Journal of General Virology (1993), 74, 1093-1102. Printed in Great Britain 1093

Typing of hepatitis C virus isolates and characterization of new subtypes using a line probe assay

Lieven Stuyver ,* Rudi Rossau , Ann Wyseur , M a d d y Duhamel , Bart Vanderborght , H u g o Van H e u v e r s w y n and Geert Maer tens

Innogenetics N.V. , Industriepark Zwijnaarde 7, Box 4, B-9052 Gent, Belgium

A reverse-hybridization assay, the line probe assay (LiPA), based on variations found in the 5' untranslated regions of the different hepatitis C virus (HCV) geno- types was developed, permitting simple and fast de- termination of four HCV genotypes and their subtypes. Using this assay, 61 PCR-positive Brazilian HCV sera were typed. Of the sera, 33% had a type 1 HCV infection, 38 % had type lb (related to HCV-J), 1.5 % had type 2a (related to HC-J6), 24.5 % had type 3 (related to E-bl and HCV-T), and 3 % of the sera were co-infected. This assay format was further evaluated using 13 sera from Belgium and the Netherlands, and all

of these could be classified. Two pools of Japanese sera were classified as either type 2a or were co-infected with types lb and 2a, but no type 2b sequences were detected. Another eight PCR-positive sera were obtained from Burundi and Gabon. The sequence of the 5' untranslated region of these African viruses was strongly divergent from the three previously described types. Therefore, these isolates were tentatively classified as type 4. These and some of the other non-type 1 sera often demon- strated weaker reactivities than type 1 isolates in currently used second generation antibody confirmation assays.

Introduct ion

Hepatitis C viruses (HCV) are a family of positive- stranded, enveloped RNA viruses, causing most non-A, non-B (NANB) hepatitis. Their genome organization indicates a close relationship to the Pestiviridae and Flaviviridae. The sequences of cDNA clones covering the complete genome of several prototype isolates have already been completely determined (Kato et al., 1990; Takamizawa et al., 1991 ; Choo et al., 1991, Okamoto et al., 1991, 1992b; Inchauspe et al., 1991). Comparison of these isolates shows a considerable variability in the envelope (E) and non-structural (NS) regions, while the 5' untranslated region (UR) and, to a lesser extent, the core region are highly conserved.

Using cloned sequences of the NS3 region, Kubo et al. (1989) compared a Japanese and an American isolate and found nearly 80 % nucleotide and 92 % amino acid identity. The existence of sequence variability was further documented when sequences of the 5' UR, and of the core and E1 regions became available (HC-J1 and HC- J4; Okamoto et al., 1990). After the isolation of several

The nucleotide sequence data reported in this paper have been assigned to the DDBJ, EMBL and GenBank databases under the following accession numbers: Dl13448, D13449, D13450, D13451, D13452, D13453, D13992, D13993, D13994, D13995, D13996 and D13997.

NS5 fragments in Japanese laboratories, the two types K1 and K2 were described (Enomoto et al., 1990). A comparison of the American-like isolate PT-1 with K1, which was more prevalent in Japan, showed that they represent closely related but different subtypes with an inter-group nucleotide identity of about 80 %. The K2 sequence was more distantly related to both K1 and PT- 1 because only 67 % of the nucleic acid and 72 % of the amino acid sequences were the same. Moreover, K2 could be divided into two groups, K2a and K2b, also possessing inter-group nucleotide identity of about 80 %. Nucleotide sequence analysis in the 5' UR showed 99 % identity between K1 and PT-1, and at most 94 % identity between K1 and K2, enabling the use of the 5' UR for studying restriction fragment length polymorphism (RFLP) and classification of HCV into the types K1 and K2 (Nakao et al., 1991). Further evidence for a second type was given by the complete sequences of HC-J6 and HC-J8, two sequences related to the K2 type (Okamoto et al., 1991, 1992b). A phylogenetic tree of HCV containing four branches (i.e. type I: HCV-1 and HCV- H; type II: HCV-J, -BK, HC-J4; type III: HC-J6; type IV: HC-J8) was proposed (Okamoto et al., 1992b). However, 79 % nucleic acid sequence similarity can be observed between types I and II, and also between types III and IV. The first group (types I and II) and the second group (types III and IV) appear to be less related, with only 67 to 68 % similarity. A further new type of HCV,

0001-1417 © 1993 SGM

1094 L. S tuyver and others

HCV-T, was detected in Thailand in a study of NS5 regions (Mori et al., 1992). HCV-T was about 65% identical to all other known NS5 sequences, and could be classified into two subtypes, HCV-Ta and HCV-Tb, of whose nucleic acid sequences 80% are the same. Elucidating the phylogenetic relationship of a similar new type in British isolates with types I to IV was possible by analysing the conserved parts of the 5' UR and the core, NS3 and NS5 regions (Chan et al., 1992a). A new phylogenetic tree was proposed whereby type 1 corresponds with types I and II, type 2 with types III and IV and type 3 with the isolates E-bl to E-b8 and HCV-T. Some sequences of the 5' UR of isolates from type 3 were also reported by Bukh et al. (1992), Cha et al. (1992) and Lee et al. (1992).

The demonstration that different HCV genotype infections resulted in different responses to interferon-0c treatment (Pozatto et al., 1991; Kanai et al., 1992; Yoshioka et al., 1992) and in distinct serological reactivities (Chan et al., 1991) stresses the importance of HCV genotyping. Until now, this could only be achieved by large sequencing efforts in the coding region or in the 5' UR, by PCR on HCV cDNA with type-specific sets of core primers (Okamoto et al., 1992a) or by RFLP analysis in the 5' UR (Chan et al., 1992b) or in the NS5 region (Nakao et al., 1991).

We used a sensitive PCR protocol for the highly conserved 5' UR with sets of nested, universal primers. The amplification products obtained were hybridized to oligonucleotides directed against the variable regions of the 5' UR, immobilized as parallel lines on membrane strips (reverse-hybridization principle). This hybridiz- ation line probe assay (LiPA), has recently been developed for the simultaneous detection of cystic fibrosis mutations, bacterial pathogens in clinical samples or for human leukocyte antigen typing (R. Rossau, unpublished results). By using this rapid assay, previously poorly described isolates related to Z5 and Z6 (Bukh et al., 1992) and GV (Cha et al., 1992) were detected. A new type 4 is proposed for these strains of HCV. The LiPA technology allows an easy and fast determination of four HCV types and their subtypes that are present in patients' sera.

