Detection of Mammalian Reovirus RNA by Using Reverse Transcription-PCR: Sequence Diversity within the λ3-Encoding L1 Gene

Abstract

Reoviruses infect virtually all mammalian species, and infection of humans is associated with mild gastrointestinal or upper respiratory illnesses. To improve reovirus detection strategies, we developed a reverse transcription-PCR technique to amplify a fragment of the reovirus L1 gene segment. This assay was capable of detecting 44 of 44 reovirus field isolate strains and was sufficiently sensitive to detect nearly a single viral particle (1.16 ± 0.13) per PCR of prototype strain type 3 Dearing. Pairwise comparisons of the 44 partial L1 gene sequences revealed that nucleotide variability ranged from 0 to 24.7%, with most of the nucleotide polymorphism occurring at synonymous positions. Phylogenetic trees generated from amplified L1 gene sequences suggest that multiple alleles of the L1 gene cocirculate in nature and that genetic diversity of the L1 gene is largely independent of the host species, geographic locale, or date of isolation. Phylogenetic trees constructed from the L1 gene sequences are distinct from those constructed from the four reovirus S-class gene segments, which supports the hypothesis that reovirus gene segments reassort in nature. This study establishes a new sensitive and specific technique for the identification of mammalian reoviruses and enhances our understanding of reovirus evolution.

Mammalian reoviruses are nonenveloped viruses that contain a genome of 10 double-stranded RNA segments (reviewed in reference 24). The double-stranded RNA molecules are classified by size into three large (L), three medium (M), and four small (S) gene segments. There are three reovirus serotypes, which can be differentiated by the capacity of anti-reovirus sera to neutralize viral infectivity and inhibit hemagglutination (29, 33). Reoviruses have a wide geographic distribution, and virtually all mammals, including humans, serve as hosts for infection (reviewed in reference 41). Reovirus infections are usually short-lived, with clearance occurring within 1 to 2 weeks under experimental virus challenge conditions (31).

Epidemiologic surveys indicate that by the end of childhood, most humans have antibodies against each of the three reovirus serotypes (14, 17, 18). In a study conducted at Boston Children's Hospital, the incidence of anti-reovirus antibodies increased from less than 25% in those less than 1 year of age to greater than 70% in those greater than 3 years of age (18). Reovirus infections do not appear to have a seasonal distribution and occur throughout the year. Reoviruses are commonly found in environmental water sources (1, 5, 28, 35), and it has been suggested that human fecal contamination is a major source of reovirus in water supplies (19).

Because of their broad host range, reoviruses are relatively easy to cultivate in the laboratory (reviewed in reference 41). Primary reovirus isolates can be propagated in a variety of commonly available cell types, but monkey kidney cells are generally considered the most permissive (30, 36). Reovirus infection is suspected on the basis of cytopathic effect, although confirmation by either immunocytochemistry or immunofluorescence is necessary (34, 36, 40). However, techniques for virus cultivation can be insensitive, which likely leads to an inability to detect reovirus in certain environmental or clinical situations. For this reason, reovirus contamination of cell culture involving media supplemented with animal-derived products is often overlooked. Diagnostic strategies based on detection of reovirus RNA have been described (6, 25, 38, 42); however, a rigorous assessment of the sensitivity and specificity of these techniques has not been reported.

In this study, we developed a sensitive and specific reverse transcription (RT)-PCR technique to detect reovirus RNA. We chose the viral L1 gene segment as the target for PCR amplification because it encodes a component of the viral RNA-dependent RNA polymerase (37) and is therefore likely to be conserved among reovirus strains. We show that the L1-based RT-PCR technique is sufficiently sensitive to amplify the L1 gene segment sequences from 44 independent reovirus strains and is robust enough to detect field isolate strains of reovirus. This work establishes a new method for the detection of mammalian reoviruses that should improve the performance of currently available reovirus diagnostic strategies.

MATERIALS AND METHODS

Cells and viruses.

