- Email: [email protected]
Spontaneous dimerization of retroviral MoMuLV RNA
Biochimie (1993) 75,681-686 0 SociCt6 franCaise de biochimie et biologie molCculaire / Elsevier, Paris
681
Spontaneous dimerization of retroviral MoMuLV J Paolettia, M Mougelb, N Tounektia, PM Girarda, C Ehresmannb, B Ehresmannb arr,_;& /I,, D;nnL;-in 1TDA lA7 PnTDC ,,-,I IllAn InrCCDnA _.._ P,-:ll, nn,-,..I:,, I..A:+..~ [email protected]._.._ D_..--. nAOAn “I‘&CC UF “,“LrUrrIcc)“,\rl IT, G,.‘\LI U~IU “/-TV IIVLIL~~UVI, IL&T UUJlUV~-nUUS&y, 7W3UV VllC~JUlJ; hUPR 9002 du CNRS, Institut de Biologie Mol&ulaire et Cellulaire du CNRS, 15, rue Rent! Descartes, 67084 Strasbowg, France LUN~LCL~T
VC3rm~Ul1113,
1llJlIlUl
~1:11-J-.X.
(Received 30 April 1993; accepted 2 May 1993)
Summary - The genome of the Moloney muriue leukemia virus (MoMuLV) is composed of two identical RNA molecules joined at their 5’ ends by the dimer linkage structure (DLS). Dimerization sequences are located within the PSI encapsidation domain. We present here an overview of the work we have performed on spontaneous dimerization of a MoMuLV RNA fragment encompassing the PSI domain in order to z,.derstand the mechanism by which retroviral RNA dimerization takes place. We present kinetical, thermodynamical and conformational evidence which leads to the conclusion that the PSI domain is a structurally independent domain and that conformational changes are triggered by the dimerization process. We conclude that at least one particular region (nucleotides 278-309) of the RNA is directly involved in the process while the conformation of some other regions is changed probably because of a long-range effect. retroviral RNA / dimerization / structure / physical chemistry
Introduction - __
Cells infected by the Moioney murine ieukemia virus (MoMuLV) synthesise virus-specific RNA by transcription of the integrated proviral DNA. The primary transcript is of genome size and the RNA must serve as a precursor for the genome of the progeny, as a messenger for the translation of gag and pol proteins and as a premessenger for generating the env mRNA
VI.
A unique property of all retroviruses examined so far, is that their unspliced RNA is packaged in virions illsa dimer [2]. Genomic RNA, extracted from retroviruses of different origins has been shown, by electron wuwtrnwnnv “y”I.*v”IvyJ,
tn LV he “I
.linked . . . ..“a.
thrnwuh . . ..VU~..
the . ..”
dimer u L... -1
Iintarrc? 1.1Y.l~~
structut c (DLS) located close to their 5’ end [3-6]. More recently, it has been suggested that the dimerization of unspliced RNA from Rous sarcoma virus (RSV) and MoMuLV is controlled by a c&element lnpatd .“VH.“U
in L.. the . ..”
pncancidatinn “..“Uy”.Ub...V..
Anmain uv...w...,
the . ..”
PST * Y*
reuinn ‘Vb”..
[7-lo], and is stimulated by the nucleocapsid protein or the gag precursor [l-14]. These data favour the conclusion that dimerization and encapsidation could be related processes and therefore the dimer linkage structure could serve as a signal for encapsidation of unspliced RNA. It has also been shown that dimerization is associated with reverse transcription through interstrand switching [ 15-171 and with genetic recombination [ 181. As a consequence, dimerization of
retroviral RNA represents an important process for driving and modulating the retroviral cycle. We then initiated a study dealing with some physical-chemicai parameters of dimerization, together with attempts to elucidate the changes in the RNA conformation which could be associated with this process. Here we summarise the main results obtained so far in studying the MoMuLV dimerization. Dimerization is a slow process which involves conformational modifications in the RNA structure. It induces specific changes in three defined locations of the highly structured encapsidation region which possesses an intrinsic conformation. Most probably it also induces, in domains outside this region, ie the splice donor tcnl site PIP i&i&on region_ ‘I allosteric tran\irl, _.I_ mrl _.._ the 1-1_ o”0 ---------sitions which may reflect functiona. relevance. Analytical study of the dimerization of in vitro mmersated bm.“” _w__ MnMnI S.--m._-_. ,V RN-A_
The analytical study of MoMuLV RNA dimerization was performed using synthetic RNA fragments obtained by in vitro transcription, containing the leader (620 nucleotides) and part of the 5’ gag sequences. As shown for RSV RNA [9] and HIV-l RNA [ 191, MoMuLV retroviral RNA fragments spontaneously dimerise in vitro [20]. When studying MoMuLV dimerization, it is possible, starting from
682
0' 0
I
20
I
40
I
I
60
fore, the value of about 50°C implies probably that the T,,,for the dimer should be greater than 55°C. By performing a denaturation of the dimer in 250 mM NaCI, we determined a T,,,value equal to 75°C which corresponds to a rather stable structure. In order to compare the thermal stability of the in vitro formed dimer with the previously measured stability of the genome length MoMuLV RNA dimer (2 x 8832 nucleotides [22]), we derived the T,,, from denaturation experiments performed in 100 mM NaCl (fig 2). The T,,, thus calculated (57.2”C) compares quite well with the T,,, of the genome length dimer, indicating that the nature of the interactions between the RNA subunits is comparable whether the 70s natural dimer or the in vitro reconstituted dimer is considered. The enthalpy and entropy parameters derived from the concentration dependence of the RNA on T,,, (AH” = -208 f 14 kcal/mol and AS” = -600 + 41 cal/mol) lead to a AG”37” value which is equal to -23 kcal/mol deg.