Methods Serum samples. BR1 to BR61, 61 Brazilian samples, were tested in

the HCV antibody ELISA (Innotest HCV Ab, Innogenetics) and/or in the Inno-LIA (line immunoassay) HCV Ab II confirmation test (Innogenetics). The first 23 serum samples (BR1 to BR23, Table 1) were from haemodialysis patient with either high alanine transferase levels or positive Inno-LIA results, or from blood donors in cases where the recipient developed NANB liver disease. Serum samples BR24 to BR37 were randomly chosen and the 24 remaining sera (BR38 to BR61) were selected because of their Inno-LIA pattern. Most of the latter showed weakly positive, indeterminate, or negative reactivity with the NS4 and

NS5 synthetic peptides on the LIA. The following sera were also included in our characterization effort: two pools of Japanese sera (JP62 and JP63, a gift from Dr T. Arima, Kagoshima University, Japan), nine Belgian sera (BE64 to BE69, BE82, BE83 and BE84), four sera from The Netherlands (NE70 to NE73), six sera from Burundi (BU74 to BU79) and two sera from Gabon (GB80 and GB81). All of them were tested with the Inno-LIA HCV Ab II. BU74 to BU78 were positive only for anti-core antibodies, and BU79 reacted only with the NS5 line. Both Gabonese sera were LIA HCV-negative and human immunodeficiency virus-negative but human T cell leukaemia virus- positive. One serum from Belgium (BE69) and one from the Netherlands (NE73) were completely antibody negative. Three of the NE-sera (NE71 to NE73) were selected because they were negative or indeterminate in the second generation RIBA test (Ortho Diagnostics).

Primers andprobes. The primers used for PCR were complementary to the conserved areas of the 5' UR of the different HCV types. Sequence degeneracy was included to allow annealing to type 1 and type 2 sequences (Kato et al., 1990; Nakao et al., 1991 ; Okamoto et al., 1991) and to the sequence of our type 3 clone (BR56). The sequences of the outer PCR primers 1 and 2 and of the nested PCR primers 3 and 4 are listed in Table 2. The probes used for the detection of the different serum types are also listed in Table 2. All oligonucleotides were synthesized on a 392 DNA/RNA Synthesizer (Applied Biosystems).

cPCR, analysis of the PCR product, and cloning. Viral RNA was extracted from serum, essentially as described by Chomczynski & Sacchi (1987) but with minor modifications. The RNA was coprecipit- ated with 20 gg dextran T500 (Pharmacia). The RNA pellet was briefly dried and resuspended in 10 gl diethylpyrocarbonate-treated water. After adding 2 gl of 150 ng/gl random primers (Pharmacia) and 10 rain denaturation at 65 °C, first-strand cDNA synthesis was carried out in 20 gl reaction mixture at 42 °C in the presence of 25 units (U) human placental ribonuclease inhibitor (Amersham), 500 gM dATP, dCTP, dTTP and dGTP, 1 x avian myeloblastosis virus (AMV) buffer (Strata- gene) and 2.5 U AMV reverse transcriptase (Stratagene). Outer PCR amplified 7 lal of the resulting cDNA over 40 cycles, each consisting of 1 min at 95 °C, 1 min at 55 °C and 1 min at 72 °C in a total volume of 50 gl. The solution was adjusted to a final concentration of 200 gM dATP, dCTP, dTTP and dGTP, 1 x Taq buffer (Stratagene), 0-2 gM of each primer, and 1 U Taq polymerase (Stratagene). From the first round of amplification, 1 pt product was amplified again with the nested primers for 40 cycles in a buffer of the same composition. For HCV typing, the nested PCR contained 40 gM Bio-1 I-dUTP (Sigma) and 160 gM dTTP. Both the outer and the nested PCR products were then subjected to electrophoresis in a 2 % low melting point (NuSieve GTG, FMC)/1% Ultra Pure (Gibco BRL) agarose gel. After ethidium bromide staining, PCR fragments were cut out from the agarose gel, the DNA was recovered by centrifugation through a 0.45 gm HV membrane (Millipore), purified by two phenol-chloroform and two ether extractions, precipitated and subsequently treated with T4 DNA polymerase and T4 kinase (Boehringer). It was then ligated in the dephosphorylated EcoRV site of pBluescript K S ( - ) (Stratagene). Plasmid DNA preparation was as described in the alkaline lysis method (Maniatis et al., 1982). Sequencing reactions were carried out on double-stranded plasmid DNA with T7 and T3 primers by using the Deaza G / A T7 sequencing mixes (Pharmacia).

Preparation of the line probe assay strips. The 16-met oligo- nucleotides, specific for the different types or subtypes of HCV (Table 2, numbers 5 to 22), were provided with a poly(dT) tail at their Y end as follows. Twenty pmol primer was incubated for 1 h at 37 °C in 25 gl buffer containing 3.2 mM-dTTP, 25 mM-Tris-HC1 pH 7.5, 0.1 M-sodium cacodylate, 1 mM-CoCI~, 0.1 M-dithiothreitol and 60 U terminal deoxy- nucleotidyl transferase (Pharmacia). The reaction was stopped by adding 2.5 I11 0.5 M-EDTA pH 8.0 and diluted with 20 x SSC (Maniatis

T a b l e 1. Final results o f H C V L i P A typing and H C V antibody assays

HCV Isolate NS4* NS5* Core* LIA* EIAI" type

BR1 3 3 3 P ND 1 BR2 3 1 2 P ND 1 BR3 0 0 0 N ND 1 BR4 1 0 0 I ND 1 BR5 3 3 2 P ND 1 BR6 3 3 2 P ND 1 BR7 9 0 1 P ND 1 BR8 0 9 1 I ND 1 BR9 0 0 0 N NO lb BR10 0 0 0 N ND lb BR11 3 0 3 P ND lb BR12 1 1 0 P ND lb BR13 1 0 0 I ND lb BR14 1 0 0 I ND lb BR15 0 0 1 P ND lb BR16 1 0 0 I ND lb BR17 1 0 0 I ND lb BR18 0 0 3 P ND 2a BR19 3 3 2 P ND 3 BR20 9 0 1 I NO 3 BR21 2 3 2 P ND 3 BR22 0 0 0 N ND 3 BR23 3 3 3 P ND 1 b + 3

