Epitope Peptide-Based Predication and Other Functional Regions of Antigenic F and HN Proteins of Waterfowl and Poultry Avian Avulavirus Serotype-1 Isolates From Uganda

Uganda is a Newcastle disease (ND) endemic country where the disease is controlled by vaccination using live LaSota (genotype II) and I2 (genotype I) vaccine strains. Resurgent outbreak episodes call for an urgent need to understand the antigenic diversity of circulating wild Avian Avulavirus serotype-1 (AAvV-1) strains. High mutation rates and the continuous emergence of genetic and antigenic variants that evade immunity make non-segmented RNA viruses difficult to control. Antigenic and functional analysis of the key viral surface proteins is a crucial step in understanding the antigen diversity between vaccine lineages and the endemic wild ND viruses in Uganda and designing ND peptide vaccines. In this study, we used computational analysis, phylogenetic characterization, and structural modeling to detect evolutionary forces affecting the predicted immune-dominant fusion (F) and hemagglutinin-neuraminidase (HN) proteins of AAvV-1 isolates from waterfowl and poultry in Uganda compared with that in LaSota vaccine strain. Our findings indicate that mutational amino acid variations at the F protein in LaSota strain, 25 poultry wild-type and 30 waterfowl wild-type isolates were distributed at regions including the functional domains of B-cell epitopes or N-glycosylation sites, cleavage site, fusion site that account for strain variations. Similarly, conserved regions of HN protein in 25 Ugandan domestic fowl isolates and the representative vaccine strain varied at the flanking regions and potential linear B-cell epitope. The fusion sites, signal peptides, cleavage sites, transmembrane domains, potential B-cell epitopes, and other specific regions of the two protein types in vaccine and wild viruses varied considerably at structure by effective online epitope prediction programs. Cleavage site of the waterfowl isolates had a typical avirulent motif of 111GGRQGR'L117 with the exception of one isolate which showed a virulent motif of 111GGRQKR'F117. All the poultry isolates showed the 111GRRQKR'F117 motif corresponding to virulent strains. Amino acid sequence variations in both HN and F proteins of AAvV-1 isolates from poultry, waterfowl, and vaccine strain were distributed over the length of the proteins with no detectable pattern, but using the experimentally derived 3D structure data revealed key-mapped mutations on the surfaces of the predicted conformational epitopes encompassing the experimental major neutralizing epitopes. The phylogenic tree constructed using the full F gene and partial F gene sequences of the isolates from poultry and waterfowl respectively, showed that Ugandan ND aquatic bird and poultry isolates share some functional amino acids in F sequences yet do remain unique at structure and the B-cell epitopes. Recombination analyses showed that the C-terminus and the rest of the F gene in poultry isolates originated from prevalent velogenic strains. Altogether, these could provide rationale for antigenic diversity in wild ND isolates of Uganda compared with the current ND vaccine strains.

