Proc. Nati. Acad. Sci. USAVol. 91, pp. 8719-8723, August 1994Medical Sciences

The physical state of the negative strand of hepatitis C virus RNAin serum of patients with chronic hepatitis CMICHIKO SHINDO*, ADRIAN M. Di BISCEGLIE*t, TOSHITAKA AKATSUKAt, TSE-LING FONG*,LUCA SCAGLIONEt, MIKHAIL DONETSt, JAY H. HOOFNAGLE*, AND STEPHEN M. FEINSTONE**The Liver Diseases Section, Digestive Diseases Branch, National Institute of Diabetes and Digestive and Kidney Diseases, National Institutes of Health andtThe Laboratory of Hepatitis Research, Center for Biologics Evaluation and Research, Food and Drug Administration, Bethesda, MD 20892

Communicated by James H. S. Gear, March 23, 1994

ABSTRACT Negative strands of the hepatitis C virus(HCV) genome (a positive-stranded RNA virus) have beenfound in a nuclease-resistant form in the serum of patients withHCV infections. We determined whether a complete negative-strand copy is present in the serum, whether the negativestrand is particle-associated, and finally, whether it is virion-associated and encapsidated like the positive (genomic) strand.Isopyknic sucrose and cesium chloride density ultracentrifu-gation followed by a strand-specific reverse transcription-polymerase chain reaction on the collected fractions was per-formed to determine whether both positive and negativestrands were associated with similar particles. Both strandscomigrated to approximately the same density (1.11-1.16g/cm3) in sucrose. After treatment ofthe plasma with detergent(0.1% Nonidet P-40) to remove the viral envelope and centrif-ugation on cesium chloride gradients, the positive strandsshifted to a density of 1.35 g/cm3, and the negative strands werenot detected. By using antibodies specific for the HCV core orenvelope glycoproteins El or E2 coated onto the wells of amicrotiter plate, it was possible to specifically bind HCV orviral cores to the solid phase. Pelleted virus particles wereresuspended in either PBS or PBS with 0.1% Nonidet P-40 toexpose the core. These pellets were then incubated in antibody-coated microtiter wells. RNA extracted from the bound andunbound fractions was tested for HCV RNA. The anti-coreantibody was able to bind positive strands but not negativestrands only in detergent-treated samples. In the nondetergent-treated pellets, the anti-El and -E2 bound the positive strand,but only anti-El bound the negative strands. These findingsindicate that while both strands of HCV RNA can be detectedin serum, the positive strand is encapsidated within the envel-oped core, and the negative strand appears to be in a membraneparticle associated with the viral envelope protein El but doesnot appear to be within the HCV core of circulating virions.

Hepatitis C virus (HCV) is an enveloped single-strandedpositive-sense RNA virus that is a member ofthe Flaviviridaefamily (1-7). Analysis of the viral genome suggests that likeother flaviviruses, it encodes an RNA-dependent RNA poly-merase that should synthesize RNA without a DNA inter-mediate and no HCV DNA sequences have been detected ininfected tissue (8). Therefore, it is assumed that the HCVgenomic RNA acts both as an mRNA for translation of viralproteins and as a template for the transcription ofthe negativeRNA strand intermediates. If this strategy is used by HCV,one would expect to find the negative (antigenomic) RNAstrands of HCV in the cytoplasm of infected cells but not inthe viral particles nor in the extracellular serum.A combination of reverse transcription (RT) ofHCV RNA

into cDNA followed by amplification of the DNA by PCR(RT-PCR) has been particularly useful in detecting HCV

genomes in clinical specimens (9-19). Since HCV is a single-stranded RNA virus, the independent detection and quanti-tation of plus and minus strands using a strand-specificRT-PCR is possible (11).We have reported (11, 20) that the negative strand ofHCV

RNA was present not only as expected in the liver where thevirus replicates but also in extracellular serum of manypatients with chronic hepatitis C and experimentally infectedchimpanzees with acute HCV infections. Both strands ofHCV detected in serum were resistant to RNase prior toextraction with guanidinium isothiocyante, detergent, andphenol/chloroform, suggesting that both were physicallyprotected from serum nucleases but were not present asdouble-stranded RNA.To further characterize the HCV RNA detected in serum,

we have carried out additional experiments on the specificityofthe strand-specific RT-PCR, the nature of the antigenomicRNA detected in serum, and the physical-chemical state ofthe viral particles that encapsidate the viral genome.

