In February of 1997, 4 of 9 infant baboons at the TBRI developed an acute respiratory infection of unknown etiology shortly after being isolated from birth in preparation for a research study on respiratory syncytial virus (Fig. 1). The first two cases (Fig. 1, baboon 1 [B1] and B2) died 13 and 5 days after the onset of symptoms despite treatment with empiric broad-spectrum antibiotics and supportive care. Symptoms began with sneezing and rapidly worsened with development of lethargy, >20% weight loss, abnormally low body temperature, and dyspnea. White blood cell (WBC) counts were normal, but a few atypical lymphocytes were present. Chest radiographs revealed a bilateral interstitial pneumonia. Initial bacterial cultures of bronchoalveolar lavage fluid and blood samples were positive for methicillin-sensitive Staphylococcus aureus (MSSA) and rare Kluyvera ascorbata (23) in only one of the two cases.

On necropsy immediately after death, the lung tissue from both cases was noted to be hemorrhagic with patchy regions of consolidation and significant neutrophilic infiltration. Notably, intranuclear inclusions (presumably from AdV infection) were evident throughout the respiratory epithelium and were most evident in the major airways. The final pathological diagnosis was bronchointerstitial pneumonia, probably viral in etiology, with accompanying tonsillitis, lymphadenitis, and mild liver necrosis (in one baboon). Tests from lung tissue were negative for Bordetella pertussis, Chlamydophila spp., Mycoplasma spp., Ureaplasma spp. Legionella spp., and hantavirus. Adenovirus was isolated in cell culture from bronchoalveolar lavage fluid and lung tissue samples from both fatal cases. The isolates were untypeable by virus neutralization testing at an off-site laboratory using sera reactive to members of species HAdV-A to HAdV-F and SAdV-A.

Although clinical samples from the two fatal cases were unavailable for further analysis, two other animals in the room were noted to be sneezing and coughing around the same time cases B1 and B2 presented with fatal pneumonia (Fig. 1, B3 and B5). Adenovirus was isolated from a nasal swab collected from one of these two symptomatic individuals during the 1997 outbreak (Fig. 1, room 1, BaAdV-3). Nasal swabs were also collected from the sole remaining asymptomatic baboon in the same room as the symptomatic cases (Fig. 1, room 1, B4) and from asymptomatic baboons in a nearby but separate room (Fig. 1, room 2, B8 and B9). From these nasal swabs, 3 additional AdV isolates were obtained (Fig. 1, BaAdV-1, BaAdV-2, and BaAdV-4). The two symptomatic baboons in room 1, cases B3 and B5, were quarantined and recovered completely within 1 week with supportive care, including intravenous fluids, oxygen administration, and empirical antibiotics. None of the 4 baboon AdV isolates was typeable as HAdV species A to F or SAdV-A by neutralization testing done in 1997.

We attempted to culture the 4 BAdVs isolated from sick and asymptomatic baboons during the 1997 outbreak in a variety of human and monkey cell lines. The majority of cells and cell lines tested resulted in productive infection as determined by the extent of cytopathic effect (CPE) (Table 1). All 4 BAdVs grew efficiently in cells from other Old World monkey species (rhesus, cynomolgus, and African green monkeys). In 2012, we tested the BaAdVs for growth in human cell lines; all 4 grew efficiently in the human lung adenocarcinoma A549 cell line, which is commonly employed in isolation of human AdVs (24). Notably, unlike the other 3 AdV strains, BaAdV-3 was also successfully cultured on two additional human cell lines. No growth was observed in lymphoblastoid B95a cells from marmosets, which are New World monkeys.

In 2012, we sought to further characterize the 4 AdVs isolated from baboons by whole-genome sequencing and phylogenetic analysis. The sequences of the AdV hexon, DNA polymerase, and fiber were initially recovered by Sanger sequencing. To sequence the entire AdV genome, early passaged cultures corresponding to the 4 BaAdV isolates were subjected to next-generation “deep” sequencing on an Illumina HiSeq 2000 (Fig. 2). Out of 32.9 to 45.2 million raw reads, from 61.3% to 93.1% of the genome was assembled de novo for each of the 4 AdV strains. Specific PCR and Sanger sequencing were then used to confirm the de novo assembly and to fill in the gaps. The genomic assemblies revealed that all 4 baboon AdV strains retained the major core adenoviral proteins, and similar to AdVs in species SAdV-B, -F, and -G, all 4 strains contained two fiber proteins (16, 25–28).

