Comparison of Molecular and Conventional Methods for Typing of Enteroviral Isolates

Abstract

Twenty-eight enteroviral isolates obtained from various clinical specimens were typed by Lim-Benyesh-Melnick (LBM) pool-based neutralization, PCR-restriction fragment length polymorphism (RFLP), and partial sequencing of the VP1 region of the enteroviral genome. Sequencing was found to be a good alternative to LBM typing, while PCR-RFLP was inadequate for identification of enteroviral isolates.

Identification of enterovirus (EV) serotypes is useful to differentiate polioviruses from nonpolio enteroviruses, to study the association of EV types with certain diseases, to identify newer EVs, and for epidemiological surveillance (8). The conventional method for typing of enteroviruses is neutralization with type-specific sera (8) or with an intersecting pool of hyperimmune equine sera, the Lim-Benyesh-Melnick (LBM) pool (5, 6). These methods, in addition to being tedious, time-consuming, and expensive, may fail to resolve the serotype when there is aggregation of EV strains, antigenic variations, or multiple serotypes of EV in the specimen (8). These disadvantages associated with conventional methods have led to the development of molecular methods like PCR-restriction fragment length polymorphism (PCR-RFLP) (3, 4) and sequencing of the EV genome for typing of EVs (9, 10, 11). This paper reports the results of a comparison of the LBM pool with PCR-RFLP or partial sequencing of the VP1 region of EV genome for typing of EV isolates.

(This research forms part of the Ph.D. thesis work of D. J. Manayani.)

Twenty-eight EV isolates tested were obtained from cerebrospinal fluid, brain biopsy, throat, and rectal swab specimens in primary monkey kidney (PMK) or rhabdomyosarcoma (RD) cell cultures between September 1998 and August 2000 from patients attending the Christian Medical College and Hospital, Vellore, India. Stock strains of poliovirus serotypes 1, 2, and 3 and coxsackievirus B (CB) serotypes 3, CB5 (Centers for Disease Control and Prevention, Atlanta, Ga.), and CB4 (courtesy Gagandeep Kang, Wellcome Research Laboratory, Vellore, India) were also tested. The LBM pools (A to H) (Staten Serum Institut, Copenhagen, Denmark) were used as recommended by the supplier.

The 304-bp nested PCR product generated using universal EV primers from the 5′ noncoding region (5′ NCR) (1, 2, 7) of EV isolates was subjected to restriction enzyme (RE) analysis with three enzymes, BglI, StyI, and Asp700 (XmnI) (Boehringer Mannheim Roche, Mannheim, Germany) (4).

For sequencing, the RNA was extracted from the infected-cell culture supernatant of the RD cell line by using the QIAamp viral RNA kit (Qiagen Gmbh, Hilden, Germany). The primer pairs 012 and 011 or 040 and 011 generated amplicons of approximately 450 bp that span the VP1-2A region of the EV genome (10). Following cycle sequencing with fluorescent dideoxy-chain terminators, the products were sequenced with an automated genetic analyzer (ABI 310 analyzer; PE Applied Biosystems, Foster City, Calif.). Sequences (GenBank) that match the VP1 sequence of the isolates were identified using BLAST. The percent identity score of the VP1 sequence of EV isolates with the GenBank EV sequences that showed maximum identity, as well as the next closest identity, was calculated using ClustalW (12). If the pairwise identity scores of VP1 sequences of the test strain were 75% or more with respect to any particular EV GenBank sequence, then it was identified as the homologous serotype (11).

The results of typing of the 28 EV isolates are shown in Table 1. Of 27 isolates tested, 26 (96%) were serotyped by the LBM pool-based neutralization. Primers from the VP1-2A region amplified all 28 EV isolates, and 27 (96%) of them could be assigned a serotype by sequencing. The pairwise identity score for EV isolates from this study with the respective homologous GenBank sequences ranged from 78 to 89%; with the exception of one isolate, the pairwise identity scores to the closest heterologous sequences were in the range of 50 to 68%. Excluding the one untypeable isolate, identical results were obtained both by LBM pool typing and by partial sequencing of VP1 for 22 (85%) of 26 tested isolates.

TABLE 1.

