Stool samples from 311 children hospitalized with acute gastroenteritis at Sheba Medical Center, Tel Hashomer, Israel, during May 15, 2005–May 15, 2006, were suspended in saline. Rotavirus dipsticks (Hy Laboratories LTD, Rehovot, Israel) were used to identify rotavirus-positive stool samples. (For comparison with rotavirus ELISA with Dako reagents [Dako Diagnostics LTD, Glostrup, Denmark], diagnostic dipsticks from Hy-Laboratories and Novamed [Novamed, Jerusalem, Israel] had sensitivities of 98.5% ± 3% and 92.3%, and specificities of 92.5% ± 1% and 93.5%, respectively [L.M. Shulman et al., unpub data].) The section of the dipstick containing the rotavirus-specific bands from 14 rotavirus-positive dipsticks that had been stored at room temperature for at least 1 week were excised and placed into 120 µL saline.

Rotaviral RNA was extracted from these rotavirus-specific bands and from the corresponding saline stool suspensions by using QIAmp Viral RNA Mini kits (QIAGEN, Valencia, CA, USA). RNA amplified from rotavirus-specific bands from dipsticks archived at room temperature 5 years earlier was similarly extracted. RNA was amplified by using generic primers for reverse transcription–PCR (RT-PCR) (QIAGEN OneStep RT-PCR kits, QIAGEN) followed by heminested PCR (AmpliTaq Gold, Applied Biosystems, Foster City, CA, USA) with genotype-specific primers to identify G and P genotypes as described by Gouvea et al. (11) and Gunasena et al. (12), respectively. RT-PCR products were gel purified (QIAQuick Gel Extraction Kit, QIAGEN) or purified directly (HiPure PCR Product Purification Kit; Roche Applied Science, Indianapolis, IN, USA) from the PCR mix and sequenced by using an ABI PRISM Dye Deoxy Terminator cycle sequencing kit (Applied Biosystems) and the genotype-specific primer and common primer from the heminested PCR reaction described. Reaction mixtures were analyzed on Applied Biosystems model 373 DNA automatic sequencing system to confirm the RT-PCR genotyping results and for phylogenetic analysis.

Sequences of the same genotype were truncated to the longest common length by using the Sequencher program (Genecodes, Ann Arbor, Michigan, USA). Nearest-neighbor phylogenetic trees were constructed from these sequences after the data were bootstrapped 1,000 times with Clustal X (13). The tree was analyzed by using NJplot (14). Sequences have been deposited in European Molecular Biology Laboratory/GenBank/DNA Data Bank of Japan under accession nos. FN298858–FN298875, FN582119–FN582124, and HQ174462–HQ174463. (Isolate names are linked to their specific accession numbers in Table A1).

This study was approved by the institutional review board of Chaim Sheba Medical Center (SMC-7606-09). Experiments were not performed on humans. All personal identification was removed from the remnants of fecal samples sent to the National Center for Viral Gastroenteritis for rotavirus analysis. The viral RNA used for this study was obtained from these anonymous samples.

To assess the suitability of RNA extracted from dipsticks for determining the VP7 genotype, we first determined the VP7 G genotype of rotavirus in rotavirus-positive clinical samples from children hospitalized during May 15, 2005–May 15, 2006. Of 311 stool samples, 61 were determined to be rotavirus positive by using diagnostic rotavirus dipsticks. The G genotype and P genotype of RNA in 54 of the 61 samples was determined by the size of the heminested PCR amplification product as described. Mixed infections, e.g., >1 G (6 samples) and P genotypes (2 samples) in the same fecal suspension, were found in 8 (15%) of the 54 positive samples. Taking into account mixed infections, we found G1, G2, G3, and G9 G genotypes in 79%, 5%, 2%, and 15%, respectively, of the 62 rotaviruses identified in the 54 clinical samples. The frequencies of associated G and P genotypes for isolates from the 2005–2006 rotavirus season were 3% G1P[4], 76% G1P[8], 5% G2P[4], 2% G3P[8], 3% G9P[4], and 11% G9P[8].

We chose 18 rotavirus-positive clinical samples with which to study the feasibility of recovering molecular information from dipsticks. Thirteen were from children infected with 1 G genotype of rotavirus and 5 from children simultaneously infected with rotaviruses belonging to 2 different G genotypes. Of the 13 infections by a single G genotype, 8 were G1, 3 were G2, and 1 each were G3 and G9. All 5 mixed infections, RoV-24_ISR05, RoV_41_ISR05, RoV_56_ISR06, RoV_57_ISR06, and RoV_60_ISR06, were simultaneous infections with G1 and G9 genotypes. Mixed infections were identified by specific product size on gels and confirmed by partial sequencing of the G1 product for all 4 patients whose G9 sequences appear in the Figure.

RNA was extracted from rotavirus-positive bands of dipsticks from the 18 clinical samples. The dipsticks had been stored for >1 week at room temperature. The quality of the RNA from all 18 extractions was sufficient to enable determination of VP7 G genotypes by heminested RT-PCR. The genotypes obtained from the RNA extracted from dipsticks were identical to those from RNA extracted from corresponding saline suspensions for 14 of the 18 rotavirus-positive stool samples, including 3 samples from children with dual infections. In 2 mixed G1–G9 infections, G9, but not G1 RNA, was recovered, and G1 RNA was recovered from dipstick RNA in addition to G9 from a G9-infected child. A fourth dipstick from a G3 sample yielded a G3 and an equivocal G1 that did not appear in subsequent replicate heminested amplifications. In other words, 21 (92%) of 23 genotypes identified from RNA extracted from 18 fecal suspensions were also identified from RNA extracted from dipsticks, including 8 of 10 genotypes from 5 samples simultaneously infected with rotaviruses of 2 different genotypes.

