Dimerization of retroviral genomic RNAs: structural and functional implications

Biochimie (1996) 78,639-653 0 SociCtC frangaise de biochimie et biologie molCculaire / Elsevier, Paris

Dimerization

of retroviral genomic RNAs: Structural and functional implications JC Paillart, R Marquet, E Skripkin, C Ehresmann,

B Ehresmann*

UPR 9002 CNRS, Institut de Biologie Mole’culaire et Cellulaire, 15, rue R-Descartes, (Received 3 May 1996; accepted

67084 Strasbourg cedex, France

11 June 1996)

Summary -

Retroviruses are a family of widespread small animal viruses at the origin of a diversity of diseases. They share common structural and functional properties such as reverse transcription of their RNA genome and integration of the proviral DNA into the host genome, and have the particularity of packaging a diploid genome. The genome of all retroviruses is composed of two homologous RNA molecules that are non-covalently linked near their 5’ end in a region called the dimer linkage structure (DLS). There is now considerable

evidence that a specific site (or sites) in the 5’ leader region of all retroviruses, located either upstream or/and downstream of the major splice donor site, is involved in the dimer linkage. For MoMuLV and especially HIV-l, it was shown that dimerization is initiated at a stem-loop structure named the dimerization initiation site (DIS). The DIS of HIV-1 and related regions in other retroviruses corresponds to a highly conserved structure with a self-complementary loop sequence, that is involved in a typical loop-loop ‘kissing’ complex which can be further stabilized by long distance interactions or by conformational rearrangements. RNA interactions involved in the viral RNA dimer were postulated to regulate several key steps in retroviral cycle, such as: i) translation and encapsidation: the arrest of gag translation imposed by the highly structured DLS-encapsidation signal would leave the RNA genome available for the encapsidation machinery; and ii) recombination during reverse transcription: the presence of two RNA molecules in particles would be necessary for variability and viability of virus progeny and the ordered structure imposed by the DLS would be required for efficient reverse transcription. dimerization

/ retrovirus

/ packaging

/ RNA structure

Introduction The Retroviridae comprise a large family of viruses that are responsible for many diseases, including rapid and longlatency malignancies, arthritis and other autoimmune diseases, neurological disorders and immunodeficiencies [ 11. Retroviruses have been traditionally divided into three subfamilies, depending on their pathogenicity: the Oncovirinae (ASLV, MLV, MMTV, MPMV, HTLV, BLV), the Lentivirinae (HIV, Visna-Maedi virus, EIAV, CAEV) and the Spumavirinae (SSRV, HSRV). Despite their variety of

*Correspondence and reprints Abbreviations: ASLV, avian sarcoma and leukosis virus; BKD, endogenous baboon virus; CAEV, caprine arthritis-encephalitis virus; DIS, dimerization initiation site; DLS, dimer linkage structure; EIAV, equine infectious anemia virus; HaMSV, Harvey mouse sarcoma virus; HBV, hepatitis B virus; HIV-l or 2, human immunodeficiency virus type 1 or 2; HSRV, human spumaretrovirus; HTLV-I, human T lymphotropic virus type I; MoMuLV, Moloney murine leukemia virus; MPMV, Mason-Pfizer monkey virus; NCp, nucleocapsid protein; PBS, primer binding site; Psi, packaging site; RD- 114, endogenous feline virus; REV-A, reticuloendotheliosis virus type A; RSV, Rous sarcoma virus; RT, reverse transcriptase; SD, splice donor; SIV, simian immunodeficiency virus; SNV, spleen necrosis virus; SSRV, simian spumaretrovirus; VL 30, viruslike retrotransposon 30; WoMV, wolly monkey sarcoma virus

pathogenicities and host cell interactions, all retroviruses share strong similitudes in virion structure, genome organization and replication mode. Retroviruses have several characteristic features: 1) they are the only RNA viruses to be truly diploid [2, 31; 2) they need a specific cellular tRNA to initiate the reverse transcription [4] and an RNA/DNA-dependent DNA polymerase to reverse transcribe their genome into proviral DNA [5, 61; 3) the produced double-stranded DNA is integrated into the host chromosomal DNA leading to the formation of a provirus; 4) their genome is produced essentially by the cellular transcriptional machinery; and finally 5) they are the only (+) RNA viruses whose genome does not serve directly as mRNA immediately after infection. The retroviral genome consists of two single stranded positive RNA molecules (8-11 kb) with a 5’ cap and 3’ poly(A) sequence as in eucaryotic mRNAs [7]. The two homologous RNAs have been shown to be associated near their 5’ end [g-lo]. The order of the genes encoding viral proteins is invariably gag-pol-env although for some viruses (lentiviruses, HTLV, BLV, HSRV), this sequence is interrupted by several open reading frames coding for viral accessory/regulatory proteins which are involved in regulation of virus expression [I 11. These viruses are called complex viruses in contrast with simple viruses (MLV) which do not code for such kind of regulatory proteins. Several retroviruses have transduced oncogenes such as the gene order on the RNA genome is also disrupted [7].

640 The present review considers dimerization of genomic RNAs of retroviruses for which a dimer linkage structure (DLS) has been identified. We first describe the localization of the DLS in the 5’ end of the genome. We describe the mechanism of initiation of RNA dimerization and finally, discuss the implications of dimerization during the viral cycle and subsequent approaches in the perspective of using dimerization as a therapeutic target.

Localization

of the DLS in the 5’ leader region

Electron microscopy studies on dimers of genomic RNA extracted from RSV virions first revealed that the two molecules are held together at many contact points [ 121. It was further demonstrated that the most stable contact is located in a region close to the 5’ end of the genomic RNA in type C oncornaviruses RD- 114, BKD and WoMV [8,9,13]. This structural feature, which was named DLS (dimer linkage structure) [8, 93 was confirmed in a wide variety of murine and avian viruses [lo, 14-161 (fig 1). The DLS was esti-

mated to involve less than 50 nucleotides surrounding the position 466 f 9 nucleotides in MoMuLV and 5 11 f 28 nucleotides in RSV, thus encompassing parts of the 5’ leader region and of the 5’ end of the gag gene [lo]. The general orientation of the two RNA strands appeared to be parallel [8,9]. The DLS was proposed to be stabilized by hydrogen bonds or other non-covalent interactions, since the dimers dissociated into monomers under relatively mild conditions [8, 9, 15, 171. Precise localization of the DLS began in the late 1980s when Darlix and coworkers showed that synthetic RNAs corresponding to the 5’ end of retroviral genomes dimerize in vitro. In the following sections, we will focus on the DLS of viruses for which in vitro dimerization sites have been identified. Human retroviruses HIV-l is with MoMuLV the retrovirus whose RNA dimerization is the most studied. The region of the genomic HIV-l RNA involved in dimerization, the nature of the interactions between the two molecules, and the physico-chemical para-

Fig 1. Electron micrographs of RNA dimers. In vitro synthesized RNA corresponding to nucleotides l-725 of HIV-l (MN isolate) were dimerized in the presence of NCpl5 and spread on a grid after protein extraction and partial denaturation of RNA dimer in urea/formamide. Two dimers (D) and two monomers (M) are shown at a final magnification of x 225 000. The dimer linkage is indicated by an arrow. By courtesy of JL Darlix and M Erard.

641 meters controlling dimer formation are well documented. Darlix et al showed in 1990 that a fragment of genomic RNA encompassing the 5’ leader region and the 5’ part of the gag gene of HIV-1Mal is able to dimerize in vitro, and that this process is stimulated by the nucleocapsid protein precursor NCp15 [ 181. By using truncated and mutant RNA fragments, they identified a lOO-nucleotide sequence, located immediately downstream of the SD site and encompassing the start codon of the gag gene, as a c&element necessary and sufficient to promote in vitro dimerization in the presence of NCp15 (fig 2a). This region overlaps the encapsidation signal [ 19-2 l] (fig 2a). Soon after, Marquet et al defined physico-chemical parameters to study in vitro dimerization in absence of NCp15 and pointed out a conserved purine-rich motif (RGGARA, R = purine) which is present in the putative dimerization-encapsidation region of more than 40 retroviruses (fig 2a) [22]. Later, we showed that elements of dimerization are located upstream of the SD site [23]. Indeed, salt-induced dimerization of an RNAs corresponding to nucleotides 1-3 11 of HIV- 1, thus lacking the putative DLS, can be observed on agarose and polyacrylamide gels provided that Mg2+ (0.1 mM) is included in the gels and in the running buffer. Furthermore, dimers formed by RNA fragments containing or lacking sequences upstream of the SD site strongly differ by their dimerization properties (kinetics and cation dependence) as well as by their thermal stability [23]. We identified the bases that are essential for initiation of RNA dimerization by chemical modification interferences and site-directed mutagenesis [24, 251. This region (nucleotides 260-290) was shown to adopt a stem-loop structure with a self-complementary sequence in the loop (274GUGCAC279) [24, 261. We called this region the dimerization initiation site (DIS) [24]. A similar localization of the sequences required for dimerization of HIV-1Lai RNA was later determined by studying dimerization of truncated and mutated RNA fragments [27, 281, and inhibition of dimer formation by antisense DNA oligonucleotides [28]. Thus, the dimerization signal seems to be bipartite in HIV- 1 with an initiating (nucleotides 260290) and a stabilizing (nucleotides 3 11415) region [25]. Compared to HIV- 1, HIV-2 genomic RNA dimerization is not so well studied. It was shown that an RNA fragment corresponding to the R and U5 regions (l-255), upstream of the PBS is sufficient to promote dimerization of HIV-2 RNA [29]. As for HIV-l, it was suggested that there might be multiple contact points between the two genomic RNAs, ie upstream and downstream of the PBS (fig 2b). Murine retroviruses Prats et al showed, by studying truncated RNA fragments and inhibition of dimerization by antisense DNA oligonucleotides, that dimerization of MoMuLV RNA occurs in the 5’ leader region, and more precisely between position 280 and 330 (fig 2c) [16]. This region overlaps the previously defined packaging signal [30-341. NCplO and NCpl2 from

