. The spectra at different concentrations of sodium perchlorate showed an isosbestic point at 203 nm, being consistent with a twostate folding mechanism. The maximal ellipticity at 222 nm was -21,000 deg cm’ dmol-‘. Figure 2 shows the sodium perchlorate-induced transition at different concentrations of the polypeptide measured by the change in ellipticity at 222 nm. The folding transition showed no significant dependence on the polypeptide concentration, as expected for a monomeric helical state. Using cd., we compared the effects of various salts on the polypeptide. Although the values of
0
I
100
IO Na Cl04
(mM
1
Figure 2. Dependence on polypeptide concentration of the sodium perchlorate-induced folding transition measured by the change in ellipticity at 222 nm in 10 rnti-Tris’ HCI buffer (pH 75) at 20°C. Polypeptide concentrations were 1.6 pM (circles), 7.9 PM (triangles) and 141 pM (squares). Cells with 5, 1 and @l-mm pathlengths were used for peptide concentrations of 1.6, 7.9 and 141 PM. respectively.
maximal c.d. intensity differed slightly depending on salt species, all of the salts examined induced similar a-helical states. However, as can be seen from Figure 3, the concentration range of salt required to bring about the transition varied greatly among the different sodium or potassium salts (Table 1). Because the transition induced by NaCl is the same as that by KCl, the large difference in the effects of various salts is due to the difference in the effects of anions. The order of effectiveness of anions is: triphosphate4= Fe(CN)z> Fe(CN)z> diphosphate3- > EDTA3- > phosphate’> SO:> trichloroacetate> Cl02 > trifluoroacetate> Cl- >F-. . . . . . . . . . . . . . . . . . . .(l) where the net charges are the values at pH 7.5: the net charges for phosphate, diphosphate, triphosphate and EDTA anions were calculated from their pK, values (Sillen & Martell, 197 1). As can be seen, an anion with higher valency is more effective than an anion with lower valency. Among the monovalent anions, chaotropic anions such as trichloroacetate or perchlorate are more effective than a kosmotropic anion, fluoride. This series is consistent with the order of anions that stabilize the molten globule state of acid-denatured proteins (see eqn (1) of Goto et aZ., 1990b), and also with the electroselectivity series of anions toward anion-exchange resins (Gregor et al., 1955; Gjerde et
Y. Goto and X. Aimoto
390
0.1
I Salt
IO
100
1000
( mM i
Figure 3. Foiding transitions of’ the model polypeptide induced by various salts in IO mPrr-Tris .HCl buffer (pN 7.5) at 20°C measured by the change in ellipticity at 222 nm. Salts used were KF (0). NaCl (A), KC1 (v); Na trifluoroacetatr (m). KaClO, (*),
(O), iYa trichloroacetate
K,Fe(C&
(O),
Sa,
(A)> Na,SO,
triphosphate
(V),
(O), Pja,H phosphate
K,Fe(CN),
al., 1980: see eqns (3) and (4) of Goto et al., 1990b). It should be noted that ferrocyanide anion, with a net charge of -4 at pH 7.5, is more effective than ferricyanide anion, with a net charge of -3. The order of effectiveness of these anions to stabilize the molten globule of cytochrome c or apomyoglobin at
Table 1 Salt or acid-induced folding transition. polypeptide at 20°C c,+
Salt or acid
bM)
Salt K,Fe(CIJ), Na, triphosphate K,Fe(CN)a Na, diphosphate Na,H, EDTA hTa,H phosphate Na,SO, Na trichloroacetate NaCIO, Na trifluoroacetate NaCl KC1 KF Acid H,SOh Trichloroacetic HC104 Trifluoroacetic HCl
of the model
oa20 @020 0.025 PO32 0.038
0529
acid
n.d.9 n.d.§ n.d.$ n.d.$ 1.3
1%
1.1 1.1 I.7
4.4
1.2
044
11
1.3
75 80
l-2 1.3 I.6
108 @28
acid
A4
2.1 54 22 103
1.1 1.6 1.6 1.6
1.8
Salt-induced transitions were measured in 10 mM-Tris HCI buffer at pH 7.5 using the change in ellipticity at 222 nm. t Midpoint concentration of transition. $ Preferential binding parameter of anion obtained from equation (5). 9 Not determined (n.d.) because we were unable to observe the later stage of the transitions because of aggregation.
(*).
Na,H,
EDTA
(C)? Ka, diphosphate
(+).
pH 2 was reversed because, at pH 2, the net charge of ferrocyanide ( -2.5) is less than that of ferricyanide (-3; Goto et al., 1990b). These results irtdicate that the electrostatic interactions between anions and the positively charged groups of Lhr model polypept,ide are responsible for the folding transition. (b) Acid-induced
cm~ormational
transitiorL
Goto et al. (1990a,b) showed that various strong acids also stabilize the molten globule state of aciddenatured proteins and that the effect’s of strong acids are comparable to the effects of corresponding salts at the same concentrat’ions. Figure 4 shows the cd. spectra of the x-helical conformation induced by perchloric acid. The c.d. spectrum in 144 rnlrperchloric acid agreed well with the spectrum in 144 mM-sodium perchlorate in 10 mrvl-Tris . HGl buffer at pH 7.5 (not shown), indicating t)he formation of an E-helical state very similar to that induced by sodium perchlorate. Various strong acids induced co-operative folding transitions. Figure 5 compares the transitions measured by the change in ellipticity at 222 nm. The order of effectiveness of various acids is consistent with equation (I) and the values of t’he midpoint concentration (c,,,) for the transitions induced by acids are similar to the values of G, for the corresponding salt-induced transitions (Table 1); although there is a tendency for the acid-induced transitions to require a slightly higher concentration of anion compared to the salt-induced transitions. (c) Alkalinknduced
con&wmational
transition
Provided that, the electrostatic repulsion is the major force of unfolding under low-salt conditions,
Conformation
-
of an Amphiphilic
Polypeptide
391
4
‘5 E Tl
;
2
$ D
P 0x
O
z -
-2 0
I
I
I
I
,
200
I
220
Wavelength
I
I
240 ( nm I
Figure 4. cd. spectra of the model polypeptide under different conditions at 20°C. The conditions were 1 mx-HCl at pH 3 (broken line), 144 miv-HClO, at pH 0.9 buffer at pH 9.6 (continuous line), or 10 mM-CHES (dotted line). deprotonation of lysyl residues under alkaline conditions would be expected to reduce the repulsion and, as a result, to favor the helical conformation. As shown in Figure 6(a), under conditions of low ionic strength, co-operative folding transition of the polypeptide was observed with increase in pH. The midpoint pH of the transition at I = 0001 was 8.0. The cd. spectrum at pH 9.6 wa,s very similar to that induced by perchloric acid (Fig. 4). Figure 6(a) also shows the pH-dependent transitions under different conditions of ionic strength measured by the change in ellipticity at 222 nm. Because high ionic strength induces the helical conformation at neutral and acidic pH values, the amplitude of the pa-induced folding transition decreased with an increase in ionic strength. However, the major transition occurred at pH values between 8 and 9 independent of ionic strength. At an ionic strength of 0.5, the amplitude of the pH-induced transition was very small. It is important to note that the conformational transition between the unfolded and molten globule states of apomyoglobin shows a similar dependence on ionic strength (see Fig. 1 of Goto & Fink, 1990). Figure 6(b) shows the NaCl-induced transitions at different pH values measured by the change in ellipticity at 222 nm. While the folding transition with a large amplitude was observed at acidic pH, the amplitude of the transition decreased significantly above pH 7.5 and no transition was observed at pH 9.5. These results are consistent with the pHdependent transition under different conditions of ionic strength (Fig. 6(a)).