Typing of EV isolates by molecular and serological methods

Isolatea	Typing results of EV isolatesb by:
	Serotyping	Partial sequencing of VP1	RE pattern
98/D1136 RS	E17	E17	E4/E14/CA2/CA3
98/D1222 CSF	E9	E4	Untypeable
98/D1222 TS	E4	E4	Untypeable
98/D1222 RS	E4	E4	Untypeable
99/D287 TS	E24	E6	Untypeable
99/D362 CSF	E14	E14	CA6
99/D412 TS	CA9	CA9	CB2/E6/PV1
99/D500 TS	E24	E6	Untypeable
99/D518 RS	E6	E6	Untypeable
99/D542 RS	ND	E25	Untypeable
99/D692 RS	E6	E6	Untypeable
99/D696 TS	CA9	CA9	CB2/E6/PV1
99/D700 RS	CA9	CA9	CB2/E6/PV1
99/D791 RS	Untypeable	Untypeable	Untypeable
99/D817 RS	E7	E7	CB2/E6/PV1
99/D978 CSF	E14	E14	CA6
99/D978 RS	CA9	E14	CA6
99/D1060 TS	E6	E6	Untypeable
99/D1121 RS	E9	E9	CB2/E6/PV1
99/D1125 RS	CA9	CA9	CB2/E6/PV1
99/D1128 TS	CA9	CA9	CB2/E6/PV1
99/D1128 RS	CA9	CA9	CB2/E6/PV1
00/D33 RS	E2	E2	CA6
00/D274 BB	CA9	CA9	E4/E14/CA2/CA3
00/D283 TS	CA9	CA9	CB2/E6/PV1
00/D283 RS	CA9	CA9	CB2/E6/PV1
00/D324 TS	CA9	CA9	CB2/E6/PV1
00/D1001 RS	E9	E9	Untypeable
PV1*c	PV1	PV1	PV1
CB3*c	CB3	CB3	N
PV2	ND	ND	N
PV3	ND	ND	PV3
CB4	ND	ND	N
CB5	ND	ND	CB5

^aIsolates were designated in accordance with the year of specimen collection, the laboratory in which the specimens were processed, the laboratory specimen identification number, and the type of specimen. Isolates from different sites of the same patient have the same laboratory specimen identification number. RS, rectal swab; CSF, cerebrospinal fluid; TS, throat swab; BB, brain biopsy.

^bE, echovirus serotype; CA, coxsackievirus A serotype; PV, poliovirus serotype; CB, coxsackievirus B serotype; ND, not done; N, does not fit with the digestion pattern suggested by Kuan (4).

^c*, stock strains.

Of the 28 clinical isolates, 11 were untypeable by PCR-RFLP; in the remaining 17, the assigned serotype or type complex was different from the serotype assigned by the other two methods (Table 1). The term “type complex” refers to an RE digestion pattern indicative not of a single serotype but of two or more serotypes. Isolates from different sites of the same individual had the same RE digestion pattern.

This study showed that there was a high degree of agreement between conventional serotyping and typing by sequencing. We were able to assign a serotype for 26 of 27 clinical isolates by sequencing, and in 22 of them the serotype assignment was identical to that obtained by conventional typing. Discordant typing results were obtained for four isolates. This discordance between serotyping by LBM pool and sequencing could occur due to antigenic variations or viral aggregation (11). Typing of isolates with discordant results by using monospecific sera has shown sequencing to be the accurate method of typing (11). In the study reported here, typing with monospecific sera was not carried out. The isolate that could not be typed by sequencing was also not typeable by conventional methods and may represent a unique type. Serial subcultures to achieve a sufficient concentration of virus for conventional typing delayed the typing results. On the other hand, typing data were available following PCR and sequencing within a few days. Thus, PCR and subsequent sequencing offers the dual advantage of rapidity and the ability to detect previously undescribed serotypes.

This study showed greater agreement between PCR sequencing and conventional typing than did that of Oberste et al. (11). Also, we were able to obtain sequencing data with the use of primers 011 and 012, compared to the five primer pairs used by Oberste et al. (11). These differences are most probably because fewer serotypes were represented in our set of clinical samples than were present in the other study (11). Since regional laboratories may have to deal with fewer serotypes of EV, it is possible that they could obtain typing data for most isolates by using fewer primer pairs for sequencing.

In contrast to the results with PCR and sequencing, the results of typing by PCR-RFLP were disappointing. We used a different primer pair in the nested round for PCR-RFLP than was used by Kuan (4). However, our primer pairs spanned the same 5′ NCR, generating a slightly larger PCR product (304 versus 297 bp) (4). Also, half of the prototype strains did not match the specific patterns described by Kuan (4). Variations may occur in the 5′ NCR of EVs (13), and this possibly will occur even in prototype strains due to repeated subculturing, which may explain the differences in findings. Despite the lack of correlation in type assignment between PCR-RFLP and conventional typing, this method could still be used for outbreak investigations, where comparison of isolates and differentiation between outbreak strains and nonoutbreak strains is the primary objective.