MoMuLV and RSV, respectively, are able to stimulate in vitro dimerization of synthetic MoMuLV RNAs [16, 35, 361, thus demonstrating the ubiquitous and general action of NC proteins on viral RNAs [37]. Analytical studies of MoMuLV RNA dimerization performed on synthetic RNAs (620 nucleotides) showed that dimerization is a slow process [38] that involves conformational modifications of the MoMuLV RNA structure [39]. More recently, the DLS was localized in the 278-303 region and was suggested to adopt two conformations as a function of temperature. Interestingly, one of them contains a self-complementary sequence in the loop (288UAGCUA293) [40]. HaMSV results from multiple recombinations between MoMuLV sequences, rat retrotransposon virus-like 30s (VL30) sequences and the c-ras cellular gene [41]. Mutagenesis and analytical studies showed that the 5’ end of HaMSV RNA (l-1088) is able to dimerize in vitro and that the 205-271 region of VL30 is sufficient to promote dimerization (fig 2d) [42, 431. This region contains guaninerich-, pyrimidine-richand purine-rich sequences in 5’, central and 3’ regions respectively [43]. The nucleocapsid protein NCplO from MoMuLV stimulates in vitro dimerization of VL30 RNA fragments containing the 205-271 region, which is also required for efficient RNA packaging [44]. Recently, the HaMSV dimerization signal was localized in the 205-226 region, which was proposed to adopt a stem-loop structure with a self-complementary sequence in the loop (213GGCC216) [45]. Avian retroviruses Electron microscopy [lo] and sequence analysis [46] localized the DLS of ASLV in the 5’ part of the gag gene, between positions 521 and 548. This region consists of an imperfect inverted repeat of 28 nucleotides that may adopt a stem-loop structure and is well conserved in the ASLV group [47]. In agreement with these studies, Bieth et al reported that: i) dimerization can be inhibited by antisense DNA oligonucleotides mapping around position 550; and ii) that region 208-270 is also required for efficient in vitro dimerization (fig 2e) [48]. NCp12 stimulates in vitro dimerization of RNA containing the two sites [48]. Both sites were shown to be involved in genomic RNA encapsidation [46,49,50]. More recently, it was also suggested from mutagenesis experiments that the 28 nucleotide inverted sequence could be involved in dimerization and would be part of a larger dimerization signal involving a guanine-rich region located between positions 496 and 530 [5 11. Dimerization studies on another avian retrovirus, REVA, revealed that the sequence required for genomic RNA encapsidation [52, 531 is also needed for efficient in vitro dimerization [54]. This region mapped between nucleotides 208 and 452 (fig 2f) and is part of a two stem-loop structure [55]. NCp 10 from MoMuLV enhanced dimerization of REV-A RNA containing the DLS/Psi domain [54].

642

PO’

I -----_____

v

_-----____ -------------

8

HIV-l

R

C

----A--_________

__*_____________

415

290 311

HIV-2

---------*II

MoMuLV R ___________ I

HaMSV R ____----__

1

330

280

420

_________________ __%&JT* * *___________*_ ____ 205

MoMuLV

e

-_----._____

US

215

d

----------

_____.._______.__L_______________pBs________-

260

b

u3R

env

1_]

L,

226

271

VL30

RSV

R ---___...---__L 496

REV-A R .-----.-e-1

Au--------

208

531 544

564

452

BLV

R -____----___ 322

369

Fig 2. Schematic representation of the 5’ leader region of some retroviruses. Top, general genetic organization repeat sequence; U5, unique 5’ end; PBS, primer binding site; SD, splice donor site; gag, pal, env, genes coding and surface proteins; U3, unique 3’ end). a) HIV-l; b) HIV-2; c) MoMuLV, d) HaMSV, e) RSV; f) REV-A; sequence, the main and secondary dimerization signals are indicated by thick black and gray lines respectively, on the RNA genome. The encapsidation site (thick black dashed lines) and sequences increasing its efficiency indicated above each sequence for comparison. Gray spots correspond to the consensus purine-rich sequences of most retroviruses (see text).

of the retroviral genome (R, for the structural, enzymatic and g) BLV For each leader together with their position (thick gray dashed lines) are found in the 5’ leader region

643 Bovine retroviruses The dimerization signal of bovine leukemia virus was first located in a region encompassing U5, the PBS and 30 bases 3’ of the PBS by using synthetic mutated RNAs and dimerization inhibition by antisense DNA oligonucleotides (fig 2g) [56,57]. Computer analysis of the 5’ 600 nucleotides of the BLV genomic RNA showed that the PBS and immediate downstream region is structured as a stem-loop with a selfcomplementary sequence in the loop (340UGAUCA345) [57,58]. This sequence, together with a sequence in the gag coding region, is part of the genomic RNA packaging signal [58, 591. Surprisingly, in vitro dimerization was not enhanced by the BLV nucleocapsid protein NCp12, but by the matrix protein MApl5 [56,57]. MAp15 was shown to form a specific complex with the dimer of RNA encompassing the U5-5’ gag region supposed to contain the BLV packaging signal.

Mechanism Contribution

of RNA dimerization of the purine-rich

motifs

The hypothesis that RNA dimerization could involve purine quartets came from a phylogenetic study [22]. A purine-rich consensus sequence, RGGARA (R = purine), was identified in the putative dimerization-encapsidation region of more than 40 retroviruses (fig 2). Regarding HIV-l, which contains three purine tracts in the putative DLS, physicochemical data were consistent with the fact that dimer formation of RNA 1-715 containing the putative DLS region (nucleotides 3 11-415) could involve non-canonical interactions (strong dependence of the cation size, high stability of dimer, no dimerization of the antisense RNA, formation of tetramers) [22]. Moreover, structural studies of a monomer species obtained at low Mg*+ concentration and a dimer form, obtained at higher Mg*+ concentration, revealed reactivity changes in the first purine-rich motif (nucleotides 338-349): positions Nl of guanine and N7 of adenine became protected in the dimer while they were accessible in the monomer [60]. These observations raised the possibility that dimerization could involve purine quartets composed of adenines and guanines (A2G2) [22]. The second purine-rich motif (nucleotides 377-385) was also proposed to be involved in the dimerization process forming G quartets (G4) stabilized by K+ [61, 621 or NH4+ [63]. Extensive studies of salt-induced dimerization of HIV1 RNA were done in our laboratory to discriminate the relative contribution of the DIS and of the purine-rich sequences (fig 3) [23, 251. First, deletion of the purine-rich motifs on RNA fragments corresponding to nucleotides l615 of HIV-1Mal had no effect on dimer formation [25]. Second, the thermal stability (T,,,) of the dimers formed by an RNA corresponding to nucleotides 1-3 I 1 of HIV- lMa1 or by an RNA lacking the four purine tracts decreases by

11-13°C compared to the dimer RNA encompassing the whole dimer linkage region (nucleotides 1-615) (40-42”C versus 53°C) (fig 3) [23, 251. Otherwise, Sundquist and Heaphy [62] and Awang and Sen [61] studied dimerization of short HIV-l NL4.3 or HIVlMa1 RNA fragments, respectively, starting immediately downstream of the SD site (corresponding to nucleotides 294-418 and 305-415 of HIV- 1Mal). These authors proposed that dimerization is mediated by G-quartets. However, we showed that the dimers formed by such RNAs totally differ from those formed by RNAs having a complete 5’ region. Thus, these RNA dimers formed with truncated fragments lacking the 5’ end are most likely artifacts (fig 3) [23]. Noteworthy, under identical ionic conditions, RNA corresponding to nucleotides 3 11-612 of HIV-1Mal forms highly stable dimers (T,,, = 6 1“C), while no dimerization of an RNA corresponding to nucleotides 305-615 was detected (fig 3) [64]. Indeed, we showed by chemical modification interference that dimerization of RNA 311-612 directly involves the three guanines at its 5’ end [64]. In this case, modification of these three residues on Watson-Crick and N7 positions totally prevented RNA dimerization. Thus, the purine-rich motifs seem to be involved in dimer stabilization event rather than in dimerization initiation [25]. Furthermore, the implication of guanine tetrads in the dimer of HIV-l genomic RNA was recently rejected on the basis of the thermal stability, under different ionic conditions, of a full-length RNA dimer extracted from virions [65]. The loop-loop

