Each year, >100,000 diarrhea patients are admitted to the Dhaka hospital of the International Centre for Diarrheal Disease Research, Bangladesh (ICDDR,B). Fecal samples from 2% of these patients are collected and examined as part of systematic routine surveillance system for the presence of enteric pathogens (11). All procedures were performed in compliance with relevant laws and institutional guidelines and in accordance with the Declaration of Helsinki.

Stool specimens from a subset of patients from routine surveillance during 2007–2009 (1,614 samples total) were available for further studies. These specimens were prescreened for adenovirus and rotavirus A by using TaqMan EZ RT-PCR Core Reagents (Applied Biosystems, Foster City, CA, USA), rotavirus primers RVNSP3R and RVNSP3F and probe 5′-FAM-AGTTAAAAGCTAACACTGTCAAA-TAMRA-3′ (12), and TaqMan Universal Mastermix (Applied Biosystems) (13). Sequence-independent nucleic acid amplification and next-generation sequencing were performed on 105 stool specimens from diarrhea patients enrolled during 2007 by using a 454 GS Junior Instrument (Roche, Indianapolis, IN, USA) as described (6,7). More than 725,000 trimmed reads were assembled by using de novo assembly and analyzed according to BLAST searches (Technical Appendix Figure 1 and Table 1) (6,7). Sequences were classified on the basis of the taxonomic origin of the best-hit sequence (6,7). An E (expect) value of 0.001 was used as cutoff value of significant virus hits. The largest proportion of virus-related sequences in human diarrhea samples from Bangladesh in 2007 was related to known bacteriophages and mammalian viruses (Technical Appendix Figure 1).

One sample, no. 289, yielded a novel mammalian virus from the family Caliciviridae that we further characterized by near full-length genome sequencing using random amplification with next-generation sequencing, specific reverse transcription PCRs, and 3′ rapid amplification of cDNA ends PCR (Figure 1, panel A) as described (6,7). We named the virus isolate calicivirus Bangladesh/289/2007 (GenBank accession no. JQ745645).

The Caliciviridae genome encodes a polyprotein precursor for nonstructural proteins, and 2 structural capsid proteins, viral protein (VP) 1 and VP2, from 2 or 3 open reading frames (ORFs) (Figure 1, panel B) (8,9). The genome organization of Bangladesh/289/2007 is most closely related to that of viruses in the genus Norovirus, with ORF2 encoding VP1 overlapping with ORF1 (Figure 1). The partial polyprotein precursor and complete VP1 and VP2 proteins were aligned with corresponding sequences of representative caliciviruses. Divergence analysis demonstrated that the calicivirus Bangladesh/289/2007 showed 57.7%, 49.5%, and 53.1% amino acid identity in the polyprotein, VP1, and VP2, respectively, to its closest relative, Recovirus Tulane (GenBank accession no. EU391643), which was identified from rhesus macaques (Table 1; Figure 2).

Additionally, on the basis of the identity profiles of noroviruses (14)—strains, genotypes, and genogroups showed 85.9%–100%, 56.2%–85.7%, and 38.6%–55.1% identity, respectively—Bangladesh/289/2007 may represent a new genogroup in the proposed genus Recovirus. This conclusion was corroborated by comparison of a partial RNA-dependent RNA polymerase (RdRp) sequence of calicivirus Bangladesh/289/2007 to known recovirus RdRp sequences (Technical Appendix Figure 2).

We performed a diagnostic real-time recovirus PCR targeting the RdRp of all 1,614 available samples from patients with diarrhea (11). Reverse transcription PCR–grade viral nucleic acid was extracted by using the MagNA Pure LC Total Nucleic Acid Isolation Kit (Roche) and amplified by using reverse transcription PCR with primers VS665 and VS666 and probe VS664 (Technical Appendix Table 2) and TaqMan EZ RT-PCR Core Reagents (Applied Biosystems). The cycling program consisted of 50°C for 2 min, 60°C for 30 min, 95°C for 5 min, and 50 cycles of 95°C for 20 s and 59°C for 1 min, resulting in a 164-bp amplicon.

In addition to sample 289, 5 other human diarrhea samples (Table 2) were sequence-confirmed to be positive for a recovirus, with high homology (>98%) to calicivirus Bangladesh/289/2007; this finding indicates that recovirus Bangladesh infects humans. Clinical data indicate that all patients with a recovirus present in their feces had 6 to >21 watery stools in the first 24 hours after illness onset (Table 2); 4 patients experienced vomiting and 2 patients had fever. Patient ages ranged from 3 months to 50 years. Three recovirus-positive patients showed evidence of co-infection with other pathogens that are known to cause diarrhea in humans, such as rotavirus A, adenovirus, Vibrio cholerae, or Salmonella spp.; the other 3 patients did not. Although viruses such as norovirus, sapovirus, and astrovirus were not detected in the recovirus-positive samples by sequence-independent amplification assays, all samples were not analyzed for all known enteric pathogens. Of 514 diarrhea samples gathered by the Diagnostic Unit, Department of Virology, Erasmus Medical Center, Rotterdam, the Netherlands, for diagnosis of gastrointestinal infections during 2007 and 2009, none was positive for Recovirus Bangladesh (data not shown).

For a large proportion of human diarrhea cases, no etiologic agent can be identified, despite multiple metagenomic studies of viruses in human stool aimed at identifying new etiologic agents (3–5). In addition, it cannot be known when and where emerging and reemerging viruses will appear in the human population. To identify potential etiologic agents of diarrhea in humans, we performed a metagenomic viral inventory in diarrhea samples from Bangladesh, which led to the identification of a novel calicivirus.

Although no species demarcation criteria have been defined for the family Caliciviridae by the International Committee on the Taxonomy of Viruses, we classified calicivirus Bangladesh/289/2007 in the proposed genus Recovirus, primarily on the basis of phylogenetic analyses (8). Bangladesh/289/2007 may also be considered a new genogroup of the genus Recovirus for the following reasons: 1) the genetic distance between Recovirus Tulane and Bangladesh/289/2007 VP1 is similar to that of VP1 capsids of noroviruses belonging to different genogroups (14); 2) the genetic distances between macaque recoviruses and Recovirus Bangladesh/289/2007 RdRP are similar to that of RdRP sequences of recoviruses belonging to different genogroups (15); 3) the genome organization of Recovirus Tulane and Bangladesh/289/2007 differs; and 4) Recovirus Tulane and Bangladesh/289/2007 were identified in different host species.

In conclusion, this identification of a novel calicivirus, classified as Recovirus Bangladesh/289/2007, from human diarrhea samples provides PCR-based evidence that recoviruses can infect humans. A previous study found high prevalence of virus-neutralizing antibodies against a closely related calicivirus, Recovirus Tulane, in serum samples from animal caretakers (15). Larger epidemiologic studies using genetic and serologic screening will be necessary to provide more insight into the distribution and pathogenic potential of recoviruses in humans.