MATERIALS AND METHODSSera from 33 patients with chronic hepatitis C were tested forHCV RNA by using a strand-specific RT-PCR. Four serumsamples with high titers of both positive and negative strandsof HCV RNA were pooled to have a sufficient quantity of auniform reagent for further studies. The pool had a PCR titerof 10W for positive strands and 104 for negative strands ofHCVRNA.Both the positive and negative strands ofHCV RNA were

separately detected and quantitated by strand-specific RT-PCR using a modification of previously described methods(11, 20). Oligonucleotide primers used were chosen from thehighly conserved 5' noncoding (5'NC) region of the HCVgenome (8-11). In selected experiments, primers from thenonstructural regions 3 (NS3) and 5 (NS5) were also used.Double PCR using nested primers was employed to achievethe sensitivity required to detect the low levels ofHCV RNAfound in most clinical samples.

Strand-specific RT was performed in 20 gl containing 5 plof serum RNA, 0.2 ,umol of either the sense or antisenseprimer, all four dNTPs (each at 200 ILM), 20 units of RNasin(Promega), PCR buffer (10 mM Tris*HCl, pH 8.3/50 mMKCI/1.5 mM MgC12/0.01% gelatin), and 0.5 unit of avianmyeloblastosis virus reverse transcriptase (Promega). Thereverse transcriptase was then inactivated by heating at 950Cfor 30 min followed by quenching rapidly on ice.PCR amplification was performed by adding 80 pd of PCR

buffer containing 0.2 pnmol of the opposite-sense primer and

Abbreviations: RT, reverse transcription; NP-40, Nonidet P-40;HCV, hepatitis C virus; NS, nonstructural region.tTo whom reprint requests should be addressed at: The LiverDiseases Section, Building 10, Room 9C 103B, National Institute ofDiabetes and Digestive and Kidney Diseases, National Institutes ofHealth, Bethesda, MD 20892.

8719

The publication costs of this article were defrayed in part by page chargepayment. This article must therefore be hereby marked "advertisement"in accordance with 18 U.S.C. §1734 solely to indicate this fact.

8720 Medical Sciences: Shindo et al.

2 units of Taq polymerase (Perkin-Elmer/Cetus). Twenty-five PCR cycles were run followed by a second round of 35cycles using 10 M4 of the first reaction products and the innerprimer set. PCR products were analyzed by agarose gelelectrophoresis with ethidium bromide staining. The ex-pected size of the PCR product using the 5'NC primers was257 bp (Fig. 1). End-point titers ofHCV RNA were estimatedby testing 1:10 serial dilutions of extracted RNA.For each set of strand-specific RT-PCR assays, a series of

controls was performed. To exclude the possibility that thereaction mixture was contaminated with HCV cDNA and toexclude the possibility that the Taq polymerase had reversetranscriptase activity under the conditions of this reaction, acontrol was performed using all the components exceptreverse transcriptase. To rule out the possibility of residualreverse transcriptase activity after the heat-inactivation stepand the possibility that the RNA could self-prime resulting ina false positive-strand-specific reaction, a reaction was doneusing all components except the primer during the RT step(for this experiment, both sense and antisense primers wereadded to the PCR mixture after heat inactivation of thereverse transcriptase). To rule out contamination ofthe senseprimers with antisense primers, the RT reaction and firstPCRwas carried out using only the sense primer.

