Investigation of Zinc-Binding Affinities of Moloney Murine Leukemia Virus Nucleocapsid Protein and Its Related Zinc Finger and Modified Peptides YVES MELY,’** FABRICE CORNILLE,2 MARIE-CLAUDE FOURNIE-ZALUSKI,’ JEAN-LUC DARLIX,3 BERNARD P. ROQUES,* and D O M I N I Q U E GERARD’ ’Labor,itoire de Biophysique, CNRS UA 491, Universit6 Louis Pasteur, Facult6 de Pharmacie de Strasbourg, 74, route du Rhin B P 24 67401 lllkirch Cedex, 2D6partement de Chimie Organique, INSERM U266, CNRS UA498, UFR des Sciences Pharmaceutiques et Biologiques, 4, avenue de I’Observatoire, 75270 Paris Cedex 06, and ’Ecole Normale Sup6rieure de Lyon, 46, a1li.e d’ltalie, 69364 Lyon Cedex 07, France
SYNOPSIS
Nucleocapsid proteins of retroviruses are small basic, nucleic acid-binding proteins with either one or two “Cys-His” boxes, which have been shown to be involved in genomic RNA dimerization, encapsidation, and replication primer tRNA annealing to the initiation site for reverse transcription. The nucleocapsid ( NC ) protein of Moloney murine leukemia virus (MoMuLV NCplO) is made up of 56 residues with one Cys-His motif. The Zn‘+binding affinities and induced conformational changes of NCplO were investigated by following the fluorescence of Trp 35 located in the Cys-His domain. At pH 7.5, NCplO was shown to bind Zn2+ at a 1 : 1 ratio with a very high apparent binding constant of 1.2 (k0.3) 1013M-’.A similar apparent binding constant was obtained for a 19-residue peptide encompassing the Cys-His box, designated the “zinc finger motif,” indicating that it contains most if not all the information to bind Zn2+tightly. Changing Trp 35 to Phe in the peptide did not affect the Zn2+affinity, indicating that Trp 35 is not crucial for Zn2+binding. On the contrary, replacing Cys 29 by Ser, the chemical modification or oxidation of the three Cys sharply reduced Zn2+affinity, confirming the essential role of Cys in Zn2+binding. In addition, fluorescence and energy transfer data suggested that Zn 2+ binding modifies the Trp 35 environment but not its solvent exposure, and increases the average distance between Tyr 28 and Trp 35 by about 2 A. These data suggest that Zn2’ binding to retroviral NC protein is biologically relevant.
INTRODUCTION The nucleocapsid protein ( N C ) of retroviruses, encoded by the 3’ end of the gag gene, is a small, basic, single-stranded nucleic acid binding protein.’ In avian, murine, and human retroviruses, the NC protein has been shown to be involved in retroviral RNA dimerization 1-4 and e n ~ a p s i d a t i o n as ~-~ well as replication primer tRNA annealing to the initiation site for reverse transcription on the viral g e n ~ r n e . In ~.~ addition, the NC protein molecules have been found
Biopolvrnt.rs, Vol 31, 899-906 (1991) (r 1991 .Johp Wiley & Sons, Inc
CCC 0006-3525/91/070899-08$0400
* To whom correspondence should be addressed.
t o be firmly bound to the genomic RNA dimer in Retroviral NC proteins contain either one or two “Cys-His” motifs, thought to form zincbinding domains, by analogy with the zinc fingers of the transcription factor TFIIIA.” T h e binding of Zn to retroviral NC proteins and its biological relevance have been a matter of controversy. Data on Moloney murine leukemia virus ( MoMuLV) and on avian myeloblastosis virus ( AMV ) suggested that the NC proteins of both viruses7*’’did not bind stoichiometric amounts of Zn2+. It has also been reported that interactions between Rauscher MuLV NCplO and a nonspecific RNA, l 3 or between AMV NC protein and a homopolymer RNA,” were independent of Zn”. In contrast, an 18-residue peptide encompassing the first putative zinc finger motif
’+
a99
900
MELY ET AL.
of HIV-1 NC and Rauscher MuLV NCp10'"16~'7have been reported to bind Zn *+ tightly and stoichiometrically. Moreover, mutation of a Cys residue in the putative zinc finger of MoMuLV NC protein prevented genomic RNA encapsidation believed to be mediated by NC protein-viral RNA int e r a c t i o n ~ .Furthermore, ~ only freshly prepared MuLV NC protein interacted tightly with the homologous native viral RNA, whereas its denaturation decreased binding to the same RNA.'' Recent observations have indicated that MuLV NCplO can dimerize viral RNA and anneal replication primer tRNA a t the origin of reverse transcription, but not upon o ~ i d a t i o n . ~ ~ ~ ~ ~ ~ ' ~ T o assess the Zn*+-binding properties of MoMuLV NCplO, its Zn*+-bindingaffinity and associated conformational changes were investigated. To this end, the single Trp 35 residue, located in the zinc finger motif, seemed a well-suited reporter since its fluorescence is highly sensitive to Zn2'.19 T h e Zn2+-binding properties of NCplO zinc finger peptides were investigated in the same manner and compared with those of NCplO. In addition, since Cys and His residues should be protonated a t physiological pH, the p H dependence of Zn2+-binding properties was investigated.
was shown to be fully active in biological assays.lg The amino acid composition of NCplO 24-42 and the various point mutants is given in Figure 1. Lyophilized protein and peptides were stored under vacuum. Prior to use, they were dissolved in freshly degassed buffer (e.g., H E P E S 50 m M , KCl 0.1M p H 7.5) and poured immediately into anaerobic quartz cells that maintain an inert argon atmosphere. T h e mean time to prepare the samples was around 30 s and corresponded t o the so-called "starting recording time." NCplO and tryptophancontaining peptide concentrations were assessed using a n extinction coefficient of 7000M-' X cm-' a t 280 nm. The -SH content of NCplO and the various peptides, checked a t the beginning and the end of each experiment by titration with 5,5'-dithiobis(2-nitrobenzoic acid) , l o was as previously described.Ig Oxidation of -SH groups was always negligible throughout each experiment. EDTA and zinc sulfate were from Aldrich, EGTA from Sigma, and KCl Normatom from Prolabo.
Spectroscopic Measurements
All spectroscopic measurements were performed in anaerobic quartz cells. Absorption spectra were recorded with a Cary 219 spectrophotometer, while fluorescence spectra and kinetic measurements were recorded with a Perkin-Elmer M P F 66. Excitation wavelength was set either a t 295 nm for selective excitation of Trp, or a t 280 nm for excitation of both T r p and Tyr residues. Quantum yields were determined taking either L-Trp or L-Tyr in water, as a reference.*' T o determine the experimental Zn *+ -binding constants, fluorimetric titrations were performed by
MATERIALS A N D METHODS Materials
Synthesis and purification of NCplO and related zinc finger peptides: NCplO 24-42, (Ser") NCplO 24-43, ( Phe35)NCplO 24-42 and [ ( S-Acm) cYs26-29-39 INCplO 24-43 were performed as previously d e ~ c r i b e d . 'T~h e synthetized protein NCplO
NCplO
A T V V S G
0
OKODROGGERRRSOLDRD
Is-Rcm C y ~ ~ ~ ' ~ ~ ' ' ~ 1 N C24-45 p10
DO
C fl I
Y
Acm
ISer2'1
[Phes51
C K E K 6 H W fl K 0 C
P K K
1
I
Acm
Acm
HW
NCplO 24-45
DO
C fl Y K K E K 6
NCplO 24-42
DO
C fl
Y
C K E K 6 H W A K 0 C PKK
NCplO 24-42
DO
C fl
V
C K E K 6 H - fl K 0 C P
-
1
fl K 0 C P
P
K K P
K K
Figure 1. Primary sequence of NCplO and related zinc finger-likepeptides. Point mutants of the zinc finger peptide corresponded to individual changes in amino acids to the indicated underlined amino acids. In [ ( S-Acm)Cysz6~zg~3g] NCplO 24-43, each cysteine was coupled
to an acetamidomethyl group.