By pairwise analysis of nucleotide identity (Fig. 3A), BaAdV-1, the strain isolated from an asymptomatic baboon in the outbreak room (Fig. 1, room 1, B4), was found to be closely related to a previously described SAdV-B strain isolated from a stool sample from a captive rhesus monkey (SAdV-A1335, GenBank accession number JN880456, 97.8% identity) (18). Strains BaAdV-2 and BaAdV-4, both isolated from asymptomatic baboons housed away from the outbreak location in a separate room (Fig. 1, room 2, B8 and B9), were 100% identical to each other, while strain BaAdV-3, isolated from a symptomatic baboon positioned near the two baboons who died from adenoviral pneumonia (Fig. 1, room 1, B5), was 91.2% identical to strains BaAdV-2/-4 (Fig. 3B). The closest relatives to BaAdV-2/-4 and BaAdV-3 were members of AdV species F and G (Fig. 3B and 4). Phylogenetic analysis of the individual hexon, penton base, DNA polymerase, and fiber proteins and amino acid pairwise identity comparisons revealed that BaAdV-1 is a member of the SAdV-B species (Fig. 4 and 5A), while BaAdV-2/-4 and BaAdV-3 appear to be members of a new species intermediate between SAdV-F and SAdV-G, provisionally named “simian adenovirus C (SAdV-C)” (Fig. 4 and 5B). Both BaAdV-2/-4 and BaAdV-3, like members of the SAdV-A and SAdV-B species, contained the RGD motif in the penton base protein which mediates binding to α_v integrins during virus internalization (29).

Pools of hyperimmune rabbit antisera containing neutralizing antibodies to the 41 HAdV prototype strains (representing species A to F) were tested against the BaAdVs. No neutralization was observed at a starting serum dilution of 1:8 with the exception of the species F (HAdV-40 and -41) antisera pool that showed low-level neutralization (1:16) with BaAdV-1 (SAdV-B) and BaAdV-2/-4 (SAdV-C). Sera from control baboon B107 that was positive for neutralizing antibody to BaAdV-1 and BaAdV-2/-4 did not neutralize species G HAdV-52.

Notably, despite sharing 91.2% overall nucleotide identity across the genome (Fig. 3B), the sequence of BaAdV-3, the only AdV among the 4 that had been isolated from a symptomatic baboon, diverged significantly from that of BaAdV-2/-4 in the short fiber 1 gene. While the other major adenoviral proteins were >90% identical, the short fiber 1 of BaAdV-3 shared only 58% amino acid identity relative to that of BaAdV-2/-4 (Fig. 5B). Similarity and bootscanning plots of BaAdV-3 relative to BaAdV-2/-4 (species SAdV-C) and related AdVs in species SAdV-A, -F, and -G suggested high likelihood of a recombination event involving the short fiber 1 gene (Fig. 6). However, if the short fiber 1 gene of BaAdV-3 is the result of a recombination event, the presumptive AdV donor strain has yet to be identified, as the BaAdV-3 short fiber 1 lacked any closely related phylogenetic neighbors (~58% to BaAdV-2/-4 and ≤50% amino acid identity to other sequenced fiber proteins in GenBank).

Of note, many staff members at the TBRI had anecdotally reported experiencing “flu-like” symptoms around the time of onset of the 1997 baboon outbreak. To investigate the possibility that a cross-species transmission event, either zoonotic (from baboon to human) or anthroponotic (from human to baboon), may have occurred, preoutbreak and postoutbreak sera from potentially exposed human staff personnel at the TBRI (Table 2, human 1 [H1] to H6) were tested for antibodies to BaAdV-1, BaAdV-3, and BaAdV-2/-4 by virus neutralization in a blinded fashion. As additional controls for baseline seroprevalence, we also tested sera from a random selection of 5 young children and available sera from 10 baboons born approximately the same time as affected baboons with pneumonia, but not part of the 1997 outbreak. Significantly, 5 of 6 (83%) and 6 of 6 (100%) personnel, while seronegative prior to the outbreak, had evidence of neutralizing antibody titers to BaAdV-1 and BaAdV-2/-4, respectively, after the outbreak. The highest neutralizing antibody response, 1:160, corresponded to the researcher in closest contact with sick baboons during the 1997 outbreak (Table 2, H1). Interestingly, no neutralization to BaAdV-3 was observed in any of the staff personnel. The specificity of the neutralization assays was further confirmed by the screening of 5 epidemiologically unassociated human children under 5 years of age, all of whom were negative for neutralizing antibodies to BaAdV-1, BaAdV-2/-4, and BaAdV-3. Among 11 healthy baboons in the colony who were not part of the 1997 outbreak, 4 of 11 (36%) (from sera collected in 2002–2003) and 3 of 11 (27%) (from sera collected in 1996 and 2003) harbored antibodies to BaAdV-1 and BaAdV-2/-4, respectively (Table 2, B101 to B111). Very little to no neutralization was observed to BaAdV-3; neutralizing antibody at 1:8 titer was detected in only one baboon, B108, who also had a titer of 1:64 to closely related strain BaAdV-2/-4.