BR24 2 2 2 P 6.7 1 BR25 3 2 3 P 6.9 1 BR26 1 2 3 P 6.4 1 BR27 9 0 2 P 7-1 lb BR28 2 3 2 P 6.9 lb BR29 3 3 9 P 6.1 lb BR30 3 3 2 P 6.6 lb BR31 1 0 2 P 6.8 lb BR32 2 3 2 P 6.8 lb BR33 0 2 0 P 1.1 3 BR34 1 3 3 P 6.6 3 BR35 9 1 1 P 5.2 3 BR36 2 0 2 P 6-2 3 BR37 1 0 3 P 6.8 l b + 3 BR38 1 3 0 P 7.1 1 BR39 9 0 1 I 2.9 1 BR40 0 0 2 P 0.9 1 BR41 0 0 3 P 0.3 1 BR42 0 0 1 I 6.0 1 BR43 0 0 2 P 4.7 1 BR44 0 0 2 P 6.7 1 BR45 0 0 1 P 4.7 1 BR46 0 0 9 I 2.2 1 BR47 3 3 2 P 0.5 lb BR48 0 1 2 P 0.2 lb BR49 0 l 1 P 5.4 1 b BR50 1 0 1 P 7.4 lb BR51 0 0 3 P 7.4 lb BR52 2 3 2 P 4.7 lb BR53 0 0 1 P 7"9 lb BR54 2 3 2 P ND lb BR55 0 9 1 I 1"4 3 BR56 2 1 2 P 2"9 3 BR57 1 3 2 P 1"5 3 BR58 1 1 2 P 7"3 3 BR59 0 1 0 I 0.4 3 BR60 0 3 1 P 1"6 3 BR61 0 0 3 P 2'6 3

JP62 1 2 2 P ND 2a JP63 3 3 2 P ND l b + 2 a

Typing o f H C V isolates 1 0 9 5

T a b l e 1. (cont.)

HCV Isolate NS4* NS5* Core* LIA* EIAt type

BE64 3 3 3 P NO lb BE65 0 0 3 P ND lb BE66 1 0 2 P ND lb BE67 3 3 2 P ND lb BE68 2 1 3 P ND 2a BE69 0 0 0 N ND 3 BE82 3 3 2 P ND i BE83 3 3 2 P ND lb BE84 2 3 2 P ND lb

NE70 3 1 3 P NO 1 NE71 9 9 1 P ND 3 NE72 0 0 1 I No 3 NE73 0 0 0 N ND 3

BU74 0 0 1 P NO 4 BU75 0 0 1 P NO 4 BU76 0 0 1 P NO 4 BU77 0 0 2 P ND 4 BU78 0 0 1 P ND 4 BU79 0 1 0 I ND 4

GB80 0 0 0 N 0"4 4 GB81 0 0 0 N 3.4 2 + 4

* The Inno-LIA HCV Ab assay contains one line with NS4 epitopes, one line with NS5 epitopes and four lines with core epitopes. Only the highest score for the core lines is given. The intensity of the signal is given by a number : 0, negative; 9, indeterminate; 1 to 3, positive. The final interpretation of the antibody test is given in the LIA column: P, positive; N, negative; I, indeterminate.

t Signal to noise ratio in the Innotest HCV Ab. ND, Not done.

et al., 1982) until a final concentration of 6 x SSC and 2.5 pmol oligonucleotide/gl was reached. This solution was applied at 1 pmol over every 4 mm, on a nitrocellulose membrane. As control for the conjugate, biotinylated D N A was applied alongside. The oligo- nucleotides were fixed to the membrane by baking at 80 °C for 2 h. The membrane was then sliced into 4 m m strips.

LiPA test hybridization and colour development. Equal volumes (10 111 each) of the nested PCR amplification product containing incorporated Bio-11-dUTP and of 400 mN-NaOH/10 mM-EDTA were mixed and incubated at room temperature for 10 min. Then 1 ml prewarmed (to 37 °C) hybridization buffer, containing 3 M-tetramethylammonium chloride (TMAC1, Merck), 50 mg- sod ium phosphate buffer pH 6.8, 1 mM-EDTA, 5 x D e n h a r d t ' s solution (Maniatis et al., 1982), 0.6% (w/v) SDS and 100 gg /ml sheared sa lmon sperm DNA, was added and the hybridization was carried out in a shaking water bath at 42 °C for 2 h (Jacobs et al., 1988). The strips were washed twice at room temperature for 5 rain with 1 ml prewarmed (to 37 °C) wash buffer (3 M-TMAC1, 0"2% SDS, 50 mN-Tris HC1 pH 8.0), followed by a stringent wash at 51 °C for 30 min and two brief washing steps at room temperature. The wash buffer was then replaced by rinse solution (Inno-LIA kit) and the strips were rinsed twice with 1 ml at room temperature. Finally, the strips were rinsed with conjugate diluent ( Inno-LIA kit) and incubated with conjugate diluent containing 4000- fold diluted streptavidin labelled with alkaline phosphatase (Gibco BRL) for another 30 min at room temperature. The strips were washed again three times with rinse solution and once with substrate diluent ( Inno-LIA kit). Colour development was achieved by adding BCIP and

1096 L. S tuyver and others

Table 2. Nucleot ide sequence, posi t ion and orientation o f the pr imers and probes

Number Type Position Polarity Sequence from 5' to 3'* Reference

1 Universal -299 + CCCTGTGAGGAACTWCTGTCTTCACGC Kato et al. (1990) 2 Universal - 1 - GGTGCACGGTCTACGAGACCT Okamoto et al. (1991) 3 Universal - 264 + TCTAGCCATGGCGTTAGTRYGAGTGT This work 4 Universal - 29 - CACTCGCAAGCACCCTATCAGGCAGT This work 5 1 - 170 + AATTGCCAGGACGACC Kato et al. (1990) 6 1 - 117 - TCTCCAGGCATTGAGC Kato et al. (1990) 7 lb - 103 + CCGCGAGACTGCTAGC Kato et al. (1990) 8 2 -83 + TAGCGTTGGGTTGCGA Nakao et al. (1991) 9 2 -126 -- ATAGAGTGGGTTTATC Okamoto et al. (1991)

10 2a - 165 + CCGGGAAGACTGGGTC Okamoto et al. (1991) 11 2a -136 + ACCCACTCTATGCCCG Okamoto et al. (1991) 12 2b - 165 + CCGGAAAGACTGGGTC Okamoto et al. (1992a, b) 13 2b - 136 + ACCCACTCTATGTCCG Okamoto et al. (1992a, b) 14 3 - 170 + AATCGCTGGGGTGACC This work 15 3 - 117 - TTTCTGGGTATTGAGC This work 16 3 - 103 + CCGCGAGATCACTAGC This work 17 3 -146 + TCTTGGAGCAACCCGC Chan et al. (1992a, b) 18 3 - 146 + TCTTGGAACAACCCGC Chan et al. (1992a, b) 19 4 - 170 + AATYGCCGGGATGACC Bukh et al. (1992) 20 4 - 117 - TTTCCGGGCATTGAGC This work 21 Universal -115 + TTGGGCGYGCCCCCGC Kato et al. (1990) 22 Universal - 195 + TCTGCGGAACCGGTGA Kato et al. (1990)

* Some of the primers contain degeneracy at one or two nucleotide positions; W is A or T; R is A or G, and Y is C or T.