Uganda is a Newcastle disease (ND) endemic country where the disease is controlled by vaccination using live LaSota (genotype II) and I 2 (genotype I) vaccine strains. Resurgent outbreak episodes call for an urgent need to understand the antigenic diversity of circulating wild Avian Avulavirus serotype-1 (AAvV-1) strains. High mutation rates and the continuous emergence of genetic and antigenic variants that evade immunity make non-segmented RNA viruses difficult to control. Antigenic and functional analysis of the key viral surface proteins is a crucial step in understanding the antigen diversity between vaccine lineages and the endemic wild ND viruses in Uganda and designing ND peptide vaccines. In this study, we used computational analysis, phylogenetic characterization, and structural modeling to detect evolutionary forces affecting the predicted immunedominant fusion (F) and hemagglutinin-neuraminidase (HN) proteins of AAvV-1 isolates from waterfowl and poultry in Uganda compared with that in LaSota vaccine strain. Our findings indicate that mutational amino acid variations at the F protein in LaSota strain, 25 poultry wild-type and 30 waterfowl wild-type isolates were distributed at regions including the functional domains of B-cell epitopes or N-glycosylation sites, cleavage site, fusion site that account for strain variations. Similarly, conserved regions of HN protein in 25 Ugandan domestic fowl isolates and the representative vaccine strain varied at the flanking regions and potential linear B-cell epitope. The fusion sites, signal peptides, cleavage sites, transmembrane domains, potential B-cell epitopes, and other specific regions of the two protein types in vaccine and wild viruses varied considerably at structure by effective online epitope prediction programs. Cleavage site of the waterfowl isolates had a typical avirulent motif of 111 GGRQGR'L 117 with the exception of one isolate which showed a virulent motif of 111 GGRQKR'F 117 . All the poultry isolates showed the 111 GRRQKR'F 117 motif corresponding to virulent strains. Amino acid sequence variations in both HN and F proteins of AAvV-1 isolates from poultry, waterfowl, and vaccine strain were distributed over the length of the proteins with no detectable pattern, but using INTRODUCTION Avian Avulavirus serotype-1 (AAvV-1) is a member of the genus Avulavirus in the family of Paramyxoviridae that causes ND in poultry. The antigenic serotypes evolving in this group under environmental or vaccination pressure, evade the immune system of birds and account for vaccine failures (1,2). The AAvV-1 serotype displays a great genetic and antigenic diversity in the fusion (F) and hemagglutinin neuraminidase (HN) genes that encode for the structural envelope spike proteins playing host recognition, infection, and pathogenesis roles, but importantly influencing the antigenicity and immunogenicity of the ND viruses (3)(4)(5)(6).
Infectivity assays of ND viruses employ the use of animal models (1-day-old Gallus gallus chick) which requires a lot of time and monetary inputs. Low-virulent strains are defined by the monobasic amino acid sequences at positions 112-113 and 115-116 of the C-terminus of fusion protein cleavage site (FPCS) with leucine (L) at position 117 and/or intracerebral pathogenicity indices (ICPI) of ≤0.7 (7,8). Such strains induce antibody titers sufficient to prevent ND with the infected birds frequently remaining asymptomatic or presenting mild respiratory form of ND. The FPCS together with the hypervariable regions of the F protein are the basis of phylogenetic analysis of pathotypes (9). Genotypes and lineages emerge among other typing methods that classify NDV strains (10,11). To date, two classes I and II, are recognized and each further subdivided into three different subgenotypes (1.1.1, 1.1.2, and 1.2) and 21 clades or subgenotypes (I-XXI), respectively (9,10,12,13). In order to avoid discrepancies of nomenclature systems and incorrect assigning of genotypes, the most recent classification emphasizes the use of full F gene to classify these viruses further (6,12,14). Notably, the class II genotype VII viruses are the most commonly reported in ND outbreaks in poultry, pet, and wild birds throughout the world, while class I which is commonly isolated from waterfowl, shore birds, and some poultry is less virulent and is exploited for potential vaccine candidates (10).
The potential targets for immune system response to ND virus during ND infection are the two membrane-anchored glycoproteins F and HN, also known for cell-binding and infection (6,15). Genomic and antigenic differences between wild isolates and vaccine strains are among the reported causes of ND vaccine failure. These differences accrue from several cumulative mutations at the F and HN genes of the wild strains in response to vaccination pressure (1,13,16). Enzymatic restriction of the viral precursor F 0 molecule by the host enzymes in the respiratory and intestinal tracts facilitates replication of virulent viruses in these tissues (17). The conformational change induced in the precursor F 0 protein after proteolytic cleavage and the suspension of two disulfide-linked subunits of the hydrophobic N-terminus of F1 and the C-terminus of F 2 fragments off the membrane initiates the cellular fusion interaction. Thus, structural protein changes produced by point mutations involving one or two amino acids or the changes resulting from recombination at the F protein antigenic site of these viruses circulating in an endemic wild birds or poultry are implicated in ND outbreaks (13,16,18).
In ND pathogenesis, HN initiates infection while Fglycoprotein mediates viral attachment and penetration into host cells (19). Both proteins induce host-immune response and are required for producing neutralizing antibodies induced by vaccines. Antibodies against F proteins have been demonstrated in in vivo study as importantly necessary in neutralizing the ND infectivity (20,21). In fact, the passive immunization of chicks using antibodies directed to the two antigenic sites I and II of F protein, which reacted with antibodies 83(F) and 313(F), respectively, have demonstrated complete suppression of viral growth and death of chicks (22). Seven major F protein neutralizing epitopes involved in fusion inhibition and neutralization are shown at specific residues 72, 74, 75, 78, 79, 157-171, and 343 for epitopes A1, A2, A3, A4, and A5, respectively. The amino acid residues show that both F 1 and F 2 are involved in the formation of a single antigenic site vital in the structure and function of the active F epitopes (23,24). In the HN protein, five conformational and linear antigenic sites are crucial in inducing antibodies at amino acid residues of 193-201, 345-353, 494, 513-521, and 569 (25,26). Amino acid changes in the linear epitopes of wild isolates from glutamate (E) to lysine (K) at position 347 have been known to alter the character of HN protein making some strains of ND viruses escape from epitope recognition by mABs (27).
Genomic changes in the non-segmented, negative-sense RNA viruses are known to arise from the inherent error rate of polymerase, generating a large number of variants in the nonconserved proteins upon which nature selects suitable changes in the viral genome (28). Lower recombination rates are also reported in HN and F genes of non-segmented, negative-sense AAvV-1s, that account for strain variation (29,30). However, strain-specific variations in AAvV-1s, arising from mutations are available in the public domain repositories. Nonsynonymous mutation rates have been associated with the codons at the cleavage site motif of the F protein resulting into diversity and classification of the strains. The mono-or multibasic amino acid tendencies at this site suggest that increase in virulence increases the rate of virus evolution (28,31). Wild aquatic birds are considered to be natural reservoirs for ND viruses and constituting a threat in spreading these recombinant viruses to the domestic birds as they come into contact with them along their migration pathways across the world (32). Studies have revealed similarities between strains recovered from aquatic birds and shorebirds with those isolated from live bird markets (LBMs) in some parts of the world (10).
ND is known to be endemic in Uganda; phylogenic analysis involving F and HN sequences of wild isolates from Uganda have indicated unique ND viruses in relation to those found elsewhere in the world (33,34). Understanding the forces of adaptive evolution or increased genetic variation in both virulent and non-virulent AAvV-1s can help prevent future outbreaks. Currently, two live monovalent ND vaccines (LaSota and I 2 ) are available in Uganda for the control of ND infection. Due to the frequent and recent ND outbreaks, there is a renewed interest to understand the forces that drive positive or negative (purifying) pressures on specific amino acid sites within surface proteins. Importantly, the possible continued cocirculation of the virulent AAvV-1s in the apparently healthy-looking birds marketed at LBMs (33) needs to be studied. Recently, we isolated thermostable, low-virulent ND viruses in waterfowl and compared their immunogenicity with those of the vaccine (LaSota and I 2 ) strains. We challenged the immunized birds with a circulating wild virulent strain and noted some protection failure rates (35).
Solutions to ND vaccine failure rates and the observed presence of virulent viruses harbored by healthy domestic birds can only be envisaged by establishing the role of selection pressures and virulence on evolutionary change of the NDV F protein and comparing the immune-dominant epitopes of the currently used vaccine strains with that of wild viruses using in silico approaches. The amino acid patterns, which are apparently hidden at sequence level become evident when mapped onto experimentally derived 3D structure. Carrying out 3D mapping of the amino acid patterns provides timely and inexpensive insights into antigen homology studies for ND control. In this study, we focused on the HN and F proteins of the Ugandan AAvV-1s, which are the main surface proteins important in viral entry, carrying functionally important structures like signal peptide, cleavage activation sites, and B-cellmediated antibody response. Additionally, the sequential and conformational variations at the hydrophobic and hydrophilic sites and transmembrane sites establish the diversity of wild ND viruses and the current vaccine strains.

