Abstract

The genome of human adenovirus (HAdV) D30 was sequenced in depth. Sequence assembly and analysis revealed two distinct viral sequences with identical hexon genes, which were the same as the one previously reported for HAdV-D30. However, one of the two viruses was found to be a recombinant of HAdV-D29. Exclusive reliance on serum neutralization can lead to mischaracterization of adenoviruses and miss coinfections. Whole-genome sequencing remains the gold standard for proper classification of HAdVs.

TEXT

Human adenoviruses (HAdVs) belong to the genusMastadenovirus within the family ofAdenoviridae and cause a wide array of infections, including those of the respiratory, gastrointestinal, and ocular surface mucosas (4,5,29). Since the first adenovirus was characterized, over 60 types have been formally recognized in GenBank and classified into seven species (A to G) based on serology, whole-genome sequencing, and phylogenomics. Previously, two serological methods were employed to determine the identification of a HAdV type: serum neutralization and hemagglutination inhibition. Serum neutralization is based on recognition of specific epitopes on the viral capsid hexon protein, which is encoded by two small regions on the hexon gene, accounting for approximately 2.6% of the total genome (3,13,28,30), while hemagglutination inhibition is based on an interaction with the HAdV fiber knob. Recently, whole-genome analysis of HAdVs has shown that serum neutralization, by relying on a small fraction of the viral genome as a means of virus identification, can be misleading (21,27,28). For example, HAdV-D53 isolates were originally mistyped as HAdV-D8, -D22, or -D37 on the basis of serum neutralization or limited sequencing (7).

As part of an effort to characterize all the remaining unsequenced HAdV prototypes, HAdV-D30 was obtained from the American Type Culture Collection (ATCC; Manassas, VA;http://www.atcc.org/) and sequenced. Virus stock was grown in A-549 cells (CCL-185; ATCC), purified by a CsCl gradient, and sequenced on a Roche 454 DNA sequencer (Branford, CT) by Operon (Eurofins MWG Operon; Huntsville, Alabama) to at least 17-fold depth (Next Gen) with an accuracy of greater than 99% (Q20 or better). The sequencing reads were assembled using CLC Genomics Workbench (http://www.clcbio.com/index.php?id=1240), with anN₅₀ average of 5,260 bp. TheN₅₀ metric is a measure of the quality of a draft genome sequence;N₅₀ is the length of the sequence such that using equal-length or longer contigs yields half the bases of the genome. Annotation was performed using a custom annotation engine (J. Zaitshik, T. Schmidt, and D. W. Dyer, unpublished data) and the Genome Annotation Transfer Utility (26), with confirmation from NCBI's open reading frame (ORF) finder (http://www.ncbi.nlm.nih.gov/projects/gorf/). Artemis (http://www.sanger.ac.uk/resources/software/artemis/) was used to evaluate the data (2,22). Open reading frames were BLAST analyzed against GenBank sequences for confirmation and protein similarity. Splice sites were predicted using the GenScan web server at MIT (http://genes.mit.edu/GENSCAN.html).