‘kissing’ complex model

As the mechanisms of HIV-l and MoMuLV RNA dimerization are better characterized than those of other retroviruses, the following section will mainly deal with these two viruses. We first identified the DIS of HIV-l by combining chemical modification interference with site-directed mutagenesis on a synthetic RNA corresponding to nucleotides l-6 15 of the HIV-lMa1 genome [24, 251. Modification interference allowed us to discriminate primary effects (nucleotides directly involved in RNA interaction) from secondary effects (conformational changes induced by dimerization). The 6 base self-complementary sequence (274GUGCAC279) was shown to be of importance in the process of RNA dimerization since: 1) single modification of three nucleotides in the loop (U275, G276 and A278); or 2) single base substitutions (U275 to C, A278 to G) abolishes dimerization even if the initially postulated DLS (region 311-415) is present in the RNA fragment (fig 3) [25]. Previous work demonstrated that HIV-l RNA dimerization is a fast process (dimerization constant of 2 l_tM-t min-i for the wild-type 1-615 RNA) which follows a conformation kinetic model [23]. Dimerization and dissociation occurred in the same time scale and equilibrium between these two processes is reached within a few minutes. In order to further confirm the direct involvement

644

311

415

++

311

612

305

>SO’=C

++

615

61 ‘C no dimer

278

no dimer 311

A

AM

A

++

42 “C

615

+++

53 T

615

+t

40°C

615

wild-tm.‘a

g

g

aggugc

u g c a

no dimer

1

no dimer

+++

53 “C

Fig 3. In vitro dimerization of various HIV-l RNA fragments. The upper diagram shows the 5’ leader region of HIV-l genomic RNA with the TAR region (truns-acting responsive element), the primer binding site (PBS), the dimerization initiation site (DIS), the SD site, the gag coding region (AUG at position 350) and the four consensus purine-rich sequences (gray). The position of the previously defined DLS [ 181 is also indicated and renamed 3’ DLS. RNAs are numbered from position +l of genomic RNA, deletions are mentioned by a break in the RNA and substitutions in the DIS loop are shaded in gray. The right part of the diagram summarizes the dimer yield (electrophoresis on agarose gel in the presence of 0.1 mM Mg’+) and the thermal stability of the dimers; (++) and (+++) correspond to 4&60% and 60-80% of dimers, respectively. Data are from [22-2.51.

of the loop sequence in the dimerization process, compensatory mutant restoring the self-complementarity in the loop (274GCGCGC279), as well as two truns-complementary mutants (U275 to C on one RNA, A278 to G on another one) were tested for dimerization (fig 3). They definitively proved that dimerization proceeds by Watson-Crick interactions between the self-complementary loop of each monomer [24, 251. These results were extended to the Lai isolate by studying truncated and deleted RNA fragments, and using antisense DNA oligonucleotides directed against the DIS in order to inhibit dimer formation [27, 28, 661. It was initially suggested that the loop-loop interaction might be followed by formation of an extended duplex [24,25,27, 28,661. In order to test the formation of an extended duplex, we constructed four pairs of truns-complementary mutants of RNA 1-615 that allows to discriminate the loop-loop

‘kissing’ complex from the extended duplex (fig 4) [67]. None of these mutants containing a point mutation in the loop (DIS C275 or DIS G278) are able to form homodimers, but do form heterodimers [25]. In addition, some of these mutants contained an inverted stem sequence (DIS HxC275 or DIS HxG278), so that pairs DIS C275/DIS HxG278 and DIS HxC275/DIS G278, which have incompatible stems can only form the ‘kissing’ complex but not the extended duplex (fig 4) [67]. We found that neither thermal stability nor equilibrium dissociation constant were significantly altered compared to wild-type values (T,,, = 53°C and & = 21 nM), thus demonstrating that the HIV-l dimer linkage is limited to the loop-loop ‘kissing’ complex and that fusion of the stem does not occur under conditions close to physiological conditions (37°C 5 mM Mg2+) (fig 5a) [67]. Moreover, the ‘kissing’ complex was shown to be dynamic, since

645

Wild-type 260 294

DIS C279DISHxG278 260

DIS C275/DIS6238

294

260 294 294

260

DIS HxC279DIS6278 260

DISHxC279DISHxG278

294

260 294 294

Kissing complex or extended duplex ?

260

Kissing complex only

Fig 4. Strategy

to discriminate between ‘kissing’ complex and extended duplex. Mutations (shaded nucleotides) were introduced in RNA 1-615. In DIS HxC27.5 and DIS HxG278, the stem sequence is partly inverted (gray boxes). Wild-type RNA and truns-complementary pairs DIS C275/DIS G278 and DIS HxC275/DIS HxG278 are able to form a ‘kissing’ complex or an extended duplex, while DIS C275/DIS HxG278 and DIS HxC275/DlS G278 can only form the ‘kissing’ complex [67].

RNAs of different size (55, 300, 600 nucleotides) containing the DIS are able to interfere with the preformed HIV-l RNA dimer [67]. This situation contrasts with the interactions between natural sense and antisense RNAs which usually form loop-loop ‘kissing’ complexes that are converted into extended intermolecular base-pairing (for reviews see [68, 691). Functional and structural analysis of wild-type and mutant RNAs suggest that the nine nucleotide loop has an intrinsic conformation and that the non complementary nucleotides surrounding the self-complementary sequence are needed to optimize the loop-loop recognition (Paillart JC, Skripkin E, Ehresmann B, Ehresmann C, Westhof E, Marquet R, to be published). However, recent studies showed that truncated RNAs from HIV- 1Lai are able to form two type of dimers depending of the temperature of dimer formation and electrophoresis conditions

[66]. The high-stability dimer, formed at 6O”C, resists semidenaturing conditions (without Mg2+) while the lowstability dimers, formed at 37”C, cannot. The first one would correspond to an extended duplex, the second one to a ‘kissing’ complex [66]. Similar results were obtained by studying the thermal stability of HIV-1Lai RNA dimer (RNA 70-402) formed at 37°C or 55°C [70]. Since these experiments were done either with truncated RNA fragments [66] and/or formed in the absence of Mg2+ [70], they can be hardly compared with our own results. For example, dimerization of the HIV-1Lai RNA in the absence of Mg2+ [70] is 25-fold slower than in the presence of 5 mM Mg2+ during the dimerization procedure (Paillart JC, Skripkin E, Ehresmann B, Ehresmann C, We&of E, Marquet R, to be published). The requirement for dimerization of MoMuLV RNA of a short self-complementary sequence in the 5’ leader region

646

AG Gu

360

G c

194

c

A

‘A 294

G

260

‘GGA

ACA

Loop-loop Kissing Complex

260 294

Extended Duplex

*

b

260

294

294

260

was first proposed by Prats et al [ 161 and further supported by structural probing [39], mutagenesis and thermodynamic analysis [40]. In vitro dimerization of MoMuLV RNA is slow and temperature dependent (optimum at 5O’C) [38, 401. At low temperature, the self-complementary 288UAGCUA293 sequence would be embedded in a stem, but a structural rearrangement would place it in a loop at high temperature (fig 5b) [40]. Dimerization initiation by the six bases in the loop (‘kissing’ loop-loop model) would constraint the loop and induce subsequent opening of the stems leading to stabilization of the structure by formation of 16 Watson-Crick base pairs (fig 5b) [40]. The orientation imposed by a loop-loop interaction is not in contradiction with the overall parallel orientation of the two molecules of murine retroviruses observed by electron microscopy [ 10, 141. The relative overall orientation of the molecules is imposed by the number of base pairs in the DIS stem and by the geometry of the loop-loop interaction. Although it is now quite clear that HIV-l and MoMuLV RNA dimerization occurs via the formation of a loop-loop ‘kissing’ interaction, the mechanism of dimerization of the other retroviral RNAs (HIV-2, RSV, BLV, HaMSV) is a matter of speculation. Conservation

MoMuLV

Temperature conformational

change

Fig 5. Loop-loop ‘kissing’ complex model of HIV-l and MoMuLV RNA dimerization. The self-complementary sequence in the loop is shaded. a. HIV-l RNA dimerization is restricted to the loop-loop ‘kissing’ complex in physiological conditions (see text). b. Dimerization model for MoMuLV (from [40]). An extended duplex is postulated to occur after ‘kissing’ complex formation.

