Tài liệu Báo cáo khoa học: Template requirements and binding of hepatitis C virus NS5B polymerase during in vitro RNA synthesis from the 3¢-end of virus minus-strand RNA docx

[20] studied RNA synthesis by recombinant HCV NS5B using deletion mutants of the 3¢ terminus of the minus-strand RNA but deletions were made on the basis of the structure of the 5¢UTR of

NS5B polymerase during in vitro RNA synthesis from the 3¢-end of virus minus-strand RNA

The´re`se Astier-Gin, Pantxika Bellecave, Simon Litvak and Michel Ventura

UMR-5097 CNRS, Universite´ Victor Segalen Bordeaux 2, Bordeaux, France

Hepatitis C virus (HCV) is the major causative agent

of non-A, non-B hepatitis [1] This virus has a

posit-ive-stranded RNA genome and belongs to the

Flavivir-idae family The RNA contains a large open reading

frame that encodes a polyprotein which is cleaved into

10 viral proteins: C, E1, E2, p7, NS2, NS3, NS4A,

NS4B, NS5A and NS5B [2] Recently, a frame shift

product of HCV core encoding sequence, the F

pro-tein, was described [3,4] This protein has no known

functions The large open reading frame is flanked by

two untranslated regions (UTR) The 341-nucleotide

(nt) 5¢UTR in association with the first nucleotides of

the core protein contains an internal ribosome entry

site (IRES) that directs cap-independent translation of

the viral RNA [5,6] The 3¢UTR is composed of a

short variable region, a polypyrimidine tract (poly U-UC) of variable length and a highly conserved 98-nucleotide segment (3¢X) The two latter domains are essential for viral infectivity in vivo [7] and RNA replication of HCV in the HCV replicon system [8,9] HCV RNA replication occurs in two steps In the first step the viral replicase synthesizes a minus-strand RNA that serves as a template for the synthesis of new plus-strand RNA molecules Initiation of RNA synthesis at the 3¢-end of the plus- and minus-strand RNA most probably involves interactions between the protein components of the replication complex,

in particular with the viral polymerase (NS5B), and structures and⁄ or sequences of the viral RNA templates The secondary structure of the 3¢-end of the

Keywords

HCV; minus strand RNA; RdRp

Correspondence:

The´re`se Astier-Gin, CNRS UMR5097,

Universite´ Victor Se´galen Bordeaux 2, 146,

rue Le´o Saignat, 33076 Bordeaux cedex,

France

Fax: +33 5 57571766

Tel: +33 5 57571742

E-mail: Therese.astier@reger.u-bordeaux2.fr

(Received 23 March 2005, revised 24 May

2005, accepted 3 June 2005)

doi:10.1111/j.1742-4658.2005.04804.x

In our attempt to obtain further information on the replication mechanism

of the hepatitis C virus (HCV), we have studied the role of sequences at the 3¢-end of HCV minus-strand RNA in the initiation of synthesis of the viral genome by viral RNA-dependent RNA polymerase (RdRp) In this report, we investigated the template and binding properties of mutated and deleted RNA fragments of the 3¢-end of the minus-strand HCV RNA in the presence of viral polymerase These mutants were designed following the newly established secondary structure of this viral RNA fragment We showed that deletion of the 3¢-SL-A1 stem loop significantly reduced the level of RNA synthesis whereas modifications performed in the SL-B1 stem loop increased RNA synthesis Study of the region encompassing the 341 nucleotides of the 3¢-end of the minus-strand RNA shows that these two hairpins play a very limited role in binding to the viral polymerase On the contrary, deletions of sequences in the 5¢-end of this fragment greatly impaired both RNA synthesis and RNA binding Our results strongly sug-gest that several domains of the 341 nucleotide region of the minus-strand 3¢-end interact with HCV RdRp during in vitro RNA synthesis, in parti-cular the region located between nucleotides 219 and 239

Abbreviations

HCV, hepatitis C virus; IRES, internal ribosome entry site; nt, nucleotide; RdRp, RNA-dependent RNA polymerase; TCA, trichloroacetic acid; UTR, untranslated region.