4. Discussion (a) Conformation
of the a-helical
state
An anion-induced folding transition very similar to that of acid-denatured proteins was observed at
0.01
1
I
I
I
0.1
I
IO
100
Acld
I
I
1000
( mM )
Figure 5. Folding transitions
of the model polypeptide induced by various acids measured by the change in elliptieity at 222 nm at 20°C. Acids used were HCl (A), trifluoroacetic acid ( n ), HC104 ([7), trichloroacetic acid (A) and H,S04 (0). The ellipticity in the absence of a’cids was measured in distilled water at pH 5.
neutral pH for the amphiphilic polypeptide. The folding t,ransition induced by sodium perchlorate was independent of peptide concentration (Fig. 2), indicating that the helical conformation is monomeric. The maximal ellipticity at’ 222 nm induced by sodium perchlorate was - 21,000 deg cm2 dmolV’. On the basis of this value and the method of Chen et al. (1972), the helical content of the folded state is 60%. This value is smaller than that expected for the designed polypeptide. We do not know the precise conformation of the folded state: in particular, the conformation around the turn sequence. However, the global conformation is probably not so different from that which we initially designed, i.e. a monomeric conformation in which two amphiphilic helices interact with each other intramolecularly through hydrophobic Leu residues. This must be a conformational state equivalent to the molten globule of acid-denatured prosince the mechanism of conformational teins, transition of the polypeptide is the same as that of the acid-denatured protein. While proteins in the molten globule state can fold up further to a more compact native structure, based on the precise packing of side-chains, the molten globule-like structure is the ultimate folded conformation for the model polypeptide because it does not show such close packing as that of natural proteins obtained during the course of evolution. (b) Role of anions in the conformational
trawdion
The effectiveness of various anions in inducing the helical state varied greatly, and the series obtained (eqn (1)) is consistent with the electroselectivity series of anions toward anion-exchange resins. The conformation of the model polypeptide is assumed to be determined by the balance of repulsive forces between protonated amino groups and the opposing
Y. Goto and 8. Aimoto
392
6
8
IO
PH (a)
I 100 NaCl
I 1000
( mM 1
Figure 6. Dependencies of the ellipticity at 222 run of the model polypeptide on (a) pH at different ionic strength, and (b) on h’aC1 concentration at different pH values? at PO”C. The values of ionic strength in (a) were 0.001 (0). @Ol (A), @05 (0). @1 (0) and 0.5 (O), where ionic strength of the buffer component was 001. except for the case of an ionic strength of 0.001, and the total ionic strength was controlled by KaCl. Filled circles represent the HCI-induced transition taken from Fig. 5, where the ionic st,reugth is minimal at each pH. The pH vaiues in (b) were 2.1 (10 mLw-HC1: 0). 4.4 (10 mM-Ka acetate: a). 68 (10 rnx-Tris.HCI: I-J), 8.0 (10 mM-‘fris.HCl: O), 8.3 (10 rnM-Tris.HCl: 0) and 95 (10 mh&!HES: 0).
(b) mainly weak hydrophobic interactions, forces, which stabilize the folded state. As described in detail by Goto et al. (1990b), anions can induce the folding transition by three possible mechanisms: shielding of the repulsive forces by (1) the Debye-Hiickel screening effect or (2) interaction with positive charges by anion binding, or (3) increasing the hydrophobic interactions of the polypeptide by affecting the water structure. The consistency of equation (1) with the electroselectivity series (eqns (3) and (4) of Goto et al. (1990b)) clearly indicates that, as is the case for acid-denatured proteins, electrostatic anion binding is the major factor responsible for the folding transition. Agreement of the acid-induced transitions of the model polypeptide with the corresponding salt induced transitions further confirms this conclusion.
Thus, the conformational transition of polypeptides induced by anion binding is not limited to aciddenatured proteins, but can take place even at neutral pH, provided that the polypeptide is basic and that the polypeptide has the potential to adopt an a-helical conformation when the electrostatic repulsion is reduced.
(c) Phase
diagram
of the conformationat
transition.
The co-operative folding transition was aiso induced when the pH was increased above 7.5. The cd. spectrum of the helical conformation at pH 9% (Fig. 4) was very similar to that induced by salt or acid. This indicates that the deprotonation of amino groups reduces the electrostatic repulsion and, as a
Conformation
2
of an Amphiphdic
4
6
Polypeptide
393
8
IO
PH
Figure 7. Phase diagram for the unfolded (U) and folded (F) helical states of the model polypeptide at 2O’C. The contour lines corresponding to 10 %, 30 %, 50 %, 70 %, 80 o/0 and 90 O/’ progress of the folding transition were coneltructed from the ?l’aCl-induced transitions at different pH values (circles), the pa-induced transitions at different ionic strengths (triangles). and the HCl-induced transition (squares). The ionic strength of the buffer component was less than 0.01 and the total ionic strength was controlled by NaCl. Hatched area is prohibited owing to the increase in the minimal ionic strength with decrease in pH.
result, the polypeptide takes a conformation similar to that induced by anion binding. As a first approximation, we assumed that the helical conformation induced at alkaline pH is the same as that induced by anion binding at neutral and acidic pH values. Then, we constructed a phase diagram for the pHand ionic strength-dependent conformational states on the basis of the pH-, of the model polypeptide NaCl- and HCl-induced conformational transitions (Fig. 7). The unfolded state is stable below pH 8 under conditions of low ionic strength. The left-hand boundary of the phase diagram corresponds to the ionic strength of HCl alone at the given pH and the hatched area is prohibited. The contour lines are relatively insensitive to pH below pH 7. Above pH 7, they show hyperbolic curvature toward the region of low ionic strength. Recently, Goto & Fink (1990) have reported a phase diagram for the acidic conformational states of apomyoglobin. An important point is that the phase diagram of the model polypeptide is very similar in shape to the phase diagram for the transition between the unfolded and molten globule states of apomyoglobin. We have been studying the pHand salt-dependent conformational transition of melittin. The phase diagram for the conformational states of melittin is also similar to that of the model polypeptide (Y. Goto, unpublished results). These results suggest that, not only the conformational stability of the model polypeptide, but also the conformation of many polypeptides, including aciddenatured proteins and melittin, are determined by a similar mechanism. Therefore, to elucidate the mechanism of protein folding, it is of importance to
know why the phase diagram shown in Figure 7. (d) Anion-
has a shape like that
and pH-dependent transition
conformational
To consider the effects of pH and anion on the conformational stability of the model polypeptide, we assume that the linked reactions of proton and anion bindings to an amino group and the conformational transition are represented by Mechanism 1: FA
Mechanism
1
where U and F represent the unfolded and folded states, respectively, species with the superscript H represent the protonated species, species with the subscript A represent the species with anion binding, and KFH, KiH, KtH, etc. represent the equilibrium constants of the respective processes indicated by arrows.
