Rearrangements of Rotavirus Genomic Segment 11 Are Generated during Acute Infection of Immunocompetent Children and Do Not Occur at Random

Abstract

Group A rotaviruses are the main cause of viral gastroenteritis in infants. The viral genome consists of 11 double-stranded RNA (dsRNA) segments. Dysfunction of the viral RNA polymerase can lead to gene rearrangements, which most often consist of partial sequence duplication of a dsRNA segment. Gene rearrangements have been detected in vivo during chronic infection in immunodeficient children or in vitro during passages at a high multiplicity of infection in cell culture, suggesting that these replication conditions lead to selective advantages favoring the recovery of viruses with rearranged genes. During acute rotavirus infection, the replication level is high, but the occurrence of rearrangement events has never been reported. By the use of a reverse transcription-PCR assay specifically designed to detect small numbers of copies of rearranged forms of segment 11 in a high background of its standard counterpart, we detected 12 rearrangement events among 161 cases (7.5%) of acute rotavirus infection in immunocompetent children. Strikingly, in all but one case, rearrangement took place at the same location within the short direct repeat AUGU sequence. For the unique case with a different rearrangement pattern, the rearrangement occurred within the direct repeat ACAAGUC that was specific for this isolate. In conclusion, we report the occurrence of segment 11 rearrangements during acute rotavirus infection in immunocompetent children. We show that under such conditions of infection, the viral RNA polymerase generates rearrangements which occur not at random but within direct repeats which might constitute hot spots for RNA recombination.

Group A rotaviruses are the main cause of viral gastroenteritis in infants and in the young of many animal species. The viral genome consists of 11 segments of double-stranded RNA (dsRNA) which can be separated by polyacrylamide gel electrophoresis (PAGE). Electropherotype profiles of rotavirus dsRNA typically exhibit four well-defined size classes of segments (10). However, some group A rotaviruses show unusual electropherotypes, in which segments of a standard size are replaced by rearranged forms, which most often are of a larger size. Rotaviruses with a rearranged genome (for a review, see reference 9) were first isolated from chronically infected immunodeficient children (18, 29) and animals (4, 30, 34) and were eventually generated in cell culture by serial passages at a high multiplicity of infection (MOI) with animal (17, 32) or human (19) rotavirus strains.

Gene rearrangements in rotaviruses mostly involve segment 11 which encodes the two nonstructural proteins NSP5 and NSP6 and less frequently involve segments 5, 6, 7, 8, 9, and 10. It is not known whether the rearrangements in segment 11 occur more frequently or if viruses with a rearrangement in segment 11 have some selective advantage so that they are detected more easily (10). Sequence analyses of rearranged segments have shown that gene rearrangement usually results from a partial head-to-tail duplication of the dsRNA sequence (3, 7, 11, 13, 14, 16, 19, 20, 22, 23, 26, 28, 31, 32, 35, 36). In most cases, sequence duplication starts after the stop codon and extends up to the 3′ end, leading to a long 3′ untranslated region (UTR) (9). As a result, rearranged segments contain a normal 5′ UTR and open reading frame (ORF) and keep their ability to encode normal proteins (7, 11, 26, 32). Less frequently, sequence duplication may occur within the ORF. Gene rearrangements leading to a modified ORF have been described for segments 5 (35) and 7 (11). In both cases, the resulting viruses retained their capacity to grow in cell culture, although they expressed a modified NSP1 (16, 35) or NSP3 (11) protein. Rearrangements involving sequence deletions have occasionally been reported. Tian et al. (35) have described a mutant virus, P9delta5, with a single 308-bp deletion in its gene 5 ORF, resulting in a NSP1 protein truncated to its first 150 amino acids (aa). Additionally, deletions can occur in the duplicated part of a rearranged gene, as described for segment 11 (14, 23, 28) or for segment 7, in which the deletion resulted in a modified NSP3 protein (11). It may be expected that deletions or duplications can occur within the ORFs of all the genes, but these may be detectable only for genes 5 (NSP1) and 7 (NSP3), since their protein products might be not essential for virus replication; for other genes, most of these changes may result in lethal mutations that yield progeny incapable of growth. Hypothetically, the length of the 3′ UTR may also be a factor affecting the ability of the gene to rearrange without perturbing the ORF, as a longer 3′ UTR would provide a greater target region for rearrangement to take place.