of the DIS

The 5’ untranslated leader region of retroviruses contains several sequences which are required for viral replication. As described in figure 2, this region contains all cis-elements controlling transcription (TAR), polyadenylation, initiation of reverse transcription (PBS), dimerization of the genomic RNA (DLS), encapsidation (Psi), splicing (SD) and initiation of translation of gag gene. The secondary structure of retroviral leader RNAs was studied using a panel of methods including chemical and enzymatic probing, free energy minimization, sequence comparison and mutagenesis. Studies on HIV-l [26,60] and HIV-2 [71] (for review [72]), MoMuLV [39, 73-751, RSV [47, 761 and MPMV [77] revealed that the 5’ leader is highly structured and that the main structural features are conserved in different isolates of each virus. In the previous section, we mentioned that RNA dimerization requires a stem-loop structure, called DIS in HIV-l [24], located in the leader region. We performed an extensive sequence comparison of the DIS region of 29 HIV-l isolates and the two sequences from the closely related SIV isolated from chimpanzee (SIV-CPZ) [78] and putative DIS in the 5’ leader region of HIV-2 (23 strains including SIVSMM and SIV-MM) [78], avian (13 sequences) [76] and murine (10 sequences) [75] retroviruses (fig 6). The structure of the HIV-l DIS and of the putative DIS in other retroviruses is supported by phylogenetic analysis [26, 39, 60, 7 1,761, The DIS always consists of a hairpin structure with a conserved self-complementary sequence in the loop. Depending on the virus, two kinds of loops are observed. The first loop family (HIV-l, HIV-2, most avian retroviruses,

647

a

b

C-G-

Cl3.14

G-C-

~14

G-C -

A’9

C-G G-U 409

HIV-1 ; SIVCPZ I,OYI

436

HIV-2 ; SIV

: 2, MAL; 3, RF: 4. U455; 5. IBNG; 6. SM;

12. NDK: 13.3202A21; 14, ANT70; 15. CPZGAB:

d

C Gl

U-A-U-A G-C A-U U-A 2,3,4(3

U-G -

A6

-

C-GG-C

C5.6

258

+A197 A6 214

Avian

Murine

1. UR-2; 2. CT-IO: 3, RC: 4. AW, 6. RAV6; 7. SRA-I

1, MLV4070A : 2, MLM: 3. MLAPRO; 4. MSMPROCG; 5. MoMLV; 6. MaMSV; 7. FBRMSV

5. RAV-2;

a-,- :q

e fYwG

G-C

‘A

G-C U-A C

U

C

U

U-A G-C G-C

524

A

RSV

G

A-U’ A-U C-G

G-C

333

C-G C-G G-C

c-_(

G-C’ G-C C-G C-G

543

c

G

G-C 251

BLV

G-C 205

226

HaMSV

Cc--c C-G U--A 798

811

MMPV

Fig 6. Phylogenetic comparison of DIS stem-loop structure in the 5’ leader region of HIV-l and related regions in other retroviruses. Analysis results from comparison of: (a) 29 HIV-l sequences and two closely related SIV isolated from chimpanzee (SW-CPZ) [78]. b. 23 sequences from HIV-2 or SIV isolated from sooty mangabeys (SIV-SMM). c. 10 murine retroviruses [75]. d. 13 avian retroviruse sequences [76]. The drawn sequence corresponds to the most representative one: consensus B for HIV-l (eg Lai isolate), consensus A for HIV-2/SIV (eg, ROD isolate), consensus for 13 avian isolates (AMV isolate), and AKV isolate for murine retroviruses. Nucleotide changes occurring in the different isolates are indicated together with the number of the particular isolate. Deletions are shown by (A), and insertions by (+). The self-complementary sequence in the loop is shaded. e. Hairpin structure determined in the 5’ leader of RSV [.51], BLV [57, 581, HaMSV [45] and MMPV [77].

648 BLV) contains a six self-complementary sequence flanked by non-complementary nucleotides which are most often purines (fig 6a, b, d, e). The loops in the second class contains only four nucleotides and are totally self-complementary (murine, HaMSV and MMPV) (fig 6c, e). Noteworthy, true conservation is observed although the DIS sequence is quite heterogeneous, and the overall structure is well conserved. Most sequence changes in the stem do not alter the base-pairing scheme because changes are conservatives, ie G-C changed to A-U, or semi-conservatives, ie G-U changed to A-U. Regarding HIV-l/SIV-CPZ, the strongest co-variation is observed for HIV-lANT70 and SIVCPZANT isolates where three and two base changes, respectively, on one side of the DIS stem are compensated by base substitutions in the opposite strand (fig 6a) [24, 26, 721. Interestingly, the self-complementarity of the loop is observed in all retroviruses, including the avian group which shows the highest sequence diversity (fig 6d). Only two different self-complementary sequences are observed in the HIV-l/SIV-CPZ group, 274GCGCGC279 and GUGCAC (four isolates), with the unique exception of the HIV-1IBNG isolate where the C275 is replaced by U giving the GUGCGC sequence (fig 6a). However, if this sequence is still able to form a loop-loop ‘kissing’ complex by the formation of two U-G base-pairs. It was indeed shown by site-directed mutagenesis that an RNA with this sequence can dimerize even though the corresponding dimer is quite unstable [25]. In HIV-l and SIV-CPZ, the nature of the nucleotides surrounding the self-complementary loop sequence is also well conserved, with one (ANT70, CPZANT isolates) two (OYI, CPZGAB isolates) or mainly three nucleotides which are always purines (fig 6a). In ANT70, CPZANT isolates, the presence of only one nucleotide 5’ of the complementary loop sequence is the result of a co-variation of two purines resulting in the addition of an additional base-pair that reduces the size of the loop (fig 6a). The hairpin of the HIV-2/SIV group is also well conserved with an invariant self-complementary loop sequence (420GGUACC42.5) and, as for HIV-l, the presence of purines surrounding the self-complementary sequence (l-6 nucleotides) (fig 6b). The avian and murine system present the same hairpin conservation as for human retroviruses (fig 4d, 6~). The murine self-complementary loop sequence is invariably 288UAGCUA293 with the only exception of MLV4070A which has a UGGCCG sequence (fig 6~). Interestingly, this isolate compensates the loss of stability resulting from the formation of two external G-U base-pairs in the putative loop-loop complex by two internal G-C basepairs. In avian RNAs, three different self-complementary sequences are frequently observed: 264CGGCCG269, GGGCCC or CUGCAG, and sequence heterogeneity is rather high (fig 6d). Only RAV-0 does not possess a hairpin equivalent to the DIS. Thus, there is a high pressure of selection to conserve the overall secondary structure of the DIS in HIV- 1 or related structures in other retroviruses, that

probably reflects the biological role of this region in RNARNA interactions. Recently, it was pointed out by computer analysis or sitedirected mutagenesis that a stem-loop structure located in the 5’ untranslated region of HaMSV [45] and BLV [57-591 or in the 5’ gag region of RSV [48, 511 could be involved in RNA dimerization (fig 6e). The analysis of the 5’ leader region of MMPV also revealed the presence of an hairpin structure with a self-complementary loop sequence [77]. Role of the nucleocapsid

protein in RNA dimerization

The nucleocapsid protein (NCp) is derived from the C-terminal region of the Gag precursor polyprotein. Retroviral NC proteins are small, basic proteins, which bind singlestranded nucleic acids with no apparent specificity in vitro (for recent review [37]). They are characterized by one (MLV, SNV) or two (ASLV, HIV, SIV) zinc finger domains (except the spumaretrovirus group) with the motif C&C&H&C [79,80] and were shown to bind one or two zinc ions with high affinity [81, 821. NCp are associated with the genomic RNA in the mature virion [83,84] and are required for selection and encapsidation of genomic RNA. NCp was shown to stimulate the dimerization of synthetic RNA from RSV [4X], HIV-l [18, 85-871 and MoMuLV in vitro [16, 36, 88, 891, while full-length genomic RNA extracted from virions has never been observed to dimerize in vitro [16]. UV cross-linking [18, 861 and gel retardation experiments [86, 901 indicated that NCp7 binds around position 320 of HIV-l RNA in a pyrimidine-rich region. It was proposed that binding of NCp7 would induce conformational changes in the RNA structure leading to the dimerization of the genomic RNA [ 18, 861. However, it is now well established that dimerization of HIV-l RNA does not initiate at this position. In parallel, it was shown by filter binding assays that HIV- 1 NCp strongly binds the 5’ leader region [91-961. Moreover, NCp was shown to bind with high affinity to the DIS stem-loop (I& = 200 nM), and to the Psi hairpin located at nucleotides 320-350 [94,97]. Mutant proteins allowed to show that basic amino acids surrounding the zinc fingers are required for the annealing activity of NCp from HIV-l [87] and MoMuLV [36,88,89]. One cannot exclude that this activity would allow transition from the ‘kissing’ to the extended duplex [66, 67, 701, in a maturation step of the dimer [65, 981. In HIV-l, this transition is not observed under physiological conditions in the absence of NCp proteins [66, 701, but might be observed under particular conditions (high temperature, RNAs with truncated ends) in HIV-l [66, 701 or in HaMSV [45].