Y. Got0 and 8. Bimolo
394
From constant,
Mechanism II; the apparent equilibrium of the folding transition is defined as:
By using the equilibrium can be expressed as:
constants,
equation
(2)
+ KEH/[H] + KgH/[H]
+ K:[A]K;H/[H]) + K;[A]KgH/[H]
H
H
reactions
in
F
KUHKF!, = KUHKUHA F
’
(3)
Representation of the mechanism involving multiple (n) binding sites is complicated. However, provided that n sites are independent, the apparent equilibrium constant is represented simply by multiplying the contribution of the respective binding sites: @=”
of the linkage of the 1, i.e.: K”HKFH = KUHKt!
and:
x (1 + Kp[A] 1 + KyH[A]
=
The former corresponds to a decrease value upon folding and the latter to an the binding constant for anion upon
F
Kapp = KFH
K aPP
respectively. in the pK, increase in folding. Because Mechanism
fi i=l
(I+ K:;[Ai+ ’ (1 + K:“[A]+
K:;/[H] K:“/[H]
+ K:JA]K;y/[H]) + K: KUH H H KFH > KU”. A A
A
A
equilibria
F
3
must be:
KUH < KUF F and: Ku”
< KUHA
F
F
:
indicating that the conformational equilibrium shifts to stabilize the folded state upon either deprotonation or anion binding. Although we cannot- estimate the contribution of electrostatic interactions quant,itatively, the reduction of the electrostatic repulsive forces must be the major factor responsible for the change in conformational equilibrium upon deprotonation or anion-binding. To simula.te the conformational transition, we now assume two extreme cases. In one case, proton and anion bindings are independent (i.e. t,he independent binding of proton and anion): KIT = di:y>
Kl;i, = KY”, KF” H, HZ = KF”” and: Ku” = KUHA H, H,
(5)
where a is the activity of the ligand (Tanford, 1970). From the alkaline transition at, an ionic strength of 0.001, An is estimated to be - 1.8, suggesting that the number of amino groups responsible for the two. Analysis of the transition is at least SaCl-induced transition using molarity instead of activit’y gave a value of 1.3, suggesting that the number of binding sites for anions responsible for the transition is at least two. We also analyzed the folding transitions induced by various salts and acids and the results are shown in Table 1. Thus, while the anion-induced transition can be explained in terms of the preferential binding to the folded state compared with the binding to the extended unfolded state, the alkaline-induced transition is explained in terms of the preferential deprotonation of the folded state compared with the deprotonation of the unfolded state. Here, it would be useful to illustrate the property of one binding site in some detail using Mechanism 1. The preferential interactions of proton or anion with the polypeptide are represented by: and:
the eonformational
Then, equation
(4) becomes:
In another case, anions bind only to the protonated binding sites (i.e. the dependent bindings of proton and anion). In ot.her words t,he species rjl, and i,?, are absent from Mechanism 1. Equat’ion (4) reduces to: KdeP ~PP-
KUHn F
We do not know the exact number of amino groups responsible for the conformational transition. As the simplest case; we assumed that, on the basis of the preferential binding parameters obtained, only two amino groups are important and that these two groups are equivalent,. We estimated the values of KzH2, KIHH, KiH, KgH and KgH to be 0.01, 10031--“, 1 M-l, 10~7~s M. and 10-9.5 M respectively, so that t,he calculateb curves fit both the pH-induced transition curve at the low-est ionic strength (I = 0.001) and t,he NaCl-induced transition curves at acidic pH va.lues. It should be mentioned that there are so many parameters to
Conformation
0
0
2
4
6
8
of an Amphiphilic
IO
PH
0
2
4
Polypeptide
"0
2
395
4
6
ii----
6
8
PH
6
8
IO
PH
IO
PH
Figure 8. Simulation of the phase diagram (a, c) and the pH-induced conformational transition (b. d), on the basis of the dependent bindings (a. b) or independent (c: d) bindings of protons and anions to the 2 equivalent amino groups of the polypeptide (see the text.). In (a) and (c), contour lines corresponding to lo%, 30%, SO%, 70%, 80% and 90% progress of the folding transition are shown. In (b) and (d), folding transition curves at ionic strengths of @OOl: 0.01, @05, 0.1 and 0.5 are shown. Broken line shows the calculated folding transition curve induced by HCl.
determine that the set of values used is not unique, although they account for most of the observed conformational transitions. Figure 8(a) and (c) shows the phase diagrams calculated based on the assumptions of dependent binding (eqn (7)) and independent binding (eqn (6)), respectively. While the two phase diagrams agree with each other at zero ionic strength (i.e. on the pH axis) and at acidic pH below 5, they disagree largely between these two conditions. The contour lines based on the dependent bindings show a hyperbolic curvature at pH above 7 toward the region of low ionic strength, intersecting the pH axis with a steep angle (close to 90”). On the other hand, the contour lines based on the independent bindings show a sigmoidal curvature above pH 7. The differences in the shape of the phase diagrams are independent of the values of the parameters chosen or the number of responsible amino groups. It is evident that the shape of the observed phase diagram is similar to that obtained on the basis of the assumption of dependent bindings of proton and anion. Figure 8(b) and (d) shows the pH-induced transitions under different conditions of ionic strength, simulated on the basis of the assumptions of dependent and independent bindings, respectively. The transition curves calculated for the two assumptions are largely different. The pH region where co-operative conformational transition occurs shifts significantly to the lower pH region with an increase in ionic strength when the assumption of independent
binding is used. On the other hand, it is insensitive to ionic strength when the assumption of dependent binding is used. The observed transition curves (Fig. 6(a)) are similar to the curves based on the assumption of dependent binding. Thus, both the phase diagram and transition curves calculated indicate that the anion- and pHdependent conformational stability of the model polypeptide can be explained satisfactorily provided that anions are assumed to bind only to the protonated amino groups. This is what would be expected from the electrostatic nature of the interactions and substantiates the mechanism of how the conformational stability of the model polypeptide is determined by both salt and pH conditions. (e) Concludi~ng remarks Electrostatic interactions among the charged groups are an important factor in determining the conformation and stability of proteins (Matthew, 1985; Harvey, 1989). However, it is generally considered that they make a relatively small contribution compared the with contributions of hydrophobic interactions or hydrogen bonds (Hollecker & Creighton, 1982; Pace et al., 1990; Pace, 1990). However, present results anld those obtained by Goto et al. (1990a,b) show that the electrostatic repulsive forces are critically important in determining the conformation of acid-denatured proteins and the model polypeptide, in which the
Y. Goto and X. Aimoto
396
stability of the folded conformational states is relatively low. The anion- and pa-dependent folding transitions of the model polypeptide can be explained simply by the linked reactions of the conformational transition and proton and anion bindings to a limited number of amino groups, in which anions bind only to the protonated groups. The mechanism may be fundamental to the conformational stability of the acidic molten globule state and basic peptide such as melittin. We thank Professor Kozo Hamaguchi, Osaka Universit’y, for stimulating discussions. This work was supported in part by a Grant-in-Aid for Scientific Research from the Ministry of Educat.ion, Science and Culture of Japan. References Chen, Y.-H., Yang, J. T. & Marinez, H. M. (1972). Determination of the Secondary Structures of Proteins by Circular Dichroism and Optical Rotatory Dispersion. Biochemistry, 11, 4120-4131. De Grado, W. F. & Lear, J. D. (1985). Induction of Peptide Conformation at Apolar/Water Interfaces. J. Amer. Chem. Sot. 107, 7684-7689. Ebert, G. b Kuroyanagi, Y. (1982). Salt, Effects on the Conformation of a “Statistical” Copolymer of L-Leutine and L-Lysine. Polymer, 23, 1147-1153. Fink, A. L., Calciano, L. J., Goto, Y. & Palleros, D. R. (1990). Acid-denatured States of Proteins. In Current Research in Proteins Chemistry, (Villafranca, J. J., ed.), pp. 417-424, Aca.demic Press, San Diego. Gjerde, D. T., Schmuchler, G. & Fritz, J. S. (1980). Anion Chromatography with Low-conductivity Eluents. II. J. Chromatoyr. 187, 35-45. Goto, Y. & Fink, 8. L. (1989). Conformational States of. Beta-Lactamase: Molten-Globule States at Acidic and Alkaline pH with High Salt. Biochemistry 28, 945-952. Goto, Y. & Fink, A. L. (1990). Phase Diagram for Acidic Conformational States of Apomyoglobin. J. Mol. Biol. 214, 803-805. Goto, Y.? Calciano, L. J. & Fink, A. L. (1990a). Acid-induced Folding of Proteins. Proc. Nat. Acad. Xci., U.S.A. 87, 573-577. Edited
Goto,