Thus, gene rearrangements have been proposed to take a part in the evolution of rotaviruses (besides point mutations and gene reassortments) and to contribute to their diversity (9, 33). Although rotavirus strains with a rearranged segment 11 (11R) of unknown (5, 12) or animal (20, 23) origin have seldom been found to circulate among immunocompetent children, gene rearrangements have been detected mostly in vivo during chronic infection in immunodeficient children or in vitro during passages at a high MOI in cell culture. It is assumed that these unusual replication conditions lead to selective advantages favoring the recovery of viruses with rearranged genes. This might explain why rearrangement events are not reported to occur during acute rotavirus infection, despite high levels of viral replication. Indeed, if a rearrangement occurred in the course of acute infection, the resulting rearranged viral population would not have enough time to expand and would remain in a minority compared to the wild-type viral population. Consequently, the resulting rearranged segment would be difficult to detect among the high background of its standard counterpart. We developed a nested reverse transcription (RT)-PCR assay specifically designed to detect a small number of copies of 11R in a high background of standard segment 11 (11S), in order to determine whether gene rearrangements containing a sequence duplication generated by the viral RNA polymerase could be detected in the course of acute rotavirus infection. By the use of this assay, we detected the occurrence of rearrangement events in 12 out of 161 cases (7.5%) of acute rotavirus infection in immunocompetent children. In all but one case, the rearrangement occurred at the same location within the segment 11 sequence, suggesting a nonrandom mechanism for gene rearrangement.

MATERIALS AND METHODS

Viruses and cells.

Human rotaviruses (HRV) M0 and M2 are two previously described cell culture-adapted viral clones isolated from the stool of a chronically infected child with severe combined immunodeficiency syndrome (11). Virus M0 has 11 standard dsRNA segments, whereas virus M2 harbors a rearranged segment 7 (7R) and a 11R. Virus propagation on confluent monolayers of MA-104 cells was performed as previously described (11). The viral stocks titers were 5 × 10⁶ PFU/ml for M0 and 1 × 10⁶ PFU/ml for M2.

To obtain cell culture-adapted viruses from stool samples, 10% stool suspensions in 150 mM NaCl were clarified by low-speed centrifugation and inoculated to MA-104 cells, either directly or after pelleting the viruses by ultracentrifugation through a 45% sucrose cushion at 40,000 rpm for 4 h. Cell lysates were further serially propagated on MA-104 cells in 25-cm² flasks for five 72-h culture passages, and aliquots of each passage were kept frozen at −80°C for further analysis.

Clinical samples.

Stool samples were obtained from children attending the Emergency Unit of Trousseau Hospital (Paris, France) for acute rotavirus gastroenteritis. The diagnosis of rotavirus infection was established by the detection of group A VP6 antigen in stools, using a commercial enzyme-linked immunosorbent assay technique (IDEIA RV; Dako, Cambridgeshire, United Kingdom). Patients considered for the study were immunocompetent children aged less than 3 years. Rotavirus-positive stool samples used for further analysis (n = 161) were randomly chosen among available samples from these patients (maximum of one stool sample per patient) during two distinct epidemic periods: December 1998 to February 1999 (n = 50) and December 2000 to February 2001 (n = 111). Stool samples were stored at −20°C until analysis.

Nucleic acid analysis.

Rotavirus genomic dsRNA was extracted either from stool suspensions (∼10% [wt/vol] in 150 mM NaCl) clarified by low-speed centrifugation, or from cell-culture lysates, using RNA-PLUS (Bioprobe Systems, Montreuil, France) according to the manufacturer's recommendations. Concentration of dsRNA stocks of virus M0 and M2 used in all experiments were adjusted to 200 ng per μl. Rotavirus RNA genomic profiles were determined by PAGE in 10% polyacrylamide gels for 16 h at 200 V at room temperature, followed by ethidium bromide staining. The cDNA probe used for the detection of 11S and 11R, corresponded to the full-length segment 11 of virus M0, and was labeled with [α-³²P]dCTP by random priming using a Megaprime DNA Labeling System (Amersham Biosciences, Little Chalfont, United Kingdom). For Northern blotting experiments, rotavirus dsRNAs were separated by electrophoresis on 1% agarose, denatured in 0.1 M NaOH-0.25 M NaCl for 7 min at room temperature, and transferred overnight to Hybond-N+ membrane (Amersham Biosciences, Little Chalfont, United Kingdom) in 20× SSC (1× SSC is 0.15 M NaCl plus 0.015 M sodium citrate). The membrane was rinsed with water, baked for 2 h at 70°C, and prehybridized for 2 h at 42°C in hybridization buffer (50% formamide, 10% dextran sulfate, 5× Denhardt's solution, 50 mM Tris [pH 7.5], 0.8 M NaCl, 0.1% sodium pyrophosphate, 0.5% sodium dodecyl sulfate, and 100 μg/ml of denatured salmon sperm DNA), before the addition of the heat-denatured cDNA probe to the buffer. The membrane was hybridized for 48 h at 42°C, washed four times at 30°C with 2× SSC-0.1% sodium dodecyl sulfate, and exposed to X-ray films with intensifying screens (Amersham Biosciences, Little Chalfont, United Kingdom) at −70°C for 1 to 4 days.