plus-strand RNA has been determined [10,11] and the

involvement of the three stem loops of the 3¢X has

been extensively studied both in vitro, in the replicon

system and in vivo [7–9,12] The secondary structure of

the 3¢-end of the minus-strand RNA has been

estab-lished more recently [13,14] It was shown that the

341 nt from the 3¢-end of the minus-strand RNA,

com-plementary to the HCV 5¢UTR, folds into six stem

loops With the exception of the short SL-A1 stem

loop, the one closest to the 3¢-end, these structures

dif-fered from those of the plus-strand RNA Thus the

minus-strand 3¢-terminal domain is not the mirror

image of its antisense sequence corresponding to

5¢UTR Its role in the initiation of the plus-strand

RNA synthesis cannot be directly assessed in the HCV

replicon system because translation and replication are

linked in this system However, it has been shown that

the first 125 nt of the 5¢UTR are essential for RNA

replication of HCV replicons in HuH7 cells [15,16]

Most probably the 125 nt present at the 3¢-end of the

complementary minus-strand RNA is an important

element for de novo synthesis of plus-strand RNA

Data obtained from in vitro experiments using the

recombinant NS5B protein argue for this hypothesis

We have previously shown that the recombinant HCV

RNA polymerase efficiently replicates the 341 nt of the

3¢-end of the HCV minus-strand RNA in vitro and that

deletion of the 3¢ 45 nt greatly impaired RNA synthesis

[17] Oh et al [18] showed that RNA synthesis from the

HCV minus-strand RNA required a minimum of 239 nt

from the 3¢-end Furthermore, we have reported that an

oligonucleotide complementary to the SL-B1 domain

in the 3¢-end of the HCV minus-strand RNA inhibits

in vitro initiation of RNA synthesis by the viral

poly-merase [19] Kashiwagi et al [20] studied RNA synthesis

by recombinant HCV NS5B using deletion mutants of

the 3¢ terminus of the minus-strand RNA but deletions

were made on the basis of the structure of the 5¢UTR of

the plus-RNA, the structure of the 3¢-end of the

minus-strand RNA being at that time unavailable

In the present study we investigated the involvement

of sequences and⁄ or structures in RNA synthesis

direc-ted in vitro by HCV NS5B by using new mutants of

the 3¢-end of minus-strand RNA

Results

Effect of mutations in the 3¢- or the 5¢-end of the

(–)IRES HCV RNA template on RNA synthesis

The secondary structure of the 3¢-terminal nucleotide

region of the HCV minus-strand RNA is illustrated in

Fig 1 This fragment contains two domains: domains

I and II Structures of both domain I (A), as deter-mined by Schuster et al [13] and by Smith et al [14], and domain II, as determined, respectively, by Smith

et al [14] (B), Schuster et al [13] (C) or predicted by RNA Draw software (D) are represented Nucleotides are numbered increasingly from the 3¢-end of the RNA; the five first stem-loops (A) are named as repor-ted by Schuster et al [13] The 228 nt of domain I fold into five stable stem-loops (A) It has been found to display the same secondary structure in the three mod-els presented here: the fragment containing 365 nucleo-tides described in [14], the 416 nt fragment described

by [13] and the 341 nt fragment, used in this work On the contrary, the 5¢-end of the different RNA frag-ments (137 nt in Fig 1B, 188 nt in Fig 1C, and 113 nt

in Fig 1D) is less stably organized, thus giving differ-ent structures in the three cases

Effect of modifications at the 3¢-end: mutations or deletions in the SL-A1 and SL-B1 stem loop

We have shown previously that the 3¢-end of the HCV minus-strand RNA is replicated highly efficiently

in vitro by purified viral polymerase NS5B [17] This high level of RNA synthesis is associated with the presence of a cytidine residue at the 3¢-end Upstream sequences and⁄ or structures also seem to be involved,

as deletion of 45 nt at the 3¢-end greatly reduces RNA synthesis Moreover, as another indication of the importance of this region, we have recently shown that ODN7, an antisense oligonucleotide complementary to

a domain comprised between nt 85 and 103 of the 3¢-terminal minus-strand HCV RNA, was able to inhi-bit RNA initiation [19]

