The N-terminal part of VP1 was sequenced for 43 enterovirus isolates that could not initially be neutralized with LBM pools or in-house antisera. Most isolates were found to belong to human enterovirus type A (HEV-A) and HEV-B (18 isolates of each). All HEV-A isolates could be typed by sequencing, with CV (coxsackievirus)-A16 and EV (enterovirus)-71 being dominant (nine and seven isolates, respectively). These types thus seem to have diverged more from their prototypes than the other types. Among the HEV-B isolates, E-18 dominated with five isolates that became typable after filtration. The virus type obtained by molecular typing was verified for 28 of the other patient isolates by neutralization using high-titre monovalent antisera or LBM pools. Twenty-two of the other 30 'untypable' isolates had substitutions in the VP1 protein within or close to the BC loop. Two closely related HEV-B isolates diverged by 19?4 % from E-15, the most similar prototype. Two non-neutralizable HEV-C isolates split off from the CV-A13/CV-A18 branch, from which they diverged by 15?7-18?2 %. Three of the six non-neutralizable isolates, W553-130/99, W543-122/99 and W137-126/99, diverged by >24?2 % from the most similar prototype in the compared region. The complete VP1 was therefore sequenced and found to diverge by >29 % from all prototypes and by >28 % from each other. Strains similar to W553-130/99 that have been identified in the USA are tentatively designated EV-74. The two other isolates fulfil the molecular criterion for being new types. Since strains designated EV-75 and EV-76 have been identified in the USA, we have proposed the tentative designations EV-77 and EV-78 for these two new members of HEV-B.

We used the polymerase chain reaction to study the presence of DNA from cytomegalovirus (CMV) and human herpesvirus (HHV)-6, HHV-7, and HHV-8 in saliva from 44 human immunodeficiency virus (HIV) type 1 -infected patients at different stages of disease and in 15 healthy HIVseronegative controls. CMV DNA, HHV-6 DNA, and HHV-7 DNA were found in all groups, but HHV-8 DNA was found only in symptomatic HIV-1-infected patients (5 [17%] of 29). One of the patients with HHV-8 DNA in saliva had oral Kaposi's sarcoma at the time of sampling, and another later developed the tumor. CMV DNA was found most often in the patients with AIDS. HHV-6 shedding tended to be less frequent among HIV-1-infected patients than among healthy controls. HHV-7 DNA was detected least frequently in the group of patients with AIDS. The presence of viral DNA was not correlated either with antiherpesvirus drug therapy or with oral symptoms, apart from Kaposi's sarcoma.

Molecular typing of enteroviruses should ideally focus on regions encoding determinants for neutralization. Mapping of monoclonal neutralizing antibodies has shown the VPI protein, in particular its aminoterminal part, encompassing the B-C loop, to be one major antigenic region. We therefore sequenced 570 nucleotides from the 5'-end of the VP1 region of the genome for all 28 echovirus prototypes, and for 61 clinical isolates representing all different echovirus types. An analysis of 133 sequences, including 39 sequences retrieved from GenBank, classified all echoviruses in enterovirus group B confirming results from sequencing within the VP2 region. The nucleotide and amino acid divergence of VP1 sequences of homotypic strains varied from 7.5-23.0% and from 0.0-5.3%, respectively, when compared to their corresponding prototypes, whereas strains belonging to different serotypes these divergences were 22.1-38.9 % and 4.9-16.4 %, respectively. Despite these minimal overlaps, the VP1 sequence was always more similar to that of the homotypic prototype than to that of any heterotypic strain. For 13 out of 14 echovirus types, where multiple isolates were available, the corresponding VP1 sequences diverged more from those of the prototype than from the other homotypic sequences as a reflection of genetic drift. Because there was a complete concordance between the sequences of the region encoding the VP1 aminoterminus and the serotype (P< 0.00001) sequence analysis of this region might complement typing by neutralization, and classify correctly echovirus isolates that may not be typed conveniently by the antisera in hand.

Molecular typing of enteroviruses should ideally focus on regions encoding determinants for neutralization. Mapping of monoclonal neutralizing antibodies has shown the VPI protein, in particular its aminoterminal part, encompassing the B-C loop, to be one major antigenic region. We therefore sequenced 570 nucleotides from the 5'-end of the VP1 region of the genome for all 28 echovirus prototypes, and for 61 clinical isolates representing all different echovirus types. An analysis of 133 sequences, including 39 sequences retrieved from GenBank, classified all echoviruses in enterovirus group B confirming results from sequencing within the VP2 region. The nucleotide and amino acid divergence of VP1 sequences of homotypic strains varied from 7.5-23.0% and from 0.0-5.3%, respectively, when compared to their corresponding prototypes, whereas strains belonging to different serotypes these divergences were 22.1-38.9 % and 4.9-16.4 %, respectively. Despite these minimal overlaps, the VP1 sequence was always more similar to that of the homotypic prototype than to that of any heterotypic strain. For 13 out of 14 echovirus types, where multiple isolates were available, the corresponding VP1 sequences diverged more from those of the prototype than from the other homotypic sequences as a reflection of genetic drift. Because there was a complete concordance between the sequences of the region encoding the VP1 aminoterminus and the serotype (P< 0.00001) sequence analysis of this region might complement typing by neutralization, and classify correctly echovirus isolates that may not be typed conveniently by the antisera in hand.

Phylogenetic analysis within the VP1 region now enables molecular typing of enteroviruses consistent with neutralization results. Three untypable isolates, 2776/82, 57/99 and 22/00, from Korea, North India and Bangladesh, respectively, showed within this region 98.0-99.0% amino acid identities. These were less than 77% to the previous enterovirus prototypes, but 91.5-92.5% to CA55-1988, the recently identified enterovirus 73 (EV73) prototype from California. All three strains were, however, most similar to CA64-4454, an EV73 prime strain, to which they shared 96.5-98.5% identity. Seven compared EV73 strains formed two clusters in the VP1 dendrogram, one cluster with strains from South and East Asia and CA64-4454, and the other with strains from Oman and California including the prototype. When sequencing the complete open reading frame of 2776/82, its non-structural region was found to be divergent from all human enterovirus B (HEV-B) strains, including CA55-1988, indicating that one or other strain was recombinant. Boot scanning of the genomes showed a recombination point within the P2 region. Therefore, part of this was sequenced for 57/99 and 22/00 and was found similar to 2776/82, while CA55-1988 was similar to coxsackievirus B3, demonstrating that CA55-1988 was the recombinant. Since all strains of EV73 isolated so far outside California originate from Asia, where it has a broad geographical distribution, it seems that EV73 may have been introduced to California from Asia. Further analysis of EV73 strains will reveal if the recombination occurred in the USA or in Asia and will help to elucidate the origin of this virus.