RT-PCR procedures and sequencing.

Primers used for RT-PCR amplifications of 11S and 11R are described in Table 1. After denaturation at 95°C for 5 min followed by a 1-min incubation on ice, 5 μl of the dsRNA extract was reverse transcribed in a 50-μl reaction mixture containing 20 μM EDTA, 10 mM dithiothreitol, 0.5 mM dNTP, 0.1 μM of primer, 10 U of RNase inhibitor (Life Technologies, Cergy, France), 200 U of SuperScript II (Life Technologies) and SuperScript buffer. After a 45-min incubation at 45°C, the reaction was stopped by adding 1 μl of 0.5 M EDTA and 150 μl of distilled water. cDNAs were purified and concentrated by a standard procedure of phenol-chloroform extraction and ethanol precipitation. Further PCR amplifications were performed using one-fourth of the obtained cDNAs. PCRs were performed in a 50-μl reaction mixture containing 10 mM Tris-HCl (pH 8.3), 50 mM KCl, 4 mM MgCl₂, 0.2 mM dNTP, 0.25 μM of each primer, and 1.25 U of AmpliTaq DNA polymerase (Perkin Elmer, Villebon, France). Amplification was achieved with a 9700 Perkin Elmer thermocycler. PCR conditions for full-length amplification of segment 11 sequence were 35 cycles at 94°C for 30 s, 58°C for 1 min, and 72°C for 45 s. Cycles conditions for nested PCR A and B systems were of 94°C for 30 s, appropriate annealing temperature for 30 s, and 72°C for 30 s. First steps of both PCR were 20 cycles, with annealing temperatures of 42°C and 44°C for PCR A and B, respectively. Second steps of the nested PCR were performed using 1 μl of the first-step PCR mixture, under PCR conditions of 35 cycles and an annealing temperature of 58°C. Strict precautions were taken to avoid any carryover when performing PCR, and negative controls (H₂O) were included in each PCR run. PCR products were analyzed after electrophoresis in 1.5% agarose gels.

TABLE 1.

Primers used in RT-PCR procedures

Primer name	Nucleotide sequence (5′ to 3′)	Locationa	Use
11f	GGCTTTAAAAGCGCTACA	1-18 (f)	RT (11f) and full-length PCR of segment 11
11r	CCCACTCCCGTTTTGTGACC	645-664 (r)
11R1f	GTCTTGATTCATGTGTATCAATTT	383-406 (f)	RT (11R1r) and first step of PCR A
11R1r	GGTATTAATGTTAGTGCTAGTCTT	364-387 (r)
11R2f	CATGTGTATCAATTTCAACTAACCA	392-416 (f)	Second step of PCR A; RT(11R2r) and first step of PCR B
11R2r	CTCCTTAACTAAAGGTATTAATGTTAGT	351-378 (r)
11R3f	CAGATAGTGATGACGGTAAATGTA	509-532 (f)	Second step of PCR B
11R3r	TCATCAACACAATCACGACCTTCA	298-321 (r)

^aPosition on the sequence of segment 11 of virus M0 (GenBank accession no. AF338244); f, forward primer; r, reverse primer.

Sequencing was carried out by the dideoxynucleotide chain terminator method by using an ABI Prism 310 automatic sequencer (Applied Biosystems) with an ABI PRISM Big Dye terminator cycle sequencing Ready Reaction kit. Nucleotide sequences of full-length segments 11 and PCR products A and B were determined directly for both strands from two independent PCR products.

Nucleotide sequence accession numbers.

Rotavirus segment 11 sequences of samples S7, S22, S33, S35, S56, S79, S111, S125, S141, S142, S143, and S160 were deposited in GenBank under accession numbers EF590980 to EF590991 for standard sequences and EF590992 to EF591003 for rearranged sequences.

RESULTS

Rearranged segment 11 is specifically detected in a background of its standard counterpart.

Because rotavirus gene rearrangements have most frequently been described for segment 11, we focused on the specific detection of rearranged forms of this segment. Additionally, as rearrangement is probably a rare event during viral replication, we aimed to detect a small number of rearranged sequences among a high background of their standard counterparts. We first evaluated the abilities of PAGE, Northern blot hybridization, and RT-PCR to specifically detect a 11R. Two HRV clones, M0 and M2, with known segment 11 sequences (11) were used for this purpose. Virus M0 contains a 11S, while virus M2 contains a 11R (Fig. 1A, lanes 1 and 2). Virus M2 dsRNA was serially 10-fold diluted in a constant amount of virus M0 dsRNA in order to evaluate whether, and to what extent, 11R could be detected in the presence of 11S.

FIG. 1.