Biological

relevance

As far as we know, all genomic RNAs extracted from retroviral particles are in dimeric form. The localization of the dimer linkage sequence in the 5’ leader region, in proximity

649 of known functional elements, raises the possibility that RNA dimerization may regulate several functions such as translation, encapsidation and recombination. Translation Viral replication requires a highly regulated mechanism. Indeed, viral RNA synthesized in the nucleus may be either spliced and used as mRNA, or transported into the cytoplasm without splicing and used as mRNA for synthesis of the Gag and Gag-PO1 precursors or packaged by the viral machinery. Existence of two pools of unspliced genomic RNA, one serving as message and the other one as genome, was first proposed in 1973. MoMuLV 60-70s RNA was shown to be located at the cell surface, while 35s RNA was essentially identified in polyribosome fractions [99]. These fractions had half-lives of 3-4 h and 12 h, respectively [ 100, 1011. Thus, translation seems to act as a negative signal for encapsidation of unspliced viral mRNA. Otherwise, in vitro translation of RNAs containing the 5’ leader region of RSV [48] and HIV-l [60] was reduced by 50 to 90% when RNA was dimeric. As discussed in the preceding sections, the 5’ leader region of retroviruses is highly structured and conformational changes were shown to take place in vitro in the vicinity of the AUG gag gene upon dimerization of HIV-l and MoMuLV RNAs [60,74]. It is well known that secondary structure elements may inhibit translation of eucaryotic and procaryotic genes proceeding by the ribosomal scanning model [102-1051. Recently, this mechanism was shown to apply for the translation of HIV- 1 Gag, and it was proposed that the structural features of the dimerization-encapsidation region, rather than RNA dimerization, down regulate translation of this mRNA, at least in vitro [106]. The low translation level of the full-length genomic RNA molecules would allow binding of Gag proteins on the packaging signal and direct RNAs in the budding particles. A similar translation-encapsidation regulation was proposed for RSV [ 107, 1081. However, we showed recently that deletion of the DIS in a molecular clone of HIV-l did not significantly alter the translational level of structural Gag proteins (Paillart J-C, Berthoux L, Ottmann M, Darlix J-L, Marquet R, Ehresmann B, Ehresmann C, in press). Encapsidution Encapsidation and dimerization of genomic RNA have been suggested to be related processes in the course of replication. Indeed, the c&acting elements required for encapsidation of avian [46, 49, 50, 52, 109, 1101, murine [30-32, 441, human [19, 21, 111-1151 and bovine [58, 591 retroviruses overlap with those required for in vitro RNA dimerization [ 16, 18, 23, 24, 29,42, 48, 54,571. Moreover, NC proteins were shown to be crucial for in vivo RNA encapsidation (for review see [37]), via specific selection of the packaging site [ 116, 1171, and to truns-activate in vitro RNA dimerization [37]. Specific encapsidation of unspliced

genomic RNA implies that at least parts of the encapsidation signals should be located downstream of the SD site to prevent encapsidation of spliced viral RNAs. Recently, it was shown by us (Paillart J-C, Berthoux L, Ottmann M, Darlix J-L, Marquet R, Ehresmann B, Ehresmann C, in press) and by others [ 1181 that deletion of the DIS in a pNL4.3 molecular clone of HIV-l reduced the packaging efficiency by five- and two-fold, respectively. In the latter, authors showed that integrity of the stem of the DIS, together in coordination with the main packaging hairpin [97], was required for efficient encapsidation while interstrand base pairing between the two stem regions within the DIS hairpin was not [ 1181. Thus, in HIV- 1 as well as in RSV, for which both encapsidation and dimerization signals are multipartite and located upstream and downstream of the SD site, interactions involving regions separated by the SD site should occur to specify the genomic RNA recognition. The question whether dimerization takes place before or after encapsidation is still subject to controversy. On one hand, based upon isolation of fresh budding RSV particles, it was suggested that genomic RNA is packaged as two monomers and dimerizes after particle release [2, 1191. On the other hand, analysis of MoMuLV genomic RNA extracted viruses produced by cells treated with actinomycin D [ 1001 or from viruses containing NCp mutated in the zinc finger [89, 120-1221 revealed that, although the yield of packaged RNA is decreased, the RNA in the particles is dimeric, thus suggesting that dimerization takes place before encapsidation. This was also the case for REV-A for which dimers of genomic RNA were isolated from the cytoplasm of infected cells [54]. Furthermore, analysis of the thermal stability of the RNA dimer extracted from fresh particles [ 17,981 or protease deficient virions [65] suggests that the RNA dimer undergoes a stabilization, ie, ‘a maturation event, after release. These results provide strong support for encapsidation of an immature RNA dimer which is further stabilized after maturation of the Gag polyprotein precursor [37, 951. Recombination

during reverse transcription

Reverse transcription of the viral genome is ensured in the nucleocapsid by the virus-encoded reverse transcriptase (RT). Reverse transcription is at the origin of the high genetic variability of retroviral genome [ 123, 1241. Indeed, viral genomic RNA packaged as a dimer is replicated with a low fidelity [ 1251 and is an efficient template for recombination. The dimeric genome was shown to facilitate genetic exchange between the two RNA molecules [123], and frequent retroviral recombination during reverse transcription requires heterodimeric RNA [ 126, 1271 (for review see [128]). Experimental data indicated that recombinations could occur during the synthesis of minusstrand DNA, following the copy-choice model [128, 1291, or at a lower rate during the synthesis of plus-strand DNA, according to the displacement/assimilation model [ 130-

650 1321. In both cases, independently of the nature of the first DNA-strand transfer, which was shown to be either intra[126, 1331 or intermolecular [134, 1351, and the second DNA-strand transfer, which is essentially intramolecular [ 126, 1341, it seems that the two RNA molecules are necessary for recombination events and virus viability. Thus, even though the Psi/DLS structure of SNV might be dispensable for replication and recombination events [ 136, 1371, recent results showed that the structured Psi/DLS domain of MLV [39] is a preferred region for recombinational patch repair [ 1381 and that the HIV-l DIS stem-loop structure is required for efficient reverse transcription (Paillart J-C, Berthoux L, Ottmann M, Darlix J-L, Marquet R, Ehresmann B, Ehresmann C, in press). Taken together, these results support the idea that: i) the retroviral genome is encapsidated as a dimer; and ii) the highly ordered structure promoted by the DIWDLS region is required for retroviral replication and recombination during reverse transcription.

Acknowledgments We thank JL Darlix for providing electron micrographs of HIV- 1 RNA dimers. This work was supported by grants from the French Agency against AIDS (ANRS). JCP and ES are fellows from the ANRS.

References 1 Coffin JM (1990) Retroviridae and their replication. In: virology. (Fields BN, Knipe DM, eds) Raven Press, Ltd, New York. 1437-1.500 2 Canaani E, Helm KVD, Duesberg P (1973) Evidence for 3040s RNA as precursor of the 60-70 S RNA of Rous sarcoma virus. Proc Nutl Acad Sri USA 72,401-405 3 Beemon K, Duesberg P, Vogt P (1974) Evidence for crossing-over between avian tumor viruses based on analysis of viral RNAs. Proc Narl Acad Sci USA 7 1,425U258 4 Marquet R, Isel C, Ehresmann C, Ehresmann B (1995) tRNAs as primer of reverse transcriptases. Biochimie 77, 113-124 5 Baltimore D (1970) Viral RNA-dependent DNA polymerase. Nature 226,209-2 11 6 Temin HM, Mizutani S (1970) RNA-dependent DNA polymerase in virions of RSV. Nature 226, 211-213 7 Varmus HE (1982) Form and function of retroviral proviruses. Science 216.812-820 8 Bender W, Davidson N (1976) Mapping of poly (A) sequences in the electron microscope reveals unusual structure of type C oncomavirus RNA molecules. Cell 7, 595-607 9 Kung HJ, Hu S, Bender W, Bailey JM, Davidson N, Nicolson MO, McAllister RM (1976) RD-114, baboon, and wolly monkey viral RNAs compared in size and structure. Cell 7, 609-620 10 Murti KG, Bondurant M, Tereba A (1981) Secondary structural features in the 70 S RNAs of Moloney murine leukemia and Rous sarcoma viruses as observed by electron microscopy. J Viral 37,4111119 11 Cullen BR (199 1) Human immunodeficiency virus as a prototypic complex retrovirus. J Viral 65, 1053-1056 12 Mange1 WF, Delius H, Duesberg PH (1974) Structure and molecular weight of the 60-70s RNA and the 30-40s RNA of Rous sarcoma virus. Proc Nat1 Acad Sci USA 7 1,454 l-$545 13 Kung HJ, Bailey JM, Davidson N, Nicholson MO, McAllister RM (1975) Structure subunit composition, and molecular weight of RD-114 RNA. J Viral 16, 397-411