Temperature ("C) Fig 1. Percentages of dimerization obtained with the RNA fragment encompassing nucleotides l-725 as a function of temperature. The RNA at a concentration of 1.3 x 10-7 M was denaturated by heating at 95°C for 2 min. Aliquots were then incubated for I h in 50 mM Tris-HCl (pH 7), 250 mM NaCl at different temperatures. The percentage of dimcr was derived from analysis of 1% agarose gel ekctrophoresis under native conditions.
pure tnonomeric forms, to obtain pure dimeric stsuctures and we took advantage of this result to study the kinetics of the dimerir.ation process of a 725~nucleotide fragtnent starting at position +I of the genome. The dimer formation is a slow process whiLh depends an RNA, Na61 and magnesium concentrations. The apparent rate constant was estimated from the rate data by applying the equation for a bimolecular reaction of the type M + M < = > D, and the value of the rate Ctinstant calculated for experimental conditions corresponding to 250 mM NaCl and 37°C was 550 + 1SO M--W 1201. Such a value is far from corresponding to a pure diffusion-controlled reaction [21] and probably implies a conformational change in the structure of the RNA molecule. In the same manner, the extent of dimerization depends on temperature and reaches a maximum between 50 and 55OC with a rate constant (50 mM TrisHCI, pH 7, 250 mM NaCI) approximately 2500 M-W (tig 1) when, in the same conditions, dimerization of HIV-l RNA is optimum at about 37°C [ 191. This optimum in the temperature of dimerization represents a balance between an enhancement of the rate constant and the thermal stability of the dimer. There-
loo: . 808
60 40 20
OO._‘20
40
60
Temperature
80
100
(“Cl
Fig 2. Thermal stability of the l-725 RNA dimer as a function of temperature. After denaturation at 95°C for 2 min and chilling on ice, the RNA was incubated at 55°C for 30 min at a concentration of 1 x 10-S M (complete dimerization). The dimer thus obtained was then diluted in 10 mM Tris-HCl (pH 7), 100 mM NaCl, 1 mM EDTA at a RNA concentration of 1.3 x 1D-7 M and dialysed for 45 min in the same buffer. Aliquots were then incubated for 5 min at temperatures ranging from 20 to 80°C and analysed on 1% agarose gel electrophoresis under native conditions.
683 Following the free energy parameters used to predict RNA duplex stability [23], these thermodynrtnlic parameters do correspond to a minimum of 1O-l 5 fully paired nucleotides, assuming a classical WatsonCrick base pairing mechanism and a random sequence for the dimerization process. This represents the minimum RNA length involved in the dimer linkage structure. In order to check if the assumption of a dimerization process based on classical Watson-Crick pairing was correct and to check the specificity of the viral RNA dimerization, the dimerization of an antisense RNA (nucleotides 567 to -30 nt) was followed in parallel with the dimerization of the sense RNA (nucleotides l-725 nt). The results reported by Tounekti et al [24] show that RNA dimerization is not observable when an antisense RNA is used, in accordance with the result found by Marquet et al [ 191 in the case of HIV-l antisense RNA. These findings strongly suggest that RNA dimerization does not proceed only through regular hybridisation implying Watson-Crick base pairing; a mechanism involving non-canonical base-pairs should be considered. This hypothesis is supported by the great stability of the dimer toward denaturation by urea and formamide under conditions in which most tertiary and secondary structures are disrupted. More than 50% dimer is still present in 2.9 M urea and 37% formamide and some dimeric forms persist even in 4.7 M urea and 59% formamide [24]. Nevertheless, an apparent discrepancy exists between the urea-formamide stability and the thermal stability since a 57°C T,, for the dimer in 0.1 M NaCl is quite reasonable for a regular Watson-Crick base pairing. Since it has been suggested that the dimer linkage structure is part of the encapsidation region (PSI region), we compared the kinetics of dimer formation of the RNA fragment encompassing nucleotides l725 to the kinetics obtained when using the 350nucleotide long PSI fragment (2 15-565). As shown in figure 3, kinetics are about the same for both fragments with a rate constant for PSI dimerization which is three times higher than for the l-725 fragment. In addition, the thermal stability and the thermodynamic parameters corresponding to the dimer formed from the PSI fragment are about the same as those observed when studying the dimerization of the l-725 fragment [20]. These results seem to imply that the PSI region contains all the information in terms of structure and sequence required for dimerization. In an attempt to gain further understanding of the conformational modifications associated with dimer formation, we initiated a study to investigate the conformation of the PSI region in the 725-nucleotide fragment and of the PSI fragment itself. In both cases, ihe monomeric and the dimeric forms have been studted using chemical probing [25].
0)' 0
20
40 60 Time (mn)
80
100
Fig 3. Kinetics of dimerization
of the l-725 nucleotide RNA fragment and of the PSI fragment (2 15-565 nt) at a concentration of 1.3 10-7 M. These experiments were done at 50°C in 50 mM Tris-HCI (pH 7), 250 mM NaCl and the percentage of dimer was followed on 1% agarose gel electrophoresis. 0, l-725 nt RNA fragment. , PSI fragment.