Y., Takahashi, N. & Fink, A. I[,. (199%). Mechanism of Acid-induced Folding of Proteins. Biochemistry, 29, 3480-3488. Gregor. H. P.? Belle, J. & Marcus, R. A. (1955). Studies on Ion Enchange Resins. XIII. Selectivity Coefficients of Quaternary Base Anion-exchange Resins Toward Univalent Anions. J. Amer. Chem. Sot. 77,
2713-2719. Harvey, S. C. (1989). Treatment of Electrostatic Effects in Macromolecular Modeling. Proteins, 5, 78-92. Holleeker. M. & Creighton, T. E. (1982). Effect on Protein Stabilit’y of Reversing the Charge on Amino Groups. Biochim. Biophys. Acta, 701, 395-404. Kuwajima, K. (1989). The Molten Globule State as a Clue for Cnderstanding the Folding and Co-operativity of Globular Protein Structure. Proteins, 6, 87-103. Matthew, J. B. (1985). Electrostatic Effects in Proteins. Awnu. Rev. Biophys. Biophys. Chem. 14, 387-417. Ohgushi, M. & Wada, A. (1983). ‘Molten-Globule State’: A Compact Form of Globular Proteins with Mobile Side Chains. FEBS Letters, 164, 21-24. Pace, C. N. (1990). Conformational Stability of Globular Proteins. Trends Biochem. Sci. 15, 14-17. Pace, C. N., Laurents, D. V. & Thomson, J. A. (1990). pH Guanidine Urea and Dependence of the Hydrochloride Dena,turation of Ribonuclease A and Ribonuclease Tl. Biochemistry, 29, 2564-2572. Hypotheses and Ptitsyn, 0. B. (1987). Protein Folding: Experiments. J. Protein Chem. 6: 273-293. Ptitsyn, 0. B., Pain, R. H., Semisotnov, G. V.; Zerovn.ik; E. & Razgulyaev, 0. I. (1990). Evidence for a Molten Globule State as a General Intermediate in Protein Folding. FEBS Letters, 262. 20-24. Sillen, L. G. & Martell; A. E. (1971 j. StabiEity Constants of Il/letal-zon Com$exes, Supplement No. 1, AidPn Press, Oxford. Talbot. J. C., Dufourcq, J., deBony, J., Faucon, J. F. & Lussan: C. (1979). Conforma.tional Change and Self Association of Monomeric Melittin. FEBS Letters, 102; 191-193. Takakuwa, T., Konno, T. & Meguro, H. (1985). A jSl’ew Standard Substance for Calibration of Circular Ammonium Dichroism: d- lo-Camphorsulfonate. Anal. Sei. 1, 215-218. Tanford, C. (1970). Protein Denaturation. Part C. Models Mechanism of Theoretical for the Denaturation. Advan. Protein Chem. 24. l-95.
by P. E’. Wriyht
(1991) 218; 387-396
Anion and pH-dependent Conformational Amphiphilic Polypeptide
Transition
of an
Yuji Gotot Department of Biology, Faculty of Science Osaka University, Toyonaka, Osaka 560, Japan
and Saburo Aimoto Institute for Protein Research Osaka University, Suita, Osaka 565, Japan (Received 3 August
1990; accepted 9 October 1990)
While several proteins, including /?-lactamase, cytochrome c and apomyoglobin, are maximally unfolded at pH 2 by HCl in the absence of salt, the addition of anions, either from salt or acid, co-operatively induces the unfolded proteins to refold to a molten globule state, because anions bind preferentially to the compact molten globule state compared to the extended unfolded state. To study the role of the anion-dependent conformational transition at neutral pH, we synthesized a model polypeptide of 51 amino acid residues, consisting of tandem repeats of a Lys-Lys-Leu-Leu sequence and containing a turn sequence, Asn-Pro-Gly, at the center of the molecule. The model polypeptide showed no significant conformation by circular dichroism under conditions of low salt at neutral pH. However, addition of anions, either from salt or acid, induced the folding transition to an a-helical conformational state. The order of effectiveness of various anions in inducing the folding transition was consistent with the series of anions in inducing the molten globule of the acid-denatured protein. This suggests that the helical state of the model polypeptide is equivalent to the molten globule state. At pH values above 9, the model polypeptide also took an a-helical conformation, which was very similar to that induced by anions. On the basis of the chloride and pa-dependent conformational transitions, a phase diagram for the conformational states was constructed. The phase diagram was explained simply by assuming that the conformational transition is linked to the proton and the anion bindings to a limited number of amino groups and that anions bind only to the protonated groups.
1. Introduction
mediate conformation upon addition of anions, either from salt or acid. The intermediate conformation is compact and contains a significant amount of secondary structure, but has a disordered tertiary structure. The intermediate conformation is called a molten globule and is assumed to be a major intermediate of protein folding (Ohgushi & Wada, 1983; Ptitsyn, 1987; Kuwajima, 1989; Ptitsyn et al., 1990). Goto et al. (1990b) examined the salt and aciddependent formation of the molten globule of cytochrome c and apomyoglobin and indicated that the mechanism involves electrostatic interaction of anions with the positively charged groups of the proteins. The conformation of acid-denatured proteins is determined by the balance of the repulsive forces between positive residues, which favor the extended conformation, and mainly hydrophobic forces, which favor the compact molten globule
The conformation of acid-denatured proteins, including /I-lactamase, cytochrome c and apomyoglobin, depends critically on ionic conditions (Goto & Fink, 1989, 1990; Fink et al., 1990; Goto et al., 1990a,b). While these proteins are largely unfolded by HCl at low ionic strength to a conformation similar to that obtained with high concentrations of Gdn-HClf, they refold co-operatively to an intert Author to whom all correspondence should be addressed. 1 Abbreviations used: Gdn-HCl, guanidine hydrochloride; h.p.l.c., high-pressure liquid chromatography; Acm, acetamidomethyl; Boc, t-butyloxycarbonyl; CAPS, 3-(cyclohexylamino)1-propanesulfonic acid; c.d., circular dichroism; CHES, 2-(N-cyclohexylamino)-ethanesulfonic acid; Z, carbobenzoxy.
387 0022~2836/91/060387-10
$03.00/O
0
1991 Academic
Press Limited
Y. Goto and S. Aimoto
388
conformation. The preferential binding of anions to a limited number of basic groups of the compactly folded molten globule state, compared to the extended unfolded state, results in the co-operative refolding transition with an increase in the concentration of anion. Alternatively, the shielding of intramolecular electrostatic repulsive forces in the unfolded state by the anion binding reflects the intrinsic forces that favor the formation of the molten globule. These results indicate the importance of electrostatic interaction in determining the conformational stability of acid-denatured proteins. Such anion-induced conformational transition may not necessarily be limited to acid-denatured proteins, but may have an important role even at neutral pH in determining the conformational stability of proteins and peptides. For example, the conformation of melittin, a bee venom toxin consisting of 26 amino acid residues and containing five basic residues but no acidic residues, depends on salt conditions (Talbot et al., 1979). While melittin is largely unfolded under conditions of low salt’ at pH 75, it assumes a tetramerie helical structure in the presence of a high salt concentration. As another example, Ebert & Kuroyanagi (1982) reported that the conformation of a statistical copolymer of Lys and Leu at a molar ratio of 1 : 1 at pH 7.0 depends on salt conditions (see also De Grado & Lear, 1985). While the polymer is unfolded under conditions of low salt, it assumes a helical conformation in the presence of a high salt concentration. The order of effectiveness of several anions in inducing the coilto-helix transition of the copolymer is consistent with the series of anions in inducing the refolding transition of acid-denatured proteins. These saltinduced conformational transitions at neutral pH may be explained by the same mechanism as that used for the conformational transition of aciddenatured proteins. It is possible that such electrostatic interactions and consequent conformational changes play an important role in the function of some proteins. To understand the anion-dependent formation of the molten globule further, and to clarify the role of this conformational transition at neutral pH, we synthesized a model polypeptide, which consisted of tandem repeats of a Lys-Lys-Leu-Leu sequence (De Grado & Lear, 1985) and contained a turn sequence at the center of the molecule. W’e designed the polypeptide assuming that, while unfolded in the absence of salt at neutral pH due to the electrostatic repulsion between the amino groups, it would form a folded conformation in the presence of salt, in which two amphiphilic helices interact with each other intramolecularly through hydrophobic Leu residues. We report here that the model polypeptide shows an anion-dependent conformational transition very similar to that of acid-denatured proteins, demonstrating tha,t the interactions between anions and basic groups of polypeptides and consequent conformational
changes can occur even at neutral
also show that t.he anion- and pa-dependent
pH. We
forma-
tion of a molten globule-like conformation of the model polypeptide can be explained simply by assuming the linkage of conformational transit’ion with proton and anion bindings to a limited number of amino groups, in which anions bind only to the protonated groups.