Murine L929 cells were grown in either suspension or monolayer cultures in Joklik's modified Eagle's minimal essential medium (Irvine Scientific, Santa Ana, Calif.) supplemented to contain 5% fetal bovine serum (Intergen, Purchase, N.Y.), 2 mM l-glutamine, 100 U of penicillin per ml, 100 μg of streptomycin per ml, and 0.25 μg of amphotericin per ml (Irvine). Human neonatal kidney (HNK) cells were grown in monolayer culture in E-199 medium (Abbott Laboratories, North Chicago, Ill.) supplemented to contain 10% fetal bovine serum (Life Technologies, Grand Island, N.Y.). The reovirus strains used in this study are shown in Table 1. Reovirus strains type 1 Lang (T1L/53), type 2 Jones (T2J/55), and type 3 Dearing (T3D/55) are laboratory stocks available from the American Type Culture Collection (Manassas, Va.). Reovirus strains 234/99, 469/99, 491/99, 514/99, 758/99, and 780/99 are from the Abbott Laboratories culture collection. Virus stocks were prepared by using L929 cells as previously described (43).

TABLE 1.

Reovirus strains used for RT-PCR assay development

Virus straina	Abbreviation	GenBank accession no.	Reference(s)
T1/human/Ohio/Lang/1953	T1L/53	M24734	26, 33
T2/human/Ohio/Jones/1955	T2J/55	M31057	27, 33
T3/human/Ohio/Dearing/1955	T3D/55	M31058	27, 33
T3/human/Washington, D.C./clone 93/1955	T3C93/55	AY007420	13
T1/human/Washington, D.C/clone 62/1957	T1C62/57	AY007411	13
T1/human/Washington, D.C/clone 63/1957	T1C63/57	AY007414	13
T3/human/Washington, D.C/Abney/1957	T3A/57	AY007396	32
T3/human/Washington, D.C./clone 84/1957	T3C84/57	AY007419	7
T1/human/Washington, D.C./clone 15/1958	T1C15/58	AY007398	13
T1/bovine/Maryland/clone 22/1959	T1C22/59	AY007408	13
T1/bovine/Maryland/clone 23/1959	T1C23/59	AY007388	13
T1/bovine/Maryland/clone 24/1959	T1C24/59	AY007412	13
T1/bovine/Maryland/clone 28/1959	T1C28/59	AY007413	13
T1/bovine/Maryland/clone 29/1959	T1C29/59	AY007407	13
T1/human/Mexico/clone 1/1959	T1C1/59	AY007400	13
T1/human/Washington, D.C/clone 11/1959	T1C11/59	AY007415	13
T3/bovine/Maryland/clone 81/1959	T3C31/59	AY007383	13
T1/bovine/Maryland/clone 16/1960	T1C16/60	AY007399	13
T1/bovine/Maryland/clone 33/1960	T1C33/60	AY007402	13
T1/bovine/Maryland/clone 34/1960	T1C34/60	AY007403	13
T1/bovine/Maryland/clone 35/1960	T1C35/60	AY007404	13
T1/bovine/Maryland/clone 39/1960	T1C39/60	AY007405	13
T1/bovine/Maryland/clone 49/1960	T1C49/60	AY007401	13
T1/bovine/Maryland/clone 50/1960	T1C50/60	AY007406	13
T3/bovine/Maryland/clone 43/1960	T3C43/60	AY007397	13
T3/bovine/Maryland/clone 44/1960	T3C44/60	AY007417	7
T3/bovine/Maryland/clone 45/1960	T3C45/60	AY007418	7
T3/human/Tahiti/clone 8/1960	T3C8/60	AY007393	13
T1/bovine/Maryland/clone 19/1961	T1C19/61	AY007410	13
T3/bovine/Maryland/clone 18/1961	T3C18/61	AY007409	13
T3/murine/France/clone 9/1961	T3C9/61	AY007384	13
T1/human/Netherlands/1/1967	T1Neth/67	AY007385	12
T2/human/Netherlands/1/1973	T2Neth/73	AY007422	12
T3/human/Netherlands/1/1983	T3Neth/83	AY007416	12
T1/human/Netherlands/1/1984	T1Neth/84	AY007395	12
T2/human/Netherlands/1/1984	T2Neth/84	AY007423	12
T1/human/Netherlands/1/1985	T1Neth/85	AY007421	12
T2/human/Tokyo/1/1990	T2Tokyo/90	AY007394	12
Cell culture/Chicago/234/1999b	234/99	AY007389	This study
Cell culture/Chicago/469/1999b	469/99	AY007390	This study
Cell culture/Chicago/491/1999b	491/99	AY007391	This study
Cell culture/Chicago/514/1999b	514/99	AY007386	This study
Cell culture/Chicago/758/1999b	758/99	AY007387	This study
Cell culture/Chicago/780/1999b	780/99	AY007392	This study

