Of the 10 cases from the 2008 outbreak in Bangladesh, 5 were confirmed positive for NiV infection by at least 1 laboratory test at CDC. Of those 5 positive cases, 4 were positive for IgM; 2 for IgG; 3 by rRT-PCR; and 2 by conventional RT-PCR; 2 throat swab samples yielded live NiV, 1 from Manikgonj (NiV/BD/HU/2008/MA [BD, Bangladesh; HU, human]; accession no. JN808857) and 1 from Rajbari (NiV/BD/HU/2008/RA; accession no. JN808863) (). Despite the isolates having come from patients from 2 districts, sequence analysis of the entire genome of the 2 isolates indicated that they were identical. Phylogenetic analysis of ORFs from each NiV gene indicated that this strain was similar to, but distinct from, the 2007 isolate from India (NiV/IN/HU/2007/FG [IN, India]; accession no. FJ513078) (, panel A; , panels A–E). To rule out the possibility of laboratory contamination, we performed 2-step conventional RT-PCR by using RNA from duplicate samples of the original throat swab samples from which the 2 viruses were isolated. We amplified the entire N gene ORF from each sample and confirmed that the sequences were identical. Although there were 4 isolated cases of NiV infection in Bangladesh in 2009 as confirmed by IgM or IgG ELISA, or both, we were not able to obtain NiV sequences from those case-patients ().
Results from patients with confirmed Nipah virus infection, Bangladesh, 2008–2010*
Figure 1 Phylogenetic analyses of sequences from the complete Nipah virus N ORF (A) and the 729-nt proposed N ORF genotyping window (B). Tree created with maximum parsimony, close-neighbor-interchange algorithm, 1,000 bootstrap replicates (16). Branch lengths (more ...)
Figure 2 Phylogenetic analyses of sequences from the complete Nipah virus P ORF (A), M ORF (B), F ORF (C), G ORF (D), and L ORF (E). Tree created with maximum parsimony, close-neighbor-interchange algorithm, 1,000 bootstrap replicates (16). Branch lengths are (more ...)
Of the 17 cases from the 2010 outbreak in Bangladesh, 12 were confirmed positive. All 12 were positive for IgM, 2 for IgG, 5 by rRT-PCR, and 3 by conventional 2-step RT-PCR (). Although we detected NiV RNA by rRT-PCR from urine, CSF, and throat swab samples, we were unable to isolate virus from any of those sources. Of the 3 samples from which we were able to amplify NiV sequences, 1 was from a 10-year-old girl from the initial cluster (NiV/BD/HU/2010/FA2; accession no. JN808859) and the other 2 were from patients with isolated cases. The patients with isolated cases were a medical intern (NiV/BD/HU/2010/FA1; accession no. JN808864) who was working in the pediatric department at Faridpur Medical College Hospital and a 7-year-old girl (NiV/BD/HU/2010/GO; accession no. JN808860) who was examined by the same medical intern; both died. The illness in the intern developed only 6 days after the 7-year-old girl died; this incubation period was atypically short for NiV infection, indicating the possibility of separate infections (6
). Sequence analysis of the N ORFs amplified from throat swab samples confirmed that the intern and the girl were infected with distinct lineages of NiV (, panel A). Our attempts to recover NiV sequences from prior contacts of the medical intern who were IgM positive for NiV infection were unsuccessful. Phylogenetic analysis indicated that the sequence from the 7-year-old girl was similar to, but distinct from, the 2007 isolate from India, whereas the sequence from the intern was closer to that of the 2004 isolate from Bangladesh (NiV/BD/HU/2004/RA1; accession no. AY988601). The N sequence obtained from the 10-year-old girl from the initial cluster was shown to be slightly more similar to the 2007 isolate from India (, panel A). We were only able to amplify the complete N ORF from the throat swab samples from the 7-year-old and 10-year-old girls because our rRT-PCR indicated the presence of ≈103
copies of NiV N RNA (cycle threshold ≈26–30). The rRT-PCR conducted on the throat swab sample from the medical intern indicated the presence of ≈106
copies of NiV N RNA (cycle threshold ≈20), which corroborated our ability to amplify nearly the entire genome except for the 3′ leader and 5′ trailer (data not shown) from this sample.
Since the initial molecular characterization of NiV from Bangladesh in 2004 (11
), there has been a shortage of full-length NiV ORF sequences from Bangladesh. However, the sequence data obtained from the 2008 and 2010 Bangladesh outbreaks in this study, along with the recent characterization of the 2007 isolate from India (4
), support the previous observation of relative heterogeneity among NiV nucleotide sequences from humans affected by outbreaks in Bangladesh compared with sequences from Malaysia (11
). Phylogenetic analysis indicated that these new sequences from Bangladesh and India group substantially closer to the sequences from Bangladesh in 2004, which led us to propose a system to describe the distinct lineages of NiV (; , panels A–E). We propose to designate the current sequences obtained from Malaysia (MY) and Cambodia (KH) as genotype M and the sequences obtained from Bangladesh and India as genotype B. By using a 729-nt window in the N terminal region of the N gene ORF (N ORF nt 123–852, NiV genome positions 236–964), we were able to determine 25 distinct nucleotides that universally differentiated the genotypes (, panel B). The topology of the phylogenetic tree and the positions of the branches generated from this smaller nucleotide window were similar to those of the tree generated with the full-length N ORF sequences and have reasonably high bootstrap values at the root branch junctions, albeit with lower bootstrap values at the distal branch junctions (, panels A, B). In support of this scheme, we observed similar topologies and branching patterns in phylogenetic trees generated for the complete P, M, F, G, and L ORFs, all with strong bootstrap values (, panels A–E).
