PHYLOGENETIC ANALYSIS OF THE GENOME OF AN ENTERITIS-ASSOCIATED BOTTLENOSE DOLPHIN MASTADENOVIRUS SUPPORTS A CLADE INFECTING THE CETARTIODACTYLA Open Access

Adenoviruses are common pathogens found in all tetrapod groups (Davidson et al. 2003). They are nonenveloped DNA viruses that replicate within host nuclei, with genomes of 26–45 kilobase pairs (Davidson et al. 2003). The family Adenoviridae may be divided into six genera that often show predilection for host clades (Doszpoly et al. 2013). Mastadenovirus is currently reported only in mammals and Aviadenovirus only in birds (Davidson et al. 2003). Atadenovirus likely originated in squamates but has been found in other host species including birds, ruminants, and marsupials (Wellehan et al. 2004). Siadenovirus has an unknown origin (Doszpoly et al. 2013). A single member of the genus Ichtadenovirus was characterized in the white sturgeon (Acipenser transmontanus) (Doszpoly et al. 2013). After the identification of a clade of novel viruses from testudinoid turtles, the provisional genus Testadenovirus has been proposed (Doszpoly et al. 2013).

Placental mammals are divided into four main superorders (Murphy et al. 2001). The two less-speciose superorders are the Afrotheria (including elephants, manatees, aardvarks, and others) and Xenarthra (armadillos, sloths, anteaters). The two larger sister clades (collectively known as Boreoeutheria) contain most mammal species and are classified as Laurasiatheria (carnivores, Cetartiodactyla [whales and even-toed ungulates], Perissodactyla [odd-toed ungulates], insectivores, bats) and Euarchontoglires (rabbits, rodents, primates) (Murphy et al. 2001). There is not yet consensus on the branching order of Afrotheria and Xenarthra (Bininda-Emonds et al. 2007; Song et al. 2012), but adenoviruses have yet to be sequenced from hosts in these taxa. Within the Laurasiatheria, a clade known as the Cetartiodactyla is formed by the camelids, suids, hippos, cetaceans, and ruminants (Murphy et al. 2001). The Cetruminantia are a monophyletic subgroup within the Cetartiodactyla that contains cetaceans, ruminants, and hippopotami (Murphy et al. 2001).

A trend toward host-virus codivergence has been noted in adenoviruses. This phenomenon is also observed in other families, such as the herpesviruses, that share complexity and site of replication (Pulliam and Dushoff 2009). Genomes that are rich in adenine/thymidine are associated with more-recent host-jumping events (Wellehan et al. 2004; Poss et al. 2006; Doszpoly et al. 2013). Within Mastadenovirus, support exists for an evolutionary division between strains infecting the mammalian superorders Laurasiatheria and Euarchontoglires (Hall et al. 2012). Within these classifications, closely related host species may make it easier for host-jumping events to occur. As one example, titi monkey adenovirus 1 caused severe respiratory disease in titi monkeys (Callicebus cupreus) in a research facility, with interprimate monkey to human transmission observed (Chen et al. 2011). The only previous complete gene set sequenced from an adenovirus from a marine mammal is from Zalophus californianus, California sea lion adenovirus 1 (Cortés-Hinojosa et al. 2015). Sea lions are in the superorder Laurasiatheria but are in order Carnivora, not the Cetartiodactyla. Several genomes of adenoviruses from hosts in the Cetartiodactyla, more-closely related to cetaceans, are available for comparison.