^aStrains were named by using the following scheme: serotype/species of origin/place of origin/strain designation/year of isolation.

^bSerotype not known.

Reovirus particle count determination.

Cell culture stocks of T3D/55 were processed for transmission electron microscopy by the method of Miller et al. (21). One-hundred-microliter aliquots of virus stock diluted 1:10 in cell culture medium were placed in a Beckman EM90 particle counting rotor and sedimented onto Millipore filter supports with a Beckman Airfuge air-driven ultracentrifuge operating at 30 lb/in² for 60 min. Filters were fixed in 3% glutaraldehyde in Sorensen's phosphate-buffered saline (PBS; pH 7.4), postfixed in 1% OsO₄ in Sorensen's PBS, dehydrated in isopropanol, and embedded in Embed 812 Resin. Two blocks were ultrathin sectioned, stained with 2% methanolic uranyl acetate, followed by Reynolds' lead citrate, and then examined with a LEO 910 transmission electron microscope. Micrographs for particle counting were taken at a magnification of ×4,000, and virus particles were counted (six to eight fields each) on negatives using a grid with a field length of 14 μm. The number of virus particles per milliliter of the original suspension was calculated by using the formula N = 250 V_f(D)/T′L, where N is the number of virus particles per milliliter of the diluted sample, V_f is the average number of virus particles per field, D the dilution factor, T′ is the effective section thickness, and L is the field length in millimeters. T′ is a constant based upon virus diameter; in this case, a value of 0.0001 mm was used. In the present case, the titer of the T3D/55 stock was determined to be (1.16 ± 0.13) × 10¹⁰ viral particles per ml.

Nucleic acid extractions.

Nucleic acid was extracted by using the Viral Nucleic Acid Extraction kit as prepared for Abbott Laboratories (Qiagen Inc., Valencia, Calif.). A 100-μl aliquot was diluted to 200 μl with PBS; a 200-μl aliquot of PBS was extracted as a negative control. For extractions in the presence of HNK cells, 10⁶ viable cells (viability, 95%) diluted in 100 μl of PBS were substituted for PBS. Samples were lysed by the addition of 200 μl of a lysis solution containing 10⁶ copies of an RNA template (pAW109; Perkin-Elmer, Branchburg, N.J.). This template was used as a positive control to monitor the efficiency of nucleic acid extraction and RT. Samples were treated with 25 μl of a protease mixture and incubated at 70°C for 10 min. Ethanol (200 μl) was then added and mixed. Nucleic acid was collected on a spin column and washed twice with 500 μl of AW1 buffer and twice with 500 μl of AW2 buffer. Nucleic acid was eluted from the column in a volume of 100 μl by the addition of 110 μl of RNase- and DNase-free water at 70°C.

RT.

RT was performed in accordance with the manufacturer's (Perkin-Elmer) instructions. A 15-μl aliquot of nucleic acid was incubated at 94°C for 2 min in the presence of 250 pM random hexamers and then snap cooled on ice. Eighty microliters of chilled RT mixture was added, resulting in a final concentration of 1× PCR buffer II, 5 mM MgCl₂, 1 mM each deoxynucleoside triphosphate (dNTP), 100 U of RNase inhibitor, and 250 U of murine leukemia virus reverse transcriptase (Perkin-Elmer). Reaction mixtures were incubated at 20°C for 10 min and then at 42°C for 30 min. Each RT assay mixture included an RT control that contained water in place of nucleic acid.

