Molecular epidemiology of Avian Rotaviruses Group A and D shed by different bird species in Nigeria

In 2011 and 2013, fecal samples (n = 349) were collected from domestic birds in Ogun and Oyo States, Southwestern Nigeria (Table 1). The samples were collected from different farm types (classified as backyard farms, Farm type 1: farms with less than 500 birds, Farm type 2: farms with between 500 and 10,000 birds, and Farm type 3: farms with over 10,000 birds) and at live bird markets (Table 1). The collection sites (n > 40) were randomly selected to represent the avian RV situation within the different farm types. In Oyo state, no birds were sampled in backyard farms. Besides chickens (Gallus gallus domesticus; n = 255), ducks (Anas platyrhynchos; n = 24), guinea fowls (Numida meleagris; n = 25), pigeons (Columba livia; n = 19), quails (Coturnix coturnix; n = 16) and turkeys (Meleagris gallopavo; n = 10) were included. In 2011, age and health status of the flocks were recorded. In 2013, various bird species were sampled irrespective of their health status.

Samples were directly placed on ice, transported to the laboratory and preserved at −20 °C until shipment to Luxembourg, where they were stored at −80 °C until further processing.

Feces were resuspended in 500 μl of virus transport medium [30]. Samples were cleared at 2200 rpm for 20 min and 140 μl of supernatant medium was used for RNA extraction using QIAamp Viral RNA Mini Kits (Qiagen, Venlo, The Netherlands).

Prior to the RT-PCR, double-stranded RNA was denatured at 95 °C for 2 min followed by cooling on ice. Detection and sequencing PCRs were performed using the Qiagen one-step RT-PCR kit (Qiagen). RVA positivity was detected by real-time RT-PCR targeting the VP6 gene [13] and/or by the conventional PCR targeting the NSP4 gene [25]. RVD was detected by a real-time RT-PCR targeting the VP6 gene [13]. Suboptimal probe binding due to frequent mutations at the binding sites were suspected based on low fluorescence signals in both real-time RT-PCRs and detection was therefore confirmed by gel electrophoresis. The Avian RVA (cell culture supernatant, strain RVA/Chicken-tc/DEU/02V0002G3/2002/G19P[30]) and RVD (intestinal content, strain RVD/Chicken-wt/NLD/10 V0133/2010/GXP[X]) strains, kindly provided by Dr. P. Otto (Friedrich-Loeffler Institute, Germany), served as positive controls.

For sequencing of RVA and RVD positive samples, additional primers were designed and evaluated with Geneious software (version 7.1.7; Biomatters Limited; Auckland, New Zealand [http://​www.​geneious.​com]) [31] and Primer3Plus (http://​primer3plus.​com/​cgi-bin/​dev/​primer3plus.​cgi) for partial amplification of VP4, VP6, VP7 of RVA and VP6 and VP7 of RVD (Additional file 1). NSP4 sequences were obtained by using the detection primers [25]. Amplification conditions were as follows: 50 °C for 30 min, 94 °C for 15 min with subsequent 40 cycles of denaturation at 94 °C for 30 s, annealing at 53 °C for 30 s, extension at 72 °C for 60 s. RT-PCR was performed with 3 μl of RNA and 22 μl of PCR mix (containing 1× Qiagen OneStep RT-PCR Buffer, 1.25 mM MgCl₂, dNTPs at 400 μM each, 1 μl of Qiagen OneStep RT-PCR Enzyme Mix, 0.5 μM of each primer and PCR-grade H₂O to adjust the final volume). Positive samples were identified by gel-electrophoresis and amplicons of the appropriate size excised from the gel and purified with the QIAquick gel extraction kit (Qiagen). Specific PCR products were directly purified using the JetQuick™ extraction kit (Genomed, Löhne, Germany). Sequencing was performed using the BigDye terminator kit (Applied Biosystems, Foster City, CA) on an ABI 3130 sequencer (Applied Biosystems). RVA genotype classification of partial gene sequences was done as recommended by RCWG [24, 32].

RV sequences were assembled in Geneious software [31] and trimmed before further analysis. RV sequences from this study were compared to all avian RV sequences available in GenBank database (http://​www.​ncbi.​nlm.​nih.​gov/​Genbank/​index.​html). Alignments were obtained by applying the ClustalW multiple alignment method (EMBL; Heidelberg, Germany) and poorly aligned regions trimmed using Gblocks as implemented in SeaView software (version 4; CNRS; Villeurbanne, France [http://​doua.​prabi.​fr/​software/​seaview]) [33‐35]. For each alignment, the best-fit model of nucleotide substitution was selected using JModeltest (https://​github.​com/​ddarriba/​jmodeltest2) [36] (Additional file 2). Bayesian analyses were applied as statistical inference methods of the phylogenetic analysis, as described before [37]. Representative isolates for which sequences from several genes were obtained or that were detected in new RV host species are shown in Figs. 1, 2 and 3. Additional phylogenetic trees based on all long sequences from this study and including also more GenBank sequences can be found in the supplementary material (Additional files 3, 4, 5 and 6). Only posterior probability values >0.7 are depicted in the phylogenetic trees.