Rotaviral VP7 RNA suitable for genotyping was also recovered from all 5 air-dried rotavirus-positive dipsticks that had been stored at room temperature for 5 years. The G genotypes of the rotavirus in the fecal suspensions from the 5 archived dipsticks from 2002 were unknown, and the suspensions were no longer available for comparison. One of the dipsticks was manufactured by Hy Laboratories; the remaining 4 were manufactured by Novamed. Initial amplification with generic G-type primers yielded generic G genotype amplification products of expected size (≈1,062 nt) upon gel electrophoresis. Three of the dipsticks yielded G3 RNA; the other 2 yielded G1 RNA. As proof of concept that P genotypes could be determined in addition to G genotypes, the P genotypes of RNA recovered from 2 of the 5-year-old rotavirus dipsticks with G3 RNA were also determined by direct sequencing of the RT-PCR products by using the generic P-type primers. Both were P[8] by sequence analysis (Table A1).

The G genotypes of 3 G1s, 2 G2s, 1 G3, and 1 G9 RNA extracted from dipsticks were confirmed by sequencing their heminested amplification products. Moreover, the sequences of the RNA extracted from each of the 7 dipsticks were identical to the sequence of the RNA extracted directly from the corresponding stool suspension. The lineages and sublineages of these and other isolates from the 2005–06 season were inferred from phylogenetic comparisons to equivalent segments of the 11 lineages of G1, the 2 lineages of G2, the 4 lineages of G3, and the 6 lineages of G9 described by Phan et al. (15), Page and Steele (16), Wang et al. (17), and Martinez-Laso et al. (18), respectively. The G3 phylogenetic tree also includes 3 sequences obtained from the 5-year-old dipsticks and 4 G3 isolates from among those with highest homology to G3 isolates from Israel identified in BLAST searches (http://blast.ncbi.nlm.nih.gov/Blast.cgi) (19). Specifically, all VP7 sequences for a given G genotype were aligned and truncated to the longest common segment. Four separate phylogenetic trees (Figure) were constructed because of the size difference between G genotype–specific heminested amplification products. The lineage or sublineage of each isolate from Israel (Table A) was inferred from the relative difference of each sequence compared with equivalent segments of previously determined lineages. For each genotype, a new lineage or sublineage was suggested when the sequence from Israel failed to closely group with the equivalent segment of >1 reference strains and the nucleotide differences between the strain from Israel and the reference strains were similar or greater than the differences among reference lineages and sublineages. The existence of such new lineages and sublineages would need to be confirmed with longer full-length open reading frame sequences. Finally, sequences from RNA recovered from dipsticks and/or from saline suspensions were used in BLAST searches to identify the most similar contemporary sequences. G1 isolates were most similar to isolates from Europe, the Far East, and South America; G2 isolates to isolates from the Far East; and G3 isolates to isolates from Europe and the Far East (Figure). G9 isolates were most similar to isolates from Africa, Europe, and the Far East.

The G1P[8] genotype predominated in central Israel during the 2005–06 rotavirus season as it did throughout Israel during 1991–1994 (20) and in northern Israel during 2007–2009 (2,10). However, the relative distributions of other genotypes differed. For example, G4P[8], absent during 2005–2008, was present in 32.3% of samples a decade earlier and reappeared in 2008–09 (2). Conversely, G9 genotypes, absent a decade earlier, were present in 15.7% of the 2005–06 isolates and 9.3% of the 2006–07 isolates. This substantial prevalence of G9 mirrors the global emergence of G9 among hospitalized children in the mid-1990s (21,22). In addition, double infections indicated by >1 G or P genotype rose from <2% during 1991–1994 (20) to 16.4% for the isolates in this study and 27% during 2007–08 (2). As indicated by Muhsen et al (2), evidence is good for rotavirus reassortants emerging in vivo from appropriate double infections. Phylogenetic analysis in the present study suggested that some G2 and G9 rotaviruses from Israel belonged to new sublineages or lineages, respectively.

In conclusion, air-dried affinity-concentrated virus from rotavirus-positive bands of dipsticks yielded RNA suitable for G genotyping and sequencing, even for dipsticks stored at room temperature for 5 years. Even though P genotyping and sequencing was performed only on RNA isolated from 2 of the 5-year old dipsticks, it is reasonable to assume that P genotyping and sequencing would have been possible for the rest based on the equivalent quality of the RNA extracted from the dipsticks and RNA extracted from stool suspensions. Thus, dipsticks can be used for recovery of molecular epidemiologic data by reference laboratories from diagnostic tests routinely performed in many clinical laboratories and point-of-care facilities. RNA extracted from these archived dipsticks will enable centralized laboratories to easily recover epidemiologically useful data from the samples of other laboratories and from regions lacking adequate laboratory facilities. In addition, centralized laboratories will be able to assess whether the introduction of universal rotavirus vaccination changed the distribution of rotavirus genotypes associated with severe rotavirus-associated acute gastroenteritis.