PCR.

PCR was performed in accordance with the manufacturer's (Perkin-Elmer) instructions in a 9600 DNA Thermocycler. In the primary PCR, 25 μl of the RT reaction mixture was used in a 50-μl final reaction volume, with final concentrations of 1× PCR buffer II, 2.5 mM MgCl₂, 500 μM each dNTP, 1 μM each oligonucleotide primer, and 1.25 U of AmpliTaq DNA polymerase (Perkin-Elmer). Samples were incubated at 94°C for 1 min and then amplified for 35 cycles of 94°C for 20 s, 50°C for 30 s, and 72°C for 30 s, followed by an additional incubation at 72°C for 10 min. A secondary PCR contained 1× PCR buffer II, 2 mM MgCl₂, 200 μM each dNTP, 0.5 μM each oligonucleotide primer, 0.625 U of AmpliTaq DNA polymerase, and 1 μl of the primary PCR amplification product (416 bp) in a final volume of 25 μl. Secondary amplification was done with the same cycling parameters as the primary amplification. The expected size of the L1 gene secondary PCR product is 344 bp. The oligonucleotide primers used to detect the control RNA transcript were DM155 and DM156 (Perkin-Elmer), which generate a product of 306 bp.

Oligonucleotide primers.

The oligonucleotide primers for the primary L1 gene PCR were L1.rv5 (forward; 5′-GCATCCATTGTAAATGACGAGTCTG-3′) and L1.rv6 (reverse; 5′-CTTGAGATTAGCTCTAGCATCTTCTG-3′). The oligonucleotide primers for the secondary L1 gene PCR were 1.rv7 (forward; 5′-GCTAGGCCGATATCGGGAATGCAG-3′) and L1.rv8 (reverse; 5′-GTCTCACTATTCACCTTACCAGCAG-3′).

Agarose gel electrophoresis and sequencing of PCR products.

Secondary reovirus L1 gene PCR products and pAW109 products were visualized by agarose gel electrophoresis. Gels contained 1× Tris-acetate-EDTA, 2% agarose, and 200 μg of ethidium bromide per liter; the electrophoresis running buffer contained 1× TAE and 200 μg of ethidium bromide per liter. Amplification products were visualized with UV light and photographed with an EagleEye Still Video System (Stratagene, La Jolla, Calif.). Samples were scored positive for reovirus L1 gene RNA if cDNA fragments of the appropriate sizes were detected in both the secondary PCR product and the pAW109 control template. PCR products were isolated with the QIAquick PCR Purification Kit (Qiagen) and directly sequenced with an Applied Biosystems 377 DNA sequencer (Applied Biosystems, Foster City, Calif.).

Phylogenetic analysis of L1 gene nucleotide sequences.

Sequences of the primary PCR products were aligned by using the Genetics Computer Group (Madison, Wis.) software, version 9.1. Phylogenetic trees were constructed from variation in the unique (non-primer-containing) L1 gene nucleotide sequences (365 bp corresponding to nucleotides 1913 to 2277) by using the branch-and-bound algorithm of the phylogenetic analysis program PAUP, version 3.1.1 (39). Phylogenetic trees were also constructed from the same data by using the neighbor-joining algorithm included in the Phylogeny Inference Package PHYLIP (10). Branching orders of the phylograms were verified statistically by resampling the data 1,000 times in a bootstrap analysis with the branch-and-bound algorithm as applied in PAUP.

Nucleotide sequence accession numbers.

The nucleotide sequences obtained by using our optimized RT-PCR assay have been deposited in GenBank and are listed in Table 1.

RESULTS

Detection of reovirus L1 gene segment sequences by RT-PCR.