Polymerase sequence differences greater than 15% are consistent with distinct species, and differences less than 15% are seen with distinct types within a species (Harrach et al. 2011). Each type may be associated with a wide variety of pathogenic implications ranging from subclinical infection to high levels of morbidity and mortality (Echavarria 2008). In general, more-severe clinical illness has been associated with host-jumping events (Pulliam and Dushoff 2009). Infections may be highly contagious in endemic hosts but they typically suffer from lesser disease (Chen et al. 2011). Human mastadenoviruses have been associated with conjunctivitis, gastroenteritis, fulminant hepatitis, pneumonia, and encephalitis (Jawetz 1959; Yolken et al. 1982; Gray et al. 2000). Respiratory diseases caused by mastadenoviruses affect domestic equids, bovines, porcines, ovines, and caprines (Maclachlan and Dubovi 2016). Canids can be infected with the species Canine adenovirus A, which is subclassified into two types; canine adenovirus 1 and 2 (CAdV-1 and CAdV-2; Harrach et al. 2011). These are associated with viral hepatitis and respiratory disease, respectively (Maclachlan and Dubovi 2016). Additional novel mastadenoviruses have been found in wildlife including bats, rodents, tree shrews, sea lions, and primates (Phan et al. 2011; Kohl et al. 2012; Cortés-Hinojosa et al. 2015).

The presence of adenoviral infection in cetaceans was first established when adenovirus-like particles were found in sei (Balaenoptera borealis), bowhead (Balaena mysticetus), and beluga whales (Delphinapterus leucas), but no further analysis of these samples was pursued (Van Bressem et al. 1999). An adenovirus polymerase was amplified from a mesenteric lymph node of a stranded harbor porpoise (Phocoena phocoena) that was found dead (van Beurden et al. 2017; GenBank accession no. JN377908). Recently, an adenovirus was described from bottlenose dolphins in Spain with anorexia, diarrhea, and vomiting, including characterization of partial hexon and polymerase genes (Rubio-Guerri et al. 2015). The PCR positivity of the animals was seen at the same time as were clinical signs. According to the International Committee on Taxonomy of Viruses (https://talk.ictvonline.org/) rules of nomenclature, we hereafter refer to that virus as bottlenose dolphin adenovirus 1 (BdAdV-1; Harrach et al. 2011). We describe a second novel bottlenose dolphin adenovirus from a dolphin with diarrhea including the full gene complement of a cetacean adenovirus, the phylogenetic analysis, and development of a quantitative PCR (qPCR) assay for rapid, sensitive, and specific diagnostic testing.

Samples were collected from a bottlenose dolphin after the emergence of clinical signs. As a member of an open-water managed population, this individual was noted to suffer from acute anorexia, diarrhea, and lethargy. Blood samples were taken for complete blood count and plasma chemistry panels. Fecal samples were also obtained for negative-staining electron microscopy using previously described methodology (Atkins et al. 2009) and viral testing.

We extracted DNA from the fecal sample of the affected dolphin using the Qiagen DNeasy kit (DNeasy Blood and Tissue Kit, Qiagen Inc., Valencia, California, USA) following manufacturer instructions. We first identified an adenovirus sequence using previously published consensus primers in a nested PCR protocol (Wellehan et al. 2004). Additional regions of sequence were obtained at various points in the genome by use of further consensus primers based on mastadenoviruses from other laurasiatherian hosts, including the inverted terminal repeat sequence (see Supplementary Material Table). The amplicons were sequenced in both directions using Sanger methodology on ABI 3130 automated DNA sequencers at the University of Florida Interdisciplinary Center for Biotechnology Research (Life Technologies, Carlsbad, California, USA). We then used the obtained sequences to design new primers specific for this virus (see Supplementary Material Table) and amplified products were sequenced as above. Each nucleotide was sequenced at least twice in each direction.

A standard PCR protocol was used for each primer pair, with annealing temperatures adjusted based upon manufacturer predictions of melting temperature for each set of primers. The Takara Ex Taq TM (Hot Start Version, TaKaRa Bio Inc., Otsu, Japan) reagents were used for these reactions. Each PCR began with an initial denaturing step of 94 C for 5 min. The following three steps were then grouped and repeated for 45 cycles. These steps started with further denaturation at 94 C for 30 s followed by annealing at a temperature that was generally determined as 6 C below the manufacturer-predicted melting temperature for each primer pair. Extension at 72 C for 5 min was the final step in this group, with times adjusted for the expected size of the amplicon. A single final elongation step was then performed at 72 C for 10 min followed by a 4 C holding temperature. Upon completion, 1% agarose gels were used to visualize PCR products. Fragments that fell within the expected size range were then cut from the gel and extracted with the QIAquick Gel extraction kits (Qiagen). Sequencing was completed of the extracted PCR products as mentioned earlier.