During the development of the strand-specific RT-PCR, itwas found that it was difficult to completely eliminate allresidual avian myeloblastosis virus reverse transcriptaseactivity by heating. When an excess of reverse transcriptasewas used, heating the reaction mixture to 950C for 30 min didnot always completely inactivate reverse transcriptase activ-ity. When the reaction mixture was heated for longer (60-120min), the reverse transcriptase activity was destroyed, butthe sensitivity of the reaction was decreased. To avoid theneed for excess heating, the optimal amount of reversetranscriptase activity that could be completely inactivated byheating to 950C for 30 min without reducing the sensitivity ofthe PCR assay was determined. The optimal amount ofreverse transcriptase per reaction mixture was determined tobe 0.5 unit for the 5'NC primers, 1.5 units for the NS3primers, and 0.5 unit for the NS5 primers (data not shown).As an additional control for specificity, phenol/chloroform

extraction of the RT reaction mixture after heat inactivationand prior to PCR was also used to ensure the completeelimination of residual reverse transcriptase activity. Afterheating and phenol/chloroform inactivation of the reversetranscriptase, strand-specific RT-PCR results still indicatedthe presence of negative-strand HCV RNA in serum.

'El .'::..I.

Sucrose gradient ultracentrifugation was performed todetermine whether both positive and negative strands wereassociated with particles that cosedimented. The pooledHCV serum was layered onto 20-60% sucrose gradients,centrifuged in a SW 60 rotor at 46,000 rpm at 4°C for 20-24hr. Eight 500-,u fractions were collected from the top of thetube. RNA was extracted from each fraction and subjected tostrand-specific RT-PCR.

Ultracentrifugation was performed in cesium chloride gra-dients with or without the addition of 0.1% Nonidet P-40(NP-40). The pooled serum was layered on cesium chlorideor cesium chloride/NP-40 gradients (0.78-4.8 M) and cen-trifuged at 54,000 rpm in a SW 60 rotor at 4°C for 20 hr.Sixteen 250-,lI fractions were collected from the top of thegradient. RNA was extracted from each fraction and sub-jected to the strand-specific RT-PCR.RNase sensitivity of the positive and negative strands was

evaluated before and after removal of the viral envelope bydetergent. Viral particles were pelleted by ultracentrifugationof the pooled serum through 3.5 ml of 20% (wt/vol) sucrosein a SW 60 rotor at 58,000 rpm for 3 hr at 4°C. Pellets wereresuspended in 200 Al of PBS and divided into two equalaliquots. NP-40 (0.1% final concentration) was added to onealiquot, which was incubated at room temperature for 3 hr toremove the viral lipid envelope. Portions of these resus-pended pellets were then incubated with or without RNase A(1 mg/ml) at 37°C for 30 min. RNA was then extracted andsubjected to strand-specific RT-PCR.

Solid-phase immunocapture assays using human monospe-cific antibodies to HCV core antigen or mouse monoclonalantibodies raised against either the core or envelope glyco-proteins El and E2 were used to test whether both RNAstrands were incorporated in a particle that could be boundby HCV-structural-protein-specific antibodies. The humanantibodies were produced by Epstein-Barr virus-trans-formed human B cells (21) and the mouse monoclonal anti-bodies were made by standard hybridoma technology frommice that had been hyperimmunized with HCV antigensproduced by recombinant baculoviruses expressing specificHCV proteins (H. Hsu, M.D., T.A., and S.M.F., unpub-lished data; and ref. 22).The same HCV serum pool used for the centrifugation

studies was used for the antibody-capture assays. Particleswere pelleted from the serum and either resuspended in PBSor treated with PBS plus 0.1% NP-40 for 3 hr. The pelletswere then incubated in wells of microtiter plates coated withantibody to HCV core protein, El, or E2 for 1 hr at 37°C.Supernatants were removed and saved for testing as the

NS3 NS5

270 bp -_257 bp _-*

148 bp .-*

FIG. 1. Ethidium bromide-stained agarose gels demonstrating detection ofthe negative-stranded HCV RNA in serum by using strand-specificPCR with primers from 5'NC, NS3, and NS5 regions of the HCV genome. Each primer set consisted of four primers for nested PCR listed inorder of the outer sense, outer antisense, inner sense, and inner antisense primers with the nucleotide numbering from the 5' to the 3' base ofthe primer. 5'NC primers: nt 1-21, GGCGACACTCCACCATAGATC; nt 324-304, GGTGCACGGTCTACGAGACCT; nt 28-48, CTGTGAG-GAACTACTGTCTTC; nt 284-264, CCCTATCAGGCAGTACCACAA. NS3 region primers: nt 5112-5135, AGCCACCGTGTGCGCTAGGGC-TCA; nt 5354-5333, GCCAAAGCAGCCAGGACGCCG; nt 5166-5190, AGTGTTTGATTCGCCTCAAGCCC; nt 5314-5295, TCGTGAC-GACCTCCAGGTCG. NS5 region primers: nt 8150-8177, ATGGGAAGCTCCTACGGATTCCAATACT; nt 8522-8494, AGTCCTGGAGC-CCTGCGGCTCGACAGGCT; nt 8192-8219, GTTGAATTCCTCGTGCAAGCGTGGAAGT; nt 8462-8435, TACCACAGCTAGTTGT-CAGTACGCCGCT.