2. Materials
and Methods (a) Materials
wit.h the A sequence: model polypeptide H-(Acm-Cys)-(Lys-Lys-Leu-Leu)~-Lys-Zys-~4sn-Pro-G1~-(Leu-Leu-Lys-Lys)G-Tyr-NH, was synthesized on an ABI 430A peptide synthesizer employing the Boc-amino acid anhydride method. The side-chain protected Boc-amino acid derivatives used were Cys(Acm), Lys(ZCl-Z) and Tyr(2Br-Z). 4-Methylbenzhyrylamin resin was used as a starting material. The pept,ide resin obtained was treated with anhydrous hydrogen fluoride containing 10 “/0 anisole at 0°C for 75 min. The peptide was extracted with 65 M-acetic a,cid and freeze-dried. The crude peptide was applied to a preparative C,, h.p.1.c. column and the main fractions. eluted with a gradient of acetonitrile in 61 Y0 trifluoroacetic acid. were collected. The amino acid composition of the polypeptide was determined with an Trica amino acid analyzer, model A-5500 (Kyoto, Japan). The sample was hydrolyzed in sn evacuat,ed, sealed tube wit,h 6 X-HCl at 110 “C for 72 h. The Asp : Gly : Leu : Tyr : Lys : Pro composition was 1.1 : I.1 : 21.6 : I,3 : 23.8 : 1.1, being consistent with the ratio expected from the designed sequence: 1 : 1 : 22 : 1 : 24 : 1. However, the analgtma! h.p.1.c. pattern of the prepared polypeptide indicated that it consisted of several polypeptides with a sequence close to the designed one. We did not try to obtain the pure material in view of the purpose of the present research. (b) Methods (i) Circular dichrokm measwements All measurements in this work were carried out at 20 “(> with thermostatically control!ed cell holders. c.d. measurements were carried out with a Jasco spectropolarimeter. model J-50OA, equipped with an interface and a personal computer. The instruments were ammonium d-lo-camphorsulfonate calibrated wit.h (Takakuwa et al., 1985). The results are expressed as mean residue ellipticity [Ol, which is defined as [flj = in. 100 x e,,,,/(le), where dobsdis t,he observed ellipticity degrees, c is the concentration in residue mole/l, and 2 is the length of the light patb in cm. The c.d. spectra were measured, except where otherwise specified, at a protein concentration of 10 pM with a l-mm cell from 250 to 195 nm. The polypeptide concentration was determined from amino acid analysis. (ii) Conformational transitions Usually; 61 ml of the polypeptide soiution, dissolved in deionized water, was mixed with 0.9 ml of acid or salt solution. The acid and salt solutions were prepared before the measurements and used as soon as possible. The buffers used were sodium acetate (pH 3 to 65) Tris HCJ (pH 6.5 to 85), CHES (pH 8.5 to 9.5) and CAPS (pH 9.5 to 11). The solutions below pH 3 were prepared by using t’he appropriate concentration of HCl. pH was measured soon after spectroscopic measurement with a Radiometer PHM83 at 20°C. In the measurements with K,Fe(CX),, K,Fe(CX),, sodium triphosphate and sodium diphos-
Conformation of an Amphiphilic
Polypeptide
389
-5
* b x -10 5 ii 0 -15 z
,-
I
I
I
200
I
I
220 Wavelength
Figure 1. c.d. spectra function of sodium 10 mw-Tris’HCl buffer rhlorat,e concentrations: 4.9 mM: 5. 18 rnM. The
I
-2c
240 ( nm )
of the model polypeptide concentration perchlorate (pH 7.5) at 20°C. Sodium 1; 0 mM; 2,27 mM; 3, 63 polypept,ide concentration
as a in permM; was
10 /lM.
phate, the polypeptide showed a tendency to form aggregates at the later stage of the transition. The formation of aggregates resulted in disruption of the isosbestic point seen in Fig. 1. At higher concentrations of these salts, the solutions became turbid and a sudden decrease in c.d. intensity at 222 nm was observed. In cases where the maximal ellipticity could not be estimated because of aggregation, it was assumed to be the same as that for the transition induced by sodium perchlorate, in which we did not observe aggregation.
3. Results (a) Salt-induced conforma,tional transition Figure 1 shows the effect on the c.d. spectrum of adding sodium perchlorate to the model polypeptide in 10 mM-‘l’ris. HCI buffer at pH 7.5. In the absence of salt, the c.d. spectrum had a minimum at 200 nm,
showing that the polypeptide takes no significant secondary structure. Addition of sodium perchlorate changed the spectrum to one with minima at 208 and 222 nm, representing the formation of an a-helical conformation. The spectra at different concentrations of sodium perchlorate showed an isosbestic point at 203 nm, being consistent with a twostate folding mechanism. The maximal ellipticity at 222 nm was -21,000 deg cm’ dmol-‘. Figure 2 shows the sodium perchlorate-induced transition at different concentrations of the polypeptide measured by the change in ellipticity at 222 nm. The folding transition showed no significant dependence on the polypeptide concentration, as expected for a monomeric helical state. Using cd., we compared the effects of various salts on the polypeptide. Although the values of
0
I
100
IO Na Cl04
(mM
1
Figure 2. Dependence on polypeptide concentration of the sodium perchlorate-induced folding transition measured by the change in ellipticity at 222 nm in 10 rnti-Tris’ HCI buffer (pH 75) at 20°C. Polypeptide concentrations were 1.6 pM (circles), 7.9 PM (triangles) and 141 pM (squares). Cells with 5, 1 and @l-mm pathlengths were used for peptide concentrations of 1.6, 7.9 and 141 PM. respectively.