This study was performed in strict accordance with the Guide for Care and Use of Laboratory Animals (39). The use and care of all animals followed policies and guidelines established by the Institutional Animal Care and Use Committee (IACUC) and the Institutional Biohazards Committee of the TBRI. The protocol for the maintenance and breeding of the baboon study was approved by the TBRI IACUC. No specific animal research protocol was drafted for this study, as only excess clinical samples were analyzed for diagnostic purposes. Extensive veterinary care was provided to all animals affected by the outbreak to minimize pain and distress. Written informed consent was obtained from the staff of the TBRI for analysis of their samples and analyzed under protocols approved by the University of California, San Francisco (UCSF) institutional review board (IRB). Any potentially identifying information has been provided with the explicit permission of the individual involved.

Analysis of necropsy tissues and cultivation of the novel baboon AdVs described in this study were performed under biosafety level 2 (BSL-2) conditions as approved by the Institutional Biohazards Committees of the TBRI and Centers for Disease Control and Prevention (CDC). Nucleic acid extractions of AdV cultures were performed at the University of California, San Francisco in BSL-2 facilities specifically certified by the UCSF Biosafety Committee for handling novel simian AdVs. Serological analysis of human and baboon sera was performed in a BSL-2 laboratory at the California Department of Public Health (DPH).

The baboon outbreak lasted approximately 3 weeks from 21 February to mid-March of 1997. Affected baboons were quarantined immediately after development of respiratory symptoms. The two baboons with fatal cases died or were humanely euthanized 5 and 13 days after the onset of clinical signs. Daily reports on clinical and epidemiological parameters were tracked and recorded by veterinary and management staff. In response to the outbreak, all incubator rooms were decontaminated with paraformaldehyde gas. Cages, walls, floors, and all exposed work area surfaces were cleaned with 2.6% buffered glutaraldehyde (Metricide) or bleach. Disposable protective suits and gloves were worn at all times when feeding or otherwise in contact with infant baboons for a period of at least 2 weeks. Hematological testing and cultures for bacteria, mycoplasma, and fungi were performed at the University of Texas Health Science Center at San Antonio (UTHSCSA). Samples were also tested for respiratory syncytial virus (RSV), influenza virus, parainfluenza, human adenovirus, and herpesviruses, including cytomegalovirus (CMV), at the UTHSCSA. In cultures manifesting cytopathic effects, the presence of adenovirus was confirmed by immunofluorescence microscopy with an anti-hexon antibody conjugate. Respiratory samples were sent out to an off-site laboratory and tested for Bordetella pertussis, Chlamydophila spp., Mycoplasma spp., Ureaplasma spp., Legionella spp., and hantavirus. AdV typing by virus neutralization testing using sera reactive against HAdV species A to F and SAdV-A viruses was performed at Viral Reference Laboratories, Inc., in San Antonio, TX, in 1997. Culturing of baboon AdVs was performed at the UTHSCSA, CDC, and California DPH.

Gross and histopathological analyses of necropsy tissues were performed by a board-certified veterinary pathologist at the Primate Research Center of the TBRI. Necropsy tissues were fixed in 10% formalin and embedded in paraffin. Five-micron-thick sections were then cut using a microtome, stained with hematoxylin and eosin (H&E), and examined by light microscopy.