NBT to the substrate diluent and incubation of the strips at room temperature for 30 min. The colour development was stopped by replacing the buffer with distilled water. Samples were considered reactive in the LiPA when the intensity of the hybridization signal reached the intensity of the conjugate control line, which was always stronger than the corresponding signal on the negative control strip.

Results

Complemen tary po lymerase chain reaction

Serum R N A from HCV-infected pat ients was used as a

template for c D N A synthesis, the product of which in tu rn served as a template for nested PCR. Two sets of

P C R primers were designed, primers 1 and 2 for the outer reaction, primers 3 and 4 for the nested react ion

(Table 2). These four primers were chosen to match the published sequences (Kato et al., 1990; N a k a o et al.,

1991 ', O k a m o t o et al., 1991) and the sequence of a clone ob ta ined f rom the U R of isolate BR56 ((3. Maertens,

__~__1 __2_2 3 4 5 M 6 7 8 9 I0 M A B A B A B A B A B A B A B A B A B A B

Fig. 1. Ethidium bromide-stained agarose gel showing the length of the nested PCR fragments. Lane A of each pair shows the PCR fragment with incorporation of Bio-ll-dUTP. Lane B is the PCR fragment without Bio-ll-dUTP. Sera in lanes 1, BR28; 2, BR24; 3, BR29; 4, BR33; 5, BR36; 6 and 7, negative control sera; 8, JP62; 9, BR23; 10, cPCR control without template; M, molecular size markers.

unpubl i shed results). The result ing amplif icat ion product

of the nested P C R was 2 3 5 b p long. Owing to the incorpora t ion of Bio-11-dUTP there was a decrease in

mobi l i ty which was clearly visible after agarose gel electrophoresis (Fig. 1). The size of the D N A fragments was the same for all the different H CV types, suggesting

that a second experiment, such as restr ict ion enzyme

digestion or hybridizat ion, was necessary to allow classification. A m e m b r a n e strip method using immobi l - ized HCV genotype-specific ol igonucleot ide probes was

therefore developed.

A line probe assay f o r discrimination between types 1, 2

and 3

The sequences for the probes against type 3 were derived from a cPCR clone from serum BR56. W h e n compar ing the publ ished type 1 sequences with BR56, two regions of

16 nucleotides, f rom - 170 to - 155 and - 132 to - 117,

con ta in ing four or five muta t ions could be observed each time. W h e n type 2 sequences became available, var ia t ion was again found in these two regions. Therefore, the

pos i t ion of the characterizing probes was chosen in regions with relatively little similarity between types, bu t good conservat ion within one type. I n a first version of

the strips, eight separately immobi l ized oligonucleotides were applied. Two were directed against H C V type 1 (probes 5 and 6), four against type 2 (probes 10 and 11

for type 2a, probes 12 and 13 for type 2b) and two against type 3 (probes 14 and 15) (Table 2). cPCR

products were synthesized f rom 23 Brazil ian sera (BR1 to BR23) and 17 of them recognized the 16-mers of type

Typing of H C V isolates 1097

aP62

HCJ8 2b

-Z30 -220 -210 - ~ 0 -19~ -180 -~70 -160 150 type

~cvJ I b SR29 ~b BEg2 I b B E ~ l b SES~ Ib

Z6 4 8u79 4 8U74 4 6680 4

t y p e -Iz, o 130 - IZO -~10 1oo -90 ~0 -7o -6o - %

NCV1 3 ~ TCTTG GATC.AACCC T T]GGgCGIGCC£ CC~C AAGACIGCTAGC CGAG T A ~ G ~ T G O ~ I C 0 ~ A A A ~ C C ~ G ~ a l

.eva ab . . . . . . . . ZZII221ZII.~ZI ~ZZZZZZZZIZZ 55~ ~iSiilZiZiiSSi25 ii~iiiiiiiii BESZ l b . . . . . . . . . . . . . G - - - - C . . . . . . . . . . . . . . . . . . . . . . . . . . . . . BE84 l b ~

~CJ6 ~a . . . . . . . . A . - - - ~ A T C TC . . . . . . . . . . . . . . . . . . . . .

. . . . . . . . . . . . . . . . . . . . . . . . . . iiiiiiiiiiii iiii iiiii+iiiiii GB81 2 . . . . . . . A + . . . . . A . . . m . . . . . C..TC+ ~CJ8 Zb . . . . A . - - - - - ~ - - - J - - / r - ~ - ~ C

-"'° .... '3' ::::: ............. :: ~ : : : : : : : , , 1 , ~ , , iiiiiiiiiiii ~ o i i ~ ~i~iii!!!i!!! ii!iiii!iiiii

6 u 7 9 ~ . . . . . . . . . . . . . . . . . . . . . I i . . . . . . . . . . . . . . . ~ . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . ~uT~ 4 . . . . . ~ . . . . . . . . . . . . . . . . . . . . . . . . . . . .

Fig. 2. Alignment o f the 5' U R nucleotide sequences of four different types of HCV. Boxed nucleotides indicate the positions of probes used for typing of the four different groups. The underlined nucleotides are used for subtyping within each group. (.) between nucleotides - 138 and - 1 3 7 in most sequences corresponds to the insertion in some of the type 1 and 4 isolates. The numbering of the probes corresponds with that in Table 2.

1 by hybridization. Four type 3 sera were found, and one type 2a serum. Serum BR23 was co-infected with type 1 and type 3. Two pools of Japanese sera were subsequently tested: JP63 reacted with the type 1 and type 2a probes, and the majority of the JP62 pool contained type 2a sequences. After cPCR cloning and sequencing the region between the primers 3 and 4, the sequence of JP62 (Fig. 2) was confirmed as type 2a. The type 2b probes 12 and 13, to which JP62 did not react, differed by only one nucleotide each from the sequence of JP62. This means that the chosen hybridization and washing conditions were very stringent and that even single mismatches abolished hybridization in this assay.