To identify more precisely sequences or structural elements important for RNA synthesis we constructed mutants in the SL-A1 and SL-B1 domains comprised

in a 341 nucleotide fragment of the 3¢-end minus-strand RNA called (–)IRES This fragment is effi-ciently replicated in vitro by the HCV NS5B [17] The mutants were designed in such a way that the deletions

or the base changes did not alter the structure of the other domains of the (–)IRES RNA as determined by predicted secondary structure with RNA Draw soft-ware The structure of the first 151 nt nucleotides of wild-type and mutated RNA is shown in Fig 2 The mutated RNAs were used as templates in the RdRp assay and the levels of RNA synthesis were compared with that of the wild-type (–)IRES Very different results were obtained when changing either stem-loop As shown in Table 1, the deletion of the SL-A1 stem loop in the (–)IRES DSL-A1 mutant reduced the RNA synthesis by 39% These results were

in accordance with those of Kashiwagi et al [20],

which showed that deletion of SL-A1 reduced the

RNA synthesis by 25%

We then performed different deletions or

site-direc-ted mutagenesis in the sequence of the hairpin SL-B1

(Fig 2) We carried out the following mutations of

SL-B1: (a) [(–)IRESD91-97)] that contains a deletion of

nt 91–97 corresponding to a bulge where the ODN7

antisense hybridized; (b) [(–)IRES Dhp2] that contains

a deletion of the 39 nt corresponding to the apical part

of SL-B1; (c) [(–)IRES hp2b] with a change of four nt that induces a dissociation of the stem at the base of SL-B1; (d) [(–)IRES LDH2] with a complete deletion

of SL-B1 In contrast to the results obtained when changes were introduced in SL-A1, none of the modifi-cations in SL-B1 reduced RNA synthesis (Table 1) In all cases RNA synthesis was increased

Altogether, these results indicated that while the presence of the SL-A1 domain is necessary for efficient RNA synthesis, the SL-B1 region does not contain

Fig 1 Secondary structure of the 3¢ ter-minal sequences of HCV minus-strand RNA Model structures of domain I (A) and domain

II (B, C and D) are shown (A) Domain I is composed by 228 nt located at the 3¢ end The first five stem loops are named as des-cribed in [13] (B) Secondary structure of

a fragment spanning from nt 229–365 in domain II as described in [14] (C) Secondary structure of domain II, spanning nt 229–416,

as described by in [13] (D) Secondary struc-ture of a fragment spanning from nt 229–341

in domain II (predicted by RNA DRAW soft-ware).

sequences or structural domains necessary for the

high level of RNA synthesis obtained when incubating

in vitroNS5B and the 3¢-end of the minus-strand HCV

RNA as template Moreover, modifications of the

lat-ter domain even resulted in enhanced RNA synthesis

(Table 1)

Effect of deletions at the 5¢-end

It has been shown that almost all the 5¢UTR region is

required for efficient RNA replication of HCV RNA

when using the replicon system [15] Thus, we

exam-ined the effect of 5¢-end deletions in the (–)IRES 341

template on the amount of RNA synthesized in vitro

by the HCV NS5B 1a The predicted secondary struc-ture of three deletion mutants is displayed in Fig 3

As shown in Table 1 deletion of 102 nt at the 5¢-end

of (–)IRES RNA leading to (–)IRES 239 reduced RNA synthesis by 51% These results are in agreement with those obtained by Oh et al [18] Further deletion

of the 5¢-end by 20 nt to give (–)IRES 219 showed a striking reduction of RNA synthesis to only 19% of that obtained with the (–)IRES wild-type RNA Struc-ture prediction by computer analysis showed that the four bases at the 3¢-end of the (–)IRES 239 and (–)IRES 219 were unannealed as in the wild-type

Fig 2 Secondary structure of RNA mutated

in the SL-A1 and SL-B1 domains Only the

secondary structure of the 151 nt from the

3¢-end of the minus-strand RNA is shown.

The computer predicted structure at 25
C

of the same domain is shown for each

mut-ant RNA The DG of the wild-type (–)IRES

RNA and those of the five mutant RNA are

indicated The arrows or the curly bracket

showed the location of deletions or

muta-tions, respectively.