Two hundred microliters of sample was first passed through a 0.4-µm-pore-size filter to remove bacteria and cellular debris and then treated with RNase (Invitrogen, Carlsbad, CA). Total nucleic acid was then extracted from cultured AdV supernatant using commercially available kits (Qiagen, Valencia, CA).

All inoculations of monkey cells (PMK, or primary monkey kidney; CyMK, or cynomolgus monkey kidney; and Vero [African green monkey kidney]) were made using primary nasal swab specimens. Inoculations of human cells and cell lines were made with P1 virus after a single passage in monkey cells. The cells or cell lines were grown in media consisting of Hank’s medium (for A549 cells) or Dulbecco’s modified Eagle’s medium (DMEM) (for other cells) supplemented with 1× nonessential amino acids (Invitrogen, Carlsbad, CA), 10% fetal bovine serum (FBS), 100 U of penicillin/ml, and 100 µg of streptomycin/ml. After 80 to 90% confluence was achieved, cell culture media were changed to maintenance media with 2% FBS and inoculated with 200 µl of clinical sample or 100 µl of passaged viral supernatant. Prior to inoculation, nasal samples were clarified by centrifugation at 4,000 × g for 10 min; lung tissue was homogenized using a tissue homogenizer in 5 volumes of buffer. Prior to passaging, cell culture supernatant was subjected to 3 freeze-thaw cycles and clarified as described above. Viral replication was monitored over 2 weeks by visual inspection under light microscopy for CPE. Supernatants of viral cell lysates were quantified using an endpoint dilution assay.

Deep sequencing libraries were prepared for whole-genome AdV sequencing using a variation of the TruSeq protocol (Illumina, San Diego, CA) (3). Briefly, nucleic acid extracts were randomly amplified to cDNA using a round A/B procedure as previously described (40, 41), and then digested using the restriction enzyme BpmI (New England Biolabs, Ipswich, MA) for 2 h at 37°C, followed by end repair and A-tailing with Klenow and Taq polymerase, respectively (Invitrogen, Carlsbad, CA). Size selection targeting 200- to 300-bp fragments was then performed using AMPure beads, followed by attachment of adapters containing 6-nucleotide barcode tags using DNA ligase. Final libraries were quantified using the Bioanalyzer DNA 12000 chip (Agilent) and SYBR Fast quantitative PCR (qPCR) system (KAPA Biosystems), pooled into a single lane, and sequenced on an Illumina HiSeq 2000 instrument (100-bp paired-end sequencing).

Raw deep sequencing reads were initially trimmed by removal of adapters, primers, and low-complexity/low-quality sequences. De novo assembly of partial AdV genomes was performed using the PRICE assembler (42). Despite high overall coverage (Fig. 2), de novo assembly of the entire AdV genome was not achieved because of (i) the significant degree of variation in coverage (especially for BaAdV-1) and (ii) the fact that PRICE is a “seed-based” assembler that assembles into contiguous regions and not a true de novo assembler. All gaps were subsequently filled by PCR using specific primers designed from assembled regions followed by Sanger sequencing. PCR and Sanger sequencing were also used to confirm the correctness of the assembly. After assembly of the full viral genomes corresponding to the 4 BAdV isolates, trimmed deep sequencing reads were then mapped to the AdV genome using Geneious software (43) with the following parameters (no gaps allowed, maximum mismatches allowed per read of 5%, and maximum ambiguity of 1).

The predicted coding regions in BaAdV genomes were identified using the fully annotated genome sequences of species F and G AdVs in GenBank as a reference. First, each BaAdV genome was aligned to the most similar reference genome in GenBank, followed by identification of open reading frames (ORFs) using Geneious. The selection of ORFs required the presence of an ATG start codon, a 100-amino-acid (aa) minimum size, and a match to a corresponding ORF in an annotated AdV reference genome. The GT-AG intron start-stop signal was used to predict the splice donor and acceptor sites for spliced genes. To confirm the accuracy of the predicted coding regions, each identified ORF was then aligned using BLASTx to a reference database consisting of all adenoviral proteins in GenBank. Whole-genome nucleotide pairwise identity plots (window size of 100) and amino acid pairwise identity calculations were performed in Geneious. Similarity and bootscanning plots were generated using Simplot (44), with a window size of 1,000 bp and a step size of 50 bp.