Standard techniques for the detection of a 11R. The ability of PAGE (A), Northern blot hybridization (B), or full-length RT-PCR (C) to specifically detect a 11R was evaluated with serial 10-fold dilutions of virus M2 dsRNA in a constant amount of virus M0 dsRNA (M2:M0 ratios from 1:1 to 10⁻⁷:1). The dsRNA profiles of virus M0 and M2 are shown in the two first lanes of panel A. Compared to its standard counterpart M0, virus M2 contains a 11R instead of a 11S (indicated by large arrows) and a 7R instead of a standard segment 7 (7S) (indicated by thin arrows). Within the M2/M0 mixtures, 11R was clearly detectable at a molar ratio of 1 to 1 and 1 to 10 11S by PAGE and Northern blot hybridization, respectively. (11R was faintly visible at a molar ratio of 1:10 [PAGE] and 1:100 [Northern blot hybridization] on overexposed pictures). Full-length RT-PCR performed with primers pair 11f and 11r gave PCR products of 664 bp for M0 (11S) and of 1,237 bp for M2 (11R). With the M2/M0 mixtures, only 11S was detected. MW, molecular mass marker (100-bp ladder).

As expected, PAGE and Northern blot hybridization were poorly sensitive in detecting 11R within the M2/M0 dsRNA mixtures (Fig. 1A and B). 11R was repeatedly detected at molar ratios to 11S of only 1:10 by PAGE, and 1:100 by Northern blot hybridization, despite the large amount (≥1 μg) of dsRNA used for these assays.

We next evaluated a RT-PCR strategy based on the amplification of full-length segment 11 sequences. Primers 11f and 11r (Table 1), designed to amplify the full-length sequence of both 11S and 11R, were used on the M2/M0 dsRNA mixtures. In all cases, the resulting PCR products corresponded to the full-length 11S, even for the dsRNA mixture containing a 1:1 molar ratio of 11R to 11S (Fig. 1C). This result confirmed that a RT-PCR strategy based on full-length RNA segment amplification cannot be used to detect rearranged sequences in a background of homologous standard sequences, because it favors the amplification of the shortest template.

We thus developed a RT-PCR strategy based on the specific amplification of the region surrounding the rearrangement (Fig. 2). To amplify this region, we used a forward primer in the 3′ part and a reverse primer in the 5′ part of the sequence. These primers are in a back-to-back position on the 11S sequence, with the forward primer located downstream of the reverse primer. Thus, the back-to-back primer pair can amplify only rearranged forms of segment 11 containing a head-to-tail duplication of the sequence (Fig. 2). We developed two nested RT-PCR assays (termed A and B), using different combinations of back-to-back primer pairs (Table 1 and Fig. 2), which can theoretically amplify all the 11R sequences described in the literature (7, 11, 13, 14, 20, 22, 23, 28, 31, 36). Considering the location of the primer sets, these assays can detect all sequence duplications occurring after nucleotide (nt) 416 and reinitiating before nt 351. Although this system can detect a large range of possible rearrangements, we cannot exclude the possibility that duplications involving sequences not assayed here may be missed. The abilities of both assays to specifically detect 11R were first evaluated using M0 and M2 dsRNA individually. As shown in Fig. 3A (lanes 1 and 2), RT-PCR assays A and B gave concordant results. When M0 dsRNA was used as template, no PCR product was obtained, indicating that the assays were specific and unable to amplify 11S sequences with no head-to-tail duplication. When M2 dsRNA was used as template, PCR products of an expected size of 560 bp and 386 bp were obtained with assays A and B, respectively. Additionally, the sequence of the PCR products exactly matched the corresponding sequence of virus M2 11R.

FIG. 2.

RT-PCR strategy for the specific detection of segment 11 rearrangements. 11S and 11R of viruses M0 and M2 are shown. The large box indicates the NSP5 ORF, and the small boxes indicate the UTRs. The duplicated sequence generated by a nt 614 to nt 42 rearrangement in 11R is shaded. White arrows indicate primer pairs 11R1f and -r, 11R2f and -r, and 11R3f and -r, in a back-to-back position on the segment 11 sequence, with the forward (f) primer located downstream of the reverse (r) primer (no possible amplification). Black arrows indicate the location of the same primers on the 11R sequence, in a position allowing specific amplification of the rearranged sequence. PCR A and B products were obtained by the use of nested primers 11R1f and -r and 11R2f and-r, and primers 11R2f and -r and 11R3f and -r, respectively. Gray arrows indicate primers 11f and 11r used for full-length amplification of segment 11.

FIG. 3.

Specificity and sensitivity of the RT-PCR assays in detecting a segment 11 rearrangement. The nested RT-PCR assays A and B were evaluated with virus M2 dsRNA serially 10-fold diluted (A) in H₂O (10⁻¹ to 10⁻⁷) or (B) in a constant amount of virus M0 dsRNA (M2:M0 ratios from 1:1 to 10⁻⁶:1). 11S of virus M0 (first lane) was not amplified, while 11R of virus M2 gave a PCR product of 560 bp and 386 bp with RT-PCR assay A and B, respectively. MW, molecular mass marker (100-bp ladder); the open arrowhead indicates the band migrating at 500 bp.