Discrimination between subtypes

After careful comparison of all available type 1 coding sequences, two subtypes (la and lb) could be clearly distinguished, with an average genome identity of 79 %. Although only two mutations were observed in the 5' UR of HCV-1 (la) and HCV-J (lb), i.e. they were 98-8 % identical, the A to G transition observed at position - 99 occurred in all the type lb isolates so far studied. Therefore, hybridization to probe 7, which spans the region of the G substitution, is indicative of a type lb isolate. If no hybridization occurs with probe 7, the

isolate is probably a type la, but a variant of type lb can also be present. Therefore, no subtyping is given for such sera.

When comparing all available 5' UR sequences of type 3 (this work; Chan et al., 1992a, b; Bukh et al., 1992; Lee et al., 1992), the isolates could be divided into two groups according to the presence of a common G (probe 17) or a more rare A (probe 18) at position - 139. This division of type 3 isolates was done only to generate more information during characterization. These two groups do not correspond to the type V (3a) and VI (3b) of Mori et al. (1992). Discrimination between types 2a and 2b could be made in the variable regions as reported above.

The combination of all these type- and subtype- specific probes (Table 2) allowed us to separate the 17 Brazilian sera, which had been previously characterized as type 1, into eight type 1 and nine type lb sera. Material from three of the four type 3 sera formed hybrids with probe 17. Different molecules in the cPCR fragment of the co-infected serum BR23 hybridized with the lines for type lb and type 3 (Fig. 3, strip 12).

Another 38 Brazilian sera (BR24 to BR61) were also tested in the LiPA. The major criterion for the selection of these sera was the absence of antibodies for NS4 and NS5 epitopes, since earlier reports showed that there was a low degree of cross-reactivity of type 2 and type 3 anti- NS4 antibodies with type 1 NS4 antigens (Chan et al., 1991). Of the 38 Brazilian sera, 12 could be classified as type 1, 14 as type lb, 11 as type 3 (of which nine hybridized to probe 17) and one was a co-infection of types 1 b and 3. We concluded that all the Brazilian sera that were tested could be divided into types.

Identification of type 4 isolates and incorporation of type 4-specific probes in the line probe assay

PCR fragments amplified from six Burundian sera (BU74 to BU79) failed to react with the type-specific probes on the strip; only probe 7 was reactive with some of the sera. We cloned and sequenced PCR fragments from these samples and obtained sequences that were clearly different from most of the previously described types (BU74 and BU79; Fig. 2). The Burundian samples are related to each other, and to Z5 and Z6 (Bukh et al., 1992) and show greater similarity to type 1 than to type 3 or type 2. However, most of the differences from type 1 were again located within the variable regions. As reported by Bukh et al. (1992), an extra nucleotide was detected in BU74 and BG80, between positions -140 and - 139, but we prefer to locate this insertion between - 1 3 8 and -137 because this results in closer similarity to other isolates. These results argue in favour of the existence of new HCV type(s) or subtype(s), which we provisionally called type 4.

1098 L. Stuyver and others

1 2 3 4 5 6 7 8 9 10 11 12 13

i

:7::

~/,~---~ 2C1/22

//-7

, _ _ _ _ _ _ _ ~ 10/11

- - ~ 12/13

[ Z ~ 14

15

]6

17

18

19

2O

Fig. 3. HCV LiPA typing results for some exceptional and representative sera. The strip contains 16 parallel probe lines and (C) a control line for conjugate binding. The position of each probe is given by its number, which corresponds with the numbering in Table 2. The strips were hybridized with cPCR products of the following sera: strip 1, BR29; strip 2, BE82; strip 3, BE84; strip 4, BE83; strip 5, JP62; strip 6, BG81; strip 7, BR56; strip 8, BR36; strip 9, BR33; strips 10, BU79; strip 11, BG80; strip 12, B R23; strip 13, no RNA added to the cPCR reactions. Strip 14 was hybridized with biotin-labelled oligo(dA).

After obtaining these new data, the LiPA was improved in four ways. First, probes 21 and 22 were chosen from highly conserved regions for use as universal HCV probes for the confirmation of the presence of the PCR product (Table 2). Secondly, two oligonucleotides were synthesized for identification of the type 4 sequences (probes 19 and 20). Thirdly, two universal type 2 probes were selected (probes 8 and 9), since a type-specific reaction could not be obtained with the four type 2 16- mers from the variable regions. Fourthly, another type 3 specific probe was added (probe 16).

With this last version of the LiPA, the six PCR fragments from the Burundian sera were classified as type 4, as expected (Fig. 3, strip 10). Two Gabonese sera, four sera from The Netherlands and nine Belgian sera were also included in the evaluation. From GB80, a type 4 HCV 5' UR could be amplified; this was then cloned and sequenced (Fig. 2). The other Gabonese serum GB81 had a co-infection of a variant of type 2 (cloned and sequenced, Fig. 2) and type 4. The latter gave the same typing pattern as BU79. To establish whether the reaction of GB81 with the type 2 and type 4 probes was caused by unexpected cross-reactivity between typing probes, or was merely the result of a co-infection, the cPCR product was cloned and 17 individual colonies

were subjected to PCR and HCV LiPA. Type 4 inserts were found in 10 (59%) colonies and seven (41%) possessed type 2, indicating the co-existence of two types of HCV in the same serum. For the three NE sera that were negative (NE71 and NE72) or indeterminate (NE73) in the Ortho RIBA test and positive (NE71), indeterminate (NE72) or negative (NE73) in the Inno- LIA HCV Ab II test, we could show that type 3 isolates were present. The fourth NE serum (NE70), which demonstrated good reactivity in both the Ortho RIBA and the Inno-LIA, contained a type 1 isolate. Finally, from the nine Belgian sera we analysed, BE64 to BE67 were infected with type lb strains, one patient of Italian origin (BE68) had a type 2a infection and BE69 contained type 3 sequences. The latter was obtained from a case of chronic, viral-like NANB hepatitis, but negative in all second generation assays and anti-NS3, anti-E1 and anti-E2 research assays. This serum had a very low virus titre and became weakly positive only after the second round of PCR in four different samples taken over 2 years, showing the need for nested cPCR in HCV diagnosis.