Pairwise sequence comparisons conducted across each individual NiV gene ORF indicated a nucleotide variation range of 6.32%–9.15% between genotype M and B viruses and an amino acid variation range of 1.42%–9.87% (; Technical Appendix
). The ranges of nucleotide and amino acid variation of sequences within genotype M were 0.19%–2.21% and 0.18–3.67%, respectively, and within genotype B were 0.28%–1.06% and 0.28% –0.56%, respectively. The apparently higher levels of variation found among ORFs within genotype M is mostly caused by the comparatively divergent sequences obtained from Pteropus vampyrus
(NiV/MY/BA/2010/MY; accession no. FN869553) and P. lylei
(NiV/KH/BA/2004/KHM; accession nos. AY858110, AY858111) bats. Not only is the proposed genotyping scheme supported by consistent phylogenetic tree topologies, but pairwise nucleotide comparisons of the 729-nt region yield similar percentages of variability as seen in the full-length N ORF comparisons. This finding indicates that this sequence window is a relatively accurate indicator of overall nucleotide variability within and across genotypes M and B (Technical Appendix Figure 1, panels A, C
Percentage nucleotide and amino acid variability among available complete Nipah virus open reading frame sequences
A comprehensive amino acid alignment of currently available complete N protein ORFs indicated that the 4 residues that distinguish between genotype M and B viruses are almost all located in the COOH-terminus (). Of these residues, only 1 (position 387) is located within the putative minimum contiguous sequence required for capsid assembly (17
), and none were located in the 29 COOH-terminal and 10 NH-terminal residues required for interaction with the P protein (18,19
). Of note, there are 4 residues (V429, E432, D457, and T521) in the COOH-terminal region common to all genotype B sequences that are shared with 2 of the comparatively divergent genotype M sequences. In light of the overall nucleotide and amino acid sequence comparisons, however, the divergent genotype M sequences from the bats still differ substantially from genotype B sequences (, panel A; , panels A–E; Technical Appendix Figure 1, panels A, B
Amino acid differences among available complete Nipah virus N gene open reading frame sequences*
Amino acid alignments of the P protein indicated numerous differences between genotype M and B sequences in the first 400 residues, which comprise the shared N terminal region between the P, V, and W proteins. Of the differences in this region, there were neither changes that would be predicted to alter the STAT-1 binding ability of the P, V, and W proteins nor changes that could adversely affect RNA replication (20–22
). There were only 4 changes in the COOH-terminal region of P, which is required for direct N–P interactions, 2 of which were nonconservative changes (N590→S, E635→G) (18
). The P sequence derived from P. vampyrus
bats had an intriguing sequence of amino acids from residues 408–440, in which there was substantial sequence divergence from genotypes M and B at the nucleotide and amino acid levels (23
). These particular nucleotide changes in the P sequence also introduced several amino acid changes in the unique COOH-terminal regions of the V (11 changes) and W (9 changes) ORFs, which distinguish them from any genotype M and B sequences.
We observed the M protein to be highly conserved across genotypes M and B, and we found just 2 aa residues exclusive to genotype B that are not located in any region of the protein with a known function, such as budding (24,25
), nuclear localization, or ubiquitination (26
). In the F protein, the predicted cleavage site, F1 amino-terminal domain, transmembrane domain, and predicted N-linked glycosylation sites are all conserved across both genotypes. Although the percentage of amino acid variation in the G protein is higher than that in all other NiV proteins (except the P protein), it is not surprising that the residues implicated in Ephrin B2 and B3 binding are conserved across the genotypes (27,28
). The amino acid differences between genotypes M and B sequences are predominantly found at residues that are distant from the receptor binding site. Two differences were found in the intracellular domain, 3 differences in the stalk region (positions 72–182), 9 differences in a span of ≈100 aa (positions 236–344) along the side of the globular head domain, 4 differences closer to the top of the globular head domain (positions 385–424), and only 2 differences (positions 498 and 502) that were close to the tryptophan residue at position 504, which is part of the receptor-binding pocket. As in other NiV proteins, several amino acids were shared between genotype B sequences and 2 genotype M sequences derived from the bat isolates. The significance of these changes has yet to be explored.
The level of amino acid conservation throughout the L proteins was high; the purported GDNE catalytic site and the K-X21
-GEGSG ATP binding site were conserved across genotypes M and B. Most distinct differences between genotypes M and B sequences (26 of 32) were located outside the 6 linear domains typically found in nonsegmented negative strand virus polymerases (29,30
). The cis-acting control sequences in NiV are usually well conserved. The tri-nucleotide intergenic sequences amplified from the throat swab sample from the medical intern in 2010 had GAA for all 6 intergenic regions, which was identical to the 2007 isolate from India. For the 2008 isolates, the intergenic sequence between the N and P genes was AAA, and the rest of the intergenic sequences were GAA. The biological implications of finding adenosine in the first position of NiV intergenic sequences is unknown. The 3′ leader and 5′ trailer sequences of the 2008 NiV isolates were identical to those found in the 2004 Bangladesh and 2007 India isolates.