Open reading frames (ORF) were predicted using CLC Main Workbench version 5.5 (CLC Bio, Katrinebjerg, Denmark), and homologies to other known proteins were determined using BLASTP (Altschul et al. 1990). Based on homologous adenoviral sequences and the use of NNSplice (Reese et al. 1997), predicted splicing patterns were determined.

We designed a qPCR assay specific to the polymerase gene of this virus using a forward primer (TtAdVqPCRF1, TGATGTGTTACCGCCGTTTT), reverse primer (TtAdVqPCRR1, CGTAATGGTTCGTTCGTCCA), and probe (TtAdVprobe1, 6FAM-TTCACGAAGAGGAGGAAGACTTTG-MGB), amplifying a 66-base pair (bp) product. Primers and probes were designed in the program Primer Express version 3.0 (Applied Biosystems, Foster City, California, USA). The probe was designed with minor groove binder (MGB) as a quencher and 6-carboxyfluorescein (FAM) as a reporter dye. To obtain a template for standard curves, the DNA-dependent DNA polymerase was amplified from the index case using a previously published consensus PCR (Wellehan et al. 2004). The DNA concentration was determined with a NanoDrop™ spectrophotometer (Thermo Fisher Scientific, San Jose, California, USA) and double-checked by comparison to a DNA mass ladder. The standard curve was included in triplicate on each plate using a serial 10-fold dilution of the PCR amplicon from the index case with values ranging from 10 to 10⁶ copies. The 20-μL reactions were run in duplicate. These consisted of 7 μL of extracted DNA and 10 μL of a commercial qPCR universal master mix (TaqMan™ FastUniversal PCR Master Mix 2X, Applied Biosystems) with 1 μL of each primer added at a concentration of 18 μM and 1 μL of the probe at 5 μM. The reactions were processed using a standard fast protocol. A control for each sample was added in a separate well using a Eukaryotic 18S rRNA Endogenous Control primer/probe set (Applied Biosystems). Amplification of the reactions was performed with a FAST 7500 Real-Time PCR System (Applied Biosystems) with the standard fast protocol: initial denaturation at 95 C for 20 s and then 50 cycles of 95 C for 3 s followed by 60 C for 30 s.

Additional DNA extracts were used to test this qPCR assay against other bottlenose dolphins from both wild and managed populations. These included a combination of historic samples extracted from both respiratory and fecal specimens of 42 dolphins in 2012. The DNA extracts from 23 individuals of other species known to be positive for different adenovirus species were also used to ensure a lack of cross-reactivity of the assay. These included samples of adenoviruses from three genera: Aviadenovirus, Mastadenovirus, and Siadenovirus.

Five conserved predicted proteins in the core area of this new bottlenose dolphin genome were chosen (polymerase, pTP, penton base, hexon, and p100K). Thirty-five fully sequenced adenoviruses were selected from GenBank: 22 mastadenoviruses, five aviadenoviruses, four atadenoviruses, and four siadenoviruses in addition to our virus. A shorter region of polymerase for which homologous 147–150 amino acid sequences from BdAdV-1 and the harbor porpoise adenovirus 1 were available as well as testadenoviruses and additional atadenoviruses and siadenoviruses, which were also examined. Homologous amino acid sequences were aligned using MAFFT (Katoh and Toh 2008). Bayesian analyses of the amino acid alignments were performed using MrBayes 3.2.2 (Ronquist and Huelsenbeck 2003) on the CIPRES server (Miller et al. 2015). Bayesian analyses were run for 2,000,000 generations with the implementation of a stopping rule when the average deviation of the split frequencies was <0.01%. Chains were sampled every 100 generations and the first 25% of Markov chain Monte Carlo samples were discarded as a burn-in. To evaluate whether different tree topologies generated with different proteins were statistically different, Bayes factors were examined. We used the same set of proteins to run maximum likelihood (ML) analysis in RAxML-HPC2 on the CIPRES server (Stamatakis et al. 2008; Miller et al. 2015) with a gamma distributed rate variation and 1,000 bootstrap replicates to determine node support.