(–)IRES RNA and that the secondary structures of the

stem loops SL-A1, SL-B1, SL-C1 and SL-D1 were

unmodified The 5¢-SL-E1 stem loop was unchanged in

(–)IRES 239 whereas in (–)IRES 219, the base of the

stem was replaced by a six nt bulge (Fig 3) We also

performed a deletion of 237 nt at the 5¢-end to give

the (–)IRES 104 This deletion preserved the structure

of the SL-A1 and of the SL-B1 hairpin and the 4 nt at

the 3¢-end remained free as in the wild-type (–)IRES

RNA synthesis obtained with the (–)IRES 104 RNA

was only 24% of that of the wild-type (–)IRES RNA,

a value similar to that of the (–)IRES 219 RNA The

fact that the (–)IRES 104 can be used as a template by

the recombinant HCV NS5B differed from the data

obtained by Oh et al [18] These authors showed that

the 122 nt of the 3¢-end of the HCV minus-strand

RNA was unable to sustain RNA synthesis Structure

prediction by computer analysis revealed that the

sec-ondary structure of the 3¢-end of the 122 nt RNA

frag-ment was unmodified compared to wild-type RNA

This discrepancy can be explained by differences

between the sensitivity of the assays used in the two

studies Finally, we performed an RdRp assay with an RNA fragment corresponding to 20 nt of the 3¢-end of the minus-strand HCV RNA that formed the four base single-strand region and the SL-A1 hairpin As shown

in Table 1, our recombinant NS5B was unable to use this RNA as template for RNA synthesis It could be argue that this data results from a misfolding of this short RNA due to a bimolecular association However, this last hypothesis is unlikely because the (–)IRES 20 migrates as one RNA species of the expected size in native gel (data not shown) Further studies are needed

to clarify this point

Taken together, these results indicate that regions located at the 5¢-end of the (–)IRES RNA are crucial to obtain a high level of RNA synthesis by NS5B in vitro,

in particular the 122 nt located at the 5¢-end Alternat-ively, the elimination of this less stably structured domain decreased RNA synthesis by increasing the relative amount of structured regions giving rise to templates poorly replicated by the NS5B

Analysis of RNAs synthesized using wild-type and mutant (–)IRES RNA as templates

To examine the products synthesized in the presence of the mutated (–)IRES, an RdRp assay was performed with the NS5B 1a (Fig 4) The same amount of each product (833 Bq) was loaded onto a 6% denaturing polyacrylamide gel All templates with mutations or deletions in the SL-A1 or the SL-B1 stem loop allowed synthesis of a major RNA product with the size of the input template (Fig 4A) No arrest bands were observed during the synthesis with the exception of RNA (–)IRES DSL-A1 that gave a small amount of a product about 39 nt shorter than the template (Fig 4A) The relative quantity of this short product was variable in different experiments In all cases, slower migrating bands were also observed The major one migrated to a position corresponding to an RNA two times larger than the template For the wild-type (–)IRES RNA we have previously shown that this product corresponds to two successive copies of the template [17] Products of higher molecular weight in very low amounts were also visible They may corres-pond to three (or more) successive copies of the tem-plate

When RNA synthesis was performed with (–)IRES RNA templates that have deletions in the 5¢-end, a major product migrating to the same position as the template and slower ones were observed (Fig 4A) In the case of the (–)IRES 104, the product which was two-fold the size of the template was almost as abun-dant as the product the same size as the template In

Table 1 RNA synthesis obtained with mutants of (–)IRES RNA

without or with heparin Determination of Kdfor various (–)IRES

RNAs An RdRp assay was performed with the purified NS5B1a

(150 n M ) and mutant RNAs as templates as described in

Experi-mental procedures The amount of RNA synthesized was

deter-mined after TCA precipitation and counting in a Wallac scintillation

counter The results were expressed as a percentage of the value

obtained with the wild-type (–)IRES in the absence or in the

pres-ence of heparin (The addition of heparin during RNA synthesis

reduced RNA synthesis by 72%) Data were corrected following

the number of A residues in each RNA template and were the

mean of at least three experiments For Kddetermination, a

rena-tured32P-labeled RNA (13 n M , 167 Bq) was incubated with NS5B

(50 n M , 100 n M , 200 n M , 500 n M and 1 l M ) for 20 min at 25
C.