Deep sequencing and assembly of the ATCC stock of HAdV-D30 revealed two distinct viral genomes. These two genomes (human adenovirus 30 strain human/USA/BP-7-1/1959/30 [P30H30F30], designated HAdV-D30, and human adenovirus 63 strain human/USA/BP-7-2/1959/63[P30H30F29], designated HAdV-D63) were aligned using mVISTA Limited Area Global Alignment of Nucleotides (LAGAN;http://genome.lbl.gov/vista/lagan/submit.shtml) (15,17,19,20). HAdV-D8 was included as a negative control. mVISTA performs a global pairwise alignment between two sequences side by side, reflecting the percentage of sequence conservation in the height of each data point along they axis. Conservation in the two genomes identified in the HAdV-D30 sample from ATCC was observed in the penton base and hexon genes, with divergence beginning at the E3 transcription unit and continuing through the fiber gene (Fig. 1A). Further analysis showed that the coding regions for the hexon protein epitopes responsible for serum neutralization were 100% identical between the sequences and identical to the hexon gene sequence archived in GenBank for HAdV-D30 (GenBank accession no.AB330111) (6). The fiber gene sequence for the first isolate, designated HAdV-30, matched 100% to a partial fiber sequence previously archived in GenBank as HAdV-D30 (GenBank accession no.AJ831473) (13), confirming its identity. Protein identity between the fiber of the proposed HAdV-D63 isolate and the same archived HAdV-D30 fiber sequence was only 55% (data not shown), but identity was 100% with HAdV-D29 (GenBank accession no.AB562587) (8). Notably, by mVISTA, sequence conservation of HAdV-D63 was observed with HAdV-D29 from the E3 region onward (Fig. 1A). Further analysis showed that nucleotide sequence identity between HAdV-D30 and HAdV-D63 from the 5′ end of the genome to the beginning of the E3 region was 99.8% (25,615/25,653 nucleotides), with 11 out of 13 genes being 100% identical at the protein level (data not shown). For the remainder of the genome—from the E3 region to the 3′ end—nucleotide identity between HAdV-D63 and HAdV-D29 was 99.56% (8,989/9,028 nucleotides). The overall nucleotide identity between HAdV-D63 and HAdV-D30 was 97.0%, while the overall identity between HAdV-D63 and HAdV-D29 was 97.2%. For further examination of a possible recombination event as the source of HAdV-D63, Bootscan analysis with whole-genome sequences was performed for the same four viruses using Simplot (http://sray.med.som.jhmi.edu/SCRoftware/simplot/) (11). This software searches for recombination by matching specific sequences in one of at least two sample viruses with sequences present in a reference virus. In HAdV-D63, a single recombination event between HAdV-D30 and HAdV-D29 was identified for nucleotides 26,000 to 34,700, a range which includes the E3, E4, L4, and fiber-coding regions (Fig. 1B). The apparent drop in the Bootscan line between nucleotides 8,000 and 10,500 comes about because high identity between 3 or more sequences in a Bootscan analysis is always interpreted by the software as an absence of recombination and HAdV-D29 and HAdV-D30 show high identity with each other in this region. Together, these data justify identification of HAdV-D63 as a unique virus type, caused by recombination between HAdV-D30 and HAdV-D29. Designation of HAdV-D63 as a unique type was approved by the Human Adenovirus Working Group (http://hadvwg.gmu.edu/) under criteria established in collaboration with GenBank. Allocation of a new type number was based on recognition of a rarely observed recombination event including approximately one-fourth of the genome, likely leading to a new tropism for the virus, based on acquisition of a new fiber protein.

Finally, bootstrap analysis, confirmed with neighbor-joining phylogenetic trees, was performed for all genes of both genomes using MEGA (Molecular Evolutionary Genetics Analysis, version 4.0.2) (25). Phylogenetic analysis revealed that the open reading frames within almost three-fourths of the genomes of HAdV-D30 and HAdV-D63, including E1A, E1B, E2B, L1, L2, L3, and E2A (data not shown) and, in particular, the penton base (Fig. 1C) and hexon (Fig. 1D) genes, are closely related. However, the remaining open reading frames at the 3′ end of the HAdV-D63 genome, specifically E3, E4, L5 (data not shown), and the fiber gene (Fig. 1E), appear closest phylogenetically to those of HAdV-D29, consistent with a recombination between HAdV-D29 and HAdV-D30 in the ontogeny of HAdV-D63.

In summary, a detailed genomics and bioinformatics analysis of HAdV-D30, obtained from ATCC, was performed and revealed two viruses with identical hexon genes, illustrating the power of high-resolution genomic analysis. On the basis of previously deposited partial sequence data for HAdV-D30, the whole-genome sequence of the first isolate should become the HAdV-D30 reference strain. Considering the importance of recombination in adenovirus evolution (12,14,18,21,23,27) and its role in the emergence of new pathogens (21,27,28), identification of the second genome as a novel recombinant genome type is justified (24). The ATCC virus stock was not recognized as two distinct viruses due to exclusive reliance on serum neutralization. On the basis of a prior fiber gene sequence incorrectly archived in GenBank as HAdV-D30 (GenBank accession no.AF447393) (10), we believe HAdV-D63 has been in existence since at least 2002.

Our findings are consistent with previous data in which types HAdV-D19a, -D53, -D54, -B55, and -D56 were individually shown to be unique types generated by homologous recombination events which preserved the hexons of parent viruses with different tissue tropisms (1,8,9,13,16,20,21,27,28). It should be noted that this parallels our recent genomic analysis of HAdV-D15 (unpublished data) which found, with the exception of the hexon open reading frame, a genome sequence quite different from that recently submitted for the same ATCC stock with the same lot number (GenBank accession no.AB562586.1) (8). We conclude that human adenoviruses, particularly within HAdV-D species, readily undergo recombination such that a serotype, as identified by serum neutralization or limited DNA sequencing, can represent two or more distinct viruses. Thus, serology and/or hexon gene sequencing at best provides a low-resolution view and at worst can lead to a mischaracterization of a particular HAdV.

ACKNOWLEDGMENTS

This study was supported by NIH grants EY013124 and P30EY014104, a Senior Scientific Investigator Award (to J.C.) from Research to Prevent Blindness, Inc., New York, NY, and the Massachusetts Lions Eye Research Fund.

Footnotes

REFERENCES