The adult dolphin was part of a managed population housed in open-ocean enclosures. The patient presented with signs consisting of acute anorexia, diarrhea, and lethargy that resolved with supportive care. Negative-staining electron microscopy of feces revealed the presence of icosahedral particles approximately 75 nm in diameter, morphologically consistent with an adenovirus (Fig. 1). Surface projections were seen that resembled adenoviral fiber proteins (Fig. 1).

The sequence we obtained was composed of 29,474 bp. The sequence was submitted to GenBank under accession no. KR024710. The inverted terminal repeat regions were estimated to add 23 bp to each end of the genome. The virus was named bottlenose dolphin adenovirus 2 (BdAdV-2). The total guanine/cytosine (G/C) content was calculated to be 36%. The BdAdV-2 has a core region similar to other adenoviruses, with some deviations at certain locations in the genome. The presence of a fiber gene was not identified on the basis of sequence homology, but there is an ORF of roughly the expected size in the region where the fiber gene is predicted. This possible fiber protein includes a region with homology to an immunoglobulin domain identified using BLASTP. The BdAdV-2 genome possesses two exons homologous to dUTPase, and the expected U exon was not identified in this virus. The E3 and E4 regions each contain a predicted protein (E3-11.5 kDa and E4-23 kDa, respectively) that does not show significant homology to any known protein. A genome map was generated to display the BdAdV-2 genome (Fig. 2).

The BdAdV-2 genome possesses signals for splicing at many of the locations found in previous adenovirus studies (Davidson et al. 2003). These splicing patterns are typically seen in the E2 and late genes of all adenoviruses (Fig. 2). These predicted sites will require experimental evidence to confirm.

No additional animals of the 42 bottlenose dolphins tested were identified as being positive for BdAdV-2 infection. The standard curve showed amplification at all values from 10 to 10⁶ copies. The R² values of the standard curve ranged from 0.97–0.98 and efficiencies ranged from 94.4–96.7%. The primers and probe were assessed for cross-reactivity with 23 samples known to be positive for 23 different adenovirus types. No cross-reactivity was observed.

Results of phylogenetic analyses were similar for all genes, and a concatenated five-gene analysis is shown (Fig. 3). Both ML and Bayesian analyses grouped BdAdV-2 into a clade containing mastadenoviruses infecting other members of Cetartiodactyla; bovine adenovirus 1, bovine adenovirus 2, porcine adenovirus 3, and porcine adenovirus 5. This clade was supported with 100% posterior probability and a 99.1% ML bootstrap value. The phylogeny of BdAdV-2 parallels that of its bottlenose dolphin host. While the deeper branching of the mastadenoviruses was less well resolved than the multiple gene analysis, analysis of the shorter polymerase fragment for which other cetacean adenoviruses were available suggested the cetacean adenoviruses form a clade, supported with 82.4% posterior probability and a 57.6% ML bootstrap value (see Supplementary Material Figure). The 444-bp region of polymerase with comparable cetacean adenovirus sequence showed 73% nucleotide identity with the Spanish bottlenose dolphin adenovirus sequence and 81% identity with the harbor porpoise adenovirus 1 sequence, and the 318-bp region of hexon showed 75% nucleotide identity with BdAdV-1. This is consistent with distances seen between distinct species.