Three Belgian serum samples were derived from a group of patients prior to interferon-~ treatment. All three samples belong to type 1, but PCR fragments were

Typing o f H C V isolates 1099

Table 3. Interpretation o f the results shown in Fig. 3

S e r u m

T y p e P r o b e B R 2 9 B E 8 2 B E 8 4 B E 8 3 J P 6 2 G B 8 1 B R 5 6 B R 3 6 B R 3 3 B U 7 9 G B 8 0 B R 2 3

C o N u g a t e c o n t r o l + + + + + + + + + + + +

C o n t r o l P C R 2 1 / 2 2 + + + + + + + + + + + +

1 5 + + - - + . . . . . . . +

1 6 + + + + . . . . . . . +

l b 7 + - + + . . . . . + + +

2 8 . . . . + + . . . . . .

2 9 . . . . + + . . . . . .

2 a 1 0 / 1 1 . . . . + + . . . . . .

2 b 1 2 / 1 3 . . . . . . . . . . . .

3 14 . . . . . . + + + - - +

3 15 . . . . . . + + + - - +

3 16 . . . . . . + - + - - +

3 17 - + . . . . + + - - - +

3 18 - - - + . . . . + - - -

4 19 . . . . . . . . . . + -

4 2 0 . . . . . . . . . + + -

C l a s s i f i c a t i o n l b 1 l b l b 2 a 2 a 3 3 3 4 4 l b + 3

cloned and sequenced because abnormal patterns on the LiPA appeared (Fig. 3, strips 2, 3 and 4). BE82 possesses a G at position -99 and fulfils one criterion for type lb, but the mutation at position - 9 4 negates this identi- fication. The reaction of BE82 with probe 17 could not be confirmed by sequencing. However, for BE83, which recognized type 3 probe 18, the sequence was confirmed. The interpretation of signals obtained with type 1 isolates on type 3 probes 17 and 18 is therefore dependent on sequence analysis.

In total, 18 different oligonucleotides were used for the final version of the LiPA strips, Table 3, as shown in Fig. 3. The interpretation of these strips is given in Table 3. A summary of the classification in relation to the serology is presented in Table 1.

Discussion

In order to study the natural variation of HCV isolates obtained from different geographical areas throughout the world, we developed a rapid method for the typing and subtyping of HCV isolates, in the form of an LiPA.

Essentially, a cPCR fragment is synthesized from the 5' UR of any HCV genome by using sets of primers targeting highly conserved regions. The oligonucleotides used for typing are directed against the internal type- specific variable parts of the cPCR fragment. In fact, the two variable regions between positions - 170 and - 155 and between - 132 and - 117 in the linear sequence may be part of a stem in the folded viral RNA (Brown et al., 1992), and mutations in the first region may be complemented by another mutation in the second region to allow or prevent RNA duplex formation. Variation

and conservation are expected to occur at the same positions in other new types of HCV as well and, therefore, this variable region might remain instrumental for discrimination between all the current and as yet undiscovered types of HCV. Moreover, since more variations compared to the 5' UR are observed in the core, NS3 and NS5 regions, typing in these regions employing universal sets of primers and genotype-specific probes might no longer be tenable.

Several attempts have been undertaken already to classify the different known genotypes of HCV (Mori et al., 1992; Cha et al., 1992; Chan et al., 1992a; Okamoto et al., 1992b). We propose a system of types and subtypes (Table 4). In this system, isolates belonging to different types show average nucleotide identity of about 68 %, those belonging to different subtypes of the same type are about 79 % the same and those belonging to the same subtype are more than 90 % identical. These degrees of similarity refer to the complete genome or to the region between positions 8261 and 8600 in NS5 according to Kato et al. (1990). Classifications proposed by others do not discriminate subtypes (Chan et al., 1992a) or do not recognize the closer relationship between groups I and II (about 79%) (Okamoto et al. 1992b; Cha et al., 1992; Mori et al., 1992) compared to that between group I and any other group (only about 68 %). Cha et al. (1992) classify all type 2a and 2b (our notation) isolates into group GIII without making any subdivision, as proposed by Okamoto et al. (1992b), Nakao et al. (1991) and this work, but they do propose that GI and GII (cor- responding to our subtypes la and lb) should be discriminated although equal genetic distances can be observed between la and lb, or 2a and 2b. Our system

1100 L. S tuyver and others

Table 4. Overview o f H C V nomenclature

This work

la lb 2a 2b 3 4

Chan et al. (1992a) 1 1 2 2 3 Okamoto et al. (1992a) I 11 III IV Mori et aL (1992) I II III IV V and VI Cha et al. (1992) GI GII Gill GIII GIV Nakao et al. (1991) Pt K1 K2a K2b K3 Prototype HCV- 1 HCV-J HC-J6 HC-J8

GV

will allow the classification of new groups of HCV genotypes together with their related subtypes. A new group that might, for instance, be related to subtypes la and lb could then be named lc, whereas such a group might be called group VII by Mori et al. or group GVI by Cha et al., not illustrating the relationship with their groups I and II.

In this study, 61 PCR-positive Brazilian HCV sera were typed. Sera with a type 1 HCV infection amounted to 20 (33 %) of the samples, 23 (38 %) were type lb, one (1-6%) was type 2a, 15 (25%) was type 2a, 15 (25%) were type 3 and two (3 %) sera with co-infections were found. The recognition of co-infected sera is illustrated by BR23 (Fig. 1, lane 9; Fig. 3, strip 12). The remaining 20 sera were collected from five different countries, with eight of the sera originating from two African countries. The sequences of the 5' UR of the virus that could be amplified from these African sera were strongly divergent from the previously described types. Therefore, we tentatively classify these isolates as type 4. Similar sequences communicated in the study of Bukh et al.

(1992) also originated from African sera, although one was from Denmark. We recently found type 4 isolates in Belgium as well. Fig. 2 shows that in the region between nucleotides -238 and -55 , as many as seven variations are possible. It is likely that type 4 is composed of several subtypes, or that these subtypes are divergent subtypes of type 1. It also remains to be determined whether a discrimination between the type 3 subtypes Ta and Tb (Mori et at., 1992) can be made in the 5' UR. More data about the serology, the sequences of the 5' UR and the open reading frames of type 3 and type 4 isolates are needed to allow subtyping.