Conformational study of the PSI domain involved in dimerization of MoMuLV RNA The susceptibility of the PSI region (nucleotides 215565) to chemical probes has been monitored on in vitro transcribed fragments corresponding either to the isolated PSI domain or to the 5’ terminal 725-nucleotide fragment. For details see [24]. In short, the reactivity pattern of the PSI region in the monomeric form is essentially the same in the large fragment ( l-725) or in the PSI fragment (215-565) . This result strongly suggests that the PSI domain possesses an intrinsic conformation independent of the neighbouring sequences whether they are located downstream from the 5’ end or upstream from the 3’ end of the PSI sequence. Based upon a computer prediction using the program of Zucker [26], a secondary structure model which integrates the experimental data of the chemical probing and a phylogenetic comparison has been derived [241. Figure 4a shows the secondary structure structure model of the PSI region in its monomeric form, together with the reactivity changes induced by dimerization, leading to an alternative model corresponding to the PSI region in its dimer form (fig 4b). These
684
G C-W
#G--C
ec -
G
od - =tJ
G
OA 220-.:
9
C G OA G CO 223’ ecel uo c’=
G
GG
Co
nou oOC i’m-G
A0 G* C
Euo
OA F CO UO
c
G
c
GO
G
UO
G
C
eou
u
clOC
G
210-G
U
uo-230
Fig 4, Secondary structure rrodcl of’ region PSI of MoMuLV (figure from (241). a) Proposed secondary structure of the monomer. The reactivities at the tested Watson-Crick positions A(NI ). G(Nl), U(N3), C(N3) are indicated: (0). highly reactive: (0). moderately reactive: and (0). marginally reactive nucleotides. Unreactive nucleotides are not indicated: (o), nucleotides whose reactivity was undetermined. (Cl). nucleotides that become reactive under semi-denaturing conditions. Dimerization induced reactivity changes are indicated: ong or weak reduction ( ,CI ). strong or weak enhancement (+,+). Adenine residues that are reactive at N7 are encircled: ). highly reactive: (0). m rately reactive: (0). marginally reactive. Base-pairs are represented by hyphens only when helices are supported by both reactivity data and phylogenetic comparison. Different possible pairings are shown by broken lines. A possible alternative conformation for region 282 to 309 that may exist as a dynamir equilibrium is shown at the left. A putative alternative conformation for nucleotides 210 to 236 is shown at the right. b) Possible secondary structure that could account for the dimerization-induced Reactivity changes. Reactivity at Watson-Crick positions and at N7: same symbols as in (a); (*), nucleotides that display reactivity changes upon dimerization. Long-range interactions are numbered l-3. It is not yet clear whether junction 2 or 3 occurs in the dimer.
685 models are formed of three distinct domains: domain 1 (nucleotides 237-381), domain 2 (nucleotides 38 l437), and domain 3 which corresponds to the 3’ terminal domain of the PSI region). These domains are further condensed by long-range interactions (junctions I, 2, 3). Domains 1 and 3 are affected by the dimerization process whereas domain 2 is unchanged whether the PSI region is monomeric or dimeric. Domain 1 This domain is essentially organised in five stem-loop structures. Two of these hairpins (3 10-352 and 35% 374) are very stable due to their high content in GC pairs and do not show any significant changes of l’tactivity after dimerization. On the contrary, the region corresponding to nucleotides 278-309 shows a high level of reactivity in the monomer form indicating a rather unstable structure. Our experimental results can be interpreted as the co-existence of two weak stemloop structures differing by the location of bulged residues. Once dimerization takes place, we observe a strong reduction of the reactivity of these nucleotides which can be explained by a direct interaction between two RNA molecules or a conformational rearrangement. These findings are in accordance with the fact that in the monomer form this region can be hybridised with an oligodeoxyribonucleotide complementary to the sequence 280-3 10 whereas this sequence is inaccessible to the same oligodeoxyribonucleotide in the dimer [ 141. Domain 2 This domain, which contains two hairpins, does not show any changes of reactivity after dimerization. The first hairpin (382-399) is very stable and corresponds to a regular helix made of five adjacent GC pairs and a seven-base loop. The second one (405-434) corresponds to a regular AU rich stem closed by a ten-base loop. Nucleotides from this loop present a low extent of reactivity which could be better expkined, through phylogenetic considerations, by an internal organisation than by a long-range interaction. I. Domain 3 This domain, which corresponds to the 3’ terminal domain of the PSI region, has a conformation which is clearly dependent upon the nature, monomeric or dimeric, of the RNA molecule. When the reactivity pattern in the monomer form is consistent with a long stem-loop structure (468-5 16); this structure evolves toward a much shorter stem-loop (468-488) when the RNA molecule has been subjected to dimerization. This change of conformation can be interpreted either