901
ZINC-BINDING AFFINITIES OF MoMuLV NCplO
adding increasing Zn2+concentrations to the protein or to a given peptide in 50 m M buffer: MES (4morpholine ethane sulfonic acid) or HEPES (N- (2hydroxyethyl ) -piperazine-N '-2-ethane sulfonic acid), 0.1 M KC1 a t the desired p H in the presence of 1 m M complexant (EDTA or EGTA). EDTA was used in the presence of high-affinity species a t p H above 7, while EGTA was used for lower affinity species or at acidic pH. For each addition of Zn2+, the fluorescence intensity changes a t the maximum emission wavelength (351 nm) were followed for a t least 30 min to ensure that equilibrium was attained. Excitation wavelength was set a t 280 nm, while excitation and emission slits were set respectively a t 2 and 5 nm. T o determine dissociation rate constants, kinetic measurements were performed by adding 1 m M EDTA to the Zn2+-loaded protein or peptide and following the time course of fluorescence intensity changes. Atomic absorption was performed with a Varian SpectAA-40. Contaminating Zn2+ from buffers was always below 1 pM. Data Analysis Determination of Experimental Zn2+-BindingConstants. From the fluorimetric titrations, the average number v of moles of Zn2+bound per mole of protein
was evaluated by
where ZF was the fluorescence measured for any added Zn2+concentration, while IF,,and ZFT correg 1.0
.6 0
sponded respectively to the fluorescence intensities of the apo- and the fully Zn2f -saturated protein or peptide. T h e concentration of free Zn2+was given by the positive root of the second degree equation
+ l ] [ Z n ] + v [Pt] [Zn,] -
=
0
(2)
where [ E,] , [ Zn,] , and [ P,] were respectively the total concentrations of complexant, zinc, and protein ( o r peptide), while KE was the affinity of the complexant for Zn2+computed from Martell and Smithy2 (e.g., a t p H 7.5, log KE was 13.65 and 9.15 for EDTA and EGTA, respectively). Finally, the experimental Zn2+-bindingconstant Kexp,was computed by fitting the experimental values of v and [ Zn] to the equation
Determination of Dissociation Rate Constants.
Experimental kinetic data were fitted by a single exponential theoretical curve, computed according to the equation
where u was calculated as in Eq. ( 1) , while koffcorresponded t o the dissociation rate constant and d t , the "starting recording time." As d t was rather high ( > 30 s ) , it was introduced as a parameter and the identity between the computed and the measured value of d t was a good check on the validity of the model. All calculations and curve fitting were carried out using non linear least square SAS computer procedures.
2 0.8 &
m
g 0.6 2
'R
RESULTS
0.4
\
ZnZ+-BindingProperties
73
5 0.2
0 n
r; 0.0
/I
15 14 13 12 11 10 9
8
7 6 5
-log(free Zn) Figure 2. Zn2+-bindingisotherms of NCplO and related zinc finger peptides. NCplO (A),NCplO 24-42 (H), and (Ser'g)NCp10 24-43 ( 0 ) were 40-60 ~ L M in 50 m M HEPES, 0.1 M KCl pH 7.5. Experimental values were obtained from fluorescence titrations, in the presence of 1 m M EDTA or EGTA, as indicated under Materials and Methods. The solid lines were drawn by using Eq. ( 3 ) , and the experimental binding constants of Table I.
The fluorescence of the single T r p residue of NCplO, NCplO 24-42, and ( Ser2') NCplO 24-43, a t p H 7.5, increases linearly with substoichiomet ric additions of Zn2+ (suggesting a high binding affinity) and reaches a plateau for a 1 : 1 stoi~hiometry.'~ The degree of saturation u of the protein or the peptide by Zn2+ could thus be determined unambigously from fluorescence titrations performed in the presence of metal complexants: EDTA and EGTA solubilized in large excess ( 1 m M ) over protein or peptide concentration to buffer very low free Zn2+concentrations.
902
Table I
MELY ET AL.
Experimental Zn2+-BindingConstants of NCplO and Related Zinc Finger Peptides at pH 7.5
NCplO
K (M-')
1.2(20.3) *
1OI3
( SerZ9)
NCplO 24-42 (Phe35)NCplO 24-42
NCplO 24-43
l.l(kO.1) * 101:j
2.6(20.1). lo7
[ ( S - A ~ m ) C y s ~ ~ ~ ' ~ ~ " 24-43 ~]NCp10 Oxidized NCplO 24-42"
< lo2
The fully oxidized peptide was obtained by keeping NCplO 24-42 under atmospheric conditions, a t room temperature, for 24 h. No detectable Zn2+binding was observed, so only upper limits are given. NCplO and the various peptides were 10 to 100 p M in 50 m M HEPES, O.1M KCI, pH 7.5. Experimental association binding constants, computed from fluorescence titrations, as indicated under Materials and Methods, were expressed as means (+ standard deviation) for a t least 3 independent determinations.
At p H 7.5, the experimental binding constants of NCplO and NCplO 24-42, computed from the binding curves (Figure 2 ) using Eq. ( 3 ), were extremely high and quite indistinguishable (Table I ) . An identical experimental binding constant was also observed for ( Phe35)NCplO 24-42 where the Z n 2 + binding process was monitored from the Zn2+-induced increase (about 60% ) in Tyr 28 fluorescence. In sharp contrast, the affinity of the point mutant (Ser") NCplO 24-43 was decreased by a 106factor, while no detectable Zn2+binding was observed when Cys residues were blocked by acetamidomethyl groups or oxidized. In the case of NCplO 24-42, the p H dependence of the experimental Zn2+-binding constants Kexpwas investigated and the binding of the metal to the finger domain proved critically dependent on the p H (Table 11). From the high KeXp values a t neutral and basic pH, it was inferred that dissociation rate constants were probably sufficiently low to be measured despite the long "starting recording time" (about 30 s ) due essentially to the argon-bubbling system needed t o keep Cys residues in their reduced form. Indeed, dissociation rate constants were measured for p H above 6 (Table 11) and in each case, kinetic curves fitted by Eq. ( 4 ) were clearly monoexponential (Figure 3 ) with a calculated starting recording time corresponding to the measured one. Moreover to further ascertain that true koffrates were measured, it was checked that kOffvalues were independent of EDTA concentration in the 0.5-10 mM range (data not shown). Znz+-lnduced Conformational Changes. Emission
quantum yields and maximum emission wavelength of NCplO, NCplO 24-42, and ( SerZ9)NCplO 24-43 in their apoforms a t p H 7.5 are quite indistinguishable.lg Moreover, the addition of an excess of Zn2i induces, in each case, a similar high quantum yield increase without affecting the maximum emission wavelength: 351 k 1 nm (Figure 4).T o further investigate the characteristics of the T r p 35 environment, the p H dependence of NCplO 24-42 quantum
yield was studied. At p H 4, both in the presence and in the absence of Zn", fluorescence quantum yields were low and quite indistinguishable. When the p H was increased, the apoprotein fluorescence quantum yield increased sigmoidally, with a midtransition point a t about p H 6.2 (Figure 5A) while in the presence of Z n 2 + ,the fluorescence quantum yield increased much more sharply, reaching a plateau a t about p H 5.5 (Figure 5B ) . Radiationless energy transfer that gives information on the distance between Tyr 28 and T r p 35 residues was a convenient means to follow the Zn 2+ induced reorganization of NCplO (or a given peptide) since these two residues are the sole aromatic amino acids in the protein and are strategically located in the zinc finger sequence.23The distance R between the two chromophores is related to the energy transfer efficiency 7 according to
where Ro is the Forster critical distance for a given donor-acceptor pair. For the Tyr-Trp couple, Rowas calculated by
where the overlap integral is J A D = 4.8 X 10-"M-' X cm6,24and the refractive index is n = 1.335. The orientation factor K' was taken as 2/3, as is usual when the donor and acceptor undergo complete dynamic isotropic orientational averaging.2s The quantum yield 46'' of the donor ( T y r ) was measured directly on ( Phe35)NCplO 24-42, a peptide whose T r p residue was replaced by Phe, which is not fluorescent when excited a t 280 nm. T h e quantum yields measured for this modified peptide were 0.022 (+0.001) in the absence and 0.036 (-tO.OOl) in the presence of Z n 2 + ,regardless of the p H (between 4 and 8 ) . Under these conditions, the calculated distances Ro were respectively 11.1and 12.1 A for the apo- and Zn2+-loaded species. Since the fluorescence quantum yield of the donor ( T y r ) in the presence
ZINC-BINDING AFFINITIES OF MoMuLV NCplO
903
Table I1 pH Dependence of NCplO 24-42 Experimental Zn2+-BindingConstants and Dissociation Rate Constantsa PH K (M
4.9 1)
how ( S ' )
5.45
i . ~ ( t ~ 104 . i ) . 1.2(+0.1).lo6
> 0.05
> 0.05
5.95 4.5(+-0.5).10' > 0.05
6.45 1 . 5 ( t o . i ) .lo9 4.7(&0.2)* lo-*
6.9 2.0(+-0.1).lo1!
