Currently, the only basis for recognizing a new type of HAdV is by serology, as specified in the 2010 report of the ICTV Adenovirus Subgroup:
“Adenovirus serotypes are differentiated on the basis of neutralization assays. A serotype is defined as one that either exhibits no cross-reactions with others, or shows a homologous/heterologous titre greater than 16 (in both directions). For homologous/heterologous titre ratios of 8 or 16, a serotype assignment is made if either the viral haemagglutinins are unrelated (as shown by lack of cross-reaction in haemagglutination-inhibition tests), or if substantial biophysical, biochemical or phylogenetic differences exist. Antigens at the surface of the virion are mainly type-specific. Hexons are involved in neutralization, and fibers in neutralization and haemagglutination-inhibition.”
The 2010 ICTV report specifies the use of “type” for “serotype”:
“The natural host range of adenovirus types (note that ‘type’ is the correct description even if it is based on the earlier serotypes) is usually confined to a single species, or to closely related species.”
The 2010 ICTV report does not mention the use of genomics and complete genome data as a basis for typing. However, it is pointed out that genomics use is implied by the acceptance of DNA sequences representing the serum neutralization epitope (instead of wet bench SN data) and the hemagglutination inhibition epitope (instead of wet bench HI data), as well as the ICTV points under the “species demarcation criteria” (note, the acceptance of partial and limited DNA sequences for typing is not explicitly stated nor does it constitute ‘full genomic analysis’):
1. Phylogenetic distance (>5-15%, based primarily on distance matrix analysis of the DNA polymerase amino acid sequence)
2. Genome organization (characteristically in the E3 region)
3. Nucleotide composition (G+C)%
7. Ability to recombine
8. Number of VA RNA genes
“If virus neutralization data are available, lack of cross-neutralization combined with a phylogenetic distance of more than 15% separates two types into different species. If the phylogenetic distance is less than 5%, any additional common grouping criteria from the list above may classify separate types into the same species even if they were isolated from different hosts.”
This is inadequate; it does not constitute a formal genomics analysis and pertains to species rather than types. Reviewers and editors ask for a formal explicit non-ambiguous definition, hence the “Letter to the Editor”.
Three major objections to the genomics-based algorithm are: 1) the exact criteria are not defined precisely, in particular, no cut-off numbers are stated; 2) the recognition of emergent types, including recombinants, is based on “sufficient genomics, biological or pathogenic differences from related types”; and 3) the inclusion of the species designation with the type number in the name may be confusing.
It is clear from examining the published and original serology data, and from applying genomics and bioinformatics techniques to the examination of the novel peer-reviewed types HAdV-G52, -D53, -B55 and -D56, that the genome data and its analysis is preferable, logical, more informative and more accurate than “serology-only” data for characterizing, defining, understanding, typing and naming human adenoviruses. The recognition of biological or pathogenic differences as important attributes reflects the 21st century approach to viruses as model organisms in Systems Biology.
Addressing the objections:
1. “No precise criteria.”
1A) The lack of simple numerical cut-offs as guidelines is due to the biology of adenoviruses and the consideration of the adenovirus research field’s desire to accept the previously designated serotypes as genotypes (simply “type”). It is pointed out that papillomavirus taxonomy is defined by “genomics” and “DNA sequencing”, and has been “accepted without question since 2004”. While it is true DNA sequencing is used, depending on the researcher queried, this represents either the L1 gene or the whole genome. The biology of the papillomaviruses is such that their genomes contain 8 genes in a small (8kb) non-recombinant genome (“…strong evidence that PV genomes are very stable, and sequence changes by mutation or recombination are very rare events”) (deVilliers, et al., Virology 324:17 (04) (PMID: 15183049); Bernard, et al., Virology 401:70 (10) (PMID: 20206957)). This allows for the L1 gene, or any of the 8 genes, to represent the entire genome and provide absolute cut-off numbers: “A new PV isolate is recognized as such if the complete genome has been cloned and the DNA sequence of the L1 ORF differs by more than 10% from the closest known PV type. Differences between 2% and 10% homology define a subtype and less than 2% a variant” (PMID: 15183049).
In contrast, human and simian adenoviruses contain large genomes that undergo evolution, including recombination, as noted for HAdV-D53, -B55 and D-56 recently; it is also recognized in the literature that many of the original serotypes are also recombinants, for example, “18 of the 39 prototypes of group II” (Wigand and Fliedner, Archiv fur die gesamte Virusforschung (Arch Virol.) 24:245 (68) (PMID: 5698889). Inspection of the HAdV genome data shows it is not possible to have strict cut-off numbers, which are arbitrary, artificial and not reflective of Nature’s rules.