Viral Sequence Compilation
For analysis, 70 and 55 selected sequences of the F and HN protein were used, respectively. Of the 70 F protein, 40 were retrieved from the GenBank database, comprising 25 described earlier in our pathogenicity and evolution study of Avulavirus serotype-1 isolated from poultry sampled from LBM in Uganda (33). Three ND viruses isolated from waterfowl in 2011 (36), one sequence described 10 years earlier in Uganda (34), and 11 sequences representing various class II NDV genotypes were used. The rest of the 30 sequences were partial F gene sequences of viruses isolated from freshly voided fecal matter of several species of waterfowl including the long-tailed cormorant, grayheaded gull, white-winged tern, little egret, greater cormorant, Egyptian goose, slender-billed gull, African open-billed stork, black-headed heron, gull-billed tern, African jacana, yellowbilled duck, black crake, yellow-billed stork, sacred ibis, pinkbacked pelican, long-toed plover, pied kingfisher, African darter, hadada ibis, cattle egret, fish eagle, hamerkop, white-faced whistling duck, malachite, common squacco, Terek sandpiper, and wood sandpiper. Accordingly, all the isolates from LBM were highly pathogenic and of low evolution placed in genotype V in the new classification system described by Byarugaba et al. (33). Similarly, of the 55 HN protein sequences, 32 were retrieved from the GenBank database as part of our poultry LBM isolates during our pathogenicity study and evolutionary study (33) Table 1.
Sampling, Isolation, Confirmation, and Sequencing of Waterfowl AAvV-1s Isolates Sampling, virus isolation, confirmation, and sequencing of the waterfowl AAvV-1 isolates were done as described by Byarugaba et al. (33) and Wanyana et al. (36). Samples of freshly voided fecal matter were swabbed using sterile Dacron swabs from land/rock surfaces into cryovials containing viral transport medium supplemented with antibiotics (isotonic PBS, 2,000 U/ml penicillin, 2 mg/ml streptomycin, 50 µg/ml gentamycin, 50 U/ml nystatin, and 0.5% BSA). The sample vials were transported in dry shippers to the laboratory and stored at −80 • C until analysis. Each sample was inoculated (in triplicate) by the allantoic route into 9-10-day embryonated chicken eggs (ECEs) for virus isolation according to the OIE Manual of Standards  for Diagnostic Tests and Vaccines (37). Hemagglutinin inhibition tests, specific RT-PCR test, RNA extraction, cDNA synthesis, and sequencing were made on the harvested allantoic fluid (AF) accordingly. The AF, harvested 3 days post-inoculation was tested for hemagglutination (HA) using 1% chicken erythrocytes and ND virus confirmed by hemagglutination inhibition (HI) with an in-house rabbit-generated polyclonal anti-AAvV-1 sera as described (37). The HI-positive samples were confirmed further by RNA extraction and cDNA synthesis by Qiagen one-step RT-PCR that extracts and amplifies the F and HN genes according to the conditions described by Byarugaba et al. (33). The amplified cDNAs of the NDV-positive extracts were separated, purified on 1% Agarose gel then visualized and documented in a Bio-Rad Gel Doc XR imager. Sequencing was done by Sanger sequencing method on a 3130XL Applied Biosystems capillary sequencer at the Plateau de Genomique GeT-Purpan, UDEAR UMR 5165 CNRS/UPS, CHU PURPAN, Toulouse France. The positive samples were also biologically characterized using standard pathogenicity tests (mean death time, intracerebral pathogenicity test, and intravenous pathogenicity test described by 33). In vivo evaluation of immunogenicity and protection efficacy on four of these AAvV-1 isolates was performed previously as described in Omony et al. (35). These waterfowl AAvV-1 isolates were sequenced for partial F gene targeting the cleavage site and the hypervariable region of the F gene. No sequenced data for the HN gene of these isolates were availed for this study.

Sequence Data and Phylogenetic Analysis
Both nucleotide and amino acid sequences of F protein were individually retrieved from the NCBI GenBank in fasta format, matching those present in viral zone UniProtKB/Swiss-Prot entries using accession numbers. These sequences were edited to equal length and aligned using ClustalW provided in Bioedit software version 7.0.9 (38). Nucleotide similarity calculated using MegAlign Pro (DNASTAR program for life sciences, version 9, Inc., Madison, WI, USA). The waterfowl partial F sequences were aligned separately and converted into protein sequences using Emboss software (EMBL-EBI). Phylogenetic analysis was performed by Bayesian methods in MEGA version X program (39) and or Geneious Prime 2020 with the NJ Kimura 2-parameter method and 1,000 bootstrap replicates. The phylogenetic analysis was done on nucleotides based on the pilot tree proposed by Dimitrov et al. (12) to maintain the tree topology and to ascertain the genotypes of the waterfowl isolates. The same nucleotide sequences were later translated into amino acid sequences in MEGA X program v10.1.8 to compare our waterfowl, poultry isolates, and vaccine (LaSota: AY845400 and I 2 : AY935499) strains at amino acid levels.