i . i ( + o . i ) . 1013
l . l ( k 0 . 1 ) . lo-'
6.4(&0.3)
1.5
7.9 i.6(+0.2). 1014 4.2(+0.2).
* K ( ' p 1 0 21-42 was 10-60 p M in 50 mM buffer, 0.1M KCI, at the desired pH. For pH < 6.5, the buffer was MES, while for pH > 6.5. it was HEPES. Experiniental binding constants were computed and expressed as in Table I. Dissociation rate constants, expressed as means (? standard deviation), were determined from kinetic experiments, as described under Materials and Methods. For pH < 6.0, hoe were too high to be measured in our system and so only lower limits are given.
of the acceptor ( T r p ) was low and difficult to evaluate from the fluorescence spectra of the protein or peptide, the efficiency of the energy transfer process was computed from the T r p fluorescence enhancement by a slightly rearranged form of the classical Forster equation:
where f ;'Po and f i8" were respectively the fractional absorption of Tyr and T r p a t 280 nm, while and 6"' represented respectively the measured quantum yields of the protein or peptide a t 280- and 295-nm excitation wavelengths. Calculated distances R between Tyr 28 and T r p 35 chromophores were respectively 10.8 ( k 0 . 2 ) and 12.8 ( a 0 . 2 ) A in the absence and in the presence of excess Zn2+,regardless of pH (between 5.5 and 8 ) .
DISCUSSION The NC protein NCplO of MoMuLV has one CysHis motif, where the single Tyr and T r p residues are located in positions 28 and 35, respectively. Zn2+ binding to NCplO was monitored using the fluorescence of T r p 35 or, alternatively, Tyr 28. We report here that NCplO, NCplO 24-42, and ( Phe3")NCp10 24 42 bind one Zn2+per protein molecule, with a n experimental binding constant a s high as 10'3M-1. Such a high value was not unexpected since the nucleoprotein from the related Rauscher murine leukemia virus bind Zn2+ with a n estimated binding constant of 10"MM-'a t p H 7.0.17 Moreover, apparent Zn" -binding constants greater than 10'*M-' were also inferred for the zinc finger domains of Xenopus transcription factor IIIA.26The zinc finger domain of NC protein appears to possess most, if not all, of the information necessary and sufficient for Zn2+ binding, since its experimental Zn2+-binding constant is very close to that of NC protein. However, the finger peptide cannot activate dimerization of the retroviral RNA and annealing of replication
primer tRNA. In fact, both these processes that take place in the course of MoMuLV virion assembly require the complete NC protein (Darlix, unpublished results). It can be concluded that the finger domain is necessary but not sufficient for the full biochemical activity of NCplO. From the dependence of the experimental Zn2+binding constants upon p H in the range of 5-8, the presence of ligands with pKa in this p H range was deduced. Since both Cys and His residues could be protonated in this p H range, they are likely candidates for this p H dependence. T o investigate this, a simple model was built assuming that only the species with fully deprotonated Cys and His could bind Zn2+with a high binding constant K M e . Thus, taking a pK1 of 6.2 for His ( a s discussed further) and assuming that each Cys protonates with a n identical pK2 constant, the measured Zn '+ -binding constant Kexpcould be readily expressed as
log(Kexp)= 10g(KMe)- log[l X [H]
+ lopK'
+ ( 1 0 p K 2 )X2 [ H I 2+ (10L'"2)'3 X [ H I 3 X (1
3 0.21
+ lopK1X
[ H I ) ] (8)
J
\,
c 0.0
N
0
100
200
300
400
500
Time (s) Figure 3. Kinetics of Zn2+release from NCplO 24-42. Zn2+-loadedNCplO 24-42,8.85 pM, in 50 m M HEPES, 0.1M KC1, pH 7.5, was mixed with 1 m M EDTA. The solid curve was the experimental record while the dashed curve corresponded to the computer fit to the monoexponential Eq. ( 4 ) , using the dissociation rate constant of Table 11.
904
MELY ET AL.
>- 1.4
t
* 1.2 z 1.0
z
w 0.8 0
5 0.6 0 0.4
8 0.2 i 0.0 3
340
300
380
420
460
500
WAVELENGTH (nm) Figure 4. Fluorescence spectra of apo-NCplO 24-42 (a,b) and zinc-saturated (c,d) NCplO 24-42. The peptide ( 2 0 p M ) was dissolved in 50 m M HEPES, 0.1 M KCl, pH 7.5. Spectra were recorded for excitation wavelengths set a t 280 (a,d) and 295 (b,c) nm, respectively.
Experimental data were fitted very satisfactorily with Eq. ( 8 ) giving pK2 = 8.6 (k0.2), which is fully compatible with the usual pK, range of Cys and KMe = 1.8 ( k0.2 ) 1016M-’. Including supplementary Zn2+-binding species (e.g., monoprotonated ones) worsened the fit, which validates the hypothesis that probably, as for yeast transcription factor zinc finger domains, 27 only species with fully deprotonated His and Cys could bind Zn2+. In addition, as the p H dependence of the experimental binding constant Kexpwas much stronger than that of the dissociation rate constant kOff, it was tempting to deduce that Kexpvariations stemmed essentially from variations in the association rate constant k,, . However, a t p H above 7, the calculated k,, became unrealistically high ( k o n> 5.10’’ M-’ X s-’), suggesting that the binding scheme is not governed by a simple twostate equilibrium between the Zn 2+ -complexed and uncomplexed forms, but probably involves a more complicated pathway with different conformational species. It could thus be inferred that the measured binding constants Kexpin fact correspond to apparent binding constants and that further information is needed to fully describe this binding mechanism. Replacing Cys 29 by Ser causes a drastic decrease in the Zn2+-bindingaffinity and results in a strong inhibition of genomic RNA encapsidation in vivo7 indicating that the Zn2+-saturated NCplO protein is necessary for genomic RNA packaging into virions. However, we do not yet know a t which stage of genome packaging nor how Zn2+ bound to NC intervenes. This is a t present under investigation. When all Cys residues were oxidized or chemically modified, the Zn2+-binding properties vanished, confirming the critical involvement of these residues in Zn2+binding. In clear contrast, replacing T r p 35
by Phe left the Zn2+affinity of the zinc finger domain unaffected. Similarly, replacing T r p 35 by Ser did not affect Zn2+binding to NCplO, but did result in the absence of MuLV RNA dimerization by mutant NC protein in vitro and genomic RNA packaging in vivo ( Prats and Darlix, unpublished results). These data indicate that T r p 35 within the zinc finger is not essential for Zn2+binding, but does play a critical role in MuLV dimerization and packaging, probably through specific NC-RNA interactions. Since NCplO, NCplO 24-42, and (Ser”) NCplO 24-43 fluoresced, both in the absence and presence of Zn2+,with a maximum at about 351 nm, a fully polar environment for T r p 35 in the apo- as well as in the Zn2+-loadedspecies is suggested. In the absence of Zn2+, the quantum yields of these compounds were rather low and sigmoidally dependent _____
0.07
-
A
0.06 -
6 0.03 4
5
6
7
8
7
8
PH
4
5
6
PH Figure 5. pH dependence of NCplO 24-42 quantum yield. Quantum yields of both apo-NCplO 24-42 ( A ) and Znz+-loaded ( B ) NCplO 24-42 were determined as described under Materials and Methods. Excitation wavelength was set either a t 280 nm (.), where both Tyr and Trp residues absorb, or 295 nm ( 0 ) , where only Trp residues absorb. Each quantum yield was the mean of 3-8 independent determinations, and its standard deviation was between 0.001 and 0.003. The solid lines in Figure 5A were drawn by using Eq. 10, while those in Figure 5B were hand-drawn.