Proc. Natl. Acad. Sci. USA 91 (1994)

Proc. Natl. Acad. Sci. USA 91 (1994) 8721

Fraction

Density

POSITIVESTRAND

257 bp -_

NEGATIVESTRAND

257 bp -.

1 2 3 4 5 6 7 8 9 10 1112131415 16110 1.28 1.69

23.45>6 7 8 9 1 |1 2 1 4 1 1 e4 35

257 bp -_

257 bp-

1.801.70 -

1.60 -

1.501.40 -1.30 }--1.20-

1.1 [ii..-..I- ......±. I I 1 1 1 11 2 3 4 5 6 7 8 9 1011 12131415 16

80 r

60 F

.50V140 K

2 0r

1 0 L

? 4 ;) 8 Q1'' .^ 10d<A1 sF.

FIG. 2. Detection of both positive- and negative-stranded HCV RNA in serum after isopyknic ultracentrifugation in cesium chloride densitygradients. (Left) Without NP-40 treatment. Positive-stranded HCV RNA was detected in fractions 2-7 (densities, 1.10-1.28 g/cm3), andnegative-stranded HCV RNA was detected in fractions 2-4. (Right) With NP-40 pretreatment. Positive strands of HCV RNA were detectedonly in fraction 9 (1.35 g/cm2) and negative strands of HCV RNA were not detected in any fraction.

unbound fraction. The microtiter wells were then washedfour times with NET buffer (150 mM NaCl/0.01 mM EDTA/100mM Tris-HCl, pH 8.0) and RNA was extracted by addingthe guanidium extraction buffer directly to the microtiterwells. The unbound fraction was extracted in the usual wayand both fractions were tested by strand-specific RT-PCR.The specificity of the antibody capture assays was dem-

onstrated by blocking experiments using the HCV core, El,and E2 baculovirus-expressed HCV antigens. Cell lysatesfrom recombinant baculovirus-infected SF-9 cells were usedas the source of the antigen. These blocking antigens (100 IL)were preincubated in the wells ofthe microtiter plate that hadbeen coated with 100 pd of the human or mouse monoclonalantibodies followed by 200 p.1 of 10% (wt/vol) bovine serumalbumin. The antigens were aspirated from the wells, and theserum pellets were added, incubated, and treated as de-scribed above.

RESULTS

HCV RNA was detected in serum of all 33 patients studiedby using primers from the 5'NC region. By strand-specificRT-PCR, all 33 had the positive strand of HCV RNA while13 (39%) also had detectable negative strand. In every case,controls for specificity of the detection of the negative strandwere negative.

Since all the strand-specific RT-PCRs were performedwith primers from the conserved 5'NC region, there was thepossibility that only a fragment of the negative-strand RNAwas actually present in the serum. Therefore, we also per-formed the strand-specific RT-PCR using primers from theputative NS3 and NS5 coding regions of the genome inaddition to the 5'NC region. The negative strand could bedetected with all three sets of primers, suggesting that mostof the genome of HCV was represented as negative strandsin the serum (Fig. 1).The pooled serum was subjected to ultracentrifugation on