1B) The desire of the community to retain all original types dictates the typing should not be rigid with respect to numbers, especially in light of some of the prototypes isolated in the 1950s and in comparison to newer circulating field strains of the same type. While some genomes are highly conserved over time (PMID: 15681456, 20627349, 19932910), others may diverge considerably (PMID: 17360747, 19492050, 21056888) or moderately (PMID: 19553309, 20042633).
Specifically, for HAdV-B11 and -B35, the genome percent difference is 1.8%; for HAdV-B14 and -B55 it is 1.1%, with the latter two types differing by the recombination of 2.6% of genome representing the epitope from HAdV-B11 into the genome of HAdV-B14 (97.4% of the genome). This results in a significant change in its pathogenic effects, e.g., the host response where the body ‘sees’ HAdV-B55 as HAdV-B11, a renal pathogen, but the recombinant virus is a respiratory pathogen, with a different tissue tropism.
Additionally, a small genome percent difference between established prototypes is not unusual: the difference between HAdV-B21 and –B50 is 1.16%. A comparison of the difference between HAdV-B3 prototype and a current field strain circulating fifty years later is 1.83%. Two isolates of HAdV-D37 differ by 1.64%: the “GW” prototype versus a 1991 isolate.
All of the 56 type are sequenced and either analyzed or being analyzed. The data are available on an adenovirus wiki site (http://www.binf.gmu.edu/wiki/index.php/Main_Page).
1C) As to the objection of “imprecisely defined” criteria in the definition and recognition of new types resulting from “sufficient genomics, biological or pathogenic differences from related types”, this statement follows the ICTV guidelines of serotype definition: “or if substantial biophysical, biochemical or phylogenetic differences exist”.
Given the advances and applications of technology and methodology across the biological fields, virology is in step with 21st century science with the recognition that these viruses are not single independent entities, but do have a dramatic effect on the host and on populations. Therefore, the recognition and inclusion of biological and/or pathogenic properties to define a new type is important, regardless of the size or seemingly insignificant changes in the genome.
Again, this consideration reflects the access to more and better data, and with the understanding that a virus now is more than just two epitopes and is important in the context of its host as a System Biology model organism, as studied by Berk and colleagues (PMID: 18719283, 18719284 and 20861261). These are important for the understanding of the molecular evolution of HAdVs and the role the genome changes play in epidemiology, and in human and public health.
1D) The exact cut-offs for recognizing serotypes are not absolute and change according to the data. This provides for closely cross-reacting HAdVs to be recognized as “new” serotypes: Hierholzer, et al., 1991 (PMID: 1759904); Stevens, et al., 1967 (PMID: 6081385); deJong, et al., 1999 ( PMID: 10565911). These studies report 4x, 8x, 16x and 32x cross-reactions. Therefore, the ICTV recognizes new serotypes by both “case-by-case” and ambiguity (currently): ”titre ratios of 8 or 16, a serotype assignment is made … (lack of cross-reaction in HI tests) or if substantial biophysical, biochemical or phylogenetic differences exist”. In addition, there are serotypes that type identically: HAdV-D10 and -D29, along with the imputed molecular type of -D56.
It should be NOTED that an updated consistent enforcement of the current ICTV serotyping definition of the previously accepted serotypes 1-51 will result in several serotypes no longer recognizable as unique serotypes.
1E) SN titrations vary with lots and with animals- rabbits, horses, goats; reagents are not readily accessible unlike genome data. Genomic data are absolute, comprehensive, and portable. Fewer laboratories are doing wet bench SN and none is performing wet bench HI. Reagents are not readily accessible nor are exactly reproducible. Imputed epitopes as molecular typing constitute partial data in the context of the whole genome primary sequence.
1F) The ICTV criteria have been modified, in the past and recently, to accommodate previously unanticipated needs, e.g., the 2010 report for adenoviruses includes changing “…the required calculated phylogenetic distance as species demarcation criterion from 5-10 to 5-15%. This gives a more realistic criterion and space for taking other criteria into consideration, too, if in doubt.”
2. “Recognition of recombinants.”