ZINC-BINDING AFFINITIES OF MoMuLV NCplO
on the pH, with a midtransition point a t about pH 6.2. The localization of the His 34 residue in the vicinity of the T r p residue, the usual pK, range of His residues (about 6-7) and the well-known potential of His residues to act as fluorescence quenchers under their protonated formz8 suggest that the midtransition point corresponds to the pK of His 34. T o further verify this point, let us consider the equilibrium between M and MH, the protein (or peptide) species with deprotonated and protonated His respectively, governed by the His protonation constant K , . T h e measured quantum yield @,,, a t a given pH can be expressed as
where f M is the fraction of protein with deprotonated His, while @M and @MH are the quantum yields of M (measured a t p H 8) and MH (measured a t p H 4), respectively. From Eq. ( 9 ) and the protonation equilibrium, the following relation can be derived:
where [ H ] is deduced from the p H value. Experimental quantum yields measured with either 280or 295-nm excitation wavelengths were adequately fitted with Eq. (10) (Figure 5 A ) , giving pKl = 6.2 ( kO.1) , thus strengthening the foregoing hypothesis. In the presence of Z n Z f , since the maximum emission wavelength is almost unchanged, the high quantum yields above p H 5.5 are not due to a change in the polarity of T r p 35 environment but rather to a rearrangement of T r p environment, which takes the Trp 35 residue away from neighboring quenching groups. These quenching groups clearly not include His 34 residue, since a t p H 8, where His was fully deprotonated ( a n d thus no longer able to quench fluorescence), the apoprotein quantum yield remained much lower than that of the Zn2+-loaded form. Moreover, radiationless Tyr + T r p energy transfer suggests that Tyr 28 and T r p 35 residues move off by about 2 A. Due to the probable flexibility of the apo species, this value represents in fact a n average distance between the various conformations of the apo species and the zinc-saturated one. This small hut highly reproducible value thus confirmed the existence of a zinc-induced rearrangement. These Zn 2 i -induced structural modifications associated with fluorescence quantum yield changes probably represent a discrete local reorganization rather than a n important conformational change, since only minor Zn 2+ -induced secondary structure changes have been detected in the related Rauscher
905
MuLV NCplO protein.':' The decrease in 2n"binding affinities at acidic p H is reflected by decreased quantum yields, and thus, below pH 5.5, NCplO or NCplO 24-42 were present as a mixture of apo- and Zn*+-Ioadedspecies. This was further confirmed by the relationship between the quantum yield a t a given acidic p H and the concentration of added Zn2+(data not shown). Since similar spectral characteristics were observed a t p H 7.5 for NCplO, NCplO 24-42, or (Ser") NCplO 24-43, it is suggested that the T r p environment is not greatly modified by amino acids outside the zinc finger sequence nor by the nature of the amino acid a t position 29, and that NCplO and the two finger peptides have similar conformations in both their apo- and Zn'+-loaded forms. In conclusion, our data indicate that Moloney MuLV NCplO is a Zn2+-bindingprotein and suggest that Zn2+bound to NC protein is biologically relevant. This tight binding of one Zn" per N C molecule probably plays a structural role in the retroviral NC protein and work is now in progress to verify this point. This work was supported by grants from Centre National de la Recherche Scientifique, Institut National de la Santb et de la Recherche Mkdicale (Grant INSERM No. 861010), Agence Nationale de la Recherche sur le SIDA, and Universitk Louis Pasteur.
REFERENCES 1. Coffin, J. M. (1984) in RNA Tumor Viruses, vol. I, Weiss, R., Teich N., Varmus, H. & Coffin, J. M., Eds.,
2.
3. 4.
5. 6. 7.
8.
9.
Cold Spring Harbor Laboratories Press, Cold Spring Harbor, NY, pp. 261-368. Murti, K. G., Bondurant, M. & Tereba, A. (1981) J. Virol. 37,411-419. Bieth, E., Gabus, C. & Darlix, J. L. (1990) Nucleic Acids Res. 18, 119-127. Prats, A. C., Roy, C., Wang, P., Erard, M., Housset, V., Gabus, C., Paoletti, C. & Darlix, J. I,. (1990) J. Virol. 64, 774-783. Mbric, C. & Spahr, P. F. (1986) J . Virol. 60, 450459. Mbric, C. & Goff, S.P. (1989) J . Virol. 63, 15581568. Gorelick, R. J., Henderson, L. E., Hanser, J. P. & Rein, A. (1988)Proc. Natl. Acad. Sci. USA 85,84208424. Prats, A. C., Sarih, L., Gabus, C., Litvak, S., Keith, G. & Darlix, J. L. (1988) EmboJ. 7 , 1777-1783. Barat, C., Lullien, V., Schatz, O., Keith, G., Nugeyre, M. T., Gruninger-Leitch, F., Barrk-Sinoussi, F., Le Grice, S. F. J. & Darlix, J. L. ( 1989) EmboJ. 8,32793285.
906
MELY ET AL.
10. Darlix, J. L. & Spahr, P. F. (1982) J . Mol. Biol. 189, 147-161. 11. Berg, J. M. (1986) Science 232, 485-487. 12. Jentoft, J. E., Smith, L. M., Fu, X., Johnson, M. & Leis, J. ( 1988) Proc. Natl. Acad. Sci. U S A 85, 70947098. 13. Roberts, W. J., Pan, T., Elliott, J. I., Coleman, J. E. & Williams, K. R. (1989) Biochemistry 28, 1004310047. 14. South, T. L., Kim, B. & Summers, M. F. (1989) J. Am. Chem. SOC. 111, 395-396. 15. Summers, M. F., South, T. L., Kim, B. & Hare, D. R. (1990) Biochemistry 29,329-340. 16. Green, L. M. & Berg, J. M. (1989) Proc. Natl. Acad. Sci. U S A 86, 4047-4051. 17. Green, L. M. & Berg, J. M. (1990) Proc. Natl. Acad. Sci. U S A 87, 6403-6407. 18. Nissen-Meyer, J. & Abraham, A. K. (1980) J . Mol. Biol. 142, 19-28. 19. Cornille, F., Mely, Y., Ficheux, D., Savignol, I., GCrard, D., Darlix, J. L., Fournih-Zaluski, M. C. & Roques, B. P. ( 1990) Znt. J. Peptide Protein Res. 36,551-558.