The Kdwas estimated from the concentration of NS5B resulting in

50% shifting of 32P RNA Results correspond to mean values of

3–4 independent experiments for each RNA N.D., not determined.

RNA

RNA synthesis percentage (–)IRES

K d (nM) (–)-heparin (+)-heparin

addition to the product twice the template size, the

(–)IRES 239 RNA gave a product of about 375 nt

indicated by a star (Fig 4A) Again no prominent

arrest of RNA synthesis was observed when (–)IRES

RNAs harboring 5¢ deletions were used as templates

Data presented in Fig 4A were obtained by using

the NS5BD21 from HCV H77 genotype 1a To

exam-ine whether a NS5B of another strain of HCV would

give the same results, we performed an RdRp assay

with a recombinant NS5BD21 purified from the HCV

J4 of genotype 1b The products obtained with both

enzymes in the presence of wild-type (–)IRES (–)IRES

239 and (–)IRES 104 are shown in Fig 4B The

migra-tion patterns of the products synthesized by NS5B of

both HCVs were identical Altogether, these results

indicated that the mutations and deletions performed

in the RNA fragment corresponding to the 3¢-end of

the HCV minus-strand RNA did not significantly alter

the initiation site of RNA synthesis by recombinant

HCV NS5B from different viral strains

Analysis of RNA synthesized in one round

of synthesis

RNA products shown in Fig 4 were obtained under

conditions where HCV NS5B could reinitiate RNA

synthesis several times To analyze the RNA synthes-ized in one round of synthesis, we performed RdRp assays in the presence of heparin in order to prevent reinitiation In a first step we determined the heparin concentration needed to prevent reinitiation during a

2 h RdRp assay As illustrated in Fig 5A, heparin at

a concentration of 200 lgÆmL)1 (220 times heparin molar excess with respect to the enzyme) reduced RNA synthesis by 72% Incubation with a higher con-centration (400 lgÆmL)1) did not significantly modify the level of RNA synthesis, suggesting that the amount

of heparin was sufficient to sequester all enzyme mole-cules These data were confirmed by performing a kinetic experiment in the presence of heparin at

200 lgÆmL)1 As shown in Fig 5B, the amount of syn-thesized RNA greatly increased in the first 10 min of the reaction to reach a plateau after 30 min Analysis

of the RNA products on polyacrylamide gel showed that the size of the products did not increase, indica-ting that the elongation step was achieved (data not shown)

We then compared the total level of RNA synthes-ized in the presence or absence of heparin at

200 lgÆmL)1, using the various templates Results reported in Table 1 show that the level of RNA syn-thesis obtained with the mutated templates (compared Fig 3 Predicted structure of the (–)IRES RNA with deletions of the 5’-end The computer predicted structure at 25
C and the DG values are shown for three 5’-deleted mutants.

to wild-type RNA) was very similar under both

condi-tions The only exception was observed in the case of

the (–)IRES LDH2 template, where the entire SL-B1

stem loop has been deleted Results with this

construc-tion displayed a slightly but significantly reduced RNA

synthesis in the presence of heparin compared to that

observed in the absence of this compound

Our further step was to study RNA products formed

under single-round conditions, i.e in the presence of

heparin as the trapping molecule As shown in Fig 6A

when the wild-type (–)IRES was used as template for

HCV NS5B in the presence of heparin, a template

sized RNA was the major RNA product Interestingly,

the high molecular weight RNA corresponding to two

(or more) successive copies of the template

disap-peared Heterogeneous RNA of smaller sizes were also

observed but in very low amounts These data indicate:

(a) that the HCV NS5B is highly processive and

(b) that the high molecular weight product was the

result of a reinitiation process The same results were

obtained when (–)IRES RNA mutated or deleted in

the SL-A1 or in the SL-B1 stem loops were used as

templates in the presence of heparin (data not shown)