Generation of Structural Homology Models
To determine the conserved regions, homology models were generated for the translated F and HN proteins by aligning sequences for each gene separately using multiple sequence alignment (MSA) with the aid of ClustalW as implemented in the Bioedit program version 7.0 (38) and or Jalview. The consensus regions for each protein in each group (poultry, waterfowl, and vaccine) were used in a BLAST search against the protein Data Bank (PDB), to identify the available homologs or orthologous of known proteins. Structural homology modeling was also done using tools of SWISS-MODEL modeling server and viewed by deep viewer v4.01 (40). Overall, PDB entries 3MAW and 3T1E were used as templates for the F protein and HN protein, respectively. The experimental 3D structure (3MAW) of F protein from ND virus had identity and similarity scores of LaSota (71.1 and 74.3%), waterfowl (71.6 and 74.3%), and poultry (72.3 and 74.5%), respectively. Selected templates were subjected to chimeras to replace amino acids with Avulavirus F and HN protein consensus sequence. Energy minimization was carried out with YASARA force field in YASARA version 18.4.24, to obtain the most stable local minimum protein conformation.
Later on, correctness of models was established using PROSTAT (module in homology) and PROCHECK (41), and the candidate epitopes were analyzed by different predication tools for structurally conserved regions (SCRs), loops, and accuracy.

Mapping of Mutations, Variable Positions, and Specific Functional Regions
Following MSA, sequence variation was mapped onto the protein structures and entropy calculations with the aid of Scop3D, a tool, which visualizes variations across multiple sequences on the protein structures (43). F protein and HN protein numbering was based on LaSota using GenBank accession numbers AY845400 and P35743, respectively. The functional regions were defined based on literature. All functional regions were mapped on to the structures and Jalview or Chimera-analyzed models for diversity as visualized to the predicted structure models. Also using aligned sequences, single nucleotide polymorphisms (SNPs) were detected by trimming all gene reads and assembled to LaSota (AY845400) using Geneious assembler in Geneious Prime 2020 program (version11.0.4). During the analysis, phylogenetic tree of isolates based on selected functional regions were considered using Jalview.

Determination of F or HN Protein Recombination Among Isolates
F and HN nucleotide sequences were separately aligned using MegaAlign software (DNAStar, 5.01) and analyzed for recombination using the split decomposition method in seven local statistical recombination tools (RDP4, GeneConv, BootScan/RecScan, MaxChi, Chimera, LARD) integrated in the RDP4 program version 4 (44). These tools were used to estimate any recombination event, recombination hotspot, recombination rate plots, etc., and to detect any putative recombination breakpoint. These methods were applied using the following parameters: window size = 20, highest acceptable p-value < 0.001, and Bonferroni correction. For reliable results, any four putative recombination events were detected and translated into corresponding amino acid sequences.  Table 1. Others were the 27 F gene waterfowl isolates characterized for fusion cleavage sites and were not deposited in the GenBank database.

Animal Ethics
The

Phylogenetic Analyses Based on the Fusion Gene of AAvV-1s
The phylogenetic analyzes was performed on the full F gene (1,659 nucleotides) encoding for full F protein (553-amino acids), including the hypervariable and cleavage site regions of partial F gene (603 nucleotides) of isolates from waterfowl, encoding for a 201-amino acid protein. In order to compare the genetic relatedness of NDV isolates from waterfowl and poultry used in this study to the vaccine strains and other NDV genotypes, the phylogenetic analysis was done by comparing our isolates with the already known F sequences of class II genotype. The analyses inferred two subclusters of AAvV-1 isolates from poultry and waterfowl. These virus isolates clustered differently from the commonly used vaccine strains and historic strains available in the GenBank database. Notably, the phylogenetic tree clustered AAvV-1 isolates from waterfowl (n = 29) differently from genotype II strains, which had been historically described as genotype II. This was in exception of the virulent strain NDV/WF/MU0118/2011 isolated from Queen Elizabeth (QE) park, which clustered together with isolates from poultry (n = 25). Overall, the AAvV-1 isolates from waterfowl clustered differently from the isolates derived from poultry, which were recently subtyped in genotype Vd as shown in Figure 1.