20. Ellman, G. L. (1959) Arch. Biochem. Biophys. 82, 70-77. 21. GBrard, D., Lemieux, G. & Laustriat, G. (1975) Photochem. Photobiol. 22, 89-95. 22. Martell, A. E. & Smith, R. M. (1974) Critical Stability Constants, Vol. 1, Plenum Press, New York. 23. Schiller, P. W. & St. Hilaire, J. (1980) J. Med. Chem. 23, 290-294. 24. Eisinger, J., Feuer, B. & Lamola, A. A. (1969) Biochemistry 8, 3908-3915. 25. FourniC-Zaluski, M. C., Belleney, J., Lux, B., Durieux, C., GCrard, D., Gacel, G., Maigret, B. & Roques, B. P. ( 1986) Biochemistry 25, 3778-3787. 26. Shang, Z., Liao, Y. D., Wu, F. Y. H. & Wu, C. W. ( 1989) Biochemistry 28,9790-9795. 27. Parraga, G., Horvath, S., Hood, L., Young, E. T. & Klevit, R. E. (1990) Proc. Natl. Acad. Sci. USA 87, 137-141. 28. Shinitzky, M. & Goldman, R. ( 1967) Eur. J . Biochem. 3, 139-144. Received November 26, 1990 Accepted March 12, 1991
SYNOPSIS
Nucleocapsid proteins of retroviruses are small basic, nucleic acid-binding proteins with either one or two “Cys-His” boxes, which have been shown to be involved in genomic RNA dimerization, encapsidation, and replication primer tRNA annealing to the initiation site for reverse transcription. The nucleocapsid ( NC ) protein of Moloney murine leukemia virus (MoMuLV NCplO) is made up of 56 residues with one Cys-His motif. The Zn‘+binding affinities and induced conformational changes of NCplO were investigated by following the fluorescence of Trp 35 located in the Cys-His domain. At pH 7.5, NCplO was shown to bind Zn2+ at a 1 : 1 ratio with a very high apparent binding constant of 1.2 (k0.3) 1013M-’.A similar apparent binding constant was obtained for a 19-residue peptide encompassing the Cys-His box, designated the “zinc finger motif,” indicating that it contains most if not all the information to bind Zn2+tightly. Changing Trp 35 to Phe in the peptide did not affect the Zn2+affinity, indicating that Trp 35 is not crucial for Zn2+binding. On the contrary, replacing Cys 29 by Ser, the chemical modification or oxidation of the three Cys sharply reduced Zn2+affinity, confirming the essential role of Cys in Zn2+binding. In addition, fluorescence and energy transfer data suggested that Zn 2+ binding modifies the Trp 35 environment but not its solvent exposure, and increases the average distance between Tyr 28 and Trp 35 by about 2 A. These data suggest that Zn2’ binding to retroviral NC protein is biologically relevant.
INTRODUCTION The nucleocapsid protein ( N C ) of retroviruses, encoded by the 3’ end of the gag gene, is a small, basic, single-stranded nucleic acid binding protein.’ In avian, murine, and human retroviruses, the NC protein has been shown to be involved in retroviral RNA dimerization 1-4 and e n ~ a p s i d a t i o n as ~-~ well as replication primer tRNA annealing to the initiation site for reverse transcription on the viral g e n ~ r n e . In ~.~ addition, the NC protein molecules have been found
Biopolvrnt.rs, Vol 31, 899-906 (1991) (r 1991 .Johp Wiley & Sons, Inc
CCC 0006-3525/91/070899-08$0400
* To whom correspondence should be addressed.
t o be firmly bound to the genomic RNA dimer in Retroviral NC proteins contain either one or two “Cys-His” motifs, thought to form zincbinding domains, by analogy with the zinc fingers of the transcription factor TFIIIA.” T h e binding of Zn to retroviral NC proteins and its biological relevance have been a matter of controversy. Data on Moloney murine leukemia virus ( MoMuLV) and on avian myeloblastosis virus ( AMV ) suggested that the NC proteins of both viruses7*’’did not bind stoichiometric amounts of Zn2+. It has also been reported that interactions between Rauscher MuLV NCplO and a nonspecific RNA, l 3 or between AMV NC protein and a homopolymer RNA,” were independent of Zn”. In contrast, an 18-residue peptide encompassing the first putative zinc finger motif
’+
a99
900
MELY ET AL.
of HIV-1 NC and Rauscher MuLV NCp10'"16~'7have been reported to bind Zn *+ tightly and stoichiometrically. Moreover, mutation of a Cys residue in the putative zinc finger of MoMuLV NC protein prevented genomic RNA encapsidation believed to be mediated by NC protein-viral RNA int e r a c t i o n ~ .Furthermore, ~ only freshly prepared MuLV NC protein interacted tightly with the homologous native viral RNA, whereas its denaturation decreased binding to the same RNA.'' Recent observations have indicated that MuLV NCplO can dimerize viral RNA and anneal replication primer tRNA a t the origin of reverse transcription, but not upon o ~ i d a t i o n . ~ ~ ~ ~ ~ ~ ' ~ T o assess the Zn*+-binding properties of MoMuLV NCplO, its Zn*+-bindingaffinity and associated conformational changes were investigated. To this end, the single Trp 35 residue, located in the zinc finger motif, seemed a well-suited reporter since its fluorescence is highly sensitive to Zn2'.19 T h e Zn2+-binding properties of NCplO zinc finger peptides were investigated in the same manner and compared with those of NCplO. In addition, since Cys and His residues should be protonated a t physiological pH, the p H dependence of Zn2+-binding properties was investigated.
was shown to be fully active in biological assays.lg The amino acid composition of NCplO 24-42 and the various point mutants is given in Figure 1. Lyophilized protein and peptides were stored under vacuum. Prior to use, they were dissolved in freshly degassed buffer (e.g., H E P E S 50 m M , KCl 0.1M p H 7.5) and poured immediately into anaerobic quartz cells that maintain an inert argon atmosphere. T h e mean time to prepare the samples was around 30 s and corresponded t o the so-called "starting recording time." NCplO and tryptophancontaining peptide concentrations were assessed using a n extinction coefficient of 7000M-' X cm-' a t 280 nm. The -SH content of NCplO and the various peptides, checked a t the beginning and the end of each experiment by titration with 5,5'-dithiobis(2-nitrobenzoic acid) , l o was as previously described.Ig Oxidation of -SH groups was always negligible throughout each experiment. EDTA and zinc sulfate were from Aldrich, EGTA from Sigma, and KCl Normatom from Prolabo.
Spectroscopic Measurements
All spectroscopic measurements were performed in anaerobic quartz cells. Absorption spectra were recorded with a Cary 219 spectrophotometer, while fluorescence spectra and kinetic measurements were recorded with a Perkin-Elmer M P F 66. Excitation wavelength was set either a t 295 nm for selective excitation of Trp, or a t 280 nm for excitation of both T r p and Tyr residues. Quantum yields were determined taking either L-Trp or L-Tyr in water, as a reference.*' T o determine the experimental Zn *+ -binding constants, fluorimetric titrations were performed by
MATERIALS A N D METHODS Materials
Synthesis and purification of NCplO and related zinc finger peptides: NCplO 24-42, (Ser") NCplO 24-43, ( Phe35)NCplO 24-42 and [ ( S-Acm) cYs26-29-39 INCplO 24-43 were performed as previously d e ~ c r i b e d . 'T~h e synthetized protein NCplO
NCplO
A T V V S G
0
OKODROGGERRRSOLDRD
Is-Rcm C y ~ ~ ~ ' ~ ~ ' ' ~ 1 N C24-45 p10
DO
C fl I
Y
Acm
ISer2'1
[Phes51
C K E K 6 H W fl K 0 C
P K K
1
I
Acm
Acm
HW
NCplO 24-45
DO
C fl Y K K E K 6
NCplO 24-42
DO
C fl
Y
C K E K 6 H W A K 0 C PKK
NCplO 24-42
DO
C fl
V
C K E K 6 H - fl K 0 C P
-
1
fl K 0 C P
P
K K P
K K
Figure 1. Primary sequence of NCplO and related zinc finger-likepeptides. Point mutants of the zinc finger peptide corresponded to individual changes in amino acids to the indicated underlined amino acids. In [ ( S-Acm)Cysz6~zg~3g] NCplO 24-43, each cysteine was coupled
to an acetamidomethyl group.