When one round of RdRp assay was performed with

the 5¢ deleted (–)IRES RNA 239 and 104 different

patterns were observed (Fig 6B) With the (–)IRES

239 the major product was always a template size RNA, while the high molecular weight RNA twofold the size of the template was absent In contrast, the

375 nt RNA product was synthesized (Fig 6B) The same type of experiment undertaken with the (–)IRES

104 showed that all the products two to four times the size of the template (and visible in Fig 4A) disap-peared almost completely, as only a faint high mole-cular weight band was visible (Fig 6B) In this latter case, in addition to the template size product, a short RNA product was present in relatively high amounts suggesting that under these conditions HCV NS5B often released from the 104 nt RNA template after ini-tiation of RNA synthesis

In addition to RNA synthesized from the 3¢-end of the HCV minus-strand RNA, we have previously des-cribed the products using the 3¢-end of the plus-strand HCV RNA as template [17] To assess whether the high molecular weight RNA produced from one of these plus-strand RNA fragments called (+) 3¢UTRNDX also disappeared, we performed experi-ments in the presence of heparin The (+) 3¢UTRNDX template corresponded to 150 nt 3¢ of the NS5B cod-ing sequence plus the 3¢UTR sequence deleted from

B A

Fig 4 RNA synthesized by NS5B in the presence of wild-type and mutated RNAs Wild-type and mutated (–)IRES RNAs were used in RdRp assays An aliquot of the reac-tion products was precipitated by 10% TCA and the radioactivity incorporated in newly synthesized RNA determined as described

in Experimental procedures section The remaining of the products was purified by phenol ⁄ chloroform extraction (1 : 1, v ⁄ v) and precipitated by one volume of isopropanol in the presence of 0.5 M ammonium acetate.

32 P-labeled reaction products (833 Bq each) were denatured and loaded onto a 6% denaturing polyacrylamide gel (A) Products synthesized by HCV NS5B genotype 1a (B) Products synthesized by HCV NS5B genotype 1a or NS5B 1b as indicated in the figure A labeled 0.16–1.77 kb RNA ladder (Gibco-BRL) and the labeled RNA templates were used as size markers.

the 3¢X 98 nt As shown in Fig 6A, in the absence of

heparin this RNA gave a major RNA product of the

template size (282 nt) and a slower migrating product

twice the size of the template This high molecular

mass RNA was not observed in the presence of

hep-arin (Fig 6A) indicating that the mechanism of

reiniti-ation operates in RNA synthesis using both the plus

and the minus 3¢-ends of HCV RNA as templates

Finally, we wanted to see if a recombinant NS5B

purified from a highly related virus, the GBV-B was

able to synthesize the same type of RNA as HCV

NS5B from the wild-type (–)IRES RNA Data

presen-ted in Fig 6A showed that as in the case of HCV

NS5B, the GBV-B NS5B synthesized a major product

of the template size but, even in the absence of hep-arin, no high molecular weight RNA were observed suggesting a different initiation mechanism of RNA synthesis for both viral polymerases

Binding of wild-type and mutated (–)IRES RNA

to HCV NS5B Data described above indicate that deletions of the stem loop SL-A1 at the 3¢-end or of 102, 122 or 237 nucleotides at the 5¢-end of the (–)IRES 341 nt RNA diminished in vitro RNA synthesis directed by the HCV NS5B To assess whether the low level of RNA synthesis was related to the binding of these templates

to the HCV RdRp, we performed gel shift assays Results of such an experiment are presented in Fig 7 They showed that the binding of both wild-type and

239 (–)IRES RNA was complete at 1 lm NS5B In a native polyacrylamide gel, the two RNA migrated as one species (Fig 7A); however, in some experiments

a slowly migrating band of RNA was observed (see below) In our experimental conditions the [32P]RNA⁄ NS5B complex remained at the top of the gel The Kdvalues for the wild-type and deleted RNAs were determined from curves obtained as in Fig 7B

As shown in Table 1, with the unique exception of the (–)IRES20 RNA that did not bind the NS5B, all four mutated RNAs bound the viral polymerase with the same affinity as the wild-type (–)IRES RNA