maximal c.d. intensity differed slightly depending on salt species, all of the salts examined induced similar a-helical states. However, as can be seen from Figure 3, the concentration range of salt required to bring about the transition varied greatly among the different sodium or potassium salts (Table 1). Because the transition induced by NaCl is the same as that by KCl, the large difference in the effects of various salts is due to the difference in the effects of anions. The order of effectiveness of anions is: triphosphate4= Fe(CN)z> Fe(CN)z> diphosphate3- > EDTA3- > phosphate’> SO:> trichloroacetate> Cl02 > trifluoroacetate> Cl- >F-. . . . . . . . . . . . . . . . . . . .(l) where the net charges are the values at pH 7.5: the net charges for phosphate, diphosphate, triphosphate and EDTA anions were calculated from their pK, values (Sillen & Martell, 197 1). As can be seen, an anion with higher valency is more effective than an anion with lower valency. Among the monovalent anions, chaotropic anions such as trichloroacetate or perchlorate are more effective than a kosmotropic anion, fluoride. This series is consistent with the order of anions that stabilize the molten globule state of acid-denatured proteins (see eqn (1) of Goto et aZ., 1990b), and also with the electroselectivity series of anions toward anion-exchange resins (Gregor et al., 1955; Gjerde et
Y. Goto and X. Aimoto
390
0.1
I Salt
IO
100
1000
( mM i
Figure 3. Foiding transitions of’ the model polypeptide induced by various salts in IO mPrr-Tris .HCl buffer (pN 7.5) at 20°C measured by the change in ellipticity at 222 nm. Salts used were KF (0). NaCl (A), KC1 (v); Na trifluoroacetatr (m). KaClO, (*),
(O), iYa trichloroacetate
K,Fe(C&
(O),
Sa,
(A)> Na,SO,
triphosphate
(V),
(O), Pja,H phosphate
K,Fe(CN),
al., 1980: see eqns (3) and (4) of Goto et al., 1990b). It should be noted that ferrocyanide anion, with a net charge of -4 at pH 7.5, is more effective than ferricyanide anion, with a net charge of -3. The order of effectiveness of these anions to stabilize the molten globule of cytochrome c or apomyoglobin at
Table 1 Salt or acid-induced folding transition. polypeptide at 20°C c,+
Salt or acid
bM)
Salt K,Fe(CIJ), Na, triphosphate K,Fe(CN)a Na, diphosphate Na,H, EDTA hTa,H phosphate Na,SO, Na trichloroacetate NaCIO, Na trifluoroacetate NaCl KC1 KF Acid H,SOh Trichloroacetic HC104 Trifluoroacetic HCl
of the model
oa20 @020 0.025 PO32 0.038
0529
acid
n.d.9 n.d.§ n.d.$ n.d.$ 1.3
1%
1.1 1.1 I.7
4.4
1.2
044
11
1.3
75 80
l-2 1.3 I.6
108 @28
acid
A4
2.1 54 22 103
1.1 1.6 1.6 1.6
1.8
Salt-induced transitions were measured in 10 mM-Tris HCI buffer at pH 7.5 using the change in ellipticity at 222 nm. t Midpoint concentration of transition. $ Preferential binding parameter of anion obtained from equation (5). 9 Not determined (n.d.) because we were unable to observe the later stage of the transitions because of aggregation.
(*).
Na,H,
EDTA
(C)? Ka, diphosphate
(+).
pH 2 was reversed because, at pH 2, the net charge of ferrocyanide ( -2.5) is less than that of ferricyanide (-3; Goto et al., 1990b). These results irtdicate that the electrostatic interactions between anions and the positively charged groups of Lhr model polypept,ide are responsible for the folding transition. (b) Acid-induced
cm~ormational
transitiorL
Goto et al. (1990a,b) showed that various strong acids also stabilize the molten globule state of aciddenatured proteins and that the effect’s of strong acids are comparable to the effects of corresponding salts at the same concentrat’ions. Figure 4 shows the cd. spectra of the x-helical conformation induced by perchloric acid. The c.d. spectrum in 144 rnlrperchloric acid agreed well with the spectrum in 144 mM-sodium perchlorate in 10 mrvl-Tris . HGl buffer at pH 7.5 (not shown), indicating t)he formation of an E-helical state very similar to that induced by sodium perchlorate. Various strong acids induced co-operative folding transitions. Figure 5 compares the transitions measured by the change in ellipticity at 222 nm. The order of effectiveness of various acids is consistent with equation (I) and the values of t’he midpoint concentration (c,,,) for the transitions induced by acids are similar to the values of G, for the corresponding salt-induced transitions (Table 1); although there is a tendency for the acid-induced transitions to require a slightly higher concentration of anion compared to the salt-induced transitions. (c) Alkalinknduced
con&wmational
transition
Provided that, the electrostatic repulsion is the major force of unfolding under low-salt conditions,
Conformation
-
of an Amphiphilic
Polypeptide
391
4
‘5 E Tl
;
2
$ D
P 0x
O
z -
-2 0
I
I
I
I
,
200
I
220
Wavelength
I
I
240 ( nm I
Figure 4. cd. spectra of the model polypeptide under different conditions at 20°C. The conditions were 1 mx-HCl at pH 3 (broken line), 144 miv-HClO, at pH 0.9 buffer at pH 9.6 (continuous line), or 10 mM-CHES (dotted line). deprotonation of lysyl residues under alkaline conditions would be expected to reduce the repulsion and, as a result, to favor the helical conformation. As shown in Figure 6(a), under conditions of low ionic strength, co-operative folding transition of the polypeptide was observed with increase in pH. The midpoint pH of the transition at I = 0001 was 8.0. The cd. spectrum at pH 9.6 wa,s very similar to that induced by perchloric acid (Fig. 4). Figure 6(a) also shows the pH-dependent transitions under different conditions of ionic strength measured by the change in ellipticity at 222 nm. Because high ionic strength induces the helical conformation at neutral and acidic pH values, the amplitude of the pa-induced folding transition decreased with an increase in ionic strength. However, the major transition occurred at pH values between 8 and 9 independent of ionic strength. At an ionic strength of 0.5, the amplitude of the pH-induced transition was very small. It is important to note that the conformational transition between the unfolded and molten globule states of apomyoglobin shows a similar dependence on ionic strength (see Fig. 1 of Goto & Fink, 1990). Figure 6(b) shows the NaCl-induced transitions at different pH values measured by the change in ellipticity at 222 nm. While the folding transition with a large amplitude was observed at acidic pH, the amplitude of the transition decreased significantly above pH 7.5 and no transition was observed at pH 9.5. These results are consistent with the pHdependent transition under different conditions of ionic strength (Fig. 6(a)).
4. Discussion (a) Conformation
of the a-helical
state
An anion-induced folding transition very similar to that of acid-denatured proteins was observed at
0.01
1
I
I
I
0.1
I
IO
100
Acld
I
I
1000
( mM )
Figure 5. Folding transitions
of the model polypeptide induced by various acids measured by the change in elliptieity at 222 nm at 20°C. Acids used were HCl (A), trifluoroacetic acid ( n ), HC104 ([7), trichloroacetic acid (A) and H,S04 (0). The ellipticity in the absence of a’cids was measured in distilled water at pH 5.
neutral pH for the amphiphilic polypeptide. The folding t,ransition induced by sodium perchlorate was independent of peptide concentration (Fig. 2), indicating that the helical conformation is monomeric. The maximal ellipticity at’ 222 nm induced by sodium perchlorate was - 21,000 deg cm2 dmolV’. On the basis of this value and the method of Chen et al. (1972), the helical content of the folded state is 60%. This value is smaller than that expected for the designed polypeptide. We do not know the precise conformation of the folded state: in particular, the conformation around the turn sequence. However, the global conformation is probably not so different from that which we initially designed, i.e. a monomeric conformation in which two amphiphilic helices interact with each other intramolecularly through hydrophobic Leu residues. This must be a conformational state equivalent to the molten globule of acid-denatured prosince the mechanism of conformational teins, transition of the polypeptide is the same as that of the acid-denatured protein. While proteins in the molten globule state can fold up further to a more compact native structure, based on the precise packing of side-chains, the molten globule-like structure is the ultimate folded conformation for the model polypeptide because it does not show such close packing as that of natural proteins obtained during the course of evolution. (b) Role of anions in the conformational
trawdion
The effectiveness of various anions in inducing the helical state varied greatly, and the series obtained (eqn (1)) is consistent with the electroselectivity series of anions toward anion-exchange resins. The conformation of the model polypeptide is assumed to be determined by the balance of repulsive forces between protonated amino groups and the opposing
Y. Goto and 8. Aimoto
392
6
8
IO
PH (a)
I 100 NaCl
I 1000
( mM 1
Figure 6. Dependencies of the ellipticity at 222 run of the model polypeptide on (a) pH at different ionic strength, and (b) on h’aC1 concentration at different pH values? at PO”C. The values of ionic strength in (a) were 0.001 (0). @Ol (A), @05 (0). @1 (0) and 0.5 (O), where ionic strength of the buffer component was 001. except for the case of an ionic strength of 0.001, and the total ionic strength was controlled by KaCl. Filled circles represent the HCI-induced transition taken from Fig. 5, where the ionic st,reugth is minimal at each pH. The pH vaiues in (b) were 2.1 (10 mLw-HC1: 0). 4.4 (10 mM-Ka acetate: a). 68 (10 rnx-Tris.HCI: I-J), 8.0 (10 mM-‘fris.HCl: O), 8.3 (10 rnM-Tris.HCl: 0) and 95 (10 mh&!HES: 0).