901
ZINC-BINDING AFFINITIES OF MoMuLV NCplO
adding increasing Zn2+concentrations to the protein or to a given peptide in 50 m M buffer: MES (4morpholine ethane sulfonic acid) or HEPES (N- (2hydroxyethyl ) -piperazine-N '-2-ethane sulfonic acid), 0.1 M KC1 a t the desired p H in the presence of 1 m M complexant (EDTA or EGTA). EDTA was used in the presence of high-affinity species a t p H above 7, while EGTA was used for lower affinity species or at acidic pH. For each addition of Zn2+, the fluorescence intensity changes a t the maximum emission wavelength (351 nm) were followed for a t least 30 min to ensure that equilibrium was attained. Excitation wavelength was set a t 280 nm, while excitation and emission slits were set respectively a t 2 and 5 nm. T o determine dissociation rate constants, kinetic measurements were performed by adding 1 m M EDTA to the Zn2+-loaded protein or peptide and following the time course of fluorescence intensity changes. Atomic absorption was performed with a Varian SpectAA-40. Contaminating Zn2+ from buffers was always below 1 pM. Data Analysis Determination of Experimental Zn2+-BindingConstants. From the fluorimetric titrations, the average number v of moles of Zn2+bound per mole of protein
was evaluated by
where ZF was the fluorescence measured for any added Zn2+concentration, while IF,,and ZFT correg 1.0
.6 0
sponded respectively to the fluorescence intensities of the apo- and the fully Zn2f -saturated protein or peptide. T h e concentration of free Zn2+was given by the positive root of the second degree equation
+ l ] [ Z n ] + v [Pt] [Zn,] -
=
0
(2)
where [ E,] , [ Zn,] , and [ P,] were respectively the total concentrations of complexant, zinc, and protein ( o r peptide), while KE was the affinity of the complexant for Zn2+computed from Martell and Smithy2 (e.g., a t p H 7.5, log KE was 13.65 and 9.15 for EDTA and EGTA, respectively). Finally, the experimental Zn2+-bindingconstant Kexp,was computed by fitting the experimental values of v and [ Zn] to the equation
Determination of Dissociation Rate Constants.
Experimental kinetic data were fitted by a single exponential theoretical curve, computed according to the equation
where u was calculated as in Eq. ( 1) , while koffcorresponded t o the dissociation rate constant and d t , the "starting recording time." As d t was rather high ( > 30 s ) , it was introduced as a parameter and the identity between the computed and the measured value of d t was a good check on the validity of the model. All calculations and curve fitting were carried out using non linear least square SAS computer procedures.
2 0.8 &
m
g 0.6 2
'R
RESULTS
0.4
\
ZnZ+-BindingProperties
73
5 0.2
0 n
r; 0.0
/I
15 14 13 12 11 10 9
8
7 6 5
-log(free Zn) Figure 2. Zn2+-bindingisotherms of NCplO and related zinc finger peptides. NCplO (A),NCplO 24-42 (H), and (Ser'g)NCp10 24-43 ( 0 ) were 40-60 ~ L M in 50 m M HEPES, 0.1 M KCl pH 7.5. Experimental values were obtained from fluorescence titrations, in the presence of 1 m M EDTA or EGTA, as indicated under Materials and Methods. The solid lines were drawn by using Eq. ( 3 ) , and the experimental binding constants of Table I.
The fluorescence of the single T r p residue of NCplO, NCplO 24-42, and ( Ser2') NCplO 24-43, a t p H 7.5, increases linearly with substoichiomet ric additions of Zn2+ (suggesting a high binding affinity) and reaches a plateau for a 1 : 1 stoi~hiometry.'~ The degree of saturation u of the protein or the peptide by Zn2+ could thus be determined unambigously from fluorescence titrations performed in the presence of metal complexants: EDTA and EGTA solubilized in large excess ( 1 m M ) over protein or peptide concentration to buffer very low free Zn2+concentrations.
902
Table I
MELY ET AL.
Experimental Zn2+-BindingConstants of NCplO and Related Zinc Finger Peptides at pH 7.5
NCplO
K (M-')
1.2(20.3) *
1OI3
( SerZ9)
NCplO 24-42 (Phe35)NCplO 24-42
NCplO 24-43
l.l(kO.1) * 101:j
2.6(20.1). lo7
[ ( S - A ~ m ) C y s ~ ~ ~ ' ~ ~ " 24-43 ~]NCp10 Oxidized NCplO 24-42"
< lo2
The fully oxidized peptide was obtained by keeping NCplO 24-42 under atmospheric conditions, a t room temperature, for 24 h. No detectable Zn2+binding was observed, so only upper limits are given. NCplO and the various peptides were 10 to 100 p M in 50 m M HEPES, O.1M KCI, pH 7.5. Experimental association binding constants, computed from fluorescence titrations, as indicated under Materials and Methods, were expressed as means (+ standard deviation) for a t least 3 independent determinations.
At p H 7.5, the experimental binding constants of NCplO and NCplO 24-42, computed from the binding curves (Figure 2 ) using Eq. ( 3 ), were extremely high and quite indistinguishable (Table I ) . An identical experimental binding constant was also observed for ( Phe35)NCplO 24-42 where the Z n 2 + binding process was monitored from the Zn2+-induced increase (about 60% ) in Tyr 28 fluorescence. In sharp contrast, the affinity of the point mutant (Ser") NCplO 24-43 was decreased by a 106factor, while no detectable Zn2+binding was observed when Cys residues were blocked by acetamidomethyl groups or oxidized. In the case of NCplO 24-42, the p H dependence of the experimental Zn2+-binding constants Kexpwas investigated and the binding of the metal to the finger domain proved critically dependent on the p H (Table 11). From the high KeXp values a t neutral and basic pH, it was inferred that dissociation rate constants were probably sufficiently low to be measured despite the long "starting recording time" (about 30 s ) due essentially to the argon-bubbling system needed t o keep Cys residues in their reduced form. Indeed, dissociation rate constants were measured for p H above 6 (Table 11) and in each case, kinetic curves fitted by Eq. ( 4 ) were clearly monoexponential (Figure 3 ) with a calculated starting recording time corresponding to the measured one. Moreover to further ascertain that true koffrates were measured, it was checked that kOffvalues were independent of EDTA concentration in the 0.5-10 mM range (data not shown). Znz+-lnduced Conformational Changes. Emission
quantum yields and maximum emission wavelength of NCplO, NCplO 24-42, and ( SerZ9)NCplO 24-43 in their apoforms a t p H 7.5 are quite indistinguishable.lg Moreover, the addition of an excess of Zn2i induces, in each case, a similar high quantum yield increase without affecting the maximum emission wavelength: 351 k 1 nm (Figure 4).T o further investigate the characteristics of the T r p 35 environment, the p H dependence of NCplO 24-42 quantum
yield was studied. At p H 4, both in the presence and in the absence of Zn", fluorescence quantum yields were low and quite indistinguishable. When the p H was increased, the apoprotein fluorescence quantum yield increased sigmoidally, with a midtransition point a t about p H 6.2 (Figure 5A) while in the presence of Z n 2 + ,the fluorescence quantum yield increased much more sharply, reaching a plateau a t about p H 5.5 (Figure 5B ) . Radiationless energy transfer that gives information on the distance between Tyr 28 and T r p 35 residues was a convenient means to follow the Zn 2+ induced reorganization of NCplO (or a given peptide) since these two residues are the sole aromatic amino acids in the protein and are strategically located in the zinc finger sequence.23The distance R between the two chromophores is related to the energy transfer efficiency 7 according to
where Ro is the Forster critical distance for a given donor-acceptor pair. For the Tyr-Trp couple, Rowas calculated by
where the overlap integral is J A D = 4.8 X 10-"M-' X cm6,24and the refractive index is n = 1.335. The orientation factor K' was taken as 2/3, as is usual when the donor and acceptor undergo complete dynamic isotropic orientational averaging.2s The quantum yield 46'' of the donor ( T y r ) was measured directly on ( Phe35)NCplO 24-42, a peptide whose T r p residue was replaced by Phe, which is not fluorescent when excited a t 280 nm. T h e quantum yields measured for this modified peptide were 0.022 (+0.001) in the absence and 0.036 (-tO.OOl) in the presence of Z n 2 + ,regardless of the p H (between 4 and 8 ) . Under these conditions, the calculated distances Ro were respectively 11.1and 12.1 A for the apo- and Zn2+-loaded species. Since the fluorescence quantum yield of the donor ( T y r ) in the presence
ZINC-BINDING AFFINITIES OF MoMuLV NCplO
903
Table I1 pH Dependence of NCplO 24-42 Experimental Zn2+-BindingConstants and Dissociation Rate Constantsa PH K (M
4.9 1)
how ( S ' )
5.45
i . ~ ( t ~ 104 . i ) . 1.2(+0.1).lo6
> 0.05
> 0.05
5.95 4.5(+-0.5).10' > 0.05
6.45 1 . 5 ( t o . i ) .lo9 4.7(&0.2)* lo-*
6.9 2.0(+-0.1).lo1!
i . i ( + o . i ) . 1013
l . l ( k 0 . 1 ) . lo-'
6.4(&0.3)
1.5
7.9 i.6(+0.2). 1014 4.2(+0.2).