20-60% sucrose gradients. Eight 0.5-ml fractions were col-lected from the top of the tube and tested for positive andnegative strands of HCV RNA. Both positive and negative

strands were detected in fractions 3-5, with densities rangingfrom 1.11 to 1.16 g/cm3. The peak titers were in fraction 4(1.15 g/cm3). Thus, both strands cosedimented to approxi-mately the same density in sucrose. HCV RNA was alsodetected in small amounts at the top of the gradient (possiblydue to lipid binding) and at the bottom (possibly due to releaseof free viral RNA during centrifugation).To band nucleocapsids expected to have a density greater

than 60% sucrose, ultracentrifugation in cesium chloridegradients containing NP-40 was performed. Sixteen fractionsfrom gradients with or without 0.1% NP-40 were separatelytested for both positive- and negative-strand HCV RNA.Without detergent, the positive-stranded HCV RNA wasdetected over a broad range in fractions 2-7 (1.10-1.28g/cm3) with peak titers in fraction 3 (1.15 g/cm3) (Fig. 2).Again, small amounts ofRNA were detected near the bottomof the gradient probably due to released RNA. The negativestrand was detected only in fractions 2-4 (1.10-1.28 g/cm3).In contrast, with detergent added to the gradient, positive-stranded HCV RNA was found only in fraction 9 (1.35g/cm3), and negative strands were not detected.RNase sensitivity of both strands was tested in viral

particles that had been pelleted through 20% sucrose. When

Table 1. RNase A sensitivity of the positive and negative strandsof HCV RNA in pellets of serum from HCV chronic carriers

Serum pellet treatment Strand-specific PCR

RNase Positive NegativeNP-40 RNase A inhibitor strand strand

_- + +

_ + -+ +

+ + -+_+ --+ +

+ + + + +

Pooled HCV positive sera were pelleted through 20%1b sucrose asdescribed in the text. The pellets were treated with various combi-nations of NP-40, RNase A, and placental RNase inhibitor asdescribed in the text. Treatment: -, not added; +, added. Results:+, detected; -, not detected.

c

Medical Sciences: Shindo et al.

8722 Medical Sciences: Shindo et al.

Table 2. PCR titers for positive and negative strands of HCVRNA by capture assays using specific antibodies to HCV core,El, and E2 antigens

PCR titer

Capture Bound fraction Unbound fractionantibody + strand - strand + strand - strand

Anti-core- NP-40 - - 10-3 10-3+ NP-40 10-2 - 10-2 10-2

Anti-El- NP-40 10-2 10-2 10-2 10-2+ NP-40 10-1 - 10-2 10-2

Anti-E2- NP-40 10-1 _ 10-2 10-2+ NP-40 100 - 10-2 10-2Capture antibodies were used to coat the wells of microtiter plates

as described in the text. The antibodies and their specificities aredescribed in the text. In other experiments, the specific binding ofHCV or the HCV cores to the antibody-coated wells was completelyblocked by the homotypic recombinant baculovirus antigen but nota heterotypic antigen (i.e., baculovirus-expressed core antigenblocked binding of viral cores but the El antigen did not).

the pellets were resuspended in PBS, both positive andnegative strands ofHCV RNA were detected despite RNasetreatment. When pellets were resuspended in 0.1% NP-40,RNase digestion resulted in loss ofthe negative strand but thepositive strand remained detectable. The RNase sensitivityof the negative strand of HCV RNA after NP-40 treatmentsuggested that the negative strand was not contained withinthe HCV nucleocapsid. However, if placental RNase inhib-itor was added before detergent treatment, both positive andnegative strands could be detected (Table 1).The specific antibodies to core, El, and E2 proteins were

used to coat wells of microtiter plates in which HCV-positiveserum pellets with and without NP-40 treatment were incu-bated. The amounts ofboth bound and unbound negative andpositive strands of HCV RNA were then quantified bystrand-specific RT-PCR (Table 2). Anti-El bound both pos-itive and negative strands ofHCV RNA at a 1:100 dilution forboth while the anti-E2 bound only the positive strand. Thetiter of the positive strand was reduced to 101 for anti-El and100 for anti-E2 if the pellets were pretreated with NP-40 andthe negative strands could not be detected. In contrast,microtiter plates coated with antibodies toHCV core retainedHCV RNA only in samples that had been pretreated withNP-40 and, importantly, retained only positive-strandedHCV RNA.Antigen blocking tests were performed to prove that the

reactions were specific. The binding of both strands ofHCVRNA by anti-El- and anti-E2-coated microtiter plates couldbe blocked by preincubation of the microtiter wells withbaculovirus-produced recombinant El and E2 antigen but notby the core antigen. Furthermore, binding of the positivestrand ofHCV RNA in the detergent-treated samples to theanti-HCV-core-antigen-coated microtiter plates was com-pletely blocked by preincubation of the NP-40-treated pelletswith baculovirus-produced recombinant HCV core antigenbut not the El/E2 antigen. Finally, both positive and nega-tive strands of HCV RNA could be detected in the unboundfractions in these blocking experiments in titers of 102 to 103.