as a direct involvement of the region around nucleotide 480 in the dimerization process or as an allosteric effect induced by dimerization. It has been shown --- . __ (Girard et al, manuscript in preparation) that a RNA fragment containing nucleotides 215-364 is able to dimerize and that the stability of the dimer is not very different from the one corresponding to the PSI or the l-7 15 dimer. This result, together with results dealing with deletion mutants in region 374-480, would favour the fact that the conformational change induced by dimerization in the region around nucleotide 480 is probably due to an indirect allosteric effect. From these results it appears that the PSI region forms an independent domain in which all of the information necessary for dimerization is present. Kinetic and thermodynamic considerations indicate that the thermal stability of the dimer formed from a PSI fragment is not quite different from the stability observed for the l-725 fragment [20] or the one measured for the genomic dimer [22]. From the kinetic point of view, lengthening the fragment from PSI to l-725 inhibits slightly the dimerization: the rate constant for dimerization of the PSI fragment is about three times higher than the one corresponding to dimerization of the l-725 fragment leading to slower kinetics. This is conformed by looking at the dimerfzation kinetics of longer fragments encompassing nucleotides l-1030 or l-1515. These kinetics show a much lower rate of dimerization without any apparent change in the thermal stability of the dimer which is formed. From a structural point of view, the PST region appears as an independent and organised domain which has the same structure whether studied alone or as part of the l-725 fragment. This observation was confirmed by studies dealing with the structure of the PSl region in the complete MoMuLV genomic RNA isolated from viral particles or in the virion [27]. The PSI region, in these cases, appears to fold into a structure quite comparable to the one observed for synthetic fragments. The most important changes in the chemical reactivity pattern of the nucleotides take place in between nucleotides 278 and 309, a region which seems necessary for the dimerization. It could be postulated that this particular region of the genome could be directly involved in the dimerization process. On the other hand, the conformation of the region neighbouring nucleotide 480 is strongly changed upon dimerization but this change is more likely due to an allosteric effect triggered by dimerization. These long-range conformational effects are important in terms of function since dimerization could then induce consequences in regions not directly involved in the process (Mougel et al, manuscript in preparation). However, the precise mechanism, in terms of hybridisation. by
686 which the association of two RNA molecules takes place is not very clear yet. It is difficult to accommodate the high stability of the dimer in presence of urea and formamide and the possibility to form heterodimers between MoMuLV RNA fragments and other retroviral RNA fragments [ 193, with a purely WatsonCrick type of hybridisation. Among possible non-Watson-Crick interactions, the formation of four stranded structures could be a good candidate to explain the dimerization of retroviral RNA [ 191. Such structures have been shown to appear in both deoxyribo- and ribonucleic acids. Acknowledgments
9 IO 11 12 13 14 IS 16 17 18 I9
Bieth E, Garbus C, Darlix JL (1990) Nucleic Acids Res 18, 119-127 Pratz AC, Roy C, Wang P, Erard M, Housset V, Gabus C, Paoletti C, Darlix JL ( 1990) J Viral 64,774-783 Merit C, Spahr PF ( 1986) J Viral 60,450-459 Ciorelik RJ, Henderson JE, Hanser JP, Rein A (1988) Proc Nat1 Acad Sci USA 85,8420-8424 Merit C, Goff S ( 1989) J Viral 60,450-459 ?ratz AC, Sarih L, Gabus C, Litvak S, Keith G, Darlix JL (1988) EMBO J 7,1777-1783 Panganibam A, Fiore D ( 1988) Science 241, 1064-1069 Huo G, Taylor J ( 1990) J Viral 64,4321-4338 Temin HM (1991) I?XJI&SGenetics 7,71-74 Hu WS, Temin HM (1990) Proc Nat1 Acad Sci USA 87, 1556-1560 Marquet R, Baudin F, Gabus C, Darlix JL, Mougel M, Ehresmann C, Ehresmann B ( 1991) Nucleic Acids Res 19, 2349-2357
This work was supported by grantsfo JP and BE from Agence Nationale de Recherche sue le SODA (ANRS) and from Centre National de In Recherche Scientifique (CNRS): Interface Chimie-Biologie.
20
28
Roy C, Tounekti N, Mougel M, Darlix JL, Paoletti C, Ehresmann C, Ehresmann BR, Paoletti J ( 1990) Nucleic Acids Res 18,7287-7292 Alberty RA, Hammes GG (1958) J Phys Chem 62, 156 I59 Maize1 J, Bender W, Hu S, Duesberg PH, Davidson N (1978) J Wrol25,384-394 Freier SM, Kierzek R, Jaeger JA, Sugimoto N, Caruthers MH, Neilson T, Turner DH (1986) Pm Nat! Acad Sci USA 83.9373-9377 Tounekti N, Mougel M, Roy C, Marquet R, Darlix JL, Paoletti J, Ehresmann B, Ehresmann C (1992) J MO/ Biol 223,205-220 Ehresmann C, Baudin F, Mougel M, Rornby P, Ebel JP, Ehresmann B ( 1987) Nucleic Acids Res 22,9 109-9128 Zucker M ( 1989) Science 244,48-52 Alford RL, Honda S, Lawrence S, Belmont JW (1991) J Vi/v/ 183, 61 I-619 Lee JS, Evans DH, Morgan AR (1980) N~cclcic* Acids Rc’s
29
Guschlbauer
21 22
References
23
Coffin JM (1984) IN: RNA Trrrnnr~ Virrrses (Weiss R. Teich N, Varmus H, Coffin G, eds) Cold Spring Harbor Laboratory, Cold Spring Harbor, NY, vol I, 261-368 Varmus HE ( 1982) S<*ier~c*p 2 16, 8 12-820 Bender W, Davidson N (1976) C& 7.595-607 Kung HJ, Hu S, Bender W, Bailey JM, Davidson N, Nicolson MO, MC Allister RM ( 1976) C& 7,609-620 Bender W, Chicn YH, Chattopadkny S, Vogt PK, Gardner MR, Davidson N ( 1978) ,/ Vi~r)/25, X88-896 Murti KG, Bondurant M, Tercba A (1981) J I/~/Y)/37, 41 l-419 rZ;:;n R, Mulligan RC, BaltinIore D ( I983) Cdl 33, I S3tiilllll
R,
Bi~ltimOre D ( 1985) ./ VIM 54, 401-407
24 2s 26 27
8,4305-4320 Strwt
W, Chantot JF, Thiele 1-s I I
L9yrr 8, 49
D (1990) ,I Bicmol
681
Spontaneous dimerization of retroviral MoMuLV J Paolettia, M Mougelb, N Tounektia, PM Girarda, C Ehresmannb, B Ehresmannb arr,_;& /I,, D;nnL;-in 1TDA lA7 PnTDC ,,-,I IllAn InrCCDnA _.._ P,-:ll, nn,-,..I:,, I..A:+..~ [email protected]._.._ D_..--. nAOAn “I‘&CC UF “,“LrUrrIcc)“,\rl IT, G,.‘\LI U~IU “/-TV IIVLIL~~UVI, IL&T UUJlUV~-nUUS&y, 7W3UV VllC~JUlJ; hUPR 9002 du CNRS, Institut de Biologie Mol&ulaire et Cellulaire du CNRS, 15, rue Rent! Descartes, 67084 Strasbowg, France LUN~LCL~T
VC3rm~Ul1113,
1llJlIlUl
~1:11-J-.X.