* K ( ' p 1 0 21-42 was 10-60 p M in 50 mM buffer, 0.1M KCI, at the desired pH. For pH < 6.5, the buffer was MES, while for pH > 6.5. it was HEPES. Experiniental binding constants were computed and expressed as in Table I. Dissociation rate constants, expressed as means (? standard deviation), were determined from kinetic experiments, as described under Materials and Methods. For pH < 6.0, hoe were too high to be measured in our system and so only lower limits are given.
of the acceptor ( T r p ) was low and difficult to evaluate from the fluorescence spectra of the protein or peptide, the efficiency of the energy transfer process was computed from the T r p fluorescence enhancement by a slightly rearranged form of the classical Forster equation:
where f ;'Po and f i8" were respectively the fractional absorption of Tyr and T r p a t 280 nm, while and 6"' represented respectively the measured quantum yields of the protein or peptide a t 280- and 295-nm excitation wavelengths. Calculated distances R between Tyr 28 and T r p 35 chromophores were respectively 10.8 ( k 0 . 2 ) and 12.8 ( a 0 . 2 ) A in the absence and in the presence of excess Zn2+,regardless of pH (between 5.5 and 8 ) .
DISCUSSION The NC protein NCplO of MoMuLV has one CysHis motif, where the single Tyr and T r p residues are located in positions 28 and 35, respectively. Zn2+ binding to NCplO was monitored using the fluorescence of T r p 35 or, alternatively, Tyr 28. We report here that NCplO, NCplO 24-42, and ( Phe3")NCp10 24 42 bind one Zn2+per protein molecule, with a n experimental binding constant a s high as 10'3M-1. Such a high value was not unexpected since the nucleoprotein from the related Rauscher murine leukemia virus bind Zn2+ with a n estimated binding constant of 10"MM-'a t p H 7.0.17 Moreover, apparent Zn" -binding constants greater than 10'*M-' were also inferred for the zinc finger domains of Xenopus transcription factor IIIA.26The zinc finger domain of NC protein appears to possess most, if not all, of the information necessary and sufficient for Zn2+ binding, since its experimental Zn2+-binding constant is very close to that of NC protein. However, the finger peptide cannot activate dimerization of the retroviral RNA and annealing of replication
primer tRNA. In fact, both these processes that take place in the course of MoMuLV virion assembly require the complete NC protein (Darlix, unpublished results). It can be concluded that the finger domain is necessary but not sufficient for the full biochemical activity of NCplO. From the dependence of the experimental Zn2+binding constants upon p H in the range of 5-8, the presence of ligands with pKa in this p H range was deduced. Since both Cys and His residues could be protonated in this p H range, they are likely candidates for this p H dependence. T o investigate this, a simple model was built assuming that only the species with fully deprotonated Cys and His could bind Zn2+with a high binding constant K M e . Thus, taking a pK1 of 6.2 for His ( a s discussed further) and assuming that each Cys protonates with a n identical pK2 constant, the measured Zn '+ -binding constant Kexpcould be readily expressed as
log(Kexp)= 10g(KMe)- log[l X [H]
+ lopK'
+ ( 1 0 p K 2 )X2 [ H I 2+ (10L'"2)'3 X [ H I 3 X (1
3 0.21
+ lopK1X
[ H I ) ] (8)
J
\,
c 0.0
N
0
100
200
300
400
500
Time (s) Figure 3. Kinetics of Zn2+release from NCplO 24-42. Zn2+-loadedNCplO 24-42,8.85 pM, in 50 m M HEPES, 0.1M KC1, pH 7.5, was mixed with 1 m M EDTA. The solid curve was the experimental record while the dashed curve corresponded to the computer fit to the monoexponential Eq. ( 4 ) , using the dissociation rate constant of Table 11.
904
MELY ET AL.
>- 1.4
t
* 1.2 z 1.0
z
w 0.8 0
5 0.6 0 0.4
8 0.2 i 0.0 3
340
300
380
420
460
500
WAVELENGTH (nm) Figure 4. Fluorescence spectra of apo-NCplO 24-42 (a,b) and zinc-saturated (c,d) NCplO 24-42. The peptide ( 2 0 p M ) was dissolved in 50 m M HEPES, 0.1 M KCl, pH 7.5. Spectra were recorded for excitation wavelengths set a t 280 (a,d) and 295 (b,c) nm, respectively.
Experimental data were fitted very satisfactorily with Eq. ( 8 ) giving pK2 = 8.6 (k0.2), which is fully compatible with the usual pK, range of Cys and KMe = 1.8 ( k0.2 ) 1016M-’. Including supplementary Zn2+-binding species (e.g., monoprotonated ones) worsened the fit, which validates the hypothesis that probably, as for yeast transcription factor zinc finger domains, 27 only species with fully deprotonated His and Cys could bind Zn2+. In addition, as the p H dependence of the experimental binding constant Kexpwas much stronger than that of the dissociation rate constant kOff, it was tempting to deduce that Kexpvariations stemmed essentially from variations in the association rate constant k,, . However, a t p H above 7, the calculated k,, became unrealistically high ( k o n> 5.10’’ M-’ X s-’), suggesting that the binding scheme is not governed by a simple twostate equilibrium between the Zn 2+ -complexed and uncomplexed forms, but probably involves a more complicated pathway with different conformational species. It could thus be inferred that the measured binding constants Kexpin fact correspond to apparent binding constants and that further information is needed to fully describe this binding mechanism. Replacing Cys 29 by Ser causes a drastic decrease in the Zn2+-bindingaffinity and results in a strong inhibition of genomic RNA encapsidation in vivo7 indicating that the Zn2+-saturated NCplO protein is necessary for genomic RNA packaging into virions. However, we do not yet know a t which stage of genome packaging nor how Zn2+ bound to NC intervenes. This is a t present under investigation. When all Cys residues were oxidized or chemically modified, the Zn2+-binding properties vanished, confirming the critical involvement of these residues in Zn2+binding. In clear contrast, replacing T r p 35
by Phe left the Zn2+affinity of the zinc finger domain unaffected. Similarly, replacing T r p 35 by Ser did not affect Zn2+binding to NCplO, but did result in the absence of MuLV RNA dimerization by mutant NC protein in vitro and genomic RNA packaging in vivo ( Prats and Darlix, unpublished results). These data indicate that T r p 35 within the zinc finger is not essential for Zn2+binding, but does play a critical role in MuLV dimerization and packaging, probably through specific NC-RNA interactions. Since NCplO, NCplO 24-42, and (Ser”) NCplO 24-43 fluoresced, both in the absence and presence of Zn2+,with a maximum at about 351 nm, a fully polar environment for T r p 35 in the apo- as well as in the Zn2+-loadedspecies is suggested. In the absence of Zn2+, the quantum yields of these compounds were rather low and sigmoidally dependent _____
0.07
-
A
0.06 -
6 0.03 4
5
6
7
8
7
8
PH
4
5
6
PH Figure 5. pH dependence of NCplO 24-42 quantum yield. Quantum yields of both apo-NCplO 24-42 ( A ) and Znz+-loaded ( B ) NCplO 24-42 were determined as described under Materials and Methods. Excitation wavelength was set either a t 280 nm (.), where both Tyr and Trp residues absorb, or 295 nm ( 0 ) , where only Trp residues absorb. Each quantum yield was the mean of 3-8 independent determinations, and its standard deviation was between 0.001 and 0.003. The solid lines in Figure 5A were drawn by using Eq. 10, while those in Figure 5B were hand-drawn.