DISCUSSIONHCV resembles other members of the flavivirus family (2-5,23, 24). While nothing directly is known about the replicativestrategy of HCV, it is assumed to be similar to that of otherbetter-studied flaviviruses. Assumptions based on close anal-

ogy with flaviviruses or pestiviruses, however, may be mis-leading as HCV appears to be unique in several ways, suchas the size of the 5'NC region and the structure of envelopeantigens. While virion RNA is clearly positive-stranded,negative strands ofHCV RNA have been detected (11) in theextracellular serum. The specificity of assays used to detectnegative-stranded HCV RNA in serum has been questioned(25), but methodologic considerations alone do not explainthese findings. In this study, specificity controls were carriedout with each experiment and showed that the Taq polymer-ase was not acting as a reverse transcriptase, that the viralRNA was not contaminated by cDNA, that the viral RNAwas not folding back on itself and self-priming the RTreaction, and that the single primers used in the RT step werenot contaminated with the opposite-sense primer. The spe-cific concern regarding total inactivation ofreverse transcrip-tase activity in the sensitive RT-PCR assay was addressedby extensive inactivation steps using heat and phenol/chloroform extraction and by careful titration of the optimal(but least) amount of reverse transcriptase required for theassay. These results reconfirm the finding of negative-stranded HCV RNA in serum of a proportion of patients withchronic hepatitis C. In addition, if the negative strands wereonly an artifact due to a false positive-strand-specific RT-PCR, the negative strands should have been found every-where the positive strands were found. As we could detectthe negative strands in only 13 out of 33 patients withpositive-strand HCV RNA and the detection of negativestrand was not directly dependent on the titer of the positivestrand (11), the artifact hypothesis is clearly wrong. In thisstudy we also were able to physically separate the positivestrands from the negative strands by density gradient ultra-centrifugation, further demonstrating the validity of thestrand-specific detection system.We have sought in this study to define the state that the

HCV negative-strand RNA exists in the serum. We havereconfirmed the previous result that the negative strand issensitive to RNase A after the serum is extracted. Thissuggests that the negative strand is not protected by virtue ofits being duplexed with the positive strand. We have at-tempted to determine whether the negative strand is associ-ated with the viral particle and perhaps encapsidated like thegenomic RNA. While it has not been possible to determineprecisely the physical state of the negative strand, it has beenpossible to get a reasonably clear idea of how it exists in theserum. First primer sets from the 5' end (5'NC), the middlesection (NS3), and the 3' end (NS5) of the viral genome wereused successfully in the strand-specific RT-PCR assays forthe negative strand of HCV RNA, indicating that negativestrands included most, if not all, the HCV genome. Further-more, the susceptibility of the extracted HCV RNA toRNases suggested that the negative strand circulated sepa-rately from the positive strand and was not present in theextracellular space as duplexed RNA. Both the positive andnegative strands of HCV RNA detected in serum of infectedpatients were associated with lipids and presumably lipidmembrane particles as demonstrated by their resistance toRNase digestion before the lipid/protein extraction and bytheir sedimentation characteristics. Both strands of HCVRNA banded in isopyknic sucrose gradients at a density of1.15 g/cm2, which is similar to banding density of other

flaviviruses, and suggests that the virus has a lipid-richenvelope. By using RT-PCR for HCV RNA as the detectionsystem, HCV particles were found to band at a heavierdensity (1.35 g/cm2) after lipid extraction, suggesting that thevirion is composed of a nucleocapsid inside of the lipidenvelope. Importantly, however, only the positive strand ofHCV RNA was found to be associated with this detergent-resistant (presumably nucleocapsid) particle, indicating thatthe negative strands of HCV RNA detected in serum were

Proc. Natl. Acad. Sci. USA 91 (1994)

Proc. Natl. Acad. Sci. USA 91 (1994) 8723

associated with a lipid membrane but were not encapsidatedlike the positive strands. It remains possible that the differ-ence in titer of the positive and negative strands resulted inthe different ability to detect them after detergent treatment.However, since both the positive and negative strands wereinitially present in easily detected quantities, we feel that thisis not the reason for the loss of negative strands afterdetergent treatment.

