Whole-genome sequencing of genotype VI Newcastle disease viruses from formalin-fixed paraffin-embedded tissues from wild pigeons reveals continuous evolution and previously unrecognized genetic diversity in the U.S.

Spleen, kidney and liver FFPE tissue samples from 14 Eurasian collared-doves and 3 rock pigeons, collected from 10 different U.S. mortality events between 2010 and 2016 were used in this study. Samples were collected from the following locations: Plymouth, Massachusetts (MA); Brewster, Texas (TX); Gallatin, Montana (MT); Lea, New Mexico (NM); Hidalgo, TX; Washington, Utah (UT); Dallam, TX; Montgomery, Pennsylvania (PA); and Garden City, Kansas (KS). Tissues were originally fixed in formalin, embedded in paraffin and evaluated by light microscopy and IHC [9] before being analyzed in this study. The same set of organs from one ECDO and one ROPI, negative for NDV and with an unrelated cause of death, were used as negative controls.

Five micrometers thick tissue sections were collected in 1.5 ml centrifuge tubes and immediately de-paraffinized by CitriSolv (Thermo Fischer Scientific, USA). Total RNA was extracted using RNeasy FFPE Kit (Qiagen, USA) as per manufacturer’s instructions. Briefly, de-paraffinized tissue was digested by proteinase K, treated with DNase, precipitated by 100% ethanol, adsorbed to an RNeasy MinElute spin column, washed twice and eluted in 50 μl RNase-free water directly from the spin column membrane. Total RNA was quantified using RNA High Sensitivity Assay kits (Thermo Fischer Scientific, USA) on a Qubit® fluorometer 3.0 and fragment size of RNA was determined using RNA 6000 Pico kit on an Agilent Bioanalyzer® (Agilent Technologies, Santa Clara, USA) as per manufacturer’s instructions.

Paired-end sequencing was conducted from cDNA libraries prepared from total RNA using KAPA Stranded RNA-Seq kit (KAPA Biosystems, USA) as per manufacturer’s instructions. Briefly, the protocol involved the synthesis of both the first strand and second strand of DNA from total extracted RNA by using random-priming, marking and A-tailing of cDNA fragments with 3′-adenine residues to allow ligation of 3′-thymine-containing adaptors, and random PCR amplification to create the DNA library. The fragment length distribution for each library was assessed using the Agilent High Sensitivity DNA Kit (Agilent Technologies, USA) on the Agilent 2100 Bioanalyzer (Agilent Technologies, USA). The Qubit® fluorometer and the Qubit® dsDNA HS Assay Kit was used to measure the concentration of the dsDNA libraries. All libraries for NGS underwent equimolar dilution and were pooled. The library pool was loaded into the 300 cycle MiSeq Reagent Kit v2 (Illumina, USA) and pair-end sequencing (2 × 150 bp) was performed on the Illumina MiSeq instrument (Illumina, USA). After automated cluster generation in MiSeq, the sequencing reads were processed by bioinformatics. Pre-processing and de-novo assembly of the raw sequencing data was completed within the Galaxy platform [20] using a previously described approach [17].

Phylogenetic analyses were performed using MEGA6 software (MEGA, version 6) [21]. Preliminary analysis was performed to infer the evolutionary history of genotype VI NDV. The sequences obtained here along with 82 complete genome concatenated coding sequences (CDS), 305 complete fusion gene sequences, and 1234 sequences of the initial 374 nucleotides (nt) of the fusion gene, which were available in GenBank, were initially analyzed (data not shown). Smaller datasets, including the most closely related viruses (complete genome CDS, n = 41, complete fusion gene n = 72, and 374 nt partial fusion n = 931) (see Additional file 1: Tables S1, S2, and S3), were parsed from the initial datasets and further analyzed. Determination of the best-fit substitution models was performed using MEGA6, and the goodness-of-fit for each model was measured by corrected Akaike information criterion corrected (AICc) and Bayesian information criterion (BIC) [22]. The final trees with 1000 bootstrap replicates were constructed using the maximum-likelihood method based on the General Time Reversible model for the complete genome CDS tree, the Tamura-Nei model for the fusion tree, and Kimura 2-parameter model for the 374 nt partial fusion tree as implemented in MEGA 6. For all analyses, the codon positions included were 1st +, 2nd +, 3rd +, noncoding, and all positions containing gaps and missing data were eliminated.

The estimates of average evolutionary distances were inferred using the complete fusion gene sequences. Analyses were conducted using the maximum composite likelihood model [23]. The rate variation among sites was modeled with a gamma distribution (shape parameter = 1). Previously described criteria based on the phylogenetic topology and evolutionary distances between different taxonomic groups were used to determine genotypes and sub-genotypes [24]. In addition, the USDA validated real-time RT-PCR assay primers and probe [25] used to detect virulent pigeon NDV were aligned with all available sequences obtained from pigeons in the U.S. utilizing ClustalW [26] as implemented in MEGA6.

Forty-eight FFPE tissue samples (kidney, liver, and spleen) from 17 pigeons, which were collected postmortem in seven states of the U.S., were sequenced using a target-independent NGS approach through random sequencing, and analyzed through the bioinformatics workflow described previously [17]. Twenty-nine of the 48 tissue samples analyzed by NGS were IHC positive [9]. From 17 birds submitted for sequencing, nearly complete genomes were recovered from 12 ECDO and 2 ROPI birds, which represents a total positive characterization rate of 82% (Table 1). Out of 48 tested samples, 23 nearly complete genomes were recovered, and they were all identified as members of genotype VI NDV. The identity among sequences obtained from different tissues of the same bird was 100%. In all sequences, the deduced amino acid cleavage site of the fusion protein was ₁₁₂RRKKR↓F₁₁₇, which is specific for virulent viruses. Those positive samples were 87.5% of the total kidneys, 40% of the total spleens, and 18% of the total livers tested. All negative control samples were negative using this NGS approach. The comparison with IHC showed that the 23 samples that were identified as positive for genotype VI by NGS, were also previously identified by Isidoro-Ayza, et al. as positive NDV samples by IHC (Table 1). Interestingly, six samples that were originally examined and classified as IHC positive (21%) were determined to be NGS negative with no NDV detected (see Additional file 2: Table S4). Aligning of the USDA validated real-time RT-PCR assay primers and probe [25] used to detect virulent pigeon NDV to all available sequences obtained from pigeons in the U.S. showed two to four mismatches in each of the oligonucleotides (data not shown).