Sixteen genes are common between all five recognized genera of adenoviruses. Many of these proteins are highly conserved, but significant variation has been seen in the fiber gene, likely due to the strong positive selective pressure placed on this protein by the immune system (Robinson et al. 2011). The fiber protein has been identified as the first viral component to interact with host cells through a variety of receptors and may be involved in the initial stages of infection (Robinson et al. 2011). Fiber and other core structural genes have also been implicated in activation of the initial innate inflammatory response (Russell 2009). An alteration in this gene may change the way in which the host immune system responds to the virus. Type BdAdV-2 is particularly divergent in this region, not displaying any homology to known fiber sequences. Based on the fiber-like surface projections seen on electron microscopy, the virus likely possesses a fiber protein with an unrecognizably distinct sequence that includes an immunoglobulin domain. The immunoglobulin domain in the fiber protein may potentially interact with the host immune system (Nimmerjahn and Ravetch 2008). The U exon is also not clearly represented in the expected region in BdAdV-2. The U exon is common to most members of the adenovirus family, but it has been lost in other mastadenoviruses such as porcine adenovirus 5 and murine adenovirus 1 (Davidson et al. 2003).

Two genes homologous to the dUTPase superfamily are present in the E4 region of BdAdV-2. The presence of dUTPase protein in viruses has been primarily explained by horizontal gene transfer across different virus taxa (Baldo and McClure 1999). In some viruses, dUTPases have an important role enabling virus replication in nondividing cells (Pyles et al. 1992; Oliveros et al. 1999). Viral dUTPases have been studied in poxviruses, retroviruses, and herpesviruses (Chen et al. 2002), but they are also found in other adenoviruses where they have been associated with oncogenic potential (Weiss et al. 1997; Baldo and McClure 1999). The predicted BdAdV-2 dUTPases are more-closely related to nonadenoviral dUTPases than are those of other mastadenoviruses; BLASTP searches found higher scores with dUTPases of fungi, but dUTPase1 did show some similarity to dUTPases found in aviadenoviruses.

The E3 and E4 regions of adenoviruses generally consist of proteins that are less conserved and often specific to a genus or single virus (Davidson et al. 2003). The E3 region of BdAdV-2 has two ORFs with uncharacterized protein products. The E3-12.5 kDa includes a putative E3 domain most similar to related regions of porcine and bovine mastadenoviruses. This reinforces the phylogenetic analysis that places BdAdV-2 into a clade with other cetartiodactylan adenoviruses. Predicted E3-11.5 kDa and E4-23 kDa proteins do not have close homology with any defined sequence.

A strong trend toward similarity of host and virus phylogenies has been documented in adenoviruses. When examining congruent relationships between host phylogenies and virus phylogenies, we must consider whether a relationship represents true codivergence or a phylogenetic congruence of paralogous viruses (Jackson 2005). Barriers to host jumping are important: the more host jumping, the more any signal of host-pathogen codivergence is to be obscured. One study has found that the factor most-strongly correlated with a virus' tendency to host jump is the ability to replicate in the cytoplasm, whereas adenoviruses replicate in the nucleus (Pulliam and Dushoff 2009). A previous analysis of the genera Atadenovirus and Mastadenovirus found indications that codivergence may have affected adenoviral phylogeny, in conjunction with duplication events (Jackson 2005). However, data at the time were limited to single genes, resulting in relatively low confidence values in the maximum parsimony analyses. All but one of the atadenoviruses in that study were from nonsquamate hosts; later data show that the diversity of atadenoviruses is predominantly in squamate hosts, and their presence in nonsquamate hosts may represent host-jumping events (Wellehan et al. 2004; Ball et al. 2014). The genome of a second squamate atadenovirus has recently been completed, and more taxa should provide additional clarification (Pénzes et al. 2014). Examination of multiple genes for phylogenetic signal is important; recombination is well documented in adenoviruses, and individual gene phylogenies may not reflect the history of the rest of the viral genome (Lukashev et al. 2008).

Congruence of host taxa and viruses on shorter branches with lack of congruence at deeper branchings may be more consistent with preferential host switching than with host-virus codivergence (Jackson and Charleston 2004). It is interesting to note that, with midpoint rooting, the earliest division of the adenoviruses divides Mastadenovirus, which utilizes mammal hosts, from the other genera that are endemic in sauropsid (reptile) hosts. The sauropsid-mammal split is the earliest divergence of extant Amniota. In the clade primarily using reptile hosts, the earliest divergence is the genus Atadenovirus, much like their squamate hosts represent the earliest divergence within the extant reptiles (Chiari et al. 2012). This congruence at deeper levels is more suggestive of a significant role for codivergence. Longer sequence data from additional adenoviruses of diverse vertebrate hosts are needed to clarify the relative roles of codivergence and congruence.