This hypothesis for the structure of HCV and the separatestructures containing positive- and negative-stranded HCVRNA was further supported by results of antibody-captureassays using monoclonal antibodies to HCV core and enve-lope antigens. Mouse monoclonal antibodies to El capturedboth positive and negative HCV RNA strands and a mono-clonal antibody to the E2 glycoprotein captured only thepositive strand. In contrast, specific antibody to HCV coreantigen captured only positive-stranded HCV RNA and onlyin samples that had been pretreated with detergents to stripoff the outer lipid-containing envelope. These antibody bind-ing assays could be blocked by the specific recombinantantigens used to select the human antibodies and to immunizethe mice for the mouse monoclonal antibodies.These data support the concept that HCV is like other

flaviviruses in that it has a lipid-membrane envelope with thegenomic RNA contained within a core particle. The antibody-capture data also support the suggestion that HCV has twoenvelope glycoproteins, El and E2 (24). The E2 glycoproteinis coded by a region of the HCV genome that is analogous tothe NS1 protein in flaviviruses and has been labeled E2/NS1.These capture data provide direct evidence that E2 is indeeda structural protein that is associated with the viral envelope.The genome of HCV is a single-stranded positive-sense

RNA molecule that is contained within the nucleocapsid. Theantigenomic strand of HCV RNA that is frequently detectedin the serum of infected individuals by highly sensitive PCRtechniques seems to be associated with a lipid membranestructure that is also associated with at least the El glyco-protein. It remains possible that this membrane structure isactually the virion particle, but it seems more likely that it ispart of the replication complex that has spilled out of dam-aged cells. Since the PCR technique that we have used candetect as few as 10 RNA molecules, and other positive-stranded RNA viruses have not been studied with techniquesas sensitive, it is possible that low levels of extracellularnegative-strand RNA could be found for other viruses if asimilar technology were applied.The clinical and biologic significance of the finding of

negative strands of HCV RNA in serum remains undefined.An important implication of our results is that the detectionof the negative strand of HCV RNA is not necessarily amarker of viral replication in some particular tissue sample.It is not possible to easily eliminate all the blood from tissuesamples. Therefore, negative strands of HCV RNA detectedin nonhepatic tissues may have originated in the liver andwere detected only as a contaminant in the organ studied. Inthe few studies that have tried to quantify the amount ofnegative and positive strands of HCV RNA in tissue, itappears that the relative amounts of the two strands canreflect replication, since the relative amounts of negative and

positive strands ofHCV RNA in liver are close to equivalent(11).

1. Choo, Q. L., Kuo, G., Wiener, L. R., Overby, D. W. &Houghton, M. (1989) Science 244, 359-362.

2. Choo, Q. L., Richman, K. H., Han, J. H., Berger, K., Lee, C.,Dong, C., Gallego, C., Coit, D., Medina-Selby, A., Barr, P. J.,Weiner, A. J., Bradley, D. W., Kuo, G. & Houghton, M. (1991)Proc. Nati. Acad. Sci. USA 88, 2451-2455.

3. Hijikata, M., Kato, N., Ootsuyama, M., Nakagawa, M. &Shimotohno, K. (1991) Proc. Natl. Acad. Sci. USA 88, 5547-5551.

4. Kato, N., Hbikata, M., Ootsuyama, Y., Nakagawa, S., Ohi-koshi, Y., Sugiyama, T. & Shimotohno, K. (1990) Proc. Natl.Acad. Sci. USA 87, 9524-9528.

5. Kubo, Y., Takeuchi, S., Boonmar, S., Katayama, T., Choo,Q. L., Kuo, G., Weiner, A. J., Bradley, D. W., Houghton, M.,Saito, I. & Miyamura, T. (1991) Nucleic Acids Res. 17, 10367-10372.