Competition experiments were then performed with all mutated RNAs and the wild-type (–)IRES RNA for binding to the enzyme NS5B (500 nm) and wild-type [32P](–)IRES RNA (13 nm) were incubated with increasing amounts of cold RNAs and analyzed by electrophoresis on nondenaturing polyacrylamide gel Free 32P-labeled (–)IRES migrated as two bands, a major one indicated by an arrow and a minor one that migrated more slowly (Fig 8) Repeated experiments indicated that the latter band was not visible in every electrophoresis analysis of a same 32P-labeled RNA preparation Small variations in denaturation–renatur-ation conditions or during electrophoresis could explain these differences In the presence of NS5B,

a clear shift of both RNA species was observed As shown in Table 2 and illustrated in Fig 8A, 44 nm of wild-type RNA released half of the labeled RNA in the complex with NS5B This dissociation is specific as

4 lm yeast tRNA was unable to release the (–)IRES RNA bound to NS5B (data not shown) The five RNAs with deletions or mutations in the two stem loops SL-A1 or SL-B1 located close to the 3¢-end dis-placed the 32P-labeled (–)IRES RNA at slightly higher concentrations ranging from 53 nm for (–)IRES Dhp2b

Fig 5 Effect of heparin on RNA synthesized by NS5B (A)

Wild-type (–)IRES RNA was preincubated for 30 min at 25
C in the

RdRp reaction mixture without ATP and UTP Various

concentra-tions of heparin were then added followed by ATP and 3H-UTP.

The reaction mixture was further incubated at 25
C for 2 h The

amount of radioactivity incorporated into the nucleic acids was

measured after TCA precipitation and plotted against heparin

con-centration (B) An RdRp assay using 32P-UTP as labeled nucleotide

was performed as above in the presence of heparin (200 lgÆmL)1)

with wild-type (–)IRES RNA as template Twenty microliters were

removed from the reaction mixture at different times The amount

of radioactivity incorporated into the nucleic acids was measured

after TCA precipitation and plotted against the incubation time in

minutes.

to 97 nm for (–)IRES hp2b On the contrary, higher

amounts of 5¢ deleted RNAs were needed to dissociate

the NS5B⁄ [32P](–)IRES RNA complex (Table 2 and

Fig 8B) As expected no competition was observed

between the wild-type (–)IRES and the (–)IRES 20

RNAs (data not shown) These results strongly suggest

that multiple domains of the 341 nt of the

minus-strand RNA 3¢-end are involved in the binding to

NS5B in particular the region located between nt 219

and 239 Consequently, one can hypothesize that the

5¢ deleted RNA did not bind NS5B in the same

man-ner as the wild-type RNA and could not efficiently

dis-placed this RNA in complex with NS5B

Discussion

The replication mode of Flaviviridae, which involves

synthesis of a minus-strand RNA serving as template

for synthesis of the plus RNA genomic strand, would

seem to indicate that the 3¢-end of HCV minus-strand RNA should play an important role in the initiation of synthesis of the viral genome

In this report we investigated the template and bind-ing properties of mutated and deleted RNA fragments

of the 3¢-end of the minus-strand HCV RNA in further detail This study should lead to interesting interpretations since it is facilitated by the recent deter-mination of the secondary structure of this region [13,14] We first analyzed the effect of mutations or deletions in the two stem loops SL-A1 and SL-B1 located near the 3¢-end The deletion of SL-A1 signifi-cantly reduced the level of RNA synthesis directed by NS5B but mutations or deletions performed on SL-B1 (including its complete deletion) do not have a deleteri-ous effect on the level of RNA synthesis in one or sev-eral rounds of synthesis Our results showed that, unlike for SL-A1, the deletion of the entire stem loop SL-B1 or part of this domain increased in vitro RNA

Fig 6 Effect of heparin on RNA synthe-sized by NS5B Wild-type and mutated (–)IRES RNAs were preincubated for 30 min

at 25
C in the RdRp reaction mixture with-out ATP and UTP Heparin (200 lgÆmL)1) was then added followed by ATP and [ 32 P]UTP The reaction mixture was further incubated at 25
C for 2 h [ 32 P]RNA prod-ucts were quantified after TCA precipitation

of an aliquot of the reaction mixture as des-cribed in Experimental procedures section.