Predicted F Protein 3D Structure, Functional and Antigenic Domains
The complete translated fusion protein amino acid sequences from poultry isolates spanning cDNA coding sequence (CDS) of 553 amino acids and the partial fusion protein from waterfowl virus isolates of 182 amino acids were used to predict 3D fusion structure and to assess the functional and antigenic domains relative to LaSota vaccine strain. Comparing these sequences at hypervariable region, proteolytic cleavage site (F 0 ), N-terminus of the fusion protein (F2), cleavage motif, transmembrane regions, and amino acid involved in neutralization sites yielded   Table 2. Deduced amino acid sequences of the F gene cleavage site (F 0 ) in all domestic poultry isolates contained multiple basic amino acid residues. The conserved amino acid residues 112 R-R-Q-K-R 116 , with phenylalanine at position 117 between Cterminus of F 2 and N-terminus of F 1 fragment of this gene confirmed the poultry isolates to belong to a virulent group of viruses. All the waterfowl AAvV-1 isolates had the conserved amino acid residues 112 G-R-Q-G-R'L 117 at the cleavage site with the exception of one NDV118/WF/UG/2011QE. This was a characteristic of the avirulent strains like the vaccines strain currently in use in the country.
In both poultry and waterfowl F proteins, the six potential glycosylation sites were conserved. These sites had the motif asparagine (N)-X-serine (S)/threonine (T) where X is any amino acid except of proline (P) and aspartate (A). These six asparagine-linked glycosylation sites were at 85NRT, 191NNT, 366NTS, 447NIS, 471NNS, and 541NNT. Although the 12cysteine residues reported at 25,27,76,199,338,347,362,370,394,399,401, and 424 were conserved (Figure 2A), some of them were not available for the partial F protein analyses in waterfowl virus isolates. Compared with the representative vaccines used in Uganda, the CDS of poultry and waterfowl NDV isolates, had a number of mutations at the fusion peptide (117-142), hydrophilic region A (143-185), hydrophilic site B (268-299), hydrophilic region C (471-500), and transmembrane domains of F protein ( Table 3). The substitution especially at the I508A, I509V, V513L, I516V, L517T, and I520V in the transmembrane site suggests limited conservation at this site in the F gene.
At structural level, by mapping sequence variation in poultry isolates onto the homology model PDB-entry 3MAW built on, F protein which is the F 0 protein for Avulavirus serotype−1, > 80% sequence identity was observed. The predicted 3D structure of poultry AAvV-1 isolates was similar to that of reference LaSota (RSMD, Cα = 3.203). To map these variations onto the PDB   4  8  9  11  13  14  16  17  18  19  20  21  22  26  28  29  30  32  44  69 3MAW_A chain - Variable positions found in the fusion protein mutated in poultry and waterfowl compared with LaSota vaccine virus strain-AY845400 used as a reference. Bold positions are surface-exposed residues in the variable situation predicted by discrimination of protein secondary structure class. Underlined positions contribute to the hydrophobic stability of the protein. ANT, antigenic region. T All these mutations were in only two waterfowl isolates but conserved in other waterfowl and in all poultry isolates.  model and perform entropy calculations, we used Scop3D and Chimeras for sequence manipulations and final visualization was done in Discovery Studio Visualize v20.1.0. We confirmed that the seven residues involved in virus neutralization from a stretch of 157-171 and individual residues D72, E74, A75, K78, A79, and L343 were conserved across all poultry virus isolates compared to LaSota. Two AAvV-1 isolates from waterfowl; NDV/WF/MU131/LUT and NDV/WF/MU159/MUS showed 13 amino acids variability at a stretch of 157-171 site with a shift from more hydrophobic to hydrophilic amino acid residues seen in LaSota/poultry and to the two waterfowl isolates respectively (Table 3; Figures 2A,B). We predicted linear B-cell epitopes and their surface accessibility and antigenicity scores using the BepiPred linear Epitope prediction method, Emini surface accessibility tool, and Kolaskar and Tongaonkar antigenicity, which use epitope scores and immunogenicity predictions through IEDB online (www.iedb.org) facilities. Both BepiPred-2.0 prediction tool and Vaxijen 2.0 tool gave 13 effective B-cell epitopes for LaSota vaccine and 11 for chicken isolates. Two effective epitopes at position 271-293 and 451-458 were undetected in chicken. Overall, these gave seven nonoverlapping sites above antigenicity score (0.8) and surface accessibility score (0.6); 33 485 . All these were conserved in all the vaccine strain, poultry and waterfowl ND virus isolates except for the two waterfowl isolates NDV/WF/MU159/MUS and NDV/WF/MU131/LUT having substitution at amino acid 164-167. However, structure-based antibody prediction based on DiscoTope of the F protein consensus in poultry isolates showed four B-cell epitope locations ( Table 4). Linear epitopes of F consensus for poultry and waterfowl isolates at (1-182 amino acid) indicated 16 substituted positions; L32I, M68L, E82D, R112G, K115G, S/D124G, V135I, N145K, K146Q, L451Q, T453S, V457I, D479N, A482E, N486R, and N489D, suggesting that these surface-exposed amino acid residues do modify the effectiveness of the predicted epitopes thus are speculated for the antigenic differences between these wild viruses and vaccine group.
Organizing these variations and others into functional domains like cytosolic, transmembrane, and flanking regions are shown in Table 5. However, assembling the HN nucleotide sequences on the CDS of LaSota vaccine (AY845400), all the 32 reads predicted only four polymorphic amino acid changes that had effect on the HN protein shown as V35M (75.8%), M48V (36.4%), A66T (81.8%), and S440N (48.5%) with codon changes G103A, A142G, G196A, and G1319A, respectively. Importantly, 28 charge position variations observed were due to substitutions between the poultry isolates and LaSota strain sequences from the model (PDB-3T1E). Positions 42A→S and 43S→I are located in the transmembrane domain affecting the charge and physiochemical properties of HN protein in poultry isolates. 84I→V and 98K→S lie proximately at the hydrophobic and hydrophilic sites predicted at MHC I alleles and linked to known T-cell epitope affecting the charge of the peptide. Comparison of the HN antigen-exposed surfaces and the predicted potential B-cell epitopes of consensus sequence of isolates from poultry and LaSota strain using the BepiPred-2.0 prediction tool on IEDB server, 25 epitopes were discovered. Analysis of their antigenicity with Vaxijen 2.0 tool and accessibility with Emini Surface Accessibility Prediction tool showed 15 out of 25 effective epitopes. Six (positions: 555-566, 108-115, 526-537, 236-254, 370-381, and 147-166) out of 15 effective epitopes were similar to those in LaSota vaccine strain. The other   Antigenic site  Antigenic   328  329  342  369  387  404  432  440  453  464  466  474  476  477  479  484  495  502  508  540  570   LaSota strain  T  V  D  I  V  I  S  S  I The underline amino acid positions alter markedly the charge of the HN protein of the 32 poultry isolates analyzed. Variations at Cysteine-123 and N-glycosylation-508 positions are high-lightened.
The original five N-glycosylation sites in the HN protein of the isolates from poultry were all conserved between themselves at positions 119NTS, 341NNT, 433NKT, 481NHT, and 538NKV but with differences in the patterns of amino acids at three positions 119NSS, 341NDT, and 538NKT when compared with LaSota and I 2 vaccine strains commonly used in Uganda. N-Glycosylation reported at amino acid position 508 was absent in all our isolates from poultry but replaced by 508N→Y. Comparisons of the HN protein total length of vaccine strain with wild isolates revealed several point mutations at transmembrane domains including the hydrophobic (HRa) and hydrophilic (HRb) sites and in the six continuum of antigenic sites 2, 3, 4, 12, 14, and 23 that form 3D conformation of the HN molecule ( Table 6).