To construct the amino acid phylogeny trees corresponding to the hexon, penton base, DNA polymerase, and short/long fiber proteins, the translated protein sequences corresponding to representative human and simian AdVs in species A to G, SAdV-A, and SAdV-B, as well as nonprimate AdVs, were first downloaded from GenBank. Multiple sequence alignments were then performed using the FFT-NS-i×1000 algorithm of MAFFT at default parameters (45). A phylogenetic tree was constructed in Geneious using the Jukes-Cantor neighbor-joining method and 5,000 bootstrap replicates and a support threshold percentage of 25, using murine adenovirus 2 (MAdV-A) as an outgroup.

Viral stocks of BaAdV-1, BaAdV-2, and BaAdV-3 were generated by passaging in Vero E6 cells, aliquoted, and quantitated by endpoint dilution. To perform the virus neutralization assay, 100 µl of viral supernatant mixed with serum or control serum was incubated for 1 h at 37°C. After incubation, the mixture was inoculated into wells containing 4,000 Vero E6 cells per well and incubated at 37°C and 5% CO₂. The cells in the wells on the plate were observed every other day for evidence of CPE. For cells in wells that showed inhibition of viral CPE at the screening dilution of 1:10, the corresponding serum samples were diluted 2-fold from 1:8 to 1:128 and then retested. The reciprocal of the highest dilution where replicate well monolayers showed no CPE was taken as the neutralizing antibody titer.

Five pools of rabbit hyperimmune reference sera at the California DPH, collectively containing antibodies to human AdV species A to E, were available for testing. An individual rabbit serum sample reactive to HAdV-40 and HAdV-41 was also available for cross-neutralization testing of HAdV-F. For each pool, rabbit sera at a screening dilution of 1:8 and 100 µl of viral supernatant at a 50% tissue culture infective dose (TCID₅₀) of 10³/ml were mixed and inoculated into Vero E6 cells. Cells in wells on the plate were observed every other day for 2 weeks for evidence of CPE. For cells in wells that showed inhibition of viral CPE at the screening dilution of 1:8, the corresponding serum samples were diluted 2-fold from 1:8 to 1:128 and then retested.

Since neutralizing reference sera to human HAdV-52 (species G) was not available, the serum sample from baboon B108, shown previously to be positive for neutralizing antibody to species SAdV-B and SAdV-C AdVs (Table 2), was tested for cross-neutralization of HAdV-52 in an indirect neutralization assay. To perform the assay, 100 µl HAdV-52 supernatant at a TCID₅₀ of 10³/ml was mixed with serum from baboon B108 at the screening dilution of 1:8 and used to inoculate Vero E6 cells. The cells in wells were observed every other day for 2 weeks for evidence of CPE.

GenBank accession numbers for the adenoviral sequences used in Fig. 3 to 5 are as follows: CAdV-1 (canine adenovirus 1), AC_000003; HAdV-1, AC_000017; HAdV-3, DQ086466; HAdV-4, AY458656; HAdV-7, AC_000018; HAdV-8, AF532578; HAdV-12, NC_001460; HAdV-14, FJ822614; HAdV-18, GU191019; HAdV-21, AY601633; HADV-36, GQ384080; HAdV-40, NC_001454; HAdV-41, DQ315364; HAdV-48, EF153473; HAdV-49, DQ393829; HAdV-50, AY737798; HAdV-52, DQ923122; MAdV-2 (murine adenovirus 2), NC_014899; PAdV-3 (porcine adenovirus A), AC_000009; SAdV-1, AY771780; SAdV-3, NC_006144; SAdV-6, JQ776547; SAdV-7, DQ792570; SAdV-18, FJ025931; SAdV-20, HQ605912; SAdV-21, AC_000010; SAdV-22, AY530876; SAdV-48, HQ241818; SAdV-49, NC_015225; SAdV-50, HQ241820; SAdV-A1139, JN880448; SAdV-A1163, JN880449; SAdV-A1173, JN880450; SAdV-A1258, JN880451; SAdV-A1312, JN880454; SAdV-A1335, JN880456; TMAdV (titi monkey adenovirus), HQ913600; TSAdV (tree shrew adenovirus), NC_004453. The complete genome sequences of BaAdV-1, BaAdV-2, BaAdV-3, and BaAdV-4 (genetically identical to BaAdV-2) have been submitted to GenBank (accession numbers KC693021 to KC693024). Deep sequencing reads have been submitted to the NCBI Sequence Read Archive (accession number SRA067619).