2A) There is a question as to whether recombinants constitute new types and are “deserving” of new type numbers. Molecular evolution takes place using the mechanisms of base substitutions and insertion/deletion (indels), as well as recombination. These are observed for HAdVs as “genetic drift” (HAdV-G52) and “antigenic shift” (HAdV-B55). The resultant recombinant HAdV “product” may be biologically relevant and pathogenically significant (or may not). Understanding the pedigree of a current circulating field strain may lead to prophylaxis and prevention or other methods of abatement of an outbreak. ”Saying” HAdV-D53 is just another HAdV-D22, HAdV-B55 is just another HAdV-B11 or HAdV-D56 is just another HAdV-D15 and 29 is not productive.
The arguments of “then there will be too many type numbers” (papillomaviruses count 189 types amongst the members) and “a recombination of known serotypes is not a new type” and “the size of the recombination sequence determines if it is a new type” do not make sense from a scientific point of view.
2B) There are proposals to have two or three tracks of HAdV names. One is to have a track of “serotypes” with a separate track of “genome types”. The result will be a serotype designated “52” with the concurrent genome type designation of “1”.
Another proposal is an “alternative could be to mark these ‘types’ with an R, to clearly show that it consists of, and possesses identical serological properties of, previously described, well-known HAdV serotypes. Our proposal would be to name such recombinants as HAdV-R1…”
Recombinants as new types appear to be a strongly contested concept, even stripped of the “size matters” problem. It is very clear that intra-species recombination is a fact of life for HAdVs, and the idea of naming everything with a piece of something else “R” seems to totally miss this basic point of biology. Moreover, to imply that one must know the true extent of variance before assigning types is perfectly reasonable, IF one gets rid of the serotypes already assigned. The bottom line is that there is only minor intra-species whole genome variance in HAdV’s. Rather variance is restricted for the most part to hyper variable regions. Any typing methodology must take this into account, just as serotyping did.
Again, Wigand and Fliedner (1968; PMID: 5698889) reported “18 of 39 adenovirus strains of group II…. are serologically intermediate” in their paper entitled “Serologically intermediate adenovirus strains: A regular feature of group II adenoviruses”. This has been corroborated with the recent determination of complete genome sequences of these and all original serotypes of HAdVs.
2C) A potential third naming track based solely on the “hexon SN-epitope” arises from some researchers’ thought that any recombinants with the same SN marker as an older serotype is “merely just a variant of that serotype” and therefore should have the same type number regardless of the extent of the parental genome or the biological characteristics of the virus. This hexon-centric argument that the SN datum is the only important feature of the genome, especially to clinicians, and will be lost and untraceable if these recombinants are given type numbers other than the type number reflective of the SN marker is addressed easily by internet resources; for example, one tool is presented on this site as a “serotyping reference tool” where the user may enter a type number, e.g., 01, 02….57, and a display of SN-types are shown. This third naming proposal is absolutely inadequate. Examples include HAdV-D53 with only the SN epitope region of HAdV-D22 and HAdV-B55 with only the SN epitope region of HAdV-B11. The former is a difference of an ocular pathogen versus a non-pathogen and the latter is a difference of a respiratory pathogen versus a renal pathogen.
3. “Including species name with type number.”
3A) To provide a convenient reference to all researchers, the HAdV name is proposed to include the species desigation, so that HAdV-C1 would show type 1 as being a species C member. This is contested in that it would lead to confusion: “In regards, to the ‘HAdV-C2’ format of type names, a question of how a simian AdV belonging to HAdV-G would be named occurs; it should be SAdV-1 and not be HAdV-G1 as that would point to a human adenovirus serotype.
First, there is support by several researchers to include the simian adenoviruses with the HAdVs, that is, they appear to clade within the same species format. It should be noted that genomics and bioinformatics analyses reveal human and simian adenoviruses are highly similar to each other and their naming could be done without mention of the host, that is, humans are no more important that any of the other simians (or vertebrates). The concern from some researchers is that with SAdV-G1, researchers cannot be sure if G stands for SAdV-G or HAdV-G species. It is pointed out that the simians would have “S” as the first designation and the humans would have “H”. No confusion should result! Furthermore, the release of additional simian adenovirus genomes recently have been prefaced with a two-letter abbreviation for organism, for example “ch” for chimp and “go” for gorilla. This Letter does not touch upon this controversy.
Another perspective on this is, if truly Homo sapiens are just among the vertebrate species, then why use “human” or “simian” in the name? Why not just use “adenovirus”? Our data indicate most if not all of the simian adenoviruses segregate into the species A-G phylogenetic trees. Disregarding this, a reasonably intelligent person should know the difference between HAdV-G52, SAdV-G1 and SAdV-G7. Incidentally, the percent ID between these versus HAdV-G52 and HAdV-F40 and -F41 show a close relationship between the members of Gs, which include one HAdV and two SAdVs, than G to F.”