To enhance the sensitivity with which a previously described RT-PCR technique detects the reovirus L1 gene segment (42), we aligned the nucleotide sequences of prototype reovirus strains T1L/53, T2J/55, and T3D/55 (46) to identify regions of maximum sequence conservation. Oligonucleotide primers for RT-PCR were designed such that sequence regions of absolute conservation (7 to 10 bp) were placed at the 3′ ends of the primers, while positions further 5′ were selected based on the most highly conserved nucleotide. A nested PCR protocol was designed to amplify a 344-bp fragment of the L1 gene segment corresponding to nucleotides 1930 to 2273. Preliminary primer sensitivity was examined by using the prototype strains, as well as reovirus isolates not detected by the previously described L1 gene RT-PCR technique (42). The identities of these amplified L1 gene PCR products were confirmed by nucleotide sequence analysis. Alignment of the partial L1 gene sequences of these newly identified reovirus strains was used to further modify the primers to enhance conservation with respect to the L1 gene segment.

Having established the theoretical utility of the optimized L1 gene oligonucleotide primers, we tested the RT-PCR assay for the ability to detect L1 gene sequences from 44 independent reovirus isolates (Table 1). The strains used were isolated from different mammalian hosts and geographic locales over a period of approximately 50 years. Additionally, representatives of all three reovirus serotypes were included in the analysis. In all of the samples tested, the L1 gene fragment was successfully amplified by using the optimized L1 gene RT-PCR assay (data not shown). The amplified fragments were subjected to DNA sequence analysis to confirm that the PCR products corresponded to the reovirus L1 gene segment and to eliminate the possibility of PCR contamination. The cDNA sequences of these PCR products confirm that each corresponds to the reovirus L1 gene segment.

L1 RT-PCR assay sensitivity.

To define the sensitivity of the L1 gene RT-PCR assay, 10-fold serial dilutions, ranging from 10⁵ to 10⁻⁵ PFU, of a reovirus T3D/55 stock were used as templates in independent RT-PCRs. This virus stock was determined to be 1.0 × 10⁸ PFU/ml (performed in duplicate with a variance of <20%) by plaque assay (43) and 1.16 × 10¹⁰± 0.13 × 10¹⁰ particles per ml by transmission electron microscopy. Therefore, the particle-to-PFU ratio of this virus stock was 116:1. Viral nucleic acids were extracted in the presence of either HNK cells or PBS to establish assay sensitivity in the presence or absence of background nucleic acids. Following RT-PCR, the L1 gene fragment was detected in the dilutions that contained ≥10⁻² PFU (1.16 ± 0.13 reovirus particles) per RT-PCR in the presence of either PBS or HNK cell lysates (data not shown). The robustness of the assay at the limit of detection was established by two independent operators on two different days testing 10 identical samples on each occasion (40 total data points). At all data points, a positive signal was detected where expected. Therefore, the optimized L1 gene RT-PCR assay can detect nearly a single particle of reovirus, and extraneous nucleic acids in the reaction do not interfere with assay sensitivity.

L1 gene RT-PCR assay specificity.

To define the specificity of the newly described oligonucleotide primers for reovirus, nucleic acids from viral isolates corresponding to a number of distinct virus families were extracted and utilized as templates in the L1 gene RT-PCR assay. High-titer stocks of bovine viral diarrhea virus, herpes simplex viruses 1 and 2, Moloney murine leukemia virus, measles virus, parainfluenza virus type 3, pseudorabies virus, and three strains of rotavirus were prepared from cell culture. Nucleic acids were extracted from these stocks and tested by using both primer pairs individually and in the nested format. In no case did the oligonucleotide primers amplify sequence fragments of the appropriate sizes (data not shown). Therefore, the newly described oligonucleotide primers appear to possess a high degree of specificity for the reovirus L1 gene segment.

Sequence diversity of the reovirus L1 gene and λ3 protein.

The 365-nucleotide primary PCR products were used to analyze the L1 gene and λ3 protein sequence diversity of the 44 strains studied. Analyses of the L1 gene sequences determined that 235 (64.4%) of the 365 nucleotide positions were occupied by identical residues. Most of the observed nucleotide sequence variability occurred at synonymous positions, ranging from 0 to 24.7% in pairwise comparisons (data not shown). In this analysis, variability at synonymous positions between T2J/55 and all of the other strains was greater than any other pairwise comparison (data not shown).