Statistical analyses were performed in R software (version 3.1.0.; R Foundation for Statistical Computing, Vienna, Austria [https://​www.​r-project.​org/]) [38]. Two-sided Fisher’s exact test was applied to identify potential factors influencing virus shedding in chickens: age group, symptoms (diarrhea, increased mortality or no overt symptoms) and RVA-RVD-coinfection (yes/no). Logistic regression analysis was applied to predict whether RVA shedding is affected by age group and RVA-RVD-coinfection. The overall significance of the models was assessed by chi-squared tests and the individual effect of the categorical factors of the model by Wald tests. The packages aod [39] and Rcpp [40] in R were used to test for association between the categorical variables and RV shedding.

In total, 36.1% (126/349, 95% CI: 31–41%) and 31.8% (111/349, 95% CI: 27–37%) of all fecal samples were positive for avian RVA and RVD (Table 1). RVA and RVD co-infection was revealed for 15.2% (53/349, 95% CI: 12–19%) of the birds. RVD shedding among chickens was increased in farms with more than 10,000 birds (p = 0.04) and significantly higher shedding rates were observed in Oyo than in Ogun state (41.3 and 30.5%; p = 0.007). There was no association between RVA shedding among chickens and collection site or state. Mostly asymptomatic chickens shed RVs. When taking into account only the chickens for which the health status was recorded (n = 225), only 18.9% (28/148) of the RVA positive chickens and only 29.7% (44/148) of the RVD positive chickens and were symptomatic. Thus symptomatic chicken were significantly less likely to shed RVA and RVD (Odds Ratio = 0.2 and 0.3, 95% CI = 0.1–0.4 and 0.3–0.8, p < 0.001 and 0.01). Logistic regression analysis predicted that RVA positivity is affected by age and RVD coinfection (p = 0.002). RVA shedding was significantly increased in chickens that shed also RVD (Odds Ratio = 2.4, 95% CI = 1.5–4.6, p = 0.004). Although the overall effect of age group was significant in the Wald test (χ2 = 39.9, degrees of freedom = 3, p < 0.001), a significant difference between age groups could only be revealed for age groups 1 and 2 (χ2 = 5.0, degrees of freedom = 1, p = 0.025). Chickens aged 26–140 days were more likely to shed RVA than 1–25 days old chickens (Odds Ratio = 1.11, 95% CI = 0.57–2.16).

Besides chickens, other bird species were also positive for RVA (i.e. 23/25 guinea fowls, 13/19 pigeons, 2/16 quails, 5/24 ducks and 1/10 turkeys) and RVD (i.e. 3/25 guinea fowls, 11/19 pigeons, 3/16 quails, 1/24 ducks and 1/10 turkeys). The majority (63.8%, 60/94) of those non-chicken birds were sampled at live bird markets.

Sequencing success was generally low and dependent on host species, RV group and genome segment. This was most probably due to poor sample quality, low copy numbers in fecal samples and/or the low specificity of primers designed on the basis of only a few avian RV sequences available in public databases. Nevertheless, sequencing data was obtained from 69% (87/126) of RVA (Table 2) and 100% (111/111) of RVD positive samples.

RVD sequences showed little diversity (>98% nucleotide identity for VP6 and >99% nucleotide identity for VP7) and were clearly distinct from RVA (Fig. 1). Both phylogenetic and nucleotide identity analyses of partial RVD sequences showed that all VP6 sequences grouped into two major clusters (Fig. 1a) with less than 90% nucleotide identity. Similarly, partial VP7 sequences grouped into two major clusters with less than 83% nucleotide identity (Fig. 1b).

Genotype assignment of RVA can be retrieved from Table 2. In general, RVA strains from the same place and time point had highly similar sequences. On the phylogenetic tree of NSP4, the first diverging event led to two well supported lineages: one comprising the avian and the other the mammalian RVA strains (Fig. 2). All avian NSP4 strains fell into four clusters, corresponding to four distinct NSP4 genotypes (Fig. 2, Additional file 3). Genotype E10 (reference strain RVA/Chicken-tc/DEU/02V0002G3/2002/G19P[30]) regrouped most of our sequences from chickens and guinea fowls. The remaining sequences from various bird species (including pigeons, ducks, chickens, guinea fowls and quail) were closely related to two fox and dove sequences (strains: RVA/Fox-tc/ITA/288356/2011/G18P[17] and RVA/spotted dove/CHN/ARONSP4/2014/GxP[X]). Those sequences were assigned to the new NSP4 genotype, E19 (Table 2).

Most partial VP4 sequences from chickens and guinea fowls clustered together with RVA/Chicken-tc/DEU/06 V0661/2006/G19P[31] within genotype P[31] (Fig. 3a; Additional file 3). Similar to NSP4, all pigeon and duck sequences and a few chicken and guinea fowl sequences clustered together and were assigned to genotype P[17], that includes also the above fox strain and the reference strain RVA/Pigeon-tc/PO-13/1983/G18P[17]. The phylogenetic trees of NSP4 and VP4 had topologies similar to those obtained for VP6 and VP7 (Figs. 2 and 3; Additional files 5 and 6). Most chicken and guinea fowl strains were assigned to VP6 genotype I11 and VP7 genotype G19. The remaining strains were closely related to the fox and the pigeon PO-13 strains and assigned to VP6 genotype I4 and VP7 genotype G18 (Fig. 3b and c).

Thus, the majority of our sequences had one of the following two genotype constellations: G19-P[31]-I11-E10 or G18-P[17]-I4-E19. However, 7/87 samples showed different genome constellations possibly indicative of genetic reassortment (Table 2).