14 Maisel J, Bender W, Hu S, Duesberg PH. Davidson N (1978) Structure of 50 to 70s RNA from Moloney sarcoma viruses. J Viral 25, 384-394 15 Bender W, Chien YH, Chattopadkyay S, Vogt PK, Gardner MB, Davidson N (1978) High-molecular weight RNAs of AKR, NZB and wild mouse viruses and avian reticuloendotheliosis virus all have similar dimer structures. J Viral 25, 888-896 16 Prats AC, Roy C, Wang P, Erard M, Housset V, Gabus C, Paoletti C, Darlix JL (1990) C&elements and rrans-acting factors involved in dimer formation of murine leukemia virus RNA. J Viral 64, 774-783 17 Stoltzfus CM, Snyder PN (1975) Structure of B77 sarcoma virus RNA: stabilization of RNA after packaging. J Viral 16, 1161-l 170 18 Darlix J-L, Gabus C, Nugeyre MT, Clavel F, Barr&Sinoussi F (1990) C&elements and rruns-acting factors involved in the RNA dimerization of HIV-l. J Mol Biol216, 689-699 19 Aldovini A, Young RA (1990) Mutations of RNA and protein sequences involved in human immunodeficiency virus type 1 packaging result in production of noninfectious virus. J Viral 64, 1920-1926 20 Clavel F, Orenstein JM (1990) A mutant of HIV-l with reduced RNA packaging and abnormal particle morphology. J Viral 64, 5230-5234 21 Lever A, Gottlinger H, Haseltine W, Sodroski J (1989) Identification of a sequence required for efficient packaging of HIV-l RNA into virions. J Viral 63.40854087 22 Marquet R, Baudin F, Gabus C, Darlix JL, Mougel M, Ehresmann C, Ehresmann B (1991) Dimerization of human immunodeficiency virus (type 1) RNA: stimulation by cations and possible mechanism. Nucleic Acids Res 19, 2349-2357 23 Marquet R, Paillart J-C, Skripkin E, Ehresmann C, Ehresmann B (1994) Dimerization of human immunodeficiency virus type 1 RNA involves sequences located upstream of the splice donor site. Nucleic Acids Res 22, 145-151 24 Skripkin E, Paillart JC, Marquet R, Ehresmann B, Ehresmann C (1994) Identification of the primary site of human immunodeficiency virus type 1 RNA dimerization in vitro. Proc Nat1 Acad Sci USA 91,49454949 25 Paillart JC, Marquet R, Skripkin E, Ehresmann B, Ehresmann C (1994) Mutational analysis of the bipartite dimer linkage structure of HIV-l genomic RNA. J Viol Chem 269,2748&27493 26 Harrison GP, Lever AML (1992) The human immunodeficiency virus type-l packaging signal and major splice donor region have a conserved stable secondary structure. J Viol 66, 4144-4153 27 Laughrea M, JettC L (1994) A 19-nucleotide sequence upstream of the 5’ major splice donor site is part of the dimerization domain of human immunodeficiency virus 1 genomic RNA. Biochemistry 33, 1346413474 28 Muriaux D, Girard PM, Bonnet-Mathonikre B, Paoletti J (1995) Dimerization of HIV-llai RNA at low ionic strength. J Biol Chem 270, 8209-8216 29 Berkhout B, Oude Essink BB, Schoneveld I (1993) In vitro dimerization of HIV-2 leader RNA in the absence of PuGGAPuA motifs. FASEE J 7, 181-187 30 Schwatzberg P, Colicelli J, Goff SP (1983) Deletion mutants of Moloney murine leukemia virus which lack glycosylated gag protein are replication competent. J Viral 46, 538 31 Mann R, Mulligan RC, Baltimore D (1983) Construction of a retrovirus packaging mutant and its use to produce helper-free defective retrovirus. Cell 33, 153-159 32 Mann R, Baltimore D (1985) Varying the position of a retrovirus packaging sequence results in the encapsidation of both unspliced and spliced RNAs. J Viral 54, 401-407 33 Adam MA, Miller AD (1988) Identification of a signal in a murine retrovirus that is sufficient for packaging of nonretroviral RNA into virions. J viral 62, 3802-3806 34 Bender MA, Palmer TD, Gelinas RE, Miller AD (1987) Evidence that the oackaeine signal of Molonev murine leukemia virus extends into the &g reiioi. J-Virol61, 163911646 35 Prats AC, Sarih L, Gabus C, Litvak S, Keith G, Darlix JL (1988) Small finger protein of avian and murine retroviruses has nucleic acid annealing activity and positions the replication primer tRNA onto genomic RNA. EMBO J 7, 1777-1783

651 36 de Rocquigny H, Ficheux D, Gabus C, Allain B, Foumie-Zaluski MC, Darlix JL,Rbques BP (1993) Two short basic sequences surrounding the zinc finger of nucleocapsid protein NCplO of Molonev murine leukemia vi&s are critical for RNA annealing activity. Nuclk Acids Res 2 1, 823-829 37 Darlix JL, Laoadat-Tanolski M. de Rocauianv H, Roaues BP (1995) First glimpses at structure-function rela$ot%ips of the nucleocapsid protein of retroviruses. J Mel Biol254, 523-537 38 Roy C, Tounekti N, Mougel M, Darlix JL, Paoletti C, Ehresmann C, Ehresmann B, Paoletti J (1990) An analytical study of the dimerization of in vitro generated RNA of Molonev murine leukemia virus MoMuLV. Nucleic Acids Res 18,7287-7292 39 Tounekti N, Mougel M, Roy C, Marquet R, Darlix JL, Paoletti J, Ehresmann B, Ehresmann C (1992) Effect of dimerization on the conformation of the encapsidation Psi domain of Moloney murine leukemia virus RNA. .I Mol Biol223, 205-220 40 Girard P-M, Bonnet-Mathonibre B, Muriaux D, Paoletti J (1995) A short autocomplementary sequence in the 5’ leader region id responsible for dimerization of MoMuLV genomic RNA. Biochemisfry 34, 9785-9794 41 Van Beveren C, Coffin J, Hughes S (1984) Appendix C. In: RNA Tumor Viruses (Weiss R, ef al, eds) Cold Spring _ - Harbor Laboratory Press, NY, 928-939

42 Torrent C, Gabus C, Darlix JL (1994) A small and efficient dimerization/uackaninp. signal of rat VL30 RNA and its use in murine leukemia virus-VL3&d&ived vectors for gene transfer. J Viral 68, 661-667 43 Torrent C, Bordet T, Darlix J-L (1994) Analytical study of rat retrotransposon VL30 RNA dimerization in vitro and packaging in murine leukemia virus. / Moi Rio1 240,434-444 44 Torrent C, Wang P, Darlix JL (1992) A murine leukemia virus derived retroviral vector with a rat VL30 packaging psi sequence. Bone Marrow Transplant 9, 143-147 45 Feng Y-X, Fu W, Winter AJ, Levin JG, Rein A (1995) Multiple regions of harvey sarcoma virus RNA can dimerize in vitro. J Viral 69.2486-2490 46 Schwartz DE, Tizard R, Gilbert W (1983) Nucleotide sequence of Rous sarcoma virus. Cell 32, 853-869 47 Darlix J-L (1986) Control of Rous sarcoma virus RNA translation and uackaging by the 5’ and 3’ untranslated sequences. J Mol Bioll89,421-434 48 Bieth E, Gabus C, Darlix J-L (1990) A studv of the dimer formation of Rous sarcoma virus RNA and‘of its’effect on viral protein synthesis in vitro. Nucleic Acids Res 18, 119-127 49 Katz RA, Terry RW, Skalka AM (1986) A conserved cis-acting sequence in the 5’ leader of avian sarcoma virus RNA is required for packaging. J Viral 59, 163-167 50 Pugatsch T, Stacey DW (1983) Identification of a sequence likely to be required for avian retroviral packaging. Virolonv 128. 505-511 5 1 Lear AL, Haddrick M, Heaphy S (r995) A study of the dimerization of Rous sarcoma virus RNA in vitro and in viva. Virology 211, 47-57 52 Watanabe S, Temin HM (1982) Encapsidation sequences for spleen necrosis virus, an avian retrovirus, are between the 5’ long terminal repeat and the start of the Rag gene. Proc Natl Acad Sci USA 79, 5986-5990 53 Embretson JE, Temin HM (1987) Lack of competition results in efticient packaging of heterologous murine retroviral RNAs and reticuloendotheliosis virus encapsidation-minus RNAs by the reticuloendotheliosis virus helper cell line. J Viral 61, 2675-2683 54 Darlix J-L, Gabus C, Allain B (1992) Analytical study of avian reticuloendotheliosis virus dimeric RNA generated in vivo and in vitro. J Viral 66,7245-7252 55 Yang S, Temin HM (1994) A double hairpin structure is necessary for the efficient encapsidation of spleen necrosis virus retroviral RNA. EMBO J 13,713-726 56 Katoh I, Kyushiki H, Sakamoto Y, Ikawa Y, Yoshinaka Y (1991) Bovine leukemia virus matrix-associated protein MA (~15): further processing and formation of a specific complex with the dimer of the 5’-terminal genomic RNA fragment. J Viral 65,6845-6855 57 Katoh I, Yasunaga T, Yoshinaka Y (1993) Bovine leukemia virus RNA sequences involved in dimerization and soecific gag urotein binding: close relation to the packaging sites of avian, murine, and human retroviruses. J Viral 67, 1830-1839 1

I.