The F and HN Protein Recombination Among Isolates
Using all recombination detection methods in RPD4, there was no recombination detected while analyzing the HN gene of both isolates from waterfowl and poultry compared within, among themselves, or with vaccine strain used in Uganda. However, using F gene of individual groups or combined in the presence of LaSota, I 2 and I 2 progenitor as reference vaccines, there were two breakpoints that signaled the recombination events which were attributable to an evolutionary process in isolates of waterfowl using the stringent criteria of any four recombination detection methods. The seven statistical methods used in the RPD4 were automated RPD(R), GeneConv (G), Bootscan (B),  MaxChi(M), Chimaera (M), SiScan (S), and 3SEQ (T). Several potential recombination points were detected; two of these in the isolates from waterfowl and one isolate from poultry are shown in Table 7.
Using only GeneConv to search for the possibility of significant recombination, which uses the homologous segments, we detected few recombination breakpoints in the isolates of poultry. By using less-stringent criteria of two recombination detection methods, we found 25 statistically significant recombinant fragments in isolates from poultry, observed at lengths 11-437 nt and 439-1,659 derived from major parent (LaSota: 93.9% similarity) and 12-438 nt regions derived from minor parent (NDV/C/UG/MU032/2011IGA: 100% similarity) in the fusion gene producing a recombinant strain NDV/C/UG/MU074/2011KOB. At the same time, 30 recombinant fragments were observed in isolates from waterfowl at lengths 10-438 to 438-546 nt and 439-1,654 nt derived from minor parent (LaSota: 99.2% similarity) and major parent (NDV/WF/MU159/MUS: 100% similarity) respectively producing a recombinant strain NDV/WF/MU131/LUT. However, only two most important breakpoint events compatible with recombination with a beginning breakpoint at nucleotide positions 316 to 436 nt and 317 to 435 nt using the five recombination detection methods (Figure 3A: A-E) were identified in the F gene of waterfowl NDV isolate NDV/WF/MU118/QE isolated from a conservation area (Queen Elizabeth National park). Both FPCS and intracerebral pathogenicity indices showed the daughter virus to be virulent. They also indicated that it appeared to have arisen from a recombination between a non-virulent NDV/WF/MU159/MUS and LaSota vaccine with statistically significant probabilities (p < 0.01). The lines in open rectangles seen in the GeneConv plots ( Figure 3A: B) connect the probable beginnings and ends of the fragments that underwent recombination.
Recombination signals were also detected between I 2 and I 2 progenitor vaccines and the NDV isolates from waterfowl at varying lengths along the fusion gene. GeneConv in the RDP sorted and scored all the significant fragments in the alignment and listed them in decreasing Bonferroni-corrected p-values. Testing this assortment of sequences provided a maximum parsimony pedigree suggestive of recombination. However, the exact point of fragmentation together with the relationships of the ND viruses waterfowl isolates involved can be identified by the tree-like network. The p-values produced guided in the identification of the fragments and locations of recombination as shown in Figure 3B.