In contrast to the high degree of variability among the L1 gene nucleotide sequences, the deduced amino acid sequences of the L1-encoded λ3 protein were highly conserved. The L1 gene nucleotide sequence fragment amplified by the primary PCR corresponds to residues 633 to 753 of the deduced λ3 sequence (46) (Fig. 1). This sequence includes a glycine-aspartate-aspartate motif, which is found in many RNA-dependent RNA polymerases (2, 16, 22). Few sequence polymorphisms were observed in this region, and none were identified in the glycine-aspartate-aspartate motif. Of the 121 λ3 protein positions analyzed, 111 (91.7%) were occupied by identical residues. In pairwise amino acid sequence comparisons of these strains, identities ranged from 95.0 to 100% (Fig. 1). Thus, residues 633 to 753 in the deduced λ3 amino acid sequence are highly conserved among this sampling of reovirus strains.

FIG. 1.

Alignment of deduced amino acid sequences of λ3 protein. Amino acid residues 633 to 753 of the T1L/53 λ3 protein are shown in the single-letter code. Residues that are identical to the T1L/53 sequence are indicated by dashes. Amino acid positions are numbered above the sequences, and the glycine-aspartate-aspartate motif is underlined. Strains with identical amino acid sequences in this region of λ3 are (i) T1L/53, T3C93/55, T1C62/57, T3C84/57, T1C15/58, T1C28/59, T1C1/59, T1C11/59, T1C16/60, T1C49/60, T3C44/60, T3C45/60, T1C19/61, T3Neth/83, T1Neth/84, T2Neth/84, and T1Neth/85; (ii) T3D/55, T3A/57, T3C31/59, T3C43/60, T3C9/61, and T1Neth/67; (iii) T1C22/59, T3C18/61; (iv), T1C23/59, T1C24/59, 514/99, and 758/99; (v) T1C29/59, T1C33/60, T1C34/60, T1C35/60, T1C39/60, and T1C50/60; and (vi) 469/99, 491/99, and 780/99.

Phylogenetic analysis of nucleotide sequence diversity within the L1 gene segment.

The availability of partial L1 gene sequences from 44 reovirus strains provided an opportunity to rigorously define the relationship of L1 gene sequence diversity to the viral serotype, host origin, geographic site, and date of isolation. Variability among the L1 nucleotide sequences was analyzed at each nucleotide position by using the branch-and-bound algorithm of the phylogenetic analysis program PAUP (39). The resulting minimum-length phylogenetic tree delineates the most probable evolutionary relationships among these L1 gene segments (Fig. 2). The most noteworthy features of this L1 gene phylogeny are that the T2J/55 L1 gene segment was selected as the root by the midpoint rooting procedure and that the L1 gene sequences do not segregate into discrete lineages defined by serotype. Exemplifying this pattern is the clustering of L1 gene segments of T1Neth/67 and T3C31/59. A serotype-independent mode of L1 gene evolution is further supported by the phylogeny of the T3C93/55 L1 gene segment, which has a more recent ancestor in common with L1 sequences of type 1 and 2 strains than with other type 3 strains. The major branching order of the L1 gene phylogram was corroborated by bootstrap analysis and an exhaustive search of possible phylogenetic trees (39) (Fig. 2). An identical tree topology was produced by the neighbor-joining algorithm as applied in the Phylogeny Inference Package PHYLIP (10) (data not shown). These results suggest that the L1 gene segment has evolved independently of the viral serotype.

FIG. 2.

Minimum-length phylogenetic tree based on partial L1 gene nucleotide sequences of 44 reovirus strains. The phylogenetic tree for 365 nucleotides of the L1 gene (nucleotides 1913 to 2277) of the strains shown in Table 1 was constructed by using the branch-and-bound algorithm of the parsimony program PAUP. The tree is rooted at its midpoint, and nucleotide changes for each branch are shown above the branch lines. Bootstrap values of >50% (indicated as percentages of 1,000 repetitions) for major branches are shown at the nodes. Strains with identical nucleotide sequences in this region of L1 are indicated as follows: a, T3C31/59 and T3C9/61; b, T1C23/59 and T1C24/59; c, T1C62/57, T1C63/57, and T1C28/59; d, T1C22/59 and T3C18/61; e, T1C33/60, T1C34/60, T1C35/60, and T1C39/60; f, T1L/53, T1C15/58, T1C1/59, T1C16/60, and T1C49/60; g, T3C93/55, T3C84/57, T3C44/60, and T3C45/60; h, 469/99, 491/99, and 780/99.