(Received 30 April 1993; accepted 2 May 1993)
Summary - The genome of the Moloney muriue leukemia virus (MoMuLV) is composed of two identical RNA molecules joined at their 5’ ends by the dimer linkage structure (DLS). Dimerization sequences are located within the PSI encapsidation domain. We present here an overview of the work we have performed on spontaneous dimerization of a MoMuLV RNA fragment encompassing the PSI domain in order to z,.derstand the mechanism by which retroviral RNA dimerization takes place. We present kinetical, thermodynamical and conformational evidence which leads to the conclusion that the PSI domain is a structurally independent domain and that conformational changes are triggered by the dimerization process. We conclude that at least one particular region (nucleotides 278-309) of the RNA is directly involved in the process while the conformation of some other regions is changed probably because of a long-range effect. retroviral RNA / dimerization / structure / physical chemistry
Introduction - __
Cells infected by the Moioney murine ieukemia virus (MoMuLV) synthesise virus-specific RNA by transcription of the integrated proviral DNA. The primary transcript is of genome size and the RNA must serve as a precursor for the genome of the progeny, as a messenger for the translation of gag and pol proteins and as a premessenger for generating the env mRNA
VI.
A unique property of all retroviruses examined so far, is that their unspliced RNA is packaged in virions illsa dimer [2]. Genomic RNA, extracted from retroviruses of different origins has been shown, by electron wuwtrnwnnv “y”I.*v”IvyJ,
tn LV he “I
.linked . . . ..“a.
thrnwuh . . ..VU~..
the . ..”
dimer u L... -1
Iintarrc? 1.1Y.l~~
structut c (DLS) located close to their 5’ end [3-6]. More recently, it has been suggested that the dimerization of unspliced RNA from Rous sarcoma virus (RSV) and MoMuLV is controlled by a c&element lnpatd .“VH.“U
in L.. the . ..”
pncancidatinn “..“Uy”.Ub...V..
Anmain uv...w...,
the . ..”
PST * Y*
reuinn ‘Vb”..
[7-lo], and is stimulated by the nucleocapsid protein or the gag precursor [l-14]. These data favour the conclusion that dimerization and encapsidation could be related processes and therefore the dimer linkage structure could serve as a signal for encapsidation of unspliced RNA. It has also been shown that dimerization is associated with reverse transcription through interstrand switching [ 15-171 and with genetic recombination [ 181. As a consequence, dimerization of
retroviral RNA represents an important process for driving and modulating the retroviral cycle. We then initiated a study dealing with some physical-chemicai parameters of dimerization, together with attempts to elucidate the changes in the RNA conformation which could be associated with this process. Here we summarise the main results obtained so far in studying the MoMuLV dimerization. Dimerization is a slow process which involves conformational modifications in the RNA structure. It induces specific changes in three defined locations of the highly structured encapsidation region which possesses an intrinsic conformation. Most probably it also induces, in domains outside this region, ie the splice donor tcnl site PIP i&i&on region_ ‘I allosteric tran\irl, _.I_ mrl _.._ the 1-1_ o”0 ---------sitions which may reflect functiona. relevance. Analytical study of the dimerization of in vitro mmersated bm.“” _w__ MnMnI S.--m._-_. ,V RN-A_
The analytical study of MoMuLV RNA dimerization was performed using synthetic RNA fragments obtained by in vitro transcription, containing the leader (620 nucleotides) and part of the 5’ gag sequences. As shown for RSV RNA [9] and HIV-l RNA [ 191, MoMuLV retroviral RNA fragments spontaneously dimerise in vitro [20]. When studying MoMuLV dimerization, it is possible, starting from
682
0' 0
I
20
I
40
I
I
60
fore, the value of about 50°C implies probably that the T,,,for the dimer should be greater than 55°C. By performing a denaturation of the dimer in 250 mM NaCI, we determined a T,,,value equal to 75°C which corresponds to a rather stable structure. In order to compare the thermal stability of the in vitro formed dimer with the previously measured stability of the genome length MoMuLV RNA dimer (2 x 8832 nucleotides [22]), we derived the T,,, from denaturation experiments performed in 100 mM NaCl (fig 2). The T,,, thus calculated (57.2”C) compares quite well with the T,,, of the genome length dimer, indicating that the nature of the interactions between the RNA subunits is comparable whether the 70s natural dimer or the in vitro reconstituted dimer is considered. The enthalpy and entropy parameters derived from the concentration dependence of the RNA on T,,, (AH” = -208 f 14 kcal/mol and AS” = -600 + 41 cal/mol) lead to a AG”37” value which is equal to -23 kcal/mol deg.