DISCUSSION
The economic impact of ND on both the backyard and commercial poultry industry in Uganda and elsewhere is significant. The annual disease resurgence, vaccine failures, and potentially virulent strains circulating in apparently healthy birds become an issue. The concerns of viral evolution, type of vaccine used to protect flocks, and post-vaccine evaluation are always raised. This study evaluated the predicted B-cell epitopes and functional domains of F and HN proteins in AAvV-1 isolated from Ugandan aquatic birds and poultry and compared them with those of the vaccines to determine variations between the two groups that might explain ND vaccine failures. To date, AAvV-1 vaccine evaluation is based on empirical crossprotection of birds. Many of such studies indicate an increasing evidence that antigenically matched ND vaccines provide better immunity (45,46). However, such studies are costly and laborious and require extended field studies. Notwithstanding other factors that might lead to vaccine failures, insufficient cold-chain maintenance, insufficient immunity titer, hygiene status, etc. have been of great concern. Here, we focused on in silico study of the neutralizing epitopes of HN and F-glycoproteins of endemic wild ND viruses and vaccine (genotype II) to ascertain the presence of virus variants and any recombinants at these two antigenic and surface glycoproteins in our AAvV-1 isolates that could result into inefficient vaccination.
In Uganda, vaccination using live avirulent LaSota and recently introduced I 2 strain, belonging to genotypes II and I, respectively, of class II offers the best ND control. Phylogenetic analysis of AAvV-1s reported in the two previous independent studies showed that the partial and full F gene sequences of AAvV-1 isolates from waterfowl (36) and poultry (33) clustered Ugandan viruses with those of genotypes II and Vd viruses, respectively. This is consistent with previous reports of a predominance of genotypes II and V viruses in wild and domestic birds. However, based on the current unified system of classification described by Dimitrov et al. (12), the subgenotype Vd becomes V due to lack of the branch support. However, phylogenetic analysis of the same viruses combined in the current study showed two subclusters; one cluster of genotype V where poultry belonged and other (unassigned) where waterfowl isolates belonged. No virus isolates from waterfowl clustered with genotype II as previously reported except that one virulent (NDV/WF/MU118/2011QE) strain clustered with poultry isolates (genotype V). Although knowledge of circulating genotypes established through phylogenetic analyses is important in epidemiology or evolutionary changes of NDV, little about the antigenic diversity at the B-cell epitopes is gained from these tree analyses, suggesting that selection of neutralizing variants arising from mutations or fragments of gene recombination implicated in vaccine failures have to be identified. Besides, the proper genotype to where the AAvV-1 isolates from waterfowl are placed in class II viruses warrants investigation.
The importance of antibodies directed toward both F and HN epitopes that strongly neutralizes ND infectivity is important for understanding the diversity of these genes in wild virus isolates and vaccine strains. In this study, we demonstrated amino acid substitutions in the entire length of F and HN proteins in AAvV-1 isolates from waterfowl and poultry as compared with vaccine strains used in Uganda. Analysis of the hypervariable site of F protein revealed two groups, one with multiple basic acid residues at the proteolytic cleavage site 112 RRQKR 116 F 117 at the N-terminus of F2 and F1, respectively, where viruses from poultry belonged. The other group had 112 GRQGR 116 L 117 where those from waterfowl and vaccine strains belonged. This confirms the incidences of AAvV-1s in susceptible domestic and  feral bird species and suggestive of the potential role of feral birds in transmitting these viruses to poultry. The reported (RK/RQRR↓F) motif cleavage site among AAvV-1 isolates within the same genotype from domestic poultry and other wild hosts have been attributed to native selection, host pressure selection, viral evolution, or wide vaccination (14). There was no evidence of variation of pathogenicity among strains in paired samples isolated from cloaca and oropharynx of a single bird (33), and the presence of AAvV-1s with different pathogenicity based on the cleavage site motif within one geographical area suggests a long time geographical isolation of these different pathotypes. Waterfowl harboring viruses with lentogenic F-cleavage motif site generally, except for one isolate, and healthy-looking poultry harboring virulent AAvV-1 strains confirmed by the RRQKR↓F F-cleavage motif site, suggests that the two groups differ in their fusion activity and virulence in their hosts (31). This is of great importance and should be noted by both epidemiologists and vaccinators since variability in amino acid residues at the FPCS of viruses existing in the same geographical location suggests presence of different viral genotypes and pathotypes and potential for vaccine failure (47). The presence of avirulent ND viruses in waterfowl indicates that wild birds may be a potential source of virulent AAvV-1 for poultry due to possible point mutations at FPCS (7,48). At the same time, these wild birds might therefore serve as a source of potential vaccine isolates (35). Conservation of F and HN proteins at N-glycosylation, cysteine sites, and other specific structural and functional sites like receptor (sialic acid) binding sites between AAvV-1 isolates from waterfowl and poultry together with vaccine strains indicate these are required for biological active state especially among amino acids residues involved at the direct interactions that form a high-level structural integrity (49,50). The substitutions particularly in the surface-exposed amino acid residues in signal peptide (position: 13-29 aa) is not surprising considering that the amino acids in this region undergo constant positive selection making them highly variable. However, the mutations in the fusion peptide (position: 117-142 aa) is least expected except for the two nucleotides necessary for a change from a trypsinlike to a furin-like site. The exceptional changes at the nearby position of this site especially at V118I, V121I, S124G, V134I, and N145K in isolates from poultry and D124G and A129V in AAvV-1 isolates from waterfowl warrant further investigation in view of recombination detected in this region. The hypervariable hydrophobic regions (HRa, HRb, and HRc) especially I269V,  T288N, I330L, T385A, Q451L, S453T, I457V, and K494N whose side chains are surface-exposed (51), together with T508A, I509V, V513L, and I520V of the transmembrane region of F protein are implicated in structural variations between the wild and vaccine strains. The 13 observed mutations in the F protein of the two AAvV-1 isolates from waterfowl at the known antigenic aa position 157-171 account for variation. These favored more of hydrophilic aa residues (n = 5) at positions N157S, K162N, K167E, N170D, and E171G than hydrophobic aa residues (n = 2) at positions M158I and I169T while other aa were neutral in nature (n = 6). Only K167E and E171G of the hydrophilic aa decreased the protein stability by the predictor of effects of single-point protein mutation in I-Mutant Suite. Taken together, increased antigenicity and the neutralizing epitopes of F protein of these two viruses could result into altered fusion activity and neutralizing escape variants (52,53). The implications of these changes in the critical regions of F and HN genes (neutralizing epitopes) could be serious as they may mean escape of the virus from immunity.
Structurally, the seven and five major epitopes on the F protein and HN proteins, respectively, have been mapped on to experimental 3D structure in the PDB. Our findings located these functional residues and found those of F protein to be more conserved at positions 72, 78, 79, 343, and 157-171 in all virus isolates from poultry and vaccine strains emphasizing the role of mABs elicited by this glycoprotein in the neutralization of NDV (54). However, the 13 substitutions occurring in two virus isolates from waterfowl at a stretch of 157-171 aa between LaSota strain, displaying mutations N157S, K162N, K167E, N170D, and E171G, which are hydrophilic and exposed on the outer periphery of the protein surface had improved antigenicity of these epitope which are necessary for antibody binding and are crucial for the AAvV-1 isolate antigen diversity. This suggests cocirculation of these novel viruses similar to those identified earlier as vaccine candidates (35). On the other hand, the eight substitutions in the HN protein-neutralizing epitopes together with other functional sites between LaSota strain and AAvV-1 isolates in this study could explain why these are antigenic variants which fail to be neutralized by antibodies produced in healthy poultry birds vaccinated with LaSota vaccine and sold in the market.
However, our previously reported vaccination results in Omony et al. (35) did not fully prevent ND infection under our experimental conditions where we used four avirulent NDV isolates from waterfowl as vaccine strains and challenged the immunized birds with a virulent NDV isolates from poultry. In any case, excretion of the challenge virus might have occurred. Not ignoring other factors of immune status of the vaccinated flocks, level of hygiene, and cold-chain maintenance to safeguard vaccine doses, protection against ND is dependent on the genotype and immunogenicity of the vaccine virus (35). Therefore, it would be important to develop an effective vaccine that is able to produce and maintain a high antibody level in vaccinated flocks.
In our case, the variations in amino acid between the wild isolates and vaccine strains could be implicated in antigenic variants and vaccine failures leading to annual ND incidence.
Evolution of AAvV-1 isolates has continuously posed threats of emergence of new virulent strains and challenges of diagnosis of ND (55). Many studies have suggested that different strains of AAvV-1 evolve through various degrees of accumulation of point mutations and gene exchange by recombination (23). However, our analysis did not detect any recombination in the HN gene of wild AAvV-1 isolates and from the LaSota vaccine strain, but we obtained signals of recombination in the F gene confirming previous findings in the genotype V viruses (29,30). The virulent AAvV-1 isolates from poultry possessing the virulent motif 112 RRQKRF 117 which is different from that of LaSota strain 112 GRQGRL 117 can be explained by the acquisition of the F2terminus of F gene from virulent viruses retaining their avirulent F1-terminus giving a novel clade of genotype (V) of Ugandan ND-viruses. On the other hand, AAvV-1 isolates from waterfowl maintain in nature their avirulent consensus of F1-terminus. This is again corroborated by studies that have showed that AAvV-1 isolates from poultry highly select pathogenic derivatives of the virus from nonpathogenic precursors (56). The N-terminal of F gene and the rest of the genes in the AAvV-1 isolated from waterfowl had high degree of similarity with those of the genotype II (LaSota strain). Therefore, it was more likely that the F2 terminal of F gene in AAvV-1 isolates from poultry originated from one of the prevalent LaSota (genotype II) strain or isolates from waterfowl. Although it has been noted that the F cleavage site of ancestor viruses, such as the long used live LaSota vaccine strain tends to be conserved, the potential of these viruses mutating to highly virulent strains with passage in chicken is a notable concern. Our current study noted recombination at 316-436 nt region that codes for cleavage site and its surrounding F1terminus region (106-145 aa) which produced the only virulent daughter strain among the low virulent strains. It is not surprising considering that this region was predicted as a potential B-cell epitope by structural-based DiscoTope tool in AAvV-1 isolates from waterfowl. However, such event suggests that further work is warranted to investigate this role in the waterfowl NDV virus isolates. Recombination in F gene AAvV-1 viruses is suggested here as an important parameter to be considered in vaccination with live, attenuated viruses. The F gene recombination in NDV might be a potential reason for immunization failure.