32

P-labeled reaction products (833 Bq each) were denatured and loaded onto a 6% denaturing polyacrylamide gel (A) RNA products synthesized without or with hep-arin (200 lgÆmL)1) by HCV or GBV-B NS5B The templates used corresponded to 3’ domains of plus or minus-strand HCV RNA (B) RNA products synthesized by HCV NS5B

in the presence of heparin with wild-type or 5’ deleted (–)IRES RNA.

synthesis by NS5B It looks like, when increasing the

relative amount of unstructured domains in the

tem-plate, the RNA synthesis by NS5B is enhanced

How-ever, the level of RNA synthesis observed with the

RNA mutant (–)IRESD91–97 did not fit with this

hypothesis, suggesting that interactions of the bulge

formed by the deleted sequences with other regions of

the RNA or with NS5B could occur

Analysis of the products showed that the synthesized RNAs are homogenous in size suggesting that initi-ation occurred by a de novo mechanism at the 3¢-end

of all SL-A1 or SL-B1 mutated templates, as previ-ously shown for the wild-type (–)IRES RNA [17] The HCV NS5B polymerase appeared to be highly proces-sive in all cases as no major product shorter than the template could be observed even in the presence of heparin

We have also shown that the NS5B is unable to syn-thesize RNA from the (–)IRES20 RNA corresponding

to the four free nucleotides of the 3¢-end and the SL-A1 hairpin RNA products were only obtained when sequences corresponding to the SL-B1 stem loop were added at the 5¢-end giving the (–)IRES104 RNA

A significant level of RNA produced from the (–)IRES

104 RNA was obtained even though it is about 25% that of the wild-type (–)IRES However in the latter case the synthesis is less processive as a pause is observed in about half of the initiation events These results indicate that the polymerase frequently dissoci-ates from the template after initiation from (–)IRES

104 RNA suggesting that sequences between nt 104 and 341 stabilize the RNA polymerase complex during the elongation process Data obtained with other 5¢ deleted mutants (–)IRES 219 and (–)IRES 239, con-firmed these results In both cases, these two templates can sustain efficient elongation without polymerization arrest and NS5B release from the template The level

of RNA synthesized from (–)IRES 219 is in the range

of that observed with (–)IRES 104 but the addition of

20 nucleotides at the 3¢-end to give (–)IRES 239 enhances this process by doubling the amount of RNA product (Table 1)

A striking observation when analyzing the products obtained with our recombinant NS5B-1a is the pres-ence of high molecular weight RNA two to three times the size of the template These products correspond to successive copies of the (–)IRES RNA [17] In this study, we showed that these high molecular weight products were also synthesized with the recombinant RdRp of an HCV of genotype 1b (Fig 4B) but not with the NS5B of the highly related GBV-B virus (Fig 6A) These products were not specifically pro-duced from templates derived from the 3¢-end of HCV minus-strand RNA as they were also present when RNA fragments of the 3¢UTR were used as templates Data from RdRp assays performed in the presence of heparin indicated that these products occurred after a reinitiation event on a different template except in the case of a product of about 375 nt when using the (–)IRES 239 as template The nature of the latter

375 nt RNA remains to be elucidated

A

B

Fig 7 Gel shift assay with wild-type or 239 (–)IRES RNA After

denaturation and renaturation, 32P-labeled RNA 341 (–)IRES and

239 (–)IRES (13 n M , 167 Bq) were incubated with NS5B (50 n M ,

100 n M , 200 n M , 500 n M , 1000 n M and 2500 n M ) in the RdRp

reac-tion mixture without nucleotides for 20 min at 25
C The reaction

products were analyzed on a 4% polyacrylamide nondenaturing gel.

The gel was autoradiographied and the amount of unbound RNA

determined by scanning with NIH image (A) Autoradiogram of the

gel (B) The percentage of wild-type (d) or 239 (s) RNA bound with

enzyme was plotted against NS5B concentration.