Temperature ("C) Fig 1. Percentages of dimerization obtained with the RNA fragment encompassing nucleotides l-725 as a function of temperature. The RNA at a concentration of 1.3 x 10-7 M was denaturated by heating at 95°C for 2 min. Aliquots were then incubated for I h in 50 mM Tris-HCl (pH 7), 250 mM NaCl at different temperatures. The percentage of dimcr was derived from analysis of 1% agarose gel ekctrophoresis under native conditions.
pure tnonomeric forms, to obtain pure dimeric stsuctures and we took advantage of this result to study the kinetics of the dimerir.ation process of a 725~nucleotide fragtnent starting at position +I of the genome. The dimer formation is a slow process whiLh depends an RNA, Na61 and magnesium concentrations. The apparent rate constant was estimated from the rate data by applying the equation for a bimolecular reaction of the type M + M < = > D, and the value of the rate Ctinstant calculated for experimental conditions corresponding to 250 mM NaCl and 37°C was 550 + 1SO M--W 1201. Such a value is far from corresponding to a pure diffusion-controlled reaction [21] and probably implies a conformational change in the structure of the RNA molecule. In the same manner, the extent of dimerization depends on temperature and reaches a maximum between 50 and 55OC with a rate constant (50 mM TrisHCI, pH 7, 250 mM NaCI) approximately 2500 M-W (tig 1) when, in the same conditions, dimerization of HIV-l RNA is optimum at about 37°C [ 191. This optimum in the temperature of dimerization represents a balance between an enhancement of the rate constant and the thermal stability of the dimer. There-
loo: . 808
60 40 20
OO._‘20
40
60
Temperature
80
100
(“Cl
Fig 2. Thermal stability of the l-725 RNA dimer as a function of temperature. After denaturation at 95°C for 2 min and chilling on ice, the RNA was incubated at 55°C for 30 min at a concentration of 1 x 10-S M (complete dimerization). The dimer thus obtained was then diluted in 10 mM Tris-HCl (pH 7), 100 mM NaCl, 1 mM EDTA at a RNA concentration of 1.3 x 1D-7 M and dialysed for 45 min in the same buffer. Aliquots were then incubated for 5 min at temperatures ranging from 20 to 80°C and analysed on 1% agarose gel electrophoresis under native conditions.
683 Following the free energy parameters used to predict RNA duplex stability [23], these thermodynrtnlic parameters do correspond to a minimum of 1O-l 5 fully paired nucleotides, assuming a classical WatsonCrick base pairing mechanism and a random sequence for the dimerization process. This represents the minimum RNA length involved in the dimer linkage structure. In order to check if the assumption of a dimerization process based on classical Watson-Crick pairing was correct and to check the specificity of the viral RNA dimerization, the dimerization of an antisense RNA (nucleotides 567 to -30 nt) was followed in parallel with the dimerization of the sense RNA (nucleotides l-725 nt). The results reported by Tounekti et al [24] show that RNA dimerization is not observable when an antisense RNA is used, in accordance with the result found by Marquet et al [ 191 in the case of HIV-l antisense RNA. These findings strongly suggest that RNA dimerization does not proceed only through regular hybridisation implying Watson-Crick base pairing; a mechanism involving non-canonical base-pairs should be considered. This hypothesis is supported by the great stability of the dimer toward denaturation by urea and formamide under conditions in which most tertiary and secondary structures are disrupted. More than 50% dimer is still present in 2.9 M urea and 37% formamide and some dimeric forms persist even in 4.7 M urea and 59% formamide [24]. Nevertheless, an apparent discrepancy exists between the urea-formamide stability and the thermal stability since a 57°C T,, for the dimer in 0.1 M NaCl is quite reasonable for a regular Watson-Crick base pairing. Since it has been suggested that the dimer linkage structure is part of the encapsidation region (PSI region), we compared the kinetics of dimer formation of the RNA fragment encompassing nucleotides l725 to the kinetics obtained when using the 350nucleotide long PSI fragment (2 15-565). As shown in figure 3, kinetics are about the same for both fragments with a rate constant for PSI dimerization which is three times higher than for the l-725 fragment. In addition, the thermal stability and the thermodynamic parameters corresponding to the dimer formed from the PSI fragment are about the same as those observed when studying the dimerization of the l-725 fragment [20]. These results seem to imply that the PSI region contains all the information in terms of structure and sequence required for dimerization. In an attempt to gain further understanding of the conformational modifications associated with dimer formation, we initiated a study to investigate the conformation of the PSI region in the 725-nucleotide fragment and of the PSI fragment itself. In both cases, ihe monomeric and the dimeric forms have been studted using chemical probing [25].
0)' 0
20
40 60 Time (mn)
80
100
Fig 3. Kinetics of dimerization
of the l-725 nucleotide RNA fragment and of the PSI fragment (2 15-565 nt) at a concentration of 1.3 10-7 M. These experiments were done at 50°C in 50 mM Tris-HCI (pH 7), 250 mM NaCl and the percentage of dimer was followed on 1% agarose gel electrophoresis. 0, l-725 nt RNA fragment. , PSI fragment.