Sensitivity of the assays was evaluated with 10-fold serial dilutions of virus M2 dsRNA (Fig. 3A). RT-PCR assay B was positive up to a 10⁻⁶ dilution (corresponding to a dsRNA input of 0.06 pg in the PCR), and was more sensitive than assay A (positive up to a 10⁻⁴ dilution), probably because of the shorter length of the target sequence. Considering that the 11R in virus M2 (1,237 bp) roughly represents 1:16 of the total genome length, the detection limit of assay B was estimated at 0.004 pg of 11R, corresponding to approximately 3,000 copies.

When tested with M2/M0 dsRNA mixtures (Fig. 3B), RT-PCR assay B gave a positive signal up to a 10⁻⁴ dilution, corresponding to an input of 6 pg of M2 dsRNA within a background of 60 ng of M0 dsRNA in the RT-PCR, and was more sensitive than assay A (positive up to a 10⁻³ dilution). Thus, RT-PCR assays A and B were able to detect the 11R at molar ratios to 11S of 1:1,000 and 1:10,000, respectively, although the presence of 11S lowered the assay sensitivity.

Taken together, these results indicate that nested RT-PCR allows specific and sensitive detection of rearranged forms of segment 11 within a high background of 11S, thus providing a useful tool to detect a minority of rearranged segments in stool samples during acute rotavirus infection.

Rearranged forms of segment 11 are detected in the feces of immunocompetent children with acute rotavirus infection.

Stool samples (n = 161) were collected from 161 immunocompetent children with acute rotavirus gastroenteritis. dsRNA was extracted for RT-PCR procedures to detect standard and rearranged forms of segment 11. In all of the 161 samples, full-length 11S was successfully amplified. This was performed as a control to confirm the absence of PCR inhibitors in the dsRNA extracts (see the control PCR in Fig. 4A). Samples were then tested with both assay A and assay B for the presence of rearranged forms of segment 11. Each assay was repeated twice for each sample. The RT-PCR results obtained with assay B are shown in Fig. 4A. A segment 11 rearrangement was reproducibly detected in 12 out of 161 samples (7.5%), as judged by the presence of a discrete band on agarose gels. RT-PCR assay A, which used a distinct primer combination, was 100% concordant with assay B, although PCR signals obtained with assay A were often weak, in agreement with the low sensitivity of this assay compared to that of assay B (Fig. 4B). The 12 samples containing a rearranged gene 11 were collected at different times during two distinct nonconsecutive epidemics periods: 4 (out of a total of 50) during the 1998-to-1999 epidemics, and 8 (out of a total of 111) during the 2000-to-2001 epidemics, indicating that these 12 samples were not issued from a common localized outbreak.

FIG. 4.

Detection of rearranged forms of segment 11 in stool samples. In each RT-PCR experiments, dsRNA from viruses M0 and M2 and no-RNA (H₂O) were used as controls. The size of the PCR products expected for M0 11S and M2 11R are indicated by open and filled arrowheads, respectively. (A) RT-PCR B results showing the 12 samples positive in the specific detection of a segment 11 rearrangement (S7 to S160). S7, S22, S33, and S35 were collected during the 1998-to-1999 epidemics (28 December 1998 and 3 January, 2 February, and 4 February 1999, respectively). S56, S79, S111, S125, S141, S142, S143 and S160 were collected during the 2000 to 2001 epidemics (5 December 2000 and 1 January, 7 January, 9 January, 30 January, 2 February, 9 February, and 22 February 2001, respectively). All PCR products had a similar size, except for sample S141. Four negative samples (S136 to S159) are shown as an example. Control PCR showed that full-length amplification of 11S was positive for all the samples (S7 to S159). (B) Comparison of RT-PCR A and B results obtained for samples S56 and S141. MW, molecular mass marker (100-bp ladder).

Interestingly, although amplification of distinct rearranged sequences was expected to generate PCR products of various size according to the rearrangement location and the length of the duplication, all but one of the PCR products had the same apparent size. Eleven out of 12 PCR products migrated slightly above the amplicons obtained from virus M2 (i.e., 560 bp and 386 bp for RT-PCR A and B, respectively), and one (sample S141) migrated slightly below them (Fig. 4). These results indicated that at least two forms of segment 11 rearrangements were detected in the samples and that most of these rearrangements were probably very similar, considering the sizes of the PCR products.