Phylogenetic analysis also demonstrates that L1 diversity correlates with neither the host species nor the geographic origin of the isolates. For example, the L1 gene segments of reovirus strains T1L/53 and T1C33/60, viruses isolated from a human in Ohio and a cow in Maryland, respectively, have an ancestor in common. Also, the L1 gene segment of a bovine reovirus strain isolated in Maryland (T1C22/59) segregates with the L1 gene segment of a human strain isolated 24 years later in The Netherlands (T3Neth/83). Thus, evolution of the L1 gene segment apparently does not restrict the host range of mammalian reoviruses, at least among bovine, human, and murine species.

Since the results of our phylogenetic analysis did not support an association between L1 sequence diversity and the viral serotype, we hypothesized that the L1 gene segment and the S-class gene segments evolve independently. We formally tested this hypothesis by using PAUP to construct phylogenetic trees of the S1 (7, 23), S2 (3, 8), S3 (12, 45), S4 (15), and L1 gene sequences from nine reovirus strains for which sequence information is available (Fig. 3). The major branching orders of phylogenetic trees constructed from the L1 and S-class gene segments were found to be markedly different. For example, the S1 sequences of strains T1L/53 and T2J/55 have an immediate ancestor in common, whereas the S2, S3, S4, and L1 sequences of these strains are more distantly related. A similar relationship of S1, S2, S3, S4, and L1 sequences exists for strains T3C31/59 and T3C18/61. Thus, phylogenetic analyses of nucleotide sequence diversity among these reovirus strains indicate that the evolutionary histories of the S1, S2, S3, S4, and L1 gene segments are independent of one another.

FIG. 3.

Minimum-length phylogenetic trees based on S1, S2, S3, S4, and L1 gene nucleotide sequences of T1L/53, T2J/55, T3D/55, and six type 3 reovirus field isolate strains. S1, S2, S3, and S4 phylogenetic trees were constructed from full-length δ1-encoding (7, 23), δ2-encoding (3, 8), δNS-encoding (12, 45), and δ3-encoding (15) sequences, respectively. The L1 phylogenetic tree was constructed from nucleotides 1913 to 2277 by using the branch-and-bound algorithm of the parsimony program PAUP. Each tree is rooted at the midpoint of its longest branch. Bootstrap values of >50% (indicated as percentages of 1,000 repetitions) for major branches are shown at the nodes.

DISCUSSION

In this report, we describe a new RT-PCR assay for detection of reovirus L1 gene sequences. This assay was designed to detect a broad array of divergent reovirus isolates based on conserved regions of the L1 gene segment. The L1 gene RT-PCR assay was shown to detect a variety of geographically distinct reovirus strains isolated over the past 50 years. This assay has exceptional sensitivity and is capable of detecting virtually a single virus particle (1.16 ± 0.13) if it is present in the reaction mixture. Therefore, the L1 gene RT-PCR assay should substantially improve diagnostic strategies for reovirus.

Prior to the initiation of this study, we evaluated the sensitivity and specificity of several previously reported RT-PCR assays for detection of reovirus RNA (6, 38, 42). In each case, the L1 gene RT-PCR assay described here was found to be superior. A number of differences in assay procedures may explain these findings. First, random hexamers were used instead of gene-specific oligonucleotides to prime the RT reaction. The enhanced sensitivity most likely results from the capacity of random hexamers to generate multiple cDNA copies per available RNA template in comparison to specific priming. Second, the new L1 RT-PCR assay utilizes a three-step process for RT and nested PCR. Each of the RT-PCR assays described previously combined both RT and PCR in a single step, and in one case (6), a nested PCR was not performed. In our experience, uncoupling RT and the primary PCR results in enhanced sensitivity, as does performance of a nested reaction. Third, PCR primers were modified to enhance conservation with template as more sequences became available. This iterative process of primer design resulted in PCR primers that were maximally homologous to the L1 gene sequences of the strains studied.