Conformational study of the PSI domain involved in dimerization of MoMuLV RNA The susceptibility of the PSI region (nucleotides 215565) to chemical probes has been monitored on in vitro transcribed fragments corresponding either to the isolated PSI domain or to the 5’ terminal 725-nucleotide fragment. For details see [24]. In short, the reactivity pattern of the PSI region in the monomeric form is essentially the same in the large fragment ( l-725) or in the PSI fragment (215-565) . This result strongly suggests that the PSI domain possesses an intrinsic conformation independent of the neighbouring sequences whether they are located downstream from the 5’ end or upstream from the 3’ end of the PSI sequence. Based upon a computer prediction using the program of Zucker [26], a secondary structure model which integrates the experimental data of the chemical probing and a phylogenetic comparison has been derived [241. Figure 4a shows the secondary structure structure model of the PSI region in its monomeric form, together with the reactivity changes induced by dimerization, leading to an alternative model corresponding to the PSI region in its dimer form (fig 4b). These
684
G C-W
#G--C
ec -
G
od - =tJ
G
OA 220-.:
9
C G OA G CO 223’ ecel uo c’=
G
GG
Co
nou oOC i’m-G
A0 G* C
Euo
OA F CO UO
c
G
c
GO
G
UO
G
C
eou
u
clOC
G
210-G
U
uo-230
Fig 4, Secondary structure rrodcl of’ region PSI of MoMuLV (figure from (241). a) Proposed secondary structure of the monomer. The reactivities at the tested Watson-Crick positions A(NI ). G(Nl), U(N3), C(N3) are indicated: (0). highly reactive: (0). moderately reactive: and (0). marginally reactive nucleotides. Unreactive nucleotides are not indicated: (o), nucleotides whose reactivity was undetermined. (Cl). nucleotides that become reactive under semi-denaturing conditions. Dimerization induced reactivity changes are indicated: ong or weak reduction ( ,CI ). strong or weak enhancement (+,+). Adenine residues that are reactive at N7 are encircled: ). highly reactive: (0). m rately reactive: (0). marginally reactive. Base-pairs are represented by hyphens only when helices are supported by both reactivity data and phylogenetic comparison. Different possible pairings are shown by broken lines. A possible alternative conformation for region 282 to 309 that may exist as a dynamir equilibrium is shown at the left. A putative alternative conformation for nucleotides 210 to 236 is shown at the right. b) Possible secondary structure that could account for the dimerization-induced Reactivity changes. Reactivity at Watson-Crick positions and at N7: same symbols as in (a); (*), nucleotides that display reactivity changes upon dimerization. Long-range interactions are numbered l-3. It is not yet clear whether junction 2 or 3 occurs in the dimer.
685 models are formed of three distinct domains: domain 1 (nucleotides 237-381), domain 2 (nucleotides 38 l437), and domain 3 which corresponds to the 3’ terminal domain of the PSI region). These domains are further condensed by long-range interactions (junctions I, 2, 3). Domains 1 and 3 are affected by the dimerization process whereas domain 2 is unchanged whether the PSI region is monomeric or dimeric. Domain 1 This domain is essentially organised in five stem-loop structures. Two of these hairpins (3 10-352 and 35% 374) are very stable due to their high content in GC pairs and do not show any significant changes of l’tactivity after dimerization. On the contrary, the region corresponding to nucleotides 278-309 shows a high level of reactivity in the monomer form indicating a rather unstable structure. Our experimental results can be interpreted as the co-existence of two weak stemloop structures differing by the location of bulged residues. Once dimerization takes place, we observe a strong reduction of the reactivity of these nucleotides which can be explained by a direct interaction between two RNA molecules or a conformational rearrangement. These findings are in accordance with the fact that in the monomer form this region can be hybridised with an oligodeoxyribonucleotide complementary to the sequence 280-3 10 whereas this sequence is inaccessible to the same oligodeoxyribonucleotide in the dimer [ 141. Domain 2 This domain, which contains two hairpins, does not show any changes of reactivity after dimerization. The first hairpin (382-399) is very stable and corresponds to a regular helix made of five adjacent GC pairs and a seven-base loop. The second one (405-434) corresponds to a regular AU rich stem closed by a ten-base loop. Nucleotides from this loop present a low extent of reactivity which could be better expkined, through phylogenetic considerations, by an internal organisation than by a long-range interaction. I. Domain 3 This domain, which corresponds to the 3’ terminal domain of the PSI region, has a conformation which is clearly dependent upon the nature, monomeric or dimeric, of the RNA molecule. When the reactivity pattern in the monomer form is consistent with a long stem-loop structure (468-5 16); this structure evolves toward a much shorter stem-loop (468-488) when the RNA molecule has been subjected to dimerization. This change of conformation can be interpreted either