The dsRNA profiles of 5 out of the 12 positive samples (S7, S22, S56, S141, and S143) for which sufficient material was available were examined by PAGE and Northern blot hybridization to investigate whether a rearranged dsRNA segment 11 of a larger size could be detected. Among the five strains that were analyzed by PAGE, four exhibited distinct electropherotypes (Fig. 5). As expected from the poor sensitivity of these techniques, all isolates had a typical group A rotavirus dsRNA profile and no additional segment 11 of a larger size was detected. This result confirmed that RT-PCR had detected rearranged forms of segment 11 that were in a minority within the viral population.

FIG. 5.

dsRNA profile of five rotavirus isolates with a PCR positive detection of a segment 11 rearrangement. dsRNA of the five rotavirus samples S7, S22, S56, S141, and S143 were analyzed by (A) PAGE and (B) Northern blot hybridization using a segment 11-specific probe. All the isolates had a typical rotavirus group A RNA profile, with a 11S. M0, M2, and an M0/M2 dsRNA mixture were used as controls. The locations of M0 11S and M2 11R are indicated.

Additionally, these five rotavirus isolates were successfully adapted on MA-104 cell culture from the stool supernatants. Viruses recovered from the first five cell culture passages were tested for the presence of a 11R by PAGE and RT-PCR. In all cases, dsRNA profiles of cell culture-adapted viruses were identical to those obtained directly from the stool samples, but the 11R was no longer detected by RT-PCR (results not shown). The failure to recover cell culture-adapted viruses with a 11R might be related to a too-small number of viruses with a rearranged genome in the stool sample or to a selective cell culture disadvantage. Plaque isolation would be useful to specify this point. However, with HRV, it is not possible to obtain plaques directly from the stool sample. Plaque isolation from the cell culture-adapted viruses could have been an alternative, but considering the RT-PCR sensitivity threshold, the number of plaques to analyze to have a chance of detecting a non-wild-type virus would exceed ten thousand.

Taken together, these results indicate that rearranged forms of segment 11 can be generated during acute rotavirus infection in immunocompetent children, although they remain in a minority compared to their standard counterparts. Moreover, 11 out of the 12 PCR products had the same apparent size in agarose gels, suggesting that they might have closely related sequences resulting from a similar rearrangement event. To clarify this point, the sequence of the PCR products was determined.

Segment 11 rearrangements do not occur at random within the sequence.

For each of the 12 samples found to contain rearranged forms of segment 11, the full-length 11S and the PCR product obtained with assay B were sequenced. The sequence of the PCR product obtained with assay A could be determined for only seven samples, as the remaining PCR products were in amounts too low to be sequenced.

The 12 full-length 11S sequences were 664 nt long. Only two sequences were identical (S56 and S79), while the others differed from one another by 1 to 44 nt substitutions (99.8% to 93.4% identity), resulting in 0 to 11 aa changes in the NSP5 protein sequences (100% to 94% identity). Each PCR B product corresponded, as expected, to a sequence encompassing a segment 11 rearrangement. The 5′ part of the amplicon sequence matched the 11S 3′ region, while the 3′ part matched the 5′ region, indicating a typical head-to-tail duplication of the segment 11 sequence. The seven available PCR-A product sequences were identical to the corresponding PCR-B product sequences in the region of overlap. Most importantly, for each sample, the rearranged sequences were 100% identical to the corresponding 3′ and 5′ sequences of their 11S counterpart, with the duplicated sequence carrying the same nucleotide changes. Since it is most unlikely that identical nucleotide changes occur independently at the same location in the standard and duplicated part of the sequence, the nucleotide changes that we observed most probably occurred before the rearrangement event. This indicated that among the 12 samples, independent events of segment 11 rearrangements had occurred in at least 11 distinct rotavirus strains.

Eleven of the 12 rearranged sequences had the same length (580 nt and 406 nt for PCR A and B, respectively) and, strikingly, for all of them, rearrangement took place at the same location (Fig. 6). Compared to the 11S sequence, the sequence was interrupted downstream of the NSP5 ORF stop codon at nt 618 (U), within the short sequence AUGU (nt 615 to 618), and reinitiated at nt 26 (C) just downstream of the same short sequence AUGU (nt 22 to 25), which includes the initiation codon AUG. Thus, the rearrangement consisted of a partial duplication of the sequence potentially leading to a putative 1,257-bp-long segment 11 containing a normal 5′ UTR and NSP5/NSP6 ORF followed by a long 3′ end containing a complete duplication of the NSP5/NSP6 ORF and a 3′ UTR.

FIG. 6.

Identical pattern of segment 11 rearrangement in 11 rotavirus isolates. 11Ss and putative 11Rs are shown. For 11R, location of PCR A and B products that were sequenced are indicated by arrows. The ORFs are represented by large boxes and the UTRs by small boxes. Nucleotide sequences involved in the rearrangement are detailed, and the short direct repeat AUGU is boxed in. Numbers in parentheses indicate the samples which have the corresponding sequence: (1) S22, S33, S56, S79, S111, S143, S160; (2) S7, S35; (3) S142; and (4) S125. The duplicated sequence generated by the nt 618 to nt 26 rearrangement in 11R is shaded.