From a bioinformaticist’s point of view, “with regard to the HAdV only approach, one wholeheartedly agree and would very much like to see ICTV remove host names from AdV species, as this is a scientifically ill-advised approach that will most certainly cause issues down the road. Unfortunately, ICTV is unlikely to do this, and as such, any classification/genotyping approach will have to remain piecemeal and host organism specific. This also means that by a fact of policy ICTV does NOT consider Homo sapiens just another vertebrate species, no matter the educational training of its constituent members. I am pretty sure that all [of the] discussion[s] included the use of species names in isolate naming, nor do I see any mention otherwise on the linked web site, so his comments to that regards appear baseless. This of course should be pointed out.”
4. “Comments of ‘genomics is already being used for defining serotypes’.”
Again, it is pointed out that genomics use is implied by the acceptance of DNA sequences representing the serum neutralization epitope (instead of wet bench SN data) and the hemagglutination inhibition epitope (instead of wet bench HI data). This is not the way to set standards and no way to do science! It is ambiguous. There is no formal declaration in the ICTV reports noting that the DNA sequences are acceptable for typing human adenoviruses and no explicit guidelines as to the use and appropriateness of these data. As a result, phylogenetic trees are presented, in the literature, with qualitative evidence, i.e., “this branches differently”. Furthermore, it should be noted that the values and cut-offs, of the phylogenetics metrics, of both epitopes are not understood, particularly for the HI epitope.
5A) Genomics-based HdV typing is already in place and vetted by peer reviewers. Since the publication of HAdV-G52 as the first adenovirus to be typed using genomics and bioinformatics (Jones, et al., JVirol 81:5978 (07)(PMID: 17360747)), there has been only one formal objection and response (de Jong, et al. and Jones, et al. JVirol 82:3809 (08) (PMID: 18334604). Genomics and bioinformatics are valid approaches for characterizing, typing and naming novel HAdVs.
Still, the overriding problem is that reviewers and editors point out that “genomics is not accepted by the ICTV guidelines as a criterion for typing”; “type” is not an appropriate substitute for “serotype”; and currently, as of March 2011, there are only “51 serotypes of human adenoviruses recognized”.
5B) The serology methods, serum neutralization (SN) and hemagglutination inhibition (HI), have been replaced by DNA sequencing (imputing) of the epitopes despite the fact that a 1:1 relationship has not been established. Apparently no laboratory is currently performing wet-bench HI and very few labs perform SN tests. No antisera for the new types HAdV-G52 through HAdV-D56 have been made. Therefore, the current ICTV adenovirus definition of only, and formally accepting, serology for characterizing, determining and naming novel types of HAdV is outdated and is limiting research in and understanding of adenoviruses.
5C) Naming is simply a means of differentiating two viruses with distinct properties, to facilitate communication. Providing new names is not really the point. It is more important to reach consensus on what to call viruses newly characterized genomically, i.e., those that are not identical to previously sequenced “prototypes”, so that researchers can have a conversation about the science, and know what each other is talking about.
A problem lies in the conferring of a new number to this virus. The number is a link to a GenBank deposit of the entire genome. It would be more fruitful for the community, and lead to more progress scientifically, to be discussing the genomes, not the numbers. The recent workshop at NCBI was focused on facilitating submission of new adenovirus sequences to GenBank.
There are at least 200 fully sequenced adenoviral genomes that will be deposited in GenBank. Most of these will not be numbered, as that would require close analysis of each virus individually, and in most cases, showing unique biological properties will not be possible (the available metadata won’t be sufficient to do this).
The genomes and data emanating from them may be discussed, but arguing about naming the virus is counter-productive. We chose to name B55 (and peer reviewers accepted it) to distinguish it from other known viruses, for the purpose of communication. B55 is a paradigm changer, not because of its number, but because of its genome.
5D) Fifty-seven adenovirus researchers across disciplines have agreed, as co-signers, to the value and appropriateness of genome data for identifying, characterizing and typing HAdVs. Several others have agreed as well, at least in principle, but have refrained from signing the letter due to other considerations.
5E) Note, despite the widespread misinformed belief that the ICTV has a mandate to provide guidelines below the “species” taxa, it does not- according to the official website and to the current chairman of the appropriate study group.
Again, this is an on-going project in developing and refining typing criteria that is open to all in the adenovirus research community for input. The collective knowledge of the whole is greater than that of an individual; the inclusion of experts in the particular subfields is both important and critically necessary.