(b) mainly weak hydrophobic interactions, forces, which stabilize the folded state. As described in detail by Goto et al. (1990b), anions can induce the folding transition by three possible mechanisms: shielding of the repulsive forces by (1) the Debye-Hiickel screening effect or (2) interaction with positive charges by anion binding, or (3) increasing the hydrophobic interactions of the polypeptide by affecting the water structure. The consistency of equation (1) with the electroselectivity series (eqns (3) and (4) of Goto et al. (1990b)) clearly indicates that, as is the case for acid-denatured proteins, electrostatic anion binding is the major factor responsible for the folding transition. Agreement of the acid-induced transitions of the model polypeptide with the corresponding salt induced transitions further confirms this conclusion.
Thus, the conformational transition of polypeptides induced by anion binding is not limited to aciddenatured proteins, but can take place even at neutral pH, provided that the polypeptide is basic and that the polypeptide has the potential to adopt an a-helical conformation when the electrostatic repulsion is reduced.
(c) Phase
diagram
of the conformationat
transition.
The co-operative folding transition was aiso induced when the pH was increased above 7.5. The cd. spectrum of the helical conformation at pH 9% (Fig. 4) was very similar to that induced by salt or acid. This indicates that the deprotonation of amino groups reduces the electrostatic repulsion and, as a
Conformation
2
of an Amphiphdic
4
6
Polypeptide
393
8
IO
PH
Figure 7. Phase diagram for the unfolded (U) and folded (F) helical states of the model polypeptide at 2O’C. The contour lines corresponding to 10 %, 30 %, 50 %, 70 %, 80 o/0 and 90 O/’ progress of the folding transition were coneltructed from the ?l’aCl-induced transitions at different pH values (circles), the pa-induced transitions at different ionic strengths (triangles). and the HCl-induced transition (squares). The ionic strength of the buffer component was less than 0.01 and the total ionic strength was controlled by NaCl. Hatched area is prohibited owing to the increase in the minimal ionic strength with decrease in pH.
result, the polypeptide takes a conformation similar to that induced by anion binding. As a first approximation, we assumed that the helical conformation induced at alkaline pH is the same as that induced by anion binding at neutral and acidic pH values. Then, we constructed a phase diagram for the pHand ionic strength-dependent conformational states on the basis of the pH-, of the model polypeptide NaCl- and HCl-induced conformational transitions (Fig. 7). The unfolded state is stable below pH 8 under conditions of low ionic strength. The left-hand boundary of the phase diagram corresponds to the ionic strength of HCl alone at the given pH and the hatched area is prohibited. The contour lines are relatively insensitive to pH below pH 7. Above pH 7, they show hyperbolic curvature toward the region of low ionic strength. Recently, Goto & Fink (1990) have reported a phase diagram for the acidic conformational states of apomyoglobin. An important point is that the phase diagram of the model polypeptide is very similar in shape to the phase diagram for the transition between the unfolded and molten globule states of apomyoglobin. We have been studying the pHand salt-dependent conformational transition of melittin. The phase diagram for the conformational states of melittin is also similar to that of the model polypeptide (Y. Goto, unpublished results). These results suggest that, not only the conformational stability of the model polypeptide, but also the conformation of many polypeptides, including aciddenatured proteins and melittin, are determined by a similar mechanism. Therefore, to elucidate the mechanism of protein folding, it is of importance to
know why the phase diagram shown in Figure 7. (d) Anion-
has a shape like that
and pH-dependent transition
conformational
To consider the effects of pH and anion on the conformational stability of the model polypeptide, we assume that the linked reactions of proton and anion bindings to an amino group and the conformational transition are represented by Mechanism 1: FA
Mechanism
1
where U and F represent the unfolded and folded states, respectively, species with the superscript H represent the protonated species, species with the subscript A represent the species with anion binding, and KFH, KiH, KtH, etc. represent the equilibrium constants of the respective processes indicated by arrows.
Y. Got0 and 8. Bimolo
394
From constant,
Mechanism II; the apparent equilibrium of the folding transition is defined as:
By using the equilibrium can be expressed as:
constants,
equation
(2)
+ KEH/[H] + KgH/[H]
+ K:[A]K;H/[H]) + K;[A]KgH/[H]
H
H
reactions
in
F
KUHKF!, = KUHKUHA F
’
(3)
Representation of the mechanism involving multiple (n) binding sites is complicated. However, provided that n sites are independent, the apparent equilibrium constant is represented simply by multiplying the contribution of the respective binding sites: @=”
of the linkage of the 1, i.e.: K”HKFH = KUHKt!
and:
x (1 + Kp[A] 1 + KyH[A]
=
The former corresponds to a decrease value upon folding and the latter to an the binding constant for anion upon
F
Kapp = KFH
K aPP
respectively. in the pK, increase in folding. Because Mechanism
fi i=l
(I+ K:;[Ai+ ’ (1 + K:“[A]+
K:;/[H] K:“/[H]
+ K:JA]K;y/[H]) + K: KUH H H KFH > KU”. A A
A
A
equilibria
F
3
must be:
KUH < KUF F and: Ku”
< KUHA
F
F
:
indicating that the conformational equilibrium shifts to stabilize the folded state upon either deprotonation or anion binding. Although we cannot- estimate the contribution of electrostatic interactions quant,itatively, the reduction of the electrostatic repulsive forces must be the major factor responsible for the change in conformational equilibrium upon deprotonation or anion-binding. To simula.te the conformational transition, we now assume two extreme cases. In one case, proton and anion bindings are independent (i.e. t,he independent binding of proton and anion): KIT = di:y>
Kl;i, = KY”, KF” H, HZ = KF”” and: Ku” = KUHA H, H,
(5)
where a is the activity of the ligand (Tanford, 1970). From the alkaline transition at, an ionic strength of 0.001, An is estimated to be - 1.8, suggesting that the number of amino groups responsible for the two. Analysis of the transition is at least SaCl-induced transition using molarity instead of activit’y gave a value of 1.3, suggesting that the number of binding sites for anions responsible for the transition is at least two. We also analyzed the folding transitions induced by various salts and acids and the results are shown in Table 1. Thus, while the anion-induced transition can be explained in terms of the preferential binding to the folded state compared with the binding to the extended unfolded state, the alkaline-induced transition is explained in terms of the preferential deprotonation of the folded state compared with the deprotonation of the unfolded state. Here, it would be useful to illustrate the property of one binding site in some detail using Mechanism 1. The preferential interactions of proton or anion with the polypeptide are represented by: and:
the eonformational
Then, equation
(4) becomes:
In another case, anions bind only to the protonated binding sites (i.e. the dependent bindings of proton and anion). In ot.her words t,he species rjl, and i,?, are absent from Mechanism 1. Equat’ion (4) reduces to: KdeP ~PP-
KUHn F
We do not know the exact number of amino groups responsible for the conformational transition. As the simplest case; we assumed that, on the basis of the preferential binding parameters obtained, only two amino groups are important and that these two groups are equivalent,. We estimated the values of KzH2, KIHH, KiH, KgH and KgH to be 0.01, 10031--“, 1 M-l, 10~7~s M. and 10-9.5 M respectively, so that t,he calculateb curves fit both the pH-induced transition curve at the low-est ionic strength (I = 0.001) and t,he NaCl-induced transition curves at acidic pH va.lues. It should be mentioned that there are so many parameters to
Conformation
0
0
2
4
6
8
of an Amphiphilic
IO
PH
0
2
4
Polypeptide
"0
2
395
4
6
ii----
6
8
PH
6
8
IO
PH
IO
PH
Figure 8. Simulation of the phase diagram (a, c) and the pH-induced conformational transition (b. d), on the basis of the dependent bindings (a. b) or independent (c: d) bindings of protons and anions to the 2 equivalent amino groups of the polypeptide (see the text.). In (a) and (c), contour lines corresponding to lo%, 30%, SO%, 70%, 80% and 90% progress of the folding transition are shown. In (b) and (d), folding transition curves at ionic strengths of @OOl: 0.01, @05, 0.1 and 0.5 are shown. Broken line shows the calculated folding transition curve induced by HCl.