The sequencing of 23 positive NDV samples produced a total of 8,548,074 raw reads, which were reduced to 2,624,877 (30.71%) reads once the host genome and the internal PhiX control were filtered out. The assembly analyses showed that 430,542 reads were mapped to genotype VI NDV. The mean fragment lengths ranged from 121 to 226 nucleotides and the assembled genomes had a percent coverage ranging from 97.34 to 99.86% (Table 1). The mean depth was 9–1498 with 19 out of 23 samples having a mean depth above 20. Detailed information on total and filtered number of reads and fragment length size for each individual sample is provided in Additional file 3, Table S5.

The evolutionary distances between the U.S. pigeon sequences and other genotype VI viruses were estimated. Seventeen sequences obtained in the current study from tissues sampled between 2014 and 2016 from Texas, Utah and Kansas were closely related to each other (99.8% mean identity within the group). These sequences were most closely related (96.8 to 98%) to sub-genotype VIa sequences from Texas and Louisiana from 2004, 2006 and 2010 and single sequences from Rhode Island (RI) (2000), Nevada (2003) and Minnesota (MN) (2007). Three sequences retrieved from birds sampled in Pennsylvania and Massachusetts in 2012 and 2014, respectively (99.8% mean identity between them), were most closely related to sub-genotype VIa sequences recovered from ROPI from Michigan (MI), Maryland (MD) and Pennsylvania in 2013 (99.3–99.4% nucleotide distance) and from pigeons and chickens from Pennsylvania, Minnesota and New Jersey (NJ) between 2004 and 2009 (98.2–98.6% distance). Three more sequences, obtained from a single ECDO sampled in Montana in 2010 and identical to each other, were closely related to sequences from the Northeast U.S. described above and also Connecticut (CT), Maine (ME), New York (NY), and Ohio (OH) during the same period.

In order to confirm the evolutionary relationship between the NDV sequences studied here and viruses isolated in other geographical regions, phylogenetic analyses using the complete genome concatenated CDS, complete fusion gene and 374 nt partial fusion gene sequences were performed. Currently, the accepted method for NDV classification uses the complete coding sequence of the fusion gene [24]. The isolates used in the phylogenetic analysis of the complete coding sequence of the fusion gene expectedly grouped together within currently designated sub-genotype VIa with the viruses that showed highest nucleotide sequence identity to them (Fig. 1). The 17 isolates from Texas, Utah and Kansas, that showed high genetic similarity within the lineage, formed a separate branch with other isolates from Texas, Louisiana, Minnesota, Nevada and Rhode Island that shared high genetic identity with them (Fig. 1). The remaining six sequences studied here clustered together in a separate branch with the sequences retrieved from pigeons and chickens sampled in the Northeast between 2004 and 2009 (Fig. 1). To assess the genetic diversity within sub-genotype VIa, the nucleotide distance between these two groups of sequences was estimated and they were found to be distantly related (4.2% nucleotide distance). In addition, both groups were distantly related (7.21–15.30% and 5.70–14.74%, respectively) to the rest of the sub-genotypes in genotype VI (Table 2 green and red, respectively) supporting the designation of a new sub-genotype, namely VIn, as per the nomenclature criteria for NDV put forth by Diel et al. [24]. The results from the complete genome CDS tree, although less representative due to lack of complete genome sequences in GenBank, show that the phylogenetic relationships identified in the complete fusion gene analyses extend to complete genome level (Fig. 2).

To study the epidemiological relations of the pigeon-origin NDV from the U.S. and the rest of the world, additional phylogenetic analysis was performed utilizing the 374 nt hypervariable region at the beginning of the fusion gene coding sequence (Additional file 4: Figure S1). The analysis was performed on this region due to the abundance of genotype VI 374 nt sequences and the scarcity of complete fusion gene sequences from Europe and China in GenBank. The obtained results suggest that the sub-genotypes VIa and VIn NDV circulating in the U.S. are not directly related to the ones circulating in Europe and Asia. Indeed, viruses from Europe (1987–2005) and the U.S. (isolation from Florida in 2008) grouped into a small branch (Additional file 4: Figure S1, group 1). However, the abundance of sub-genotype VIa viruses from the U.S. (isolated between 2000 and 2014 from North Carolina, South Dakota, PA, RI, TX, MN, NY, NJ, OH, MA, ME, MI, MD, and MT) formed a separate branch in the phylogenetic tree (see Additional file 4: Figure S1, group 2). A third group of sub-genotype VIa sequences was observed, consisting of sequences from Europe (1993–2000), South Africa (2005–2006), Japan (1995–1996) and the U.S. (single isolations from Maryland in 1998 and Florida in 2006) (see Additional file 4: Figure S1, group 3). A fourth branch separated from the latter and contained the isolates identified as novel sub-genotype VIn in the complete fusion gene analysis, along with viruses isolated from Texas, Arizona, Louisiana and Rhode Island between 2000 and 2010 (see Additional file 4, Figure S1, VIn).