Since isolate-specific mutations are scattered through- out the 5' UR, it is possible that an isolate of a given type or subtype could fail to hybridize with one of the typing probes (Fig. 3, strip 3) or the subtyping probes (Fig. 3, strip 2). The presence of at least two type-specific probes for each type partially overcomes this loss of information and such a serum sample can therefore be classified. Typing lb isolates as type 1 is possible with this version of the LiPA (as demonstrated by BE82) and can be

explained by the absence of a type la (related to HCV- 1)-specific probe. We are searching for ways to avoid such shortcomings. Owing to mutations, a given isolate can also hybridize to a subtyping probe for another type. Such reactivities merely indicate the presence of the sequence of the subtyping probe in the isolate studied (Fig. 3, strip 4). In a few cases, cross-hybridization of sequences of types 1 and 4 with probes 17 and 18 was observed, although the sequences of these isolates were not homologous and the hybridization was carried out in TMAC1 (Fig. 3, strip 2). We never observed reactivities with multiple typing probes, unless a serum was co- infected as investigated for GB81.

In general, when a type 1 cPCR product hybridizes in the LiPA, the sequences of the probes 5, 6, 21 and 22 must be present in the nested cPCR fragment. Conse- quently, 48 of 184 bp (26%) (Fig. 2) are immediately known. For the same reasons 26 to 41% of the sequence is known for other isolates but, owing to the degeneracy of some of the 16-mers, these percentages are maximum scores. Nevertheless, this approach supports the idea of the sequencing by hybridization principle (Strezoska et

al., 1991). When comparing LiPA results with antibody reactivity

of these sera in our Inno-LIA HCV Ab II assay (Table 1) some correlations between genotypes and their pheno- types (serotype) emerged. Types 3 and 4 sera from Belgium, The Netherlands, Gabon and Burundi all reacted as very weakly positive, indeterminate, or negative in the current second generation antilzody assays. The weakly positive reaction is mostly caused by anti-core antibodies, whereas antibodies against the LIA NS4 and NS5 epitopes are usually absent. This is consistent with the conservation of core sequences encoding only slightly different epitopes allowing immunological cross-reaction. Epitopes for the NS4 and NS5 region are located in highly variable regions, disabling most cross-reaction. Because current antibody assays involve type 1 epitopes, it is possible that a small proportion of type 2-, 3- or 4-infected sera will give a negative result. From the 14 randomly chosen sera (BR24 to BR37), the nine type 1-infected sera were

Typing of HCV isolates 1101

positive in the LIA. From four type 3 sera, two (BR34 and BR36) reacted with NS4 and three (BR33, BR34 and BR35) with NS5. BR37 was not taken into account because of the co-infection. When the serological data for the 80 sera infected by a single type were analysed, 58 % and 44 % of the type 1 sera recognized the NS4 and NS5 epitopes, respectively. These percentages are rather low and due to the selection criteria. For the type 3 sera, 37% and 53% were reactive with the NS4 and NS5 epitopes, respectively. It is possible that higher cross- reactivities are observed in high-risk groups, such as in those samples obtained from Brazil. Such cross-reacting sera could be induced by multiple infections, some of which occur simultaneously, but others might occur one after another. A previous anti-HCV memory could be boosted by new HCV infections and result in co- circulation of viruses of one type with antibodies mainly directed against another type. Such an explanation is plausible for serum BR56, which has been typed as HCV type 3 but contained antibodies to type 1 core, El, E2, NS3, and NS5 (data not shown). It remains to be determined whether anti-type 3 antibodies are present in this serum. In addition to the differences in immune response, each different HCV type could also show a different progression to long-term liver disease, as has already been reported (Okamoto et at., 1992a).

In conclusion, the LiPA permits rapid determination of the type of HCV infection. The assay has the ability to discriminate between four different HCV types and four subtypes, and might aid the detection of new types of HCV. Moreover, the LiPA can be further improved by replacing the cPCR reactions by the RNA-capture PCR (Van Doom et al., 1992). Finally, this assay could prove instrumental in further elucidating the relationship between genotypes, serotypes, the clinical status or outcome of the disease, and the response to treatment. We are currently interested in evaluating such relation- ships.

The authors would like to acknowledge Eric Delaporte, Guy De Groote, Geert Leroux-Roels, Dirk Pollet, Caroline Rouzere and Vestine Ntakarutimana for the collection of sera. We also thank Kris De Keyser and Danielle Peeters for the LIAs, Wouter Van Arnhem and Thierry Scarcez for nucleotide sequencing, Conny Van Loon for oligonucleotide synthesis, Gonda Verpooten for the production of strips and Fred Shapiro for carefully editing the manuscript. This work was supported in part by the Eureka project EU 680 (hepatitis C).

References

BROWN, E.A., ZHANG, H., PING, L.H. & LEMON, S.M. (1992). Secondary structure of the 5' nontranslated regions of hepatitis C virus and pestivirus genomic RNAs. Nucleic Acids Research 20, 5041-5045.

BuKn, J., PURCELL, R. H. & MILLER, R. H. (1992). Sequence analysis of the 5' noncoding region of hepatitis C virus. Proceedings of the National Academy of Sciences, U.S.A. 89, 494~4946.

CHA, T.-A., BEAL, E., IRVINE, B., KOLBERG, J., CHIEN, D., KUO, G. &

URDEA, M. S. (1992). At least five related, but distinct, hepatitis C viral genotypes exist. Proceedings of the National Academy of Sciences, U.S.A. 89, 7144-7148.

CHAN, S.-W., SIMMONDS, P., MCOMISH, F., YAP, P.-L., MITCheLL, R., DOW, B. & FOLLETT, E. (1991). Serological reactivity of blood donors infected with three different types of hepatitis C virus. Lancet 338, 1391.

CHAN, S.-W., McOMISH, F , HOLMES, E- C., DOW, B., PEUTHERER, J. P., FOLLETT, E., YAP, P. L. & SIMMONDS, P. (1992a). Analysis of a new hepatitis C virus type and its phylogenetic relationship to existing variants. Journal of General Virology 73, 1131 1141.

CHAN, S.-W., HOLMES, E. C., McOMISH, F., FOLLETT, E., YAP, P. L. & SIMMONDS, P. (1992b). Phylogenetic analysis of a new, highly divergent HCV type (type 3): effect of sequence variability on serological responses to infection. Hepatitis C Virus and Related Viruses, Molecular Virology and Pathogenesis, First Annual Meeting, Venezia, Italy. Abstract D5, 73.

CHOMCZYNSKI, P. & SACCHI, N. (1987). Single step method of RNA isolation by acid guanidinium thiocyanate-phenol-chloroform ex- traction. Analytical Biochemistry 162, 156-159.