determine that the set of values used is not unique, although they account for most of the observed conformational transitions. Figure 8(a) and (c) shows the phase diagrams calculated based on the assumptions of dependent binding (eqn (7)) and independent binding (eqn (6)), respectively. While the two phase diagrams agree with each other at zero ionic strength (i.e. on the pH axis) and at acidic pH below 5, they disagree largely between these two conditions. The contour lines based on the dependent bindings show a hyperbolic curvature at pH above 7 toward the region of low ionic strength, intersecting the pH axis with a steep angle (close to 90”). On the other hand, the contour lines based on the independent bindings show a sigmoidal curvature above pH 7. The differences in the shape of the phase diagrams are independent of the values of the parameters chosen or the number of responsible amino groups. It is evident that the shape of the observed phase diagram is similar to that obtained on the basis of the assumption of dependent bindings of proton and anion. Figure 8(b) and (d) shows the pH-induced transitions under different conditions of ionic strength, simulated on the basis of the assumptions of dependent and independent bindings, respectively. The transition curves calculated for the two assumptions are largely different. The pH region where co-operative conformational transition occurs shifts significantly to the lower pH region with an increase in ionic strength when the assumption of independent
binding is used. On the other hand, it is insensitive to ionic strength when the assumption of dependent binding is used. The observed transition curves (Fig. 6(a)) are similar to the curves based on the assumption of dependent binding. Thus, both the phase diagram and transition curves calculated indicate that the anion- and pHdependent conformational stability of the model polypeptide can be explained satisfactorily provided that anions are assumed to bind only to the protonated amino groups. This is what would be expected from the electrostatic nature of the interactions and substantiates the mechanism of how the conformational stability of the model polypeptide is determined by both salt and pH conditions. (e) Concludi~ng remarks Electrostatic interactions among the charged groups are an important factor in determining the conformation and stability of proteins (Matthew, 1985; Harvey, 1989). However, it is generally considered that they make a relatively small contribution compared the with contributions of hydrophobic interactions or hydrogen bonds (Hollecker & Creighton, 1982; Pace et al., 1990; Pace, 1990). However, present results anld those obtained by Goto et al. (1990a,b) show that the electrostatic repulsive forces are critically important in determining the conformation of acid-denatured proteins and the model polypeptide, in which the
Y. Goto and X. Aimoto
396
stability of the folded conformational states is relatively low. The anion- and pa-dependent folding transitions of the model polypeptide can be explained simply by the linked reactions of the conformational transition and proton and anion bindings to a limited number of amino groups, in which anions bind only to the protonated groups. The mechanism may be fundamental to the conformational stability of the acidic molten globule state and basic peptide such as melittin. We thank Professor Kozo Hamaguchi, Osaka Universit’y, for stimulating discussions. This work was supported in part by a Grant-in-Aid for Scientific Research from the Ministry of Educat.ion, Science and Culture of Japan. References Chen, Y.-H., Yang, J. T. & Marinez, H. M. (1972). Determination of the Secondary Structures of Proteins by Circular Dichroism and Optical Rotatory Dispersion. Biochemistry, 11, 4120-4131. De Grado, W. F. & Lear, J. D. (1985). Induction of Peptide Conformation at Apolar/Water Interfaces. J. Amer. Chem. Sot. 107, 7684-7689. Ebert, G. b Kuroyanagi, Y. (1982). Salt, Effects on the Conformation of a “Statistical” Copolymer of L-Leutine and L-Lysine. Polymer, 23, 1147-1153. Fink, A. L., Calciano, L. J., Goto, Y. & Palleros, D. R. (1990). Acid-denatured States of Proteins. In Current Research in Proteins Chemistry, (Villafranca, J. J., ed.), pp. 417-424, Aca.demic Press, San Diego. Gjerde, D. T., Schmuchler, G. & Fritz, J. S. (1980). Anion Chromatography with Low-conductivity Eluents. II. J. Chromatoyr. 187, 35-45. Goto, Y. & Fink, 8. L. (1989). Conformational States of. Beta-Lactamase: Molten-Globule States at Acidic and Alkaline pH with High Salt. Biochemistry 28, 945-952. Goto, Y. & Fink, A. L. (1990). Phase Diagram for Acidic Conformational States of Apomyoglobin. J. Mol. Biol. 214, 803-805. Goto, Y.? Calciano, L. J. & Fink, A. L. (1990a). Acid-induced Folding of Proteins. Proc. Nat. Acad. Xci., U.S.A. 87, 573-577. Edited
Goto,
Y., Takahashi, N. & Fink, A. I[,. (199%). Mechanism of Acid-induced Folding of Proteins. Biochemistry, 29, 3480-3488. Gregor. H. P.? Belle, J. & Marcus, R. A. (1955). Studies on Ion Enchange Resins. XIII. Selectivity Coefficients of Quaternary Base Anion-exchange Resins Toward Univalent Anions. J. Amer. Chem. Sot. 77,
2713-2719. Harvey, S. C. (1989). Treatment of Electrostatic Effects in Macromolecular Modeling. Proteins, 5, 78-92. Holleeker. M. & Creighton, T. E. (1982). Effect on Protein Stabilit’y of Reversing the Charge on Amino Groups. Biochim. Biophys. Acta, 701, 395-404. Kuwajima, K. (1989). The Molten Globule State as a Clue for Cnderstanding the Folding and Co-operativity of Globular Protein Structure. Proteins, 6, 87-103. Matthew, J. B. (1985). Electrostatic Effects in Proteins. Awnu. Rev. Biophys. Biophys. Chem. 14, 387-417. Ohgushi, M. & Wada, A. (1983). ‘Molten-Globule State’: A Compact Form of Globular Proteins with Mobile Side Chains. FEBS Letters, 164, 21-24. Pace, C. N. (1990). Conformational Stability of Globular Proteins. Trends Biochem. Sci. 15, 14-17. Pace, C. N., Laurents, D. V. & Thomson, J. A. (1990). pH Guanidine Urea and Dependence of the Hydrochloride Dena,turation of Ribonuclease A and Ribonuclease Tl. Biochemistry, 29, 2564-2572. Hypotheses and Ptitsyn, 0. B. (1987). Protein Folding: Experiments. J. Protein Chem. 6: 273-293. Ptitsyn, 0. B., Pain, R. H., Semisotnov, G. V.; Zerovn.ik; E. & Razgulyaev, 0. I. (1990). Evidence for a Molten Globule State as a General Intermediate in Protein Folding. FEBS Letters, 262. 20-24. Sillen, L. G. & Martell; A. E. (1971 j. StabiEity Constants of Il/letal-zon Com$exes, Supplement No. 1, AidPn Press, Oxford. Talbot. J. C., Dufourcq, J., deBony, J., Faucon, J. F. & Lussan: C. (1979). Conforma.tional Change and Self Association of Monomeric Melittin. FEBS Letters, 102; 191-193. Takakuwa, T., Konno, T. & Meguro, H. (1985). A jSl’ew Standard Substance for Calibration of Circular Ammonium Dichroism: d- lo-Camphorsulfonate. Anal. Sei. 1, 215-218. Tanford, C. (1970). Protein Denaturation. Part C. Models Mechanism of Theoretical for the Denaturation. Advan. Protein Chem. 24. l-95.
by P. E’. Wriyht