ZINC-BINDING AFFINITIES OF MoMuLV NCplO
on the pH, with a midtransition point a t about pH 6.2. The localization of the His 34 residue in the vicinity of the T r p residue, the usual pK, range of His residues (about 6-7) and the well-known potential of His residues to act as fluorescence quenchers under their protonated formz8 suggest that the midtransition point corresponds to the pK of His 34. T o further verify this point, let us consider the equilibrium between M and MH, the protein (or peptide) species with deprotonated and protonated His respectively, governed by the His protonation constant K , . T h e measured quantum yield @,,, a t a given pH can be expressed as
where f M is the fraction of protein with deprotonated His, while @M and @MH are the quantum yields of M (measured a t p H 8) and MH (measured a t p H 4), respectively. From Eq. ( 9 ) and the protonation equilibrium, the following relation can be derived:
where [ H ] is deduced from the p H value. Experimental quantum yields measured with either 280or 295-nm excitation wavelengths were adequately fitted with Eq. (10) (Figure 5 A ) , giving pKl = 6.2 ( kO.1) , thus strengthening the foregoing hypothesis. In the presence of Z n Z f , since the maximum emission wavelength is almost unchanged, the high quantum yields above p H 5.5 are not due to a change in the polarity of T r p 35 environment but rather to a rearrangement of T r p environment, which takes the Trp 35 residue away from neighboring quenching groups. These quenching groups clearly not include His 34 residue, since a t p H 8, where His was fully deprotonated ( a n d thus no longer able to quench fluorescence), the apoprotein quantum yield remained much lower than that of the Zn2+-loaded form. Moreover, radiationless Tyr + T r p energy transfer suggests that Tyr 28 and T r p 35 residues move off by about 2 A. Due to the probable flexibility of the apo species, this value represents in fact a n average distance between the various conformations of the apo species and the zinc-saturated one. This small hut highly reproducible value thus confirmed the existence of a zinc-induced rearrangement. These Zn 2 i -induced structural modifications associated with fluorescence quantum yield changes probably represent a discrete local reorganization rather than a n important conformational change, since only minor Zn 2+ -induced secondary structure changes have been detected in the related Rauscher
905
MuLV NCplO protein.':' The decrease in 2n"binding affinities at acidic p H is reflected by decreased quantum yields, and thus, below pH 5.5, NCplO or NCplO 24-42 were present as a mixture of apo- and Zn*+-Ioadedspecies. This was further confirmed by the relationship between the quantum yield a t a given acidic p H and the concentration of added Zn2+(data not shown). Since similar spectral characteristics were observed a t p H 7.5 for NCplO, NCplO 24-42, or (Ser") NCplO 24-43, it is suggested that the T r p environment is not greatly modified by amino acids outside the zinc finger sequence nor by the nature of the amino acid a t position 29, and that NCplO and the two finger peptides have similar conformations in both their apo- and Zn'+-loaded forms. In conclusion, our data indicate that Moloney MuLV NCplO is a Zn2+-bindingprotein and suggest that Zn2+bound to NC protein is biologically relevant. This tight binding of one Zn" per N C molecule probably plays a structural role in the retroviral NC protein and work is now in progress to verify this point. This work was supported by grants from Centre National de la Recherche Scientifique, Institut National de la Santb et de la Recherche Mkdicale (Grant INSERM No. 861010), Agence Nationale de la Recherche sur le SIDA, and Universitk Louis Pasteur.
REFERENCES 1. Coffin, J. M. (1984) in RNA Tumor Viruses, vol. I, Weiss, R., Teich N., Varmus, H. & Coffin, J. M., Eds.,
2.
3. 4.
5. 6. 7.
8.
9.
Cold Spring Harbor Laboratories Press, Cold Spring Harbor, NY, pp. 261-368. Murti, K. G., Bondurant, M. & Tereba, A. (1981) J. Virol. 37,411-419. Bieth, E., Gabus, C. & Darlix, J. L. (1990) Nucleic Acids Res. 18, 119-127. Prats, A. C., Roy, C., Wang, P., Erard, M., Housset, V., Gabus, C., Paoletti, C. & Darlix, J. I,. (1990) J. Virol. 64, 774-783. Mbric, C. & Spahr, P. F. (1986) J . Virol. 60, 450459. Mbric, C. & Goff, S.P. (1989) J . Virol. 63, 15581568. Gorelick, R. J., Henderson, L. E., Hanser, J. P. & Rein, A. (1988)Proc. Natl. Acad. Sci. USA 85,84208424. Prats, A. C., Sarih, L., Gabus, C., Litvak, S., Keith, G. & Darlix, J. L. (1988) EmboJ. 7 , 1777-1783. Barat, C., Lullien, V., Schatz, O., Keith, G., Nugeyre, M. T., Gruninger-Leitch, F., Barrk-Sinoussi, F., Le Grice, S. F. J. & Darlix, J. L. ( 1989) EmboJ. 8,32793285.
906
MELY ET AL.
10. Darlix, J. L. & Spahr, P. F. (1982) J . Mol. Biol. 189, 147-161. 11. Berg, J. M. (1986) Science 232, 485-487. 12. Jentoft, J. E., Smith, L. M., Fu, X., Johnson, M. & Leis, J. ( 1988) Proc. Natl. Acad. Sci. U S A 85, 70947098. 13. Roberts, W. J., Pan, T., Elliott, J. I., Coleman, J. E. & Williams, K. R. (1989) Biochemistry 28, 1004310047. 14. South, T. L., Kim, B. & Summers, M. F. (1989) J. Am. Chem. SOC. 111, 395-396. 15. Summers, M. F., South, T. L., Kim, B. & Hare, D. R. (1990) Biochemistry 29,329-340. 16. Green, L. M. & Berg, J. M. (1989) Proc. Natl. Acad. Sci. U S A 86, 4047-4051. 17. Green, L. M. & Berg, J. M. (1990) Proc. Natl. Acad. Sci. U S A 87, 6403-6407. 18. Nissen-Meyer, J. & Abraham, A. K. (1980) J . Mol. Biol. 142, 19-28. 19. Cornille, F., Mely, Y., Ficheux, D., Savignol, I., GCrard, D., Darlix, J. L., Fournih-Zaluski, M. C. & Roques, B. P. ( 1990) Znt. J. Peptide Protein Res. 36,551-558.
20. Ellman, G. L. (1959) Arch. Biochem. Biophys. 82, 70-77. 21. GBrard, D., Lemieux, G. & Laustriat, G. (1975) Photochem. Photobiol. 22, 89-95. 22. Martell, A. E. & Smith, R. M. (1974) Critical Stability Constants, Vol. 1, Plenum Press, New York. 23. Schiller, P. W. & St. Hilaire, J. (1980) J. Med. Chem. 23, 290-294. 24. Eisinger, J., Feuer, B. & Lamola, A. A. (1969) Biochemistry 8, 3908-3915. 25. FourniC-Zaluski, M. C., Belleney, J., Lux, B., Durieux, C., GCrard, D., Gacel, G., Maigret, B. & Roques, B. P. ( 1986) Biochemistry 25, 3778-3787. 26. Shang, Z., Liao, Y. D., Wu, F. Y. H. & Wu, C. W. ( 1989) Biochemistry 28,9790-9795. 27. Parraga, G., Horvath, S., Hood, L., Young, E. T. & Klevit, R. E. (1990) Proc. Natl. Acad. Sci. USA 87, 137-141. 28. Shinitzky, M. & Goldman, R. ( 1967) Eur. J . Biochem. 3, 139-144. Received November 26, 1990 Accepted March 12, 1991