CHOO, Q.-L., RICHMAN, K. H., HAN, J. H., BERGER, K., LEE, C., DONG, C., GALLEGOS, C., COLT, D., MEDINA-SELBY, A., BARR, P. J., WEINER, A. J., BRADLEY, D. W., KUO, G. & HOUGHTON, M. (1991). Genetic organization and diversity of the hepatitis C virus. Proceedings of the National Academy of Sciences, U.S.A. 88, 2451-2455.

ENOMOTO, N., TAKADA, A., NAKAO, T. & DATE, T. (1990). There are two types of hepatitis C virus in Japan. Biochemical and Biophysical Research Communications 170, 1021-1025.

INCHAUSPE, G., Z~BEDEE, S., LEE, D.-H., SUGITANI, M., NASO•F, M. & PRINCE, A. M. (1991). Genomic structure of the human prototype strain H of hepatitis C virus: comparison with American and Japanese isolates. Proceedings of the National Academy of Sciences, U.S.A. 88, 10292-10296.

JACOBS, K.E., RUDERSBORF, R., NELL, S.D., DOUGHERTY, J.P., BROWN, E.L. & FRITSCH, E.F. (1988). The thermal stability of oligonucleofide duplexes is sequence independent in tetraalkyl- ammonium salt solutions: application to identifying recombinant DNA clones. Nucleic Acids Research 16, 4637-4650.

KANAI, K., KAKO, M. & OKAMOTO, H. (1992). HCV genotypes in chronic hepatitis C and response to interferon. Lancet 339, 1543.

KATO, N., HIJIKATA, M., OOTSUYAMA, Y., NAKAGAWA, M., OHKOSHI, S., SUGIMURA, T. & SHIMTOHNO, K. (1990). Molecular cloning of the human hepatitis C virus genome from Japanese patients with non-A, non-B hepatitis. Proceedings of the National Academy of Sciences, U.S.A. 87, 9524--9528.

KUBO, Y., TAKEUCHI, K., BOONMAR, S., KATAYAMA, T., CHOO, Q.-L., KuO, G., WEINrR, A. J., BRADLEY, D. W., HOUGHTON, M., SA1TO, I. & MIYAMURA, T. (1989). A cDNA fragment of hepatitis C virus isolated from an implicated donor of post-transfusion non-A, non-B hepatitis in Japan. Nucleic Acids Research 17, 10368-10372.

LEE, C.-H., CHENG, C., WANG, J. & LUMENG, L. (1992). Identification of hepatitis C viruses with a nonconserved sequence of the 5" untranslated region. Journal of Clinical Microbiology 30, 1602-1604.

MANIATIS, T., FRITSCH, E.F. & SAMBROOK, J. (1982). Molecular Cloning." A Laboratory Manual. New York: Cold Spring Harbor Laboratory.

MORt, S., KATO, N., YAGYU, A., TANAKA, T., IKEDA, Y., PETCttCLAI, B., CHIEWSILP, P., KURIMURA, T. & SHIMOTOHNO, K. (1992). A new type of hepatitis C virus in patients in Thailand. Biochemical and Biophysical Research Communications 183, 334-342.

NAKAO, T., ENOMOTO, N., TAKADA, N., TAKADA, A. &. DATE, T. (1991). Typing of hepatitis C virus genomes by restriction fragment length polymorphism Journal of General Virology 72, 2105-2112.

OKAMOTO, I~., OKADA, S.-I., SUGIYAMA, Y., YOTSUMOTO, S., TANAKA, T., YOSHIZAWA, H., TSUDA, F., M1YAKAWA, Y. & MAYUMI, M. (1990). The 5' terminal sequence of the hepatitis C virus genome. Japanese Journal of Experimental Medicine 60, 167-177.

OKAMOTO, H., OKADA, S.-I., SUGIYAMA, Y., KURAI, K., hZUKA, H., MACHIDA, A., MIYAKAWA, Y. & MAYUMI, M. (1991). Nucleotide sequence of the genomic RNA of hepatitis C virus isolated from a human carrier: comparison with reported isolates for conserved and divergent regions. Journal of General Virology 72, 2697-2704.

1102 L. Stuyver and others

OKAMOTO, H., SUGIUAMA, Y., OKADA, S.-I., KURAI, K., AKAHANE, Y., SUGAI, Y., TANAKA, T., SATO, K., TSUDA, F., MIYAKAWA, Y. & MAYUMI, M. (1992a). Typing hepatitis C virus by polymerase chain reaction with type-specific primers: application to clinical surveys and tracing infectious sources. Journal of General Virology 73, 673-679.

OKAMOTO, H., KURAI, K., OKADA, S.-I., YAMAMOTO, K., IIZUKA, H., TANAKA, T., FUKUDA, S, TStrDA, F. & MISI-nRO, S. (1992b). Full- length sequences of a hepatitis C virus genome having poor homology to reported isolates: comparative study of four distinct genotypes. Virology 188, 331-341.

POZATTO, G., MORETTI, M., FRANZIN, F., CROC~, L. S., TIRIBELLI, C., MASAYU, T., KANEKO, S., UNOURA, M. & KOBAYASHI, K. (1991). Severity of liver disease with different hepatitis C viral clones. Lancet 338, 509.

STREZOSKA, Z., PAUNESKU, T., RADOSAVLJEVIC, D., LABAT, I., DRMANAC, R. & CRKWNJAKOV, R. (1991). DNA sequencing by hybridization: 100 bases read by a non-gel-based method. Pro-

ceedings of the National Academy of Sciences, U.S.A. 88, 1008% 10093.

TAKAMIZAWA, A., MORI, C., FUKE, I., MANABE, S., MURAKAMI, S., FUJITA, J., ONISHI, E., ANDOH, T., YOSHIDA, I. & OKAYAMA, H. (1991). Structure and organization of the hepatitis C virus genome isolated from human carriers. Journal of Virology 65, 1105-1113.

YOSHIOKA, K., KAKUMU, S., WAKITA, T., ISHIKAWA, T., ITOH, Y., TAKAYANAGI, M. HIGASHI, Y., SHIBATA, M. & MORISHIMA, T. (1992). Detection of hepatitis C virus by polymerase chain reaction and response to interferon-0~ therapy: relationship to genotypes of hepatitis C virus. Hepatology 16, 293-299.

VAN DOORN, L.-J., VAN BELKUM, A., MAERTENS, G., QUINT, W., KOS, T. & SCr~LLEr,~NS, H. (1992). Hepatitis C virus antibody detection by a line immunoassay and (near) full length genomic RNA detection by a new RNA-capture polymerase chain reaction. Journal of Medical Virology 38, 289-304.

(Received 19 October 1992; Accepted 31 December 1992)