Only 1 of the 12 rearranged sequences (sample S141) had a shorter length (526 nt and 352 nt for PCR A and B, respectively) and corresponded to a different pattern of rearrangement, taking place upstream of the stop codon of the NSP5 ORF. The sequence was interrupted in the NSP5 ORF at nt 588 (C), within the short sequence ACAAGUC (nt 582 to 588), and reinitiated at nt 50 (U) in a different frame (+1), just downstream of the same short sequence ACAAGUC (nt 43 to 49) (Fig. 7). Interestingly, this direct repeat was present only in this particular isolate. This rearrangement should potentially lead to a putative 1,203-bp-long segment 11 with a normal 5′ UTR, a modified NSP5/NSP6 ORF, and a 307-nt-long 3′ end containing the duplicated last 258 nt of the NSP5 ORF followed by the 3′ UTR. In this case, the rearrangement which links nt 588 to nt 50 within the NSP5 ORF resulted in a frame shift (+1) corresponding to the NSP6 frame. Consequently, the rearranged gene 11 sequence should encode a modified NSP5 protein consisting of the first 189 aa of NSP5 (nt 22 to nt 588), followed by the 10-aa sequence “FPQFLLVSLK” resulting from the translation in the NSP6 frame of the duplicated sequence (nt 50 to nt 79) and ending with the 92 aa of NSP6 (duplicated sequence nt 80 to nt 355).

FIG. 7.

Pattern of segment 11 rearrangement in sample S141. Sample S141 11S and putative 11R are shown. For 11R, the locations of PCR A and B products that were sequenced are indicated by arrows. The ORFs are represented by large boxes and the UTRs by small boxes. Nucleotide and amino acid sequences involved in the rearrangement are detailed, and the short direct repeat ACAAGUC is boxed. The duplicated sequence in 11R is shaded. The rearrangement which links nt 588 to nt 50 within the NSP5 ORF results in a frame shift (+1) corresponding to the NSP6 frame. NSP5m indicates the modified protein potentially encoded by 11R. NSP5m consists of the first 189 aa of NSP5, followed by a 10 aa polypeptide (black box) resulting from the translation in the NSP6 frame of the duplicated sequence (nt 50 to nt 79), and ending with the 92 aa of NSP6.

Taken together, these results strongly suggest that segment 11 rearrangements do not occur at random within the sequence, but rather at specific sites, probably favored by short direct repeats.

DISCUSSION

To our knowledge, this is the first report showing that gene 11 rearrangements are detectable quite frequently (7.5%, according to our technique) in the course of acute rotavirus infection in immunocompetent children. Until now, rotavirus gene rearrangements have been detected only during specific circumstances of viral replication (9, 18, 19, 28, 29), namely, chronic infection in immunodeficient children or cell culture at a high MOI (≥1,000 PFU per cell). Our results indicate that the detection of rearrangement events is not restricted to such circumstances and strongly suggest that the ability to generate gene rearrangements is a general feature of the viral RNA polymerase.

However, if gene rearrangements—at least those concerning segment 11—are actually generated during acute infections, which are far more frequent than chronic infections, one might expect that viruses with a rearranged genome should be regularly encountered among rotavirus field isolates, which is not the case. Two reasons for this apparent paradox can be proposed. First, rearranged forms of segment 11 generated during acute infection are kept in a minority in the viral progeny compared to their standard counterparts. In chronic infection, viruses with rearranged segments progressively overgrow and replace wild-type viruses over time (18, 29); in contrast, the short duration of acute infection allows only a limited number of viral replication cycles to occur, leaving no time to the viral progeny with a rearranged genome to reach a significant level compared to the wild-type viral progeny. Moreover, it is presumed that unusual replication conditions exert a positive selection leading to the recovery of viruses with a rearranged genome over wild-type viruses. Previous experiments have shown that serial propagation in cell culture of wild-type rotaviruses mixed with rotaviruses with a rearranged genome at either a low or a high MOI, results in a viral progeny with a wild-type or a rearranged genome, respectively (17). Our results indicate that the ratio of 11Rs to 11Ss in the viral progeny recovered from the feces during acute infection is most probably <1:100 (sensitivity threshold of Northern blot hybridization) and ≥1:10,000 (sensitivity threshold of RT-PCR). This ratio might be too low to allow rotaviruses with a rearranged genome to emerge during acute infection and to disseminate to new hosts. However, this might sometimes happen, since rotavirus strains with a 11R of unknown origin have been reported to circulate among immunocompetent children in a few epidemics (5, 12). In addition, it has been shown that, in HRV with the so-called “short” or “supershort” electropherotypes, the segment 11 derived from a sequence rearrangement (22, 27). Second, viruses with a 11R produced during acute infection might be defective or have a selective disadvantage compared to wild-type viruses. The failure to recover cell culture-adapted viruses with a 11R from stool samples could favor this hypothesis. However, in all but one case, segment 11 rearrangement led to a segment consisting of a normal 5′ UTR followed by an unmodified NSP5/NSP6 ORF and a long 3′ UTR containing the sequence duplication, and rotaviruses with such a type of rearranged segments have been reported to be not defective (7, 11, 13, 14, 20, 22, 31, 36). Moreover, such rearranged segments can reassort in vitro and replace their normal counterparts structurally and functionally (2, 6, 15). Additionally, one-step growth curve experiments have shown that the porcine rotavirus strain CC86 with a 11R replicates as well as its wild-type counterpart (8, 25). In one case, the rearrangement occurred within the NSP5 ORF, so that the resulting rearranged gene 11 should potentially encode a modified protein consisting of the first 189 aa of NSP5 linked to the 92 aa of NSP6 by a 10-aa polypeptide (FPQFLLVSLK). In that case, the functionality of the modified NSP5 protein as well as the viability of viruses with such a 11R remain highly hypothetical. Rearrangements that modify the ORF have been identified for gene 5 of a bovine rotavirus that encodes a truncated NSP1 protein (16, 35) and for gene 7 of a HRV that encodes an elongated NSP3 protein (11). In both cases, the resulting viruses had retained their capacity to grow in cell culture, although viruses expressing the modified NSP1 protein gave lower yields and formed smaller plaques than the wild-type virus (35).