Higher G/C contents and lower pathogenicity have been associated with longstanding host-virus relationships. In contrast, lower G/C contents have been related to recent host jumps (Wellehan et al. 2004; Poss et al. 2006; van Hemert et al. 2007). This could be a way to avoid the innate immunity of the new host through toll-like receptors that recognize unmethylated CG dinucleotides (Aderem and Hume 2000). Although hypotheses for mechanisms leading to this nucleotide bias have not been tested, its presence has been shown experimentally (Poss et al. 2006). The first known members of Atadenovirus had low G/C contents with a range of 32–47%, which ultimately gave the genus its name. These were discovered within ruminant, marsupial, and avian hosts. Later discoveries within this genus found higher G/C contents within viruses from squamates (43.75–58.09%), leading most to believe that this group may represent the original hosts for Atadenovirus (Wellehan et al. 2004). With a relatively low G/C content of 36%, BdAdV-2 could represent a recent host-jumping event. No further cases were identified in other bottlenose dolphins by qPCR surveillance. Further studies will be important for determining the virulence of BdAdV-2 in bottlenose dolphins and the host range.

Our phylogenetic analyses classified BdAdV-2 within the genus Mastadenovirus and grouped it with adenoviruses of laurasiatherian hosts. More specifically, BdAdV-2 and most mastadenoviruses infecting Cetartiodactyla appear to be monophyletic. Bovine adenovirus 3 is outside this clade, which could mean that this strain represents a host-jumping event from another host group or alternatively may represent a separate clade that evolved in parallel in the Laurasiatheria. The overall evolutionary pattern of BdAdV-2 closely resembles that of the bottlenose dolphin. This finding is in contrast to the prediction of a recent host-jumping event, based on the low G/C content of BdAdV-2. Considering both aspects of the data, it is likely that the origin of this virus is another member of Cetartiodactyla. Further studies should investigate the relationship between adenoviruses of cetaceans and other Cetartiodactyla. The relationship between adenoviruses found in cetaceans with those found in ruminant species may present strong implications for management of agricultural runoff and other open-water contaminants. Other examples of cetacean viruses that likely originated in bovids include Tursiops truncatus parainfluenza virus 1 and bovine enterovirus (Nollens et al. 2008, 2009). Dolphins and domestic cattle will have similar host proteins and receptors because they are closely related, allowing viruses to pass between these species more readily (Nollens et al. 2009). However, based on limited data, BdAdV-2 appears to form a clade with other cetacean adenoviruses; it may have originated in another cetacean species. Phylogenetic analysis of a small section of the polymerase found that BdAdV-2 was more-closely related to harbor porpoise adenovirus 1 than to BdAdV-1, most consistent with host jumping between cetacean species, and the paraphyly of the bottlenose dolphin adenoviruses is not consistent with codivergence on a finer scale. The divergence within these viruses is consistent with what is seen with differences between species, suggesting a longer relationship of these viruses within the Cetacea, or at least within the suborder Odontoceti (toothed whales). Further surveillance and sequence data of other cetacean adenoviruses is needed.

We discovered BdAdV-2 infection in an animal suffering from acute anorexia, lethargy, and diarrhea. This clinical presentation is similar to that seen in BdAdV-1–infected dolphins in Spain (Rubio-Guerri et al. 2015). The sensitive and specific qPCR we developed provides a means for rapid diagnostic evaluation of potentially infected individuals.

Our work was funded by grant N00014-09-1-0252 from the Office of Naval Research and by National Oceanic and Atmospheric Administration contract 00090868 to J.F.X.W. We thank Hubbs-SeaWorld Research Institute, the Navy Marine Mammal Program, and the Hervey Family Non-Endowment Fund at The San Diego Foundation for their support with sampling and assays.

Supplementary material for this article is online at http://dx.doi.org/10.7589/2017-03-052.