The L1 gene RT-PCR assay described in this report was sufficiently robust to detect L1 gene sequences from 44 independent reovirus strains. In no case did the assay fail to detect L1 gene sequences in samples demonstrated by other methods to contain reovirus, and in some cases, unique reovirus L1 gene sequences were detected when other methods or RT-PCR assays failed. Additionally, the use of RT-PCR is considerably faster than conventional methods of virus isolation. This improvement in efficiency can be extremely important in instances in which early detection of reovirus is important, for example, to prevent contamination of cell lines in a laboratory environment or to eliminate costly downstream processes that are dependent upon virus-free initiating materials.

Analysis of the partial L1 gene sequences determined in this study allowed us to extend current hypotheses about reovirus evolution. The L1 gene segment encodes the λ3 protein (20), which is a minor constituent of the viral core (4). The function of λ3 is not entirely understood; however, existing evidence indicates that the protein is involved in viral RNA synthesis. The λ3-encoding L1 gene determines strain-specific differences in the pH-optimum of transcription (9), and the deduced amino acid sequence of λ3 contains a sequence motif that is common to many RNA-dependent RNA polymerases (2, 16, 22). Baculovirus-expressed λ3 mediates poly(C)-dependent poly(G) polymerase activity but does not catalyze transcription of viral gene segments (37). This finding indicates a requirement for proteins in addition to λ3 in viral RNA synthesis. In contrast to the L1 gene nucleotide sequences of the 44 strains studied, the corresponding region of λ3 (amino acids 633 to 753) was highly conserved; sequence polymorphism was limited to only 10 of 121 residues. This high degree of conservation also was observed between strains isolated from different host species. These findings suggest that at least some regions of λ3, perhaps because of its role in viral RNA synthesis, cannot accommodate significant primary sequence variability and that λ3 is well adapted to different host species. From this analysis, it appears that divergent λ3 sequences do not confer a selective advantage in the reovirus-host interaction.

Reoviruses are capable of serotype-independent gene segment reassortment in experimental settings, such as coinfections of cultured cells (11) or mice (44). Analysis of phylogenetic trees constructed from protein coding sequences of the S1 (7), S2 (3), S3 (12), and S4 (15) genes derived from field isolate strains indicates that reovirus genes also reassort in nature. Findings made in the present study support this idea in that we found no relationship between the L1 gene sequences and the viral serotype. Moreover, phylogenetic trees constructed from the L1 gene sequences were distinct from those constructed from the S1, S2, S3, and S4 gene segments. Therefore, like the S-class gene segments of mammalian reoviruses, the L1 gene segment appears to be evolving independently of other reovirus gene segments.

Unlike many animal viruses, host range determinants for mammalian reovirus have not been identified. Nucleotide sequence variability of S1 (7), S2 (3), S3 (12), and S4 (15) genes of field isolate strains does not correlate with the host species, geographic site, or date of isolation. Like the S-class gene segments, the L1 gene does not appear to be a host range determinant for reoviruses. These observations suggest that reovirus strains containing gene segments with divergent sequences cocirculate among different host species at a given geographic site. It is likely that maintenance of divergent reovirus genes in a given population at the same time is due to gene segment reassortment (7), and no particular allelic combination appears to offer a selective advantage. These findings suggest that reoviruses are well adapted to mammalian species and sequence polymorphism does not confer enhanced viral fitness. In this sense, reoviruses can be considered evolutionarily ancient viruses that have reached genetic equilibrium with their host species.

Despite the sequence diversity within the reovirus L1 gene segment, we have been able to develop a highly sensitive and specific RT-PCR technique to detect reovirus RNA. This new assay should substantially improve reovirus diagnostics and enhance the detection of these viruses in cell culture. Moreover, application of RT-PCR diagnostics may be useful in linking reoviruses to certain disease states in humans (42).