as a direct involvement of the region around nucleotide 480 in the dimerization process or as an allosteric effect induced by dimerization. It has been shown --- . __ (Girard et al, manuscript in preparation) that a RNA fragment containing nucleotides 215-364 is able to dimerize and that the stability of the dimer is not very different from the one corresponding to the PSI or the l-7 15 dimer. This result, together with results dealing with deletion mutants in region 374-480, would favour the fact that the conformational change induced by dimerization in the region around nucleotide 480 is probably due to an indirect allosteric effect. From these results it appears that the PSI region forms an independent domain in which all of the information necessary for dimerization is present. Kinetic and thermodynamic considerations indicate that the thermal stability of the dimer formed from a PSI fragment is not quite different from the stability observed for the l-725 fragment [20] or the one measured for the genomic dimer [22]. From the kinetic point of view, lengthening the fragment from PSI to l-725 inhibits slightly the dimerization: the rate constant for dimerization of the PSI fragment is about three times higher than the one corresponding to dimerization of the l-725 fragment leading to slower kinetics. This is conformed by looking at the dimerfzation kinetics of longer fragments encompassing nucleotides l-1030 or l-1515. These kinetics show a much lower rate of dimerization without any apparent change in the thermal stability of the dimer which is formed. From a structural point of view, the PST region appears as an independent and organised domain which has the same structure whether studied alone or as part of the l-725 fragment. This observation was confirmed by studies dealing with the structure of the PSl region in the complete MoMuLV genomic RNA isolated from viral particles or in the virion [27]. The PSI region, in these cases, appears to fold into a structure quite comparable to the one observed for synthetic fragments. The most important changes in the chemical reactivity pattern of the nucleotides take place in between nucleotides 278 and 309, a region which seems necessary for the dimerization. It could be postulated that this particular region of the genome could be directly involved in the dimerization process. On the other hand, the conformation of the region neighbouring nucleotide 480 is strongly changed upon dimerization but this change is more likely due to an allosteric effect triggered by dimerization. These long-range conformational effects are important in terms of function since dimerization could then induce consequences in regions not directly involved in the process (Mougel et al, manuscript in preparation). However, the precise mechanism, in terms of hybridisation. by
686 which the association of two RNA molecules takes place is not very clear yet. It is difficult to accommodate the high stability of the dimer in presence of urea and formamide and the possibility to form heterodimers between MoMuLV RNA fragments and other retroviral RNA fragments [ 193, with a purely WatsonCrick type of hybridisation. Among possible non-Watson-Crick interactions, the formation of four stranded structures could be a good candidate to explain the dimerization of retroviral RNA [ 191. Such structures have been shown to appear in both deoxyribo- and ribonucleic acids. Acknowledgments
9 IO 11 12 13 14 IS 16 17 18 I9
Bieth E, Garbus C, Darlix JL (1990) Nucleic Acids Res 18, 119-127 Pratz AC, Roy C, Wang P, Erard M, Housset V, Gabus C, Paoletti C, Darlix JL ( 1990) J Viral 64,774-783 Merit C, Spahr PF ( 1986) J Viral 60,450-459 Ciorelik RJ, Henderson JE, Hanser JP, Rein A (1988) Proc Nat1 Acad Sci USA 85,8420-8424 Merit C, Goff S ( 1989) J Viral 60,450-459 ?ratz AC, Sarih L, Gabus C, Litvak S, Keith G, Darlix JL (1988) EMBO J 7,1777-1783 Panganibam A, Fiore D ( 1988) Science 241, 1064-1069 Huo G, Taylor J ( 1990) J Viral 64,4321-4338 Temin HM (1991) I?XJI&SGenetics 7,71-74 Hu WS, Temin HM (1990) Proc Nat1 Acad Sci USA 87, 1556-1560 Marquet R, Baudin F, Gabus C, Darlix JL, Mougel M, Ehresmann C, Ehresmann B ( 1991) Nucleic Acids Res 19, 2349-2357
This work was supported by grantsfo JP and BE from Agence Nationale de Recherche sue le SODA (ANRS) and from Centre National de In Recherche Scientifique (CNRS): Interface Chimie-Biologie.
20
28
Roy C, Tounekti N, Mougel M, Darlix JL, Paoletti C, Ehresmann C, Ehresmann BR, Paoletti J ( 1990) Nucleic Acids Res 18,7287-7292 Alberty RA, Hammes GG (1958) J Phys Chem 62, 156 I59 Maize1 J, Bender W, Hu S, Duesberg PH, Davidson N (1978) J Wrol25,384-394 Freier SM, Kierzek R, Jaeger JA, Sugimoto N, Caruthers MH, Neilson T, Turner DH (1986) Pm Nat! Acad Sci USA 83.9373-9377 Tounekti N, Mougel M, Roy C, Marquet R, Darlix JL, Paoletti J, Ehresmann B, Ehresmann C (1992) J MO/ Biol 223,205-220 Ehresmann C, Baudin F, Mougel M, Rornby P, Ebel JP, Ehresmann B ( 1987) Nucleic Acids Res 22,9 109-9128 Zucker M ( 1989) Science 244,48-52 Alford RL, Honda S, Lawrence S, Belmont JW (1991) J Vi/v/ 183, 61 I-619 Lee JS, Evans DH, Morgan AR (1980) N~cclcic* Acids Rc’s
29
Guschlbauer
21 22
References
23
Coffin JM (1984) IN: RNA Trrrnnr~ Virrrses (Weiss R. Teich N, Varmus H, Coffin G, eds) Cold Spring Harbor Laboratory, Cold Spring Harbor, NY, vol I, 261-368 Varmus HE ( 1982) S<*ier~c*p 2 16, 8 12-820 Bender W, Davidson N (1976) C& 7.595-607 Kung HJ, Hu S, Bender W, Bailey JM, Davidson N, Nicolson MO, MC Allister RM ( 1976) C& 7,609-620 Bender W, Chicn YH, Chattopadkny S, Vogt PK, Gardner MR, Davidson N ( 1978) ,/ Vi~r)/25, X88-896 Murti KG, Bondurant M, Tercba A (1981) J I/~/Y)/37, 41 l-419 rZ;:;n R, Mulligan RC, BaltinIore D ( I983) Cdl 33, I S3tiilllll
R,
Bi~ltimOre D ( 1985) ./ VIM 54, 401-407
24 2s 26 27
8,4305-4320 Strwt
W, Chantot JF, Thiele 1-s I I
L9yrr 8, 49
D (1990) ,I Bicmol