58 Kurg A, Sommer G, Metspalu A (1995) An RNA stem-loop structure involved in the packaging of bovine leukemia virus genomic RNA in viva. Virology 211, 434-442 59 Mansky LM, Krueger AE, Temin HM (1995) The bovine leukemia virus encapsidation signal is discontinuous and extends into the 5’ end of the gag gene. J Viral 69,3282-3289 60 Baudin F, Marquet R, Isel C, Darlix J-L, Ehresmann B, Ehresmann C (1993) Functional sites in the 5’ region of human immunodeficiency virus type-l RNA form defined structural domains. J Mol Rio1 229, 382-397 61 Awang G, Sen D (1993) Mode of dimerization of HIV-l genomic RNA. Biochemistry 32, 11453-l 1457 62 Sundquist WI, Heaphy S (1993) Evidence for interstrand quadruplex formation in the dimerization of human immunodeficiency virus-l genomic RNA. Proc Nat1 Acad Sci USA 90,3393-3397 63 Weiss S, Hlusl G, Famulok M, Konig B (1993) The multimerization’ state of retroviral RNA is modulated by ammonium ions and affects HIV-l full-length cDNA synthesis in vitro. Nucleic Acids Res 21,48794885 64 Marquet R, Paillart J-C, Skripkin E, Ehresmann C, Ehresmann B (1996) Localization of the dimerization initiation site of HIV-l genomic RNA and mechanism of dimerization. In: Biological Structure and Dynamics. (Sarma RH, Sarma MH, eds) Adenine Press, Albany, NY, in press 65 Fu W, Gorelick RJ, Rein A (1994) Characterization of human immunodeficiency virus type 1 dimeric RNA from wild-type and proteasedefective virions. J Viral 68, 5013-5018 66 Laughrea M, Jette L (1996) Kissing-loop model of HIV- 1 genome dimerization: HIV-l RNA can assume alternative dimeric forms, and all sequences upstream or downstream of hairpin 248-271 are dispensable for dimer formation. Biochemistry 35, 1589-1598 67 Paillart J-C, Skripkin E, Ehresmann B, Ehresmann C, Marquet R (1996) A loop-loop ‘kissing’ complex is the essential uart of the dimer linkage of genomic HIV-l RNA. Proc Nat1 Acad Sci .!JSA, in press 68 Simons RW (1993) The control of prokaryotic and eukaryotic gene expression by naturally occurring antisens RNA. In: Antisense Research and Applications (Crooke ST, Lebleu B, eds) CRC Press Inc. Boca Raton, 97-124 69 Wagner EGH, Simons RW (1994) Antisens RNA control in bacteria, phage and plasmids. Annu Rev Microhiol48, 713-742 70 Muriaux D; Foss6 P, Paoletti J (1996) A kissing complex together with a stable dimer is involved in the HIV-lt_,, RNA dimerization process in vitro. Biochemistrv 35. 5075-5082 71 Berkhout B, Schoneveld I(l993) Secondary structure of HIV-2 leader RNA comprising the tRNA-primer bindine site. Nucleic Acids Res 21. 1171-1178 72 Berkhout B (1996) Structure and function of the human immunodeficiency virus leader RNA. Prog Nucleic Acids Res Mol Biol, in press 73 Konings DAM, Nash MA, Maize1 JV, Arlinghaus RB (1992) Novel GACG-hairnin nair motif in the 5’ untranslated region of tvue C retrovirus related tomurine leukemia virus. J Viral 66y632-646’ 74 Mougel M, Tounekti N, Darlix JL, Paoletti J, Ehresmann B, Ehresmann C (1993) Conformation analysis of the 5’ leader and the gag initiation site of MoMuLV RNA and allosteric transitions induced bv dimerization. Nucleic Acids Res 21,4677-4684 75 Alford RL, Honda S, Lawrence CB, Belmont JW (1991) RNA secondary structure analysis of the packaging signal for Moloney murine leukemia virus. Virology 183, 61 l-619 76 Hackett PB, Dalton MW, Johnson DP, Petersen RB (1992) Phylogenetic and physiscal analysis of the 5’ leader RNA sequences of avian retrovirus. Nucleic Acids Res 19, 6929-6934 77 Harrison GP, Hunter E, Lever AML (1995) Secondary structure model of Mason-pfiser monkey virus 5’leader sequence: identification of a structural motif common to a variety of retroviruses. J Viral 69, 2175-2186 78 Myers G, Korber B, Hahn BH, Jeang K-T, Mellors JW, McCutchan FE, Henderson LE, Pavlakis GN (eds) (1995) A compilation and analysis of nucleic acid and amino acid seauences. In: Human Retroviruses and AIDS. Los Alamos National Laboiatory, Los Alamos, New Mexico, USA 79 Henderson LE, Copeland TD, Sowder RC, Smythers GW, Oroszlan S (1981) Primary structure of the low-molecular-weight nucleic acid

652 binding proteins of murine leukemia viruses. / Biol Chem 256, 84008406 80 Covey SN (1986) Amino acid sequence homology in gag region of reverse transcribing elements and the coat protein gene of cauliflower mosaic virus. Nucleic Acids Res 14, 623433 8 1 MCly Y, Comille F, Four&-Zaluski M-C, Darlix JL, Roques BP, GCrard D (1991) Investigation of zinc-binding affinities of Moloney murine leukemia virus nucleocapsid protein and its related zinc finger and modified peptides. Biopolymers 3 1, 899-906 82 Bess JW, Powell PJ, Issaq HJ, Schumack LJ, Grimes MK, Henderson LE, Arthur LO (1992)Tightly bound zinc in human immunodeticiency virus type 1,human T-cell leukemia virus type 1, and other retroviruses. J Viral 66, 84&847 83 Darlix JL, Spahr P-F (1982) Binding sites of viral protein P19 onto Rous sarcoma virus RNA and possible controls of viral functions. J MO! Biol 160, 147-161 84 MBric C, Darlix JL, Spahr PF (1984) It is Rous sarcoma virus P12 and not P19 that binds tightly to Rous sarcoma virus RNA. JMol Biol 173, 531-538 85 Weiss S, Konig B, Morikawa Y, Jones I (1992) Recombinant HIV-1 nucleocapsid protein ~15 produced as a fusion protein with glutathione S-transferase in Escherichia coli mediates dimerization and enhances reverse transcription of retroviral RNA. Gene 12 1, 203-2 12 86 Sakaguchi K, Zambrano N, Baldwin ET, Shapiro BA, Erickson JW, Omichinski JG, Clore GM, Gronenborn AM, Appella E (1993) Identification of a binding site for the human immunodeficiency virus type- 1 nucleocapsid protein. Proc Nut1 Acad Sci USA 90, 5219-5223 87 de Rocquigny H, Gabus C, Vincent A, FoumiC-Zaluski MC, Roques BP, Darlix JL (1992) Viral RNA annealing activities of human immunodeficiency virus type- 1 nucleocapsid protein require only peptide domains outside the zinc fingers. Proc N&l Acad Sci USA 89,6472X476 88 Prats AC, Housset V, de Billy G, Comille F, Prats H, Roques B, Darlix JL (199 1) Viral RNA annealing activities of the nucleocapsid protein of Maloney murine leukemia virus are zinc independent. Nucleic Acids Res 19, 3533-3541 89 Housset V, de Rocquigny H, Roques BP, Darlix JL (1993) Basic amino acids flanking the zinc finger of Moloney murine leukemia virus nucleocapsid protein NCplO are critical for virus infectivity. J Viral 67, 2537-2545 90 Tummino PJ, Scholten JD, Harvey P, Holler TP, Maloney L, Gogliotti R, Domagala J, Hupe D (1996)The in vitro ejection of zinc from human immunodeficiency virus (HIV) type 1 nucleocapsid protein by disulfide benzamides with cellular anti-HIV activity. Proc Nat1 Acad Sci USA 93,969-973 91 Luban J, Goff SP (1991) Binding of the human immunodeficiency virus type 1 (HIV- 1) RNA to recombinant HIV- 1 gag polyprotein. J Viral 65, 3203-3212 92 Berkowitz RD, Luban J, Goff SP (1993) Specific binding of human immunodeficiency virus type 1 gag polyprotein and nucleocapsid protein to viral RNAs detected by RNA mobility shift assays. J Viral 67, 7190-7200 93 Berkowitz RD. Goff SP (1994) Analysis of binding elements in the human immunodeficiency virus type 1 genomic RNA and nucleocapsid protein. Virology 202, 233-246 94 Clever J, Sassetti C, Parslow TG (1995)RNA secondary structure and binding sites for gag gene products in the 5’ packaging signal of human immunodeficiency virus type 1. J Viral 69,2101-2109 95 Geigenmiiller U, Linial ML (1996) Specific binding of human immunodeficiency virus type 1 (HIV- 1) gag-derived proteins to a 5’ HIV-I genomic RNA sequence. J Viral 70.667-671 96 Surovoy A?+Waidelich D, Jung G (1992) Nucleocapsid protein of HIV- I and its Zn complex formation analysis with electrospray mass spectrometry. FEBS Lett 3 11, 259-262 97 Hayashi T, Ueno Y, Okamoto T (1993) Elucidation of a conserved RNA stem-loop structure in the packaging signal of human immunodeficiency virus type-l. FEBS Lett 327, 213-218 98 Fu W, Rein A (1993) Maturation of dimeric viral RNA of Moloney murine leukemia virus. J Viral 67, 5443-5449 99 Fan H, Baltimore D (1973)RNA metabolism of murine leukemia virus: detection of virus-specific RNA sequences in infected and uninfected