Remarkably, all but one of the segment 11 rearrangements that were detected in the course of acute infection shared the same rearrangement pattern. This strongly suggests that rearrangements do not occur at random sites within the sequence. The mechanism for genome rearrangements is not yet elucidated, and different models have been proposed (9, 24). Current hypotheses suggest that the viral RNA-dependent RNA polymerase may jump back on its template during either the transcription (plus-strand synthesis) (20, 24) or the replication (minus-strand synthesis) (9, 11) step. Previous studies have reported preferential sites for gene rearrangements suggesting a nonrandom mechanism (11, 19). Recently, Alam et al. (1) reported that three independent virus clones obtained in vitro have an identical rearrangement of the NSP3 gene, suggesting a preferential site for gene 7 rearrangement. In some cases, direct repeats that might favor the polymerase switch at a specific location on the sequence are found close to the rearrangement site (3, 19, 20, 32). In our case, all but one of the segment 11 rearrangements were located within the same short direct repeat AUGU that was present in all segment 11 sequences. Interestingly, for the unique case with a different rearrangement pattern, the rearrangement occurred within the direct repeat ACAAGUC that was present only in this particular segment 11 sequence and not in others. These direct repeats might correspond to hot spots for RNA recombination, as reported for poliovirus and other RNA viruses (see reference 21 for a review). This might also explain why some RNA segments, such as segment 11, could rearrange more frequently than others, depending on the presence and the location of direct repeats in the RNA sequence. However, direct repeats are not an absolute requirement for genome rearrangements, since several reports have described rearranged segments for which no direct repeat could be found close to the duplication site (7, 11, 19, 22, 23). In that case, secondary structures between the 5′ and 3′ ends of the mRNA have been proposed as an alternate way of facilitating and directing the transfer of the RNA polymerase at a specific site from the 5′ to the neighboring 3′ end of the mRNA template (11). Moreover, it should be considered that after the initial event of rearrangement, rearranged genes may evolve rapidly during subsequent cycles of viral replication, leading to deletions in the noncoding duplicated sequence that may conceal the initial location of the rearrangement (11, 14, 23, 28). Thus, some of the rearranged segments previously characterized, and particularly those recovered from viruses isolated from chronically infected children, might be the result of a multiple-step process. Conversely, during acute rotavirus infection, viruses have only a short time to undergo a genetic drift. Thus, the rearranged sequences that are detected are most likely related to the initial rearrangement event and are probably the most relevant ones upon which to draw hypotheses on rearrangement mechanisms. The predicted secondary structures formed by the complementary folding of the 5′ and 3′ ends of the segment 11 RNA could not explain the rearrangements that we observed in this study. Conversely, the sequences contained direct repeats. Thus, our results clearly support the model according which direct repeats might favor the polymerase switch at a specific site during the transcription step.

In conclusion, we demonstrate that viral replication steadily generates rearranged forms of segment 11 during acute rotavirus infection. Moreover, the similitude of the rearranged sequences suggests that rearrangements do not occur at random sites. Further studies are needed to determine whether such nonrandom rearrangements can occur for other dsRNA segments during acute rotavirus infection or are restricted to RNA segments containing direct repeats suitable for the polymerase to fall back on its template.