CONCLUSION
We have evaluated AAvV-1s isolates originally assigned to genotypes V and II isolated from chicken and waterfowl, respectively, in Uganda. Based on the deduced amino acid analysis, structures mapping various mutations were found throughout F and HN genes of the isolates that affect the predicted B-cell epitopes; some were exclusive to either AAvV-1 isolates from waterfowl or poultry. Recombination was also confirmed in the F gene of this non-segmented negative-strand RNA virus. These changes could have real impact on the immunity and disease dynamics of the virus in vaccinated as well as immunologically naïve flock. Mixed or closed feeding of poultry with waterfowl in NDV-prevalent areas and frequent and uncontrolled use of vaccines could have serious implications in the control of ND where vaccination was done. Whether all these changes individually or combined influence the antigenicity of the virus leading to serious consequences of vaccine efficacy need to be established.

DATA AVAILABILITY STATEMENT
The datasets presented in this study can be found in online repositories. The names of the repository/repositories and accession number(s) can be found below: https://www.ncbi.nlm. nih.gov/genbank/, LT549451-53, HG937535-91.

ETHICS STATEMENT
The animal study was reviewed and approved by College of Veterinary Medicine Animal Resources and Biosecurity Higher Degrees Research Committee and Uganda National Council of Science and Technology (Approval #HS 776) approved this study. Written informed consent was obtained from the owners for the participation of their animals in this study.

AUTHOR CONTRIBUTIONS
JBO, DKB, HK, JLN, and MO-O designed the study. JBO and DKB analyzed the sequenced data. AW, JBO, and KKM collected the samples, isolated AAvV-1 strains, and identified the viruses. All authors read and approved the final manuscript.

FUNDING
The World Bank Millennium Science Initiative Project (Grant #MSI/03/32/2010) to BD through the Uganda National Council of Science and Technology supported the sample collection, virus isolation, and sequencing of the isolates.