100

101

102 103 104

105

106

107

108

109

110

I1 1

112

113

114

115

116 117

118

119

120

121

cells and identification of virus-specific messenger RNA. J Mol Biol 80,93-l 17 Levin JG, Grimley PM, Ramseur JM, Berezesky IK (1974) Deficiency of 60 to 70s RNA in murine leukemia virus particles assembled in cells treated with actinomycin D. J Viral 14, 152-161 Messer LI, Levin JG, Chattopadhyay SK (1981) Metabolism of viral RNA in murine leukemia virus-infected cells: evidence for differential stability of viral message and virion precursor RNA. J Viral 40, 683690 Gold L (1988) Post-transcriptional regulatory mechanisms in Escherichia coli. Annu Rev Biochem 57, 199-233 Kozak M (1986) Influences of mRNA secondary structure on initiation by eukaryotic ribosomes. Proc Nat1 Acad Sci (ISA 83,28X&2854 Kozak M (1989) Circumstances and mechanisms of inhibition of translation by secondary structure in eukaryotic mRNAs. Mol Cell Biol9, 5134-5142 de Smit M, van Duin J (1990) Control of prokaryotic translational initiation by mRNA secondary structure. Progr Nucleic Acid Res Mol Biol38, l-35 Miele G, Mouland A, Harrison GP, Cohen E, Lever AML (1996) The human immunodeficiency virus type 1 5’ packaging signal structure affects translation but does not function as an internal ribosome entry site structure. J Viral 70, 944-95 1 DonzC 0, Spahr PF (1992) Role of the open reading frames of Rous sarcoma virus leader RNA in translation and genome uackaeinz. . -EMBO J 11, 3147-3757 DonzC 0, Damav P. Suahr PF (1995) The first and third uORFs in RSV leader RNA are efficiently translated: implications for translational regulation and viral RNA packaging. Nucleic Acids Res 23, 861-868 Linial ML, Miller AD (1990) Retroviral RNA packaging: sequence requirements and implications. Curr Top Micro&o1 Immunol 157, 125152 Linial M, Medeiros E, Hayward WS (1978) An avian oncovirus mutant (SE 21Qlb) deficient in genomic RNA: biological and biochemical characterization. Cell 15, 1371-1381 Clavel F, Hoggan MD, Willey RL, Strebel K, Martin MA, Repaske R (1989) Genetic recombination of human immunodeficiency virus. J Viral 63, 1455-1459 Hayashi T, Shioda T, Iwakura Y, Shibuta H (1992) RNA packaging signal of human immunodeficiency virus type-l. Virology 188, 590599 Kim HJ, Lee K, O’Rear JJ (1994) A short sequence upstream of the 5’ major splice site is important for encapsidation of HIV-I genomic RNA. Virology 198, 336-340 Garzino-Demo A, Gallo RC, Arya SK (1995) Human immunodeficiency virus type 2 (HIV-2): packaging signal and associated negative regulatory element. Hum Gene Ther 6, 177-184 Rizvi TA, Panganiban AT (1993) Simian immunodeficiency virus RNA is efficiently encapsidated by HIV- 1 particles. J Viral 67,268 l2688 Zhang Y, Barklis E (1995) Nucleocapsid protein effects on the specificity of retrovirus RNA encapsidation. J Viral 69, 57165722 Berkowitz RD, Ohagen A, Hiiglund S, Goff SP (1995) Retroviral nucleocapsid domains mediate the specific recognition of genomic viral RNAs by chimeric gag polyproteins during RNA packaging in viva. J Viral 69, 6445-6456 McBride MS, Panganiban AT (1996) The human immunodeficiency virus type 1 encapsidation site is a multipartite RNA element composed of functional hairpin structures. J Viral 70, 2963-2973 Cheung KS, Smith RE, Stone MP, Joklik WK (1972) Comparison of immature (rapide harvest) and mature Rous sarcoma virus particles. Virology 50, 851-864 Gorelick RJ, Nigida Jr SM, Arthur LO, Henderson LE. Rein A (1990) Roles of nucleocapsid cysteine arrays in retroviral assembly and replication: possible mechanisms in RNA encapsidation. In: Advances in Molecular Biology and Targeted Treatment of AIDS (Kumar A, ed) Plenum Press, New York, 257-272 Dupraz P, Oertle S, MCric C, Damay P, Spahr PF (1990) Point mutations in the proximal cys-his box of Rous sarcoma virus nucleocapsid protein. J Viral 64, 49784987

653 122 MCric C, Goff S (1989) Characterization of Moloney murine leukemia virus mutants with single amino-acid substitution in the Cys-His box of the nucleocapsid protein. / Vim1 63, 1558-1568 123 Temin HM (199 1) Sex and recombination in retroviruses. Trends Gener I, 71-74 124 Temin HM (1993) Retrovirus variation and reverse transcription: abnormal strand transfers result in retrovirus genetic variation. Proc Natl Acad Sci USA 90,6900-6903 125 Ricchetti M, But H (1990) Reverse transcriptase and genomic variability: the accuracy of DNA replication is enzyme-specific and sequence-dependent. EMSO J 9, 1583-1593 126 Hu WS, Temin HM (1990) Retroviral recombination and reverse transcription. Science 250, 1227-1233 127 Stuhlmann H, Berg P (1992) Homologous recombination of copackaged retroviruse RNAs during reverse transcription. J Virol66, 2378-2388 128 Hu W-S, Pathak VK, Temin HM (1993) Role of reverse transcriptase in retroviral recombination. In: Reverse Transcriprase. Cold Spring Harbor Laboratory Press, New York, 25 l-274 129 Coffin JM (1979) Structure, replication and recombination of retrovirus genomes: some unifying hypothesis. J Gen viral 42, l-26 130 Hunter E (1978) The mechanisms for genetic recombination in the avian retroviruses. Curr Top Microhiol Immunol79, 295-309

13 1 Junghans RP, Boone LR, Skalka AM (1982) Retroviral DNA H structures: displacement-assimilation model of recombination. Cell 30, 5362 132 Skdlka AM, Boone L, Junghans R, Luk D (1983) Genetic recombination in avian retroviruses. J Cell Biol294, 75-86 133 Jones JS, Allan RW, Temin HM (1994) One retroviral RNA is sufficient for synthesis of viral DNA. J Viral 68, 207-2 16 134 Panganiban A, Fiore D (1988) Ordered interstrand and intrastrand DNA transfer during reverse transcription. Science 241, 1064-1069 135 Luo G, Taylor J (1990) Template switching by reverse transcriptase during DNA synthesis. J. Viral. 64, 43214328 136 Jones JS, Allan RW, Temin HM (1993) Alteration of localisation of dimer linkage structure sequence in retroviral RNA: little effect on replication or homologous recombination. J Viral 67, 3551-3558 137 Zhang J, Temin HM (1996) The recombination rate is not increased when retroviral RNA is missing an encapsidation sequence. J Viral 70, 2019-2021 138 Mikkelsen JG, Lund AH, Kristensen KD, Duch M, Sorensen MS, Jorgensen P, Pedersen FS (1996) A preferred region for recombinational patch repair in the 5’ untranslated region of primer binding site-impaired murine leukemia virus vectors. J Viral 70, 1439-1447