6. Okamoto, H., Okada, S., Sugiyama, S., Yotsumoto, S., Ta-naka, H., Yoshizawa, H., Tsuda, F., Miyamura, Y. & Mayumi,M. (1990) Jpn. J. Exp. Med. 60, 167-177.

7. Takamizawa, A., Mori, C., Fuke, I., Manabe, S., Murakami,S., Fujita, J., Onishi, E., Andoh, T., Yoshida, I. & Okayama,H. (1991) J. Virol. 65, 1105-1113.

8. Weiner, A. J., Kuo, G., Bradley, D. W., Bonino, F., Saracco,G., Lee, C., Rosenblatt, J., Choo, Q. L. & Houghton, M.(1990) Lancet 355, 1-3.

9. Cristiano, K., Di Bisceglie, A. M., Hoofnagle, J. H. & Fein-stone, S. M. (1991) Hepatology 14, 51-53.

10. Farci, P., Alter, H. J., Wong, D., Miller, R. H., Shih, J. W.,Jett, B. & Purcell, R. H. (1991) N. Engl. J. Med. 325, 98-104.

11. Fong, T. L., Shindo, M., Feinstone, S. M., Hoofnagle, J. H. &Di Bisceglie, A. M. (1991) J. Clin. Invest. 88, 1058-1060.

12. Garson, J. A., Tedder, R. S., Brigg, M., Tuke, J. A., Glaze-brook, J. A., Trute, A., Parker, D., Barbara, J. A. J., Conreas,M. & Aloysis, S. (1990) Lancet 335, 1419-1422.

13. Kaneko, S., Unoura, M., Kobayashi, K., Kuno, K., Mura-kami, S. & Hattori, N. (1990) Lancet 335, 976.

14. Kato, N., Yokosuka, O., Omata, M., Hosoda, K. & Ohoto, M.(1990) J. Clin. Invest. 86, 1764-1767.

15. Okamoto, H., Okada, K., Sugiyama, Y., Yotsumoto, S., Ta-naka, H., Yoshizawa, H., Tsuda, F., Miyamura, Y. & Mayumi,M. (1990) Jpn. J. Exp. Med. 60, 215-222.

16. Simmonds, P., Zhang, H. G., Watson, H. G., Rebus, E. D.,Fergason, P., Balfe, G. H., Leadbetter, P. L., Yap, J. F.,Peuthere, J. F. & Ludlam, C. A. (1990) Lancet 336,1469-1472.

17. Shimizu, Y. K., Weiner, A. J., Rosenblatt, J., Wong, D. C.,Shapiro, M., Popkin, T., Houghton, M., Alter, H. J. & Purcell,R. H. (1990) Proc. Natl. Acad. Sci. USA 87, 6441-6444.

18. Shindo, M., Di Bisceglie, A. M., Cheung, L., Shih, J. W.,Feinstone, S. M., Cristiano, K. & Hoofnagle, J. H. (1991) Ann.Intern. Med. 115, 700-704.

19. Ulrich, P. P., Romeo, J. M., Lane, P. K., Kelly, I., Daniel,L. J. & Vyas, G. N. (1990) J. Clin. Invest. 86, 1609-1614.

20. Shindo, M., Di Bisceglie, A. M., Biswas, R., Mihalik, K. &Feinstone, S. M. (1992) J. Infect. Dis. 166, 424-427.

21. Akatsuka, T., Donets, M., Scaglione, L., Ching, W.-M., Shih,J. W.-K., Di Bisceglie, A. M. & Feinstone, S. M. (1993) He-patology 18, 503-510.

22. Hsu, H. H., Donets, M., Greenberg, H. B. & Feinstone, S. M.(1993) Hepatology 17, 763-771.

23. Grakoui, A., McCourt, D. W., Wychowski, C., Feinstone,S. M. & Rice, C. M. (1993) J. Virol. 67, 2832-2843.

24. Houghton, M., Weiner, A. J., Han, J., Kuo, G. & Choo, Q. L.(1991) Hepatology 14, 381-387.

25. McGuiness, P. H., Bishop, G. A., McCaughan, G. W., Trow-bridge, R. & Gowans, E. J. (1994) Lancet 341, 904-905.

Medical Sciences: Shindo et al.